Relationships among plumage coloration, blood selenium concentrations and immune responses of adult and nestling tree swallows

In many animals, individuals display elaborate ornaments that are used as signals to conspecifics (Andersson, 1994). Because these ornaments are often costly to produce or maintain, only individuals in the best condition will be capable of possessing the most elaborate ornaments, making ornament expression a reliable indicator of phenotypic and/or genetic quality (Zahavi, 1975; Hamilton and Zuk, 1982; Grafen, 1990). In males, these ornaments appear to be maintained primarily by sexual selection because more ornamented males are preferred mates or are superior competitors, and as a result obtain more mating opportunities and greater reproductive success (Andersson, 1994). However, in a number of taxa, females display ornamentation that is also associated with individual quality (Amundsen and Pärn, 2006; Kraaijeveld et al., 2007). Traditional sexual selection likely plays a limited role in maintaining female ornamentation because female reproductive success is largely limited by their ability to produce young rather than their access to mates (LeBas, 2006; Tobias et al., 2012). Female ornamentation is more likely maintained by social selection, competition among conspecific females for resources other than mates (LeBas, 2006; Tobias et al., 2012). Because different selective pressures maintain ornamentation in each sex, it is possible that similar ornaments are related to different aspects of individual and/or offspring quality in each sex (Massaro et al., 2003; Kekäläinen et al., 2010; Kelly et al., 2012; Pickett et al., 2013).

In many birds, both sexes display colorful plumage that may be produced by pigments deposited in feathers (carotenoid- and melanin-based colors) or by the feather microstructure (structural coloration). One aspect of individual quality that plumage coloration is frequently linked to is the ability to resist parasite infection or mount an immune response (Dunn et al., 2010; Kelly et al., 2012). While previous work focused largely on pigment-based, particularly carotenoid, coloration, recent work has shown that structural coloration may also be related to immunity or levels of parasitism (Doucet and Montgomerie, 2003; Bonato et al., 2009; Griggio et al., 2010). For example, in male satin bowerbirds (Ptilonorhynchus violaceus), brighter blue or greater UV chroma is associated with lower parasite loads (Doucet and Montgomerie, 2003). Male ostriches (Struthio camelus) and south polar skuas (Catharacta maccormicki) with larger or more colorful white plumage patches mount a stronger humoral immune response (Bonato et al., 2009; Hanssen et al., 2009). Finally, stronger cutaneous immune responses [response to phytohemagglutinin (PHA) injection] have been associated with greater UV reflectance of green plumage in budgerigars (Melopsittacus undulates) (Griggio et al., 2010). Only recently have studies examined the relationship between female coloration and immunity and have primarily focused on white plumage patches. In female common eiders (Somateria mollissima), smaller white wing bands are associated with immunosuppression (Hanssen et al., 2006), while larger wing patches are associated with lower parasite loads in Eurasian black-billed magpies (Pica pica) (Blanco and Fargallo, 2013). A greater number of white spots and spots with greater UV reflectance are associated with a stronger cutaneous immune response in female diamond firetails (Stagonopleura guttata) (Zanollo et al., 2012). It seems that in both sexes, structural plumage coloration may be associated with ‘good genes’ for greater immunocompetence or more colorful individuals may be in better condition and thus mount stronger immune responses.

The plumage coloration of adults may also be related to immune responses of their offspring. For instance, white plumage coloration of male ostriches is positively associated with the humoral immune response of their offspring (Bonato et al., 2013). In female great tits (Parus major), more immaculate white cheek patches are associated with greater cutaneous immune responses in offspring (Remeš and Matysioková, 2013). Relationships between parental coloration and offspring immunity could arise in several ways. First, more colorful adults may have better immune responses and if the basis for immunity is heritable, then offspring of more colorful parents will also have stronger immune responses (Folstad and Karter, 1992; von Schantz et al., 1999). Transgenerational epigenetic effects can also enhance offspring immunity (Grindstaff et al., 2003; Hasselquist and Nilsson, 2009; Clairardin et al., 2011) and could be associated with parental coloration. Alternatively, more colorful adults may provide greater or higher quality resources to their young (Linville et al., 1998; Siefferman and Hill, 2003; Garcia-Navas et al., 2012), which could enhance offspring immunity because mounting an immune response can be energetically costly and influenced by condition (Norris and Evans, 2000; Lifjeld et al., 2002; Palacios et al., 2012; Pickett et al., 2013).

