Is the structural and psittacofulvin-based coloration of wild burrowing parrots Cyanoliseus patagonus condition dependent?

J. Avian Biol. 39: 653662, 2008 doi: 10.1111/j.1600-048X.2008.04417.x, # 2008 The Authors. J. Compilation # 2008 J. Avian Biol. Received 28 October 2007, accepted 12 March 2008 Is the structural and psittacofulvin-based coloration of wild burrowing parrots Cyanoliseus patagonus condition dependent? Juan F. Masello, Thomas Lubjuhn and Petra Quillfeldt J. F. Masello (correspondence), School of Biological Sciences, University of Bristol, UK. Present address: Max Planck Institute for Ornithology, Vogelwarte Radolfzell, Schlossallee 2, D-78315 Radolfzell, Germany. E-mail: masello@orn.mpg.de. T. Lubjuhn, Institut für Evolutionsbiologie und O kologie, Rheinische Friedrich-Wilhelms-Universität Bonn, Germany. P. Quillfeldt, Max Planck Institute for Ornithology, Vogelwarte Radolfzell, Germany. The bright colours of parrots are caused by psittacofulvin pigments, which appear unique to this Order, and by structural colours. We measured red (psittacofulvin), green (mixed) and blue (structural) colours of wild burrowing parrots Cyanoliseus patagonus of northeastern Patagonia, Argentina, and measured nestlings regularly to obtain data on breeding success and nestling growth. As adult feathers are moulted outside the breeding season, adult body condition could not be measured directly during feather growth, and climatic conditions were used as an indirect parameter. The colony of burrowing parrots is surrounded by Monte steppe habitat, the breeding success has been shown to depend strongly on the climatic patterns. The area experienced a drought with very poor breeding success as well as a year of above-average precipitation during the study period, serving as a natural experiment. We thus analysed the variability of colouration within the population among and within breeding seasons. We observed strong inter-annual differences in nestling and adult colouration. Nestlings grew blue feathers with lower achromatic brightness during better conditions, and when controlling for year effects, nestlings with higher mass and from more successful families also had blue feathers with lower achromatic brightness. Adult blue feathers showed the same trend, with lower achromatic brightness in the moult following breeding seasons of better conditions. In contrast, during better conditions, adults grew red feathers with higher achromatic brightness and the colour hue was also affected, and the hue of the red region of nestlings varied with the hatching order. The colour of all three regions of nestlings varied between nests, and the colour of the red region of adult males positively correlated with breeding success (clutch size, brood size). In summary, the present data suggest that environmental conditions contribute to variability in both structural and the psittacofulvin-based colours of wild burrowing parrots. Parrots (Aves, Psittaciformes) have bright red-to-yellow hues in their feathers. Unlike most bird species, that deposit dietary or modified carotenoids to produce these colours, parrots use lipochromes (Krukenberg 1882, Völker 1936, 1937, 1942) that are not acquired from the diet (Völker 1936, McGraw and Nogare 2005). Recently, the biochemical identity, structure and distribution of the polyenal lipochromes of parrots, called psittacofulvins (Krukenberg 1882), has been described (Veronelli et al. 1995, Stradi et al. 2001, Morelli et al. 2003, McGraw and Nogare 2005). In a study across 27 parrot genera, McGraw and Nogare (2005) found that the 5 known psittacofulvins occurred in all species they examined. Despite the absence of carotenoids in their feathers (see Völker 1936, 1937, 1942, Stradi et al. 2001, McGraw and Nogare 2004), parrots have both dietary and metabolically derived carotenoids in the blood at the time of feather growth (McGraw and Nogare 2004, 2005), at levels comparable to those found in many other carotenoid-coloured birds, suggesting that parrots avoid depositing carotenoids in growing feathers (McGraw and Nogare 2004, 2005). Because psittacofulvins were not detected in parrot blood at the time of feather growth (McGraw and Nogare 2004), it was concluded that these pigments are locally synthesized by growing feathers. Red parrot feathers show a similar reflectance curve to that of carotenoid-based reds in other birds (see Burkhardt 1989, Bennett et al. 1996, Hunt et al. 2003), with predominant reflection in the red portion of the spectrum, and a smaller UV reflectance peak (e.g. grey parrot Psittacus erithacus, chestnut-fronted macaw Ara severa, scarlet macaw Ara macao, white-fronted amazon Amazona albifrons in Burkhardt 1989; red-tailed black cockatoo Calyptorhynchus banksii, red lory Eos bornea, and grey parrot in McGraw and Nogare 2005). Across species, McGraw and Nogare (2005) found that total concentration of psittacofulvins present in red parrot feathers was correlated with the hue and saturation. Parrots exhibit also very conspicuous blue structural colours produced by feather nanostructures and greens, which are a combination of structural colour and pigments in particular arrays (e.g. Dyck 1971a,b, 1992, Finger et al. 1992, Prum et al. 2003). Studies over the last decade 653

highlighted the importance of such UV-blue structural colours in sexual signalling in birds (e.g. Andersson et al. 1998, Doucet and Montgomerie 2003, Shawkey and Hill 2005). It has been hypothesized that psittacofulvins might be costly to produce (Masello et al. 2004, McGraw and Nogare 2005). Only recently, condition dependence of these unusual pigments has started to be investigated (Masello et al. 2004). A relationship between condition and the expression of structural plumage colouration has also been suggested by recent studies (e.g. Andersson 1999, Keyser and Hill 1999, Doucet 2002), and may be based on the ontogenetic costs of the formation of a regular spongy structure that is responsible for the coherent scattering of short wavelengths (e.g. Dyck 1971a, Andersson 1999, Prum et al. 2003, Shawkey et al. 2003). In Psittaciformes, structural colours have received little attention so far. Burrowing parrots Cyanoliseus patagonus are colonial Neotropical Psittaciformes. Burrowing parrots inhabit scrubland and grassland of Argentina and Chile, requiring cliffs where to excavate their nest burrows (e.g. Masello et al. 2006a). Each burrow is occupied by a single pair that