Module 13: African Savannas — Fire, Grazers & Megafauna
African savannas occupy roughly half of the continent and concentrate the last great terrestrial migrations and megafauna assemblages on Earth. This module develops the fire–grazing–rainfall triangle of savanna maintenance, Walter’s two-layer soil-water coexistence theory, the Serengeti–Mara wildebeest system, Sahel drought–greening oscillations, and the intertwined fates of lions, elephants, vultures and pastoralists under CMIP6 projections.
1. The Savanna Biome & Tree–Grass Coexistence
Savannas are open-canopy woodlands dominated by a continuous C4 grass understorey and discontinuous C3 tree cover. They cover roughly 13×106 km2 in sub-Saharan Africa and support more large-bodied mammals than any other biome. The C4 photosynthetic pathway (Hatch–Slack) concentrates CO2 around RuBisCO, suppressing photorespiration at high temperatures and maintaining high water-use efficiency where radiation and temperature are high and moisture is seasonal.
The Fire–Grazing–Rainfall Triangle
Bond (2008) showed that savannas globally are under-canopied relative to the tree cover rainfall could support, with the difference closed by fire and herbivory. The stable mosaic arises as a dynamical balance:
\[\frac{dT}{dt} = g(R)\,T\!\left(1-\tfrac{T}{K(R)}\right)-\mu_f\,F(G)\,T-\mu_h\,H(T,G)\]
where \(T\) is tree cover, \(F\) fire frequency, \(H\) herbivore browsing, and \(R,G\) rainfall and grass biomass.
Walter’s Two-Layer Hypothesis
Walter (1971) proposed that trees and grasses coexist because their roots exploit different soil-water pools: grasses access the shallow layer, trees the deep layer. Rewritten as a resource-partitioning model:
\[W_{\text{shallow}}\to \text{grass},\quad W_{\text{deep}}\to \text{tree},\quad W_{\text{deep}}=\alpha(R-R_c)\]
Deep infiltration only occurs when rainfall exceeds a critical value \(R_c\approx 400\) mm.
Modern demographic models (Higgins, Bond & Trollope 2000) replace the static two-layer view with a stochastic fire-trap dynamic in which tree saplings must escape a fire-prone height band (<2 m) before the next burn; rainfall sets maximum growth rate and fire return interval sets mortality.
Fire–Grazing–Rainfall Triangle
2. Serengeti–Mara & the Wildebeest Migration
The Serengeti–Mara ecosystem (~25,000 km2) hosts the annual migration of approximately 1.5 million white-bearded wildebeest (Connochaetes taurinus mearnsi), 350,000 Thomson’s gazelle, and 200,000 plains zebra. The migration tracks the rainfall gradient between the short-grass plains in the south-east (wet season) and the Mara in the north-west (dry season). Holdo, Holt & Fryxell (2009) showed that GPS-collared wildebeest movement velocity scales with the spatial gradient of NDVI greenness, itself modulated by rainfall.
\[\vec{v}_{\text{migr}}\;\propto\;\nabla\!\!\;\text{NDVI},\qquad \text{NDVI}=f(R,T,\text{soil})\]
Rinderpest Release (1960s)
Prior to the 1960s, rinderpest (a morbillivirus introduced in 1887 with Italian cattle) suppressed wildebeest and buffalo at roughly a fifth of carrying capacity. Vaccination of domestic cattle throughout the 1950s removed the reservoir; Sinclair (1979) documented the subsequent numerical response, with wildebeest rising from ~250,000 to 1.4 million by 1977. This top-down cascade increased grazing pressure, shortened fire-return intervals, and is now a textbook example of a trophic release.
Tri-Trophic Dynamics
The Sinclair–Hilborn–Mduma (Mduma et al. 1999) system couples wildebeest population to rainfall-driven calf recruitment and to lion-plus-hyena predation with a Holling Type-II response:
\[\frac{dN}{dt}=rN\!\left(1-\tfrac{N}{K(R)}\right)-\frac{aN}{1+ahN}\cdot P\]
\(K(R)=c_K R\); prey-limited predation is swamped once \(N\gg (ah)^{-1}\).
