Addressing uncertainty in extinction risk due to projection length

Michelle DePrenger-Levin, Department of Research, Denver Botanic Gardens, Denver, CO; Department of Integrative Biology, University of Colorado Denver, Denver, CO and Michael Kunz, North Carolina Botanical Garden, Chapel Hill, NC; Environment, Ecology and Energy Program, University of North Carolina at Chapel Hill, Chapel Hill, NC

Background/Question/Methods
Lefkovitch matrix models are widely used to predict plant population dynamics and estimate risk from climate change, habitat alterations, and management actions. The change in extinction risk or relative risk among and within species, across projected climate conditions, or due to disturbance regimes informs listing decisions (IUCN or ESA) and management actions. The time point at which to estimate extinction risk can be an arbitrary decision meant to match the time scale of a disturbance regime or duration of a conservation action. The same projection length will cover several generations of a fast species and fewer or only a fraction of a generation of a long-lived species. Projection lengths can be scaled by pace of life (i.e. generation time) but may still lead to biased extinction risk estimates. Confidence intervals that integrate observation, process, or model error and uncertainty are rarely included in extinction risk estimates. The resulting uncertainty can cause misleading multi-species comparisons. In this study, we create virtual matrix models that reflect pace of life (fast to slow) and reproductive strategy (iteroparity to semelparity) for stochastic simulations to examine the impact of life history traits and population size on uncertainty of extinction risk.
Results/Conclusions
results. Life history traits of plants (pace and parity) change the speed and variability of population fluctuations and the length of time over which extinctions are expected. Across population sizes typical of a species of conservation concern, bias across life history traits will be reduced in very small populations (10 individuals) where the chance of extinction is high whereas at intermediate sizes (50, 100, or 500 individuals), slow pace of life tends to increase the variability in time to extinction while variation among semelparous simulations was greater than iteroparous. Multi-species or population comparisons projected over the same project period will detect a larger proportion of expected extinctions for short-lived species than longer-lived species. Scaling projection lengths by generation time may not reduce this bias. Instead, quantifying the variability in time to extinction across life history traits should be used for more robust extinction risk comparisons.