Systematic differences in phenology estimates from unstructured and structured biodiversity datasets

Abstract Phenology, the timing of biological events such as first flowering or emergence, plays a key role in understanding ecological responses to climate change. With the rapidly growing number of opportunistic citizen science records, it is important to understand whether and how phenological estimates differ between unstructured and structured recording approaches for animals and plants. We compared phenological metrics derived from unstructured records in iRecord Butterflies with those from structured, transect‐based monitoring in the UK Butterfly Monitoring Scheme (UKBMS). Focusing on single‐generation species, both data sources produce similar estimates of temporal trends in phenology, but unstructured data shows systematically earlier weighted mean flight dates (7.57 ± 1.57 days). It also reports earlier onsets (9.38 ± 1.24 days) and later ends (−8.46 ± 4.55 days) than UKBMS, resulting in longer inferred flight periods. Results may reflect intrinsic differences in sampling between the two datasets, as well as recorder behaviour and motivations. Moreover, these differences are independent of sample size, suggesting that iRecord Butterflies observers may systematically search for early and late flying adults. Recognising and accounting for phenology sampling differences will be important when using citizen science data for phenological inference in butterflies and other taxa.