Abstract
The impact of feather-degrading bacilli on feathers depends on the presence or absence of melanin. In vitro studies have demonstrated that unmelanized (white) feathers are more degradable by bacteria than melanized (dark) ones. However, no previous study has looked at the possible effect of feather-degrading bacilli on the occurrence of patterns of unmelanized patches on otherwise melanized feathers. The pied flycatcher Ficedula hypoleuca Pallas, 1764 is a sexually dimorphic passerine with white wing bands consisting of unmelanized patches on dark flight feathers. These patches are considered to be a sexually selected trait in Ficedula flycatchers, especially in males, where the patches are more conspicuous (larger and possibly whiter) than in females. Using in vitro tests of feather bacterial degradation, we compared the degradability of unmelanized and melanized areas of the same feather for 127 primaries collected from the same number of individuals in a population breeding in central Spain (58 males and 69 females). In addition, we also looked for sex differences in feather degradability. Based on honest signalling theory and on the fact that there is stronger sexual selection for males to signal feather quality than in females, we predicted that unmelanized areas should be more degradable by bacteria than melanized ones within the same feather, and that these unmelanized areas should also be more degradable in males than in females. We confirmed both predictions. Microstructural differences between cross-section dimensions of unmelanized and melanized barbs, but not differences in the density of barbs within unmelanized and melanized areas of feathers in males and females, could partly explain differences in degradability. This is the first study to show differences in bacterial degradability among markings on the same feather and among unmelanized feather patches between males and females as predicted by sexual selection theory.
INTRODUCTION
Feather-degrading bacteria (FDB), a polyphyletic assemblage of microorganisms that colonize the plumage of most wild birds (Burtt & Ichida, 1999; Whitaker, Cristol & Forsyth, 2005; Shawkey et al., 2007; Gunderson, Forsythe & Swaddle, 2009; Shawkey, Pillai & Hill, 2009), have recently received increasing attention from avian ecologists (for a review, see Gunderson, 2008; Burtt, 2009). Although no experimental study has to date demonstrated the detrimental effects of FDB on feathers under natural conditions, correlative evidence suggests that these bacteria are active on the plumage, and can damage it (Gunderson et al., 2009; Shawkey et al., 2009). Moreover, the evolution of antibacterial defences, such as uropygial oils (Shawkey, Pillai & Hill, 2003; Reneerkens et al., 2008), suggests that FDB are a real threat for avian plumage in the wild.
An important issue in the context of avian optical communication is the association of the keratinolytic activity of Bacillus licheniformis (Weigmann, 1898), the most studied FDB, with feather coloration. Previous studies have shown that the expression of carotenoid-dependent plumage signals is negatively related to endogenous (Nolan, Hill & Stoehr, 1998; Brawner, Hill & Sundermann, 2000; Hill, Farmer & Beck, 2004) and exogenous (Shawkey et al., 2009) bacterial infections. The FDB also alter the blue structural coloration of feathers (Shawkey et al., 2007; Gunderson et al., 2009). Evidence from domestic (Goldstein et al., 2004; Gunderson et al., 2008) and wild (Burtt et al., 2010) avian species suggests that unmelanized (white) feathers are more degradable by bacteria than melanized (dark) feathers. Recent studies of sexual selection have, however, neglected the potential role of FDB as an evolutionary force driving the honesty of the expression of unmelanized or melanized plumage signals (Jawor & Breitwisch, 2003; Moreno & Møller, 2006; McGraw, 2008).
