Exploring spatial patterns of vulnerability for diverse

Journal of Environmental Management 95 (2012) 9e16

Contents lists available at SciVerse ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

Exploring spatial patterns of vulnerability for diverse biodiversity descriptors in regional conservation planning Ruppert Vimal a, Pascal Pluvinet a, Céline Sacca b, Pierre-Olivier Mazagol b, Bernard Etlicher b, John D. Thompson a, * a b

Centre d’Ecologie Fonctionnelle et Evolutive, UMR 5175, CNRS, 1919 route de Mende, 34293 Montpellier cedex 5, France Environnement Ville Société, UMR 5600, CNRS, Université Jean Monnet, 6, rue Basses des Rives, 42023 Saint-Étienne cedex 2, France

a r t i c l e i n f o

a b s t r a c t

Article history: Received 19 May 2011 Received in revised form 2 August 2011 Accepted 26 September 2011 Available online 30 October 2011

In this study, we developed a multi-criteria assessment of spatial variability of the vulnerability of three different biodiversity descriptors: sites of high conservation interest by virtue of the presence of rare or remarkable species, extensive areas of high ecological integrity, and landscape diversity in grid cells across an entire region. We assessed vulnerability in relation to (a) direct threats in and around sites to a distance of 2 km associated with intensive agriculture, building and road infrastructure and (b) indirect effects of human population density on a wider scale (50 km). The different combinations of biodiversity and threat indicators allowed us to set differential priorities for biodiversity conservation and assess their spatial variation. For example, with this method we identified sites and grid cells which combined high biodiversity with either high threat values or low threat values for the three different biodiversity indicators. In these two classes the priorities for conservation planning will be different, reduce threat values in the former and restrain any increase in the latter. We also identified low priority sites (low biodiversity with either high or low threats). This procedure thus allows for the integration of a spatial ranking of vulnerability into priority setting for regional conservation planning. Ó 2011 Elsevier Ltd. All rights reserved.

Keywords: Conservation priorities Mediterranean Risk assessment Systematic conservation planning

1. Introduction The loss of natural areas as a result of land-use change has become a major threat to biodiversity (Sala et al., 2000; Wilcove et al., 1998). Around the world, over the last fifty years, agricultural intensification, urbanization, industrial land development and the construction of transport networks have given rise to an unprecedented destruction and fragmentation of natural environments (Stanners and Bourdeau, 1995), with an ever increasing human footprint on natural ecosystems (Sanderson et al., 2002). There has thus been growing recognition of the need to provide an explicit spatial framework to assess the vulnerability of species, habitats and ecological connectivity to this processes of landscape transformation (Noss et al., 2002; Pressey and Taffs, 2001; Rouget et al., 2003b; Schipper et al., 2008; Spring et al., 2010; Wilson et al., 2005). When combined with methods developed for priority setting in systematic conservation planning based on irreplacability and complementarity * Corresponding author. Tel.: þ33 (0) 4 67 61 32 14; fax: þ33 (0) 4 67 41 06 16. E-mail addresses: [email protected] (R. Vimal), [email protected] (P. Pluvinet), [email protected] (C. Sacca), pierre.olivier.mazagol@ univ-st-etienne.fr (P.-O. Mazagol), [email protected] (B. Etlicher), [email protected] (J.D. Thompson). 0301-4797/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2011.09.018

(Margules and Pressey, 2000; Pressey, 1994), the assessment of vulnerability and its spatial variability enables the establishment of conservation priorities for species and sites based not only on their conservation value but also on their probability of being lost (Brooks et al., 2006; Pressey and Taffs, 2001; Rodrigues et al., 2004). Vulnerability is an inclusive term which integrates three main dimensions: exposure to threats, the intensity or sensitivity to threats and the resilience or adaptability of ecological systems or human activities to the threats (Wilson et al., 2005). It can be integrated into spatial conservation planning by assessing threats to particular habitat types (Rouget et al., 2003b), rare or endangered species (Abbitt et al., 2000) and habitat connectivity (Spring et al., 2010). A comprehensive assessment of these threats is a difficult task as a complex array of factors is involved, making it necessary to adopt an approach based on multiple threat criteria (Noss et al., 2002). These should include, at least to a certain degree, the ecological and evolutionary processes required for biodiversity to persist in space and time (Balmford et al., 1998; Poiani et al., 2000; Pressey et al., 2003; Rouget et al., 2003a; Smith et al., 1993). In this study, our objective is to define conservation priorities in the Mediterranean region of southern France based on exposure to different threat processes with three different biodiversity

