Module 5: Biomes & Climate

Wladimir Köppen (1884, revised 1918, 1936) classified the world’s climates using only two widely available meteorological variables: monthly mean temperature and monthly mean precipitation. The scheme was subsequently refined by Rudolf Geiger (1954, 1961). Köppen’s great insight was to define climate boundaries by vegetation-relevant thresholds — for example the 10 °C warmest-month isotherm that divides polar from boreal climates corresponds to the polar tree line. The scheme therefore encodes an implicit biogeographic hypothesis: that present-day vegetation distributions are predictable from present-day climate.

Five top-level climate groups (with representative thresholds):

• A — Tropical: coldest-month mean temperature ≥ 18 °C; sub-types Af (rainforest), Am (monsoon), Aw/As (savanna).
• B — Arid: annual precipitation less than aridity threshold \(P_{\text{th}} = 2\bar T + k\) where \(k \in \{0, 140, 280\}\) depending on seasonality; sub-types BW (desert), BS (steppe).
• C — Temperate: coldest-month mean between -3 and 18 °C; sub-types Cfa (humid subtropical), Cfb (marine west coast), Csa/Csb (Mediterranean), Cwa (humid subtropical monsoon).
• D — Continental: coldest month < -3 °C, warmest ≥ 10 °C; sub-types Dfa/Dfb (humid continental), Dfc (subarctic / taiga), Dsa (dry summer cold).
• E — Polar: warmest month < 10 °C; sub-types ET (tundra), EF (ice cap).

Peel, Finlayson & McMahon (2007, Hydrology & Earth System Sciences 11, 1633–1644) produced the canonical modern Köppen–Geiger map at 0.1° resolution from CRU TS climatology, now embedded in most biogeographic analyses. Beck et al. (2018 Scientific Data) released a 1 km Köppen–Geiger product from ERA5 reanalysis, with projected-2100 variants under SSP2-4.5 and SSP5-8.5 that show 22–38% of the land surface shifting Köppen zone by 2100.

Robert H. Whittaker (Communities and Ecosystems, 2nd ed., 1975, Macmillan) plotted the world’s biomes as polygons in a 2-D space of mean annual temperature vs mean annual precipitation. The Whittaker diagram became the pedagogical icon of macroecological biogeography because it captures, in two variables, the first-order partition of Earth’s terrestrial vegetation into biomes:

• Tundra: very cold (< 0 °C MAT), any precipitation.
• Boreal forest / taiga: 0–5 °C MAT, 300–900 mm MAP.
• Temperate rainforest: 5–15 °C MAT, > 2000 mm MAP.
• Temperate deciduous forest: 5–15 °C MAT, 750–1500 mm MAP.
• Temperate grassland: 5–15 °C MAT, 300–600 mm MAP.
• Desert: hot or cold, < 250 mm MAP.
• Tropical savanna / seasonal forest: > 15 °C MAT, 700–1800 mm.
• Tropical rainforest: > 15 °C MAT, > 2000 mm MAP.

The Whittaker diagram is a dimensional reduction of the full climate space. The two chosen axes are not arbitrary: temperature sets the thermal energy for photosynthesis and the length of the growing season, while precipitation sets water availability. Evapotranspiration, which combines them, is the single best climatic predictor of net primary productivity (Lieth 1975). More complex climate classifications (Holdridge life zones, 1967) use up to 30 biotemperature and evapotranspiration-ratio classes, but the two-dimensional Whittaker plot remains the dominant teaching tool.

Alexander von Humboldt (1807) and Alfred Russel Wallace (1878) recorded that species richness increases toward the equator in almost every terrestrial and marine taxon. Hillebrand (2004 American Naturalist 163, 192–211) performed the canonical meta-analysis: across > 600 independent taxa he found a mean log-linear slope of \(\partial \log S / \partial \text{lat}\)of approximately −0.02 per degree latitude, with the overwhelming majority of taxa showing tropical-peak patterns and virtually none showing temperate peaks.

Exceptions exist (ichneumonid wasps, peatland bryophytes, some benthic marine groups) but they are the minority. The LDG is the single most robust biogeographic generalisation.

