2.6. Statistical analysis Ž
Interactions between the four treatment combinations hereafter abbreviated to treat- .
ment and focal pig weight category on drinking behaviour, lesion score and ADG were investigated using analyses of variance with weight class, treatment and replicate as
factors. Interactions between drinker ratio and group size on the drinking behaviour of the pooled focal pigs were examined using two-way analyses of variance. Changes in
lesion score value over time were investigated using a one-way analysis of variance. The existence of significant correlations between the individual parameters of drinking
behaviour and between mean lesion score and parameters of drinking behaviour were examined using Pearson Product Moment Correlations. A Wilcoxon Signed Rank Test
was used to examine differences in drinking bout length during the period of the day with the maximum compared to minimum duration of drinking behaviour.
3. Results
Unless explicitly stated the results presented below describe the drinking behaviour, Ž
. lesion score and performance of the nine focal pigs per pen only. Turner et al. 1999
presented an analysis of these parameters considering the mean for all members of the pen, but without a detailed account of the findings for differential weight categories of
pigs.
3.1. Ambient temperature The mean daily temperature ranged from 14.3
8C to 18.28C and did not significantly differ between treatments. Replicates, which were allocated sequentially between Jan-
Ž .
Ž .
uary replicate 1 and July replicate 4 , did not significantly influence the drinking patterns described below.
3.2. Drinking behaÕiour of pooled focal pigs The parameters of drinking behaviour, irrespective of weight category or treatment,
were investigated by pooling the data from all focal pigs. The percentage of total daily drinking time occurring during each hour within the 24-h period of video analysis was
Ž .
calculated Fig. 1a . The maximum hourly percentage of total drinking time occurred Ž
. during the hour of 1800–1900 h 10.1 S.E.M. 0.73 of total , and the minimum
Ž .
occurred between 0700 and 0800 h 0.9 S.E.M. 0.28 . The mean number of drinking Ž
. bouts per day was 30.9 S.E.M. 1.41 visitsrpigrday Fig. 1b with a range of 7–98. The
frequency distribution of drinking bout length was skewed with a median of 21.0 s. The mode of this distribution occurred in the range 10 to 14 s, and represented 18.4 of
Ž .
drinking bouts Fig. 1c . The mean total time spent at the drinkers in a 24-h period was Ž
. 832 S.E.M. 43.8 s, with a maximum recorded duration of 2450 s Fig. 1d .
S.P. Turner
et al.
r Applied
Animal Beha
Õ iour
Science 67
2000 321
– 334
326
Ž . Ž .
Fig. 1. Drinking behaviour of pooled focal pigs: a Mean percentage of total daily drinking time occurring during each hour of a 24-h period. b Frequency Ž .
Ž . distribution of number of drinking bouts per pig per day. c Frequency distribution of drinking bout length. d Frequency distribution of total daily drinking time.
3.3. Drinking behaÕiour of separated focal pig weight categories and treatments Heavy, medium and light weight focal pigs visited the drinkers a similar number of
Ž .
times 28.1, 31.1 and 30.3 S.E.D. 3.05 visitsrpigrday, respectively , and there was no interaction between weight category and treatment on the frequency of visits. The mean
Ž drinking bout duration was not significantly affected by weight category 28.1, 24.4 and
. 25.9 S.E.D. 2.70 s for heavy, medium and light weight pigs , nor was there an
interaction with treatment. Consequently, the total time spent drinking per day was Ž
. similar for each weight category 787, 777 and 810 S.E.D. 94.7 srpigrday , and again
there was no interaction between weight category and treatment. With the three weight categories pooled, focal animals in groups of 20 pigs
Ž performed more drinking bouts per day than groups of 60 36.3 vs. 22.5 S.E.D. 2.86
. boutsrpigrday, P - 0.01, respectively, Table 1 . Focal pigs offered one drinker per 10
animals also performed drinking bouts more frequently than those offered one drinker Ž
. per 20 individuals 34.7 vs. 24.2 S.E.D. 2.86 boutsrpigrday, P - 0.05 . No significant
interaction occurred between the two main factors of group size and drinker ratio. A trend was apparent for pigs in groups of 20 to perform longer drinking bouts than those
Ž .
in groups of 60 28.9 vs. 22.9 S.E.D. 2.55 s, P - 0.1 . Consequently, those in groups of 20 spent a significantly longer duration at the drinkers each day than animals in groups
Ž .
of 60 1055 vs. 517 S.E.D. 86.9 srpigrday, P s 0.001 . Those animals offered one drinker per 10 pigs also spent more time at the drinkers each day than pigs offered one
Ž .
drinker per 20 933 vs. 638 S.E.D. 86.9 srpigrday, P - 0.05 . An interaction between Ž
group size and drinker ratio on daily time at the drinkers was apparent 490, 787, 544 and 1323 S.E.M. 94.2 srpigrday for 60 pigs, three drinkers; 20 pigs, one drinker; 60
. pigs, six drinkers and 20 pigs, two drinkers, respectively, P - 0.05 .
When the focal animal weight categories and treatments were pooled, drinking bout Ž
length and total daily drinking time were significantly and positively correlated r s 0.79, .
