Carotenoids are minor antioxidants for birds


  • D. Costantini,

    Corresponding author
    1. Dipartimento di Biologia Animale e dell’Uomo, University Sapienza, Viale dell’Università 32, I-00185 Rome, Italy;
    2. Division of Neuroanatomy and Behaviour, Institute of Anatomy, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland; and
      *Correspondence author. E-mail:
    Search for more papers by this author
  • A. P. Møller

    1. Laboratoire de Parasitologie Evolutive, CNRS UMR 7103, 6, Université Pierre et Marie Curie, Bât. A, 7ème étage, 7 quai St. Bernard, 7 Case 237, F-75252 Paris Cedex 05, France
    Search for more papers by this author

*Correspondence author. E-mail:

Carotenoids are fat-soluble natural pigments with antioxidant properties (e.g. Møller et al. 2000; Surai 2002; Krinsky & Yeum 2003), but they also have a number of other additional physiological functions, such as immunostimulation (Blount et al. 2003; Faivre et al. 2003; McGraw & Ardia 2003). Animals are incapable of synthesizing carotenoids de novo, and must obtain them through their diet. Carotenoids are often used in visual displays through deposition in skin or feathers. Given these multiple uses that all require substantial amounts of carotenoids for normal functioning, carotenoids have been suggested to be in limited supply for reproduction, health-related functions, or the expression of sexual colouration. For example, it has been suggested that carotenoids may limit vital functions, such as scavenging of free radicals, eliminating peroxides and enhancing immune function (production of lymphocytes, enhancement of phagocytic ability of neutrophils and macrophages, production of tumour immunity), in which they have been shown to be involved (Møller et al. 2000). In particular, bird species with carotenoid-based colouration have been hypothesized to trade antioxidant response against sexual signalling due to the fact that mounting an antioxidant response uses carotenoids that are no longer available for production of the sexual signal (von Schantz et al. 1999).

It is known that maternally transferred carotenoids are useful antioxidants in developing embryos. Carotenoids seem to play an important role in maintaining redox homeostasis during embryonic development and the first days after hatching (Surai 2002; McGraw, Adkins-Regan & Parker 2005). Together, these phases represent a critical period during which the ability to cope with free radicals depends to some extent on maternal effects, that is, antioxidants such as vitamins A and E, and carotenoids invested by the mother into the egg. These compounds will be pivotal for the immediate post-hatching period during which oxidative stress (imbalance toward pro-oxidants) increases because of exposure of the chick to atmospheric oxygen, to the shift to pulmonary respiration and to the increase in metabolic rate.

Recently, Hartley & Kennedy (2004) suggested that carotenoids are unlikely to have a main function as antioxidants in birds. They suggested that carotenoid-based colouration might not display antioxidant capacity of carotenoids themselves, but rather reflect (i) the concentration of other colourless antioxidants (e.g. vitamins A, C and E) that would mitigate against the oxidative decolouration of carotenoids making them available for sexual signalling and/or (ii) carotenoids for quite different functions, such as for their roles in the immune system or in the embryonic development. Consistently, a recent study on captive zebra finches Taeniopygia guttata showed that increased availability of melatonin, that is, a non-pigmentary antioxidant, actually enhanced the expression of the carotenoid-based sexual colouration, but had no effect on circulating carotenoids (Bertrand, Faivre & Sorci 2006).

Recent analyses of the antioxidant activity of carotenoids in birds produced conflicting results, with some studies showing a significant effect, while others did not. Here we briefly review these studies, quantify the magnitude of the effects, and test for heterogeneity in effects among studies.

Recent experimental evidence from free-living birds on the antioxidant properties of carotenoids (e.g. Costantini et al. 2006; Hõrak et al. 2006; Costantini, Fanfani & Dell’Omo 2007b; see Table 1 for a full list of papers) corroborates the hypothesis of Hartley & Kennedy (2004) that carotenoids might be minor antioxidants for birds. This recent evidence actually challenges previous emphasis for these compounds as important antioxidants, suggesting that the antioxidant role of xanthophylls (i.e. oxygenated carotenoids) has been exaggerated (Table 1). A fixed-effect meta-analysis based on the Pearson product moment correlation coefficient (Rosenthal 1991) as a measure of effect size derived from the test statistics of the different studies, using the formulae in Rosenthal (1994, table 16·1), weighting the z-transformed effect sizes by using (N – 3), provided an overall weighted estimate of effect size of –0·0048 that did not differ significantly from 0 (t =–0·11, df = 22, P = 0·91). This implies that carotenoids account for < 0·002% of antioxidant capacity. We can also test the hypothesis that the reported effect size was not small [sensu Cohen (1988) explaining 1% of the variance]. Indeed, the observed mean effect size differed significantly from an effect size of 0·10 (one-sample t-test, t = –2·38, df = 22, P = 0·026).

