The significance of this new phylogeny to the origin of amniotes lies in the recognition of an entirely new set of outgroups beyond diadectomorphs. The new phylogeny indicates that future phylogenetic studies of amniote phylogeny will have to include lepospondyls as outgroups. This has not been a common practice, but it will be necessary if the phylogeny presented here is accepted. Using lepospondyls could certainly help to polarize several characters used in phylogenetic studies of amniotes because they are very different from the seymouriamorphs that were previously used as outgroups.
Carroll (1970b, 1991) suggested that early amniotes showed evidence of having passed through a phase in which their skull was reorganized to be adapted to small size (snout-vent length 100 mm).
Carroll's is the most explicit functional scenario on the origin of amniotes and needs to be summarized before it is discussed. Carroll suggested that amniotes evolved from anamniote tetrapods that had direct development and laid their eggs on land, like plethodontid salamanders. According to Carroll, the size of an anamniotic egg is limited by the absence of extraembryonic membranes and by the absence of a mineralized shell because support and gas exchange would be problematic beyond a certain size (10 mm in diameter). Caecilians have large eggs, but this may be possible because of the elastic fibers in the egg capsule and the large filamentous gills of the embryos. Carroll suggests that external gills would not have been present in the embryos of the ancestors of amniotes because the gas exchange is performed by the allantois. These constraints on the egg size of the ancestors of amniotes, together with the correlation between egg size and adult size found in modern amniotes and plethodontids, suggests that our anamniotic ancestors would have been small and that their size could have increased only after the acquisition of the amniotic egg (according to Carroll, extant amphibians can only reach large adult sizes through a long growth period in a larval stage). This apparently explained why the groups suggested by Carroll as the earliest amniotes and their sister groups were generally small. Furthermore, the extensive ossification found in the vertebrae, carpus, and tarsus of early amniotes suggests that the adults were small, like plethodontids in which precocious ossification appears to result in small adult size (growth stops when all the cartilage is replaced by bone). Therefore, there were theoretical reasons to expect the earliest amniotes to be small (Carroll 1970b, 1991), and there was some support in favor of this scenario (such as the small size and extensive ossification of the earliest amniotes).
One problem with this scenario is that factors other than egg size may constrain plethodontids. The most obvious of these is the absence of lungs. Without lungs, plethodontids are restricted to small adult size because the surface/volume ratio of the animal must remain high enough to allow the animal's needs in oxygen to be accommodated by cutaneous respiration. Therefore, the small size of terrestrial plethodontids does not necessarily result from the small size of their eggs and from their direct development. Indeed, the polarity of this argument may be reversed. Direct development and terrestrial eggs are also found in some gymnophiones (Duellman and Trueb, 1986). The small size of many gymnophiones may yield further support for Carroll's scenario, but again, other factors may constrain the size of many gymnophiones. The fossorial existence of most of these tetrapods may select for a small size. Furthermore, the correlation between egg size and adult size claimed by Carroll (1991) for most tetrapods is not strong among lissamphibians (Duellman and Trueb, 1986).
The suggestion that lepospondyls are close relatives of amniotes may at first appear to lend support to Carroll's scenario because most lepospondyls are small. However, optimization of size on the tetrapod phylogeny indicates that lepospondyls are the only small relatives of amniotes and that small size (snout-vent length 100 mm) does not appear to be primitive for Amniota or larger clades that also include diadectomorphs and lepospondyls. Diadectomorphs, embolomeres, and some, if not all, seymouriamorphs are large. Utegenia, Ariekanerpeton, and Discosauriscus, the only small seymouriamorphs, seem to be only represented by incomplete growth series in which the adults are missing. It is possible that the relatively small size of the earliest amniotes may simply be coincidental or result from taphonomic biases. For instance, most amniotes found at Joggins and Florence, Nova Scotia (two of the localities in which the earliest amniotes were found) were preserved in hollow tree stumps. As the largest stumps had a diameter of only about 30-60 cm, the vertebrates preserved in them were necessarily relatively small. Even here, some of the amniote remains found at Joggins and Florence (Reisz, 1972) were only moderately small (centrum length 5-9 mm), with an estimated snout-vent length of up to 350 mm. In fact, the smallest presumed amniote from Joggins, Archerpeton with an estimated snout-vent length of 60 mm (Carroll, 1964, 1970b), is a tiny microsaur (R. Reisz, personal observation). A survey of Middle Pennsylvanian amniote bearing localities indicates that the size of amniotes found in these sites ranges from 130 to 300 mm in snout-vent length. Furthermore, Carroll's scenario was predicated on the assumption that the earliest known amniotes were basal, an assumption that is not supported by the most recent phylogeny (Laurin and Reisz, 1995).
Extended embryo retention unordered present H absent ililllill polymorphic
Figure 6. Evolutionary history of egg retention in extant choanates. This cladogram indicates that the primitive state for egg retention cannot be determined for amniotes or their extant sister taxon on the basis of the available evidence. Therefore, the hypothesis that the origin of amniotes included an intermediate stage in which anamniotic eggs were laid on land is not more parsimonious than the suggestion that the extraembryonic membranes evolved to facilitate extended egg retention.
Therefore, very small size is probably not a primitive character for amniotes and this aspect of the scenario proposed by Carroll (1970b, 1991) is not supported by the fossil record.