lays one clutch of two to five eggs per year (Masello and Quillfeldt 2002, 2004a). Burrowing parrots have a socially and genetically monogamous breeding system with intensive biparental care (see Masello et al. 2002, 2006a, Masello and Quillfeldt 2002, 2003, 2004b). Burrowing parrots show slight size dimorphism, males being larger (about 5%) than females (Masello and Quillfeldt 2003). HPLC confirmed the absence of carotenoids in burrowing parrots feathers (J. F. Masello, P. Quillfeldt and K. Gruenewald unpubl. data). Instead, adults, juveniles and nestlings in the last weeks of the nestling period, have a bright psittacofulvin-based (R Stradi pers. comm.) red abdominal patch in the centre of the abdominal region (see Masello and Quillfeldt 2003, 2004b, Masello et al. 2004). Adult males have larger abdominal red patches than females (Masello and Quillfeldt 2003, Masello et al. 2004). Although not obviously different in colouration to humans, differences in hue of the red abdominal patch between male and female burrowing parrots were detected using numeric measures of hue, value, and chroma (Masello et al. 2004). It was found that the colouration of the red patch of burrowing parrots is a good predictor of female body condition and male size (Masello et al. 2004). These previous studies suggested that the red colouration of the abdominal patch acts as a signal of individual condition, quality and parental investment (see Masello and Quillfeldt 2004b, Masello et al. 2004). However, those studies used methods based on the humanvisible spectrum (400700 nm), while birds can see UVA wavelengths (320400 nm), and will perceive colours differently to humans so they may be using visible features that humans cannot perceive (e.g. Bennett et al. 1994, Cuthill et al. 2000). This suggestion is supported by the use of UVA wavelengths in mate choice by many birds species so far tested (e.g. Bennett et al. 1996, 1997), including the only parrot so far tested, the budgerigar (Pearn et al. 2001). Burrowing parrots have a conspicuous blue colouration in the wings (primaries P1 to P10, secondaries S1 to S5 sometimes also S6, the alular remiges, the dorsal major primary and secondary coverts). Because of its pronounced visibility, particularly during flight, this bright blue colour could play a role in condition signalling. Here, we report a spectrometric study of structural and psittacofulvin-based ornamentation of wild burrowing parrots of north-eastern Patagonia, Argentina. Specifically, we test whether the variability of colouration within the population reflect the conditions that nestlings and adults experience during feather growth. Methods The study was carried out during four breeding seasons (Oct. 1998 to Feb. 1999, Nov. 1999 to Jan. 2000, Nov. 2001 to Jan. 2002, and Nov. 2003 to Jan. 2004) at the largest colony of burrowing parrots located in a cliff facing the Atlantic Ocean in north-eastern Patagonia, Argentina (Masello et al. 2006a). The habitat in the surroundings of the colony corresponds to the north-eastern Patagonian sector of the phytogeographical province of Monte, characterized by bushy steppes and xerophytes forests. According to accessibility, 79 nests were closely monitored in one sector of the colony as part of an ongoing study of the breeding behaviour of the species (e.g. Lubjuhn et al. 2002, Masello et al. 2006a and references therein, Masello et al. 2006b). Nests were inspected every five days by climbing the cliff face. As in previous studies (e.g. Masello and Quillfeldt 2002, 2003, 2004a,b), the following parameters of breeding success were recorded: (1) clutch size, the number of eggs laid per nest, (2) number of eggs hatched, (3) brood size, at the time of hatching of the last nestling and 30 d after, and (4) number of fledglings. Adult burrowing parrots were captured in their nests only during the breeding season and while attending nestlings. As in earlier studies (e.g. Masello and Quillfeldt 2002, 2003, 2004a,b), the following morphometric parameters of the attending adults and the nestlings were recorded each time the nest was inspected: (1) body mass, using a digital balance to the nearest 1 g, (2) wing length, the distance from the anterior surface of the radio carpal joint to the tip of the longest primary, using a wing rule to the nearest 1 mm, and (3) length of the internal tail feather using a feather rule to the nearest 1 mm. At the time of the first measurement, when nestlings were still clearly different sizes, the hatching rank was determined, and nestlings were individually marked. Mass growth in nestling burrowing parrots follows a quadratic regression reaching a peak mass around a mean age of 38 d, which is followed by a large mass recession (Masello and Quillfeldt 2002). Burrowing parrots have a red feather patch in the centre of the abdominal region of the coelom. Using a sheet of transparent plastic the outline of the red patch of adults were copied in order to calculate the surface area, length, and width of the patch. The red patch was copied the first time the adults were captured in each year. Birds were sexed using PCR amplification of a highly conserved W-linked gene as previously described (Lubjuhn and Sauer 1999, Masello and Quillfeldt 2004b). Blind duplicate and triplicate blood samples of burrowing parrots were analysed in order to ensure the accuracy of the gender 654

determination. In all cases duplicates and triplicates confirmed the results. Burrowing parrots tend to desert their nests in response of disturbance during incubation and the first week after hatching (Masello et al. 2002). In order to reduce observer influence, nests were not disturbed until about five days after the estimated hatching date of the last nestling of a clutch. Blood sampling had no detectable adverse effects on the birds. After measuring and sampling the birds were released in their burrows. No desertion occurred. The number of fledglings and the pre-fledgling nestling size of handled nests were within previously reported ranges (Masello and Quillfeldt 2002, 2003, 2004a). Feather sampling, colour measurements and analysis The first time the adults were captured in the nest one feather from the centre of the abdominal red patch and the green-blue fourth secondary covert of the right wing were sampled for further analyses. All adult feathers were sampled during December. In the case of nestlings, sampling of the feathers was partial during the first