Application of numerical classification in ecological investigations of water pollution

Numerical classification encompasses a variety of techniques for the grouping of entities based on the resemblance of their attributes according to mathematically stated criteria. In ecology this usually involves classific~tion of collections, representing sites or sampling per1ods, or classification of species. Classification can thus simplify patterns of collection resemblance or species distribution patterns in an instructive and efficient manne r. Procedures of numerical classification are thoroughly reviewed, including data manipulations, computation ~f r e semblance measures and clustering methods. The 1mportance and e ffects of transformations and standardizations are di s cussed. It is particularly critical to choose an appropriate resemblance measure which best corresponds with the investigator's concept of ecological resemblance. Clus tering methods form groups on the basis of patter~s of inte r-entity similarity. Various types of clusterlng me thods e xist but currently the most useful and best de veloped are those which are exclusive, intrinsic, . hierarchical and agglomerative. Agglomerative clusterlng me thods which distort spatial relationships and intensely c luster are often most useful with ecological data. The value of post-clustering analyses in the interpretation of the results of numerical classifications is stressed. Thes e include reallocation of misclassified entities, comparison of classifications of collections with thos~ of species (nodal analysis), comparing alternate c~assl­ fi cations, testing differences among groups, relatlng clas sification to extrinsic environmental factors and i nterfacing classification with other multivariate analyses. The use f ulne ss of numerical classification is demonstrated f or obj ec tive analysis of the data sets resulting from field s urveys and monitoring studies conducted for the assessme nt of effe cts of pollution. However, to date few. pol l ution bio logists have applied the more powerful classlf i catory tec h nique s and post-clustering analyses.