What counts when counting birds

Not choosing monitoring sites randomly or not surveying them every year can heavily influence the estimates of population size and trend.

Who wouldn’t love to spot a young Peregrine Falcon? During monitoring, observers often favour areas that offer a greater chance of detecting the species, but this tendency can cause methodological problems.
Photo © Mathias Schäf

When conducting a species survey, it can be tempting to visit the best and most promising locations first. Places that are harder to access or irregularly occupied are visited later or not at all. Such behaviour leads to a biased and overly positive sample of sites, and hence to problems when calculating population trends.

This was found to be the case in a recent study that analysed data from a Peregrine Falcon monitoring programme that started in the 1960s. In an area encompassing the entire Jura mountain range, 420 Peregrine Falcon nest sites were monitored between 2000 and 2020, but not all sites could be visited every year.

To solve the problem of non-random site selection in the data, so-called Bayesian occupancy models were used, with remarkable results: the counts suggested an increase in the population, while the models showed that it was in fact declining.

Citizen-science projects are especially prone to such non-random site coverage, because understandably, the volunteers want to survey as many locations as possible where the species in question is likely to be present, or to occur in greater density. It is all the more important to aim for random coverage, or else to employ models that correct the bias, so that better trend estimates can be achieved.

Kéry, M., G. Banderet, C. Müller, D. Pinaud, J. Savioz, H. Schmid, S. Werner & R. Monneret (2021): Spatiotemporal variation in postrecovery dynamics in a large Peregrine Falcon (Falco peregrinus) population in the Jura mountains 2000–2020. Ibis 156: 217–239. doi.org/10.1111/ibi.12999.