Eeva M. Soininen

Herbivorous rodents in boreal, alpine and arctic ecosystems are renowned for their multi‐annual population cycles. Researchers have hypothesised that these cycles may result from herbivore–plant interactions in various ways. For instance, if the biomass of preferred food plants is reduced after a peak phase of a cycle, rodent diets can be expected to become dominated by less preferred food plants, leading the population to a crash. It could also be expected that the taxonomic diversity of rodent diets increases from the peak to the crash phase of a cycle. The present study is the first to use DNA metabarcoding to quantify the diets of two functionally important boreal rodent species (bank vole and tundra vole) to assess whether their diet changed systematically in the expected cyclic phase‐dependent manner. We found the taxonomic diet spectrum broad in both vole species but with little interspecific overlap. There was no evidence of systematic shifts in diet diversity metrics between the phases of the population cycle in either species. While both species' diet composition changed moderately between cycle phases and seasons, these changes were small compared to other sources of diet variation—especially differences between individuals. Thus, the variation in diet that could be attributed to cyclic phases is marginal relative to the overall diet flexibility. Based on general consumer‐resource theory, we suggest that the broad diets with little interspecific overlap render it unlikely that herbivore–plant interactions generate their synchronous population cycles. We propose that determining dietary niche width should be the first step in scientific inquiries about the role of herbivore–plant interactions in cyclic vole populations.