1. Introduction
Housing with fully perforated floors is commonly used for early-weaned pigs, since this encourages the physical separation of the pig and its dung, and allows increased
stocking density. However, the behavioural and other disadvantages of this type of Ž
. housing are well documented for example, van Putten and Dammers, 1976 . Bedded
housing systems may redress the behavioural problems but can encourage enteric Ž
. problems for example, Silver, 1989 , and have other practical management disadvan-
tages such as increased labour requirement. The Straw-Flow
w
housing system was developed to provide the welfare advantages of straw without the associated manage-
Ž .
ment disadvantages Bruce, 1990 , and was shown to be successful in this for
Ž .
growingrfinishing pigs by Lyons et al. 1995 . In evaluating a new housing system for its effects on animal welfare, many factors
Ž .
must be taken into consideration. Broom 1989
stated that ‘‘The welfare of an individual is its state as regards its attempts to cope with its environment’’. Welfare can
therefore be assessed by measuring the extent of coping strategies, one form of which involve behavioural modification. Abnormal behaviour is not necessarily a sign of
Ž .
suffering Dawkins, 1980 , and should be interpreted with caution. However, a compari- son of behaviour in different environments can provide indications of modified patterns,
and identify whether one system leads to the performance of more potentially harmful behaviours than another. The built environment may lead directly to abnormal be-
haviours, for example by restricting movement, or providing slippery floors which may
Ž lead to abnormal methods of lying and rising, and a reluctance to change posture Fraser
. and Broom, 1990 . It may also give rise indirectly to abnormal behaviour by failing to
provide adequate circumstances for appropriate expression of highly motivated be- haviours. Thus, changes in activity patterns, pen-directed behaviours and pig-directed
behaviours may be indicative of housing inadequacy. Such behavioural assessment should then be combined with assessment of other welfare parameters to arrive at an
overall conclusion. This paper describes a behavioural investigation into the suitability of Straw-Flow for early-weaned pigs, when compared with systems chosen to reflect the
range of other housing in curent commercial use. The long-term effects of weaner housing are also considered, since residual influences of the early housing environment
Ž .
have previously been reported in pigs for example, Boe, 1993 . Other, nonbehavioural Ž
. Ž
. aspects of the investigation are described by Kelly et al. 1998 , and Kelly et al. 2000 .
2. Materials and methods
2.1. Pen design Four weaner pens were constructed within two thermostatically controlled rooms.
Each pen was designed for 20 pigs from 6 to 20 kg. All weaner pens were provided with two nipple drinkers and a seven-space feeder with shoulder guards. Two ‘‘straw’’ pens
Ž .
a and b were built within the same room. Both initially had a kennel covering the lying area. This was constructed of two overlapping sheets of 10-mm plywood, with 125-
mm-wide plastic-strip curtain across the front to allow easy access for pigs while conserving heat. The roof sheets moved independently, so that the kenneled area could
Ž .
be expanded as the pigs grew. Two flatdecks c and d were built within a second room, both with fully perforated floors. There are many types of flooring available; however
2073F galvanised expanded metal was considered to be the most commonly used type on farms. The pen floor was 790 mm above the ground, providing under-floor slurry
storage. Waste was disposed of after each group of pigs had been removed. No bedding was provided. Pens in all systems were thoroughly cleaned between replicates.
Ž .
2 2
Ž a Deep-straw. The pen measured 2.75 = 1.76 m providing 0.23 m rpig excluding
. feeder area . Bedding was allowed to build up, straw being supplied as necessary to
maintain a clean, dry lying area. Manure was removed about three times per replicate, as necessary to maintain a clean, dry lying area. Straw was supplied, 141.6 kg in total
Ž .
mean of three replicates , 4.4 times the amount provided to the Straw-Flow pen. There was a double-glazed window in the pen wall to allow observation of behaviour within
the kennelled area. Ž .
Ž .
b Straw-Flow see Fig. 1 . This pen had the same dimensions as a. The concrete Ž
. floor sloped at 1:16, with a 100 mm step down into the dunging area 600 mm wide .
There was a gate at the front of the pen with a 75 mm gap beneath it to allow dung to pass out of the pen There was a double-glazed window in the pen wall to allow
observation of behaviour within the kennelled area. One kilogram straw was placed in
Ž the pen before the pigs arrived. A further 1 kg was supplied each day 0.5 kg at 0900 h,
. and 0.5 kg at 1700 h, after the day’s observations were complete . The straw was shaken
apart inside the kennel. Waste straw and manure accumulated outside the pen and were removed daily. The straw provided was not sufficient to allow a bedded layer to
accumulate, but a ‘‘crust’’ formed over the surface of the lying area, comprising straw, feed, skin, dung etc.
