HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Veissier, I. Articles by Delval, E. Search for Related Content PubMed PubMed Citation Articles by Veissier, I. Articles by Delval, E.

Cubicle housing systems for cattle: Comfort of dairy cows depends on cubicle adjustment¹

I. Veissier^*,2, J. Capdeville and E. Delval^*

^* INRA, Centre de Clermont-Ferrand–Theix, URH-ACS, 63122 Saint Genès Champanelle, France; and and Institut de l’Elevage, Antenne de Toulouse–Castanet Tolosan, BP 18, 31 321 Castanet Tolosan cedex, France

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion Implications Literature Cited

 
Housing is important for the welfare of cows. Although recommendations have been proposed, abnormal movements and injury problems are still observed in cubicle houses. We conducted a survey on 70 French dairy farms that used cubicles. We examined the design of the cubicles, and the behavior, injuries, and cleanliness of the cows. Most of the cubicles did not comply with the recommendations, often being too narrow and/or too short. Difficulties in lying behavior and injuries were more common when the neck rail was high. No improvement was noted when cubicles of a recent design were used ("U.S." cubicles), apparently because these cubicles were most often cantilevered on a double head rail rather than fixed on freestanding posts. An experiment was conducted, making similar measurements, on 84 cows to compare two configurations for U.S. cubicles (cantilevered on a double head rail as observed in the survey with a high and rear neck rail vs. fixed on freestanding posts as recommended) and another recent cubicle type (Euroconfort, cantilevered on head rails, but with a large space between the rails and fixed as recommended), with and without a brisket board. In U.S. cubicles on rails, cows spent more time lying and less time fully standing inside the cubicles than in the other cubicles (percentage of time: lying, 53.9 vs. 51.5; fully standing, 7.3 vs. 8.5); in Euroconfort cubicles, they hit bars more often when getting up than in U.S. Cubicles (percentage of observations: 42.4 vs. 26.4. Without a brisket board, cows lay down more often in a fore position in U.S. cubicles than in Euroconfort ones. Somatic cell counts increased with time in U.S. cubicles on rails and decreased in the other cubicles. It is suggested that the position of the neck rail in U.S. cubicles cantilevered on rails did not leave enough space for the cow to stand inside the cubicle, thereby encouraging the cow to lie down. This could in turn favor udder contamination and/or inflammation. It is concluded that the positioning of the neck rail is of prime importance, that U.S. cubicles should be used with a brisket board and with correct positioning of the neck rail (even when a head rail is used), and that leaving a large space between head rails does not offer an adequate remedial solution for keeping a free head space in front of the cubicle.

Key Words: Cubicles • Dairy Cows • Injuries • Lying Behavior • Milk • Welfare

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion Implications Literature Cited

 
Housing is recognized as an important welfare factor for farm animals. In its five freedoms, the Farm Animal Welfare Council (1992) stipulates that animals must be kept free of physical discomfort by appropriate shelter and resting areas (Freedom 2); however, current housing systems are not always satisfactory. When cattle are accommodated indoors (as in winter in many countries and even all around the year in some cases) and when straw is not easily available, they are often housed in sheds with cubicles or tethered. Tethering restricts animal movements and behavior such as social behavior and locomotion. This decreases Freedom 3 (i.e., the possibility to express normal behavior). Restriction of movement and of social contacts is likely to be experienced as negative by animals, as evidenced by their physiological responses to stress when tethered for the first time (Ladewig and Smidt, 1989) and by the enhanced locomotion and social behavior observed in restricted animals released in large enclosures and/or with congeners (Dellmeier et al., 1985). Cubicle housing systems offer the animals more possibility of movement. However, they can also have negative effects in some circumstances, such as making it more difficult for animals to get up and lie down (Ferrouillet et al., 2001) or inducing injuries, and even making animals prone to lameness (Colam-Ainsworth et al., 1989; Webster, 2002). Recommendations have been produced for the design and the adjustment of cubicles (CIGR, 1994; Juan, 2001), but problems still arise in practice, from cows presenting mild injuries to cows lying down in exercise areas rather than cubicles.

The current study aimed first to analyze welfare problems among cows housed in cubicle systems in relation to the nonobservance of recommendations, and second to propose improvements for the use of cubicles based on observations in commercial farms and on an experiment in controlled conditions.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion Implications Literature Cited

 
A survey was conducted on a sample of French commercial dairy farms in 1999. An experiment comparing cubicle settings was subsequently conducted from 2000 to 2003.

Survey
Subjects.
Seventy farms where dairy cows were housed in cubicles at least in winter were selected. These farms were located in 10 French regions (West, North, East, and Central France) and their cattle sheds had been built between 1970 and 1998. We expected the variations in location and age to offer variations in cubicle size and/or design.

Observations on Cubicles.
On each farm, we recorded the following information on the cubicles in use: 1) number of cubicles per animal; 2) curb height; 3) width and length (i.e., distance from the curb to the first obstacle in the cubicle (e.g., a front rail, a wall); 4) height of the neck rail and its position (distance between the curb and the projection of the neck rail on the floor); 5) presence or absence of a brisket board; 6) presence or absence of a head rail; 7) width of the passage in front of the cubicles classified as no passage (cubicle partitions ending on a wall or face to face with no space in between), narrow passage (less than 125 cm), or wide passage (more than 125 cm); 8) type of partitions between cubicles classified as standard (two horizontal lines made of metal, wood, strap, or rope, with both ends fixed on posts), loop with a single fixing (at its front end), loop with a double fixing (at its front end), mushroom (i.e., a loop with a double fixing at its center), U.S., and other types (Figure 1); 9) shape of the cubicle base (hollow; flat); 10) flooring classified as slatted floor, clay, asphalt, concrete, mat, or mattress; 11) bedding classified as no bedding material, manure, sawdust, or other; and 12) frequency of cleaning of the rear part of cubicles (regular or irregular).

View larger version (26K): [in this window] [in a new window]   Figure 1. Types of cubicle partitions. Standard, loop with single or double fixing, mushroom, and U.S. partitions were examined in the survey. The U.S. and Euroconfort partitions were used in the experiment.

