Molecular studies reveal highly ordered geographic patterns in plant and animal distributions. The tropics illustrate these patterns of community immobilism leading to allopatric differentiation, as well as other patterns of mobilism, range expansion, and overlap of taxa. Integrating Earth history and biogeography, Molecular Panbiogeography of the Tropics is an alternative view of distributional history in which groups are older than suggested by fossils and fossil-calibrated molecular clocks. The author discusses possible causes for the endemism of high-level taxa in tropical America and Madagascar, and overlapping clades in South America, Africa, and Asia. The book concludes with a critique of adaptation by selection, founded on biogeography and recent work in genetics.
Molecular Panbiogeography of the Tropics
Evolution in Space
Many different ways of analyzing spatial variation in biological diversity - the biogeographic patterns - have been employed by different authors and some of the assumptions in these methods are discussed here. The chronological aspect of evolution is discussed in the next chapter.
Every kind of plant or animal has its own particular distribution and ecology, and this was already well-understood in ancient times. Yet portraying a distribution is not straightforward. New collections are always being made and ideas on the delimitation of taxonomic groups change. Outline maps are generalized simplifications only but are useful for comparative purposes. Although dot maps showing sample localities give more detail, they are always incomplete, the accuracy of the dot locations can often be questioned, and the entities that the dots represent - the populations or individuals - are constantly changing position due to birth, death and movement. A distribution is dynamic and so a distribution map represents an approximation, a probability cloud, not an actual distribution. Nevertheless, the fact that so many distribution maps have been made reflects the high value that biologists and many others have put on them.
Knowledge of organic distribution is useful for simple survival and economic development, as the plants, animals and microbes of a particular place are often among its most distinctive and valuable features, and also its most poisonous and dangerous. Many groups have particular, idiosyncratic distributions; the details of these are known by local people and broader scale distributions are documented in the literature.
Organisms are distributed spatially in three dimensions and while the questions treated in this book mainly involve differentiation in the horizontal plane, in latitude and longitude, the altitudinal component of a clade's distribution must also be considered. While the altitudinal elevation of a group is sometimes assumed to reflect its ecological preference, in some cases there is an ecological lag and historical effects are important. For example, an area may be uplifted along with its biota
and some of the biota will likely survive to become montane taxa. Depending on where it is located, a population may be uplifted or not during an episode of mountain building and so biogeography can determine ecology, rather than the reverse.
The method of multiple working hypotheses
The focus in this book is on distribution patterns and their interpretation in terms of evolutionary processes. Most biogeographic interpretation over the last two thousand years has been based on a single paradigm, the center of origin/dispersal model of historical development. But having only a single working hypothesis to explain a set of phenomena can lead to problems and over time it becomes easy to accept that the single hypothesis is the truth.
Although much modern work in biogeography stresses supposed consensus, in science and philosophy, as in art and literature, a diversity of views and approaches can be a good thing. Puritans of all sorts (whether Oliver Cromwell or Louis XIV) cannot stand anyone having a different view from themselves. The inflexible schemes of these great simplifiers, levellers, and systematizers can hold up progress for decades. In contrast, geologists (Chamberlin, 1890 reprinted 1965) and now molecular biologists (Hickerson et al., 2010) cite the method of 'multiple working hypotheses'. This proposes that it is never desirable to have just one working hypothesis to explain a given phenomenon. Accepting a single interpretation as definitive can be counter-productive and lead to the decline of a subject.
It is unfortunate that the interpretations of the data currently given in most molecular studies are all based on the same fundamental concepts. This 'plug-and-play' biogeography involves the following steps: assume that the study group has a center of origin and use a suitable program to find one; accept that fossil-calibrated clock dates give the maximum age of the group; describe possible dispersal routes from the center of origin. The axioms that are assumed here can be questioned though and a Socratic approach may be useful. Canetti (1962) wrote that: 'A scholar's strength consists in concentrating all doubt onto his special subject' and a healthy scepticism is one of the pillars of science, both in history and in everyday practice. When identifying unfamiliar plants and animals on the reef or in the rainforest it is tempting, but often dangerous, to jump to conclusions before considering a wide range of possibilities and the same is true for biogeographic interpretation.
The case studies of different groups discussed below adopt certain assumptions and concepts, and some of these are outlined next.
Phylogenies, classifications, and nested sets: hierarchical summaries of character distributions
A related group of organisms forms a branch or clade in a phylogeny or evolutionary family tree. A clade may or may not be be formally named, as a taxon (plural: taxa). The closest relative of a group is termed its sister group. In most published phylogenies, the clades in a group are shown in a strictly hierarchical system of nested clades. Phylogeny is the general process of the evolution or genesis of clades, and 'a phylogeny' is also a term for a branching diagram or a tree, a symbolic arrangement of hierarchical, nested sets of clades. Nested sets of groups are depicted in traditional dichotomous keys, nomenclatural systems, cladograms, phylogenies, trees and so on; all represent the same thing, an Aristotelian classification. This is only one way of representing variation and others include ordination, a method which show trends rather than groups and which is often used in ecology.
Ideas on the evolutionary process still reflect the Aristotelian, classificatory approach in many ways. This sometimes leads to a misplaced emphasis on the clades rather than on the morphological and molecular characters that underlie them. The usual units of analysis in this book are indeed clades, as presented in molecular phylogenies, but these should not be taken too literally. Biogeographic areas may be problematic and biological groups - 'monophyletic clades' - may also be complex. Most groups have characters/genes that show phylogenetic and geographic variation within the group that is 'incongruent' with those of other characters/genes and this will be discussed below. Ultimately, in a hypocladistic approach, the focus is on the evolution of the underlying characters rather than on particular combinations of characters, including the clades.
Is the species special? The Darwinian species concept
Evolution results in a continuum of differentiation. Entities may differ by a smaller or greater amount and depending on this level may be recognized as barely distinct populations, subspecies, species, genera, families etc. The focus here is on the process of differentiation rather than any of its particular products, and the species is seen here simply as a point on a trajectory between subspecies and genus, it has no 'special' value. This is the species concept used by Darwin (1859) and Croizat (1964) (see also Ereshefsky, 2010). Neither the species nor any of the other taxonomic categories have any absolute value and a species or genus in one group cannot necessarily be compared with species- or genus-level differentiation in another group.
In contrast with the Darwinian species, the species in the neo-darwinian synthesis are very special indeed as they have a reality that subspecies, genus and the other categories do not. In this return to mediaeval nominalism, subspecies, genera and other 'universals' are seen as really just names and not things. Only species are real things ('individuals'). This distinction is not accepted in the Darwinian approach used here, in which clades (monophyletic groups) of any rank and their characters replace the species as the basic units of analysis. The most detailed information available on geographic differentation happens to concern 'monophyletic' clades in morphological and molecular phylogenies, although geographic variation in any single characters would be just as useful.
Degree of difference
The interpretation of degree of difference (branch length) between groups is discussed in the next chapter. The particular degree of difference of a group and its taxonomic rank are not necessarily related to time and instead they may reflect aspects of prior genome architecture in the ancestor. The focus here is on spatial differentiation in any clades, whatever rank or branch length is involved.
Spatial analysis of groups: the center of origin/dispersal model and the vicariance model
These two models are often contrasted in biogeographic studies and their applications have caused a great deal of debate.
Center of origin/dispersal model
How did the distribution of a plant or animal develop? Consider a hypothetical distribution (Fig. 1-1). One theory is that such a pattern originated by a plant or animal evolving at a point somewhere within its current area and spreading out from there to the limits of its present range.
Researchers attempt to locate the 'center of origin' or 'ancestral area' by studying the distribution and phylogeny in the group itself and by using different criteria. The center of origin has been thought to occur in the area that shows one or more of the following:
the highest diversity of forms within the group ('1' in Fig. 1-1),
the oldest fossil ('2' in Fig. 1-1),
the most 'advanced' form (cf. Darwin, 1859, Briggs, 2003) ('3' in Fig. 1-1)
the most 'primitive' form (cf. Mayr, 1942, Hennig, 1966) ('4' in Fig. 1-1)
the 'basal' clade or grade of the group (most modern studies) ('5' in Fig. 1-1)
Several computer programs designed to find the center of origin of a group are now available, for example, DIVA (Ronquist, 1997) and Lagrange (Ree and Smith, 2008).
Many other criteria for locating a group's center of origin have been proposed in addition to those listed above and the confusion that this implies was pointed out by Cain (1943). This paper led to the modern critique of the center of origin that has been developed in panbiogeography (Craw et al., 1999) and paleontology (López-Martínez, 2003, 2009, Cecca, 2008).
Cecca (2008) characterized two models of evolutionary biogeography: center of origin/dispersal theory as developed by Darwin (1859) and Wallace (1876), and vicariance, as developed by Sclater (1864) and Croizat (1964). In discussions of these models, the phrase 'center of origin' does not simply mean a center where a group has originated (all taxa originate somewhere), but refers to a specific concept used in the dispersal model. In this model, a group's ancestor evolves as a monomorphic, homogeneous entity in a restricted area (the 'center of origin') following a chance dispersal event and the group attains its distribution by physical movement out of this center.
Finding the center of origin of a group is a fundamental aim of many studies and groups may be analysed in ever increasing detail in order to locate the center. The center of origin of a group is often located by examining the group itself. An alternative approach considers a group not on its own, but in relationship to its closest relative or sister group (Fig. 1-2). It may be difficult to understand the origin of a group by studying the group itself, especially if groups come into existence together with at least one other, by vicariance.
In many cases two sister groups have neatly allopatric (vicariant) distributions, with one group replacing or representing the other in a second area, often nearby or even adjacent to the first. Each of the two groups may have arisen not by spreading out from a point, but by geographic (allopatric) differentiation in its respective area from a widespread ancestor. In this process ('vicariance'), there is no physical movement, only differentiation, with populations in area A evolving into one form, populations in area B into another. This vicariance theory is a basic component of panbiogeography. In a vicariance event, the distribution of a group comes into existence with the group itself. A group's 'center of origin' may be more or less the same as its distribution, especially if it is part of an allopatric series. (A distribution range may expand or contract after its initial formation, leading to secondary overlap; this is discussed below).
Despite the development of the vicariance model, the center of origin/dispersal model of the evolutionary process is still widely assumed by paleontologists (Eldredge et al., 2005), ecologists (Levin, 2000, Gaston, 2003: 81), and some biogeographers. For example, Cox and Moore (2010: 204) wrote: 'Let us imagine that a species has recently evolved. It is likely, to begin with, to expand its area of distribution or range until it meets barriers of one kind or another...'. But this does not necessarily happen in a vicariance event, as the new species already abut their relatives. If a globally widespread form evolves by breaking down into, say, two allopatric species, one in the northern hemisphere and one in the southern hemisphere, neither one may expand its range. In a vicariance model, a new clade is just as likely to contract its range as to expand it.
In the vicariance approach, the focus is on tracing the originary breaks between groups, not on locating a point center of origin within a group. In a dispersal analysis, the first question is: where is the center of origin? In a vicariance analysis, the first question is: where is the sister group? The focus is not on the group itself or on details of its internal geographic/phylogenetic structure, but on its geographic and ecological relationship with its sister group and other relatives.
In this model, a group originates by the breakdown of a widespread ancestor, not by evolving at a point and spreading out from there. Analysis of any group can start either with a point center of origin or, alternatively, with a widespread ancestor. In the latter model, a group evolves on a broad front over the region it occupies, by 'fracturing' with its sisters (vicariance) at phylogenetic and biogeographic breaks or nodes. A node is not a center of origin or an ancestor, it is a break where the distributions of two or more groups meet.
In one example, Chakrabarty (2004) supported a vicariance history for the freshwater fish family Cichlidae. He compared the process with a mirror being struck several times with a hammer. There is no movement of the individual shards, which are all neatly vicariant. The sequence of the hammer blows (i.e. phases of differentiation) is seen in the phylogeny or cladogram. The process also resembles the development of vascular tissue in a young organ out of ground tissue. There is no physical movement and the veins do not grow by pushing their way through tissue, but by differentiating in situ, in accordance with the genetic program.
Modes of speciation: dispersal and vicariance
The mode of differentiation of groups in general (phylogenesis) and of species in particular (speciation) is problematic and the interpretation of even the simplest cases is debated. As with clades in general, two allopatric species can be explained as the result of either vicariance in a widespread ancestor (dichopatric speciation) or founder dispersal from a center of origin (peripatric speciation).
Origin of the ancestor
Two descendant groups may have originated by vicariance, but what about the ancestor of the two? Surely the ancestor must have dispersed to achieve its wide range? In fact this is not necessary, as the ancestor of the two groups A and B, in areas A and B, may itself have originated as an allopatric member of a broader complex, the ancestor of A+B+C, that also occurred in areas C. This in turn may have differentiated from the ancestor of A+B+C+D, as indicated by the four clades in Fig. 1-3. Here there is no center of origin. In the center of origin/dispersal theory, each of the four allopatric groups in Fig. 1-3 would have a separate center of origin and their distributions are not directly related to their origins - the groups formed first and the distributions were established later. The boundaries of the four groups are secondary and the distributions only met after the four groups spread out from their respective centers of origin. Instead, in panbiogeography the mutual boundaries are interpreted as phylogenetic and geographic breaks or nodes. These recur at the same localities in many different groups with different ecology and so a chance explanation is unlikely.
