Sibley's Classification of Birds
by Eric Salzman
Birding, December, 1993
How are birds related? This is not just an academic question. Every living thing is the end-product of an evolutionary history and carries the evidence of that descent. Until recently there were only two ways to reconstruct this history: from the fossil record and from the constitution and distribution of living things. Now there is a third: the comparative studies of DNA and its byproducts, which form the basis of the work of Charles Sibley and his collaborators.
Compared with microbiology, ecology, and field work, taxonomy has not been very sexy, at least until now. But no one with a serious interest in birds can ignore the big picture from the micro-shuffle of races and species to the macro-levels of families, orders, and subclasses -- even if it is just a matter of finding your way around the field guides, keeping up with what you can count on a life list, or trying to make sense of a new and exotic avifauna.
The classification of birds and other living things is based on a few simple premises. Life on earth has a single origin. All birds are of a feather -- descended from a reptilian or reptile-like ancestry. The key to the reconstruction of the evolutionary history (or, the phylogeny) of the 9000 to 10,000 living kinds of birds is getting a satisfactory picture of how they are related to one another.
Many students have tried to find that grand design. The modern classification of birds goes back to 1867 and the work of Darwin's great disciple, Thomas Henry Huxley. Huxley (1867) was followed by, among others, Hans Friedrich Gadow (1891, 1893) and Erwin Stresemann (1934) in Germany, and Alexander Wetmore (1951) and Ernst Mayr and Dean Amadon (1951) in the US. These men all proposed avian classifications based on physical characters supplemented by behavior. Anyone can recognize a parrot, and it is immediately obvious that all parrots are more closely related to one another than any of them is to anything else. In this case, the physical relationships are external and easily seen. Other relationships are less obvious. Swallows are unrelated to swifts, and hummingbirds are unrelated to sunbirds, but swifts and hummingbirds are distant cousins of each other; the evidence for this relationship comes more from internal physical characters -- anatomy rather than external morphology. Two Empidonax flycatchers that show almost no consistent physical differences may frequent different habitats, sing different songs, build different kinds of nests, and refrain from interbreeding; behavior keeps the birds apart and populations that live together and do no interbreed are good species by definition.
Nevertheless, form (morphological characters, physical structure) or form-plus-function (physical characters plus behavior) do not always provide enough information. Interbreeding or the lack thereof is useful only at the species level and for overlapping populations. Closely related species may take up different modes of life and drift apart in appearance and lifestyle while unrelated creatures with similar lifestyles may come to look alike -- the well-known phenomenon of convergent evolution. There is no certain way to know if a shared character is the result of common ancestry or an artifact of convergent evolution.
Selection, natural or otherwise, works on the individual, on the so-called "phenotype" (the realization of one organism's genetic program as shaped by its environment), by removing or not removing its particular genotype from the population. In most natural circumstances, selection pressure produces a fairly stable phenotype that is well adapted to its environment but a genotype that contains far more information than is actually used at any given time. Some of this represents the potential for change, and even small genetic differences can be reflected in big structural or behavioral differences (the reverse is also true, that big genetic differences may only add up to small differences in the phenotype). But, surprisingly, most genes and most genetic changes are neutral; that is, they have no discernible effect on the phenotype and they are relatively immune to selection pressure. Therefore, most genetic change in most populations is slow and governed by the laws of chance. Accidents, "misprints," or "slip-page" produce the phenomenon known as "genetic drift."* Allowing for certain adjustments, this rate of drift averages out over time to a constant rate. If this is so, then the comparison of DNA from different species would give an objective view of the amount of time that has passed since these species embarked on their separate evolutionary paths, and a hierarchy of relatedness could be discovered.
The theory is simple enough, but the sheer amount of information in the genetic material is vast, and the sheer magnitude of the task of analysis is challenging. The solution devised by Sibley and his cohorts is the so-called DNA-DNA hybridization. Thanks to Maurice Wilkins, Rosalind Franklin, Francis Crick, and James Watson (the last-named a birder, by the way), we know that DNA takes the form of two intertwined helical chains or strands -- the well-known double helix -- bonded together by cross-beams; the structure has been compared to a twisted ladder with rungs binding the two sides firmly together. These rungs can be melted by high heat, permitting the strands to be separated; cooling them down permits the rungs to reform and the strands to reunite. Hybrid DNA molecules are made out of separated strands form different species that are then allowed to unite -- the so-called "hybridization." The rate and extent to which these unmatched strands can unite provides an objective measure of similarity or difference.