Behavior.
Getting-up movements were observed preferentially before milking. For every getting-up movement, we noted the number of intentions (the cow extends its head forward), whether the movement was interrupted (the cow lies down again after lifting its hindquarters) or abnormal (the cow lifts its forequarters before its hindquarters). Lying movements were observed preferentially after milking. For any lying-down movement, we noted the number of intentions (the cow sniffs the ground and swings its head; CIGR, 1994) and whether the lying movement was interrupted (the cow stands back up after bending one foreleg). When the animals were lying down, we noted how many were in cubicles, and their diagonal position was recorded in four classes: shoulder and hip on the same side in contact with or directly below a partition, shoulder in contact with a partition and hip on the same side inserted below a partition, shoulder and opposite hip in contact with or directly below partitions, and shoulder in contact with a partition and opposite hip inserted below the opposite partition. We also noted the longitudinal position of animals lying or standing in cubicles (the animal was considered to be in a rear position when the distance between its rearmost part [its rump when lying or a hoof when standing] and the cubicle curb was less than 5 cm, and in a fore position when it was more than 30 cm) and whether they were in reversed position (forequarters at the rear of the cubicle). We also recorded any animal sitting in a dog-sit posture (hindquarters on the floor and forelegs stretched out (Albright and Arave, 1997).

Injuries and Health.
After the above items had been recorded, external injuries (skin injuries and inflammation of hocks) and cleanliness of animals were recorded by close examination (from approximately 1 m) of standing animals. If less than two-thirds of the animals were standing, the lying animals were gently encouraged to stand up. Skin injuries were classified with reference to their location (back, rump, thighs, hocks, tail, udder [except teats], teats, thorax, shoulders, shoulder tips, dewlap, knees, and neck) and severity. Except for hock injuries, a lesion was classified as mild when some hair was absent and/or the skin was abraded, but without edema or open lesion, and as severe in cases of edema or an open lesion. The severity of a hock inflammation was classified as mild when the edema was less than 10 cm thick, medium when the edema was more than 10 cm thick, and severe in the case of an open lesion and/or suppuration. In addition, farm records were used to assess the occurrence of lameness and clinical mastitis.

Cleanliness.
The cleanliness of the animals was recorded according to the method proposed by Faye and Barnouin (1985). Briefly, a cow’s rear was divided into five areas: the anogenital area (the triangle between the two pin bones and the top of the udder), the back of the udder (seen from behind), the bottom part of the hind legs (from the hock to the dewclaws), the udder sides and the belly (region between the udder and the navel, seen from the sides), and the thighs. The cleanliness of each area was scored on a scale, with 0 = no dirt, 0.5 = some small dirty parts, 1 = large dirty parts covering less than half the area, 1.5 = large dirty parts covering more than half the area, 2 = area completely covered with dirt. When an area had two sides, only the dirtier side was scored.

Size.
Height at withers (H), width at last rib (W), and diagonal body length (L) were measured on the 20% tallest cows in the herd.

Three observers participated in the survey. They had been trained together by an instructor before they started visiting farms. During training, they received a guidebook explaining precisely how to conduct the visit, record information on cubicles, and observe animals (Patouillard, 1999; a short description is given in Capdeville and Veissier, 2001). They were also given oral explanations, and they visited three farms. Consistency between observers was checked from data recorded during these first three visits.

Statistical Analyses.
The farm was taken as the observation unit. Preliminary calculations were done to assess the observance of recommendations for cubicles. Owing to the relatively high number of data we recorded to describe the cows’ state (47) compared with the number of farms observed (between 59 and 69 due to missing data for some measures), we decided to define summary data for the animals’ state and to run multiple ANOVA to analyze the effect of the causal factors (cubicle design) on these summary data to avoid misleading interpretations when variables were linked to each other. The summary animal variables were derived from principal component analyses (PCA) on behavior, injuries, and cleanliness, respectively. The SAS software (SAS Inst., Inc., Cary, NC) was used for all the statistical analyses.

Preliminary Calculations on Items Observed on Cubicles.
We used the recommendations of the CIGR (1994) amended by Capdeville (Juan, 2001) to set optimal cubicle size: curb height = 15 to 20 cm; cubicle width = (0.86 xH) + [0.7 x (W –68)]; cubicle resting length (distance between the curb and the back of the neck rail) = (1.01 xL) + 10 cm; length of the head space = (0.32 xH) for space sharing cubicles, and (0.56 xH) for nonspace sharing; cubicle total length = sum of the two previous values; place of the brisket board (distance from the curb and the back of the brisket board) = (0.96 xL) + 15 cm; and height of the neck rail = (0.75 xH). We then considered that the observed value for a cubicle dimension was lower than recommended when it was at least 10% smaller than the optimal value, and that it was higher than recommended when it was at least 10% greater than the optimal value (except for the length of the head space, for which there is no recommended upper limit).

Principal Component Analyses on Items Observed on Animals.
The proportion of each item observed in animals was calculated by dividing the frequency of that item by the number of animals observed on a given farm and for a given item. For instance, the number of interruptions of lying movements was divided by the number of lying movements observed, the number of lying intentions was divided by the number of successful lying movements, the number of animals with a particular injury was divided by the number of animals that were examined for injuries, and so on. We assumed that not all the animal items were independent of each other. For instance, a dirty shed can make animals dirty in different body areas, and two items can express different degrees of severity of a similar cause (e.g., a moderately narrow cubicle can result in many lying intentions and a very narrow cubicle can result in many intentions and interruptions of lying), etc. To avoid using redundant information, we carried out PCA on blocks of items. A first PCA was run on behavior, a second on injuries, and a third on cleanliness. Only the first axes of the PCA were taken into account because their eigen value was much higher than that of the subsequent axes.

The first axis of the PCA on behavior had an eigenvalue of 3.64. The coordinates of the behavioral items on this axis were getting-up intentions, 0.40; getting-up interruptions, 0.21 for a normal movement and 0.33 for an abnormal one; abnormal getting up, 0.20; lying intentions, 0.28; lying interruptions, 0.30; lying outside cubicles, 0.00; diagonal position in cubicles, 0.25 for animals with a shoulder in contact with a partition and the hip on the same side inserted below a partition and 0.24 for animals with a shoulder in contact with a partition and the opposite hip inserted below the opposite partition; animals in a fore position, 0.33 for animals lying and 0.31 for animal standing; presence of dog-sit postures, 0.20. Thus, the high proportions of rear positions, of difficulties in getting up or lying down, and of cows lying partly in adjacent cubicles result in a high positive value on this axis.

The first axis of the PCA on injuries had an eigenvalue of 7.70. The coordinates of animal injuries on it were mild lesions, 0.23 for lesions on the rump and 0.27 for lesions on the back, thorax, or shoulder tips; severe lesions, 0.22, 0.26, 0.29, 0.27, 0.28, and 0.28, respectively, for lesions on the back, rump, thighs, dewlap, knees or neck; and hock inflammation, 0.19 and 0.29, respectively, for medium and severe lesions. Thus, a high occurrence of lesions (regardless of their location or severity) on a farm resulted in a high positive value on this axis.

