Content uploaded by Joy Ann Mench
Author content
All content in this area was uploaded by Joy Ann Mench
Content may be subject to copyright.
Content uploaded by Joy Ann Mench
Author content
All content in this area was uploaded by Joy Ann Mench
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
ENVIRONMENT, WELL-BEING, AND BEHAVIOR
Effect of Sand and Wood-Shavings Bedding
on the Behavior of Broiler Chickens
S. J. Shields,
1
J. P. Garner,
2
and J. A. Mench
3
Department of Animal Science, University of California, One Shields Avenue, Davis, California 95616
ABSTRACT The purpose of this study was to determine
the effect of 2 different bedding types, sand and wood
shavings, on the behavior of broiler chickens. In experi-
ment 1, 6 pens were divided down the center and bedded
half with sand and half with wood shavings. Male broilers
(10/pen) were observed by scan sampling at 5- or 12-min
intervals throughout the 6-wk growth period during the
morning (between 0800 to 0900 h), afternoon (1200 to
1500 h), and night (2300 to 0600 h). There was a significant
behavior × substrate × week interaction during the day (P
< 0.0001) and at night (P < 0.0002). Drinking, dustbathing,
preening, and sitting increased in frequency on the sand
side but decreased on the wood shavings side during the
(Key words: broiler chicken, behavior, bedding, sand, wood shavings)
2005 Poultry Science 84:1816–1824
INTRODUCTION
Broiler chick ens beco me incre asi ngl y inactive as they
near market we ight , spending a s much a s 80% of their
time resti ng (Murp hy a nd Prest on, 1988; Newberry and
Hall, 1990; Weeks et al., 2000). Inactivit y is probably largely
a consequence of selection and management for particular
growth characteristics. The rapid growth of breast muscl e
in broiler s mo ves the c ent er of gravity forward and the
legs outward, producing a gait pat ter n that is pro babl y
energetic all y inefficient and t iri ng (Corr et a l., 2003). In
addition, broilers may find walking painful (McGeown, et
al., 1999; Danbury, et al., 2000) as they approach slaughter
weight because they are increasingly prone to leg disorders
(Mench, 2004). However, a lack of a cti vity could in turn
increase the incidenc e of gait and skeletal disorders (Haye
and Simons, 1978; Thorp and Duff, 1988). The normal stress
and strain that is caused by exercise is important for me-
chanicall y organizing the growth process of bone into the
2005 Poultry Science Association, Inc.
Received for publication July 30, 2004.
Accepted for publication July 11, 2005.
1
Current address: University of Nebraska-Lincoln, Department of
Animal Science, PO Box 830908, Lincoln, NE 68583-0908.
2
Current address: Purdue University, Department of Animal Science,
125 South Russell Street, West Lafayette, IN 47907.
3
To whom correspondence should be addressed: jamench@
ucdavis.edu.
1816
day, as did resting at night. In general, broilers performed
a greater proportion of their total behavioral time budget
on the sand (P < 0.0001) as they aged. Broilers used the
divider between the 2 bedding types to perch; perching
behavior peaked during wk 4. In experiment 2, male broil-
ers were housed in 8 pens (50 birds/pen) bedded only
in sand or wood shavings. Bedding type had no effect
on behavioral time budgets (P = 0.8946), although there
were age-related changes in behavior on both bedding
types. These results indicate that when given a choice,
broilers increasingly performed many of their behaviors
on sand, but if only one bedding type was provided they
performed those behaviors with similar frequency on
sand or wood shavings.
proper patterns of twisting an d angulation (r evi ewed in
Lanyon, 1993; Rath et al., 2000).
A number of attempts have been made to increase the
activity levels of broilers. Example s include increasing the
distance between food and wate r sources (Haye and Si-
mons, 1978) or placing barriers between the food and water
(Bizeray et al., 2002a) to make broilers walk further to reach
resources . However, using restricted food and water access
as a means to promote locomotion in a commercial house
could create a potential welfa re problem, because broile r
chickens with seve re gait diso rde rs may have difficulty
reaching the food and water and, thus, starve or become
dehydrate d. Light management has also been used to in-
crease activity, for example by increasing the light intensi ty
(Newberry et al. 1985, 1988) or providing intermi tten t daily
lighting (Simons and Haye, 1985). Both methods stimulate d
activity, but only intermitte nt lighting decreased leg prob-
lems. However, complete control over the lighting schedule
under commerc ial conditi ons is not always possible. There
are also other drawbacks to using certain lightin g programs
to reduce leg probl ems , su ch as a poten tial inc rease in
breast blisters from longer periods resting on the keel bone
(Deaton et al., 1978) and a greater incidence of hock and
footpad burns (Sørensen et al., 1999).
Abbreviation Key: CI = confidence interval.
BEDDING TYPE AND BROILER BEHAVIOR 1817
Another way to increase activity levels might be to en-
courage broilers to display normal behaviors that require
energetic movemen t that includes exercise of the legs, for
example walking, foraging, and dustbathi ng behaviors (Ar-
nould et al., 2004). Providing broilers with an enrichment
device that stimulate d them to perch, traverse inclines,
forage, and dustba the resulted in a slight improvement in
gait score (Mench et al. 2001). A simpler and more cost-
effective way to incre ase broiler activity in commercial
houses might be to provide a bedding substrate that stimu-
lates particular behaviors. Sand appears to be one such
potential substrate. Broilers that are depri ved of bedding
and subsequen tly given a choice between sand, pine wood
shavings, rice hulls, and a recycled paper bedding product
choose to dustbathe and forage more in sand than in any
of the other substrates (Shields et al., 2004). When sand-
filled trays are placed in pens, broilers dustbathe and forage
preferent ial ly in the sand rather than in the wood shavings
covering the pen floor (Arnould et al., 2004).
