Module 3: Continental Drift & Vicariance

In January 1912, Alfred Wegener presented two lectures at the Frankfurt Geological Association, published that year as Die Entstehung der Kontinente(“The Origin of the Continents”, Geologische Rundschau 3, 276–292), and expanded in 1915 into the full monograph Die Entstehung der Kontinente und Ozeane. Wegener’s thesis was simple and radical: the continents are not fixed, but drift horizontally over geological time. He synthesised four classes of evidence:

• Geometric fit of the South American east coast to the African west coast at the 1000 m isobath, not the coastline.
• Stratigraphic continuity of Karoo sediments from South Africa to the Santa Catarina sequence of Brazil, the Permian-Triassic Gondwanan glacial tillites that track contiguous ice sheets from South America through Africa, India, Antarctica, and Australia.
• Palaeontological links: the Permian freshwater reptile Mesosaurus occurs only in the Irati and Whitehill formations of Brazil and South Africa; the seed fern Glossopteris flora is shared among all five Gondwanan continents; and the Triassic terrestrial tetrapods Lystrosaurus and Cynognathus show identical distributions.
• Structural alignment of orogenic belts — the Appalachians continue into the Caledonides of Scotland and Scandinavia.

Wegener lacked a viable mechanism. He proposed that continents ploughed through sima oceanic crust driven by centrifugal (Polflucht) and tidal forces, a physical model that Harold Jeffreys (1924) showed was two orders of magnitude too weak. The hypothesis was rejected by most geophysicists for half a century even while palaeobotanists (Du Toit 1937, Our Wandering Continents) and biogeographers continued to accept it.

The revival of continental drift came from marine geology and palaeomagnetism. Four breakthroughs between 1960 and 1968 converted drift into the unified theory of plate tectonics:

• Hess (1962) and Dietz (1961) proposed seafloor spreading: mantle convection forms new oceanic crust at mid-ocean ridges and consumes it at trenches, carrying continents passively on top.
• Vine and Matthews (1963), Nature 199, 947–949, showed that the symmetric magnetic anomaly stripes flanking the Reykjanes Ridge record the alternating polarity of Earth’s magnetic field as new basalt crystallises and spreads away from the ridge axis. Each stripe pair is a geological tape recorder of spreading.
• Wilson (1965), Nature 207, 343–347, introduced the transform fault as a third class of plate boundary beyond ridges and trenches, completing the kinematic framework and predicting the now-famous “Wilson cycle” of ocean-basin opening and closing.
• Morgan (1968) and Le Pichon (1968) formalised the rigid-plate model on a sphere using Euler poles of rotation, completing the quantitative theory.

The arithmetic of spreading is straightforward. The half-spreading rate \(v\) relates magnetic anomaly age \(\tau\) to distance from the ridge \(x\):

Reconstructed plate motions on a sphere are rotations. Finite motion of plate B relative to plate A is described by a rotation vector \(\boldsymbol{\Omega}\) about an Euler pole \((\theta, \phi)\):

Wegener’s reconstructed supercontinent Pangaea (“all earth”) assembled by the late Permian (~270 Mya) from the collision of Laurussia and Gondwana. It began to fragment ~200 Mya in the early Jurassic, splitting first into a northern Laurasia (future North America, Europe, northern Asia) and a southern Gondwana (future South America, Africa, Madagascar, India, Antarctica, Australia-Zealandia).

The chronology is set by the oldest seafloor flanking each new ocean basin and is broadly:

• ~200 Mya: Central Atlantic begins opening between North America and Africa.
• ~180 Mya: Gondwana begins internal fragmentation. Africa and South America remain contiguous for tens of millions of years after the first rifts appear.
• ~160 Mya: India–Madagascar separates from Antarctica/Australia.
• ~130 Mya: South Atlantic begins opening between Africa and South America.
• ~90 Mya: India separates from Madagascar and begins its rapid northward drift.
• ~70 Mya: India’s Deccan flood basalts erupt as it passes over the Reunion hot spot; northward drift continues at 15–20 cm/yr, one of the fastest plate motions in Earth history.
• ~55 Mya: India collides with Eurasia; the Tethys Sea closes; uplift of the Himalayas begins. The collision vicariantly separates Afrotropical and Oriental biotas along an orogen that is still rising today.
• ~35 Mya: Australia separates from Antarctica and begins northward drift, opening the Southern Ocean and initiating the Antarctic Circumpolar Current.
• ~3 Mya: Isthmus of Panama closes, triggering the Great American Biotic Interchange.

