Abstract
Scatter-hoarding rodents act as both seed predators and dispersers, strongly influencing seed fate and plant recruitment. Their foraging decisions are influenced not only by the traits of individual seeds but also by the traits of neighboring seeds. Although the effects of seed traits such as size and chemical defense are well-established, how the relative frequency of co-occurring seeds influences these decisions remains poorly understood. In this study, we conducted a field experiment in a subtropical forest in southwestern China using artificial seeds. A total of 19,200 artificial seeds with different seed sizes (diameters of 0.5, 1.5, and 2.5 cm) or tannin content (0%, 5%, and 10%) were deployed across 240 seed patches. Within each patch, the ratios of paired seeds were set across five levels (9:1, 7:3, 5:5, 3:7, and 1:9). We found that seed removal exhibited clear frequency-dependent patterns. Specifically, large seeds were removed less frequently as their relative abundance increased, whereas small seeds showed the opposite trend when paired with larger seeds. Similar frequency-dependent responses were observed for tannin treatments, although these effects were context-dependent. In contrast, caching probability and dispersal distance remained largely unaffected by relative frequency. These results demonstrate that rodent foraging is jointly shaped by seed traits and frequency-dependent processes, primarily during the initial removal stage. Our findings provide a mechanistic basis for understanding context-dependent seed dispersal and highlight the role of frequency-dependent selection in shaping plant community dynamics.