Sand is being consider ed as an alterna tive to pine wood
shavings as bedding for broiler chickens in some areas of
the United States (Grimes et al., 2002 ). Litter quality and
bird performance paramet ers are similar for sand and
wood bedding, and sand is advantag eou s in that it harbors
fewer harmful microorganisms like Escherichia coli (Bilgil i
et al., 1999a,b). However , it is not known how the behavior
of broilers would be affected by the use of sand bedding
in commercial houses. The objectives of the 2 expe rime nts
presented here were to determine 1) whether broilers dif-
ferential ly perform particul ar behavior s on sand or wood-
shavings bedding when given a choice between bedd ing
types and 2) if behavioral time budget s differ between
broilers reared on sand and those reared on wood shavings.
The larger goal of thes e experi men ts was to det erm ine
whether sand bedding would promote the expression of
more active behaviors and, perhaps, improve leg condition.
MATERIALS AND METHODS
Experiment 1
Male Ross × Ross broiler chicks (n = 60) were purchased
from a local hatchery. At d 1 of age, the chicks were sepa-
rated into 6 different pens. Each pen measu red 3.05 × 3.05
m and ha d an overhead brood er and 2 circular feeders.
Each pen contained 2 cup waterers, which, along with the
feeders, were arrange d symmetricall y on each side. Feed
(Purina Mills Meat Builde r withou t added medi cat ion;
http://ww w.p urin ami lls .com) an d water were available
ad libitum . Each pen was divided down the center with a
3.8 cm w ide × 8.9 cm h igh pine board and filled to a
depth of 2.5 cm with pine wood shavi ngs on one side and
masonry-g rad e (construc tion ) sand on the other side. The
location of the 2 su bst rat es was alternate d be tween the
right and left sides of the pens. There were windows along
the length of the build ing that allowed daylight to enter,
but the study was conducted in the winter, so the days were
short. There was also fluorescen t light ing set to provide a
16L:8D cycle with the lights coming on at 0700 h and going
off at 2300 h. Although many different lighting programs
are used by the broiler industry, an 8-h scotophase is recom-
mended in some industry standa rds as a management
practice to reduce leg problems (e .g. , Certified Humane ,
2004). The birds were managed in accordance with the
Guide for the Care a nd Use of Agricultural Animals in
Agricultu ral Research and Teachi ng (FASS, 1999), and the
experimen ts described below were approved by the Uni-
versity of Califor nia Davis Institutional Animal Use and
Care Administrative Advisory Committee.
Behaviora l observations were conducted 5 d per week
starting when the chicks reached 7 d of age and continuing
through 49 d of age. Each pen was observed 5 times a week
during the afternoon (1200 to 1500 h). Each observation was
1 h long. Because there are circadi an rhythms in behavior,
pens were also observed once per week beginning at 0800
h to ensure a better approximation of behav ior s that are
more commo nly performed in the morning. There were
more afternoo n observ ati ons than mor ning obs ervations
because parallel observa tion s were being conducted to per-
form a d eta ile d analysis of the structure of du stb athing
behavior, whi ch occurs primar ily in the afternoon. The
results of these pa ral lel studies are reported el sewh ere
(Shields, 2004).
At the beginning of an obse rvat ion session, an observer
satquietlyabout2mawayfromthefrontofapen,allowed
5 min for the chicks to habituate to the observer’s presence,
then start ed a stopwatch and recorded data with pen and
paper. Instantaneous scan samples (Martin and Bateson,
1986) of all the birds in a pen were conducted at the start
of the observation and continued through the hour at 5-
min intervals. During a scan sample, the location (i.e., side
of the pen) and the number of individua ls engaged in a
behavior were recorded. Behaviors were divided into the
following categories : stand, locomote, preen standing,
scratch, feed, peck, sit, preen sitting, drink, dustbathe, perch
(on the board that divided the 2 types o f bedding), and
other. Mos t be havi ors recorded in the “other” category
were aggressive pecks, threats, or chases. A Latin square
was used to determine the order in whic h the pens were
observed each week. Six different observers (S. J. Shield s
and 5 assistants) collected data.
Behaviors occurring at night were videotaped under red
light when the birds were 34, 35, 38, 41 to 43, and 46 to 50
d of age. On each of these nights, video recordi ng started
at 2300 h and continued until 0600 h. Data were then taken
from 1-h segments of the videos starting at 2300, 0200, and
0500 h. For each hour-long segment, an instantane ous scan
sample was performed every 12 min, providing 6 samples
per hour of video. The location and behavior of each bird
were recorded during each scan. The only behav ior s dis-
cernible on the videos were stand, locomote, feed, drink,
perch, and rest. Be cau se it was difficult to see whether
the broile rs h ad th eir eyes open, the c ateg ory of resting
included sleeping and sitting awake.
