Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Pays Bas → Modelling the spatial distribution of snake species under changing climate scenario in Spain

International Institute for Geo-Information Science and Earth Observation (ITC) 2010

Modelling the spatial distribution of snake species under changing climate scenario in Spain

Muthoni, F.K. (Francis Kamau)

Titre : Modelling the spatial distribution of snake species under changing climate scenario in Spain.

Auteur : Muthoni, F.K. (Francis Kamau)

Etablissement de soutenance : University of Twente International Institute for Geo-Information Science and Earth Observation (ITC)

Grade : Master of Science in Geo-Information Science and Earth Observation 2010

Résumé
Climate change has been postulated to be one of the main drivers of biodiversity loss globally as it results to alteration of the species habitats. Hence, this study investigated the effect of climate change on the spatial distribution of four snake species Coronella girondica, Natrix maura, Malpolon monspes sulanus and Rhinechis scalaris in Spain. The specific objectives were to ; (1) identify the predictor variables that had the highest predictive power in the potential distribution models of the four snake species ; (2) generate potential distribution models for the four snake species using the current climatic and biophysical explanatory variables ; 3) project the future potential distribution of the four snakes using the projected year 2050 climatic and biophysical variables ; and 4) investigate whether the projected range shifts among the specialist species was different than that of generalist species. Maxent algorithm was trained using species presence data and a set of climatic and biophysical environmental variables. To identify which set of predictors had the highest contribution to the model, two types of models were run ; one with biophysical variables only and another with both climatic and biophysical variables. To assess the average behaviour of the algorithm, ten random partitions were run each comprising of 70% of presence data for training and 30% for testing. Jackknife test of variable importance was used to identify one variable that resulted to the least drop in the training gains when omitted from the models. This variable was eliminated and the process continued until only one variable remained. Mann- Whitney U statistic was used to test the statistical difference between the different sized models. The model with the least number of predictors and the training gains not significantly different from that of the full model was selected as the best model. Similarly, Mann-Whitney U statistic was used to test the statistical difference between the two sets of models. The best current conditions model from either of the two model suites was used to project the future ranges based on scenario A2 of the HadCM3 model. The models were evaluated using the a rea under the ROC curve (AUC), binomial tests and sensitivity. To reveal the future range shifts, the current and future maps were cross-tabulated to derive kappa index of agreement and crammers V statistic. Results showed that climatic variables had the highest predictive power suggesting that the distribution of these species at meso-scale is largely set by climate. Moreover, all species were projected to shift their ranges by the year 2050 due to changing climates. Furthermore the generalist’s species range was projected to expand while that of specialist’s species tended to contract. Nevertheless, factors such as biotic interactions, dispersal abilities and evolutionary adaptations need to be incorporated into the models before a concrete conclusion that climate is the main driver of the species range shifts. Moreover, re-testing of these hypotheses with higher resolution dataset that captures fine ecological details of these species was recommended.

Version intégrale (ITC)

Page publiée le 30 janvier 2018