4.1. Good Feeding
Inside the Good Feeding principle, only one animal-based measure is used: body condition. Sixty percent of the farms assessed had at least one animal that was too lean, and more than 50% of the farms had more than 1% of the animals that were too lean (with the maximum reaching a 6% and 7% in one farm according to each auditor), so this was the most frequent problem seen in the farms in terms of animal based measures (
Table 6). Although farms from auditor one were less affected (41%) than farms from auditor two (75%), diseases may also affect body condition [
9], so the differences found between auditors in the health status of the farms could explain it. Although in other species, such as poultry, pigs and cattle, diseases also affect body condition, the measure is considered reliable for the assessment of prolonged hunger [
6,
8,
10]. It is stated that one feeder per 3–4 meat rabbits should be sufficient [
9]. But under commercial conditions even 8–10 rabbits can be held per feeder without causing any problems [
4]. In the present study, 67% of the farms had seven or less animals per feeder (and 34% up to 4 animals), while a 25% arrived to nine. However, it is not clear any relationship between these numbers and the presence of lean animals in the farms studied, so any refinement of the protocol at this respect need of further studies.
Water plays an important role in the digestive process. Rabbits with an insufficient supply of water will limit their feed intake [
11]. In fact, rabbits have great water requirements and consume approximately twice as much water as feed, but there are no animal-based indicators available to measure prolonged thirst in a farm by visual inspection. For, other species, no animal-based parameters are currently in use for this criterion and, alternatively, the number of drinking points and sometimes also the cleanliness and functioning of drinkers is assessed [
6,
8,
10]. According to Verga et al. (2006) [
12], the minimum provided should be one nipple for every 10 rabbits. In the present study, 75% of the farms provided one drinker for every eight animals and, only in punctual cases such as Farms 11 and 32, with 12 and 13 animals, respectively, had more than 10 animals per drinker. In any case, rabbits must have continuous access to safe drinking water and, for this, it is important that nipples are clean, e.g., no hairs visible, no green rash, and are working perfectly. In this case, all farms assessed obtained the maximum score because of the cleanliness of the drinkers.
In contrast to other protocols [
6,
8,
10], in the present one more weight is given to the absence of prolonged hunger (65%) than to that of prolonged thirst (35%). This was the result of long discussions between experts. Although everybody agreed that thirst is worse than hunger, it was considered that when assessing body condition, you are closer to assessing chronic hunger than you are to thirst when you are assessing the number of animals per drinker, which is just assessing competition for resources. This is an important point to take into account as, in fact, although the name of the criterion is absence of prolonged thirst, the indicator used in this protocol is not really assessing this, as the mental experience of thirst cannot be assessed. This limitation should also be considered for the overall protocol for reflecting the holistic welfare state. Nevertheless, if in the future a good animal-based parameter is validated and feasible for the assessment of thirst in a growing-rabbit farm, the weights of absence of prolonged hunger and absence of prolonged thirst should be reconsidered in the present protocol. For instance, recently Kells et al., (in press) [
13] validated a skin test for the assessment of dehydration in healthy calves of 4–5 days old deprived of feed and water for 24 h. Although the species and the context are not comparable, these types of approaches could be of interest for further improvements of the protocol.
4.2. Good Housing
Inside the good housing principle, three main aspects should be considered: comfort around resting, thermal comfort and ease of movement. Two animal-based measures related to comfort around resting are wet and dirty animals. In the present study, only in three farms were wet animals found, and they represented less than 2% of the animals assessed, so this is not a frequent problem. In the case of dirtiness, something similar is happening, as, when found, only moderate cases were described in up to 1% of the animals affected. The respiratory tract of rabbits is irritated by fine dust in the air [
14], so dust levels should not be too high. Nonetheless, there is no literature showing which dust levels are acceptable or not. For other species, Welfare Quality applies a dust-sheet test, which is a simple procedure indicating the amount of dust in the air [
8]. In this case, 63% of the farms had excellent results with this test, but the rest 37% had some rests of dust in the test. Although, in general, the farms with these conditions had lower scores for nasal and ocular discharge, at the moment, it is suggested to maintain both indicators in an assessment protocol. Finally, light quality is another important aspect to assess when comfort around resting is considered. Lighting should provide uniform illumination and permit the effective observation of rabbits. Continuous lighting (i.e., no dark period in a 24-h cycle) negatively impacts welfare and health, and a natural light–dark pattern enables the rabbit to apply its natural rest–activity rhythm [
11]. In this case, 31% of the farms did not provide enough light to the animals, so this is a critical point to consider.
