Integrating decision analysis and quantitative ecology to support wildlife management


Organizers: Marjorie Liberati, Michigan State University; Michael Runge, U.S. Geological Survey; Anna Tucker

Supported by: Biometrics Working Group

This symposium will highlight synergies between decision analysis and quantitative ecology—two emerging areas of importance in wildlife management. Decision analysis, such as Structured Decision Making, provides a way of systematically deconstructing complex decisions into discrete components, notably separating science-based tasks from values-based tasks. This breakdown allows decision makers and stakeholders to clearly frame the decision problem, guard against hidden biases, articulate assumptions, and contemplate trade-offs. It also allows quantitative scientists to understand what predictions are needed to support the decision analysis, and how uncertainty might be an impediment to the decision. The process of prediction, with the necessary consideration of uncertainty, provides the link between decision analysis and quantitative modeling.

Statistical analysis of historical data can provide the foundation for developing predictive models or it can resolve uncertainties though collection of empirical data (for example, in the context of adaptive management). Formal methods of expert judgment can be used to generate prior distributions in a Bayesian context, both for initial forecasting and for assessment of empirical observations. The relevance of quantitative ecological science to a decision maker is enhanced when decision analysis and biometrics methods are developed in a collaborative fashion.

This symposium will highlight these connections by using paired presentations. The first presentation in a pair will describe a project’s decision framing and the second will highlight the analytical research that grew from the decision-making process. .

Integrating Decision Analysis and Quantitative Ecology to Support Wildlife Management
Marjorie Liberati
Species Status Assessments to Support Listing Decisions for Cryptic Creatures: Focusing on the Four R’s – Resiliency, Redundancy, Representation, and Risk.
Melanie Olds, Brian Crawford, John Maerz, Clinton Moore
Since 2016, the U.S. Fish and Wildlife Service (USFWS) has used Species Status Assessments (SSAs) to review the status of at-risk species. The decision faced is to designate a species as “not warranted” for listing or to list it under the Endangered Species Act as “threatened” or “endangered.” This decision can be considered a risk problem and requires assessing the probability (and uncertainty) of species’ outcomes as well as the decision recommenders’ risk tolerances. The SSA framework was developed as a transparent, repeatable process to characterize species’ outcomes using principles of structured decision making that integrates best-available science with expert judgment. SSAs evaluate outcomes related to three fundamental objectives: resiliency, redundancy, and representation (the “3Rs”). Briefly, resiliency describes the condition of populations and their ability to persist over time; redundancy describes the number of populations and is related to the species’ ability to withstand large-scale catastrophic events; and representation describes the breadth of genetic diversity or ecological settings occupied by the species. We highlight a case study of an SSA we conducted in 2018 for the southern hognose snake (Heterodon simus) – a highly cryptic species in the southeastern Coastal Plain. We evaluated resiliency using population persistence probabilities, redundancy using the number and spatial distribution of populations predicted to persist, and representation using the number of populations predicted to persist in each of nine representative units (unique ecological settings). As with other SSAs, we evaluated current species conditions as well as future conditions under scenarios (i.e., alternatives) varying in levels of stressors and management. We tailored our analysis to capture and summarize results for a range of persistence probability thresholds at multiple future time points to accommodate any differences in decision recommenders’ risk tolerances. The SSA was presented to the recommenders and informed the status decision of the USFWS.