The first axis of the PCA on cleanliness had an eigenvalue of 3.67. The coordinates of cleanliness items on it were 0.27, 0.47, 0.48, 0.49, and 0.49 for the cleanliness score of the anogenital area, the back of the udder, the bottom part of the hind legs, the udder sides and the belly, and the thighs, respectively. Thus, the dirtier the cows on a farm were, the higher the value on this axis.

The first three axes of the PCA were further used as summary variables for behavior, injuries, and cleanliness. We calculated parametric correlations between them.

Analysis of Causal Factors.
We looked for cubicle aspects that could explain variations in the summary variables obtained for each farm. Because factors can be confounded in a survey, we ran multiple ANOVA to reduce misattributing an effect to a given cubicle aspect. The following model was used:

where Y_stuvvvxyz represents a summary variable; a_s, b_s, c_s, d_s, e_s, f_s, g_t, h_u, i_s, j_v, k_w, l_x, m_y, and n_z represent the effects of the number of cubicles per animal, of their curb height, their width, their length, the height of the neck rail, its position, the presence of a brisket board, the presence of a head rail, the width of the passage in front of the cubicles, the type of partitions between cubicles, the shape of the cubicle base, the flooring, the bedding, and the frequency of cleaning, respectively. Owing to the large number of factors and covariates, no interactions were included in the model. The results section will focus on significant effects (P < 0.05) and tendencies (0.05 < P < 0.10).

Experiment
The results of the survey suggest that the adjustment of the cubicles, specifically of the height of the neck rail, is of prime importance for the position of the cows in the cubicles, and in turn for their getting-up and lying-down movements, and external injuries. They also suggest that cubicles of a new shape, such as "U.S." cubicles, provide no extra benefit, possibly because they are cantilevered on double rails rather than fixed on individual posts. The double front rail may oblige farmers to fix the neck rail in a high rear position (see below in Discussion).

The aim of the experiment was to compare cubicles with U.S. partitions fixed on freestanding vertical posts (as recommended) vs. cantilevered on two front rails. In the first case, the cows can place their head forward when lying, whereas in the second case, the front rails limit this position. A third treatment consisted of cubicles with partitions of the recently released Euroconfort type. This partition looks like a U.S. partition cantilevered on front rails but with more space between rails due to greater distance between the horizontal bars of the partition (Figure 1). In addition, a prototype brisket board was designed to prevent injuries to the cows’ forelegs, while efficiently preventing them from lying down in a fore position.

Animals and Housing.
The experiment was run in three consecutive years at the INRA experimental farm at Les Monts-Dore, France. The experimental shed contained three pens placed off a common alley. The pens were 6.45 m wide and between 11 and 13 m long. Each pen contained a feeding rack (along the alley), an exercise area (scraped five times a day), and one cubicle per cow (nine cubicles in the pens on the sides and ten in the middle pen). The cubicle base had a 7% slope, was covered with a mattress (INRAmat, CAT Rochefort-Montagne, France), and was spread with 500 g of wood shavings per day. The rear part of the cubicles was cleared of feces twice a day. Three types of cubicle partitions were used: U.S. partitions fixed on freestanding vertical posts (U.S. on posts); U.S. partitions cantilevered on two front rails (U.S. on rails); and Euroconfort partitions cantilevered on rails (Euroconfort, Jourdain, Escrennes, France). In 1999, before the experiment started, we measured the H, W, and L of each cow on the farm. Mean H, W, and L calculated over the 20% largest cows were 143 and 174 cm, respectively. According to the recommendations of the CIGR (1994) amended by Capdeville (Juan, 2001; see above), the optimal size of cubicles was as follows: cubicle width = 123.3 cm; cubicle resting length = 185.7 cm; and height of the neck rail = 107.3 cm

In all pens, the cubicle width was 124 cm. In Euroconfort and U.S. on posts cubicles, the resting place was 186 cm long and the neck rail was 107 cm above the mattress (as recommended). For the U.S. on rails cubicles, the neck rail was 112 cm above the mattress and positioned so that the resting area was 175 cm long (corresponding to the adjustments observed in the survey). The neck rail was placed above the lateral bars in both types of U.S. cubicles, and below that bar in Euroconfort cubicles (to leave 26 cm between the low lateral bar and the top of the mattress to prevent leg trapping). The prototype brisket board was made of sheet steel. It was 22 cm high and 18.5 cm wide, with the top and back edges chamfered (Figure 2). When the brisket board was used, it was fixed at 182 cm from the cubicle curb.

View larger version (14K): [in this window] [in a new window]   Figure 2. Side view of the prototype brisket board used in the experiment.

Conditions of Observation in yr 3.
Each pen was assigned a cubicle type on a given year, and this was changed between years so that the experimental design was balanced for cubicle types, pens, and years. When the cows were turned into the shed in mid-October each year, the pens were separated by fences and no brisket board was present in the cubicles. The prototype brisket board was installed 3 mo later to assess its beneficial effect as a correcting measure on the cows’ position. Finally, for the last month of observation, the fences between pens were removed to give free access to the different cubicle types to the cows. Each cow was identified by a number painted on its flanks. Three video cameras (SPT-M128CE, Sony Corp., Tokyo, Japan) were placed, one in front of each pen, and connected to time-lapse video recorders (Sony SVT-1000P, Sony Corp.). Time switches allowed 1-min recordings every 10 min.

Measurements.
The same measurements were made each year, except for preferences, which were assessed in the third year only. Table 1 summarizes the timing of the measurements.

Behavior
24-h Observations.
Every 3 wk from 3 wk after the cows had been turned into the shed, we observed their behavior for 24 h. During the day (from 0700 to 1700), an observer stood in front of the pens. At night (from 1700 to 0700), the animals were video-recorded using one camera (Sony SPT-M128CE) and one time-lapse video recorder (Sony SVT-1000P) per pen. Every 10 min, we noted the activity and the position of each cow as observed at first glance. For the activity, five exclusive classes were considered: eating (head in or above the trough), moving (taking at least one step), miscellaneous activity (drinking, licking salt, making a social encounter, etc.), idling (standing still), and lying down. The number of legs each cow had in a cubicle was noted (0 to 4).

