A Body Condition Scoring Chart for Holstein Dairy Cows

A. J. EDMONSON! I. J. LEAN,"> L. D. WEAVER,' T. FARVER,3and G. WEBSTER'

Veterinary Medicine Teaching and Research CenterUniversity of California

Tulare 93274and

Department of Epidemiology and Preventive MedicineUniversity of California

Davis 95616

ABSTRACT

A chart for body condition scoring offreely moving Holstein dairy cows wasdeveloped using an iterative processconsisting of literature review, interviewswith experts, field testing, statisticalanalysis, and comments from chartusers. The chart consists of text anddiagrams that detail changes in con­formation with body condition changefor eight body locations identified asimportant in body condition scoring. Theprecision with which a prototype chartwas used to give location specific condi­tion scores to cows was examined, andthe variability among the assessorsdescribed. This chart gave consistentresults with small variability amongassessors, no significant difference at­tributable to experience of assessors, andno significant cow assessor interaction.

Minor modifications were made to thechart, which was then used to assesslocation specific and overall body condi­tion scores. Assessors scored cows in theeight body locations and rescored thecows in a different order to assign anoverall score. The chart produced con­sistent scores over a wide range of bodyconditions with small variance amongassessors. The overall score was mostclosely related to the condition scores ofthe pelvic and tailhead areas of the cow.

Received October 7, 1987.Accepted August 22, 1988.'Veterinary Medicine Teaching and Research

Center, Tulare, CA 93274.>Reprint requests.3Department of Epidemiology and Preventive

Medicine, Davis, CA 95616.

1989 J Dairy Sci 72:68-78

Both location specific scores within cowsand the overall body score for a cow werestrongly correlated, demonstrating thatthe chart was internally consistent. Thechart is an effective field tool for bodycondition scoring Holstein cows.

INTRODUCTION

Body condition scoring (BCS) is a sub­jective method of assessing the amount ofmetabolizable energy stored in fat and muscle(body reserves) on a live animal. Jefferies (12)initially developed a BCS system for ewes. Thesystem involved palpating the backbone andlumbar processes, feeling for the sharpness andcovering of the bones. Ewes were scored on ascale from 0 to 5, where 0 was on the point ofdeath and 5 was very fat. His technique wasadapted for scoring beef cattle by Lowmanet al. (16) using a 0 to 5 scale, with intermediatevalues for animals whose condition falls be­tween these numbers, functioning as an 11­point scale. This system also used palpation ofthe backbone and lumbar processes and in­cluded palpation of the tailhead region. Sub­sequently Mulvany (19) modified the systemfor use in dairy cattle but introduced adjust­ment factors if the scores in the tailhead andloin areas differed. In Australia, an 8-gradesystem for scoring dairy cows was developed byEarle (6) and a similar lO-point system de­veloped in New Zealand (10). Both the NewZealand and Australian scoring systems usedphotographs of individual cattle to. definecondition scores and have accompamed thephotographs with a limited text description ofthe areas to be scored.

Body condition scoring of dairy cows inthe US is generally performed according to a1 to 5 scale (26). This method, like thoseused in the United Kingdom, involves palpating

68

BODY CONDITION SCORING DAIRY COWS 69

COWS to assess the amount of tissue under theskin. These systems therefore require animals tobe under restraint while scoring is performed.In many production systems, especially thosewith large herd sizes, the opportunity for thistype of evaluation is limited. The Australianand New Zealand body scoring techniques,however, use only visual inspection, a preferredmethod when large numbers of freely movingcattle are involved. Body condition scoringperformed in this way is a rapid and easymethod of assessing the condition of cattlewithout the use of scales and is relativelyunaffected by body size (10). To the authors'knowledge neither the New Zealand, Aus­tralian, nor American systems have been fullyvalidated. Cattle condition scores have beenrelated to milk yield and reproductive per­formance (2, 4, 5, 9, 10, 11, 17, 22, 23).Advice has been given regarding condition forstage of production, management decisions (16,19, 21, 24), and in the evaluation of dairyproduction and nutrition (15).

Body condition scoring dairy cows is cur­rently performed using a variety of scalesand systems, and difficulty exists in inter­preting the literature because of variabili­ty in the way authors apply scoring meth­ods. The objectives of this study were todevelop a condition scoring chart for freelymoving Holstein dairy cows and to evaluatethe precision that this chart gave when usedby different assessors under practical fieldconditions.

MATERIALS AND METHODS

Chart

A chart for condition scoring Holstein cowswas prepared after reviewing and applying theprocedures currently used for condition scoringin the United Kingdom, Australia, New Zealand,and US (6, 10, 12, 13, 16, 19, 25, 26). Theoriginal template was subsequently modified byinterviewing three people experienced in bothAustralian and US dairy cattle conditionscoring methods and repeating this process untilagreement on all the areas of the chart wasachieved. Diagrams were added to the text toconvey the gradation of body changes andreduce the dependance on written descriptions.

