12. Speciation: con't

(iv) Modes of speciation: the role of geography

(a) Allopatric speciation.

The evolution of reproductive isolation (or evolutionary independence) owing to geographic isolation is what defines allopatric (aka "geographic") speciation. Allopatric speciation has long been recognized as a principle mechanism of speciation, but is most forcefully championed in the writings of Ernst Mayr (e.g. Mayr 1963).

Allopatric speciation may proceed through random genetic change in isolated populations (indirectly through genetic drift) or it may proceed via genetic changes as a result of selection in contrasting environments (drift and selection may also interact to promote speciation in allopatry). Details aside, the key element is the accumulation of genetic changes as a result of physical isolation (i.e. unconstrained by gene flow).

There is some experimental data supporting the role of geographic isolation for simulated systems (e.g. Drosophila) where isolation (and sometimes selection) is imposed across generations and reproductive isolation is tested subsequently (see review by Rice and Hostert 1993), but the vast majority of evidence for allopatric speciation comes from basic biogeographical observations.

Part of this evidence stems from the vast amount of geographic variation that biologists observe in a species across its geographic range in morphological, behavioural, biochemical/molecular, life history, and physiological traits. The examples below are just two. The lower example in the "drongo" is the kind of geographic variation that so impressed Mayr and fuelled, in large part, in arguments in favour of the importance of allopatric divergence to speciation.

The importance of geographic isolation in driving speciation is often inferred from observing reproductive isolation between forms that are now in contact, but which are thought to have diverged in allopatry. For instance, the finches of the Galapagos Islands shown below are thought to have diversified after a single colonist dispersed to one island, subsequently colonized the other islands in the archipelago. Many of the islands represent contrasting environments particularly with respect to seed sizes. These differing selective environments appear to have been responsible for divergence in bill morphology (and associated morphological and behavioural traits). Such changes may have driven (indirectly) sufficient genetic changes that have resulted in reproductive isolation when two or more species come into contact. For instance, some islands have as many as four species that are sympatric and do not interbreed. Other islands have sympatric species with some evidence of hybridization and introgression (see Grant and Grant 1998).

A further example is the case of "ring species"; divergent forms that exist and intergrade along a chain encircling a barrier and where the terminal forms overlap, but do not interbreed. Ring species have been cited as an example of where geographic variation in space can be used to infer variation in times of divergence (i.e., geographic isolation promotes divergence along the path of the "ring" and as a result the forms at the opposite ends of the ring are sufficiently divergent that they do not interbreed where they meet). The distribution of subspecies of Hoplitis (a bee) is a classic example (see Mayr 1963).

A recent example of possible ring species was published by Irwin et al. (2001). They studied six subspecies of green warbler, Phylloscopus, that encircle the treeless Tibetan Plateau (see below). Two of the taxa co-exist without interbreeding in central Siberia (the cross-hatched region of red-blue samples), but are interconnected by a series of intergrading populations that encircle the Tibetan Plateau along the southeast and southwest.

This figure shows the distribution of the six taxa around the Tibetan Plateau and representative "sonagrams" from eight locations. The authors used such data, along with molecular, plumage, and behavioural (response to sonagram playback experiments) data to show that the two taxa (P. trochiloides trocholoides (red shading) and P. t. viridanus (blue shading)) that co-exist in central Siberia are reproductively isolated while taxa further south in the ring were not. The authors suggested that the two isolated taxa came into contact following dispersal northward around the western and eastern boarders of the plateau and that the resultant speciation was a consequence of isolation of populations around the barrier.

Finally, consider the example of sympatric "species pairs" of lake whitefish (Coregonus clupeaformis) in eastern Canada and northeastern USA. The figure below shows the distribution of major phylogroups of whitefish in North America. One group (encircled in the solid line) is very widespread and is thought to stem from isolation and postglacial colonization from the Mississippi Refugium. A second major group (encircled in the dashed line) is thought to have been derived from isolation and postglacial colonization from an "Acadian Refugium" and is much more restricted in distribution.

This figure above shows the distribution of these two (plus another) phylogroups between so-called dwarf and normal lake whitefish that co-exist in many lakes in northeastern North America. These kinds of whitefish are considered by many to be valid biological species as they are ecologically and genetically distinct in sympatry indicating that they are to a large degree reproductively isolated in lakes where they co-exist.