Observation of Movements and Positions Associated with Lying Down.
Two days after the first four 24-h observations and 2 wk before the last one, we observed lying-down and getting-up movements of cows and their position in cubicles when lying. For this purpose, the cows were restrained at the trough for 1 h after the morning milking. When the cows were released, an observer noted on a tape recorder the beginning and the end of intentions to lie down, whether a lying-down movement was interrupted, the position of the head during the movement (in front of the animal between or below rails, on the side and between or below the horizontal bars), and contacts between the animal and a rail or a bar. These contacts were considered to be mild when the animal lightly touched a rail or a bar (no vibration of the rail or a bar detected), medium (some vibration detected, movement of the animal continuous), or strong (movement of the animal interrupted for 1 s or more). When at least two-thirds of the cows of a given pen had lain down, their position in the cubicle was recorded as in the survey (see above). They were then encouraged to stand up by a person approaching them from the back, and getting-up movements were observed as in the survey (see above); in addition the position of the head during the movement and contacts with a rail or a bar were noted.

Preferences.
When the partitions between pens were removed in yr 3, the position of each cow (in a cubicle of a given type or outside cubicles) was recorded at 0600, 1100, and 1530 every day for 3 wk.

Injuries and Cleanliness.
Cows were inspected individually on the day they were turned into the shed and then every 3 wk thereafter (on the day after we recorded lying-down and getting-up movements; see above). The cows were restrained at the trough after the morning milking. We recorded the external injuries and the cleanliness of each cow using the same method as in the survey (see above). We also recorded the extent (largest dimension) of hair loss around each injury and calculated their sum over all injuries of an animal. The cows were released after inspection.

Milk Production and Health.
The milk yield of each cow was recorded with milk meters every day during milking. Two days each week, milk samples were taken from each cow at each milking time. Somatic cell counts were further determined by a colorimetric method (Fossmatic 5000 automatic counter, Foss Electric, Hillerod, Denmark). Animals were inspected twice a day and received appropriate medical treatment when necessary. Lameness was controlled twice daily when the cows were led to milking. Any disease or other health problem was recorded.

Statistical Analyses.
The SAS software was used for all statistical analyses.

Preliminary Calculations.
From the size of the cows and for each cow, we calculated optimal cubicle width, length of the resting area, and height of the neck rail using the recommendations of the CIGR (1994) and Capdeville (Juan, 2001) (see above). We considered that the cubicle setting was lower than recommended when the width, length, or height was at least 5% less than the optimum for that cow, and higher than recommended when these dimensions were at least 5% greater than the optimum. From 24-h observations, we calculated the time spent in each activity by a cow (expressed in percentage of time per 24 h and arcsine-transformed), and the mean duration of its lying bouts. For milk yield and somatic cell counts, only data obtained when cows had calved for more than 8 wk and would not calve for the next 12 wk were kept, so that milk production was almost stable. Somatic cell counts in the milk were converted into decimal logarithms.

Comparisons Between Treatments.
Variance analyses were run on quantitative data. We expected that (putative) differences between treatments would be due to lying behavior. Because the behavior of an animal might depend on that of the other animals from the same group, we first tested group effects on lying frequency. As the group had no effect (F = 1.77, P > 0.10), the cow was taken as the unit of observation in all subsequent analyses. For cows that were used for more than 1 yr, we checked that their lying frequency was not consistent among years (F = 1.27, P > 0.10). We thus considered observations on a given cow in a different year as independent. The following mixed model was used to analyze the effect of treatments under study:

where Y_ijklmn represents data obtained on a cow and on a given day; a_i, b_j, c_k, d_l/k, and e_m represent the effects of the cubicle (U.S. on posts, U.S. on rails, and Euroconfort), the brisket board (presence or absence), the pen, the cow (random factor nested in the pen), and the time of observation (covariate taken for repeated effects); and f_ij represents the interaction between the cubicle and the brisket board. For analyses of milk yield and somatic cell counts, initial values calculated from the 3 wk before the animals were turned into the shed were added to the model as covariates. The covariance matrix was considered to be unstructured for all behavioral data because the behavior of an animal depends on many events that cannot always be controlled (e.g., estrus, external temperature). The covariance matrix was considered to have a compound symmetry for injuries, cleanliness, and milk production because these data were assumed to display a consistent evolution with time. The choice of the covariance matrix was confirmed by null model likelihood ratio tests. In the case of a significant global effect, treatments were compared using the least squares means procedure. For qualitative data, proportions of animals were compared using ² calculations. Cubicles were compared before and after the installation of the brisket board, and the effect of the brisket board was assessed by comparing cubicles before and after its installation. The results section will focus on the comparison between cubicles and the effects of the brisket board, and on significant effects (P < 0.05) or tendencies (0.05 < P < 0.10). For quantitative data results are expressed as means ± standard errors.

Evaluation of Preference Between Cubicles.
To assess the preference of a cow for a given type of cubicle (when partitions between pens were removed), we calculated the percentage of observations in which the cow was in a cubicle of this type out of all observations for that cow in a cubicle of any type. A t-test for paired data was used to compare this ratio with the theoretical ratio corresponding to no preference.

	Results

Top Abstract Introduction Materials and Methods Results Discussion Implications Literature Cited

 
Survey
Observations on Cubicles.
Few farms had cubicles with dimensions that corresponded to the recommendations (± 10%). The cubicle width, its total length, the length of the resting area, the length of the head space, and the height of the neck rail complied with the recommendation in 44, 26, 23, 57, and 20% of the farms respectively. Half the farms had cubicles that were narrower and/or shorter than recommended, and 43% were lacking in headspace. The neck rail was too low or too far from the curb in 74% and 59% of instances, respectively. In one-third of the farms, the cubicle curb was too high. When present, the brisket board was not far enough from the curb in 28% of instances and too far from it in 36% of instances.

The cubicles were equipped with a head rail in half of the farms. A head rail was found in 25% of the farms with cubicles made of standard or mushroom partitions, 58% of the farms with cubicles made of a loop with double fixing, and 76% of the farms with cubicles made of U.S. or single fixed loop partitions.

Observations on Animals.
Mean values obtained on farms for behavior, injuries, and cleanliness are shown in Table 2.

High values on the first axis of the PCA on behavior were observed when the cubicle base was flat (F = 7.07, P < 0.05), when the neck rail was high (regression coefficient (ß) for the height of the neck rail: ß = 0.17, F = 4.47; P = 0.05), and when the rear part of the cubicles was not cleaned regularly (F = 4.37, P = 0.05). Values on this axis tended to vary with the height of the cubicle curb (regression coefficient for the height of the cubicle curb: ß = 0.51, F = 4.09; P = 0.06).

