The Correlation between Play Behavior, Serum Cortisol and Neutrophil-to-Lymphocyte Ratio in Welfare Assessment of Dairy Calves within the First Month of Life

Round 1

Reviewer 1 Report

This article entitled “The correlation between play behavior, serum cortisol and neutrophil to lymphocyte ratio in welfare assessment of dairy calves within the first month of life”, by Ogi et al., is an original research article falling into the scope of the journal “Dairy”. Its topic is about the correlation between physiological and behavioral parameters for the welfare assessment of dairy calves during their month of life, including stressful events such as disbudding. The authors particularly focused on the play behavior and the serum cortisol and N/L ratio as parameters, the latter being especially promising but very few studied in cattle. Animal-based measures in the welfare assessment of farm animals are still needed and deserve scientific investigations. The work presented here is in line with this requirement and the data presented here are of sufficient novelty and international interest.

General comments and recommendations:

The context and interest of this research are well explained, but there are several inaccuracies (notably inadequate references) in the introduction that I would like the authors to correct. The objectives are clearly stated. Some elements of Material and Methods deserve further clarifications/explanations. The precision of some results could be enhanced. The discussion is appropriate, relating to the results and interpreting them from the perspective of previously published papers. The conclusions are supported by the data and are reasonable.

Please find hereunder my detailed comments:

• L17: It seems there is a mistake as the Pearson coefficient indicated here is negative while you wrote before there is a positive correlation
• L30-35: I would like to add a caution about this part, since in more recent literature, Broom has revised and explicated his theory about animal welfare definition, encompassing the emotional aspects as well. Broom denies limiting his conception, his definition of welfare to a functional approach only and argues that his definition also encompasses feelings as exposed in Broom (1991). He provided arguments for the evolution of feelings as part of animal functioning that are explained in Broom (1993;1998) and Broom and Fraser (2007). There is now a more consensual view of what is animal welfare between the different scientists, involving previously defined approaches, as stated in Wemelsfeder and Mullan, 2014. Then, I would like to ask the authors to consider these updates on animal welfare definitions
• “seem to indicate a state of good welfare” (L42): Or less negative welfare? I think this is more subtle... please see Ahloy-Dallaire et al, 2018 on this topic.
• L57, ref [22,23]: I am sorry to say this, but these references do not seem to be the most appropriate since they are about the link between the salivary oxytocin level/the oxytocin receptor polymorphism in mothers and their maternal behaviour, but not about the development of the newborns per se. Instead, there are plenty of most adequate references on this topic, e.g. Veenema (2012) or Mellor (2015) for reviews, or Foyer et al (2016), and of course Guardini et al (2016 & 2017). Please change for more appropriate references.
• L68, ref [29,30]: Again, I think there are much more appropriate references to cite to illustrate the use of cortisol as a biomarker of stress in mammals, including very complete reviews.
• “but the scientific literature about the possible correlations between these variables is scarce” (L81-82): It is not accurate... There are many publications combining physiological (of any kind, such as heart rate variability, eye white, brain activity or neuromodulator measurements) and behavioral parameters to assess animal emotions/animal welfare. For examples in livestock species, please see Briefer et al, 2015 in goats; Désiré et al, 2004 in lambs; Rietmann et al, 2004, Bachmann et al, 2003, and Mendonça et al, 2019 in horses; Geverink et al, 2002, Rault et al, 2013, Carreras et al, 2016 and Marcet-Rius et al, 2018 (moreover in the context of play) in pigs; and finally in cows, Ternman et al, 2012, Proctor et al, 2015, and Doerfler et al, 2017.
• L95: do you mean between 30 min and 2h?
• L111: Was this kit previously validated to assay cortisol in cow serum or was it previously used as such in other publications?
• L126-129: What about the intra-rater observer reliability? Have you calculated it for the single chosen observer? Because you might have selected the observer based on this intra-rate reliability instead of randomly, as it seems here. Anyway, it is always preferable to ensure the accuracy of the behavioral assessments based on two behavioral observations from two independent observers, and a good agreement between the two observers calculated only on 12% of the videos is not a sufficient reason to discard one behavioural observation.
• L140-158: Why didn't you run correlation analyses with the other types of play behaviours? Even if the results can be considered as "negative" because non-significant, it is still interesting and informative for the scientific community to show them. Could you please add these correlation analyses? or at least mention their outcomes?
• L160-162: same remark as in the abstract about the Pearson’s coefficient. Also, because of the coefficient values (between 0.2 and 0.4), you should specify that both are weak correlations.
• L164: all locomotor play as listed in Table 1 (the 6 types) or only the running? Please clarify.
• Figure 1 legend: the correlation being positive, as shown in the figure and written in the text, the coefficient cannot be negative! Be careful to this mistake repeated several times.
• L200-201: this is a result. You should have mentioned it before, and possibly provided the detailed statistical results. Also, please clarify your sentence to state exactly what parameters you attempted to correlate.
• L205-208: Could you please clarify your hypothesis? Sorry, I don't really understand what you mean since running time is increasing over the time of the first month... don't you consider running as a play behavior?
• “measured non-invasively in saliva” (L256): And in hair too. Hair cortisol seems to be a more integrative biomarker of cortisol secretion, and as such a better predictor of chronic/ long-term stress, like N/L. It could be of interest to compare N/L with this kind of cortisol measurement by the way. See the recent review of Heimbürger et al, 2019 for instance. Moreover, this type of measurement has already been performed in cows: Comin et al (2011, 2013) and Burnett et al (2014, 2015).
• L257: please remove “it”.
• References: ref. 23 is incomplete (and inappropriate by the way, as noted here above) and the format of ref [40] does not seem correct.