years of our study: in the breeding season 1998 only secondary coverts were sampled, in the breeding season 19992000 only red feathers were samples, while in the rest of the years complete sets of samples were collected. Nestling feathers were sampled before they fledged, i.e. in most cases during the first week of January. As in early studies of feathers (e.g. Cuthill et al. 1999, Langmore and Bennett 1999, Quesada and Senar 2006), we sampled feathers that were representative of the colour of the rest of the patch. The feathers of the abdominal red patch were selected for colour analyses following previous studies (see Masello and Quillfeldt 2003, 2004b, Masello et al. 2004) that revealed this patch as a conspicuous secondary sexual character signalling individual quality and parental investment. Likewise, the blue colouration of the wing feathers is very conspicuous in burrowing parrots, particularly during flight. As in earlier studies (e.g. Mays et al. 2004, Moreno et al. 2007), just one representative feather of the colour was sampled in order to reduce disturbance to the birds: a) the green-blue secondary coverts are the only feathers with blue colouration that can be easily collected without affecting flight capability, and b) sampling of feathers of the abdominal red patch can affect the thermoregulation capability of the adult birds, which brood their nestlings overnight during the entire nestling period (see Masello et al. 2006a). The secondary coverts of burrowing parrots are moulted after the end of the breeding season (January; JFM and PQ pers. obs.). In contrast to the wing feathers, the red abdominal feathers are not moulted during the breeding season (JFM and PQ pers. obs.), but according to aviculturists they are moulted following wing moult (C. Doty and D. Willis pers. comm.). Thus, the feathers sampled for our study grew 9 to 11 months before sampling. For adults, feathers may change their quality as the birds move in and out of the sandy burrows. We expect the effects of this abrasion effect to be homogeneous among sampled feathers, as all studied burrows belong to the same geological formation (see Masello et al. 2006a) and thus, have similar physical characteristics (e.g. sand of similar granulometry, humidity). Feather reflectance was measured at the University of Bristol following earlier developed procedures at the School of Biological Sciences (e.g. Bennett et al. 1996, Langmore and Bennett 1999, Pearn et al. 2001). The distal region of the red feathers of the abdominal patch of burrowing parrots is red, while the medial region is increasingly yellow and the basal region is greyish, so four measurements were taken only in the exposed part of the centre of the red region (referred to throughout this paper as the red region ). The outer web of the green-blue fourth secondary coverts has a distal green region and a blue basal region (referred to throughout this paper as the green region and the blue region ). Four measurements (each from a 2 mm diameter spot) were taken within the exposed part of each region. Avoiding the overlap of feather barbs, all feathers were carefully mounted on black velvet during measurement to eliminate stray reflections. Within feathers, regions were randomly allocated for spectrometric measurements over time, and feathers from each individual were allocated over time in a randomised block design (see Bennett et al. 1997). Feathers were illuminated from the proximal end, at 458 to the surface, using a Zeiss CLX 500 Xenon lamp. Reflected light was collected at 908 to the surface, using a Zeiss GK21 goniometer and the spectrum determined with a Zeiss MC 500 UV-VIS spectrometer. Reflectance was measured relative to a 99% Spectralon TM white standard, at a wavelength range of 300700 nm. White references were taken between each region and between each bird, and the reflectance standard was crosschecked against a virgin standard prior to the study. Dark references were performed before each sample. We restricted spectral analyses to wavelengths from 300700 nm, as most birds are sensitive to ultraviolet UVA wavelengths and 700 nm is likely the upper limit of the vertebrate visual spectrum (Jacobs 1981, Bennett et al. 1994). Measurements were done blind to the sex and age (adult or nestling) of the individual. Statistical procedures Unrotated principal components analyses (PCA; see Sokal and Rohlf 1994) on reflectance spectra data at 2.43 nm intervals were performed separately for each feather region (Fig. 1). Several authors (e.g. Endler 1990, Bennett et al. 1997, Cuthill et al. 1999) recommended PCA analysis of reflectance spectra as an objective way of describing variation in reflectance due to the fact that principal components are independent of the visual system of humans. The PCA extracts three principal components for each colour region and the corresponding PCA-scores for each individual, called hereafter PC1, PC2 and PC3. The first principal component is essentially flat and therefore describes variation in mean reflectance (also termed brightness or achromatic variation, sensu Endler 1990, Bennett et al. 1997, Cuthill et al. 1999). The first principal component by this definition explains most of the betweenspectra variation (cf. Endler 1990, Endler and Théry 1996, Bennett et al. 1997). As in other studies, the second and third principal components represent variation in spectral shape, i.e. chromatic variation and are therefore indirectly 655

10 (a) Green d) Green 8 0.8 0.4 6 0.0 4-0.4 2 8.0 (b) Blue e) Blue Reflectance (%) 7.5 7.0 6.5 6.0 PC coefficients 0.8 0.4 0.0 5.5-0.4 12 (c) Red 0.8 e) Red 10 8 6 4 females male 0.4 0.0 PS1 PS2 PS3 2-0.4 300 400 500 600 700 300 400 500 600 700 Wavelength (nm) Figure 1. Plumage characteristics for three feather regions of adult burrowing parrots. a) and d) corresponds to the green region of the outer web of the green-blue fourth secondary covert of the right wing, b) and e) corresponds to the blue region of the outer web of the green-blue fourth secondary covert of the right wing, and c) and f) corresponds to the red region of the feathers of the centre of the abdominal red patch. Mean reflectance values, plotted every 22.3 nm, are shown to enhance readability (a, b, and c). Each mean and SE correspond to 89 females and 83 males for the green and the blue regions, and 86 females and 85 males for the red region, from all years of the study. Black indicates females, while grey