(Please note that this is a co-review prepared by a PhD student and their supervisor.)

Sex reversal, meaning a mismatch between genotypic sex and phenotypic sex, can have important consequences in nature, and the scientific community is only starting to realize this. Therefore, it would be a significant novel finding if several avian species presented 3-6% sex reversal in the wild. Unfortunately, however, the data presented in this manuscript cannot convincingly demonstrate the existence of sex reversal. Much more information is needed on the methods (as detailed below) to ensure the credibility of the results and conclusions. We recommend re-working the manuscript into a full-length article that details the methodological steps and results of marker optimization (as these are novel findings, too) and provides multiple lines of evidence of sex reversal for the concerned samples. We think the revised paper would better fit in a journal that does not force such a hard limit on word count as Biology Letters does.

Major concerns:

First, there is way too little information provided in the manuscript for evaluating the reliability of the genetic sexing methods used. Although primers designed for the CHD1 gene can be applied to many species, they do not always work reliably, and even when they do, they produce species-specific size fragments. When such primers are applied to certain species for the first time, the methods usually need to be optimized for each species, and the results must be verified in detail. The manuscript reports that one out of four different PCR primer pairs “was successful in sexing” each species, but it remains unclear what “successful” means: what settings were tried with how many male and female samples, and what criteria were used for concluding that those settings were appropriate for reliably identifying genotypic sex. Finding and validating the right technical parameters for sexing new species is a novel result that should be documented in detail when publishing those results. Without those details, a multitude of concerns arise that cast doubt on the validity of the conclusions. Ideally, photographs should be provided to show how the gel results look like: what is the bp size of the bands; are the two bands similarly strong/faint? When the PCR and gel settings are not optimized, false results can emerge, e.g. ZW individuals may appear ZZ if there is large difference in the faintness of Z and W bands on the gel or when the Z and W fragments are of similar size and the gel is too thin or not run for long enough. Unexpectedly long fragments or aspecific PCR products can also cause problems. Such findings, or the lack thereof, should be reported to support the claims that a certain method can or cannot be used for diagnosing genetic sex in a certain species. This information is entirely missing from this manuscript, so the readers cannot trust if the “successful” markers were truly reliable and the “unsuccessful” markers cannot be used for sexing those species (or maybe they were just not tried with the right settings). For P2/P8, separation of Z and W products may require lengthy electrophoresis and is better suited to at least 2% gel, as differences between products are typically small across species. For CHD1F/CHD1R, it would be worth considering raising the annealing temperature from the low 48°C to avoid possible artefacts or to consider using a "touchdown" PCR protocol, as the accumulation of aspecific products also makes it difficult to validate results and can be misinterpreted as sex-reversal. Not all markers may work well at one universal annealing temperature, but the PCR parameters were not reported for the other 3 markers, not even for P2/P3 that was used for 2 species.

Second, even when the PCR setting has been optimized and verified, positive confirmation of sex reversal requires multiple lines of validation to confidently exclude other explanations for finding an apparent mismatch between phenotypic and genotypic sex. In the present study, the PCR was re-run for discordant individuals, which is great, but this kind of validation is insufficient because it does not address the following sources of error:
1. Samples can get accidentally swapped or mislabeled, especially when the staff is not trained or motivated enough to handle the samples with sufficient care (as might perhaps happen with people working under great pressure in a hospital). This kind of error is very difficult to track and prove, but in our experience (and talking with colleagues, also in other research labs) it does occasionally happen when working with a large number of samples, despite all efforts to prevent mistakes. More often than not in such cases, one can deduce the error from the circumstances of taking and processing the sample; if any such suspicion arises, we consider the data unreliable and hence missing.
2. Similarly, there is a chance for contamination, either during taking the sample or during the process of DNA extraction and PCR. For example, failing to use flame-sterilized equipment for taking liver samples might easily explain the occurrence of ZW DNA in some males. Contamination can be tested by extracting DNA from an independent sample if multiple samples have been taken from the same individual. Post-sampling contamination can be tested if the initial sample is divided into smaller parts and the parts are processed independently.
3. Measuring DNA concentration and ensuring that all samples are in the ideal concentration range can be helpful. Too low concentration might lead to false negatives, but too high concentration can also yield unreliable results. Even if DNA concentrations are not known, one should check if the bands of discordant individuals are similarly strong/faint on the gel as those of the concordant individuals; conspicuous differences between the two groups would suggest that the apparent discordance might be a genotyping artefact.
4. When all the above aspects suggest that the PCR result is valid, it may still provide false information on genotypic sex due to marker error. For example, one of the primers may fail to bind to the DNA on the primer binding site, producing one band instead of two (making a ZW individual appear ZZ). This can happen by rare mutation, recombination, or other rearrangement
in the sex chromosomes. Such polymorphisms on the sex chromosomes can even lead to a few “normal” ZZ males having a W-specific marker (Ehl et al., 2021). Therefore, the safest way of diagnosing sex reversals is to apply more than one marker or method. For example, the authors could try using primers for other genes that show sexual dimorphism in their intronic regions, such as intron 16 of the NIPBL gene (Suh et al., 2011) or intron 3 of the ATP5A1 gene (Ellegren & Carmichael, 2001), to test if the mismatch between phenotypic sex and genotype is consistent across multiple loci along the sex chromosomes. If that is not possible, it is strongly advised to check for CHD1 polymorphisms with Sanger sequencing, which should not be prohibitively costly for a small sample size of putatively sex-reversed individuals.
5. If genotypic sex is confirmed by all the above tests, it is still possible that phenotypic sex was mis-categorized. For example, Lee et al. (2008) attributed the 12.7% mismatch rate across 25 avian species to erroneously recorded phenotypic sex in the tissue bank data. To avoid doubts in this respect, it is recommended to provide photographs of all the sex-reversed individuals (e.g. in supplementary material), showing both their gonads and their external characteristics that may be informative about their sex (e.g. dichromatic plumage). For example, the “intersex individuals” described in L236-242 might be gynandromorphs (Adolfi et al., 2019), which would be a very interesting finding from the wild, but it needs to be distinguished from sex reversal. Ideally, the gonads of discordant individuals should be further investigated by histological examination, as this would help clarifying their phenotypic sex and inferring their reproductive potential.

