- Email: [email protected]
Effect of feeding duration and rumen fill on behaviour in dairy cows
Applied Animal Behaviour Science 70 (2000) 83±97
Effect of feeding duration and rumen ®ll on behaviour in dairy cows Tina LindstroÈm*, Ingrid Redbo Department of Animal Nutrition and Management, KungsaÈngen Research Centre, Swedish University of Agricultural Sciences, S-753 23 Uppsala, Sweden Accepted 27 April 2000
Abstract The aim of the present experiment was to test the hypothesis that oral manipulation of feed is a behavioural need in cattle, irrespective of actual rumen load. Twelve rumen ®stulated cows were used and subjected to four different treatments: low rumen contentlong duration of eating (A), high rumen contentshort duration of eating (B), high rumen contentlong duration of eating (positive control) (C) and low rumen contentshort duration of eating (negative control) (D). To obtain treatment A and B, rumen content was transferred by hand from cow A to B through the rumen ®stulaes. Each treatment lasted for 3 days with 2 weeks of recovery between each new treatment. The experiment was repeated twice during two consecutive years. All cows were fed the same mixture of silage, concentrate and hay. The cows were videotaped under normal conditions (24 h), and on the third day of the experiment. From these videotapes, the behaviours (frequency and duration per 24 h) have been analysed. Time spent eating differed between the four treatments (P<0.001), with shortest eating-times in B and D. The cows with low rumen content (A and D) spent shorter time ruminating (P<0.001) than the cows with ®lled rumen (B and C). The B and D cows (short duration of eating) spent longer time (P<0.001) with behaviours related to feed-searching than the cows with long duration of eating (A and C). The C cows had fewer (P<0.001) bouts of behaviours related to feed-searching than the A, B and D cows. Time spent with stereotypies (tongue-rolling) was longer (P<0.01) in D than in the other treatments. There was a difference (P<0.001) between treatments in eating bouts. The A cows had more (P<0.05) eating bouts than the cows in B, C and D. The cows with low rumen content (A and D) had fewer ruminating bouts (P<0.001) than the cows with ®lled rumens (B and C). The number of bouts with stereotypies differed (P<0.01), the cows in D having the highest ®gures compared with all the other treatments. In conclusion, our results support the hypothesis that oral manipulation of feed is a behavioural need in cattle irrespective of rumen load. A low duration of feeding behaviours combined with a
* Corresponding author. Tel.: 46-1867-2699; fax: 46-1867-2946. E-mail address: [email protected] (T. LindstroÈm).
0168-1591/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 1 5 9 1 ( 0 0 ) 0 0 1 4 8 - 9
84
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
low rumen load, which is a common practice in certain categories of growing cattle or dry dairy cows, seriously impairs the welfare in cattle. # 2000 Elsevier Science B.V. All rights reserved. Keywords: Cattle-feeding and nutrition; Feeding behaviour; Stereotypies; Behavioural need
1. Introduction The feeding of dairy cows during the stable period is often accomplished within a rather short time, compared with the duration of eating and rumination among cows on pasture. A study of heifers (Redbo and Nordblad, 1997) showed that both the development as well as the performance time of stereotypies were related to the duration of feeding behaviours. Corresponding results have been found in dairy cows where restricted feeding may lead to signi®cant increases in behavioural disturbances, such as stereotyped oral behaviours, compared with giving feed ad libitum (Redbo et al., 1996). This has also been found in studies of other species. Rushen (1985) found in an early study that the occurrence of stereotypies was strongly associated with feeding periods in tethered sows. In a study with pregnant gilts and sows (Robert et al., 1993), bulky feeding regimes reduced stereotypies over 24 h and hardly any oral behaviours were performed by the gilts fed ad libitum. A study of dry sows (Brouns et al., 1994) showed that the occurrence of stereotypies was related to a combination of a lack of suf®cient amount of food to produce satiety and to a frustration of feeding/foraging behaviour. In hens, sterotyped pacing was found by Duncan and Woodgush (1972) to be associated with long-term and intense feeding frustration. In this investigation, hens exposed to a short duration of feed-deprivation responded to a frustrated feeding situation with displacement preening, while birds strongly motivated to feed due to a long deprivation time developed stereotyped pacing. In another study with caged and feed-deprivated pigeons, the stereotypy spot-pecking disappeared completely when the birds got free access to food (Palya and Zacny, 1980). Even though the cows physiological needs are most often ful®lled in modern dairy systems, studies have indicated that the behavioural, or psychological, need of heifers and dairy cows for the actual performance of feeding behaviours may be unful®lled (e.g. Redbo, 1990; Redbo et al., 1996; Redbo and Nordblad, 1997). A behavioural need has, according to Jensen and Toates (1993), been used to describe ``the need to perform a speci®c behaviour pattern whatever the environment is like and even if the physiological needs which the behaviour serves are ful®lled''. Jensen and Toates argue that ``animals do possess needs with regards to their behaviour, i.e. prevention of performance of speciesspeci®c behaviour may cause signs of reduced welfare''. All this raises the question if there are two different `kinds' of hunger Ð one that is more physiologically determined, and one more linked to a psychological need for the performance of the oral manipulation of feed. However, there is no clear distinction between these two systems since they are both parts of the complicated hunger-satiety system. Although the very duration of oral feed manipulation obviously is important for the wellbeing of cattle, we do not know how important a long feeding duration is in relation to physical ful®lment. It is likely that species evolved for long feeding sessions, such as cattle,
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
85
need a long sensory and oral stimulation by the act when feed is passing through the mouth in order to receive necessary negative feedback on feeding motivation. It has been found that behaviours related to feed-searching, such as nosing and licking in or around the empty feed trough or feeding site, adjoins a stereotypy bout signi®cantly more often than expected (Redbo, 1992a). This relation between behaviours related to feed-searching and stereotypies may indicate that these stereotypies can be considered as displacement behaviours and that the switch from behaviours related to feed-searching to stereotypies occurs as a consequence of the frustration of the former behaviour, i.e. the motivation to ingest feed (Redbo, 1992b). Thus, both stereotyped oral behaviours as well as behaviours related to feed-searching are interesting as markers for frustrated feeding behaviours in cattle. To be able to correctly interpret the effect of a long or a short eating duration on the well-being of the cow, the effect of oral feed intake must be separated from the effect of rumen load. It is possible to achieve such a separation by the use of cows with rumen ®stulaes through which already swallowed feed can be extracted or rumen content from a donor cow can be inserted. In this study, we have used such cows. For readers not familiar with rumen ®stulaes in cows, we must explain that the experimental cows all are treated as normal dairy cows, i.e. they calve once a year, spend the summer on pasture, etc. The aim of the present experiment was to test the hypothesis that oral manipulation of feed is a behavioural need in cattle, irrespective of actual rumen load. The objective of this study was to examine how feeding duration and rumen ®ll, respectively, would in¯uence behavioural variables possibly re¯ecting frustrated feeding motivation. 2. Materials and methods 2.1. Animals and study design Twelve rumen-®stulated, lactating, cows of the Swedish Red and White Cattle Breed with ages ranging from 3 to 10 years, were used and subjected to four different treatments:
A: low rumen contentlong duration of eating, B: high rumen contentshort duration of eating, C: high rumen contentlong duration of eating (positive control), D: low rumen contentshort duration of eating (negative control).
Each animal passed through every treatment but in different orders. Each treatment lasted for 3 days, followed by a 2-week recovery period when the cows were kept in accordance with ordinary management and feeding routines, before a new 3-day experimental period started. The experiment was repeated twice during two consecutive years. All cows were housed in the same stable in individual stalls with an empty stall or a solid partition in between them. All the cows included in the experiment were continuously controlled for symptoms of illness or ailments in addition to body temperature control. All cows were healthy and in good shape during the experimental days both years, except one cow that had to be culled the ®rst year, due to mastitis which was not caused by the experiment.
86
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
2.2. Feeding All cows were fed a mixture consisting of 60% grass clover silage, 5% grass hay and 35% concentrate in individual amounts calculated according to their individual milk production. The energy concentration in the mixture varied during the 2 years between 11.7 and 12.0 MJ metabolizable energy per kg dry matter. One day before the experiment started the feed was restricted to 75% of normal individual feed amount for all four treatments. The cows were given food individually in troughs twice a day at 06.35 and 13.50 h year 1, and at 05.55 and 14.20 h year 2. Amounts of left-overs were recorded during the whole experimental period. All cows had free access to water in individual bowls and free access to salt lick. To obtain treatments A and B we transferred rumen content by hand from cow A to B through the rumen ®stulaes. The rumen content was always taken from the upper layer in the same area in the centre of the rumen of cow A. The rumen content was weighed continuously during every transfer. Pilot studies had shown that rumen content derived from the present feed mixture and taken from that area, had a DM content of 17%. Cows in treatment A were offered 150% of normal feed amount to eat orally (in principle ad libitum) but rumen content was regulated to stay at 75% of normal feed allotment. Cows in treatment B were offered 50% of normal feed amount to be eaten orally, while 50% (in the form of rumen content from cows in A) was ®lled directly into the rumen. Cows in C received 150% of normal amounts, while cows in D only received 50% to eat. Neither treatment C or D involved any transfer of rumen content, the complete feed amounts were eaten orally. Transfer of rumen content from cow A to B was made three times a day (07.00, 10.00 and 13.00 h) and at 16.00 h rumen content was extracted from the rumen from cow A without passing it to cow B. The aim of this control of rumen ®ll was to standardise the effects of the treatments, i.e. to control that the cow had a ®lled or non-®lled rumen, respectively. This was made to separate the physiological effects of feed intake that originates from the rumen itself, from the effects of the oral manipulation of the feed. 2.3. Behavioural observations The cows were videotaped both under normal conditions (24 h) 2 days before the onset of the treatments, and on the third day of the experiment, in all four treatments. From these videotapes the following behaviours have been analysed according to their total frequency and duration per 24 h: eating Ð when the cow was obtaining food from the trough or from the ¯oor in her mouth or when chewing or swallowing food; ruminating Ð chewing or regurgitating already swallowed food; behaviours related to feed-searching Ð nosing or licking the empty feed-trough, or on the ¯oor, or ®ttings close of the trough; licking salt Ð licking on the salt lick; stereotypy Ð when the cow consistently rolled its tongue outside or inside her mouth or chewed on a bar in an unvaried, repeatedly and stereotyped manner; total standing Ð when the cow was standing up on all four legs irrespective of any other activities; total lying Ð when the cow was lying irrespective of other activities; other active Ð included all other behaviours except eating, ruminating, behaviours related to feedsearching and stereotypy.
