How blue are British tits? Sex, age and environmental effects

Transcription

1 Bird Study ISSN: (Print) (Online) Journal homepage: How blue are British tits? Sex, age and environmental effects Peter N. Ferns & Shelley A. Hinsley To cite this article: Peter N. Ferns & Shelley A. Hinsley (2010) How blue are British tits? Sex, age and environmental effects, Bird Study, 57:3, To link to this article: Published online: 30 Jul Submit your article to this journal Article views: 732 View related articles Citing articles: 1 View citing articles Full Terms & Conditions of access and use can be found at Download by: [ ] Date: 05 January 2018, At: 02:33

2 Bird Study (2010) 57, How blue are British tits? Sex, age and environmental effects PETER N. FERNS 1 * and SHELLEY A. HINSLEY 2 1 School of Biosciences, Cardiff University, Cardiff, CF10 3AX, UK and 2 Centre for Ecology & Hydrology, Maclean Building, Crowmarsh Gifford, Wallingford, OX10 8BB, UK Capsule The blue colour of the coverts and crowns of Blue Tits Cyanistes caeruleus and Great Tits Parus major is influenced by sex, age and environmental factors. Aims To quantify the blueness of two species of tits breeding in different habitats in Britain. Methods By manipulating daylength in the laboratory, adult Blue Tits were induced to moult at fast and slow speeds. When they had finished, the blue colour of their wing coverts and crowns was measured (in the range visible to human observers), and compared with the colour of birds caught in the field. Results As well as highly significant sex and age differences in colour, Blue Tits were 26% bluer in 1998 than 2000, and male Great Tits were 15% less blue in small woods than in large woods in our study population in East Anglia, England. Both species became more saturated with blue up to the age of three years. British Cy. c. obscurus were darker blue than Cy. c. caeruleus of Continental Europe. Conclusions There are significant environmental influences on the blueness of British tits in addition to the well-known age and sex effects. The discovery that many birds have tetrachromatic vision and can detect UV light in the range nm that is invisible to human beings (Huth & Burkhardt 1972, Wright 1972, Chen et al. 1984, Hart 2001, Cuthill 2006) has led to a wealth of studies investigating the possible use of UV light in foraging (Willson & Whelan 1989, Honkevaara et al. 2002) and communication (reviewed in Hill & McGraw [2006]), including the extent to which bird plumages are reflective in the UV part of the spectrum. Instead of being produced solely by pigments, short wavelength plumage colours (including UV) are usually produced by coherent light scattering from the interfaces between keratin and minute air bubbles contained within the spongy medulla of feather barbs (Prum et al. 1998, 1999). In some species, such as Blue Tits Cyanistes caeruleus, sexual dimorphism is maximal in the UV range and has been shown to provide a basis for mate choice (Andersson et al. 1998, Hunt et al. 1998, 1999). Fitzpatrick (1998) was the first to suggest that nutritional conditions during the moult might influence the structural coloration of bird plumage. In a review of *Correspondence author. fernspn@cardiff.ac.uk integument coloration in birds, Prum (2006) suggested that the lightness and saturation (chroma) of such tissues might be more susceptible to developmental perturbation than their hue. This seems to be in agreement with studies showing that captive male brown-headed cowbirds subjected to periodic fasts during the moult, grew less saturated iridescent plumage (McGraw et al. 2002). Keyser & Hill (1999) also showed that male blue grosbeaks, which moulted faster in the field (as indicated by the spacing of tail feather growth bars, which they interpreted as an indication of good nutritional conditions), had brighter blue breast and rump plumage. However, Hill (2006) emphasized the need for more experimental studies on the effects of nutritional deprivation on structural coloration. We used experimental manipulation of daylength to change the speed of moult, and hence feather quality (Dawson et al. 2000), in captive Blue Tits, and measured the blueness of their feathers in the range visible to human observers. In the field, nutritional conditions are likely to manifest themselves through differences in the size, body condition and colour of birds in habitats of different quality (van Balen 1967, Ulfstrand et al. 1981, Figuerola et al. 1999). We therefore measured the blue coloration of Blue Tits and Great Tits Parus major 2010 British Trust for Ornithology

3 316 P.N. Ferns and S.A. Hinsley breeding in different habitats. These comprised large woods (good habitat) with a relatively high breeding success, and small woods (poor habitat) with a poor breeding success (Hinsley et al. 1999, Hinsley et al. 2003). Like the UV coloration, the blueness of the coverts and crowns of Blue Tits (and the coverts of Great Tits) which is visible to us, is produced structurally. Furthermore, Örnborg et al. (2002) found that the average peak reflectance of Blue Tit crown feathers shifted upwards, from about 359 nm for males and 373 nm for females immediately after the moult, into the range visible to human observers by the time of chick-feeding (to 404 nm for males and 413 nm for females) as a result of feather wear and soiling. Delhey et al. (2006) found a similar trend, though average peak reflectances remained in the UV during chickfeeding. In this paper we describe the blue colour of the wing covert and crown feathers of both species in the field, excluding the ultraviolet part of the spectrum, though blue colours including and excluding the UV are highly correlated. Age and sex-related differences in the blue plumage coloration of Blue Tits in both the visible and invisible (to humans) parts of the spectrum are well established and play an important role in mate choice (Perrins 1979, Cramp & Perrins 1993, Andersson et al. 1998, Hunt et al. 1998, 1999, Johnsen et al. 2003, Delhey & Kempenaers 2006). We thus took account of these major sources of variation. Even though some parts of the blue plumage of Blue Tits are most dimorphic at wavelengths greater than 400 nm, which human beings can detect, and despite the fact that subjective dimorphism has been described along with ageing information (Svensson 1992, Cramp & Perrins 1993, Scott 1993, Harper 2000), the extent of our ability to identify the sex of this species has been questioned. For example, Guilford & Harvey (1998) described the species as being largely monomorphic in our visible spectrum. We also, therefore, carried out discriminant analysis of Blue Tit coloration to see in what proportion of our population it might be possible to distinguish the sexes by colour, and then tested our own ability to do so based on an additional sample. METHODS Moult speed in the laboratory We have already described the protocol for inducing different rates of moult in captivity (Ferns & Hinsley 2008), and so present only a summary here. Ten birds were captured under licence in July and August in Monks Wood, Huntingdon and housed in individual cages in rooms with different artificial daylengths (Dawson et al. 2000). In one room (containing five females), the daylength was kept constant at 18 hours (its value in the field at the time), while in the other (containing four females and one male) it was reduced to 12 hours at the rate of one hour each week and then kept constant. Each bird was provided with food and water ad libitum. Plumage colour was recorded as soon as the wing, crown and flank feathers of each individual bird had finished moulting, which took about a month longer, on average, in the long daylength group. To ensure that all individuals moulted all of their feathers, we used only birds that were more than a year old. Birds were aged using the criteria in Svensson (1992) and sexed by examination of the brood patch (females were identified by the presence of a refeathering brood patch). We caught only one adult male, but excluded it from the main calculations and used it merely to confirm that males are affected in a similar way to females. The birds were transferred to an outdoor aviary at the end of the experiment and released after they had completed another wing moult under normal daylight conditions. Colour of tits in different habitats The colour of wild birds in good and poor habitats (large and small woods in East Anglia, c. 100 ha and c. 1 ha respectively, see Hinsley et al. (1999) for further details) was measured whilst they were feeding chicks in six years in Blue Tits ( ) and in three years in Great Tits ( ). The most overtly blue parts of the plumage of Great Tits, to human eyes, are the proximal upper wing coverts, which appear slate grey. The proximal coverts also form one of the bluest parts of the plumage of Blue Tits, and we therefore measured the feathers in this region in both species. To avoid the complication of the differently coloured coverts of juveniles, we only examined the moulted feathers in all age groups, thus avoiding the primary coverts and any unmoulted distal greater coverts. The wing was held in the closed position so that the feathers were folded together exposing mostly their blue central areas. Six measurements of the colour were made, each bird being moved away from the sensor (see later) after each measurement and successive measurements being made from different (but sometimes partially overlapping) parts of the coverts. The colour of only one wing (usually the right) was measured.