Overlap in distribution
In many cases the allopatry between close relatives is not perfect and the groups show marginal overlap or interdigitation. This can represent local, secondary overlap by range expansion following the original, allopatric differentiation.
Sometimes a group shows extensive secondary overlap with its sister and this occurs mainly at higher taxonomic levels such as families and orders. A worldwide group and its worldwide sister may have each occupied half the Earth before the overlap developed. A vicariance analysis of the groups' history does not involve finding each group's 'center of origin', but tracing possible original breaks between the two groups. For example, one of the worldwide sister groups may be fundamentally northern, the other southern.
A few species, more genera, and many higher level taxa, such as birds and flowering plants, are world-wide and overlap with each other everywhere. This pattern probably reflects the older age of the higher categories. The overlap of groups shows that vicariance cannot be the only biogeographic process. If it was, there would be pure allopatry - each point on Earth would only have one, locally or regionally endemic, life form. As it is, most places have biotas that include many kinds of plants and animals, indicating the overlap of clades. The overlap may be due to early phases of large-scale range expansion by whole communities.
If the affinities of any group are traced far enough these will be found to make up a world-wide complex. Beyond this stage, if not before, there must be overlap with relatives. The widespread overlap among, for example, families and orders of birds, indicates phases of range expansion. In primates, for example, the main branches - Old World monkeys, New World monkeys, lemurs, tarsiers - are notable for their high degree of allopatry. On the other hand, primates as a whole, (together with their close relatives) show wide overlap with their sister group, the rodents and lagomorphs. At some stage (probably between the origin of primates and the origin of the main primate sub-groups) there was a phase of overlap between the proto-primate complex and the proto-rodent complex.
Phases of vicariance affecting whole communities are often attributed to geological changes in the past, such as the opening of the Atlantic. In the same way, phases of population mobilism, with range expansion and colonization, probably occurred during and following the great geological revolutions. For example, the last, great phase of marine transgressions, in the Mesozoic, produced dramatically extended coastlines and associated habitats. Marine transgressions occurred on all the continents and, at the same time, rifting and continental breakup also produced new seaways. During this phase of mobilism, groups with suitable coastal, marginal ecology colonized vast areas of new habitat and became widespread globally. This was followed by a phase of immobilism through the Cenozoic during which local differentiation predominated.
Geology and vicariance
It is often suggested that the main factor distinguishing evolution by vicariance and by dispersal is the time of appearance of the 'barrier' - before evolution in dispersal and during evolution in vicariance. A more general point is that in a vicariance model the Earth and its life evolve together, whereas in the dispersal model they do not. In dispersal theory, every taxon has its own unique history caused by one-off, chance events and there are no community-wide biogeographic patterns with single causes. (Molecular clock studies based on dispersal theory also support this notion; this will be discussed in Chapter 2). Yet many biologists would be reluctant to abandon the idea that geology can cause community-wide vicariance and generate both large- and small-scale community patterns. The idea that Earth and life evolve together is seen in the geographic concordance of many aspects of biogeography. This agreement even occurs among groups with completely different ecology, such as intertidal marine groups and montane groups. To cite just one example, the Hawaiian Islands and the Marquesas Islands form a center of endemism that is unexpected, given the direction of the currents, and yet it is defined by reef fishes, insects and montane plants (see Chapter 7). Geographic congruence among different groups and also a general congruence between biogeography and geology have been known for a long time. For example, 'The primary geographical divisions in the global mammal fauna clearly coincide with geology and plate boundaries...' (Kreft and Jetz, 2010: 19). For example, in mammals major phylogenetic/geographic breaks (nodes) occur between South America and Africa, and between Madagascar and Africa.
The four processes proposed in biogeography and the two that are accepted here
Four key processes have been proposed in biogeography. As discussed above, differentiation (e.g. speciation) can be due to vicariance of a widespread ancestor or to founder dispersal from a center of origin. In addition, two overlapping sister clades can be explained as the result of range expansion (by normal ecological dispersal, simple physical movement) or by sympatric differentiation.
Of the four processes just cited, vicariance and normal ecological dispersal are accepted as important by all authors. They are the two processes that are accepted in this book as explaining distributions. Normal ecological dispersal can involve movement within the distribution area or outside it, and this may lead to range expansion. Range expansion explains overlap, it does not explain allopatry.
The third process, sympatric differentiation, was controversial although it is now accepted in some cases (Schluter, 2001, Friesen et al., 2007, Bolnick and Fitzpatrick, 2007). If it does occur it is probably quite rare and many cases of supposed sympatry between sister groups prove, on closer examination, to involve only partial geographic overlap and significant allopatry. Other apparent cases of sympatry may involve allopatry at a small scale. As noted above, low level clades are often allopatric with their sisters whereas higher level clades show more overlap, so overlap can generally be regarded as a secondary process that has developed over time from original allopatry, rather than by sympatric evolution.
The fourth process, differentiation by founder dispersal, is controversial and may not exist.
'Dispersal': one word, several concepts
Several quite different processes have all been termed 'dispersal' and they can be contrasted as follows.
1. Normal ecological dispersal
This is the normal physical movement seen in plants and animals. It includes daily and annual migrations along with the dispersal of juveniles. The movement is made possible by the well-known mechanisms observed in different groups. Normal ecological dispersal occurs every day and does not lead to differentiation (speciation etc.). It may take place over long distances, for example, in sea-birds, or over much shorter distances, depending on the organism. Following their origin by reproduction, all individual organisms have dispersed to where they are by this process. Normal ecological dispersal is seen in the weeds that soon colonize a disturbed area, whether this is a newly dug garden, an area of burnt vegetation, the area in front of a retreating glacier, a landslide, or a volcanic island such as Krakatau in Indonesia that has been devastated by a recent, explosive eruption.
Despite appearances, this process of simple movement does not necessarily explain the distribution area occupied by a taxon - in particular any allopatry with related taxa - as it does not account for evolutionary differentiation and this can, by itself, produce a distribution. Thornton (2007) titled his book 'Island colonization: The origin and development of island communities', yet studies on the colonization of Krakatau, for example, only concern the ecological origin and development of communities, not their evolutionary origin. The community on Krakatau is a subset of the weedy community that already existed on the islands in the region and its evolutionary origin dates to long before the last eruption on Krakatau.
2. Range expansion
Following a 'normal dispersal' event, an organism's new position may lie within the former range of the taxon or it may lie outside it and represent a range expansion. Range expansion is seen in historical times in the anthropogenic spread of weeds and at other times in geological and evolutionary history. Range expansion, when it does occur, may be very rapid and a more or less local plant or animal may become world-wide in hundreds rather than millions of years. This takes place by normal ecological dispersal using the normal means of dispersal in the group, not the rarely-used or unknown means sometimes cited to explain the more spectacular events of founder dispersal.
A global ancestor may have achieved its range during Mesozoic range expansion. This mobilism eventually stabilized and was replaced with a phase of immobilism through the Cenozoic. This was a period of in situ evolution that produced local differentiation, mainly at species and subspecies level. Phases of mobilism may alternate with phases of immobilism in which allopatric evolution (vicariance) takes place. Earlier phases of population mobilism would have occurred as new landscapes emerged from the devastation of the Permo-Carboniferous ice ages, centered in the southern hemisphere and much more severe and long-lasting than the Pleistocene ice ages. These cycles of biogeographic mobilism and immobilism may take tens of millions of years to complete, as with the geological cycles of mountain uplift, erosion, deposition and further uplift.
Naturally it is far more difficult to analyse the biogeography of a period prior to the one in which the 'modern', extant patterns developed and many aspects of the pre-modern patterns will never be known. The modern centers of endemism in their turn will not last forever. A new geological or major climatic catastrophe will eventually lead to massive extinction and renewed mobilism, with weedy taxa taking over before they settle and establish new regional blocks of endemic taxa.
3. 'Long distance dispersal'/'speciation by founder dispersal'
The defining feature of this process is not so much the long distance but the fact that it involves a unique, extraordinary, dispersal event by a founder across a barrier. This leads to isolation and speciation (or at least some differentiation). As Clark et al. (2008) emphasized, there is an important distinction between dispersal as normal individual movement and range expansion on one hand, processes that are seen every day, and long-distance dispersal involving founder speciation on the other. The latter (termed 'dispersal-mediated allopatry' in Clark et al., 2008) is a theoretical construction. Normal ecological movement and range expansion, along with other kinds of 'dispersal' such as daily and annual migrations, are accepted here; long distance dispersal/founder speciation is not.
As noted, every individual plant and animal moves as part of its normal means of survival, at least during one stage of its life-cycle. With the exception of some colonial taxa, all individual organisms have reached their present position by dispersing there. This normal ecological movement should not be confused with 'long distance' or 'founder' dispersal, which leads to new lineages. Dispersalists argue that 'When lineages arrive in new habitats they will usually diverge and sometimes speciate' (Renner, 2005). But any patch of newly cleared garden will soon be colonized by 'weedy' flora and fauna, later by less weedy taxa, and none of these will speciate there. Again, founder dispersal is quite distinct from normal dispersal. Authors supporting the center of origin/founder dispersal view for one or other group have often concentrated on proving that ecological dispersal, or ordinary movement, does occur, but this may not be relevant to the issue of founder dispersal.
For differentiation or speciation to occur, a fundamental change in the population ecology from a state of mobilism to one of relative immobilism has to occur. Dispersal on its own might explain how primates came to be in America, if the American primates were the same as those of Africa or Asia. But physical movement on its own cannot explain why the American ones are different and form their own group. One main problem with founder dispersal is explaining how movement between populations could be occurring at one time, but then at some point stop or at least decrease (leading to differentiation). What is the reason for the crucial change from high rates of dispersal to low rates? In theory this might be due to changing behavioral patterns in animals or means of dispersal in plants, but this cannot explain repeated patterns in unrelated animals and plants. Geological or climatic change is one obvious possibility and this is the basis of vicariance. In center of origin theory, dispersal and speciation are instead determined by chance - the change from movement to no (or less) movement is created by a 'barrier' which is permeable to a chance crossing by a single founder, but is then, somehow, impermeable to all others. In this view, the evolutionary biogeography of a group is due to chance and so there is no need to examine any details of distributions that cannot be attributed to local ecology.
Dispersalists have sometimes suggested that island endemic taxa, for example, had much more effective means of dispersal in the past than they do now, and that these were lost with evolution, 'trapping' taxa on an island (Carlquist, 1966a,b). This is an ingenious and logical solution to the general problem that is often not mentioned - what causes the change from a phase of dispersal to a phase of no dispersal? Unfortunately, the idea of 'loss of means' is probably wrong, as most endemics in most places have not lost their means of dispersal. But the fact that the idea was proposed at all indicates there is a problem that cannot be solved simply by citing 'chance'. Profound geological and ecological change is a more likely reason for cycles of immobilism - mobilism - immobilism.
Dispersal theory accepts that normal ecological dispersal and founder dispersal both occur in nature, whereas vicariance theory only accepts the first, but in any case it is important to distinguish between the two processes. The fact that a distinction is not made between contiguous range expansion (by normal dispersal) and across-barrier, founder dispersal is a serious drawback with programs such as DIVA (Kodandaramaiah, 2010). This conflation of the two different processes is a defining feature of dispersal biogeography (Matthew, 1915) and is also the basis of the confusing criticism that panbiogeography denies 'dispersal'.
To summarise: most modern biogeographers follow Mayr (1982, 1997) in accepting that allopatry can be caused either by vicariance (dichopatry) or by founder dispersal (peripatry), but only vicariance is accepted here. Allopatry is accounted for by immobilism and vicariance, while overlap among groups can be attributed to range expansion and population mobilism ('dispersal').
Dispersal: 'any and all changes in position'
Many birds, primates and other groups show daily and annual migrations that involve significant distances and are repeated through the millennia. These need to be accounted for in biogeography and ecology, although are usually dealt with separately. Clements and Shelford (1939) realised the problem and introduced the highly generalized, rigorously geometric concept of 'any and all changes in position'. This would include changes in position due to physical movement or to evolution. The authors' suggestion that this concept be termed 'dispersal' or 'migration' was elegant but confusing and never caught on. This does not detract from the value of the concept. As with global phylogenies and the evolution of major groups, daily migrations of animals, even at a local scale, may reflect either current ecological conditions or past conditions such as former streams, rivers or coastlines.