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
87
Behavioural recordings from the `normal day' and from treatments C and D were only performed in year 1. In year 2 the analyses only concerned the two treatments A and B. 2.4. Statistical analyses The analyses of behavioural data were performed using Minitab (1996) and Systat (1998). All behavioural variables were tested for normality and variables not normally distributed were logarithmically transformed. In most cases, the transformations did not improve the distribution and, thus, non-parametric statistics were used for these variables. Differences between treatments (year 1) in duration of normally distributed behaviours (eating, ruminating, total standing and total lying) were tested with one-way Anovas followed by Tukeys tests. Differences between treatments in duration of behaviours related to feed-searching, licking salt and stereotypies (not normally distributed behaviours) were tested using non-parametric statistics (Kruskal±Wallis k-sample test). This test was followed by a non-parametric Tukey-type multiple comparison (Zar, 1984). Differences between treatments (year 1) in frequency of normally distributed behaviours (ruminating, total standing and total lying) were tested with Anovas followed by Tukeys tests. Differences between treatments in frequency of behaviours related to feed-searching, licking salt and stereotypies (not normally distributed behaviours) were tested using nonparametric statistics (Kruskal±Wallis k-sample test). The ®gures for frequencies of bouts of eating were transformed (log) and thereafter analysed with Anovas followed by Tukeys tests. Differences (year 2) between treatments A and B in duration of normally distributed behaviours (eating, ruminating, total standing and total lying) were tested with the Two Sample T-tests. Differences between A and B in duration of behaviours that were not normally distributed (behaviours related to feed-searching, licking salt, and stereotypies) were tested with the Mann±Whitney test. Differences (year 2) between A and B in frequencies of normally distributed behaviours (ruminating, total standing and total lying) were tested with Two Sample T-test. Differences between A and B in frequencies of behaviours that were not normally distributed (eating, behaviours related to feed-searching, licking salt, and stereotypies) were tested with Mann±Whitney test. Differences between normal day and experimental day in duration were tested with the Paired Samples T-test for normally distributed behaviours (eating, ruminating, total standing and total lying). For behaviours related to feed-searching, licking salt and stereotypies (not normally distributed) the differences were tested with the Wilcoxon Signed Ranks test. Differences between normal day and experimental day in frequency were tested with the Paired Samples T-test for normally distributed behaviours (ruminating, total standing and total lying). For eating, behaviours related to feed-searching, licking salt and stereotypies (not normally distributed) the differences were tested with the Wilcoxon Signed Ranks test.
88
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
3. Results The experiments were accomplished during two consecutive years. Since the results in duration of rumination in treatment A differed signi®cantly between years, the two experimental periods could not be treated as replicates. Consequently, the data from the 2 years have been treated as two similar experiments and are analysed separately. Even though the magnitude of the ®gures for duration of rumination in treatment A differ between years, a comparison between treatments A and B clearly show the same pattern both years, with high ®gures for treatment B but low for treatment A. 3.1. Feed consumption The oral feed consumption in year 1 was, on average, 25.8 kg dry matter (DM) in treatment A, 9.1 kg DM in B, 24.7 kg DM in C and 8.5 kg DM in D on the third day of experiment. In year 2 the oral consumption was 24.5 kg DM in A, 9.1 kg DM in B, 22.1 kg DM in C and 8.8 kg DM in D on the third day of experiment. In addition to the oral intake, the cows in treatment B, both year 1 and 2, received on average 9.1 kg DM per day as rumen content via the rumen ®stulaes. 3.2. Duration of certain behaviours Ð year 1 There was a signi®cant (P<0.001) difference in time spent eating between the four treatments, with shortest eating-times in treatments B and D (Table 1). When it comes to time spent ruminating there was a signi®cant (P<0.001) difference between the treatments with low rumen content (A and D) compared with the treatments with ®lled rumen (B and C). The cows in treatments B and D (with short duration of eating) spent signi®cantly (P<0.001) longer time with behaviours related to feed-searching than cows with long duration of eating (treatments A and C). The cows in treatment A spent signi®cantly Table 1 Duration of certain behaviours in percent of total observation time (24 h), year 1 experimental day 3, in the four different treatments (A, B, C and D)a Behaviour
Treatment A
Eating Rumination Behaviour related to feed-searching Licking salt Stereotypies Total standing Total lying
B a
23.2 (6.5) 20.3a (6.3) 1.5a (0.8) 1.4 4.4a (3.3) 4.9 0.3a (0.5) 0.0 51.1 (9.5) 48.9 (9.5)
C b
4.4 (1.8) 31.7b (4.4) 3.2b (1.0) 3.1 0.4b (0.4) 0.3 1.8a (2.2) 0.6 45.8 (8.8) 54.2 (8.8)
D a
22.4 (4.0) 37.8c (4.7) 1.3a (1.8) 0.3 0.7b(0.8) 0.5 0.2a (0.3) 0.0 51.1 (10.9) 48.9 (10.9)
7.7b (5.7) 15.9a (4.9) 3.4b (1.5) 2.7 1.6b (1.7) 1.4 8.6b (11.0) 1.5 44.7 (8.0) 55.3 (8.0)
a Means with different superscripts within a variable row are significantly different at (P<0.05). Figures within brackets are standard deviations. Figures in italics are medians.
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
89
Table 2 Frequencies (bouts) of certain behaviours during observation time (24 h), year 1, experimental day 3 in the four different treatments (A, B, C and D)a Behaviour
Treatment A
Eating Rumination Behaviour related to feed-searching Licking salt Stereotypies Total standing Total lying
B a
61.0 (30.3) 51.0 11.4a (3.2) 103.7a (46.7) 113.0 126.0a (68.5) 121.0 11.4a (23.8) 0.0 10.3 (2.7) 10.0 (2.6)
C b
4.9 (4.7) 3.0 20.5b (3.5) 160.1a (63.0) 151.0 10.9b (18.0) 7.0 14.0a (16.6) 7.0 11.6 (4.3) 11.3 (4.8)
D c
21.9 (8.0) 24.0 25.2b (6.9) 18.3b (15.7) 11.0 11.7b (12.3) 8.0 3.6a (8.0) 0.0 14.3 (6.2) 14.3 (6.2)
14.6c (5.3) 13.0 11.8a (2.7) 87.5a (42.0) 85.0 26.8b (21.8) 26.0 34.9b (33.0) 17.0 10.2 (2.4) 9.4 (2.6)
a Means with different superscripts within a variable row are significantly different at (P<0.05). Figures in brackets are standard deviations. Figures in italics are medians.
(P<0.01) longer time licking salt than cows in the other three treatments. Time spent with stereotypies was signi®cantly (P<0.01) longer in treatment D than in the other treatments. There were no differences between treatments either in duration of total standing, or in duration of total lying. 3.3. Frequencies of behaviours Ð year 1 There was a signi®cant (P<0.001) difference between treatments in the frequency of eating bouts. The cows in treatment A showed a signi®cantly (P<0.05) higher number of eating bouts than the cows in treatments B and C as well as D (Table 2). There were no differences in the number of eating bouts in treatment C compared with treatment D. The cows with low rumen content (treatments A and D) had signi®cantly (P<0.001) fewer ruminating bouts than the cows with ®lled rumens (treatments B and C). When it comes to behaviours related to feed-searching there was a signi®cant (P<0.001) difference between cows in treatment C and the other three treatments that showed a larger number of bouts. The cows in treatment A licked the salt lick signi®cantly (P<0.001) more often than the cows in the other treatments. There was a signi®cant (P<0.01) difference in the number of bouts with stereotypies Ð the cows in treatment D had the highest ®gures compared with all the other treatments. There were no differences between treatments in number of bouts of total standing or total lying. 3.4. Duration of certain behaviours Ð year 2 There was a signi®cant difference between treatments A and B in duration of eating, with higher ®gures for cows in treatment A (Table 3). Time spent with rumination showed signi®cantly higher ®gures for cows in treatment B compared with cows in A. The cows with short eating-time (B) spent signi®cantly more time with behaviours related to feed-
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
90
Table 3 Duration of certain behaviours in percent of total observation time (24 h), and frequencies (bouts), year 2, experimental day 3a Behaviour
Year 2 Duration
Frequency
A
B
Significance
A
B
Significance
Eating
26.3 (6.2)
6.0 (4.7)
P<0.001
27.6 (5.9) 1.7 (1.3)
33.7 (4.9) 3.1 (1.7)
P<0.05 P<0.05
9.5 (8.7) 6.0 20.4 (5.9) 58.0 (30.2)
P<0.001
Rumination Behaviour related to feed-searching
76.5 (42.4) 66.0 13.8 (3.3) 37.4 (36.5)
1.2 6.3 (2.3) 5.5 0.5 (0.7) 0.2 49.9 (6.2) 50.1 (6.1)
3.1 0.6 (0.6) 0.5 3.7 (4.4) 0.7 45.9 (9.1) 54.1 (9.1)
30.0 132.6 (66.9) 114.0 14.6 (21.7) 4.5 11.4 (2.8) 10.4 (2.8)
53.0 11.4 11.5 19.0 12.5 12.1 11.1
Licking salt Stereotypies Total standing Total lying a
P<0.001 n.s. n.s. n.s.
