DIGITALNA ARHIVA ŠUMARSKOG LISTA
prilagođeno pretraživanje po punom tekstu




ŠUMARSKI LIST 7-8/2022 str. 74     <-- 74 -->        PDF

Leopards can easily adapt to their habitat and live anywhere with sufficient vegetation and sufficient prey animals (Edgaonkar and Chellam 2002; Bailey 2005; Sarı et al. 2020). Leopards’ habitat selection is mainly based on prey abundance. The leopard is a predator with one of the widest ranges of food sources in the world, and it can adapt to various climatic zones and ecological environments as long the quantity and quality of prey are high enough (Bailey 1993; Nowell and Jackson 1996; Edgaonkar and Chellam 2002; Bailey 2005; Sarı et al. 2020). The other ecological parameters such as vegetation, elevation, slope etc. are indirectly important for the species’ ecological needs (Sarı et al. 2020). When the literature records of the Anatolian leopard in Türkiye are examined, it will be seen that it is recorded in a wide variety of habitats, from semi-desert areas to evergreen forests, from sea level to top of mountains, from forest areas with high closure rate to open areas. When evaluating leopard habitat preference in this study, only climatic data have been examined before and, although climate is a crucial variable affecting the dispersal and layout of plant and animal communities, should also be taken into account (Walck et al. 2011; Zhang et al. 2019a). Another reason for using only climate data is that the exact location of the leopard literature records as points is unknown due to the diversity of observations source/type (shooting, observations, signs of presence). Ninety-seven variables were selected related to leopard distribution, representing bioclimatic factors (Bio 1–Bio 19), these were downloaded from the World Climate Database (www.worldclim.org) (Hijmans et al. 2005) (Table 1).
Data Analysis – Analiza podataka
The main principle of the MaxEnt software is that there are a random variable and uncertainty related to it (Elith et al. 2010). In the maximum similarity method of MaxEnt, the probability of a species being present at each pixel in the studied area is generalized to the entire studied area (Yost et al. 2008). In this study, ten replications were used for each model, and 10% of the dataset included in the analysis was evaluated as test data. To obtain the best results from the modelling methods, the data used in the analyses were obtained at the highest resolution obtained from WorldClim 30” (~900m²). All datasets obtained were in the decimal coordinate system and WGS84 map datum. These datasets were generated by interpolating average monthly climatic data from climate stations around the world (Hijmans et al. 2005). These datasets include data on monthly total precipitation and data on average, minimum, and maximum temperature, and 19 climatic datasets derived from these data. These 19 ecological factors were evaluated to identify potential distribution areas that can provide the same conditions in different regions. Geographical coordinates of the areas where the leopard was found were entered into the program by taking into account the current climate data consisting of 19 climate dataset; then these coordinates were analysed.
MaxEnt (version 3.3.3e), a software that models species niches and distributions, was used to complete ENM (Philips et al. 2006). MaxEnt predicts species’ distributions from bioclimatic data (Elith et al. 2006; Phillips et al. 2006). The MaxEnt algorithm uses bioclimatic and locality point data to find the maximum entropy distribution, and from there predicts species’ niches (0 = lowest probability and 1 = highest) (Philips et al. 2006, Kozak et al. 2008; Phillips and Dudik 2008). Ninety-one presence points were used for the model, and ArcGIS 10.2 software (Esri 2012) was used to estimate distribution maps and climatic variables (Figure 1). The area under the receiver operating characteristic curve (AUC) helped to determine the importance of the model, evaluate its results, and perform a sensitivity analysis on it; a jackknife test was performed to assess the relative importance of variables in the species distribution prediction (Phillips et al. 2006; Zhang et al. 2019b). Threshold independent receiver-operating characteristic analysis (ROC) was used to calibrate the model and evaluate its accuracy. The AUC values of the ROC curve were analyzed to determine the model’s success. AUC values close to 1 were considered excellent, descriptive if close to 0.7 and non-informative if close to 0.5 (Philips et al. 2006; Elith et al. 2006).
3. RESULTS
3. REZULTATI
The model determined the habitat suitability for leopards in Türkiye, and the relationship between suitability and climatic variables was analyzed. When the results were evaluated, the habitat suitability model was found to be highly reliable (Figure 2).
The ROC value of the obtained habitat suitability model was 0.914 (Figure 3). The data analysis found the training data to be AUC values close to 1 as a result of the data