关键词: 生境质量, InVEST模型, CA-Markov模型, 地理探测器, 驱动因子, 伊犁谷地

Abstract: Based on land use classification from 1995 to 2015, we predicted temporal and spatial change of land use in Yili Valley in 2025 using the CA-Markov model. We assessed temporal and spatial patterns of habitat quality in Yili Valley from 1995 to 2025 using the InVEST model, and analyzed the characteristics of habitat quality changes using a spatial statistical method and geographical detector model. We further analyzed the spatial distribution and change trends of habitat quality by driving factor detection, interaction detection and ecological detection analyses. From 1995 to 2025, the area of cultivated land and construction land shows a gradually increasing trend, while that of forest land and unused land fluctuates. Water area shows a decreasing trend, with a slowly shrinking amplitude. The high-value areas of ecological degradation are mainly in the farmland and urban construction areas of the river valley and present as a stripe distribution, suggesting that ecological restoration projects should be strengthened. The low-value areas are alpine forests and grasslands, which are generally surrounded by high-value areas, with a nested distribution. Emphasis in those areas should be placed on the protection of ecological functions. The highand low-value areas of habitat quality show opposite trends to those of ecological degradation. The ecological degradation of each county is generally stable, with relative low degradation. The habitat quality shows a downward trend from 1995 to 2015, but increases from 2015 to 2025. The use of geographic detectors identifies that the spatial distribution of habitat quality is affected by a combination of natural and anthropogenic factors. The dominant influencing factor is land use, followed by slope. In the interaction detection and ecological detection, each factor significantly differs in affecting the spatial distribution of habitat quality, and shows a twofactor enhancement.

Key words: habitat quality, InVEST model, CA-Markov model, geographical detector, driving factor, Yili Valley.