Injuries and Health.
On average, two injuries were observed per cow. The areas most commonly injured were the hocks (57% of the cows), knees (51%), the neck (24%), the rump (19%), and the dewlap (17%). Lesions on other areas were observed in fewer than 12% of the cows. High values on the first axis of the PCA on injuries were observed when the cubicle base was flat (F = 11.02, P < 0.01), when the curb was high (ß = 0.50, F = 8.41; P < 0.01), when the neck rail was high (regression coefficient for the height of the neck rail: ß = 0.18, F = 6.92; P = 0.01), when the number of cubicles per animal was low (regression coefficient for the ratio of cubicles to cows: ß = –3.17, F = 5.23; P < 0.05), when manure was used for bedding compared with no bedding or other materials (F = 4.33, P = 0.01), when the cubicle was not cleaned regularly (F = 4.07, P = 0.05), and when the floor of the resting area was covered with a mat compared with a mattress or nothing (concrete; F = 3.30, P = 0.05). The occurrence of clinical mastitis and lameness varied greatly between farms, with no relation with cubicles characteristics (F = 0.60 and 0.56, P > 0.10).

Cleanliness.
The overall cleanliness score was 3.1 on average. The dirtiest area was the bottom part of the hind legs (mean score = 0.8) and the cleanest, the udder sides and belly (mean score = 0.35). The cleanliness varied little between farms (SE between 0.04 and 0.05 depending on the area of the body). Cows were dirtier (that is they obtained higher values on the first axis of the PCA on cleanliness) when the floor of the resting area was covered with a mat compared with a mattress (F = 3.56, P < 0.05). A decrease in the number of cubicles per cow tended to be linked to a higher dirtiness (regression coefficient for the ratio of cubicles to cows: ß = –3.74, F = 3.54, P < 0.10).

Correlations Between Behavior, Injuries, and Cleanliness.
The first axes of the PCA on behavior, injuries and clealiness were positively correlated to each other (r = 0.61 to 0.68, P < 0.01).

Experiment
Animal Size.
The cows used for the experiment were on average 144 ± 0.5 cm tall, 75 ± 0.8 cm wide, and 178 ± 0.8 cm long, with no significant differences between housing treatments. Forty-nine cows (out of 84) were larger than the 20% largest cows of the herd as measured in 1999. Cubicles were narrower than recommended for ten cows in Euroconfort cubicles, and three cows in U.S. cubicles either on rails or on posts (² = 7.9, P < 0.05). They were shorter than recommended for 21 cows in U.S. cubicles on rails, one cow in U.S. cubicles on posts and eight cows in Euroconfort cubicles (² = 33.5, P < 0.01). The neck rail was higher than recommended for six cows in U.S. cubicles on rails and one cow in Euroconfort cubicles (² = 9.9, P < 0.01). The cubicles were rarely larger, longer, or had a lower neck rail than recommended (less than two cows per cubicle type), with no difference found between cubicles.

Behavior
24-h Observations.
Means and standard errors corresponding to 24-h observations are shown in Figure 3. Statistics are given in Table 3.

View larger version (34K): [in this window] [in a new window]   Figure 3. Time budget of cows (n = 28 per treatment) with access to U.S. cubicles fixed on free-standing posts (----) or cantilevered on a double head rail (——), or to Euroconfort cubicles cantilevered on a double head rail (——). The cows were observed every 10 min for 24 h every 3 wk before and after installation of a brisket board. Top left: time spent lying; top right: time spent idling (standing immobile); bottom left: time spent idling with all four hooves in a cubicle; bottom right: mean duration of lying bouts. See Figure 1 for a description of U.S. and Euroconfort cubicles.

View this table: [in this window] [in a new window]   Table 3. Effect of time, type of cubicles (U.S. cubicles fixed on freestanding posts, U.S. cubicles cantilevered on a double head rail, Euroconfort cubicles cantilevered on a double head rail; n = 3 x 28), presence of the brisket board, and their interaction on the behavior of cows (means and SE are shown on Figure 3 or given in the text)a

Movements and Positions Associated with Lying.
Data and statistics on positions during lying and on getting-up movements are given in Table 4 for comparisons between cubicles and in Table 5 for assessment of the effect of the brisket board. Data and statistics on lying-down movements are given in the text below.

Once the cows were lying down, their bodies were in the middle of the resting area (that is, not in a fore or rear position) in 68% of instances, and their heads were forward in 65% of instances. Before the brisket board was installed, the cows in U.S. cubicles were more often lying in a fore position and less often in a rear position than the cows in Euroconfort cubicles, and the cows in U.S. cubicles on posts or in Euroconfort cubicles more often had their head forward than cows in U.S. cubicles on rails. The presence of the brisket board reduced the incidence of lying in a fore position in U.S. cubicles and the incidence of lying with the head forward in Euroconfort cubicle. Differences between cubicles in the position of the body disappeared after the brisket board was installed. When cows had their head to the side, they placed it between the lateral bars in Euroconfort cubicles and below these in U.S. cubicles. The cows were lying in a diagonal position (i.e., with a hip inserted below a lateral bar) in 38% of instances, with no variations due to the type of cubicles or the presence of the brisket board.

Getting up was always preceded by only one intention. During these movements, the cows placed their head forward on 90% of instances; if not, they placed their head between the lateral bars in Euroconfort cubicles and below these bars in U.S. cubicles. Before the brisket board was installed, the cows in U.S. cubicles on posts or in Euroconfort cubicles touched the neck rail more often than the cows in U.S. cubicles on rails; contacts with a front rail were very rare in Euroconfort cubicles but more common in U.S. cubicles on rails. The brisket board reduced the frequency of contacts with the neck rail in U.S. cubicles, and decreased the contacts with a front rail in U.S. cubicles on rails. After the brisket board was installed, the contacts with the neck rail were still more common in Euroconfort than in U.S. cubicles, but no difference in the frequency of contacts with the front rail was observed. The contacts with the neck rail were generally mild (53%) or medium (45%), and rarely strong (2%).

Injuries and Cleanliness.
Means and standard errors for injuries recordings are shown in Figure 4. The F-and P-values are given in the text below for significant effects.

View larger version (18K): [in this window] [in a new window]   Figure 4. External injuries and hock inflammatories on cows (n = 28 per treatment) with access to U.S. cubicles fixed on free-standing posts (----) or cantilevered on a double head rail (——), or to Euroconfort cubicles cantilevered on a double head rail (——). The cows were observed every 3 wk before and after the installation of a brisket board. Left: total number of injuries; right: total extent of hair losses (taken on all injuries). See Figure 1 for a description of U.S. and Euroconfort cubicles.