Each of these tectonic events produced a corresponding biogeographic signature. Ratite birds (ostriches in Africa, rheas in South America, emus and cassowaries in Australia, kiwis in New Zealand, extinct moas in New Zealand) have a molecular divergence structure broadly consistent with Gondwanan vicariance (Cracraft 2001), though recent phylogenomic work (Mitchell et al. 2014 Science) demonstrates that elephant birds and kiwis are sister taxa, implying additional trans-oceanic dispersal beyond pure vicariance.

The Indian plate presents perhaps the most spectacular single episode of plate motion. After separating from Madagascar ~90 Mya, it accelerated northward at an average 15–20 cm/yr — a Pacific-style spreading rate applied to a cratonic fragment — closing the Tethys Sea that had lain between Gondwana and Laurasia. The ~55 Mya collision with Eurasia continues today, producing both the Himalayan orogen and the Tibetan Plateau, with present convergence of ~4 cm/yr (GPS geodesy; Molnar & Tapponnier 1975).

Biogeographically, India arrived with a “Gondwanan cargo” of lineages shared with Madagascar and Africa, but spent tens of millions of years as an isolated island continent. Purported Indian-endemic survivors include:

• Sooglossidae frogs (sister to Nasikabatrachidae; Biju & Bossuyt 2003 Nature 425, 711–714 described the purple frog Nasikabatrachus sahyadrensis of the Western Ghats, with its closest relatives on the Seychelles).
• Caecilian amphibians (Ichthyophiidae) with deep Gondwanan splits.
• Ratite affinities (elephant birds of Madagascar and kiwis of New Zealand were once thought to share Gondwanan ancestry with Indian lineages).

The collision also acted as a filter: Oriental faunas expanded westward into India after contact with Eurasia, while African and Malagasy affinities persist principally in the ancient uplands of the Western Ghats and Sri Lanka. The signal-to-noise problem in separating Gondwanan vicariance from Eocene dispersal motivates the molecular clock and likelihood-ancestral-range methods of Sections 7–8.

Through most of the Cenozoic, North and South America were separate continents with fundamentally different mammalian faunas. North America held a placental fauna of carnivorans, perissodactyls, artiodactyls, proboscideans, and lagomorphs. South America had an autochthonous fauna of marsupials (didelphimorphs, paucituberculates, sparassodont predators), xenarthrans (armadillos, glyptodonts, ground sloths), and the endemic ungulate orders Litopterna and Notoungulata. Isolation ended with the closure of the Central American Seaway and the final emergence of the Isthmus of Panama ~3.0 Mya (Coates et al. 2004; more recent work places the closure slightly earlier, 4.2–3.5 Mya, Bacon et al. 2015 PNAS).

Stehli and Webb (1985, The Great American Biotic Interchange, Plenum) synthesised fossil evidence across the South American Land Mammal Age (SALMA) stages Huayquerian through Lujanian and quantified the exchange. Their headline results:

• Approximately 50% of modern South American mammalian genera descend from North American immigrants (“northerners” Webb 1991).
• Only ~10% of modern North American mammalian genera descend from South American immigrants (“southerners”).
• Surviving southern immigrants in North America: opossums (Didelphis), nine-banded armadillos (Dasypus), porcupines (Erethizon). Most other southern immigrants (giant ground sloths, glyptodonts, phorusrhacid terror birds, toxodont ungulates) went extinct in North America after a brief flourish, many surviving only until the Pleistocene megafaunal collapse (~11 kya).
• Surviving northern immigrants in South America: jaguars, pumas, foxes, bears, deer, peccaries, tapirs, camelids (llamas, vicuñas), skunks, rabbits, mice—essentially the dominant modern Neotropical fauna.

The asymmetry is striking and demands explanation. Webb (1976, 1991) argued that the extinction of native South American ungulates (Litopterna, Notoungulata) was competitive, driven by the arrival of ecologically similar but more derived placental ungulates. Marshall et al. (1982) gave a more nuanced view: South American endemic predators (Sparassodonta) had already declined before the isthmus closed, leaving vacant carnivore niches for immigrant felids and canids. Modern analyses (Woodburne 2010 Journal of Mammalian Evolution) show the interchange had multiple pulses, with the major asymmetric northward extinction of South American immigrants occurring in the Late Pleistocene rather than at first contact.

Beringia, the shallow continental shelf between Siberia and Alaska, has been emergent for extended periods whenever eustatic sea level dropped more than ~50 m below present, which occurred repeatedly during Pleistocene glaciations. Hopkins (1967, The Bering Land Bridge, Stanford) synthesised the geological evidence. Key emergent phases:

• Last Glacial Maximum ~26–19 kya: a land bridge several hundred kilometres wide.
• MIS 5b/5d glacial stages (~90–100 kya).
• Multiple earlier Pleistocene glacials back to ~2.6 Mya, matching the intensification of Northern Hemisphere ice sheets.