4 Blueness of British tits 317 We also measured the colour of the crown, mainly because this is the bluest part of the plumage of Blue Tits (Cramp & Perrins 1993). There is an area of white feathers at the front of the crown, and the extent of this varies between individuals. To avoid this area, we therefore measured the crown in three places just to the front of the apex of the head, at the apex of the head and just behind the apex, and repeated this procedure twice (to make up the six measurements). Great Tits have a glossy blue-black crown, that does not show any seasonal fading (Figuerola & Senar 2005), and we measured the colour of this in the same way as for Blue Tits. Re-traps were only included in a separate analysis of age-related colour change, and were excluded from all other analyses of the coloration of breeding birds. The colour of both wild and captive birds was measured with a Minolta CR221 Chroma Meter which records the colour of a 3 mm diameter circle in the nm wavelength region, with a spectral response matching that of the CIE 1931 Standard (human) Observer curve. The measuring area was illuminated by an inbuilt light source at 45. The instrument was calibrated using a certified standard white plate (CRA45) prior to the measurement of every bird. Three colour parameters were recorded lightness (L; black = 0%, white = 100%), chroma (C; colour saturation, 0 100%) and hue (H; 0 = 360 = red, 90 = yellow, 180 = green, 270 = blue). When describing differences in saturation and chroma between different groups in this paper we use relative rather than absolute percentages. The repeatability of the six individual measurements of each bird in the field for all plumage areas was quite high, the least repeatable value being for the lightness of Blue Tit coverts in 2001 (L, F 34,175 = 12.80, P < ), and the most being for the hue of Great Tit crowns in 1998 (H, F 46,281 = 24.10, P < ). Hues were more repeatable because they varied more widely between individuals than lightness or chroma. The repeatability of the means of the six measurements we used in all analyses (based on individuals that were accidentally or deliberately measured twice on the same day) was even higher ( ). In our experience, the repeatability of measurements made with the Minolta colorimeter is higher than those made with a fibre optic spectrophotometer (even when the probe is enclosed within a sheath), partly because the area being measured can be precisely located in the hole in the baseplate of the instrument before the light projection tube is lowered onto it. Statistical analyses Two wild Blue Tits had lost all the feathers from the tops of their heads (probably due to parasite infestation) by the time they were measured and so the crown colours of these could not be measured. The lightness and chroma of breeding Blue Tit coverts were normally distributed, as were these measures of crown colour after removing two and one outliers (standardized residuals >3) respectively. These were, therefore, analysed using ordinary parametric techniques (after removing the outliers). The lightness of breeding Great Tit coverts was normally distributed, as was covert chroma after the removal of one outlier, and so we also analysed these parametrically. In each case, glm anova s were undertaken with age, sex, wood and year as explanatory variables, and only first order interactions included. Non-significant terms were then removed unless they were involved in a significant interaction. The hues of breeding Blue Tit plumage, and all the remaining measures of breeding Great Tit plumage colour were not normally distributed. This was due to the presence of more than ten outliers in each case whose influence could not be reduced by transformation. Since there were no clear grounds for excluding these individuals from any statistical analysis, we used medians and quartile deviations instead of means and standard deviations as measures of central tendency and variation. We started with whichever of the age, sex, year and wood differences was largest and randomly re-sampled the measured colours times to determine the two-tailed probability of obtaining differences in medians as large as, or larger than, those actually observed. If these proved to be significant, we proceeded to test the next largest difference within each of the significantly different groups. In the case of year differences, we calculated the largest resampled median difference between all pairs of years and compared it with the largest observed difference. To facilitate comparisons we also provide trimmed means (i.e. means after the largest and smallest 5% of values were removed) for these variables. This procedure effectively removed outliers from the non-normal samples. To determine whether size influenced colour, we repeated the above analyses on the whole population, within each age and sex group, including wing length as