The second criterion to consider inside good housing, is thermal comfort. The results show that most of the farmers (75%) have registers of temperature, not arriving to the maximum values established in the protocol. In addition, 50% were aware of the importance of considering the risk that hair burning has in achieving a high environmental temperature, locally, at specific areas inside the building. The range of temperatures proposed by the experts for growing-rabbits is a maximum of 28 °C, and although thermal stress in rabbits has been defined at 30 °C [
15,
16], temperatures from 25 °C to 30 °C should be considered already as slightly or moderately stressful for these animals [
17]. In fact, Dalmau et al., 2015 [
17], described an increase in faecal cortisol metabolites and behavioural changes in rabbits subjected to mean temperatures of 27 °C for six hours a day, when compared to animals subjected to 20 °C. Therefore, in the future, the protocol should reconsider this temperature to be closer to the 15–20 °C defined as ideal by EFSA (European Food Safety Authority) (2005) [
9]. Another even better option, is to consider not just the temperature, but also the temperature and humidity by means of an index combining both factors, such as the Temperature–humidity index (THI) discussed by Marai et al. (2002) [
18]. According to these authors, absence of thermal stress is considered when the THI is below 27.8, as moderate heat stress when it ranges from 27.8 to 28.9, as severe heat stress when it ranges from 28.9 to 30.0 and as very severe stress when it is above 30.0. Nevertheless, if nowadays 25% of the farmers assessed did not have registers, to include the need for humidity assessment could dramatically reduce the number of farms with available data for the assessment, so this change should be considered as well in terms of feasibility. On the other hand, although the experts considered that rabbits are well adapted to temperatures below 10 °C, any future improvement of the protocol should consider moving the low value of the range allowed (1–28 °C) closer to the optimal temperatures of 15–20 °C as well [
9].
Ease of movement considers the height of the cage and space allowance. The height of the cage is important because rabbits inspect their environment by showing ”standing up” behaviour [
9]. According to EFSA (2005) [
9], fattening rabbits should be kept in collective cages with a minimum 75–80 cm depth, 35–40 cm width and 38–40 cm height. In the present study, only 41% of the farms provided at least 38 cm height to the animals and a 15% were below 32 cm, so this is a critical point to be considered. Fattening rabbits in medium-sized groups (7–10 rabbits) is common practice in all commercial rabbit-producing countries except Italy and Hungary, where rabbits are usually kept two per cage from weaning until slaughter. The higher slaughter age (80–90 days) necessary to reach the high market weight requested by consumers (2.5–2.6 kg on average) compared to France (2.3–2.4 kg) and Spain or Portugal (1.8–2.2 kg), and the consequent possibility of increased aggressive behaviour and wounds are the two main reasons for this housing system [
19]. There are several studies, with respect to stocking density for meat rabbits, and, in general, a maximum stocking density of 16 rabbits/m
2 or 40 kg/m
2 is seen as acceptable in terms of welfare [
20,
21]. The 500 cm
2 per animal considered as acceptable in the present protocol is exactly 40kg/m
2 for an animal of 2 kg body weight. If the final slaughter weight was 1.8 kg, the space allowance would be 450 cm
2 per animal, and at 2.2 kg would be 550 cm
2 per animal. Therefore, a change of the parameter to kg/m
2 should be considered in the future for the protocol. In any case, according to the expert consultation performed in EFSA (2020) [
22], restriction of movement was the main welfare problem identified in growing-rabbits, so a possible improvement could be to increase the weight of this parameter in the final score of the protocol (now it is 15 points out of 100).
