Predator–Prey Dynamics and Optimal Foraging Theory

Predator–prey dynamics and optimal foraging theory unite population-level consumer–resource mathematics with individual behavioral optimization to explain how energy and biomass flow across trophic levels and how predation structures ecological communities. Initiated independently by American biophysicist Alfred J. Lotka (1925) and Italian mathematician Vito Volterra (1926) to explain coupled population cycles in Adriatic fisheries and boreal fur-trapping records (such as the 9–11-year periodicity of snowshoe hares Lepus americanus and Canadian lynx Lynx canadensis), classical predator–prey theory models the demographic coupling between resource density (N) and consumer density (P).

To ground population differential equations in behavioral mechanisms, C. S. Holling (1959) decomposed the response of predators to prey density into the numerical response (changes in predator population density P through fecundity, mortality, and aggregation) and the functional response (the per-capita consumption rate of an individual predator as a function of prey abundance N). Concurrently, Robert H. MacArthur, Eric R. Pianka (1966), and Eric L. Charnov (1976) developed Optimal Foraging Theory (OFT), applying microeconomic optimization models to predict which prey items a consumer should attack and how long it should exploit a depleting resource patch to maximize its net rate of energy intake (E / T) under natural selection.

Type I, Type II (Disc Equation), and Type III Functional Responses

In his landmark 1959 sandpaper-disc experiments and small-mammal predation studies on European pine sawfly cocoons (Neodiprion sertifer), C. S. Holling identified three canonical functional response curves relating the number of prey consumed per predator (N_a) during total foraging time T to prey density N:

1. Type I (Linear / Filter-Feeding Response): Per-capita capture rate increases linearly with prey density (f(N) = aN, where a is the attack rate or clearance volume per unit time) up to an abrupt saturation ceiling where mechanical ingestion reaches capacity. Type I responses occur in passive filter feeders (such as Daphnia cladocerans, baleen whales, or web-building spiders) whose handling of one prey item does not interrupt searching for others.
2. Type II (Cyrtocrinid / Saturating Hyperbolic Response): In most active predators, total time T is partitioned between searching time (T_s) and handling time (T_h N_a, encompassing pursuit, subjugation, consumption, and post-prandial digestion of each item). Substituting T_s = T − T_hN_a into the capture relation N_a = a T_s N yields Holling's Disc Equation:

f(N) = N_a / T = aN / (1 + a T_h N) As N → ∞, the per-capita consumption rate asymptotes at a maximum physiological ceiling of 1 / T_h. Because the proportion of the prey population killed per predator (f(N) / N = a / (1 + aT_hN)) decreases monotonically with N, a Type II response exerts inverse (destabilizing) density-dependent mortality on the prey population.

1. Type III (Sigmoidal / Prey-Switching Response): At low prey densities, per-capita capture rates accelerate faster than linearly (f(N) = aN^k / (1 + aT_hN^k) with k ≥ 2) before saturating at high N due to handling time. Sigmoidal Type III curves arise when prey occupy spatial refuges at low densities, or when generalist vertebrate predators form a cognitive search image (Nikolaas Tinbergen, 1960) and exhibit prey switching (F. H. Murdoch, 1969), concentrating attacks on whichever prey species is locally common while ignoring rare prey. Over the low-to-intermediate density range where ∂(f(N)/N)/∂*N* > 0, Type III predation exerts stabilizing direct density-dependent regulation.

