Confronting assumptions about prey selection by lunge-feeding whales using a process-based model

The relative energetic benefits of foraging on one type of prey rather than another are not easily measured, particularly for large free-ranging predators. Nonetheless, assumptions about preferred and alternative prey are frequently made when predicting how a predator may impact its environment, adapt to environmental change, or interact with human activities.
We developed and implemented a process-based model to investigate the potential energetic benefit (PEB) of in situ foraging opportunities in rorqual whales. The model integrates and evaluates the energetic importance of measured prey patch characteristics (prey distribution, energy content and predator avoidance) and predator characteristics (morphometrics, foraging tactics and feeding rates).  We applied the model to test the assumption that hatchery-released juvenile salmon are an “easy meal” for humpback whales compared to more common prey, herring and krill.
In eleven out of the thirteen foraging situations considered, whales were found to be feeding in a manner where net energy gain was greater than the energetic costs of non-foraging swimming.  Humpback whale PEB for hatchery-released juvenile salmon fell within the range of the PEB for krill and herring but varied by species, from relatively high PEB for chum salmon to relatively low for coho salmon. Our model provides behavioral insight as well, indicating that shallow feeding may be more important for reducing energy expenditure through slower lunge speeds than for increasing prey capture.  The model also provides a means of identifying prey patch characteristics, with prey aggregation playing the largest role in determining PEB despite being a poor overall proxy for PEB, supporting the use of the complex model framework.
Modeling approaches are especially valuable where they can use reasonable assumptions to substitute for lack of reliable observations, thereby integrating a range of interacting factors into a single framework.  Additionally, because process-based models can make predictions outside the range of previously observed conditions, they will be increasingly useful in a changing climate.

Methods
Methods are described in a manuscript in review