ECOPHYSIOLOGY

Clotilde Biard Æ Peter F. Surai Æ Anders P. Møller

Effects of carotenoid availability during laying on reproductionin the blue tit

Received: 10 October 2004 / Accepted: 7 February 2005� Springer-Verlag 2005

Abstract Carotenoids are antioxidant pigments involvedin several physiological processes and signalling in ani-mals that cannot synthesise them and therefore mustacquire them from food. We experimentally investigatedthe effects of carotenoid availability in the diet duringegg laying on antioxidant deposition in egg yolk and therelated effects on nestling condition, female body con-dition and parental investment in the blue tit (Paruscaeruleus). Carotenoid supplementation of egg-layingfemales resulted in a significant increase in carotenoidconcentration in egg yolk, but not in vitamin E or Aconcentration. There was no relationship between yel-low plumage colour of adult females and carotenoiddeposition in eggs, and no differential effect of feedingtreatment depending on female colour. Nestlings fromeggs laid by carotenoid supplemented females hadlonger tarsi, had faster development of the immunesystem as reflected by leukocyte concentration in blood,and grew brighter yellow feathers than nestlings fromcontrol females. However, nestlings from the two groupsdid not differ significantly in body mass, plasma an-tioxidants or plumage colour hue. At the time of chickrearing, carotenoid-fed females had increased plasmavitamin E levels compared to controls. However, fe-males from the two treatment groups did not differsignificantly in body condition or feeding rate. Theseresults suggest that carotenoid availability is limiting

during egg laying, and that females may have to balancethe benefits of investing in egg quality against the po-tential costs of impairing their own future antioxidantprotection. In addition, there may be considerable var-iation in carotenoid availability not only across seasons,but also among different stages of the breeding season.

Keywords Antioxidants Æ Carotenoid-based plumagecolour Æ Fledgling condition Æ Parental investment ÆYolk composition

Introduction

From a life history perspective, an individual’s invest-ment in current reproduction and its expected benefits interms of reproductive success will be balanced against itspotential costs on future fecundity or survival (Stearns1989). However, from a proximal point of view, thesereproductive strategies will also be constrained by theindividual’s body condition, the quality of the environ-ment in which it is reproducing, and their interaction.The availability of nutrients that are essential for bothself-maintenance and reproduction may mediate suchcosts of reproduction.

Carotenoids are a family of more than 600 biologi-cally active pigments produced by plants, algae andfungi, involved in several physiological processes andsignalling in animals, who cannot synthesise them andtherefore must acquire them from food (Goodwin 1984).The biochemical structure of carotenoids makes themefficient quenchers of reactive oxygen species, and theyare part of the antioxidant system of vertebrates pro-tecting DNA, proteins and biological membranesagainst oxidative damage (reviewed in Edge et al. 1997;Møller et al. 2000). Carotenoids are used in physiolog-ical processes as antioxidants, but also have a protectiveand recycling role for other antioxidants like vitamins Eand A (Surai and Speake 1998), and they modulateantioxidant enzyme function (Surai et al. 2001a). Fur-

Communicated by Roland Brandl

C. Biard (&) Æ A. P. MøllerLaboratoire de Parasitologie Evolutive - CNRS UMR 7103,Universite Pierre et Marie Curie - Paris 6, case 237,7 quai Saint Bernard, 75252 Paris Cedex 05, FranceE-mail: Clotilde.Biard@u-bourgogne.fr

P. F. SuraiLipid and Antioxidant Group - Department of Biochemistry andNutrition, Avian Science Research Centre, Scottish AgriculturalCollege, Auchincruive Ayr, KA6 5HW UK

Present address: C. BiardBiogeosciences CNRS UMR 5561, Universite de Bourgogne,6, bd Gabriel, 21000 Dijon, France

Oecologia (2005)DOI 10.1007/s00442-005-0048-x

thermore, some carotenoids are precursors of other an-tioxidants (i.e. b-carotene is a precursor of vitamin A)(Surai 2002). They are also directly involved in activa-tion of the immune system and in detoxification ofreactive oxygen species by-products (reviewed in Chew1993; Møller et al. 2000).

In birds, females allocate large amounts of antioxi-dants and particularly carotenoids to their eggs (Blountet al. 2000). These molecules have been shown in captivebirds to be of crucial importance during embryo devel-opment and hatching (reviewed in Surai et al. 2001a), asthis period of intense growth is associated with increasedoxidative stress (Surai et al. 1999; Surai 2002). They havethus been suggested to have profound effects on chickhealth and immunity (Blount et al. 2000, 2002; Surai2002). However, carotenoids may be a limiting resourceas a consequence of environmental variability in theiravailability, differences in foraging ability or differencesin absorption or metabolising efficiency among individ-uals (reviewed in Olson and Owens 1998; Møller et al.2000). They may thus be of crucial importance duringreproduction for female birds, as they are not only di-rectly involved in egg production and thus potentially inreproductive success, but also in other physiologicalfunctions highly activated during this period (the im-mune system: Møller et al. 2003, and consequently theantioxidant system). Egg production may impair bodycondition and/or immunity of adult females, and con-sequently their ability to efficiently rear chicks (Mona-ghan et al. 1998) and survive (Visser and Lessells 2001).Very few studies have investigated whether carotenoidsare a limiting resource in wild bird populations: Carot-enoids could be limiting during moult in great tit (Parusmajor) and house finch (Carpodacus mexicanus), andenvironmental variation in their availability at that timedetermined plumage colour (Slagsvold and Lifjeld 1985;Hill et al. 2002). In the lesser black-backed gull (Larusfuscus), carotenoid availability during reproductionlimited egg laying capacity (Blount et al. 2004). How-ever, much is still unknown about patterns of carotenoidinvestment in reproduction and their consequences forfitness in natural populations.

We experimentally investigated the effects of carot-enoid availability in the diet during egg laying on anti-oxidant deposition in egg yolk and the subsequent effectson nestling condition, female body condition andparental investment in a natural population of blue tits(Parus caeruleus). The blue tit is a hole nesting passerinethat breeds in gardens or woodland areas, laying one eggper day and starting incubation at clutch completion.The yellow colour of the blue tit’s breast plumage iscarotenoid-based and determined by two xanthophyllpigments, lutein and zeaxanthin (Partali et al. 1987).Pigments deposited in feathers are unavailable after-wards for other physiological functions and feathercolour may reflect the amount and type of carotenoidsavailable to the individual at the time of feather growth.We carried out a carotenoid supplementation experi-ment during egg laying, in which females were either

provided with extra carotenoids in vegetable fat(‘carotenoid’ group) or with vegetable fat only (‘control’group) to test for carotenoid limitation and its conse-quences.

If there is a limit to carotenoid availability during egglaying, then increasing the availability of carotenoids toegg laying females should allow them to deposit morepigments in their eggs thereby increasing egg quality.Furthermore, if carotenoid availability is limited duringthe entire reproductive period, then females having ac-cess to a carotenoid-rich diet during egg-laying may alsoduring chick rearing show a superior body condition andparental investment. If there is a limit to carotenoidavailability across seasons, then less brightly colouredfemales (i.e. females that had fewer pigments availablefor investment in feather colour during moult) shouldbenefit more from supplementation than more brightlycoloured females. Finally, we hypothesised that an in-crease in egg carotenoid content would result in benefi-cial effects on hatching success, nestling condition and/or fledging success. Because of the physiological inter-actions between carotenoids and vitamin E and vitaminA (Edge et al. 1997; Surai et al. 2001a; Surai 2002), wealso measured these two other major lipid-soluble an-tioxidants.

Materials and methods

Data collection and experimental protocol

Data were collected in 2002 in a population of blue titsbreeding in eastern France (48�17¢N, 4�18¢E). The studyarea (about 250 ha) contained 400 nestboxes evenlydistributed among homogenous deciduous old wood-land composed mainly of oak (Quercus spp.), hornbeam(Carpinus betulus) and beech (Fagus sylvatica). Nestswere regularly inspected to determine laying date, clutchsize, incubation date, hatching date, number of hatch-lings and unhatched eggs, and number of fledglings.

