Optimal Foraging Theory: How Animals Calculate the Cost of Every Meal

A hummingbird visits roughly 1,500 flowers per day to meet its energy needs. It cannot afford to probe a flower that was already drained ten minutes ago. An otter diving for sea urchins must decide whether to eat the small one it just found or keep searching for a larger one. These choices look deliberate — but they are not conscious. They are the product of natural selection acting over millions of years to produce behavior that maximizes energy intake per unit time. This is the central claim of Optimal Foraging Theory.

By the end of this lesson you will be able to: • State the central assumption of Optimal Foraging Theory (OFT) and explain what it does not claim about animal cognition • Apply the prey selection model's profitability formula to rank food types and predict diet choices • Use the Marginal Value Theorem to predict when a forager should leave a depleting patch • Evaluate OFT's limits using real experimental evidence and the concept of satisficing

Optimal Foraging Theory (OFT) was formally developed by MacArthur and Pianka in 1966, and extended through the 1970s–80s by Charnov, Krebs, and others. Its central assumption: natural selection has favored foraging behaviors that maximize net energy gained per unit of total time spent — including travel time, search time, and handling time. Foraging efficiency = Net energy gained / Total time invested Critically, OFT does not claim animals consciously calculate this ratio. It predicts that behavior shaped by evolution will approximate the optimal strategy as if the animal were calculating it — exactly as a ball rolling downhill finds the lowest point without choosing to. The test is empirical: do real animals behave as the model predicts?

The prey selection model asks which food types a forager should include in its diet. Each prey type has two key properties: Ei = energy value (calories or joules) hi = handling time (seconds to catch, subdue, and consume it) Profitability = Ei / hi (energy returned per second of handling) The model's prediction: rank all prey types by profitability. Always accept the highest-ranked type when encountered. Add lower-ranked types only when doing so raises the overall rate of energy intake. The counterintuitive key prediction: whether to take a low-ranked prey item depends on the ABUNDANCE of high-ranked items, not on the abundance of the low-ranked item itself. When profitable prey is plentiful, pass on less-profitable prey even if it is right in front of you. When profitable prey is rare, accept less-profitable prey because the search time for something better exceeds the handling cost of eating what is available. Krebs, Erichsen, Webber, and Charnov (1977) tested this with great tits (Parus major) offered large and small mealworm pieces on a conveyor belt. At high densities of large (high-profitability) pieces, birds ignored small pieces almost entirely — exactly matching the model's prediction.

Food often occurs in patches — a tree of fruit, a mussel bed, a patch of flowers. As a forager depletes a patch, the rate of gain falls. Eric Charnov's Marginal Value Theorem (MVT, 1976) predicts the optimal moment to leave. The core logic: leave a patch when your current gain rate (the marginal value) drops to the average gain rate available across the whole environment. Staying longer means earning less than you could by moving on. A key prediction: the optimal stay time depends on travel time between patches. Long travel time between patches → stay longer in each one, because leaving is expensive. Short travel time → leave sooner, because a fresh patch is cheap to reach. Animals in habitats where patches are far apart should tolerate lower gain rates before leaving than animals in dense patch habitats. Bee studies confirm this: nectivorous bees leave individual flowers at a threshold closely matching MVT predictions adjusted for the specific travel distances in that meadow. Load-carrying bumblebees returning to a hive also adjust their collection threshold based on round-trip flight cost — exactly as the MVT predicts for central-place foragers.

OFT predictions have held up remarkably well in controlled experiments. The great tit conveyor-belt study and numerous bee foraging studies provide strong empirical support for both the prey selection model and the MVT. However, real animals deviate from pure energy-maximization in predictable ways: Risk sensitivity: animals near starvation often choose lower but more reliable food sources over higher-variance options — even if expected energy is the same. They are risk-averse, violating simple energy maximization. Multiple currencies: energy is not the only constraint. Foragers must balance energy gain against predation risk (a bird may eat suboptimal food near cover rather than optimal food in the open), water intake, micronutrient needs, and social demands. Incomplete information: animals cannot perfectly sample their environment. They use rules of thumb — leave-after-n-prey rules, giving-up time thresholds — that approximate optimal behavior without computing it exactly. The upshot: animals are satisficers, not optimizers. They do well enough, shaped by selection toward the optimum — but real behavior is messier, and understanding the deviations is often as scientifically interesting as the predictions themselves.