One aspect of nutritional condition that may influence the strength of the immune response is the availability of micronutrients such as carotenoids, vitamins or trace elements (reviewed in Hasselquist and Nilsson, 2012). These factors may directly enhance the immune response by facilitating immune cell proliferation, production of cytokines and activity of immune cells, or by altering gene expression (reviewed in Wintergerst et al., 2007; Hasselquist and Nilsson, 2012). These factors may also influence the strength of the immune response through their antioxidant activity and ability to scavenge reactive oxygen species that can be immunosuppressive (reviewed in Wintergerst et al., 2007; Hasselquist and Nilsson, 2012). In birds, the majority of micronutrient studies have focused on the role of carotenoids in immunity because these pigments are also used extensively in plumage coloration (Lozano, 1994; von Schantz et al., 1999). While carotenoid supplementation may be beneficial, high carotenoid intake may be detrimental (Huggins et al., 2010; Giraudeau et al., 2013), suggesting that there is an optimal level of micronutrient supplementation. Given that brighter plumage coloration is associated with greater antioxidant defenses in some species (Perez-Rodriguez et al., 2010; Marko et al., 2011), it is possible that the point at which micronutrients become toxic varies among individuals. Additionally, the role of trace elements in modulating the immune response has not been well examined in wild birds, despite the fact that optimal concentrations of some elements, such as selenium, are necessary for mounting robust immune responses and are important components of enzymes with antioxidant functions (Surai, 2002,, 2006; Wintergerst et al., 2007; Huang et al., 2012).

Tree swallows, Tachycineta bicolor (Vieillot 1808), provide a model system for examining the relationships between structural coloration, immunity and micronutrient status. All male tree swallows have iridescent blue–green dorsal plumage with white ventral plumage (Robertson et al., 2011). Females display delayed plumage maturation; after second year (ASY) females have blue–green dorsal plumage identical to males, while second year (SY) females have brown dorsal plumage and both have white ventral plumage (Stutchbury and Robertson, 1987b). Because of this dichotomy in female coloration, most studies to date have focused on ASY females and males, and found that plumage coloration is a signal used in mate choice and associated with aspects of individual quality. Tree swallows pair assortatively by plumage brightness (Bitton et al., 2008) and males that are brighter and/or older produce more extra-pair offspring (Bitton et al., 2007). Males that are brighter, with lower wavelength hues are more likely to be recaptured in subsequent breeding seasons, suggesting they have greater site fidelity or higher survival than males that are duller, with green-shifted hues (Bitton and Dawson, 2008). ASY females with bluer plumage hues fledge more offspring while brighter dorsal plumage is positively associated with mean egg mass (Bitton et al., 2008; but see Bentz and Siefferman, 2013). However, another study found that more colorful ASY females produce smaller swellings in response to PHA injection, suggesting they have weaker immune responses (Bentz and Siefferman, 2013). Overall, these results suggest tree swallows that are brighter with bluer hues are higher quality than dull (lower brightness) individuals with green-shifted hues (but see Bentz and Siefferman, 2013). However, no study to date has examined the reflectance of white breast plumage to determine whether it is a signal of quality or how parental coloration relates to offspring immunity. Additionally, concentrations of the micronutrient selenium are positively associated with the bactericidal capacity of nestling plasma (Beck et al., 2014) and with wing growth (Beck et al., 2015) in this species, revealing the need for further examination of the role of selenium in immunity.

The purpose of this study was threefold. We examined two types of structural plumage coloration, the iridescent blue/green back and white breast, of both sexes of adult tree swallows. We first examined variation in plumage coloration between the sexes and determined whether they pair assortatively by coloration. Second, we determined whether structural plumage coloration relates to the immune response of adults or that of their nestlings by examining bactericidal capacity (adults and nestlings) and the response to injection of PHA (nestlings only). In this case, we were specifically interested in whether we would find different relationships between coloration and immunity in the two sexes given that different selective pressures maintain coloration in males and females (LeBas, 2006; Kelly et al., 2012). Previous studies found individuals with brighter white plumage produced stronger immune responses (Bonato et al., 2009) and that tree swallows with lower wavelength hues and brighter blue plumage are of higher quality (Bitton et al., 2007,, 2008; Bitton and Dawson, 2008). Thus, we predicted that adults with brighter white breast plumage and/or with brighter dorsal plumage with bluer hues and greater blue chroma would have greater bactericidal capacity and produce young with stronger immune responses than adults with dull breast plumage or dorsal plumage with greener hues and greater green chroma. Third, we examined blood concentrations of the micronutrient selenium, to determine whether it related to immune function. We examined a range of selenium concentrations by sampling birds at colonies impacted by a spill that increased selenium availability and at nearby reference colonies with background selenium concentrations. We also examined interactive effects between blood selenium concentrations and coloration to address the possibility that more attractive, higher quality individuals or their offspring respond differently to selenium supplementation than less colorful individuals. We predicted that individuals with greater blood selenium concentrations would mount stronger immune responses than those with lower blood selenium concentrations.

