Abundance‐related systematic bias in the quantification of phenology using site presence‐based survey methods

Abstract

Reliable quantification of the timing of spring phenology is important to the understanding of ecological responses to climate change. Citizen science data have a potentially useful role supporting these types of studies. Site presence‐based methods represent a relatively simple means of following spring phenology, including migrant bird arrival. However, binary presence/absence observations are ill‐suited to following the build‐up of individuals during the spring emergence or arrival period because presence may be detected uniformly after a small proportion only of the population is present. The reliability of this survey method has been evaluated by mathematical modelling, further supported by comparison of model predictions with arrival date estimates determined in previous field studies. Modelling demonstrates a systematic abundance‐dependent bias in median arrival dates estimated using site presence‐based survey methods in which the apparent distribution is shifted increasingly ahead of the true distribution with increasing abundance, a feature also evident in field observations. Since this error in the estimate changes with abundance, abundance changes with time will lead to a distorted picture of the phenological trend with time, impacting on the reliability of these methods for characterizing phenological events. The model provides a general framework for identifying when this inherent bias will arise, and for compensating for it by reference to count data obtained from other sources, thereby assisting in the provision of improved estimates of phenological change. To ensure that the most appropriate conclusions concerning phenological change are drawn from studies using site presence data, it is imperative that these identified methodological limitations are recognized and properly considered during data interpretation.