Individual dietary specialization and dive behaviour in the California sea otter: Using archival time–depth data to detect alternative foraging strategies

The existence of individual prey specializations has been reported for an ever-growing number of taxa, and has important ramifications for our understanding of predator–prey dynamics. We use the California sea otter population as a case study to validate the use of archival time–depth data to detect and measure differences in foraging behaviour and diet. We collected observational foraging data from radio-tagged sea otters that had been equipped with Mk9 time depth recorders (TDRs, Wildlife Computers, Redmond, WA). After recapturing the study animals and retrieving the TDRs it was possible to compare the two data types, by matching individual dives from the TDR record with observational data and thus examining behavioural correlates of capture success and prey species. Individuals varied with respect to prey selection, aggregating into one of three distinct dietary specializations. A number of TDR-derived parameters, particularly dive depth and post-dive surface interval, differed predictably between specialist types. A combination of six dive parameters was particularly useful for discriminating between specialist types, and when incorporated into a multivariate cluster analysis, these six parameters resulted in classification of 13 adult female sea otters into three clusters that corresponded almost perfectly to the diet-based classification (1 out of 13 animals was misclassified). Thus based solely on quantifiable traits of time–depth data that have been collected over an appropriate period (in this case 1 year per animal), it was possible to assign female sea otters to diet type with 490% accuracy. TDR data can thus be used as a tool to measure the degree of individual specialization in sea otter populations, a conclusion that will likely apply to other diving marine vertebrates as well. Our ultimate goals must be both to understand the causes of individual specialization, and to incorporate such variation into models of populationand community-level food web dynamics. r 2007 Elsevier Ltd. All rights reserved.