16 

Evolutionary Reversions: The Course of Evolution Is Not Unidirectional

Sometimes, when ancestral conditions of living reccur, metazoans are capable to revolve lost ancestral phenotypic (behavioral, morphological, physiological and life history) characters.

The extreme complexity of the processes of integration of new structures in metazoan systems on the one hand, and the negligible probability of the occurrence of “reverse mutations”, made biologists to intuitively believe that reversion to ancestral structures, especially after long evolutionary periods of millions of years, when many genes have changed, is impossible.

The neoDarwinian tenet that lost ancestral  phenotypes are irreversible was clearly not Darwinian. Idea of reversion to ancestral states as a mode of metazoan evolution was embraced by Darwin, who even speaks of a Law of reversion (Darwin, 1859g0). He considered the occurrence of atavisms as a proof of the possibility and reality of reversion of ancestral characters:

I have stated that the most probable hypothesis to account for the reappearance of very ancient characters is that there is a tendency in the young of each successive generation to produce the long-lost character, and that this tendency, for unknown causes sometimes prevails. (Darwin, 1859g)

After Darwin, in 1883, a Belgian biologist, Louis Antoine Marie Joseph Dollo (1857-1931), formulated the law of irreversibility of evolutionarily lost traits, known as Dollo’s law in his honor. As defined by one of its coauthors, the Law says: “What in the course of ages has phylogenetically disappeared cannot again recur” (Hall, 1998). This law explicitly denies the possibility of evolutionary reversions.

At that time, under circumstances of wide gaps in the paleontological record, Dollo’s law raised no doubts about its validity and later it was incorporated into the general neoDarwinian scheme of evolution. It was argued that a lost trait cannot reappear after long periods of time because, in the absence of selection on genes responsible for the trait, genes will mutate to such an extent that would become nonfunctional. It was estimated that the silenced genes might retain their function for no longer than 6 million years (Marshall et al., 1994) because

Degradation of genetic information is sufficiently fast that genes or developmental pathways released from selective pressure will rapidly become non-functional. (Marshall et al., 1994)

One should keep in mind that the estimation is made for a single gene, whereas in most cases more than one gene are involved in the development of morphological traits. Empirical evidence, however, has shown that genes may remain functional for periods of time longer than 6 milion years (odontogenic genes in birds, for example, are still functional presently, ~80 million years after the loss of dentition in this vertebrate class). Authors do not explain what prevents expression of these genes in the lost organs alone, and what reactivates their function during evolutionary reversals.

Now, more than one century after formulation of Dollo’s law, adequate solid evidence on reversion of ancestral structures is accumulated and hardly any biologist would reasonably question the occurrence of evolutionary reversions, which now seem to have been a leitmotif in the evolution of the Animal Kingdom.

The neoDarwinian biology finds it impossible to reconcile its basic tenet of accumulation of mutations, as a prerequisite of the evolutionary change, with the established facts of evolutionary reversions occurring independently of changes in genes.

In an attempt to show that Dollo’s law is still valid, Bull and Charnov present what they call “7 possible examples” of irreversible evolution: all-female parthenogenesis (thelytoky), polyploidy, selfing in hermaphroditic populations, dioecy evolved from hermaphroditism, heteromorphic sex chromosomes, Muller’s ratchet, and haplo-diploidy.

It is noteworthy that their list does not include the irreversibility of morphological characters, the most visible aspect of evolution and diversity of the animal world. They believe that the irreversibility of the above non-morphological characters may be deduced from the uniqueness and irreversibility of the history of living organisms (Bull and Charnov, 1985). By inferring possible examples of irreversible evolution “from the uniqueness and irreversibility of the history of living organisms” authors only make a circular reasoning for the irreversibility of evolution is nothing but the irreversibility of the history of living organisms.

Theoretical arguments against the Bull and Charnov’s examples aside, recent empirical evidence shows that one of the assumed impossible reversions, the reevolution of sexuality from parthenogenesis, occurred in mites of the Crotoniidae family. These sexually reproducing mites evolved from the parthenogenetically reproducing ancestors of the family Camisiidae. Investigators believe this case defies Dollo’s law of irreversibility of evolution and proves that “parthenogenesis is not necessarily an evolutionary dead end” (Domes et al., 2007).

B. Rensch considered two striking exceptions from the principle of the “phylogenetical irreversibility” as questionable (Rensch, 1960b). First, the transformation of the heterodont teeth  (teeth of different morphology) of the primeval whales (Archaeoceti) into isodont teeth of modern whales, which represents a return to the reptilian dentition from which mammal heterodentition originated, and second, the reappearance of the undifferentiated type of vertebral column in snakes and slow worms (legless lizards of Anguidae and Amphisbaenidae families). But he argues that “Such reversibilities seem to be extremely rare in the major steps of transspecific evolution”. Rensch’s admission that these evolutionary reversions occurred, contradicts Dollo’s law. The difficulties in explaining their origin compelled him to “downgrade” Dollo’s law into a “rule”.

Rensch attempted to relate such phenomena of reversible evolution to the reversibility of mutations (Rensch, 1960b) but this would raise a serious theoretical objection. Such “macroreversions” are difficult to be conceived as products of adaptive mutations (all the experimentally produced mutations are deleterious or have no adaptive value) of a single gene, for such transformations require multiple adaptive mutations. Furthermore, there is no evidence on mutations in genes related to transition from heterodonty to isodonty and from differentiated to undifferentiated type of vertebral column. On the contrary, as shown earlier, the GRNs (gene regulatory networks) and  genes involved in these transitions are conserved across the vertebrate taxa.

There is some confusion on the meaning of the “evolutionary reversion” of a species to its ancestral phenotype. In regard to the problem of phylogenetic irreversibility and to evolutionary reversions, B. Rensch believes that what reappears during reversions is only the general appearance of the character not the identical structure (Rensch, 1960d).

Probably every biologist would agree that  “identical” return to ancestral features is inherently impossible. However, by recognizing the occurrence of evolutionary reversions biologists do not imply any “identical” return to the ancestral state.

When we speak of a trait, such as a fin, a head, tail, tooth, etc., we use such words to describe structures which, being distinct in different animal species, are related to each other by the origin, function and patterning. No principle of identicalness could apply in determining whether a structure is of the same kind for, sensu stricto, no identical structures could ever evolve.

The concept of “identicalness” in our biological context is irrelevant and cannot be a defining criterion of evolutionary reversions. So, e.g., huge as they are, differences between the head of a mammal and that of a fish, they are not essential enough to force us to look for new terms for describing the same organ. When we say that phasmid insects have lost and regained their wings, this is a clear statement of evolutionary reversion of a structure with a specific function that was lost somewhere in phylogeny no matter whether, or how much, it differs from the ancestral structure. In this meaning, evolutionary reversions are as real as losses of structures are. It would make not much sense to look at the modern metazoans for structures that are “identical” to the ones their ancestors had. If the modern biology does not apply any criterion of “identicalness” to the study of homologous organs or parts, why should we expect evolutionary reversions to be identical to ancestral structures?

The impossibility of returning to a structure that would be identical to its ancestral state has its specific causal basis. It is related to the unavoidable differences that evolve over time in their developmental and genetic contexts. Remember that not only in different species and different individuals, but even in the different parts of the same organism, different developmental contexts may determine different patterns of gene expression and phenotypic outcomes. So, e.g., the pdm and apterous genes show distinctive patterns of expression in wings and legs in Drosophila (Cohen et al, 1992; Ng et al., 1995; Averoff and Cohen, 1997).

Hence, what is to be expected in the cases of evolutionary reversals is not “identicalness” to ancestral structures rather than recurrence of ancestral “design”, with the last word used, in Webster’s meaning, for describing “instructions for making something which leave the details to be worked out.” It is namely a “common design” related to a common developmental pathway, executed under different developmental and genetic contexts, that makes us viscerally think of the locomotor appendages of reptiles, birds and mammals as limbs, despite the obvious differences in their structure and morphology.