A total of 79 nests were followed for this experiment,six of which failed before incubation, and they weretherefore excluded from the study. Carotenoids wereobtained from Kemin Foods (FloraGlo 20% #80447) inthe form of a dietary supplement made of crystallinelutein derived from marigold flowers and suspended incorn oil. Xanthophyll concentration in the product was20% lutein and 0.86% zeaxanthin, confirmed withHPLC analysis, and both carotenoids were in freealcohol forms readily available for absorption. Supple-mental units were made of either pure solid vegetable oil(control treatment) or the same oil melted with carote-noids (v/v) to a final concentration of xanthophylls of10% (carotenoid treatment). Mean mass of a supple-mental unit was 5 g. Supplemental units were freshlyprepared each evening and stored at �10�C and in thedark until use the following day. On the day it was laid,the first egg was numbered with a pencil, and the nestwas randomly attributed to a treatment group (control

or carotenoids). A feeding unit was placed inside thenestbox, just under the entrance hole and above the nestand females were then allowed to feed freely on thesupplementation. Nests were subsequently visited everysecond day, and the two new eggs numbered. Theamount of food eaten was estimated (0: not eaten, 1:pecked, 2: 1/5 to 1/4 eaten, 3: 1/4 to 2/3 eaten, 4: morethan 2/3 eaten) and supplemental units replaced amaximum of three times either if lost or completely ea-ten. At the first visit after supplemental food was pro-vided, 89% of the females were found to have startedto feed on it (n=65). Among the remaining females,8% (n=6) started to feed before the second visit, 1.5%(n=1) before the third, and 1.5% before the fourth visit(n=1).

When the clutch was close to completion, i.e. whenthe 10th and 11th eggs were laid, the first and one of thetwo last eggs were removed. Eggs were not sampled ifthe clutch was found incubated with fewer than ten eggs,as incubation would have begun more than 48 h before,in order to avoid any bias in biochemical analysesresulting from embryo development (n=1 control nest).Sampled eggs were stored in a cooling bag in the field,and immediately frozen when back in the lab, within 5 hof removal. After egg sampling, nests were visited everysecond day until incubation had begun, clutch size wasrecorded (mean clutch size ± SE=11.6±0.1, n=70,minimum 9, maximum 14), and supplemental feedingstopped. Total quantity of supplemental food eaten wascalculated summing the estimated amounts eaten foreach supplemented unit provided. Thus, for each nest weobtained the first egg, i.e. the egg before treatment, andone of the 9th to 11th eggs, i.e. the post-treatment egg.Eggs were collected for 72 nests and yolks analysed for69 nests: 34 carotenoid treatment and 35 controls (eggsfrom three nests and the egg before treatment of onecontrol nest were broken during storage and could notbe analysed). Once incubation started, nests were notvisited until the day before the estimated hatching date(incubation date + 14 days), and then visited each dayuntil hatching. Four nests were deserted or depredatedbefore hatching.

Adults were trapped while feeding young aged at least5 days to limit risk of clutch desertion, and were sexedaccording to the presence (female) or absence (male) of abrood patch. Adults were individually identified with anumbered aluminium ring and they received an addi-tional unique combination of three colour rings for vi-sual identification (46 females: 23 carotenoid and 23control nests). Among the 69 broods hatched, 34 wereincluded in another experiment that will be reportedelsewhere. Thus results concerning nestlings reportedhere concern the remaining 35 nests. Chicks were cap-tured in the nest a few days before fledging, when 12–18 days old (mean ± SE: 15.3±0.3 days). A total of227 nestlings from 27 nests (13 carotenoid nests: 107nestlings and 14 control nests: 120 nestlings) were indi-vidually identified with a numbered aluminium ring.Sample size was reduced due to nest desertion or pre-

dation (four nests) or early fledging (three nests). Inaddition, one nest was excluded from the experiment asthe female incubated during 52 days and only two nes-tlings finally hatched from the nine eggs left after eggsampling. We measured tarsus length to the nearest0.1 mm with a calliper and body mass to the nearest0.25 g with a Pesola spring balance. A sample of five toeight yellow feathers was plucked from the centre of theyellow breast for each bird, and stored in individualplastic bags in the dark until later colour analysis.

A blood sample (50–100 ll) was taken from thebrachial vein in heparinized micro-haematocrit tubes for42 females and 163 nestlings. Nestlings were not blood-sampled if captured too close to fledging in order toavoid premature fledging. Blood samples were stored ina cooling bag in the field, and when back in the lab,placed in a fridge and left upright to measure sedimen-tation after 8 h. Samples were then centrifuged for 5 minat 5,000 rpm. Length of the red blood cell layer wasmeasured to the nearest 0.5 mm and length of the ‘buffycoat’ layer was measured to the nearest 0.01 mm with agraduated magnifying ocular. Plasma was then sepa-rated from blood cells and stored at �20�C. Sedimen-tation and hematocrit were measured as the amount ofred blood cells divided by the total length of the tubefilled with blood (n=163 nestlings). The relative amountof leukocytes in total blood volume was measured as theratio of the ‘buffy coat’ layer to total length (n=161nestlings). Mean ± SE relative error in the estimationof sedimentation rate and the proportion of leukocytesin blood was 26±1% and 26±1%, respectively (medi-ans were 22% and 20%, respectively). Blood sedimen-tation is useful to detect elevated levels ofimmunoglobulins and fibrinogen, high sedimentationrate being indicative of acute infections and inflamma-tory diseases (e.g. Sturkie 1986; Svensson and Merila1998). The number of leukocytes increases in case ofstress and infection (e.g. Sturkie 1986; Ots et al. 1998),and in chicken and quail chicks the number of leuko-cytes in blood increases to reach adult levels by 3 weeksof age (Sturkie 1986).

Observations of feeding behaviour were conductedwith a telescope at a distance of at least 20 m from thenestbox, once the adults were identified with colourrings. Nests were observed during (mean time ± SE)32±1 min (range 21–61 min), and female entries in thenestbox were recorded. It was impossible to accuratelydetermine the size or type of prey brought to the nest,and thus we only recorded whether and how many fooditems females carried. Observations were conducted on43 nests, when chicks were six (mean nestling age ± SE:6.7±0.2 days, ten observations), nine (9.2±0.1 days, 42observations) and/or 12 days old (12.2±0.1 days, 39observations). Most nests were observed more thanonce: four nests were observed once, 30 nests twice andnine nests three times. The number of prey brought tothe nest by the female was highly and positively corre-lated with female entry frequency (r=0.96, P<0.0001)which was significantly and moderately repeatable

(intra-class correlation coefficient (Lessells and Boag1987): r=0.44, P=0.0006). Female entry frequency wasthus used as a measure of female feeding rate.

Antioxidant analysis of yolk and plasma

Before antioxidant analysis, eggs were weighed if theshell was intact, and yolks separated from albumen andweighed if not damaged (mean egg mass ±SE=1.161±0.009 g, n=135, range=0.768–1.384 g;mean yolk mass ± SE=0.263±0.003 g, n=129,range=0.200–0.368 g). Antioxidants were extractedfrom eggs using half the yolk. Samples were homogen-ised with 0.7 ml NaCl (5%) and 1 ml ethanol, afterwhich antioxidants were extracted adding 2 ml hexaneand further homogenisation, centrifugation and collec-tion of the hexane phase (extraction repeated twice). Thesame protocol was followed for plasma: 20 ll of plasmawas mixed with 40 ll ethanol, then extracted twice with500 ll hexane. For both types of samples, hexane ex-tracts were pooled and evaporated at 60–65�C undernitrogen flow, the residue was then dissolved in 0.1 mldichloromethane and 0.1 ml methanol.