The chart was prepared so that each area ofthe cow that was considered important in

assigning an overall body condition score couldbe examined individually for changes along a 1to 5 scale, using .25-unit increments, func­tioningas a 17-point scale. A score of 1 in­dicated an emaciated condition, and a score of5 indicated an obese condition. Initially, duringtrial 1, the chart examined nine body areas withlocation B8 being divided into B8 (spinous andtransverse processes of the coccygeal vertebrae)and B9 (ischiorectal fossa). This was sub­sequently modified to the eight body locationsoutlined below and shown in Figure 1. Themodifications were a result of user commentsand the variance found for the tailhead regionin trial 1. The eight areas of the cow's bodywere examined and criteria within each areawere used to indicate the body condition. Theeight locations (B1 to B8) examined were inthree major regions:

1) Loin - B1 spinous processes, (thevertical prominances of the lumbar vertebrae);B2 depression between the spinous and trans­verse processes; B3 transverse processes (thetransverse prominances of the lumbar ver­tebrae); B4 overhanging shelf formed by thetransverse processes above the flank.

2) Pelvis - B5 tuber coxae (hooks) andtuber ischii (pin bones) bony prominances; B6depression between the hook and pin bones; B7depression between the hooks.

3) Tail head - B8 spinous and transverseprocesses of the coccygeal vertebrae andischiorectal fossa (depression beneath the tail).

Precision: Trial 1

The chart precision was evaluated by nineassessors, each scoring the same 59 cows, andrescoring 16 of these animals as a conveniencesample.

Assessors. The .nine assessors were con­sidered in three groups: 1) three experts in­volved in the development of the scale; 2) threenovices with some experience in conditionscoring cattle; and 3) three beginners who hadnever condition scored cattle or seen the chartprior to this trial, but who were familiar withcattle.

Cattle. Seventy-two cattle were selectedfrom one dairy, which had 2000 lactatingHolstein cows available. Animals were chosenby a stratified random procedure, the stratabeing parity (first, second, or subsequent

Journal of Dairy Science Vol. 72, No.1, 1989

Spinous processes (SP) ISpilous to Transvers~ Transverse Ioverhanging..sheitJ TUb~ cox~e (hoOkS~Betw;n pins and1Between the hooks r Tailhead to pins(anatomy varies) processes processes (care • rumen fill) & lIber ischii (pins) hooks (anatomy varies)

- --- -- ---- -

(3

ttl

~oZrJJoZttl..,>\'

,(

first evidenceoffat

bones prominent'U' shaped cavityformed under tail

-V

.;. bones rounded with: fat and slight fat-tilled'+ dep~er tail

~ ,(-: bones buried in fat,: cavity filled with fat+forming tissue folds---- .

rounded .

T

or·

sloping

rounded

slightdepression

depression

~: flat:~

flat -:

T

·

T.· . .- - -: thin flesh covering: definite depression:

.~..· .:.~.:.· .· .moderate -:bones smooth, cavity

depression : under tail shallowT & fatty tissue lined

~: ,..--.....,.t slight depression t i r

T

smooth

covered

prominent

buried in fat

=- rounded with fat=-· .· .· .- -·

·

T

·

=-n.:.· .· .· .':" .: -:-· .· .·

~\

slight shelf

)----1-

\----1-

moderate shelf

(----1-bulging

prominent shelf

cr

edge barelydiscernable

buried in fat

1/2 length ofprocess visible

... ...: appears smooth, :-?P's just discemable+

: distinct ridge, no :: invidual processes :

-;- di~Cmabie -;-. .. .: """ .smooth, rounded

edgenearly flat

-r:

~smooth slope

··· -· ;· .i- ..!····-;-L~: rounded (convex): .

T·····

lbilled in fal

-L

sharp, prominentridge

smooth ridge, theSP's not evident

flat, no processesdiscernable

individual processes : obvious depression:avident Ii'\":" IL":" _~ : --.0::::= : between 1/2 to 1/3 :

: of processes visible:

:J=r~r':" -;- 1/3 - 1/4 visible -;-

1-bl ~smooth concave; <1/4 visible •

• curve

individual processes : deep depression: very prominent, : definite shelf, : extremely sharp, : severe depression,: severely depressed: bones very prominentdistinct, givinga' ilL' >1/2 length visible • gaunt, tucked • no tissue cover • deVoid of flesh • J".. f\ ..A- • with deep 'V' shapec

saw-tooth appearance + -0:== + + I +('J\/>+ +\ ~ '-'" 1 + cavity under tail

-1- : :~:~~: : : :V• <II '" '" '" '" '"

'" '" '" '" '" '" '"'T" "r "T" "T" "T" ... .,..· .· .· .· .- -

~: :· .· .

very sunken t~ t

5.00

1.50

1.00

4.50

4.00

4.75

4.25

3.75

1.25

3.25

2.75

2.00

2.25

1.75

3.50

3.00

2.50

SCORE

FRAME NOT ASVISIBLE ASCOVERING

FRAME OBVIOUS

SEVEREOVERCONOlTIONlNG

FRAME & COVERINGWELL BALANCED

SEVEREUNDERCONOITIONING

(emaciated)

~

....'"00

'"

....g....~a.o~.

