High brood patch temperature of less colourful, less pheomelanic female Barn Swallows Hirundo rustica

Transcription

1 Ibis (2016), 158, High brood patch temperature of less colourful, less pheomelanic female Barn Swallows Hirundo rustica MASARU HASEGAWA, 1, * EMI ARAI, 2 SHOSUKE ITO 3 & KAZUMASA WAKAMATSU 3 1 Department of Evolutionary Studies of Biosystems, Sokendai (The Graduate University for Advanced Studies), Hayama-machi, Kanagawa , Japan 2 Division of Ecology and Evolutionary Biology, Graduate School of Life Sciences, Tohoku University, Sendai, Miyagi, Japan 3 Department of Chemistry, Fujita Health University School of Health Sciences, Toyoake, Aichi , Japan In birds, even a minor difference in egg temperature (1 1.5 C) has been shown to affect the fitness of offspring by changing hatching success, incubation period and nestling quality. Female, but not male, passerines develop brood patches. Thus if there are traits, such as plumage ornamentation, that indicate optimal egg temperature, males should pair with females that exhibit those traits. However, no study has yet investigated the relationship between female brood patch temperature, which would directly affect egg temperature, and female plumage ornamentation. In this study, we examined the surface temperature of female brood patches during nocturnal incubation and examined its relationship with female plumage ornaments in Asian Barn Swallows Hirundo rustica gutturalis. After controlling for ambient air temperature, brood patch temperature was negatively associated with colour saturation of the female throat patch. No other female ornaments, such as tail-length, white tail spots or throat patch size, predicted brood patch temperature. When oral (mouth) temperature was statistically controlled, females with less colourful throats and longer tails showed higher brood patch temperature, indicating that these females had hotter brood patches in relation to the temperature of other body parts. Furthermore, we found a negative relationship between pheomelanin pigmentation and brood patch temperature after controlling for ambient air temperature or oral temperature. To our knowledge, this is the first study to show that female ornaments can predict the absolute/relative thermal investment in brood patches. This relationship, together with other aspects of female quality, may affect male mate preference and female ornamentation. Keywords: female ornaments, Hirundo rustica gutturalis, incubation, pheomelanic plumage coloration. In many passerine birds, females develop brood (incubation) patches through the loss of ventral feathers and increased skin vascularization to allow effective heat transfer from the adult to the egg clutch (Lea & Klandorf 2002). High egg temperature has recently been shown not only to improve hatching success and shorten the incubation period, but also positively to impact nestling life-history traits that persist thereafter, affecting survival and reproduction of juveniles and adults (e.g. *Corresponding author. perorobomusadiobe@gmail.com tarsus-length, Nord & Nilsson 2011; post-fledging survival, Bernstein & Bech 2016; reproduction, Hepp & Kennamer 2012, DuRant et al. 2013, Hepp et al. 2015). Even a small difference in egg temperature (1 1.5 C) can affect a wide variety of phenotypic or life-history traits (DuRant et al. 2013). Because males often lack a brood patch in passerines (Lea & Klandorf 2002), they do not perfectly function as a substitute for female incubation in many species (e.g. Kleindorfer et al. 1995, Voss et al. 2008, but see Auer et al for exceptions). Thus, for males, pairing with females that can attain a high egg temperature will

2 Female ornaments and brood patch temperature 809 impact hatching success, incubation period, offspring quality and ultimately male reproductive success. Because incubation takes place long after pair formation (the incubation period follows the mating, nesting and laying periods; c. 20 days in the Barn Swallow Hirundo rustica, Møller 1994), males are unable to use direct cues about incubation at the time of pairing. Males may focus on female plumage as an indicator of incubation temperature if it impacts reproductive success, as has been shown for other parental traits (Amundsen & P arn 2006, Griffith & Pryke 2006). From an evolutionary perspective, it should be beneficial for males to pair with females that have such traits, which would drive male mate choice. Male mate choice would in turn drive the evolution of female plumage via sexual selection (i.e. via positive coevolution between male mate preference and female traits; Andersson 1994, Kokko et al. 2002, Griffith & Pryke 2006). Some recent studies have focused on female incubation behaviour (e.g. on-bouts, off-bouts) that affects egg temperature in relation to female plumage ornamentation (e.g. Matysiokova & Remes 2010). However, female brood patch temperature, which would also affect egg temperature, has received little attention (Deeming 2008). This may be because an efficient and non-invasive device to measure temperature was not available until relatively recently (e.g. infrared thermometer, or thermography, Deeming & Du Feu 2008, McCarffery 2013). Female brood patch temperature is a major determinant of egg temperature, particularly at night, because females of diurnal species spend the entire night on the nest (i.e. without foraging trips; e.g. Zebra Finch Taeniopygia guttata, Nord et al. 2010). Due to the scarcity of studies focusing on this topic, it remains unclear whether (and which) female plumage characteristics predict brood patch temperature. The Barn Swallow is a classic model species of sexual selection in monogamous birds. Males have several sexual ornaments, including long tails (Møller 1988), large white spots on the tail (Kose & Møller 1999, Kose et al. 1999) and plumage coloration (including throat patch size and colour; Ninni 2003, Safran & McGraw 2004, Hasegawa & Arai 2013, Scordato & Safran 2014). These characteristics signal several male traits to potential mates, such as male age (Møller & de Lope 1999), viability (Møller 1991), territory quality (Hasegawa et al. 2014, Wilkins et al. 2016) and hormonal states (Turner 2006, Safran et al. 2008). Likewise, corresponding traits can be found in females. Traits in females are somewhat less conspicuous, possibly due to a sex difference in selection pressure, but these traits may also be the targets of intersexual selection (tail-length, Møller 1993, but see Cuervo et al. 1996; plumage coloration, Safran & McGraw 2004). These female traits may signal quality to conspecifics (e.g. hormonal state, Vitousek et al. 2013), although males and females signal different qualities with different ornaments, as reported