Modelling of early infection of cereal roots by the take-all fungus: a detailed mechanistic simulator.

Abstract

A stochastic model is constructed to simulate the spatial and temporal spread of infection of the take-all fungus, Gaeumannomyces graminis var. tritici (Sacc.) Arx & Olivier var. tritici Walker on seminal roots of wheat. The model is designed to synthesize information on the dynamics and spatial orientation of the growth of main seminal root axes of wheat and the dynamics of primary and secondary infection of the pathogen. Primary infection is initiated by the soil inoculum. Three types of secondary infection by runner hyphal growth are distinguished; re-infection of the same root, infection of another root on the same plant via the crown, and cross-infection between roots on different plants. There are nine pathogen parameters, 14 host parameters, as well as four system parameters in addition to the location and orientation of seedlings. The pathogen parameters comprise estimates for the size and density of inoculum, the rate of growth of the fungus on roots, and the mean distances and probability of occurrence for primary and secondary infection. The host parameters concern orientation, density, emergence, rates of growth and size of roots. The principal output variables are total and infected root length, numbers of infections, proportion of infected roots and the numbers of primary and cross infections. Results of sensitivity analysis of the output variables to selected input parameters are presented. The model is tested against independent data-sets from inoculum-density experiments for different soil temperatures and ranges of inoculum density. Statistical methods of response curve analysis are used to compare the behaviour of inoculum density-disease response curves for the simulated and experimental data. The model fitted the data satisfactorily for the majority of host and infection variables. Inclusion of secondary infection in the model improved the goodness-of-fit but the density of secondary infections was small relative to primary infections. Practical and conceptual problems in the validation of complex simulation models for fungal infection are discussed. The advantages and limitations of this and related models are critically assessed.