Introductory comments
Conservative geographic speciation
Revolutionary speciation
Cascading revolutionary speciation
Geography of revolutionary speciation
In a future work of this series I will attempt a detailed comparative study of speciation in the whole sceloporine radiation to test the revolutionary speciation hypothesis presented in the beginning of the present report. However, now that a phylogeny of the crevice-using and clarkii group radiation has been synthesized, a preliminary analysis of speciation patterns in it may be made. In this analysis one should note that the conclusions drawn from it are no better than the phylogeny on which it is based. However, if the phylogeny is realistic and the chromosomal speciation hypothesis valid, one should find several distinct phenomena in this radiation:
All three of these phenomena seem to have occurred in the radiation of the clarkii group and the crevice-users. However, before they are examined, the biological background against which the speciation of the radiation occurred should be reviewed, as this speciation did not take place in a vacuum. At the beginning of the radiation, there were a host of related and unrelated lizard species competing for xeric and subxeric habitats along the NW coast of Mexico. Minimally there must have been at least one small-sized, small-scaled Sceloporus (i.e., the nelsoni prototype), at least one Uta, at least one Urosaurus, at least one sand lizard (Callisaurus, Holbrookia, or Uma), probably a Petrosaurus, and at least one Phrynosoma among the sceloporines; and a Crotaphytus or two, probably at least three iguaninines, and possibly some Anolis (in less xeric areas) among the other Iguanidae; not to mention several representatives of at least two other lizard families. Later, as the present lineage radiated, several other Sceloporus species besides those resulting from the radiation would be added to the fauna of competing lizards. In such a complex community, one might expect all Sceloporus to evolve rather sharp niche specializations, as have Anolis species in complex faunas. However, many Sceloporus, such as clarkii for example, still show surprisingly great ecological plasticity, perhaps because they have evolved especially good "general" adaptations that allow them to function in a variety of habitats. This background should be kept in mind in the discussion to follow.
There are several clear examples of conservative geographic speciation in the present radiation. At least five species pairs appear to have formed across the lower Rio Grande de Santiago. This river, though not impressive during the present, comparatively dry climatic regime, drains almost all of the high plateau enclosed by the Sierra Madre Oriental, the Sierra Zacatecas, the Sierra Volcanica Transversal, and the southern half of the Sierra Madre Occidental. Although the lower reach of this drainage now meanders gently across the short stretch of coastal plane, it must have been an important barrier during Pleistocene pluvial periods. Furthermore, for species distributed at higher elevations along the western fronts of the Sierras, the deep, hot barrancas of the Rio Santiago and its major tributaries would add major ecological barriers to the river itself. The river approximately divides at least five pairs of sister species: neIsoni/pyrocephalus, clarkii/melanorhinus, possibly the primitive rock-crevice-user/primitive plant-crevice-user, asper/shannorum, shannorum/heterolepis, and possibly jarrovi/dugesii. To review these cases, both pyrocephalus and melanorhinus are karyotypically derived with respect to their sister species, but in melanorhinus the difference is due to the spread of a Y-autosomal fusion which clearly could not have had a net adverse effect on fitness or it would have spread through the population. In pyrocephalus, the fixed inversion difference which differentiates it from nelsoni may have been adaptive as a balanced or transient polymorphism, as indicated by Cole's (1970) report of a similar inversion polymorphism in clarkii. In many organisms such inversion heterozygosity does not appear to reduce fertility, apparently because chiasma formation in inhibited in the mutually inverted region. The remaining species pairs separated by the Rio Santiago are chromosomally identical, and except for the important divergence between the primitive rock- and plant-crevice-users, the ecological differences of the species pairs are slight and clearly the result of climatic differences in their respective ranges. Also, even the early divergence between the rock- and plant-crevice-users can be explained by climatic differences. North of the Rio Santiago, the slopes of the Sierra Madre Occidental are much more xeric than are the slopes of the Sierra Volcanica Transversal to the south of the river. Presumably rock crevices would be the most common cover on the xeric mountainsides then, as it is now (pers. obs.), while tree holes and crevices would be more available in the humid habitats to the south. Any specializations for the different crevice types would be enhanced when the sister species again became sympatric.
