Nonlinear Associations of Carbon Density and Habitat Quality with Climate, Vegetation and Topography in the Qilian Mountains
Year: 2026
Type: article
Abstract: Habitat quality and carbon density showed contrasting environmental associations, with several SHAP zero crossings recurring across three boosting models. Background and Objectives: Nonlinear ecosystem-service relationships can reveal spatial heterogeneity relevant to conservation, but their sensitivity to model choice remains an important source of uncertainty. Materials and Methods: We combined the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model, 12 machine-learning algorithms, and Shapley additive explanations (SHAP). Model fitting and SHAP interpretation used five historical snapshots from 2000 to 2020; extrapolated estimates for 2025 were used only for illustrative temporal comparison. Results: Both indicators were generally higher in the southeastern than northwestern Qilian Mountains. Mean annual temperature and elevation ranked highest for habitat quality, with zero crossings near −4 °C and 3.84 km. Normalized difference vegetation index ranked highest for carbon density, with zero crossings at 0.16–0.17; evapotranspiration crossings occurred around 147–150 mm. Precipitation boundaries varied more among models. Conclusions: The contrasting associations identify candidate environmental gradients for monitoring and restoration assessment. Agreement among boosting models supports consistency within this model family, while the crossings remain statistical descriptors of model-derived indicators rather than demonstrated ecological tipping points.
Source: Forests
DOI: 10.3390/f17111375
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
SDGs: Life on land, Climate action
Open Access status: gold
Estimated APC paid: $3,137
Award: G3689591211
OpenAlex ID: W7235542520
