Contents

Introduction


Chromosomal Speciation Theory

Basics

genetic revolutions

chromosomal differentiation

hybridization

The geographic component

Cascading revolutionary speciation

cascade initiation

chain termination

evidence

Karyotypes as a phyletic tool


Basics

Wallace (1959), White (1969, 1973), Nadler (1969), Mayr (1969a, 1970) and others have noted that certain classes of chromosomal rearrangements are frequently found fixed between closely related species but are only rarely found as polymorphisms within populations of these species. Such mutations usually have no obvious adaptive functions as polymorphisms, and in many cases are suspected to cause meiotic assortment difficulties in heterozygotes that would reduce their fertility. Other kinds of rearrangements, such as the paracentric inversions of Drosophila (Dobzhansky, 1970) and the reciprocal translocations of Oenothera (Stebbins, 1950) have clear adaptive functions as polymorphisms within the genetic systems of the species they characterize. Robertsonian rearrangements clearly belong to the former class of mutations. For example, about one of every four Drosophila species differs by a Robertsonian rearrangement; but, in spite of the extensive cytological sampling of Drosophila, Wallace (1959) could find only one instance of Robertsonian heterozygosity from nature; and this occurs only in the zone of hybridization between D. americana americana and D. a. texana (Stone and Patterson, 1947; Hsu, 1952; Carson and Blight, 1952). Sceloporus would appear to provide another example of this situation, and many other similar examples from cytologically well known groups are cited by White (1969, 1973).

On theoretical grounds we might expect meiotic malassortment from Robertsonian trivalents to produce some percentage of aneuploid gametes (Ohno, 1965), which would then reduce the effective fertility of the heterozygotes if these gametes competed for fertilization. Among vertebrates such meiotic malassortment and consequent reduced reproductive fitness in heterozygotes has been demonstrated in humans (Hamerton, 1971), mice (Tettenborn and Gropp, 1970; Gropp et al., 1972), and cattle (Gustavsson, 1971a, 1971b) . Certainly cases are also known where Robertsonian rearrangements survive as apparently stable polymorphisms (e.g.. Ford et al., 1957; Meylan, 1964, 1965; Matthey, 1966; Jotterand, 1972; etc.; and see below), but these are not frequent, and to my knowledge, no reasonable mechanism whereby the Robertsonian polymorphism as such would have an important adaptive function in the genetic system of a species has been documented.

On the other hand, the frequent fixation between species of mutations which theoretically might reduce heterozygote fertility has led many cytogeneticists and systematists to propose models of chromosomal speciation based on the assumption that such chromosomal mutations do in fact reduce heterozygote fertility during the speciation process (Callan and Spurway, 1951; Spurway, 1953; Spurway and Callan, 1960; Lewis, 1953, 1966; Lewis and Raven, 1958; Wallace, 1959; Matthey, 1964, 1965; John and Lewis, 1965, 1966; White et al., 1967; White, 1968, 1969, 1973; Key, 1968; Mayr, 1969a, 1970; Todd, 1970; Arnason, 1972; etc.). However, rather than attempt to review each specific model here, I will summarized what I think are some of the more important aspects in the development of a more-or-less synthetic theory of revolutionary (frequently, but not always chromosomal) speciation.

With an admitted (over-?) simplification, two extremes of species formation may be recognized: conservative or "classical" allopatric speciation, and a revolutionary mode of speciation. The conservative type involves the geographic separation of comparatively large populations followed by their probably slow genetic divergence in response to the average differences in selection pressures operating on the respective isolates, and is a process which no rational biologist now disputes. Such speciation probably needs geological times to reach completion. The revolutionary mode of speciation invokes the founder principle and genetic revolution (Mayr, 1942, 1954, 1963, 1970).

Genetic revolutions

A genetic revolution involves large shifts in the frequency of alleles at polymorphic loci (or even their fixation) in local populations as a consequence of their restriction to such effectively small sizes (say N < 20) that large sampling errors may occur in the reproduction of gene frequencies from one generation to the next (Wright, 1940, 1951, 1970). This effectively small population size is possible only if gene exchange with other populations is minimal or absent. Given the necessary isolation, the chance effects plus local environmental selection may occasionally produce in only a  few generations a new adaptive genotype which has also achieved a new epistatic balance. Although this new genotype will contain no new alleles, it may differ enough from the parental stock in epistatic balance so that hybrids with the parental stock are considerably less fit than are progeny from matings within the respective "pure" populations. The case of apparent incipient speciation in the geographically isolated Bogota population of Drosophila pseudoobscura described by Prakash (1972) appears to exemplify this effect.

Chromosomal differentiation

Such post-revolution populations may survive as biological species if they are well enough adapted to the local ecology to compete successfully with other forms seeking to use it and if they are well enough isolated from the parental stock to keep introgressive gene flow below that which would break down either the new environmental adaptations or the new epistatic balance. However, under most conditions lacking strict geographic isolation, it would seem that occasional spurts of gene flow from neighboring populations would inevitably break down or swamp the new balances (Wallace, 1959; Mayr, 1963). But, on the other hand, chromosomal differentiation during revolutionary speciation in strict geographic isolation (as presumed for the Bogota D. pseudoobscura) would seem to play no role in the process sufficient to account for the apparently frequent fixation of possibly heterozygously semisterilizing chromosomal differences between sister species-especially between sister species1 of parapatrically distributed swarms as exemplified by cytologically comparatively well known groups such as: the morabine grasshoppers (White et al., 1967, 1969; White, 1973, etc.); the pocket mice, Perognathus (Patton, 1967a, 1967b, 1969a, 1967b); pocket gophers, Thomomys (some populations only--Thaeler, 1968); mole rats, Spalax (Wahrman et al., 1969a, 1969b; Lay and Nadler, 1972); the equally fossorial tuco-tucos, Ctenomys (Reig and Kiblisky, 1969); etc.; and especially the cryptic species of the Sceloporus grammicus complex (see below). However, as I will now show, chromosomal differentiation may afford a revolutionary population which is not well isolated geographically from its parental stock with an intrinsic barrier to gene flow that would allow it to survive and differentiate as a species even when exposed to the occasional spurts of potential gene flow that would swamp a karyotypically ancestral population.

