An objective of plant ecology is to understand how environmental gradients affect the structure of plant communities. Soil fertility is one such gradient that is hypothesized to strongly influence the assembly of species. However, methods of quantifying this gradient, developed in agronomy and tuned precisely for specific species, specific soils and specific climates cannot be used (i) either at the global scale, (ii) or to compare places where the causes of the soil fertility are different, (iii) or at the interspecific scale of natural communities. In this study, we propose a definition of the soil fertility that can be applied at the multispecific scale of plant communities, is independent of its causes, and we provide a method to measure and quantify it. To do this, we used structural equation modelling. The measurement submodel predicts values for the latent generalized fertility (FG) using four indicator variables: the relative growth rates (RGR) of Festuca rubra, Trifolium pretense, Triticum aestivum and Arabidopsis thaliana. The model was calibrated using 76 soils of the south of Quebec, Canada and independently tested using the aboveground net primary productivity (NPP) of the natural vegetation over two growing seasons.
Results/Conclusions
Both the measurement submodel and the full structural equation model fit the data well. In this paper, the direct causes of the latent variable FG were the supply rates of NO3-, P and K while the indirect causes were the texture, the organic matter content and the water holding capacity. The FG values predict 51% of the variance in NPP and was a better predictor than any other single variable or actual nutrient flux rates. Furthermore, because of the modular nature of this model in which the causes and effects of soil fertility are clearly separated, the FG values remain comparable even if the causes of this fertility (i.e. the soil characteristics) can vary from a place to another. Therefore this model can be applied to other regions and other grassland soils anywhere.