Mittelbach et al. (2007 Ecology Letters, “Evolution and the latitudinal diversity gradient”) reviewed 25 distinct explanations. They cluster into five families:

• Productivity hypothesis (Wright 1983): more solar energy at the equator supports more individuals, which by the more-individuals hypothesis supports more species. Empirical fits yield \(S \propto NPP^{\alpha}\) with \(\alpha \approx 0.3\text{--}0.5\).
• Area hypothesis (Rosenzweig 1995): the tropics occupy more area than the subtropics (because climate zones form bands on a sphere). Under a species-area law \(S = cA^{z}\), tropical area excess translates directly to richness excess.
• Time / climatic stability(Fischer 1960; Jablonski 1993): the tropics have been climatically stable since at least the Eocene, allowing long accumulation of species without the Pleistocene glacial extinctions that cleared temperate and boreal biotas.
• Biotic interactions(Dobzhansky 1950): at the equator biotic interactions (predation, parasitism, competition) are stronger selective forces than abiotic stresses, driving faster speciation and narrower niches.
• Ecological limits / niche saturation(Rabosky 2009 Ecology Letters): the LDG reflects a higher carrying capacity for species in the tropics; diversification rates equilibrate to the energetic and spatial resources available, not to the rate of species production.

These mechanisms are not mutually exclusive. Modern analyses (Jablonski et al. 2017 Science) favour a synthesis in which the tropics are both a cradle of speciation and a museum of lineage accumulation, with productivity and stability setting the carrying capacity and biotic interactions driving the rate of filling it.

Within tropical and temperate mountain systems, species richness often peaks at mid-elevations rather than at the base (Rahbek 2005 Ecology Letters, meta-analysis of 204 elevational gradients). Four common shapes are recognised: monotonic decline with elevation; mid-elevational peak (the most common pattern); low-plateau with sharp decline; and bimodal peaks. The mid-elevation peak is commonly interpreted as the mid-domain effect of McCain (2004) plus a secondary moist-belt peak in cloud forest at ~1500–2500 m in the tropics.

Elevational gradients offer a natural analogue of the latitudinal gradient over a vertical kilometre: every ~170 m of elevation corresponds roughly to 1° of latitude in temperature. This compressed climate gradient is a standard setting for climate-change range-shift studies and for the “escalator to extinction” phenomenon (Harris 2018) where species on tropical sky islands are pushed upslope with no remaining habitat.

Species distribution models (SDMs) relate presence/absence data to climatic predictors and map the species’ realised niche into geographic space. The dominant current algorithm is MaxEnt (Phillips, Anderson & Schapire 2006 Ecological Modelling), which uses maximum-entropy inference on presence-only data.

MaxEnt computes the distribution \(P(x)\) of maximum entropy subject to feature constraints:

MaxEnt and its relatives (Boosted Regression Trees, Random Forest SDMs, generalised additive models) predict present-day distributions well in the climate space where calibration data exist but extrapolate poorly into novel-climate regions — the fundamental limit of correlative SDMs when projecting to future climates. Mechanistic niche models (Kearney & Porter 2009 Ecology Letters) use species-specific physiological limits instead and are more robust to extrapolation, though harder to parameterise.

Parmesan & Yohe (2003 Nature), updated in Parmesan (2006 ARES) documented a mean range shift of ~17 km per decade poleward and ~11 m per decade upslope across > 1700 species for which long-term distribution records exist. The figure has been validated and extended by Chen et al. (2011 Science, mean 16.9 km/decade poleward). These are mean shifts; some species shift an order of magnitude faster, while others (sedentary trees with long generation times) lag behind their climate.

The Coupled Model Intercomparison Project Phase 6 (CMIP6, IPCC AR6) projects under the SSP2-4.5 “middle of the road” scenario a ~2.7 °C global warming by 2100. Projected biome-scale impacts:

• Boreal treeline advance of 100–500 km northward (Kaplan et al. 2003); tundra loss of 30–50% by area.
• Mediterranean-type biome expansion in southern Europe, southwestern Australia, Cape Floristic Region; upward shift of Mediterranean/montane ecotone.
• Amazon “savannisation”: tropical rainforest converts to seasonal savanna under drying scenarios (Cox et al. 2000, 2013; Lovejoy & Nobre 2018 Science Advances).
• Sahel greening (Herrmann et al. 2005): CO2 fertilisation and localised monsoon intensification.
• Polar and subpolar marine ecosystem reorganisation: sea-ice loss, ocean acidification, and warming drive polar-to-temperate range shifts of fish.

Beck et al. (2018) produced CMIP6-derived Köppen classifications for 2071 –2100 at 1 km resolution, showing that under SSP2-4.5 approximately 22% of the land area changes Köppen zone, rising to 38% under SSP5-8.5. The most-affected classes are tundra (ET, -37%), boreal (Dfc, -25%) and cold semi-arid (BSk, +13%).

