University of Groningen Determinants of the spatial ...

University of Groningen

Determinants of the spatial distribution of tree species in a neotropical forestGarzon-López, Carol

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2013

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Garzon-López, C. (2013). Determinants of the spatial distribution of tree species in a neotropical forest. s.n.

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne-amendment.

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 22-04-2022

https://research.rug.nl/en/publications/a4f22f8c-5ce2-4679-b8e9-12023f923879

Carol X. Garzon-LopezPatrick A. JansenStephanie A. BohlmanAlejandro OrdoñezHan Olff

CHAPTER3Effects of Sampling Scale on Patternsof Habitat Association in Tropical Trees

Journal of Vegetation Science. In press

Chapter 3

40

Studies of habitat associations in tropical forest trees have yieldedhighly variable results in terms of the percentage of species withsignificant habitat associations. Here, we determine how species-specific spatial aggregation characteristics and sampling scale affectthe outcome of plant-habitat association studies in a tropical forest.Using distribution maps of canopy-statured individuals, we evaluatedpatterns of habitat association in five tropical tree species on BarroColorado Island, across a wide range of sampling scales (from 50 –1600 ha). We used point pattern analysis to evaluate the scale-dependency of species’ clumping patterns and the association ofspecies distributions with important environmental variables (forestage, topography and geological formation). Analyses were performedwith and without the inclusion of species-specific aggregation charac-teristics. We found that the percentage of significant habitat associa-tions varied dramatically with spatial scale, but tended to increasewith plot size. Including dispersal constraints in the model of speciesdistributions decreased the number of significant associations. Esti-mates of clumping characteristics (e.g. clump size) had high varianceswithin and among spatial scales. At smaller plot sizes, there was largevariation in the occurrence and even sign (positive or negative) ofsignificant associations depending on plot location. Our studysuggests that patterns of habitat association (and hence conclusionson the importance of niche vs. neutral processes) are strongly affectedby species-specific aggregation characteristics and sampling scale.Explicit consideration of both is critical for studies of habitat associa-tion and the main processes that structure communities of tropicaltrees.

Abstract

Scale-dependence of plant-habitat associations

41

Los estudios de asociaciones planta-hábitat en bosques tropicales hangenerado resultados variables en términos del porcentaje de especiescon asociaciones significativas. En este estudio determinamos cómolas características de agregación espacial, específicas para cadaespecie, y la escala afectan el resultado del estudio de asociacionesplanta-hábitat en bosques tropicales. Evaluamos patrones de asocia-ción en 5 especies de árboles en la Isla de Barro Colorado, en un rangoamplio de escalas de muestreo (desde 50 hasta 1600 ha), usandomapas de distribución de individuos que han alcanzado el dosel.Usamos análisis de patrones espaciales para evaluar el efecto de laescala en los patrones de agregación de las especies y su asociacióncon variables ambientales (edad del bosque, topografía y formacionesgeológicas). Los análisis se llevaron a cabo incluyendo y excluyendolas características de agregación específicas. En este estudio encon-tramos que el porcentaje de asociaciones significativas varía de formadrásticas con la escala espacial, pero tiende a aumentar con elaumento en el tamaño del cuadrante. El número de asociacionessignificativas disminuyó al incluir las limitaciones por la estrategia dedispersión en el modelo de distribución de especies. Los estimativosde las características de agregación (e.g. tamaño) variaron bastantedentro de muestras de la misma escala y entre escalas diferentes. Encuadrantes de menor tamaño hubo amplia variación en la presencia ytipo (negativo o positivo) de asociación dependiendo de la localizacióndel cuadrante. Nuestro estudio sugiere que los patrones de asociaciónplanta-hábitat (y, por consiguiente, las conclusiones sobre la impor-tancia de los procesos neutrales y de nicho) son fuertemente afectadospor las características de agregación de cada especie y la escala demuestreo. La inclusión explícita de estas dos es crítica en los estudiosde asociaciones planta-hábitat y de los principales procesos queestructuran las comunidades de árboles en bosques tropicales.

Resumen

Introduction

Numerous hypotheses have been proposed to explain the high level of speciescoexistence that maintains the high diversity of tropical forests (Wright 2002). Oneof the most common explanations is niche theory, which argues that species adap-tation to specific conditions separates the distribution of different species alongenvironmental gradients in space and time (Whittaker et al. 1975, Hubbell 2001,Wright 2002). Thus, each species occupies a specific niche formed by a combina-tion of abiotic conditions (light, soil factors) that facilitate their establishment andsurvival (Pulliam 2000). However, recent studies indicate that dispersal limitationis also an important determinant of the spatial distribution of tree species (e.g.,Svenning 2001, Vormisto et al. 2004, Svenning et al. 2006, John et al. 2007,Bohlman et al. 2008) that may promote coexistence of species. The degree to whichdispersal limitation versus habitat specialization drives species distributions iscurrently a major point of scientific discussion (Wiegand et al. 2007, Lin et al.2011).

A simple indicator of the importance of niche differentiation is the strength ofthe association of different species to particular habitats or environmental factors.However, studies on habitat associations in tropical forest trees have yieldedhighly variable results in terms of the percentage of species with significanthabitat associations (Table 3.1). Possible causes for this inconsistency includedifferences among studies in: (1) the environmental factors examined, whichinclude soil chemistry, topography and forest age (2) the statistical techniques usedto detect associations, which range from Mantel tests and randomization tests tomultivariate analyses (Itoh et al. 2010); (3) species-specific characteristics (e.g.,body size, dispersal strategy; Nathan & Muller-Landau 2000); (4) the spatial scaleof the study (Lam & Quattrochi 1992, Wu 2004, Cottenie 2005), which varies fromas little as 0.3 ha (Balvanera et al. 2002) to the 25 and 50 ha scale typically used forlong-term forest dynamics plots (Harms et al. 2001, Russo et al. 2005) to as muchas 158 ha (Phillips et al. 2003) (Table 3.1). Throughout this paper, we will use‘scale’ to refer to the spatial extent of the area studied, rather than the level ofdetail, resolution or ‘granularity’ of sampling (van Gemerden et al. 2005).

Spatial scale is especially important because environmental factors are hetero-geneous at different spatial scales (Whittaker et al. 2001, Kneitel & Chase 2004,Snyder & Chesson 2004, Wu & Li 2009, McGill 2010). At large scales (approxi-mately 1,000–10,000 ha), species distributions may reflect climatic gradients(Rhode 1992), whereas at small scales (<10 ha), they may rather reflect soil-relatedheterogeneity with respect to nutrient or water availability (which may bepartially driven by individual trees) and heterogeneity in canopy openness (Keddy1982, Ceccon et al. 2003). Spatial heterogeneity at intermediate scales (approxi-mately10–1000 ha) may be caused by environmental factors, such as geology,topography and historical events (e.g., forest management history, Clark et al.

Chapter 3

42

1995). Until now, no study has evaluated how the sampling scale affects the detec-tion of habitat associations.

