For this study, I compared sequentially molted feathers for a few captive raptors from year to year and symmetrically matched feathers (left/right pairs) for many raptors to see if color patterns of sequential feather pairs were identical or if symmetrical pairs were mirror-image identical. Feather pairs were found to be identical only when without color pattern (e.g., the all-white rectrices of Bald Eagles [Haliaeetus leucocephalus]). Complex patterns were not closely matched, but some simple patterns were sometimes closely matched, although not identical. Previous claims that complex color patterns in feather pairs are fingerprint-identical (and therefore that molted feathers from wild raptors can be used to identify breeding adults from year to year with certainty) were found to be untrue: each feather is unique. Although it is unwise to be certain of bird of origin using normal feathers, abnormal feathers can often be so used.
Naturalists, falconers, and aviculturalists have long known that some individual birds can be identified by peculiarities in physical features. The most distinctive birds are those with gross abnormalities (e.g., Seton-Thompson  wrote about White Spot, the American Crow [Corvus brachyrhynchos]). Identification of falcons by persistent patterns in toe scales also seems secure (Smith et al. 1993). Much less likely, Dixon (1937) claimed the ability to identify wild female Golden Eagles (Aquila chrysaetos) by the peculiarities of color patterns on eggs. Concerning normal plumage, various authors have claimed that it was possible to identify adult Peregrine Falcons (Falco peregrinus) at the eyrie from year to year by comparing the amount of light and dark coloration on the head (Nelson 1988; Craig and Enderson 2004), and Rymon (1993) claimed to be able to identify Ospreys (Pandion haliaetus) at the nest with certainty by comparing head and body plumage from year to year.
That individual remiges and rectrices of adult raptors are grossly similar in pattern to their replacements in subsequent plumages is self-evident (except where injuries, advanced age, etc., impinge). Also, symmetrically matched feathers are normally much alike, as is consistent with the rules governing bilateral symmetry (see reviews of asymmetry in Neville 1976; Møller and Swaddle 1997).
In recent decades, some authors have further claimed that wild breeding accipiters could be individually identified through the years from molted feathers (Opdam and Müskens 1976). However, they interjected a caution that “small differences in feathers of subsequent moults may occur.” Ziesemer (1983) used year-to-year consistency in feather patterns to identify Goshawks (Accipiter gentilis) at the nest. Bortolotti and Honeyman (1985) identified some Bald Eagles (Haliaeetus leucocephalus) by spotting on otherwise immaculate feathers. Newton (1986) claimed that the details of feather patterns from year to year “gave a means of identification analagous to human fingerprints.” He also asserted that “any particular feather in one wing had the same pattern as its counterpart in the other wing.” More recently, Struwe-Juhl and Schmidt (2002) used feather length and rectrix color pattern to identify individual breeding White-tailed Eagles (H. albicilla) from year to year within a small local population. They cautioned, however, that closely related females had similar color patterns, thus limiting usefulness of the technique. It is noteworthy that, in the feather series illustrating all of the above studies, no sequential feathers were identical.
In those studies, researchers suggest that feather pattern identification could be useful for other species having complex color patterns. The implication is that it would be more difficult to use the technique on feathers with very simple patterns (e.g., it would serve poorly for the black and white rectrices of adult Pallas's Fish-Eagles [H. leucoryphus] and be totally useless for the all-white rectrices of adult Bald Eagles). Conversely, it should work best for species with the most complex patterns, such as the Golden Eagle.
In this study, I focus on melanin-based patterns (Roulin et al. 2008) easily visible to humans. For an expanded review of the influence of color on sexual selection see Hill and McGraw (2006a, 2006b). Color saturation of individual feathers not only varies between individuals but is also condition dependent (Wiebe and Bortolotti 2002), so caution must be employed in plumage comparisons.
Here, I examine seven questions. First, for normal feathers: (1) Are the color patterns produced by the corresponding follicles by all birds in a population unique? (2) Is the color pattern produced by one follicle for an individual adult bird identical from year to year? (3) Are symmetrical pairs of feathers on the wing and tail of an individual mirror image identical? (4) If color patterns are not identical but only similar, can we be certain that one bird produced any two feathers?
