ArticlePDF Available

Geometric analysis of macronutrient selection in breeds of the domestic dog, Canis lupus familiaris

  • GA Pet Food Partners
  • Mars Petcare

Abstract and Figures

Although many herbivores and omnivores have been shown to balance their intake of macronutrients when faced with nutritionally variable foods, study of this ability has been relatively neglected in carnivores, largely on the assumption that prey are less variable in nutrient composition than the foods of herbivores and omnivores and such mechanisms therefore unnecessary. We performed diet selection studies in 5 breeds of adult dog (Canis lupus familiaris) to determine whether these domesticated carnivores regulate macronutrient intake. Using nutritional geometry, we show that the macronutrient content of the diet was regulated to a protein:fat:carbohydrate ratio of approximately 30%:63%:7% by energy, a value that was remarkably similar across breeds. These values, which the analysis suggests are dietary target values, are based on intakes of dogs with prior experience of the respective experimental food combinations. On initial exposure to the diets (i.e., when naive), the same dogs self-selected a diet that was marginally but significantly lower in fat, suggesting that learning played a role in macronutrient regulation. In contrast with the tight regulation of macronutrient ratios, the total amount of food and energy eaten was far higher than expected based on calculated maintenance energy requirements. We interpret these results in relation to the evolutionary history of domestic dogs and compare them to equivalent studies on domestic cats.
Content may be subject to copyright.
© The Author 2012. Published by Oxford University Press on behalf of
the International Society for Behavioral Ecology.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License
(, which permits unrestricted non-commercial use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Original Article
Geometric analysis of macronutrient selection
in breeds of the domestic dog, Canis lupus
Adrian K.Hewson-Hughes,a Victoria L.Hewson-Hughes,a AlisonColyer,a Andrew T.Miller,a Scott
J.McGrane,a Simon R.Hall,a Richard F.Butterwick,a Stephen J.Simpson,b and DavidRaubenheimerc
aWALTHAM® Centre for Pet Nutrition, Freeby Lane, Waltham-on-the-Wolds, Melton Mowbray,
Leicestershire LE14 4RT, UK,bSchool of Biological Sciences and the Charles Perkins Centre, University
of Sydney, New South Wales, Australia, and cInstitute of Natural Sciences, Massey University, Auckland,
New Zealand
Although many herbivores and omnivores have been shown to balance their intake of macronutrients when faced with nutri-
tionally variable foods, study of this ability has been relatively neglected in carnivores, largely on the assumption that prey are
less variable in nutrient composition than the foods of herbivores and omnivores and such mechanisms therefore unnecessary.
We performed diet selection studies in 5 breeds of adult dog (Canis lupus familiaris) to determine whether these domesticated
carnivores regulate macronutrient intake. Using nutritional geometry, we show that the macronutrient content of the diet was
regulated to a protein:fat:carbohydrate ratio of approximately 30%:63%:7% by energy, a value that was remarkably similar across
breeds. These values, which the analysis suggests are dietary target values, are based on intakes of dogs with prior experience of
the respective experimental food combinations. On initial exposure to the diets (i.e., when naive), the same dogs self-selected a
diet that was marginally but signicantly lower in fat, suggesting that learning played a role in macronutrient regulation. In con-
trast with the tight regulation of macronutrient ratios, the total amount of food and energy eaten was far higher than expected
based on calculated maintenance energy requirements. We interpret these results in relation to the evolutionary history of
domestic dogs and compare them to equivalent studies on domestic cats. Key words: Canis lupus, carnivore nutrition, domesti-
cation, domestic dog, geometric framework, macronutrient regulation, predation, right-angled mixture triangles. [Behav Ecol]
In the study of animal foraging, there has long been a focus
on single nutritional currencies, most notably energy (in
optimal foraging theory, sometimes with nutrients or second-
ary compounds included as linear constraints) or nitrogen (in
nutritional ecology) (Raubenheimer etal. 2009). Over recent
years, however, the application of an alternative approach,
geometric analysis, has demonstrated that the balance of
energetic macronutrients exerts a powerful, and often domi-
nant, effect on the nutrition-related behavior, physiology, and
performance of animals (Simpson and Raubenheimer 2012).
This is reected, inter alia, in the divergent macronutri-
ent regulatory responses among feeding guilds, with dietary
breadth, and in association with development mode and the
possession of nutrition-related symbionts (see Discussion).
A fundamental, but little-researched, area in the evolution
of nutritional regulatory strategies concerns the impacts of
domestication and the associated selection pressures on the
preferred dietary macronutrient proles of animals. In a
recent study of domestic cats, Hewson-Hughes et al. (2011)
used nutritional geometry to demonstrate that cats actively
selected a diet with a protein:fat:carbohydrate energy bal-
ance of 52:36:12, and interpreted this to reect their largely
unchanged status as an obligate predator. That the selected
macronutrient balance in that study was not an artifact of the
captive history of the cats, or of the experimental conditions,
but rather provides insights into the evolutionary history of
these felids, is suggested by 2 facts. First, not only did the cats
select a specic macronutrient balance but they also altered
the proportions of different experimental food combina-
tions eaten to maintain the target nutrient balance. Second,
Plantinga etal. (2011) have subsequently demonstrated that
a very similar macronutrient prole is selected by feral cats in
From an evolutionary viewpoint, among the most
fascinating instances of domestication is the dog. Domestic
dogs, the only large carnivore ever to be domesticated,
were derived from wolves (Canis lupus), probably in several
independent events (Vilà etal. 1997; vonHoldt etal. 2010).
Unlike other domesticated animals, which originated in
agrarian societies 10 000 years B.P. or less, the initial stages
of dog domestication took place among hunter-gatherers
at least 15 000 (Driscoll et al. 2009; Wayne and vonHoldt
2012) and possibly in excess of 100 000years ago (Vilà etal.
1997). In the early stage, this was most likely more a case
of spontaneous coevolution between wild wolf populations
and humans, but with the transition to agriculture there
Address correspondence to A.K. Hewson-Hughes. E-mail: adrian.
Received 18 April 2012; revised 31 August 2012; accepted 14
September 2012.
Behavioral Ecology
Behavioral Ecology Advance Access published October 23, 2012
by guest on December 6, 2012 from
was increasingly intense selection for smaller, more docile
dogs (Wayne and vonHoldt 2012). Over the last 200years,
intensication of articial selection has resulted in a
proliferation of breeds, with the consequence that dogs are
now the most phenotypically diverse of all animal species.
They range in size from 1 kg (Chihuahua) to 100 kg (Mastiff),
display more variation in skeletal and cranial proportions
than the entire carnivore order, and show behavioral
and physiological attributes that are comparably diverse
(Wayne and Ostrander 2007; Wayne and vonHoldt 2012).
This diversity is all the more interesting because it exists
contemporaneously with the ancestral species, from which
it was derived in what is from an evolutionary perspective a
very short time period.
Here we describe a series of experimental studies in which
we explored the patterns of dietary protein, fat, and carbo-
hydrate selection in 5 diverse breeds of domestic dogs, the
papillon (PAP), miniature schnauzer (MS), cocker spaniel
(COS), Labrador retriever (LR), and St Bernard (SB). Our
aims were to determine whether the patterns of macronutri-
ent selection in these breeds were as diverse as other pheno-
typic traits, to compare these to comparable experiments on
domestic cats (Hewson-Hughes et al. 2011), and relate the
results to the composition of prey in the wild.
Animals, housing, and welfare
Diet selection studies were performed in adult dogs (C.lupus
familiaris) of both sexes (neutered and entire) representing 5
breeds of dog from toy to giant (PAP, MS, COS, LR, and SB)
at the WALTHAM Centre for Pet Nutrition (WCPN), Melton
Mowbray, UK. Throughout each study, dogs were pair-housed
in purpose-built, environmentally enriched facilities with con-
stant access to an outdoor area where they could interact with
other dogs. Extensive socialization with people and on-lead
walking were provided each day. Food was offered to individ-
ually housed dogs in two 1-h periods per day (08:00–09:00 h
and 14:00–15:00 h) and drinking water was available at all
times. Given the potential for weight gain in these studies due
to the provision of foods in excess of energy requirements,
a limit of 15% above individual dogs’ ideal bodyweight was
set by the WCPN Ethical Review Committee as the maximum
amount of weight a dog could gain before being removed
from the study and dogs that did gain weight underwent a
weight management phase after completing the study to
return them to their ideal bodyweight.
Diets and general protocols
We performed 3 experiments, which differed in the diet for-
mat (wet vs. dry) and the combination of foods (differing in
macronutrient balance) offered to the dogs. We used com-
mercial dog foods that were either dry (kibbles) or wet (cans/
pouches), which differ not only in water content and texture
but also macronutrient proles, most notably a higher carbo-
hydrate level in dry foods. The rst aim of our experiments
was to assess and compare the dietary balance selected by the
different breeds of dog fed either dry or wet foods with vari-
able macronutrient levels (with foods available in excess of
calculated maintenance energy requirements [MERs]). The
second aim was to evaluate whether naivety to the experimen-
tal foods affected the macronutrient balance selected.
All experiments comprised 3 phases (the durations of
which are given below). The rst was a naive self-selection
(NSS) phase, in which the dogs were exposed to all of the
experimental foods simultaneously. The aim of this phase was
to measure nutrient self-selection by the dogs when naive to
the experimental foods. In the second, monadic “learning”
phase, the dogs were exposed to each of the foods on suc-
cessive days. This phase provided the dogs with further expe-
rience of the foods before they were once again allowed to
self-select a diet in the experienced self-selection (ESS) phase.
We considered the ESS phase of the experiments to provide
the best measure of the macronutrient balance preferred by
the dogs, because in this phase they could freely compose a
diet from nutritionally complementary foods of which they
had prior experience. To evaluate the impact of this, we com-
pared macronutrient selection in Phase 1 (NSS) and Phase 3
Experiment 1: dry foods—variable protein, carbohydrate,
Three dry-format foods (high fat, pFc; high carbohydrate,
pfC; and high protein, Pfc) were manufactured using stan-
dard processing (extrusion) procedures at Mars Petcare,
Verden, Germany. These foods were formulated based on
Mars Inc. commercial recipes with the inclusion level of poul-
try meal, maize gluten, ground rice, wheat our, and beef tal-
low altered to achieve differences in the macronutrient:energy
ratios of the foods (Table S1, SupplementaryData).
The following protocol was carried out in 5 breeds of dog,
details of which are shown in Table S2 (Supplementary Data).
Of the 51 dogs that started the experiment, 16 were removed
as a result of reaching the weight gain limit set by the Ethical
Review committee. Prior to the experiment, dogs were fed
standard commercially available dry diets (Pedigree® com-
plete adult small dog [PAP], Pedigree® Advance adult mini
[MS and COS], and Pedigree® complete adult [LR and SB])
at 200% of their estimated MER (calculated as 460 kJ kg−0.75
[110 kcal kg−0.75]) for at least 1 week to accustom them to
being offered excess food during the experiment. This calcu-
lation for MER falls between the National Research Council
(NRC) recommended energy requirements of 544 kJ kg−0.75
(130 kcal kg−0.75) and 377 kJ kg−0.75 (90 kcal kg−0.75) for active
and inactive dogs, respectively (NRC 2006).
Phase 1:NSS
For 7days, dogs were given foods pFc, pfC, and Pfc simultane-
ously in 3 separate bowls from which to self-select a diet. Each
food was offered at 50% of each dog’s calculated MER at both
the morning (08:00–09:00 h) and afternoon meal (14:00–
15:00 h; i.e., 100% MER of each food was offered in total each
day and 300% MER offered in total each day). To avoid posi-
tional bias the location of each food was rotated daily.
Phase 2: monadic “learning”phase
Dogs were cycled through eight, 3-day periods in which they
were conned to a different food (pFc, pfC, or Pfc) on each
of the 3 days. Each food was offered at 100% of each dog’s
calculated MER at both the morning (08:00–09:00 h) and
afternoon meal (14:00–15:00 h; i.e., 200% MER offered in
total each day). This phase served as a conditioning phase in
which the dogs gained experience of the nutritional prole
of each of the foods separately.
Phase 3:ESS
In this phase, the regimen of Phase 1 was repeated on the
now experienced dogs. However, it was noticed that some
dogs were consuming all of one or more of the foods, and
so to ensure that the composition of the diet self-selected by
the dogs was not constrained by availability of the foods, this
phase was extended by 4 days during which each food was
offered at 100% of each dog’s calculated MER at both the
Behavioral EcologyPage 2 of 12
by guest on December 6, 2012 from
morning (08:00–09:00 h) and afternoon meal (14:00–15:00 h;
i.e., 200% MER of each food offered in total each day and
600% MER offered in total each day). Only the data collected
during this 4-day extension were used in the statistical analy-
sis of this experiment (because on reviewing the data it was
apparent that some dogs also depleted one or more foods
during the NSS phase).
Experiment 2: wet foods—variable protein, carbohydrate,
It appeared from analysis of Experiment 1 that dogs were
selecting a diet composition toward the highest possible fat
intake and lowest possible carbohydrate intake. However, due
to the macronutrient proles of the foods provided in that
experiment, they were unable to compose a diet with less
than 22% of energy from carbohydrate or more than 54%
of energy from fat. We hypothesized that providing foods
with lower carbohydrate and higher fat content would allow
dogs to demonstrate their target intake for protein, fat, and
carbohydrate free from the constraints on fat and carbohy-
drate intake imposed by the foods offered in the dry food
Wet food formulations provided an opportunity to
explore lower dietary carbohydrate concentrations because,
unlike dry foods, they do not require starch as a bind-
ing agent. Three commercially available wet foods (i.e.,
processed in cans or aluminum foil trays) providing dif-
fering macronutrient energy ratio proles (Chappie® orig-
inal = pfC; Cesar® chunks in loaf = Pfc; Pedigree® puppy
original=pFc) were used in this experiment (see Table S3,
Supplementary Data). Forty-two dogs representing 4 dif-
ferent breeds started this experiment and 5 were removed
before data were collected (Table S2, Supplementary Data).
Of the dogs that took part in the dry food experiment, 3
PAPs, 2 MSs, 2 COSs, and 3 LRs also took part in this wet
food experiment. Prior to the study, dogs were fed a stan-
dard commercially available canned food (Pal® chunks in
jelly) at 200% of their MER (460 kJ kg−0.75) for at least 1
week to accustom them to being offered excess food during
the study.
Phase 1:NSS
For 7days, dogs were given foods pFc, Pfc, and pfC simultane-
ously in 3 separate bowls from which to self-select a diet. Each
food was offered at 100% of each dog’s calculated MER at
both the morning (08:00–09:00 h) and afternoon meal (14:00–
15:00 h; i.e., 200% MER of each diet was offered in total each
day). If the full allocation of any food was eaten at 1 meal the
amount of that food was increased (in 15% steps) at subsequent
meals to guarantee that some of all foods was left uneaten. This
ensured that the balance of macronutrient intake selected was
not constrained by availability of the foods. To avoid positional
bias, the position of each food was rotated daily.
Phase 2: monadic “learning”phase
Dogs were cycled through eight, 3-day periods in which they
were conned to a different food (pFc, pfC, or Pfc) on each
of the 3days. In order to limit potential for weight gain each
food was offered at 100% of MER (i.e., 50% MER at the
morning meal and 50% MER at the afternoon meal) for the
rst six 3-day cycles and increased to 200% MER (100% each
at morning and afternoon meal) for the nal 2 cycles.
Phase 3:ESS
This was the same as the naive phase with each food initially
offered at 200% of MER and increased if necessary for any
dog that ate the full allocation of any food offered.
Experiment 3: wet foods—xed protein
The results of the above Experiments 1 and 2 suggested
that protein was relatively invariant in the diets selected by
the dogs at close to 30% of total energy, whereas the bal-
ance of fat:carbohydrate varied more widely with experi-
mental circumstances. We performed a third experiment
in which protein was xed at 30% and the foods differed
in their fat:carbohydrate ratio (foods X, Y, and Z, Table S3,
Supplementary Data). Because in this design the foods var-
ied only in 2 dimensions, the NSS, monadic learning, and
experienced simultaneous self-selection phases of the experi-
ment each involved only 2 foods. The food pairings repre-
sented all pairwise combinations of high, medium, and low
fat:carbohydrate foods. This experiment also provided an
opportunity to demonstrate that the macronutrient intake
selected using commercially available wet diets described
above was not driven by differences in “palatability” between
the diets. Foods were prepared fresh each day by mix-
ing (using an electric food mixer) appropriate amounts of
drained skinless chicken breast (steam sterilized in cans at
Mars Petcare, Verden, Germany), lard (Tesco, UK; melted in
a microwave), and pregelatinized wheat our plus vitamins
and minerals. The nutritional composition of the diets was
formulated to meet recommendations for adult dogs (NRC
Twenty-six MSs (Table S2, Supplementary Data) were used
in 3 studies with n = 12 per study (10 dogs were used in 2
studies) with each study offering a choice between a pair of
foods (X vs. Z, Y vs. Z, and X vs. Y) using the methodology
described above with the following modications.
Phase 1:NSS
Each food was offered at 150% of each dog’s MER equally
divided into two 1-h meals per day (10:00–11:00 h and 14:00–
15:00 h; i.e., 600% MER offered in total each day).
Phase 2: monadic “learning”
Each food was offered at 100% of MER (i.e., 50% MER at the
morning meal and 50% MER at the afternoon meal) for the
rst six 2-day cycles and increased to 200% MER (100% each
at morning and afternoon meal) for the nal 2 cycles.
Phase 3:ESS
This was the same as the naive phase with each food offered
at 150% of MER equally divided into two 1-h meals per day
(i.e., 600% MER offered in total each day).
Statistical analyses
For all studies, the percentages of total energy from protein,
fat, and carbohydrate were analyzed by mixed-model analyses
to take into account the repeated measures for individual dogs
within a phase. The carbohydrate data were log10 transformed
prior to analyses due to positively skewed distributions;
subsequently means and 95% condence intervals (CIs) have
been back transformed for illustration in gures. Statistical
analyses were carried out using GenStat® v13.1 software (VSN
International Limited, Hemel Hempstead, UK). An overall
signicance level of 5% was adopted for all analyses.
Wet and dry foods—variable protein, carbohydrate, andfat
For the ESS wet and dry models, dog was dened as a random
effect and breed as a xed effect to investigate if there were
differences in the macronutrient balance selected between
breeds. For the NSS vs. ESS models (wet foods only), phase
nested in dog was dened as the random effect and breed,
Hewson-Hughes etal. • Macronutrient selection indogs Page 3 of 12
by guest on December 6, 2012 from
phase, and their interaction were dened as xed effects to
investigate if there were differences in the macronutrient bal-
ance selected between breeds and if this changed with experi-
ence. Estimates of the target intakes (kJ) for protein, fat, and
carbohydrate as well as total energy intake of dogs during the
ESS of the wet food experiments were determined by mixed-
model analysis with dog as a random effect and breed as a
xed effect. Data were log10 transformed prior to analysis due
to increasing variance of residuals with increasing energy con-
sumed. Means and 95% CIs were back transformed to kJ d−1
for presentation in tables.
Wet foods—xed protein
For the ESS analysis, diet pair nested in dog was dened as
the random effect and diet pair was dened as a xed effect
to investigate if there were differences in the balance of car-
bohydrate and fat selected (protein was xed) between diet
pairs. For the NSS vs. ESS models, phase nested in diet pair
nested in dog were dened as the random effects and diet
pair, phase and their interaction were dened as xed effects
to investigate if the balance of fat and carbohydrate selected
within diet pairs changed with experience.
For both analyses, the variability between diet pairs was
found to differ signicantly and this variability was modeled
by allowing the residual error to change with diet pair.
Experiments 1 and 2: wet and dry foods—variable protein,
fat, and carbohydrate
Figure 1 shows the proportional protein, fat, and carbohy-
drate compositions of the experimental foods used in the dry
and wet diet experiments, together with the nutrient intakes
of each dog breed associated with the phases of the 2 variable
protein environments. Arst thing to note from the gure is
the composition of the experimental foods (lled squares for
wet foods and hollow squares for dry foods). For each experi-
ment, the triangle formed by joining the points representing
the foods denes the accessible area for a dog provided with
ad libitum access to that food combination. The plot shows
that dogs fed the wet and dry food combinations could theo-
retically compose diets with similar maximum carbohydrate
concentration (i.e., >50% energy from carbohydrate), but
those fed the wet foods could achieve an intake with lower
proportional carbohydrate content than those fed dry foods.
Conversely, dogs on the dry foods could achieve an intake
with higher proportional protein content than those on the
wet foods, but the 2 diet formats allowed similar minimal pro-
tein concentrations. Both the minimal and maximal attain-
able fat concentrations were higher on wet than dry foods.
The area of overlap between the triangles shows the range
of protein:fat:carbohydrate dietary concentrations that was
accessible both to dogs on the wet and the dry foods. The rela-
tionship between the accessible region in diet space and the
intakes selected by the dogs in the ESS phase of the experi-
ment provides information about the macronutrient balance
of the targetdiet.
Target macronutrient balance—ESS
Statistical comparison of macronutrient intake in dogs in the
wet and dry experiments is not appropriate because these
experiments were done separately. There are nonetheless
some interesting contrasts apparent from visual comparison
of the scatter of the intake points self-selected by experienced
dogs in the 2 experiments (Figure1). First, intakes of all of
the dogs in the dry food experiment fell within the region
of overlap between the 2 diet triangles, which was equally
accessible to the dogs on both diet formats. However, dogs in
the wet diet treatment did not select this region, suggesting
that the composition selected by dogs on the dry foods was
not a true macronutrient target but a constrained outcome
imposed by the composition of the dryfoods.
Rather, dogs in the wet diet treatments composed a diet
that had similar protein concentration to those in the dry diet
treatment (all dogs fell within the band spanning 25–35%
total energy as protein), but was considerably lower in car-
bohydrate and higher in fat than dogs in the dry diet treat-
ments. This pattern, taken together with the fact that dogs
in the dry diet treatment selected intake points that were
close to the minimum carbohydrate concentration available
to them, suggests that the dry diets are appreciably higher in
carbohydrate than the target diet composition. Indeed, even
dogs on wet foods appear to have minimized the propor-
tional carbohydrate content of their diet. Overall, these data
suggest that the preferred diet composition of the dogs has
low carbohydrate:fat balance, with between 25% and 35% of
energy contributed by protein.
The proportional carbohydrate, fat, and protein compositions of
the experimental foods used in the dry (hollow squares) and wet
diet (lled squares) experiments, together with the nutrient intakes
of dogs fed these foods (circles). In these plots, known as right-
angled mixture triangles (Raubenheimer 2011), 2 components of a
3-component mixture are represented in the normal way as x and y
axes (in this case carbohydrate and fat, respectively), and the third
component (in this case protein) varies inversely with the distance
from the origin. Thus, 2 mixtures that have the same balance of
fat:carbohydrate will fall on a line projecting from the origin, and
the mixture with the higher protein content will be closer to the
origin than the mixture with lower protein. The gray- and black-
lined triangles show the areas that are accessible to dogs with access
to the dry and wet foods, respectively. Red solid circles show intake
points selected in the ESS phase and green hollow circles show mean
intakes selected in the NSS phase (wet foods only) by the various
breeds of dog. The dotted diagonals show the highest (closest to the
origin) and lowest % protein energy potentially achievable by the
dogs in the experiment. Dashed diagonals dene the band of actual
protein intakes of dogs during the experienced self-selecting phase.
Dog breeds are not distinguished in the plot.
Behavioral EcologyPage 4 of 12
by guest on December 6, 2012 from
Although it appears that dogs were unable to achieve their
target macronutrient composition from the choice of dry
foods offered to them, the selected balance of percentage
energy intake from protein, fat, and carbohydrate was not sig-
nicantly different between the 5 breeds (fat: F(4,32.3)= 1.27,
P = 0.303; protein: F(4,32.4) = 1.22, P = 0.323; carbohydrate:
F(4,32.3) = 1.38, P = 0.261), suggesting that mechanisms for
active regulation of macronutrient intake may be similar
across these breeds. Total energy intake obviously increased
in relation to the size of each breed of dog, with all breeds
apart from PAP consuming >100% of their calculated MER
(Table1) and over the course of the experiment body mass
increased on average by 0.8% in PAP, 5.4% in MS, 11.1% in
COS, 13.1% in LAB, and 4.6% in SB (for those dogs that
completed the experiment).
The choice of wet foods removed the constraints on fat
and carbohydrate intake and the estimated target intakes
(kJ d−1) for protein, fat, and carbohydrate for each breed
are shown in Table2. Similar to the situation seen in the dry
food experiments, PAP consumed close to their calculated
MER, whereas the other breeds all consumed >200% of
their calculated MER (Table 2) and over the course of the
experiment body mass changed on average by −1.2% in
PAP, +9.3% in MS, +8.4% in COS, and +11% in LAB (for
dogs that completed the experiment). When the intake data
were expressed as percentage of energy intake from protein,
fat, and carbohydrate, there were small but statistically
signicant differences in the proportions of protein and
fat (but not carbohydrate) selected by the 4 breeds (fat:
F(3,27.7) =10.51, P< 0.001; protein: F(3,27.9) = 6.5, P = 0.002;
carbohydrate: F(3,27.9)= 1.57, P = 0.22). Figure 2 shows that
this was principally due to PAP and COS composing a diet
with a higher percentage of energy from protein and lower
percentage of energy from fat than LAB and MS. The biggest
difference observed was for percentage energy from fat
between PAP and MS (8.8%, 95% CI 4.8–12.7). The average
macronutrient energy composition of the diet self-selected by
all dogs during the ESS phase of the wet food experiment was
ca. 29% protein (95% CI 27.3, 30.9), 63% fat (62.0, 64.6) and
7% carbohydrate (6.4,7.9).
A comparison of Phase 1 (NSS) and Phase 3 (ESS) of the
experiments enabled us to assess the effects of experi-
ence gained in Phases 1 and 2 (monadic phase) on the
nutrient selection by the dogs. In the dry food experiment
(Experiment 1), however, dogs in Phase 1 frequently depleted
1 or more of the experimental foods, even though each was
provided at 100% of estimated MER. Consequently, the com-
position of the selected diet partly reected an unanticipated
constraint due to the experimental design and could thus not
be interpreted to indicate the ad libitum selected macronu-
trient balance. The patterns of food depletion did, however,
provide information about the nutritional regulatory priori-
ties of the dogs. Figure 3 shows that the majority of deple-
tions involved the pFc food, which supports the conclusion
reached from our analysis of the ESS phase (above), that the
target diet has a higher fat:carbohydrate ratio than was avail-
able to dogs on the dry food treatments.
Comparison of nutrient selection in the NSS and ESS
phases of the wet foods experiment suggested that when fed
wet foods, naive dogs selected a region in nutrient space that
was close to the target region (Figure 1). Closer inspection,
however, revealed that on average (i.e., over all 4 breeds)
there was a small but statistically signicant decrease in the
percentage of energy intake from protein (F(1,28.7) = 8.27,
P= 0.008) and increase in percentage of energy intake from
fat (F(1,29.1) = 12.17, P = 0.002) in the ESS phase compared
with the NSS phase, whereas carbohydrate intake did not dif-
fer (F(1,29.3) =0.59, P = 0.449) (Figure 4). The breed effect
seen in the ESS analysis reported above was also seen in the
NSS phase, but the phase by breed interaction was not sig-
nicant, indicating that the differences seen between breeds
were consistent between NSS andESS.
Experiment 3: wet foods—xed protein
The diet compositions and selected intakes of dogs in NSS
and ESS phases of the wet foods, variable protein experiment,
are shown in Figure 5. Unlike Experiments 1 and 2, where
choices involved 3 foods and the dogs could thus move freely
within a triangle (Figure1), the dogs in this experiment had
2-way choices and were thus conned to move along the line
joining the composition points for the 2 foods in their respec-
tive diet pairings (X + Y, Y + Z, and X +Z).
Mean daily energy intake (kJ d−1) and percentage of maintenance
energy requirement (%MER) consumed in 5 breeds of dog during
ESS phase of the dry food experiments
Breed Energy intake (kJ d−1) %MERa
PAP 1010 (872–1149) 96 (87–105)
MS 3427 (3033–3821) 146 (129–163)
COS 4616 (4086–5147) 164 (148–180)
LAB 9979 (8941–11 019) 168 (150–187)
SB 11 705 (10 253–13 157) 109 (95–124)
Values shown are means with 95% CIs in parentheses. PAP, papillon;
MS, miniature schnauzer; COS, cocker spaniel; LAB, Labrador
retriever; SB, St Bernard.
aMER based on the calculation 460 kJ kg−0.75.
Mean daily macronutrient and energy intake (kJ d−1) and percentage of maintenance energy requirement (%MER) consumed in 4 breeds of dog
during ESS phase of the wet foods with variable protein, fat, and carbohydrate experiments
Macronutrient energy intake (kJ d−1)
Total energy intake (kJ d−1) %MERa
Protein Fat Carbohydrate
PAP 365 (281–475) 652 (476–894) 79 (52–121) 1114 (841–1477) 108 (99–117)
MS 1087 (920–1283) 2973 (2436–3628) 336 (257–439) 4403 (3684–5262) 208 (197–220)
COS 2254 (1919–2648) 4392 (3622–5327) 435 (335–564) 7172 (6034–8525) 266 (247–285)
LAB 3101 (2541–3785) 7900 (6222–10 032) 933 (676–1287) 12 029 (9713–14 897) 226 (208–244)
Values shown are means with 95% CIs in parentheses. PAP, papillon; MS, miniature schnauzer; COS, cocker spaniel; LAB, Labrador retriever.
aMER based on the calculation 460 kJ kg−0.75.
Hewson-Hughes etal. • Macronutrient selection indogs Page 5 of 12
by guest on December 6, 2012 from
Target macronutrient balance—ESS
Figure5 shows that in the ESS phase the dogs in the Y + Z and
X + Z treatments selected a diet that was tightly convergent
(mean % energy intake from fat: Y + Z=61.2 [95% CI 59.7,
62.8]; X + Z = 60.8 [58.7, 63.0], P = 0.759. Mean % energy
from carbohydrate: Y + Z = 7.9 [6.8, 9.2]; Z + X = 6.3 [5.2,
7.7], P = 0.063). This point corresponds very closely to the
selected intake point by dogs in the 3-way choice of wet foods
(Figure1), providing further support that this represents the
selected target balance of dogs in our experiment. The macro-
nutrient balance selected in Y + Z and X + Z pairings was sig-
nicantly different from the macronutrient balance selected
in the X + Y food pairing (mean % energy intake from fat: X
+ Y= 50.8 [49.2, 52.4]; Y + Z=61.2 [59.7, 62.8], P <0.001; X
+ Z = 60.8 [58.7, 63.0], P < 0.001. Mean % energy from car-
bohydrate: X + Y = 20.1 [18.0, 22.4]; Y + Z = 7.9 [6.8, 9.2],
P< 0.001; Z + X=6.3 [5.2, 7.7], P <0.001). This reects that
dogs in the X + Y treatment could not move closer to the tar-
get point than the point representing the composition of food
Y, and their intake thus represents a constrained compromise.
Mixed-model analysis revealed a signicant phase effect
(P<0.001) and a signicant phase by diet–pair interaction
(P< 0.001) indicating that differences in fat and carbohy-
drate energy intake seen between phases were not consis-
tent between the different diet pairings. Thus, in pairing
Mean macronutrient: energy composition (with 95% CI) of the diet composed from a choice of 3 wet foods with variable protein, fat, and
carbohydrate content in the experienced self-selecting phase by 4 breeds of dogs (Experiment 2). COS, cocker spaniel; LAB, Labrador
retriever; MS, miniature schnauzer; PAP, papillon. Letters a–e indicate homogenous breed groupings within a macronutrient at the 5% level
(i.e., those macronutrients with the same letter are not signicantly different between breeds).
Frequencies at which various individual foods and 2- and 3-way combinations of foods were depleted in the NSS phase of the dry foods
experiment (Experiment 1).
Behavioral EcologyPage 6 of 12
by guest on December 6, 2012 from
Y + Z (the 2 foods that were closest together and encom-
passed the target) the intakes for both self-selection phases
of the experiment were superimposed (Figure 5, green
circles) and there was no signicant difference in the %
energy intake from fat (and hence carbohydrate, because
protein density was held constant in the foods and there-
fore changes in the proportion of fat energy in the food
were offset by carbohydrate energy) between NSS and ESS
(Figure6). In contrast, the largest difference in % energy
intake from fat between NSS and ESS phases (5.5%, 95% CI
2.9, 8.1; Figure5, red circles) was seen in food pairing X +
Z (the 2 foods furthest apart; P< 0.001; Figure6), whereas
the difference in % fat energy intake between ESS and NSS
for pairing X + Y, which did not encompass the target, was
intermediate between the other diet pairings (Figure 5,
blue circles; Figure6).
Effect of experience on the proportional macronutrient content of the self-selected diet of 4 breeds of dog in the 3-choice wet food
experiment (Experiment 2). The plot shows the mean differences (with 95% CI) in percentage energy intake from fat, protein, and
carbohydrate (CHO) between the experienced and NSS phases. Apositive value thus indicates that experience resulted in an increase in
proportional intake of that macronutrient.
Right-angled mixture triangle showing proportional protein, fat, and carbohydrate intakes of miniature schnauzers fed wet format diets
with variable fat:carbohydrate balance and protein content xed at ~30% of energy (Experiment 3). Solid squares show the composition
of the 3 foods: X=low fat:carbohydrate; Y=intermediate fat:carbohydrate; Z=high fat:carbohydrate. Colors represent different diet
pairings (blue=X + Y; green=Y + Z; red=X + Z), and symbols distinguish different phases in the experiment (hollow circles=NSS; lled
circles=ESS). The diagonal lines show the minimum and maximum selected proportional protein intakes from the variable protein, fat, and
carbohydrate wet and dry experiments, as in Figure1.
Hewson-Hughes etal. • Macronutrient selection indogs Page 7 of 12
by guest on December 6, 2012 from
Evolutionary perspectives
We have investigated the selection by domestic dogs of the
macronutrient balance as well as metabolizable energy con-
tent of their diet. Diet selection in animals has long been eval-
uated in relation to energy intake, but research over the past
2 decades has demonstrated a critical role for macronutrient
balance (Raubenheimer et al. 2009). This work, involving a
wide range of animals, has linked macronutrient balance to
many aspects of evolutionary tness, including growth rates
and size (Raubenheimer and Simpson 1997; Simpson et al.
2004), obesity (Simpson and Raubenheimer 2005; Warbrick-
Smith et al. 2006), fecundity (Lee et al. 2008), longevity
(Piper et al. 2011), disease resistance (Cotter et al. 2010),
sexual advertisement (Maklakov etal. 2008), and risk of pre-
dation (Hawlena and Schmitz 2010).
Such results underpin a strong expectation that animals
would have evolved by natural selection an ability to balance
their macronutrient intake through selecting appropriate
foods and combining them in the required proportions. It
has long been expected that herbivores and omnivores would
possess such mechanisms, based on the premise that these
animals generally feed on nutritionally imbalanced and vari-
able foods (Westoby 1978). Carnivores, in contrast, are widely
assumed to feed on nutritionally balanced and relatively
invariant foods, and therefore not to require the mechanisms
for regulating the balance of macronutrient intake (Westoby
1978). However, several recent experiments involving both
invertebrate and vertebrate predators have demonstrated that
they do, indeed, regulate their intake of foods to balance the
gain of macronutrients, and that failure to do so can result
in tness penalties (Mayntz etal. 2005, 2009; Raubenheimer
et al. 2007; Wilder and Rypstra 2008; Hewson-Hughes et al.
2011; Jensen et al. 2011, 2012). We therefore expected that
the same would be true of domesticdogs.
Dogs are interesting with regard to the evolution of nutri-
tional strategies because of their peculiar evolutionary
history (Akey etal. 2010). Derived from Eurasian gray wolves,
domestication began over 14 000 years ago, and much more
recently dogs have undergone an intense period of articial
selection generating phenomenal phenotypic divergence
among the more than 400 breeds (Akey et al. 2010). There
are therefore at least 3 clearly identiable periods in the evo-
lutionary history of extant domestic dogs, each characterized
by different environmental and selective circumstances: pre-
domesticated wolf ancestry, early domestication, and rapid
articially selected diversication. Successive periods of dif-
ferent selection pressures can each leave a distinct imprint
on the nutritional regulatory responses of a species. Different
aspects of the nutritional biology of humans, for example,
can be traced to our catarrhine primate ancestry (Milton
2000), our Palaeolithic hunter-gather prehistory (Lindeberg
et al. 2003), and the rapid changes in nutritional ecology
wrought by agriculture (Beja-Pereira etal. 2003; Burger etal.
2007). Although a good deal of more research is needed, our
study provides some interesting pointers as to how the vari-
ous stages of the evolution of domestic dogs might have inu-
enced their pattern of macronutrient selection, the subject to
which we now turn.
Macronutrient selection
Our results showed, as expected, that dogs do indeed
regulate the macronutrient balance of their diet. In all
experimental treatments where it was possible, the diets of
experienced self-selecting dogs converged on an intake with
approximately one-third of energy derived from protein and
a low carbohydrate:fat ratio. These treatments comprised
the wet foods, 3-way diet choice experiment (Experiment
2; Figure 1), and 2 of the treatments (Y + Z and X + Z) in
the wet foods, xed protein 2-way diet choice experiment
(Experiment 3; Figure 5). Substantially different diet com-
positions were selected only by the dogs in the dry foods
experiment (Experiment 1), and in the X + Y pairing of
the xed protein experiment (Experiment 3), both of
Effects of experience on the proportional fat content of the selected diet of miniature schnauzers fed different food pairings (see Figure5;
Experiment 3). The plot shows the mean (with 95% CI) of the difference in proportional fat intake in the selected diet of dogs between the
experienced and NSS phases. Apositive value thus indicates that experience resulted in an increase in proportional fat intake. Because in this
experiment the proportion of protein in the diets was xed, changes in proportional fat intake are exactly offset by changes in carbohydrate
content and for this reason carbohydrate is not presented.
Behavioral EcologyPage 8 of 12
by guest on December 6, 2012 from
which were constrained by available foods from reaching
the point of convergence. Dogs in the dry food experiment
also selected a diet with between 25% and 35% protein, but
with a fat:carbohydrate ratio that was close to the maximum
possible given the available diet choice. When the protein
content was xed at 30% in wet foods, the dogs with diet
pairing X + Y similarly selected a diet with a fat:carbohydrate
ratio close to the maximum possible given their diet pairing.
Taken together, these results suggest that the target diet of
dogs in our study consists of approximately 30% of energy
from protein, 63% of energy from fat, and 7% of energy
from carbohydrate.
Two previous studies have investigated macronutrient
intake in domestic dogs. Romsos and Ferguson (1983)
offered 2-year-old female beagles ad libitum access to 1
of 2 pairings of 3 foods differing in protein content. The
foods contained metabolizable nutrient energy proportions
of 20:38:42 (protein:fat:carbohydrate) (food 1), 25:55:20
(food 2), and 46:27:27 (food 3). Dogs were offered a choice
between foods 1 and 3, or foods 2 and 3.In both cases dogs
selected 30% of energy as protein, which is consistent with
the results of our study, although it is not possible to discern
the roles of fat and carbohydrate from the experimental
designs used. In a more recent study, Tôrres et al. (2003)
conned beagles to 1 of 5 choices between 2 isocaloric foods
differing in protein content (0% vs. 25%, 9% vs. 32%, 18%
vs. 32%, 18% vs. 48%, or 25% vs. 48%). Again in agreement
with our study, the dogs adjusted their food choices to main-
tain dietary protein intake at close to 27% of metabolizable
An interesting nding of our study is the remarkable
consistency in the proportional composition of the target
diet across breeds. In the dry diets experiment, there was
no signicant difference between breeds in proportional
macronutrient composition of the selected diet, whereas
breed explained a small but signicant proportion of the
variance in the 3-choice wet diet experiment. Assuming
that the discrepancy between breeds is robustly associated
with the different food formulations, this would suggest
that when fat is limiting and carbohydrate excessive (i.e.,
dry foods) all breeds prioritize maximizing proportional fat
intake whereas when fat is not limiting (wet food formula-
tions) different breeds are able to express preferences for
their respective macronutrient targets. It is unclear, how-
ever, why LAB and MS should select slightly higher fat and
lower protein than PAP and COS; expanding the study to
a wider range of breeds would provide a more substantial
comparative perspective which could shed light on this
question. In the meantime, however, the overriding conclu-
sion is that the recent rapid divergence among dog breeds is
not substantially reected in their macronutrient priorities
compared with other phenotypic features such as size, color,
and temperament.
In contrast with the similarity between dog breeds, experi-
ments have shown that some more distantly related vertebrate
carnivores select a diet with very different macronutrient bal-
ance than the P:F:C balance of 30:63:7 selected by the dogs in
our study. Using a similar experimental protocol to the pres-
ent study, Hewson-Hughes etal. (2011) found that domestic
cats selected a diet that is substantially higher in protein and
lower in fat (52:36:12). Interestingly, a subsequent meta-anal-
ysis showed that the diets of free-roaming feral cats also com-
prised 52% of energy from protein. On the other hand, the
relative contributions of fat and carbohydrate to the energy
intake of feral cats (46% and 2%, respectively) differed from
the 36%:12% selected under experimental conditions in the
study of Hewson-Hughes etal. (2011), a difference to which
we returnbelow.
Why should cats select a diet with higher proportional
protein and lower proportional fat content than dogs? In
common with all of the extant members of the cat fam-
ily, domestic cats are metabolically reliant on a diet that in
the wild can be satised by eating largely vertebrate prey
(Bradshaw 2006). Because historically very few households
would regularly have had surplus meat or sh, there was up
until recently strong selection pressure on cats to satisfy their
specialized nutrient needs through hunting small vertebrates
(Bradshaw 2006), which typically have a high protein:fat con-
tent. For example, the data on body compositions of several
species of typical prey of domestic cats presented by Eisert
(2011) range from 33% energy from protein (common vole)
to 68% energy from protein (bank vole), with a midpoint of
50.5% which is very similar to 52% selected both by domes-
tic cats in experimental conditions (Hewson-Hughes et al.
2011) and free-roaming feral cats (Plantinga et al. (2011).
Domesticated dogs, in contrast, have not been under the
same constraint and would have relied on a wider range of
human-provided foods during the process of domestication.
Furthermore, being social hunters, the wolf ancestors of
domestic dogs were able to utilize larger prey. Because the
proportional body fat content of mammals increases with
size (Prothero 1995), the fat:protein balance of the diet of
the ancestors of domestic dogs would likely have been higher
compared with the small, solitary-hunting ancestor of domes-
tic cats. Finally, compared with dogs, the domestication of
cats is very recent (possibly <200 years B.P.) and might be
incomplete (Driscoll et al. 2009). This is reected in the
minimal morphological divergence of domestic cats from
their wildcat ancestors (Driscoll etal. 2009) and might also
explain the similarity in the protein content of experimen-
tally selected and wilddiets.
Interestingly, however, Darwin (1868) noted that domes-
tic cats have longer intestines than wildcats, a difference he
attributed to a less carnivorous diet associated with feed-
ing on kitchen scraps. If Darwin is correct on this, then the
intensication of articial selection for divergent morphol-
ogy in cats, initiated in the Victorian era, might also have had
nutritional implications. This could explain why cats in the
experiments of Hewson-Hughes etal. (2011), although select-
ing a similar proportion of protein in the diet as feral cats
in the wild (Plantinga etal. 2011), selected a higher propor-
tion of carbohydrate (12% vs. 2%, respectively). If the 1.7%
average carbohydrate content of the 8 species of small verte-
brate prey listed by Eisert (2011) is representative, then feral
cats would be ecologically constrained from achieving an
intake of 12% carbohydrate even if this did reect an arti-
cially selected dietary preference. But why, if the macronutri-
ent preference of cats has been evolutionarily altered since
Victorian times, has comparable divergence not taken place
among the diverse breeds included in our study? Owing to
the longer history of dog domestication, it is likely that the
pattern of macronutrient selection in domesticated dogs
had already diverged signicantly from their wolf ancestors
by 200years ago when articial selection intensied (Wayne
and vonHoldt 2012), and all of the breeds in our study were
derived from a common ancestor that was already “adapted”
to a human-associateddiet.
Several species of carnivorous shes have also been found
to select a diet high in protein (ca. 55%—Ruohonen et al.
2007). By contrast, an obligate predator, the mink, selects a
diet more similar to that of dogs, consisting of 35% protein
by energy (Mayntz et al. 2009). This is interesting, because
like cats mink feed on small vertebrates and therefore almost
certainly have a natural diet that is higher in protein than
the self-selected preference. It is possible, however, that the
macronutrient preference of the mink in the experiments
Hewson-Hughes etal. • Macronutrient selection indogs Page 9 of 12
by guest on December 6, 2012 from
of Mayntz et al. (2009) is not representative of their wild
counterparts, but inuenced by articial selection in cap-
tivity. Suggestively, the standard diet of the experimental
population, which had been in captivity for 75 generations,
contained 32% protein, which is very similar to the 35%
self-selected by these animals. An equivalent example of
tight adaptation in captive-bred animals to the macronutri-
ent composition of the diet has been described by Warbrick-
Smith etal. (2009).
Effects of experience
Experiments on herbivorous (Simpson and White 1990;
Raubenheimer and Tucker 1997) and omnivorous (Gadd and
Raubenheimer 2000) insects have demonstrated that learning
can play an important role in balancing macronutrient gain.
Recently, Hewson-Hughes et al. (2011) showed that adult
domestic cats offered a 3-way choice between foods varying
in the protein, fat, and carbohydrate content selected a diet
higher in protein and lower in carbohydrate following experi-
ence with the diets than when naive. The dogs in the present
experiments similarly selected a different macronutrient bal-
ance when experienced (ESS) than when naive (NSS), both
in the 3-choice wet food study and the 2-choice xed protein
study. In the 3-choice study, experienced dogs selected a sig-
nicantly higher proportion of fat and lower proportion of
protein in ESS than NSS, with no signicant difference in
the proportion of energy from carbohydrate. In the 2-choice
study, dogs similarly selected a higher proportion of energy
from fat in ESS than NSS (Figure 6), but in this case the
proportion of protein was xed and fat thus displaced car-
bohydrate. It seems, therefore, that the main effect of expe-
rience with the experimental foods was to enable the dogs
to increase the proportional contribution of fat to the diet.
Interestingly, the effect of experience in cats and dogs thus
echoes the relative proportions of different macronutrients
in the target diet, with cats learning to increase the propor-
tional protein content of the diet, and dogs the proportional
fat content.
Results of the 2-choice xed protein study suggested that
the role of learning was dependent on diet pairing (Figures
5 and 6). There was no signicant difference in propor-
tional fat intake for dogs given the Y + Z pairing, indicating
that the dogs were able to regulate to the target point even
when rst exposed to the experimental foods (in the NSS
phase). In contrast, when given food pairing X + Z, which
also encompassed the macronutrient target but had widely
discrepant fat:carbohydrate balances, the dogs only reached
the target composition when experienced (in ESS) and
when naive ingested a low fat:carbohydrate diet. Learning
similarly played a role for dogs given the Y + X diet pair-
ing, which comprised foods less dissimilar than the X + Z
pairing but did not encompass the target balance. In this
case, the compositions of the foods precluded the dogs from
achieving the target balance, but they approached the target
balance more closely when experienced. It is interesting to
speculate on why the need for learning should be context
dependent in this way. The common factor in the 2 treat-
ments that did require learning is that both involved food
X, which had a high carbohydrate:fat (38%:33%) balance.
It is plausible that the reason that learning was required on
these 2 diet pairings is that short-term regulatory mecha-
nisms, which enabled inexperienced dogs to reach the
target when given food pairing Y + Z, are ineffectual on
extreme foods with carbohydrate:fat balance as high as food
X because they are beyond the evolutionary experience of
dogs. In such circumstances more general mechanisms,
such as positive associative and aversion learning, might be
required to enable dogs to meet their macronutrient target
(Berthoud and Seeley 2000).
Whereas the evidence is compelling that dogs regulated the
relative proportions of protein, carbohydrate, and fat in the
diet, it was apparent that the total amount of food and energy
eaten was far higher than expected. In fact, as set out in the
Methods, we had to make various adjustments in food pro-
visioning to ensure that our subjects were maintained under
conditions of excess food availability. The daily metabolizable
energy requirement for adult dogs at maintenance (i.e., to
support energy equilibrium) recommended by the NRC is 544
kJ kg−0.75 (130 kcal kg−0.75) although it is recognized that there
is considerable individual variation, even between dogs kept
under the same conditions (NRC 2006). In the present stud-
ies, we used 460 kJ kg−0.75 (110 kcal kg−0.75) as the basis for cal-
culating the “maintenance” amount of food to offer the dogs
and for expressing their subsequent energy intake relative to
their “maintenance” requirement. Our results show that of
the breeds used in these studies, MS, COS, and LAB all con-
sumed well in excess of their calculated energy requirement
during the self-selection phases of both dry (146%, 164%,
and 168% of MER, respectively) and wet (208%, 266%, and
226% of MER, respectively) food experiments. When offered
ad libitum access to 2 different diet pairings, adult beagles
consumed similar amounts of energy (~5100 and 5400 kJ
d−1) in both choice treatments over the course of the 4-week
experiment (Romsos and Ferguson 1983). The authors con-
cluded that dogs regulate both protein (see above) and total
energy intake when offered appropriate food choices. It is
interesting to note that the energy intake at which the bea-
gles “regulated” their intake in both choice treatments was
approximately 208% and 224% of MER (based on 460 kJ
kg−0.75)—similar to the values seen in dogs in the wet choice
experiments in the presentstudy.
These ndings may reect the feeding behavior of the
wild ancestors of domestic dogs, wolves. Wolves may only kill
prey every few days or even less frequently and therefore are
adapted to a feast or famine foraging pattern, facing great
uncertainty in the availability and intake of protein and
energy. Hence, it is not surprising that when wolves do make
a kill they can consume large amounts of food, enabling them
to sustain periods of limited or no food availability. Amini-
mum daily food requirement of 3.25 kg wolf−1 d−1 (5× basal
metabolic rate) has been estimated for a 35 kg wolf (Peterson
and Ciucci 2003) and consumption rates of 5.7 kg wolf−1 d−1
and 10.4 kg wolf−1 d−1 have been estimated for larger wolves
(~45 kg) depending on kill rates (Stahler etal. 2006). It seems
that dogs still have this propensity to ingest a large amount
of food/energy if given the opportunity. However, now that
they are domesticated and have regular access to food with
no need to expend energy obtaining it, it is easy to see how
dogs could become overweight if the amount of food offered
is not controlled.
Our study has added to the growing list of experiments
demonstrating that carnivores, like herbivores and omni-
vores, are able to combine foods of varying composition
to compose a diet of xed macronutrient balance. The
diet composed by the dogs in our study (P:F:C = 30:63:7)
was lower in protein-derived energy than previous stud-
ies have demonstrated for the domestic cat and predatory
shes, but comparable to farmed mink. Many interesting
Behavioral EcologyPage 10 of 12
by guest on December 6, 2012 from
questions remain regarding the origins of the diversity of
nutritional priorities among predators. In domesticated
animals, these questions are complicated but particularly
interesting because of the different timescales involved
(predomestication, early domestication, and recent), and
a considerable amount of work remains to be done before
robust generalizations regarding the inuence of the differ-
ent periods can be drawn. On a shorter timescale, there are
many interesting questions concerning the contributions
of genetically evolved regulatory responses versus the role
of individual experience. The nding in the present study
that macronutrient selection in dogs is inuenced by expe-
rience, combined with our earlier demonstration that the
same is true for cats, demonstrates the need for integrated
studies that span timescales from the evolutionary to the
developmental. Finally, the present study has focused on
macronutrition, based on a large body of data showing that
protein, fat, and carbohydrates exert a powerful inuence
on the nutritional responses of many animals (Simpson and
Raubenheimer 2012). We suggest that nutritional geometry
provides a powerful framework for future studies to inves-
tigate the roles of micronutrients (e.g., minerals) in the
dietary responses of companion animals and their interac-
tions with macronutrients.
Supplementary material can be found at http://www.beheco.
This work was funded by the WALTHAM Centre for Pet
Nutrition, part of Mars Petcare. D.R. is part-funded by the
National Research Centre for Growth and Development,
New Zealand. S.J.S. was supported by an Australian Research
Council Laureate Fellowship.
The authors would like to acknowledge the skills and expertise of
many people at the WALTHAM Centre for Pet Nutrition involved
in feeding and caring for the dogs in these experiments. The tech-
nical assistance of Matthew Gilham and Gaëlle Thomas is greatly
Akey JM, Ruhe AL, Akey DT, Wong AK, Connelly CF, Madeoy
J, Nicholas TJ, Neff MW. 2010. Tracking footprints of arti-
cial selection in the dog genome. Proc Natl Acad Sci USA.
Beja-Pereira A, Luikart G, England PR, Bradley DG, Jann OC,
Bertorelle G, Chamberlain AT, Nunes TP, Metodiev S, Ferrand N,
et al. 2003. Gene-culture coevolution between cattle milk protein
genes and human lactase genes. Nat Genet. 35:311–313.
Berthoud H-R, Seeley RJ. 2000. Neural and metabolic control of mac-
ronutrient intake. Boca Raton: CRC Press.
Bradshaw JWS. 2006. The evolutionary basis for the feeding behav-
ior of domestic dogs (Canis familiaris) and cats (Felis catus). J Nutr.
Burger J, Kirchner M, Bramanti B, Haak W, Thomas MG. 2007.
Absence of the lactase-persistence-associated allele in early
Neolithic Europeans. Proc Natl Acad Sci USA. 104:3736–3741.
Cotter SC, Simpson SJ, Raubenheimer D, Wilson K. 2010.
Macronutrient balance mediates trade-offs between immune func-
tion and life history traits. Funct Ecol. 25:186–198.
Darwin C. 1868. The variation of animals and plants under domesti-
cation. London: John Murray.
Driscoll CA, Macdonald DW, O’Brien SJ. 2009. From wild animals
to domestic pets, an evolutionary view of domestication. Proc Natl
Acad Sci USA. 106:9971–9978.
Eisert R. 2011. Hypercarnivory and the brain: protein requirements
of cats reconsidered. J Comp Physiol B. 181:1–17.
Gadd CA, Raubenheimer D. 2000. Nutrient-specic learning in an
omnivorous insect: the American cockroach Periplaneta americana
L.learns to associate dietary protein with the odors citral and car-
vone. J Ins Behav. 13:851–864.
Hawlena D, Schmitz OJ. 2010. Herbivore physiological response to
predation risk and implications for ecosystem nutrient dynamics.
Proc Natl Acad Sci USA. 107:15503–15507.
Hewson-Hughes AK, Hewson-Hughes VL, Miller AT, Hall SR,
Simpson SJ, Raubenheimer D. 2011. Geometric analysis of macro-
nutrient selection in the adult domestic cat, Felis catus. J Exp Biol.
Jensen K, Mayntz D, Tøft S, Clissold F, Hunt J, Raubenheimer
D, Simpson SJ. 2012. Optimal foraging for specic nutri-
ents in predatory beetles. Proc Royal Soc B. doi:10.1098/
Jensen K, Mayntz D, Tøft S, Raubenheimer D, Simpson SJ. 2011.
Nutrient regulation in the wolf spider Pardosa prativaga. Anim
Behav. 81:993–999.
Lee KP, Simpson SJ, Clissold FJ, Brooks R, Ballard JWO, Taylor PW,
Soran N, Raubenheimer D. 2008. Lifespan and reproduction in
Drosophila: new insights from nutritional geometry. Proc Natl Acad
Sci USA. 105:2498–2503.
Lindeberg S, Cordain L, Eaton SB. 2003. Biological and clini-
cal potential of a Palaeolithic diet. J Nutr Environ Med.
Maklakov AA, Simpson SJ, Zajitschek F, Hall MD, Dessmann J,
Clissold F, Raubenheimer D, Bonduriansky R, Brooks RC. 2008.
Sex-specic tness effects of nutrient intake on reproduction and
lifespan. Curr Biol. 18:1062–1066.
Mayntz D, Nielsen VH, Sørensen A, Toft S, Raubenheimer D,
Hejlesen C, Simpson SJ. 2009. Balancing of protein and lipid by
a mammalian carnivore, the mink (Mustela vison). Anim Behav.
Mayntz D, Raubenheimer D, Salomon M, Toft S, Simpson SJ. 2005.
Nutrient-specic foraging in invertebrate predators. Science.
Milton K. 2000. Back to basics: why foods of wild primates have rel-
evance for modern human health. Nutrition. 16:480–483.
National Research Council (NRC). 2006. Nutrient requirements of
dogs and cats. Washington (DC): National Academy Press.
Peterson RO, Ciucci P. 2003. The wolf as a carnivore. In: Mech LD,
Boitani L, editors. Wolves: behaviour, ecology and conservation.
Chicago (IL): University of Chicago Press. p. 104–130.
Piper MDW, Partridge L, Raubenheimer D, Simpson SJ. 2011.
Dietary restriction and aging: a unifying perspective. Cell Metab.
Plantinga EA, Bosch G, Hendriks WH. 2011. Estimation of the dietary
nutrient prole of free-roaming feral cats: possible implications for
nutrition of domestic cats. Br J Nutr. 106:S35–S48.
Prothero J. 1995. Bone and fat as a function of bodyweight in adult
mammals. Comp Biochem Physiol A. 111:633–639.
Raubenheimer D. 2011. Toward a quantitative nutritional ecology:
the right-angled mixture triangle. Ecol Monogr. 81:407–427.
Raubenheimer D, Mayntz D, Simpson SJ, Tøft S. 2007. Nutrient-
specic compensation following overwintering diapause in a gener-
alist predatory invertebrate: implications for intraguild predation.
Ecology. 88:2598–2608.
Raubenheimer D, Simpson SJ. 1997. Integrative models of nutrient
balancing: application to insects and vertebrates. Nutr Res Rev.
Raubenheimer D, Simpson SJ, Mayntz D. 2009. Nutrition, ecology
and nutritional ecology: toward an integrated framework. Funct
Ecol. 23:4–16.
Raubenheimer D, Tucker D. 1997. Associative learning by locusts:
pairing of visual cues with consumption of protein and carbohy-
drate. Anim Behav. 54:1449–1459.
Romsos DR, Ferguson D. 1983. Regulation of protein intake in adult
dogs. J Am Vet Med Assoc. 182:41–43.
Ruohonen K, Simpson SJ, Raubenheimer D. 2007. A new approach to
diet optimisation: a re-analysis using European whitesh (Coregonus
lavaretus). Aquaculture. 267:147–156.
Simpson SJ, Raubenheimer D. 2005. Obesity: the protein leverage
hypothesis. Obes Rev. 6:133–142.
Hewson-Hughes etal. • Macronutrient selection indogs Page 11 of 12
by guest on December 6, 2012 from
Simpson SJ, Raubenheimer D. 2012. The nature of nutrition: a
unifying framework from animal adaptation to human obesity.
Princeton: Princeton University Press.
Simpson SJ, Sibly RM, Lee KP, Behmer ST, Raubenheimer D. 2004.
Optimal foraging when regulating intake of multiple nutrients.
Anim Behav. 68:1299–1311.
Simpson SJ, White PR. 1990. Associative learning and locust feed-
ing: evidence for a “learned hunger” for protein. Anim Behav.
Stahler DR, Smith DW, Guernsey DS. 2006. Foraging and feeding
ecology of the gray wolf (Canis lupus): lessons from Yellowstone
National Park, Wyoming, USA. J Nutr. 136:1923S–1926S.
Tôrres CL, Hickenbottom SJ, Rogers QR. 2003. Palatability affects the
percentage of metabolizable energy as protein selected by adult
beagles. J Nutr. 133:3516–3522.
Vilà C, Savolainen P, Maldonado JE, Amorim IR, Rice JE,
Honeycutt RL, Crandall KA, Lundeberg J, Wayne RK. 1997.
Multiple and ancient origins of the domestic dog. Science.
vonHoldt BM, Pollinger JP, Lohmueller KE, Han E, Parker HG,
Quignon P, Degenhardt JD, Boyko AR, Earl DA, Auton A, et al.
2010. Genome-wide SNP and haplotype analyses reveal a rich his-
tory underlying dog domestication. Nature. 464:898–903.
Warbrick-Smith J, Behmer ST, Lee KP, Raubenheimer D, Simpson SJ.
2006. Evolving resistance to obesity in an insect. Proc Natl Acad Sci
USA. 103:14045–14049.
Warbrick-Smith J, Raubenheimer D, Simpson SJ, Behmer ST. 2009.
Three hundred and fty generations of extreme food specializa-
tion: testing predictions of nutritional ecology. Entom Exp App.
Wayne RK, Ostrander EA. 2007. Lessons learned from the dog
genome. Trends Genet. 23:557–567.
Wayne RK, vonHoldt BM. 2012. Evolutionary genomics of dog domes-
tication. Mamm Genome. 23:3–18.
Westoby M. 1978. What are the biological bases of varied diets? Am
Nat. 112:627–631.
Wilder SM, Rypstra AL. 2008. Diet quality affects mating behaviour
and egg production in a wolf spider. Anim Behav. 76:439–445.
Behavioral Ecology
Page 12 of 12
by guest on December 6, 2012 from
... Food is typically easily accessible, consumed quickly, and on average more effective than attention or praise for dogs (Feuerbacher & Wynne, 2012. Previous studies suggest dogs can engage in a discrimination of foods by scent and flavor and prefer foods that have meat, sugars, or fat (Hewson-Hughes et al., 2013;Houpt et al., 1978;Pétel et al., 2018;Rao et al., 2018;Thompson et al., 2016;Tôrres et al., 2003). Nevertheless, individual dogs have unique food preferences (Cameron et al., 2013;Riemer et al., 2018;Vicars et al., 2014). ...
... Further, individual dogs' preference results often diverged from dog food preferences identified in the literature. For example, studies suggest that dogs tend to choose foods with higher meat content (Callon et al., 2017;Houpt et al., 1978;Pétel et al., 2018;Riemer et al., 2018;Thompson et al., 2016), sucrose over bland foods (Houpt et al., 1978;Tôrres et al., 2003), moist over dry foods (de Brito et al., 2010;Rao et al., 2018), and fat over protein or carbohydrates (Hewson-Hughes et al., 2013). Although several dogs' preferences in the current investigation matched these general guidelines F I G U R E 5 Progressive ratio and fixed ratio reinforcer assessments. ...
Behavioral interventions for animals typically require the inclusion of programmed reinforcers. Although pet owners and human caregivers can often identify items that the animal will consume, preference assessments can more accurately determine relative preference rankings between various stimuli, which is important given that higher preferred items tend to function as more effective reinforcers than lower preferred items. Preference assessments have been developed to identify rankings for a variety of stimuli across species, including the domesticated dog (Canis lupus familiaris). However, previous preference assessments for dogs were developed for laboratory use and could be challenging for dog owners to perform alone. The purpose of this study was to modify existing dog preference assessment methods to produce a valid and feasible preference assessment for dog owners. Results suggest that the preference assessment identified preference rankings for individual dogs. Owners were able to implement the protocol with high integrity and found the protocol acceptable.
... While the specific minimum nutrient requirements of cats and dogs have been established, research has shown that cats and dogs are able to select a 'target intake' of protein, fat, and carbohydrates to achieve nutritional adequacy when given the choice between diets differing in macronutrient composition [48,49]. ...
... An extensive study by Hewson-Hughes et al. [49] used geometric analysis to assess macronutrient selection in dogs when presented with six dry-format (extruded) diets and six wet-format (retorted) diets for five different dog breeds. It was found that after initially selecting a diet significantly lower in fat, dogs were able to regulate their dietary macronutrient level based on the metabolisable energy compositions of 30% protein, 63% fat, and 7% carbohydrate, with values showing similarities across the different breeds. ...
Full-text available
Simple Summary: The pet food industry is growing rapidly globally. Although new products continue to be developed, research into their palatability still largely uses traditional methods. Testing focuses on the amount of food consumed, but little consideration is given to why differences are observed and which ingredients are most important. This review will discuss the feeding behaviour and nutritional requirements of dogs and cats, the main types of pet foods produced currently, and the current methods used for assessing palatability. Finally, the opportunities to use better methods to develop foods that are more palatable and understand the nutritional factors responsible for driving intake are discussed. Abstract: The pet food industry is an important sector of the pet care market that is growing rapidly. Whilst the number of new and innovative products continues to rise, research and development to assess product performance follows traditional palatability methodology. Pet food palatability research focuses on the amount of food consumed through use of one-bowl and two-bowl testing, but little understanding is given to why differences are observed, particularly at a fundamental ingredient level. This review will highlight the key differences in feeding behaviour and nutritional requirements between dogs and cats. The dominant pet food formats currently available and the ingredients commonly included in pet foods are also described. The current methods used for assessing pet food palatability and their limitations are outlined. The opportunities to utilise modern analytical methods to identify complete foods that are more palatable and understand the nutritional factors responsible for driving intake are discussed.
... individuals can select from a variety of foods offered ad libitum) with species under human care increased knowledge on the nutrient composition that would be ingested when no restrictions are imposed (e.g. Erlenbach et al., 2014;Felton et al., 2016;Hewson-Hughes et al., 2013). Furthermore, the nutrients typically considered are mainly restricted to (digestible) protein and carbohydrates, and fat (Bosch et al., 2015;Coogan et al., 2018;Erlenbach et al., 2014), but little attention has been given to fiber (indigestible but fermentable carbohydrates or protein) as a separate main nutrient. ...
Full-text available
The dietary nutrient profile has metabolic significance and possibly contributes to species' foraging behavior. The brown bear (Ursus arctos) was used as a model species for which dietary ingredient and nutrient concentrations as well as nutrient ratios were determined annually, seasonally and per reproductive class. Brown bears had a vertebrate- and ant-dominated diet in spring and early summer and a berry-dominated diet in fall, which translated into protein-rich and carbohydrate-rich diets, respectively. Fiber concentrations appeared constant over time and averaged at 25% of dry matter intake. Dietary ingredient proportions differed between reproductive classes; however, these differences did not translate into a difference in dietary nutrient concentrations, suggesting that bears manage to maintain similar nutrient profiles with selection of different ingredients. In terms of nutrient ratios, the dietary protein to non-protein ratio, considered optimal at around 0.2 (on metabolizable energy basis), averaged around 0.2 in this study in fall and around 0.8 in spring and summer. We introduced the minimal non-fat to fat ratio necessary for efficient maintenance metabolism. This ratio varied across seasons but never fell beneath the theoretically estimated minimum to ensure metabolic efficiency. This population thus managed to ingest diets that never exerted a lack of glucogenic substrate, suggesting that metabolic efficiency may either be a driver of active diet selection or that natural resources available to bears did not constitute a constraint in this respect. Given the considerable proportion of fiber in the diet of brown bears, the relevance of this nutrient and its role in foraging behavior might be underestimated.
... However, only protein and fat are essential to the dog, they do not have dietary requirements for carbohydrates 43 . In addition, research indicates that dogs intuitively select a dietary macronutrient composition dominated by protein and fat 29,44 . Feeding dogs a NPMD is known to affect serum, urine and fecal metabolite concentrations when compared to feeding an UPCD (dry dog food) diet 45,46 although the meaning of these changes still requires further research. ...
Full-text available
Diet has a key role in the homeostasis of the gut microenvironment, influencing the microbiome, the gut barrier, host immunity and gut physiology. Yet, there is little information on the role of early diet in the onset of inflammatory gastrointestinal disorders later in life, especially in dogs. Therefore, the aim of the present cross-sectional, epidemiological study with longitudinal data, was to explore associations of companion dogs’ early life diet style and food items with owner-reported chronic enteropathy (CE) incidence in later life. Food frequency questionnaire data from Finnish companion dogs was analyzed using principal component analysis and logistic regression. We found that feeding a non-processed meat-based diet and giving the dog human meal leftovers and table scraps during puppyhood (2–6 months) and adolescence (6–18 months) were protective against CE later in life. Especially raw bones and cartilage as well as leftovers and table scraps during puppyhood and adolescence, and berries during puppyhood were associated with less CE. In contrast, feeding an ultra-processed carbohydrate-based diet, namely dry dog food or “kibble” during puppyhood and adolescence, and rawhides during puppyhood were significant risk factors for CE later in life.
... For example, many different fat sources, particularly the highly polyunsaturated fats and/or the lower saturated fats, can condition flavour preferences (Ackroff et al., 2005). Other researchers have suggested that, if dogs are offered a range of nutritionally variable foods, they make food choices that maintain a specific ratio of macronutrients (Hewson-Hughes et al., 2013). These studies imply that food consumption reflects a physiological need rather than choices based on palatability or food availability (Hewson-Hughes et al., 2016). ...
Full-text available
Background: Including adequate concentrations of antioxidants in dog diets has been recommended to reduce their vulnerability to the action of free radicals and reactive oxygen species (ROS). Oxidative stress in dogs has been associated with a wide range of diseases and disorders, as well as with ageing. There are few reports about the influence of diet on dog's antioxidant profile and oxidative stress. Objective: The objective of this study was to evaluate the effect of four types of dry dog food on the oxidative/antioxidant profile of dogs. Methods: Six Beagle dog males were used. The study included four experimental diets (dry foods A-D). Each dry food was supplied for 5 weeks to all dogs, for a total of 24 weeks, including an adaptation week between one food and another. For each dry dog food, the total phenolic content (TPC), total antioxidant capacity (TAC) and cytotoxicity were evaluated. Each week, a blood sample was collected to measure ROS and TAC of plasma. A crossover repeated measures design was used. Mixed models were adjusted, and means were compared using the Tukey test. Results: Food A had the highest values for TPC and TAC. Food C had the lowest levels of ROS, whereas food B had the highest TAC in the blood plasma. The dog had a significant influence on the redox state of its blood plasma, even when the same dog was fed the different dry foods. Conclusion: Dry dog food influences the oxidative/antioxidant profile of dog's blood plasma; however, this seems to be unrelated to the antioxidant profile of the food.
... Establishing the ecological consequences of dietary specialisation ("extrinsic" outcomes) is usually a focus for isotopic (δ-and p-space) approaches, but implications regarding consumer resource acquisition, physiological performance and fitness ("intrinsic" outcomes) are equally important, albeit underexplored (but see Costa-Pereira, Toscano, et al., 2019b). Studies across trophic guilds consistently demonstrate how nutrition, and regulation of macronutrient (protein, lipid, carbohydrate) balance, mediates feeding choices (Behmer, 2009;Coogan et al., 2017;Erlenbach et al., 2014;Felton et al., 2016;Hewson-Hughes et al., 2013;Raubenheimer et al., 2005;Rowe et al., 2018), individual fitness and performance (Jensen et al., 2012;Simpson et al., 2004) and broader ecological phenomena (e.g. migrations, Nie et al., 2015;Raubenheimer et al., 2009;Simpson et al., 2006). ...
Full-text available
Dietary specialisations are important determinants of ecological structure, particularly in species with high per‐capita trophic influence like marine apex predators. These species are, however, among the most challenging in which to establish spatiotemporally integrated diets. We introduce a novel integration of stable isotopes with a multidimensional nutritional niche framework that addresses the challenges of establishing spatiotemporally integrated nutritional niches in wild populations, and apply the framework to explore individual diet specialisation in a marine apex predator, the white shark Carcharodon carcharias . Sequential tooth files were sampled from juvenile white sharks to establish individual isotopic (δ‐space; δ ¹³ C, δ ¹⁵ N, δ ³⁴ S) niche specialisation. Bayesian mixing models were then used to reveal individual‐level prey (p‐space) specialisation, and further combined with nutritional geometry models to quantify the nutritional (N‐space) dimensions of individual specialisation, and their relationships to prey use. Isotopic and mixing model analyses indicated juvenile white sharks as individual specialists within a broader, generalist, population niche. Individual sharks differed in their consumption of several important mesopredator species, which suggested among‐individual variance in trophic roles in either pelagic or benthic food webs. However, variation in nutrient intakes was small and not consistently correlated with differences in prey use, suggesting white sharks as nutritional specialists and that individuals could use functionally and nutritionally different prey as complementary means to achieve a common nutritional goal. We identify how degrees of individual specialisation can differ between niche spaces (δ‐, p‐ or N‐space), the physiological and ecological implications of this, and argue that integrating nutrition can provide stronger, mechanistic links between diet specialisation and its intrinsic (fitness/performance) and extrinsic (ecological) outcomes. Our time‐integrated framework is adaptable for examining the nutritional consequences and drivers of food use variation at the individual, population or species level.
