Relationship between quality of facilities, animal-based welfare indicators and measures of reproductive and productive performances on dairy farms in the northwest of Spain

Abstract The aim of the present study was to evaluate the effect of facility comfort level on animal-based welfare indicators as well as on productive and reproductive parameters of 187 dairy farms in the northwest of Spain. Fifty-eight facility-based parameters, which included both housing facilities and management parameters, were evaluated in each farm and used to classify the 187 farms in 5 groups (from 1: top farms to 5: worst farms). Animal-based welfare indicators and reproductive and productive parameters were compared among farm groups. Lameness score significantly differed (p < .005) in top farms (67% of cows having normal gait) versus groups 4 (55%) and 5 (53%). The proportion of cows with dirty lower legs and udder was lower (p < .05) in top farms (80% and 31%, respectively) than in groups 4 (92% and 51%) and 5 (95% and 56%). Other animal-based indicators were not different among farm groups. Heat detection rate was higher (p < .001) in top farms (56.5%) versus groups 4 (51%) and 5 (47%), and milk production/cow/day was also significantly higher (p < .001) in top farms (34.4 L) versus groups 3 (31.2 L), 4 (30.5 L) and 5 (27.4 L). Other reproductive or productive parameters were not different among farm groups. Facility-based measures were compared only between the top and worst farms. In top farms, most facility-based parameters met the welfare-related objectives, whereas in the worst farms only a few facility-based parameters met the welfare objectives. Highlights Fifty-eight facility-based parameters, including housing measures and some management practices, were evaluated and scored in 187 dairy farms. Top-performing farms (i.e. with the best scores) had lower incidence of lameness and of cows with dirty legs and udder than poor-performing farms. Top-performing farms had higher rates of heat detection and higher milk production/cow/day than poor-performing farms.


Introduction
Animal welfare has become a topic of great interest to EU citizens, and legislation concerning animal welfare in modern livestock production is frequently revised. The livestock industries are not only sensitive to the welfare of farm animals, in part because poor welfare conditions are often associated with reduced productive performance of animals, but also because they need to project a welfare-friendly image of their products to be competitive (Hemsworth and Coleman 2011).
In Europe, public opinion perceives that, compared with laying hens or pigs, dairy cows are reared in fairly good welfare conditions (European Commission 2005). Because the public opinion may be influential in determining legislative decisions on animal welfare standards, this may be the reason for the lack of specific regulations concerning animal welfare in dairy farms. Among all factors influencing animal welfare, perhaps the most evident is the quality and comfort ability of facilities, which may widely differ among the dairy farms in a region or a country.
Inadequate animal welfare conditions are known to cause chronic stress and numerous health issues in dairy farms (Abeni and Bertoni 2009), but objective and systematic assessment of animal welfare aspects, including comfort ability of facilities, is not routinely done by veterinarians and therefore, insufficient or poor cow comfort may be overlooked.
Assessment of animal welfare conditions provides an opportunity to address ethical issues and also to improve the productive performance of dairy farms (Abeni and Bertoni 2009), as cow discomfort is known to hinder full development of the productive potential. Various different studies have demonstrated that some facility-related aspects, as well as management practices, may influence animal health, longevity and reproductive and productive performance (Cook and Reinemann 2007;Erina et al. 2008). Bach et al. (2008) found that milk production differed considerably among herds of similar genetic background and fed with the same ration. In the cited study, non-dietary factors (i.e. management practices) accounted for 56% of the variation in milk production not attributable to diet.
The dairy industry in developed countries is characterised by high production costs and a final product of low price, and farmers therefore operate with narrow profit margins. In this context, despite that a positive relationship between cow comfort and productivity has long been known, farmers are often reluctant to renovate existing facilities or to build new ones with larger stall dimensions (Grant and Miner 2015), unless they are convinced that the cost is financially worthwhile. As milk production is affected by many different variables, estimating the impact that improved facilities may have on farm productivity is a difficult task, and it may be the reason not to undertake reforms or improvements since the benefit that can be reported is not known.
In the present study, the level of cow comfort provided by facilities and some management practices was objectively assessed and quantified in a representative number of dairy farms located in the province of Lugo, which is the most important dairy province of Spain. Then, animal-based indicators of welfare and productive and reproductive performance were compared among farms with different degrees of cow comfort. Feeding practices and genetic background were uncontrolled factors, but they were similar for all the farms in the region. The global aim of the study was to determine if herd health and production differed among farms with better or worse facilities and/ or management, and the magnitude of the difference. More concretely, the specific objectives of the study were: 1. To assess and score 58 facility-based parameters, which included both housing measurements and evaluation of some management practices, in 187 dairy farms located in the province of Lugo. 2. To determine if the quality of facilities was related with: (i) animal-based measures of welfare (body condition score, lameness score, hock lesions and hygiene), (ii) reproductive performance (postpartum intervals, fertility, heat detection rate and cull rate), and (iii) productive parameters (somatic cell count, milk production and percentages of fat and protein in milk).