The score for cows’ cleanliness was on average 3.33 ± 0.07, with no variation with time, cubicles or the presence of the brisket board.

Milk Production and Health.
Cows produced on average 21.6 ± 0.21 kg milk/d, with no variations due to the type of cubicle or the presence of the brisket board. The somatic cell counts averaged 108,500 ± 500/mL of milk. It increased with time in cows with access to U.S. cubicles on rails, while it decreased in cows with access to U.S. cubicles on posts or to Euroconfort cubicles (coefficients of regression for week effect on the log of cells counts were 0.09, –0.05, and –0.06, F-value for time x cubicle = 2.96; P < 0.05; Figure 5). No major effect of the type of cubicle or of the presence of the brisket board was observed on somatic cell counts. Over the 3 yr, between two and four cows per housing treatment suffered from mastitis or lameness.

View larger version (20K): [in this window] [in a new window]   Figure 5. Somatic cell counts in the milk of cows (n = 28 per treatment) with access to U.S. cubicles fixed on free-standing posts (----) or cantilevered on a double head rail (——), or to Euroconfort cubicles cantilevered on a double head rail (——), before and after installation of a brisket board. See Figure 1 for a description of U.S. and Euroconfort cubicles.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion Implications Literature Cited

 
The results of the survey and the experiment suggest that in a loose housing system with cubicles the behavior of cows, the occurrence of external injuries to them and their cleanliness are affected by cubicles characteristics.

The survey showed that French loose housing systems with cubicles rarely complied with the proposed recommendations (CIGR, 1994; Juan, 2001).

First, cubicles were too narrow in 50% of the farms. This may force cows to lie down with part of their body in an adjacent cubicle. Indeed, more than half the cows observed in our survey were found adopting such a lying position.

Second, according to CIGR (1994) and DEFRA (U.K. Department for Environment Food and Rural Affairs, 2002), cubicles must be long enough to enable the cow to rest comfortably on the floor or to stand with all feet in the cubicle. Too short a cubicle can increase the frequency of standing with only the forefeet inside the cubicle, a position that exposes the cow to hoof lesions (Galindo and Broom, 2000; DEFRA, 2002). In our survey, cubicles were too short in 50% of the farms. Nevertheless, only 14% of the cows lying in a cubicle had their rump less than 5 cm away from the cubicle curb, and 29% of the cows standing in a cubicle had a hoof as close to the curb. Farmers may have corrected the shortness of cubicles by fixing the neck rail and the brisket board further from the curb than recommended (in 59% farms for the neck rail and 36% for the brisket board). This can help to maintain a proper length for the resting area but limits the length available for the head. Also, this in turn can render lying-down and getting-up movements more difficult because cows cannot thrust their head forward properly, especially for lunging while getting up (CIGR, 1994; DEFRA, 2002). However, we did not observe a high frequency of lying-down or getting-up intentions (1.4 and 1.1 per successful movement, respectively), the proportion of interrupted movements remained low (12.5 and 4% for lying-down and getting-up movements), and dog-sit positions were observed only in 4% of the farms. It is likely that cows adjusted their behavior by placing their head to the side of the cubicle rather than forward.

Large variations were observed between farms in the state of the cows. The multivariate analyses helped us to understand these variations. First, the composition of the first axis of the PCA run on behavioral data revealed that difficulties in lying down or getting up are associated with animals lying in a fore position in the cubicles and/or with their body partly in adjacent cubicles. This confirms the assumption that fore positions in cubicles limit the movements of cows. Cows that lack space for their head probably position it to the side, and this may result in more diagonal positions of the cows in cubicles, as suggested by the high value of diagonal positions on the first axis of the PCA. The difficulties in movements and positioning in the cubicles were more frequently observed in farms where the neck rail was high. The neck rail prevents cows from stepping forward in cubicles. Accordingly, it should be located below the cows’ withers (15 to 25 cm below the withers height for DEFRA [2002] or at 75% of that height for CIGR [1994]). Our results confirmed that a high neck rail does not allow good positioning of cows.

Second, the composition of the first axis of the PCA run on external injuries revealed that farms can be ranked according to the number of external injuries to the cows, regardless of their location or severity. Injuries can have various causes: inappropriate feeding facilities or cubicle partitions, sharp edges on a brisket board, hard bedding, etc. (Wechsler et al., 2000; DEFRA, 2002; Livesey et al., 2002; Hansen et al., 2003). Again, the position of the neck rail is evidently of prime importance: injuries were more common when the neck rail was high. The cubicle partitions should be built so that certain zones remain unobstructed: a zone for the head, a fairly small low zone in the middle of the cubicle to allow the passage of the front legs, and a large zone at the end of the cubicle to prevent injuries to the rump (CIGR, 1994). If the positioning of the cows is inappropriate (e.g., if the cows step too far into the cubicle because of too high a neck rail), the animals are likely to hit obstructions in the cubicle, which may cause injuries.

Injuries were more frequent when the number of cubicles per cow was reduced. It is likely that in such situations, dominant cows displace subordinates to take their place in a cubicle, which could expose them to injuries due to collisions with partitions during increased lying-down and getting-up movements. Our observations confirm the recommendation of at least one cubicle per cow (DEFRA, 2002).

As already reported by Livesey et al. (2002), mattresses on the cubicle base were associated with a reduction of injuries compared with mats; however, we found no significant differences between farms where cubicles were covered with a mattress and those where cubicles were not covered by specific flooring. In our survey, no bedding was used in two-thirds of farms using mats whereas some bedding was generally put on the other floors. Despite no evidence of a clear effect of bedding on injuries in our survey, flooring and bedding probably interact with each other, leading to more injuries when mats are used without bedding. In addition, cows display a strong preference for mattresses over concrete, preference for mats being intermediate (Natzke et al., 1982; Sonck and Daelemans, 1999; Bony and Barbet, 2000). Hence, mattresses, and not mats, should be recommended in cubicles.

We did not find any relation between the characteristics of cubicles on a farm and clinical mastitis or lameness; however, the occurrence of these diseases was assessed from the farmers’ own records over the previous year. These records may not have been accurate enough to highlight effects.

Third, the composition of the first axis of the PCA on cleanliness revealed that farms could be ranked according to the overall cleanliness of cows, whatever the body area concerned. Like injuries, cleanliness depends on a large number of factors: feeds, regularity of cleaning of the shed by the farmer, design of the cubicles, type of flooring, etc. The results of our survey suggested that mats make cows dirtier than other flooring.