According to Wright (1941), the conditions required for a genetic revolution are also precisely those which would allow the chance fixation of a chromosomal mutation which reduced heterozygote fertility. He calculated that a mutation not effecting fitness when homozygous, but reducing heterozygote fertility by 50%, still had better than one chance in 10-3 of being fixed after an indefinite number of generations if it occurred in a population with an effective size of 10. Of course, the probability of fixation of such a mutation becomes impossibly low as population size increases; or, more importantly, the probability of fixation becomes greater if heterozygote fertility is higher or the effective population size is lower. However, once a local population becomes fixed for a heterozygously semi-sterilizing mutation, the reduced fertility of  chromosomally heterozygous hybrids and backcrosses, etc. with immigrants from the ancestral stock would reduce the effective introgression of ancestral genes considerably below that which would result from the same immigration rate into a non-chromosomally differentiated population. Therefore, the chromosomally differentiated post-revolution population would have an immediate degree of intrinsic protection against potentially disruptive gene flow lacked by non-chromosomally differentiated populations.

Hybridization

It might be argued, notwithstanding the partial sterility barrier provided by chromosomal differentiation, that any hybridization and consequent introgression would slow or disrupt the speciation process (e.g., see Mayr, 1963). But the fact that all species in some of the swarms of parapatric species cited above seem to be chromosomally differentiated and the fact that this differentiation would seem to play no function in purely allopatric speciation suggests an alternative view. Because the low fertility of chromosomally heterozygous hybrids would function to reduce the effective fitness of hybridizing parents, at least before the post-revolution population grows large, hybridization could serve as an intrinsic selection mechanism to encourage the perfection of any ecological or behavioral differentiation which served to reduce the frequency of hybridization (Fisher, 1930; Dobzhansky, 1940, 1941, etc.).

Mayr (1963), arguing against this theory, supposed that pre-mating isolation could not be evolved by reduced hybrid fertility until postulating isolation was essentially complete (which it certainly is not in the present model, at least initially), because he thought that introgression through the reduced fertility barrier would tend to break down whatever genetic difference(s) caused the reduced hybrid fertility. However, hybrid chromosomal semisterility would serve as an intrinsic selective mechanism to maintain itself in the local population against anything up to a 50% frequency of ancestral-type chromosomes (although, if this 50% frequency is exceeded, selection would then work to eliminate the mutant chromosome from the local population). It also seems likely that genie selection within the post-revolution population could maintain new coadaptive and environmental balances against small amounts of potentially disruptive gene flow penetrating the essentially fixed hybrid semisterility barrier (e.g., Bigelow, 1965). Moore (1957) suggested several other reasons why hybrid infertility probably could not serve as an effective selective mechanism to promote premating isolation (see below). However, these all rest on the basic assumption that only small fractions of the hybridizing populations occur in contact or overlap zones where they risk hybridization; and, of course, this assumption does not apply to the extremely localized, early post-revolution population where all of its members would bear essentially the same risks of hybridization. Only if the post-revolution population grows large enough so that many of its individuals are geographically removed from the potential dispersal paths of the ancestral stock do Moore's arguments become effective.

Possibly much more interestingly, limited introgression might provide an important source of genetic variability which could be selectively incorporated into the new adaptive and coadaptive balances to allow the post-revolution population to differentiate much more rapidly from the ancestral condition than could a strictly isolated and therefore genetically depauperate founder population of Mayr (1942, 1963, etc.). Key (1968) supposes that the semisterility of the hetero-zygotes will act like a semipermeable barrier to selectively hold back or trap genes or rearrangements that favor hybridization, and to pass those which would reinforce the degree of isolation or improve the local adaptations of the homozygous mutant population. This semi-permeable barrier, generally a zone of parapatric hybridization which Key not too appropriately called a tension zone (see below), will be pushed across the countryside to a point of equilibrium where the population pressures of the respective pure populations are equal; i.e., until the populations on either side of the hybrid zone are equally well adapted to the local conditions. In the absence of premating isolation, the hybridization will also serve as a barrier to prevent any inter-penetration of the populations (Zaslevskii, 1963). Therefore, as long as the new genie equilibria of the chromosomally differentiated post-revolution population are not swamped by too many genes filtering through the semipermeable partial sterility barrier, limited hybridization with the ancestral stock might serve both as a selective mechanism and as a source of genie variability to speed the evolutionary differentiation.

If the chromosomally differentiated population achieves a successful adaptation and grows large enough so that many of its individuals no longer risk hybridization themselves, because they are separated from the ancestral stock by a zone of parapatric hybridization, then a further, not-so-obvious effect of the low hybrid fertility will work to reduce introgression into this "protected" population even below that which successfully passes the partial sterility barrier of the chromosomal heterozygotes. If chromosomal heterozygotes formed in the zone of hybridization are less fecund than the chromosomal homozygotes, it then follows that the population pressure competing for limiting resources in the hybrid zone will be reduced by an amount directly related to the proportion of heterozygotes in the total population and the reduction in their fertility with respect to the homozygotes. This reduced population pressure will very likely result in a net dispersal of individuals from areas of pure populations into the hybrid zone, Consequently, the hybrid zone will serve as a partial vacuum, or a "sink," for gene flow; and to get out of the immediate area of the hybrid zone, any ancestral genes penetrating its semisterility barrier would still have to migrate out against a net gene flow into the hybrid zone.

 The effectiveness of this sink effect as a trap for introgressed genes (i.e., those which have successfully backcrossed into chromosomally homozygous individuals) will of course depend on the degree to which the population pressure of the hybrid zone is reduced with respect to pure populations and on the vagilities of the individuals comprising the various populations. This is because these factors will determine the steepness of the diffusion gradient against which the introgressed genes must migrate. It is not intuitively clear to me what values these parameters must have before the sink would become completely effective as a trap for gene flow, but these should be easy to determine mathematically or with computer simulation experiments. However, given a sufficiently reduced heterozygote fitness and an appropriate population structure, it seems evident to me that the hybrid zone could become completely effective as a barrier to gene flow even without complete heterozygote sterility. Therefore, given the chromosomal differentiation of a post-revolution population and a chance for it to spread enough to form a central population protected from the risk of hybridization, this central population might be able to evolve completely independently from the parental stock even without the evolution of complete reproductive isolation in the contact zone. If this situation pertains, further spread and differentiation of the chromosomally mutant population would then depend only on maintenance of a high enough population pressure on its side of the contact zone to counterbalance the population pressure of the ancestral stock and any surface tension effects due to concavity of the hybrid zone (Key, 1968; and see below).