A score for each behavior was obtained for each observa-
tion by summin g the num ber of bi rds engaged in that
behavior over all of the 5-min (or 12-min, for night observa-
tions) scan sa mple s in a particular observation. Separate
SHIELDS ET AL.1818
scores wer e calculated for the sand side an d th e wood
shavings side of the pen. Perching could not occur on one
side or the other because the perch actually separated the
2 bedding substrates; thus, this behavior was excluded
from these calculatio ns. Because some behaviors were rare,
there were several scores with the value zero, which re -
sulted in a substantial floor eff ect (Martin and Bateson ,
1986). Therefore, we summed the behavioral scores for all
of the observation s so tha t there was one total score for
each behavior in each pen f or each week. Thi s removed
most of the zeros from the data set. Summing the observa-
tions over the week also prevented the problem of colinear -
ity (lack of independence among in dep ende nt variables)
within the behavior × substrate term or other higher-orde r
terms. Because there were 6 pens and 6 wk, summing the
values from each observation created a total of 36 behav-
ioral scores o n sand and 36 behavioral scores on wood
shavings for each of the differ ent behavior categories. To
express each behavioral score as a proportion of the total
behavior performed in a given week and pen, the behav-
ioral score on each substrate was then divided by the total
score for all the behaviors in that pen on both sides for that
week. Dividin g each behaviora l score by the total possible
behaviora l score crea ted a popula tio n time budge t that
could then be subjecte d to statistica l analysis.
The category of “other” was excluded from the statistical
analysis because there we re relatively few sc ore s above
zero in this catego ry. As a result the total budget for each
pen in each week used in the analysis su mmed to less
than 100%, the reby elimin atin g colinearity between the
behaviors within the analysis.
Analyses were performed in the SAS software (SAS,
2000) usin g the PROC GLM procedure . All analyses were
blocked by pen, which was treated as a fixed effect. Trans-
formation s were applied where needed to meet the as-
sumptions of GLM (homogen eity of variance, norma lity
of error, and linearit y). The effectivene ss of these tran sfo r-
mations was confirmed using posthoc plots.
To examine the effect of bedding upon the overall time
budget of the birds in a single analysis, the effects of behav-
ior, week, and substrate on time budget proportion were
examined. The GLM model for the analysis was proportio n
of the t ime bud get taken up by ea ch behavior = pen +
behavior|substrate|week, with week as a cont inu ous vari-
able. In such an analysis the behavior term descri bes the
shape of the time budget, a behavior × substrate interacti on
describes how the budget differs on the 2 bedding types,
and a behav ior × substrate × wee k in ter acti on describes
how this differenc e changes with time (for further details
see Chu et al. 2 004) . Therefore, th e behavior × substrate
× week interac tion was exa mine d to test for differ enti al
changes in the time budget s over time (i.e. , week) on the
2 beddings. Week was treated as a continuou s variable to
explicitl y test for progressive c han ges over tim e and to
account for th e fact th at observations made cl ose to on e
another in time will be more simila r to each other than
those s epa rat ed in time. The data in th is an alysis were
Box-Cox trans form ed with k = 1.25 to pr oduc e the best
error structure.
Bonferron i-c orre cte d posthoc estimates were calculated
for the rate of change in each behavior with time on each
bedding type. Bonferroni-correc ted posthoc contrasts were
performed to compare these rates o f ch ange and , th us,
determine whether any change in a behavior with time
(i.e., week) differed between the 2 beddings. The change
in overall use of each bedd ing t ype with time and the
overall budgets on each bedding type were examine d using
posthoc contrasts equivalent to the week × substrate and
substrate × behavior interaction s, r espe cti vel y. For each
posthoc test, a combined P ≤ 0.05 for all compar iso ns was
considere d statisticall y significant.
Pecking, scratchi ng, and feeding often occurred together,
and so they were combined together into a new behavioral
category, foraging. This behavioral categor y includ es be-
haviors associated with high levels of locomotion and so
can be considered a good general indicator of activity (Biz-
eray et al., 2002b). The 3 behaviors were summe d into a
single score and conve rte d to a propor tion of the total
behaviora l time budget to obtain a separate analysis for
foraging. We analyzed thi s for agi ng score using a GLM
blocked by pen. The independent variables of interest were
substrate and week (which again was treated as a covari-
ate). Diff ere ntia l ch ang es in foraging behavior on th e 2
bedding types over time (i.e., week) were test ed by the
substrate × week interact ion. However, the error structure
revealed nonl inea rit y, and a quadratic model was found
to provide a better fit. Data were angula r transf orm ed for
this analysis.
Because perching occu rre d between th e 2 sides of the
pen and could not be included in the analysis of the behav-
ioral time budget, it was also analyze d separately in a GLM
blocked by pen, in which week was the independe nt factor.
Week was treated as a continuous variable; however, post-
hoc assessment of the error structure revealed evidence of
nonlinear ity . Therefore a GLM blocked by pen was used to
perform a polynomial regressi on, which provided a much
better fit to the data. To confirm these results the analysis
was also run with week as a categori cal variable. Data were
angular transformed for this test. Tukey posthoc tests were
used to det erm ine significant differen ces between weeks
in the anal ysi s whe re w eek was treated as a cate gorical
variable.
Because ni ght video was only taken during the last 3
wk, day-of -ag e was used a term in the analysis inst ead of
the week te rm to provide sufficient d ata resolut ion over
time. The anal ysis was otherwise the same a s that p er-
formed for the daytime time budgets. The behavior × sub-
strate × day-of-age interacti on was examined to test f or
different ial changes in the time budgets over tim e (i. e.,
days of age) on the 2 beddings. Perching behavior was not
included in this analysis because it could not be categorized
as occurring on a particular side. Data were log trans-
formed for this analysis. Posthoc analyses were performed
as for the daytime time budgets.
Experiment 2
Four hundred male Cobb broiler chicks were purchased
from a local hatchery and distributed evenly among 8 pens.