Serengeti–Mara Migration Loop
3. Sahel: Drought, Greening & the Hydrological Paradox
The Sahel is the semi-arid belt (250–600 mm/yr) between the Sahara and the Sudanian savanna. Hulme (2001) documented a severe drying trend from the late 1960s through the 1980s — one of the largest regional climate anomalies of the 20th century — followed by a partial recovery since the 1990s. Dai (2013) quantified the drought signal through the Palmer Drought Severity Index (PDSI):
\[\text{PDSI}\approx \frac{P-\text{PET}_{\text{Penman}}}{\sigma_{P-\text{PET}}}\]
PDSI integrates precipitation anomaly and potential evapotranspiration from the Penman–Monteith equation.
The Hydrological Paradox
Counter-intuitively, Descroix et al. (2009) showed that runoff coefficients in the Niger and Mekrou basins increased during the drought and have remained elevated despite precipitation recovery. The mechanism is a soil-crust/land-cover hysteresis: repeated dry years destroyed herbaceous cover, sealed the soil surface by raindrop impact, and shifted the water balance away from infiltration:
\[Q/P\;=\;\phi(\text{vegetation cover},\;\text{crust area})\;\nearrow\;\text{despite}\;P\nearrow\]
Re-Greening & CMIP6 Outlook
Satellite records (NOAA AVHRR, MODIS) show widespread greening of the Sahel since ~1985 (Herrmann et al. 2005) attributable to rainfall recovery and (locally) farmer-managed natural regeneration. Almazroui et al. (2020) show CMIP6 multi-model ensemble Sahel warming exceeds 2× the global mean under SSP5-8.5, with wet-season rainfall projections diverging across models — some drying, others strongly wetting.
4. Congo Forest–Savanna Mosaic & Elephants
The Congo basin forms a gradient from closed-canopy rainforest to derived savanna. Doumenge et al. (2015) mapped the mosaic and documented a net forest-cover loss of 2.5% per decade at the northern and southern fringes. The savanna elephant (Loxodonta africana) and forest elephant (Loxodonta cyclotis) were confirmed as distinct species (Rohland et al. 2010); forest elephants have a reproductive generation time of ~31 years versus ~22 in savanna populations.
Keystone Engineering & Stable States
Dublin, Sinclair & McGlade (1990) showed that elephants and fire maintain alternative stable states in the Serengeti–Mara: a “grassland attractor” and a “woodland attractor,” separated by a tipping point parameterised by elephant density and burn frequency. The switching manifold satisfies:
\[\mu_f\,F+\mu_e\,E\;>\;g(R)\quad\Longrightarrow\quad \text{grassland attractor}\]
Poaching Dynamics
Wittemyer et al. (2014) modelled carcass-ratio data to infer that ~8% of African elephants were killed annually during the 2010–2012 ivory spike, with forest elephants declining by ~62% between 2002–2011 (Maisels et al. 2013). Population dynamics combine low intrinsic growth with poaching mortality:
\[\frac{dN}{dt}=rN\!\left(1-\tfrac{N}{K}\right)-(\mu+q\,I_{\text{ivory}})N\]
\(I_{\text{ivory}}\) is the standardized ivory price index; \(q\) is poaching catchability.
5. Megafauna, Carnivores & Giraffe Scaling
Sub-Saharan savannas host the world’s richest intact large-mammal community: bush elephant (Loxodonta africana), Cape buffalo (Syncerus caffer), giraffe (Giraffa camelopardalis), hippopotamus, black and white rhino, topi, hartebeest, eland, kudu.
Giraffe Neck & Allometric Scaling
Giraffes have seven cervical vertebrae — the mammalian norm — but each is elongated to ~28 cm. Vertebral elongation scales with a power law \(L_v\propto M^{0.28}\), while cardiovascular pressure required to perfuse the brain scales as \(P\propto h\rho g\). A 2.5 m neck plus heart-to-brain elevation of ~3 m yields arterial pressures of ~280/180 mmHg — nearly twice human values (Mitchell & Skinner 2003).
