Comparative Assessment of Different Earth System Models for Habitat Suitability of Cuminum cyminum (Linn.) Crop: A Machine Learning Evaluation from Arid and Semi-Arid Hot Areas of the India

Abstract

Species distribution model has been widely used to predict the distribution of Rare Endangered and Threatened species based on bioclimatic variables, including forestry, medicinal, and others. Crop species modeling is limited to prime crops like wheat, rice, maize, soyabean, etc., and few attempts have been made for spies’ crops grown in arid and semi-arid regions. This study examines the habitat suitability of Cuminum cyminum (cumin), a crop grown in Rajasthan and Gujarat, India. The widely used WorldClim dataset and CMCC-BioClimInd bioclimatic dataset were tested for predictive and elucidative power. This evaluation calculated AUC and omission rate using the MaxEnt entropy method. The WorldClim dataset includes three-time frames and four greenhouse gas scenarios. The CMCC-BioClimInd dataset included five Earth System Models (ESMs): GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR, MIROC-ESM-CHEM, and NorESM1-M, which included two emission scenarios (4.5 and 8.5). The results indicate that both data sets have similar predictive accuracy. However, the optimal predictive areas for this crop differed significantly between the two model types. Annual precipitation, precipitation seasonality, precipitation during the coldest quarter, and potential evapotranspiration Hargreaves are the main factors affecting this crop's growth and expansion into new regions. Our research offers a novel standpoint for the implementation of this crop in potential new areas within Rajasthan and Gujarat.