General Comments

In the study, the authors use 5 years worth of grooming and protozoan infection data in free-ranging mandrills to test the hypothesis that females that make risk-averse decisions about grooming infected conspecifics themselves exhibit higher richness of infection with intestinal protozoa. They also find that these grooming related hygienic phenotypes are repeatable within females across the years of study, perhaps representing an aspect of their personality. The authors also suggest that hygienic phenotypes are inherited through social rather than genetic mechanisms, since maternal kin, but not paternal kin or unrelated individuals, have similar behavioral patterns.

I find the study interesting, and using 5 years worth of data in a well-studied population is commendable. My main concern, however, is whether or not the hygienic index created for each female is justifiable, since some females appear in grooming events apparently much more often than others, and indices are created for females for which grooming events are as scarce as 10 events over an entire year. I am skeptical that such a measure adequately reflects the reality of behavioral phenotypes in the group. I’m not sure the authors can justify data from 10 grooming events over a year. Most behavioral studies use far more, and what we know about variation in behavior across individuals, seasons, contexts, etc, suggest that such a small number is simply insufficient.

I also would like to see more justification for the use of protozoan richness as the index for risk profile associated with each female. This may be a useful index of risk or the outcome of risk aversion versus proneness, but it may not be. Since no justification was provided for it, readers cannot know how they should interpret the results, or whether they can be trusted. In addition to this being important for interpretation, I would also like to see more information in general about the parasites themselves, their likelihood of transmitting through grooming, something about their patterns of infection and what might drive them (other than PAR grooming interactions), etc. Partly because of the study design, aggregating behavioral data and infection data at the scale of a full year, it’s not easy to see the link between behavior and infection that underlies the premise of this study. So I think more can be done in that regard.

Lastly, I tried to make some suggestions for grammatical and structural improvement of the manuscript, but there are a bunch of English issues that could be improved. I hope my suggestions are useful to the authors. I did not proofread the whole manuscript for English, though.

Overall, I think this is a study that could provide important information but my concerns above and below leave me skeptical about the value of the results as currently shown. I think a more in depth look at behavior profiles across females at higher resolution, for example within ecologically or physiologically relevant seasons, or even using specific grooming events themselves as the unit of the analysis (e.g. response variable in the models) might allow for more robust analysis. I hope the authors find these comments useful in reconsidering the approach used.

Specific Comments

ABSTRACT
L15 - “such strategies”

L15 - “deprive individuals of…”

L24-25 - “sociality level” might be too vague for readers

The results in the abstract are presented in the present tense, but I would argue this should be wrapped in the framework of the study period, so past tense should be used; e.g. “…consistently avoided…”, “…while others did not…”, “maternal kin exhibited…”, and so on

L25-26 - nothing here in the abstract can distinguish between genetic inheritance and social inheritance, so this seems odd here

L28 - why “unforeseen consequences”?

INTRODUCTION
L38 - is EBOLA really spreading at large scales? That would be terrifying, but I don’t think it belongs in the same category as influenza or COVID-19. I would also argue that EBOLA is not a worldwide threat and is not currently causing wildlife declines or largely affecting human populations (though it has done so on occasion and remains a threat). I suggest more caution in writing, and use tense correctly when identifying current versus historical or recurring and localized threats. This comment probably applies throughout the manuscript.

L43 - I’m not sure “coined” is approporate here. The authors borrowed the term and applied it to animal social systems, but they did not come up with it in the first place.

L46 - the cited paper (Butler et al 2015) do not even mention the term “extinction” in their paper, so I would not go so far as to make this claim and attribute it to them. Their point was about preventing epizootics through social immunity, and that’s what their simulations show. Also, this sentence and especially the phrase “…has prevented…” makes it sound like models were applied to real data, i.e. a real extinction was prevented by social distancing. In fact, the entire results of those simulations were theoretical, so you should take a bit more care in sharing their results

L49 - remove “system” or change to something like “use of a waste management system”

L50 - I think that you can start a new paragraph beginning with “Recently, …”, because your opening paragraph is far too long at one and a half pages, and covers too many ideas

L65-66 - While the authors may have a point, I would advise caution with statements like “…no study on animal models has ever…”

L69 - “to comply with”

L70 - I’m not sure what has been done in animal populations represents a “simple descriptive approach of anti-parasitic behaviors”. Plenty of that literature has used careful experimentation, including in the field, and has used statistical models to parse out sources of variation in expressed antiparasitic behavior. These same studies have also already identified some sources of variation between individuals as well, as the authors note with sex biases, for example. This sentence should be deleted or rephrased accordingly

L74-75 - is it accurate to call this a “natural population”? They are ex-captive and fenced in, and called “free-ranging” in the methods, so I would argue this is not accurate.