Other conservatively evolved species pairs appear to have been formed by the divide of the Sierra Madre Oriental. These divisions probably resulted from separations during Pleistocene cold periods. Examples are: shannorum/standard grammicus and possibly jarrovii/bulleri. All speciation in the torquatus group appears to be conservative. Once the primitive torquatus became specialized for the exclusive use of rock crevices, populations could survive only in areas of rock outcropping. On the highlands of the Mexican Plateau, there are many rocky mountains surrounded by "seas" of alluvial soil. Although colonists can probably cross these seas easily enough to found new populations, the "seas" should provide enough isolation to allow abundant opportunities for conservative speciation, with or without a genetic revolution in the founder population. And in this respect, it is worth noting that, excepting the cases of sympatry which prove that the overlapping populations are good species, the present species taxonomy of the torquatus group is arbitrary to a high degree. Each mountain range rising above the alluvial floor has its own torquatus group populations which can usually be taxonomically distinguished from those of other mountains if someone wishes to look at enough characters.
The origin and speciation of the megalepidurus group are less clear. First, although megalepidurus and pictus are treated as full species in the present work, samples collected in the area where they meet are morphologically intermediate. This problem is still under study, but it seems likely that these two "species" are only the geographic extremes of a single intergrading population. Secondly, it is difficult to determine where and when the ancestral megalepidurus originated from the standard grammicus that probably gave rise to it, because standard grammicus. have probably displaced megalepidurus from a good part of its former range.
Likely examples of revolutionary speciation involving chromosomal differentiation are shown by the divergence of the 2n=32 neo XY stock [now assumed to be extinct] from the primitive 2n=34 lineage (assuming that the sex chromosomal modification does not account for the reduction in 2n) and by the split between the 2n=40 clarkii group and the 2n=32 neo XY stock. Then, of course, the grammicus complex seems to provide an ideal example of cascading speciation. However, the two older cases must be mentioned first. Of these, the 2n=34 XY/2n=32 neo XY speciation event is so old that it has been greatly obscured by subsequent speciations and extinctions, and therefore analysis of it would not be profitable; but the 2n=32 neo XY/clarkii speciation seems less obscure, though it must also be rather old. If the 2n=32 neo XY stock was basically ground and rock dwelling (see page 119), then the shift of clarkii up into the trees represents a major ecological shift which might be associated with its chromosomal derivation and revolutionary speciation. However, if the ecologies of the present day clarkii and the crevice-using derivatives of the 2n=32 stock are any indication, the primitive clarkii was sufficiently plastic in its ecology to exert a continual pressure on the more conservative ground dwelling 2n=32 XY stock. This pressure, " and other competitive pressures from a series of species in the more f xeric area north of the Rio Grande de Santiago would tend to force the still ground dwelling 2n=32 XY stock up the slopes of the Sierra Madre, where it evolved the live-bearing and crevice-using habits and the Y-autosome fusion spread throughout the primitive crevice-using stock.
It should be noted that, in the present approximately parapatric distributions of the clarkii and crevice-using stocks along the whole length of the Sierra Madre, there is no evidence for a past geographic separation of the stocks. However, it should be recalled that the derivation of the 2n=40 clarkii lineage from the 2n=32 ancestral stock must have been older than separations of the species pairs discussed in the preceding section. From this we may conclude only that the derivation of clarkii is too old to provide any evidence for or against speciation involving the geographic isolation of the primitive and derived stocks, As will be seen in the following section, the evidence from grammicus is considerably clearer, presumably because the speciation in this group is much more recent than that which gave rise to clarkii.
From this distance in time, and since no intermediates survive, it is impossible to say whether the four fissions fixed in the clarkii group karyotypes were fixed simultaneously in a single event or in a sequence of speciation events. On the other hand, there is no doubt that a sequence of karyotypes exists in grammicus, and that in at least some instances the karyotypically different populations are good species. Furthermore, the karyotypic sequence in grammicus has most of the characteristics predicted by the cascading speciation model. And, in comparison to the obvious geographic differentiation of the torquatus group species which live on comparative isolated and geologically stable rock "islands," the grammicus live in the much more continuous vegetational phase of the environment.