Carroll's hypothesis of transformation from the presumed basal amphibian to the amniotes includes a number of evolutionary modifications of the reproductive process. These involve internalfertilization, abbreviation and elimination of larval stages, laying of eggs in terrestrial locations; and development of extraembryonic membranes and shell. According to Carroll, all of these changes are more or less interdependent and must have occurred concurrently. However, he assumed that the evolution of the amniote reproductive pattern proceeded via an intermediate stage in which anamniotic eggs were laid in damp places on land. We are convinced that this assumption cannot be made because the evolutionary innovation of internal fertilization immediately raises the possibility of embryo retention and embryo-mother interaction (Lombardi, 1994). This is supported by the development of placentation in extant synapsids and frequent placentation in squamates, regardless of whether they lay eggs or not. It is quite reasonable to suggest (Lombardi, 1994) that some of the extraembryonic membranes evolved in an embryo-retaining form, as a pathway for fetal-maternal interaction, rather than a protection from the harsh external environment. Carroll's hypothesis of amniote origins also assumes that the primitive amniote condition involves the laying of eggs on land. Unfortunately, we cannot deduce the reproductive pattern of diadectomorphs, the nearest sister taxon of amniotes. We also have no evidence on the reproductive strategies of the numerous fossil members of the amniote clade because neither embryo-retaining specimens nor nests of Paleozoic or Early Mesozoic amniotes have been found. If we use a cladogram of extant tetrapods, with Latimeria and Dipnoi as outgroups, and investigate the evolutionary history of embryo retention among amniotes, we see the following ambiguous distribution pattern of the two character states (extensive embryo retention absent or present): The nearest relatives of tetrapods have both conditions, with Latimeria showing extensive embryo retention, whereas extant Dipnoi lay eggs at an early developmental stage. Lissamphibia have both conditions. Among extant squamates we see a bewildering array of reproductive strategies, with frequent placentation as well as extended embryo retention (Blackburn, 1992), but egg laying without extended embryo retention appears to be primitive for sauropsids because all other extant members of this clade crocodiles, birds, and turtles lack extended embryo retention. All extant synapsids (including monotremes and therians) show extended embryo retention. The resultant pattern does not allow us to determine the primitive condition for amniotes (Fig. 6). Therefore, the scenario that the evolution of the amniotic condition involved the intermediate stage of anamniotic eggs being laid on land is not more parsimonious than the alternative suggested here (that extraembryonic membranes appeared to facilitate extensive embryo retention). It is obvious that this aspect of the origin of amniotes needs further study.
The fossil record does not support the thesis that the earliest amniotes and their ancestors were small. If the evolutionary innovation of the amniotic egg permitted an increase in the size of the adult, as suggested by Carroll, the fossil record should provide a test for this hypothesis. However, the fossil record indicates that the adult size of amniotes does not show a significant increase in size from the earliest appearance of amniotes in the lower part of the Middle Pennsylvanian until the upper part of the Late Pennsylvanian. Therefore, the main adaptive advantage of the amniotic egg was probably not to allow increase in the size of the adults in the absence of extended larval stages but rather increased chances of survival of the hatchling by reducing the probability of starvation or predation. It is generally assumed that tetrapods primitively practiced a reproductive strategy involving large numbers of eggs laid in water, with little parental investment per egg. This is the presumed reproductive strategy of lepospondyls and seymouriamorphs. In contrast, the assumed primitive condition for amniotes involves a reproductive strategy of producing fewer eggs than their anamniotic relatives with greater parental investment in each egg, including nourishment and possibly parental care. Perhaps the origin of amniotes includes a shift between these two reproductive strategies. Therefore, though the origin of amniotes is probably associated with terrestrial reproduction, it may also be associated with a shift toward increased parental investment in each egg. Early amniotes may have laid larger eggs than their distant ancestors, and they may have laid them later if they had extensive embryo retention. Their hatchlings may have been larger and vulnerable to predators to a lesser degree or for a shorter period than the eggs and hatchlings of groups that laid smaller eggs and that lacked embryo retention. By investing energy into parental care, early amniotes would have ensured high survival chances to their offspring.
The new phylogeny and the scenario of the origin of amniotes associated with a shift in reproductive strategies do not specify when the amniotic egg appeared. However, according to the new phylogeny, the amniotic egg can only have appeared in the lineage that led to Amniota, no sooner than the divergence of this lineage from the lineage that led to lissamphibians, and at least slightly before the divergence between synapsids and sauropsids. Determining more accurately the timing of the appearance of the amniotic egg requires knowledge of the type of egg laid by diadectomorphs (or Westlothiania, if it is a close relative of amniotes). Unfortunately, fossil eggs are rare in the Paleozoic (and amniote eggs have not been discovered) and only indirect arguments may allow us to solve this question. Aquatic larvae are known in temnospondyls and seymouriamorphs, and lateral-line canal grooves are known in several groups of early terrestrial choanates. Therefore, the status of diadectomorphs may one day be established using such evidence. No diadectomorph larva is known, and as far as we know diadectomorphs lacked lateral-line canal grooves. These facts, as well as the large size of diadectomorphs, do not preclude the possibility that diadectomorphs may have laid amniotic eggs, but the fossil record is currently insufficient to safely settle this question.
The new phylogeny presented here is of course incomplete and more detailed phylogenies will no doubt appear in the future, but it is hoped that the present study will stimulate research on the anatomy and phylogeny of early choanates. The large amount of missing and partially uncertain data present in the matrix (Appendix 2) clearly illustrates that progress in this field will require more anatomical data.
Was this article helpful?