There are a number of vague areas in vertebrate phylogeny, but few so unsettled as the base of the neornithine radiation. The problem is that the early fossils are mosaics. That is, the relationships don't look like successive beads on a string. They look like a fish net, with all kinds of impossible lateral connections. For example, the one Paleocene bird fossil we have in abundance is Presbyornis. Presbyornis has a head like a duck, including the flat, upturned beak, tweezer-shaped mandibles, long rounded cranium, and other features including the peculiar "double-piston" filter-feeding system of ducks. But ducks are tied to chickens by covalent bonds. That is, their DNA is so similar (both by hybridization and sequence tests) that they have been united by some into a single taxon, the Galloanserae. However chickens are closely related to tinamous and the ratites by post-cranial anatomy, particularly the structure of the clavicle.
On the other side, the post-cranial anatomy of Presbyornis looks a great deal like a long-legged shore bird, particularly a flamingo. Flamingos are conceded by (almost) all to be closely linked to pelicans, albatrosses, loons, probably penguins, and the like -- the charadriomorph lineage. As much as Allen Feduccia has been reviled in other contexts, perhaps unjustly, he has convinced most ornithologists of the importance of these shore birds in the initial avian radiation after the KT extinction event. In fact, he and others have assembled a convincing "pseudo-phylogeny" or transformation series between Presbyornis's duck-like feeding and the advanced filter-feeding apparatus of flamingos.
On the third hand, so to speak, Presbyornis seems not too distantly related to the gruiforms: phorusrhacids, cranes, rails and their first cousins, the grebes. Presbyornis itself is not a gruiform, but their common ancestor could not have lived long before the Paleocene.
So just what is Presbyornis? Phylogenetically, it was most likely a duck, although a very early duck. Physically, it looked a little like the Roadrunner -- not like real roadrunners, but like the cartoon Roadrunner, the effortless champion over gravity, air resistance, logic, fate, technology, and the execrable Wile E. Coyote, Certified Genius. Unlike roadrunners, real or animated, Presbyornis was highly colonial and apparently gathered in large flocks to feed by the shores of lakes. Several species are known ranging from perhaps 0.5 to 1.5 m in height. The large open area in front of and below the orbit suggests that it may have had a salt gland. In fact, its remains are found at many levels in the Eocene Green River lake system of Wyoming and Utah, reflecting feeding in both brackish and fresh water environments over long periods of time. Similar remains are known from Mongolia and elsewhere.
Unlike modern ducks, Presbyornis had long legs and presumably was a wader by choice, rather than a swimmer. The upper surface of the upper jaw overshoots the framework of the jaw itself, leaving a sort of ridge along the outside of the upper jaw. It has been suggested that this used to scrape algae off logs and debris in shallow lakes. However, its principal diet was likely based in filter feeding, using the tongue to draw bottom water in like a straw, extracting food by pressing the tongue against the roof of the mouth, and expelling the excess water out the sides. The relative length of the neck and legs of Presbyornis are quite consistent with a wading, bottom-feeding style. However, compared to a flamingo, the feet are large and set relatively far back. That is, Presbyornis might have been capable of limited diving, as well as flight and wading. Alternatively, it may simply have been adapted as a wading filter-feeder on rocky or uneven bottoms. In any case, Presbyornis was a versatile and successful bird.
One of the interesting features of Presbyornis is its very lack of specialization. It flew, it waded on varying surfaces in fresh and salt water, it may have swum and dived. It scraped and filter fed. It did so for millions of years until the Green River lake system itself dried up and disappeared. Most of the modern "orders" of birds had already developed by the Eocene. Environments like the famous London Clay deposits of Essex boast a wide variety of Eocene birds. However, it seems that the avifauna of at least one major lake system remained dominated by an ancient generalist, despite dramatic changes in both the regional climate and the lakes themselves.
On the basis of this limited data, we may make two observations. First, species sorting did not occur. Species sorting is a concept which is used in discussions of punctuated equilibrium, for example, to explain sudden changes in a taxon. The idea is that species will gravitate toward their optimal environments. What appears to be an evolutionary leap may be no more than climatic change followed by migration. Here, the climate and lake conditions changed dramatically over 5-10 My without any significant -- much less dramatic -- change in avifauna. The same cannot be said of the aquatic lake fauna: fish, arthropods, and so on, much of which shows both intra-lake migration and replacement in response to salinity and climate change.
One reason that sorting did not occur in Presbyornis is likely that the lakeshore environment is inherently discontinuous. No migration into or out of an interior lake system is possible without traversing geographic barriers. However, this may not be an adequate explanation. Fish populations, for example, are recruited and replaced throughout the history of Green River even though the barriers to entry (to borrow a term from economics) may be nearly as high. Thus, an additional mechanism is proposed here: a second observation, also borrowed from economics. Taxa which dominate an ecosystem guild demonstrate substantial genetic and ecological inertia analogous to monopoly power in an economy.
Detailed discussion of this phenomenon is deferred to another day. However, this inertia involves both resistance to competition from immigrant species and resistance to change in response to selective pressures inside the ecosystem. Very briefly, a dominant species occupies all of the ecological space in its niche. An immigrant competitor, in addition to overcoming any barriers to entry, will face competition from a population which is at the maximum carrying capacity of the ecosystem. The immigrant may be more efficient. However (a) it will be swamped in numbers; (b) it will spend an unreasonable amount of time and energy in interactions with the native species; and (c) it will lack the learned or inherited microadaptations to the particular system -- it doesn't know the territory. Internally, the resident species will not change because of stabilizing selection. All but the rarest genetic changes will be drowned in the gene pool, so to speak. One might also speculate that in the case of colonial animals, infant mortality is far more determined by the fitness of the flock, herd or tribe, than by individual fitness. If so, the only genetic changes likely to survive are those related to relative mating success, such as display and dominance behaviors.
This essay has, unfortunately, wandered far from its intended path. The principle point is that the last common ancestor of ducks and chickens seems to have been at or near the base of a rather broad radiation which occurred at or around the time of the KT event, likely from a shore bird with long legs and a generalist habit. Presbyornis is the closest morphological and ecological model to this transitional form which is represented by a reasonably good fossil record. It shows a number of specializations for filter feeding which were probably not present in the transitional shorebird ancestor. However, a small flock of something that looked quite a lot like Presbyornis may well have looked up from pecking at a dead fish on the beach about 65 My ago to see the horizon darkening with ash and dust -- and then simply rode out the storms and bad seasons as they had so often done before.