Figure 2.5. Cladograms and phylogenies: a cladogram for two taxa, with the possible phylogenies that might be implied.

Note that characters 5 and 6 define a group that is inconsistent with the rooted cladogram. One might conclude that the inconsistency falsifies the hypothesis of relationship derived from the majority of the characters. We might favor the clado-gram as illustrated in Figure 2.7, as the information from four characters defines the cladogram, whereas the inconsistent hypothesis is defined by only two characters. This is often characterized as the adoption of the least refuted hypothesis (e.g., Lynch 1982). It is used typically by those who employ nonnumerical algorithms to generate rooted cladograms. These inconsistencies are the crucial problem in the resolution of evolutionary relationships and are discussed further below (see Homoplasy: The Fundamental Problem)

Rooting the cladogram. The construction of a root for a cladogram is one of the most difficult problems of phylogenetic systematics. One needs a criterion to locate the part of the cladogram that bears the most ancestral states. This requires a criterion for identifying ancestral character states. Two main ones have been employed: outgroups and ontogeny. In a three-taxon case, one taxon is more distantly related than two others that are more closely united by synapomorphies. Cladograms can be constructed by successive additions based on three-taxon statements (see Nelson and Platnick 1981; Wiley 1981, but also see Nelson and Platnick 1991; Platnick, Humphries, Nelson and Williams 1996). The choice of an outgroup involves picking a character or set of characters that is widely agreed to have more ancestral states than is present in the first three taxa. Thus, an outgroup for the bivalves might lack a shell but have spiral cleavage and a mantle. Its attachment to the cladogram defines polarity.

Figure 2.7. A matrix of character states by taxa for a hypothetical group. For each character, 0 is taken to be ancestral and 1 is derived. At right is a Hennigian analysis for characters 1 through 4, which are compatible. Character numbers on tree define groupings delineated by the next highest node (e.g., group [A,C] is defined by character 1; [A,C,E] is defined by character 2).

Figure 2.7. A matrix of character states by taxa for a hypothetical group. For each character, 0 is taken to be ancestral and 1 is derived. At right is a Hennigian analysis for characters 1 through 4, which are compatible. Character numbers on tree define groupings delineated by the next highest node (e.g., group [A,C] is defined by character 1; [A,C,E] is defined by character 2).

The use of ontogeny derives from Haeckel's biogenetic law that ontogeny recapitulates phylogeny (Nelson 1978). The reality of the biogenetic law has long been in hot dispute (see discussion in Gould 1977). This criterion could only have utility if evolution by terminal addition occurs and either descendants comprise a simple addition of the stages or a form of acceleration compresses the stages into the same developmental period. Alternatively, an extension of Karl von Baer's laws of development into evolution (von Baer didn't believe in evolution) pose hypotheses of polarity (Nelson 1978; Patterson 1982). General (ancestral) features are believed to occur early in the ontogeny of related taxa. Later ontogenetic stages represent specializations (derived states). This hypothesis supposes implicitly that early stages of ontogeny are less subject to evolutionary alteration than later stages. See chapter 4 for further discussion of this issue.

Ontogenetic considerations show that apparent ancestry cannot be identified under certain conditions (Fink 1982):

1. When the common ancestor of two taxa evolved state b by adding a stage to the ontogenetic trajectory, but one of the two descendants went to state a by loss of the terminal state b

2. Same as condition 1, only ancestor exhibits acceleration and one descendant shows slowing of development

3. Same as condition 1, only contrasting a movement up in onset of development in an ancestor, followed by relative delay of onset in development of the descendant

These three conditions will erase the record of character polarity. Any shuffling of stages within a sequence would destroy the directional utility of the ontogenetic order of the character states in descendants.

Alberch and Gale (1983) have investigated ontogeny in frogs and salamanders and demonstrated that developmental regularities might be a valid key to character state sequence. During development, digit number one is the last to appear in the frogs Xenopus laevis, whereas digit number five is the last to appear in the axolotl, Ambystoma mexicanum. This seems to correspond to evidence for evolutionarily derived digit loss. The last digit produced during ontogeny is that one which is lost first. This cannot be used to determine polarity, but at least a predicted sequence defining a linear order of character states in evolution might be established from such data. Evolution could go in either direction along the sequence. Another encouraging example is McGowan's (1984) study of the development of the avian tarsus. It had been previously suggested that ratites were derived from carinates (flying birds) by an arrest of development. Carinates, however, can be shown to have an ontogenetically unique pretibial bone, whereas the ratites share the ascending process of the astralegus with the theropod dinosaurs. Ratites therefore are in a relatively ancestral state and their ancestral stock is thus more ancestral than that which defines the flying birds.

It is likely that many evolutionary sequences involve terminal addition - or at least resolvable alterations of developmental sequences. Alterations of ontogenetic sequences, particularly terminal addition, have been suggested in several studies of fossil mollusks (Fisher, Rodda, and Dietrich 1964; Miyazaki and Mickevich 1982;

Newell 1937). The evolutionary adjustment of ontogenetic patterns in the tropical American salamander genus Bolitoglossa is another example. Hand and foot morphology of the species represents all stages of intermediacy between the slightly webbed, large-digited structures of the upland species and the diminutive fully webbed small-digited ones of the lowlands (Wake and Brame 1969). Diminutive lowland species are paedomorphic and seem to result from the retention of juvenile characters of the larger ancestors in the adults of the smaller descendants. The webbing and small size seems adaptive for the relatively more arboreal habit of the lowland forms (Alberch 1981; Wake and Brame 1969).

In his functional study, Alberch (1981) identifies the plesiomorphic state for the genus Bolitoglossa by an outgroup comparison. The outgroup is intermediate along the ontogenetic track, relative to the two derived species he considers carefully. Thus, the root of the network (not done by Alberch in quite this manner or with this terminology) is near a point where species bear intermediate character states of the ontogenetic track. We must assume, therefore, that reversals are possible; one can move backward or forward along an evolutionary-ontogenetic trajectory. But suppose that we had no good outgroup for comparison. It is likely that the network might be rooted (if only ontogenetic characters were employed) near the taxon with the greatest representation of early ontogenetic character states. Given our information, this could lead to incorrect judgments about the history of the group. We can imagine two closely related sister taxa whose difference rests on one synapomorphy. If most of the useful characters are associated only with ontogeny, and one cannot be sure as to the ancestral state of any character, only the sequence, then it may be difficult to root one group with respect to another. In other words, a conflict in rooting cladograms might arise between the use of outgroup comparisons and ontoge-netic character sequences.

If there is a correlation between the order of ontogenetic stages and the strati-graphic sequence, then we would be justified in invoking fossil sequence as corroborative evidence favoring rooting of the network near the species that is both stratigraphically oldest and ontogenetically "earliest" (Miyazaki and Mickevich 1982). This conclusion is based on the sensible argument that it is more likely that evolution has proceeded forward in time, rather than backward.

The Problem of Ancestors. The root of cladograms provides information on the nature of ancestry but does not define ancestors. Consider the cladograms and possible phylogenies in Figure 2.5. In many cases, the ancestor will never be identified, simply because no independent criteria would delimit a choice among the possible phylogenetic hypotheses. In these cases, the study of macroevolution is restricted to the study of the possible mode of transition from a taxon to its closest relative. We can only speculate about what combination of character states ancestral transitional taxa might have borne.

0 0

Post a comment