Predator-Prey Cycles

How a predator-prey cycle starts

Picture a boreal forest in spring. Snowshoe hares are abundant — last year was good, browse was plentiful, females raised three or four litters. Lynx, the specialist predator, are still recovering from a lean winter. Their numbers are low but climbing because food is everywhere. This is the first quarter of the cycle: prey high, predators rising.

By summer the lynx population has caught up. Each lynx now finds prey easily, kits survive, and the predator density passes the threshold at which hare reproduction can no longer outrun mortality. Hares start to crash. This is the second quarter: prey falling, predators still high and still hunting.

A year or two later the hares are sparse. Lynx litters fail, kits starve, adult lynx wander into trapper territory looking for food. The fur records spike — predator captures peak roughly two years after prey peaks, the famous lag. This is the third quarter: both species low, predators starving fastest.

With predators reduced, surviving hares — released from heavy predation, finding browse regrown after years of low grazing pressure — reproduce explosively. Their numbers double every breeding season. The fourth quarter: prey climbing fast, predators still scarce. The cycle has come back to where it started, ready to repeat.

The Lotka-Volterra equations

The mathematics behind the cycle was worked out twice independently — by Alfred Lotka in 1925 (motivated by chemical autocatalysis) and Vito Volterra in 1926 (asked by his son-in-law, a marine biologist, to explain why selachian fish increased during World War I when fishing collapsed). Their two coupled differential equations are the foundation of population ecology:

dN/dt = aN − bNP        (prey: grow on their own, get eaten)
dP/dt = cNP − dP        (predators: grow by eating, die otherwise)

Here N is prey density, P is predator density, and a, b, c, d are positive rate constants for prey reproduction, predation efficiency, predator conversion, and predator mortality. The system has a non-trivial equilibrium at N* = d/c, P* = a/b — but the equilibrium is neutrally stable, meaning trajectories near it loop around it forever rather than spiraling in.

In phase space (predators on the y-axis, prey on the x-axis) every trajectory is a closed loop. Bigger initial perturbations make bigger loops. There is no damping. This is the simplest nontrivial dynamical system that produces sustained oscillation, and it remains the textbook entry point even though every real population deviates from it.

Worked example: the Canadian lynx-hare cycle

The Hudson's Bay Company kept meticulous records of fur returns from 1820 onward — pelts of snowshoe hare (Lepus americanus) and Canada lynx (Lynx canadensis) trapped across the boreal forest. Charles Elton, in 1924, plotted them and saw something extraordinary: both species oscillated in lockstep, peak to peak about 10 years apart, with the lynx peak trailing the hare peak by roughly two years. The amplitude was enormous — hares varied by two orders of magnitude.

For decades the lynx-hare cycle was the textbook example of pure two-species predator-prey dynamics. Recent work has complicated the story. Hares cycle on islands where there are no lynx; the cycle period correlates with vegetation regrowth time; and hare physiology shows chronic-stress effects when predator density is high (cortisol, suppressed reproduction, heritable to offspring). The current view is a tritrophic story: vegetation-hare-predator coupled together, with lynx riding the hare cycle but not creating it on their own.

Even with that revision, the lynx-hare data remain the cleanest century-long record of population oscillation in any wild system. Anyone learning ecology meets this graph in their first textbook chapter.

Predator-prey cycles vs related dynamics

	Predator-prey	Host-parasite	Prey-switching predator
Interaction sign	+/− (predator/prey)	+/− (parasite/host)	+/−, but partner depends on density
Lethal per encounter?	Usually yes	Usually no	Yes (when targeted)
Generation time ratio	Comparable	Parasite ≪ host	Predator generalist
Cycle period	≈ generation × 2π	Often within-host weeks	Damped or absent
Stabilizing forces	Refugia, prey switching	Acquired immunity	Density-dependent diet
Classic example	Lynx-hare, wolf-moose	Measles in pre-vaccine era	Songbirds tracking caterpillars
Lotka-Volterra fit	Decent	Needs SIR extensions	Needs functional response

The same coupled-equation logic underlies all three, but the time scales, lethalities, and stabilizing mechanisms differ. Real ecosystems are usually closer to "switching" than to pure Lotka-Volterra — most predators eat several things and shift to whichever is easiest.

