[Summary and Contents]




To account for the proliferation of species in chromosomally variable lineages of Sceloporus, I proposed a speciation model (Hall, 1973) which resembles White's "stasipatric" model (White, et al., 1967; White, 1968, 1969; Key, 1968; Patton, 1969), which Bush generalized (Bush, 1975; Wilson, et al., 1975). However, my model for Sceloporus differs significantly from those of White and Bush in three areas: 

  1. The chromosomal differentiation of a local population and the initiation of chromosomal speciation may occur anywhere within a species' range where conditions are favorable, rather than only on its,geographic periphery as supposed by Key (1968), Patton (1969), and Bush (1975). 
  2. Detailed studies of hybridization between the chromosomally differentiated Sceloporus grammicus populations (Hall and Selander, 1973; Hall, Stamm, and Reichlin, unpub.) indicate that parapatric hybridization plays a very significant role in the completion of genetic isolation which is quite different from that envisioned by Bush (1975) and the others. 
  3. I find that certain genetic effects inherent in the mechanics of chromosomal speciation imply some rather interesting consequences relating to patterns of speciation, and evolutionary changes in the genetic systems themselves, which quite adequately account for the associations of chromosomal diversity and phyletic evolution without any requirement to suppose any direct role of the chromosome mutation in altering developmental pathways. 

The arguments for points 1 and 3 follow fairly directly from initial assumptions of the basic chromosomal speciation model and will be discussed first. 

However, the function of hybridization in chromosomal speciation, as I understand it from my studies of the grammicus hybrid zones, and presumably in any case of contact hybridization (Littlejohn, et al., 1971) where hybrids meet conditions of reduced hybrid fitness and low vagility, is so paradoxical that it must be discussed in detail. Such zones of contact hybridization (parapatric or narrow allopatric hybridization in  Woodruff's 1973 terminology) paradoxically appear to be as effective as complete geographic separation in genetically insulating one population from another, in spite of the fact that hybrids and presumably also backcrosses are at least partially fertile. Some of the evidence and the detailed model to support this assertion will be presented after arguments 1 and 3 of the chromosomal speciation model are discussed immediately below. 

Chromosomal differentiation; initial conditions

The process of chromosomal speciation (if it is assumed to be different from allopatric speciation--Mayr, 1963) is initiated by the fixation of a chromosomal rearrangement in some local population which is not completely or permanently isolated from its parental species. It is also assumed that meiotic problems in chromosomally heterozygous individuals somehow result in at least a partial barrier to gene flow between the two kinds of chromosomally homozygous individuals, which will then favor the evolution of completed isolating mechanisms even in the absence of geographic isolation (Spurway, 1953; Spurway and Callan, 1960; Lewis, 1953, 1966; Lewis and Raven, 1958; Wallace, 1959; John and Lewis, 1965, 1966; White, 1968, 1969; Mayr, 1970; Bush, 1975; Wilson, et al., 1975). To make this more explicit, if the conditions described below are met by the genetic system of an ancestral species or population, then over evolutionary time, occasional local populations will become fixed by drift for chromosomal mutations which would serve as intrinsic mechanisms to partially isolate the mutated populations from genes of the ancestral species.

  1. To provide a mechanism for the partial isolation, heterozygosity for some kinds of chromosomal rearrangements in meiosis should induce a relatively high frequency of meiotic malassortment, which would then lead to the production of an appreciable fraction of zygotically lethal aneuploid gametes. A 10 to 50% aneuploidy would seem to be reasonable in a chromosomal speciation model, pending tests of this variable in computer simulation models. Malassortment in this frequency range has been shown in heterozygotes for single centric fusions in mice (Cattanach and Moseley, 1973; Gropp, et al., 1972, 1974; Ford and Evans, 1973) and cattle (Gustavsson, 1971a, 1971b), and for inversions, translocations, and D/G fusions in ) humans (Polani, et al., 1965; Hamerton, 1971). 
  2. Such rearrangements, when they occur should have little or no negative effects on developmental fintess in heterozygous balanced or homozygous individuals or on reproductive fitness when they are homozygous. 
  3. The mutation rate for such rearrangements should be appreciable (say more than one per 10,000 gametes). Mutation rates of this magnitude are seen in nature (White, 1973; Hamerton, 1971). 
  4. The population structure and mating system should be such that many demes in the ancestral species have effective population sizes of 10 or less. Wright (1941) showed with a path coefficient calculation that even mutations which reduced the fertility of heterozygotes by as much as 50% still had about one chance in 1000 of achieving fixation by drift (random sampling error) in a population with an effective size of 10. In still smaller populations or where heterozygosity reduced fitness by lesser amounts, the probability of fixation would be higher. 
  5. The interdeme vagility should be low enough so that demes infrequently exchange individuals; or if such exchange is frequent, then the species' mating system and social structure should be such that immigrants only infrequently are able to reproduce in the deme (Wilson, et al., 1975). This assumes that interdeme gene flow is no more than 5-10% per generation. Since selection against the heterozygote in a polymorphic population would work against drift to selectively eliminate whichever of the chromosomal arrangements is rarest in the deme, low rates of ancestral chromosome injection from surrounding populations should not greatly reduce the probability of the initial chance fixation. Pending tests in a computer simulation model, 10% gene flow per generation should have less than an order of magnitude effect on the probability of fixation.

