Andrew Dobson and colleagues overstate the argument against a road across the Serengeti National Park in Tanzania (Nature 467, 272–273; 2010). Migrant wildebeest already regularly cross transportation arteries in the Serengeti and in adjacent linked ecosystems. Established evidence of impacts on declines in migratory wildlife populations is centred not on roads, but on fencing — a separate and more remote possibility.

Conservation policy in the Serengeti has too often been based on idealized notions of what landscapes should look like. The Serengeti ecosystem has been shaped by thousands of years of human use, and is now groomed by park managers rather than by local inhabitants.

The ecosystem is supported and enhanced by pastoralist land use in surrounding buffer zones. Dobson et al. do not consider the likely importance of the road for the land use, mobility, livelihoods and welfare of people living near the park.

Wildlife tourism is a major contributor to Tanzania's economy. But local people, whose land-use practices have created and now maintain the ecosystems valued by Western conservationists, are being dispossessed by state and international elites who capture much of the tourism revenue and reinvest it in conservation-incompatible land use.

Wherever the road goes, it will have an environmental impact as well as socio-economic costs and benefits that require rigorous analysis of good data.