Land Use Change
-
Upload
ayue-ghana -
Category
Documents
-
view
207 -
download
9
Transcript of Land Use Change
Land UseChange
Science, Policyand Management
42963_C000.indd 1 11/6/07 11:18:13 AM
42963_C000.indd 2 11/6/07 11:18:13 AM
Land UseChange
Science, Policyand Management
Edited byRichard J. Aspinall
Michael J. Hill
CRC Press is an imprint of theTaylor & Francis Group, an informa business
Boca Raton London New York
42963_C000.indd 3 11/6/07 11:18:13 AM
CRC PressTaylor & Francis Group6000 Broken Sound Parkway NW, Suite 300Boca Raton, FL 33487-2742
© 2008 by Taylor & Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government worksPrinted in the United States of America on acid-free paper10 9 8 7 6 5 4 3 2 1
International Standard Book Number-13: 978-1-4200-4296-2 (Hardcover)
This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the conse-quences of their use.
No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.
Library of Congress Cataloging-in-Publication Data
Land use change : science, policy, and management / Richard J. Aspinall and Michael J. Hill [editors].
p. cm.Includes bibliographical references and index.ISBN 978-1-4200-4296-2 (hardcover : alk. paper)1. Land use--Environmental aspects. 2. Land use--Management. 3. Land
use--Case studies. I. Aspinall, Richard J. II. Hill, Michael J. (Michael James), 1937- III. Title.
HD108.3.L362 2008333.73’13--dc22 2007027752
Visit the Taylor & Francis Web site athttp://www.taylorandfrancis.com
and the CRC Press Web site athttp://www.crcpress.com
42963_C000.indd 4 11/6/07 11:18:15 AM
RJA — to Chloe, for her company
MJH — to my mother, Mavis Hill, for her lifelong friendship
42963_C000.indd 5 11/6/07 11:18:16 AM
42963_C000.indd 6 11/6/07 11:18:16 AM
vii
ContentsFigures......................................................................................................................ixPermissions...............................................................................................................xiPreface.................................................................................................................... xiiiAcknowledgments...................................................................................................xvIntroduction...........................................................................................................xviiEditors.................................................................................................................. xxiiiContributors..........................................................................................................xxv
Part I theory and Methodology
Chapter 1 Basic.and.Applied.Land.Use.Science....................................................3
Richard J. Aspinall
Chapter 2 Developing.Spatially.Dependent.Procedures.and.Models.for.Multicriteria.Decision.Analysis:.Place,.Time,.and.Decision.Making.Related.to.Land.Use.Change................................................. 17
Michael J. Hill
Part II Comparative regional Case Studies
Chapter 3 Spatial.Methodologies.for.Integrating.Social.and.Biophysical.Data.at.a.Regional.or.Catchment.Scale........................... 43
Ian Byron and Robert Lesslie
Chapter 4 An.Integrated.Socioeconomic.Study.of.Deforestation.in.Western.Uganda,.1990–2000.............................................................. 63
Ronnie Babigumira, Daniel Müller and Arild Angelsen
Chapter 5 Modeling.Unplanned.Land.Cover.Change.across.Scales:.A.Colombian.Case.Study.................................................................... 81
Andres Etter and Clive McAlpine
Chapter 6 Landscape.Dynamism:.Disentangling.Thematic.versus.Structural.Change.in.Northeast.Thailand...........................................99
Kelley A. Crews
42963_C000.indd 7 11/6/07 11:18:16 AM
viii Contents
Chapter 7 Developing.a.Thick.Understanding.of.Forest.Fragmentation.in.Landscapes.of.Colonization.in.the.Amazon.Basin........................... 119
Andrew C. Millington and Andrew V. Bradley
Chapter 8 Urban.Land.Use.Change,.Models,.Uncertainty,.and.Policymaking.in.Rapidly.Growing.Developing.World.Cities:.Evidence.from.China......................................................................... 139
Michail Fragkias and Karen C. Seto
Part III Synthesis and Prospect
Chapter 9 Synthesis,.Comparative.Analysis,.and.Prospect............................... 163
Michael J. Hill and Richard J. Aspinall
Index....................................................................................................................... 179
42963_C000.indd 8 11/6/07 11:18:17 AM
ix
FiguresFigure.I. Conceptual.model.of.sustainability.choice.space.Figure.2.1. The.concept.of.grazing.piospheres.interacting.with.landscape.structure.Figure.2.2. Time.series.approaches.to.extracting.signal.summary.indicators.Figure.2.3. Nonbiophysical.time.series.also.provide.potential.indicators.Figure.2.4. The.spatial.pattern.of.temporal.signals.may.be.grouped.by.applying.prin-
cipal.components.analysis.Figure.2.5. Temporal.signals.are.usually.based.on.biophysical.or.human.phenomena.Figure.2.6. A.framework.for.application.of.multicriteria.analysis.of.ecosystem.services.
to.the.range.land.environment.Figure.2.7. Cognitive.mapping.interface.Figure.2.8. Schema.for.transformation.of.spatiotemporal.information.Figure.3.1. Land.use.change. in. the.Barmera,.Berri,.and.Renmark.areas.of.South.
Australia.Figure.3.2. Location.of.the.Glenelg.Hopkins.region.Figure.3.3. Land.managers’.perception.of.salinity.and.mapped.salinity.discharge.sites.Figure.3.4. Land.managers.who.manage.properties.near.areas.of.high.conservation.
value.Figure.4.1. Uganda.study.area.showing.the.distribution.of.reforestation.Figure.4.2. Conceptual.framework.for.analysis.Figure.4.3. Cumulative.deforestation.by.parishes.across.Uganda.Figure.4.4. Relationship.between.deforestation.and.distance.from.roads.Figure.5.1. Location.of.Colombia.Figure.5.2. Predicted.forest.presence.according.to.the.best.model.Figure.5.3. Predicted.deforestation.hot.spots.Figure.5.4. Forest.maps.of.the.colonization.front.Figure.5.5. Spatial.location.of.the.local.hot.spots.of.deforestation.and.regeneration.Figure.5.6. Logistic.pattern.of.forest.cover.decline.during.the.transformation.process.Figure.5.7. Forest.cover.change.at.the.local.scale.in.the.Colombian.Amazon.Figure.6.1. The.panel.process,.conducted.at.both.the.pixel.and.patch.levels.Figure.6.2. Typical.nuclear.village.settlement.as.seen.in.1:50,000.scale.Figure.6.3. LULC.in.the.greater.study.area.in.several.water.years.Figure.6.4. Stylized.LULC.trends.observed.and/or.reported.in.northeast.Thailand.Figure.6.5. Stylized.LULC.pattern.metric.change.for.the.interspersion/juxtaposition.
index.Figure.6.6. The.change.in.configuration.from.1972/1973.to.1975/1976.Figure.7.1. Thick.and. thin.understandings.of. forest. fragmentation.along. roads. in.
lowland.forests.of.the.Amazon.Basin.Figure.7.2. Sequence.of.images.showing.progressive.deforestation.Figure.7.3. Six-phase.conceptual.model.of.forest.fragmentation.Figure.7.4. Metrics.calculated.for.the.conceptual.model.Figure.7.5. Selected.metrics.
42963_C000.indd 9 11/6/07 11:18:17 AM
x Figures
Figure.7.6. Metrics.calculated.for.Communidad.Arequipa.Figure.8.1. Population.distribution.of.the.world’s.urban.agglomerations.Figure.8.2. The.Pearl.River.Delta.in.southeast.China.Figure.8.3. Predicted.probability.of.change.to.urban.areas.between.2004.and.2012.
for.Shenzhen.Figure.8.4. Predicted.probability.of.change.to.urban.areas.between.2004.and.2012.
for.Foshan.Figure.8.5. Predicted.probability.of.change.to.urban.areas.between.2004.and.2012.
for.Guangzhou.Figure.8.6. Shenzhen,.Guangzhou,.and.Foshan.urban/nonurban.prediction.for.2012.Figure.9.1. Integrating.frameworks.that.seek.interdisciplinary.definition.and.focus.
on.key.questions.of.management.of.land.use.change.in.coupled.human.environment.systems.
Figure.9.2. A.diagram.of. the.major. elements.of. the. coupled.human.environment.system.
Figure.9.3. Connecting.the.paradigms.Figure.9.4. The.context.for.spatial.and.temporal.analysis.in.determining.the.mix.of.
landscapes.of.fate.and.desire.
42963_C000.indd 10 11/6/07 11:18:18 AM
xi
PermissionsFigure.I. Potschin,.M..and.Haines-Young,.R.,.“Rio+10,”.sustainability.science.
and.landscape.ecology,.Landscape and Urban Planning,.75,.162–174,.2006..Reprinted.with.permission.from.Elsevier.
Figure.5.2. Etter,.A..et.al.,.Regional.patterns.of.agricultural.land.use.and.defor-estation.in.Colombia,.Agriculture, Ecosystems & Environment,.114,.369–386,.2006..Reprinted.with.permission.from.Elsevier.
Figure.5.3. Etter,.A..et.al.,.Regional.patterns.of.agricultural.land.use.and.defor-estation.in.Colombia,.Agriculture, Ecosystems & Environment,.114,..369–386,.2006..Reprinted.with.permission.from.Elsevier.
Figure.5.4. Etter,. A.. et. al.,. Characterizing. a. tropical. deforestation. wave:. the.Caquetá. colonization. front. in. the. Colombian. Amazon,. Global Change Biology,.12,.1409–1420,.2006..Reprinted.with.permission.from.Blackwell.Publishing.
Figure.5.5. Etter,. A.. et. al.,. Characterizing. a. tropical. deforestation. wave:. the.Caquetá. colonization. front. in. the. Colombian. Amazon,. Global Change Biology,.12,.1409–1420,.2006..Reprinted.with.permission.from.Blackwell.Publishing.
Figure.5.6. Etter,.A..et.al.,.Modeling.the.conversion.of.Colombian.lowland.eco-systems.since.1940:.drivers,.patterns.and.rates,.Journal of Environmental Management,.79,.74–87,.2006..Reprinted.with.permission.from.Elsevier.
Figure.5.7. Etter,. A.. et. al.,. Unplanned. land. clearing. of. Colombian. rainforests:.spreading.like.disease?,.Landscape and Urban Planning,.77,.240–254,.2006..Reprinted.with.permission.from.Elsevier.
Table.5.1. Etter,.A..et.al.,.Modeling.the.conversion.of.Colombian.lowland.eco-systems.since.1940:.drivers,.patterns.and.rates,.Journal of Environmental Management,.79,.74–87,.2006..Reprinted.with.permission.from.Elsevier.
Figure.9.1a. IGBP.Secretariat,.Global Land Project: Science Plan and Implementation Strategy,.GLP,.IGBP.Report.53/IHDP.Report.19,.64.pp,.2004..Reprinted.with.permission.from.the.International.Geosphere.Biosphere.Program,.Stockholm..
Figure.9.1d. Steinitz,.C.. et. al.,.Alternative Futures for Changing Landscapes: The Upper San Pedro River Basin in Arizona and Sonora,.2003..Reprinted. in. modified. form. with. permission. from. Island. Press,.Washington.D.C.
42963_C000.indd 11 11/6/07 11:18:18 AM
42963_C000.indd 12 11/6/07 11:18:18 AM
xiii
PrefaceThere.is.a.growing.international.community.of.scholars.who.work.on.themes.and.issues.that.are.central.to.understanding.land.use.change.as.a.fundamental.factor.in.the.operation.of.environmental.and.socioeconomic.systems.at.scales.from.local.to.global..This.book.presents.a.series.of.chapters.that.address.spatial.theories,.methodologies,.and.case.studies.that.support.an.integrated.approach.to.analysis.of.land.use.change..Case.studies.provide.a.series.of.regional.test.beds.for.theories.and.methodologies,.and.the.empirical.content.of.the.case.studies.allows.a.comparative.analysis.of.land.use.change.issues.from.diverse.places..Case.studies.thus.are.an.important.mecha-nism,.not.only. for.understanding. the.multiscale.nature.and.consequences.of. land.use.change.and.the.particular.history.of.changes.in.case.study.locations,.but.also.for..eliciting.general.principles. and. factors.of. importance. in.different. socioeconomic,.cultural,.and.environmental.contexts..This.generalization.from.case.studies.provides.input.to.decision.making.related.to.possible.future.trajectories.of.change.
The.chapters. in. this.book.were.written. to.present. this. interaction.of. theories,.methodologies,.and.case.studies..Additionally,.all. the.authors.are.concerned.with.links. between. science. and. decision. making,. especially. in. relation. to. policy. and.practice..Each.author.attempts.to.enable.effective.communication.between.the.aca-demic.content.of.his.or.her.work.with.decision.makers,.including.those.concerned.with.policy.and.those.concerned.with.land.management.
All.the.chapters.were.first.presented.in.a.paper.session.at.the.6th.Open.Meeting.of.the.International.Human.Dimensions.of.Global.Environmental.Change.Programme.in.Bonn,.October.2005—Spatial.Theory. and.Methodologies. for. Integrated.Socio-economic.and.Biophysical.Analysis.and.Modeling.of.Land.Use.Change:.An.Inter-national.Test.of.Theory.and.Method.and.a.Comparative.Synthesis.of.Change.at.Local.and.Regional.Scales.
42963_C000.indd 13 11/6/07 11:18:19 AM
42963_C000.indd 14 11/6/07 11:18:19 AM
xv
AcknowledgmentsWe.thank.the.authors.of.the.chapters.for.their.prompt.responses.to.all.our.requests..We.also.thank.Professor.Roy.Haines-Young.and.Dr..Marion.Potschin,.University.of. Nottingham,. for. their. participation. in. the. session. at. the. International. Human..Dimensions.of.Global.Environmental.Change.Programme.open.meeting.in.Bonn,.Germany,. in.October.2005.at.which. the.papers.and. ideas. in. this.book.were.first..presented.and.discussed..We.also.thank.Tai.Soda.and.Amber.Donley.from.Taylor.&. Francis/CRC. Press. for. their. patient. prompting. and. management. during. the..preparation.of.this.book.
42963_C000.indd 15 11/6/07 11:18:19 AM
42963_C000.indd 16 11/6/07 11:18:19 AM
xvii
IntroductionThis.book.addresses.spatial.theories.and.methodologies.that.support.an.integrated.approach.to.analysis.of.land.use.change..The.work.focuses.on.spatial.representation.and.modeling. for. scientific.study.and.development.of.management.understanding.of. complex,.dynamic. land.use. systems..Case. studies. are.used. to.develop. specific.examples,.not.only.of.change.in.the.study.areas.used.by.the.different.case.studies,.but.also.to.illustrate.the.variety.and.commonality.of.data.sources,.methods,.and.issues.faced.when.studying.and.developing.understanding.of.land.use.change.
Land.use.and.land.cover.change.reflect.a.variety.of.environmental.and.social..factors.1,2.This.necessitates.that.an.equally.varied.suite.of.data.be.used.for.effective.analysis..Remote. sensing,. from.both. satellites. and.air. photos,. provides. a. central.resource.for.study.of.land.use.and.land.cover.change..Socioeconomic.surveys.and.censuses. provide. an. equally. important. source. of. data. on. social. and. economic.systems..Atlases.and.other.map.sources.can.provide.data.on.specific.environmental.and. socioeconomic. characteristics. of. an. area.. Similarly,. household. and. other..surveys.can.give.information.on.motivations,.values,.behaviors,.and.actions.of.deci-sion.makers.and.land.managers,.and.there.are.many.other.specific.data.of.relevance.for.study.of.land.use.change..These.different.data.do.vary,.however,.in.their.avail-ability,.currency,.and.relevance.for.study.of.land.change,.and.this.is.reflected.in.the.variety.of.case.studies.in.the.published.literature.and.the.specific.issues.and.ques-tions.they.address.
Similarly,.methods.that.are.appropriate.for.analysis.of.the.different.data.sources.are.varied,.and.study.of. land.change. typically.may. involve.methods. from.remote.sensing,.GIS.(geographic.information.systems),.process.and.empirical.modeling,.and.spatial.and.statistical.analysis,.among.others,.as.well.as.methods.that.can.address.qualitative.and.quantitative.data..The.need.to.study.land.use.as.a.coupled.natural.and.human.system.adds.to.the.complexity.of.methods.needed.since.interaction.of.systems.and.integration.in.analysis.present.major.methodological.challenges.
Thus,.the.purpose.of.this.book.is.to.illustrate.issues.and.opportunities.for.study.of.land.change.by.discussing.relevant.science.and.methods.and.by.providing.a.series.of. exemplary. case. studies. that. present. approaches,. data,. and. methods. to. study.land.use.change..Various.themes.run.through.the.book,.including.the.wide.use.of.remote.sensing,.GIS.(often.implicit.and.supporting.analysis.rather.than.explicit.and.a.goal.of.the.work),.and.a.variety.of.statistical.and.other.modeling.methods.aimed.at.understanding.spatial.and.temporal.trends,.patterns,.and.dynamics..The.central.importance.of.understanding.the.social,.economic,.policy,.cultural,.and.institutional.contexts.and.influences.on.land.change.is.also.a.recurrent.theme..
A.further.set.of.issues,.and.a.major.context.for.interpretation.of.results.of.the.chapters,.is.associated.with.sustainable.development.and.the.emergence.of.sustain-ability.science.3,4,5.Sustainable.development.provides.a.context.and.focus.for.scientific.research4,6,7.as.well.as.a.guide.and.impetus.for.civic.debate.and.decision-making.5,8.Potschin.and.Haines-Young7.present.a.conceptual.model.(the.“tongue.model”).of.the.
42963_C000.indd 17 11/6/07 11:18:20 AM
xviii Introduction
state.of.a.landscape.and.its.development.trajectory.over.time.(Figure.I)..This.treats.landscapes.as.multifunctional.spaces.and.is.founded.on.a.paradigm.of.natural.capital.and.ecosystem.services.9,10.The. tongue.model.places.boundaries.on. sustainability.defined.by.a.combination.of.biophysical.limits.of.ecological.systems.in.a.landscape;.this.reflects.not.only.environmental.conditions.and.ecosystem.services.that.can.be.achieved,.but.also.social.and.cultural.values,.as.well.as.costs.and.uncertainties.that.can.be.accepted..The.tongue.model7.provides.a.useful.framework.for.development.and.evaluation.of. the.models.and.case.studies.of. land.use.and. land.cover.change.presented.in.Chapters.3.through.8..The.case.studies.also.provide.a.test.of.the.gen-eral.applicability.of.the.tongue.model.since.they.explore.a.diverse.range.of.socio-economic,.cultural,.and.biophysical.contexts.for.land.use.systems..Effective.linkage.of.science,.policy,.land.management.practices,.and.decision-making.related.to.land.and. ecological. systems. will. increasingly. require. frameworks. such. as. this. with. a.focus. on. integration. of. human. and. natural. systems. that. are. explicitly. directed. at.achieving.consensus.based.on.a.strong.scientific.foundation.
The.case.studies.also.focus.attention.on.terminology,.especially.related.to.land.use. and. land. cover..Land.use. refers. to. the. social,. economic,. cultural,. political,. or.other. value. and. function. of. land. resources.. This. contrasts. with. land. cover,. which.refers.to.the.biophysical.properties.of.the.land.surface.11.As.such,.land.use.and.land.cover,.although.related,.are.distinct.from.one.another..The.distinctions.are.important.for. understanding. causes. and. consequences.of. change. in. land. systems,. as.well. as.for.mapping.and.other.measurement.of. land.system.properties.. Indeed,.one.of. the.main.challenges.to.the.study.of.land.systems.may.be.the.tendency.for.conflation.and.confusion.of. the.terms. land use.and. land cover..This.partly.relates. to.wide.use.of.remotely.sensed.imagery.as.a.source.of.land.system.data.but.also.to.ambiguity.and.lack.of.distinction.in.definition.of.categories.of.interest..For.example,.the.Anderson.
Figure i Conceptual.model.of.sustainability.choice.space..(From.Potschin.and.Haines-Young.7.With.permission.)
Development options defined bydifferent interest groups
State indicator (e.g., area of a particular habitat)
Set of landscape configurations that more or less sustainrequired landscape outputs for society. Boundary ofenvelope set by minimum and maximum value of statevariable required to deliver outputs, given people’sattitudes to risks and costs, and long-term uncertainties.
+
–
Z1 – Z0
time
42963_C000.indd 18 11/6/07 11:18:22 AM
Introduction xix
classification,12.which.is.widely.used.as.a.taxonomy.for.land.surface.description.based.on.imagery.from.different.sources.(as.well.as.field.mapping),.and.many.other.clas-sifications.confound.cover.and.use.classes.within.the.same.formal.mapping.legend..This.may.be.based.on.apparent.similarity.in.cover.and.use..For.example,.forestry.(use).and.forest.(cover).may.refer.to.the.same.place.on.the.ground..It.does,.however,.have.important. consequences. for. process-level. understanding. and. modeling. of. change,.presenting. a. limitation. for. recording. change. and. consequently. for. understanding..Recently,.efforts.have.been.made.to.produce.separate.classifications.of.land.use13.and.land.cover,14.and.the.Global.Land.Project15.identifies.cover.and.use.along.a.continuum.spanning.natural.and.social.systems.
STruCTure OF THe BOOK
The.book.is.organized.in.nine.chapters..Chapters.1.and.2.address.theoretical.and.methodological. issues. and. mechanisms. for. study. of. land. use. systems. as. coupled.human.and.natural.systems..Chapters.3.through.8.provide.a.suite.of.regional.case.studies,.including.a.discussion.of.change.in.rapidly.urbanizing.areas..Collectively,.these.six.chapters.provide.insight.into.the.nature.of.both.land.use.change.and.the.diverse.range.of.socioeconomic,.cultural,.and.biophysical.contexts.of.land.use.sys-tems.across.the.planet..The.case.study.chapters.provide.a.series.of.illustrations.for.many.of.the.frameworks,.issues,.and.methodologies.described.in.the.first.two.chap-ters..The.empirical.content.of.the.case.studies.also.allows.a.comparative.analysis.of.land.use.change.issues.from.diverse.places..
Chapter. 1. considers. basic. science. questions. that. underpin. study. of. land. use.change,. as. well. as. a. suite. of. applied. science. issues. that. influence. the. utility. and.use. of. scientific. information. about. land. use. change. for. policy. and. land. manage-ment..The.chapter.focuses.on.issues.that.influence.the.analysis.of.dynamics.of.land.change;.the.need.for.understanding.the.integration.and.feedbacks.between.different.components.of.land,.climate,.socioeconomic,.and.environmental.systems;.resilience,..vulnerability,.and.adaptability.of.land.use.systems;.scale.issues;.and.accuracy..The.chapter.also.examines.some.issues.relevant.to.effective.communication.and.partner-ship. between. scientists. and. practitioners. responsible. for. decision-making,. either.policy.or.management,.about.land.use.
Full. understanding.of.both. land.use. change.and.decision-making. requires. an.appreciation.of.multiple.characteristics.of. land.use.systems.and.factors. that. influ-ence.change..Multicriteria.decision.analysis.can.provide.a.series.of.tools.to.help.with.both.scientific.study.and.decision-making..Chapter.2.examines.the.development.of.spatially.dependent.procedures.and.models.for.such.multicriteria.decision.analysis..The.approaches.focus.on.reducing. the.complex.responses.and.patterns.associated.with.land.use.change.to.simpler.meaningful.metrics..These.are.based.on.use.of.both.quantitative.and.qualitative.measures.of.patterns.and. trends.and. include.methods.from.signal.processing,.time.series.analysis,.and.analysis.of.spatial.patterns..
The.next.six.chapters.present.a.series.of.case.studies.that.are.exemplars.of.dif-ferent. aspects. of. land. use. science.. In. Chapter. 3. Byron. and. Lesslie. explore. inter-actions.between.people.and.the.environment.as.a.critical.component.in.regional-.and..catchment-scale.natural.resource.management..The.chapter.uses.a.case.study.from.
42963_C000.indd 19 11/6/07 11:18:22 AM
xx Introduction
Australia.to.bring.data.on.the.environment,.from.a.time.series.of.orthophotos,.and.social.systems,.from.a.property.survey,.together..This.integrated.analysis.is.complemented.by.use.of.spatially.referenced.survey.data.to.understand.how.land.managers’.values,.perceptions,.and.practices.relate.to.the.biophysical.environments.they.manage.
Chapter.4,.by.Babigumira,.Müller,.and.Angelsen,.links.deforestation.in.western.Uganda.in.the.1990s.to.the.socioeconomic,.spatial,.and.institutional.contexts.within.which. it.occurred..The.authors.develop.an.empirical.model. that. integrates. socio-economic.data.from.a.national.census.with.spatial.data.derived.from.remote.sensing..The.socioeconomic.survey.informs.on.poverty.and.economic.opportunities,.whereas.the.remotely.sensed.data.represent.the.costs.and.feasibility.of.forests.
Chapter.5,.by.Etter.and.McAlpine,.examines.the.patterns,.processes,.and.drivers.of.unplanned.land.cover.change.in.Colombia.as.representative.of.change.in.the.tropics..Statistical.modeling.is.used.to.predict.changes.in.forest.cover.at.local,.regional,.and.national.levels,.over.times.ranging.from.a.decade.to.a.century..Explanatory.variables.include. both. biophysical. and. socioeconomic. data,. and. these. are. obtained. from. a.range.of.sources,.including.remote.sensing,.maps,.and.surveys..
Chapter.6,.by.Crews-Meyer,.is.a.case.study.of.land.change.in.northeast.Thailand.that.uses.a.time.series.of.satellite-derived.data.within.a.longitudinal.approach,.panel.analysis,.for.modeling.temporal.dynamics..The.approach.also.draws.on.landscape.ecology. to. emphasize. the. importance. of. the. spatial. scale. of. observation. on. the.inference.of.process. and.attempts. to. examine. changes. in. landscape. composition.and.configuration..
Chapter.7,.by.Millington.and.Bradley,.uses.a.case.study.of.deforestation.associ-ated.with.planned.colonization.schemes.in.the.Amazon.Basin.to.develop.a.detailed—thick—understanding.of.deforestation..They.argue.that.the.impacts.of.roads.on.forest.fragmentation.are.agents.of.deforestation.at.one.scale.only.and.that.at.another.scale.the.pattern.of.property.ownership.represents.that.scale.at.which.land.owners.make.decisions.about.forest.clearance.and.regrowth.as.household.responses.to.economic.and.policy.signals.
Chapter. 8,. by. Fragkias. and. Seto,. discusses. issues. of. urban. land. use. change.modeling.and.explores.the.intersection.of.land.use.modeling.with.urban.policy-making.at.different.scales..The.work.also.concentrates.on.the.effects.of.uncertainties.in.data.sources.and.reviews.a.predictive.model.of.rapid.urban.transformation.that.extends.a.standard.modeling.approach.to.provide.a.policy-making.framework.that.explicitly.reduces.uncertainty..Chinese.cities.are.used.as.a.case.study.as.important.exemplars.of.developing.world.cities.
Chapter.9.reviews.findings.of.the.full.set.of.chapters.and.case.studies.and.points.to.the.potential.developments.in.spatial.methods.needed.to.advance.integrated.model-ing.of. land.use. change. and. improve. linkage.between. scientific. study.of. land.use.change. with. policy. and. management. decision-making. at. regional,. national,. and.international.scales.
Richard J. Aspinall Michal J. Hill
42963_C000.indd 20 11/6/07 11:18:23 AM
Introduction xxi
reFereNCeS
. 1.. Geist,.H.,.and.Lambin,.E..F..Proximate.causes.and.underlying.driving.forces.of.tropical.deforestation..Bioscience.52(2),.143–150,.2002.
. 2.. Geist,.H..J.,.and.Lambin,.E..F..Dynamic.causal.patterns.of.desertification..Bioscience.54(9),.817–829,.2004.
. 3.. Kates,.R..W.,.Sustainability Science. Research and Assessment Systems for Sustainability Program. Discussion.Paper.2000-33..Belfer.Center.for.Science.and.International.Affairs,.Kennedy.School.of.Government,.Harvard.University,.Cambridge.MA,.2000..Available.at.http://ksgnotes1.harvard.edu/BCSIA/sust.nsf/pubs/pub7/$File/2000-33.pdf.
. 4.. Kates,.R..W..et.al..Sustainability.science..Science.292,.641–642,.2001.
. 5.. Clark,.W..C.,.and.Dickson,.N..M..Sustainability.science:.The.emerging.research.program..Proceedings of the National Academy of Sciences.100(14),.8059–8061,.2003..
. 6.. Haines-Young,.R..Sustainable.development.and.sustainable.landscapes:.defining.a.new.paradigm.for.landscape.ecology..Fennia.178(1),.7–14,.2000.
. 7.. Potschin,. M.,. and. Haines-Young,. R.. “Rio+10,”. sustainability. science. and. landscape.Ecology,.Landscape and Urban Planning.75,.162–174,.2006.
. 8.. Cash,.D..W..et.al..Knowledge.systems.for.sustainable.development..Proceedings of the National Academy of Sciences.100(14),.8086–8091,.2003.
. 9.. Costanza,.R.,.and.Daly,.H..E..Natural.capital.and.sustainable.development..Conservation Biology.6(1),.37–46,.1992.
. 10.. Daily,.G..C..Nature’s Services: Societal Dependence on Natural Ecosystems..Washington,.DC:.Island.Press,.392.pp..1997.
. 11.. Comber,.A..J.,.Fisher,.P..F.,.and.Wadsworth,.R..A..What.is.land.cover?.Environment and Planning B: Planning and Design.32,.199–209,.2005.
. 12.. Anderson,.J..R..et.al..A LandUse and Land Cover Classification System for Use with Remote Sensor Data..Geological.Survey.Professional.Paper.No..964..Washington,.DC:.U.S..Government.Printing.Office,.1976.
. 13.. Jansen,.L..J..M..Harmonization.of. land.use.class.sets. to. facilitate.compatibility.and.comparability. of. data. across. space. and. time.. Journal of Land Use Science. 1(2–4),.127–156,.2006.
. 14.. Herold,.M..et.al..Evolving.standards.in.land.cover.characterization..Journal of Land Use Science.1(2–4),.157–168,.2006.
. 15.. GLP Science Plan and Implementation Strategy.. IGBP.Report.No.53,. IHDP.Report.No.19..Stockholm,.IGBP.Secretariat..64.pp..2005.
42963_C000.indd 21 11/6/07 11:18:24 AM
42963_C000.indd 22 11/6/07 11:18:24 AM
xxiii
EditorsRichard J. Aspinall. received. a. BSc. (Hons). from. the. University. of. Birmingham.and.a.PhD.from.the.University.of.Hull,.both. in.geography..He.has.worked.in. the.United.Kingdom.including.10.years.at.the.Macaulay.Land.Use.Research.Institute.in..Scotland,.and.in.the.United.States.in.the.Department.of.Earth.Sciences.at.Montana.State.University,.as.chair.of.the.School.of.Geographical.Sciences.at.Arizona.State.University,. and. as. program. director. for. geography. and. regional. science. at. the.U.S..National.Science.Foundation..He.is.now.professor.and.chief.executive.at. the.Macaulay. Institute. in. Scotland,. an. interdisciplinary. research. institute. addressing.sustainable.development.and.land.use..His.research.interests.are.in.the.areas.of.land.use.and.land.cover.change,.analysis,.and.modeling.of.coupled.natural.and.human.systems,.and.methods.and.applications.in.environmental.geography.including.GIS,.landscape.ecology,.biogeography,.geomorphology,.and.hydrology.
Michael J. Hill. received. the. BAgrSci. and. MAgrSci. from. LaTrobe. University,..Bundoora,.Australia,.and.a.PhD.from.the.University.of.Sydney,.Australia,.in.1985..He.spent.12.years.in.the.CSIRO.Division.of.Animal.Production.and.then.6.years.in. the.Bureau.of.Rural.Sciences. in. the.Department. of.Agriculture,.Fisheries. and.Forestry.of. the.Australian.government.where.he.carried.out. research. in.and.con-tributed. to. the.management.of. the.Co-operative.Research.Centre. for.Greenhouse.Accounting..In.2006,.he.became.professor.of.earth.systems.science.in.the.Depart-ment.of.Earth.Systems.Science.and.Policy.at.the.University.of.North.Dakota,.Grand.Forks..He.has.a.background.in.grassland.agronomy,.but.has.been.working.with.spa-tial. information. and. remote. sensing. of. land. systems. for. the. past. 12. to. 15. years..He.has.published.widely.on.agronomy,.ecology,.biogeography,.production.of.grass-lands,.and.radar,.multispectral,.and.hyperspectral.remote.sensing.of.grasslands,.and.more.recently.has.been.involved.in.the.development.of.scenario.analysis.models.for.assessment.of.carbon.dynamics.in.agricultural.and.rangeland.systems..His.current.interests. are. in. the. use. of. MODIS. land. product. data. in. model-data. assimilation,.application.of.quantitative.information.from.hyperspectral.and.multiangle.imaging.to.vegetation.description,.multicriteria.and.decision.frameworks.for.coupled.human.environment.systems,.and.methods.and.approaches.to.application.of.spatial.data.for.land.use.management.
42963_C000.indd 23 11/6/07 11:18:25 AM
42963_C000.indd 24 11/6/07 11:18:25 AM
xxv
Contributors
Arild AngelsenArild. Angelsen. is. an. associate. professor. of. economics. in. the. Department. of.Economics.and.Resource.Management.at.the.Norwegian.University.of.Life.Sciences..His. research. interests. are. in. economic. analysis. and. assessment. of. projects. and..policies. in. developing. countries,. particularly. within. agriculture. and. forestry,. and.related.to.poverty,.environmental.effects,.and.use.of.natural.resources.
Richard J. AspinallRichard. J..Aspinall. is.professor.and.chief. executive.at. the.Macaulay. Institute,. an.interdisciplinary.research.institute.addressing.sustainable.development.and.land.use..His.research.interests.are.in.the.areas.of.land.use.and.land.cover.change,.analysis,.and.modeling.of.coupled.natural.and.human.systems,.and.methods.and.applications.in.environmental.geography.including.GIS,.landscape.ecology,.biogeography,.geo-morphology,.and.hydrology.
Ronnie Babigumira.Ronnie.Babigumira.is.a.PhD.student.in.the.Department.of.Economics.and.Resource..Management.at.the.Norwegian.University.of.Life.Sciences..His.research.interest.is.in.land.use.change.in.Africa.
Andrew V. BradleyAndrew. V.. Bradley. is. a. research. scientist. at. the. Natural. Environmental. Research..Council.(NERC).Centre.for.Ecology.and.Hydrology.(CEH).at.Monks.Wood,.United..Kingdom..His.research.interests.focus.on.understanding.socioeconomic.drivers.of.forest.loss,.forest.fragmentation,.and.agricultural.land.use.and.land.cover.change.
Ian ByronIan.Byron’s.research.interests.include.analysis.of.sociological.survey.data.and.integra-tion.of.survey.results.with.biophysical.spatial.data..He.co-authored.the.chapter.while.a.social.scientist.with.the.Bureau.of.Rural.Sciences,.a.science-policy.agency.within.the.Australian.Department.of.Agriculture,.Fisheries.and.Forestry.
Kelley A. CrewsKelley. A.. Crews. is. associate. professor. of. geography. and. the. environment. at. the..University. of. Texas,. Austin,. where. she. also. directs. the. GIScience. Center.. Her..thematic.research. interests. include.remote.sensing,.population–environment. inter-actions,. landscape.ecology,.and.policy.analysis..Geographically.her.work. focuses.on.tropical.and.subtropical.forest/savanna/wetland.ecotones.in.the.western.Amazon,.the.Okavango.Delta.of.Botswana,.and.Northeast.Thailand.
42963_C000.indd 25 11/6/07 11:18:25 AM
xxvi Contributors
Andres EtterAndres.Etter.is.a.professor.in.the.Department.of.Ecology.and.Territory.of.the.School.of.Environmental.and.Rural.Studies.at.Javeriana.University.(Bogotá-Colombia)..He.has.20.years’.experience.in.landscape.ecological.mapping.and.research.in.the.Colombian.Amazon.forests,.Orinoco.savannas,.and.Andean.Montane.forest.regions.integrating.biophysical,.socioeconomic,.and.historic.data,.using.remote.sensing.and.GIS..His.cur-rent.research.deals.with.the.modeling.and.understanding.of.land.use.change.processes.aimed.at.more.informed.and.dynamic.conservation.planning.processes.
Michail FragkiasMichail. Fragkias. is. the. executive. officer. of. the. IHDP. Urbanization. and. Global..Environmental.Change.Core.Project.hosted.by.the.Global.Institute.of.Sustainability.at.Arizona.State.University..His.interests.focus.on.urban.land.use.change.and.the.interaction. of. urban. spatial. structure. with. the. environment.. He. employs. (spatial)..statistical. analysis,. simulations,. and. geographical. information. systems. (GIS). to.study.the.significance.of.social,.economic,.and.political.drivers.of.urban.land.use.change. in. China.. He. completed. his. undergraduate. studies. in. economics. at. the.National.University.of.Athens.in.Greece.and.his.MA.and.PhD.in.economics.at.Clark..University.in.Massachusetts..He.co-authored.the.present.chapter.while.being.a.post-doctoral.scholar.at.the.Center.for.Environmental.Science.and.Policy.(CESP).at.the.Freeman.Spogli.Institute.for.International.Studies.(FSI).at.Stanford.University.
Michael J. HillMichael.J..Hill.is.a.professor.of.earth.systems.science.in.the.Department.of.Earth.Systems.Science.and.Policy.at.the.University.of.North.Dakota..His.research.inter-ests.include.hyperspectral.remote.sensing,.biogeochemical.processes,.and.land.use.change. in. savanna. systems,. and. analysis. of. coupled.human.environment. systems.using.spatial.multicriteria.analysis.and.spatial.analysis.methods.
Robert LesslieRob. Lesslie. is. a. principal. research. scientist. in. the. Bureau. of. Rural. Sciences,. a..science.policy.bureau.with.the.Australian.government’s.Department.of.Agriculture,.Fisheries.and.Forestry..He.is.an.ecologist.by.training.and.retains.a.keen.interest.in.landscape.ecology..He.currently.manages.the.national-.and.catchment-scale.land.use.mapping.project.for.Australia,.which.includes.work.on.land.management.practices..He. has. been. developing. multicriteria. software. and. has. a. specific. interest. in. arid.lands.and.rangelands.
Clive McAlpineClive. McAlpine. is. a. senior. research. fellow. with. the. Centre. for. Remote. Sensing.and.Spatial.Analysis.with.the.School.of.Geography,.Planning.and.Architecture,.the..University.of.Queensland,.Brisbane..His.research.interests.are.in.landscape.ecology,.biodiversity.conservation,.land.cover.change.modeling,.and.the.climate.impacts.of.land.cover.change.
42963_C000.indd 26 11/6/07 11:18:26 AM
Contributors xxvii
Andrew C. MillingtonAndrew.C..Millington.is.a.professor.in.the.Department.of.Geography.at.Texas.A&M.University..He.has.previously.worked.in.three.other.geography.departments:.he.was.formerly.professor.and.departmental.chair.at.the.University.of.Leicester,.England;.reader.in.geography.at.the.University.of.Reading,.and.lecturer.at.Fourah.Bay.College.(part.of.the.University.of.Sierra.Leone)..His.research.focuses.on.the.impacts.of.human.and.institutional.agents.on.spatiotemporal.patterns.of.land.use.and.land.cover.change.and.the.impacts.of.land.use.and.land.cover.change.on.biological.phenomena..He.uses.hybrid.methodologies.from.GIScience,.environmental.science,.ecology,.and.social.science. to.research. these.phenomena..His.current.research. includes.analyzing.the.effects.of.policy.initiatives.on.land.use.and.land.cover.change.in.Argentina,.Bolivia,.and.Peru,.and.he.is.initiating.work.on.land.use.and.land.cover.change.in.forest.lands.in.Texas..He.received.his.BSc.in.geography.and.geology.from.Hull.University,.his.MA.in.geography.from.the.University.of.Colorado.at.Boulder,.and.his.DPhil.in.geog-raphy.from.Sussex.University.
Daniel MüllerDaniel. Müller. is. an. agricultural. economist. now. working. in. the. Department. of..Economic. and. Technological. Change,. Center. for. Development. Research. at. the..University.of.Bonn..He.contributed.to.this.book.during.a.postdoctoral.appointment.at.Humboldt.University..His.research.interests.are.in.development.and.natural.resource.economics,.patterns,.and.process. in. land.cover.and. land.use.change,.econometric.analysis,.and.spatial.data.analysis.
Karen C. SetoKaren.C..Seto.is.an.assistant.professor.in.the.Department.of.Geological.and.Environ-mental.Sciences.and.a.fellow.at.the.Woods.Institute.for.the.Environment.at.Stanford.University.. Her. research. focuses. on. monitoring. and. forecasting. land. use. change,.especially.urban.growth.in.Asia..Her.current.research.efforts.include.analyzing.the.effects. of. policy. reforms. on. urban. growth. in. China,. India,. and. Vietnam.. She. is.the.Remote.Sensing.Thematic.Leader.for.the.World.Conservation.Union’s.(IUCN)..Commission. on. Ecosystem. Management. and. is. a. recipient. of. the. NASA. New.Investigator. Program. in. Earth. Science. Award. and. an. NSF. Faculty. Early. Career.Development.(CAREER).Award..She.received.her.BA.in.political.science.from.the.University.of.California.at.Santa.Barbara,.her.MA.in.international.relations,.resource.and.environmental.management.from.Boston.University,.and.her.PhD.in.geography.from.Boston.University.
42963_C000.indd 27 11/6/07 11:18:27 AM
42963_C000.indd 28 11/6/07 11:18:27 AM
Part I
Theory and Methodology
42963_S001.indd 1 8/17/07 2:29:35 PM
42963_S001.indd 2 8/17/07 2:29:35 PM
�
1 Basic and Applied Land Use Science
Richard J. Aspinall
Contents
1.1 Introduction.....................................................................................................31.1.1 TheoreticalFoundations......................................................................4
1.2 BasicScience...................................................................................................51.2.1 DynamicsofChangeinSpaceandTime.............................................51.2.2 IntegrationandFeedbacksbetweenLandscape,Climate,
Socioeconomic,andEcologicalSystems.............................................71.2.3 Resilience,Vulnerability,andAdaptabilityofLandSystemsas
CoupledNaturalandHumanSystems.................................................81.2.4 ScaleIssues..........................................................................................81.2.5 Uncertainty..........................................................................................8
1.3 AppliedScience..............................................................................................91.3.1 AddressingEvolvingPublicandPrivateLandManagement
IssuesandDecisions.......................................................................... 101.3.2 InterpretationandCommunicationofScientificKnowledgefor
AdaptiveManagementofChangeinLandUseSystems................... 111.3.3 HumanandEnvironmentalResponsestoChange............................. 11
References................................................................................................................ 11
1.1 IntroduCtIon
Landusesciencecanbedefinedasaninclusive,interdisciplinarysubjectthatfocusesonmaterialrelatedtothenatureoflanduseandlandcover,theirchangesoverspaceand time, and the social, economic, cultural, political, decision-making, environ-mental,andecologicalprocessesthatproducethesepatternsandchanges.1Avarietyoftheories,methodologies,andtechnologiesunderpinresearchonlandusescience,and,consequently,anumberofbasicandappliedsciencethemesthatarecharac-teristic of land use research can be identified. These reflect the interdisciplinaryand integrated analysis required to comprehend landuse, aswell as the role andimportanceoflanduse,landusechange,andlandmanagementandpolicy,andtheimportanceoflanduseforsustainability.2LanduseisalsoconsideredacentralpartofthefunctioningoftheEarthsystem3aswellasreflectinghumaninteractionswiththeenvironmentatscalesfromlocaltoglobal.
42963_C001.indd 3 10/29/07 10:47:46 AM
� Land Use Change: Science, Policy and Management
Basicsciencequestionsinlandusescienceincludethosethatfocuson(a)dynamicsofchangeinspaceandtime;(b)integrationandfeedbacksbetweenlandscape,climate,socioeconomic,andecologicalsystems(c)resilience,vulnerability,andadaptabilityof coupled natural and human systems (d) scale issues and (e) accuracy. Appliedscienceaddressespolicyandmanagementquestionsinlandusescienceincluding(a)addressingevolvingpublicandprivate landmanagement issuesanddecisions;(b)interpretationandcommunicationofscientificknowledgeforadaptivemanage-mentofchangeinlandusesystems;and(c)humanandenvironmentalresponsestochange.Theappliedissuesalsoshouldbesetagainstaneedforexplicitmanagementofuncertainties.Thiswillincludedefinitionofthelimitsofapplicabilityofchangeprojectionsandotheranalyses,particularlyastranslatedintodecisionsupportandparticipatoryapproaches.Theneedandroleforspatiallyintegrateddynamicmodelsofcouplednaturalandhumansystemsinthecontextsofstudyandmanagementoflandusechangeunderpinthisdiscussion.
1.1.1 TheoreTical FoundaTions
Therehasbeensomediscussionofthepotentialandneedforanintegrated,orover-arching,theoryforlandchange.Lambinandcolleagues4notethreerequirementsforanoverarchingtheory:(a)toengagethebehaviorofpeopleandsocietyandreciprocalinteractionwith landuse, (b) tobemultilevelwith respect tobothpeopleand theenvironment,and(c)tobemultitemporalinordertoincludeboththecurrentandpastcontextsinwhichland,people,andenvironmentinteract.Integratedstudyoflanduseandlandcoverchangestypicallyisinterdisciplinaryormultidisciplinaryinapproachand thus involves theories from multiple participating disciplines.4 The practicalneedsofinterdisciplinaryresearchhaveledempiricalcasestudiestouseavarietyofmechanismsforencouragingdialoguebetweendisciplines,includingarangeofinte-gratingframeworks,mostbasedonsomeformofsystemsrepresentation.5,6Empiricalstudiesalsorecognizesomequalitiesoflandsystemsthatarecommonacrosscasestudies,andthesesuggestcharacteristicsthatatheoryoflandchangeneedstobeabletoincorporate:
(a) Complexcauses,processes,andimpactsofchange7
(b) Differencesandinter-relationshipsbetweenlanduseandlandcover8,9,10
(c) Interactionofsocioeconomicandbiophysicalprocesses11,12,13,14
(d) Multiplespatialandtemporalscalesatwhichprocessesoperate11,15,16
(e) Interactionacrossmultipleorganizationallevels17
(f) Feedbacksandconnectionsinbothsocialandgeographicalspaces (g) Multiplelinksbetweenpeopleandland7
(h) Influenceofsocial,historical,andgeographicalcontextonlandusechange18
(i) Importance of individual, social, demographic, economic, political, andculturalfactorsindecisionmaking19,20
(j) Combineduseofqualitativeandquantitativedataandmethods21,22
42963_C001.indd 4 10/29/07 10:47:47 AM
Basic and Applied Land Use Science �
Reviewofmultiplecasestudiesinmeta-analyseshasalsoprovidedinsightintolandchange,providinggeneralizationaboutfactorsthatleadtochange.Examining152publishedcasestudiesoftropicaldeforestation19and132casestudiesofdeserti-fication20 Geist and Lambin identified a relatively small set of underlying causescommon to the landusechangesobserved indifferent regionsandplaces.Theseunderlyingcausesaredescribedasproximate,havingapparent immediate impactonchange,andultimate,whichrepresentfundamentalcausesofchange.23Therearefivebroadgroupsofunderlyingfactorscommontobothsetsofcasestudies:demo-graphic,economic,technological,policyandinstitutional,andcultural;desertifica-tion also included climatic factors. Proximate causes of change common to bothtropicaldeforestationanddesertificationincludedinfrastructureextension,agricul-tural activities and expansion, and wood extraction; increased aridity also was aproximate cause for desertification. A meta-analysis of 91 published case studiesofagriculturallandintensificationinthetropics,24intendedasacompaniontothemeta-analysesoftropicaldeforestationanddesertification,usedthesamefactorsasGeistandLambin’sstudiesandrecordedaverydetailedandvariedlistofprocessesassociatedwithagriculturalintensification.Themainfactorsidentifiedweredemo-graphic,market,andinstitutional,particularlypropertyregimes.Themostcommonprocesses of agricultural intensification in the tropics included adoption of newcrops,plantingoftrees,anddevelopmentofhorticulture.24
Theseconcerns fordevelopment anduseof theory, and for improvingunder-standing of social and natural processes, as well as their interaction, in study oflanduse,provideaguideforcasestudiesandattemptsatintegrationandsynthesisacrosscasestudies.IntheremainderofthischapterIdiscusssomebasicandappliedscienceissuesthatmayhelpnotonlytheprocessofstudyinglanduseandchange,butalsothecommunicationandinvolvementofawidevarietyofinterestedparties,includingdecisionmakersandlandmanagers,inboththeconductofresearchanditsimplicationsforlandmanagementandpolicy.
1.2 BasIC sCIenCe
1.2.1 dynamics oF change in space and Time
Landuseandlandcoverchangesareinherentlyspatialanddynamic.Themagnitudeandimpactofchangesinlanduseandlandcoveraresuchthatlandusechangeisrecognizedasachangethatisglobalinextentandimpact.25,26,27Landusedynamicsisalsorecognizedasoneofthegrandchallengesinenvironmentalscience.28
Avarietyoflanduseandlandcoverchangeprojectshavemeasuredandmonitoredchangeinlanduseandlandcover,particularlyforperiodsoverthepast30years,usingsatelliteimagery,29,30,31butalsoforperiodsuptothepastseveralhundredyears.32,33Theseindividualprojectshavenotonlyidentifiedcommonlanduseandlandcovertransitions but also raised awareness and interest in the processes that producechange.Forexample,theInternationalHumanDimensionsProgrammeonGlobalEnvironmental Change/International Geosphere-Biosphere Programme Land-UseandLand-CoverChange(IHDP/IGBPLUCC)program,6aninternationalprogram
42963_C001.indd 5 10/29/07 10:47:47 AM
� Land Use Change: Science, Policy and Management
running from 1995 to 2005, coordinated a broad network of local, regional, andcontinentalscaleprojectsexamininglandcoverandlandusechange.14Thetaskofestablishingcauseandeffect inrelation to landusechangedynamicsfacesmanyofthechallengesofotherempiricalfield-basedsciencesthatrelyonobservationofEarthsurfacephenomena,suchasgeology.34
Studyoflandusedynamicsisfurthercomplicatedbyavarietyoftime-relatedfactors.Landusesystems,aswellas theunderlyingfactorsandprocesses, them-selvesmaychangethroughtime.Thisproducesavarietyofpathdependence35andlegacyeffects,36resultinginlandusepatternsandsystemsthatmayreflectavarietyofnot only contemporaryprocesses, but alsoprocesses and responses tohistoricdrivers of change. Additionally, land cover change involves both conversion andmodificationofcover37andmaybegradualorepisodic.7Typicallychangethroughtime,especiallyforspatialmodels,isstudiedquantitativelyforaplacewithaseriesofsnapshotsof landcover (sometimes treatedasequivalent toor interchangeablewithlanduse).Thismaynotonlyunderestimatetheextentofchangebutalsofailto capture whether changes are gradual or episodic. Measurement, analysis, andmodeling of land use systems need frameworks, tools, and methods that help toseparatemultipleinfluencesandasynchronouscausesofchangeinlandusesystemdynamics,aswellasprovideanimprovedabilitytodetectagreaterrangeoftypesandratesofchange.Crews’research38(Chapter6)usespanelanalysistofocusonthelongitudinal(time)sequenceofchange.
The use of snapshots of land cover to study change and develop quantitativemodels may also ignore the rich source of insight and methods from a study ofhistoricaldevelopmentinthatplace.7,36Forexample,environmentalhistory,narrative,andstorytellingallofferinsightintochangethroughtime.39Combiningquantitativeapproacheswithenvironmentalhistorywouldofferadvantagesforstudyofchangeinspatialpatternandtemporaldynamicsoflandsystems,andforunderstandingtherolesofmultiplecausesandprocessesofchange.
Arangeofotherissuesneedstobeaddressedtosupportresearchondynamicsof land change, especially if the goals are improved understanding of processesandchangesproducedandbettermodelingofplace-basedchangeinlandsystems.Specifically,thereisaneedfornewexperimentalandobservationaldesignsaswellas researchprotocols that supportquantitativeandqualitativeanalysisofchange.Thiswillenablecloseranddirectcomparisonofresultsfromdifferentcasestudies,leadingtoimprovedmeta-analysesandsynthesisbetweencasestudies.Thereisalsoaneedforresearchintohowspatialandtemporaldynamicsmightbestberepresentedtosupportquantitativeandqualitativeanalysis.Althoughgeographicalinformationsystems(GIS)andremotesensingprovidemechanismsforrecording,representation,measurement,andanalysisofthespatialstructureofpropertiesofthelandsurface,GISisstillpoorlydevelopedforrepresentationandanalysisofspatiotemporaldata.ImprovedGISdatastructuresforlandcoverandusedatathatexplicitlyincorporatetimeareneeded.Thiswillbeabenefitnotonlyfornewformsofanalysis,butitalsowillprovideanempiricalfoundationforthebroadrangeofnoveltools, includingagent-basedmodeling40andcellularautomata41,42thathaveaddedtoourabilitytorepresentandmodelsocialandspatialprocessesthatarecentraltoamorefullunder-standingoflandsystemdynamics.Regular,repeatedremotelysensedmeasurements
42963_C001.indd 6 10/29/07 10:47:48 AM
Basic and Applied Land Use Science �
arealsoneededtohelpdevelopthespatialandtemporalhistoryoflanduseandlandcoverasaprecursortoimproveddynamicspatialmodelsofchange.43,44
1.2.2 inTegraTion and Feedbacks beTween landscape, climaTe, socioeconomic, and ecological sysTems
Landsystemsare increasinglybeing treatedasexemplarsof couplednatural andhuman systems (frequently socioecological systems,45 which are considered heretobeexactlysynonymous).Thissystems-basedviewof landrecognizes that landuseandlandcoversystemsaredefinedattheinterfaceofnatural/biophysicalandhumansystems.Indeed,theIHDP/IGBPGlobalLandProjectestablishesaframe-work for combined study of land change, ecosystem services, and sustainabilitywithinanEarthsystemsciencecontextthroughexploringlanduseandlandcoverasrepresentingacontinuumbetweensocioeconomic(use)andbiophysical(cover)systems.3 This attempt at a coupled and holistic view of land systems presents arichframeworkforunderstandingmultidirectionalimpactsandfeedbacksbetweenelementsoflandsystemsandthemanyunderlyingfactorsandprocessesthatoperatetoshapeandchangeland.Forexample,notonlydolandcoverandlandusereflectenvironmentalsystemsandtheopportunitiesandconstraintstheyprovide,butlandcover and land use also have a strong and direct influence on climate and otherenvironmentalchangeandmustbeincludedinnewmodelsoftheEarth’sclimatesystem.27,46Similarly,changesinlandsystemsreflectsocioeconomicprocessesthatoperateataverywiderangeofspatialandtemporalscales,includingglobalization,tradeandmarkets,policyandlandmanagementdecisionsatthenational,regional,local,orhousehold/individuallevel.
Therehavebeenseveralframeworksforstudyoflandsystemsasexemplarsofcouplednaturalandhumansystems.Forexample,Machlisandcolleagues5describeahumanecosystemmodelthatattemptstolinksocialsystemsandnaturalsystemsthroughafocusonsocialsystemslinkedtonatural,cultural,andsocioeconomicsascriticalresources.Thisexplicitlyincludessocialstructures,butdoesnotintegrateanecologicalsystemsview,anddoesnotaddressprocess-levelunderstanding.Despitethese attempts and the claims of certain disciplines to provide an overarchingintegrativecontent, there remainsaneedforbetterconceptualmodelsof integra-tion and feedbacks between landscapes, climate, ecological (environmental), andsocioeconomicsystems.Steinitzandcolleagues47providearesearchframeworkforlandscapedesignbasedondifferent typesofmodeland thequestions themodelsanswer.Again,thishasmanyelementsthatsuggestintegrationbutwithafocusonmodelsandcorequestionsthatmaynotbefullyinclusive.
Ingeneral,qualitiesofsuccessfulframeworksmightinclude(a)afocusoninter-disciplinaryortransdisciplinary48approaches,(b)atheoreticalgroundinginarangeof sciences to allow for participation from multiple disciplines, (c)being systemsbased,toaddressprocessesandstructures,(d)beingspatiallyandtemporallyexplicittoallowgeographicandhistoricalcontextsandcontingenciestobeincludedinunder-standingandanalysisofdynamics,and(e)beingexplicitinaddressingscale,includingspatial,temporal,andorganizationalscales,49includingmultiscaleeffects.50
42963_C001.indd 7 10/29/07 10:47:49 AM
� Land Use Change: Science, Policy and Management
1.2.3 resilience, VulnerabiliTy, and adapTabiliTy oF land sysTems as coupled naTural and human sysTems
Inconjunctionwithagrowingunderstandingoflandsystemsasexemplarsofcouplednatural and human systems, there is a move to interpret land systems within thecontextofsustainabilityscience.3,51Asanintegralpartofthis,researchaddressesthevulnerability(fromthehazardsliterature),resilience(fromtheecologicalliterature),andadaptabilityoflanduseandlandsystems.Sustainability,vulnerability,resilience,andadaptabilityallhavestrongtemporalelementsthatrefernotonlytotheabilityoflandsystemstorespondtochangebutalsototheabilityoflandsystemstoinflu-encetheresponsesofsocioeconomicandenvironmentalsystemstochange.Whatwouldbethecharacteristicsofaresilientlandsystemoravulnerablelandsystem?Canweassesswhetheralandsystemorlanduseisresilientorvulnerabletochangesintheexternaldrivingprocesses(suchasininternationalpricesforcrops)?Doesalandsystemconferresilienceorvulnerability tosocialsystems?Forexample,aretherepatternsof landuseand landcover,andasuiteofassociated landmanage-mentmechanisms, thatmakea community resilient/vulnerable todrought,flood,or other environmental or socioeconomic change? What are the time scales forresilience/vulnerability thataremostappropriateforsustainability?Landsystemsclearlyofferthepotentialtoexploreeconomic,social,institutional,environmental,andecosystemresilience,vulnerability,andsustainabilitywithbenefitstosociety,environment,andland.
1.2.4 scale issues
Scale issues are always central to discussions of land system change,52 but theirresolutionislinkedtothemanydifferentmeaningsof“scale”intheinterdisciplinaryandmultidisciplinarycommunitiesworkingonlandusescience.Scalereferstoallof(a)spatialextentandresolution,(b)time,includingthedurationofastudyandtheresolutionofdatasnapshots,(c)taxonomiclevel,forexample,thelevelofinterestinlanduseorlandcoverasinthedetailinalanduse/coverkeyorintheinstitutional/geopolitical hierarchy of global-international-national-regional-local, or (d) otheranalyticaldimensionsusedforstudy.Giventheinterdisciplinarynatureofmuchoftheresearchandliteratureonlanduse,someconsistencyandexplicitreportingoftheuseofterminologyrelatedtoscalewouldbevaluable.52Gibsonandcolleagues49provideadetaileddiscussionofscaleissues.Scaleissuesinlandusechangemayalso benefit from the space-time approaches developed in landscape ecology.53,54Asystematicanalysisandreviewofscalesatwhichprocessesandresponsesoperateinlandusesystemsmaybevaluableinguidinganalysisandmodelingofchange.
1.2.5 uncerTainTy
Uncertaintyunderliesmanyaspectsoflandusescienceandisassociatednotonlywithmeasurement,analysis,andmodeling,butalsowithdecisionmakingandtheconsequencesofmodelsforprojectingcurrentandfutureconditionsandstatesoflandusesystems.Theaccuracyofmodelsthatpredictchangeisaconstantconcernof land use scientists,55 but there are also concerns related to the accuracy and
42963_C001.indd 8 10/29/07 10:47:50 AM
Basic and Applied Land Use Science �
uncertaintiesofsystemsmodelsasawholecomparedwiththeaccuracyanduncer-taintiesofcomponent(sub-)modelsoflandusesystems.Errorpropagationthroughcoupledmodels of landuse systems as coupledhumanandnatural systems is ofparticularconcernandinterest.
Although statistical methods for management of uncertainty associated withmeasurement,models,andpredictionsarerelativelywelldeveloped,othertypesofuncertaintyrequireattention.Conroy56describesfourtypesofuncertainty:
1.Statisticaluncertainty:reflectinginabilitytomeasure 2.Structural uncertainty: reflecting inability to describe and model system
dynamics 3.Partialcontrollability:reflectinginabilitytocontroldecisions 4. Inherentuncertainty:presentinallsystems(stochasticprocesses)
All of these aspects of uncertainty need research for effective use of models topredict changeand tousemodels fordecision support related topolicy and landmanagement.Betterintegrationandunderstandingofscientificuncertaintyrelatedtodecisionmakingunderconditionsofuncertaintyshouldhelptoenhancecommu-nicationbetweenscientists,policymakers,andlandmanagers.
Policy-relevantmodelsraisesomespecialconcernsbeyondthoseofuncertainty.Lee57andKingandKraemer58havediscussedrequirementsformodels thatare tobeused inpolicycontexts.Lee57 identifiesfivequalitiesofmodels themselves: (1)transparency,(2)robustness,(3)reasonabledataneeds,(4)appropriatespatiotemporalresolution,and(5)inclusionofenoughkeypolicyvariablestoallowforlikelyandsig-nificantpolicyquestionstobeexplored,whileKingandKraemer58concentrateontheroleservedbymodels:(1)toclarifytheissuesinthedebate,(2)toenforceadisciplineofanalysisanddiscourseamongstakeholders,and(3)toprovideaninterestingformofadvice,mainlyintheformofwhatnottodo.Agarwalandcolleagues59usethesecriteria, criteria from Veldkamp and Fresco60 describing space, time, biophysicalfactors,andhumanfactors,aswellasscaleandcomplexityofmodelsascross-cuttingcriteriatoreviewandcomparelandusechangemodels.Theytakeapragmaticviewthatlandusemodels,althoughnotideal,willbegoodenoughtobetakenseriouslyinthepolicyprocess.Theyconcludebyidentifyinganeedforgreatercollaborationbetweenlandusemodelersandpolicymakers.Suchcollaborationwillneedtoiden-tifythekeyvariablesandsectorsofinterest,scalesofanalysis,andchangescenariosanticipated.Thiswillalsorequiretranslationbetweenthedemographic,economic,technology,institutionalandpolicy,andculturalfactors19,20,24usedforanalysis,under-standing,andmodelingoflandusechangeandthespecificneedsofpolicymakers.59Collaboration in this form requires closer focuson applied science from landusescientistsandconsiderationofmechanismsusedforcollaborationandparticipationbetweenscientistswithpolicymakersandlandmanagers.
1.� applIed sCIenCe
Landusescienceisalsoanappliedsciencewithclearlinkstopolicyandpracticethroughdecisionmakingandotherhumaninterventionandactiononlanduseand
42963_C001.indd 9 10/29/07 10:47:50 AM
10 Land Use Change: Science, Policy and Management
landcover.Threeissuesseemparticularlyimportantinrelationtoemergingtrendsandneedsforlandusescienceasanappliedscience:
1.Addressingevolvingpublicandprivatelandmanagementissuesanddecisions 2. Interpretation and communication of scientific knowledge for adaptive
managementofchangeinlandusesystems 3.Understandinghumanandenvironmentalresponsestochange
1.3.1 addressing eVolVing public and priVaTe land managemenT issues and decisions
The multiscale nature and consequences of land use change present a compel-lingcase for translating science intopolicyandpractice.Since landuse isat theinterfaceofhumanandnaturalsystems,improvedunderstandingofthesocialandbiophysicalprocessesthatproducechangeinlandusesystemscanplayausefulroleinbothpolicyandpracticeforbothprivateandpubliclandmanagementissuesanddecisions.Improvedunderstandingoflandusechangecanprovideinputtoevidence-basedpolicyandbeusedtodevelopalternativescenariosforlanduseresponsetodifferent policies.61,62,63 This may help to inform wide participation of interestedgroupsandindividualsindebateanddiscussionandalsohelptoleadtoimproveddecisionmaking.Thismaybethroughimprovedandinformedconsensusbyhelp-ingtoexploreimpactsandconsequencesofparticularpolicies(andalternatives)orotheractions.Landusechangealsoreflectslinksthatoccuracrossorganizationalscalesand thus is important in linkingbroadernational, international,andglobaltrendsandconditionswithconsequentunderstandingandconcern for livelihoodsandsustainabilityofcommunitiesatmorehumanlocalscales.
