Available via license: CC BY 4.0
Content may be subject to copyright.
J Anim Physiol Anim Nutr. 2017;1–8. wileyonlinelibrary.com/journal/jpn
|
1
Received:16October2016
|
Accepted:11July2017
DOI: 10.1111/jpn.12794
ORIGINAL ARTICLE
Macronutrient intake of dogs, self- selecting diets varying in
composition offered ad libitum
M. T. Roberts1,2 | E. N. Bermingham1 | N. J. Cave2 | W. Young1 | C. M. McKenzie3 |
D.G. Thomas2
1Food Nutrition & Health Team, Food &
Bio-basedProducts,AgResearchGrasslands,
Palmerston North, New Zealand
2InstituteofVeterinaryAnimalBiomedical
Sciences, Massey University, Palmerston
North, New Zealand
3Bioinformatics&StatisticsTeam,AgResearch
Grasslands,PalmerstonNorth,NewZealand
Correspondence
DavidThomas,AnimalProductionandHealth,
InstituteofVeterinary,AnimalandBiomedical
Sciences, Massey University, Palmerston
North, New Zealand.
Email:d.g.thomas@massey.ac.nz
Funding information
AgriculturalandMarketingResearch
andDevelopmentTrust,Grant/Award
Number:A15013;K9NaturalFoodLimited;
AgResearchLimited,Grant/AwardNumber:
A21247
Summary
Thedietofthedomesticdoghaschangedsignificantlyfromthatofitswolfancestor,
withto dateonlytwostudies having examinedmacronutrientself-selectionin dogs.
Whilstthefirstfocusedsolelyonprotein intake, determining an intake of 30% me-
tabolisableenergy(ME),thesecondinvestigateddietaryprotein,fatandcarbohydrate
(PFC),indicatinganintakeratioof30:63:7%byenergy.Thisstudy’saimwastofurther
elucidatemacronutrientintakebyprovidinggreatermacronutrientrange,energycon-
tent, and to investigate over a longer duration than previous studies. Fifteen adult
dogswere given access to threewet diets providing 500% ofdaily ME, twice daily
over10days.The diets were nutritionally complete andformulatedusing the same
fouringredientsindifferentproportionstosupplyhighlevelsofprotein(58%ME),fat
(86%ME)orcarbohydrate(54%ME).Overallfatandcarbohydrateconsumptionsig-
nificantly declined from 6,382 to 917kcals per day (p<0.001) and 553 to
214kcalsday−1(p<.01) respectively. Protein intake, however, remained constant
overthestudyandrangedfrom4,786to4,156kcalsday−1. Such results impacted on
percentagetotal energy intake, withfatdecreasingfrom68%to 52% (p<.001) and
proteinincreasingfrom29%to44%(p<.01).Ourfindingssuggestthatdogsstillpos-
sessa“feastorfamine”mentality,whereinenergydensefatisprioritisedoverprotein
initially.Withcontinuedfeedingover10days,atransitiontoamorebalancedenergy
contributionfrombothmacronutrientsisevident.Thestudyalsoshowsthatgiventhe
option,dogs do not select carbohydrateto be a significantportionof the diet. The
health implications of such dietary selection are of interest.
KEYWORDS
dietarycomposition,dog,macronutrient,self-selection
1 | INTRODUCTION
Whilst archaeological records cannot determine whether domestic
dogsoriginated from a single wolfpopulationorarose from multiple
populations at different times (Frantz etal., 2016; Vilà etal., 1997),
dogs are the only largecarnivore to have been domesticated, most
likelyoverawidegeographicarea(vonHoldtetal.,2010).Byinheriting
suchwolfancestry,thedomesticdogisclassifiedasa carnivore,with
teeth adapted for grasping and tearing; however, they also possess
omnivoroustraits(Serpell,1995).The doghasarequirementforboth
proteinandfat(AssociationofAmericanFeedControlOfficials,2016;
NationalResearchCouncil,2006),butnotforcarbohydrate,despitere-
centfindingsthatshowthatdomesticdogsmayhaveevolutionaryad-
aptationsforimprovedcarbohydratedigestion(Axelssonetal.,2013).
ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsuse,distributionandreproductioninanymedium,
providedtheoriginalworkisproperlycited.
©2017TheAuthors.Journal of Animal Physiology and Animal NutritionPublishedbyBlackwellVerlagGmbH.
2
|
ROBERTS ET al.
The macronutrient composition of modern dog foods canvary
significantlydepending onthe format fed.This is largely due to the
manufacturingprocessesrequiredtoproducethefood.Forexample,
kibbled diets typically contain 16%–38% protein, 6%–18% fat and
40%–60% carbohydrate (dry matter basis). However,wet/raw diets
typically containno or low levels (<10%) of carbohydratecombined
with higher levels of protein and fat (45% and 50% respectively).
Fromadietaryperspectivecommercialdrydogfoodisbyfarthemost
popularfeedingoption,beingfedto over88%ofdogs(NewZealand
CompanionAnimalCouncil2016).Indeedwhilstanyimpactonhealth
fromfeedingadifferingdietaryformat(dryorwet)andmacronutrient
composition has yet to be determined, obesity and its many associated
conditions such as diabetes, cardiorespiratory disease and urinary dis-
ordersare an increasing health risk forcompanion animals (German,
2006). Indeed in the past 10years, approximately 30%–40% of pet
dogsareclassifiedasbeing overweight,whilstanadditional5%–20%
termed obese (Witzel etal., 2014).Although the establishment of a
targetedmacronutrientintakewillnotinitselfhighlightanyimpacton
health,itmayserveas astartingpointforfutureresearch,wherebya
specific dietary macronutrient composition could be assessed in refer-
encetotheimpactonmarkersofhealth.
