Melanin pigmentation negatively correlates with plumage. preening effort in barn owls ALEXANDRE ROULIN

Functional Ecology 2007 Melanin pigmentation negatively correlates with plumage Blackwell Publishing Ltd preening effort in barn owls ALEXANDRE ROULIN Department of Ecology & Evolution, University of Lausanne, Biophore, 1015 Lausanne, Switzerland Summary 1. According to the handicap principle of sexual selection, colourful ornaments honestly signal absolute quality only if they entail fitness costs. The degree of melanism often covaries positively with aspects of individual quality, and hence melanin-based coloration should be costly to produce or to maintain in a good shape. This is, however, unlikely because melanin-based coloration is often strongly heritable and in birds the rate of feather wear decreases with the amount of melanin pigments packed in feathers. 2. The hypothesis that melanin pigments reduce the cost of maintaining colourful ornaments in a good shape predicts a negative correlation between the degree of melanism and both the size of the uropygial gland that produces preening secretions and the intensity of preening behaviour. 3. Using a correlative approach, I evaluated these two predictions in the barn owl Tyto alba in which the body underside varies from immaculate to heavily marked with black spots, a eumelanin-based trait, and from white to reddish-brown, a phaeomelaninbased trait. I correlated plumage traits with preening behaviour in nestlings and with the size and mass of the uropygial gland in dead adults. I also weighed nonornamental wing and tail feathers to assess whether the quality of nonornamental feathers is positively correlated with the degree of melanism of an ornamental plumage trait. 4. The degree of phaeomelanism was neither associated with preening behaviour nor with the size and mass of the uropygial gland. In line with the two predictions, individuals with more and larger black spots had a lighter uropygial gland and preened less frequently. Because nonornamental wing and tail feathers of spottier individuals were heavier per unit of surface area, the entire plumage of eumelanic individuals may be more robust and in turn require less care than the plumage of nonmelanic conspecifics. 5. In conclusion, the degree of eumelanism can be associated with aspects of individual quality even if eumelanic ornaments are neither costly to produce nor to maintain in a good shape. Key-words: colour polymorphism, handicap principle, melanin, preening, Tyto alba, uropygial gland Functional Ecology (2007) doi: 10.1111/j.1365-2435.2006.01229.x Ecological Society Introduction The handicap principle states that an ornament can honestly signal the absolute quality of individuals only if the ornament entails significant fitness costs. This principle is based on the premise that only individuals in prime condition can simultaneously allocate resources into an ornament and other phenotypic traits such as parasite defence (Hamilton & Zuk 1982). This property ensures that sexually active individuals can assess the expression of an ornament to determine the absolute Author to whom correspondence should be addressed. E-mail: Alexandre.Roulin@unil.ch quality of potential mates (Zahavi 1975). Although carotenoid-based coloration is considered an honest signal of individual quality (Olson & Owens 1998; McGraw, Regan & Parker 2005), there is still no consensus about whether continuous variation in melaninbased coloration signals honestly individual quality. Indeed, in many animals the degree of melanism is strongly heritable (Buckley 1987; Hearing & Tsukamoto 1991; Majerus 1998; Majerus & Mundy 2003) and not, or weakly, sensitive to environmental factors (Hill & Brawner 1998; Roulin, Richner & Ducrest 1998; Gonzales et al. 1999; McGraw et al. 2002; Roulin & Dijkstra 2003; Poston et al. 2005; Siefferman & Hill 2005; but see Griffith, Owens & Burke 1999). If different variants of a melanin-based ornament entail similar production 264

265 Maintenance cost of melanism costs, they should not be equally costly to maintain in a good shape or to bear as the degree of melanism is often positively correlated with immunocompetence (Roulin et al. 2000; Galeotti & Sacchi 2003), resistance to parasites (Roulin et al. 2001a; Armitage & Siva- Jothy 2005), developmental homeostasis (Rohwer & Wingfield 1981; Roulin et al. 2003) and the level of aggressiveness (e.g. Järvi & Bakken 1984; Senar et al. 1993; West & Packer 2002). A previous study indeed suggested that the costs to maintain ornaments in good condition should reinforce the honesty of such signals (Walther & Clayton 2004). However, in contrast to what we would predict under the handicap principle, in birds the indentation hardness of melanic keratin is 39% greater than that of nonmelanic keratin (Bonser 1995). Under the same abrasive conditions melanic feathers wear more slowly than nonmelanic feathers (Burtt 1979) and break less frequently particularly when individuals cannot preen them correctly (Kose & Møller 1999). Therefore, I propose the hypothesis that birds displaying a melanin-based trait to a larger extent invest less effort to preen their feathers using their beak, a behaviour that protects feathers against abrasion and breakage (Jacob & Ziswiler 1982; Moyer, Rock & Clayton 2003; Zampiga, Hoi & Pilastro 2004), ectoparasites (Moyer et al. 2003; Clayton et al. 2005) and bacteria (Shawkey, Pillai & Hill 2003). From this hypothesis, I propose the following two predictions. First, melanic birds should preen less frequently their plumage than less melanic conspecifics because their plumage is more robust. Second, because uropygial gland