Plumage coloration differed significantly between males and ASY females (Pillai's trace=0.410, F_6,117=13.6, P<0.001). The dorsal plumage of males had lower green chroma, greater blue and UV chroma, and lower hue than that of ASY females (Table 1, all F_1,124≥13.7, P<0.001). The brightness of the blue dorsal plumage did not differ between the sexes nor did the brightness of the white breast plumage (all F_1,124≤0.003, P≥0.96). Coloration did not differ significantly between the spill-impacted and reference colonies (Pillai's trace=0.070, F_6,117=2.8, P=0.20) nor was there a significant interaction between sex and colony (Pillai's trace=0.063, F_6,117=1.3, P=0.26).

We calculated principal components scores using color data from 603 males and 583 ASY females captured in 2011 and 2012. Principal components analysis produced two principal components that explained 89.3% of the variance in dorsal plumage coloration of adults (Table 2). PC1 explained 65.7% of the variance and received strong positive loadings for green chroma and hue, and strong negative loadings for blue and UV chroma. Thus, individuals with high PC1 scores have hues shifted toward higher wavelengths (greener) and greater green chroma while individuals with low PC1 scores have hues shifted toward lower wavelengths (bluer) and greater blue and UV chroma. PC2 explained an additional 23.6% of the variance in blue coloration and received a strong positive loading for blue brightness and a weaker positive loading for blue chroma. Individuals with greater PC2 scores have brighter blue coloration and a greater proportion of the curve in the blue portion of the spectrum. We used the PC scores to examine assortative pairing in 2012 of 197 pairs of adults. Across the entire study area, males and females paired assortatively by PC1 scores (r=0.19, P=0.01) but not by PC2 scores (r=0.06, P=0.40) or by white brightness (r=0.05, P=0.50).

We quantified the blood selenium concentrations of 29 males, 15 females and 77 nestlings (one per nest box) and used these three groups (age/sex) as categories in a two-way ANOVA to determine whether they differ in selenium concentrations or bactericidal capacity at exposed (greater selenium bioavailability) and reference (background selenium bioavailability) colonies. We found that blood selenium concentrations were significantly greater at exposed than at reference colonies (F_1,114=49.2, P<0.001), and differed between adults and nestlings (F_2,114=93.5, P<0.001), but the interaction between age/sex and colony was not significant (F_2,114=2.1, P=0.13). Tukey post hoc tests indicated that adults had significantly greater blood selenium concentrations than nestlings (Table 3, all P≤0.001). Exposed and reference colonies did not differ in bactericidal capacity (Table 3, F_1,114=0.001, P=0.98). However, nestlings had significantly lower bactericidal capacity than adults of both sexes (F_2,114=15.2, P<0.001) regardless of colony (interaction: F_2,126=2.4, P=0.10).

We found that bactericidal capacity of adult plasma was related to different aspects of plumage coloration in female and male tree swallows. For ASY females, the final model indicated that the brightness of white breast plumage was positively associated with bactericidal capacity (Table 4, Fig. 1, r²=0.31, d.f.=11, P=0.05). For males, the final regression model included PC1 score, blood selenium concentration and their interaction (Table 4, r²=0.42, d.f.=20, P=0.01). Males with lower PC1 scores (greater blue and UV chroma, hues shifted toward blue) had the greatest bactericidal capacity, but only if they had blood selenium concentrations around 10 μg g⁻¹ wet mass (Fig. 2). In contrast, males with the highest PC1 scores (greater green chroma and hue) with blood selenium concentrations in excess of 10 μg g⁻¹ wet mass had the lowest bactericidal capacity (Fig. 2). For both sexes, sample collection date, residual body mass, blood selenium concentration, other aspects of coloration and other interaction terms were unrelated to bactericidal capacity (Table 4).

