Differential effects of endoparasitism on theexpression of caro tenoid- and melanin-basedornamental colorationKev in J. M cGraw*a nd G eo¡rey E. HillDepartment of Biological Sciences and Alabama Agricultural Experiment Station, 331 Funchess Hall, Auburn University, Auburn,AL 36849, USAThe striking diversity of sexual dimorphisms in nature begs the question: Why are there so many signaltypes? One possibility is that ornamental traits convey di¡e rent sets of information about the quality ofthe sender to the receiver. The colourful, pigmented feathers of male birds seem to meet the predictions ofthis hypothesis. Evidence suggests that carotenoid pigmentation re£ects the nutritional condition of malesduring moult, whereas in many instances melanin pigmentation is a reliable indicator of soci al status.However, as of yet there have been no experimental tests to determine how these two ornament typesrespond to the same form of environmental stress. In this study, we tested the e¡ect of endoparasiticinfection by intestinal coccidians (Isosporasp.) on the expression of both carotenoid- and melanin-basedornamental coloration in captive male American gold¢nches (Carduelis tristis). We found that thecarotenoid-based plumage and bill coloration of parasitized males was less saturated than that developedby unparasitized males, but that the brightness and size of melanin-based black caps did not di¡erbetween the groups. These ¢ndings provide the most robust empirical support to date for the notion thatcarotenoid and melanin ornaments reveal di¡erent information to conspeci¢cs.Keywords:American gold¢nch;Carduelis tristis; coccidia; honest advertisement;Isospora;plumage coloration1. INTRODUCTIONSince Darwin (1871) developed his theory of sexual selec-tion to explain the remarkable sex di¡erences in size,shape, colour and behaviour that exist among animals innature, biologists have been fascinated by the evolution ofsexually dimorphic traits. Yet the information content ofornate male characteristics has been subject to criticalempirical testing only in the last few years. Visual signals,and in particular the colourful feathers that are displayedby male b irds, have been the focus of many tests of thesignalling function of ornamental traits (e.g. Butcher &Rohwer 1989; Savalli 1995).In many of the ¢rst attempts to understand the originand maintenance of brightly coloured plumage, colorationwas characterized as a single trait (e.g. Hamilton & Zuk1982; Johnson 1991). However, the bright colours infeathers can b e produced by several di¡erent mechanismsthat may follow completely di¡erent evolutionary trajec-tories (Fox & Vevers 1960). The most well studied of theseforms of plumage coloration is pigment-based ornamenta-tion, and plumage pigmentation can result from thedeposition of either melanin pigments, which are black orbrown in coloration, or carotenoid pigments, which varyin hue from red to orange to yellow (Fox 1976; Brush 1978).Most of the early studies that evaluated the informationcontent of variation in avian plumage pigmentationfocused on melanin-based coloration in males. Rohwer(1975) proposed the idea that variation in colour or patchsize may function as a badge of social status (the `statussignalling’ hypothesis), and it has been demonstratedsubsequently in a number of species that melanin-basedornaments function as indicators of social rank (reviewedin Senar 1999). Interestingly, carotenoid-based plumagecoloration has been largely ignored i n tests of statussignalling. Results from a few correlational studies suggestthat carotenoid coloration may also be related todominance (Marler 1955; Searcy 1979; Shawcross &Slater 1984; Eckert & Weatherhead 19 87; Maynard Smith& Harper 1988; McGraw & Hill 2 000a), but the onlytwo experimental studies on carotenoid-based plumagebrightness and aggression suggest that these ornamentsare unreliable predictors of social status (Wolfenbarger1999; McGraw & Hill 2000b).Unlike those studies that have considered thebehavioural mechanisms underlying bright plumage,which have been dominated by species with melaninpigmentation, studies on how ornamental plumage i sproduced have focused on species that deposit carotenoidpigments. Carotenoids cannot be synthesizedde novobyvertebrates, so they must be ingested i n their diet beforebeing absorbed, transported, processed and deposited infeathers (Brush 1981; Goodwin 1984). Such acquisitionaland usage demands make carotenoid ornaments particu-larly sensitive to environmental factors (Hill 1995, 1996,1999; Olson & Owens 1998). Several studies have docu-mented the in£uence of ecological stressors (e.g. food orpigment limitation, parasitic infection) on the expressionof carotenoid-based coloration (Slagsvold & Lifjeld 1985;Hill 1992, 1993, 2000; Hill & Montgomerie 1994; Olson1996; Linville & Breitwisch 1997; Brawneret al. 2000),corroborating the assertion that carotenoid ornamentsProc. R. Soc. Lond. B (2000) 267, 1525^1531 1525 © 2000 The Royal SocietyReceived 31 March 2000. Accepted 28 April 2000doi 10.1098/rspb.2000.1174*Author and address for correspondence: Department of Neurobiologyand Behavior, Cornell University, Ithaca, NY 14853, USA([email protected]).re£ect the health and condition of males. In contrast,although a handful of recent studies suggest thatnutritional stress during moult may a¡ect the develop-ment of melanin-based coloration (Slagsvold & Lif jeld1992; Veiga & Puerta 1996; Otter & Ratcli¡e 1999), noexperimental studies have been conducted to consider theproximate bases of variation in melanin-b ased orna-mental plumage pigmentation.The pattern that emerges from previous studies ofcarotenoid- and melanin-based plumage coloration isclear. Melanin and carotenoid pigmentation seem tocommunicate di¡erent information to conspeci¢cs, withcarotenoid pigmentation representing the physiologicalcondition of males during moult and melanin colorationindicating social status. However, these c onclusions canbe drawn only from comparisons of information contentacross species. As of yet, no study has considered thesignalling functions of carotenoids and melanins in aspecies in which both types of ornaments are present. Asa result, the goal of our study was to examine how bothcarotenoid and melanin ornamentation may act as indi-cators of health and condition in the American gold¢nch(Carduelis tristis). Males of this passer ine species displaycarotenoid-based plumage (yellow) and bill