secretions are preened into the plumage using the beak, the degree of melanism should be negatively correlated with the size and/or mass of the uropygial gland, a good proxy of the amount of preen oil as capsules where this oil is produced occupy 68% of the uropygial gland (Sandilands, Savory & Powell 2004). This gland produces mono-, di- and trimester waxes, sterols and hydrocarbons that are important to maintain the plumage supple, rigid and to protect it against abrasion (Jacob & Ziswiler 1982; Shawkey et al. 2003). This second prediction is based on the assumption that production of preen oil and that preening behaviour are costly (Oka & Okuyama 2000; Yorinks & Atkinson 2000). Using a correlative approach, I evaluated these two predictions in the barn owl Tyto alba by correlating the degree of plumage melanism with preening behaviour in nestlings and with the size and mass of the uropygial gland in dead adults. This species is particularly suited because the body underside varies from immaculate to heavily marked with black spots of varying size, a eumelanin-based trait, and from white to reddish-brown, a phaeomelanin-based trait. A negative relationship between the degree of melanism measured on the body underside and effort invested in preening may be explained not only by the fact that in melanic individuals ornamental feathers of the body underside require less care but also because nonornamental feathers (e.g. wing and tail feathers) of melanic individuals are of higher quality. To assess indirectly whether nonornamental feathers of spottier or darker reddish-brown individuals are of higher quality, I weighed one wing and one tail feather in another sample of dead owls with the assumption that heavier feathers are more robust (Aparicio, Bonal & Cordero 2003) and of higher quality (Dawson et al. 2000). To my knowledge, this is the first study that investigates whether individuals displaying different variants of a melanin-based trait invest differentially in preening behaviour. Materials and methods model organism The barn owl is nocturnal, medium-sized (range in body mass is 241 and 478 g) and preys mainly upon small mammals. Three-week-old nestlings are thermoindependent and hence no longer brooded by their mother, parents sleeping at some distance from their nest at daytime. First-year body feathers are grown at 35 60 days of age. Nestlings take their first flight at 55 days of age. In females, the size of black spots covaries positively with immunocompetence, parasite resistance, developmental homeostasis and calcium physiology (Roulin 2004a; Roulin et al. 2006). Wing and tail lengths are not genetically correlated with plumage spottiness (Roulin 2006). Males mate nonrandomly with respect to female plumage spottiness and they invest more effort in reproduction when paired with heavily spotted partners (Roulin 1999). This suggests that female plumage spottiness is sexually selected, and as a consequence I consider this plumage trait as ornamental. In males and in some years only, reddish-brown individuals feed their brood at a higher frequency than white ones, produce more offspring and have a heavier heart (Roulin et al. 2001b). Reddish-brown and white individuals have a different diet (Roulin 2004b). Replacement of wing feathers starts in the bird s second year (Taylor 1994). preening behaviour Between July and September 1997, I measured preening behaviour in a Swiss population breeding in nest boxes (1 0 0 6 0 5 m) in which direct sunlight does not get in. Because it is difficult to record nestling behaviour in crowded broods, I created two-chick broods by randomly choosing two 21 44-day-old (mean ± SD = 32 ± 7) siblings in 13 broods. Although clutches of two eggs are rare in my population (four of 672 complete clutches, 0 6%), broods containing two chicks at fledging are common (53 of 625 broods, 8 5%). On one day, one of the chicks was food-supplemented by keeping it close to the nest box in a large plastic pail without any lamp (diameter = 0 6 m; height = 0 8 m) between 09.00 and 21.00 h with three dead laboratory mice, whereas its nonfood-supplemented sibling was in another pail without any food. On another day, I reversed the

266 A. Roulin treatment, and on another day both individuals were nonfood-supplemented. This procedure ensures that food treatments were assigned randomly with respect to plumage traits. The three treatments were carried out on three successive days and their order was randomized. Between 09.00 and 21.00 h the food-supplemented individuals consumed on average 1 7 mice (34 g). At 21.30 h the two siblings were returned to their nest box, and their nest-mates, which had remained in the nest during daylight hours, were put in a similar can at some distance from the nest until 23.30 h. At night, I recorded the amount of time each nestling preened their plumage using their bill during 15 min before a parent brought the first prey item between 20.57 and 23.00 h (mean = 22.23). To normalize the data set, I square-root transformed the proportion of time nestlings spent preening during the 15 min. Siblings were ringed on different legs to recognize them on video footage. Because in some cases the infrared sensitive video camera was not properly installed the day before the experiment started, I could not record preening behaviour in some individuals. More specifically, I successfully monitored 17 food-supplemented individuals from 12 broods, 17 nonfood-supplemented individuals in the presence of a food-supplemented sibling (11 nests), and 22 nonfood-supplemented individuals in the presence of a nonfood-supplemented sibling (13 nests). The method to quantify