Many studies of sexual selection have focused on the role of ornaments in mate choice (see Andersson 1994, Espmark et al. 2000). Birds have been a favorite taxon for those studies because of the prevalence of conspicuous ornaments, including brightly colored feathers, skin, and bills, and dramatically elongated feathers or feathers of elaborate structure. The colors of avian ornaments are the result of carotenoids, melanins, tissue structure, or some combination thereof. Carotenoid ornaments (e.g. bright reds, oranges, and yellows) have generated much interest. Those pigments can not be synthesized de novo by animals, but rather must be included in the diet (Brush 1990). That allows for the possibility that they are limited in nature or difficult to obtain (Hill 1994, 1996; Hudon 1994; Linville and Breitwisch 1997). Additionally, carotenoids have been implicated in a variety of critical physiological functions in animals. That recognition has stimulated much recent research on mate choice of carotenoid advertisement and the indicator value of carotenoid ornaments for several important aspects of individual condition.

Avian melanin ornaments have also received considerable research attention. In contrast to studies of carotenoid ornaments, however, the primary focus of studies of melanin ornaments has been on the advertisement of melanin “badges” in intrasexual competition. Thus, the associations among expressions of melanin ornaments, mate choice, and individual condition have not been made to the same degree as for carotenoids. There are several possible reasons for that. First, melanins not only color ornaments but are also the basis of most inconspicuous appearances (i.e. camouflage), in contrast to the highly conspicuous carotenoids. Second, melanins are synthesized by animals and thus do not need to be included in the diet. Finally, it is perhaps incorrectly assumed that melanin ornaments are typically black or blackish-brown and thus show both less variation within species and less diversity across species than do carotenoid ornaments.

Here, we review a variety of aspects of the biology of avian melanin ornaments. We demonstrate that the color diversity of melanin ornaments is quite broad, including red (e.g. Red Junglefowl and Barn Swallow [Hirundo rustica]), orange (e.g. Red Junglefowl), yellow (e.g. Western Tanager [Piranga ludoviciana]), and green (e.g. Mallard), in addition to black and brown. Color differences in all melanin ornaments are, in part, a function of the ratio of the two types of melanins contained, and we outline the intriguingly different metabolic pathways to the formation of the two different categories of melanins. There are likely to be both significant physiological benefits and costs to melanin production, and those may well differ for the two types of melanins. Finally, we address new avenues of investigation concerning melanin ornaments in the context of sexual selection.

Diversity of Melanin Ornaments

All melanin-colored ornaments contain a mixture of the two types of melanin, eumelanin and pheomelanin. The reflectance spectrum of the ornament is, in part, a function of the ratio of the two types of melanin in feathers (Haase et al. 1992, 1995; Prota et al. 1995; Shiojiri et al. 1999). Additionally, the absolute amount of eumelanin is important, because its presence may overwhelm the contribution of pheomelanin to the reflectance spectrum. The two types of melanin have different spectral qualities (Sarna and Swartz 1998). Eumelanin is perceived by humans (and presumably by birds) as black to dark brown, whereas pheomelanin is perceived as light brown, buff, dull red, orange, and yellow (Prota et al. 1995). Some pheomelanin pigments have the potential to produce brighter reds, oranges, and yellows than may typically be thought to be due to melanins (e.g. the bright orange feathers of Red Junglefowl; J. Hudon pers. comm.). A field guide to the birds of any geographic region shows quite clearly that ornaments of such colors are widespread. Despite that diversity, investigators have focused predominantly on black to brown eumelanic ornaments (Table 1Table 1.Table 1.).

The question of the diversity of melanin ornaments is complicated by the findings that both melanins and carotenoids can occur in the same ornament (Hudon 1991, Saino et al. 1999; K. McGraw pers. comm.). Melanins occur in ornaments that were once considered to be pigmented solely by carotenoids, for example, the yellow feathers of Western Tanagers (Hudon 1991) (see also Stradi 1998). In addition, carotenoids have been found in what are commonly classified as melanin ornaments, for example, the rusty-orange feathers of Barn Swallows (Saino et al. 1999; see also Howell 1952, Nero 1954). Reflectance spectra (and color perception) of such mixed ornaments may well depend on both the carotenoid and melanin pigments contained. There are, however, possible technological problems in verifying that hypothesis, insofar as the carotenoid pigment contribution to the spectrum in some mixed ornaments may overwhelm that due to melanins (J. Hudon pers. comm.), or melanins may overwhelm carotenoids (Theron et al. 2001). The alternative would thus be chemical extraction of pigments from an ornament (see below) and comparison of pigment amounts to reflectance spectra for a sample of individuals. To our knowledge, no one has yet done that. Moreover, no one has evaluated mate choice as a function of pigment composition in mixed ornaments assessed in the process of choice.