In this context, case studies of land use change can be valuable in two ways.First,theycanprovidegenericunderstandingthatcanbeusedasevidenceinsupportofdiscussion toproduceevidence-basedpolicy.Second, theycanprovide specificscenarios and impacts for an area of interest to help focus decision making in adeliberateandlocallyrelevantmanner.Thusstudiesoflandusechangecanbeusednotonlytoelucidategeneralissuesandprinciplesconcerninglandusechange,butalso to provide insight into applied issues associated with land use and potentialmeaningofchangeforcommunitiesandplaces.Thelatterwouldbeofvalueinestab-lishingimprovedunderstandingofsocialfeedbacksandindevelopingparticipatoryprocessesandconsultationindecisionmaking.Theseopportunitiesandusessuggestthatthereareanumberofimportantconcernsthatshouldbemadeexplicitincasestudiesoflandusechange.Forexample,arecasestudiesconcernedwithmanagementand/orpolicy?Arestakeholders involved in the researchand inwhatways?Whatarethereactionstoresultsandprocessoftheresearch,especiallyfromavarietyofdifferentgroups?Whatlessonsarelearnedfromacasestudyaboutthepoliciesandmanagementpracticesthatinfluencelandusechangeinthestudyareaandsystems?Betterunderstandingof theseaspectsof landusechangecasestudieswillhelp toimprovelinksbetweenscienceandpractice37andalsoplacelandusechangeevidenceinbothascientificanddecisionandmanagementcontext,gettinglanduse/systemssciencetranslatedfromsciencetomanagementandpolicy,andviceversa.
42963_C001.indd 10 10/29/07 10:47:51 AM
Basic and Applied Land Use Science 11
1.3.2 inTerpreTaTion and communicaTion oF scienTiFic knowledge For adapTiVe managemenT oF change in land use sysTems
Interpretationandcommunicationofscientificknowledgeareofincreasingimpor-tanceacrossallofscience.Landuse–relatedresearch,notablyforagricultureandforestry, has a longhistoryofdirect and strong application to landmanagement;examples may be found in agricultural extension services or forestry worldwide.Therearealsoexamplesoflanduseinformationdirectlyguidingpolicy(e.g.,LandUtilisation Survey in Great Britain64). Communication of case study content andinformation on land use change is no less important today, not only for input topolicyandlandmanagementdecisions,butalsoforunderstandinglocalhumanpro-cessesthatproducemanyaspectsofcontemporarychangeinlanduseandlandcover(asreflectedinthegenericunderlyingcausespreviouslydiscussed).Thisraisesthequestionof theextent towhichcase studies involve landmanagersandpolicyorotherdecisionmakersaspartoftheresearchteam(transdisciplinaryinthesenseofTressandcolleagues48).Suchatwo-waycollaborationwouldbebeneficialbothtoscientistsandpractitioners.65,66
1.3.3 human and enVironmenTal responses To change
Applicationoflandusesciencetounderstandingpracticalconsequencesofchangeforhumanandenvironmentalsystemsandtheircomponentsubsystemspotentiallyprovideslinkstoawidevarietyofareasofconcerninenvironmentalmanagementandprovisionofecosystemservices.67Improvedunderstandingandpredictiveandscenario-basedtoolsalsohaveapplicationinlanduseplanningandenvironmentalmanagementandcare.68Keyappliedissuesincludethemannerinwhichdifferentinstitutions,society,andindividualsrespondtochange;howenvironmentalsystemsrespond; space and time scales of response and consequences; development ofstrategiestoadapttoandmanagechangeanditsimpacts;andlong-andshort-termconsequencesofchangeanddecisions.
Theseapplied issues shouldbesetagainstaneed forexplicitmanagementofuncertainties, which is a recurrent theme in scientific communication. Increasedawarenessofuncertaintyshouldincludedefinitionofthelimitsofapplicabilityofprojectionsoflandusechangeandotheranalyses,includingscenarios,particularlyastranslatedintodecisionsupportandparticipatoryapproaches.Boundaryorgani-zations65 and transdisciplinary48 approachescanhelp this explicitmanagementofuncertaintysinceawidercommunityisengagedinmodelingandanalysis,whichboth helps the scientists and practitioners through improved communication andgreaterunderstandingofbothdecision-makingneedsandscientificprocesses.
referenCes
1. Aspinall,R.J.Editorial.Journal of Land Use Science1(1),1–4,2006. 2. Raquez,P.,andLambin,E.F.Conditionsforasustainablelanduse:Casestudyevidence.
Journal of Land Use Science1(2–4),109–125,2006. 3. GLP.Science Plan and Implementation Strategy. IGBPreportNo.53,IHDPReport
No.19.Stockholm,IGBPSecretariat.64pp,2005.
42963_C001.indd 11 10/29/07 10:47:51 AM
12 Land Use Change: Science, Policy and Management
4. Lambin,E.F.,Geist,H. J., andRindfuss,R.R. Introduction: Local processeswithglobal impacts. In:Lambin,E.F. andGeist,H. J.,Eds.Land-Use and Land-Cover Change: Local Processes and Global Impacts.IGBPSeries.Springer-Verlag,Berlin,1–8,2006.
5. Machlis,G.E.,Force,J.E.,andBurch,W.R.Thehumanecosystem,PartI:Thehumanecosystemasanorganizingconceptinecosystemmanagement.Society and Natural Resources10,347–367,1997.
6. Turner,B.L.etal.,eds.Land Use and Land Cover Change: Science/Research Plan.IGBPReportNo.35,HDPReportNo.7.Stockholm,Sweden,InternationalGeosphere-BiosphereProgramme,132pp,1995.
7. Lambin,E.F.,Geist,H.J.,andLepers,E.Dynamicsofland-useandland-coverchangeintropicalregions.Annual Review of Environment and Resources28,205–241,2003.
8. Brown,D.G.,Pijanowski,B.C.,andDuh,J.D.Modelingtherelationshipsbetweenland use and land cover on private lands in the upper midwest, USA. Journal of Environmental Management59(4),247–263,2000.
9. Brown,D.G.,andDuh,J.D.Spatialsimulationfortranslatingfromlandusetolandcover.International Journal of Geographical Information Science 18(1),35–60,2004.
10. Xu,J.C.etal.Land-useandland-coverchangeandfarmervulnerabilityinXishuang-bannaprefectureinsouthwesternChina.Environmental Management36(3),404–413,2005.
11. Walsh,S.J.etal.Scale-dependentrelationshipsbetweenpopulationandenvironmentinnortheasternThailand.Photogrammetric Engineering and Remote Sensing 65(1),97–105,1999.
12. Veldkamp, A., and Verburg, P. H. Modelling land use change and environmentalimpact.Journal of Environmental Management 72(1–2),1–3,2004.
13. Urama, K. C. Land-use intensification and environmental degradation: Empiricalevidencefromirrigatedandrain-fedfarmsinsoutheasternNigeria. Journal of Environ-mental Management75(3),199–217,2005.
14. Lambin, E. F., and Geist, H. J., eds. Land-Use and Land-Cover Change: Local Processes and Global Impacts.IGBPSeries.Springer-Verlag,Berlin,222pp,2006.
15. Riebsame,W.E.,Gosnell,H.,andTheobald,D.M.Landuseandlandscapechangein the Colorado Mountains. 1. Theory, scale, and pattern. Mountain Research and Development16(4),395–405,1996.
16. Veldkamp, A., and Fresco, L. O. CLUE-CR: An integrated multi-scale model tosimulate land use change scenarios in Costa Rica. Ecological Modelling 91(1–3),231–248,1996.
17. Bousquet,F., andLePage,C.Multi-agent simulations and ecosystemmanagement:Areview.Ecological Modelling176(3–4),313–332,2004.
18. Adger,W.N.Evolutionofeconomyandenvironment:Anapplication to landuse inlowlandVietnam.Ecological Economics 31(3),365–379,1999.
19. Geist,H.,andLambin,E.F.Proximatecausesandunderlyingdrivingforcesoftropicaldeforestation.Bioscience52(2),143–150,2002.
20. Geist,H.J.,andLambin,E.F.Dynamiccausalpatternsofdesertification.Bioscience54(9),817–829,2004.
21. Hill, M. J. et al. Multi-criteria decision analysis in spatial decision support: TheASSESSanalytichierarchyprocessandtheroleofquantitativemethodsandspatiallyexplicitanalysis.Environmental Modelling and Software20(7),955–976,2005.
22. Madsen,L.M.,andAdriansen,H.K.Understandingtheuseofruralspace:Theneedformulti-methods.Journal of Rural Studies20(4),485–497,2004.
23. Geist,H.J.,andLambin,E.F.What Drives Tropical Deforestation? A Meta-Analysis of Proximate and Underlying Causes of Deforestation Based on Subnational Case Study Evidence. LUCC Report Series No. 4., Land-Use and Land-Cover Change
42963_C001.indd 12 10/29/07 10:47:52 AM
Basic and Applied Land Use Science 1�
(LUCC)ProjectIV.InternationalHumanDimensionsProgrammeonGlobalEnviron-mentalChange(IHDP)V.InternationalGeosphere-BiosphereProgramme(IGBP),116pp,2001.
24. Keys,E.,andMcConnell,W.J.Globalchangeandtheintensificationofagricultureinthetropics.Global Environmental Change15,320–337,2005.
25. Vitousek, P. M. Beyond global warming: Ecology and global change. Ecology 75,1861–1876,1994.
26. Vitousek,P.M.etal.HumandominationofEarth’secosystems.Science277(15July),494–499,1997.
27. Foley,J.A.etal.Globalconsequencesoflanduse.Science 309,570–574,2005. 28. National Research Council (NRC). Grand Challenges in Environmental Sciences.
ReportfromtheCommitteeonGrandChallengesinEnvironmentalSciences.NationalAcademyPress,Washington,DC.96pp,2001.
29. Skole, D., and Tucker, C. Tropical deforestation and habitat fragmentation in theAmazon:Satellitedatafrom1978to1988.Science260(5116),1905–1910,1993.
30. Brown,D.G.LanduseandforestcoveronprivateparcelsintheuppermidwestUSA,1970to1990.Landscape Ecology18(8),777–790,2003.
31. Ferraz,S.F.D.etal.LandscapedynamicsofAmazoniandeforestationbetween1984and2002incentralRondonia,Brazil:Assessmentandfuturescenarios.Forest Ecology and Management 204(1),67–83,2005.
32. Etter,A.,andvanWyngaarden,W.PatternsoflandscapetransformationinColombia,withemphasisintheAndeanregion.Ambio29(7),432–439,2000.
33. Goldewijk,K.K.Estimatinggloballandusechangeoverthepast300years:TheHYDEDatabase.Global Biogeochemical Cycles15(2),417–433,2001.
34. Schumm, S. D. To Interpret the Earth: 10 Ways to Be Wrong. 2nd ed. CambridgeUniversityPress,Cambridge.143pp,2006.
35. Brown,D.G.etal.Pathdependenceandthevalidationofagent-basedspatialmodelsoflanduse.International Journal of Geographical Information Science19(2),153–174,2005.
36. Aspinall,R.J.Modellinglandusechangewithgeneralizedlinearmodels—amulti-modelanalysisofchangebetween1860and2000inGallatinValley,Montana.Journal of Environmental Management72(1–2),91–103,2004.
37. Lesslie, R., Barson, M., and Smith, J. Land use information for integrated naturalresources management—a coordinated national mapping program for Australia.Journal of Land Use Science1(1),45–62,2006.
38. Crews-Meyer, K. A. Characterizing landscape dynamism using paneled-patternmetrics.Photogrammetric Engineering and Remote Sensing 68(10),1031–1040,2002.
39. Wyckoff,W.,andHansen,K.Settlement,livestockgrazingandenvironmentalchangeinsouthwestMontana,1860–1990.Environmental History Review15,45–71,1991.
40. Parker,D.C.etal.Multi-agentsystemsforthesimulationofland-useandland-coverchange:Areview.Annals of the Association of American Geographers93(2),314–337,2003.
41. Batty,M.Agents,cells,andcities:newrepresentationalmodelsforsimulatingmulti-scaleurbandynamics.Environment and Planning A 37(8),1373–1394,2005.
42. Lau,K.H.,andKam,B.H.Acellularautomatamodelforurbanland-usesimulation.Environment and Planning B-Planning and Design32(2),247–263,2005.
43. Turner, D. P., Ollinger, S. V., and Kimball, J. S. Integrating remote sensing andecosystemprocessmodelsforlandscape-toregional-scaleanalysisofthecarboncycle.Bioscience54(6),573–584,2004.
44. Tralli,D.M.etal.Satelliteremotesensingofearthquake,volcano,flood,landslideandcoastal inundationhazards.ISPRS Journal of Photogrammetry and Remote Sensing59(4),185–198,2005.
42963_C001.indd 13 10/29/07 10:47:53 AM
1� Land Use Change: Science, Policy and Management
45. Carpenter, S. et al. From metaphor to measurement: Resilience of what to what?Ecosystems4(8),765–781,2001.
46. Foley,J.A.etal.Incorporatingdynamicvegetationcoverwithinglobalclimatemodels.Ecological Applications 10(6),1620–1632,2000.
47. Steinitz,C.etal.Alternative Futures for Changing Landscapes: The Upper San Pedro River Basin in Arizona and Sonora.Washington,D.C.:IslandPress.202pp,2003.
48. Tress, B. et al., eds. Interdisciplinary and Transdisciplinary Landscape Studies: Potential and Limitations,DELTASeries2.Wageningen,192pp,2003.
49. GibsonC.C.,OstromE.,andAnhT.K.Theconceptofscaleandthehumandimen-sionsofglobalchange:Asurvey.Ecological Economics32,217–239,2000.
50. Redman,C.L.,Grove,J.M.,andKuby,L.H.Integratingsocialscienceintothelong-termecologicalresearch(LTER)network:Socialdimensionsofecologicalchangeandecologicaldimensionsofsocialchange.Ecosystems7(2),161–171,2004.
51. Kates, R. W., and Parris, T. M. Long-term trends and a sustainability transition.Proceedings of the National Academy of Sciences100,8062–8067,2003.
52. Verburg,P.H.etal.Landusechangemodelling:Currentpracticeandresearchpriori-ties.GeoJournal61,309–324,2004.
53. Delcourt, H.R.,Delcourt, P. A., andWebb T. A., III. Dynamic plant ecology: Thespectrum of vegetation change in space and time. Quaternary Science Reviews 1,153–175,1983.
54.Delcourt,H.R.,andDelcourt,P.A.Quaternarylandscapeecology:Relevantscalesinspaceandtime.Landscape Ecology2,23–44,1988.
55. Pontius,R.G.,andSpencer,J.Uncertaintyinextrapolationsofpredictiveland-changemodels.Environment and Planning B-Planning and Design 32(2),211–230,2005.
56. Conroy,M.J.Conservationandlandusedecisionsunderuncertainty:Models,data,andadaptation.In:Hill,M.J.,andAspinall,R.J.,Eds.,Spatial Information for Land Use Management.OPAOverseasPublishingAssociates,Ltd.,Reading,U.K.,145–158,2000.
57. Lee,D.B.,Jr.Requiemforlarge-scalemodels.AIP Journal (May),163–177,1973. 58. King,J.L.,andKraemer,K.L.Models, facts,and thepolicyprocess:Thepolitical
ecology of estimated truth. In: Goodchild, M. F., Parks, B. O., and Steyaert, L. T.,eds.Environmental Modeling with GIS.OxfordUniversityPress,NewYork,353–360,1993.
59. Agarwal,C.etal.A Review and Assessment of Land-Use Change Models: Dynamics of Space, Time and Human Choice.CIPECCollaborativeReportSeriesNo.1.CIPEC,IndianaUniversity.90pp,2002.
60. Veldkamp,A.,andFresco,L.O.CLUE:Aconceptualmodeltostudytheconversionoflanduseanditseffects.Ecological Modelling85,253–270,1996.
61. Steinitz, C. et al. A delicate balance: Conservation and development scenarios forPanama’sCoibaNationalPark.Environment47(5),24–39,2005.
62. Berger,P.A.,andBolte,J.P.Evaluatingtheimpactofpolicyoptionsonagriculturallandscapes:Analternative-futuresapproach.Ecological Applications14(2),342–354,2004.
63. vanDijk,T.ScenariosofCentralEuropeanlandfragmentation.Land Use Policy20(2),149–158,2003.
64. Sheail,J.Scott revisited:Post-waragriculture,planningandtheBritishcountryside.Journal of Rural Studies13(4),387–398,1997.
65. Guston,D.H.Boundaryorganisationsinenvironmentalpolicyandscience:Anintro-duction.Science, Technology and Human Values26(4),399–408,2001.
66. Lemos, M. C., and Morehouse, B. J. The co-production of science and policy inintegratedclimateassessments.Global Environmental Change15,57–68,2005.
42963_C001.indd 14 10/29/07 10:47:53 AM
Basic and Applied Land Use Science 1�
67. Rounsevell, M. D. A. et al. Future scenarios of European agricultural land use II.Projectingchangesincroplandandgrassland.Agriculture Ecosystems and Environ-ment107(2–3),117–135,2005.
68. Haberl,H.,Wackernagel,M.,andWrbka,T.Landuseandsustainabilityindicators.Anintroduction.Land Use Policy 21(3),193–198,2004.
42963_C001.indd 15 10/29/07 10:47:54 AM
42963_C001.indd 16 10/29/07 10:47:54 AM
17
2 Developing Spatially Dependent Procedures and Models for Multicriteria Decision AnalysisPlace, Time, and Decision Making Related to Land Use Change
Michael J. Hill
Contents
2.1 Introduction................................................................................................... 172.2 Concept......................................................................................................... 192.3 TransformationIssues................................................................................... 192.4 TransformationDomainsandMethods........................................................ 212.5 ExampleLandscapeContext—AustralianRangelands................................23
2.5.1 SpatialPatternsandRelationships.....................................................252.5.2 TemporalPatternsandInfluences......................................................262.5.3 DataandInformation:ScaleofRepresentation.................................292.5.4 SomeSpatiotemporalInputstoaRangelandMCA...........................30
2.6 AFrameworkforaMulticomponentAnalysiswithMCA........................... 322.7 Conclusions................................................................................................... 332.8 Acknowledgments......................................................................................... 37References................................................................................................................ 37
2.1 IntRoDUCtIon
Land use change occurs within a space-time domain. Frameworks for assessingappropriatelanduseandprioritiesforchangemustcapturethecomplexity,reduce
42963_C002.indd 17 11/6/07 7:17:13 AM
18 Land Use Change
dimensionality,summarizeahierarchyofmaineffects,transfersignalsandpatterns,andtransforminformationintothelanguageofthepoliticalandeconomicdomains,1yetretainthekeydynamics,interactions,andsubtleties.Spatialinteraction,temporalcycles,responsesandtrends,andchangesinspatialpatternsthroughtimeareimpor-tant sources of information for condition, planning, and predictive assessments.Spatiallyappliedmulticriteriaanalysis2enablesdiversebiophysical,economic,andsocialvariablestobemappedintoastandardizedrankingarray;usedasindividualindicators;combinedtodevelopcompositeindexesbasedonobjectiveandsubjec-tivereasoning;andusedtocontrastandcomparehazards,risks,suitability,andnewlandscapecompositions.3,4,5,6Themulticriteriaframeworkallowsthecombinationofmulti-andinterdisciplinarity.7Thesystemdefinitiondependsuponthepurposeoftheconstruct,scaleofanalysis,andsetofdimensions,objectives,andcriteria.7
Whenmappingbothquantitativeandordinaldataintofactorlayers,retentionof,andaccessto,rationaleandreasoningforinclusionandweightingorcontributiontocompositesisimportantformaintenanceofthelinkbetweentheoutcomeoftheanalysisandtherealorapproximatedatausedasinput.Thisparticularlyappliestospatialandtemporalinformation.Hereitisimportanttoknowwhatthemeaningofaspatialortemporalmetricmightbewhenitisincludedamongotherdataindevel-opmentofanassessmenttoaiddecisionmakers.Themeaninghastwocomponents:(1)thefirstrelatestothedirectdescriptionofthemetricsuchastheaveragepatchsizeofremnantvegetationwithinaparticularanalyticalunit,or theamplitudeoftheseasonaloscillationingreennessfromanormalizeddifferencevegetationindex(NDVI)profile;(2)thesecondrelatestowhatthemetricmeasuresintermsofinflu-enceon the target issue; forexample,patchsizesgreater thanx indicateahigherwaterextractiontowaterrechargeratio,resultinginaloweringofthewatertable,oranamplitudeequaltoyindicatesa75%probabilitythattheareaisusedforcerealcroppingandhencehasnowaterextractioncapacityinsummer.Inthecontextofmulticriteriaanalysis(MCA),assignmentofmeaningtospatialandtemporalmetricsdepends on project-based research, wherein a relationship is established betweensomeaspectoflandusechangeorcondition,orsomederivedpropertyofaninputvariablelayer,andametricthatisrobustandtranslatablefromstudytostudy.Intrin-sically,somemetricshavemoreeasilyascribedmeaningthanothers—themeaning-fulnessbeinginverselyproportiontothedegreeofabstractionandextentofremovalfrom biophysical, economic, or social measures that are a directly related to themanifestationoflandusechange.
ThereisaverywidearrayofpotentialanalyticaladjunctstoMCA.8Thesecanbesummarizedintoseveralgroupsofmethods:thosefordealingwithinputuncer-tainty;thoseappliedtoweightingandranking;modelsanddecisionsupportsystems(DSS)deliveringhighlyprocessedandsummarizedderivedlayersintotheanalysis;various cognitive and soft systems methods requiring transformation for use, orperhapssittingoutsideofthestandardMCA;optimizationapproaches;andintegratedspatialDSS,participatorygeographical information(GIS)andmultiagentsystems.However, the quantification, metrication, and summary of spatial and temporalsignals and temporal change in spatial patterns represent a level of sophisticationandderivationthathasyet tobefullyexplored.Recentexperiencewith thedevel-opmentof simple scenario tools forassessingcarbonoutcomes frommanagement
42963_C002.indd 18 11/6/07 7:17:14 AM
Developing Spatially Dependent Procedures and Models 19
change inrangelands9,10hasemphasized the importanceofspatialgradients, inter-actionsandpatterns,and temporal trendsand transitions in response toanthropo-genicandenvironmentalforcing.Inthischapter,theAustralianrangelandsareusedasanexamplecoupledhuman-environmentsystemtoexaminetherolethatspatialandtemporalinformationcanplayinamulticriteriaframeworkaimedatinformingpolicyandbydefinitionrequiringasubstantialelementofsocialcontext.
Thereisalargeandlong-standingliteraturebasedealingwithsignalprocessing11andtimeseriesanalysis12,13,14andmergingmethodsacrossthesetwoareas.15Thisliteratureindicateshowthepropertiesofdemographic,economic,social,andbio-physicalpoint-basedtimeseriesdatacanbecaptured.Withspatiallyexplicittimeseriesweareinterestedinhowthesepropertiescanbemeaningfullymappedintoamulticriteriaanalysisframework.
2.2 ConCePt
Thepremisebehindthischapteristhatsomeformofmulticriteriaframeworkisuse-fulforexplorationofcomplexcoupledhumanenvironmentsystemsandforinformingpolicydecisionmaking.Integrationofnonscientificknowledgeisofkeyimportance,andtheuserperspectivemaybetheultimatecriterionforevaluation.16Arequirementofthisanalysisisthatitissimpleandtransparenttotheclient,stakeholder,partici-pant,anddecisionmaker,butthatithasthecapabilitytocapturecomplexspatialandtemporalinteractionsandtrendsthatinfluencethenatureofbothsystembehaviorand evolution and the consequencesof decisions. Inprinciple, it is necessary formulticriteriaframeworkstoincludemeasuresofsystemdynamics—bothspatialandtemporal.Therefore,theunderlyingthemeinthischapteristheefficacy,efficiency,andinformationcontentoftransformationsofspatialandtemporaltrends,patterns,and dynamics into standardized, indexed layers for use in spatial multicriteriaanalysis.Theensuingdiscussiondoesnot imply thatmulticriteriaapproachesareeithertheonlywayorthebestwaytoapproachanalysisforpolicydecisionmakingincoupledhumanenvironmentsystems.Itissimplyoneapproachthathasproventobeuseful,4,6,17,18anditprovidesacontextfordiscussionoftheissueoftransformationofspatialandtemporalsignalsoutofacomplexmultidimensionalresponsespaceintostandardized,unitless,ordinalscalarstoassistinhumanproblemexplorationanddecisionmaking.
2.3 tRAnsFoRMAtIon IssUes
Intermsofdefinition,transformationistakentomeanamethodbywhichamorecomplexspatialpatternorrelationship,ortemporalpatternortrend,ismappedintoonetomanyquantitativemetrics thathavesomefunctionalrelationshiporunder-standabledescriptivecontribution thatcanbe ranked in termsof theobjectiveofamulticriteriaapproach.Thistransformationcanthereforebeasimpleregressionfunctionwhereintheslopeisusedasthemetric,oritcanbeasetofpartialmetricsthat together provide a composite indicator capable of being ranked.Examples ofthelattermightincludeseveralspatialpatchmetricssuchasnumber,size,andedgelengthorseveralcurvemetricssuchastimings,amplitude,andareaunderthecurve.
42963_C002.indd 19 11/6/07 7:17:14 AM
20 Land Use Change
Sextonetal.19definefourdimensionsofscale:
(a) Biological—fromcell,organism,population,community,ecosystem,land-scape,biometobiosphere;withfourusefullevels:(1)genetic,(2)species,(3)ecosystem,and(4)landscape.
(b) Temporal—differentspansoftimefordifferenteventsandprocesses. (c) Social—examplescheme:(1)primaryinteraction—physicalhumancontact
withecosystem,(2)secondaryinteraction—emotional(laws,policies,regu-lation,votes,plans,assessments,andso forth), (3) tertiary—indirectandqualitative(values,interests,cultures,heritage,andsoforth).
(d) Spatial—manyhierarchiesbasedonnumerousattributes.
Possiblythegreatestissueintransformationrelatestoscale-dependenteffects.Thisisparticularlysoinhumanenvironmentinteractionswheregeographicalvaria-tioninhumanbehaviorandbiophysicalfactorsatdifferentscalesinteract.20Thisisalsoparticularlysowhencombiningbiophysicaldatawitheconomicandsocialdatawherepixelsandpolygonswithdiscretespatialpropertiesmustbecombinedwithindividualbehaviors and institutional arrangements thatoperate in amultivariatepseudospatialsphereofinfluence21andhavenonequivalentdescriptions.7Forexample,aregionmaybeboundbycertainrulesthatgovernthedegreeofeconomicsupportforcertainactivities.Thepotentialspatialdimensionsaretheregionboundary,buttheeffectivespatialpatterninsidetheregionisgovernedbyarangeofexistingconditions,humancharacteristicsandbehaviors,economicconditions,andbiophysicallimita-tions, someofwhichcanbedirectly suppliedas spatialdata layers, andsomeofwhichrequireamodelofpotentialinfluenceoreffecttocreateanindexoflikelihoodofadoptionorcompliance.Itispossibletoestablishequivalencerulesbetweenbio-physicalandsocial landscapeelementsusingstructural (e.g., speciescompositionandhydrologicalsystemversuspopulationcompositionandtransportationandcom-municationinfrastructure),functional(e.g.,patchconnectivityversuscommuting),andchange-based(e.g.,desertificationversusurbanization)22approaches.Itisalsopossibletoestablishdemographicscaleequivalencebetweenbiophysicalandsocialdomains using a spatial hierarchy based on individuals (e.g., plants and people),landscapes (e.g., watersheds and counties), physiographic regions (e.g., ecoregionandcensusregion),andextendedregions(e.g.,biomeandcontinent).22
Relationships of information derived within one scale category are relianton assumptions from others.19 In a more general sense, the modifiable areal unitproblem(MAUP),wherecorrelationsbetweenlayersvarywithdifferentreportingboundaries, requires excellent transformation methods, using finer scale data toinformthebroaderscaleanalysis,23andconstantawarenessofthepotentialproblemofunderstandingandmanagingpatterns,processes,relationships,andhumanactionsatseveralscales.19Multiagentsimulationapproaches24haveconsiderablebenefitsindealingwithindividualbehaviorsinurbananddenselypopulatedsystemproblems25aswellaslandcoverchangeproblems26andtechnologydiffusionandresourceusechange.27Theymayalsobeappliedtoexamineemergentpropertiesatthemacro-scalefromdifferentmicroscaleoutcomesandincorporatespatialmetrics.28
Thesecondmajorissueintransformationrelatestoameaningorquantifiablerela-tionship with an attribute that affects or contributes to assessment of the objective
42963_C002.indd 20 11/6/07 7:17:15 AM
Developing Spatially Dependent Procedures and Models 21
oftheanalysis.Akeyelementhereis thefieldworkandanalyticalworktodevelopspecific and general quantitative, probabilistic, or qualitative relationships betweenpatternsandprocesses29,30,31thatcanbeusedeitherlocallyorgloballytoassignarankin termsof somemulticriteriaobjective.Laney32describes twoapproaches: studiesidentifying the land cover and change pattern, then seeking to develop amodel toexplainthesepatterns(pattern-ledanalysis)andstudiesthatdevelopatheorytoguidepatterncharacterization(process-ledanalysis).Bothapproachesmayhaveflaws,withpattern-ledanalysisbeinghighlydatadependentandabletoidentifyonlyprocessesassociatedwiththatdata,andprocess-ledanalysisdependentonthepriortheoreticalmodel,adherencetowhichmayprecludetreatmentofotherequallyvalidprocessesandpaths.Theultimateintegrationoftransformationandmeaningmightberepresentedbythe“syndrome”approach,33whereinalternativearchetypal,dynamic,coevolutionpatternsofcivilization-natureinteractionaredefined(e.g.,desertificationsyndrome).Thesesyndromesmightbecharacterizedbyhighlydevelopedcomposite indicatorsthatincorporatecomplexderivedspatiotemporalrelationshipsandpatterns.
2.4 tRAnsFoRMAtIon DoMAIns AnD MetHoDs
Theeffectivenessofamulticriteriaframeworkisprobablyproportionaltotheextenttowhichsystemelementsandinteractionsarecaptured.Representationoftimeintradi-tionalGISplatformsisverypoor,34whileimage-processingsystemsthathandletimeseriesofspatialdatalackthetoolsforextractionandsummaryofinformationfromthetimedomain.Moreaccessiblespace-timeanalyticalfunctionalityisneededtomakeawidevarietyoftransformationapproachesavailabletothoseotherthanexpertspatialanalystsandsignalprocessors.Thechallengeliesinacquiringdatainallofthepoten-tialresponsedomainsatasuitablescaleandwithacceptablequality.AlistofpossibleinformationdomainsisgiveninTable2.1alongwiththekindoftransformationissueinvolvedandsomepossiblemethods.Where individualsare involved,demographicinformationcoupledwithsurveysandunitsofcommunityaggregationformthebasisfortransformation—spatiallyintermsofthelocationofbehaviorsandrecordedpref-erencesinrelationtolandusepatternsandchanges,andtemporallyinthesensethattrajectoriesinopinionandbehaviorleadtolandusechange.Socialsystemsarereflex-ivelycomplex(i.e.,havingawarenessandpurpose).Therefore,withinasocialmulti-criteriaanalysiswithnonequivalentobserversandnonequivalentobservations,thereisaneedtodefineimportanceforactorsandrelevanceforthesystem.7Theactorsinsocialnetworksthatinfluencethelanduseoutcomemustbespatiallyrepresented,35butthereisachallengeincapturingthelinkbetweeninfluenceandbiophysicaloutcome.36
Atthelevelofsocialandeconomicstatistics,collectionunitsoftendeterminethenatureoftheanalysis.Socialindicatordatamaybeidiosyncraticatthelocalscale,haveincompletetimeseries,havedefinitionalchangesovertime,andhavemisalignedreportingboundaries.37ThisresultsinMAUP,ecologicalfallacy,expedientchoiceofstatistics,arbitrarychoiceofmeasures,anddifficultyinestablishinganycausalrela-tionships.37,38Transformationsarerequiredtosummarizetemporaltrendsandcyclesand todefinespatialpatternsandrelationshipsatafinerscale,whichmayhelp todistributetheinformationdownwardfromthecollectionunitinscaleinaspatiallyexplicitway.Dasymetricmappingcanbeusedonlytoassignpopulationstoremotely
42963_C002.indd 21 11/6/07 7:17:16 AM
22 Land Use Change
tAble 2.1transformation Domains for spatiotemporal Multicriteria Frameworks
Information Domain transformation Issue Methods
Individualbehaviorsandpreferences
Representationofindividualatresolutionofanalysis
Transformsurveyinformationintostatisticsandmetricsthatsummarizethetendenciesinthepopulationforthatspatialunit
Individualperceptions Representationofabstractconceptssuchasbeauty,degreeofspacecontamination,etc.
Uselandscapeimagemetrics,spatialdistancesandlandscapecontents
Institutionalarrangements,governmentregulations,andincentives
Representationoftheinfluenceorlikelihoodofadoptionorcompliance
Developprobabilitymodelsbasedonpriorsurveysofimpactandcreateprobabilitylayers
Economicvariables Relatingcollectionunittoanalysisunit
Self-organizationofspatialunits;temporaltrends,metrics,andtimeperiodsummaries
Socialstatistics,societalsystems,transportandsurveys
Conversiontoafactorlayer—attachingameaningandarank
Developprobabilitymodelsandpartialregressionmodelstoascribesomeofvariationintargetissuetothesocialfactors.Createfactorlayersbasedonthepercentagevariationdescribed,direction(+or–)andstrength(slope)oftrend
Climate Impact/responseanoutcomeofcomplextemporalsequencesandspatialpatterns
Developimpactthresholdandseveritylayersbasedonmultiplescenarioruns
Disturbances—fire,grazing,clearing,flooding,desertification,urbanization,abandonment
Representationofspatialextent,spatialgradient,timing,duration,impact,agents(i.e.,activeunitssuchasanimals)
Derivemetricsdescribingspatialandtemporalpatterns,harmonics,limits,responses,demographicsthatcanberankedintermsofthetargetissue
Landuse Representationofpersistenceandchangeatlevelofcovertype,species,managementpractice,seasonalmagnitude
Derivemetricsthatcapturepattern,change,persistence,sequence,andallquantitativepropertiesofthechangeinahierarchicalstructure
Bio/geochemicalprocess—hydrology,sedimentation,nutrients,gasexchange,emissions,consumption
Representationofprocessintermsofoutcomeaffectingorinfluencingtargetissue
Aggregated,averaged,summarizedandprobabilityconvertedoutputsfromprocessmodeling
42963_C002.indd 22 11/6/07 7:17:16 AM
Developing Spatially Dependent Procedures and Models 23
sensedurbanclasses,andpopulationsurfacescanbecreatedbyassociatingthecountwithacentroidanddistributingitaccordingtoaweighteddistancefunction.38,39Therelationshipbetweenpeopleandtheirenvironmentiscapturedbycognitiveappraisalfromperceivedenvironmentalquality indicators.40 Indicatorsof residentialqualityand neighborhood attachment40 might be transformed into spatial properties byassigningproximityfunctionstoservices,assigningdistancemetricstoroadaccessand access to green space, ranking buildings for aesthetics and quality of humanenvironment,andmappingthesewithspatiallyexplicitviewabilityconstraints.
Climateprovidesanoverarchinginfluencethatisbothspatiallygeneralizedandlocally spatiallydependent, and it is a fundamentally time-dependentandcyclicalfactor.Here the transformations includespatialpatternsofmicroclimaticvariationandtemporaltrendsinclimatechange,metricsofseasonalcyclesandtrends,orvari-anceinextremes.Theremaininginformationdomainsarethemostspatiallyandtem-porallyinteractive,withbiogeochemicalprocessesinteractingwithlandusetypeandchangehighlyinfluencedbyhumanandotherdisturbances.Thesedomainsrequiremanyspatialandtemporalmetricsaswellashigherlevelmeasuresofsystemresponseintheformofoutputsofspatiallyandtemporallyexplicitmodels(e.g.,hydrology).
Somemethodsfortransformingcomplexspatialandtemporalpatterns,relation-ships,andsignalsaregiveninTable2.2.Theseareconsideredintermsofthegeneralspatialcontext,thespecificsocialnetworkdatawherespatialandnonspatialcogni-tivedomainsmix,40 thevisual contextwhereviewsandbeautyperceptions inter-mingle with functional and locational considerations,41 and the temporal contextwheremethodsfromnonspatialtimeseriesanalysiscomplementmethodsspecificallydeveloped for time series of satellite data. The spatial and temporal contexts arediscussedinmoredetailinthefollowingsections;however,theexamplelandscapecontextusedinthediscussionmustfirstbedescribed.
2.5 eXAMPle lAnDsCAPe ConteXt—AUstRAlIAn RAnGelAnDs
TheAustralianrangelandsprovideasuitablecombinationofspatialandtemporaldynamics and dependencies for illustration of issues surrounding transformationofspatialandtemporalsystempropertiesintoanMCAframework.Thissystemischaracterizedbyahierarchyofscaleswithinandacrosswhichinfluences,effects,relationships,andfunctionsoperate.Allofthescaledomainsofbiological,temporal,social,andspatialarerelevant.Thesystemisaffectedbyverylarge-scaleclimateandeconomicfactorsandverysmall-scalespatialdependenciesinhabitatsandland-scapefunction.Therangelandshavethefollowingcharacteristics:
1.Diversityinclimate,soils,andvegetationtypes(Figure2.1). 2.Heavilyutilizedbydomesticlivestock. 3.Substantiallyinfestedwithferalanimals. 4.Asignificantbiomassandsoilcarbonreserveandasourceofgreenhouse
gasemissionsthroughannualwildfire. 5.Systemprincipallylimitedbywateravailability. 6.Spatialinteractions,patterns,andgradientssubstantiallyrelatedtolandscape
scaleterrain–waterdynamicsandanthropogenicwatersupply(bores).
42963_C002.indd 23 11/6/07 7:17:17 AM
24 Land Use Change
7.Temporal dynamics heavily influenced by interaction between climate(watersupply),grazingandfire.
8.Meso-scalelandscapepropertiesstronglylinkedtooveralllandscapefunction,particularlyinrelationtowaterharvestingandconsequenthabitatdevelopment.
9.Significantsocialissuesthroughindigenousrightsandsacredsitesandsite-basedtourism.
10.Managementofthelandscapeisinfluencedbyexogenoustemporalvaria-tionincostoffinanceandinputs,tradebarriersandrestrictions,priceofcommodities,specificallybeefcattle,andchangesinfamilystructuresandruralemployment.
tAble 2.2Methods for exploration and transformation of Complex spatial and temporal Patterns, Relationships, and signals
spatial23
Convolutionfiltering(movingwindoworkernel)containingfunctionsfromsimplestatisticstotexturalindexestocomplexregressiontospatialautocorrelation
Distancemeasuresinspatialneighborhoods—associationofpatch,gapandshapewithsocioeconomicchangefactors29
Cost–distancegenerationofuserandpurposedefinedanalysisunitsGeographicallyweightedregression49toovercomenonstationarity,spatialdependencies,andnonlinearspatialdistributions,allowingclassificationofsystemparametersbyalearningalgorithm—self-organization
spatial/social networks35–50
Resilience,fastandslowadjustment,perturbation,catastrophe,turbulence,andchaosmodels51
Bioecologicalmodels—analysisofdynamicphenomenaofcompetition-complementarity-substitution(networkasaniche);sociallandscapeanalysisinlandscapeecology22
Neuralnetworks—noteasilyinterpretablefromeconomicviewEvolutionaryalgorithms—geneticalgorithmswithbinarystrings;evolutionaryalgorithmswithcontinuoussettingandfloatingpointvalues
Visual41
Characteristicfeatures—lower-upperfeaturerelationships;contourblockdrawings;imagetextures;contoursandhorizon;spatialrelationsofspacesandelements;proportionsoflandscapezonesinview;hierarchicalproperties;typologyoffringes
Spatialdistancemeasures—viewtexture;intrusionintoskylineandlandscapeline;relativestructuralcomplexity;relativeproportions;distance–sizerelationships
Sensitivity—functionaldistanceinlandscapes;structuraldistancestobekeptfree
temporal
Traditionaltimeseriesanalysis12,14,52,53—trends,cycles,seasonality,lags,phase,irregularity,smoothing,differencing,autocorrelations,spectralanalysis
Curvemetrics43,54—limits,amplitude,periodicity,timings,areas,slopes,trajectories55;phenologySignalprocessing—Fouriertransforms56;Wavelettransforms15,44
Principalcomponentanalysisoftimeseries44
Complexbio-socioeconomiccycles(e.g.,Kondratieffwaves57);syndromesofchange33
42963_C002.indd 24 11/6/07 7:17:17 AM
Developing Spatially Dependent Procedures and Models 25
Thesystemrepresentsatypeofexamplewherehumandemographicsarenotamajor factorsince largepastoral leasesareessentiallyunpopulatedexcept for thestationhomesteadandassociatedbuildings.Humaninfluenceinthisenvironmentisprovidedthroughmanagement,whichreachesoutfromthehomesteadtoinfluenceverylargetractsofland.Hence,superficiallyitmightbedifficulttodrawmethod-ological parallelswith themany coupledhuman environment systems worldwideandhighhumanpopulationdensities.However, in this system,demographicsarestill important since themajor influential population is thatofdomesticatedbeefcattle,withancillaryinfluencefromferalanimalpopulations.Theyareindividualeconomicunitswithcostsassociatedwithparasiteanddiseasecontrolandhumanhandlingandvalueintermsoffoodandbreedingpotential.Thedecision-makingframework forcattle ismuch lesscomplex than forhumans;cattle requirewater,feed,shade,andsocializationandwilloptimizetheirbehaviorwithinthisresponsespace. Nevertheless, they influence and respond to spatial and temporal patterns,and,therefore,thissystemcanstillprovideusefulmethodologicalinsights.
2.5.1 SPATIAL PATTERNS AND RELATIONSHIPS
Thespatialinterrelationshipsinthisrangelandsystemcanbeillustratedbyastylizedlandscapecontainingartificialwaterpointssurroundedbypiospheresofinfluencebygrazinganimalsuponthevegetationuptoadistancelimit(Figure2.1).Thesewaterpointsoccurwithinfencedpaddocks,partsofwhichareinaccessibletostocksincethey are outside the water access limit. The paddocks also contain different landcovertypeswithdifferenthabitatsuitability,firesusceptibility,andlivestockcarryingcapacities.Thelandscapehasrockyareas,areaswiththickshrublandinaccessibletostock,swampyandsalineareaswithlowproductivity,andanaboriginalsacredsite.Theareaalsohasanaestheticcomponentwithaviewpointandrestarealocatedonamajorroad,withbasicpicnicfacilitiesoutsidethemappedextent.Themajorspatial
Water point piosphere of grazing intensity
Fenced paddocks
Heavily thickened woodland with shrubs
Poor, light soil
Inaccessible rocky outcrop
Swampy area with unpalatable plants
Saline scald area
Elevation contours
Sacred aboriginal site
FIGURe 2.1 TheconceptofgrazingpiospheresinteractingwithlandscapestructuretocreatespatiallyandtemporallydependentresponsezonesinAustralianrangelands.Thesearemoreprevalentwhererainfallislessreliable,paddocksaresmaller,andstockingpressureishigher.
42963_C002.indd 25 11/6/07 7:17:20 AM
26 Land Use Change
gradients in this landscape are createdby the effect of grazingonvegetation andhabitat,theconnectivitybetweenhabitats,thestructureinrelationtoshelter,waterharvestandstockaccess,andtheappearanceofthelandscapefromaspecificdirec-tionandangleofview.
Inordertocapturespatialattributes,alevelofspatialpatternreportingmustbedefined,andthislevelofaggregationmustbecompatiblewiththeresolutionofotherdataintheanalysis.Thescaleofaggregationmightrelatetosomefunctionaldistanceandsphereofinfluenceinthelandscape,andpatternextractionmightbeundertakenforanumberofdifferentaggregationunits,42anestedsetofpatchscales,30inordertospecificallycapturetheinfluenceoflandscapestructurefromdifferentelementsofthesystemsuchasbirdhabitat,cattlegrazingbehavior,scaleofmicrotopography,andsoforth.
2.5.2 TEMPORAL PATTERNS AND INFLUENCES
The temporal behaviors of, and influences on, this rangeland system could bedescribedbya timeseriesofweatherandsatellitedata,whichrecordssequencesofdetectablelandcoverchangeandvegetationstate,aswellasderivedmeasuresofsystemfunctionintegratedthroughmodels.Amonthlytimeseriesofnetecosystemcarbonexchange(Barrett,personalcommunication)providesanexampledatasetfor illustrationofapproaches todisaggregationanddecompositionof signals intomeaningfulindexes(Figure2.2).Aseriesofseasonallybasedsystemresponsespro-videthebasisforextractionof:
1.Curvemetricsthatdescribethetiming,duration,magnitudeandperiodicityoftheresponse43
2.Acumulativeaggregateofthenetsystembehaviorthroughtime 3.Trendinsignalfromwaveletorothertransforms15,44
4.Temporalautocorrelationtoseehowstrongthe“memory”isinthesystem—astrongmemoryindicatesmoreregularcyclicalbehavior
5.PowerspectrumandFourier transformsonoriginaldataandfirstdiffer-ences or first derivatives to detect major cyclical patterns—in this caseoccurringatabout22,44,and66months
6.Cumulative probability curves to identify the relative behavior for someproportionofcases(Figure2.2)
Thesemetricsandmeasuresoftimeseriesattributescanbederivedspatiallyandconvertedtosingleorpartialcomponentindicatorsofsystemproperties.
Thetemporalinfluencesarealsorepresentedbynonbiophysicaltimeseriessuchas livestock numbers, climate cycle indexes, prices and costs, and human activitymeasures(Figure2.3).Thesedatamayonlybeavailableatacoarselevelofspatialresolution,suchascattlenumbersfromtheagriculturalcensus,orindividualbehaviorsfromsocialsurveyswithlimitedsamples.Alternatively,theymaybeglobalvariablessuchascattleprices,interestrates,andclimateindexessuchasthesouthernoscilla-tionindex(SOI).Inthesecases,ameansmustbefoundtoapplythesespatiallyviasomefilteringlayerthatassignstheattributesonlytothosepixelswheretheinfluenceoccurs,ortothosepixelsnotconstrainedbyotherfactors.
42963_C002.indd 26 11/6/07 7:17:21 AM
Developing Spatially Dependent Procedures and Models 27
–2
0
–4
0
–0.0 2
0.00
0.02
–0.4
0.0
0.4
0.8
0.0
0.2
–0.2
0.0
0
40
80
NEP 1981 – 2000
NEP
Cum
ulat
ive N
EPRe
spon
seSi
gnal
Aut
ocor
rela
tion
Pow
erSp
ectr
um
Pow
erSp
ectr
um
Firs
tD
iffer
ence
Cum
ulat
ive
perc
enta
ge
Normal score
Wavelet transformTrend line
Fourier transform
Running cumulation
Time in months Time intervals
Metrics
Amp Integral
IntervalPeriod
0 12T1 T24 8
Max
Min
0.5 Amp
0.2
0.0
0 24 48 72 96 120 144 168 192 216 240
0.00000
0.00008
0.00016
–2 –1 0 1 2 3 4
0.00000
0.00010
FIGURe 2.2 Timeseriesapproaches toextractingsignalsummary indicators.Atimeseriesofnetecosystemproductivityindicatesthebasepotentialforcarbonfixation.Thismaybetrans-formedinto indicatorsbycalculatingmetrics, includingarunningintegral,extractingthefre-quencyofcyclicpatternsusingfastFouriertransform(FFT)orpowerspectrumsonoriginaldataorfirstdifferencesandderivatives,definingdirectionoftemporalchangethroughtrendanalysisorwavelettransforms,andestimatinglikelihoodofvariouslevelsthoughcumulativeprobability.
42963_C002.indd 27 11/6/07 7:17:22 AM
28 Land Use Change
–20
0
20
SOI
0
2
0
4
8
12
Pow
er
0
40
80
120
0
200
400
600
0
0
20
40
60
1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
Cattl
e No.
(mill
ion)
1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 20008
9
10
11
12
Interest paymentsHandling/marketingWages – hired hands
Age – ownerAge – spouseHours worked – ownerHours worked – spouse
Cattl
e ($/
head
)
Cost
s ($)
40000
80000
120000
Age
/Hou
rs
IPO
FIGURe 2.3 Nonbiophysicaltimeseriesalsoprovidepotentialindicatorsbutmaynotbespatially explicit at the required scale.Theseneed tobe transformed into indicators suchastrendsindemographics(e.g.,cattle),patternsofclimateandfrequencyofoccurrenceofcertainclimate types, trends inpricesforcommodities, trends incostsofproduction,andtrendsinhumanactivitiespotentiallyaffectingmanagementandeconomicoutcomes.
42963_C002.indd 28 11/6/07 7:17:24 AM
Developing Spatially Dependent Procedures and Models 29
Thespatialextentofdiscreteandconsistenttemporalpatternsmaybedefinedby,forexample,aprinciplecomponentsanalysis(PCA)onthetimeseriesandsubsequentclassification (Figure 2.4a). This reveals distinct temporal patterns that representa regional summary (Figure2.4b; the temporal net ecosystemproductivity (NEP)signalforoneoftheseclassesisusedinFigure2.2).Bycontrast,timeseriesattributesmaybecalculatedonapixel-by-pixelbasis,andthedata,suchasstandarddeviationinNEP,aremappedtoprovideacontinuousfactorlayer.Theanalysisofthetrendsinthetimeseries(Figure2.2)mayprovideinformationaboutmajorperiodsofdifferingbehaviors,suchasthenetlosscarbonthroughoutnorthernAustraliabetween1985and1993,versusthenetgainincarbonbetween1994and2000(Figure4.2c).
2.5.3 DATA AND INFORMATION: SCALE OF REPRESENTATION
Movingdownscaletotheregionofinterestforanalysis,theVictoriaRiverDistrict(VRD)inthenorthernAustralianrangelandsprovidesagoodbasisforassessmentof
(a) (b)
(c) (d)
PCA classes12345678910111213141516171819202122
St. Dev. NEP0.004 – 0.0270.027 – 0.050.05 – 0.0730.073 – 0.0960.096 – 0.1180.118 – 0.1410.141 – 0.1640.164 – 0.1870.187 – 0.210.21 – 0.233
NEP 8593< –2–2 – –1–1 ––0.5–0.5 – –0.2–0.2 – 00 – 0.20.2 – 0.50.5 – 1> 1
NEP 9400< –2–2 – –1–1 ––0.5–0.5 – –0.2–0.2 – 00 – 0.20.2 – 0.50.5 – 1> 1
FIGURe 2.4 (See color insert following p. 132.) (a) The spatial pattern of temporalsignalsmaybegroupedbyapplyingprincipalcomponentsanalysis to the timeseriesandcreatingaclassificationbasedonthemajorprincipalcomponents.(b)Thetemporalmetricsmaybecalculatedonthespatialtimesseriestocreatemapsof,forexample,standarddevia-tionofnetecosystemproductivity(NEP).(c)Arunningintegration,trend,orwaveletanalysismaydefineperiodsofdistinctbehaviorinthetimeseriesthatcanthenbesummarizedbymetricssuchasanintegralofNEPforperiodsofdeclineandincrease.Shownfor1985–1993and1994–2000here.
42963_C002.indd 29 11/6/07 7:17:27 AM
30 Land Use Change
datascalesandtransformationthroughspatialfiltering(Figure2.5).TheregionhasthreeofthePCAclassesgiveninFigure2.4a.ThetemporalsignalandassociatedmetricsandtimeseriesattributescouldbebroadlyassignedtoallareaswithintheclasszoneintheVRD.However,theNEPdatarepresenttheresponseofthesystemundisturbedbylivestock.Therefore,intheabsenceofspecificestimatesthattakethespatiallyvariableimpactofgrazingintensityaroundwaterpointsintoaccount,onecouldapplyanarbitraryscalingofeffectonNEPthatvariesfromusingthesuppliedsignalsforungrazedareas,tocompletelysuppressingtheaccumulationsignalatthewaterpoint,wherestockpressureishighest.Verybroadscaleestimatesofprofitperhectareatfullequityprovideanindicationofprofitability.Minepresenceisindicatedbyacountforeach1kmpixel—adecisionmayneedtobemadeaboutabufferingruleforradiusofdisturbance.Thenumberofthreatenedbirdsisderivedfromaverycoarseresolutiondataset.However, ifany information isavailableon thehabitatforthesebirds, then,forexample,anindexofpotential threattogrounddwellingbirdscouldbecreatedwithveryfinespatialresolutionusingarulegoverningdegreeofdisturbancewithdistancefromcattlewateringpoints.Completelyaspatial,butimportant,systemattributesandmetricsmaybedownscaledtoappropriateresolu-tionifrelationshipstospatialdataatanappropriateresolutionareknownorcanbederived.Themajorchallengearises inascribingcausal relationshipsandareasofinteresttohumanpopulationcentersbasedonsocialstatisticsaboutactivitiesandpreferencesofhumans.However,certainkeyvariablessuchasbusinessenterprisedebttoequityratiosmaybeimportant.Ifspatialdataareofsufficientqualityandeffectsofdisturbanceonkeyenvironmentalmeasuresareknown,thenaspatialeco-nomicandsocialdatamaybeusedalongwithbiophysicaldatatoascribeintegrativeindexesof economicbenefit, costofdegradation, and socialbenefit to individuallandscapepixelshavingparticularsuitesofbiophysicalattributes.
2.5.4 SOME SPATIOTEMPORAL INPUTS TO A RANgELAND MCA
Therangelandexampleprovidesaveryspecificopportunitytoelucidatetheetiologyofspatialandtemporaltransformationstosummarizecomplexsystemresponsesandbehaviorsinmulticriteriaanalysis.Ifwetaketheeffectofgrazingonvegetationcondi-tionasanexampleofacomplexbiophysicalprocessinfluencedbyhumanmanagement,inordertocapturetheelementsoftheconditionofaparticularpixelonemightneed:
1.Distancetowaterpoint(seeFigure2.1)—simpledistancemetric. 2.Valueofaveragegrazingpressure—complexfunctionalcalculationbasedon
distancefunctionsdescribingthecost–distancerelationshipbetweendistancefromwaterpointandlivestocktendencytotraveldistancefromwater.Thenthevalueofaveragegrazingpressureiscalculatedfromtheratioofaveragedensityoflivestockunitstothenominalsafecarryingcapacityofthevegeta-tiontype.Thisisrelatedtotime-useanalysis,45wheretheperiodoftimethatanindividualunitisinaparticularspatiallocationisimportant.
3.Aproximityfunctiontodenseshrubbyvegetation—modifies(2)above,sincegrazingpatternmaybeperturbedbyanimalbehaviorwithrespecttoshelter.
4.Averagemaximumseasonalbiomassforthatpixel—simpletemporalcurvemetric.
42963_C002.indd 30 11/6/07 7:17:27 AM
Developing Spatially Dependent Procedures and Models 31
(a) (b)
(c) (d)
FIGURe 2.5 (See color insert following p. 132.) Temporalsignalsareusuallybasedonbiophysicalorhumanphenomenathatoperateatalargescale(e.g.,climate,interestrates).Demographicchangesatafinescalemayhavescalelimitationsduetolevelofaggregationinreporting.Temporalsignalsandindicatorsarefilteredbyspatialvariation.TheVictoriaRiverDistrictintheNorthernTerritoryofAustraliaishighlyproductive.(a)Cattlearedis-tributedoffreehold-leaseholdlandbutconfinedbywaterpoints.(b)Bothproductivityandecologicalimpactvarywithvegetationtype,whichisassociatedwithsoils,topography,andrainfallgradient.(c)Costsarelowandenterprisesareprofitablebuttheincrementissmallonaperhectarebasis.(d)Miningwithmajorphysicaldisturbanceoccurssporadicallyacrossthearea.Therearethreatenedbirdspeciesintheregionandthesemaybegroundnestingandimpactedbygrazing.
42963_C002.indd 31 11/6/07 7:17:29 AM
32 Land Use Change
5.Average amplitude for annual biomass for that pixel (max minus min)temporalcurvemetric.
6.Timeofhalfmaximumseasonalbiomassforthatpixel—simpletemporalmetricforstartofperiodofgreenfeedavailability.
Thesedatamightbeusedtodefineanindexofanimalimpactforeachpixelthatintegrates spatial and temporal relationships, trends, and influences.Thiskindofcombinationofspatialandtemporalrelationshipsmaybeusedtoconstructspatiallyexplicit indexes forotherelementsof thesystemsuchas landscapefunction,bio-diversityimpact,andsocioeconomicbenefit(Table2.3).Thetemporalcomponentisfilteredonthebasisofspatialrelevance(i.e.,grazedareasarerelevantbutungrazedareasarenot),exceptwherethetemporaltrendofresponsehasasphereofinfluencebeyondthelocalpixelandisthengovernedbyaspatialfunctionforrelativeeffectandadjustedbyanyspatialconstraints.
2.6 A FRAMewoRk FoR A MUltICoMPonent AnAlysIs wItH MCA
Finally, a broad framework for application of MCA to assessment of ecosystemservicefromarangelandenvironmentisdescribed(Figure2.6).InthisframeworkMCAisusedtoprovideassessmentofcurrentecosystemservicelevels,andthentoassessstrategiesforimprovingecosystemservicesthroughmanagementchange.ThespatiallyexplicitecosystemserviceratingforanareacouldbeconstructedusingthesuiteofindividualandcompositeindicatorsinaMCAenvironmentsuchasMCAS-S(multicriteriaanalysisshell–spatial46;Figure2.7).Thisframeworkcouldmakeuseofspatialandtemporalmeasuresandmetricsaspartofthesuiteofindicatorsusedtodefinetheconditionofthelandscapeintermsofarangeofusesandfunctions.TheapproachdescribedinFigure2.6usesstateandtransitionmodelsofvegetationcondition.10,11Therangelandlandscapeisclassifiedintostatesbasedondisturbanceof original natural vegetation.The states are described in termsof structure andfoliagecoverandtypeofvegetation.47
Theecosystemservicefromthevegetationstatesisdescribedintermsofanumberofthemes:biodiversity,landscapefunction,waterharvesting,carbonstock,grazingpotential, indigenous utility, economic return, and aesthetic value. Each of thesethemesismadeupofasetofindicatorlayersdescribingattributes.Theseattributescouldbeindividualmeasuressuchasnumberofthreatenedbirds,carbonbiomass,orproximitytoaboriginalsacredsites.Alternativelytheycouldbecompositeindicatorsbasedonaggregationofindividualmeasuresincomplexspatiotemporalrelationshipstogiveanimalimpactperpixel.Theseindividualandcompositeindicatorsmaybecombinedbyvariousmethodsintoasingleindexofecosystemserviceforthattheme.Overallecosystemservicefromthatlandscapepixelisrepresentedbythecombina-tionoftheindividualthemeindicatorsintooneoverallindex.Theoverallecosystemserviceslevelcanbeimprovedbymovingthethemeindicatorstohigherlevels.Eachtheme can be improved by applying a number of strategies—individual strategiesmayimprovemorethanonetheme,butmayhaveanegativeeffectinotherthemes.Forexample,economicpotentialmaybeincreasedbyincreasingstockingrates,butthismayaffectvariouselementsoflandscapefunction.
42963_C002.indd 32 11/6/07 7:17:29 AM
Developing Spatially Dependent Procedures and Models 33
tAb
le 2
.3D
eriv
ed F
acto
r la
yers
tha
t In
tegr
ate
spat
ial a
nd t
empo
ral t
rend
s w
ith
Func
tion
and
eco
syst
em o
utco
me
Fact
orIn
put
data
spat
ial
Inpu
t da
tate
mpo
ral
Inte
grat
ion
Ani
mal
impa
ct
per
pixe
lB
uffe
red
wat
erp
oint
sw
ith
anim
ald
istr
ibut
ion
assi
gned
by
expo
nent
ial
dist
ance
fun
ctio
n;m
apo
fin
acce
ssib
leo
run
attr
activ
eve
geta
tion
Prob
abili
tyo
fan
imal
ut
iliza
tion
ina
nyti
me
peri
od—
from
dis
tanc
efu
nctio
nof
fre
quen
cyo
fvi
sita
tion
mod
ified
by
spat
iala
cces
sibi
lity
and
divi
ded
byf
eed
requ
irem
ent
for
spec
ified
wei
ghtg
ain
Mod
eled
tim
ese
ries
of
gras
slan
dgr
owth
;es
timat
eof
fee
dre
quir
emen
tper
uni
tof
time;
est
imat
eof
saf
eut
iliza
tion
rate
for
ve
geta
tion
type
s
Lik
elih
ood
that
ani
mal
de
man
dw
illn
ote
xcee
dsa
feu
tiliz
atio
nra
tein
an
ype
riod
—fr
om
hist
oric
alti
me
seri
eso
ffe
eda
ccum
ulat
ion
in
grow
ing
seas
ons
Com
bine
d,s
cale
din
dex
ofr
elat
ive
anim
alim
pact
on
any
pios
pher
epi
xeld
ueto
spa
tialc
ontr
olb
yw
ater
acc
ess
and
tem
pora
lpr
obab
ility
of
over
graz
ing
Lan
dsca
pe
func
tion
Ele
vatio
n,s
lope
,hig
hre
solu
tion
imag
ery
defin
ing
tree
san
dbu
shes
;de
rive
dla
ndsc
ape
drai
nage
pat
tern
and
wat
er
accu
mul
atio
n
Text
ural
ana
lysi
sof
pat
tern
of
wat
erh
arve
stc
onve
rted
to
mag
nitu
dea
ndd
ensi
ty
mea
sure
s
Tim
ese
ries
of
rain
fall
and
evap
orat
ion;
mod
eled
de
man
dfo
rw
ater
by
vege
tatio
nty
pes
Tim
ing,
fre
quen
cy,a
nd
peri
odic
ityo
fw
ater
flo
ws
and
prob
abili
tyo
fw
ater
har
vest
suf
ficie
nt
tos
usta
ins
tabl
esy
stem
Com
bine
d,s
cale
din
dex
ofw
ater
ha
rves
tfun
ctio
nfo
ran
ypi
xeld
ue
tos
patia
lpat
tern
of
land
scap
eel
emen
tsa
ndte
mpo
ral
prob
abili
tyo
fsu
cces
sful
cap
ture
if
har
vest
able
rai
nfal
leve
nts
occu
r
Soci
oeco
nom
ic
bene
fitp
er
graz
edu
nit
Liv
ew
eigh
tgai
npe
ran
imal
;ani
mal
di
stri
butio
n;v
alue
of
live
wei
ght(
pric
epe
ran
imal
);
cost
of
prod
uctio
n(c
ost
per
anim
al)
Perc
enta
geo
fto
tall
ive
wei
ghtg
ain
attr
ibut
able
to
each
gra
zed
pixe
lthe
nco
st,
pric
ean
dpr
ofitp
erg
raze
dpi
xel;
spat
ialc
orre
latio
nw
ithv
eget
atio
nty
pe.
Tim
ese
ries
of
anim
al
pric
es,c
ostc
hang
es,
anim
aln
umbe
rs(
stoc
king
ra
tes)
,fee
dav
aila
bilit
y,
live
wei
ghtg
ain
per
unit
time.
Len
gth
ofp
erio
dso
fin
crea
seo
rde
clin
ein
te
rms
oftr
ade
and
in
profi
tper
uni
tof
graz
eda
rea
Com
bine
d,s
cale
din
dex
of
econ
omic
ret
urn
per
unit
area
due
to
spa
tiald
istr
ibut
ion
ofa
nim
als
and
feed
,and
tem
pora
lcha
nges
in
pri
ces
and
cost
spe
run
itar
ea
Bio
dive
rsity
im
pact
per
pix
elD
istr
ibut
ion
ofe
ndan
gere
dsp
ecie
s;a
ssoc
iatio
nof
sp
ecie
sw
ithv
eget
atio
n,
land
scap
efe
atur
es
Con
nect
ivity
,pat
tern
of
habi
tat;
focu
san
dex
tent
of
dist
urba
nce
from
ani
mal
im
pact
per
pix
el
Tim
ese
ries
of
spec
ies
num
bers
;clim
ate,
fee
dsu
pplie
s,s
tock
num
bers
,pr
edat
orn
umbe
rs
Peri
odic
ityo
fpo
pula
tion
fluct
uatio
ns;t
imin
gof
lo
ws
inp
opul
atio
n;
tren
dsin
pop
ulat
ion
Com
bine
d,s
cale
din
dex
of
prob
abili
tyo
fne
gativ
ebi
odiv
ersi
tyim
pact
fro
mr
educ
ed
conn
ectiv
itya
ndd
istu
rban
ce
duri
ngp
opul
atio
nlo
ws
42963_C002.indd 33 11/6/07 7:17:30 AM
34 Land Use Change
Strategies for improving ecosystem service within a land cover state may beassessed by a mixture of spatial and nonspatial MCA where drivers of changeswithineachstatearedefined,andfeasibilityoreffectivenessofadjustmenttothesedriversisassessedusingarangeofspatialattributes.AnexampleofthecombinationofavarietyofdatalayersbyarelativelysimplemethodtoformasingleindexofpotentialproductivityfromgrazinginAustralianrangelandsisgiveninFigure2.7.Thiscouldrepresentasinglethemewithinafullecosystemservices’assessment.Thesethemesmaybecomparedbytwo-wayanalysis18,46ormaybecombinedandanoverallconditionassessedusingspiderdiagrams/radarplots.Somedetailedcontex-tualanalysisofthiskindofecosystemserviceassessmentincorporatingtheuseofspatialdataandspiderplotsisprovidedbyDefriesetal.48
2.7 ConClUsIons
Transformation of complex spatial and temporal patterns and behaviors intosimpledata forms isan importantenabling technicalcapability required inordertomaximize the informationcontentandeffectivenessofdigitaldecisionsupportsystems.Anessentialframeworkofanalyticalcapabilityinvolvesseamlessaccesstotoolsandmethodsforspatialanalysisandextractionofspatialpatternsandinter-relationships,toolsfortimeseriesanalysisonspatialdata,toolsfordevelopingsimplemodelsanddefiningrelationshipsandcausalityacrossspatialscales,andgenerationof uncertainty and error measures and incorporation in analysis along with basedata.AschemafortransformationofspatiotemporalcomplexityintoindicatorsforMCA is given in Figure2.8. The schema uses rangeland and urban examples toillustratethekindsofspecificmeasuresandderivedinformationrequired.Process
FIGURe 2.6 Aframeworkforapplicationofmulticriteriaanalysisofecosystemservicestotherangelandenvironment.