Note the distribution of phylogroups between dwarf and normal whitefish in Cliff Lake in northern Maine. The two "species" are fixed for the different phylogroups (dwarfs are all part of the Atlantic lineage (a subgroup of the Mississippian lineage) while normal whitefish are all fixed for the Acadian lineage. Because these two phylogroups are so strongly patterned geographically (see previous figure) it is reasonable to argue that the isolation in distinct refugia (allopatry) has driven the divergence between dwarf and normal whitefish and they have colonized the lake independently resulting in two biological species of whitefish in one lake – allopatric speciation.

Note, however, that other lakes (e.g. Webster Lake, Como Lake) with dwarf and normal whitefish are not so cleanly partitioned along phylogenetic lines (the dwarfs and normals may have arisen within a single lineage or they may have exchanged genes to a greater degree in some of these other lakes). Nevertheless, the pattern in Cliff Lake especially is a good example of the role of biogeographic pattern that allows some inference about the process of speciation.

In summary, allopatric speciation is widely accepted as a major mode of speciation given the kinds of evidence discussed above and its theoretical appeal. The continuing issues are related to: (i) are other (non-geographic) modes of speciation possible, (ii) at what rate does geographic speciation proceed, and (iii) what specific processes are involved (selection, drift, founder events, or some combination?)

(b) Parapatric speciation

Parapatric speciation involves the evolution of species from populations that are in contact at their range margins (i.e. they are contiguous). The populations are largely geographically separate, but remain in contact at their edges.

Parapatric speciation is suggested by the observation of clines; a geographic gradient in some measureable character (morphology and molecular or biochemical traits most commonly). The clines are thought to stem from diversifying selection along an environmental gradient ("ecotones") where one phenotype is favoured one one side of the cline, intermediate (hybrid) phenotypes are favoured (or less selected against) through the transition zone, and the alternative phenotype is favoured on the other side of the cline. Hybrid zones (areas where genetically divergent populations are in contact and hybridize to produce offspring of mixed ancestry) are also possible signs of parapatric speciation. Hybrid zones are common. Below is a map of Europe showing the origin of a major hybrid zone in the Pyrenees Mountains between Spain and France.

Another example involves contact between taxa in northwestern North America. Several forms come into contact along the Coastal-Cascade mountain crest. For example, Dolly Varden and bull trout (Salvelinus malma and S. confluentus) are two species of char that hybridize in several watersheds along this geographic contact zone (see Taylor et al. 2001).

Hybrid zones could result, however, from secondary intergradation (or "secondary contact") between allopatrically-derived forms (e.g. whitefish example above, Dolly Varden and bull trout, Chorthippus grasshoppers in Europe) or from primary intergradation from selection along an environmental gradient (i.e., without geographic isolation).

As an example, consider the figure below. It shows the distribution of two related toads, Bombina variegata and B. bombina, in central Europe. The two species meet along a NE-SW transect zone (indicated by the thin lines) and areas of known hybridization are shown in thick black lines. In the right panel, note the transition of alleles (black and white shading) as one moves south to north through a single transect point. The central contact zone tends to have intermediate frequencies where the two species meet and hybridize.

The figure below shows the transition of enzyme allele frequencies (top) and morphology (bottom) for B. variegata as samples are taken along two different transition zones (the environmental gradient). Note how the frequency of alleles initially is very low, rises to intermediate levels in the contact zone of hybridization and then to near fixation (1.0) at the other end of the transect. These are classic examples of clines. The cline is thought to be maintained by divergent selection on either side of the contact zone. The width of the contact zone (area of intermediate frequencies) is influenced by the strength of the selective transition. A sharp selective "boundary" would produce a very steep transition zone (a "stepped cline"). See Szymura and Barton (1986) for more details.

As above, such clinal variation could stem from secondary contact of allopatrically derived forms - where they meet is the contact zone and various levels of interbreeding and selection will influence the width of the contact zone. Alternatively, the cline could stem from primary contact and parapatric speciation.

