Phenotypic plasticity alone cannot explain climate-induced change in avian migration timing

Abstract

Recent climate change has been linked to shifts in the timing of life-cycle events in many organisms, but there is debate over the degree to which phenological changes are caused by evolved genetic responses of populations or by phenotypic plasticity of individuals. We estimated plasticity of spring arrival date in 27 species of bird that breed in the vicinity of an observatory in eastern North America. For 2441 individuals detected in multiple years, arrival occurred earlier during warm years, especially in species that migrate short distances. Phenotypic plasticity averaged -0.93 days °C(-1) ± 0.70 (95% CI). However, plasticity accounted for only 13-25% of the climate-induced trend in phenology observed over 46 years. Although our approach probably underestimates the full scope of plasticity, the data suggest that part of the response to environmental change has been caused by microevolution. The estimated evolutionary rates are plausible (0.016 haldanes).

Keywords: Bird migration; evolution; phenology; phenotypic plasticity; temperature.

Figure 1

Temperature-induced plasticity in spring arrival date of birds at Powdermill Nature Reserve in western Pennsylvania, USA. Values are estimates of plasticity (±1 SE) from a mixed-effects linear model regressing arrival date against temperature. Temperature is averaged over a large region extending 1200 km to the south of the study area. Sample sizes are the number of species. Individuals of species that overwinter in North America reacted more strongly to warm years, as reflected in the interaction between temperature and migration distance (Table 1).

Figure 2

Rate of change in the date of spring arrival for 27 bird species at Powdermill Nature Reserve between 1961 and 2006 (filled square), and the rate of change expected if the entire response arose from individual-level plasticity induced by warming temperatures (open circles). Plasticity was estimated from a single hierarchical mixed-effect model conducted on the entire dataset (“pooled”), and from separate models for each species (“separate”). Temperature was averaged over southeastern North America. Error bars represent ± 95% CI.