Published online by Cambridge University Press: 20 January 2017
Accurate and precise estimation of weed seed bank populations is critical to studying weed seed bank demographics. Research was conducted over 3 site-years in Montana to (1) examine the spatial distribution of wild oat seed banks on a small scale (1 m2 plots), (2) compare wild oat seed bank density sample means and precision between two soil samplers, and (3) predict the sample area needed to quantify a range of wild oat seed bank densities at several levels of precision. Seed bank sample means obtained with a large sampler (10- by 10-cm box) were greater than means obtained with a small sampler (4.4- or 3.8-cm-diam cylinder) for 15 of 18 seed banks. There was no clear advantage in precision (SE/mean) when sampling seed banks using a large number of small soil samples rather than using a small number of large soil samples. Furthermore, at very low seed bank densities, using a small number of large samples gave better precision. Precision improved as sample number increased for each seed bank at each site-year. High-density seed banks tended to have better precision than low-density seed banks at any given sample number. Seed banks had an aggregated spatial distribution when sampled with either soil sampler. As the desired precision level decreases (becomes more precise), the predicted sample area required increases greatly. A seed bank containing 6,000 seeds m−2 has a predicted sample area of 0.5, 0.7, 1.3, and 2.9% of the total area to obtain precision levels of 0.5, 0.4, 0.3, and 0.2, respectively.