Therefore, it is unclear to which extent variable management routines on farms affect the welfare of the animals, and whether the effect on welfare is positive or negative. The
Ž . present study addresses two questions: 1 does an unpredictable feeding schedule affect
Ž . the behaviour, production or health of dairy calves and 2 do animals that are used to a
predictable feeding schedule react otherwise to occasional deviations in this schedule, than animals that are fed according to an unpredictable schedule.
2. Materials and methods
2.1. Housing The calves were kept in two identical units in the same building complex. Room
temperature was kept at 208C and the lights were turned off from 1500 to 0700 except for a 30-min inspection period at 2200. When the behaviour was recorded during the
nights, the lights were left on. Windows in the walls and roof provided natural day light. Ž
. The calves were kept in individual pens 0.85 = 1.85 m with concrete floor covered
with 5–15 cm of straw. New straw was added daily. 2.2. Animals
Thirty six Holstein–Friesian calves of mixed sexes were used in the experiment, but only 24 calves were subjected to behavioural recordings. Those were randomly selected
within each treatment. The calves were removed from their mothers immediately after calving. After 4–5 days with colostrum feeding, they were bucket fed with 2 l of milk
twice a day. The amount was gradually increased to a maximum of 2 = 4.5 l of milk. All calves had free access to hay, concentrate, and water. To minimise the age variation
within groups, the behavioural observations were conducted in two parallel trials, with 12 calves — four from each treatment — in each trial. The first trial ran two weeks
earlier than the second. Data from the two trials were pooled for statistical analysis.
2.3. Treatments Ž
. On day 5, the calves were designated to one of three treatments C, OD and IR .
Group C served as a control and received milk on the same time every day: at 0710 and Ž
. 1430. Group OD occasional deviation was treated the same way except on 2 days
Ž .
later referred to as ‘‘observation days’’ when they were 5 and 8 weeks old. Here, they Ž
. were fed both meals 3 h later than usual. Group IR irregular feeding schedule received
milk according to a feeding schedule with stratified random feeding times. The first time at 0600, 0700, 0800, 1000, 1100 or 1300 and the second time at 1300, 1400, 1500,
1730, 1800 or 2100. The calves were assigned to the groups according to age Žmean SD age at first behaviour recording: C: 34 4 days; OD:33 5 days; IR:
. Ž
. Ž
35 5 days , birth weight C: 43 9 kg; OD: 44 7 kg; IR: 43 6 kg , sex 7 out of .
12 males in each group , and sire. The minimum interval between the first and the second feeding was set to 4 h. The calves had access to the milk buckets for 15 min.
2.4. Recordings 2.4.1. BehaÕiour
The behaviour of the calves was recorded with VHS time-lapse equipment. For each of the two trials, four calves per treatment were recorded simultaneously, giving a total
sample of 24 animals. The recordings were made when the calves averaged 5 and 8 weeks old for 48 h each time. In the subsequent analysis, the first 24 h are referred to as
‘‘control day’’ and the second 24 h as ‘‘treatment day’’ but the whole 48-h period is referred to as ‘‘observation days’’. The observation period of 2 = 24 h was found to be a
sufficient sample to represent the true behaviour of the calves, especially as the eight calves in each treatment were not filmed the same day. For practical reasons, the
recordings started at 2200 the day before the control day and ended at 2200 the treatment day. Therefore, for each treatment video recordings were analysed for eight
calves; 48 h at 5 weeks of age and 48 h at 8 weeks of age.
The following behaviours were recorded: Ø
Periods of lying and standing. Ž
. Ø
Licking or sucking the interior of the pen frequency per hour . Ž
. Ø
Licking or sucking other calves frequency per hour . Ž
. Ø
Tongue playing frequency per hour . Ž
. Ø
Comfort behaviour, defined as scratching or licking own body frequency per hour . Ž
. Ø
Eating concentrate or hay frequency per hour . Ž
. Ø
Drinking water frequency per hour . Ž
. Ž Ø
Extending the head through the feeding barrier HTB-behaviour frequency per
. hour .
HTB behaviour is, in contrast to the other behaviours, not commonly used in behavioural research but prior observations showed that this behaviour is very common
prior to feeding in our housing system. We suggest it can be used to indicate the calves’ expectations for feeding.
2.4.2. Production and health All the calves were weighed at birth and weekly thereafter. Milk consumption was
measured at every feeding and concentrate and hay consumption was measured weekly. Ž
. On the basis of these recordings ‘‘mean daily feed intake’’ SFUrday , ‘‘growth rate’’
Ž .
Ž .
grday and ‘‘feed conversion rate’’ SFUrkg growth were calculated for each calf. ŽSFU stands for Scandinavian Feeding Units which roughly equals one kg grain; see
. Andersen and Foldager, 1980 for details . All incidences of disease were recorded and,
if necessary, treated by a veterinarian. 2.5. Statistics
2.5.1. BehaÕiour The frequency of licking or sucking the interior of the pen, other calves and tongue
playing proved to be very low and not suitable for statistical analysis. Consequently, those data were excluded.
Ž .
The frequencies for each of the 24-h periods control day and treatment day of comfort behaviour, eating, drinking, and HTB-behaviour were calculated for each calf.
The averages of number of lying periods, length of lying period and total lying time were also calculated individually for the calves. These parameters were tested for normal
Ž .
distribution using PROC UNIVARIATE SAS institute, 1987 . Differences between treatments were tested by the model:
Y s m q a T q a G q e ;
i j 1 i
2 j
i j
Ž where: Y s the response variable frequency of comfort behaviour, eating, drinking,
i j
. Ž
. and HTB-behaviour for trial i, group j; m s the general mean; T s trial, i s 1,2 ;
i
Ž .
Ž .
G s group treatment effect, j s C,OD,IR ; e s random error.
j i j
The analysis for difference between treatments were made separately for each combination of age and observation days to prevent errors caused by repeated measure-
ments of the same individuals. Residuals were tested for normality. To evaluate the effect of observation days, the results for each animal were paired and the difference
between the treatment day and the control day was tested with a standard t-test for
Ž .
paired observations H0: m s 0 . This was done separately for each treatment.
D
2.5.2. Production and health Ž
. Ž
. The variables ‘‘mean daily feed intake’’ FUrday , ‘‘growth rate’’ grday and
Ž .
‘‘feed conversion rate’’ FUrkg growth and frequencies of diarrhoea and pneumonia were tested in the following model:
Y s m q a T q a G q a B e
;
i jk 1 i
2 j
3 k
i jk
Ž where: Y
s the response variable mean daily feed intake, growth rate, feed conversion
i jk
. Ž
. rate for trial i, group j and birth weight k; m s the general mean; T s trial, i s 1,2 ;
i
Ž .
Ž .
G s treatment group effect, j s C,OD,IR ; B s birth weight; e s random error.
j k
i jk
The model was subjected to analysis of variance to test the effects of treatments, controlling for trial effect. The residuals were tested for normality. This was done by
Ž .
PROC GLM and PROC UNIVARIATE SAS Institute, 1987 .
3. Results