P<0.01 P<0.05
(10.0)
P<0.001
(24.9)
n.s.
(5.1) (5.1)
n.s. n.s.
Figures in brackets are standard deviations and figures in italics are medians.
searching than cows with a long eating-time (A). Even this year the cows in treatment A spent signi®cantly more time licking the salt lick compared with the cows in treatment B. There were no differences between treatments in time spent with stereotypies, total standing or total lying. 3.5. Frequencies of behaviours Ð year 2 There was a signi®cant difference in number of eating bouts between treatments A and B (Table 3) with higher ®gures for the A cows. The number of ruminating bouts showed signi®cantly higher ®gures for the B cows than for the A cows. The cows in treatment B showed signi®cantly higher frequencies of behaviours related to feed-searching than the cows in treatment A. The number of bouts of licking the salt lick was signi®cantly higher in treatment A than in treatment B. There were no signi®cant differences between treatments in numbers of stereotypy bouts or numbers of bouts of total standing or total lying. 3.6. Normal day compared with experimental day 3 Ð duration In treatment A, there were signi®cantly higher ®gures in duration of eating, licking salt and total time spent standing (P<0.01) on experimental day 3 compared with normal day (Fig. 1). Duration of rumination, behaviours related to feed-searching and stereotypies showed signi®cantly lower ®gures on the experimental day compared with a normal day in treatment A (Fig. 1). In treatment B, the cows spent signi®cantly less time with eating on the experimental day compared with a normal day, but signi®cantly more time with rumination and behaviours related to feed-searching (Fig. 2). There were no signi®cant differences in time spent
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
91
Fig. 1. Duration of certain behaviours in percent of total observation time (24 h) year 1, normal day and experimental day 3 in treatment A. Figures above the bars show their standard deviations.
Fig. 2. Duration of certain behaviours in percent of total observation time (24 h) year 1, normal day (N) and experimental day 3, treatment B. Figures above the bars show their standard deviations.
92
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
Fig. 3. Frequencies of certain behaviours during total observation time (24 h) year 1, normal day and experimental day 3, treatment A. Figures above the bars show their standard deviations.
licking salt, stereotypies, total standing or total lying on a normal day compared with experimental day in treatment B. 3.7. Normal day compared with experimental day 3 Ð frequencies In treatment A, the cows had signi®cantly more bouts of eating and licking salt on the experimental day than on a normal day (Fig. 3). The number of bouts of ruminating was
Fig. 4. Frequencies of certain behaviours during total observation time (24 h) year 1, normal day and experimental day 3, treatment B. Figures above the bars show their standard deviations.
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
93
signi®cantly lower on the experimental day than on a normal day in treatment A. There were no signi®cant differences in frequencies of behaviours related to feed-searching, stereotypies, total standing or total lying. In treatment B, the cows showed a signi®cantly higher frequency of eating on a normal day than on an experimental day (Fig. 4). On the experimental day, the number of bouts was signi®cantly higher for rumination and behaviours related to feed-searching than on a normal day. There were no signi®cant differences in treatment B between a normal day and an experimental day in frequencies of licking salt, stereotypies, total standing or total lying. 4. Discussion In the present study, we have found that a long duration of oral manipulation of feed by eating and ruminating lowered the levels of stereotypies and behaviours related to feedsearching in lactating dairy cows irrespective of their rumen load. This indicates that cattle have a behavioural need to perform feeding behaviours. The cows with free access to feed in the present study (treatments A and C) showed three to four times longer duration of eating (about 5.5 h of 24 h) than the cows in the other two treatments. This was expected, since this was the very aim of the treatments. This is in line with other studies, such as Senn et al. (1995) reported that cows in a loose-housing system fed ad libitum hay, corn silage, and grass silage spent about 5.5 h eating. Freer and Campling (1965) offered hay ad libitum to cows, which resulted in a mean eating duration of 4.23 h, and results from Freer et al. (1962) showed that cows fed ad libitum hay ate 4 h per 24 h. In our study, the B and D cows were able to consume all the feed at once due to their restricted feeding, and this gave them rather short eating times (about 1 and almost 2 h, respectively). Such low ®gures of eating duration are usually seen when cattle are fed high levels of concentrate and low levels of roughages (Gonyou and Stricklin, 1981). In the present study, all cows in all treatments started to eat at the same time as the feed was delivered in the mornings and in the afternoons, meaning that the addition of fresh feed to the trough was an important stimulus for the cows to initiate eating. Campling and Balch (1961) reported that the eating time increased from 3±4 h to 6.5±8 h when swallowed food was removed during the ®rst 3 h of eating in rumen ®stulated cows. In another study (Carr and Jacobson, 1967), 1.4 kg DM was removed from the rumen daily 3 h after feeding and this resulted in a signi®cant increase in voluntary intake of 0.5 kg DM. This indicates that the amount of digesta in the reticulo-rumen is one of the factors that governs the feed intake in cattle. On the other hand, Gordon (1963) made some experiments with rumen content transfer in sheep and stated that the quantity of food in the alimentary tract below the reticulum and rumen (i.e. the omasum, the abomasum and the intestine) in¯uences the quantity of food a sheep will eat to a greater extent than the quantity of food in the rumen itself. In our study, the A and C cows spent approximately the same time eating, even though rumen content of the A cow was continuously extracted, which might indicate that rumen ®ll is not the only factor that governs feed intake in cattle. Our results are in accordance with several earlier studies indicating that receptors in the buccal cavity and throat also are important to encourage further feeding or to cease feeding (see e.g. Forbes, 1995). The time spent feeding found in A and C cows is almost on the level
94
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
as for freely grazing cattle (Hafez and Bouisson, 1975) and might be suf®ciently long to give negative feed-back on motivation for eating. Time spent ruminating depends mainly upon the characteristics of the roughage and the rate at which the roughage is eaten (Freer et al., 1962). Cows fed long hay ad libitum ruminated for 9.14 h (Freer and Campling, 1965), which is in accordance with our ®ndings in treatment C, which had the longest duration of rumination (9.40 h). Cows on restricted feeding (treatment D) had rather short duration of ruminating (3.49 h), which was statistically similar to the A cows even though the A cows were fed ad libitum and, thus, had much longer eating times. The relatively low ®gures for rumination in A are, of course, explained by the fact that rumen content was continually removed from the rumen of A. The relatively long ruminating times in B are due to the circumstance that the B cows had to ruminate food eaten but not ruminated by the A cows. Ad libitum feeding of cows generally leads to many eating occasions during the 24 h (Balch, 1971). In a study by Senn et al. (1995), meal frequency was on average 12 meals per day (24 h) when lactating cows were fed ad libitum hay, corn silage and grass silage. This ®gure differs from the corresponding treatment in our study (C) where the cows were fed ad libitum. This may partly be a question of how to de®ne a meal; we did not make any de®nition of this but considered an eating session as a bout when it was preceded and followed by any other behaviour. The very high frequency of eating bouts found in treatment A may at least partly have been caused by physical and possibly physiological stimuli from the un®lled rumen. If these kind of stimuli triggered the frequent initiation of eating, it is very interesting that the total eating duration did not differ between treatments A and C, even though the feed was available ad libitum. This might indicate that a feedback mechanism working by oral stimulation determined eating duration in total, while the initiation of bouts might be more linked to stimulation originating from the rumen alimentary tract. Freer et al. (1962) found in one experiment that the total time spent eating and ruminating was almost constant, i.e. that there should exist a complementary relationship between the time spent eating and the time spent ruminating. Freer et al. (1962) suggest that the mean time spent eating roughages depends upon the rate at which the roughage is broken down in the rumen and on the contribution that mastication, during eating and ruminating, makes to achieve this breakdown. Thus, time spent ruminating depends upon the characteristic of the roughage, and the rate at which the roughage is eaten. The cows in treatment D had very low ®gures for feeding behaviours, i.e. eatingrumination (23.6% of 24 h time) compared with the time spent feeding that has been found in cattle fed ad libitum hay during non-competition situations (about 60% of 24 h time; see Albright and Arave, 1997). In the present study, it is very interesting to note that the D cows also had the highest ®gures of stereotypies, far beyond the levels of stereotypies seen in the other three treatments, both in duration and frequency. Several previous studies have shown that those stereotyped oral behaviours that are typical for cattle, i.e. tongue-rolling, barbiting and excessive licking, are strongly affected by feeding regime (e.g. Redbo et al., 1996; Redbo and Nordblad, 1997). The latter studies showed that restrictive feeding of cows and a decreased duration of feeding behaviours, respectively, signi®cantly increase the development and performance time of oral stereotypies, while an increased feeding duration reduces stereotypy levels. Similar effects have
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
95