BEDDING TYPE AND BROILER BEHAVIOR 1819
The same types of pens were used in this study as in the
first experiment. Overh ead fluoresce nt light ing was the
same as in experiment 1 (8D:16L) with the lights coming
on at 0700 h. Because th e study started in Octob er, the
days began to get shorter as the study progressed. Four
pens were bedded approximately 17.8 cm deep in masonry-
grade sand, and the other 4 pens were bedded in pine wood
shavings to approximatel y the same depth. Husbandry was
as previously described.
Beginning when the chi cks reached 1 wk of age, e ach
pen was observed fo r 1 h, 4 d per week. Ob serv ati ons
continued until the chicks reache d 7 wks of age. Two of
the weekly observations on each pen were performed in
the morning (between 0800 and 1200 h), and 2 observations
were performed in the afternoon (between 1200 and 1600
h). During each 1-h observatio n, a trained observer sat in
front of the pen and recorded behavioral data as described
for experiment 1.
Scores for each behavior were summed and analyzed as
in experiment 1. The behavioral categories used in experi-
ment 2 were slightl y different from the ones used in the
first study. The category “feed” was split into the catego-
ries, “feed sitting” and “feed standing” for the time budget
analysis, although both were include d in the “foragin g”
category. The category of “other” was excluded from the
statistic al analysis to reduce the floor effect, but the behav-
ioral categories of “chase” and “a ggr essi on” were left in
the analysis because inspecti on of the data revealed that
there were mor e scores above zero than in the pr evi ous
experimen t. Analyses of the overall behavioral time budget
and foraging behavior were performed as previous ly de-
scribed, except that pen was now nested within substrate
rather than crossed with substrate. Hence, the GLM model
for the time budget analysis became: Proporti on of the time
budget taken up by e ach behavio r = pen (substrate) +
behavior|substrate|week, with week as a continu ous
variable.
RESULTS
Experiment 1
The time budgets on the 2 bedding types changed differ-
ently over time (be hav ior × substrate × week in ter acti on
F
9,675
= 4.39, P < 0.0001; Figure 1). Posthoc contrasts compar-
ing the rate of change in behavior between the 2 bedding
types showed a significant difference in the rate of change
of drinking, dustbathing , preening while sitting, and sitting
behaviors on the 2 different beddi ngs . Postho c estimates
of the rate of c hang e in beh avio r on eac h bedding type
revealed that there was a significant decrease in locomotion
and standing behavior on both bedding types. In addition,
the behaviors, drink, peck and scratch, did not change
significantly on the sand, but decre ase d significant ly on
the wood shavi ngs. The re w as a sign ificant incre ase in
sitting on the sand.
Posthoc contr asts rev eal ed th at, ove rall, the behaviors
drink, dustbathe, locomote, peck, preen standin g, and
standing were significantly different on the 2 sides of the
pen (F
9,675
= 6.61, P < 0.0001). The proportion of the total
behaviora l time budget made up by each of these behaviors
was significantly higher on the sand side of the pens (Figure
2). Post hoc contrasts also revealed that the birds performed
a greater proportion of their total behaviora l time budget
on the sand as the experime nt progr ess ed (F
1,675
= 57.31;
P < 0.0001).
Foraging beha vior on the 2 bedding types chang ed dif-
ferently over time (F
1,61
= 7.89; P = 0.0067; Figure 3). Posthoc
analysis showed that foraging did not change significantly
in the sand but decreased significantly in the wood shav-
ings. The p rop orti on of the total behavior al time budget
that was spent foraging in sand did not change significantly
over the cours e of the stud y. The proportion of the total
behaviora l budget that was spent foraging in the w ood
shavings decreased significantly.
Perching was first noticed when chic ks were less than 1
wk old. Visual inspection of the data suggested that perch-
ing decreased from wk 1 to 2, then increased at wk 4, and
then decreased again to wk 6. The data were, therefore, fit
to a cubic polynomial. There was a significant week × week
× week interac tion (F
1,63
= 8.30, P = 0.005), indicating that
perching decr ease d, increased, and then decr eas ed again
as a pro gre ssi ve function of age. When the a nalysis was
rerun with week as a cat ego rica l va ria ble, these results
were confirmed (F
5,61
= 6.73, P < 0.001). Perching declined
from wk 1, when it made up 3.0% [95% confidence interval
(CI): 2.1 to 3.9%] of the popula tio n time budge t, to wk 2,
when it was 1.8% (95% CI: 1.1 to 2.5% ) of the population
time budget. In wk 3 perching was 1.9% (95% CI: 1.3 t o
2.7%) of the popula tio n ti me budge t. I ts frequency then
increased in wk 4, when it reached 2.9% (95% CI: 2.1 to
3.9%) of the ti me budget, and then finally decreased to
1.5% (95% CI: 0.9 to 2.2%) and 0.7% (95% CI: 0.3 to 1.2%) of
the total behavioral time budget in wk 5 and 6, respecti vel y.
Tukey pairwis e compariso ns revealed that there was more
perching during wk 1 than in wk 6 and significantly more
perching in wk 3 and 4 than in wk 6.