The occurrence of patterns of white unmelanized patches on otherwise dark melanized feathers is widespread among birds, with a diversity of patterns in different parts of the plumage serving as potential signals of individual quality (Jones, 1990; Price & Pavelka, 1996; Brooke, 1998; Fitzpatrick, 1998; Kose & Møller, 1999; Moreno-Rueda, 2005; McGlothlin et al., 2007; Morales et al., 2007; Galván, 2008; Hegyi, Garamszegi & Eens, 2008; Hegyi et al., 2008; Hanssen et al., 2009). Sexual selection theory predicts that the honesty of sexual and social advertisements may result from signalling costs (Andersson, 1994, Zahavi & Zahavi, 1997). The costs implied by white wing bands consisting of unmelanized patches on otherwise dark melanized flight feathers could derive from producing more resistant barbs, for instance, in terms of molecular organization (Vágási, Pap & Barta, 2010), which can withstand physical abrasion (Barrowclough & Sibley, 1980; Burtt, 1986; Bonser, 1995) and biotic degradation by feather-degrading bacilli (Goldstein et al., 2004; Gunderson et al., 2008; Burtt et al., 2010) or feather lice (Kose & Møller, 1999), even in the absence of melanization. Unmelanized feather markings would therefore be more damaged in individuals of poorer quality (Fitzpatrick, 1998). For instance, Kose & Møller (1999) found that male barn swallows reliably indicated their superior quality by the presence of large white tail patches without parasite damage. Producing larger patches of resistant barbs could be limited by the increasing costs of producing such resistant feather structures. Unmelanized wing patches could thus reveal the capacity of an individual to produce high-quality resistant plumage over a larger surface by removing the defences offered by melanin from certain patches, which would function as windows onto feather resistance. Females would preferably select males of high quality, as expressed reliably by their more conspicuous (larger and possibly whiter) unmelanized wing patches, thereby potentially favouring the increase in conspicuousness of the ornament in the population. Poor-quality males trying to produce large patches would suffer stronger degradation given the poorer structural quality of their feathers. Females could detect such poor-quality males through feather damage and/or barb loss in the unmelanized sections. Moreover, the evolutionary stability and optimization of plumage signals depends on the interactions between sexual and natural selection (Olsen et al., 2010). Thus, wing feathers may offer the clearest index of resistance, as damage of these structures may lead to impaired flight, which would increase the risk of predation (Dale & Slagsvold, 1996; Møller, Erritzoe & Nielsen, 2010), reduce the foraging capacity in aerial insectivores (Swaddle et al., 1996), and increase migration costs (Møller, De Lope & Saino, 2004). Thus, unmelanized wing patches could reveal feather resistance to degradation where it matters most.
The previous argument pertains to the sexual selection of males, and is not applicable to intersexual comparisons. At an intersexual level, it is widely accepted that sexual selection acts more intensely in males than in females, and thus males are generally the strongest signalling sex (Andersson, 1994). We assume that females do not experience a strong selection pressure to signal feather quality through the demelanization of feather patches, as there is much weaker evidence for sexual selection on wing patches in females (Hegyi et al., 2007; Morales et al., 2007) than in males (Sheldon & Ellegren, 1999; Sirkiä & Laaksonen, 2009; Sirkiä, Virolainen & Laaksonen, 2010) in Ficedula flycatchers, and also because the wing patches in females are much smaller than in males (Lundberg & Alatalo, 1992). Only the males would strongly benefit by signalling feather quality through the demelanization of patches.
Unmelanized wing patches used as signals are a widespread phenomenon in birds, although they have been studied little (but see Török, Hegyi & Garamszegi, 2003; Garamszegi et al., 2006; McGlothlin et al., 2007; Morales et al., 2007; Hegyi, Garamszegi & Eens, 2008; Hegyi et al., 2008). The pied flycatcher (Ficedula hypoleuca Pallas, 1764) is a small hole-nesting sexually dimorphic passerine with melanin-based plumage, showing variable combinations of jet-black to brown or grayish brown on the back and wing coverts (Lundberg & Alatalo, 1992), with an unmelanized white forehead patch (Dale et al., 1999; Osorno et al., 2006; Morales et al., 2007; Galván & Moreno, 2009; Moreno et al., 2011) and unmelanized white patches at the base of dark primaries and secondaries. These white wing patches are larger in central Spanish populations than in other central–northern European populations (Curio, 1960), and are larger in males than in females (Lundberg & Alatalo, 1992). They are also present in the closely related collared flycatcher Ficedula albicollis (Temminck, 1815) (Hegyi, Török & Tóth, 2002). The size (length and area) of the unmelanized wing patch is important to sexual selection in Ficedula flycatchers, as shown by males with larger unmelanized wing patches having a higher mating (Sheldon & Ellegren, 1999; Sirkiä & Laaksonen, 2009) and reproductive (Sirkiä, Virolainen & Laaksonen, 2010) success, and a higher survivorship (Török et al., 2003). Literature is limited and less conclusive about the possible role of the unmelanized patch in the sexual selection of females. Hegyi et al. (2007) and Morales et al. (2007) found that unmelanized wing patches indicate female phenotypic quality, and suggest that it may have a role in sexual selection. Furthermore, because the unmelanized patch is maintained throughout the year in males and females, it may be important for non-breeding social interactions such as territorial competition (e.g. for F. albicollis, see Garamszegi et al., 2006; Hegyi et al., 2008). Ficedula hypoleuca thus appears as an appropriate species in which to look for sex differences in the effects of feather-degrading bacilli on an unmelanized plumage ornament potentially exposed to sexual selection.
Based on in vitro feather degradation tests, we tested the following predictions derived from the fact that unmelanized feathers are more degradable by bacteria than melanized ones (Goldstein et al., 2004; Gunderson et al., 2008), and from the fact that sexual selection generally acts more intensely in males than in females (Andersson, 1994).
