Predicting future community composition from random soil seed bank sampling – evidence from a drained lake bottom

Abstract

AbstractQuestions: What is the accuracy and reliability of the commonly used random soil sampling methodology for predicting seedling density, species richness and composition of the emerging seedling community?Location: Lake Kraenepoel, western Belgium.Methods: We compared density, species composition and observed and rarefactioned species richness of the seedling community emerging on a soft water lake bed exposed after drainage with the seedling community germinating in the laboratory from random soil samples in the same plots.Results: Seedling density did not differ between the two methods and there was a significant correlation between seedling density on the exposed lake bed and in the soil samples. This indicates that future seedling density can be reliably predicted based on soil sampling, in particular for the most abundant species. The most frequently occurring and abundant species among the seedlings in the soil samples were also the most frequent and abundant species germinating on the exposed lake bed. In contrast, species richness was much higher on the exposed lake bed than in the soil samples, and this difference was still significant for annual species after correction for differences in sampling intensity by rarefaction. We found no correlation between the number of species retrieved by the two methods. Although seedlings of rare and target species emerged on the lake bed, random soil sampling clearly failed to detect seeds of most of these species.Conclusions: Random soil sampling at a commonly used intensity and using the standard germination conditions can accurately predict future total seedling density and the density of the most abundant species. However, the method is not reliable for predicting the probability of establishment of populations of uncommon species. When executing a seed bank study, sampling intensity and germination conditions need to be adapted to the nature and the level of detail of the research question to be answered.