Module 7: Ungulate Herds — Collective Biophysics of the Savanna

Two species of wildebeest occupy different biogeographical zones: the blue wildebeest (Connochaetes taurinus), abundant across East and Southern Africa, and the black wildebeest (C. gnou), endemic to the highveld grasslands of South Africa. The Serengeti-Mara blue wildebeest population is estimated at 1.3–1.5 million individuals—a global biomass peak for a single large-herbivore species.

Rain-food regulation (Mduma 1999)

Mduma, Sinclair & Hilborn (1999) showed through 30 years of census data that wildebeest population dynamics in the Serengeti are bottom-up regulated by dry-season rainfall. The regulation mechanism runs through grass biomass. Let \(R\) = dry-season rainfall, \(G\) = grass biomass, \(N\) = wildebeest density:

The migration loop: 3000 km per year

The wildebeest herd traces an annual loop approximately 3000 km long, tracking greenness through the ecosystem. December through March: southern Serengeti short-grass plains (calving). April–May: northwest movement through the Western Corridor. June–July: northern crossings of the Mara River. August–October: Masai Mara grazing. November–December: southern return. Each leg is punctuated by river crossings that exact enormous predation and drowning tolls.

Calving synchrony — predator swamping

Approximately 80% of wildebeest calves are born within a three-week window in late January / early February. This extraordinary synchrony is a classic example of predator swamping: by producing far more neonates than the local predator guild can consume during the vulnerable window, the herd ensures per-capita predation mortality crashes. If we denote predator saturation rate \(P_{\max}\) and neonate flux \(B(t)\):

The Mara River crossing — hydrodynamics and crocodiles

The Mara River crossings are defining events. Subalusky et al. (2017) estimated that approximately 6200 wildebeest carcasses per yeardrown or are killed crossing, delivering an enormous pulse of carbon, nitrogen, and phosphorus into the river ecosystem—a continent-scale trophic cascade. Physics of the crossing:

With river current \(U \sim 2\) m/s and peak flows higher still, adult wildebeest (250 kg) are often swept off their feet. Narrow hoofprint area and powerful neck extension allow swimmers to keep muzzles above water; juveniles suffer highest mortality.

Wildebeest locomotion efficiency

Wildebeest achieve minimum cost-of-transport at approximately 60 km/h gallop. Their narrow hoof (contact patch \(\sim 20\,\text{cm}^2\)) concentrates load on dry firm ground. Bone stress at mid-stance approaches the Alexander-predicted safety margin of 2–4: the tibia and metatarsals are loaded near their fatigue limit, consistent with a species that covers 3000 km/yr.

Ungulate herds exhibit the same three classes of self-organized behavior as starling flocks and fish schools: alignment, cohesion, and separation. Reynolds (1987) showed these three local rules alone produce realistic group motion. Krause & Ruxton (2002) applied fish-school physics to ungulates: each rule has a characteristic distance scale.

Reynolds boid model

Confusion effect (Treisman 1975)

Predators attacking a herd experience a confusion effect: the visual system cannot lock onto a single target when surrounded by many near-identical moving objects. Attention is modeled as a fixed-capacity channel:

Dilution effect

For a predator that captures exactly one prey per attack, each individual in a group of \(N\) has capture probability \(1/N\):

Selfish herd (Hamilton 1971)

Hamilton’s selfish-herd theory states that each individual minimizes its domain of danger—the Voronoi region around it, within which it is the nearest prey to any predator entering. The optimal strategy is movement toward the centroid, producing spontaneous aggregation without explicit cohesion.

Migration as a traveling wave

The spatial advance of the wildebeest front across the Serengeti can be modeled as a Fisher–KPP traveling wave with invasion speed

Two zebra species occupy the African savanna. The plains zebra (Equus quagga) is abundant across East and Southern Africa, while Grevy’s zebra (E. grevyi) is the larger, narrower-striped northern species. Why do zebras have stripes? Four hypotheses have been tested:

Hypothesis 1: crypsis — largely disproven

How & Zanker (2014) measured zebra spectral reflectance and modeled lion visual system detection. At 10 m range under African daylight, lions readily detect zebras. Crypsis cannot account for stripes.

Hypothesis 2: social cohesion

Caro (2014) found weak evidence that striped coats aid kin and conspecific recognition in groups. Statistical signal exists but is small.

Hypothesis 3: tabanid-fly deterrence — LEADING

Caro et al. (2019) ran painted-horse experiments in which horses were dressed in black, white, and striped coats and blood-feeding tabanid flies were counted. Landings were reduced by approximately 80% on striped horses. The proposed mechanism is optical-flow disruption at the scale of fly vision: stripe-period matches the insect’s motion-detecting receptive field, producing aliasing during final approach.

Hypothesis 4: thermoregulation — disproven but nuanced

Larison et al. (2021) tested the hypothesis that alternating black/white stripes set up micro-convection currents that cool the animal. High-resolution thermal imaging showed no meaningful cooling effect. However the temperature difference between black and white stripes (~15 °C at noon) does set up small-scale convection that may aid pelage turnover.

Polarized-reflectance optics

How & Zanker (2014) also examined polarization of reflected light. Tabanid flies preferentially orient to strongly polarized reflections from smooth dark surfaces (often water). Zebra stripes break up the polarization signature, reducing attractiveness:

The African buffalo is a 600–900 kg bovid numbering approximately 800,000 in Africa. It has a fearsome reputation among human hunters (200+ deaths per year) and is a keystone mover of both vegetation structure and pathogen dynamics.