The ability of animals to select a macronutrient ratio that optimises
fitness costs (such as lifespanand rate of reproduction) has to date
beenproveninarangeofspecies(Leeetal.,2008;Simpson,Sibly,Lee,
Behmer,& Raubenheimer,2004).Moreover,establishing the macro-
nutrientprofile“targeted”bydogscouldhighlightthepotentialdiffer-
ence between what they want to consume, and what most commercial
diets are providing. Further questions may also be addressed as to
whetheradogwouldunder-eatsomenutrients,andover-eatothersin
an attempt to reach an intuitive predetermined macronutrient profile
when provided with an inappropriate dietary composition.
Froma nutritional standpoint, whilst feeding biologically appro-
priate diets to pet dogs has not currently been shownto provide
any health benefits, raw meat diets have been demonstrated to be
highlydigestible,resultinginlowfaecalvolumeand desirablefaecal
quality(Beloshapka,Duclos,VesterBoler,&Swanson,2012).Inaddi-
tion,itisclearlyapparentthatdomesticateddogsarecurrentlyeating
diets that differ substantially from what their ancestors consumed.
Highlightingthis,Bosch,Hagen-Plantinga,andHendriks(2015)found
that the dietary composition of wild wolves showed the selected
protein–fat–carbohydrateprofilewas54:45:1byenergy.
To date, only two studies have examined dietary macronutrient
selectioninthedog.Whilstthe firststudy appeared to demonstrate
apreferenceforproteinovercarbohydrates(consuming30% protein
by energy), the impact of fat was not fully determined (Romsos &
Ferguson,1983).Amorerecentstudydid,however,allowforallthree
macronutrientstobeself-selectedbydogsofdifferingbreeds,suggest-
ingan overallprotein/fat/carbohydrateratio(P:F:C) ofapproximately
30:63:7%byenergywhenfedcompleteandbalancedwet-baseddiets
(Hewson-Hughesetal.,2012). However,the restriction of dailytotal
foodintakeincertainexperimentalstages (e.g.,100%ofMERforthe
firstsix3-daycyclesofthelearningphase)mayhavelimitedtheextent
by which the animals could fully select the provided diets. In addition,
the structuring ofdifferent feeding phases and diet composition se-
lectedmaypotentiallyhaveinfluencedthedogsfeedingpatterns.
Existingliteraturesuggeststhatwhendogsareprovidedwiththe
ability to self-select a macronutrient ratio, theywill consume 30%
oftheir maintenance energy requirements from protein.However,a
numberofcommercialwetdiets containin excessofthisvalue,with
reportsfromdogownersthatanincreaseinpalatabilityislinkedwith
this factor. Therefore, the hypothesis of the study was that dogs
wouldselectadietconsistingofmorethan30%oftotalenergyfrom
protein. The aim of this study was, therefore, to establish the self-
selectivemacronutrientintakeofdogsbyprovidingthemwitharange
ofdiets, each specifically higherinenergysourced fromproteinand
fatoveralongerduration.Thisconsequentlywillenabletheintuitive
macronutrient capabilities ofthe domestic dog to be examined in a
deeper manner than has previously been conducted. Subsequently
ourfindingswill eitherreinforceorchallengethoseofthepreviously
conductedstudies,withthe potentialtohighlight thatadogmaystill
possessasimilarmacronutrientintaketothatoftheirwildancestors.
2 | MATERIALS AND METHODS
2.1 | Ethics
EthicalapprovalwasgainedfromtheMasseyUniversityAnimalEthics
Committee (MUAEC 15/75), before commencing the experiment.
The dogs were housed at Massey University Canine Nutrition Unit
(Palmerston North, New Zealand), in accordance with the Animal
Welfare(CompanionDogs)CodeofWelfare(2007).
2.2 | Animals
Fifteen Harrier hound dogs (five male and 10 female) were used
throughout the study, comprising of four neutered and one entire
maleandthreeneuteredandsevenentirefemales.Thedogswereall
deemedhealthy based on a physical examination. The mean ageof
thedogsusedinthestudywas7.68years±0.73SEM.Thedogswere
housedinpairsin10mx10m(100m2)outdoorpensoringroupsof
4in grass paddocksmeasuring700m2 for 8hraday. Overnight the
dogswerehousedindoorsinpairswithwaterandbeddingprovided.
2.3 | Diets
Ahighprotein(HP),highfat(HF)andhighcarbohydrate(HC)diet(Table1)
was formulated to meet AAFCO Dog Food Nutrient Profiles for adult
maintenance(AssociationofAmerican FeedControlOfficials,2015).All
diets consisted of the same four ingredients at different inclusion lev-
els, namely maize, lamb loin fat, green tripe and venison mechanically
debonedmeat(MDM)(TableS1).Thelevelsofprotein,fat,ash,moisture
andNFE(nitrogenfreeextract)wereanalysedforeachdiet(Table1).
A 5-dayperiod was used to adapt the dogs onto the test diets,
consistingofa20%dayondayincreaseofanequalmixtureoftheHF,
HPand HC diets,whilst concurrently decreasingtheirexistingcom-
mercialdrydiet(protein–fat–carbohydrateprofile21:23:56byenergy)
|
3
ROBERTS ET al.
by20%. Therefore,by the last day of the adaption period, the dogs
werebeingfedsolelyanequalcombinationoftheexperimentaldiets,
at which point they were deemed to have been fully transitioned.
The10-dayexperimentalphaseofthestudythenstarted,consist-
ingofthe dogsbeingoffered250% oftheir dailyME requirementof
eachdiet,twiceaday(8a.m.and2p.m.;1,500%ME,totalperdayfor
allthreediets).Allthreedietswereofferedsimultaneously.