A phylogeny often has its main division, its basal break, between a small group and a more diverse sister group containing several clades. The smaller group is termed 'basal', although strictly speaking only the nodes or breaks between groups are basal; no group is more or less basal than its sister group. The term 'basal group' is thus potentially misleading as a basal group is no more primitive than its sister, and is not ancestral to it (Krell and Cranston, 2004; Crisp and Cook, 2005; Santos, 2007; Omland et al., 2008). Nevertheless, by now the term 'basal group' is widely used and understood in its purely topological sense and it is a useful term for a smaller sister group. It should probably always be used in quote marks, to indicate the problem, but then terms such as 'clade', 'monophyletic', 'dispersal', 'center', 'gene', etc. would have to be treated in the same way.
The phrases 'sister to the rest of' and 'basal in' are used here more or less interchangeably. The difference between the two is arbitrary and mainly nomenclatural - a basal group is considered to be part of the sister group and has the same name, a sister is a separate group and has a different name.
Although an ancestor would be basal in a phylogeny, a basal group is not necessarily ancestral or structurally primitive. For example, Amborella is likely to be the basal angiosperm, sister to all the rest, but Pennisi (2009: 28) went one step further and suggested: 'Given that placement, Amborella's tiny flowers may hint at what early blossoms were like'. In fact there is no reason why one (Amborella) or the other (all the other flowering plants) of the two sister branches should have a flower that is more primitive.
In the same way, the basal clade in a group is often interpreted as occupying the center of origin for the group, although this cannot be justified (Crisp and Cook, 2005). Likewise, morphological analysis may show that the oldest fossil clade in a group is phylogenetically basal to the rest, but it cannot be assumed to be ancestral to the others; it may simply be an extinct sister group.
Thus the idea that basal groups in a phylogeny are ancestral can be rejected as a generalization. Basal groups are simply less diverse sister groups and their distribution boundaries may represent centers of differentiation in what were already widespread ancestors, not centers of origin for the whole group (Heads, 2009a).
Despite these arguments, modern phylogeographic studies often assume that a 'basal' clade is primitive, ancestral, and located near the group's original center of origin, while advanced members of a clade have migrated away (Avise, 2000). The idea is derived from Mayr (1942) and Hennig (1966), who proposed that the primitive member of a group occurs at the center of origin. This is in contrast with the Darwinian model, which assumes that an advanced form would outcompete the older forms and force them to migrate away (Darwin, 1859, Matthew, 1915, Darlington, 1966, Frey, 1993, Briggs, 2003). In this view, the center of origin is occupied by derived forms. The conflict between the Darwinians and the Mayr/Hennig/phylogeography school over the center of origin is irrelevant in the vicariance model, where there is no center of origin to begin with.
To summarize, a 'basal' group is not ancestral, it is simply the smaller of two sister groups. Both will have the same age and neither one is derived from the other, or more advanced or primitive than the other.
Basal groups and centers of origin
Good examples of basal group/center of origin analyses are seen in the extensive literature proposing dispersal into and out of the Caribbean. In the mockingbirds, Mimidae, a Yucatán endemic is basal to a largely Caribbean clade. Lovette and Rubenstein (2007: 1045) argued that this 'is suggestive of a pathway of colonization into the Antilles from central America via Cuba...'. Conversely, in butterflies, the Greater Antilles genus Antillea is basal to a widespread clade (Phyciodina) of North, South and Central America. So from a center of origin in the Antilles, 'The ancestral Phyciodina colonized the [Antilles-Venezuela] landspan and spread south to the Guyanan Shield and then quickly to the Brazilian Shield' (Wahlberg and Freitas, 2007: 1265). (The subsequent scenario involved a convoluted history of transcontinental dispersals and back-dispersals, although the authors described these butterflies as 'well-known to be relatively sedentary').
In fact, no migration into or out of the Caribbean is required for the mockingbirds or the butterflies. In both groups, the location of the basal clade in the Yucatán/Greater Antilles region and the distribution of the rest of the group elsewhere can be explained by simple vicariance somewhere around Yucatán/Greater Antilles in an already widespread ancestor. The basal node represents an early center of differentiation in already widespread groups, not a center of origin. In a similar example, Sturge et al. (2009) wrote that molecular phylogeny 'confirms' that the New World oriole Icterus (Icteridae) colonized South America from the Antilles, but this was only because South American species were nested in an otherwise Antillean clade and a widespread ancestor is a more parsimonious solution.
Phylogenies can represent sequences of dispersal events or sequences of differentiation events in a widespread ancestor
Consider the group of four taxa A-D shown in Fig. 1-3 that are found in allopatric areas A-D and have a phylogeny (D (C (B + A))) In modern dispersal theory, the sequence of nodes in a phylogeny is read as a sequence of dispersal events, with taxa invading a new region, differentiating there, and then invading another region. The center of origin is occupied by the basal population in the basal group, D, and the phylogenetic sequence reflects a series of dispersal events from area D to C, B and A. Each of the four taxa has its own individual center of origin somewhere within its range.
The model has been criticized by vicariance biogeographers because the simple allopatry among the four clades might not be due to dispersal but to a sequence of in situ differentiation events in an ancestor that was already widespread in A-D. The phylogeny (D (C (B + A))) would then reflect a sequence of breaks among the areas: D vs. A+B+C, C vs. A+B, A vs. B. In this case, the sequence of differentiation shows a simple progression from east to west in Fig. 1-3 and there is no physical movement.
In other cases the phylogeny does not follow a simple geographic progression. Fig. 1-4 shows a pattern in which the two sequential basal clades in a group, A and B, are not adjacent geographically and are separated by other groups, C and D. Dispersal theory would attribute this to jump dispersal. In vicariance theory it indicates that a widespread ancestor differentiated first at two basal nodes (between A and the rest, then between B and C+D) and finally at a node geographically between A and B.
Thus a phylogeny may convey the impression of a center of origin at the locality of the basal clade and 'dispersal' from there, but if there was a widespread ancestor this is not necessary. Major disjunctions of tens of thousands of kilometers often occur between taxa at consecutive nodes on a phylogeny, even in groups in which long-distance, colonizing dispersal is improbable. Here it is especially likely that a phylogeny reflects a sequence of vicariance events. In many groups, differentiation has taken place repeatedly and more or less simultaneously around just a few globally significant nodes, such as the south-west Pacific basin and south-west Indian Ocean basin. Consider a phylogeny in five groups: Australia (Madagascar (Australia (Madagascar (Australia)))). In a dispersal interpretation, this would require repeated long-distance dispersal events backwards and forwards between the two centers. A vicariance interpretation of the same pattern proposes repeated differentiation at the same nodes in a widespread ancestor, perhaps caused by reactivation of tectonic features.
Vicariance interpretations of phylogenetic sequences are given in some recent literature. For example, in the New World snake Bothriechis earlier studies deduced a process of dispersal from Costa Rica to southern Mexico. Instead, Castoe et al. (2009: 98) interpreted the phylogeny as indicating 'a more simplistic northward progression of cladogenesis that requires no inference of dispersal' (italics added). A similar pattern is also seen in other snakes of the area, 'suggesting vicariance as the primary driving force underlying speciation'. The Great American Biotic Interchange theory proposes dispersal across central America, so the idea that the evolution in the region may not have involved physical movement is of special interest. In another example, Doan (2003) interpreted a phylogeny of Andean lizards as reflecting a northward sequence of speciation in a widespread ancestor, rather than northward dispersal. In a botanical study, the phylogeny of Rhododendron (Ericaceae) in Malesia was interpreted as a geographical progression of cladogenesis (Brown et al., 2006). The same method of interpreting phylogeny used in these papers is adopted here.
Summing up, phylogenies of extant clades can indicate a sequence of divisions (nodes) between sister groups, rather than a sequence of ancestors and descendants. Dispersalists have adopted the second option, but this has only led to long-lasting, unresolved debates about the center of origin in particular groups, about how to locate the center of origin in the first place, and what the means of dispersal could be. In the model of evolution proposed here, there is no center of origin (other than the point of break, which is a margin rather than a center), there is no founder dispersal speciation, and there is no 'radiation' from a center. If the ancestor is already widespread geographically (and probably also ecologically) before the differentiation of the descendant groups, the issue is no longer about how the modern groups 'reached' a certain area, but how they evolved there; in other words, where the breaks occurred which led to their differentiation. Once the spatial context is clarified, the question of timing should be more straightforward.
Dispersal-vicariance analysis (DIVA)
In the 'dispersal-vicariance analysis' of Ronquist (1997), inferences of dispersal events are minimized as they attract a 'cost'. Extinction also attracts a cost but vicariance does not. It was not explained why this approach should be taken and as suggested above it is based on a confusion of the two different concepts of 'dispersal'. Dispersal in the sense of ordinary movement should not attract any cost in any model. Jump or founder dispersal would attract no cost in a traditional dispersalist model, although in a vicariance model of speciation or evolution it is rejected a priori.
In most modern studies, the spatial analysis of phylogeny has been based on the idea of a center of origin and so authors employ programs, such as DIVA, that will often find one. Authors looking for a particular center of origin sometimes complain that DIVA will find a widespread ancestor if, for example, all the extant groups are allopatric. But even when they are not, a widespread ancestor can still be proposed, as original allopatry may have been obscured by subsequent range expansion or extinction. There is no logical need to interpret a phylogeny as a series of dispersal events.
Groups that are reciprocally monophyletic in two areas
If one group occurs on a continent, A, and its sister group occurs on a much smaller island, B, (Fig. 1-5), the island group is often assumed to have been derived from the mainland group by dispersal. The island forms are predicted to be related to particular populations in their large sister group. Yet well-sampled molecular studies now show that in many of these cases the phylogeny has the pattern: (A1, A2, A3...) (B1, B2, B3...), where the superscripts indicate different areas within A and B. The groups in the two areas are reciprocally monophyletic and the group in B is not related to any one population in A. In this type of pattern, dispersal can still be salvaged as an explanation, but only if it occurred prior to any other differentiation in the groups and this is often unlikely. On the other hand, reciprocal monophyly is the standard signature of simple vicariance of a widespread ancestor at a break between A and B. Even if groups in the two areas A and B are not reciprocally monophyletic, vicariance is still possible and this is discussed next.
Groups with a basal grade in one region or habitat type
A monophyletic clade includes all the branches derived from a single node. A paraphyletic group or grade comprises several sequential branches of a phylogeny, but does not include all the branches derived from a node (e.g. in Fig. 1-3 the clades B, C, and D, but not A,). Many groups comprise a basal grade located in one area, A, and a disjunct population or clade in a second area, B. The pattern is usually explained as the result of dispersal of the clade from A to B. Instead, a grade located in a single area may represent a phase of differentiation there, not a center of origin (Heads, 2009b). For example, within a widespread ancestor already in A and B (Fig. 1-6) allopatric evolution may occur around a node at A. If this is followed by secondary overlap at A and extinction of populations between A and B, this will produce a basal grade in area A (Fig. 1-6D). In actual cases the overlapping clades in area A often show slight but significant differences in their distribution (as indicated in Fig. 1-6D) and these may represent traces of the earlier phase of allopatry. Many biogeographic analyses treat all the species in an area such as A as having the same distribution, overlooking any allopatry, and this can confuses analysis. To summarize: a dispersal analysis interprets the pattern described here, with a basal grade in A, by long distance dispersal from A to B 'across a barrier'. Instead, a vicariance analysis infers differentiation in a widespread ancestor followed by local overlap within A by normal means of dispersal.
Examples of groups with basal grades in one area
The Arctotidinae (Asteraceae) are a good example of the basal grade pattern. The group comprises a basal grade of three southern African clades and also Cymbonotus of southern Australia which is sister to two other southern African genera (Fig. 1-7; Funk et al., 2007). Cymbonotus is sister to a southern African clade and embedded in an Australian-southern African clade; it is not embedded in a southern African clade. It is likely that the southern African clades show significant differences in their distributions within the region.
In another example, Protea (Proteaceae) is diverse in the Cape region of South Africa and also has a few species widespread throughout tropical Africa. Valente et al. (2010) found that 'non-Cape species are nested within a wider radiation of Cape lineages and all except two of them belong to a single clade'. 'Therefore', the authors suggested, 'most extant lineages outside the Cape originated by in situ diversification from a single ancestor that arrived there from the Cape' (p. 746). This logic is not accepted here (cf. Fig. 1-6). The authors only interpreted the phylogeny in terms of a dispersal model because of the programs they used and an alternative vicariance scenario was not considered. Nevertheless, the sister genus of Protea is Faurea, centered in tropical Africa (where Protea has low diversity) and Madagascar (where Protea is absent); a vicariance analysis would be simple and of considerable interest.
In a third example, Grandcolas et al. (2008: 3311) argued that 'within certain New Caledonian groups, multiple species are nested within larger clades with taxa from Australia, New Zealand or New Guinea, calling for explanations in terms of recent dispersal [to New Caledonia]'. Thus the phylogeny: (Australia (Australia (Australia (New Caledonia)))) is taken to reflect a center of origin in Australia, where the 'basal grade' occurs. An alternative explanation for the pattern would be differentiation in a widespread Australian-New Caledonian ancestor, as in Fig. 1-6.