1 White unmelanized wing patches on primary feathers are more degradable by bacteria than the contiguous dark melanized areas within the same feather. This would be expressed as a higher concentration of by-products of keratin degradation per unit of feather fragment in unmelanized rather than melanized patches.
2 White unmelanized wing patches on primary feathers are more degradable by bacteria in males than in females, as there is stronger sexual selection on males to signal feather quality through the exhibition of unprotected patches. This sex difference would be expressed as a higher concentration of by-products from keratin degradation per unit of feather fragment in unmelanized patches in males than in females. Because dark melanized areas on primary feathers are not sexually selected, their bacterial degradability does not differ between the sexes.
Additionally, we checked for possible differences in barb density between unmelanized and melanized areas of feathers, and between sexes, and for possible differences in the microstructural dimensions of these barbs.
To our knowldege, this is the first study aiming to describe intrafeather differences in the bacterial degradability of a melanin-based sexually dimorphic plumage pattern in a wild bird, and to explore possible sex differences in the effect of feather-degrading bacilli in the context of sexual selection and honest signalling.
MATERIAL AND METHODS
STUDY AREA AND SPECIES
We conducted this study in the spring of 2009 and 2010 on a population of F. hypoleuca breeding in nest boxes in a montane oak forest (Quercus pyrenaica), at 1200 m a.s.l. in Valsaín, central Spain (40°54′N, 04°01′W). We followed the breeding activity in each nest box occupied by F. hypoleuca from early stages of nest construction to fledging. Ficedula hypoleuca is a small (12–13 g) hole-nesting passerine of European woodlands that accepts artificial nest boxes for breeding and is used as a model species in ecology (Lundberg & Alatalo, 1992). Monitoring of the study population has been in progress since 1991 (Moreno et al., 2005).
FEATHER SAMPLING
We captured 61 males and 72 females in 2009, and 20 males and 20 females in 2010 during the nestling stage (nine males and seven females were recaptured in 2010, although feathers collected in each season of the same individuals were used in different and independent analyses, thereby preventing pseudoreplication). Feathers collected in 2009 were used for in vitro bacterial degradation tests and in measurements of the microstructural cross-section dimensions of barbs (see details below). Feathers collected in 2010 were used in measurements of the surface of the white wing feather patch and barb density of melanized and unmelanized areas of the feathers (see details below).
We captured birds within nest boxes when they entered them for feeding their young using standard traps. We removed the fourth primary on the right (in 2009) and left (in 2010) wing from each bird, and immediately put the feather into a sterile tube, which was kept in a portable cooler (for a maximum period of 5 h) until being stored in a dark cold room at 4 °C. In the field we used aseptic techniques. We used latex gloves and forceps for all manipulations of birds and feathers. We replaced gloves and disinfected them with ethanol (97%) for each new manipulation. Forceps were flame sterilized using an alcohol burner. Feathers collected at this stage in each reproductive season (2009 or 2010) had been worn by birds since the last moult at the end of the previous breeding season. All feathers collected were stored in the cold room until the autumn of 2010 in the case of feathers from 2009, when they were degraded by bacteria in vitro (see details below), and until the spring of 2011 in the case of feathers from 2010, when patch areas and barb densities were measured (see details below).
We assume that the possible effects of this storage period within the cold room does not influence the in vitro bacterial degradability tests of feathers collected in 2009. First, any effects of this storage period on the natural bacterial community of feathers should be equivalent for all feathers, being of the same duration. In addition, Bacillus licheniformis (the most important FDB) growing naturally on feathers was presumably forced to sporulate, and thus remained inactive because it grows optimally at high temperatures. We cannot exclude that other less prevalent and harmful FDB were active on feathers, and could have contributed to any possible damage observed. However, we checked feathers before starting in vitro degradation tests, and none of them showed any macroscopic (detected with the naked eye) physical damage.
MICROSTRUCTURAL DIMENSIONS OF UNMELANIZED AND MELANIZED BARBS
Using the scanning electron microscope, we measured six unmelanized and six melanized barbs from six fourth primary feathers from three females and three males captured in 2009 that were not included in the degradation tests (one unmelanized and one melanized barb was examined from each feather from each individual). As barbs are elliptical in cross section, we measured the width with and without the cortex, and the maximum length of the cross section of each barb, as well as the width of its cortex at the blunt end of the barb. All 12 barbs were cut as close as possible to the rachis and mounted vertically in hot glue on a scanning electron microscope stub, with the cross section nearest the rachis being uppermost. The barbs, glue, and stub were gold-coated in an SPI sputter coater (SPI Supplies, Division of Structure Probe Inc., West Chester, PA, USA). The stubs were mounted on the stage of an LEO 435 VP scanning electron microscope (Carl Zeiss, NTS Inc., Peabody, MA, USA) and observed at 100–300× magnification for measuring cross-section dimensions.