2.4 | Experimental protocol
Thedogswereweighedatthestart(day1),middle(day 5) and end
(day10)oftheexperimentalperiod(Table2).In orderto assessself-
selected macronutrient consumption, three large plastic bowls each
containing250%ofthe dailyenergyrequirement oftheHF, HCand
HPdietswereprovidedtoeachdog(twicedaily,at8a.m.and2p.m.)
for 10days (Figure1). The position of each bowl was interchanged
at each feeding time to prevent positional bias. A number of feed-
ingdynamicswerealsoobservedbothdirectlybyanobserverduring
eachfeeding period andafterwardsvia the use ofavideo recording
camera(SonyHandycamHDR-SR11E/SR12E)toverifyresults.These
observations involved which diets were approached first, which
diets consisted of any consumption first and which diets were com-
pletelyavoided.Dogswereofferedthedietsuntilsatiatedstatuswas
achieved. This was defined as the point whereby the animal lost inter-
est in any of the diets.
2.5 | Calculations
Protein, fat and carbohydrate energy intakes were determined by
applying modified Atwater factors (protein/carb 3.5Kcalg−1 fat
8.5Kcalg−1) (National Research Council, 2006). As these data were
knownforeachspecificdiet,totalenergyconsumptionwascalculated
bysubtracting thetotalof eachdietprovided toeachdog from that
remaining after each dietary exposure. Additionally macronutrient
ratiowas determined as the overall percentageenergy contribution
thateachmacronutrientmadetoeach diet.Therefore,byaddingthe
differentquantitiesofeachdietconsumedandtherespectiveprotein,
fatandcarbohydrateenergycontribution,thetotalenergyfromeach
macronutrient could be established.
2.6 | Statistical analysis
Separate analyses were conducted for each of the response variables
(i.e.,protein,fat,carbohydrateandprotein:fatratio)againstmeasure-
mentday,usingarandomcoefficientsregressionmodelwhichallowed
forseparate slopes and intercepts to be fitted for each dog.As the
experimentinvolved dogsofboth sexes(5maleand10 female), the
factors“sex”and“reproductive”wereassessedseparately,butnosig-
nificant differences were found, so these factors were not included in
themodel.Modellingwasundertakenusingrsoftware(RCoreTeam,
2016).Alldatawerereportedas interceptand slopewith associated
standarderror(SE).
Fisher’sexacttestwas used to compare the proportionsoffirst
approachedandfirstconsumedforeachofthediets(HP,HFandHC).
Thetestwasperformedwiththestatisticalsoftwarepackageminitab®
16(2010).
Binarylogisticregressionanalysiswasusedtotesttheeffectofdiet
on diet avoidance, with diet avoidance as the binary response variable
(avoidedvs.notavoided)andthediet(HP,HFandHC)aspredictor.
Bodyweight was analysed with a repeated measurements lin-
earmixed model (REML) with the factormeasurementday (levels1,
TABLE1 Macronutrientprofilesofhighprotein(PFC
57.6:41.7:0.7%),highfat(PFC12.8:86.7:0.5%)orhighcarbohydrate
(PFC17.8:27.7:54.5%)dietsofferedat500%maintenanceenergy
requirementstoadultdogs(n=15)for10days
Nutrient DM (g/100 g) HF HC HP
Moisture(asfed) 41.2 26.2 73.0
Protein 23.9 19.3 71.2
Fat(etherextract) 66.4 12.4 21.2
Ash 7.5 4.8 5.5
Carbohydrate 0.9 59.3 0.9
Crudefibre 1.3 4.2 1.2
ME(Kcalkg−1)a6,512 3,805 4,325
DM,DryMatter; HP,HighProtein;HF,High fat;HC,HighCarbohydrate;
ME,Metabolisableenergy.
aCalculated from modified Atwater factors (National Research Council,
2006)
TABLE2 Meanbodyweightofdogs(n=15)offeredhighprotein
(PFC57.6:41.7:0.7%),highfat(PFC12.8:86.7:0.5%)orhigh
carbohydrate(PFC17.8:27.7:54.5%)dietsat500%maintenance
energyrequirementsfor10days
Day 1 Day 5 Day 10 p- value
Mean 25.9c27.0b27.5a<.001
SEM 0.72 0.77 0.77
SEM, Standard error of mean.
Thesuperscriptsaresignificantlydifferentfromoneanother(p<.05).
FIGURE1 Experimentaldesigninvolvingdogsofferedhigh
proteinHP(PFC57:42:1%byenergy),highfatHF(PFC13:86:1%by
energy)orhighcarbohydrateHC(PFC18:28:54%byenergy)dietsfor
10 days
TEST DOG OBSERVOR
RECORDING
CAMERA HC
WATER HF
HP
4
|
ROBERTS ET al.
5and 10). Analysiswasconducted using GenStat18thedition(VSN
International,2016). Results are presented as means and associated
standarderrorofthemean(SEM).
3 | RESULTS
3.1 | Bodyweight
Bodyweightincreased significantly (p<.001) over the 10-day study
(Table2).Atthestartofthestudy,themeanbodyweightofthedogs
was 25.9kg±0.72 SEM which increased to 27.5kg±0.77 SEM on
day 10.
3.2 | Energy intake
Over the course of the study, the dogs reduced (p<.001; Table3)
theirpercentageofenergyconsumedfrom363to162percentofen-
ergyintakeaccordingtothequadraticequation:%ME=419.1(±31.8
SEM)–60(±8.78SEM)xDay+3.43(±0.78SEM)xDay2(Figure2).