Species Status Assessments to Support Listing Decisions for Cryptic Creatures: Estimating Population Persistence Using Community Science Data
Brian Crawford, Melanie Olds, John Maerz, Clinton Moore
Estimates of demographic rates, population abundance, and risk of extinction are valuable criteria for conservation decisions. However, acquiring these estimates is challenging for most rare, cryptic species due to data limitations. The southern hognose snake (Heterodon simus) is a highly cryptic species that underwent a Species Status Assessment (SSA) in 2018 that was used for a listing decision under the Endangered Species Act. To evaluate species’ outcomes as part of the SSA, we used a comprehensive dataset of southern hognose occurrences to estimate persistence of 205 populations across the species’ range under current conditions and future scenarios representing various levels of urbanization, sea level rise, and management. We adapted a Bayesian formulation of the Cormack-Jolly-Seber model, designed to estimate survival of individuals, to analogously estimate persistence of populations between 1950 and 2018. We used estimates to simulate population persistence through 2080. We accounted for imperfect detection by developing a search effort index from occurrence records of 13 snake species commonly observed in southern hognose habitats. Estimated persistence varied considerably by population but was positively influenced by habitat suitability and proportion of protected area within a population’s boundary and was negatively related to time since last observation. Future habitat loss due to sea level rise and urbanization substantially decreased persistence probability for most populations, but predicted persistence increased with modest improvements to habitat suitability and land protection. In addition to being used in the SSA and informing decision makers’ assessment of species’ risk and subsequent status decisions, our results can aid partners in prioritizing sites and strategies at the population, regional, and range-wide level for the southern hognose snake. Our analytical approach is useful for estimating persistence for other data-limited species and is currently being used in the Florida pine snake SSA.
Reducing Uncertainties in Eastern Black Rail Conservation: Identifying Species and Management Needs for a Rare and Highly Cryptic Marsh Bird
Maureen Correll, Abby Lawson, Mitch Hartley, Michelle Stantial, Craig Watson, Aimee Weldon, Jim Lyons
Eastern Black Rail populations have declined by more than 90% in their Atlantic Coast range since the 1990’s. Eastern Black Rail populations historically have suffered from widespread conversion and alteration of wetland habitat, with recent declines likely driven by sea-level rise and nest inundation from increased tidal flooding. This rapid decline (~ 9% annually) and the Black Rail’s shrinking range triggered federal listing of the subspecies as “Threatened” in 2020. Despite recent attention from its heightened conservation status, little is known about successful management strategies to benefit the species, due in large part to the combined challenges of its relative scarcity on the landscape and the difficulty in detecting individuals even when they are present. To identify paths forward despite this uncertainty, the Atlantic Coast Joint Venture created a partner-led Black Rail Conservation Plan to identify recovery criteria for this species. We then held an adaptive management workshop in January 2020 with twenty-six workshop participants to 1) frame the conservation problem, 2) identify objectives, and 3) create influence diagrams including alternatives, consequences, and tradeoffs for the established relationships. These results were then used to inform experimental design to test partner-developed hypotheses regarding the species.   
Reducing Uncertainties in Eastern Black Rail Conservation: Qualitative Value of Information to Prioritize Conservation Actions
Abby Lawson, Michelle Stantial, Maureen Correll, Mitch Hartley, Kevin Kalasz, Michael Runge, Amy Schwarzer, Craig Watson, Aimee Weldon, Mark Woodrey, Jim Lyons
Value of information (VoI) evaluates the importance of different scientific uncertainties in the context of management. VoI is a useful approach in adaptive management, though the quantitative information needs can be substantial. We used a new qualitative approach to estimating VoI to prioritize scientific uncertainties about management to benefit eastern black rails (Laterallus jamaicensis jamaicensis) on the U.S. Atlantic coast. Thirteen hypotheses were identified by a group of black rail experts and habitat managers participating in an adaptive management workshop held in January 2020. Twenty-six workshop participants scored the alternative hypotheses based on their expert judgment about 1) the degree of theoretical foundation and empirical support in the literature, 2) relevance to management decisions, and 3) degree to which uncertainty could be reduced. Using the participant’s ordinal-scale assessments of these three aspects, we calculated a qualitative VoI and assigned each hypothesis to one of four prioritization categories. More than half of the hypotheses (n = 7) were designated as “Highest” priority, meaning that their theoretical foundations were highly uncertain but resolving the uncertainty was deemed relevant to management decisions and attainable. Two hypotheses were assigned to the “High” priority category, followed by Medium (n = 1), and Low (n = 3). Using the QVoI framework, we ultimately selected two hypotheses to test with management experiments: “Water Application” and “Fire vs. Herbicide”. Here we discuss the qualitative VoI framework and its use in endangered species recovery planning, as well as the development of predictive Bayes Net models for each experiment that will be implemented in a resilience experimentalist adaptive management framework for eastern black rail populations on the Atlantic coast.
