David K. Skelly
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
David K. Skelly has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- A. Z. Andis Arietta
School of the Environment Yale University New Haven Connecticut USA
5 shared publications - Mark C. Urban
Department of Ecology and Evolutionary Biology, University of Connecticut, Storrs, CT, USA
3 shared publications - Erin L. Sauer
Department of Ecology, Evolution, and Natural Resources, Rutgers University, New Brunswick, NJ, USA
2 shared publications - Logan S. Billet
School of the Environment, Yale University, New Haven, CT, USA
2 shared publications - Adriana D. Rubinstein
School of the Environment Yale University New Haven Connecticut USA
1 shared publication Adriana Rubinstein
Yale Univ., School of the Environment CT USA
1 shared publication
Latest Publications
- Open Access
- 2 November 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (45)
Thermal ecology drives alternate responses to rising temperatures in Caribbean anoles
Rapidly rising temperatures threaten populations of ectothermic animals worldwide, with the greatest impacts predicted in the tropics—hotspots for biodiversity. Ectotherms can respond to warming through two primary compensatory responses: behavioral thermoregulation, which buffers exposure to extreme temperatures; and heat tolerance increases via plasticity, adaptation, or both. The relative contribution of these responses is predicted to depend on ecological context, but whether these principles can anticipate organismal responses to contemporary warming remains largely untested. Here, using a decadal resurvey of anole lizards from Hispaniola, Jamaica, and Puerto Rico, we show that simple principles of thermal ecology can reliably anticipate organismal responses to rising temperatures. Species occupying thermally heterogeneous open/edge habitats, such as dry forests, behaviorally thermoregulate and have shown little to no increase in heat tolerance over the past decade. By contrast, species from thermally homogeneous closed-canopy environments are poor thermoregulators and have increased their heat tolerance by 2 °C, on average, over the same period. Seasonal comparisons, acclimation experiments, and laboratory-rearing experiments indicate that contemporary increases in heat tolerance likely reflect contributions from rapid evolutionary change and physiological plasticity. Thermal ecology, not phylogenetic or spatial proximity, is the key factor driving responses to rising temperatures. Our results consistently link organismal responses to habitat complexity and thermoregulatory behavior; we suggest that these factors are key to predicting how ectotherms respond to climate warming.
- 2 November 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (45)
Open AccessThermal ecology drives alternate responses to rising temperatures in Caribbean anoles
Rapidly rising temperatures threaten populations of ectothermic animals worldwide, with the greatest impacts predicted in the tropics—hotspots for biodiversity. Ectotherms can respond to warming through two primary compensatory responses: behavioral thermoregulation, which buffers exposure to extreme temperatures; and heat tolerance increases via plasticity, adaptation, or both. The relative contribution of these responses is predicted to depend on ecological context, but whether these principles can anticipate organismal responses to contemporary warming remains largely untested. Here, using a decadal resurvey of anole lizards from Hispaniola, Jamaica, and Puerto Rico, we show that simple principles of thermal ecology can reliably anticipate organismal responses to rising temperatures. Species occupying thermally heterogeneous open/edge habitats, such as dry forests, behaviorally thermoregulate and have shown little to no increase in heat tolerance over the past decade. By contrast, species from thermally homogeneous closed-canopy environments are poor thermoregulators and have increased their heat tolerance by 2 °C, on average, over the same period. Seasonal comparisons, acclimation experiments, and laboratory-rearing experiments indicate that contemporary increases in heat tolerance likely reflect contributions from rapid evolutionary change and physiological plasticity. Thermal ecology, not phylogenetic or spatial proximity, is the key factor driving responses to rising temperatures. Our results consistently link organismal responses to habitat complexity and thermoregulatory behavior; we suggest that these factors are key to predicting how ectotherms respond to climate warming.