been found in other species, such as mink (Bildsoe et al., 1991) and pigs (e.g. Robert et al., 1993) where stereotypy levels were reduced with higher feeding levels. In this study, we have regarded stereotyped oral behaviours as indicators of frustrated feeding motivation. Even though these stereotypies can be in¯uenced by other factors than feeding, such as e.g. tethering (Redbo, 1992a), the cows in this study were all exposed to the same management and environmental factors irrespective of treatment. In the beginning of the experiment we were somewhat uncertain if 3 days of treatment would be enough to achieve any responses in stereotypy levels. It turned out to be enough to detect differences between treatments, but it is quite possible that longer periods on the various treatments would have accentuated the results. The reason why we limited the treatments to 3 days only was that the A treatment would have involved a cumulative physiological strain on the cows if it had proceeded for a longer period. The distribution of stereotypies, whether it concerns their occurrence, duration or frequency, is usually skew, with typical high standard deviations and low medians. This requires rather large differences to be regarded as signi®cant. When the normal days for A and B treatments were compared, the ®gures for all behavioural categories were almost similar, except for stereotypies (Figs. 1 and 3). Nonetheless, there was a signi®cant decrease in stereotypy duration between a normal day and a treatment day in the A treatment. This is in accordance with the studies mentioned above, where an increased feeding duration decreased the performance time of stereotypy levels in dairy heifers (e.g. Redbo and Nordblad, 1997). Brouns et al. (1994) got similar results when gilts that were fed a high ®bre diet that took longer to consume, decreased their performance time of oral abnormal behaviours in general. The two treatments with a long eating time, A and C, had very low duration of stereotypies, and even though these values were not signi®cantly different from those in treatment B, the tendency for higher levels in B was rather obvious in both years. We had expected that the cows in treatment B should respond to the short eating time with higher stereotypy levels. An important reason for this lack of response might be that the B cows had relatively high ®gures for ruminating, a behaviour that may have compensated for the low time spent with oral feed intake. The time spent with rumination in B, notwithstanding the small portion of time spent with eating, might have provided the necessary contribution to the oral stimulation needed to decrease feeding motivation. Treatment D imposed, undoubtedly, the highest strain on the cows with its low duration of eating and ruminating together with a low rumen load, and the cows responded with much higher stereotypy levels to this treatment compared with the other three. Consequently, when measured in stereotypy responses, a short duration of feeding behaviours combined with low feed allotments/low rumen content constituted the worst possible condition for the cows among the treatments used in this study. The second orally expressed behaviour that we have used as markers for frustrated feeding motivation in this study was non-stereotyped nosing and licking in or around the feeding site or trough. We have called this behavioural category `behaviours related to feed-searching' since the behaviour is composed of elements of appetite behaviours and probably may be triggered by feeding motivation. In previous studies, these behaviours have been found to be closely related to stereotypies since they occur signi®cantly more often immediately before and after a stereotypy bout than any other behaviour (e.g. Redbo, 1992a,b). Thus, we would have expected them to increase in the treatments with short
96
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
eating duration and to be highest in treatment D. This was also what we found; cows in B and D had higher duration of feed-searching related behaviours than A and C, and the same trend was obvious in year two when treatments A and B were compared. Similarly, when normal days were compared with treatment days in A and B, the time spent on these behaviours was decreased in treatment A and increased in treatment B. In addition, the frequency of these behaviours increased between normal days and treatment days in B, while no such difference was found in A. Taken together, these results indicate that motivation for feeding was persistent in treatments with short eating duration. However, the unexpectedly high frequency of behaviours related to feed-searching in A in year 1 is dif®cult to explain. The other treatment with long duration of eating, C, had very low corresponding ®gures. A partial explanation might be that the A cows could have had a generally increased, but unsatis®ed, appetite for macro-minerals caused by the continuous removal of rumen content. This was most probably also the reason why the A cows had such high ®gures for licking on the salt lick, which, however, only contained pure sodium chloride. An estimation of the amount of sodium lost from the A cows by removal of rumen content gives a ®gure of about 230 g sodium per day per cow. The frequent use of the salt lick probably restored this loss, but a side-effect might have been an excess of chloride, a load that would probably have led to ailments, such as acidosis if the treatment had proceeded long enough. However, all the A cows were continually controlled for symptoms of illness, and during the 3 days of treatment no such disturbances were detected in either of the two experimental years. In the present study, the total time standing was longer in treatments with free access to feed (A and C) than in the treatments with restricted feeding (B and D). This must of course have been an effect of the treatments as cows usually stand up while eating. Freer and Campling (1965) got similar results in a study with cows offered ad libitum long hay; the time spent resting varied inversely with the time spent eating and ruminating. In conclusion, our results support the hypothesis that oral manipulation of feed is a behavioural need in cattle irrespective of rumen load. A low duration of feeding behaviours combined with a low rumen load, which is a common practice in certain categories of growing cattle or dry dairy cows, seriously impairs the welfare in cattle. Acknowledgements This study constitutes part of a Nordic cooperation project (NKJ 94) where behavioural needs in cattle are considered. The research was ®nanced by the Swedish Council for Forestry and Agricultural Research and The Swedish Farmer's Foundation for Agricultural Research. The authors are grateful to Ms. Annette Herrloff and Ms. Anna SaÈaÈf for excellent technical assistance, and to Dr. Peter UdeÂn and Prof. Bo Algers for valuable discussions concerning the planning and design of this study. References Albright, J.L., Arave, C.W., 1997. The Behaviour of Cattle. CAB International, Wallingford, 306 pp. Balch, C.C., 1971. Proposal to use time spent chewing as an index of the extent to which diets for ruminants possess the physical property of fibrousness characteristics of roughages. Br. J. Nutr. 26, 383±392.
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
97
Bildsoe, M., Heller, K.E., Jeppesen, L.L., 1991. Effect of immobility stress and food restriction on stereotypies in low and high stereotyping female ranch mink. Behav. Proc. 25, 179±189. Brouns, F., Edwards, S.A., English, P.R., 1994. Effect of dietary fibre and feeding system on activity and oral behaviour of group housed gilts. Appl. Anim. Behav. Sci. 39, 215±223. Campling, R.C., Balch, C.C., 1961. Factors affecting the voluntary intake of food by cows. Br. J. Nutr. 15, 523± 530. Carr, S.B., Jacobson, D.R., 1967. Intraruminal addition of mass or removal of rumen contents on voluntary intake of the bovine. J. Dairy Sci. 50, 1814±1818. Duncan, I.J.H., Woodgush, D.G.M., 1972. Thwarting of feeding behaviour in domestic fowl. Anim. Behav. 20, 444±451. Forbes, J.M., 1995. Voluntary Food Intake and Diet Selection in Farm Animals. CAB International, Wallingford, 532 pp. Freer, M., Campling, R.C., 1965. Factors affecting the voluntary intake of food by cows. Br. J. Nutr. 19, 195±207. Freer, M., Campling, R.C., Balch, C.C., 1962. Factors effecting the voluntary intake of food by cows. Br. J. Nutr. 16, 279±295. Gonyou, H.W., Stricklin, W.R., 1981. Eating behavior of beef cattle groups fed from a single stall or trough. Appl. Anim. Ethol. 7, 123±133. Gordon, J.G., 1963. Animal behaviour and food intake. In: Cuthbertson, D.P. (Ed), Progress in Nutrition and Allied Sciences. Oliver & Boyd, Edinburgh, pp. 281±286. Hafez, E.S.E., Bouisson, M.F., 1975. The behaviour of cattle. In: Hafez, E.S.E (Ed.), The Behaviour of Domestic Animals. BaillieÁre Tindall, London, pp. 203±245. Jensen, P., Toates, F.M., 1993. Who needs `behavioural needs'? Motivational aspects of the needs of animals. Appl. Anim. Behav. Sci. 37, 161±181. Minitab, 1996. Release 11 for Windows. Minitab Inc., State College, USA. Palya, W.L., Zacny, J.P., 1980. Stereotyped adjunctive pecking by caged pigeons. Anim. Learning Behav. 8, 293±303. Redbo, I., 1990. Changes in duration and frequency of stereotypies and their adjoining behaviours in heifers, before, during and after the grazing period. Appl. Anim. Behav. Sci. 26, 57±67. Redbo, I., 1992a. The influence of restraint on the occurrence of oral stereotypies in dairy cows. Appl. Anim. Behav. Sci. 35, 115±123. Redbo, I., 1992b. Stereotypies in dairy cattle and their relation to confinement, production-related factors, physiological reactions and adjoining behaviours. Thesis Report 212, Swedish University of Agricultural Science, Uppsala. Redbo, I., Nordblad, A., 1997. Stereotypies in heifers are affected by feeding regime. Appl. Anim. Behav. Sci. 53, 193±202. Redbo, I., Emanuelson, M., Lundberg, K., Oredsson, N., 1996. Feeding level and oral stereotypies in dairy cows. Anim. Sci. 62, 199±206. Robert, S., Matte, J.J., Farmer, C., Girard, C.L., Martineau, G.P., 1993. High-fibre diets for sows: effect on stereotypies and adjunctive drinking. Appl. Anim. Behav. Sci. 37, 297±309. Rushen, J.P., 1985. Stereotypies, aggression and the feeding schedules of tethered sows. Appl. Anim. Behav. Sci. 14, 137±147. Senn, M., Durst, B., Kaufmann, A., Langhans, W., 1995. Feeding patterns of lactating cows of three different breeds fed hay, corn silage, and grass silage. Physiol. Behav. 58, 229±236. Systat 8.0. Statistics 1998. SPSS Inc. Zar, J.H., 1984. Biostatistical Analysis, 2nd Edition. Prentice-Hall, Englewood Cliffs, NJ.
