Module 2: Island Biogeography

In The Theory of Island Biogeography (Princeton, 1967), Robert MacArthur and Edward O. Wilson proposed that species richness on an island is maintained at a dynamic equilibrium between immigration of new species from a mainland source pool and extinction of species already present. Let \(P\) be the size of the mainland pool and \(S\) the number of species currently on the island. Then:

The equilibrium \(S^{\ast}\) is obtained by setting \(I(S) = \mu(S)\):

Area and Isolation Enter via Rate Parameters

The beauty of the MacArthur–Wilson (MW) model is how island area and isolation enter:

• Larger islands have smaller \(E_0\) (lower per-species extinction rate, because larger populations are less prone to demographic stochasticity).
• More isolated islands have smaller \(I_0\) (fewer colonists arrive per unit time).

Both effects push \(S^{\ast}\) down as islands become smaller or more isolated, and both are confirmed empirically across countless island systems.

Arrhenius (1921) observed empirically that species richness scales with area as a power law:

MacArthur & Wilson gave the relation a mechanistic grounding. If island area controls the extinction rate through population size, and if populations follow the lognormal abundance distribution (Preston 1962), the resulting species–area relation has an exponent \(z\) that depends on how rapidly extinction increases as populations shrink:

The difference between oceanic-island \(z \approx 0.25\) and nested-continental \(z \approx 0.1\) is itself a theoretical prediction of MW: continental samples share one species pool and many dispersal events, so extinction is mitigated by the “rescue effect” (Brown & Kodric-Brown 1977). True oceanic islands do not enjoy this rescue and therefore have steeper species–area curves.

The Target-Area Effect

Larger islands also receive more colonists per unit time because they present a larger target to dispersing propagules (Gilpin & Diamond 1976). This effect adds an area contribution to the immigration rate as well as the extinction rate:

Daniel Simberloff and E. O. Wilson performed the first experimental test of MW theory in the Florida Keys in 1966–1967 (reported 1969 Ecology 50, 1970 Ecology 51). They selected six small red-mangrove (Rhizophora) islets ranging from 11 to 18 m diameter and at different distances from a source mangrove forest. Each islet was fumigated with methyl bromide under a plastic tent, removing all arthropods while leaving the trees alive. Arthropod recolonisation was then monitored for ~1 year.

The results were paradigm-confirming:

• Species richness recovered to within \(\pm\)10% of pre-fumigation equilibrium in 200–300 days.
• The recovery followed the MW saturation curve \(S(t) = S^{\ast}(1 - e^{-(I_0 + E_0)t})\).
• Equilibria scaled with both islet area and distance-to-source, as predicted.
• Species composition turned over substantially: identity of recolonists was stochastic, but total richness was deterministic. This is the signature prediction of a dynamic equilibrium.

The experimental manipulation by Simberloff & Wilson remains the gold-standard test of dynamic equilibrium theory. It has since been repeated for microbial communities, intertidal invertebrates, and pond zooplankton.

On 27 August 1883, the Krakatau volcano between Java and Sumatra erupted catastrophically. The explosion sterilised three small islands (Rakata, Panjang, Sertung), leaving no surviving biota above the level of possibly some bacteria. Indonesian, Dutch, and British botanists began censusing the recolonising flora and fauna almost immediately, and the census has continued to the present.

The Krakatau longitudinal record (Thornton 1996, Whittaker & Fernández-Palacios 2007) shows a saturation trajectory entirely consistent with MW theory:

• By 1908 (25 years post-eruption), Rakata held 115 plant species.
• By 1934: 271 plant species.
• By 1989: 314 plant species, consistent with a MW-like saturation around 320.
• Bird species: ~30 by 1920, ~50 by 1950, stabilised at ~36 (turnover exceeds net change).

The Krakatau birds have shown measurable turnover, with some colonists going extinct on the islands and others arriving. This demonstrates the turnover prediction of MW theory at a spatial scale three orders of magnitude larger than the Simberloff–Wilson mangroves.

Although MW theory is fundamentally about equilibrium, sufficiently isolated oceanic islands can accumulate species faster through in situ speciation than through immigration. This shifts the dynamic from colonisation-driven to speciation-driven and produces the classic island radiations.

Darwin’s Finches (Galápagos)

The 18 recognised species of Darwin’s finches (Thraupidae: Geospizini) are descended from a single South American colonising finch population, ca 1.0–1.5 Ma (Lamichhaney et al. 2015). Beak-shape divergence is driven by diet specialisation (seed size, cactus flowers, insects), with the sharpest adaptive shifts observed in response to the 1977 El Niño drought documented by Grant & Grant (2002, Science).

Hawaiian Drosophilids

The Hawaiian Drosophila and Scaptomyza fauna comprises over 800 described species and an estimated 1,000–1,200 total, all descended from a single ancestor ca 25 Ma. Hawaiian Drosophila represent roughly one quarter of the world’s described drosophilids despite occupying less than 0.002% of its land area. The radiation is progressive with island age: the oldest main island (Kauai, ca 5 Ma) holds a phylogenetic subset that repeatedly colonised younger islands, creating stepwise species accumulation.