Author Response

This article entitled “The correlation between play behavior, serum cortisol and neutrophil to lymphocyte ratio in welfare assessment of dairy calves within the first month of life”, by Ogi et al., is an original research article falling into the scope of the journal “Dairy”. Its topic is about the correlation between physiological and behavioral parameters for the welfare assessment of dairy calves during their month of life, including stressful events such as disbudding. The authors particularly focused on the play behavior and the serum cortisol and N/L ratio as parameters, the latter being especially promising but very few studied in cattle. Animal-based measures in the welfare assessment of farm animals are still needed and deserve scientific investigations. The work presented here is in line with this requirement and the data presented here are of sufficient novelty and international interest.

General comments and recommendations:

The context and interest of this research are well explained, but there are several inaccuracies (notably inadequate references) in the introduction that I would like the authors to correct. The objectives are clearly stated. Some elements of Material and Methods deserve further clarifications/explanations. The precision of some results could be enhanced. The discussion is appropriate, relating to the results and interpreting them from the perspective of previously published papers. The conclusions are supported by the data and are reasonable.

We gratefully acknowledge the interest in our work and the valuable suggestions.

Please find hereunder my detailed comments:

• L17: It seems there is a mistake as the Pearson coefficient indicated here is negative while you wrote before there is a positive correlation

We thank the reviewer for spotting this. It was the same mistake (we added a sign – by mistake) in other parts as we copied and pasted the same result. The results now changed as we considered repeated measures correlations as advised by reviewer 2. LL 17-19

• L30-35: I would like to add a caution about this part, since in more recent literature, Broom has revised and explicated his theory about animal welfare definition, encompassing the emotional aspects as well. Broom denies limiting his conception, his definition of welfare to a functional approach only and argues that his definition also encompasses feelings as exposed in Broom (1991). He provided arguments for the evolution of feelings as part of animal functioning that are explained in Broom (1993;1998) and Broom and Fraser (2007). There is now a more consensual view of what is animal welfare between the different scientists, involving previously defined approaches, as stated in Wemelsfeder and Mullan, 2014. Then, I would like to ask the authors to consider these updates on animal welfare definitions

We rephrased the statement. LL 31-40

• “seem to indicate a state of good welfare” (L42): Or less negative welfare? I think this is more subtle... please see Ahloy-Dallaire et al, 2018 on this topic.

We rephrased the sentence and added the reference. L 47

• L57, ref [22,23]: I am sorry to say this, but these references do not seem to be the most appropriate since they are about the link between the salivary oxytocin level/the oxytocin receptor polymorphism in mothers and their maternal behaviour, but not about the development of the newborns per se. Instead, there are plenty of most adequate references on this topic, e.g. Veenema (2012) or Mellor (2015) for reviews, or Foyer et al (2016), and of course Guardini et al (2016 & 2017). Please change for more appropriate references.

Amended. [26]

• L68, ref [29,30]: Again, I think there are much more appropriate references to cite to illustrate the use of cortisol as a biomarker of stress in mammals, including very complete reviews.