Second, for two types of abnormal feathers: (5) Are color patterns in abnormally pigmented feathers identical from year to year? (6) Are misshapen feathers identical from year to year? (7) If abnormal feathers are not identical from year to year, can we be certain that they are from the same bird?
For perfect surety in identification of individual birds by color patterns in normal feathers produced by the same follicle in succeeding generations, two conditions must be met: first, no two individuals in a population can have identical feather patterns. Second, the pattern in a feather produced one year must be identical to the pattern in succeeding years. Only if both conditions are met, can an observer state positively that two feathers were from the same bird.
I collected feathers from a female Golden Eagle held in captivity long term, beginning when it was ca. 24 yr old and concluding at its death around age 44 yr. Details of the eagle's handling are in Ellis and Kéry (2004). I also collected feathers from a second captive Golden Eagle over a 3-yr period, beginning when the bird was in juvenile plumage. Molted feathers from three captive adult Peregrine Falcons (two females, one male) were also used, but only for 4 yr each. Feathers from many other raptors (and a few non-raptors) were also inspected for this study (Table 1, 2).
Species examined for feather pattern congruency and showing conspicuous dissimilarity in color patterns for right and left matched pairs of feathers.
Species showing extreme similarity, but not congruence in color patterns for right and left matched pairs of feathers. For each species, color patterns were not complex.
For this study, I considered that identical feathers have bars, streaks, dots, and spots that match in size, shape, and location. In the ornithological literature, bars are broad, dark lines roughly perpendicular to the rachis (Fig. 1, left three remiges are barred). Streaks parallel the rachis as exemplified in the marks closest to the rachis on the inner web of the feather second from right. Dots (smaller) and spots are round, are usually darker than the background color, and can best be seen in the right two feathers in Fig. 1. Blotches are irregularly shaped large marks. Blotches and streaks intergrade on the inner webs of the right two feathers in Fig. 1. Variation in color saturation due to fraying, bleaching, condition-dependent growth, soiling, and fault bars were not considered.
Feathers from the same follicle in different years are hereafter referred to as “sequential pairs” (or “sequential series” if consisting of three or more feathers). Feathers from matched follicles on opposite sides of the body are here termed “symmetrical pairs” (or “symmetrical series” if three or more). “Feather pairs” refers to either or both. Consistent with standard practice, rectrices are numbered from the center (L1 and R1) outward. Primary remiges are numbered from the wrist outward (the most distal are L10 and R10); secondaries from the wrist inward (the most distal are L1 and R1).
Comparisons of symmetrical pairs of feathers using museum specimens were difficult (because the wings and tail are typically folded) except for outer primaries and outer and central rectrices. No attempt was made to quantify or codify color pattern differences. Rather, I present numerous photos to illustrate coloration trends.
Because of the need to photograph feathers separate from the bird, most of the photographs that accompany this text were, except for molted feathers from the two eagles and three falcons, taken from feather bag series from museum specimens or from feathers plucked from dead birds. Most photos were taken in the shade, above a black velvet background, with digital cameras, and in flash/portrait mode.
Results and Discussion
I addressed questions of identification by feather patterns most completely using Golden Eagles. In the darkest individuals, light areas in wing (Fig. 1: rightmost feather) and tail (Ellis and Lish 2006:6) are restricted. In pale individuals, light areas in some feathers are much more extensive than dark (Fig. 1: second from right). Some adults show evenly sized and evenly spaced barring (Fig. 1: left two feathers), whereas in others, barring is replaced by blotches and/or streaks (Fig. 1: right two feathers), or even variegation/vermiculation (Fig. 1: seen most clearly in the pale zone on the basal half of the inner web of second feather from left).
Differences in Color Patterns in Sequential Feathers
In Figure 2, feathers selected from sequential series of Golden Eagle primaries, representing two loci, show general similarity, but each individual feather pattern is clearly unique. Further, two loci for the tail also differ greatly (Fig. 3). Some feathers are sufficiently different from others in their series (e.g., the primaries on the far left in Fig. 2 and the rectrices on the far right in Fig. 3) that if found in the wild it would be easy to conclude that they came from different eagles.