Beringia permitted dispersal of Eurasian mammals into North America (Mammuthus, Bison, Ovis, Equus, Homo sapiens ~18–15 kya based on genetic and archaeological data) and of North American lineages into Eurasia (camelids, which originated in North America before going extinct there in the Late Pleistocene; dogs; some proboscidean taxa). The bridge acted as a selective filter: cold-tolerant species crossed during glacials, and warm-adapted taxa were excluded (Guthrie 1982, Frozen Fauna of the Mammoth Steppe).

The deeper mammalian dichotomy between Australasian marsupial-dominated faunas and Eurasian/American placental-dominated faunas is a textbook example of vicariance. Metatheria and Eutheria diverged from a common tribosphenic mammalian ancestor ~160 Mya (molecular clock calibrations, dos Reis et al. 2012 PNAS). The Australian radiation of marsupials proceeded in isolation after Australia broke from Antarctica (~35 Mya), producing the modern kangaroos, wombats, koalas, Tasmanian devils, and extinct thylacine, all descended from a single pulse of marsupial colonisation ~55 Mya via Antarctica (Beck 2008 Journal of Mammalian Evolution). The strict marsupial–placental competitive-exclusion model is no longer supported: early Tertiary Australia had a significant placental fauna (Godthelp et al. 1992 reported Tingamarra on the basis of the famous Tingamarra porterorum, though this has been contested) and today still supports rodents and bats that arrived via Asia.

Léon Croizat, in Panbiogeography (self-published 1958, 3 volumes), introduced the concept of the generalised track — a line on the Earth’s surface joining the ranges of taxa with similar distributions. Where multiple generalised tracks coincide, they define a node. Croizat rejected chance dispersal as the dominant explanation for biogeographic patterns and argued instead that the Earth’s geographic evolution is the primary explanation.

Croizat’s method is graphical. For a given taxon, plot disjunct ranges on a map, then connect them by a minimum-length “individual track” following plausible current or past dispersal routes. Superpose tracks of independent clades with broadly similar distributions. The resulting generalised track identifies geographic structure common to many clades — typically matching old continental connections or oceanic current corridors.

Croizat identified five major generalised tracks spanning the globe:

• Trans-Atlantic (South America–Africa).
• Atlantic-Caribbean.
• Austral (South America–Australia–New Zealand via Antarctica).
• Indo-Pacific (Madagascar–India–SE Asia–Pacific).
• Boreal (Eurasia–North America).

Croizat’s work was largely dismissed in the English-speaking world during the 1960s because he was explicitly anti-Darwinian in rhetoric, opposed to “chance dispersal” as an explanation of anything, and personally difficult. Yet his generalised tracks were revived within the cladistic vicariance biogeography of the 1970s, as Nelson, Platnick, Rosen and colleagues recognised that the spatial congruence he demanded was in effect the assumption of vicariance (Craw, Grehan & Heads 1999, Panbiogeography, Oxford U. Press, reassessed his framework).

Cladistic biogeography emerged in the 1970s from the marriage of plate tectonics and Hennig’s cladistic methodology. Its canonical papers are:

• Rosen (1978), Systematic Zoology 27, 159–188, “Vicariant patterns and historical explanation in biogeography.” Rosen used Caribbean fish cladograms to demonstrate that vicariance, not dispersal, was the null expectation.
• Nelson & Platnick (1981), Systematics and Biogeography, Columbia U. Press. The classic exposition of the method: for each clade, convert a cladogram of taxa into an “area cladogram” by substituting each taxon with its geographic range. Multiple area cladograms are then compared for congruence.
• Humphries & Parenti (1986)provide a popular textbook treatment.

The vicariance cladistic research programme has three steps:

1. Construct taxon cladograms using morphological or molecular characters.
2. Replace each terminal taxon with its area of endemism, obtaining an area cladogram.
3. Compare area cladograms across independent clades using congruence tests (Brooks Parsimony Analysis, Component Analysis, Paralogy-Free Subtrees, Three-Item Analysis).

If many independent clades share the same area cladogram, a common geological history (vicariant fragmentation) is the most parsimonious explanation. If the area cladogram matches the known tectonic fragmentation sequence of the regions involved, the match is taken as strong evidence of vicariance.

The descriptive cladistic-vicariance programme was quantified in the 1990s– 2000s with explicit likelihood models for ancestral range reconstruction. Two approaches dominate modern analyses:

• DIVA (Ronquist 1997, Systematic Biology 46, 195–203): a parsimony method minimising a cost matrix for dispersal (gain of areas), extinction (loss of areas) and vicariant speciation (range splitting).
• DEC (Ree & Smith 2008, Systematic Biology 57, 4–14): a continuous-time Markov likelihood model on the set of non-empty subsets of areas, with dispersal rate \(d\), extinction rate \(e\), and explicit cladogenetic range-inheritance rules.
• DEC+J (Matzke 2014) adds a cladogenetic “jump” event mimicking founder-event speciation, improving fit for young island radiations but producing statistical controversy (Ree & Sanmartín 2018).
• BioGeoBEARS (Matzke 2013) is the dominant R implementation that lets the user compare DEC, DIVA-LIKE, BAYAREA, and their +J variants on the same phylogeny.

