Es

Ra

b

c

d

e

a

ARRAA

KBEGD

1

rp

(0

0d

Small Ruminant Research 102 (2012) 135– 141

Contents lists available at ScienceDirect

Small Ruminant Research

jou rn al h om epa ge: www. elsev ier .com/ locate /smal l rumres

conomic values for disease resistance traits in dairy goat productionystems in Kenya

.C. Betta,b,c,∗, M.G. Gichehad, I.S. Kosgeye, A.K. Kahie, K.J. Petersa

Department of Animal Breeding in the Tropics and Sub-Tropics, Humboldt University of Berlin, Philippstraße 13, 10115 Berlin, GermanySwedish University of Agricultural Sciences (SLU), Department of Animal Breeding and Genetics, P.O. Box 7023, SE-750 07 Uppsala, SwedenInternational Livestock Research Institute (ILRI), P.O. Box 30709, Nairobi 00100, KenyaDepartment of Agricultural Sciences, Faculty of Agriculture and Life Sciences, Lincoln University, Lincoln 7647, P.O. Box 84, Canterbury, New ZealandAnimal Breeding and Genetics Group, Department of Animal Sciences, Egerton University, P.O. Box 536, Egerton 20115, Kenya

r t i c l e i n f o

rticle history:eceived 22 February 2011eceived in revised form 7 July 2011ccepted 12 July 2011vailable online 17 August 2011

eywords:reeding objectivesconomic valuesoatsisease resistance

a b s t r a c t

This study estimated economic values (EVs) for disease resistance traits for dairy/crossbredgoats in Kenya. The traits mean somatic cell count (SCC, cells/�l) and faecal worm eggcount (FEC, epg) were taken as indicator traits for the most prevalent diseases in thesmallholder farms i.e., mastitis and helminthiosis, respectively. Economic weights wereobjectively assigned to these indicator traits in a selection index such that the overall gainsin the breeding objective traits were maximised. Four options for calculating EVs for SCCand FEC were considered. Option 1, response from single trait selection was set equiva-lent to index response for the trait. Option 2, response from single trait selection was setequivalent to maximum gains achievable. Option 3, level of FEC/SCC was set to zero; andoption 4, response in FEC/SCC was set to the minimum gains achievable. In all the options,EVs with/without risk for breeding objective traits 12-month live weight (LW-kg); ADG,average post-weaning daily gain (ADG-g); DMY, average daily milk yield (DMY-kg) wereused. For each production trait selected for improvement, a less positive response in thetraits FEC and SCC would be desirable. Maximum negative EVs were achieved at a pointwhere the response in SCC was set at zero (option 3) while EVs for SCC were zero whenresponse for DMY was maximised (option 2). In addition, considerable differences in EVs

for SCC were obtained when EVs with/without risk were used. Similar results were alsoobserved for FEC when LW was the objective of improvement. However, more positive EVsfor FEC were estimated relative to ADG and DMY. The results confirm that there is a scopeto incorporate disease resistance traits in a breeding program with objective of reducingdisease incidences and the costs of disease control.

. Introduction

Dairy goats (manly crossbreds) potential have been wellecognised in Kenya and their contribution to the denselyopulated areas remains high to-date. Despite this, the sec-

∗ Corresponding author at: Swedish University of Agricultural SciencesSLU), Department of Animal Breeding and Genetics, P.O. Box 7023, SE-7507 Uppsala, Sweden.

E-mail address: rawlynce@yahoo.com (R.C. Bett).

921-4488/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.smallrumres.2011.07.008

© 2011 Elsevier B.V. All rights reserved.

tor remains largely marginalised when compared to otherruminants i.e., cattle and sheep, due to lack of well designedand executed breeding programs. Recently, Bett et al.(2011) observed that a crossbreeding program targeting75% crossbreds was optimal and desirable for implemen-tation in the smallholder production systems. In the study,breeding objectives incorporating farmer’s preferences and

risk, which were lacking, were defined. Economic values(EVs) for production (12-month live weight, average post-weaning daily gain, average daily milk yield) and functional(mature weight and number of kids weaned) traits were

nant Research 102 (2012) 135– 141

Table 1Economic values with (� = 0.002)a and without risk for average daily milkyield (DMY), average post-weaning daily gain (ADG) and 12-month liveweight (LW) obtained from crossbreds with 75% German Alpine bloodlevel (B1).