'<V'Jn,,'::snro<:~'-31"Z9

Figure 1. Body condition scoring chart for Holstein cows.

BODY CONDITION SCORING DAIRY COWS 71

lactation), and within each parity, days in milk(DIM) (0 to 120, 121 to 240, 241 to dry, anddry). The final selections were made using atable of random numbers and the last two digitson the cow's ear tag. The 16 animals for re­scoring were a random convenience samplefrom several pens.

Design. The nine assessors were given thescoring chart 1 d prior to the trial, and thechan design was discussed before arriving at thedairy. Photographs of dairy cattle were used todiscuss the areas included on the scoring chartand the ranges of condition in each area.

Selected cows moved freely with other cattlein dry lot or freestall housing during scoring.The assessors moved from cow to cow as agroup, viewing each animal and assigning ascore to each body location. The conditionscore assigned to a body location was notdiscussed between assessors. After all availablecattle had been scored, 16 cows were rescoredusing the same procedure, without reference tothe previously assigned scores.

Precision: Trial 2

The usefulness of the chart was examined bysix assessors; five of whom had used the chartin trial 1, and one who had not scored cowswith the chart previously but was experiencedwith dairy cattle. Each assessor scored the same25 cows. The 25 cows were a conveniencesample selected from the corrals of a dairy with2000 Holstein cows available. The cows wereselected to represent a wide range of bodyconditions.

Each assessor scored the 25 cows in the eightbody locations by marking the chart where thecriteria on the chart matched the appearance ofthe cow. The 2S cows were then rescored in adifferent order to minimize the correlationbetween the two assessments. When rescored,the cow was given an overall body conditionscore without reference to the location specificscores previously assigned. Cows moved freelywith other cattle in dry lot or freestall housingduring scoring. Consultation among assessorsdid not occur.

Analysis

Trial 1. Preliminary data description wasperformed using a statistical graphics program(Statgraphics, 1985 STSC, Inc., Rockville MD).

The body condition scale was consideredcontinuous (even though the scores werecorrected to the nearest .25 point) and normali­ty assumptions made. This allowed the scoresfrom each body location to be examined byANOVA using statistical software (SAS In­stitute, Inc., Cary, NC). Initially, the effects ofassessor, cow, expert category, parity, and DIMwere examined using a partially hierarchal(nested) analysis of covariance. A final evalua­tion, using only the variables found to bestatistically significant in the preliminaryanalysis, was made using a random effectsANOVA model.

The final model is given below:

Yjln =u + aj + cr + (ac)jl + ejln

where u is a constant, aj (assessor), CI (cow),and (ac)jl (interaction) are independent normalrandom variables with expectations zero and

. . 2 2 d 2respective varIances 0a, 0c, an 0a6 ejln areindependent N(o, 0

2), and independent of ai,

cr, and (ac)jl j = 1,...9; I = 1,...59, n = 2 for16 cows that were rescored, otherwise n = I.

Trial 2. Using statistical software (BMDPStatistical Software, 1985, Los Angeles, CA),the data were examined by cluster analysis todetermine which body locations were scoredsimilarly, and the correlations among each bodylocation with the overall body condition scoreassigned. By considering the body conditionscale as continuous (even though the scoreswere corrected to the nearest .25 point) andmaking normality assumptions, the data werefurther examined by ANOVA. A randomeffects model was used to estimate the mag­nitude of the factors determining the conditionscore. In this trial, there was no estimate of theerror of the variance, because no replication inthe scoring of each body location occurred. Todetermine if the interaction term amongassessor and cow could be used as a proxy forthe error term, the interaction term was com­pared with both the estimate of the error termand the interaction term obtained in theprevious trial.

RESULTS

Trial 1

Of the 72 cows selected in the sample, S9were available for scoring. Box plots suggested

Journal of Dairy Science Vol. 72, No.1, 1989

72 EDMONSON ET AL.

Figure 3. Notched box and whisker plots illus­trating the distribution of the condition scores(mean from the 9 body locations) given by the dif­ferent categories of assessors.

not explained by the variation among assessorsor among cows is given as error (Table 3). Forall locations, variability among cows had amuch greater impact on the variance of thelocation specific mean scores than the vari­ability among assessors (Table 4). In locationsB8 and B9, the variability among assessors waslarger than in other body areas. Thus, the