VegetationStructure, Growthforms, Foliage cover
Vegetation stateTypes I to VII
Transitions betweenvegetation states–also changes ES
Vegetationzones – atany scale
Land useBiodiversity
Water quality
Water yieldAgriculturalproduction
Timber yield
Amenity uses
Carbonsequestration
Landmanagementpractices
Ecosystem Service (ES)states with categoryattributesBiodiversityWater qualityWater quantityFood and FiberAmenityWood productsCarbon stock
Transitions betweenES states withinvegetation states
Land usechange
Land managementchange
Climatechange
Drivers of change in ESS and vegetation states
Attributes of drivers of change, e.g., profitability, cost, social impact, feasibility,likelihood, effectiveness, permanence, negative effects, etc.
Water usechange
Climatevariability
State andTransitionModel
State andTransitionModel
MCA to define ecosystemservice states and indexes
MCA to assessmerits of changestrategies andattributes ofoutcomes
42963_C002.indd 34 11/6/07 7:17:32 AM
Developing Spatially Dependent Procedures and Models 35
Temporal variation, e.g., climate; economic conditions
Space with gradients andobjects
e.g., landscape, city
Spatial relationshipsand patterns
e.g., terrain/soil;buildings/open space
Mobile agent makingdecisions
e.g., cattle/humans
Spatial relationshipsand patterns
e.g., feed and waterrequirement; play areas
Temporal dependenciesvariation in feed supplyprobability of demand > supplyfrequency of droughttravel timerecreation time
Spatial impactsgrazing/utilization pressuretrampling/trackingpre–condition for woodyingress ease of businessinteraction attractivenessfor tourists
Conceptual Basis for Transformation
Interrelation andcombination
Measures of system properties and processes – Indicators
Spatial constraints, e.g., distance from water or business clients and servicesTemporal constraints, e.g., time spent in location or decision/consulting time
Variation inavailable resources
function,appearance density,
range
texture,association
quantity,type
FIGURe 2.8 Schemafor transformationofspatiotemporal information into indicatorsofsystempropertiesformulticriteriaanalysisevaluations.
Potential productivity for grazingRainfall reliabilityRainfall rel. (w/sp)
Rainfall rel. (ann)
Forage potential
Growth Foliage proj cover Soil nutrient
Accessibility (ARIA)
Soil PSoil NSoil carbonNDVI meanNPP mean soil_fert
FIGURe 2.7 (See color insert following p. 132.) Cognitivemappinginterfacesuitableforcombinationofdiversespatiotemporalmetrics,indices,anddatalayersdescribingmean-ingfulpropertiesofasystemunderanalysis.Thisexampleshowstheconstructionofacom-positeindextorepresentpotentialgrazingproductivityfromrangelands.46,58
42963_C002.indd 35 11/6/07 7:17:34 AM
36 Land Use Change
andothercomplexmodelsoperateexternallytothisframeworkandprovidespatiallayersandtemporalsignalsrepresentinganintegrationofcomplexprocessesforusewithintheframework.
Theexampleusedinthischapterlargelyaddressesthebiophysicaldomainsincerangelandshavelowhumanpopulationsandindividualshavecontroloverlargelandareas.Thismaymean that sociological factors play a smaller relative role in themanagementofthesystemexceptwhenconcernedwithindigenousrightsandissues.However,theanalyticalproblemsareuniversal—forexample,thereisstillaneedtodeterminethelevelandimportanceofchangesinspousalworkcontributionstotheoperationofcattlestationsandtothefunctioningofisolatedfamilieseveniftheseeffectsareverysmall.However,throughtheprincipleofequivalenceofbiophysicalandsocial landscapeelements, cattledemographics, likehumandemographics incities,operateasamajordriverinthesystem,andassigningrulestobehavioranddefining spatial relationships and temporal trends and influences assume criticalimportance.Thischapterhasprovidedsomeexamplesofanalyticalmethodsand
(a)
(b)
FIGURe 2.9 VegetationtypesinAustralianrangelands.(a)Northernsavannawoodlandsmayhaveexcellentherbaceouscoverduetolargepaddocks,lowerstockdensityandreliableseasonalrainfall(PhotoM.J.Hill).(b)Saltbushplainsmaybeingoodcondition(asshownhere)butcanbesusceptibletodegradationwithdroughtsandovergrazing(PhotocourtesyofR.Lesslie).
42963_C002.indd 36 11/6/07 7:17:36 AM
Developing Spatially Dependent Procedures and Models 37
approaches needed. The case studies to follow will examine spatial methods ingreaterdetailforarangeofgeographicallydistinctsystemsandproblems.
2.8 ACknowleDGMents
TheMCASsoftwareinterfaceillustratedinthischapter(Figure2.7)hasbeendevel-opedbyRobLesslieasteamleader,AndrewBarryasprogrammer,andtheauthorastechnicaladvisor.Iamgratefultomycolleaguesforpermissiontoreproduceworkfromourjointeffortsinthissoleauthorchapter.IthankDamianBarrettforaccesstocontinentalcarboncyclemodeloutputsusedinFigures2.2and2.4.
ReFeRenCes
1. Grant,W.E.,Peterson,T.R.,andPeterson,M.J.Quantitativemodellingofcouplednatural/human systems: simulation of societal constraints on environmental actiondrawingonLuhmann’ssocialtheory.Ecological Modelling158,143–165,2002.
2. Jankowski,P.Integratinggeographicalinformationsystemsandmulti-criteriadecisionmaking methods. International Journal of Geographical Information Systems 9,251–273,1995.
3. Varma,V.K.,Ferguson,I.,andWild,I.Decisionsupportsystemforthesustainableforestmanagement.Forest Ecology and Management128,49–55,2000.
4. Store, R., and Jokimaki, J. A GIS-based multi-scale approach to habitat suitabilitymodelling.Ecological Modelling169,1–15,2003.
5. Mazzetto,F.,andBonera,R.MEACROS:Atoolformulti-criteriaevaluationofalter-nativecroppingsystems.European Journal of Agronomy18,379–387,2003.
6. Giupponi, C. et al. MULINO-DSS: A computer tool for sustainable use of waterresourcesatthecatchmentscale.Mathematics and Computers in Simulation64,13–24,2004.
7. Munda,G.Socialmulti-criteriaevaluationforurbansustainabilitypolicies.Land Use Policy23,86–94,2006.
8. Hill, M. J. et al. Multi-criteria decision analysis in spatial decision support: TheASSESSanalytichierarchyprocessandtheroleofquantitativemethodsandspatiallyexplicitanalysis.Environmental Modelling and Software20,955–976,2005.
9. Hill,M.J.etal.VegetationstatechangeandconsequentcarbondynamicsinsavannawoodlandsofAustraliainresponsetograzing,droughtandfire:Ascenarioapproachusing113yearsofsyntheticannualfireandgrasslandgrowth.Australian Journal of Botany53,715–739,2005.
10. Hill,M.J.etal.AnalysisofsoilcarbonoutcomesfrominteractionbetweenclimateandgrazingpressureinAustralianrangelandsusingRange-ASSESS.Environmental Modelling and Software21,779–801,2006.
11. Smith,S.W.The Scientist and Engineer’s Guide to Digital Signal Processing,2nded.CaliforniaTechnicalPublishing,SanDiego,Calif.643pp,1999.
12. Hamilton,J.D.Time Series Analysis. PrincetonUniversityPress,N.J.820pp,1994. 13. Gourieroux, C., and Monford, A. Time Series and Dynamic Models. Cambridge
UniversityPress,UK.668pp,2000. 14. Brockwell,P., andDavis,R. Introduction to Time Series and Forecasting.Springer
Verlag,Netherlands.434pp,2002. 15. Percival, D.B.,andWalden,A.T.Wavelet Methods for Time Series Analysis (WMTSA).
CambridgeUniversityPress,Cambridge.594pp,2000.
42963_C002.indd 37 11/6/07 7:17:37 AM
38 Land Use Change
16. Siebenhuner,B.,andBarth,V.Theroleofcomputermodellinginparticipatoryinte-gratedassessments.Environmental Impact Assessment Review25,391–409,2004.
17. Walker,J.etal.Assessment of Catchment Condition.CSIROLandandWater,Canberra.36pp,2002.
18. Hill,M.J.etal.Multi-criteriaassessmentof tensions in resourceuseatcontinentalscale:aproofofconceptwithAustralianrangelands.Environmental Management 37,712–731,2006.
19. Sexton,W.T.,Dull,C.W.,andSzaro,R.C.Implementingecosystemmanagement:Aframeworkforremotelysensedinformationatmultiplescales.Landscape and Urban Planning40,173–184,1998.
20. Osborne,P.E.,andSuarez-Seoane,S.Shoulddatabepartitionedspatiallybeforebuild-inglarge-scaledistributionmodels?Ecological Modelling157,249–259,2002.
21. Geoghegan,J.etal.ModellingtropicaldeforestationinthesouthernYucatanpeninsularregion:comparingsurveyandsatellitedata.Agriculture, Ecosystems and Environment85,25–46,2001.
22. Field, D. R. et al. Reaffirming social landscape analysis in landscape ecology:aconceptualframework.Society and Natural Resources16,349–361,2003.
23. Nelson,A.AnalysingdataacrossgeographicscalesinHonduras:Detectinglevelsoforganisationwithinsystems.Agriculture, Ecosystems and Environment85,107–131,2001.
24. Bousquet,F., andLePage,C.Multi-agent simulations and ecosystemmanagement:Areview.Ecological Modelling176,313–332,2004.
25. Loibl, W., and Toetzer, T. Modeling growth and densification processes in suburbanregions—simulationoflandscapetransitionwithspatialagents.Environmental Modelling and Software18,553–563,2003.
26. Evans,T.P.,andKelley,H.Multi-scaleanalysisofahouseholdlevelagent-basedmodeloflandcoverchange.Journal of Environmental Management72,57–72,2004.
27. Berger,T.Agent-basedspatialmodelsappliedtoagriculture:Asimulationtoolfortech-nologydiffusion, resourceusechangesandpolicyanalysis.Agricultural Economics25,245–260,2001.
28. Parker,D.C.,andMeretsky,V.Measuringpatternoutcomesinanagent-basedmodelofedge-effectexternalitiesusingspatialmetrics.Agriculture, Ecosystems and Environment101,233–250,2004.
29. Peralta, P., and Mather, P. An analysis of deforestation patterns in the extractivereservesofAcre,Amazoniafromsatelliteimagery:alandscapeecologicalapproach.International Journal of Remote Sensing21,2555–2570,2000.
30. Crews-Meyer,K.A.AgriculturallandscapechangeandstabilityinnortheastThailand:Historicalpatch-levelanalysis.Agriculture, Ecosystems and Environment101,155–169,2004.
31. Laney,R.Aprocess-ledapproachtomodellinglandusechangeinagriculturalland-scapes:AcasestudyfromMadagascar.Agriculture, Ecosystems and Environment 101,135–153,2004.
32. McConnell,W.J.,Sweeney,S.P.,andMulley,B.Physicalandsocialaccesstoland:Spatio-temporalpatternsofagriculturalexpansioninMadagascar.Agriculture, Ecosystems and Environment101,171–184,2004.
33. Cassel-Gintz,M.,andPetschel-Held,G.GIS-basedassessmentof the threat toworldforestsbypatternsonnon-sustainablecivilisation-natureinteraction.Journal of Environmental Management 59,279–298,2000.
34. Sui,D.Z.GIS-basedurbanmodelling:Practices,problemsandprospects.International Journal of Geographical Information Science12,651–671,1998.
35. Faust,K.etal.SpatialarrangementsofsocialandeconomicnetworksamongvillagesinNangRongDistrict,Thailand.Social Networks21,311–337,1999.
42963_C002.indd 38 11/6/07 7:17:38 AM
Developing Spatially Dependent Procedures and Models 39
36. Leenders, R. Th. A. J. Modeling social influence through network autocorrelation:Constructingtheweightmatrix.Social Networks24,21–47,2002.
37. Jackson, J. E., Lee, R. G., and Sommers, P. Monitoring the community impacts ofthe Northwest Forest Plan: An alternative to social indicators. Society and Natural Resources17,223–233,2004.
38. Martin,D.,andBracken,I.Theintegrationofsocio-economicandphysicalresourcedataforappliedlandmanagementinformationsystems.Applied Geography13,45–53,1993.
39. Martin, D. Automatic neighbourhood identification from population surfaces.Computers, Environment and Urban Systems22,107–120,1998.
40. Bonaiuto,M.,Fornara,F.,andBonnes,M.Indexesofperceivedresidentialenviron-mentqualityandneighbourhoodattachment inurbanenvironments:AconfirmationstudyonthecityofRome.Landscape and Urban Planning65,41–52,2003.
41. Krause,C.L.Ourvisuallandscape:managingthelandscapeunderspecialconsider-ationofvisualaspects. Landscape and Urban Planning54,239–254,2001.
42. Croissant,C.Landscapepatternsandparcelboundaries:Ananalysisofcompositionand configuration of land use and land cover in south-central Indiana. Agriculture, Ecosystems and Environment101,219–232,2004.
43. Reed,B.C.etal.Measuringphenologicalvariabilityfromsatelliteimagery.Journal of Vegetation Science 5,703–714,1994.
44. Li,Z.,andKafatos,M.InterannualvariabilityofvegetationintheUnitedStatesanditsrelationtoElNino/SouthernOscillation.Remote Sensing of Environment71,239–247,2000.
45. Zhang,M.Exploringtherelationshipbetweenurbanformandnonworktravelthroughtimeuseanalysis.Landscape and Urban Planning73,244–261,2005.
46. Hill,M.J.etal.Asimple,portable,multicriteriaanalysis tool fornatural resourcemanagement assessments. In: Zerger, A, and Argent, R. M., eds., MODSIM 2005, Proceedings of the International Symposium on Modelling and Simulation.UniversityofMelbourne,12–15December,2005,CDROM(availableat:http://www.mssanz.org.au/modsim05/authorsH-K.htm#h).
47. Thackway,R.,andLesslie,R.J.ReportingvegetationconditionusingtheVegetationAssets, States and Transitions (VAST) framework. Ecological Management and Restoration7,s53–s61,doi:10.1111/j1442-8903.2006.00292.x.,2006.
48. Defries,R.S.,Asner,G.P.,andHoughton,R.Trade-offsinlandusedecisions:Towardsa framework for assessing multiple ecosystem responses to land-use change. In:Defries,R.S.,Houghton,R.,andAsner,G.P.,eds.Ecosystems and Land Use Change.GeophysicalMonographSeries153,AmericanGeophysicalUnion,2004.
49. Platt, R. V. Global and local analysis of fragmentation in a mountain region ofColorado.Agriculture, Ecosystems and Environment101,207–218,2004.
50. Reggiani,A.,Nijkamp,P.,andSabella,E.Newadvancesinspatialnetworkmodelling:Towards evolutionary algorithms. European Journal of Operational Research 128,385–401,2001.
51. Nijkamp,P.,andReggiani,A.Non-linearevolutionofdynamicspatialsystems.Therelevance of chaos and ecologically-based models. Regional Science and Urban Economics 25,183–210,1995.
52. Roderick,M.L.,Noble,I.R.,andCridland,S.W.Estimatingwoodyandherbaceousvegetation cover from time series satellite observations. Global Ecology and Biogeography 8, 501–508,1999.
53. Lu,H.etal.DecompositionofvegetationcoverintowoodyandherbaceouscomponentsusingAVHRRNDVItimeseries.Remote Sensing of Environment86,1–18,2003.
54. Malingreau,J.-P.Globalvegetationdynamics:satelliteobservationsoverAsia.International Journal of Remote Sensing9,1121–1146,1986.
42963_C002.indd 39 11/6/07 7:17:38 AM
40 Land Use Change
55. Zhang,X.etal.MonitoringvegetationphenologyusingMODIS.Remote Sensing of Environment84,471–475,2003.
56. Moody, A., and Johnson, D. M. Land-surface phenologies from AVHRR using thediscretefouriertransform.Remote Sensing of Environment75,305–323,2001.
57. Devezas,T.,andCorredine,J.T.Thebiologicaldeterminantsoflongwavebehaviorin socio-economic growth and development. Technological Forecasting and Social Change68,1–57,2001.
58. Lesslie, R. et al. Towards Sustainability for Australia’s Rangelands. Analysing the Options.DepartmentofAgriculture,FisheriesandForestry,Australia,16pp,2006.
42963_C002.indd 40 11/6/07 7:17:39 AM
Part II
Comparative Regional Case Studies
42963_S002.indd 41 8/17/07 2:29:21 PM
42963_S002.indd 42 8/17/07 2:29:21 PM
43
3 Spatial Methodologies for Integrating Social and Biophysical Data at a Regional or Catchment Scale
Ian Byron and Robert Lesslie
Contents
3.1 Introduction...................................................................................................443.2 MappingChangeinLandUseataRegionalScale....................................... 45
3.2.1 WhyUnderstandingLandUseChangeIsImportant........................463.2.2 TypesofLandUseChange................................................................463.2.3 ChangesinLandUseintheLowerMurrayRegion..........................46
3.3 MappingCorrespondencebetweenBiophysicalDataandLandManagerPerceptions,ValuesandPractices..................................................483.3.1 IntegratingDataSources...................................................................48
3.3.1.1 TheGlenelgHopkinsLandholderSurvey............................483.3.1.2 SalinityDischargeSitesBasedonGroundwaterFlows
Systems.................................................................................483.3.1.3 LandUseCategorizedasConservationofNatural
Environment......................................................................... 493.3.2 SpatialMethodsforAssessingCorrespondenceinAssessments
andResponsestoSalinity..................................................................503.3.2.1 Context.................................................................................503.3.2.2 Approach..............................................................................503.3.2.3 Analysis................................................................................50
3.3.3 MappingtheRelationshipsbetweenAreasofHighConservationValueandLandManagers’ValuesandPractices....... 533.3.3.1 Context................................................................................. 533.3.3.2 Approach.............................................................................. 533.3.3.3 Analysis................................................................................54
3.4 InsightsandImplicationsfromIntegratingSocialandBiophysicalDataataRegionalScale...............................................................................54
42963_C003.indd 43 11/6/07 7:10:14 AM
44 Land Use Change
3.5 FutureDirections.......................................................................................... 593.6 Conclusions................................................................................................... 59References................................................................................................................60
3.1 IntroduCtIon
Catchment or watershed-based approaches to natural resource management andplanninghavebeenwidelyadopted inmanycountriesacross theglobe includingAustralia.1,2,3 These approaches seek to meld the benefits of building local com-munity engagement with the need to develop better integrated and coordinatedapproaches for addressing landscape-scale changes in the condition of land andwaterresources.4
In Australia, regional catchment management organizations now manage alargeproportionofnationalinvestmentinnaturalresourcemanagementthroughtheNaturalHeritageTrustandtheNationalActionPlanforSalinityandWaterQuality.TheNaturalHeritageTrustrepresentsAustralia’slargestinvestmentinenvironmentalmanagement.5Aspartof thedeliveryof funds,catchmentgroupsare required todevelopregionalplansthatsetouthowthenaturalresourcesoftheregionaretobemanaged.EachregionalplanistobeendorsedbystateandAustraliangovernmentagenciespriortotheirimplementation.Althoughtherearestateandregionaldiffer-ences,thesecatchmentgroupsaretypicallyaskedto:
Describetheircatchmentconditionintermsofenvironmental,economic,andsocialassetsIdentifythedesiredfutureconditionofthoseassetsIdentify the key processes that might mitigate the achievement of thedesiredconditionsIdentify management actions and targets that will help achieve desiredconditionsMonitorandevaluateprogress
Clearlytheserolesrequirecatchmentgroupstobeabletounderstandthedriversandbarriersaffectinglandmanagersandunderstandtheimpactsoflandmanagementpracticesonkeyregionalassets.Unfortunately,thereareverylimiteddataavailablethathavebeendesignedwiththesepurposesinmind.AlthoughmostregionalgroupsinAustraliahaveaccesstoarangeofbiophysicaldatasources,veryfewhaveaccesstodetailedsocialdataspecifictotheirregion.Endter-Wadaetal.6assertedthatnatu-ralscientistshavebeenreluctanttoincludesocialsciencedimensionsinecosystemassessments.Atthesametime,Brown7suggestedthatinpartthelackofsocialdataincorporatedinlandscapeplanningreflectsanabsenceofsystematicapproachesforcollectingandanalyzingthisinformationwithbiophysicaldata.
Nevertheless,thereisincreasingrecognitionoftheneedtointegratesocialandbiophysical data to achieve improved natural resource management outcomes. AreviewconductedaspartofAustralia’sNationalLandandWaterResourcesAuditconcludedthat thereisastrongneedforapproachesthat integratesocioeconomicandbiophysicaldataataregionalscale.8Similarly,Endter-Wadaetal.6concluded
•
••
•
•
42963_C003.indd 44 11/6/07 7:10:15 AM
Spatial Methodologies for Integrating Social and Biophysical Data 45
thateffectiveecosystemmanagement ispredicatedonbringing togetherscientificanalysisofsocialfactorsandbiophysicalfactors.
Therelativelyrecentemergenceofgeographicalinformationsystems(GIS)hasprovidedanimportantsetoftoolstofacilitateinterdisciplinaryresearchthatinte-gratessocialandbiophysicaldataatalandscapescale.9Inrecentyearstherehavebeenanumberofimportantstudiesusingcensusdatatolinkchangesinsocialstructurewithecologicalfactorsandvisaversa.9,10,11Althoughthesestudieshaveclearlyhigh-lightedhowintegratingsocialandbiophysicaldataiscriticaltoimprovingnaturalresource management outcomes, they also acknowledge significant limitationswithnationallycollectedcensusdata.Inparticular,censusdataareonlyavailableataggregatelevelsthatrequireresearcherstoassumethatthesocialvariablesarehomogenousacrossthesmallestcensusunit(typically200households).9
Inadditiontoconcernsaboutspatialresolution,Endter-Wadaetal.6suggestthatwhileimportant,understandingdemographictrendsaloneisinsufficientforunder-standingcomplexsocialsystemsandtheirrelationshiptoresourceconditionsanddynamics.Insummarizingthepotentialcontributionsofsocialsciencetoecosystemmanagement,Endter-Wadaet al.6 concluded thatunderstanding spatialvariabilityin resourceneeds,values,anduseswascriticalbuthighlighteda lackofsystem-aticdataanalysisrequiredtomovebeyondtherhetorictotherealityofintegratinghuman values in ecosystem management. According to Grove et al.,11 exploringquestionsabouthowmotivationsandcapacitiesinfluenceandareinfluencedbythebiophysicalenvironmentwillbebestexploredbyadaptingtraditionalsocialsciencefieldmethodsthathavebeenappliedtonaturalresourcemanagement.
Althoughnumerousresearchershaveintegratednationallycollectedcensusdatainto landscapeanalyses, therearevery fewexamplesofattempts topurposefullycollectsocialdatathatcanbeintegratedwithspecificbiophysicaldatalayers.Brown7providessomeinsightsintotheapplicationoftheseapproachesasdoesearlierworkbyCurtis,Byron,andMcDonald12andCurtis,Byron,andMacKay,13uponwhichthischapteraimstoextend.
Thischapterdrawsonfindingsfromspatiallyreferencedsurveysoflandmanagerstohighlightmethodologiesforintegratingsocialandbiophysicaldataataregionalorcatchmentscale.Specific issuesandapproachescovered includemapping landusechangeandexploringtheextentandnatureoflinksbetweenmappedbiophysicalresourceconditionsandlandmanagerperceptions,values,andpractices.
3.2 MAPPInG CHAnGe In LAnd use At A reGIonAL sCALe
3.2.1 Why Understanding Land Use Change is important
Acapacityfordetectingandreportinglandusechangeiscriticaltoevaluatingandmonitoring trends in natural resource conditions and the effectiveness of publicinvestment innatural resourcemanagement.Australia’sNationalLandandWaterResourcesAudit,forexample,hasidentifiedthereportingofchangeovertimeandtheintegrationoflanduseinformationwithothernaturalresourceinformationasakeytoeffectivelyaddressingmajorsustainabilityproblemssuchassalinity,waterquality,andsoilloss.
42963_C003.indd 45 11/6/07 7:10:15 AM
46 Land Use Change
3.2.2 types of Land Use Change
Aparticulardifficultywithlandusechangereportingisdiscriminatingthediffer-entdimensionsofchange.Protocolsforreportinglandusechangeinanagriculturalcontext,forexample,shouldbecapableofdistinguishingthetemporalcharacteris-ticsoffarmingsystems(e.g.,rotations),seasonalvariability,andlonger-termindus-tryandregionaltrends.Lesslie,Barson,andSmith14identifyfourbroadapproachestomeasuringandreportinglandusechange:
1.Arealchange:lossorgaininthearealextent.Thisprovidesanindicationofwhethertargetlandusesareincreasingordecreasinginareaovertime.Changescanbepresentedstatistically,graphically,orspatiallyandidenti-fiedchangescomparedandtrendsobserved.
2.Transformation:thepatternoftransitionfromonelandusetoanother.Forexample,anareamaybecroppedoneyear,grazedthenextyear,andthencroppedagaintheyearafter.Alternatively,landunderimprovedpasturefordairymaybepermanentlyconvertedtovineyards.15Landusetransforma-tionsbetweentimeperiodsmaybeexpressedusingachangematrix.
3.Dynamics:ratesofchangeandperiodicityinarealextentortransforma-tions.Thetemporalnatureofchangemaybefurtherexploredbyanalyzingwhetherratesofchangeareincreasingordecreasing,arelong-orshort-termtrends,orcyclic(forexample,changesasaresultofdifferencesingrow-ingseasons,structuraladjustment,farmingsystems,orrotationregimes).Thismayrevealkeytrendsinlanduseandlandmanagementnotevidentinexpressionsofsimplearealchangeortransformations.Successfulanaly-sisoflandusedynamicsrequiresconsistent,high-qualitytime-seriesdata.Oftenitisnotpossibletoobtainsufficientlyconsistentdataoverconsecu-tiveyearsorseasons.
4.Prediction: modeling spatial or temporal patterns of change. The useofmodels topredictpast,present,andfuture landusesbasedoncertainrules,relationships,andinputdatamayhelpidentifykeydriversoflandusechange,implementscenarioplanning,andfillgapsindataavailability.
3.2.3 Changes in Land Use in the LoWer mUrray region
Thecapacitytoreportchangealsodependsontheavailabilityofconsistenttime-seriesdatacapableofprovidinginsightsintorelevantaspectsofchange.Wherefine-scaledtime-seriesdataareavailable,spatialanalysiscanprovideimportantinsightintothenatureoflandusechange.
Using time-series data from fine-scaled mapping based on orthophoto inter-pretationanddetailedpropertysurveys,itispossibletohighlightspatialtrendsinlandusepatterns.Forexample,Figure3.1showstrendsintheexpansionofirrigatedhorticulturearoundthetownsofRenmark,Berri,andLoxtonintheLowerMurrayregionofsoutheasternAustraliaproducedbytheAustralianCollaborativeLandUseMappingProgram (ACLUMP), a partnershipofAustralian and state governmentagenciesproducingcoordinatedlandusemappingforAustralia.14Thistime-series,1990to2003,isdrawnfrom1:25,000catchment-scalelandusemappingcompleted
42963_C003.indd 46 11/6/07 7:10:16 AM
Spatial Methodologies for Integrating Social and Biophysical Data 47
usingorthophotointerpretationanddetailedpropertysurveys.16Themappingrevealsapatternof landusetransformationandintensificationfromdrylandcerealcrop-pingandgrazingtoirrigatedhorticulture,andatrendfromsmall-scaleenterprisesclusteringaroundtownareastodispersed,large-scaleestablishmentsatincreasingdistancesfromirrigationwatersupply(rivers).
Time-series land use mapping at catchment-scale in Australia is producedby ACLUMP to agreed to national standards, facilitating its use in national andregionalnaturalresourceassessments.Themappingprocessinvolvesstagesofdatacollation, interpretation, verification, independent validation, quality assurance,andtheproductionoflandusedataandmetadata.Thisincludescollectingexistinglanduse informationandcompiling it intoadigitaldata setusingaGIS. Impor-tantinformationsourcesincluderemotelysensedinformation,landparcelboundaryinformation,forestandreserveestatemapping,landcover,localgovernmentzoning
Statistical Local AreasIrrigated horticulture first mapped prior to 1990 (mapped in 1988 for SA)Irrigated horticulture first mapped in 1995 and between 1990–1995Irrigated horticulture first mapped between 1995–1999Irrigated horticulture first mapped in 2001Irrigated horticulture first mapped in 2003
Irrigated horticulture data provided bySA Department of Environment and Heritage
Loxton
BerriBarmera
Renmark
FIGure 3.1 (See color insert following p. 132.) LandusechangeintheBarmera,Berri,andRenmarkareasofSouthAustralia.
42963_C003.indd 47 11/6/07 7:10:18 AM
48 Land Use Change
information, other land management data, and information collected in the field.Agreed-tostandards includeattribution toanationalclassification, theAustralianLandUseandManagement(ALUM)Classification.14Fine-scaleddataofthetypeillustratedinFigure3.1are,however,expensivetoproduceandarepresentlyoflim-ited availability. More cost-effective methods for wider application are presentlyunderdevelopment.
3.3 MAPPInG CorresPondenCe BetWeen BIoPHYsICAL dAtA And LAnd MAnAGer PerCePtIons, VALues And PrACtICes
3.3.1 integrating data soUrCes
The approachoutlined in this chapter used aGIS to integrate social survey datacollectedintheGlenelgHopkinsregionwithbiophysicaldata.TheGlenelgHopkinsregionislocatedintheStateofVictoriainthesoutheastofAustralia.Theregioncovers an area of approximately 26,000 square kilometres or approximately 11%ofthestate17(Figure3.2).Agriculturerepresentsamajorcontributortotheregionaleconomy,andin1999to2000itwasworthapproximatelyAU$650millionorapprox-imately10%ofthegrossvalueofagriculturalproductionintheStateofVictoria.
Thethreemajordatalayersusedinthischapterare:
1.Aspatiallyreferencedsurveyofrurallandholders18
2.Amapofsalinitydischargebasedonthegroundwaterflowsystems19
3.Land use categorized as Conservation of Natural Environment (Class 1)undertheALUMclassificationsystem20
3.3.1.1 the Glenelg Hopkins Landholder survey
In2003theBureauofRuralSciencesandGlenelgHopkinsCatchmentManagementAuthorityconductedasurveyofapproximately1,900rurallandholdersfromacrossthe region.18The survey focusedongatheringbaseline information regarding thekey social and economic factors affecting landholder decision making about theadoptionofpracticesexpectedtoimprovethemanagementofnaturalresourcesinthe Glenelg Hopkins region. The survey was sent to a random selection of ruralpropertyowners,withpropertiesover10hectaresinsize,identifiedthroughlocalrate payer databases. A final response rate of 64% was achieved for this survey.Allsurveydata(some250variables)wereenteredintoageographicalinformationsystem(ArcViewGIS)andassignedtoapropertycentroidusingxandycoordinatesincludedintheratepayerdatabases.
3.3.1.2 salinity discharge sites Based on Groundwater Flows systems
ThemapofsalinitydischargesitesintheGlenelgHopkinsregionwasundertakenaspartofthegroundwaterflowsystemsprojectconductedbyDahlhaus,Heislers,andDyson.19ThegroundwaterflowsystemsweredevelopedbytheNationalLandand Water Resources Audit as a framework for dryland salinity management in
42963_C003.indd 48 11/6/07 7:10:18 AM
Spatial Methodologies for Integrating Social and Biophysical Data 49
Australia.21Thisworkcategorizeslandscapesbasedonsimilaritiesingroundwaterprocesses,salinityissues,andmanagementoptions.Dahlhaus,Heislers,andDyson19stated thatwhilegroundwaterflowsystemsareauseful tool inhelping tounder-standsalinity,therehasbeenlittlescientificvalidationoftheflowsystemsorsalinityprocessesintheGlenelgHopkinsregion.
3.3.1.3 Land use Categorized as Conservation of natural environment
LandusemappingfortheGlenelgHopkinsregionintheStateofVictoriawasunder-takenusingathree-stageprocess.20Thefirststageofmappinginvolvedthecolla-tionofexistinglanduseinformation,remotelysensedinformation(satelliteimageryandaerialphotography), andcadastre.Other important information sourceswerereserveestatedata,landcover,localgovernmentzoninginformation,andotherlandmanagementdata.ThesecondstageinthemappingprocessinvolvedinterpretationandassignmentoflanduseclassesaccordingtotheALUMclassificationtocreateaninitialdraftlandusemap.Thefinalstagesofmappingincludedfieldverification,theeditingofdraftlandusemaps,andvalidation.Themappingisdatedat2001andisproducedatscalesof1:25,000and1:100,000.
FIGure 3.2 LocationoftheGlenelgHopkinsregion.
V I C T O R I A
TownsSurvey AreaSLA BoundaryMain Road
Legend
Survey Area
N
BALMORAL
COLERAINECASTERTON
MERINO
DARTMOOR
HEYWOOD MACARTHUR
PORTLAND
Km150100500
PORT FAIRYWARRNAMBOOL
KOROITBUSHFIELD–WOODFORD
ALLANSFORD
TERANGNOORAT
MORTLAKE
PENSHURST
DUNKELD
WILLAURA
ARARAT
BEAUFORT
SKIPTONSNAKE VALLEY
LEARMONTHMINERS REST
HAMILTON
42963_C003.indd 49 11/6/07 7:10:20 AM
50 Land Use Change
3.3.2 spatiaL methods for assessing CorrespondenCe in assessments and responses to saLinity
3.3.2.1 Context
TheGlenelgHopkinsregionisoneof21priorityregionsidentifiedundertheNationalActionPlanforSalinityandWaterQualityasbeingaffectedbysalinityandwaterqualityproblems.TheGlenelgHopkinsSalinityPlan22identifiedheavyimpactsofsalinityonagriculture,theenvironment,andinfrastructurewithanestimatedcosttotheregionofoverAU$44millionannually.
3.3.2.2 Approach
The land manager survey included a question that asked respondents to indicateif they had any areas of salinity on their property. By assigning land managers’responsestothepointdatacontainingpropertycentroidsforeachpropertysurveyed,itispossibletoexploretheextentthatlandmanagerperceptionsarespatiallylinkedtomappedsalinitydischargesitesusingthegroundwaterflowsystems(representedaspolygons).
Asdatafromthelandmanagersurveycouldonlybejoinedtoapointfilebasedonapropertycentroid,anymeasureofdirectcorrespondencewouldfailtoallowfordifferencesinpropertysizeandshape.Althoughthereisawiderangeoftechniquesavailablefor interpolatingcontinuoussurfacesfrompointdata, theextent thatanychangeinsocialvariablescanbepredictedbyalgorithmsbasedonaspatialrelation-shipisquestionable,particularlywherethepointsaredispersedacrossalargearea.23
Forthesereasons,nearestneighboranalysis24wasusedtoidentifythedistanceto the closest edge of the nearest mapped salinity discharge site for each surveyrespondent.Thesedistancescanthenbecomparedforrespondentswhosaidtheyhadsalinityontheirpropertyandthosewhodidnotoracrossarangeofothervariables.
Althoughinterpolatingsurfacesfromthelandmanagerpointdataisproblem-atic, creating a raster-based surface of distance from any grid cell to the nearestsalinitydischargesiteprovidesaquickvisualdisplaythatcanbeoverlayedwiththelandmanagerperceptionsandsalinitydischargelayers(Figure3.3).
3.3.2.3 Analysis
Theresultsof thenearestneighboranalysisclearlyshow that landholders incloseproximitytomappedsalinitydischargesitesweresignificantlymorelikelytoidentifyareasofsalinityontheirproperty(Table3.1).Withoverhalfofallrespondentswithin0.5kmofadischargesiteidentifyingsalinityontheirproperty,applyingthismethod-ologyalsosuggeststhatmostlandholdersareawareofsalinityontheirproperty.
By adopting the nearest neighbor technique it is also possible to explore theextentthatlandmanagersclosertomappedsalinitydischargesitesaremorelikelytobeconcernedabouttheimpactsofsalinityandundertakingpracticesexpectedtohelpmitigatesalinity(Table3.2).
Althoughmostrespondentsclosetomappedsalinitydischargesitesappeartobeawareoftheissue,therewerestillalargenumberofrespondentsnearmapped
42963_C003.indd 50 11/6/07 7:10:20 AM
Spatial Methodologies for Integrating Social and Biophysical Data 51
FIGure 3.3 (See color insert following p. 132.) Landmanagers’perceptionofsalinityandmappedsalinitydischargesites.
42963_C003.indd 51 11/6/07 7:10:21 AM
52 Land Use Change
tABLe 3.1Land Managers’ Perception of salinity and distance to Mapped salinity discharge
distance to nearest mapped salinity discharge site (m)
Land manager identified salinity (%)
Yes no
0–499 61 39
500–999 47 53
1,000–1,999 35 65
2,000–2,999 27 73
3,000–3,999 31 69
4,000–4,999 17 83
Over5,000 11 89
tABLe 3.2distance to Mapped salinity discharge and Land Manager Attitudes and Practices
Land manager attitudes and practicesMedian distance to nearest
mapped salinity discharge site (m)
Concern about salinity reducing productive capacity of their propertya
High 1,639
Moderate 2,028
Low 3,092
Concern about salinity reducing productive capacity of the local areaa
High 1,951
Moderate 2,386
Low 3,202
Concern about salinity reducing water quality in the local areaa
High 1,824
Moderate 2,262
Low 3,216
Planted trees or shrubsa
Yes 2,493
No 3,498
Established deep rooted perennial pastureb
Yes 2,764
No 2,648a Differencewasstatisticallysignificant(p<0.05)usingtheKruskal-Wallisnon-
parametricchi-squaretest.b Differencewasnotstatisticallysignificant(p>0.05)usingtheKruskal-Wallis
nonparametricchi-squaretest.
42963_C003.indd 52 11/6/07 7:10:22 AM
Spatial Methodologies for Integrating Social and Biophysical Data 53
salinitythatappeartobeunawareoftheproblem.Forthepurposesofthisexamplewehaveassumedthatlandholderswithin0.5kilometerofadischargesitethathavenotidentifiedsalinityontheirpropertyareunawareoftheproblem.
Thespatialidentificationofrespondentswhoappeartobeunawareofsalinityontheirpropertyprovidesanimportantopportunitytoidentifykeycharacteristicsofthisgroupofrespondentsandthusdevelopbettertargetedcommunityengagementstrategies.Forexample,Table3.3clearlyhighlightsadistinctivesetofcharacteris-ticsoflandholdersthatappeartobeunawareofsalinityontheirproperty.
3.3.3 mapping the reLationships betWeen areas of high Conservation vaLUe and Land managers’ vaLUes and praCtiCes
3.3.3.1 Context
Akey aim fornatural resourcemanagement in theGlenelgHopkins region is tomaintainandenhanceremnantnativevegetation.TheGlenelgHopkinsregionhasanextensivehistoryoflandclearing,andnativevegetationnowcoverslessthan13%oftheregion,with8%inparksandreservesfragmentedacrosstheregion.
3.3.3.2 Approach
ThecombinationofdatacollectedthroughtheregionallandholdersurveywithlandusemappingdatafortheGlenelgHopkinsregionprovidesanopportunitytoidentifythoselandmanagerswhoaremostlikelytohaveanimpactonareasofhighconser-vationvalue.
Nearestneighboranalysisforeachsurveyrespondent(representedasapropertycentroid)tothenearestedgeofanareaclassifiedashavinghighconservationvalue(polygon)canbeusedtohelpidentifykeygroupsoflandmanagers.Oncethedis-tancetothenearestareaofhighconservationvaluehasbeencomputed,standard
tABLe 3.3differences between Land Managers Who Were Aware of salinity and those unaware
Characteristics of land managersa
Land managers within 500 m of salinity
Aware unaware
Primaryoccupationfarming 85% 56%
MemberofaLandcaregroup 70% 37%
Completedatrainingcourserelatedtopropertymanagement
69% 25%
Hadworkundertakenontheirpropertythatwasatleastpartiallyfundedbygovernment
53% 19%
Plantedtreesandshrubs 85% 67%
Establishedperennialpasture 73% 44%
Medianpropertysize 465ha 136haa Alldifferenceswerestatisticallysignificant(p<0.05)usingthePearsonchi-squaretest.
42963_C003.indd 53 11/6/07 7:10:22 AM
54 Land Use Change
statisticalanalysescanbeusedtohelpdiscernpatternsbetweenspatialproximityandkeycharacteristicsoflandmanagersandtheirproperty.Asoutlinedpreviously,araster-basedsurfaceofdistancefromanygridcelltothenearestareaofhighcon-servationvalueprovidesausefuloverlaytohelpgraphicallyrepresentrelationships(Figure3.4).
3.3.3.3 Analysis
WhenappliedtosurveyandlandusedatafromtheGlenelgHopkinsregiontheseanalyses show some very clear differences in the characteristics of land manag-ersandtheirpropertybasedontheirproximitytoareasofhighconservationvalue(Table3.4).
Applyingthesametechniqueitisalsopossibletoexploreifthevalueslandholdersattachtotheirpropertyintermsofthesocial,economic,andenvironmentalbenefitsarelinkedspatiallytoareasofhighconservationvalue.Theseanalysesshowthatrespondentswhosaidbeingclosetonaturewasanimportantvalueoftheirpropertywereinfactsignificantlyclosertoareasofhighconservationvalue,whileinturnthosewhosaidprovidinghouseholdincomewasimportantweresignificantlyfurtherfromtheseareas.However,itisinterestingtonotethatthepropertyprovidingthesortoflifestyledesiredwasnotlinkedtotheproximitytoareasofhighconservationvalue(Table3.4).
Finally,surveydataandlandusemappingdatacanbecomparedtoseeiflandmanagersnearareasofhighconservationaremorelikelytohaveadoptedmanage-mentpracticesaimedatimprovingbiodiversity.Theseanalysesshowmixedresults.Landmanagersclosertoareasofhighconservationvalueweresignificantlymorelikelytohaveencouragedregrowthofnativevegetationandfencedoffareasofnativevegetationontheirproperty.However,theserespondentswerealsosignificantlylesslikelytohaveplantednativetreesandshrubsontheirproperty(Table3.4).
3.4 InsIGHts And IMPLICAtIons FroM InteGrAtInG soCIAL And BIoPHYsICAL dAtA At A reGIonAL sCALe
Theuseofspatialmethodologiesforintegratingsocialandbiophysicaldata,asdem-onstratedinthischapter,hassomeimportantinsightsandimplicationsforeffortstoimprovenaturalresourcemanagementoutcomes.
In thefirst instance, these approacheshighlighted anumberof importantdis-crepanciesbetweenrespondents’assessmentsofsalinityandthosemadeusingthegroundwaterflowsystem.Overa thirdofall respondentswithin500metersofamappedsalinitydischargesitedidnotidentifyareasofsalinityontheirproperty,andoverhalfofallthelandmanagerswhoidentifiedareasofsalinitydidnotcorrespondwithmappedsalinitydischargesites.ManypartsoftheGlenelgHopkinsregionarecharacterizedbyrelativelyhighrainfall,anditispossiblethatsomelandholdershavemistakenwaterloggingasasignofsalinity. It isalsopossible thatsmall localizedareasofsalinityidentifiedbylandholderscouldbemissedthroughinterpretationoflarge-scale aerial photographs. Similarly, the identification of salt indictor speciesthrough the interpretationofaerialphotographsaspartof themappingofsalinity
42963_C003.indd 54 11/6/07 7:10:22 AM
Spatial Methodologies for Integrating Social and Biophysical Data 55
N
S
W E
Kilometers
0 – 0.0124435330.012443533 – 0.0248870660.024887066 – 0.0373305980.037330598 – 0.0497741310.049774131 – 0.0622176640.062217664 – 0.0746611970.074661197 – 0.0871047300.087104730 – 0.0995482620.099548262 – 0.1119917950.111991795 – 0.124435328
Glenelg Hopkins regionFarmerNon–farmerConservation of natural environments
Std. Dev = 1595.02Mean = 1951N = 640.00
Std. Dev = 1249.48Mean = 1276N = 355.00
8000 – 8500
7000 – 7500
6000 – 6500
5000 – 5500
4000 – 4500
3000 – 3500
2000 – 2500
1000 – 1500
0 – 500
8000 – 8500
7000 – 7500
6000 – 6500
5000 – 5500
4000 – 4500
3000 – 3500
2000 – 2500
1000 – 1500
0 – 500
Distance to nearest area of high conservation value (meters)
Distance to nearest area of high conservation value (meters)
Non–farmer
Farmer
Freq
uenc
y
0 30 60 120
Freq
uenc
y
140
120
100
80
60
40
20
0
140
120
100
80
60
40
20
0
Distance to nearest area of high conser vation value<VALUE>
Legend
FIGure 3.4 (See color insert following p. 132.) Landmanagerswhomanagepropertiesnearareasofhighconservationvalue.
42963_C003.indd 55 11/6/07 7:10:42 AM
56 Land Use Change
tABLe 3.4distance to nearest Area of High Conservation Value and Land Managers’ Characteristics, Values, and Practices
Land managers’ characteristics, values, and practicesMedian distance to nearest area of
high conservation value (m)
Primary occupationa
Farmer 1,614
Non-farmer 794
Property sizea
Small(<100ha) 829
Medium(100–499ha) 1,425
Large(500haandover) 1,642
Landcare membershipa
Yes 1,063
No 1,600
Completed a short course related to property managementa
Yes 1,429
No 1,129
Had work undertaken on their property that was at least partially funded by governmenta
Yes 1,512
No 1,152
Value attached to property in providing an attractive place to liveb
High 1,276
Moderate 1,470
Low 969
Value attached to property in providing habitat for native animalsa
High 1,074
Moderate 1,357
Low 1,479
Value attached to property in providing majority of household incomea
High 1,616
Moderate 1,012
Low 794
Planted tree and shrubsa
Yes 1,470
No 881
42963_C003.indd 56 11/6/07 7:10:43 AM
Spatial Methodologies for Integrating Social and Biophysical Data 57
dischargesitescouldalsobeconfoundedbyhighrainfallandatendencyforwater-logging.Incorporatingdataonrainfallandrainfallreliabilityacrosstheregionisoneoptionthatmayhelpfurtheridentifyreasonsforconflictingperceptionsofsalinity.
Totheextentthatweassumethatthesalinitydischargemapshavecorrectlyiden-tifiedareasofsalinity,theintegrationofspatiallyreferencedsurveydataprovidesimportantinsightsforthedevelopmentoftargetedmanagementstrategies.Throughthis process it is possible to spatially locate not only areas affected by salinitybut areas within those, where the land managers are unaware of, or not activelymanaging,theproblem.Takeninisolationeitherthemapsofsalinitydischargeorlandholderidentifiedsalinityareinadequatefortargetedstrategiestomitigatesalin-ity.Forexample, ifmost landholders inaregionarenotaffectedordirectlycon-tributingtosalinity,thecapacityoflandmanagersatanaggregatescaletomodifyland management practices is largely irrelevant. That is, rather than undertakingbroadoutreachactivitiesaboutsalinity,themethodspresentedinthischapterallowregionalcatchmentmanagerstodevelopmorestrategicinvestments.
Similarly, linking remnant areasofvegetation through targeted investment isanintegralpartofplanstoimprovebiodiversityoutcomesintheGlenelgHopkinsregion(Figure3.5).Theintegrationoflandusemappingofareasofhighconserva-tionvaluewithspatiallyreferencedlandholderinformationclearlyhighlightedthatlandmanagersnearareasofhighconservationareaswerequitedifferentandappeartopreferentiallypurchasepropertieswithhighamenityvalue.
Thefactthatlandmanagerswhowereunawareofsalinityandthosenearareasofhighconservationvaluetendedtobenonfarmers,ownsmallerproperties,werelesslikelytobeinvolvedintheLandcareprogramorhavecompletedacourserelatedtopropertymanagementhaveimportantimplications.Thesmallnumbersoflargefamily farming operations that manage much of the land area of Australia havebeen a key target of policies and programs aimed at improving natural resourcemanagement.Althoughtheseprogramsandpolicieshavebeenlargelysuccessful,as
tABLe 3.4 (continued)distance to nearest Area of High Conservation Value and Land Managers’ Characteristics, Values, and Practices
Land managers’ characteristics, values, and practicesMedian distance to nearest area of
high conservation value (m)
Encouraged regrowth of native vegetationa
Yes 1,122
No 1,352
Fenced areas of native busha
Yes 856
No 1,357a Difference was statistically significant (p < 0.05) using the Kruskal–Wallis non-parametric
chi-squaretest.b Differencewasnotstatisticallysignificant(p>0.05)usingtheKruskal–Wallisnon-parametric
chi-squaretest.
42963_C003.indd 57 11/6/07 7:10:43 AM
58 Land Use Change
(a)
(b)
(c)
FIGure 3.5 (a) Pastures, (b) rolling country-side and (c) canola cropping in theGlenelgHopkinsregion.
42963_C003.indd 58 11/6/07 7:10:45 AM
Spatial Methodologies for Integrating Social and Biophysical Data 59
inthecaseofflagshipprogramssuchasLandcare,thereisincreasingevidenceofagrowingnumberofrurallifestylemotivated“tree-change”orsubcommerciallandholdings.25There isaclearneedforregionalcatchmentmanagers tobetter targetandengagetheselandmanagerswhoarelikelytooperateoutsidetheirnormalinfor-mationchannelsandnetworks,particularlytotheextentthattheymanagecriticalnaturalresources,asisthecaseintheGlenelgHopkinsregion.
3.5 Future dIreCtIons
Theexamplespresented in thischapter linking spatially referenced social surveydatawithbiophysicaldataata regional scaleprovidesonlyafirstglimpseat thepotentialoftheseapproaches.Otherapplicationsmayincludeexploringon-propertyprofitabilityagainstrainfallreliability,croppingtechniquesagainstsoilerosionandturbiditylevels,ormanagementofriparianzonesandwaterquality.
Moresophisticated techniquesformappingsalt,suchasairborneelectromag-netics(AEM),alsoprovideanopportunity tomoreprecisely identify locations inacatchmentwherevariousmanagementactions (suchasplanting trees)willhelpmitigate salinity.AEMprovides a three-dimensionalmodel of salt bymeasuringtheelectricalconductivityofthegroundatdifferentdepths.26Furthermore,arecentstudybyBakerandEvans27usingAEMfoundthattreeplantinginareaspreviouslythoughttobebeneficialmayactuallycontributetosalinitybyreducingfreshwaterflows.ThecombinationofAEMdatawithspatiallyreferencedsurveydataappearslikely to hold much promise in allowing better targeted and more site-specificapproachesformanagingdrylandsalinity.
Generally,thecostoflandusechangedetectionandreportingusingfine-scaledtime-seriesdatabasedonorthophotointerpretationanddetailedpropertysurveys,asoutlinedinthischapter,isprohibitiveandmorecost-effectivemethodsareneeded.The capacity to adequately characterize change is highly dependent on matchingthespatiotemporalaccuracyandprecisionofavailabledatatotherelevantlandusedynamic.Thecouplingof time-seriessatellite imagery toregularlycollectedagri-culturalstatisticspresentsonepracticalandwidelyapplicableapproachtomappingagriculturallandusechange.Aprocedureadoptedforregional-scalelandusemappinginAustraliabyACLUMPusingagriculturalcensusandsurveydata,AdvancedVeryHighResolutionRadiometer(AVHRR)imagery,andastatisticalspatialallocationprocedureispromisingforthedevelopmentofannualtime-seriesanalyses,providingthatlimitationsinspatialaccuracycanbeaccommodated.28,29Thereisscopetooforimprovementifhigherresolutionsatelliteimagery(e.g.,ModerateResolutionImagingSpectroradiometer [MODIS]) can be applied successfully. ACLUMP partners arecurrentlypursinginvestigationsinthisarea.
3.6 ConCLusIons
Thischapterhaspresentedarangeofsimpleandpracticalapproachesforintegratingspatiallyreferencedsocialdatawithbiophysicaldatalayers.Furthermore,bypre-sentingresultsfromcasestudiesdemonstratingtheapplicationofthesetechniques,
42963_C003.indd 59 11/6/07 7:10:46 AM
60 Land Use Change
hopefullywehavehelpedshowhowtheseapproachescancontributetoimprovedmanagementoutcomesatalandscapescale.
Thespatialmethodologiesoutlined in thischapterarederivedfromourearlyattemptsatcreatingmeaningfulintegrationbetweensocialandbiophysicalsciences,andbarelyscratchthesurfaceofthemanypotentialapplicationsofthistechnique.Inparticular,theuseoflongitudinalsurveydatacombinedwithlongitudinallandusechangedataislikelytohelpmoredefinitivelyunderstandtheinteractionsbetweenlinkedsocialandbiophysicalsystems.
reFerenCes
1. Kenney,D.S.Arecommunity-basedwatershedgroupsreallyeffective?:Confrontingthethornyissueofmeasuringsuccess.ChronicleofCommunity3(2),33–38,1999.
2. McGinnis,M.V.,Woolley,J.,andGamman,J.Bioregionalconflictresolution:Rebuild-ing community in watershed planning and organizing. Environmental Management24(1),1–12,1999.
3. Curtis,A.,Shindler,B.,andWright,A.Sustaininglocalwatershedinitiatives:Lessonsfrom Landcare and watershed councils. Journal of the American Water ResourcesAssociation38(5),1–9,2002.
4. Dale, A., and Bellamy, J. Regional Resource Use Planning in Rangelands: AnAustralianReview.LandandWaterResourcesResearchandDevelopmentCorporation,Canberra,1998.
5. CommonwealthofAustralia.NHTAnnualReport2003–2004.NaturalHeritageTrust,Canberra,2004.
6. Endter-Wada, J. et al. framework for understanding social science contributions toecosystemmanagement.EcologicalApplications8(3),891–904,1998.
7. Brown,G.Mappingspatialattributesinsurveyresearchfornaturalresourcemanage-ment:Methodsandapplications.SocietyandNaturalResource18,17–39,2004.
8. CommonwealthofAustralia.Australia’sNaturalResources1997–2002andBeyond.NationalLandandWaterAudit,Canberra,2002.
9. Radeloff,V.C.etal.Exploringthespatialrelationshipbetweencensusandland-coverdata.SocietyandNaturalResources13,599–609,2000.
10. Field, D. R. et al. Reaffirming social landscape analysis in landscape ecology: Aconceptualframework.SocietyandNaturalResources16,349–361,2003.
11. Grove,J.M.etal.DataandmethodscomparingsocialstructureandvegetationstructureofurbanneighbourhoodsinBaltimore,Maryland.SocietyandNaturalResources19,117–136,2006.
12. Curtis,A.,Byron,I.,andMcDonald,S.Integratingspatiallyreferencedsocialandbio-physicaldatatoexplorelandholderresponsestodrylandsalinityinAustralia.JournalofEnvironmentalManagement68,397–407,2003.
13. Curtis,A.,Byron,I.,andMacKay,J.Integratingsocio-economicandbiophysicaldatatounderpincollaborativewatershedmanagement.AmericanJournalofWaterResourcesAssociation41(3),549–563,2005.
14. Lesslie, R. Barson, M., and Smith, J. Land use information for integrated naturalresources management—a coordinated national mapping program for Australia.JournalofLandUseScience1(1),1–18,2006.
15. Flavel,R.,andRatcliff,C.MountLoftyRangesImplementationProject:EvaluatingRegionalChange.PrimaryIndustriesandResourcesSouthAustralia,Adelaide,2000.
16. Smith,J.,andLesslie,R.LandUseDataIntegrationCaseStudy:TheLowerMurrayNAP Region. Unpublished report to the National Land and Water Resources Audit.BureauofRuralSciences,Canberra,2005.
42963_C003.indd 60 11/6/07 7:10:46 AM
Spatial Methodologies for Integrating Social and Biophysical Data 61
17. GlenelgHopkinsCMA.GlenelgHopkinsRegionalCatchmentManagementStrategy.GHCMA,Hamilton,2003.
18. Byron,I.,Curtis,A.,andMacKay,J.ProvidingSocialDatatoUnderpinCatchmentPlanningintheGlenelgHopkinsRegion.BureauofRuralSciences,Canberra,2004.
19. DahlhausP.,Heislers,D.,andDysonP.GHCMAGroundwaterFlowSystems.DahlhausEnvironmentalConsultingGeology,GHCMA,Hamilton,2002.
20. BureauofRuralSciences.GuidelinesforLandUseMappinginAustralia:Principles,ProceduresandDefinitions,3rded.AustralianGovernmentDepartmentofAgriculture,FisheriesandForestry,Canberra,2006.
21. NationalLandandWaterResourcesAudit.AustralianDrylandSalinityAssessment2000.Extent, Impacts,Processes,MonitoringandManagementOptions.NLWRA,Canberra,2001.
22. GlenelgHopkinsCMA.SalinityPlan.GHCMA,Hamilton,2002. 23. Sinischalchi,J.M.etal.Mappingsocialchange:Avisualisationmethodusedinthe
Monongahelanationalforest.SocietyandNaturalResources.19,71–78,2006. 24. Jenness,J.Nearestfeatures(nearfeat.avx)extensionforArcView3.x,v.3.8a.Jenness
Enterprises. (Available at http://www.jennessent.com/arcview/nearest_features.htm).2004.
25. Aslin,H.et al.Peri-urbanLandholdersandBio-security Issues—AScopingStudy.BureauofRuralSciences,Canberra,2004.
26. Spies,B.,andWoodgate,P.SalinityMappingMethodsintheAustralianContext.LandandWaterAustralia,Canberra,2004.
27. Baker, P., and Evans W. R. Mid Macquarie Community Salinity Prioritisation andStrategicDirectionProject.BureauofRuralSciences,Canberra,2002.
28. BureauofRuralSciences.LandUseMapping for theMurray-DarlingBasin:1993,1996,1998,2000Maps.AustralianGovernmentDepartmentofAgricultureFisheriesandForestry,Canberra,2004.
29. Walker, P., and Mallawaarachchi, T. Disaggregating agricultural statistics usingNOAA-AVHRRNDVI.RemoteSensingofEnvironment63,112–125,1998.
42963_C003.indd 61 11/6/07 7:10:47 AM
42963_C003.indd 62 11/6/07 7:10:47 AM
63
4 An Integrated Socioeconomic Study of Deforestation in Western Uganda, 1990–2000
Ronnie Babigumira, Daniel Müller and Arild Angelsen
Contents
4.1 Introduction................................................................................................... 634.2 Background...................................................................................................64
4.2.1 Uganda...............................................................................................644.2.2 TheForestrySector...........................................................................65
4.3 Deforestation.................................................................................................664.3.1 DefinitionsofDeforestation...............................................................664.3.2 GoodorBadDeforestation................................................................ 674.3.3 AConceptualFramework..................................................................68
4.4 DataandMethods.........................................................................................694.4.1 DataSources......................................................................................694.4.2 EconometricModel............................................................................ 704.4.3 MethodologicalIssues....................................................................... 71
4.5 ResultsandDiscussion.................................................................................. 724.5.1 DescriptiveStatistics.......................................................................... 724.5.2 EconometricResultsandDiscussion................................................. 74
4.5.2.1 SocioeconomicContext........................................................ 754.5.2.2 SpatialContext..................................................................... 764.5.2.3 InstitutionalContext............................................................. 76
4.6 ConcludingRemarks.....................................................................................77References................................................................................................................ 78
4.1 IntRoDUCtIon
Thepast20yearshasbeenaperiodofintensivestatisticalinvestigationintothecausesoftropicaldeforestation,withtheworkofAllenandBarnes1commonlyreferredtoasthearticlethatkicked-offthiseffort.Yetthereissurprisinglylimitedconvergenceonthebasicquestion:“whatdrivesdeforestation?”Thereareanumberofreasonsfor
42963_C004.indd 63 11/6/07 7:16:10 AM
64 Land Use Change
this.First,thesimplefactisthattheanswertothisquestioniscontextspecific—itisnotthesameconstellationoffactorsthatcanexplaindeforestationacrossthetropics.Second,onecanexpectsomeresearcherbias,inthesensethattheanswersprovidedreflect the researchers’background:geographical focus,discipline,politicalview,andsoforth.Third,thevariablesincludedhavedifferedgreatly—oftendeterminedbywhateverdataareeasilyavailable.Thesefactorshaveleadtodifferentandevencontradictorydeforestationstoriesbeingtold.Onewaytowardaconsensuswouldbebetterandmoreintegratedandholisticmethodologies.Thisbookmakesthecasefortheneedandroleforspatiallyintegratedmodelsofcouplednaturalandhumansystemsinthecontextsofstudyandmanagementoflanduse.
ThischapterisanempiricalapplicationofanintegratedapproachusingdatafromWesternUganda.Ourobjectiveistoanalyzetherolethatthecontextwithinwhichlanduseagentsoperateplaysintheirlandusedecisions.Todothisweintegratespatiallyexplicit socioeconomicandbiophysicaldata aswell asdataon landcover changesderivedfromremotesensingtoestimateaneconometricmodelofdeforestation.
Wearguelikeothersthatdeforestationismainlyaresultofactionsofagentsrespondingtoincentives.Indeed,overthepast20yearsmostanalystshavearguedthattropicaldeforestationoccursprimarilyforeconomicreasons,thatis,landusersconvertforesttononforestusesifthenewlandrenttheycangetishigherthanforforestuses.Thisapproachisbasedonthefactthatpeopleandsocialorganizationsare themost substantial agentsof change in forested ecosystems throughout theworld.2 Although this perspective is important, it is not the complete story oftropical deforestation. The incentives (land rent) are determined by the contextwithinwhichagentsoperate,andamorecomprehensiveanalysisneedstoincorpo-ratetheseaswell.
Followingabroadreviewofeconomicmodelsofdeforestation,AngelsenandKaimowitz3 recommended incorporation of agricultural census and survey datainto a geographic information systems framework. They argued that models thatcombineremoteobservationswithgroundbasedsocialdatawouldallowmodelerstotakeintoaccounttheroleofsocioeconomicfactorsandhavepotentialtoimproveourunderstandingofthedeterminantsoflandcoverchanges.3,4
Thischapterintroducesthreekeyaspectsofcontext,namelythesocioeconomic,spatial,andinstitutionalaspects.AfterabriefbackgroundonUgandaandthedefores-tationdebate,wepresentaframeworkofanalysisandthendataandmethods.Thekeyresultsarethenpresentedanddiscussed.
4.2 BACKGRoUnD
4.2.1 Uganda
Ugandaisalandlockedcountrycoveringabout236,000km2,81%ofwhichissuit-ableforagricultureowingtoarichendowmentofsoilsandaclimatethatisgenerallyfavorableforfarmingthroughouttheyear.5,6,7UgandaistoalargeextentdependentonnaturalresourcesbecausethemajorityofUgandansliveintheruralareaswithlow-inputlow-outputagricultureasthemainsourceoflivelihood.7,8
42963_C004.indd 64 11/6/07 7:16:11 AM
An Integrated Socioeconomic Study of Deforestation in Western Uganda 65
The country has enjoyed an impressive economic growth rate since the early1990s,amongthehighestinSub-SaharanAfrica.Thisisinsharpcontrasttoitsrecentpast.Thelate1970sandtheearly1980swerecharacterizedbyeconomicchaosthatresultedfromthecivilunrestoftheperiod.Macro-andmicroindicatorsofeconomichealthwerepoor,with low savings rates, high inflation rates, and ahigh externaldebtburden.Atippingpointinthistrend,however,wasthechangeingovernmentin1986.Thenewgovernmentthenembarkedonanumberofinitiativestorehabilitate,stabilize,andexpand theeconomy.Theresultof these initiativeswas theonsetofUganda’sownroaringnineties.Theexceptiontothispictureisthenorthernpartofthecountry,wherepoliticalinstabilityandviolencehaveemptiedthecountrysideinmanydistricts.Itisforthisreasonthatwedonotfocusonthewholecountry.
