Habitat and Time Are More Important Predictors of Weed Seed Predation than Space on a Diversified Vegetable Farm in Maine, USA

Abstract

Postdispersal weed seed predation is a significant source of weed mortality in agroecosystems. The magnitude of seed predation, however, is variable. Understanding the relative importance of factors driving variability in seed predation rates will increase the potential utility of seed predation to farmers. We conducted landscape-scale field experiments to quantify and compare the effects of space, time of sampling, and habitat on weed seed predation. Seed predation assays, with and without vertebrate exclosures, measured seed predation rates at spatially explicit sample sites across 8.5 ha of crop and noncrop habitats on a diversified organic vegetable farm in Maine. Total and invertebrate seed predation averaged 8% and 3% d−1, respectively. Vertebrate seed predators detected by motion-sensing cameras included small mammals and birds. A ground beetle,Harpalus rufipes, was highly dominant in pitfall traps, comprising 66% of invertebrate seed predators captured within crop fields. Seed predation was randomly distributed in space. However, time of sampling and habitat were highly significant predictors of seed predation. Variance partitioning indicated that habitat factors explained more variation than did time of sampling. Total seed predation was greater in crop and riparian forest habitats than in mowed grass, meadow, or softwood forest. Generally, invertebrate seed predation was greatest at sites with an intermediate degree of vegetative cover, whereas habitat type was the chief biotic determinant of vertebrate seed predation rates. These results suggest cover cropping and wetland conservation as practices that may bolster seed predation rates.