Author's personal copy...Author's personal copy 328 M. Piquer-Rodríguezet al. / JournalforNature...

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Journal for Nature Conservation 20 (2012) 326– 336

Contents lists available at SciVerse ScienceDirect

Journal for Nature Conservation

j our na l ho mepage: www.elsev ier .de / jnc

Future land use effects on the connectivity of protected area networks insoutheastern Spain

María Piquer-Rodrígueza,∗, Tobias Kuemmerleb, Domingo Alcaraz-Segurac,d,Raul Zurita-Millae, Javier Cabelloc

a Geography Department, Humboldt Universität zu Berlin, Unter den Linden 6, 10099 Berlin, Germanyb Potsdam Institute of Climate Impact Research (PIK), Telegraphenberg A 31, 14473 Potsdam, Germanyc Departamento de Biología Vegetal y Ecología, Centro Andaluz para la Evaluación y Seguimiento del Cambio Global (CAESCG), University of Almería, Ctra, de Sacramento s/n, 04120La Canada de San Urbano, Almería, Spaind Departamento de Botánica, Facultad de Ciencias, Planta 6 Sección de Biología, Campus Universitario de Fuentenueva, Universidad de Granada, 18071 Granada, Spaine Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente, P.O. Box 217, 7500 AE Enschede, Netherlands

a r t i c l e i n f o

Article history:Received 30 October 2011Received in revised form 29 February 2012Accepted 1 July 2012

Keywords:Cellular automataConservation planningFragmentationLand abandonmentLand-use intensificationMarkov chainsMorphological spatial pattern analysisNatura 2000PolarisationUrbanisation

a b s t r a c t

Land-use change is a major driver of the global biodiversity crisis, mainly via the fragmentation andloss of natural habitat. Although land-use changes will accelerate to meet humankind’s growing demandfor agricultural products, conservation planning rarely considers future land uses and how they mayaffect the connectivity of ecological networks. Here, we integrate land-use models with landscape frag-mentation and connectivity analyses, to assess the effects of past and future land-use changes on theconnectivity of protected area networks for a highly dynamic region in southeast Spain. Our results showa continued geographical polarisation of land use, with agricultural intensification and urban develop-ment in the coastal areas, and the abandonment of traditional land use in the mountains (e.g., 1100 km2

of natural vegetation are projected to be lost in coastal areas whereas 32 km2 of natural vegetation wouldrecover in interior areas from 1991 to 2015). As a result, coastal protected areas will experience increasingisolation. The connectivity analyses reveal that the two protected area networks in place in the study area,the European “Natura 2000” and the Andalusian “RENPA” networks, include many landscape connectors.However, we identify two areas that currently lack protection but contain several important patchesfor maintaining the region’s habitat connectivity: the northwestern and the southwestern slopes of theSierra Cabrera and Bédar protected area. Our results highlight the importance of considering future land-use trajectories in conservation planning to maintain connectivity at the regional scale, and to improvethe resilience of conservation networks.

© 2012 Elsevier GmbH. All rights reserved.

Introduction

Land-use change is the main driver of current biodiversitydecline due to the loss, fragmentation, and degradation of naturalhabitats (CBD 2010; MA 2005). Protected areas are a cornerstoneof conservation efforts worldwide and are generally effective inpreventing habitat loss inside them (Andam et al. 2008; Radeloffet al. 2010). However, land-use changes in the surroundings ofprotected areas increasingly isolate protected areas (Newmark2008; Seiferling et al. 2011). To avoid such isolation, conservation

∗ Corresponding author. Tel.: +49 0 30 2093 9341.E-mail addresses: [email protected]

(M. Piquer-Rodríguez), [email protected] (T. Kuemmerle),[email protected] (D. Alcaraz-Segura), [email protected] (R. Zurita-Milla),[email protected] (J. Cabello).

planning needs to move from site-based protection towards moreholistic approaches that consider the landscape as a whole (Naveh2000). Hence, the challenge for decision makers is to prioritise con-servation efforts that maintain the functionality and connectivity ofhabitat networks (Knight et al. 2006). This is particularly importantin the face of climate change, since maintaining functionally con-nected landscapes is the single most effective conservation actionto allow species to adapt to a changing environment (Heller andZavaleta 2009; Pullin et al. 2009).

Many regions of the world are simultaneously experiencingboth loss and recovery of natural habitats due to land-use change(Barbier et al. 2010). On the one hand, the expansion and inten-sification of human land uses to satisfy the ever-growing demandof natural resources (e.g. food, wood, fiber, bio-energy) is a majordriver of habitat loss (Keys and McConnell 2005). On the otherhand, urbanisation and industrialisation often result in a decline ofrural populations and traditional land uses, triggering ecosystem

1617-1381/$ – see front matter © 2012 Elsevier GmbH. All rights reserved.http://dx.doi.org/10.1016/j.jnc.2012.07.001


M. Piquer-Rodríguez et al. / Journal for Nature Conservation 20 (2012) 326– 336 327

recovery in abandoned lands (Kuemmerle et al. 2011; MacDonaldet al. 2000). The co-occurrence of land-use expansion (here: con-verting land that was under alternative use to agriculture) andland abandonment (here: conversion of agriculture to naturalvegetation) result in complex land-use change effects on habitatconnectivity. This poses challenges to conservation planning withthe goal to identify effective and resilient protection networks.

Traditionally, prioritising areas for conservation has been basedon the selection of a number of sites to reach an overall conservationgoal (e.g., protection of 10% of all habitats or 95% of all threatenedspecies) (Margules and Pressey 2000; Moilanen et al. 2009). Yet,most conservation planning efforts have been exclusively based oncontemporary landscape patterns (Rouget et al. 2003). We are onlyaware of a handful of studies that have (at least partly) consideredfuture land-use trajectories when selecting conservation sites. Forexample, current and potential land-use threats and future vul-nerability values were incorporated to map conflict areas betweenland transformation and conservation sites at the regional scale(Joshi et al. 2011; Reyers 2004; Wessels et al. 2003) or between landtransformation and biodiversity hotspots (Perez-Vega et al. 2012).Similarly, using the potential extent of future habitat loss owing toland-use and climate changes and global scenarios showed land-use change pressures that drive the endangerment of biodiversityat the global and continental scale (Araújo et al. 2008; Giam et al.2010; Visconti et al. 2011). However, linking land-use projectionswith connectivity analyses for networks of protected areas hasrarely been done. Major reasons for the scarcity of studies linkingland use modelling, landscape connectivity analysis, and conser-vation prioritisation is the lack of future land-use maps at scalesdetailed enough to allow for conservation planning, as well as thelack of tools to assess the fragmentation and connectivity of land-scapes in a spatially explicit way.

Relatively simple and robust land-use simulation models areincreasingly becoming available for conservationists and land-use planners to explore future pathways of land-use systems(e.g. Clarke and Gaydos 1998; Pontius et al. 2001; Verburg andOvermars 2009; Verburg et al. 2006). Similarly, novel spatiallyexplicit approaches that analyse landscape connectivity have beendeveloped that allow identifying those landscape elements that areof particular importance for habitat connectivity. For instance, mor-phological image segmentation can stratify habitat patches intocore, edge, and corridor habitat. Integrating such spatially explicitfragmentation approaches with graph theory tools (Kindlmannand Burel 2008), which asses the structural connectivity of habi-tat networks, provides powerful tools for identifying priority sitesfor conservation (Saura and Torné 2009; Ziółkowska et al. 2012).Such approaches refer to the structural connectivity of landscapes(i.e., the topological relations among natural habitat patches).Landscape structural connectivity is a good proxy for functionalconnectivity for many species, since highly connected landscapesallow for better dispersal of intermediate-mobility species and,hence, for lower risk of isolated populations (Saura and Rubio2010). Our goal here was to combine a spatially explicit land-usemodel and fragmentation analysis in order to assess the effects offuture land use on the connectivity of current protected area net-works. As a study area, we selected a relatively large region in thesoutheast of Spain which undergoes rapid land-use changes, withco-occurrence of both land-use expansion and abandonment. Ourspecific research questions were:

1. How will future land-use patterns develop during the next yearsin southeast Spain and how will they affect the fragmentationand connectivity of natural habitats?