Lion Population Decline
Bauer et al. (2015) pooled 47 long-term lion (Panthera leo) monitoring sites, finding a 43% continent-wide decline between 1993 and 2014, with West and Central African populations collapsing by ~66%. East and Southern African populations inside large, well-funded protected areas remained stable or grew; the pattern is consistent with a human-footprint-driven extinction filter.
Cheetah & African Wild Dog
Cheetahs (Acinonyx jubatus) have the largest per-capita home range of any African carnivore (mean ~1,600 km2) and are consequently vulnerable to habitat fragmentation. The African Lion and Cheetah Vision (ALV) habitat-shrinkage map shows an ~91% loss of historic range. African wild dogs (Lycaon pictus) face an additional thermo-regulatory problem: chase-hunting at high body-core temperatures (>40°C) forces them to hunt at dawn and dusk, and Woodroffe et al. (2017) showed that each 1°C of warming correlates with 14% fewer pups weaned.
6. Birds, Small Mammals & Pastoralism
Africa’s Vulture Crisis
Ogada et al. (2016) showed that seven of eleven African vulture species have declined ~80–97% over three generations. The dominant drivers are carbofuran / diclofenac poisoning (targeted and secondary), belief-based use and power-line collisions. Egyptian vultures (Neophron percnopterus) fell ~92%; white-backed vultures (Gyps africanus) ~90%. Vulture collapse cascades into carcass persistence and zoonotic disease risk.
The African fish eagle (Haliaeetus vocifer), secretary bird (Sagittarius serpentarius, declining ~37%), red-billed and yellow-billed oxpeckers (Buphagus), ground hornbills (Bucorvus) and southern masked weavers round out a richly structured avian community whose phenology tracks rainfall onset.
Oxpecker–Ungulate Mutualism
Oxpeckers remove ticks from the hides of ungulates while also feeding at wounds. Modelled as a facultative mutualism with cheating:
\[\frac{dU}{dt}=r_U U\!\left(1-\tfrac{U}{K_U}\right)+\beta\,O U -\gamma\,O U\]
with \(\beta\) the tick-removal benefit and \(\gamma\) the wound-parasitism cost.
Meerkat Thermal Ecology
Clutton-Brock et al. (1999) documented meerkat (Suricata suricatta) social behaviour in the southern Kalahari, with cooperative breeding and pup-feeding correlated with rainfall and insect availability. Meerkats regulate body temperature by alternating basking (morning) and burrow-retreat (midday) — a two-mode Newton’s-cooling problem.
Maasai Pastoralism & Fencing
Homewood et al. (2001) linked Maasai livestock density to wildlife abundance across the Serengeti–Mara, showing that moderate pastoralism coexists with wildlife via nutrient pulses (boma manure hotspots) but that fencing truncates migration corridors. Hobbs et al. (2008) modelled the herding-vs-fencing trade-off with a movement-cost function \(C(x)=c_0+c_1 \|\nabla \phi_{\text{fence}}\|\) that captures rising energetic expenditure near barriers.
7. Pyrogeography & CMIP6 Projections
Archibald et al. (2013) synthesised “pyromes” worldwide — clusters of fire regime defined by return interval, mean fire size and seasonality. African savannas occupy the high-frequency/low-intensity pyrome, with mean fire return intervals of 2–5 years. MODIS burned-area products (MCD64A1) estimate ~180–200×106 ha burnt annually across Africa — roughly two-thirds of the global burnt area.
\[\text{Fire intensity}\;I\;=\;H\,w\,R_s\quad(\text{kW m}^{-1})\]
\(H\) fuel heat yield, \(w\) fuel load, \(R_s\) rate of spread (Byram 1959).
Wet-Bulb Exceedance
Wet-bulb temperature \(T_w\) above 35°C is the physiological survival threshold for humans in shade (Sherwood & Huber 2010). CMIP6 projections show East Africa approaching recurrent \(T_w > 32\)°C days by late century under SSP5-8.5, with large-ungulate heat-budget implications.
Almazroui 2020 Warming
Almazroui et al. (2020) show CMIP6 median warming across the Sahel & East Africa reaches 4.5°C by 2100 under SSP5-8.5, with amplified variance. Projections of wet-season rainfall are bimodal: dynamical models favour strengthened WAM (West African Monsoon) and wetter Sahel, while statistical downscaling favours drying.
