Performance of multiscale landscape metrics as indicators of soil C, N, P, and physicochemical properties of NFV in ML reaches of Yellow River Basin, China

• NFV in the middle reaches had higher soil C and N, but lower soil P, bulk density, EC, and pH. • Differences in forest area between reaches resulted in spatial variation of NFV landscape pattern. • Forest PD and ED performed well in predicting soil C and N before the threshold in larger scales. • Forest area metrics performed well in predicting soil P, bulk density, EC, and pH in most scales. Landscape metrics are significant indicators of ecological functions. However, limited data are available regarding the relationship between landscape metrics and soil ecological functions. In this study, using 99 National Forest Villages (NFVs) in the middle and lower reaches of the Yellow River Basin as study objects, landscape metrics were computed for circular areas of ten extents in FRAGSTATS. Differences in soil carbon (C), nitrogen (N), phosphorus (P), physicochemical properties, and landscape metrics between the middle and lower reaches were compared using a one-way ANOVA. Multiscale performance of 14 forest landscape metrics as indicators of 10 soil parameters was explored using several methods: partial correlation, linear mixed model, linear mixed-effects model, and threshold model. The results showed that NFVs in the middle reaches had higher soil C and N and better physicochemical properties, but lower soil P than the lower reaches. The landscape pattern showed large variations between the middle and lower reaches. Soil C and N were positively correlated with patch number and shape metrics, respectively, before the threshold. Soil P and physicochemical properties were negatively correlated with forest area metrics before and after the threshold. The correlation between landscape metrics and soil parameters depended on kilometer scale. Our findings highlight that soil characteristics and forest landscape patterns of National Forest Villages are linked, and landscape regulation can improve forest soil C and N, as well as the physicochemical properties. The insights from this study are vital for informing land management decisions aimed at enhancing carbon sequestration, mitigating climate change, and maintaining soil fertility. Ultimately, this research provides a reference for other river basins and landscapes, and future research should consider human activities at larger kilometer scales to predict soil properties.