1. Introduction
Free-ranging ungulates forage in landscapes where food quality and quantity vary in Ž
. both space and time Senft et al., 1987 . The ability to track changes in forage resources
should allow large herbivores to graze more efficiently because they would know where Ž
. and what to graze Bailey et al., 1996 . Individual learning and social learning are the
Ž most probable mechanisms for livestock to obtain this critical information Provenza et
. Ž
. al., 1988 . Instinct genetics could be used to transfer forage resource information from
one generation to another, but individual and social learning should be favored more on Ž
rangelands due to spatial and temporal variation in forage quantity and quality Provenza .
and Balph, 1988 . Several studies have shown that cattle individually can learn the location of food and
Ž then adjust their foraging patterns to take advantage of this knowledge Bailey et al.,
. 1989a; Laca, 1998; Howery et al., 2000 . Cattle also can associate locations with the
Ž .
quantity and quality of food found there Bailey et al., 1989b; Bailey and Sims, 1998 . Potentially, cattle could readily learn the location of feeding areas from other animals.
Ž .
For example, Howery et al. 1998 observed that domestic calves learn where to graze from their dam. In feral cattle, the integrity of social groups is high, and individuals
Ž .
regularly interact with herd members Lazo, 1994 . If an animal located an area with high food resources, this information could be transferred to other members of the social
group. Familiarity, social ranking and experience may affect the efficiency with which
Ž .
information is exchanged Mosley, 1999 , but few studies have specifically examined Ž
. the effect of social facilitation on cattle during foraging. Howery et al. 1999 showed
that pen-mates transferred foraging information to other animals more efficiently than Ž
. non-pen-mates. In another study, Sato 1982 suggested that movements of a cattle herd
during grazing was directed by active movement of cattle with high-social ranking or the independent movement of low-ranking cattle.
The role of previous foraging experience on the ability of cattle to transfer foraging Ž .
information has received little examination. The objectives of this study were: 1 to Ž
. determine if cattle followers could learn the location of a new source of food directly
Ž . Ž
from a social model and 2 to determine if the foraging experiences i.e., fixed, .
variable, or no-experience of the social models would affect the rate of learning by followers.
2. Methods
2.1. Study apparatus Ž
. The study was conducted in an eight-arm radial maze Fig. 1 . The central decision
area was 30 m in diameter, and arms were 5 = 35 m. The maze was constructed from electric fence in a level area that had been mowed to a 5-cm stubble height. Animal
sorting pens were located approximately 100 m from the central decision area. Large hay stacks prevented animals in the holding and sorting pens from seeing the maze.
Ž .
Ž Opaque feed pans 0.5 m diameter and 0.2 m sides were placed at the entrance 5 m
. Ž
. from decision area and at the end of each arm Fig. 1 .
Fig. 1. Diagram of eight-arm radial maze containing eight feeders near arm entrances and eight feeders at arm ends.
Ž .
Eight-arm radial mazes have been used for several species rats, pigeons and cattle Ž
to study spatial memory Beatty and Shavalia, 1980; Roberts and Van Veldhuizen, 1985; .
Bailey et al., 1989a . This apparatus is useful for examining spatial choices of feeding sites because the distances to all feeding sites are similar, and the sequence of feeding
Ž .
site selections is relatively independent of the path taken Fig. 1 . 2.2. Animals
A total of 36 yearling heifers were divided into four groups. Nine of the heifers were 3 4 Hereford and 1 4 Tarentaise and were placed in one group. The remaining 27
Hereford heifers were randomly allotted as three groups of nine animals. Groups of
animals grazed in the same pastures or were fed in the same drylot for at least 6 months before the study. During the study, groups of heifers grazed in a common pasture from
Ž .
14:00 to 08:00 h interval between trials . Within a group, three animals were randomly selected as leaders. The remaining six
heifers were randomly assigned as three pairs of followers. Each pair of followers was randomly assigned to one of the leaders. For the purposes of this study, leaders refer to
animals that have received different training and experience in the maze than followers. As a result of the initial training, leaders were expected to locate highly palatable feed at
arm ends at different rates than followers when leaders and followers were in the maze during the same sessions.
