Almost certainly most speciation occurs allopatrically (that is, when the forms occur in different places). The main driving forces appear to be stochastic -- drift and/or mutation (both stochastic forces) drive the divergence between the forms.  Reduced gene flow (lack of "migration" in the population genetics sense) is also important.  For example, the rise of the Andes appears to have been very important in producing patterns of genetic differentiation in neotropical plants and animals (Gascon et al., 2000). The best-worked examples of sympatric speciation in animals used to be for Rhagoletis flies (Feder et al., 1988). More recently, it has become a very active area for fish (Schliewen et al., 1994) and brood parasitic birds (Sorenson et al., 2003). A major driving force here may be specialization on different host plants or different habitats and resources (in the case of fish or birds). As the host plants diverge in phenology, secondary compounds or other ecological features, it becomes more and more disadvantageous to fall "somewhere in the middle". The eventual result may be the evolution of reproductive isolating mechanisms that reduce or prevent successful offspring production across forms. Some of the spectacular radiation of African cichlid fishes is probably also a result of sympatric speciation, perhaps driven by a combination of ecological and sexual selection (Schliewen et al. 1994; but see Meyer et al. 1996). Recent work (Verheyen et al., 1996) on the cichlids also suggests that water level lowering created separated lakes that further promoted speciation processes (i.e., more traditional allopatric speciation). One of the "fly" guys reviewed sympatric speciation in Trends in Ecology and Evolution (Bush, 1994). We will come back to an example of "ecological" sympatric speciation in a later section.  Sympatric speciation can occur more readily in plants (Savolainen et al., 2006), because plants can change their chrosome number (polyploidy) and persist or spread by asexual reproduction.  Parapatric speciation occurs where the ranges of two forms abut but do not overlap extensively. In such cases, lower fitness of hybrids drives increased differentiation, eventually resulting in premating isolation.

II. Sexual selection and speciation

Ever since Darwin (1871) it has been apparent that sexual selection may be a potent force in speciation. Perhaps the most succinct theoretical treatment of the idea is by Russ Lande (1981). Strong selection on secondary sexual traits (ornaments such as plumes or behaviors such as courtship displays) can lead to rapid divergence that produces reproductive isolation. See also the work by Schluter and Price (1993). Mary Jane West-Eberhard (1988) produced a classic treatment of sexual selection and the intriguing additional idea of social selection. Sexual selection produces the classic sexually dimorphic traits that distinguish, for example, peacocks from peahens. Social selection may be equally intense but produces monomorphic exaggerated traits. Good examples would include keel-billed toucans (Ramphastos sulfuratus) and puffins (Fratercula spp.). In such species, both mates compete intensively for scarce resources (e.g., nesting cavities for toucans and burrows for puffins) as well as for access to the best mates of the opposite sex (both sexes compete for mates and both parents invest heavily in parental care). Speciation here was probably NOT ecological -- the two sister species (horned and Atlantic) have different breeding microhabitats (rock crevices and earth burrows respectively), whereas the two most different species (Atlantic and tufted) are both burrowers. Whether the force is sexual or social selection a common feature is a certain degree of arbitrariness (stochasticity). Toucans in one location may become red-breasted and yellow-billed, while those elsewhere may become yellow-breasted and green-billed. Natural selection, in contrast, usually has a certain directionality or even optimality attached to it. John Endler has been a proponent of the idea that traits such as plumage color are predictable from features of the environment -- in a paper on manakins, he and Marc Théry argued that the particular colors of male plumage are a fine-tuned fit to the ambient light environment (Endler and Théry, 1996).  Sexual selection may not always be a diversifying force. An interesting example of how sexual selection can help mask speciation is provided by my graduate school colleagues Dave and Jean Zeh (1994). They studied several different species of pseudoscorpions (Cordylochernes) that ride on harlequin beetles. Although the pseudoscorpions have undergone considerable genetic divergence, sexual selection has maintained a consistent morphology across the set of different species. [See Turner and Burrows (1995) for a model for sympatric speciation via sexual selection].
Behavior and hybrid zones: Two recent examples exist for how the dynamics of hybrid zones may be affected by the mating system, sexual selection and behavioral differences among hybridizing forms.  Sievert Rohwer and his colleagues have studied the interaction between Townsend's Warblers (Dendroica townsendi) and Hermit Warblers (Dendroica occidentalis).  The Townsend's males are more aggressive and the hybrid zone is moving rapidly into the range of the Hermit Warbler.  Behavioral asymmetries are therefore driving Townsend's genes into Hermit country -- Hermit females mate with Townsend's males but not vice versa (Pearson and Rohwer, 2000; Rohwer and Wood, 1998).  I worked on a hybrid zone between two species of manakins in Panama.  The Golden-collared Manakin (Manacus vitellinus) and the White-collared Manakin (M. candei) come together in a narrow hybrid zone along the Caribbean coast of Panama near the Costa Rican border.  Beyond the hybrid zone is a zone of introgression.  Here the birds are genetically and morphologically indistinguishable from White-collared Manakins (Parsons et al., 1993).  One trait, though, the golden collar, has crossed the species boundary.  The clines for the genetic and morphological traits are narrow and coincident, but the cline for collar color is displaced 50 km to the west (Brumfield et al. 2001).  Using taxidermic mount experiments, my colleagues and I were able to show that Golden-collared males are more aggressive and that the introgression zone "lemon-collared" males resemble Golden-collared males not only in their collar color but also in their behavior toward other males (McDonald et al., 2001). A recent study by Stein and Uy (2006) updates that work by demonstrating that female choice may be the most important driving force behind the advantage to the introgressive yellow plumage.