P - 0.01 as were the daily number of visits to the drinker and total daily drinking time Ž
. r s 0.87, P - 0.001 . Bout length was not significantly correlated with the daily
Ž .
number of visits to the drinker r s 0.39 . However, considering these relationships
within individual pens, but again with weight category pooled, the correlation between bout length and total daily drinking time reached significance in only three pens. The
relationship between daily number of drinking visits and total drinking time reached
Table 1 Drinking behaviour of the pooled focal pigs in each drinker ratio and group size combination
Treatment S.E.M. Significance
60 pigs, 20 pigs,
60 pigs, 20 pigs,
Drinkers Group Drinkers= 3 drinkers 1 drinker 6 drinkers 2 drinkers
size group size
UU U
Daily number of drinking 18.6
29.8 26.5
42.8 3.10
Ž .
bouts boutsrpigrday Ž .
Mean bout length s 24.8
26.4 21.0
31.4 2.76
P - 0.1
U U
Total daily drinking 490
787 544
1323 94.2
P - 0.1 Ž
. time srpigrday
significance in 10 pens and that between bout length and daily number of drinking visits reached significance in four pens, although three of these relationships occurred in the
treatment of 20 pigs with two drinkers.
3.4. Diurnal pattern of drinker use The 24 h period of video analysis of each pen was divided into six blocks of 4 h.
When the percentage of the total daily drinking time occurring during each block was plotted, no notable effect of treatment on the diurnal pattern of drinking in any focal pig
weight category was apparent. Fig. 2, therefore, shows the diurnal pattern when the four treatments were pooled. The period during which the greatest percentage of the total
daily drinking time occurred was between 1700 and 2059 h. The percentage of the total daily drinking time occurring during this period was not significantly different between
Ž treatments 29.4, 27.4, 31.3 and 28.8 S.E.M. 5.29 for 60 pigs, three drinkers; 20 pigs,
. one drinker; 60 pigs, six drinkers and 20 pigs, two drinkers, respectively, P 0.05 . The
period during which the smallest percentage of total daily drinking time occurred was between 0500 and 0859 h, and the percentage of daily drinking time occurring during
Ž this period was also similar for each treatment 9.7, 3.2, 7.4 and 6.6 S.E.M. 1.65,
. respectively, P 0.05 . When treatments were pooled, focal pigs of different weight
category did not significantly differ in the percentage of total drinking time which Ž
occurred during the maximum 29.2, 25.8 and 32.8 S.E.D. 3.99 for heavy, medium .
Ž and light weight pigs, respectively or minimum drinking periods 8.5, 7.3 and 4.9
. S.E.D. 1.86, respectively . Focal pig weight category did not interact with treatment to
significantly influence the percentage of total drinking time occurring during either of these periods. Mean drinking bout length was similar between the two periods and was
not significantly affected by weight category.
Fig. 2. Percentage of total daily drinking time occurring during 4-h blocks for each weight category when treatment was pooled.
3.5. Aggression Analysis of the lesion scores for each body area separately showed no significant
influences of treatment. The total lesion score for the whole body was therefore used in Ž
. the analyses below. Turner et al. 1999 presented the effects of group size and drinker
allocation on the pooled focal pig lesion score. Here the lesion score data for specific weight categories of pig are described. Heavy weight pigs had significantly more lesions
Ž than light weight animals over the whole trial period 13.2, 10.8 and 8.8 S.E.D. 1.18
. lesionsrpig for heavy, medium and light weight categories, P - 0.005 . By 3 days
Ž .
post-mixing, this pattern was evident P - 0.05 . Treatment and weight category did not significantly interact to affect the lesion score during either the whole period, or at 3
days post-mixing alone. When treatments were pooled, the lesion score peaked at 3 days Ž
. post-mixing, significantly P - 0.001 declined to half this value by the end of the first
week and reached a plateau by the end of the second week. This pattern varied little Ž
. between weight categories Fig. 3 .
The mean of the five lesion score recordings over time per pig did not significantly Ž
. Ž
. correlate with the daily number of drinker visits r s 0.24 , mean bout length r s 0.02
Ž .
and total daily drinking time r s 0.18 . 3.6. Performance
Considering the focal pigs, the heavy individuals had a higher ADG than the medium Ž
and light weight animals 0.73, 0.65 and 0.66 S.E.D. 0.030 kgrday, respectively, .
P - 0.05 . The ADG of the focal pigs was investigated for a possible detrimental effect on the light weight animals. No such interaction between treatment and weight category
Ž occurred, and the ADG of the pooled focal pigs was similar for each treatment Table
. 2 . When considering all the animals within the pen, pigs beginning the experiment with
an upper inter-quartile start weight had a higher ADG than those in the lower quartile, Ž
. irrespective of treatment 0.69 vs. 0.62 S.E.D. 0.027 kgrday, P - 0.01 , although this
Fig. 3. Development of mean lesion score over time for each of the three weight categories of focal pig.
Table 2 Ž
. ADG
kgrpigrday of focal pigs and, considering all members of the pen, pigs of lower and upper inter-quartile start weight in the four combinations of group size and drinker allocation
Treatment S.E.M.
Significance 60 pigs,
20 pigs, 60 pigs,
20 pigs, Drinkers
Group Drinkers=
3 drinkers 1 drinker
6 drinkers 2 drinkers
size group size
Pooled focal pigs 0.70
0.70 0.65
0.67 0.024
U
Lower quartile 0.53
0.68 0.68
0.58 0.037
UU
Upper quartile 0.63
0.80 0.70
0.64 0.037
pattern only reached significance after 2 weeks on trial. The lower quartile animals Ž
. gained most slowly in a group of 60 with three drinkers P - 0.05 . This pattern was
Ž .
repeated by the animals with an upper inter-quartile start weight P - 0.01 , and the
interaction between start weight and treatment was not significant.
4. Discussion