For nighttime behavior, there was a significant behavior
× day-of-age × substrate interaction (F
4,755
= 5.50, P =
0.0002), indicating that the time budgets on the 2 bedding
types changed with age. Bonferron i-c orre cte d posthoc esti-
mates show ed t hat r est ing increased significantly from
13.9% (95% CI: 9.4 t o 20.5% ) of the tota l behavi oral time
budget at 34 d of age to 61.2% (95% CI: 42.2 to 88.7%) of
the total behavioral time budget at 50 d of age) on the sand
and decrea sed nonsignificantly from 27.5% (95% CI: 18.7
to 40.4%) to 12.0% (95% CI: 8.2 t o 17.6%) on the wood
shavings as the birds aged. Posthoc contra sts showed that
this diffe ren ce in rates of change of r esti ng b ehavior on
the 2 be ddi ng types was sig nificant. None of the o the r
behaviors showed significant diffe ren ces between the 2
bedding ty pes in their rate of c han ge w ith a ge. Posthoc
contrasts at the mean age of the birds in the analysis (42
d) showed that the beha vio ral time budget differe d on the
2 bedding types (F
4,755
= 5.36, P = 0.0003) and that drinking,
feeding, resting, and standing were all performed more on
the sand side of the pen (Figure 4).
SHIELDS ET AL.1820
Figure 1. Significant change in behavior as a proportion of the total behavioral time budget over the course of experiment 1. All figures are
residual plots with each data point plotted as its residual from the least squares lines (LSL). The data points represent the behavioral time budget
score corrected for pen. The significant posthoc tests for each behavior are indicated beneath the panel letter as follows: SνW = the change in
behavior over time differed between sand and wood shavings, S+ = the behavior increased significantly in the sand over time, S− = the behavior
decreased significantly in the sand over time, and W− = the behavior decreased significantly in the wood shavings over time. Each panel of the
graph indiciates a different component of the total time budget: (a) drinking, (b) dustbathing, (c) preening while sitting, (d) sitting, (e) locomotion,
(f) standing, (g) pecking, (h) scratching, (i) preening while standing.
Experiment 2
Rather than providing both bedding types in each pen,
in experiment 2 each pen contained only 1 of the 2 bedding
substrate s. The behavioral time budgets did n ot change
different ly with time on the 2 types of bedding (the week
× treatmen t × b ehav ior interaction was not significant:
F
12,566
= 0.53, P = 0.8946). However , planne d posthoc con-
trasts revealed that overall the behavioral time budgets did
change significantly with week on both substrat es (F
12,566
=
20.78, P < 0.00 01; Table 1). Thus, posthoc Bonferroni-cor -
rected comparisons showed that aggre ssio n, chasing, feed
standing, locomotion, pecking, and standing all decreas ed
significantly with time in both treatments, whereas preen
sitting and sitting increased significantl y on both treat-
ments with time.
The separate analysis of foraging confirmed the lack of
an effect of bedding. There was no differen ce in foraging
behavior betw een t he 2 bedding typ es ( F
1,38
= 2.62, P =
0.1141). However, there was an overall effect of time (i.e.,
week; F
1,38
= 26.90, P < 0.0001), such that the time spent
foraging on both bedding types declined from 26.2% (95%
CI: 24.4 to 28.1%) in wk 1 to 18.7% (95% CI: 17.1 to 20.4%)
in wk 6.
DISCUSSION
In experiment 1, bedding type did influence behavior but
not only in the way expected. Active behaviors (dustbathe,
locomote, and peck) were p erfo rme d mo re of ten on the
sand, but so too were inactive behaviors (resting and sit-
ting). Furthermore, ther e was no significant difference be-
BEDDING TYPE AND BROILER BEHAVIOR 1821
Figure 2. Overall distribution of behaviors on each side of the pen as shown by the proportion of total behavioral time budget for each behavior.
Error bars depict SEM. *Significant difference at P ≤ 0.05.
tween the behavior s of broilers given only sand or wood
shavings in experi men t 2. Thus there was no suppor t for
the idea that exercise could be increased and leg problems
decreased by housing broilers on sand bedding. Although
the present experiment did not show that active behaviors
were increased by the provision of sand bedding, i t did
show th at the broilers preferred sand to wood shav ings
when they were given a choice.
It was interesting that the birds sat and rested more on
the sand rather than on the wood shavings, because wood
shavings subjectively seem to be a softer substra te. There
might be a perc eptu al difference in the way sand appears
to broilers or in the way it feels on their feet and in their
plumage. Clea nlin ess , temper atu re at lower depth in the
bedding, odor, or some other characteristic of the bedding
may be more important than the softnes s of the bedding
for resting. Bilgi li et al. (1999a) found that sand bedding
in commercial houses is cleaner than wood shavings in
that it harbors fewer microorg ani sms , such as Escherichi a
coli. Althoug h they found no differen ce between sand and
wood shavings in moisture content, temperature, or ammo-
nia production, the qu ali ty of sand an d wood shavings
may be differe nt deeper in the sand, where broiler chickens
create small depressions in which to rest. Bedding depth
might also be a factor in the preference observed in experi-
ment 1. The bedding depth in this experiment was rather
shallow, particularl y as compared with the shavings depth
typically found in a comme rcia l house. It is possible that
shallow sand provides more comfort or has better insula-
tive properti es than shallow wood shavings, although this
would need to be assessed experime nta lly.
Pecking and scratching decreased on the wood shavings
side of the pen but increas ed on the sand side, which could
occur for several reasons. One possibilit y is that scratching
and pecking are usually perfor med pri or to a dustbath,
and there was signi ficantly more dustba thin g on the sand
side of the pens as the birds aged. Anot her possibility is
that the decrease in these behaviors reflects the changing
condition of the litte r. Sand may stay more fri abl e than
wood shavings and, theref ore, might be an easier or more
rewarding surface in whic h to scratch and peck . Foraging
also decli ned significantly on the wood shavings and in-
creased on sand. The texture or mix of colors of the sand
might have elicited feeding more than the wood shavings,
perhaps because it was more similar to the type of surfa ce
on which fowl would forage naturally. It is also possible
that sand particles resembled the small stones that chickens
ingest if they have access to them. The motivation to find
and ingest small pebbles for grit may still be strong despite
the lack of need for poultry to have grit to dige st commer-
cially prepared poultry feed.