10

R. Vimal et al. / Journal of Environmental Management 95 (2012) 9e16

descriptors. To do so, we first quantify local scale threats linked to agricultural intensification, urbanization and road infrastructures and a more diffuse estimate of threat based on human population density in a wider area. We then assess how exposure to these threats is spatially distributed in relation to three different biodiversity descriptors: (i) small sites of high ecological interest that contain rare or remarkable species, (ii) extensive areas of high ecological integrity that often surround the former and (ii) an estimate of fine-scale landscape diversity across the entire region. This combination of diverse threats and multiple biodiversity descriptors allows us to progress toward the integration of a comprehensive assessment of vulnerability in regional conservation planning. 2. Methods 2.1. The study region This study was carried out in the Languedoc Roussillon region (27 376 km2) in southern France (Fig. 1), which encompasses most of the Mediterranean region west of the Rhône valley. The main landscape types which occur in this region are coastal landscapes with lagoons, marshes, cliffs and dunes, lowland garrigues often as a mosaic with cultivated areas, vast areas of vines, extensive upland limestone plateau areas, and hilly or mountainous landscapes on granite and schist in the southern tip of the Massif Central and the south-eastern Pyrenees. In the last 50 years profound modifications to the landscape of the region have occurred. First, extensive and rapid urbanization has occurred around towns and villages across the lowland plains and in conjunction with massive proliferation of coastal tourist resorts. Second, human population decline in rural areas has been accompanied by the abandonment of vineyards and grazing activity in many areas, changes which have set the scene for rapid natural reforestation of fields (IFEN, 2003). 2.2. Biodiversity descriptors We considered three biodiversity descriptors: landscape diversity (LD), species of high regional conservation interest (CI) and

Fig. 1. The location of the study region (in dark grey).

ecological integrity (EI). The former is based on grid cells across the whole study region, the latter two are based on an inventory of sites of high ecological interest conducted at the national scale which identified key sites in our study region. 2.2.1. Landscape diversity (LD) A landscape diversity index based on a grid scale of 100 m 100 m across the whole study region was developed using five criteria. (i) Land-use heterogeneity in each grid cell was calculated with a Shannon diversity index using pixel neighborhood analysis of Corine Landcover, with 12 12 moving window and the “GRASS r.le” module. (ii) The density of watercourses which promote regional biodiversity due to the interface of terrestrial and aquatic environments (Naiman et al., 1993) was quantified based on the Carthage database and using a “line density” tool included in the “spatial analyst” extension for “Esri ArcGIS 9.3”. (iii) The national geological map (1:1 000 000) of France was used to establish a typology with six categories of subsoil based on substrate nature, geochemistry and interest for overall floristic diversity. (iv) Anthropogenic impact was quantified as the difference between the observed vegetation and its climax with a range of values from zero (for urbanization) to 5 (for climax vegetation). The expected climax vegetation was obtained from vegetation maps of France (1:200 000; Ozenda, 1985). (v) The rarity of a given environment was estimated by the ratio between its surface in a given grid cell to that in the total study area. Each criterion was normalized and a multi-criteria evaluation (MCE) module in Idrisi, with a Weighted Linear Combination (WLC) function was used to produce the LD index. We ranked grid cells with high (the 20% of sites with the highest values), intermediate (the 40% of sites with intermediate values) and low (the 40% of sites with the lowest values) values of LD. 2.2.2. Sites with species of high conservation interest (CI) As part of the national inventory of zones of high ecological interest in each region of France, a list of priority species has been established in each region (INPN, 2006). To identify the list of priority species, regional specialists weighted and noted each species of a given taxonomic group following a general methodology to define their conservation interest (Dreal, 2009). A score between zero and eight was attributed to each species according to their rarity and to the regional responsibility for such species; Species with an international and national protection status were automatically included. Plants with a score between two and eight and animals with a score between three and eight were selected as priority species for the delimitation of 793 sites (16.8% of the surface of the study region). We used the inventory of these sites for our study in which we used taxonomic groups for which we had a sufficient knowledge of the regional distribution and abundance, namely 629 vascular plant species, and 112 vertebrate species (fishes, n ¼ 14; reptiles, n ¼ 8; amphibians, n ¼ 6; birds, n ¼ 61; mammals, n ¼ 23). Only species which reproduce in a site were considered to be present. Non-native species were not assessed. To assign priorities we assessed the irreplaceability value of all sites. We used the scores attributed in the regional inventory to define four species categories of conservation interest, each with specific conservation targets. Analyses were performed using an adaptive annealing schedule in the “Marxan” software (Ball and Possingham, 2000; Possingham et al., 2000). Neither boundary parameter nor surface area costs were used. Irreplaceability for a given site is equal to the number times in which a site is selected among the 1000 solutions obtained. We defined a conservation target as the minimum coverage of a species range to be included in priority areas. For the purpose of this study (of course values can be adapted to local contexts) we adopted a target of 80% for species with