Differences in dispersal capacity of species may also strongly affect the spatialstructure of communities and confound the detection of environmental effects(Duque et al. 2002, Gilbert & Lechowicz 2004). Seed dispersal patterns determinethe initial potential distribution of individuals, but if the habitat is also importantfor survival, the final distribution will resemble the initial dispersal pattern modi-fied by the habitat requirements of the species. To discriminate between factorsthat determine the spatial distributions of tree species in tropical forests, it isessential to incorporate environmental factors and dispersal limitations simultane-ously in appropriate statistical models (Nathan & Muller-Landau 2000).

In this paper, we consider the effect of sampling scale on the detection ofhabitat associations in the tropical moist forest of Barro Colorado Island (BCI),Panama. Four studies of habitat association have previously been conducted onBCI at different sampling scales with varying conclusions. Two studies of the BCI50-ha plot found that 64% and 29% of 171 and 75 tree species, respectively,showed significant habitat associations and that topography and soil chemistrywere important factors (Harms et al. 2001, John et al. 2007). Two studies at thescale of the whole island (Svenning et al. 2004, 2006), which sampled 32 and 7 haof forest, respectively, found that 25% and 68% of the 94 and 26 species studied,respectively, showed significant habitat associations and that forest age was animportant factor (Table 3.1). Here, we not only encompass the spatial scales ofthese previous studies but exceed the sampling scale of any previously publishedstudy. Our prediction was that the percentage of species with associations and thestrength and sign of habitat associations varies with sampling scale. We testedfour specific hypotheses: (1) the number of habitat association found for a speciesis inflated by species-specific aggregation characteristics; (2) aggregation charac-teristics of species are consistent across spatial scales, signifying a characteristicdispersal pattern for each species; (3) The variability of habitat associationpatterns tests will decrease with increasing sampling scale, and (4) the number ofhabitat associations found for a species increases with sampling scale, as largerplots encompass more environmental heterogeneity. Support for these hypotheseswould indicate that sampling scale and species-specific aggregation patterns needto be explicitly considered in tests of habitat association.


43

Chapter 3

44

Tab

le 3

.1.

Met

hods

an

d ou

tcom

es o

f pu

blis

hed

stu

dies

on

hab

itat

ass

ocia

tion

s of

tro

pica

l fo

rest

tre

es.

The

‘pe

rcen

tage

exp

lain

ed’

colu

mn

in

dica

tes

the

perc

enta

ge o

f sp

ecie

s w

ith

sign

ific

ant

habi

tat

asso

ciat

ions

. M

VA

ref

ers

to m

ulti

vari

ate

anal

ysis

. A

n X

in

the

“dis

pers

al l

imit

atio

n” c

olum

n in

di-

cate

s st

udi

es t

hat

att

empt

ed t

o ac

cou

nt

for

disp

ersa

l li

mit

atio

n d

uri

ng

the

anal

ysis

. T

he

fin

al f

our

stu

dies

(bl

ack

box)

wer

e co

ndu

cted

at

Bar

roC

olor

ado

Isla

nd, P

anam

a.

Ref

eren

ceSa

mpl

ing

N.

Ana

lysi

s%

E

xpla

nato

ry v

aria

ble

Hol

drid

geC

ount

ryD

ispe

rsal

ar

ea (h

a )

spec

ies

expl

aine

dH

ydro

.To

po.

Soil

prop

.G

eo.

syst

emlim

itat

ion

Bal

vane

ra e

t al

,0.

311

9A

NO

VA57

xT

ropi

cal

Mex

ico

2002

MVA

dry

fore

stO

livei

ra-F

ilho

0.54

55M

VA77

xm

onta

ne

Bra

zil

et a

l, 19

94fo

rest

Miy

amot

o 2

55ra

ndom

izat

ion

80x

xhe

ath

fore

stIn

done

sia

Xet

al,

2003

test

sC

osta

et

al, 2

005

2.95

87M

VA<

33

xx

trop

ical

B

razi

lra

in f

ores

tC

anno

n &

6.9

197

rand

omiz

atio

n25

xx

xlo

wla

nd

Indo

nesi

aX

Leig

hton

, 200

4te

sts

fore

sts

chi s

quar

eM

VASr

i-N

gern

yuan

g 7.

520

rand

omiz

atio

n 65

xm

onta

ne

Tha

iland

Xet

al,

2003

test

s ra

in f

ores

tPe

lissi

er e

t al

, 10

120

MVA

54x

trop

ical

Fren

ch G

uian

a20

01ra

in f

ores

tC

lark

et

al,

11.7

267

MVA

30x

trop

ical

C

osta

Ric

a19

99

rain

for

est

Dal

le e

t al

, 14

23M

VA26

xse

mid

ecdu

ous

Pana

ma

2002

moi

st f

ores

tSa

bati

er e

t al

,19

195

MVA

25x

trop

ical

Fren

ch G

uian

a19

97ra

in f

ores

tPi

tman

et

al,

36.3

829

Rab

inow

itz

15-2

5x

xtr

opic

al

Peru

1999

m

etho

d ra

in f

ores

t It

oh e

t al

, 52

267

Mon

te-c

arlo

10

0x

dipt

eroc

arp

Mal

aysi

aX

2003

perm

utat

ion

test

s


45

Ref

eren

ceSa

mpl

ing

N.

Ana

lysi

s%

E

xpla

nato

ry v

aria

ble

Hol

drid

geC

ount

ryD

ispe

rsal

ar

ea (h

a )

spec

ies

expl

aine

dH

ydro

.To

po.

Soil

prop

.G

eo.

