Development of a Group-Adapted Housing System for Pregnant Sows: A Field Study on Performance and Welfare Aspects

Round 1

Reviewer 1 Report

The study holds promise in two areas: 1. As a reference in the circadian patterns of feeding. 2. the incidence of lameness based on gestation day is valuable and would suggest that the authors discuss current literature on time frame between trauma (interaction and lesions at the feeder) and subsequent clinical indications (lameness). Welfare occurs at the level of the stockperson-pig interface - thus providing stockpersons will a window of observation. Scientific soundness - 1. The study is designed around 2 feeding systems.Therefore, it is a major flaw in the study to be deficient on amount of feed consumed versus disappearance. 2. The study uses 114 animals and outcomes such as lameness, affect on litter performance, in spite of observations, is restricted to 114 sample size. The low incidence of moderate lameness, subjective scoring method (~2 steps per limb) and lack of description of observation (side or back or top view) required a statistical power analysis of the study to measure lameness is weak. This should be included in the discussion and conclusion.

Author Response

Reviewer 1

Research design, methods, results and conclusions can be improved

Comments and Suggestions for Authors

The study holds promise in two areas: 1. As a reference in the circadian patterns of feeding. 2. the incidence of lameness based on gestation day is valuable and would suggest that the authors discuss current literature on time frame between trauma (interaction and lesions at the feeder) and subsequent clinical indications (lameness).

Thank you for your comments and suggestions for improvement.

We now included a part in the discussion to discuss the lameness prevalence, including also detail on the time frame between trauma (hierarchical fights) and subsequent clinical indications (lameness) (L363-373): “Maes et al. [19] and Bos et al. [22] reported a general higher lameness prevalence in group-housed sows compared to individually housed sows. Bos et al. [22] also observed a higher lameness risk within the first three days after grouping sows. When looking at our results, considering both systems, the highest lameness prevalence was detected on the 38th day of gestation, i.e., the first measured time-point after group-housing. This higher lameness prevalence also concurs with higher injury index, and both are assumed to result from increased hierarchy fights due to grouping. Ziron [11]and Pluym et al. [23] drew the same conclusion, the former specifying that rank fights increase the danger of claw injuries that can thus lead to serious foundation problems. While Bos et al. [22] as well report lower lameness scores in sows housed in static groups compared to dynamic groups (where more hierarchical fights occur), Pluym et al. [23] found no difference.”.

Welfare occurs at the level of the stockperson-pig interface - thus providing stockpersons will a window of observation.

Though we agree that human-animal relationship play a main role in animal welfare, we are not sure to understand what is mean here.

Scientific soundness - 1. The study is designed around 2 feeding systems. Therefore, it is a major flaw in the study to be deficient on amount of feed consumed versus disappearance.

We can only agree with your comment. Unfortunately, in the current field study, it was not possible to measure directly or indirectly the amount of feed consumed or that landed on the floor and could therefore be mixed with feces. As explained (L100-101 existing system and L120-122 SWOF system), this was not measured in any of the two systems. Therefore, this “bias” was the same for both systems. Besides, we looked for references in the literature that would have measured this “wasted food” and did not find a good one.

We reformulate how the amount of food delivered/consumed was assessed (L100-101 existing system) “The amount of feed delivered by the feeders was considered as the amount of feed consumed by the sows, since no food was visible anymore in the trough” and (L120-122 SWOF system) “Although the floor was visibly soiled with food and/or feces, the amount of feed delivered by the feeders was considered as the amount of feed consumed by the sows as the wasted food could not be measured”. We additionally added detail in the Measurements’ section (L147-151): “This average feed consumption per sow was estimated based on the total amount of food delivered divided by the number of sows allocated to the feeding area. In the SWOF system, the total amount of food delivered was recorded every day. In the existing system, the feed dispensers were calibrated (to control this calibration, the food delivered was weighed at two time-point during the study period)”.

