Issues affecting the measurement of disturbance response patterns in herbaceous vegetation - A test of the intermediate disturbance hypothesis

The 'intermediate disturbance hypothesis' (IDH) predicts maximum species diversity at intermediate levels of disturbance. Disturbance levels are measured by intensity or frequency of disturbance or by the time since disturbance. The IDH has been tested over a wide range of studies and communities with results either accepting or rejecting the hypothesis. The objectives of this study were to determine how observed disturbance response patterns for the same herbaceous plant community are influenced by modelling techniques, particularly in relation to the expression of disturbance gradients, choice of species diversity indices, and time of assessment since disturbance. Response patterns were examined using a multi-factorial disturbance experiment involving grazing (absent and present); fire (absent and present); soil cultivation (none, 5 and 20 cm); and amendment (none, fertiliser, and fertiliser plus clover seeds). Generalized linear models with a b-spline function were used to define response patterns for five different disturbance gradients over a 24-month period using three indices of diversity. Seven basic disturbance response models were recognized ranging from the classical IDH bell-shaped pattern through increasing or decreasing trends, to a no-change model with increasing levels of disturbance. Only 6.7% of all models were consistent with the IDH. The no-change model was found in nearly half the cases investigated and increased in occurrence with time since disturbance. The choice of the disturbance intensity gradient, the species diversity index used and the time of assessment after disturbance significantly influenced the frequency of occurrence of the disturbance response models observed. Consequently, the responses of vegetation to disturbance gradients show many patterns depending on how they are defined and modelled rather than the simple bell-shaped curve as predicted by the IDH, but the ecological mechanisms supporting some of these patterns need to be further investigated.