2. Do current protected areas comprise those landscape elementsthat are important for maintaining the regional landscape

connectivity? Which important gaps should be considered forconservation?

Methods

Study area

Our study focuses on the province of Almería in the southeast ofSpain, Andalusia (Fig. 1). This area is located in the Mediterraneanbasin, which is highly dynamic in terms of land-use changes withclose co-occurrence of both loss and recovery of natural vegetation(SOER 2010). On the one hand, the province of Almería has under-gone land use expansion over the last decades, especially near thecoast, leading to habitat loss due to the expansion of greenhousesand urbanisation (the latter mainly for tourism) (Mota et al. 1996;Penas et al. 2011). Indeed, the current economy of Almería mainlydepends on semi-industrial intensive agriculture (Sánchez-Picón2005) and tourism (Junta de Andalucía 2009), which are concen-trated along the coast. On the other hand, rural exodus from innerregions towards the coast, has led to the abandonment of tradi-tional land-use practices (del Barrio et al. 2010), followed either bythe subsequent vegetation recovery (Penas et al. 2011) or increasedsoil erosion (Vogiatzakis et al. 2006).

The province is a representative site of the Mediterranean bio-diversity hotspot (Médail and Quézel 1999), including 2800 plantspecies (Blanca et al. 2009), 20% of the Iberian endemic plant gen-era (Cabello 2002), and 3300 km2 of unique Habitats of EuropeanCommunity Interest (Fig. 1, Directive 92/47/CEE). Thanks to its greatclimatic and relief variability, the region harbours diverse vegeta-tion communities ranging from semi-arid scrublands and steppesin the valleys and coastal hills, over to Mediterranean sclerophy-lous and coniferous forests in the mid-mountains, and to alpinegrasslands in the higher mountains (up to 2800 m.a.s.l) (Valle 2004).

There are currently two protected area networks in forcein the study area: (1) the Andalusian network of protectedareas, Red de Espacios Naturales Protegidos de Andalucía (RENPA,Decreto 95/2003), consists of 19 sites in Almería and protects33% of the province surface; and, (2) the European Natura 2000network contains six additional sites, increasing the protec-tion to 52% of the province. Natura 2000 sites were classifiedas sites of community importance (SCIs) in 2006 becausethey contribute to maintain (1) priority habitats or species ofcommunity interest, and (2) the coherence of the Natura 2000 net-work (Directive 92/43/EEC, http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG: 1992L0043:20070101:EN:PDF). TheSCIs are fully protected in Spain since their approval by the Euro-pean commission in 2006 and have the same legal status as theRENPA protected areas. Development is only fully prohibited inSpanish National Parks, where there are core zones. In Almeríathere is only part of a trans-boundary National Park (National Parkof Sierra Nevada) with no core areas (zone R). All other protectedareas in Almería have a lower protection status (i.e., natural parks,natural reserves, natural monuments and natural places) and land-use changes can thus, to some extent, take place inside them. Forthe sake of simplicity, we further refer to the Andalusian networkof protected areas as the RENPA network, to the six SCIs enlargingthe RENPA as Natura 2000 protected areas and to the whole Euro-pean Natura 2000 network of protected areas (RENPA and SCIs) asthe Natura 2000 network (Fig. 2).

Land use data

We used two regional land-use maps from 1991 and 1999 in thiswork. These maps were produced via photointerpretation of aerial,Landsat Thematic Mapper, and Indian Remote Sensing Satellite


328 M. Piquer-Rodríguez et al. / Journal for Nature Conservation 20 (2012) 326– 336

Fig. 1. Study area in the Mediterranean Basin. Eco-regions (numbered polygons) and Priority Habitats of European Community Importance (in grey) are shown for theAlmería province (SE Spain) in Andalusia (south Spain).

Fig. 2. Vegetation changes for the period 1991–2015 with the RENPA and Natura 2000 protected areas (PAs) overlaid.



Table 1Description of the land use (LU) classes (relative area share) as well as accuracy measures for 2005.

LU class LU description Area change(%) 2015–1991

Area change(%) 1999–2007

Area-adjusted produceraccuracy (%) 2005

Area-adjusted useraccuracy (%) 2005

Urban Urban nucleus, rural settlements, roads and quarries 1.64 −0.577 18.25 75.00Agriculture Non-irrigated and irrigated arable lands 6.16 −2.166 84.72 63.16Vegetation Natural vegetation −12.16 4.336 86.07 84.72Freshwater Rivers, water reservoirs and marshes 0.18 −0.095 0.00 0.00Greenhouses Intensive agricultural practices 4.18 −1.498 3.74 71.43

Overall Accuracy (%) 77.22Allocation disagreement (%) 12.00Quantity disagreement (%) 8.50

images at a scale of 1:50,000. Both land use maps have a pixel size of30 m × 30 m and an overall accuracy of 94.4% (Junta de Andalucía.Consejería de Medio Ambiente 1999). A more recent map (2007)was not suitable for our analysis due to differences in the thematicand spatial resolution. We reclassified the 1991 and 1999 land-usemaps into five thematic classes: (1) urban areas; (2) greenhouses;(3) agriculture; (4) natural vegetation; and, (5) freshwater (Table 1).Next, we followed the methodology suggested by Jongman et al.(2006) to stratify land-use patterns by using an eco-region zonation(Fig. 1) based on climate, terrain, potential vegetation, and regionalhydrology (Borja Barrera and Montes del Olmo 2008).

Land use change modelling

To estimate potential land-use conversions and to producechange maps, we projected future land uses from the transitionsoccurred between 1991 and 1999 based on Markovian chains andon a cellular automata model (Eastman 2003). Markovian chainswere used to derive the conditional probabilities of change bycalculating the land-use transitions from one land use (1991) toanother (1999) and treating them as a stochastic process with dis-crete time-steps (Bell 1974). Thus, using the 1991 and the 1999land-use maps, we calculated transition probabilities matrices andtransition areas matrices (i.e., amount of pixels changing from onecategory to another) for the years 2005, 2007 and 2015. The lattertwo years were chosen according to the 8-year time period betweenthe input maps (1991-1999). The year 2005 was used for an in situfield validation. Thus, our projection can be interpreted as a baselinescenario depicting the ‘Business as Usual’ pathway.

Markov transition probabilities depend on the time interval (T)used to simulate land-use change (NRCS 2000). If T equals the timelag between the input datasets (i.e., for the years 2007 and 2015),transition probabilities are calculated using Eq. (1):

Pi,j,T = ni,j∑ini,j

(1)

where Pi,j,T is the transition probability from land use i to j in timeinterval T, and ni,j is the number of pixels that change from one landuse to another.