2. The study uses 114 animals and outcomes such as lameness, affect on litter performance, in spite of observations, is restricted to 114 sample size. The low incidence of moderate lameness, subjective scoring method (~2 steps per limb) and lack of description of observation (side or back or top view) required a statistical power analysis of the study to measure lameness is weak. This should be included in the discussion and conclusion.

We had indeed around 114 focus animal per time point. This should be enough but you are right that the lameness had a low incidence (i.e., 11.92% in SWOF and 6.22% in existing system) so the statistical power was not so high. We now acknowledge this in the discussion (L359-362): “However, we recorded a low lameness prevalence (11.92% in the SWOF and 6.22% in the existing system) in the 114 focus sows studied. Therefore, it is possible that the statistical power might have been too low to detect a difference between the two systems”. In a future study, more observations should be performed to validate our findings. Besides, the lameness prevalence we found in agreement with other studies that found between 6 and 17% of lame sows (e.g., Maes et al., 2016). In the conclusion, we  now specified that the system did not seem to affect the lameness prevalence (L359-362) “However, we recorded a low lameness prevalence (11.92% in the SWOF and 6.22% in the existing system) in the 114 focus sows studied. Therefore, it is possible that the statistical power might have been too low to detect a difference between the two systems” and (L443-445) “As a consequence, the results presented in this study, in particular as regards to the injury index and the low lameness prevalence, remain to be further validated taking into account the group sizes and management practices”.

As regards to the litter performance, the statistical power is more than enough since data of 50 sows in the existing system and 51 sows in the SWOF system were included. The number of degrees of freedom available is well within the sample sizes to reach statistically significant results. It is generally considered necessary that the total degrees of freedom available are about 5 times the degrees of freedom required for the fixed effects (Gygax, 2014; Friston, 2012).

We now added a more detailed description (side view) on how we observed and scored lameness (L167-170): “Sows were observed from the side when walking at least 10 steps on concrete floor. The assessor, a trained veterinarian, was not further than 4 meters away and had a clear and unobstructed view of the moving sow as recommending in the Welfare Quality protocol [14]”. Though only one observer scored all the sows on a short distance only, it was a trained veterinarian. In the Welfare Quality 2009 report, no minimal distance is mentioned but only “a certain distance”. It would have been scientifically better to have more than one observer to avoid subjective scoring, or to do an intra-observer agreement.

Reviewer 2 Report

The aim of the study is to evaluate a new liquid feeding system for group housed sows. This system present two ad libitum feeding areas, one with a low-energy food and one with high-energy food. According to the authors, this system present several advantages as possible synchronous feeding and natural biphasic feeding rhythm of the sows. The effect of this new feeding station is assessed through a comparison of two groups of sows, one fed with this new system and one fed with a dry died distributed in a long trough. Indicators are sows’ body weight, lameness, integument injuries, vulva injuries, displacements at the trough, feed intake rhythm and litter performance.

General comment

This manuscript is well written and easy to read. It studies an interesting feeding system, allowing the respect of the sows’ behavior (synchronization of feeding, biphasic feeding rhythm) and reducing fight and restricted food frustration. However, the new food distribution system should to be more clearly explain. The experimental design presents some major biases, from which some have been identified by the authors. These biases are differences in housing systems that overlap with the two feeding systems: 1) stable vs. dynamique groups, which can generate differences in fights to establish hierarchy, 2) fully vs. partial slatted floor, which can generate differences in locomotory disorders (and lameness), 3) diet composition, which has a direct impact on the sows’ weight. Thus, the results have to be considered with caution, as underligned by the authors, even if they are considered as “first insights”. In the discussion part, the authors discuss their results on body weight, body and vulva lesions, but their do not discuss the results on lameness and on the litter performance characteristics and there is no mention of synchonisation. These results have to be discussed.

Specific comment

Lines 42-45: from whom point of view an optimal feeding system during gestation should match with these requirements?