Farms
The study involved 187 dairy farms in the province of Lugo (Galicia, northwest of Spain). A substantial proportion (19%) of the 844,000 dairy cows registered in Spain in 2015 was from farms in the province of Lugo. Farms were recruited through the veterinary surgeons responsible for reproductive control on dairy farms in the region. Only free stall farms were included in the study, as tie stall farms usually have old facilities and are becoming increasingly less frequent in the study area. In most farms (160) cows were all year round indoors without access to pasture.
All farms had Holstein cows; no other breeds were present in any farm. The average number of lactating cows per farm was 55 (range: 10-240), and a total of 9228 lactating cows were evaluated. Average parity per farm was 2.5 ± 0.5 (range: 1.7-4.9).
All farms milked twice a day and all had conventional milking parlour. Feeding management was not evaluated in each farm, but all farms in the region tended to use the same type of raw materials to formulate rations, and all used mixer trucks. In general, rations were based on corn and grass silages supplemented with concentrates, which were adjusted depending on production.
Farms were visited once, by the same evaluator, for collection of the data specified below.

Reproductive parameters
Reproductive data from the previous 12 months were obtained from the reproduction records kept by the veterinary surgeons attending each farm. The following parameters were considered: interval between calving and first artificial insemination (AI), conception rate at the first AI, interval between calving and conception, heat detection rate, average fertility and cull rate.

Productive parameters
The following milk production data were obtained from the Galician Dairy Control Programme records provided by veterinary surgeons: somatic cell count (SCC; mean value for the previous 12 months), milk production per cow and day (L/Cow/day) and percentages of fat and protein (mean values for the previous 12 months).

Facility-based measures
The facility-based measures considered are specified in Table 1. Facility-based parameters that complied with the welfare objective were scored 0 points, whereas those that did not comply with the welfare goals were scored 1 point. Distance measurements were done using a laser metre or, for small distances, a roll metric tape. Data collection in the different farm areas was done following always the same order.

Animal-based measures
Eighty to one hundred per cent of cows on each farm were evaluated for the parameters specified in Table 2. Animal-based measures were collected around the time of the first milking. At the start of the evaluation, all cows were restrained in headlocks. Each cow was first scored for body condition (BCS), hock injuries and hygiene, and then released and scored for lameness from an average distance of 3 metres. Each of these measurements was recorded once for each herd and all scores were done by the same evaluator.
Body condition score (BCS) was based on a scale of 1 to 5 with 0.25 point intervals (Ferguson et al. 1994). For each cow, BCS and days in milk (DIM) data were compared against the objectives proposed by PennState Extension (2017). The proportions of cows with normal, low or high BCS according to the objectives for DIM were determined for each farm.
Hygiene score was based on a scale of 1 to 4, as proposed by Cook (2002), where 1 represents the highest and 4 the lowest level of cleanliness. Cows were scored separately for leg, udder and flank hygiene. The proportion of cows with dirty legs, udder or flanks (i.e. scored 3 or 4 for any of the 3 measures) was calculated separately for each farm.
Hock injuries were recorded as present or absent, and different degrees of injury were not considered. Tarsal joints of each cow within the herd were evaluated. The percentage of cows with lesions (scratch, swelling, abrasion or trauma) in one or the two limbs was used for the analysis. Lameness was scored as described by Sprecher et al. (1997) on a scale of 1 to 5, where healthy cows were awarded a score of 1 and severely lame cows were awarded a score of 5.