Within this vegetational phase, at least standard and F5+6 grammicus seem plastic enough in their physiological tolerances that they can live anywhere from desert to rain forest, where they can find suitably large plants or plant products which have crevices that are not occupied by other grammicus races. And even areas that are largely uninhabitable by grammicus because they contain no large plants may still be crossed along water courses: for example, grammicus have been taken from trees along the Rio Nazas near the Coahuila-Durango border, which is well out in the otherwise completely uninhabitable area of the Chihuahuan Desert (Bogert, 1949). From this it should be obvious that grammicus will have had much less opportunity for conservative speciation than the torquatus have had. And where apparently conservative speciation has occurred in the grammicus group, i.e., between the pairs shannorum/heterolepis and shannorum/standard grammicus, this speciation seems to have been associated with obvious geographic barriers and is not associated with chromosomal differentiation. On the other hand, there are no obvious geographic barriers associated with the speciation in the grammicus complex (which does not prove that such did not exist" there is just no evidence for them) . And, most significantly, it is precisely these grammicus populations, and not the populations of the torquatus group, etc., which are differentiated by the chromosomal mutations likely to reduce heterozygote fertility.
The ecological shifts predicted by the revolutionary speciation model have not occurred in the grammicus complex. However, any major ecological differentiation between species would seem to be precluded because all of the ecologically adjacent habitats seem to be fully occupied by other Sceloporus or other lizards. On the other hand, the grammicus complex species exclude one another geographically, which is the alternative prediction of the hypothesis.
The phyletic structure of the species cascade in grammicus is adequately described in the section on the phylogeny of karyotype evolution (page ) and need not be described again here. It should be sufficient to note that it is the terminal FM populations which are highly polymorphic for fissions, and that the polymorphic P1 may also represent the end of a sequence (S → F6 → Pl). Also, it is pertinent that these polymorphisms involve only the largest and most symmetrical chromosomes, which might not function well in speciation because assortment from their trivalents would be most nearly regular (White, 1963) and because their malassortment products may be gametically rather than zygotically lethal. In other words, in the FM populations the substrate for further speciation by fissioning has essentially been exhausted; and in the P1 population, the chromosome involved in the polymorphism is the one least likely to be successful in initiating further speciation, and therefore most likely to be involved in an aborted speciation event. These situations are, of course, completely consistent with the cascading speciation model. Another coincidence with the cascade model is the failure to find populations karyotypically intermediate between F5+6 and FM, which may indicate that these intermediates had little chance to spread before they, in turn, spawned chromosomally more derived species by the revolutionary mode. These missing intermediates then suggest that the rate of speciation may have accelerated as the sequence of derivation progressed. Finally, although I have treated them as one species above--on only flimsy evidence--it is entirely possible that the coastal and plateau F5+6 populations independently differentiated from an F6 ancestry. This would provide another short branch to the cascade; but, even so, the cascade is much more linear than branched. In short, where relevant data are available, the observations are fully consistent with the cascading speciation model and cannot be readily explained by anything other than some kind of * chromosomal speciation model.
As discussed in the previous sections, there are three somewhat different and not necessarily exclusive geographic circumstances under which chromosomal speciation may occur: one which supposes that speciation can occur only on the geographic periphery of the parental stock if it has a continuous distribution within that range (the "peripheral version"), one which supposes that speciation can occur anywhere within the normal range of a parental stock whose population is subdivided into effectively very small semi-isolated breeding pools (the "interstitial version"), and an intermediate one which supposes that a species may be formed on one of the many "internal peripheries" of a parental stock which is restricted to habitats which show mosaic distributions within the broad geographic range of the population. Of the three, the interstitial speciation would seem to deviate roost extremely from classical allopatric models.
