Species diversity of nematode communities in the digestive tract of domestic ruminants: multivariate versus univariate estimations

The evaluation of species diversity can be performed using species richness (number of species), diversity (number and relative proportions of species ) or evenness (equitability of the proportion of each species). A large num ber of indices related to diversity or evenness is available but their math ematical properties remain unexplained, and hence their use is somewhat unr eliable. We intended to present an array of well defined methods, based on discriminant analysis, to evaluate diversity in helminth communities of rum inants. The database used concerned seven cattle farms and seven sheep farm s located in temperate areas, from which at least three necropsies were und ertaken in autumn. The hosts were chosen because they had experienced nearl y a complete grazing period and no anthelmintic treatment was performed for a minimum of 2 months before the necropsies were undertaken, so that the h elminth communities could be considered as representative of each farm at o ne specific moment. We used discriminant analyses to evaluate diversity, ba sed either on proportions of species, or on proportions of species arranged in decreasing order, or on a set of six indices of diversity selected from the most commonly used. We showed that actual proportions were the most ef ficient in cattle but not in lambs, and that discriminant analyses based on six indices of diversity or based on actual proportions of species ranked in decreasing order yielded very similar results in both host communities. We rejected the use of one single index for evaluating diversity in all cas es and proposed the following procedure: actual proportions of species prov ide information on interactions between species and. when there is limited or no interaction, a combination of diversity indices or single indices cou ld be used for estimating diversity. When interactions between species have to be taken into account, the best information will be obtained either fro m species proportions arranged in decreasing order or a combination of biod iversity indices.