There was one big hitch with this technique. Sibley and his colleagues began with the assumption that the rate of change in the genotype was universal and statistically predictable over time. But species that begin breeding early accumulate changes at a faster rate than those that mature more slowly, requiring what we might call a "length-between-generations" correction factor. Even with this complication, the DNA-DNA hybridization technique proved to be remarkably simple, consistent, and objective. At least this is its strong claim.
Between 1975 and 1986, Sibley, his associate Jon E. Ahlquist, and various assistants produced thousands of these DNA-DNA hybrid molecules between and among some 1700 species, representing all but two or three of the families of living birds. The results have been published in two hefty volumes from Yale University Press: a 976-page Phylogeny and Classification of Birds: A Study in Molecular Evolution (1990) by Sibley and Ahlquist, and a 1111-page volume on the Distribution and Taxonomy of Birds of the World (1990) by Sibley and Burt L. Monroe, Jr., chairman of the AOU Checklist Committee on Classification and Nomenclature and former president of the AOU.
Since the days of Linnaeus (1758), the basic unit of biology has been recognized as the species, usually defined (in some way) as a breeding population or a group of populations that interbreed; closely related species are grouped into the higher category of genus (plural: genera). To choose an obvious example, there are many closely related thrushes around the world in the genus Turdus; only one, the species migratorius, occurs in most of North America. Turdus migratorius is, of course, the American Robin.
So far, so good. But what do we do with birds that we call thrushes but that differ in essential ways from Turdus? Clearly there are a number of genera that make up the thrush family; but how inclusive should this family be? Are the obviously related chats, European robins, and nightingales also thrushes? And what about the Old World flycatchers, warblers, kinglets, and babblers? The "true" thrushes, Turdus and relatives, turn out to be only a subfamily in a huge assemblage of hundreds of species and any number of families and subfamilies. The issue becomes one about drawing lines. The more subtle the system, the more divisions: subspecies and superspecies, subfamilies and superfamilies. Finally, all the thrushes, all their allies, and all the other songbirds (or perching birds, the passerines) make up an order. There are a number of orders, sometimes themselves divided into suborders or combined into superorders, making up the class of animals with feathers that we call Aves, or birds.
The family of birds might be shown in the form of a tree or as a genealogy. Trees grow from the ground up; genealogies work from the top down (there are other representations that read like books, from left to right). In these representations, the older, more "primitive" forms are meant to come first and the more recently evolved ones later. Most recent reference books, including modern field guides, begin with penguins, ratites (Ostrich, rheas, Emu, cassowaries, kiwis), loons, grebes, and tubenoses, and they end with perching birds (finches or corvids depending on whether the point of view is American or European). Old and primitive lead to modern and complex. Closeness on the list signifies closeness of relationship.
In 1960, Alexander Wetmore divided living birds into two superorders, one consisting of penguins, the other of everything else. This arrangement, often referred to as Wetmorean, is typical; it is the arrangement of orders and families that you now see in the AOU, ABA, and Clements checklists, for example. North Americans (who have lots of sparrows and bunting) put the finches last, arguing that seedeating is the latest and fastest-growing trend in bird evolution,. European-trained systematists prefer to end up with the corvids, bowerbirds, and Birds of Paradise; they argue that intelligence, architectural ability, and complex behavior represent advanced evolution. Europeans also lump the thrushes and their relatives into one huge family, the Muscicapidae. Other ornithologists hold that a finch is not always a finch and split them into weavers and Old World sparrows, waxbills, northern or "true" finches, and finally, the American sparrows, buntings, and southern grosbeaks lumped with the wood warblers, tanagers, and icterids in that massive pile-up known as the "New World nine-primaried oscines." Check your field guide. Peterson sticks pretty closely to Wetmore. The National Geographic Society field guide follows the pre-Sibley revisionists. These views -- now incorporated in the AOU and ABA checklists -- are, however, only cosmetic compared with what is being put forth as the result of the DNA-DNA hybridization studies.