BARB DENSITY IN UNMELANIZED AND MELANIZED AREAS OF FEATHERS
To estimate barb density in the unmelanized and melanized areas of feathers we used the set of 40 fourth primary feathers from the left wings of F. hypoleuca (20 from males and 20 from females) collected in 2010. For each feather, we quantified the number of barbs along 0.3 cm of rachis for the unmelanized and melanized areas (number of barbs counted were subsequently recalculated for a 1-cm length). We related barb density in equivalent unmelanized and melanized areas of each feather to the areas that we considered in the in vitro bacterial degradation tests for feathers collected in 2009 (see details below). We used a stereoscopic microscope (Olympus SZX7) to quantify barbs, and all counts were made by the same observer.
DEGRADATION OF UNMELANIZED AND MELANIZED FRAGMENTS OF FEATHERS
We included a total of 127 fourth primary wing feathers from 58 males and 69 females captured in 2009 in the in vitro bacterial degradation tests. Using a digital caliper we measured the length of the unmelanized patch at the base of each primary feather (excepting three male feathers, because of technical problems), and separated the white patch from the rest of the melanized feather. We also removed an adjoining dark melanized fragment of feather of exactly the same length as the white unmelanized fragment of the same feather. Using the naked eye, unmelanized and melanized areas within each feather looked clearly and distinctly white and dark (brown–black), respectively. We have assumed that our degradability measures are relative to mass or surface of feather fragments. However, we were unable to weigh the fragments as the extremely low masses were below the accuracy of standard precision balances (± 0.02 mg). On the other hand, the length of fragments was strongly and positively correlated with the surface area of fragments in a different sample of fourth primary feathers from the left wings of 20 males and 20 females collected in 2010 (previously used in barb density quantification) (r = 0.96, P < 0.001; r = 0.86, P < 0.001; for males and females, respectively). We have therefore assumed that by correcting for fragment length in the present analyses we are relativizing for fragment area, and therefore presumably for fragment mass.
We placed each unmelanized and melanized fragment of feather in independent test tubes with 3 mL of media (10 mL of five different saline solutions, 0.10 g yeast extract, 950 mL deionized water; see Williams et al., 1990 for details), autoclaved to sterilize the feather media (previous media plus the fragment of feather to be degraded). Autoclaving could affect the molecular structure of feathers (e.g. by the denaturation of proteins), and thus their subsequent bacterial degradability in vitro (Gunderson et al., 2008). This may be particularly true for melanized feathers, which contain more proteins (melanin and keratin) than unmelanized ones (Goldstein et al., 2004). Thus, if our prediction that unmelanized areas are significantly more degradable than contiguous melanized areas within the same feather is confirmed consistently, we would assume that autoclaving has no relevant confounding effect (Gunderson et al., 2008). Autoclaving has been used in other recent studies of feather degradation, and therefore seems adequate (Gunderson et al., 2008; Burtt et al., 2010). Each of the tubes was inoculated with feather-degrading bacilli (B. licheniformis, OWU 138B, ATCC 55768) following the method described by Goldstein et al. (2004). We incubated the tubes for 120 h in an incubation shaker at 37 ± 1 °C. We removed two samples of 200 µL from each tube, beginning immediately before inoculation (0 h) and the 120 h after inoculation. We transferred each sample to a microcentrifuge tube and centrifuged at 14 000 g for 10 min. The clear supernatant was transferred to a second microcentrifuge tube and stored in a cold room (4 °C) until the end of the experiment, 120 h after inoculation, when we measured the absorbance at 562 nm. A bovine serum albumen standard curve was used to convert absorbance measurements to micrograms of oligopeptides per mL. Oligopeptides are a by-product of the bacterial degradation of feather keratin, and their concentration in the medium increases as the feather degrades (Goldstein et al., 2004). We subtracted the initial value (0 h) from the final value (120 h) to obtain the absolute concentration of oligopeptides. This value informs us about the vulnerability of feathers to bacterial degradation, here termed as bacterial degradability. Five and six cases for unmelanized and melanized fragments of feathers from females, respectively, and 11 cases for melanized fragments of feathers from males were excluded because of methodological errors resulting in a lower final concentration than the initial concentration. Thus, we finally obtained valid measurements of bacterial degradation for 64 unmelanized and 63 melanized fragments from females, and 58 unmelanized and 47 melanized fragments from males.