indicates males. d), e) and f) illustrate the PC coefficients (PC) plotted against wavelength, shown separately for PC1 (indicated by a continuous line), PC2 (indicated by a dotted line) and PC3 (indicated by a broken line) related to hue and saturation (e.g. Endler and Théry 1996, Bennett et al. 1997, Cuthill et al. 1999). As body mass is partly the result of structural body size and does not necessarily reflect the quantity of body reserves, an index of nestling body condition was calculated as previously described in Masello and Quillfeldt (2004b). Also for adult we scaled body mass to body size to obtain a condition index as previously described in Masello and Quillfeldt (2003). Data were analysed using SPSS 11.0. Throughout the following analyses, we used only data obtained for the first time for each bird or each breeding pair. Only data from one brood per breeding pair was considered in the analyses. Note that sample sizes for different analyses and figures varied as not all parameters of breeding success and/or nestling growth could be taken on all coloured measured birds. This was caused by breeding pairs moving to inaccessible parts of the burrow, swapping of families to adjacent inaccessible nest chambers (as some nests have more than one or interconnect to neighbouring nests, see Masello et al. 2006a) and harsh weather conditions. For some of the analyses t-values are given in addition to F- values in order to show the direction of the studied relationships. As a measure of effect sizes we included partial eta-squared values (h 2 ) i.e. the proportion of the effecterror variance that is attributable to the effect. The sums of the h 2 values are not additive (e.g. http:// web.uccs.edu/lbecker/spss/glm_effectsize.htm). In order to control for multiple testing, we provide P-values corrected for the number of tests (P corr ) in addition to 656

uncorrected P-values. We calculated corrected p-values using the equation P corr 1(1a?) k, a conversion of the Dunn-Šidák equation (Sokal and Rohlf 1994), in which a equals the originally derived P-value, and k equals the number of tests. In all cases where general linear models (GLM) were performed, covariates were always entered in separate models. In those cases where homogeneity of variance was not achieved according to Levene s test, we followed the suggestion of Zöfel (2002) and considered the test results only significant when P B0.01. In nestlings, we decided to pool both sexes together, because a multivariate analysis of nestling colouration revealed a sexual difference only for one of the nine colour parameters, namely the PC1 of the green region (green region of feathers of 228 nestlings, Wilks s l0.892, P B 0.001, PC1 F 1, 226 25.65, PB0.001). The high Wilks s l suggests a very small difference and the range of PC1 values for the green region overlapped entirely. The same analysis showed no differences among the sexes for the blue and red regions. To test whether nestling colouration depends on nestling condition, we first carried out GLM including year and nest as factors and nestling hatching order as covariate. In this analysis, year represents the environmental conditions that provisioning parents experienced, nest represents individual differences among provisioning parents ( adult quality or experience ) and we included nestling hatching order to test for differences within broods, because last-hatched chicks of a brood often experience considerably poorer provisioning rates than first-hatched nestlings (Masello and Quillfeldt 2002, 2004a). We initially included the interaction between the factors and the covariate in the model but removed it as it did not reveal significance (all P 0.2). In order to study possible correlations of colouration with parameters of adult body condition and size, GLMs based on Type III sums of squares, were carried out for each of the PC of the green, blue and red colour regions, which were included as dependent variables. GLM were carried out for each sex and parameter separately. The year was included as categorical independent variable ( factor ) in the models and body condition, wing and tail length, and the size of the abdominal red patch were included as covariates. Used data corresponded to 74 males and 79 females in the blue and green regions, 76 and 77 in the red region respectively. The age of the adult birds remained unknown. Results Fig. 1 shows the plumage characteristics for three feather regions of burrowing parrots. Reflectance curves of all three regions revealed some reflection in the UV (Fig. 1 ac). The main variability was in the first principal component (PC1, achromatic variation) for all three colours (71% in green, 66% in blue and 78% in red colour regions). The second and third principal components (PC2 and PC3) describe variation in spectral shape, i.e. chromatic variation (Fig. 1 df). PC2 explained 20% of variation of the green colour, 26% of blue and 14% of red, while PC3 explained 5% of variation of the green colour, 6% of blue and 6% of red (Fig. 1 df). Nestling coloration The colour of all three regions of nestlings varied between nests (Table 1). The green and blue regions also differed between years (Table 1). The red region of nestlings, moreover, varied with the hatching order (Table 1). Posthoc tests of the differences between years revealed that in particular, the blue region of nestlings differed in the years 1998 and 2003 (Dunn s Method, PB0.001). During 1998 (mean PC1 0.5890.17, n 39) the blue feather regions had higher than average achromatic brightness (mean PC1 all years 0.0290.11, n97), while in 2003 (mean PC1 0.5590.15, n44) feathers showed lower than average achromatic brightness. In order to study condition-dependence of coloration while controlling for the variation among years and hatching ranks, we carried out general linear models including only first-hatched chicks, and including year as factor, and nestling parameters (mean body condition, peak mass, hatching date, clutch and brood size, number of fledglings) as covariates. We found that the PC1 of the blue region negatively correlated with family size and breeding success (clutch size: F15.04, df88, t3.88, PB 0.001, P corr B0.001, h 2 0.152; brood size at the time of Table 1. Relationships between nestling coloration, years, nests and hatching order. Feather region PCA-scores Year Nest Hatching order P P corr h 2 P P corr h 2 P P corr h 2 Green PC1 0.211 0.020 B0.001 B0.001 