Line 45: Relationship between “production-related diseases (e.g., lameness)” and feeding system need to be developed.

Lines 59-61: this reference should be kept for discussion

Lines 80-83: I am not sure that these precision are needed here

Lines 83-86: when do the sows arrive into the group after service?

Line 98: you miss 3 sows?

Lines 104-107: how many time per day is the food distributed?

Figure 2: you should draw the sow flow in the plan to help comprehension. You should also draw the doors (or gates) through which the sows get. In the plan, there seems to be only solid walls and there is no understanding of how sows move from one compartment to another.

Line 118-129: I do not understand how the food is delivered. Ad libitum? So how can you evaluate the amount of food eaten by each sow? How can you know the exact number of sows coming to each trough? Does the gate stop to allow the sows to enter the compartment after the eighteenth sow? What is the dilution rate of the feed?

Line 140-142: I am not sure that these precision are needed here

Line 178: why vulva injuries were not recorded at the same dates than integument scoring?

Lines 205-207: what was the trouble with these sows? Are their illness can be caused by the feeding system?

Line 241: should be Table 4?

Lines 245-246: does this result take into consideration the difference at D31?

Lines 340-354: what conclusion can you draw on individual energy adapted feeding and on sow satiety of the SWOF?

Author Response

Comments and Suggestions for Authors

The aim of the study is to evaluate a new liquid feeding system for group housed sows. This system present two ad libitum feeding areas, one with a low-energy food and one with high-energy food. According to the authors, this system present several advantages as possible synchronous feeding and natural biphasic feeding rhythm of the sows. The effect of this new feeding station is assessed through a comparison of two groups of sows, one fed with this new system and one fed with a dry died distributed in a long trough. Indicators are sows’ body weight, lameness, integument injuries, vulva injuries, displacements at the trough, feed intake rhythm and litter performance.

This is indeed a good summary of our study.

General comment

This manuscript is well written and easy to read. It studies an interesting feeding system, allowing the respect of the sows’ behavior (synchronization of feeding, biphasic feeding rhythm) and reducing fight and restricted food frustration. However, the new food distribution system should to be more clearly explain.

Thank you very much for your comment and further suggestions for improvement. We answered in detail in the point-by-point answers hereafter.

The experimental design presents some major biases, from which some have been identified by the authors. These biases are differences in housing systems that overlap with the two feeding systems: 1) stable vs. dynamique groups, which can generate differences in fights to establish hierarchy, 2) fully vs. partial slatted floor, which can generate differences in locomotory disorders (and lameness), 3) diet composition, which has a direct impact on the sows’ weight. Thus, the results have to be considered with caution, as underligned by the authors, even if they are considered as “first insights”.

We can only agree with your comment. As you recognise, we acknowledged these confounding factors and tried to consider our results as cautiously as possible.

In the discussion part, the authors discuss their results on body weight, body and vulva lesions, but their do not discuss the results on lameness and on the litter performance characteristics and there is no mention of synchonisation. These results have to be discussed.