SHIELDS ET AL.1822
Figure 3. Change in foraging behavior on each side of the pen by
week. Foraging behavior did not change significantly in the sand but
did decrease significantly in the wood shavings.
One une xpe cte d result fro m the first experiment was
that drinking behavio r tended to increase significantl y on
the sand side of the pen and that at night birds also drank
more on the sand side. This could simply be beca use the
birds drank at the d rin ker nearest to them, and because
they spent a greater proportion of their time on the sand
tended to be closer to the sand side drinker . However,
there was no differenc e in the amou nt of feeding behavior
that occurred on each side during the day. Beca use eating
and drinking usually occur together, it would be expected
that drinking would also occur approximat ely equally on
Figure 4. Overall distribution of behaviors on each side of the pen at night. *Significant differences at P ≤ 0.05.
both sides of the pen, but this was not the case. Another
possibili ty is that pr ope rtie s of the bedding affected the
quality of the water. Bedding sometimes got into the auto-
matic cup waterers, and the wood shavings became sus-
pended in the water to a greater extent than the heavier
sand. The result was that the water tended to stay cleaner
on the sand side of the p en, which might be on e reaso n
that the birds walked to the sand side to drink. Also, when
water spilled out onto the bedding the wood shavings in
the area around the drinker became wetter than the sand.
One advantage of using sand as bedding, then, would be
that the water supply and the area around the water supply
stays cleaner, at least when cup watering systems are used,
as in the present study.
Our results agree with previous work showing that
broiler chickens become increasingl y inactive (Murphy and
Preston, 1988; Newberry and Hall, 1990; Weeks et al., 2000).
In the experiments presented here, behaviors such as sitting
and preen sitting increased with time, whereas behaviors
that required more energy expenditur e such as locomotion,
stand feeding, standing, aggression and chas e decreased
on both bedding types. In contrast , lighter-t ype breeds such
as Leghorns are much more active than broilers , displaying
behaviors such as aggression, running, and frolickin g when
they are 6 wk old (Mench, 1988). As mentione d previousl y,
this increasi ng inactivity is probably largely a consequence
of rapid growth rate and its associ ate d effects on body
conformat ion and leg problems . Slow-growing broilers are
much more active at 6 weeks of age than are fast-gro wing
broilers (Bokkers and Koene, 2003). The motivati on to move
around in familiar surroundings may also be lacking when
the housing environment for broilers does not provide
BEDDING TYPE AND BROILER BEHAVIOR 1823
Table 1. Change in the behavioral time budget over the 6-wk period of experiment 2
Change in behavioral time budget
Week 1 Week 6 Significant
Behavior mean (95% CI)
1
mean (95% CI)
1
change?
2
Aggressive 1.1% (0.7% − 1.5%) 0.2% (0.1% − 0.5%) Yes
Chase 0.6% (0.3% − 0.9%) 0.0% (0.0% − 0.0%) Yes
Drink 3.0% (2.3% − 3.8%) 3.0% (2.3% − 3.7%) No
Dustbathe 1.0% (0.6% − 1.5%) 1.0% (0.6% − 1.5%) No
Feed Sitting 1.1% (0.7% − 1.6%) 1.2% (0.8% − 1.7%) No
Feed Standing 17.1% (15.5% − 18.7%) 12.2% (10.8% − 13.7%) Yes
Locomote 5.0% (4.1% − 6.0%) 1.8% (1.3% − 2.4%) Yes
Peck 6.5% (5.5% − 7.6%) 4.0% (3.2% − 4.9%) Yes
Preen sitting 3.3% (2.6% − 4.2%) 7.4% (6.3% − 8.6%) Yes
Preen standing 2.2% (1.6% − 2.9%) 1.0% (0.6% − 1.5%) No
Scratch 0.6% (0.3% − 1.0%) 0.1% (0.0% − 0.3%) No
Sit 47.0% (44.9% − 49.2%) 64.2% (62.1% − 66.2%) Yes
Stand 10.1% (8.8% − 11.4%) 2.9% (2.2% − 3.7%) Yes
1
Mean percentage of time the behavior occupied in the time budget; the 95% confidence interval (CI) is shown
in parenthesis
2
Significance level of P < 0.05.
much complexity or novelty (Newbe rry , 1999). There are
undoubted ly other factors that contribut e to the decrease
in activity as broilers age, and these may become apparent
with additional experimental work.
In the first experiment, in which pen s were divid ed,
broilers used the divider to perch. As in other experiment s
that have reported that broilers do not use perches exten-
sively (Hughes and Elson, 1977; Pettit-Ri ley and Estevez,
2001; Estevez et al., 2002 ), perch ing behavior was only a
relativel y small percentage of the time budget (0.7 to 3.0%).
Also in agreement with other studi es (LeVan et al., 2000;
Pettit-Ri ley and Estevez, 2001), perching peaked in wks 3
and 4 and then decreased, proba bly as a consequenc e of
growth-re lat ed changes in body conformation or leg
soundness that make perching increasingly difficult.
In experiment 1, the finding that dustbathing was per-
formed more on the sand than on the wood shavings is in
agreement with previous work in our laboratory showing
that broilers prefer sand to wood shavings, paper bedding,
or rice hulls for dustbathing (Shields et al., 2004). Laying
hens also prefer to dustbathe in sand rather than in wood
shavings or straw (Petherick and Duncan, 1989; van Liere,
1991; Sanotra et al., 1995). The results of experiment 2 were
more surprising, because based on experiment 1 it was
expected t hat broi ler s ho used on sand bedding would
show more d ust bath ing behavio r than those hou sed on
shavings. Instead , the pattern of dustbathing behavior, as
well as other behaviors, was similar on both bedding types.
This finding suggests that the chick ens ’ behavi ora l time
budget was relatively inflexible and that they adjusted to
the less preferred substrate.
The results of the second experime nt may suggest that
the prefer enc e for sand is a weak preferenc e. But it is also
possible that the motivat ion to perfo rm the behaviors that
we meas ure d is so high that broiler s will perform them
even if the conditi ons are not idea l. Or it may be that
other factors that differed between the 2 experim ents , for
example the strain of broi lers used or the dept h of the
bedding, caused a differe nt pattern of substrate usage in
the 2 studies. Obviously, more testing is warranted.
Several concl usio ns can be drawn from these studies.
First, under the conditions in our experiment, broiler chick-
ens rested more on sand, suggestin g that there was some
differenc e between sand and wood shavings that made the
sand a preferred resting substrate. These observations also
confirmed that broil ers become increasingl y inactive as
they age. Finally, these studies demonstrate d that, when
given a choice, the broilers dustbathed, foraged, and drank
more on sand than wood-shaving s beddi ng but that their
behaviora l time budget was similar on wood-shavi ngs bed-
ding when no choice was provided. Beca use there was no
differenc e in activity levels when broilers were raised on
only shavings or sand, it is unlike ly that the provision of
sand bedding would improve leg condition due to exercise-
related effec ts. Simi lar ly, Arno uld et al. (2004) found that ,
although broilers are attracted to trays of sand placed in
their pens and use the sand preferentially for dustbathin g
and foraging, providing sand trays has little effect on over-
all locomotor activity and does not decrease leg problems.
The present experiments demonstrate d a complex rela-
tionship between broiler behavior and bedding type. From
these resu lts it could not be determined definitive ly
whether the broilers chose the sand to perform their activi-
ties or whether instead they performed more of their activi-
ties on the sand side simpl y becaus e they prefe rre d to
spend thei r time on the sand. Further testi ng is, therefore,
needed to determine the various motivating factors under-
lying b roi ler s’ ch oic es of particular beddin g ty pes. Also,
the experi men tal conditions in these studies differed in
several important respects from commercial rearing condi-
tions, and so it would is important to examine the effects on
behavior of such variables as bedding depth and condition,
stocking density, and lighting cycle.
ACKNOWLEDGMENTS
This project w as supported by the National Research
Initiativ e of the USDA Cooperative State Researc h, Educa-
SHIELDS ET AL.1824
tion and Extension Service (grant number 2001-02498). The
authors also thank the William and Charlotte Parks Foun-
dation for financial sup por t, th e student assis tan ts who
helped collec t data (Emil y Blake, Cle ide Falc one , Sarah
McClellan d, Leslie Gu sta fson , Kristen Ro berson, Bradley
Swagart, Sharon Tam, Ashley Thorne, and Bret Weaver),
Brian Bennett for help setting up pens, and the staff at the
Hopkins Avian Science Research Center at the University
of California, Davis, for their expert animal care.
REFERENCES
Arnould, C., D. Bizeray, J. M. Faure, and C. Leterrier. 2004. Effects
of the addition of sand and string to pens on use of space,
activity, tarsal angulations and bone compos ition of broiler
chickens. Anim. Welf. 13:87–94.
Bilgili, S. F., G. I. Montenegro , J. B. H ess, and M. K. E ckman.
1999a. Live performance, carcass quality, and deboning yields
of broilers reared on san d as a litter source. J. Appl. Poult.
Res. 8:352–361.
Bilgili, S. F., G. I. Montenegro , J. B. H ess, and M. K. E ckman.
1999b. Sand as litter for rearing broiler chickens. J. Appl. Poult.
Res. 8:345–351.
Bizeray, D., I. Estevez, C. Leterrier, and J. M. Faure. 2002a. Influ-
ence of increased environmental complexity on leg condition,
performance, and level o f fearfulness in broilers. Poult. Sci.
81:767–773.
Bizeray, D., C. Leterrier, P. Constantin, G. Le Pape, and J. M. Faure.
2002b. Typology of activity bouts and effect of fearfulness on
behaviour of meat-type chickens. Behav. Proc. 58:45–55.
Bokkers, E. A. M., and P. Koene. 2003. Behaviou r of fast- and
slow growing broilers to 12 weeks of age and the physical
consequences. Appl. Anim. Behav. Sci. 81:59–72.
Certified Humane. 2004. Broiler chickens. Humane Farm Animal
Care, Herndon, VA.
Chu, L., J. P. Garner, and J. A. Mench. 2004. A behavioral compari-
son of New Zealand White rabbits (Oryctolagus cunicu lus)
housed individually or in pairs in conv ention al laboratory
cages. Appl. Anim. Behav. Sci. 85:121–139.
Corr, S. A., M. J. Gentle, C. C. McCorquodale, and D. Bennett.
2003. The effect of morphology on walking ability in the mod-
ern broiler: A gait analysis study. Anim. Welf. 12:159–171.
Danbury, T. C., C. A. Weeks, J. P. Chambers, A. E. Waterman-
Pearson, and S. C. Kestin. 2000. Self-selection of the analgesic
drug carprofen by lame broiler chickens. Vet. Rec. 146:307–311.
Deaton, J. W., F. N. Reese, and J. L. McNaughton. 1978. Effect of
intermittent light on broilers reared under moderate tempera-
ture conditions. Poult. Sci. 57:785–788.
Estevez, I., N. Tablante , R. L. Pettit-Riley, and L. Carr. 2002. Use
of cool perches by broiler chickens. Poult. Sci. 81:62–69.
FASS. 1999. Guide for the Care and Use of Agricultural Animals
in Agricultural Research and Teaching. Fed. Anim. Sci. Soc.,
Savoy, IL.
Grimes, J. L., J. Smith, and C. M. Williams. 2002. Some alternative
litter materials used for growing broilers and turkeys. World’s
Poult. Sci. J. 58:515–526.
Haye, U., and P. C. M. Simons. 1978. Twisted legs in broilers. Br.
Poult. Sci. 19:549–557.
Hughes, B. O., and H. A. El son. 1977. The use of perches by
broilers in floor pens. Br. Poult. Sci. 18:715–722.
Lanyon, L. E. 1993. Skeletal responses to physical loading. Pages
485–505 in Physiology and Pharmacology of Bone, Handbook
of Experimental Pharmacology. G. R. Mundy and T. J. Martin,
ed. Springer-Ver lag, Berlin.
LeVan, N. F., I. Estevez, and W. R. Stricklin. 2000. Use of horizontal
and angled perches by broiler chickens. Appl. Anim. Behav.
Sci. 65:349–365.
Martin, P., and P. G. Bateson. 1986. Measuring Behaviour. Cam-
bridge Univ. Press, Cambridge, UK.
McGeown, D., T. C. Danbury, A. E. Waterman-Pearson, and S.
C. Kestin. 1999. Effect of carprofen on lameness in chickens.
Vet. Rec. 144:668 –671.
Mench, J. A. 1988. The developme nt of aggressive behavior in
male broiler chicks: A comparison with laying-type males and
the effects of feed restriction. Appl. Anim. Behav. Sci.
21:233–242.
Mench, J. A. 2004. Leg problems. Pages 3–17 in Measuring and
Auditing Broiler Welfare. C.A. Weeks and A. Butterworth ed.
CAB, Wallingford , UK.
Mench, J. A., J. P. Garner, and C. Falcone. 2001. Behavioral activity
and its effects on leg problems in broiler chickens. Pages 152–
156 in Proc. 6th Eu r. Symp. Poult. Welf. H. Oester and C.
Wyss, ed. World’s Poultry Scien ce Association, Zollikofen,
Switzerland.
Murphy, L. B., and A. P. Preston. 1988. Time-budgetin g in meat
chickens grown commercially. Br. Poult. Sci. 29:571–580.
Newberry, R. C. 1999. Exploratory behaviour of young domestic
fowl. Appl. Anim. Behav. Sci. 63:311–321.
Newberry, R. C., and J. W. Hall. 1990. Use of pen space by broiler
chickens: Effects of age and pen size. Appl. Anim. Behav. Sci.
25:125–136.
Newberry, R. C., J. R. Hunt, and E. E. Gardiner. 1985. Effect of
alternating lights and strain on behavior and leg disorders of
roaster chickens . Poult. Sci. 64:1863–1868.
Newberry, R. C., J. R. Hunt, and J. E. Gardiner. 1988. Influence
of light intensity on behavior and performance of broiler chick-
ens. Poult. Sci. 67:1020–1025.
Petherick, J. C., and I. J. H. Duncan. 1989. Behaviour of young
domestic fowl directed towards different substrates. Br. Poult.
Sci. 30:229–238.
Pettit-Riley, R., and I. Estevez. 2001. Effects of density on perching
behavior o f broiler chickens. Appl. Anim. Behav. Sci.
71:127–140.
Rath, N. C., G. R. Huff, and J. M. Balog. 2000. Factors regulating
bone maturity and strength in poultry. Poult. Sci. 79:1024–1032.
Sanotra, G. S., K. S. Vestergaard, J. F. Aggery, and L. G. Lawson.
1995. The relative preferences for feathers, straw, wood-shav-
ing and sand for dustbathing, pecking and scratching in do-
mestic chicks. Appl. Anim. Behav. Sci. 43:263–277.
SAS Institute. 2000. Version 8 for Windows. SAS Institute Inc.,
Cary, NC.
Shields, S. J. 2004. Dustbathing by broiler chickens: Characteristics,
substrate preference, and implications for welfare. Ph.D. Diss.
University of California, Davis, CA.
Shields, S. J., J. P. Garner, and J. A. Mench. 2004. Dustbathing by
broiler chickens: A comparison of preference for four different
substrates. Appl. Anim. Behav. Sci. 87:69–82 .
Simons, P. C. M., and U. Haye. 1985. Intermittant lighting has a
positive effect on twisted legs. Poultry Mag. (Mar.):34–37.
Sørensen, P., G. Su, and S. C. Kestin. 1999. The effect of photope-
riod:scotoperio d on leg weakness in broiler chicken s. Poult.
Sci. 78:336–342.
Thorp, B. H., and S. R. I. Duff. 1988. Effect of exercis e on the
vascular pattern in the bone extremities of broiler fowl. Res.
Vet. Sci. 45:72–7 7.
van Liere, D. W. 1991. Function and organization of dustbathing
in laying hens. Ph.D. Diss. Wageningen Univ., Wageningen,
The Netherlands.
Weeks, C. A., T. D. Danbury, H. C. Davies, P. Hunt, and S.
C. K estin. 2000. The behaviour of broil er chickens and its
modification by lameness. Appl. Anim. Behav. Sci. 67:111–125.