Center of origin/basal grade theory and ancestral habitat reconstruction
As discussed, a basal grade does not necessarily indicate a geographic center of origin and this argument also applies to ecology, as a basal grade is often thought to occupy the ancestral habitat. For centropagid crustaceans, Adamowicz et al. (2010) concluded that 'Species occupying saline lakes are nested within freshwater clades, indicating invasion of these habitats via fresh waters rather than directly from the ocean or from epicontinental seas' (p. 418). But using the same argument given above for geography, this ecological phylogeny: (freshwater (freshwater (freshwater (saline lakes)))), does not necessarily mean that freshwater habitat was the center of origin for the saline lake clade. The ancestor of the whole group may have occupied both freshwater and saline lakes and itself be derived from a marine ancestor, for example following marine transgression and regression.
In other crustaceans the spiny lobsters, Palinuridae, are widespread in warmer seas and especially common around Australasia. Tsang et al. (2009) reasoned that the three genera restricted to the southern high latitudes (Jasus, Projasus and Sagmariasus) are the basal lineages in the family, 'suggesting a Southern Hemisphere origin for the group'. In the same way, the authors assumed that the basal groups indicated the ecological center of origin. For one clade they wrote: 'the shallow-water genus Panulirus is the basal taxon in Stridentes, while the deep-sea genera Puerulus and Linuparus are found to be derived. This indicates that the spiny lobsters invaded deep-sea habitats from the shallower water rocky reefs and then radiated'. Again, the habitat of the basal taxon is not necessarily a center of origin for the other groups and the ancestor may have already been widespread in deep and shallow water before it differentiated into the modern genera.
Groups with a basal grade in one region and widespread distal clades
Many groups have a basal grade in one region, as in the last pattern, and also have a widespread distal clade.
The dandelion tribe Cichorieae (= Lactuceae) of Asteraceae is cosmopolitan and has its basal clades and also its sister group (Gundelieae), i.e. a basal grade, in the Mediterranean region (Funk et al., 2005). As usual, this does not mean the Mediterranean basin was a center of origin from which the tribe itself spread. Instead, the region may have been an early center of differentiation in an already global ancestral Cichorieae.
The family Asteraceae as a whole, the largest plant family with ~24 000 species, is another example of a group with a basal grade in one region and a widespread distal clade. Its basal group, subfam. Barnadesioideae, is in South America. The sister-groups of the family are centered on the eastern Pacific margin (Calyceraceae of South America) and western Pacific (Goodeniaceae, mainly in Australia; Stevens, 2010). Bremer (1994) concluded that 'the geographic origin of the Asteraceae most probably involved the South America and the Pacific' - a vast area - whereas Stuessy et al. (1996) instead suggested a much smaller area (about 200 miles across) in southern Argentina. The conflict reflects different approaches to the center of origin concept and to fossils: Stuessy et al. (1996) wrote that a vicariance model 'seems unlikely because available fossils provide no evidence...', whereas Bremer (1994) accepted that 'groups may be much older than their fossil record..'. Stuessy et al. (1996) suggested that Barnadesioideae are not only the sister group to all the other Asteraceae, they actually 'gave rise' to the rest of the family. This is unnecessary and unlikely. The 'sister as ancestor' theory is often adopted as it is compatible with a localized center of origin model, but Bremer (1994) took a broader perspective in suggesting that the origin of the Asteraceae could have been linked to the history of the Pacific area.
A recent phylogeny for the Asteraceae (Fig. 1-8, from Funk et al., 2005, and Panero and Funk, 2008) is: (Barnadesioideae (Mutisioideae (Stifftioideae (Wunderlichioideae (Gochnatioideae + seven remaining subfamilies))))). The sister group of the whole family, Calyceraceae, and the five basal branches are all small groups mainly found in South America, although some have outliers elsewhere. (Mutisioideae have a few representatives in Africa and North America, and one genus, Liebnitzia, is in China and Mexico. Wunderlichioideae have two genera in Southeast Asia. Gochnatioideae also occur in Central America, North America and Asia).
Fig. 1-8 here
Most authors accept that this phylogeny indicates a center of origin in South America. Instead, it may reflect evolution of a world-wide ancestor in which the modern groups differentiated at breaks in or around what is now South America. The process is the same as that shown in Fig. 1-6, with South America equivalent to the area on the left, although in Asteraceae the distal clade has subsequently expanded its range to include South America. Funk et al. (2005) wrote 'it appears incontrovertible' that the family itself had a center of origin in southern South America. But there is no reason why the site of the initial splits in the modern group should indicate the distribution of the ancestor and the authors did not consider the possibility of a globally widespread ancestor.
Some of the oldest known fossil flowers of Asteraceae are from the Eocene of Patagonia. These show affinities with subfamily Mutisioideae and also the distal clade Carduoideae. Barreda et al. (2010) regarded the fossils as support for the hypothesis of a South American origin of Asteraceae and an Eocene age of divergence. Yet the fossils are also compatible with much broader area of origin for the family and much earlier divergence within the family. The authors suggested that 'an ancestral stock of Asteraceae may have formed part of a geoflora developed in southern Gondwana'. Another possibility is that the group was already more or less world-wide when it originated by splitting from its sister group, Calyceraceae, in or around its range in southern South America.
After the five basal South American groups diverged from the remaining Asteraceae, the monotypic Hecastocleis of California and Nevada (mountains in the Mojave Desert) is sister to the rest, within which a large African group is sister to the remaining, cosmopolitan clade. Funk et al. wrote that the South American differentiation followed by the African phase 'might suggest a Gondwanan origin for the family'. Again, the observed phylogeny could instead have developed from an ancestor that was already cosmopolitan. Differentiation in the global ancestor occurred first around phylogenetic/biogeographic nodes or breaks in or near South America, then California (part of the Pacific, not Gondwana), and finally Africa, and this could reflect a sequence of differentiation events, not a route of dispersal.
Funk et al. (2005) wrote that 'the few data from pollen records and geology seem to indicate a more recent [Cenozoic] origin for the family' and relied on this date, along with the phylogeny, in deducing intercontinental dispersal. However, this is not necessary if the fossil pollen dates and calibrations are only minimum, not absolute dates and migration leading to overlap occurred over the very different geography of the Mesozoic. As Funk et al. stressed, the general perception of ecology in Asteraceae as simply 'weedy' is incorrect. Some species are indeed cosmopolitan or pantropical weeds yet the vast majority are restricted-range endemics. The main clades are notable for their conspicuous geographic centers of diversity in different areas, for example, Stifftioideae in eastern Brazil, Liabeae in Peru, Heliantheae s.lat. from Mexico to the northern Andes, Calenduleae in South Africa, and Gundelieae around the Mediterranean. The extensive regional endemism means that while the family is very large there are relatively few global clades (several of the larger tribes and subtribes, a few large genera). Thus in the history of the family there have only been a small number of widespread ancestors (for example, groups such as Senecioneae and Astereae each require their own global ancestor). These few ancestors may have undergone a phase of active mobilism in the Mesozoic during which they occupied much of Earth's land surface, before settling down into a Cenozoic phase of immobilism and speciation. Although many modern species of Asteraceae are narrow endemics, even these often retain a weedy ecology within their local centers of endemism, occupying unstable sites such as cliffs and rocky outcrops on steep mountain slopes.
Other groups with basal clades or grades in South America:
The following groups resemble Asteraceae in having their basal nodes in or around South America and provide further illustrations of the principles discussed here.
The distribution of the plant Gunnera, the only genus in Gunneraceae, is shown in Fig. 1-9 along with its sister, the Myrothamnaceae. The molecular phylogeny of Gunnera (Wanntorp et al., 2002) matches the morphological classification into subgenera and sections (van der Meijden, 1975). The basal clade is a very small plant of Uruguay and southern Brazil (Rio Grande do Sul, Santa Catarina) that forms dense mats on seepages in coastal sand dunes. The rest of the genus occurs in the mountains of the tropics and the south temperate zone, although it is absent from eastern tropical America and western tropical Africa. Fossil pollen from the Early Cretaceous is recorded in the southern continents and also from North America (Stevens, 2010). The Myrothamnaceae replaces Gunnera in large areas of south-central Africa.
Fig. 1-9 here
Reading the phylogeny of Gunnera from the bottom to the top, the first phylogenetic break occurred around what is now south-eastern Brazil, isolating the basal clade from adjacent populations in South America and Africa. The second break occurred in the Indian Ocean (between clades 2 and 3), the third in the Pacific (between 3 and 4). The phylogeny follows a geographic sequence from South America eastwards and this would usually be interpreted as a sequence of dispersal events. Instead, starting with a widespread ancestor, the sequence of phylogeny could represent a series of vicariance events, a wave of evolution passing around the Earth through the population.
In a dispersal analysis, the focus is on the question: how did Gunnera cross the Atlantic, Indian and Pacific Oceans? In a vicariance analysis, the focus is instead on the phylogenetic and geographic breaks, their exact locations, and their possible geological causes. If Gunnera evolved by vicariance with Myrothamnaceae (perhaps at a node between western and eastern Africa), it may have already been widespread in the southern continents at the time of its origin.
Verbenaceae, Solanaceae, and Bignoniaceae
Marx et al. (2010) proposed that these plant families all originated in South America and colonized the Old World on multiple occasions, but this was only because the basal clades in each family are in South America. A widespread ancestor/vicariance interpretation is an alternative possibility.
The basal clade in the grass family is Anomochlooideae (Bouchenak-Khelladi et al., 2010), which comprises two rainforest genera:
Streptochaeta: throughout mainland tropical America,
Anomochloa: coastal Brazil (Bahia).
The pair highlights the significance of a break at or near coastal Brazil.
Genlisea occurs in tropical America (mainly in the east; absent in the Andes) and in Africa. It comprises two main clades (Flieschmann et al., 2010):
1. Subg. Tayloria: south-eastern Brazil
2. Subg. Genlisea: Africa + South America, including south-eastern Brazil.
Flieschmann et al. (2010) adopted a center of origin model and because of the basal grade in South America inferred dispersal from South America to Africa (and then back again to South America). In an alternative, widespread ancestor/vicariance model, the primary split is between Africa + America (except south-eastern Brazil) on one hand, and south-eastern Brazil on the other. The only dispersal required is a range expansion of clade 2 into south-eastern Brazil to account for the overlap with clade 1. The split in clade 2 at the Atlantic occurred after the split between clades 1 and 2 around south-eastern Brazil.
A group with a basal grade in South and Central America: Oxyura and Nomonyx ducks
Allocation of clades to a priori geographic areas, such as the continents, in the initial stages of biogeographic analysis has often involved incorrect assumptions of sympatry. This in turn has led to the idea that the 'areas of sympatry' were centers of origin. biogeographic analysis does not reqire the use of any area, other than those defined by the taxa themselves. For example, the ducks Oxyura and Nomonyx (Fig. 1-10) form a globally widespread clade with a phylogeny that could be presented as: (America (America (America + Old World))). Although this is accurate descriptively it would be misleading as the basal clades are presented as sympatric and 'America' could be misinterpreted as an area and as a center of origin. In fact, the groups are largely allopatric and the phylogeny is better summarized as: (eastern America (southern America (western America + Old World)))). Although there are clades in different parts of America, no clade defines South America or America and these geographic entities may not have been important for the biogeographic history.
Fig. 1-10 here
For the Oxyura-Nonomyx group, McCracken and Sorenson (2005) suggested a center of origin in lowland South America, as the two basal members occur there. Subsequently a complex series of intercontinental dispersal events led to the modern distribution. Yet this scenario does not explain the fact that the three main clades are largely allopatric, with secondary overlap restricted to the Greater Antilles, Colombia and northern Argentina. There is an overall east/west split (with, for example, Nomonyx in eastern Mexico, Oxyura in western Mexico) and the three main clades meet at a node in north-western Argentina. This may be near the original break. The three Old World clades are also allopatric (in Eurasia, Africa and Australia), and so the simplest scenario is a world-wide ancestor breaking up into six vicariant clades, with some local overlap. Testing this idea would require examining the patterns in detail at intercontinental and also local, ecological scales and seeing if they are shared with other groups.
If we break 'America' down into its components, the phylogeny for the group of ducks is: (Brazil + Greater Antilles (Argentina (western North America + Greater Antilles + Andes) (Old World))). The first division is between Brazil-Greater Antilles (Nomonyx) and the rest of the world (Oxyura), with breaks around the Greater Antilles, Colombia/Ecuador, and northern Argentina. America does not appear as a monophyletic area, but as a composite of eastern and western sectors, each with endemic taxa that are not sisters. This is compatible with the tectonic division of the Americas into eastern (cratonic) and western (orogenic/accreted terrane) provinces. The second division in the phylogeny involves breaks in western Argentina, perhaps around the basins that were later incorporated in the uplift of the Andes. The third division implies differentiation of the western North America-Antilles-Andes group and its Old World sister. Ducks are known from Cretaceous fossil material and this provides a useful minimum age for the group (Clarke et al., 2005).