for the Barn Owl Tyto alba (Roulin et al. 2001a,b). In Barn Swallows, females have a well-vascularized brood patch and contribute more than 90% of diurnal incubation attentiveness (94% in Hirundo rustica gutturalis, Hasegawa et al. 2012a; 91% in Hirundo rustica erythrogaster, Smith & Montgomerie 1992, reviewed in Turner 2006). Males thus incubate for a limited time in some subspecies (e.g. 6% in H. r. gutturalis, 9%inH. r. erythrogaster) but because they do not have brood patches, males cannot compensate for female incubation (Voss et al. 2008). Female incubation attentiveness increases with her mate s ornamentation but is not associated with male incubation attentiveness, their own body condition or ornamentation (Hasegawa et al. 2012a). Sex differences in incubation attentiveness are more prominent at night, because females almost certainly incubate all night even in populations where males participate in diurnal incubation (Smith & Montgomerie 1992). Thus, the nocturnal incubation of Barn Swallows is a suitable system in which to study brood patch temperature in relation to female ornamentation. In this study, we examined female brood patch temperature at night in relation to female ornamentation in Asian Barn Swallows H. r. gutturalis, in which females perform most of the incubation task (94% diurnal incubation). In contrast to subspecies with a colourful ventral plumage (e.g. H. r. erythrogaster), this subspecies has a whitish ventral plumage and a large pheomelanic throat patch, twice as large as the nominate subspecies Hirundo rustica rustica (Turner 2006, Hasegawa et al. 2010). As in other subspecies, throat coloration is an important sexual ornament in males (Hasegawa et al. 2010, Hasegawa 2011, Scordato & Safran 2014) and perhaps also in females (M. Hasegawa unpubl. data). Due to the large

3 810 M. Hasegawa et al. brood patch of this species, we can measure the surface temperature of a brood patch using an infrared thermometer (Fig. 1). Likewise, the wide gape of this aerial insectivore allows us to measure oral temperature to assess whether brood patch temperature is controlled independently of the temperature of other body parts. Among four measures of female ornaments (tail-length, the size of the white tail spots, throat patch size and throat coloration, Hasegawa et al. 2010), we predicted that throat coloration is related to brood patch temperature because its underlying structural basis, comprising melanin pigments (eumelanin and pheomelanin), is genetically linked to thermoregulation (Fan et al. 2005, Ducrest et al. 2008, Koskenpato et al. 2016). Melanin-based coloration has been demonstrated to correlate with body temperature in some empirical studies (eu- and pheomelanic traits of the Japanese Quail Coturnix japonica, Minvielle et al. 2007, 2010; eumelanin but not pheomelanic traits in the Barn Owl, Roulin & Randin 2015, Dreiss et al. 2016). For this reason, we studied brood patch temperature in relation to pheomelanin pigmentation, to distinguish its effect from the confounding effects of post-moulting changes in feather coloration (e.g. stain: Montgomerie 2006, Hasegawa et al. 2008). Because Barn Swallows nest under the eaves of houses in our study population, as in other Japanese populations (e.g. Hasegawa et al. Figure 1. Illustration of the method used to measure the surface temperature of a female brood patch using an infrared thermometer. An optic-aiming system allowed us immediately to find the correct distance for measurements, at which the temperature was measured. 2014, Hasegawa & Arai 2015a), female brood patch temperature may be particularly important in this population compared with colonial populations (i.e. this population could experience greater temperature cooling relative to those nesting in warm livestock stalls, Turner 2006, p. 123). We discuss the evolutionary implications of our findings. METHODS Study site This study was conducted in a residential area of Hayama-machi, Kanagawa Prefecture, Japan ( N, E, altitude 43 m), from March to May In this area, Barn Swallows nest under the eaves of houses along streets (Tajima & Nakamura 2003, Hasegawa et al. 2010). We inspected nests every other day to record the laying date (expressed as the date of clutch initiation) and clutch size. To determine the hatching date, nests were inspected daily from approximately 10 days after the laying date (Hasegawa et al. 2010). We recorded hatching success or failure (note the high nest predation rate in outdoor populations: Hasegawa et al. 2012b, see also Fujita 1993) and hatching date. Sweep-nets were used to capture adult males and females soon after the start of the incubation period (mean = 3 days, range = 0 9; one female was captured during the incubation period, but her incubation date was unknown). Birds were captured from 23:00 to 03:00 h and were marked with an individual combination of two or three coloured rings (Arai et al. 2009). The sex of each individual was determined on the basis of the presence or absence of a brood patch (Turner 2006). At capture, we measured body mass (to the nearest 0.1 g), wing-length, tarsus-length, tail-length, size of the white tail spots (the sum of the lengths of the white spots of the two outermost right tail feathers; all measurements to 0.01 mm precision) and throat patch depth (see below). We also collected approximately throat feathers to measure throat coloration and to quantify the pheomelanin content. Detailed information on the measurements can be found in previous publications (Hasegawa et al. 2010, Hasegawa & Arai 2013). Throat patch depth was defined as the depth of the red throat patch (Hasegawa & Arai 2013,

4 Female ornaments and brood patch temperature b, Hasegawa et al. 2014). We placed a transparent plastic sheet on the throat region, ensuring that the feathers lay flat in their natural position. We traced the size of the patch on the sheet with a marker pen (cf. Lendvai et al. 2004) and scanned the sheet to measure the depth of the patch (to 0.1 mm) using SCION IMAGE software (Scion, Frederick, MD, USA). For each bird, the throat patch was traced twice and the mean of the two measurements was used. Throat patch depth yielded high repeatability during the study period: repeatability = 0.84 (using function lmer in R package lme4, Nakagawa & Schielzeth 2010); there was no significant sex difference in throat patch depth in the study sample (n male = 52, n female = 37, v 2 = 0.80, P = 0.37). Note that we included individuals whose oral and brood patch temperatures could not be measured to estimate repeatability (see Supporting Information Appendix S1 for details). Once in the laboratory, we placed five feathers on a piece of white paper such that the perimeters of the feathers