The sink effect is implicit in Key's (1968) "tension zone" concept, and other discussions of the maintenance of narrow zones of parapatric hybridization by semisterility (Woodruff, 1972), but I do not know that the possible effects of the reduced population pressure resulting from the semisterility have ever been explicitly considered. Perhaps this is because Key's term, "tension zone," is not a good metaphor for this reduced population pressure, although it does signify the disruptive selective effects of the chromosomal discontinuity and the surface tension effects of the hybrid zone. On the other hand, the idea of a "sink" does suggest that the zone of hybridization is an area of comparatively low pressure that would tend to suck in migrants and the genes they carry--genes which would eventually "go down the drain" when they found themselves in lethally unbalanced zygotes formed as a result of malassortment by the chromosomal heterozygotes.

When hybrid zones formed as a result of reduced hybrid fitness are thought of as zones of reduced population pressure, it becomes immediately obvious why, as Key (1968: p. 19) suggested but did not convincingly explain, that factors of any kind having negative heterotic effects should interact to stack up in a hybrid zone to "lead eventually to full reproductive isolation, in the sense of complete infertility or inviability of hybrids between the populations on each side of the zone (i.e., to speciation)"--without the evolution of premating isolation. Furthermore, the sink effect of the hybrid zone also seems to provide a completely adequate explanation for the puzzling long-term stability of zones of parapatric hybridization between populations which have large ranges in proportion to their hybrid zones, such as are found in Sceloporus grammicus (Hall and Selander, in press; and see below), the grasshoppers cited by Key, etc., and in many other vertebrates (e.g., see Mayr, 1963, 1970; Woodruff, 1972, MS). Not only do genes filtering past the heterozygotes, i.e., genes which might otherwise tend to break down genie differences between the populations, find it difficult to get out of the hybrid zone against the diffusion pressure of a net immigration, but the same problems would also apply to any genes favoring premating isolation (Woodruff, 1972). It seems quite unlikely that any single gene would ever enable its chromosomally homozygous carriers to successfully discriminate among its "conspecifics," homozygous "others," and the host of chromosomally heterozygous F1, F2, and backcrosses, because clearly the carriers of the gene would have to avoid mating with the carriers of the alternative chromosome arrangement to avoid eventually losing the gene in lethally unbalanced zygotes. In all probability, any genes allowing only partial discriminatory ability to their carriers would still be trapped before they found their ways into combinations that afforded their carriers with the required complete discriminatory ability.

Besides the diffusion gradient tending to keep these "discriminator" genes out of the main population, Moore (1957) suggested two more reasons why they would not spread out of the hybrid zone:

  1. Any specialization of population A in the hybrid zone which reduced competition with the similar population B would probably be maladaptive for A away from the area where B is encountered, because A would not then be making full use of all the resources potentially available to it,
  1. Similarly, genes which allowed A in the contact zone to successfully discriminate against mating with the similar B would probably prevent A from mating with some of its conspecifics in the area away from the contact zone.

From the discussion above, it should be clear that if a heterozygously semisterilizing chromosome mutation can once become fixed in a local population large enough to be protected by a zone of parapatric hybridization, it will then be comparatively easy for this population to spread further and evolve independently as a new species if it has by chance and/or local adaptive response achieved a genotype more suited to its local habitat than are those of the adjacent populations of the ancestral stock. Also, given a reasonable population structure, mutation rate, and meiotic system in the ancestral stock, it seems likely that local populations within it will occasionally happen to be sufficiently isolated at the right time and for the necessary few generations to allow one of them to become by chance chromosomally differentiated. However, the most critical stage in the speciation process will occur after the initial fixation, when the homozygous mutant population is still small, because to survive and grow it must be able to counterbalance the population pressure of the much larger ancestral stock where the two populations contact. Then, the reproductive fitness of the homozygous mutants must be high enough to push the same absolute number of migrants into the hybrid zone as does the much more extensive ancestral stock, or otherwise the zone will contract around the mutant population until it is swamped out of existence. On the other hand, it is also at this early stage when the semipermeability of the semisterility barrier can most effectively change a post-revolution population by simultaneously adding genie variability and selectively removing genes which favor hybridization; and it is at this stage when a major adaptive change may be most critical to the survival of the post-revolution population. For these reasons, in this early stage of differentiation the geographic relationships with the parental stock will be most important for the derived population, because these will determine the effective pressures that the derived and parental populations can bring to bear on the hybrid zone.

The Geographic Component

Although revolutionary speciation involving chromosomal differentiation may occur in rough sympatry, i.e., without prolonged or absolute geographic isolation (at least in comparison to the conservative mode of speciation), in an area historically inhabited by the parental stock, it cannot be over-emphasized that the parental and speciating populations must still be separated a least microgeographically during the first stage of the speciation process to provide the necessary degree of inbreeding for the revolution and chromosomal differentiation. Three ways this separation might occur within the broad geographic range of a parental species may be postulated. Two extremes may be termed the "peripheral" and "interstitial" versions; with a third, "internal periphery," version representing somewhat of an intermediate version.

As the name implies, the peripheral version supposes that adequate genetic isolation for the genetic revolution and fixation of a heterozygously semisterilizing chromosome mutation can occur only on the geographic periphery of a species; perhaps as a propagule passes some local environmental barrier or as a local breeding unit becomes isolated from the main population during climatic fluctuations (Callan and Spurway, 1951; Spurway, 1953; Lewis, 1953, 1966; Wallace, 1959; Mayr, 1963, 1969a, 1970; Key, 1968). In this type of speciation, recently derived post-revolution species would be found only peripherally to their parental stocks. Possible examples of this situation are provided by Spalax (Wahrman et al., 1969a, 1969b; Lay and Nadler, 1972), Perognathus (Patton, 1969b), and morabine grasshoppers according to Key (1968).

The interstitial version supposes that opportunities for revolutionary speciation might occur within the normal range of a species if it has a normally scattered and clumped population structure, as do morabine grasshoppers living on scattered bushes (White, 1968, 1969, in press), many terrestrial and fossorial vertebrates (e.g., burrowing rodents in pockets of loose soil--Arnason, 1972; White, 1969, in press), and Sceloporus grammicus in areas where they are found living on scattered fallen logs (Moody et al., in prep,; and see below). If some of the local populations of such a species are effectively small and also effectively (but not necessarily absolutely) isolated for probably a minimum of 10 to 20 generations at a time, then incipient speciation of one of the populations would depend only on the chance fixation of a heterozygously semisterilizing mutation while it was reaching a new adaptive equilibrium by chance drift and local selection. If this interstitial type of speciation is common, and if recent post-revolution populations can be found, some of them should be completely surrounded by populations of the parental stock. This is similar to the "stasipatric" model (White et al., 1967; White, 1968, etc.), which White believes is demonstrated by the "speciation" in the morabine grasshoppers, in that the initiating chromosomal rearrangement becomes fixed in an interstitial population (but see Key, 1968, for an alternative interpretation of the morabine situation). It also seems possible that speciation in the Sceloporus grammicus complex conforms to this pattern (see below).

