On Avian Subspecies: Buyer Beware? - Part 2
by Nate Swick
By Steve N.G. Howell
Shortcomings of the Subspecies Concept
Like any attempt to draw lines on natural processes, the subspecies concept is not perfect. Frequent criticisms of the subspecies concept have led some ornithologists to refine its definition to something that seems scientific, to move away from the concept of average differences between populations and toward strict application of some statistically significant criteria. But is this realistic? Let's look at three major problems with the subspecies concept, regardless of percentages, as outlined by E. O. Wilson and W. L. Brown (1953).
First, it is common for different, diagnosable characters to be reflected in different ways or in different geographic patterns. A simple example of this discordant variation might be that wing length decreases from north to south, tail length increases from northwest to southeast, and color tones vary from east to west. Which character is used to define a subspecies? And what happens if five or even more characters are evaluated and they each vary, but not concordantly? For example, how would you divide the five populations in Figure 1? Do you recognize no subspecies, or do you recognize numerous subspecies, each with its own matrix of features? This is not an academic question but a very real occurrence in nature. Wilson and Brown used studies of towhees (Dickinson 1952) and grackles (Huntington 1952) as examples of this problem.
Second, if a subspecies is described based upon a single character, it is quite possible that this character will show up in different, unrelated populations. For example, Wrentits in drier habitats of southern Oregon have pale upperparts that match birds in southern California, and these two populations have at times have been united in a single subspecies (AOU 1957); other Wrentits in Oregon resemble disjunct populations in coastal central California (Phillips 1986). While any or all of these birds could be united as a single subspecies based upon a single feature, it is difficult not to view this shared plumage coloration as simply a local reflection of convergent climatic variables.
When describing subspecies it is important to compare like with like. The plumage tones of Wrentits vary considerably with geographic variation, but also with time of year and associated plumage wear and fading. Compare these two birds from the same site in central coastal California, one in mid-summer, the other in early winter. Marin County, California, 26 June 2010 (A) and 24 November 2007 (B). © Steve N. G. Howell.
A similar example may be found in Europe, where two subspecies of European Golden Plover were traditionally recognized - a paler southern subspecies with less black on the belly in breeding plumage, and a darker northern subspecies with more black. When larger samples were analyzed, the breeding distributions of these two types showed a complex mosaic pattern related to both latitude and altitude such that no subspecies are now recognized (Cramp and Simmons 1983, Byrkjedal and Thompson 1998). One could argue that as long as 75% of these Wrentits or plovers are separable from other populations, then they might qualify as subspecies; but how much useful geographic information is conveyed by a subspecies that includes widely separated populations with a mosaic distribution? One could just as easily say: "birds in drier areas typically have paler upperparts."
Third, and perhaps the most difficult criticism to address, is: how does one define distinguishable? What level of differentiation does one need? There is no lower limit to defining a subspecies. For example, one observer may perceive subtle color tones that another does not, and thus be able to distinguish valid (and repeatable) differences among series of specimens - while another observer would condemn all specimens as the same. Thus, it is possible that valid subspecies might be discounted by modern ornithologists lacking the patience or perception of their predecessors. Science demands repeatability, yet no two pairs of human eyes are alike. Machines such as spectrometers, which measure color differences, offer surrogates for the integrative powers of human cognizance, but as noted by Kevin Winker (1997), the capabilities of this method still fall short of what the human eye and brain can perform quickly and accurately.
For example, the human eye might readily classify 20 specimens of species X into two groups of 10 birds that differ in appearance - one group of darker and redder birds with buffier streaking, the other of paler birds with whiter streaking, but to quantify this "obvious" difference is not so easy. Thus, subspecies descriptions often include phrases such as "streaking buff rather than whitish, and rump darker cinnamon than subspecies X" and so on. Conversely, convoluted multivariate statistics involving plumage and morphology may reveal or support differences to scientifically justify subspecies, as done by Engelmoer and Roselaar (1998) for a selection of shorebirds, including Red Knot. That sort of work is admirable, but hard going for most of us to read, and for better or worse few people have the patience to try and replicate it. For example, despite all that work, Peter Pyle apparently threw up his hands and considered Red Knot as monotypic (Pyle 2008).