a score of 7 or 8 in the regional inventory (46 plants, 6 mammals, 10 birds, 4 herptiles and 2 fish species), 40% for species with a score of 5 or 6 (78 plants, 10 mammals, 15 birds, 3 herptiles and 6 fish species), 10% for species with a score ¼ 3 or 4 (132 plants, 7 mammals, 36 birds, 7 herptiles and 6 fish species), 5% for the 373 plant species with a score ¼ 2. Such targets are important for the identification of gaps in current protection and for priorities (see also Vimal et al., 2011), so we included them here for the calculation of irreplaceability. We ranked sites according to their irreplaceability value in three classes: high (irreplaceability value > 900), intermediate (irreplaceability value of 200e900) and low (irreplaceability value < 200). Since many sites had either a high or low irreplaceability value the intermediate class had a wider range than the two extreme classes. 2.2.3. Ecological integrity (EI) sites As part of the above inventory, sites of ecological integrity which go beyond the boundaries of the CI sites were also designated by regional ecological experts. The delimitation of these sites is based on the identification of contiguous interrelated natural habitats which form a cohesive unit in terms of ecological structure and function. The designation of these sites was frequently made to surround CI sites. For these sites, we calculated their biodiversity value as the combination of the mean value of LD for the grid cells which are included in the site and the proportion of the site which is covered by CI sites. These two variables were normalized and a mean value for each EI site calculated. Sites were ranked into one of three classes with high (the 20% of sites with the highest values), intermediate (the 40% of the sites with intermediate values) and low (the 40% of sites with the lowest values) values. This ranking is done for the purpose of our study and limits can of course be changed. By limiting the upper values to the highest 20% we identify the highest priority sites. 2.3. Threat indicators We assessed exposure to local threats and a more global threat due to human population density. Three types of local threats were quantified: roads, intensive agriculture and built-up areas aggregated in a 100 m 100 m grid cell (for a total of 2 780 078 cells). For roads, we used the 2008 “road” layer of BD TOPOÒ/RGE GIS database (Institut Géographique National) which contains five types of roads. We attributed a coefficient to each type of road which reflects potential impacts (in terms of numbers of vehicles): highways ¼ 1, primary (mostly national) roads ¼ 0.6, secondary roads ¼ 0.35, small country roads ¼ 0.15 and small roads within towns and villages ¼ 0.05. The value of a grid cell is obtained using the highest coefficient value among all road types which intersect it. For agriculture, we used the 2006 Corine Landcover database which contains four types of agricultural land use. We attributed a coefficient to each kind of agricultural land use based on the likelihood of turnover each year and levels of fertilizer and pesticide application: arable land ¼ 1, mixed agriculture ¼ 0.65, permanent cultures ¼ 0.4 and grasslands ¼ 0.15. The value of a grid cell is obtained using the highest coefficient value of the different agricultural types which intersect it. For urbanization, we used the 2008 “built-up” layer of BD TOPOÒ/RGE GIS database (Insititut Géographique National). Here, the threat value of a given grid cell is the proportion of undifferentiated buildings. For each threat type independently and each grid cell, the threat values of the cells present in a 2 km zone around the cell were weighted by their distance to a given cell. Weighting values were estimated using a Gaussian kernel smoother with a bandwidth of 0.8 km. The precise threat value for a cell is thus equal to the sum of the weighted values. On a larger scale (50 km), we assessed threats in relation to overall human population density in and around a given site using

11

census data of the human population (Institut National de la Statistique et des Etudes Economiques) in each of the 1546 municipal districts in the region. For each municipal district, all the population values of the districts present in a 50 km buffer from the centroid were weighted by their distance to the municipality in question (weights were estimated using a Gaussian kernel smoother with a bandwidth of 25 km). The human population threat value for a given municipal district is equal to the sum of these weighted values. For each cell, the threat value for human population density is thus the value for the district in which it occurs. All the threat values were normalized from 0 to 1 prior to their combination. An important methodological point in this study is the use of a smoothed function. Although several studies have already investigated the indirect impact of human presence on biodiversity surrogates using buffer zones (e.g. Harcourt et al., 2001; Luck, 2007; Vazquez and Gaston, 2006), our method allows us to integrate a threat which is likely to decrease in relation to the distance, of a road or construction for example, from a given site (see also Jarvis, 2001). The distances of 2 km and 50 km are arbitrary and we assume that the relative threat does not vary among sites when the distance is increased. For each grid cell or site, we then calculated a combined threat value based on the mean values of the four different threat values using weighting coefficients (human population density ¼ 2, agriculture ¼ 3, built-up areas and road presence ¼ 6). Here we thus considered that local threats, in particular urbanization and road, have more important implication for the identification of conservation priorities. For LD, the threat value was automatically calculated for each of the 100 m 100 m grid cells. For each of the CI and EI sites, we calculated the threat value based on the mean value of all grid cells present in the site. For each biodiversity descriptor, we then ranked the threat values of each site in three classes: high (the 20% of sites with the highest values), medium (the 40% of sites with intermediate values) and low (the 40% of sites with the lowest values). Again limits were made simply for the purpose of the study to identify high priority sites and spatial variability. 2.4. Spatial priorities For each biodiversity descriptor we established a priority ranking of all sites or cells according to a classification based on three classes of biodiversity from low (C) to high (A) with three levels of threat values (Table 1). All the analyses in this study were performed in R software. The data manipulation and the maps were realized using Postgre SQL, Postgis and ArcGis. 3. Results 3.1. Threat values The different threats vary in their spatial distribution (Fig. 2a) and their normalized values show differences in frequency

