Mechanistic approaches to modeling wildlife response to climate change


Organizers: Marta P. Lyons, Midwest Climate Adaptation Science Center; Owen P. Mckenna, USGS Northern Prairie Research Center;  Olivia E. LeDee,U.S. Geological Survey

Global change will lead to novel environments, non-analogous to current or recent climates, effective conservation planning will require robust modeling of wildlife responses to these future novel climatic conditions. Population, habitat, and distribution models that explicitly incorporate the mechanisms or processes underlying responses to environmental conditions, rather than relying only on associations between observed patterns, offer a flexible tractable framework for anticipating wildlife response to climate change and management scenarios. This symposium will highlight recent work using mechanistic approaches to model population and habitat responses to climate change. Speakers will address how mechanistic approaches can be utilized to gain important insights at multiple organizational levels on how wildlife will respond to climate change and opportunities of climate adaptation.

Mechanistic Niche Models Predict Less Climate Change Induced Range Loss by Incorporating Temporal Shifts in Foraging Activity
Marta Lyons, Kenneth Kozak
Niche models, also known as species distribution models, are a valuable tool in forecasting how species will respond to climate change. However, future distribution predictions are highly sensitive to choices in modeling approach, species data, and climate data. Correlative niche models predict dire futures for mountaintop salamanders in the Southern Appalachians. These models’ inability to extrapolate to non-analogous climate or account for behavioral plasticity may lead to overpredictions of distribution loss. We used a mechanistic niche model that directly incorporates species-specific physiology, morphology, and behavior to predict current and future annual available energy on the landscape. This mechanistic modeling approach not only predicted less range loss than correlative niche models, but also allowed us to analyze the potential for temporal shifts in the time of year that the salamanders were expected to forage. Future foraging windows and suitable area varied depending on choices in global climate models, stressing the need to include multiple global climate models when modeling species distributions.
Climate Change Effects on Wetland Waterfowl Habitat in the Prairie Pothole Region
Owen McKenna
The Prairie Pothole Region (PPR) is recognized as one of the most productive areas for waterfowl in North America and is used by an estimated 50–80 % of the continent’s breeding duck population. The ongoing acquisition program of the U.S. Fish and Wildlife Service National Wildlife Refuge System has conserved approximately 1.3 million hectares of critical breeding-waterfowl habitat. A major assumption inherent to the current conservation approach is that past distributions of waterfowl habitat and populations are relatively representative of future distributions. Our goal with this interagency collaboration is to co-produce useable information to better plan for future impacts of climate change on the wetland habitat for breeding waterfowl pairs in the U.S. Prairie Pothole Region. We are using a mechanistic hydrology model in combination with multi-decadal monitoring data and predictive breeding waterfowl pair statistical models to simulate wetland-waterfowl responses under different climate futures.
Managing and Promoting the Resiliency of Winter-Adapted Species to Climate Change
Benjamin Zuckerberg, Jonathan Pauli
For many organisms, winter is a period of resource scarcity and energy deficits. Global climate change, arguably one of the most pressing anthropogenic forces influencing wildlife populations and communities, is rapidly altering winter conditions throughout the Northern Hemisphere. Adaptation to climate change requires understanding species vulnerabilities in terms of demographic sensitivity and exposure to climatic and other environmental factors across full life cycles and in diverse landscapes. As part of a long-term research study, we summarize recent findings from a pair of winter-adapted species, snowshoe hare and ruffed grouse, to highlight how warming and more variable winters may effect wildlife. We discovered how a warming winter climate and more variable snow cover can impacts a host of ecological responses from stress to survival to population cycling. Further, we will demonstrate how spatially-explicit demographic modeling and habitat management can identify how winter climate stressors can influence population dynamics both historically and into the future. Both winter temperature and precipitation extremes are critically important for considering in the full life cycle of climate-vulnerable vulnerable species. Mechanistic approaches to species distribution modeling provides a framework for linking demographic and distributional dynamics to climate change, and can provide unique information for conservation planning.
Forecasting Population Responses of Monarch Butterflies to Climate Change
Erin Zylstra, Elise Zipkin
Butterflies are declining, and there is growing evidence that climate change may be one of the primary drivers of changes in population sizes. For example, dynamics of the eastern migratory population of monarch butterflies (Danaus plexippus) – a population that has declined substantial in the last quarter century – have been largely driven by temperature and precipitation on the spring and summer breeding grounds over the last 15 years. To assess how monarchs may respond to future changes in climate, we forecasted population sizes under a range of climate scenarios based on a full-annual-cycle population model. Specifically, we relate counts of butterflies on the summer breeding grounds, and subsequent overwintering population size, to weather conditions on the spring breeding grounds in eastern Texas and on the summer breeding grounds in the midwestern U.S. and southern Ontario. We then project future climate on the spring and summer breeding grounds for early (2023-2043), mid (2050-2070), and late (2080-2100) 21st century based on several global circulation models under four emission scenarios. During the early- and mid-century, mean values of forecasted overwintering population sizes were similar to the mean population size between 2004-2018. However, the probability that the annual overwintering population size will be smaller than the past minimum size (e.g., as documented between 1994-2020) at least once prior to 2070 was >0.4. By the end of the century, forecasted population sizes under uncurbed emission scenarios were well below those observed between 2004-2018. The distribution of monarchs across the summer breeding grounds is also likely to shift, with higher forecasted counts in areas where climate models project increases in precipitation and little change in temperature. Our fine-scale population forecasts can be used to prioritize areas for conservation and habitat restoration activities.   