- Open Access
A shared pathogen reservoir can tip widespread infection into mass mortality
Pathogens that persist as low-intensity infections across many wildlife populations can drive mass mortality in others. Mass mortality is often abrupt, and the timing can be difficult to predict from host traits or environmental conditions alone. In a recent field study tracking ranavirus epizootics in wood frog ( Rana sylvatica ) breeding ponds, we found that environmental conditions and host features did not reliably distinguish die-off from non-die-off pond-years, whereas within-season trajectories of infection prevalence and environmental viral concentration did. Infection prevalence predicted subsequent environmental viral concentration throughout epizootics, whereas environmental viral concentration predicted subsequent infection prevalence and intensification primarily near die-off onset. This temporal asymmetry is consistent with a potential feedback operating through the shared viral environment. Here, we develop a compartmental model linking waterborne viral concentration to the rate at which low-intensity infections progress to high-intensity, high-shedding states within already-infected hosts. We show that a dose-dependent progression model generates the two-phase epizootic trajectory observed in natural die-offs: prolonged low-intensity circulation followed by a rapid increase in high-intensity infection and mortality as environmental viral concentration rises. The model exhibits a sharp mortality threshold between low-intensity circulation and mass mortality, governed mainly by the high-to-low intensity shedding ratio, host density, and pond volume. Existing explanations for die-off variation emphasize individual-level susceptibility, but our model demonstrates that dose-dependent environmental feedback acting on established infections can generate the transition from low-intensity infection to mass mortality in a homogeneous host population, even though variation in host susceptibility and environmental conditions may modify its timing and severity. This mechanism may apply in any system where hosts share a bounded environment, pathogen dose influences disease severity, and pathogen shedding increases with infection intensity.
Open AccessA shared pathogen reservoir can tip widespread infection into mass mortality
Pathogens that persist as low-intensity infections across many wildlife populations can drive mass mortality in others. Mass mortality is often abrupt, and the timing can be difficult to predict from host traits or environmental conditions alone. In a recent field study tracking ranavirus epizootics in wood frog ( Rana sylvatica ) breeding ponds, we found that environmental conditions and host features did not reliably distinguish die-off from non-die-off pond-years, whereas within-season trajectories of infection prevalence and environmental viral concentration did. Infection prevalence predicted subsequent environmental viral concentration throughout epizootics, whereas environmental viral concentration predicted subsequent infection prevalence and intensification primarily near die-off onset. This temporal asymmetry is consistent with a potential feedback operating through the shared viral environment. Here, we develop a compartmental model linking waterborne viral concentration to the rate at which low-intensity infections progress to high-intensity, high-shedding states within already-infected hosts. We show that a dose-dependent progression model generates the two-phase epizootic trajectory observed in natural die-offs: prolonged low-intensity circulation followed by a rapid increase in high-intensity infection and mortality as environmental viral concentration rises. The model exhibits a sharp mortality threshold between low-intensity circulation and mass mortality, governed mainly by the high-to-low intensity shedding ratio, host density, and pond volume. Existing explanations for die-off variation emphasize individual-level susceptibility, but our model demonstrates that dose-dependent environmental feedback acting on established infections can generate the transition from low-intensity infection to mass mortality in a homogeneous host population, even though variation in host susceptibility and environmental conditions may modify its timing and severity. This mechanism may apply in any system where hosts share a bounded environment, pathogen dose influences disease severity, and pathogen shedding increases with infection intensity.
- Open Access
Epizootic tipping points: Environmental viral feedbacks predict amphibian die-offs
Virulent pathogens often circulate in wildlife populations without causing mass mortality, but what distinguishes outbreaks that escalate to die-offs remains poorly understood. We tracked ranavirus epizootics across 40 wood frog breeding ponds over three years, where infection was widespread but die-offs occurred in only 13 of 102 pond-years. Models based on viral state received greater support than abiotic and host conditions for subsequent infection prevalence, viral load, and environmental viral concentration (eDNA). Infection prevalence predicted later eDNA concentration throughout epizootics, consistent with continued shedding from infected hosts. In contrast, relationships between eDNA concentration and subsequent infection prevalence and viral load were strongly attenuated when die-off-onset visits were excluded. Dynamic viral trajectories distinguished die-off from non-die-off pond-years more clearly than static pond and host features. Our results point toward a phase-dependent feedback in which pathogen accumulation in the shared environment contributes to the escalation from widespread infection to mass mortality.