Phase-space portrait

Plotting predator density against prey density (rather than each against time) reveals the structure most cleanly. Time becomes implicit; what you see is a closed loop. Pick any starting point on the loop. Move counterclockwise:

1. East side (prey high, predators low). Prey are growing exponentially. Predators are recovering. The system moves north — predator density rises.
2. North side (both high). Predation overwhelms prey reproduction. Prey crash. The system moves west.
3. West side (prey low, predators high). Predators starve faster than they reproduce. Predator density falls. The system moves south.
4. South side (both low). Released prey grow. Predators still scarce. The system returns east, completing one cycle.

Three things break this idealized loop in the real world: stochastic shocks (a hard winter, a disease outbreak), density-dependent prey reproduction (which pulls trajectories inward toward equilibrium), and external resources for predators (alternative prey, scavenging) that prevent the predator population from crashing all the way.

Real-world examples beyond lynx-hare

• Isle Royale wolves and moose (1958-present). The longest continuous predator-prey study in the world. Cycle structure is messier than lynx-hare — disease (canine parvovirus arrived 1980), inbreeding (wolves dropped to two related individuals in 2018), and severe winters all shape the dynamics.
• Didinium and Paramecium (Gause, 1934). Classic lab experiments in test tubes. Pure cultures crashed to extinction; adding refugia (sediment, glass wool) let cycles persist. Demonstrated that real cycles need spatial structure.
• Soay sheep on Hirta, Scotland. No predators, but prey crash anyway when winter forage and parasite load conspire. Shows oscillation can arise from one-species feedback alone, not just predation.
• Lemmings and stoats in Fennoscandia. The classic 3-4 year northern small-mammal cycle. Period scales with latitude — collapses to non-cyclic behavior in southern Scandinavia where milder winters break the feedback.
• Spruce budworm and balsam fir in eastern Canada. An outbreak insect that defoliates fir over 30-60 year cycles. The "predator" is bird and parasitoid mortality; the resource recovers slowly because trees are long-lived.
• Yellowstone wolves and elk (post-1995 reintroduction). The most famous trophic cascade. Wolves cut elk numbers, elk avoided river bottoms, willows recovered, beavers returned, river morphology changed. Predator-prey dynamics that ripple through the entire ecosystem.

Variants and refinements

• Functional responses (Holling, 1959). The simple Lotka-Volterra term bNP assumes predators eat in proportion to encounter rate. Holling's Type II response adds a saturation term — predators can only handle so many prey per day. This stabilizes some systems and destabilizes others.
• Numerical responses. Predators don't reproduce instantaneously; they need time to convert food into offspring. Adding a delay reflects biology and dampens oscillation.
• Refugia. Adding a fixed safe zone for prey converts a closed orbit into a damped spiral that settles to coexistence. Real habitat heterogeneity is the main reason coupled species don't go extinct.
• Three-species systems. Adding a top predator (apex) or alternative prey often stabilizes cycles. Hairston-Smith-Slobodkin (1960) argued the world is green because three trophic levels keep herbivores in check.
• Stochastic Lotka-Volterra. Adding noise to the rate constants converts neutrally stable orbits into noisy spiraling trajectories that match data far better than the deterministic version.
• Disease in the loop. Adding a parasite that attacks the predator (or prey) can convert cycles into chaos, drive extinctions, or stabilize them depending on parameters. Anderson and May's 1980s work formalized this.

Common pitfalls

• Assuming all wild populations cycle. Most don't. Pure cycles require tight two-species coupling, generation-time matching, and minimal alternative resources — uncommon in nature.
• Confusing cycle with seasonal pulse. Annual peaks and troughs from breeding seasons are not predator-prey cycles. Look for multi-year periodicity exceeding generation time.
• Treating lynx-hare as wolves-and-deer. Specialist predators on a single key prey are rare. Most large carnivores eat multiple prey species and don't show clean cycles.
• Ignoring vegetation. The hare cycle isn't lynx alone — vegetation regrowth time is part of the period. Two-species models leave this out.
• Forgetting density-dependent reproduction. Even without predators, prey crash when food runs out. Don't attribute everything to predation.
• Reading Lotka-Volterra as predictive. The model is a heuristic, not a forecast tool. Real cycles are noisy, drift in period, and respond to climate. Use it to build intuition, then move to fitted statistical models.