Modeling population dynamics and forecasting population sizes is a major focus of basic and applied ecological research. When applied to the management of rare species, it is critical that the estimates from population viability analyses (PVAs) have high accuracy and low uncertainty. Long-term data sets help to reduce uncertainty in estimates of PVAs, but are often difficult to collect. A possible solution to reducing uncertainty in these models is to include information from similar studies as prior data in a Bayesian demographic matrix model. The objective of this study was to evaluate the ability of prior probability distributions to provide more accurate and less uncertain PVAs of rare species. Bayesian demographic models specifying uninformed and informed priors were constructed for four rare plant species (Calochortus howellii, C. pulchellus, C. tiburonensis, and Pedicularis furbishiae) that were the focus of demographic studies in the 1980s. Informative priors were derived from demographic data collected on congeners of each species (C. obispoensis, C. albus, and P. lanceolata). The models were projected from the 1980s until 2009, and the originally studied populations of the four species were re-censused in 2009 to validate the estimates of population size and quasi-extinction risk generated by the models.  

Results/Conclusions

The inclusion of prior data decreased the variance around estimates of instantaneous growth rates (λ) and quasi-extinction risks. The inclusion of prior data did not, however, consistently increase the accuracy of most of the projections of population size and quasi-extinction risk. When biotic interactions or stochastic events not included in the models caused population declines, neither models with uninformed or informed priors accurately predicted the fate of a population. For instance, succession of nearby vegetation caused the population size of C. pulchellus to be much lower than predicted by models with or without informed prior probability distributions. Models specifying informed priors were also inaccurate when the populations studied for the prior specifications were not on the same demographic trajectories as the focal species. The results presented here show that while the inclusion of prior data decreased the variance around estimates from Bayesian PVAs, ecologists should take great care in specifying prior probability distributions with demographic models. If available the inclusion of prior data on the effects of biotic or stochastic processes on vital rates may be more informative. Further, prior probability distributions should ideally be specified from a congeneric population that is on a similar demographic trajectory as the focal population.