Amended. [33]

• “but the scientific literature about the possible correlations between these variables is scarce” (L81-82): It is not accurate... There are many publications combining physiological (of any kind, such as heart rate variability, eye white, brain activity or neuromodulator measurements) and behavioral parameters to assess animal emotions/animal welfare. For examples in livestock species, please see Briefer et al, 2015 in goats; Désiré et al, 2004 in lambs; Rietmann et al, 2004, Bachmann et al, 2003, and Mendonça et al, 2019 in horses; Geverink et al, 2002, Rault et al, 2013, Carreras et al, 2016 and Marcet-Rius et al, 2018 (moreover in the context of play) in pigs; and finally in cows, Ternman et al, 2012, Proctor et al, 2015, and Doerfler et al, 2017.

We added “in dairy calves”. L 97

• L95: do you mean between 30 min and 2h?

Amended: “between 30 min and 2 h after birth”. L 110

• L111: Was this kit previously validated to assay cortisol in cow serum or was it previously used as such in other publications?

The ELISA kit was not validated in cow serum, but it was previously used in other publications to assay salivary cortisol:

• Harewood, E.J. and McGowan, C.M., 2005. Behavioral and physiological responses to stabling in naive horses. Journal of Equine Veterinary Science, 25(4), pp.164-170.
• De Palo, P., Maggiolino, A., Albenzio, M., Caroprese, M., Centoducati, P. and Tateo, A., 2018. Evaluation of different habituation protocols for training dairy jennies to the milking parlor: Effect on milk yield, behavior, heart rate and salivary cortisol. Applied animal behaviour science, 204, pp.72-80.

and plasma/serum cortisol:

• Gutiérrez, J., Gazzano, A., Pirrone, F., Sighieri, C. and Mariti, C., 2019. Investigating the role of prolactin as a potential biomarker of stress in castrated male domestic dogs. Animals, 9(9), p.676.
• Ogi, A., Mariti, C., Baragli, P., Sergi, V. and Gazzano, A., 2020. Effects of stroking on salivary oxytocin and cortisol in guide dogs: preliminary results. Animals, 10(4), p.708.
• Naddaf, H., Varzi, H.N., Sabiza, S. and Falah, H., 2014. Effects of xylazine-ketamine anesthesia on plasma levels of cortisol and vital signs during laparotomy in dogs. Open veterinary journal, 4(2), pp.85-89.
• Tatara, M.R., Krupski, W., Tymczyna, B. and Luszczewska-Sierakowska, I., 2012. Biochemical bone metabolism markers and morphometric, densitometric and biomechanical properties of femur and tibia in female and gonadectomised male Polish Landrace pigs. Journal of Pre-Clinical and Clinical Research, 6(1).
• Ekiz, B., Ekiz, E.E., Kocak, O., Yalcintan, H. and Yilmaz, A., 2012. Effect of pre-slaughter management regarding transportation and time in lairage on certain stress parameters, carcass and meat quality characteristics in Kivircik lambs. Meat Science, 90(4), pp.967-976.

Moreover, the aim of the study was to compare the cortisol basal concentration at different time points and no comparison with other studies was performed. We added some references. LL 127-128

 

• L126-129: What about the intra-rater observer reliability? Have you calculated it for the single chosen observer? Because you might have selected the observer based on this intra-rate reliability instead of randomly, as it seems here. Anyway, it is always preferable to ensure the accuracy of the behavioral assessments based on two behavioral observations from two independent observers, and a good agreement between the two observers calculated only on 12% of the videos is not a sufficient reason to discard one behavioural observation.

We agree with the reviewer, but we assessed only six (only one with measurable duration) locomotory behaviors quite simple to recognize (low response rate). Moreover, Hausman et al. (2021) suggested that “no significant differences in inter observer agreement were obtained at the various total inter observer agreement cutoffs (i.e., 30%, 25%, 15%, and 10%)”. We added the reference LL 144-145

• Hausman, N.L., Javed, N., Bednar, M.K., Guell, M., Schaller, E., Nevill, R.E. and Kahng, S., 2021. Interobserver agreement: A preliminary investigation into how much is enough?. Journal of applied behavior analysis.
• L140-158: Why didn't you run correlation analyses with the other types of play behaviours? Even if the results can be considered as "negative" because non-significant, it is still interesting and informative for the scientific community to show them. Could you please add these correlation analyses? or at least mention their outcomes?