Simulation 1: Serengeti Wildebeest Dynamics
Tri-trophic wildebeest model following Sinclair–Hilborn–Mduma, forced by Serengeti wet-season rainfall reconstructions extended with CMIP6 SSP scenarios. The simulation includes rinderpest-release calibration, rainfall anomaly-driven calf recruitment (Holdo 2009), logistic competition and Holling Type-II lion–hyena predation. Cumulative forage is integrated using \(\texttt{np.trapezoid}\).
Click Run to execute the Python code
Code will be executed with Python 3 on the server
Simulation 2: Lion Range Suitability SDM
Species-distribution model for African lion (Panthera leo) with four covariates: prey biomass, rainfall envelope, human footprint (Venter 2016 proxy) and livestock conflict (Kissui 2008). Logistic suitability maps are combined with a temporal decline model calibrated to Bauer et al. (2015) — 43% pan-African decline 1993–2014.
Click Run to execute the Python code
Code will be executed with Python 3 on the server
8. Biochemistry of Savanna Adaptation: How Animals Survive the Heat
Savanna fauna face a triple stress: air temperatures exceeding 40 °C for months, water restricted to a short rainy season, and forage dominated by C4 grasses that are nutritionally poor and chemically defended. The adaptive solutions span six biological scales — vascular anatomy, renal chemistry, haemoglobin allostery, evaporative physiology, rumen microbiology, and heat-shock molecular biology.
8.1 Selective Brain Cooling: The Carotid Rete Mirabile
Oryx (Oryx gazella), Grant’s gazelle and Thomson’s gazelle tolerate core body temperatures of 46.5 °C — lethal for most mammals — because arterial blood is pre-cooled at the base of the skull by the carotid rete (Taylor & Lyman 1972). Several hundred millimetre-thin arterioles pass through the cavernous sinus where venous blood, chilled by nasal evaporation, flows in counterflow. The heat-exchanger efficiency is
\[ \varepsilon = \frac{T_{a,in} - T_{a,out}}{T_{a,in} - T_{v,in}} = 1 - \exp\!\left[-\text{NTU}\,(1-C_r)\right]\cdot\left[1 - C_r\,\exp\!\left(-\text{NTU}\,(1-C_r)\right)\right]^{-1} \]
with NTU ≈ UA/(ṡcp) typically 2–3 in the rete. Brain blood thus arrives 2–3 °C cooler than rectal core, sparing the hypothalamus while the body stores heat that can later be dumped at night — a strategy dubbed adaptive heterothermy (Schmidt-Nielsen 1964; Mitchell 2002).
8.2 Water Economy: Urea Recycling and Renal Countercurrent Extremes
The dik-dik (Madoqua) and oryx achieve urine osmolarities of 3 000–5 000 mOsm/L — compared with 1 200 in humans — via exceptionally long loops of Henle in the inner medulla. The maximum concentration ratio U/P follows the renal-length scaling
\[ \frac{U_{osm}}{P_{osm}} \approx C_1 \left(\frac{L_{Henle}}{D_{papilla}}\right)^{0.5} \]
and is amplified by urea recycling between the collecting duct and the loop via UT-A/UT-B transporters. Ungulates with access to low-quality forage (giraffe, Grant’s gazelle) divert 60–80% of blood urea back to the rumen, where microbial urease converts it to NH4+for bacterial protein synthesis (Houpt 1963, Schmidt-Nielsen 1957). This tightens the nitrogen cycle and lets animals survive on < 4% dietary crude protein during the dry season.
8.3 Haemoglobin Allostery at Altitude and Heat
Highland savanna species (Ethiopian gelada, Kilimanjaro elephant shrew) show elevated 2,3-bisphosphoglycerate (2,3-BPG)which right-shifts the oxy–haemoglobin curve and releases O2at tissue p50. The Hill equation with cooperativity n ≈ 2.7
\[ Y(P_{O_2}) = \frac{P_{O_2}^{\,n}}{P_{50}^{\,n} + P_{O_2}^{\,n}}, \qquad P_{50} = P_{50}^{0}\,10^{-0.48\,\Delta\text{pH} + 0.024\,\Delta T} \]
captures the Bohr and temperature effects. For a gazelle running at 60 km/h in 41 °C heat, blood-lactate pH drops to 7.10 and arterial blood temperature rises to 41.5 °C, together pushing p50 from 26 to 38 mmHg — unloading an extra 24% of O2 per pass (Jessen 2001).
