Methods for Planning Optimal Elimination Strategies Adaptively in Wide-Ranging Structured Populations

ROOM: CC, Room 25A
This half-day symposium will convene researchers and practitioners focused on the development of methodologies and strategies for planning elimination of invasive species populations or disease in wide-ranging structured populations. The symposium will cover applications of structured decision making for management of invasive species or disease. We will emphasize methods for considering population connectivity and other types of surveillance data to estimate elimination certainty. The primary goal is to present and discuss practical methodology for integration of demographic connectivity in cost-optimization frameworks that consider spatial processes and are developed using realistic surveillance data sources.

12:50PM Optimal Allocation of a Fixed Budget for Planning Elimination of Genetically Structured Pest Populations
  Kim Pepin; Tim Smyser; Amy Davis; Ryan Miller; Michael Tabak; Gabriel Gellner; Chris Slootmaker
Elimination programs for feral swine are challenged both logistically by limited funding and ecologically by population structure. Studies have shown that populations typically consist of multiple subpopulations that cluster in space but show some level of genetic mixing across long-distances, presumably due to human-mediated movement. We use a dynamic population model to identify conditions where the connectivity structure and magnitude of movement between subpopulations (demographic connectivity) would influence the outcome of spatial management strategies. We determine the optimal management strategies with an economic model that determines the likelihood of elimination success based on the resource budget and movement patterns. Our main objective is to develop tools for identifying optimal spatial control strategies and the financial budget needed for elimination of feral swine in a target area. We found that elimination can be achieved faster and require removing fewer individuals when control efforts are preferentially allocated to source subpopulations (those that provide migrants or translocated individuals) before shifting allocations to recipient subpopulations (those that receive migrants or translocated individuals), but these benefits are only achieved when source to recipient movements are frequent and unidirectional. Our results suggested that allocating all resources equally among subpopulations initially is optimal if it results in large initial decreases for all populations. We also showed the proportional budget increase that is necessary when populations have higher connectivity and when translocation of individuals from other populations occurs. We discuss how these tools can be applied to management data to support planning of large scale feral swine removal activities. As this presentation is part of a symposium on planning optimal elimination strategies adaptively in wide-ranging structured populations, we also give insights on gaps, challenges, and opportunities for using bioeconomic models to address management objectives in highly structured populations.
1:10PM Tracking Invasive Hybridization Through Time: Using Temporally Spaced Genomic Data to Prioritize Conservation Areas
  Brad Shaffer; Evan McCartney-Melstad; Erin Toffelmier; Robert Cooper
Hybridization between endangered and introduced species is a key concern for conservation biology. The federally protected California tiger salamander (Ambystoma californiense) has been hybridizing with the introduced barred tiger salamander (Ambystoma mavortium) since the 1950s when it was intentionally introduced into central California. Given its strong ecological effects, the movement of invasive alleles into native California tiger salamander populations and genomes represents a key management focus for state and federal regulators. Our earlier work with 68 nuclear SNP markers identified a large hybrid swarm in the central coast of California as well as the rough edges of the hybrid zone to the north and east. To further explore how natural selection has shaped this recent contact zone, we mapped 5,237 exons on the genome, sequenced these exons using a target enrichment approach in over 3,000 salamanders collected over the past 30 years, and used repeated temporal sampling over the range of the species to map fluctuations in hybrid indices over time and space. We discuss two separate barred tiger salamander invasions, in Monterey and Santa Barbara Counties, where non-native genes threaten native populations, and strategies for using this temporal series to help determine conservation actions. Finally, we briefly discuss an ongoing large-scale field experiment that seeks to ask whether we can use pond hydroperiod manipulations as a tool to select for native genes and genotypes, essentially using natural selection to increase native salamander genotypes on the landscape.
1:30PM A Population Genomic Approach to Estimating Migration Rate When Sample Size Is Small and Populations Are Closely Related
  Timothy J. Smyser; Michael A. Tabak; Kim M. Pepin; Amy J. Davis; Ryan S. Miller
Quantifying the movement of individuals among subpopulations can be difficult when subpopulations are closely related or sample sizes are small. We developed a population genomics approach that can resolve migration rates under such challenging conditions. Specifically, we conducted a supervised analysis with ADMIXTURE using a leave-one-out (LOO) approach in which we iteratively queried the genetic origin of a single individual while all other individuals served as reference samples for the subpopulation in which they were sampled. Using the observed genotypes within each subpopulation, we bootstrapped across loci to assemble genotypes for parental populations, descendants of migrants, and grand descendants of migrants. We repeated the LOO analysis on the bootstrapped dataset. By comparing the assignment of observed individuals among respective subpopulations to parallel distributions generated from the bootstrapped dataset, we were able to classify observed individuals as residents, migrants, descendants of migrants, and grand descendants of migrants. We validated our method with a genetically explicit individual-based model in which we incorporated biologically-informed demographic parameters, independent pairwise migration rates, and the physical linkage structure among loci. We then evaluated the ability of the LOO approach to return the true genetic origin of simulated individuals given their known ancestry derived from the model. Having quantified the discriminatory power of the LOO approach, we applied this method to quantify migration rates of feral swine (Sus scrofa) among subpopulations in Missouri, USA. We documented high rates of translocation among subpopulations and introductions from exogenous sources. Evaluating the influence of environmental and sociological covariates on translocation rates, we found translocation rates were positively associated with the density of Cervid farms, frequency of natural resource violations, and availability of public land. Further, our results demonstrate that feral swine introduction pressure increases as deer harvest decreases, suggesting that hunters are substituting feral swine as a large game species.