We used all the variables in the correlation. We realized it was not clear that we divided between play point events (for which we summed up all the events of play) and play state events (for which we only have running behavior). We have now clarified this in the data analysis section.

• L160-162: same remark as in the abstract about the Pearson’s coefficient. Also, because of the coefficient values (between 0.2 and 0.4), you should specify that both are weak correlations.

We have now checked this in the results as well and added the term “weak”. L 183

• L164: all locomotor play as listed in Table 1 (the 6 types) or only the running? Please clarify.

We have now clarified in the data analysis section.

• Figure 1 legend: the correlation being positive, as shown in the figure and written in the text, the coefficient cannot be negative! Be careful to this mistake repeated several times.

Thanks again!

• L200-201: this is a result. You should have mentioned it before, and possibly provided the detailed statistical results. Also, please clarify your sentence to state exactly what parameters you attempted to correlate.

We realized we missed some information in that sentence that was based on results shown in the results section.

• L205-208: Could you please clarify your hypothesis? Sorry, I don't really understand what you mean since running time is increasing over the time of the first month... don't you consider running as a play behavior?

We considered running as a play behavior. We have rephrase the sentence: “The relative time spent running by the calves tended to increase over time during the first 24 days of observation (D24 > D17 > D10 >D3)”. LL 233-235

• “measured non-invasively in saliva” (L256): And in hair too. Hair cortisol seems to be a more integrative biomarker of cortisol secretion, and as such a better predictor of chronic/ long-term stress, like N/L. It could be of interest to compare N/L with this kind of cortisol measurement by the way. See the recent review of Heimbürger et al, 2019 for instance. Moreover, this type of measurement has already been performed in cows: Comin et al (2011, 2013) and Burnett et al (2014, 2015).

We thank the reviewer for the kind advice. We added the references L 295 [70-72]

• L257: please remove “it”.

Amended. L296

• References: ref. 23 is incomplete (and inappropriate by the way, as noted here above) and the format of ref [40] does not seem correct.

Amended. The inappropriate reference has been replaced [26] and the format has been corrected [46]

Reviewer 2 Report

General comments

• Power calculation and blinded observers are not mentioned.
• There are issues with statistical analyses
• Disbudding should be mentioned but no inferences/conclusions should be drawn as the study was never designed to assess the effect of disbudding in the first place

 

Abstract

Could the authors mention (briefly) why these markers were chosen?

L19-22 I am not sure to understand as disbudding was never mentioned before. Were calves tested before and after disbudding? Please clarify

Introduction

L28 “European” arguably elsewhere as well!

L39 I would precise “positive welfare” as argued in a recent review (Keeling LJ, Winckler C, Hintze S and Forkman B (2021) Towards a Positive Welfare Protocol for Cattle: A Critical Review of Indicators and Suggestion of How We Might Proceed. Front. Anim. Sci. 2:753080.)

L42 “seems”

L47 not sure to understand what “relaxed field” means. Can the authors paraphrase?

L48 I would probably cite Jensen 1998 for this.

L51 to 53 The authors may be right, but the study cited only considered the effect of low energy intake on play behaviours. Please provide additional citations to support the claims made. Other factors may affect play behaviors such as the space given to these animals.

L54 not sure if this is a key factor but it is a key target at least.

L54 is stimulation of play behaviours really the point or what it means in terms of welfare?

L70 I’m quite sure there are many studies showing correlations between chronic stress and cortisol concentrations. To avoid cherry-picking, the authors may want to highlight the inconsistencies of using cortisol as an affective states marker (see Ede et al., 2019 for a discussion on this point).

L72 not sure how the N/L ratio can be more reliable than cortisol concentrations when it is directly dependent.

L76 “be done carefully”

L80 not sure what the authors mean by “the animal’s sensory experience”. Do the authors mean emotional or affective states?

L85 it is the first time that disbudding is mentioned. Please introduce this before or delete if disbudding was not truly part of the study.

 

Methods

“(T3, T10, T17, T24, T31)” Shouldn’t it be D for days instead of T for time I presume?

L137 “The calves received no analgesic treatment with nonsteroidal anti-inflammatory drug (NSAID).” Can the authors explain why? At this stage it is still not clear whether pain due to hot-iron disbudding is part of the study so one may wonder why calves did not receive the best pain mitigation strategy available (that is later mentioned in the discussion!).

