engleski

Preliminary habitat suitability model for Fraxinus angustifolia Vahl in Croatia using Maximum Entropy

Predictive species distribution modelling is a useful tool for developing effective management and conservation programmes for species and ecosystems. The main objective of this study is to test the usefulness of habitat suitability model in predicting the potential distribution of Fraxinus angustifolia, a wide ranged economically important tree species in Croatia, to explore which ecological factors influence the species occurrence and to assess a possible regional niche variation. We applied a maximum entropy modelling method using software MAXENT 3.3.1 which uses species presence only data. A total set of 802 occurrence points was used, obtained from Flora Croatica Database, Forest Inventory and our own sampling. Twelve independent environmental layers were included in the model as predictor variables: WorldClim bioclimatic variables, topographic variables and a habitat type variable. We performed 30 replicate runs with default modelling parameters, each time randomly selecting 70% of the occurrence points as training and remaining 30% as test data. The Maxent model performed well with high average training and test AUC values across 30 runs. Predictions were significantly different from random for all runs at all thresholds measured by the binomial omission tests. Single most important predictor variable was habitat type followed by the elevation. Predicted distribution was well congruent with present known distribution of Fraxinus angustifolia in Croatia. The overlap between the distribution in Continental and Mediterranean biogeographical region was low, suggesting regional niche variation. Our results indicate that close approximation to the reality distribution pattern can be obtained using this approach, even for species with specialized ecological requirements.

species distribution modelling; Maxent; Fraxinus angustifolia

nije evidentirano