Comparing Soil and Bison δ13C to Field Estimates of C4 Plant Abundances in North America

Abstract:

Stable carbon isotope data (i.e., δ13C) from soils and herbivore tissue are commonly used as a proxy for the relative abundance of C4 and C3 plants at a site. These data are also increasingly used to represent other climatologically relevant properties of vegetated environments, such as productivity, aridity, water use efficiency, and tree cover. The δ13C values of soils and herbivore tissues are generally assumed to resemble their source vegetation, after accounting for diverse fractionation processes during litter decomposition and tissue metabolism and turnover. However, δ13C values have rarely been compared to source vegetation at a regional to continental scale. As a result, the quality of δ13C as a proxy has not been thoroughly evaluated, and the importance of modifying factors have not been assessed at biogeographically relevant scales. To address some of these issues, we combined three multi-source datasets from North America: herbaceous C4 plant abundances from thousands of vegetation plots, hundreds of soil δ13C measurements, and hundreds of bison collagen, hair, and enamel δ13C data (tissues with different turnover rates). These datasets were resampled to common grid for comparison. A stronger relationship with C4 vegetation existed for bison as compared to soil δ13C. To determine which factors might explain deviations in the vegetation plot and isotopic data, we used statistical models to quantify the influence of soil variables, mean annual precipitation and temperature, tree cover, presence of invasives, and fire frequency in conjunction with plot-based C4 abundance data. Our bison model was only improved by the addition of invasives. In contrast, our soil model was significantly improved when accounting for tree cover (C3 vegetation and shade), precipitation, various soil parameters, and invasive grasses, suggesting that soils are more likely to be biased from source vegetation in ways that could influence interpretation as a proxy at broad scales.