We observed positive correlations between cow behavior, injuries, and cleanliness. One might conclude that difficulties in lying down or getting up result in more injuries or that lying too fore in a cubicle leads to dirtier cows. However, these elements could be all linked to common external factors, like poor management. Cow behavior, injuries, and cleanliness were also linked to a number of factors other than the ones listed above. At this time, we cannot account for all these links by direct cause-and-effect relations. For instance, we observed that cows displayed more behavioral problems (difficulties in lying down or getting up, diagonal or rear positions) and had more external injuries in sheds that were not cleaned regularly. A similar result has been reported in calves with a better feed efficiency in clean veal units compared with dirty ones (Lensink et al., 2001). As suggested by Lensink (2000), the cleanliness of a shed can reflect the stock handler’s implication in his/her farming work. In turn, stock handling proficiency is known to affect many aspects of animal production and welfare: a positive attitude toward animals and farming tasks is generally accompanied by more positive behavior toward animals, low fear reactions of animals, and higher production (in terms of growth, milk, reproduction success, etc.; Hemsworth, 1997). Although we could not separate the effects of the physical environment of the cows and those of the management by the farmer, our survey suggests that cleanliness is a good index of proper stock handling.

As recommended by CIGR (1994), newer cubicles, such as those with U.S. partitions, seem better able to meet animals’ need for free space for the head and rump. However, we found no evidence of an effect of the type of partition on cow behavior, injuries, or cleanliness. This suggests counteracting factors. Partitions of recent cubicle types, such as those of the U.S. type, were often cantilevered on head rails, whereas cubicles of older types (standard or mushroom partitions) were generally fixed without a head rail. The head rail limits the space available for the cow’s head. Hence, improvements due to partitions with a more ergonomic design might be offset by an inappropriate use of these partitions. In addition, informal discussions with farmers involved in the survey suggested that they were aware of problems caused by the head rail. Some of them seem to have readjusted their cubicles by fixing the head rails in a high position, leaving enough space below the rails for the cows to thrust their head forwards. Because the neck rail is usually placed on top of lateral bars (fixed on the head rails when they are cantilevered), this can result in a position of the neck rail that is higher than recommended. Our results suggest that a high neck rail can result in more behavioral difficulties and more injuries. However, due to the limited number of observations, such interactions between type of partition and position of the neck rail could not be assessed adequately in our survey.

According to DEFRA (2002), insufficient width or length, lack of a comfortable base, and inappropriate divisions are major factors that put at risk the welfare of cows housed in a cubicle system. The results of the survey lend support to these guidelines. However, some factors (such as management) or interactions (for instance between cubicle type and cubicle adjustments) could not be assessed accurately. In a survey, causal factors are inevitably linked. We thus decided to conduct a controlled experiment to compare two types of cubicle partition and two possible fixings of these partitions corresponding to a different position of the neck rail, and to assess the benefits provided by a brisket board.

In this experiment, we specifically compared two fixings of U.S. partitions (cantilevered on head rails and with a neck rail higher and in a more rearward position than recommended vs. fixed on free-standing posts with a neck rail placed as recommended) and the Euroconfort partitions (cantilevered on head rails but a neck rail placed as recommended).

Our cows spent 12.5 h of the day lying down in the cubicles. Results in the literature report that cows spend between 10.6 and 12.6 h lying (Sonck et al., 1999; Raasch et al., 2000; Wechsler et al., 2000; Stefanowska et al., 2001). Mattresses increase the time spent lying (Chaplin et al., 2000), probably because they offer more comfortable bedding. This explains why the time the cows spent lying in our experiment, where the cubicles were covered with mattresses, corresponds to the higher values found in the literature.

The cows offered U.S. cubicles cantilevered on rails spent more time lying down and less time idling than the cows offered Euroconfort or U.S. cubicles fixed on freestanding posts, especially after the brisket board was installed. When a resting place is uncomfortable, cows lie down less and this is compensated for by more idling (Haley et al., 2001). Such compensation was observed in our experiment, the cows spending a total of 16.8 h lying or idling, whatever their cubicle and with or without a brisket board. The increase in the time cows spent lying down in U.S. cubicles on rails was accompanied by longer lying bouts. Long lying bouts can result from difficulties in lying down or getting up (CIGR, 1994). However, we noted no difficulties in lying down or getting up. Hence, difficulties in lying down or getting up cannot account for the longer lying bouts of cows in U.S. cubicles on rails.

In U.S. cubicles on rails, the neck rail was 11 cm closer to the cubicle curb than in U.S. cubicles on posts or Euroconfort cubicles. Hence the resting area of U.S. cubicles on rails was too short for most cows. This may make standing with all four legs in a cubicle more difficult. In fact, cows spent less time idling inside U.S. cubicles on rails, whereas no difference between cubicles was found in the time spent idling outside cubicles or with the front legs only in a cubicle. Hence, it is likely that the lower proportion of time spent idling and the higher proportion of time spent lying in U.S. cubicles on rails was due to difficulties in standing fully in the cubicles owing to the rear position of the neck rail.

When getting up, cows in U.S. cubicles on posts or in Euroconfort cubicles touched the neck rail more often than those in U.S. cubicles on rails, especially before the brisket board was installed. Again, this is likely to be due to the position of the neck rail, which was lower and further from the curb in these cubicles than in the U.S. cubicles on rails. In our experiment, getting-up movements were induced. In response to a stimulus, sows get up quickly and their movements are more affected by the animal size than spontaneous getting-up (Marchant and Broom, 1996). This suggests more difficulties in getting up when the movement is induced. However, in our experiment, the contacts with the neck rail did not modify the getting-up movements, which were mostly continuous and always successful.

The cows lie down in the same position in U.S. cubicles fixed on posts or cantilevered on rails. Hence, a rear but high neck rail, as in U.S. cubicles cantilevered on rails, seems not to affect the position of the cows when lying. The cows lie down more often in a rear position and less often in a fore position in Euroconfort cubicles than in U.S. cubicles, despite the same placement of the neck rail in Euroconfort and U.S. cubicles on posts. It is likely that the low front rail of Euroconfort cubicles prevented the cows from advancing too far in the cubicle. In addition, in Euroconfort cubicles, we observed that cows tried to place their head below the front rails during about one in five lying-down movements. They could only insert their nose since the bottom rail was only at 32 cm from the floor. In Euroconfort cubicles, the low rail may have hindered the cows while they were lying down, but this hindrance disappeared once they had lain down; in Euroconfort cubicles cows had their head forward when lying as often as those in U.S. cubicles on posts, where there was no obstacle in front of the cows. In U.S. cubicles on rails, cows lie down more often with the head to the side than in other cubicles. In these cubicles, the low rail was 80 cm above the mattress, which may not be enough for a large cow to keep its head up. Hence, a low front rail can disturb animals while they are lying down but a high one can disturb them once they are lying.

