We would like to clarify points raised in William Sutherland's criticism of the treatment of pollinators in the UK National Ecosystem Assessment (Nature 503, 167; 2013).

The estimated economic costs of pollinator decline are only as robust as the natural science on which they rest, as Sutherland indicates. If we could predict with certainty the effects of changes in pollinator populations on agricultural production, then evaluating them would be trivial.

It was because of uncertainty in the underlying population ecology that we omitted estimates of pollination services from our economic analysis of the impacts of land-use change in our report, which was extensively peer-reviewed (see also I. J. Bateman et al. Science 341, 45–50; 2013).

The Delphi technique — a consensus method that Sutherland mentions for synthesizing research findings — can be helpful in some situations, but should be applied with caution to environmental valuation. The rapid expansion of empirical literature in this field means that conventional beliefs can rapidly become group-think norms, with dangerous consequences.

For example, we rejected the popular consensus in favour of using survey techniques as a way of valuing biodiversity, choosing instead to estimate the costs of ensuring species conservation. We stand by our approach, which we believe conforms with Sutherland's appeal for quality over quantity.