0.480 0.019 () 0.158 0.035 PC2 0.106 0.028 0.017 0.145 0.394 0.201 0.010 PC3 B0.001 0.002 0.104 0.002 0.018 0.430 0.485 0.003 Blue PC1 B0.001 B0.001 0.271 B0.001 0.002 0.460 0.286 0.007 PC2 0.083 0.031 0.007 0.063 0.409 0.834 B0.001 PC3 0.107 0.028 0.371 0.312 0.046 () 0.345 0.025 Red PC1 0.954 B0.001 0.005 0.048 0.443 0.152 0.015 PC2 0.410 0.013 0.001 0.011 0.467 0.099 0.020 PC3 0.041 0.315 0.047 B0.001 B0.001 0.588 0.005 0.043 0.058 General linear models (GLM), based on Type III sums of squares, were carried out for each of the PC of the green, blue and red colour regions, which were included as dependent variables. Year and nest were included as factors in the models, while hatching order was included as covariate. Sample size green and blue region: 72 nestlings in 1998, 69 in 2001 and 87 in 2003. Sample size red region: 68 nestlings in 1999, 46 in 2001 and 85 in 2003. Negative t values are marked with (). Significant p-values after correcting for the number of tests (P corr ), using a conversion of the Dunn-Šidák equation (Sokal and Rohlf 1994), are marked bold. 657

hatching F7.31, df88, t2.70, P0.008, P corr 0.025, h 2 0.080; number of nestlings fledged: F9.01, df 96, t3.00, P0.003, P corr 0.010, h 2 0.089), and with nestling peak mass (F7.81, df 72, t2.79, P0.007, P corr 0.020, h 2 0.102). The PC2 of the blue region positively correlated with brood size at the time of hatching (F6.24, df 88, t2.50, P0.014, P corr 0.043, h 2 0.069). Adult coloration, body condition and size Parameters reflecting condition during the period of moult (tail length and wing length) showed some relationships with colouration. The tail length of males positively correlated with PC1 (GLM, F 6.35, df 74, t 2.52, P0.014, P corr 0.042, h 2 0.084), and negatively with PC2 (F18.45, df74, t 4.30, PB0.001, P corr B 0.001, h 2 0.211) of the green region. Wing length was negatively correlated to the PC2 of the green region of females (F6.80, df 79, t2.61, P0.011, P corr 0.032, h 2 0.084). The size of the abdominal red patch was not correlated with the colour of the red region when controlling for the influence of annual differences. We found all P0.05 after correcting for multiple testing. As expected, when controlling for the influence of annual differences, no colour parameter of adult parrots correlated with body condition at the time of capture, i.e. several months after the time of feather growth. Adult coloration and breeding success Adult plumage colouration could express the quality of the birds, for example if colourful parents had higher breeding success. In order to test for correlations between adult colouration and their breeding success, we carried out general linear models where hatching date, clutch and brood size, and number of fledglings were included as dependent variables, year as factor and PC 1, 2 and 3 of the three plumage regions of the adults as covariates. We found that the PC3 of the red region of adult males positively correlated with clutch size (F9.93, df57, t3.15, P 0.003, h 2 0.160) and brood size (F6.23, df56, t 2.5, P0.016, h 2 0.109). Other results were not significant after correcting the p-values for multiple testing. Inter-annual differences in adult colouration Among years, males differed in all three plumage regions while females differed in the blue and red regions (Table 2). Both adult males and females showed highly significant differences particularly for the PC1 of the blue and the red regions (Table 2). In particular, the years 1999 and 2003 were different (Tukey post-hoc test, P B0.001), with high PC1 in the blue region in feathers samples in 1999, while in 2003 feathers showed lower than average PC1 in the blue region (Fig. 2; see also blue region in Fig. 1). In the red region (Table 2), feathers sampled in the year 2003 had higher than average PC1 (Tukey post-hoc test, P 0.005) in both sexes, and in addition males had the highest PC3 for the studied years (Tukey post-hoc test, P 0.002). Table 2. Annual differences in the colouration of adult burrowing parrots. Feather region PCA-scores F females P P corr F males Green PC1 0.88 0.457 0.94 0.426 PC2 1.01 0.393 2.98 0.037 0.289 PC3 0.99 0.402 4.00 0.011 0.094 Blue PC1 12.05 B 0.001 B0.001 10.49 B 0.001 B0.001 PC2 7.60 B 0.001 0.002 1.04 0.382 PC3 0.77 0.514 3.28 0.026 0.209 Red PC1 5.18 0.003 0.024 4.95 0.004 0.031 PC2 1.54 0.212 2.23 0.092 PC3 1.01 0.393 5.90 0.001 0.010 Inter-annual differences in colouration among the years were studied using PC of the three feather regions of adult males and females and one-way analysis of variance. Data for the green and blue regions correspond to 24 females and 22 males in 1998, 24 and 21 in 1999, 9 and 11 in 2001, 22 and 20 in 2003. Data for the red region correspond to 23 females and 22 males in 1998, 24 and 21 in 1999, 8 and 11 in 2001, 22 and 22 in 2003 respectively. Significant P- values after correcting for the number of tests (P corr ), using a conversion of the Dunn-Šidák equation (Sokal and Rohlf 1994), are marked bold. Coloration of breeding pairs We found significant correlations between the male and the female of a pair in two plumage regions. In the blue region, PC1 were correlated between the male and the female of a pair (R0.38, n71, PB0.001, P corr 0.006). In the red region, PC3 from the male and the female of a pair were Reflectance (%) 8.5 1998 1999 8.0 2001 2003 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 300 400 500 600 700 Wavelength (nm) Figure 2. As an example for inter-annual differences in colouration we present reflectance curves of the blue region of the fourth secondary covert of the right wing of wild burrowing parrots. Reflectance curves correspond to feather samples obtained from 22 males in 1998, 21 in 1999, 11 in 2001 and 22 in 2003. 658