Carotenoid composition and concentration, andvitamin A and E concentration were determined fol-lowing previously published procedures (Surai 2000;Surai et al. 2001b). Carotenoid composition was deter-mined by using high-performance liquid chromatogra-phy (HPLC) with a Spherisorb type ODS2 5-lm C18reverse-phase column, 25 cm·4.6 mm (Phase separation,Clwyd, UK). An isocratic HPLC at a flow rate of 2 mlmin�1 with a mobile phase of acetonitrile–methanol(85:15) for 10 min was used with a change to acetoni-trile–dichloromethane–methanol (70:20:10) and runningfor the next 10 min and re-equilibration with the firstmobile phase for the next 4 min, using detection byabsorbance at 445 nm. Peaks were identified by com-parison with the retention times of a range of carotenoidstandards (obtained from Sigma, Poole, UK, Fluka,Gillingham, UK, Apin, Abingdon, UK, and Hoffmann-LaRoche, Basel, Switzerland). Total carotenoid con-centration was determined by HPLC with a Spherisorbtype S5NH2 reverse-phase column 25 cm·4.6 mm (Phaseseparation, Clwyd, UK) with a mobile phase of metha-nol–distilled water (97:3), at a flow rate of 1.5 ml min�1

as described by (Horak et al. 2002). Lutein was used forthe calibration (Sigma, Poole, UK).

Concentrations of vitamins A and E were determinedby injection of samples onto a Spherisorb type ODS2 3-lm C18 reverse-phase column, 15 cm·4.6 mm (PhaseSeparation, Clwyd, UK) with a mobile phase of meth-anol–distilled water (97:3), at a flow rate of 1.05 mlmin�1 using fluorescence detection by excitation andemission wavelength of 295 and 330 nm, respectively,for vitamin E and 330 and 480 nm for vitamin A. Peaksof retinol, d-, c- and a-tocopherol were identified bycomparison with the retention time of standards oftocopherols (Sigma, Poole, UK).

All sampled eggs were analysed for total carotenoidconcentration. A sub-sample of these eggs was randomlychosen to determine vitamin A and E (14 nests fromeach treatment, with both before- and after-treatmenteggs) and individual carotenoid composition (five nestsfrom each treatment, for both before- and after-treat-ment eggs, i.e. 20 eggs in total). All female plasmasamples of at least 20 ll (n=37) were analysed for totalcarotenoid and vitamin concentration. All nestlingplasma samples of at least 20 ll (n=81) were analysedfor total carotenoid and a random sub-sample wasanalysed for vitamin A and E (n=40). Concentrationsare given in lg/g or lg/ml. Vitamin E calculated as thesummed concentrations of d-, c-, a-tocopherol is used insubsequent analyses. However, results obtained consid-ering only the major antioxidant form, a-tocopherol(Surai 2002) were qualitatively the same.

Feather colour analysis

Yellow breast feather colour was analysed in the labo-ratory, using a spectroradiometer (Ocean Optics) (Sainoet al. 1999; Horak et al. 2000), blindly with respect to theexperimental group of the samples. In brief, featherswere illuminated at an angle of 90� with a deuterium-halogen lamp, and reflected light was measured at anangle of 45�. Feathers were measured against a blackshort velvet background regularly changed to keep itsreflectance constant and equal to 0. Percentage ofreflectance at each 1 nm interval was calculated between300 nm and 700 nm, with respect to white and darkreferences, as R(k)=100 · [(sample-white)/(white-dark)].From these values, we calculated two independentparameters, brightness and hue to characterise caroten-oid-based plumage colour. Brightness is the totalintensity of light reflected by the feather. As we wereinterested in yellow feathers, brightness was calculatedas the integral of reflectance on the restricted part of thespectrum corresponding to the yellow band [550–625] nm (Saks et al. 2003). Brightness in the yellow ac-counted for 20±0.1% and 19.01±0.03% of the totalreflectance in females and nestlings, respectively. Hue iscalculated as an angle value (0�: red–90�: yellow) usingthe segment classification and is correlated to thewavelength of maximum slope and the sign of this slope(Endler 1990). Plumage colour was analysed for all 46females and for 220 nestlings (feathers from all six nes-tlings of one nest and from one nestling of one nest, allfrom the control group could not be measured). Twofeathers were analysed for each individual with fourmeasures per feather. Repeatability of measurementscalculated as the intra-class correlation coefficient(Lessells and Boag 1987) was always highly significant(P<0.0001) with the following values (females; nest-lings): (a) repeatability within feathers: yellow bright-ness: 0.68; 0.50, yellow hue: 0.50; 0.41; (b) andrepeatability within individuals: brightness: 0.61; 0.42,hue: 0.44; 0.34. Average values for the eight measures

per individual were used in subsequent statistical ana-lyses.

Statistical analyses

All statistical analyses were made using SAS ver. 8.2(1999–2001, SAS Institute Inc., Cary, NC, USA). Testsof the residuals for normality and homoscedasticity wereused to check the validity of the model, and dependentvariables were transformed to meet these constraints ifneeded. In accordance with Freckelton (2002) we did notuse residuals to control for the effect of size on bodymass and of the proportion of red blood cells on sedi-mentation. Instead, tarsus length and hematocrit wereentered as covariates in models with body mass andsedimentation, respectively, as dependent variables. Inaddition, initial models concerning females always in-cluded tarsus length, laying date and nestling age, inorder to account for female body size and seasonalvariation, respectively. In particular when investigatingplasma antioxidants in adult females, tarsus lengthcontrolled for metabolic rate which is a function of bodysize (Sturkie 1986) and the related rate of production ofreactive oxygen species. Initial models concerning nes-tlings always included hatching date, nestling age andbrood size in order to account for seasonal variation,nestling growth and sibling competition, respectively.Carotenoid concentration was included in analysesinvolving vitamin A and E concentration in plasma innestlings and adult females to control for correlationbetween plasma antioxidants resulting from their phys-iological interactions (see Introduction). Interactionsand main effects were sequentially dropped from themodel when not significant.

Effect of treatment on egg characteristics (egg massand yolk mass) and yolk carotenoid composition wasinvestigated using a MANOVA model with the proce-dure GLM. Difference in yolk antioxidant compositionbetween before- and after-treatment eggs was testedusing repeated measures ANOVA with the before-treatment egg as a contrast. Yolk mass of both eggs wasfirst entered as additional covariates in these models, butdid not explain a significant part of the variance. Mixedlinear models (Goldstein 2003) were used with theMIXED procedure to analyse feeding rates, with nest asa repeated effect, and time of day, nestling age andbrood size as initial covariates in the fixed effects. Mixedlinear models were also used to analyse nestling char-acteristics with nest as a random effect. Declaring nest arepeated or random effect sets up a common correlation

among all observations having the same level of nest inorder to account for non-independence. When analysingnestling characteristics, both genetic and common rear-ing environment effects are included in the sources ofcovariance existing among nestlings raised in the samenest. The Wald test of the covariance parameter wasused to assess whether the variance was significantlystructured by the repeated or random effect. The nullmodel likelihood ratio v2 test was used to assess whetherthe model with repeat or random effect(s) provided asignificantly better fit than the same model without re-peat or random effect(s). Degrees of freedom for Type 3Tests of Fixed Effects were calculated using the Residualmethod. Models were compared with Akaike’s Infor-mation Criterion (AIC), and the most parsimonious wasretained (lowest AIC, Burnham and Anderson 1998).

Significant models were followed by comparisons ofmeans or least square means between treatments withadjusted values of P, using the Tukey-Kramer method.Values are reported as mean ± 1SE.

Results

There was no difference in laying date between treat-ments (F1,71=0.07, P=0.79). Females consumed thesame amount of control and carotenoid-supplementedfeed (control: 1.2±0.2 supplemental unit, carotenoid:0.76±0.07 supplemental unit, Wilcoxon two-sampletest, P=0.81). Female treatment did not significantlyaffect egg mass and yolk mass (MANOVA, Wilks’Lambda=0.95, F2,57=1.48, P=0.24), or clutch size(F1,68=0.05, P=0.83).