Additionally,populationhasbeengrowingatanaverageof2.5%peryear,9almostdoublinginjust22yearsfrom12.6millionin1980toabout24.7millionin2002.Duringthelatterpartofthisperiodgrowthwasevenhigher,withanaveragegrowthrateof3.4%between1991and2002.10Thepopulation isprojected to increase to32.5millionby2015.7
Giventhehighdependenceonnaturalresources,thecombinationofeconomicandpopulationgrowthwillundoubtedlyexertalotofpressureontheseresources.Ugandathereforeprovidesaninterestingstudyintohowthesesocioeconomicdimen-sionscouldhaveimpacteddeforestation(Figure4.1).
4.2.2 The ForesTry secTor
Priortothelate1990s,theextentofUganda’sforestestatewasbasedoneducatedguesses.Lackofcomprehensivedata limitedthedeterminationofforestareaandratesofdeforestation. Initialestimatesby theFoodandAgricultureOrganization(FAO)puttheforestandwoodlandcoverat45%ofthetotallandcoverin1890.Morerecentfigureshavebeeninthe20%to25%range.Forestandwoodlandareimportantbecauseonly3%ofUgandanhouseholdsinruralareasand8%inurbanareashaveaccesstogridelectricity;therestrelyonbiomassforenergysources.11Itisestimatedthatforestsprovideanannualeconomicvalueof$360million(6%ofGDP).Treesthroughfuelwoodandcharcoalprovide90%oftheenergydemandswithaprojec-tionof75%in2015.
ThefirstefforttomapUganda’soriginalvegetationwasdonebyLangdale-Brown12in1960whoestimatedtheextentofforestcoverfor1900,1926,and1925.13Thesedatashowanincreasingtrendintheannualrateofchangeofforestcover(Table4.1).
Thenext effort tomapUganda’s forest estatewasundertakenbyHamilton.13Usingsatelliteimagery,Hamiltontriedtomapoutclearstandingforest.Ourunder-standingofthismapisthatitfocusesonwhatissubsequentlyreferredtoastropicalhighforestbytheNationalBiomassStudy(NBS).ThemaprevealsthatforestisnotaparticularlycommontypeofvegetationinUganda.ThisledHamiltontoconcludethat visions of vast sweeps of mahogany-rich jungles, such as are entertained bysomeplanners,werequiteillusory.
AmorerecentandcomprehensiveattemptwasundertakenbyNBSinaprojectstartedin1989withtheobjectiveofprovidinguniqueinformationonthedistribu-tionandindirectlyconsumptionofwoodybiomassinthecountry.
42963_C004.indd 65 11/6/07 7:16:11 AM
66 Land Use Change
4.3 DeFoRestAtIon
4.3.1 deFiniTions oF deForesTaTion
Deforestationhasbeenusedtodescribechangesinmanydifferentecosystems.Itisgenerallydefinedaslossofforestcoverorforestedland,1,14whileVanKootenandBulte15defineitastheremovaloftreesfromaforestedsiteandtheconversionoflandtoanotheruse,mostoftenagriculture.FAOappliesasimilardefinition—aperma-nentchangefromforesttononforestlandcover,withforestbeingdefinedasanareaofminimum0.5hawithtreesofminimum5mheightinsitu,minimum10%canopycover,andthemainusenotbeingagriculture.
N
S
W E
Kilometers
Major roadAll–year roadWater body
YesNo
0 5025 100 150
Legend
TanzaniaRwanda
Kenya
Sudan
Congo, DRC
Kampala
North
Central East
West
Deforestation
FIGURe 4.1 (See color insert following p. 132.) Ugandastudyareashowingthedistri-butionofdeforestationwithinthewesternregionofthecountry.
42963_C004.indd 66 11/6/07 7:16:13 AM
An Integrated Socioeconomic Study of Deforestation in Western Uganda 67
Moredetaileddefinitionstakeintoaccountwhathappenstothedeforestedland,transitions among classes, the magnitude of change, the threshold in area abovewhichdeforestationissaidtohaveoccurred,aswellthetemporaldimensionsofthechange.16,17Astheprecisionindefinitionincreases,sodoesthelevelofcomplexityandthechallengesofempiricalwork.However,evenrecognizingtheimportanceofexactdefinitions,thecaseforprecisionshouldnotbeexaggerated.Causesofmajorundesirableforestinterventionscanbeanalyzedandpracticalimplicationsforpolicymakingderived,eveninaworldwitharelativelackofpureconceptualdefinitions.18
4.3.2 good or Bad deForesTaTion
Thedebateondeforestationcentersonwhethertropicaldeforestationisanimpendingenvironmentaldisaster,onewhichifnotaddressedwouldhavedireenvironmentalconsequences,orisjustanotheroverhypedagendabyenvironmentalistsandsomealarmistresearchers.
Fortheever-worseningschoolofthought,tropicaldeforestationisconsideredtobeamajorenvironmentalcrisis,becauseofitsinternationalimpactsonbiodiversitylossandclimateandbecauseofitslocalimpactssuchasanincreaseinfloodoccurrence,thedepletionofforestresources,andsoilerosion.19Suchfearsabouttheimminentextinctionoflargenumbersofplantsandanimalshavepromptedanoutpouringofconcernandanalysisabouttropicaldeforestationinthepasttwodecades.20
However, there is an it’s-not-that-bad school that is a less pessimistic schoolarguingthattherearenogroundsforthealarmistclaims.21Proponentsofthisschoolwouldgoontoarguethatdeforestationisanatural,beneficialcomponentofeconomicdevelopmentespeciallyindevelopingcountriesandisthereforenothingmorethanagradualhumanalterationofanabundantnaturalresource(land)inordertoincreaseproductivityandwelfare.
Theformerschool isgenerallymoreprominent,owingto thevisibilityof theimpactsofchangesinlocalandinternationalclimate,andhasresultedintheemer-genceofthesocialmovementdevotedtoreducingdeforestation.Importantquestionstherefore remainaboutwhy,despite the emergenceof this and thepublicationofhundredsofstudiesthatanalyzeditscauses,thedestructionoftropicalrainforestsdidnotappeartoslowdownmuch,ifatall,duringthe1990s.20
tABle 4.1early estimates of Forest Cover and Deforestation Rates
YearForest and moist thicket
(Ha) total area (%)Annual forest lossa
(HaY–1)
1900 3.1×106 12.7
1926 2.6×106 10.8 1.8×104
1958 1.1×106 4.8 4.7×104
a Owncalculations.Source: Langdale-Brown(1960).12
42963_C004.indd 67 11/6/07 7:16:14 AM
68 Land Use Change
4.3.3 a concepTUal Framework
Deforestationistheresultoftwobroadsetsofprocesses:naturalandhumaninducedprocesses. In the former, forest reduction is inducedbybioticandabioticgrowthreducingfactorswithintheforestecosystemorasaresultofbroadclimaticchangesorcatastrophessuchasfiresandlandslides.1Thesenaturalprocesses,however,areoftensoslowandsubtleastobeimperceptible.
Ontheotherhand,thechangesinitiatedbyhumanactivitytendtoberapidinprogression,drasticineffects,widespreadinscale,andthusmorerelevanttousonaday-to-daybasis.Understandingtherelationshipbetweenhumanbehaviorandforestchangethereforeposesamajorchallengefordevelopmentprojects,policymakers,andenvironmentalorganizationsthataimtoimproveforestmanagement.22
Toshedsomelightonthisrelationship,wetakeasourstartingpoint,ashaveothermodelsofdeforestationinthevonThünen(1826)tradition,thatanypieceoflandisputintotheusethathasthehighestnetbenefitsorlandrent.Thecenterofthediscussionisthenhowvariousfactorsdetermineandinfluencetherentaccruedfromforestversusnonforestuses,andtherebytherateofdeforestation.ArecentextensivereviewofthisapproachisgivenbyAngelsen.23
This approach is operationalized by modeling an agent (land use decisionmaker)livingatorwithaccesstotheforestmargin,whoseaimistomaximizethelandrent.(Wearemindfulofthepitfallsofapplyingaprofitmaximizingapproachtoruralhouseholds;however,westillbelievethisapproachisinformative.)Agentsareindividuals,groupsofindividuals,orinstitutionsthatdirectlyconvertforestedlandstootherusesorthatinterveneinforestswithoutnecessarilycausingdeforest-ationbutsubstantiallyreducetheirproductivecapacity.Theyincludeshiftingculti-vators, private and government logging companies, mining and oil and farmingcorporations, forestconcessionaires, and ranchers.18Themainculpritoragent isgenerally thought tobe theagriculturalhouseholddwellingat the forest frontier(this setting is plausible in Uganda given the dependence on forests for energyhighlightedabove).
Theagent’sdecisionsare influencedbyanumberof factorssuchaspricesofagricultural outputs and inputs, available technologies, wage rates, credit accessandconditions,householdendowments, forestaccess (bothphysicalandpropertyrights),andbiophysicalvariableslikerainfall,slope,andsoilsuitability.Location,thecenterofattentioninvonThünen’soriginalwork,doesinfluenceanumberofthesevariables(e.g.,pricesandwagerates).Thesefactorsaffecttheagent’sdecisionsdirectlyandare,therefore,referredtoasdecisionparametersorimmediatecausesofdeforestation(cf.theterminologyusedbyAngelsenandKaimowitz3).
Atthenextlevelisthecontextwithinwhichtheagentsoperate.Thesecontextualforcesdeterminedeforestationviatheir impactonthedecisionparameters.Thesecausesaremorefundamentalandoftendistancedinthesensethatitisdifficulttoestablishclearlinksbetweenthissetoffactorsanddeforestation.Theyareacomplexdynamicmixofthesocioeconomic,spatial,andinstitutionalsystemsofcommuni-tiesrepresentingthefundamentalorganizationofsocietiesandinteractinginwaysthataredifficulttopredict.TheabovediscussioncanbesummarizedinFigure4.2.
42963_C004.indd 68 11/6/07 7:16:14 AM
An Integrated Socioeconomic Study of Deforestation in Western Uganda 69
4.4 DAtA AnD MetHoDs
4.4.1 daTa soUrces
Landuseandlandcoverdataforthisstudycomefromlanduse/covermapsfromtheUgandaNBSandFAOAfricover.Althoughwerefertothemasthe1990and2000maps,thesatelliteimagesusedintheirproductionarefrom1989to1992and2000to2001,respectively,owingtotheneedtousecloud-freeimages.
The1990mapwasproducedbyvisualinterpretationofSpotXSsatelliteimageryfromFebruary1989 toDecember1992.Followingpreliminary interpretation, themapwasverifiedthroughsystematicandextensivegroundtruthing.The2000mapistheFAOAfricoverlandcovermapproducedfromvisualinterpretationofdigitallyenhancedLandsatThematicMapper(TM)images(Bands4,3,2)acquiredmainlyin theyear2000/2001.ThelandcoverclassesweredevelopedusingtheFoodandAgriculture Organization/United Nations Environmental Program (FAO/UNEP)internationalstandard(LCCS)landcoverclassificationsystem.The2000mapwasreclassifiedbystaffatNBStoenablecomparisonbetweenthetwomaps.
Administrative boundaries, infrastructure, and river maps come from theDepartment of Surveys and Mapping, Ministry of Lands, Housing and UrbanSettlementsandtheDepartmentofSurveysandMapping.Socioeconomicdataare
Natural Causes
Agents
Context
Subsistenceorientedfarmers LoggersCommercial
farmers
Spatial InstitutionalSocioeconomic
Rent (Agricultural)
Deforestation
FIGURe 4.2 Conceptual framework for analysis. Deforestation is influenced by naturalcausesandhumanactivities.Thehumanactivitiesaredrivenbytherentalcostoflandwithinsocioeconomic,spatial,andinstitutionalcontexts.
42963_C004.indd 69 11/6/07 7:16:16 AM
70 Land Use Change
fromtheNationalPopulationandHousingCensus1991,bytheStatisticsDepartment,MinistryofFinanceandEconomicPlanning.
The slopeandelevationwere calculated from thedigital elevationdataof theShuttleRadarTopographicMission(SRTM)(CGIAR-CSISRTM).Void-filledseam-lessSRTMdataV1,accessedJanuary2005,availablefromtheCGIAR-CSISRTM90m Database: http://srtm.csi.cgiar.org. Soil data are from Uganda’s agroecologi-calzones(AEZ)database24andfromtheresultsofasoilreconnaissancesurvey.25Followingconsultationswithoneof the authorsof thismap,weuse soil organicmatterandsoiltextureasthevariablestocapturesoilsuitability.Wethencalculateaweightedindexfrombothrastermaps.Thisindexactsasaproxyforagriculturalpotentialinherentinaparcel.
ThedifferentmapswereprojectedintoUniversalTransverseMercator(UTM)Zone36,southoftheequatorandthenassembledinarastergeographicalinforma-tionsystem(GIS)whereweresampledthedatatoacommonspatialresolutionof250m.Thechoiceofresolutionwasprimarilyguidedbytheneedforamanageabledatasize.
AGISwasused togenerateadditionalspatialvariables,specifically thecost-adjusteddistancetoroads,theeuclideandistancetowater,andtheeuclideandistancetoprotectedareas.WethenexportallthegridsasASCIIfilesandimportthemintoStata9,26whichweusetocarryoutthedescriptiveandeconometricanalysis.
4.4.2 economeTric model
Toanalyzetherolethatcontextplaysinlandusechange,weestimateaneconomet-ricmodelfortheprobabilitydeforestation.Ourunitofanalysisisa6.25hapixel.Underlyingthiseconometricmodelisalatentthresholdmodelbasedontheideathatthelandusedecisionregardingtheparcelismadebyanoperatorwhocanbeasingleperson,household,orgroupofpeopleinthecaseofcommonpropertyownership.27Thisoperatormayormaynotowntheparcel(ourdatadoesnotallowustomakethatdistinction).However,weassumethatforanygivenparcel,thereisanoperatorwhoisabletomakealandusedecisionpertainingtothisparcel.Aparcelwillbeclearedifitiseconomicallyprofitable.Thatis:
R R
nft f ft f+ +≥1 1| |
where Rnft f+1| representsthepresentvalueoftheinfinitestreamofnetreturnsfrom
convertingaparcelthatwasoriginallyunderforest( f)inperiodttononforest(nf)landuse inperiod t +1, whichwewill refer toasagricultural rent.This typeofmodelisfurtherdiscussedelsewhere.27,28Inlinewiththisintegratedapproach,theeconomicprofitabilityofaparcel isa functionof threesetsof factors: thesocio-economic,spatial,andinstitutionalcontexts.
1.The socioeconomic context within which the parcel is embedded has abearingonoutputpricesandinputcosts.Higheroutputpriceswillincreaseagricultural rent, while higher wages translate into higher input costs,whichreducetherentandmaythusreducetheprobabilityofdeforestation.
42963_C004.indd 70 11/6/07 7:16:18 AM
An Integrated Socioeconomic Study of Deforestation in Western Uganda 71
Wearguethatbecausetheopportunitycostoflaborinpoorcommunitiesistypicallyverylow,theprobabilityofdeforestationwillbehigherinpoorercommunities.Moreover,inequalitymayhaveabearingwithinthisframe-work. For any given average income, higher inequality implies a largerproportion of the population has an opportunity cost of labor below thelevelthatmakesforestclearingprofitable.Thuswehypothesizethathighinequalitywillbecorrelatedwithhigherprobabilitiesofdeforestation.
2.Thespatial contexthasaninfluenceontheagriculturallandrent.Includedinthisistheinsituresourcequality,thatis,theresponseofthelandtotheusewithout regard to its locationdetermines thequantityofagriculturalharvestpossiblefromagivenparcel,whichinturnaffectstheprobabilityofclearance.Alsoincludedistheaccessibilityand,byextension,allcostsandbenefitsassociatedwithaspecificlocationasopposedtoresourcequalityaswellasidiosyncraticlocation-specificcharacteristicsoftheparcel.Moreaccessibleparcelsaremorelikelytobecleared,andthisdoesnotnecessarilymeanthatagriculturewillbethesubsequentlanduse.Theseparcelswillbeclearedmainlyforthesaleoftimber.
3.Finally,theinstitutional contextwithinwhichtheagentsoperatealsohasaninfluenceonagriculturallandrent.Thisprimarilyreferstothepropertyrightsregimesinthecommunitiesthatdetermineaccessanduserights.Totheextentthattheyareenforceable,restrictionsonclearancetranslateintoacostandtherebyloweragriculturalrent.
We therefore select a number of explanatory variables that best capture the con-textsurroundingthemanagementoftheparcel.ThevariablesandtheiroriginsaredescribedinTable4.2togetherwithouraprioriexpectationsontheirrelationshipwiththelikelihoodofdeforestation.
Ourfocusisonagriculturalrentonly,whileforestrentisignored.Thissimpli-ficationcanbejustifiedontwogrounds:First,muchoftheforestisofdefactoopenaccessandtheforestrentthereforeisnotcapturedbytheindividuallanduser(unlikeagriculturalrent).Second,duringearlystagesintheforesttransition(characterizedbyhighlevelsofdeforestation,suchasinWesternUganda),changesinagriculturalrentratherthanforestrentarethekeydriver(cf.Angelsen23).
4.4.3 meThodological issUes
Conventional statistical analysis frequently imposes a number of conditions orassumptions on the data it uses. Foremost among these is the requirement thatsamplesbe random.Spatialdata almost alwaysviolate this fundamental require-ment,andthetechnicaltermdescribingthisproblemisspatial autocorrelation.29
Spatial autocorrelation (dependence) occurs when values or observations inspacearefunctionallyrelated.Spatialautocorrelationmayarisefromanumberofsourcessuchasmeasurementerrorsinspatialdatathatarepropagatedintheerrortermsorfrominteractionbetweenspatialunits.Itmayalsoarisefromcontiguity,clustering,spillovers,externalities,orinterdependenciesacrossspace.
42963_C004.indd 71 11/6/07 7:16:18 AM
72 Land Use Change
Threeapproachesforcorrectingforspatialeffectsareoftenmentionedin theliterature:regularsamplingfromagrid,purespatiallagvariablesusinglatitudeandlongitude indexvalues, and spatial lagvariables involving ageophysical variablesuchasaslopeorrainfall.30
Beforecarryingouttheeconometricestimation,wetestforspatialdependenceusing the SPDEP package31 in R language.32 We find evidence of spatial auto-correlationatboththepixelandparishlevels.Weminimizetheeffectsofspatialautocorrelationbyincludinglatitudeandlongitudeindexvariables,andbydrawingasamplefromagridwithadistanceof500mbetweencells.
4.5 ResUlts AnD DIsCUssIon
4.5.1 descripTive sTaTisTics
Mostdeforestationwasconcentratedinafewareas.Aplotofcumulativedistribu-tionofdeforestationshowsthat15%oftheparishesaccountedfor70%ofthetotaldeforestation(Figure4.3).Furthermore,mostofthedeforestation(60%)waswithin10kmfrommainroads(Figure4.4).
tABle 4.2Description of Variables
Variable Description source expected signa
socioeconomic Context
head_emp Employedhouseholdheads Census91 –
educ_Gini EducationGinicoefficientb Census91 +
popdens Populationdensity Census91 +
mig_share Shareofmigrantsinparish Census91 +
spatial Context
cdcity_allrds Costadjusteddistancetoroads Infrastructuremap –
dwater Distancetowater Infrastructuremap –
slp Slope DEM –
elev Elevation DEM –
soil+2cl Proportionofsuitablesoils CIAT +
rain Rainfall CIAT ?
x Latitudeindexvalue LUC&infrastructuremaps ?
y Longitudeindexvalue LUC&infrastructuremaps ?
Institutional Context
dprotect Distancefromprotectedareas LUC&infrastructuremaps ?
prtct Protectedareadummy LUC&infrastructuremaps –a A priori expectationson theeffectofvariablesondeforestation (–) less; (+)moredeforestation;
(?)ambiguous.CIAT,InternationalCenterforTropicalAgriculture;DEM,digitalelevationmodel;LUC,landusecover.
b TheEducationGinicoefficientisameasureofinequalityrangingfromzero(perfectequality)toone(perfectinequality).
42963_C004.indd 72 11/6/07 7:16:19 AM
An Integrated Socioeconomic Study of Deforestation in Western Uganda 73
Somedescriptive statistics arepresented inTable4.3.Compared to all otherparcels,forestparcelshadmorerainfall,wereatalowerelevation,andwereonlesssteepslopes.Not surprisingly, forestparcelsweregenerally located farther fromurban centers and farther from themain roads.This is typical of avonThünendevelopment process, with areas close to urban centers being cleared first and
1009080706050403020100Parishes – cumulative %
Obs
erve
d D
efor
esta
tion
– cum
ulat
ive %
Deforestation is accumulated starting with highest valueSource: NBS, FAO Africover
100
90
80
70
60
50
40
30
20
10
0
FIGURe 4.3 CumulativedeforestationbyparishesacrossUgandastartingwiththelargestpercentagearea.
Parcels within 10 KM
60.75
28.32
10.93
Parcels between 10 KM and 20 KM Parcels 30 KM or more
Perc
enta
ge lo
ss
60
40
20
0
FIGURe 4.4 Relationshipbetweendeforestationanddistancefromroads.
42963_C004.indd 73 11/6/07 7:16:23 AM
74 Land Use Change
expandingaspopulationand theeconomygrows.Most forestparcelswere inorclosetoprotectedareas.
Withintheforestedparcels,theonesthatweredeforestedwereatalowereleva-tionwithlesssteepslopes.Thissuggeststhataccessibilitywasakeyfactorinthedecisiontoclearaforestparcel,consistentwithourexplanationabove.
4.5.2 economeTric resUlTs and discUssion
Giventhebinarynatureofthedependentvariable,thatis,thelandiseitherclearedor it isnot,weestimateabinary logitmodel.Wecorrect forpossiblecorrelationintheerrortermsofpixelswithinaparishandusetheHuberandWhitesandwichestimatortoobtainrobustvarianceestimates.
The econometric results are presented in Table4.4. The dependent variableisacategoricalvariable indicatingwhetherornot theparcelwasdeforested.The
tABle 4.3Descriptive statistics
All Parcels
(n = 697,060)
Mean Minimum Maximum
Proportionofsuitablesoils 64.89 32 91
Rainfall(mm) 1,091.05 701 1,949
Elevation(meters) 1,296.52 601 4,391
Slope 6.01 0.00 63.66
Cost-adjusteddistancetoroads 0.51 0.00 1.67
Distancetourbancenters(km) 64.68 0.00 167.38
Distancetoroad(km) 8.74 0.00 38.57
Distancefromprotectedareas(km) 6.83 0.00 38.21
Distancefromwater(km) 16.29 0.25 57.34
Forest Parcels
All Deforested nondeforested
(n = 194,601) (n = 46,420) (n = 148,181)
Mean Mean Mean
Proportionofsuitablesoils 62.51 63.78 62.11
Rainfall(mm) 1,177.46 1,148.02 1,186.69
Elevation(m) 1,230.12 1,052.06 1,285.91
Slope 5.2 3.77 5.65
Cost-adjusteddistancetoroads(km) 0.65 0.62 0.66
Distancetourbancenters(km) 82.91 85.97 81.95
Distancetomainroad(km) 9.98 8.97 10.29
Distancefromprotectedareas(km) 3.23 4.78 2.75
Distancefromwater(km) 19.3 18.28 19.62
42963_C004.indd 74 11/6/07 7:16:23 AM
An Integrated Socioeconomic Study of Deforestation in Western Uganda 75
regressionmodelwas significantat thep= .001 level.Mostcoefficientshave theexpectedsigns.Belowwediscussthestatisticallysignificantresults.
4.5.2.1 socioeconomic Context
Theresultsshowthatdeforestationismorelikelytooccurinbetter-offcommunities.Ourproxy forwealth in thecommunity—proportionofhouseholdheads that areemployedoutsidethefarm—isstatisticallysignificantatthe5%level,withanoddsratio(foronestandarddeviationincrease)of1.2.Therearetwocontradictoryeffectsof the better-off farm employment opportunities reflected by this variable. First,goodemploymentopportunitiesincreasetheopportunitycostsoflaborandtherebyloweragriculturalrentandreducethepressureonforestconversion.But,ahigher
tABle 4.4logit Model Results of Deforestation in Western Uganda, 1990–2000
Variable Coef. p-value e^(b*sdx)a
socioeconomic Context
Proportionofheadsthatareemployed 0.828* 0.034 1.202
EducationGinicoefficient 0.329 0.675 1.028
Populationdensity(pp/ha) 0.291** 0.008 1.166
Shareofmigrantsinparish 1.363** 0.008 1.302
spatial Context
Costadjusteddistancetoroads –1.318** 0.004 0.709
Distancetowater –0.004 0.503 0.948
Slope 0.051*** 0.000 1.366
Elevation –0.002*** 0.000 0.339
Proportionofsuitablesoils –0.004 0.579 0.948
Rainfall –0.002 0.054 0.784
X –0.003*** 0.000 0.450
Y –0.003*** 0.000 0.409
Institutional Context
Protecteddummy –1.912*** 0.000 0.388
Distancefromprotectedarea –0.003 0.843 0.985
Constant 8.250*** 0.000
No.ofparcels 43,760
Modelp-value 0.000
PseudoR2 0.191a ChangeinoddsforoneSDincreaseinx.Dependentvariable=1ifparcelchangedfromforesttononforestclass,0otherwise.Percentageofcorrectpredictions=76.3.*,p<.05;**,p<.01;***,p<.001.
42963_C004.indd 75 11/6/07 7:16:24 AM
76 Land Use Change
shareofoff-farmemploymentisalsocorrelatedwitheconomicdevelopment,creat-inghigherdemandforagriculturalproducts.OurresultssuggestthatintheWesternUgandancontextthelattereffectisdominating.
Consistentwith thisexplanation,wealsofindthathigherpopulationdensitieshaveapositiveimpactonforestconversion.Ahighshareofmigrantsintheparishalsopullsinthesamedirection.Migrantshaveinitiallynoorverysmallparcelsofagriculturallandandcanthereforebeexpectedtobecomemajoragentsofdeforesta-tion.Thustheempiricalmodelsuggeststhatmigrationtotakeadvantageofforestlandsuitableforagriculturalconversionplaysamajorrole.Inequality,asmeasuredbytheeducationalGini,doesnotappeartohaveanysignificanteffectonthelikeli-hoodofdeforestation.
4.5.2.2 spatial Context
Parcelsclosertoroadsweremorelikelytobedeforested.Thedescriptivestatisticshave already shown this, and the econometric results confirm the importance ofdistanceasafactor.Thisresultsuggests,inlinewithnumerousotherstudies,thatinterventionsthatreducethecostofaccesstoforestedlandwillincreasethelikeli-hoodofdeforestation.
Notsurprisinglywefindthatparcelsatlowerelevationweremorelikelytobedeforested, suggestingagain that the spatial contexthadabearingon thecostofaccess.Aonestandarddeviationincreaseintheelevationreducedtheoddsofdefor-estationby0.34.Asurprisingresultis,however,theeffectofslopeontheprobabilityofdeforestation.Weanticipatedthattheprobabilityofdeforestationwouldbenega-tivelycorrelatedwiththeslope(mainlyowingto thehighercostsassociatedwithworkingathigherelevationandsteeperslopes).However,wefindinsteadthatsteeperslopesweremorelikelytobedeforested,andwedonothaveasatisfactoryexpla-nationof this result.Aplausibleexplanationcouldbe that lower landshavebeenconvertedandthepressuremayhaveshiftedtothemarginallands.Supportforthiscanbefoundintheargumentthatfragilelandsinsub-SaharanAfricaarefacingaworseningsocialandenvironmentalcrisis.33
Distance towater, rainfall, andproportionof suitable soilswerenot statisti-cally significant. One reason for the insignificance of the soil variable might bethatthemaponwhichthisvariableisbasedisrathercoarse,andthereforedoesnotcapturetherelevantlocalspecificvariationthatmayexist.Thesamemaybetrueforrainfall.
4.5.2.3 Institutional Context
Aninterestingresultthatemergesfromthisworkisthefactthatinstitutionalinterven-tionsseemtohavemitigateddeforestation.Parcelsinprotectedareaswerelesslikelytobedeforested.Thisisalsowhatonecanobserveonthemapsandontheground:theprotectedareasareindeed“greener.”Theconservationareashavebeenbackedbyrelativelystrongenforcementatthelocallevelswithpunishmentofviolators.
In addition, we tested if the conservation led to more land being deforestedoutsidetheconservationareas,akindofnegativespillovereffect.Ourhypothesiswould thenbe that,aftercontrolling forother factors, landclose to theprotected
42963_C004.indd 76 11/6/07 7:16:24 AM
An Integrated Socioeconomic Study of Deforestation in Western Uganda 77
areasshouldexperiencehigherforestconversion.Butthisvariableisnotstatisticallysignificantthuswecouldnotrejectthenullhypothesis.
4.6 ConClUDInG ReMARKs
The process of land use change is driven by a complex web of factors that cutsacrossdisciplines.Thismeansthateffortstoaddressthelandusechangeprocessshouldsimilarlybeholisticandcutacrossdisciplines.Thischapterisanexampleof how such a study couldbe empirically carriedout.Weargued that additionalinsightscouldbegainedfromintegratingspatiallyexplicitsocioeconomic,institu-tional,biophysical, and landcoverdata. Increasingavailabilityofhigh resolutionspatialdatameansthatsuchanapproachispossibleinmostplaces.Thefact thatthevariablesusedtocapturethesocioeconomiccontextaresignificantshowsthatsuchaframeworkcanbeofpolicyrelevance,forexample,byincludingtheeffectonnaturalresourcesinthedesignandimplementationofhumanresettlementprogramsorinfrastructuredevelopmentprojects.
Four main stories emerge from our econometric analysis: the poverty cumaffluence,thepopulation,theprotection,andthespatialstory.Althoughwesetoutthinkingdeforestationwasdrivenbypoorhouseholds,wedonotfindanyevidenceinsupportofthis.Ratheritappearsthatdeforestationismorelikelyinthebetter-offcommunities. Second, high population densities, together with a high proportionofmigrantswhomaybe ingreaterneedforagricultural land,hasalsoplayedanimportantrolefordeforestationinWesternUgandaduringthe1990s(Figure4.5).
(a) (b)
(c)
FIGURe 4.5 Landusechange inUganda. (a)Forestclearing. (b)Bananaplantationsonclearedland.(c)Pastorallanduseonclearedland.
42963_C004.indd 77 11/6/07 7:16:26 AM
78 Land Use Change
Third,thereisalsoastrongspatialstorytothisintermsoffactorssuchasclosenesstoroadsandlowelevation,leadingtomoredeforestation.Finally,ourstudyshowsthatprotectionhadbeeneffectiveinreducingthelikelihoodofdeforestation.
ReFeRenCes
1. Allen, J.C., andBarnes,D.F.The causesof deforestation indeveloping countries.Annals of the Association of American Geographers75(2),163–184,1985.
2. Wear,D.N.,Abt,R.,andMangold,R.People,spaceandtime:Factorsthatwillgovernforestsustainability.In:Transactions of the 63rd North American Wildlife and Natural Resources conference.WildlifeManagementInstitute,Washington,D.C.,1998.
3. Angelsen,A.,andKaimowitz,D.Rethinkingthecausesofdeforestation:lessonsfromeconomicmodels.World Bank Research Observer14(1),73–98,1999.
4. Rindfuss,R.R.,andStern,P.C.Linkingremotesensingandsocialscience:Theneedand challenges. In: Liverman, D., et al., eds., People and Pixels: Linking Remote Sensing and Social Science.NationalAcademyPress,Washington,D.C.,1–27,1998.
5. Mukiibi,J.K.AgricultureinUganda.In:Mukiibi,J.K.,ed.,Agriculture in Uganda.FountainPublishers,Kampala,2001.
6. Esegu,J.F.O.ForesttreegeneticresourcesinUganda.In:Mukiibi,J.K.,ed.,Agricul-ture in Uganda.FountainPublishers,Kampala,2001.
7. NEMA.State of the Environment Report for Uganda 2000/2001.MinistryofWater,LandsandtheEnvironment,Kampala,2000.
8. MAAIFandMFPED.Plan for Modernisation of Agriculture: Eradicating Poverty in Uganda.Kampala,2000.
9. NEMA.State of the Environment Report for Uganda 1998.MinistryofWater,LandsandtheEnvironment,Kampala,1998.
10. MFPED.Background to the Budget for Financial Year 2004/05,Kampala,2004. 11. MFPED.Poverty Eradication Action Plan (2004/5–2007/8),Kampala,2004. 12. Langdale-Brown, I., The VegetationofUganda.UgandaDepartment ofAgriculture
2(6),1960. 13. Hamilton,A.C.Deforestation in Uganda.OxfordUniversityPress,Nairobi,1984. 14. Geist,H.J.,andLambin,E.F.What Drives Tropical Deforestation?Louvain-la-Neuve,
2001. 15. Van Kooten, G. C., and Bulte, E. The Economics of Nature: Managing Biological
Assets.Malden,MA,Blackwell,2000. 16. FAO. State of the Worlds Forests, 1997. Food and Agriculture Organization of the
UnitedNations(FAO),Rome,1997. 17. Myers,N.Conversion of Tropical Moist Forests: A Report Prepared by Norman Myers
for the Committee on Research Priorities in Tropical Biology of the National Research Council.NationalAcademyofSciences,Washington,D.C.,1980.
18. Contreras-Hermosilla,A.The Underlying Causes of Forest Decline.CenterforInter-nationalForestryResearch(CIFOR),Bogor,Indonesia,2000.
19. Caviglia,J.L.,andKahn,J.R.Diffusionofsustainableagriculture in theBraziliantropical rain forest: A discrete choice analysis. In: Orlove, B. S., ed., Current Anthropology.UniversityofChicagoPress,Chicago,2001.
20. Rudel,T.K.Tropical Forests: Regional Paths of Destruction and Regeneration in the Late Twentieth Century.Chichester,ColumbiaUniversity,NewYork,2005.
21. Lomborg,B.The Skeptical Environmentalist: Measuring the Real State of the World.CambridgeUniversityPress,Cambridge,2001.
42963_C004.indd 78 11/6/07 7:16:27 AM
An Integrated Socioeconomic Study of Deforestation in Western Uganda 79
22. Southworth, J., and Tucker, C. The influence of accessibility, local institutions, andsocioeconomicfactorsonforestcoverchangeinthemountainsofwesternHonduras.Mountain Research and Development21(3),276–283,2001.
23. Angelsen, A. Forest cover change in space and time: combining von Thünen andthe forest transition. In: World Bank Policy Research Working Paper, World Bank,Washington,D.C.,2006.
24. CIAT.Bean Database for Africa and Uganda’s Agroecological zones (AEZ) Database. CD-ROM,Ver.1.0,Kampala,1999.
25. MemoirsoftheResearchDivision,KawandResearchStation.DepartmentofAgricul-ture,Uganda,1960.
26. StataCorp.Stata Statistical Software: Release 9,CollegeStation,TX,2005. 27. Nelson, G. C., and Geoghegan, J. Deforestation and land use change: Sparse data
environments.Agricultural Economics27(3),201–216,2002. 28. Munroe,D.,Southworth,J.,andTucker,C.M.Thedynamicsofland-coverchangein
western Honduras: Spatial autocorrelation and temporal variation. Prepared for the2001AAEAAnnualMeetings,2001.
29. O’Sullivan,D.,andUnwin,D.Geographic Information Analysis.JohnWiley&Sons,Hoboken,NJ,2003.
30. de Pinto, A., and Nelson, G. C. Correcting for spatial effects in limited dependentvariableregression:Assessingthevalueof“ad-hoc”techniques.PresentedatAmericanAgriculturalEconomicsAssociationAnnualMeeting,LongBeach,CA,2002.
31. Bivand, R. Implementing spatial data analysis software tools in R. Geographical Analysis38(1),23–40,2006.
32. RDevelopmentCoreTeam.R: A language and environment for statistical computing.RFoundationforStatisticalComputing,Vienna,Austria,2005.
33. Babu,S.,andHazell,P.Growth,povertyalleviation,andtheenvironmentinthefragilelandsof sub-SaharanAfrica. In:Strategies for Poverty Alleviation and Sustainable Resource Management in the Fragile Lands of Sub-Saharan Africa.DeutscheStiftungfürInternationaleEntwicklung,Kampala,1998.
42963_C004.indd 79 11/6/07 7:16:27 AM
42963_C004.indd 80 11/6/07 7:16:28 AM
81
5 Modeling Unplanned Land Cover Change across Scales
A Colombian Case Study
Andres Etter and Clive McAlpine
Contents
5.1 Introduction................................................................................................... 815.2 StudyRegion................................................................................................. 83
5.2.1 RegionalLevel...................................................................................845.2.2 LocalLevel........................................................................................85
5.3 Methods.........................................................................................................855.3.1 NationalLevel....................................................................................865.3.2 RegionalLevel...................................................................................865.3.3 LocalLevel........................................................................................88
5.4 Results...........................................................................................................885.4.1 NationalDeforestationThreatHotSpotsandMainDrivers.............885.4.2 AmazonColonizationFronts.............................................................905.4.3 DeforestationPatternsattheLocalLevel.......................................... 91
5.5 DiscussionandConclusions..........................................................................92Acknowledgments....................................................................................................96References................................................................................................................96
5.1 IntRoDUCtIon
Landuseistheinteractionbetweenhumansandthebiophysicalenvironmentwithcumulativeimpactsonthestructure,function,anddynamicsofecosystemsatthelocal, regional, and global levels of ecological organization.1 Human impacts ontheEarth’senvironmentareleavinganincreasingecologicalfootprint,whichnowthreatensmanyglobalecosystems.2,3Humanactivitiesthatproducechangesinlandcover, such as agriculture, mining, and urban development, are a major cause ofthe ecological footprint. Over the past century, the global human population hasincreased3.5 times,while the areaof agricultural landhasdoubled.4From1980thetropicshaveexperiencedhigherdeforestationthantemperateregions,withthe
42963_C005.indd 81 11/6/07 7:14:27 AM
82 Land Use Change: Science, Policy and Management
largestconcentrationintheAmazonBasin.5Changesinlandcoversuchasdefor-estationaffect the functioningofecological systemsatmultiple scales,withcon-sequencesrangingfromglobalclimatechange,soilandhydrologicaldegradation,toincreasedbiologicalextinctions.6,7,8,9,10Humanpressuresontheenvironmentareexpectedtoincrease,atleastinthenearfuture,asaresultoftheexpandingglobalpopulationandincreasinglevelsofconsumptionandwasteaccumulation.3Althoughthestakesarehigh,knowledgeoftheprocessesthatunderliehuman-inducedlandcoverchangeisstilllimited.7Thisisanemergingissueinspatiallyexplicitenviron-mentaldisciplinessuchaslandscapeecology.11,12
Theproximatecausesofhuman-inducedchangesinlandcoverarisefrombothbroad-scaleclearingofnaturalvegetationforagriculture,miningandurbandevel-opment,andfromhabitatmodificationsresultingfromforestryandalteredgrazingand fire regimes. However, the ultimate causes are the biophysical and culturalfactors that influence where and at what pace habitat clearing and modificationoccurs,fueledbytheglobalpopulationgrowthandtheconsumptionofresources.7Toreducethehumanecologicalfootprintandencouragesustainablelanduse,weneedtounderstandtheseultimatecauses.
Most human-induced land cover change is directly related to land use. Landuse involves the exploitation of land and its resources and largely represents theinteractionbetweenhumansandthebiophysicalenvironment.1Landcoverchangeoccursatdifferentspatialscales,includingthelocal(102to103km2),regional(104to106km2), and national (106 to 108 km2), or continental and global geographicalscales.Landuse isaprocessdrivenbysocial,economic,political, technological,and cultural factors, but is spatially and temporally constrained by biophysicalconditions such as climate, soils, water availability, accessibility, and biologicalresources. These drivers and constraints, and their interactions, need to be con-ceptualizedandanalyzed inan integratedmultidisciplinary framework ifwearetodevelopanadequateunderstandingofthewaysinwhichlanduseprocessesleadto land cover change. Bürgi et al.12 point out that convincing integrative studiesarestilllargelylacking,withseveralkeyareasrequiringinvestigation:(i)movingfrompattern-orientedtoincreasinglyprocess-orientedstudies,whichincludebroadtime frames and historical approaches; (ii) devising methods for extrapolationandgeneralitybuilding; (iii)findingways to linkdataof different qualities; and(iv)effectivelyincorporatingcultureasadrivingforceofchange.Globalpressuresofhuman-landscapeinteractionsandtheirfeedbackshaveincreasedasaresultoftheworld’ssociopolitical,economic,andculturalglobalizationprocessesduringthelate20thcentury.Becauseofthis,opportunitiesandconstraintsforlandusechange,inparticularnewlanduses,increasinglydependontheglobalfactorsinfluencingmarketsandpoliciesatdifferentorganizationallevels.
Historically, the major changes in global land cover occurred in temperateregions, but land cover change, especially deforestation, is now concentrated intropicalregions,leadingtolargereductionsinnetprimaryproduction.8ItiswidelyrecognizedthatdeforestationratesinLatinAmerica,especiallyinthetropicalrainforestsoftheAmazonBasin,arecurrentlythehighestintheworld.6,13,14AbouthalfofthelandclearinginLatinAmericahastakenplacesince1960,15withtheAmazonBasinaccounting fora largeproportionof recentglobaldeforestation.5Themost
42963_C005.indd 82 11/6/07 7:14:27 AM
Modeling Unplanned Land Cover Change across Scales 83
studied region of the Amazon is the Brazilian Amazon,16,17,18,19,20,21,22,23,24 whichshowsthatmostofthelandclearinghasresultedfromgovernment-orientedcoloni-zationschemesandinfrastructurebuilding.
TheColombianAmazonregion is lesswellstudied,although it iscurrentlyamajor global deforestation hot spot.5,25 Unlike the Brazilian Amazon, clearing isoccurring in an unplanned way, with peasants migrating freely into the regionattractedbyanillegaldrugeconomyandthepossibilitytoownland.OtherregionsofColombiasuchastheAndesandtheCaribbeanarealreadyheavilyimpactedbyhuman land use.26 In Colombia, a growing population and increasing integrationintotheglobaleconomyarebringingincrementalpressuresonnaturalecosystems.Colombiaisinternationallyrecognizedforitshighbiologicaldiversity,bothintermsof species richness and endemism.27,28,29,30 However, knowledge of the processesanddriversofdeforestationandconsequentthreatsarelimited.28Thereneedstobeamuchbroaderanddeeperunderstandingof thespatialand temporalpatternsofchangeandthefactorsdrivingthesechanges.Thisnewknowledgewillallowmoremeaningfulassessmentsoftheimpactsofdeforestationonecosystemsandbiodiver-sityandprovideusefulinputsforlanduseandconservationplanning.
Inthischapterwepresentamultiscalesynthesistocontributetotheunderstandingofthepatterns,processes,anddriversofunplannedlandcoverchangeinthetropics,basedonColombiaasastudycase.31Weaddresslandcoverchangeatthenational,regional,andlocallevels,andanalyzethevariabilityofdrivers,spatialpatterns,andratesofchange,focusingonthespatiallydominantforestecosystems.Atthenationallevel,attentionisdirectedtowardinvestigatingthebroadpresentandfuturelandcovertrendsacrossthemainbiogeographicregions.Attheregionallevel,wefocusonthecolonizationoftheAmazonregion,whiledetailedanalysisatthelocalleveliscarriedoutinsixselectedsampleareasoflowlandhumidtropicalforests.Finally,wediscusstheimplicationsoftheresearchoutcomesforconservationplanning.
5.2 stUDY ReGIon
Colombia covers a land areaof 1.1millionkm2,with largevariations in altitude(0to5,800m),meanannualrainfall(300to10,000mm),lengthofgrowingperiod(60to360daysperyear1),andahighdiversityofgeologicalsubstrates,givingrisetoanunusualhighenvironmentalvariabilityrelativetoitsgeographicsize.Colombianecosystemsrangefromdesertandtropicalsavannastoveryhumidrainforestsandtropical snow-coveredmountains.Asa consequenceof thisvariability,Colombiahashighlevelsofendemismandspeciesrichnessandhasbeenclassedasamega-diversecountry.28,32Thecountrycanbedividedintofivebiogeographicregionswithspecificbiophysicalandlandusecharacteristics(Figure5.1a):Andes(278,000km2),Caribbean (115,400 km2), Pacific Coast (74,600 km2), the Colombian Amazon(455,000 km2), and Orinoco plains (169,200 km2); and two smaller regions, theMagdalena(37,100km2)andCatatumbo(7,000km2),whicharegenerallyincludedintheAndeanregion.
ThetotalpopulationofColombiaiscurrentlyaround40million,25%ofwhichstill live in rural areas. The population has been historically concentrated in theAndean andCaribbean regions.33,34These regions currentlyhave an average rural
42963_C005.indd 83 11/6/07 7:14:28 AM
84 Land Use Change: Science, Policy and Management
populationdensityofapproximately33personsperkm2,whilethePacific,Orinoco,andAmazonregionshavelowerdensitiesrangingfrom5to17personsperkm2.Duringthe20thcentury,thepopulationgrew10-fold,havingastrongimpactonthelandscape.Howeversincethe1970s,Colombiahasbecomeanincreasinglyurbanandindustrial-izedcountry,with thenationalpopulationgrowthratefallingbelow2%in the late1990s,andtheruralpopulationstabilizing.AlthoughthecountryunderwentastrongracialmixingresultinginadominantMestizopopulation,theculturaldiversityisstillhigh,withregionallycontrastingruralculturesof99ethnicgroupswith101languages,varyingfromAmerindiantoAfro-AmericanandEuropean-American.35
The economy is based on mining (oil, coal, and nickel), agriculture (coffee,flowers),andindustrialexports.Inrecentdecades,Colombiahasexperiencedcon-siderablesocialandpoliticalunrest,drivenbyextremistleft-andright-wingarmedforces, triggering large internalpopulationmovements andeconomicdestabiliza-tion.Paralleltothisunrest,apervasiveeconomyofillegalcrops(cocaandopium)forexporthasdevelopedinmanyremotefrontierareas,causingfurthersocialandpolitical instability. These internal social and political pressures have importantconsequencesforregionalpatternsoflandcoverchange,includingaredirectionofcolonizationpatternsandlandabandonmentincertainareas.
5.2.1 Regional level
Theregionalstudyarea(23,000km2)islocatedintheCaquetáDepartmentintheAndeanfoothillsoftheeasternAmazonregionofColombia(Figure5.1b).Itincludes17municipalities(wholeorinpart)withatotalruralpopulationofabout180,000in1993.Populationdensitiesrangefrom7to16inhabitantsperkm2ofclearedland.The
Regional case study
N
a)
Caribb
eanM
agdala
na
Paci
ficAndes
b)Cleared forestsExtant forests
Local‒level case studies
750 Km5002500
Catatumbo
Orinoco
Amazon
FIGURe 5.1 LocationofColombia showing: (a) the seven regions; and (b) extentof theoriginal(gray+black)andremnantforestedecosystems(black),andnonforestedecosystems(white);andlocalandregionalstudyareas(red).
42963_C005.indd 84 11/6/07 7:14:30 AM
Modeling Unplanned Land Cover Change across Scales 85
mostextensivelanduseiscattleranching,withillegalcropsofcoca(Erythroxylumcoca)coveringasmallerareabutincreasingineconomicimportancesincetheearly1980s.Thelatestfiguresaredebated,butindicateadropofalmost50%inillegalplantationsareainColombiafromahighin2001.36
5.2.2 local level
The local-levelanalysis focusedonsix landscapes located in thehumid lowlands(>2,000mmmeanannualrainfall)ofthecentral(Magdalena)andeastern(Orinoco,Amazon, and Catatumbo) regions of Colombia. The landscapes comprise areasrangingfrom68to128km2,whichhavebeensubjecttosubstantiallandclearingduringthepast30to60years(Figure5.1b,Table5.1).Lessthantwomillionpeople(5%)currentlyliveintherurallowlands.Intheseareas,theruralpopulationdensityisverylow,varyingfromless than5to15inhabitants/km2,with lowestdensitiesoccurringintheextensivecattlegrazinglandscapesintheCaribbean,Orinoco,andAmazonregions.
5.3 MetHoDs
Thedatasetsandmethodsdescribedhereinwereusedtoanalyzethespatialpatternsof forest conversion, determine underlying drivers of this process, and predict
table 5.1General biophysical Characteristics and Data sets of the local study sites
study area
area (km2) Region
Dates of air-photo coverage
Mean altitude
(m)
Mean annual rainfall (mm)/
no. dry months (<100 mm)
original vegetation in
landscape
LaBalsa 128 Orinoco 1938,1961,1979,1987,1992,2001
250 2500/4 Forest–Savannamosaic
Guamal 140 Orinoco 1938,1961,1979,1987,1992,1997,2001
300 3000/3 Forest–Savannamosaic
Caquetá 100 Amazon 1946,1975,1985,1992,2000
200 3100/2 Forest
Opón 68 Magdalena 1971,1985,1996,2002
200 3200/1 Forest
Berrío 82 Magdalena 1950,1961,1977,1985,1996,2002
120 3000/3 Forest
Tibú 74 Catatumbo 1961,1975,1985,2000
150 2700/3 Forest
Source: Etter,A.etal.40Withpermission.
42963_C005.indd 85 11/6/07 7:14:31 AM
86 Land Use Change: Science, Policy and Management
forestedareaswithahighprobabilityofconversioninthefuture.Thestudyapplieda multiple-scale modeling approach using logistic and ordinal regression, regres-siontreeanalysis,andgeographicalinformationsystems(GIS).Landcoverchangesweremodeledwithbiophysicalandsocioeconomicvariablestoanalyzeandpredictdeforestationandregenerationprocesses.Foralllevelsofanalysis,theresponsevari-ablewasabinary(1,0)mapofforestcover,whiletheexplanatoryvariablesincludedsoilfertility,rainfall,moistureavailabilityindex,slope,accessibility,andprotectedareasforthenationallevel,andsoilfertility,accessibility,andlandcoverneighbor-hoodofforestandsecondaryvegetationfortheregionalandlocalanalyses.Weusedabroadrangeof informationsources, includingremotelysenseddata fromaerialphotographsandsatelliteimages,andsecondarysourcesofbiophysicalandsocio-economicdata,aswellashistoricaldata(Table5.2).
5.3.1 national level
Atthenationallevel,thestatisticalmodelingcomprisedtwosteps:(i)predictingthespatial location of forest conversion using logistic regression and regression trees;(ii)refiningthepredictionswiththepopulationgrowthratedata.Weemployedbothmodeling techniquesas they treat thespatialvariation in theeffectofexplanatoryvariablesdifferently.37Logisticregressionmodelstheeffectofvariablesinaspatiallyhomogeneousmanner,whereasclassificationtreescantreattheeffectinaspatiallyheterogeneousmanner.Usingtheresultsofthebest-performingmodel,weidentifiedareas with a high probability of conversion as those with a predicted probabilitygreaterthan70%.Weoverlaidtheseareaswitharecentruralpopulationgrowthratemaptoidentifyareaswithahighprobabilityofconversionandapopulationgrowthrategreaterthan2%.Wedefinedtheseareasas“deforestationhotspots.”AllanalyseswereperformedusingS-PLUS.38AqualitativeassessmentoftherelativeriskofforestconversionforthedifferentremnantnaturalforestecosystemswasdonebyoverlayingthepredicteddeforestationhotspotswiththeecosystemmapbyEtter.26
5.3.2 Regional level
Using thebinary forest/nonforestmaps, forestproportionzoningmapswerepro-ducedbysmoothingthedataovera50by50gridcellwindowandthenslicedinto10%intervals.39Toanalyzethespatialdynamicsofdeforestationandregeneration,wecomparedhowdeforestationandregenerationchangedineach10%forestcoverzoneforeachtimeperiod.Inasecondstep,weperformedmapcross-tabulationsofthe10%zonemapsforthethreeperiods(1989to1996,1996to1999,1999to2002)(e.g.,cellsmakingtransitionfromthe70%to80%forestcoverzonetothe60%to70%forestcoverzoneinagivenperiodoftime).Thisallowedustoclassifyzonetransitionsinaspatiallyexplicitwayandtoidentifytheintensityofchange(negativefordeforestationandpositive for regeneration)dependingon thenumberof zonejumpsperperiod.Wethendefinedhotspotsofchangeonlyasthoseareaswith“highspeed”transitions,definedasareasshowingtwoormorezonejumps(e.g.,fromthe30%to40%zonetothe10%to20%zone).
42963_C005.indd 86 11/6/07 7:14:31 AM
Modeling Unplanned Land Cover Change across Scales 87
tab
le 5
.2D
ata
and
Dat
a so
urce
s fo
r th
e th
ree
leve
ls o
f ana
lysi
s
leve
l of
anal
ysis
Dep
ende
nt v
aria
ble
Inde
pend
ent
vari
able
type
Dat
a so
urce
type
Dat
a so
urce
Nat
iona
lB
inar
yfo
rest
/non
fore
stE
cosy
stem
map
(E
tter,
1998
)C
limat
e,s
oil,
slop
e,d
ista
nce
tor
oads
,di
stan
ceto
riv
ers,
rur
alp
opul
atio
ngr
owth
(19
85–1
993)
Inst
ituto
de
Est
udio
sA
mbi
enta
les
-ID
EA
M,2
004;
In
stitu
toG
eogr
áfico
Agu
stín
Cod
azzi
,198
3,1
985,
20
00;I
nter
natio
nalW
ater
Man
agem
entI
nstit
ute
(IW
MI)
,200
4)
Reg
iona
lB
inar
yfo
rest
/non
fore
stL
ands
atE
TM
(19
89,1
996,
19
99,2
002)
Soil,
dis
tanc
eto
roa
ds,d
ista
nce
tor
iver
s,
rura
lpop
ulat
ion
grow
th(
1985
–199
3),
neig
hbor
ing
land
cov
er
Inst
ituto
de
Est
udio
sA
mbi
enta
les
-ID
EA
M,2
004;
In
stitu
toG
eogr
áfico
Agu
stín
Cod
azzi
,200
0;
Mal
agón
eta
l.,1
993)
Loc
alB
inar
yfo
rest
/non
fore
stA
ir-p
hoto
s(1
938–
2000
);
Lan
dsat
ET
M(
2000
,200
2)So
il,d
ista
nce
tor
oads
,dis
tanc
eto
riv
ers,
ne
ighb
orin
gla
ndc
over
Inst
ituto
Geo
gráfi
coA
gust
inC
odaz
zi,1
983,
200
0;
Air
-pho
tos
(193
8–20
00)
42963_C005.indd 87 11/6/07 7:14:32 AM
88 Land Use Change: Science, Policy and Management
5.3.3 local level
Forestdynamics,asquantifiedthroughforestclearingandregrowth,weremodeledandanalyzedusingfivedifferentlogisticregressionmodels,includinganincreasingnumberofindependentvariables,inordertoexplainthespatiotemporalpatternsandsomeofthedriversofchange.40Thepurposeofmodelingwasthreefold:(i)toanalyzethegeneraltemporaltrendsofthedeforestationprocessofhumidlowlandforestsindifferent regions; (ii) assess the relative importanceof thepredictorvariables foreachapriorimodel;and(iii)selectabestapproximatingmodelfromtheapriorisetofcandidatemodels.
5.4 ResUlts
5.4.1 national DefoRestation thReat hot spots anD Main DRiveRs
The best performing model to predict deforestation was the regression tree thatincludedtheregionsasavariable(Figure5.2).Atthenationallevel,themostimpor-tantvariablesexplainingthepresenceandabsenceofforestcoverweredistancetoroadsanddistancetotownsforboththelogisticregressionandclassificationtreemodels.However,theeffectofthesevariablesonthepresenceofforestcovervariedfromregiontoregion.Theeffectofothervariablessuchassoilfertility(Andean,Pacific,Orinoco)andnumberof raindays (Caribbean,Magdalena)are importantexplanatoryvariablesinonlyafewregions.Ingeneral,thefollowingrelationships
N
Low
High
0 . 0.1110.111 – 0.2220.222 – 0.3330.333 – 0.4440.444 – 0.5560.556 – 0.6670.667 – 0.7780.778 – 0.8890.889 – 1No Data
Km5002500
FIGURe 5.2 Predicted forest presence according to the best model (classification tree).(FromEtteretal.50Withpermission.)
42963_C005.indd 88 11/6/07 7:14:33 AM
Modeling Unplanned Land Cover Change across Scales 89
between deforestation and the explanatory variables were observed across allmodels: deforestation was predicted to be greater in unprotected areas that havefertilesoils,gentleslopes,andareneartosettlements,roads,andrivers.Therela-tionshipbetweenthenumberofraindaysandforestconversionwaspositiveintheCaribbeanandAmazon,whilenegativeintheAndeanregion.Theregressiontreemodeladjustedwiththepopulationgrowthdatashowsasetofdistinctgeographical“hotspots”ofpredictedfuturedeforestationareas(Figure5.3).
Thefiveforestedecosystemspredictedtobemostvulnerabletoforestconver-sionaccordingtothepredictedfuturedeforestationhotspotsrankedaccordingtopredicteddeforestationareawere:thehumidtropicalforestsoftheundulatingplains
N
Km5002500
1
2
3
45 6
7
8
FIGURe 5.3 (See color insert following p. 132.) Predicted deforestation hot spotsobtainedbycombiningareaspredictedtohavethehighestprobabilityofforestconversion(>70%)fromthebestmodel(theregion-specificclassificationtree)withtheareaswithgreaterthan2%ruralpopulationgrowthrate(1985–1993).Reddepictsthedeforestationhotspots(areaswith>70%probabilityofforestconversionand>2%ruralpopulationgrowth).Orangeandreddepictareaswith>70%probabilityofforestconversion.Greendepictsforestedareas,grayrepresentsclearedforestedareas,andwhiterepresentsnonforestedareas.Whitecircledareasindicatecurrenthotspotsofdeforestation,whicharealsoareasofhigh-valuebiodiversityvalue:(1)Quibdó-Tribugá,(2)Farallones-Micay,(3)Patía-Mira,(4)Fragua-Patascoy,(5)AltoDuda-Guayabero,(6)Macarena,(7)Guaviare,and(8)Perijá.BlacklineistheAndeanregion,andlightgreenlinesarenationalparks.(FromEtteretal.50Withpermission.)
42963_C005.indd 89 11/6/07 7:14:34 AM
90 Land Use Change: Science, Policy and Management
of thenorthernAmazon, thehumid sub-Andean forests, thehumidhigh-Andeanforests,thehumidmid-altitudeAndeanforests,andthehumidtropicalforestsoftheundulatingplainsintheMagdalena.However,whenriskwasmeasuredintermsoftheproportionoftheremnantecosystemareapredictedtobetransformed,therank-ingof riskchanged to:verydry tropical forests in theCaribbean,humid tropicalforestsof the rolling landscapes in theMagdalena, the tropicaldry forestsof thehills,andthelowlandswampforestsoftheCaribbean.
5.4.2 aMazon colonization fRonts
TheanalysesshowthecolonizationfrontintheAmazonmovinglikeawaveorigi-nating in themainurbancentersof the regionand followingprimarily the rivers(Figure5.4). Between 1989 and 2002, the 90% forest cover boundary, used as aproxy for the “colonization frontline,” moved to the east an average distance of11.2km,representingarateof0.84kmperyear.However,theratevariedbetween0.5to3.2kmperyeardependingonthelocationalongthefront.Thetotalannualnetdeforestationratesshowedapeakof40,400hain1996to1999,increasingfrom18,600hain1989to1996,anddecliningto23,830hain1999to2002.Anaverageof25,400haofforestswasclearedeachyearduring1989to2002,ofwhich3,400haperyearwasclearedintheperforationprocessbeyondthe90%frontline.
The net forest change rate within the colonization front was the result ofthe combined effect of deforestation and forest regeneration and varied with theforestcoverzones.Thehighest rateofdeforestationoccurred in the50% to80%forest cover zones, with rates exceeding 4% per year during 1996 to 1999. Theforestregenerationratepeakedinthe20%to50%forestcoverzones,withannualratesofupto0.9%,suggestingtheprogressiveincreaseofsecondaryforestsagainst
Forest cover zone (%)
Cleared
1989 1996 1999 2002
Forest
0–10 10–20 20–30 30–40 40–50 50–60 60–70 70–80 80–90 90–100
(a)
(b)
FIGURe 5.4 (See color insert following p. 132.) Forestmapsofthecolonizationfrontforeachstudydate:(a)extentofforestcover(black=forest);and(b)percentageforestcoverat10%incrementzones.(FromEtteretal.39Withpermission.)
42963_C005.indd 90 11/6/07 7:14:36 AM
Modeling Unplanned Land Cover Change across Scales 91
“matureforests”astheleveloflandscapetransformationincreases.Thesamepatternwasmaintainedforalltimeperiods.
Highratesofforestcoverchangewereobservedinlocalizedareas,or“localhotspots”offorestlossandforestregeneration(Figure5.5).Someofthesedeforestationhotspotswerespatiallystableduringtheentirestudyperiod,suchastheoneinthenortheastofthestudyarea;othersadvanced;whilesomelocationsshowedintensedeforestationinoneperiodfollowedbyaperiodofintenseregeneration,suchasthesouthwesternpartofthestudyarea.Themoredynamiclandscapeswereinvariablysituatedin theforestcoverzoneswithanintermediateamountofremnantforests(30%to70%).Thenortheastof thestudyarea,whichwasdeclared inDecember1999asamilitaryexclusionzonebythegovernmentforthepeaceprocesswiththeFARCguerrillas,showedthemostrapiddeforestationinthe1999to2002period,whiletherestoftheregionshowedtheoppositetrend.
5.4.3 DefoRestation patteRns at the local level
Theanalysisofthedeforestationtrendinallofthesixlowlandforestareasduringthepast60yearsshowedarapiddeclineoftheforestcover,andwastypicallyfollowedby an increase of pasturelands, initially supporting extensive cattle grazing, butintensifying where infrastructure development permits. However, a major findingwasthat inallcasestheforest lossfollowsanasymmetrical logisticpattern,withfourrecognizablephasesofforestloss(Figure5.6):
Phasei:initialphasewithlowrateofchangePhaseii:middlephasewithhighestrateofchangePhaseiii:midtolatephasewithrateofchangeslowingdownPhaseiv:finalphasewithanapparentnewdynamicequilibriumofre-growth
balancingforestloss
LEGENDRapid deforestationRapid forest regenerationMinor changes
TownsRiversRoads
1999–2002 militaryexclusion zone
1989–1996 1996–1999 1999–2002
FIGURe 5.5 (See color insert following p. 132.) Spatiallocationofthelocalhotspotsofdeforestation(red)andregeneration(green)forthethreetimeperiodsofstudy.(FromEtteretal.39Withpermission.)
42963_C005.indd 91 11/6/07 7:14:37 AM
92 Land Use Change: Science, Policy and Management
Anadditionalresultisthatasaconsequenceofthelogisticdeclinepatternoftheforestcoverinthelandscape,theratesofdeforestationshowaquadraticrelationshipwiththeproportionofforestinthelandscape(Figure5.7a)andadirectlinearrela-tionshipwiththeamountofexposedforestedge(Figure5.7b).Themaximumratesofclearingoccurredatintermediatevaluesofforestproportioninthelandscapeandatthehighestvaluesofexposedforestedgedensities,ofapproximately40mha–1.
Also,when comparing thevarious regions, differences in colonizationwavesand rates of deforestation were evident for this period in the Amazon, Orinoco,andMagdalenaCaribbean.The logistic curvedepicting thedeforestationprocessshiftsintimedependingonwhenthelandclearingprocessbegan.Onaverage,thetimespanbetweenbeginning(phasei)andend(phaseiv)oftheprocessisattainedafter30to40years,atatimewhentheremnantsofforestecosystemsreachvaluesbetween2%and10%oftheiroriginalcover.Inallstudiedlandscapes,thedominantandmorepersistentreplacementcoverwasintroducedpastures,withcropsrepre-sentingaminorproportionoflandusefollowingclearing.
5.5 DIsCUssIon anD ConClUsIons
Our study contributes to broadening the understanding of patterns and drivers oftropicaldeforestationprocessesinunplannedcolonizationfronts.Therearefourmajorcontributionsaboutpatternsandprocessesof landcoverchangeinColombia thatcanbedrawnfromthisstudywithpotentialapplicationinothertropicalregions.