STATISTICAL ANALYSES
We tested for possible differences in the microstructural cross-section dimensions of barbs in relation to colour (white unmelanized versus dark melanized) and sex (males versus females) using the VARIANCE COMPONENT module of STATISTICA 7.0. We included individual as a random factor.
As barb counts per cm of feather were not normally distributed, we used non-parametric tests for the analysis where this variable was dependent. First, we used the Wilcoxon matched-pairs test to test for possible differences in barb density between the unmelanized and melanized areas within each feather in males and females, independently. Second, we used the Mann–Whitney U-test to test for possible differences between sexes in barb density in unmelanized and melanized areas of feathers, independently.
To analyse the variation in the bacterial degradability of fragments of feathers (including all fragments, both matched and unmatched), we constructed all possible models including colour (white unmelanized versus dark melanized) and sex (males versus females) as binomial factors, together with the interaction of these two factors (colour and sex) and feather fragment length as a continuous covariate. We also included individual as a random factor in all models. We performed model selection among all nine possible models using SAS 9.1. We used the corrected Akaike information criterion = AIC to select the model with the highest plausibility. We obtained AICc values and weights for each model with SAS 9.1. We specifically selected the most plausible model (lowest AIC and highest explanatory weight) and ran this model using the VARIANCE COMPONENT module of STATISTICA 7.0. When using number of barbs in each fragment exposed to degradation (based on the highly conserved barb density values observed for feathers from 2010) instead of fragment length as a covariate in the model, the results did not change. Thus, we will only present the tests controlling for fragment length. Finally, using the GLM module of STATISTICA 7.0, we performed complementary analyses to compare specifically the degradability of the unmelanized and melanized feather fragments between males and females.
RESULTS
We found that the cross-section total and partial widths of melanized barbs were significantly larger than those of unmelanized barbs (F1,12 = 24.83, P = 0.01; F1,12 = 29.98, P = 0.01, respectively), but we found no significant differences in cross-section total length or width of the cortex (F1,12 = 5.41, P = 0.08; F1,12 = 0.31, P = 0.61, respectively). None of the microstructural dimensions of barbs differed between the sexes (all F1,12 < 6.20, P > 0.07) (Fig. 1).
Elliptical cross section of a melanized barb of the fourth primary feather of the right wing of Ficedula hypoleuca. The maximum length and total width (including cortex) are delimited with pairs of crosses, P1–P1R and P2–P2R, respectively. Dark boxes show the measurements in µm, delimited by each pair of crosses.
The number of barbs in the unmelanized and melanized areas of each feather was highly conserved in males and females (males, N = 20, mean number of unmelanized barbs per cm of feather = 40.67 ± 0.39, mode = 40.00, mean number of melanized barbs per cm of feather = 36.67 ± 0.00, mode = 36.67; females, mean number of unmelanized barbs per cm of feather = 40.33 ± 0.41, mode = 40.00, mean number of melanized barbs per cm of feather = 35.67 ± 0.35, mode = 36.67). The unmelanized areas of feathers had significantly more barbs per cm than their contiguous melanized areas in both males and females (N = 20, Z = 3.82, P < 0.001; N = 20, Z = 3.92, P < 0.001, respectively). When looking specifically at each area of the feathers (unmelanized or melanized), males showed significantly more barbs per cm than females in the melanized part (Mann–Whitney U-test, N = 40, Z = 2.62, P = 0.01), but not in the unmelanized one (Mann–Whitney U-test, N = 40, Z = 0.57, P = 0.57).
Males had significantly longer unmelanized fragments than females (17.92 ± 0.26 versus 15.60 ± 0.31 mm, F1,124 = 30.74, P < 0.001).
Among the nine different models obtained, the most plausible one (lowest AICc and highest weight) included colour, sex, interaction of colour and sex, and length of feather fragment as explanatory predictors (individual was included as random factor) (Table 1).