preening behaviour was not stressful to the birds, as on video footage I did not notice that individuals were disturbed in any way. This statement is corroborated by the fact that nestlings behaved in a predictable manner following theories of parent offspring conflict (e.g. Roulin 2001) and sibling negotiation (e.g. Roulin, Kölliker & Richner 2000). Furthermore and as expected, food-supplemented nestlings vocalized less intensely in the absence of parents than its fooddeprived sibling (Kölliker & Richner 2000), but spent more time preening (Roulin 2002). This indicates that the experimental design used in the present study was powerful to detect a trade-off between nestling vocalization behaviour and preening behaviour. I am therefore confident that results obtained from this experiment are biologically relevant. When 55 days of age, I assessed nestling plumage coloration on the breast, belly, flank and underside of the wings by comparison with eight colour chips ranging from I for reddish-brown to VIII for white (Roulin 1999). The mean value was used in the statistical analyses. On the same four body parts, I placed a 60 40 mm frame within which I counted spots. Within the same frame I measured spot diameter to the nearest 0 1 mm, and calculated a mean value for each body part. The percentage of the plumage surface covered with black spots was obtained with the formula (100 π number of spots (mean spot diameter/2) 2 /60 40). Percentages found on the two flanks were averaged, as well as percentages found on the two wings. The mean value of the four body parts was denoted plumage spottiness and used in the statistical analyses. I used this index rather than number and size of black spots as in other studies (e.g. Roulin 2004a) because quality of the plumage should be positively correlated with the plumage surface covered with black eumelanic patches. Methods of assessing plumage traits are reliable (Roulin 1999, 2004a). Sex of the nestlings was determined with molecular markers (Roulin, Ducrest & Dijkstra 1999). To analyse the relationship between preening intensity (i.e. proportion of time spent preening) and both nestling plumage coloration and spottiness, I removed variation in these plumage traits explained by sex because females are on average darker coloured and spottier than males (Roulin & Dijkstra 2003). To this end, I extracted residuals from an anova with each plumage trait entered in turn as the dependent variable and sex as a factor. For each nest I calculated mean residual sibling values. Because nonfood-supplemented nestlings preened at a similar rate in the presence of a foodsupplemented as a nonfood-supplemented sibling (paired t-test: t 19 = 0 22, P = 0 83), I pooled the data. Mean residual nestling plumage spottiness and date when preening behaviour was recorded were not significantly correlated (Pearson correlation on mean sibling values: r = 0 48, n = 13, P = 0 10). uropygial gland size and mass Between January 2004 and 2005, 36 dead barn owls were collected along highways in the Champagne and Lorraine regions of France. Bodies were collected daily, and hence stayed no more than 1 day on the roadside before being frozen at 20 C. In January 2005, the bodies were thawed and I identified sex after gonad inspection. Individuals with a bursa of Fabricius were assigned to the age category yearling, while others were classed as adult (Glick 1983), a factor that might be important, as in birds the uropygial gland increases in size with age (Sandilands et al. 2004). Body mass (mean ± SD: 290 ± 21 g) was calculated as the mass measured on the day of collection minus the mass of the stomach contents (13 ± 12 g). I measured the length (12 3 ± 0 9 mm), width (7 1 ± 0 6 mm) and height (12 6 ± 1 2 mm) of the uropygial gland to the nearest 0 1 mm. To obtain an index of the size of this organ, I extracted the first component of a principal components analysis that explained 63 5% of total variation (eigenvalue = 1 91) with loading factors of length being 0 63, width 0 56 and height 0 55. This first component was denoted gland size. Eigenvalues for the second and third components were 0 68 and 0 41, respectively. I cut off the uropygial gland at its basis and weighed it to the nearest 0 01 g (mean ± SD: 0 347 ± 0 096 g). Plumage coloration and spottiness were measured in the same way as explained above for nestlings. Collection date was defined as the number of days between the 1 January 2005 and the date when dead bodies were collected along highways. I controlled for date in the statistical analyses because uropygial glands increase

267 Maintenance cost of melanism in size during reproduction and the post-breeding period (Battacharyya & Chowdhury 1995). Bill length was not significantly associated with plumage spottiness (r = 0 24, n = 34, P = 0 16; bill was damaged in two individuals), a trait that may play a role in preening behaviour (Clayton & Cotgreave 1994; Clayton et al. 2005). mass of nonornamental wing and tail feathers Between January 2000 and April 2001, 97 dead barn owls were collected in the same region as where uropygial glands were measured. Bodies were collected daily, and hence stayed no more than 1 day on the roadside. In April 2001, the bodies were thawed and I assessed plumage coloration and spottiness in the same way as explained above. Sex (46 males and 51 females) was identified after gonad inspection, and individuals assigned to the age class yearling (n = 74) if I found the organ bursa of Fabricius and adult otherwise (Glick 1983). First of June was an arbitrary reference of birth date allowing to estimate the age at which yearlings died. In the present study, I considered