Population viability analysis

Population Viability Analysis (PVA) is a species-specific method of risk assessment frequently used in conservation biology. It is traditionally defined as the process that determines the likelihood that a population will go extinct within a given number of years. More recently, PVA has been described as a marriage of ecology and statistics that brings together species characteristics and environmental variability to forecast population health and extinction risk. Each PVA is individually developed for a target population or species, and consequently, each PVA is unique.

Uses

Population viability analysis is used to estimate the likelihood of a population’s extinction, indicate the urgency of recovery efforts, and identify key life stages or processes that should be the focus of recovery efforts. PVA is also used to compare proposed management options and assess existing recovery efforts. PVA is frequently used in endangered species management to develop a plan of action, rank the pros and cons of different management scenarios, and asses the potential impacts of habitat loss.

History

In the 1970s, Yellowstone National Park was the center of a heated debate over different proposals to manage the Park’s problem grizzly bears (Ursus arctos). In 1978, Mark Shaffer proposed a model for the grizzlies that incorporated random variability, and calculated extinction probabilities and Minimum Viable Population size. The first PVA is credited to Shaffer.

PVA gained popularity in the United States as federal agencies and ecologists required methods to evaluate the risk of extinction and possible outcomes of management decisions, particularly in accordance with the Endangered Species Act of 1966, and the National Forest Management Act of 1976.

In 1986, Gilpin and Soulé broadened the PVA definition to include the interactive forces that affect the viability of a population, including genetics. The use of PVA increased dramatically in the late 1980s and early 1990s following advances in personal computers and software packages.

Components

Question

PVA begins with a scientific question to be explored. Questions may be broad, such as “Is this species at risk of becoming endangered? If so, why?”. Questions may also be specific, such as “What amount of take is acceptable for native steelhead trout?”, “Is it better to re-introduce 30 condors to Arizona, or re-introduce 15 condors each to Arizona and California?”, or “Which potential forest management strategy will be of greatest benefit to Kirtland’s Warbler?”.

Model

The model structure for a PVA is primarily determined by the question at hand, the available data, and important characteristics of the species, such as complex social structures or age of first reproduction. Depending on available data, scientists may also incorporate stochasticity—random variability over time—into the model.

Model selection is critical: an overly complicated model can reduce precision, while an overly simple model is not likely to be accurate. Before settling on a particular PVA model, it is best to examine a few potential models to gain understanding of the effects of different model structures on results.

Projection Length

PVA uses historical and current conditions to predict future outcomes; thus, the accuracy of PVA decreases with increasing extrapolated time. It is approximated that PVA projections are valid for 5-10% of the length of the dataset, and projection lengths of 50 or more years are unacceptable. In this sense, PVA has been compared to weather forecasting: data from a numerous sources is brought together to predict the likelihood of certain patterns, and cannot be accurately predicted far in advance because of environmental variability.

Sensitivity Analysis

Sensitivity analysis highlights which vital rates or parameters primarily affect model outcomes and which parameters have little impact on model results. Sensitivity analysis can highlight key factors for further examination and indicate the potential management strategy that will lead to the quickest recovery.

For example, a sensitivity analysis for loggerhead sea turtles (Caretta caretta) found that survivorship through the egg stage had little effect on population outcomes, while survivorship through the oceanic juvenile stage was critical to population health. Recovery efforts, which had been focused on egg survival on nesting-beaches, were redirected towards increasing juvenile survivorship by incorporating turtle-excluder devices (TEDs) into fishing nets.

Interpretation of Results

PVA validity depends on the appropriateness of model structure and quality; model quality can be improved through testing with independent field data. The accuracy and relevance of PVA increases when the model is regularly revisited and updated.

Examples

A PVA for the endangered Fender’s blue butterfly (Icaricia icarioides) was recently performed with a goal of providing additional information to the United States Fish and Wildlife Service, which was developing a recovery plan for the species. The PVA concluded that the species was more at risk of extinction than previously thought and identified key sites where recovery efforts should be focused. The PVA also indicated that because the butterfly populations fluctuate widely year-to-year, to prevent the populations from going extinct the minimum annual population growth rate must be kept much higher than at levels typically considered acceptable for other species.

Following a recent outbreak of canine distemper virus, a PVA was performed for the critically endangered island fox (Urocyon littoralis) of Santa Catalina Island, California. The Santa Catalina island fox population is uniquely composed of two subpopulations that are separated by an isthmus, with the eastern subpopulation at greater risk of extinction than the western subpopulation. PVA was conducted with the goals of 1) evaluating the island fox’s extinction risk, 2) estimating the island fox’s sensitivity to catastrophic events, and 3) evaluating recent recovery efforts which include release of captive-bred foxes and transport of wild juvenile foxes from the west to the east side. Results of the PVA concluded that the island fox is still at significant risk of extinction, and is highly susceptible to catastrophes that occur more than once every 20 years. Furthermore, extinction risks and future population sizes on both sides of the island were significantly dependent on the number of foxes released and transported each year.

Controversy

Debates exist and remain unresolved over the appropriate uses of PVA in conservation biology and PVA’s ability to accurately assess extinction risks.

An enormous amount of data is required for PVA— it is estimated that for a precise extinction probability assessment of t years, 5t -10t years of data are required. Datasets of such magnitude are typically unavailable for rare species; it has been estimated that suitable data for PVA is available for only 2% of threatened bird species. PVA for threatened and endangered species is particularly a problem as the predictive power of PVA plummets dramatically with minimal datasets. Ellner et al. (2002) argued that PVA has little value in such circumstances and is best replaced by other methods. Others argue that PVA remains the best tool available for estimations of extinction risk.

Even with an adequate dataset, it is possible that a PVA can still have large errors in extinction rate predictions. PVA is a scientific best guess of what will happen in the future, based on past and current conditions. It is impossible to incorporate all future possibilities into a PVA: habitats may change, catastrophes may occur, new diseases may be introduced. PVA is a model—an oversimplified representation of the world—and no model is 100% correct. For this reason, several argue that PVA is always best used in a relative sense, such as comparing proposed management plans against each other or comparing management proposals with current conditions.

Future Directions

Improvements to PVA likely to occur in the near future include: 1) creating a fixed definition of PVA and scientific standards of quality by which all PVA are judged, 2) developing reliable PVA for plants, and 3) incorporating recent genetic advances into PVA.

References

Beissinger, Steven R. and McCullough, Dale R. (2002). “Population Viability Analysis”, Chicago: University of Chicago Press

Gilpin, M.E. and Soulé, M.E. (1986). “Conservation biology: The Science of Scarcity and Diversity”, Sunderland, Massachusetts: Sinauer Associates

Perrins, C.M., Lebreton, J.D., and Hirons, G.J.M. (eds.) (1991). “Bird population studies: relevance to conservation and management”, New York: Oxford University Press

Gourt Home::Gourt :: Population viability analysis