Traits

DMY ADG LW

Economic values without riskb 49.50 51.94 77.65b

136 R.C. Bett et al. / Small Rumi

also estimated. However, EVs for disease resistance traitswere not considered.

Disease resistance traits have multi-fold influence oninput and output of a production system, which in turnaffects profits and EVs (Sivarajasingam, 1995; Gichehaet al., 2005). They are further complicated by envi-ronmental factors, nonlinearity effects and interactions(Sivarajasingam, 1998), and therefore cannot be estimatedusing the conventional approaches. Economic values fordisease resistance traits can be predicted by assigningrelative weights of indicator traits to matched specificbreeding objectives (Sivarajasingam, 1995; Gicheha et al.,2005).

In Kenya, disease resistance traits were perceivedby dairy goat farmers to be of primary importance intheir production systems (Bett et al., 2009a). The mostimportant diseases noted in these systems were masti-tis and helminthiosis. Economic consequences of mastitisinclude loss of milk production, increased culling rate,and increased cost of labour for detection and veterinarytreatment. Infection to gastro-intestinal parasites is asso-ciated with delayed and reduced productivity, increasedsusceptibility to other infections and increased use ofanthelminthics or cost of controlling helminths. Conven-tionally, use of drugs (antibiotics) and anthelmintics havebeen used to control mastitis and helminthiosis, respec-tively. However, with the emergence of drug resistantparasites, high costs of pharmaceutical products, chemi-cal residues in animal products and changes in consumerpreferences (Baker, 1995; Barillet et al., 2001; Barillet,2007), selection for improved disease resistance in ani-mals is becoming more common in livestock breeding(Stear and Murray, 1994; Gicheha et al., 2005). Selectionfor disease resistance however requires its incorpora-tion in the breeding objectives. Development of breedingobjectives involves identification of important traits of aproduction system and estimation of their economic val-ues. This study estimates EVs for disease resistance traitsfor dairy goats in the smallholder production systems inKenya.

2. Methodology

2.1. Data source

The selected study sites were in Central, Rift Valley,Coast and Nyanza administrative provinces of Kenya (Bettet al., 2009a). In these regions, there is widespread dairygoat production under low input smallholder systems sup-ported by different donor organizations (see Bett et al.,2009a,b). Biological and economic data used in the param-eterization of the models (Bett et al., 2011) were derivedfrom dairy goat records (Krause, 2005) and follow-upfield studies (Bett et al., 2009a,c). Performance traits arerecorded by farmers registered with the Dairy Goat Asso-

ciation of Kenya (DGAK) since 1992. The EVs estimated byBett et al. (2011) (Table 1) for production traits were usedas input parameters in the selection index to derive EVs fordisease resistance traits.

Risk-rated economic values 35.91 45.42 65.59

a Arrow Pratt coefficient of absolute risk aversion see Bett et al. (2011).b In Kenya Shillings – KES, where 1 USD = 70.00 KES.

2.2. Disease resistance traits

Even well-researched definition of breeding objectivesand selection criteria may never be used in practise if thosedefinitions do not take into account the perception andwishes of the breeders for whom they are designed. Inthis study, disease resistance traits were ranked highly byfarmers in a field survey (Bett et al., 2009a). Mastitis andhelminthiosis were the most common diseases in thesesmallholder farms. In genetic evaluation schemes howeverindirect selection is necessary in order to lower the preva-lence of these two diseases. Mean somatic cell count (SCC,cells/�l) and faecal worm egg count (FEC, epg) can be takenas indicator traits for mastitis and helminthiosis, respec-tively (Baker et al., 1999; Rodriguez-Zas et al., 2000; Barilletet al., 2001; Barillet, 2007). The indicator traits SCC andFEC can be selected for in a breeding program without anydetrimental effect on each other (Sechi et al., 2009).