In the light of the modern knowledge on the relationship between species’ genome and its morphology it is not the genes or groups of genes involved in the formation of a biological structure that count but the patterns of their expression. Evolutionary reversions would, thus, necessarily differ somehow from the ancestral original because:

1. The evolution of the genome implies quantitative and qualitative changes in genes and overall organization of the genome. It creates a new and different genetic context, which may affect the result of activation of signal cascades and GRNs (gene regulatory networks).

2. The biochemical and cytological environments in which the products of genes will act and interact also may change in the course of evolution. A different developmental context will arise that also might influence the phenotypic outcome.

Thus, the changed genetic-developmental background would lead to unavoidable differences in the phenotypic results and evolutionary reversions will not be identical to the ancestral structures. What reverses is the ancestral morphological design rather than an identical structure (in the above Webster’s meaning).

Hence, evolutionary reversion consists in reappearance of a phenotype that is similar, rather than identical, to the lost ancestral phenotype. It arises as a result of activation of a suppressed ancestral developmental pathway under conditions of the changed developmental and genetic context.

This definition allows us to predict that

1. Evolutionary reversions can occur whenever an inactivated ancestral developmental pathway is be reactivated, and hence

2. Evolutionary reversions  may be reproducible and can be experimentally induced.

By the early 70es of the 20^th century, biologists came to realize that evolutionary reversions were not induced by genetic mechanisms:

The reappearance of a complex trait suggests that much of the organizational basis for the feature has survived intact in the genome but has been deactivated (through introduction of a negative feedback element or loss of a positive feedback element, etc.). The probability that a large portion of the genome responsible for the feature has survived pleiotropic substitution or mutation and that only certain critical regulatory genes have been affected by these “random” processes would seem a remote combination of probabilities indeed to explain a common pattern. (Regal, 1977)

From present-day knowledge, it can be argued that extant species have conserved in a functionally unaffected state genes involved in the development of ancestral structures (remember: birds have conserved in a functionally intact state genes involved in teeth development for ~80 million years after having lost dentition).

.............................................................................

.............................................................................

Biologists denied the possibility of the occurrence of reversions of features lost for longer periods of time as it is the case with aquatic mammals.. No evolutionary reversion would be possible to occur after more than 6 million years. And the fact that in the North American wildcat, Lynx, a third cusp in the carnassial teeth of the lower jaw reversed after 20 million years, was explained by assuming that the loss of that cusp did not involve the loss of any structural gene (Marshall et al., 1994).

Other reversals have been noted in complex structures, including the reappearance of “lost” muscles in the limbs of some birds and limbs in usually limbless tetrapods. It has also been possible to experimentally reverse toe number in guinea pigs by selective breeding from three toes to a more primitive four toes. This reversal appears to be the consequence of the continued maintenance of an ancestral developmental pathway that can produce more toes in guinea pigs and can be elicited in the appropriate genetic background. As is the Lynx molar, toe number is a meristic trait: once the anlage is provided, the “toe program” is played out automatically. (Marshall et al., 1994)

The authors do not elaborate on the “toe program”: is that program activated by a regulatory gene or not. If the first would be the case then that gene too, being silenced for 20 million years, according to their estimation, would be nonfunctional and the reversion of the ancestral developmental pathway and ancestral cusp would be impossible. If the second, that is if no gene is involved in the loss of the cusp, then the activation of the ancestral cascade is non-genetically determined.

Mos of developmental pathways are remarkably conserved in metazoans. Hence, it would be evolutionarily advantageous to activate those conserved pathways than re-evolve them. This is what some of experiments mentioned by Raff seem to suggest. In one of those experiments it has been demonstrated that a relatively simple treatment with testosterone in an all-female species of fish induces a number of complex processes: “reactivation” of “complex morphogenetic pathways” and reversion of the “lost” complex morphologies of “complex insemination apparatus”, and male body proportions and pigmentation as well as male sexual behavior and spermatogenesis. Another example: evolutionary reversion of tetrapods to the marine habitat led to the loss of hind limbs. In whales this occurred 40-50 million years ago, but cases of atavistic reappearance of hind limbs, although in a reduced form, in whales show that genes, GRNs (gene regulatory networks) and developmental pathways determining hind limb development are still present and functional in these animals.

Based on the theoretical arguments, on the one hand, and on the empirical evidence on long term evolutionary reversions in fish and lizards, especially on the experimental observations on spontaneous reversions in ambystomatid salamanders from Mexican highlands [these neotenic species that evolved 0.5-1 million years ago occasionally reverse to metamorphosing species (Schaffer, 1984)], on the other, Raff accepts the possibility of evolutionary reversions for short periods of time but not for longer periods of time required for evolution of fish-like morphology  in marine mammals:

It is unlikely that genes governing development of a hydrodynamically streamlined body shape and fins were saved for a rainy day through more than 300 million years of terrestrial evolution. (Raff, 1996b)

With the benefit of evidence on gene regulatory networks accumulated in the meantime, it is impossible to believe that transition of mammals from terrestrial to aquatic mode of life, that is the development of fins instead of limbs, would require any loss (by deletion or otherwise) of “genes for tetrapod limbs”. In the light of the modern biological research and the comparative results of the sequencing of genomes of various species, it seems more plausible that transformation of mammal limbs into fins of marine mammals did not require or involve the loss of “genes for fins” and/or evolution of new genes for fish-like morphology. Recent evidence generally suggests that epigenetic repatterning of gene expression, rather than any loss or gain of genes, drove the evolution of these terrestrial mammals into marine fish-like animals.

It is becoming clear that although, over time, genes unavoidably evolve via mutations, evolutionary changes of the phenotype result from specific changes in developmental pathways and in patterns of gene expression. These evolutionary changes do not depend on changes in genes. Loss and reversion of a structure implies not silencing or loss of genes; most of genes are conserved in the course of phylogeny because they are necessary for the development of many other structures in the body. Reviewing a number of experimental studies on the loss and induction of organs in amphibians, B.K. Hall came to the conclusion:

An organ may be lost without loss of the entire developmental system for producing that organ…Loss of organs is often mediated through modification (not loss) of inductive reactions. (Hall, 1998l)

Hundreds and thousands of genes are involved in the development of each structure in the animal body (~2500 genes are involved in the development of eye, e.g.) and most of those genes are involved in the development of most of the rest of animal structures.

Metazoan organisms develop (or prevent the ectopic development of) different structures by specifically activating different developmental pathways in different parts of the body, although the same genes are present in all over the body. They succeed in doing this because they are capable to selectively switch off/on different developmental pathways in different regions of the body. In principle, there is no visible reason why metazoans would not use this ability to selectively switch off/on specific developmental pathways for suppressing and activating developmental pathways for producing evolutionary loss and reversion of ancestral phenotypes.

Atavisms: Ancestral Developmental Pathways May Be Conserved and Reactivated

Atavisms are sudden reversions to ancestral morphological features in small proportions of individuals of a population. According to de Beer (1958), the fundamental criterion of an atavistic structure is morphological resemblance to that of an ancestor, regardless of its genetic basis (Lande, 1978). No “hypothesis of reverse mutations” could explain their origin. Firstly, because no reverse mutation is known to systematically occur at frequencies many atavisms occur and, secondly, emergence of atavistic structures requires reactivation and occurrence of “useful” mutations simultaneously in more than one gene. The suddenness of the appearance of lost ancestral structures taking place during atavisms proves that

1. While losing structures, metazoans can still conserve developmental pathways for the lost ancestral structures and re-evolve them, and

2. Reversion to ancestral structures does not require new or changed genes or genetic information.

According to R.A. Raff, “relatively weak selection could lead to limb reduction and virtual loss in as little as one million years” (Raff, 1996d). He also estimated that atavistic appearance of hind limbs in marine mammals is retained for as long as 10⁶ to 10⁷ generations. R. Lande (1987) notes that the process of vestigialization of hind limbs in whales may have taken a few million years until they were lost ~ 40 million years ago but atavistic recovery of hind limbs still occurs after this even evolutionarily very long period of time.