Effect of feeding duration and rumen ®ll on behaviour in dairy cows Tina LindstroÈm*, Ingrid Redbo Department of Animal Nutrition and Management, KungsaÈngen Research Centre, Swedish University of Agricultural Sciences, S-753 23 Uppsala, Sweden Accepted 27 April 2000
Abstract The aim of the present experiment was to test the hypothesis that oral manipulation of feed is a behavioural need in cattle, irrespective of actual rumen load. Twelve rumen ®stulated cows were used and subjected to four different treatments: low rumen contentlong duration of eating (A), high rumen contentshort duration of eating (B), high rumen contentlong duration of eating (positive control) (C) and low rumen contentshort duration of eating (negative control) (D). To obtain treatment A and B, rumen content was transferred by hand from cow A to B through the rumen ®stulaes. Each treatment lasted for 3 days with 2 weeks of recovery between each new treatment. The experiment was repeated twice during two consecutive years. All cows were fed the same mixture of silage, concentrate and hay. The cows were videotaped under normal conditions (24 h), and on the third day of the experiment. From these videotapes, the behaviours (frequency and duration per 24 h) have been analysed. Time spent eating differed between the four treatments (P<0.001), with shortest eating-times in B and D. The cows with low rumen content (A and D) spent shorter time ruminating (P<0.001) than the cows with ®lled rumen (B and C). The B and D cows (short duration of eating) spent longer time (P<0.001) with behaviours related to feed-searching than the cows with long duration of eating (A and C). The C cows had fewer (P<0.001) bouts of behaviours related to feed-searching than the A, B and D cows. Time spent with stereotypies (tongue-rolling) was longer (P<0.01) in D than in the other treatments. There was a difference (P<0.001) between treatments in eating bouts. The A cows had more (P<0.05) eating bouts than the cows in B, C and D. The cows with low rumen content (A and D) had fewer ruminating bouts (P<0.001) than the cows with ®lled rumens (B and C). The number of bouts with stereotypies differed (P<0.01), the cows in D having the highest ®gures compared with all the other treatments. In conclusion, our results support the hypothesis that oral manipulation of feed is a behavioural need in cattle irrespective of rumen load. A low duration of feeding behaviours combined with a
* Corresponding author. Tel.: 46-1867-2699; fax: 46-1867-2946. E-mail address: [email protected] (T. LindstroÈm).
0168-1591/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 1 5 9 1 ( 0 0 ) 0 0 1 4 8 - 9
84
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
low rumen load, which is a common practice in certain categories of growing cattle or dry dairy cows, seriously impairs the welfare in cattle. # 2000 Elsevier Science B.V. All rights reserved. Keywords: Cattle-feeding and nutrition; Feeding behaviour; Stereotypies; Behavioural need
1. Introduction The feeding of dairy cows during the stable period is often accomplished within a rather short time, compared with the duration of eating and rumination among cows on pasture. A study of heifers (Redbo and Nordblad, 1997) showed that both the development as well as the performance time of stereotypies were related to the duration of feeding behaviours. Corresponding results have been found in dairy cows where restricted feeding may lead to signi®cant increases in behavioural disturbances, such as stereotyped oral behaviours, compared with giving feed ad libitum (Redbo et al., 1996). This has also been found in studies of other species. Rushen (1985) found in an early study that the occurrence of stereotypies was strongly associated with feeding periods in tethered sows. In a study with pregnant gilts and sows (Robert et al., 1993), bulky feeding regimes reduced stereotypies over 24 h and hardly any oral behaviours were performed by the gilts fed ad libitum. A study of dry sows (Brouns et al., 1994) showed that the occurrence of stereotypies was related to a combination of a lack of suf®cient amount of food to produce satiety and to a frustration of feeding/foraging behaviour. In hens, sterotyped pacing was found by Duncan and Woodgush (1972) to be associated with long-term and intense feeding frustration. In this investigation, hens exposed to a short duration of feed-deprivation responded to a frustrated feeding situation with displacement preening, while birds strongly motivated to feed due to a long deprivation time developed stereotyped pacing. In another study with caged and feed-deprivated pigeons, the stereotypy spot-pecking disappeared completely when the birds got free access to food (Palya and Zacny, 1980). Even though the cows physiological needs are most often ful®lled in modern dairy systems, studies have indicated that the behavioural, or psychological, need of heifers and dairy cows for the actual performance of feeding behaviours may be unful®lled (e.g. Redbo, 1990; Redbo et al., 1996; Redbo and Nordblad, 1997). A behavioural need has, according to Jensen and Toates (1993), been used to describe ``the need to perform a speci®c behaviour pattern whatever the environment is like and even if the physiological needs which the behaviour serves are ful®lled''. Jensen and Toates argue that ``animals do possess needs with regards to their behaviour, i.e. prevention of performance of speciesspeci®c behaviour may cause signs of reduced welfare''. All this raises the question if there are two different `kinds' of hunger Ð one that is more physiologically determined, and one more linked to a psychological need for the performance of the oral manipulation of feed. However, there is no clear distinction between these two systems since they are both parts of the complicated hunger-satiety system. Although the very duration of oral feed manipulation obviously is important for the wellbeing of cattle, we do not know how important a long feeding duration is in relation to physical ful®lment. It is likely that species evolved for long feeding sessions, such as cattle,
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
85
need a long sensory and oral stimulation by the act when feed is passing through the mouth in order to receive necessary negative feedback on feeding motivation. It has been found that behaviours related to feed-searching, such as nosing and licking in or around the empty feed trough or feeding site, adjoins a stereotypy bout signi®cantly more often than expected (Redbo, 1992a). This relation between behaviours related to feed-searching and stereotypies may indicate that these stereotypies can be considered as displacement behaviours and that the switch from behaviours related to feed-searching to stereotypies occurs as a consequence of the frustration of the former behaviour, i.e. the motivation to ingest feed (Redbo, 1992b). Thus, both stereotyped oral behaviours as well as behaviours related to feed-searching are interesting as markers for frustrated feeding behaviours in cattle. To be able to correctly interpret the effect of a long or a short eating duration on the well-being of the cow, the effect of oral feed intake must be separated from the effect of rumen load. It is possible to achieve such a separation by the use of cows with rumen ®stulaes through which already swallowed feed can be extracted or rumen content from a donor cow can be inserted. In this study, we have used such cows. For readers not familiar with rumen ®stulaes in cows, we must explain that the experimental cows all are treated as normal dairy cows, i.e. they calve once a year, spend the summer on pasture, etc. The aim of the present experiment was to test the hypothesis that oral manipulation of feed is a behavioural need in cattle, irrespective of actual rumen load. The objective of this study was to examine how feeding duration and rumen ®ll, respectively, would in¯uence behavioural variables possibly re¯ecting frustrated feeding motivation. 2. Materials and methods 2.1. Animals and study design Twelve rumen-®stulated, lactating, cows of the Swedish Red and White Cattle Breed with ages ranging from 3 to 10 years, were used and subjected to four different treatments:
A: low rumen contentlong duration of eating, B: high rumen contentshort duration of eating, C: high rumen contentlong duration of eating (positive control), D: low rumen contentshort duration of eating (negative control).
Each animal passed through every treatment but in different orders. Each treatment lasted for 3 days, followed by a 2-week recovery period when the cows were kept in accordance with ordinary management and feeding routines, before a new 3-day experimental period started. The experiment was repeated twice during two consecutive years. All cows were housed in the same stable in individual stalls with an empty stall or a solid partition in between them. All the cows included in the experiment were continuously controlled for symptoms of illness or ailments in addition to body temperature control. All cows were healthy and in good shape during the experimental days both years, except one cow that had to be culled the ®rst year, due to mastitis which was not caused by the experiment.
86
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
2.2. Feeding All cows were fed a mixture consisting of 60% grass clover silage, 5% grass hay and 35% concentrate in individual amounts calculated according to their individual milk production. The energy concentration in the mixture varied during the 2 years between 11.7 and 12.0 MJ metabolizable energy per kg dry matter. One day before the experiment started the feed was restricted to 75% of normal individual feed amount for all four treatments. The cows were given food individually in troughs twice a day at 06.35 and 13.50 h year 1, and at 05.55 and 14.20 h year 2. Amounts of left-overs were recorded during the whole experimental period. All cows had free access to water in individual bowls and free access to salt lick. To obtain treatments A and B we transferred rumen content by hand from cow A to B through the rumen ®stulaes. The rumen content was always taken from the upper layer in the same area in the centre of the rumen of cow A. The rumen content was weighed continuously during every transfer. Pilot studies had shown that rumen content derived from the present feed mixture and taken from that area, had a DM content of 17%. Cows in treatment A were offered 150% of normal feed amount to eat orally (in principle ad libitum) but rumen content was regulated to stay at 75% of normal feed allotment. Cows in treatment B were offered 50% of normal feed amount to be eaten orally, while 50% (in the form of rumen content from cows in A) was ®lled directly into the rumen. Cows in C received 150% of normal amounts, while cows in D only received 50% to eat. Neither treatment C or D involved any transfer of rumen content, the complete feed amounts were eaten orally. Transfer of rumen content from cow A to B was made three times a day (07.00, 10.00 and 13.00 h) and at 16.00 h rumen content was extracted from the rumen from cow A without passing it to cow B. The aim of this control of rumen ®ll was to standardise the effects of the treatments, i.e. to control that the cow had a ®lled or non-®lled rumen, respectively. This was made to separate the physiological effects of feed intake that originates from the rumen itself, from the effects of the oral manipulation of the feed. 2.3. Behavioural observations The cows were videotaped both under normal conditions (24 h) 2 days before the onset of the treatments, and on the third day of the experiment, in all four treatments. From these videotapes the following behaviours have been analysed according to their total frequency and duration per 24 h: eating Ð when the cow was obtaining food from the trough or from the ¯oor in her mouth or when chewing or swallowing food; ruminating Ð chewing or regurgitating already swallowed food; behaviours related to feed-searching Ð nosing or licking the empty feed-trough, or on the ¯oor, or ®ttings close of the trough; licking salt Ð licking on the salt lick; stereotypy Ð when the cow consistently rolled its tongue outside or inside her mouth or chewed on a bar in an unvaried, repeatedly and stereotyped manner; total standing Ð when the cow was standing up on all four legs irrespective of any other activities; total lying Ð when the cow was lying irrespective of other activities; other active Ð included all other behaviours except eating, ruminating, behaviours related to feedsearching and stereotypy.