2.3. Estimation of economic values

Incorporation of disease resistance traits in a breedingprogram requires calculation EVs for indicator traits, FECand SCC. Methods for estimating EVs for disease resistancetraits in a single trait index (Sivarajasingam, 1995) andmulti-trait index (Sivarajasingam, 1998; Gicheha and Bett,2010) were applied. The approach objectively assigns eco-nomic weights to an indicator trait in a selection index suchthat the overall gains in the breeding objective traits aremaximised. This method is based on a selection index the-ory (Hazel, 1943), thus a vector of selection index weightsis calculated as:

b = P−1 Ga (1)

where b is a vector containing the coefficients of theindex traits, and P−1 the inverse phenotypic (co)variancesmatrix of the characters in the selection index. The genetic(co)variance matrix of selection criteria traits and traits inthe breeding objective is represented by the G matrix, anda is a vector of economic weights in Kenya Shillings (KES)of traits in the breeding objective.

Response (g) after one round of selection on the indexfor each trait, assuming a selection intensity of 1, is calcu-lated as:

′

g = b G

�I(2)

where �I is the standard deviation of the selection index(�I =

√b′Pb) which equals overall response for traits.

R.C. Bett et al. / Small Ruminant Research 102 (2012) 135– 141 137

Table 2Phenotypic standard deviations (�p), heritabilities along the diagonal and phenotypic (above diagonal) and genetic (below diagonal) correlations.

Traitsa �p LW ADG DMY FEC SCC

LW 2.93 0.26 0.69 0.08 −0.08 0.00ADG 0.11 0.44 0.10 0.06 0.00 0.00DMY 3.01 0.34 0.07 0.38 0.00 −0.14FEC 2 (1000×) 0.30 −0.05 −0.21 0.31 0.00SCC 1.46 0.00 0.00 0.15 0.00 0.15

a LW, 12-month live weight (kg); ADG, average post-weaning daily gain (kg); DMY, average daily milk yield (kg); FEC, faecal worm egg count (epg) andSCC, mean somatic cell count (cells/�l).

Table 3Economic values (EVs) for SCC and response to selection for DMY and SCC under different breeding optionsa and EVs.

Option Response using EVs without risk Response using EVs with risk

EV (KES) DMY (kg) SCC Overall response rIH EV (KES) DMY (kg) SCC Overall response rIH

1 −58.80 1.14 0.05 53.54 0.58 −42.65 1.14 0.05 38.84 0.582 0.00 1.18 0.10 56.62 0.63 0.00 1.18 0.10 42.21 0.633 −109.43 1.06 0.00 52.21 0.51 −79.39 1.06 0.00 37.87 0.514 −56.45 1.15 0.05 53.66 0.58 −41.10 1.15 0.05 38.93 0.58

sponse;g SCC sett

g

g

wa

rttlDttatmfpd

c

O

O

OO

a Option 1: response from single trait selection set equivalent to index reains achievable; option 3: level of SCC set to zero; option 4: response inraits SCC and DMY; rIH, accuracy of selection.

The overall response expressed in KES (gKES) is thusiven by:

KES = b′GM

�I(3)

here M is a vector of economic weights in the diagonalnd zeros off diagonals.

Breeding objectives were matched to expectedesponses in production traits and responses in theraits maximised relative to the overall gains. A total of fiveraits consisting of three production traits (LW), 12-monthive weight (kg); ADG, average post-weaning daily gain (g);MY, average daily milk yield (kg) and disease resistance

raits (FEC and SCC) were used. Economic values for resis-ance to helminths were estimated relative to LW, ADGnd DMY, and for resistance to mastitis estimated relativeo DMY. The indicator traits for resistance to helminths and

astitis were FEC and SCC, respectively. Consequently,or each production trait selected for improvement a lessositive response in the traits FEC and SCC would beesirable.