Yolk composition

Individual carotenoids

The two main carotenoids in blue tit eggs were lutein(with isomer cis-lutein comprising less than 5% of trans-lutein content) and zeaxanthin, with small amounts ofb-cryptoxanthin and b-carotene (Table 1). As theexperimental food was mainly enriched in lutein, weexpected a difference in relative proportions of luteinand zeaxanthin in post-treatment eggs between treat-ments, but not in the general profile of individual car-otenoids in the yolk. There was indeed a marginal effectof treatment on carotenoid composition of the yolk inpost treatment eggs (MANOVA, Wilks’ Lambda=0.06,F5,4=5.71, P=0.06), female treatment mainly affecting

Table 1 Mean yolk carotenoid composition (%±SE) of eggs produced after treatment by blue tits from carotenoid and control feedinggroups (n=5)

Lutein Zeaxanthin Cis-lutein b-cryptoxanthin b-carotene Unidentified

Carotenoid 77.7±2.1 13.3±0.7 0.6±0.1 2.1±0.3 4.6±1.1 1.8±0.4Control 63.0±1.6 18.5±0.5 1.5±0.1 3.2±0.4 10.6±2 3.3±0.6

the proportion of lutein (F1,8=29.25, P=0.0006),cis-lutein (F1,8=18.86, P=0.002) and zeaxanthin(F1,8=38.25, P=0.0003), but only moderately b-caro-tene (F1,8=7.01, P=0.03). The relative proportion oflutein and zeaxanthin in eggs laid after treatment bycarotenoid-fed females was 5.8:1 whereas it was 3.4:1 forcontrol females.

Total carotenoids

Before treatment, there was no significant differencebetween groups in carotenoid deposition in eggs(F1,66=0.01, P=0.92), and carotenoid concentration infirst egg yolk was on average 19.8±1.6 lg/g (range: 6.5–80.6 lg/g). There was a significant effect of femaletreatment on the difference between after- and before-treatment eggs (repeated measures ANOVA, contrasts:

F1,66=13.94, P=0.0004), total carotenoid content inafter-treatment eggs being higher for carotenoid(33.7±3.4 lg/g, range: 3.0–82.0 lg/g) than controlfemales (21.0±1.6 lg/g, range: 9.0–56.9 lg/g)(F1,67=12.38, P=0.0008) (Fig. 1a). In the carotenoidgroup, total carotenoid concentration in after-treatmenteggs increased with the quantity of experimental foodeaten by females, while no such increase was observed inthe control group (Fig. 2). In the control group, therepeatability of carotenoid concentration between thefirst egg and the egg post-treatment was r=0.66,P<0.0001. Carotenoid concentration in first egg yolkwas not significantly related to brightness of the yellowplumage of the mother (F1,41=1.72, P=0.19). Carot-enoid concentration in after-treatment egg yolk was notsignificantly related to brightness of the yellow plumage(F1,39=1.78, P=0.19), and there was no significantinteraction between female colour and feeding treatment(F1,39=0.19, P=0.67). Likewise, no significant rela-tionships were found between yolk carotenoids and hueof female plumage (carotenoid concentration in first eggyolk: F1,41=0.46, P=0.50, carotenoid concentration inafter-treatment egg yolk: F1,38=0.17, P=0.68; interac-tion with treatment: F1,38=0.09, P=0.76).

Vitamins

Vitamin A (retinol) concentration in the first egg yolk inthe carotenoid group was higher than in the control(F1,24=5.46, P=0.03, model controlling for clutch size),and concentration in the post-treatment egg yolk in thecarotenoid group was also higher than in the controlgroup (F1,25=11.07, P=0.003, model controlling forclutch size) (Fig. 1b). However, female feeding treatmentdid not significantly affect the difference between after-and before-treatment eggs in vitamin A concentration(repeated measures ANOVA, contrasts: F1,24=0.01,P=0.93, model controlling for clutch size), which alsodid not vary significantly with laying sequence (repeated

Fig. 1 Mean±1SE antioxidant concentration (lg/g) in the yolk ofeggs produced before and after treatment by blue tits receivingdifferent experimental diets: a Carotenoid concentration (caroten-oid treatment: n=34, and control group: n=35). b Vitamin A(retinol) concentration (carotenoid treatment: n=14 and controlgroup n=14). c Vitamin E (summed d-, c-, a-tocopherol)concentration (carotenoid treatment: n=14 and control group:n=14)

Fig. 2 Carotenoid concentration (lg/g) in the yolk of post-treatment egg laid by supplemented females as a function of thequantity of experimental food eaten. Carotenoid group:F1,32=9.33, R2=0.22, slope estimate ± SE=20.9±6.8, P=0.004.Control group: F1,33=0.04, P=0.84

measures ANOVA, univariate test for within subjecteffect, source: time F1,24=0.68, P=0.42).

In the case of vitamin E, carotenoid and controlgroups were not significantly different from each otherbefore treatment (F1,24=1.66, P=0.21, model control-ling for clutch size). Yolk concentration of vitamin E didnot vary significantly with laying sequence (repeatedmeasures ANOVA, univariate test for within subjecteffect, source: time F1,24=1.03, P=0.32). Female feed-ing treatment did not significantly affect the differencebetween after- and before-treatment eggs in vitamin Econcentration (repeated measures ANOVA, contrasts:F1,24=2.16, P=0.15, model controlling for clutch size),and both treatment groups showed comparable con-centrations of vitamin E in post-treatment eggs(F1,25=0.56, P=0.46, model controlling for clutch size)(Fig. 1c).

Egg hatchability

The proportion of hatchlings in relation to the numberof egg laid did not differ significantly between treatments(logistic regression: F1,64=3.63, P=0.06), and was onaverage 84.6±3.8% in the carotenoid group and91.9±1.6% in the control group.

Nestling characteristics

Body condition

Nestling tarsus length was affected by female treatment(F1,225=4.70, P=0.03, Wald’s Z=2.97, P=0.001;Likelihood ratio test: v21 df=62.16, P<0.0001): nest-lings from carotenoid-fed females were larger (meantarsus length ± SE: 19.1±0.1 mm) than nestlings fromcontrol-fed females (18.8±0.1 mm).

Nestling body mass was not significantly affectedby female feeding treatment (F1,224=0.31, P=0.58,

carotenoid group: 10.95±0.10 g, control group:10.88±0.07 g, model controlling for tarsus lengthF1,224=79.11, P<0.0001, Wald’s Z=3.21, P=0.0007;Likelihood ratio test: v21 df=113.8, P<0.0001).

The proportion of leukocytes in nestling blood(square-root-transformed) was affected by female treat-ment (F1,156=16.44, P<0.0001) and its interaction withnestling age (age main effect: F1,156=5.78, P=0.02,interaction with treatment: F1,156=14.29, P=0.0002,model controlling for body mass: F1,156=15.57,P=0.0001 and nest effect: Wald’s Z=1.72, P=0.04;Likelihood ratio test: v21 df=8.72, P=0.003). Runningthe model excluding individuals for which the error inestimation of the proportion of leukocytes in blood wasabove the mean value of 0.26 or median value of 0.22(n=113 and 81 nestlings used in the analysis, respec-tively) provided qualitatively similar estimations of ef-fects as those presented here. In nestlings fromcarotenoid-fed females the proportion of leukocytes inthe blood remained constant with age, while it increasedwith age in nestlings from control fed females (Table 2and Fig. 3). Nestlings from carotenoid-fed females had ahigher proportion of leukocytes in their blood(mean ± SE: 0.0146±0.0008) than nestling from con-trol-fed females (0.0122±0.0007) (difference of leastsquares means: t=2.81, df=156, P=0.006). However,this difference was not biologically significant as it wasless than the precision in the estimation of mean values(carotenoid group: precision of 0.004 and control group:precision of 0.003).

Sedimentation rate increased with hatching date(F1,158=6.01, P=0.01) and decreased with nestling age(F1,158=7.5, P=0.007), but was not significantly af-fected by female feeding treatment (F1,158=0.00,P=0.99, carotenoid group: 64.2±0.1%, control group:68.2±0.8%, model controlling for hematocritF1,158=56.65, P<0.0001, Wald’s Z=2.76, P=0.003;Likelihood ratio test: v21 df=122.85, P<0.0001).