Full-text available
Fat has high energy density and is considered one of the primary energy sources for dogs, however, increasing fat level in dry dog food has been challenging due to the lubrication and limitation of the coating system. The objective was to determine the effect of whole soybeans (WSB) on nutrient digestibility, stool quality, and palatability by dogs. The corn gluten meal, chicken fat, and brewers rice were replaced by WSB at 10, 20, and 30% (WSB10, WSB20, and WSB30, respectively) in the base diet (WSB0). Twelve beagles were randomly assigned. The digestibility trial was duplicated 4 × 4 Latin square design where dogs were allowed a 9-d adaptation followed by a 5-d total fecal collection for each period. Least-square means were analyzed with a single degree of freedom contrasts and significance at α = 0.05. Palatability was determined with a 2-bowl test by 20 beagles for 2 d with each WSB diet compared to the WSB0. First choice preference between two diets and total food consumption were recorded. Individual intake ratios (IR) were calculated (intake of each diet/total intake) for each dog. First choice (FC) was analyzed by a Chi-square probability, and the diet consumption was compared by a Wilcoxon signed rank test and a 2-way analysis of variance. Fecal moisture, output, and defecation frequency increased linearly ( P < 0.05) as WSB increased. Apparent total tract digestibility of dry matter, organic matter, crude protein, fat, and gross energy decreased linearly ( P < 0.05) as dogs fed the increased level of WSB. The fresh fecal pH in dogs decreased linearly ( P < 0.05) as WSB content increased. The acetate, propionate, and the total short-chain fatty acid concentration increased linearly ( P < 0.05) while the total branched-chain fatty acid concentration decreased linearly ( P < 0.05) as WSB increased. Dogs had greater ( P < 0.05) FC for WSB diets than WSB0, but there was no difference among treatments for diet consumption and IR. In conclusion, additional thermal processing before extrusion may improve nutrient digestibility of WSB. The stool quality and palatability were not affected, and fermentation in hindgut increased by WSB by dogs.
Full-text available
A previously healthy 5-year-old, female, spayed, mixed breed dog presented for acute abdominal distension and non-productive retching after consuming a large volume of kibble dog food. Abdominal radiographs confirmed food engorgement, and standard medical therapy was initiated with intravenous fluids and analgesics. After 36 hours of persistent abdominal distension and progressive hyperlactatemia, an exploratory laparo-tomy revealed generalised ischaemia throughout the gastrointestinal tract, spleen, and liver. Due to a poor prognosis, humane euthanasia was elected. Postmortem examination confirmed severe necro-haemorrhagic gastroenteritis and multifocal microthrombi. Clostridium perfringens was cultured from the small intestines, which may represent either an underlying cause or an opportunistic proliferation.
Full-text available
Animals regulate their food intake to maximize the expression of fitness traits but are forced to trade off the optimal expression of some fitness traits because of differences in the nutrient requirements of each trait ("nutritional trade-offs"). Nutritional trade-offs have been experimentally uncovered using the geometric framework for nutrition (GF). However, current analytical methods to measure such responses rely on either visual inspection or complex models of vector calculations applied to multidimensional performance landscapes, making these approaches subjective or conceptually difficult, computationally expensive, and, in some cases, inaccurate. Here, we present a simple trigonometric model to measure nutritional trade-offs in multidimensional landscapes (nutrigonometry) that relies on the trigonometric relationships of right-angle triangles and thus is both conceptually and computationally easier to understand and use than previous quantitative approaches. We applied nutrigonometry to a landmark GF data set for comparison of several standard statistical models to assess model performance in finding regions in the performance landscapes. This revealed that polynomial (Bayesian) regressions can be used for precise and accurate predictions of peaks and valleys in performance landscapes, irrespective of the underlying structure of the data (i.e., individual food intakes vs. fixed diet ratios). We then identified the known nutritional trade-off between life span and reproductive rate in terms of both nutrient balance and concentration for validation of the model. This showed that nutrigonometry enables a fast, reliable, and reproducible quantification of nutritional trade-offs in multidimensional performance landscapes, thereby broadening the potential for future developments in comparative research on the evolution of animal nutrition.
Conference Paper
Full-text available
The foraging and feeding ecology of gray wolves is an essential component to understanding the role that top carnivores play in shaping the structure and function of terrestrial ecosystems. In Yellowstone National Park (YNP), predation studies on a highly visible, reintroduced population of wolves are increasing our understanding of this aspect of wolf ecology. Wolves in YNP feed primarily on elk, despite the presence of other ungulate species. Patterns of prey selection and kill rates in winter have varied seasonally each year from 1995 to 2004 and changed in recent years as the wolf population has become established. Wolves select elk based on their vulnerability as a result of age, sex, and season and therefore kill primarily calves, old cows, and bulls that have been weakened by winter, Summer scat analysis reveals an increased variety in diet compared with observed winter diets, including other ungulate species, rodents, and vegetation. Wolves in YNP hunt in packs and, upon a successful kill, share in the evisceration and consumption of highly nutritious organs first, followed by major muscle tissue, and eventually bone and hide. Wolves are adapted to a feast-or-famine foraging pattern, and YNP packs typically kill and consume an elk every 2-3 d. However, wolves in YNP have gone without fresh meat for several weeks by scavenging off old carcasses that consist mostly of bone and hide. As patterns of wolf density, prey density, weather, and vulnerability of prey change, in comparision with the conditions of the study period described here, we predict that there will also be significant changes in wolf predation patterns and feeding behavior.
Full-text available
A recent area of progress in nutritional ecology is a growing awareness that nutritional phenotypes are best understood in a multidimensional context, where foraging is viewed as a process of balancing the intake and use of multiple nutrients to satisfy complex and dynamic nutrient needs. Numerous laboratory studies have shown that this view can yield novel insights into unresolved questions and provide a framework for generating new hypotheses. By contrast, progress with this multidimensional view has been slow in the arena of ultimate interest to functional biologists, the field. One reason for this is that the Geometric Framework for nutrition that has been extensively used in laboratory experiments focuses on amounts of nutrients (e.g., required, eaten, or retained), and such data are typically very difficult or impossible to collect for most free-ranging animals. Further, many problems in field-based nutritional ecology involve comparisons of mixtures that are expressed as proportions (e.g., food, diet, body, or fecal compositions), rather than absolute amounts. As yet, however, no geometric framework has been established in nutritional ecology for this. Here I recommend an approach for the geometric analysis of nutritional mixtures, and illustrate its use in a variety of contexts by reanalyzing published data. Despite its simplicity, this approach holds considerable promise for furthering the study of field-based nutritional ecology.
Full-text available
Nutrient balancing is well known in herbivores and omnivores, but has only recently been demonstrated in predators. To test how a predator might regulate nutrients when the prey varies in nutrient composition, we restricted juvenile Pardosa prativaga wolf spiders to diets of one of six fruit fly, Drosophila melanogaster, prey types varying in lipid:protein composition during their second instar. We collected all fly remnants to estimate food and nutrient intake over each meal. The spiders adjusted their capture rate and nutrient extraction in response to prey mass and nutrient composition irrespective of energy intake. Intake was initially regulated to a constant lipid plus protein mass, but later spiders fed on prey with high proportions of protein increased consumption relative to spiders fed on other prey types. This pattern indicates that the spiders were prepared to overconsume vast amounts of protein to gain more lipids and energy. The spiders also regulated protein after ingestion, and ingested protein was incorporated less efficiently into body tissue when the prey was protein rich. Despite both preand postingestive nutrient regulation, the body lipid:protein compositions of the spiders were highly affected by the nutrient compositions of their prey, and growth in carapace length and lean body mass increased with increasing prey protein:lipid ratio. Our results demonstrate that prey nutrient composition affects these predators, but also that the spiders possess behavioural and physiological adaptations that lead to partial compensation for these effects.
Conference Paper
The dentition, sense of taste and meal patterning of domestic dogs and cats can be interpreted in terms of their descent from members of the order Carnivora. The dog is typical of its genus, Canis, in its relatively unspecialized dentition, and a taste system that is rather insensitive to salt. The preference of many dogs for large infrequent meals reflects the competitive feeding behavior of its pack-hunting ancestor, the wolf Canis lupus. However, its long history of domestication, possibly 100,000 years, has resulted in great intraspecific diversity of conformation and behavior, including feeding. Morphologically and physiologically domestic cats are highly specialized carnivores, as indicated by their dentition, nutritional requirements, and sense of taste, which is insensitive to both salt and sugars. Their preference for several small meals each day reflects a daily pattern of multiple kills of small prey items in their ancestor, the solitary territorial predator Felis silvestris. Although in the wild much of their food selection behavior must focus on what to hunt, rather than what to eat, cats do modify their food preferences based on experience. For example, the "monotony effect" reduces the perceived palatability of foods that have recently formed a large proportion of the diet, in favor of foods with contrasting sensory characteristics, thereby tending to compensate for any incipient nutritional deficiencies. Food preferences in kittens during weaning are strongly influenced by those of their mother, but can change considerably during at least the first year of life.
Nutrition has long been considered more the domain of medicine and agriculture than of the biological sciences, yet it touches and shapes all aspects of the natural world. The need for nutrients determines whether wild animals thrive, how populations evolve and decline, and how ecological communities are structured.The Nature of Nutritionis the first book to address nutrition's enormously complex role in biology, both at the level of individual organisms and in their broader ecological interactions. Stephen Simpson and David Raubenheimer provide a comprehensive theoretical approach to the analysis of nutrition--the Geometric Framework. They show how it can help us to understand the links between nutrition and the biology of individual animals, including the physiological mechanisms that determine the nutritional interactions of the animal with its environment, and the consequences of these interactions in terms of health, immune responses, and lifespan. Simpson and Raubenheimer explain how these effects translate into the collective behavior of groups and societies, and in turn influence food webs and the structure of ecosystems. Then they demonstrate how the Geometric Framework can be used to tackle issues in applied nutrition, such as the problem of optimizing diets for livestock and endangered species, and how it can also help to address the epidemic of human obesity and metabolic disease.
There is a need to develop new methods for formulating optimal feeds for farmed fish, and for making explicit assessments of the criteria by which the success of a formulation may be judged. We show how a combination of mixture design theory and state-space models of nutrition (the Geometric Framework, GF) can be used to derive a 5-step protocol for multi-criterion diet optimisation. Step 1 involves selecting the focal nutritional axes for modelling; step 2 uses mixture theory to choose an optimal selection of experimental diets to test in experiments; step 3 entails using GF to plot and interpret intake and growth arrays; step 4 involves plotting response variables onto intake arrays, and step 5 uses multi-criterion optimisation to combine and weigh several relevant response variables. We illustrate the approach by re-analysing data from dietary studies on European whitefish.
Summary • The science of nutritional ecology spans a wide range of fields, including ecology, nutrition, behaviour, morphology, physiology, life history and evolutionary biology. But does nutritional ecology have a unique theoretical framework and research program and thus qualify as a field of research in its own right? • We suggest that the distinctive feature of nutritional ecology is its integrative nature, and that the field would benefit from more attention to formalizing a theoretical and quantitative framework for developing this. • Such a framework, we propose, should satisfy three minimal requirements: it should be nutritionally explicit, organismally explicit, and ecologically explicit. • We evaluate against these criteria four existing frameworks (Optimal Foraging Theory, Classical Insect Nutritional Ecology, the Geometric Framework for nutrition, and Ecological Stoichiometry), and conclude that each needs development with respect to at least one criterion. • We end with an initial attempt at assessing the expansion of our own contribution, the Geometric Framework, to better satisfy the criterion of ecological explicitness.