only individuals of known age (i.e. yearlings) because for adults I cannot determine the age of the wing and tail feathers, a factor that influences feather abrasion (e.g. Brommer et al. 2003) and in turn weight (Newton 1966; Wijnandts 1984). In 70 yearlings I pulled out the 10th primary feather of the left wing and one of the central tail feathers (in the four other individuals feathers were damaged). These two feathers were chosen arbitrarily. In April 2004, the 140 feathers were weighed to the nearest 0 001 g blind to plumage characteristics. An index of the surface of the feathers was determined by photocopying them and weighing to the nearest 0 001 g the surface of the paper covered with the feathers. This method is reliable as feather mass was positively correlated with the index of feather surface (Pearson correlation, wing: r = 0 38, n = 70, P = 0 001; tail: r = 0 22, n = 70, P = 0 066). A relationship between feather density (i.e. feather mass statistically corrected for the index of feather surface) and plumage coloration or spottiness may be due to the fact that plumage characteristics are correlated with mass of the rachis and/or mass of barbules (i.e. feather without the rachis). To test this possibility, the most proximal 22 mm of the calamus of wing feathers of 72 yearlings was weighed to the nearest 0 001 g. I considered 22 mm because this part was always free of feather barbules. The same procedure was applied to the rachis of tail feathers of 68 yearlings. Body mass (mean ± SD: 286 ± 23 g) was calculated as the mass measured on the day of collection minus the mass of the stomach contents (11 ± 11 g). statistical procedure All statistical analyses were performed with the package JMP IN 5 1 (SAS Institute Inc., Cary, NC, USA), are two-tailed and P-values smaller than 0 05 considered Fig. 1. Relationship between the proportion of time spent preening during the 15 min before the first nocturnal visit of a parent with a prey item (square-root transformed) and mean residual nestling plumage spottiness. Residuals were extracted from an anova with plumage spottiness as the dependent variable and sex as a factor. Each data point represents mean sibling values. Pearson correlation is r = 0 63, n = 12, P = 0 028. Regression line is drawn. significant. I used stepwise regression analyses using the BACKWARD option with the probability of leaving a factor being 0 10. In the text, I nevertheless report statistics for all variables even if some of them were rejected from the model. Results preening behaviour Spottier food-supplemented nestlings spent less time preening during the 15 min preceding the first nocturnal arrival of a parent with a prey item (four-way stepwise regression, residual plumage spottiness: F 1,10 = 6 55, P = 0 028, Fig. 1; residual plumage coloration: F 1,10 = 0 17, P = 0 69; nestling age: F 1,10 = 0 06, P = 0 81; hour: F 1,10 = 0 55, P = 0 48). In nonfood-supplemented nestlings, preening behaviour was not significantly associated with residual plumage spottiness (similar stepwise regression, F 1,11 = 0 09, P = 0 77) and coloration (F 1,11 = 2 14, P = 0 17) after controlling for nestling age (F 1,11 = 2 00, P = 0 19) and hour (F 1,11 = 5 17, P = 0 04, preening intensity increased along the night). Because plumage spottiness is strongly heritable (Roulin & Dijkstra 2003), a relationship between this plumage trait and preening behaviour implies that siblings should preen at a same rate. This was the case when siblings were foodsupplemented (Spearman correlation: r s = 0 90, n = 7, P = 0 006; Fig. 2) but not when they were nonfoodsupplemented (Pearson correlation: r = 0 38, n = 13, P = 0 19). The above results are not inflated by ectoparasitism (Clayton 1991; Yorinks & Atkinson 2000) and body mass. Indeed, number of flies Carnus hemapterus on

268 A. Roulin Fig. 2. Proportion of time spent preening during the 15 min before the first nocturnal visit of a parent with a prey item (square-root transformed) in food-supplemented siblings. Regression line is drawn. each nestling, the most common ectoparasite in Swiss nestling barn owls, was not correlated with nestling plumage spottiness (Pearson correlation, r = 0 33, n = 12 nests, P = 0 29; parasites were not counted in one nest). Nestling body condition given by the residuals of body mass on age from a linear regression analysis was not associated with nestling plumage spottiness (r = 0 08, n = 13 nests, P = 0 79). Fig. 3. Relationship between residual uropygial gland mass and plumage spottiness. Residual uropygial gland mass was extracted from a multiple regression analysis with uropygial gland mass as the dependent variable and as two independent variables collection date and uropygial gland size. Filled circles represent males and open circles females. Pearson correlation is r = 0 42, n = 36, P = 0 011. Regression line is drawn. uropygial gland mass and size In a stepwise six-way ancova, the uropygial gland was heavier in the least spotted individuals (F 1,32 = 7 16, P = 0 012; Fig. 3) after controlling for collection date (F 1,32 = 15 80, P = 0 0004; Fig. 4) and index of gland size (F 1,32 = 70 56, P < 0 0001). In the same model, body mass (F 1,32 = 0 84, P = 0 37), sex (F 1,32 = 1 49, P = 0 23), age (F 1,32 = 2 51, P = 0 12) and interaction between plumage spottiness and sex (F 1,32 = 0 78, P = 0 47) were not significant. A similar stepwise five-way ancova with the index of gland size as the dependent variable showed no significant relationship with plumage spottiness (F 1,34 = 0 26, P = 0 62) indicating that the uropygial gland of the least spotted individuals is heavier but not larger. Similar analyses revealed that