Open AccessEpizootic tipping points: Environmental viral feedbacks predict amphibian die-offs
Virulent pathogens often circulate in wildlife populations without causing mass mortality, but what distinguishes outbreaks that escalate to die-offs remains poorly understood. We tracked ranavirus epizootics across 40 wood frog breeding ponds over three years, where infection was widespread but die-offs occurred in only 13 of 102 pond-years. Models based on viral state received greater support than abiotic and host conditions for subsequent infection prevalence, viral load, and environmental viral concentration (eDNA). Infection prevalence predicted later eDNA concentration throughout epizootics, consistent with continued shedding from infected hosts. In contrast, relationships between eDNA concentration and subsequent infection prevalence and viral load were strongly attenuated when die-off-onset visits were excluded. Dynamic viral trajectories distinguished die-off from non-die-off pond-years more clearly than static pond and host features. Our results point toward a phase-dependent feedback in which pathogen accumulation in the shared environment contributes to the escalation from widespread infection to mass mortality.
- Open Access
Prey metabolic responses to predators depend on predator hunting mode and prey antipredator defenses
Predation risk causes prey to react in numerous ways, from life history changes to shifts in habitat. These responses give some insight into how different types of predators affect prey, but prey responses are often not comparable across taxa and experimental systems. Metabolism is a ubiquitous trait among living organisms, which offers a way to examine predator effects and create generalizable outcomes. We examined metabolic responses of terrestrial and aquatic prey from three antipredator functional groups under varied contexts of predator cues. We found that constitutively defended prey did not exhibit metabolic responses to any type of predator cues, while deimatic and freeze‐flight prey exhibited metabolic responses that were dependent on both the cue type and predator hunting mode. Consistent with previous studies on nonconsumptive effects, we also found that ambush predators elicited a metabolic response while active predators did not. We propose that future work on this topic should continue to take a metabolic approach as a unified, scalable response variable to the sensory ecology of nonconsumptive effects, while identifying additional systems that can evaluate more fully both predator hunting mode and prey antipredator defenses.
Open AccessPrey metabolic responses to predators depend on predator hunting mode and prey antipredator defenses
Predation risk causes prey to react in numerous ways, from life history changes to shifts in habitat. These responses give some insight into how different types of predators affect prey, but prey responses are often not comparable across taxa and experimental systems. Metabolism is a ubiquitous trait among living organisms, which offers a way to examine predator effects and create generalizable outcomes. We examined metabolic responses of terrestrial and aquatic prey from three antipredator functional groups under varied contexts of predator cues. We found that constitutively defended prey did not exhibit metabolic responses to any type of predator cues, while deimatic and freeze‐flight prey exhibited metabolic responses that were dependent on both the cue type and predator hunting mode. Consistent with previous studies on nonconsumptive effects, we also found that ambush predators elicited a metabolic response while active predators did not. We propose that future work on this topic should continue to take a metabolic approach as a unified, scalable response variable to the sensory ecology of nonconsumptive effects, while identifying additional systems that can evaluate more fully both predator hunting mode and prey antipredator defenses.