If T does not match the interval between the input datasets (i.e.,for the year 2005), then transition probabilities are calculated usingEq. (2):

Pi,j = 1 − exp[In(1−Pi,j,T )]

T(2)

All Markovian probabilities matrices were corrected for poten-tial errors (Pε

i.j) derived from spatial variability in the input land-use

maps between the land-use values mapped and the real in situ

land-use values, following Eq. (3) (NRCS 2000):

Pε

i.j=

⎧⎪⎪⎨⎪⎪⎩Pi,j(1 − ε) for i = j

Pi,j1 − Pi.j(1 − ε)/∑i

∀i /= jPi,j for i /= j and∑i

∀i /= jPi,j ≥ 0

1 − Pi,j(1 − ε)/(N − 1) for i /= j and∑i

∀i /= jPi,j = 0(3)

where ε is the overall error, i.e., input land-use maps overall pro-duction error, and N is the number of land-use classes. In our case, εwas assumed to be 0.15 taking into account that land-use maps arerecommended to have as an average 85% overall accuracy (Foody2002). Higher production accuracies were reported by the inputmaps producers but we were skeptical on the validation mea-sures undertaken (which were spatially distributed only along mainroads) and preferred to consider lower accuracies as a precaution.

To spatially allocate the amount of land-use change predicted bythe Markov chain probabilities, we used a cellular automata (CA)model. The CA model uses the Markov transition areas matrix andthe latest input land-use map (1999) to allocate future land usesbased on the distribution of local land-use neighbours. In otherwords, the area of change calculated in the transition matrix isthe same as the total area of change spatially allocated by thecellular automata. A 5 × 5 cell moving window was selected tospatially allocate the projected land-use quantities in the surround-ings of previous land-uses’ cells. The size of the moving windowwas defined by the median patch size of the 1999 land-use map(5.928 m2) since the patch size distribution was skewed (Levin andFox 2004). This resulted in three projected land-use maps for 2005,2007 and 2015, as well as a vegetation change map from 1991 to2015 that contained the classes (1) natural vegetation loss, and (2)natural vegetation gain. For the vegetation change map other land-use transitions were not considered. Both, the Markovian chainsand the cellular automata model were implemented using the soft-ware IDRISI Kilimanjaro (Eastman 2003).

Because our land-use model derived current and future land-use maps, we could validate the current land-use map. To do so,we collected field data in September 2005 where we evaluated 240random plots of 30 m × 30 m. These plots were stratified based onthe eco-regions map (Stehman and Czaplewski 1998). Each plotwas labeled according to the five land-use classes (Table 1) andwas evaluated as correct when the assigned label matched thatof the 2005 projected land-use map. We estimated two measuresthat inform about different aspects of the model accuracy (Foody2002): (1) we calculated the confusion matrix, the area-adjustedproducer’s and user’s accuracies; and (2) the quantity and allocationdisagreement (Pontius et al. 2008; Pontius and Millones 2011). Theproducer’s and user’s accuracies were adjusted to the area actuallycovered by the land cover classes to correct for the bias inherent tostratified sampling procedures (Card 1982; Cochran 1977).



Inter-annual changes of habitat connectivity

To analyse the fragmentation and structural connectivity ofnatural habitat, we used an approach that combines image segmen-tation with graph theory (Saura and Rubio 2010). Fragmentationassesses the configuration of individual landscape elements, e.g.the shape of individual patches, and characterises how land-usechanges affect habitat topology and configuration (Fahrig 2003).We used morphological spatial pattern analysis (Vogt et al. 2007)to quantify the fragmentation of natural habitats. This approachsegments an input image (e.g. the natural vegetation map in ourcase) into a set of fragmentation components that reveal informa-tion about patch size, shape, geometry and connections (Soille andVogt 2009): core; bridge (i.e., connections among core areas); islet(i.e., small habitat patches without core habitat); edge; and, per-foration (i.e., edge habitat inside larger core patches). We used an8-neighbour rule and a 1-pixel edge width (without distinguishingbetween internal and external edges, (Vogt et al. 2009)). To char-acterise inter-annual changes in habitat fragmentation, we carriedout the same analysis for 1991, 1999, 2007, and 2015.

Structural connectivity refers to the topological connectionsamong natural habitat patches across the entire landscape. Graphtheory (Kindlmann and Burel 2008) provides a comprehensiveframework to analyse landscape connectivity and to characterisethe relative importance of particular patches (i.e., nodes) and cor-ridors (i.e., links) for the regional connectivity of an entire habitatnetwork. Using the fragmentation components maps derived fromthe morphological spatial pattern analysis, graph models wereobtained for 1991, 1999, 2007, and 2015. We used the index ofpatch probability of connectivity (dPC, Saura and Rubio, 2010), asan estimator of the contribution of each patch to overall landscapeconnectivity (Baranyi et al. 2011). Patches were grouped into nodesand links depending on their area and relative position in the land-scape (i.e., nodes are core areas that may connect to other nodes,while links are usually smaller and connect nodes). For each of ourland-use maps, we identified the most important habitat patchesfor the maintenance of overall structural landscape connectivity.Finally, to evaluate the inter-annual changes of habitat connectiv-ity, we calculated changes in node importance, area of the mostimportant node, number of connectors, and mean distance betweenconnectors.

The role of protected area networks in preserving landscapeconnectivity

Targeting priority conservation settings for a broad range ofusers (landscape planners, conservationists, decision makers, parkmanagers, etc.) should consider an ensemble of connectivity areaswhich would do the most to create a network of natural landscapesrather than a ranked list of areas (Beier et al. 2011). To evaluate therole of protected areas in preserving the most important areas forlandscape connectivity in our study area, we integrated the resultsfrom the land-use projections and from the landscape connectiv-ity analysis. First, we identified the most important patches forthe maintenance of the structural connectivity (patches with high-est dPC values) for the period under study. Second, we assessedwhether protected area networks (RENPA and Natura 2000) effi-ciently preserved important areas for connectivity. Finally, wehighlighted important areas for landscape connectivity by averag-ing the contribution to connectivity (dPC value) of each patch overthe whole study period and by overlaying the RENPA and Natura2000 protected areas networks. Landscape connectors with high-est dPC values were considered good connectors and importantareas for conservation. If these important areas for conservation did

Table 2Markov land use transition probability matrix for 2007 based on the cross-tabulationof preceding land use quantities (1991 and 1999), it is read from left (1999) to top(2007). Markov transition probability matrices for 2005 and 2015 were similar tothose of 2007.

1999 2007

Urban Agriculture Vegetation Freshwater Greenhouses

Urban 0.8927 0.0530 0.0286 0.0014 0.0243Agriculture 0.0097 0.9085 0.0572 0.0021 0.0225Vegetation 0.0064 0.0673 0.9050 0.0010 0.0203Freshwater 0.0014 0.0100 0.0867 0.9018 0.0000Greenhouses 0.0123 0.0916 0.0190 0.0003 0.8768

not fall inside the protected area networks, they were identified aspotential conservation gaps.

Results

Land use projections

The Markov land use transition probability matrix for 2007showed that around 10% of the area of each land-use class wasexpected to change from 1999 to 2007 (Table 2). The greatestprobabilities of change were observed for the transitions fromnatural vegetation to agriculture, from freshwater to natural veg-etation, from agriculture to natural vegetation, from greenhousesto agriculture and from urban areas to agriculture. Markov tran-sition probabilities of change for 2005 and 2015 were similarto those of 2007, especially for natural vegetation changes. Interms of area, the rate of expansion in agricultural land increased(100 km2 in 1991–1999 and 234 km2 in 2007–2015). The sameincreasing trend with a remarkable acceleration was found forgreenhouse developments (55 km2 increase in 1991–1999 and177 km2 in 2007–2015) and urban activities (23 km2 increase dur-ing 1991–1999 and 102 km2 during 2007–2015).

