PARENT SESSION
Oral Session # 16: Ecological Modeling I.
Presiding: C Ray
Tuesday, August 5. 8:00 AM to 11:30 AM, SITCC Meeting Room 102.
Does correlation among vital rates matter? The effect of correlation structure on estimates of population viability.
Kaye, Thomas*,1, Pyke, David, 1 Institute for Applied Ecology, Corvallis, OR
ABSTRACT- Transition matrix models are one of the most widely used tools for assessing population viability. Demographic parameters may be correlated across years and environments and inclusion of this correlation structure in stochastic models may be necessary to avoid overly optimistic estimates of population viability. However, negative correlations among vital rates are also possible, and these tend to counteract the effects of positive correlations. Therefore, the effect of correlation structure on population viability estimates may depend on the nature of the correlations among vital rates, which, in turn, may differ among species and environments. We used empirically derived data from 27 populations of five perennial plant species collected over a span of five to ten years to examine the effect of correlation among transition elements on population viability estimates. We also compared different methods of including stochasticity and checked for interactions between stochastic method and correlation structure. For each method, we estimated stochastic population growth rate (a measure of viability) as our response variable. Temporal correlation among vital rates in our stochastic matrix models altered estimates of population viability, but this effect differed among species and was generally weak. The magnitude of change in estimated stochastic growth rate for each species examined here was largely explained by the ratio of positive to negative cross-correlations of transition matrix elements; as the relative number of negative correlations decreased, the impact of correlation structure increased. When deciding whether or not to include such correlation structure in viability models, conservationists may want to examine the correlations in their species as a means of assessing their anticipated effect, and we provide a preliminary linear regression model for doing so. These results are applicable to a range of perennial plants and possibly other life histories.
Key words: correlation structure, stochastic, matrix model, population viability analysis