Premium
Sampling rate and misidentification of Lévy and non‐Lévy movement paths: comment
Author(s) -
Auger-Méthé Marie,
Clair Colleen Cassady St,
Lewis Mark A.,
Derocher Andrew E.
Publication year - 2011
Publication title -
ecology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.144
H-Index - 294
eISSN - 1939-9170
pISSN - 0012-9658
DOI - 10.1890/10-1704.1
Subject(s) - biological sciences , auger , library science , geography , archaeology , biology , computer science , computational biology
In a recent paper, Plank and Codling (2009) critique the use of Levy walks to describe animal movement, arguing that non-Levy walk processes could be misidentified as Levy patterns and, conversely, movement patterns actually generated by Levy processes may be wrongly attributed to other mechanisms. The authors suggest that this ambiguity is partly caused by sampling paths at scales that do not reflect actual movement decisions and this despite the theoretical scale-independence of Levy walks. These findings, if true, would be an important contribution, as the Levy walk is a popular, although controversial, model in the animal movement literature. Here, we support Plank and Codling’s (2009) contention that movement patterns must be attributed to the correct process and that animal movement is likely not truly scale-invariant. However, we challenge their methodology, and hence that they showed that Levy and non-Levy processes could be misidentified for one another and that this ambiguity partly depends on the sampling scale. Our main criticism is that using the relative fit of poorly chosen models, without verifying for the absolute fit of the best model, is insufficient evidence for either the identification or the misidentification of a process. To demonstrate this methodological problem, we first describe the models used to simulate the data and thus representing the movement processes. Then we describe how the models that were fitted to the data differed from the ones used to simulate the data. Finally, we argue that the authors failed to consider the importance of examining the absolute fit of the best model. Without this information, it is impossible to determine whether either model provides a reasonable explanation for a given data set, whether those data are generated by simulations or actual animals.