8.4 Sweating, Panting and the Elephant Radiator
Cape buffalo (Syncerus caffer) possess only apocrine sweat glands of moderate density (~150/cm2) and rely on wallowing. Lions and cheetahs pant at 200–300 breaths/min, exploiting the resonant frequency of the respiratory system to dissipate heat via nasal turbinate evaporation without hyperventilatory alkalosis (Richards 1970). African elephants (Loxodonta africana) lack functional sweat glands and instead use their ears as radiators: flapping at 0.5 Hz augments convective coefficient h from 5 to 25 W·m-2·K-1, dumping up to 100 W of metabolic heat through the ear pinna’s rich subcutaneous vascular network (Williams 1990, Koffi 2013). Interestingly, only African elephants (not Asian) have five TRPV1 copies, a variant linked to heightened heat sensing (Weissenböck 2010).
8.5 Rumen Biochemistry on C4 Grass
Savanna C4 grasses are fortified with silica phytoliths (~15% dry weight) and condensed tannins. Wildebeest and zebra coevolved proline-rich salivary proteins (PRPs)that bind tannins with affinities Kd≈10-6 M, preventing enzymatic inhibition in the rumen (Mehansho 1987). In parallel, rumen microbial communities (Fibrobacter succinogenes, Ruminococcus albus) express cellulase/xylanase cocktails tuned to the 40 °C rumen, with optimum activity at pH 6.3–6.7. Oxalate-rich dry-season forage is detoxified via Oxalobacter formigenes, whose formyl-CoA transferase converts oxalate to formate + CO2. Ruminants also secrete Ca2+-oxalate crystalsin urine to excrete otherwise toxic oxalate loads.
8.6 Trypanotolerance: Disease Biochemistry under Tsetse Pressure
In tsetse-fly (Glossina) belts, the N’Dama cattle breed and native buffalo tolerate Trypanosoma congolense while European zebu die within months. N’Dama show elevated serum haptoglobin that scavenges haem released by parasite lysis, plus a polymorphism in the APOL1 gene encoding apolipoprotein L1, a trypanolytic factor that disrupts trypanosome lysosomal membranes via ionic pore formation (Murray 1982, Vanhamme 2003). This is the same locus that, in humans of West-African descent, confers resistance to T. brucei rhodesiense but predisposes to kidney disease — a textbook example of antagonistic pleiotropy.
8.7 Heat-Shock Proteins and Thermal Adaptation
Savanna endemics constitutively express HSP70at levels 2–4× higher than temperate-zone congeners (Collier 2006 in cattle; Basiricò 2011). The stress-induced HSP70-1A promoter carries G3144C and C1128T SNPs associated with thermotolerance; Boran and N’Dama cattle carry the favourable alleles near fixation whereas Holstein imports do not. Transcriptional burst kinetics
\[ \frac{d[\text{HSP70}]}{dt} = k_0 + k_{max}\,\frac{(T-T_{th})^{n}}{K^{n}+(T-T_{th})^{n}} - \lambda [\text{HSP70}] \]
with threshold Tth≈38.5 °C, Hill coefficient n≈6 and decay λ ≈ 0.05 h-1 produce the characteristic pulse response that protects denaturing enzymes. Climate warming narrows the safety margin ΔT = Tlethal − Tacclim; by 2050 large portions of East Africa may exceed the thermotolerance ceiling for indigenous zebu (Rahimi 2021), with cascading effects on pastoralist livelihoods.
8.8 Simulation: Oryx Heat-Balance and HSP Induction
Below we solve a two-compartment heat-balance model for an 80 kg oryx during an 11-hour daytime transient at Tair = 42 °C, tracking core vs. brain temperature (cooled by the rete) and the induced HSP70 response via a Hill transcriptional burst.
