From Carlos Alonso-Alvarez

June 7, 2008
Manuscript: 08-PONE-RA-040407

Response to Dr Brosnan

In our new version of the manuscript, as required, we tempered the interpretation, recognizing that other mechanisms may be involved. Of course, this seems to be a very complex system and other studies support complementary mechanisms (e.g. testosterone regulation, calcium and amino acid availability in diet; see text).

Also as required, we added a figure representing the pathway.

Finally, Introduction was shortened, removing the paragraph related to the nature of signalling costs (also suggested by referee 2).

Response to both referees (referees’ comments in italics)

Response to Referee 1
Referee 1 did not consent in publishing his comments.


Response to Referee 2

REFEREE2: This is a nice piece of work identifying an effect of levels of a key cellular antioxidant on melanin synthesis. The proposal that antioxidant defence and signal production is not a new one, but in terms of melanin synthesis this is a novel piece of work. The effects seem very clear and are nicely presented.

There have been many attempts to find the magic solution to signal production and many researchers have suggested one important hormone, enzyme or metabolic pathway is THE answer. I suggest that there is no ONE answer to this question. You may well be right that levels of glutathione are important for breast stripe size - but I suggest you reword to suggest they are an important contributing factor.

RESPONSE: We fully agree that there should be not a single mechanism explaining the production of melanin-based signals. We now tempered our assertions. We believe that the proposed mechanism is very relevant, but not the only one (see abstract and fourth paragraph of Discussion).

REFEREE2: I would strongly suggest that you consider a diagram of your proposed pathways as a figure. Clearly BSO inhibits the biosynthesis of reduced glutathione (GSH), meaning that there is more glutathione present in its oxidized form (GSSG). GSH meanwhile has a negative effect on melanin production through inhibition of tyrosinase. The importance of glutathione for skin colour has been well recognized for some time. However, how you propose the inhibition of tyrosinase should converge with the management of free radicals is unclear. If increased levels of GSH mean better antioxidant defence, this also means less melanin production. Are you therefore proposing that melanin is a handicap? Is the trade-off between oxidative stress and melanin production or are you proposing that individuals manipulate management of reduced/oxidized versions of glutathione in order to maximize melanin production? Is melanin a by-product of the attempts to reduce oxidative stress as suggested by some authors for insect immune responses and fungal growth? If you are to propose a new hypothesis it must be quite clear how you propose it functions.

RESPONSE: We added a diagram with the proposed pathway. We indeed defend that the mechanism implies a cost for signalling (page 5, second paragraph). Only those individuals able to pay the cost of mobilizing antioxidants to counteract the impact of low GSH levels on oxidative stress would be able to fully develop the signal. We must however mention that we did not measure the GSH fractions, but the total amount of glutathione (tGSH). This was clearly stated in the first version. BSO is a selective inhibitor of a key enzyme in the glutathione biosynthetic pathway. The decline in glutathione must be mostly attributed to a decline in the production of the non-oxidized intact form (i.e. reduced glutathione: GSH), which would be subsequently oxidized when interacting with free radicals (GSSG). It is also possible that the decline affected both fractions in parallel. We can only speculate on this particular issue, and so we did not address it in the text.

We explicitly stated that melanin production implies a handicap (page 5, second paragraph and Figure 1). The diagram describes how the cost of signaling is produced. Although it could have been posed in terms of a physiological trade-off, it was not because the limiting resource involved in the trade-off is not evident (resources would not be glutathione or melanin, but the whole antioxidant machinery). We emphasize that it is the fact that melanin is obligatorily linked to low glutathione levels that implies a cost. We do not believe that the hypothesis agrees with the idea proposed for insects and fungi.

REFEREE2: A lot rides on the specific effects of BSO and how this results in a change in breast stripe. The authors identify themselves that its crucial to know whether the dose has had some toxic effect or not since the doses have been (relatively) randomly chosen. The authors state that the effects of BSO are specific - and this is true however glutathione is not. It's ubiquitous throughout the vertebrate body and your manipulations have no doubt changed levels in all sorts of tissues. There are published studies relating BSO administration in relation to embryo production, tissue culture proliferation, intestinal enzyme function... etc. So almost certainly it is not just breast stripe which is affected. The authors counter this by measuring body condition (size or mass corrected for size) and concluding that the treatments did not affect this - but what is the statistical power here? This is important if we are to accept an important null result.

RESPONSE: We did not detect significant effects of BSO doses on body condition, body mass gain, and body mass or body size. However, the statistical power of these tests was low. Hence, we cannot conclude that no effects on these parameters were present. Now we have included this information in the text (Discussion, end of first paragraph). Nonetheless, we never excluded that other subtle effects could have been present, but mostly associated to the GSH decrease. Furthermore, it is well known that adverse circumstances during the period of early development promote physiological changes determining the life history trajectory (also mentioned in Discussion of first version). The fact that the study was carried out during a sensitive window (chick growth) suggests that effects could have been long term or delayed in time. We hope in future years recapture at least part of the birds in order to know long-term effects.