Table 1 Comparison of biodiversity classes and threat classes to produce three levels of conservation priority (A being the highest) for a given biodiversity descriptor with an associated threat class. Biodiversity classes

Low Intermediate High

Threat classes Low

Medium

High

C1 B1 A1

C2 B2 A2

C3 B3 A3

12


Fig. 2. (a) The spatial distribution of threat classes in the study region in five percentile classes ranging from yellow (lowest 20% of values) to dark brown (highest 20% of values); (b) the frequency distribution of normalized threat values according to kernel density estimates. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

distributions (Fig. 2b). In general, threat values (at the grid cell level) are maximized and spatially concentrated in the lowland plain of the region where human populations are increasing dramatically at the current time. Some parts of the region (the south western part of the study area and the southern border of the Massif Central) are less threatened by the human activities we used to assess threat levels (Fig. 2a). The frequency distribution of threat values associated with urbanization is highly skewed, the large majority of sites have low values and only a few grid cells have high values. The frequency distributions of the other threat values, particularly those associated with human population density, are more wide ranging (Fig. 2b). Overall, as witnessed by the quantile values, for each of the three types of local threat (agriculture, urbanization, and roads) and for the combined indicator, threat values tend to be weaker for the CI and EI sites than for LD grid cells (Table 2). The fact that CI and EI sites do not have the same values as LD grid cells reflects their spatial occurrence in certain parts of the region where threat values are lower than average for the region. CI sites tend to have higher q90 values than EI sites for agriculture and urbanization but not for

roads (Table 2), for which threat values on CI and EI sites are very similar. The q25 value is systematically lower in the CI sites than in the EI sites, whereas the q75 and q90 values are generally higher in the CI sites. Hence CI sites are exposed to a wider range of threat classes than EI sites. In relation to threats associated with human population density, CI sites and LD cells have higher q50, q75 and q90 values than EI sites. We detected significant variation in the surface area of EI sites in relation to their threat class for the three local threats and the combined threat value (Table 3). For urbanization and roads, the surface area of the EI sites in the most threatened class was significantly lower than that of EI sites in the lowest and intermediate threat classes, while for agriculture and the combined threat values the surface area of EI sites in the most threatened class was only significantly smaller than sites in the least threatened class (Table 3). There was no significant variation in surface area of sites more or less threatened by human population density. For CI sites, the surface area of the sites most exposed to threats associated with roads and urbanization was significantly lower than those in the two lower threat classes. There was no significant variation in surface area of

R. Vimal et al. / Journal of Environmental Management 95 (2012) 9e16 Table 2 Distribution of normalized threat values in four percentiles for the three biodiversity descriptors: Landscape Diversity (LD) cells, Conservation Interest (CI) sites, Ecological Integrity (EI) sites. The 25% of lowest values are inferior to the value in the q25 column; intermediate values are for the lowest 50% or the lowest 75% of values (the q50 and q75 columns respectively). The highest 10% of values are equal to or above the value given in the q90 column. Threat

Percentile q25

q50

q75

q90

Human population density

LD cells CI sites EI sites

0.0903 0.0887 0.0775

0.2177 0.2293 0.1661

0.3978 0.4024 0.2706

0.6184 0.6437 0.4909

Agriculture


0.0153 0.0004 0.0015

0.1189 0.0393 0.0387

0.3187 0.1827 0.1449

0.4520 0.4031 0.3069

Urbanization


0.0011 0.0002 0.0005

0.0043 0.0016 0.0019

0.0137 0.0062 0.0057

0.0472 0.0193 0.0139

Road


0.0686 0.0379 0.0520

0.1110 0.0774 0.0850

0.1696 0.1279 0.1301

0.2491 0.1890 0.1844

Combined


0.0542 0.0385 0.0411

0.0993 0.0736 0.0681

0.1719 0.1287 0.1091

0.2325 0.1977 0.1589

sites in the different classes of threat associated with intensive agriculture, human population density and combined threats. 3.2. Spatial priorities For LD cells, only 0.4% of grid cells combine a high biodiversity value and a high threat (priority class A3) and most of these are localized on the coast (Fig. 3c). 5.9% of EI sites and 3.8% of CI sites combine high threat and high biodiversity values and thus occur in priority class A3. These sites also occur primarily near the coast or in the lowland plain (Fig. 3c). 14.4% of EI sites and 16.3% of CI sites combine low or intermediate threat and high biodiversity values (i.e. priority classes A1 and A2 respectively). These sites are more widely distributed across the region but tend to be away from the coastal areas. 22.8% of EI sites and 21.3% of CI sites combine intermediate or high threat and intermediate biodiversity values (i.e. priority classes B2 and B3 respectively). These sites are more mostly localized in the lowland plains and near to the coast (Fig. 3c). Table 3 Mean surface area (m2) of conservation interest (CI) sites and ecological integrity (EI) sites in each of the three threat classes in relation to exposure to different threats. F values are based on ANOVA of mean surface area of sites in the three threat classes for each threat and their combination. Values with a different code letter within each biodiversity descriptor and threat are significantly different from one another after a Tukey means test. CI sites F