overlapped. The feather samples were scanned at 800 dpi resolution (GT X970; Epson, Japan), and the images imported into the Adobe PHOTOSHOP ELEMENTS 10.0 software (Adobe Systems, San Jose, CA, USA). We measured mean RGB values in a square of 30 pixels near the distal end of the feather sample. These mean RGB values were converted to hue saturation brightness values using the algorithm of Foley and van Dam (1984). Because the throat feathers did not reflect in the UV region, this method captured most of the variation in throat coloration (Safran & McGraw 2004, E. Arai unpubl. data). We used saturation as our index of plumage coloration (e.g. Hasegawa et al. 2010) because saturation does not require correction for plumage colour fading, unlike the other two measures of coloration, which fade linearly with time (Hasegawa et al. 2008). The saturation value (unitless; range: 0 255) was negatively associated with pheomelanin (reddish) and eumelanin (black) pigmentation; thus, lesssaturated throat feathers contained more pheomelanin and eumelanin pigments, producing redder and darker feathers (Arai et al. 2015, see also Hasegawa et al. 2008, 2010). Although the range of saturation is sometimes expressed as a percentage (i.e %) in some studies of plumage coloration (e.g. Safran et al. 2005), we used the raw data range in our study (Hasegawa et al. 2016) but also present percentage values to enable comparison with other studies (our estimates can be converted to percentages by dividing by 2.55; Yabusaki et al. 2014). After statistically controlling for sex as a covariate (males: , n = 44 on the scale, on 0 100% scale; females: , n = 32 on the scale, on the 0 100% scale; v 2 = 20.47, P < ; see Appendix S1 for details of sample sizes), saturation values yielded modest repeatability in the study period (repeatability = 0.63). To achieve a reliable saturation value from moderate repeatability, we averaged the saturation value from the two independent samples of five feathers whenever possible (Falconer & MacKay 1996). Averaged saturation for the two samples was used for 24 of 27 females for which complete plumage measurements were available. For the remaining three females, we had only one feather sample from which the saturation value was measured. Measuring temperature Surface temperatures of the brood patch and inside the mouth were measured using an infrared thermometer, a non-intrusive and effective method (2 3 s per measurement; Deeming & Du Feu 2008). At capture, we measured the surface temperature of the brood patch using a handheld infrared thermometer (Thermofocus Pro, Technimed, SRI Corp., Varese, Italy; accuracy 0.2 C at C and 0.3 C otherwise) at a distance of c. 3 cm, while the breast feathers were fixed with plastic sticks (Fig. 1). Similarly, we measured the surface temperature inside the mouth (i.e. oral temperature). To accurately measure surface temperature, we did not touch the brood patch or inside the mouth; in the case of the oral temperature, we opened and held the mouth open by solely touching the bill tips. Hourly mean air temperature at the study site was obtained from the Japan Meteorological Agency (2015). We recorded hourly mean air temperature at the nearest sampling site ( N, E, altitude 5 m; 10 km from the centre of our study site) as representative of air temperature at the study site. When we recorded hourly mean air temperature at the second nearest sampling site ( N, E, altitude 42 m; 13 km from the study population; 22 km from the nearest sampling site noted above), the two measures of hourly mean temperature were highly correlated

5 812 M. Hasegawa et al. (r = 0.94, n = 27, P < ) and qualitatively similar results were recovered when we used this temperature instead of those at the nearest sampling site. This indicates that hourly mean air temperature at the nearest sampling site is likely to be a reliable measure of air temperature around the study site (although micro-climate differentially affects air temperature at each nest within the population; see Discussion). Pigment analysis The concentration of pheomelanin was determined by HPLC according to the method described previously (Ito et al. 2011). The HPLC system consisted of a JASCO 880-PU pump (JASCO Co., Tokyo, Japan), a Shiseido C18 column (Capcell Pak MG; mm; 5 lm particle size, Shiseido, Tokyo, Japan) and a JASCO UV-970 UV/VIS detector at 269 nm. The mobile phase was 0.1 mol/l potassium phosphate buffer (ph 2.1) : methanol = 99 : 1 (v/v). Analyses were performed at a flow rate of 0.7 ml/min and at 45 C. For this analysis, we used sub-samples of females in which 10 feathers could be obtained for colour measurements (n females = 26 for its relationship with coloration; n females = 19 for oral and brood patch temperatures). First, the length of the pheomelanic parts of each feather was measured to the nearest 0.01 mm. After removing the eumelanic proximate part from each feather sample, the remaining pheomelanic parts were put into a 10-mL screw (Te)-capped conical glass test tube, to which 40 ll water, 150 ll 1 M K 2 CO 3 and 10 ll 30% H 2 O 2 (final concentration: 1.5%) were added. The mixture was mixed vigorously at 25 C (range, 24 26) for 20 h. The residual H 2 O 2 was decomposed by adding 20 ll 10% Na 2 SO 3 and the mixture was then acidified with 56 ll 6 M HCl. After vortexing, the reaction mixture was transferred to a 1.5-mL Eppendorf tube and centrifuged at g for 1 min, and an aliquot (80 ll) of the supernatant was directly injected into the HPLC system. Given the amount of dilution used for the current study, pheomelanin pigmentation (ng/mm) was calculated by the thiazole-2,4,5-tricarboxylic acid (TTCA) value divided by the length of the pheomelanic feather part. TTCA is a degradation product from pheomelanin and serves as a specific marker of pheomelanin (Ito et al. 2011). After controlling for sex difference in pigmentation (difference male female se: , F = 14.03, P < 0.001), the throat saturation value was significantly negatively related to pheomelanin pigmentation (coefficient se = , F = 7.89, P < 0.01; overall: n male = 46, n female = 26, F = 21.41, P < ; see Appendix S1). Data analysis We compared female brood patch temperature with the oral temperature of the same individuals using a paired t-test. Because surface temperature should be affected by ambient air temperature, we first studied the relationship between brood patch and oral temperature with hourly mean air temperature using Pearson s product-moment correlation coefficient. Thereafter, to control for the influence of ambient air temperature, we used residual temperature controlling for hourly mean air temperature using univariate statistics. In the multivariate models, we included hourly mean air temperature