3.3 | Feeding dynamics
Throughout the duration of the experiment, the percentage of
dogswhich first approached and first consumed a diet wasdeter-
mined(Figure3).ForHP,thepercentageofdogswhichapproached
thediet firstwas47%(±3.7 SEM)andfirstconsuming it64%(±6.0
SEM)(p<.001),forcarbohydrate24%(±3.0 SEM) first approached
thediet, with 4% (±1.9SEM)first consuming it (p<.001). Thehigh
fatdietdisplayednosignificantdifferencesbetweenthose firstap-
proached 29% (± 3.5 SEM) and first consumption 30% (±5.5 SEM)
FIGURE2 Overthe10-dayperiod,thepercentageofenergy
consumedbythedogs(n=15)reduced(p<.001)according
tothequadraticequation:%ME=419.1(±31.8SE)–60(±8.78
SE)xDay+3.43(±0.78SE)whenofferedhighprotein(P:F:C
57:42:1%byenergy),highfat(P:F:C13:86:1%byenergy)orhigh
carbohydrate(P:F:C18:28:54%byenergy)dietsfor10days
0
50
100
150
200
250
300
350
400
450
0246
810
Day
Mean % of energy consumed
FIGURE3 Percentageofhighprotein(P:F:C57:42:1%byenergy),
highfat(P:F:C13:86:1%)orhighcarbohydrate(P:F:C18:28:54%
byenergy)experimentaldietswhichwereapproachedfirstbyadult
dogs(n=15)for10dayswere47±3.7%,29±3.5%and24±3.0%.
Thoseinvolvinganyleveloffirstconsumptionwere64.3±6.0%,
29.7±5.5%and3.7±1.9%.Forboththehighproteinandhigh
carbohydratediets,thereweresignificantdifferencesbetweenthe
percentagefirstapproachedandfirstconsumed(p<.001)
0
10
20
30
40
50
60
70
80
High protein High fatHigh carb
Percentage
Diet
1st diet approached
1st diet consumed
FIGURE4 Percentageofhighprotein(PFC57:42:1%),highfat
(PFC13:86:1%)orhighcarbohydrate(PFC18:28:54%)experimental
dietscompletelyavoidedbydogs(n=15)over10days.The
percentageofhighcarbohydratedietswhichwerecompletely
avoidedwassignificantlydifferenttothepercentageofhighfat
diets,whichinturnwassignificantlydifferenttothehighproteindiet
(p<.001)
0
10
20
30
40
50
60
70
High proteinHigh fa
tH
igh carbohydrate
Percentage
Diet
b
a
FIGURE5 Meanmacronutrientdailyconsumption(kcalsday−1)
foradultdogs(n=15)offeredhighprotein(P:F:C57:42:1%by
energy),highfat(P:F:C13:86:1%byenergy)orhighcarbohydrate
(P:F:C18:28:54%byenergy)dietsfor10days.Theconsumptionof
carbohydrate(kcals=284.09(±64.12SE)–26.04(±8.33SE)xDay),
andfat(kcals=6,989.38(±1,197.65SE)–607.24(±124.10SE)xDay)
declinedoverthestudy(p < .01 and p<.001respectively).
Consumptionofproteinremainedconstant(kcalsday−1=4,856.21
(±921.20SE)–70.00(±96.95SE)xDay)
0
2000
4000
6000
8000
10,000
12,000
14,000
12345678
91
0
Kcals per day
Day
Fat Protein Carbohydrate
|
5
ROBERTS ET al.
(p=.720).Significantdifferences(p<.001)wereobservedbetween
thepercentageofeachdietcompletelyavoided,with58%(±2.9SE)
of the carbohydrate diet being completely avoided, 20% (±2.3 SE)
of the fat diet and 3% (±1.0 SE) of the protein diet (Figure4). No
changesin this behaviour were observedover the duration of the
study(p=.206).
3.4 | Kcals per day of each macronutrient consumed
Over the course of the study, the daily consumption of carbohydrate
reduced(p<.01; Figure5) from 554 on day 1 to214kcalsday−1 on
day10(kcals=284.09(±64.12 SE)–26.04(±8.33SE)xDay (Table3).
Thekcalsper day of fat consumed also reduced (p<.001; Figure5)
from6,382onday1to917kcalsday−1onday10(kcals=6,989.38
(±1,197.65 SE)–607.24 (±124.10 SE)xDay (Table3). Consumption
ofprotein remained constant over the study ranging from 4786on
day1to4,156kcalsday−1onday10(kcalsday−1=4,856.21(±921.20
SEM)–70.00(±96.95SEM)xDay(Table3).
3.5 | Macronutrient consumption and ratio
Protein intake (as a proportion of total ME) increased (p<.01;
Figure6)from29.4%MEonday 1to 44%ME(ME%=27.77(±3.17
SE)+1.60(±0.36SE)xDay; Table3)byday10.Fatintakedecreased
(p<.001;Figure6)from68%MEonday1to52%ME(ME%=69.95
(±3.14SEM)–1.81(±0.37SE)xDay)(Table3)byday10.Nosignificant
difference in carbohydrate intake was observed (Figure6) over the
study(2.5%MEonday1and4.4%MEbyday10:ME%=2.28(±0.62
SE)+0.21(±0.27SE)xDay;Table3).
The P:F ratio reflects these differences, increasing significantly
(p<.001)fromday1 to10ofthestudy(P:F=0.40(±0.07SE)+0.05
(±0.01SE)xDay;Table3).AP:F:Cratioof34:62:4%wasselectedby
thedogsonday1, which gradually changed to 45:51:4% by day10
(see Figure6, raw data solid lines), driven by this increase (p<.01)
in protein intake (ME day−1) and decrease (p<.001) in fat intake
(MEday−1).
Using the fitted regression line of Protein%=27.8+1.6 Day,
Protein%intakeonDay5was calculatedtobe 35.8%,increasingto
43.8%byday10(Table3,Figure6).