EI sites Low

Medium

High

Human population density

1.04 ns

532 a

617 a

647 a

2.77 ns

13,342 a

15,564 a

7433 a

Agriculture

0.46 ns

606 a

603 a

526 a

3.36 *

15,401 a

13,897 ab

6664 b

Urbanization

5.49 **

667 a

617 a

375 b

8.96 ***

16,535 a

14,628 a

3011 b

Roads

13.37 ***

742 a

591 a

280 b

7.29 **

15,348 a

15,439 a

3749 b

2.47 ns

675 a

553 a

491 a

5.41 **

16,642 a

13,257 ab

5487 b

Combined

Low

Medium

High

F

*p < 0.05, **p < 0.01, ***p < 0.001.

13

High biodiversity grid cells for LD are principally located in one of three geographic sectors: the part of the study region which extends north to the Massif Central, in the south-west toward the Pyrenees and in the coastal wetlands (Fig. 3b). Seventy six percent of the LD cells in the high biodiversity class occur in EI sites; hence, expert delimitation of the EI sites is fairly coherent with the spatial distribution of the highest values of landscape diversity (Fig. 3b). Indeed, in these sites, the biodiversity values of LD cells are significantly higher than which do not occur in EI sites (t ¼ 835.3, p < 0.001). 43% of the LD cells in priority class A3 are present in EI sites. 82.7% of the surface area of the CI sites are included in EI sites. 83.1% of the surface area of the CI sites in the high biodiversity class are included in EI sites. However, the biodiversity values of CI sites that occur within EI sites are not significantly higher than CI sites that do not occur within the perimeter of an EI site (t ¼ 0.06, p > 0.9). 28% of the high biodiversity LD cells occur in CI sites and 18.1% of the LD cells in priority class A3 are present in the CI sites. The biodiversity values of the LD cells present in CI sites are significantly higher than the value of LD cells which do not occur in CI sites (t ¼ 395.7, p < 0.001). Moreover, the biodiversity values of LD cells present in high biodiversity CI sites are significantly higher than those that occur in the other CI sites with intermediate and low biodiversity values (t ¼ 131. 9, p < 0.001). 4. Discussion 4.1. Assessing vulnerability for diverse biodiversity descriptors Information on different threat processes which determine the risk of habitat destruction and biodiversity decline is crucial for effective conservation planning (Wilson et al., 2005). The purpose of this study has been to illustrate a method to develop and combine multiple criteria with which to assess spatial variability in the vulnerability of different biodiversity descriptors and their spatial variation. The three biodiversity descriptors we examined displayed marked differences amongst each other in terms of the distribution and levels of exposure to human-based threats. Their joint analysis allows us to make a fairly comprehensive assessment of regional conservation priorities and their spatial delimitation. Landscape diversity (LD) was quantified on a grid scale for the whole study region, whilst the two other biodiversity descriptors were based on regional inventories of sites recognized as being of ecological interest. Our study shows that the LD descriptor highlights areas of high biodiversity value based on the two other descriptors and thus the use of regional inventories in conservation planning. The sites of high conservation interest reflect the need to protect the habitat of important species based on their regional rarity and localized distribution. An important point concerning the CI sites is that the establishment of priorities for the conservation of these sites was not based on species richness but on the complementarity of the sites to meet predefined target representations for each species (e.g. Pressey, 1994). As a result, a site can have high irreplaceability value but low species richness. In contrast to EI sites, CI sites experience a wider range of threat values. Overall, while the EI sites are generally located outside of the areas where human-based threats are most important, the CI sites are more widely distributed (Fig. 3b) and thus experience a greater range of threats, hence their slightly higher values for the q75 and q90 percentiles (Table 2). The designation of EI sites is based on a conservation objective geared toward maintaining ecological cohesion and integrity and it is thus reassuring that their current vulnerability is lower than CI sites and that their biodiversity values reflect high biodiversity values in the LD cells. The EI sites, the most vulnerable in terms of current exposure to the three direct threats we studied, have

14


Fig. 3. Spatial conservation priorities for landscape diversity (LD), ecological integrity (EI) and species of conservation interest (CI) sites. (a) The different categories of combined threat values (pale grey e low, dark grey e medium, black e high). (b) The different categories of biodiversity values: low (pale green), intermediate (medium green) and high (dark green). (c) The different combinations of biodiversity and combined threat described in Table 1 to identify low conservation priorities from classes C1,2,3 and B1 (pale orange), intermediate conservation priorities from classes B2,3 (orange) and from classes A1,2 (red) and high conservation priorities from class A3 (brown). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

a smaller surface area than sites less exposed to these threats. This result can be explained in two ways. First the threatened elements due to urbanization, agriculture and roads are rarely present inside the sites but tend to be located around their perimeter. Thus, the EI sites with the highest threat value are logically the smallest. Second, the sites with small surface area can be delimitated in more intensively human-dominated and fragmented parts of the study region (Fig. 3). Hence, for threats associated with human

