Contributions of Covariance: Decomposing the components of stochastic population growth in the endangered Lady's Slipper orchid Cypripedium calceolus

Background/Question/Methods

Correlations between vital rates are common in nature and can have important effects on both evolution and demography. However, few studies have directly investigated the impact of vital rate correlations on population growth rates. Here, we extend Life Table Response Experiments (LTREs) to variable environments, showing how to precisely quantify contributions made by: (1) mean vital rates, (2) differential variability driven by environmental fluctuations, (3) correlations reflecting demographic tradeoffs and synchrony of stage transitions and (4) vital rate elasticities describing local selection pressures.  These new methods employ a singular value decomposition of the small noise approximation of the stochastic growth rate that apportions population-level differences in stochastic growth rates on the bases of vital rate covariance and elasticities.  We illustrate the value of our new decomposition by analyzing empirical data on the endangered Lady’s Slipper orchid Cypripedium calceolus. 

Results/Conclusions

We found that stochastic effects on the dynamics of these populations were driven largely by tradeoffs associated with adult dormancy transitions and that the strong negative correlations imposed by the algebra describing dormancy, emergence and stasis transitions resulted in conjugate pairing of stochastic contributions.  This resulted in near-stationary population growth rates despite observed differences in vital rates.  This confirms previous research that found that adult dormancy was important for individual survival and population persistence of this species and that differences in dormancy rates were an important driver of population performance.  Our results advance ecological and evolutionary understanding about the role vital rate covariance plays in stochastic population dynamics, in particular for this endangered species.  Because some of this information is unavailable using previous methods, our new decomposition should therefore be broadly useful for applications in basic ecology and conservation biology.