syst

emlim

itat

ion

Rus

so e

t al

, 52

764

mor

talit

y an

d 73

xdi

pter

ocar

p M

alay

sia

X20

05gr

owth

rat

esfo

rest

Laur

ance

et

al,

65dr

yM

VA

23,3

xtr

opic

al

Bra

zil

1999

bi

omas

sra

in f

ores

tH

all e

t al

, 10

04

toru

s tr

ansl

atio

n75

xTe

rra

firm

e C

entr

al a

fric

an

X20

04

test

for

CSR

fore

stre

publ

icPh

illip

s et

al,

158,

484

9 m

ante

l cor

rela

tion

s40

x tr

opic

al

Peru

X

2003

rand

omiz

atio

n te

sts

rain

for

est

mul

tipl

e re

gres

sion

Boh

lman

et

al,

4084

9 m

ante

l cor

rela

tion

s40

x tr

opic

al

Bra

zil

X20

08ra

ndom

izat

ion

test

sra

in f

ores

tm

ulti

ple

regr

essi

onG

unat

illek

e et

al,

2512

5 to

rus

tran

slat

ion

79x

dipt

eroc

arp

Sri L

anka

X

2006

fore

st

Har

ms

et a

l,50

171

chi s

quar

e75

sem

idec

duou

s Pa

nam

aX

2001

rand

omiz

atio

n te

sts

51x

moi

st f

ores

tto

rus

tran

slat

ion

64Sv

enni

ng e

t al

, 31

,594

MVA

25-3

1 x

xx

sem

idec

duou

s Pa

nam

a20

04m

oist

for

est

Sven

ning

et

al,

7,92

26

Ord

inar

y le

ast-

68x

x se

mid

ecdu

ous

Pana

ma

X20

06sq

uare

s re

gres

sion

moi

st f

ores

tCA

RJo

hn e

t al

, 50

301

NM

DS

ordi

nati

ons

29se

mid

ecdu

ous

moi

st,

Pana

ma

2007

2522

8M

onte

-car

lo

25x

ever

gree

n m

onta

neC

olom

bia

X50

1104

sim

ulat

ion

test

s40

and

low

land

for

ests

Ecu

ador

Methods

Study siteBarro Colorado Island (BCI), Panama (9.9° N, 79.51° W) is a 1560-ha island thatbecame isolated from the surrounding mainland tropical forest during 1910–1914,when the Chagres River was dammed to form the central part of the PanamaCanal (Leigh 1999). The island is covered with tropical moist forest and has anannual rainfall of 2623 mm, with a four month wet season from December to April(Leigh 1999). BCI is home to a 50-ha forest dynamics plot, established in 1982(Condit 1998, Hubbell et al. 1999, Hubbell 2005), and the island has been the focusof many studies, including vegetation structure, geology, hydrology, soil dynamicsand tropical ecology (Losos & Leigh 2004).

The island is heterogeneous at various spatial scales and in several aspectsincluding forest age, soil type and topography. About half of the forest is approxi-mately 100 years old while the remaining is old-growth forest (Enders 1935, Leigh1999; Figure 3.2). BCI has a variety of soil types that vary systematically with thetype of underlying rock and the topography (Leigh 1999). The island hosts twomain geologic formations dating from the Oligocene, known as Bohio and Caimito(Figure 3.2). The top of the island is covered with the only non-sedimentarylithology, the Andesite flow (Johnson & Stallard 1989, Barthold et al. 2008). Theelevation ranges from 27 to 160 m above the sea level, and there is a large varia-tion in slope (Svenning et al. 2004).

Tree distributionsThe spatial distributions of canopy-statured individuals of five study species –three arborescent palms and three large-canopy tree species (Table 3.2) – across theentire island were measured from high-resolution aerial photographs, on which thecrowns of each species could be easily recognized (Garzon-Lopez et al. in press).The five species varied in size, dispersal mode and shade tolerance (Table 3.2).

The photographs were taken in April 2005 and April 2006. Flights were flownin overlapping north-south swaths at an altitude of 400 m in 2005 and 700 m in2006. In 2005, each photo on average covered 8.6 ha with a spatial resolution of0.085 m/pixel. In 2006, coverage and resolution averaged 15.9 ha and 0.114m/pixel. The aerial photographs were registered to a georeferenced Quickbirdsatellite image of BCI (DigitalGlobe, Longmont, CO, USA) from March 2004, usingthe ERDAS IMAGINE v8.7 program (Leica Geosystems, GA, USA). Features visiblein both photos and satellite image, such as flowering Tabebuia guayacan (a canopyspecies with conspicuous yellow flowers that can be readily observed in aerialphotographs and satellite images; Sanchez-Azofeifa et al. 2011), telemetry towersand large tree crowns were used as registration points for warping and georefer-encing the individual photographs to produce an island-wide orthorectifiedmosaic.

Chapter 3

46

A key for identifying the crowns of the tree species was developed based oncrown structure, pattern and colour following the criteria Trichon (2001) and veri-fied using mapped positions of trees of each species in the 50-ha plot (Garzon-Lopez et al. in press). All aerial photos were surveyed to map canopy-statured indi-viduals belonging to the five tree species. Polygons were drawn to trace thecontours of each crown, and the centroids of each crown were used to produce amap of all sampled individuals for each species. Validation was conducted using 75ha of ground-mapped distributions, including those in the 50-ha plot, to test accu-racy using an error matrix (Garzon-Lopez et al. in press). This confirmed that thedistribution maps reflected the actual distributions of adult trees of each species(Garzon-Lopez et al. in press).

Environmental factorsFive environmental factors (Table 3.3) commonly used in habitat-distribution asso-ciation studies (cf. Table 3.1) were used to test for associations with the distribu-tions of canopy individuals. Of these, geological formation (mapped at 1:15000scale) has a large impact on soil nutrient availability, as well as other soil relatedcharacteristics (e.g., particle size, water retention capacity, pH); Forest age indi-cates different stages of succession, canopy density and complexity, and lianacoverage, and; Slope and elevation (mapped at a 1:25000 scale) as well as distanceto shore serve as proxies of soil moisture, soil runoff and other hydrologicalvariables. For this analysis, we combined slope, elevation and distance to shoreinto one topographic variable with five habitat types (Table 3.4). The remainingvariables (i.e., forest age and geological formation) are also discrete orthogonalvariables, so they were analysed as separate factors.


47

Table 3.2. Characteristics of canopy tree species studied. Crown diameter is the mean sun-exposedcrown diameter calculated from the aerial photos.

Species Attalea Astrocaryum Jacaranda Tabebuia Oenocarpus butyracea standleyanum copaia guayacan mapora

Dispersal mode Animal Animal Wind Wind AnimalHeight (m) 30 20 30 30 20Characteristics palm tree palm tree Deciduous Deciduous palm treeShade tolerance Low Medium Low Low HighIndividuals mapped 3121 2456 977 688 7555Seed size (cm) 4 ! 2 3 ! 2 2 ! 1.8 3 ! 1 2 ! 2Crown diameter (m) 8.9 6.4 15.1 17.3 3.8Density (ha-1) 2.05 1.73 0.67 0.45 5.41

Determining the level of spatial aggregation and plant-habitat associationsWe used six different sizes of rectangular plots representing increasing samplescales (i.e., 50, 100, 200, 400 and 800 ha, and the entire island -1560 ha). Thesmaller sample sizes (50–400 ha) were replicated by placing polygons at randomlygenerated points. Overlap among equal-sized samples was avoided. The number ofreplicates was 10 for 50 and 100 ha plots, 5 for 200 ha plots, 3 for 400 ha and 1 forthe 800 ha plot and whole-island plot. Areas with homogeneous environmentalvariables did not allow the analysis of habitat association thus reducing thesample sizes for some plot sizes (Appendix S2). Each replicate was evaluated as anindependent sample.

Spatial aggregationWe used Ripley’s K(d) function (representing the aggregation of trees at distance d)as a measure of spatial aggregation (Ripley 1976). To observe deviations from arandom distribution, the K(d) function was transformed to L(d)/d = (√K(d)/π-d)/d

Chapter 3

48

Table 3.3. Environmental variables and levels within each variable.