Among atavisms recognized by Rensch in his Evolution Above the Species Level is the formation of a fourth toe (which is normally reduced) in guinea pigs, appearance of rudimentary hind limbs in whales and dolphins, formation of supernumerary nipples in mammals (Rensch, 1960c), secondary lack of shells in snails, and secondary development of a cap-shaped shell in snails such as Ancylus (Rensch, 1960d).

Rare cases of atavistic development of hind limbs have been reported to occur in the humpback whale, Megaptera nodosa, and the sperm whale, Physeter catadon, with an estimated frequency of 0.02% of the general population (Lande, 1978). In another study, 37% of a population of 72 individuals of minke whale, Balaenoptera acutorostrata, developed a bony femoral rudiment. Balaena mysticetus even develops a vestigial femur and tibia. Skeletal elements, distinct from rudimentary pelvic girdle appear in humpback whales at a frequency of 1:5000 and completely developed hindlimbs have been observed on another whale (Bejder and Hall, 2005). The fact that such atavistic phenomena repeatedly occur proves that developmental pathways for the lost ancestral tetrapod structures are conserved in marine mammals for evolutionarily long periods of ~50 million years.

The number of digits is 5 in amphibians (some forms have four) but in mammals it varies from 5 (humans) to 1 (horse). It is well known the fact that with a certain frequency horses develop two additional toe bones, one on each side.

One in several hundreds of pintail ducks in Kerguelen islands of southern hemisphere shows many markings of the northern pintail, Anas acuta (Omland, 1997).

Musculus iliofemoralis externus atavistically reappears in individuals of many bird species in Hawai, Australia and New Zealand and atavisms of musculus caudiliofemoralis pars iliofemoralis are recorded in birds in USA and in Tuamotu Archipelago in the Pacific Ocean (Hall, 1998j).

Evidence is presented on the anomalous reappearance of ancestral muscles in individuals of species of other birds and mammals that presently lack these muscles. Such is the case with musculus caudiliofemoralis observed on the left side of an individual of the bird Artamus leucorhynchus; musculus abductor cruris caudalis in the hind limb of the rodent jerboa (Jaculus jaculus); musculus latissimus dorsi pars caudalis in bird wings has been found on both sides and/or on one side in individuals of the passerine bird, Thraupis palmarum; two cases of reestablishment of musculus iliofemoralis externus are observed in birds of the family of sturnids (Raikow et al., 1979).

An atavism in humans is the sudden appearance of the “werewolf syndrome” (congenital generalized hypertrichosis, characterized by a very intense hair growth all over the human body) in man. It is assumed, that the developmental pathway for hair-coverage was silenced after humans diverged from our primate ancestors, but occasionally it is reactivated to produce the atavism.

Neodarwinian Explanation of Atavisms

I am not aware of any serious neoDarwinian interpretation of atavistic reversions, but none of the known neoDarwinian mechanisms of evolutionary change (gene mutations, genetic recombinations, neutral mutations, gene drift and the implied natural selection) are applicable as explanations for the occurrence of atavisms. Even if the highly speculative idea that genes that have been silenced during the phylogeny may be reactivated to produce atavisms would be proven to be true, it will not fit into the neoDarwinian paradigm.

Epigenetic Explanation of Atavisms 

One of the basic tenets of the epigenetic theory of evolution presented in this work is that loss of  various phenotypic (behavioral, morphological, physiological, and life history) characters is not necessarily associated with the loss of relevant genes or loss of developmental and neural pathways determining the development of the lost structure. As is extensively shown in chapters 11 and 12 on circumevolutionary phenomena, switching of developmental pathways, involving no changes in genes for producing alternative (in some cases inherited) phenotypes, is a common phenomenon in metazoans. Such switches to alternative developmental pathways or even to ancestral developmental pathways in some cases has been possible to be experimentally induced (see on experimentally induced reversions later in this chapter) and these cases represent nothing less than experimental atavisms or experimental reversions.

What takes place in the cases of the appearance of atavisms in nature is that a “forbidden” developmental pathway is unpredictably activated. From this perspective, atavisms can be considered to result from accidental activation of developmental pathways that have been switched off in the course of the species phylogeny. 

Let’s briefly review the evidence on the evolutionary reversions in nature as they appear in the comparative anatomy and the paleontological evidence.

Evolutionary Reversions in Nature 

Digestive Tract  

Alternation of teethed and toothless forms is repeatedly encountered in all vertebrate classes:  

Toothlessness occurs repeatedly in many forms of nearly all classes of vertebrates, with birds lacking them during all stages of their development. (Montagna, 1959)  

As already noted, transition of mammals from terrestrial to aquatic life (whales, dolphins, porpoises) is surprisingly associated with a transition from heterodontia (the teeth are different in their shape and size) to homodontia, a dentition which is typical for lower vertebrates such as some fish, amphibians, and reptiles. Empirical evidence shows that genes involved in odontogenesis are the same and conserved in both homodont and heterodont vertebrates.

The labyrinthine structure of teeth (enamel is infolded along longitudinal grooves, often making a complicated pattern in the interior) characteristic for extinct crossopterygians disappears in the rest of fish and amphibians to reappear in a group of frogs Labyrinthodontia, which owes its name to that teeth structure.

In the digestive tract some blind diverticuli, known as caeca (sing. lat. caecum - blind) appear and disappear during the evolution of vertebrates. Such caeca are known to be present even in organisms as simple as amphioxus. In birds a caecum appears as crop sack. Pyloric and duodenal caeca are numerous in some fish and up to 200 caeca are described in the mackerel, a fish in the North Atlantic. An ileocolic caecum is common in amniotes, including man (Kent, 1973h).  

Respiratory System

 Evolution of the respiratory system in vertebrates (gills and lungs) also is characterized by several instances of evolutionary reversions. In vertebrates, the lung appears since the earliest jawed fish (gnathostomes) and is a general feature of ostheichthyos (Carter, 1967f). One fossilized placoderm fish Bothriolepis, had a paired lungs (Carter, 1967d). The lung then disappears in most of the modern fish to reappear in amphibians, reaching its most complicated form in birds.  

Eyes in Snakes  

Living in darkness in burrows, lizards from which snakes originated, lost vision and some important components of their eyes (pigments of their visual cells, lacrimal glands, the iris diaphragm, and focusing muscles). Evolving into snakes, later they reevolved the complete original eye structure.  

Warmbloodedness 

Warmbloodedness generally has been considered to be a characteristic of the classes of birds and mammals, although for a long time it has been known that warmbloodedness evolved first in the class of fish. Numerous warmblooded species and genera of the class of fish predate the appearance of warmbloodedness in birds and mammals. But it seems to have been lost in reptiles and independently, convergently, evolved again in birds and mammals, accompanied by a series of morphological and physiological adaptations (feathers, fur, subcutaneous fat, temperature homeostatic mechanism, etc.).  