as a covariate. We used non-parametric statistics when investigating residual temperature in relation to laying date and incubation date at capture (Shapiro Wilk test: W < 0.91, P < 0.02). A linear model was used to investigate the relationship between predictor variables and brood patch temperature. Because of the small sample size we used the robust linear model (lmrob in the robustbase package), which is better suited to such data than are ordinary least-squared models. In addition to four female ornaments saturation value of throat patch, throat patch depth, tail length and the size of white tail spots we included hourly mean air temperature or oral temperature as a covariate to account for environmental variability. An alternative approach using residual brood patch temperature controlling for hourly mean air temperature or oral temperature as a response variable yielded qualitatively similar results. As this information is also meaningful, we included it in Supporting Information Tables S1 and S2. Because we captured only four females that were ringed during breeding in previous seasons, perhaps due to high breeding dispersal (Arai et al. 2009), we could not control for female age in the current study (although there was no significant difference in residual brood patch temperature, residual oral temperature or residual brood patch temperature controlling for oral temperature between females with and without previous

6 Female ornaments and brood patch temperature 813 breeding experience; n with = 4, n without = 23, t < 1.88, P > 0.10). All plumage ornaments were standardized to a mean of zero and variance of one (Lande & Arnold 1983). When we tested for multicollinearity among variables using the variance inflation factor (VIF), we found low VIF values (maximum = 1.2), indicating that multicollinearity should have only a minor effect on the relationships (Graham 2003). The significance of the variables was based on Wald statistics using the full model with and without the focal variable in question (R function anova.lmrob with the test = Wald option: Mundry & Nunn 2009, Forstmeier & Schielzeth 2011). We have a priori information that pheomelanin coloration should be linked to the response variable, which is the least problematic in the issue of multiple hypotheses testing (Forstmeier & Schielzeth 2011). Nevertheless, we applied Bonferroni correction to address the problem. In the current case, apart from the covariates that needed to be controlled (i.e. surrounding temperature; Forstmeier & Schielzeth 2011), we had four variables and thus we used a = 0.05/4 = Finally, from a subset of the data, we assessed the relationship between pheomelanin pigmentation and brood patch temperature. In this analysis, which had a reduced sample size (n = 19), we used residual brood patch temperature based on the whole dataset rather than including air temperature or oral temperature as a covariate to control accurately the correlation between these temperatures. All data analyses were performed using the R statistical package (ver ; R Development Core Team, 2011). RESULTS Brood patch temperature and environmental factors The mean ( sd) brood patch temperature was C, which was significantly higher than the oral temperature measured inside the mouth of the same individuals ( C, n = 30; paired t-test; t = 19.21, P < ; Fig. 2). Both variables were positively correlated with hourly mean air temperature of the study population at the point of capture (brood patch temperature = 0.30*air temperature ; n = 30, r = 0.44, P = 0.01; oral temperature = 0.56*air temperature ; n = 30, r = 0.66, Figure 2. Plot of female brood patch temperature vs. the same individual s oral temperature. A simple regression line is shown (thick line: brood patch temperature = 0.46*oral temperature ; F 1,28 = 17.72, P < 0.001), together with the equivalence line (dashed line). P < 0.001). Thus, we used residual brood patch or oral temperature to control for hourly mean air temperature in univariate statistics or included hourly mean air temperature as a covariate in the multivariate models. Oral temperature and brood patch temperature were positively correlated, with and without controlling for hourly mean air temperature (with: n = 30, r = 0.50, P < 0.01; without: n = 30, r = 0.62, P < 0.001; Fig. 2). Neither residual brood patch temperature nor residual oral temperature was strongly correlated with laying date (mean date = 24 April, range = 17 April 13 May; brood patch: n = 29, Spearman s correlation coefficient, r s = 0.17, P = 0.38; oral: r s = 0.34, n = 29, P = 0.07; residual brood patch temperature controlling for oral temperature: r s = 0.07, n = 29, P = 0.71). Likewise, there was no significant correlation between incubation date at capture and residual brood patch temperature (r s = 0.11, n = 29, P = 0.58), residual oral temperature (r s = 0.21, n = 29, P = 0.27), or residual brood patch temperature controlling for oral temperature (r s = 0.12, n = 29, P = 0.52). Residual brood patch temperature and residual oral temperature were not significantly different between females with a clutch size of five or a clutch size of six (brood patch: t = 0.85, n = 30, P = 0.39; oral: t = 0.91, n = 30, P = 0.36; residual brood patch temperature controlling for oral temperature: t = 0.77, n = 30, P = 0.44; no

7 814 M. Hasegawa et al. other clutch sizes were recorded among the females studied). Brood patch temperature and female plumage traits Of the 30 females studied, complete plumage measurements were available for 27; of the remaining three, two had broken tail tips, and the white tail spots were missing in the third. Regarding the relationship between brood patch temperature and female plumage ornamentation while controlling for hourly mean air temperature as a covariate, throat saturation value, but not other values, significantly predicted brood patch temperature, even after Bonferroni correction (a = 0.05/ 4 = ; Table 1a). Females with more saturated throat feathers (i.e. less colourful females) showed higher brood patch temperature (Fig. 3). Qualitatively similar results were obtained when we used residual brood patch temperature controlling for hourly mean air temperature as a response variable (i.e. saturation value, but not other values, significantly predicted response variables; Table S1). Wing-length, tarsus-length and body mass were not significant when included in the model alone or in combination (data not shown). Instead of using hourly mean air temperature as a covariate, we could use oral temperature as a covariate to study thermal output at the brood patch compared with oral temperature. When oral temperature was statistically controlled, throat saturation and tail-length significantly predicted