Populations of Oxyura and Nomonyx inhabit lakes, swamps, and sometimes brackish water (del Hoyo, 1992, Kear, 2005). Nomonyx is recorded in mangrove. In North America, O. jamaicensis breeds inland, in the northern prairies and south into the intermontane basins and valleys of the western United States, reaching the coast in southern California. In the northern Andes, O. j. andina occurs at 2500-4000 m. These populations are now montane and even alpine, but they may have been derived from ancestral complexes which lived in the mangrove swamps and lakes of the pre-Andean Cretaceous basins. Uplift in the Andean region began in the Cretaceous and the birds here would have been lifted up with the land, a process suggested for Andean parrots by Ribas et al. (2007). In the ducks, most of the populations which differentiated in Brazil as Nomonyx lay too far east to be caught up in the main Andean orogeny and remained in the lowlands.
Case studies: basal nodes around the Philippines and Madagascar
The same methods used above can be applied to distribution centered in and around the Indian Ocean basin. This is an important biogeographic pattern and the three examples cited next illustrate sequences of differentiation that are typical for groups in the region.
Old World rats and mice
All four subfamilies of the family Muridae are restricted to the Old World. Strangely, despite the weedy ecology of some members the group is completely absent from the Americas. In contrast, all five subfamilies of the related Cricetidae are present in America. The two families overlap in Eurasia.
One of the murid subfamilies, the Murinae, include typical rats and mice and occur throughout the Old World. In Murinae, one of the clades in the Philippines, the pair Batomys and Phloeomys, is sister to all the rest (Steppan et al., 2005, Rowe et al., 2008). Nevertheless, 'Whether the Philippine Old Endemics represent a relictual distribution from the periphery or the core of the ancestral range of the Murinae cannot be determined'(Steppan et al., 2005: 382). In its geography and its phylogeny the Philippines group is peripheral to the main bulk of the Old World group and it might be thought to have 'budded off' its more widespread sister. But this would have to have occurred before any other differentiation in the group and instead, the Philippines group could represent the sole remains of a group that was formerly more widespread in the Pacific region.
As with America, the Philippines archipelago is a geological composite, made up of many fault-bounded blocks of crust with independent histories, or terranes (see the Glossary for this and other geological terms). Some Philippines terranes have an Asian origin while others formed further east in the central Pacific (Metcalfe, 2006). It is possible that the two main clades of Murinae - the small basal clade in Philippines and the rest - are respectively West Pacific and Indian Ocean in origin. Secondary juxtaposition of the two followed the collision of many Pacific terranes with New Guinea and the Philippines.
Terrane accretion in the west Pacific also explains the strong biogeographic connection between the Philippines and northern New Guinea, including its offshore islands. In the main group of Murinae, beyond the basal genera, Steppan et al. (2005) and Rowe et al. (2008) found a close relationship between another Philippines clade (Apomys, Archboldomys etc.) and a large clade termed the 'Old endemic genera' of New Guinea and Australia. In this group, three of the four tribes are mainly in New Guinea. Steppan et al. (2005) described this Philippines - New Guinea/Australia connection as 'perhaps [their] most surprising finding'. In geology, the sector: New Guinea - northern Moluccan Islands - eastern Philippines, is recognized as a belt of deformation and Pubellier et al. (2003, 2004) suggested that the portion of the Philippine Sea plate carrying the Taiwan-Philippine arc may have originated closer to Papua New Guinea. Biogeographic connections between Philippines and New Guinea may be explained in this way.
Steppan et al. (2005) suggested that the murines 'appear to have originated in Southeast Asia and then rapidly expanded across all of the Old World', but this was only because 'three of the four basal branches... include taxa almost entirely restricted to South east Asia'. In fact, the four clades have quite different distributions (Rowe et al., 2008), as follows:
the Philippines (the basal group),
Southeast Asia and west to Eurasia and Africa,
Southeast Asia and areas to the east and south (Philippines and New Guinea/Australia).
In a simple vicariance model a widespread Old World/Pacific ancestor has undergone early differentiation around the Philippines/southeast Asia. Subsequently, the clades have been juxtaposed and there has been range overlap. Rowe et al. (2008: 97) suggested that 'The pattern that emerges from these phylogenies [of Murinae] is of rapid and probably adaptive radiations after colonization of landmasses previously unoccupied by muroid-like rodents'. Yet there is no special reason to assume that the phylogeny reflects a sequence of dispersal events rather than differentiation in an already widespread ancestor.
The main group in the owl genus Otus comprises three allopatric clades, structured as follows (Fuchs et al., 2008):
1. Africa, Pemba Island (Tanzania), Mediterranean to central Asia (O. scops etc.).
2. Philippines (O. mirus and O. longicornis).
3. Madagascar, the Comoros, the Seychelles, Sri Lanka, and eastern mainland Asia to northern
China (O. sunia, O. rutilus etc.).
The phylogeny is: 1 (2 + 3). The first break is in Mozambique Channel, the next is around the Philippines. As usual, the two breaks may represent successive divergence events in an already widespread ancestor, rather than dispersal across the Indian Ocean and back again. It is easy to see how even a simple phylogeny of a widespread group with allopatric clades can generate a convoluted dispersal scenario if every differentiation event requires physical movement.
The freshwater gastropod family Pachychilidae (Köhler and Glaubrecht, 2007) has a widespread Indo-Pacific distribution with the clades distributed as follows:
3. India to southern China and Java.
4. Sulawesi, Torres Strait, Central America.
The phylogeny is: 1 (2 (3 + 4). As in Otus, the sequence of differentiation 'jumps' across the Indian Ocean, in this case from around the Philippines to around Madagascar. Clade 4 is a typical trans-tropical Pacific affinity and this pattern is discussed further below (see Chapter 6).
The magpie robins, Copsychus s.s. (Muscicapidae), comprise three main clades (Lim et al., 2010):
4. India to China and Borneo.
The phylogeny is 1 (2 (3 + 4), giving a sequence of differentiation events similar to that of the pachychilids, despite the very different ecology and means of dispersal in the two clades.
Character incongruence in modern clades and polymorphism in ancestral complexes
The center of origin/dispersal model of biogeography is based on the idea that the ancestor of a clade (either a single parent pair or at most a uniform species) is monomorphic and that the clade had a small, single center of origin. The new taxon is separated from the ancestor by a chance, one-off dispersal event. In this model, each character in the ancestor has only one state and so all characters in the new taxon are either primitive (resembling the ancestor) or new. The new taxon is not simply a recombination of ancestral polymorphism and does not 'emerge' over a broad region. .
In contrast, vicariance theory stresses that a hierarchical tree diagram does not represent all aspects of phylogeny. A phylogeny is a summary diagram only and cannot portray all aspects of differentiation, such as characters showing variation that is 'incongruent' with the phylogeny. For example, in three genera with a phylogeny: a (b +c), a character may occur in a and b, but not c. Although hierarchical classifications are often useful they are only summaries and have limitations. The many structural features, both morphological and molecular, that underlie the clades can be distributed in different ways in the phylogeny and may show different geographic patterns. The characters and their variation within a clade may date back to before the origin of the clade as such, for example if evolution has proceeded by hybridism or incomplete lineage sorting. In the latter case, character incongruence in a phylogeny reflects polymorphism that was already present in the ancestor before it began to differentiate into the modern taxa. In this way, as in hybridism, new clades may emerge over a broad range as new recombinations of ancestral characters.
Many examples of groups of are now known in which the variation and geography of one gene is incongruent with that of another and this cannot be depicted in a single tree. Workers on some groups do not take the tree metaphor as seriously as others. For example, in prokaryotes horizontal gene transfer is pervasive and 'very few gene trees are fully consistent, making the original tree of life concept obsolete' (Puigbó, 2009). Botanists often feel the same way and there is increasing appreciation of the role of hybridism in animals. In the case of incomplete lineage sorting, the ancestral polymorphism may be topologically incongruent with the phylogeny of the modern descendants and also geographically incongruent in its geographic distribution.
Fig. 1-11 shows a hypothetical clade in which patterns of variation in two parts of the genome show geographic trends lying at right angles to each other. One of the patterns may represent variation in the ancestor. Fig. 1-12 shows an actual example from the fish Nemadactylus in which the pattern of variation in cytochrome-b is phylogenetically and geographically incongruent with clades shown in D-loop sequences (Burridge, 1997). Burridge discussed whether this pattern was due to hybridism or to incomplete lineage sorting and concluded in favor of the latter. In any case, patterns of 'incongruent' variation have long been known in morphological studies and are now well-documented in many molecular accounts. It seems that lineages do not necessarily evolve in a linear or hierarchical way and that phylogeny may develop through the recombination of ancestral characters, rather than the evolution of any new, uniquely derived characters.
Fig. 1-11 here
Fig. 1-12 here
A typical example of incongruence occurs in Bystropogon (Lamiaceae), endemic to the Canary Islands and Madeira. Chloroplast DNA supports a relationship of Bystropogon with New World groups, while nuclear DNA shows connections with Old World groups (Trusty et al., 2004). The authors concluded that due to the 'apparent conflict... we are not certain of the true biogeographic relationship of Bystropogon' (p. 2004). In a vicariance framework, both sets of affinities are accepted as valid and there is no real conflict. The New World, Macaronesian, and Old World genera each evolved in situ out of a widespread ancestor by recombination of ancestral characters.
Extinction, fossils and centers of diversity: Nothofagus (Nothofagaceae), the southern beeches
All clades go extinct and so extinction is clearly a major factor in distribution. How does it affect analysis of larger clades? Other things being equal, extinction may occur first in areas of lower diversity and this would be most likely in large, widespread clades. This idea was tested in the tree genus Nothofagus, the only member of Nothofagaceae (Heads, 2006a). This genus was studied as it is a diverse, widespread group with a fossil record that is exceptionally good relative to most other groups.
Nothofagus trees dominate forests in many parts of Australasia and southern South America. In a dispersal paradigm, this trans-oceanic distribution presents two problems: where is the center of origin? And how exactly did Nothofagus move from Australasia to South America (or vice versa)? Instead, if the problem is seen in a broader phylogenetic and geographic context the first question is how to integrate the distribution of Nothofagaceae with that of its relatives. Nothofagaceae are the basal group in the Fagales (oaks and beeches etc.), a group that is distributed all around the world. The different families of Fagales have their respective centers of diversity in different places: Nothofagaceae in Australasia and southern South America, Casuarinaceae in Australia, Rhoipteleaceae in China and Southeast Asia, Fagaceae in North America and Eurasia, Ticodendraceae in Central America, etc. This suggests that Nothofagaceae and the other families arose by vicariance in a widespread ancestor that was already present in Australasia and South America before Nothofagaceae evolved.
There has probably been dispersal (range expansion) in the early Fagales, before they attained their world-wide distribution and differentiated into families. In contrast, the families themselves show a large degree of allopatry and so there may be no need to infer any great range expansion of Nothofagus itself, for example across the South Pacific.
After the allopatric evolution of the families, overlap has developed around their margins. For example, Fagaceae extend south from the northern hemisphere to New Guinea and overlap there with Nothofagus (which is usually at higher altitude). Casuarinaceae have a clear center of diversity in Australia, but a few widespread species occur in New Guinea and, a few fossil species are known from New Zealand and South America. In New Caledonia both families are diverse. Thus although the distribution of Casuarinaceae overlaps that of Nothofagus completely, with one exception this only occurs in areas where the levels of diversity in Casuarinaceae are low. The exceptional overlap in New Caledonia, geologically a composite island, is a frequent pattern.
Vicariance with some secondary overlap occurs between Nothofagus and its relatives and also occurs within Nothofagus. Counting extant species, the four subgenera have vicariant centers of diversity as follows:
Subg. Fuscospora New Zealand,
Subg. Brassospora New Guinea and New Caledonia,
Subg. Lophozonia Chile north of Chiloé Island,
Subg. Nothofagus Chile south of Valdivia.
All four subgenera are known from Upper Cretaceous fossils. Thus the process of differentiation from within the order to within the genus may have been caused by vicariance and there is no need for any center of origin or long-distance, trans-oceanic dispersal. The only dispersal required is that needed to explain the secondary overlap of Nothofagus with Fagaceae (in New Guinea) and Casuarinaceae (especially in New Caledonia), and local overlap among the Nothofagus subgenera.
What has been the affect of extinction on the biogeography of Nothofagus? Nothofagus is an abundant, wind-pollinated tree with an extensive and diverse fossil record from the Cretaceous on. Many fossil species have been described and if these are added to the extant ranges, the total distributions of the subgenera are all enlarged and overlap. Nevertheless, the centers of subgeneric diversity do not change; they are the same as those cited above for the extant species alone (Heads, 2006a). This is compatible with the idea that, in general, total extinction occurs outside a group's center of diversity rather than in the center of diversity itself. If the areas of overlap among the Nothofagus subgenera do represent secondary range expansions, the original differentiation of the subgenera may have been purely allopatric.
