Integrated analysis for population estimation, management impact evaluation, and decision-making for a declining species
Graphical abstract
Introduction
Conservation management requires addressing problems involving complex interactions between social and ecological systems; multiple, concurrent threats to natural resources; and potential strategies whose outcomes are uncertain (Game et al., 2014). Concomitantly, ecological modeling can help infer and forecast system dynamics, upon which management decisions can be based. Modeling approaches that are realistic in the representation of context-specific processes and transparent in the treatment of key uncertainties are a means to robust conservation decisions (Schmolke et al., 2010).
Population viability analysis (PVA) is an effective tool for predicting outcomes of interest (e.g., population abundance, growth, persistence) for wildlife species, (Akçakaya and Sjögren-Gulve, 2000; Morris and Doak, 2002). PVAs are highly customizable to a species' life history (e.g., life stages, behavioral states) and context-specific factors that affect demographic rates on which predictions are based (Akçakaya and Sjögren-Gulve, 2000; Morris and Doak, 2002; Rhodes et al., 2011; Wilson et al., 2016). Modeling multiple, concurrent threats within a single PVA is crucial for decision-making, since factors not addressed may render targeted management actions ineffective (Heppell et al., 1996; Rhodes et al., 2011; Crawford et al., 2014a). PVAs also provide decision-relevant information via efficient evaluation of the sensitivity of model outcomes to changes in parameter values, including values estimated by expert opinion (Wade, 2002). Still, obtaining reliable predictions from PVAs remains challenging within many conservation contexts given multiple sources of parameter uncertainty. These issues magnify as PVAs are commonly applied to rare, declining, and cryptic species with sparse datasets. Population parameters (e.g., abundance, survival, productivity) are estimated from observation data; thus, uncertainty around parameter estimates inherently includes variation of demographic process as well as observation error that should be separated before making inferences (Clark and Bjørnstad, 2004). Overestimation of demographic rates and increases in uncertainty can occur for species with limited data or low detection (e.g., Zipkin et al., 2014). In some cases, there may be no current data on which to estimate parameters, such as for rare species or novel management actions, and PVAs may rely on expert elicitation associated with higher degrees of uncertainty (Krueger et al., 2012). Finally, the effectiveness of management strategies may be difficult to estimate for cryptic species or those with low productivity or delayed maturity because longer post-management periods are necessary to detect changes in population growth (Heppell et al., 1996; Moore et al., 2012; Tempel et al., 2014).
Novel modeling approaches have been developed to improve the accuracy of parameter estimates and population predictions associated with PVAs. First, an integrated model is a unified analysis that can leverage information contained in multiple, partial datasets to estimate shared demographic processes for a population (e.g., Wilson et al., 2016). Integrated models increase precision, ensure consistency of estimates across datasets, and reduce effects of potential bias of individual datasets (Schaub and Abadi, 2011). Examples of these frameworks include the joint live-dead encounter model for mark-recapture and dead-recovery data developed by Burnham (1993) and, more recently, integrated population models (IPMs) for the unified analysis of mark-recapture, population count, and other datasets (Schaub and Abadi, 2011). Second, recent PVA formulations have been developed to improve the accuracy of population predictions by formally incorporating uncertainty around parameter estimates while separately modeling annual stochasticity in population simulations (e.g., Moore et al., 2012; Shoemaker et al., 2013). These models have been constructed in Bayesian (e.g., Bayesian PVAs: Wade, 2002, Kéry and Schaub, 2012) and frequentist frameworks (e.g., McGowan et al., 2011), and we refer to this general class of models as robust PVAs. Robust PVAs reduce the risk of overestimating population outcomes, such as probability of persistence (McGowan et al., 2011), and have also been used to explicitly evaluate effects of management alternatives on population outcomes (Moore et al., 2012; Hegg et al., 2013; Servanty et al., 2014; Green and Bailey, 2015). To date, the application of integrated models to conservation issues is limited (Schaub and Abadi, 2011; Zipkin and Saunders, 2018). The application of robust PVAs to evaluate management actions is growing, but these efforts have not been coupled with integrated models for improved parameter estimation in the context of conservation decision-making (but see Hoyle and Maunder, 2004, Maunder, 2004, Lieury et al., 2015, Saunders et al., 2018). Here, we use integrated models and robust PVAs to estimate context-specific demographic rates, evaluate management actions, and predict population outcomes to inform decision-making for a declining species of conservation concern, the diamondback terrapin (Malaclemys terrapin).