Rosenzweig–MacArthur Isoclines and the Paradox of Enrichment

Whereas the original Lotka–Volterra equations (d*N*/d*t* = rN − aNP; d*P*/d*t* = baNP − mP) exhibit neutrally stable frictionless cycles whose amplitude depends on initial conditions, Michael L. Rosenzweig and Robert H. MacArthur (1963) incorporated logistic prey self-limitation (K) and a saturating Holling Type II functional response:

d*N* / d*t* = rN(1 − N / K) − [aNP / (1 + aT_hN)], and d*P* / d*t* = e [aNP / (1 + aT_hN)] − mP

In the (N, P) phase plane, the predator zero-growth isocline (d*P*/d*t* = 0) is a vertical line at N̂ = m / [a(e − mT_h)], while the prey zero-growth isocline (d*N*/d*t* = 0) forms a humped parabola peaking at N_peak = (K − 1/(aT_h)) / 2. When the vertical predator isocline intersects the prey isocline to the right of the hump (N̂ > N_peak), prey self-limitation damps oscillations toward a locally stable equilibrium point. However, Michael L. Rosenzweig (1971) discovered the Paradox of Enrichment: increasing the nutrient carrying capacity K shifts the hump of the prey isocline to the right until N̂ < N_peak, crossing a Hopf bifurcation that destabilizes the equilibrium into a stable limit cycle with large-amplitude booms and busts that expose both populations to stochastic extinction at cycle troughs.

The Contingency Prey Model (Diet Breadth)

Formulated by Robert H. MacArthur and Eric R. Pianka (1966) and formalized by Eric L. Charnov (1976), the contingency prey model asks whether a searching predator should narrow its diet to high-value items or generalize across k available prey types, each characterized by encounter rate λ_i, net assimilate energy E_i, and handling time h_i. Ranking prey types in descending order of profitability (E₁/h₁ > E₂/h₂ > ... > E_n/h_n), a forager maximizes its long-term rate of energy intake (R = [∑ λ_iE_i] / [1 + ∑ λ_ih_i]) by including prey type j in its diet if and only if its profitability exceeds the average intake rate of the higher-ranked diet set {1, ..., j − 1}:

E_j / h_j > [∑_i=1^j−1 λ_iE_i] / [1 + ∑_i=1^j−1 λ_ih_i]

This inequality yields three counterintuitive Three Rules of Optimal Diet:

1. Zero–One Rule: A prey type is either always attacked upon encounter (p_j = 1) or always ignored (p_j = 0); partial preferences occur only at knife-edge thresholds or under nutrient constraints.
2. Independence from Own Abundance: Whether a lower-ranked prey item j should be included in the diet depends solely on the encounter rates (λ₁, ..., λ_j−1) of more profitable items, and is independent of its own encounter rate λ_j.
3. Specialization under Richness: As high-profitability prey (λ₁) become abundant, optimal foragers drop lower-ranked items and specialize; when preferred prey are scarce, foragers expand their diet breadth to generalize. John R. Krebs and colleagues (1977) verified these predictions experimentally using great tits (Parus major) foraging on large and small mealworm segments delivered on a moving conveyor belt.

Charnov's Marginal Value Theorem (MVT)

When resources are clumped in discrete patches separated by travel time τ, a forager's cumulative energy gain G(t) inside a patch exhibits diminishing returns (∂*G*/∂*t* > 0, ∂²G/∂*t*² < 0) as local prey are depleted. Eric L. Charnov's Marginal Value Theorem (1976) proves that to maximize overall net energy intake rate across the habitat, the forager should depart a patch at the residence time *t*\* where its instantaneous marginal capture rate inside the patch drops to equal the long-term average intake rate across the entire environment (including travel time τ):

∂*G*(*t*\*) / ∂*t* = G(*t*\*) / (τ + *t*\*)

Geometrically, the tangent line drawn from (−τ, 0) to the gain curve G(t) defines *t*\*; hence, longer inter-patch travel times τ or poorer habitat quality require longer patch residence times *t*\*.

Trait-Mediated Cascade Dynamics and Giving-Up Densities

Predators shape ecosystems not only by killing prey (Density-Mediated Indirect Interactions, DMII), but also by intimidating prey into altering their habitat use, vigilance, and physiology (Trait-Mediated Indirect Interactions, TMII). Joel S. Brown (1988) quantified perceived predation risk using the Giving-Up Density (GUD)—the amount of food left behind in a depletable experimental patch when a forager quits, where metabolic harvest rate (H) balances metabolic cost (C), predation risk cost (P), and missed-opportunity cost (MOC): H = C + P + MOC.

