Premium
Applications of Multivariate Statistical Methods in Fisheries
Author(s) -
Paukert Craig P.,
Wittig Timothy A.
Publication year - 2002
Publication title -
fisheries
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.725
H-Index - 79
eISSN - 1548-8446
pISSN - 0363-2415
DOI - 10.1577/1548-8446(2002)027<0016:aomsmi>2.0.co;2
Subject(s) - fishery , multivariate statistics , statistics , biology , mathematics
Abstract We summarized the uses of cluster analysis, discriminant function analysis (DFA), multivariate analysis of variance (MANOVA), and principal components analysis (PCA) in Transactions of the American Fisheries Society (TAFS) and North American Journal of Fisheries Management (NAJFM) from 1981–2001. The use of all multivariate methods, except cluster analysis, increased in TAFS over the last 21 years, whereas it did not for NAJFM. Two common goals of multivariate methods were to ordinate or group habitat variables and to relate fish population characteristics with habitat. No paper fully reported the statistical test or addressed all associated assumptions. Researchers conducting diet analysis studies or relating fish to habitat characteristics should consider using PCA or MANOVA, whereas scientists conducting morphological experiments should consider DFA. Cluster analysis was widely accepted in genetic studies of allele or loci frequencies. Statistical assumptions need to be addressed when the objective of the analysis is hypothesis testing or calculation of confidence intervals, which are based on statistical distributions (e.g., F , X 2 ) that require these assumptions.