Optimized MaxEnt Models Predict Species-Specific Habitat Suitability Shifts for Three Dryland Fabaceae Species in Xinjiang

Abstract

Climate warming is altering plant range limits in drylands, where water limitation, soil salinity, and thermal extremes jointly constrain species niches. Yet it remains unclear whether dryland legumes with contrasting life forms respond differently to future climates, a problem that remains poorly resolved. We used ENMeval-optimized MaxEnt models to estimate the current and future potential distributions of three ecologically important legumes in arid Xinjiang: Alhagi camelorum, a perennial semi-shrub; Glycyrrhiza inflata, a perennial herb; and Trigonella arcuata, an annual herb. Models were evaluated under four Shared Socioeconomic Pathways for 2061–2080 and 2081–2100 and showed high discriminatory performance with test AUC values of 0.910–0.947. Under the current climate, G. inflata formed a concentrated core in the Tarim Basin oases, T. arcuata showed a fragmented northern-biased distribution, and A. camelorum was widely dispersed across both northern and southern Xinjiang. The strongest predictors differed among species: minimum temperature of the coldest month for G. inflata (45.9% contribution), precipitation seasonality for T. arcuata 76.8%, and NDVI for A. camelorum 34.0%. Bootstrap-based niche analyses further quantified these differences: A. camelorum and G. inflata had similar niche breadths, Levins’ B2 = 0.329 and 0.315, respectively, whereas T. arcuata had the narrowest niche breadth, B2 = 0.225; 95% CI: 0.162–0.241. Pairwise niche overlap was generally low, particularly between G. inflata and T. arcuata (Schoener’s D = 0.019; Hellinger’s I = 0.038). Projections based on BCC-CSM2-MR indicated species-specific and scenario-dependent suitability shifts across both future periods. G. inflata showed the largest projected net gains across periods and scenarios; A. camelorum showed moderate but consistently positive gains, whereas T. arcuata showed weaker and more uncertain responses, including significant net losses under SSP1-2.6. These asymmetric trajectories suggest species-specific responses among the three legumes and provide hypotheses for how life-form-related ecological strategies may shape dryland habitat shifts. Given this limited replication, these trajectories should be treated as species-specific hypotheses rather than evidence for life-form effects. Future SDM assessments should broaden taxonomic sampling and incorporate key functional traits to improve forecasts of dryland community reassembly and conservation planning.