Behaviour · 2007 · 57 citations · 39 references
Life SciencesMarine SciencesZoologyMarine EcologyMarine MammalogyMarine ZoologyAnimal IdentificationDiving PhysiologyCumulative Frequency AnalysisCumulative Frequency DistributionsDiving BehaviorHistogram AnalysisMaximum Likelihood InferenceMethod Of Maximum LikelihoodSealsTemporal Environmental Variation
Foraging behaviour often occurs in bouts, and defining these bouts is challenging because they reflect varying environmental scales, yet existing methods are diverse, subjective, and rarely compared for accuracy and precision. The study examined whether a maximum likelihood estimation method (MLM) could identify dive bouts more effectively than a sequential differences analysis (SDA). MLM was applied to real interdive duration data from Antarctic fur seals and compared with SDA, while also testing both methods on simulated data to assess bias and parameter estimation. MLM produced shorter bout‑ending criteria, more precise parameter estimates, better agreement with observed cumulative frequency distributions, and less biased parameter estimates than SDA, and its performance was affected by SDA’s histogram bin width choice, indicating MLM as a more accurate, precise, and objective tool for bout identification.
Abstract Foraging behaviour frequently occurs in bouts, and considerable efforts to properly define those bouts have been made because they partly reflect different scales of environmental variation. Methods traditionally used to identify such bouts are diverse, include some level of subjectivity, and their accuracy and precision is rarely compared. Therefore, the applicability of a maximum likelihood estimation method (MLM) for identifying dive bouts was investigated and compared with a recently proposed sequential differences analysis (SDA). Using real data on interdive durations from Antarctic fur seals (Arctocephalus gazella Peters, 1875), the MLM-based model produced briefer bout ending criterion (BEC) and more precise parameter estimates than the SDA approach. The MLM-based model was also in better agreement with real data, as it predicted the cumulative frequency of differences in interdive duration more accurately. Using both methods on simulated data showed that the MLM-based approach produced less biased estimates of the given model parameters than the SDA approach. Different choices of histogram bin widths involved in SDA had a systematic effect on the estimated BEC, such that larger bin widths resulted in longer BECs. These results suggest that using the MLM-based procedure with the sequential differences in interdive durations, and possibly other dive characteristics, may be an accurate, precise, and objective tool for identifying dive bouts.
39
The Problem of Pattern and Scale in Ecology: The Robert H. MacArthur Award Lecture
Simon A. Levin · Ecology · 1992 · 6.8K citations · Full text
Life Sciences, Biology, Ecology +13
Population Responses to Patchy Environments
John A. Wiens · Annual Review of Ecology and Systematics · 1976 · 1.1K citations · Full text
Critical Thresholds in Species' Responses to Landscape Structure
Kimberly A. With, Thomas O. Crist · Ecology · 1995 · 753 citations