Paleontological Evidence on Evolutionary Reversions 

Vast evidence on evolutionary reversals is presented by paleontologists of the end of 19^th -early 20^th century. At the time, the iterative (from Latin iterare - to repeat) formation of species (iterative Arten- und Formenbildung), under which was not understood a perfect and full repeat of an ancestral form because, in Karl Beurlen’s expression “a form returns, but in a different stage of development, i. e., phylogenetically different” (Beurlen, 1937h). In many gastropod groups, especially that of Pleurotomariide from Silurian to present-day forms, he observed that  

Sometimes a certain form for long periods of time remains quite unchanged, but repeatedly becomes the starting point of a rampant divergent species formation” (Beurlen,1937d)  

So, e.g., Pleurotomaria strain repeats Worthenia type. The same phenomenon was demonstrated by H. Salfeld in ammonites (cephalopod molluscs). The Cardioceras genus of ammonites that is not immediately related to Amaltheides repeats the same developmental stages of Amaltheides. Initially, both of them are normal forms, without a keel. Later a full keel develops in both genera, and this marks the end of their development. In a similar way Schlothemia type of lower Lias will be repeated by Parkinsonia in upper Dogger age, Oxynoticeras type of lower Lias by Dorsetensia in lower Dogger, and Coeloceras type of middle Lias by Stephanoceras in the middle Dogger, etc. The same phenomenon could be demonstrated in numerous other cases; it appears almost systematically where an analysis of sufficient material is possible (Beurlen, 1937e).

Similar “iterations” are observed in other ammonite groups: Garantiana - Cosmoceras, Aspidoceras - Physodoceras, Rasenia-Polyptychites, etc.

Vola type of the family of bivalved pectinides (order Filibranches) is a derivative type that independently evolved three times from Pecten (Chlamys-type): the first time in Lias (Weyla), second time in Cretaceous (Neithea), and third time in Tertiar (Vola) (Beurlen, 1937f). Again here it is important to point out that the same type appears three times under three very different conditions of living, what makes very difficult perception of a possible role of the environment and natural selection in this evolutionary reversion.

Another important reversion (“iteration”) is observed in the relationship between bivalved marine oyster genera Gryphaea and Ostrea (Liostrea). Gryphaea type has repeatedly intermittently appeared three times during Lias (Gr. arcuata), Dogger (Gr. dilatata), and Cretaceous Senon (Gr. vesiculata) from the conservative strain Ostrea (Liostrea). Each of those iterative forms represents the respective phylogenetic stage of its parental Ostrea.

Higher in a side phylogenetic line, cartilagineous fish, Chondroichthyes (whose ancestors had bony skeletons) have almost completely lost the bone tissue, which is replaced by cartilaginous tissue. In Osteoichthyes the skeleton is composed mostly of bone. Higher, in Chondrosteans, ossification is lost but again still higher (later) in Holostei, and especially in Teleostei, the bony skeleton reappears in an intensified form. 

Reversion of Shell Coiling in Gastropods 

Although evolution of shell coiling in gastropods has been associated with several adaptive advantages, it has been repeatedly lost (Vermeij, 1987) and the taxa that lost shell coiling have been considered to be unable to revert to the coiled shell because of the developmental and constructional constraints that evolved after the loss of shell coiling, which presumably restricted the number of possible morphologies and prevented the reversion to regular coiling (Gould and Robinson, 1994).

However, reliable evidence shows that evolutionary uncoiling of shells in gastropods does not represent an evolutionary dead end. Paleontologists have described a great number of evolutionary convergences in the shell form of molluscs. As a result of the volcanic character of the Steinheim basin in Southwestern Germany  and of periodic appearance of warm springs, the snail Gyraulus multiformis has been subject to environment temperatures that varied over a wide range. Accordingly (as a consequence of the appearance and disappearance of the warm springs), the snail changed its shell form from that of a flat helix to a high helix and back to the original flat helix. Paleontological studies of E. Hennig and Fejervary have shown that the return to ancestral forms of shells is a widespread phenomenon in ammonites (Beurlen, 1937b).

Recent studies on a number of species of genera of the Calyptraeida family, have revealed that the loss of shell coiling in mollusks leads to no evolutionary dead end as it has beeen previously believed; reversions of coiled shells have been identified in two populations of Trochita calyptraeformis, and possibly in three other species of Zegalerus and Sigapatella (Zegalerus tenuis, Sigapatella terranovae, and S. novazealandica). Trochita appeared first during Miocene, 23.8 to 20.5 million years ago, and  paleontological evidence shows that during the evolution it lost shell coiling, whereas modern Trochita calyptraeformis has reevolved regular shell coiling  (Collin and Cipriani, 2003; figure 16.1).

Figure 16.1. Coiled Trochita calyptraeformis  (left) and a typically uncoiled, bilaterally symmetrical Crepidula species Crepidula norrisiarum (right) (From Collin and Cipriani, 2003). 

Investigators argue that the Trochita case of reversion to ancestral shell coiling is different from many other cases of evolutionary reversions in which genes and developmental pathways for reversion were maintained because they were used for development of other structures, for those structures (toe number in horses and guinea pigs, number of molars in Lynx, etc.), in distinction from the mollusc shell, are meristic characters regulated by heterotopic expression of genes. They believe that the mechanism of shell coiling is retained in the larval stage of these molluscs and a heterochronic mechanism may have been operating in these cases. They also believe that  

The genetic and developmental pathways for shell coiling have been retained in the larval stages of the uncoiled. (Collin and Cipriani, 2003) 

and only a heterochronic change in the activation of these pathways, not any gene mutations, would be necessary to re-evolve coiled shells (Collin and Cipriani, 2003). However, the concept of heterochrony does not contribute to the understanding of the mechanism of the reversion of shell coiling because the heterochrony is a phenomenon whose mechanism is not known and, hence, needs itself to be explained.  

Reversion of Cartilaginous Skeleton in Fish 

Embolomers and earliest stegocephals had fully ossified endocranium while later, in stegocephals and amphibians, endocranium turns cartilaginous like that of their earlier ancestors (Beurlen, 1937b). Ostracoderms, the oldest class of fish lived in fresh water for nearly 500 million years. They had no jaws and no paired fins but they had bony plates and scales. Their survivors, the living cyclostomes, have no bones on the skin or any place else. It was thought that the “enzymatic complex necessary for the deposit of bone is no longer present” in cyclostomes (Kent, 1973b).

Cartilaginous skeleton of sharks is a secondary development and the bone seen in the skeleton of ostracoderms, placoderms, and the first bony fish is truly primitive (Colbert and Morales, 1991). Cartilaginous skeleton of sharks, thus, is a reversion to the cartilaginous skeleton of primitive chordates. Cartilaginous fish derive from bony fish (Carter, 1967c).  

Reversion of the Hydrodynamic Body Shape in Marine Mammals 

The most impressive and well known example of the reevolution of ancestral body shapes is the adaptation of terrestrial mammals to the aquatic life. In the process of adaptation to aquatic habitat, their whole body transformed into fish-like shape by shortening of the neck and head, addition of vertebrae, reduction of the pelvic girdle, loss of hind limbs, isolation of ear-bones, etc.

Scientists believe that the evolution of the first whale forerunners from their ungulate terrestrial mammal ancestors occurred about 60 million years ago. This is the estimated age of fossils of small primeval whales that still possessed carnivorous heterodont dentition and articulate mobile cervical vertebrae. 52 million years ago a whale ancestor existed that still used its hind- and forelimbs for locomotion. Biologists named it Ambulocetus natans. It was still able to walk on land, probably similarly to modern sea lions, and to swim by undulating its spine and propelling itself by feet movements.

All cetaceans (whales, dolphins, porpoises, etc.) are unanimously recognized to have evolved from terrestrial mammals. They have developed a fish-like body shape as well as fins and fluke. Within no more than 10 million years these terrestrial mammals evolved into a whole order of marine mammals (Eldredge, 1989). Given that smaller morphological changes necessary for evolution of Hyracotherium to a horse took 50-60 million years to occur (Wesson, 1991a), it is reasonable to think that there should be some factor that facilitated the evolutionary reversion from terrestrial ungulates to modern aquatic mammals. As has been repeatedly pointed out, this may be related to conservation in an inactive form of the developmental pathways of their marine ancestors.