Statistical analysis
Regarding facility-based measures, farms were scored for each parameter specified in Table 1 and were awarded a score of 0 for compliance with the objective and a score of 1 for non-compliance with the objective. The sum of all scores awarded to each farm corresponded to the number of parameters that did not comply with the objective; and therefore, low scores represent high quality facilities. The total score awarded to farms was used to calculate quintiles and thus divide the farms into five groups (Table 3). Farms were divided in quintiles to try to satisfy two conditions: (1) to have a sufficient number of farms per group, and (2) that farms in group 1 (considered the top-performing farms) and those in group 5 (considered the poorest-performing farms) were as far each other as possible, so that differences could be detected. Animal-based measures and productive and reproductive parameters of farms in group 1 were compared with those of the other groups by means of ANOVA and Dunnet's t test. A Fisher's exact test was used to compare facility-based measures between the top-performing group and poorest-performing group of farms. ANOVA was also used to compare animalbased measures and productive and reproductive parameters between farms with or without outdoor run.

Results
For facility-based measures, the lowest score obtained was 6 out of 58 points and the highest score, 39 out of 58 points. The ranges of scores awarded to the five groups of farms, as well as the mean farm size, parity and percentage of farms with outside access per group, are shown in Table 3.
Comparison of animal-based measures and of measures of productivity and reproductive performance between the top-performing farms and the other groups of farms revealed significant differences in the following items: lameness score, leg and udder hygiene, heat detection rate and daily milk production per cow (Table 4). The proportion of cows with normal gait (scored 1) was higher in the top-performing group than in the other groups (67% relative to 55-54% in groups 4 and 5), as was the proportion of cows with dirty udders (31% relative to 49-56% in groups 3-5). The rate of heat detection was also higher in the top-performing group (56.5% relative to 51-47% in groups 4-5). Finally, the daily milk production per cow was higher in the top-performing group (34 relative to 31-27 L in groups 3-5).
When animal-based measures, as well as productivity and reproductive performance, were compared between farms with outdoor run (n ¼ 27) and onlyindoor farms (n ¼ 160), significant differences were found in some parameters: the proportion of cows scored 2 for lameness was higher (p < .05) in farms with outdoor run (31.7 ± 17.4% vs. 26 ± 12.8%); percentages of cows with normal gait or with higher lameness scores did not differ.
Regarding facility-based measures, the main differences between the top-performing (group 1) and poorestperforming (group 5) farms (Table 5) were associated with the type and maintenance of bedding, type of floor, dimensions and cleaning of alleys, care of cows, accessibility to water and food, and the farm environment. However, stall dimensions were not different in these two groups of farms, being inadequate in both.

Discussion
The results of this study indicated a difference of 33 points, out of a possible 58, between the top-performing farm (scored 6 points) and the poorest-performing farm (scored 39 points). However, the distance between the two groups, top-performing (score: 6-20) and poorest-performing farms (score: 30-39) was shorter, which made it difficult to identify differences between the middle groups. This was the reason why facility-based parameters were only compared between the top-performing and the poorest-performing farms.

Hygiene
The proportions of cows with hygiene score >2 for legs, udder and flanks in all groups of farms included in the present study (Table 4) were higher than reported by Cook and Reinemann (2007) both for the 25% top-performing farms (47%, 11%, and 8% for legs, udder and flanks, respectively) and for the average-performing farms (59%, 19%, and 15%, for legs, udder and flanks, respectively). Although hygiene scores were far from optimal in any of the groups of farms in the present study, there were significant differences between top-performing farms and those in groups 3, 4 and 5. This was probably related to the better quality of bedding materials and bedding maintenance, and with the better cleaning practices used on top-performing farms. Clean resting areas and clean alleys contribute to cow cleanliness (Cook and Reinemann 2007) and lower incidence of hoof disease because humidity and dirtiness cause soft hoofs that are more likely to become damaged and infected (M€ ulling et al. 2006;Lagger 2007). A high incidence of lameness will in turn contribute to poor bedding and cow hygiene, as lame cows tend to lie down more time than healthy ones (Ito et al. 2010).
Poor cow hygiene is known to be associated with a high incidence of mammary infections as manure and bedding materials are the main sources of E. coli and environmental Streptococcus (S. uberis, S. dysgalactiae, Enterococcus spp; Cook and Reinemann 2007). However, there were no differences in somatic cell counts between the groups of farms included in the present study, probably because there were high percentages of dirty cows in all farm groups. In farms with outdoor run, there were more cows with dirty legs and udder, however, somatic cell counts were lower than in indoor farms. Although there is not an obvious explanation for this, dirtiness of cows with outside access might be due to soil and mud more than to manure, that is considered the main contaminating source.