REFEREE2: Related to this - how do these breast stripe sizes for juvenile sparrows compare to other published studies? Also how does the breast stripe size at 15 days relate to adult breast stripe size? Please give this information in the ms text.

RESPONSE: See also the answer to comment 2 in Dr McGraw’s review. We believe that the surface of the breast stripe at 15 days should correlate with the surface in adulthood. This assertion is based in the finding that the surface is correlated between successive moults during adulthood in this species (r = 0.91; Senar and Quesada 2006 Behaviour 143: 589-595; see Discussion, penultimate parag.). However, this still needs confirmation (new sentence at the end of the cited paragraph).

REFEREE2: Page 3 para 2 - the introduction relating to the kinds of costs associated with trait production and maintenance is too general and contains too little detail to be useful. Cut this paragraph or review costs comprehensively.

RESPONSE: We removed the paragraph about the different conceptual costs of signalling.

REFEREE2: The authors suggest at the start of the discussion that these results show how important environmental effects are in comparison to genetic control I fund this a puzzling statement since the physiological pathways controlling metabolism of GSH :GSSG will be under tight genetic control. Rewording needed.

RESPONSE: We removed the cited sentence from the first paragraph of Discussion. We reworded the sentences addressing this point (page 9, last paragraph), also including new references. The assertions at the end of the abstract were also tempered.

REFEREE2: Clearly the signal which juvenile sparrows produce in their first moult reflects a level of melanin production over a relatively short period of time ( 2+ weeks?) which they then hold for many months. If the benefits are substantial would it not be better for an individual to sustain some oxidative stress in this time to 'pay the cost' for a better signal?

RESPONSE: Short term benefits may imply long term or delayed costs (see also point 4). The benefits for sacrificing antioxidant power so early (e.g. acquirement of food resources during first months) could be paid in the future (e.g. delayed sexual maturity). For instance in zebra finches the potential benefits of accelerated growth during first weeks of life (attaining optimal body size as adults) are paid in terms of reduced resistance to oxidative stress at the age of sexual maturity (Alonso-Alvarez et al 2007 Funct Ecol 21: 873-879).

REFEREE2: P9 last paragraph 'The hypothesis that...' sentence is unclear suggest rewording. Of course for females there is still a potential cost of T, set at a lower level.

RESPONSE: The paragraph was rewritten (now between page 8 and 9). We also included the suggestion that females can also pay a cost of testosterone.

REFEREE2: Minor points
P 5 4th last line caused by the depletion
P11 l15 'GSH levels in domestic fowl chicks (Gallus species?)
No supplier or product code for BSO

RESPONSE: All the cited minor points were corrected. BSO provider was cited in the first version, but now we added the product code.
September 2008
Manuscript: 08-PONE-RA-04047R1

Response to Dr Brosnan

In our new version of the manuscript, we have followed the suggestions made by both referees. Particularly, we included referee 2’s suggestion about discussing the way to demonstrate that the mechanism we are proposing represents a handicap. All our changes and responses are described below (referees’ comments in italics).

Since the discussion produced during this correspondence has been very fruitful, we would want to ask you if it could be possible to include all the correspondence as free access commentaries in the website.

In any case, thanks for your kind assistance and help.

Sincerely,

Carlos Alonso-Alvarez & Ismael Galvan


Referee 2

REFEREE2: Comments from Referee 2 to our previous responses to Dr McGraw’s comments:

1)Fine
2) Authors should give the data range of stripe sizes given in Galvan et al in this paper and demonstrate here that this range is overlapping, so that the reader doesn't need to go and consult more literature to be convinced of this point.

RESPONSE: Sorry, but we do not have data on breast stripe size in great tit nestlings from previous manuscripts. What we meant is that we have made several observations of great tits nestlings during the course of previous studies in the same population. Unfortunately, however, we had not measured this trait before.

3) Fine
4) Fine
5) Ok

Referee 2’s own comments:

REFEREE2: The diagram clearly helps, but it is the first part of the diagram, which is overly simplistic (ROS - Gluthathione-Melanin) and therefore doesn't really help convince me that this is more likely to work as a handicap than not. I would urge the authors to include some discussion of whether it is a handicap and the data which is necessary to obtain to show this. I think your paper would be much improved by thinking about what the theoretical mechanisms are relating GSH metabolism with melanin production for dominance signalling.
You have convincingly shown that glutathione has an important role to play in determining melanin production in birds, but I am much less convinced by the argument that this works like a handicap. Glutathione levels go down in response to trauma over a few hours-months and the levels recover as the trauma/disease/pathogen is dealt with. There is ample data on the time course in mammals. It seems to me then that having a low level of GSH doesn't mean that you are coping poorly with oxidative stress - it could - and indeed is LIKELY to mean quite the opposite. Low GSH may indicate efficient defence against ROS, but this would be associated with a large signal.
I suggest that the top part of that diagram could be adapted to show that low GSH could be associated with a handicap OR it could be associated with high quality physiological coping mechanisms, which are functioning well.