5

w qa:0u'"z0

I- 3etU0.-J

ZetLJJE

2+

1

that higher parity cows may have significantlyhigher body condition scores than lower paritycows, but the number of cows scored in thefifth, sixth and seventh parity groups was verysmall (Figure 2). Box plots display batches ofdata, the middle line shows the median, the topand bottom lines of each box show the upperand lower quartiles, the vertical lines showthe extremes, and the plus symbols ind icateoutliers. Notches indicate the 95% confidenceintervals of the medians; overlap of the notchessuggests no significant difference between thedata sets (18). Analysis of variance indicatedno significant source of variability attributableto parity above the degree of variability amongcows within the parity groups. Days in milk wasnot a significant covariate. No significant sourceof variability could be attributed to expertisecategory above the degree of variability amongassessors within these categories (Figures 3 and4). These results were consistent for the datafrom all nine body locations. The analyticaldesign, therefore, reduced to one involving thefac;tors "cow" and "assessor". both beingconsidered random. The two-way analysis ofvariance showed no significant interactionbetween "assessor" and "cow" in all nine bodylocations where scoring rook place in thereplicated trial.

"Assessor" and "cow" main effects weresignificant (P<.OOOl) in all nine body locations(Tables 1 and 2). Residual variation in the data

5 5

•..,4

~ ~*t~w 4

a: a:0 0u uV> V>

Z z:0 0

S3 ;:: 3e:t

U U0 0-' -'z: z:e:t e:tW W:E: 2 :E: 2 +

+

1 1A B D E F G H

ASSESSOR

Figure 2. Notched box and whisker plots illus­trating the effect of parity on the distribution of thecondition scores (mean score from the 9 body loca­tions).

Figure 4. Notched box and whisker plots illustrat­ing the di'stribution of the condition scores (meanfor the 9 body locations) given by each of the 9assessors.

