An ecologist notes that important details are missing from climate-change models.

Unmitigated climate change will gravely reduce Earth's biodiversity. How much this will happen is calculated by combining data on how the species richness of different habitats varies with their area and projections of how much various habitat types will shrink as the planet warms.

But such grand analyses are blunt instruments; they miss numerous local processes. I have seen, for example, rosy finches and ptarmigans feeding on the contracting ring of vegetation that surrounds melting snow patches on Alpine slopes. Would these creatures survive the summer if the snow patches melted in late spring rather than late summer? Formed in existing mountainside hollows, snowbeds will not march uphill as the climate warms.

This question was recently answered by Robert Björk and Ulf Molau, then both at the University of Gothenburg in Sweden. They reviewed how the release of water and nutrients from the contracting edge of lingering snow patches sustains alpine life in midsummer by providing nourishing vegetation (R. Bjork and U. Molau Arctic Antarctic Alpine Res. 39, 34–43; 2007). The duo propose that bryophytes, grasses, sedges and rushes will be worst hit by the patches' earlier annual disappearance, and that these easy-to-graze species will be replaced by shrubs and trees, hitting Alpine herbivores hard.

This is just one example of the many impacts on biodiversity that fall through the cracks of current, coarse projections. Life and climate intersect on fine spatial and temporal scales — in the microclimates provided by terrestrial 'nurse plants' and in rock pools that form fleetingly in bedrock depressions. The disruption of these delicate intersections may add up to even more damage to biodiversity than the large-scale models predict. This deserves more study.

Discuss this paper at http://blogs.nature.com/nature/journalclub