The flowering of avian orders after the massive extinctions at the Cretaceous-Tertiary Boundary is one of the most difficult, yet critical, subjects to recreate in the evolution of the vertebrates. Ideally, one would begin by studying fossil birds from early Paleocene deposits. So far, there is insufficient fossil material to accomplish this, but going back from the present evidence is fraught with traps and stumbling blocks.
Information on phylogeny is obtainable from at least two present sources: the whole-animal biology of the organisms and molecular biology. The most accurate phylogenies will result from those sets of data in which there is the closest agreement. Especially desired is information that can help explain the stages through which organisms have passed to reach their current condition.
To a large extent, however, cladistic analyses of phylogeny overlook biology and paleontology and focus, instead, on the analysis of large numbers of characters, which are employed without regard to possible convergences. The presumption, evidently, is that if enough characters are used, any complications resulting from convergent evolution will be swamped out and thus not be significant. However, because the fewer convergent characters that are included, the more accurate the analysis will be, it follows that characters suspected of convergence should be omitted. The results of pruning may be surprising and may show why some phylogenies are far off the mark.
A recent cladistic analysis proposing a sister-group relationship between the grebes (Podicipedidae) and the flamingos (Phoenicopteridae) (Mayr 2004) exemplifies this problem. Mayr used characters described in two of my papers (Storer 1982, 2000). In the first (Storer 1982), I pointed out that in both groups (and also the tinamous, gallinaceous birds, pigeons, sandgrouse, ibises, spoonbills, some cormorants, some falcons, and four of the nine suborders of the complex order Gruiformes), variable numbers of the thoracic vertebrae are fused into a notarium. According to my understanding, this structure has arisen independently in at least 10 phylogenetic lines of birds, presumably to strengthen sections of the vertebral column. In falcons, a notarium might prevent damage from the hard jolt in striking prey on the ground. In heavy-bodied but poorly maneuverable flyers (e.g., tinamous and Galliformes—an accepted convergence—and grebes), a stronger vertebral column could be advantageous in hard landings. In flamingos, it could mitigate the tendency of the downward pull of their long legs and neck to stretch gaps between the vertebrae during flight. Whatever the selective advantage might be, the presence of a notarium is not evidence for a relationship between flamingos and grebes.
The structure of the legs and feet and how these elements are moved are integral to understanding how locomotion evolved in grebes and flamingos. In most birds (and tetrapods), the joints between the phalanges and between these and the tarsometatarsus are hinge-like; those of the grebes are unique among living birds in having the beginning of a ball-and-flange type of joint, which provides rotation between these joints. This type of joint was carried to an extreme in Hesperornis (Marsh 1880), which provides another obvious case of convergence.
J. Fjeldså (pers. comm.) and I (Storer 2002) independently concluded that the foot structure of the protogrebe arose as an adaptation for moving through dense stands of upright, hard-stemmed vegetation by rotating the toes 90° so that the side of the foot passed through the vegetation first. The unwebbed toes were then rotated back to their original position and separated and placed on the ground, permitting some of the reeds to pass between them. When the grebes moved to an aquatic habitat, the toes became lobed and the movements of the feet used in passing through vegetation became adapted for foot-propelled swimming. If so, the evolution of grebes' feet from those of other swimming birds would have been impossible. With regard to flamingos, Mayr (2004) proposed ancestry from a group of swimming birds. However, if the above explanation of the evolution of grebes' feet is valid, grebes and flamingos could not have come from a common swimming ancestor. If, on the other hand, flamingos were derived from shorebirds, as paleontological evidence indicates (Feduccia 1999), their webbed feet (as in some shorebirds; e.g., avocets) could have evolved primarily as support for foraging on a soft substrate.
The one character used by Mayr that is unique to both the grebes and the flamingos is the presence of a chalky covering of the eggs. Its function is unknown. Both groups build their nests in water—the grebes on floating vegetation, the flamingos on mud in shallow water—and in some cases, the eggs are subject to immersion and possible clogging of pores in the shell, which would suffocate the embryo. Other birds using similar habitats (e.g., jacanas) lack this shell structure. Its origin and primitiveness, then, are unresolved; but if convergence is involved, the character has no use in phylogenetic studies. An analysis omitting this structure and the notarium would, ipso facto, diminish support for a sister-group relationship between the flamingos and grebes.
Data from parasites are also relevant. Usually, a parasite is acquired by a new definitive host when the latter ingests an intermediate host of the parasite and the parasite becomes able to reproduce in the new host. Mayr's (2004) notation that all the known definitive hosts of the cestode family Amabiliidae are grebes, with the exception of a single species (in a very distinct subfamily) in flamingos, cannot be used as evidence of a phylogenetic relationship. The intermediate host (or hosts) for the amabiliid species in flamingos is unknown, but of 13 intermediate hosts in grebes, 11 are dragonfly or damselfly nymphs (Odonata) and the other two are a mayfly nymph (Ephemeroptera) and a water boatman (Corixidae, Hemiptera); all of these insects are aquatic. In a phylogenetic study, the parasite character can be used only after crossing-over is eliminated as the means by which members of the same family of parasites came to parasitize members of both definitive host groups. Van Tuinen et al. (2001) were also in error in using this character as evidence of a phylogenetic relationship. In this case, the odds favor the likelihood that any crossover would have occurred after both grebes and flamingos became aquatic.
Several figures in Mayr's (2004) paper require discussion. In figure 2, character 46, the ratio of length to width of the basal phalanx of “major digit of the wing” is a proportion. Because its boundary can be set at will, the boundary should be justified. In figure 3, the tubercle on the cnemial crest of the grebe is not clearly depicted, and the structure connected to the tubercles in the three species shown should be identified to confirm that it is the same in all groups. In character 57, the crests on the hypotarsus are present in both grebes and loons as well as flamingos. This suggests that it is advantageous in diving birds and therefore convergent. Apparently, Mayr was unaware that the hypotarsal canal for M. flexor perforatus digiti II is present in the primitive genera of grebes but not in Podiceps and Aechmophorus. This figure should have illustrated the primitive rather than derived condition, which is irrelevant in this context. There are other inaccurate representations of skeletal characters that lead the reader to suppose that skeletons of grebes and flamingos are more similar than is the case. In grebes, for example, the fourth toe is the longest; but in figure 4, the fourth is shoved between toes two and three, making the proportions appear similar to those of flamingos and most other birds.
There are two basic problems with Mayr's (2004) paper. The first involves the author's failure to use the mass of basic natural-history information relevant to the subject. He failed to appreciate that crossing-over is a common way for a parasite to move from one definitive host to another. Similarly, the inability to explain how the structure of the feet could change from webbed in flamingos to the far more complex structure of grebes' feet, or vice versa, indicates that any such connection must have occurred before either line took to the water. These points weaken his argument for any sister-group relationship between grebes and flamingos. The second problem is his not considering whether convergence might be involved in some characters, such as the notarium. Convergence is important, and not screening for possible examples before making an analysis is like failing to remove the rotten apples before filling a barrel for storage.