Plasma antioxidants

Carotenoid concentration in nestling plasma was notsignificantly affected by female feeding treatment (log-transformed concentration: F1,79=0.75, P=0.39, carot-enoid group: 17.01±3.23 lg/ml, control group:26.11±5.0 lg/ml) (Wald’s Z=1.32, P=0.09; Likeli-hood ratio test: v21 df=5.98, P=0.01). Concentration ofvitamin A in nestling plasma was positively correlatedwith carotenoid concentration (log-transformed con-centrations: F1,29=25.87, P<0.0001), but was not sig-nificantly affected by female treatment (F1,29=1.59,P=0.22, carotenoid group: 0.53±0.05 lg/ml, controlgroup: 0.66±0.09 lg/ml) (Wald’s Z=1.28, P=0.10;Likelihood ratio test: v21 df=5.54, P=0.02). Similarly,vitamin E concentration in nestling plasma was posi-tively correlated with carotenoid concentration (log-transformed concentration: F1,36=109.40, P<0.0001)and decreased with nestling age (F1,36=8.41, P=0.006)

Fig. 3 Proportion of leukocytes in nestling blood as a function ofage and female feeding treatment (statistics are given in Table 3).For clarity of the figure, data points are separated around agevalues according to female treatment (carotenoid group on the leftand control group on the right), but they actually correspond to thesame nestling age

but was not significantly affected by female treatment(F1,36=2.69, P=0.11, carotenoid group: 1.34±0.84 lg/ml, control group: 4.07±1.92 lg/ml).

Feather colour

Feather brightness in the yellow part of the spectrumwas affected by female treatment (F1,218=5.08, P=0.02,Wald’s Z=3.20, P=0.0007; Likelihood ratio test:v21 df=132.13, P<0.0001): nestlings from carotenoid-fed females were more brightly coloured (mean bright-ness in the yellow ± SE: 3104±77) than nestlings fromcontrol-fed females (2858±77). However, feather colourhue was not significantly affected by female treatment(F1,217=0.08, P=0.78, carotenoid group: 86.8±1.3�,control group: 85.2±1.4�), but increased with nestlingage (F1,217=6.54, P=0.01, Wald’s Z=3.07, P=0.001;Likelihood ratio test: v21 df=99.83, P<0.0001).

Fledging success

The proportion of nestlings fledged in relation to thenumber of hatchlings did not differ significantly betweentreatments (logistic regression: F1,25=2.58, P=0.12),and was on average 97.3±2.0% in the carotenoid groupand 93.3±2.8% in the control group.

Breeding female characteristics

Body condition

Adult female body mass decreased with nestling age(F1,42=10.45, P=0.002), but was not significantly re-lated to laying date (F1,42=2.27, P=0.14) and was notsignificantly affected by female feeding treatment(F1,42=0.00, P=0.95; carotenoid group: 10.95±0.19 g;control group: 10.84±0.11 g).

Plasma antioxidants

Carotenoid concentration in adult female plasma at thetime of chick rearing was not significantly related tocarotenoid investment in egg yolk before treatment (log-transformed concentrations: F1,33=1.76, P=0.19) or inegg yolk after treatment (log-transformed concentra-tions: F1,34=0.76, P=0.39). Total carotenoid concen-tration in adult female plasma was not significantlyaffected by feeding treatment (F1,35=0.69, P=0.41;carotenoid group: 12.4±3.5 lg/ml; control group:8.4±2.3 lg/ml).

Vitamin A concentration in adult female plasma atthe time of chick rearing was affected by feeding treat-ment (F1,33=5.60, P=0.02; model controlling for plas-ma carotenoid concentration (log-transformed):F1,33=22.65, P<0.0001 and tarsus length: F1,33=6.22,P=0.02) and was significantly higher in the carotenoidgroup (1.32±0.08 lg/ml) than in the control group(1.09±0.1 lg/ml). Furthermore, vitamin A concentra-tion in adult female plasma was not significantly relatedto the concentration of vitamin A in egg yolk before(F1,12=0.06, P=0.81) or after treatment (F 1,12=0.19,P=0.67).

Vitamin E concentration in breeding female plasmawas related to plasma carotenoid concentration (log-transformed concentrations: F1,30=32.86, P<0.0001),laying date (F1,30=16.00, P=0.0004), and the interac-tion between tarsus length and feeding treatment (tarsuslength main effect: F1,30=1.87, P=0.02; female feedingtreatment main effect: F1,30=8.07, P=0.008; interac-tion: F1,30=7.92, P=0.009). There was no difference inmean vitamin E concentration between feeding groups(carotenoid group: 3.73±1.51 lg/ml; control group:6.37±2.41 lg/ml; comparison of least squares means:t=�1.27, df=30, P=0.21). In both feeding groupsvitamin E concentration increased with carotenoidconcentration in plasma. However, in control-fed adultfemales vitamin E concentration also increased withlaying date and decreased with female body size, while

Table 2 Mixed linear model of multiple regression of the proportion of leukocytes in nestling blood (square-root-transformed) on nestlingage while controlling for body mass, as a function of female feeding treatment

Parameter Treatment

Carotenoida Controlb

Wald’s Z p Wald’s Z p

Covariance parameter estimatesNest 1.14 0.13 1.30 0.09Residual 5.00 <0.0001 6.75 <0.0001

Effect Estimate ± SE df t p Estimate ± SE df t p

Fixed effects parameter estimatesIntercept 0.063±0.035 55 1.78 0.08 �0.070±0.029 100 �2.39 0.02Body mass 0.0078±0.0030 55 2.59 0.01 0.0062±0.0022 100 2.85 0.005Nestling age �0.0017±0.0020 55 �0.85 0.40 0.0072±0.0014 100 4.98 <0.0001

aLikelihood ratio test v21 df=4.42, P=0.03bLikelihood ratio test v21 df=4.68, P=0.03

no such relationships were found in carotenoid-fed adultfemales (Table 3).

Nestling provisioning

Feeding rate (square-root-transformed) of nestlings bytheir mother was not significantly affected by femalefeeding treatment (F1,87=0.09, P=0.76, carotenoidgroup: 8.11±1.32 visits/h; control group: 8.81±1.44visits/h) nor by time of day (F1,87=2.78, P=0.09) orbrood size (F1,89=3.11, P=0.08) (Wald’s Z=2.44,P=0.01, Likelihood ratio test: v21 df=8.18, P=0.004).

Discussion

The amount of carotenoids available to a female forinvestment in her eggs depends on carotenoid avail-ability in the food, but also on female foraging,absorption and metabolising efficiency (Olson andOwens 1998; Møller et al. 2000; Surai 2002). Femalesfrom the carotenoid fed group increased yolk caroten-oid content compared to controls by a factor 1.5. Thisincrease is consistent with previous supplementationexperiments both in wild (Blount et al. 2002) andcaptive birds (Surai and Speake 1998; Surai and Sparks2001; Bortolotti et al. 2003). Furthermore, yolk carot-enoid concentrations in eggs laid by carotenoid sup-plemented females remained within the natural rangefor this species as found in eggs laid before treatment,as well as in eggs laid by non-manipulated birds duringthe same or previous breeding seasons (author’sunpublished data). If blue tit reproductive physiology isclose to that of the zebra finch (Taeniopygia guttata),we can assume that yolk transfer to the follicle iscompleted 24 h before laying (Salvante and Williams2002). Thus the second egg is not likely to be affectedby female treatment, but we expect carotenoid transferto the yolk to increase from the third egg, i.e. 48 h afterthe female started to feed on experimental food (fordata on chicken, see Surai and Speake 1998). There-fore, dietary supplementation of females with carote-noids during egg laying has probably resulted in agradual increase in yolk pigment concentration withlaying sequence. When carotenoids were readily avail-able in the diet, females were able to deposit morepigments in their eggs. Thus the most important limitmay not be metabolising efficiency i.e. individual ability

to absorb and use carotenoids (Olson and Owens1998), but more likely carotenoid availability, due topigment scarcity in the environment, female ability toforage on relevant prey items, and/or high prevalenceof parasites which decrease carotenoid assimilationfrom the diet or cause carotenoids to be used for theproduction of immune responses (reviewed in Surai2002). In the closely related great tit (Parus major), eggyolk carotenoid concentration has been shown to varydepending on breeding habitat, with carotenoid con-centration in eggs from a rural population being twicethat in a urban population (Horak et al. 2002). Thisdifference was hypothesised to be a consequence offemales in the urban habitat being more subject tooxidative stress and/or carotenoid availability beinghigher in the rural habitat (Horak et al. 2002). Carot-enoid availability during egg laying has also beenshown to be limited in the lesser black-backed gull(Larus fuscus) in which female supplementation withcarotenoids increased the probability to lay a replace-ment clutch (Blount et al. 2004).