1:50PM Dynamic Occupancy Applications to Estimate the Probability of Elimination Across Space and Time
  Amy Davis; Amy Gilbert; Antoinette Piaggio; Colleen Webb; Richard Chipman; Kim Pepin
Normal 0 false false false EN-US X-NONE X-NONE /* Style Definitions */ table.MsoNormalTable {mso-style-name:”Table Normal”; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-parent:””; mso-padding-alt:0in 5.4pt 0in 5.4pt; mso-para-margin-top:0in; mso-para-margin-right:0in; mso-para-margin-bottom:8.0pt; mso-para-margin-left:0in; line-height:107%; mso-pagination:widow-orphan; font-size:11.0pt; font-family:”Calibri”,sans-serif; mso-ascii-font-family:Calibri; mso-ascii-theme-font:minor-latin; mso-hansi-font-family:Calibri; mso-hansi-theme-font:minor-latin; mso-bidi-font-family:”Times New Roman”; mso-bidi-theme-font:minor-bidi;} Some of the largest threats to wildlife and human health come from invasive species and diseases. Invasive species and disease management often face similar challenges particularly with respect to surveillance and understanding when negative surveillance implies the species or disease is absent and when it was just undetected. Prematurely declaring elimination comes with considerable risk to allowing persistence or reinvasion which comes with potentially high economic costs. To accurately estimate the probability of elimination we need to have unbiased estimates of detection probability. Dynamic occupancy models are well suited to address issues of species or diseases that are near elimination. These models directly model changes in occupancy probability across space and time while accounting for imperfect detection. We demonstrate the utility of this approach to management surveillance methods for an invasive species, wild pigs (Sus scrofa), and for circulation of rabies virus in raccoon (Procyon lotor) populations across a broad management area. We show how multiple surveillance methods can be incorporated in a single framework and evaluate the relative detection probabilities and thus make management recommendations. Additionally, we can make resource allocation recommendations and provide levels of confidence for local elimination probabilities. We discuss how this framework can be easily incorporated into an adaptive management framework.
2:10PM Optimization of Control and Surveillance Activities to Achieve and Prove Freedom From Bovine Tuberculosis (Tb) in New Zealand Wildlife
  Andrew M. Gormley; Graham Nugent; Dean P. Anderson; Kevin Crews
Bovine TB is a bacterial disease typically associated with cattle, but with a wide host range including many domestic and wild animals. In New Zealand the predominant wildlife maintenance host is the introduced brushtail possum (Trichosurus vulpecula). In areas of high possum density, localised populations can maintain TB and readily transmit infection to adjacent farmed cattle and deer. The TBfree New Zealand program is a dual program of livestock testing and management of the disease in possums. The current program has the targets of freedom from TB in wildlife by 2040 and complete biological eradication of bTB by 2050. Management of TB in wildlife consists of an initial phase of long-term large-scale lethal control of possums in order to suppress the local population to sufficiently low levels such that TB can no longer persist. Control is followed by the ‘Proof of Freedom’ (PoF) phase: a surveillance framework used to assess the probability that TB in wildlife has been locally eradicated based on quantitative assessment using Bayesian belief updating of negative surveillance data, thereby proving local freedom from the disease. In this talk I will present an overview of the wildlife component of the TBfree programme, with a particular focus on the PoF, including how we include the various available data sources such as direct surveillance of possums as well as using feral pigs, ferrets and livestock as sentinels. I will also present a number of novel approaches related to the optimisation of control and surveillance that having been recently developed due to existing methods being impractical or expensive. Finally I discuss how the proof of freedom framework can, and has been applied to a number of other wildlife management scenarios.
2:30PM Refreshment Break
3:20PM Models to Evaluate Management Activities on Structured Populations Using Unmarked Data
  Elise Zipkin
Understanding the impacts of management activities on a target species requires inferences on population dynamics. Ideally, this is achieved through estimation of spatio-temporal demographic rates such as survival and fecundity across life history stages. Yet estimating these quantities can be difficult, requiring years of intensive data collection. Often this is accomplished through the capture and recapture of individual animals, which is generally only feasible at a limited number of locations. In contrast, recently developed models using a state-space formulation allow for the estimation of abundance and spatial variation in abundance from occupancy and count data for both closed (e.g., Royle-Nichols model, N-mixture model) and open (Dail-Madsen model, dynamic N-occupancy model) populations. These models require repeated survey events during a time period when the population is closed and thus detection errors can be explicitly attributed to false-negatives in the data (e.g., failure to detect an individual when it is present). We present recent extensions to the unmarked modeling framework that allow for state-specific estimation of demographic rates from stage-structured population data. We also demonstrate how detection/nondetection (e.g., occupancy) data can be used either separately or in conjunction with count data to estimate recruitment, survival, colonization, and extinction rates. We discuss the data requirements (e.g., number of survey locations, years, and replicate sampling events) for both accurate and precise estimates of the parameters of interest.