Melanin Production

The examinations of potential costs, controls, and material requirements of melanin production have focused on mammals, with less research on avian melanogenesis. We therefore present findings from studies on both mammals and birds in the following discussion. Melanogenesis involves the conversion of L-tyrosine into the two types of melanin. The biochemical pathway along which L-tyrosine is converted to melanin is complex and may have significant energy costs associated with it (V. Hearing pers. comm., Fig. 1). Both end products, eu- and pheomelanin, are large, negatively charged, insoluble, nitrogenous heteropolymers.

Melanogenesis takes place in melanocytes, which are derived from neural crest cells (Duval et al. 2002, Han et al. 2002). Melanocytes responsible for skin, hair, and feather pigmentation are found in the basal layer of the epidermis and in hair or feather follicles (Duval et al. 2002). Melanin production occurs in melanosomes, a melanocyte-specific organelle. The two types of melanosomes are rod-shaped eumelanosomes and spherical pheomelanosomes (Wolff 2003). Melanosomes move from the melanocytes into keratinocytes and become incorporated into growing hair or feathers, thus providing pigmentation (Duval et al. 2002). Pigments in skin, fur, or feathers represent a mixture of the two types of melanins that originate from multiple melanocytes.

There are at least three enzymes involved in melanogenesis that require metal cofactors. The copper-containing enzyme tyrosinase catalyzes the rate-limiting step in melanogenesis (Wolff 2003). The two enzymes Tyrp-1 and Tyrp-2 are required for eumelanogensis. The latter enzyme requires zinc, and the former is thought to require iron (Furumura et al. 1998). There are adverse health effects on birds when the diet has either too little or too much copper. Copper deficiency affects production of hemoglobin, with anemia as the result, whereas iron deficiency affects both production and maintenance of blood cells and the constituents of eggs (Arnall and Keymer 1975). The functions of zinc are not known with certainty. Whether deficiencies in those metal cofactors affect melanogenesis is also unknown.

Melanogenesis begins with the oxidation of L-tyrosine and DOPA to dopaquinone by tyrosinase, the rate-limiting step in melanogenesis (Wolff 2003; Fig. 1). In addition to production of dopaquinone, tyrosinase also oxidizes two other eumelanin intermediates. Eumelanogenesis is stimulated by the binding of α-melanocyte stimulating hormone (α-MSH) to the G protein-coupled receptor MC1R (also called CMC1 in birds) in the melanocyte membrane (Takeuchi et al. 1996a, b; Bowers et al. 1997; Theron et al. 2001; Voisey and Van Daal 2002; Wolff 2003). When α-MSH binds to MC1R, levels of tryosinase, Tyrp1, and Tyrp2 increase, and levels of intracellular cysteine decrease. That favors production of eumelanin. Pheomelanogenesis occurs without α-MSH stimulation. During pheomelanogenesis, following the production of dopaquinone, cysteine combines with dopaquinone to form cysteinyldopas that are oxidized via benzothiazines to pheomelanin (Wolff 2003). During pheomelanogenesis, the enzymes important to eumelanogenesis do not increase in concentration, and intracellular cysteine levels remain high. That favors the formation of pheomelanin. Additionally, pheomelanogenesis is increased when agouti signaling protein (ASP) interacts with MC1R (Hunt and Thody 1995, Wolff 2003).

Pheomelanogenesis has been suggested as the default pathway during pigment production (Wolff 2003). Presence of a default pathway is important. If there are significant differences in production costs of the two melanins, at least one production cost is nontrivial, and a changing ratio of melanin types affects the reflectance spectrum, then the expression of a melanin ornament can serve as an honest indicator of condition. It is thus important that investigators begin to examine production costs more closely, measure pheomelanin and eumelanin content in melanin ornaments, and test whether different ratios of melanins are perceptually distinct to avian vision and affect the behavior of the perceiver.

