Optimal Foraging Theory (OFT): Energy Maximization, Evolutionary Constraints, and Ecological Applications

OFT as an economics model. Classic OFT, broadly construed, is one example of optimization concepts and models developed originally in engineering and economics (Maynard Smith 1978a). Despite terminology that can be circular or misconstrued when used sim- plistically, optimization does not mean that selection produces optimal solutions, but rather optimization concepts help to identify specific adaptations and the operating constraints (Parker and Maynard Smith 1990; see also Pyke 1983; Stephens and Krebs 1986). Ultimately, the ability to understand the criteria and the extent to which species forage efficiently could help predict their distribution. Nevertheless, it is useful here to recall several caveats noted in Chap. 1: (i) ‘what is an adaptive trait’ (Dobzhansky 1956) in the context of attempts to ‘atomize’ an organism; (ii) that natural selection does not start with a blank slate; and (iii) natural selection can only act on what is available in the gene pool.

All optimization models, of which OFT is an example, share implicitly or explicitly several features (Parker and Maynard Smith 1990). In brief these are (i) operational limits or constraints such as the range of phenotypes on which selection can act; (ii) specification of the optimization criterion, in other words what attribute is being maximized; and (iii) the assumption that the population is responding to the contemporary environment (i.e., not a historical legacy) and that its behavior will be reflected in its contribution to the next generation. In OFT, a typical situation would be briefly as follows (see Box 1 in Parker and Maynard Smith, 1990): (i) benefits and costs, expressed in a common currency such as calories, of adopting a particular action (strategy) are identified; (ii) an indirect optimization criterion typically is selected, usually implicitly being to maximize the net rate of energy intake per unit time spent foraging.

In the case of foraging distance to catch prey, this would be as a function of movement distance, with movement continuing until the marginal value of taking the next step becomes equal to or less than the marginal costs of that step; the optimal move being the one that maximizes caloric benefits minus costs; (iii) the indirect criterion of energy gain is translated implicitly or explicitly into direct fitness (number of progeny) associated with the various strategies; (iv) predictions are tested against observations.

Pyke (1984) summarized the assumptions inherent specifically in OFT, the most important being that: (i) an individual’s fitness depends on its behavior while foraging; (ii) foraging behavior is heritable; (iii) the relationship between foraging behavior and fitness is known; and (iv) the animals forage ‘optimally’ i.e., their behavior changes in pace with or more rapidly than the presumed relevant conditions change. Pykes and Maynard Smith’s notation that foraging behavior must be or is assumed to be directly related to fitness is important, especially since, in routine tests of the theory with macroorganisms, fitness (numbers of descendants) is generally impossible to measure. In the case of microorganisms, attributes such as growth rate, competitive success, total population size, i.e., direct measures of fitness are routinely determined, as discussed later.

While much of OFT has dealt with the foraging behavior of birds, there are other interesting situations where foraging efficiency and energy economy have been studied (see Table 9.1 in Stephens and Krebs 1986 for synopsis and tests). Some are direct tests of the theory but many are not. For example, consider the giant panda bear and its food, almost exclusively bamboo (a good example of specialization; see later section). Actually, the animal appears not to forage efficiently for a couple of reasons. First, pandas are not even anatomically streamlined for holding the stalks of bamboo and stripping off the leaves, as recounted wittily by Steven Jay Gould in his 1980 book ‘The Panda’s Thumb.’ Second, its alimentary tract is poorly adapted to this relatively low-energy, nonnutritious diet. So the animal spends most of its time eating; it also adapts by living passively, along with having evolved physiological adjustments reflected in reduced sizes of typically high-metabolic organs (brain, liver, kidneys) (Nie et al. 2015). In contrast, many predators have episodes of extremely high-energy expenditure.