An additional issue is the statistical analyses (although this is much less important than the doubts about “sex reversals” being genuine). For each species separately, several measurements of external morphology were compared between discordant and concordant individuals using various statistical tests. However, the sample sizes for discordant individuals are so low (n = 1 - 5 across Tables 2-4) that none of these statistical tests can be used reliably. The only valid statistical approach we can imagine here is either some kind of randomization test (bootstrapping) or, alternatively, comparing the data points of discordant individuals qualitatively to the 95% confidence intervals of the data calculated from the concordant individuals. We strongly recommend presenting the data in box plots to show the median, interquartile range, minimum and maximum value for each variable in the concordant individuals, and indicating the position of discordant individuals along these distributions as points (symbols) added onto the plots. If this information is given, we do not think any other statistical inference is needed – because any statistical test would be powerless to answer the question “do sex-discordant individuals differ in size measurements from concordant conspecifics?” at such tiny sample sizes.

Finally, we must point out that the data tables provided at the Dryad link fall short of the guidelines for open access transparency. Although 5 species were studied, data tables are provided for 4 only. For each species, there is one excel file with a lot of sheets, with no explanation as to what each sheet represents; some of the sheets are empty. There should be a single sheet for each species, providing all the data for all individuals, using headers and values that are easy to interpret (or else explained in another sheet or file). For example, it is unclear what “Intersex in place” means, while it is impossible to find out from the tables which individuals were intersex.

Further, relatively minor comments on the manuscript in the order of appearance in the text:

L37-38: This sentence should be deleted from the abstract, because the study does not address any of the issues mentioned here. We agree that these issues are relevant as they provide motivation for studying sex reversal, but the results in this manuscript do not allow for any inference to be made about these topics.

L52: Distinction between genotypic and phenotypic sex is important for so many reasons beyond the potential disconnect between “brain sex” and “gonad sex”. According to theoretical models, the presence of sex-reversed individuals in wild populations may catalyze various demographic and microevolutionary processes, including skewed sex ratios, reduced effective population sizes, evolutionary transitions between sex-determination systems, evolutionary changes of mate-choice preferences, and even population extinction (e.g. Grossen et al., 2011; Bókony et al., 2017; Schwanz et al., 2020; Nemesházi et al., 2021).

L63: Temperature-dependent sex determination is only one (albeit the best known and apparently most widespread) form of environmental sex determination, so please change “i.e.” to “e.g.” here.

L93-95: This statement should be backed up with evidence. What publications demonstrate that sex reversal commonly has such deleterious effects? We very much doubt that any evidence of that sort exists in birds. The fitness consequences of sex reversal are extremely poorly known even in taxa where sex reversal has been studied more often, and in more detail, than in birds (e.g. reptiles, amphibians).

L110: Please clarify in the text that this sentence does not refer to birds but to reptiles and humans.

L113 and throughout the entire text: Please use clear terminology. Phrases like “sex-reversed females” and „sex-reversed males” are unclear because “female” and “male” can refer to genotypic sex or to phenotypic sex. For example, a “sex-reversed female” could be a genotypic female with male phenotype, or a genotypic male with a female phenotype. When talking about birds, we recommend using phrases like “sex-reversed ZW males” and “sex-concordant ZZ males”. Phrases like “ZW males” clearly communicate whether there was a mismatch, and adding “sex-reversed” to this phrase should facilitate easy interpretation for a broad readership (i.e. even for those who find it difficult to keep in mind whether ZW birds are genotypically male or female because they do not work in this research field).