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
87
Behavioural recordings from the `normal day' and from treatments C and D were only performed in year 1. In year 2 the analyses only concerned the two treatments A and B. 2.4. Statistical analyses The analyses of behavioural data were performed using Minitab (1996) and Systat (1998). All behavioural variables were tested for normality and variables not normally distributed were logarithmically transformed. In most cases, the transformations did not improve the distribution and, thus, non-parametric statistics were used for these variables. Differences between treatments (year 1) in duration of normally distributed behaviours (eating, ruminating, total standing and total lying) were tested with one-way Anovas followed by Tukeys tests. Differences between treatments in duration of behaviours related to feed-searching, licking salt and stereotypies (not normally distributed behaviours) were tested using non-parametric statistics (Kruskal±Wallis k-sample test). This test was followed by a non-parametric Tukey-type multiple comparison (Zar, 1984). Differences between treatments (year 1) in frequency of normally distributed behaviours (ruminating, total standing and total lying) were tested with Anovas followed by Tukeys tests. Differences between treatments in frequency of behaviours related to feed-searching, licking salt and stereotypies (not normally distributed behaviours) were tested using nonparametric statistics (Kruskal±Wallis k-sample test). The ®gures for frequencies of bouts of eating were transformed (log) and thereafter analysed with Anovas followed by Tukeys tests. Differences (year 2) between treatments A and B in duration of normally distributed behaviours (eating, ruminating, total standing and total lying) were tested with the Two Sample T-tests. Differences between A and B in duration of behaviours that were not normally distributed (behaviours related to feed-searching, licking salt, and stereotypies) were tested with the Mann±Whitney test. Differences (year 2) between A and B in frequencies of normally distributed behaviours (ruminating, total standing and total lying) were tested with Two Sample T-test. Differences between A and B in frequencies of behaviours that were not normally distributed (eating, behaviours related to feed-searching, licking salt, and stereotypies) were tested with Mann±Whitney test. Differences between normal day and experimental day in duration were tested with the Paired Samples T-test for normally distributed behaviours (eating, ruminating, total standing and total lying). For behaviours related to feed-searching, licking salt and stereotypies (not normally distributed) the differences were tested with the Wilcoxon Signed Ranks test. Differences between normal day and experimental day in frequency were tested with the Paired Samples T-test for normally distributed behaviours (ruminating, total standing and total lying). For eating, behaviours related to feed-searching, licking salt and stereotypies (not normally distributed) the differences were tested with the Wilcoxon Signed Ranks test.
88
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
3. Results The experiments were accomplished during two consecutive years. Since the results in duration of rumination in treatment A differed signi®cantly between years, the two experimental periods could not be treated as replicates. Consequently, the data from the 2 years have been treated as two similar experiments and are analysed separately. Even though the magnitude of the ®gures for duration of rumination in treatment A differ between years, a comparison between treatments A and B clearly show the same pattern both years, with high ®gures for treatment B but low for treatment A. 3.1. Feed consumption The oral feed consumption in year 1 was, on average, 25.8 kg dry matter (DM) in treatment A, 9.1 kg DM in B, 24.7 kg DM in C and 8.5 kg DM in D on the third day of experiment. In year 2 the oral consumption was 24.5 kg DM in A, 9.1 kg DM in B, 22.1 kg DM in C and 8.8 kg DM in D on the third day of experiment. In addition to the oral intake, the cows in treatment B, both year 1 and 2, received on average 9.1 kg DM per day as rumen content via the rumen ®stulaes. 3.2. Duration of certain behaviours Ð year 1 There was a signi®cant (P<0.001) difference in time spent eating between the four treatments, with shortest eating-times in treatments B and D (Table 1). When it comes to time spent ruminating there was a signi®cant (P<0.001) difference between the treatments with low rumen content (A and D) compared with the treatments with ®lled rumen (B and C). The cows in treatments B and D (with short duration of eating) spent signi®cantly (P<0.001) longer time with behaviours related to feed-searching than cows with long duration of eating (treatments A and C). The cows in treatment A spent signi®cantly Table 1 Duration of certain behaviours in percent of total observation time (24 h), year 1 experimental day 3, in the four different treatments (A, B, C and D)a Behaviour
Treatment A
Eating Rumination Behaviour related to feed-searching Licking salt Stereotypies Total standing Total lying
B a
23.2 (6.5) 20.3a (6.3) 1.5a (0.8) 1.4 4.4a (3.3) 4.9 0.3a (0.5) 0.0 51.1 (9.5) 48.9 (9.5)
C b
4.4 (1.8) 31.7b (4.4) 3.2b (1.0) 3.1 0.4b (0.4) 0.3 1.8a (2.2) 0.6 45.8 (8.8) 54.2 (8.8)
D a
22.4 (4.0) 37.8c (4.7) 1.3a (1.8) 0.3 0.7b(0.8) 0.5 0.2a (0.3) 0.0 51.1 (10.9) 48.9 (10.9)
7.7b (5.7) 15.9a (4.9) 3.4b (1.5) 2.7 1.6b (1.7) 1.4 8.6b (11.0) 1.5 44.7 (8.0) 55.3 (8.0)
a Means with different superscripts within a variable row are significantly different at (P<0.05). Figures within brackets are standard deviations. Figures in italics are medians.
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
89
Table 2 Frequencies (bouts) of certain behaviours during observation time (24 h), year 1, experimental day 3 in the four different treatments (A, B, C and D)a Behaviour
Treatment A
Eating Rumination Behaviour related to feed-searching Licking salt Stereotypies Total standing Total lying
B a
61.0 (30.3) 51.0 11.4a (3.2) 103.7a (46.7) 113.0 126.0a (68.5) 121.0 11.4a (23.8) 0.0 10.3 (2.7) 10.0 (2.6)
C b
4.9 (4.7) 3.0 20.5b (3.5) 160.1a (63.0) 151.0 10.9b (18.0) 7.0 14.0a (16.6) 7.0 11.6 (4.3) 11.3 (4.8)
D c
21.9 (8.0) 24.0 25.2b (6.9) 18.3b (15.7) 11.0 11.7b (12.3) 8.0 3.6a (8.0) 0.0 14.3 (6.2) 14.3 (6.2)
14.6c (5.3) 13.0 11.8a (2.7) 87.5a (42.0) 85.0 26.8b (21.8) 26.0 34.9b (33.0) 17.0 10.2 (2.4) 9.4 (2.6)
a Means with different superscripts within a variable row are significantly different at (P<0.05). Figures in brackets are standard deviations. Figures in italics are medians.
(P<0.01) longer time licking salt than cows in the other three treatments. Time spent with stereotypies was signi®cantly (P<0.01) longer in treatment D than in the other treatments. There were no differences between treatments either in duration of total standing, or in duration of total lying. 3.3. Frequencies of behaviours Ð year 1 There was a signi®cant (P<0.001) difference between treatments in the frequency of eating bouts. The cows in treatment A showed a signi®cantly (P<0.05) higher number of eating bouts than the cows in treatments B and C as well as D (Table 2). There were no differences in the number of eating bouts in treatment C compared with treatment D. The cows with low rumen content (treatments A and D) had signi®cantly (P<0.001) fewer ruminating bouts than the cows with ®lled rumens (treatments B and C). When it comes to behaviours related to feed-searching there was a signi®cant (P<0.001) difference between cows in treatment C and the other three treatments that showed a larger number of bouts. The cows in treatment A licked the salt lick signi®cantly (P<0.001) more often than the cows in the other treatments. There was a signi®cant (P<0.01) difference in the number of bouts with stereotypies Ð the cows in treatment D had the highest ®gures compared with all the other treatments. There were no differences between treatments in number of bouts of total standing or total lying. 3.4. Duration of certain behaviours Ð year 2 There was a signi®cant difference between treatments A and B in duration of eating, with higher ®gures for cows in treatment A (Table 3). Time spent with rumination showed signi®cantly higher ®gures for cows in treatment B compared with cows in A. The cows with short eating-time (B) spent signi®cantly more time with behaviours related to feed-
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
90
Table 3 Duration of certain behaviours in percent of total observation time (24 h), and frequencies (bouts), year 2, experimental day 3a Behaviour
Year 2 Duration
Frequency
A
B
Significance
A
B
Significance
Eating
26.3 (6.2)
6.0 (4.7)
P<0.001
27.6 (5.9) 1.7 (1.3)
33.7 (4.9) 3.1 (1.7)
P<0.05 P<0.05
9.5 (8.7) 6.0 20.4 (5.9) 58.0 (30.2)
P<0.001
Rumination Behaviour related to feed-searching
76.5 (42.4) 66.0 13.8 (3.3) 37.4 (36.5)
1.2 6.3 (2.3) 5.5 0.5 (0.7) 0.2 49.9 (6.2) 50.1 (6.1)
3.1 0.6 (0.6) 0.5 3.7 (4.4) 0.7 45.9 (9.1) 54.1 (9.1)
30.0 132.6 (66.9) 114.0 14.6 (21.7) 4.5 11.4 (2.8) 10.4 (2.8)
53.0 11.4 11.5 19.0 12.5 12.1 11.1
Licking salt Stereotypies Total standing Total lying a
P<0.001 n.s. n.s. n.s.
