Short-term costs may be long-term benefits: Using stochastic demographic models to understand a complex symbiosis

Background/Question/Methods

Interactions between endosymbionts and their hosts are highly variable and often context dependent. Endosymbionts can significantly alter host demographic rates and thereby influence population dynamics. Additionally, costs or benefits from the symbiosis can be magnified or dampened by inter-annual environmental variation (e.g., climate). Effects of symbionts on host responses to inter-annual variability should have consequences at the population level, since variance in demography generally reduces the long-term population growth rate (in a geometric series, a bad year is more costly than an equally good year is beneficial). We collected demographic data from experimental plots of a cool-season grasses that naturally hosts fungal endosymbionts (endophytes), Poa autumnalis. We compared demographic rates between hosts that were naturally symbiotic and hosts from which the endophyte was experimentally eliminated. We asked whether fungal endophytes altered their host demography, and specifically how these changes affect long term population growth.

Results/Conclusions

We found that endophytes do alter host demographic rates. There was support for models including endophyte effects on host survival, growth and reproduction, however, the effect was not consistent across demographic rates. Interestingly we observed effects on the mean, effects on the variance, and effects on both the mean and variance for different vital rates. These results are impossible to interpret independently so we implemented an integral projection model to synthesize the results and examine the effect of endophytes on their hosts’ population dynamics.  The IPM’s showed that endophytes affect the population dynamics of their grass hosts via effects on both mean vital rates as well as variance in vital rates. Our study demonstrates the important effects that symbionts can have on host population dynamics in temporally fluctuating environments.