Variable Levels Source

Geological formation - Bohio volcanic Barthold et al. 2008 - Caimito volcanic- Caimitio marine- Andesite

Forest age - Secondary forest (after 1880) Enders 1935- Old-growth forest (before 1880)

Slope - Flat (0°–5°) Johnson & Stallard 1989(proxy of soil water drainage) - Shallow (5°–10°)

- Steep ( ≥10°)

Distance from the shore - Shore (land mass ≤150m from ArcGIS – Quickbird derived the island edge)

- Inland (>150 m from the shore)

Elevation - Low (≤63m ASL) Johnson & Stallard 1989- High (63–128m ASL)

Table 3.4. Levels composing the topography variable.

Habitat Distance to shore (m) Slope ( º ) Elevation (m)

Shore 0 to 150 - -Flat >150 < 5 < 63Ridge >150 < 5 63 to 128Shallow >150 5 to 10 -Steep >150 > 10 -

and plotted against distance at the whole island (Besag & Diggle 1977). Wemeasured aggregation for the entire island, and at each of the seven smaller spatialscales, in order to determine the sampling scale effect on estimates of K(d).Following Plotkin et al. (2000), clump size (!) and clump density ("), were esti-mated from K(d) by assuming that its distribution follows a Poisson point processdescribed by the function:

K(d)PCP = # d2 + "-1 (1 – exp ( – d2)) Eq(1)

4!2

The parameters ! and r were estimated at each spatial scale using the ‘pcp’ func-tion in the R-package ‘splancs’. Clump size (!) is expressed in meters and clumpdensity (r) in clumps per m2.

SimulationsTo determine the effect of the environment on the observed spatial patterns, wesimulated the spatial distributions of adult trees based on the aggregation charac-teristics (! and r) of each species. We used two simulation approaches (i.e.,complete spatial randomness - CSR and a Poisson cluster process - PCP) to deter-mine if habitat associations were significant even when clumping patterns weretaken into account. First, we simulated 1,000 distributions for each one of thesubsamples with complete spatial randomness (CSR), which is determined only bythe number of trees. Second, we generated 1,000 spatial clumped distributions ateach spatial scale, assuming a Poisson cluster process (PCP) in which spatialpatterns are determined by species-specific clumping parameters determined fromthe Ripley’s K analysis. For this, we implemented Plotkin et al. (2000) approach,using both the clump density (r) and size (!) estimated form K(d) at each scale asinput parameters. PCP’s were simulated by placing a number of clusters equal to"i*A +0.5, where A corresponds to the area of the sample for the ith species. Thenumber of stems per cluster was given by ni /"i*A, where n corresponds to thetotal number of individuals in species i. Visual inspection of simulation outputsbased on PCP resembled the observed distributions more closely than simulationswith CSR, showing the importance of dispersal limitation as a basic determinantof species distributions (Figure 3.1).

Habitat associationsTo test for plant-habitat associations, we used the Gamma test (Plotkin et al. 2000),which is a modified version of a Chi-square test. A standard Chi-square test is notsuitable in this case due to its assumption of independence in the locations ofconspecific trees. The Gamma statistic is defined as the proportion of trees foundon habitat type A, which is defined as (1/ni,) niA; where ni is the number of trees ofspecies i in the sample, and niA is the number of trees of species i in the sampleoccurring on environmental factor A. To test if the species were significantly asso-


49

ciated with the environmental factors we studied, we compared observed Gammavalues to those generated in 1000 PCP and CSR simulations. If observed Gammavalues fell outside the 95% confidence interval of simulated Gamma values,habitat associations were consider significant and either negative (ObservedGamma

< SimulatedGamma) or positive (ObservedGamma > SimulatedGamma). We performedall the analyses using the ‘Splancs’ and ‘grdevices’ package in the R program (RDevelopment Core Team 2005)

Results

We mapped a total of 15,209 crowns of the five tree species. The density rangedfrom 0.45 (T. guayacan) to 5.4 trees ha–1 (Oenocarpus mapora) (Table 3.2). Visualinspection of the island-wide distribution maps suggested that all species hadbiased/clustered spatial distributions, with more individuals occurring in somehabitat types than in others (Figure 3.2). For example, O. mapora seemed to bemore abundant on ridges, Jacaranda copaia seemed in the old-growth forest andAttalea butyracea in regrowth forest.

Aggregation patterns differed among species as indicated by L(d)/d values at thescale of the entire island (Figure 3.3). T. guayacan and A. butyracea had the lowestL(d)/d of all five species, indicating that these were less aggregated (fewertrees/clump) than other species in the range from 0 to 1500 m. O. mapora had thehighest L(d)/d value, indicating the strongest aggregation. T. guayacan had a peakL(d)/d at the largest distance (32 m), which suggests it had the largest clump size(!) of any species (Figure 3.3).

Chapter 3

50

regrowthold growth

A B C

Real distribution Poisson cluster process Complete spatial randomness

Figure 3.1. Three spatial distributions of Attalea butyracea on Barro Colorado Island, Panama: (A)observed, (B) distribution simulated with Poisson cluster process that incorporates species-specificaggregation and (C) distribution simulated with complete spatial randomness, which does not. Eachplot has the same number of individuals. The background is a map of forest age.


51

regrowthold growth

bohio formationcaimito marine faciescaimito volcanic faciesandesite flows

flat (<5 deg slope)shallow slope (5-10 deg)

Forest age Topography Geological formationsAs

troca

rym

stand

leyan

umAt

talea

but

yrac

eaJa

cara

nda

copa

iaOe

noca

rpus

map

ora

Tabe

buia

guay

acan

steep slope (>10 deg)ridgeshore

Figure 3.2. Spatial distribution of five tropical tree species plotted against a background of envi-ronmental factors on Barro Colorado Island, Panama.

Spatial clumpingThe spatial clumping parameters showed several consistent trends among species(Fig. 3.4; Appendices S3, S4). The size of clumps and the number of trees per clumpincreased while the clump density (ha–1) decreased with increasing plot size indi-cating larger spatial scales are needed to identify the large-scale clumping patternsexhibited by all five species. Clump size was fairly consistent between the 100-

Chapter 3

52

00

1

2

3

4

5

6

10 20 30 40distance (m)

distance = 0 to 1800 m





Oenocarpus mapora

L(d)

/d

distance (m)0

0

1

2

3

4

5

6

10 20 30 40

Jacaranda copaia

L(d)

/d

0 10 20 30 40

Tabebuia guayacan0

0

1

2

3

4

5

6

10 20 30 40

Attalea butyraceaL(

d)/d

0 10 20 30 40

Astrocaryum standleyanum

00123456

18001200600

00123456

18001200600 00123456

18001200600

00123456

18001200600 00123456

18001200600

Figure 3.3. Ripley’s K function (L(d)/d) for five species on Barro Colorado Island, Panama. Plotsshow association at small distances (2 – 40 m) and at large distances (2 – 1800 m, Inserts). Values of(L(d)/d) above 1 indicate clumping, and below 1 indicate even distributions. Solid lines representobserved (L(d)/d), and dashed lines represent the 95% envelope generated through 99 Monte Carlosimulations with complete spatial randomness.

and 200-ha plot sizes but showed little consistency between adjacent plot sizes inthe rest of the sampling size range (Figure 3.3). The variance in clump densitydecreased with plot size for all species, but the variance in cluster size andtrees/clump showed no clear pattern with plot size (data not shown).