population density, the surface area of highly threatened EI sites is also the lowest (Table 3). 4.2. Methodological issues for comprehensive assessment of vulnerability The assessment of conservation priorities we have performed at regional scale is subject to a several limitations that require


consideration in the practical application of such work. First, the thresholds we used to create classes of vulnerability and biodiversity are arbitrary and thus different classifications for a given site could be obtained if limit values are changed for the different classes. It may be preferable to identify natural breaks in the data. If however arbitrary threshold are fixed, conservation planners should take account of the range of values in each class. Second, we used several coefficients in order to account for the different impacts of the threat indicators although in an ideal situation differential coefficients should be defined for each species in each biogeographic region. For instance, a meadow in the lowland plain does not contain the same biodiversity as one in the Pyrenees and bird species vary dramatically in their sensitivity to agricultural practices. Third, our assessment of vulnerability is a photographic vision of the current threat status. As we mention below, a comprehensive assessment of vulnerability will require a prospective study in order to integrate the probability that a given site will be impacted in the future (on different time scales) by a given type of threat. Such analyses are of course highly complex and merit a full study which is beyond the scope of this paper. Even if it is probable that future technical developments will reduce these biases and limitations, our discussion of them is simply to highlight the difficulty and complexity of constructing a more comprehensive and objective assessment of conservation priorities. 4.3. Defining conservation priorities No more than 6% of EI sites and 4% of CI sites and only 0.4% of LD sites have combined biodiversity and threat values that make them part of the high priority class A3. The majority of these cells and sites occur in coastal areas (Fig. 3). This is typical of both the Mediterranean region where coastal areas have been and will continue to be the most impacted by development projects (Benoit and Comeau, 2005) and also other regions with a coastal zone which incur the highest threats of natural habitat transformation (Rouget et al., 2003b). The priority sites A1 and A2 occur in more inland places and rarely near the coastal fringe of the region. The priority sites B2 and B3 occur more in the lowland plain. Variation in the spatial distribution of threats in terms of exposure to different risks of transformation is thus highly apparent and should not be neglected in assessing conservation priorities associated. Sites in the high priority class A3 probably require more strict conservation status e reglementary control e to reduce current threats and to assure biodiversity conservation. Sites of high ecological value but low threat value (A1 and A2); a situation which is common to our study area; do not necessarily need to have a reglementary protection status. This highlights the different meaning of the term “conservation priorities”. If sites in class A3 need urgent active conservation plan to reduce current threats, sites in class A2 and particularly in class A1 require special attention for future transformations and thus must be considered in land-use planning process. From this point of view, sites in class B2 and particularly in class B3 can require more attention in term of current risk of transformation. However, the latter sites, whose vulnerability is particularly high, may in some regions be doomed to development and thus of low priority if resources for conservation investment are scarce (Rodrigues et al., 2004). Flexible conservation management of such sites should be envisaged with monitoring of any increase in threat status. As Noss et al. (2002) remark, conservation priorities change in time and thus correct assessment of vulnerability requires frequent modification or otherwise integration of scenarios of future change as new sites become protected and others are subject to development (Pressey and Taffs, 2001). High biodiversity sites with a current low

15

threat should thus not be neglected in a regional conservation plan and should be given priority if future land-development projects emerge. As work such as ours moves closer to the practical realities of conservation planning, attention should thus be made to refine the different priority classes.

4.4. Conservation management at the site level Once the choice of sites based on a combination of biodiversity and threats is made, the next step is to envisage the implementation of conservation management for each site. This requires precise knowledge of the biodiversity present and its vulnerability to different potential threats. The three biodiversity descriptors used in our study will clearly require different conservation actions in relation to the levels of threat incurred. For example, new road projects should be avoided in the EI sites to limit fragmentation where ecological integrity is recognized as a priority. Here particular types of management and very often the maintenance of traditional human activities should be promoted. In contrast, in the CI sites it may be more pertinent to attribute specific funds to species protection. In addition to this need for a diversity of conservation actions, the use of several biodiversity descriptors also illustrates the need to consider the interaction between different spatial patterns of biodiversity and to integrate them into a comprehensive conservation plan. Two situations are particularly interesting in this respect. First, most of the CI sites and LD cells occur within an EI site. Indeed, 83% of the surface area of the CI sites and 76% of the high biodiversity LD cells occur within the perimeter of EI sites. The latter represent a functional entity containing several components and thus provide the opportunity to implement conservation actions for CI sites and LD cells of high interest and perhaps define precise areas of EI sites for particular conservation actions. Second, for the CI sites and LD cells of high biodiversity value but which are independent of EI sites, special attention should be given to the threat elements that occur in and around them and the biotic interaction they require with their surrounding landscape to persist (DeFries et al., 2007; Thompson et al., 2011).

Acknowledgments This work was funded by the Agence Nationale de la Recherche contract 05-BDIV-014 (ABIME), and the Languedoc-Roussillon regional government. We are particularly grateful to staff of the Conservatoire des Espaces Naturels (Languedoc Roussillon) and the Conservatoire Botanique National Méditerranéen de Porquerolles for their advice and data.