In a species with regular barring, a Peregrine Falcon (Fig. 4), color patterns are grossly similar. However, on closer inspection, even coarse features such as the number and width of the dark bars on the outer web at rectrice R6 (Fig. 4a) vary from year to year. The offset in spacing of the bars on the inner and outer webs also changes through time. Fine details of the boundaries between light and dark bars are everywhere different from year to year.
In Figure 4b, there is a temporal trend toward interruption of the fifth bar (from the tip) on the inner web. In the 1982 R1 (a first-generation adult feather), the bar is continuous to the rachis; year by year there is a graded transition until 1985, when this bar stops short of the rachis. Lesser differences for other features are also readily visible.
Differences in Mirror-Image Color Patterns of Symmetrical Feathers
Figure 5 illustrates these differences for a Golden Eagle at two tail loci. In Figure 6, two wing loci are shown. In none of the comparisons are patterns of matched feathers even approximately congruent. In Figure 5a, it could easily be concluded that each of the left feathers came from a different bird than those from the right side of the tail.
Two pairs of eagle primaries in Figure 6 show great similarity in general pattern, but fine details at the borders between light and dark areas differ. Differences are great enough in Figure 6b that it would be easy to conclude that the feathers came from different birds.
Comparisons for non-eagles (Fig. 7, 8) illustrate the same general trends: symmetrical pairs are often very similar but not identical. Extremely complex patterns are less closely matched in symmetrical pairs (Fig. 7a, 7c, 7d). The owl feathers (Fig. 7b) have relatively simple patterns that are more alike than feather pairs with complex patterns, but even for the owl, the patterns are not mirror-image identical. The relatively simple patterns of juvenile Saker Falcon (F. cherrug) rectrices (Fig. 8) are clearly divergent when symmetrical pairs are compared. In five comparisons, a spot on one feather is matched with a bar on its mate.
Differences in Abnormally Pigmented or Deformed Feathers
Sequential series for three feather loci (Fig. 9) from the same captive adult female Golden Eagle show that aberrant feathers are also each unique. However, the abnormal secondaries (Fig. 9a) are so different from normal feathers that it would be possible to state with near-absolute certainty that each feather is from the same eagle even if the feathers were found in the wild.
Figure 9b demonstrates a caveat: less severe color abnormalities, such as the pale-to-white spot at the tip of rectrix L5, can transform, disappear, and even reappear from year to year. Such variability would provide unconvincing evidence for identification of the bird if the feathers were collected in the field.
The third abnormality, a physical flaw on the inner web of rectrix L6 (Fig. 9c), resulted in a weakness in all barbs on the distal portion of the inner vane severe enough to sometimes result in portions of the vane falling free. Finding a series of feathers with this or a similarly severe abnormality would provide convincing evidence of identity.
Overview of Color Pattern Comparisons
Preliminary observations (above) from a few species may be further refined by examining symmetric pairs from a broader array of species (Fig. 10, 11, 12). In addition, 24 species of Falconiformes, five owls, and eight non-raptors were evaluated (Table 1, 2). Although differences within feather pairs were evident for all, the patterns for the male American Kestrel (Falco sparverius; Fig. 10c) and for two small accipiters (Fig. 11c R1 and L1; Fig. 11d) are very alike. From the Gabar Goshawk (Melierax gabar) feathers (Fig. 11d), it is easy to see how Newton (1986) could conclude that symmetrically matched feathers were mirror-image identical. However, that species exhibits simple bars with diffuse boundaries between light and dark areas, whereas feathers from species with complex patterns and sharp boundaries are clearly unique (Fig. 10a).
Some additional generalizations are: first, broad zones with uniform pigmentation (e.g., Fig. 11c, 11d) on one feather are closely matched on its symmetric mate. Second, boundary lines where pale and dark zones meet are, if jagged, poorly matched (Fig. 10a, 11a). However, straight and smoothly curving boundaries are often closely approximated in paired feathers (Fig. 11c, L1 and R1), especially when the boundaries between light and dark zones are diffuse and/or have little contrast (Fig. 11d). Where boundaries are distinct, they are more visibly different in shape (Fig. 7a, 8, 10a).
