Chemical identity predicts carbon mineralization and non-additive effects of leaf mixtures

Background/Question/Methods Non-additive effects—deviations from the mean response of individual species—have been well-documented in the decomposition of litter mixtures. Species richness and functional group richness have proven insufficient to predict process rates of mixtures or the strength and direction of their interactions. While species identity is known to influence the decay of mixtures, a species roster cannot be extended to predict the degradation of mixtures of differing species composition. Nor are standard measures of litter quality such as initial N concentration, or lignin:N ratio without limitation, as their relationship to decomposition may shift with time as well as environmental context. The purpose of this study was to determine the ability of chemical identity, the comprehensive chemical characterization by mid-infrared spectroscopy, to predict rates of carbon mineralization of two-species leaf mixtures and their non-additive effects. Dried foliage of ten tropical species formed the basis of 32 treatments consisting of 10 single species, 21 pairs and a foliage-free control. Three “key” species exhibiting slow, medium and fast carbon mineralization rates in a preliminary study were paired with seven “companion” species displaying a range of mineralization rates. Microbial respiration of the 32 treatments (1:50 tissue: sand at 28ºC x 4 replicates) was followed for 12 weeks. Multiple linear regression models constructed from all possible pairs of spectral points from the key and companion spectra of leaf mixtures were evaluated for their predictive capacity of cumulative respired CO₂, C mineralization rate, and their non-additive effects.

Results/Conclusions Chemical identity of key and companion species accounted for as much as 87% of the variation in the C mineralization rates and cumulative respired CO₂ of the 21 leaf mixtures. Furthermore, of the suite of spectral traits that generated predictive models, a subset was able to explain 71% of the variation in the C mineralization rate of an independent test set of 15 leaf mixtures decomposed under different soil and temperature conditions. The magnitude and direction of mixture interactions could not be predicted on the basis of only two spectral points; however, mixtures were successfully separated into synergistic, antagonistic and no non-additive effects by linear discriminant analysis at p<0.05 using the first four principal components of the infrared spectra of their key and companion species. With a simple, limited data set, our results suggest that the calibration of infrared spectra to the C mineralization rate of mixed litters is tenable with a data set of sufficient sample size.