Hawaiian honeycreepers (~50 species derived from a single Carpodacus-like rosefinch ancestor, ca 5 Ma) follow the same progressive pattern. The honeycreepers have suffered catastrophic post-contact extinction: perhaps half of all species have been lost since human arrival ca 1,000 years ago, a rate that makes them the textbook example of island avifaunal collapse (Pratt 2005).

E. O. Wilson (1961) proposed that island biotas pass through sequential stages: a colonising species first expands across multiple islands, then evolves into island endemics, becomes ecologically specialised, and finally goes extinct—only to be replaced by a later generation of colonists. This taxon cycle is documented most clearly in Melanesian ants and in West Indian birds (Ricklefs & Bermingham 2002).

Four stages are recognised:

• Stage I: recent colonist, widespread across many islands, similar to mainland source.
• Stage II: differentiated into island subspecies, still present on many islands.
• Stage III: distinct endemic species, restricted to few islands.
• Stage IV: relict endemic on one or two islands, often specialised, prone to extinction.

The cycle is a consequence of two opposing processes: adaptive evolution to local island conditions reduces fitness for cross-ocean dispersal, while the smaller populations of each new island subspecies increase extinction risk. Ricklefs & Cox (1972, 1978) argued this generates a characteristic age structure of island assemblages that matches the qualitative taxon-cycle prediction.

MW assumed islands are geomorphically static. For volcanic oceanic islands this is false: islands grow, erode, and eventually subside. Robert Whittaker, Kostas Triantis and Richard Ladle (2008, Journal of Biogeography 35, 977) proposed the General Dynamic Model (GDM) of oceanic island biogeography to capture this time-dependence.

Under the GDM, island species richness and endemism are hump-shaped functions of geological age: young islands are small and depauperate, middle-aged islands are large and rich, and old islands have eroded to low small atolls with reduced richness. The model writes:

The GDM is empirically confirmed for the Hawaiian archipelago. Kauai (5.1 Ma, old) has fewer endemic plants per unit area than Maui-Nui complex (1.2 Ma, middle-aged); the Big Island (< 0.5 Ma, young) is the richest overall in absolute terms because of its area, but per unit area it ranks below Maui-Nui. The progressive colonisation from old to young islands, combined with age-dependent within-island diversification, produces the observed species richness peak.

Brown & Kodric-Brown (1977) showed that for weakly isolated islands, immigration can be frequent enough to prevent extinction of declining populations. This rescue effect means that extinction rate depends not just on population size but also on immigration rate:

The rescue effect is the bridge between classical island biogeography and Levins’s (1969) metapopulation theory, in which a species persists globally through a balance of local extinctions and local recolonisations among habitat patches. Hanski’s (1994) incidence function model uses the same ingredients on patchworks that are not necessarily surrounded by ocean.

Island biogeography theory supplies the quantitative backbone for reserve design. Two classic applications:

• Habitat-fragmentation species loss: cutting a forest into fragments reduces effective area. Under \(S = cA^z\) with fragmentation \(z \approx 0.3\), reducing area to 10% leaves \(S / S_0 = 0.1^{0.3} \approx 50\%\) of species.
• SLOSS debate (Single Large Or Several Small): under a naive species–area model, one large reserve typically holds more species than several small reserves summing to the same area, but small reserves may capture beta-diversity better. The modern consensus (Fahrig 2017) is that the answer is taxon-specific.

Whittaker (2008)’s GDM has been applied to the projected submergence of Pacific atolls under sea-level rise: young islands at an accumulative phase of the GDM are expected to gain species faster than older subsiding atolls, so conservation triage should concentrate resources on old islands first.

The pattern of \(z\) values across systems confirms the MW theoretical expectation that more isolated, more clearly bounded systems (oceanic islands, mountaintop “sky islands”, habitat fragments embedded in hostile matrix) have steeper species–area curves than weakly bounded or contiguous systems (continental nested quadrats, islands close to the mainland).

Several important extensions and criticisms have sharpened MW theory over the decades:

• Williamson’s neutral-theory critique: Hubbell’s (2001) unified neutral theory predicts species–area curves from dispersal limitation without a dynamic-equilibrium assumption; fit to oceanic islands is often comparable to MW.
• The problem of habitat heterogeneity: larger islands almost always contain more habitat types; the area effect is partly a habitat-heterogeneity effect (Williams 1943). Disentangling these requires controlled comparisons across islands of equal habitat diversity.
• The “small island effect”(Lomolino 2000): below a critical area threshold, species richness is roughly constant rather than continuing to decline with area. Proposed explanations include habitat saturation at minimum-viable-area limits and stochastic occupancy of a minimal species set.
• Neutral phylogenetic tests: Rosindell & Phillimore (2011) embed MW in a neutral speciation framework and show that observed Hawaiian and Canary Island phylogenies can be produced with a small set of parameters, supporting a tight coupling between island biogeography and macroevolution.