5 318 P.N. Ferns and S.A. Hinsley a covariate measure of size (van Balen 1967). It was necessary to separate the ages and sexes in this way since males have brighter plumage and longer wings than females and likewise for adults compared with juveniles. Thus age and sex groups analysed together would, in most cases, produce an apparently significant wing length effect. To determine the influence of age on colour more precisely, we examined the smaller number of re-traps whose age was known in more detail, either because they were ringed as pulli (in which case we knew their exact age), or because they were ringed as juveniles (in which case we knew their age to within a couple of weeks based on the number of days elapsed between the average population hatching date in the year they hatched and the date on which they were measured). By including only birds measured on more than one occasion, we avoided any undue influence of differential survival on apparent age-related colour change (Delhey & Kempenaers 2006). We could not make any comparison with the colour trend in those birds of known age that were measured only once, because the sample size of birds 2 years old was too small. We used the raw colour measurements as a basis for discriminant analyses of the sexes. In one year (2001), one of us identified the sex of captured Blue Tits using the brood patch (SAH), while the other did so using overall plumage brightness (PNF). Parametric analyses were undertaken using minitab and genstat 9.1, re-sampling was undertaken with resampling statistics , and circular statistics were calculated with oriana Since three parameters are necessary to define any colour precisely, none of our three (lightness, chroma or hue) can be considered superfluous, and thus we did not apply Bonferroni corrections (Nakagawa 2004). Had we done so, the minimal level for statistical significance would have been P = The largest shared variance between any pair of colour parameters in birds measured in the field was 43% between carpal hue and chroma in Great Tits (0.3% in Blue Tits). The largest shared variance in either of the other two pairs of colour parameters within either species was 11%. RESULTS Colour and speed of moult in captive Blue Tits Female Blue Tits subjected to short (12 hour) daylengths underwent moult significantly faster than those on long (18 hour) days, though the rate of primary moult was still about 25% slower than that of birds in the field (Ferns & Hinsley 2008). There was no significant difference in the lightness or chroma of the wing coverts of fast and slow moulting birds ( P > in both cases). However, they did differ significantly in wing covert hue (Fig. 1 ; H fast = ± 1.9 ; H slow = ± 6.2 ; t 7 = 4.67, P = 0.002). The slow moulting birds were very close to pure spectral blue (270 in the cylindrical co-ordinates of the LCH colorimetric system), whilst the fast moulters had a trace of green (pure spectral green = 180 ), giving them a significantly lower hue than 270 ( t 4 = 6.89, P < 0.025). The feathers of fast moulting birds were of poorer quality and thus more prone to breakage, while the overall tone of their plumage was greyer than that of slow moulting birds. The single male Blue Tit in the fast moulting group had a wing covert hue of 265.1, which was not quite significantly greater than that of fast-moulting females (two-tailed t 5 = 3.67, P = 0.067). Adult females that moulted in the same year as the laboratory birds, and which came from the same wood, had a hue of ± 8.6 ( n = 5). This is a significantly lower value than slow moulters in the laboratory (Fig. 1 ; t 8 = 2.35, P = 0.043), but not significantly greater than fast moulters ( t 8 = 0.61, P > 0.200). There was no significant difference in any of the colour parameters of the crowns of fast and slow moulting birds. Covert colour in breeding adults The large size of the standard deviations of the hues of the coverts and crowns of both species (Tables 1 and 2 ) shows the powerful influence of the outliers, in turn necessitating the use of non-parametric statistics to analyse them. In the case of the hues, the outliers contained an excess of females compared with the bulk of the population. For example, all of the outlying Great Tits were female, though the proportion of these that were juvenile was no greater than expected by chance (Fisher Exact Test, P > 0.200). Thus, the hues of male Blue Tits and Great Tit coverts were much more tightly clustered around pure spectral blue (270 ) than those of females (Fig. 2 ), especially in those individuals that were more saturated (higher chroma). Hue becomes increasingly difficult to measure, as well as losing its importance as a colour parameter, when chroma approaches zero i.e. as a colour approaches black, white or (as in this case) grey. However, even in the most variable group of individuals we measured i.e. juvenile female Great Tits with chromas less than 1%, both covert and crown hue were highly significantly clustered in the blue region of the spectrum (Rayleigh s Uniformity

6 Blueness of British tits 319 Figure 1. Box and whisker plots (whiskers = range; box = 25th 75th percentiles; line = median; sample size above) of the dominant hue (in degrees) of the wing coverts of Blue Tits that moulted at fast and slow speeds in the laboratory experiment, compared with those that moulted under natural daylight conditions in the field during the same year. The hue of fast moulters was significantly lower than pure spectral blue (270 ), and lower than that of slow moulters, but not significantly different from birds that moulted naturally. Table 1. Covert and crown colour parameters in Blue Tits of different ages and sexes, all wood sizes combined. Plumage area Colour parameter Sex Age n Mean ± sd Median ± QD Coverts Lightness Male*** Adult ± 1.4 Juvenile ± 1.5 Female Adult ± 1.6 Juvenile ± 1.3 Chroma Male*** Adult*** ± 2.11 Juvenile ± 2.06 Female Adult*** ± 1.67 Juvenile ± 1.36 Hue Male*** Adult ± ± 2.6 Juvenile ± ± 1.7 Female Adult ± ± 2.7 Juvenile ± ± 3.2 Crown Lightness Male Adult* ± 4.0 Juvenile ± 4.7 Female Adult* ± 4.4 Juvenile ± 4.8 Chroma Male*** Adult** ± 3.02 Juvenile ± 3.49 Female Adult** ± 2.84 Juvenile ± 2.42 Hue Male*** Adult ± ± 2.2 Juvenile ± ± 1.4 Female Adult ± ± 1.5 Juvenile ± ± 1.9 Lightness and chroma (saturation) are measured on a percentage scale, while pure spectral blue has a hue of 270 ; *P = ; **P = ; ***P < (based on GLM ANOVAS or re-sampling [see Methods]; asterisks are placed next to males for sexually significant differences and next to adults for age differences); a median ± quartile difference (QD) is given when a sample was not normally distributed and in these cases a trimmed mean ± sd is also provided for comparison.

