Citation
Overview
Like many things, long-term monitoring of bird populations was interrupted during the COVID-19 lockdowns. Here, BTO scientists look at the effects of restrictions on participation in the Breeding Bird Survey during 2020 compared to previous years, and devise analytical methods to overcome the issues that emerged.
In more detail
We are thankful to the volunteers who participated in the BBS during 1994–2020. We thank Philipp Boersch-Supan and Rob Robinson for statistical input, David Noble (BTO), Paul Woodcock (JNCC) and Mark Eaton (RSPB) for further advice, and two reviewers for comments that improved the paper. This work was funded through a partnership with JNCC (on behalf of Natural England, NatureScot, Natural Resources Wales and Department of the Environment, Northern Ireland) and RSPB and also received funding from BTO Gifts in Wills. The BTO/JNCC/RSPB Breeding Bird Survey is a partnership jointly funded by the BTO, RSPB and JNCC, with fieldwork conducted by volunteers.
Abstract
Capsule: COVID-19 restrictions significantly biased BTO/JNCC/RSPB Breeding Bird Survey coverage across the UK allowing indicative trends to be produced for approximately one third of species in England only.
Aims: To investigate the effect that COVID-19 restrictions had on participation in and coverage of the Breeding Bird Survey, and to quantify the likely impacts on population change reporting based on 2020 data.
Methods: We determined geographic, seasonal and habitat coverage for the BBS in 2020 and compared this to previous years, and quantified the scale of biases and reductions in sample size for target species. We degraded existing BBS data (1994–2019) to simulate 2020 coverage and produced population change estimates using three methods to the complete and degraded data to assess the impacts of 2020 coverage on emergent trends.
Results: In 2020, 49% fewer survey squares were visited compared to 2019. Reductions were greatest in Wales, Scotland and Northern Ireland, and in the early breeding season, with 90% fewer visits made. The few early visits completed were on atypical dates and showed marked habitat biases. Individual species were detected in 23–96% fewer squares than normal. Population change estimates derived using routine trend models were negatively biased in up to 96% of species, with errors greatest for species normally detected on early visits. Alternative trend models using visit-specific parameterisation or focussing only on late-season visits overcame coverage biases for some species.
Conclusions: Lockdown restrictions meant it was not possible to produce population trend information for UK, Wales, Scotland or Northern Ireland in 2020. Indicative long-term trends could be produced in England only for a subset of c.40 species. We recommend managers of citizen science schemes undertake similar analyses to assess the scale of coverage biases when unforeseen events cause temporary, but substantial changes, in sampling effort.
