Priority areas for a cold-adapted snake 335
and is categorized as Endangered on the IUCN Red List (Mizsei et al., 2018). The species has only 17 known popula- tions, inhabiting alpine–subalpine meadows above 1,600 m on isolated mountaintops (Mizsei et al., 2016). The occupied habitats are the highest and coldest in the region; as the spe- cies is adapted to cold environments it is sensitive to climate change. These alpine habitats are threatened by overgraz- ing and habitat degradation, which simplify vegetation struc- ture and reduce cover against potential predators. The spe- cies is strictly insectivorous, specializing on bush crickets and grasshoppers (Mizsei et al., 2019), and thus it is vulner- able to changes in its primary prey resource caused by land use. Shepherds are known to intentionally kill these snakes, which are held responsible for 1–4% of lethal bites to sheep annually in Albania (E. Mizsei, unpubl. data). Here we identify priority areas that could facilitate the
long-term persistence of V. graeca. We performed single species spatial prioritization to identify areas based on conservation value attributes of the currently suitable landscape. We aimed to identify any populations likely to disappear by the 2080s and any populations likely to be strongholds where the species could benefit from targeted management and conservation. To provide guidelines for conservation, we studied the overlap between priority areas and protected areas and explored the interrelationships of the conservation value variables to identify opportuni- ties for potential interventions.
Methods
Approach We used spatial prioritization, an approach rarely applied to single species (Adam-Hosking et al., 2015), which is based on quantifying known threats and transforming them into variables describing conservation value based on current conditions. The process consisted of (1) delimiting the study area, (2) collecting and processing data on threats to derive conservation value variables, (3)spa- tial prioritization of the study area based on the conservation value variables, and (4) evaluation of overlaps with current protected areas and analysis of any interrelationships between conservation value variables (Fig. 1).
Habitat suitability modelling to define the study area We defined our study area to include all V. graeca habitats identified by a habitat suitability model (Fig. 2). The model was constructed using 351 records of V. graeca, 83%of which were taken from our previous publications (Mizsei et al., 2016, 2017) and 17% of which were new, unpublished records. Our dataset contained all known localities of V. graeca, and also included absence data from 31 localities (55 records) predicted in an earlier habitat suitability model (Mizsei et al., 2016), and where the species has not
been found despite extensive searches. To account for po- tential spatial biases in sampling effort among sites, we re- sampled the presence dataset to obtain a spatially balanced subset of 71 presence records, which represented all known populations (Fig. 2). We used BIOCLIM climate variables, which have been successfully used to model the distribution of several European reptiles, including vipers (Scali et al., 2011; Martínez-Freiria, 2015; Mizsei et al., 2016). To ensure com- patibility with our previous work (Mizsei et al., 2016), we used the same set of predictors: annual mean temperature (BIO1), temperature seasonality (BIO4), annual precipita- tion (BIO12) and precipitation seasonality (BIO15). These variables had high predictive performance and low in- tercorrelations (r,0.7). We obtained these variables for current climatic conditions (mean of 1950–2000) from the WorldClim 1.4 database at 30 arc seconds resolution (Hijmans et al., 2005). We generated the habitat suitability model using ensemble modelling (Thuiller et al., 2012) in the BIOMOD2 package in R 3.3.0 (R Core Team, 2016). Weused two linear model algorithms (Generalized Linear Models, GLM; Generalized Additive Models, GAM) and three machine learning algorithms (Artificial Neural Networks, ANN; Random Forest, RF; Maximum Entropy, MaxEnt). Default settings were used to build the models (Thuiller et al., 2012). To increase model accuracy we generated 10 da- tasets of pseudo-absences, which included real absence data and random points at least 5 km from the presence local- ities (a total of 10,000 points per dataset). We ran 10 repli- cates with each of the five modelling algorithms for the 10 pseudo-absence datasets, which resulted in 500 habitat suit- ability model replicates. In eachmodel replicatewe random- ly divided the presence data into training (70%) and testing subsets (30%). We used the four BIOCLIM variables as pre- dictors. We scored all individual model replicates by the true skill statistic (TSS; Allouche et al., 2006), retaining only models with TSS.0.95. This filtering resulted in 275 mod- els, which always correctly classified the testing subset and that were used to produce ensemble projections by consen- sus (i.e. raster cells predicted suitable in each of the 275 mod- els). Finally, we generated a binary suitability map using the ensemble projection of the best models and defined the re- sulting patches as the study area for the analysis (Fig. 2).
Variables for estimating conservation value We used nine environmental variables of three classes to estimate the con- servation value of planning units (1 ha raster cells) over the entire study area: (1) habitat suitability: climate suitability, habitat size, habitat occupancy, vegetation suitability; (2) cli- mate change: future persistence, potential for altitudinal range shift, (3) land-use impact: habitat alteration, habitat degradation, disturbance (Fig. 1). Each variable was standard- ized to 0–1, where 0 represented minimum and 1 repre- sented maximum conservation value.
Oryx, 2021, 55(3), 334–343 © The Author(s), 2020. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605319000322
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102 |
Page 103 |
Page 104 |
Page 105 |
Page 106 |
Page 107 |
Page 108 |
Page 109 |
Page 110 |
Page 111 |
Page 112 |
Page 113 |
Page 114 |
Page 115 |
Page 116 |
Page 117 |
Page 118 |
Page 119 |
Page 120 |
Page 121 |
Page 122 |
Page 123 |
Page 124 |
Page 125 |
Page 126 |
Page 127 |
Page 128 |
Page 129 |
Page 130 |
Page 131 |
Page 132 |
Page 133 |
Page 134 |
Page 135 |
Page 136 |
Page 137 |
Page 138 |
Page 139 |
Page 140 |
Page 141 |
Page 142 |
Page 143 |
Page 144 |
Page 145 |
Page 146 |
Page 147 |
Page 148 |
Page 149 |
Page 150 |
Page 151 |
Page 152 |
Page 153 |
Page 154 |
Page 155 |
Page 156 |
Page 157 |
Page 158 |
Page 159 |
Page 160 |
Page 161 |
Page 162 |
Page 163 |
Page 164