Fig. 1. Elevation of the Straw-Flow
w
weaner pen.
Ž . Ž .
Ž . c ‘‘Large’’ flatdeck. This pen had the same dimensions as a and b , providing
0.23 m2rpig. Ž .
d ‘‘Small’’ flatdeck. This pen was constructed to represent commercial practice, measuring 3.05 = 1.22 m
2
and providing 0.17 m
2
rpig. Ž .
e Grower pens. The grower pens, which were together within the same room, were all of Straw-Flow design, measuring 2.7 = 4.0 m
2
providing 0.68 m2rpig for 16 pigs. The step down into the dunging area was 100 mm, with a 100-mm gap under the gate.
Each pen was provided with one double-space feeder with shoulder guards, and two bite drinkers which had guards and a stepping block to enable smaller pigs to reach the
Ž .
drinkers easily. Straw 1.9 kg per pen per day was provided via a self-help dispenser, giving the same amount of straw per floor area as the weaner pen.
2.2. Animals In each replicate, 20 three- to four-week-old entire male pigs from the same farm
were weaned into one pen of each system on the same day, with initial weight balanced across the treatments. All of the pigs received from the breeding farms had been tail
Ž .
docked shortly after birth. Replicate one Landrace = Large White came from one farm, Ž
. replicates two and three Landrace = Duroc= Large White from another. The grower
pens could accommodate only 16 pigs at 90 kg liveweight. In the first replicate, the four ‘‘extra’’ pigs were removed from each pen at 40 kg liveweight. In the second and third
replicates, the four ‘‘extra’’ pigs were removed at the time of transfer to the grower pens.
2.3. Management Ž
. During the post-weaning period, pigs were inspected twice daily 0900 and 1700 h .
Low-level lighting was used continuously to permit video recording of pig behaviour, although no video data are presented in this paper. Fluorescent lighting was available to
permit inspection and observations. Both rooms were maintained under the same lighting regime, with fluorescent lighting switched on between the morning and evening
inspections. The set temperature within each room was adjusted according to the pigs’
Ž changing requirements, as determined by the computer model ‘‘PigCrit’’ Bruce and
. Clark, 1979 , which calculates upper and lower critical temperatures for young pigs.
Straw bedding reduces the lower critical temperature of a group of pigs relative to lying on bare concrete, and a kennel over the lying area retains heat produced by the pigs.
This made it appropriate to operate the straw room at a lower temperature than the flatdeck room, with the pigs remaining within their thermoneutral zones. The recorded
mean temperature over the three replicates in the straw room was initially 23
8C, decreasing to 16
8C by the end of the weaner period, and in the flatdeck room 278C, decreasing to 18
8C. The pigs were transferred to the grower pens after 4 weeks Ž
. Ž
. Ž
replicates 1 and 2 or 5 weeks replicate 3 in the weaner pens at a liveweight of 19.6 .
kg . During the grower period, pigs were inspected twice daily. At 0900 h, the feed was replenished, manure removed from the external dung channel and the area washed.
2.4. Daytime behaÕiour recording In this study, scan sampling was the method of choice for behaviour recording
because it provides ‘‘a reasonably accurate estimate of the amounts or percentages of Ž
. time individuals or group members spend in particular activities’’ Banks, 1982 . It also
allowed all four treatment groups to be observed contemporarily by the same observer. Ž
. In replicate one, 12 animals per pen were selected for observation ‘‘focal animals’’
to provide a spread of weights at weaning. In replicates two and three, 14 focal animals were selected. The same animals were observed throughout the study, regardless of
subsequent performance. Some focal animals were removed when the groups were split; however, 10 focal animals remained in each group during the grower stage.
Ž At each timepoint, each focal animal’s activity was recorded in terms of posture 1 of
. Ž
. Ž
. 5 possible , behaviour 1 of 18 and substrate 1 of 17 . Some activities were observed
to occur only rarely, and so were subsequently combined for statistical analysis where appropriate. A detailed description of the categories analysed is provided in Table 1. Of
Table 1 Categories used in the analysis of daytime behavioural observations
Category Definition
Posture Standing
weight carried on four feet Sitting
dog-sitting position Kneeling
front knees and hind feet in contact with floor Lying
combined category of lying laterally and lying ventrally Substrate
Straw-directed any of the behaviours nose, root, carry, paw, biterchew with straw as the substrate
Pen-directed any of the behaviours nose, root, carry, paw, biterchew with floor, walls or fence as
the substrate Ž
Pen-fittings-directed any of the behaviours nose, root, carry, paw, biterchew with feeder, drinker or block .
or guard , kennel or straw dispenser as the substrate, except maintenance behaviours Ž
. see below
Ž .
