Accounting for seasonal patterns in bird availability prevents biased population trend estimates with advancing spring phenology

Abstract Advancing spring phenology has been observed around the world, including changes in the timing of breeding of birds. When singing rates are tied to breeding stage, the rate at which birds are available for detection by surveyors can also show seasonal patterns that may shift with spring phenology. As the timing of peak bird availability changes over years, monitoring programs that do not account for changing availability could incorrectly conclude that there is a change in population size. We used a 20-year point-count dataset to test for relationships between bird availability and spring vegetation phenology for 27 species in boreal Alaska. Nine of 22 migratory species showed a significant effect of day of spring (DOS) on availability, usually with availability declining over the survey window (late spring and early summer). In contrast, 3 of 5 resident species showed availability increasing over the survey window. We then conducted a simulation study to evaluate how changing spring phenology could affect estimates of population trend under a static survey window. We found that including DOS in the model as a covariate of availability prevented bias in the trend estimates and did not reduce precision. However, when the model ignored the effect of DOS on availability, population trend estimates were often significantly biased when spring phenology was advancing. Our study adds to previous evidence that bird availability is often related to spring phenology, and demonstrates that failing to account for seasonal changes in availability could result in the spurious estimation of a population trend when spring phenology changes over time. In some cases, the bias could be large enough to change species status assessments under IUCN Red List Criteria. Monitoring programs for birds and other taxa with seasonally varying availability could avoid bias by simply measuring and modeling the relationship between DOS and availability.