In the peripheral version, only a small fraction of the total population of a species is likely to occur in the small peripheral isolates where revolutionary speciation could take place. Furthermore, these peripheral populations would seem to suffer a high probability of going extinct when their ecological tolerance is exceeded by slightly abnormal environmental fluctuations (which is why they are peripheral). On the other hand, their isolation is likely to be more extreme than is that of the interstitial populations, and they would likely be exposed to more extreme disruptive selection for differentiation from the central populations. If revolutionary speciation and ecological differentiation occurred, a nascent species could then easily spread geographically into areas not occupied by its parental stock (Wallace, 1959; Patton, 1969b). And, probably most importantly, because of its peripheral location, the derived population would form a hybrid zone only on the front facing the ancestral stock, thereby avoiding or at least reducing the surface tension effect that would otherwise cause the hybrid zone to have a tendency to close around and extinguish the derived population.

In the interstitial version, a comparatively large proportion of a species may live in local inbred populations where revolutionary speciation might occur. Although these local populations may not always maintain sufficient isolation for successful revolutions, they would be far less likely to go extinct for environmental reasons than would peripheral isolates. And, even though an interstitial population is not peripheral, it is quite possible that selection pressures in its local environment would differ from those average for the species. For example, in a species which lives in more than one habitat type (dense forest and open woodland, flood plane and hillside, etc.), but most individuals live in only one of the habitats, they would probably be best adapted to this favored habitat. Then, if a genetic revolution provided sufficient reproductive isolation and resulted in a genotype which was more effective in a less favored habitat than the usual genotype of the ancestral stock, this incipient species could then geographically exclude the ancestral population from this less favored habitat. Though less likely, before the nascent species grows large enough to be completely protected by a zone of parapatric hybridization, it might be forced by competition and selective hybridization with the parental stock to differentiate enough ecologically to allow the two species to coexist sympatrically. Given a fortuitous population structure, many more revolutions and initial chromosomal differentiation events would be expected to occur interstitially than would take place on the much more limited periphery of a species which had a more continuous distribution through its range. However, the newly differentiated population would face the possibility of hybridization all around its circumference, and it would therefore require an especially superior local adaptation to counterbalance the tendency for the hybrid zone to contract.

An intermediate, and possibly more realistic version of the revolutionary speciation model acknowledges that many species living in complex environments have areas within their broad geographic ranges where they normally do not live, though they may occasionally cross these areas (e.g., different climatic zones on a mountain, forests ^s meadows, rocky hillsides vs alluvial flats, etc.). On a microgeographic scale within the range of interest for revolutionary speciation, ecotones between these different habitats will be ecologically similar to the periphery of a more continuously distributed species (Key, 1968). The presence of such internal peripheries within the broad range of a species would increase by several orders of magnitude the opportunities for revolutionary speciation in comparison to those available to a species with a more-or-less continuous distribution within its range. As in the external periphery situation, the internal periphery would considerably reduce the extent of the hybrid zone formed with the ancestral stock, and would thereby facilitate the survival of the derived stock. After a species formed on an internal periphery has expanded its range, as in the interstitial version, one would expect to find the derived species completely surrounded by its parental stock. However, in a geographic exclusion situation of the interstitial version, the derived species should be found to occupy a habitat type clearly within the ecological range inhabited by the parental stock away from the area inhabited by the derived stock. Whereas, in the internal periphery version, the derived species should be found in an ecology definitely peripheral to that of the ancestral stock.

Whether revolutionary speciation is supposed to be peripheral, interstitial, intermediate, or all of these, it is worth noting that the fixation of chromosomal rearrangements between species seems to be most frequent in groups of organisms of apparently limited vagility; while, on the other hand, highly mobile and presumably outbred groups such as the Cetacea, Pinnipedia, birds, and bats generally show remarkably stable karyotypes (Arnason, 1972; White, in press). Particularly notable examples of the former situation are the burrowing rodents cited above (Perognathus, Spalax, Thomomys, and Ctenomys) and probably at least some populations of the karyotypically diverse Sceloporus grammicus (see below). So, it would seem that whatever the geographic details of the revolutionary speciation process, if the above mentioned species have been formed by it, then their limited vagility has probably aided the speciation process, presumably by promoting inbreeding,

Cascading Revolutionary Speciation

The revolutionary speciation model suggests extensions which have interesting implications for systematics. One offers a possible explanation for the occasionally explosive proliferations of species that some groups (such as Sceloporus) form by contrast to their closely related stocks which appear to have had similar ecological opportunities but have speciated much more conservatively (e.g., the other sceloporines). Not only can revolutionary speciation occur without the degree of geographic isolation needed for conservative speciation, but occasionally, in comparison to its ancestral stock, a revolutionarily derived species may show an intrinsically amplified probability of further speciation by revolution; and so on to form further generations of species by a cascading process, until the process terminates, either for intrinsic reasons or because there are no longer ecological niches available for further derived species. In other words, an initial revolutionary speciation event may in some instances start a process which leads to the formation of not just one or two new species, but, rather, to the rapid proliferation of a whole swarm of sister species. The model is subdivided into two categories of conditions and events: those which initiate and accelerate the cascade process, and those which terminate its chains of derivation. The development of the model here is largely intuitive; but, besides the qualitative testing which can be gained from comparative studies of appropriate natural radiations (see below), most aspects should be amenable to mathematical testing and verification by computer simulation (not yet attempted).

Cascade initiation

The conservative mode of speciation depends mainly on the extrinsic interposition of geological, ecological, or climatic barriers between subdivisions of an ancestral population, and then on the differential effects of extrinsic natural selection on the separated populations. Intrinsic characteristics become important only when (and if) the separated populations come into secondary contact.

On the other hand, speciation by revolution depends almost exclusively on intrinsic characters of the genetic system of the ancestral stock, e.g.:

  1. on genes which influence population structure and vagility;
  2. on aspects of the recombination system which determine
    1. how much genetic variation is carried,
    2. how this variation is distributed geographically, and
    3. how fast it can respond to natural selection;
  3. on the availability of karyotypic possibilities for rearrangements which reduce heterozygote fertility;
  4. on genes which affect mutation rates for rearrangements that reduce heterozygote fertility (Ives, 1950);
  5. 5) on genes which affect the frequency of mal-assortment in structural heterozygotes;
  6. etc....