Consider a population on one side of an environmental transition zone. Across this zone selection acts against alleles within this population preventing its colonization of the other habitat Alternatively, perhaps some dispersal barrier prevents the population from colonizing the habitat on the other side of the transition zone. Then some change in the environment facilitates dispersal across the zone or a mutation arises in the parent population that is favoured in the other habitat. These factors promote colonization of the new habitat where the alternative allele is favoured and rises in frequency. Interbreeding is found only through the transition zone (area of intermediate habitat) and divergent seclection eliminates hybrid genotypes in either habitat. As the new population becomes established and adapts to the new habitat and disperses farther from the zone of contact, it becomes increasingly genetically divergent from the parent population. The associated genetic changes may become so pronounced that outside the contact zone, a complete barrier to interbreeding results.

Unfortunately, primary and secondary contact are rather difficult to distinguish as causes of clinal variation and hybrid zones. Some consider that parapatric divergence is unlikely to result in character "break points" that coincide spatially very closely as in the example of Bombina enzymes and morphology above. For instance, how likely is it that the environment would select for particular enzymes AND a particular morphology at exactly the same point? It may be that secondary contact and the mixing divergent populations could produce concordant breaks in selectively neutral characters more easily. Nevertheless, parapatric divergence remains a possible, if difficult to prove, mechanism of speciation.

(c) Sympatric speciation

Finally, we consider sympatric speciation - the evolution of reproductive isolation (or its equivalent) in the complete absence of any geographic isolation. Essentially, it envisions the evolution of two genetically distinct populations from a single randomly mating one with no geographic isolation. Mayr in his writings consistently downplayed the possibility or applicability of sympatric speciation to animal speciation (with the possible exception of things like genome duplication, polyploidy, etc that could "instantly" generate reproductively incompatible populations within the same area).

Models of sympatric speciation typically begin with the existence of a polymorphism within a population that is maintained in the population by disruptive selection. Again, consider a single habitat with two niches, say an island with two trophic niches represented by a bimodal distribution of seed sizes ("big" and "small") or a lake with two feeding niches: a limnetic (plankton specialist) and one benthic (macroinvertebrate specialist).

The lack of an intermediate habitat (to be extreme) means that selection acts against individuals in the population that are intermediate in the morphological or behavioural traits necessary to exploit the different food niches. From this scenario (which is a common observation in many bird and fish communities) its relatively easy to see how two phenotypes would result from disruptive selection without geographic isolation, but how do we get two reproductively isolated populations (species)? To achieve speciation, the feeding phenotypes must mate assortatively (like with like), otherwise random mating would restore a unimodal distribution of phenotypes with each new generation.

Assortative mating could be achieved by the close linkage between loci that control the trophic polymorphism and those that control assortative mating. Say, for instance the allele A at locus 1 encodes limnetic feeding and allele B at locus 2 controls mate preference. Allele A' encodes benthic feeding and allele B' controls the alternative mating preference. Therefore, the combination AB would result in limnetic types always mating with limnetics and the combination A'B' would result in benthic types always mating with benthics.

Unfortunately, recombination will continually act to breakdown any association between loci that might lead to such assortative mating, particularly if the traits are controlled by many loci. This is the major theoretical objection to this common model of sympatric speciation. The debate also centres on the strength of selection that would be required to offset recombination reshuffling any allelic combinations that could drive sympatric speciation, which appears to be prohibitive.

Another more likely way in which assortative mating could be achieved is if mate choice is based on the same traits that are the focus of disruptive selection or on traits that are correlated with those under disruptive selection (i.e. via pleiotropy). A possible example involves size differences that result as a correlated response to adaptation to distinct trophic niches in some fish.

Sockeye salmon and kokanee are the sea-going and freshwater resident forms, respectively of Oncorhynchus nerka. The forms co-exist in many watersheds throughout the North Pacific. Because sockeye go to sea (they must make arduous migrations) whereas kokanee remain in lakes for their entire lives, we suspect there is strong divergent selection for size at maturity in these fish. Sockeye are much bigger than kokanee (see figure below; sockeye top, kokanee bottom. Both are 4 years of age) and experiments have shown that the forms tend to mate assortatively based on size when they occur together in the same streams at the same time or even when they occur alone. Hence, selection for different sizes at maturity appears to have had a correlated response on assortative mating which could provide a mechanism for speciation in sympatry.