Species-habitat associationsWhen compared against simulations complete spatial randomness (CSR), allspecies showed an increasing percentage of significant habitat associations (eitherpositive or negative) with spatial size (Fig. 3.5; Appendix S5). At spatial scales≥400 ha, all species were associated with ≥40% of the environmental variables.All three palm species showed significant associations with ≥40% of the environ-mental variables starting at plot sizes of 100 ha. A. butyracea, Astrocaryum stand-leyanum, and O. mapora all had similarly high percentages of significant associa-tions with forest age, geologic substrate and topography. For J. copaia there werefewest associations with topography. For T. guayacan, the greatest number ofassociations was with forest age.

The simulations under PCP, which incorporated species-specific differences inaggregation, yielded fewer plant–habitat associations than CSR simulations (Figure


53

0

100

200

300

400

500

clum

p si

ze (m

)

1600100 200 40050 800scales

Tabebuia guayacan

1600100 200 40050 800

Oenocarpus mapora

0

100

200

300

400

500cl

ump

size

(m)

1600100 200 40050 800


1600100 200 40050 800

Attalea butyracea

1600100 200 40050 800

Jacaranda copala

Figure 3.4. Effect of sampling scale on clump size (m) as determined by Poisson point clusteranalysis for five tropical tree species on Barro Colorado Island, Panama. Box-plots summarize allsubsamples at a particular sampling scale (sample sizes are 10, 10, 10, 4, 1, 1 subsamples for plotsizes on the x-axis). Plot shows median and interquartile range (box) and standard error (whiskers).

3.5; Appendix S6). For all environmental variables considered together, thepercentage of significant habitat associations was ≤40% for all species and all plotsizes. The number of significant habitat associations with the PCP simulationsincreased with sample size, but less strongly than with the CSR simulations. For A.butyracea and O. mapora, the PCP simulations actually showed a drop in thepercentage of significant correlations at the largest spatial sizes for some environ-mental variables. For simulations under PCP, forest age had the most significantassociations for nearly all the species, especially at large plot sizes. The decrease inassociations from CSR to PCP simulations was greatest for geological formationand lowest for forest age (Figure 3.5). O. mapora showed significant reductions inthe percentage of significant habitat associations between PCP and CSR simula-tions at all spatial scales. A. butyracea, A. standleyanum and J. copaia showed thelargest decreases in the percentage of habitat associations under PCP simulationsat larger spatial scales.

There was a high degree of variability in sign of the habitat associations (posi-tive or negative) within and among sampling sizes, especially under CSR simula-tions (Appendix S5 and S6). For example, there were both significantly positive andnegative associations at the 100-ha and 200-ha scales found for A. butyracea withthe ridge habitat type depending on plot location. At the 100 ha scale, 50% of theCSR simulations that had 2 or more significant associations for the same speciesand habitat type showed both positive and negative associations. Under PCP simu-lations, this was reduced to 20%. This indicates that the actual spatial placementof plots had a strong influence (via environmental heterogeneity, clustering charac-teristics, species density, etc.) on the observed association patterns. Forest age hadconsistently positive or consistently negative associations for both PCP and CSR atnearly all spatial scales and for all species. Geologic substrate and topography hadmore instances of having both negative and positive associations for the samespecies within and between spatial scales, especially with the CSR simulations.

Under PCP simulations, several consistent combinations of species andenvironmental variables showed some consistency within or among scales interms of sign and significance of association (Appendix S5 and S6). All palm speciesplus T. guayacan were positively associated with the secondary forest and nega-tively associated with old-growth forest (Figure 3.5). In contrast, J. copaia waspositively associated with the old-growth forest and negatively associated withsecondary forest. These associations with forest age were most consistent at thelargest plot sizes (800–1600 ha). O. mapora had the most numerous habitat associ-ations. It was significantly and positively associated with the Andesite and Caimitovolcanic types, regrowth forest and the shore and ridge habitats, but negativelyassociated with the old-growth forest and steep habitats.

Chapter 3

54


55

020406080

100

% o

f hab

itat a

ssoc

iatio

ns

1600100 200 40050 800scales (ha)

CSRPCP

020406080

1000

20406080

1000

20406080

1000

20406080

100

1600100 200 40050 800 1600100 200 40050 800 1600100 200 40050 800

all variables forest geology topography

O. m

apor

aT.

gua

yaca

nJ.

cop

aia

A. s

tand

leya

num

A. b

utyr

acea

Figure 3.5. Variation in habitat associations with different sampling scales assuming PCP (PoissonCluster Process) and CSR (Complete Spatial Randomness) in generating the simulated datasets forthe five study species. The y-axis (% habitat associations) is the mean number of levels within thegiven environmental variable (forest, geology, topography) showing a significant association (eitherpositive or negative) with the species distribution (Appendix S5 and S6). The error bars indicate thestandard error of significant associations among plots at the given sampling scale.

Discussion

Habitat association is a distinctive feature of niche specialization (Harms et al.2001), but the strength of the habitat associations found in tropical forests ishighly inconsistent across studies (Table 3.1; Cottenie 2005). Our study suggeststhat an important part of these inconsistencies can be attributed to the variationin the sampling scale across studies and/or a failure to explicitly account forspecies-specific aggregation patterns due to dispersal limitation. The number ofthe plant–habitat associations detected in this study varied strongly with thesampling scale for the five studied species, but differences among plot sizesdecreased when dispersal limitation was included. Changes in the number ofhabitat associations depended on the environmental factor being considered.Forest age produced fairly consistent results across spatial scales (especially spatialscales ≥400 ha) both with and without incorporating dispersal limitation, whereasgeological formation and topographical class did not. Svenning et al. (2004) alsofound forest age to be the environmental variable with the highest number ofsignificant relationships with plant distributions. For some species and environ-mental factors, consistent results showing habitat associations for canopy-staturedtropical forest species may require large sampling scales (>200 ha).

Due to the large variation in environmental and species arrangement withinreplicates at this scale, we find that a single 50-ha plot, the standard in studies oftropical tree diversity and the distributions may not be suitable for identifyingconsistent trends of habitat associations that could be applicable at landscapescales for many species and environmental variables. The mean percentage ofsignificant habitat associations was the lowest for the 50-ha plot scales for nearlyall combinations of species and environmental factors. Variance in percentage ofsignificant habitat associations among 50-ha plot was also high. Different plotsmay yield entirely opposite conclusions. This emphasizes that the particular place-ment of a single plot can determine the results of a study. Also, we found thatanalysis at larger spatial scales can reveal habitat associations that were not foundat smaller scales.