Incorporating Competition and Climate in Management of a High-Elevation Endemic Salamander: Decision Framing
Evan Childress, Evan Grant, Adrianne Brand, John Wofford
Even in protected areas, climate change is presumed to threaten high-elevation biota. Assessing climate change impacts is required for efficient spending of funds, suitable management of rare and endangered species, and effective conservation of National Park Service biological resources. National Park Service managers must choose among management actions that might mitigate the potential negative effects of climate change. Managers currently face such choices for high-elevation endemic salamanders, such as the Shenandoah salamander (Plethodon shenandoah), a federally endangered species that is endemic to Shenandoah National Park. Shenandoah National Park and the U.S. Geological Survey, along with stakeholders, conducted two Structured Decision Making workshops to frame the park’s decision problem. USGS and NPS developed an iterative decision process for managing P. shenandoah extinction risk that includes updates from ongoing research and monitoring. To give workshop participants a sense of the decision landscape, we conducted an initial modeling effort and framed the problem to decide among a restricted set of management alternatives when consequences accrue over a 60-year period. In the first workshop, participants developed a prototype of the decision, including framing the problem statement, identifying management objectives and potential management strategies, evaluating the alternatives to identify potential solutions, and identifying data needs to reduce key uncertainties in the decision. Participants in the second workshop included experts in National Park Service policy at multiple administrative levels, who refined objectives, further evaluated the initial management alternatives against NPS policy, and discussed constraints on implementing active management for the species and its high-elevation habitat. After reviewing the current state of information and refining objectives, workshop participants determined that active management, within the constraints of agency policy, should be considered to reduce the long-term extinction risk and suggested research to support development of a management policy for the species.
Incorporating Competition and Climate in Management of a High-Elevation Endemic Salamander: Modeling
Evan Grant, Graziella DiRenzo, Jo Werba, Adrianne Brand
Managers at Shenandoah National Park need validated information to better manage the endangered Shenandoah salamander in the face of climate change. Importantly, over the last several years, substantial efforts have occurred to develop an improved understanding of the species distribution and ecology. The USGS has implemented a research program designed to estimate demographic rates and collect microhabitat and meteorological data for extinction risk assessment and conservation planning for the Shenandoah salamander. This work has included field- and laboratory-based studies of the salamander, field studies and modeling of climate change projections of temperature and relative humidity within the salamander’s habitat, and an updated  population viability analysis to estimate the future extinction risk. Given the endangered status of the species and the large degree of uncertainty in climate change projections and species response, a science panel review of current information is required by NPS in evaluating support for management actions. We present a summary of current information on the Shenandoah salamander and how we are using Expected Value of Information to present uncertainties to a NPS scientific review committee alongside the evaluation of alternative management actions designed to reduce extinction risk.
