Understanding the soil nutrients that predict the species richness across a productivity gradient in a semi-arid grassland

Background/Question/Methods: The unimodal humped-back model (HBM) of the species richness-productivity relationship can be explained in part by resource availability. Few studies have explored the role of soil nutrients to predict richness in sites of differing productivity. We ask whether soil primary and secondary nutrients can predict patterns of species richness and productivity in low and high productivity sites within semi-arid in the Lac du Bois Grasslands Protected Area; a 15,712 hectare protected area located northwest of Kamloops, British Columbia, Canada. We selected four sites, two of the sites were of low productivity (~50 – 340 g/m²) and two high productivity (~570 – 1850 g/m²). The sites were located approximately 6 km apart. At each site, two 8 x 8 m grids, each containing 64, 1 m² plots, were established between 300 – 700 m apart. Plant species were identified and above-ground biomass harvested in each of the 1 m² plots. In addition, a 10 cm × 10 cm × 10 cm soil core was taken from the center of each plot. Soil was analyzed for primary and secondary nutrients. We used random forest models to identify the most important nutrient predictors of species richness in low and high productivity sites, and linear regressions to examine the relationship between soil nutrient data. Distance and correlation biplots of the important axes from principal component analysis were done to select the important soil nutrients for regression analyses.

Results/Conclusions: The relationship between richness and productivity across the full dataset was consistent with the HBM of species richness. Productivity was unrelated to richness in low productivity sites, but decreased with richness in high productivity sites. The PCA of the soil nutrients indicated that the cumulative percentage of variation in the data accounted for by the first two axes was 76.5%. Different nutrients were important in predicting species richness in low and high productivity sites. In low productivity sites, C, S, Na, in addition to other nutrients were important while in high productivity sites, Mg and productivity were chosen as important factors. In multiple linear regressions, primary and secondary nutrients were important predictors of species richness. Multiple soil nutrients have been shown to help predict plant species richness and biomass, and therefore may be used to better explain variation in species richness and biomass patterns in grasslands.