Such a major evolutionary event as this rapid appearance of a whole group of aquatic mammals would be unthinkable from the point of view of the neoDarwinian gradualism that implies the occurrence and accumulation of numerous useful point mutations under the action of natural selection. Some calculations made by R. Wesson (1991) give a general idea of the improbability of evolution of the whale from its terrestrial ancestor via gene mutations: 

By Mayr’s calculation, in a rapidly evolving line an organ may enlarge about 1 to 10 percent per million years, but the organs of the whale-in-becoming must have grown about ten times more rapidly over 10 million years. Perhaps 300 generations are required for a gene substitution. Moreover, mutations need to occur many times, even with considerable selective advantage, in order to have a good chance of becoming fixed. Considering the length of whale generations, the rarity with which the needed mutations are likely to appear, and the multitude of mutations needed to convert a land animal into a whale, it is easy to conclude that gradualist natural selection of random variations cannot account for this animal. (Wesson, 1991a)

If the changes that occurred in the morphology, physiology, and behavior of marine mammals, belong, and unquestionably they belong, to the category of changes involving more than one gene, any evolutionist, by right, would ask: “How is it that such an almost improbable event occurred more than once, i.e., in whales, dolphins, porpoises, and seals?”

Miracles ruled out, no convincing answer could be given to the above question based on the tenets of the neoDarwinian paradigm.  

Reversion of Wings in Stick Insects 

Reversion of ancestral wings in wingless insects has been considered impossible because of the complexity of the events that would enable their reevolution:  

Entomologists have long assumed that re-evolution of wings in apterous lineages was impossible, because functional wings require complex interactions among multiple structures, and the associated genes would be free to accumulate mutations in wingless lineages, effectively blocking the path for any future wing reacquisition. However, this assumption requires that developmental pathways for wing formation are largely independent of pathways required for development of other structures. For instance, in Drosophila and other insects, leg and wing imaginal discs have a common origin from a single group of cells and the development pathway for wing determination has been largely coopted (recruited) from the pathway required for limb formation. (Whiting et al., 2003) 

The order of stick insects, phasmids, consists of 3000 described species belonging to three families with about 500 genera. 60 percent of the extant phasmid species are fully or partially wingless. Although their ancestral condition is wingless, phasmids have independently evolved wings at least 4 times (Whiting et al., 2003). Wings in phasmids did not evolve de novo but are result of activation of an ancestral wing patterning pathway. This was possible to occur because the wingless insects have conserved   

the neural structures and basic functional circuitry required for flight. (Whiting et al., 2003)  

However, the study has been challenged by Trueman et al. (2004), who believe that, while representing an important advance in our understanding of insect and gene evolution, it is at variance with the long-held view that wings evolved only once in insects and have been repeatedly lost (Trueman et al., 2004). In turn Whiting et al., responded to the critique by arguing that their analyses, as well as those of Trueman et al., suggest that both parsimony and likelihood methods support the notion that the ancestral state in stick insects has been wingless and wings have independently evolved many times in this group, and wing reversions represent an evolutionary phenomenon that is more widespread than generally assumed (Whiting and Whiting, 2004).

Other biologists also believe that 

The studies of these insects illustrate that the basic blueprints for complex developmental structures can remain largely intact even over large evolutionary spans (i.e. radiations of higher level taxonomic groups). (Porter and Crandall, 2003) 

Reversion of wings in phasmids involved no changes in genes. Despite the loss of wings, insects have conserved the wing patterning pathway and GRNs (gene regulatory networks) and, as pointed out earlier, the neural circuitry necessary for flight. The reactivation of the wing developmental pathway, in the absence of genetic changes, is epigenetically determined, as most regulatory processes of gene activation are. 

Reversion of Mandible in a Collembolan Insect 

Insect mandibles are complex characters that have been considered to be stable characters. However, insects of the Brachystomellidae family, order Colembola, have lost mandibles several times, independently and adaptively, as a result of switching to a sucking feeding behavior for ingesting food particles in liquid suspension.  Recently is reported that this complex character has also reevolved at least once, in the case of Probrachystomellides nicolaii (Najt et al., 2005).  

Reversion of Eyes in Ostracods 

Two podocopid ostracodes (bivalved crustaceans), Dutoitella lesleyae Dingle and Poseidonamicus whatley Dingle, presently live in shallow water around the Marion Island between the South Africa and Antarctica, at a depth of 113-474m and 240-355m, respectively. Scientific evidence suggests that they have colonized this habitat by migrating from sea regions deeper than 600 to 900 m, where ostracodes are functionally blind. These sighted ostracodes evolved from blind ostracodes of Eocene (56-34 Mya) in response to the lighter aquatic habitat they moved in (Dingle, 2003). The reversion of eyes in these crustaceans implies that the complex ancestral eye-patterning developmental pathways have been conserved for long evolutionary periods in deep sea-dwelling eyeless ostracodes, despite the occurrence of unavoidable changes in genes. 

Reversion of Limbs in Snakes 

An interesting instance of reversion of limbs in snakes is inferred from the study of a 95-million-year-old fossil snake from the Middle East. It represents the most extreme hindlimb development seen so far in snakes. The limb consists of tibia, fibula, tarsals, metatarsals, and phalanges. The snake is nested with basal snakes, macrostomatans, which retained rudimentary hind limbs and represents a reversion to the ancestral limbed state (Tchernov et al., 2000).  

Reappearance of musculus iliofemoralis externus (IFE) in the Bowerbird Loria loriae and the New Zealand Thrush, Turnagra capensis 

Musculus iliofemoralis externus (IFE), is lost in most Passeriform birds. It was found that in Loria’s Bird-of-paradise (Loria loriae) and the New Zealand thrush (Turnagra capensis; Turnagridae), the IFE has reevolved. That muscle and the musculus iliotrochantericus are separate at the insertion sites but fused over most of their lengths. Based on extensive supporting evidence, Raikow et al. (1979) conclude that the loss of the muscle in these birds “was limited to its expression in the phenotype” and  they believe that  

The loss of a structure in the phenotype does not necessarily mean that the genetic information controlling its development has also been eliminated from the genome. (Raikow et al., 1979)

Reversion of an Ancestral Digit in Guinea Pigs

All the species of the family of Caviidae, including guinea pig have lost one digit (digit I) in the front feet and 2 digits (I and V) in the hind feet. The fact that the same is observed in species of the related family of Hydrochoeridae suggests that the loss may have occurred in their common ancestor, i.e. long ago. Nevertheless, it is not uncommon that individuals with an extra digit are born from normal guinea pigs. By selection and inbreeding of such individuals, by the beginning of the 20^th century, Castle succeeded in creating a strain of guinea pigs that constantly produced offspring with the same fully developed extra digit. The breeding results could be accounted for by neither recessive nor dominant inheritance (Wright, 1934).

Evolutionary Reversion of Life Histories  

H.B. Shaffer studied a number of metamorphosing, facultative and permanently paedomorphic (neotenic) ambystomatid salamander species from the Mexican highlands that diverged from their common metamorphosing ancestor some 0.5 - 1 million years ago. Paedomorphic species occasionally have been shown to revert to metamorphosing species (Shaffer, 1984).

Species of an amphibian group, Stereospondylii, showed the same disgust for the terrestrial life, when after 150 million years of amphibian status they headed back to the sea to live a wholly aquatic life (Taylor, 1983f).

No change in genes involved in the signal cascade determining metamorphosis has occurred  in the paedomorphic and facultative paedomorphic salamanders. 