Hunting styles vary considerably among felids (e.g., lions, cheetahs, pumas) consistent with their different sizes, body designs, prey species, and habitat. Pumas hunt in rugged terrain and exemplify a strategy of stalk/ambush/pounce. The energetics of each phase has been continuously monitored in the wild from animals with radio tracking collars (Williams et al. 2014). These cats have an efficient energy budget overall because most of their time awake is associated with low-energy expenditure (sitting and stalking), while the attack phase, although demanding high-energy expenditure, is of short duration and the animal’s pouncing force is calibrated to prey size. Of course, these two examples, while interesting and relevant to foraging, are not direct tests of the theory.

In formal tests of OFT there are various complications, notwithstanding its conceptual elegance and intuitive appeal. Apart from semantic difficulties alluded to above and noted by Perry and Pianka (1997), examples of some of the complexities include the following situations. (These do not mean that OFT is wrong. Rather, optimal foraging can mean different things to different species and accordingly such cases require modifications to the conventional model [see Chap. 9 in Stephens and Krebs 1986].) The standard criterion of maximizing energy return from the resource may not be the overriding factor in foraging, but rather something else limiting in the diet, such as salt intake or protein acquisition. An interesting case pertains to sodium, which is an essential element often in limiting supply in the diet of moose (Belovsky 1978, 1984).

The primary source of the mineral is aquatic plants, which are available only in the summer as they are under ice in the winter, so the animals must fulfill seasonally their entire annual requirement. However, they supply less energy per unit dry mass than do terrestrial plants and, if consumed exclusively, the aquatics are too bulky and quickly fill the rumen. Thus, theory predicts and Belovsky’s experiments support the postulate that moose forage on an ‘appropriate’ mix of both deciduous leaves and aquatic plants. This turns out to be about 18% aquatics. Analogously, herbivores and detritivores (particularly wood-decomposing fungi) typically are faced with a shortage of nitrogen needed for protein, not carbon. Microbial population size and thus decomposition rate of many forms of plant debris—wood especially because the C:N ratio is high—is driven by nitrogen availability.

Even where energy acquisition is the appropriate currency, organisms may be viewed (simplistically) as foraging either to maximize efficiency of energy intake (energy gain per unit expended) or net rate, i.e., per unit time (see microbial example in the following section). Honeybees in visiting flowers often do not fill their crops with nectar, which runs counter to optimal foraging based on a criterion of net rate of return. When metabolic costs of transporting nectar are accounted for, however, the better foraging prediction accords with crops being only partially filled (Parker and Maynard Smith 1990). With respect to rate, a further consideration is whether the organism is maximizing its instantaneous intake rate of digestible energy or maximizing over some longer period, such as its daily rate. Babin et al. (2011) studied the distribution of bison with respect to a plant community in Saskatchewan, Canada.

They found that the animals were grazing on a species distribution of plants that allowed the animals to maximize instantaneous rate, even though foraging on smaller plants (which required longer cropping times and ultimately provided slower satiation) provided more digestible energy. This strategy is apparently not universal because shorebirds forage consistently with longer rather than shorter time energetic predictions (Quaintenne et al. 2010). Beyond these more global considerations is the complication of animals foraging differently depending on whether they are doing so alone or among competitors or when predators are in the vicinity.

In overview, Stephens and Krebs (1986) summarize the reaction of ecologists to classic OFT into five categories ranging from its being trivial and tautological at one extreme, to often having been verified and useful at the other. That spectrum of opinion was the status in the theory’s heyday of the late 1980s and would apply now as well. The data from their extensive review of the literature (Stephens and Krebs Tables 9.1-9.5) support a conclusion that optimal foraging models generally have performed well. Enthusiasm for econometrics-centered OFT, like all theories in ecology, has peaked and waned over the ensuing decades, and attention shifted to subsequent refinements such as Levy walks previously described (James et al. 2011; Pyke 2015). Like most if not all major theories in ecology, OFT has provoked intense debate and controversy (Perry and Pianka 1997). Nevertheless, both as a formal model and as a framework for empirical tests, it remains a fundamental guide to investigations (see, e.g., Plante et al. 2014).