Using climatic variables alone overestimate climate change impacts on predicting distribution of an endemic species.

Abstract

Plant species distribution is constrained by both dynamic and static environmental variables. However, relative contribution of dynamic and static variables in determining species distributions is not clear and has far reaching implications for range change dynamics in a changing world. Prunus eburnea (Spach) Aitch. & Hemsl. is an endemic and medicinal plant species of Iran. It has rendered itself as ecologically important for its functions and services and is currently in need of habitat conservation measures requiring investigation of future potential distribution range. We conducted sampling of 500 points that cover most of Iran plateau and recorded the P. eburnea presence and absence during the period 2015-2017. In this study, we evaluated impacts of using only climatic variables versus combined with topographic and edaphic variables on accuracy criteria and predictive ability of current and future habitat suitability of this species under climate change (CCSM4, RCP 2.6 in 2070) by generalized linear model and generalized boosted model. Models' performances were evaluated using area under the curve, sensitivity, specificity and the true skill statistic. Then, we evaluated here, driving environmental variables determining the distribution of P. eburnea by using principal component analysis and partitioning methods. Our results indicated that prediction with high accuracy of the spatial distribution of P. eburnea requires both climate information, as dynamic primary factors, but also detailed information on soil and topography variables, as static factors. The results emphasized that environmental variable grouping influenced the modelling prediction ability strongly and the use of only climate variables would exaggerate the predicted distribution range under climate change. Results supported using both dynamic and static variables improved accuracy of the modeling and provided more realistic prediction of species distribution under influence of climate change.