L78 - either “allows social partners access to social resources” or “allows social partners to access social resources”

L85-87 - B. coli is not typically thought to be pathogenic in wild primates, and is pretty ubiquitous. E. hystolytica is generally known to be pathogenic, but is not typically found in wild primates (my understanding is that most infections that resemble E. hystolytica are in fact the commensal E. dispar). Although I don’t disagree with the premise, I worry that the use of examples or phrasing here do not strongly support the claims made or may misrepresent the host-parasite relationship somewhat

L113 - try “We did not exclude the possibility, however, that adult females might be less hygienic because of their matured physiological immune system”

L116 - I’m not sure that “studying its frequency across different kin categories” is sufficient to understand how genetic and social inheritance are distinguished, expecially when following this statement with sentences about maternal kin, which on the surface could reflect either mechanism.

METHODS
L133-134 - “seasonal year” sounds awkward, and since it’s a 12-month period anyway, and your ordering is arbitrary, just call it a “year” or maybe better a “study year” and explain how it was considered thereafter, as is done

L147 - “groom their group mates dramatically less…”

L151 - “sampling”

L174 - don’t need to restate that a seasonal year is from Feb-Jan

L219-220 - this sounds like a result of the study. If so, omit it from here. If not, clarify why this is relevant to the study methods.

Section (e) - I think it needs to be justified why protozoan richness was chosen as the measure of infection. This may be an appropriate measure of the outcome of risky behavior, but that should be stated clearly in the text. I also wonder whether the authors might also include an index of the groomee’s parasite richness in the analysis. Since the authors have created an index of risk for each female based on their proportion of perianal (versus other areas) grooming, in theiry they could create an index that describes the riskiness of that behavior relative to the infection status of their grooming partners, e.g. creating an average protozoan richness for groomee’s throughout each year. I understand this needs to be balanced by how often females groomed their various partners, but it’s not impossible to imagine such an analysis. Also, as I mention below, if the unit of the analysis is changed from risk index per year to the actual behavior of interest that is being observed PAR grooming events/bouts, then this idea of groomee infection status can more easily be incorporated into the analysis.

Section (e) and (f) - I’m also concerned that the resolution of the data - aggregated to the scale of an entire year - might be too crude to test the hypotheses set by the authors. For example, this completely ignores seasonal variation in both infection and grooming tendencies, as well as the possibility that there is relevant seasonal variation in the observation and sampling schedule for each individual. Especially concerning male-female relationships and reproductive seasonality (including interannual variability within females), this might introduce key differences in the expected grooming relationships females have with other both within and between years. Yet, that is not accounted for in the model structure here. I would also argue that risk profiles should vary throughout the year as parasites tend to have seasonal patterns driven by a combination of host and environmental variables. It’s also somewhat startling that the indices can be based on as few as 10 grooming events for a female in a given year. Across an entire year, with no information about how the grooming bouts/events are distributed across the year, 10 seems like an insufficient number. And, there is no information provided about the distribution of grooming events per female throughout the year. I feel this is a critical missing component, but also that 10 may simply be too few behavioral events on which to base robust estimates of behavioral phenotypes across an entire year of study.

Section (g) - here, the use of a minimum of 5 grooming events over a year seems even more insufficient than the 10 stated above. But another comment here is that, if the authors claim that 5 is just as good as 10 (via the high correlation), then by their own logic couldn’t they use 5 as the minimum for the other parts of the study? Of course, as stated above, my suggestion is that even 10 is too limited to make strong inference from such data.