Model-Predicted Forage for Bison in Northern Great Plains National Parks in Plausible Future Climate Scenarios
Amy Symstad
The contiguous Great Plains landscape has been permanently impacted by Euro-American colonization and development, and the vast bison herds and the unique people who depended on them were nearly extirpated. National parks in the Northern Great Plains preserve remnants of this landscape, as well as nearly wild bison populations, thereby helping to maintain the influence of bison on natural and cultural landscapes in this region. Regional and national strategies for maintaining or improving the viability of bison as a wildlife species in the limited areas to which they are confined require understanding the population sizes each bison refuge can support. Climate change challenges the maintenance of these confined bison populations not because of direct physiological impacts on these widely adapted animals, but through its effects on the vegetation they depend on for forage. This talk will describe how two very different mechanistic models were used to project plausible future forage production outcomes for two national parks in South Dakota. MC1, an existing dynamic global vegetation model that simulates vegetation distribution, biogeochemical cycling, and fire in a highly interactive manner, was parameterized to represent conditions at Wind Cave National Park, then used to simulate the effects of three climate-CO2 projections, coupled with a variety of grazing-pressure and fire-frequency scenarios, on the park’s overall grassland productivity. For Badlands National Park, a landscape-scale state-and-transition simulation model (STSM) was built from available ecological site descriptions and expert knowledge using ST-sim software, then used to simulate grassland composition and production under four future climates, coupled with four bison-fire-invasive species management alternatives. The two modeling approaches provide different types of information, with different types of uncertainties, but managers at each park found both outcomes valuable for future bison management planning.
Incorporating Physiological Diversity into Mechanistic Models Reveals New Estimates of Winter Tick Impacts on Moose
Kennan Oyen, Troy Koser, Thomas Arya, Krishna Woerheide, Janet Rachlow, Weyand Logan, Jacob DeBow, Henry Jones, Kevin Monteith, Rebecca Levine, Samantha Dwinnell, Kent Hersey, Pia Olafson, Randy Ryan, Richard Davis, Neil Pople, Agnieszka Gigiel, Gerald S
Moose are declining in several regions of North America and particularly at their southern range limits. Although a myriad of factors likely contribute to these declines, unprecedented epizootics of Dermacentor albipictus (winter ticks) are strongly correlated with high juvenile mortality, particularly in the Northeastern United States. Winter ticks are single-host ticks that parasitize many ungulates but are particularly detrimental to moose, causing alopecia, chronic anemia, and emaciation that often leads to mortality, particularly around 10 months old. The frequency and severity of winter tick epizootics vary throughout North American moose ranges, and the factors that drive these events, including environmental change, tick physiology, and host dynamics remain unclear. We compared physiological metrics between winter ticks collected from sites where epizootics are frequent and uncommon. Preliminary data suggest that winter ticks from different geographic regions have variable physiological tolerances. Ticks from Wyoming, USA had over 40% higher survivorship following acute cold exposures and survived chronic low temperature and humidity conditions for approximately 4 months longer than those from Maine, USA. These results suggest that ecological niche models based on species tolerances from only one region may be strongly biased and not capture full adaptive capacity of this species. Using an alternative approach combining integral projection models (IPMs) and matrix population models, we test whether differences in winter tick physiological thresholds might explain tick population dynamics and the frequency and distributions of epizootic events. Integral projection models can be applied to systems where several attributes vary across life stages that influence demography and there is substantial individual variability within a stage class which drive population dynamics. Winter ticks exhibit substantial variability in physiological traits which may explain differences in regional tick population dynamics.