Lameness scoring
According to Juaristi et al. (2004), dairy farmers should aim for lameness scores to be distributed as follows: >75% of cows with normal gait, <15% with slight lameness, <9% of cows moderately lame, <0.5% of cows scored lame and <0.5% severely lame. None of the five groups of farms considered in this study complied with this objective (Table 4), indicating that lameness may be somewhat neglected by farmers in the region. Nonetheless, the proportion of cows scoring 1, 2 and 3 differed significantly between the topperforming group and groups 4 and 5. Such Table 5. Proportions (expressed as percentages) of the top-performing and poorest-performing farms that met the objectives established for the facility-based measures evaluated. Parameters and numbers shown in bold type indicate significant differences between the two farm groups.
differences may be related to the use of suitable material for bedding and floors, rigorous cleaning practices, adequate dimensions of alleys and preventive measures against hoof disease, all which were observed on most top-performing farms (Table 5). For optimal hoof health, cows should be lying down for around 12 h daily to minimise hoof stress and the incidence of lameness (M€ ulling et al. 2006;Erina et al. 2008). Adequate stall dimensions (M€ ulling et al. 2006;Espejo and Endres 2007) and comfortable, clean and dry bedding (M€ ulling et al. 2006;Fregonesi, Veira, et al. 2007) are essential factors in this regard. Ninety percent of the farms included in this study showed some deficiencies in relation to stall dimensions, although the beds were cleaner and bedding materials of higher quality in the top-performing group of farms than in group 5. Moreover, although the difference was not significant, 25% of the top-performing farms and 40% of the farms in group 5 provided less than 1 stall per cow. Higher than 100% stocking density is known to be associated to reduced resting time and increased idle standing in alleys (Fregonesi, Tucker, et al. 2007;Hill et al. 2009;Krawczel et al. 2010), which will favour lameness. Lame cows and heifers are the most affected by high stocking density as they are not competitive.
Another important aspect related to hoof disease is the type of floor and the cleanliness and dimensions of alleys (M€ ulling et al. 2006). In free stall farms, cows walk long distances (up to 1 Km a day) to gain access to feeding tables, water troughs and the milking parlour (Fjeldaas et al. 2011). The type of floor, as well as the state of conservation and hygiene will determine the comfort provided, the abrasive capacity, level of humidity and degree of contamination. Seventy-four percent of the farms included in the present study had textured (striped or grooved) concrete floors, and only a few of the top-performing farms had rubber floors, which have been reported to be best for hoof health (Telezhenko et al. 2007), although it has also been suggested (Kremer et al. 2007) that the scarce abrasive effect of rubber may favour some claw diseases if claws are not cut frequently. Concrete floors tend to be more slippery (M€ ulling et al. 2006) unless textured, in which case they provide a rough, abrasive surface. However, the abrasive capacity of concrete floors should not be so high as to cause excessive wear of hoofs nor as low as to favour abnormal growth of claws and slipping (M€ ulling et al. 2006). Slatted floors tend to be less abrasive and slippier than compact floors (Telezhenko and Bergsten 2005).
The slats can also become misaligned and difficult for cows to walk on.
Adequate cleaning of alleys is essential for hoof hygiene and to prevent cows slipping. Alley dimensions should be such that two cows walking in the opposite direction should not have to touch each other. Narrow or blind alleys, when existing dominant cows, may favour stressful situations and sudden avoidance movements causing hoof injuries (M€ ulling et al. 2006).
Most of the top-performing farms in the present study applied measures to prevent hoof diseases. These included regular claw trimming and installation of footbaths, which may partly account for group 1 having lower (i.e. better) lameness scores (Lagger 2007) than groups 4 and 5, in which such preventive measures were less frequently applied.
Farms with outdoor run were more frequent in groups 4 and 5 than in the other farm groups (Table  3). Having an external area for cows to walk around helps to maintain hoof health, as walking on meadows or yards would prevent excessive wear or growth of claws caused by concrete floors, and the prolonged exposure of hoofs to humidity and manure typical of indoor farms. However, in the present study, in some of the farms with outdoor access, cows were allowed to go outside for short periods of time daily or only in good weather, or sometimes yards were not properly drained. So, these handicaps added to poor facilities and/or poor management would account for outdoor run-farms not having better lameness scores than indoor farms.
Lameness is known to decrease dry matter intake and body condition (Bach et al. 2007) and to reduce reproductive performance and milk production (Huxley 2013), leading to important economic losses. According to Liang et al. (2017), the mean cost associated with each lame cow on a dairy farm is around 160 e for primiparous and 300 e for multiparous cows, including production and fertility losses and treatment costs.