Diamondback terrapins inhabit salt marshes along the Eastern and Gulf Coasts of the United States – regions experiencing the fastest annual increases in developed area, road density, and traffic loads (Baird, 2009). Multiple anthropogenic threats contribute to terrapin population declines, which has prompted many states to list the species as “of special concern” or a higher protection status (Roosenburg, 1991; Gibbons et al., 2001; Grosse et al., 2011; Crawford et al., 2014a; Chambers and Maerz, in press; Maerz et al. in press). Terrapins are frequent bycatch in commercial and recreational crab pot fisheries (Roosenburg et al., 1997; Grosse et al., 2011; Chambers and Maerz, in press), and in areas where roads fragment salt marsh, adult females are struck by vehicles while searching for elevated nesting habitat (Butler et al., 2006; Szerlag-Egger and McRobert, 2007; Crawford et al., 2014b). Terrapins share characteristics with the majority of turtles (e.g., long-lived, delayed maturity, naturally high adult survival) that are likely to make populations susceptible to even low rates (3–10%) of additive mortality due to roads (Gibbs and Shriver, 2002; Steen and Gibbs, 2004; Butler et al., 2006; Maerz et al. in press). Human-subsidized predators, such as raccoons (Procyon lotor), contribute to high rates (50–90%) of nest mortality on roadsides and other developed areas (Crawford, 2015; Maerz et al. in press). The density of roadside vegetation can also influence terrapin demographic rates. Grosse et al. (2015) observed higher predation rates and higher proportions of male hatchlings for nests laid in planted hedgerows (commonly cedar and wax myrtle Myrica cerifera), relative to cleared, open areas along roadsides. Like many reptiles, terrapins exhibit environmental sex determination (ESD) where warmer incubation temperatures produce greater proportions of female offspring (Ewert et al., 1994). While existing management practices have targeted road mortality (Aresco, 2005) and predation (Munscher et al., 2012), vegetation management practices also have the potential to increase population growth (Maerz et al. in press).
The aim of this research was to apply an integrated analysis to evaluate the consequences of management strategies to inform decision-making within the context of road impacts on wildlife. We used a population of terrapins that nest on the causeway to Jekyll Island, Georgia, USA as a model system. Our specific objectives were (i) to develop an integrated model to jointly estimate demographic rates from two mark-recapture datasets, (ii) to directly estimate impacts of road mortality and management actions deployed during the study on demographic rates, and (iii) to incorporate estimates from this and other studies, as well as expert opinion, in a robust PVA to project population persistence under simulated management strategies. This work builds on previous research that estimated the effects of road-associated threats and identified management targets (Crawford et al., 2014a; Crawford et al., 2014b; Crawford et al., 2017; Grosse et al., 2015). It precedes research that will incorporate population persistence outcomes for each strategy in the context of other socioeconomic objectives for road management on Jekyll Island. Our approach, linking integrated models and robust PVAs in a unified analysis, is applicable across conservation contexts for using limited data efficiently, tailoring models to represent system complexity, and prioritizing threats and management actions that impact at-risk populations.