According to our projections, between 1991 and 2015, landuse expansion throughout the study region will be mainly at theexpense of natural vegetation, as agriculture (6.2% increase), green-houses (4.2% increase), and urban areas (1.6% increase) all expand(Fig. 3). Land abandonment (and subsequent natural vegetationrecovery, 32 km2) was not very widespread in our projections andwas restricted to a few spots in rural interior mountains or insideprotected areas (Fig. 2b). The major areas of agricultural expansionwere observed in the interior eco-regions, while the constructionof new greenhouses was clustered in coastal eco-regions (Table 3).Urbanisation was also concentrated along the coast (Fig. 3). Accord-ingly, most of the expected vegetation loss from 1991 to 2015(1100 km2) occurs in coastal eco-regions (Table 3).

The validation of our land-use projections for 2005 showedgood reliability of the land-use maps (overall area-adjusted accu-racy ∼77.2%; percentage of landscape with quantity disagreement∼8.50%; percentage of landscape with allocation disagreement∼12%). Classification accuracies differed among land-use classes(Table 1). Land-use classes with a large spatial extent (natural veg-etation and agriculture) showed higher producer’s accuracies thanland uses with lower surface (freshwater, urban areas, and green-houses, Table 1). The 2005 projection agreed with field observationsin 189 points and disagreed in 51 points.

Changes in habitat fragmentation and connectivity

Our results suggest that the landscape of Almería provincewill become more fragmented from 1991 to 2015. According tothe projected land-use changes, the area of core natural vegeta-tion decreases by 6.5% (1077 km2) covering 64% of the provincein 2015, edge area increases by 6.1% (92 km2), and bridges (0.18%



Fig. 3. Land use maps for 1991, 1999 (both inputs) and 2007 and 2015.

change) and islets (0.07% change) remain relatively stable (Fig. 4).The total number and the area of perforation elements decreased(from 6321 in 1991 to 2558 in 2015, and 4.19% in 1991 to 3.67% in2015, respectively), as perforation elements tend to be convertedinto edge elements (Table 3). The greatest increase in edge areaoccurred in mountainous eco-regions (Surco-Bético Oriental, Sierra

de Baza-Filabres and Sierra de Gádor) and in the coastal eco-regionof Cuencas Litorales Almerienses (Table 3). The decline of islets(Fig. 4) mainly occurred in the coastal areas (Fig. 5b), for examplein the Cuencas Litorales Almerienses eco-region where 1% of theregion experiences loss of islets which means a 42% disappearanceof islets and a loss of 22.6% of total islets. In general, bigger islets

Table 3Relative changes (%) from 1991 to 2015 in the area occupied by each land use and fragmentation element in nine different eco-regions as simulated by a Markovian cellularautomaton model.

Increase in the provincial area (%)

2015 Interior eco-regions Coastal eco-regions

Total Surco-Betico C. Alto S. Betica Sierra Sierra C. L. F. L. Sierras F. L.(sq km) Oriental Almanzora Externa Almeriense Baza Filabres Gádor Almeriense Contraviesa Prelitorales Cabo de Gata

Agriculture expansion 715 12 10 10Urban constructions 153 4 3Greenhouse developments 385 17 60 16Vegetation recovery 32Vegetation loss 1100 23 59.6 20

Perforations 18.2 23.3 19Perforations to edge 19.6 17.6 16.7Core to edge 20.7 14.2 15.2



Fig. 4. Temporal changes in fragmentation components in the study area.

tend to persist in the landscape (432 islet patches covering 1.25%area in 1991 vs. 288 patches covering 1.65% area in 2015).

As a consequence of the projected increase in landscape frag-mentation, structural connectivity decreased in terms of smallerarea of the most important connector with increasing connectiv-ity importance through time (Table 4). Accordingly, the numberof habitat connectors (nodes and links) lowered through time(Table 4) and distances among connectors increased (Table 4). Inaddition, our results indicate that the location of the most impor-tant nodes to preserve connectivity changes (Fig. 5a). In 1991, the

Table 4Spatio-temporal characteristics of structural landscape connectors.

Year Highest nodeimportance

Area most imp.node (sq km)

No. elements(nodes and links)

Observed meandistance amongelements (m)

1991 0.8 3019.86 1056 871.51999 4.075 182.73 1093 914.112007 6.00 60.85 815 1008.752015 13.04 23.66 480 1506.86

most important node was found in the southern part of the studyarea (Fig. 5a) while in 1999 the most important node was locatedin the upper part of the Almanzora basin (Fig. 1) and the relativeimportance of the southern node decreased. Results for 2007 high-light an important area for connectivity in the lower part of theAlmanzora basin. Finally, in 2015 the most important area for con-nectivity was located in and northwest of Sierra de Bédar (Fig. 5a).

The role of protected area networks in preserving landscapeconnectivity

Overlaying the RENPA network to the map of landscape connec-tors (Fig. 5a) showed that this network largely fails to protect thenatural vegetation elements (nodes and links) that are most impor-tant for enhancing or maintaining landscape connectivity in 1999and 2007. However, although habitat priorisation principles of theEU Habitats Directive do not consider the protection of connectivityelements (nodes and links), RENPA protected areas include morethan half (68.5%) of the landscape connectors that were found tobe important in 1999 (Fig. 6). Although the Natura 2000 network

Fig. 5. Increase in habitat fragmentation and spatial shifts in the landscape elements important for connectivity between 1991 and 2015. (a) Change in the structuralconnectivity elements and (b) change in the location of the most important landscape connector.



0

10

20

30

40

50

60

70

80

0

200

400

600

800

1000

1200

1400

1600

1800

2000

1991 1999 2007 2015

Per

centa

ge

(%)

of

the

tota

l ar

ea o

fla

ndsc

ape

connec

tors

under

pro

tect

ion

Are

a (s

qK

m)

of

lan

dsc

ape

connec

tors

under

pro

tect

ion

RENPA PAs

Natura 2000 PAs area

Natura 2000 Network area

RENPA PAs %

Natura 2000 PAs%

Natura 2000 Network %

Fig. 6. Comparison of the area of important connectors within protected areas (PAs).In 1991 and 1999 there are no data for the Natura 2000 protected areas because thisnetwork was established in 2006.

would preserve a larger area of the important landscape connectors(1493.2 km2 in 2015, Fig. 6) than the RENPA network (677.1 km2,Fig. 6), a smaller share of the connector elements will be protectedin 2015 by the Natura 2000 Network (36.2% by Natura 2000 vs.46.7% by RENPA protected areas, Fig. 6). In short, our results showthat the protection of landscape connectors (total area and pro-portion) tend to decline for both the RENPA and the Natura 2000networks, as the location and extent of natural vegetation elementscrucial for overall landscape connectivity change (Fig. 6).

Despite the restrictions regarding land use within protectedareas, our model suggests that an additional area representing19.31% of the natural vegetation is expected to be lost between1991 and 2015 in the province of Almería, of which 100.54 km2

would be located in RENPA protected areas and 104.54 km2 inNatura 2000 protected areas. Among the previous 19.31% vege-tation lost, a net loss of 480 km2 of EU-priority habitats (mainlylocated near coastal Natura 2000 protected areas) is projected, outof which 108 km2 would be located inside the Natura 2000 network.The main vegetation losses in the vicinities of protected areas wereprojected to be distributed around the first 2 km of the boundariesof Natura 2000 protected areas (29.2% of vegetation lost, Fig. 2a).Yet, the general tendency in our results was towards an increasein vegetation loss with increasing distance to the protected areasboundary (Fig. 2a).

