Ecosystem transformation by emerging infectious disease: loss of large tanoak from California forests

Summary 1.  Few pathogens are the sole or primary cause of species extinctions, but forest disease has caused spectacular declines in North American overstorey trees and restructured forest ecosystems at large spatial scales over the past 100 years. These events threaten biodiversity associated with impacted host trees and other resources valued by human societies even when they do not directly cause host extinction. 2.  Invasion of Phytophthora ramorum and emergence of the forest disease sudden oak death has caused a large‐scale decline of tanoak ( Notholithocarpus densiflorus ) in Californian coastal forests. Here, we describe structural changes to tanoak forests and develop predictive models of infection rates, mortality rates and changes in tanoak biomass and abundance by combining regionally extensive longitudinal field studies and mathematical modelling. 3.  Pathogen‐invaded stands had smaller average tanoak tree size and higher proportions of large dead tanoak trees compared with uninvaded stands. This pattern is caused in part by a positive relationship between tanoak size and mortality rate, as well as prolific basal sprouting from trees killed by the disease. Tanoak infection, mortality and biomass decline rates were positively related to the prevalence of infection in sporulation‐supporting species, especially California bay laurel ( Umbellularia californica ). 4.  We developed a stage‐structured and spatially explicit mathematical model including species dynamics and P. ramorum transmission, where the long‐term outcome of disease ranges from host extinction when densities of bay laurel are high to limited or no disease outbreak. Low densities of tanoak in a matrix of non‐susceptible neighbouring species resulted in slow‐enough transmission to retain overstorey tanoak, suggesting host‐density thresholds may exist in real forests. 5.Synthesis . Tanoak is likely to persist in many disease‐impacted forests via vegetative reproduction, but overstorey trees may be eliminated or greatly reduced in abundance, a pattern similar to other forest diseases that have emerged in the last century including chestnut blight and beech bark disease. Our results support a general model of disease‐caused changes to forest trees useful for the analysis of emerging forest pathogens where vegetative reproduction, community‐level epidemiology and stage‐specific mortality rate interact to determine local disease intensity and host decline.