In theory, over time any population of a species may develop its own unique characters, especially if somewhat separated from other populations. It's simply a matter of how differences are manifest, and whether humans can see them, or at least measure them. Many subspecies distinguishable on morphological grounds may not be distinguishable by some genetic techniques (e.g., Ball and Avise 1992, Zink et al. 2005), providing fuel to those who consider that subspecies should be done away with.
Beyond these three theoretical considerations, reliance on specimens from the breeding grounds to define a subspecies brings its own problems. In the breeding season, birds are often in variably worn plumage, when it is difficult, if not impossible, to make meaningful comparisons of plumage tones. For example, the sometimes subtle differences in color tones apparent in fresher plumage may not be apparent in late summer. But fewer people were out collecting specimens in fall than in spring, and fresh-plumaged specimens are often the exception, not the rule. Thus, if a species includes migratory populations, there might be only a short time window, if that, when fresh-plumaged local breeders are present in an area before either they leave, or there is an influx of migrants from another region. This could easily confuse the issue of regional plumage variation, and thus of which subspecies occurs where (see the example of Sage Sparrows, discussed later).
One of the biggest practical problems with many subspecies described from continental North America is that they represent, or may represent, points on a continuum of variation - that is, points along a cline. As a theoretical example of a cline, a species may be strikingly blackish in the far east of its range, and strikingly pale silvery in the far west; these two extremes can be separated with near 100% confidence. But if these extremes are bridged by all manner of intermediates, finding a point, or points, where one can draw any line is arbitrary at best (see Figure 2). But lines often have been drawn, and subspecies named, usually when only points on the cline were sampled and it was not realized that variation was continuous. In this case, the problem lies not with the subspecies itself but with unfortunate misapplication of the subspecies concept.
Figure 2. A is 100% separable from E, and C is 75% separable from both A and E, but B is not 75% separable from A or C, and D is not 75 % separable from C or E. Where do you draw subspecies lines? Or do you?
Subspecies: A B C D E
Character: Paler ----------------------Darker
Subspecies in the Modern Era
The AOU website cited earlier also noted that one mission of the AOU is to produce a revised checklist that describes geographic variation and subspecies of North American birds. This is a noble mission - or a pointless one - depending on your philosophical view. To complete such a task thoroughly would require many lifetimes. Moreover, the chances a uniform approach would be applied to evaluating all subspecies seem slimmer than discovering a live Dodo. Let's look at just a few modern studies of subspecies to appreciate some of the challenges - and to see some of the benefits of understanding and recognizing valid subspecies.
In a seemingly rigorous statistical analysis, Michael Patten and Philip Unitt (2002) attempted to carefully apply the 75% rule to subspecies of the Sage Sparrow. Of the five widely recognized subspecies, Patten and Unitt concluded that only three should be recognized. The two interior breeding subspecies, canescens and nevadensis, were considered not to satisfy the 75% rule (although 73.7% of male nevadensis proved larger than the largest canescens - exemplifying the arbitrariness of definitions); thus, canescens was synonymized with nevadensis.
However, two other ornithologists, who studied Sage Sparrows - Carla Cicero and Ned Johnson - revisited the question (2006) and concluded from the same data that the subspecies canescens and nevadensis were fully diagnosable, far in excess of the 75% rule! "Contamination" of Patten and Unitt's sample by migrants was identified as the problem: when presumed non-local migrants were included, no subspecies were diagnosable, but when only known breeding birds were used, the subspecies were readily diagnosable.
Sage Sparrow may be the only species of North American bird that has received such critical subspecies scrutiny with respect to the 75% rule. And if four competent and respected ornithologists come to different conclusions from the same data, then what hope is there for wholesale review of avian subspecies in North America? Such a task would have to be farmed out to numerous people, few of whom would be as qualified as those who studied the Sage Sparrow...
Sage Sparrows have been subject to much subspecies debate. This individual was deep in the heart of the range of the desert subspecies cinerea, which is not supposed to have dark back streaking. But then again, birds can't read and don't know what they are supposed to look like. Baja California Sur, Mexico, 25 October 2008. © Steve N. G. Howell.