Averaging the importance for connectivity over the whole studyperiod (Fig. 7) revealed that the Natura 2000 network would stillpreserve more landscape connector area (2739.9 km2) than theRENPA network (1354.9 km2). We detected two regions that couldenhance the protected areas’ network connectivity substantiallywere they protected from intensive land conversion in the future:(1) the region northwest of the Sierra de Cabrera and Bédar pro-tected area; and, (2) the region between the Sierra de Cabrera andBédar and the Sierra Alhamilla protected areas (Fig. 7). The areaconnecting these two regions contains EU-priority habitats (Fig. 1)and their protection would not necessarily need to restrict futuredevelopment of the region.

Fig. 7. Average importance of connectors across the study period 1991–2015. Landscape connectors in dark grey highlight their persistence in time as good landscapeconnectors. Landscape connectors in black highlight their importance in maintaining structural connectivity. (1) Sierra Alhamilla protected area (PA) and (2) Sierra deCabrera and Bedar protected area.



Discussion

Our land-use model projects continued vegetation loss in south-east Spain, mainly due to the assumed expansion of agriculturalactivities, which is in strong agreement with other studies. Forexample, Munoz-Rojas et al. (2011) showed that the main land-use changes for the south of Spain were an increase in intensiveagriculture between 1956 and 2007. Our results also suggest thatthe threat of natural vegetation losses is highly clustered through-out the study area and especially widespread in the coastal areas(Cuencas Litorales Almerienses, due to greenhouse and urbandevelopments) and in the interior eco-regions (Surco-Betico Orien-tal, due to agriculture, Table 3). Overall, natural habitats in Almeríacontinue to be fragmented heavily, resulting in an increasing isola-tion of the remaining patches (Piquer et al. 2004). Our land-usemodel was a ‘Business as Usual’ type of scenario and furtherresearch should assess the range of alternative land-use scenariosthat may occur in Almería province, considering socio-economicdrivers (e.g., continued rural exodus), climate change, and policyreforms such as the upcoming CAP reform in 2013. In this con-text, European-wide future scenarios and land-use maps have astrong potential for conservation which is far from fully exploited;however, they are still too coarse in scale for regional planning.

Fragmentation did not occur uniformly throughout our studyregion and understanding the spatial variation in fragmentationprocesses is therefore a major step towards choosing appro-priate conservation management actions (Riitters and Coulston2005). Here, habitat fragmentation was mainly characterised by anincrease in edge elements in interior mountainous areas (e.g., dueto the creation of holiday houses and sparse greenhouse develop-ments) and a loss of islets in coastal eco-regions (due to massiverecreational complexes for tourism and a dense fabric of green-houses, especially in the surroundings of protected areas, Fig. 5).Both fragmentation and isolation processes have important eco-logical consequences. Increasing edge length translates into moreedge effects (e.g., species invasions, pollution, higher predation)(Murcia 1995), which especially affects species habitat specialists.Losing islets means that stepping stones acting as refugia or facil-itating dispersal or genetic flows are lost (Cox and Underwood2011). This process especially threatens those ecosystems dis-tributed along coastal eco-regions (Fig. 5), such as the endemicarborescent matorral of Ziziphus lotus, a priority habitat under theHabitat Directive (code 5220) close to extinction in Europe due tothe ongoing greenhouse expansion in the study area (Mota et al.1996). The relationship between fragmentation, structural connec-tivity and species extinction risk is not linear but characterised bythresholds (Fahrig 2001 2002 2003; Metzger and Décamps 1997).Determining these thresholds for threatened species, and assessingwhether such thresholds would be breached based on future land-use protections, such as those we generated here, are importantsteps towards sustainable land-use systems that align productionand conservation goals.

While the study showed that the Natura 2000 network was rel-atively effective in safeguarding natural vegetation patches until2015 inside Natura 2000 protected areas (where some land-usepractices are allowed), a main result from the projections is also thatoverall landscape connectivity of natural habitats could stronglydecline. A major reason for this likely to be that conservation plan-ning has so far not adequately considered landscape connectivityand potential changes therein. This is surprising given that pre-venting habitat fragmentation and the isolation of protected areasin regions with rapid land-use changes is one of the top conserva-tion issues in Europe (Heller and Zavaleta 2009; Pullin et al. 2009)and important in light of accelerating climate change (Heller andZavaleta 2009). Indeed, the European Habitats Directive takes intoaccount the importance of connecting elements in the landscape

(article 10 of the Habitat Directive). However, the definition of con-servation status of natural habitats lack guidelines (Mehtälä andVuorisalo 2007; Velázquez et al. 2010) and priority conservationareas are defined solely based on expert knowledge, without usinga systematic approach to evaluate the future risk of habitat loss,often resulting in a patchwork of protected areas (Gurrutxaga et al.2010; Maiorano et al. 2007).

Increasing protected area isolation in coastal regions is furtherhighlighted by the finding that although most land-use changesin the study area occurred outside protected areas (Fig. 2b), high-est vegetation loss rates were projected for the land closest toNatura 2000 protected areas (i.e., within the first 2 km, Fig. 2a). Thisis in agreement with Piquer et al. (2004) that detected intensiveland uses and vegetation losses in the surroundings of coastal pro-tected areas in Almería. One explanation for this is the expansionof the protected area network itself. As more land is designated toreserves, the remaining natural vegetation areas increasingly areunder much higher risk of conversion. Ultimately, this calls intoquestion the current conservation paradigm on (relatively strictly)protected areas if such a paradigm leads to an increasing isolationof natural habitats. In mountainous areas the situation was slightlydifferent, with vegetation cover staying relatively stable throughtime. However, land use in the mountains is not economically prof-itable and therefore rare.

The connectivity analysis also revealed that regional conser-vation planning has so far not considered the most importantlandscape connectors for natural habitat networks in southeastSpain (Fig. 5 a), although substantial areas of landscape elementsproviding a connector function are within the RENPA network ofprotected areas in Almería (Fig. 6). Furthermore, the Natura 2000network of protected areas, although covering a much greater area,would include proportionately less important connector elementsgiven the land-use projections (Fig. 7). These results highlight thechallenges that conservation planners face, because both currentand potential future land use need to be considered when decidingupon conservation measures. This is not easy in regions where landuse is dynamic, and where opposite trends, such as land-use expan-sion (natural vegetation lost, Fig. 2b) and abandonment (naturalvegetation gain, Fig. 2b) of land use take place side by side.

The results from the connectivity analyses show interestingdynamics in the central upper part of the province (Almanzora riverbasin, Fig. 1), where the projections indicate a marked shift in patchimportance for connectivity due to land-use change (Fig. 5a). Thisriver basin has experienced one of the most aggressive develop-ment trends in South Spain (H. Castro, pers. comm.), because ofintensive mining for marble in the west, urban constructions in thecentre, and irrigation of extensive cropland expansion in the lowerparts of the basin (mining is part of the urban class, see Table 1).The drastic changes in overall landscape connectivity due to theseintensive land uses further illustrates the lack of integrated man-agement of the basin where all pressures and outcomes (includingoverall habitat connectivity) are evaluated when assessing theenvironmental impact of development. Including feedback mech-anisms (Claessens et al. 2009) or implementing multi-objectivemanagement measures such as the Ecosystem Based Managementapproach (Curtin and Prellezo 2010) or cumulative environmen-tal impact assessment, especially in the context of the HabitatsDirective’s Appropriate Assessments (Söderman 2009), could be animportant step towards integrated landscape management in suchcases (Crain et al. 2009).

