Combining ground count, telemetry, and mark-resight data to infer population dynamics in an endangered species.
Abstract
To successfully manipulate populations for management and conservation purposes, managers must be able to track changes in demographic rates and determine the factors driving spatial and temporal variation in those rates. For populations of management concern, however, data deficiencies frequently limit the use of traditional statistical methods for such analyses. Long-term demographic data are often piecemeal, having small sample sizes, inconsistent methodologies, intermittent data, and information on only a subset of important parameters and covariates. We evaluated the effectiveness of Bayesian state-space models for meeting these data limitations in elucidating dynamics of federally endangered Sierra Nevada bighorn sheep Ovis canadensis sierrae. We combined ground count, telemetry, and mark-resight data to: (1) estimate demographic parameters in three populations (including stage-specific abundances and vital rates); and (2) determine whether density, summer precipitation, or winter severity were driving variation in key demographic rates. Models combining all existing data types increased the precision and accuracy in parameter estimates and fit covariates to vital rates driving population performance. They also provided estimates for all years of interest (including years in which field data were not collected) and standardized the error structure across data types. Demographic rates indicated that recovery efforts should focus on increasing adult and yearling survival in the smallest bighorn sheep population. In evaluating covariates we found evidence of negative density dependence in the larger herds, but a trend of positive density dependence in the smallest herd suggesting that an augmentation may be needed to boost performance. We also found that vital rates in all populations were positively associated with summer precipitation, but that winter severity only had a negative effect on the smallest herd, the herd most strongly impacted by environmental stochasticity. Synthesis and applications. For populations with piecemeal data, a problem common to both endangered and harvested species, obtaining precise demographic parameter estimates is one of the greatest challenges in detecting population trends, diagnosing the causes of decline, and directing management. Data on Sierra Nevada bighorn sheep provide an example of the application of Bayesian state-space models for combining all existing data to meet these objectives and better inform important management and conservation decisions.