Asynchrony, density dependence, and persistence in an amphibian
Understanding drivers of metapopulation dynamics remains a critical challenge for ecology and conservation. In particular, the degree of synchrony in metapopulation dynamics determines how resilient a metapopulation is to a widespread disturbance. In this study, we used 21 years of egg mass count data across 64 nonpermanent freshwater ponds in Connecticut, USA to evaluate patterns of abundance and growth and to assess regional as well as local factors in shaping the population dynamics of wood frogs (Rana sylvatica = Lithobates sylvaticus). In particular, we asked whether a species known to undergo metapopulation dynamics exhibited spatial synchrony in abundances. With the exception of a single year when breeding took place during severe drought conditions, our analyses revealed no evidence of synchrony despite close proximity (mean minimum distance < 300 m) of breeding ponds across the 3213‐ha study area. Instead, local, pond‐scale conditions best predicted patterns of abundance and population growth rate. We found negative density dependence on population growth rate within ponds as well as evidence that larger neighboring pond populations had a negative effect on focal ponds. Beyond density, pond depth was a critical predictor; deeper ponds supported larger populations. Drought conditions and warm winters negatively affected populations. Overall, breeding ponds vary in critical ways that either support larger, more persistent populations or smaller populations that are not represented by breeding pairs in some years. The infrequency of spatial synchrony in this system is surprising and suggests greater resilience to stressors than would have been expected if dynamics were strongly synchronized. More generally, understanding the characteristics of systems that determine synchronous population dynamics will be critical to predicting which species are more or less resilient to widespread disturbances like land conversion or climate change.
Asynchrony, density dependence, and persistence in an amphibian
Understanding drivers of metapopulation dynamics remains a critical challenge for ecology and conservation. In particular, the degree of synchrony in metapopulation dynamics determines how resilient a metapopulation is to a widespread disturbance. In this study, we used 21 years of egg mass count data across 64 nonpermanent freshwater ponds in Connecticut, USA to evaluate patterns of abundance and growth and to assess regional as well as local factors in shaping the population dynamics of wood frogs (Rana sylvatica = Lithobates sylvaticus). In particular, we asked whether a species known to undergo metapopulation dynamics exhibited spatial synchrony in abundances. With the exception of a single year when breeding took place during severe drought conditions, our analyses revealed no evidence of synchrony despite close proximity (mean minimum distance < 300 m) of breeding ponds across the 3213‐ha study area. Instead, local, pond‐scale conditions best predicted patterns of abundance and population growth rate. We found negative density dependence on population growth rate within ponds as well as evidence that larger neighboring pond populations had a negative effect on focal ponds. Beyond density, pond depth was a critical predictor; deeper ponds supported larger populations. Drought conditions and warm winters negatively affected populations. Overall, breeding ponds vary in critical ways that either support larger, more persistent populations or smaller populations that are not represented by breeding pairs in some years. The infrequency of spatial synchrony in this system is surprising and suggests greater resilience to stressors than would have been expected if dynamics were strongly synchronized. More generally, understanding the characteristics of systems that determine synchronous population dynamics will be critical to predicting which species are more or less resilient to widespread disturbances like land conversion or climate change.
- 14 September 2021
- Published by Oxford University Press (OUP) in Evolution
- Vol. 75 (11) , 2930-2943
Rapid microgeographic evolution in response to climate change
Environmental change is predicted to accelerate into the future and will exert strong selection pressure on biota. While many species may be fated to extinction, others may survive through their capacity to evolve rapidly at highly localized (i.e. microgeographic) scales. Yet, even as new examples have been discovered, the limits to such evolutionary responses have not often been evaluated. One of the first examples of microgeographic variation involved pond populations of wood frogs (Rana sylvatica). Although separated by just tens to hundreds of meters, these populations exhibited countergradient variation in intrinsic embryonic development rates when reared in a common garden. We repeated this experiment 17 years (approx. 6–9 generations) later and found that microgeographic variation persists in contemporary populations. Furthermore, we found that contemporary embryos have evolved to develop 14% to 19% faster than those in 2001. Structural equation models indicate that the predominant cause for this response is likely due to changes in climate over the intervening 17 years. Despite potential for rapid and fine-scale evolution, demographic declines in populations experiencing the greatest changes in climate and habitat imply a limit to the species’ ability to mitigate extreme environmental change. This article is protected by copyright. All rights reserved
- 14 September 2021
- Published by Oxford University Press (OUP) in Evolution