Point pattern analysis identified pronounced species-specific aggregationpatterns, which is in agreement with Condit et al. (2000) and Plotkin et al. (2002).However, the calculated clumping characteristics changed when larger samplingscales were considered. In general, larger sample scales yielded larger clump sizes.If there were a single ‘best’ clump characteristic for these species, we wouldexpect clumping characteristics to remain the same for at least several adjacentsampling scales. There was some consistency in clumping characteristics betweenthe 100 and 200 ha sampling sizes, but overall there was little consistency amongadjacent sample sizes. The variation in clumping characteristics with samplingsize suggests different mechanisms are causing spatial aggregation depending onthe spatial scale. For example, different dispersal agents may cause seeds to aggre-

Chapter 3

56

gate at different scales. In the case of animal-dispersed seeds, seeds rain willinitially cause seeds to clump near the stems of the parent tree. For species withmultiple animal dispersers, such as A. butyracea, different dispersers (agoutis,squirrels) may move seeds different distances from the parent tree. Also, tree-fallgap formation may generate clumps of characteristics sizes, especially for light-demanding species, which includes three of the five species in this study, whoseestablishment is favoured by tree-fall gaps (Table 3.2). Previous studies (Plotkin etal. 2002; Wiegand et al. 2009) have quantified clustering properties at differentspatial scales, but all within plots ≤ 50 ha. Our results suggest different criticalclustering sizes continue to develop at larger spatial scales.

Our results also show that habitat associations that emerge in tests under CSRoften disappear once the effect of spatial aggregation (dispersal limitation) isaccounted for. While some previous studies of habitat association have treatedtrees as independent units without accounting for the natural clumping that mayarise from dispersal limitation (e.g., Baillie et al. 1987, Clark et al. 1999, Dalle et al.2002, Blundell & Peart 2004, Costa et al. 2005), many recent studies includedispersal limitations and, as in our case, find that aggregation pattern has a strongeffect on the number of plant–habitat associations found for a species (Plotkin etal. 2000, Svenning et al. 2006, Bohlman et al. 2008, Leithead et al. 2009). Our studyis unique in showing how clumping patterns vary across spatial scales, and in factwe found the greatest reduction between CSR and PCP occurred at the largestspatial scales rarely used in previous studies. Dispersal limitation appears to be acritical factor in determining tree distributions even at large spatial scales (Nathan& Muller-Landau 2000) and may lead to wrong conclusions about species-habitatassociations if not accounted for.

Our study supports the emerging view that the spatial distribution of tropicaltree species involves both niche differentiation and dispersal limitation, wheretheir relative importance varies with spatial scale. Rather than being twocontrasting processes, they interact with each other. Also, this implies thatdispersal limitation needs to be accounted for when niche differences are studiedand vice versa.

AcknowledgementsWe thank Helene Muller-Landau for her valuable help with the aerial photos taken by

Marcos Guerra, and the Smithsonian Tropical Research Institute for their logistical support.

We are very grateful to Joe Wright for his vital support. We wish to thank Sara Pinzon-

Navarro, Noelle Beckman and Eduardo Medina for all their support and help in the analysis.

This work was funded by the Ubbo Emmius scholarship program of the University of

Groningen (CXG).


57

Chapter 3

58

Appendix S2. Samples per plot size for each level of the environmental variables. Total possiblesamples per plot size are listed in the row headings. Samples that lack spatial coverage of the levelfor the plot size were not included.

50 ha 100 ha 200 ha 400 ha 800 ha 1600 haN = 10 N = 10 N = 6 N = 3 N = 1 N = 1

GeologyAndesite 6 9 5 3 1 1Bohio volcanic 7 8 6 3 1 1Caimito marine 6 7 5 3 1 1Caimito volcanic 3 2 2 2 1 1

Forest ageRegrowth 9 10 6 3 1 1Old growth 10 10 6 3 1 1

HabitatShore 5 9 6 3 1 1Flat 7 9 6 3 1 1Ridge 10 10 6 3 1 1Shallow 7 9 5 3 1 1Steep 10 10 6 3 1 1


59

0

1

2

3

num

ber o

f clu

mps

per

ha

1600100 200 40050 800scales

Tabebuia guayacan

1600100 200 40050 800

Oenocarpus mapora

0

num

ber o

f clu

mps

per

ha

1600100 200 40050 800


1600100 200 40050 800

Attalea butyracea

1600100 200 40050 800

Jacaranda copala

1

2

3

Appendix S3. Effect of sampling scale on the number of clumps per ha for five tropical tree specieson Barro Colorado Island, Panama.

0

10

20

30

40

50

num

ber o

f tre

es p

er c

lum

p

1600100 200 40050 800scales

Tabebuia guayacan

1600100 200 40050 800

Oenocarpus mapora

0

10

20

30

40

50

num

ber o

f tre

es p

er c

lum

p

1600100 200 40050 800


1600100 200 40050 800

Attalea butyracea

1600100 200 40050 800

Jacaranda copala

0

1200

800

400

0

100

200

300

400

0

100

200

50

500

150

Appendix S4. Effect of sampling scale on the average number of trees per clump for five tropicaltree species on Barro Colorado Island, Panama.

Chapter 3

60

Ap

pen

dix

S5.

Tabl

e of

hab

itat

ass

ocia

tion

s in

CSR

-gen

erat

ed s

imul

atio

ns.

The

sig

n of

hab

itat

ass

ocia

tion

was

ide

ntif

ied

as:

‘-’

nega

tive

ass

ocia

tion

,‘+

’ pos

itiv

e as

soci

atio

n, ‘0

’ no

asso

ciat

ion,

and

‘na’

whe

n th

e w

hen

the

habi

tat

type

was

not

pre

sent

in t

he s

ampl

e si

ze.

Env

iron

men

tal

Sam

ple

size

(ha)