Mediterranean-type ecosystems (Csa/Csb in Köppen; hot dry summers and cool wet winters) occur on five continents in geographically disjunct regions: the Mediterranean Basin, California, central Chile, the Cape Floristic Region of South Africa, and southwestern Australia. The five regions have converged on remarkably similar physiognomic vegetation — evergreen sclerophyll shrubland dominated by Quercus, Ceanothus, Cryptocarya, Protea and Banksia respectively (Mooney & Dunn 1970 American Naturalist).

Despite the convergent physiognomy, the five regions have very different species richness: the Cape Floristic Region holds ~9000 plant species in 90,000 km² (about 100 species per km²), more than twice the density of any other Mediterranean-type biome. SW Australia is a close second. The California and Chile systems, recently derived from tropical climate, are much poorer. The Mediterranean Basin holds rich flora but shows strong human modification over 8000 years. The Cape and SW Australia are classic examples of Myers’ (2000 Nature) biodiversity hotspots.

Arid biomes (Köppen B) cover ~33% of the terrestrial surface. They are not biodiversity deserts: the Succulent Karoo of southern Africa hosts ~5000 plant species, the richest arid flora on Earth (Jürgens 1991 Madoqua). Deserts illustrate the decoupling of biomass and species richness: low productivity yet high endemism, driven by topographic micro-heterogeneity, fog-belt coastal enrichment (Atacama, Namib), and long evolutionary history of arid-adapted clades (Cactaceae in the New World, Aizoaceae in the Old).

Desert biogeography shows strong convergence: cacti (Americas) and euphorbs (Africa) converge on succulent stem architecture; Namib beetles (Stenocara) and Californian rove beetles independently evolve fog-drinking behaviour. The convergence underscores the climate-vegetation coupling that motivates Whittaker’s diagram.

The boreal forest (taiga, Köppen Dfc) is the largest terrestrial biome by area (~15 × 10&sup6; km²), dominating Canada, Siberia and Scandinavia. Its species richness is low (a dozen dominant tree species globally) but its carbon stocks are enormous: boreal soils hold 1000–1700 Pg C, roughly twice atmospheric CO2, mostly in deep peat and permafrost (Bradshaw & Warkentin 2015 Global & Planetary Change). Warming-driven permafrost thaw and boreal wildfire represent the largest single terrestrial carbon-climate feedback in CMIP6 projections.

Biogeographically the boreal forest exhibits a circumpolar but compositionally disjunct structure: North American taiga is dominated by Picea glauca, Picea mariana, Pinus banksiana and Populus tremuloides, while Eurasian taiga is dominated by Larix sibirica/dahurica, Picea obovata, Pinus sylvestris. This split reflects both the Bering filter and the climate tolerance differences (Larix is deciduous-needled, tolerating the extreme continentality of northeastern Siberia that kills North American evergreen conifers).

Marine biome classification is not based on Köppen or Whittaker but on three physical variables: sea-surface temperature (SST), salinity, and photic depth (Longhurst 2007, Ecological Geography of the Sea, 2nd ed., Academic Press). Longhurst’s scheme divides the world ocean into 56 provinces clustered into four domains:

• Polar Domain: seasonal ice-edge blooms, high latitude, Arctic and Southern Ocean provinces.
• Westerlies Domain: mid-latitude, strong spring bloom, high production (N. Atlantic, N. Pacific).
• Trade-winds Domain: subtropical oligotrophic gyres, low production, high clarity.
• Coastal Domain: shelf seas and upwelling zones, high production, high diversity.

The marine LDG is weaker than the terrestrial one and inverted in some taxa (calanoid copepods, Chaetognatha). Coral reefs show the classic tropical peak with the Indo-Malayan-Philippine “Coral Triangle” containing the highest species density of reef fishes and corals on the planet (Hoeksema 2007 Biogeography of the Sea).

Tropical mountain systems isolated by low-elevation tropical lowlands form sky-island archipelagos. The classic case is the East African “montanes”: Mt Kenya, Kilimanjaro, the Aberdares, the Rwenzoris and Mt Elgon each host endemic tree heath (Erica) forest, Afroalpine Dendrosenecio and Lobelia giant rosettes, and endemic birds. Similar sky-island patterns are found in the Madrean archipelago of SW USA / N Mexico, the Brazilian Atlantic mountains, and the Papuan New Guinea highlands.

Sky-island biogeography combines island biogeographic theory (Module 2) with the vertical climate gradient of this module: each summit is a thermal “island” whose area shrinks as climate warms. La Sorte & Jetz (2010 Proc. Roy. Soc. B) projected that ~80% of tropical montane bird species will lose >50% of their thermal habitat by 2100 under SSP5-8.5. Because upward migration is constrained by the mountain summit itself, tropical sky-island biotas face the sharpest climate-change extinction risk of any terrestrial biome.