Four options for calculating EVs for SCC and FEC wereonsidered.

ption 1 Response from single trait selection was setequivalent to index response for the trait. Forinstance, response in LW from single trait selec-tion is equals to index response in LW. This alsoapplies for ADG and DMY. Single trait selection

implies that only that trait is in the breedingobjective and selection index.

ption 2 Response from single trait selection was setequivalent to maximum gains achievable.

ption 3 Level of FEC/SCC was set to zero.ption 4 Response in FEC/SCC was set to the minimum

gains achievable.

option 2: response from single trait selection set equivalent to maximum to the minimum gains achievable. See Table 2 and text for definition of

In all the options, EVs with and without risk for traitsLW, ADG and DMY were used, Table 1 (Bett et al., 2011).Response in all the traits studied was calculated for a rangeof EVs. The candidates for selection had records for LW, ADGand FEC, while DMY and SCC were recorded in the dams ofthe selection candidates only. Phenotypic standard devia-tions, heritabilities and genetic and phenotypic correlationare shown in Table 2. These estimates are from crossbredgoats in Kenya (Ahuya et al., 2009). Where not available,averages in literature from the tropics were used (Baker,1995; Els, 1998; Neopane and Pollot, 1998; Goncalves andWechsler, 2000; Queiroz et al., 2000; Ribeiro et al., 2000;Barillet et al., 2001; Barillet, 2007).

3. Results

3.1. Economic values for SCC and response to selection

The EVs for SCC and response to selection in DMY andSCC evaluated under different breeding options and EVs(with and without risk) are shown in Table 3. Since lowerand more negative EVs for SCC were more desirable, max-imum EVs were achieved at a point where the responsein SCC was set at zero (option 3) while EVs for SCC werezero when response for DMY was maximised (option 2).However, option 3 registered the lowest overall responsesand accuracy of selection and option 2 the largest. Consid-erable differences in EVs for SCC were observed when EVswith and without risk were used (Table 3) e.g., −109.43 and−79.39 (option 3). Similarly, response to selection for thetraits varied depending on the scenarios and EVs for thematched trait.

3.2. Economic values for FEC and response to selection

The EVs for FEC and response to selection in ADG, LW,DMY and FEC under different breeding options and EVs(with and without risk) are shown in Table 4. The EVs

nant Research 102 (2012) 135– 141

ues

(EV

s)

for

FEC

and

resp

onse

to

sele

ctio

n

for

AD

G, L

W, D

MY

and

FEC

un

der

dif

fere

nt

bree

din

g

opti

onsa

and

EVs.

pti

on

Res

pon

se

usi

ng

EVs

wit

hou

t

risk

Res

pon

se

usi

ng

EVs

wit

h

risk

EV

(KES

)

AD

G

(kg)

LW

(kg)

DM

Y

(kg)

FEC

(ep

g)

Ove

rall

resp

onse

r IH

EV

(KES

)

AD

G

(kg)

LW

(kg)

DM

Y

(kg)

FEC

(ep

g)

Ove

rall

resp

onse

r IH

−0.0

8

0.01

––

−0.1

2

0.58

0.32

−0.0

7

0.01

––

−0.1

2

0.51

0.32

0.00

0.01

–

–

−0.0

7

0.57

0.32

0.00

0.01

–

–

−0.0

7

0.50

0.32

0.11

0.01

––

0.00

0.56

0.32

0.10

0.01

–

–

0.00

0.49

0.32

0.10

0.01

–

–

−0.0

2

0.57

0.32

0.05

0.01

–

–

−0.0

2

0.50

0.32

−38.

05

–

0.76

–

0.17

52.6

7

0.47

−32.

15

–

0.76

–

0.17

44.4

9

0.47

0.00

–0.

82–

0.40

63.9

1

0.55

0.00

–0.

82–

0.40

53.9

8

0.55

−58.

75

–

0.66

–

0.00

50.8

9

0.44

−49.

62

–

0.66

–

0.00

42.9

8

0.44

−30.

70

–0.

79–

0.22

54.1

3

0.49

−26.

05

–

0.79

–

0.22

45.6

9

0.49

−23.

30

–

–

0.57

−0.4

8 39

.51

0.39

−16.

90

–

–

0.57

−0.4

8

28.6

6

0.39

0.00

–

–

0.61

−0.3

0

30.1

9

0.33

0.00

–

–

0.61

−0.3

0

21.9

0

0.33

20.5

5

–

–

0.55

0.00

27.0

0

0.30

14.9

0

–

–

0.55

0.00

19.5

9

0.30

10.5

5

–

–

0.59

−0.1

6

27.7

9

0.31

7.50

–

–

0.59

−0.1

6

20.1

8

0.31

resp

onse

from

sin

gle

trai

t

sele

ctio

n

set

equ

ival

ent

to

ind

ex

resp

onse

;

opti

on

2:

resp

onse

from

sin

gle

trai

t

sele

ctio

n

set

equ

ival

ent

to

max

imu

m

gain

s

ach

ieva

ble;

opti

on

3:

leve

l of S

CC

set

to

zero

;on

se

in

SCC

set

to

the

min

imu

m

gain

s

ach

ieva

ble.

See

Tabl

e

2

and

text

for

defi

nit

ion

of

trai

ts;

r IH

, acc

ura

cy

of

sele

ctio

n.

138 R.C. Bett et al. / Small Rumi

for FEC estimated relative to ADG were positive in all theoptions except in the scenario where the response in sin-gle trait selection (ADG) was set equivalent to the indexresponse (option 1). Similar results were obtained whenEVs for FEC were estimated relative to DMY. However, EVsfor FEC were more negative in this case. For LW, lowerand more negative EVs for FEC were obtained. The low-est EVs KES −58.75 (without risk) and −49.62 (with risk)were estimated in option 3 when the response in FEC waszero.

Responses in all breeding objective traits varied amongthe scenarios (options 1–4). Only responses for ADG werenot affected among the scenarios. Subsequently, the cor-responding responses in FEC were negative in options 1, 2and 4 and were similar to responses estimated in relationto DMY though the values in the latter were more nega-tive (Table 4). Responses in FEC in relation to LW were allpositive.

3.3. Economic values for FEC and response to selectionwhen more than one trait was considered in the index

The EVs for FEC and response to selection in ADG, LW,DMY and FEC under different breeding options (1 and 2)and when more than one trait was considered in the indexare shown in Table 5. When both DMY and LW were used asindex traits and selection on DMY assumed, the index didnot produce any results in option 1. However, when LW wasassumed, responses were positive and EVs for FEC was KES-5.45. In option 2, the EVs for FEC when DMY and LW werethe objective of improvement were KES −75.00 and 40.00,respectively. In comparison to a one trait index (Table 4),the EVs for FEC when LW was the objective of improvementwere large and more positive for both options, while moredesirable EVs were obtained in option 2 when DMY wasused.

Inclusion of a third trait in the index did not produceany results for DMY and ADG for the first scenario (option1); a sign that LW which apparently had the largest EVs(Table 1) dictated the gains and EVs for FEC. When LWwas the objective of improvement, EVs for FEC was lowerand more negative in option 1 than the two trait index(DMY and LW). Increasing the number of traits in the indexalso resulted in improved overall responses and accuracyof selection. In option 2, EVs were negative when DMY orADG were the objective of improvement and positive whenLW was considered. Selecting for maximum gains in LWresulted in undesirable (positive) EVs for FEC; KES 40.00.Larger negative EV for FEC i.e., −98.00 was reached whenresponses for ADG was maximised (Table 5).

4. Discussion

Essentially, SCC is used as an indicator of udder healthstatus for management and selection purposes. Nielsenet al. (2006) incorporated mastitis resistance as a non-market value trait in the breeding objective by allowing

a certain percentage loss in the selection response to milkyield that farmers or companies were willing to sacrifice.In this study, EVs for SCC were predicted by matchingthis trait to the expected responses in the production trait Ta

ble

4Ec

onom

ic

val

Trai

tO

AD

G

1 2 3 4

LW

1 2 3 4

DM

Y

1 2 3 4

aO

pti

on

1:op

tion

4:

resp

R.C. Bett et al. / Small Ruminant Research 102 (2012) 135– 141 139

Table 5Economic values (EVs) for FEC and response to selection for ADG, LW, DMY and FEC when more than one production trait was considered in the indexunder different breeding optionsa and EVs without risk.