4 | DISCUSSION
Our study shows that when dogs are allowed to self-select from
diets varying in macronutrient composition, they will consume at
least30% of theirenergyfromprotein, thus inagreementwithour
hypothesis.Moreover,whilstmeanproteinintakeoverthecourseof
thestudy was37%,the energy consumptionalteredover thedura-
tionofthestudy,withanincreasefrom29%totalenergyonday1to
TABLE3 Linearandquadraticresponsestoanalysisoftotalenergyconsumed,gramsofmacronutrientsconsumed,specificoverall
macronutrientenergyintakeandratiosindogs(n=15)offeredhighprotein(PFC57.6:41.7:0.7%),highfat(PFC12.8:86.7:0.5%)orhigh
carbohydrate(PFC17.8:27.7:54.5%)dietsat500%maintenanceenergyrequirementsfor10days
Response Model αSE β1SEM β2SE
TotalEnergyConsumed(unit) Linear 373.10*** 40.42 −23.97*** 3.28 – –
Quadratic 419.10*** 31.8 −60.00*** 8.78 3.43*** 0.78
Proteinintake(%ofoverallME) Linear 27.77*** 3.17 1.60** 0.36 – –
Fatintake(%ofoverallME) Linear 69.95*** 3.14 −1.81*** 0.37 – –
Carbohydrateintake(%of
overallME)
Linear 2.28*** 0.62 0.21 0.27 – –
Protein(kcalsday−1) Linear 4,856.21*** 921.20 −70.00 96.95 – –
Fat(kcalsday−1) Linear 6,989.38*** 1,197.65 −607.24*** 124.10 – –
Carbohydrate(kcalsday−1) Linear 284.09*** 64.12 −26.04** 8.33 – –
Protein:Fat Ratio Linear 0.40*** 0.07 0.05*** 0.01 – –
α, Intercept; SE, Standard error; β1,CoefficientofLinearterm;β2,CoefficientofQuadraticterm.
*p<.05.
**p < .01.
***p < .001.
FIGURE6 Meanself-selectedmacronutrienttotalenergy
intake(solidline)andlinearfittedresponse(dottedline)ofadult
dogs(n=15)(27.77±3.17protein%+1.60±0.36xDay),
(69.95±3.14fat%–1.81±0.37xDay),(2.28±0.62carbohydrate
%+0.21±0.27xDay),offeredhighprotein(PFC57:42:1%),
highfat(PFC13:86:1%)orhighcarbohydrate(PFC18:28:54%)
dietsfor10days.Protein(%energyintake)increased(p<.01)
andfatdecreased(p<.001)
0
10
20
30
40
50
60
70
80
12345678
91
0
% energy intake
Day
Protein Fat Carbohydrate
6
|
ROBERTS ET al.
44%by day 10.Thisincrease in proteinintakewas associated with
a decrease in fat consumption over the experiment, with the dogs
consuming68% on day1and 52% by day10.Thus, the protein:fat
ratioincreasedfrom0.45onday1to0.90byday10.Althoughboth
proteinandfat intake altered significantly during the study, carbo-
hydrateconsumption remained steadyat3% throughout thestudy.
These changes in macronutrient selection by the dogs are illus-
tratedin Figures7a,b.Anutrition trianglewasutilised, to represent
a multidimensional assessment of dietary composition information
(Raubenheimer,2011). Collectively,thesechanges in macronutrient
intakeresultedinenergy consumptiondecreasingfrom of363%on
day1to162%onday10.
Based on raw data energy intake, the dogs selected an average
macronutrientP:F:C ratio of38:59:3%(byenergy) during the study.It
mustbe noted,however,thattheP:F:Cratio onday1(35:62:3byen-
ergy)wasdifferenttothat consumedon day10 (45:51:4% byenergy).
Thisdifferencewasdriven by the decreaseinfat energy consumption
(6,382–917kcalsday−1) ratherthan any drop in protein intake(4,786
to4,156kcalsday−1)overthedurationofthestudy.Suchareductionin
energyconsumptionislikelyasaresultofincreasingbodyfatinthedogs
astheexperimentprogressed,withplasmaleptinlevelslikelytoberising
(Ishiokaetal., 2002).Asleptin serves asasignallingpathway between
adiposetissueandthecentralnervoussystem,theconsequenceofthis
maybeareductioninenergyintake(Akers&Denbow,2008).
The initial targeting of fat dense food sources has also been
demonstrated in the predatory beetle Agonum dorsale (Carabidae)
(Raubenheimer,Mayntz,Simpson,&Tøft,2007).Withinthestudy,the
beetleswereassessedinregardtotheirnutrientintakeover10days,
withthe first 2days involvingtargetinga diet rich in fatafterwhich
protein intake increased.Although this study involved determining
macronutrientintakeafteremergencefromhibernation,andthusdif-
feredfromourinvestigation,wedidobservethesamemacronutrient
pattern. Our dogsalso targeted a high fat diet initially, with energy
contributionfromproteinincreasingthereafter,whichmayindicatean
evolutionary past, whereby limited prey availability would predispose
dogstoinitiallyselectfatsources.Althoughthedogsusedinourstudy
were maintained at a healthy body condition score, variations in how
a score relates tobody fat content can occur (Ishioka etal., 2005).
Further studies in dogs investigatingthe association between body
composition,macronutrientselection,totalenergyintakeandfactors
suchas leptin involved in influencing food intake would help better
understandboththemacronutrientandenergyintakeofdogs.
WhencomparingtheaverageP:F:Cratioof38:59:3%selectedby
the dogs in our study to thatdetermined by Hewson-Hughes etal.
(2012)of30:63:7(%byenergy),severalkeyfactorscouldexplainthe
differences,namelythelengthofthe studyperiod, thecalculationof
theP:F:Candtheexperimentalstructure.