Click Run to execute the Python code
Code will be executed with Python 3 on the server
Key References
• Bond, W. J. (2008). “What limits trees in C4 grasslands and savannas?” Annual Review of Ecology, Evolution, and Systematics, 39, 641–659.
• Walter, H. (1971). Ecology of Tropical and Subtropical Vegetation. Oliver & Boyd.
• Higgins, S. I., Bond, W. J. & Trollope, W. S. W. (2000). “Fire, resprouting and variability: a recipe for grass–tree coexistence in savanna.” Journal of Ecology, 88, 213–229.
• Sinclair, A. R. E. (1979). “The eruption of the ruminants.” In Serengeti: Dynamics of an Ecosystem, University of Chicago Press.
• Mduma, S. A. R., Sinclair, A. R. E. & Hilborn, R. (1999). “Food regulates the Serengeti wildebeest: a 40-year record.” Journal of Animal Ecology, 68, 1101–1122.
• Holdo, R. M., Holt, R. D. & Fryxell, J. M. (2009). “Opposing rainfall and plant nutritional gradients best explain the wildebeest migration in the Serengeti.” American Naturalist, 173, 431–445.
• Hulme, M. (2001). “Climatic perspectives on Sahelian desiccation: 1973–1998.” Global Environmental Change, 11, 19–29.
• Dai, A. (2013). “Increasing drought under global warming in observations and models.” Nature Climate Change, 3, 52–58.
• Descroix, L. et al. (2009). “Spatio-temporal variability of hydrological regimes around the boundaries between Sahelian and Sudanian areas of West Africa.” Journal of Hydrology, 375, 90–102.
• Herrmann, S. M., Anyamba, A. & Tucker, C. J. (2005). “Recent trends in vegetation dynamics in the African Sahel.” Global Environmental Change, 15, 394–404.
• Almazroui, M. et al. (2020). “Projected change in temperature and precipitation over Africa from CMIP6.” Earth Systems and Environment, 4, 455–475.
• Doumenge, C. et al. (2015). “Tropical Africa’s forests: carbon, biodiversity, and ecological functioning.” Forests, 6, 4019–4042.
• Dublin, H. T., Sinclair, A. R. E. & McGlade, J. (1990). “Elephants and fire as causes of multiple stable states in the Serengeti-Mara woodlands.” Journal of Animal Ecology, 59, 1147–1164.
• Wittemyer, G. et al. (2014). “Illegal killing for ivory drives global decline in African elephants.” PNAS, 111, 13117–13121.
• Maisels, F. et al. (2013). “Devastating decline of forest elephants in Central Africa.” PLOS ONE, 8, e59469.
• Bauer, H. et al. (2015). “Lion (Panthera leo) populations are declining rapidly across Africa, except in intensively managed areas.” PNAS, 112, 14894–14899.
• Woodroffe, R. et al. (2017). “Hot dogs: High ambient temperatures impact reproductive success in a tropical carnivore.” Journal of Animal Ecology, 86, 1329–1338.
• Ogada, D. et al. (2016). “Another continental vulture crisis: Africa’s vultures collapsing toward extinction.” Conservation Letters, 9, 89–97.
• Clutton-Brock, T. H. et al. (1999). “Selfish sentinels in cooperative mammals.” Science, 284, 1640–1644.
• Homewood, K. et al. (2001). “Long-term changes in Serengeti-Mara wildebeest and land cover: pastoralism, population, or policies?” PNAS, 98, 12544–12549.
• Hobbs, N. T. et al. (2008). “Fragmentation of rangelands: Implications for humans, animals, and landscapes.” Global Environmental Change, 18, 776–785.
• Archibald, S. et al. (2013). “Defining pyromes and global syndromes of fire regimes.” PNAS, 110, 6442–6447.
• Mitchell, G. & Skinner, J. D. (2003). “On the origin, evolution and phylogeny of giraffes.” Transactions of the Royal Society of South Africa, 58, 51–73.
• Venter, O. et al. (2016). “Sixteen years of change in the global terrestrial human footprint.” Nature Communications, 7, 12558.
• Rohland, N. et al. (2010). “Genomic DNA sequences from mastodon and woolly mammoth reveal deep speciation of forest and savanna elephants.” PLOS Biology, 8, e1000564.