What are the conservation implications of this study? Thereexists a vast array of land-use models that predict land-use changesinto the future without further application into the planning con-text. Often, these models do not generate land-use maps at scalesfine enough and across areas large enough for conservation plan-ning. Where available, land-use and conservation planning also



does not make full use of these projections. Cost-effective conserva-tion strategies must prioritise places that are of high conservationvalue but that at the same time are threatened (Brooks 2001). Ourstudy further highlights the value of combining land-use mod-els and conservation planning tools to identify such areas at risk(Reyers 2004; Rouget et al. 2003; Wessels et al. 2003), and to iden-tify dynamics in important hotspots of landscape connectors thatwould need to be protected to ensure the ecological integrity ofthe landscape as a whole. Our study showed three main regionsthat lack protection and that could be important to integrate in theprotected area network:

1. The Cuencas Litorales Almerienses (coastal basins of Almería)eco-region that faces high vegetation loss due to greenhouseand urban development, very high risk of patch (islets) vege-tation being lost, and a strong trend towards conversion fromperforation to edge elements.

2. The northwestern surroundings of the Sierra de Cabrera andBédar protected area (Fig. 7) one of the most important connect-ing elements in our analysis, and where edge habitat increasesdramatically by 2015 (Fig. 5b); and

3. The region connecting the Sierra de Cabrera and Bédar protectedarea and Sierra Alhamilla protected area (Fig. 7), a region whereland use resulted in much natural vegetation loss and an increasein edge habitat.

For these three areas we urge regional conservation managersand landscape planners to closely investigate the conservationstatus, fragility and future development options in light of the out-standing importance for the habitat network of Almería province asa whole. Landscape connecting patches sometimes fall outside pro-tected areas (such as in 1999 and 2007 in Almería), and elementsimportant for connectivity may change through time. Generally,we therefore recommend revising conservation planning in theAlmería province to (1) reinforce protected area networks (spe-cially coastal Natura 2000 protected areas for southeast Spain) bycautiously extending the protected area system (see above), (2) toreevaluate the current focus on protected areas alone to expandthe scope of conservation planning beyond the boundaries of pro-tected areas (Araújo et al. 2007; Santos et al. 2008), with a specialemphasise on overall landscape connectivity, and (3) to considerfuture landscape connectivity in a holistic manner (Naveh 2000) toavoid social conflicts (Roca et al. 2011) and safeguard the integrityof ecosystems.

In combining a land-use simulation model and connectivityanalyses we have shown that even relatively simple tools (cellu-lar automata of Markov chains and morphological spatial patternanalyses), if used jointly, can substantially advance understandingof hotspots of landscape connectivity and how they may changein the future. Such approaches are urgently needed to allow con-servation planners to evaluate the effectiveness of protected areanetworks and future threats to conservation. Preserving land-scape connectivity is an increasingly important conservation goalin an ever more dynamic world, and approaches such as the oneimplemented here for southeast Spain highlight a way forwardfor conservation planning to deal with the uncertainties of thefuture.

Acknowledgements

The authors are grateful to A. Bregt, J. Crompvoets, R. Bunce, L.Daniele, K. Kok, Y. Cantón, P. Escribano and two anonymous review-ers for useful comments on earlier manuscript versions. We thankDr. Pontius for making available his accuracy assessment tools. Wegratefully acknowledge funding from the Regional Government of

Andalusia, Spain (project number RNM 1280), the European Com-mission (integrated project VOLANTE, FP7-ENV-2010-265104), theElsa-Neumann-Stipendium, and the Einstein Foundation.

References

Andam, K. S., Ferraro, P. J., Pfaff, A., Sanchez-Azofeifa, G. A., & Robalino, J. A. (2008).Measuring the effectiveness of protected area networks in reducing deforesta-tion. Proceedings of the National Academy of Sciences, 105(42), 16089–16094,doi:10.1073/pnas0800437105

Araújo, M. B., Lobo, J. M., & Moreno, J. C. (2007). The effectiveness of Iberian pro-tected areas in conserving terrestrial biodiversity. Conservation Biology, 21(6),1423–1432, doi:10.1111/j.1523-1739.2007.00827.x.

Araújo, M. B., Nogués-Bravo, D., Reginster, I., Rounsevell, M., & Whittaker, R. J. (2008).Exposure of European biodiversity to changes in human-induced pressures.Environmental Science & Policy, 11(1), 38–45.

Baranyi, G., Saura, S., Podani, J., & Jordán, F. (2011). Contribution of habitat patchesto network connectivity: Redundancy and uniqueness of topological indices.Ecological Indicators, 11(5), 1301–1310.

Barbier, E. B., Burgess, J. C., & Grainger, A. (2010). The forest transition: Towards amore comprehensive theoretical framework. Land Use Policy, 27(2), 98–107.

Beier, P., Spencer, W., Baldwin, R. F., & McRae, B. H. (2011). Toward best practicesfor developing regional connectivity maps. Conservation Biology, 25(5), 879–892,doi:10.1111/j.1523-1739.2011.01716.x.

Bell, E. J. (1974). Markov analysis of land use change—an application of stochas-tic processes to remotely sensed data. Socio-Economic Planning Sciences, 8(6),311–316, doi:10.1016/0038-0121(74)90034-2.

Blanca, G., Cabezudo, B., Cueto, M., Fernandez Lopez, C., & Morales Torres, C. (2009).Flora vascular de Andalucia Oriental Sevilla: Consejería de Medio Ambiente. Juntade Andalucía.

Borja Barrera, F., & Montes del Olmo, C. (2008). La gestión ecosistémica comoherramienta territorial para la toma de decisiones. Ecorregions e integraciónfuncional de carreteras y espacios naturales protegidos en Andalucía, Paper Pre-sented at the II Congreso Nacional de Medio Ambiente en Carreteras, Santander,Spain.

Brooks, D. (2001). Terrestrial ecoregions of the World: Chaco. Available from30 August, 2010 http://www.worldwildlife.org/wildworld/profiles/terrestrial/nt/nt0210 full.html.

Cabello, J. (2002). El reto de la conservación de la biodiversidad. In S. López, J. Cabello,& M. Piquer-Rodríguez (Eds.), Agricultura, agua y sostenibilidad en la provincia deAlmería (p. p306). Almería: Junta de Andalucía; Posidonia.

Card, D. (1982). Using known map category marginal frequencies to improve esti-mates of thematic map accuracy. Photogrammetric Enginneering and RemoteSensing, 48(3), 431–439.

CBD. (2010). Global Biodiversity Outlook 3. Montréal: Secretariat of the Conventionon Biological Diversity., p. 94

Claessens, L., Schoorl, J. M., Verburg, P. H., Geraedts, L., & Veldkamp, A. (2009). Mod-elling interactions and feedback mechanisms between land use change andlandscape processes. Agriculture, Ecosystems & Environment, 129(1–3), 157–170.

Clarke, K. C., & Gaydos, L. J. (1998). Loose-coupling a cellular automaton modeland GIS: Long-term urban growth prediction for San Francisco and Washing-ton/Baltimore. International Journal of Geographical Information Science, 12(7),699–714.

Cochran, W. (1977). Sampling techniques (3rd ed.). New York: Wiley.Cox, R. L., & Underwood, E. C. (2011). The importance of conserving biodiversity

outside of protected areas in mediterranean ecosystems. PLoS One, 6(1), e14508.Crain, C. M., Halpern, B. S., Beck, M. W., & Kappel, C. V. (2009). Understand-

ing and managing human threats to the coastal marine environment. Annalsof the New York Academy of Sciences, 1162(1), 39–62, doi:10.1111/j.1749-6632.2009.04496.x.