The pathway of melanogenesis is complex, requiring several different enzymes and ions, appropriate amino acid precursors, a favorable pH environment, and the influence of kerotinocytes and hormones. The products of >80 gene loci are involved in melanogenesis (Haase and Schmedemann 1992; Aroca et al. 1993; Haase et al. 1995; Hearing 1999, 2000; Ito et al. 2000; Takeuchi et al. 2000; Duval et al. 2002; Hoogduijn et al. 2003). In addition to complicated controls and requirements, melanogenesis is known to produce cytotoxic byproducts that are lethal to melanocytes (Tomita et al. 1984, Thody et al. 1991, Bowers et al. 1994; R. Bowers pers. comm.). Avian melanocyte cultures can be rescued by addition of antioxidants, suggesting that the cytotoxic byproducts are predominantly oxygen free radicals (Bowers et al. 1994; R. Bowers pers. comm.). It is not known, however, whether cytotoxic chemical production in melanogenesis affects either ornament expression or the body at large. In any case, complicated pigment production may provide information concern the health, genetic quality, or condition of an individual, as has been argued for other complicated types of pigments (Arnold et al. 2002).

Measuring Melanin Ornament Expression and Variation

The majority (58%) of studies on melanin ornaments cited in Table 1Table 1.Table 1. measured only the size (or area) of discrete patches of melanic feathers (44 of 76 studies). Thirty-three of the 76 studies (43%) in some way measured color of melanin ornaments (including four studies that measured both size and color). Sixteen of the 33 studies measuring color recorded only gross color changes (e.g. brown vs. black), six used color chips (either Munsell or Methuen), and seven used color reflectance spectrophotometry (see below). Three studies measured melanin content of feathers.

In some species, there is obvious variation in color of melanin ornaments (e.g. a gradation from dark brown to black). In such species, only gross color differences have been measured (e.g. Slagsvold and Lifjeld 1988, 1992; Sætre et al. 1995, 1997 for Pied Flycatchers). Prior to the recent introduction of portable reflectance spectrophotometers, there were few other options for measuring color differences. The newly employed spectrophotometer technology can reveal variation unapparent to human vision (Johnsen and Zuk 1996, Hill 1998, Hill and Brawner 1998, McGraw and Hill 2000, Smiseth et al. 2001, Jawor 2002). Whether that variation is perceived visually by birds must be empirically established, but obviously it is unwise to simply assume that it is not. That line of argument is even more relevant for melanin ornaments other than black, where even human vision can perceive variation. Reflectance spectrophotometry also yields measures of hue, saturation, and brightness of the color being measured (Hill 1998). Unfortunately, the correspondence between reflectance spectra (or measures of hue, saturation, and brightness) and melanin concentration or melanin ratios is unknown.

In a study of Northern Cardinals, we found variation in hue, saturation, and brightness of the black face mask of males (Jawor 2002). We also measured hue, saturation, and brightness of melanin ornaments of several other species (four eumelanic ornaments and one pheomelanic ornament) and found both intra- and interspecific variability (Table 2). Those data demonstrate that, in some species, there is substantial variation in melanin color variables, whereas in others, there is essentially none (Table 2, Figs. 2 and 3).

A final measure of ornament expression involves extraction, identification, and quantification of the pigment contained in the ornament. That has been done for carotenoid pigments in feathers (Hudon 1991, Stradi 1998, McGraw et al. 2001). Both eu- and pheomelanin are difficult pigments to isolate and measure. The most common method is the analysis of the byproducts of eumelanin oxidation and pheomelanin acid hydrolysis (Haase et al. 1995, Ito and Wakamatsu 1998, Ito et al. 2000). That technique gives an estimate of the two types of melanin within feathers. Still to be determined are the relationships among melanin content and concentration in ornamental plumage, reflectance spectra, and color perception by birds.

Melanin Ornaments as Honest Indicators

Environmental variables and physiological condition can influence melanogenesis and thus may also affect melanin ornament expression. Relevant influences on biosynthesis and deposition of melanin into ornaments include food availability and changing hormonal condition. Food availability is known to affect both gross color changes in and area of melanin plumage ornaments in some species (Slagvold and Lifjeld 1992, Veiga and Puerta 1996, Griffith et al. 1999a), but not in others (Hill 2000, McGraw et al. 2002). An unanswered question is whether a trade-off occurs at low food levels between melanin ornament expression and production of other proteins dependent on tyrosine. A similar trade-off has been suggested between expression of carotenoid ornaments and the need for those pigments in other cellular processes (Lozano 1994, von Schantz et al 1999).