3:40PM Cost-Effective Eradication Through Optimized Early Detection Investments Across Heterogeneous Landscapes
  Rebecca Epanchin-Niell
Eradication and control efforts to reduce long term damages from invasive species generally require prior detection of target populations. Detection of a new invader population can be the result of a formal surveillance program, such as visual survey or trapping programs implemented by state or federal agencies, or more haphazard reporting by the public or other pathways. As established invaders spread across the landscape, control and damage costs increase and the likelihood of successful eradication decreases. Thus, expected long term control and damage costs are expected to decrease with increasing investments in early detection survey programs that enable control to begin sooner. However, the value of such a survey program will vary across species and sites, dependent on their characteristics. Here we describe a bioeconomic model for optimizing early detection surveillance investments across a heterogeneous landscape and across multiple potential invaders to minimize the long term costs from the invader and its control. The model accounts for uncertainty in species’ characteristics and in the costs and effectiveness of control measures. We use the model to show how the value of information from surveillance investments depends on a species’ introduction rate, spread rate, damage potential, control costs and effectiveness, the likelihood of detection by the public, and the costs and effectiveness of survey methods. The decision-support tool can be used to inform efficient allocation of surveillance resources across target species and sites for early detection of new invasive species populations.
4:00PM What and How to Manage? Choosing Good Management Actions for Multiple Threats.
  Joslin Moore
Most threatened species, communities or protected areas are subject to multiple threats, and the distribution of these threats varies in space and time. How do we decide which threats to address first and in which locations? Existing decision frameworks can be helpful when choosing between actions targeting a specific threat. However, they rarely account for multiple threats and the consequent future losses associated with threats left unmanaged. Nor that there is often more than one candidate action to address any given threat. Furthermore, limited resources mean that deciding which threats to manage how is context dependent in that it will depend on the distribution and relative management feasibility of other threats as well. These interdependencies and unattributed losses are important considerations when allocating resources to the management of a specific population, community or national park facing multiple threats. We present a framework that considers the potential impact of current and future threats and accounts for the combined benefit associated with managing a suite of threats while also accounting for the consequences of unmanaged threats. Formulating the decision as a multiple-choice knapsack problem enables us to simultaneously account for multiple threats that can each be addressed using multiple actions. Because uncertainty is high and the number of parameters can be substantial, we use global sensitivity analysis and value of information analysis to identify important uncertainties. We apply the framework to a range of case studies that have been developed with land agencies using structured decision making including prioritising management of introduced plant species, managing multiple threats to endangered peatlands and developing a surveillance plan for early detection of invasive species in marine protected areas.
4:20PM Navigating the Tradeoffs in Invasive Control with Multi-Criteria Decision Analysis
  Michael C. Runge
Eliminating invasive species is often difficult and costly. Moreover, both the benefits and costs of attempting elimination may be multi-faceted. The presence of invasive species may imperil native species and disrupt important ecosystem services, helping make the case for elimination. But eradication or containment efforts themselves may also cause inadvertent harm to native species and ecosystem services, and may create disturbances in the ecosystem that invite new invasions. Thus, embedded in the question about how to design optimal elimination strategies is a larger question—whether to attempt elimination at all. Multi-criteria decision analysis provides a four-step framework for evaluating the tradeoffs and weighing the benefits and costs of an elimination effort. First, articulate the full set of objectives of interest, both benefits and costs, recognizing that although elimination of the invasive species may be an end in itself, it is also a means to other fundamental outcomes. Second, design alternative strategies for consideration, from no intervention to containment to full eradication attempts. Third, forecast the effects of the alternatives on all the objectives, using the best available science and explicitly accounting for uncertainty. Fourth, employ decision analysis tools to help the decision maker weigh the tradeoffs among objectives. Even in cases where it is clear that an elimination attempt is warranted, consideration of the wider range of objectives can influence the optimal design of an elimination strategy. Further, when uncertainty impedes the design of a strategy, and alternative designs affect other objectives differently, consideration of the multiple objectives may influence the implementation of adaptive management.
4:40PM Panel Discussion

Organizers: Kim Pepin, National Wildlife Research Center, USDA-APHIS-WS, Fort Collins, CO; Amy Davis, National Wildlife Research Center, USDA-APHIS-WS, Fort Collins, CO;
Supported by: Biometrics Working Group and Wildlife Damage Management Working Group

Location: Cleveland CC Date: October 11, 2018 Time: 12:50 pm - 5:00 pm