Lemmings construct nests of grass and moss under the snow during winter, and counting these nests in spring is 1 method of obtaining an index of winter density and habitat use. We counted winter nests after snow melt on fixed grids on 5 areas scattered across the Canadian Arctic and compared these nest counts to population density estimated by mark-recapture on the same areas in spring and during the previous autumn. Collared lemmings were a common species in most areas, some sites had an abundance of brown lemmings, and only 2 sites had tundra voles. Winter nest counts were correlated with lemming densities estimated in the following spring (r(s) = 0.80, P < 0.001), but less well correlated with densities the previous autumn (r(s) = 0.55, P < 0.001). Winter nest counts can be used to predict spring lemming densities with a log-log regression that explains 64% of the observed variation. Winter nest counts are best treated as an approximate index and should not be used when precise, quantitative lemming density estimates are required. Nest counts also can be used to provide general information about habitat-use in winter, predation rates by weasels, and the extent of winter breeding.

Determining the manner in which food webs will respond to environmental changes is difficult because the relative importance of top-down vs. bottom-up forces in controlling ecosystems is still debated. This is especially true in the Arctic tundra where, despite relatively simple food webs, it is still unclear which forces dominate in this ecosystem. Our primary goal was to assess the extent to which a tundra food web was dominated by plant-herbivore or predator--rey interactions. Based on a 17-year (1993-2009) study of terrestrial wildlife on Bylot Island, Nunavut, Canada, we developed trophic mass balance models to address this question. Snow Geese were the dominant herbivores in this ecosystem, followed by two sympatric lemming species (brown and collared lemmings). Arctic foxes, weasels, and several species of birds of prey were the dominant predators. Results of our trophic models encompassing 19 functional groups showed that <10% of the annual primary production was consumed by herbivores in most years despite the presence of a large Snow Goose colony, but that 20-100% of the annual herbivore production was consumed by predators. The impact of herbivores on vegetation has also weakened over time, probably due to an increase in primary production. The impact of predators was highest on lemmings, intermediate on passerines, and lowest on geese and shorebirds, but it varied with lemming abundance. Predation of collared lemmings exceeded production in most years and may explain why this species remained at low density. In contrast, the predation rate on brown lemmings varied with prey density and may have contributed to the high-amplitude, periodic fluctuations in the abundance of this species. Our analysis provided little evidence that herbivores are limited by primary production on Bylot Island. In contrast, we measured strong predator-prey interactions, which supports the hypothesis that this food web is primarily controlled by top-down forces. The presence of allochthonous resources subsidizing top predators and the absence of large herbivores may partly explain the predominant role of predation in this low-productivity ecosystem.