7 320 P.N. Ferns and S.A. Hinsley Table 2. Covert and crown colour parameters in Great Tits of different ages and sexes, all wood sizes combined. Plumage area Colour parameter Sex Age n Mean ± sd Median ± QD Coverts Lightness Male Adult ± 1.1 Juvenile ± 1.2 Female Adult ± 1.4 Juvenile ± 1.3 Chroma Male*** Adult ± 0.69 Juvenile ± 0.82 Female Adult ± 0.41 Juvenile ± 0.43 Hue Male* Adult** ± ± 1.6 Juvenile ± ± 2.9 Female Adult ± ± 20.5 Juvenile ± ± 14.4 Crown Lightness Male*** Adult ± ± 0.7 Juvenile ± ± 0.6 Female Adult ± ± 0.8 Juvenile ± ± 0.8 Chroma Male*** Adult** ± ± 0.54 Juvenile ± ± 0.72 Female Adult ± ± 0.36 Juvenile ± ± 0.46 Hue Male Adult* ± ± 3.7 Juvenile ± ± 6.2 Female Adult ± ± 16.5 Juvenile ± ± 12.8 Lightness and chroma (saturation) are measured on a percentage scale, while pure spectral blue has a hue of 270 ; *P = ; **P = ; ***P < (based on GLM ANOVAS or re-sampling [see Methods]; asterisks are placed next to males for sexually significant differences and next to adults for age differences); a median ± quartile difference (QD) is given when a sample was not normally distributed and in these cases a trimmed mean ± sd is also provided for comparison. tests; coverts, z = 16.35, P < ; crowns, z = 7.04, P = ). An examination of Blue Tits in Fig. 2 shows that chromas below 5% tend to become more variable, whereas the same degree of variability does not occur in Great Tits until their chromas are below 2%. Male Great Tit hues are just as tightly clustered as those of Blue Tits, albeit around a very slightly greenish-blue compared with the pure spectral blue of male Blue Tits. However, female Great Tit hues are more variable than those of female Blue Tits, and the majority in both species are blue with a little green, though the most unsaturated Great Tits appear predominantly grey. Blue Tits were in fact about three times as saturated with blue as Great Tits as well as being closer to pure spectral blue (Fig. 2 ). Thus, the brightest adult male Great Tits had coverts that were only as colourful as some juvenile female Blue Tits, but there was more overlap in dominant wavelengths (i.e. hues) in the two species. Sex and age effects We deal with sex and age differences first because these are the major sources of variability. Female Blue Tits were significantly lighter in colour than males (L, F 1,225 = 290.0, P < ), but males were more than twice as saturated (C, F 1,219 = 402.0, P < ) and their hues (H) were closer to pure spectral blue (Table 1, resampling, P < ). Male Great Tits were also more saturated with colour than females ( F 1,107 = 325.3, P < ) and closer to pure spectral blue (re-sampling, P =0.018) (Table 2 ). Adult Blue Tits were significantly more saturated than juveniles (17% more in males and 50% in females) ( F 1,219 = 28.4, P < ), but there were no age-related differences in lightness or hue. Adult male Great Tits were significantly closer to pure spectral blue than juveniles (resampling, P = 0.008). There were no significant age differences in female Great Tits ( P > 0.200). Environmental effects Blue Tits did not differ significantly in any colour parameter in woods of different sizes. Male Great Tits were 15% less saturated with blue in small woods (Fig. 3, mean ± sd = 2.46 ± 0.67) than large woods (2.90 ± 0.77), but females were only 5% less saturated in these

8 Blueness of British tits 321 Figure 2. Plot of Blue Tit (a) and Great Tit (b) covert chroma against hue showing the much greater scatter of the latter, and its shift towards green in females. Dashed line, pure spectral blue.

9 322 P.N. Ferns and S.A. Hinsley Figure 3. Influence of wood size on the covert chroma of male and female Great Tits (mean + se bar, sample size above). Age groups have been combined as they did not differ significantly in chroma. two habitats (0.66 ± 0.45 and 0.70 ± 0.39 respectively). This was the consequence of a significant wood effect on chroma ( F 1,107 = 4.66, P = 0.033), but the interaction of wood with sex was not significant ( F 1,107 = 3.43, P = 0.067). Female Great Tits were significantly closer to pure spectral blue in large woods (median = ) than small woods (median = 247.3, P = 0.041). Blue Tit coverts were significantly more colourful in some years than others, birds being about 26% more saturated in 1998 than in 2000 ( F 1,219 = 3.66, P = 0.003). They were also closer to pure spectral blue in 2001 than in 1996 (15 purer in males, re-sampling, P = 0.023; and 10 in females, P = 0.040). In one year (1997), females were significantly less close to pure spectral blue in small woods (257.4 ) than in large woods (median = ) (re-sampling, P = 0.044). There were no significant year effects in Great Tits. Crown colour in breeding adults Despite the fact that the crown colours in the two species appear very different (i.e. blue in one species and black in the other), there is a close similarity in their dominant wavelengths (Tables 1 and 2 ). It may seem surprising that the black of the Great Tit crown also has its dominant wavelength in the blue part of the spectrum, but this is because relatively little light is reflected at all wavelengths. In fact its peak wavelength is just into the UV (Ferns & Hinsley 2004, Hegyi 2007). Blue Tit crowns are about three times as saturated with colour and also reflect twice as much light, as indicated by their higher lightness values (Table 1 ). Sex and age effects There was no significant difference in lightness between male and female Blue Tits, but the crowns of males were significantly more saturated with colour (51% more in large wood adults and 67% in juveniles) (C, F 1,217 = 108.7, P < ). The crown hues of male Blue Tits were closer to pure spectral blue than those of females (Table 1, re-sampling, P < ). Female Great Tits were significantly lighter than males and significantly less saturated with colour (Table 2, re-sampling, P < in both cases). They were also significantly less close to pure spectral blue ( P = 0.007).