Sockeye salmon (top) and kokanee (bottom) collected from a B.C. stream. Both forms are 4 years of age, mature, and spawn in close proximity to one another.

The same mechanism may explain, in part, assortative mating in species pairs of limnetic (bottom) and benthic (top) sticklebacks found in some lakes in southwestern B.C. In this case, mate choice could be based on morphological differences between the species that are the result of divergent selection or on size differences. The selective mechanisms driving changes in body size per se, however, are less clear than in the case of sockeye salmon and kokanee. We do know, however, that size-based mating is also important in sticklebacks.

Benthic (top) and limnetic (bottom) gravid female sticklebacks collected from Paxton Lake on Texada Island, Strait of Georgia, B.C.

What's the geographic evidence for sympatric speciation?

(a) Narrow endemics and geographic distribution:

First, sympatric speciation (or at least the absence of large scale allopatric speciation) is often suggested when the two species have completely overlapping ranges or when one species has a geographic range contained completely within the larger range of the other species. For instance, kokanee tend to be found in the North Pacific within the range of sockeye salmon. Also, benthic and limnetic sticklebacks have been recorded from only six lakes in southwestern B.C. despite the occurrence of G. aculeatus in literally thousands of lakes throughout the Holarctic!

Some of the most compelling evidence for sympatric speciation comes from the distributions of so-called "species flocks" of freshwater fishes. Species flocks are simply a collection of very closely related species that are endemic to a single lake.

Classic examples include the cichlid fishes of the Rift Valley lakes of eastern Africa. In lakes Victoria, Tanganyika, and Malawi there are 247, 163, and 445 endemic species of cichlids, respectively! A few smaller lakes in western Africa have smaller numbers of endemics (4-10). We've seen some of these fish before, but here is an array of species and their general head (feeding) morphologies from Lake Malawi.

Other examples include: the Cyprindon species flock from a small lake on the Yucatan Peninsula and a flock of 50 or so species of cottids (sculpins) in Lake Baikal, eastern Russia.

  A species flock of whitefish in Bear Lake (Utah and Idaho)

The extremely limited geographic distribution of these fish strongly hints that they may have arisen within each lake (contrast with the dwarf and normal whitefish example in the allopatric speciation example). This evidence for sympatric speciation is not ideal, however, because the species with the smaller range could have existed over a broader area in the past, but became extinct there. Also, such "nested" distributions do not eliminate the possibility of microallopatric speciation from having operated (allopatry with the range of the ancestral species followed by dispersal).

(b) Molecular phylogenetic studies

Combined with distributional evidence, molecular phylogenetic studies have been used to test for phylogenetic relationships that would be consistent with different models of speciation.

For instance, the figure below shows phylogenetic patterns (right panel) that would be expected under different geographic models of speciation (left panel). :

(a) allopatric speciation: geographic separation (curved line separation A,B,C from E,E,F) coincident with phylogenetic separation

(b) allopatric speciation: geographic and phylogenetic separation not coincident (what might cause this - dispersal??)

(c) allopatric speciation: peripheral population model

(d) allopatric speciation: founder event

(e) sympatric speciation: the small grey dot within the distribution of f (F) is the distribution of the sister species, f '

Note how some of the phylogenetic patterns are the same (e.g. (d) and (e)). The geographic distribution data is crucial to differentiate between them!

Molecular phylogenies of several of the African lake cichlid communities also result in data that is consistent with sympatric speciation; the species within the lakes are monophyletic strongly suggesting that they arose from a common ancestor within each lake.

A sample of the different species from Lake Malawi. Note the different head forms that are associated with different feeding mechanics and niches of the different species. Could these different head forms form the basis for assortative mating?? The species are also very different in colouration that likely contributes to mate choice.

An example is shown below. Meyer et al. (1990) used mtDNA to reconstruct the phylogeny of representative haplochromine cichlids from Lake Victoria (347 endemic species) and Lake Malawi (445 endemic species) in eastern Africa. The pattern observed (within lake monophyly) is exactly what what would be expected if the species had arisen sympatrically. If the species had arisen by (a very large number!) of allopatric speciation events, one would not expect such a dramatic clustering of species by lake.