These characters of the genetic system will vary between species and populations, and hence some species or populations may be genetically more apt to generate new species by the revolutionary mode than will others. Selection may affect all aspects of the genetic system. However, selection would be able to work only extremely slowly on things like mutation rates and the meiotic assortment of the heterozygous mutations, which would occur only rarely, but would nonetheless be of primary importance in determining the rate at which an ancestral stock initiated chromosomal speciation events. The reasons for this slow response to selection are twofold:

  1. selection on mutation rate would be effective only when a mutation actually occurred, and for meiotic assortment only in the (usually) rare heterozygotes produced by the rare mutations; also
  2. these phenomena are likely to be only rarely penetrant pleotropic effects of genes much more stringently controlled for their normal functions.

With only infinitesimal rates of selection, characters of this kind will probably show considerable variation by random drift. Then, given:

  1. that important characteristics of the genetic system involved in revolutionary speciation may vary among local populations of a species, and
  2. that all extrinsic factors relating to revolutionary speciation are held constant,

it follows that the essentially chance event of revolutionary speciation will occur most probably within a local population which has the most favorable genetic system for it. And, in turn, because its genes are inherited from the local population which founded it, the post-revolution species may on the average be more likely to produce further derived species by the revolutionary mode than will be the ancestral stock. Then, of course, the same type of selective sampling process which amplified the frequency of favorable alleles in the origin of the first generation reyolutionarily derived species may be repeated in it to produce a second generation of even more derived species, which in turn may be still more likely to produce a third generation of derived species, and so on.... until something happens to terminate the cascading process (see below). In other words, this genic amplification process will accelerate the rate of speciation as the cascade proceeds.

Two predictions may be made about the phyletic structures of recent radiations formed by the cascade process: the process of derivation is likely to result,

  1. in the formation of one or a few long sequences of chromosomal derivation in the cascade rather than many short sequences deriving independently from comparatively primitive species; and
  2. species formed later in a chain of derivation may be able to occupy a much smaller geographic area and/or ecological range before giving rise to further derivatives than will species formed early in the chain.

These predictions will be discussed in turn.

Presumably the ecological and/or geographical area available for a radiation will be limited. It is also reasonable to assume that a new species can be formed only peripherally to an ancestral stock, whether this periphery is ecological or geographical. Since any species formed will expand to fill the space for which it is suitably adapted, it is quite likely that a derived species will then expand along the periphery of its parental stock, and thereby limit opportunities for this parental stock to form additional derived species along this periphery. This may be called latera1 inhibition. Furthermore, because of the amplification effect of the derivation process, it is probable that the derived species will form a third species to further saturate the available habitat for speciation before the first (ancestral) species tries to form another derivative. Similarly, because of the amplification, the third species in the sequence is likely to form a fourth species before either the first or second will. Hence, the net effects of the amplification and lateral inhibition will produce a sequence which tends to be linear rather than branched. That is not to say that a cascade will not be branched, because many extrinsic effects will influence the speciation and some chains may be terminated for one reason or another (see below); but, in general, the cascade will tend to be more chain-like than fan-like.

The second prediction of the cascading chromosomal speciation model follows directly from the amplification effect. Since species formed late in a sequence of derivation will likely form further derived species, much more rapidly than will species formed early in the sequence, these late derivatives in general may have much less opportunity to grow in number or to spread geographically before their "progeny" are formed. These late derivatives may therefore be quite ephemeral with respect to either early derivatives or to species which terminate their sequences of derivation. As will be seen below, both phyletic predictions of the cascade model seem to be fulfilled by the data from several mammalian genera and the Sceloporus grammicus complex.

Chain termination

It should be obvious that cascading speciation cannot go on indefinitely at a high rate. If nothing else happens, the process will end because the available niches become completely saturated with derived species. Or before that, the available chromosomal substrate for a favored class of rearrangements may be used up (e.g., centric fissions are no longer possible when there are no more bi-armed chromosomes in the karyotype), or natural selection may eliminate the factors which account for the reduction in heterozygote fertility; or other, more subtle mechanisms may work to otherwise change the genetic system so revolutionary speciation is no longer likely. The mechanisms for the first two kinds of chain termination are obvious; those for determining heterozygote fertility will be discussed in detail, both because they are easily analyzed and because the degree to which fertility is reduced is clearly a controlling variable in the speciation process. However, the mutation rates for semisterilizing mutations must also be discussed, because the extent of the natural selection operating to modify heterozygote sterility is directly related to the frequency of heterozygosity, which, in turn, is related to the mutation rate for the heterozygously semisterilizing rearrangements. For the discussion to follow we will assume that the mutation rate is controlled by "imitator" genes, which Ives (1950) has shown to exist in Drosophila populations, probably more frequently in the wild than in the lab. The critical point in the selective interaction between mutation rate and the degree of heterozygote semisterility is that the mutation rate is likely to be raised by the amplification effect of the cascade process. This is because, all other things being equal, the frequency of speciation in a lineage will be directly related to the rate at which heterozygously semisterilizing rearrangements that might facilitate the speciation are generated, Mutator genes that increased the mutation rate would then probably be concentrated by the amplification effect.

All would agree that selection will work to eliminate genes which reduce the fertility of their carriers, and that the magnitude of this selection will be directly related to the degree by which the fertility is reduced. However, even a considerably reduced fitness of the chromosomal heterozygote will have little effect on the mutator gene which caused the chromosome mutation after the first or second generation following the initial rearrangement event (assuming that the gene and the rearrangement it "caused" are not closely linked). Furthermore, it is probably reasonable to assume that the mutations caused by mutator genes are the result of rarely penetrant effects of an allele whose normal functions are adaptive. Given this assumption, then selection will probably have little effect on mutation rate, at least during an active cascading sequence, when major shifts in gene frequency and even chance fixations are likely to result from the various sampling processes included under the rubric of the amplification effect.

