Relating sampling effort to the detection of spatial patterns of species richness in the soil seed bank of a mediterranean shrubland.

Studying the soil seed bank is a time and space-consuming task, and therefore only a small fraction of the soil is sampled. It is then critical to optimize sampling effort to reliably estimate soil seed bank properties. Here, we test whether the spatial patterns of species richness in the soil seed bank differ under increasing sampling effort. For this, we used data of germination from soils in a mediterranean shrubland in Central Spain. Two data sets were used, one of the seedlings emerging after heating the soil to break dormancy, and one with the combined germinations of heated and non-heated soil subsamples. We simulated increased sampling effort with sample-based rarefaction curves, extrapolating the species richness corresponding to a 2x and 3x increase in the number of individuals (seedlings) per sample. We then analyzed the spatial pattern of the original and extrapolated species richness using linear regression and semivariograms. Species richness increased by 34% and 52% in the 2x and 3x estimations, however the spatial pattern of species richness remained largely unchanged. For the long-distance spatial pattern, the slope of the plot-scale trend (i.e., the regression coefficient) increased only slightly with increasing sampling effort, while the adjusted R-squared of the regression decreased with increasing sampling effort. For the short-distance spatial pattern we could only fit spherical model semivariograms to the data from soils exposed to a heat shock, and the intensity of the spatial pattern (spatial dependence) increased very slightly with increased sampling effort. These results suggest that even with a doubled or tripled sampling effort, as provided by the simulations, the spatial pattern of species richness would have remained unchanged. We argue that increased effort in detecting species in the seed bank needs not necessarily improve the detection of spatial pattern.