Table 1.  Summary of studies testing for antioxidant effects of carotenoids in birds. Effect size is the Pearson product-moment correlation coefficient derived from the formulae for different test statistics reported by Rosenthal (1994, table 16·1)
SpeciesExperimental studyTest statisticSample sizePearson rReference
Falco tinnunculusNegative covariation between reactive oxygen metabolites and carotenoids; field studyF1,104·53 = 0·16, P = 0·69N = 261 nestlings0·04Costantini et al. (2006)
Falco tinnunculusNegative covariation between antioxidant capacity (OXY test) and carotenoids; field studyF1,106·14 = 0·56, P = 0·46N = 261 nestlings–0·07Costantini & Dell’Omo (2006)
Falco tinnunculusSupplementation of carotenoids; decrease in reactive oxygen metabolites; field studyTreatment × repeated measure: F2,22 = 0·15, P = 0·86N = 61 nestlings (30 controls and 31 supplemented)0·08Costantini et al. (2007b)
Falco tinnunculusSupplementation of carotenoids; decrease in antioxidant capacity (OXY test); field studyTreatment × repeated measure: F2,22 = 0·11, P = 0·89N = 61 nestlings (30 controls and 31 supplemented)–0·07Costantini et al. (2007b)
Falco tinnunculusCarotenoid supplementation; reactive oxygen metabolites increased; captive conditionsTreatment × time: F4,56 = 6·76, P < 0·001N = 18; both adult males and females0·33Costantini et al. (2007a)
Falco tinnunculusCarotenoid supplementation; the antioxidant capacity (OXY test) did not respond; slightly decreased captive conditionsTreatment × time: F4,56 = 1·36, P = 0·26N = 18; both adult males and females–0·15Costantini et al. (2007a)
Gallus gallusLaying hens; carotenoid supplementation produced a small decrease of the antioxidant capacity (TEAC test); captive conditionst-test, t = –3·15, df = 7, P = 0·015 for plasma, t = –2·2, df = 7, P = 0·06 for serumN = 9; adult females–0·77, –0·41Cohen et al. (2007)
Larus fuscusYolk susceptibility to lipid peroxidation (TBARS-MDA test) is lower in eggs produced by carotenoid-supplemented females; field studyt = 2·41, df = 14, P = 0·03N = 160·54Blount et al. (2002a)
Larus fuscusCarotenoid-fed females had higher plasma antioxidant capacity (TEAC test); field studyF1,13 = 14·98, P = 0·002N = 15; adult females0·73Blount et al. (2002b)
Parus majorCorrelation between antioxidant capacity (BIOXYTECH AOP-490 test) and carotenoids; field studyr = –0·02, p = 0·9N = 37 adult females–0·02Tummeleht et al. (2006)
Parus majorCorrelation between antioxidant capacity (TEAC test) and carotenoids; field studyr = 0·10, P = 0·5N = 37 adult females0·10Tummeleht et al. (2006)
Parus majorTotal plasma carotenoids did not correlate with non-enzymatic antioxidant activity in plasma of adult great tits; field studyF 1,50 = 0·05, P = 0·83; r = –0·031N = 52–0·031Isaksson et al. (2007)
Parus majorTotal plasma carotenoids did not correlate with non-enzymatic antioxidant activity in plasma of nestling great tits; field studyF 1,45 = 1·36, P = 0·25; r = 0·17N = 470·17Isaksson et al. (2007)
Taeniopygia guttataSupplementation of carotenoids; response of antioxidant capacity (KRL test); captive conditionsF1,115 = 0·57, P = 0·45; N = 120; both males and females0·07Alonso-Alvarez et al. (2004)
Taeniopygia guttataSupplementation of carotenoids; correlation between change in plasma carotenoids and response of antioxidant capacity (KRL test); captive conditionsSlope ± SE = 0·054 ± 0·018, P = 0·003; the direction is positiveN = 120, both males and females0·28Alonso-Alvarez et al. (2004)
Taeniopygia guttataSupplementation of carotenoids; decrease in antioxidant capacity (KRL test) was modulated by carotenoid availability, as individuals with carotenoid-supplemented water maintained similar antioxidant capacity whatever the number of eggs laid; captive conditionsEffect of carotenoid availability: F1,102 = 8·52, P = 0·004; Number of eggs laid × carotenoid availability: F1,102 = 6·83, P = 0·01N = 1180·28, –0·25Bertrand et al. (2005)
Taeniopygia guttataCorrelation between yolk carotenoids in eggs and lipid peroxides (TBARS-MDA test); captive conditionsr2 = 0·35, P = 0·03N = 12–0·59McGraw et al. (2005)
Carduelis chlorisCorrelation between antioxidant capacity (TEAC test and BIOXYTECH AOP-490 test) and carotenoids; wild birds under captive conditionsr = –0·10–0·19, P = 0·2–1N = 25–55 adult males0·05Hõrak et al. (2006)
Carduelis chlorisCarotenoid supplementation decreased lipid peroxidation (MDA test); wild birds under captive conditionsCarotenoid: F1,70 = 6·96, P = 0·01; immune challenge × carotenoid supplementation: F1,70 = 0·11, P = 0·74N = 74, adult males0·30, 0·04Hõrak et al. (2007)
Carduelis chlorisCarotenoid supplementation did not modulate the effect of immune challenge on the dynamics of the antioxidant capacity (BIOXYTECH AOP-490 test); wild birds under captive conditionsTime × immune challenge × carotenoid supplementation: F2,130 = 1·82, P = 0·17N = 77, adult males–0·12Hõrak et al. (2007)