 These assumptions are at least tacitly accepted by most published chromosomal speciation models. 

One way I differ from the Key (1968), Patton (1969), Bush (1975) version of the stasipatric model is in the probable location of the chromosomal differentiation which initiates a speciation event. These authors all believe that this differentiation can occur only on a species' geographic periphery. As implied by Wilson, et al. (1975), fixation of the mutation can occur anywhere within the total range of the parent species where the population and cytogenetic parameters allow it, and not just on the species' periphery (Hall, 1973). Note that Arnason (1971), White (1973), Bush (1975), and Wilson,et al. (1975), among others, emphasize that comparative data from all radiations reviewed show the most evidence for chromosomal speciation in those lineages which have limited vagility and subdivided populations, just where suitable conditions of population subdivision and small deme size would not be restricted to the geographic periphery. Also, while the initial fixation of a rearrangement depends only on the genetic system parameters noted above, it should not be forgotten that survival of the mutant population, at least during the cr critical early stages of the speciation process, when it would be at its smalles, is completely at the mercy of the local environment. At the species' periphery, early extermination is likely to result from only slightly more than normal environmental fluctuation; while more central populations are likely to be relatively long-lived, at least with respect to their local environment, because they presumably are better adapted to it. Also, assuming the favorable population structure mentioned above, there will be many more "central" than peripheral populations suitable for the initial differentiation. Although Key (1968) debates this. White, et al. (1967) believed that several of the chromosomally differentiated morabine grasshopper species formed within the range of the ancestral species. 

Cascading speciation

Assuming that we have a reasonable model for completing the genetic isolation of the already partially isolated, chromosoraally differentiated founder population of the nascent species (see below), the sampling effect inherent in the speciation process itself may serve as a mechanism to amplify the frequency of genes favorable to further chromosomal speciation in the chromosomally derived species. This may lead to a burst of rapid speciation which I call a cascade (Hall, 1973). 

Many variable gene loci concerned with the parameters of a species' genetic system may have minimal effects on individual fitness, but may have alleles which profoundly influence the probability of chromosomal speciation. Variation at any or all of these loci may be affected by the amplification process of chromosomal speciation.

An example of such a locus might be one whose alleles affected the frequency of chromosomal mutation (e.g. Ives, 1950; Kidwell, et al., 1973; Voelker, 1974). Consider 2 alleles; a, which produces an efficient repair enzyme which allows only one rearrangement per 100,000 gametes; and b, a slightly defective enzyme which allows rearrangements at a rate of one per 1,000 gametes. Both frequencies are probably within the range of natural mutation .rates for chromosomal rearrangements. Then, if all other aspects of a genetic system which is otherwise favorable for chromosomal speciation are held constant, a population fixed for allele b would initiate chromosomal speciation events 100 times as frequently as would one fixed for allele a. In regard to this kind of difference, even though the dogma of population genetics states that natural selection will always work to minimize mutation rates because most mutations are deleterious, note that selection will have minimal effects on this kind of polymorphism. Certainly the heterozygously semisterilizing chromosomal rearrangements that we are considering here are deleterious to the individuals which are heterozygous for them, but even assuming that all of the excess rearrangements which allele b allowed were zygotic lethals, given its "penetrance" of only one per 1000 gametes, it would take many thousands of generations of selection to appreciably change its frequency in a polymorphic population. Another kind of locus which would have minimal effects on individual fitness, but which could have similarly profound effects on the probability of speciation would be one which influenced the frequency of meiotic malassortment in chromosomally heterozygous individuals. Selection could affect this locus only by working on the very rare chromosomally heterozygous individuals (Ford and Evans, 1973).