Journal of Dairy Science Vol. 72, No.1, 1989

BODY CONDITION SCORING DAIRY COWS 73

Oll'lOOOO"'lI'lo-o-",'0\0 lO'1"\OtI'\VN'\lnmmMmmmmmm

O\.NN"'lvr---.....-4'f""'4tn.-l

~~~~~~~~~

lroI"'--I""""1N.q-\CCO'1"V'IoNNmNNNm~N

u-) 0\ '1""'1- ci r-: 0\ M n\ '1"-mmmNmNNO'l""'1mmmmmmmmm

1""""1\t"i\OOOl"'--trll'-Oll"lo'mor...:OONo'.q:r....:.-l'l""'1I""""1N NI""""1",,"m

\t"i\O'C·OC,oOON ....ooooooom",vll'lvIn .... NOONMlJ'''loNO\t--N'o:trt"liO\NtI'\I""""1N'o:tV'lo'1"mvvmNM",,;,.,.;mmmmMMM

'000 f'O 0 o-o-lI'l '0 VoOm-oooO~-o~ol'oooor....oor....r....V\\O

co

~ ";;"tl ojo u~.2

"1:l~oj

5-.- oj.... w'" 0w ....

'""...0u'"C.g

:0c0u>,

"1:l0

..0

"'5.5>,.:::.0oj

'Coj

>.....0

'""~::s~

"'5.....

~0... >,

3 :.5E 5::s

" CJluC

.~oj

>'".;:

.....0

~oj

5'a""'5

"1:lCoj

Coj

"5"..c........0

'""...oj

'E'""-;i::".....5

"1:lcoj

c:'0c.~

W...:l~

<...

.13051, .27826

.16053, .34286

.16659, .35601

.15372, .32896

.15534, .32977

.13646, .29160

.14287, .30496

.19389, .41273

.22903, .48691

95% Confidenceinterval

95% Confidencein terval

.00288, .03126

.00224, .02948

.00157, .03874

.00566, .05623

.00127, .01914

.00374, .03922

.00333, .03611

.01497, .17534

.01349, .12345

95% Confidenceinterval

.05789, .07250

.07638, .09565

.08121, .10170

.07895, .09887

.05572, .06978

.06630, .08303

.06643, .08320

.07976, .09989

.11313, .14167

Cowvariance

.18383

.22628

.23487

.21685

.21842

.19239

.20133

.27293

.32295

Assessorvariance

.00682

.00555

.00445

.01311

.00325

.00878

.00789

.03353

.03060

Error

.06458

.08534

.09074

.08821

.06226

.07408

.07423

.08912

.12640

8.96.14.7

12.14.99.99.0

29.219.2

Bodylocation F-Statistic'

I All significant (P<.OOOl).

Bl 33.5B2 31.3B3 30.6B4 29.1B5 41.1B6 30.6B7 32.0B8 36.0B9 30.2

TABLE 3. Point and interval estimates of the error(Trial 1).

I All significant (P<.OOOl).

BlB2B3B4B5B6B7B8B9

TABLE 2. Point and interval estimates of varianceof the cow and F-statistics for testing Ho : varianceof the cow =0 (Trial 1).

B1B2B3B4B5B6B7B8B9

Bodylocation F-Statistic'

Bodylocation

TABLE 1. Point and interval estimates of varianceof the assessor and F-statistics for testing Ho : vari­ance of the assessor = 0 (Trial 1).

Journal of Dairy Science V?1. 72, No.1, 1989

74 EDMONSON ET AL.

among cow main effects to be significant(P< .000 1) in all eight body locations (Table 5).The among assessor variability was significant(P<.OOOI) in all body locations except B6 and

largest estimates and confidence intervals of thevariance of the estimated mean were obtainedfrom these areas. The estimate of the meanfrom all nine body locations was similar;however, certain locations tended to give highscores (B2 and Bs) and others gave low scores(B8, B9 and B4) (Table 4).

Trial 2

The condition scoring chart is shown inFigure 1. The eight locations on the cows bodyare defined, and criteria within each area aredescribed. The scale for the body condition iscontinuous, but for convenience, the chart ismarked in .25 increments.

Cluster analysis indicated that body condi­tion scores given for the first four body loca­tions (BI to B4) tended to cluster, as did scoresgiven to the last four body locations (B5 to B8)(Figure Sa). The overall body condition scoregiven by the assessors fell within the group ofpelvic and tailhead location scores (B5 to B8)(Figure 5b). Overall body condition score wasassociated most closely with the scores givenfor the bony prominences of the hook and pinbones (Bs) (Figure 5b). Correlation coefficientswere greater than .92 among alI the bodylocations, including the overall score.

The ratios of the interaction term comparedto the estimate of the error or the interactionobtained from the first trial were all close to l.The interaction term could, therefore, be usedas a proxy for the error term in the analysisof variance. The analysis of variance showed the

TABLE 5. Point and interval estimates of varianceof the cow and F-statistics for testing Ho : varianceof the cow'" 0 (Trial 2).

B1

B2

B3

B4

B5

B6

B7

B8

B1

B2

B3

B4

B5

Ov

a

b

Bodylocation F-Statistic'

Cow 95% Confidencevariance interval

B6

B7

1 All significant (P<.0001).

Bl82B3B485B6B7BllOverall

27.0238.3933.0039.0946.8835.9342.9847.4763.42

.39567

.46766

.51277

.48446

.53495

.45205

.47773

.58397

.48812

.23727- .78894

.28185- .92396

.30844-1.01665

.29204- .95679

.32309-1.05287

.27222- .89443

.28828- .94172

.35274-1.14910

.29555- .95636

B8Figure 5. Tree diagrams summarizing the cluster

analysis of body condition socres. The most closelyrelated body locations are connected in successivesteps from left to right (1). 1. Relationship amongthe condition scores given for each location by allthe assessors. b. Relationship among the conditionscores given for each location and the overall scoresby all the assessors.

Journal of Dairy Science Vol. 72, NO.1, 1989

BODY CONDITION SCORING DAIRY COWS 75

TABLE 6. Point and interval estimates of varianceof the assessor and F-statistics for testing Ho : varianceof the assessor = 0 (Trial 2).

Body Assessor 95% Confidencelocation F-Statistic variance interval

81 7.5P .02385 .00837-.21381B2 5.26a .01279 .00425-.14460B3 7.03a .02318 .00807-.2146184 4.94a .01203 .00395-.14523B5 8.46a .02088 .00742-.1789686 2.08 .00336 .00077-.0083387 .93 .00019 I

88 8.19a .01268 .00768-.18802Overall I1.57 a .Q1983 .00724-.15222

ap <.OOO1.

I Too small to calculate.

TABLE 7. Point and interval estimates of the errors.

Body 95% Confidencelocation Error interval

Bl .09124 .07193-.11956B2 .07505 .05917-.0983583 .09614 .07579-.1259884 .07631 .06016-.10000B5 .06995 .05515-.0916686 .07764 .06121-.10174B7 .06828 .05383-.0894888 .07540 .05944-.09881Overall .04692 .03699-.05868

B7, where no significant variability was foundamong the assessors (Table 6). The residualvariation not explained by assessor or cow isgiven in Table 7. For all eight locations, vari­ability among cows had a much greater impacton the variance of the location specific meanscores than the variability among assessors(Table 8).

In assigning the overall body conditionscore, the assessor and cow variances were alsosignificant. However, the variance of the overallscore fell within the confidence intervals of thelocation specific variances (Tables 5 and 6).

DISCUSSION

Body condition scoring systems have beendifficult to interpret due to inadequate detail.Some have been based on photos with minimalinterpretation and others on lengthy writtendescriptions. This may limit the repeatabilityof the system to assessors working closelytogether. Scoring systems reflect body nutrientreserves of cattle (10, 14, 27). Wright andRussel (27) examined 73 dairy and beef cowsand showed that the condition score wasrelated to the proportion of fat in the liveweight and to body water, protein, ash, andbody energy. They also found that breedsdiffer in the partitioning of fat among thevarious deposits, which resulted in these breedsdiffering in the proportion of total bodyfat at the same body condition score. In all butthe thinnest cows, intermuscular and intramu­scular fat constituted the major deposit.

TABLE 8. Point and interval estimates of the mean and the estimate of its variance (Trial 2).

Summary of the sources of variability in thebody condition scores

% Assessor % Cow % Other sourcesvariability variability of error

Body Estimated 95% Confidence Estimatedlocation mean interval total

Bl 3.40833 3.28-3.53 .020482 3.33667 3.05-3.62 .0213B3 3.234-33 2.92-~.54 .0250B4 3.21000 2.92-3.50 .0219B5 3.23000 2.92-3.54 .025386 3.26667 2.99-3.54 .0192B7 3.22167 2.95-3.50 .0196B8 3.21833 2.89-3.54 .0275Overall 3.29500 3.00-3.59 .0232

19.510.015.4

9.113.7

2.9.15

13.214.3

77.687.882.088.484.594.397.585.084.2

2.92.22.62.31.72.72.31.81.3

Journal of Dairy Science Vol. 72, No.1, 1989

76 EDMONSON ET AL.

Several workers (3, 9, 16) found scorescorrelated with subcutaneous fat depth mea­sured by ultrasound, and relationships withbody weight and heart girth measurements havebeen found (3). Johnson (13) showed that thechange in condition score followed the patternof live weight change in dairy cattle, and others(8, 10) have estimated the live weight changeassociated with changing condition score.Surface profiles of cows around the loins andrump have been correlated with the conditionscore (10). Wildman (26) similarly found bodyweight and frame measurements correlated withbody condition score. Condition scores havealso been related to biological measures (25),milk yield (10, 11, 23), and reproductiveperformance (2, 4, 5, 10, 22).

The chart developed in this study resultedfrom an iterative process of literature review,interviews with experts, field testing, statisticalanalysis, and comments from chart users.Scoring with a chart removes the influence ofthe individual cow by using diagrams, ratherthan photographs of single cows, to depictchange in conformation with weight gain orloss. Further, this format minimizes the dif­ficulty of interpreting written description andfocuses the assessors attention on each bodylocation before assigning an overall score.

The analyses of variance indicate the relativemagnitude of the components that affect theassigned condition score. The variance alsoindicates the precision with which a score maybe assessed by several observers (7). Our an­alyses found that significant variation generallyexists between assessors when each bodylocation was scored. Trial 1 found no sig­nificant difference between assessor expertisecategory, suggesting that the chart enabledbeginners to condition score cattle with asimilar precision to experienced assessors. Thevariability of the score given by an assessor onthe first and second observation in trial 1 wasvery small. Other sources of variability inaddition to assessor (i.e., expert category) andcow (i.e., replication, days in milk, and parity)represented a very small portion of the totalvariability (coefficient of variance .27 to .41 %).

Trial 1 demonstrated that scores given bythe assessors were almost parallel across thecows scored (no significant cow-assessor in­teraction). This indicates that all assessorsincreased and decreased the score assigned from

Journal of Dairy Science Vol. 72, No.1, 1989

cow to cow by a similar amount, even if theydid not assign the same score (similar accuracyon all cows). Thus there was consensus amongassessors on when a body condition score washigh and when it was low. The lack of in­teraction indicates that the chart enabledassessors to score cows without bias in theirinterpretation of the subjective criteria ondifferent animals. Both Evans and Nicoll (7, 20)found this interaction term to be significantand in some cases to be larger than the variancebetween assessors.

The overall mean scores obtained in trial 1reflect the average condition score of cows onthe dairy, since the stratified sample of cowsscored was representative of all the cows on thedairy. The between cow variability is con­sistently larger than the variability betweenassessors, indicating that the scoring procedureworks.