First, lossof forests followsa logisticpattern.Thecasestudies from lowlandforestsinColombiashowthatthedeclineofforestcoverduringtheprocessofdefor-estationconformstoalogisticpattern,reachingasemistabletransformedlandscapestateafterundergoingfourtransformationphaseswithvaryingratesofdeforestation.
FIGURe 5.6 Logisticpatternofforestcoverdeclineduringthe transformationprocessinColombia,showingphasesitoiv.LargeplotcorrespondstoLaBalsacasestudy,withequation: Y = 0.033 + [0.967 / (1 + exp(0.219 * (X – 1947)))]. (From Etter, A. etal.40Withpermission.)
1925 1945 1965 1985 2005
Ori
gina
l for
est c
over
1
0.8
0.6
0.4
0.2
0
i ii iii iv
42963_C005.indd 92 11/6/07 7:14:39 AM
Modeling Unplanned Land Cover Change across Scales 93
Thismeansthattherateofforestlossassumesaquadraticshapewithapeakrateoflosswhenapproximatelyhalfthelandscapeistransformed.
Second, deforestation fronts move as waves. In the unplanned colonizationfronts inColombia, deforestationprogresses in a relativelyuniformpattern at anapproximatespeedof1kmperyear,mimickingawave.Forestcoverdeclineswithintheadvancingwave,followingasimilarlogisticpatternofforestdeclinementionedabove,therebylinkingthelocalandregionallevelsofanalysis.
Third,changesinlandscapestructure,suchasanincreaseinedgedensityperunitarea,facilitatetheclearingprocess.Thereisadirectrelationshipbetweentheexposed forest edge density in the landscape and the rate of deforestation, withmaximumratesofclearingoccurringatthehighestforestedgedensitiesofapproxi-mately40mha–1.Therelationshipbetweenproportionofforestinthelandscapeandforestedgeassumesaquadraticform,withmaximumedgedensityatintermediateforestproportionswhentheclearingrateisalsohighest.Thissuggestsastronglink
Forest proportion in landscape
Forest edge (m.ha–1)
R2 = 0.8819
R2 = 0.7643
Ann
ual d
efor
esta
tion
rate
Ann
ual d
efor
esta
tion
rate
0
0 10 20 30 40
0.2 0.4 0.6 0.8 1
5.0%4.5%4.0%3.5%3.0%2.5%2.0%1.5%1.0%0.5%0.0%
5.0%4.5%4.0%3.5%3.0%2.5%2.0%1.5%1.0%0.5%0.0%
iii
iii
ii
ii
i
i
a)
b)
FIGURe 5.7 ForestcoverchangeatthelocalscaleintheColombianAmazon:(a)Relationbetweendeforestationratesandforestproportioninlandscape;(b)Relationbetweenratesofdeforestationandforestedgedensity.(FromEtteretal.51Withpermission.)
42963_C005.indd 93 11/6/07 7:14:40 AM
94 Land Use Change: Science, Policy and Management
between spatial pattern (forest proportion and edge density) and process (rate offorest clearing) in unplanned tropical colonization fronts. Furthermore, this rela-tionshipalsoconfirmsthebroaderapplicabilityofthelogisticpatternofforestlossacrossanentirecolonizationfront.
Fourth, secondary forests replaceclearedmature forestsandbecomeadomi-nantcomponentofthetropicalforestmosaicinthehighlytransformedlandscapes.Thehistoricalanalysisofdeforestationacrossallcasestudiesdemonstratesthatinhighlytransformedtropicalforestlandscapestheforestcomponentismadeoftwoparts: astable componentofmature remnant forest andadynamiccomponentofsecondaryforestsofdifferentages.Thesecomponentsarespatiallymixed,andformasuccessionalforestmosaic.
Landscapemetrics1,41aremeanttoprovidethemeanstointegrateissueslinkingpatternandprocess.However, landscape indicesarestillmostlyused todescribethespatialpatternsoflandscapechangewithoutmakingalinktoprocessesinthelandscape.42Ourstudyprovidesanexamplewherealinkbetweenlandscapemetrics,suchasforestproportionandamountofexposedforestedges,andthespeedoflandconversionismade.Thelogisticmodelofforestdeclineatthebaseofthesefindingsprovidesan importantgeneralprinciple forexplainingdeforestation inunplannedcolonization fronts (Figure5.8).However, itneeds tobe tested in tropical forestsoutsideColombia.
Studiesandassessmentsonmultiscaleanalysesoflandusemodelshavebeendone for Ecuador43 and Central America.44,45 We provide, for the first time inColombia,amultiple-leveloverviewofdeforestation,whichshouldgivebothgeneral(i.e.,whereareforestslikelytobelostinthefuture)andspecific(i.e.,howandwhyisdeforestationadvancing)guidelinesfordetectingspatiallyexplicitlandcoverchangedynamicsboth for further studiesand todirect theattentionofplanners.Weusespecificdatasetsforeachlevelofanalysis.
LanduseplanninginColombiaislimitedwithregardtotheuseofup-to-dateandrelevantspatialdata,whichiscriticalconsideringthatColombiaisacountrywherethebiologicalnaturalresourcesareanimportantassetandthegainsandlossesfrominappropriate landcoverchangeareveryhigh.Understandingthe levelof threat tonaturalecosystemsisfundamentaltohelpmakemoreinformeddecisionsinconser-vationplanning.46Thegeneralpatternsofdeforestationdescribedmakeasignificantcontributiontothisaimandcanpotentiallybeappliedinothertropicalandsubtropicalcountriesoftheworldwhereunplannedlandclearingoccurs.Thestudyhasspecificrelevancetolanduseandconservationplanninginhighlydynamiclandscapesbecausethemodelshelppredictandanticipatethespatialprogressionofunplanneddeforesta-tionfronts.Theresultsshould,inprinciple,betransferableandapplicabletootherunplannedcolonization fronts in the tropicsandsubtropicsandhelp improve landuseandconservationplanningby: (i)providingmethods tocalculate the threat toforestecosystemsandcharacterizethedynamicsofclearingassociatedwithcoloni-zationfronts;(ii)assistingtheanalysisofcarbonsequestrationandgreenhousegasemissionsbudgets,47,48 through the spatial and temporaldynamicsofdeforestationandforestregeneration;(iii)guidingbiodiversityconservationplanningandrestora-tionecologyinhighlytransformedanddynamiclowlandforestlandscapes,therebyhelpingdiscriminatebetweenrapidlychanging(hotspots)andmorestableareas.49
42963_C005.indd 94 11/6/07 7:14:41 AM
Modeling Unplanned Land Cover Change across Scales 95
(a)
(b)
(c)
FIGURe 5.8 (a)Andeanlandscapeshowinghighlevelsoftransformationwithdominatingcattle grazing land uses. (b) High diversity forests in the Amazon region of Colombia.(c)ColonistforestclearingsinanAmazoncolonizationfrontinColombia.
42963_C005.indd 95 11/6/07 7:14:44 AM
96 Land Use Change: Science, Policy and Management
Weshowedthatanunplannedtropicalforest landscape transformationresultsinseverelyclearedand impacted landscapes.The logisticmodelof forestdeclineprovidesan importantgeneralprinciple forexplainingdeforestation inunplannedcolonizationfronts.However,itneedstobetestedintropicalforestssuchasotherregionsofLatinAmerica,Africa,andAsia.However,animportantquestionremains:does government controlled land-use planning make a difference? An intriguingresearchagendawouldbetoinvestigatetheextenttowhichtheselandcoverchangepatternsandprocessesoccur incountrieswherelandclearingisplanned,suchasBrazil,orpartiallyregulated,suchasAustralia.
aCknowleDGMents
We especially thank Hugh Possingham, Kerrie Wilson, Stuart Phinn, and DavidPullarfortheirinputatseveralstagesofthisresearch.UniversidadJaverianaandtheUniversityofQueenslandprovidedfinancialsupporttoAE.
ReFeRenCes
1. Forman,R.T.T.LandMosaics:TheEcologyofLandscapesandRegions.CambridgeUniversityPress,NewYork,1995.
2. Wackernagel,M.etal.Trackingtheecologicalovershootofthehumaneconomy.PNAS99(14),9266,2002.
3. Vitousek,P.M.etal.HumandominationofEarth’secosystems.Science277(5325),494,1997.
4. Houghton,R.A.Theworldwideextentofland-usechange:Inthelastfewcenturies,andparticularly in the lastseveraldecades,effectsof land-usechangehavebecomeglobal.Bioscience44(5),305,1994.
5. Lepers,E.etal.Asynthesisofinformationonrapidland-coverchangefortheperiod1981–2000.Bioscience55(2),115,2005.
6. Laurance,W.F.Reflectionsonthetropicaldeforestationcrisis.BiologicalConservation91,109,1999.
7. Lambin,E.F.etal.Thecausesoflanduseandland-coverchange:Movingbeyondthemyths.GlobalEnvironmentalChange11,261,2001.
8. DeFries,R.,andBounouaL.Consequencesoflandusechangeforecosystemservices:Afutureunlikethepast.GeoJournal61,345,2004.
9. Fearnside, P. M. Global warming and tropical land-use change: Greenhouse gasemissionsfrombiomassburning,decompositionandsoilsinforestconversion,shiftingcultivationandsecondaryvegetation.ClimateChange46(1–2),115,2000.
10. Myers, N. et al. Biodiversity hotspots for conservation priorities. Nature 403, 853,2000.
11. Wu, J., and Hobbs R. Key issues and research priorities in landscape ecology: Anidiosyncraticsynthesis.LandscapeEcology17,355,2002.
12. Bürgi, M., Hersperger, A. M., and Schneeberger, N. Driving forces of landscapechange—Currentandnewdirections.LandscapeEcology19(8),857,2004.
13. Bilsborrow,R.E.,andOgendo,O.H.W.O.Population-drivenchangesinlanduseindevelopingcountries.Ambio21(1),37,1992.
14. Achard, F. et al. Determination of deforestation rates of the world’s humid tropicalforests.Science297(5583),999,2002.
42963_C005.indd 96 11/6/07 7:14:45 AM
Modeling Unplanned Land Cover Change across Scales 97
15. Houghton,R.A.,Skole,D.L.,andLefkowitz,D.S.ChangesinthelandscapeofLatinAmericabetween1850and1985II.Net releaseofCO2to theatmosphere.EcologyManagement38(3–4),173,1991.
16. Frohn,R.C.etal.Usingsatelliteremotesensinganalysistoevaluateasocio-economicandecologicalmodelofdeforestation inRondonia,Brazil. InternationalJournalofRemoteSensing17(16),3233,1996.
17. Fearnside,P.M.SoybeancultivationasathreattotheenvironmentinBrazil.EnvironmentalConservation28(1),23,2001.
18. Moran,E.F.etal.Effectsofsoilfertilityandland-useonforestsuccessioninAmazo-nia.ForestEcologyandManagement139(1–3),93,2000.
19. Portela, R., and Rademacher, I. A dynamic model of patterns of deforestation andtheireffecton theabilityof theBrazilianAmazonia toprovideecosystemservices.EcologicalModelling143(1–2),115,2001.
20. Skole, D. L. et al. Physical and human dimensions of deforestation in Amazonia.Bioscience44(5),314,1994.
21. Alves, D. S. Space-time dynamics of deforestation in Brazilian Amazônia. InternationalJournalofRemoteSensing23(14),2903,2002.
22. Laurance,W.F.etal.PredictorsofdeforestationintheBrazilianAmazon.JournalofBiogeography29,737,2002.
23. Laurance,W.F.,andFerreira,L.V.EffectsofforestfragmentationonmortalityanddamageofselectedtreesinCentralAmazonia.ConservationBiology11(3),797,1997.
24. Laurance,W.F.etal.The futureof theBrazilianAmazon.Science291(5503),438,2001.
25. Viña,A.,Echavarria,F.,andRundquist,D.C.SatellitechangedetectionanalysisofdeforestationratesandpatternsalongtheColombia-Ecuadorborder.Ambio33(3),118,2004.
26. Etter,A.MapaGeneraldeEcosistemasdeColombia(1:2000000).In:Chaves,M.E.and Arango, N., eds., Informe Nacional sobre el Estado de la Biodiversidad enColombia—1997.InstitutoAlexandervonHumboldt,Bogotá,1998.
27. Myers,N.Biodiversityhotspotsrevisited.Bioscience53(10),916,2003. 28. Chaves,M.E.,andArango,N.,eds.InformeNacionalsobreelestadodelaBiodiversidad
enColombia—1997.I.A.vonHumboldt,Bogotá,689,1998. 29. ConventionofBiologicalDiversity(CBD).GlobalBiodiversityOutlook,2005.Available
from:http://www.biodiv.org/gbo/. 30. Orme, C. D. L. et al. Global hotspots of species richness are not congruent with
endemismorthreat.Nature436(7053),1016,2005. 31. Etter,A.Modelingunplannedlandscapechange:AColombiancasestudy.Dissertation,
UniversityofQueensland,Brisbane,2006. 32. Hernández, J. et al. Estado de la biodiversidad en Colombia. In: Halffter, G., ed.,
LaDiversidadBiológicadeIberoamérica,Vol.I.CYTED,Mexico,1992. 33. Herrera,M.OrdenamientoTerritorialyControlPoliticoenlasLlanurasdelCaribeyen
losAndesCentralesNeogranadinos,SigloXVIII.Dissertation,UniversityofSyracuse,NewYork,2000.
34. Colmenares,G.HistoriaEconomicaySocialdeColombia.5thed.Vol.1,TMEditores,Bogotá,357pp,1999.
35. Loh,J.,andHarmon,D.Aglobalindexofbioculturaldiversity.EcologicalIndicators5(3),231,2005.
36. UnitedNationsOfficeforDrugControlUNDOC.Colombia:Cocacultivationsurvey.UnitedNationsOfficeforDrugControlandGovernmentofColombia,Bogotá,72pp,2004.
37. McDonald,R.I.,andUrban,D.L.Spatiallyvaryingrulesoflandscapechange:lessonsfromacasestudy.LandscapeandUrbanPlanning74(1),7,2004.
42963_C005.indd 97 11/6/07 7:14:46 AM
98 Land Use Change: Science, Policy and Management
38. Insightful-Corporation.S-Plus6.1forWindows.Seattle,Washington,2002. 39. Etter,A.etal.Characterizingatropicaldeforestationfront:Adynamicspatialanalysis
of a deforestation hotspot in the Colombian Amazon. Global Change Biology 12,1409–1420,2006.
40. Etter,A.etal.ModelingtheconversionofColombianlowlandecosystemssince1940:drivers,patternsandrates.JournalofEnvironmentalManagement79(1),74–87,2006.
41. McGarigal,K. et al.FRAGSTATS:Spatialpatternanalysisprogram for categoricalmaps. Computer software program produced by the authors at the University ofMassachusetts, Amherst. Available at: www.umass.edu/landeco/research/fragstats/fragstats.html,2002.
42. Li,H.,andWu,J.Useandmisuseoflandscapeindices.LandscapeEcology19(4),389,2004.
43. deKoning,G.H.J.etal.Multi-scalemodellingoflandusechangedynamicsinEcuador.AgriculturalSystems61,77,1999.
44. Kok, K., and Veldkamp, A. Evaluating impact of spatial scales on landuse patternanalysisinCentralAmerica.Agriculture,EcosystemsandEnvironment85(1–3),205,2001.
45. Kok,K.etal.Amethodandapplicationofmulti-scalevalidationinspatiallandusemodels.Agriculture,EcosystemsandEnvironment85(1–3),223,2001.
46. Wilson,K.etal.Measuringandincorporatingvulnerabilityintoconservationplanning.EnvironmentalManagement35(5),527,2005.
47. Houghton,R.A.Abovegroundforestbiomassandtheglobalcarbonbalance.GlobalChangeBiology11(6),945,2005.
48. Foley,J.A.etal.Globalconsequencesoflanduse.Science309(5734),570,2005. 49. Meir,E.,Andelman,S.,andPossingham,H.Doesconservationplanningmatterina
dynamicanduncertainworld?EcologyLetters7(8),615,2004. 50. Etter,A.etal.RegionalpatternsofagriculturallanduseanddeforestationinColumbia.
AgriculturalEcosystemsandEnvironment114,369–386,2006. 51. Etter,A.etal.UnplannedlandclearingofColumbianrainforests:Spreadinglikedis-
ease?LandscapeandUrbanPlanning77,240–254,2006.
42963_C005.indd 98 11/6/07 7:14:47 AM
99
6 Landscape DynamismDisentangling Thematic versus Structural Change in Northeast Thailand
Kelley A. Crews
Contents
6.1 Introduction...................................................................................................996.1.1 TemporalFrequency:TensionsandLimits...................................... 1006.1.2 FromPatternandStructuretoProcessandFunction...................... 101
6.2 APanelApproach....................................................................................... 1026.2.1 ExtensiontoPatternMetrics............................................................ 103
6.3 ThaiTestingGrounds.................................................................................. 1046.3.1 LocalLessonsLearnedThusFar.................................................... 107
6.4 PaneledPatternMetrics:MeansorEnd?.................................................... 112Acknowledgments.................................................................................................. 115References.............................................................................................................. 115
6.1 IntRoDUCtIon
Landchangeresearchnecessarilydrawsuponaninterdisciplinarymilieuoftheoriesand practices ranging from ecology to geography to policy and beyond; a domi-nantapproachsuccessfullyusedin thisarenaover thepast fewdecadeshasbeenthatofscale-pattern-process.1Choiceofscaleinfluenceswhichlandscapepatternscanbediscerned,inturnusedtoinferprocess.Thenumberofresultinglandscapestudieshaveincreasedsubstantiallyoverthepastdecade.2,3,4,5Assessingsensitivitiesofpatterndetectionandsubsequentinferableprocessestochangesinscale(typicallyspatial resolutionorpixelsize)of remotelysenseddatahasbecomean importantresearchagendaforremotesensingspecialists.6,7Thisworkdrawsinparticularonprinciplesoflandscapeecologythatpositthepossibleimpactsthatscalecanhaveonlandscapecharacterization.8Scaleiscomprisedoftwoprimaryfacets:grain,thesizeofanobservationalunit(e.g.,thedimensionsofasinglepixel),andextent,typicallyrepresentedasthesizeoftheoverallstudyarea.Althoughthesecomponentpartsaretypicallyappliedtospatialscale,theyaseasilymaybeappliedtotemporalscale(e.g.,scaleasthefrequencyofobservationandextentasthetotallengthofstudy).
42963_C006.indd 99 11/6/07 7:13:44 AM
100 Land Use Change: Science, Policy and Management
Implicitintheseargumentsistheseparationoflandscapeconfigurationfromland-scapecomposition.Inotherwords,thespatialassociationofdifferentelementsisasimportantastheoverallproportionofthelandscapeoccupiedbytheelements.Manylanduse/landcoverchange(LULCC)applications,rangingfrombiologyconserva-tion tohydrological assessments to landuseplanning,nowroutinelyprovide thisdecoupledinformation.4Thisworkreviewsthesuccesses,limitations,andpossibili-tiesofenrichingLULCCresearchwithincreasedtemporalgrainorobservationalfrequencyforextricatingcompositionalor thematicchangefromconfigurationorstructuralchange.AcasestudyfromNortheastThailand isused to illustrate thispairedapproach,underscoringtheneedtofurtherdevelopandrefinethismethodinecosystemsfromelsewhereonthenaturallyoranthropogenicallydrivenspectrumandwithvaryingdegreesofspatiotemporalheterogeneity.
6.1.1 Temporal Frequency: Tensions and limiTs
Although most scale-pattern-process work has focused on spatial scale, temporalscalehasnonethelessbeenexplicitlyincludedintheoreticaldiscussionsevenifseldomanalyzed.4,9 Environmental remote sensing defines scale more broadly to includespatial,temporal,spectral,radiometric,anddirectionalscales.10Spatialandtemporalscaleareparticularly importantwhenextracting thematicdata for satellite image-basedchangedetection.11Spatialscale,bothgrainandextent,isregardedasamajorinfluenceondetectionanddefinitionoflandscapepatterns.12,13,14Temporalfrequency,orthetimescalebetweenavailabledataacquisitions,islessstudiedinLULCCwork,inlargepartduetothelimitedavailabilityofhighqualityandhighresolutionmulti-temporalimagesequences.6,15,16Thetemporalgrainofimagery,thoughtypicallynotreferredtoassuch,hasbeenexaminedinenvironmentswhereseasonality(whetherduetophenological,climatic,oranthropogenicchanges)caninterferewithassessmentof longer-term(read: interannual)LULCC.17,18,19,20The temporalextentofLULCCprojectstypicallydefaultstoeithertheearly1970s(concomitantwiththe1972launchofEarthResourceTechnologySatelliteorERTS1,laterrenamedLandsat1)or,inafewcases,toafewdecadesearlierwhenmilitaryreconnaissanceaerialphotographywas available.Thenecessarily truncated temporal extentof these studiespresentsproblemsinestablishingbaselines,acriticalissuegiventhenecessityofdeterminingwhatchangehasoccurredandplacingitintheappropriatehistoricalcontext.21,22Thetermspatiotemporal scaleordomain iswidelyusedwithin theLULCCmodelingcommunity,butthisdescriptionmaycausesomeconfusion.Thecoinageofthetermpresentsanunderstandablecommitmenttoconsiderhowlandscapeschangeacrosstimeandspace,thoughcurrentlythestateofthesciencetendstomodelspatialinter-actionsovertimeratherthanofferingapathfordigitallyrepresentingaspatiotemporalscaleasinteractiveratherthanonlycombinatory.
Compounding the confusion is the dialogue concerning the use of the termslandscape scale and landscape level. For geographers, the term landscape scaleconnotesacertainsize(spatialextent)ofastudyarea—largerthanaplot,smallerthan a continent,4 and the term level is often used to denote study across spatialscales,23whetherornotsuchworkisspatiallyexplicit(e.g.,multilevelmodeling24).For many ecologists who tend to focus first on biotic components of landscapes
42963_C006.indd 100 11/6/07 7:13:45 AM
Landscape Dynamism 101
(e.g.,populationsandcommunities),thetermlandscapescaleisnonsensical,sincedepending upon organism size and range, a landscape can be incredibly small(considermicrobeslivinginasmallpuddle)oraslargeasaplanet.25Thistensionmanifests itself in incongruities and inconsistencies in theuseof these terms (aswellaswhether theyareseenas interchangeableornot),a troublesomeglitchforLULCCscholarsdrawinguponecologythroughthelensof landscapeecology.1,26ForthepurposesofdiscussingLULCCinthiswork,thetermlandscape scalewillbeusedtoconnotespatialandtemporalgrainandextentcommonlyusedinLULCCwork.Thetermlandscape levelwillbeusedtorefertoorganizationalortheoreticalconstructswherethelandscapeliesonaspectrumoffunctionalunits,rangingfrompatchestolandscapestometalandscapes.27
Notetheaboveissuesrevolveprimarilyaroundspatialscale;rarelydoLULCCpractitionersmentionalandscapescalewhenreferringtoacertaintime,asopposedto those studying longer-term landscapes (e.g., ingeomorphology, sedimentology,or palynology). Temporal matters are receiving more scholarly attention of late,particularly in both empirical and process-based modeling efforts.28 LandscapesintemporallyshallowLULCCstudiesarebeingincreasinglyconsideredasactingupon their previous incarnations22 and seen, therefore, as temporally contingentuponthosepastdrivers.Pathdependenciescanandarebeingtrainedintomodel-ing scenarios, and presumably other temporal analogues of spatial concepts willbeoperationalized(e.g.,spatialneighborhoodeffectscouldbeusedasamodelformoresophisticatedrepresentationsofpathdependencyviatemporalneighborhoodeffects). In spatialneighborhoodeffects, it isunderstood that theprecise locationoftheneighborrelativetotheareaofinterestisoftenunimportantaslongasthatneighboriswithinacertainthresholdeddistance.Sowhileanalysismaytakeplaceinaspatiallyexplicitenvironment,conditionsorrulescanbewrittentoloosenthatexplicitnessandquery,forexample,forneighborswithinacertainspatialdistance(withoutregardtothatdistance),withoutregardtodirection.Theparallelintemporalstudieswouldbetorelaxtheassertionoftemporalexplicitnesssuchthatitmaynotmatterwhenaparticularprecedingeventhappened,onlythatitdidhappenorhowoften ithappened.Although temporalmodeling is fairly straightforward in termsofassessingcausality(giventhepresumptionoflineartimemovingonlyforward),thechallengeremainstosortoutfrommyriadperiodicitiesoflandscapedriversandchange,whichareimportantenoughinanygivenlandscape,andhowthentobestdefineatemporallandscapescale.
6.1.2 From paTTern and sTrucTure To process and FuncTion
Several decades of LULCC research have shown that understanding landscapechangerequiresdetectingchangesinbothcompositionandconfiguration.6,8Typi-callythesecomponentsareassessedsequentially:first,alandscapeisclassifiedintoathematiclanduse/landcover(LULC)schemeforatleasttwotimes;second,theconfigurationofeachofthoseclassifiedlandscapesarequantifiedthroughsometypeofpatternassessment(oftenpatternmetrics29);andthird,postclassificationchangedetectionisperformedonthosethematicclassificationstoproducethematicchangemap(s).30Althoughtheprocessusuallystopsthere,someresearchershavealsothen
42963_C006.indd 101 11/6/07 7:13:45 AM
102 Land Use Change: Science, Policy and Management
quantified the configuration of the change map(s) with pattern metrics as well,11thoughconcernsoferrorpropagationhavelimitedthisapproach.31Theimportanceofascertainingspatialstructure(andchangesinsaidspatialstructure)stemsfromlandscapeecology,wherespatialconfigurationfacilitatesandmitigatestheflowofenergyandmaterialsacrossthelandscape.8Thatis,thelandscapeinteractionsthatbothcauseandaremanifestedaslandscapechangenecessarilyoccurinspace,andlocationmatters.Thatisnottosayallchangesoccurasdiffusivemovementssince,dependinguponthevectorofmovement,energyormaterialsmaybeimpartedbyjumpingorpercolatingacrossthelandscape.4Definingthetemporalnatureofspatialstructurewillassistintakingthesemeasurementsandconvertingthescale-patternintoprocess.
Process can be defined in two primary ways. The first will be referred to asdynamics, and it is a mechanistic concept rooted in patterns of change, growth,and activity; this definition is embraced by the Geographic Information Sciencecommunity(GISc)studyinglandscapedynamics,andfitsthenecessarilypiecemealfashion by which LULC is extracted, studied, and modeled. The second type ofprocess will be referred to as dynamism, which is a more gestaltic concept thatinvolves continuous change, growth, or activity; this definition comes from theecology community (particularly landscape ecology) and fits the more continualnatureoftheprocessesstudiedbyecologists,whetherparticulartolandscapestudiesornot.32Thenuanceofthedifferenceinthesetwoapproachesisslight,buttheimpli-cationsareeasilyobservableinthevaryingoperationalizationofbothepistemologyandmethodologynowevidencedinlandscapechangestudiesfromthesetwocom-munities.Here,panelanalysisofLULCandpaneledpatternmetricsareofferedasonemethodofbridgingthisgap,suggestingthatLULCCscholarsshifttowardanapproachofunderstandinglandscapedynamismviaimprovedassessmentofLULCCdynamics.11,27Thatis,improveddescriptionofmechanicsshouldleadtoimprovedexplanationandpredictionofprocessand,perhaps,function.
6.2 A PAneL APPRoACH
Panelanalysissimplyreferstoalongitudinalmethodwherebyunitsofanalysisareheldconstant.Longusedinpsychologyandsociology,panelorlongitudinalanalysisfollowed the same subjects over time (as opposed to a census or cross-sectionalapproach,wheredifferentsubjectsareevaluatedineachobservationperiod).Tech-nically,allfrom-toremotesensing-basedchangedetectionispanelanalysis,sinceeachpixelor instantaneousfieldofview(IFOV)isfollowedindividually throughtime,presumingaccurategeometricrectification.30However,from-tochangedetec-tionusuallyisperformedonpairsofimages,whereaspanelanalysis(inLULCC)isnowusedtorefertoatimeseriesofthreeormoreclassifications.33,34,35Inpanelanalysis,pixelhistoriesortrajectoriesareconstructedthatmaintaintheentiretem-poralpatternofLULCinordertorevealgreaterinformationaboutprocess(es)behindobservablepatterns.Forexample,considerahumidtropicalareaclassifiedonlyintoforest(F)andnonforest(N)andobservedovertwodecadeseveryotheryear.Withpanel analysis, trajectories that might be calculated would include those suggest-ingsemipermanentdeforestation(e.g.,F-F-F-F-N-N-N-N-N-N-N),deforestationand
42963_C006.indd 102 11/6/07 7:13:46 AM
Landscape Dynamism 103
successionalregrowth(e.g.,F-F-F-F-N-N-N-N-N-F-F),afforestationorreforestation(e.g., N-N-N-N-N-N-N-F-F-F-F), silviculture of fast growing tree species (e.g.,F-N-N-N-F-N-N-N-F-N-N-N),orfallowcycling(e.g.,N-F-F-F-F-F-N-F-F-F-F-F-N).Withtraditionalfrom-tochangedetectionofthefirstandlastyears,thosetrajectorieswouldhavehadtheirchangecharacterizedasfollows:semipermanentdeforestationwithF-Nwouldstillbecalleddeforestation(correct);deforestationandsuccessionalregrowthwithF-Fwouldbecalledstableorpermanentforest(incorrect);afforesta-tionorreforestationwithN-Fwouldstillbecalledassuch(correct);silviculturewithF-Nwouldbecalleddeforestation(incorrect),andfallowcyclingwithN-Nwouldbecalledpermanentnonforest(incorrect).UltimatelythepanelapproachtoLULCCdoesnothingtoimproveattributionofclassesthatarestableovertime,andlittletoimproveattributionofclasseswhosechangeisunidirectional.Butlandscapecompo-nentsthatundergoveryquickchange,cyclethroughmultiplestages,switchbetweentwoormoreclassesfrequently,orareinfluencedbyrelativelyshort-termphenomena(e.g.,seasonality)areopentobettermultitemporalcharacterization.Thatis,panelanalysisimprovesourabilitytodetectthekindsofchangethatLULCCresearchislargelydesigned tocapture,model, andmanage;bycorollary, traditional from-tochangedetectionisbiasedtowarddetectingstable,slow-changing,orunidirection-allychangingclasses.Asthenumberofclassificationsinthetimeseriesincreases,quite obviously the ability to detect greater nuanced or more quickly switchingchangeincreases.ThequestionforLULCCprojectsthenishowmanyimagesareenough?Thetextbookansweristhatitdependsuponthetimefootprintoflandscapeprocessesonthelandscape(e.g.,humidtropicalforestsreachsuccessionalcanopyclosuremorequicklythantheaveragetemperateforest);thepracticalansweristhatitdependsonhowmanyqualityimagesareavailableinanareagivenatmosphericinterference,sensorproblems,costofacquisition,andaccesstoarchives,tonameonlyafewoftheproblemsfacingtheLULCCcommunity.
6.2.1 exTension To paTTern meTrics
Thoughpatternmetricanalysisistypicallyoutputasstatisticsatthepatch,class,andlandscapelevels,somepackagessuchasFragstats29allowforoutputtingpatch-basedimageswherebyeachpatch(fromwhichallpatch,class,andlandscapestatisticsaregenerated)ismappedwithauniqueidentifierorobject(whethercomputedinrasterorvector,bitdepthlimitationsnotwithstanding).Thegoalofpaneledpatternmetricanalysisistoassessthechangingstructureoflandscapepatcheswithoutregardtothematic class. That is, in building pattern metric panels we explicitly choose toexaminethenatureof,forexample,fragmentationwithoutregardtowhether it isanurbanarea,forestedexpanse,oragriculturalfieldthatisbeingfragmented.Bydoingso,theexplicitcontributionofconfigurationasopposedtocompositioncanbetested,assessed,andmodeled.Currentresearchatthispointhasfocusedontheformulation and sensitivity analyses of paneled pattern metrics, and this methodrequiresfurthertestinginotherecosystemsandlandscapeswithdifferinglevelsofspatial,temporal,andspatiotemporalheterogeneity.Incaseswheretherobustnessandsensitivityofthepaneledpatternmetricmethodisvalidated,thenextstepisto
42963_C006.indd 103 11/6/07 7:13:47 AM
104 Land Use Change: Science, Policy and Management
notonlytesttheseparateimpactsofcompositionandconfiguration,butalsotheirinteractionandconfoundingaswell.
Theconstructionofpaneledpatternmetricsfollowslogicallyfrompanelanalysis,andtheentirepanelmethodispresentedinFigure6.1.First,atimeseriesofimageryiscategorized into thematicclassifications;aminimumoffour temporalobserva-tionsissuggested,thoughifpatchboundariescanbederived,generated,orfoundelsewhere,threeclassificationsmaysuffice(inabsenceofpreexistingpatchdelinea-tions,abaselineyearofthetimeseriesisused,requiringthreefurtherclassificationsformovingbeyondtraditionaltwo-imagechangedetection).Fromtheseclassifica-tions,apanelLULC iscreatedasdepictedandasdescribedabove.Additionally,patternmetricanalysis isrunoneachclassification,outputtingbothstatisticsandpatchimagesforallobservationsforeachmetricofinterest.Forpurposesofthisdis-cussion,presumethemetricofinterestistheinterspersion/juxtapositionindex(IJI).Changeimagesbetweenconsecutivepairsofpatchimagesarecalculatedandmayinitiallybeleftasfloatoutputbutmusteventuallybebinnedintocategoriesofchange(e.g.,increaseby>20%,increaseby10%to20%,increaseby5%to10%,changeby±5%,decreaseby5%to10%,decreaseby10%to20%,decreaseby>20%).Oncebinnedappropriately,thechangebetweeneachsetofIJImetricimagesisstackedtobuildatrajectoryofchangeatthepatchlevelandthenexportedtoindividualpixelsandbuiltback toafinalmappedproductofpaneledpatternmetricsoutputat thepatchlevel.11Theprocessisrepeatedforeachmetricofinterest,witheachmetricbinnedaccordingtoappropriatehypothesizedorobservedthresholdsorflippoints.
Currentlyboundedorconstrainedmetricshavebeentestedinordertolimitthesubjectivityinvolvedincategorizationofthemetricoutput.Thatis,metricssuchasIJI,doublelogfractaldimension,andpercentagelandscapeall—asoperationalizedinFragstatsandotherpatternmetricprograms—havetheoreticalboundswhereboththeupperandlower limitsareknown.Unboundedorunconstrainedmetrics(e.g.,meanpatchsize,showninFigure6.1forcontrast)presentgreatersubjectivityincat-egorizationsincethereisnotheoreticallimitforthesemetrics(thoughinanygivenlandscape and with any given classification scheme an empirical limit obviouslyexists).Ascurrentlywritten,thepaneledpatternmetricalgorithmpresumesequalintervals between time steps since the original time series used for testing metthoseconditions;modificationtoaccountfordifferingtimelagsiseasilydoneviaaweightingmechanismoncecategorizationthresholds(numberandplacement)havebeendetermined.As such, themethod is suitable forboth interannual and intra-annualanalyses.
6.3 tHAI testInG GRoUnDs
The concern over interannual and intraannual LULCC stems from building thisapproach in an environment with strong phenological, climatic, and anthropogenicseasonalpulses,renderingassessinglonger-termLULCCproblematicwhenanythingbut anniversary date imagery was used for deriving LULC information. NortheastThailandishometoaregionknownasIsaan,wheretheformerNangRongdistrictresides(duetogrowthandredistrictingthisareanowincludesnotonlytheNangRongdistrictbutalsoNonSuwan—denotedonsomemapsasNonguWuan,Chamni,and
42963_C006.indd 104 11/6/07 7:13:47 AM
Landscape Dynamism 105
IJI = 14.2MPS = 8.1
IJI = 17.9MPS = 4.7
IJI = 23.7MPS = 2.2
IJI = 19.2MPS = 3.1
(2)
(2a)
(2b)(3a)
(3)
(1)
FIGURe 6.1 (See color insert following p. 132.) The panel process, conducted atboth the pixel and patch levels: (1) four multispectral satellite images are each catego-rized intoa thematicLULCclassification; (2)patternmetricsare runoneachof the fourLULCclassifications,eachproducingasetofpatch,class,andlandscapestatistics(heretheinterspersion/juxtapositionindex[IJI]andmeanpatchsize[MPS]areshown)aswellasanoutputimageofthedelineatedpatches;(2a)patternmetricoutputforeachofthefourtimesisusedtocalculatethreepiecemealchangemapsforeachpatternmetricandeachconsecu-tivepairofimages(e.g.,showingfluctuationsinIJIorMPSbetweentwotimeperiods)asperCrews-Meyer11,27;(2b)threepatternchangemapsarestackedintoonepanelofallstruc-turalchangeforeachgivenmetric(e.g.,showingfluctuationinIJIorMPSthroughalltimeperiods)asperCrews-Meyer11,27;(3)threethematicchangemapsarecreatedforeachofthetimeperiodsrepresentedbythefourclassifications;(3a)thethreethematicchangemapsarestackedtorepresentthefullrecordofallthematicchangeacrossthefourclassificationsasperCrews-Meyer.3
42963_C006.indd 105 11/6/07 7:13:49 AM
106 Land Use Change: Science, Policy and Management
ChalermPrakeat;thisworkwastestedprimarilyincurrentdayNangRongandNonSuwan).SituatedinbothBuriramProvinceandthenorth-flowingMekongRiverDeltasystem,theareaisthepoorestareaofapoorcountry36,37anddominatedculturally,ecologically,andfinanciallybyastrongmonsoonalpulse,poorsoils,38andconcomi-tant lowland wet rice production.39 Villagers typically live in a nuclear settlementpattern(seeFigure6.2),withresidenceslocatedinlowlandwoodedremnantsandricefieldsradiatingoutinmostdirectionsforthetypical2to5kmdailywalktofields.40,41ThoughthisareawasnotinfluencedbytheGreenRevolution,agriculturehasdriventheconversionofthelandscapeinitiallyopenedbymilitaryroadbuildingeffortsandfacilitated by the gradual building toward a market economy.37 Wet rice replacedsavannainthelowlands,whiledrought-deciduouscropssuchascassavaandsugar-canefollowedthe1970sfactorpriceincreaseintotheuplanddrydipterocarpforests.Followingacurrencycollapse in the late1990s,manyyoungadultswhotypicallymigratedtoBangkokortheeasternseaboardforlaborreturnedtothedistrictatthesametimethegovernmentunderwentanewwaveofdecentralizationacrossfederaltolocallevels.40Anincreasinglydensenetworkofroadbuildingandwaterimpound-ments,33combinedwithpoorenvironmentalmanagement(e.g.,lackofdrainingriceirrigation waters increases soil salinity), has compounded the intensification cycleseeninpartsofSoutheastAsiaandelsewhere.Althoughtheselonger-termdynamicshavebeendocumentedthroughanextensivehouseholdandcommunitysurveyseries
FIGURe 6.2 Typical nuclear village settlement as seen in 1:50,000 scale panchromaticaerial photo from 1994, with approximate settlement boundary indicated. Note remnantforestpatchesusedforshaderelief,andricepaddysurroundingvillageradially.
42963_C006.indd 106 11/6/07 7:13:50 AM
Landscape Dynamism 107
aswellasremotesensingandgeographicalinformationsystems(GIS)analyses,theseasonalpulsesalsodetected(whenimagery,fieldwork,andweatherpermit)cancausedetectablelandscapechangeaslargeinmagnitude(ifnotecologicalimportance)astwodecadesofinterannualchange.18,19Thepresenceofamonsoonallydrivenclimateaddstothelogisticalproblemsofobtainingcloud-freeimageryforderivingLULCinformation.However,adeep timeserieshasbeenestablishedaspartofa largerprojectandhasprovenmorethanadequatefortestingthepanelLULCandpaneledpatternmetricmethods.41,42Figure6.3 illustrates interannual trends inLULCCinthelargerstudyareaovera25-yearperiod;easilydiscerniblearetherapiddeclineinmorehighlyvegetatedLULC(particularlyintheuplandsouthwesternsection)andtheexpansionofriceintothelowlandsavannas.
6.3.1 local lessons learned Thus Far
Figure6.4illustratesastylizedrepresentationoffourLULCclassesandtheircompo-sitionalchangeovertimeasobservedand/orreportedelsewhere.Figure6.4ashowsthe interannualor longer-termchange in forest (primarilyuplanddrydipterocarpandgallery remnant forestsalong ripariancorridors), savanna (primarily lowlandgraminoids with some standing trees), wet rice agriculture, and other agriculture(uplandordroughtdeciduouscropsandcashcrops,includingcassava,kenaf,jute,andsugarcane).33These“real”changescanbecontrastedwiththestylizedrepresen-tationofintraannualchangeinagivenyearduetopreviouslymentionedseasonalityshowninFigure6.4b.ThisgraphisorderedbytheThaiwateryearthatrunsApril1throughMarch31,withearlymonsoonalshowers(knownasmangorains)commenc-inginMayandfollowedbyseveralmonthsofheavyprecipitationthatisextremely
(a) (b) (c)
FIGURe 6.3 (See color insert following p. 132.) (a)LULCinthegreaterstudyareainthe1972/1973wateryear;(b)1985;and(c)1997.
BackgroundHigher Density ForestLower Density ForestSavannaBare SoilRice AgricultureMixed AgricultureCash Crop AgricultureWater
42963_C006.indd 107 11/6/07 7:13:53 AM
108 Land Use Change: Science, Policy and Management
variable in both time and space; rice is typically harvested in late November orDecember,withfieldsburnedusuallyinJanuaryandthedriestmonthsendingthewateryear.These“changes”arepartreal(e.g.,phenologicalchangewithagriculturalcropsordeciduouscycles)andinpartartifact(e.g.,green-upfromshowerswithoutactualcanopyorbiomasschange).
Typical forest changes in this part of northeastThailand represent a familiarstory:fromthe1970sthroughthe1990s,forestsgenerallydeclined(asdidsavannas)duetoagriculturalextensification.Anearlyriseinotheragricultureintheuplandsattheexpenseofforests(nowrelegatedtoextremelythinripariancorridorsandsmallremnantsatopthemostuplandsitesonvolcanicsoils)wasfollowedbyasharpriseinwetriceagricultureinthelowlandareas.Villagesettlementandexpansionoccurintheselowlandareasaswell,althoughtheseareasaccountforlittlechangeintermsof
FIGURe 6.4 (a) Stylized LULC trends observed and/or reported in Northeast Thailandfrom the1970s to late1990s (annualchange,holdingseasonalityconstant). (b)Thesametrendswithinagiventypicalyear(intraannual).
Time (Interannual)
Perc
ent L
ands
cape
504540353025201510
50
ForestSavannaRice Ag.Other Ag.
ForestSavannaRice Ag.Other Ag.
0
10
20
30
40
50
60
Time (Intraannual)
Perc
ent L
ands
cape
(a)
(b)
42963_C006.indd 108 11/6/07 7:13:54 AM
Landscape Dynamism 109
spatialfootprint.Sometradeoffissuggestedbetweenwetriceagricultureandotheragricultureindryversuswetyears,likelyexplainedbyterraceposition(notanthro-pogenic rice terraces,butfluvial terracesormiddleelevationgrounds thatdonotfloodeachyear)sincericemaybeplantedatslightlyhigherelevationsinwetyearswithcashcropsoccupyingthoseareasindryeryears.Overalltheinterannualcom-positionalchangesexhibitsomesharpincreasesordecreasesovertimebutrepresentfairlystabletrajectories(notethisgraphisoverallcomposition;aperpixelcompari-sonwouldrepresentmuchmoreswitchingamongclasses,andagreaterclassificationschemedepthormoveawayfromananniversarydate/stylizedrepresentationwouldshowmuchgreaterfluctuationaswell).
Theintraannualrepresentationdepictsmuchmoremarkedfluctuationsincom-positionof these fourLULCclasses.Forests appear tohavegreater coverdue tophenologicalchanges(green-upfromrains)andlessercoverduringdeciduousevents.Rice agriculture changes dramatically with crop calendar and related monsoonaltiming,asricepaddymovefromfloodedtoplantedtofloweringtoharvesttoburn-ingtobarren;sotoodolandsofotheragriculturechange,althoughtoalesserextentthan rice owing to the drought resistant nature of some of these crops. Savannacoverappears tochangedramaticallyaswellanddoessoinresponsenotonlytogreen-upofgrassesbutmoreover toallagricultural landsspectrallymixingwithgrassysavannasduringdryerperiods.Themagnitudeofintraannualchangecom-paredtointerannualchangeunderscorestheneedforanniversarydateimagery,butalsoforunderstandingtheprocessimplicationsoftheperiodicitiesofdifferenttypesofchange.Withoutaphysicalprocessguideintermsofthecriticalityofacertainlossorgaininonecovertypeoveranother,perhapsLULCCscholarscanatleastbracketinterannualchangeinthecontextofseasonaldynamicstypicallyobservedtounderstandwhenthelandscapehaschangedbeyondits“natural”resilience.9,43
Figure6.5movestotheconsiderationofstructuralchangebyshowingtwotypi-calmetrictrajectoriesforinterannualchanges,againfromthe1970stothe1990s.Figure6.5agraphstheIJIforthesamefourclasses,andfromthisillustrationthelandscapenarrativequicklybecomesapparent.AmappedviewofthisforasubsetofthestudyareaisalsopresentedinFigure6.6,astypicallyininterpretationbothpatch-derivedstatistics(e.g.,Figure6.5)andmapsofchangeinconfiguration(Figure6.6)areused.Here,bothforestcoverandwetriceagricultureexperiencedasharpincreaseinIJIfollowedbyasharpdecrease,indicatingincreasesininterspersionfollowedby decreases. Taken together, these two trends exhibit evidence of an importantecologicalchange in the landscape,andone thatwhenmappedshowselevationaldifferences. Forested areas, notably in the southwest in Non Suwan district, hadbeenthematrixordominantclassinuplandareasintheearly1970s,butbecameincreasinglyfragmentedand interspersedasotheragriculturewas introduced.Bytheearly1990s, the foresthadbeendesiccated to littlebut remnantpatches, andalthoughstillinterspersedwithotherLULCtypes,theseforestedpatchesweresosmallthatthemetricplummetsaslessandlessforestedgeremainstoneighborotherLULCtypes.Asimilartrendofriceagricultureoccursinthelowlandareasbutfortheoppositereason:riceexperiencesasharpincreaseininterspersionasitbecamemoreandmorewidespread,untilbythelate1980sitbecamesoubiquitousthatitsspatialcohesionresultsinlowerIJIscores.SmallerchangesinotheragricultureIJI
42963_C006.indd 109 11/6/07 7:13:55 AM
110 Land Use Change: Science, Policy and Management
supportthehypothesisthatnotonlythecompositionbutalsotheconfigurationofthisclasschangesinthemiddleelevationareas,whileexperiencingsomechangesinuplandareasattheexpensesofforestedlands.Savanna,particularlyintheeasternportion of the area, becomes increasingly interspersed with wet rice agriculture,particularlyinareasmoreproximatetotheprimaryriverchannels(i.e.,thatfloodthemostfrequently).Thislandscapenarrativeisbolsteredbyinterestinglyparallelchangesinmeanpatchsize(MPS)(Figure6.5b),wheretypicalagriculturepatches(regardlessoftypeortopographicposition)increaseasagricultureoverwhelmsthelandscapeatthespatialexpenseofbothforestandsavanna.
Movingbeyondthecombinedchangesincompositionandconfigurationtocon-figuration only11,27 illustrates trends in structure related to topography and acces-sibilitynotdetectableintheaboveefforts.First,wavesoffragmentationareeasilydiscernibleinthesouthwesternoruplandareasaspeopleliterallymovedupthehillsfromsurroundinglow-lyingareas.Typicallyuplandcropsarelesslaborintensive,
Inte
rspe
rsio
n / J
uxta
posit
ion
Inde
x 30
25
20
15
10
9876543210
(a)
(b)
Mea
n Pa
tch
Size
(Hec
tare
s)
Time (Interannual)
Time (Interannual)
ForestSavannaRice Ag.Other Ag.
ForestSavannaRice Ag.Other Ag.
FIGURe 6.5 (a)StylizedLULCpatternmetricchangefortheinterspersion/juxtapositionindex (IJI) observed and/or reported in northeast Thailand from the 1970s to late 1990s(annualchange,holdingseasonalityconstant).(b)Metricoutputformeanpatchsize(MPS)forthesametimeandlocation.
42963_C006.indd 110 11/6/07 7:13:57 AM
Landscape Dynamism 111
(a)
(b)
(c)
FIGURe 6.6 (See color insert following p. 132.) (a) The change in configurationfrom1972/1973to1975/1976,revealingthattheentiresubsetareahasexperiencedagreaterthan10%increaseininterspersion/juxtapositionindex(IJI)scores(duetoincreasedfragmentationandconcomitantinterdigitation).Notethatmostoftheareaexperiencedthesametypeofchange.(b)Illustrationofadifferent trendbetween1975/1976and1979,wherebyincreases,decreases,andrelativestabilityinIJIvaryspatially.Moreuplandareas(mostcentral in thesubset)experiencedaconsolida-tionon the landscape,whileperipheral areas remain relatively stable in termsofconfigurationwithnotableexceptionsonthesoutheasternperimeter.(c)IllustrationofthecontinuedspatialheterogeneityinIJI,withlowland/peripheralareasundergo-ingcontinuedfragmentation,whilethelessaccessible,uplandareasappeartohaveleveledoff intermsoflarger-scalefragmentationorconsolidationbutcontinuetoexperiencesmallpocketsoffragmentationthroughout.
42963_C006.indd 111 11/6/07 7:13:58 AM
112 Land Use Change: Science, Policy and Management
anddonotrequireproximitytothenuclearvillagesettlementsthatricepaddydo.Thus,whenfactorpricesforcassavaincreasedinresponsetoEuropeandemandforcattlefeed,itwasrelativelyeasyforlowlandvillagerstoplantuplandareasnotpre-viouslyclaimedgiventhattheyhadbeenseenasundesirableforriceorresidence,giventheir(uphill)distancetorivers.Interestingly, thesouthernedgeof thisareadidnotseeanadvancementofagricultureasdidtheotherareas,possiblyduetothatside’sproximitytotheKhmer(Cambodian)borderandrelatedhistoryofmilitaryviolenceandminefields.40Inthelowlandareas,thetrendsovertimerespondmoretoaccessibility toripariancorridorsandthusassuranceofyearlyfloodingfor thebestricepaddylands.27Structurally,though,theuplandandlowlandareaswereverysimilar,despitethedifferentdominantLULCtypesandvariousdriversofLULCC.First,intermsofIJI,thecoreareas(eithermostcentralintheuplandareasorclosesttoriparianchannels)exhibitedthegreatestpossiblefluctuationinIJIacrosstime.RegardlessofLULCtype,theseareasexperiencedincreases,decreases,andperiodicepisodesofstabilityinIJIoverthe1970sand1980s.Sowhilethecovertypesmayhavebeenrelativelyslowtochangeovertime,bycomparisonthestructure,specifi-callytheinterspersionofLULCclasses,wasconstantlychangingandchangingindifferentways.Thatis,notonlywastherefirstorderchangebutalsosecondorderchange.TheareasproximatetothesecoreregionsalsoexperiencefairlydramaticstructuralchangeasreflectedbyIJI,althoughslightlymoreconsistentinthesecondorder(roughlytwoperiodsofincreasingIJIwithoneperiodofdecrease).Intermsofpercentagelandscape(PCT),thesimilaritiesinuplandversuslowlandareaswerelesspronounced.Thecoreuplandareas (the leastaccessible) sawearlydecreasesinPCT, followedby increases and subsequentdecreases, similar to the trends inLULCCandLULCpatternchangediscussedearlier.Theregionsproximatetotheuplandcorebutmoreaccessible(lowerelevations)showedtwoperiodsofdecreaseas well, although the increase in PCT came at slightly different times (a notableexceptionwas the southernmost regionmentionedearlier,whichwas structurallystable in thedynamic sense in that it experiencedadecrease inPCT throughallobservations).Lowlandareas,incontrast,nearlyuniformlydisplayedthedecrease-increase-decreasepatternforPCT,despiteobservedstabilityinLULCcomposition,suggestingthateveninthefaceofrelativelystableLULCCthestructuraldynamismoftheareawasin“constantfluctuation.”
6.4 PAneLeD PAtteRn MetRICs: MeAns oR enD?
Thesubjectivityofthepanelapproach,andparticularlythepaneledpatternmetricmethod,presentssignificantchallengesandmerits testinginecosystemswithdif-ferentlevelsofhumanimpactandlandscapeheterogeneity(acrosstimeandspace).Aparticularconcernisthedelineationofpatchboundaries;whentestingexistingpatches(betheyforestrefugia,controlplots,orcadastre-definedparcelsofland),theboundariestouseinanalysisareclear.Butotherwise,boundariesareconstructedfromayearof the imagery,and thedeterminationof thebaseyear impactswhattrends arepossible todetect and inwhatdirection.By testing the sensitivity androbustnessofresultstochangesinthebaseyear,researchershaveinpaneledpatternmetricsapotentialtoolforcreatingecologicallymeaningfulunitsofanalysisthat
42963_C006.indd 112 11/6/07 7:13:58 AM
Landscape Dynamism 113
fitwithinahierarchical(nestedornon-nested)frameworksuitablefordrawingupontheoriesoflandscapeecologyandhierarchicalpatchdynamics.4,8,9,44Withincreas-inggraininthetimeseries(greaternumberofobservationsorimages),thepossiblebiasinbaselinedeterminationshoulddrop.However,traditionalnotionsofaccuracyassessmentcalculationpositthatasthenumberoftemporalobservationsincreases,theamountofaccuracyassessmentdataneededincreasesatanincreasingrate31andthelikelihoodofanacceptablecumulativeproductdecreasessharply.If10imageswith95%accuracywereused,thetraditionalmannerofrepresentingchangeaccuracywouldmeanthat,atbest,the10-imagechangeproductwouldhaveanaccuracyofjustunder60%(0.9510):anaccuracylevelhardlyworthpursuing.Andyetitappearsthatthesemulti-inputproductsofferthemostpromiseforextractingprocess.Perhapsanewframeworkforconsideringhowtothinkaboutchangeandaccuracyisneeded.Fromphysics,Griffiths’s45consistenthistoryapproachholdspromiseasametaphorforaccuracy,wherethegreatertheobservations,thelessimportantanygivenerrorinanygivenobservationis,sincethepatternisindicativeoflargertemporalrelation-shipsatworkandnotaproductofanyonesingulareventormistakenpixelclassifica-tion.Rectifyingthislineofreasoningwithdisturbanceand(dis)equilibriumtheorypresents a challenging but fruitful avenue of epistemological and methodologicalreasoning, especially given continued sensor systems failures requiring analyststobuild timeseries frommultiplesensorplatformsand thuspossibly introducingartifactsintoclassificationswhencomparedacrosstime.30,31
IntegratingnotonlyclassificationsderivedfrommorethanonesensorsystembutotherkindsofproductspresentsanotherfrontierofLULCCwork.Duetofactorsvaryingfromcost toclimaticconditions,establishingsufficient temporalgrainorfrequencychallengesmanyresearchteams;moreover,thetemporalextentofmostLULCC work lacks the historical depth necessary to test, assess, or account forlandscape impactsoccurringover longer timeframesor in themoredistantpast:geomorphologicalchange,speciesevolution,pastcivilizationlanduses,andclimaticchangeresearchallofferevidencethattheforcesatworkonlandscapesthousandstohundredsofthousandsofyearsagostillimpactlandscapesandlandscapecompo-nentstoday(Figure6.7).22LinkingofmappedproductstorecentLULCclassifica-tionshasbeendoneinfrequentlytoextendatimeseries,buttherealchallengeliesinlinkingnon–wall-to-wallornonspatiallyexplicitproductstotoday’sdigitalproducts.Traveljournals,agriculturaltaxationrecords,andartworkofferrareglimpsesintouncharteredtemporalextentsoflandscapesoflongago,22eveninthoselackinginspatialextentandexplicitness.
The panel method generally is, perhaps obviously, of most use in data setswith a rich time series. As digital archives of LULC and LULCC become morewidelyavailable,thisapproachcanbefurthertestedindifferentlyimpactedsocialandenvironmentallandscapes.Panelanalysishasmostcommonlybeenappliedtoinformationextractedfromopticalsensorsystems,butitcouldbeextendedtootherimagerysources.Panelanalysisoffersparticularappealforlandscapeswithtemporalheterogeneity.Paneledpatternmetrics,morespecifically,offerthegreatestpotentialinsightwhenlandscapecoverchangeappearstobeseparablefromchangesinland-scapeconfiguration.Forexample,inareasofshiftingorswidden(anareaclearedfortemporarycultivationbycuttingandburningthevegetation)cultivation,theamount
42963_C006.indd 113 11/6/07 7:13:59 AM
114 Land Use Change: Science, Policy and Management
(b) (c)
(a)
(d)
FIGURe 6.7 (a)Seasonalityimpactsthislandscapeinseveralways.Here,ricestubbleisatypical“winter”ordryseasonlandscapecomponent.Hazeinbackgroundissmokefromtraditionalburningofthericefieldstoboostnutrientsinthesedegradedsoils.(b)Cashcropagriculture tends tooccur in topographic regionsnot suitable forannual riceproduction.Shownaresugarcane(formarket)andeucalyptus(typicallyforlocalbuildingsupplies,fieldborders,orsoilstability).(c)Typicalrecentlyimproved(elevated)roadwithriceagricultureinbackground.“Borrowpits”provideroadelevationmaterialsandareoftensubsequentlyusedassmallwaterimpoundmentsforirrigation.(d)Whilemostwaterimpoundmentsinthe area are small in spatial extent, the impact of the network of such impoundments iscriticalinpopulation-environmentimpacts.Waterimpoundments,whileusedforirrigationandfloodcontrolthroughoutmanyfields,aretypicallyringedbyasmallfringeofwater-demandingcropsforhouseholdconsumption,suchaswatermelonandbananas.
42963_C006.indd 114 11/6/07 7:14:01 AM
Landscape Dynamism 115
oflandsunderanyparticularusemaychangeverylittlebetweenobservations;buttheconfigurationcouldstillbefluctuating,particularlyunderlocalizedmanagementstrategies.LandscapesundergoingevendrasticLULCCmayinfactbestructurallystablewhenwholesalereplacementofclassesoccurs.Landscapesundergoingbothcompositional and configurational change present the most complex situation forlandscapeassessment,anddisentanglingthesetypesoflandscapechangeiscriticalforextractingabetterunderstandingofprocessandfunctionfrompattern.Tempo-rally,landscapeswithheterogeneouschange(differentratesofchangeoverdifferentperiods) would benefit more from the general pattern approach than those withconsistenttemporalprocesses,butproperextractionofthesepatternspresumesanadequatelyrichtimeseries.
The panel method generally, as used here and elsewhere,34 offers LULCCresearchersonewayofslidingcloser toprocesson thepattern-processspectrum.Assessing dynamism (continuous change) from dynamics (changes and driversassessed via snapshots in time) remains a critical area of concentration for theLULCC community. The paneled pattern metric approach provides a means forexploring stronger linkages to process and function from patterns of LULC andLULCC.Moreover,paneledpatternmetricsconstituteanexplicittestofthevalueof landscapeecologytoLULCCworkbyexploring therelativecontributionsandinteractionsoflandscapecompositionandconfiguration.Theimplementationofthistool is,currently,pronetosubjectivity.Althoughpatternmetricshavebeenfoundtorevealcriticaldifferencesinlandscapes,rarelyaretheyexplicitlyandquantita-tively linked to human or biophysical processes. As such, determining appropri-ate thresholds forcategorizationofconstrainedorunconstrainedmetrics remainscriticaltodobutdifficulttojustify.Aswiththeintroductionofvegetationindicesseveraldecadesago,astatisticalcorrelationmayconvincesomeoftheutilityofanapproach,butultimatelyitistheempiricalandquantitativetietoprocessthatcon-vincespractitionersoftheapproach’sworth.ThepursuitofthatlinkageisaripeareaforresearchofecologicallyandbiophysicallygroundedLULCCteams.Itmaybethattheprocesslinkagebetweenpaneledpatternmetricsandlandscapeprocessesisneverdiscoveredorverified,oreventhatitisrejectedanddisproven.Butuntilsucha time,requiringaprocess linkagemaybepremature,whenthegreatestpromiseofpaneledpatternmetricsmaylieinapplicationofadataminingapproachtofirstuncovercriticalthresholdsorflippointsofLULCC,andthenbringtobeartheoriesandmethodsforfleshingouttheprocessesatwork.
ACKnoWLeDGMents
Iamgratefultothefollowingforsupportofthiswork:CarolinaPopulationCenterandDepartmentofGeography,UniversityofNorthCarolina;SigmaXiScientificResearchSociety;andMahidolUniversity,Thailand.
ReFeRenCes
1. Walsh,S.J.,andCrews-Meyer,K.A.Linking People, Place, and Policy: A GIScience Approach.KluwerAcademicPublishers,Boston,2002.
42963_C006.indd 115 11/6/07 7:14:02 AM
116 Land Use Change: Science, Policy and Management
2. Meyer, W. B., and Turner, B. L., II, eds. Changes in Land Use and Land Cover: A Global Perspective.CambridgeUniversityPress,Cambridge,1994.
3. Liverman,D.etal.People and Pixels: Linking Remote Sensing and Social Science.NationalAcademyPress,Washington,D.C.,1998.
4. Turner,M.G.,Gardner,R.H.,andO’Neill,R.V.Landscape Ecology in Theory and Practice: Pattern and Process.Springer,NewYork,2001.
5. NationalResearchCouncil.Population, Land Use, and Environment: Research Directions.PanelonNewResearchonPopulationandtheEnvironment.Entwisle,B.,andStern,P.C.,eds.,CommitteeontheHumanDimensionsofGlobalChange,DivisionoftheBehavioralandSocialSciencesandEducation.NationalAcademiesPress,Washington,D.C.,2005.
6. Frohn, R. C. Remote Sensing for Landscape Ecology: New Metric Indicators for Monitoring, Modeling, and Assessment of Ecosystems.LewisPublishers,BocaRaton,Fla.,1998.
7. Rindfuss,R.R.etal.Developingascienceoflandchange:Challengesandmethodolog-ical issues. Proceedings of the National Academy of Science 101(39), 13976–13981,2004.
8. Forman,R.T.T.Land Mosaics: The Ecology of Landscapes and Regions.CambridgeUniversityPress,Cambridge,1995.
9. Wu, J., and Loucks, O. L. From balance of nature to hierarchical patch dynamics:Aparadigmshiftinecology.Quarterly Review of Biology 70(4),439–466,2005.
10. Jensen, J. R. Remote Sensing of the Environment: An Earth Resource Perspective.PrenticeHall,UpperSaddleRiver,N.J.,2000.
11. Crews-Meyer,K.A.Characterizinglandscapedynamismviapaneled-patternmetrics.Photogrammetric Engineering and Remote Sensing 68,1031–1040,2002.
12. Ahl,V.,andAllen,T.F.H.Hierarchy Theory: A Vision, Vocabulary, and Epistemology.ColumbiaUniversityPress,NewYork,1996.
13. Gustafson, E. J. Quantifying landscape spatial pattern: What is state of the art?Ecosystems1,143–156,1998.
14. O’Neill,R.V.etal.Landscapepatternmetricsandecologicalhealth.Ecosystem Health5,225–233,1999.
15. Skole, D., and Tucker, C. Tropical deforestation and habitat fragmentation in theAmazon:Satellitedatafrom1978to1988.Science260,1905–1910,1993.
16. Plummer, S. E. Perspectives on combining ecological process models and remotelysenseddata.Ecological Modelling 129,169–186,2000.
17. Oetter,D.R.etal.LandcovermappinginanagriculturalsettingusingmultiseasonalThematicMapperdata.Remote Sensing of Environment76,139–155,2001.
18. Walsh,S.J.etal.Amulti-scaleanalysisofLULCandNDVIvariationinNangRongDistrict,NortheastThailand.Agriculture, Ecosystems, and Environment 85, 47–64,2001.
19. Walsh, S. J. et al. Patterns of change in land use, land cover, and plant biomass:separating inter- and intra-annual signals in monsoon-driven northeast Thailand. In:Millington,A.,Walsh,S.J.,andOsburn,P.,eds.,GIS and Remote Sensing Applications in Biogeography and Ecology.KluwerAcademicPublishers,TheNetherlands,2001.
20. Norman,A.L.IsolatingSeasonalVariationinLanduse/LandcoverChangeUsingMulti-temporal Classification of Landsat ETM Data in the Peruvian Amazon. MAthesis,UniversityofTexas,Austin,Tex.,2005.
