Chemical composition and diversity influence non-additive effects of litter mixtures on soil carbon and nitrogen cycling: Implications for plant species loss

Diverse plant litter mixtures frequently decompose differently than expected compared to the average of the component species decomposing alone, and it remains unclear why decomposition may respond non-additively to diversity. Here, we hypothesized that litter chemical composition and chemical diversity would be important determinants of the strength and direction (synergistic versus antagonistic) of non-additive soil carbon (C) and nitrogen (N) cycling responses to litter mixtures. To test this, we performed a soil incubation experiment using litter mixtures comprised of up to four plant species, and we measured three components of decomposition: respiration, net N mineralization, and microbial biomass N accumulation. We used nine chemical traits to calculate the chemical composition and diversity of the litter mixtures. First, we found that respiration responded as the average of the individual species in the mixture (i.e. additively), rather than non-additively as initially predicted. Second, litter mixtures stimulated significantly more net N immobilization than expected in 64% of cases, and non-additive responses were highly dependent on mixture chemical composition, and were influenced to a lesser degree by chemical diversity. Specifically, concentrations of tannins and certain low molecular weight phenolics in the mixtures were positively correlated with greater N immobilization than expected. Non-additive N mineralization responses were poorly correlated with traditional measures of litter chemistry like N concentration, C:N, lignin:N, and phenolic:N. Our results also show that non-additive N mineralization responses were affected by loss of some species significantly more than others, and the effects of species loss could depend on 1) whether a species contains compounds with strong effects on non-additive responses; and 2) whether those compounds are also found in other species. Finally, litter mixtures stimulated more microbial biomass N than expected in 45% of cases, but non-additive responses were only weakly dependent on the litter chemistry variables that we measured.