10 Blueness of British tits 323 Adult Blue Tits had crowns that were slightly lighter (about 4%) than those of juveniles ( F 1,219 = 4.07, P = 0.045) (Table 1 ). Adult Blue Tits were significantly more saturated than juveniles, but this was entirely due to the difference in large woods where adult males were 31% more saturated than juveniles and adult females were 45% more ( F 1,217 = 9.57, P = 0.001). Environmental effects In small woods, juvenile Blue Tits of both sexes were slightly, but not significantly more saturated than adults, making them brighter than juveniles in large woods (30% more in males and 23% in females) as indicated by the significant interaction between age and wood size ( F 1,217 = 11.21, P = 0.002). Juvenile male Great Tits were more saturated with blue in large woods (median chroma = 3.18%) than in small woods (2.29%), but this difference was not quite significant ( P = 0.054). There was a significant year effect in female Blue Tit hues, which were closer to pure spectral blue (267.4 ) in 1996 and least close (260.6 ) in 2000 ( P = 0.047). There were no significant year effects in Great Tits. Longer-term age effects Age clearly influenced covert chroma in both species, but the relationship was not linear (Fig. 4 ). We therefore fitted quadratic polynomials. In the case of male Blue Tits, both linear and quadratic terms were significant ( F 1,20 = 6.33, P = 0.021, R 2 = 0.250; F 1,20 = 10.6, P = 0.004, R 2 = 0.279; respectively). In females, only the linear term was significant ( F 1,25 = 9.39, P = 0.005, R 2 = 0.159; F 1,25 = 0.74, P > 0.200, R 2 = 0.023; respectively). In male Great Tits, only the quadratic term was significant ( F 1,25 = 6.99, P = 0.018, R 2 = 0.298), the linear term in males, and both terms in females were not significant ( P < 0.200). Unlike covert chroma, the relationship between Blue Tit crown chroma and age showed no sign of departure from linearity, or of differences in slope between males and females, but a linear regression was significant ( F 1,46 = 5.46, P = 0.024, R 2 = 0.166). There was no significant relationship between crown chroma and age in Great Tits. Size effects Two significant wing length effects were found, both in juveniles. The first was an increase in male Blue Tit covert chroma of 0.48 ± 0.23% (mean ± se) ( F 1,45 = 4.42, P = 0.041, R 2 = 0.089) per additional mm of wing length. The regression line differed significantly in elevation, but not slope, from the line for females ( F 1,109 = 91.7, P < ), which itself was not significantly different from zero ( F 1,63 = 1.37, P > 0.200). The second wing length effect was a significant increase in covert hue in juvenile male and female Great Tits with increasing wing length. In the this case the slope for both males and females was significant and positive (males, slope = 0.97 ± 0.44, F 1,23 = 4.96, P = 0.036, R 2 = 0.177; females, slope = ± 4.87, F 1,31 = 5.20, P = 0.030, R 2 = 0.144) and differed between the sexes ( F 1,54 = 4.29, P = 0.043). Thus, juvenile females with the shortest and longest wings had hues that were spread out much more widely below and above pure spectral blue than those of juvenile males. There were no significant effects of wing length on the colours of adults. Identifying male and female tits by colour Regardless of age, discriminant analysis of covert lightness and chroma correctly categorized the sexes of 91.6% of Blue Tits, whilst that of all three crown colour components discriminated only 75.3%. This is because of the much greater variability and overlap in covert hue. In the case of Great Tits, 94.6% were correctly classified on the basis of covert chroma alone, or on the basis of crown lightness and chroma combined. Six out 71 Blue Tits were incorrectly sexed by eye by the author (PNF) using plumage brightness. The error rate of 8.3% is lower than expected using discriminant analysis, but this is probably because both members of the pair were caught at 15 nests and they were examined at random, but in turn (all members of such pairs were sexed correctly). The error rate excluding pairs was 14.6%. DISCUSSION Colour variation and speed of moult Why is the blue colour of the coverts and crowns of individual Blue Tits and Great Tits so variable, especially in hue? While high variability is generally characteristic of sexually selected features, and they are often sensitive to environmental influences, this is not the whole explanation (Alatalo et al. 1988, Barnard 1991). We found that the carotenoid-pigmented yellow flank feathers, also subject to sexual selection in the same sample of birds, were much less variable than their structural blue feathers (Ferns & Hinsley 2008).

11 324 P.N. Ferns and S.A. Hinsley Figure 4. Influence of age on covert chroma in (a) breeding Blue Tits and (b) breeding Great Tits. The regression lines are fitted quadratic polynomials and are statistically significant in the case of males (solid lines), but not females (dashed lines), though a linear regression is significant for female Blue Tit covert chroma.

12 Blueness of British tits 325 While the underlying evolutionary reason why particular colours are favoured by mates may be that they indicate the quality of potential sexual partners, the proximate mechanisms that make them such indicators are also of interest. The blue coloration is most variable when its chromaticity is low (<1%) and quite a few individuals, mostly female Great Tits, fall into this part of the range (Fig. 1 ). Structural blue may be more sensitive to environmental perturbation than pigment colour alone, because it depends on the presence of both spongy keratin and underlying melanin pigment for its expression (Prum et al. 1999). It is also possible that the spongy keratin responsible for blue hues may sometimes be insufficiently regular in structure (Andersson 1999), perhaps as a consequence of poor nutrition during the moult (Dawson et al. 2000), that this changes the wavelength of the reflected light in a variable fashion. The spongy keratin is located beneath the cortex in the outer parts of the medulla of the feather barbs (Auber 1957, Lucas & Stettenheim 1972, Prum et al. 1998, 1999), and so should not deteriorate in wavelength through abrasion until the whole of the dorsal part of the cortex has been worn away. Such severe wear is plausible in the case of the erectable contour feathers of the crown, which may be abraded during foraging amongst vegetation, but seems less likely for the central parts of the covert feathers that need to remain intact to minimize air leakage between the bases of the primary and secondary feather shafts during flight. The edges of the coverts do abrade, however, and repeated feather flexure during flight could also degrade the internal fine structure of the spongy layer, even in the absence of external wear, leading to changes in colour. The underlying eumelanin layer that absorbs any unreflected light is dark brown, with a hue in the region of 50 (personal observations). Only one female Great Tit had coverts with a hue in this region of the spectrum (see Fig. 2 ), and only two Blue Tits, both female, had crowns with similar hues. Contrary to Prum s (2006) suggestion, lightness and saturation (chroma) were less susceptible to environmental perturbation than hue in our laboratory Blue Tits, since the former colour parameters did not differ significantly in birds induced to moult at different speeds. This difference in sensitivity is not surprising given the wide variation in hue recorded in the field, and its susceptibility to habitat and year effects. Birds that moulted in the field were significantly less close to pure spectral blue than slow moulters in the laboratory, but not significantly closer than fast moulters. This is probably due to the fact that they were measured much later in the season (at least six months later), rather than immediately after the moult, by which time their plumage would have become worn and faded (Örnborg et al. 2002, Delhey et al. 2006). The moult of our short-daylength birds was accelerated rather more than we had originally intended (Ferns & Hinsley 2008), and so they ought to have been less close to pure spectral blue than most individuals from the field, as indeed they were. This is despite the fact that females moult later than males in the field (Flegg & Cox 1969, Ginn & Melville 1983), and thus moult faster. The results for our captive Blue Tits are also somewhat different from those of Keyser & Hill s (1999) for male Blue Grosbeaks in that there was no significant difference in the general brightness of fast moulters, but they did differ significantly in hue. However, the grosbeaks were assumed to moult faster because of superior condition, whereas ours were forced to moult faster by manipulating daylength which is known to lower feather quality (Dawson et al. 2000). Factors affecting colour variation in the field The fact that Blue Tits are bluer than Great Tits is obvious, but the fact that they are only three times as saturated with blue, and that the hues of the two species are so similar, is perhaps less expected. It is also no surprise that in Blue Tits, males are distinctly bluer on average than females, and that adults are usually slightly bluer than juveniles. This is both because females and juveniles are less saturated with colour, and because females have more variable hues than males, and are significantly lighter, thus appearing much paler. It is much less obvious to the casual observer that very similar trends are present in Great Tits. Our observations provide the first quantification of these trends, which are well established in subjective descriptions in the literature, for example, Perrins (1979), Cramp & Perrins (1993) and Gosler (1993). For example, even the slight greenish tinge to Great Tits coverts is accurately described by Cramp & Perrins (1993). Why are females so much more variable in colour, especially hue, than males? It is partly a consequence of their tendency to moult later than males, and partly because the duration of their moult is more variable (Dhondt, 1973, 1981, Cramp & Perrins 1993, Bojarinova et al. 1999). These proximate factors may be a consequence of their greater investment in parental care, which ultimately, and almost inevitably, lead to a reduced investment in sexually selected characters.