What follows here is a quick and vastly simplified overview of what Sibley and his co-workers are proposing, with brief notes on some of the implications. The birds themselves are the same, but our way of looking and thinking about them has been altered, and the birder who is interested in the larger issues of bird evolution, kinship, and biogeography should be fascinated and stimulated.
The first thing to notice is that the bird universe has been carved up into many more categories and subdivisions than we have been used to. In this system there are -- reading from top to bottom -- subclasses, infraclasses, parvclasses, superorders, orders, suborders, infraorders, parvorders, superfamilies, families, subfamilies, even tribes and subtribes, and all before we get to the level of genus and species.
Right off the bat, the subclass of living birds (Neornithes) is divided up into two Infraclasses and seven Parvclasses. There are big differences between these groups, and, therefore, their ancestry is considered to be very ancient. One should not jump to the conclusion that, for example, gamebirds and waterfowl are closely related -- although the fact that they might share a common ancestor is of interest. If we follow the breakdown to the lower levels -- not shown here -- we find that:
* All of the waterfowl, except for screamers, Australian magpie-goose, and whistling-ducks, belong to a single family;
* American (New World) quail are in their own family separated from all the other quail, grouse, and pheasants (which are together in a different family of their own);
* Woodpeckers are distant relatives of barbets and toucans are not as close to jacamars and puffbirds as previously thought;
* Jacamars, puffbirds, hornbills, hoopoes, and trogons each get an order of their own;
* The families of rollers, cuckoo-rollers, motmots, todies, bee-eaters, and three families (!) of kingfishers make up another single order;
* American barbets are in a separate family, not very closely related to African and Asian barbets; toucans are merely a variety of American barbet.
At the end of the line is the Parvclass Passerai -- "all other living birds". Note that the mysterious South American Hoatzin is, as has been proposed, a kind of cuckoo; the parrots keep their unity; and the swift/hummingbird alliance is confirmed, as is the owl/nightjar/frog-mouth/oilbird/potoo connection. The African touracos may also belong here, but a lot of the data appear to be ambiguous.
Again, let's go to the end of the line. The superorder of Passerimorphae ("everything else") breaks down. This is almost too neat: four orders including a fascinating waterbird/hawk complex. The idea that the passerines or perching birds form a single huge order, the Passeriformes, is well known and the DNA work confirms it. Sibley now proposes a completely parallel order of water birds -- if you don't mind thinking of hawks and vultures as water birds -- the Ciconiiformes, or stork-like birds. This scheme deserves a closer look. Shorebirds come first.
Not unexpectedly, jaegers, skimmers, and larids (gulls and terns) -- and, unexpectedly, alcids -- are all in a single family. They are distant relatives of the plovers, oystercatchers, avocets, and stilts, and even farther away from the snipes and sandpipers. A closer look reveals a few oddities. Great Auk and Least Tern are (or were) kissing cousins, and both are distant relatives of the Piping Plover and Black-necked Stilt; and all of these are more closely related to each other than any of them are to any sandpiper. These arrangements may yet be subject to reshuffles, but the message is clear: Shorebirds are an older and more diversified group than their appearance and way of life might suggest.
The relationship of water birds to the diurnal raptors is even more striking and unsuspected. Classifying American vultures with storks is not the most radical thing about this rearrangement of life forms. There used to be widespread agreement that penguins, grebes, loons, and tubenoses belonged at or near the base of the avian family tree and that pelicans, cormorants, darters, boobies, and tropicbirds formed a single order, also low on the totem pole. Sibley et al. have quite a different view. If they are correct, then all these birds, along with the pigeons and doves and the bustard/crane/rail complex, are not at all "primitive," but, in fact, share a common ancestor with the passerines!
This brings us to the great order of perching birds. Sibley and Monroe are generous in their estimates: 5712 species, constituting 59 percent of their total of 9672. They do not disagree with their predecessors in regarding the passerines as a relatively recent and not highly diversifies offshoot of the great tree of avian life. All of the songbirds are much more closely related to one another than, say, swifts are to hummingbirds. Like their predecessors, these authors also divide the passerines into suboscines (New Zealand wrens, Old World pittas and broadbills, New World flycatchers and cotingas, and furnarids), and oscines ("songbirds").