You are completely right. We are not sure what happened but we initially wrote a section on lameness and litter performance characteristics that disappeared in the version of the manuscript we sent. We added these points in the discussion (L353-373 for lameness): “The crude fiber in the silages in the SWOF system lead to visible larger amount of feces on the floor [17]. According to Ebertz et al. [18], crude fibers caused more smeared slots on the slatted floor and the increased excrements produced poor floor cleanliness. Contrary to Maes et al. [19], that observed a 2.8 times higher risk of lameness in pens with dirty, wet slatted floors, than in pens with good floor hygiene, we observed no difference of the lameness prevalence between the two systems. Contrary to Willgert et al. [20] that reported higher risk of lameness for younger sows (two parities or less), we did not find a parity effect on lameness. However, we recorded a low lameness prevalence (11.92% in the SWOF and 6.22% in the existing system) in the 114 focus sows studied. Therefore, it is possible that the statistical power might have been too low to detect a difference between the two systems. Previous studies reported similar lameness prevalence in pregnant sows, i.e., between 6 and 17% (for a review see [21] and [19]. Maes et al. [19] and Bos et al. [22] reported a general higher lameness prevalence in group-housed sows compared to individually housed sows. Bos et al. [22] also observed a higher lameness risk within the first three days after grouping sows. When looking at our results, considering both systems, the highest lameness prevalence was detected on the 38th day of gestation, i.e., the first measured time-point after group-housing. This higher lameness prevalence also concurs with higher injury index, and both are assumed to result from increased hierarchy fights due to grouping. Ziron [11]and Pluym et al. [23] drew the same conclusion, the former specifying that rank fights increase the danger of claw injuries that can thus lead to serious foundation problems. While Bos et al. [22] as well report lower lameness scores in sows housed in static groups compared to dynamic groups (where more hierarchical fights occur), Pluym et al. [23] found no difference”.

Concerning the mention of sow behavioural synchronisation (i.e., the behaviour of several individuals related in time; Docking et al., 2008), though it is an interesting point, this is out of the scope of this study. We did not record other behaviours than the (biphasic) feed intake rhythm and already discussed it L426-430.

Specific comment

Lines 42-45: from whom point of view an optimal feeding system during gestation should match with these requirements?

The herein given requirements for an optimal feeding system during gestation are from our point of view and based on the three welfare aspects defined by Fraser (2008). We reformulated this section to precise it (L42-47) ”To ensure animals with a good welfare, three aspects should be considered: (1) the health and biological functioning, (2) the ability to express a natural behavior, and (3) the affective state [1]. Consequently, in our view, an optimal feeding system for pregnant sows during gestation should (1) provide sows with optimal nutrients allowing the best body condition for farrowing and avoid production-related diseases, as well as (2) enable sows to follow their natural feeding behavior.” and (L49-50) “Yet, these conventional feeding systems do not meet the requirements for a good welfare”.

Line 45: Relationship between “production-related diseases (e.g., lameness)” and feeding system need to be developed.

Lameness is an important production-related disease in sows. Therefore, we think that when developing a new feeding system, one should consider this aspect to avoid increasing the lameness prevalence (or even to reduce it). Indeed, different factors related to the feeding system influence lameness including, according to Maes et al. (2016), the group size, (hierarchical) agonistic interactions, pen design (e.g., use of divider panels), stock density, floor cleanliness (e.g., type of food). A high BCS (over-conditioning due to feed over-consumption) could also lead to lameness. We now reformulated this paragraph (L42-47), as explained in our answer to your previous comment, and we now discuss the lameness prevalence and its link to the feeding system in the discussion (L353-357) “The crude fiber in the silages in the SWOF system lead to visible larger amount of feces on the floor [17]. According to Ebertz et al. [18], crude fibers caused more smeared slots on the slatted floor and the increased excrements produced poor floor cleanliness. Contrary to Maes et al. [19], that observed a 2.8 times higher risk of lameness in pens with dirty, wet slatted floors, than in pens with good floor hygiene, we observed no difference of the lameness prevalence between the two systems” and (L359-373) “However, we recorded a low lameness prevalence (11.92% in the SWOF and 6.22% in the existing system) in the 114 focus sows studied. Therefore, it is possible that the statistical power might have been too low to detect a difference between the two systems. Previous studies reported similar lameness prevalence in pregnant sows, i.e., between 6 and 17% (for a review see [21] and [19]. Maes et al. [19] and Bos et al. [22] reported a general higher lameness prevalence in group-housed sows compared to individually housed sows. Bos et al. [22] also observed a higher lameness risk within the first three days after grouping sows. When looking at our results, considering both systems, the highest lameness prevalence was detected on the 38th day of gestation, i.e., the first measured time-point after group-housing. This higher lameness prevalence also concurs with higher injury index, and both are assumed to result from increased hierarchy fights due to grouping. Ziron [11]and Pluym et al. [23] drew the same conclusion, the former specifying that rank fights increase the danger of claw injuries that can thus lead to serious foundation problems. While Bos et al. [22] as well report lower lameness scores in sows housed in static groups compared to dynamic groups (where more hierarchical fights occur), Pluym et al. [23] found no difference”.

