Distribution modeling for predicting habitat suitability for citron (Citrus medica L.) under climate change scenarios

Abstract

Citron (Citrus medica) is one of the major citrus fruits of commerce. It is cultivated mostly in the warm temperate Mediterranean coastal region, and in the tropical and subtropical regions of South and Southeast Asia. It also grows naturally in northeast India, Myanmar, and Yunnan (China). Citron populations are declining in their natural habitats due to forest degradation and change in land use patterns. To predict the habitat suitability for citron, we evaluated four widely used species distribution models (SDMs), namely Generalized Linear Models (GLM), Support Vector Machine (SVM), Random Forest (RF), and Maximum Entropy (MaxEnt) in ‘sdm’ package. Based on Area under ROC Curve (AUC; >0.97) and True Skill Statistics (TSS; >0.9) of all four models, an ensemble modeling was carried out using 111 geo-referenced datasets and 19 bioclimatic and three topographic variables under three climate change scenarios (current, 2050 and 2070). Precipitation of warmest quarter (bio18, 69.6%), elevation (11.1%), mean temperature of wettest quarter (bio8, 10.6%), and temperature seasonality (bio4, 10.1%) were the most important environmental variables for determining the distribution of citron in the study area. Potentially suitable habitat for citron was predicted to occur from 12 – 29 °N to 79 – 108 °E. The total suitable habitat covers about 875,868 km2, including 195,846 km2, 274,386 km2, and 405,636 km2, representing high, moderate, and least suitable areas, respectively. Increase in habitat suitability was predicted for 2050 and 2070 under four representative concentration pathways (RCPs), i. e., RCP2.6, RCP4.5, RCP6.0, and RCP8.5. This study predicted that northeast India and northern Myanmar would remain moderately or highly suitable for citron.