Idiosyncratic relationships between pond invertebrates and environmental, temporal and patch-specific predictors of incidence

The incidence of eight species of microcrustacea and seven species of Dytiscidae (adults) was recorded from 30 small, temporary pools, in January and early summer, over six years. Incidence of each species, both in January and in summer, was modelled using logistic regression. Three categories of predictor variables were included in the models. Firstly systematic environmental variables; length of previous summer's dry phase, inter-pond links caused by flooding, distance between ponds and macrophyte structural density. Secondly temporal factors; year and previous incidence. Thirdly individual ponds, representing patch-specific effects. Logistic regression provided effective models of incidence for most species. However individual species varied in the factors that provided significant prediction of presence and absence. Systematic, temporal and patch-specific predictors were all significant predictors for different species of both microcrustacea and Dytiscidae. No single factor dominated most models. Individual factors showed positive or negative relationships with incidence with different species. Length of previous summer dry-phase was the most common predictor for microcrustacea and beetles in January. Summer models were more varied; length of previous dry phase and flood linkage between ponds and macrophyte density were significant predictors in many models, year and distance less important. Patch-specific predictors were significant for five of the microcrustacea and four of the beetles. The results suggest that individual species show idiosyncratic responses to systematic, temporal and patch-specific factors, rather than most species responding in a similar way to one or two dominant influences.