Reproduction and milk production
The heat detection rate and milk production were significantly higher in the top-performing group of farms than in groups 4 and 5. This finding may be associated with the higher incidence of lameness in groups 4 and 5. Lame cows feel pain and may move less than healthy cows, thus affecting dry matter intake and reducing the chance that heat symptoms will be detected. Lameness may therefore have a significant impact on heat detection and milk production (Cha et al. 2010).
The difference in milk production between the topperforming and the poorest-performing farms was 7 L/ cow present on farm/day. Differences in milk production may reflect differences in genetic and in feeding management. It is likely that larger farms with modern facilities and good management practices had also animals of higher genetic merit, but this could be highly variable among farms within groups, so in this study we cannot speculate about the influence of genetic on milk production. Although rations offered by the different farms were not evaluated, they probably did not differ too much since practically all farms recruited worked with a nutritionist. However, the feeding environment provided by top-performing farms was better than that in farms of group 5. Clean and dry feeding tables, continuous feed availability, and adequate space of manger and water trough per cow will favour optimal dry matter intake . Feeding tables that are unsuitable or of difficult access may lead to insufficient or imbalanced intake of food and water by cows, being the lame cows and heifers the most harmed as they cannot be competitive. Although not evaluated in this study, it is very likely that management practices known to increase dry matter intake and milk production, such as feeding for feed refusals or feed push up (Bach et al. 2008), were more frequent in the top-performing group of farms than in the poorest-performing group.
Other facility-related factors such as the level of comfort of stalls may also affect milk production, as cows may rest for shorter periods in uncomfortable stalls (Espejo and Endres 2007;Fregonesi, Veira, et al. 2007). During rest periods, blood flow to the mammary gland increases by 21.6%, which enhances milk production (Erina et al. 2008). As already mentioned, stocking density >100% is associated to shortened resting times and reduced milk production (Hill et al. 2009).
Inadequate ventilation systems may be ineffective in counteracting heat stress in summer, another factor known to affect milk production (Bernabucci et al. 2014;De Rensis et al. 2015). Heat stress is known to reduce the feeding, resting and rumination behaviours of cows, and thus, feeding intake and milk production (Tapki and Sahin 2006). Facilities designed to allow good ventilation and avoid humidity and condensation were present in most of the top-performing farms, but they were less frequent in the poorest-performing farms.

Conclusions
The present study showed that comfort and quality of facilities differed significantly among the dairy farms recruited for the study, but some defects were common to all farm groups. Most farms, either in the topperforming or in the other groups, had stalls with insufficient dimensions, and an important proportion of farms (25-40%) provided less than one stall per cow. Proportions of lame cows were high in all groups of farms, varying between 33% and 46%. Inadequate facilities and poor facility maintenance were associated with significantly higher incidence of lameness and with significantly poorer hygiene in legs and udder, with lower heat detection rate and lower milk production. The observed differences in productivity among farm groups, although not entirely due to facilityrelated aspects, should serve as incentive for farmers to undertake reforms and/or changes in management practices so that cow comfort could be significantly improved. On the other hand, lameness and cow hygiene measures can be used as rapid indicators of unsuitable facilities.