Ž .
Pig-directed nose ‘‘social behaviour’’ , root, paw, biterchew ‘‘nonsocial behaviour’’ any part of
Ž .
another pig tail, ear and abdomen recorded separately ; shoverpush pig, avoid, fleer Ž
. squeal aggression or playfighting recorded separately ; mount pig
BehaÕiour
a
Maintenance feed at feeder , drink at drinker, await feeder or drinker, eliminate urine or dung, rub,
stretch, yawn, cough, vomit Motion
walking around pen; no obvious other behaviour Ž
. Inactive
activity and substrate both recorded as ‘‘none’’ any posture Total play
runningrscampering, playfighting Vacuum chewing
chewing with no visible substrate Total nosing
nasal disc applied to any substrate without pressure Total rooting
nasal disc applied to any substrate with pressure and forward movement Total chewing
any substrate taken into mouth and jaw movements made
a
Defined as head in feeder: it was not possible to establish whether pigs were actually eating; similarly, pigs with drinker in mouth were recorded as drinking.
these, behaviours directed at the pen and pen fittings were considered to be detrimental as there was the possibility of injury to the pig as well as the development of
stereotypies. Nonsocial pig-directed behaviour was also considered to be undesirable because of the risk of injury to the recipient. Some other behaviours, although not
necesssarily overtly harmful, were also considered to be detrimental as indicative of coping difficulty if performed over long periods of time, for example inactivity and
vacuum chewing. Behaviours which were considered to be ‘‘beneficial’’ included play Ž
. Dybkjaer, 1992 , and straw-directed behaviour, as straw offers the opportunity for a
variety of different manipulations and provides a substrate for the strong foragingrex- ploratory motivation of young pigs. Similar distinctions have been made by other
Ž .
authors, for example Boe 1993 . The pigs were allowed to settle for 30–60 min after the daily feeding routine was
completed. This period was considered necessary, as the stock people had to enter the pens to clean and refill the hoppers, check drinkers and deliver straw, and thus
influenced the pigs’ behaviour. Behaviour was recorded at 30-min intervals between 1030 and 1700 h. Where it proved impossible to complete the observation cycle during
the 30 min, the recording interval was extended. This was particularly necessary during the weaner period, when additional time was required to move quietly in and out of the
separate weaner rooms.
The observer was always positioned outside the pens to gain a clear view of the pigs’ activity but making no attempt to hide from them. After entering the room, the observer
allowed 2–3 min to elapse before starting to record behaviour, to allow any interest caused by her entry to subside. Generally the pigs ignored the observer, who was
familiar to them. All observations were recorded by the same person throughout the study. The pigs within each pen were always observed in the same order. The weaner
pens were always observed in the order: small flatdeck, large flatdeck, Straw-Flow, deep straw. The grower pens were also always observed in the same order but, as the groups
of pigs were transferred from the weaner pens in a randomised manner, the grower pen order did not necessarily reflect the weaner pen order.
Behaviour was recorded on 3 days in each week. Since the behaviour of pigs in the same pen was not independent, all the observations collected in each pen over the 3-day
period were summed, to produce a frequency of occurrence of each behaviour in each pen for that ‘‘week’’. These are expressed in the results as a percentage of the total
number of recorded observations. Behaviour was recorded and analysed each week in the weaner pens, and for the first, second, fifth and eighth week in the grower pens. In
addition, these results were summed to produce a total for the weaner and grower periods.
2.5. Statistical analysis Data, as defined above, were analysed as pen means by two-way analysis of variance.
Ž .
Ž .
Treatment 3 df and replicate 2 df were used as factors in the analysis, leaving 6 df for the error term. Where the data were found to be not normally distributed, they were
Ž .
normalised using a square root transformation. A GLM procedure Minitab 9.2 , which Ž
. fits the general linear model GLM , allowed analysis of unbalanced samples. Where the
GLM analysis indicated significance, pairwise comparison of the treatment means was made using the method of least significant difference.
3. Results