brood patch temperature, even after Bonferroni correction (a = 0.05/4 = ; Table 1b). Females with more saturated throat feathers (less colourful females) and longer tails showed higher brood patch temperatures relative to their oral temperature (Fig. 4). The size of white tail spots and throat patch size did not significantly predict brood patch temperature in this model (Table 1b). Using residual brood patch temperature controlling for oral temperature yielded qualitatively similar results (saturation value and tail-length, but not other values, predicted the response variables; Table S1). Winglength, tarsus-length and body mass were not significant when included in the model individually or in combination (data not shown). Analysis of oral temperature controlling for hourly mean air temperature showed no significant predictors (Table 2). Again, qualitatively similar Table 1. Multivariate robust linear models predicting female brood patch temperature after controlling for (a) hourly mean air temperature and (b) oral temperature (n = 27). Coefficient se Wald v 2 P (a) Controlling for hourly mean air temperature Air temperature Saturation value Throat patch depth Tail-length Size of white tail spots (b) Controlling for oral temperature Oral temperature < Saturation value Throat patch depth Tail-length < Size of white tail spots Note that plumage ornaments (saturation value, throat patch size, tail-length and the size of the white tail spots) were standardized to a mean of zero and variance of one before analysis. Significant test results (P < 0.05) are indicated in bold. The mean and sd data of saturation value, throat patch depth, tail-length and size of the white tail spots (n = 27) were as follows: saturation value: on the scale and on the scale (see Methods for details); throat patch depth: ; tail-length: ; size of the white tail spots: Figure 3. Plot of residual female brood patch temperature after controlling for covariates in relation to the throat saturation value (see Table 1a). A larger saturation value indicates a less colourful throat patch. A simple regression line is shown (residual brood patch temperature = 0.35*standardized saturation value 0.05; F 1,25 = 4.69, P = 0.04). The mean and sd data of saturation value (n = 27) were on the scale and on the scale see Methods for details.

8 Female ornaments and brood patch temperature 815 (a) (b) Figure 4. Plot of residual female brood patch temperature after controlling for covariates in relation to the throat saturation value (a) and tail-length (b; Table 1b). A larger saturation value indicates a less colourful throat patch. Simple regression lines are shown: (a) residual brood patch temperature = 0.34*standardized saturation value 0.03; F 1,25 = 6.47, P = 0.02; (b) residual brood patch temperature = 0.32*standardized tail length 0.03; F 1,25 = 5.78, P = The mean and sd data of saturation value and tail-length (n = 27) were as follows: saturation value: on the scale and on the scale (see Methods), tail-length: results were observed when we used residual oral temperature (neither variable was significant; Table S2). Wing-length, tarsus-length and body mass were not significant when included in the model alone or in combination (data not shown). Brood patch temperature and female pigmentation We used a subsample (n = 19) of the data above to study the direct relationship between brood patch temperature and female pheomelanin Table 2. Multivariate robust linear model predicting female oral temperature after controlling for hourly mean air temperature (n = 27). Coefficient se Wald v 2 P Air temperature < Saturation value < Throat patch depth Tail-length Size of white tail spots Note that plumage ornaments (saturation value, throat patch depth, tail-length and the size of the white tail spots) were standardized to a mean of zero and variance of one before analysis. Significant test results (P < 0.05) are indicated in bold. The mean and sd of the above variables are given in Table 1. pigmentation (as TTCA, ng/mm). For the reduced sample sizes, we used residual temperatures rather than using models including air temperature or oral temperature as a covariate to better control for the positive correlation between these temperatures (see above). Residual brood patch temperature after controlling for hourly air temperature significantly decreased with increasing pheomelanin pigmentation (robust linear regression: coefficient se = , Wald v 2 = 4.83, P = 0.028; Fig. 5a). Likewise, residual brood patch temperature after controlling for oral temperature decreased significantly with increasing pheomelanin pigmentation (robust linear regression: coefficient se = , Wald v 2 = 15.07, P < 0.001; Fig. 5b). Residual oral temperature after controlling for hourly air temperature was not significantly related to pheomelanin pigmentation (robust linear regression: coefficient se = , Wald v 2 = 0.14, P = 0.70). Brood patch temperature and incubation period Among the 27 females in which hatching success/ failure could be observed, 22 succeeded in hatching the clutch and five failed (c. 19%), perhaps due to nest predation. The probability of hatching

9 816 M. Hasegawa et al. (a) (b) Among females whose clutch hatched, the length of the incubation period (mean = 14 days, range = 12 16, n = 22) was not significantly related to residual brood patch temperature (r s = 0.10, P = 0.67) or residual oral temperature (r s = 0.23, P = 0.30). This was also the case when we used residual brood patch temperature controlling for oral temperature (r s = 0.25, P = 0.26). Qualitatively similar, non-significant relationships were found when we used hourly mean air temperature or oral temperature as a covariate (Spearman s partial correlation coefficient, r sp < 0.28, P > 0.20). Neither female ornament nor pheomelanin pigmentation explained the length of the incubation period (ornament: n = 21, r s < 0.29, P > 0.20; pigmentation: n = 14, r s = 0.37, P = 0.20). Figure 5. Plot of pheomelanin pigmentation (as TTCA, ng/ mm) in relation to residual female brood patch temperature after controlling for (a) air temperature and (b) oral temperature. Simple regression lines are shown: (a) residual brood patch temperature = 0.15*pheomelanin pigmentation ; F 1,17 = 4.05, P = 0.06; (b) residual brood patch temperature = 0.13*pheomelanin pigmentation 2.01; F 1,17 = 6.30, P = 0.02). was not significantly predicted by brood patch temperature, when controlling for hourly mean air temperature or oral temperature as a covariate (robust GLM with binomial error distribution; all: Wald v 2 < 2.37, P > 0.12). This was also the case when we studied oral temperature, controlling for hourly mean air temperature (robust GLM with binomial error distribution: both variable: Wald v 2 < 1.70, P > 0.19). Neither female ornament nor pheomelanin pigmentation explained the probability of hatching in univariable models (ornament: n = 25, Wald v 2 < 2.32, P > 0.13; pigmentation: n = 17, Wald v 2 = 1.32, P = 0.25; note that the multivariate model could not be performed due to complete separation). DISCUSSION The main finding of the current study was that brood patch temperature could be predicted by a measure of plumage ornamentation, the throat saturation value we recorded in female Barn Swallows. This relationship reflects internal thermogenesis rather than the influence of ambient air temperature, as we statistically controlled for air temperature (Table 1a). This inference is further supported by the observation that the relationship between brood patch temperature and throat coloration remained after controlling for oral temperature (Table 1b), thus excluding the effect of local spatiotemporal variation in microclimate around focal individuals. Furthermore, this result indicates that females maintain brood patch temperature partially independent of oral temperature, and that this is in part dependent on their plumage coloration. Although plumage coloration is affected by several post-moulting processes (Montgomerie 2006), direct inspection of pheomelanin pigmentation recovered a negative relationship with brood patch temperature, indicating the importance of the pigmentation itself. To our knowledge, this is the first study to show that plumage ornamentation predicts active thermal output at brood patches. The throat patch of the Barn Swallow consists mainly of pheomelanin (McGraw et al. 2004, 2005, Arai et al. 2015), and genetic upregulation of pheomelanin pigment deposition (via the melanocortin system) is known to decrease body temperature (Minvielle et al. 2007, 2010, Ducrest et al. 2008). Thus, our finding is consistent with

10 Female ornaments and brood patch temperature 817 the genetic linkage between melanin-based coloration and body temperature, although further studies are needed to determine the generality of the result. In another model system, the Barn Owl, there was no relationship between pheomelanic coloration and body temperature, although information on brood patch temperature is lacking (Roulin & Randin 2015, Dreiss et al. 2016). It has also recently been proposed that feather structure plays a role in relation of pheomelanin pigmentation (Roulin et al. 2013, Koskenpato et al. 2016) such that barb density, and thus the extent of thermal insulation, is lower in pheomelanic birds. Hence, even if thermogenesis is similar, the surface temperature under pheomelanin plumage patches may be different. Although Barn Swallows with chestnut coloration were virtually absent in our population, there is still variation in pheomelanin plumage coloration that is correlated with throat coloration (Arai et al. 2015, M. Hasegawa unpubl. data). In Barn Swallows the above observation may explain our result that brood patch temperature, but not oral temperature (which is not covered by plumage and thus affected by ambient temperature directly via respiration), was correlated with feather coloration. The relative importance of these and other potential explanations (e.g. differential thermoregulation strategy; Dreiss et al. 2016) requires further study. Although we found a negative relationship between throat coloration and brood patch temperature, this relationship might not be the primary concern of males in choosing their mate. Our study also established that incubation period was not predicted by brood patch temperature, whereas incubation period is predicted by female incubation behaviour (Hasegawa et al. 2012a, Hasegawa & Arai 2016). This suggests a relatively minor influence for a mate s brood patch temperature on male reproductive success (and male mate preference; see introductory text). Rather, because colourful females are older and more experienced and hence tend to have greater reproductive output in this species (Turner 2006), males should prefer colourful females as suggested by previous correlative studies (Safran & McGraw 2004). The direction and intensity of male mate preference based on female coloration and its consistency with female quality remain to be clarified in future studies by experimentally manipulating plumage coloration. Even if the high brood patch temperature has a minor influence on a female s reproductive output compared with other female attributes, it may indirectly influence offspring phenotype (DuRant et al. 2013, Hepp et al. 2015). High egg temperatures typically enhance embryo growth, which is consistent with the expectation of a negative relationship between egg temperature and incubation period (Hepp et al. 2015), although overheating might be maladaptive (Nord & Nilsson 2016). This might also be applicable in Barn Swallows. In contrast to artificial control of egg temperature, which is used to test the influence of temperature on clutches (e.g. Bernstein & Bech 2016), the heat energy of incubation is lost through the interaction between the bird and the egg (Barn Swallows: mean egg temperature, 36.2 C; Voss et al. 2008, mean brood patch temperature, C; this study, Deeming 2008). In addition, brood patch temperature might affect the diurnal behaviour (e.g. foraging) of incubating birds via differential energy requirements (Ardia et al. 2008). It therefore remains to be tested whether the influence of brood patch temperature on the fitness of offspring is similar to the results obtained from studies using artificial incubation temperatures. Regarding tail ornaments, another well-known target of sexual selection (Møller 1994, Turner 2006), we observed a positive relationship between tail-length and brood patch temperature when controlling for oral temperature (but not for hourly mean air temperature), indicating that females with longer tails did not provide a higher absolute brood patch temperature, but did invest relatively more thermal energy into brood patches compared with oral temperature. Female taillength, but not other ornaments, predicted reproductive expenditure (sensu Clutton-Brock 1991) in the Japanese subspecies H. r. gutturalis, with longtailed females being likely to rear more offspring (M. Hasegawa unpubl. data), as found in the European subspecies H. r. rustica, partially due to their high body condition or genetic quality (Turner 2006). Thus, high thermal output of the brood patch may be one kind of reproductive expenditure of long-tailed females with higher thermal expenditure on the brood patch for incubation rather than on other body parts for somatic maintenance. It remains to be clarified why female tail-length, but not other ornaments, predicts reproductive expenditure.