Curtin, R., & Prellezo, R. (2010). Understanding marine ecosystem based manage-ment: A literature review. Marine Policy, 34(5), 821–830.

del Barrio, G., Puigdefabregas, J., Sanjuan, M. E., Stellmes, M., & Ruiz, A.(2010). Assessment and monitoring of land condition in the IberianPeninsula, 1989–2000. Remote Sensing of Environment, 114(8), 1817–1832,doi:10.1016/j.rse.2010.03.009.

Eastman, J.R. (2003). IDRISI Kilimanjaro Tutorial.Fahrig, L. (2001). How much habitat is enough? Biological Conservation, 100(1),

65–74.Fahrig, L. (2002). Effect of habitat fragmentation on the extinction threshold: A

synthesis. Ecological Applications, 12(2), 346–353.Fahrig, L. (2003). Effects of habitat fragmentation on biodiversity. Annual Review of

Ecology Evolution and Systematics, 34, 487–515.Foody, G. M. (2002). Status of land cover classification accuracy assessment. Remote

Sensing of Environment, 80(1), 185–201.Giam, X., Bradshaw, C. J. A., Tan, H. T. W., & Sodhi, N. S. (2010). Future habitat

loss and the conservation of plant biodiversity. Biological Conservation, 143(7),1594–1602.

Gurrutxaga, M., Lozano, P. J., & del Barrio, G. (2010). GIS-based approach for incor-porating the connectivity of ecological networks into regional planning. Journalfor Nature Conservation, 18(4), 318–326.

Heller, N. E., & Zavaleta, E. S. (2009). Biodiversity management in the face of cli-mate change: A review of 22 years of recommendations. Biological Conservation,142(1), 14–32.



Jongman, R. H. G., Bunce, R. G. H., & Mateus, V. L. (2006). Objectives and applica-tions of a statistical environmental stratification of Europe. Landscape Ecology,21, 409–419.

Joshi, P., Yadav, K., & Sinha, V. (2011). Assessing impact of forest landscape dynam-ics on migratory corridors: A case study of two protected areas in Himalayanfoothills. Biodiversity and Conservation, 20(14), 3393–3411, doi:10.1007/s10531-011-0123130.

Junta de Andalucía (Producer). 2009. Retrieved from http://www.juntadeandalucia.es/turismocomercioydeporte/export/sites/ctcd/archivos/estadisticas/estadisticas-oficiales-de-ctcd/balance 2009 def.pdf. (08.08.2011).

Junta de Andalucía. Consejería de Medio Ambiente (Producer, 1999). Retrievedfrom http://www.juntadeandalucia.es/medioambiente/documentos tecnicos/uso cobertura/12 56.pdf (17.07.2011).

Keys, E., & McConnell, W. J. (2005). Global change and the intensification of agricul-ture in the tropics. Global Environmental Change Part A, 15(4), 320–337.

Kindlmann, P., & Burel, F. (2008). Connectivity measures: A review. Landscape Ecol-ogy, 23(8), 879–890, doi:10.1007/s10980-008-9245-4

Knight, A. T., Driver, A., Cowling, R. M., Maze, K., Desmet, P. G., Lombard, A. T., et al.(2006). Designing systematic conservation assessments that promote effectiveimplementation: Best practice from South Africa. Conservation Biology, 20(3),739–750, doi:10./j.1523-1739.2006.00452.x.

Kuemmerle, T., Olofsson, P., Chaskovskyy, O., Baumann, M., Ostapowicz, K., Wood-cock, C. E., et al. (2011). Post-Soviet farmland abandonment, forest recovery, andcarbon sequestration in western Ukraine. Global Change Biology, 17, 1335–1349.

Levin, J., & Fox, J. A. (2004). Elementary statistics in social research. Boston: The Essen-tials.

MA. (2005). Ecosystems and human well-being: Synthesis. Washington D.C: IslandPress., p. 137.

MacDonald, D., Crabtree, J. R., Wiesinger, G., Dax, T., Stamou, N., Fleury, P., et al.(2000). Agricultural abandonment in mountain areas of Europe: Environmentalconsequences and policy response. Journal of Environmental Management, 59(1),47–69.

Maiorano, L., Falcucci, A., Garton, E. O., & Boitani, L. (2007). Contribution of the Natura2000 Network to Biodiversity Conservation in Italy. Conservation Biology, 21(6),1433–1444, doi:10.1111/j.1523-1739.2007.00831.x.

Margules, C. R., & Pressey, R. L. (2000). Systematic conservation planning. Nature,405(6783), 243–253.

Médail, F., & Quézel, P. (1999). Biodiversity hotspots in the Mediterranean Basin:Setting global conservation priorities. Conservation Biology, 13(6), 1510–1513,doi:10.1046/j.1523-1739.1999.98467.x.

Mehtälä, J., & Vuorisalo, T. (2007). Conservation policy and the EU Habitats Directive:Favourable conservation status as a measure of conservation success. EuropeanEnvironment, 17(6), 363–375, doi:10.1002/eet.458.

Metzger, J.-P., & Décamps, H. (1997). The structural connectivity threshold: Anhypothesis in conservation biology at the landscape scale. Acta Oecologica, 18(1),1–12.

Moilanen, A., Wilson, K. A., & Possingham, H. P. (2009). Spatial conservation prior-ization quantitative methods & computational tools. Oxford: Oxford UniversityPress.

Mota, J. F., Penas, J., Castro, H., Cabello, J., & Guirado, J. S. (1996). Agricultural devel-opment vs biodiversity conservation: The Mediterranean semiarid vegetationin El Ejido (Almería, southeastern Spain). Biodiversity and Conservation, 5(12),1597–1617, doi:10.1007/bf00052118.

Munoz-Rojas, M., De la Rosa, D., Zavala, L. M., Jordán, A., & Anaya-Romero, M. (2011).Changes in land cover and vegetation carbon stocks in Andalusia, Southern Spain(1956–2007). Science of The Total Environment, 409(14), 2796–2806.

Murcia, C. (1995). Edge effects in fragmented forests: Implications for conservation.Trends in Ecology & Evolution, 10(2), 58–62.

Naveh, Z. (2000). What is holistic landscape ecology? A conceptual introduction.Landscape and Urban Planning, 50(1–3), 7–26.

Newmark, W. D. (2008). Isolation of African protected areas. Frontiers in Ecology andthe Environment, 6(6), 321–328.

NRCS. (2000). Markov matrices of landscape change. Ecological Indicators for the Nation.Washington DC, USA: National Research Council Staff. National Academies Press.

Penas, J., Benito, B., Lorite, J., Ballesteros, M., Canadas, E., & Martinez-Ortega, M.(2011). Habitat fragmentation in arid zones: A case study of Linaria nigricansunder land use changes (SE Spain). Environmental Management, 48(1), 168–176,doi:10.1007/s00267-011-9663-y.

Perez-Vega, A., Mas, J. F., & Ligmann-Zielinska, A. (2012). Comparing two approachesto land use/cover change modeling and their implications for the assessmentof biodiversity loss in a deciduous tropical forest. Environmental Modelling &Software, 29(1), 11–23, doi:10.1016/j.envsoft.2011.09.011.