Assuming humans to be "normal," Polani et al. (1965) have estimated that new centric fusions involving chromosome 21 occur at a frequency of 2 x 10-5 . The frequency for all possible fusions in the human karyotype is then probably not over 1 x 10-4, assuming that all acrocentrics show similar probabilities of being involved in fusions. An increase in the mutation frequency by an order of magnitude, to 1 x 10-3 , would also increase the rate or probability of chromosomal speciation by an order of magnitude, but even a mutation frequency of 1 x 10-3 will have minimal effects on the frequency of a rarely penetrant mutator gene. The selective disadvantage of a nearly fixed mutator gene would be approximated by the frequency of mutations it caused if heterozygotes had zero fitness. And, as heterozygote fitness improves, the selective disadvantage of the mutator gene diminishes, although in a complex relationship, since the net disadvantage must be figured over several generations. However, we can take the base mutation rate as a reasonable approximation of the selective disadvantage of a common mutator gene. Then, even with a disadvantage as great as s = 1 x 10-3 , according to Crow and Kimura (1970), more than 9000 generations would be required to change a gene with no dominance effecf (the simplest case) from a frequency of 0.5 to 0.01. Given the capability of populations to increase exponentially if their overall fitness is high, a well adapted chromosomally derived new species should be able to saturate the available niche, and hence be ready to form further new species, long before selection would have a chance to greatly change the frequency of the mutator genes controlling the probability of that further speciation. And, it does not seem unreasonable to suppose that other, more subtle aspects of the genetic system will be modified similarly.

However, selection for reducing heterozygote sterility will behave somewhat differently. Every chromosomal heterozygote will provide an opportunity to selectively increase fertility. With low mutation rates and no history of chromosomal speciation, these opportunities will be infrequent. However, during each episode involving fixation of a chromosomal difference, there will be a short period when a large fraction of the local population will be chromosomal heterozygotes; and, if a mutation once becomes fixed in a local population, there will be a much longer period of time when chromosomally heterozygous hybrids are formed with the ancestral stock. So, for as long as the population remains too small to benefit from the sink effect of a well established zone of parapatric hybridization, each chromosomally heterozygous hybrid formed will provide an instance of selection favoring the increase of heterozygote fertility. Also, note that if heterozygote fertility in the hybrid zone increases significantly, this will diminish the sink effect of the zone, which will allow longer survival and greater dispersal to the genes. If the sink effect is initially strong, this should cause no problem to the speciation process, because the fertility increasing genes would still be lost in the sink. But if the sink is weak, such genes could eventually combine to improve fertility to the point where the sink was no longer functional, and the hybrid zone would decay into a condition of local polymorphism. If such an aborted speciation event occurs after several successful chromosomal speciation events in a sequence, it is quite likely that the basal rate for chromosome mutations in the terminal species would have been boosted to a level where mutations and the consequent heterozygosity were not uncommon. Nor is it unreasonable that there would be more than one aborted speciation attempt in the terminal species. These situations would insure that genes providing full fertility to the chromosomal heterozygotes would spread throughout the terminal species of the chain of derivation, thereby eliminating any further possibility of chromosomal speciation from this species and coincidentally also eliminating the selection pressure that would tend to reduce the rate at which new chromosome mutations were added to the population. Therefore, if the cascade is stopped by this intrinsic mechanism, the chain terminating species would almost inevitably carry many chromosomal polymorphisms of the kind usually fixed between more primitive species of its lineage and those of related lineages.

Other aspects of the genetic system may be similarly affected during a cascade, but it is not as obvious what they might be or what changes one might predict as results of the cascade. However, the cascade model does make definite and testable predictions about the phyletic relations one should find as a result of a cascade and about some of the changes in the genetic system that should occur during a sequence of derivation. These predictions will be tested by the radiation of crevice-using Sceloporus discussed below. However, it will first be useful to review other comparative data that are available which may also provide tests of these predictions.

Evidence

To summarize the preceding discussion, several potentially testable predictions can be made from the cascading speciation model. These assume that chromosomal semisterility provides the initial isolation; other bases for the isolation are theoretically possible, but would be much more difficult to verify in practice so they will not be discussed here. The predictions are:

  1. Under favorable circumstances when many niches are available for newly formed species, cascading speciation in a lineage may rapidly proliferate a large swarm of sister species; when otherwise comparable lineages in the same circumstances, but with more conservative genetic systems, form few if any new species. This is a direct result of the amplification process discussed on p. 28.
  2. If the cascading speciation is based on chromosomal semi-sterility, all species formed by the cascade will be chromosomally differentiated from one another by mutations which might be expected to reduce heterozygote fertility (although species which terminate sequences of derivation may carry apparently stable polymorphisms for these kinds of mutations). This follows from the basic revolutionary speciation model"-otherwise what function is served by the chromosomal differentiation?
  3. Species formed by a cascade will generally have parapatric distributions or may even be sympatric. Species formed originally in a parapatric situation are unlikely to immediately acquire allopatric distributions.
  4. Parapatric species formed by a cascade will be genetically isolated, even though no premating isolation may have been evolved and hybrids still show some degree of fertility. This follows from the analysis of the "sink effect" of parapatric hybridization (see p. 15).
  5. Sequences of karyotypic derivation will coincide with sequences of phyletic derivation, and these will be more linear than highly branched. This is obvious, and branching patterns are discussed on p. 29.
  6. Sequences of karyotypic derivation will frequently go to extremes in geologically short times (e.g., a completely acrocentric karyotype will be produced if fissions are the main mutation type, or all chromosomes will become bi-armed if fusions are the main mutation type). This follows from the amplification process.
  7. If there are gaps in a recent cascade due to extinction, the extinct species will likely have been formed late in the sequence, when species are expected to be rather ephemeral in comparison to either the terminal or relatively primitive species. This also follows from the amplification process (see p. 30).
  8. Sequences of derivation which have terminated for other than ecological reasons will end in species which either have used up the karyotypic substrate for chromosomal differentiation or are polymorphic for the kind of karyotypic differences fixed between species formed earlier in the sequence (see discussion p. 30).

All of these predicted relationships or conditions are logically derived from the revolutionary speciation model, and so far as I know, none of them would be expected from any other model of speciation or would be accounted for by any other cytogenetic mechanism of which I am aware. Therefore, if these conditions and relationships can be demonstrated in suitable natural radiations of species, the demonstrations would tend to verify the model.

However, for a radiation to provide useful tests, it should meet several criteria. It should contain many species and be recent enough that only few species formed by it are likely to have become extinct. The taxonomy of the species,in the radiation and their bio-geographic relationships should be very well known. Most importantly, a great deal must be known both about the comparative population cytogenetics and the phylogenetic relationships of the species. And, finally, there should be available similar information from proper control groups against which the patterns found in the test radiation can be compared. Needless to say, very few groups have been well studied in any of these respects, let alone all. However, as outlined in the introduction of the present report, Sceloporus--and especially the radiation of crevice-using Sceloporus--are well enough known in most of the required aspects to provide nearly ideal tests of most of the predictions of the cascading revolutionary speciation model. However, before turning to my analysis of the crevice-using radiation of Sceloporus, it will be useful to review some of the non-reptilian radiations which might provide good test cases if more were known about them.