It has been suggested that carotenoid displays in fe-male birds may advertise their ability to produce highquality eggs (Blount et al. 2000). There was no rela-tionship between yellow plumage colour of adult femalesand carotenoid deposition in eggs, and no differentialeffect of feeding treatment depending on female colour.Carotenoid availability during moult and/or femalecapacity to develop bright plumage thus does not reflectpigment availability during reproduction or femaleability to invest these pigments in eggs. This suggeststhat female colour does not constitute a reliable signal ofthe ability of the female to produce eggs with highconcentration of antioxidants. This may arise from highvariability in carotenoid availability across seasons,resulting from variability in the occurrence of caroten-oid-rich food and in female ability to obtain pigmentsfrom the environment across seasons, but also from thetrade-off in allocation of carotenoids to different func-tions. It has been shown that great tit females experiencean increase in resting metabolic rate of 12% during nestbuilding (corresponding to the ovary-oviduct recrudes-cence period), 27% during egg laying and 20% duringchick feeding period as compared to that in winter(Nilsson and Raberg 2001). These patterns are probablythe same in the blue tit, resulting in an increase in oxi-dative stress and a need for carotenoids to be used forantioxidant function. Furthermore, seasonal variation in

Table 3 Multiple regressions of female vitamin E concentration in plasma (log-transformed) in relation to plasma carotenoid concen-tration, laying date and tarsus length, in females from carotenoid (n=18) and control feeding groups (n=19)

Treatment Carotenoid Control

Effect Estimate±SE t P Estimate±SE t P

Intercept �44.03±22.43 �1.96 0.07 25.87±11.76 2.20 0.04Total carotenoid (log-transformed) 1.33±0.32 4.11 0.001 0.89±0.24 3.73 0.002Laying date 0.27±0.14 1.95 0.07 0.21±0.05 4.43 0.0005Tarsus length 0.81±0.98 0.82 0.43 �2.53±0.65 �3.92 0.001

the prevalence of parasites causes hosts to invest more inimmune function during the breeding season than dur-ing non-breeding (Møller et al. 2003). As a consequence,the requirements of carotenoids for immune function arelikely to be higher, causing these pigments to be morelimiting during reproduction than during moult.

During development, fat-soluble antioxidants aretransferred from the yolk to the tissues of the embryoand particularly to the liver (Surai et al. 1998; Surai andSpeake 1998), which will concentrate and redistributethem after hatching (Surai et al. 2001a). Therefore, aswith vitamin E (Surai et al. 1999), an increase in carot-enoid concentration first in egg yolk and later in the liverof the newly hatched chick could be beneficial formaintaining its antioxidant system. This would be par-ticularly important during hatching, as this process isassociated with increased oxidative stress (Surai et al.1999; Surai 2002). Eggs of chicken high in carotenoidsare characterised by higher hatchability in comparisonto eggs with low carotenoid levels (Inborr 1996; Llau-rado et al. 1997; Kemp et al. 2001). However, thesepositive effects of carotenoids on hatchability found inpoultry studies have not been confirmed in wild birds todate. In an experimental manipulation of egg yolk car-otenoids in barn swallow (Hirundo rustica) (Saino et al.2003) there was no significant effect of an increase inyolk carotenoids on hatchability, and a similar resultwas found in our study of blue tits. These discrepanciesmay be explained by the fact that poultry studies con-trasted dramatically different carotenoid concentrationsin eggs, and that carotenoid concentrations in eggs ofwild birds have generally been reported to be naturallyhigher than in captive birds (reviewed in Surai 2002).

A long lasting effect of yolk carotenoids on nestlingphenotypic development may be expected, as nestlingshatched from carotenoid-rich eggs will benefit for alonger period from more yolk-derived carotenoidsstored in the liver. Furthermore, the concentration ofyolk carotenoids influences the subsequent capacity ofthe chick to assimilate dietary carotenoids and accu-mulate them in its tissues (Koutsos et al. 2003). Atfledging, nestlings from carotenoid-supplemented fe-males were characterised by longer tarsi, more leuko-cytes in their blood and more brightly colouredjuvenile plumage than controls. We did not detect anyeffect of female treatment on nestling circulating an-tioxidants. However, nestlings were blood sampledbefore fledging when 15 days old, and in poultrystudies effects of yolk carotenoid concentration oncirculating levels of carotenoids in nestling have gen-erally been observed before 9 days of age (reviewed inSurai et al. 2001a; Surai 2002). An increase in theconcentration of carotenoid pigments first in egg yolk,and later in the liver may facilitate growth in nestlingsby preventing oxidative stress which could impairphysiological functions like efficient nutrient utilisationor result in damage to lipid-rich tissues. In addition,carotenoids are involved in gap-junction and cellcommunication, gene expression and protection

against mutation (reviewed in Burri 2000; Surai 2002).This may have enabled nestlings from carotenoid-fedfemales to achieve larger body size at fledging thancontrols. This may also result from the naturallyhigher concentration in vitamin A of eggs from thecarotenoid group, due to the ability of vitamin A toregulate cell growth and promote cell differentiation(Zile 2001). However, these differences in tarsus lengthare more likely to result from the increase in carot-enoid concentration in eggs, as the treatment effects onimmune system and plumage colour of nestlings sug-gest specific action of carotenoids. Indeed, the increasein yolk concentration of carotenoids due to treatmentseems to have affected the development of the immunesystem in nestlings. There was no difference betweennestlings from the two treatment groups in sedimen-tation rate thus nestlings from carotenoid supple-mented nests were unlikely to be more infected thancontrols (e.g. Sturkie 1986; Svensson and Merila 1998).However, in nestlings from control nests, the propor-tion of leukocytes in blood increased with age untilfledging, while in nestlings from carotenoid supple-mented nests, the proportion of leukocytes in bloodwas constant and 12 days old nestlings already had thesame level of leukocytes as fledglings. Carotenoids playimportant roles in immuno-regulation and immuno-stimulation in vertebrates (reviewed in Chew 1993;Møller et al. 2000), and an increase in yolk carote-noids has been shown to enhance chick spleen andbursal lymphocyte function in vitro (Haq et al. 1996).The increase in yolk carotenoids may have allowed amore rapid development of the immune system andmaturation of immune cells in nestlings from carot-enoid-fed females. This would provide a functionalexplanation of the results of a recent study showingthat barn swallow nestlings originating from eggsexperimentally enriched in carotenoids showed higherantibody response at fledging (Saino et al. 2003). Bluetit nestlings from carotenoid-supplemented femalesalso grew brighter yellow feathers than control nes-tlings. In the blue tit, breast feather colour is deter-mined by the xanthophyll pigments lutein andzeaxanthin (Partali et al. 1987). Nestlings from carot-enoid-fed females had access to more yolk-derivedcarotenoids and therefore may have been able to de-posit more pigments in feathers than nestlings fromcontrol females and/or to grow feather of betterstructural quality. The differences in nestling tarsuslength, dynamics of increase in leukocytes abundancein blood and feather colour between carotenoid andcontrol fed females are probably mediated throughpre-hatching maternal effects and yolk carotenoidcontent, but may also have occurred as a result of adifference in post-hatching maternal effects in relationto treatment. However, there was no evidence thatfemale feeding treatment affected subsequent parentalinvestment: females from the two treatment groupsshowed comparable investment in brood rearing, asreflected by brood provisioning rates.