The DEC state space is the \(2^{k} - 1\) non-empty subsets of \(k\) areas. The rate matrix \(Q\) is \((2^{k}-1)\times (2^{k}-1)\) with:

Transition probabilities across a branch of length \(t\) are \(P(t) = e^{Q t}\). At each cladogenetic event the DEC model allows one of three inheritance rules: subset sympatry (one daughter inherits a subset of the ancestral range, the other inherits the whole), vicariance (the ancestral range splits into two complementary subsets), or single-area sympatry (both daughters inherit the ancestral single-area range). The full likelihood of the phylogeny is computed by Felsenstein’s pruning algorithm with cladogenetic weighting at each internal node.

Empirical applications of DEC/DIVA have clarified long-standing biogeographic puzzles:

• The genus Nothofagus (southern beeches) has an austral distribution (S America, Australia, New Zealand, New Guinea). DEC reconstruction on the combined phylogeny (Cook & Crisp 2005) supports a primarily vicariant Gondwanan break-up, but with late Cenozoic dispersal across the Tasman Sea.
• Chameleons (Chamaeleonidae) were once considered a Gondwanan relict, but a DIVA-based analysis (Tolley et al. 2013 Proc. Roy. Soc. B) supports a Malagasy origin with multiple transoceanic dispersals to Africa, India and the Seychelles.
• Ratites: DEC reconstruction (Mitchell et al. 2014) on genomic-scale data rejects pure Gondwanan vicariance; multiple independent losses of flight and transoceanic dispersals of flying ancestors fit best.

Ancestral-range inference is only as reliable as its time calibration. Bayesian and maximum-likelihood phylogenetic analyses use fossil occurrences both as tip calibrations (fossils included as terminal taxa with dated ages) and as node calibrations (minimum or soft-maximum priors on internal nodes). Heath, Huelsenbeck & Stadler (2014 PNAS) formalised this in the fossilised birth–death (FBD) model, which combines speciation, extinction and fossil sampling into a single generative process for time-calibrated trees.

Canonical vicariance calibrations:

• South Atlantic opening (~130 Mya)used to calibrate ostrich/rhea, cichlid (Africa/South America), and many plant clades (e.g., Malpighiaceae, Davis et al. 2002).
• India–Madagascar (~90 Mya)used for Nasikabatrachidae/Sooglossidae and chameleons (with competing hypotheses of Paleogene dispersal).
• Australia–Antarctica (~35 Mya)used for Nothofagus, marsupials, southern beeches.
• Isthmus of Panama (~3 Mya)used for marine geminate species pairs split on the two sides of the isthmus (Lessios 2008 Annual Review of Ecology, Evolution and Systematics).

The circularity risk is real: if one calibrates a clade with an assumed vicariant event at a geological age, one cannot then use the inferred tree to test whether the event was vicariant. Modern practice sidesteps this by calibrating only with fossils or with vicariant events involving different clades, and then testing the target hypothesis with a molecular clock derived independently.

The Croizat–Nelson–Rosen cladistic vicariance programme was at one stage framed as diametrically opposed to dispersalism. The data now show that the opposition was too polar. Molecular-clock and ancestral-range reconstructions have revealed that many classical “Gondwanan vicariance” patterns are in fact recent transoceanic dispersals (e.g. the New Zealand biota: Landis et al. 2008 New Zealand Journal of Geology & Geophysics; the classic Oligocene “drowning” of NZ would imply that most extant clades are post-drowning colonists).

De Queiroz (2005 TREE), “The resurrection of oceanic dispersal in historical biogeography,” reviewed the evidence and argued that — far from being the rare, ad hoc explanation of Croizat’s caricature — long-distance transoceanic dispersal has been the dominant process for many clades. Biogeographers now routinely use DIVA/DEC to quantify the contribution of each process to each node. The modern consensus: both vicariance and dispersal are fundamental and their relative contributions are clade-specific, time-specific, and ecosystem-specific.

Each row represents a clade where biogeography has had to be re-evaluated as molecular clocks and likelihood ancestral-area analyses matured. Notice that some classical vicariance cases (cichlids, amphisbaenians) remain well supported, while others (ratites, Malpighiaceae) have shifted to dispersal after phylogenomic reassessment.