Complete set of predictive models for the variation of bacterial degradability of melanized and unmelanized fragments of feathers from 58 males and 69 females from a Spanish pied flycatcher (Ficedula hypoleuca) population, including individual as a random factor
| Model | K | AICc | ΔAICc | Weight |
|---|---|---|---|---|
| Sex, colour, sex × colour, length of fragment | 5 | 2676.3 | 0 | 1.0 |
| Sex, colour, length of fragment | 4 | 2706.5 | 30.2 | 0.0 |
| Colour, length of fragment | 3 | 2716.2 | 39.9 | 0.0 |
| Sex, length of fragment | 3 | 2750.6 | 74.3 | 0.0 |
| Sex, colour, sex × colour | 4 | 2758.3 | 82 | 0.0 |
| Length of fragment | 2 | 2761.3 | 85 | 0.0 |
| Sex colour | 3 | 2789.5 | 113.2 | 0.0 |
| Colour | 2 | 2805.8 | 129.5 | 0.0 |
| Sex | 2 | 2835.5 | 159.2 | 0.0 |
| Model | K | AICc | ΔAICc | Weight |
|---|---|---|---|---|
| Sex, colour, sex × colour, length of fragment | 5 | 2676.3 | 0 | 1.0 |
| Sex, colour, length of fragment | 4 | 2706.5 | 30.2 | 0.0 |
| Colour, length of fragment | 3 | 2716.2 | 39.9 | 0.0 |
| Sex, length of fragment | 3 | 2750.6 | 74.3 | 0.0 |
| Sex, colour, sex × colour | 4 | 2758.3 | 82 | 0.0 |
| Length of fragment | 2 | 2761.3 | 85 | 0.0 |
| Sex colour | 3 | 2789.5 | 113.2 | 0.0 |
| Colour | 2 | 2805.8 | 129.5 | 0.0 |
| Sex | 2 | 2835.5 | 159.2 | 0.0 |
Corrected Akaike information criterion (AICc) values are computed using the Satterthwaite method. Models are ranked according to their AICc values. ΔAICc indicates the difference between a model and the best model (lowest AICc). Weight indicates the relative plausibility of each model.
Complete set of predictive models for the variation of bacterial degradability of melanized and unmelanized fragments of feathers from 58 males and 69 females from a Spanish pied flycatcher (Ficedula hypoleuca) population, including individual as a random factor
| Model | K | AICc | ΔAICc | Weight |
|---|---|---|---|---|
| Sex, colour, sex × colour, length of fragment | 5 | 2676.3 | 0 | 1.0 |
| Sex, colour, length of fragment | 4 | 2706.5 | 30.2 | 0.0 |
| Colour, length of fragment | 3 | 2716.2 | 39.9 | 0.0 |
| Sex, length of fragment | 3 | 2750.6 | 74.3 | 0.0 |
| Sex, colour, sex × colour | 4 | 2758.3 | 82 | 0.0 |
| Length of fragment | 2 | 2761.3 | 85 | 0.0 |
| Sex colour | 3 | 2789.5 | 113.2 | 0.0 |
| Colour | 2 | 2805.8 | 129.5 | 0.0 |
| Sex | 2 | 2835.5 | 159.2 | 0.0 |
| Model | K | AICc | ΔAICc | Weight |
|---|---|---|---|---|
| Sex, colour, sex × colour, length of fragment | 5 | 2676.3 | 0 | 1.0 |
| Sex, colour, length of fragment | 4 | 2706.5 | 30.2 | 0.0 |
| Colour, length of fragment | 3 | 2716.2 | 39.9 | 0.0 |
| Sex, length of fragment | 3 | 2750.6 | 74.3 | 0.0 |
| Sex, colour, sex × colour | 4 | 2758.3 | 82 | 0.0 |
| Length of fragment | 2 | 2761.3 | 85 | 0.0 |
| Sex colour | 3 | 2789.5 | 113.2 | 0.0 |
| Colour | 2 | 2805.8 | 129.5 | 0.0 |
| Sex | 2 | 2835.5 | 159.2 | 0.0 |
Corrected Akaike information criterion (AICc) values are computed using the Satterthwaite method. Models are ranked according to their AICc values. ΔAICc indicates the difference between a model and the best model (lowest AICc). Weight indicates the relative plausibility of each model.
The length of the fragment exposed to degradation was significantly and positively associated with feather degradation (Table 2). Interestingly, the interaction of colour and sex was significant (Fig. 2; Table 2). Specifically, the unmelanized fragments of feathers were more degradable by bacteria than the melanized fragments in both males (N = 99, F = 56.77, P < 0.001) and females (N = 127, F = 8.86, P < 0.01) (Fig. 2; Table 2). Unmelanized patches were significantly more degradable in males than in females (N = 119, F = 6.89, P < 0.01), but no significant differences were found in the melanized part of the feathers (N = 107, F = 0.19, P = 0.66) (Fig. 2).