We quantified the bactericidal capacity of a single nestling from each of 76 nest boxes, and captured 38 females and 56 males that were attending these boxes. Bactericidal capacity of nestling plasma was related to aspects of plumage coloration in ASY females and social males (the male attending the nest). For ASY females, the final regression model indicated that Julian sample date, female white brightness, nestling blood selenium concentration and the interaction between nestling selenium concentration and residual body mass were related to nestling bactericidal capacity (Table 5, r²=0.45, d.f.=31, P=0.002). We calculated the residuals from a regression of sample date against nestling bactericidal capacity to better focus on the other independent variables. We found that female white brightness had a positive relationship with nestling residual bactericidal capacity (Fig. 3A). Additionally, nestlings with low residual body mass and low blood selenium concentrations had the lowest residual bactericidal capacity, while the greatest bactericidal capacity was associated with nestlings with high blood selenium concentrations (Fig. 3B). For social males, the final regression model indicated that white brightness and the interaction between white brightness and nestling blood selenium concentration were related to nestling bactericidal capacity (Table 5, r²=0.30, d.f.=51, P=0.004). This result shows that nestlings with greater blood selenium concentrations and a brighter white social male had the greatest bactericidal capacity, whereas young of dull social males and/or with low blood selenium concentrations had low bactericidal capacity (Fig. 4). Nestling bactericidal capacity was not related to the bactericidal capacity of the social male (r=0.06, N=16, P=0.81) or female (r=0.06, N=12, P=0.85).

In contrast to bactericidal capacity, the PHA response of nestlings was unrelated to the plumage coloration of the female and social male, selenium concentrations in nestling blood, sample date, residual mass and the interaction terms (Table 6, final model social male, r²=−0.05, d.f.=17, P=0.73; final model female, r²=0.05, d.f.=15, P=0.38).

In a number of avian species as well as many other taxa, both sexes display ornaments that may serve as honest signals of quality to potential mates or rivals (Amundsen and Pärn, 2006; Kraaijeveld et al., 2007). In many cases, these ornaments may be similarly expressed by each sex, but may be the product of different sources of selection, sexual selection in males and social selection in females. Because of these different sources of selection, similar ornaments expressed in both sexes could relate to different aspects of quality, though only a few recent studies have found such differences (Kelly et al., 2012; Pickett et al., 2013). Here, we demonstrate that plumage coloration in female and social male tree swallows is similarly related to nestling immunity, while different aspects of coloration are related to bactericidal capacity in adults. Bactericidal capacity of nestlings and adult males was also related to their blood selenium concentrations, suggesting that the role of micronutrients other than carotenoids in mediating immunity in wild birds warrants further examination.

In ASY females, we found that the brightness of white breast plumage was positively associated with the bactericidal capacity of her plasma. White brightness may be associated with either the quantity or activity of complement enzymes or lysozyme that are primarily responsible for this aspect of the innate immune response (Matson et al., 2006). The results for females were similar to other recent studies that indicated the size, number or reflectance of white plumage patches is related to the strength of the immune response or parasite load in female birds (Hanssen et al., 2006; Zanollo et al., 2012; Blanco and Fargallo, 2013). Our results suggest that greater attention should be paid to the reflectance of large, white patches of plumage, which are infrequently examined in most avian studies despite being relatively common.

In male tree swallows, we detected an interactive relationship between male PC1 plumage scores, blood selenium concentration and plasma bactericidal capacity. The relationship between selenium and bactericidal capacity looked much like an optimal response curve, but the selenium concentration associated with peak bactericidal capacity differed between males with different plumage characteristics. Selenoproteins play a number of important roles in the immune system, such as increasing the proliferation and activity of immune cells, altering gene expression, and regulating the redox state of the organism (Maggini et al., 2007; Wintergerst et al., 2007; Huang et al., 2012). One possibility is that males with bluer dorsal plumage modulate these processes differently from greener males when provided with supplemental selenium. However, a stronger innate immune response may also impose a cost because its activation leads to the production of free radicals and oxidative stress (Hasselquist and Nilsson, 2012). Thus, a second possibility is that bluer males balance the trade-off between immunity and oxidative stress differently from greener males when provided with supplemental selenium. In collared flycatchers (Ficedula albicollis) and pied flycatchers (Ficedula hypoleuca), larger white plumage patches in males are associated with greater antioxidant capacity and lower oxidative stress than in males with smaller patches (Marko et al., 2011; Moreno et al., 2011). This suggests that structural plumage coloration can be associated with measures of oxidative stress and/or antioxidant capacity, though this remains to be addressed in tree swallows.