plumage coloration was neither correlated with gland mass nor with gland size (P > 0 13). mass of nonornamental wing and tail feathers Fig. 4. Relationship between uropygial gland mass (g) and collection date. Filled circles represent males and open circles females. Pearson correlation is r = 0 57, n = 36, P = 0 0003. Regression line is drawn. Mass of the 10th primary wing feather was positively associated with plumage spottiness (stepwise five-way ancova, F 1,66 = 9 60, P = 0 003, Fig. 5) after controlling for the index of feather size (F 1,66 = 7 51, P = 0 008). In the same model, sex, body mass, coloration, estimated age and interaction between sex and plumage spottiness were not significant (P > 0 35). I obtained a similar relationship between mass of the central tail feather and plumage spottiness (F 1,65 = 9 55, P = 0 003, Fig. 6) after controlling for the index of feather size (F 1,65 = 3 56, P = 0 06) and body mass (F 1,65 = 8 14, P = 0 006; heavier birds had heavier feather tails). In this model, sex, plumage coloration, index of feather size, estimated age and interaction between sex and plumage spottiness were not significant (P > 0 33). Mass of the most proximal 22 mm of calamus of the 10th primary wing feather was not associated with plumage spottiness (two-way ancova, F 1,71 = 0 20, P = 0 65; sex: F 1,71 = 0 28, P = 0 60). In contrast, the most proximal 22 mm of the calamus of the central tail feathers were heavier in spottier individuals

269 Maintenance cost of melanism Fig. 5. Relationship between residual mass of the 10th primary feather of the left wing and plumage spottiness. Residuals were extracted from a linear regression of feather mass on the index of feather surface. Filled circles represent males and open circles females. Pearson correlation is r = 0 35, n = 69, P = 0 004. Regression line is drawn. Fig. 6. Relationship between residual mass of one central tail feather and plumage spottiness. Residuals were extracted from a linear regression of feather mass on the index of feather surface. Filled circles represent males and open circles females. Pearson correlation is r = 0 42, n = 69, P = 0 0004. Regression line is drawn. (two-way ancova, F 1,64 = 6 24, P = 0 015, Fig. 7; sex: F 1,64 = 5 23, P = 0 026, rachis were heavier in females than males). Discussion A comparative analysis has suggested that ornamental plumage traits are more costly to maintain in a good shape than nonornamental feather traits (Walther & Clayton 2004) thus reinforcing the honesty of such sexually selected signals. In their study, the authors did not discuss the potential role of alternative pigments Fig. 7. Relationship between mass of the most proximal 22 mm of the calamus of one central tail feather and plumage spottiness. Filled circles represent males and open circles females. Pearson correlation is r = 0 38, n = 67, P = 0 0014. Regression line is drawn. that are packed in feathers to explain their results. To the best of my knowledge, the present study is the first one to examine whether melanin-based coloration predicts investment in maintenance of the external body surface. According to the handicap principle of sexual selection blacker individuals should allocate more resources to preen their plumage given that the level of blackness often reflects aspects of quality including aggressiveness (e.g. Mennill et al. 2003) and immunocompetence (Figuerola, Munoz & Gutierrez 1999; Roulin et al. 2000; Galeotti & Sacchi 2003) and this trait plays a role in mate choice (Jawor & Breitwisch 2003). However, given that in birds melanin pigments protect feathers against abrasion (Bonser 1995) and ectoparasites (Kose & Møller 1998), I hypothesized the opposite pattern, namely that blacker individuals invest less effort to maintain their plumage supple and rigid. The present correlative study carried out in the barn owl is consistent with this hypothesis. Preening behaviour plays an important role in keeping the plumage supple, rigid, resistant to abrasion (Moyer et al. 2003; Zampiga et al. 2004) and perhaps to avoid the fading of bright colours (Piersma, Dekker & Sinninghe Damsté 1999). When nestling owls were food-satiated, they preened less often if the plumage of the body underside was blacker at 55 days of age. This pattern was not detected when owls were nonfoodsupplemented perhaps because they invest more effort in sibling competition and wait until they have been fed to take care of their plumage (Roulin 2002). I quantified preening behaviour in nestlings before or during the time when feathers of the body underside were growing, and nestling age was not associated with preening intensity. Therefore, the negative relationship between plumage spottiness and preening behaviour cannot be explained by the degree of eumelanism of the ornamental feathers but probably because nonornamental

