Summary
1.The ubiquitous, positive abundance-occupancy relationship is of potential value to conservation and pest management because of the possibility of using it to predict species abundance from occupancy measures. 2. He & Gaston (2000a) developed a model, and a parameterization method, for the prediction of abundance from occupancy based on the negative binomial distribution. There are to date few empirical tests of either the estimation method or model. Here we conduct such a test in a field-based mesocosm experiment using a Drosophilidae assemblage associated with decaying fruit. 3. With individual (and groups of ) fruit as minimum mapping units, abundance estimates derived using the parameterization method of the He-Gaston model differed significantly from measured values, and were least accurate for the most abundant species. 4. Substitution of k -values corrected for species density in the model did not improve abundance predictions significantly. However, substitution of k -values calculated directly from the negative binomial distribution yielded highly accurate abundance predictions. 5. Although the distribution of fly species did not deviate significantly from the negative binomial distribution, and the finest possible minimum mapping units were used (individual fruit), the parameterization method in the He-Gaston model consistently underestimated the abundance of species in the assemblage because individuals were very highly aggregated within fruit. 6. Because of its potential importance, this model and parameterization method require further exploration at fine scales, commonly represented by individual habitat units, for highly aggregated species. The incorporation of spatially explicit information may provide a means of improving abundance predictions in this regard.