Microbial biomass and composition influence ecosystem functions such as soil C accrual during grassland restoration. Soil microbes can contribute to soil C directly, e.g., recalcitrant necromass, and indirectly through soil aggregation (e.g. physical entanglement of hyphae and aggregate promoting exudates). Plant inputs and physicochemical mechanisms of aggregate formation also contribute to soil C accrual. As such, we sought to develop a robust prediction of physically sequestered C (microaggregate-within-macroaggregate C) during grassland restoration and identify the major controls on this physically protected soil fraction using multiple restoration chronosequences located in Kansas (KS), Nebraska (NE), and South Africa (SA). We used a mixed effect model to determine microaggregate-within-microaggregate C response to restoration age with location as a random effect. We then used a confirmatory factor analysis to test hypothesized relationships, developed from KS data. This analysis revealed the relative strength of percent clay (texture analysis), root biomass, root quality (C:N ratio), soil structure (latent variable indicated by bulk density, microaggregate-within-macroaggregate proportional mass, and macroaggregate proportional mass), and microbial composition (latent variable indicated by microbial biomass C biomass, fungi:bacteria ratio, and arbuscular mycorrhizae biomass) as causal influences on physically protected soil C.
Results/Conclusions
Physically protected C was accrued at a rate of 16 g m-2 y-1 across all sites. Highest levels of physically protected C occurred in Nebraska, where soil contained the highest clay content. The model of protected C was tested with data from NE, SA, and all sites. Root quality (C:N ratio), rather than microbial composition, explained more variance in physically protected C in SA compared to KS. Less microbial influence on physically sequestered C in South Africa may have resulted from less precipitation received than other sites, which could reduce arbuscular mycorrhizae activity. Using data from all sites resulted in lack of model fit, suggesting that the major control on physically sequestered C was not measured. Differences in path coefficients among site-specific models suggest that major controls on protection of C vary by site. Carbon saturation deficit and factors affecting quality of soil organic matter (e.g. black carbon and N deposition) and physicochemical mechanisms of soil aggregation (e.g. wet-dry cycling and bioturbation) deserve further study and have the potential to improve understanding of controls on physically sequestered soil C.