4.3. Good Health
This principle includes the absence of injuries, diseases and pain induced by management. Wounds on the body were seen in 50% of the assessed farms, being the most common problem inside the absence of injuries criterion, while fallen ears and lameness were seen in few animals. Wounds can be caused by inadequate equipment (e.g., sharp parts of cages), or by mutilative or aggressive behaviour of other rabbits [
23]. In the present protocol, only lesions of 2 cm or bigger were considered. This is an important point to take into account, because according to Rauterberg et al. (2019) [
24], rabbits have a high prevalence of lesions of less than 1 cm, especially in the area of the ears. However, these small lesions are difficult to assess in commercial conditions due to the light conditions of some farms and the angle of vision of the animals in the cages. As the present protocol avoids catching the animals in order to reduce the stress associated with human contact, to include these small lesions without catching the rabbits dramatically reduced the inter-observer repeatability of this parameter, so, as decided in the Welfare Quality for piglets [
6], only lesions clearly visible for any auditor (at least 2 cm in size) are considered. This means that although the assessment of wounds showed a good capacity of discrimination between farms, the protocol does not consider all the wounds that a rabbit could have, but rather only a portion. Hairless lesions and wounds on the ears were found in just 10% and 20%, respectively, of the farms assessed by auditor one and in 50% of the farms from auditor two. Although problems of reliability should not be discarded for these two measures, differences in the health status of the farms assessed by the two auditors could explain these differences. In fact, auditor two reported a high incidence of signs compatible with dermatophytosis in the farms assessed, as most of the lesions found were circular with erythema, with a rough and dry surface [
21], compatible with this explanation.
Accordingly, auditor two found more health problems in their farms than auditor one. These differences were focused in the incidence of skin condition problems, such as dermatophytosis, with just 20% of the farms from auditor one and 66% of the farms from auditor two affected, and respiratory problems, with few observed in farms one to 20, but with a high incidence in farms from 21 to 32 (
Table 6). For instance, auditor one registered animals sneezing in just one farm, while for auditor two 75% of the farms were affected with this problem. Skin alterations and respiratory problems are easier to train for auditors than other parameters such as lameness, that needs more exercises to ensure good repeatability between auditors. In rabbits, it is easy to check the eyes and nose for alterations, and sneezing, when focused on a cage, is also very easily assessed, with a high agreement between auditors during trainings. Therefore, although it is not possible to discard an auditor effect, other reasons would better explain the differences found. For instance, it is described that dermatophytosis appears in environments with a high relative humidity, temperature changes, contact with contaminated soil and poor hygiene [
25]. Auditor one reported no problems of cleanliness in their farms, while auditor two reported that ten of the 12 farms assessed had a problem of dirtiness. Again, three reasons could be argued for these differences. (1) A different perception of what is dirty or not by part of the auditors. (2) The fact that most of the farms from auditor one were in a regimen of all in all out. This means that growing-rabbits were moved to specific cages for these animals and after finishing the growing period, cages were sanitated. In the farms from auditor two, growing-rabbits were maintained in the doe’s cage after weaning (mothers were moved) and does at different stages of production were maintained in the same building, impairing sanitation. (3) Another reason could be the fact that farms from auditor one were inside an animal welfare certification scheme and farms from auditor two were not. In any case, another frequent disease problem in rabbit farms associated with poor hygiene is Pasteurellosis [
18,
26]. Signs of Pasteurellosis can include discharge (from the eyes and nose), matted forepaws, sneezing, respiratory distress, abscesses and neck torsion [
18]. Only in one farm from auditor one, but in five from auditor two (42%) was a problem of neck torsion found. This, again, could be an auditor effect, but the most suitable explanation is a higher incidence of health problems (skin condition and respiratory disorders) related with sanitation in farms from auditor two.