Both types of melanins are initially produced from L-tyrosine, and cysteine is necessary for the production of pheomelanin (Ito et al. 2000). In birds, tyrosine is considered an essential amino acid (Murphy 1996, Hebert et al. 2002), whereas cysteine is synthesized from methionine, another essential amino acid (Griminger and Scanes 1986). Historically, amino acids have not been considered limiting, but that assumption needs to be reassessed. Indeed, some seed diets fed to cage birds are known to be protein deficient and specifically deficient in tyrosine, cysteine, or methionine (Arnall and Keymer 1975). Changes in melanin coloration in molting captive birds have been observed when diets are insufficient in protein (e.g. Dark-eyed Juncos; E. Ketterson pers. comm.). It is not known whether levels of amino acids—and tyrosine in particular—become limiting during molt and ornament production. Given that amino acid availability may differ among individuals, that potential constraint on melanin ornament production needs to be addressed more fully.

Food availability may influence melanin ornaments by affecting the hormonal controls of melanogenesis as well as affecting essential amino acid availability. In mammals, the type of melanin produced by melanocytes is controlled by an interaction between α-MSH and ASP (Lu et al. 1994, Hunt and Thody 1995, Hearing 1998, Graham et al. 1997). Eumelanogenesis is initiated by the stimulation of MC1R by α-MSH, whereas pheomelanogenesis occurs without α-MSH signaling, and is influenced by ASP (see above). Birds do not possess ASP but rather agouti related protein (AGRP), which has been suggested as functioning in a similar manner to ASP (Takeuchi et al. 2000). Agouti related protein is expressed in most tissues in birds and is known to interact with leptin (a hormone produced by adipocytes) and other melanocortin receptors (MC1R–MC5R or CMC1–CMC5) to regulate feeding behavior, metabolism, and neuroendocrine function both during periods of starvation and periods of satiation (Wilson et al. 1999, Takeuchi et al. 2000). Agouti related protein expression, by influencing pheomelanin production during molt, may affect the color of developing melanin ornaments, thus providing information concerning an individual's food intake during molt. For example, the color of the black face mask in male Northern Cardinals is correlated with body condition early in the breeding season (J. M. Jawor and R. Breitwisch unpubl. data), which, in turn, may reflect nutritional health during molt the previous autumn when the ornament is built. In contrast, in a controlled study of the effects of diet on melanin coloration in House Sparrows, McGraw et al. (2002) found that food deprivation did not influence brightness of melanin coloration. However, they did not measure ratios of the two melanin types in this study. The interaction between AGRP and α-MSH, their effect on ornament pigmentation, and avian ability to perceive differences in the ratio of the two melanins in ornaments clearly warrants further investigation.

The production of cytotoxic compounds during melanogenesis also indicates that melanin ornaments can serve as indicators of condition and the ability to withstand physiological perturbance. Production of oxygen free radicals in melanogenesis should be particularly interesting in species that display both melanin and carotenoid ornaments, insofar as carotenoids can act as oxygen free radical scavengers (von Schantz et al. 1999). We hypothesize that in such species, the two types of ornaments have a combined signaling function. Individuals can demonstrate their ability not only to withstand generation of cytotoxins but also their ability to ingest enough carotenoids to satisfy pigmentation, other physiological needs, and deactivation of the damaging byproducts from melanogenesis. That proposal may explain the apparent amplifier effect of melanin ornaments on carotenoid ornaments in, for example, guppies (Poecilia reticulata; Brooks 1996).

Melanin ornament expression may be influenced by titers of sex hormones during molt, during the breeding season, or both. Melanocytes are known to express cell surface receptors for both estrogen and progesterone, although the precise effects of those two hormones on melanogenesis are unknown (Hearing 1998). Although sex hormones are known to affect ornamental characteristics in birds, the type of effect varies with species (Haase and Schmedemann 1992, Haase et al. 1995). For example, in Mallards, higher levels of testosterone preceding molt influence the eclipse, nonbreeding plumage of males, with higher testosterone inducing a more complete eclipse plumage (Haase and Schmedemann 1992). Testosterone appears to affect melanogenesis so that feathers that are predominantly eumelanic in breeding plumage become mixed or predominantly pheomelanic in nonbreeding plumage (Haase et al. 1995). In House Sparrows, the area of the black bib is enlarged by testosterone during molt (Gonzalez et al. 1999, Evans et al. 2000, Gonzalez et al. 2001), but whether the ratio of the two types of melanins is affected is not known. Also in House Sparrows, testosterone enhances bill color, with higher levels of testosterone producing a blacker bill (Keck 1933, Haase 1975). There is an opposite effect of testosterone in European Starlings (Sturnus vulgaris), with higher levels of testosterone inhibiting melanin deposition in the bill and favoring carotenoid deposition (Witschi and Miller 1938). Those studies indicate that melanin ornaments are influenced by, and indicative of, levels of hormones that are important for breeding behavior.

