Using causal inference to detect a climate change signal on infectious tree disease

Andrew M. Latimer, Plant Sciences, University of California Davis, Davis, CA and Joan Dudney, Department of Plant Sciences, University of California, Davis, Davis, CA

Background/Question/Methods
Though climate change is predicted to cause major shifts in infectious disease risk, definitive evidence is often elusive due to data limitations and confounding factors. Thus, disease outbreaks are often interpreted as stochastic events, rather than a response to changing environmental conditions. Nonlinear stochastic events, such as droughts, are also predicted to increase in frequency and severity throughout various parts of the globe and negatively affect fungal pathogens. Here we take advantage of a unique long-term dataset (two survey periods spanning ~19 years; over 8,000 individual hosts) of the fungal tree disease, white pine blister rust (Cronartium ribicola Fisch., blister rust). We predicted that blister rust was nonlinearly related to climate. Due to this nonlinear relationship, we expected that climate change over the past nineteen years shifted blister rust nonlinearly, increasing infections in colder regions and decreasing infections in hotter, drier regions. We hypothesized that a mechanism driving the nonlinear range shift was an interaction between increased water stress and blister rust infections.
Results/Conclusions
Using a first differences panel modeling approach, we found evidence of nonlinear responses of disease spread to rising temperatures. This nonlinear effect increased infections in colder climates and may have decreased new infections in the hottest, driest conditions. We demonstrated that there was a significant interaction between disease and water stress in infected hosts, which may be contributing to nonlinear shifts in pathogen infections. Though many studies predict nonlinear climate effects on disease spread, we present some of the first empirical evidence of this relationship. Our results underscore the importance of quantifying nonlinearities in climate–disease interactions to improve predictions of disease outbreaks.