Modeling forest disease using a macroparasite framework
Forest pathogens are typically modeled using variants of SI models, in which the infection states of trees are binary. However, many forest pathogens only infect trees locally, in isolated bole infections or leaf infections. Individual trees can have multiple, independently reproducing infections. In these cases, models based on macroparasite frameworks, where multiple parasites occupy an organism, may better represent disease dynamics.
I developed a model for the host-pathogen dynamics of Sudden Oak Death (SOD), a disease caused by the invasive pathogen Phytophthora ramorum in California oak woodlands and forests, based on the macroparasite framework of Anderson and May (1978). I compared the dynamics of this model to an SI model, parameterizing both using data from SOD field studies, and examined the conditions under which each could reproduce patterns in field data.
The macroparasite model reproduces several patterns observed in field data which the SI model does not. First, the macroparasite model, even without age-related variation in parameters, generates data with positive age-mortality relationships due to accumulating infections in older trees. The SI model requires explicit age structure in model parameters to produce similar patterns. Secondly, the macroparasite model generates observed density-mortality patterns, which can not be produced by the SI model. Thirdly, the macroparasite model generates aggregated distribution of infections in hosts, which can explain the phenomenon of "super spreaders", i.e. individuals that produce high amounts of inoculum and cause a large fraction of new infections. The SI model only generates this pattern when extended by including individual variation in disease parameters. All these phenomena are more pronounced under conditions corresponding to late stages of disease invasion, and when the time scales of host and pathogen life stage parameters are similar.
The macroparasite framework is a parsimonious approach that can explain several important phenomena in forest disease. Data and models to with only binary infection status can lead to misinterpretation of observed patterns.
Slides and related materials will be archived at http://doi.org/rpm prior to the conference.