Third, dark markings in extensive pale zones (Fig. 7c, 12a) are not closely matched (Table 3, Axiom 2c) even though the general appearance of the zones under comparison are typically very alike in hue and brightness. For all species inspected so far (raptor and non-raptor) which have very narrow, tortuous, transverse bars (Fig. 12b), such bars are widely disparate in matched pairs (Table 3, Axiom 2e).
Trends in color pattern congruency/similarity in sequential pairs and symmetrical pairs of feathers of raptorial birds.
Applying the principles of Ancestral-state Reconstruction (Cunningham et al. 1998, Cunningham 1999), the ancestral color pattern for the wings and tails of raptorial birds consisted of alternating bands of broad uniformly dark bars and broad uniformly light bars (e.g., Fig. 4, 7b, 8, 11c, 11d). It appears that the amount of difference between paired feathers is generally proportional to the degree to which the species' feather pattern has diverged from the juvenile or ancestral pattern. Good examples are seen in Figures 3b, 7c, 7d, 11a, 11b. Exceptions to this rule include species with extreme pattern reduction for purposes of social signaling (e.g., the Bald Eagle [solid white tail] and the adult Black-chested Buzzard-Eagle [Geranoaetus melanoleucus; solid black tail]).
Deviation from the ancestral form can take several directions. Light bars may be flecked with dark as in Figures 11b and 12c. Bars may be replaced by streaks as seen to a small degree in the right feather in Figure 12c (also Fig. 1, 3, 5). Bars may be narrowed and multiplied as in the subadult Black-chested Buzzard-Eagle (Fig. 11b). Bars can be widened and reduced in number, as in many species of raptors for which tail pattern signals age, status, or gender (Fig. 12a). Extreme deviation from the coarsely barred pattern can yield vermiculation as in some feathers of the Great Horned Owl (Bubo virginianus; Fig. 12d) and many caprimulgids, or anastomosing dark bars as in the Red-backed Hawk (Buteo polyosoma; Fig. 7d).
Positive Identification of Individual Birds
Although most individual raptors cannot be identified with certainty from color patterns in normal feathers, under some circumstances, individual birds can be identified even in the wild. The pale secondary remige in Figure 9a could be diagnostic, and an adult female Golden Eagle in Montana (see Ellis 1979:43) was readily identifiable long-term, even in flight and often at great distances, because she had two adjacent rectrices permanently missing (R4 and R5). When the tail was fanned, R6 flared far beyond the normal position. Other abnormalities in other birds may be just as diagnostic.
A second method that could be used to identify an individual with certainty from year to year requires that the molt be incomplete (as in the Golden Eagle) or that the bird be photographed periodically through the molt. If an eagle were photographed with enough detail in one year and then rephotographed the following year, many of the exact same feathers (albeit more faded and worn) would be present and therefore the bird would be identifiable with certainty in the subsequent year (or in some cases years). Note, however, that this conclusion depends on each feather being unique.
From the many comparisons, three primary generalizations emerge. First, complex patterns are not identical for either sequential feathers or symmetrical pairs. Second, simple patterns are more closely duplicated than complex patterns. Third, grossly abnormal feathers, while not identical from year to year, can be safely used to identify an individual. These generalizations, although true for the species so far studied and likely true for all species, must be considered tentative until each species of raptor is evaluated.
In closing, the hypothesis that individual raptors can be identified from year to year by normal feather patterns required that two conditions be met: (1) that feather pairs be identical and (2) that all individuals in a population differ. I found instead, for species with complex patterns and therefore great potential interindividual variation, that paired feathers were not identical (thus the first condition was not met). For species with simple patterns and therefore much less interindividual variation, there is greater similarity between feathers but also greater likelihood that another individual in the population has similar feathers (Fig. 13). Because of both conditions, certainty of identification is not possible.
I thank the curatorial staff at various museums for access to feathers. For pertinent permits, I thank the U.S. Fish and Wildlife Service, the Mongolian Ministry of Nature, and the former Soviet Ministry of Agriculture. Publication costs were paid by the National Aeronautics and Space Administration-Goddard Space Flight Center. Brent Sigafus cleaned and formatted the electronic images in the figures. Gary R. Bortolotti provided helpful comments on the early manuscript and supplemented the literature. Helpful reviews were also provided by Cheryl R. Dykstra and Edward H. Burtt, Jr.