L141 How were the different behaviours included in locomotory play aggregated? Did the authors run such analyses for each time-point? That would be a lot of tests!

 

I am really not sure pearson correlations are appropriate in this case. My understanding, based on the limited information provided, is that the authors used every time point collected per individual on all time-points. This artificially increases the sample size (n= 105 instead of 21). A more appropriate analysis would be to run a linear model including calves as random factors, cortisol or N/L as outcome variable, N/L or cortisol as dependent variable and time as repeated measures. This would account for the repeated measurements.

 

Results

L161 the authors report a positive correlation but a negative r score “-0.341”.

L173 not clear from the figure when disbudding was performed neither why it is mentioned here.

Is only running time assessed? What about other play behaviours? The authors need to explain what happens to the other variables or if they were not used, it needs to be mentioned too.

I would like the authors to consider a different statistical approach. By simply considering time as continuous the authors could explore the change over time of their outcome variables. This would greatly simplify the statistical analyses and avoid so many, arguably useless, statistical tests.

 

Figure 2: it is not clear what y axes refer to, especially for running time. What does “relative duration of running” mean? Please clarify.

 

Discussion

L197 but only time spent running is analyzed! What happened to the behaviours described in table1?

L198 These behaviours are not typical described as play behaviours so I would delete this sentence.

By the way, I still wonder why disbudding is mentioned in this study. The study was clearly not designed to assess disbudding pain.

My understanding is that calves were disbudded during the experiment, but the effect of disbudding was not assessed per se.

If I am right, disbudding should only be mentioned as a note for the readers that calves were disbudded on XX days regardless of the study objectives. The authors should not, however, comment on the effect of disbudding given it was not in the study objectives and their data do not allow for strong inferences as calves were disbudded at different ages.

L200 ok but this should be mentioned before!

L203 “during the 24 days of observation”

L205 “the time spent running at T31 was slightly lower than at T24. It could be hypothesized that the calves tended to gradually decrease their motivation to play at the turn of the first month of life. The reason of this decrease could be due to a reduction in the novelty of the pen [27,51] or it could reflect an effect of age and gradual weaning [9,51].”

This is extremely speculative, please delete.

L209 the authors should not comment these results as they obviously never intended to include the effect of disbudding in this study. Please delete.

L215 to 239: the authors cannot discuss these results simply because they haven’t disbudded calves on one specific day. It is arguably completely different to disbud a calf 7 days or 3 days before assessing play behaviours. Please delete and/or replace by a note that play behaviours on day 24 did not seem to be affected by disbudding occurring between day 17 and 21, consistent with previous results showing acute effect of disbudding on play behaviours.

L244 “since the physiological responses to the environment could be considered a more objective measurement of stress than behavioral observation [63], investigating both N/L and serum cortisol helped us to better legitimate the significance of locomotor play behavior.”

I agree we can read that in the literature but mostly by scientists more interested in physiological measures. I am not sure why physiological measures would be more objective and how the current results would “legitimate” locomotor play behaviors as markers of positive welfare.
In addition, I urge the authors to critically look at Figure 1b. Despite the significant relationship (that may not hold when repeated measures are accounted for), it is obvious that the relationship is extremely weak. Just looking at the calves that did not play they range from the two extremes in N/L ratio measures.

I honestly don’t really understand what this relationship would mean anyway? The authors need to be clearer in what they think the biological significance is.

Conclusions
This is a little clearer but please try to clarify the meaning of the potential results in the discussion.

Author Response

General comments

• Power calculation and blinded observers are not mentioned.
• There are issues with statistical analyses
• Disbudding should be mentioned but no inferences/conclusions should be drawn as the study was never designed to assess the effect of disbudding in the first place

We thank the reviewer for the valuable comments which significantly refined our manuscript. 

Abstract

Could the authors mention (briefly) why these markers were chosen?

The limited number of allowed words in the abstract did not give us the chance to explain the reason why we chosen those markers, but we argued the significance of serum cortisol, N/L and locomotor play behavior in the introduction and discussion sessions.

L19-22 I am not sure to understand as disbudding was never mentioned before. Were calves tested before and after disbudding? Please clarify

We replace the world “disbudding” with “the 24^th day of life”. We expanded on disbudding in the intro and discussion as we still think including that was important since the ages considered included disbudding. LL 21-23

Introduction

L28 “European” arguably elsewhere as well!