It has been suggested that melanin coloration is generally unaffected by parasites. Hill and Brawner (1998) and McGraw and Hill (2000) found no effect of an intestinal coccidian parasite, Isospora spp., on melanin coloration in House Finches and American Goldfinches, respectively. Additionally, Hill and Brawner (1998) found no effect of Mycoplasma gallicepticum (the cause of mycoplasmal conjunctivitis) on melanin coloration. In those same experiments, however, coccidial and mycoplasma infections negatively influenced carotenoid ornament expression. We present two caveats concerning those studies. First, the melanic feathers measured in Hill and Brawner (1998) were tail retrix barring, which the authors acknowledge may not have been sexually selected as an ornament and thus unlikely to be so affected by condition. In McGraw and Hill (2000), on the other hand, the melanic feathers were the black forehead patch, and very likely an ornament subject to sexual selection. Second, melanins are derived from amino acids, and there is no reason to expect that coccidial parasites, which affect carotenoid absorption across the intestinal membrane, similarly affect the absorption of those melanin precursors. Active absorption of amino acids is mediated by transport proteins in healthy intestinal epithelium, whereas carotenoids are endocytosed by epithelial cells. When the epithelium is injured by coccidian activity, amino acids may still be able to enter or pass between epithelial cells, while carotenoids can no longer be absorbed (K. Crane pers. comm.). Although melanin ornaments may not be affected by endoparasites, they may be affected by ectoparasites. Fitze and Richner (2002) found that ectoparasites (hen fleas [Ceratophyllus gallinae]) affected the area of melanin ornamentation in Great Tits, but the color of both melanin and carotenoid pigmentation was not affected.

It has been hypothesized that different types of ornaments, and different aspects of a single ornament, may convey information on different components of condition. Through selection, ornaments have thus become associated only with those qualities for which they are most informative (Møller and Pomiankowski 1993, Hill and Brawner 1998, McGraw and Hill 2000). Carotenoid ornaments may thus best convey information concerning endoparasitism and health of the immune system. In contrast, melanin ornaments may best display information concerning nutritional health, ectoparasitism, and hormonal titers related to aggression and intrasexual competitive abilities. We need a broader assessment of components of condition in relation to ornament expression to determine if that proposed division is generally the case. We hypothesize that birds displaying both carotenoid and melanin ornaments provide more reliable indication of their overall quality than birds displaying only one or the other type of ornament.

Sexual Selection on Melanin Ornaments

Gray (1996) and Badyaev and Hill (2000) have proposed that lack of sexual dimorphism in melanin ornaments in contrast to carotenoid ornament dimorphism is evidence against sexual selection of melanin ornaments. That proposition is based on two assumptions. The first is that sexual selection acts much more forcefully on males than on females (see Lande 1980), and the second is that the primary mechanism of sexual selection is unidirectional mate choice. However, there is reason to question both of those assumptions. First, Amundsen (2000) reviewed studies on ornaments in female birds and concluded that those frequently function both in mate choice and in intrasexual competition. Second, bidirectional mate choice has been demonstrated in some species, and that pattern should be common in monogamous, biparental birds (Burley 1981, Burley and Coopersmith 1987). If melanin ornaments function in bidirectional mate choice or female intrasexual competition, then the prediction of sexual dimorphism in those ornaments does not necessarily follow.

Ornaments function in both mechanisms of sexual selection (Berglund et al. 1996, Qvarnström and Frosgren 1998). Melanin ornaments, in particular, appear to function more commonly in intrasexual competition than in mate choice (Badyaev and Hill 2000). If intrasexual competition occurs in both sexes for resources, mates, or both, selection should favor ornaments that function in both sexes and thereby constrain evolution of sexual dimorphism. In that view, the apparent pattern of limited dimorphism in many melanin ornaments is not evidence against their evolution by sexual selection but of their current use by both sexes.