13 326 P.N. Ferns and S.A. Hinsley This, and a number of other possibilities, is discussed by Andersson (1994). Our results also quantify for the first time the more intense blue coloration of older birds of both species referred to by Perrins (1979). Furthermore, the quadratic relationships (Fig. 4 ) show that chroma deteriorated in birds which survived more than three years. Similar declines in the territory size, breeding performance, wing length and weight of older birds have been attributed to senility (van Balen 1967, Perrins 1979, Hudde 1985), and reproductive senescence in individual female Great Tits has been observed to start at the age of three (Bouwhuis et al. 2009). Increases in the UV coloration of Blue Tit crowns have also been observed between juvenile and adult stages (Delhey & Kempenaers 2006, Korsten et al. 2007). One of our most interesting findings is that the colour of Blue Tits is not as sensitive to habitat differences as that of Great Tits. Thus male Great Tit coverts were less saturated with blue in small woods and those of females were less close to pure spectral blue, whereas Blue Tits showed no colour differences in woods of different sizes. This difference may reflect differences in the size of the two species and may be related to a more marked dispersal tendency in Blue Tits (Matthysen et al. 2005). The breeding and moulting performance of Blue Tits is probably less compromised in our small fragmented woodlands because of their ability to exploit smaller prey, and to use multiple habitat patches more efficiently (Hinsley et al. 1999, Hinsley et al. 2008). On the other hand, Blue Tit colour is more sensitive to between-year variation in environmental conditions than that of Great Tits, perhaps because the latter species is dominant at feeding sites. This potential effect of competition may be most significant in years when moult occurs relatively late and hence is faster. Thus, the only significant habitat effect we recorded in Blue Tits was also a year effect. When captured at the nest in 1997, female Blue Tits in small woods had lower covert hues than birds in large woods. The previous breeding season of 1996 was the latest in 17 years of observations since 1993 in our study area (Hinsley et al. unpubl. data) and therefore the Blue Tits measured in the 1997 breeding season had probably moulted late. In contrast, Blue Tits coverts were most saturated with blue in 1998, the year following the earliest start to breeding (and therefore probably also the earliest moult) during the same period. Although we found no effect of moult speed on the chroma of laboratory birds, this suggests that such an effect might exist in the field (Hinsley et al. 2003). If wing length is inversely related to juvenile hatching date, and early hatching individuals are of better quality (Lack 1966, Perrins 1979, Garnett 1981, Smith et al. 1989), the fact that we found significant relationships between wing length and colour indicates that covert chroma might provide a clue to quality in juvenile male Blue Tits, and covert hue in both sexes of juvenile Great Tits. This could be especially important if birds are capable of identifying whether potential partners are juvenile or not. The structural blue and ultraviolet colours of Blue Tits have been measured in numerous studies (Andersson et al. 1998, Hunt et al. 1998, 1999, Sheldon et al. 1999, Örnborg et al. 2002, Griffith et al. 2003, Johnsen et al. 2003, 2005, Alonso-Alvarez et al. 2004, Delhey et al. 2006, Delhey & Kempenaers 2006, Korsten et al. 2007, Szigeti et al. 2007), and those of Great Tits, in at least one (Hegyi et al. 2007). Differences in the areas of plumage measured and the units in which the measurements are expressed make comparisons difficult, but one factor that emerges from these studies is a tendency for the wavelength component of the UV to respond most to environmental factors. Our results suggest the existence of a similar tendency within the violet and blue range visible to humans, in that hue was the most variable colour character we measured, and the most sensitive to environmental conditions. Peak reflectance of Blue Tit crown feathers is in any case partly within the human visible range during the breeding season (Örnborg et al. 2002, Delhey et al. 2006). The study most directly comparable to ours is that of Figuerola et al. (1999), who measured tit colour in evergreen oak and mixed oak and pine forest near Barcelona, Spain, an area where Cy. c. caeruleus begins to grade into the slightly darker and bluer Cy. c. ogliastrae (Kvist et al. 1999, Harrap & Quinn 1996). The repeatabilities of their three replicate measurements were lower than those of our six, and thus they were unable to analyse Blue Tit crown lightness, or wing and crown hues. The standard deviations of their measurements within habitats were lower than ours, but this reflects the fact that they found more significant habitat differences and therefore could not combine habitats as we have done in Tables 1 and 2. Despite the possible influence of nearby Cy. c. ogliastrae, the Blue Tits measured in Spain were much lighter than our British Cy. c. obscurus, their coverts being about 30% more reflective across all age and sex groups (e.g. our adult males = 27.7 ± 1.4; Spanish adult males in oak woodland = 37.5 ± 0.9, in mixed