Lines 59-61: this reference should be kept for discussion

We moved and reformulated/included around reference to Whittaker et al. (2000)’s study to the discussion (L375-381) “This is in agreement with Petherick and Blackshaw [10], Whittaker et al. [13] and van der Peet-Schwering [24] that looked at restrictively vs. (non group-adapted) ad libitum fed group-housed pregnant sows. On the contrary, Cools et al. [25] documented a greater number of total born and born alive piglets for sows fed restrictively. Moreover, such as Whittaker et al. [13], we found no effect of the parity on the litter performance, whereas Petherick and Blackshaw [10] observed that sows in their 7th parity and over had fewer born alive and more stillborn piglets compared with sows in their 2nd to 6th parity”.

Lines 80-83: I am not sure that these precision are needed here

We considered your suggestion and removed all the detail about prophylaxis and vaccinations as these are usual for pregnant sows and in this farm.

Lines 83-86: when do the sows arrive into the group after service?

The sows arrived into the group around 4 weeks after service. We added this in the Materials and Methods (L80): “Sows were transferred to the group-housing systems approximatively four weeks after service”.

Line 98: you miss 3 sows?

That is exact, we made a mistake in our calculations. We thoroughly checked our data and corrected the numbers in the sentence. There was on average 46 sows and not 44 sows as previously stated with subdivided groups of 18 to 25 sows. The exact repartition was: (Batch 1) 25 and 25 sows, and (Batch 2) 18 and 24 sows. We changed the number of sows involved (L91-92): “In the existing system, sows were kept in a stable group of 46 sows on average divided by the trough in the middle into two groups of 18 to 25 sows each”, as well as in Table 1 (L124) and L333. We accordingly changed the animal-place-ratio in Table 1 (L124).

Lines 104-107: how many time per day is the food distributed?

Restrictively fed sows received food once or twice in the morning. We included this (L99-100): “The feed was delivered once or twice between 6 am and 10 am”.

Figure 2: you should draw the sow flow in the plan to help comprehension. You should also draw the doors (or gates) through which the sows get. In the plan, there seems to be only solid walls and there is no understanding of how sows move from one compartment to another.

As you thoughtfully suggested, we worked on Figure 2 (L110-111) to make the sows’ flow (using arrows), as well as the sorting and reverse gates clearer. Doors are now shown by dotted black lines.

Line 118-129: I do not understand how the food is delivered. Ad libitum? So how can you evaluate the amount of food eaten by each sow?

Indeed, the food was provided ad libitum in both the A and B areas of the SWOF system, as stated L112-115. We now explain more in detail how we evaluated the amount of food eaten by each sow (L147-151): “This average feed consumption per sow was estimated based on the total amount of food delivered divided by the number of sows allocated to the feeding area. In the SWOF system, the total amount of food delivered was recorded every day. In the existing system, the feed dispensers were calibrated (to control this calibration, the food delivered was weighed at two time-point during the study period)”. Besides, as explained (L145-147) “In both systems, the amount of feed delivered by the feeders was considered as the amount of feed consumed by the sows”. Though this is not ideal, there was no technical possibility to measure more precisely the amount of food eaten or wasted (i.e., that landed on the floor).

How can you know the exact number of sows coming to each trough? Does the gate stop to allow the sows to enter the compartment after the eighteenth sow?