2.3. Initial training — leaders Ž
Treatments were based on three initial training regimes of the leaders i.e., experi- .
enced-fixed, experienced-variable, and no-experience; Table 1 . One of the three leaders within a group was not allowed to travel to the end of maze arms and consume barley
Ž .
during the 7 contiguous days of initial training no-experience treatment . Gates placed in each arm 10 m from the decision area restricted access of the no-experience treatment
leaders to the end of the arms. The ‘‘experienced-fixed’’ treatment was based on a leader that received 0.8 kg of
ground barley at the end of the same two arms during each of the 7 days of initial training. Maze arms with barley were determined by first randomly selecting one of the
eight arms and then randomly selecting one of the remaining five arms that were not adjacent to the arm selected first. The purpose of this restriction was to prevent heifers
from selecting a specific section of the maze. After much training, cattle may adopt a
Ž .
strategy of choosing adjacent arms in radial maze studies Bailey and Sims, 1998 . The ‘‘experienced-variable’’ treatment was based on a leader that also received 0.8
kg of ground barley at the end of two maze arms, but the maze arms changed daily during the 7 days of initial training. The locations of the maze arms with barley were
assigned randomly with the restrictions that arms were not adjacent and that the same arm was not used 2 days in a row.
All eight inside feeders, near arm entrances, contained 0.2 kg of whole oats during initial training of all leaders. The oats were placed at arm entrances so that animals
would enter all maze arms, and failure to travel to arm ends would not be the result of animals being reluctant to enter a particular arm. For the experienced-fixed and
experienced-variable leaders, 0.05 kg of straw was placed in the six feeders at the end of maze arms that did not contain barley. Heifers rarely consumed straw throughout the
study.
Leaders were individually herded to the maze and released in the decision area for 20 min or until all feeders with grain had been visited. If a feeder with grain had not been
Ž .
visited after 20 min usually feeders at arm ends with barley , observers herded the heifer to the feeders containing grain, and the visit was recorded as an assisted choice.
2.4. Initial training — followers The goal of the initial training for followers was to train them to expect straw at arm
ends. Followers were trained in pairs for 6 contiguous days, and each pair of heifers was
considered as an experimental unit. Initial training for all followers was identical. All feeders at the end of maze arms contained 0.05 kg of straw. Inside feeders contained 0.4
Ž .
kg of whole oats 0.2 kgrheifer . For the first 4 days of initial training, heifers were herded to the end of two randomly chosen arms after 20 min in the maze. The purpose
of this herding was to ensure that heifers had the opportunity to observe and expect straw at arm ends. At the end of initial training, followers usually entered all arms and
consumed the oats in the inner feeders, but rarely traveled to arm ends.
2.5. Pretest — followers Ž
. In previous maze studies, Bailey et al.
1989a showed that cattle used spatial memory rather than olfactory cues to locate grain. The purpose of the pretest was to
confirm that heifers were not using olfactory or some other cue to locate grain in this study. On day 7 after initial training, 0.8 kg of ground barley was placed at the end of
two maze arms. The location of barley was randomly selected with the constraint that selected arms were not adjacent to one another. Heifers were allowed 20 min in the
maze. Observers did not attempt to herd or assist heifers in any way.
2.6. Exposure to leaders and test sessions Ž
. After initial training and the pretest Table 1 , a pair of followers were exposed in the
Ž .
maze with their associated leader for one 30-min session exposure session and then Ž
evaluated without the leader for two 20-min sessions on the following 2 days test .
sessions . This process was repeated a second time for each leader and pair of followers. During the exposure sessions, the leader and associated followers were herded into the
maze together and released. During test sessions, only followers were evaluated in the maze, while leaders received the same protocol as during initial training. For both the
Ž .
exposure and test sessions, ground barley 0.8 kgrfeeder was placed at the end of two arms in the identical locations as the follower pair received during the pretest, and whole
oats were placed in the inside feeders at a rate of 0.2 kgrheifer.
2.7. BehaÕiors obserÕed Ž
. Prior to the study, oats and barley 0.5 kg were offered in feeders in a separate
25 = 25 m pen. Each heifer was observed individually for 10 min, and the number of vocalizations, movement, the amount and type of grain consumed and a 1 to 5
Ž .
temperament rating 1 s calm, 5 s wild and unmanageable were recorded. During all phases of the study, the following information was observed for each
individual animal: time in maze, sequence of arms entered, time that each animal Ž
entered and exited each arm, distance traveled down each arm e.g., 1 4, 1 2, 3 4 and .
end , time spent at arm ends and whether an animal placed its muzzle in the feeders at arm ends.
2.8. Analyses 2.8.1. Leaders
The behaviors of leaders in the experienced-fixed and experienced-variable treat- ments were compared using paired t-tests to examine whether the distribution of two
Ž .
food resources nearby smaller quantities of oats and distant greater quantities of barley
D.W. Bailey
et al.
r Applied
Animal Beha
Õ iour
Science 68
2000 93
– 105
98
Table 1 Study protocol for leaders and followers in the experienced-fixed, experienced-variable and no-experience treatments
Phase Day
Experienced-fixed Experienced-variable
No-experience Purpose
Initial training — 1–7
inner feeders — oats in all inner feeders — oats in all
inner feeders — oats in all train leaders to respond to
Ž leaders 20 minr
outer feeders — barley in outers feeders — barley in
outer feeders — no access different spatial treatments
. session
two fixed locations and two locations, changed daily
straw in six locations and straw in six locations
Initial training — 1–6
inner feeders — oats in all locations train followers to expect straw
Ž followers 20 minr
outer feeders — Straw in all locations at arm ends
. session
Pretest — followers 7
inner feeders — oats in all inner feeders — oats in all
inners feeders — oats in all confirms that followers were not
Ž .
