Using integrated population models to improve conservation monitoring: California spotted owls as a case study

Integrated population models (IPMs) constitute a relatively new approach for estimating population trends and demographic parameters that makes use of multiple, independent data sources (e.g., count and mark-recapture data) within a unified statistical framework. In principle, IPMs offer several advantages over more conventional modeling approaches that rely on a single source of data, including greater precision in parameter estimates and the ability to estimate demographic parameters for which no explicit data are available. However, to date, the IPM literature has focused primarily on model development and evaluation, and few “real-world” applications have demonstrated that IPMs can strengthen inferences about population dynamics in a species of conservation concern. Here, we combined 23 years of count, occupancy, reproductive, and mark-recapture data into an IPM framework to estimate population trends and demographic rates in a population of California spotted owls (Strix occidentalis occidentalis). Using this framework, we observed a significant population decline, as evidenced by the geometric mean of the finite annual rate of population change (λt¯ˆ=0.969, 95% CRI 0.957–0.980) and the resulting realized population change (proportion of the initial population present in 2012; Δˆ2012=0.501, 95% CRI 0.383–0.641). The estimated decline was considerably greater than the approximately 30% decline estimated using conventional mark-recapture and occupancy approaches (Tempel and Gutiérrez, 2013). The IPM likely yielded a greater decline because it allowed for the inclusion of three years of data from the beginning of the study that were omitted from previous analyses to meet the assumptions of mark-recapture models. The IPM may also have yielded a greater estimate of decline than occupancy models owing to an increase in the number of territories occupied by single owls over the study period. All demographic parameters (adult and juvenile apparent survival, reproductive rate, immigration rate) were positively correlated with λtˆ, but immigration was fairly high (immtˆ=0.097, 95% CRI 0.055–0.140) and contributed most to temporal variation in λtˆ, suggesting that changes in owl abundance were influenced by processes occurring outside of our study area. More broadly, our results indicated that the IPM framework has the potential to strengthen inference in population monitoring and demographic studies, particularly for those involving long-lived species whose abundance may be slowly declining. In our case, the conservation implications from the results of the IPM suggested a decline in the population of owls that was steeper than previously thought.