- Vol. 75 (11) , 2930-2943
Rapid microgeographic evolution in response to climate change
Environmental change is predicted to accelerate into the future and will exert strong selection pressure on biota. While many species may be fated to extinction, others may survive through their capacity to evolve rapidly at highly localized (i.e. microgeographic) scales. Yet, even as new examples have been discovered, the limits to such evolutionary responses have not often been evaluated. One of the first examples of microgeographic variation involved pond populations of wood frogs (Rana sylvatica). Although separated by just tens to hundreds of meters, these populations exhibited countergradient variation in intrinsic embryonic development rates when reared in a common garden. We repeated this experiment 17 years (approx. 6–9 generations) later and found that microgeographic variation persists in contemporary populations. Furthermore, we found that contemporary embryos have evolved to develop 14% to 19% faster than those in 2001. Structural equation models indicate that the predominant cause for this response is likely due to changes in climate over the intervening 17 years. Despite potential for rapid and fine-scale evolution, demographic declines in populations experiencing the greatest changes in climate and habitat imply a limit to the species’ ability to mitigate extreme environmental change. This article is protected by copyright. All rights reserved
- Open Access
- 7 September 2021
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 14 (2) , 431-443
Estimating species relative abundances from museum records
Dated, geo‐referenced museum specimens are a rich data source for reconstructing species' distribution and abundance patterns. However, museum records are potentially biased towards over‐representation of rare species, and it is unclear whether museum records can be used to estimate relative abundance in the field. We assembled 17 coupled field and museum datasets to quantitatively compare relative abundance estimates with the Dirichlet distribution. Collectively, these datasets comprise 73,039 museum records and 1,405,316 field observations of 2,240 species. Although museum records of rare species overestimated relative abundance by 1‐fold to over 100‐fold (median study = 9.0), the relative abundance of species estimated from museum occurrence records was strongly correlated with relative abundance estimated from standardized field surveys (r2 range of 0.10–0.91, median study = 0.43). These analyses provide a justification for estimating species relative abundance with carefully curated museum occurrence records, which may allow for the detection of temporal or spatial shifts in the rank ordering of common and rare species.
- 7 September 2021
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 14 (2) , 431-443
Open AccessEstimating species relative abundances from museum records
Dated, geo‐referenced museum specimens are a rich data source for reconstructing species' distribution and abundance patterns. However, museum records are potentially biased towards over‐representation of rare species, and it is unclear whether museum records can be used to estimate relative abundance in the field. We assembled 17 coupled field and museum datasets to quantitatively compare relative abundance estimates with the Dirichlet distribution. Collectively, these datasets comprise 73,039 museum records and 1,405,316 field observations of 2,240 species. Although museum records of rare species overestimated relative abundance by 1‐fold to over 100‐fold (median study = 9.0), the relative abundance of species estimated from museum occurrence records was strongly correlated with relative abundance estimated from standardized field surveys (r2 range of 0.10–0.91, median study = 0.43). These analyses provide a justification for estimating species relative abundance with carefully curated museum occurrence records, which may allow for the detection of temporal or spatial shifts in the rank ordering of common and rare species.
- 11 December 2020
- Published by Wiley in Journal of Experimental Zoology Part A: Ecological and Integrative Physiology
- Vol. 335 (1) , 146-157
Temperature‐mediated trade‐off between development and performance in larval wood frogs (Rana sylvatica)
Countergradient variation has been detected in diverse taxa. In a common manifestation, individuals from colder environments develop faster than conspecifics from warmer environments when placed in a common garden. Where such a pattern exists, it implies a trade‐off: Individuals from warmer environments have intrinsic rates of development lower than those demonstrated by other individuals of the same species. We explored a trade‐off between development rate and locomotor performance in the wood frog (Rana sylvatica), an amphibian for which countergradient variation has been well documented. We reared wood frogs from 10 populations under two temperature regimes, bracketing the temperatures observed in local natural ponds. Individuals reared under warmer conditions developed more rapidly but exhibited burst speeds 20% lower than individuals reared under colder conditions. The slope of the reaction norm was consistent across the 10 populations and thus, we found no evidence of countergradient variation in performance. Burst speed assays of wild‐caught tadpoles from the same populations showed a similar but nonsignificant trend, with greater variability among ponds. Overall, our findings support the existence of a development–performance trade‐off that may be of broad importance and which may help explain the widespread occurrence of countergradient variation.