270 A. Roulin feathers of spottier individuals are of higher quality. Unfortunately, I did not measure feather quality in birds used to measure preening behaviour. However, in another sample of birds, I found that wing and tail feathers of spottier individuals were significantly heavier. Assuming that feather mass is an appropriate measure of feather quality (Dawson et al. 2000), most feathers of deeply melanic birds are more rigid and require less care than feathers of less melanic conspecifics. The relationship between preening behaviour and the degree of melanism may be caused by an uncontrolled variable such as susceptibility to parasites or body condition. This is, however, unlikely to be the case because in the sample of birds used in the present study there was no relationship between ectoparasitism, body mass and plumage spottiness. Being aware of the correlative nature of the data and that I measured preening behaviour in nestlings and not in adults (which is logistically difficult in natural conditions), I tried to obtain other pieces of evidence for the hypothesis that the extent to which birds display a melanin-based ornament predicts investment in preening behaviour. To this end, I measured and weighed the uropygial gland of dead birds, the organ that is responsible for the production of preening secretions (Jacob & Ziswiler 1982). As expected, blacker individuals had a lighter gland, a measure that correlates with the amount of preening lipid secretions (Oka & Okuyama 2000). This finding further suggests that blacker barn owls invest less effort in preening behaviour. In conclusion, I could not find any evidence that the production of a black eumelanic ornament (Roulin & Dijsktra 2003) and need to preserve this trait in good shape (present study) entail significant costs. This leads me to propose the novel hypothesis that eumelanic individuals are more immunocompetent, resistant to parasites and developmentally stable (Roulin 2004a) not only because genes that determine production level of eumelanic pigments may participate in the regulation of these physiological processes (Roulin 2004a) but also because eumelanic individuals divert energy (Croll & McLaren 1993; Giorgi et al. 2001) and time into other activities than preening behaviour (Redpath 1988; Cotgreave & Clayton 1994; Christe, Richner & Oppliger 1996). This hypothesis could be tested experimentally for instance by forcing deeply and lightly eumelanic individuals to preen at the same level with the prediction that any phenotypic difference between these two categories of birds, as observed in natural conditions (e.g. Roulin 2004a), would decrease. Acknowledgements I thank Céline Ohayon for having weighed wing and tail feathers, Hughes Baudvin for organizing the collection of dead barn owls along French roads by the SAPRR (Société des Autoroutes Paris-Rhin-Rhône), the Swiss National Science Foundation (grant PP00A- 102913) for financial support, and Marietta Wilhelm, Philippe Christe and two anonymous referees for useful comments on a previous draft. References Aparicio, J.M., Bonal, R. & Cordero, P.J. (2003) Evolution of the structure of tail feathers: implications for the theory of sexual selection. Evolution 57, 397 405. Armitage, S.A.O. & Siva-Jothy, M.T. (2005) Immune function responds to selection for cuticular colour in Tenebrio molitor. Heredity 94, 650 656. Battacharyya, S.P. & Chowdhury, S.R. (1995) Seasonal variation in the secretory lipids of the uropygial gland of a subtropical wild passerine bird, Pycnonotus cafer (L.) in relation to the testicular cycle. Biological Rhythm Research 26, 79 87. Bonser, R.H.C. (1995) Melanin and the abrasion resistance of feathers. Condor 97, 590 591. Brommer, J.E., Pihlajamäki, O., Kolunen, H. & Pietiäinen, H. (2003) Life-history consequences of partial-moult asymmetry. Journal of Animal Ecology 72, 1057 1063. Buckley, P.A. (1987) Mendelian genes. Avian Genetics, a Population and Ecological Applicationsroach (eds F. Cooke & P.A. Buckley), pp. 1 44. Academic Press, Orlando, FL. Burtt, E.H. (1979) Tips on wings and other things. The Behavioural Significance of Color (ed. E.H. Burtt), pp. 75 100. Garland STPM Press, New York. Christe, P., Richner, H. & Oppliger, A. (1996) Of great tits and fleas: sleep baby sleep. Animal Behaviour 52, 1087 1092. Clayton, D.H. (1991) Coevolution of avian grooming and ectoparasite avoidance. Bird Parasite Interactions: Ecology, Evolution, and Behaviour (eds J.E. Loye & M. Zuk), pp. 258 289. Oxford University Press, Oxford. Clayton, D.H. & Cotgreave, P. (1994) Relationship of bill morphology to grooming behaviour in birds. Animal Behaviour 47, 195 201. Clayton, D.H., Moyer, B.R., Bush, S.E., Jones, T.G., Gardiner, D.W., Rhodes, B.B. & Goller, F. (2005) Adaptive significance of avian beak morphology for ectoparasite control. Proceedings of the Royal Society of London B 272, 811 817. Cotgreave, P. & Clayton, D.H. (1994) Comparative analysis of time spent grooming by birds in relation to parasite load. Behaviour 131, 171 187. Croll, D.A. & McLaren, E. (1993) Diving metabolism and thermoregulation in common and thick-billed murres. Journal of Comparative Physiology B 163, 160 166. Dawson, A., Hinsley, S.A., Ferns, P.N., Bonser, R.H.C. & Eccleston, L. (2000) Rate of moult affects feather quality: a mechanism linking current reproductive effort to future survival. Proceedings of the Royal Society of London B 267, 2093 2098. Figuerola, J., Munoz, E. & Gutierrez, R. (1999) Blood parasites, leucocytes and plumage brightness in the cirl bunting, Emberiza cirlus. Functional Ecology 13, 594 601. Galeotti, P. & Sacchi, R. (2003) Differential parasitaemia in the tawny owl (Strix aluco): effects of colour morph and habitat. Journal of Zoology (London) 261, 91 99. Giorgi, M.S., Arlettaz, R., Christe, P. & Vogel, P. (2001) The energetic grooming costs imposed by a parasitic mite (Spinturnix myoti) upon its bat host (Myotis myotis). Proceedings of the Royal Society of London B 268, 2071 2075. Glick, B. (1983) Bursa of Fabricius. Avian Biology, Vol. 7 (eds D.S. Farner, J.R. King & K.C. Parkes), pp. 443 500. Academic Press, New York. Gonzales, G., Sorci, G., Møller, A.P., Ninni, P., Haussy, C. & de Lope, F. (1999) Immunocompetence and conditiondependent sexual advertisement in male house sparrows (Passer domesticus). Journal of Animal Ecology 68, 1225 1234. Griffith, S.C., Owens, I.P.F. & Burke, T. (1999) Environmental determination of a sexually selected trait. Nature 400, 358 360.