RESPONSE: We believe that the alternative interpretation the referee is suggesting here is actually the same than that we are providing in the manuscript. We are not suggesting that low GSH levels mean that individuals are coping poorly with oxidative stress, but rather the opposite. Indeed, our interpretation of the possible handicap is that highly expressed melanic signals may indicate individual quality because this needs low GSH levels. As we found an increase in TAS levels in those individuals in which GSH levels were experimentally decreased as compensation, the handicap could be explained by assuming that high quality individuals should be able to afford that compensation more efficiently than low quality individuals (see our response to the point below). Thus, the alternative mechanism suggested by the referee that low GSH levels may be indicators of quality is already mentioned in the manuscript. However, we must take into account that TAS levels increased probably as a compensatory mechanism, an observation that permits us to suggest that melanin-based traits may act as handicap signals. We believe we should not ignore the increase in TAS levels before adding another hypothetical mechanism.

Nonetheless, we also included a sentence suggesting that, “if the capacity to compensate GSH depletion by resource allocation to other branches of the antioxidant machinery does not represent the base of a handicap, GSH-mediated melanin-based coloration would still act as a cue of individual quality, with highly expressed melanic traits indicating that oxidative stress is efficiently dealt with by the cell metabolism (e.g. a better respiratory efficiency reducing ROS production)”.

REFEREE2: You then set yourself up for gathering further data to document what is going on in the bird. If you are to show that it's a handicap
- you need data on how the time course of GSH levels
- How this relates to signal production
- how repeatable in the wild GSH values are during signal production
- manipulate oxidative stress levels and monitor GSH response and melanin production
You can document this in a paragraph explaining what is necessary to determine the evolutionary pathways which have led to this link.

RESPONSE: The referee mentions some aspects we should cite as information that would be needed to demonstrate that the mechanism we are proposing actually acts as a handicap. These aspects are very interesting and necessary, but current signal theory predicts, as a key point in a handicap system, that high quality individuals can cope better with the cost of a handicap than low quality individuals. In other words, low quality individuals pay more for having a trait at a certain degree of expression than high quality ones. In our particular case, we are proposing that the cost of having low GSH levels is the compensation with other antioxidants. Thus, to demonstrate that this is a handicap, we should show, by using any measure of individual quality, that high quality individuals experience higher increases in TAS levels after a decrease in GSH levels than low quality individuals. If there were no differences between these groups of birds in TAS, low quality individuals could still delay the payment of the cost, which should be reflected in any fitness component (i.e. reduced future fecundity or survival). We have also expressed this idea in the manuscript.

REFEREE2: P5 l 9 'with low enough levels of glutathione, such as those required for melanogenesis, must have sufficient alternative antioxidant..' rewording for English expression.
I also suggest you add in here the alternative interpretations e.g.
1) such individuals can afford the damage sustained (also a handicap mechanism)
2) Alternatively low GSH levels indicate that ROS are being efficiently dealt with by the cell metabolism and therefore signal size acts to indicate individual quality.
If you cover all possibilities then regardless of which turns out to be correct you will continue to be cited as the seminal paper in this area!

RESPONSE: The first was changed as suggested. With regard to the other suggestion, please, see previous comments above.

REFEREE2: Within the control group was there any relationship between GSH level and stripe size? Please give result (N = ?) in text at the end of the Results.

RESPONSE: The relationship is not significant. The test is F1, 40.1= 0.89, P = 0.351, n = 53. Please, see justification for avoiding the inclusion of this test in the manuscript body in the previous correspondence (Response to point 3 from Dr McGraw).

REFEREE2: Please give the sample sizes in the different experimental groups in the methods.

RESPONSE: Samples sizes were included in methods (second paragraph, second sentence).

REFEREE2: Figure legend for Fig 1. I suggest you rework the top part of this figure to make it clear that low GSH levels could be good or bad and reword the legend to read 'Hypothesised mechanisms relating to the relationship between oxidative stress management and melanin-based signals..'

RESPONSE: We again emphasize that this is a hypothesized mechanism that does not exclude the presence of other alternatives. These are discussed in the final version, as well as the way to demonstrate the presence of a handicap.

REFEREE2: Sorry to ask you to do more. This is a really nice paper, but I do think it will be much improved by not taking one slant on the data.

RESPONSE: Thanks to the referee contribution and ideas.