Interestingly, Johnson et al. (1994) presented geological evidence that despite the old total age of Lake Victoria (a couple of million years), the lake was most recently completely dry 12,000 years ago! Given the monophyletic pattern observed among species and the fact that they are endemic to the lake, this is a possible example of extraordinarily rapid speciation.

This kind of evidence is often used to argue for sympatric speciation, but some caveats should be noted.

Some of the lakes have had significant changes in lake level which have created semi-isolated lake sub-basins and have raised the possibility of microallopatric speciation in isolated lake basins. This problem is particularly relevant to these large lakes that have very convoluted basins. In addition, many of the species hybridize under changed environmental conditions, especially changes in water clarity (see Seehausen et al. 1997). It's possible that mtDNA monophyly could be more apparent than real, and stem from mtDNA transfer between species through hybridization (see Taylor and McPhail (2000) for a possible example in the stickleback species pairs.)

The issue of microallopatric speciation in African cichlids is perhaps less of a problem in the smaller western African lakes in Cameroon (see Schliewen et al. (1994)).

Two such lakes, Bermin and Barombi Mbo (see below) are much smaller, crater lakes that are almost perfect circles. They have many fewer (< 10), but still endemic species of cichlids, and have single inlets and outlets. Fallinf lake levels, therefore, would probably tend to concentrate populations rather then isolated them.

Molecular phylogenetic data (mtDNA) also tend to reveal within-lake monophyly of these species flocks of cichlids (boxed areas in figures below) - again suggestive of sympatric speciation in these much smaller lakes.

Clearly, from all these examples of allopatric, parapatric, and sympatric speciation, basic biogeographical information and analyses have been central aspects of trying to understand that most vexing problem in evolutionary biology - the origin(s) of species!

References:

Bernatchez, L. and J.J. Dodson. 1990. Allopatric origin of sympatric populations of lake whitefish (Coregonus clupeaformis) as revealed by mitochondrial DNA restriction analysis. Evolution 44: 1263-1271.

Bush, GL,1994. Sympatric speciation in animals: new wine in old bottles. Trends Ecol. Evol. 8: 285-288.

Grant, P.R., Grant, B.R. & Petren, K. 2000.The allopatric phase of speciation: the sharp-beaked ground finch (Geospiza difficilis) on the Galapagos islands. Biological Journal of the Linnean Society 69: 287-317.

Irwin, D.E., S. Bensch, and T. Price. 2001. Speciation in a ring. Nature 409: 333-337.

Johnson, T.C. et al. 1997. Late Pleistocene desiccation of Lake Victoria and rapid evolution of cichlid fishes. Science 273: 1091-1093.

Mayr, E. 1963. Animal species and evolution. The Belknap Press, Harvard University.

Meyer, A. et al. 1990. Monophyletic origin of Lake Victoria cichlid fishes suggested by mitochondrial DNA sequences. Nature 347: 550-553.

Rice, WR, Hostert, EE, 1993. Perspective: laboratory experiments on speciation: what have we learned in forty years? Evolution 47: 1637-1653.

Schliewen, UK ,Tautz, D, Paabo, S, 1994. Sympatric speciation suggested by monophyly of crater lake cichlids. Nature 368: 629-632.

Seehausen,O, Van Alphen, JM, Witte,F, 1997. Cichlid fish diversity threatened by eutrophication that curbs sexual selection. Science 277: 1808-1811.

Szymura, JM, Barton, NH,1986. Genetic analysis of a hybrid zone between the fire-bellied toads, Bombina bombina, and B. variegata, near Cracow in southern Poland. Evolution 40: 1141-1159.

Taylor, E.B. and J.D. McPhail. 1999. Evolutionary history of an adaptive radiation in species pairs of threespine stickleback (Gasterosteus aculeatus): insights from mitochondrial DNA. Biol. J. Linn. Soc. 66: 271-291.

Taylor, E.B. and J.D. McPhail. 2000. Contingency and determinism interact to prime speciation in sticklebacks. Proc. Royal Soc. Lond. (B) 267: 2375-2384.

Taylor, E.B.,Z. Redenbach, A.B. Costello, S.M. Pollard, and C.J. Pacas. 2001. Nested analysis of genetic diversity in northwestern North American char, Dolly Varden (Salvelinus malma) and bull trout (S. confluentus). Can. J. Fish. Aquat. Sci. 58: 406-420.