21. Hall,F.G.,Strebel,D.E.,andSellers,P.J.Linkingknowledgeamongspatialandtem-poralscales:Vegetation,atmosphere,climateandremotesensing.Landscape Ecology2(1),3–22,1988.
22. Butzer, K., and Helgren, D. Livestock, land cover and environmental history: TheTablelandsofNewSouthWales,Australia,1820–1920.Annals of the Association of American Geographers95,80–111,2005.
42963_C006.indd 116 11/6/07 7:14:02 AM
Landscape Dynamism 117
23. Ramos, F. A multi-level approach for 3D modelling in geographical informationsystems.International Archives of Photogrammetry and Remote Sensing34(4),43–47,2002.
24. Axinn,W.G.,Barber, J.S.,andGhimire,D.J.Theneighborhoodhistorycalendar:Adata collection method designed for dynamic multilevel modeling. Sociological Methodology27,355–392,1997.
25. Allen,T.F.H.Thelandscape“level”isdead:Persuadingthefamilytotakeitofftherespirator.In:Peterson,D.L.,andParker,V.T.,eds.,Ecological Scale: Theory and Applications.ColumbiaUniversityPress,NewYork,1998.
26. Millington,A.C.,Walsh,S.J.,andOsborne,P.E.GIS and Remote Sensing Applica-tions in Biogeography and Ecology.KluwerAcademicPublishers,Boston,2001.
27. Crews-Meyer,K.A.AgriculturallandscapechangeandstabilityinnortheastThailand:Historicalpatch-levelanalysis.Agriculture, Ecosystems and Environment101,155–169,2004.
28. Brown,D.G.,Aspinall,R.J.,andBennett,D.A.Landscapemodelsandexplanationin landscape ecology—A space for generative landscape science? Professional Geographer,Focussection58(4),2006.
29. McGarigal,K.,andMarks,B.J.FRAGSTATS: Spatial Pattern Analysis Program for Quantifying Landscape Structure.ForestScienceDepartment,OregonStateUniversity,Corvallis,Oregon,1993.
30. Jensen,J.R.Introductory Digital Image Processing: A Remote Sensing Perspective,3rded.PrenticeHall,UpperSaddleRiver,N.J.,2005.
31. Congalton,R.G., andGreen,K.Assessing the Accuracy of Remotely Sensed Data: Principles and Practices.LewisPublications,BocaRaton,Fla.,1999.
32. Crews-Meyer,K.A.Temporalextensionsof landscapeecology theoryandpractice:LULCCexamplesfromthePeruvianAmazon.Professional Geographer,Focussection58(4),421–435,2006.
33. Crews-Meyer,K.A.IntegratedLandscapeCharacterizationviaLandscapeEcologyandGIScience:APolicyEcologyofNortheastThailand.Doctoraldissertation,UniversityofNorthCarolina,ChapelHill,2000.
34. Mertens,B.,andLambin,E.F.Land-cover-changetrajectoriesinSouthernCameroon.Annals of the Association of American Geographers90,467–494,2000.
35. Crews-Meyer,K.A.Assessing landscape change andpopulation-environment inter-actionsviapanelanalysis.Geocarto International16,69–80,2001.
36. Fukui, H. Food and Population in a Northeast Thai Village. University of HawaiiPress,Honolulu,1993.
37. Donner,W.The Five Faces of Thailand: An Economic Geography.St.Martin’sPress,NewYork,1978.
38. Arbhabirama,A.etal.Thailand Natural Resources Profile.OxfordUniversityPress,Singapore,1988.
39. Craig, I. A., and Pisone, U. Overview of rainfed agriculture in northeast Thailand.In:Proceedings of a Workshop on Soil, Water and Crop Management Systems for Rainfed Agriculture in Northeast Thailand. Khon Khaen University, Khon Khaen,Thailand,1985.
40. Crews-Meyer,K.A.Personalcommunicationwithvillagersduringfieldwork,JanuarytoFebruary,unpublished,1999.
41. Walsh,S.J.etal.Scale-dependentrelationshipsbetweenpopulationandenvironmentin Northeastern Thailand. Photogrammetric Engineering and Remote Sensing 65,97–105,1999.
42. Fox, J. et al. People and the Environment: Approaches for Linking Household and Community Surveys to Remote Sensing and GIS.KluwerAcademicPublishers,Boston,2003.
42963_C006.indd 117 11/6/07 7:14:03 AM
118 Land Use Change: Science, Policy and Management
43. Holling,C.S.Resilienceandstabilityofecologicalsystems.Annual Review of Ecology and Systematics4,1–23,1973.
44. Peterson,D.L.,andParker,V.T.Ecological Scale: Theory and Applications.ColumbiaUniversityPress,NewYork,1998.
45. Griffiths, R. B. Consistent histories and the interpretation of quantum mechanics. Journal of Statistical Physics36,219–272,1984.
42963_C006.indd 118 11/6/07 7:14:03 AM
119
7 Developing a Thick Understanding of Forest Fragmentation in Landscapes of Colonization in the Amazon BasinAndrew C. Millington and Andrew V. Bradley
Contents
7.1 Introduction................................................................................................. 1197.2 CaseStudyandMethods............................................................................. 122
7.2.1 Chapare............................................................................................ 1227.2.2 Methods........................................................................................... 123
7.3 AConceptualModelofFragmentation......................................................1247.4 BehaviorofLandscapeMetrics.................................................................. 1287.5 Discussion................................................................................................... 133Acknowledgments.................................................................................................. 135References.............................................................................................................. 135
7.1 IntRoDUCtIon
Landscapefragmentationisakeyconcernoflandscapeecologistsandconservationbiologists.Landscapesprovidethehabitatsanddeterminetheresourcesnecessaryforplantandanimalspeciestosurvive.Aslandscapesfragment,theproportionsofdifferentelementsinanylandscapechange,asdotheirspatialproperties.1Effortsto understand spatial patterns of, and relationships between, these elements havebeenakeyobjectiveoflandscapeecology,whileconservationbiologistshavetriedto relate the responsesof species to these spatialpatterns,2,3,4 sometimes throughdirectmanipulationoflandscapes(cf. reviewofsuchexperimentsbyDebinskiandHolt5)butmoreoftenbyobservation.Manycommentatorsonenvironmentalissuesin the humid tropics cite landscape fragmentation, along with the more general“forestloss,”asmajordeterminantsofbiodiversitylossandecologicaldeterioration.
42963_C007.indd 119 10/29/07 10:48:54 AM
120 Land Use Change
Researchtounderstandecologicalresponsestoforestlossovertimeisfragmentaryin itself, but research carried out under the auspices of the Biological DynamicsofForestFragments Project in northern Amazonia6,7,8 has provided a plethora ofnotableexceptions,theresultsofwhichwererecentlyreviewedbyLaurenceetal.9
Researchundertakenbybiologistson theecologicalandconservationaspectsof fragmentationof tropical forests isvoluminouswhencompared to thenumberofresearchpapersthatlinksocioeconomicprocessestothespatialphenomenonofforestfragmentationinthehumidtropics.Therelativepaucityofresearchonhowparticularpatternsoffragmentationaregeneratedshouldbeofconcernbecauseonlywhenwe fullyunderstandhowandwhy fragmentationoccursatawide rangeofgeographicalscaleswillwebeable toplanfor,manage,andaccrueconservationbenefits.Researchundertakenonthistopicgenerallyfocusesonthespatialpatternsthatresultfrom,oraretheend pointsof,ageneralizedprocess(orsetofprocesses)oftropicalforestconversion.10,11,12,13Thesespatialpatternshavebeentermedclearance typologiesbyHussonetal.10orclearance morphologiesbyLambin.11Examplesofthelinkagesbetweenspatialtypologiesormorphologiesandgeneralizedeconomicactivities are: planned settlement creates the so-called fishbone pattern of forestloss;spontaneoussettlercolonizationalongroadnetworkscreateslinearcorridorsofclearance;large-scalecommercialranchingandothertypesofcommercialagri-culturecreate largeblocksofpasture, cultivation, and forest; subsistenceagricul-turecreatesadiffusemosaicofsmallclearings;veryhighruralpopulationdensitiesleaveanagriculturallandscapewithforestislands;andislandsofforestsurroundingurbanareasoccurwhenperi-urbanplantationspredominate.Itisthefirstofthesetypologies—plannedsettlementleadingtofishbonepatternsoflandclearance—thatweaddressindetailinthischapterbydrawingonevidencefromChapare,Bolivia.
There are limitations to these generalized process-pattern relationships.ImbernonandBranthomme13notedvariations in theprocess-pattern relationshipswithin relatively small studyareasacross the tropics,possibly indicating that thespatialscaleatwhichmuchofthisresearchhasbeencarriedoutonlyallowsverygeneralizedobservationsofthelinkagesbetweendriversoflandusechangeandtheresultingpatternsoffragmentationtobemade.Lambin11andHargisetal.14 describehow spatial patterns can morph from one to another over time as fragmentationevolves,indicatingthatsometemporaldependencymightexistandthatpatternsmaynot,inthemselves,beend points.Moregermanetoourworkisthattheseprocess-patterngeneralizationstypifya“thin”understandingofhowparticularpatternsofforestfragmentationarecreatedbythedrivers(agents)oflandcoverchange,despitethefactthattheactionsofsuchagentsinthehumidtropicsarewellunderstoodfromthesynthesesthathavebeenundertaken.15,16,17
Acase inpoint is the relatively“thin”understandingbetween roadconstruc-tionandforestfragmentationthathasdevelopedfortheAmazonBasin.18Figure7.1illustrateswhat ismeantbya“thin”understandingandindicateswherea“thick”understandingneedstobedevelopedtomeet theneedsofconservationplanning.We acknowledge that research in a limited number of colonization zones in theAmazonBasinhasmodeledlandcolonization.19,20,21,22Thefociofthesestudieshasgenerallybeenonsocietalprocessesandimpactsandontheresultingforestcoverinageneralsense.Althoughthishasdeepenedourunderstandingoflandusedynamics
42963_C007.indd 120 10/29/07 10:48:55 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 121
incolonizationzones,knowledgeofthecausallinkagesbetweenprocessesthatleadindividuallandownerstofragmenttheforestsontheirpropertiesinparticularways,andhowthefragmentationpatternsonindividualpropertiesmeshtogetheracrossa community or a number of communities, has rarely been investigated, thoughits importance is recognized.23,24Anotableexception is the researchbyPerzandWalker25whoappliedaneo-Chayanoviananalysistosecondaryforestregrowthonsmallcolonistfarms,arguingthatmoreattentionneedstobepaidtohouseholdsasthemostproximatecontextinlandusedecisionmaking.
Forest
City A
City A
City B
City B
Conservationunit
Major road
Flow of species betweenconservation units
1
2
3
FIgURe 7.1 ThickandthinunderstandingsofforestfragmentationalongroadsinlowlandforestsoftheAmazonBasin.Theprogressionfromstage1to2showsaforestblockdissectedbyaroadconnectingcitiesAandB.Stage3illustrateslarge-scalefragmentationbetweentwoconservationunitsandtheconnectivitybetweenthemthatisrequiredisillustratedbytheblackarrow.A“thin”understandingoffragmentationisrepresentedinStages2and3;thedevelopmentof the“thick”understandingthatwearguefor in thischapter isrequiredforthewhiteareaalongthemainroad.
42963_C007.indd 121 10/29/07 10:49:01 AM
122 Land Use Change
Ifplanningandmanagementinterventionsaretobemadeduringforestconver-sioninareasundergoingcolonization,thenadetailedunderstandingofhowagentsareoperatinginthelandscapeatdifferentgeographicalscalesisessential.Forexample,theinfluenceofroadsandotherlinesofaccessoccuratonescale.Generalizationsabouttheenvironmentalimpactsofroadsatthisscalehavebeenrecognized26,27,28andused,somewhatcontentiously,tomodeltheimpactofdevelopmentpoliciesinBrazil.29,30,31,32,33Butnestedbelowtheroadnetworkingeographicalspaceisalmostalways a cadastre or land property grid. Although the roads can be constructedbothbefore andafter a cadastrehasbeen surveyed,weargue that the roadsand the cadastre provide two spatial imprints connected through a scale hierarchy inforestedlandscapesthataredestinedtofragment.Moreover,attemptstomodelfrag-mentationspatiallybasedonroadbuildinghavemissedafundamentalpoint.Thatis,itisthecolonisthouseholdswithinthelimitsoftheirpropertiesthatcreatethepatternsofforestfragmentationbyrespondingtoeconomicandpolicysignalswithmachetes and chain saws, rather than planners and road builders with maps andbulldozers.Itcouldbearguedthattheplannersandroadbuildersspatiallyconstrainwhat farmers canclear, aswell asprovide thewherewithal to extract timber andproduce fromtheir farms.Weacknowledge that theargument thatwemakeheremayonlyapplytofarmersinplannedcolonizationschemes:itmaynotapplytoothertypesofhumidtropicalforestcolonizationintheAmazonBasinorelsewhere.
Developinga“thick”understandingoffragmentationatcontemporarydeforesta-tionfrontsthereforerequiresintegratingtheactionsoflandmanagersonindividualpropertiesovertimeandmeshingthemtogetherwithintheroadnetworksandlandpropertygrids.Inanappliedvein,whatisrequiredspecificallytoplanandmanagelandscapesofcolonizationisresearchintothespatialandtemporaldynamicsofforestfragmentation,whichcovermultiplescales,consideringallagentsofchange,andthelinksbetweenagentsandscales.Theresultsofsuchresearchwillallowanimportantquestiontobeanswered.Thatis,howdothecollectiveactionsoflandmanagersinaparticulararealeadtoparticularpatternsofforestfragmentationovertrajectoriesoftime?Ifthisquestioncanbeansweredrobustly,thentwofurtherquestionsofconcerntolandscapeecologistsandconservationplannerscanalsobetackled:
1.Canzonesofcolonizationbeplannedsothattheycandevelopintomulti-purpose landscapes that allow rural production systems to co-exist withbiologicalconservation?
2.Howcanexisting,partiallyfragmentedlandscapesbeplannedfor?
Inthischapterweexplainhowwehaveattemptedtodevelopa“thick”under-standingofthedynamicsoflandscapefragmentationinacolonizationzoneinthelowlandhumidtropicsofBolivia,andthenreflectonfurtherresearchneeds.
7.2 CAse stUDY AnD MetHoDs
7.2.1 Chapare
WeusedobservationsfromtheChapareregionofBoliviainthisresearch.ChapareisacolonizationzoneinthehumidtropicallowlandsofBoliviadatingbacktothe
42963_C007.indd 122 10/29/07 10:49:02 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 123
1930s, though most colonization and forest conversion has taken place since the1960s.34,35,36,37,38Theareaisboundedtothenorthbyrelativelyundisturbedlowlandtropicalforestsandtothesouthbymontaneforests.Theseforestsarelikelytoremainrelativelyundisturbedintheforeseeablefuturebecausetothenorththeyareeitherpermanentlyorseasonallyinundated,andtothesouththeyareprotectedbyParqueNacional(PN)Carrasco.Thezoneofcolonizationcreatesawedgeoflivelihoodsanddisturbancebetweenthesetwoforestblocks,therebycompromisingtheexchangeofanimalsand,lessobviously,plantmaterialbetweenthetwo.Giventhestrongaffini-tiesbetweentheanimalsandplantsinthelowlandmontaneforestsinPNCarrascoandthelowlandforests,thisisacauseofconcernforconservationists.
Chaparebenefits fromadensenetworkofprimaryandsecondary roadsaug-mentedbyfoottracks.34,37,39Thisnetworkhasdevelopedprogressivelysincethe1960sintwoways.First,byitsphysicalextension;second,byupgradingtheroadsurfacesfromdirttotarmacorcobble.Alandpropertygridhasdevelopedinparallelwiththeroadnetwork,andthetwoareintegraltocolonizationofthearea.ThelandnowoccupiedbyeachofthecommunitiesinChaparewassurveyedandmarkedoutdownto the limitsofeach landparcelby theInstitutoNacionaldeColonización(INC)before itwassettled.Titlesweregiven tocolonistsmoving intoeachcommunity.TheserecordsareheldbyINC.Thetransportationnetworkandthelandpropertygridcombinetoprovidethespatialstageonwhichcolonistsactouttheirlivelihoods,whilesimultaneouslyspatiallyconstrainingtheiractivities.
7.2.2 Methods
Tounderstand thespatialand temporal relationshipsbetween thedifferentagentsof change, one of us (Bradley) conducted detailed surveys in three communitiesbetween2000and2003.34SalientdetailsofeachcommunityarelistedinTable7.1.In each community the following research was undertaken to develop an under-standingofthespatialandtemporaldynamicsoflandcoverchange:
1.ThelandpropertygridforeachcommunitywasobtainedfromINCandtheownersofeachpropertyidentified.
2.Permissionwasobtainedfromthecommunitysindicatotointerviewlandowners/managers.Subsequentlythoseinterviewedwereselectedrandomly.
tAble 7.1salient Information Concerning the three Communities studied
Community Area (ha)Altitude (m.a.s.l.)
number of properties
Year of first settlement
Prevailing economic activities 2000–2003
Arequipa 1,220 250 60 1983 Banana,blackpepper,cassava,heartofpalm,andricecultivation
Bogotá 3,196 250–350 90 1972 Cattlerearing:beefanddairy
Caracas 1,745 220 110 1963 Banana,mandarin,andorangecultivation
42963_C007.indd 123 10/29/07 10:49:02 AM
124 Land Use Change
3.LandcovermapsofeachcommunitywerecreatedusingLandsatMSS,TM,andETM+imageryacquiredinthedryseasonsof1975,1976,1986,1992,1993,1996,and2000usingclassificationalgorithmsinERDASImagine(fulldetailsofwhichcanbefoundinBradley34).Fieldverificationwascar-riedoutonthemapsderivedfromimageryacquiredin2000(Figure7.2).Thesemapswerethensimplifiedintobinaryforestandnonforestcovers.
4.Eachlandowner/managerselectedwasshownthetimesequenceofforest/nonforestmapsfortheirpropertyandaskedtorecallaspectsofforestclear-anceandwhatcropshadbeengrownatthetimestheimageswereacquired.Thiswasdoneusingparticipatoryruralappraisalmethods,themostinfor-mativeofwhichwastowalkeachfarmer’spropertywithhim.Thisenabledfarmerstoverifytheirrecallofwhathadbeengrownatparticulartimes,andalsoenabledgeolocationoftheseobservationsusingaglobalposition-ingsystem(GPS)receiverinnondifferentialmode.
5.For each property surveyed, a forest/nonforest map—a property forest/nonforest map—wasannotatedwiththeowner/manager’sobservations.
Propertiesaretypically20hainareasofcultivationand50hainareasoflive-stockrearing.Fifteen,13,and17propertiesweresurveyedindetailforArequipa,Bogotá,andCaracas, respectively(Table7.1).Thenamesof thecommunitiesandthefarmersweinterviewedhavebeenmadeanonymousinaccordancewithnormalsocial science survey practices, and because some of the farmers have illegallygrowncocainthepast.Theobservationsmadeaboutfarmsandfarmer’sresponsesto questions were used in two ways. First, to understand the drivers of land usechangeinChaparefromthe1970stothepresenttime,34andsecond,toverifytheforest/nonforestmapsofeachcommunity—community forest/nonforest maps—thattheproperty forest/nonforest mapsofeachpropertysurveyedwereextractedfrom.Theverifiedcommunity forest/nonforest mapswerethenusedtomapareasofforestandnonforestforeachcommunity.
7.3 A ConCePtUAl MoDel oF FRAgMentAtIon
Bycomparingtheprogressivedevelopmentofspatialpatternsofforestfragmenta-tionbetweenthethreecommunitieswedevelopedaconceptualmodelofforestfrag-mentationthathassixphases(Figure7.3).Thefirstorplanningphaseoccursbeforecolonization,and,consequently,noforesthasbeenclearedatthistime(Figure7.3a).However, thisstage is importantfor it isat this timethat thegeneralspatialcon-figuration of forest and agriculture patches that will ultimately populate thisgeographicalspaceisdetermined.Thisisbecausethepropertygridissurveyedandlaidout,andthe(unimproved)accessroadsandtracksareconstructed.Thisphaseexistsforashortperiodoftimebeforecolonistsarrive.Inthesecondphase—earlycolonization—olonists clear forest at the primary ends of each property, therebycreatingasimplepatternoffragmentationoneithersideoftheprimaryaccessroad(Figure7.3b). In the three communities we researched intensively, all propertieswereoccupiedalmost immediately,and theareasofeachplotcleared(calculatedfromcommunity forest/nonforest maps)were similar because the colonists eitherarrivedtogetherorwithinafewmonthsofeachother.Moreover,theirmotivations
42963_C007.indd 124 10/29/07 10:49:03 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 125
TM B
and
4 im
age:
11th
Apr
il 19
86TM
Ban
d 4:
18th
July
199
3ET
M B
and
4: 1
4th Ju
ly 2
000
FIg
UR
e 7.
2 S
eque
nce
of im
ages
sho
win
gpr
ogre
ssiv
ede
fore
stat
ion
that
is s
pati
ally
con
stra
ined
wit
hin
ane
twor
kof
pri
mar
yan
dse
cond
ary
feed
erro
ads
in b
etw
een
whi
ch a
rere
ctan
gula
rlan
dhol
ding
s. T
heim
ager
yis
fro
me
aste
rn C
hapa
reb
etw
een
1986
an
d20
00 a
ndis
10
by 1
0km
in a
rea;
all
imag
esa
re a
tmos
pher
ical
lya
ndg
eom
etri
call
yco
rrec
ted
TM
/ET
M+
Ban
d4.
(a)
Show
sth
ele
astf
ores
t cle
aran
cea
nd w
as a
cqui
red
on 1
1A
pril
198
6. T
he d
ark
gray
tone
sth
atd
omin
ate
the
imag
ear
edi
ffer
entt
ypes
of
low
land
trop
ical
fore
st.T
he m
ain
Coc
haba
mba
-San
taC
ruz
bise
cts
the
imag
ean
dfo
rest
is c
lear
eda
bout
500
mo
nei
ther
sid
eof
th
ero
ad. A
net
wor
kof
pri
mar
yfe
eder
road
sca
nju
st b
ese
enin
the
fore
sted
are
as o
nei
ther
sid
eof
the
mai
nro
ad, a
ltho
ugh
ther
eis
ver
yli
ttle
cle
aran
ce in
any
of t
he c
omm
unit
ies
thes
ero
ads
serv
e. T
heb
lack
sin
uous
pix
els
are
smal
lriv
ers
that
bis
ectC
hapa
re
and
eman
ate
from
the
And
es f
ooth
ills
to th
eso
uth
of th
eim
ages
.(b)
Im
age
was
acq
uire
don
18
July
199
3. T
heli
ght g
ray
area
sal
ong
the
prim
ary
feed
er r
oads
are
are
as o
fcl
eara
nce.
(c)
Imag
ew
as a
cqui
red
on 1
4Ju
ly2
000
and
show
sfu
rthe
rde
fore
stat
ion.
D
iffe
rent
siz
efo
rest
pat
ches
tha
t hav
ebe
enis
olat
ed b
yde
fore
stat
ion
fron
tsc
oale
scin
gfr
omd
iffe
rent
dir
ecti
ons
can
be s
een
in
the
cent
ero
fth
eim
age.
The
cas
tell
ated
nat
ure
ofd
efor
esta
tion
inth
isty
peo
fco
loni
zati
ons
chem
eca
nbe
cle
arly
see
n.
42963_C007.indd 125 10/29/07 10:49:06 AM
126 Land Use Change
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
1 2 4 5 6 7 8 9 10 11 12 13 14 15 16 17 3 1 2 4 5 6 7 8 9 10 11 12 13 14 15 16 17 3
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
BA
C
C
1 2 4 5 6 7 8 9 10 11 12 13 14 15 16 17 3
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
BA
C
1 2 4 5 6 7 8 9 10 11 12 13 14 15 16 17 3
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
C
BA
1 2 4 5 6 7 8 9 1011121314151617 3 1 2 4 5 6 7 8 9 1011121314151617 3
Stream
Property grid
Primaryaccessroad
Secondaryaccessroad(s)
(a)
(b) (e)
(f )(c)
(d)
FIgURe 7.3 Six-phaseconceptualmodelof forest fragmentationbasedonacommunitywith34plots(numbered1to34)ofequalarea.Aprimaryaccessroad(black)runsthroughthecenterofthecommunityandastreamcutsthroughproperties1to7.Imagesrepresent:(a) the planning phase, (b) the early colonization phase, (c) the illicit coca phase, (d) theimprovedaccessphasewithsecondaryroads(grayroads),(e)thecomplexclearancephase,and(f)theplotexhaustionphase.Thelightgraycellsinphases(e)and(f)representsecondaryregrowthforest.
42963_C007.indd 126 10/29/07 10:49:08 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 127
forclearanceweresimilar,thatis,toclearforestforlandtoplantsubsistencecrops(followedbycashcropsinsubsequentyears)andtoacquireconstructionmaterialsforhouses.Inaless-detailedexaminationoftheimageseriesforallofChapare,wesawthisphasereplicatedinallcommunities.
Morecomplexspatialpatternsof forest fragmentationestablish themselves inthethirdphaseofthemodel—theillicitcocacultivationphase(Figure7.3c).BecausecocacultivationinthelowlandsofBoliviahasalwaysbeenillegal(incomparisontocultivationforchewinginthesubtropicalmontaneforestswhereitislegal),farmersgenerallyadoptedstrategiestocultivatecocathatfragmentedforestsinparticularways. However, in the 1970s when coca cultivation and cocaine production wasbarelycontrolledbythegovernment,cocawasgrownopenlyat theprimaryendsofmanyproperties.Asgovernmentcrackdownsoncocagrowingtookeffect,manyfarmersgrewlegalcashandsubsistencecropsattheprimaryendsoftheirplotsandretreatedintotheremainingforestontheirpropertiestoclearsmallareastogrowcoca.Thiswasthemaincauseofforestperforation,andtheextentofperforationdepictedinthismodel(Figure7.3c)ishighbecauseofillegalcocacultivationandisprobablygreater than itwouldbe inothercolonizationareas.Thisassumptionhasyet tobe tested.Thecolonist footprintmodeldevelopedbyBrondizioetal.20predictshighratesofforestclearanceatthisstageasfarmerspreparelandtoplantperennialcashcrops.Butourevidenceindicatesthatalthoughafewfarmersclearedforestatmuchfasterratesthanothers(e.g.,property28,Figure7.3c),thiswasexcep-tionalbecausethevastmajorityoffarmersonlyhadtoclearsmallareastocultivatetheperennialcropofchoice—coca—which,becauseithasahigh-sellingreturn,isconservativeinitslandrequirements.Differencesinforestclearanceratesbetweenindividualpropertiesoccuratthisstagebecausefewfarmerscultivatedland-hungryperennial crops at this time, as predicted by Brondizio et al.,20 rather than coca.Thesedifferencesleadtothecastellatedpatternoftheforest/nonforestboundariesthatcharacterize“fishbone”deforestation.
Thedevelopmentofsecondary,unimprovedfeederroadsatsomepointintimeduringcolonizationistypicalofmostcommunitiesinChapare.Roadsandtracksareconstructedalongtheboundariesofcommunitiestoconnectwiththeroadsthatwereconstructedinitially.Wehavecharacterizedthisphaseasimprovedaccess,andinFigure7.2dsecondaryfeederroadsaredrawnalongthesecondaryendofproperties1to11and18to27.Theestablishmentofsuchroadsandtracksallowsplotstobecultivatedfrombothends,buttheactualreasonsfortheirconstructionareunclearatpresent.Ourinterviewssofarsuggesttheymaybeconstructedtoconsolidatecom-munityboundaries,buttheymaysimplyimproveaccess.Whateverthereason,theycanbeusedtosplituppropertiestosatisfyactualandpotentialdisputesoverinheri-tance,orallowfarmers tocultivatemorefertilesoilsatoneendof theirpropertywhileallowingrecoveryofvegetationandsoilattheotherendoftheplot.
Inthefifthphaseofthemodel—complexclearance—asignificantamountofthelandinacommunityisundersometypeofcultivation(Figure7.3e).Thetermcomplexarisesbecausemanylandscapeecologymetricsattaintheirhighestvaluesduringthisphase.Theformationofbothforestpatchesandtheextensionoftheforestperimeterare due to differential rates of clearance between farmers, and the continuation offorestclearancefromboththeprimaryandsecondaryendsofsomeproperties.The
42963_C007.indd 127 10/29/07 10:49:08 AM
128 Land Use Change
formationofaforestpatchduetothedifferencesinratesofforestclearanceisshowninFigures7.3dand7.3e.Thefarmersinproperties21and23haveclearedforestatfasterratesthanthefarmerinproperty22.Asaconsequenceapatchofforest(B)onproperty22,whichonceshieldedacocaclearing(A),isnowsurroundedbyagricul-turalland.ThismethodofpatchformationiscommonplaceinChapare,andoccursbecause of the intersection of government coca eradication policies (which causesperforation-styleclearanceof forestdeep inproperties),differences incropchoicesbetweenfarmers(whichleadstodifferentlandrequirementstogrowparticularcrops),and differences in household circumstances and aspirations. The creation of forestpatchConproperty10isavariantonthewayinwhichforestpatchBwasformed.Inthiscasetheratesofforestclearancebetweenproperties9,10,and11arenotonlydif-ferentintheamountsofforestclearedannually,butalsothedirectionsofclearancearedifferentbecauseoftheinfluenceofthesecondaryaccessroadonproperty9.
Wehavetermedthefinalphaseplotexhaustion(Figure7.3f).Bythiswemeanthatmostoftheforesthasbeencleared.Someisolatedpatchesremain,andtherearealsopatchesofsecondaryregrowthforestandforestsinareasthataredifficulttoclearorarelocatedonlandthatcannotbecultivated(e.g.,theriparianforestinproperties1to7).
We have evidence that properties are already changing hands by the timethepenultimateandfinal stagesof themodelare reached.Althoughwehavenotrecordedlandbeingsoldinthe45propertieswehavesurveyedindetail,wehavecomeacrossthisonfarmswehavevisitedinothercommunities.Somepropertiesare being sold to new owners and some wealthy farmers purchase adjacent plotstoincreasetheircontiguouslandholdings.Wehaveindicatedthisinthemodelbycombiningproperties28and29inFigure7.3f.
7.4 beHAVIoR oF lAnDsCAPe MetRICsThe conceptual model outlined above is based on detailed observations made inthreecommunities,andtoevaluateitsutilityforanalyzingtheecologicalimplica-tionsoffragmentationweusedmetricscommonlyemployedbylandscapeecologistsand conservation biologists. We calculated proportional forest cover, the numberofforestpatches,andtheforest/nonforestedgelengthforeachphaseinthemodel(Table7.2).ThesedataarevisualizedinFigure7.4.
Lambin11postulatedthatlandscapemetricsusedtocharacterizefragmentationwouldfollowaparticulartrajectoryastropicalforestlandscapeschangedfromthosethatwere entirely forested, through landscapesof agricultural patches in a forestmatrix,toentirelyagriculturallandscapes(i.e.,afewforestpatchesinanagricul-turalmatrix).Hedidnotquantify thispostulatedbehavior,buthypothesized thatthemetricswouldattainpeakvaluesintheheterogeneous,intermediatelandscapesand would be low for homogenous forest or agricultural landscapes. Trani andGiles40 simulated deforestation of a hypothetical forest and calculated metrics atvariouspoints alongadeforestation/fragmentation trajectory.Threemetrics fromtheir analysis—mean forest patch size, the forest/nonforest edge length, and themeannearestneighbordistancebetweenforestpatches—areshowninFigure7.5.We calculated the same landscape ecology metrics as Trani and Giles40 for eachcommunitywestudiedusingFragstats.41Asthemetricsfollowedsimilartrendsineachcommunity,weonlyillustratethemetricsforCommunidadArequipainthis
42963_C007.indd 128 10/29/07 10:49:09 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 129
tAble 7.2selected landscape Metrics Calculated for the six Phases of the Conceptual Model
Phase in conceptual model
Proportional forest loss
(%)
number of forest
patches
number of cultivation
patches
Mean patch size (nominal
units)
Forest edge length
(nominal units)
Planning 0 1 0 306 0
Earlycolonization 10.8 2 1 137 37
Illicitcoca 29.7 4 20 71 140
Improvedaccess 36.6 6 15 49 151
Complexclearance 58.8 12 7 11 203
Plotexhaustion 83.3 20 2 3 135
100
350
300
250
200
150
100
50
80604020
Mea
n pa
tch
size
100
250
200
150
100
50
80604020
Tota
l for
est e
dge l
engt
h
Proportion of forest cleared (%)(b)
Proportion of forest cleared (%)(a)
FIgURe 7.4 Metricscalculatedfortheconceptualmodel:(a)meanforestpatchsize,and(b)totaledgelength.Bothareinnominalunits.
42963_C007.indd 129 10/29/07 10:49:10 AM
130 Land Use Change
10080604020
Mea
n pa
tch
size (
km2 ) 1000
800
600
400
200
Tota
l for
est e
dge l
engt
h (k
m)
MN
N d
istan
ce b
etw
een
fore
st p
atch
es (k
m)
10080604020
250
200
150
100
50
10080604020
1.2
1.0
0.8
0.6
0.4
0.2
Proportion of forest cleared (%)(a)
Proportion of forest cleared (%)(b)
Proportion of forest cleared (%)(c)
FIgURe 7.5 SelectedmetricsfromTraniandGiles40: (a)meanforestpatchsize inkm2,(b)totaledgelengthinkm,and(c)meannearestneighbordistancebetweenforestpatchesinkm.
42963_C007.indd 130 10/29/07 10:49:12 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 131
chapter(Figure7.6)and,asforestlosshadonlyreached46%by2000forthiscom-munity,thedatadonotextendtoveryhighproportionalforestlosses.Comparingthe limited number of metrics in Figures7.4, 7.5, and 7.6 provides an initial testof the robustnessof theconceptualmodel.Adecline in forestcoverover time intheconceptualmodelisclearinTable7.2.ThisallowsthesuccessivephasesofthemodeltobeparameterizedasproportionalforestlossesinthegraphsinFigure7.4.BothLambin11andTraniandGiles40usedforestcoverintheirgraphs.
Mean patch size declines in a consistent manner (Figures7.4a, 7.5a, and 7.6a).Initially thedecline is rapid,more so in reality (inCommunidadArequipa) than inthemodelorthesimulation.Ataroundhalftheareadeforested,therateofdecreaseinmeanpatchsizedeclinesandthentherateofdecreaseinforestpatchsizetapersoff.Totaledgelengthinitiallyincreasesasforestsbegintobecleared,andtheedgelengthisatitsgreatestatintermediateforestcoversandthendeclines.Thisisevidentintheconceptualmodel, the simulation, and in the realdata,Figures7.4b,7.5b, and7.6b,respectively.Therearedifferencesinthepeakvaluesoftotaledgelength,whichsuggestthevariationinthepeakmayberelatedtothespatialconfigurationoffragmentation.WhereasTrani andGiles40 simulateda somewhat randompatternof fragmentation,thatinCommunidadArequipaandthesimulationmodelareforregularlystructuredlandscapes,which,becausepropertiesarelongandthin,hasatendencytohavehighedge lengths at intermediate forest losses compared to lower edge lengths in morerandomlyfragmentedforests.Thethirdmetric—meannearestneighbor(MNN)dis-tancebetweenforestpatches—hasonlybeencalculatedfortheactualdatafromCom-munidadArequipaandextractedfromTraniandGiles’s40simulation.Forthismetricthere isadifference inbehavior.There is relatively littlevariation in theMNNdis-tanceinCommunidadArequipa,andthedistancesovertherangeofforestcoversinthe community describe a shallow U-shape. However, Trani and Giles’s simulationdescribesanupturned-UdistributioninMNNdistanceasforestisprogressivelylost.
Insummary,ouranalysisoflandscapemetricssuggeststhat:
1.Meanpatchsize(MPS)andtotaledge length(TEL)areveryrobustandconsistentmeasuresoffragmentation.AlthoughinallthreecasestherearesimilartrendsinMPSandTELwithprogressivedeforestationandincreas-ingforestfragmentation,therearedifferencesintheprecisenatureofthecurves,whichmaybeafunctionoftheeffectthatthespatialconfigurationof land properties have on fragmentation. From the viewpoint of modelvalidationthelatterpointisnotthatimportant,butitdoeslendweighttotheargumentthatunderstandinglandowner’sdecisionsatthesmallscaleisimportantfordevelopingourunderstandingoffragmentationandfutureplanningofcolonizationzones.
2.Thebehaviorofthemeannearestneighbordistancebetweenforestpatchesdoes, however, vary between the studies and is due either to the differ-encesinspatialscalebetweenthetwostudiesorthenatureofforestloss.InCommunidadArequipathespatialimprintofthecadastralgridandthewaysinwhichpeopleclearforestwithinthegridmayleadtoarestrictedset of possible distances between forest patches and, as these have onlybeenobservedintheearlystagesoffragmentation,theymaychangesig-nificantlyasmoreforestisclearedandfragmentationproceeds.
42963_C007.indd 131 10/29/07 10:49:12 AM
132 Land Use Change
10080604020
Mea
n pa
tch
size (
km2 )
MN
N p
atch
es (k
m2 )
60
50
40
30
20
10
Tota
l for
est e
dge l
engt
h (k
m)
10080604020
120
100
80
60
40
20
10080604020
60
50
40
30
20
10
Proportion of forest cleared (%)(a)
Proportion of forest cleared (%)(b)
Proportion of forest cleared (%)(c)
FIgURe 7.6 MetricscalculatedforCommunidadArequipa:(a)meanforestpatchsize in km2, (b) total edge length in km, and (c) meannearest neighbor distancebetweenforestpatchesinkm.
42963_C007.indd 132 10/29/07 10:49:13 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 133
3. In theanalysisof landscapemetricsdescribed in thischapter the resultsfromTraniandGiles’s studymaydiverge fromourconceptualmodelattheplotexhaustionstageinwaysthatwedonotyetunderstand.ThismayoccurbecausewhereasthedeforestationsimulatedbyTraniandGileswasfor progressive deforestation, what happens in colonization schemes isthattherecanbesignificantregrowthinthemediumtolaterstageinthesequenceofdeforestation.
7.5 DIsCUssIon
Wesee the conceptualmodelwehavedeveloped fromChapareas apreliminaryattempt to thickenourunderstandingofwhyparticularpatternsof fragmentationoccur in a so-called fishbone colonization scheme. The regularity of the spatialimprintsofroadsandpropertygridsinsuchareasmadetheminterestingcandidatesforthisinitialinvestigation(Figure7.7).Thisstudyis,however,limited,partlybytherelativelysmallnumberofcommunitieswehaveresearchedintensively,whichmight lead tocontext specificgeneralizations,42 andpartlybecausewehaveusedretrospective analyses of decision making by colonists. While taking a differentapproachtoPerzandWalker25—byfocusingoncolonist’sresponsestochangesineconomicconditionsandanti-cocapoliciesandthelossofprimaryforest—wejointhemintheirclarioncallforresearcherstofocusontheforestoutcomesofsmall-farmcolonists.Theystatethat“most land use models … do not take account of land taken out of production and left to fallow”(p.1009),andwewouldarguethatmostlandusemodelsdonottakeaccountoflandtakenoutofproduction,lefttofallow,or how land under primary and secondary forest is spatially configured. Onlyifweresearchalongtheselineswillwebeabletoapplymethodssuchasthoseadvocatedtoplancolonizationinforestareas43orbeabletorestorefragmentedareas.44
Thefieldthenisopenforfurtherresearch,andwearguethefollowinglinesofinvestigationareneededtodeepenourunderstandingfurther:
1.More communities in Chapare could be investigated, particularly thosewithphysicalandsocioeconomiccharacteristicsotherthanthoseoutlinedinTable7.1.However,amoreprofitablelineofinvestigationwouldbetoresearchclustersofadjacentcommunities.Thiswouldallowinteractionsbetweencommunitiesattheirboundariestobeinvestigatedandmightalsorevealtheextenttowhichcooperationbetweencommunitiestakesplace,orgaugethepotentialofcooperationinthefuture.
2.ComparativeresearchbetweenChapareandsimilar—fishbone—coloniza-tionschemesintheAmazonBasinwouldenableustofurtherconsidertherobustnessofmanyaspectsoftheconceptualmodelwehavedevelopedandenableustomovetowardageneraltoolthatcouldbeusedforbasin-wideintegratedplanningincolonizationschemes.
3.So far,ourmodel reliesona retrospectiveanalysisofdata.However, itsutilityforplanningliesinitsabilitytointegrateconservationintheplan-ningoffuturecolonizationintheAmazonBasin.Therefore,researchintodecisionmakingbyindividuallandownersandsindicatos underdifferent
42963_C007.indd 133 10/29/07 10:49:14 AM
134 Land Use Change
futureeconomic,political,andenvironmentalscenariosisbothattractivetoresearchersandessentialtoplanners.
4.Aswenotedin the introduction, thisstudyisgroundedinoneclearancetypology—planned settlement leading tofishbonepatternsof land clear-ance.Othertypologiesrequiresimilarresearch.
Basedonourobservationsanaggingquestion remains: is itpossible tobuildstructuresforconservationinplansforcolonizationschemes?Therequirementisto
a)
b)
c)
FIgURe 7.7 (a)Chapareroadbuilding.TheextensionoftheroadnetworkinChaparecon-tinues.Thisphotographwas taken inAugust2003,andshowsa roadbeingpusheddeepintorelativelyintactlowlandtropicalforestalongthelineofaformerfootpathwhichlinkedsomeisolatedsettlementstotheformerroadhead.(b)IntheeastofChaparecattlerearingisthemainfarmingactivity.Landparcelshereare50hainarea,comparedto20hainsettle-mentsdominatedbycultivation.(c)BananasareoneofthemainalternativecropstococainChapare.Theyoftenresultinlargeareasbeingclearedtocompensateforlossofcocaincomewhichleadstolargeareasofnonforestmonoculture.Theroadcuttingthroughthisbananaplantationisanunimprovedprimaryfeederroad.
42963_C007.indd 134 10/29/07 10:49:15 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 135
setasidesomelandwithinthecolonizationschemesthatwilleitheractas“sinks”or“reservoirs”withthecolonizationzone,ortoplanwildlifecorridors—eithercon-tiguous forest corridors or stepping-stones of forest patches for migration acrossand within the colonization landscapes. Theoretically this is possible, but wheredoes thelandcomefrom?InChaparetheoriginalcadastrehadlandthatwasnotassignedtosettlers;oftenstripsofforestalongriversor, to thenorth,forests thatare inundated for many months each year. In other words wildlife corridors andforest patches were “planned by accident” but were not afforded protection untilforestsalongwatercourseswerespecificallyprotectedinthenewForestryLawthatwaspassedin1996.45Disappointinglyfromtheviewpointofconservation,butnotsurprisinglygiventhedemandsonlandandthelaissezfaire attitudetospontaneouscolonization, many of these unallocated lands have been occupied, the exceptionbeingtheinundatedforests.Forexample,adjacenttoCommunidadArequipaastripofforestlandalongariverhasbeenillegallycolonized,andanunallocatedforestareaadjacenttoCommunidadCaracaswasaddedtothecommunityasitexpandedaftertheyhadpetitionedINC.
Ifwe fail to thickenourunderstandingofhow landmanagersmake landusedecisionsinlandscapesofcolonization,primaryforestswillcontinuetodisappearinwayswedonotcomprehend,secondaryforests(whichhavedifferentecologicalproperties and conservation values to primary forest) will come and go, and thespatialoutcomesmaycontinuetosurpriseus.Theselandscapeswilllosetheirabilitytoallowfaunalandfloralinterchangebetween“intactforestblocks”asthedeforesta-tionfrontstheyrepresentclosedown.Newdeforestationfrontswillopenup.Ifweareunabledevelopa “thick”understandingand inject it intoplanningprocesses,thediseaseof“thin”understandingwillcontinuetoprevailandinalllikelihoodthescenariosoutlinedforroadcorridorsinBrazilbyFearnside29andLaurenceetal.30,31willbecomecommonplaceinAmazonia.
ACKnoWleDgMents
PartofthisresearchwasfundedbytheEuropeanUnion(ContractERBIC189CT80299).AndrewBradley’sPhDresearchwaspartlyfundedbythisgrant,aswellasaSlawsonAwardfromtheRoyalGeographicalSocietywiththeInstituteofBritishGeographers,andfundingfromtheUniversityofLeicester.WearegratefultoRichardJ.AspinallandMichaelHill for invitingus topresentourresearch, toChristianBrannstromforcriticallycommentingonanearlyversionofthischapter,andtoFelixHuancaViracaforaccompanyingusinthefield.
ReFeRenCes
1. Godron,M.,andForman,R.T.T.Landscape Ecology.JohnWiley&Sons,NewYork,1986. 2. Fahrig,L.Effectsofhabitatfragmentationonbiodiversity.Annual Review of Ecology
and Systematics34,487–515,2003. 3. Ries,L.etal.Ecologicalresponsestohabitatedges:mechanisms,modelsandvariability
explained.Annual Review Ecology, Evolution and Systematics 51,491–522,2004. 4. Turner,I.M.Specieslossinfragmentsoftropicalrainforest:Areviewoftheevidence.
Journal of Applied Ecology33,200–209,1996.
42963_C007.indd 135 10/29/07 10:49:16 AM
136 Land Use Change
5. Debinski,D.M.,andHolt,R.D.Asurveyandoverviewofhabitatfragmentationexperi-ments.Conservation Biology14,342–355,2000.
6. Bierregaard,R.O.,Jr.,andLovejoy,T.E.Thebiologicaldynamicsoftropicalrainforestfragments.Bioscience42,859–866,1992.
7. Laurence,W.F.,andBierregaard,R.O.Fragmented tropical forests.Bulletin of the Ecological Society of America77,34–36,1996.
8. Laurence,W.F.,andBierregaard,R.O.,eds.Tropical Forest Remnants.UniversityofChicagoPress,Chicago,1996.
9. Laurance, W. F. et al. Ecosystem decay of Amazonian forest fragments: A 22-yearinvestigation.Conservation Biology16,605–618,2002.
10. Husson,A. et al.Studyof forestnon-forest interface:Typologyof fragmentationoftropicalforest.TREESSeriesB,ResearchReportNo.2,EuropeanCommissionEUR16291EN,Brussels,1995.
11. Lambin, E. F. Modelling and monitoring land-cover change processes in tropicalregions.Progress in Physical Geography21,375–393,1997.
12. Mertens,B.andLambin,E.F.SpatialmodellingofdeforestationinSouthernCameroon.Spatial desegregation of diverse deforestation processes. Applied Geography 17,143–162,1995.
13. Imbernon,J.,andBranthomme,A.Characterizationoflandscapepatternsofdeforestationintropicalrainforests.International Journal of Remote Sensing 22,1753–1765,2001.
14. Hargis,C.D.,Bissonette,J.A.,andDavid,J.L.Understandingmeasuresoflandscapepattern.In:Bissonette,J.A.,ed.,Wildlife and Landscape Ecology: Effects of Pattern and Scale,Springer-Verlag,NewYork,231–261,1997.
15. Rudel,T.,andRoper,J.Thepathstorainforestdestruction:Crossnationalpatternsoftropicaldeforestation,1975–90.World Development25,53–65,1997.
16. Geist,H.J.,andLambin,E.F.Whatdrivestropicaldeforestation?LUCCReportSeriesNo.4,CICAO,Louvain-la-Neuve,2001.
17. Lambin,E.J.,Geist,H.J.,andLepers,E.Dynamicsandland-useandland-coverchangeintropicalregions.Annual Review of Environment and Resources28,205–241,2003.
18. Alves,D.Space-timedynamicsofdeforestationinBrazilianAmazonia.International Journal of Remote Sensing23,2903–2908,2002.
19. Arima, E. Y. et al. Loggers and forest fragmentation: Behavioral models of roadbuildingintheAmazonBasin.Annals of the Association of American Geographers95,525–541,2005.
20. Brondizio, E. S. et al. The colonist footprint: Toward a conceptual framework ofdeforestation trajectories among small farmers in frontier Amazonia. In: Wood C.,andPorro,R.,eds.,Deforestation and Land Use in the Amazon.UniversityPressofFlorida,Gainesville,Fla.,133–161,2002.
21. Walker,R.T.MappingprocessandpatterninthelandscapechangeoftheAmazonianfrontier.Annals of the Association of American Geographers93,376–398,2003.
22. Walsh,S.J.etal.Characterizingandmodelingpatternsofdeforestationandagricul-turalextensificationintheEcuadorianAmazon.In:Walsh,S.J.,andCrews-Meyer,K.,eds.,Linking People, Place and Policy.Kluwer,Boston,187–215,2002.
23. Moran, E. F. et al. Integrating Amazonian vegetation, land-use and satellite data.BioScience44,329–338,1994.
24. Pichón,F.J.Theforestconversionprocess:adiscussionofthesustainabilityofpre-dominantlandusesassociatedwithfrontierexpansionintheAmazon.Agriculture and Human Values13,32–51,1996.
25. Perz,S.G.,andWalker,R.T.HouseholdlifecyclesandsecondaryforestcoveramongsmallfarmcolonistsintheAmazon.World Development30,1009–1027,2002.
26. Reid, J. W., and Bowles, I. A. Reducing the impacts of roads on tropical forests.Environment39,10–17,1997.
42963_C007.indd 136 10/29/07 10:49:16 AM
Developing a Thick Understanding of Forest Fragmentation in Landscapes 137
27. Holdsworth,A.R.,andUhl,C.FireinAmazonianselectivelyloggedrainforestandthepotentialforfirereduction.Ecological Applications7,713–725,1997.
28. Nepstad,D.etal.Roadpaving,fireregimefeedbacks,andthefutureofAmazonforests.Forest Ecology and Management154,395–407,2001.
29. Fearnside, P. M. Avança Brasil: Environmental and social consequences of Brazil’splannedinfrastructureinAmazonia.Environmental Management30,735–747,2002.
30. Laurence, W. F. et al. The future of the Brazilian Amazon. Science 291, 438–439,2001.
31. Laurence,W.F.etal.DeforestationinAmazonia.Science304,1109–1111,2004. 32. Nepstad,D.etal.AvançaBrasil:Os Custos Ambientais para Amazônica.Institutode
PesquisaAmbientaldeAmazônica(IPAM),Belém,Brazil,24pp,2000.(Availableat:http://www.ipam.org.br/avanca/politicas.htm).
33. Soares-Filho,B.etal.Simulatingtheresponseofland-coverchangestoroadpavingand governance along a major Amazon highway: The Santarém-Cuiabá corridor.Global Change Biology 10,745–764,2004.
34. Bradley,A.V.Land-UseandLand-CoverChangeintheChapareRegionofBolivia.PhDthesis,UniversityofLeicester,UK,2005.
35. Henkel, R. The Chapare of Bolivia: A Study of Tropical Agriculture. PhD thesis,UniversityofWisconsin,Madison,Wisc.,1971.
36. Henkel,R.Coca(Erythroxylum coca)cultivation,cocaineproduction,andbiodiversitylossintheChapareregionofBolivia.In:Churchill,S.P.,Balslev,H.,Forero,E.,etal.,eds.,Biodiversity and Conservation of Neotropical Montane Forests.TheNewYorkBotanicalGarden,Bronx,N.Y.,551–560,1995.
37. Millington, A. C., Velez-Liendo, X., and Bradley A. V. Scale dependence in multi-temporal mapping of forest fragmentation in Bolivia: Implications for explainingtemporal trends in landscape ecology and applications to biodiversity conservation.Photogrammetry and Remote Sensing57,289–299,2003.
38. Steininger,M.C.etal.TropicaldeforestationintheBolivianAmazon.Environmental Conservation 28,127–134,2002.
39. MACA:MinisteriodeAsuntosCampesinosyAgropecuarios.ModelsofExploración–Tipo,parasieteSubregionesdelsubtropicohumidodeCochabamba.IBTA,Chapare,Bolivia,1991.
40. Trani,M.K.,andGilesR.H.,Jr.Ananalysisofdeforestation:Metricsusedtodescribepatternchange.Forest Ecology and Management114,459–470,1999.
41. McGarigal,K.,andMarks,B.Fragstats.Spatialpatternanalysisprogramforquantifyinglandscapestructure.ForestScienceDepartment,OregonStateUniversity,Corvallis,1994.
42. Fujisaka,S.,andWhite,D.Pastureorpermanentcropsafterslash-and-burncultivation?Land-usechoiceinthreeAmazoncolonies.Agroforestry Systems42,45–59,1998.
43. Venema, H. D., Calamai, P. H., and Fieguth, P. Forest structure optimization usingevolutionary programming and landscape ecology metrics. European Journal of Operational Research16,423–439,2005.
44. Lamb, D. J. et al. Rejoining habitat remnants: Restoring degraded forest lands. In:Laurence,W.F.,andBierregaard,R.O.,eds.,Tropical Forest Remnants.UniversityofChicagoPress,Chicago,1997.
45. GovernmentofBolivia.LeyForestal1700.LaPaz.Bolivia,1997.
42963_C007.indd 137 10/29/07 10:49:17 AM
42963_C007.indd 138 10/29/07 10:49:17 AM
139
8 Urban Land-Use Change, Models, Uncertainty, and Policymaking in Rapidly Growing Developing World Cities: Evidence from China
Michail Fragkias and Karen C. Seto
Contents
8.1 Introduction................................................................................................. 1398.2 ModelingUrbanLandUseChange,Policymaking,andUncertainty........ 140
8.2.1 ModelingUrbanLandUseChange................................................. 1408.2.2 PolicyMaking.................................................................................. 1428.2.3 TheIntersectionofModelingandPolicyMaking........................... 1438.2.4 PolicyEvaluationandUncertainties................................................ 146
8.3 ANewApproachinModelingUrbanGrowthinDataSparseEnvironments.............................................................................................. 1478.3.1 MechanicsoftheModel.................................................................. 1478.3.2 ApplicationtoThreeCitiesofthePearlRiverDelta,China........... 150
8.4 DiscussionandConclusions........................................................................ 154References.............................................................................................................. 159
8.1 IntroduCtIon
Projectionssuggestthatastheworld’surbanpopulationwilljumpto61%by2030(from today’s50%mark),mostof thisurbangrowthwilloccurprimarily in lessdeveloped countries, and in Asia in particular.1 Much interest already exists inmegacities—citieswithpopulationsof10millionormore—onwhichasignificantamountofinformationisbeingcollected.Ithasbeennotedthoughthatthemajorityofurbangrowthwilloccurinmedium-sizedcities.2Giventhaturbangrowthisa
42963_C008.indd 139 11/6/07 7:12:47 AM
140 Land Use Change
major component of global environmental change3,4 and the danger of potentialundesirableenvironmentalandsocialeffectscausedbyhighratesofgrowthisever-present,therelativeimportanceofstudyingmedium-sizedcitiesversusmegacitiescitiesinthenextcenturyishigh.Furthermore,developingworldcitieshavelimitedhumanandfinancialresourcesemployedinvariousaspectsofpolicymaking.Con-sequently,thecollectionofreliabledataandtheuseofmoreadvancedmethodsinplanningpracticeandpolicymakingbecomesextremelydifficult(Figure8.1).
Policymakersindevelopingworldcitiesareincreasinglyfacedwithpressuretoassesstheimpactoftheirlandusestrategiesandpolicies5ashighpopulationgrowthtrends are predicted for at least the next 25 years. Potential socioeconomic andenvironmentalimpactsofpoliciescanbeassessedwithquantitativemodels.Giventhe number and underlying motives of different approaches to modeling, policy-makers,especiallythoseindevelopingworldcities,couldbenefitfromassistanceinchoosingthemostappropriatemodel.Iscurrenttechnologyormethodologyadvance-mentbasedoncurrentandrecentfuturerealitiesofmedium-sizeddevelopingworldcities?Prosandconsofdifferentmodelingapproachesforlandusepolicymakingneed to be evaluated given the particularities of such cities (e.g., the problem ofincompleteandscarceinformation).Thesuccessofsustainabledevelopmenteffortsreliessignificantlyontheidentificationofbetter(asaccurateaspossible)forecastingschemesregardingratesandpatternsoffutureurbandevelopmentthatalsoconnectbetterwiththeprocessofpolicymaking.Thus,thischapterprovidesaninquiryintoquestionsandtradeoffsapolicymakerfaceswhenitcomestothechoiceofcontext-specific suitable modeling tools and the establishment of guidelines assisting thedecision-makingprocess.
Thepurposeofthischapteristhreefold.First,itdiscussesissuesofurbanlandusechangemodelingandexplorestheintersectionoflandusemodelingwithurbanpolicymakingatdifferentscalesinthecontextofdevelopingworldcities.Second,it discusses the effects of uncertainties in the data sources, theories, and modelsmethodsofaddressingtheseissues.Third,itreviewsapredictivemodelofrapidurbantransformationthatrelatesastandardmodelingtraditiontoanexplicituncertainty-reducingpolicy-makingframeworkusingChinesecitiesasacasestudy.
8.2 ModelIng urban land use Change, PolICyMakIng, and unCertaInty
8.2.1 Modeling Urban land Use Change
Urbanareas,andtheirformandfunction,havebeenstudiedinthecontextsofurbanplanning,urbaneconomics,urbangeography,andurbansociology,muchofwhicharegroundedinthespatiallandusemodelsofvonThünen.6Aneedforquantita-tiveanswersregardingtheeffectsofextent,rateofchange,andpatternsofglobalurbanlandusechangehasledtothedevelopmentofurbanlandusechangemodels(ULCM).AULCMisasimplificationofreality,anditssuccessliesinretainingthefundamentalcharacteristicsofthesystembysimplifyingrealityasmuchasneces-sary(butnotbeyondthat).Thoughtofasatool,aULCMtargets“usefulness”;unfor-tunately,thiscapacityisnotalwayssuccinctlystatedordemonstrated.
42963_C008.indd 140 11/6/07 7:12:47 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 141
-180
-60
Robi
nson
Pro
ject
ion
Cent
ral M
erid
ian
0.00
Sour
ce: E
SRI D
ata &
Map
s CD
Kilo
met
ers
0
1,45
0
2
,500
5,50 0
8,70
0
11,6
00
C
ount
ry B
ound
ary
POPU
LATI
ON
7500
00 –
1000
000
1000
001
– 300
0000
3000
001
– 500
0000
5000
001
– 100
0000
0
1000
0001
– 23
6200
00
Anta
rctic
Circ
le
Trop
ic of
Cap
ricor
n
Equa
tor
Pacifi
c Oce
an
Trop
ic of
Can
cer
Indi
an O
cean
Pacifi
c Oce
an
Atla
ntic
Oce
an
–180
–160
–140
–120
–100
–80
–60
–40
–20
020
4060
8010
012
014
016
018
0
–160
–140
–120
–100
–80
–60
–40
–20
–60
–40
–20
20
40
60
80
0
–40
–2020
40
60
80
0
020
4060
8010
012
014
016
018
0
Arcti
c Cir
cle
FIg
ur
e 8.
1 (S
ee c
olor
ins
ert
follo
win
g p.
132
.) Po
pula
tion
dis
trib
utio
n of
the
wor
ld’s
urb
ana
gglo
mer
atio
nsw
ith
750,
000
peop
leo
r m
ore
arou
ndy
ear
2000
.
42963_C008.indd 141 11/6/07 7:13:05 AM
142 Land Use Change
Theusefulnessofanurbanlandusechangemodelisjudgedinconnectiontothegoalofthemodelingexercise;thesegoalscanbetightlyorlooselyconnectedwithgoalsofpolicymakers.InthischapterwediscusstwodistinctfunctionsforaULCM:explanationandprediction/forecasting(thatleadstoprescription).Initsfirstfunc-tion,ithelpstheresearcherorthepolicymakerimprovehisorherunderstandingofprocessesthatleadtochangeandshedlightonelementsofcausalityguidedbyandtestingalternativetheoriesofurbangrowth.Initssecondfunction,itcandescribeandpredictthetypesoflandusechangethatoccur(type,amount,rate,pattern,andtimingofchanges)andmorepromptlyleadtoprescription.
Therearenowdozensoflandusemodelsavailable;areviewandtypologyof(urban)landusechangemodelshasbeenpresentedindetailelsewhere.7,8,9,10Manynew“flavors”ofmodeling arebeingdeveloped.11,12This proliferation reflects themethodologicalprogress in theattempt tounderstandorpredict thenatureof thelandscape, the types of changes occurring, the causal structure connecting theunderlyingfactorsofchange,andthehypothesestobetested.Alternativeclassifica-tionsofurban landusechangemodels includea three-dimensionalcontinuumofspatial scale, time scale, andhumandecision-making,11 overlapping categoriesofequation-basedsystem,statistical technique,expert,evolutionary,cellular,hybrid,andagent-based,13anddistinctcategoriesof large-scale, rule-based, state-change,andcellularautomata.14
ULCMsoftenclaimpolicyrelevancebutlackacleardefinitionofthedegreeofthisrelevance.Landusechangemodelingiscurrentlyweaklycoupledwithlandusepolicymaking.Althoughwedonotclaimaneedforaverystrongcoupling(duetotheadverseresourceandpolitical realityforsucha task indevelopingcountries),wesuggestthatitneedstobestrengthenedforoptimalknowledgeutilizationinthepolicy-makingprocess.Thiscanbeachievedbyexplicitlyintroducingmechanismsformodel uncertainty reduction and a policy-makingmodule in landuse changemodels.Itisveryimportantthattherelevanceofmodelsismoreclearlyunderstoodandfuturedirectionsreevaluated.Inwhatfollows,weaddressissuesexistingatthemodeling–policy-makinginterface.
8.2.2 PoliCy Making
Policy-relevant landusechangemodelsmay target avarietyof types, levels, andstagesofpolicy-makingactivitythatheavilyinfluencesobservedlandusepatterns.Some facets of urban land use change derive in part from policies implemented(synchronously or asynchronously) at different administrative unit levels: at localmunicipal,county,state,prefecture,andregionallevels.Nationalmacroeconomic,regional, and local policies have dramatic direct and indirect effects on agents’choicesofcurrentandfuturelanduse.Policymakersattheselevelsincludearangeofpublicofficials,suchasurbanandenvironmentalplanners,andvariousadminis-tratorsatlocalgovernmentagencies.
Atthelocalandnationallevels,concernsregardingsocialwelfaremeasuredinlevelsofconsumption,productiveactivity, cityamenitiesanddisamenities, exter-nalitiesandideasofsustainabilityguidepolicy-makingeffortsintargeting—amongothergoals—an“optimal”urbanareasize,shape,andpopulationmix.Localurban
42963_C008.indd 142 11/6/07 7:13:06 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 143
andexurbangovernmentsconsistentlyutilizezoning,growthcontrols,andtaxation/subsidiestodriveurbangrowthandregulate,distribute,orredistributegainsfromurban development (while implicitly targeting that the maximization of propertyvaluesinurbanareas).Increasingly,environmentalconcernsregardingtheimpactofurbanlanduseconversionsalsodirectpolicymaking.Atthegloballevel,institution-allydesignedpoliciesinfluenceprocessesofurbanlandusechangeinamultitudeofwaysthroughtheestablishmentofincentiveschemesandstructuraladjustmentpro-grams.Closemonitoringofurbanpopulationtrendssuggeststheheightenedinterestofinternationalfinancialandotherinstitutions(suchastheUnitedNationsandtheWorldBank)thatdriveglobalchange.
Manytheoriesofpolicyformationexist,withdifferentassumptionsregardingknowledgeutilizationwithintheformationprocess.Mosturbangrowthmodelsarenotusuallyexplicitontheirassumptionsregardingthepolicy-makingprocess;themostwidelyadoptedviewofthepolicy-makingprocessisthatoftherationallinearprocessoragendasettingtheory.Aswithmosttechnicalanalysisenteringthepolicyrealm,thepolicyrelevanceofalandusemodelisofamoreinformationalnatureratherthanaconcretepolicydrivernature.
Fromtherationalpolicymaker’sstandpoint,theuseofalandusechangemodelinvolvesasequenceofdecision-makingstepsandactionsthatrequires(i)theexami-nationofavailablemodelingoptions,(ii)thechoiceofmodelevaluationcriteriaandtheirweights(dependingonthepreferencesofthepolicymakerandtherealitiesofthepolicy-makingsetting), (iii)evaluating themodelby theselectedcriteria,and(iv)deriving the overall evaluation through the collection of individual weighedcriterionevaluations.5Criteriafortheselectionofamodelingprocessmayincludetheemphasisonpredictionversusexplanation, thedata sparsenessor richnessofthepolicy-makingenvironment,levelsofuncertaintyinthequalityofthedata,theemphasis on probabilistic versus heuristic/mechanical approaches, the flexibilityof the model to alternative variable specifications, the sophistication in accuracyassessment(validation)ofpredictions,theneedofdeepversusbasicunderstandingof themodelingapproach, theneed forweakversusstrongcouplingofmodelingwiththeprocessofpolicymaking,themodel’scapacitytoinformaboutavarietyofpolicy-makinggoalsandatdifferentlevelsofpolicymaking,andtheemphasisonthetheoreticalfoundationofthemodelingapproach.Severalofthesecriteriaarediscussedinmoredetailbelow.