We removed the word “European”. L 29

L39 I would precise “positive welfare” as argued in a recent review (Keeling LJ, Winckler C, Hintze S and Forkman B (2021) Towards a Positive Welfare Protocol for Cattle: A Critical Review of Indicators and Suggestion of How We Might Proceed. Front. Anim. Sci. 2:753080.)

We added the word “positive” and the suggested reference. L 45

L42 “seems”

Amended. L 47

L47 not sure to understand what “relaxed field” means. Can the authors paraphrase?

We added the statement “(free from threats or intense resource competition)”. L 51

L48 I would probably cite Jensen 1998 for this.

We added the reference [22]

L51 to 53 The authors may be right, but the study cited only considered the effect of low energy intake on play behaviours. Please provide additional citations to support the claims made. Other factors may affect play behaviors such as the space given to these animals.

We rephrase the sentence and change the reference. LL 59-64

L54 not sure if this is a key factor but it is a key target at least.

We rephrase the sentence. LL 59-64

L54 is stimulation of play behaviours really the point or what it means in terms of welfare?

We rephrase the sentence. LL 59-64

L70 I’m quite sure there are many studies showing correlations between chronic stress and cortisol concentrations. To avoid cherry-picking, the authors may want to highlight the inconsistencies of using cortisol as an affective states marker (see Ede et al., 2019 for a discussion on this point).

We add the term “serum/plasma” cortisol. L 77

We acknowledge “that hair cortisol analysis provides a valid and reliable reflection of long-term cortisol secretion”, as reported in Do Yup Lee, E.K. and Choi, M.H., 2015. Technical and clinical aspects of cortisol as a biochemical marker of chronic stress. BMB reports, 48(4), p.209.

L72 not sure how the N/L ratio can be more reliable than cortisol concentrations when it is directly dependent.

“an increase in N/L ratio was reported to be”. We just reported assumptions made elsewhere. Anyway, we agree that the N/L changes are secondary to glucocorticoids release. L78

L76 “be done carefully”

Amended. L 83

L80 not sure what the authors mean by “the animal’s sensory experience”. Do the authors mean emotional or affective states?

We meant “every experienced sensations which arise as the integrated outcomes of sensory and other neural inputs from within the animal’s body and from its environment” as reported in the cited paper, but we replaced “the animal’s sensory experience” with “the animal’s emotional/affective states”. L 96

L85 it is the first time that disbudding is mentioned. Please introduce this before or delete if disbudding was not truly part of the study.

 We rephrased the statement. LL 87-101

Methods

“(T3, T10, T17, T24, T31)” Shouldn’t it be D for days instead of T for time I presume?

We used T for time as that is usually what is presented but we agree that in our case it makes more sense to use D as they correspond to specific days.

L137 “The calves received no analgesic treatment with nonsteroidal anti-inflammatory drug (NSAID).” Can the authors explain why? At this stage it is still not clear whether pain due to hot-iron disbudding is part of the study so one may wonder why calves did not receive the best pain mitigation strategy available (that is later mentioned in the discussion!).

We rephrased the statements in the abstract and introduction. Now we should have clarified that disbudding is not part of the study and the disbudding treatment is only part of the farm management. Please, see also LL 262-279.

L141 How were the different behaviours included in locomotory play aggregated? Did the authors run such analyses for each time-point? That would be a lot of tests!

We have now clarified it as also reviewer 1 was confused about it. Mainly, there is only one state event (i.e., running) and we used time of running for it, the other point events (the other locomotory behaviours that happen very quick and we cannot estimate the duration) were all sum up to obtain the variable point events of locomotory play.

I am really not sure pearson correlations are appropriate in this case. My understanding, based on the limited information provided, is that the authors used every time point collected per individual on all time-points. This artificially increases the sample size (n= 105 instead of 21). A more appropriate analysis would be to run a linear model including calves as random factors, cortisol or N/L as outcome variable, N/L or cortisol as dependent variable and time as repeated measures. This would account for the repeated measurements.

We really appreciate this comment as it made us discover a new technique. We used correlations and not LMM as we wanted to test an association between two variables rather than a unidirectional influence of a variable over the other. We found a way to incorporate repeated measurements in correlations, possible with the package rmcorr in R. We have now edited the data analysis and results sections.

Results

L161 the authors report a positive correlation but a negative r score “-0.341”.

Many thanks for spotting this mistake (i.e., added sign - by mistake). The results changed since we ran repeated measures correlations.