Findings from manipulative aviary studies of melanin ornaments have suggested that those types of ornaments are unreliable indicators of quality unless there is social control of “cheats” (Rohwer 1977, Møller 1987a). It seems likely that some proportion of social control is a result of experimental dissociation of behavior and signal. Birds manipulated to signal dominance or subordinance frequently do not act in accordance with their altered signal (Rohwer 1977). In contrast, signal integrity was maintained where either behavior was modified to match the new signal or birds were not allowed to interact physically with manipulated conspecifics (Rohwer and Rohwer 1978, Senar and Camerino 1998). In any case, social testing need not indicate signal dishonesty. Clearly, condition can vary seasonally and on much shorter time scales, even daily. We predict that social testing should become common if several conditions are met: ornaments are built during distinct periods, condition can change through time, potential benefits of social testing are high, and the cost of social testing is relatively low for the initiator.

Any social testing associated with melanin ornaments represents a cost of signaling by those ornaments. As signals of dominance rank, melanin ornaments stimulate aggressive interactions between individuals of similar dominance status (Rohwer 1977). As previously mentioned, individuals with experimentally altered melanin ornaments frequently experience aggression from unaltered conspecifics. House Sparrows that interact frequently with conspecifics during molt grow larger badges (McGraw et al. 2003), but whether badge color changes is not known. Whereas House Sparrows with large badges are dominant to those with smaller badges (Møller 1987b, 1988; Gonzalez et al. 2002), interactions between individuals with similarly sized badges can be prolonged and violent (J. M. Jawor pers. obs.). We suggest, then, that in addition to physiological costs of melanin production, melanin ornaments may be costly to individuals in terms of the time and energy spent (and risks of injuries incurred) defending dominance status from similarly ornamented and ranked individuals. Those ornaments thus would satisfy the requirement that honest signals be costly.

An intriguing finding in several species is that males with smaller melanin ornaments have higher reproductive success (Lemon et al. 1992, Qvarnström and Frosgren 1998, Griffith et al. 1999b). In some cases, less ornamented males are also preferred by females (Lemon et al. 1992, Qvarnström and Frosgren 1998, Griffith et al. 1999b). Those males are often better parents than more ornamented males, in terms of nestling feeding rates (Studd and Robertson 1985a, b, 1988; Griffith et al. 1999b; but see Voltura et al. 2002). Melanin ornaments are positively affected by testosterone in some species (see above), as is aggressive behavior (Rohwer and Rohwer 1978). In contrast, paternal behavior can be negatively affected by high testosterone titers (Ketterson et al. 1992, Ketterson and Nolan 1994). The preference for less ornamented—and presumably less dominant or aggressive males—may represent a stronger preference for high-quality parents. Owens and Hartley (1998) found that male parental care is reduced in species with greater sexual dimorphism in the size of melanin ornaments. That suggests that melanin ornament size, which is affected by testosterone, can act as a reliable predictor of parental care behavior. Size of melanin ornaments in males, therefore, should be relevant to female mate choice in species to which the differential allocation hypothesis (Burley 1986) applies. Importance of paternal care, and the trade-offs in reproductive effort that females are willing to incur as a consequence of mate choice, need to be considered in studies addressing melanin ornaments that signal dominance status.

A Holistic View of Ornaments and Sexual Selection

Our conclusion is that investigators should broaden their study of avian melanin ornaments. Techniques now available allow for increasingly rigorous measurement of melanin ornaments, and investigators should no longer simply assume that the apparently invariant color of eumelanin ornaments is so perceived by birds. Further, the known use of black eumelanic ornaments in intrasexual competition should encourage investigators to focus on the question of the types of information being communicated by each class of ornament. Any type of ornament may best display information on only limited aspects of condition, and we see no theoretical reason why melanin ornaments cannot be as informative as carotenoid ornaments regarding particular aspects of condition. Carotenoid ornaments may best indicate endoparasitism and general health, whereas melanin ornaments may best predict dominance, hormonal titer, foraging abilities, and ectoparasitism. Intrasexual competition undoubtedly precedes or constrains mate choice in many species of birds, and the use of ornaments in those two mechanisms of sexual selection should receive equal investigative effort. Our hope is that investigators of avian ornamentation will develop a broader view of both the components of sexual selection and the diversity of ornaments displayed by birds.