8.2.3 The inTerseCTion of Modeling and PoliCy Making
Recently we have witnessed a scarcity of application of ULCMs for developingworldcities.ThisreflectsanunderestimationofthepotentialeffectsofurbangrowthinLDCs,theproblematicnatureofempiricalworkinLDCs,alackofunderstandingofwhatcouldbethebestmodelingoptionavailabletoadecisionmakerinadevel-opingworldcity,andadearthofapplicableULCMs.Throughourworkwearguethatpresentpressingpredictiveneedselevate theimportanceofstatisticalmodelsthatutilizeaminimalinputscheme.Modelswithsimpleinputrequirementscanfindwiderapplicationinaddressingcurrentandfutureneedsinthesecities.DatasetsinLDCsarescarceandinmanycasesinexactduetoinstitutionalfactorsandlimited
42963_C008.indd 143 11/6/07 7:13:07 AM
144 Land Use Change
resources. A future increased allocation of resources toward the collection ofdetailedgeoreferencedsocioeconomicdatabythegovernmentsofthesecountriesisnotcertain,andalthoughdataarebeingcollectedataninternationallevel, thisoccursataveryslowpaceandataquiteaggregatedlevel.Furthermore,problematicmeasurementcanbecatastrophicforthepredictivepowerofmodelsthataresuc-cessfulincapturingthetruedata-generatingprocess(DGP).
Understanding the importance of knowledge utilization in decision making,thequestionoftherelativeimportanceofexplanationversuspredictionforpolicy-makersarises.Whendoesapolicymakerneed(i)predictionsregardingthelocationandtimingoflandusechangeunderalternativescenariosand/or(ii)theknowledgeof whether theoretical hypotheses stand up to statistical tests and of magnitudesoftheexpectedchangesassociatedwithshiftsinavarietyofpolicyleveragesandviceversa?Itisnotclearifthepolicymakeralwaysneedsadeeperunderstandingofprocessesandknowledgeofthecausationchain.Possibly,theanswertosuchaquestiondependsontheactualpolicy-makingformationprocessandthetype/levelofgovernmentor institution responsible for thedecision.Variousauthors suggestthat,ataminimum,policymakersshouldbeabletounderstandthefoundationsofamodelingapproachoratleastbeabletoidentifyhowtheresultsaregenerated.10,15Giventhenumberandlevelofcomplexitiesofalternativemodelingapproaches,thismaybeanunrealistictarget.Ourexperiencewithdevelopingworldcitiesshowsthatpolicymakersaredefinitelymoreinterestedinknowinghowshiftsinpoliciesaffectoutcomes;theymaynotwanttoknowtheinnerworkingsofthemodel.
Policymakerpreferencesoveroutputdefines if andwhen thepolicymakerhasa stake in the choice of methodology (e.g., process-based or mechanistic models).Althoughsocioeconomicprocessesgeneratetheobservedlandscapeoutcomes,modelsthatbelongtoarule-basedapproachmayinfactresultinbetterpredictionsthanpro-cess-basedmodelsutilizingsocioeconomicdata.Thiscanbepartlyattributedtodataimperfection:variablescapturingthesocioeconomicprocessescanbeinaccurateorsimplytheseprocessesmightbehardorimpossibletoquantify.Mechanisticmodelsusedataconstructs thatarebasedonproximate(rather thanunderlying)causesoflandusechange.Unfortunately,thesemodelsarealsomoresensitivetoomittedorinexistentinformation,afactthatcanpotentiallymisguidepolicymaking.Process-basedmodelscanstillbesuccessful tovariousdegrees for forecasting,dependingonthegeographicallocation,methods,andaerialunitlevelofanalysisemployedforprediction.Suchmodelswithprovenhighpredictivepowerarealsousuallybasedonproximateratherthanunderlyingcauses.Naturally,successesinpredictivecapabilityofrule-basedmodelsdonotvoidthesearchforaDGP.
Models—asopposed to theories—of landusechange theoryhavebeenmorepopulartoolsforpolicymakingandhavebeendevelopedmoreforbothsubstantiveandpracticalreasons.10Substantivereasonsincludethecomplexityofthelandusechangephenomenaandthecomplexinterrelationsofvariousinstitutional,cultural,political,economic,andsocialchangedeterminants in theoreticalwork.Practicalreasonsincludetheavailabilityofresourcesandthe“demandsoftheofthedecisionmaking‘clientele’.”Inshort,solidquantitativeresultsthataremarketable,visually
42963_C008.indd 144 11/6/07 7:13:08 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 145
powerfulandreadyforuseastoolsforawiderangeofdecisionmakersarevaluedmorehighly—andmodelsproduceresultsmuchbetterthantheory.�
Established approaches in different scientific disciplines and pressures regard-ingpeeracceptanceandcareeradvancementalsodrivemethodology-relatedchoicesforurbanlandusechangemodelsandarepartlyresponsibleforlooseconnectionsofmodelswithpolicymaking.Theevidenceforthisisanecdotal,astheauthorshavebeenexposedtosuchcomplaintsinpersonaldiscussionswithotherresearchers.Inshort,theproducer’s(anacademicresearcher)incentivesforconsiderableoutputintheformofjournalpublicationsmayleadtomodelslooselyorvaguelyconnectedtothepracticeofpolicymaking.Thisissueisadmittedlydifficulttoresolveunderthecurrentpractices.
Awarenessofthetheoreticalfoundationofanapproachmayalsobeimportantforthepolicymaker’schoice.Urbancellularautomata(CA)models,forexample,haveatheoreticalgroundingonideasofcitiesasself-organizedandemergentphenomenain bottom-up complex systems and fail to capture urban growth in the top-downpoliticaldimension.Unfortunately,thesearestill“largelyabstractarguments.”16Evencutting-edgeadvancedmultiagentsystemCA(MAS/CA)simulatingcities“atthefinescaleusingcells,agents,andnetworks”arefornowfarfrombeingreadyforanyprac-ticaluseor“largely…pedagogic”value.16
Asthedecision-makingclienteleofurbanlandusechangemodelstargetsavari-etyofgoals,agoodmodelshouldaccommodatesuchavariety.Policymakerscarefordifferentsizeadministrativeareasdependingonwhethertheyareemployedatalocal,provincial,ornationallevel.Modelsshouldbeabletoaddressneedsofeachlevelofdecisionmakinganddistillresultsderivedatthehighestleveltothelowestlevel and vice versa. Connected to this issue is the capacity to address single ormultipleneighboringurbanareasinthesamemodel.Singleurbanmetropolitanareaanalysisisnotinclusiveofsurroundingregionalspatialdynamics(immigrationandout-migrationflows,tradeflows,andsoforth),animportantlimitationsincecitiesareinterconnectednodeswithinanetworkofflows,aswellascomponentsofasystemofcentral(urban)places.
Whenmodelsareweaklyinformedbytheory,anadvantageofaULUCmodelisitsflexibilityinallowingtheusertomakedecisionsonmodelspecification(althoughthedangerofmodelselectioniseverpresent—thisisaddressedinthenextsection).Currentdesignofmechanicalrule-basedmodelsshowssomeinflexibilitytoalterna-tivespecifications,witharesulting“onesizefitsall”/“cookie-cutter”feel.Finally,animportantconsiderationisthelimitationsinspatialrepresentationsofalternativescenariosimposedbytheULCM,assumingthatquantifiableinformationonpoten-tial alternative directions in local, regional, or national policies can be provided.Policiessuchaszoningorgrowthcontrolsaretheeasiesttorepresent,whileopen-nesstoin-migrationorothereconomicinformationsuchasmarketconditionsmaynot be easily incorporated into models. Highly stylized (input-restricted) modelsareonlyabletoincorporatepoliciesreflectingroaddevelopmentand“off-limitstodevelopment”zoning;thisisalimitingfactorinthecapacityofthemodelstoincludeotherformsofpolicymaking.
�Pre-1981literatureonthepoliticsofmodeluseindecisionandpolicymakingisreviewedinBriassoulis10(1999,chap.5).
42963_C008.indd 145 11/6/07 7:13:09 AM
146 Land Use Change
8.2.4 PoliCy evalUaTion and UnCerTainTies
Model and expert knowledge utilization targets the reduction in the uncertaintyofoutcomepredictions and the consequent effectsof theseoutcomes.Statisticaldecision theory provides quantitative tools for the reduction of uncertainty inoptimalpolicymaking.17Giventhatanyurbanpolicysimulationresultsaredepen-dent on models, how is the “best” model defined and how is it chosen amongall possible models? Amodel is a single representation of reality, and, althoughstatisticalcriteriacanbeusedtoidentifythe“best”one,itrepresentsjustoneofmanypossibledatageneratingprocesses;thus,modelselectionshouldbeavoided.Modelselectionhasbeencriticizedasbeingaweakbasisforpolicyevaluationandderivationoffutureprescriptions;thesearchforasinglebestpredictivemodelismisguidedwhenitcomestopolicy-relevantmodels.Robustnessacrossmodels,ontheotherhand,isbeingadvancedforpolicy-relevantwork.Unfortunately,modelselectionignoresthefundamentaldimensionof“modeluncertainty,”butmethod-ologiesforrobustnessofthepolicyprescriptionacrossalternativemodelspecifica-tionscanbealternativelyutilized.
Methodologiesaddressingtheissuesoftheoryandmodeluncertaintyarenowavailableforincorporationinpolicy-relevantresearch.17,18Uncertaintyovercompet-ing theories results fromuncertaintyaboutwhich theoryofurbangrowth shouldbeutilizedduetoinstitutionalandculturalfactorsaffectinglandmarketsindevel-opingcountriesordifferingassumptions regardingagentdecisionmaking; it canleadtomodelsthatarenotwellinformedbytheory.Modeluncertaintyresultsfromuncertaintyover functional formspecification for statisticalmodelsand isdue tosubjective perceived relevance and endogeneity issues as well as the question ofappropriatespatialandtimelags,proxyvariables,andsoforth.Incompleteknowl-edgeregardingthebestmodelofasystemshouldforcetheresearchertoexplorethesensitivityofmodelingapproaches toalternative specifications.18This frameworkpartiallysolvestheproblemsofsubjectivityandadhocspecificationsinuncertainenvironmentsregardingthecapacityofavarietyofproxiestocapturetheeffectofvariablesenteringthedatageneratingprocess.
An applied statistical framework of policy-relevant urban growth modelingthataccounts formodeluncertaintymakesanexplicit reference toapolicymaker(PMhereafter)whoexaminesasetofurban–growth-relatedpoliciesPforadmin-istrativeunits(e.g.,sub-citydistricts,cities,counties,provinces)throughtheselec-tion of a single policy p.17,18 The PM utilizes data d about a metropolitan area’sland-use and transportation systems (realizations of a process), and the choice isconditionalonamodelmoftheurbaneconomy(mcanconstitutealternativetheoriesandstatisticalspecifications).ThePMminimizestheexpectedvalueofanobjective(loss)functionl(p,θ)whereθisatheexogenousstateofnature(notcontrolledbythePMbutaffectingtheinfluenceofpontheloss,l)�.
�ConsiderX,avectoroftargetedurbangrowthratesperadministrativeunitpertimeperioddefinedbythePM.Differentpolicieswilldrivedifferentratesandpatternsofurbangrowth.Thedeviationofthesegrowthratesfromthetargetedgrowthratescanbeexpressedasalossfunction.Eachadminis-trativeunitisweightedaccordingtothepreferencesofthePM.Theweightsareassignedaccordingtotheimportanceofconvergencetothetargetforeachadministrativeunit:thegreatertheimportanceofmeetingthetarget,thehighertheweight18.
42963_C008.indd 146 11/6/07 7:13:10 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 147
Unknownexogenousfactorsthatinfluencelandusedecisions(thestateofnaturein a decision-theoretic framework) such as monetary or fiscal macroeconomicpoliciesleadtotheminimizationoftheexpectedvalueofthelossfunctionbythePM.Usuallyprobabilitiesofthestatesofnatureareconditionedonexistingdataandselectedmodels:μ(θ|d,m).Accounting formodeluncertainty, theyarenot condi-tionedonmsincethePMunderstandsthatthereisprobablyno“best”modelfortheurbanlandusesystem.Theprobabilitydensityfunction(pdf)forθ,μ(θ|d),isassumedconditionalontheexistingdatadonly(andnotonm).Withawell-definedlossfunc-tionandpdf, theoptimalpolicyis theonethatminimizes theexpectedloss—thesolutiontotheoptimizationproblem.WethusarguethataULCMshouldgenerateaprobabilitydistributionofestimatesandpredictionsaswellastheirdistributioncharacteristics/properties for eachpixel for each timeof change.AmathematicalformulationofthemodelcanbefoundinFragkiasandSeto18andBrocketal.17Thefollowingsectiondescribestheworkingsofanurbangrowthmodelthattakesintoaccounttheaboveconsiderations.
8.3 a new approach in Modeling urban growth in data sparse environments
8.3.1 MeChaniCs of The Model
Wepresenthereanapproachinlandusechangemodelingthatcanexplicitlyevaluatepoliciesunderuncertaintieswithinaspatialsocioeconomicenvironmentbyincorpo-ratingamethodologythataddressesissuesoftheoryandspecificationuncertainty.ThemodelproposedbyFragkiasandSeto18 isahybridspatiallyexplicitmodelofurbanlandusechangewithafoundationoneconomicandstatisticaldiscretechoicemodelsoflandusechange,8adjustedforuseindata-sparseenvironments.
Themodelfirstreadstheinputdataavailableprovidedthroughremotesensinganalysisandothersourcesatadefinedspatialresolution.Initscurrentimplementationtheseare:urban/non-urbanland-usemaps,atransportationnetwork,areasexcludedfromdevelopment,andacentralbusinessdistrictlocation.Itprocessesthisdatapro-ducing new images/matrices such as new urban growth between examined years(abinaryimagefromwhichweextractinformationonadependentvariable)andacollectionofnewimagesfromwhichthemodelextractsindependentvariables.�
Next, themodelemploysstatisticalanalysisfor thecalibrationstage.Separat-ing the study area into two parts (its East and West half), it performs a randomsamplingofdevelopablepixelsoftheinitialEasthalfoftheurban/non-urbanimage(attimet0).Itcreatesacalibrationdatasetwithasingledependentvariableandmulti-ple setsof independent variables ready for regression analysis.Using two (t0, t1)calibrationimages,themodelrunsmultiplelogisticregressionswithbinarydepen-dentvariableyas‘changetourbanornochange’betweentimeperiodst0andt1.The
�Inthismodelwefocusonpredictionofchangesanddonotutilizedataonsocioeconomicprocessesthatresultinlandusepatterns.Inanefforttodevelopamodelwithminimaldatarequirementsweusespatialdensityanddistancevariablestocheckthepredictiveaccuracyofamodel.Themodelalsoutilizesdistrictdummyvariables,eachrepresentingone(oracollection)ofthedistrictsofeachurbanareainthestudy.FragkiasandSeto18presentsamoredetaileddescriptionofthemodel.
42963_C008.indd 147 11/6/07 7:13:10 AM
148 Land Use Change
modelingapproachpresentedinthiscasestudysystematicallyincorporatesavarietyofmodels(orspecifications)andaccountsformodeluncertaintyinlandusechangerelatedpolicymaking�.Themultiple specificationmodel runs reflect theneedsoftheemployedmodelaveraging technique.Fornexplanatoryvariables thatcanbeselectedforthemodels,atotalof2nsetsofalternativespecificationexistandareutilizedintheanalysisasalternativeregressorsets.
Pseudo-Bayesian model averaging is then performed using the calibrationsample.Each2nlogitmodelrungeneratespredictedprobabilitiesofchange(fittedvaluesofthedependentvariable)foreachsamplepoint,andaweightedaverageofthepredictedprobabilitiesiscalculated;the2nsetsoffittedvaluesareweightedbytheirrespective(normalized)pseudo-R2statistic�.Aseriesofbinarysamplesetsofpredictedurban/non-urban land are then createdutilizing an arrayofprobabilitycut-offpoints (thresholdvalues) that range from0 to1.Themodel compares theseriesofpredictedurban/non-urbanvalueswith theactualrealizationof landuseduringthetimeperiodunderstudyandselectsthe“optimal”thresholdlevelforthecalibrationperiod(thecut-offpointthatgeneratestheminimumdifferencebetweenpredictedurbanlandandactualurbanland).�
Modelvalidationoccursattwospatialscales:theindividualpixel(throughPCPvalidation)andachosenadministrativeunitlevel(byaggregatingpixellevelinfor-mationandvalidationthroughsampleenumeration).Thevalidationsampleisderivedthroughasecondrandomspatialsamplewithinthesecond(West)halfofthestudyarea.All2nsetsofindependentvariablevaluesareextractedforthenewvalidationsample.Togetherwiththeestimatedsetsofvariablecoefficientsfromthecalibrationstagetheyareappliedtothefittedprobabilitylogitformulaforeachmodel.Thisgen-eratespredictedprobabilitiesofchangeforthesampleddevelopablepixelsfortimeperiodt1utilizingtheaverageofthefitted/predictedprobabilitiesofallthemodels(weightedbythenormalizedpseudo-R2scorethateachmodelachieves).
Thefirsttypeofvalidationoccursthroughthegoodness-of-fitmeasureof“percentcorrectlypredicted”(PCP)�.Typically,inPCPvalidation,choice(orprediction)isdefinedasthealternativewiththehighestpredictedprobability.TheseclassificationsarethencomparedwithactualchangesandthePCPmeasureiscalculated.Apartfromtheintuitivebinarycut-offvalueof0.5,anyprobabilitythresholdvaluecanbesetforthegenerationofbinarypredictedchangevalues.Weautomatetheselectionofthisthresholdinthecalibrationstageutilizingthecriterionof“bestgrowthrate
�Probabilisticmodelsaresensitivetotheproblemsofpredictivebiasandlackofcalibration:predictivebiasisaproblemofbalanceor“the systematic tendency to predict on the low side or the high side”(p.391)19,andaveragingmodelswithalternativespecificationsincreasesthechancesofaveragingouttheproblem;lackofcalibration,is“a systematic tendency to over- or understate predictive accuracy”(p.391)19andisalsoanegativefactorforvalidationthroughthresholdingduetoanincreasedsensitivitytoit.
�Thestandarddeviationsofthepredictedprobabilityofchangeestimatesarealsocalculated(butonlyatthestageofthefullimageapplication).
�Thisisaformofanexternalimpositionofanurbangrowthratescenarioonthemodel.Athresholdcanalsobeselectedinsuchawaythatanalternativeurbangrowthscenarioisportrayed.
�PCPvalidationoccursintheformofseparationbyspacewithinanout-of-samplemodelingframework.20
42963_C008.indd 148 11/6/07 7:13:11 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 149
matching.”�Thisprobabilitythresholdvalueisappliedinthenewvalidationsample,theclassificationisperformed,andthePCPmeasureiscalculated.
Thesecondtypeofvalidationoccursatalargerscalethantheindividualpixelthroughsampleenumeration.Thetechniqueofsampleenumerationsumspredictedprobabilities over a set of agents or observationswith thegoal of generatingcon-sistentestimationsofaggregateoutcomes.21Themodelutilizesaspatiallyexplicitversionofthisaggregationmethodforthepurposesofvalidationandforecasting,summingupprobabilitiestothedistrictlevel.Itemployssampleenumerationforthevalidationdatasetandcomparestheaggregateestimateofurbanchangetotheactualaggregatechangeinthesample.Predictiveaccuracyisdefinedbythesuccessofthesummedaveragedpredictedprobabilitiesinaccuratelycapturingaggregatechangeattheselectedadministrativeunit�.
Themodelpredictsland-useintwoways.First,wethresholdthepredictedprob-abilities.Predictionresultsarepresentedinitiallyforthepopulationofdevelopablepixelsint2andpotentiallyforanydiscretenumberoffutureiterations(t3,t4,andsoforth).Eachiterationaccountsforthepassingofasingle—equivalenttothelogitmodels—timeperiod.Thepredictionsutilizetheestimatedsetsofregressioncoeffi-cientsand(potentiallyforeachiteration)apartiallynewsetofindependentvariables(reflectinglandscapeevolution).Theoptimalcalibrationthresholdisappliedagainforthe“translation”fromprobabilitytopredictedchange.Second,predictionoccursthroughthesampleenumerationtechniqueforforecastingataggregateadministra-tiveunitlevelsutilizingahypotheticalscenariodataset.
Amultiplescenarioexaminationisusuallyanintegralpartofapolicydecision-makingprocess.Giventhenatureofthecurrentlyincorporatedvariables,asimula-tionmodulecanbeusedfortheexaminationofpolicy-relevantalternativescenariosregarding simple policy leverages: first, the researcher can define new areas ofundevelopedlandthatareexcludedfromdevelopment;second,alteredtransporta-tionroutescanbedesignedaccordingtoexistingplansofroadorrailwayexpansion;third,theusercancreatepatchesofnewdevelopedlandofhighintentionality(e.g.,anewairport)capturingspill-overeffectsofsuchdevelopmentsthatwouldotherwisebedifficulttopredict.Theuser/decisionmakercanfeedthemodelacollectionofscenarioimagesanddefinealossfunctionthatconnectspredictionsofurbangrowthwith the objective the decision maker is trying to achieve (e.g., minimization ofagriculturallandloss).Themodelcanprovidepredictionsofchangeandassociated
�Thisautomatedcalibrationprocessthoughhasadisadvantage.Sinceitisbasedonanadhoccriterionselectedbytheresearcher,theselectionofaprobabilitythresholdover0.5issubjective.
�TheobviousvalueofPCPvalidationistheeaseofvalidationatthepixellevelandanyaggregatedlevelofgroupsofpixels(forexample,withinadministrativeboundaries).Unfortunately,theapplica-tionofprobabilitythresholdsinstatisticalmodelsiscountertothenotionofapredictedprobabilityinsuchmodels.Limited informationdue to theunobservablecomponent in the formulationof thechoiceprocessforbidsthepredictionofthechoiceofanalternativeforasingleunitofobservation.Thenatureofthepredictedprobabilitiesindiscretechoicemodelshasastandardstatistical“largesamplerepetition”interpretation:thealternativewiththehighestprobabilityisnottheunit’schoiceeachtime.Thisinterpretationmakesamodeldesignthatattemptsacrossfromprobabilityspacetochoicespaceamorecomplexanddifficulttask.21Thus,PCPmeasuresmaynotprovidethebestwaytovalidateastatisticalmodel.
42963_C008.indd 149 11/6/07 7:13:12 AM
150 Land Use Change
lossesforeachscenario,thusgivingthedecisionmakerthechoiceofthepolicythatminimizesthelossaccountingformodeluncertainty.
8.3.2 aPPliCaTion To Three CiTies of The Pearl river delTa, China
While urban land use change is a worldwide phenomenon, it is most dynamic inAsia—andparticularlyinChina—whereunprecedentedratesofurbangrowthhaveoccurredoverthelasttwoandahalfdecades.1Theinteractionofcompellinglocal,regional,andglobalfactorshasresultedinremarkablyvariedurbanconfigurations.22OurstudyareaiscomprisedofthreeofthemostdevelopedcitiesinthePearlRiverdelta(PRD)regioninthecoastalsoutheastChina,Shenzhen,Guangzhou,andFoshan(Figures8.2and8.3);theseandothercitiesinthePRDhaveexperienceddramaticurbanlandgrowthratesinthepasttwodecades.Duringan11-yearperiod—from1988to1999—urbanlandgrew�inthePRDby451.6%orapproximately16.5%ayear23.TheDeltageneratesmorethan70%oftheprovincialGDPandishometo21millionpeople, nearly one-third of the province’s official population.23 Previous researchshowsthattheregionisundergoingrapidurbantransformation.24Althoughthecitiesareincloseproximity,theydifferinhistory,demographics,andeconomics.18,25,26
Themodel is runfor the threecitiesat twopixel resolutions,30mand60m.Usinga randomcalibrationsample fromthedevelopablepixelsof the initial1988urban/nonurbanimage�themodelexploresallpotentialcombinationsoflogitmodelscreatedbyallowingfivevariablestoenteralternativespecifications,whichresultsin32models.�Afterderivingallthesetsofestimatedcoefficientsandcalculatingthemodel-averagedpredictedprobabilities,thecalibrationstageidentifiesthethresholdsthatbestfittheobservedurbangrowthratefortheperiod1988to1996forallcities.
Validation occurs with the use of the second random sample and the model-averaged predicted probabilities for all cities and specified resolutions. In a caseoftwodiscretestatesoflanduse(urbanandnonurban),onefindstwocaseswherepredictions are accurate (correctly predicted urban and nonurban) and two caseswherepredictionsarewrong(wronglypredictedurbanandnonurban).InthecaseofPCPvalidationthetotalpercentageofwronglypredictedpixelsrangesbetween23%27%.Themodelconsistentlygeneratesadisproportionatelylargerpercentageofwronglypredictednonurbanpixelsrelativetothepercentageofwronglypredictedurbanpixels,amanifestationofthepredictivebiasproblem:weobserveasystematictendencyofsmallerprobabilityvalues.Themodelalsoreportscomparisonsbetweentheaggregate—atthedistrictlevel—actualchange,theaggregatepredictedchangethroughsampleenumeration,and theaggregatepredictedchange through thresh-olding for all cities and resolutions.The sample enumeration techniqueperformsconsistentlybetter in thevalidationof theaggregatecountsofchange.Acrossallcities, we observe that the vector distances between the aggregate actual changewithaggregatesampleenumerationpredictedchangeandtheaggregatethresholding
�In particular, the metropolitan areas of cities such as Shenzhen, Guangzhou, and Foshan grew by132.3%(8%annually),247.6%(12%annually),and140%(8.3%annually),respectively
�Forthe30mresolution,thisnumberiscloseto14,000pixelsintheShenzhenstudyarea,3,300pixelsintheFoshanstudyarea,and10,200pixelsintheGuangzhoustudyarea(Figure8.3).
�Excludingdistrictdummyvariableswouldprovideanadditional32models.
42963_C008.indd 150 11/6/07 7:13:13 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 151
predictedchangedropsignificantlyasweloweredtheresolutionfrom30mto60m.The performance of aggregate prediction through thresholding improves relativetoaggregatepredictionthroughsampleenumerationbutisstillnotsatisfactory.Inshort,thereexistsaclearfailureofspatiallyaccuratepredictionwhenusingthresholdprobabilitiesfortransitiontochoices.Thepixellevelpredictionscanbesuccessfully
Kilometers6030150
KilometersFoshan study areaGuangzhou study areaShenzhen study areaGraticule (5 degrees)Province BoundariesGuangdong Province
25012562.50
Urban land use Urban land 1988 Growth 1988–1996
20N
25N11
0E
115E
SOUTH CHINA SEA
GUANGXI PROVINCEGUANGDONG PROVINCE
HUNAN PROVINCE JIANGXI PROVINCEFUJIAN PROVINCE
PEARL RIVER DELTA
N
FIgure 8.2 (See color insert following p. 132.) The Pearl River Delta in southeastChinaandtheShenzhen,Foshan,andGuangzhoustudyareas(urbanlandin1988,newurbanlandbetween1988and1996).
42963_C008.indd 151 11/6/07 7:13:15 AM
152 Land Use Change
usedthroughaggregationtoalargeradministrativeunitsuchastheneighborhoodorthetownshipthroughtheprocessofsampleenumeration.
Atthepredictionstage,themodelaveragingprocessthatgeneratesthepredictedprobabilityvaluesisrepeatedforthefullpopulationofobservations;thet1period(1996)dataprovide thevaluesof independentvariables,anda full imageofpre-dictedprobabilitiesofdevelopmentisgenerated.Eachpixelisassignedtheaveragedpredicted value from the logit formula using the sets of estimated coefficients.Figures8.4,8.5,and8.6mapthesepredictedprobabilitiesofdevelopmentbetween2004and20012andtheassociatedstandarddeviationsofthepredictions.
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
100
200
300
400
500
600
100
200
300
400
500
600
0.12
0.1
0.08
0.06
0.04
0.02
0
100 200 300 400 500 600 700 800 900
100 200 300 400 500 600 700 800 900
FIgure 8.3 (See color insert following p. 132.) Predicted probability of change tourbanareasbetween2004and2012andstandarddeviationofpixelpredictedprobabilitiesforShenzhen(60mresolution;valuesinpercentagepoints).
42963_C008.indd 152 11/6/07 7:13:16 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 153
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
50
100
150
200
250
300
350
400
450
500
550
50
100
150
200
250
300
350
400
450
500
550
100 200 300 400 500 600
100 200 300 400 500 600
FIgure 8.4 (See color insert following p. 132.) Predicted probability of change tourbanareasbetween2004and2012andstandarddeviationofpixelpredictedprobabilitiesforFoshan(30mresolution;valuesinpercentagepoints).
42963_C008.indd 153 11/6/07 7:13:18 AM
154 Land Use Change
For predictions based on thresholding, the model generates images of urban/nonurbanlanduseforanyfuturetimeperiod;Figure8.7plotstheimagefor2012.Amountsofurban landusechange foreachdistrictbetweenanysetofyearsarepredictedthroughthetechniqueofsampleenumeration.
8.4 discussion and Conclusions
Accountingformodeluncertaintyinlandusechangerelatedpolicymakingthroughtheemployedmethodologyreducesuncertaintiesforadecisionmakerthatareinher-entintheuseofmodelsinthedecisionmakingprocess.Theproposedmethodology
0.6
0.5
0.4
0.3
0.2
0.1
0
0.12
0.1
0.08
0.06
0.04
0.02
0
50
100
150
200
250
300
350
400
50
100
150
200
250
300
350
400
100 200 300 400 500 600 700
100 200 300 400 500 600 700
FIgure 8.5 (See color insert following p. 132.) Predicted probability of change tourbanareasbetween2004and2012andstandarddeviationofpixelpredictedprobabilitiesforGuangzhou(60mresolution;valuesinpercentagepoints).
42963_C008.indd 154 11/6/07 7:13:19 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 155
9008007006005004003002001000
7006005004003002001000
6005004003002001000
0
100
200
300
400
500
600
0
100
200
300
400
500
0
50
100
150
200
250
300
350
400
FIgure 8.6 Shenzhen,Guangzhou,andFoshanurban/nonurbanpredictionfor2012.
42963_C008.indd 155 11/6/07 7:13:21 AM
156 Land Use Change
FIg
ur
e 8.
7 C
ity
ofS
henz
hen
–M
arch
200
2.
42963_C008.indd 156 11/6/07 7:13:23 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 157
potentiallyreducestheproblemsofpredictivebiasandlackofcalibration.Throughtheabovemethodology,aPMisabletocalculateavarietyofcharacteristicsofprob-abilitydistributionsofoutcomes (suchas themeanandvariance) thatmayaffectpolicy-makingchoices.Furthermore,sincethemethodologyenhancesthecapacityfor convergence of the true value of underlying parameters and the estimates ofthoseparameters,aresearchermayworrylessaboutpossiblestatisticalmodelbiasesandcanfeelmoresecure in thedecreasedcapacityofmanipulationof themodelby policy stakeholders toward particular answers. Unfortunately, though, movingfromthissimplesinglepolicyleveragemodeltomultiplesimultaneousleveragelevelchoicesbyPMsincreasessignificantlythecomplexityofthemodel.Furthermore,thisrationaldecision-makingframeworkisalsofarfromaperfectdepictionoftheactualpolicy-makingprocess;itignoresissuessuchastheconflictsoverideology,powerdifferentials,problemsincommunicationandcollaboration,andtheinfluenceofbureaucraciesandspecialinterestgroups.
Theconceptofuncertaintyreductionthroughmechanism/modeldesign—andtherelevanceofthismethodology—issupportedbymodernpoliticaleconomy.Sincearealisticviewofgovernmentsupportsthatpoliticians,bureaucrats,andpolicymakerswithin national and local governments are a mix of actors with public-good andprivate-gain motivation, policies should be designed with the assumption that allpolicymakersoperateunderthehomoeconomicusmodel(thatis,personalgainmoti-vationdrivesactionandisahugefactorinanydecision-makingprocess27).Giventhatmodelsandmodelselectioncanbeemployedforthesupportofprivateinterests,amodelingapproachthatprovidesadefenseagainstmodelselectionisdesirable.
Themodelinputaccuracyissue(“junkin—junkout”)naturallyaffectsland-usechangemodelstoo.Socioeconomicdatacanbenonexistent,incomplete,inaccurate,unreliable,oralltheaboveinadevelopingworldsetting.Evendataaccuraciesofafundamentalvariablesuchasurbanizationanditsforecastsareseverelycriticized.28Resultsofmodelingapproachesthataresensitivetoomittedornonexistentinforma-tionarepotentiallymisguidingforpolicymaking.Wefindthatenrichingdatasetswithremotelysensedinformation,incombinationwithmodelsofastatisticalnature,isapositivestepinanymodelingexerciseregardingdevelopingworldcities.Onthe one hand, given data inaccuracies in socioeconomic data, statistical modelsprove tobe lesssensitive to inputdata imperfections,since theyassignerrors inmeasurementinthestochasticpartof themodelandproduceunbiasedestimateseven when errors plague the data (if those errors are not systematic—randomlydistributed).Ontheotherhand,theuncertaintyovertheaccuracyofinputscanbecontrolled/measuredmoreeasilyfordataderivedfromsatelliteimageryratherthatsocioeconomiccensussources(assumingtheminimizationofuncertaintyandnobiasinsatelliteimageryclassification).
The application of probability thresholds in statistical models is problematicdue to the notion of a predicted probability in probabilistic models; pixel-levelvalidationtechniquesmaybemisguidingwhenusedinstatisticalmodels.Thestan-dardstatistical“largesamplerepetition”interpretationofprobability(whichappliestoourstatisticalmodel)conflictswiththeideaofPCPmeasuresofgoodness-of-fitdue to thisfundamentalcharacteristicofprobabilisticdiscretechoicemodels.An
42963_C008.indd 157 11/6/07 7:13:23 AM
158 Land Use Change
improvementtostandardthresholdingtechniques(whichalsoautomaticallydiscardinformationprovidedbytheneighboringpixels’predictedprobabilities)shouldbeofferedby the research community and is a researchgoal for the authorsof thischapter.Wesuggestthatuntilamoreadvancedvalidationmechanismisproposed,statisticalmodelsshouldbeeffectively limited tomainly thenonspatiallyexplicitsampleenumerationvalidationandprediction.Theseestimatescouldpotentiallybecoupledwithamoreadequatemoduleoperatingasanpixelallocationmechanismofpredictedgrowth.
Theaforementionedproblemofpredictivebiasmanifests itself in themodelthrough an imbalance in the percentages of wrongly predicted nonurban/urbanpixels. The model presented here partially corrects for this limitation throughpseudo-Bayesianmodelaveraging.Still,wefindthatasignificantlylimitingchar-acteristicofpolicy-relevant landusechangemodelsis that theydonotallowtheuser to define which type of error is more important.� Future versions of urbangrowthmodelsshouldincorporateamechanismthatwouldallowaPMtocontroltheamountofdifferenttypesoferrors(optimally,bypolicyregion),whentherela-tivecostsofwrongpredictionsbymodelsaresignificantlydifferent,andthusattainhigherpolicyrelevance.
Weidentifyseveralkeypointsfromtheaboveanalysis.First,theuserofproba-bilisticchoicemodelsshouldprimarilyfocushisorherattentiononthepredictedprobabilitymapsso thatprobablehotspots fordevelopmentare identifiedratherthan generating predicted maps of choices through simple thresholding mecha-nisms(changeornotchangeinabinarysetting).Second,theadvantageofutiliza-tionofavarietyofmodelsismanifestedinthefactthatmapsofstandarddeviationsfor the predicted probability maps convey important information regarding thespatiallyexplicitagreementofthevariousmodels.Third,inlightofscarceinfor-mationanddifficultyofdatasetenrichment,predictedprobabilitymapsshouldbecoupledwithadditionalinformationaboutdevelopmenttrendsthatmaybederivedfromsurveyswithimportantstakeholdersinlandusechangedecisionmakingandlocalknowledge.
Generally,similarlytoothermodelsinitsclass,thesuccessofthemodeliscon-strainedbytheavailabilityofquantifiableproxies.Similartoothermodelsofthistype,scenariobuildingrequirestheuseofvariablesforwhichfuturevaluescanbeprovided.Futureimplementationsofthemodelwillconsider(i)validationthroughseparationbytime;(ii)themoreimmediateincorporationofcalibrationschemestodifferentlocalities/districtsthatcalibratethemodelandcaptureinmoredetailtheurbangrowthrates(thequantityofchange)atthedistrictlevel;(iii)thecouplingofthismodelwithothermodelsoperatingatotheraggregationlevels,sheddingmorelightintotherealisticprospectsofdevelopmentofanymodel-identifiedhotspot;and(iv)thecollectionandanalysisofscenariosandpolicymakerpreferencesincollabo-rationwithChinesecityplanningdepartments.
�Instatistics,forexample,whentestingahypothesisaboutapopulation,theresearchercancontroltheprobabilityoftwodistincttypesoferrors,typeIand.typeIIerrors,basedontheseverityofapossiblemistake.Thedilemmaofwhethertoadoptahighorlowsignificancelevelisstrongerwhenthestakesarehigher.
42963_C008.indd 158 11/6/07 7:13:24 AM
Urban Land-Use Change, Models, Uncertainty, and Policymaking 159
reFerenCes
1. UnitedNations.WorldUrbanizationProspects:The2003Revision.UNPress,NewYork,2004.
2. UNCHS.TheStateoftheWorld’sCitiesReport2001.UnitedNationsCentreforHumanSettlements(Habitat),NewYork,2002.
3. Lambin,E.F.etal.Thecausesofland-useandland-coverchange:Movingbeyondthemyths.GlobalEnvironmentalChange11,261,2001.
4. National Research Council. Human Dimensions of Global Environmental Change:Global Environmental Change: Research Pathways for the Next Decade. NationalAcademyPress,Washington,D.C.,293pp,1999.
5. Boulanger, P.-M. and Bréchet, T. Models for policy-making in sustainable devel-opment:Thestateoftheartandperspectivesforresearch.EcologicalEconomics.55,337,2005.
6. von Thünen, J. H. Der isolierte staat in beziehung auf landwirtschaft und national-ökonomie. Gustav Fisher, Stuttgart; translation by C. M. Wartenburg (1966), TheIsolatedState.OxfordUniversityPress,Oxford,U.K.,1826.
7. Brown,D.G.,etal.Modelinglanduseandlandcoverchange.In:Gutman,G.etal.,eds.,LandChangeScience:Observing,Monitoring,andUnderstandingTrajectoriesofChangeontheEarth’sSurface.KluwerAcademicPublishers,Dordrecht,Netherlands,395,2004.
8. Irwin,E.G.,andGeoghegan,J.Theory,data,methods:Developingspatiallyexpliciteconomicmodelsoflandusechange.AgricultureEcosystemsandEnvironment85,7,2001.
9. EPA.ProjectingLand-UseChange:ASummaryofModelsforAssessingtheEffectsofCommunityGrowthandChangeonLand-UsePatterns.U.S.EnvironmentalProtectionAgency,OfficeofResearchandDevelopment,Cincinnati,Ohio,2000.
10. Briassoulis, H. Analysis of land use change: Theoretical and modeling approaches.In:LoveridgeS.,ed.,The Web Book of Regional Science.RegionalResearchInstitute,WestVirginiaUniversity,Morgantown,W.Va.,2000(Availableat:http://www.rri.wvu.edu/regscweb.htm).
11. Agarwal,C.etal.A Review and Assessment of Land-Use Change Models: Dynamics of Space, Time, and Human Choice.U.S.DepartmentofAgricultureForestService,NortheasternForestResearchStation,UFSTechnicalReportNE-297.Burlington,Vt.,2002.
12. Brown,D.G.etal.Pathdependenceandthevalidationofagent-basedspatialmodelsoflanduse.International Journal of Geographic Information Systems19,153,2005.
13. Parker,D.C.etal.Multi-agentsystemsforthesimulationofland-useandland-coverchange:Areview.Annals of the Association of American Geographers93,314,2003.
14. Klosterman, R. E., and Pettit, C. J. Guest editorial. Environment and Planning B: Planning and Design32,477,2005.
15. Szanton,P.Not Well Advised.RussellFoundationandtheFordFoundation,NewYork,1981.
16. Batty, M. Editorial. Environment and Planning B: Planning and Design 31, 326,2004.
17. Brock,W.A.,Durlauf,S.N.,andWest,K.D..Policyevaluationinuncertaineconomicenvironments.Brookings Papers on Economic Activity1,235,2003.
18. Fragkias,M., andSeto,K.C.Modelingurbangrowth indata sparseenvironments:Anewapproach.Environment and Planning B: Planning and Design34,858,2007.
19. Hoeting,J.A.etal.Bayesianmodelaveraging:Atutorial.Statistical Science14,382,1999.
42963_C008.indd 159 11/6/07 7:13:25 AM
160 Land Use Change
20. Pontius, R. G., Huffaker, D., and Denman, K. Useful techniques of validation forspatiallyexplicitland-changemodels.Ecological Modelling179,445,2004.
21. Train,K.Discrete Choice Methods with Simulation.CambridgeUniversityPress,NewYork,2003.
22. Webster, D. R. On the edge: Shaping the future of peri-urban East Asia. StanfordUniversityAsia/PacificResearchCenterUrbanDynamicsDiscussionPaper,2002.
23. Seto,K.C.,andKaufmann,R.K.ModelingthedriversofurbanlandusechangeinthePearlRiverDelta,China:Integratingremotesensingwithsocioeconomicdata.Land Economics79,106,2003.
24. Seto,K.C.etal.Monitoringland-usechangeinthePearlRiverDeltausingLandsatTM.International Journal of Remote Sensing23,1985,2002.
25. Seto,K.C.,andFragkias,M.Quantifyingspatiotemporalpatternsofurbanland-usechangeinfourcitiesofChinawithtimeserieslandscapemetrics.Landscape Ecology,20,871,2005.
26. Seto,K.C.UrbangrowthinSouthChina:WinnersandlosersofChina’spolicyreforms.Petermanns Geographische Mitteilungen148,50-57,2004.
27. Brennan,G.,andBuchanan,J.M.The Reason of Rules: Constitutional Political Economy.CambridgeUniversityPress,Cambridge,U.K.,1985.
28. Cohen, B. Urban growth in developing countries: A review of current trends and acautionregardingexistingforecasts.World Development32,23,2004.
42963_C008.indd 160 11/6/07 7:13:25 AM
Part III
Synthesis and Prospect
42963_S003.indd 161 8/17/07 2:29:27 PM
42963_S003.indd 162 8/17/07 2:29:28 PM
163
9 Synthesis, Comparative Analysis, and Prospect
Michael J. Hill and Richard J. Aspinall
Contents
9.1 Introduction................................................................................................. 1639.2 ASummationoftheChapters..................................................................... 1649.3 OverallMessages........................................................................................ 169
9.3.1 RealizingtheFullPotentialfromRemoteSensing......................... 1699.3.2 ApplicationofIntegratedMethods.................................................. 1719.3.3 PlacingAnalysisinContext............................................................. 173
9.4 Conclusions................................................................................................. 174References.............................................................................................................. 176
9.1 IntRoDUCtIon
This book has examined the issue of integrated analysis of spatial structure andspatiotemporal processes related to landuse change in terrestrial coupledhumanenvironmentsystems.TheconsequencesoflandusechangehavebeentotransformalargeproportionofthelandsurfaceoftheEarth,andthesechangesareinfluenc-ing the global carbon cycle, regional climate, water quality and distribution, andbiodiversitythroughhabitatloss.1Thebookpresentsanumberofcasestudiesthatinclude urban, wilderness, wet tropical forest, and arid desert-like environments,humanpopulationdensitiesfromhightoverylow,andthosethatdemonstratebothdirectandindirecthumaninfluences(Table9.1).Thecasestudiesincludeseveralthatfocusonanalysisofclearingoftropicalforestenvironments(Chapters4,5,6,and7;Table9.1).Theseareenvironmentsofhighsignificancetoglobalcarbonstocks,bio-diversity,regionalclimate,andAfrican,Asian,andSouthAmericaneconomies.2,3However,bothintensiveagriculturallands(Chapter3;Table9.1)andextensivegraz-inglands(Chapter2;Table9.1)arealsocovered,whilespecificattentionispaidtothelate20th-centuryphenomenonoftheriseofmegacities(Chapter8;Table9.1).
Thesecasestudiesandother literature1,2,3,4,5 suggest thata trade-offapproachisprobablytheonlypragmaticoptiontomoderatingtheimpactofhumansontheterrestrial system.Thepaceofmodificationof the landsurface showsnosignofslowingwhilehumansmodifytheirapproachesinresponsetochangesindemandandpriceforproductandtoavoiddetectionandcontrol.6Inthischapterwebrieflysummarizethemainpointsfromeachchapterandthenexamineinturnthemain
42963_C009.indd 163 11/6/07 7:11:47 AM
164 Land Use Change: Science, Policy and Management
messagesdescribedaboveandlookatthepotentialfordeliveryofsocietalbenefitfromaholisticunderstandingof,andapproachto,landusechange.
9.2 A sUMMAtIon oF tHe CHAPteRs
Theapproachinthisbookcanbesummarizedintermsoftheproblemcontext,meth-odological approaches, and geographical-system context (Table9.1). The sciencecontextforthethemeofthebookhasbeenoutlinedinChapter1.Fivebasicsciencequestionswereidentified:dynamicsofchangeinspaceandtime;integrationoffeed-backsbetweenlandscape,climate,socioeconomicandecologicalsystems;resilience,
tAble 9.1A summary of the Key Issues, Contexts, and Methods examined in each Chapter of the book
Chapter Problem contextMethodological
approach(es)Geographical-system
context
1. Aspinall Technicaloverviewofdynamics,scale,accuracy,uncertainty,patternandprocess
Models—conceptual,GIS,RS,CA,MAS,simulation,statistical,empirical,visualization,space-timescaling
Internationalresearchframeworks—GLP,bio-complexity,etc.
2. Hill Transformationofspatiotemporaldataandrelationshipstosimpleindexes
Integrationoftime-seriesanalysis,spatialanalysis,numericalandheuristicmethods
Savannaandgrasslandbiomes/livestockasagents
3. Byron/Lesslie Socialsurveysofattitudestonaturalresourcemanagementissues
Assignmentofrelationshipsbetweenlandholderperceptionandopinionandbiophysicalfeaturesormanagementpractices
RuraleasternAustralia
4. Babigumira,Müller,andAngelsen
Tropicalforestclearing Spatiallyexplicitlogisticmodels
Africantropicalforests
5. EtterandMcAlpine
Tropicalforestclearing Regressiontreeandregressionmodelsofdeforestation/regeneration
Amazoniantropicalforest
6. Crews-Meyer Fragmentationofforest Patchpanelmetrics—pixel-patchhistories
Thailandtropicalforestandregrowth
7. MillingtonandBradley
Tropicalforestclearing Spatialimprintofcadastralgrids;differentialbehavioroffragmentationmetricsasforestcoverdeclines
Amazoniantropicalforest
8. FragkiasandSeto
Urbanexpansion/megacities
Multiplelogisticregression,pseudo-Bayesianmodelaveragingtogetpredictedprobabilitiesofchange
CitiesofPearlRiverDelta—Shenshen,Guangzhou,andFoshan,China
42963_C009.indd 164 11/6/07 7:11:48 AM
Synthesis, Comparative Analysis, and Prospect 165
vulnerability,andadaptabilityofcoupledhumanandnaturalsystems;scaleissues;andaccuracyanduncertaintyissues.Approachesincorporatingalloftheseconsider-ationscanbegroupedunderthegeneralheadingof“models,”butthisincludesadiver-sityoftypesfromconceptualthroughtoempirical(Table9.1).Abroadercontextforincorporatingthedifferentelementsofcasestudiesoflandusechangeistoconsidera rangeof integrating frameworks (Figure9.1).These include theanalytical struc-turefortheGlobalLandProject7; theHumanEcosystemModel8; theverygeneralframeworkofexplicitfocusonlinkagesbetweenthedynamicsofhumanandnaturalsystemsoftheU.S.NationalScienceFoundationBiocomplexityintheEnvironmentprogram(http://www.nsf.gov/geo/ere/ereweb/fund-biocomplex.cfm.),and,withaviewto a more design-oriented approach to landscape change, the Landscape DesignResearch Framework.9 All of these frameworks seek interdisciplinary definitionandfocusonkeyquestionswithinthebroadscopeoflandusechangeandlinkstomanagementofchangeincoupledhumanenvironmentsystems.
InChapter2,theroleofmultiplecriteriaandtrade-offanalysisisdiscussedinthecontextofmethodsandapproachesforcaptureandtransformationofcomplexprocesses.Thechapterproposessimpleindex-basedcomparativeframeworksinaninteractiveenvironmentthatassistdecisionmaking,butretainthelegacyandcriti-calinformationcontentneededforcompleteappreciationofissuesassociatedwithland use change. The analysis seeks to balance evidence-based science and soft-systemsapproachesbyintegrationofharddatawithvaluejudgment,publicopinion,andpolicy andmanagement goals.Temporal analysis is important for entraininglegacyandhistorical factors indecisionmaking; spatial analysis is important forappreciationofsocial,economic,andbiophysicalimpactsoutsidetheboundaryofthedirectlyaffectedareaorgeographicallocationofinterest.Theapproachseekstomeasureandaggregatetheperformanceofalternativeoptions.Itrequiresahighlysystematicandtransparentapproachtomanagementofinformation.
Theapplicationofmethodsthataddressthescienceissuesidentifiedabove,andaggregationofdiverseinformationanddatasourcesintoframeworkstoassistdecisionmaking,mustbemediatedbythepolicycontextfortheanalysisandmodeling.IntheIntroductionwesuggestedthatsustainabilityscience,andspecificallythetonguemodelofPotschinandHaines-Young,10offersacontextand“choicespace”forlinkingscienceanddecisionmakinginpolicyandmanagement.Thecasestudiesprovideamixofanalysisacrosssocial,economic,andbiophysicalperspectivesnecessarytodevelopanunderstandingoflandusechangerelevanttosustainability.
InChapter3,asocialsurveyapproachisusedtodevelopunderstandingoftheattitudes and perceptions of rural communities. The work introduces importantmethodological issues surrounding relationships between point survey data andspatiallyexplicitbiophysicalfeaturesofalandscape.Theanalysislooksatrelation-shipsbetweenanswers to surveyquestionsabout landmanagementpracticesandundesirablelandscapefeaturesbasedonadistanceanalysis.Thisimmediatelyintro-duces interestingquestions abouthowhumansperceive their spatial environmentandhowtheirspatialsensitivitiesandawarenessdiffer.Italsorequiressomeatten-tiontothelegacyeffectsofhistoricalexperienceandviewsofthelandscape,rootedinhistoricparadigms,and the impactofaspatialviews,norms,andmedia issues
42963_C009.indd 165 11/6/07 7:11:48 AM
166 Land Use Change: Science, Policy and Management
DecisionMaking
EcosystemsServices
Distu
rbance
How
shou
ld th
e la
ndsc
ape
be d
escr
ibed
?
How
doe
s the
land
scap
e op
erat
e?
Is th
e cu
rren
t lan
dsca
pe w
orki
ng w
ell?
How
mig
ht th
e la
ndsc
ape
be a
ltere
d?
Wha
t pre
dict
able
diff
eren
ces m
ight
the
chan
ges c
ause
s?
How
shou
ld th
e la
ndsc
ape
be ch
ange
d?
Dat
a In
form
atio
n Cu
ltura
l kno
wle
dge
Repr
esen
tatio
n M
odel
s
Proc
ess M
odel
s
Eval
uatio
n M
odel
s
Chan
ge M
odel
s
Impa
ct M
odel
s
Dec
isio
n M
odel
s
Ecol
ogic
alSy
stem
sBi
ogeo
chem
istry
Biod
iversi
tyW
ater
Air
Soil
Soci
alSy
stem
s
T1. D
ynam
ics of
land
–sys
tem
sT2
. Con
sequ
ence
s of l
and–
syste
m ch
ange
T3
. Int
egra
ting a
nalys
is an
d m
odell
ing f
or la
nd su
stain
abili
ty
Land
Use
&M
anag
emen
t
Popu
latio
nSo
cial/E
cono
mic
struc
ture
Polit
ical/I
nstit
utio
nal
regim
esCu
lture
Tech
nolo
gy
Earth
Syste
m
Land
Syste
ms
T2.
1
T2.
2
T2.4
T3.1
Crit
ical
pat
hway
s of c
hang
eT3
.2 V
ulne
rabi
lity a
nd re
silien
ce of
Lan
d–Sy
stem
sT3
.3 E
ffect
ive go
vern
ance
for s
usta
inab
ility
T2.
3
T1.
1
T1.
2
T1.3
Nat
ural
Cultu
ral
Soci
o–ec
onom
ic
Soci
alIn
stitu
tions
Soci
alCy
cles
Soci
alO
rder
Cri
tical
Res
ourc
esH
uman
Soc
ial S
yste
m
Nat
ural
Sys
tem
Hum
an S
yste
m
(a)
(b)
(c)
(d)
FIG
UR
e 9.
1 (S
ee c
olor
inse
rt f
ollo
win
g p.
132
.) A
var
iety
of
inte
grat
ing
fram
ewor
kst
hat
seek
int
erdi
scip
lina
ryd
efini
tion
and
foc
uso
nke
yqu
es-
tion
sw
ithi
nth
ebr
oad
scop
eof
man
agem
ent o
fla
nd u
se c
hang
ein
cou
pled
hum
an e
nvir
onm
ents
yste
ms.
(a)
Ana
lyti
calf
ram
ewor
kfo
rth
eG
loba
l Lan
dP
roje
ct o
f IG
BP
and
IHD
P.(
From
GL
P, 2
005,
wit
hpe
rmis
sion
.)(b
) The
Hum
an E
cosy
stem
Mod
el. (
From
Mac
hlis
eta
l.,8 1
997,
wit
hpe
rmis
sion
.)(c
) The
U
.S. N
atio
nalS
cien
ceF
ound
atio
npr
ogra
mo
nB
ioco
mpl
exit
yin
the
Env
iron
men
t(ht
tp://
ww
w.n
sf.g
ov/g
eo/e
re/e
rew
eb/f
und-
bioc
ompl
exit
y.cf
m);
(d)
The
L
ands
cape
Des
ign
Res
earc
hFr
amew
ork.
(Fr
omS
tein
itz
eta
l.,9
2003
,wit
hpe
rmis
sion
.)
42963_C009.indd 166 11/6/07 7:12:23 AM
Synthesis, Comparative Analysis, and Prospect 167
insociety.Thechapterisimportantbecauseitbeginstoaddressspatialanalysisofopinions,perceptions,andattitudes.
In Chapter 4, an economic analysis centered on the concept of land rent asadriverof landusechange is applied toassessmentofdeforestation inUganda.Aneconometricmodel(binary—logistic)incorporatingexplanatoryvariablesthatdescribesocioeconomic,spatial,andinstitutionalcontextsisusedtoestimatetheprobability of deforestation. The analysis tests a set of hypotheses and seeks todefine reasons for change. This analysis is also interesting for what it does notinclude:socialsurveyofattitudes(suchasdescribedinthepreviouschapter)couldaddimportantmissingmotivationalinformationonagentbehaviors.Theanalysisimplicitlyassumes thatmotivationwill solelybebasedonmaximizationof landrent. This is one of the main issues for analysis of coupled human environmentsystems:iseconomicreturnafullyadequatedriverforlandusechangeandcon-sequentlandcoverchangeinmostcircumstances?Anassessmentofthefirsttwocase studies might conclude that each would benefit if both of their approacheswerecombined(i.e.,bothsocialsurveyandeconometricmodelingwereapplied).However, evidence from the other case studies shows that further improvementmightbeobtainedifmoresophisticatedspatialandtemporalanalysiswasappliedinconjunctionwithsocialsurveyandeconomicmodeling.
Thenextthreechaptersalladdressdeforestationintropicalforests.InChapter5,regressiontreeanalysisandlogisticregressionareappliedtoanalysisofdeforesta-tionandregeneration.Socialandeconomicprocessesareimportant,asislocalandregionalcontext,anddeforestationfollowsatemporallyexplicittrajectorydescribedbyasigmoidalcurve.Patternsare linked toprocessatdifferentscales:national,regional,and local.Althoughdistances to roadsand towns,proximate factors inthe terminology of Geist and Lambin,11,12 are the best predictors at the nationallevel, deforestation and regeneration occur at local hot spots at a regional level.However,atalocallevelmoreexplicitrelationshipsareobtainedwithaccessibilityand soil type, sincedeforestation rates are strongly related to a spatialmetric—forestedgedensity.Thisprovidesanexampleofspatiallyexplicitanalysisderivingametricwithdirectmeaninginrelationtodeforestationpotentialassociatedwithaccessibility.Hence,spatialanalysisandcalculationofpatternmetricscanbeusedtogenerateindicatorsoflikelihoodofdeforestationatscalesinwhichpatternandprocessaredirectlyconnected.
ThisthemeofpatternmetricsasdescriptorsorindicatorsoflandscapeprocessesiscontinuedfurtherinChapter6.Here,changesinpatternmetricsareanalyzedforlandscapepatchesthroughanumberoftimestepsusingremotelysensedimagery.Inparticulartheinterspersion-juxtapositionindexandmeanpatchsizeindexpro-videmeasuresoffragmentation.Theseindicesgivetemporalprofilesfordifferentlandcoverclassessuchasforest,savanna,andriceagriculture.Theanalysisinthechapter combines the definition of landscape change in terms of pattern metrics,forexample,fluctuations through time in the interspersionof landuse/landcoverclasses,withassignmentofmeaningtothechangesinpatternmetrics,forexample,reductioninforestinterspersionasforestpresencedeclineswiththespreadofriceagriculture.Explanationofregionaldifferencesisbasedonuseofcontextualinfor-mation,forexample,proximitytoareasofmilitaryinstabilitydeterringagricultural
42963_C009.indd 167 11/6/07 7:12:23 AM
168 Land Use Change: Science, Policy and Management
encroachment.Thischapterillustratestheapplicationofsophisticatedspatiotemporalanalysis combined with the explanatory contextual information, as discussed inChapters1and2.Thequalityofthisanalysisishighlydependentonthesensitivityandaccuracyofchangedetectionfromremotesensing.
The assessment of forest fragmentation is continued in Chapter 7 where thegoal is to increase understanding of relationships between road construction andforestfragmentationinAmazonia.Here,thefocusisonthebehavioroftheagentsofchangeratherthanonthestructureofthelandcover.Hence,ifChapter6approachesthe issue with a pattern to process orientation, Chapter 7 is examining the sameissuewithaprocess topatternorientation.13Asix-phaseconceptualmodelof thedevelopment of forest fragmentation is described. The approach uses a combina-tion of initial context (colonization) and resulting spatial arrangement of distur-bance(theroadsystem)asthefoundationtodeveloptheconceptualmodel,whichateachphasehasasocio/econo-politicalcontextthatdrivesestablishmentofmoreelaboratespatiallyexplicitlandscapestructure,leadingtoincreasedfragmentation.Patternmetricssuchasmeanpatchsizeandtotaledgelengtharegooddescriptorsoffragmentation,aligningwellwiththeedgemetriccorrelationwithdeforestationin Chapter5, and the interspersion and patch size metric relating to agriculturalexpansioninChapter6.However,thegoaltothickenunderstandingofthehumandimensionofthechangeleadstothedevelopmentofamodelwithanemphasisoncontextinthepredictiveprocess.Thislandscapeisoneofaparticularpattern(i.e.,herringboneclearancepattern),asaresultofthecolonizationcontext,whichestab-lishesthespatialskeleton,thatisthefoundationforthefinalfragmentationpattern.
Thethreechaptersthatdealwithspatialanalysisoftropicaldeforestationandfragmentation provide a powerful case for combination of spatial analysis andmetricationofspatialpatternsindisturbedlandscapes;analysisofchangesinspatialmetrics through timeas indicatorsofparticularprocessesandparticular trajecto-riesinlandscapestructuretowhicheconomicandbiophysicalfunctionalitycanbeascribed; and application of detailed analysis of socio/econo-political contexts inorder to explain evolution of spatial patterns and regional differences in patternsdescribedbyspatialmetrics.
ThefinalcasestudyinChapter8addressestheissueofrapidurbantransforma-tion.Adecisiontheoryframeworkispresentedthatusespolicies,economicdata,an economic model, and an optimization procedure that minimizes an objectivefunction to produce probability maps of predicted urban expansion. There is nobest or true outcome; the output is a probability surface. The expected rates ofgrowthfortheglobalmegacities—therewerealready19citieswithpopulationsinexcessof10millionin2000—introducesanurgencytodevelopmentofpredictivemodelsforplanningduetoexpandedneedforenergy,sanitation,transport,educa-tion,emergencymanagement,healthandsafety,andcleanairandwater.Captureofsocial,psychological,andeconomicdriverswithinacomplexspatialcontextisevenmoreimportant.Manycitiesarealready“landscapesoffate”thathavemostoftheundesirablepropertiesdescribedlaterinthischapter,anddependuponwealthgener-ationandgentrificationofpooreroruglierareasforsignificanttransformationbacktoamore“desirable”landscape.Therefore,themodelingdescribedinChapter8isofparamountimportanceinprovidingspatiallyexplicitprobabilitiesindicatingareas
42963_C009.indd 168 11/6/07 7:12:24 AM
Synthesis, Comparative Analysis, and Prospect 169
fordevelopmentthatmayenablesomebalancetobeattainedbetweendesirableandfatefulurbanlandscapesforhumanhabitation.Theurbancaserequiresmoreatten-tiontospatiallyexplicitattributionofsociospatialpropertiesandmeasuresofqualityofbuiltspatialhabitatsforhumanactivitiesinordertobalancethepowerfulinflu-enceofcitylandvaluesandcity-basedcommerceandfinancialenterprise.14
9.3 oVeRAll MessAGes
TheframeworkoftheGLPprovidesausefultemplatewithinwhichtoexplorethekeymessagesarisingfromboththeoverviewandcasestudychapterspresentedinthisbook.WehaveusedtheelementsfromFigure2oftheGLPSciencePlanandImplementationStrategy7(“Thecontinuumofstatesresultingfromtheinteractionsbetweensocietalandnaturaldynamics”)toillustratethekeyenablingtechnologies,methods,andapproachesneededtoproviderealsocietalbenefitfromanalysisandstudyoftheseinteractions(Figure9.2).Simplyput,fourofthemajormessagesfromthecasestudiesare:
1. Inordertodetectandaccuratelymeasurechange,remotesensingdataandassociated methodologies must deliver the highest possible informationcontentandaccuracyinchangediscrimination.
2.Remotely sensed data should be complemented with detailed householdand other socioeconomic data from field and census surveys to addressdecision-makingprocesses indetailandgainabetterunderstandingandcapacitytomodelhumanandothersocialandeconomicprocessesinflu-encinglandusechange.
3.Usingthebasicremotesensingandsurveydataresources(numbers1and2) together with spatially explicit descriptions of social, financial, juris-dictional,political,andpsychologicalunitsandinfluences,thereneedstobeconcertedandintegratedapplicationofavarietyofnumerical,heuristic,spatial,andtemporalmethodstoderivethehighestlevelsofunderstandingandquantificationofdependencybetweenpatternsandprocesses.
4.Theanalysisfromnumber3shouldbeplacedinapragmaticcontextthroughcloser relations between science with management and policy related tolanduseandlandusechange.