Most clades go extinct eventually but the main centers of diversity, as in the families of Fagales and the subgenera of Nothofagus, may survive extensive and long-lasting phases of 'erosion' and local extinction. A group with 100 species in South America, one on Hawaii, and one in Taiwan may suffer extinction but in terms of biogeography it is likely to be the outlying single species that disappear, not the center of diversity in South America.
It is sometimes suggested that a local or regionally endemic clade may have formerly occurred everywhere and that extinction alone has produced its distribution. Fossils show that extinction has reduced the range of many taxa, for example, the Nothofagus subgenera, and wiped out others completely. But biogeography has to explain the data available, both fossil and extant records, rather than invoking possible extinction as a reason to give up analysis and, in any case, extinction may tend to occur in areas that were only occupied secondarily to begin with. As Allwood et al. (2010: 676) concluded: 'without fossils, hypotheses on extinction are ad hoc, unable to be tested and can be used to explain almost any biogeographic pattern... while acknowledging the potentially confounding effects of extinction, we have no evidence that this process has misled our biogeographic interpretations'.
Dispersal and two concepts of chance
A key factor often cited in chance dispersal is the 'long distance' involved. On the other hand, chance or jump dispersal is also used to explain differentiation over short distances, for example, across a river. Thus the defining feature of chance dispersal is not so much distance, as the fact that the physical movement by the founder is a single, freak event. The process is 'chance', not in the sense of being analyzable statistically, as normal, ecological dispersal is often analysed, but in the sense of not being analyzable at all. In this model, plants and invertebrates that are restricted to dense rainforest and move only meters in their lifetime are proposed to have moved - just once - thousands of kilometers across open ocean to attain a distribution in, say, New Guinea and Colombia. No further analysis is given, even if there are no obvious means of dispersal. Chance dispersal is justified because 'anything can happen given enough geological time', but the process is unfalsifiable and explains all distributions and none at the same time.
The new, revolutionary idea of chance as a calculated probability began with Pascal and Fermat in the 17th century and became a founding principle of modern science. For example, a probability-cloud diagram of seed dispersal depicts how 'chance' (in the new, modern sense) determines aspects of normal ecological dispersal. On the other hand, 'chance dispersal' uses chance in the old sense of 'factors beyond our understanding' or 'given enough time anything can happen'; 'chance' here is a very different concept and is not a real explanation for anything. If the process leading to biogeographic distribution is due to 'chance' in this old sense, every clade would have its own individual biogeographic history independent of all others. There would be no real biogeographic patterns, only pseudopatterns including pseudovicariance, that is, vicariance caused by chance geographic coincidence of chance dispersal events at different times. Community-wide patterns of allopatric differentiation caused by a new mountain range or a new channel of the sea, for example, would not occur. In contrast, molecular work has shown high levels of precise vicariance in most taxa, including widespread marine species and even bacteria (Fenchel, 2003). In protozoa, the amoeba Nebela vas was found to have a typical Gondwana plus South Pacific islands distribution (Smith and Wilkinson, 2007), contradicting the usual paradigm of microbial distribution in which 'everything is everywhere'. Foissner (2006) reviewed microbial biogeography and also stressed endemism in different regions, including Laurasia, Gondwana, and west Gondwana (eastern South America, Africa). Foissner compared the patterns with those of spore plants, 'many of which occupy distinct areas, in spite of their minute and abundant means of dispersal' (p. 111).
The profound geographic structure evident in communities is not explained simply by appealing to chance. Many authors who conclude that long distance dispersal has occurred often admit that the evident, normal means of dispersal in the studied organism do not appear sufficient for the postulated dispersal events. Sometimes authors suggest that more study is required to clarify the means of dispersal, but the dispersal events suggested are one-off events, unique in geological time, and it is hard to imagine how they could ever be studied. By definition they do not conform to any general laws and are completely unpredictable. Many authors who support dispersal wisely avoid discussing the problem.
A good example is the alpine cushion-plant genus Abrotanella (Asteraceae). There are three clades, distributed in:
western Tasmania and Australasia,
Stewart Island (southern New Zealand) and South America.
The last two clades overlap only on the summit ridge of the highest mountain in Stewart Island, Mount Anglem. Wagstaff et al. (2006) concluded that distributions in Abrotanella 'undoubtedly' reflect a 'convoluted history of dispersal' but this idea (based on a fossil-based chronology) does not explain the precise dovetailing of the clades at intercontinental, regional and even local scales. In addition, these plants do not have the feathery pappus on the 'seed' that is usual in members of Asteraceae such as dandelion and allows effective wind-dispersal. Wagstaff et al. (2006) concluded that long distance dispersal across the South Pacific (10 000 km) must have occurred, but they did not speculate as to how it could have actually happened. Perhaps it never did.
What is an 'origin' in evolution?
In the vicariance model of evolution, a group does not originate at a point within itself, but over a large area and by articulation at its margins with another group. Instead of the group having an absolute origin at an ideal point, its ancestor was always already widespread and differentiated and so the modern group 'originated' only in the sense that it 'became recognisable'. In the 'unfolding' of organic evolution, entities originate by differentiation of former entities, not by starting at a point and expanding. In contrast, in center of origin/dispersal theory an ancestor develops at a single point and is homogeneous - there is no differentiation. Its descendant taxa start from 'nothing' in this sense and develop into biological entities. Other biological entities that have an absolute origin include the uniquely derived characters of cladistics and the taxa of special creation. Again, if the ancestor is polymorphic, the clades do not necessarily represent an absolute beginning and may instead be recombinations of features and genes that already existed.
Critique of founder speciation
Founder speciation has been criticized in studies of genetics and biogeography, with the critiques based on different data sets, and the concept may not be necessary.
The critique of founder speciation in population genetics studies
In 'founder dispersal', unlike normal ecological dispersal, the founder is isolated from its parent population by 'dispersing over a barrier' - an apparent contradiction - and the new population then diverges into a new species. In the biogeography of the modern synthesis, this is achieved by a second unlikely process, the 'genetic revolution' produced by the founder effect. The founder effect is well-established in genetics but whether it can lead to speciation, via some sort of a 'genetic revolution', is much more controversial. Nevertheless, until the recent revival of ecological speciation, dispersal theory was based on it.
Mayr (1954) stressed geography in speciation and did not accept that speciation was simply ecological. He introduced founder effect/genetic revolution as a mode of geographic differentiation in order to account for various distribution patterns in New Guinea birds. He developed the idea from field observations, not from genetic studies. He was especially interested in birds endemic to small islands north of New Guinea and separated by relatives on the mainland only by narrow sea passages. The idea he proposed to explain the pattern, a combination of chance dispersal and founder effect/genetic revolution, became almost universally accepted in explanations of island endemism and as a paradigm for geographic evolution in general. One of the most studied localities in this research has been Hawaii and the classic case there is the fruit-fly Drosophila. There are about a thousand species of Drosophila species in Hawaii (O'Grady et al., 2009), many more than would be expected given the area of the islands, and geneticists developed an explanation for this diversity anomaly based on Mayr's theory (Templeton, 1980, 1998, Carson and Templeton, 1984). This work has provided a theoretical basis for modern accounts of founder speciation/chance dispersal.
While many biogeographers have accepted the argument from genetics, geneticists themselves have been less convinced. Tokeshi (1999) argued that the genetic founder effect does not seem to be an effective means of speciation and Nei (2002) cited 'one of the most important findings in evolutionary biology in recent years: that speciation by the founder principle may not be very common after all'. Orr (2005) wrote that despite the early popularity of the idea, 'it is difficult to point to unambiguous evidence for founder effect speciation, and the idea has grown controversial'. The experiments of Moya et al. (1995) failed to corroborate predictions of founder effect speciation and subsequent studies have also found no evidence for it (Rundle et al., 1998, Mooers et al., 1999, McKinnon and Rundle, 2002, Rundle, 2003). Crow (2008) called the idea of genetic revolution 'vague and misguided' (see also Crow, 2009). Even in birds, founder effects 'may be unnecessary' (Grant, 2001, cf. Walsh et al., 2005). The passerine Zosterops is often cited as the classic case of a taxon that has evolved by founder speciation (Mayr and Diamond, 2001), yet a detailed study of clades in the south-west Pacific concluded that the focus on founder effects in this group 'has been overemphasized' (Clegg et al., 2002).
Florin (2001) described how 'The vicariance model of allopatric speciation has been repeatedly confirmed empirically, while peripatric [founder effect] speciation has suffered severe criticism for being both implausible and empirically unsupported'. In her own studies on flies she found 'no support for speciation through founder effects'. In recent years the debate has heated up and advocates of dispersal theory have found it necessary to publish an article stressing 'The reality and importance of founder speciation in evolution' (Templeton, 2008). This was a reply to Coyne and Orr's (2004: 401) conclusion that 'there is little evidence for founder effect speciation'. Coyne (1994) wrote that the idea that the theory 'has infected evolutionary biology with a plague of problematic work'.
Templeton (1998) had earlier introduced a new method of analysis - nested clade analysis - that incorporated founder effect speciation (as 'range expansion') and promised to distinguish between the results of this process and those due to the fragmentation of populations, or vicariance. Apart from its problematic use of founder effect speciation (and confusion with range expansion) this method has been criticized on other grounds (Knowles and Maddison, 2002, Beaumont and Panchal, 2008, Petit, 2008). Neigel (2002) wrote that in nested clade analysis 'it is predicted that gene genealogies will exhibit a "star phylogeny"... in a population that has been expanding in size... These descriptive patterns are useful as heuristics and help us understand how we can use population genetic data to reconstruct population histories (Templeton 1998). However it is not clear that they provide reliable evidence of particular kinds of population histories. Often the same pattern may be created by any of several different population histories. For example, a star phylogeny could also result if a once continuous population became fragmented...'.
In another critique of nested clade analysis, Knowles (2008) stressed the simulation studies that have found consistently high error rates using the method. Knowles noted that the method had been cited 1700 times and questioned why it is so popular despite having such obvious problems. Nested clade analysis is a logical development of Templeton's earlier work supporting founder effect speciation and dispersal biogeography. As a formalization of earlier work by Mayr (1954) and Carson and Templeton (1984), nested clade analysis incorporates the core concepts of orthodox dispersal theory and the modern synthesis, and so it may remain popular largely for historical reasons.
Mayr's critique of ecological speciation
Although geneticists such as Coyne and Orr (2004) dispute speciation by the genetic founder effect this does not mean that they reject center of origin/dispersal and founder events. Nevertheless, founder effect speciation was one of the pillars of modern dispersal theory and rejecting it has important consequences. In particular, if it is rejected but dispersal itself is retained, other modes of differentiation must be invoked and ecological speciation is often inferred. Yet although many sister species have different ecologies, ecological differentiation seems unlikely to be the initial cause of many variants. Mayr (1954: 168) recognized this and discussed the issue when he proposed his ideas on 'genetic revolution'.
Mayr had an extensive field knowledge of New Guinea and he developed his founder theory to account for a striking fact. This was the marked morphological difference between bird taxa on the New Guinea mainland and their representatives on small, offshore islands close to the mainland (Mayr, 1942, 1954, 1992). Biological and physical differences between the two nearby environments, both with lowland rainforest, seemed relatively minor (Mayr 1954: 158, 168). Of course, any two areas show ecological differences, but in this case the environmental variation seemed much less than that within the vast areas of rainforest through mainland New Guinea, where there was often less differentiation in the sister group. As Mayr later emphasized: 'The crucial process in speciation is not selection, which is always present, even when there is no speciation, but isolation' (Mayr, 1999: xxv). In the case of the islands, any appeal to ecology seemed ad hoc and with no supporting evidence; instead, the obvious factor seemed to be the 'water barrier' isolating the island and mainland forms. Although Mayr's ideas on the particular genetic mechanism were probably incorrect, as geneticists now suggest, they were attempts to explain a genuine, fundamental problem in biogeography, ecology and evolution, exemplified by the New Guinea birds.
If chance dispersal with genetic revolution and founder effect speciation is untenable and ecological speciation is unlikely, what are the alternatives? The answer may involve isolation, as Mayr thought, but mediated by tectonics, not dispersal (Heads, 2001, 2002). Authors have recognized that tectonic geology is relevant for evolution and biogeography but its importance has been underestimated. In addition, the knowledge of tectonics in places such as New Guinea has increased greatly since the 1930s when Mayr was formulating his ideas. Many tectonic structures and processes have only been recognized recently by geologists and some of these are geographically coincident with biogeographic features.
