Much of our current understanding of what determines the timing of cyclical seasonal events (i.e. phenology), such as plant flowering or adult insect emergence, boils down, one way or another, to the accumulation of heat. This is especially the case for animals that are dependent on external sources of body heat, such as insects in temperate regions of the world. As climate warms, more heat accumulates faster, shifting key seasonal events earlier across multiple taxa. At least that is the textbook story of one major impact of climate change on ecosystems. But models just focused on growing degree days (a measure of heat accumulation) are only so good, and it begs the question of what else might determine timing of biological events important for how ecosystems function, such as pollination of plants by insects?
Hidden in the literature are indications that other weather-related and climatic factors, such as unusually warm or dry conditions prior to and during insect life stages, may also be important for shaping phenology. For example, these events may temporarily limit adult insect flight, reducing energetic costs, which in turn may lead to longer overall flight durations. Extreme or unusual weather events are increasing in frequency and intensity, and because these events occur over a short time period, the timing and magnitude can be lost when averaging seasonal or annual values. This means that important drivers of phenological shifts may go unrecognized when focusing only on climatic averages. The impact of these short duration unusual events have not been examined at large scale, so how do you test something like this?
We focused on using natural history collections (NHC) data because they provide a critical resource for calculating phenological trends over multiple decades and geographic extents. In recent work, we showcased best practices in assembly, cleaning and downstream phenology modeling using NHC data. In particular, we focused on how climate and traits interact to determine the phenology of flight timing for butterflies and moths. Those data served as a starting point for us, as we began digging deeper into testing the importance of unusual weather conditions on adult insect flight phenology.
What we ultimately decided to do was to delineate a coarse set of grid cells across the Eastern USA and estimate yearly flight onset, offset and duration within those grids for species where we had sufficient data. Next, we calculated average annual temperature and precipitation and seasonality in a similar way to previous studies. However, we also calculated the number of anomalous weather days (values > 2 standard deviations from the long term mean) prior to and during the adult flight season for temperature, precipitation, and growing degree days. Our statistical analyses included three separate models, one each for flight onset, duration, and flight termination. Each model included all species of Lepidopterans as random effects, a set of associated life history traits (one or many broods per year, larval overwintering stage, nocturnal or diurnal, and flight season), and a phylogenetic tree to account for autocorrelation.
Our results were unexpected. We found that weather anomalies were more important than climate averages in driving Lepidopteran phenology. We expected there might be a signal, but this result was much stronger than we anticipated. While more work needs to be done, we expect that these results may generalize to additional insects and maybe more broadly. In particular, we found unusual warm days lead to earlier start of insect flight periods and unusual cold days slightly delay the end of flight periods. We also found that unusual cold and warm days were much more important than average temperatures for flight duration, with one unit extending duration by nearly 20 days vs. 8 days for annual temperatures.
These results emphatically suggest it is critical to consider unusual or anomalous weather events when understanding shorter or longer-term trends in phenology responses. Focusing on average warming alone is insufficient to understand phenological change, or the potential for mismatch within or across trophic levels. Going forward, we would like to further investigate the order or cadence of anomalous weather events and how this affects different aspects of insect phenology. For example, does it matter whether an unusually warm event is followed by unusually high precipitation, or what are the implications for an anomalously wet spring followed by an unusually cold summer? Now that we understand that anomalous weather events are important drivers of insect phenology, we can continue to unlock key information to predict how insects will respond under future environmental change. Given that anomalous weather events are thought to be increasing in frequency due to environmental change, our work highlights the pressing need to incorporate these events into ecological forecasting.