The transmission dynamics of the coffee rust disease: Multiple processes and critical transitions

Kevin Li, School for Enviroment and Sustainability, University of Michigan, Ann Arbor, MI, Zachary Hajian-Forooshani, Minh Chau N. Ho and John H. Vandermeer, Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI, Chenyang Su and Ivette Perfecto, School for Environment and Sustainability, University of Michigan, Ann Arbor, MI

Background/Question/Methods
The coffee leaf rust disease (CLR), caused by the fungus Hemileia vastatrix, had an economically devastating outbreak in 2012-2013, threatening the livelihoods of coffee farmers and coffee production in Mesoamerica and the Caribbean. Infection by CLR spores may arise from either local transmission, through physical contact between plants and spreading from rain splash or local air turbulence; or from long-range dispersal, through wind carrying spores over a wider region. These modes of transmission may be affected by long-term climate change, which is changing rainfall patterns, and shade tree management, which may mediate spore dispersal. We examined data on CLR intensity and transmission in monthly surveys of sentinel coffee plants in 128 sampling quadrats on a 45-ha plot in southern Mexico from 2013 to 2020. We tested for patterns of initial infection that we hypothesized would be characteristic of the two modes of transmission. Within a subset period, we examined the moderating effect of shade tree cover on infection. The two scales of transmission processes suggest the theoretical possibility of a critical transition, or a sudden change in response to slow shifts in environmental drivers, which we observed in the initial onset and eventual collapse of the disease. We examined the seasonal patterns of the longitudinal data for indicators of “critical slowing down” that can anticipate such a transition.
Results/Conclusions
The pattern of initial CLR infections suggests that splash plays a central role in the establishment of an infection, as initial infections tended to begin at the bottom of the coffee plant but shifted towards the middle with more rainfall. Shade tree cover over coffee bushes appears to mediate the effects of rain by decreasing relative infection probability with high rainfall. These patterns support the importance of splash transmission processes in the dynamics of the disease. Within the long-term seasonal dynamics, we find evidence of critical slowing down in the disease initiation timing and deceleration to peak infection, which precede the critical transition from a persistent seasonal epidemic to a benign post-outbreak state after 2019. These signals were correlated with reduced rainfall, perhaps induced by climate change, and a region-wide shift in management that increased the proportion of rust-resistant coffee varieties. More generally, critical transitions may not be uncommon in socio-ecological systems. Early warning signals could be useful for evaluating not only the risk of a critical transition, but also the mitigation effectiveness following an initial transition.