Statistics: I think the description of the statistical models constructed is not sufficient. PRG seems to be proportional data, but the authors state they constructed linear mixed models, which aren’t appropriate for such data. There is also no mention of what was done for model validation. More information is needed so that we can (1) evaluate the results and even (2) replicate the study, if desired. Much more attention to the details of the statistics is thus warranted here.

RESULTS
L269-270 - First, why not present exact p values? Second, here it says that the p-value for mother-daughter pairs is “p<0.01”, but in Figure 1.a, it is shown as “p<0.001”. Please check.

DISCUSSION
L284-286 - It’s unclear what is being suggested here. This seems most relevant for intersexual interactions (male grooms female), but males would be able to discern this through other means such as inspections, I would assume. Grooming is not often interpreted as a means to obtain this kind of information, so if the others want to highlight this possibility, perhaps some data about the relationship between grooming and inspection-type behavior would be helpful. Additionally, is this relevant for female-female grooming dyads? Without more information, it’s hard to know how to take this suggestion for the function of perianal grooming. Why can’t it have the same function as grooming anywhere else on the body?

L291 - should “contamination” actually be “infection” here? Contamination is something that happens outside the body in the environment, or potentially on the body causing a risk to others, but here it seems you are talking about being infected

The first paragraph of the discussion is part summary of results, part speculation about behavioral versus physiological immune system. I recommend splitting these, so that the results are summarized first and interpretation/speculation is reserved for subsequent paragraphs.

L298 - change “show” to “showed”, again reflecting that the results should be presented within the context of this study, which happened in the past. This logic should be applied throughout the discussion.

L318 - the authors could split the paragraph at “Alternatively”, to avoid an overly long block of text.

L329 - “Great-Ape” should neither be hyphenated or capitalized. “…great ape…” is correct

General Comments

The revised version of this manuscript has dealt with many of the original concerns and in my opinion is thus a stronger work testing variation in hygienic behaviour across females. I do believe that the study is of general interest and tackles an important and interesting question, but I still have reservations about a couple of points that I highlight below.

The first is that I still feel strongly that 10 grooming bouts across a year is far too few to represent general behavioural tendencies of a female, and I don’t think that the authors make this caveat clear enough in the current version, just as in the first. I understand that the authors provide some additional analyses in the manuscript (e.g. repeatability across years) and in the supplementary materials to justify including such females, and it seems the statistical models are robust to leave-one-out validation tests, but what happens if different inclusion thresholds are used instead of 10, e.g. 15, 20, 30, etc? Are the results robust to such varying thresholds, accepting that sample sizes get progressively smaller the higher the threshold? Is there perhaps any correlation between repeatability in PRG across years within females and the number of grooming bouts included? Such additional analyses might be possible to include, but at the absolute least the authors need to show some skepticism about their own results resting on such small data sets, and have a clear statement to this effect in the discussion.

The second is that I think the authors can improve their rhetoric and analysis around social and genetic inheritance. First, it’s rare to find a behavior that is one or the other, so having that as a null hypothesis is questionable. Second, there is no quantitative genetics study provided in these pages, so there should be much more caution in the phrasing about how much genetic influence may or may not be observable here. Though, this could be done using “animal model” style Bayesian analyses incorporating the pedigree data directly (Wilson et al 2009, Journal of Animal Ecology, DOI: https://doi.org/10.1111/j.1365-2656.2009.01639.x). I think such an approach would strongly benefit the story being tested here, as the overall genetic contribution could be estimated, particularly since such strong maternal and paternal pedigree data exist! Third, the category of full sisters has only two dyads and is not included here, but should be relevant to the story. In particular, looking at differences in hygienic similarity between full versus half-sisters should be useful in parsing social-environment from genetic influences, while comparing and contrasting each with mother-daughter pairs. This limitation should be noted somewhere, at least in the discussion as another caveat of the study. Fourth, the idea that you can assess genetic relatedness and hygiene index in some 1:1 ratio is dubious to me. We rarely see such mapping of genotype to phenotype, so it seems unnecessary to expect “twice as much” similarity in hygienic tendency between mothers and daughters as that between non-kin dyads. I think it’s enough to compare and contrast the different dyadic categories with respect to their relatedness without trying to map them quantitatively onto one another without foundation to do so.