Our results indicate that the same aspect of adult immunity was related to different types of plumage coloration in male and female tree swallows. We suggest that this results from the different selective pressures acting on males and females. Male coloration is likely maintained by traditional sexual selection (Andersson, 1994). Associations between brighter and/or lower wavelength hues and extra-pair mating success (Bitton et al., 2007), higher survival or site fidelity (Bitton and Dawson, 2008), and a stronger innate immune response (this study) suggest dorsal coloration in male tree swallows is related to indirect, genetic benefits that are passed to offspring. In contrast, female coloration is likely maintained by social selection resulting from competition for breeding and non-breeding resources. One possibility is that brighter white females obtain nest sites located closer to prime foraging areas and are in better condition and able to mount a stronger immune response as a result. At a proximate level, it is possible that sex differences in the timing of molt (Hemborg, 1999; Dawson et al., 2009), seasonal hormone concentrations (Dawson et al., 2009; Small and Schoech, 2015) or immunity (Hasselquist, 2007) lead to different relationships between coloration and immunity in males and females.

A relationship between white intensity and resource quality or parental care could explain why nestling bactericidal capacity was positively associated with white brightness in both sexes. If white brightness is associated with better resource availability or food quality, then offspring may be in better condition and able to devote more resources to development of their immune system. In one model, we detected an interactive effect of nestling residual mass and blood selenium concentration, suggesting that parental provisioning could influence nestling bactericidal capacity. Additionally, blood selenium concentrations were positively associated with nestling bactericidal capacity and with a stronger relationship between social male breast brightness and nestling bactericidal capacity. This again suggests a relationship between nutritional conditions and offspring immunity. Indeed, the high rate of extra-pair mating in tree swallows (Whittingham and Dunn, 2001) makes it unlikely that the relationship between male coloration and offspring immunity is attributable to genetic inheritance alone. This relationship could also be the result of a maternal effect if brighter white females deposit more carotenoids (Saino et al., 2003; Biard et al., 2007), antibodies (Hasselquist and Nilsson, 2009), lysozyme (Board and Fuller, 1974) or testosterone (Clairardin et al., 2011) in eggs, which can all influence the immunocompetence of their offspring. While beyond the scope of the current study, future work should assign parentage to offspring and employ a cross-fostering design to better disentangle the contributions of genetics, micronutrient availability and other parental effects on offspring immunity.

Nestlings and bluer adult males showed stronger innate immune responses at elevated blood selenium concentrations, but this was not the case for females. Rapid growth in avian young (Costantini et al., 2006; Alonso-Alvarez et al., 2007) and activation of the immune system (von Schantz et al., 1999; Bertrand et al., 2006; Hasselquist and Nilsson, 2012) are both associated with increased production of reactive oxygen species and/or reduced antioxidant defenses. Given that the generation of reactive oxygen species is associated with reduced survival and fecundity in birds (Vleck et al., 2007; Bize et al., 2008), nestling tree swallows may not be able to afford the combined oxidative cost of rapid growth and development of a robust immune response. However, nestlings with greater blood selenium concentrations may be able to offset the combined oxidative costs of growth and immune system development and activation. Males and females had very similar blood concentrations of selenium, but it may be allocated to tissues differently in each sex. In mammals, there are sex differences in selenium metabolism and expression of selenoproteins in tissues (reviewed in Schomburg, 2012), and in poultry, there are sex differences in selenoenzyme activity in different tissues (Shaaban et al., 2008). Thus, female tree swallows, unlike males, may allocate their selenium to tissues or processes that do not lead to enhanced bactericidal capacity. The sample size for the female analysis was relatively small and further research with more robust sample sizes is necessary to determine whether there are sex differences in selenium usage in wild birds.