Founder dispersal and new ideas on rift tectonics
New developments in geology indicate that founder dispersal may not be required and this adds to the problems with founder effect speciation raised by geneticists. The northern part of New Guinea and its offshore islands are one of the most mobile parts of the Earth's crust and is now interpreted as a plate margin or series of margins. It is traversed by major transform faults and strike-slip movement on these has transported terranes, fault slivers, and biota in different directions over hundreds of kilometers. This has produced whole series of centers of endemism, along with dramatic disjunctions and juxtapositions (Heads, 2001, 2002). The latter include the 'strange' and 'intriguing' distributions reported in Murinae by Musser et al. (2008). Following Mayr's New Guinea surveys, another important study of New Guinea birds involved a transect across the main tectonic boundary in New Guinea, the craton margin (Diamond, 1972). As in Mayr's studies, Diamond did not refer to structural geology and instead attributed the many faunistic and
phylogenetic differences found along the transect to ecological factors. Some of the geology was only clarified more recently, after these two biologists were writing. Nevertheless, a later, book-length collaboration explaining evolution and biogeography in the region (Mayr and Diamond, 2001) continued to rely on founder dispersal over current or recent geography. Instead, the distributions can be explained by tectonics.
The critique of founder dispersal in studies of biogeography
Biogeographers have known for years that similar distributional phenomena occur in different, unrelated taxa, each with different means of dispersal and ecology. Because of this, and because the shared patterns are intricate and precise, many biogeographers have rejected a concept of long distance or founder dispersal that relies on chance. Instead they have developed concepts of the orderly evolution of entire communities. The global vicariance biogeography produced by 19th century systematists such as the zoologist T.H. Huxley and the botanist J.D. Hooker was rejected by the dispersalists of the twentieth century as 'land-bridge building'. But in retrospect these early, global analyses of the main groups were often more or less valid. By the late 19th century museum had extensive collections from around the world and the main geographic patterns were well understood. Many biogeographers of the time analyzed the groups without reliance on centers of origin or dispersal and it was this research tradition, not that of dispersalism, that led to Wegener's (1912, 1915) prescient synthesis of geology and biogeography. Pure 'chance dispersalism' denies any true patterns in distribution other than those related to ecology and means of dispersal. Yet two botanists working on the diverse flora of Indonesia and Malaysia (Malesia) concluded in a similar way: van Steenis (1936) wrote: 'On the whole I cannot trace any relation between distribution and what is known of [means of] dispersal' and Kalkman (1979) suggested: 'The dispersal method [of dispersal] is of subordinate importance for areogenesis as compared with other factors'.
Vicariance biogeography is sometimes portrayed as a new, iconoclastic view that only developed in the 1970s after the rise of plate tectonics. This is not correct and the idea that distribution must be related to means of dispersal has often been questioned. After studying the species of Gentianaceae and their distributions, T.H. Huxley (1887) wrote that 'one conclusion appears to me to be very clear; and that is, that they are not to be accounted for by migration from any "centre of diffusion"...'. Mueller (1892) discussed biogeographic affinities between Australia and Africa and wrote that 'the enigma cannot be explained by migration (p. 432)... Floras of certain districts are not homogeneous, but exhibit amongst their warp a woof which has nothing in common with it, but exhibits the stamp of a totally different flora... I do not deny [migrations] when they are opportune, and I know very well that wind and weather, animals and men, are able to distribute species sometimes over large areas; but it is quite a different thing when we have to deal with the spreading of whole floras, sufficient to impress one district with the stamp of another, where all the species are united in an organic association, so that one cannot be understood without the other. This cannot ever have been accomplished by a migration of a mechanical nature' (p. 433).
Molecular studies and ecological speciation
It is sometimes observed that while Darwin's (1859) book is titled 'On the origin of species by means of natural selection', it does not actually cite any examples of this. Does natural selection really lead to speciation? Darwin himself stressed that regions with a similar environment have very different biotas, indicating that ecology does not explain the difference. Later authors such as Mayr (1942) and Croizat (1964) have also accepted geographic factors as the main cause of speciation, with ecology as a secondary factor.
Many clades comprise species that are allopatric but show little or no obvious differences in their ecology. These are often interpreted as non-adaptive 'radiations', rather than adaptive radiations caused by ecological differentiation. Examples include the plant Nigella in the Aegean Archipelago (Comes et al., 2008) and landsnails in the Azores (Jordaens et al., 2009). Rundell and Price (2009) suggested that sympatric species that are ecologically differentiated could arise if speciation occurs through geographical isolation and non-adaptive radiation, and is followed by ecological differentiation and range expansion into sympatry.
The environments of any two allopatric sister groups always differ in some variable or other, but this does not necessarily explain the differentiation. Simple ecological speciation would lead to species with sisters that are sympatric, with more or less exactly the same geographic range, and different ecology, but this unusual. If allopatric evolution is the usual mode of speciation there should be a positive relationship between the relative divergence time of taxa and their degree of geographic range overlap. Barraclough and Vogler (2000) studied a range of animal groups and found 'a general pattern of increasing sympatry with relative node age... consistent with a predominantly allopatric mode of speciation'. In a similar study, Kamilar et al. (2009) examined biogeography and phylogeny in 19 species of cercopithecid monkeys in Africa and also found a positive relationship between age and overlap. This is good evidence for allopatric speciation being the usual mode of diversification in cercopithecids and there is no indication that this group has unusual biogeography.
The best evidence for ecological speciation (to eliminate the possibility of geographical factors) is provided by sympatric, ecologically separated sister species. Examples include the reef fishes Hexagrammos agrammus and H. otakii in the Sea of Japan, with the former species in seaweed beds, the latter at rocky sites (Crow et al., 2010). In practice, true sympatry of sister species, not just secondary overlap at some point of their range, is uncommon. Usually there is a significant degree of geographic allopatry between the two and this probably reflects the original, allopatric break (as shown in many examples in this book). Unfortunately, studies of ecological speciation (e.g. Rundle and Nosil, 2005, Funk and Nosil, 2008, Schluter 2009) often lack geographic analysis or even distribution maps, so it is difficult to assess the geographic component in the speciation. Bolnick and Fitzpatrick (2007) observed that 'Biogeographic comparative studies of range overlaps have not been conducted for the two groups most widely thought to exhibit sympatric speciation, phytophagous insects and lacustrine fishes'.
Studies of particular clades that have included biogeographic information show that supposed cases of sympatric and ecological speciation have little support. For example, the different races of the apple maggot fly Rhagoletis pomonella (Tephritidae) on hawthorn and apple trees are often cited as evidence for sympatric speciation. Sequence data instead suggested a geographic scenario with allopatric differentiation followed by overlap (Feder et al. 2003).
The threespined stickleback Gasterosteus aculeatus (Gasterosteidae) is widespread in the north temperate zone. In a very small part of this range, near Vancouver Island, six lakes on three separate islands contain sympatric 'species' of G. aculeatus that occupy benthic habitat near the lake floor and limnetic habitat higher in the water column. The two forms are morphologically and reproductively isolated from each other. (They are referred to as species by geneticists studying ecological speciation, e.g. Taylor et al., 2006, but not by taxonomists). The pattern has been interpreted as the result of sympatric, ecological speciation in the six lakes and this model predicts that the pair in each lake should form a monophyletic clade. In fact, though, molecular studies found that this is not the case (Taylor and McPhail, 2000). The sympatric pairs of sticklebacks are only known from a very small part of the total species range, the Strait of Georgia, suggesting that the history of the region rather than ecological speciation has promoted 'species pair' evolution. The area has been submerged by the sea twice, at 12000 BP and 1500 BP, and Taylor and McPhail (2000) concluded that the presence of two forms in each lake was due to invasion of the coastal lakes by different populations of marine sticklebacks on two separate occasions, rather than ecological speciation (cf. Rundle and Schluter, 2004).
Yeung et al. (2011) studied the population genetics of the royal spoonbill Platalea regia in Australasia and its allopatric sister group, the black-faced spoonbill P. minor in eastern Asia. Their results 'do not support founder effect speciation in Pl. regia'. The authors suggested that the divergence between the two species was probably driven by selection. Gay et al. (2009) found high levels of divergence between species of gulls (Larus) and concluded: 'Such divergence is unlikely to have arisen randomly and is therefore attributed to spatially varying selection'. Yet whether or not all evolution is due to selection is controversial and there is great confusion about the role of natural selection. Wilson (2009: vii) proposed that 'All elements and processes defining living organisms have been generated by evolution through natural selection', but in the same book, Ruse and Travis (2009: x) wrote that: 'natural selection is not the only evolutionary force...'.
Although ecology and natural selection explain why there are no alpine plants with large, membranous leaves, it does not explain why the particular families, genera and species that are found on a mountain occur there in the first place. Natural selection might explain why the black form of the peppered moth Biston bistularia became more common than the mottled form during the industrial revolution, due to increased pollution. But it does not explain why there were black forms there to start with, before the industrial revolution. To summarize: 'Natural selection does not design an organism or its features; it merely filters existing variation' (Travis and Resnick, 2009: 114).
Island biogeography with metapopulations and without founders
Many taxa occur on 'ecological islands', small areas of suitable habitat found scattered through large regions. 'Ecological islands' are often more or less ephemeral and include puddles, landslides or areas of fresh lava. Among the classic examples of ecological islands are young, volcanic islands. Here the groups in the rainforest and on the reef exist as metapopulations, populations of populations, constantly dispersing from older islands to younger ones within a region as new islands appear and older ones subside. An island clade endemic to the central Pacific and widespread there can survive more or less indefinitely, in the same region, on islands that are individually ephemeral.
Many botanists have denied that there is a true, regional Polynesian flora, perhaps because dispersal theory suggests one should not exist. Instead, Philipson (1970) argued that plants endemic and widespread in Polynesia, such as Meryta (Araliaceae), Tetraplasandra (Araliaceae), Fitchia (Asteraceae), Sclerotheca (Campanulaceae), and many others, indicate that 'the southern Pacific islands must be credited with a flora specific to this region... Clearly land has been present for long periods in this area of the Pacific because well-marked genera are endemic to it. The flora characteristic of this region could survive provided a few oceanic volcanoes projected above the sea at all times. Such oceanic islands characteristically rise and fall relative to sea-level so that they are precarious footholds for a flora, but collectively they form a secure base.' In this view, new individual islands will be colonized by ordinary movement, on the scale seen every day. This is an observable ecological phenomenon (unlike long-distance dispersal/founder effect speciation) that functions using ordinary means of survival.
Founder theory, island biogeography and new ideas on volcanism
Founder theory and the associated 'equilibrium' theory of island biogeography both assume that volcanic islands appear at random with respect to other islands. But volcanism and volcanic islands tend to recur around the same tectonic features (subduction zones, propagating fissures, hot-spots, etc.) where terrestrial taxa survive and evolve more or less in situ as metapopulations. New ideas on volcanism in the Pacific, for example, emphasize the importance of the large, igneous plateaus emplaced in the Cretaceous. These are mainly submarine but some intercalated sedimentary strata include fossil wood, and the plateaus have many large, flat-topped seamounts that were once high islands. The simple hotspot model is not sufficient to explain many aspects of intraplate volcanism and in contrast with the traditional mantle-plume hotspots, new tectonic models explain linear volcanic chains by propagating fissures. These could be caused by stress fields induced by normal plate tectonics (Foulger and Jurdy, 2007a).
Ecology and biogeography
Darwin (1859: 346) observed that 'In considering the distribution of organic beings over the face of the globe, the first great fact which strikes us is, that neither the similarity nor the dissimilarity of the inhabitants of various regions can be accounted for by their climatal and other physical conditions'. Few taxa occur throughout all areas on Earth with similar ecology - most are restricted to particular geographic regions. Areas of rainforest, Mediterranean climate, desert, etc. in different parts of the world have similar climate and vegetation but almost completely different floras and faunas. This indicates the primary importance of historical factors rather than ecological ones in determining global distribution. Of course, within a biogeographic region, distribution may reflect ecological factors.
For example, within an area of endemism in tropical hill forest ten or a hundred kilometers across, the taxa are sorted out into the standard ecological zones and habitats. Some will always occur on the valley floors, others on the ridge crests and summits. Outside this biogeographic region, the species pool changes. Thus biogeography involves differentiation in space at a large scale, ecology involves smaller scale spatial differentiation, along ecological gradients within a biogeographic region. Some ecological phenomena are large scale, though, and the three latitudinal belts, northern, tropical, and southern, are among the most obvious phenomena in biogeography. Ecology can explain the differentiation among these, and the smaller-scale spatial differentiation in communities within a biogeographic region, but it cannot explain many large scale patterns and these are usually attributed to historical factors. Examples of important phylogenetic/geographic breaks where there are no obvious climatic or topographic breaks include Wallace's line and many others that are less well-known. A group's ecology does not usually explain its biogeography, although it explains much of its spatial distribution at a small scale - across meters rather than kilometers - within its biogeographic region.
The elevation of a community determines many aspects of its ecology and can change rapidly in geological or even ecological time. As discussed above, ducks and parrots in the Andes may have been uplifted tectonically during orogeny. Salicornia (Amaranthaceae) and Frankeniaceae are usually maritime shrubs, but occur at over 4200 m elevation in the Andes (Ruthsatz, 1978). Rapid uplift in New Guinea may have carried populations of coastal mangrove associates such as Pandanus (Pandanaceae) to treeline at ~3000 m in just one million years. In a similar way, populations of plants and animals may be lowered with erosion or subsidence, or stranded in the middle of a continent following marine transgression and regression. Thus the current ecology of a group may have little to do with its original one and a population can find itself in a new type of habitat without having moved from its initial substrate. There may be a long lag phase while the groups readjust to the new conditions, but in areas of active tectonism equilibrium may not be reached and ecological anomalies may persist. At this scale of time and space there is little distinction between ecology and biogeography; factors determining the elevation of a group (uplift, subsidence, erosion) may also result in biogeographic differentiation in latitude and longitude.