L113-114: Please specify in the text that these statements refer to the domestic chicken.

L116: This statement suggests that the male-biased sex ratios of threatened bird species are due to sex reversals. This is an interesting and potentially important possibility, but it needs more context so that readers can correctly infer its plausibility. Male-biased adult sex ratios are typical for birds, and there can be multiple reasons for this (e.g. (Pipoly et al., 2015; Xirocostas et al., 2020; Schacht et al., 2022). Based on existing evidence, it does not seem likely that these biases would be driven by hatchling sex ratios, at least not in comparisons across species and populations (e.g. Székely et al., 2014; Eberhart-Phillips et al., 2018). If sex reversals were affecting sex ratios, this should be more apparent closer to conception, e.g. at hatching rather than in adulthood, yet hatchling sex ratios are not typically male-biased (Donald, 2007; Székely et al., 2014). We recommend deleting this sentence from the Introduction and explaining this issue in more detail in the Discussion (L305-310).

L117-118: This sentence is a bit misleading, as it suggests that the offspring of sex-reversed individuals are more sensitive to environmental disturbances in various kinds of traits. However, what the cited studies show is that these offspring are more likely to undergo sex reversal in response to environmental stimuli, compared to the offspring of sex-concordant parents. There is no data to show whether this sensitivity extends to other traits like survival or disease susceptibility. Please re-phrase the sentence accordingly.

L123-125: It would be worth mentioning among the objectives that attempts were made to test if sex-reversed individuals differ from sex-concordant conspecifics in terms of gonad size and body morphology.

L130: How does the set of chosen species represent the species in decline? Please provide information on the conservation status of each species studied. You could present this in a table that would also provide information on the level of sexual dimorphism, e.g. is it known to be present, in which bodily measurements; do males and females differ in plumage coloration – this latter is relevant for recognizing chimeras or gynandromorphs.

L140: The fact that the samples came from birds that died in wildlife hospitals should be mentioned in the Abstract and in the Discussion. There is a chance that these samples do not adequately represent the wild populations: for example, if sex-reversed individuals have poor health, they may be over-represented in the hospital samples.

L192-208: The description of the statistical analyses is somewhat difficult to understand, as it is unclear which sex groups were compared. Please use clear terms and specify which variables were compared between which groups. The phrases “status (discordant or non-discordant)” and “gonadal makeup (i.e., testes, ovary and largest follicle)” are not clear enough. Instead, please use phrases like “testis measurements were compared between sex-concordant ZZ males and sex-reversed ZW males, whereas ovary and follicle measurements were compared between ZW females and ZZ females”. For external morphology, it seems from Table 4 that “discordant females” (probably ZW males, although this is not explicitly stated in the text) were compared to concordant ZZ males but not to concordant ZW females. The logic of this is unclear. We recommend presenting the data in such a way that both ZW males and ZZ females can be compared to both ZZ males and concordant ZW females, at least qualitatively from graphs. We suggest showing a boxplot each for ZZ males and concordant ZW females (for each morphological variable) and adding the dots of ZW males and ZZ females as different symbols.

L193: If you have categorical explanatory variables only and no numeric predictors, then the analysis is a MANOVA (not MANCOVA). However, this is probably not relevant anyway, because neither MANOVA nor MANCOVA should be used with these data (due to the small sample size of discordant birds and the issues in the next comment).

L205-206: The data themselves cannot be parametric or non-parametric; it’s the statistical tests. If some of the data do not meet the requirements of parametric tests, then they should not be used in MANOVA or MANCOVA. Please note that normality is not the only requirement for parametric tests like MANOVA and t-tests: homogeneity of variances is even more important. This requirement cannot be met when one group (concordant) has 10-100 times larger sample size than the other group (discordant). This comment and the previous one would constitute a major problem if we were not recommending the omission of the statistical analyses altogether.

Table 1: Please provide more information in this table. The numbers of discordant individuals should be broken into ZZ females and ZW males, and it would be nice to see the numbers of concordant males and concordant females, too. Additionally, we highly recommend providing further detailed information on the 25 discordant individuals in another table (either in the main text or as a supplement), showing for each individual whether it was juvenile or adult, sex-reversed (one gonad type) or intersex (both gonad types, as described in L236-242), and which locality it originated from. These data would be helpful for interpreting the results and putting them into context (e.g. see discussion about endocrine disruptors). Although this information could be extracted from the many supplementary tables with a lot of work, it would be much more accessible to the readers if the authors did this work instead.

L223: This trend is rather difficult to evaluate from Table 2. The difference between the means seems large, but the SD values are also large. The box + dot plots we recommend would be much more informative.

L245: This contradicts what is written in the Methods (“Significant differences were found in all measurements comparing immature and adult birds”).

