Population dynamics of a non-cultivated biennial plant Tragopogon pratensis infected by the autoecious demicyclic rust fungus Puccinia hysterium

Abstract

Population dynamics of the biennial plant Tragopogon pratensis have been monitored in the Park Grass Experiment at Rothamsted Research, Harpenden, UK, over many years. Observations of diseased T. pratensis, systemically infected by the autoecious demicyclic rust Puccinia hysterium, were made over the period 1995–2008, and confirmed an outbreak pattern of dynamics, characterised by an increase to a relatively high incidence followed by a reduction to low almost indiscernible levels. An epidemiological model was developed taking into account the biennial habit of the host plant, and the systemic nature of infection during the winter period, and the partial sterilisation of infected second year plants. Seedling emergence rate and natural mortality between seasons and within season were key parameters affecting host performance. The transmission rate between infected second year plants and susceptible first year seedlings, and the probability that the fungus would survive the winter systemically as mycelium producing aecia and telia on emerging second year plants, were key parameters associated with pathogenicity. Furthermore the possibility of pathogen-induced additional mortality was modelled. The model predicted that outbreak dynamics of T. pratensis would occur with high pathogenicity and medium or high host performance. In the former case the population dynamics would be cyclical with, in some cases, infected plants going to extinction. In the latter case both host and pathogen would go to extinction. The model predicted that the two pathogenicity parameters were critical in determining whether the pathogen would invade a healthy population; whereas pathogen-induced mortality had little influence, a result also obtained in some limited potted plant experiments. Fitting the model to the field data indicated that there was little or no density-dependence in seedling emergence rate, and again that pathogen-induced mortality played little role in the observed population dynamics.