The presence of the brisket board modified lying-down and getting-up movements, and the position of cows when resting. More specifically, in U.S. cubicles, it reduced the incidence of cows lying in a fore position and raised that of cows lying in the middle part of the resting area, probably because it limits the progression of the cow just before it lies down. Such effects were not observed in Euroconfort cubicles where the cows lie down rarely in a fore position already in the absence of the brisket board. The correcting effect on cows’ position in U.S. cubicles was accompanied by a reduction in contacts with the neck rail or the front rail (when present) during getting-up movements. After the brisket board was installed, cows touched the neck rail when they got up more often in Euroconfort cubicles than in U.S. cubicles. This could not be due to a fore position of the cows, since in these cubicles cows were rarely in such a position. During getting-up movements, cows usually extended their head forward and very close to the floor (CIGR, 1994). Such a movement was difficult in Euroconfort cubicles because of the position of the low front rail. As a consequence, cows in Euroconfort cubicles may have had to get up differently from those in U.S. cubicles. Difficulties in getting up due to a low front rail could explain why cows that had been kept in Euroconfort cubicles preferred lying down in cubicles without a front rail (i.e., U.S. cubicles on posts) when they could choose between cubicles afterward. However, the results for preferences must be viewed with caution because cows that had had access to U.S. cubicles on posts did not choose these later.

The differences in lying time and lying position between cubicles were not accompanied by variations in the cleanliness of cows. The lack of effect on the cleanliness of cows may result from feces being removed from the cubicles twice a day. Also, the slope of the resting area (7% in our experiment) may have facilitated the drainage of the resting area. Our cows obtained a score of 3.3 on the 0-to-10 scale proposed for cleanliness by Faye and Barnouin (1985), which corresponds to clean cows. Similar scores were obtained by Chaplin et al. (2000) in similar circumstances. Dirty cows can have high somatic counts probably because they are more in contact with microbes (Agabriel et al., 1997). We observed a regular increase in somatic cell counts in cows with access to U.S. cubicles on rails and a symmetrical decrease in cows with access to U.S. cubicles on posts or to Euroconfort cubicles. This may derive at least partly from initial differences between cows: before the start of the experiment, the somatic cell counts were high in cows allocated to U.S. cubicles on posts or to Euroconfort cubicles (1,260,000 and 938,000 cells/mL) compared with those of cows allocated to U.S. cubicles on rails (687,000 cells/mL). Hence, the somatic cell counts barely increased in the former. In addition, cows in U.S. cubicles on rails spent more time lying down than cows in U.S. cubicles on posts or in Euroconfort cubicles. During lying, the udder is in contact with the floor. Feces were removed regularly from the resting area but microbes were probably still present on the floor. Hence, an increased lying time may facilitate udder contamination, and in turn induce high somatic cell counts. In addition, during lying, the udder can undergo pressure caused by the weight of the hindquarters, and this may have been particularly marked in our experiment due to the 7% slope of the resting area. The regular increase in somatic cell counts of cows with access to U.S. cubicles on rails may thus come from the increased time they spent lying.

The differences in lying time, lying position, and contacts with rails between cubicles were also not accompanied by variations in the occurrence of external injuries. Skin injuries were frequent in our study, and became increasingly so as time progressed, so that after 4 mo in the shed, most cows were affected by a skin lesion. These lesions were not severe (i.e., they consisted of hair losses and some abrasion of the skin). They were concentrated on the hocks, suggesting a link to the flooring of the resting area (Schaub et al., 1999; Bergnsten, 2001). In our experiment, the base of the cubicles was covered with a mattress. Mattresses decrease the incidence of lesions (Livesey et al., 2002). Thus the high incidence of lesions observed in our experiment cannot reasonably have come from an uncomfortable resting area. The slope of the resting area (7%) was higher than the value of 4% recommended by CIGR (1994). This may have resulted in a transfer of part of the animal’s weight to its hindquarters, increasing the pressure of the hocks on the floor. Hence, despite possible positive effects on the drainage of the cubicle base and no apparent effect on the lying behavior of cows, a 7% slope cannot be recommended for cubicles because it may facilitate lesions on the hind legs of cows.

The frequency of injuries and their size (measured by the extent of hair losses) or severity did not increase any further after the brisket board was installed. It is unlikely that the brisket board reduced the pressure of the hocks on the floor. More likely, the size of skin lesions had reached a plateau, cows having only a limited number of points of their body where the pressure on the floor is particularly high, these points being the hocks.

In conclusion, no major problems were observed in the three cubicle types used in the experiment: U.S. cubicles fixed on freestanding posts or cantilevered on rails and Euroconfort cubicles on rails. Cows never got stuck when lying down or getting up, no major injury was observed and no differences in milk production were noted. These three types of cubicles, at least when their adjustment does not depart greatly from recommendations (CIGR, 1994; Juan, 2001) can be used on farms. However, differences were observed in the way animals used them, and slight adjustments can be proposed for a better use by animals. First, U.S. cubicles should be used with a brisket board to prevent animals advancing too far in the cubicles. Second, when U.S. cubicles are cantilevered on rails, the neck rail is generally fixed on top of the lateral bar. It is suggested that the neck rail be fixed below the lateral bars, as was done for the Euroconfort cubicles in this experiment. This makes it possible to fix it as recommended, at (0.75 xH) and in a less rearward position (0.96 xL) + 15 cm from the cubicle curb. In this position, the neck rail does not hinder animals from fully standing in the resting area. Third, Euroconfort partitions do not appear to be an optimal solution to avoid front rails hindering resting animals because their low front rail is positioned where the animals usually put their head when they lie down or get up.

The survey and experiment suggest that the adjustment of the cubicles (presence of a head rail, position of the neck rail, presence of a brisket board) and their flooring are of prime importance for the welfare of animals. The front rail can be an obstacle, forcing an animal to place its head to one side when resting, and making lunging more difficult when getting up. A neck rail in a back position forces animals to lie in the cubicle rather than stand in it. A low and fore neck rail and a brisket board prevent cows from stepping forward in the cubicle, limiting fore positions. Cows lying in a fore position in a cubicle are more likely to have difficulties getting up and incur a higher risk of injuries. There seems to be no conflict between decreasing lying restrictions and injuries and improving the cleanliness of cows.

	Implications

Top Abstract Introduction Materials and Methods Results