AZGFD and Service: Improving Desert Bighorn Sheep Population Estimates and Sonoran Pronghorn Recovery

Regarding desert bighorn sheep, this project aims to:

1) Ensure that statistical calculations in present methods are correct.
2) Irrespective of the outcome of #1, perform an evaluation to improve survey methodology and data analysis to increase survey accuracy. This requires:
a) An investigation that quantifies and reports the relationship between count accuracy and effort, and presents a range of options informing agencies the tradeoffs between count accuracy and effort.
b) Provide information describing changes we can easily make, versus changes that would require significant retooling and investment. Knowledge of expected accuracy from given methods.
3) These agencies seek a robust method applicable to estimating desert bighorn sheep numbers range-wide. Consistency in methods among disparate populations when aggregated makes range-wide predictions more robust. Achieving this objective entails knowing tradeoffs between the available options to increase survey accuracy, effort and costs.
4) Examine historic count data and methods to determine if back calculations can be conducted to provide more accurate population estimates and more accurately reflect counts gained from present methods. If such analyses are appropriate, then perform those calculations.
5) Can biometric modeling allow us to project populations for the next year? If we could generate a robust method to forecast populations, it could mean scaling back surveys to every other year or every three years at a substantial cost savings. It would also allow us to set hunt permits more accurately. Scaling back surveys would cause a decrease in our abilities to quantify certain population parameters. We would need to know these tradeoffs.

For Sonoran pronghorn, this project will:

1. Develop and evaluate an empirical predictive model relating fawn recruitment to vegetation greenup and precipitation. Incorporate recent management actions (habitat management, supplemental feed and watering) into empirical models to estimate the effects of these interventions on enhancing fawn recruitment.

2. Assuming that a reasonably strong predictive relationship exists, imbed the faunal recruitment – environment / management model into a population simulation model based on the remaining assumptions and parameter values of the Hosack et al. (2002) model.

3. Use the simulation model to evaluate alternative management strategies given stochastic variation in rainfall. This is based initially on historic patterns, subsequently modified under assumed climate trends. Focus will be on identifying appropriate population and environmental triggers at which managers should provided supplemental forage and water, but will also investigate the impacts of habitat modification, reintroduction and translocation, and other actions.

4. Evaluate whether approaches as adaptive management can reduce uncertainty about the demographic rates an environmental influences determining SP viability in order to achieve recovery goals.

Hosack, DA, P.S. Miller, JJ Hervert, and RC Lacy. 2002. A population viability analysis for the endangered Sonoran pronghorn. Mammalia 66: 207-229.

The Service will contract for this work with M.J. Conroy Wildlife Biometrics.

Presently, a “group-size-density-estimator” method is in use and has been consistently conducted since 1997. The areas surveyed and the amount of effort in each survey is standardized. Radio-collared sheep were used to evaluate sighting rates for these surveys (Hervert et al. 1998).

Over recent years, both agencies have dedicated much effort to address the level of variation associated with such survey data. One problematic issue is that the average group size of the marked animals seen on surveys can be significantly larger than the average group size of marked animals missed on surveys. This difference can be explained by the observed probabilities for discrete group sizes. For example, observers miss more small groups than large groups. This group size bias can greatly influence the accuracy of the population estimate.

This bias can generate estimates with wide confidence intervals. This in turn challenges managers’ abilities to accurately determine population declines or increases in a statistically defensible framework.

Presently, survey data are used for (but not limited to) evaluating population trends, regional meta-analyses, setting harvest objectives, quantifying lamb/ewe ratios, and lambda. The successful applicant will need to understand the connection between survey methods and the management decisions made from the survey data now, and the population based information we hope to gain from survey data. Certainly, the biometrician must know what the survey results are used for in order to help tune the technique. Such discussions will be led by AZGFD.

Presently, a “group-size-density-estimator” method is in use and has been consistently conducted since 1997. The areas surveyed and the amount of effort in each survey is standardized. Radio-collared sheep were used to evaluate sighting rates for these surveys (Hervert et al. 1998).

Over recent years, both agencies have dedicated much effort to address the level of variation associated with such survey data. One problematic issue is that the average group size of the marked animals seen on surveys can be significantly larger than the average group size of marked animals missed on surveys. This difference can be explained by the observed probabilities for discrete group sizes. For example, observers miss more small groups than large groups. This group size bias can greatly influence the accuracy of the population estimate.

This bias can generate estimates with wide confidence intervals. This in turn challenges managers’ abilities to accurately determine population declines or increases in a statistically defensible framework.

Presently, survey data are used for (but not limited to) evaluating population trends, regional meta-analyses, setting harvest objectives, quantifying lamb/ewe ratios, and lambda. The successful applicant will need to understand the connection between survey methods and the management decisions made from the survey data now, and the population based information we hope to gain from survey data. Certainly, the biometrician must know what the survey results are used for in order to help tune the technique. Such discussions will be led by AZGFD.