In conclusion, our study indicates that plumage coloration in adult tree swallows is related to their bactericidal capacity and to that of their nestlings, and that the relationships between plumage coloration and immunity can be influenced by selenium. We found that different aspects of coloration were related to bactericidal capacity in each sex, which may ultimately be due to the different selective pressures operating on each sex, sexual selection in males and social selection in females. Indeed, our results are similar to a recent study that found carotenoid-based ornamentation was related to immunity in female American goldfinches (Spinus tristis) but not in males (Kelly et al., 2012). While it is becoming increasingly clear that discrete white plumage patches can be related to individual condition (McGlothlin et al., 2007), quality (Hanssen et al., 2006,, 2009; Zanollo et al., 2012) or offspring quality (Remeš and Matysioková, 2013), very few studies have examined the reflectance of large white plumage areas (but see Bonato et al., 2009,, 2013). Bonato et al. (2013) found a positive relationship between the white brightness of the genetic father and humoral immunity in offspring, a result quite similar to the one we found for innate immunity. Together, this suggests that greater attention should be paid to the reflectance of large, white plumage areas because they may signal individual quality. We also found that blood selenium concentrations of adult males and nestlings were positively associated with one aspect of their innate immune response (current study; Beck et al., 2014). These results suggest that signal honesty may be altered by micronutrient availability, because bluer males only showed greater bactericidal capacity when they also had higher blood selenium concentrations. Studies in other wild and domesticated birds have detected positive relationships between tissue or dietary selenium concentrations and aspects of immunity, though these studies were focused on selenium supplementation or relatively high concentrations of selenium (Wayland et al., 2002; Brady et al., 2013; Surai and Fisinin, 2014). Our study suggests that more natural variation in selenium bioavailability, within or among populations, could influence the physiology of adults or young. Given the current interest in the regulation of oxidative stress and immunity in studies of behavioral ecology and life history (Hasselquist and Nilsson, 2012), and that selenium can have profound effects on these aspects of physiology (Huang et al., 2012), further examination of its role and that of other inorganic micronutrients in wild populations is warranted.

Tree swallows are secondary cavity nesters and readily settle in nest boxes. In the study population, most tree swallows are single brooded but are occasionally double brooded as has been found in other southern populations (Monroe et al., 2008). The average clutch size (N=585 nests) in this population is 5.16±0.03 (range 2–7 eggs) and the average brood size is 4.80±0.04 (range 1–6). Tree swallows are typically socially monogamous but genetically promiscuous with 80% of broods having mixed parentage in other populations (Whittingham and Dunn, 2001).

This experiment was conducted according to the Ornithological Council's guidelines to the use of wild birds in research and was approved by the Virginia Tech Institutional Animal Care and Use Committee (11-043-FIW and 12-072-FIW). Bird banding and blood sampling were approved under United States Geological Survey permit 23513 to D.M.H.

We studied a nest box breeding population of tree swallows in Roane (35.9°N, 84.5°W) and Loudon Counties (35.8°N, 84.3°W), TN, USA, during the 2011 and 2012 breeding seasons. All of the nest boxes were located within 70 m of the shore along the Emory, Clinch or Tennessee rivers and were located in nine colonies that contained 32–86 nest boxes. Boxes were spaced 15 m apart when in a single row, or 20 m apart with a staggered alignment when there were two or more rows. This study was conducted in an area where tree swallows were exposed to modestly elevated concentrations of selenium from a recently remediated coal fly ash spill and at nearby reference colonies where swallows were exposed to lower background concentrations. Fly ash contains high concentrations of many elements and exposure to fly ash can have severe negative effects on reproduction and physiology in a number of taxa (Rowe et al., 2002; Janz et al., 2010; Rowe, 2014). However, this spill occurred in December 2008 and extensive remediation efforts coupled with natural processes such as off-site transport greatly reduced element exposure well before our study began in 2011. Our previous research on swallows at this site indicated that reproductive success, nestling residual body mass and nestling stress physiology were not influenced by exposure to residual element contamination and that nestling bactericidal capacity and cutaneous immune responses did not differ among colonies (Beck et al., 2014,, 2015). The only element in nestling blood that was modestly elevated above background levels in both years of our research was selenium. Thus, the study site provided the unique opportunity to study the potential positive effects of modest dietary enrichment of selenium in a wild population of birds.

Selenium concentrations were quantified in whole blood of adults and nestlings using Inductively Coupled Mass Spectrometry at the Trace Element Analysis Core at Dartmouth College. A more detailed description of the study area and analytical methods used to quantify selenium concentrations can be found elsewhere (Beck et al., 2014).

We checked nest boxes every 4 days beginning in late March and every 3 days beginning in early June for signs of nesting activity so that we could accurately record clutch initiation and hatching dates, and capture parents at appropriate times. We captured adults using mist nets or trapped them in the nest box while they were incubating or provisioning young. Adults were sexed by the presence of a brood patch (females) or cloacal protuberance (males) or, if these were absent, by measuring the wing length, with wings shorter than 113 mm being indicative of females and wings longer than 122 mm being indicative of males (Stutchbury and Robertson, 1987a). Females were also aged as SY or ASY based on plumage coloration as described previously. Adults were banded with a combination of one metal band and one colored leg band placed on opposite legs to help us distinguish individuals breeding in neighboring boxes and between males and females from a distance. From each adult, we obtained a small blood sample (≤120 μl) and measured left and right tarsus length (each was measured twice), wing chord and body mass. When nestlings were 13 days old, we obtained morphological measurements (as described for adults), banded them with a metal leg band, obtained a small blood sample (≤120 μl), and performed the immune challenges.