The mortality rates in farms from auditor two were far lower than those found from auditor one. As this information is based on registers, no auditor effect is expected, suggesting that most of the differences found between both auditors in the health measures were due to a different management in the farms selected. Farms from auditor one had few health problems the day of the audit but had high percentages of dead animals, and farms from auditor two had more health problems the day of the audit but lower percentages of dead animals. Of course, the ideal situation is those in which the health problems and the mortality and culling rates are both low, as occurred in farm 18. In any case, an assessment protocol should penalize both situations, when a high incidence of animals with health problems is present, and when a high percentage of dead animals (even when culled) are found. Nowadays, inside the absence of disease criterion, dead and culled animals have a weight of 30% of the total score, so if a high percentage of the animals are dead, as is the case in farm 3, there is a risk of having a good overall score for absence of disease due to the lack of problems in the rest of the measures. For this reason, going forward, the protocol should address how to penalize the overall score when this is happening to avoid this compensation between measures. For instance, by minusing 10 points from the overall score when the mortality is higher than 8%, 20 points when it is higher than 15% and 30 points if higher than 20%. However, as done in previous stages, this should be done after expert consultation and, if possible, after having a high number of farms (around 100) assessed with the protocol. On the other hand, only 31% of the farmers provided the percentage of culled animals. As this is critical information, the protocol should consider this as a red line and consider an overall score of 0 points if the farmers have not registered this information.
According to EFSA (2020) [
22], digestive troubles are probably one of the main hazards for the welfare of rabbits. This can range from slight troubles (transitory low feed intake, light diarrhoea) to acute, painful ones (no feed intake, weight loss, acute diarrhoea or ceacal impaction, intestinal inflammation, gastric or intestinal dilatation or swelling, mucus excretion, etc.) [
22]. Although these problems usually affect young rabbits around the period of weaning, later, during the growing period, digestive disorders are the main cause of mortality [
22]. A possible clinical sign of digestive disorders in rabbits, in addition to body condition, is the presence of liquid manure around the anus, or a firm abdomen. To assess this, the animals need to be caught and turned over to check the anus and palpate the abdomen. This was carried out in a previous version of the protocol, but even in farms with known problems of enteropathies and high mortality rates due to this cause, the percentage of animals found with clinical signs during the assessment was very low, as in most of the cases the animal dies very quickly after the appearance of clinical signs. In addition, while does and bucks are used to being handled, growing-rabbits are not, so management during the assessment by itself is more dangerous for the welfare of the animals due to the stress caused by the auditor. As a consequence, while in the protocol developed for does and bucks this parameter was maintained [
27], for growing-rabbits it was considered that the information obtained by checking clinical signs of diarrhoea and caecal impaction (in addition to mortality rates and body condition, already assessed), does not compensate for the stress produced to obtain the data.
Pain induced by management includes how emergency killing is performed. Most of the farmers declared to use just cervical dislocation as a system for emergency killing. However, according to the recommendations of EFSA (2020) [
28], cervical dislocation is not considered a humane killing method and therefore it should be only applied on unconscious animals. In addition, the hazards related to cervical dislocation include “manual restraint” (leading to pain and fear) and “incorrect application” (leading to the absence of unconsciousness, pain, fear and distress) [
28]. For these reasons, in the present protocol, this system is not considered as correct. Certainly, this is one critical point that needs to be improved in the rabbit farms assessed, and there is probably a need for good guidelines to be provided by producer associations and supervised by researchers.
4.4. Appropriate Behaviour
The Appropriate Behaviour principle is assessed by means of the combination of three criteria: social behaviour, other behaviours and the human–animal relationship. Ear damage, scratches and wounds are indicators of negative social behaviour (i.e., aggressiveness), but they have other causes as well (see absence of injuries above), so the observation directly on the animals of this behaviour could be a better indicator. The age of growing-rabbits is an important factor, as rabbits at the end of the growing period may show more aggressive behaviour (due to maturation of the rabbits) [
20,
29,
30]. This can be easily seen in production systems raising heavier animals than those produced in Spain or Portugal, which could explain the lack of negative social behaviour found in the present study. Another explanation is the fact that all growing-rabbits were fed ad libitum, as Dalmau et al., 2015 [
31] described that when fed in a restricted manner, growing-rabbits performed eight to 12-fold more agonistic behaviours than when fed ad libitum. Nevertheless, another explanation could be an insufficient capacity of discrimination of the parameter among the assessed farms, for instance due to insufficient time of observation (just two minutes per cage) or the number of cages assessed (just 10 per farm for this parameter). Therefore, a test of the same parameter in older animals and/or restricted feeding would be necessary to ascertain if the results found in the present study were due to the lack of aggressiveness of the animals in all farms, or a problem with the sensitivity of the parameter. The results showed that none of the animals present in the farms assessed were isolated.
