Incorporating field sampling to reflect biodiversity protection function in arid region

Biodiversity is integral to achieving multiple Sustainable Development Goals, yet global change threatens its preservation in the Anthropocene. While habitat quality (HQ) has emerged as a widely adopted metric for large-scale biodiversity conservation, accurately quantifying HQ changes across same land use types remains challenging. We developed an enhanced model by integrating a composite environmental index with field data (plant biomass and biodiversity index) into the InVEST-HQ module, enabling robust spatiotemporal HQ assessment in Xinjiang’s dryland ecosystems from 1990 to 2022. The findings showed that our modifications significantly improved model accuracy, with the correlation coefficient (between actual measurement and evaluation values) increasing from 0.1 to 0.58 in forests and 0.3 to 0.76 in grasslands. InVEST-HQ model showed limited sensitivity to HQ changes, and HQ improved by NDVI exhibited habitat type limitations for HQ assessments. The comparative analysis of HQ trends revealed distinct patterns across modeling approaches. While the original InVEST-HQ model showed minimal temporal variation with slight degradation in the study area, a NDVI-modified result exhibited a marked increment in the area. The HQ modified in this study decreased for plains and increased for mountain areas. Our modified method expresses significant improvement in shrubland HQ (13.9%), grassland HQ (5.7%) and urban HQ (13.7%) and marked decline in forest ecosystems (−4.5%). By incorporating region-specific environmental parameters, our framework reliably captures HQ dynamics even in human-dominated landscapes (e.g., urban/agricultural areas), offering realistic decision-making and data support for ecological conservation in fragile arid.