To quantify plumage coloration of adults, we collected 8–10 feathers from each adult from standardized locations on the back, rump and breast. Feathers were taped to pieces of black construction paper with low reflectance from 320 to 700 nm and in a manner similar to the way they laid on the bird. We quantified plumage coloration using a JAZ reflectance spectrophotometer (Ocean Optics, Dunedin, FL, USA) with a 200 μm fiber optic probe. The probe tip was held in a sheath that excluded ambient light and kept the probe tip 5 mm from the sample surface. The spectrometer lamp was allowed to warm up for 10 min prior to making the first measurement. Spectra were produced relative to a white standard (Labsphere, North Sutton, NH, USA) and the spectrometer calibrated to the white standard between samples. All measurements were made perpendicular to the sample surface and we made four measurements of each region (back, rump and breast). We used the program CLR version 1.05 (Montgomerie, 2008) to smooth reflectance spectra and calculate color variables. We calculated brightness of the breast and dorsal plumage coloration as average percentage reflectance from 320 to 700 nm. For dorsal plumage, we calculated hue as the wavelength at which peak reflectance occurred. We also calculated three measures of chroma for dorsal plumage: UV, blue and green. We calculated chroma as the total reflectance in the UV (320–400 nm), blue (400–510 nm) and green (510–605 nm) range divided by the total reflectance from 320 to 700 nm.

We examined one aspect of the innate immune response of nestling and adult tree swallows in June and July 2012 by evaluating the bactericidal capacity of plasma (following Liebl and Martin, 2009). This bacteria killing assay (BKA) reflects the ability of complement enzymes and lysozyme to destroy cell walls and lyse cell membranes of bacteria (reviewed in Matson et al., 2006). Blood samples were obtained from adults trapped in their nest box while provisioning young or from nestlings 13 days post-hatch. The area around the puncture site was cleansed with 70% ethanol prior to blood collection from the brachial vein. We collected 120 μl of blood in heparinized capillary tubes: 60 μl was used to assess bacterial killing capacity and 60 μl was reserved to quantify selenium concentrations. Samples for the bactericidal assay were centrifuged at 10,000 rpm for 5 min within 6 h of collection and the plasma fraction placed in a sterile 0.5 ml tube. The bactericidal assay was run with fresh plasma on the day the sample was collected. Samples were run in triplicate by aliquoting 3.5 μl of plasma into sterile tubes and diluted with 31.5 μl of sterile PBS (1:10 dilution). We added 12.5 μl of 10⁵ bacteria ml⁻¹ Escherichia coli (ATCC 8739, Epower microorganisms, Microbiologics^®, St Cloud, MN, USA) solution to each tube and vortexed each sample. Samples were incubated at 37°C for 30 min and 250 μl of tryptic soy broth (TSB, Sigma-Aldrich, St Louis, MO, USA) was added to each tube, and samples were incubated for an additional 12 h at 37°C. Positive controls were created in triplicate by adding 12.5 μl of 10⁵ bacteria ml⁻¹ to 250 μl of TSB and we created duplicate blanks by combining 50 μl of PBS with 250 μl of TSB. We checked that bacteria were not introduced during bleeding or sample processing by making two additional controls without adding bacteria stock that contained 3.5 μl of plasma, 250 μl of TSB and 50 μl of PBS.

Following incubation, samples were vortexed and a Nanodrop Spectrophotometer (ND-2000, Thermo Scientific, Pittsburgh, PA, USA) was used to measure the absorbance of each sample at OD₃₀₀. The absorbance of each sample and the positive controls were each averaged and used to calculate the proportion of bacteria killed as 1−(sample absorbance/positive control absorbance). The Nanodrop arm was cleansed between each sample with 70% ethanol and the entire work area was cleansed with ethanol before and after each work day.

