Phenological mismatch between breeding birds and their surveyors and implications for estimating population trends

Author(s): Massimino, D., Harris, S.J. & Gillings, S.

Published: September 2020

Journal: Journal of Ornithology Volume: 162

Digital Identifier No. (DOI): 10.1007/s10336-020-01821-5

Several studies in recent decades, including those led by BTO, have demonstrated that many birds are migrating or breeding earlier as the climate changes. These so-called phenological shifts could have implications for monitoring, if people counting birds also change the dates on which they make surveys in ways that affect their likelihood of detecting birds. New BTO research has sought to investigate this, using data from the BTO/JNCC/RSPB Breeding Bird Survey (BBS). Data collected in South-East England were compared between two different five-year periods: 1994–98 and 2013–17. SE England was chosen because this is the region of the UK in which the largest sample of BBS squares is available, and the study's authors wanted to eliminate any regional biases in their findings.

The research showed that BBS volunteers advanced the dates of their visits by up to four days between 1994–98 and 2013–17. By contrast, the detection patterns of most species had changed less, generally by no more than two days, and were both advanced and delayed. For some species this means that surveys may better hit the peak of activity than they used to, whilst for others, surveys may now tend to miss the peak of activity. Although possible biases introduced to BBS results and population trends derived from these data were mostly small, they could have consequences, especially for the classification of Red-listed species. For example, the decline in Kestrel numbers becomes great enough for this species to qualify for the Red List if survey timing is accounted for.

BBS monitors around 120 bird species using a single method that is designed to be generally applicable to a wide range of species. The authors note that whilst it might be tempting to shift surveys progressively earlier, not all species are breeding earlier, with some now breeding later, meaning it is impossible to track every species. They suggest that surveyors should maintain fixed survey dates as much as possible, and instead visit timing should be factored into the analysis of BBS data. Overall, the research demonstrates the importance of accounting for accelerating phenological change in the design of surveys for monitoring wildlife and the statistical models used to analyse the resulting data, especially given current climate change scenarios.

Abstract

Citizen science monitoring of common birds often involves volunteers visiting selected survey locations twice per year, with interannual differences in the number of birds detected used to infer population trends. Two processes, changes in the timing of migration and breeding in response to climate change, and changes in the timing of surveys by volunteers, may cause variation in the number of birds detected, leading to biases in inferred population trends. We assessed the magnitude of potential biases using the UK Breeding Bird Survey, comparing survey timing, species phenologies and apparent trend biases between 1994–98 and 2013–17. To control for large-scale geographic effects, we focussed on a subset of 888 surveyed 1 km squares in South-East England. Survey dates became significantly earlier, advancing by 2–4 days on average. We calculated seasonal patterns of bird abundance for 68 species. After standardising these to remove long-term abundance trends, median detection dates were advanced by 0.82 days on average. At the species level, the majority of changes were ± 2 days and only five species showed a significant advancement in median detection date. However, species’ phenological changes alone are capable of inducing between an 8% suppression and 21% enhancement of species’ trends, although the majority are ± 2%. Effects of a similar magnitude are apparent if survey timings are also allowed to change, although different species are affected. Small modifications to the statistical model used to generate trends can control for changes in survey timing, but without additional survey visits, or using data from other sources, we cannot currently control for seasonal variation in detectability. Although the average effects shown here are small, biases could become increasingly important for some species, and we recommend organisers of biodiversity monitoring schemes assess whether their methods are resistant to variations in species phenology and survey timing.

Notes

The authors thank the thousands of volunteers who have taken part in the Breeding Bird Survey. The BTO/JNCC/RSPB Breeding Bird Survey is a partnership jointly funded by the BTO, RSPB and JNCC, with ieldwork conducted by volunteers.