Spatial-temporal heterogeneity of a wildland forest epidemic: Multilevel effects of individual, community and landscape drivers

Background/Question/Methods

The dynamic and inherently spatial nature of epidemiological processes presents unique challenges to managing the spread of emerging infectious diseases in natural communities. Long-term ecological studies that combine data across multiple scales of host-pathogen-environment interactions are needed to identity the principal drivers of wildland epidemics and their generality across pathosystems. However, this is not a trivial task given the logistical challenges of conducting longitudinal field surveys of host populations with measurements of corresponding biotic and abiotic conditions across broad geographical regions and over meaningful time scales. As such, we still have a limited understanding of the factors governing the distribution, abundance and impacts of emerging wildland pathogens. We designed a long-term study to capture spatial-temporal heterogeneity of wildland disease dynamics across multiple levels of host-pathogen-environment interactions (i.e., at the individual, community and landscape scale), using the emerging forest disease sudden oak death as a case study. We applied a novel survival analysis framework that incorporates multilevel, time-dependent properties of this complex pathosystem to address the following questions: (1) which scale of host-pathogen-environment interactions most strongly governs time to infection in susceptible oaks, and (2) is their evidence for temporal lag effects of climate variability on epidemic dynamics?

Results/Conclusions

We annually monitored 1125 oak hosts—across 172 plots—for P. ramorum infection and disease-induced mortality over an 8-year period. Coast live oak exhibited a steady increase in the number of trees becoming infected (n=151) and dying following infection (n=58), signaling the potential for an emerging forest disease to dramatically change oak woodlands through selective removal of a keystone species. To a lesser extent, the number of black oak becoming infected (n=35) and then dying (n=8) also increased over time. After accounting for plot-level frailties (i.e., random effects), the survival analyses revealed the collective importance of factors at the individual, community and landscape scale in mediating oak infection risk. We found that larger coast live oak are particularly susceptible to infection, and that warmer and wetter conditions increase infection risk, while species diversity has a buffering effect against disease. By measuring fluctuating environmental conditions over time, we found a lag effect of climatic variability, whereby above average precipitation and temperature during the current and previous years increased infection risk. Our results demonstrate how spatial and temporal heterogeneity in environmental conditions modulate disease impacts in a wildland pathosystem, and provide an empirical basis for assessing the risk of extinction among co-occurring hosts in pathogen-invaded communities.