vari

able

16

0080

040

020

010

050

(per

spe

cies

)40

040

0a40

0b20

020

0a20

0b20

0c20

0d20

0e10

010

0a10

0b10

0c10

0d10

0e10

0f10

0g10

0h10

0i50

a50

b50

c50

d50

e50

f50

g50

h50

i50

j

Ata

lea

buty

race

aG

eolo

gyA

ndes

ita

--

00

-+

-na

+0

++

--

na+

+0

0+

0na

00

na+

00

nana

naB

ohio

vol

cani

c-

-+

-0

00

--

0-

0na

--

--

0na

-0

0na

na0

-0

0-

0na

Caim

ito

mar

ine

--

-0

+-

+0

0+

na0

++

0na

na0

+na

0+

0+

nana

na0

0na

naCa

imit

o vo

lcan

ic+

+0

+na

nana

+0

nana

nana

na+

0na

nana

nana

nana

nana

nana

na+

00

Fore

st a

geR

egro

wth

++

++

0+

0+

+0

++

+0

++

+0

++

00

na0

++

00

+0

0O

ld G

row

th-

--

-0

-0

--

0-

--

00

--

0-

-0

00

00

-0

0-

00

Topo

grap

hySh

ore

+0

na0

00

00

-0

-0

+0

-0

-0

0na

00

na0

0na

na0

na0

0Fl

at+

+0

0+

0+

00

00

0+

+0

-0

0+

na0

00

00

na0

00

00

Rid

ge0

0-

+-

0-

++

0+

0-

0+

++

0-

00

00

00

+0

00

++

Shal

low

-0

0+

+0

+0

00

00

00

00

00

00

00

00

00

00

00

0St

eep

-0

+-

00

0-

-0

-0

00

--

00

00

00

00

0-

00

0-

0

Ast

roca

ryu

m s

tan

dley

anu

mG

eolo

gyA

ndes

ita

--

--

-+

-na

+-

++

--

na+

+0

-0

0na

00

na0

00

nana

naB

ohio

vol

cani

c-

-0

-+

-0

0-

0-

-na

-0

--

0na

00

0na

na0

00

--

0na

Caim

ito

mar

ine

0+

0+

+0

+0

0+

na+

++

0na

na+

+na

+0

00

nana

na0

+na

naCa

imit

o vo

lcan

ic+

++

+na

nana

0+

nana

nana

na0

+na

nana

nana

nana

nana

nana

na0

00

Fore

st a

geR

egro

wth

++

++

0+

00

+0

00

++

-+

00

++

00

na0

00

00

0-

0O

ld G

row

th-

--

-0

-0

+-

00

0-

-+

-0

0-

-0

00

00

-0

00

+0

Topo

grap

hySh

ore

-+

na0

00

0-

00

-0

+0

0+

-0

0na

00

na0

0na

na0

na0

0Fl

at+

++

++

0+

-+

00

0+

00

00

0+

na0

0+

00

na0

00

0-

Rid

ge0

--

--

00

0+

00

0-

00

++

0-

00

00

00

00

00

00

Shal

low

+0

++

++

00

0+

++

00

00

+0

00

+0

00

00

00

0-

0St

eep

0-

0-

--

-+

--

--

--

+-

-0

00

-0

00

00

00

-+

0

Jaca

ran

da c

opai

aG

eolo

gyA

ndes

ita

++

+0

+0

+na

0+

00

+0

na0

00

+0

0na

+0

na0

00

nana

naB

ohio

vol

cani

c-

--

+0

-0

00

00

0na

+0

+0

0na

+0

0na

na0

00

0+

0na

Caim

ito

mar

ine

0-

00

-0

-0

0-

na0

--

0na

na0

-na

00

-0

nana

na0

0na

naCa

imit

o vo

lcan

ic-

--

-na

nana

0-

nana

nana

na0

-na

nana

nana

nana

nana

nana

na-

00


61

Fore

st a

gR

egro

wth

--

--

--

-0

00

00

--

0-

00

--

00

na-

00

00

00

0O

ld G

row

th+

++

++

++

00

00

0+

+0

+0

0+

+0

00

+0

00

00

00

Topo

grap

hySh

ore

-0

na0

0-

0+

00

00

-+

00

00

0na

00

na0

0na

na+

na0

0Fl

at0

--

0-

--

00

00

0-

-0

00

0-

na0

00

00

na0

00

00

Rid

ge+

++

0+

0+

00

+0

0+

00

00

0+

00

0+

+0

00

00

00

Shal

low

+0

00

-+

-0

00

0+

--

00

00

-0

00

-0

00

00

00

0St

eep

0-

-0

0-

00

0-

00

00

-0

00

00

-0

00

00

00

00

0

Tab

ebu

ia g

uay

acan

Geo

logy

And

esit

a0

00

00

00

na+

0+

00

-na

++

0-

00

na0

0na

00

0na

nana

Boh

io v

olca

nic

--

0-

00

0-

-0

-0

na0

0-

-0

na0

00

nana

00

00

-0

naCa

imit

o m

arin

e-

0-

-+

-+

-0

0na

00

00

nana

0+

na0

0+

0na

nana

0-

nana

Caim

ito

volc

anic

++

0+

nana

na+

0na

nana

nana

+0

nana

nana

nana

nana

nana

nana

+0

0Fo

rest

age

Reg

row

th+

++

++

++

++

0+

00

++

++

0+

+0

0na

00

00

0-

00

Old

Gro

wth

--

--

--

--

-0

-0

--

+-

-0

--

00

00

0-

00

00

0To

pogr

aphy

Shor

e+

0na

00

00

00

00

0+

00

00

0+

na0

0na

00

nana

0na

00

Flat

+0

00

+0

+0

00

00

0+

00

00

+na

00

+0

0na

00

00

0R

idge

00

0+

00

0+

00

-0

00

+0

00

-0

00

00

00

00

+0

0Sh

allo

w-

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

Stee

p-

00

-0

00

-0

00

00

00

00

00

00

00

00

00

0-

00

Oen

ocar

pus

map

ora

Geo

logy

And

esit

a-

-+

-0

++

na+

++

++

-na

++

0+

-+

na0

0na

+-

-na

nana

Boh

io v

olca

nic

--

--

--

+-

--

--

na+

--

--

na+

--

nana

0-

+0

--

naCa

imit

o m

arin

e-

--

-0

+-

--

0na

+-

--

nana

+-

na-

+0

0na

nana

00

nana

Caim

ito

volc

anic

++

-+

nana

na+

+na

nana

nana

+-

nana

nana

nana

nana

nana

nana

++

0Fo

rest

age

Reg

row

th+

++

+0

-0

++

00

-0

0+

+0

-+

-+

-na

+0

-+

++

++

Old

Gro

wth

--

--

0+

0-

-0

0+

00

--

0+

-+

-+

0-

0+

--

--

-To

pogr

aphy

Shor

e+

0na

+0

-0

0-

--

-0

+0

0-

-0

na-

-na

+0

nana

0na

00

Flat

++

0+

0-

-+

+-

--

0-

00

0-

0na

-+

+0

0na

00

0+

-R

idge

++

++

0+

++

++

++

+0

++

++

00

++

00

++

+0

0+

+Sh

allo

w0

0-

+-

--

+0

-+

0-

-0

0+

00

--

--

-0

-0

0+

0-

Stee

p-

--

-0

-0

--

0-

-0

+-

--

-+

0+

-0

0-

0-

0-

--

Chapter 3

62

Ap

pen

dix

S6.

Tabl

e of

hab

itat

ass

ocia

tion

s fo

r PC

P-ge

nera

ted

sim

ulat

ions

. The

sig

n of

hab

itat

ass

ocia

tion

was

ide

ntif

ied

as:

‘-’ n

egat

ive

asso

ciat

ion,

‘+’ p

osit

ive

asso

ciat

ion,

‘0’ n

o as

soci

atio

n, a

nd ‘n

a’ w

hen

the

habi

tat

type

was

not

pre

sent

in t

he s

ampl

e si

ze.

