A Stochastic Model Incorporating the Effect of Weather Conditions on Anthracnose Development in Stylosanthes scabra

Spatial and temporal progress of anthracnose caused by Colletotrichum gloeosporioides in quantitatively resistant accessions of the tropical pasture legume Stylosanthes scabra were studied in a field experiment at the Southedge Research Station, Queensland, Australia. In a previously published work a conditional ordinal logistic regression model was developed to explain the probability of a plant developing a given disease severity level, depending on its previous disease state and that of its neighbours. In the present study this model is augmented to incorporate the effects of three weather variables which were measured daily during a growing season. Two approaches were used: (a) threshold values for relative humidity (RH), rainfall and net evaporation were used to classify days as suitable or unsuitable for anthracnose growth; (b) days were assumed to vary continuously in their rate of anthracnose growth depending on the numerical values of the weather variables. High 9am RH, low net evaporation and low 9am temperatures are significantly associated with anthracnose growth. Net evaporation proved to be a better index than rainfall and heavy rainfall was not conducive to high levels of anthracnose; however, rainfall was useful once evaporation was taken into account. The effect of 9am RH can be described either by a threshold value around 70% or by a quadratic function. A two‐variable model with net evaporation and log(rain+1) explains 97.6% of the available deviance.