The now classic case of chromosomal "speciation" in insects is provided by the radiation of the viatica group of morabine grasshoppers (White et al., 1964, 1967, 1969; White, 1968, in press). Although the radiation is cytogenetically well known and some studies of hybridization have been made, both in the field and in the lab, there is no consensus regarding the genetic relationships of the chromosomally different populations or the details of their phylogenetic derivations (e.g., see White, 1968; Key, 1968). Since these differences in interpretation can best be resolved by those with first-hand knowledge of the organisms and their biologies, I will restrict my review to some of the better known mammalian radiations, But, unfortunately, even here the data are scattered and incomplete. And, unlike the lizard radiation to be discussed below, there are usually no good comparative backgrounds available for the studies to be analyzed against.

The most detailed studies have all involved rodents, and the ones I will review have all been based on more than 100 specimens, though generally less than 500. In general, all rodent genera are karyotypically diverse, which correlates very closely with their limited vagility, as Arnason (1972) and White (in press) among others have noted. Therefore, there seem to be no really good rodent genera or groups which can serve as chromosomally conservative controls for the karyotypically variable stocks. However, among the better known groups and in terms of whole genera, possibly the most conservative of the large genera is Peromyscus, at least with respect to change in chromosome number. According to Hall and Kelson (1959), this genus contains approximately 40 continental species, although it is one of the ecologically most diverse groups-extending from the arctic tundra through all possible terrestrial habitats in North America almost to South America. Yet, among 20 species karyotyped (Hsu and Arrighi, 1968) only the 2n=52 nuttali, questionably placed in Peromyscus, deviates from a 2n=48. However, in other respects of the karyotype, Peromyscus is anything but chromosomally conservative, as the number of chromosome arms (i.e., NF) varies at least from NF=56 to NF=96 because of a great deal of interspecific and intrapopulational variation resulting from pericentric inversions (e.g., see Ohno et al., 1966; Sparkes and Arakaki, 1971; Te and Dawson, 1971; Bradshaw and Hsu, 1972; Lee et al., 1972). Presumably these polymorphisms are adaptive, as are the paracentric inversions of Drosophila, and may account for the great ecological plasticities of its species. On the other hand, Peromyscus is not notably speciose, considering its great ecological and geographic range.

An interesting comparison with the conservative 2n Peromyscus is provided by the ecologically much more restricted and more-or-less deserticolous Perognathus (Heteromyidae), which has probably more than 35 species concentrated in the xeric areas of North America (Hall and Kelson, 1959; Patton, 1969b). Patton has karyotyped 11 of the species recognized by Hall and Kelson and found 2n's fixed among them ranging from 34 to 56 as a result of Robertsonian rearrangements, with other differences due to inversions fixed between populations (Patton, 1967a, 1967b, 1969a, 1969b) . One species is polymorphic for inversions. The most striking karyotypic variation was found fixed between six parapatrically distributed cryptic species conventionally included in P. goldmani (Patton, 1969b). These cryptic species all differed by fixed centric fusions and/or inversions. Based on geographic relationships and minimal karyotypic derivations between species, Patton reconstructed a phylogeny in which the longest chain of derivation had five steps, with only the hypothetical ancestral karyotype not surviving in a present population. Two other one-step derivations trace from this ancestral stock, and a third one-step derivation branches from the second step of the chain of five species. The cryptic species are all now separated by ecological barriers (narrow flood plains inhabited by other Perognathus species) . However, these barriers are not absolute, as three chromosomally hybrid individuals were taken in a sample of 221 from near these contact zones. By comparison with Peromyscus, Perognathus is much more speciose in proportion to its ecological and geographic range. Furthermore, the cryptic speciation in the goldmani complex appears to have been recent and rapid, and conforms to the generally linear pattern predicted by the cascade model. The mole-rats. Spalax (Spalacidae) show a similar picture, although the genus contains only two or three taxonomically recognized species. Chromosome numbers in this genus range from 2n=48 to 2n=62 (Matthey, 1959; Walknowska, 1963; Soldatovic et al., 1966a, 1966b, 1967; Raicu et al., 1968; Wahrman et al., 1969a, 1969b; Lay and Nadler, 1972). Detailed collecting in Israel and on the Golan Heights revealed 2nts of 52, 54, 58, and 60 in parapatrically distributed populations of S. "ehrenbergi." Unlike Perognathus, there seem to be no ecological barriers separating the populations, although they do show a clinal distribution corresponding to annual precipitation (Wahrman et al., 1969a, 1969b; Lay and Nadler, 1972). Searches for contact zones revealed five hybrids between the 58 and 60 chromosome forms on Mt. Hermon, although these populations appear to be separated by the upper Jordan River at lower elevations. Two hybrids were found between the 52 and 54 chromosome forms, but none were found between the 52 and 58 chromosome forms, in spite of detailed searches in their contact zone. Nevo (1969) reports indications of behavioral isolation between the karyotypically different forms in laboratory crosses. On the other hand, Nevo and Shaw (1972) in an electrophoretic survey of 13 proteins controlled by 17 loci found no fixed genie differences between any of the chromosomal types, indicating a very close relationship between the populations and providing no test for barriers to gene flow between the populations. The lack of differences between populations may correspond to the generally low variability within populations. Wahrman et al., (1969a, 1969b) suggest that Spalax with a low 2n radiated into Israel from a center to the north or east which had a mesic climate, and that higher chromosome numbers were then fixed in a linear sequence as the animals encountered increasingly xeric conditions. Lay and Nadler (1972) accept this sequence of derivation and attempt to date the separation between the Israeli 2n=60 stock and a now disjunct but apparently related 2n=60 stock which has spread along the Mediterranean coast as far as central Libya, Assuming that the 2n=60 stock is actually the most recently derived, Lay and Nadler suggest that the two disjunct populations were separated for at least 10,000-25,000 years and that the sequence of karyotypic derivation must therefore have been older than that separation. However, at least until the Jordanian, Syrian, and Lebanese populations are sampled, it seems that the alternative possibility, i.e., that the 2n=60 population is primitive in the Israeli radiation, cannot be ruled out. However, in either case, the Robertsonian sequence is essentially linear and, except for the possible extinction of the 2n=56 population (if it is not found on the east side of the Jordan River--possibly below the Sea of Galilee), the parapatrically contacting populations can be traced in a linear sequence. This is of course fully consistent with the cascade model, and, if the 2n=56 population is extinct, this could easily be due to an acceleration of chromosomal differentiation during the cascade.