If we know from the video recordings how many sows ate at each trough, this sentence had a theoretical meaning: we calculated based on the trough length (4.5 m) that a maximum of 9 sows could eat synchronously per trough. For more clarity, we moved and reformulated this sentence (L118-119): “This allowed a synchronized feed intake of a maximum of 9 sows per trough at all times in each feeding area”. The gate did not stop sows to enter the respective feeding areas, even when more than 18 sows would enter them.

What is the dilution rate of the feed?

We now added that the mixture had a water-to-feed ratio of 3.5:1 on a dry matter basis (3.5 parts of water per one part of feed) (L134-135): “The WPWS was mixed with other supplemented feed components (Table 3) and water with a water-to-feed ratio of 3.5:1 on a dry matter (DM) basis”.

Line 140-142: I am not sure that these precision are needed here

The reformulated and shorten this paragraph (L133-135):” In the SWOF system, the liquid feed was calculated based on whole plant wheat silage (WPWS). The WPWS was mixed with other supplemented feed components (Table 3) and water with a water-to-feed ratio of 3.5:1 on a dry matter (DM) basis”.

Line 178: why vulva injuries were not recorded at the same dates than integument scoring?

Unfortunately, we did not initially think about documenting vulva injuries. However, during the experiment, it became clear that vulva injury scoring would be a great and meaningful addition for measuring e.g., aggression consequences. Therefore, we decided to include it as a single score when sows were moved to the farrowing compartments.

Lines 205-207: what was the trouble with these sows? Are their illness can be caused by the feeding system?

The sows that were removed from the two systems were transferred to the sick pen (infirmary) due to unknown illness (e.g., abnormal weight loss), injury (e.g., severe claw lesion, deep body scratches) or unknown reasons (not documented by the farm staff). We added “injury” as other possible cause (L202-205): “Nine focus sows were removed from the systems during the period studied due to e.g., illness or injury, i.e., 4 (batch 1) and 2 (batch 2) sows from the existing system and 2 (batch 1) and 1 (batch 2) sows from the SWOF system”.

We do not think that what happen to these sows were caused by the feeding system.

Line 241: should be Table 4?

That is exact, thank you. We changed it to Table 4 (L238). We also changed to the correct table number L268.

Lines 245-246: does this result take into consideration the difference at D31?

Yes, this result take into account the “initial” (at day 31 of gestation) body weight difference. We now re-precise it (L242-244): “After initial body weight correction (at day of gestation 31), sows in the SWOF system were lighter (p < 0.001), with on average 296.4 ± 2.6 kg [least square mean ± standard error (LSM ± SE)], than sows in the existing system, with on average 310.7 ± 2.5 kg”.

Lines 340-354: what conclusion can you draw on individual energy adapted feeding and on sow satiety of the SWOF?

The sows had the possibility to go to the feeding area and to eat as often as they wanted. Therefore, sows in the SWOF system were very probably satiate. We acknowledged (L345-347) that the energy in the ration (diet) might not have been perfectly adequate for the group neither might it have been for each individual sow. As already discussed L345-347, the SWOF sows higher activity could have led to lower weights than sows in the existing system. All in all, the weight of the SWOF sows (though it was lower) was correct. We could not find a reference to strengthen this idea for this breeding line and we therefore called the Danish Genetics breeding association that confirm that all our sows in both systems were in the upper average.

Reviewer 3 Report

This paper regards the comparison of a novel gestating sow group feeding system compared to a conventional system. The main purpose of the new system was to show that the natural feeding behavior of gestating sows is better met with the SWOF system, and that less injuries and promotion of health would occur using the new system. The complications of an on-farm trial where several confounding factors may occur (two different housing set ups, two different types of diets that might affect interpretation of the data were evident in the experimental design and does make it a challenge to interpret the statistical findings of the study, but it does not make the information unimportant and the authors deal with these challenges well and describe them fully.

