Selecting discriminant function models for predicting the expected richness of aquatic macroinvertebrates

1. The predictive modelling approach to bioassessment estimates the macroinvertebrate assemblage expected at a stream site if it were in a minimally disturbed reference condition. The difference between expected and observed assemblages then measures the departure of the site from reference condition. 2. Most predictive models employ site classification, followed by discriminant function (DF) modelling, to predict the expected assemblage from a suite of environmental variables. Stepwise DF analysis is normally used to choose a single subset of DF predictor variables with a high accuracy for classifying sites. An alternative is to screen all possible combinations of predictor variables, in order to identify several ‘best’ subsets that yield good overall performance of the predictive model. 3. We applied best‐subsets DF analysis to assemblage and environmental data from 199 reference sites in Oregon, U.S.A. Two sets of 66 best DF models containing between one and 14 predictor variables (that is, having model orders from one to 14) were developed, for five‐group and 11‐group site classifications. 4. Resubstitution classification accuracy of the DF models increased consistently with model order, but cross‐validated classification accuracy did not improve beyond seventh or eighth‐order models, suggesting that the larger models were overfitted. 5. Overall predictive model performance at model training sites, measured by the root‐mean‐squared error of the observed/expected species richness ratio, also improved steadily with DF model order. But high‐order DF models usually performed poorly at an independent set of validation sites, another sign of model overfitting. 6. Models selected by stepwise DF analysis showed evidence of overfitting and were outperformed by several of the best‐subsets models. 7. The group separation strength of a DF model, as measured by Wilks’Λ, was more strongly correlated with overall predictive model performance at training sites than was DF classification accuracy. 8. Our results suggest improved strategies for developing reliable, parsimonious predictive models. We emphasise the value of independent validation data for obtaining a realistic picture of model performance. We also recommend assessing not just one or two, but several, candidate models based on their overall performance as well as the performance of their DF component. 9. We provide links to our free software for stepwise and best‐subsets DF analysis.     

C-Type Virus associated with Gibbon Lymphosarcoma

C-TYPE viruses have been established as the causal agents of leukaemia in murine and feline species and have been characterized^1,2. C-type virus is also probably associated with fibrosarcoma in non-human primates^3–6. To determine whether viruses with identical characteristics are associated with other neoplasms in simian species, we looked for C-type viruses in cases of leukaemia. A gibbon (Hylobates lar) with a disseminated tumour (later confirmed as lymphosarcoma) was made available to the Comparative Oncology Laboratory by Dr Malcolm Jones of the University of California, San Francisco Medical Center. The principal sites of involvement (lymph node, liver and bone marrow) were extensively overrun with massive neoplastic cells, which were predominantly prolymphocytic forms. Electron microscopy revealed C-type particles identical to those observed in vitro in sections from lymph nodes, liver, spleen and bone marrow.     

Implications of community concordance for assessing stream integrity at three nested spatial scales in Minnesota,U.S.A.

1. Fish and invertebrate assemblage data collected from 670 stream sites in Minnesota (U.S.A.) were used to calculate concordance across three nested spatial scales (statewide, ecoregion and catchment). Predictive taxa richness models, calibrated using the same data, were used to evaluate whether concordant communities exhibited similar trends in human‐induced taxa loss across all three scales. Finally, we evaluated the strength of the relationship between selected environmental variables and the composition of both assemblages at all three spatial scales. 2. Significant concordance between fish and invertebrate communities occurred at the statewide scale as well as in six of seven ecoregions and 17 of the 21 major catchments. However, concordance was not consistently indicative of significant relationships between rates of fish and invertebrate taxa loss at those same scales. 3. Fish and invertebrate communities were largely associated with different environmental variables, although the composition of both communities was strongly correlated with stream size across all three scales. 4. Predictive taxa‐loss models for fish assemblages were less sensitive and precise than models for invertebrate assemblages, likely because of the relatively low number of common fish taxa in our data set. Both models, however, distinguished reference from non‐reference sites. 5. The importance of concordance, geographic context and scale are discussed in relation to the design and interpretation of stream integrity indicators. In particular, our findings suggest that community concordance should not be viewed as a substitute for an evaluation of how assemblages respond to environmental stressors.