The New World flycatchers are united with the mourners, tityras and becards, cotingas, and manakins into one big family, supporting what many ornithologists have long advocated. But the "real" antbirds are now separated from the various antthrushes and antpittas and these so-called "ground antbirds," along with the little-known gnateaters and the tapaculos, are placed nearer the ovenbirds, spinetails, and wood-creepers that they are to the other antbirds.
Sibley and his co-workers divide the true passerines or oscine songbirds into three Parvorders: a huge Australasian group, a small and little-known African group, and everything else. Here, first of all, is the evidence for the true uniqueness of the Australasian avifauna. All these birds appear to be descended from ancestors who were stranded Down Under when Australia and New Guinea were separated from the rest of the world a long time ago. As their names suggest, many of these creepers, wrens, robins, flycatchers, and babblers used to be thought of as outlying relatives of widespread groups. Instead, they turn out to be textbook examples of adaptive radiation and convergent evolution.
Most of these birds stayed in Australasia, but a few made it back into the rest of the world: the widespread shrikes, the corvids, and, most surprising of all, the American vireos and greenlets. The ancestral home of crow, magpies, and jays is Australia, although, oddly enough, the crows and ravens found there today are recent colonizers from Asia. No vireos and greenlets live anywhere in the world today except in the Americas.
In this scheme, the birds of paradise are close relatives of the crows and jays (along with the wood-swallows, Old World orioles, and cuckoo-shrikes) but not of the lyrebirds or, as was formerly widely believed, of the bowerbirds. The fantails and monarchs, almost invariable regarded as flycatchers, also turn out to be corvids of a kind. These birds, along with drongos, ioras, leaf-birds, and bush-shrikes, have also pushed out of Australia into nearby Asia and Africa.
Most of our familiar northern temperate songbirds belong to the other main branch of the oscines -- the so-called passerids (Passerida). Like Gaul, the passerids are divided in three parts. The muscicapoids (or Musicapoides) break down into families and subfamilies. Here are some old notions confirmed and a lot of new ideas. The close relationship between the Old World starlings and the New World mockingbirds, catbirds, and thrashers is a real surprise. The fact that they are in the same superfamily with the thrushes, Old World flycatchers, dippers, and waxwings is somewhat less surprising. Notice that the so-called chat-thrushes -- including the European Robin and the nightingales -- are more closely related to the Old World flycatchers than they are to the "true" thrushes. Old World warblers, babblers, kinglets, bulbuls, and their allies are conspicuously and startlingly absent. To find them, we have to turn to the next superfamily, the Sylvioids.
A lot of apple carts are upset by this collection of ten closely related families. Gnatcatchers are not, as previously thought, a New World offshoot of the Old World Warblers, but are relatives of creepers and wrens. Kinglets get their own family, as do cisticolas and prinia ("African warblers"). So do nuthatches, tits, long-tailed tits and bushtits, bulbuls, white-eyes, and (in somewhat surprising company) swallows and martins. The rest of the Old World warblers are in a single family with the babblers. At the subfamily level, the babblers (minus the laughing-thrushes but including our Wrentit) are merged with the Sylviini or Mediterranean warblers. The fact that the Wrentit lives, looks, and behaves like a Dartford Warbler may turn out to be more than coincidence or mere convergent evolution!
We are approaching the highest and bushiest branches of this family tree -- the passeroid superfamily. Larks, as well as sugarbirds, flowerpeckers, and sunbirds, turn out to be relatives of the finches.
What then are finches? Seedeating birds with conical bills suitable for cracking open seeds have always been regarded as finches, a term that traditionally includes sparrows, buntings, grosbeaks, and the like. All the finches were considered to belong to the same family or were divided into two groups. Modern systematists have tended to take the view that finches evolved several times from different ancestors in order to exploit the widespread availability of a good food source. Weaver finches (including the House Sparrow) became just plain weavers and were in turn split off from the Estrilda finches or waxbills. As we have seen, the American sparrows, all the buntings, and most of the grosbeaks (excluding a few of the northern ones) were lumped with the American wood warblers, tanagers, and icterids. The term "finch" became restricted to a relatively small group of northern seed-eaters. This four-way split is reflected in many recent checklists, papers, books, etc. But just when the finch problem appeared to be resolved, along come Sibley and his crew with another view.
