Seed predation by natural predators can contribute to the regulation of the weed seedbank, hence has been acknowledged as a promising way to significantly reduce the use of chemicals and mechanical weeding to manage arable weeds. However, the complex aspects that determine seed species preference are not yet fully understood, especially, the relationship between weed seed traits and the feeding behaviour of seed predator species. In this study, we examine predation on 42 weed species by four ground beetle species as a function of two major seed traits: seed mass and seed lipid content. We used an _original experimental design to create a statistically independent trait value distribution for evaluating the separate and interactive effects of these two traits. Our results show that all the ground beetles species consumed weed seeds, but at different rates. The smallest weed seeds with high lipid content were among the most consumed. While seed mass might be limiting for ground beetles, due to ease of handling or the thickness of the seed coat, they may preferentially consume seeds rich in lipids to increase energy uptake. To our knowledge, this study is the first to reveal that two major seed traits drive seed predation by ground beetles.

Seedbank control has been the cornerstone of agricultural management practices. Regulating weeds by using their predators, as a weed control strategy, may be a prerequisite of decreasing herbicide use, and has thus attracted much research investigating the possible contribution of both vertebrates and invertebrates as weed seed predators. Carabid beetles are considered as one of the most important invertebrate seed predators in agroecosystems. We aimed at investigating carabid beetle preferences to a single prey type, seeds of Viola arvensis. We measured the consumption of seeds in 28 species of carabid beetles, under controlled experimental conditions. Two main tribes are identified in tested species, Harpalini with 12 species and Pterostichini with ten species. We found no relationships between species body mass and Viola's seed consumption, nor with the ratio between mandible length and labrum width (ML/LW). However, trends became significant with the ratio ML/LW when restricting these analyses to species that ate at least five seeds. In addition, we detected a positive and significant relationship between consumption rate and occurrence in trapping sessions over the last 3 years. These results are in favor of weed seeds control by carabids. Clear limits of this study are the use of a single seed species and under control conditions. This experimentation calls for additional studies to check for consistencies in consumption against seed species.

No abstract