Coined by John W. Laundré, Lucina Hernández, and Kelly B. Altendorf (2001) and experimentally tested by Oswald J. Schmitz in spider–grasshopper–plant mesocosms, the Landscape of Fear demonstrates that non-consumptive predator presence often drives trophic cascades equal to or stronger than direct lethal consumption. When Schmitz rendered Pisaurina mira nursery-web spiders incapable of killing Melanoplus femurrubrum grasshoppers by gluing the spiders' chelicerae, frightened grasshoppers shifted from high-protein grasses to protective goldenrod (Solidago rugosa) refugia, generating a behavioral trophic cascade matching that of lethal spiders.

How did Carl Huffaker's mite experiments demonstrate spatial persistence in predator–prey systems?

In 1958, Carl B. Huffaker tested Gause's observation that predators rapidly eradicate prey in homogeneous laboratory arenas by rearing six-spotted herbivorous mites (Eotetranychus sexmaculatus) and predatory mites (Typhlodromus occidentalis) on arrays of oranges interspersed with rubber balls. In simple, unbarricaded arrangements, predatory mites located every occupied orange, drove the prey to extinction, and starved within weeks. However, when Huffaker created a complex 120-orange spatial labyrinth with Vaseline barriers slowing walking predators and wooden posts allowing prey mites to disperse via silk parachutes on air currents, the system sustained repeated predator–prey oscillations through asynchronous patch extinction and recolonization—an early experimental confirmation of metapopulation persistence.

What is the Nicholson–Bailey model and how do host–parasitoid systems stabilize?

Formulated by A. J. Nicholson and V. A. Bailey (1935) for insect parasitoids with discrete, non-overlapping generations, the model assumes females search at random (N_t+1 = λN_t exp(−*aP*_t); P_t+1 = cN_t[1 − exp(−*aP*_t)]). The classic Nicholson–Bailey equilibrium is unstable, producing divergent oscillations that crash both populations. Michael P. Hassell and Robert M. May (1973, 1978) showed that natural host–parasitoid systems stabilize when parasitoid attacks follow an aggregated negative binomial distribution with clumping parameter k < 1—meaning parasitoids aggregate on dense host patches while leaving low-density host patches as partial spatial refuges.

Why do risk-sensitive foragers switch between risk-averse and risk-prone choices?

Developed by Thomas Caraco (1980) in experiments with yellow-eyed juncos (Junco phaeonotus), risk-sensitive foraging theory evaluates how an animal chooses between a constant food reward and a variable reward with the same mathematical mean. When a bird's daily energy budget exceeds its overnight survival threshold, its fitness utility function is concave downward, favoring risk-averse foraging (choosing the constant option). Conversely, if the bird is starving and the mean reward is insufficient to survive cold nocturnal temperatures, only a high-variance gamble offers a probability of survival, causing the forager to switch to risk-prone behavior.

1. Holling, C. S. (1959). "Some Characteristics of Simple Types of Predation and Parasitism." The Canadian Entomologist, 91(7), 385–398. doi:10.4039/Ent91385-7
2. Rosenzweig, M. L., & MacArthur, R. H. (1963). "Graphical Representation and Stability Conditions of Predator-Prey Interactions." The American Naturalist, 97(895), 209–223. doi:10.1086/282272
3. MacArthur, R. H., & Pianka, E. R. (1966). "On Optimal Use of a Patchy Environment." The American Naturalist, 100(916), 603–609. doi:10.1086/282454
4. Charnov, E. L. (1976). "Optimal Foraging, the Marginal Value Theorem." Theoretical Population Biology, 9(2), 129–136. doi:10.1016/0040-5809(76)90040-X90040-X)