We're back to a two-way split but with some notable differences. One group includes the Old World sparrows, weavers, and waxbills (each with allies) in individual subgroups along with, quite unexpectedly, wagtails, pipits, and accentors. The other family contains the Olive Warbler, all the northern "true" finches in a second subfamily, and the American sparrows in a third together with the warblers, tanagers, grosbeaks, and icterids, all now finches of a kind.
Within the subfamilies, the groups are so close that Sibley and Ahlquist can break them out only as "tribes." In short, all the finches fall into two related families that divide into eight subfamilies. This does not confirm the complex and multiple origins that many modern workers have suggested for the finches, but rather suggests the reverse -- the diffusion and radiation of one or two lineages into multiple lines of descent, producing several kinds of finches as well as new types not normally thought of as finches at all.
For more detail, we must leave Sibley and Ahlquist (1990) and turn to the sister volume, Distribution and Taxonomy of the Birds of the World (Sibley and Monroe 1990). There are a generous 10,000 species here -- the authors tend to be splitters rather than lumpers -- arranged, above genus level at least, in a sequence based on the DNA-DNA hybridization studies completed up to 1989-1990. The accounts include notes on habitat, geographic distribution, and cross-references to various Latin and English names that have appeared in the literature over the years, as well as comments on forms and relationships down to the generic, specific, and even subspecific level. A supplement to this volume, Sibley and Monroe (1993), updating and correcting the original material, and a useful world checklist (Monroe and Sibley 1993) have just recently been published.
If the family tree proposed by Sibley and Ahlquist (1990) is accepted, then Sibley and Monroe (1990) will be the new bible. Details may change as new information becomes available, but this view of the evolutionary history and living relationships of birds is already a stimulating challenge to the traditional view of avian life. Most of the reviewers were clearly impressed by the sheer scope of these volumes, but not all the critiques -- of either volume -- have unhesitatingly endorsed the results; see, for example, the thoughtful review by Frank Gill and Frederick Sheldon (1991) of Sibley and Ahlquist and the extensive write-up of both books by Alan Knox (1991). The negative views have centered on the data analysis, the application of the age-at-first-breeding correction, and the absence of other factors that might affect rates of evolution. The DNA-DNA hybridization technique is not operative at the species level (and only partially so at the generic level) and several of the reviewers have been at particular pains to resist the trend to use Sibley and Monroe as the basis for future taxonomic and comparative work (e.g., Lanyon 1992, Peterson 1992, Peterson and Stotz 1992, Siegel-Causey 1992). And, as is perhaps inevitable, many errors and inconsistencies are being found (e.g., DeBenedictis 1992, Parkes 1992, Siegel-Causey 1992; Sibley and Monroe have corrected at least some of these errors in their 1993 supplement. O'Hara (1991) and DeBenedictis (1992) provide two of the better summaries of the problems and shortcomings of these volumes.
How seriously are we to take these critiques? The new order is by no means universally accepted and even the AOU Checklist Committee, on which Burt Monroe serves as Chairman, is moving cautiously. But, considering how many apple carts are upset, the amount of controversy generated is probably not exceptional, and most of the critics (even some of the more negative) preface or conclude their remarks with lines like "tour de force" (Parkes 1992, on Sibley and Monroe (1990)), "one of the major ornithological contributions of this generation" (Blem 1991, on Sibley and Ahlquist (1990)) and "the most complete series of hypotheses to date for reconstructing the evolution of birds" (Storer 1992, on Sibley and Monroe (1990)). Even Peterson (1992), who is highly critical, says that "the phylogeny will certainly serve as a starting point for systematic investigations for many years to come."
Obviously, these works can hardly be ignored. Robert Raikow (1991), writing about Sibley and Ahlquist (1990) in The Auk, says:
The work summarized in this book has revolutionized systematic ornithology, and only time will tell how pervasive its effects will ultimately be. Sibley and Ahlquist have given us more to ponder and debate than anyone else in 20th-century avian systematics. They will keep us busy for a long time to come.
But perhaps readers of Birding will most appreciate Carey Krajewski's (1991) remarks: Their work will be cited by virtually every avian systematist for the foreseeable future.... I asked Jon Ahlquist how he felt about this incredible accomplishment. He shrugged, smiled, and said, "Not bad for a couple of bird-watchers." Not bad indeed.
----------