References Abbitt, R.J.F., Scott, J.M., Wilcove, D.S., 2000. The geography of vulnerability: incorporating species geography and human development patterns into conservation planning. Biological Conservation 96, 169e175. Ball, I.R., Possingham, H.P., 2000. MARXAN (V1.8.2): Marine Reserve Design Using Spatially Explicit Annealing, a Manual. Balmford, A., Mace, G.M., Ginsberg, J.R., 1998. The challenges to conservation in a changing world: putting processes on the map. Conservation Biology Series (Cambridge), vol. 1, pp. 1e28. Benoit, G., Comeau, A., 2005. Méditerranée. Les perspectives du Plan Bleu sur l’environnement et le développement. de l’Aube, pp. 428. Brooks, T.M., Mittermeier, R.A., da Fonseca, G.A.B., Gerlach, J., Hoffmann, M., Lamoreux, J.F., Mittermeier, C.G., Pilgrim, J.D., Rodrigues, A.S.L., 2006. Global biodiversity conservation priorities. Science 313, 58e61. DeFries, R., Hansen, A., Turner, B.L., Reid, R., Liu, J.G., 2007. Land use change around protected areas: management to balance human needs and ecological function. Ecological Applications 17, 1031e1038. Dreal, L.R., 2009. L’inventaire Znieff en Languedoc Roussillon.

16


Harcourt, A.H., Parks, S.A., Woodroffe, R., 2001. Human density as an influence on species/area relationships: double jeopardy for small African reserves? Biodiversity and Conservation 10, 1011e1026. IFEN, 2003. L’environnement en Languedoc Roussillon. IFEN, Orléans, France, pp. 168. INPN, 2006. L’inventaire Znieff, Paris. Jarvis, J., 2001. National landscape conservation system: a new approach to conservation. Science and Stewardship to Protect and Sustain Wilderness Values, pp. 110e113. Luck, G.W., 2007. A review of the relationships between human population density and biodiversity. Biological Reviews 82, 607e645. Margules, C.R., Pressey, R.L., 2000. Systematic conservation planning. Nature 405, 243e253. Naiman, R.J., Décamps, H., Pollock, M., 1993. The role of riparian corridors in maintaining regional biodiversity. Ecological Applications 3, 209e212. Noss, R.F., Carroll, C., Vance-Borland, K., Wuerthner, G., 2002. A multicriteria assessment of the irreplaceability and vulnerability of sites in the Greater Yellowstone Ecosystem. Conservation Biology 16, 895e908. Ozenda, P., 1985. La végétation de la chaîne alpine. Masson, Paris, France. Poiani, K.A., Richter, B.D., Anderson, M.G., Richter, H.E., 2000. Biodiversity conservation at multiple scales: functional sites, landscapes, and networks. Bioscience 50, 133e146. Possingham, H.P., Ball, I.R., Andelman, S., 2000. Mathematical methods for identifying representative reserve networks. In: Ferson, S., Burgman, M. (Eds.), Quantitative Methods for Conservation Biology. Springer-Verlag, New York, NY, pp. 291e305. Pressey, R.L., 1994. Ad hoc reservations e forward or backward steps in developing representative reserve systems. Conservation Biology 8, 662e668. Pressey, R.L., Cowling, R.M., Rouget, M., 2003. Formulating conservation targets for biodiversity pattern and process in the Cape Floristic Region, South Africa. Biological Conservation 112, 99e127. Pressey, R.L., Taffs, K.H., 2001. Scheduling conservation action in production landscapes: priority areas in western New South Wales defined by irreplaceability and vulnerability to vegetation loss. Biological Conservation 100, 355e376. Rodrigues, A.S.L., Akcakaya, H.R., Andelman, S.J., Bakarr, M.I., Boitani, L., Brooks, T.M., Chanson, J.S., Fishpool, L.D.C., Da Fonseca, G.A.B., Gaston, K.J., Hoffmann, M., Marquet, P.A., Pilgrim, J.D., Pressey, R.L., Schipper, J., Sechrest, W., Stuart, S.N., Underhill, L.G., Waller, R.W., Watts, M.E.J., Yan, X., 2004. Global gap analysis: priority regions for expanding the global protected-area network. Bioscience 54, 1092e1100. Rouget, M., Richardson, D.M., Cowling, R.M., 2003a. The current configuration of protected areas in the Cape Floristic Region, South Africa e reservation bias and representation of biodiversity patterns and processes. Biological Conservation 112, 129e145. Rouget, M., Richardson, D.M., Cowling, R.M., Lloyd, J.W., Lombard, A.T., 2003b. Current patterns of habitat transformation and future threats to biodiversity in terrestrial ecosystems of the Cape Floristic Region, South Africa. Biological Conservation 112, 63e85. Sala, O.E., Chapin, F.S., Armesto, J.J., Berlow, E., Bloomfield, J., Dirzo, R., HuberSanwald, E., Huenneke, L.F., Jackson, R.B., Kinzig, A., Leemans, R., Lodge, D.M., Mooney, H.A., Oesterheld, M., Poff, N.L., Sykes, M.T., Walker, B.H., Walker, M.,

