Accurately Estimating Correlations Between Demographic Parameters: A Response to Riecke Et al. (2024).

Area of Science:

• Ecology
• Population Dynamics
• Wildlife Management

Background:

• Tag-recovery data are crucial for understanding exploited populations' demographic responses to harvest.
• Previous studies have explored bias and certainty in correlation parameters between recovery and survival probabilities.
• Refinements in prior distributions for parameterizing tag-recovery models have been proposed.

Purpose of the Study:

• To evaluate the impact of different prior distributions (Gamma(1,1) vs. Uniform(0,5)) on estimating correlations between recovery and survival probabilities.
• To assess parameter recovery and inference reliability across varying sample sizes in tag-recovery models.
• To compare discrete-time and hazard-rate parameterizations for estimating cause-specific mortality.

Main Methods:

• Fit tag-recovery models to simulated data previously used by Deane et al. (2023).
• Employed Gamma(1,1) as the prior distribution for standard deviations, comparing it with Uniform(0,5).
• Estimated recovery and survival in discrete time and cause-specific mortality as hazard rates.

Main Results:

• With large sample sizes, both prior distributions yielded similar, reliable parameter recovery and inference.
• Small sample sizes, typical for North American duck populations, resulted in uncertain and ambiguous correlation estimates.
• Decreasing sample sizes increased uncertainty in annual survival estimates relative to recovery, hindering correlation estimation.

Conclusions:

• Correlation estimates from tag-recovery models are highly uncertain with small sample sizes.
• The sign of the correlation may be the only reliably interpretable attribute under data limitations.
• Findings have implications for wildlife management and the interpretation of demographic responses to harvest.