In Hewson-Hughes etal. (2012), the experienced phase was
7 days in duration, whereas in the present study, it was 10 days. In
thecurrentstudy,whenmacronutrientselectionwasexaminedacross
thestudy period,itwas apparentthat majordifferencesintheP:F:C
selected occurred during the latter stages of the study, averaging
47:49:4(%byenergy)ondays9and10.Thus,theshortertimeframe
inHewson-Hughesetal.(2012)mayhaveresultedinmissingthisap-
parentkeymacronutrienttransitionalperiod.Itislikelythatproviding
anaveragemacronutrientratioacrossthewholeofthe experimental
periodmayfailto interpret the true nutritional movementthe dogs
madeoverrelatively short testing periods (7–10days). Forexample,
the established average macronutrient ratio observed by Hewson-
Hughes etal. (2012) overa 7-dayperiod (30:63:7% by energy) was
similartotherawdataovertheinitial7daysofourstudy(36:61:3%by
energy).However,onlywhenaveragemacronutrientvaluesareteased
apartforeach dayand examined indetail,do these keytimeframes
becomeobvious.Themacronutrientselection bythedogswithin our
studyvariedsignificantlyoverthe10-dayperiod,withadecreaseinfat
intake(68%vs.52%byenergy)andincreasein protein(29%vs.44%
byenergy)observed.Itremainstobedeterminedifthemacronutrient
selectionbythedogshadstabilisedafter10days,orwhetherprotein
intakewouldcontinuetoincrease.
FIGURE7 (a,b)Macronutrienttotalenergyintakeofindividual
adultdogs(n=15)offeredhighprotein(P:F:C57:42:1%byenergy),
highfat(P:F:C13:86:1%byenergy)orhighcarbohydrate(P:F:C
18:28:54%byenergy)diets.Theyaxisofbothaandbrepresentsfat
intake,withthexaxissignifyingthecarbohydrateintake(byenergy).
Thegraphsdepictpartialcontourplotswithlinesrepresenting
proteinintake(alsocolourcodedwiththelegendshowingtherange
ofcolours).Percentageofmacronutrienttotalenergyintakevalues
foralldogsaresymbolisedbyreddots
|
7
ROBERTS ET al.
Secondlythe experimental structure oftheHewson-Hughesetal.
(2012) study involvedthree distinct phases, of differing duration and
feedingpatterns.Theseconsistedofnaïveself-selection(havingaccess
toall threemealoptions simultaneouslyfor7days),learning (eight,3-
dayperiods,wherebythedogswererestrictedtoaspecificdiet(HC,HF,
HP)for adayof eachperiod)andexperienced (thesameasthenaïve
phase). Thus, the potential exists whereby within the self-selective
phases,thefeedingperiodendedjustasthedogswerestartingtoreg-
ulatetheirmacronutrientintake.Inbetweenthesephases,thelearning
stagemayalsohaveconfuseddogs alreadystartingto targeta macro-
nutrientintake,thatisbyconfiningeachtoa specificdietfor24hours
andrepeatingthe process eight times. Therefore,thecombination of
ashorterstudyperiod, and the inclusion ofa learningphase, limiting
thedogstospecificdietsinHewson-Hughesetal.(2012)mayhaveaf-
fectedthedog’sabilitytotargetthemacronutrientintakeweobserved.
RomsosandFerguson(1983)also addressedmacronutrientselec-
tioninthedomesticdog;however,theirprimaryaimwastounderstand
theregulationofproteinintake.Ina4-weekstudy,twodifferentsetsof
dietswereofferedtothedogs,differingnotjustinproteincontent,but
also in fat and carbohydrate. Whilst the results showed the animals se-
lected30%oftheirmetabolisableenergyfromprotein,limitationsinre-
gardtonutrientmovement,primarilyduetothecarbohydratecontent
varyingfrom 20% to 42% MEwithin the test diets,could potentially
havemaskedthetruemacronutrientratiothedogswishedtoselect.
The self- selected macronutrient profile has also been reported for
the domestic cat (Felis catus)usinganapproachsimilartothatapplied
tothe domestic dog. Hewson-Hughes etal. (2011), establishedthat
macronutrient energyprofile (P:F:C) was 52:36:12 (% byenergy). In
additionthe study also suggests that cats havea carbohydrateceil-
ingof 300kJday−1, which constrains them to deficits in protein and
fat(relativetothe determined intaketarget)when restrictedto high
carbohydrate diets (Hewson-Hughes etal., 2011). Aswith the dog
study,alackofreportingrelatingtomacronutrientintakeoverthedu-
rationof theprojectwas apparent.However,usinganother member
ofthefelidfamilythemink (Mustela vison), it was demonstrated that
withinthefirst24hrofbeingallowedtoself-selectaP:F:C(withcar-
bohydratefixedat15%)froma numberofcomplementaryfoods,the
minkselectedadietconsistingof(P:F)of35:50(%byenergy)(Mayntz
etal., 2009). This ratiowas observed throughout the 11-day study,
with additionally when confined to diets that did not allow the desired
protein:fat ratio to be achieved, the closet possible to that previously
establishedbeingtargeted.
Inthecurrentstudy,itisevidentthatoverthe10-dayexperimen-
talperiod,thedogsmadeadietary“switch,”reducingfatandincreas-
ingproteinintakeonanenergybasis.Tobetterunderstandthedietary
switch,thefeedingdynamicsofthedietswereexplored.
