How Good are {PVA} Models? testing their predictions with experimental data on the brine shrimp

Abstract

We employed replicate laboratory populations of bring shrimp {(Artemia} fanciscana) to develop a data set on extinction dynamics {(Belovsky} et al. 1999). Sets of replicate populations were established with different initial numbers of adults and with different carrying capacities (food supply rates) that were randomly varied each day by different coefficients of variation. Mean population persistence time and its variance were measured for each set of replicate populations. The experiment examined the action of demographic and environmental stochasticity on extinction.An additional experiment with replicated laboratory populations measured bring shrimp demography (birth, death and population growth rates) and non-linear (oscillatory) dynamics under the same conditions as the extinction experiment {(Belovsky} et al. 1999). Strong density dependent population dynamics were observed: birth rate increased, death rate decreased and population growth rate increased as population number relative to carrying capacity decreased. Furthermore, brine shrimp population growth rates should produce intrinsic population oscillations (nonlinear dynamics).Mean persistence time for sets of replicate populations: (1) increased with carrying capacity; (2) decreased with variability in population numbers; and (3) increased with initial population number. Population variability: (1) increased with variation in carrying capacity; (2) increased with the intensity of intrinsic oscillations in the population; and (3) increased as the ratio of starting populations numbers to carrying ca pity increased. The effect of initial population number on persistence was not as great as often hypothesized. Furthermore, variance in mean persistence time was much smaller than assumed. Our findings support qualitative perceptions about extinction and pose additional conservation concerns.Finally, experimental demographic data were used in a variety of population viability {(PVA)} models to predict mean population persistence time and its variance, and predictions were compared with the results from the extinction experiments. Many commonly used {PVA} models provided predictions that deviated considerably from experimental results, but others were very predictive. We address the reliability of commonly employed {PVA} models and their utility for conservation planning.