Computer vision enabled approaches to predicting responses to fluctuating temperatures in aquatic embryos

Abstract

Fluctuations in temperature are prevalent within environments inhabited by many ectotherms, and climate change continues to rapidly alter the thermal regimes of these environments. Early developmental stages are often the most thermally sensitive life stage, and thus can drive a species’ vulnerability to climate change. Our current understanding of the effects of fluctuating temperatures (FTs) is predominately based upon comparisons of responses to constant temperatures (CTs). Relating CT responses to those of FTs is typically achieved through the use of thermal performance curves, which are integrated over time to estimate cumulative performance and thus a developmental endpoint; an approach often termed rate summation. This approach remains widely used in the empirical and theoretical literature, despite concerns over its implicit assumptions, such as that the physiological effects of chronic and acute exposure are equivalent. A key limitation in prior empirical studies assessing rate summation is in the assessment of single endpoints, and not in the responses that lead to those endpoints, thereby concealing the drivers behind discrepancies between predicted and observed responses. Consequently, in Chapters 2, 3 and 4, I developed novel computer vision tools for measuring physiological responses with a granular temporal resolution in developing animals. These tools enabled me to assess how embryos of the great pond snail, Lymnaea stagnalis, develop differently under constant and diurnally fluctuating temperatures, and the extent to which TPCs based on performance in CTs can describe embryonic responses to FTs, not just for single endpoints but throughout embryonic development. This thesis demonstrated that 1) embryonic thermal performance does not remain consistent throughout development in response to chronic exposure, but exhibits significant reductions from early to late development (Chapter 4), 2) that chronic and acute exposure do not have the same physiological effects, evidenced by marked differences in growth under chronic and acute exposure at the same temperatures (Chapter 5), 3) developmental pathways vary based on prior thermal conditions, as shown by differing growth trajectories under constant versus fluctuating temperatures (Chapter 5), and 4) that rate summation poorly predicts developmental endpoints in embryos developing under diurnally fluctuating temperatures, but it may have application to specific developmental periods given the accurate predictions observed for early embryonic periods in this thesis (Chapter 5). Together, these findings demonstrate that responses to constant temperatures are not good predictors of how animals develop under diurnally fluctuating temperatures. Further empirical data over a wider range of thermal conditions will enable greater understanding of the drivers behind responses to fluctuating temperatures, and combining digital imaging with automated bioimage analyses would be an effective approach to accomplish this objective.