Trial 2 was designed to evaluate a widerrange of body conditions than assessed pre­viously. The cow variances calculated wereconsequently greater than those in the originaltrial and cows were more varied in conditionthan previous studies cited (7, 20). This mayhave been due to cow selection procedures usedin each study. Nicoll (20) used cattle going onto, and finishing, a feeding trial. Consequently,it is probable that these cattle were in similarbody condition. Evan's (7) cattle were scored insmaller groups (range 9 to 24, mean 14) thanthe number of cows in this study, and thesmaller numbers may have resulted in lessvariability.

Significant assessor variation was found(variance .01 to .02) except in two bodylocations. The depression between the hookand pin bones and the depression between thehooks (B6 and B7) had no significant varianceamong assessors. Despite the greater range ofconditions scored, assessor variability wassimil.a~ to or less than in trial 1, indicating thatthe 'precision with which body condition isassessed has improved. Compared with resultsof Evans (7) and Nicoll (20), the assessorvariability was similar or slightly less, despite amore varied range of cattle. This suggests thateither the new chart was easier to interpretbecause of the improvements, or assessors havelearned to score cows more consistently withthe chart, or both these factors were involved.

Cluster analysis is used to group individuals

BODY CONDITION SCORING DAIRY COWS 77

with similar attributes (1). Location specificscores from trial 2 fell into two main clusters;those related to the loin and spinous region (B1to B4), and those from the pelvic and tail­head regions (B5 to B8). Thus change in condi­tion is related most closely within similaranatomical regions of the cow, suggesting thatthe scale developed is effective. Additionally,the scores from all body locations are highlycorrelated. Both these findings indicate that thechart is internally consistent, since scoringcriteria are correctly located on the scale fordifferent body locations of the cow.

The overall body condition scores assignedin trial 2 fell within the pelvic-tailhead clusterof scores. This may have been influenced by atendency to score cows from behind. Theoverall score most closely reflects the scoresgiven to the bony prominence of the hook andthe pin bones (B5), the depression between thehook and pin bones (B6), and the depressionbetween the hooks (B7) (Figure 5b). Theassessor variance on these three body locationsis small (.0209 to .0002), being nonsignificantin 86 and 87. This suggests that these bodylocations are reliable areas to use when scoringfreely moving cows. The assessor variancefound on the overall score is within the con­fidence intervals of the variance for the in­dividual locations, indicating that a single scoremay be given to cows with confidence.

Previous authors who scored with palpationtechniques have given greater emphasis to thetail head and loin regions. Our findings suggestthat, in the range of conditions scored (overallscores 1.5 to 5.0, mean 3.3), the depression be­tween the hooks and the depression betweenthe hook and pin bones may be most im­portant, since no assessor variance was found inthese areas. The emphasis previously placed onthe loin and tailhead may have resulted fromthe palpation techniques used in other systems(16,21,26,27), since palpation of the tailheadmay be easily performed in tie stall facilities.Further, differences may exist in the relativemerit of anatomical locations in reflectingtissue deposits at different body conditions.Differences in genotype among this studypopulation and other populations previouslystudied may have resulted in a different empha­sis since genotype influences the distribu­tion and mobilization of body tissue (22).

In recommending appropriate body condi­tion scores at particular stages of the pro­duction cycle, consideration must be given tothe genotypic differences in fatness at anycondition score (27). The study populationconsisted of a sample of Holstein cows fromone dairy, and the target population for thischart is all Holstein dairy cows. The applicationof this chart to other breeds may be biased dueto genetic differences in their distribution offat.

This BeS chart provided standard pro­cedures of known precision to condition scorefreely moving Holstein dairy cows. This chartcan be used repeatedly, as described, to providea condition scoring training tool, ensuring thateach area of a cow is observed and evaluatedbefore a condition score is assigned. Once theassessors are confident with the procedureinvolved in assigning a score, only periodicreference to the chart should be required tomaintain consistency in assessing the overallcondition score between observers. This studyhas shown that the effects of parity, DIM, andexp.ert category do not significantly affect theanalysis of scores assigned by an observer usingthis chart.

Our chart indicates the score from a singlearea is a good indicator of the overall score ofthe cow. If the assessor cannot view all bodyareas, a condition score can still reliably begiven to the cow, because of the small variancein mean body location scores found in trial 1and the close relationship of the overall score to

the body location scores found in trail 2. It wasdemonstrated in trial 2 that an overall bodycondition score can be assigned with littlevariance between assessors who used this chart.