Probably the most striking example which might be ascribed to the acceleration effect of cascading speciation is the karyotypic variation among the cotton rats (Sigmodon: Cricetidae). Until the revision of Zimmerman (1970), based largely on karyology, the genus was thought to contain only five species: hispidus, ranging from South America through all of mainland Mexico and throughout the southern United States; ochrognathus, found in the SW United States and NW Mexico; leucotis, found on the Mexican plateau; alleni, found in southern Mexico; and fulviventer, found in the Sierra Madre Occidental of Mexico and extending northward into Arizona and New Mexico. Of these five taxa, three are karyotypically monomorphic and have identical 2n's: ochrognathus, 2n=52, FN=66; alleni, 2n=52, FN=64; and leucotis, 2ns=52, FN=52 (Zimmerman, 1970). The other two taxa are karyotypically variable: fulviventer, with an FN of 34, shows a Robertsonian polymorphism for a 2n range of 28 to 30 (Lee and Zimmerman, 1959); while different hispidus populations show fixed karyotypes of 2n==52, F=52 (identical to leucotis); 2n=28, FN=28; 2n=24, FN=38; and 2n=22, FN=38 (Zimmerman and Lee, 1968; Zimmerman, 1970). Based on these striking differences among hispidus, Zimmerman (1970) applied the names mascotensis to the 2n=28 population, found in the Mexican Pacific coastal states of Jalisco through Oaxaca, and arizonae to 2n=24 and 2n=22 populations found west of the Continental Divide from Jalisco through central Arizona. The revised hispidus, then, has a range from South America to the southern United States to the east of the mascotensis and arizonae populations. Karyotypic samples of hispidus, sensu stricto were from localities as far separated as Panama, Florida, North Carolina, Kansas and SE Arizona and showed essentially no variation, Based on this wide distribution, and the fact that three other species have the same 2n, one of which also has the same FN, there is little doubt that the 2n=52 pattern is primitive in the hispidus complex. Zimmerman (1970) would then derive all of the reduced 2n karyotypes in Sigmodon from this 52 chromosome pattern in a single sequence of derivation, more or less in the sequence fulviventer (2n==30-28), mascotensis (2n=28), arizonae (2n=24), arizonae (2n=22). However, if the cascading speciation model is correct, the polymorphism of fulviventer should be a chain termination situation, and this species should not have given rise to any further derivatives. In this respect it is significant that Johnson et al. (in press), in an electrophoretic study of protein variation in Sigmodon, provide convincing evidence that fulviventer has evolved quite independently of the hispidus complex derivation. Presumably, then, two independent sequences of derivation are involved, one from a 2n==52 ancestor to terminate in the polymorphism of fulviventer, which has apparently left no surviving intermediates; and the second within the hispidus complex, again which has left few intermediates until low 2n's were evolved and presumably the possibilities for further reduction in 2n's were considerably reduced. The lack of intermediate forms suggests that these populations either never grew very large and were completely displaced by the most highly derived end products, which is of course a prediction of the cascading speciation model.

Many other mammalian radiations could be analyzed in a similar manner, but all suffer from the same defects found in the cases discussed above:

  1. sample sizes are generally too small and poorly distributed to allow one to safely assume that all of the chromosomal variation within the group has been detected;

  2. in most cases the degree of isolation between the populations has not been determined--I have assumed for the discussion above (as have most other workers) that the chromosomally differentiated allopatric populations are in fact good species, but this has rarely been proven;

  3. very few of the workers are adequately trained in both cytogenetics and systematics;

  4. only very rarely have attempts been made to derive phylogenetic relationships independently of the karyotypic relationships; and

  5. mammals as a group are so chromosomally diverse that there appear to be no radiations which provide otherwise closely comparable groups which can serve as a natural experiment in chromosomal variability versus chromosomal conservatism (at least not to someone who is not an expert in the systematics of the groups concerned).

However, as will be seen below, the radiation of Sceloporus suffers from few of these defects. Although, before this is described, there remains one extension of the speciation model to be discussed

Karyotypes as a Phyletic Tool 

If fixations of chromosomal rearrangements which potentially induce partial sterility when heterozygous are in fact closely involved in some speciation processes, then analysis of this type of interspecific chromosomal variation in a highly variable group such as Sceloporus can provide a potentially extremely powerful indication of phylogenetic relationships. If a primitive karyotypic configuration can be determined for the variable group, and sequences of karyotypic derivation can be traced within it, these sequences can be used to trace sequences of phylogenetic derivation of the species characterized by the different karyotypes. One would predict, in general, that chromosomally identical species have usually speciated according to the more conservative allopatric mode, while, on the other hand, observed karyotypic differences that were not adaptive as balanced or transient polymorphisms are likely to have been fixed during episodes of revolutionary speciation which were aided by their fixation. Furthermore, the karyotypically derived population will have been the one to undergo the genetic revolution, and because the incipient post-revolution species will have been numerically small in comparison to the parental species, in any early competition between the two, the incipient species will be ecologically and behaviorally displaced with respect to the parental species. Therefore, the sequence of karyotypic derivation may serve as a guide to the evolution of the species from primitive to advanced or specialized conditions. Then, karyotypically conservative species may be expected to be more like the primitive stock from which the radiation began than will karyotypically highly derived species. Also, if genetic revolutions in a sequence of karyotypic derivation have played important roles in the evolutionary differentiations of the species in the sequence, then the species' place in the progression of karyotypic derivation probably also will indicate its place in the sequence of phylogenetic derivation, 

The karyotypic variability of Sceloporus provides an excellent test of the validity of this approach. Will an analysis of the karyotypic variation in Sceloporus according to the principles developed above provide useful and hopefully unsuspected insights for reconstructing the phylogenetic history of the genus, which can then be tested using independent lines of evidence? If it does, then we may begin to have some faith in the principles used in the reconstruction. However, the genus Sceloporus is far too large and far too much karyotypic data are available for it to allow the whole genus to be treated in this manner in one paper. Therefore, in the present report the treatment will be restricted to Smith's (1939) megalepidurus, grammicus, and torquatus (= poinsettii) species groups, plus the species asper (currently placed in the formosus group) and clarkii and melanorhinus (currently placed in the spinosus group), which I believe and will attempt to show below form a complete natural grouping within Sceloporus.


Methods and Materials

Contents