Evolutionary Reversions in Experiments

Experimental Reversion of Teeth in Birds 

Loss of teeth in birds occurred about 70-80 million years ago (Mitsiadis et al., 2003). Bird embryos develop only transient rudimentary thickenings of epithelium, reminiscent of dental thickening in other vertebrates. The avian mandibular neural crest-derived non-odontogenic mesenchyme (but not, for instance, limb bud mesenchyme) can be experimentally transformed into dental mesenchyme, to begin formation of ectopic tooth buds and express the same odontogenic genes, Msx-1, Msx-2 and Bmp-4, that other vertebrates express, after heterospecific recombination with early mouse embryo odontogenic epithelium (Wang et al., 1998). Such experiments led to the belief that the bird oral epithelium has lost the odontogenic capability, although odontogenic genes are all conserved in a functionally intact form.

In an experiment conducted thre decades ago, E.J. Kollar and C. Fisher grafted chick epithelium with mouse molar mesenchyme. They found ten aberrant structures consisting of dentin and odontoblasts in molar-like configurations. Complete teeth structure developed in four grafts. Special precautions were taken to confirm the avian source of the epithelial tissue. They came to the conclusion that the loss of teeth in Aves did not result from a loss of genetic coding for enamel synthesis in the oral epithelium but from an alteration in the tissue interactions required for ontogenesis (Kollar and Fisher, 1980).

Later experiments have shown that it is the bird neural crest-derived mesenchyme that might have lost the odontogenic epigenetic information. It has been possible to revert dentition in chicks by homotopic transplantation in chick embryos of segments of mouse neural tube, where neural crest cells for mandibular mesenchyme come from. It was observed that, 1-2 days after transplantation, mouse neural crest cells migrate to the mandibular and maxillar regions of the chick embryo. Investigators believe that the main factor that led to the loss of dentition in Aves is the failure of their oral epithelium to express BMP4 (Chen et al., 2000). If this is true, then one of the main functions of the mouse neural crest mesenchyme may be the activation of BMP4 expression in the chick oral epithelium:

Neural crest cells may play a role in the activation of BMP4 and Shh expression in tooth-forming sites of the murine oral epithelium. (Mitsiadis et al., 2003)

Chick-mouse chimera formed teeth that morphologically were of the mouse type (Mitsiadis et al., 2003). The fact that embryonic chick oral epithelium is able to properly interact with the gene regulatory network of the mouse showed that the gene regulatory network of the oral chick epithelium is conserved despite the very long time of the loss of dentition in Aves.

As with the development of any other organ, the most difficult problem metazoans have to solve for evolving teeth as an evolutionary innovation is the source of the necessary information rather than the source of material and energy. Where the information for dentition patterning and tooth development comes from? As pointed out earlier, tooth development depends on the interaction between cytological elements and signals from oral epithelium and the underlying neural crest-derived mesenchyme. What is the relative role of each of them in shaping tooth morphology? Which is the sender of instructions and which is the receiver of these instructions? The problem cannot be reasonably presented in the form of an “Egg-or-chicken” question; experimental evidence has shown that both the oral epithelium and the neural crest-derived mesenchyme are indispensable for tooth formation. Experimental evidence shows that, on the one hand, the oral epidermis forms initial teeth primordia, and on the other - that the neural crest derived mesenchyme provides the information necessary for their development into teeth. Here is the interpretation of leading experimenters in the field:

Our data and those from recombination experiments shows that first branchial arch epithelium is unique in containing instruction signals for odontogenesis and that these signals are capable of overriding any prepatterning information present in the CNC (cranial neural crest – N.C.) cells. If this is the case, then it follows that cells receiving these instructive signals must follow identical differentiation pathways. However, the mandibular and maxillary primordia develop obviously different skeletal structures and subtly different teeth yet both are covered by the same oral epithelium. In addition, the molecular and genetic differences identified between the ectomesenchymal cells of these two processes, i.e. the differences in Dlx gene expression, the different knock-out phenotypes observed for maxillary and mandibular molar and skeletal elements and the result from our cell transplantation studies, are indicative of different patterning processes. The fact that maxillary and mandibular epithelia are interchangeable as regulators of ectomesenchymal gene expression indicates that the specificity of responses to the instructive epithelial signals must be a property intrinsic to the ectomesenchymal cells. Thus in order for these cells to develop their maxilla-specific skeletal structures they may be able retain an element of prespecification to prevent development as mandible skeletal elements.

It is tempting to speculate that the different properties of the mandibular and maxillary ectomesenchymal cells are related to different origins of the neural crest cells that populate these components of the first branchial arch. Fate mapping in avian and mouse embryos shows that the mandible is mainly composed of CNC cells that migrate from the midbrain with some contribution from rhombomeres 1 and 2. The maxillary ectomesenchymal cells are derived from CNC cells migrating from both the midbrain and the forebrain. Such a difference in axial origin might explain the different responses of these cells to epithelial signals. (Ferguson et al., 2000)

NeoDarwinian Explanation 

Neo-Darwinian explanation would essentially relate the loss of dentition in Aves to gradual accumulation of mutations in genes that determine odontogenesis or genes of the GRN (gene regulatory network) for odontogenesis. Accordingly, during the evolutionarily long period of time since Aves lost their dentition, natural selection could not have acted against accumulation of deleterious mutations in odontogenic genes and such harmful changes in the genes thus rendering nonfunctional the odontogenic genes in birds.

Even for traits that are determined by a single gene in metazoans the maximal estimated time after which a silenced gene could be reactivated for producing the lost ancestral trait is 6 million years. Since in the development of most phenotypic traits, including teeth, a varying number of genes rather than a single gene is involved, reversion of dentition in birds would be possible only for periods of time shorter than 6 million years. Consequently, not only the natural reversion of teeth but even the experimental induction of tooth development in species of this vertebrate class would be impossible.

The experimental induction of teeth, about 60-80 million years after they are lost in birds, refutes the Dollo’s law and the neoDarwinian prediction that reversion of teeth in birds is impossible.

Contrary to the neoDarwinian prediction, empirical evidence shows that even after many million years, these genes are present and functional in birds although mutations that do not affect the function of proteins they code for, unavoidably have occurred in the course of the long evolution. The capability of the chick epithelium to form teeth in the presence of mouse neural crest from the mouse midbrain indicates that not only odontogenic genes, but developmental pathways for tooth formation are conserved in birds.

What prevents odontogenesis in birds is not any change in genes but an epigenetic event: the loss of ability of the bird oral epithelium to express BMP-4 gene, which in turn is related to the loss  of the ability of the bird midbrain neural crest to induce expression of the gene.

Epigenetic Explanation

The fact that in experiments the chick epithelium produced teeth of the mouse type shows that the mouse neural crest-derived mesenchyme provided the epigenetic information for odontogenesis and that the bird oral epithelium has conserved genes and and gene networks determining odontogenesis. With changes in genes or genetic information excluded as possible cause of the loss of dentition in birds, the only logical alternative for explaining the loss of teeth in birds is a regulatory change in the properties of the neural crest cells migrating from the bird midbrain and forebrain to the mandibular and maxillar regions of the chick. Long time ago, these neural crest cells ceased providing the epigenetic information necessary for inducing expression of Bmp4 and Shh in tooth-forming sites of the chick oral epithelium.

Let’s remember that the chick chimera forms teeth of the type of the donor of the neural tube (mouse) suggesting that the mouse embryonic neural crest cells are provided with instructions for inducing odontogenesis before they are split off from the embryonic mouse brain.

Figure 16.3. Phylogeny of plethodontid salamanders, showing parsimony and maximum likelihood-based reconstructions of ancestral developmental modes. Topology is that of the single most parsimonious tree based on 123 nonmolecular and 2998 mitochondrial and nuclear sequence characters. Branch shading reflects the single most parsimonious reconstruction for ancestral developmental mode with amphiumids coded as biphasic; light branches represent free-living aquatic larvae and dark branches represent direct development. Pie charts at nodes indicate likelihood-based probability of biphasic life cycle (white) versus direct development (black).

