How much can local weather conditions explain intra- and inter-annual variation in raptor migration counts at a Neotropical hotspot? A new analytical approach

Abstract Every year, millions of raptors migrate between their breeding areas in North America and their nonbreeding areas in Central and South America. Many investigations have demonstrated that weather plays a significant, yet complex role in the timing of bird migration at northern localities that, in turn, shapes their seasonal migration phenology. However, very little information is available for the Neotropical region. Here, we analyzed the relationship between local weather variables (wind-related components, humidity, and atmospheric pressure) and the intra- and inter-annual migration patterns of 8 raptor species with different migration ecologies. Our dataset spans between 23–26 years and was obtained at a raptor migration monitoring site at Mexico’s Gulf coastal plain, a well-known, globally important migration hotspot. Perhaps due to the often erratic and pulsed daily variation in migration phenology each year, classical analytical approaches do not consider quantifications of intra-annual variation. To address this, we devised a new method based on a finite approximation to the second derivative of the counts (roughness) to analyze the role of weather drivers in the patterns of daily counts. Additionally, we analyzed the inter-annual variation in migration timing, subtracting the mean number of individuals across all years for each day (deviations from the mean). Although we found significant differences in roughness across days, we observed a subtle association between weather and roughness in migration counts, regardless of the species’ flight strategy (such as thermal soaring and flapping). Only tailwind affected the intra-annual migration, although the models had a low explanatory power (marginal R2 ≤ 0.17). Similarly, we found inter-annual effects of tailwind and sidewind, depending on the type of migrant (long- and short-distance), but the models exhibited high variation and low explanatory power (marginal R2 ≤ 0.18). Moreover, the lagged effect of weather variables does not improve the variance explained by weather variables at any of the two timescales evaluated. The low association between migrant counts and weather calls for future research at larger spatial and temporal scales, as well as on assessing the cumulative effect of weather along raptor migration routes.