11 818 M. Hasegawa et al. Apart from the signalling role of female plumage coloration, the negative relationship between plumage coloration and brood patch temperature may affect the relative contribution of males and females at incubation (Parker et al. 2013). As female brood patch temperature decreases, sex differences in incubation ability decrease and thus there should be more opportunity for male participation in incubation, because male incubation becomes relatively more effective. In populations of Barn Swallows with colourful plumage, males play a greater role in incubation: the male contribution is 9% in H. r. erythrogaster, a taxon with a large throat patch and colourful ventral plumage, compared with 6% in H. r. gutturalis, which have a large red throat but have pale ventral plumage; in H. r. rustica, which have a small red throat patch, males make a negligible contribution to incubation (Turner 2006). Future studies are required to explore whether our within-population pattern can be extrapolated to patterns of male contribution to incubation across populations and taxa. We are grateful to the residents of Kanagawa Prefecture for their kind support and assistance. We also thank Nobuyuki Kutsukake and members of his laboratory at Sokendai (the Graduate University for Advanced Studies) and the laboratory of Ecology and Evolutionary Biology of Tohoku University for their very kind advice. We thank Yukiko Nakanishi for her assistance with the pigment analysis. M.H. was supported by a Research Fellowship from the Japan Society for the Promotion of Science (JSPS, 15J10000). We thank two anonymous reviewers, Rauri Bowie and Jeff Kelly for their comments that helped improve this manuscript. REFERENCES Amundsen, T. & P arn, H Female coloration: review of functional and nonfunctional hypotheses. In Hill, G.E. & McGraw, K.J. (eds) Bird Coloration, Vol. II: Function and Evolution: Cambridge, MA: Harvard University Press. Andersson, M Sexual Selection. Princeton, NJ: Princeton University Press. Arai, E., Hasegawa, M. & Nakamura, M Divorce and asynchronous arrival in Barn Swallows Hirundo rustica. Bird Study 56: Arai, E., Hasegawa, M., Nakamura, M. & Wakamatsu, K Male pheomelanin pigmentation and the breeding onset in Barn Swallow Hirundo rustica gutturalis. J. Ornithol. 156: Ardia, D.R., Perez, J.H. & Clotfelter, E.D Experimental cooling during incubation leads to reduced innate immunity and body condition in nestling tree swallows. Proc. R. Soc. Lond. B 277: Auer, S.K., Bassar, R.D. & Martin, T.E Biparental incubation in the Chestnut-vented Tit-babbler Parisoma subcaeruleum: mates devote equal time, but males keep eggs warmer. J. Avian Biol. 38: Bernstein, H.H. & Bech, C Incubation temperature influences survival in a small passerine bird. J. Avian Biol. 47: Clutton-Brock, T.H The Evolution of Parental Care. Princeton, NJ: Princeton University Press. Cuervo, J.J., de Lope, F. & Møller, A.P The function of long tails in female barn swallows (Hirundo rustica): an experimental study. Behav. Ecol. 7: Deeming, D.C Avian brood patch temperature: relationships with female body mass, incubation period, developmental maturity and phylogeny. J. Therm. Biol 33: Deeming, D.C. & Du Feu, C.R Measurement of brood patch temperature of British passerines using an infrared thermometer. Bird Study 55: Dreiss, A.N., Sechaud, R., Beziers, P., Villain, N., Genoud, M., Almani, B., Jenni, L. & Roulin, A Social huddling and physiological thermoregulation are related to melanism in the nocturnal Barn Owl. Oecologia 180: Ducrest, A.L., Keller, L. & Roulin, A Pleiotropy in the melanocortin system, coloration and behavioural syndromes. Trends Ecol. Evol. 23: DuRant, S.E., Hopkins, W.A., Hepp, G.R. & Walters, J.R Ecological, evolutionary, and conservation implications of incubation temperature-dependent phenotypes in birds. Biol. Rev. 88: Falconer, D.S. & MacKay, T.F.C Introduction to Quantitative Genetics, 4th edn. Harlow: Longman Science and Technical. Fan, W., Voss-Andreae, A., Cao, W.H. & Morrison, S.F Regulation of thermogenesis by the central melanocortin system. Peptides 26: Foley, J.D. & van Dam, A Intensity and color. In Foley, J.D. & van Dam, A. (eds) Fundamentals of Interactive Computer Graphics: Philippines: Addison-Wesley. Forstmeier, W. & Schielzeth, H Cryptic multiple hypotheses testing in linear models: overestimated effect sizes and the winner s curse. Behav. Ecol. Sociobiol. 65: Fujita, G Nest site selection and reproductive success in barn swallows preliminary report. Strix 12: (Japanese with English summary). Graham, M.H Confronting multicollinearity in ecological multiple regression. Ecology 84: Griffith, S.C. & Pryke, S.R Benefits to females of assessing color display. In Hill, G.E. & McGraw, K.J. (eds) Bird Coloration, Vol II: Oxford: Oxford University Press. Hasegawa, M Sexual selection on multiple ornaments in the Barn Swallow Hirundo rustica gutturalis. Dissertation. University of Tsukuba, Tsukuba. Hasegawa, M. & Arai, E Differential female access to males with large throat patches in the Asian Barn Swallow Hirundo rustica gutturalis. Zoolog. Sci. 30:

12 Female ornaments and brood patch temperature 819 Hasegawa, M. & Arai, E. 2015a. Infanticide on a grown nestling in a sparse population of Japanese Barn Swallows Hirundo rustica gutturalis. Wilson J. Ornithol. 127: Hasegawa, M. & Arai, E. 2015b. Experimentally reduced male ornamentation increased paternal care in the Barn Swallow. J. Ornithol. 156: Hasegawa, M. & Arai, E Long incubation off-bouts of females paired with colorful males in Barn Swallows. Wilson J. Ornithol. 128: Hasegawa, M., Arai, E., Watanabe, M. & Nakamura, M Methods for correcting plumage color fading in the Barn Swallow. Ornithol. Sci. 7: Hasegawa, M., Arai, E., Watanabe, M. & Nakamura, M Mating advantage of multiple male ornaments in the Barn Swallow Hirundo rustica gutturalis. Ornithol. Sci. 9: Hasegawa, M., Arai, E., Watanabe, M. & Nakamura, M. 2012a. High incubation investment of females paired to attractive males in Barn Swallows. Ornithol. Sci. 11: 1 8. Hasegawa, M., Arai, E., Watanabe, M. & Nakamura, M. 2012b. Female mate choice based on territory quality in Barn Swallows. J. Ethol. 30: Hasegawa, M., Arai, E., Watanabe, M. & Nakamura, M Colourful males hold high quality territories but exhibit reduced paternal care in Barn Swallows. Behaviour 151: Hasegawa, M., Watanabe, M. & Nakamura, M Promiscuous copulation attempts and discriminate pairing displays in male Barn Swallows as revealed by model presentation. Ethol. Ecol. Evol. 28: Hepp, G.R. & Kennamer, R.A Warm is better: incubation temperature influences apparent survival and recruitment of Wood Ducks (Aix sponsa). PLoS One 7: e Hepp, G.R., DuRant, S.E. & Hopkins, W.A Influence of incubation temperature on offspring phenotype and fitness in birds. In Deeming, D.C. & Reynolds, S.J. (eds) Nests, Eggs, & Incubation: New Ideas about Avian Reproduction: Oxford: Oxford University Press. Ito, S., Nakanishi, Y., Valenzuela, R.K., Brilliant, M.H., Kolbe, L. & Wakamatsu, K Usefulness of alkaline hydrogen peroxide oxidation to analyze eumelanin and pheomelanin in various tissue samples: application to chemical analysis of human hair melanins. Pigment Cell Melanoma Res. 24: Japan Meteorological Agency Miura 2015 (1jikan goto no atai). Available at: etrn/view/hourly_a1.php?prec_no=46&block_no=0392&year= 2015&month=05&day=1&view=p1 (in Japanese). (accessed 12 August 2016). Kleindorfer, S., Fessl, B. & Hoi, H More is not always better: male incubation in two Acrocephalus warblers. Behaviour 132: Kokko, H., Brooks, R., McNamara, J.M. & Houston, A.I The sexual selection continuum. Proc. R. Soc. Lond. B 269: Kose, M. & Møller, A.P Sexual selection, feather breakage and parasites: the importance of white spots in the tail of the Barn Swallow. Behav. Ecol. Sociobiol. 45: Kose, M., M and, R. & Møller, A.P Sexual selection for white tail spots in the Barn Swallow in relation to habitat choice by feather lice. Anim. Behav. 58: Koskenpato, K., Ahola, K., Karstinen, T. & Karell, P Is the denser contour feather structure in pale grey than in pheomelanic Brown Tawny Owls (Strix aluco) an adaptation to cold environments? J. Avian Biol. 47: 1 6. Lande, R. & Arnold, S.J The measurement of selection on correlated characters. Evolution 37: Lea, R.W. & Klandorf, H The blood patch. In Deeming, D.C. (ed.) Avian Incubation: Behaviour, Environment, and Evolution: Oxford: Oxford University Press. Lendvai, A.Z., Kis, J., Szekely, T. & Cuthill, I.C An investigation of mate choice based on manipulation of multiple ornaments in Kentish plovers. Anim. Behav. 67: Matysiokova, B. & Remes, V Incubation feeding and nest attentiveness in a socially monogamous songbird: role of feather colouration, territory quality and ambient environment. Ethology 116: McCarffery, D.J Application of thermal imaging in avian science. Ibis 155: McGraw, K.J., Safran, R.J., Evans, M.R. & Wakamatsu, K European Barn Swallows use melanin pigments to color their feathers brown. Behav. Ecol. 15: McGraw, K.J., Safran, R.J. & Wakamatsu, K How feather color reflects its melanin content. Funct. Ecol. 19: Minvielle, F., Gourichon, D., Ito, S., Inoue-Murayama, M. & Riviere, S Effects of the dominant lethal yellow mutation on reproduction, growth, feed consumption, body temperature, and body composition of the Japanese Quail. Poult. Sci. 86: Minvielle, F., Bed hom, B., Coville, J.L., Ito, S., Inoue- Murayama, M. & Gourichon, D Causal mutation and associated traits for the silver Japanese Quail. 9th World Congress on Genetics Applied to Livestock Production C52DFDEC3} (accessed 12 August 2016). Møller, A.P Female choice selects for male sexual tail ornaments in the monogamous swallow. Nature 332: Møller, A.P Viability is positively related to degree of ornamentation in male swallows. Proc. R. Soc. Lond. B 243: Møller, A.P Sexual selection in the Barn Swallow (Hirundo rustica), III: female tail ornaments. Evolution 47: Møller, A.P Sexual Selection and the Barn Swallow. Oxford: Oxford University Press. Møller, A.P. & de Lope, F Senescence in a short-lived migratory bird: age-dependent morphology, migration, reproduction and parasitism. J. Anim. Ecol. 68: Montgomerie, R Cosmetic and adventitious colors. In Hill, G.E. & McGraw, K.J. (eds) Bird Coloration. II. Function and Evolution: Cambridge, MA: Harvard University Press. Mundry, R. & Nunn, C.L Stepwise model fitting and statistical inference: turning noise into signal pollution. Am. Nat. 173: Nakagawa, S. & Schielzeth, H Repeatability for Gaussian and non-gaussian data: a practical guide for biologists. Biol. Rev. 85: