Grass eating patterns in the domestic dog, Canis familiaris

Abstract

Very little is known about grass eating behaviour in the domestic dog, Canis familiaris. This study is the first to investigate grass eating in dogs in a controlled experiment, and attempts to provide an initial understanding of this behaviour by describing the pattern of grass eating during the day and the relationship between grass eating and the ingestion of food. Twelve dogs were presented with both kikuyu and couch grass three times daily for 6 d and grass eating behaviours were observed using an all-occurrences sampling method. The results of this study suggest that grass eating is influenced by satiety and time of day. Dogs spent more time eating grass before ingestion of their kibble meal than after, and the time spent eating grass decreased throughout the day. Grass may be seen as a food source, as the subjects were less likely to eat grass when they were satiated. Couch and kikuyu grasses were equally preferred. We conclude that grass eating is a normal and common behaviour, as all dogs in this study were in good health and readily ate grass. As such, grass-eating should not be seen as a problematic behaviour for most dogs or as indicative of illness.

Full text

44
45
Grass eating patterns in the domestic dog, Canis
familiaris
S.J. Bjone1,2, W.Y. Brown3 and I.R. Price1
1University of New England, School of Psychology, Armidale, New South Wales, 2351, Australia;
2 sbjone@une.edu.au; 3University of New England, Animal Science, Armidale, New South Wales, 2351, Australia
Summary
Very little is known about grass eating behaviour
in the domestic dog, Canis familiaris. This study
is the rst to investigate grass eating in dogs in a
controlled experiment, and attempts to provide an
initial understanding of this behaviour by describing
the pattern of grass eating during the day and the
relationship between grass eating and the ingestion of
food. Twelve dogs were presented with both kikuyu and
couch grass three times daily for 6 d and grass eating
behaviours were observed using an all-occurrences
sampling method. The results of this study suggest that
grass eating is inuenced by satiety and time of day.
Dogs spent more time eating grass before ingestion of
their kibble meal than after, and the time spent eating
grass decreased throughout the day. Grass may be seen
as a food source, as the subjects were less likely to
eat grass when they were satiated. Couch and kikuyu
grasses were equally preferred. We conclude that grass
eating is a normal and common behaviour, as all dogs in
this study were in good health and readily ate grass. As
such, grass-eating should not be seen as a problematic
behaviour for most dogs or as indicative of illness.
Keywords: Canis familiaris, dog, feeding behaviour,
grass eating, satiety
Introduction
It has been suggested that grass eating is a common
behaviour in domestic dogs (Houpt, 2005) but there
is no known explanation for this behaviour (Lindsay,
2001) and there have been no experimental studies
investigating grass eating in this species (Hart, 1985;
UC Davis School of Veterinary Medicine Companion
Animal Behavior Program, 2005). Dogs have almost
no capacity to digest grass (Beaver, 1981). Nonetheless,
some researchers contend that grass may inuence
digestion by acting as an emetic (Fox, 1965; Beaver,
1981; Hart, 1985; de Baïracli Levy, 1992; Thorne,
1995; Lindsay, 2001; Houpt, 2005), a laxative (Hart,
1985), or by providing roughage (McKeown, 1996;
Houpt, 2005). There are several products that market
grass as a digestive aid or dietary supplement (Organic
Pet Grass Kit, ©Wheatgrasskits.com, Springville, UT;
Pet Greens® and Pet Grass®, Bell Rock Growers,
Inc., San Marcos, CA; Barley Dog®, Green Foods
Corporation, CA). However, the claims made for
these dietary supplements have not been substantiated
(Lindsay, 2001).
While there are no experimental studies, Sueda
et al. (2005) performed an owner-completed survey
and observed no relationship between plant eating
and gender, gonadal status, breed, diet or presence of
intestinal parasites. The researchers concluded that plant
eating evolved in wild canids and was preserved through
the domestication process. While the study of Sueda et
al. study provides some preliminary information about
grass eating, the scientific value of the study may have
been compromised by the subjective nature of its design
(information was provided by multiple owners about
their individual dog’s eating habits). More concrete
conclusions may be drawn from controlled scientific
experiments.
The current study is the first to scientifically
investigate grass eating behaviour in dogs in a controlled
experiment. As very little is known about grass eating
in dogs, the aim was to provide an initial understanding
of the behaviour by determining the pattern of grass
eating habits during the day as well as the relationship
between grass eating and the ingestion of food.
A preliminary pilot study indicated that dogs eat a
few grams of grass per eating episode and prefer to eat
grass presented as entire plants growing in pots rather
than as cut blades or turf (S.J. Bjone, unpublished data).
This preliminary study also illuminated the difficulty
of quantifying the amount of grass eaten. As grass
blades are light and the pots and soil that contain the
grass are heavy in comparison, it was difficult to obtain
accurate measurements of the amount of grass eaten. In
addition, the dogs often disrupted the grass by urinating
or salivating on the grass, digging or tipping the pots.
These disruptions further complicated the weighing
procedure. Therefore, the method adopted for the current
experiment was to measure grass eating behaviours:
the amount of time spent eating grass, number of grass
interactions, and vomiting events. The current study
was devised to explore the following questions:
1. Is grass eating inuenced by satiety?
2. Do dogs have a preference for one type of grass
over another?
3. Is grass eating inuenced by the time of day?
Recent Advances in Animal Nutrition in Australia 16 (2007)

46
47
Materials and Methods
Subjects, housing and diet
Twelve mixed-breed dogs which were accustomed to
kennel housing, owned by the same owner and known
to eat grass (mean age ± standard error (SE) = 6.0 ±
0.3 yr) were housed at the University of New England
Dog Research Facility for the duration of the study. The
dogs became accustomed to the daily routine and diet
during a 7 day habituation period. All dogs were fed
a nutritionally complete and balanced diet (Pedigree
Advance Adult Chicken®; MasterFoods ANZ,
Wodonga, VIC) once daily in amounts calculated to
meet maintenance energy requirements and adjusted as
necessary to maintain ideal body weight. Fresh water was
available ad libitum. Dogs were housed in compatible
groups according to owner recommendations. Each
indoor kennel was outtted with a trampoline style
bed within a secure, centrally-heated facility where
dogs slept at night. During the day, dogs were placed in
spacious, fully covered, outdoor runs.
Materials
Couch (Cynodon dactylon) and kikuyu (Pennisetum
clandestinum) grasses were used in the study because
dogs are known to eat couch grass as its “dog grass”
nickname indicates (de Baïracli Levy, 1992), and
kikuyu was readily available locally. The grasses were
grown in 20 cm pots in a greenhouse.
Ethics
All procedures were undertaken in accordance with
the Australian Code of Practice for the Care and Use
of Animals for Scientic Purposes (National Health
and Medical Research Council, 1997). All dogs were
privately owned, and written permission was obtained
from the owner for the inclusion of the animals in the
study. Animals received the highest standard of care
throughout the study, in accordance with UNE Animal
Ethics Committee guidelines. In addition, a veterinarian
assessed each dog to ensure it was healthy and t to
participate in the study. All dogs were returned to the
owner at the end of the study.
Procedure
Dogs were observed in compatible groups (n = 3)
during three 10 minute testing sessions (morning, noon
and afternoon at approximately 0900, 1200 and 1500 h)
per day for 6 d. Groups were presented with two pots of
both kikuyu and couch grass and all occurrences of each
behaviour (Table 1) were recorded by the experimenter
(SB) from the adjoining kennel (Altmann, 1974). A
mini-DV camera also recorded each session for further
analysis. The daily feeding time was rotated through the
testing session time slots every 2 d. Therefore, the dogs
were fed during the morning session for the rst two
consecutive days, the noon session for next 2 d and the
afternoon session for the remaining two testing days.
During the sessions in which the dogs were fed, two
groups were observed before ingestion of food and two
groups were observed after their meal. Group order
within testing sessions was determined using a balanced
Latin-square design.
Table 1. Behaviours recorded during testing sessions
and their denitions. Behaviours marked with an
asterisk (*) were recorded for both couch and kikuyu
grasses.
Behaviour Denition
Time spent eating grass*. A dog chewed and
swallowed grass .
Number of grass eating
events*.
An event encompassed
the dog ingesting grass
until it stopped chewing,
lifted its head or moved
to a new position.
Number of grass
interactions*.
Any interaction with the
grass which did not entail
ingestion or urination.
Number of vomiting
events.
A dog vomited.
Statistical Analysis
The total number of grass eating events, grass
interactions and vomiting events and the total time spent
eating grass for each dog was analysed using a within-
subjects repeated measures ANOVA (Tabachnick and
Fidell, 2001). A 2 × 2 ANOVA was used to analyse the
differences in the behaviours for each grass type before
and after the ingestion of food and a 2 × 3 ANOVA
was used to analyse the differences in the behaviours
for each grass type between the morning, noon and
afternoon periods. Repeated measures ANOVA was
used to analyse behaviours that did not specically
relate to a grass type.
Significance levels were set at α = 0.05 unless
otherwise noted. The strength of association was
represented by partial eta-squared, pη2 (Tabachnick and
Fidell, 2001; Levine and Hullett, 2002). If sphericity
could not be assumed for a repeated measures ANOVA,
Greenhouse–Geisser values were used and the p-value
was labelled with a “G–G” (Tabachnick and Fidell,
2001). There was a severe violation of the homogeneity
of variances assumption for the amount of time spent
eating grass and a moderate violation for the number
of grass eating events for data obtained before and after
the ingestion of food. Therefore, more stringent alphas
of 0.01 and 0.025, respectively, were chosen rather than
interpreting transformed data (Tabachnick and Fidell,
2001).
Results
The results were consistent with a preliminary study
of the dogs’ normal grass eating habits in their home
kennel environment. All dogs were observed eating
grass with a total of 709 grass eating events across all
12 dogs and all 18 testing sessions. Each dog spent an
average of 1.1 min (SE = 0.06 min) eating grass during
an average of 3.3 grass eating events (SE = 0.2) per 10
min testing session, totalling 3.3 min (SE = 0.2 min)
spent eating grass during 9.9 grass eating events (SE =
0.6) per d.
Grass Eating and Ingestion of Food
The dogs spent signicantly more time (P = 0.001, pη2
= 0.67) eating grass before ingestion of the daily kibble
meal than after the meal (Table 2; Figure 1). There
was no signicant difference between the amount of
time spent eating the two grasses (P = 0.74) and no
signicant grass and time interaction (P = 0.30). A small
to medium positive Pearson correlation (Cohen, 1988)
was present between the amount of time spent eating
grass and the number of hours since the last kibble meal
(r(214) = 0.23, P = 0.001).
Similarly, there were significantly more grass eating
events before than after the kibble meal (P = 0.001, pη2
= 0.63; Table 2, Figure 2). There was no significant
difference between the types of grass (P = 0.11) and no
significant interaction effect (P = 0.58). There were also
no significant differences in the number of interactions
before or after the meal (P = 0.953) or for grass type (P
= 0.422) and there was no significant interaction effect
(P = 0.180).
Although vomiting has been linked with the
ingestion of grass, there were only five vomiting events
involving three dogs across all 18 testing sessions. All
of the vomiting events occurred during testing sessions
in which dogs were also fed, and all events were by
dogs presented with grass before ingesting the kibble
diet. Three of the events occurred in the morning and
two occurred in the afternoon.
Behaviour Test statistics Before
meal
After
meal
Time spent eating
grass (min)
F(1,11) = 22.71
P = 0.001;
pη2=0.67
2.67 ±
0.53a
0.53 ±
0.14b
Number of grass
eating events
F(1,11) = 18.45
P = 0.001;
pη2=0.63
6.68 ±
1.42a
2.04 ±
0.57b
Day 1 Day 2 Day 3 Day 4 Day 5 Day 6
Time spent eating
grass (min)
F(5,55) = 2.48;
P = 0.042;
pη2 = 0.18
4.14 ±
0.23a
3.65 ±
0.25ab
2.32 ±
0.20b
2.46 ±
0.19ab
3.00 ±
0.26ab
4.13 ±
0.28a
Morning Noon Afternoon
Time spent eating
grass (min)
F(2,22) = 9.17;
G–G P = 0.006;
pη2 = 0.46
4.45 ±
0.91a
3.26 ±
0.72a
2.15 ±
0.53b
Number of grass
eating events
F(2,22) = 12.84
G–G P = 0.002;
pη2 = 0.54
13.73 ±
2.94a
8.92 ±
1.83b
6.92 ±
1.60b
Figure 1. The time spent eating couch and kikuyu
grass before and after the ingestion of the kibble
meal. Different superscript letters indicate signicant
differences.
Figure 2. The number of couch and kikuyu eating
events before and after the ingestion of the kibble
meal. Different superscript letters indicate signicant
differences.
Table 2. Mean and standard error values for signicant statistics. Grass represents the total of the couch and
kikuyu grasses. Different superscript letters indicate signicant differences.
0
0.5
1
1.5
2
2.5
3
3.5
Before After
Time (min)
Couch
Kikuyu
a
b
0
1
2
3
4
5
6
7
8
9
10
Before After
Number of Events
Couch
Kikuyu
a
b

48
49
Daily Pattern of Grass Eating
The dogs spent signicantly less time (G–G P = 0.006,
pη2 = 0.46) eating grass during the afternoon than the
morning and noon testing sessions (Table 2, Figure 3),
and there was no signicant difference between the
two grasses (P = 0.49) and no signicant grass-by-time
interaction (P = 0.30). The amount of time spent eating
grass did differ slightly across the six testing days (P
= 0.042, pη2 = 0.18): the dogs spent more time eating
grass on Days 1 and 6, the rst and last days of testing,
compared to Day 3 (Table 2). There was no signicant
difference between grass types (P = 0.49) and no
signicant day-by-grass interaction (G–G P = 0.56).
The number of grass eating events followed a
similar pattern to that for grass eating time; there was
a difference between the time of day (G–G P = 0.002,
pη2 = 0.54), and there was no difference between the
grasses (P = 0.40) and no significant interaction effect
(G–G P = 0.19). However, there were significantly more
grass eating events during the morning testing sessions
than during the noon and afternoon sessions (Table 2,
Figure 4).
The number of interactions were not significantly
different across time of day (P = 0.72) or grass type (P
= 0.72) and there was no significant interaction effect
(P = 0.81).
Discussion
While there are few references to grass eating in
domestic dogs in the literature, anecdotal information
suggests that couch grass is the grass of choice for dogs
(de Baïracli Levy, 1992; Engel, 2002). However, the
results of the current study do not support this theory:
the dogs did not demonstrate a statistically signicant
preference for eating either couch or kikuyu grass. This
was evident at all levels of the study: around mealtime,
throughout the day and across the six testing days.
Many of the theories about grass eating involve
digestion (Hart, 1985; de Baïracli Levy, 1992;
Overall, 1997; Lindsay, 2001; Engel, 2002). All dogs
participating in the current study passed a veterinary
health check, and all dogs were dewormed and did not
have any known digestive problems. However, in spite
of the prevalent digestion theories, all of the subjects ate
grass, spending an average of 3.3 min (SE = 0.2 min)
eating grass per day. The current study also deflates the
theory that dogs eat grass as an emetic as there were
only five vomiting events for 709 grass eating events
across all 12 dogs and all 18 testing sessions (similar to
the results of Sueda et al., 2005).
Grass eating was influenced by time of day. The
amount of time spent eating grass decreased throughout
the day with less time spent eating grass in the afternoon
than morning or noon. It is unlikely that the decrease
in grass eating as the day progressed was related to
an overall habituation effect as the dogs spent similar
amounts of time eating grass at the beginning and end
of the trial.
However, satiety may have influenced this grass
eating pattern. Possibly, by the afternoon testing session
the dogs had their fill of grass and were no longer
interested in eating more. As the dogs had already
ingested their kibble diet before the afternoon session
for 4 of the 6 testing days, they may not have been
hungry during these later sessions.
Further support for the effects of satiety on grass
eating is evident from results showing that the dogs
spent significantly more time eating grass before
they ingested their kibble meal than after the meal.
Similarly, the correlation between the amounts of time
spent eating grass and the number of hours since the last
kibble meal also supports the concept that the longer
the time sincethe last kibble meal, i.e., the hungrier the
dogs are, the more time they will spend eating grass.
The current study endeavoured to provide an
initial understanding of grass eating behaviour in dogs
by determining the pattern of grass eating during the
day and the relationship between grass eating and the
ingestion of food. Further studies are currently underway
to investigate other aspects of grass eating behaviour in
dogs such as its relationship with worm burdens and the
development of the behaviour in puppies.
Conclusions
The results of this study suggest that grass eating
is inuenced by satiety and time of day. As the day
0
1
2
3
4
5
6
Morning Noon Afternoon
Time (min)
Couch
Kikuyu
a
b
a
Figure 3. The time spent eating couch and kikuyu
grass during the morning, noon and afternoon sessions.
Different superscript letters indicate signicant
differences.
0
2
4
6
8
10
12
14
16
Morning Noon Afternoon
Number of Events
Couch
Kikuyu
a
b
b
Figure 4. The number of couch and kikuyu eating
events during the morning, noon and afternoon sessions.
Different superscript letters indicate signicant
differences.
progressed, the dogs spent less time eating grass. While
the dogs showed no preference for couch or kikuyu
grasses, dogs may see grass as a food source and are
more likely to eat grass if they are hungry, i.e., before
ingesting their regular diet. Grass eating is a normal
and common behaviour and should not be seen as a
problematic behaviour for most dogs.
Acknowledgements
This research was generously funded by the Waltham
Foundation. The authors also wish to extend their
appreciation to Jenny Frazer for allowing her dogs
to participate in the study, and to Kristy Harvey for
assistance in caring for the dogs.
References
Altmann, J. (1974). Observational study of behavior:
Sampling methods. Behaviour 49, 227–267.
Beaver, B.V. (1981). Grass eating by carnivores.
Veterinary Medicine Small Animal Clinician July,
968–969.
Cohen, J. (1988). Statistical Power Analysis for the
Behavioral Sciences (2nd ed.). Lawrence Erlbaum
Associates, Hillsdale, NJ, USA.
de Baïracli Levy, J. (1992). The Complete Herbal
Handbook for the Dog and Cat (6th (reprinted)
edn.). Faber and Faber, London, U.K.
Engel, C. (2002). Wild Health: How Animals Keep
Themselves Well and What We Can Learn from
Them. Phoenix Orion Books Ltd., London, UK.
Fox, M.W. (1965). Canine Behaviour. Charles C
Thomas Publisher, Springeld, IL, USA.
Hart, B.L. (1985). The Behavior of Domesticated
Animals. W.H. Freeman and Company, New York.
Houpt, K.A. (2005). Domestic Animal Behavior for
Veterinarians and Animal Scientists (4th ed.).
Blackwell Publishing, Ames, Iowa, USA.
Levine, T.R. and Hullett, C.R. (2002). Eta squared,
partial eta squared, and misreporting of effect size in
communication research. Human Communication
Research 28(4), 612–625.
Lindsay, S.R. (2001). Appetitive and elimination
problems. In: Etiology and Assessment of Behavior
Problems, pp. 273–299. (Vol. 2). Iowa State Press,
Ames, IA, USA.
McKeown, D.B. (1996). Eating and drinking behavior
in the dog. In: Dog Behavior and Training:
Veterinary Advice for Owners, pp. 125–140. (eds.
L. Ackerman, G. Landsberg and W. Hunthausen).
TFH Publications, Inc., Neptune City, NJ, USA.
National Health and Medical Research Council. (1997).
Australian code of practice for the care and use of
animals for scientic purposes. NHMRC, Canberra,
Australia.
Overall, K.L. (1997). Clinical Behavioral Medicine for
Small Animals. Mosby, Sydney, Australia.
Sueda, K.C., Hart, B.L. and Cliff, K.D. (2005). Plant
eating in dogs: Characterization and relationship to
signalment, illness, and behavior problems. Paper
presented at the Waltham International Nutritional
Sciences Symposium, University of California,
Davis, USA.
Tabachnick, B.G. and Fidell, L.S. (2001). Using
Multivariate Statistics (4th edn.). Allyn and Bacon,
Sydney, Australia.
Thorne, C. (1995). Feeding behaviour of domestic dogs
and the role of experience. In: The Domestic Dog:
Its Evolution, Behaviour and Interactions with
People, pp. 103–114. (ed. J. Serpell). Cambridge
University Press, Cambridge, UK.
UC Davis School of Veterinary Medicine Companion
Animal Behavior Program. (2005). A scientic
study of grass and plant-eating behavior in dogs.
http://intercom.virginia.edu/SurveySuite/Surveys/
PlantDog/index2.html, http://www.healthypet.
com/faq_view.aspx?id=14. Accessed on 18 August
2005.