To normalize or not to normalize? Fat is the question

Authors

  • Craig E. Hebert,

    Corresponding author
    1. Environment Canada, Canadian Wildlife Service, National Wildlife Research Centre, 100 Gamelin Blvd., Hull, Québec, K1A 0H3 Canada
    • Environment Canada, Canadian Wildlife Service, National Wildlife Research Centre, 100 Gamelin Blvd., Hull, Québec, K1A 0H3 Canada
    Search for more papers by this author
  • Karen A. Keenleyside

    1. Environment Canada, Ecosystem Conservation Directorate, Evaluation and Interpretation Branch, 351 St. Joseph Blvd., Hull, Québec, K1A 0H3 Canada
    Search for more papers by this author

Abstract

Concentrations of lipophilic contaminants in biota are frequently corrected for variation in tissue lipid content. Usually, this correction is accomplished by dividing tissue contaminant concentration by lipid concentration to form lipid-normalized data. This ratio-based approach is satisfactory when contaminant concentration varies in direct proportion to lipid content. However, when such a relationship does not exist, erroneous conclusions may be reached. Recent research has emphasized the potential shortcomings of the use of ratio variables. We demonstrate the importance of considering these shortcomings when lipid-normalizing data. Three examples are presented, and an alternative approach based upon the use of analysis of covariance (ANCOVA) is suggested.

Ancillary