Informing Regional Invasive Species Management: Defining the Problem, Objectives, and Actions
Jennifer Dean
Informing Regional Invasive Species Management: Developing and Parameterizing a Resource Allocation Tool
Jennifer Price Tack, Qinru Shi, Carrie Brown-Lima, Jennifer Dean, Carla Gomes, Angela Fuller
Invasive species management decisions are inherently spatio-temporal, highly uncertain, and require consideration of species-specific biological attributes and treatment feasibility. During Part 2 our case study, we describe how we used structured decision making (SDM) to guide development of a management tool for tackling these complex decisions. Our tool considers the environmental, social, and economic impacts and the cost of management of 273 invasive species in New York state and divides eight regions of the state into 5 x 5 km blocks. We considered three possible management strategies in each block: 1) search, destroy, prevent, 2) direct intervention and 3) no direct action. We used expert elicitation to parameterize species-specific components of the model when published data were unavailable, including values for costs, treatment effectiveness, and dispersal rates. We accounted for the value of each block relative to each management objective using relevant spatial datasets. Based on the species and block data, we developed a mixed-integer-programming-based optimization model that provides managers with an optimal five-year strategy within each block for each species, given the available budget. The optimal strategy largely allocates resources for direct intervention to areas with known infestations of species with fewer, more isolated patches. With the budgets available to managers, there were few to no cases in which resources were allocated to manage moderately abundant or widespread species, especially when effective treatment strategies were unavailable. Our case study demonstrates how SDM can effectively guide the development of a decision tool for managers that produces highly informative output for informing management and exploring tradeoffs among the objectives. Note: This SDM case study is presented in two talks. In Part 1 we focus on the development of the problem, objectives, and alternatives. In Part 2, we describe the development, parameterization, and results of the decision tool.
Estimating Allowable Take of Bald and Golden Eagles with Land-Based Wind as a Model: Problem
EMILY BJERRE, Brian Millsap, Guthrie Zimmerman
With an increased interest in alternative energy development and land-based wind in particular, the U.S. Fish and Wildlife Service (Service) developed an approach to managing incidental take for bald and golden eagles that had to balance regulatory responsibility, information available, and regulatory burden. Designing a ‘harvest’ management framework that incorporated iterative learning, the Service developed management objectives that included explicit statements about the management scale, population objectives, and risk tolerance given uncertainty. We use model at different scales to estimate allowable take within designated management units. Take authorizations that meet certain criteria and are within allowable take limits are considered compatible with the Service’s overall management objectives. However, once infrastructure is on the landscape, even if models suggest take may be exceeding what is allowable given the objectives, there are few management actions available to reduce mortality. For golden eagles in particular, we face the challenge of limited known opportunities to offset impacts from other sources of mortality. 
Estimating Allowable Take of Bald and Golden Eagles at Wind Farms: Modeling
Guthrie Zimmerman, Brian Millsap, EMILY BJERRE
We used the Prescribed Take Level (PTL) framework to estimate allowable take of bald and golden eagles.  The PTL framework enabled us to integrate policy decisions, including a management objective and risk tolerance, with information on growth rate and population size for estimating allowable take.  We developed integrated population models (IPMs) for both species to analyze available data.  By integrating multiple data sources in a single analysis, we gained improved parameter estimates for informing growth rates for the PTL, as well as other parameters needed to improve take estimates (e.g., proportion of adult birds breeding, proportion of individuals in each age class) that were not estimable when data were analyzed separately.  For golden eagles, our model indicates current mortality rates are at or above target levels. The models also highlight ongoing anthropogenic take as a significant source of mortality. In contrast, our analysis indicated that allowable take is above current levels authorized for the rapidly increasing bald eagle population.  Regardless of existing mortality, there is a need to authorize additional mortality associated with current and future development of wind and alternative energy infrastructure.  Although the PTL framework and IPM provided improved estimates of allowable take for both species, we still face considerable uncertainties such as actual take resulting from permitting (e.g., we can allocate a specific number of take permits, but do not control what proportion are actually filled).  Further, we lack management alternatives that can be implemented to control reductions in take if actual take is above estimated allowable levels.  Our approach is flexible and can incorporate additional information if it becomes available (e.g., linking actual take to population dynamics).
Reintroduction of Eastern Indigo Snakes in the Western Portion of Its Historic Range: What Does Successful Establishment Mean?