Abbreviations: D, Desmognathinae; P, Plethodontini; H, Hemidactyliini; and B, Bolitoglossini. (From Chippindale et al., 2004).

Embryological studies have shown that the direct-developing desmognathines, and their plethodontine ancestors, retain in the egg the larval hyobranchial apparatus, which is essential for respiration and feeding in the water. In contrast, the closely related species of bolitoglossine plethodontids, which develop only very vestigialized hyobranchial apparatus, and species of the genus Eleutherodactylus, which do not develop that larval morphology in the egg, have never returned to the biphasic development (Chippindale et al., 2004).

Thus, conservation of hyobranchial apparatus in direct-developing desmognathines seems to have been the crucial factor enabling them to reverse to the ancestral biphasic (aquatic + terrestrial) life cycle. Ecological stress, resulting from the intense competition in the terrestrial habitat, is believed to have been the major external factor of this evolutionary reversion in the life history of salamanders (Chippindale et al. 2004).

NeoDarwinian Explanation

No changes in genes or genetic information have ever been identified in relation to transition from the direct development to the biphasic life cycle in desmognathines. Hence, all the possible neoDarwinian mechanisms of evolution (gene mutations, gene drift, recombination) are not applicable as an explanans of the repeated occurrence of evolutionary reversions from direct development to the biphasic life cycle with free-swimming larvae.

Epigenetic Explanation 

Observational and experimental evidence shows that transition to biphasic life cycle is related to the patterns of the larval development. It has been observed that direct-developing desmognathine species while still in egg retain larval hyobranchial apparatus, which enables aquatic respiration and feeding, have repeatedly reversed to the biphasic life cycle. On the contrary, bolitoglossines, which develop only a  very reduced apparatus in the egg stage and species of the genus Eleutherodactylus, which do not develop it at all, have not reevolved the aquatic phase of development (Chippindale et al., 2004). With the genetic factors and genetic mechansisms excluded from the involvement in the evolutionary transition to biphasic life cycle the remaining explanation is an epigenetic activation of specific developmental pathways that made the reversion to metamorphosis possible. Recall, metamorphosis and its developmental pathways in amphibians are under strict cerebral control especially via the hypothalamus-pituitary-thyroid axis (see Neural Control of Metamorphosis in Amphibians in chapter 6).  

Reversion of Ancestral Reproductive Modes in Vertebrates 

Amphibians have switched back from viviparity to oviparity (their eggs hatch in the environment) but an amphibian species of the Bufonidae family (order Anura), Nectophrynoides viviparus, and at least populations of two salamander species (Salamandra salamandra and Salamandra algira) are viviparous, with larvae remaining in the uteri and young launched onto the land fully metamorphosed (Kent, 1973c). Salamandra atra secretes nutritive substances and produces eggs on which its viviparous young feed during the prenatal life.

As already mentioned, in 98 occasions reptiles (especially snakes) switched back from oviparity to viviparity. When ichthyosauri started aquatic life and could not make use of the sun warmth to hatch their eggs, they also switched back to viviparity, i.e. their eggs hatched inside mother’s body. However, in some mammals as monotremes, reproduction remains oviparous (they lay eggs, which hatch in the environment). And still higher, placental mammals are remarkably adapted to a perfect viviparous development. The strange loss and reevolution of oviparity, ovoviviparity and viviparity in vertebrate classes seems neither to have always been influenced by any evolutionary pressure nor to have gradually arisen. The repeated pattern of switching to alternative modes of reproduction suggests that vertebrates may have conserved ancestral developmental pathways responsible for ancestral modes of reproduction.  

Reversion to Ancestral Oviparity in Sharks and Rays 

Oviparity is the ancestral reproductive mode in these groups but presently most of their species are viviparous. Transition to viviparity in sharks and rays occurred independently in 12-15 cases. Two cases of reversion from viviparity to oviparity have been identified among these fish, in skates of the family Rajidae (25% of all species) and in the zebra shark, Stegostoma fasciata (Dulvy and Reynolds, 1997). 

Reversion of Viviparity in Reptiles 

It is commonly assumed that reptiles evolved viviparity from oviparity but the reverse has not occurred. This reversion has been considered to be unlikely because it would entail the evolution of complex structural and physiological adaptations necessary for nutrition, oxygen supply, and special maternal hormonal mechanisms.

A number of authors, however, believe that even theoretically, it is difficult to support the assumption that viviparous reptiles could not reverse to oviparous mode of reproduction (Lee and Shine, 1998). Indeed, an analysis of the reproductive mode in reptiles has shown that it has changed a minimum of 49 times in squamates (“lizards” and snakes), with 35 forward transitions, 5 reversions and 9 undetermined transitions (Lee and Shine, 1998).

Based on strong phylogenetic evidence, and the evolution of parity, it has been concluded that populations of the bimodal reproducing (viviparously ond oviparously) European lizard species, Zootoca vivipara, are not monophyletic and that although viviparity evolved only once (with ancestral reproduction mode being oviparous), a number of reversions to oviparity seem to have occurred in various populations of this species (figure 16.4). It is believed that repeated transitions to parity modes of this lizard are related to climatic changes (warmer climate favoring transition to oviparity and colder climates - viviparity) that occurred during Pleistocene in the continent (Surget-Groba et al., 2006).

Figure 16.4. Phylogenetic relationships between the oviparous and viviparous strains of Zootoca vivipara (From Surget-Groba et al., 2006).

Reversion of oviparity in populations of a single species, implying the same genotype, excludes likelihood of involvement of changes in relevant genes in the reversion and, consequently, makes neoDarwinian interpretation inapplicable to the phenomenon.  

Reversion of Arboreal Carabides to the Ground-dwelling Habitat 

Ground beetles, carabides, represent a large family of terrestrial predators. Arboreal carabides, which evolved from ground-dwelling forms, evolved some morphological characteristics (large hemispheric eyes, elongated prothorax; long elytra, long legs etc.) related to the conditions of living under bark or on leaves. It was generally believed that reversion from arboreality to the ground-dwelling life was impossible because of the impossibility of reversion to ground-dwelling morphology. Recently, K.A. Ober (2003) found that reversion from arboreality to ground-dwelling has occurred in all the phylogenies she studied and concluded that 

Reversal may be a common evolutionary process, and evolution of new ecological interactions or evolution into new habitats may not inhibit further evolution or reversals. (Ober, 2003) 

Among other insects, weevils, a very large group of 60,000 species, being ancestrally (>200 million years ago) gymnosperm feeders have shifted to angiosperm hosts and then back to gymnosperms (Marvaldi et al., 2002). 

Experimental Reversion of Ancestral Characters

Experimental Reversion of Ancestral Characters in Drosophila

Strains of Drosophila melanogaster kept under laboratory conditions for decades (hundreds of generations) have diverged from the wild strain of origin in several biochemical, physiological, and life history characters.

20 populations of these laboratory strains were experimentally returned to the ancestral environmental conditions for 50 generations and then were compared with the control laboratory population and with the ancestral Drosophila populations. These populations reverted to various degrees (complete to incomplete reversion) to most of the lost ancestral characters (starvation resistance, reproduction time, developmental time, dry body weight, lipid content, etc.) within 20 generations (Teotonio and Rose,  2000; Teotonio et al., 2002)

Three different patterns of reversion to ancestral states for different characters were observed in these experiments. Flies reverted to ancestral type very rapidly (in several to 20 generations) for some characters whereas the reversion for other characters required up to 50 generations and for some characters the reversion was incomplete. The incomplete reversions were related neither to epistasis or linkage disequilibrium nor to the absence or insufficiency of genetic variation, as is indicated by the fact that experimental hybrids did not exhibit higher reversibility (Teotonio and Rose, 2000).