P<0.01 P<0.05
(10.0)
P<0.001
(24.9)
n.s.
(5.1) (5.1)
n.s. n.s.
Figures in brackets are standard deviations and figures in italics are medians.
searching than cows with a long eating-time (A). Even this year the cows in treatment A spent signi®cantly more time licking the salt lick compared with the cows in treatment B. There were no differences between treatments in time spent with stereotypies, total standing or total lying. 3.5. Frequencies of behaviours Ð year 2 There was a signi®cant difference in number of eating bouts between treatments A and B (Table 3) with higher ®gures for the A cows. The number of ruminating bouts showed signi®cantly higher ®gures for the B cows than for the A cows. The cows in treatment B showed signi®cantly higher frequencies of behaviours related to feed-searching than the cows in treatment A. The number of bouts of licking the salt lick was signi®cantly higher in treatment A than in treatment B. There were no signi®cant differences between treatments in numbers of stereotypy bouts or numbers of bouts of total standing or total lying. 3.6. Normal day compared with experimental day 3 Ð duration In treatment A, there were signi®cantly higher ®gures in duration of eating, licking salt and total time spent standing (P<0.01) on experimental day 3 compared with normal day (Fig. 1). Duration of rumination, behaviours related to feed-searching and stereotypies showed signi®cantly lower ®gures on the experimental day compared with a normal day in treatment A (Fig. 1). In treatment B, the cows spent signi®cantly less time with eating on the experimental day compared with a normal day, but signi®cantly more time with rumination and behaviours related to feed-searching (Fig. 2). There were no signi®cant differences in time spent
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
91
Fig. 1. Duration of certain behaviours in percent of total observation time (24 h) year 1, normal day and experimental day 3 in treatment A. Figures above the bars show their standard deviations.
Fig. 2. Duration of certain behaviours in percent of total observation time (24 h) year 1, normal day (N) and experimental day 3, treatment B. Figures above the bars show their standard deviations.
92
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
Fig. 3. Frequencies of certain behaviours during total observation time (24 h) year 1, normal day and experimental day 3, treatment A. Figures above the bars show their standard deviations.
licking salt, stereotypies, total standing or total lying on a normal day compared with experimental day in treatment B. 3.7. Normal day compared with experimental day 3 Ð frequencies In treatment A, the cows had signi®cantly more bouts of eating and licking salt on the experimental day than on a normal day (Fig. 3). The number of bouts of ruminating was
Fig. 4. Frequencies of certain behaviours during total observation time (24 h) year 1, normal day and experimental day 3, treatment B. Figures above the bars show their standard deviations.
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
93
signi®cantly lower on the experimental day than on a normal day in treatment A. There were no signi®cant differences in frequencies of behaviours related to feed-searching, stereotypies, total standing or total lying. In treatment B, the cows showed a signi®cantly higher frequency of eating on a normal day than on an experimental day (Fig. 4). On the experimental day, the number of bouts was signi®cantly higher for rumination and behaviours related to feed-searching than on a normal day. There were no signi®cant differences in treatment B between a normal day and an experimental day in frequencies of licking salt, stereotypies, total standing or total lying. 4. Discussion In the present study, we have found that a long duration of oral manipulation of feed by eating and ruminating lowered the levels of stereotypies and behaviours related to feedsearching in lactating dairy cows irrespective of their rumen load. This indicates that cattle have a behavioural need to perform feeding behaviours. The cows with free access to feed in the present study (treatments A and C) showed three to four times longer duration of eating (about 5.5 h of 24 h) than the cows in the other two treatments. This was expected, since this was the very aim of the treatments. This is in line with other studies, such as Senn et al. (1995) reported that cows in a loose-housing system fed ad libitum hay, corn silage, and grass silage spent about 5.5 h eating. Freer and Campling (1965) offered hay ad libitum to cows, which resulted in a mean eating duration of 4.23 h, and results from Freer et al. (1962) showed that cows fed ad libitum hay ate 4 h per 24 h. In our study, the B and D cows were able to consume all the feed at once due to their restricted feeding, and this gave them rather short eating times (about 1 and almost 2 h, respectively). Such low ®gures of eating duration are usually seen when cattle are fed high levels of concentrate and low levels of roughages (Gonyou and Stricklin, 1981). In the present study, all cows in all treatments started to eat at the same time as the feed was delivered in the mornings and in the afternoons, meaning that the addition of fresh feed to the trough was an important stimulus for the cows to initiate eating. Campling and Balch (1961) reported that the eating time increased from 3±4 h to 6.5±8 h when swallowed food was removed during the ®rst 3 h of eating in rumen ®stulated cows. In another study (Carr and Jacobson, 1967), 1.4 kg DM was removed from the rumen daily 3 h after feeding and this resulted in a signi®cant increase in voluntary intake of 0.5 kg DM. This indicates that the amount of digesta in the reticulo-rumen is one of the factors that governs the feed intake in cattle. On the other hand, Gordon (1963) made some experiments with rumen content transfer in sheep and stated that the quantity of food in the alimentary tract below the reticulum and rumen (i.e. the omasum, the abomasum and the intestine) in¯uences the quantity of food a sheep will eat to a greater extent than the quantity of food in the rumen itself. In our study, the A and C cows spent approximately the same time eating, even though rumen content of the A cow was continuously extracted, which might indicate that rumen ®ll is not the only factor that governs feed intake in cattle. Our results are in accordance with several earlier studies indicating that receptors in the buccal cavity and throat also are important to encourage further feeding or to cease feeding (see e.g. Forbes, 1995). The time spent feeding found in A and C cows is almost on the level
94
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
as for freely grazing cattle (Hafez and Bouisson, 1975) and might be suf®ciently long to give negative feed-back on motivation for eating. Time spent ruminating depends mainly upon the characteristics of the roughage and the rate at which the roughage is eaten (Freer et al., 1962). Cows fed long hay ad libitum ruminated for 9.14 h (Freer and Campling, 1965), which is in accordance with our ®ndings in treatment C, which had the longest duration of rumination (9.40 h). Cows on restricted feeding (treatment D) had rather short duration of ruminating (3.49 h), which was statistically similar to the A cows even though the A cows were fed ad libitum and, thus, had much longer eating times. The relatively low ®gures for rumination in A are, of course, explained by the fact that rumen content was continually removed from the rumen of A. The relatively long ruminating times in B are due to the circumstance that the B cows had to ruminate food eaten but not ruminated by the A cows. Ad libitum feeding of cows generally leads to many eating occasions during the 24 h (Balch, 1971). In a study by Senn et al. (1995), meal frequency was on average 12 meals per day (24 h) when lactating cows were fed ad libitum hay, corn silage and grass silage. This ®gure differs from the corresponding treatment in our study (C) where the cows were fed ad libitum. This may partly be a question of how to de®ne a meal; we did not make any de®nition of this but considered an eating session as a bout when it was preceded and followed by any other behaviour. The very high frequency of eating bouts found in treatment A may at least partly have been caused by physical and possibly physiological stimuli from the un®lled rumen. If these kind of stimuli triggered the frequent initiation of eating, it is very interesting that the total eating duration did not differ between treatments A and C, even though the feed was available ad libitum. This might indicate that a feedback mechanism working by oral stimulation determined eating duration in total, while the initiation of bouts might be more linked to stimulation originating from the rumen alimentary tract. Freer et al. (1962) found in one experiment that the total time spent eating and ruminating was almost constant, i.e. that there should exist a complementary relationship between the time spent eating and the time spent ruminating. Freer et al. (1962) suggest that the mean time spent eating roughages depends upon the rate at which the roughage is broken down in the rumen and on the contribution that mastication, during eating and ruminating, makes to achieve this breakdown. Thus, time spent ruminating depends upon the characteristic of the roughage, and the rate at which the roughage is eaten. The cows in treatment D had very low ®gures for feeding behaviours, i.e. eatingrumination (23.6% of 24 h time) compared with the time spent feeding that has been found in cattle fed ad libitum hay during non-competition situations (about 60% of 24 h time; see Albright and Arave, 1997). In the present study, it is very interesting to note that the D cows also had the highest ®gures of stereotypies, far beyond the levels of stereotypies seen in the other three treatments, both in duration and frequency. Several previous studies have shown that those stereotyped oral behaviours that are typical for cattle, i.e. tongue-rolling, barbiting and excessive licking, are strongly affected by feeding regime (e.g. Redbo et al., 1996; Redbo and Nordblad, 1997). The latter studies showed that restrictive feeding of cows and a decreased duration of feeding behaviours, respectively, signi®cantly increase the development and performance time of oral stereotypies, while an increased feeding duration reduces stereotypy levels. Similar effects have
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
95