The increase in deposition of carotenoids in eggs as aresult of experimental increase in pigment availability tofemales during laying influenced nestling growth, im-mune function and ability to develop bright carotenoid-based signals. Although these aspects of nestling phe-notype may be associated with fitness, there was nodetectable difference in fledging success between broodsof carotenoid and control fed females. However, fitnessbenefits of increasing carotenoid investment in egg yolkmight be seen in the long-term, i.e. on offspring survivalto first reproduction.

Increasing carotenoid availability for females duringegg laying may not only have an effect on egg yolkcomposition, but also on the long-term on antioxidantlevels in adult females at the time of chick rearing. Therewas no detectable difference in plasma carotenoid levelsbetween carotenoid and control fed females. However,our treatment only lasted 9–14 days, and an increase inplasma carotenoid levels is generally obtained only afterhigh doses or long-term feeding with supplementarycarotenoids (Blount et al. 2002; reviewed in Bortolottiet al. 2003). In addition, plasma carotenoid levels infemales during the nestling phase were unrelated tolevels of carotenoids deposited in egg yolk. Plasmacarotenoid concentration is regulated by the liver, andprobably varies on a short to medium time scaledepending on quality/quantity of food ingested duringprevious days (Surai 2002). Thus plasma carotenoidconcentration at the time of chick rearing may beindependent of that during egg laying, due to possibledifferences in food constraints, linked for example to theemergence of carotenoid-rich food such as caterpillars(Partali et al. 1987; van Noordwijk et al. 1995; Visseret al. 1998). All carotenoids ingested may not have beenabsorbed and/or metabolised, and females may not havehad any carotenoid reserves left after egg production.More likely, carotenoid-fed females did not transfer allsupplemental carotenoids ingested to egg yolk, andstored the remaining pigments in the liver where carot-enoids may be involved in antioxidant function and insynergistic and recycling interactions with other an-tioxidants (reviewed in Edge et al. 1997; Surai et al.2001a).

Carotenoid-fed females had a higher concentration ofretinol in blood than controls. Although plasma retinollevels in adult female were unrelated to maternalinvestment of this vitamin in eggs, concentrations ofvitamin A in eggs laid before treatment by carotenoid-fed females were already higher compared to controlfemales. It is therefore difficult with the present data tobe sure that carotenoid supplementation during egglaying per se has led to an increase in plasma vitamin Ain females at the time of chick rearing. However, apossible long-term effect of carotenoid supply during egglaying on female antioxidant level is suggested by dif-ferences in vitamin E levels between females from thetwo treatment groups. In the control group, vitamin Econcentration in female plasma increased with layingdate and decreased with body size, indicating seasonal

and/or metabolic constraints on availability of thisvitamin to females. There may indeed be seasonal vari-ation in the quality of food available to adult femalesduring the breeding season (Partali et al. 1987; Visseret al. 1998). Alternatively, there may be differential rateof use or need for antioxidants as a function of meta-bolic rate, which is related to body size (Sturkie 1986).These relationships were not detectable in carotenoid-fed females, which suggests that dietary supplementationwith carotenoids reduced such constraints, possibly dueto the specific interactions existing between carotenoidsand vitamin E (reviewed in Surai 2002). Increasing theavailability of carotenoids during egg laying did notdirectly affect plasma carotenoid levels, but neverthelessincreased availability of vitamin E for females later inthe season. This effect may be mediated through thequantity of pigments stored in the body, although dataon carotenoid content in the liver and other tissues infemales from both treatment groups would be needed totest this prediction. There was no detectable effect offemale treatment on female body condition, as reflectedby body mass. However, benefits of an increase inantioxidant levels may be seen in terms of resistance tooxidative stress in various stress conditions and immunefunction (i.e. ability to mount an immune response), andthis may have long-term consequences for survival andsenescence.

Previous studies have shown that carotenoids arelimiting for egg laying female birds (Horak et al. 2002;Blount et al. 2004). The results of this study yield fur-ther evidence that carotenoid availability may be lim-iting during egg laying, and that this may have short-term effects in terms of egg yolk composition, but alsodelayed effects in terms of antioxidant levels in femalesand nestling quality. Indeed, increasing the availabilityof carotenoids to egg laying females resulted in an in-crease in carotenoid deposition in eggs and in positiveeffects on the development of nestling phenotypic traitsclosely associated with fitness. Maternal effects increasephenotypic plasticity in nestling traits, and therefore actas a mechanism for adaptive trans-generational phe-notypic response to environmental variability (Mous-seau and Fox 1998). Carotenoid availability may varyfrom egg laying to chick rearing, and females that areable to obtain more carotenoids during laying will en-hance the fitness of their nestlings, enabling them tobetter cope with unexpected poor conditions duringrearing. In addition, in natural and non-experimentalconditions, investment in egg carotenoid content maybe costly to females if they have limited access to car-otenoids. Females may have to balance the benefits ofinvesting in egg quality against the potential costs ofimpairing their own future antioxidant protection(Blount et al. 2000). Indeed, in this study, femaleshaving increased access to carotenoids during layingbenefited from increased availability in vitamin E dur-ing chick rearing. However, this potential trade-off andits consequences need to be experimentally investigated,for example through inducing females to produce an

immune response or to increase their investment incurrent reproduction, either during egg laying or chickrearing. More generally, carotenoids may provide amechanism underlying the already well-studied trade-offs between investment in reproduction and the im-mune system, future reproduction or survival (e.g.Linden and Møller 1989; Monaghan and Nager 1997;Monaghan et al. 1998; Nordling et al. 1998; Visser andLessells 2001). Any activity likely to increase metabolicrate, such as egg laying, incubation, brooding or chickrearing, will simultaneously increase the level of oxi-dative stress experienced by females and the rate of useof carotenoids, thereby potentially decreasing theamount of carotenoids available for other functions. Inaddition, this study suggests that there may be consid-erable variation in carotenoid availability, not onlyacross seasons, but also among different stages of thebreeding season. Further studies are needed to evaluatecarotenoid availability and physiological needs forcarotenoids in different environments at different timesof the year.

Acknowledgements Bird supplemental feeding, capture and ringingas well as egg and blood samplings were done under Ringing Li-cence (CRBPO Museum National d’Histoire Naturelle) and specialauthorisation for protected species (DIREN Champagne-Ard-enne). Kemin Europa kindly provided samples of carotenoids(OroGlo Layer Dry). Fieldwork was done under permission ofParc Naturel Regional de la Foret d’Orient and Office National desForets. We thank B. Auclair, S. Cassier and V. Coutouly for help inthe field. J. Aviles, F. Cezilly, A. and V. Huste and three anony-mous referees provided helpful comments on earlier version of themanuscript. J. M. Rossi developed software for reflectance spec-trum analysis and colour parameter calculation. C.B. was sup-ported by a doctoral grant from Ministere de l’Education et de laRecherche and P.F.S. was supported by the Scottish ExecutiveEnvironment and Rural Affairs Department.