Mechanistic Modeling for Amphibians in the US Caribbean
Jaime Collazo, Ana Rivera, Rafael Chaparro, Eloy Martinez, Adam Terando, Mitchel Eaton
Global warming will likely impinge on important biological processes of amphibians that include gas exchange, osmoregulation, escape and avoidance, and reproduction. Understanding the mechanisms that influence these physiological processes provide a stronger foundation to test hypotheses about species adaptive capacity and vulnerability, and predict where and when projected environmental conditions will affect species to implement appropriate adaptation strategies. We determined the critical thermal maximum (CTMax) of seven species of Eleutherodactylus coqui frogs in Puerto Rico. CTMax is defined as the temperature that triggered spasms or erratic behavior, impairing predator avoidance. Coqui species with narrower distribution ranges, such as E. wightmanae, had significantly lower CTMax values (36.30 ± 0.52 °C) than widely distributed species like E. coqui (38.87 ± 0.57 °C). E. coqui collected at 150 m had significantly lower CTMax and narrower tolerance range (35.87 ± 0.32 °C) than those collected at higher elevation (526 m; 38.87 ± 0.57 °C), raising questions about acclimation potential. Model projections indicate that ambient temperature in Puerto Rico could increase by 6-8 °C by 2040-2070. At present, diel average temperature (T) and relative humidity (RH) at high elevation sites (>500m; Maricao) are suitable for E. wightmanae and E. coqui. Average RH ≥ 80% and T ≤ 26 °C foster high probability (>0.6) of occupancy, abundance, and reproductive activity. At a site representing future conditions in 2040-2070 ( °C (T) for about 6-7 hours, all during daylight. Our results point at two areas of research: 1) demographic consequences of shrinking activity windows, and 2) quantifying the physiological performance of Eleutherodactylus species under sub-lethal conditions. Performance profiles would be invaluable to predict potential impacts on population viability, help identify climate refugia and the appropriate time for managed translocations.
Projecting Habitat Responses to Climate Change and Management with Coupled Ecological Models
Brian Miller
Managing wildlife and their habitat is fraught with uncertainties around how complex, dynamic ecosystems will respond to management actions and a changing climate. A range of scientific tools are available to help grapple with different aspects of this challenge, but they are often used independently. Here, I report on recent software and methodological advancements that have made it possible to couple several commonly-used tools and thereby take advantage of their complementary strengths, and I illustrate the added value of combined approaches through case study examples. Agent-based models (ABMs) can simulate a variety of agents (i.e., autonomous units, such as wildlife, people, or viruses); agent characteristics, decision-making, adaptive behavior, and mobility; and interactions between agents and their environment. However, ABMs lack statistically robust techniques for relating an organism’s distribution to climate data. Species distribution models (SDMs), on the other hand, estimate changes in habitat suitability based on observed presence and absence locations of a species and environmental and climatic covariates; notably, SDMs do not account for other relevant processes such as disturbance or management actions. State-and-transition simulation models (STSMs) are adept at representing a variety of ecological dynamics (e.g., dispersal, disturbance, succession) and management actions (e.g., invasive species detection and treatment, prescribed fire), and can track landscape attributes (e.g., biomass, carbon) and management costs. STSMs have the added benefits of being able integrate data streams from multiple tools – they can incorporate spatially-explicit SDM output and dynamically link to ABMs – and thereby evaluate the combined effects of different climate and management scenarios. Together, these tools can generate spatial and temporal patterns that better reflect interactions between wildlife and their environment, facilitating the creation of more realistic and management-relevant projections.
Modeling Hydrology of Vernal Pools in Response to Climate Change and the Impact on Amphibian Breeding Habitat
Jennifer Cartwright, Toni Lyn Morelli, Evan Grant
Vernal pools are seasonal wetlands that provide important breeding habitat to amphibians, including several species of conservation concern. Climate change effects, such as shifts in the timing and magnitude of evapotranspiration, precipitation, and drought intensity, may cause some vernal pools to become dry earlier in the year, potentially interfering with amphibian life cycle completion. However, some pools may continue to provide wetland habitat later into the year under relatively dry conditions, functioning as climate-change refugia and allowing species to persist even as summer conditions become warmer and droughts more frequent. This talk will discuss recent efforts using roughly 3,000 field observations of inundation from 450 pools from West Virginia to Maine to train machine-learning models to predict pool inundation likelihood based on day of the year, climate conditions, short-term weather patterns, pool size, and soil, geologic, and landcover attributes surrounding pools. Predictions of pool wetness were generated across five seasonal time points (from May 15 to July 15), three short-term weather scenarios (dry, wet, and average), and four sets of downscaled climate projections (2050s and 2080s under Representative Concentration Pathways 4.5 and 8.5). This process was replicated for four inundation thresholds, i.e., different definitions of amounts of water needed to provide wetland habitat. All model outputs are available through a web-enabled interface allowing users to choose the inundation thresholds, time points, weather scenarios, and future climate projections most relevant to their management needs. For species with strict habitat requirements related to hydrology and seasonal timing, approaches such as this may be helpful to anticipate—and perhaps mitigate—climate impacts to vulnerable populations.

Location: Virtual Date: November 3, 2021 Time: 2:00 pm - 3:00 pm