Both in growing and adult animals, stereotypies, that is, abnormal behaviour repeated obsessively without any apparent aim, have often been described [
32,
33]. Stereotypies and abnormal behaviour are indicators of reduced welfare in rabbits. These behaviours can be: head shaking, swaying, wire gnawing, wall pawing and over-grooming [
9,
29,
34,
35]. In the present study, only one farm with presence of these abnormal behaviours, 32, was found (with 13% of the animals affected). In fact, this farm was the worst, in general, for several measures. Therefore, as discussed previously for negative social behaviour, and although the methodology (two minutes of observation per cage) was tested previously in other farms with positive results (presence of abnormal behaviours identified, as what happened in farm 32) and the time dedicated to assess abnormal behaviours is twice than that used in other species, such as pigs, for the same purpose [
6], it cannot be discarded that some adjustment to the methodology might be necessary to increase its sensitivity. In addition, several authors have reported the importance of enrichment for growing-rabbits [
36,
37], so its presence or absence should also be considered in a future version of the protocol, since it is already considered in the protocol for does and bucks [
27].
A good human–animal relationship promotes rabbit welfare. With proper handling, rabbits experience less stress and fear, and the risk of injury to the animals and handlers is greatly reduced [
9]. Management practices have a significant impact on animal health, welfare, and productivity [
38], and the training of farmers has a great impact on this. In the present protocol, an excellent is considered when all the personnel in contact with the animals have specific training in animal welfare, but this only occurred in three farms, with another two farms having at least one of the persons trained. Therefore, this is another key point to improve in rabbit farms in the near future.
4.5. Overall Assessment
Values ranged from 44 to 82 points in the 32 farms, so from the worst to the best farm there were 38 points of difference. The lack of more farms in the excellent category, and the fact that none obtained 100 points, would show that there are possibilities for improving welfare in all farms assessed. The fact that the minimum value obtained was 44 could be due to a lack of sensitivity of the protocol in the lower levels, or to the natural bias in the selection of the farms assessed. In fact, the farms from auditor one were of farmers interested in being certified in animal welfare, with previous knowledge of the protocol (although they were never audited previously with this tool) and advised that they needed to achieve at least 55 points, so it is easy to understand that just one, with a total score of 54 points, remained out of this condition after the assessment. In fact, this could show that the protocol is easy to understand for farmers and they can apply self-assessments before being audited by external assessors. In the case of farms from auditor two, they were not in a certification program, but farmers voluntarily offered for their farms to be visited and audited, so it can be expected that these farms could be positively biased in relation to the global population of farms.
Altogether, with proper training, the protocol seems reliable for the assessment of animal welfare. However, for the criteria social behaviour and other behaviours, further research is needed to ascertain if the methodology and times of observation used are enough to ensure the absence of problems or need to be changed. Auditor one used a mean of 300 animals per farm per clinal measure, while auditor two used a mean of 170 animals per farm. Sample sizes need to find a balance between the reliability and feasibility of a protocol. In the Welfare Quality protocols, this balance was found in using 100 laying hens, 150 broilers and a maximum of 150 pigs for clinical measures [
6,
7], less than the animals used in the present study for rabbits. In terms of reliability, although sampling 10%, 12% and 33% of the population (farm, 1, 13 and 16, respectively) is always more representative than sampling 2% of the population (farms 5, 6, 14 and 18), the results found under both scenarios were similar (
Table 6). Nevertheless, this limitation should be considered for the protocol in farms with more than 10,000 growing-rabbits.
In terms of feasibility, the objective of the protocol was to fix a maximum time of 3 h to take all the data. This is because a second protocol to assess the welfare of does, bucks and kits was developed as well [
27]. As most of the farms have does with kits and growing-rabbits, the objective is to apply both protocols in a journey (maximum of seven hours). Nowadays, with 20 to 50 cages and around 170–300 rabbits assessed for clinical measures, the assessment takes from two to two and a half hours, so there is still a chance to increase the time dedicated to assess social and other behaviours, if needed.