Trait/scheme Response using EVs without risk

EV (KES) DMY (kg) LW (kg) ADG (kg) FEC (epg) Overall response rIH

DMY and LWOption 1

DMY – – – – – – –LW −5.45 0.51 0.77 – 0.20 83.81 0.49

Option 2DMY −75.00 0.61 0.50 – −0.20 83.52 0.46LW 40.00 0.38 0.81 – 0.40 97.76 0.54

DMY, LW and ADGOption 1

DMY – – – – – – –ADG – – – – – – –LW −10.90 0.56 0.88 0.00 0.26 93.88 0.55

Option 2DMY −74.00 0.65 0.63 0.00 −0.10 88.63 0.48ADG −98.00 0.64 0.49 0.00 −0.24 92.74 0.48

sponse;g of selec

si1spSwrprfaa

aclbstniLnstz

cTmedabcbmt

LW 40.00 0.44 0.93

a Option 1: response from single trait selection set equivalent to index reains achievable. See Table 2 and text for definition of traits; rIH, accuracy

elected for improvement and then maximising responsesn these traits relative to overall gains (Sivarajasingam,995; Gicheha et al., 2005). The objective in the currenttudy was to improve DMY while relatively reducing anyositive responses in SCC. Lower and more negative EVs forCC were therefore considered desirable. This was realisedhen SCC had zero response (Table 3). However, maximum

esponse for DMY was achieved when no emphasis waslaced on SCC. Response for DMY and SCC were compa-able across the EVs used (with and without risk) but EVsor SCC were more desirable when EVs without risk waspplied (Table 3). This is probably because of the high EVsssumed for the trait DMY (Table 1).

The choice of EVs for the resistance trait that can bepplied in a breeding programme and the response to beonsidered is determined by the level of disease preva-ence (Sivarajasingam, 1995). So far, no genetic antagonismetween milk yield and clinical mastitis has been demon-trated in small ruminants (Barillet et al., 2001). This meanshat selection for DMY may not necessarily be accompa-ied by a substantial increase in clinical or chronic mastitis

ncidences, which is contrary to dairy cattle (Colleau ande Bihan-Duval, 1995; Kelm et al., 2000). In this case, eco-omic concerns on mastitis incidences will not negateelection for increased milk yield. The EVs for resistanceo mastitis in a breeding programme can be assumed to beero (option 2, Table 3).

The EVs for FEC were estimated using the same pro-edure and with additional production traits in the index.he results are in agreement with those estimated foreat sheep (Gicheha et al., 2005). Based on the produc-

rs’ objectives i.e., level of disease incidences or cost ofisease control, several options on EVs that can be used in

breeding programme from a single trait index are possi-le (Table 4). If the objective is to minimise high disease

hallenges, large and more negative EVs for FEC woulde appropriate for the smallholders. This would lead toarginal losses in response for the production trait in ques-

ion (DMY, LW and ADG) and negative responses in FEC

0.00 0.46 112.53 0.62

option 2: response from single trait selection set equivalent to maximumtion.

(Table 4). When the level of disease resistance is tolerablepossibly due to routine worm control measures, suscepti-bility can be held constant. In this case, the response forFEC can be set to zero (option 3). If worm incidences arenot a significant constraint, zero EVs in FEC would be rec-ommended, because this would lead to maximum gains inthe production traits (option 2 – Table 4).

In a multi-trait index, EVs for FEC estimated relative toLW were more unfavourable when gains were optimisedunlike the traits ADG and DMY (Table 5). Additionally, anattempt to estimate EVs and gains using option 1 did notproduce any results for traits ADG and DMY. These findingsindicate that LW might be dominating the index. Increas-ing the number of traits in the index resulted in improvedoverall responses and accuracy of selection. It seems logi-cal to use more traits in the index when estimating EVs forresistance traits. Care should however be exercised whenmaking a choice on the trait(s) upon which optimisation inresponse is targeted since this will result in different EVsfor the resistance traits. The weighting given to resistancetraits is at the mercy of the economic weights of the breed-ing objective traits and the correlation between the twocategories of traits (Sivarajasingam, 1998). The challengearises when indicator traits are to be selected against butpositively correlated to breeding objective traits or whenselected for but negatively correlated to breeding objectivetraits. At present, no rule guides the consideration of whichtraits to use in a selection index to predict EVs for resistancetraits, and simulation based on the economic importanceplaced on the objective traits alone can be speculative.Breeding objective traits and potential index traits for indi-rect selection for functional traits in a breeding programmehave been discussed (Groen et al., 1997). Traits with pos-itive and high genetic correlations with indicator traits orresistance traits can actually enhance the prediction (selec-

tion for) of EVs and responses with more accuracy.

