Bringing the analysis of animal orientation data full circle: model-based approaches with maximum likelihood

Abstract

In studies of animal orientation, data are often represented as directions that can be analyzed using circular statistical methods. Although several circular statistical tests exist to detect the presence of a mean direction, likelihood-based approaches may offer advantages in hypothesis testing – especially when data are multimodal. Unfortunately, likelihood-based inference in animal orientation remains rare. Here, we discuss some of the assumptions and limitations of common circular tests and report a new R package called CircMLE to implement the maximum likelihood analysis of circular data. We illustrate the use of this package on both simulated datasets and an empirical example dataset in Chinook salmon (Oncorhynchus tshawytscha). Our software provides a convenient interface that facilitates the use of model-based approaches in animal orientation studies.