correlated (R0.36, n70, P0.002, P corr 0.011). To test whether these correlations were caused by similar conditions during moult or by assortative mating, we plotted the paired data (Fig. 3) and used Pearson correlations separately for each breeding season. Fig. 3 suggests an aggregation of the values according to the breeding season (e.g. low values for both males and females in 2003, Fig. 3a). We found that the correlation of the PC1 of the blue feather region was not significant in any season (Fig. 3), while the correlation of the PC3 of the red feather region was only significant in one season, namely 1999 (R 0.53, n24, P0.007, P corr 0.044; Fig. 3). Discussion In the present study, using reflectance spectra of plumage regions, we analysed condition-dependence of both the structural and the psittacofulvin-based colours of wild burrowing parrots of north-eastern Patagonia, Argentina. While the nestlings grew their feathers during the breeding season, and data on breeding success and nestling growth could be obtained directly, adult feathers were moulted outside the breeding season, and adult body condition could only be related to other feather parameters (wing length, tail length) and climatic conditions as indirect parameters. PC1 blue sector of males PC3 red sector of males 2 1 0-1 -2 2 1 0-1 -2 (a) (b) -2 1998 1999 2001 2003-2 1998 1999 2001 2003-1 0 1 2 3 PC1 blue sector of females -1 0 1 2 PC3 red sector of females Figure 3. Paired PC of reflectance spectra of blue and red regions of 71 breeding pairs of burrowing parrots. Data correspond to 17 pairs in 1998, 24 in 1999, 10 in 2001 and 20 in 2003 Condition-dependence in nestlings Nestling coloration depended on their condition during the development of the feather. In particular the structural colouration, in terms of achromatic brightness of the blue region, of nestlings showed a strong negative correlation with nestling peak mass within seasons, a parameter reflecting the condition of the nestlings. The study site is located on the Atlantic coast of South America, in a region experiencing increased precipitation during El Niño and severe droughts during La Niña phases of ENSO (see details in Masello and Quillfeldt 2003, 2004a, Masello et al. 2004). The ENSO dramatically impacts nestling survival and growth, and breeding success of burrowing parrots (Masello and Quillfeldt 2003, 2004a). Nestling colouration also showed high variability among years, in particular related to the blue colouration. The years 1998 and 2003 were different for the blue region of nestlings (unfortunately no samples were taken in 1999). During 1998, nestling feathers grew during a severe drought (Masello and Quillfeldt 2004a). These adverse conditions were reflected in the colouration of the nestling feathers that showed higher than average achromatic brightness in the blue regions of feathers grown in that year. Differences in nestling colouration related to nestling hatching order were weak. Last-hatched chicks of a brood often experience considerably poorer provisioning rates than first-hatched nestlings and as a consequence survival is much lower and growth worse (Masello and Quillfeldt 2002, 2004a). We found that PC3 of the red region of nestlings varied (positively) within nests in relation to the hatching rank. Condition-dependence in adults Adult burrowing parrots showed strong inter-annual variation, in particular in the blue and red regions (Table 2; Fig. 2). In the structural colouration, variability was accounted higher than average achromatic brightness in 1999 and lower than average achromatic brightness in the blue feathers sampled in 2003. In the psittacofulvin-based colouration, the year 2003 showed higher than average achromatic brightness in both sexes, while males showed the highest PC3 for the studied years. Adult feathers grew 9 to 11 months before sampling, and thus, feathers sampled in 1999 grew during a strong drought, while the feathers sampled in 2003 grew during the relatively moist conditions of a moderate to strong El Niño. The feathers grown during the drought had higher than average achromatic brightness in the structural blue colouration, while feathers grown during the good conditions of El Niño had lower than average achromatic brightness in the structural colouration and the higher than average achromatic brightness and higher PC3 in the psittacofulvin-based colouration. The average years, in terms of climatic conditions, were also average in terms of colour parameters. Several studies have shown that individual condition during moult correlates with the expression of structuralbased colouration in passerines (e.g. Keyser and Hill 1999, Doucet 2002, Ballentine and Hill 2003). In line with these studies, the observed inter-annual variability of blue 659

structural colouration of adult burrowing parrots during contrasting ENSO events suggests a relationship between environmental conditions and structural colouration in this psittaciform species. Although psittacofulvins are not acquired from the diet (Völker 1936, McGraw and Nogare 2005), the condition of the birds at the time of moulting might influence the feather growth and then pigment deposition. If psittacofulvins were costly to produce, as it has been hypothesized (Masello et al. 2004, McGraw and Nogare 2005), parrot plumage colouration is likely to be related to environmental conditions, such as ENSO events, and consequently to differences in the body condition during the time of moult. However, our data are not sufficient to test this hypothesis, as they include only two ENSO events. A long-term dataset would be needed in order to elucidate the dependence of colouration on environmental variability and the mechanisms behind those relationships. Within years, the body condition at the time of adult capture and sampling did not correlate with colour parameters. This was not unexpected, as body condition most probably did not reflect the condition of the birds during the growth of the sampled feathers for two reasons: 1) we could only measure the body mass during the breeding season, while the feathers grew several months earlier, and 2) we have shown earlier (Masello and Quillfeldt 2003) that during droughts caused by La Niña, adults invested more resources in their own body maintenance and decreased nestling provisioning, and thus maintain higher adult body masses and body conditions, despite poorer conditions and increased nestling mortality due to starvation. Instead, parameters reflecting condition during the period of moult (tail and wing length) showed relationships with the structure-pigment-based colouration of the birds (green), where larger animals had higher overall reflectance (high PC1) or higher reflectance in the visible part of the spectrum (low PC2, Fig. 1d). Tail and wing feathers, in contrast, grew at the time when our sampled feathers also grew and their length will thus better reflect the conditions during moult. This is supported by studies showing that feather growth varies with environmental conditions (e.g. Grubb 1989, Stratford and Stouffer 2001), especially resource availability, as well as the foraging