- 11 December 2020
- Published by Wiley in Journal of Experimental Zoology Part A: Ecological and Integrative Physiology
- Vol. 335 (1) , 146-157
Temperature‐mediated trade‐off between development and performance in larval wood frogs (Rana sylvatica)
Countergradient variation has been detected in diverse taxa. In a common manifestation, individuals from colder environments develop faster than conspecifics from warmer environments when placed in a common garden. Where such a pattern exists, it implies a trade‐off: Individuals from warmer environments have intrinsic rates of development lower than those demonstrated by other individuals of the same species. We explored a trade‐off between development rate and locomotor performance in the wood frog (Rana sylvatica), an amphibian for which countergradient variation has been well documented. We reared wood frogs from 10 populations under two temperature regimes, bracketing the temperatures observed in local natural ponds. Individuals reared under warmer conditions developed more rapidly but exhibited burst speeds 20% lower than individuals reared under colder conditions. The slope of the reaction norm was consistent across the 10 populations and thus, we found no evidence of countergradient variation in performance. Burst speed assays of wild‐caught tadpoles from the same populations showed a similar but nonsignificant trend, with greater variability among ponds. Overall, our findings support the existence of a development–performance trade‐off that may be of broad importance and which may help explain the widespread occurrence of countergradient variation.
- Open Access
Phenological delay despite warming in wood frog Rana sylvatica reproductive timing: a 20‐year study
Across all taxa, amphibians exhibit some of the strongest phenological shifts in response to climate change. As climates warm, amphibians and other animals are expected to breed earlier in response to temperature cues. However, if species use fixed cues such as daylight, their breeding timing might remain fixed, potentially creating disconnects between their life history and environmental conditions. Wood frogs Rana sylvatica are a cold‐adapted species that reproduce in early spring, immediately after breeding ponds are free of ice. We used long‐term surveys of wood frog oviposition timing in 64 breeding ponds over 20 yr to show that, despite experiencing a warming of 0.29°C per decade in annual temperature, wood frog breeding phenology has shifted later by 2.8 d since 2000 (1.4 d per decade; 4.8 d per °C). This counterintuitive pattern is likely the result of changes in the timing of snowpack accumulation and melting. Finally, we used relationships between climate and oviposition between 2000 and 2018 to hindcast oviposition dates from climate records to model longer‐term trends since 1980. Our study indicates that species can respond to fine‐grained seasonal climate heterogeneity within years that is not apparent or counterintuitive when related to annual trends across years.
Open AccessPhenological delay despite warming in wood frog Rana sylvatica reproductive timing: a 20‐year study
Across all taxa, amphibians exhibit some of the strongest phenological shifts in response to climate change. As climates warm, amphibians and other animals are expected to breed earlier in response to temperature cues. However, if species use fixed cues such as daylight, their breeding timing might remain fixed, potentially creating disconnects between their life history and environmental conditions. Wood frogs Rana sylvatica are a cold‐adapted species that reproduce in early spring, immediately after breeding ponds are free of ice. We used long‐term surveys of wood frog oviposition timing in 64 breeding ponds over 20 yr to show that, despite experiencing a warming of 0.29°C per decade in annual temperature, wood frog breeding phenology has shifted later by 2.8 d since 2000 (1.4 d per decade; 4.8 d per °C). This counterintuitive pattern is likely the result of changes in the timing of snowpack accumulation and melting. Finally, we used relationships between climate and oviposition between 2000 and 2018 to hindcast oviposition dates from climate records to model longer‐term trends since 1980. Our study indicates that species can respond to fine‐grained seasonal climate heterogeneity within years that is not apparent or counterintuitive when related to annual trends across years.