Response to Referee 3

REFEREE3: This is an important paper, proposing an interesting physiological
cost to melanin-based signals in plumage coloration. The data is clear
and analyses properly carried out. Their reviewed MS has properly had
into account previous comments by referees. The only remaining
drawback of the paper is that experiments were carried out with
nestlings, for which the functionality of the trait either in nest
defence or status signalling is by definition impossible or has not
been tested. In fact, the function of the carotenoid based coloration
of Great tit nestlings has been suggested to be different from adult
plumaged birds. It is also a pity that the carotenoid-based coloration
was not measured, specially having into account that recent findings
have shown a dependence of this colour on hormones and oxidative
stress; this could had been so easy as plucking a few feathers and
analysing them later in the laboratory. Nevertheless, I have found
results presented so important and interesting, that I think that the
paper merits publication as it stands. By sure that the paper will
prompt new investigations on the topic and will help us to a better
comprehension on the stability and evolution of signalling traits.

RESPONSE: We acknowledge Dr Senar for his kind words. We recognize that we did not analyse adult birds, so we cannot be completely sure about the potential impact of these effects on the adult fitness. However, manipulations during early development could have a deep impact on the physiology of individuals (see e.g. metabolic programming in rodents). If we assume that the most part of the variability in surface of this trait is determined during early development, the information content of these signals could be mostly reflecting the conditions experienced during this very sensitive time window.

Nonetheless, we must also take into account that parental favouritism associated to plumage (UV-reflectance in a patch at the nape) has been demonstrated in this species, i.e. parents feed more those chicks showing a better plumage (Galvan et al. 2008 J Avian Biology 39, 277-282), and great tit parents commonly feed chicks several days after leaving the nest, allowing favouritism on the breast stripe size.

We agree also about the potential interaction between melanin- and carotenoid-based signals, and the importance of multiple signalling for the evaluation of individual quality by the signal receiver. Interestingly, a recent and very suggestive study in great tits describes that heavy metal pollutants induced a decline in the expression of the yellow carotenoid-based plumage of great tits, but an increase in the size of the breast stripe! (Dauwe & Eens, Naturwissenschaften, in press, DOI 10.1007/s00114-008-0400-1). It is well know that heavy metals induce oxidative stress and the depletion of glutathione levels. Perhaps, birds were able to compensate for the glutathione consumption by mobilizing other antioxidant resources, which would have affected to the expression of the carotenoid-based yellow colour. This would also imply an interesting trade-off between different kinds of signals.

REFEREE3: I have a few comments to try to improve the paper:

1). The paper shows that nests with larger broods contained nestlings
with larger breast stripes. This is an interesting result. May this be
related to the fact that parents of higher quality produce larger
clutches of higher "quality"? Is there any data on the size of the
black tie of fathers? See also 2nd point.

RESPONSE: We are very sorry because this particular result was consequence of an error. We reviewed data to calculate mean values for each nest in order to check Dr Senar’s ideas (below). We found that we did not correctly calculate the brood size when summarized the number of rows (i.e. one row, one chick) for each nest (identified with visually similar codes). We have repeated all the analyses in the study. The new covariate did not influence any result. It was always removed as non-significant term. The model in the last paragraph of results was hence corrected with new statistics (very similar). The exclusion of this covariate did not influence Figure 3, because least square means only changed in the third decimal.

Also, we must mention that we avoided capturing adults for reducing disturbance.

REFEREE3: 2). It is suggested that if the resource involved in the cost (i.e.
any antioxidant) is not directly involved in the trait construction,
the cost of carrying a large breast stripe would be only related to
the fact of bearing the trait, and several examples of external costs
are provided. Within this context, I would like to suggest that the
traditional "social cost" could still be operating. For instance, it
has been shown that an increase in the level of aggression increases
stress, and it has also been found that stress induces decreases in
the level of glutathione (Kashif et al. 2004); as a consequence,
social stress could cause the increase of melanization suggested in
this paper. In fact, the authors have found that larger clutches (hence
with higher competence and stress), show increased breast stripes. Is
there any data available on the relationship between proportion of
male nestlings within the nest and breast stripe size (assuming that
males are more aggressive and a higher proportion of males within the
brood would increase competence and stress).

RESPONSE: Please see our response to the previous point.

REFEREE3: 3). It is said that a positive correlation between breast stripe
surface and body condition (residuals of body mass on tarsus length)
was detected in 15d old control birds. I assume that this has been
corrected for sex differences.

RESPONSE: Yes, it was controlled for sex differences by including the sex factor in the model (F1,46 = 4.27, P = 0.045).

REFEREE3: Finally, we must also mention that third paragraph in discussion was rewritten to clarify interpretation. Also, two references were removed (Margalida et al. 2008 and Fitzpatrick 2000) and a new one included (Barja et al. 2007 number 40).