At least in some parts of its range, grammicus seem to have ideal population structures for interstitial speciation. In areas where P1 and F6 occur on the eastern divide of the Valley of Mexico, most individuals live on dead trees and fallen logs, and only a few grammicus are found on live trees or near other classes of cover. Based on many anecdotal observations (Moody et al., in prep.), where the logs are widely scattered (as they are in many places), these logs would seem to serve as fairly well isolated "islands" until they decay, which to judge from our observations over a two-year period, probably takes ten to twenty years. That is not to say that no grammicus disperse past these logs, but only that the populations on the logs will probably remain inbred because dispersing individuals will find it difficult to establish residency on already inhabited logs. Although grammicus appear to be territorial when living in areas with dispersed cover (e.g., in Agave or on Yucca), when five or ten adults live on a log they appear to shift into a social hierarchy with despotic males, as do other iguanids when exploiting concentrated resources (Hunsaker and Burrage, 1969). Since most grammicus found in habitats unsuitable for overwintering have been young-of-the-year, it is likely that these young do most of the dispersing. Since adults actively chase young, once a juvenile is chased away from its home log--where it was probably familiar with cover and other resources--it seems unlikely that it could establish residency on another well-populated log where the cover and other resources would be initially unfamiliar. Presumably, then, once a newly fallen log is colonized, the population on it will be fairly effectively isolated by its communal defense of the log until it decayed to the point where it was no longer habitable. Since the maximum populations of these logs range from five to usually no more than 15 adults plus their young, it would seem that grammicus living in this type of habitat have an ideal population structure for interstitial speciation. A mark-and-observation analysis of dispersal and population structure on the east side of the Valley of Mexico was begun to support the anecdotal observations and proved to be completely feasible, but the time demands for other aspects of this study were too great for the analysis to be very useful. It is hoped that an effective study of this nature can be completed in the future.
Since the basic population structure of grammicus seems ideally suited for interstitial revolutionary speciation, the grammicus radiation might serve to test whether the speciation is strictly a phenomenon of the species border or whether it actually can occur interstitially. If such speciation were recent enough, a derived population should be completely surrounded by its ancestral stock. In this situation, the derived population will have either excluded its ancestral stock from that part of its former range, or the derived stock will have diverged enough to live sympatrically with the ancestral stock. However, for ecological differentiation to occur, an adjacent niche must have been available to the differentiating species; and, as we have seen above, the niches adjacent to grammicus are probably sufficiently saturated with other Sceloporus to completely preclude this possibility. Therefore, the only test would be to find encircled populations that exclude the ancestral stock from areas that it would otherwise inhabit. If Pi derives from F6, which is at best a questionable possibility, this condition would be met (excluding the possibility that speciation could have occurred along the sharp edge of the tree-line, which would function much like a continental margin, etc.). But, even here. Hall and Selander (1973) were able to propose reasonable allopatric models to explain the speciation (but, on the other hand, the allopatric models do not account for the chromosomal differentiation between the various grammicus). The other grammicus races are either too poorly mapped or appear to have spread to the geographical limits of the climatic or ecological zones within which they can exclude other grammicus from the plant crevice niche, and therefore do not provide unequivocal tests either. However, it does seem significant that all of the derived populations except F5, and perhaps the coastal F5+6, occur well within what must have been the historical geographic and ecological range of standard grammicus. This suggests that differentiation occurred either on an internal periphery or was strictly interstitial. Tending to support the interstitial alternative is the great ecological latitude shown by standard grammicus in the southern part of its range, where an apparently continuous population occurs in all wood- and plant-crevice habitats from rain forests (over a minimum altitudinal range of from 1500 m to 3600 m-confirmed by karyotyped specimens) to Yucca and Agave in the most xeric parts of the southern (and northern) plateau. All derived grammicus occur well within the ecological range inhabited by standard populations. But, to take the contrary view again, it is impossible to estimate the importance of the major climatic fluctuations of the Pleistocene in the speciation of grammicus.
So, to summarize this section, while the data presented do not prove that speciation in grammicus has occurred interstitially, they do suggest the possibility. And, as is true for most evolutionary questions, it is unlikely that field data will ever provide an absolutely unequivocal proof of the possibility. However, sufficiently detailed studies of enough other radiations comparable to that of grammicus may eventually provide enough evidence to be convincing. Furthermore, many aspects of the speciation model should be susceptible to modeling by computer simulation, which would allow values to be estimated which would be required of various parameters of population structure and genetic system before interstitial or other kinds of chromosomal speciation could become likely. Once these values have been calculated, it should then be relatively simple to go into the field to see whether natural populations conform to them. Much work can be done in this area.