Plant diversity is associated with the amount and spatial structure of soil heterogeneity in meadow steppe of China

Abstract

The link between environmental heterogeneity and diversity is a major tenet of plant ecology. Previous studies designed to test the heterogeneity–diversity hypothesis largely have only included measures characterizing the overall variation in habitat (e.g., CV of soil parameters). Rarely has the spatial structure of that variation been considered in relation to diversity. Here we examined the spatial variability (CV) and spatial structure of that variation (i.e. spatial scale of patchiness) of several main soil variables (C, N, P, pH, and conductivity) in relation to grassland plant species richness and diversity (H′). We determined the relationships of plant species richness and diversity at two spatial scales (50 × 50 m plot scale, 1 × 1 m quadrat scale) with the whole-plot soil heterogeneity within an ~750 ha natural area of Kerqin grasslands in northeastern China. We found that the best models describing species richness at the 0.25 ha and 1 m2 scales included patch size of soil conductivity and N, respectively. For each of the two spatial scales, pairs of models best described H′; a simple regression with CV of soil N and a multiple regression including soil N patch size and CV at the 0.25 ha scale, and, at the 1 m2 scale, a simple regression with soil conductivity CV and multiple regression including CV and patch size of soil N. Soil N was negatively associated with conductivity, likely due to sodium, the primary determinant of conductivity in this meadow steppe system, inhibiting plant growth and the capacity of soils to accumulate N. Consequently our results indicated that the heterogeneity of soil N was the principal control of plant species richness and H′. Moreover, our findings indicate that spatial structure (the average size of a patch), in addition to CV, was important in determining grassland species richness and diversity. Our results indicate that both components of environmental heterogeneity need to be included in future tests of the heterogeneity–diversity hypothesis.