14 Blueness of British tits 327 woodland = 37.0 ± 1.1). Thus, less than 1% of the two populations showed any overlap in lightness. Males in Britain were more saturated with blue in all habitats (e.g. chroma of our adult males = 8.77 ± 2.11; Spanish adult males in oak woodland = 7.68 ± 0.97, in mixed woodland = 5.91 ± 0.66), but in this case about 65% of the individuals in the two bluest populations overlapped. Moreover, the females in Spain were bluer than those in Britain, at least in the best habitat (e.g. chroma of our adult females = 4.07 ± 1.67; Spanish adult females in oak woodland = 4.68 ± 0.71, in mixed woodland = 2.97 ± 0.75). The crowns of British Blue Tits were bluer in all cases except for juvenile females. These trends are broadly in accord with the established differences between Cy. c. obscurus from the British Isles and Cy. c. caeruleus from the mainland of Europe (Pražak 1894, Cramp & Perrins 1993, Harrap & Quinn 1996). The best distinguishing feature is the overall darkness of British Blue Tits (i.e. their reduced lightness and higher chroma). This applies even though there is generally considered to be some admixture between these two races in parts of southern and eastern England (Clancey 1947). Sexing by discriminant analysis of the blue colour parameters of autumn moulted coverts was, surprisingly, slightly more efficient for Great Tits than for Blue Tits. Dhondt (1970) and Perrins (1979) pointed out that some young Great Tits may even be sexable in the nest on the basis of covert colour. The fact that Blue Tit crowns were less efficient than wing coverts for distinguishing the sexes probably reflects the greater exposure of the crown feathers to wear by the time of the breeding season, and the fact that differences extend into the UV. Thus, Korsten et al. (2007) were able to successfully identify the sex of 96% of Blue Tits by discriminant analysis of all three crown colour parameters (including the UV down to 320 nm), measured between November and January in Holland. This compares with our 75% (down to 400 nm) between April and June. Human observers do differ in the extent to which they can discriminate the blue coloration (Harper 2000), but appear to be able to sex Blue Tits almost as well as discriminate functions based on colorimeter measurements, and they are certainly capable of sexing the majority of individuals correctly. ACKNOWLEDGEMENTS We would like to thank A. Dawson for advice, two referees for many helpful comments, and P.E. Bellamy for maintaining the ringing database; as well as the private landowners and managers who granted us access to their woods to collect colour measurements. REFERENCES Alatalo, R.V., Höglund, J. & Lundberg, A Sexual selection models and patterns of variation in tail ornaments in birds. Biol. J. Linn. Soc. 34: Alonso-Alvarez, D., Doutrelant, C. & Sorci, G Ultraviolet reflectance affects male male interactions in the Blue Tit ( Parus caeruleus ultramarinus ). Behav. Ecol. 15: Andersson, M Sexual Selection. Princeton, NJ: Princeton University Press. Andersson, S Morphology of UV reflectance in a whistlingthrush: implications for the study of structural colour signaling in birds. J. Avian. Biol. 30: Andersson, S., Örnborg, J. & Andersson, M Ultraviolet sexual dimorphism and assortative mating in Blue Tits. Proc. R. Soc. Lond. B 265 : Auber, L The structures producing non-iridescent blue colour in bird feathers. Proc. Zool. Soc. Lond. 129: Barnard, P Ornament and body size variation and their measurement in natural populations. Biol. J. Linn. Soc. 42: Bojarinova, J.G., Lehikoinen, E. & Eeva, T Dependence of postjuvenile moult on hatching date, condition and sex in the Great Tit. J. Avian Biol. 30: Bouwhuis, S., Sheldon, B.C., Verhulst, S. & Charmantier, A Great tits growing old: selective disappearance and the partitioning of senescence to stages within the breeding cycle. Proc. R. Soc. Lond. B 276: Chen, D.M, Collins, J.S. & Goldsmith, T.H The ultraviolet receptor of bird retinas. Science 225: Clancey, P.A Parus caeruleus obscurus Prazak Observations on a comprehensive series from Great Britain. Ibis, 89: Cramp, S. & Perrins, C.M. (eds) Birds of the Western Palearctic. Vol. VII. Oxford: Oxford University Press. Cuthill, I. C Color perception. In Hill, G.E. & McGraw, K. (eds) Bird Coloration. Vol. 1. Mechanisms and Measurements : Cambridge, MA: Harvard University Press. Dawson, A., Hinsley, S.A., Ferns, P.N., Bonser, R.H.C. & Eccleston, L Rate of moult affects feather quality: a mechanism linking current reproductive effort to future survival. Proc. Roy. Soc. Lond. B 267: Delhey, K. & Kempenaers, B Age differences in Blue Tit Parus caeruleus plumage colour: within-individual changes or colourbiased survival? J. Avian Biol. 37: Delhey, K., Peters, A., Johnsen, A. & Kempenaers, B Seasonal changes in Blue Tit crown color: do they signal individual quality? Behav. Ecol. 17: Dhondt, A.A The sex ratio of nestling Great Tits. Bird Study 17: Dhondt, A.A Postjuvenile and postnuptial moult in a Belgian population of Great Tits, Parus major, with some data on captive birds. Gerfaut 63: Dhondt, A.A Postnuptial moult of the Great Tit in southern Sweden. Ornis Scand. 12: Ferns, P.N. and Hinsley, S.A Immaculate tits: head plumage pattern as an indicator of quality in birds. Animal Behaviour 67:

