Evaluation of orientation experiment data using circular statistics - doubts and pitfalls in assumptions

Busse P.,Trocinska A. 1999. Evaluation oJorientation experiment data using circular statisticsdoubts and pilJaUsin assumptions. Ring 21, 2: 107-130. The paper is devoted to the methodical analysis of the "silent" assumptions that are the basis of application of the circular statistics to evaluation of bird orientation data collected in "orientation cages" of any design. The study is based on the analysis of available published description of the methods used in evaluation of bird orientation data. The theoretical discussion is followed by a detailed analysis of application of the discussed methods to the data on migrating Robins (Erithacus rubecula), collected at the Operation Baltic field station in autumn 1996. Two different models of approach: classic one-vector model assuming only unimodal behaviour of migrating birds and multi-vector model that accepts all uniand multimodal distributions of the registered signs of bird migration restlessness significantly different from the ran dom one, are compared. Both methods can be applied at two different leveIs for an individual bird and a group of birds. The assumptions of the methods, their consequences and evaluation procedures are presented and discussed in detail. In conclusion it appears that the classic computing routine: (1) is based on wrong biological assumption of unimodality of bird behaviour that is not a case for many tests, (2) when applying classic statistic procedures, it allows to include only unimodal and axial distributions into analysis, (3) biases the results giving strong influence to side vectors that should not be included into the result vector, (4) when studying the group of birds it should not be used in most cases, as the group of birds can show multimodal distribution, instead of only one resulting mean vector. In contrast, multi-vector model evaluation procedures allow finding and analysis of any existing vector pattem at both an individual and a group levels. P. Busse, A. Trocinska, Bird Migration Research Station, University of Gdansk, Przebendowo, 84-210 Choczewo, Poland; E-mail: busse@univ.gda.pl