III. Islands as "natural laboratories" for exploring speciation

We all know how much of an influence the juxtaposition of variation and similarity among islands in the Galapagos archipelago played in crystallizing Darwin’s thoughts on evolution, speciation and natural selection. Perhaps less well known is the influence of the Indonesian archipelago on Alfred Russell Wallace, whose development of ideas on evolution and speciation paralleled Darwin's. Islands have continued to play an important role in shaping the thinking of biologists on speciation. Much of Ernst Mayr’s (1942) classic work on speciation derived from his extensive knowledge of the birds of the Indonesian archipelago. A recent classic on speciation returns to the Galapagos. Peter Grant and his wife Rosemary have long studied the evolution of Galapagos finches (Grant, 1986; see particularly Chapters 10, 11, 13; Grant and Grant, 1996, 1997).
    Two recent studies provide interesting insights into patterns of differentiation along the Hawaiian archipelago. Roderick and Gillespie (1998) recently examined the patterns of speciation in arthropods of the Hawaiian archipelago (Arthropods comprise over 75% of the endemic biota of the Hawaiian Islands). Much of their study depended on compiling the results of many previous studies. Their major conclusions included: classifying patterns of speciation within Hawaiian arthropod lineages into three categories: (i) single representatives of a lineage throughout the islands; (ii) species radiations with either (a) single endemic species on different volcanoes or islands, or (b) multiple species on each volcano or island; and (iii) single widespread species within a radiation of species that exhibits local endemism. A common pattern of phylogeography is that of repeated colonization of new island groups, such that lineages progress down the island chain, with the most ancestral groups (populations or species) on the oldest islands. While great dispersal ability and its subsequent loss are features of many of these taxa, there are a number of mechanisms that underlie diversification. These mechanisms may be genetic, including repeated founder events, hybridization, and sexual selection, or ecological, including shifts in habitat and/or host affiliation. The majority of studies that Roderick and Gillespie reviewed suggested that natural selection is a primary force of change during the initial diversification of taxa.
    Fleischer et al. (1998) used the fact that the Hawaiian Islands are know to have arisen sequentially to put a geologically-based time window on speciation events along the archipelago. Kauai is estimated to have arisen 5.1 MYA (million years ago), compared to the big island of Hawaii at 0.43 MYA. The sequential geological pattern provides a relatively rare opportunity to calibrate molecular clock against a well-documented pattern of geological change as well as phylogenetic trees.

IV. Species concepts -- biological vs. phylogenetic

Avise's Chapter 7 provides a useful overview of the several major species concepts that have been proposed. Perhaps the two major contenders at this point are Mayr (1942) and Dobzhansky’s (1937) biological species concept (BSC) based on the primacy of interbreeding as a criterion, versus the phylogenetic species concept (PSC) in which the major criterion is that species must be monophyletic units with a diagnosable shared character (synapopmorphy). Cracraft (1983) defined a phylogenetic species as "the smallest diagnosable cluster of individual organisms within which there is a parental pattern of ancestry and descent" -- where the diagnosis depends upon synapomorphies defining monophyletic clades. Zink and McKitrick (1995) provide a well-developed argument for the adoption of the PSC. Avise and Wollenberg (1997) argue that the two are not the mutually exclusive concepts that many have considered them to be -- part of their argument is that the BSC can usefully incorporate some population genetics principles that are ignored by the strictly phylogenetic PSC.