Group “voting” behavior (Prins 1996)

Prins (1996) demonstrated in Manyara buffalo that when the herd rises from a rest bout to graze, the travel direction is determined by a consensus vote. Adult females stand up, face in a chosen direction, and the herd travel-direction is the vector sum of these female orientations. Formally:

Bovine tuberculosis dynamics

Michel et al. (2007) documented bovine tuberculosis (Mycobacterium bovis) in Kruger Park buffalo herds. The pathogen transmits through droplet aerosol during close contact, particularly at shared waterholes. The SIR-like model with herd-size dependence is:

One of the most elegant hypotheses in African ecology, due to Bell (1971), explains the coexistence of multiple ungulate species on shared grassland by body-size-differentiated grass-height niches. Large herbivores process coarse, tall grass profitably because of capacious rumen volume; small herbivores cannot, and require young, tender, protein-rich short grass.

Succession order

1. Zebra first: crops tall coarse grass stems; body mass ~300 kg.
2. Wildebeest second: graze shortened intermediate-height grass; body mass ~250 kg.
3. Thomson gazelle last: takes the remaining tender, short protein-rich regrowth; body mass ~25 kg.

Facilitation vs competition

In the Serengeti the three species show grazing facilitation: zebra clearing tall stems allows wildebeest to access the medium layer, and wildebeest grazing stimulates tender regrowth favorable for gazelle. The coupling is formalized by cross-diffusion in the grass height compartments (simulated below).

Thomson’s gazelle — stotting

Eudorcas thomsonii (25 kg) produces the distinctive stotting (or pronking) behavior: stiff-legged vertical leaps to 60–80 cm even when pursued by predators. FitzGibbon (1988) showed stotting is an honest signal of condition to cheetahs: only well-conditioned gazelles can afford the energetic cost, and cheetahs preferentially abandon pursuit of strong-stotting individuals. The energy cost of a single stott is

Eland — heterothermy

The eland (Taurotragus oryx) is the largest antelope, 500–900 kg, and extreme drought adaptation rests on heterothermy. Taylor (1969) showed elands store body heat during the day, allowing core temperature to rise from 39 to 42 °C, and dump that heat by non-evaporative cooling overnight. This strategy conserves water; eland have the lowest water turnover rate of any savanna bovid:

Oryx — carotid rete mirabile

Oryx (Oryx gazella) possess a carotid rete mirabile—a heat exchanger in which warm arterial blood to the brain is cooled against cool venous blood returning from the evaporative nasal cavity. Brain temperature is held 2–3 °C below body core even as the body heats to 45 °C. Oryx also recycle urea into rumen microbes, recapturing nitrogen without excreting water. This combination (cross-referenced with savanna-adaptation biochemistry in climate-biodiversity M13) lets oryx survive indefinitely without drinking.

Kudu — tannin-induced mortality

In 1982 van Hoven documented a mass-mortality event of greater kudu (Tragelaphus strepsiceros) in South African game ranches. Prolonged browsing on fenced acacia triggered the acacias to raise leaf condensed-tannin concentrations, apparently via an airborne ethylene signal transmitted from heavily browsed trees. Tannin concentrations above 6% of dry weight are lethal to kudu, which lack a sufficient proline-rich salivary protein defence. The cascade illustrates plant-herbivore chemical warfare at the ecosystem scale.

Although the giraffe is treated in detail in Module 3, its migratory ecology couples strongly to ungulate herd dynamics. Giraffes browse the canopy stratum and remove up to 15% of annual acacia leaf production, indirectly opening understory for grazers. Giraffe abundance affects grass height distributions through trophic top-down control on woody encroachment, which in turn modulates Bell grazing succession described in Section 5.

Migration as a traveling wave — redux

Treating the wildebeest migration front as a Fisher-KPP reaction-diffusion wave:

Coupled Lotka-Volterra food web

The savanna herbivore guild is coupled to lions (Module 5), hyenas, cheetahs, and leopards (Module 6) through a generalized Lotka-Volterra system. Stability is preserved because apex predators exert top-down control that prevents any single ungulate from monopolizing forage:

Ungulate herds, starling murmurations, and fish schools share a common statistical mechanics. At the level of coarse-grained variables, the herd field behaves like a self-propelled fluid described by the Toner-Tu continuum theory:

Scale-free correlations in flocks

Cavagna et al. (2010) measured velocity correlations in starling flocks and showed that \(C_{\parallel}(r)\) decays as a scale-free power law with correlation length equal to the flock diameter. This means collective response to a perturbation (a hawk attack, an environmental cue) propagates instantly across the flock. The same phenomenon almost certainly occurs in wildebeest herds but has not been quantitatively verified at the one-million-animal scale.

Information percolation

How does a Mara crocodile alarm propagate across 10,000 wildebeest? Each animal couples to its nearest 6–7 neighbors (Ballerini et al. 2008 topological interaction). Let reaction latency \(\tau_r \sim 100\) ms and inter-animal spacing \(\ell \sim 2\) m. Information wave speed

Phase transition at critical density

Vicsek et al. (1995) showed that self-propelled agents with simple alignment rule undergo a disorder-order phase transition at a critical noise level. In herd biology this corresponds to the dichotomy between scattered grazing (disordered) and migration (ordered).