9.3.1 Realizing the Full Potential FRom Remote SenSing
The implementation plan for the Global Earth Observation System of Systems(GEOSS)15hasdefinedninekeytargetsfordeliveryofsocietalbenefitoverthenext10 years, including improving the management and protection of terrestrial andcoastal ecosystems; supporting sustainable agriculture and combatingdesertifica-tion;andunderstanding,monitoring,andconservingbiodiversity.Alargenumberofobservationalrequirementshavebeendefinedforecosystems,biodiversityassess-ment, and agricultural monitoring. These include particular properties associatedwithlandusechangesuchasburnedareas,landdegradation,speciesdistribution,alien species, extent and location of ecosystems and habitat types, fragmentationof ecosystemsandcommunitycomposition, cultivationandclearing, andgrazing
42963_C009.indd 169 11/6/07 7:12:25 AM
170 Land Use Change: Science, Policy and Management
Land
use
Hum
an co
ntro
l
Dyn
amic
land
tran
sitio
nsLa
nd co
ver
Biop
hysic
al co
ntro
l
Enab
ling t
echn
olog
ies,
met
hods
and
ap
proa
ches
• H
ighe
st p
ossib
le in
form
atio
n co
nten
t and
chan
ge
disc
rimin
atio
n fro
m re
mot
e sen
sing
• C
once
rted
and
inte
grat
ed ap
plic
atio
n of
num
eric
al,
heu
ristic
, spa
tial a
nd te
mpo
ral m
etho
ds•
Con
nect
ion
and
harm
oniza
tion
of p
arad
igm
s
bet
ween
econ
omic
, soc
ial, p
sych
olog
ical
and
b
ioph
ysic
al do
mai
ns
Dec
ision
–mak
ing
Choi
ces,
know
ledg
e, va
lues
, pre
fere
nces
and
thei
r soc
ial a
nd p
oliti
cal c
onte
xt
Soci
al ch
alle
nges
Pove
rty
Confl
ict
Soci
al ju
stic
eM
igra
tion
Cons
umpt
ion
Hea
lth
Soci
etal
Ben
efit
Ecol
ogic
al ch
alle
nges
Pollu
tion
Dise
ases
Food
/fibe
r/fu
el sh
orta
geO
verc
row
ding
Clea
n w
ater
supp
ly
Ecos
yste
m g
oods
and
serv
ices
Clea
n ai
rCl
ean
wat
erW
aste
recy
clin
gFo
od/fi
ber/
fuel
Recr
eatio
n
Soci
al sy
stem
sPo
pula
tion
Soci
al/e
cono
mic
stru
ctur
ePo
litic
al/in
stitu
tiona
l reg
imes
Cultu
reTe
chno
logy
Ecol
ogic
al sy
stem
sBi
ogeo
chem
istry
Biod
iver
sity
Air
Wat
erSo
il
FIG
UR
e 9.
2 A
dia
gram
of
the
maj
ore
lem
ents
of
the
coup
led
hum
an e
nvir
onm
ent
syst
em(
afte
rG
LP,
200
5;
Figu
re2
)au
gmen
ted
wit
hke
yen
abli
ngt
echn
olog
ies,
met
hods
and
dat
a, a
ndp
arad
igm
sfo
ran
alys
is o
fco
uple
dhu
man
env
iron
men
tsys
tem
san
dde
live
ryo
fso
ciet
alb
enefi
ts.
42963_C009.indd 170 11/6/07 7:12:26 AM
Synthesis, Comparative Analysis, and Prospect 171
impacts.Animprovement in thequalityandcoverageofobservationsof the landsurfacefromremotesensingisneededtorealizetheserequirements.
Allofthecasestudiesherethataddresstropicalrainforestclearance(inSouthAmerica,Asia,andAfrica)dependtoalargedegreeonremotesensingasaprimarysourceofdataforbasicchangedetection.CrewsinChapter6discussestheprob-lemswithmultiplicativeerrorswhenusingimagesfrommultipledates,evenwithhigh accuracies for individual classifications. The increasing availability of veryhigh-resolution imagery from space (down to 60 cm pixel resolution) means thatverydetaileddefinitionoflandcoverboundariesandindividualvegetationunitsispossibleatspecificlocations.However,highimagecostandsmallimagefootprintsmeanthatthisapproachisstillimpracticalforwidespreadchangedetection.Recentresearchhasshownthatdetectionofchangesinforestsystems,previouslylimitedbytheinsensitivityofmultispectralinstrumentssuchasLandsattosmallchangesinspectralsignaturesinheavilyfoliatedforestsystems,canbegreatlyimprovedusingspectralunmixingwith time seriesofmultispectraldata16 andwithhigh spectralresolutionspace-bornesensorssuchasHyperion.6,17Aglobalhyperspectralimagingsystemwithsufficientsignaltonoise,moderatepixelresolution(40to60m),andhighcycleforglobalcoverage(15to30days)coulddramaticallyincreaseaccuracyandsensitivityoflandcoverchangedetectionduetolandusepractices.Itisarguablethatthenaturalconclusiontothedevelopmentofremotesensingtechnologyisthecapabilitytoundertakespectroscopyofbiosphericsurfacetargetstodeliverquanti-tativevaluesforkeysurfacestructuralandbiogeochemicalproperties.18
9.3.2 aPPlication oF integRated methodS
TheapplicationofintegratedmethodsdependsonthecomprehensiveaddressingofthebasicsciencequestionsoutlinedbyAspinallinChapter1.Manyaccuracyandscaleissuescanbeaddressedbymaximizingtheinformationcontentfromremotesensing.Theinformationcontentfromremotesensingismaximizedbyprovidinga synergistic mix of imagery with full spectral fidelity and calibration, completeglobalcoverageandhightemporalfrequencyatmediumresolution,andfullspectralfidelityandcalibrationwithveryhighspatialresolutionsuchthatradiometricandspatialscalingisoptimized.ThisremotesensingsystemwouldsatisfymanyoftheinitialneedsofGEOSS,15andtheproductsofthissystemwouldprovideabench-marklevelofreliable,spectrallycomprehensive,temporalandspatialcoveragewithexplicitquantitativeuncertaintyestimates.
This improved information content in remote sensing addresses the needfor better capture of system dynamics in time and space in the future. However,historicalanalysiscanbegreatlyenhancedjustbycomprehensiveanalysisofglobalLandsat MSS and Landsat TM/ETM archives. The Australian government sup-portedan innovativeprogram toanalyzemore than30yearsofLandsatdata forAustraliainordertomonitorandmeasurelandcoverchangetosupportanationalcarbon accounting system.19,20,21 A large archive exists for the United States, forexample,butafull-timeseriesanalysisof thishasyet tobeundertaken(researchhascommencedtoexamineforestdisturbanceusingthesedatabutonlyataregionalscale;SamGoward,personalcommunication).Theprecedingisnotintendedtoover-
42963_C009.indd 171 11/6/07 7:12:27 AM
172 Land Use Change: Science, Policy and Management
emphasizetheimportanceofremotesensinginintegratedanalysis,merelytohigh-lightthecriticalroleitcanandshouldplay.
The important biophysical processes are described in detail by many mod-els of varying complexity.22,23 Often these models, developed for site-basedapplication, are difficult to supply with spatially explicit parameters and inputs,resulting in insufficient information to constrain model parameters and provideeffectivemodelpredictions.24Thecross-fertilizationbetweendisciplines,initiatedbyEarthsystemconcernsandfocus,hasprovidedmanynumerical,quantitative,andheuristicmethodsforfindingoptimalmodelfitstoobservationsusingdiversedatawithdifferentspatialandtemporalproperties.Thesemathematicaltechniquesarecollectivelyreferredtoasmultiple-constraintsmodel-dataassimilation.24Themethodsarediverseandhavemigratedfromdiversedisciplinarydomainssuchasnumericalweather prediction25 and economicoptimization andmathematics.26,27These approaches, specifically identified in the Global Land Project strategicimplementationplan(Figure9.1),andintheGEOSS10-yearimplementationplan15mayprovidewaystocombineinformationfromallavailabledatasetstocapturespatiallyexplicitsurfacesprocessesthatdirectlyinfluencelandsystemsandrepre-sentmanyoftheconsequencesoflandusechange.
Thecaptureofthesebiophysicalprocessesisonepartofthepuzzleneededtounderstandthefeedbacksincoupledhumanenvironmentsystems.28Theotherpartinvolvestheintegrationandlinkingofthespatiallyandtemporallyexplicitdynamicsfromobservationandbiophysicalmodelingwithsocialandeconomicmeasuresandmetrics,andwithculturalviewsandhumanperceptionsandopinions.Theresearcheffortsto“socializethepixel”29havebeenparalleledbydigitizingofparcelinforma-tioninpartsoftheUnitedStates,forexample(therebyprovidingdetailedownershiphistoriesandfinescalelandstatistics30),andincreasedabilitytoconsiderthespatialorganizationofpopulations, leading todevelopmentof spatiallyexplicitdata setscontaining survey information on attitudes and behaviors with concomitant con-cernsforconfidentiality.31Inadditionobservationofurbanecosystemsisexplicitlyconsideredwith,forexample,theBaltimoreandPhoenixsitesintheU.S.LongTermEcologicalResearchnetwork.32
Integrationofallthisinformationandunderstandingofdynamicsmaybepro-vided by land use change models33 that range from cellular automata types,34 tostatisticalor simulationmodels,33agent-basedmodels,35and integratedecologicaland economic modeling.36 These models must accommodate dynamics acrossscales, represent thedrivingforces,capturespatial interactionsandneighborhoodeffects,capture the temporaldynamics,and thenbecapableof integrationacrossdisciplinarydomains.33Somecautionisrequiredhere,sincewehavebeendownthispathwithverycomplexprocess-basedbiologicalandecosystemmodels.Wemayinfactneeda“qwerty”solution,apragmaticcombinationofmethodsandapproachesthatisgoodenoughandthatinterfaceswiththehumandecisionframework,butthatis not necessarily the optimum in science or computation—something judged byitseffectivenessratherthanelegance.However,thereissomereasonforoptimism,expressed earlier, that computational power, process understanding, and diversemethodswilldeliverhighlysophisticatedandeffectiveanalysisofcomplexcoupledhuman environment systems. A range of tools that enable humans to interface
42963_C009.indd 172 11/6/07 7:12:27 AM
Synthesis, Comparative Analysis, and Prospect 173
meaningfullywiththeinformationneededfordecisionsarealsorequired.Onedis-cussedbrieflyinChapter2ismulticriteriaanalysiswithaspatialinterfacedesignedforinteractivework-shopping;37however,agent-basedmodelsandotherinteractiveparadigmsmayworkjustaswell.
Throughincrementalresearchovermanyyears,crossfertilizationofmethodsbetweendisciplines,digitalcaptureofmorespatiallyexplicitbiophysicalandsocialdata,andcontinuedcomputationaladvance,theprospectsforsophisticatedanalysisofcomplexcoupledhumanenvironmentsystemsleadingtomuchbetterinformeddecisionmakinghavebecomerealizable.OneexampleofthisprocessfromAustraliaseespolicymakersandscientistsdrivingahighlysophisticated landusemappingprocess, accompanied by a sophisticated approach to definition of land manage-mentpractices38anddevelopmentofarobustecosystemscience-basedclassificationsystemforvegetationdisturbance.39,40However,theanalysisofthecoupledhumanenvironmentsystemmustalwaysbeplacedincontext.Thetrade-offbattlebetweenanthropocentricandenvirocentricviewsofthefutureoftheearthsystemwillbeamajorbattlegroundinthe21stcentury.Socontextmatters.
9.3.3 Placing analySiS in context
Therequirementfordeliveryofsocietalbenefitthathasbeenemphasizedinmanyforums7,15hasplacedanimperativeontheintegrationandharmonizationofanalysis,modeling,anddecisionmakingandlinkageofsciencetomanagementandpolicymakingbysociety.Emphasisisplacedonthecontextforproblemsandwhethertheyareimportant(orperceivedasimportant).Thisoftendependsoncommunicationofissuesinalanguageorframeworkthatimpactsdirectlyonsocietymembers.
Althoughnotwithoutcontroversy,thesimplesocietalmodelofLuhmann41pro-videsonewaytoconceptualizetheproblem(Figure9.3).Luhmannviewssocietyasacenterlesssetof“functionsystems”thatconstrainbothwhatcanbecommunicatedand how it is communicated. He labels economy, law, science, politics, religion,andeducation as themost important function systems in contemporary society.42Thesefunctionsystemshavedifferent time intervals forexternalcommunication,ranging fromdailydiscourse in science and religion,monthly courtprocesses inlaw,quarterlyteachingandreportingcyclesineconomicsandeducation,toannualelectioncyclesinpolitics(averageoveralllevelsofgovernment).Informationfromone function system only becomes active in another function system when it istranslated into the code of that function system. Hence, environmental informa-tiondoesnothaveanimpactoneconomicorpoliticalprocessuntilitistranslatedintothecodeoftheeconomicorpoliticalfunctionsystem.Globalclimatechangeis a science/environment issue that has crossed between function systems and iswidelyconsideredinthepoliticalfunctionsystem.However,thereisstillaprocessofcontinuedtransferofimprovedscientificinformationandacontinueddemandforinformationinaformthatcanbeusedtomakepoliticaldecisions.
If we therefore return to the choice between alternate future landscapes,10the“landscapesofdesire”and the“landscapesof fate,”even if informationfromenvironmental sciencehasbeen translated into thecodeof thepolitical andeco-nomicsystemstosaythat,forexample,“ifwecontinuewithacertainformofman-agement, certain consequences will ensue”, there remains a need to provide the
42963_C009.indd 173 11/6/07 7:12:28 AM
174 Land Use Change: Science, Policy and Management
methodsforexcellentanalysisofdifferentscenariosandoutcomes(Figure9.4).Forexample,thenumberofcitieswithonemillionpeoplehasgrownfrom16in1900stomorethan400in2000,andthenumberofcitieswith10millionpeoplehasgrownfromonein1950to19in2000(seeChapter8).Thereisaneedtohaveverysophis-ticatedmethodsthatcanexplorethewiderangeofinteractionsandconsequencesensuingfromsuchrapidurbangrowth.ThesemethodsmustaddressthekeysciencequestionsfromChapter1inprovidingthebestpossibleanalysisandscenariosfordecisionmaking,sincethiswillinevitablyinvolve“trade-offsbetweenimmediatehumanneedsandmaintaining thecapacityof thebiosphere todelivergoodsandservicesinthelongterm.”1Sincethesedecisionspaceswillbehighlycontested,thequalityandtransparencyofdata,methods,andassumptionswillbeofparamountimportance.Inaddition,regionallandusechangewithnegativeimpactsmayhaveglobalconsequencesthroughclimate-surfaceinteractions.43
9.4 ConClUsIons
Thisbookdocumentsthedevelopmentofanalysisoflandusechangethroughpre-sentationofarangeofcasestudies,placedincontext throughreviewof landusescience,integrativemethods,andframeworksforaddressingcomplexityandinter-relationships.Anabidingthemethatlingerssubliminallythroughoutthisbookisthetime-limiteddecisionspaceofPotschinandHaines-Young10andthetrajectoriesofSteinitzandcolleagues.9Bythesemeanshumandecisionsleadto“landscapesoffate”(perhapsacompletelyurbanizedworld),andhumanaspirationscrave“landscapesofdesire”(perhapsafairytaleland).Thechallengeistousethesophisticatedsciencemarriedtosocialandsoftsystemsparadigmstoweaveapathtoalandscapethatpreservesthefulldimensionsofhumandesireandaspirationwhileretainingafullyfunctionalearthsystem.
Data Information Cultural Knowledge
Politics
SocietalSub–systems
Active – translatedinto code of
functionsubsystem
Inactive – not yettranslated into thecode of function
subsystem
Externalcommunication
Information
Economics
Education
Law
Religion
Science
Translation
Annual
Quarterly
Monthly
Daily
FIGURe 9.3 Connecting the paradigms. Modeling coupled natural human systems.Luhmann41viewssocietyasacenterlesssetof“functionsystems”thatconstrainbothwhatcanbecommunicatedandhowitiscommunicated.Helabelseconomy,law,science,politics,religion,andeducationasthemostimportantfunctionsystemsincontemporarysociety.42
42963_C009.indd 174 11/6/07 7:12:29 AM
Synthesis, Comparative Analysis, and Prospect 175
Lan
dsca
pe
Posi
tive
prop
ertie
se.
g., A
esth
etic
ally
ple
asin
gPa
stor
alE
xciti
ngPr
oduc
tive
Mul
ti-fu
nctio
nal
Upl
iftin
gR
elax
ing
Bal
ance
d ec
osys
tem
Rem
ote
sens
ing
for
chan
gede
tect
ion
Pol
icy
cont
ext,
impo
rtan
ce,
impa
ct, e
ffec
tof
cha
nges
Lan
dsca
peof
des
ire
Cur
rent
land
scap
e
Man
agem
ent
New
land
scap
e
Neg
ativ
e pr
oper
ties
e.g.
, Ugl
yIn
dust
rial
Bor
ing
Art
ific
ial
Uni
-fun
ctio
nal
Dep
ress
ing
Stre
ssin
gD
egra
ded
ecos
yste
m
Div
erse
spa
tio-
tem
pora
lm
etho
ds f
oran
alys
is o
fpa
ttern
and
proc
ess
Mul
tiple
crite
ria
and
trad
e-of
fap
proa
che.
g.,
ecos
yste
mse
rvic
e
Lan
dsca
peof
fat
e
Prop
ortio
n de
term
ined
by
rela
tive
desi
rabi
lity
orun
desi
rabi
lity
Mix
FIG
UR
e 9.
4 T
he c
onte
xtf
or s
pati
ala
nd t
empo
ral
anal
ysis
in
dete
rmin
ing
the
mix
of
land
scap
eso
ffa
te
and
desi
re.
42963_C009.indd 175 11/6/07 7:12:31 AM
176 Land Use Change: Science, Policy and Management
ReFeRenCes
1. Foley,J.A.etal.Globalconsequencesoflanduse.Science309,570–574,2005. 2. Lewis,S.L.Tropicalforestsandthechangingearthsystem.Philosphical Transactions
of the Royal Society B361,195–210,2006. 3. Soares-Filho,B.S.etal.Modellingconservation in theAmazonbasin.Nature440,
520–523,2006. 4. Silva,J.F.etal.Spatialheterogeneity,landuseandconservationinthecerradoregion
ofBrazil.Journal of Biogeography33,536–548,2006. 5. Defries,R.S.,Asner,G.P.,andHoughton,R.Trade-offsinlandusedecisions:Towards
a framework for assessing multiple ecosystem responses to land-use change. In:Defries,R.S.,Houghton,R.,andAsner,G.P.,eds.,Ecosystems and Land Use Change.GeophysicalMonographSeries153,AmericanGeophysicalUnion,2004.
6. Asner, G. P. et al. Ecosystem structure along bioclimatic gradients in Hawaii fromimagingspectroscopy.Remote Sensing of Environment96,497–508,2005.
7. GLP.Global Land Project. Science Plan and Implementation Strategy.IGBPReportNo.53/IHDPReportNo.19.IGBPSecretariat,Stockholm,64pp,2005.
8. Machlis,G.E.,Force,J.E.,andBurch,W.R.Thehumanecosystem,PartI:Thehumanecosystemasanorganizingconceptinecosystemmanagement.Society and Natural Resources10,347–367,1997.
9. Steinitz,C.etal.Alternative Futures for Changing Landscapes: The Upper San Pedro River Basin in Arizona and Sonora.IslandPress,Washington,D.C.,202pp,2003.
10. Potschin,M.,andHaines-Young,R.“Rio+10”,sustainabilityscienceandLandscapeEcology.Landscape and Urban Planning75,162–174,2006.
11. Geist,H.,andLambin,E.F.Proximatecausesandunderlyingdrivingforcesoftropicaldeforestation.Bioscience52,2,143–150,2002.
12. Geist,H.J.,andLambin,E.F.Dynamiccausalpatternsofdesertification.Bioscience54(9),817–829,2004.
13. Laney,R.Aprocess-ledapproachtomodellinglandusechangeinagriculturalland-scapes:AcasestudyfromMadagascar.Agriculture, Ecosystems and Environment 101,135–153,2004.
14. Daniel, T. C. Whither scenic beauty? Visual landscape quality assessment in the21stcentury.Landscape and Urban Planning54,267–281,2001.
15. GEO.Global Earth Observation System of Systems. 10-year Implementation Plan and Reference Document.B.Battrick,ed.EuropeanSpaceAgencyPublicationsDivision,Netherlands,209pp,2005.
16. Asner,G.P.etal.SelectiveloggingintheBrazilianAmazon.Science310,480–482,2005.
17. Asner,G.P.etal.DroughtstressandcarbonuptakeinanAmazonforestmeasuredwithspaceborneimagingspectroscopy.Proceedings of the National Academy of Sciences101,6039–6044,2004.
18. Hill,M.J.,Asner,G.P.,andHeld,A.A.Hyperspectralremotesensingofvegetationincoupledhuman–environmentsystems—societalbenefitsandglobalcontext.Journal of Spatial Sciences32,49–66,2006.
19. Richards, G. P. The FullCAM carbon accounting model: Development, calibrationand implementation for the National Carbon Accounting System, National CarbonAccountingSystem. TechnicalReportNo.28,AustralianGreenhouse Office,Canberra,2001.
20. Furby, S. Land cover change: Specification for remote sensing analysis. NationalCarbon AccountingSystem,TechnicalReportNo.9. AustralianGreenhouseOffice,Canberra,2002.
42963_C009.indd 176 11/6/07 7:12:31 AM
Synthesis, Comparative Analysis, and Prospect 177
21. Furby,S.,andWoodgate,P.W.,eds.Remotesensinganalysisoflandcoverchange-pilottestingoftechniques.NationalCarbonAccountingSystem, TechnicalReportNo.16,AustralianGreenhouse Office,Canberra,2002.
22. Plummer, S. E. Perspectives on combining ecological process models and remotelysenseddata.Ecological Modelling129,169–186,2000.
23. Nightingale,J.M.,Phinn,S.R.,andHeld,A.A.Ecosystemprocessmodelsatmultiplescales formapping tropical forestproductivity.Progress in Physical Geography 28,1–41,2004.
24. Barrett,D.J.etal.Prospectsforimprovingsavannacarbonmodelsusingmultiplecon-straintsmodel-dataassimilationmethods. Australian Journal of Botany55,689–714,2005.
25. Todling,R.EstimationtheoryandatmosphericdataassimilationIn:Kasibhatla,P.etal.,eds., Inverse Methods in Global Biogeochemical Cycles. American GeophysicalUnions,Washington,D.C.,49–65,2000.
26. Whitley, D. An overview of evolutionary algorithms: practical issues and commonpitfalls.Information and Software Technology43,817–831,2001.
27. Yang,R.L.Convergenceofthesimulatedannealingalgorithmforcontinuousglobaloptimisation.Journal of Optimization Theory and Applications104,691–716,2000.
28. Verberg,P.H.Simulatingfeedbacksinlanduseandlandcoverchangemodels.Land-scape Ecology21,1171–1183,2006.
29. Rindfuss,R.R.etal.Developingthescienceoflandchange:challengesandmethod-ologicalissues.Proceedings of the National Academy of Sciences101,13976–13981,2004.
30. Aspinall,R.J.Modellinglandusechangewithgeneralizedlinearmodels—amulti-modelanalysisofchangebetween1860and2000inGallatinValley,Montana.Journal of Environmental Management72,91–103,2004.
31. VanWey,L.K.etal.Spatialdemographyspecialfeature:confidentialityandspatiallyexplicit data: Concerns and challenges. Proceedings of the National Academy of Sciences102,15337–15342,2005.
32. Grimm, N. B. et al. Integrated approaches to long-term studies of urban ecologicalsystems.Bioscience50,571–584,2000.
33. Verburg,P.H.etal.Landusechangemodeling:currentpracticeandresearchpriorities.GeoJournal61,309–324,2004.
34. White, R., and Engelen, G. High resolution integrated modelling of the spatialdynamicsofurbanandregionalsystems.Computers, Environment and Urban Systems24,383–400,2000.
35. Barreteau,O.,andBousquet,F.SHADOC:Amulti-agentmodeltotackleviabilityofirrigatedsystems.Annals of Operations Research94,139–162,2000.
36. Vionov,A.etal.Patuxentlandscapemodel:integratedecologicaleconomicmodellingofawatershed.Environmental Modelling and Software14,473–491,1999.
37. Hill,P.,Cresswell,H.,andHubbard,L.SpatialprioritisationofNRMinvestmentintheWestHumearea(MurrayCMAregion).TechnicalReportNo.2006,CSIROWaterforaHealthyCountryNationalResearchFlagship,Canberra,2006.
38. Lesslie, R., Barson, M., and Smith, J. Land use information for integrated naturalresources management—a coordinated national mapping program for Australia.Journal of Land Use Science1,45–62,2006.
39. Thackway,R.,andLesslie,R.Vegetationassets,statesandtransitions(VAST):Account-ingforvegetationconditionintheAustralianlandscape.BRSTechnicalreport,BureauofRuralSciences,Canberra,2005.
40. Thackway, R., and Lesslie, R. Reporting vegetation condition using the vegetation,assets, states and transitions (VAST) framework. Ecological Management and Restoration7,s53–s62,2006.
42963_C009.indd 177 11/6/07 7:12:32 AM
178 Land Use Change: Science, Policy and Management
41. Luhmann, N. Ecological Communication. Translation Bednarz, J. Jr. University ofChicagoPress,Chicago,187pp,1989.
42. Grant,W.E.,Peterson,T.R.,andPeterson,M.J.Quantitativemodellingofcouplednatural/human systems: simulation of societal constraints on environmental actiondrawingonLuhmann’ssocialtheory.Ecological Modelling158,143–165,2002.
43. Asner, G. P., and Heidebrecht, K. B. Desertification alters ecosystem-climate inter-actions,Global Change Biology10,1–13,2004.
42963_C009.indd 178 11/6/07 7:12:32 AM
179
Index
AACLUMP see Australian Collaborative Land Use
Mapping ProgramAdvanced Very High Resolution Radiometer
(AVHRR), 58AEM see Airborne electromagneticsAEZ see Agroecological zones databaseAgricultural landscape metrics, 128Agroecological zones (AEZ) database, 70Airborne electromagnetics (AEM), 59ALUM see Australian Land Use Management Amazon basin forest fragmentation, 119–134 case study and methods, 122 Chapare, 122 conceptual model, 124–127 landscape metrics behavior, 128–132 methods, 123Amazon Colonization Fronts, 90–91Amazon region of Colombia, 84 forests, 95 global deforestation, 83Analysis in context, 173Animal impact, 31Applied science, 9–11 decisions, 10 human and environmental responses to change,
11 scientific knowledge interpretation and
communication, 11Arequipa forest fragmentation, 123 metrics, 132Asia urban populations, 139Australian Collaborative Land Use Mapping
Program (ACLUMP), 46Australian Land Use Management (ALUM), 48 classification, 49Australian rangelands, 23–32, 35 biophysical attributes, 30 ecosystem assessments, 44 grazing piospheres, 25 landscape structure, 26 MCA framework, 23 representation scale, 29 scale of representation, 29–30 spatial interrelationships, 25 spatial patterns and relationships, 25–26 spatiotemporal inputs to rangeland MCA,
30–31
temporal patterns and influences, 26–29 vegetation types, 36AVHRR see Advanced Very High Resolution
Radiometer
BBasic and applied land use science, 3–11Basic science, 5–8 ecological system, 7 landscape and climate, 7 land systems resilience, vulnerability and
adaptability, 8 scale issues, 8 socioeconomic system, 7 space and time change dynamics, 5–6 uncertainty, 8Binary forest maps, 86Binary non forest maps, 86Biophysical attributes Australian rangelands, 30 land cover change, 85 land use science, 54–59Biophysical data, 50–52 data source integration, 48–49 land use science, 43–60 mapping relationships, 53–54Biophysical integration, 44–59 data and land mapping correspondence, 48–53 future directions, 59 implications, 54–58 regional scale mapping, 45Bogotá forest fragmentation, 123Bolivia see Chapare region of BoliviaBorrow pits, 114
CCA see Cellular automata modelsCaracas forest fragmentation, 123Caribbean soil fertility, 88Carrasco, 123Case studies, 10Cellular automata (CA) models, 145Central America, 94Chapare region of Bolivia, 120, 122–123, 133 Amazon basin forest fragmentation, 122 land use change, 124, 127–128
42963_Index.indd 179 11/5/07 1:51:16 PM
180 Index
progressive deforestation, 125 road building, 134China city planning departments, 158 Pearl River Delta, 150–153 urban land use change, 150Chinese urban land-use change, 139–154 modeling, 140–146 modeling and policy making intersection,
143–145 new modeling approach, 147–150 policy evaluation and uncertainties, 146 policy making, 142Clearance morphologies, 120Clearance typologies, 120Colombia Amazon region, 84 economy, 84 forested ecosystems, 84 land cover change, 83 lowland forests, 92 non forest ecosystems, 84 population, 83Colombian Amazon forest cover change, 93Colonization forest maps, 90Colonization front-line Amazon, 90Communidad Arequipa metrics, 132Community forest maps, 124Comparative analysis and prospect, 163–173Complex spatial patterns, 23, 24Complex spatial relationships, 24Complex spatial signals, 24Complex temporal patterns, 23, 24Complex temporal relationships, 24Complex temporal signals, 24Conceptual model of forest fragmentation, 127Conservation values, 55Correspondence assessment spatial methods analysis, 50 approach, 50 context, 50
DDasymetric mapping, 21Data-generating process (DGP), 144Data source integration, 48–49 Glenelg Hopkins Landholder Survey, 49 natural environment categorized, 49 salinity discharge sites, 49Decision making process models, 154Deforestation, 66–69 agents, 64 conceptual framework, 68–69 data sources, 69–70 definition, 66–67 distribution, 66
econometric results, 74–75 economic model, 70–71 economic models, 64 factors, 68 forest cover, 67 fronts, 93, 135 global land cover, 82 hot spots, 89 hot spot spatial location, 90 institutional context, 76–77 institutional intervention, 76 methodological issues, 71–72 methods, 69–72 modeling unplanned land cover change across
scales, 91 national threat, 88–90 patterns, 91, 94 poor households, 77 prediction, 88, 89 processes, 68 progressive, 125 reduction, 78 results and discussion, 72–77 socioeconomic context, 75–76 spatial context, 76 trend analysis, 91 Uganda, 66, 72, 73, 77 Western Uganda, 63–78Developing world city policy makers, 140DGP see Data-generating processDynamics land use science, 5–7
EEarth Resource Technology Satellite (ERTS), 100Earth surface phenomena, 6Earth system land use, 3Ecological deterioration, 119Economic development of Uganda, 67Economic growth rate of Uganda, 65Economic model deforestation, 70–71Economic statistics, 21Ecosystem assessments of Australia, 44Ecosystem management, 45Ecosystem services improvement strategies, 34 MCA, 34Ecuador land use models, 94Environmental crisis in Uganda, 76Environmental responses to change, 11ERTS see Earth Resource Technology SatelliteErythroxylum coca, 85
FFAO see Food and Agriculture OrganizationFFT see Fourier transform
42963_Index.indd 180 11/5/07 1:51:16 PM
Index 181
Final phase plot exhaustion, 128Fishbone deforestation, 127Food and Agriculture Organization (FAO), 65Forest(s) Amazon region of Colombia, 95 fragmentation patterns, 121 logistic pattern, 92Forest conversion management interventions, 122 planning, 122 spatial patterns, 85Forest cover Colombian Amazon, 93 decline, 92 deforestation, 67 transformation process, 92Forested ecosystems of Colombia, 84Forest energy in Uganda, 68Forest fragmentation Amazon Basin, 119–135 Arequipa, 123 Bogotá, 123 Caracas, 123 case study, 122–124 conceptual model, 124–128, 127 lowland forests, 121 progressive development of spatial patterns,
124 spatial patterns, 127 spatial relationships, 123 temporal relationships, 123 tropical lowlands, 122 tropics, 120Forest landscape metrics, 128Forest loss, 119Forest maps, 90Forest presence, 88Forest regeneration, 90Forestry sector background, 64 Uganda, 65–66Fourier transform (FFT), 27Fragmentation metrics used, 128 patterns, 121Fragstats, 103, 128 landscape, 104
GGeographical information systems (GIS), 6, 18 Glenelg Hopkins region, 48 landscape scales, 45 platforms, 21Geographic Information Science community
(GISc), 102Geology, 6
Giles metrics, 130GIS see Geographical information systemsGISc see Geographic Information Science
communityGlenelg Hopkins Landholder Survey, 48 data source integration, 49Glenelg Hopkins region, 58 GIS, 48 location, 48Glenelg Hopkins Salinity Plan, 50Global deforestation, 82 Colombian Amazon region, 83Global human population, 81Global land cover deforestation, 82Grazing piospheres, 25Green Revolution, 105Groundwater flow systems, 48–49Guangzhou nonurban area predications, 155 urban area predictions, 155
HHumidity, 90
IInstantaneous field of view (IFOV), 102Integrate methods application, 171–172Interspersion/juxtaposition index (IJI), 111
LLand cover, 5 Amazon Colonization Fronts, 90–91 biophysical characteristics, 85 characteristics of theory, 4 Colombia, 83 data sources, 87 deforestation patterns, 91–92 discussions and conclusions, 92–96 emerging trends, 10 GIS, 86 global human population, 81 human activities, 81 human-induced changes, 82 human intervention, 10 integration of theory, 4 local level, 85, 86 methods, 85–88 national level, 86 regional level, 84–85, 86 study region, 82–85 unplanned modeling, 81–96
42963_Index.indd 181 11/5/07 1:51:17 PM
182 Index
Land managers attitudes, 52 conservation value characteristics, 56–57 conservation values, 55 distance to salinity discharge, 51 landscapes of colonization, 134 land use change, 54 perceptions, 48–54 practices, 48–54 relationships, 53–54 salinity awareness, 53, 57 salinity discharge sites, 50 surveys, 45 values, 48–54Land mapping correspondence correspondence assessment spatial methods,
50–52 data source integration, 48–49 mapping relationships, 53–54Landsat Thematic Mapper images, 69Landscape analysis, 133 Australian rangelands, 26 behavior, 128–133 climate, socioeconomic, and ecological
systems integration, 7 ecology, 102, 119 fragmentation, 119 Fragstats, 104 GIS, 45 impacts, 96 level, 100, 101 LULCC, 101 metrics, 128–133 phase calculations, 129 plot exhaustion, 133 scales, 45 seasonality impacts, 114 structure, 26 tropical forest landscapes, 96Landscape dynamism, 99–115 extension to pattern metrics, 103 local lessons, 106–111 panel approach, 102–103 paneled pattern metrics, 112–115 pattern and structure, 101–102 pattern metrics, 103 process and function, 101–102 temporal frequency, 100–101 Thai testing grounds, 104–107Land systems empirical studies, 4 frameworks for study, 7 human systems, 8 interdisciplinary approaches, 7 natural systems, 7 resilience, vulnerability and adaptability, 8
societal benefits, 8 temporal dynamics, 6 transdisciplinary approaches, 7Land use, 5 conservation, 49–50 conversions, 143 dynamics, 6 Earth system, 3 emerging trends, 10 human intervention, 10 information, 11, 49 natural processes, 5 regional mapping, 45–48 social processes, 5 spatial methods, 50–53Land use change, 5 areal change, 46 biology, 20 color figure, 47 concept, 19 context of role, 70 dynamics, 46 economic statistics, 21 importance, 46 information transformations, 18 land managers, 54 lower Murray Region, 46–48 measurement approaches, 46 modeling, 142 multicriterion decision analysis, 17–37 pattern-led analysis, 21 policy decision making, 19 prediction, 46 process-led analysis, 21 related policy making, 148 scale-dependent effects, 20 social statistics, 21 socioeconomic context, 70 space-time domain, 17 spatially applied multicriterion analysis, 18 theories, 144 transformation, 46 transformation domains and methods, 21–23 transformation issues, 19–21 types, 46, 142 Uganda, 77 water supply rivers, 47Land use land cover (LULC), 106 classes, 109 classification, 105 cloud-free imagery, 106 digital archives, 113 mapped products, 113 MPS, 110 pattern metric change, 110 rice agriculture, 109 Thailand, 108
42963_Index.indd 182 11/5/07 1:51:17 PM
Index 183
Land use/land cover change (LULCC), 100 community problems, 103 constant fluctuation, 112 digital archives, 113 drivers, 112 landscape level, 101 landscapes, 115 panel approach, 103 panel method, 115 research, 101Land use models Central America, 94 Ecuador, 94 measurement tools, 149 policy decision making process, 149Land use related research communication, 11 interpretation, 11Land use science applied science, 9–11 basic and applied, 3–11 biophysical data, 54–59 biophysical data integration, 43–60 definition, 3 dynamics, 5–7 feedback with various systems, 7–8 future directions, 59 integration with various systems, 7–8 scales, 8 science questions, 4 social data, 54–59 social data integration, 43–60 theoretical foundations, 4–5 uncertainties, 8Land use systems scientific knowledge interpretation, 11 time-related factors, 6Lowland forests, 121LULC see Land use land coverLULCC see Land use/land cover change
MMagdalena soil fertility, 88Mapping relationships, 53–54 analysis, 54 approach, 53 biophysical data and land mapping
correspondence, 53–54 context, 53MAS/CA see Multiagent system CAMature forests, 91MAUP see Modifiable areal unit problemMCA see Multicriteria analysisMean nearest neighbor (MNN), 131Mean patch size (MPS), 105 LULC, 110 metrics, 131
Metrics Communidad Arequipa, 132 Giles, 130 MPS, 131 Trani, 130Migrants in Uganda, 76MNN see Mean nearest neighborModeling unplanned land cover change across
scales, 81–96 Amazon colonization fronts, 90 deforestation patterns at local level, 91 local level, 85, 88 methods, 85–88 national deforestation threat hot spots, 88–89 national level, 86 regional level, 84, 86–87 results, 88–91 study region, 83–85Moderate Resolution Imaging Spectroradiometer
(MODIS), 59Modifiable areal unit problem (MAUP), 20MODIS see Moderate Resolution Imaging
SpectroradiometerMPS see Mean patch sizeMultiagent system CA (MAS/CA), 145Multicomponent analysis framework, 32Multicriteria analysis (MCA), 18 Australian rangelands, 23 ecosystem services, 34 environments, 32 framework, 23 transformation schema, 35Multicriterion decision analysis, 17–37 Australian rangelands, 23–30 concept, 19 example landscape context, 23–30 multicomponent analysis framework, 32 procedures and models, 17–37 transformation domains and methods, 21–31 transformation issues, 19–20
NNational Biomass Study (NBS), 65National threat, 88–90Natural environment categorized, 49Natural process land use, 5Natural resource manage, 44Natural resource management, 45Natural systems, 7NBS see National Biomass StudyNDVI see Normalized difference vegetation
indexNet ecosystem productivity (NEP), 28, 29 scaling of effect, 30Nonbiophysical time series, 28Non forest ecosystems of Colombia, 84
42963_Index.indd 183 11/5/07 1:51:18 PM
184 Index
Nonurban land-use maps, 147Normalized difference vegetation index (NDVI),
18
PPanel approach, 102–103 LULCC, 115Paneled pattern metrics approach, 115 construction, 104 landscape dynamism, 112–114 research, 103Parque Nacional (PN) Carrasco, 123Pattern-led analysis, 21Pattern metric change, 110Pattern metrics extension, 103PCA see Principle components analysisPCP see Percent correctly predictedPCT see Percentage landscapePearl River Delta (PRD), 150–153Percentage landscape (PCT), 112Percent correctly predicted (PCP), 148Plot exhaustion, 133PN see Parque Nacional CarrascoPolicy decision making, 19Policy makers preferences, 144Policy making local levels, 142 national levels, 142 process, 143Policy relevant land use change models, 142Population Asia, 139 Colombia, 83 growth, 65 Uganda, 65PRD see Pearl River DeltaPrinciple components analysis (PCA), 29Private land management decisions, 10–11Process-led analysis, 21Progressive deforestation, 125Pseudo-Bayesian model, 148Public land management issues, 10–11
RRainfall in Uganda, 76Regional scale mapping land use change importance, 46 land use change in Murray region, 46–47 land use change types, 46Remote sensing, 169–170Rice agriculture, 109Road building, 134
SSalinity awareness, 57Salinity discharge, 51 groundwater flow systems, 48–49 maps, 58 sites, 48–49Scale-dependent effects, 20Scale issues, 8Scale-pattern-process, 99Seasonality impacts on landscape, 114Shenzhen nonurban area predictions, 155 urban area predictions, 155Shuttle Radar Topographic Mission (SRTM), 70Social and biophysical integration, 44–59 biophysical data and land mapping
correspondence, 48–53 future directions, 59 implications, 54–58 regional scale mapping, 45Social data integration, 43–60 land use science, 54–59Social processes, 5Social statistics, 21Societal benefits, 8SOI see Southern oscillation indexSoil fertility Caribbean, 88 Magdalena, 88Soil salinity, 105Southern oscillation index (SOI), 26Space and time change dynamics, 5–6Space-time domain, 17Spatial auto correlation, 71Spatial context Uganda, 71 Western Uganda deforestation, 76Spatial interrelationships, 25Spatially applied multicriterion analysis, 18Spatial methods analysis, 50 approach, 50 context, 50 land use, 50–53Spatial patterns Australian rangelands, 25–26 forest conversion, 85 forest fragmentation, 127Spatial relationships Australian rangelands, 25–26 forest fragmentation, 123Spatial structure, 102Spatiotemporal inputs to rangeland MCA, 30–31Spatiotemporal multicriteria frameworks, 22
42963_Index.indd 184 11/5/07 1:51:19 PM
Index 185
SRTM see Shuttle Radar Topographic MissionSynthesis, comparative analysis and prospect,
163–173
TTEL see Total edge lengthTemporal dynamics, 6Temporal patterns and influences, 26–28Temporal relationships animal impact, 31 forest fragmentation, 123Temporal signals, 31Thailand, 108Thai testing grounds, 104–106Thematic Mapper (TM) images, 69Time-related factors, 6Time-series land use mapping, 47TM see Thematic Mapper imagesTotal edge length (TEL), 131Trani metrics, 130Transdisciplinary approaches, 7Transformation domains and methods, 21–31Tropical deforestation, 5, 63, 67Tropical forest landscapes, 96Tropical forests fragmentation, 120 humidity, 90Tropical lowlands, 122
UUganda background, 64–66 community wealth, 75 cumulative deforestation, 73 deforestation, 72, 77 distribution of deforestation, 66 economic development, 67 economic growth rate, 65 environmental crisis, 76 FAO, 66 forest energy, 68 forestry sector, 65–66 institutional context, 71 land use change, 77 migrants, 76 model results of deforestation, 75 population growth, 65 rainfall, 76 spatial context, 71ULCM see Urban land use change modelUNEP see United Nations Environmental ProgramUnited Nations Environmental Program (UNEP),
69
Universal Transverse Mercator (UTM), 70Urban areas predicted probability of change,
153–154Urban cellular automata models, 145Urban land use change China, 150 data sparse environments, 147 decision-making clientele, 145 developing world cities, 139–158 issues, 140 issues of theories, 146 modeling, 140–147 policy evaluation, 146–147 policy making, 142–143 policy relevant model, 146 threshold predictions, 154 uncertainty reduction, 157 zoning, 145Urban land use change model (ULCM), 140 accuracy issue, 157 probability of threshold application, 157 scientific approaches, 145 scientific disciplines, 145 usefulness, 142Urban land-use maps, 147Urban populations in Asia, 139UTM see Universal Transverse Mercator
VVegetation types, 36von Thünen development process, 73
WWater and soil salinity, 105Watershed-based approach, 44Western Uganda deforestation, 63–78 background, 64–65 conceptual framework, 68 data and methods, 69–71 data sources, 69 defined, 66 descriptive statistics, 72–73 econometric model, 70 econometric results, 74–76 good or bad, 67 institutional context, 76 methodological issues, 71 results, 72–76 socioeconomic context, 75 spatial context, 76World’s urban population, 139 distribution, 141
42963_Index.indd 185 11/5/07 1:51:19 PM
42963_Index.indd 186 11/5/07 1:51:19 PM
(a) (b)
(c) (d)
PCA classes12345678910111213141516171819202122
St. Dev. NEP0.004 – 0.0270.027 – 0.050.05 – 0.0730.073 – 0.0960.096 – 0.1180.118 – 0.1410.141 – 0.1640.164 – 0.1870.187 – 0.210.21 – 0.233
NEP 8593< –2–2 – –1–1 ––0.5–0.5 – –0.2–0.2 – 00 – 0.20.2 – 0.50.5 – 1> 1
NEP 9400< –2–2 – –1–1 ––0.5–0.5 – –0.2–0.2 – 00 – 0.20.2 – 0.50.5 – 1> 1
Figure 2.4 (a) The spatial pattern of temporal signals may be grouped by applying prin-cipal components analysis to the time series and creating a classification based on the major principal components. (b) The temporal metrics may be calculated on the spatial times series to create maps of, for example, standard deviation of net eco system productivity (NEP). (c) A running integration, trend, or wavelet analysis may define periods of distinct behavior in the time series that can then be summarized by metrics such as an integral of NEP for periods of decline and increase. Shown for 1985–1993 and 1994–2000 here.
42963_Art_Color.indd 2 11/14/07 3:40:20 PM
(a) (b)
(c) (d)
Cattle distribution
VRD Vegetation polygonsGrassland classes Other forest and bush Sorghum woodland – low SUR Sehima woodland – moderate SUR Triodia woodland – low utilisation potential Dicanthium grassland – high SUR Triodia grassland – low utilisation potential Mitchell grassland – very high SUR Aristida woodland – Imoderate SUR Not rangeland
VRDNo. threatened birds 0 1 2 3 4
VRDMines 0 1 2 5
VRDPFE ($) NA –1 0 1 2 3 4
CattleNot grazed
Water pointsVictoria RiverDistrictRoads/Rivers
Figure 2.5 Temporal signals are usually based on biophysical or human phenomena that operate at a large scale (e.g., climate, interest rates). Demographic changes at a fine scale may have scale limitations due to level of aggregation in reporting. Temporal signals and indica-tors are filtered by spatial variation. The Victoria River District in the Northern Territory of Australia is highly productive. (a) Cattle are distributed of freehold-leasehold land but confined by water points. (b) Both productivity and ecological impact vary with vegetation type, which is associated with soils, topography, and rainfall gradient. (c) Costs are low and enterprises are profitable but the increment is small on a per hectare basis. (d) Mining with major physical disturbance occurs sporadically across the area. There are threatened bird species in the region and these may be ground nesting and impacted by grazing.
42963_Art_Color.indd 3 11/14/07 3:40:22 PM
Pote
ntial
pro
duct
ivity
for g
razin
gRa
infa
ll reli
abili
tyRa
infa
ll rel.
(w/s
p)
Rain
fall r
el. (a
nn)
Fora
ge p
oten
tial
Gro
wth
Folia
ge p
roj c
over
Soil n
utrie
nt
Acce
ssib
ility
(ARI
A)
Soil P
Soil N
Soil c
arbo
nN
DVI
mea
nN
PP m
ean
soil_
fert
Fig
ur
e 2.
7 C
ogni
tive
map
ping
int
erfa
ce s
uita
ble
for
com
bina
tion
of
dive
rse
spat
iote
mpo
ral m
etri
cs, i
ndic
es,
and
data
laye
rs d
escr
ibin
g m
eani
ngfu
l pro
pert
ies
of a
sys
tem
und
er a
naly
sis.
Thi
s ex
ampl
e sh
ows
the
cons
truc
tion
of
a c
ompo
site
inde
x to
rep
rese
nt p
oten
tial
gra
zing
pro
duct
ivit
y fr
om r
ange
land
s.46
,58
42963_Art_Color.indd 4 11/14/07 3:40:23 PM
Figure 3.1 Land use change in the Barmera, Berri, and Renmark areas of South Australia.
Statistical Local AreasIrrigated horticulture first mapped prior to 1990 (mapped in 1988 for SA)Irrigated horticulture first mapped in 1995 and between 1990–1995Irrigated horticulture first mapped between 1995–1999Irrigated horticulture first mapped in 2001Irrigated horticulture first mapped in 2003
Irrigated horticulture data provided bySA Department of Environment and Heritage
Loxton
BerriBarmera
Renmark
42963_Art_Color.indd 5 11/14/07 3:40:25 PM
N
S
W E
Kilometers
Land managers who did not identify salinity
0 – 0.0456258270.045625827 – 0.0912516530.091251653 – 0.1368774800.136877480 – 0.1825033070.182503307 – 0.2281291340.228129134 – 0.2737549600.273754960 – 0.3193807870.319380787 – 0.3650066140.365006614 – 0.4106324400.410632440 – 0.456258267
Glenelg Hopkins regionNo (land manager identified salinity)Yes (land manager identified salinity)Salinity discharge sites
Land managers who identified salinity
Std. Dev = 6674.11Mean = 6369N = 702.00
Std. Dev = 4146.41Mean = 2536N = 290.00
36,000 – 38,000
32,000 – 34,000
28,000 – 30,000
24,000 – 26,000
20,000 – 22,000
16,000 – 18,000
12,000 – 14,000
8,000 – 10,000
4,000 – 6,000
0 – 2,000
36,000 – 38,000
32,000 – 34,000
28,000 – 30,000
24,000 – 26,000
20,000 – 22,000
16,000 – 18,000
12,000 – 14,000
8,000 – 10,000
4,000 – 6,000
0 – 2,000
Distance to nearest salinity discharge site (meters)
Distance to nearest salinity discharge site (meters)
Freq
uenc
y
0 30 60 120
300
200
100
0
Freq
uenc
y
300
200
100
0
Distance to nearest salinity discharge site<VALUE>
Legend
Figure 3.3 Land managers’ perception of salinity and mapped salinity discharge sites.
42963_Art_Color.indd 6 11/14/07 3:40:27 PM
N
S
W E
Kilometers
0 – 0.0124435330.012443533 – 0.0248870660.024887066 – 0.0373305980.037330598 – 0.0497741310.049774131 – 0.0622176640.062217664 – 0.0746611970.074661197 – 0.0871047300.087104730 – 0.0995482620.099548262 – 0.1119917950.111991795 – 0.124435328
Glenelg Hopkins regionFarmerNon–farmerConservation of natural environments
Std. Dev = 1595.02Mean = 1951N = 640.00
Std. Dev = 1249.48Mean = 1276N = 355.00
8000 – 8500
7000 – 7500
6000 – 6500
5000 – 5500
4000 – 4500
3000 – 3500
2000 – 2500
1000 – 1500
0 – 500
8000 – 8500
7000 – 7500
6000 – 6500
5000 – 5500
4000 – 4500
3000 – 3500
2000 – 2500
1000 – 1500
0 – 500
Distance to nearest area of high conservation value (meters)
Distance to nearest area of high conservation value (meters)
Non–farmer
Farmer
Freq
uenc
y
0 30 60 120
Freq
uenc
y
140
120
100
80
60
40
20
0
140
120
100
80
60
40
20
0
Distance to nearest area of high conser vation value<VALUE>
Legend
Figure 3.4 Land managers who manage properties near areas of high conservation value.
42963_Art_Color.indd 7 11/14/07 3:40:30 PM
Fig
ur
e 5.
3 P
redi
cted
de
fore
stat
ion
hot
spot
s ob
tain
ed
by
com
bini
ng
area
s pr
edic
ted
to
have
th
e hi
ghes
t pr
obab
ilit
y of
fo
rest
con
vers
ion
(>70
%)
from
the
bes
t m
odel
(th
e re
gion
-spe
cific
cla
ssifi
ca-
tion
tre
e) w
ith
the
area
s w
ith
grea
ter t
han
2% r
ural
po
pula
tion
gr
owth
ra
te
(198
5–19
93).
Red
dep
icts
th
e de
fore
stat
ion
hot s
pots
(a
reas
wit
h >7
0% p
roba
-bi
lity
of
fore
st c
onve
rsio
n an
d >2
% r
ural
pop
ulat
ion
grow
th).
Ora
nge
and
red
depi
ct
area
s w
ith
>70%
pr
obab
ilit
y of
for
est
con-
vers
ion.
G
reen
de
pict
s fo
rest
ed a
reas
, gr
ay r
ep-
rese
nts
clea
red
fore
sted
ar
eas,
an
d w
hite
re
pre-
sent
s no
nfor
este
d ar
eas.
W
hite
cir
cled
are
as i
ndi-
cate
cur
rent
hot
spo
ts o
f de
fore
stat
ion,
whi
ch a
re a
lso
area
s of h
igh-
valu
e bi
o div
ersi
ty v
alue
: (1)
Qui
bdó-
Tri
bugá
, (2
) Fa
rallo
nes-
Mic
ay, (
3) P
atía
-Mir
a, (
4) F
ragu
a-Pa
tasc
oy , (
5) A
lto D
uda-
Gua
yabe
ro,
(6)
Mac
aren
a, (
7) G
uavi
are,
and
(8)
Per
ijá. B
lack
line
is t
he A
ndea
n re
gion
, and
lig
ht
gree
n li
nes
are
nati
onal
par
ks. (
From
Ett
er e
t al.50
Wit
h pe
rmis
sion
.)
N S
WE
Kilo
met
ers
Majo
r roa
dAl
l –yea
r roa
dW
ater
bod
y
Yes
No
050
2510
015
0Lege
nd
Tanz
ania
Rwan
da
Keny
a
Suda
n
Cong
o, D
RC
Kam
pala
North
Central
East
West
Def
ores
tatio
n
Fig
ur
e 4.
1 U
gand
a st
udy
area
sho
win
g th
e di
stri
buti
on o
f de
fore
stat
ion
wit
hin
the
wes
tern
reg
ion
of th
e co
untr
y.
N
Km50
025
001
2
3
45
6
7
8
42963_Art_Color.indd 8 11/14/07 3:40:33 PM
Fore
st co
ver z
one (
%)
Clea
red
1989
(a)
(b)
1996
1999
2002
Fore
st
0–10
10–2
020
–30
30–4
040
–50
50–6
060
–70
70–8
080
–90
90–1
00
Fig
ur
e 5.
4 Fo
rest
map
s of
the
col
oniz
atio
n fr
ont f
or e
ach
stud
y da
te: (
a) e
xten
t of
fore
st c
over
(bl
ack
= fo
rest
); an
d (b
) pe
rcen
tage
fo
rest
cov
er a
t 10%
incr
emen
t zon
es. (
From
Ett
er e
t al.39
With
per
mis
sion
.)
42963_Art_Color.indd 9 11/14/07 3:40:34 PM
LEG
END
Rapi
d de
fore
statio
nRa
pid
fore
st re
gene
ratio
nM
inor
chan
ges
Tow
nsRi
vers
Road
s
1999
–200
2 m
ilita
ryex
clusio
n zo
ne
1989
–199
619
96–1
999
1999
–200
2
Fig
ur
e 5.
5 Sp
atia
l lo
cati
on o
f th
e lo
cal
hot
spot
s of
def
ores
tati
on (
red)
and
reg
ener
atio
n (g
reen
) fo
r th
e th
ree
tim
e pe
riod
s of
st
udy.
(Fr
om E
tter
et a
l.39 W
ith
perm
issi
on.)
42963_Art_Color.indd 10 11/14/07 3:40:36 PM
IJI = 14.2MPS = 8.1
IJI = 17.9MPS = 4.7
IJI = 23.7MPS = 2.2
IJI = 19.2MPS = 3.1
(2)
(2a)
(2b)(3a)
(3)
(1)
Figure 6.1 The panel process, conducted at both the pixel and patch levels: (1) four multi-spectral satellite images are each categorized into a thematic LULC classification; (2) pattern metrics are run on each of the four LULC classifications, each producing a set of patch, class, and landscape statistics (here the interspersion/juxtaposition index [IJI] and mean patch size [MPS] are shown) as well as an output image of the delineated patches; (2a) pattern metric output for each of the four times is used to calculate three piecemeal change maps for each pattern metric and each consecutive pair of images (e.g., showing fluctuations in IJI or MPS between two time periods) as per Crews-Meyer11,27; (2b) three pattern change maps are stacked into one panel of all structural change for each given metric (e.g., showing fluctuation in IJI or MPS through all time periods) as per Crews-Meyer11,27; (3) three thematic change maps are created for each of the time periods represented by the four classifications; (3a) the three thematic change maps are stacked to represent the full record of all thematic change across the four classifications as per Crews-Meyer.3
42963_Art_Color.indd 11 11/14/07 3:40:37 PM
Fig
ur
e 6.
3 (a
) L
UL
C in
the
grea
ter
stud
y ar
ea in
the
1972
/197
3 w
ater
yea
r; (
b) 1
985;
and
(c)
1997
.
42963_Art_Color.indd 12 11/14/07 3:40:38 PM
(a)
(b)
(c)
Figure 6.6 (a) The change in configuration from 1972/1973 to 1975/1976, reveal-ing that the entire subset area has experienced a greater than 10% increase in interspersion/juxtaposition index (IJI) scores (due to increased fragmentation and concomitant interdigitation). Note that most of the area experienced the same type of change. (b) Illustration of a different trend between 1975/1976 and 1979, whereby increases, decreases, and relative stability in IJI vary spatially. More upland areas (most central in the subset) experienced a consolidation on the landscape, while peripheral areas remain relatively stable in terms of configuration with notable exceptions on the southeastern perimeter. (c) Illustration of the continued spatial heterogeneity in IJI, with lowland/peripheral areas under going continued fragmen-tation, while the less accessible, upland areas appear to have leveled off in terms of larger-scale fragmentation or consolidation but continue to experience small pockets of fragmentation throughout.
42963_Art_Color.indd 13 11/14/07 3:40:39 PM
-180
-60
Robi
nson
Pro
ject
ion
Cent
ral M
erid
ian
0.00
Sour
ce: E
SRI D
ata &
Map
s CD
Kilo
met
ers
0
1,45
0
2
,500
5,500
8,70
0
11,6
00
C
ount
ry B
ound
ary
POPU
LATI
ON
7500
00 –
1000
000
1000
001
– 300
0000
3000
001
– 500
0000
5000
001
– 100
0000
0
1000
0001
– 23
6200
00
Anta
rctic
Circ
le
Trop
ic of
Cap
ricor
n
Equa
tor
Pacifi
c Oce
an
Trop
ic of
Can
cer
Indi
an O
cean
Pacifi
c Oce
an
Atla
ntic
Oce
an
–180
–160
–140
–120
–100
–80
–60
–40
–20
020
4060
8010
012
014
016
018
0
–160
–140
–120
–100
–80
–60
–40
–20
–60
–40
–20
20
40
60
80
0
–40
–2020
40
60
80
0
020
4060
8010
012
014
016
018
0
Arcti
c Cir
cle
Fig
ur
e 8.
1 Po
pula
tion
dis
trib
utio
n of
the
wor
ld’s
urb
an a
gglo
mer
atio
ns w
ith
750,
000
peop
le o
r m
ore
arou
nd y
ear
2000
.
42963_Art_Color.indd 14 11/14/07 3:40:48 PM
Kilometers6030150
KilometersFoshan study areaGuangzhou study areaShenzhen study areaGraticule (5 degrees)Province BoundariesGuangdong Province
25012562.50
Urban land use Urban land 1988 Growth 1988–1996
20N
25N
110E
115E
SOUTH CHINA SEA
GUANGXI PROVINCEGUANGDONG PROVINCE
HUNAN PROVINCE JIANGXI PROVINCEFUJIAN PROVINCE
PEARL RIVER DELTA
N
Figure 8.2 The Pearl River Delta in southeast China and the Shenzhen, Foshan, and Guangzhou study areas (urban land in 1988, new urban land between 1988 and 1996).
42963_Art_Color.indd 15 11/14/07 3:40:51 PM
0.6
0.5
0.4
0.3
0.2
0.1
0 0.12
0.1 0.08
0.06
0.04
0.02
0
50 100
150
200
250
300
350
400 50 100
150
200
250
300
350
400
100
200
300
400
500
600
700
100
200
300
400
500
600
700
Fig
ur
e 8.
5 P
redi
cted
pr
obab
ility
of
ch
ange
to
ur
ban
area
s bet
wee
n 20
04 a
nd 2
012
and
stan
dard
dev
i-at
ion
of p
ixel
pre
dict
ed p
roba
bilit
ies
for
Gua
ngzh
ou
(60
m r
esol
utio
n; v
alue
s in
per
cent
age
poin
ts).
0.6
0.5
0.4
0.3
0.2
0.1
0 0.12
0.1 0.08
0.06
0.04
0.02
0
50 100
150
200
250
300
350
400 50 100
150
200
250
300
350
400
100
200
300
400
500
600
700
100
200
300
400
500
600
700
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
50 100
150
200
250
300
350
400
450
500
550 50 100
150
200
250
300
350
400
450
500
550
100
200
300
400
500
600
100
200
300
400
500
600
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
50 100
150
200
250
300
350
400
450
500
550 50 100
150
200
250
300
350
400
450
500
550
100
200
300
400
500
600
100
200
300
400
500
600
Fig
ur
e 8.
4 P
redi
cted
pro
babi
lity
of c
hang
e to
urb
an a
reas
bet
wee
n 20
04 a
nd 2
012
and
stan
dard
dev
iatio
n of
pix
el p
redi
cted
pro
babi
li-
ties
for
Fosh
an (
30 m
res
olut
ion;
val
ues
in p
er-
cent
age
poin
ts).
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
100
200
300
400
500
600
100
200
300
400
500
600
0.12
0.1
0.08
0.06
0.04
0.02
0
100
200
300
400
500
600
700
800
900
100
200
300
400
500
600
700
800
900
Fig
ur
e 8.
3 P
redi
cted
pro
babi
lity
of c
hang
e to
urb
an a
reas
bet
wee
n 20
04 a
nd 2
012
and
stan
-da
rd d
evia
tion
of p
ixel
pre
dict
ed p
roba
bilit
ies
for
Shen
zhen
(60
m r
esol
utio
n; v
alue
s in
per
-ce
ntag
e po
ints
).
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
100
200
300
400
500
600
100
200
300
400
500
600
0.12
0.1
0.08
0.06
0.04
0.02
0
100
200
300
400
500
600
700
800
900
100
200
300
400
500
600
700
800
900
42963_Art_Color.indd 16 11/14/07 3:41:03 PM
DecisionMaking
EcosystemsServices
Distu
rbance
How
shou
ld th
e la
ndsc
ape
be d
escr
ibed
?
How
doe
s the
land
scap
e op
erat
e?
Is th
e cu
rren
t lan
dsca
pe w
orki
ng w
ell?
How
mig
ht th
e la
ndsc
ape
be a
ltere
d?
Wha
t pre
dict
able
diff
eren
ces m
ight
the
chan
ges c
ause
s?
How
shou
ld th
e la
ndsc
ape
be ch
ange
d?
Dat
a In
form
atio
n Cu
ltura
l kno
wle
dge
Repr
esen
tatio
n M
odel
s
Proc
ess M
odel
s
Eval
uatio
n M
odel
s
Chan
ge M
odel
s
Impa
ct M
odel
s
Dec
isio
n M
odel
s
Ecol
ogic
alSy
stem
sBi
ogeo
chem
istry
Biod
iversi
tyW
ater
Air
Soil
Soci
alSy
stem
s
T1. D
ynam
ics of
land
–sys
tem
sT2
. Con
sequ
ence
s of l
and–
syste
m ch
ange
T3
. Int
egra
ting a
naly
sis an
d m
odell
ing f
or la
nd su
stain
abili
ty
Land
Use
&M
anag
emen
t
Popu
latio
nSo
cial/E
cono
mic
struc
ture
Polit
ical/I
nstit
utio
nal
regim
esCu
lture
Tech
nolo
gy
Earth
Syste
m
Land
Syste
ms
T2.
1
T2.
2
T2.4
T3.1
Crit
ical
pat
hway
s of c
hang
eT3
.2 V
ulne
rabi
lity a
nd re
silien
ce of
Lan
d–Sy
stem
sT3
.3 E
ffect
ive go
vern
ance
for s
usta
inab
ility
T2.
3
T1.
1
T1.
2
T1.3
Nat
ural
Cultu
ral
Soci
o –ec
onom
ic
Soci
alIn
stitu
tions
Soci
alCy
cles
Soci
alO
rder
Cri
tical
Res
ourc
esH
uman
Soc
ial S
yste
m
Nat
ural
Sys
tem
Hum
an S
yste
m
(a)
(b)
(c)
(d)
Fig
ur
e 9.
1 A
var
iety
of
inte
grat
ing
fram
ewor
ks t
hat s
eek
inte
rdis
cipl
inar
y de
fini
tion
and
foc
us o
n ke
y qu
esti
ons
wit
hin
the
broa
d sc
ope
of m
anag
e-m
ent o
f la
nd u
se c
hang
e in
cou
pled
hum
an e
nvir
onm
ent s
yste
ms.
(a)
Ana
lyti
cal f
ram
ewor
k fo
r th
e G
loba
l Lan
d P
roje
ct o
f IG
BP
and
IHD
P. (
From
GL
P,
2005
, w
ith
perm
issi
on.)
(b)
The
Hum
an E
cosy
stem
Mod
el.
(Fro
m M
achl
is e
t al
.,8 19
97,
wit
h pe
rmis
sion
.) (c
) T
he U
.S.
Nat
iona
l Sc
ienc
e Fo
unda
tion
pr
ogra
m o
n B
ioco
mpl
exit
y in
the
Env
iron
men
t (h
ttp:
//ww
w.n
sf.g
ov/g
eo/e
re/e
rew
eb/f
und-
bioc
ompl
exit
y.cf
m);
(d)
The
Lan
dsca
pe D
esig
n R
esea
rch
Fram
ewor
k. (
From
Ste
initz
et a
l.,9
2003
, wit
h pe
rmis
sion
.)
42963_Art_Color.indd 17 11/14/07 3:41:19 PM