Piquer, M., Caravias, A., Sánchez-Alcaraz, J., Alcaraz-Segura, D., & Cabello, J. (2004).Dinámica de los usos del territorio en el entorno del Parque Natural de Cabode Gata-Níjar. In J. Penas, & L. Gutiérrez (Eds.), Biología de la Conservación:Reflexiones, propuestas y estudios desde el SE Ibérico (p. 56). Almería: Institutode Estudios Almerienses. Colección Actas.

Pontius, R., Boersma, W., Castella, J.-C., Clarke, K., de Nijs, T., Dietzel, C., et al. (2008).Comparing the input, output, and validation maps for several models of landchange. The Annals of Regional Science, 42(1), 11–37, doi:10.1007/s00168-007-0138-142.

Pontius, R. G., Cornell, J. D., & Hall, C. A. S. (2001). Modeling the spatial patternof land-use change with GEOMOD2: Application and validation for Costa Rica.Agriculture, Ecosystems and Environment, 85(1–3), 191–203, doi:10.1016/s0167-8809(01)00183-189.

Pontius, R. G., & Millones, M. (2011). Death to Kappa: Birth of quantity disagreementand allocation disagreement for accuracy assessment. International Journal ofRemote Sensing, 32(15), 4407–4429, doi:10.1080/01431161.2011.552923.

Pullin, A. S., Báldi, A., Can, O. E., Dieterich, M., Kati, V., Livoreil, B., et al. (2009).Conservation focus on europe: Major conservation policy issues that need tobe informed by conservation science. Conservation Biology, 23(4), 818–824,doi:10.1111/j.1523-1739.2009.01283.x.

Radeloff, V. C., Stewart, S. I., Hawbaker, T. J., Gimmi, U., Pidgeon, A. M., Flather, C. H.,et al. (2010). Housing growth in and near United States protected areas limitstheir conservation value. Proceedings of the National Academy of Sciences, 107(2),940–945, doi:10.1073/pnas.0911131107.

Reyers, B. (2004). Incorporating anthropogenic threats into evaluations of regionalbiodiversity and prioritisation of conservation areas in the Limpopo Province,South Africa. Biological Conservation, 118(4), 521–531.

Riitters, K. H., & Coulston, J. W. (2005). Hot spots of perforated forest in the easternUnited States. Environmental Management, 35(4), 483–492.

Roca, E., Villares, M., & Fernández, E. (2011). Dealing with conflicts between con-servation and socioeconomic activities. A social perception study in the CostaBrava, Spain. Littoral 2010, 8 (Adapting to global change at the coast: Leadership,innovation, and investment).

Rouget, M., Richardson, D. M., Cowling, R. M., Lloyd, J. W., & Lombard, A. T. (2003).Current patterns of habitat transformation and future threats to biodiversityin terrestrial ecosystems of the Cape Floristic Region. South Africa. BiologicalConservation, 112(1–2), 63–85.

Sánchez-Picón, A. (2005). De frontera a milagro. La conformación histórica de laeconomía almeriense. In J. Molina Herrera (Ed.), La Economía de la provincia deAlmería (p. 84). Almeria: Instituto de Estudios de Cajamar.

Santos, K. C., Pino, J., Rodà, F., Guirado, M., & Ribas, J. (2008). Beyond thereserves: The role of non-protected rural areas for avifauna conservation inthe area of Barcelona (NE of Spain). Landscape and Urban Planning, 84(2),140–151.

Saura, S., & Rubio, L. (2010). A common currency for the different ways in whichpatches and links can contribute to habitat availability and connectivity in thelandscape. Ecography, 33(3), 523–537, doi:10.1111/j.1600-0587.2009.05760.x.

Saura, S., & Torné, J. (2009). Conefor Sensinode 2.2: A software package for quantify-ing the importance of habitat patches for landscape connectivity. EnvironmentalModelling & Software, 24(1), 135–139.

Seiferling, I. S., Proulx, R., Peres-Neto, P. R., Fahrig, L., & Messier, C. (2011). Mea-suring protected-area isolation and correlations of isolation with land-useintensity and protection status. Conservation Biology, doi:10.1111/j.1523-1739.2011.01674.x.

Söderman, T. (2009). Natura 2000 appropriate assessment: Shortcomings andimprovements in Finnish practice. Environmental Impact Assessment Review,29(2), 79–86.

SOER. (2010). The European environment—state and outlook 2010: Biodiversity. Copen-hagen: European Environmental Agency (EEA)., p. 48.

Soille, P., & Vogt, P. (2009). Morphological segmentation of binary patterns. PatternRecognition Letters, 30(4), 456–459, doi:10.1016/j.patrec.2008.10.015.

Stehman, S. V., & Czaplewski, R. L. (1998). Design and analysis for thematic mapaccuracy assessment: Fundamental principles. Remote Sensing of Environment,64(3), 331–344.

Valle, F. (2004). Datos Botánicos aplicados a la gestión del medio natural andaluz II Seriesde vegetación. Modelos de restauración forestal Junta de Andalucía: Consejería deMedio Ambiente.

Velázquez, J., Tejera, R., Hernando, A., & Victoria Núnez, M. (2010). Environmentaldiagnosis: Integrating biodiversity conservation in management of Natura 2000forest spaces. Journal for Nature Conservation, 18(4), 309–317.

Verburg, P., & Overmars, K. (2009). Combining top-down and bottom-up dynamicsin land use modeling: Exploring the future of abandoned farmlands in Europewith the Dyna-CLUE model. Landscape Ecology, 24(9), 1167–1181.

Verburg, P. H., Kok, K., Pontius, R. G., & Veldkamp, A. (2006). Modelling land-use andland-cover change. In E. F. Lambin, & H. J. Geist (Eds.), Land use and land coverchange. Local processes and global impacts (pp. 117–136). Berlin, Heidelberg/NewYork: Springer Verlag.

Visconti, P., Pressey, R. L., Giorgini, D., Maiorano, L., Bakkenes, M., Boitani, L.,et al. (2011). Future hotspots of terrestrial mammal loss. Philosophical Trans-actions of the Royal Society B: Biological Sciences, 366(1578), 2693–2702,doi:10.1098/rstb.2011.0105.

Vogiatzakis, I. N., Mannion, A. M., & Griffiths, G. H. (2006). Mediterranean ecosys-tems: Problems and tools for conservation. Progress in Physical Geography, 30,doi:10.1191/0309133306.472ra.

Vogt, P., Ferrari, J. R., Lookingbill, T. R., Gardner, R. H., Riitters, K. H., & Ostapowicz,K. (2009). Mapping functional connectivity. Ecological Indicators, 9(1), 64–71,doi:10.1016/j.ecolind.2008.01.011.

Vogt, P., Riitters, K. H., Estreguil, C., Kozak, J., & Wade, T. G. (2007). Mapping spa-tial patterns with morphological image processing. Landscape Ecology, 22(2),171–177.

Wessels, K. J., Reyers, B., van Jaarsveld, A. S., & Rutherford, M. C. (2003). Identificationof potential conflict areas between land transformation and biodiversity con-servation in north-eastern South Africa. Agriculture, Ecosystems & Environment,95(1), 157–178.

Ziółkowska, E., Ostapowicz, K., Kuemmerle, T., Perzanowski, K., Radeloff, V.C., & Kozak, J. (2012). Potential habitat connectivity of European bison(Bison bonasus) in the Carpathians. Biological Conservation, 146(1), 188–196,doi:10.1016/j.biocon.2011.12.017.

Author's personal copy...Author's personal copy 328 M. Piquer-Rodríguezet al. / JournalforNature...

Documents

Transcript of Author's personal copy...Author's personal copy 328 M. Piquer-Rodríguezet al. / JournalforNature...