Variance components analysis for the bacterial degradability of melanized and unmelanized fragments of feathers from 58 males and 69 females from a Spanish pied flycatcher (Ficedula hypoleuca) population, including individual as random factor
| Predictors | Effect | d.f. | F | P |
|---|---|---|---|---|
| Colour | Fixed factor | 1,119 | 59.02 | < 0.001 |
| Sex | Fixed factor | 1,114 | 3.31 | 0.07 |
| Colour × sex | Fixed factor | 1,101 | 31.95 | < 0.001 |
| Length of fragment | Fixed covariate | 1,120 | 10.75 | < 0.01 |
| Individual | Random factor | 120,101 | 1.74 | < 0.01 |
| Predictors | Effect | d.f. | F | P |
|---|---|---|---|---|
| Colour | Fixed factor | 1,119 | 59.02 | < 0.001 |
| Sex | Fixed factor | 1,114 | 3.31 | 0.07 |
| Colour × sex | Fixed factor | 1,101 | 31.95 | < 0.001 |
| Length of fragment | Fixed covariate | 1,120 | 10.75 | < 0.01 |
| Individual | Random factor | 120,101 | 1.74 | < 0.01 |
The d.f. errors are computed using the Satterthwaite method. Tests assume that entangled fixed effects are 0.
Variance components analysis for the bacterial degradability of melanized and unmelanized fragments of feathers from 58 males and 69 females from a Spanish pied flycatcher (Ficedula hypoleuca) population, including individual as random factor
| Predictors | Effect | d.f. | F | P |
|---|---|---|---|---|
| Colour | Fixed factor | 1,119 | 59.02 | < 0.001 |
| Sex | Fixed factor | 1,114 | 3.31 | 0.07 |
| Colour × sex | Fixed factor | 1,101 | 31.95 | < 0.001 |
| Length of fragment | Fixed covariate | 1,120 | 10.75 | < 0.01 |
| Individual | Random factor | 120,101 | 1.74 | < 0.01 |
| Predictors | Effect | d.f. | F | P |
|---|---|---|---|---|
| Colour | Fixed factor | 1,119 | 59.02 | < 0.001 |
| Sex | Fixed factor | 1,114 | 3.31 | 0.07 |
| Colour × sex | Fixed factor | 1,101 | 31.95 | < 0.001 |
| Length of fragment | Fixed covariate | 1,120 | 10.75 | < 0.01 |
| Individual | Random factor | 120,101 | 1.74 | < 0.01 |
The d.f. errors are computed using the Satterthwaite method. Tests assume that entangled fixed effects are 0.
Bacterial degradability (least square means controlling for fragment size with 95% confidence intervals) of feather fragments in relation to colour (melanized versus unmelanized) and sex (males, filled symbols, versus females, open symbols) from 127 Ficedula hypoleuca specimens from a central Spanish population.
DISCUSSION
The white unmelanized fragments of primary feathers (white unmelanized wing feather patches) were more degradable by feather-degrading bacilli than contiguous dark melanized fragments in both sexes. The unmelanized wing feather patches were also more degradable by bacteria in males than in females, but degradability of the melanized fragments of feathers did not differ between sexes when controlling for differences in the length of the fragments exposed to degradation (or number of barbs per cm of fragment). We also found that melanized barbs were wider in cross section than unmelanized barbs. The density of barbs was higher in unmelanized than in melanized fragments of each feather, and in males than in females when comparing melanized areas of feathers.
Our results are consistent with previous in vitro tests showing that the bacterial degradability of unmelanized feathers is greater than that of melanized feathers (Goldstein et al., 2004; Gunderson et al., 2008). However, these previous studies compared entire unmelanized and melanized feathers, whereas here we have compared for the first time, to our knowledge, the degradability of unmelanized and melanized areas within the same feather. The unmelanized patches on wing feathers expressed as white wing bands are widespread in birds, and are used for signalling in social and sexual contexts (Ilina, 2004; Garamszegi et al., 2006; Sirkiä & Laaksonen, 2009). Thus, the size of the unmelanized patches on flight feathers may signal the structural strength of primary and secondary feathers, thereby indicating the superior quality of an individual that can grow feathers that resist bacterial degradation over a larger surface without the protection awarded by melanin. Bacterial degradation in natural conditions may be less intense than reported here, and hence allow unmelanized patches to degrade less. Only a few recent studies have provided insight into the possible role of FDB in the context of sexual selection (Shawkey et al., 2007, 2009; Gunderson et al., 2009; Burtt et al., 2010). As predicted from models of parasite-mediated sexual selection, Shawkey et al. (2009) found that redder male house finches, Carpodacus mexicanus (Müller, 1776), had less FDB in their plumage. Shawkey et al. (2007) and Gunderson et al. (2009) showed that FDB alter the spectral properties of bluebird feathers under natural conditions.