271 Maintenance cost of melanism Hamilton, W.D. & Zuk, M. (1982) Heritable true fitness and bright birds: a role for parasites? Science 218, 384 387. Hearing, V.J. & Tsukamoto, K. (1991) Enzymatic control of pigmentation in mammals. Journal of the Federation of American Societies for Experimental Biology 5, 2902 2909. Hill, G.E. (1990) Female house finches prefer colourful males: sexual selection for a condition-dependent trait. Animal Behaviour 40, 563 572. Hill, G.E. & Brawner, W.R. (1998) Melanin-based plumage coloration in the house finch is unaffected by coccidial infection. Proceedings of the Royal Society of London B 265, 1105 1109. Jacob, J. & Ziswiler, V. (1982) The uropygial gland. Avian Biology, Vol. 6 (eds D.S. Farmer, J.R. King & K.C. Parkes), pp. 199 324. Academic Press, New York. Järvi, T. & Bakken, M. (1984) The function of the variation in the breast stripe of the great tit (Parus major). Animal Behaviour 32, 590 596. Jawor, J.M. & Breitwisch, R. (2003) Melanin ornaments, honesty, and sexual selection. Auk 120, 249 265. Kose, M. & Møller, A.P. (1999) Sexual selection, feather breakage and parasites: the importance of white spots in the tail of the barn swallow (Hirundo rustica). Behavioral Ecology and Sociobiology 45, 430 436. Majerus, M.E.N. (1998) Melanism, Evolution in Action. Oxford University Press, Oxford. Majerus, M.E.N. & Mundy, N.I. (2003) Mammalian melanism: natural selection in black and white. Trends in Genetics 19, 585 588. McGraw, K.J., Mackillop, E.A., Dale, J. & Hauber, M.E. (2002) Different colors reveal different information: how nutritional stress affects the expression of melanin- and structurally based ornamental plumage. Journal of Experimental Biology 205, 3747 3755. McGraw, K.J., Regan, E.A. & Parker, R.S. (2005) Maternally derived carotenoid pigments affect offspring survival, sex ratio, and sexual attractiveness in a colourful songbird. Naturwissenschaften 92, 375 380. Mennill, D.J., Doucet, S.M., Montgomerie, R. & Ratcliffe, L.M. (2003) Achromatic color variation in black-capped chickadees, Poecile atricapilla: black and white signals of sex and rank. Behavioral Ecology and Sociobiology 53, 350 357. Moyer, B.R., Rock, A.N. & Clayton, D.H. (2003) An experimental test of the importance of preen oil in Rock Doves (Columba livia). Auk 120, 490 496. Newton, I. (1966) The moult of the bullfinch, Pyrrhula pyrrhula. Ibis 108, 41 67. Oka, N. & Okuyama, M. (2000) Nutritional status of dead oiled rhinoceros auklets (Cerorhinca monocerata) in the Southern Japan sea. Marine Pollution Bulletin 40, 340 347. Olson, V.A. & Owens, I.P.F. (1998) Costly sexual signals: are carotenoids rare, risky or required? Trends in Ecology and Evolution 13, 510 514. Piersma, T., Dekker, M. & Sinninghe Damsté, J.S. (1999) An avian equivalent of make-up? Ecology Letters 2, 201 203. Poston, J.P., Hasselquist, D., Stewart, I.R.K. & Westneat, D.F. (2005) Dietary amino acids influence plumage traits and immune responses of male house sparrows, Passer domesticus, but not as expected. Animal Behaviour 70, 1171 1181. Redpath, S. (1988) Vigilance levels in preening dunlin Calidris alpina. Ibis 130, 555 557. Rohwer, S. & Wingfield, J.C. (1981) A field study of social dominance, plasma levels of luteinizing hormone and steroid hormones in wintering Harris sparrows. Ethology 57, 173 183. Roulin, A. (1999) Nonrandom pairing by male barn owls Tyto alba with respect to a female plumage trait. Behavioral Ecology 10, 688 695. Roulin, A. (2001) On the cost of begging: implications of vigilance. Behavioral Ecology 12, 506 515. Roulin, A. (2002) The sibling negotiation hypothesis. The Evolution of Begging: Competition, Cooperation and Communication (eds J. Wright & M. Leonard), pp. 102 127. Kluwer Academic Press, Dordrecht. Roulin, A. (2004a) Proximate basis of the covariation between a melanin-based female ornament and offspring quality. Oecologia 140, 668 675. Roulin, A. (2004b) Covariation between plumage colour polymorphism and diet in the barn owl Tyto alba. Ibis 146, 509 517. Roulin, A. (2006) Linkage disequilibrium between a melaninbased colour polymorphism and tail length in the barn owl. Biological Journal of the Linnean Society 88, 475 488. Roulin, A. & Dijkstra, C. (2003) Genetic and environmental components of variation in eumelanin and phaeomelanin sex-traits in the barn owl. Heredity 90, 359 364. Roulin, A., Richner, H. & Ducrest, A.-L. (1998) Genetic, environmental and condition-dependent effects on female and male plumage ornamentation. Evolution 52, 1451 1460. Roulin, A., Ducrest, A.-L. & Dijkstra, C. (1999) Effect of brood size manipulations on parents and offspring in the barn owl Tyto alba. Ardea 87, 91 100. Roulin, A., Jungi, T.W., Pfister, H. & Dijkstra, C. (2000) Female barn owls (Tyto alba) advertise good genes. Proceedings of the Royal Society of London B 267, 937 941. Roulin, A., Kölliker, M. & Richner, H. (2000) Barn owl (Tyto alba) siblings vocally negotiate resources. Proceedings of the Royal Society of London B 267, 459 463. Roulin, A., Riols, C., Dijkstra, C. & Ducrest, A.-L. (2001a) Female plumage spottiness and parasite resistance in the barn owl (Tyto alba). Behavioral Ecology 12, 103 110. Roulin, A., Riols, C., Dijkstra, C. & Ducrest, A.-L. (2001b) Female- and male-specific signals of quality in the barn owl. Journal of Evolutionary Biology 14, 255 267. Roulin, A., Ducrest, A.-L., Balloux, F., Dijkstra, C. & Riols, C. (2003) A female melanin-ornament signals offspring fluctuating asymmetry in the barn owl. Proceedings of the Royal Society of London B 270, 167 171. Roulin, A., Dauwe, T., Blust, R., Eens, M. & Beaud, M. (2006) A link between eumelanism and calcium physiology in the barn owl. Naturwissenschaften 93, 426 430. Sandilands, V., Powell, K., Keeling, L. & Savory, C.J. (2004) Preen gland function in layer fowls: factors affecting preen oil fatty acid composition. British Poultry Science 45, 109 115. Senar, J.C., Camerino, M., Copete, J.L. & Metcalfe, N.B. (1993) Variation in black bib of the Eurasian siskin (Carduelis spinus) and its role as a reliable badge of dominance. Auk 110, 924 927. Shawkey, M.D., Pillai, S.R. & Hill, G.E. (2003) Chemical warfare? Effects of uropygial oil on feather-degrading bacteria. Journal of Avian Biology 34, 345 349. Siefferman, L. & Hill, G.E. (2005) Evidence for sexual selection on structural plumage coloration in female eastern bluebirds (Sialia sialis). Evolution 59, 1819 1828. Taylor, I.R. (1994) Barn Owls: Predator Prey Relationships. Cambridge University Press, Cambridge. Walther, B.A. & Clayton, D.H. (2004) Elaborate ornaments are costly to maintain: evidence for high maintenance handicaps. Behavioral Ecology 16, 89 95. West, P.M. & Packer, C. (2002) Sexual selection, temperature, and the lion s mane. Science 297, 1339 1343. Wijnandts, H. (1984) Ecological energetics of the long-eared owl (Asio otus). Ardea 72, 1 92. Yorinks, N. & Atkinson, C.T. (2000) Effects of malaria on activity budgets of experimentally infected juvenile apapane (Himatione sanguinea). Auk 117, 731 738. Zahavi, A. (1975) Mate selection a selection for a handicap. Journal of Theoretical Biology 53, 205 214. Zampiga, E., Hoi, H. & Pilastro, A. (2004) Preening, plumage reflectance and female choice in budgerigars. Ethology, Ecology and Evolution 16, 339 349. Received 10 July 2006; accepted 23 October 2006 Editor: Gary Bortolotti