The possibility of using bird observations made at sea as a means of monitoring krill in the Antarctic Ocean is explored. No variation source directly linked with krill was identified by means of correspondence analysis. Contour plotting, profile analysis and tests for equality of covariance matrices proved capable of detecting changes in bird abundance and/or species composition, associated with three levels of krill abundance.

A vector close to multi-normality was obtained, thus suggesting that refinements of the technique will allow the discrimination of krill levels by means of bird counts thereby providing an efficient tool for statistical tests of significance. The most promising approach seems to be the application of discriminant analysis.

Further developments of the method are possible only with greater databases, calling for international cooperation on the subject.