V. "Instant speciation" via polyploidy etc.

Plants can speciate almost instantaneously by changing the ploidy (number of sets) of their chromosomes, accompanied by their ability to persist/disperse as "founders" via asexual reproduction. They spread vegetatively and don't necessarily have to deal with the problem of sexual reproduction for some time. In animals, less dramatically, a more rapid founder effect process (Mayr, 1942; Templeton, 1996) may sometimes be an alternative to the gradual process usually considered to dominate speciation in animals. On population genetics grounds, Slatkin (1996) argued for the potential importance of what is sometimes called "founder flush" speciation in the face of several dismissive reviews of the process. For animals, the major interest in "instant" speciation involves rapid divergence due to a few changes in major regulatory genes affecting development. Recently, however, a tetraploid rodent was discovered in South America (Gallardo et al., 1999). I suspect that few biologists would have considered this a possibility until it was reported.

VI. Ecological vs. "genetic" (drift and mutation) speciation

Dolph Schluter (1996) has recently championed the idea that speciation can occur as a result of local adaptation (i.e., emphasizing adaptation over random genetic drift as a driving force). Sockeye (anadromous) and kokanee (lake-spawners) salmon occur in the same lakes. Likewise two forms of whitefish, with different morphologies and mtDNA haplotypes, coexist in Maine and E. Canada. Smith et al. (1997) argue for a similar role of natural selection in the diversification of African passerine birds. They found high levels of genetic divergence across ecotones (habitat transitions) despite high levels of gene flow. The emphasis, in their eyes, lies in the relative importance of ecological selective forces over the stochastic forces of drift and mutation (rather than an emphasis on allopatric vs. parapatric vs. sympatric). Thus, ecology/natural selection, rather than pure genetic architecture, may drive speciation in certain cases (see a recent review in TREE by Orr and Smith, 1998). Look back at the sexual/social selection section to see how that has a stochastic flavor.  A recent paper by Rieseberg et al. (2002) used quantitative trait loci (QTLs) to argue that directional selection is the primary force behind diversification and speciation.

VII. Punctuated evolution and speciation

Stephen Jay Gould (1973), who died recently, was a major champion of the idea that speciation often precedes as relatively short bursts of accelerated evolution followed by long periods of stasis (low rate of change). "Punctuated equilibrium" is the term describing this paleontologically derived view of the evolutionary process. Along with this idea goes the idea of evolutionary constraints. Gould argued that many possible courses of evolution are simply not an option once certain paths have been followed. Ardent adaptationists, such as Richard Dawkins (1976), feel that the idea of constraints is badly overemphasized. For an amusing interchange see the back-to-back reviews of each other’s books by Gould (1997) and Dawkins (1997) in the journal Evolution.

VIII. Shifting balance theorem of Wright vs. mass selection/large size theory of Fisher

Two of the founding giants of the theories of population genetics were American Sewall Wright (U. of Chicago) and Briton R.A. Fisher. Although they agreed on most of the hard core of mathematical principles underlying population genetics, they disagreed in important ways about the relative importance of the three major forces (mutation, drift and selection) in producing the great diversity of organic form. Fisher (1958) felt that natural selection played a large role and that it did so in the context of large populations in which drift played a negligible role but rare advantageous alleles could spread by "mass selection". Wright (1978), in contrast, felt that natural selection worked in a bumpy "adaptive landscape"-- where the peaks represent gene combinations leading to high fitness and the valleys represent combinations of low fitness. Moving to higher peaks might be difficult and might require interdemic selection, migration between demes, periods of drift that could move populations across adaptive valleys towards higher peaks. Populations might reach local peaks but be unable to reach higher peaks because natural selection would keep them from moving down into the valleys. He termed this complex interplay between local natural selection and drift the "shifting balance theorem". Recently, Wade and Goodnight (1998) argued that the current balance of evidence makes Wright’s shifting balance more plausible. Metapopulation dynamics, especially, may well be the heretofore-unconsidered factor that makes Wright’s scheme work. Metapopulations involve blinking on and off, sinks and sources and other features that seem much more compatible with Wright's shifting balance theorem than with Fisher's "mass selection" view.

IX. Haldane's rule

In the event of hybridization, the heterogametic sex is the most likely to suffer reduced fertility (hybrid sterility) or viability. In mammals, this means male sterility, in birds, females are the heterogametic sex. Haldane's rule can play an interesting, if somewhat minor role in the dynamics of speciation [See Orr, 1997; Turelli, 1998; Tegelström 1987 provides a case history for Ficedula flycatchers].