APS209: Lecture 4. Optimal Foraging


The below set of notes is from APS209: Lecture 4: Optimal Foraging.

Natural Selection will favour animals that behave most efficiently, e.g. individuals that forage optimally will maximise energy intake while minimising handling time. John Maynard-Smith pioneered optimality modelling to determine how animals can best optimise their behaviour.

Selecting what to eat

In most cases it is possible to quantify the rate at which an animal obtains energy by dividing the caloric value by the time taken to find, process (e.g., time taken to open a mussel shell) and eat something. Animals often have a range of potential prey items or foraging patches. Which should they choose? Northwestern crows predate whelks, which they open by dropping from c.5m onto rocks. Small whelks are unprofitable so they only bother with large whelks. They keep dropping a whelk until it breaks. Measurements of drop height on the number of drops needed to crack whelks of different sizes gives insight into the crow strategy. First, bigger whelks require fewer drops to crack. Second, dropping from 5m is almost as good as greater heights. Third, there do not seem to be any resistant whelks. If a whelk has not cracked after, say, 5 drops it is probably just lucky so it’s not appropriate to abandon whelks that do not crack.

Predictions can be made about the optimal size of food items that should be taken by an animal. Oystercatchers for example are predicted to take very large mussels because these provide the most food. However oystercatchers in the wild do not select the largest mussels, so they appear to be behaving non-optimally. What does this tell us? Is optimal foraging theory wrong? What we often learn is that we do not understand the constraints that the animal is under. Oystercatchers do not take the largest mussels because these are often impossible to open. Similarly in leafcutter ants, daytime foraging by larger ants is reduced because of attacks from parasitic flies. Moose behave optimally within the bounds of three constraints: an energy constraint, a sodium constraint and a rumen constraint.

Charnov’s Marginal Value Theorem

Often animals must forage in a patchy environment. When doing so they need to decide when to move from one patch to another. Charnov developed his Marginal Value Theorem to predict how animals should forage in a patchy environment. This is based on the idea that animals should move from one patch to the next (or return to their young with food) when the patch at which they are feeding becomes unprofitable, this will depend on the quality of the patch and on the time it takes to travel between patches (or to and from their young).

The loading curve shows the diminishing returns of continuing to feed in the patch. The tangent to the loading curve, drawn from the x-axis at travel time (T) indicates the optimal time for an individual to forage in the patch (Topt).

With increasing travel time it is predicted that an individual should spend longer in the patch.

Kacelnik found that Starlings forage optimally when collecting food for their young. However, although the Marginal Value Theorem is a very useful model, it makes some quite restrictive assumptions. It assumes that (1) travel time between patches is known, (2) travel time and patch time have equal energetic costs, (3) patch profitability is known and (4) there is no predation or competition etc. Great Tits expend more energy during travelling time than during patch time. By adjusting for this (Cowie, 1977) observed travel and foraging times that were closer to those predicted by the model. Lima (1984) found that Downy woodpeckers faced with patches of unknown profitability, sample the patches and are thereby able to maximise their energy intake by minimising their search costs.

Models are very useful tools for studying animal behaviour. They provide quantitative predictions that we can use to test whether our theories for why animals are behaving in particular ways, are correct. When animals don’t behave in the way the models predict we can rebuild our models to incorporate constraints or alter assumptions and thereby learn more about the behaviour.

Belovsky G.E. 1978. Diet optimization in a generalist herbivore: the Moose. Theoret. Pop. Biol.14:105-134

Cowie R.J. 1977. Optimal foraging in Great tits Parus major. Nature 268:137-139.

Kacelnik A. 1984. Central place foraging in Starlings (Sturnus vulgaris). I Patch residence time. J.Anim. Ecol. 53:283-299.

Lima S. 1984. Downy woodpecker foraging behavior: efficient sampling in simple stochastoic environments. Ecology 65:166-174.

Zach R. 1979. Shell dropping: decision making and optimal foraging in Northwestern crows. Behaviour 68:106-117.