Spatiotemporal Patterns of Habitat Quality and Influencing Factors in the Loess Tableland Region Using Optimal Parameters-Based Geographical Detector: A Case Study of Qingyang City, China

Habitat quality (HQ) dynamics in ecologically fragile Loess Plateau regions serve as a critical indicator of ecosystem health and sustainability. However, the spatiotemporal patterns, underlying drivers, and optimal analytical scale of HQ change remain inadequately quantified in this region. This study systematically investigated the spatiotemporal evolution and driving mechanisms of HQ in Qingyang City (1990–2023) through an integrated approach combining InVEST modeling, spatial autocorrelation analysis, and the optimal parameters-based geographical detector (OPGD). The results revealed three key findings: (1) cropland decreased by 2274.90 km², which accounted for 31.3% of its 1990 baseline, with the majority of this loss being converted to grassland and forest. Meanwhile, impervious surface expansion was concentrated in valley flatlands, driven by topography and population agglomeration. (2) HQ exhibited a spatial pattern of “low in the south and north, high in the west and east” with high-value areas clustered in the western Ziwuling forest and eastern plains, and low-value zones in southern loess gullies and urban peripheries. Spatial autocorrelation analysis indicated strong HQ clustering, though this weakened by 2023, suggesting intensified human disturbance. (3) The OPGD model identified 4 km as the optimal spatial scale for analysis. Population (q = 0.3037) and mean annual temperature (q = 0.2795) were the dominant factors influencing HQ variation, followed by NDVI (q = 0.2568) and GDP (q = 0.1918). Interaction effects were strongest for population ∩ mean annual temperature (q = 0.4517) and mean annual temperature ∩ slope (q = 0.4515). These findings provide both a methodological framework and practical guidance for ecological management in loess tableland regions and similar fragile ecosystems worldwide.