For the Yellow-billed Cuckoo, two subspecies have been described - eastern and western, differing slightly in measurements and perhaps in plumage. In 1988, Richard C. Banks, a member of the AOU committee, examined nearly 700 specimens of Yellow-billed Cuckoo, performed some simple statistics, and concluded that no subspecies should be recognized. He arbitrarily set June-August as the breeding season (see Sage Sparrows, above), and arbitrarily separated his sample into 19 different geographical regions. He did not detect any plumage differences, although he did not conduct spectrographic analysis on birds in comparable plumage. In 1993, Kathleen Franzreb and Steve Laymon revisited the question of Yellow-billed Cuckoo subspecies using 750 specimens, including those used by Banks. Their discriminant function analysis of morphological characters correctly classified over 80% of eastern males and females, and 74.6% of western males and 85.8% of western females. They recommended provisionally that the two subspecies should be maintained. Some western populations of Yellow-billed Cuckoo are of high conservation concern, adding a pragmatic twist to the more academic question of one subspecies or two. The notion of two subspecies also has implications for a recent paper claiming that Yellow-billed Cuckoos breeding in the Southwest are the same birds that bred earlier in the East (Rohwer et al. 2012). Do these birds magically change subspecies as they migrate? What to do?
The Savannah Sparrow was the subject in 2001 of a taxonomic revision, this time by AOU committee member James Rising. He reduced 21 subspecies down to 6, and then in 2011 he split the 6 into 4 species that included a total of 7 subspecies. Rising considered the original subspecies' descriptions to be arbitrary and qualitative rather than quantitative, and consequently he synonymized some 15 subspecies. He made no attempt to apply the 75% rule, and his lumping was based on the results of multivariate statistical analyses of certain morphological features. He explicitly ignored plumage coloration, perhaps because he felt it simply represented local adaptation. However, failing to analyze plumage coloration, which was used in most if not all of the original subspecies descriptions, might be considered equally arbitrary. In many species, plumage coloration is a feature used to discriminate subspecies, so why not in Savannah Sparrow? Perhaps a careful analysis of Savannah Sparrows, using spectrographic analysis in combination with morphology, and involving sufficient samples from a sufficient number of localities, could validate some of the described subspecies. But who wants to do all that work?
When a geographically variable species is examined carefully, the value of subspecies may be appreciated, such as for determining migration patterns and nonbreeding distributions - as done in 1997 by Philip Unitt and Amadeo Rea with Brown Creepers in California. Their study revealed an undescribed subspecies of Brown Creeper breeding in coastal central California, and reviewed the features of previously described subspecies. When these subspecific characters were clarified, it could be seen that migrant Brown Creepers from both the Rockies and eastern North America reach California in fall and winter, and that western Brown Creepers become more widespread at those seasons. Unitt and Rea's paper is perhaps a classic example of what birders (and ornithologists) would like subspecies to be.
Not all modern papers about subspecies are as useful, however. Among the avian subspecies described most recently from North America is Anser albifrons sponsa, the White-fronted Goose breeding in western Alaska. This taxon was described in 2011 by Richard C. Banks. The entire subspecies description is as follows: "Similar to A. a. gambelli of interior and northern Alaska but averages smaller. Similar to A. e. elgasi of the Cook Inlet area of Alaska but smaller in all dimensions." Measurements of the type specimen alone are provided (although how bill depth and bill width were measured is not stated), no indication is given of how many specimens were examined (of any taxon), no ranges of measurements provided, no discussion of age or sex differences, and no attempt to show how sponsa might be separated from other subspecies, in the hand or in the field. If you saw or even shot a White-fronted Goose, could you distinguish it as sponsa based on this "description"?
Might one or more of these White-fronted Geese be of the recently described subspecies sponsa? Given how little information was provided in the subspecies "description" there is no way of knowing. Baja California, Mexico, 10 October 2011. © Steve N. G. Howell.
The preceding examples hint that the huge challenge of critically re-evaluating all subspecies in the 1957 AOU Checklist (and some described since) may be too great for the modern generation of ornithologists and taxonomists, whose interests largely lie in other directions. Not to mention the possibility that politics and egos might on occasion compromise scientific objectivity... But we won't mention that. However, while the concept of subspecies may be fading from modern ornithology, it has somehow struck a chord with modern birders. Witness, for example, the new subspecies maps in the 6th edition of the National Geographic Society (NGS) field guide (NGS 2011). Or the subspecies categories used in eBird. Yet any birder reading the foregoing discussion might question what if anything a subspecies means to them, and at the very least whether the subspecies names they use are valid.