The effects of modern husbandry have favoured theprevalence of parasites related mastitis and helminthiosisin the tropics (Baker, 1995; Barillet et al., 2001; Gicheha

nant Res

Congress on Genetics Applied to Livestock Production, vol. 24, Armi-

140 R.C. Bett et al. / Small Rumi

et al., 2005; Barillet, 2007). First, highly susceptible ani-mals have been allowed to survive or retained in a breedingflock besides high stocking rates, low hygiene, poor milkingand drying regimes, and continual use of facilities (milkingequipments and housing). Secondly, the scope for con-trolled grazing is limited in many smallholder farmingsystems. Lastly, even with better housing, it is possible totransfer internal parasites when pastures and forages arecut and carried to the animals. These have led to increasedmorbidity incidences in the dairy/crossbred goat flocks.Incorporation of SCC and FEC in a selection programmewill reduce disease incidences and the costs of control-ling these diseases. However, the feasibility of includingthese traits in a low-input breeding programme needsto be cautiously considered due to logistical constraints,additional measurement equipments, and the cost of mea-surement and recording (Gicheha et al., 2007). A study incattle showed that recording for disease traits is expen-sive and could cost up to 60% of the total variable costs(Itty et al., 1998). In a dairy goat breeding program, thesecosts can be reduced by measuring and recording FEC onlyin males or young males that have attained four monthsof age because acquired resistance becomes only apparentafter this age (Stear and Murray, 1994). Similarly, record-ing protocols proposed for meat sheep breeding programsthat are less costly can be adopted (Gicheha et al., 2005).For SCC, a simplified testing can be organised where sam-ples are taken only on specific test-days, that is, two tofour times during the first four test days for each doe,also referred to as part-lactation sampling design (Barillet,2007).

5. Conclusions

The findings in this study indicate that there is ascope to incorporate disease resistance traits in a breed-ing program with objective of reducing disease incidencesand the costs of disease control. Indirect selection usingindicator traits is necessary in order to lower the preva-lence of diseases in the genetic evaluation schemes. Thesomatic cell count (SCC) can be taken as indicator trait formastitis and faecal worm egg count (FEC) for helminthio-sis, and their economic values (EVs) estimated relativeto production traits in the breeding objective. However,care should be exercised when making a choice on thetrait(s) upon which optimisation in response is targetedsince this will result in different EVs for the resistancetraits. Indicator or resistance traits can be selected forwhen the genetic correlations are highly and positivelycorrelated (or against when negatively correlated) to thebreeding objective traits in order to increase prediction offavourable EVs (more negative) and accuracy of responses.

Acknowledgements

We thank Humboldt University of Berlin, Swedish Uni-

versity of Agricultural Sciences (SLU), Egerton Universityand International Livestock Research Institute (ILRI) forprovision of facilities.

earch 102 (2012) 135– 141

References

Ahuya, C.O., Ojango, J.M.K., Mosi, R.O., Okeyo, A.M., 2009. Performanceof Toggenburg dairy goats in smallholder production systems of theeastern highlands of Kenya. Small Rumin. Res. 83, 7–13.

Baker, R.L., 1995. Genetics of disease resistance in small ruminants inAfrica. In: Gray, G.D., Woolaston, R.R., Eaton, B.T. (Eds.) Breeding forResistance to Infectious Disease in Small Ruminants, Australian Cen-ter for International Agricultural Research (ACIAR), Monograph, No.34, Canberra, Australia, pp. 120–138.

Baker, R.L., Mwamachi, D.M., Audho, E.O., Thorpe, W., 1999. Genetic resis-tance to gastro-intestinal nematode in Red Maasai, Dorper and RedMaasai × Dorper ewes in sub-humid tropics. Anim. Sci. 69, 335–344.

Barillet, F., 2007. Genetic improvement for dairy production in sheep andgoats. Small Rumin. Res. 70, 60–75.

Barillet, F., Rupp, R., Mignon-Grasteau, S., Astruc, J.-M., Jacquin, M., 2001.Genetic analysis for mastitis resistance and milk somatic cell score inFrench Lacaune dairy sheep. Genet. Select. Evol. 33, 397–415.