Section snippets
Study area and population
We conducted research in conjunction with long-term monitoring efforts of the Georgia Sea Turtle Center (GSTC) on the 8.7-km Downing-Musgrove Causeway (aka Jekyll Island Causeway: JIC) to Jekyll Island, GA, USA (31.08°N, 81.47°W; Fig. 1). The JIC bisects a salt marsh peninsula consisting of a network of intertidal creeks and high marsh dominated by Spartina spp. We defined the population of interest for this study as terrapins inhabiting this peninsula and using JIC roadsides for nesting. The
Results
The Robust Design dataset included adult males (294 encounters of 194 individuals) and females (68 encounters of 56 individuals). Since we only captured 33 terrapins (with 7 total recaptures) from one of the creeks during the study period, we combined data from both creeks for analysis. Annual captures ranged from 4 to 100 individual males and 2 to 21 females. The multistate dataset contained 2307 encounters (FA,c: 1065 [46.2%]; FA,nc: 227 [9.8%]; Fd,c: 1015 [44.0%]) of 1984 individuals. The
Discussion
There are few studies to date that used joint analyses, such as integrated population models, of multiple data sources to explore the conservation status and management targets for a population (Rhodes et al., 2011; Tempel et al., 2014; Lieury et al., 2015; Wilson et al., 2016; Saunders et al., 2018). Our work contributes an example of applying integrated modeling and robust PVA approaches for rigorous parameter estimation, population prediction, and management strategy evaluation for improved
Acknowledgments
Funding to support this research was provided by the Daniel B. Warnell School of Forestry and Natural Resources and the Graduate School, University of Georgia, as well as AGL Resources Foundation through the Jekyll Island Foundation through a grant to J.C.M. We thank GSTC staff and volunteers for their assistance throughout the project – especially M. Kaylor, D. Quinn, A. Grosse, R. Cozad, A. Gillis, L. Rodriguez, and S. Diltz. We thank associates at the Savannah River Ecology Lab and the
References (70)
- et al.
Crab trapping causes population decline and demographic changes in diamondback terrapins over two decades
Biol. Conserv.
(2007) - et al.
The effect of road kills on amphibian populations
Biol. Conserv.
(2001) - et al.
The role of expert opinion in environmental modelling
Environ. Model Softw.
(2012) - et al.
Relative contribution of local demography and immigration in the recovery of a geographically-isolated population of the endangered Egyptian vulture
Biol. Conserv.
(2015) Population viability analysis based on combining Bayesian, integrated, and hierarchical analyses
Acta Oecol.
(2004)- et al.
Incorporating parametric uncertainty into population viability analysis models
Biol. Conserv.
(2011) - et al.
Using integrated population modelling to quantify the implications of multiple threatening processes for a rapidly declining population
Biol. Conserv.
(2011) - et al.
Demographic effects of road mortality in black ratsnakes (Elaphe obsoleta)
Biol. Conserv.
(2007) - et al.
Ecological models supporting environmental decision making: a strategy for the future
Trends Ecol. Evol.
(2010) - et al.
Using integrated population models to improve conservation monitoring: California spotted owls as a case study
Ecol. Model.
(2014)
Integrated population modeling to assess demographic variation and contributions to population growth for endangered whooping cranes
Biol. Conserv.
Synthesizing multiple data types for biological conservation using integrated population models
Biol. Conserv.
Population viability analyses in conservation planning: an overview
Ecol. Bull.
Mitigation measures to reduce highway mortality of turtles and other herpetofauna at a North Florida lake
J. Wildl. Manag.
Coastal urbanization: the challenge of management lag
Manag. Environ. Qual.
General methods for monitoring convergence of iterative simulations
J. Comput. Graph. Stat.
Use of an artificial nesting mound by wood turtles (Glyptemys insculpta): a tool for turtle conservation
Northeast. Nat.
A Theory for Combined Analysis of Ring Recovery and Recapture Data
Third workshop on the ecology, status, and conservation of diamondback terrapins (Malaclemys terrapin): results and recommendations
Matrix Population Models: Construction, Analysis, and Interpretation
Terrapin bycatch in the blue crab fishery
Population time series: process variability, observation errors, missing values, lags, and hidden states
Ecology
Relationships of reproductive traits and body size with attainment of sexual maturity and age in Blanding's turtles (Emydoidea blandingi)
J. Evol. Biol.
Lessons from terrapin mortality and management on the Jekyll Island Causeway, Georgia, USA
Estimating the consequences of multiple threats and management strategies for semi-aquatic turtles
J. Appl. Ecol.
Hot spots and hot moments of diamondback terrapin road-crossing activity
J. Appl. Ecol.
Mitigating road mortality of diamond-backed terrapins (Malaclemys terrapin) with hybrid barriers at crossing hot spots
Herpetol. Conserv. Biol.