15 328 P.N. Ferns and S.A. Hinsley Ferns, P.N. & Hinsley, S.A Carotenoid plumage hue and chroma signal different aspects of individual and habitat quality in tits. Ibis 150: Figuerola, J. & Senar, J.C Seasonal changes in carotenoidand melanin-based plumage coloration in the Great Tit Parus major. Ibis 147: Figuerola, J., Senar, J.C. & Pascual, J The use of a colorimeter in field studies of Blue Tit Parus caeruleus coloration. Ardea 87: Fitzpatrick, S Colour schemes for birds: structural coloration and signals of quality in feathers. Ann. Zool. Fenn. 35: Flegg, J.J.M. & Cox, C.J The moult of British Blue Tit and Great Tit populations. Bird Study 16: Garnett, M.C Body size, its heritability and influence on juvenile survival among Great Tits Parus major. Ibis 123: Ginn, H.B. & Melville, D.S Moult in Birds. British Trust for Ornithology, Tring, UK. Gosler, A The Great Tit. Hamlyn, London. Griffith, S.C., Örnborg, J., Russell, A.F., Andersson, S. & Sheldon, B.C Correlations between ultraviolet colouration, overwinter survival and offspring sex ratio in the Blue Tit. J. Evol. Biol. 16: Guilford, T. & Harvey, P.H The purple patch. Nature 392: Harper, D.G.C Sexing Blue Tits Parus caeruleus during the winter, with comments on reporting sexing criteria. Ringing & Migration 20: Harrap, S. & Quinn, D Tits, Nuthatches and Treecreepers. Christopher Helm, London. Hart, N.S Variations in cone photoreceptor abundance and the visual ecology of birds. J. Comp. Physiol. A 187: Hegyi, G., Szigeti, B., Török, J. & Eens, M Melanin, carotenoid and structural plumage ornaments: information content and role in great tit Parus major. J. Avian Biol. 38: Hill, G.E Environmental regulation of ornamental coloration. In Hill, G.E. & McGraw, K.J. (eds) Bird Coloration, Vol. 1. Mechanisms and Measurements : Harvard University Press, Cambridge, MA. Hill, G.E. & McGraw, K.J. (eds) Bird Coloration, Vol. 2. Function and Evolution. Harvard University Press, Cambridge, MA. Hinsley, S.A., Rothery, P. & Bellamy, P Influence of woodland area on breeding success in Great Tits Parus major and Blue Tits Parus caeruleus. J. Avian Biol. 30: Hinsley, S.A., Rothery, P., Ferns, P.N., Bellamy, P.E. & Dawson, A Wood size and timing of moult in birds: potential consequences for plumage quality and bird survival. Ibis 145: Hinsley, S.A., Hill, R.A., Bellamy, P.E., Harrison, N.M., Speakman, J.R., Wilson, A.K. & Ferns, P.N Effects of structural and functional habitat gaps on breeding woodland birds: working harder for less. Landscape Ecol. 23: Honkevaara, J., Koivula, M., Korpimäki, E., Siitari, H. & Viitali, J Ultraviolet vision and foraging in terrestrial vertebrates. Oikos 98: Hudde, V.H Individuelle Unterschiede im Längenwachstum Flügel bei Kohl-, Blau-, Sumpf- und Weidenmeise ( Parus major, P. caeruleus, P. palustris und P. montanus ). Vogelwelt 106: Hunt, S., Bennett, A.T.D., Cuthill, I.C., & Griffiths, R Blue Tits are ultraviolet tits. Proc. Roy. Soc. Lond. B 265: Hunt, S., Cuthill, I.C., Bennett, A.T.D. & Griffiths, R Preferences for ultraviolet partners in the Blue Tit. Anim. Behav. 58: Huth, H.H. & Burkhardt, D Der Spektrale Sehbereich eine Violettohr-Kohlibris. Naturwissenschaften 59: 650. Johnsen, A., Delhey, K., Andersson, S. & Kempenaers, B Plumage colour in nestling Blue Tits: sexual dichromatism, condition dependence and genetic effects. Proc. Roy. Soc. Lond. B 270: Johnsen, A., Delhey, K., Schlicht, E, Peters, A. & Kempenaers, B Male sexual attractiveness and parental effort in Blue Tits: an experimental test of the differential allocation hypothesis. Anim. Behav. 70: Kempenaers, B., Verheyen, G.R. & Dhondt A.A Extrapair paternity in the Blue Tit ( Parus caeruleus ): female choice, male characteristics and offspring quality. Behav. Ecol. 8: Keyser, A. & Hill, G.E Condition-dependent variation in the blue-ultraviolet coloration of a structurally based plumage ornament. Proc. Roy. Soc. Lond. B 266: Korsten, P., Vedder, O., Szentirmai, I. & Komdeur, J Absence of status signaling by structurally based ultraviolet plumage in wintering Blue Tits ( Cyanistes caeruleus ). Behav. Ecol. Sociobiol. 61: Kvist, L., Ruokonen, M., Lumme, J. & Orell, M Different population structures in northern and southern populations of the European Blue Tit ( Parus caeruleus ). J. Evol. Biol. 12: Lack, D Population Studies of Birds. Clarendon Press, Oxford, UK. Lucas, A.M. & Stettenhein, P.R Avian Anatomy Integument, Part II. Government Printing Office, Washington, DC. Matthysen, E., Van de Casteele T. & Adriaensen, F Do sibling tits ( Parus major and P. caeruleus ) disperse over similar distances and in similar directions? Oecologia 143: McGraw, K.E., Masckillop, E.A., Dale, J. & Hauber, M.E Different colors reveal different information: how nutritional stress affects the expression of melanin- and structurally based ornamental plumage. J. Exp. Biol. 205: Nakagawa, S A farewell to Bonferroni: the problems of low statistical power and publication bias. Behav. Ecol. 15: Örnborg, J., Andersson, S., Griffith, S.C. & Sheldon, B.C Seasonal changes in a ultraviolet structural colour signal in Blue Tits, Parus caeruleus. Biol. J. Linn. Soc. 76: Perrins, C.M British Tits. Collins, London. Pražak, J.P Einige Notizen über die Kohl-, Hauben- und Blaumeisen. Ornithol. Jahrb. 5: Prum, R.O Anatomy, physics, and evolution of structural colors. In Hill, G.E. & McGraw, K.J. (eds) Bird Coloration, Vol. 1. Mechanisms and Measurements : Harvard University Press, Cambridge, MA. Prum, R.O., Torres, R.H., Williamson, S. & Dyck, J Coherent light scattering by feather barbs. Nature 396: Prum, R.O., Torres, R.H., Williamson, S. & Dyck, J Two dimensional Fourier analysis of the spongy medullary keratin of structurally coloured feather barbs. Proc. Roy. Soc. Lond. B 266: Qvist, L., Ruokonen, M., Lumme, J. & Orell, M Different population structure in northern and southern populations of the European Blue Tit ( Parus caeruleus ). J. Evol. Biol. 12: Scott, G.W Sexing members of a Scottish Blue Tit Parus caeruleus population in the hand during the winter months. Ring. Migr. 14: Sheldon, B.C, Andersson, S., Griffith, S.C., Örnborg, J. & Sendecka, J Ultraviolet colour variation influences Blue Tit sex ratios. Nature 402: