Multivariate analysis of the phenotypic relationships of the species of Paragalaxias and Galaxias (Pisces: Galaxiidae) in Tasmania

Authors

  • C. R. Johnson,

    1. Department of Zoology, University of Tasmania, Box 252C, G.P.O. Hobart, Tasmania, Australia 7001
    Search for more papers by this author
    • *

      Department of Biology, Dalhousie University, Halifax, Nova Scotia, Canada.

  • D. A. Ratkowsky,

    1. Division of Mathematics and Statistics, CSIRO, Tasmanian Regional Laboratory, ‘Stowell’, Stowell Avenue, Hobart, Tasmania, Australia 7000
    Search for more papers by this author
  • R. W. G. White

    Corresponding author
    1. Department of Zoology, University of Tasmania, Box 252C, G.P.O. Hobart, Tasmania, Australia 7001
    Search for more papers by this author

To whom requests for reprints should be sent.

Abstract

The 10 species of Galaxias in Tasmania, G. olidus from mainland Australia and the four species of Paragalaxias were studied using principal co-ordinates analysis (PCOA) and cluster analysis of a standardized Euclidean distance matrix based upon variate means, and by canonical variate analysis (CVA) conducted as a stepwise multiple discriminant analysis. Thirty-five variables comprising 30 morphometric and five meristic characters were analysed. The meristic characters were not included in the CVA. Excellent separation of the two genera was achieved in all analyses. The multivariate analyses were repeated on each genus separately to see if relationships suggested by the overall analysis remain stable. When the resultant groupings of species are compared for the different analyses, no consistent, distinct groupings of species within each genus are apparent. Despite the absence of distinct groupings, some trends in the affinities of some species are evident. In particular, species affinities as indicated by the CVA are more consistent with established opinions of species relationships. From the results of the study it is suggested that caution be exercised in the application of multivariate statistical analyses of morphological data to ichthyological systematics and phylogeny.

Ancillary