This blog post is provided by Joseph Burant and Ryan Norris and tells the #StoryBehindThePaper for the paper “Early warning indicators of population collapse in a seasonal environment“, which was recently published in Journal of Animal Ecology.

In this post, Joseph Burant and Ryan Norris introduce the experiment behind their recent paper in Journal of Animal Ecology, in which they and their co-authors explore the consequences of seasonality for the detection of population declines using early warning signals. Joey is a recent PhD graduate of the University of Guelph and currently a postdoc with the Living Data Project and McGill University. Ryan is an Associate Professor at the University of Guelph, where he and his research group study the behaviour, population dynamics, and conservation of animals living in seasonal environments.

“The only constant in life is change.” – Heraclitus

The environment is in a constant state of flux, and the persistence of wildlife populations necessitates that individuals respond appropriately to these shifts. Of these environmental changes, perhaps the most powerful, and certainly the most predictable, is seasonality. Depending on the context, seasonality can mean many different things. When talking about animal populations, seasonality often refers to the occurrence of distinct periods of reproduction and no reproduction within the annual cycle (i.e., breeding versus non-breeding periods). Most often, this reproductive seasonality is a product of seasonal fluctuations in resource availability and quality, which are themselves the result of cyclical shifts in abiotic conditions ultimately caused by our planet’s movement around the Sun (pretty incredible!).

The study of the dynamics of animal populations in seasonal environments has a rich history in the field of ecology (e.g., Fretwell’s classical book on the topic). Traditionally, dividing the annual cycle of a species into breeding and non-breeding components allows researchers to more closely assess the relative contributions of reproduction (breeding) and mortality (breeding and non-breeding) to overall dynamics and stability. More generally, by dividing the annual cycle in this way, we’re better able to evaluate how changes in the environment that occur in only part of the year influence not only the population dynamics in that season, but also how the effects of these changes can carry over to affect events in subsequent periods. This becomes particularly pertinent as researchers and conservationists aim to identify and address the factors that drive wild populations to decline in the era of rapid environmental change.

To better understand how populations respond to environmental change occurring in different ‘seasons’, we conducted a multi-generation experiment in which populations of common fruit flies (Drosophila melanogaster) were exposed to habitat loss in either the breeding or non-breeding period. Many will know Drosophila as the quintessential model organism in biology, with the species’ quick generation time, low maintenance costs, and relatively small genome among the many features that make them particularly appealing for a wide range of experimental work. Those familiar with the natural history of fruit flies will also know that they are not seasonal in the sense we use here. Instead, wild Drosophila are what is referred to as “multivoltine”, and multiple successive generations will reproduce within a single breeding season. These consecutive generations of Drosophila are known to exhibit shifts in life-history traits that coincide with changes in the environment over the course of the breeding season. So how did we use this serially-breeding species to understand the dynamics and declines of seasonal populations more generally?

More than a decade ago, the Norris lab was searching for a model system to examine population dynamics in seasonal environments. In addition to having a short generation time, the species needed to have distinct breeding and non-breeding periods. However, there were no suitable existing systems because they all had continuous breeding periods. Instead, Gustavo Betini (then-PhD student in the Norris lab and co-author on our new paper) proposed the idea of ‘inducing’ a non-breeding period in Drosophila. After several preliminary trials, we finally devised a set-up where a non-breeding period could be imposed by feeding flies a sugar-water solution from a single source at the top of the holding vial, thereby providing enough food for most flies to survive but not allowing females to lay eggs. It also meant we could manipulate the amount of non-breeding habitat. Since then, this experimental approach has led to many important insights into the ecology of seasonality, including how carry-over effects influence population dynamics, how life-history trade-offs between seasons create population cycles, and how habitat loss can influence the dynamics of migratory networks. 

In a recent experiment using the seasonal fruit fly system, we established replicate populations and, after allowing them to develop for several generations under constant conditions, systematically reduced the amount of food provided in either the breeding or non-breeding period for each subsequent generation until the populations went extinct. We counted the number of individuals in each population twice per generation, and measured locomotor activity and body weight for a subset of individuals. In total, more than 400,000 fruit flies moved through our populations over the course of the study.

In an earlier paper in Ecology Letters, we used our experimental data to show how the seasonal timing of habitat loss affects different patterns of population decline and, more significantly, that breeding and non-breeding habitat loss can be distinguished using simple vital rates like reproduction and non-breeding survival. In our new article available in Journal of Animal Ecology, we combine the population abundance and individual trait data from our experiment to investigate whether habitat-loss-induced population collapse is preceded by early warning indicators. Early warning signals (EWS) are generic statistical properties of time series and are predicted to arise as external forcing results in a loss of system resiliency. Previous work has applied theory on EWS in a variety of ecological contexts –and traits like body size have been used to improve the predictive capacity of EWS– but has not considered seasonality explicitly. Our results show that the utility of EWS for detecting impending declines depends on the timing of the stressor: abundance- and trait-based early warning indicators were generally poor predictors of population collapse resulting from non-breeding habitat loss.

Many ecological systems have been brought to the brink of collapse by human activities. Applied conservation efforts are often hindered by a lack of information on the timing and location of environmental stressors that are driving species declines. Addressing these significant challenges will require simple, broadly-applicable tools to help target and focus limited resources. We envision our experimental approach as one valuable way to explore and develop such tools. By imposing a bi-seasonal regime on fruit flies in the lab, we have demonstrated how simple demographic rates and early warning signals can reveal where populations are limited. We hope that the applicability and robustness of these experimentally-derived tools will be evaluated using long-term time series of species for which season declines have been postulated, and subsequently extended to generate predictions about threatened populations.

More info:

Behrman, E. L., Watson, S. S., O’Brien, K. R., Heschel, M. S., and Schmidt, P. S. (2015). Seasonal variation in life history traits in two Drosophila species. Journal of Evolutionary Biology, 28, 1691-1704. DOI: https://doi.org/10.1111/jeb/12690.

Betini, G. S., Fitzpatrick, M. J., and Norris, D. R. (2015). Experimental evidence for the effect of habitat loss on the dynamics of migratory networks. Ecology Letters, 18, 526-534. DOI: https://doi.org/10.1111/ele.12432.

Betini, G. S., Griswold, C. K., and Norris, D. R. (2013). Carry-over effects, sequential density dependence and the dynamics of populations in a seasonal environment. Proceedings of the Royal Society B, 280, 20130110. DOI: http://dx.doi.org/10.1098/rspb.2013.0110.

Betini, G. S., McAdam, A. G., Griswold, C. K., and Norris, D. R. (2017). A fitness trade-off between seasons causes multigenerational cycles in phenotype and population size. eLife, 6, e18770. DOI: https://doi.org/10.7554/eLife.18770.001.

Burant, J. B., Betini, G. S., and Norris, D. R. (2019). Simple signals indicate which period of the annual cycle drives declines in seasonal populations. Ecology Letters, 22, 2141-2150. DOI: https://doi.org/10.1111/ele.13393.

Clements, C. F., and Ozgul, A. (2016). Including trait-based early warning signals helps predict population collapse. Nature Communication, 7, 10984. DOI: https://doi.org/10.1038/ncomms10984.

Fretwell, S. D. (1972). Populations in a Seasonal Environment. Monographs in Population Biology, vol. 5. Princeton University Press, Princeton, NJ. ISBN: 9780691081069.