been found in other species, such as mink (Bildsoe et al., 1991) and pigs (e.g. Robert et al., 1993) where stereotypy levels were reduced with higher feeding levels. In this study, we have regarded stereotyped oral behaviours as indicators of frustrated feeding motivation. Even though these stereotypies can be in¯uenced by other factors than feeding, such as e.g. tethering (Redbo, 1992a), the cows in this study were all exposed to the same management and environmental factors irrespective of treatment. In the beginning of the experiment we were somewhat uncertain if 3 days of treatment would be enough to achieve any responses in stereotypy levels. It turned out to be enough to detect differences between treatments, but it is quite possible that longer periods on the various treatments would have accentuated the results. The reason why we limited the treatments to 3 days only was that the A treatment would have involved a cumulative physiological strain on the cows if it had proceeded for a longer period. The distribution of stereotypies, whether it concerns their occurrence, duration or frequency, is usually skew, with typical high standard deviations and low medians. This requires rather large differences to be regarded as signi®cant. When the normal days for A and B treatments were compared, the ®gures for all behavioural categories were almost similar, except for stereotypies (Figs. 1 and 3). Nonetheless, there was a signi®cant decrease in stereotypy duration between a normal day and a treatment day in the A treatment. This is in accordance with the studies mentioned above, where an increased feeding duration decreased the performance time of stereotypy levels in dairy heifers (e.g. Redbo and Nordblad, 1997). Brouns et al. (1994) got similar results when gilts that were fed a high ®bre diet that took longer to consume, decreased their performance time of oral abnormal behaviours in general. The two treatments with a long eating time, A and C, had very low duration of stereotypies, and even though these values were not signi®cantly different from those in treatment B, the tendency for higher levels in B was rather obvious in both years. We had expected that the cows in treatment B should respond to the short eating time with higher stereotypy levels. An important reason for this lack of response might be that the B cows had relatively high ®gures for ruminating, a behaviour that may have compensated for the low time spent with oral feed intake. The time spent with rumination in B, notwithstanding the small portion of time spent with eating, might have provided the necessary contribution to the oral stimulation needed to decrease feeding motivation. Treatment D imposed, undoubtedly, the highest strain on the cows with its low duration of eating and ruminating together with a low rumen load, and the cows responded with much higher stereotypy levels to this treatment compared with the other three. Consequently, when measured in stereotypy responses, a short duration of feeding behaviours combined with low feed allotments/low rumen content constituted the worst possible condition for the cows among the treatments used in this study. The second orally expressed behaviour that we have used as markers for frustrated feeding motivation in this study was non-stereotyped nosing and licking in or around the feeding site or trough. We have called this behavioural category `behaviours related to feed-searching' since the behaviour is composed of elements of appetite behaviours and probably may be triggered by feeding motivation. In previous studies, these behaviours have been found to be closely related to stereotypies since they occur signi®cantly more often immediately before and after a stereotypy bout than any other behaviour (e.g. Redbo, 1992a,b). Thus, we would have expected them to increase in the treatments with short
96
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
eating duration and to be highest in treatment D. This was also what we found; cows in B and D had higher duration of feed-searching related behaviours than A and C, and the same trend was obvious in year two when treatments A and B were compared. Similarly, when normal days were compared with treatment days in A and B, the time spent on these behaviours was decreased in treatment A and increased in treatment B. In addition, the frequency of these behaviours increased between normal days and treatment days in B, while no such difference was found in A. Taken together, these results indicate that motivation for feeding was persistent in treatments with short eating duration. However, the unexpectedly high frequency of behaviours related to feed-searching in A in year 1 is dif®cult to explain. The other treatment with long duration of eating, C, had very low corresponding ®gures. A partial explanation might be that the A cows could have had a generally increased, but unsatis®ed, appetite for macro-minerals caused by the continuous removal of rumen content. This was most probably also the reason why the A cows had such high ®gures for licking on the salt lick, which, however, only contained pure sodium chloride. An estimation of the amount of sodium lost from the A cows by removal of rumen content gives a ®gure of about 230 g sodium per day per cow. The frequent use of the salt lick probably restored this loss, but a side-effect might have been an excess of chloride, a load that would probably have led to ailments, such as acidosis if the treatment had proceeded long enough. However, all the A cows were continually controlled for symptoms of illness, and during the 3 days of treatment no such disturbances were detected in either of the two experimental years. In the present study, the total time standing was longer in treatments with free access to feed (A and C) than in the treatments with restricted feeding (B and D). This must of course have been an effect of the treatments as cows usually stand up while eating. Freer and Campling (1965) got similar results in a study with cows offered ad libitum long hay; the time spent resting varied inversely with the time spent eating and ruminating. In conclusion, our results support the hypothesis that oral manipulation of feed is a behavioural need in cattle irrespective of rumen load. A low duration of feeding behaviours combined with a low rumen load, which is a common practice in certain categories of growing cattle or dry dairy cows, seriously impairs the welfare in cattle. Acknowledgements This study constitutes part of a Nordic cooperation project (NKJ 94) where behavioural needs in cattle are considered. The research was ®nanced by the Swedish Council for Forestry and Agricultural Research and The Swedish Farmer's Foundation for Agricultural Research. The authors are grateful to Ms. Annette Herrloff and Ms. Anna SaÈaÈf for excellent technical assistance, and to Dr. Peter UdeÂn and Prof. Bo Algers for valuable discussions concerning the planning and design of this study. References Albright, J.L., Arave, C.W., 1997. The Behaviour of Cattle. CAB International, Wallingford, 306 pp. Balch, C.C., 1971. Proposal to use time spent chewing as an index of the extent to which diets for ruminants possess the physical property of fibrousness characteristics of roughages. Br. J. Nutr. 26, 383±392.
T. LindstroÈm, I. Redbo / Applied Animal Behaviour Science 70 (2000) 83±97
97
Bildsoe, M., Heller, K.E., Jeppesen, L.L., 1991. Effect of immobility stress and food restriction on stereotypies in low and high stereotyping female ranch mink. Behav. Proc. 25, 179±189. Brouns, F., Edwards, S.A., English, P.R., 1994. Effect of dietary fibre and feeding system on activity and oral behaviour of group housed gilts. Appl. Anim. Behav. Sci. 39, 215±223. Campling, R.C., Balch, C.C., 1961. Factors affecting the voluntary intake of food by cows. Br. J. Nutr. 15, 523± 530. Carr, S.B., Jacobson, D.R., 1967. Intraruminal addition of mass or removal of rumen contents on voluntary intake of the bovine. J. Dairy Sci. 50, 1814±1818. Duncan, I.J.H., Woodgush, D.G.M., 1972. Thwarting of feeding behaviour in domestic fowl. Anim. Behav. 20, 444±451. Forbes, J.M., 1995. Voluntary Food Intake and Diet Selection in Farm Animals. CAB International, Wallingford, 532 pp. Freer, M., Campling, R.C., 1965. Factors affecting the voluntary intake of food by cows. Br. J. Nutr. 19, 195±207. Freer, M., Campling, R.C., Balch, C.C., 1962. Factors effecting the voluntary intake of food by cows. Br. J. Nutr. 16, 279±295. Gonyou, H.W., Stricklin, W.R., 1981. Eating behavior of beef cattle groups fed from a single stall or trough. Appl. Anim. Ethol. 7, 123±133. Gordon, J.G., 1963. Animal behaviour and food intake. In: Cuthbertson, D.P. (Ed), Progress in Nutrition and Allied Sciences. Oliver & Boyd, Edinburgh, pp. 281±286. Hafez, E.S.E., Bouisson, M.F., 1975. The behaviour of cattle. In: Hafez, E.S.E (Ed.), The Behaviour of Domestic Animals. BaillieÁre Tindall, London, pp. 203±245. Jensen, P., Toates, F.M., 1993. Who needs `behavioural needs'? Motivational aspects of the needs of animals. Appl. Anim. Behav. Sci. 37, 161±181. Minitab, 1996. Release 11 for Windows. Minitab Inc., State College, USA. Palya, W.L., Zacny, J.P., 1980. Stereotyped adjunctive pecking by caged pigeons. Anim. Learning Behav. 8, 293±303. Redbo, I., 1990. Changes in duration and frequency of stereotypies and their adjoining behaviours in heifers, before, during and after the grazing period. Appl. Anim. Behav. Sci. 26, 57±67. Redbo, I., 1992a. The influence of restraint on the occurrence of oral stereotypies in dairy cows. Appl. Anim. Behav. Sci. 35, 115±123. Redbo, I., 1992b. Stereotypies in dairy cattle and their relation to confinement, production-related factors, physiological reactions and adjoining behaviours. Thesis Report 212, Swedish University of Agricultural Science, Uppsala. Redbo, I., Nordblad, A., 1997. Stereotypies in heifers are affected by feeding regime. Appl. Anim. Behav. Sci. 53, 193±202. Redbo, I., Emanuelson, M., Lundberg, K., Oredsson, N., 1996. Feeding level and oral stereotypies in dairy cows. Anim. Sci. 62, 199±206. Robert, S., Matte, J.J., Farmer, C., Girard, C.L., Martineau, G.P., 1993. High-fibre diets for sows: effect on stereotypies and adjunctive drinking. Appl. Anim. Behav. Sci. 37, 297±309. Rushen, J.P., 1985. Stereotypies, aggression and the feeding schedules of tethered sows. Appl. Anim. Behav. Sci. 14, 137±147. Senn, M., Durst, B., Kaufmann, A., Langhans, W., 1995. Feeding patterns of lactating cows of three different breeds fed hay, corn silage, and grass silage. Physiol. Behav. 58, 229±236. Systat 8.0. Statistics 1998. SPSS Inc. Zar, J.H., 1984. Biostatistical Analysis, 2nd Edition. Prentice-Hall, Englewood Cliffs, NJ.