Specific Comments

Title
I’d suggest using the expression “hygienic behavior” or even “hygienic anti-parasite behaviour” (as on L40, L44, L66, etc) instead of “hygiene”, since this study examines a specific behavior rather than the larger concept or suite of behaviours that amount to hygiene. You might also consider “vertical transmission” in place of “mother-to-daughter” as an established term in this field; but that’s up to the authors.

Abstract
L27-28 - consider changing to “both mother-daughter and maternal half-sister dyads exhibited similar hygienic levels, whereas paternal half-sisters and non-kin dyads did not, suggesting social transmission of this behaviour.”

Introduction
There remain a large number of grammatical mistakes in the text, so I have taken some time to help with that here. But I encourage the authors to use the notes below and apply them to other places in the MS where similar issues may arise as I cannot go through all of them.
L48 remove “the” from “the environmental” (just a suggestion for readability)
L59 change “largely affecting” to “significantly affecting”
L60 “may therefore constitute…” or “may constitute, therefore, …”
L72 “…, a cercopithecine primate native to Central Africa, to study …”
L744-75 “…that allows, for example, social partners …”
L76 has a run-on sentence. I suggest ending the previous sentence after reference 27. Then, start new sentence with “However, grooming also …”
L82 “decrease host fitness”
L86-87 questionable that this is “social avoidance” since, as the other reviewer pointed out and the authors have been good to revise, they are still grooming
L88 “close maternal kin, possibly because …”
L90 “hereafter”
L92 not getting contaminated or not getting infected? Or “not encountering parasites”?
L96 “tendencies to groom (or not) the PAR of group mates should result …”
L102-103 “i.e., to groom PAR less frequently” (it’s implicit this means PAR of group mates by this point)
L103 “Second, and to …” (missing comma)
L110 I wonder if “sociality level” is too broad a term to use for this study, as it needs to be caveated later to overall grooming frequency. Why not just use grooming frequency from the start to avoid confusion? Also, grooming frequency may not be a good indicator of “sociality” more generally. This change should be made in the abstract as well.
L114 since this is a very long paragraph, I suggest beginning a new one here at “Finally, …”. Then, the part about social and genetic influences are packaged in one unit and easier to read.
L114-116 instead of creating a false dichotomy (genetic VERSUS social), I suggest more neutral phrasing such as “we evaluated the extent to which this trait was influenced by social and genetic relationships between females…”
L118 what about full sisters? It seems that the genetic data exist to also examine full sisters, and this should be highly relevant to the argument here since they’d have the same genetic relatedness as mother-daughter dyads, but perhaps less social influence on one another, particularly if further apart in age. I see that there aren’t enough full sisters in the data set, but it should at least be mentioned as a missing component of the study.

Methods
L129 “the study group” should suffice (note that “social group” is itself redundant)
L139-140 “was routinely performed”
L144 “from June to September, and an …”

Results
L295 “superior to” is not a good word here; try “higher than”
L323-324 the results are not identical. For example, the maternal half-sister dyads in the ESM data set are not significantly similar, though the ranking in similarity is the same as in the full data set. But I think it’s critical to be accurate with such statements not to overstretch claims, particularly since the authors are trying to use this analysis specifically to justify including females with so few observed grooming events.

Discussion
L350-351 change “groomed consistently less the peri-anal region…” to “consistently showed less grooming of the peri-anal region”
L351 add a comma after “than others, …”
L352-353 using the term “neither” implies two points coming up, you have three. Change to “none of female age, dominance rank or sociality level”; again though I suggest changing “sociality level” to something more specific like “overall grooming frequency”
L367-368 I find it problematic that the hypothesis requires “twice as much hygiene”. This is ill-defined and cannot be evaluated at the same level as genetic similarity/dissimilarity. There is no reason to expect that the index created for hygiene here in any way matches the scale used for genetic relatedness (I had the same feeling about the set up in the intro). I suggest refocusing on comparing and contrasting the different dyads and just reporting whether or not they fit patterns expected by genetic or social influence, without trying to quantify such differences on parallel scales.
L377-378 this is hard to read. Change to “…, we found that maternal half-sisters were more likely to share top grooming partners (7%) than were mother-daughter pairs (3.8%).”
L416 “have seldom been considered”