Wall, D.H., 2000. Biodiversity e global biodiversity scenarios for the year 2100. Science 287, 1770e1774. Sanderson, E.W., Malanding, J., Levy, M.A., Redford, K.H., Wannebo, A.V., Woolmer, G., 2002. The human footprint and the last of the wild. Bioscience 52, 891e904. Schipper, J., Chanson, J.S., Chiozza, F., Cox, N.A., Hoffmann, M., Katariya, V., Lamoreux, J., Rodrigues, A.S.L., Stuart, S.N., Temple, H.J., Baillie, J., Boitani, L., Lacher, T.E., Mittermeier, R.A., Smith, A.T., Absolon, D., Aguiar, J.M., Amori, G., Bakkour, N., Baldi, R., Berridge, R.J., Bielby, J., Black, P.A., Blanc, J.J., Brooks, T.M., Burton, J.A., Butynski, T.M., Catullo, G., Chapman, R., Cokeliss, Z., Collen, B., Conroy, J., Cooke, J.G., da Fonseca, G.A.B., Derocher, A.E., Dublin, H.T., Duckworth, J.W., Emmons, L., Emslie, R.H., Festa-Bianchet, M., Foster, M., Foster, S., Garshelis, D.L., Gates, C., Gimenez-Dixon, M., Gonzalez, S., GonzalezMaya, J.F., Good, T.C., Hammerson, G., Hammond, P.S., Happold, D., Happold, M., Hare, J., Harris, R.B., Hawkins, C.E., Haywood, M., Heaney, L.R., Hedges, S., Helgen, K.M., Hilton-Taylor, C., Hussain, S.A., Ishii, N., Jefferson, T.A., Jenkins, R.K.B., Johnston, C.H., Keith, M., Kingdon, J., Knox, D.H., Kovacs, K.M., Langhammer, P., Leus, K., Lewison, R., Lichtenstein, G., Lowry, L.F., Macavoy, Z., Mace, G.M., Mallon, D.P., Masi, M., McKnight, M.W., Medellin, R.A., Medici, P., Mills, G., Moehlman, P.D., Molur, S., Mora, A., Nowell, K., Oates, J.F., Olech, W., Oliver, W.R.L., Oprea, M., Patterson, B.D., Perrin, W.F., Polidoro, B.A., Pollock, C., Powel, A., Protas, Y., Racey, P., Ragle, J., Ramani, P., Rathbun, G., Reeves, R.R., Reilly, S.B., Reynolds, J.E., Rondinini, C., Rosell-Ambal, R.G., Rulli, M., Rylands, A.B., Savini, S., Schank, C.J., Sechrest, W., Self-Sullivan, C., Shoemaker, A., Sillero-Zubiri, C., De Silva, N., Smith, D.E., Srinivasulu, C., Stephenson, P.J., van Strien, N., Talukdar, B.K., Taylor, B.L., Timmins, R., Tirira, D.G., Tognelli, M.F., Tsytsulina, K., Veiga, L.M., Vie, J.C., Williamson, E.A., Wyatt, S.A., Xie, Y., Young, B.E., 2008. The status of the world’s land and marine mammals: diversity, threat, and knowledge. Science 322, 225e230. Smith, T.B., Bruford, M.W., Wayne, R.K., 1993. The preservation of process: the missing element of conservation programs. Biodiversity Letters 1, 164e167. Spring, D., Baum, J., MacNally, R., Mackenzie, M., Sanchez-Azofeifa, A., Thomson, J.R., 2010. Building a regionally connected reserve network in a changing and uncertain World. Conservation Biology 24, 691e700. Stanners, D., Bourdeau, P., 1995. Europe’s Environment: The Dobris Assessment. European Environmental Agency, Copenhagen, Denmark, pp. 676. Thompson, J.D., Mathevet, R., Delanoë, O., Gil-Fourrier, C., Bonnin, M., Cheylan, M., 2011. Ecological solidarity as a conceptual tool for rethinking ecological and social interdependence in conservation policy for protected areas and their surrounding landscape. Comptes Rendus de l’Académie des Science. Biologies 334, 412e419. Vazquez, L.B., Gaston, K.J., 2006. People and mammals in Mexico: conservation conflicts at a national scale. Biodiversity and Conservation 15, 2397e2414. Vimal, R., Rodrigues, A.S., Mathevet, R., Thompson, J.D., 2011. The sensitivity of gap analysis to conservation targets. Biodiversity and conservation 20 (3), 531e543. Wilcove, D.S., Rothstein, D., Dubow, J., Phillips, A., Losos, E., 1998. Quantifying threats to imperiled species in the United States. Bioscience 48, 607e615. Wilson, K., Pressey, R.L., Newton, A., Burgman, M., Possingham, H., Weston, C., 2005. Measuring and incorporating vulnerability into conservation planning. Environmental Management 35, 527e543.