The influence of survey frequency on population estimates of moorland breeding birds

Capsule A minimum of four constant-effort-search survey visits are required to generate reliable population estimates of breeding birds on moorland that are not subject to biases associated with varying levels of detectability through the season. Aims To investigate the influence of the number and the combination of survey visits on the population estimates of breeding birds on moorland. Methods Four constant-effort-search surveys (80-100 minutes per km 2 per visit) of moorland in southwest Scotland were undertaken in each of six years, 2003-2008. Using standard protocols, the numbers of apparent territories that would have been identified for each possible combination of survey visits were determined.glms were used to assess the influence of the frequency of survey visits, and different combination scenarios on the derived population estimates for Red Grouse, European Golden Plover, Common Snipe, Eurasian Curlew, Sky Lark, Winter Wren and Stonechat. Independent assessments of population density were made by transect sampling for Red Grouse and Sky Lark. Results Robust population estimates were possible from three survey visits for European Golden Plover, Eurasian Curlew and Stonechat. However, there were differences between species in the seasonal variation of their detectability. Four survey visits would underestimate the populations of Red Grouse (probably by 67-91%), Sky Lark (probably by 31-61%) and Winter Wren (by an undetermined proportion). Common Snipe were also likely to be underestimated after four survey visits, but the value of the derived estimate as an index of population density deserves further investigation. Conclusions If there is a need to carry out a multi-species survey on moorland, we suggest that a minimum of four survey visits is required to ensure the derivation of reliable population estimates for a suite of the most readily detectable species. Population estimates derived from three or fewer survey visits risk biases through uneven sampling in periods of differing detectability. With evidence for changes in the breeding phenology of birds associated with changing climate or weather patterns, it arguably becomes more important to ensure that surveys sample an adequately broad period of the breeding season.