In earlier experiments selection for late life reproduction was associated with increased longevity and stress resistance. Reversion to the ancestral state of early life reproduction, occurred within 20 generations (Service et al., 1988) and by the 100^th generation reversion to the ancestral state was observed for all the studied characters (Graves et al., 1992).

Doubts have been expressed as to whether these convergences approximate the primitive state (Porter and Crandall, 2003) but, as argued earlier, no evolutionary reversion could produce phenotypes that are identical to the ancestral phenotype; resemblance and functional similarity rather than morphological identicalness is the determining criterion of evolutionary reversions.

NeoDarwinian Explanation

Elementary knowledge from genetics and evolution theory suggests that, in the above cases, the information necessary for “approximating the primitive state” is impossible to be acquired in an “evolutionary instant” of  20-50 generations. No gene mutations, gene recombinations, or changes in allele frequencies have been related to the rapid experimental reversions to ancestral characters in D. melanogaster under laboratory conditions. However, attempts to find a neoDarwinian explanation without invoking the above neoDarwinian mechanisms have been made. So, e.g., as for the pattern of rapid reversion to ancestral state of starvation resistance, investigators write:

The rapid reversion of starvation resistance and early fecundity indicates that they were under influence of pleiotropic alleles generating a negative genetic correlation between these two characters, a correlation demonstrated in a sibling analysis by Service and Rose (1985). As a result of this correlation, these characters rapidly moved toward their ancestral values during reverse evolution, because selection focused on early fertility” (Teotonio and Rose, 2001).

The authors do not elaborate on what this hypothetical “negative genetic correlation” consists in nor do they illustrate it with any examples of how it might work. With no changes in genes and in genetic information it is difficult to imagine why these “negative genetic correlations” arise in reverting populations but not the control populations. A tentative hypothesis for explaining the reversion (an unknown) with undemonstrated existence of pleiotropic alleles (another unknown), as a tautological statement, is devoid of explanatory power.

As for the slow incomplete reversion observed for some characters we are told that

The slow response of these characters and their eventual convergence on ancestral values, after more than 100 generations in the ancestral environment, may have been a result of mutation accumulation, because this process is expected to affect evolution noticeably after a considerable time. (Teotonio and Rose, 2001)

Not only the failure to show that such mutations have occurred, but even the textbook knowledge on the extremely low frequency of occurrence of gene mutations and the several orders lower frequency of “useful” mutations that would make the reversion possible, tells us that systematic reversion of these ancestral states in laboratory pooulations of D. melanogaster, did not involve mutational changes. This is definitely felt by the investigators themselves, for immediately thereafter they admit that

There is no data as to the effect (additive or epistatic) of particular novel mutations. (Teotonio and Rose, 2001).

Epigenetic Explanation

Discussing on the causes of slower tempo of some evolutionary reversions, investigators inquired whether this may be related to the lack of genetic variability and epistasis. Based on their hybridization experiments they conclude:

If lack of genetic variability was restricting reverse evolution, randomly mating hybrid populations should be freed of this constraint, because accumulation of identical genetic changes in populations of different evolutionary history is highly unlikely. Also, if epistasis led the derived populations to converge on strong evolutionary attractors, producing stasis under reverse evolution, the large perturbation to gene frequencies caused by hybridization should allow some stalled populations to escape from these attractor states. But the results showed no difference between uncrossed and hybrid populations. (Teotonio and Rose, 2001)

Lack of studies on the developmental mechanisms of the examined characters makes it impossible to reconstruct the epigenetic mechanism of evolutionary reversion of these characters in Drosophila. However, the exclusion of genetic factors (gene mutation, genetic variability, genetic recombination, and epistatic interactions) as possible causative agents of reversion to ancestral states of investigated characters leaves open the possibility of the involvement of epigenetic factors in these evolutionary reversion experiments.

There are several facts that that would justify focusing our attention on possible epigenetic factors as causal agents of evolutionary reversions induced by the return to ancestral conditions in laboratory strains of Drosophila.

First, the return to the ancestral environment of Drosophila populations after hundreds of generations under laboratory conditions, implies that these populations are subject to an environmental stress which, according to this epigenetic theory of evolution, is a universal trigger of the process of evolutionary changes in metazoans. As shown earlier, stress conditions sometimes lead to developmental instability and to behavioral changes, which generally are the first step in the process of evolution at the supracellular level. And, needless to say, the only known way environmental stressors influence the development and change of various characters in metazoans is via the nervous system (see chapter 8, Metazoan Response to Changes in Environment).

Second, the above experiments on the evolutionary reversion of various biochemical, physiological, and life history characters in Drosophila spp. unambiguously show that these evolutionary events take place rapidly (reversion to ancestral states in some individuals occurs within a few generations), contrary to what is predicted by neoDarwinian hypotheses. Empirical evidence in other species [phenotypic plasticity, predator induced defenses, experimental reversion of “hip glands” voles (see below), etc.] shows that the only way of inducing such “sudden” inherited changes in phenotypic characters is by activating inactive ancestral pathways via neuro-hormonal cascades starting in the CNS. Hence, it is plausible to assume that only conservation in inactive state of ancestral developmental pathways makes the extraordinary rapid reversals to ancestral characters in the Drosophila laboratory strains possible.

Experimental Reversion of “Hip Glands” in Voles

The presence of specialized sebaceous glands in the skin is characteristic of a number of mammal species, including small mammal voles. But two species of voles, Microtus pennsylvanicus and Microtus longicaudus lack specialized posterolateral skin glands (hip glands). They lost these glands in the course of evolution.

Experimental evidence shows that while losing the glands these species have conserved both genes and developmental pathways involved in the development of the glands. Sebaceous hip glands are induced to form when these “glandless” animals are hormonally stimulated. Three weeks after subcutaneous administration (by injection or implantation) of appropriate doses of the hormone testosterone, animals (7 of eight males of M. pennsylvanicus, and 6 of 8 males and all the 5 females of M. longicaudus) developed “hip glands”. Individuals of each species developed “hip glands” in species-specific regions: M. pennsylvanicus, toward the tail, in a region that is characteristic for Microtus montanus and individuals of M. longicaudus developed “hip glands” in both, more anterior (toward the flanks) and more posterior (toward the tail) positions. The investigator believes that this experiment “illustrates how structures are evolutionarily either gained or lost in steps” (Jannett, 1975). 

This spectacular example of induction of the reversion of an ancestral morphological character demonstrates that developmental pathways for developing “hip glands” in these vole species are conserved although the glands have been lost somewhere in the course of their phylogeny.  

NeoDarwinian Explanation 

If the experimental induction of “hip glands” would be considered to be an example of experimental reversion of an ancestral morphological character, as it certainly is, it refutes any imaginable neoDarwinian explanation; the emergence of “hip glands” took place within the life time of animals, thus excluding the involvement of neoDarwinian mechanisms (gene mutations, gene drift, and genetic recombination). At the same time, the experiments indicate that the evolutionary loss of glands was not related to any changes in genes.

Epigenetic Explanation 

The fact that a hormonal treatment induces formation in voles of a number of glands that the species has lost in the course of its phylogeny, proves that a certain level of the testosterone is both necessary and sufficient for the development of these glands. Given the position of the hormones of the peripheral endocrine glands as downstream elements of signal cascades along the hypothalamic-pituitary axes, the reconstructed developmental cascade responsible for the formation of the “hip glands” in other vole species may look as follows:

neural signals from medial cortex à hypothalamic GnRH neurons à pituitary FSH à testosterone.

Experimental induction of “hip glands” is an intragenerational process, implying that no changes in genes have occurred (voles are able to form “hip glands” after administration of testosterone) and the signal cascade suggests that the absence of “hip glands” in M. pennsylvanicus and M. longicaudus is caused by a block of one of the elements along the cascade.