L246-247: The statistics underlying this statement are missing from the manuscript, as Table 4 only compares discordant females to non-discordant males (but not to non-discordant females).

L269: Please elaborate what you mean by “immature-like gonads”. Photographs would be helpful.

L284: It would be worth discussing whether the discordant birds came from the same area(s), as this would further suggest environmental influence. Moreover, if the “intersex” individuals came from the same area(s), it might suggest environmentally disrupted meiosis or mitosis resulting in gynandromorphy or chimerism (e.g. see Supplementary Table 1 in Szász & Rosivall, 2015) – which might be detected only at the level of gonads if the external appearance (e.g. plumage) of the species is not sexually dimorphic.

Typos and minor language errors:
L84-86: “in-ovo” is written twice, the second is not needed
L89: “and can cause a permanent sex reversal in the developing embryo”
L193: “multivariate analysis of covariance”
L195: “length” missing from the end of this line
L196: “predictor variables” (even better to say “explanatory variables”)
Table 2: t-test in the caption but Mann-Whitney test in the header; please clarify.
L284: “… zone, where endocrine …”
L295: please clarify “masses on ovary-like tissue”
L317: please clarify “reproductively redundant”

References
Adolfi, M.C., Nakajima, R.T., N, R.H. & Schartl, M. 2019. Intersex, hermaphroditism, and gonadal plasticity in vertebrates: Evolution of the Müllerian duct and Amh/Amhr2 signaling. Annu. Rev. Anim. Biosci. 7: 149–172.
Bókony, V., Kövér, S., Nemesházi, E., Liker, A. & Székely, T. 2017. Climate-driven shifts in adult sex ratios via sex reversals: the type of sex determination matters. Philos. Trans. R. Soc. B Biol. Sci. 372: 20160325.
Donald, P.F. 2007. Adult sex ratios in wild bird populations. Ibis 149: 671–692.
Eberhart-Phillips, L.J., Küpper, C., Carmona-Isunza, M.C., Vincze, O., Zefania, S., Cruz-López, M., et al. 2018. Demographic causes of adult sex ratio variation and their consequences for parental cooperation. Nat. Commun. 9: 1651.
Ehl, J., Altmanová, M. & Kratochvíl, L. 2021. With or without W? Molecular and cytogenetic markers are not sufficient for identification of environmentally-induced sex reversal in the bearded dragon. Sex. Dev., doi: 10.1159/000514195.
Ellegren, H. & Carmichael, A. 2001. Multiple and independent cessation of recombination between avian sex chromosomes. Genetics 158: 325–331.
Grossen, C., Neuenschwander, S. & Perrin, N. 2011. Temperature-dependent turnovers in sex-determination mechanisms: A quantitative model. Evolution 65: 64–78.
Lee, M.Y., Hong, Y.J., Park, S.K., Kim, Y.J., Choi, T.Y., Lee, H., et al. 2008. Application of two complementary molecular sexing methods for east Asian bird species. Genes and Genomics 30: 365–372.
Nemesházi, E., Kövér, S. & Bókony, V. 2021. Evolutionary and demographic consequences of temperature-induced masculinization under climate warming: the effects of mate choice. BMC Ecol. Evol. 21: 16.
Pipoly, I., Bókony, V., Kirkpatrick, M., Donald, P.F., Székely, T. & Liker, A. 2015. The genetic sex-determination system predicts adult sex ratios in tetrapods. Nature 527: 91–94.
Schacht, R., Beissinger, S.R., Wedekind, C., Jennions, M.D., Geffroy, B., Liker, A., et al. 2022. Adult sex ratios: causes of variation and implications for animal and human societies. Commun. Biol. 5: 1–16.
Schwanz, L.E., Georges, A., Holleley, C.E. & Sarre, S.D. 2020. Climate change, sex reversal and lability of sex-determining systems. J. Evol. Biol. 33: 270–281.
Suh, A., Kriegs, J.O., Brosius, J. & Schmitz, J. 2011. Retroposon insertions and the chronology of avian sex chromosome evolution. Mol. Biol. Evol. 28: 2993–2997.
Szász, E. & Rosivall, B. 2015. The chimeric embryo hypothesis as a mechanism of avian sex ratio manipulation? Comment on Tagirov and Rutkowska. Behav. Ecol. 26: e1–e3.
Székely, T., Liker, A., Freckleton, R.P., Fichtel, C. & Kappeler, P.M. 2014. Sex-biased survival predicts adult sex ratio variation in wild birds. Proc. R. Soc. London B Biol. Sci. 281: 20140342.
Xirocostas, Z.A., Everingham, S.E. & Moles, A.T. 2020. The sex with the reduced sex chromosome dies earlier: A comparison across the tree of life. Biol. Lett. 16.

Please see the attached pdf.