The paper is well written and the wording is clear and concise. The use of "Besides" to start out a sentence, however, was found to be distracting, and I recommend rewording every instance of it through out the paper- it was used too frequently and inappropriately, the use of "In addition" or "considering" would be more appropriate. For example, lines 52, 268, 351, 381, and 407 all contain "Besides" which could be replaced with "in addition to" or " it could be further noted".

Line 59 -61 the Whittaker reference is confusing- did they find ad libitum fed sows heavier? It says "if they were found heavier" and is hard to interpret.

Line 87 - 92 focus  animals- this  term is confusing. Not sure what is meant by that. Were these the assigned animals that all  measurements were conducted on? If so, please describe as such.

In considering the differences in weight of the two groups, it might be important to consider and describe what parameters were used to determine if a sow needed to gain more weight or not- what determined whether the sow received the low or high energy diet?

It seems the spillage of the liquid diet may have been a  problem for calculating consumption. One has to wonder why the system was not modified to reduce the amount of spillage.

In the graphic describing the SWOF system, how do sows get into and out of the feeding area near the room for feeding technology? should there be a gate somewhere on that end of the system?

Author Response

Dear Editor,

We would like to submit a revised version of our manuscript „Development of a group-adapted housing feeding system for pregnant sows: a field study on performance and welfare aspects“ by Eva Angermann, Camille M.C. Raoult, Monika Wensch-Dorendorf, Stephanie Frenking, Nicole Kemper and Eberhard von Borell, for publication in agriculture to the Special Issue „Challenges and Perspectives in Pig Farming: Breeding, Husbandry and Management“.

Thank you for the possibility to revise our manuscript. We revised it to the best of our abilities according to the reviewers' comments. We answered point-by-point to Reviewers 2 (round 2) and Reviewer 3, detailing the revisions that have been made in the manuscript (where we used the “track changes” function), citing the line number and exact change and made modifications.

Thank you for considering the revised version of our manuscript.

Sincerely,

Eva Angermann and co-authors

Comments and Suggestions for Authors

This paper regards the comparison of a novel gestating sow group feeding system compared to a conventional system. The main purpose of the new system was to show that the natural feeding behavior of gestating sows is better met with the SWOF system, and that less injuries and promotion of health would occur using the new system. The complications of an on-farm trial where several confounding factors may occur (two different housing set ups, two different types of diets that might affect interpretation of the data were evident in the experimental design and does make it a challenge to interpret the statistical findings of the study, but it does not make the information unimportant and the authors deal with these challenges well and describe them fully. The paper is well written and the wording is clear and concise.

Thank you for your good understanding and further suggestions for improving our work. Please see our point-by-point answers.

The use of "Besides" to start out a sentence, however, was found to be distracting, and I recommend rewording every instance of it throughout the paper- it was used too frequently and inappropriately, the use of "In addition" or "considering" would be more appropriate. For example, lines 52, 268, 351, 381, and 407 all contain "Besides" which could be replaced with "in addition to" or " it could be further noted".

As suggested, we replaced “Besides” by “In addition” (L54, L140, L266, L409, L435) or “It could be further noted” (L349).

Line 59 -61 the Whittaker reference is confusing- did they find ad libitum fed sows heavier? It says "if they were found heavier" and is hard to interpret.

Whittaker et al. (2000) wrote in the results’ section: “During both [studied] parities, A [ad libitum fed] sows were significantly heavier by day 50 of gestation (parity 1: P < 0.05; parity 2: P < 0-01) and at farrowing (parity l: IJ < 0.001; parity 2: P < 0.1101) than R [restrictively fed] sows (Table 2). In addition, A sows gained significantly more weight over the latter half of gestation (parity 1: P < 0.001; parity 2: P 0.001) than R sows. No difference was found in sow weight at weaning, since A sows lost more weight during lactation (parity 1: P < 0.001; parity 2: P < 0.001) than R sows”. Therefore, we do think that this reference is correctly used.