Env

iron

men

tal

Sam

ple

size

(ha)

vari

able

16

0080

040

020

010

050

(per

spe

cies

)40

040

0a40

0b20

020

0a20

0b20

0c20

0d20

0e10

010

0a10

0b10

0c10

0d10

0e10

0f10

0g10

0h10

0i50

a50

b50

c50

d50

e50

f50

g50

h50

i50

j

Ata

lea

buty

race

aG

eolo

gyA

ndes

ita

00

00

00

0na

00

0+

00

na0

00

00

0na

+0

na0

00

nana

naB

ohio

vol

cani

c0

00

-0

00

00

00

0na

0-

00

0na

00

0na

na0

00

00

0na

Caim

ito

mar

ine

00

00

+0

+0

00

na0

0+

0na

na0

0na

00

00

nana

na0

0na

naCa

imit

o vo

lcan

ic0

+0

0na

nana

00

nana

nana

na0

0na

nana

nana

nana

nana

nana

na0

00

Fore

st a

geR

egro

wth

++

++

0+

0+

+0

00

+0

0+

+0

00

00

na0

00

00

00

0O

ld G

row

th-

--

-0

-0

--

00

0-

00

--

00

00

00

00

0+

00

00

Topo

grap

hySh

ore

00

na0

00

00

00

00

00

00

00

0na

00

na0

0na

na0

na0

0Fl

at0

00

00

00

00

00

0+

00

00

00

na0

0-

00

na0

00

00

Rid

ge0

00

00

0-

+-

00

00

0+

+0

0-

00

0+

00

00

00

00

Shal

low

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

0St

eep

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

0

Ast

roca

ryu

m s

tan

dley

anu

mG

eolo

gyA

ndes

ita

00

00

-0

0na

00

0+

00

na0

00

00

0na

00

na0

00

nana

naB

ohio

vol

cani

c0

00

-0

00

0-

00

-na

00

-0

0na

00

0na

na0

00

00

0na

Caim

ito

mar

ine

00

00

+0

+0

00

na0

00

0na

na0

0na

00

00

nana

na0

0na

naCa

imit

o vo

lcan

ic+

+0

0na

nana

00

nana

nana

na0

0na

nana

nana

nana

nana

nana

na0

00

Fore

st a

geR

egro

wth

++

+0

0+

00

00

00

00

0+

00

00

00

na0

00

00

0-

0O

ld G

row

th-

--

00

-0

00

00

00

00

-0

00

00

00

00

00

00

+0

Topo

grap

hySh

ore

00

na0

00

00

00

00

00

00

00

0na

00

na0

0na

na0

na0

0Fl

at+

+0

0+

00

00

00

0+

00

00

0+

na0

0+

00

na0

00

00

Rid

ge0

00

00

00

00

00

00

00

0+

00

00

00

00

00

00

00

Shal

low

00

0+

00

00

00

++

00

00

+0

00

+0

00

00

00

00

0St

eep

00

00

-0

00

-0

--

-0

0-

-0

00

-0

00

00

00

0+

0

Jaca

ran

da c

opai

aG

eolo

gyA

ndes

ita

00

00

00

0na

0+

00

00

na0

00

+0

0na

00

na0

00

nana

naB

ohio

vol

cani

c0

00

00

00

00

00

0na

00

00

0na

00

0na

na0

00

0+

0na

Caim

ito

mar

ine

00

00

00

00

00

na0

00

0na

na0

-na

00

00

nana

na0

0na

naCa

imit

o vo

lcan

ic0

00

0na

nana

00

nana

nana

na0

0na

nana

nana

nana

nana

nana

na-

00


63

Fore

st a

geR

egro

wth

--

00

00

00

00

00

00

00

00

00

00

na0

00

00

00

0O

ld G

row

th+

+0

00

00

00

00

00

00

00

00

00

00

00

00

00

+0

Topo

grap

hySh

ore

00

na0

00

00

00

00

++

00

00

0na

00

na0

0na

na0

na0

0Fl

at0

00

00

00

00

-0

00

00

00

00

na0

-0

00

na0

00

00

Rid

ge+

++

0+

0+

00

+0

00

00

00

0+

00

0+

00

00

00

00

Shal

low

00

00

00

00

00

0+

--

00

00

00

00

00

00

00

00

0St

eep

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

0

Tab

ebu

ia g

uay

acan

Geo

logy

And

esit

a0

00

00

00

na+

00

00

-na

0+

00

00

na0

0na

00

0na

nana

Boh

io v

olca

nic

00

00

00

0-

-0

00

na0

00

00

na0

0-

nana

00

00

00

naCa

imit

o m

arin

e0

00

00

00

00

0na

00

00

nana

00

na0

+0

0na

nana

00

nana

Caim

ito

volc

anic

++

0+

nana

na+

0na

nana

nana

00

nana

nana

nana

nana

nana

nana

00

0Fo

rest

age

Reg

row

th+

++

+0

++

+0

0+

+0

+0

00

0+

00

0na

00

00

00

00

Old

Gro

wth

--

--

0-

--

00

--

0-

00

00

--

00

00

00

00

00

0To

pogr

aphy

Shor

e0

0na

00

00

00

00

0+

00

00

0+

na0

0na

00

nana

0na

00

Flat

00

00

00

+0

00

00

0+

00

00

+na

0+

0+

0na

00

00

0R

idge

00

00

00

0+

00

00

00

+0

00

00

0+

00

00

00

+0

0Sh

allo

w0

00

00

00

00

00

0-

00

00

00

00

-0

00

00

-0

00

Stee

p0

00

00

00

00

00

00

00

00

00

00

-0

00

00

+0

00

Oen

ocar

pus

map

ora

Geo

logy

And

esit

a0

00

00

00

na0

0+

00

0na

00

0+

0+

na0

0na

00

0na

nana

Boh

io v

olca

nic

00

00

00

00

00

-0

na0

00

00

na0

00

nana

00

00

00

naCa

imit

o m

arin

e0

00

00

00

00

0na

00

00

nana

0-

na0

00

0na

nana

00

nana

Caim

ito

volc

anic

++

00

nana

na0

0na

nana

nana

00

nana

nana

nana

nana

nana

nana

00

0Fo

rest

age

Reg

row

th0

+0

+0

00

+0

00

00

0+

00

00

00

0na

00

0+

00

0+

Old

Gro

wth

0-

0-

00

0-

00

00

00

-0

00

00

00

00

00

-0

00

-To

pogr

aphy

Shor

e0

0na

00

00

00

00

00

00

00

00

na0

0na

+0

nana

0na

00

Flat

00

00

00

00

+0

00

00

00

00

0na

00

00

0na

00

00

0R

idge

00

+0

0+

++

+0

++

00

++

+0

00

0+

00

++

00

00

+Sh

allo

w0

0-

+0

00

00

00

0-

00

00

00

-0

00

00

00

0+

00

Stee

p0

--

-0

00

--

0-

00

0-

--

00

00

00

0-

00

00

0-

University of Groningen Determinants of the spatial ...

Documents

Transcript of University of Groningen Determinants of the spatial ...