The CODA paradigm and pan-adaptationism
Modern interpretations of distribution are based on a center of origin and dispersal from there, and also involve the idea of adaptation. A clade moves out from its center of origin by dispersal to a new locality and habitat. Here it faces new extrinsic needs and it changes its morphology and physiology in response to these, bringing about adaptation. Lomolino and Brown (2009) referred to this as the CODA model, as it is based on center of origin, dispersal and adaptation, and this is an accurate characterization. There is no denying the intricate fit that exists between most organisms and their environment, and is necessary for their survival, but instead of structure being determined by function and extrinsic needs, structure may determine function. In this view the morphology of the teeth determine many aspects of the diet, the diet does not determine the teeth. If the teeth evolve, the diet changes.
It is usually taken for granted that adaptation is the main factor in structural evolution, although this has been questioned. Scheiner (1999: 145) wrote that 'By the end of the Modern Synthesis much of ecology and related disciplines were in the grip of... pan-adaptationism. Most traits were seen as adaptations, and organisms were viewed as being nearly perfectly adapted to their environment. This viewpoint came about through taking the theory of natural selection and attempting to fit the entire world under its banner... Many advocates of the modern synthesis were guilty of pan-adaptationism... It tended to be ecologists, rather than evolutionary biologists, who carried this banner... Natural selection was the focus of many studies so the investigated traits, not surprisingly, were shown to be adaptations... Since then we have spent much more effort on the other processes that contribute to evolution such as drift, mutation, and developmental constraints.' Scheiner (1999) did not accept natural selection as a primary driver in evolution and it is suggested below that in many groups the biogeographic history, not selection, has determined the current ecology. For example, a group may be in an area of highlands, or desert, or on an atoll because its ancestors were already in the region before the uplift, the desertification, or the subsidence and erosion of a high island. If the group already had structures and physiology suitable for the new environment (pre-adaptations) it survived, if not it went extinct; adaptation itself may have relatively minor significance.
Metapopulations and weeds
All clades go extinct in the end, but usually only after many millions of years and during that time they survive while their environment changes around them. Weeds are plants that can survive in disturbed and marginal areas, but all plants and animals are, or have been, 'weedy' to some extent. If not, they would never have survived the climatic and topographic revolutions of geological time, such as volcanism, marine transgression, orogeny, and glaciation. Populations of 'fire weeds' can survive burning in situ, while montane weeds survive uplift and thrive in highly disturbed areas around glaciers. Both of these occur around the around the Pacific margin where acitve plate margins have led to orogeny and volcanism. The Pacific 'ring of fire' maintains its own diverse and endemic biota, and this is diverse in areas such as Indonesia and the Andes. Fire-weeds found in and around the craters, fumaroles, and ash slopes include members of Ericaceae and the plants survive in these dynamic, disturbed regions as metapopulations (Heads, 2003). A well-studied 'volcano weed' in Ericaceae is Vaccinium membranaceum (Ericaceae), an animal-dispersed shrub of western North America. Yang et al. (2008) examined the genetics of a recently founded population on new volcanic deposits at Mount St Helens, Washington, 24years after the 1980 eruption. They found that 'While founders were derived from many sources, about half originated from a small number of plants that survived the 1980 eruption in pockets of remnant soil embedded within primary successional areas. We found no evidence of a strong founder effect in the new population; indeed genetic diversity in the newly founded population tended to be higher than in some of the source regions'. (p. 731).
Major geological events include phases of extensive, very large-scale volcanism, as in the central Pacific (during the Cretaceous), and also asteroid impacts, as at Chicxulub, Yucatán (Cretaceous/Paleocene boundary) and at Chesapeake Bay, Virginia (Late Eocene). These types of events will cause local, regional or even global extinction of some clades. Yet they will also provide opportunities for range expansion in many taxa, especially those that are tolerant of extreme disturbance and marginal ecology. In many cases the local endemics of to-day, even in the least disturbed, oldest forests, may be derived directly from Cretaceous weeds. Following the original range expansion - determined by geology, not chance dispersal - they have settled into a phase of immobilism through the Cenozoic.
The return of dispersalism
Throughout history it has been assumed that because animals and plants move, their distributions have been caused by this movement. The center of origin has been a recurrent motif in philosophy and literature for the last two or three thousand years, at least. In this model each kind of plant and animal originated at one point and attained its distribution by spreading out from there. This is seen in literal readings of biblical accounts, such as the stories of Eden and Ararat, and in many modern evolutionary studies. The only times that the center of origin assumption was widely questioned were in the late 19th century until the first world war, a period when systematic activity peaked, and briefly in the 1970s - 80s. The current popularity of dispersalism in molecular phylogenetics is sometimes portrayed as an exciting, new development, but in reality it is the orthodox view of the last two millennia.
The role of geographic data in evolutionary biology
Wallis and Trewick (2001) suggested that plate tectonics, Hennigian systematics and molecular clock dating 'have placed biogeographic data at center stage of evolutionary debate'. But biogeographic data have been central to evolutionary biology ever since the origin of the subject. In an 1845 letter, Darwin referred to geographic distribution as 'that almost keystone of the laws of creation' (Darwin, 1887) and he often used it to frame and test evolutionary hypotheses.
Trewick and Wallis (2001: 2178) suggested that 'The tendency to identify and focus upon repeated pattern is an important feature of biological research, but one that has usurped the role of hypothesis testing among some biogeographers'. There is more to science than testing hypotheses though. After all, medieval scholastics rigorously tested their hypotheses, by examining scripture, Aristotle and miracles. The main interest of molecular phylogenies for biogeography is not that they suggest hypotheses, but that they illustrate geographic patterns so clearly.
Authors such as McDowall (2007) have criticised panbiogeography as 'a search for general patterns'. The criticism proposes that individual taxa have their own evolutionary and biogeographic history and that the correct topic of study is a particular taxon. But many critics of panbiogeography are taxonomic specialists on particular groups, not comparative biogeographers. Many have not investigated general biogeographic patterns, described new ones, and so on, but have devoted their time to studying distributions of particular taxa in their group. Comparative biogeography cannot be concerned with one-off phenomena, such as unusual distribution patterns seen only once in a single group. Instead it focuses on general patterns and laws in distribution and explains particular cases in terms of these. Often what are assumed by specialists to be rare or unique distributions are quite common in other groups. One main conclusion of panbiogeography is that while the details of distribution patterns are virtually infinite and no distribution pattern is exactly like that of any other group, the main phylogenetic and biogeographic breaks or nodes are few. Nothing important in biology only happens once and multitudinous aspects of differentiation occur in many groups of animals and plants at the same standard nodes.
Deduction and induction in biogeography
Cain (1943: 151) concluded his critique of the center of origin idea by writing that geobotany and geozoology 'carry a heavy burden of hypothesis and assumptions which has resulted from an over-employment of deductive reasoning'. In other words, a theory such as the center of origin model has been taken as fact and used, along with other ideas ingeniously deduced from it, to explain certain observations. Cain argued that neglected to test the initial assumption: 'What is most needed in these fields is a complete return to inductive reasoning with assumptions reduced to a minimum... In many instances the assumptions rising from deductive reasoning have so thoroughly permeated the science of geography and have so long been part of its warp and woof that students of the field can only with difficulty distinguish fact from fiction'. Modern authors have agreed: 'evolutionary biology is an inductive science, one in which generalities emerge not as the result of theoretical deduction or the conduct of critical experiments, but rather through the summation of many evolutionary case studies' (Losos, 2010: 623; cf. Gavrilets and Losos, 2009). This suggests that induction is at least as important as deduction and the main section of this book (following Chapter 2) is devoted to particular case studies.
Morphological work and molecular phylogenies
The case studies discussed below are from molecular accounts. Why concentrate on these when so many morphological phylogenies are available? The main reason is the inherent biogeographic interest of the molecular clades, as they show so much orderly, complex geographic structure. For a biogeographer, molecular systematics is confirmed as a productive research program by the close correlation between phylogenetic groupings and geography. The geographic patterns shown by the molecular clades are not necessarily new. Many of the most interesting ones have already been documented in traditional taxonomic revisions, although the authors of the new studies are often unaware of the earlier work. Nearly all the maps in this book depict molecular phylogenetic arrangements of groups but the details of the geographic distributions are often based on morphological observations and collections. Working out these concordances supplements the molecular phylogeny with the more extensive samples of traditional work.
While most morphological classifications can be used for biogeographic analysis, there are problems with morphological work. Taxonomists have sometimes cherry-picked certain combinations of characters and left out others in order to achieve a desired end. One author can 'demolish' the taxonomy of another simply by selecting an appropriate combination of characters and in this way morphological classifications can be manipulated to fit preconceived notions of evolution and biogeography. Molecular work is not immune to this, but it seems to be less of a problem.
In addition to their impressive biogeography, molecular phylogenies tend to make sense morphologically, although not always with respect to the traditional homologies and categories. Structural concepts such as the 'flower' are often based on 18th century notions that require revision. Molecular phylogeny now places a raspberry (Rosaceae) and a mulberry (Moraceae) together in a novel grouping, a radically changed Rosales (Stevens, 2010). At first sight the fruit look very similar. But using traditional homologies, the first grows from a single flower, the second from an inflorescence or group of flowers, and the 'flowers' in both represent the opposite poles of floral morphology - they are as different as possible. Taking the hint from the molecular phylogeny, this traditional homology may be incorrect and the flower in one may be equivalent to the inflorescence in the other. In other words it is not only the morphological classification that is incorrect, but also the morphological concepts underlying it. (Flower-inflorescence intergrades also occur in plants such as Euphorbia; Prenner and Rudall, 2007). Thus molecular work may stimulate re-examination of the morphological categories and homologies. It is not simply a question of finding a new, cryptic character of morphology that corresponds to the new molecular grouping.
The main problems with the molecular work include sampling limitations, both phylogenetic and spatial. This is improving rapidly due to technical advances and as workers come to recognize the great value of extensive sampling. Another problem is not so much a technical limitation as a conceptual one: modern molecular work often interprets its excellent data using old, inefficient biogeographic methods, such as divining the center of origin, dating the phylogeny with fossils, and explaining common patterns by extraordinary events of chance dispersal. Molecular data provide a fabulous new source of information on biogeographic differentiation and evolution, but satisfying interpretations of the patterns have lagged far behind their description. Donoghue and Moore (2003) advocated the standard methodology now employed in hundreds of papers: use DIVA or a similar program to locate the center of origin and use fossil-based calibrations (minimum ages) to give maximum ages for clades (see next chapter). It is not surprising that the end result duplicates the biogeography of a century ago (Matthew, 1915), as both share the same principles of center of origin, fossil-based chronology, and Cenozoic evolution of the orders by long distance dispersal. Instead, all these concepts can be abandoned as the molecular phylogenies and modern tectonics suggest more likely, alternative reconstructions of groups' histories.
Although molecular and morphological phylogenies may differ in individual groups, many of the broader biogeographic patterns indicated by molecular phylogeny were known to the early workers. As Avise (2007) concluded, 'traditional non-molecular systematists generally seem to have done an excellent job in identifying and classifying salient historical discontinuities in the biological world'. Avise emphasized that intraspecific molecular groups are nearly always allopatric and that their geographic distributions 'usually make biogeographic sense', as they 'orient well' with known patterns. These comments also apply to many molecular clades above species level. Authors sometimes describe the geographic distribution of a molecular clade as idiosyncratic or peculiar when in fact it is a standard pattern and this provides good evidence for both the grouping and the distribution pattern. Many molecular clades conform to specific distribution patterns that have been of interest to comparative biogeographers for decades or longer.
To summarize, one of the main themes of molecular studies has been the precise geographic structure evident in molecular clades. Evolution can be interpreted using the methods of molecular biology and panbiogeography, as Croizat (1977) suggested, and the most useful studies on groups include detailed data on both phylogeny and comparative distribution.
In a remarkable phase of the molecular revolution, from about 1990-2000, biologists worked out the higher-level molecular phylogenies of the large, world-wide groups such as birds and angiosperms. In a second phase, researchers have produced detailed accounts of lower level groups that involve regional and local differentiation. The new molecular work shows that clades relate closely to geography and so researchers have started to annotate their phylogenies with geographic information and to map clades. The molecular work shows classic biogeographic patterns with improved clarity, revealing many details that were previously obscure or unknown, and the patterns are the subject of this book.
This chapter incorporates material previously published in the Biological Journal of the Linnean Society (Heads, 2009b), reprinted here with permission from John Wiley and Sons.