We also found that feather-degrading bacilli affected unmelanized wing feather patches more in males than in females in vitro. Although the external chemical protection against bacteria offered from uropygial oil may differ between the sexes (Shawkey et al., 2003; Reneerkens et al., 2008; Soler et al., 2008), with females producing a more protective oil (chemically and/or physically) or preening more than males (Reneerkens et al., 2008; Martin-Vivaldi et al., 2009), it is extremely unlikely that such possible differences could explain our results, because oil present on feathers before capture will certainly lose its functional properties after autoclaving, and will thus not serve as a protection of feathers during the in vitro degradation tests. However, our result is in accordance with the widely accepted fact that sexual selection in most species acts more intensely in males than in females, and thus that males are generally the strongest signalling sex (Andersson, 1994). Females may not experience a strong selection pressure to signal feather quality through demelanization of feather patches, as there is much weaker evidence for sexual selection on wing patches in female (Hegyi et al., 2007; Morales et al., 2007) than in male (Sheldon & Ellegren, 1999; Sirkiä & Laaksonen, 2009; Sirkiä, Virolainen & Laaksonen, 2010) Ficedula flycatchers, as also indicated by the much smaller wing patches in females than in males (Lundberg & Alatalo, 1992). Although future research is necessary for confirmation, unmelanized feather patches presumably differ between males and females in their level of demelanization and/or in molecular arrangements (Vágási et al., 2010), and therefore in their subsequent spectral properties, which could ultimately affect the bacterial degradability. Thus, mainly males would benefit through signalling feather quality through demelanization of patches. These patches in females could just be the consequence of a genetic correlation with male patches (Lande, 1980) and/or their maintenance could result from social selection (West-Eberhard, 1983).
We have found differences in microstructural cross-section dimensions between unmelanized and melanized barbs, potentially supporting the observed greater degradability of unmelanized barbs. Wider melanized barbs could be structurally stronger for resisting bacterial degradation. Interestingly, Voitkevich (1966) already reported on the greater thickness of the cortex of melanized barbs compared with unmelanized barbs. Apparently, there were no differences in microstructural cross-section dimensions in unmelanized barbs between males and females, but, because of limited sample sizes, we cannot discard that differences observed between the degradability of unmelanized feather patches in males and females could still be related to barb microstructure.
This should be considered as a possibility, and remains to be studied in the future with larger samples.
Interestingly the modal number of barbs per cm of feather seems to be highly conserved in males and females, possibly because of a stabilizing selection process in the context of optimizing flight performance. We found, however, that the number of barbs per cm differed significantly between unmelanized and melanized areas of feathers in both males and females, and between males and females in melanized areas but not in unmelanized areas. Because bacteria degraded unmelanized fragments of feathers more effectively, having more barbs in the more degradable area could represent a defensive strategy to ensure the conservation of a minimum number of barbs under the pressure of natural bacterial degradation. Alternatively, this observed difference in the density of barbs between unmelanized and melanized portions within the same feather could simply result from the fact that the unmelanized portion has a more basal position (Butler & Johnson, 2004). We do not have a clear explanation for the larger density of melanized barbs observed in males than in females, but we presume that this could be related to differences in the physical properties of feathers that could be important in the context of flight performance (Butler & Johnson, 2004). This is speculative, and requires future research.
Differences in the degradability observed within feathers and between sexes do not seem to be supported by our findings regarding barb density. On the one hand, females showed a greater difference in barb density between unmelanized and melanized areas, but a smaller difference in the degradability between these areas, than did males. On the other hand, the barb density of unmelanized areas of feathers did not differ between sexes, although the bacterial degradability did. In contrast, the barb density of melanized areas of feathers did differ between sexes, although bacterial degradability did not. This suggests that barb density is not a good predictor of variation in the bacterial degradability of feathers.
To conclude, unmelanized barbs are more degradable by bacteria than melanized barbs within the same feather in the primaries of birds with unmelanized wing patches. These unmelanized patches in male F. hypoleuca are more degradable by bacteria than in females. This suggests that males with larger unmelanized patches, expressed as more conspicuous wing bands, may be honestly signalling their capacity to produce highly resistant flight feathers that can withstand bacterial degradation over a larger surface without the protection awarded by melanin. Selection for these signals would be weaker in females.
ACKNOWLEDGEMENTS
We thank J. Donés (Director de Montes de Valsaín) for permission to work in the study locality. Junta de Castilla y León authorized the capturing and handling of birds. S. Merino and his research group helped in the field. We thank M. Schroeder, L. Tuhela, and C. Hamrick for guidance in laboratory work. L. Tuhela and D. Ordosh helped with the scanning electron microscopy. J. Morales helped with statistics and contributed with valuable comments on previous drafts of the article. This study was funded by projects CGL2007-61251/BOS and CGL2010-19233-C03-02 to JM from the Spanish Ministry of Science and Innovation. SGB and RRC were supported by grants FPI-MEC and JAE-CSIC, respectively. This study is a contribution to the research developed at ‘El Ventorrillo’ field station. This study complies with current Spanish laws.