L173 not clear from the figure when disbudding was performed neither why it is mentioned here.

Information added to Figure 2. L 200

Is only running time assessed? What about other play behaviours? The authors need to explain what happens to the other variables or if they were not used, it needs to be mentioned too.

We have now clarified this, see point above.

I would like the authors to consider a different statistical approach. By simply considering time as continuous the authors could explore the change over time of their outcome variables. This would greatly simplify the statistical analyses and avoid so many, arguably useless, statistical tests.

We thank the reviewer for suggesting this approach but we think the current approach is better to highlight our main results. This is because we can now present estimates for the five time periods considered, and we can clearly show the pre- vs post- disbudding (that is not a main aim of the study but it should still be considered as it happens during the study period). We do not really have a continuous predictor as we collected data at five different times, so we think using this predictor as factor rather than continuous would be better. We hope that the reviewer agrees with us. We do not think that the analysis would be simplified as we would have the same number of models (4 as the dependent variables are 4). 

Figure 2: it is not clear what y axes refer to, especially for running time. What does “relative duration of running” mean? Please clarify.

We changed to relative time spent running in the caption. This is the relative duration (percentage on total time) of running behavior, as specified in the methods.

 

Discussion

L197 but only time spent running is analyzed! What happened to the behaviours described in table1?

We agree that it was confusing, we have now clarified it, see comments above.

L198 These behaviours are not typical described as play behaviours so I would delete this sentence.

We agree that those are not typical play behaviors, therefore we rephrase the sentence not mentioning disbudding. LL 224-228

By the way, I still wonder why disbudding is mentioned in this study. The study was clearly not designed to assess disbudding pain.

My understanding is that calves were disbudded during the experiment, but the effect of disbudding was not assessed per se.

If I am right, disbudding should only be mentioned as a note for the readers that calves were disbudded on XX days regardless of the study objectives. The authors should not, however, comment on the effect of disbudding given it was not in the study objectives and their data do not allow for strong inferences as calves were disbudded at different ages.

The calves were disbudded during the experiment and the effect of disbudding was not assessed per se, therefore we entirely rephrase the statements on disbudding. We still included some information about disbudding for the reader, and discussed it briefly.

L200 ok but this should be mentioned before!

We rephrased the sentence. LL 224-228

L203 “during the 24 days of observation”

The definition of relative duration of running in the arena is “percentage of total time spent by the calves running in the arena during the 20 min of observation”.

L205 “the time spent running at T31 was slightly lower than at T24. It could be hypothesized that the calves tended to gradually decrease their motivation to play at the turn of the first month of life. The reason of this decrease could be due to a reduction in the novelty of the pen [27,51] or it could reflect an effect of age and gradual weaning [9,51].”

This is extremely speculative, please delete.

Amended. LL 236-239

L209 the authors should not comment these results as they obviously never intended to include the effect of disbudding in this study. Please delete.

We removed the statement. LL 240-245

L215 to 239: the authors cannot discuss these results simply because they haven’t disbudded calves on one specific day. It is arguably completely different to disbud a calf 7 days or 3 days before assessing play behaviours. Please delete and/or replace by a note that play behaviours on day 24 did not seem to be affected by disbudding occurring between day 17 and 21, consistent with previous results showing acute effect of disbudding on play behaviours.

We moved the first part of the statement (LL 87-94) to introduce the reason why we discussed the disbudding procedure, and we also moved the second part of the sentence (LL 273-279) in order to discuss the disbudding treatment performed at CiRAA and the pattern of physiological parameters.

L244 “since the physiological responses to the environment could be considered a more objective measurement of stress than behavioral observation [63], investigating both N/L and serum cortisol helped us to better legitimate the significance of locomotor play behavior.”

I agree we can read that in the literature but mostly by scientists more interested in physiological measures. I am not sure why physiological measures would be more objective and how the current results would “legitimate” locomotor play behaviors as markers of positive welfare.
In addition, I urge the authors to critically look at Figure 1b. Despite the significant relationship (that may not hold when repeated measures are accounted for), it is obvious that the relationship is extremely weak. Just looking at the calves that did not play they range from the two extremes in N/L ratio measures.

I honestly don’t really understand what this relationship would mean anyway? The authors need to be clearer in what they think the biological significance is.

We rephrased the sentence. LL 284-287

Conclusions
This is a little clearer but please try to clarify the meaning of the potential results in the discussion.