Optimizing the use of demographic data for viability analysis of threatened species when there are planned or unplanned missing data: The case of Eriogonum brandegeei

Background/Question/Methods

As global environmental change accelerates, conserving threatened populations is increasingly challenging and urgent. Small, isolated populations are particularly vulnerable to extirpation from genetic, environmental, and demographic stochasticity. Assessing the viability of small populations requires understanding the drivers of population dynamics. The typical approach for estimating effects of drivers on population dynamics requires detailed demographic data for many individuals every year for considerable periods of time. However, collecting individual demographic data is time and labor intensive, often resulting in datasets containing missing years of data due to anticipated and unanticipated constraints. This leads to a serious loss of information since predictions of population growth are made for annual transitions, and one missing year of data results in two missing transitions. To address this problem, we develop and apply an approach to include multi-year gaps in the modeling of vital rates and their dependence on annual climate variables. We use a Bayesian method to estimate vital rates and incorporate parameter uncertainty into downstream analyses. We integrate this into an integral projection modelling framework and apply these methods to a 15-year study of the rare, endemic plant Eriogonum brandegeei.

Results/Conclusions

Using this demographic dataset with missing years of data, we demonstrate the performance of our analysis approach for estimating vital rate parameters, including the dependence of vital rates on multiple time-varying climate drivers. We find that this approach can successfully assess current and predict future population viability from data with multi-year gaps. We then use this approach to assess the viability and population dynamics of E. brandegeei. Specifically, we estimate deterministic population growth rates, stochastic population growth, and risk of quasi-extinction. Finally, we investigate how population dynamics vary across small spatial scales within populations, and the importance of different climate variables and vital rates in driving dynamics. Our results show that the focal population of this species is at risk of extirpation, but population dynamics vary considerably across small spatial scales within the population. This approach is readily adaptable to other species and can improve the use of long-term datasets in conservation and population ecology, avoiding the winnowing of usable data due to multi-year gaps in data collection.