David Steen
Reintroduction of Eastern Indigo Snakes in the Western Portion of Its Historic Range: Predictive Modeling of Reintroduction Strategies
Conor McGowan, Brian Folt, David Steen
Wildlife repatriations can be long and expensive processes, clear release strategies and monitoring programs are essential to efficiently use resources and evaluate success. Prior evaluation of release strategies using predictive modeling can improve changes of success and save effort and cost in the long run.  We developed a simulation model and evaluated how alternative repatriation strategies influence repatriation success for the eastern indigo snake (Drymarchon couperi), a federally-Threatened species that is currently being repatriated in Alabama and Florida. Further, we demonstrate how monitoring results should be interpreted given low detection probabilities. Specifically, we built a stochastic stage-based population model and predicted population growth and extinction risk under different release scenarios. Then, we applied data from ongoing repatriations to predict success and guide future releases in an adaptive-management framework. Because D. couperi is difficult to monitor, we modeled how detection probability influenced perceptions of abundance and population growth by monitoring programs. Simulated repatriation scenarios releasing older, head-started snakes in greater abundance and frequency created wild populations with increased population growth and decreased extinction risk relative to scenarios releasing fewer and younger snakes less frequently. Ongoing repatriations in Alabama and Florida currently have a 0.23 and 0.61 probability of quasi-extinction, respectively, but extinction risk decreased to 0.07 and 0.10 once target number of released individuals is achieved. Abundances observed under realistic detection thresholds did not always predict true population growth; specifically, we demonstrate that monitoring programs during repatriations of secretive species frequently indicate that efforts have been unsuccessful when populations are in fact growing. Overall, our modeling framework informs release strategies to maximize repatriation success while demonstrating the need to consider how detection processes influence assessment of success during conservation interventions, particularly in an adaptive-management framework.
Rapid Response to Invasive Dreissenid Edna in Western Reservoirs: Framing
Adam Sepulveda, Katherine O’Donnell, David Smith, Nathan Owens
Invasive species managers across the country are grappling with using environmental (e)DNA for decision making since detections do not always indicate presence of a live animal at a site. Failure to act on a true positive could compound negative ecosystem impacts and increase already costly control measures; while acting on ‘false positive’ data could result in needless costs and efforts. To assist managers with navigating the high level of uncertainty following eDNA detections, we used a structured decision making approach to evaluate plausible tradeoffs associated with alternative management actions. We present a case study that demonstrates how decisions about management of dreissenid mussel eDNA detections at Jordanelle Reservoir (UT) can be made through a structured and participatory process by scientists and stakeholders, including Utah Department of Wildlife Resources, Bureau of Reclamation and Central Utah Water Conservation District.
Rapid Response to Invasive Dreissenid Edna in Western Reservoirs: Risk Modeling and the Value of Sample Information
Katherine O’Donnell, David Smith, Adam Sepulveda, Nathan Owens
Early detection of invasive species enables managers to respond rapidly, increasing their chances of successfully controlling an incursion. Environmental DNA (eDNA) sampling can be used to detect invasive species before they fully establish; however, managers often wait for detections from non-eDNA sampling before implementing response actions. If invaders are truly present, delaying response actions could lead to negative economic and ecological impacts; on the other hand, implementing rapid response actions if invaders are not present (i.e., eDNA was false positive) could cause unnecessary spending and, over time, declining public support (i.e., “the boy who cried wolf”). In our case study, we evaluated rapid response options after hypothetical eDNA detections of dreissenid (quagga/zebra) mussels in a Utah reservoir. We used structured decision making (SDM) to frame the problem, develop objectives, identify alternative management actions, predict consequences, and evaluate trade-offs and risks. We used decision trees to frame the decision analysis, casting the choice to delay (i.e., wait for non-eDNA confirmation) as a Value of Information problem. We elicited quantities of interest (e.g., consequences, utilities) from SDM participants and explored risks and trade-offs through sensitivity analyses. The final decision support framework helps managers evaluate rapid response options while considering multiple objectives and future scenarios.

Location: Virtual Date: November 2, 2021 Time: 3:00 pm - 4:00 pm