References

Blount JD, Houston DC, Møller AP (2000) Why egg yolk is yellow.Trends Ecol Evol 15:47–49

Blount JD, Surai PF, Nager RG, Houston DC, Møller AP, TrewbyML, Kennedy MW (2002) Carotenoids and egg quality in thelesser black-backed gull (Larus fuscus): a supplemental feedingstudy of maternal effects. Proc R Soc Lond B 269:29–36

Blount JD, Houston DC, Surai PF, Møller AP (2004) Egg-layingcapacity is limited by carotenoid pigment availability in wildgulls Larus fuscus. Proc R Soc Lond B 271(Suppl):S79–S81

Bortolotti GR, Negro JJ, Surai PF, Prieto P (2003) Carotenoids ineggs and plasma of red-legged partridges: effects of diet andreproductive output. Physiol Biochem Zool 76:367–374

Burnham KP, Anderson DR (1998) Model selection and inference:a practical information-theoretic approach. Springer, BerlinHeidelburg New York

Burri BJ (2000) Carotenoids and gene expression. Nutrition16:577–578

Chew BP (1993) Role of carotenoids in the immune response.J Dairy Sci 76:2804–2811

Edge R, McGarvey DJ, Truscott TG (1997) The carotenoids asanti-oxidants—a review. J Photochem Photobiol B 41:189–200

Endler JA (1990) On the measurement and classification of col-our in studies of animal colour patterns. Biol J Linn Soc41:315–352

Freckelton RP (2002) On the misuse of residuals in ecology:regression of residuals vs. multiple regression. J Anim Ecol71:542–545

Goldstein H (2003) Multilevel statistical models, 3rd edn. HodderArnold, London

Goodwin TW (1984) The biochemistry of carotenoids. Chapmanand Hall, London

Haq A, Bailey CA, Chinnah A (1996) Effect of b–carotene, can-thaxanthin, lutein, and vitamin E on neonatal immunity ofchicks when supplemented in the broiler breeder diets. Poult Sci75:1092–1097

Hill GE, Inouye CY, Montgomerie R (2002) Dietary carotenoidspredict plumage coloration in wild house finches. Proc R SocLond B 269:1119–1124

Horak P, Vellau H, Ots I, Møller AP (2000) Growth conditionsaffect carotenoid-based plumage coloration of great tit nes-tlings. Naturwissenschaften 87:460–464

Horak P, Surai PF, Møller AP (2002) Fat-soluble antioxidants inthe eggs of great tits Parus major in relation to breeding habitatand laying sequence. Avian Sci 2:123–130

Inborr J (1996) Astaxantin—a carotenoid with great potential.Poult Int 35:54–60

Kemp C, Wylie L, Fisher C (2001) Broiler breeder nutrition,nutrient transfer and broiler performance. In: Proceedings of13th European symposium on poultry nutrition. BlankenbergeBelgium, pp 61–67

Koutsos EA, Clifford AJ, Calvert CC, Klasing KC (2003) Maternalcarotenoid status modifies the incorporation of dietary carote-noids into immune tissues of growing chickens (Gallus gallusdomesticus). J Nutr 133:1132–1138

Lessells CM, Boag PT (1987) Unrepeatable repetabilities: a com-mon mistake. Auk 104:116–121

Linden M, Møller AP (1989) Cost of reproduction and covariationof life history traits in birds. Trends Ecol Evol 4:367–371

Llaurado LI, Francesch A, Hernandez JM, Brafau J (1997) Effectof canthaxanthin supplementation on the hatchability of eggs ofbroiler breeders. In: Proceedings of 11th European symposiumon poultry nutrition, Faaborg, Denmark, pp 280–282

Møller AP, Biard C, Blount JD, Houston DC, Ninni P, Saino N,Surai PF (2000) Carotenoid-dependant signals: indicators offoraging efficiency, immunocompetence or detoxification abil-ity? Avian Poult Biol Rev 11:137–159

Møller AP, Erritzøe J, Saino N (2003) Seasonal changes in immuneresponse and parasite impact on hosts. Am Nat 161:657–671

Monaghan P, Nager RG (1997) Why don’t birds lay more eggs?Trends Ecol Evol 12:270–274

Monaghan P, Nager RG, Houston DC (1998) The price of eggs:increased investment in egg production reduces the offspringrearing capacity of parents. Proc R Soc Lond B 265:1731–1735

Mousseau TA, Fox CW (eds) (1998) Maternal effects as adapta-tions. Oxford University Press, New York

Nilsson J-A, Raberg L (2001) The resting metabolic cost ofegg-laying and nestling feeding in great tits. Oecologia128:187–192

van Noordwijk AJ, McCleery RH, Perrins CM (1995) Selection forthe timing of great tit breeding in relation to caterpillar growthand temperature. J Anim Ecol 64:451–458

Nordling D, Andersson M, Zohari S, Gustafsson L (1998)Reproductive effort reduces specific immune response andparasite resistance. Proc R Soc Lond B 265:1291–1298

Olson VA, Owens IPF (1998) Costly sexual signals: are carotenoidsrare, risky or required? Trends Ecol Evol 13:510–514

Ots I, Murumagi A, Horak P (1998) Haematological health stateindices of reproducing great tits: methodology and sources ofnatural variation. Funct Ecol 12:700–707

Partali V, Liaaen-Jensen S, Slagsvold T, Lifjeld JT (1987) Carote-noids in food chain studies. II The food chain of Parus spp.Monitored by carotenoid analysis. Comp Biochem Physiol B87:885–888

Saino N, Stradi RD, Ninni P, Pini E, Møller AP (1999) Carotenoidplasma concentration, immune profile and plumage ornemen-

tation of male barn swallows (Hirunda rustica). Am Nat154:441–448

Saino N, Ferrari RP, Romano M, Martinelli R, Møller AP (2003)Experimental manipulation of egg carotenoids affects immunityof barn swallow nestlings. Proc R Soc Lond B 270:2485–2489

Saks L, Ots I, Horak P (2003) Carotenoid-based plumage colora-tion of male greenfinches reflects health and immunocompe-tence. Oecologia 134:301–307

Salvante KG, Williams TD (2002) Vitellogenin dynamics duringegg-laying: daily variation, repeatability and relationship withegg size. J Avian Biol 33:391–398

Slagsvold T, Lifjeld JT (1985) Variation in plumage colour of thegreat tit Parus major in relation to habitat, season and food.J Zool 206:321–328

Stearns SC (1989) Trade-offs in life history evolution. Funct Ecol3:259–268

Sturkie PD (1986) Avian physiology, 4th edn. Springer, BerlinHeidelburg New York

Surai PF (2000) Effect of selenium and vitamin E content of thematernal diet on the antioxidant system of the yolk and thedeveloping chick. Brit Poult Sci 41:235–243

Surai PF (2002) Natural antioxidants in avian nutrition andreproduction. Nottingham University Press, Nottingham

Surai PF, Sparks NHC (2001) Comparative evaluation of the effectof two maternal diets on fatty acids, vitamin E and carotenoidsin the chick embryo. Brit Poult Sci 42:252–259

Surai PF, Speake BK (1998) Distribution of carotenoids from theyolk to the tissues of the chick embryo. JNutrBiochem9:645–651

Surai PF, Ionov IA, Kuchmistova EF, Noble RC, Speake BK(1998) The relationship between the levels of a-tocopherol andcarotenoids in the maternal feed, yolk and neonatal tissues:comparison between the chicken, turkey, duck and goose. J SciFood Agric 76:593–598

Surai PF, Speake BK, Noble RC, Sparks NHC (1999) Tissue-specific antioxidant profiles and susceptibility to lipid peroxi-dation of the newly hatched chick. Biol Trace Elem Res 68:63–78

Surai PF, Speake BK, Sparks NHC (2001a) Carotenoids in aviannutrition and embryonic development 2 Antioxidant propertiesand discrimination in embryonic tissues. J Poult Sci 38:117–145

Surai PF, Speake BK, Wood NAR, Blount JD, Bortolotti GR,Sparks NHC (2001b) Carotenoid discrimination by the avianembryo: a lesson from wild birds. Comp Biochem Physiol B128:743–750

Svensson E, Merila J (1998) Molt and migratory condition in bluetits: a serological study. Condor 98:825–831

Visser ME, Lessells CM (2001) The costs of egg production andincubation in great tits (Parus major). Proc R Soc Lond B268:1271–1277

Visser ME, van Noordwijk AJ, Tinbergen JM, Lessells CM (1998)Warmer springs lead to mistimed reproduction in great tits(Parus major). Proc R Soc Lond B 265:1867–1870

Zile MH (2001) Function of vitamin A in vertebrate embryonicdevelopment. J Nutr 131:705–708