ability of individuals. In line with this, during a previous study (Masello and Quillfeldt 2003), we found strong positive correlations between tail and wing measurements of adults and brood size: large-sized burrowing parrots had a higher probability of producing more fledglings than did smaller birds, indicating that tail and wing length may reflect condition. In line with this previous study, we found that parameters of adult colouration showed correlations with two parameters of breeding success, even when controlling for inter-annual differences. The PC3 of the red region of adult males positively correlated with clutch size and brood size at the time of hatching. Summary and conclusions We found common patterns in the condition-dependent colouration of adults and juveniles. The blue plumage seemed similarly affected in adults and juveniles, with higher achromatic brightness (PC1) in the blue regions of feathers during poor conditions. This suggests that higher achromatic brightness of the blue colour is the less favoured state, i.e. an overall darker blue such as observed during good conditions is perceived as more intense. On a mechanistic level, work by Shawkey et al. (2003) suggests that the nanostructure of the spongy layer of blue feathers of eastern bluebirds correlates with parameters of spectral shape, but not with achromatic brightness. Thus, differences in total brightness may be caused by morphological features outside the spongy layer, or by the number of melanin granules within the spongy layer (e.g. Shawkey et al. 2003, Shawkey and Hill 2006). It remains to be determined what causes differences in achromatic brightness of the structural colours observed in burrowing parrots, a species where a spongy ultrastructure is also present (Dyck 1977). The red feathers, in contrast, presented a more complex pattern. Here, PC1 and PC3 were affected, thus, the colour shape (hue) and not only the overall brightness played a role. During better conditions, adults grew red feathers with higher achromatic brightness and the colour hue was also affected, and the hue of the red region of nestlings also varied with the hatching order. In summary, the above results suggest some degree of condition-dependence of both the structural and the psittacofulvin-based colours of nestling burrowing parrots. Prospects The ecological importance and condition dependence of the psittacofulvin-based colouration of wild parrots has only recently begun to be investigated (present study; see also Masello and Quillfeldt 2003, Masello et al. 2004). But the mechanisms and possible costs of the production of the psittacofulvin-based colouration of wild parrots remain to be identified. Morelli et al. (2003) have recently demonstrated the potent antioxidant action of psittacofulvins, suggesting that psittacofulvins might be involved in a tradeoff between ornamentation and immune function. Future work is needed to test this hypothesis and to establish how psittacofulvin-based colouration may act as an honest signal of individual quality subject to sexual selection. Acknowledgements We wish to thank Andrew Bennett for training on colour measuring methods, for providing the spectrometer and the lab facilities to conduct the present colour measurements at his lab at the University of Bristol. We also wish to thank Ramón Conde, Adrián Pagnossin, María L. Pagnossin, Hans-Ulrich Peter, Roberto Ure, Mara Marchesan, Tina Sommer, Mauricio Failla, Raquel Percáz and Gert Dahms for their help with the fieldwork, Kate Buchanan and Anne Peters for useful comments on the manuscript and financial support, Kaspar Delhey and Roger Mundry for useful comments on the statistical analyses, and Kay Gruenewald for lab analyses. This project was partially supported by the City Council of Viedma (Argentina), a grant of the state of Thuringia (Germany), a co-operation grant between the International Bureau of the BMBF of Germany (ARG 99/020) and the Argentinean SECyT (AL/A99-EXIII/003), a grant of the World Parrot Trust (WPT), several grants from the Liz Claiborne Art Ortenberg Foundation (LCAOF) and the Wildlife Conservation Society (WCS), a grant of the British Ecological Society (BES) to JFM and BBSRC and Leverhulme grants to Andrew Bennett. We 660

would like to thank Lisa Tell (University of California) and ZooGen Services for their assistance with the gender analysis of the 2003 blood samples. The present study was carried out with permission of the Dirección de Fauna Silvestre de la Provincia de Río Negro, Argentina (Exp. no. 143089-DF-98). References Andersson, S. 1999. Morphology of UV reflectance in a whistlingthrush: implications for the study of structural colour signalling in birds. J. Avian Biol. 30: 193204. Andersson, S.,.Örnborg, J. and Andersson, M. 1998. Ultraviolet sexual dimorphism and assortative mating in blue tits. Proc. R. Soc. B 264: 445450. Ballentine, B. and Hill, G. E. 2003. Female mate choice in relation to structural plumage coloration in blue grosbeaks. Condor 105: 593598. Bennett, A. T. D., Cuthill, I. C. and Norris, K. J. 1994. Sexual selection and the mismeasure of color. Am. Nat. 144: 848 860. Bennett, A. T. D., Cuthill, I. C., Partridge, J. C. and Maier, E. J. 1996. Ultraviolet vision and mate choice in zebra finches. Nature 380: 433435. Bennett, A. T. D., Cuthill, I. C., Partridge, J. C. and Lunau, K. 1997. Ultraviolet plumage colours predict mate preferences in starlings. Proc. Natl. Acad. Sci. U. S. A. 94: 86188621. Burkhardt, D. 1989. UV vision: a bird s eye view of feathers. J. Comp. Physiol. A 164: 787796. Cuthill, I. C., Bennett, A. T. D., Partridge, J. C. and Maier, E. J. 1999. Plumage reflectance and the objective assessment of avian sexual dichromatism. Am. Nat. 153: 183200. Cuthill, I. C., Partridge, J. C., Bennett, A. T. D., Church, S. C., Hart, N. S. and Hunt, S. 2000. Ultraviolet vision in birds. Adv. Study Behav. 29: 159214. Doucet, S. M. 2002. Structural plumage coloration, male body size, and condition in the blue-back grassquit. Condor 104: 3038. Doucet, S. M. and Montgomerie, R. 2003. Multiple sexual ornaments in satin bowerbirds: ultraviolet plumage and bowers signal different aspects of male quality. Behav. Ecol. 14: 503509. Dyck, J. 1971a. Structure and spectral reflectance of green and blue feathers of the rose-faced lovebird (Agapornis roseicollis). Biol. Sckrifter 18: 167. Dyck, J. 1971b. Structure and colour-production of the blue barbs of Agapornis roseicollis and Cotinga maynana. Zeitschr. für Zellforsch. 