If a species has the subdivided population structure and small deme sizes required for fixation of a heterozygously semisterilizing chromosome mutatation to be plausible (Wilson, et al., 1975), then gene and genotype frequencies for any polymorphism in individual denies would be expected to drift away from average values expected for the species as a whole. In an ancestral species with polymorphisms like those discussed above, which would provide favorable cytological conditions for chromosomal speciation without significantly affecting individual fitness, many domes would become fixed for alleles that favored chromosomal speciation, even if they were rare in the population as a whole (Wright, 1940, 1941, 1951; Kimura, 1970) . In other words, drifting frequencies at several loci affecting speciation probabilities could well generate an extremely high inter-deme variability for the probability of chromosomal speciation without having perceptable effects on usual parameters of individual fitness. It should then be obvious that chromosomal speciation will most likely occur first in those domes where it is most probable as determined by drifting genetic system parameters. 

Consequently, assuming that a chromosomally derived species will inherit the bulk of its genes from the original chromosomally differentiated deme, and noting the minimal effects of alleles "favorable" for speciation on individual fitness, the chromosomally derived species should perpetuate for long periods the initial homozygosity or high frequency of the "favorable" alleles which facilitated its origin. Because of this selective founder effect, on the average the frequency of such favorable alleles would be higher (possibly much higher) in the derived species than in the ancestral species. Therefore, the chromosomally derived species would be more likely to initiate further (or second, level, L2) derivations than would be the ancestral (or L0) species. And, of course, the same amplification process might be repeated when the L2 species was formed so that this L2 species would be more likely to form an L3 species than either the L0 or L1 would be to form a second species. This process would continue to concentrate or amplify the frequency of favorable genes for several generations of increasingly rapid speciation to form a cascade of species in a manner analogous to what happens in some types of multistage electronic (or "cascade") amplifiers. 

Cascade termination

On the other hand, several processes would work to counter or to halt a cascade of chromosomal speciation once it began. 

  1. Termination by substrate exhaustion. Robertsonian mutations appear to be frequently involved in the speciation process (Wallace, 1959; White, 1973; Wilson, et al., 1975; Paull, et al., 1976). The cascading process might be expected to amplify classes of favorable genes which promoted or favored the involvement of only one kind of Robertsonian mutation in the speciation process. Although there has long been considerable debate on the subject (John and Hewitt, 1968; Todd, 1970; White, 1973), it seems probable that centric fissions and centric fusions are generated by rather different kinds of mutational processes: Fissions may be produced by single breaks or "misdivisions" of the centromere (Darlington, 1939, 1940; Lima de Faria, 1956; Marks, 1957; Lewis and John, 1963; Kato, et al., 1973), while centric fusions more probably involve a reciprocal translocation process (Jackson, 1971; White, 1973). Genes which increased mutation rates and altered meiotic systems to favor speciation by one of these processes probably would not favor the other process. Therefore, Robertsonian cascades would usually involve only one or the other kind of mutation, and would be terminated when the available chromosomal substrate for that kind of mutation was used up to result in a terminal species either with all acrocentric or with all metacentric chromosomes. Possibly similar arguments nay be developed for othr classes of rearrangements, although it is not so obvious how cascades involving them might be terminated by exhaustion of the available chromosomal substrate.
  2. Termination by counter-selection. The cascading amplification process might amplify and fix the effects of certain classes of "favorable" genes at such high levels that their negative effects on individual fitness would then become great enough to selectively require the other components of the genetic system to evolve coadaptations to counteract the effects of the "favorable" genes on individual fitness. An example of a probable mechanism for such a cascade termination situation would involve interactions between mutation rate loci and loci controlling the mechanics of their meiotic assortment. Chromosomal speciation presumably depends on the fitness reducing effects of meiotic malassortment in chromosomal heterozygotes. As long as mutation rates are low enough so that chromosomal heterozygosity remains rare in the population, there will be little opportunity for directional selection as compared to drift to alter the mechanics of meiotic assortment in such heterozygotes. On the other hand, if the cascading amplification process fixes too many alleles favoring increased mutation rates, a situation may evolve where chromosomal heterozygosity becomes frequent. Here, selection would certainly favor the evolution of a spindle apparatus which would insure balanced assortment even from chromosomal heterozygotes. In the case of Robertsonian mutations, this is easily achieved by insuring 
    1. that chiasm formation is sufficiently regular to reliably link the three chromosomes (the metacentric, plus the two acro- or telocentric arms) into a trivlanet, and 
    2. that the centromeres of the trivalent always orient alternately (see diagrams in Hamerton, 1971).

     Although there are no relevant comparative studies of mutation rates in chromosomally variable radiations, it is quite possible that some of the apparently neutral Robertsonian polymorphisms in Sceloporus grammicus  discussed above, in the Mus subgenus Leggada (Matthey, 1966; Jotterand, 1972), and in Sorex (Meylan, 1964) [represent such chain terminations]. 

  3. Termination by niche saturation. If nothing else halted a cascade, available and accessible ecological niches which could be occupied by a radiation would eventually become sufficiently saturated to prevent further successful speciation. 


[Summary and Contents]