20 minrsession outer feeders — barley in
outer feeders — barley in outer feeders — barley in
using olfaction to locate grain two random locations and
two random locations and two random locations and
straw in six locations straw in six locations
straw in six locations Exposure 1 —
8 inner feeders — oats in all
inner feeders — oats in all inner feeders — oats in all
facilitate information transfer leaders and followers
outer feeders — barley in outer feeders — barley in
outer feeders — barley in from leaders to followers
Ž Ž
together 30 minr same locations as during
same locations as during same locations as during
. session
pretest and straw in six pretest and straw in six
pretest and straw in locations
locations six locations
D.W. Bailey
et al.
r Applied
Animal Beha
Õ iour
Science 68
2000 93
– 105
99 Test 1 — followers
9 and 10 inner feeders — oats in all
inner feeders — oats in all inner feeders — oats in all
test rate of information transfer Ž
. 20 minrsession
outer feeders — barley in outer feeders — barley in
outer feeders — barley in from leaders to followers
same locations as during same locations as during
same locations as during exposure 1 and straw
exposure 1 and straw exposure 1 and straw
in six locations in six locations
in six locations Post-exposure
9 and 10 same as initial — oats in
same as initial — oats in same as initial — oats in
continue training of leaders to training — leaders
inner feeders. inner feeders
inner feeders respond to diffirent spatial
Ž .
20 minrsession barley in two outer feeders
barley in two outer feeders no access to outer feeders
treatments Ž
. Ž
. fixed location and six
locations changed daily with straw
and six with straw Exposure 2 —
11 same as exposure 1
same as exposure 1 same as exposure 1
same as exposure 1 leaders and followers
Ž together 30minr
. session
Test 2 — followers 13 and 14
same as test 1 same as test 1
same as test 1 same as test 1
Ž .
20 minrsession
affected foraging strategies. Behaviors evaluated were assists, total arm entrances before 1 and 2 barley feeders were discovered, and the total number of occasions that heifers
traveled to arm ends before finding one or two feeders with barley. Heifers often entered arms and consumed oats without traveling to arm ends. Thus, arm entrances were
indicators of the search for oats, and travel to arm ends indicated a search for barley.
Ž .
Average values from the last 2 days of initial training were evaluated. Groups n s 4 were used for pairing.
2.8.2. Followers For all three treatments, followers were treated identically during initial training and
the pretest. All followers received barley in the same fixed locations during the pretest, exposure, and test sessions. For followers, the only difference among treatments
occurred during the two exposure sessions when ‘‘leader’’ heifers that had previously Ž
. received different initial training i.e., fixed, variable, no-experience were allowed to
forage with the followers. The effect of leader initial training on the behavior of heifers assigned as followers
was analyzed as a randomized complete block design. Initial training regimes of leaders Ž
. Ž
. Ž
. were the treatments
n s 3 and heifer groups nine animals were blocks n s 4 .
Dependent variables analyzed were the total visits to barley feeders by followers during Ž
. Ž
. the two exposure with leader and four test sessions without leader and the total
Ž number of visits to barley feeders by followers during the four test sessions without
. Ž
. leaders . Observational data obtained from leaders including exposure sessions were
never included in analyses of follower data. Single degree of freedom orthogonal contrasts were used to determine if followers with experienced leaders differed from
followers with inexperienced leaders. We hypothesized that followers with experienced leaders would find barley feeders sooner and visit more barley feeders than those
exposed with inexperienced leaders.
A separate analysis using pooled data from all followers was used to evaluate if leaders facilitated the discovery of barley feeding sites by followers. A non-parametric
binomial test was used to determine if followers visited arm ends with barley more often than expected by chance. The percentage of time spent by leaders at feeders during
exposure sessions was calculated from observations to evaluate the probability that leaders were present at the time followers first found barley. Because followers had been
Ž .
trained to expect straw at arm ends and previous research Bailey and Sims, 1998 showed that cattle strongly avoided maze arms that contained straw, the timing in which
Ž .
followers visited barley feeders before, during or after a leader visit was used to evaluate if leaders facilitated the discovery of barley feeders by followers. Leaders from
all three treatments could provide new information since they were not trained to avoid arm ends.
3. Results and discussion