115: 1729. Dyck, J. 1977. Feather ultrastructure of Pesquet s parrot Psittrichas fulgidus. Ibis 119: 364366. Dyck, J. 1992. Reflectance of plumage areas colored by green feather pigments. Auk 109: 293301. Endler, J. A. 1990. On the measurement and classification of colour in studies of animal colour patterns. Biol. J. Linn. Soc. 41: 315352. Endler, J. A. and Théry, M. 1996. Interacting effects of lek placement, display behavior, ambient light, and color patterns in three Neotropical forest-dwelling birds. Am. Nat. 148: 421452. Finger, E., Burkhardt, D. and Dyck, J. 1992. Avian plumage colors. Origin of UV reflection in a black parrot. Naturwiss. 79: 187188. Grubb, T. C., Jr. 1989. Ptilochronology: Feather growth bars as indicators of nutritional status. Auk 106: 314320. Hunt, S., Kilner, R. M., Langmore, N. E. and Bennett, A. T. D. 2003. Conspicuous, ultraviolet-rich mouth colours in begging chicks. Proc. R. Soc. B (Suppl.) 270: S25S28. Jacobs, G. H. 1981. Comparative color vision, Academic Press: New York. Keyser, A. J. and Hill, G. E. 1999. Condition-dependent variation in the blue-ultraviolet coloration of a structurally based plumage ornament. Proc. R. Soc. B 266: 771777. Krukenberg, C. F. W. 1882. Die Federfarbstoffe der Psittaciden. Vergleichend-physiologische Studien 2: 2936. Langmore, N. E. and Bennett, A. T. D. 1999. Strategic concealment of sexual identity in an estrildid finch. Proc. R. Soc. B 266: 543550. Lubjuhn, T. and Sauer, K. P. 1999. DNA fingerprinting and profiling in behavioural ecology, In: Epplen, J. T. and Lubjuhn, T. (eds). DNA profiling and DNA fingerprinting. Birkhäuser Verlag: Basel, pp. 3952. Lubjuhn, T., Sramkova, A., Masello, J. F., Quillfeldt, P. and Epplen, J. T. 2002. Truly hypervariable DNA fingerprints due to exceptionally high mutation rates. Electrophoresis 23: 517519. Masello, J. F. and Quillfeldt, P. 2002. Chick growth and breeding success of the burrowing parrot. Condor 104: 574586. Masello, J. F. and Quillfeldt, P. 2003. Body size, body condition and ornamental feathers of burrowing parrots: Variation between years and sexes, assortative mating and influences on breeding success. Emu 103: 149161. Masello, J. F. and Quillfeldt, P. 2004a. Consequences of La Niña phase of ENSO for the survival and growth of nestling burrowing parrots on the Atlantic coast of South America. Emu 104: 337346. Masello, J. F. and Quillfeldt, P. 2004b. Are haematological parameters related to body condition, ornamentation and breeding success in wild burrowing parrots Cyanoliseus patagonus? J. Avian Biol. 35: 445454. Masello, J. F., Pagnossin, M. L., Lubjuhn, T. and Quillfeldt, P. 2004. Ornamental non-carotenoid red feathers of wild burrowing parrots. Ecol. Res. 19: 421432. Masello, J. F., Pagnossin, M. L., Sommer, C. and Quillfeldt, P. 2006a. Population size, provisioning frequency, flock size and foraging range at the largest known colony of Psittaciformes: the burrowing parrots of the north-eastern Patagonian coastal cliffs. Emu 106: 6979. Masello, J. F., Choconi, R. G., Sehgal, R. M. N., Tell, L. A. and Quillfeldt, P. 2006b. Blood and intestinal parasites in wild Psittaciformes: a case study of burrowing parrots (Cyanoliseus patagonus). Orn. Neotropical 17: 515529. Masello, J. F., Sramkova, A., Quillfeldt, P., Epplen, J. T. and Lubjuhn, T. 2002. Genetic monogamy in burrowing parrots Cyanoliseus patagonus? J. Avian Biol. 33: 99103. Mays, H. L., McGraw, K. J., Ritchison, G., Cooper, S., Rush, V. and Parker, R. S. 2004. Sexual dichromatism in the yellowbreasted chat Icteria virens: spectrophotometric analysis and biochemical basis. J. Avian Biol. 35: 125134. McGraw, K. J. and Nogare, M. C. 2004. Carotenoid pigments and the selectivity of psittacofulvin-based coloration systems in parrots. Comp. Biochem. Physiol. B 138: 229233. McGraw, K. J. and Nogare, M. C. 2005. Distribution of unique red feather pigments in parrots. Biol. Lett. 1: 3843. Morelli, R., Loscalzo, R., Stradi, R., Bertelli, A. and Falchi, M. 2003. Evaluation of the antioxidant activity of new carotenoidlike compounds by electron paramagnetic resonance. Drugs Exptl. Clin. Res. 29: 95100. Moreno, J., Merino, S., Lobato, E., Rodríguez-Gironés, M. A. and Vásquez, R. A. 2007. Sexual dimorphism and parental roles in the thorn-tailed rayadito (Furnariidae). Condor 109: 312 320. Pearn, S. M., Bennett, A. T. D. and Cuthill, I. C. 2001. Ultraviolet vision, fluorescence and mate choice in a parrot, the budgerigar Melopsittacus undulatus. Proc. R. Soc. B 268: 22732279. 661

Prum, R. O., Andersson, S. and Torres, R. H. 2003. Coherent scattering of ultraviolet light by avian feather barbs. Auk 120: 163170. Quesada, J. and Senar, J. C. 2006. Comparing plumage colour measurements obtained directly from live birds and from collected feathers: the case of the great tit Parus major. J. Avian Biol. 37: 609616. Shawkey, M. D. and Hill, G. E. 2005. Carotenoids need structural colours to shine. Biol. Lett. 1: 121124. Shawkey, M. D. and Hill, G. E. 2006. Significance of a basal melanin layer to production of non-iridescent structural plumage color: evidence from an amelanotic Steller s jay (Cyanocitta stelleri). J. Experim. Biol. 209: 12451250. Shawkey, M. D., Estes, A. M., Siefferman, L. M. and Hill, G. E. 2003. Nanostructure predicts intraespecific variation in ultraviolet-blue plumage colour. Proc. R. Soc. B 270: 1455 1460. Sokal, R. R. and Rohlf, F. J. 1994. Biometry-The principles and practice of statistics in biological research. 3rd. ed, Freeman and Company: New York. Stradi, R., Pini, E. and Celentano, G. 2001. The chemical structure of the pigments in Ara macao plumage. Comp. Biochem. Physiol. B 130: 5763. Stratford, J. A. and Stouffer, P. C. 2001. Reduced feather growth rates of two common birds inhabiting Central Amazonian Forest fragments. Cons. Biol. 15: 721728. Veronelli, M., Zerbi, G. and Stradi, R. 1995. In situ resonance raman spectra of carotenoids in birds feathers. J. Raman Spectrosc. 26: 683692. Völker, O. 1936. Über den gelben Federfarbstoff des Wellensittichs (Melopsittacus undulatus (Shaw)). J. Orn. 84: 618 630. Völker, O. 1937. Über fluoreszierende, gelbe Federpigment bei Papageien, eine neue Klasse von Federfarbstoffen. J. Orn. 85: 136146. Völker, O. 1942. Die gelben und roten Federfarbstoffe der Papageien. Biol. Zbl. 62: 813. Zöfel, P. 2002. Statistik verstehen. Ein Begleitbuch zur computergestützten Anwendung, Addison-Wesley: München. 662