An open‐population hierarchical distance sampling model

Abstract

Modeling population dynamics while accounting for imperfect detection is essential to monitoring programs. Distance sampling allows estimating population size while accounting for imperfect detection, but existing methods do not allow for estimation of demographic parameters. We develop a model that uses temporal correlation in abundance arising from underlying population dynamics to estimate demographic parameters from repeated distance sampling surveys. Using a simulation study motivated by designing a monitoring program for Island Scrub-Jays (Aphelocoma insularis), we investigated the power of this model to detect population trends. We generated temporally autocorrelated abundance and distance sampling data over six surveys, using population rates of change of 0.95 and 0.90. We fit the data generating Markovian model and a mis-specified model with a log-linear time effect on abundance, and derived post hoc trend estimates from a model estimating abundance for each survey separately. We performed these analyses for varying numbers of survey points. Power to detect population changes was consistently greater under the Markov model than under the alternatives, particularly for reduced numbers of survey points. The model can readily be extended to more complex demographic processes than considered in our simulations. This novel framework can be widely adopted for wildlife population monitoring.

An open‐population distance sampling framework for assessing population dynamics in group‐dwelling species

“…These results indicated that our extension of the Sollmann et al. () model to include groups, group composition, group aggregation and non‐repeated transects did not induce bias and had comparable performance characteristics to the results presented by Sollmann et al. ().…”

“…Our results indicated that our expanded formulation has similar properties to the Sollmann et al. () version, and there is no evidence that the additional structure causes confounding. Credible interval coverage was slightly lower in our case, but considering we used only 50 transects and had several additional parameters to estimate, this is not surprising.…”

Integrating distance sampling with minimum counts to improve monitoring

“…The resulting integrated population model could improve precision and assessment of other demographic parameters would become possible. Although parameters representing population dynamics are estimable from repeated survey data alone (Sollmann et al , Schmidt and Rattenbury ), auxiliary data allows additional parameters (e.g., survival, recruitment) to be estimated (Schmidt et al ), potentially leading to an understanding of population drivers (Schaub et al , Abadi et al ). For these reasons we expect the basic structure of our model to be useful in extracting more information from disparate datasets, thereby improving wildlife monitoring and management outcomes.…”

Estimating demographic parameters using a combination of known‐fate and open N‐mixture models

“…The stagestructured model of Zipkin et al (2014b) may also benefit from the inclusion of known-fate data when available. It is also possible that other models could be used in place of the known-fate model, with similar benefits for other types of studies (e.g., distance sampling; Sollmann et al 2015). The integration of mark-recapture or distance sampling models with open N-mixture models should provide opportunities to improve ecological inference in a variety of settings and will help to increase our knowledge of population dynamics.…”