Optimizing demographic analysis in the face of missing data years to improve conservation of threatened species

Quantification of population dynamics and predictions of species viability rely on estimates of vital rates and an understanding of the ecological drivers of these rates. Most standard methods for assessing impacts of drivers, such as climate, on vital rates require annual demographic data for many individuals over multiple years. However, many real studies have either planned or unplanned data gaps. Vital rates are usually estimated over annual transitions, therefore one year of missing data results in two missing estimates. Additionally, relating annual climate variation to changes in vital rates is challenging if studies do not collect data annually. To avoid this loss of information due to missing data, we developed and then tested a Bayesian modeling approach for a dataset with missing years. Using an 18-year study of the rare plant Eriogonum brandegeei we estimate vital rates, their relationship to annual climate, and stochastic population growth. By comparing model performance across data subsets, as well as in tests using simulated data, we find that the approach works well with missing years of demographic data and removes the need to ignore information from multi-year transitions. This generalizable approach increases the useability of available data to study species dynamics despite patchy demographic data.