Abstract
Dividing soil respiration (Rs) into autotrophic respiration (Ra) and heterotrophic respiration (Rh) represents a pivotal step in deciphering how Rs responds to environmental perturbations. Nevertheless, in arid ecosystems beset by environmental stress, the partitioning of Rs and the underlying mechanisms through which microbial and root traits govern the distinct components remain poorly understood. This study was strategically designed to investigate Rs and its components (Ra and Rh), soil properties, and root traits within the desert-oasis ecotone (encompassing the river bank, transitional zone, and desert margin) of northwest China. Employing metagenomics, we quantitatively characterized microbial taxonomic attributes (i.e., taxonomic composition) and functional attributes (specifically, functional genes implicated in microbial carbon metabolism). Field measurements during the growing season of 2019 unveiled a pronounced decline in soil respiration rates along the environmental gradient from the river bank to the desert margin. The mean soil respiration rate was recorded as 1.82 ± 0.41 μmol m(-2) s(-1) at the river bank, 0.49 ± 0.15 μmol m(-2) s(-1) in the transitional zone, and a meager 0.45 ± 0.12 μmol m(-2) s(-1) in the desert margin. Concomitantly, the Ra and Rh components exhibited a similar trend throughout the study period, with Rh emerging as the dominant driver of Rs. Utilizing random forest modeling, we unearthed significant associations between microbial taxonomic and functional features and Rs components. Notably, both Ra and Rh displayed robust positive correlations with the abundance of phosphatidylinositol glycan A, a key player in microbial carbon metabolism. Partial least squares path modeling further elucidated that soil properties and microbial functions exerted direct and positive influences on both Ra and Rh, whereas taxonomic features failed to register a significant impact. When considering the combined effects of biotic and abiotic factors, microbial functional attributes emerged as the linchpin in dictating Rs composition. Collectively, these findings suggest that a trait-based approach holds great promise in more effectively revealing the response mechanisms of Rs composition to environmental changes, thereby offering novel vistas for future investigations into carbon cycling in terrestrial soils.