There was significant heterogeneity in effect size among results, using the procedure described by Rosenthal (1991) (χ2 = 65·34, df = 22, P < 0·001). The fail-safe number of studies needed to nullify the estimated mean effect (Rosenthal 1991) was 1692, showing that our conclusion of a small effect not significantly different from no effect at all was extremely robust. While this analysis was based on the individual estimates of effect sizes, analyses based on mean study or species estimates provided very similar conclusions (mean weighted effect size per study = 0·0766, χ2 = 42·29, df = 12, P < 0·001, fail-safe number = 403; mean weighted effect size per species = 0·0427, χ2 = 28·55, df = 5, P < 0·001, fail-safe number = 362). The significant heterogeneity implies that species differed significantly in effect size. Likewise, random effects models did not provide different conclusions (data not shown). We have no way of estimating publication bias directly (Møller & Jennions 2001), because we do not know how many studies have been initiated, but not published. However, given the weak effects and the mixture of positive and negative effects reported (see Table 1), we consider this problem to be of little significance. Given that all species studied had carotenoid-based sexual colouration, where such species are most likely to signal the physiological function of carotenoids (Møller et al. 2001), and that these species have higher levels of carotenoids than species without carotenoid-based colouration (Tella et al. 2004), this suggests that studies of species without carotenoid-based colouration are even less likely to show effects of carotenoids on antioxidant function.

The seemingly contradictory results concerning the antioxidant role of carotenoids, as reflected by the highly significant heterogeneity in effect size among studies, may be interpreted in at least two different ways: (i) results may reflect different methods and experimental protocols; or (ii) they may mirror interspecific differences that reflect the biology of the different species.

First, several methods have been developed to measure oxidative damage and antioxidant capacity of a biological sample (e.g. serum, urine). If an array of different markers is useful to thoroughly assess oxidative stress of a biological system, problems arise because specificity of the method to be used or its underlying biochemical rationale may make comparison of results difficult. For example, shortcomings can emerge in the quantification of malon-dialdehyde to assess lipid peroxidation in biological fluids or tissues because of its non-specificity (Chirico 1994). As for the quantification of antioxidant capacity, the assays (e.g. ORAC assay, OXY-adsorbent test) involving pro-oxidants (i.e. oxidants of pathological relevance) are preferred to those involving oxidants (e.g. FRAP assay, TEAC assay) that cannot have any biological relevance (Prior & Cao 1999). It should be noted, however, that most studies measuring serum/plasma antioxidant capacity by different methods converge to the same conclusion that carotenoids are minor antioxidants (see Table 1).

The second explanation would imply that species with different life-histories, intensities of sexual selection and histories of parasite-mediated selection would have evolved different mechanisms to cope with oxidative stress. This explanation must await accumulation of additional data to allow rigorous comparative analyses.

The co-evolution between mechanisms using carotenoids and life-history traits is a pivotal hallmark of adaptation we need to investigate. Therefore, in order to understand how functions of carotenoids have come about, a valuable way to go is through comparison of taxa. Life-history variation may actually be a candidate variable explaining interspecific differences because it is associated with antioxidant parameters. The fact that antioxidant defences are not uniform has been incorporated into free radical theory, and differences in antioxidant defences have been invoked to explain differences in life span among species (Beckman & Ames 1998). For example, total serum/plasma antioxidant capacity (Cohen, Klasing & Ricklefs 2007), as well as single antioxidants (e.g. carotenoids; Olson & Owens 2005) show large interspecific variation, whilst others can be synthesized or obtained from diet in a species-specific way (e.g. ascorbic acid in Del Rio 1997).

In conclusion, we emphasize that empirical evidence for carotenoids being important antioxidants in birds is very weak at best. This implies that three different approaches are needed in future research. First, we should increase our knowledge by analysing additional species under different phases of their life cycle in an effort to understand whether carotenoids may only be antioxidants in certain taxa or under specific phases of the life cycle. Second, we should evaluate if the poor antioxidant activity of xanthophylls also applies to carotenes (i.e. non-oxygenated carotenoids) that are less widespread, but still very common in some species of birds. Finally, we should continue investigating alternative functions of carotenoids, such as their role in enhancing immune function. Several recent studies have indeed provided experimental evidence consistent with an important functional role of carotenoids in immunity (Blount et al. 2003; Faivre et al. 2003; McGraw & Ardia 2003; Saino et al. 2003).


C. Alonso-Alvarez, A. Cohen, P. Hõrak and C. Isaksson kindly provided additional information about effect sizes; one associate editor and one anonymous reviewer provided helpful suggestions and constructive criticisms. D. Costantini was supported by a PhD fellowship issued from the University La Sapienza and by the scientific association Ornis italica.