Whentheoverallpercentageofdogswhichfirstapproachedand
firstconsumeda givendiet was determined (Figure3), itwasclear
thattheHFdietdisplayedsimilarvaluesof29%and31%,thusindi-
catingmost ofthe dogswhich approachedthediet first,consumed
someofit.However,whentheHPdietwasexamined,47%ofdogs
approachedit first, with 64% then consuming some ofitfirst.This
differencecanbeexplainedbyresultsfromtheHCdiet,which24%
ofdogsapproachedfirst;however,only4%thenconsumedany.The
majorityofthedogswhichapproachedtheHCdietdecidedtomove
awayandconsumeatleastsomeoftheHPdietinstead.Throughout
the study,the percentage of times that each diet was approached
andconsumed remainedconsistent.Thishighlightedthat theinitial
decisionto consumeaspecific diet atthestart oftheinvestigation
wasmaintainedduringthestudy.Inaddition,thedataalsoshowthat
theHCdietwasmuchmostlikelytoremainuntasted(58%),thanthe
HFand HPdiets(20%and 3% respectively)(Figure4).Collectively,
these feeding dynamics may indicate that therewas an olfactory
differencebetween the diets. As with the percentage of diets first
approached and consumed, the proportion of each diet completely
avoided were similar over the duration of the study. This would in-
dicatethe preferenceofdogstotarget oravoidspecific diets from
day1oftheinvestigation,remainedconsistentoverthesubsequent
9 days.
Indeed whilst we did not attempt to ensure palatability of our
diets were consistent (e.g., with the use of a palatant), the same
key ingredientswere used in all the diets, just in different propor-
tions. Interestingly research conducted by Salaun, Le Paih, Roberti,
Niceron, and Blanchard (2016) foundthat whilst the application of
a palatability enhancerincreased food intake in domestic cats, they
werestill capable of macronutrient regulationwhenofferedpairs of
differingdiets. Moreover,a recent study has also indicated thatthe
domestic cat is able to detect and maintain a macronutrient prefer-
ence,despitechangesinflavour(Hewson-Hughes,Colyer,Simpson,&
Raubenheimer,2016),withcatsstillpreferringadietcontainingapro-
tein:fatratioof70:30(byenergy),evenwhenthe dietwas flavoured
with(apparently)negativeflavours.
Inthecurrentstudy,lambgreentripewasusedastheingredientto
manipulatethedietaryproteincontent.Whilstthedogsmigratedover
thecourseofthestudytoamacronutrientratiowithagreaterenergy
contributionfromprotein,itmaybearguedthatthisindicatedapref-
erenceforgreentripe,ratherthanadesireforproteinper say.Asimilar
argumentcouldalsoberaisedregardingthecarbohydratesourceused
throughout the experiment (maize). Evidently carbohydrates played
aminimal roleinregard toselecteddietary compositioninthe dogs;
however, it is possible that this specific carbohydrate source was dis-
likedmoresothanothersthatarealsotypicallyusedindogfoods(e.g.,
rice orbarley). Future studies could addressthese questions, where
dogsareoffereddietsofsimilarmacronutrientratios,usingdifferent
protein, fat or carbohydrate sources. Similarly, moisture content was
notconsistentbetween diets in thecurrent study, with the HC diet
havinglessmoisturethantheHPdiet.Atpresent,itisunknownifthis
had any impact on the resulting macronutrient profile, but studies
have indicatedin cats that energy intake and food consumption re-
duceasthelevelofwaterinadietincreases(Wei,Fascetti,Villaverde,
Wong,&Ramsey,2011).
In conclusion, the study clearly demonstrated that over a ten- day
experiment, the test dogs selected a diet dominated by consump-
tionofenergy derived primarily from fat and protein,withcarbohy-
drateplaying a minimal role in contributing to overall energyintake.
However, only after the completion of much deeper investigations
8
|
ROBERTS ET al.
into the selective capabilities and mechanisms influencing these di-
etarydecisions,willwetrulyhaveagrasponwhatitisundoubtedlya
fascinatingandhighlycomplexareaofstudy.
ACKNOWLEDGEMENTS
We thank the staff at the Massey University Canine Nutrition Unit
fortheirhelpinconductingthestudy.ThisprojectwasfundedbyK9
Natural,AgResearchCore Funding(A21247) andAgmardt (A15013).
MrMarkRobertsissupportedbyaMasseyUniversityPhDscholarship.
ORCID
M. T. Roberts http://orcid.org/0000-0001-8057-2065
REFERENCES
Akers,R.M.,&Denbow,D.M.(2008). Anatomy and physiology of domestic
animals.1stedn.Ames,Iowa:BlackwellPub.Professional.
AssociationofAmericanFeedControlOfficials.(2015).2015 Official publi-
cation.Champaign,IL:AAFCOPublications.
AssociationofAmericanFeedControlOfficials.(2016).2016 Official publi-
cation.Champaign,IL:AAFCOPublications.
Axelsson, E., Ratnakumar,A., Arendt, M.-L., Maqbool, K., Webster, M. T.,
Perloski, M., … Lindblad-Toh, K. (2013). The genomic signature of dog
domestication reveals adaptation to a starch- rich diet. Nature, 495,
360–364.
Beloshapka, A. N., Duclos, L. M.,Vester Boler, B. M., & Swanson,K. S.
(2012).Effectsofinulinoryeastcell-wallextractonnutrientdigestibil-
ity, fecal fermentative end- product concentrations, and blood metab-
oliteconcentrationsinadultdogsfedrawmeat–baseddiets.American
Journal of Veterinary Research, 73, 1016–1023.
Bosch,G.,Hagen-Plantinga,E. A., & Hendriks, W.H.(2015).Dietarynu-
trient profilesof wild wolves: Insightsfor optimal dog nutrition? The
British Journal of Nutrition, 113,S40–S54.
Frantz,L. A. F., Mullin,V. E., Pionnier-Capitan, M., Lebrasseur, O., Ollivier,
M.,Perri,A.,… Larson,G. (2016).Genomic andarchaeologicalevidence
suggest a dual origin of domestic dogs. Science (New York, N.Y.), 352,
1228–1231.