Line 87 - 92 focus animals- this term is confusing. Not sure what is meant by that. Were these the assigned animals that all measurements were conducted on? If so, please describe as such.

We now precise that animal-related measurements (body weight, integument and vulva injuries, lameness and litter performance) were conducted on the focus animals (L81-83): “A total of 114 pregnant sows (second to eleventh parity) were included in this study as focus animals, i.e., measurements were conducted on 58 sows from the existing system (restrictive feeding) and 56 sows from the SWOF system (group-adapted ad libitum feeding)”.

In considering the differences in weight of the two groups, it might be important to consider and describe what parameters were used to determine if a sow needed to gain more weight or not- what determined whether the sow received the low or high energy diet?

As explained L113-116, “Sows entered the feeding area only passing through a sorting gate (Hölscher + Leuschner GmbH & Co. KG, Emsbüren, Germany) that assigned them to a low- (area A) or high-energy liquid ad libitum diet (area B) based on their current weight in relation to their parity (for more detail, see Supplementary Table S1)”.  We now added a supplementary Table (Table S1) detailing the sow body weight objective according to its parity and how this sow was assigned to one or the other energy diet.

It seems the spillage of the liquid diet may have been a problem for calculating consumption. One has to wonder why the system was not modified to reduce the amount of spillage.

Indeed, spillage of the liquid diet probably happened and may have been a problem for calculating consumption. Unfortunately, in the current field study, it was not possible to measure directly or indirectly the amount of feed consumed or that landed on the floor and could therefore be mixed with feces. As explained (L100-101 existing system and L121-123 SWOF system), this was not measured in any of the two systems. In addition, we looked for references in the literature that would have measured this “wasted food” and did not find a good one. Therefore, we remain uncertain how we could have modified the system to reduce the amount of spillage.

In the graphic describing the SWOF system, how do sows get into and out of the feeding area near the room for feeding technology? should there be a gate somewhere on that end of the system?

Based on Figure 2: Sows could theoretically get into the reserve area (the area near the room for feeding technology) passing through the sorting gate and get out of the SWOF system passing through a door on the right side (not visible on the current figure) to go to the farrowing compartments. In our study, this reserve area was not used.

Round 2

Reviewer 2 Report

The authors have taken into account the remarks we made and the manuscript has been improved with their corrections. However I still think that the paper focus mainly on feeding system although other housing parameters, which can influence the results, are different between systems. I would recommend to slightly change the title of the article in “Development of a group-adapted housing system for pregnant sows: a field study on performance and welfare aspects” and keep “ad libitum feeding” in key words.

Author Response

Dear Editor,

We would like to submit a revised version of our manuscript „Development of a group-adapted housing feeding system for pregnant sows: a field study on performance and welfare aspects“ by Eva Angermann, Camille M.C. Raoult, Monika Wensch-Dorendorf, Stephanie Frenking, Nicole Kemper and Eberhard von Borell, for publication in agriculture to the Special Issue „Challenges and Perspectives in Pig Farming: Breeding, Husbandry and Management“.

Thank you for the possibility to revise our manuscript. We revised it to the best of our abilities according to the reviewers' comments. We answered point-by-point to Reviewers 2 (round 2) and Reviewer 3, detailing the revisions that have been made in the manuscript (where we used the “track changes” function), citing the line number and exact change and made modifications.

Thank you for considering the revised version of our manuscript.

Sincerely,

Eva Angermann and co-authors

Author response：

The authors have taken into account the remarks we made and the manuscript has been improved with their corrections. However I still think that the paper focus mainly on feeding system although other housing parameters, which can influence the results, are different between systems. I would recommend to slightly change the title of the article in “Development of a group-adapted housing system for pregnant sows: a field study on performance and welfare aspects” and keep “ad libitum feeding” in key words.

Thank you for your additional comments. You are right, we mainly compared two different housing system. We therefore took your suggestions: changing the title and adding “ad libitum feeding” to the key words.