A stage-based population model for loggerhead sea turtles and implications for conservation
Ecology
Patterns of temperature dependent sex determination in turtles
J. Exp. Zool.
Nesting ecology and predation of diamondback terrapins, Malaclemys terrapin, at Gateway National Recreation Area, New York
J. Herpetol.
Conservation in a wicked complex world: challenges and solutions
Conserv. Lett.
Demographic and ecological factors affecting conservation and management of the diamondback terrapin (Malaclemys terrapin) in South Carolina
Estimating the effects of road mortality on turtle populations
Conserv. Biol.
Modeling the effects of crab potting and road traffic on a population of diamondback terrapins
Using Bayesian population viability analysis to define relevant conservation objectives
PLoS One
Cited by (15)
Long-term demography of a reintroduced population of endangered falcons
2022, Global Ecology and ConservationCitation Excerpt :Conversely, immigration can rescue populations from extinction (Brown and Kodric-Brown, 1977) and might be especially important for stability or growth of raptor populations (Brown and Collopy, 2013; Altwegg et al., 2014; Lieury et al., 2015; McClure et al., 2021). Management actions can similarly increase population viability (McClure et al., 2017a; Crawford et al., 2018; Saunders et al., 2018). The Northern Aplomado Falcon (Falco femoralis septentrionalis; hereafter, Aplomado Falcon) is listed under the United States Endangered Species Act (United States Department of Interior, Fish and Wildlife Service 1986).
Population viability analysis for a pond-breeding amphibian under future drought scenarios in the southeastern United States
2022, Global Ecology and ConservationCitation Excerpt :Importantly, PVAs can produce quantitative metrics – as well as transparently capture uncertainty around those metrics – that can directly inform conservation decision-making (Beissinger and Westphal, 1998; Bonnot et al., 2011; McGowan et al., 2017). PVA models are highly customizable to species-specific life history and context-specific factors, such as mechanistic effects of environmental stressors or management interventions (Akçakaya and Sjögren-Gulve, 2000; Crawford et al., 2018; Morris and Doak, 2002; Rhodes et al., 2011; Saunders et al., 2018). PVAs often use stage-based matrices as a framework for modeling species with life histories more easily characterized by stages (e.g., juvenile, adult) rather than by ages (Caswell, 2001; Crowder et al., 1994; Lefkovitch, 1965).
The lost road: Do transportation networks imperil wildlife population persistence?
2021, Perspectives in Ecology and ConservationCitation Excerpt :For instance, a study on endangered spotted turtle (Clemmys guttata) in Canada used a population viability analyses to find that probability of quasi-extinction within 150 years increased from 20–24% to 93–94% when road mortality was included in the model (Howell and Seigel, 2019). A study on vulnerable diamondback terrapin (Malaclemys terrapin) in USA obtained similar findings, as models predicted population decline in 50 years if roadkills are not mitigated (Crawford et al., 2018). Overall, studies agree in that adult (and very specially, female) survival is the key parameter to ensure population viability (Crawford et al., 2018; Howell and Seigel, 2019).
Matrix and agent-based modeling of threats to a diamond-backed terrapin population
2021, Mathematical BiosciencesCitation Excerpt :In the early 20th century, terrapins were commercially harvested, which almost led to their extinction [2,24]. More recently, studies on declining diamond-backed terrapin populations across many habitats in the United States have indicated that crab traps and nest disturbances from predators like raccoons (Procyon lotor) that benefit from human development are major threats to terrapins [25,26]. Modeling of terrapin populations in locations with limited human pressures can provide insight into how the species can cope with human influences in the long term.
Evaluating uncertainty to improve a common monitoring method and guide management decisions for diamond-backed terrapins
2024, Journal of Wildlife ManagementJuvenile proportion as a predictor of freshwater turtle population change
2023, Austral Ecology