German,A.J.(2006).Thegrowingproblemofobesityindogsandcats.The
Journal of Nutrition, 186, 1940S–1946S.
Hewson-Hughes, A. K., Colyer, A., Simpson, S. J., & Raubenheimer, D.
(2016). Balancing macronutrient intake in a mammalian carnivore:
Disentangling the influences of flavour and nutrition. Royal Society
Open Science, 3, 160081.
Hewson-Hughes, A. K., Hewson-Hughes,V. L., Colyer, A., Miller, A. T.,
McGrane,S.J.,Hall,S.R.,…Raubenheimer,D.(2012).Geometricanal-
ysisofmacronutrient selection in breeds of the domestic dog, Canis
lupus familiaris. Behavioral Ecology, 24, 293–304.
Hewson-Hughes, A. K., Hewson-Hughes,V. L., Miller, A.T., Hall, S. R.,
Simpson,S.J.,&Raubenheimer,D.(2011).Geometricanalysisofmac-
ronutrient selection in the adult domestic cat, Felis catus. Journal of
Experimental Biology, 214,1039–1051.
vonHoldt,B., Pollinger,J., Lohmueller,K., Han, E., Parker, H., Quignon,
P., … Wayne, R. (2010). ‘Genome-wide SNP and haplotype analy-
sesreveala richhistory underlyingdog domestication’.Nature, 464,
898–902.
Ishioka,K.,Hatai,H.,Komabayashi,K.,Soliman,M.,Shibata,H.,Honjoh,T.,
…Saito,M.(2005).Diurnalvariationsofserumleptinindogs:Effectsof
fastingandre-feeding.The Veterinary Journal, 169,85–90.
Ishioka,K.,Soliman,M.M.,Sagawa,M.,Nakadomo,F.,Shibata,H.,Honjoh,
T.,…Saito,M.(2002).Experimentalandclinicalstudiesonplasmaleptin
inObesedogs.Journal of Veterinary Medical Science, 64,349–353.
Lee,K.P.,Simpson,S.J.,Clissold,F.J.,Brooks,R.,Ballard,J.W.O.,Taylor,
P. W., … Raubenheimer, D. (2008). Lifespan and reproduction in
Drosophila:Newinsightsfromnutritionalgeometry.Proceedings of the
National Academy of Sciences, 105,2498–2503.
Mayntz,D.,Nielsen,V.H.,Sørensen,A.,Toft,S.,Raubenheimer,D.,Hejlesen,
C.,& Simpson, S. J.(2009).Balancing of proteinand lipid intake bya
mammalian carnivore, the mink, Mustela vison. Animal Behaviour, 77,
349–355.
NationalResearchCouncil. (2006). Nutrient requirements of dogs and cats.
Washington,DC:NationalAcademiesPress.
New Zealand CompanionAnimal Council. (2016). Companion Animals in
New Zealand 2016.Auckland,NewZealand:NewZealandCompanion
AnimalCouncil.
RCoreTeam.(2016).R: A language and environment for statistical computing.
https://www.R-project.org/.
Raubenheimer,D. (2011). Toward a quantitative nutritional ecology:The
right-angledmixturetriangle.Ecological Monographs, 81, 407–427.
Raubenheimer,D.,Mayntz,D., Simpson,S.J., &Tøft,S.(2007).Nutrient-
specific compensation followingdiapause in a predator: Implications
forintraguildpredation.Ecology, 88,2598–2608.
Romsos,D.R.,&Ferguson,D.(1983).Regulationofproteinintakeinadult
dogs.Journal of the American Veterinary Medical Association, 182, 41–43.
Salaun,F.,LePaih,L.,Roberti,F.,Niceron,C.,&Blanchard,G.(2016).Impact
of macronutrient composition and palatability in wet diets on food se-
lection in cats. Journal of Animal Physiology and Animal Nutrition, 101,
320–328.
Serpell,J.(1995).The domestic dog: Its evolution, behaviour, and interactions
with people.Cambridge,NewYork:CambridgeUniversityPress.
Simpson,S.J., Sibly, R. M., Lee, K. P.,Behmer,S.T.,& Raubenheimer,D.
(2004).Optimalforagingwhen regulatingintakeofmultiplenutrients.
Animal Behaviour, 68, 1299–1311.
Vilà,C.,Savolainen,P.,Maldonado,J.E.,Amorim,I.R.,Rice,J.E.,Honeycutt,
R.L.,&Wayne,R.K.(1997).Multipleandancientoriginsofthedomes-
ticdog.Science, 276, 1687–1689.
VSNInternational.(2016).GenStat for Windows 18th Edition.GenStat.co.uk.
Wei,A.,Fascetti,A. J.,Villaverde, C.,Wong,R. K.,&Ramsey,J.J. (2011).
Effect of water content in a canned food on voluntary food intake
and bodyweight in cats. American Journal of Veterinary Research, 72,
918–923.
Witzel,A.,Kirk,C.,Henry,G.,Toll,P.,Brejda,J.,&Paetau-Robinson,I.(2014).
‘Useofanovel morphometric methodandbody fatindexsystem for
estimationofbodycompositioninoverweightandobesedogs’.Journal
of The American Veterinary Medical Association, 244, 1279–1284.
SUPPORTING INFORMATION
Additional Supporting Informationmay be found online in the sup-
portinginformationtabforthisarticle.
How to cite this article:RobertsMT,BerminghamEN,Cave
NJ,YoungW,McKenzieCM,ThomasDG.Macronutrient
intakeofdogs,self-selectingdietsvaryingincomposition
offered ad libitum. J Anim Physiol Anim Nutr. 2017;00:1–8.
https://doi.org/10.1111/jpn.12794
