Characterizing soil seed banks and relationships to plant communities

Abstract

Estimates of soil seed banks are important to many ecological investigations and plant conservation, yet seed banks are among the most difficult plant community attributes to accurately quantify. To compare extraction and emergence seed bank characterization methods, we collected 0- to 5-cm soil seed bank samples and measured plant community composition in six microsite types (below different perennial plant species and interspaces) at 10 field sites in the Mojave Desert, USA. Extraction detected five times more species sample−1 and orders of magnitude greater seed density than emergence, though evaluating viability of extracted seed was not straightforward. Only 13 % of 847 tested seeds from extraction emerged in follow-up assays. Considering all sites, species detection was more similar between methods: 21 taxa for emergence and 28 for extraction. Results suggest that: (i) capturing microsite variation is critical for efficiently estimating site-level desert seed banks; (ii) method comparisons hinged on the scale of analysis for species richness, as differences in species detection between methods diminished when increasing resolution from the sample to the regional scale; (iii) combining data from all seed bank methods provided the strongest correlation with vegetation; and (iv) improving knowledge of seed germinability is important for advancing both seed bank methods, including for extraction to evaluate the proportion of extracted seeds that are viable. Multifactor approaches that balance several effectiveness measures (e.g., both seed density and species detection at multiple scales) and procedural challenges are most likely to accurately represent complexity in tradeoffs for choosing methods to quantify soil seed banks.