Stoichiometric and structural uncertainty in soil food web models

Abstract

Soil food web models are a powerful tool for calculating the carbon and nitrogen mineralized by different soil organisms. Two limitations of the current models are that they use (a) fixed parameters for determining the relative efficiency of carbon and nitrogen conversion and (b) a generic set of trophic species (i.e. nodes). We propose a new method for analysing soil food webs that varies production efficiency and diet mechanistically with resource stoichiometry. Then, we calculated how the lumping or splitting of trophic species affects carbon and nitrogen mineralization for both net mineralization rates as well as the distribution of mineralization (i.e. sources of mineralization). Our models with additional stoichiometric details better represent organisms that consume basal resources with high C:N ratios. This is important because we show that lumping together trophic species at low trophic levels, which often differ in C:N ratios (i.e. microbial taxa), causes the largest deviations in estimates of carbon and nitrogen mineralization. One reason for the large effect of C:N ratio is that it impacts diet and production efficiency in our stoichiometric model. Conversely, lumping together species at higher trophic levels causes the largest errors when those species have different production and assimilation efficiencies. Finally, we demonstrate that differences in death rate and biomass are mostly important when lumping species that have different diets. We suggest that C:N ratios are especially important when grouping microbial taxa because microbial C:N ratios vary widely and microbes feed at low trophic levels where differences in C:N ratio mattered the most. Conversely lumping higher consumers will be most problematic when they have differences in the conversion efficiencies. We provide an approach to calculate structural error and report it with parameter uncertainty. Then, we demonstrate, using oribatid mites as an example, that considering the life history of different taxa can help us group organisms more efficiently into trophic species. In doing so, we provide a way to reduce and report the uncertainty in soil food web models and to stimulate future empirical work measuring key life history parameters. A free Plain language summary can be found within the Supporting Information of this article.