We also quantified the response of nestlings to PHA (Sigma-Aldrich) by injecting the patagium (wing web) of nestlings with 0.15 mg of PHA dissolved in 30 μl phosphate-buffered saline (after Smits et al., 1999; Hawley et al., 2009). The injection of PHA leads to localized swelling due to the influx and proliferation of leukocytes and T-cells at the injection site (Martin et al., 2006) and a build-up of free radicals that are produced during phagocytosis by leukocytes (Peretz, 1989). Feathers were first cleared from the wing web and the area sterilized using 70% ethanol. One individual held the nestling with its right wing extended in a standardized position while a second made all measurements and performed the injections. Prior to and approximately 24 h following injection, the thickness of the wing web was quantified to the nearest 0.01 mm using a micrometer. The micrometer dial was not visible to the measurer while making the measurement. We made five pre- and five post-injection measurements and discarded the lowest and highest values in each set and used the remaining three to produce average pre- and post-injection widths for each individual. We divided the difference between the post- and pre-injection measurements by the pre-injection measurement to adjust for pre-injection differences in wing web width and used this as our measure of the cell-mediated immune response. The mean (±s.d.) time between injection and measurement was 24.8±2.3 h and there was no correlation between the time between injection and measurement and the PHA response or the average post-injection swelling (both r≤0.25, d.f.=31, P≥0.18).

We used Kolmogorov–Smirnov tests and normality plots to determine whether variables met the assumptions of parametric tests. We log transformed continuous variables (plumage hue and blood selenium concentration) or used arcsine square root transformations for proportion data (plumage chroma and brightness, immune responses). Using data collected in 2011 and 2012, male and ASY female plumage variables were first compared using a MANCOVA to determine whether they were sexually dichromatic. For all males and ASY females, the dorsal color variables for blue, structural color were highly correlated and we used principal components analysis (PCA) to reduce the dimensionality of the data set while producing variables that were not linearly related. We used transformed color variables from both years in the PCA and used Varimax factor rotation to simplify the factor loadings. For adults captured in both years, we selected a single year of data to include in the PCA. For white breast plumage, we calculated its average brightness and included it in the analyses as well. We did not perform any analyses with SY females because we had a much smaller sample size for this age class and their plumage coloration is melanin based (Bentz and Siefferman, 2013) and differs dramatically from the structural coloration of ASY females.

We used correlations to determine whether males and ASY females paired assortatively by PC scores rather than the individual color variables to minimize the number of statistical comparisons made. Colonies were classified as exposed to modestly elevated selenium (sites previously impacted by the spill) or as reference colonies with background exposure to selenium (Beck et al., 2014). Swallows were categorized as adult males, ASY females or nestlings, and we then used a two-way ANOVA to compare blood selenium concentrations and bactericidal capacity between impacted and reference colonies and between the sexes and nestlings. We used residual body mass (residuals from a regression of mass on tarsus length) as a measure of body condition. While the use of residuals as a measure of body condition is controversial (Green, 2001), studies have shown that residuals do correlate well with lipid reserves (Ardia, 2005; Schulte-Hostedde et al., 2005), and residual body mass is related to the immune response in tree swallows (Palacios et al., 2012). Using data pooled from all of the colonies, we used linear regressions and backward elimination of non-significant terms to examine the effects of plumage coloration, sample date, residual body mass and selenium concentration on the immune responses of adults and nestlings. We allowed terms to remain in the model as long as P≤0.10 but only considered their contribution to be statistically significant when P≤0.05. We used the percentage bacteria killed and wing web swelling as dependent variables. We evaluated the response to PHA injection and bactericidal capacity in a single nestling from each box to avoid issues with pseudoreplication. We ran separate models to examine relationships for males and females to determine how adult phenotype related to their bactericidal capacity. We also ran separate models for male and female parents when examining nestling immune responses for two reasons. First, males and females pair assortatively by coloration (see Results), and including their color variables simultaneously in the model introduces issues with multicollinearity. Second, we caught only a single parent at some boxes and limiting our analysis to nests where both parents were captured would significantly reduce sample sizes. Sample sizes varied slightly among tests due to incomplete sampling of individuals (e.g. failure to collect feathers for spectroscopy). All statistical tests were two-tailed with α=0.05. All statistical analyses were performed using PASW 18 (SPSS, 2009).

We thank J. Hallagan, M. Hepp, D. Sedgewick, M. Hepner, E. Burton, J. Morris, D. Blood, J. Botero, A. Garica, B. Hopkins, A. Love, S. Munoz, B. Nickely, E. Smith and L. Trapp for assistance in the field, and J. Favara, S. Walls, W. James, N. Carriker and the USDA APHIS Knoxville for logistical support. We thank B. Jackson and the Trace Element Analysis Core at Dartmouth College for quantifying selenium concentrations.