Trials have demonstrated that while bodycondition scoring is a subjective technique, itcan be related to objective measures of bio­logical change (2, 4,5,9,10,11,22,23,26). Inthis paper, use of a body condition chart as afield tool is demonstrated. The chart proved tobe a method of ensuring precise responses froma group of assessors scoring freely movingHolstein cows. An overall score may be givenwith confidence to cows by using the chart.The process of chart development described inthis study produced a practical tool to reducesubjectivity in body condition scoring. Theauthors caution that, as with other body

Journal of Dairy Science Vol. 72, No.1, 1989

78 EDMONSON ET AL.

condition score systems, biological relationshipsfound with body condition scores evaluated bythis method should not necessarily be ex­trapolated to scores performed by anothermethod.

ACKNOWLEDGMENTS

The authors acknowledge W. Goodger, E.Bouchard, S. R. Ostrowski, Al Harper, H. F.Troutt, J. Wiseman, and C. Daley for theircontributions and the dairy that provided cattlefor scoring.

REFERENCES

1 Afifi, A. A., and V. Clark. 1984. Cluster analysis.in Computer-aided multivariate analysis. Publ.Lifetime Learning, Belmont, CA.

2 Baishya, N., S. U. Morant, G. S. Pope, and J. D.Leaver. 1982. Rearing of dairy cattle. 8. Rela­tionships of dietary energy intake, changes in liveweight, body condition and fertility. Anim. Prod.34:63.

3 Boisclair, Y., D. G. Grieve, J. B. Stone, O. B. Allen,and G. K. Macleod. 1986. Effects of prepartumenergy, body condition, and sodium bicarbonateon production of cows in early lactation. J. DairySci. 69:2636.

4 Ducker, M. J., and S. V. Morant. 1984. Observa­tions on the relationships between the nutrition,milk yield, live weight and reproductive perform­ance of dairy cows. Anim. Prod. 38:9.

5 Ducker, M. J., S. V. Morant, W. J. Fisher, and R.A Haggett. 1985. Nutrition and reproductiveperformance of dairy cattle. 2. Prediction ofreproductive performance in first lactation dairyheifers subjected to controlled nutritional regi­mens. Anim. Prod. 41:13.

6 Earle, D. F. 1976. A guide to scoring dairy cowcondition. Aust. Dep. Agric. J. Victoria 74:228.

7 Evans, D. G. 1978. The interpretation and analysisof subjective body condition scores. Anim. Prod.26:119.

8 Frood, M. J., and D. Croxton. 1978. The use ofcondition scoring in dairy cows and its relationshipwith milk yield and live weight. Anim. Prod.27:285.

9 Garnsworthy, P. c., and J. H. Topps. 1982. Theeffect of body condition of dairy cows at calvingon their food intake and performance when givencomplete diets. Anim. Prod. 35:113.

10 Grainger, C., and A A. McGowan. 1982. Thesignificance of pre-calving nutrition of the dairycow. Pages 134-171 in Proc. Con£. Dairy Prod.Pasture, Ruakura Anim. Res. Stn., Hamilton, NZ.Occas. Pub!. No.8, NZ Soc. Anim. Prod.

11 Grainger, C., G. D. Wilhelms, and A A. McGowan.1982. Effects of body condition at calving andlevel of feeding in early lactation on milk pro-

Journal of Dairy Science Vol. 72, No. I, 1989

duction of dairy cows. Aust. J. Exp. Agric. Anim.Husb. 22:9.

12 Jefferies, B. C. 1961. Body condition scoring andits use in management. Tasmanian J. Agric., Min.Agric. 32: 19.

13 Johnson, C. L. 1984. The effect of feeding in earlylactation on feed intake, yields of milk, fat andprotein and on live-weight change over one lac­tation in dairy cows. J. Agric. Sci., Carob. 103 :629.

14 Lamont, D. I., D. R. Neilson, G. C. Emmans, J.Fraser, and J.H.D. Prescott. 1984. The relationshipbetween live animal measurements and the physicaland chemical composition of dairy cows. Br. Soc.Anim. Prod. 38(3):529.

15 Lean, I. J. 1985. Assessing sub-optimal nutrition.Pages 191-266 in Dairy cattle production. Proc.No. 78. Pub!. Sydney Univ. Post-Graduate Comm.Vet. Sci., Sydney, Aust.

16 Lowman, B. G., N. A Scott, and S. H. Somerville.1976. Condition scoring of cattle. Bull. No.6. EastScotland Coil. Agric., Anim. Prod., Advisory Dev.Dep.

17 Macmillan, K. L., and A. M. Bryant. 1980. Cowcondition and its relation with production andreproduction. Pages 165-171 in Proc. RuakuraFarmers Conf. 32.

18 McGill, R., J. W. Tukey, and W. A. Larsen. 1978.Variations of box plots. Am. Stat. 32(1): 12.

19 Mulvany, P. 1981. Dairy cow condition scoring.Handout No. 4468. Nat!. Inst. Res. Dairying,Shinfield, Reading, UK.

20 Nicoll, G. B. 1981. Sources of variation in thecondition scoring of cows. Ir. J. Agric. Res. 20:27.

21 Perkins, B. L., R. D. Smith, and C. J. Sniffen.1985. Body condition scoring: a useful tool fordairy herd management. Fact Sheet p150.00. Dairymanagement, Coop. Ext. New York State, CornellUniv., Ithaca, NY.

22 Richards, M. W., J. C. Spitzer, and M. B. Warner.1986. Effect of varying levels of postpartumnutrition and body condition at calving on sub­sequent reproductive performance in beef cattle. J.Anim. Sci. 62: 300.

23 Rogers, G. L., C. Grainger, and D. F. Earle. 1979.Effect of nutrition of dairy cows in late pregnancyon milk production. Aust. J. Exp. Agric. Anim.Husb. 19:7.

24 Weaver, L. D. 1984. Periparturient events andsubsequent fertility in dairy cows. Pages 82-84 inProc. 17th Annu. Conv. Am. Assoc. Bovine Pract.West Lafayette, IN.

25 Wildman, E. E. 1979. The effects of a dairy cowbody condition scoring system on selected pro­duction and metabolic parameters. Ph.D. Thesis,Virginia Polytechnic Inst. State Univ., Blacksburg.

26 Wildman, E. E., G. M. Jones, P. E. Wagner, R. L.Boman, H. F. Troutt, and T. N. Lesch. 1982. Adairy cow body condition scoring system and itsrelationship to selected production characteristics.J. Dairy Sci. 65:495.

27 Wright, I. A., and AJ.F. Russe!. 1984. Partition offat,0ody composition and body condition score inmature cows. Anim. Prod. 38:23.