Mon, Aug 02, 2021:On Demand
Background/Question/Methods
Changes in timing of life history events (phenology) in response to warming temperatures have been documented in many taxa, but predicting the strength and direction of these shifts remains challenging. Ecological traits may provide a powerful means of predicting species’ responses to environmental change, but evidence of traits as predictors of phenological shifts is limited. Additionally, few long-term phenological analyses account for the possibility of changes other than in the mean timing of an event or in first and last observations (which are sensitive to population size). Here, we present a trait-based approach to analyzing patterns of long-term phenological change in solitary bees, a diverse group of insects that serve as important pollinators of both crops and wild plants. Using quantile regression, a method robust to varying sample size, we estimated changes in the onset, median, end, and duration of annual activity for 73 bee species using 160 years of museum and citizen science data. We then grouped species by nesting location (above and below ground), diet breadth (generalist and specialist), and flight season (spring, summer, and fall) to evaluate whether the magnitude and/or direction of phenological changes vary between species that differ in each of these ecological traits.
Results/Conclusions On average, we found significant advances in early and median bee activity (-0.45 and -0.41 days/decade, respectively) and no change in late activity. Overall, duration of species’ activity grew significantly longer over the study period (+0.45 days/decade). Among dietary generalists, the strength of shifts varied by nesting location: in spring, median activity advanced faster in species nesting above ground (-1.12 days/decade) than below ground (-0.53 days/decade), while the opposite pattern occurred in early activity of summer species (above-ground: 0.09 days/decade; below-ground: -0.98 days/decade). Direction of shifts differed significantly between flight seasons. Median activity of ground-nesting specialists advanced in spring and summer (-0.26 and -0.53 days/decade, respectively) but was delayed in fall (+0.42 days/decade). Similarly, late activity was delayed in fall ground-nesting specialists (+0.78 days/decade), but did not change significantly in spring or summer species. Diet breadth did not predict patterns of phenological change. Our findings demonstrate a range of phenological shifts across a diversity of bees and provide evidence of the value of traits in predicting phenological responses, inviting further investigation of the drivers of these patterns. Finally, our results underscore the importance of evaluating shifts in phenological events at multiple points in time to fully characterize patterns of change.
Results/Conclusions On average, we found significant advances in early and median bee activity (-0.45 and -0.41 days/decade, respectively) and no change in late activity. Overall, duration of species’ activity grew significantly longer over the study period (+0.45 days/decade). Among dietary generalists, the strength of shifts varied by nesting location: in spring, median activity advanced faster in species nesting above ground (-1.12 days/decade) than below ground (-0.53 days/decade), while the opposite pattern occurred in early activity of summer species (above-ground: 0.09 days/decade; below-ground: -0.98 days/decade). Direction of shifts differed significantly between flight seasons. Median activity of ground-nesting specialists advanced in spring and summer (-0.26 and -0.53 days/decade, respectively) but was delayed in fall (+0.42 days/decade). Similarly, late activity was delayed in fall ground-nesting specialists (+0.78 days/decade), but did not change significantly in spring or summer species. Diet breadth did not predict patterns of phenological change. Our findings demonstrate a range of phenological shifts across a diversity of bees and provide evidence of the value of traits in predicting phenological responses, inviting further investigation of the drivers of these patterns. Finally, our results underscore the importance of evaluating shifts in phenological events at multiple points in time to fully characterize patterns of change.