Bett, R.C., Kosgey, I.S., Kahi, A.K., Peters, K.J., 2009a. Analysis of productionobjectives and breeding practises of dairy goats in Kenya. Trop. Anim.Health Prod. 41, 307–320.

Bett, R.C., Kosgey, I.S., Kahi, A.K., Peters, K.J., 2009b. Realities in breedimprovement programmes for dairy goats in the East and CentralAfrica. Small Rumin. Res. 85, 157–160.

Bett, R.C., Bett, H.K., Kahi, A.K., Peters, K.J., 2009c. Evaluation and effec-tiveness of breeding and production services for dairy goat farmers inKenya. Ecol. Econ. 68, 2451–2460.

Bett, R.C., Kosgey, I.S., Kahi, A.K., Peters, K.J., 2011. Definition of breedingobjectives and optimum crossbreeding levels for goats in the small-holder production systems. Small Rumin. Res. 96, 16–24.

Colleau, J.J., Le Bihan-Duval, E., 1995. A simulation study of selection meth-ods to improve mastitis resistance of dairy cows. J. Dairy Sci. 78,659–671.

Els, J.F., 1998. Heritability estimates for the improved Boer goat. In: Pro-ceedings of the Sixth World Congress on Genetics Applied to LivestockProduction, vol. 24, Armidale, NSW, Australia, 11–16 January, 1998,pp. 193–196.

Gicheha, M.G., Kosgey, I.S., Bebe, B.O., Kahi, A.K., 2005. Economic valuesfor resistance to gastrointestinal helminthes in meat sheep in Kenya.J. Anim. Breed. Genet. 122, 165–171.

Gicheha, M.G., Kosgey, I.S., Bebe, B.O., Kahi, A.K., 2007. Efficiency of alter-native schemes breeding for resistance to gastrointestinal helminthesin meat sheep. Small Rumin. Res. 69, 167–179.

Gicheha, M.G., Bett, R.C., 2010. Increasing accuracy in estimation of eco-nomic values for resistance to gastrointestinal helminths in meatsheep. Trop. Anim. Health Prod. 42, 391–396.

Goncalves, H.C., Wechsler, F.S., 2000. Genetic and environmental factorsaffecting milk production of dairy goats in Brazil. In: Proceedings of theSeventh International Conference on Goats, 15–18 May, 2000, Tours,France, p. 240.

Groen, A.F., Steine, T., Colleau, J.J., Pedersen, J., Pribly, J., Reinsch, N., 1997.Economic values in dairy cattle breeding, with special reference tofunctional traits. Report of an EAAP-Working Group. Livest. Prod. Sci.49, 1–21.

Hazel, L.N., 1943. The genetic basis for constructing selection indexes.Genetics 28, 476–490.

Itty, P., Chema, S., D’ieteren, G., Durkin, J., Leak, S., Maehl, J., Maloo, S.,Mukendi, F., Nagda, S., Rarieya, J., Thorpe, W., Trail, J.,1998. Economicaspects of cattle production and of chemoprophylaxis for control oftrypanosomiasis in village East African Zebu cattle in Kenya. In: Pro-ceedings of a meeting held in Nairobi, Kenya from the 23rd to 27thNovember 1987 organized by the International Livestock Centre forAfrica and the International Laboratory for Research on Animal Dis-eases. ILCA/ILRAD.

Kelm, S.C., Freeman, A.E., Committee, N.C.T., 2000. Direct and corre-lated responses to selection for milk yield: results and conclusions ofregional Project NC-2, “Improvement of dairy cattle through breeding,with emphasis on selection”. J. Dairy Sci. 83, 2721–2732.

Krause, A.-N., 2005. Breeding programmes for small ruminants in the trop-ics with special reference to the crossbreeding programme of the DairyGoat Association of Kenya (DGAK). Ph.D. Thesis, Humboldt University,Berlin, Germany, p. 149.

Neopane, S.P., Pollot, G.E., 1998. Genetic parameter estimation of weighttraits in Nepalese hill goats. In: Proceedings of the Sixth World

dale, NSW, Australia, 11–16 January 1998, pp. 189–192.Nielsen, H.M., Christensen, L.G., Odegard, J., 2006. A method to define

breeding goals for sustainable dairy cattle production. J. Dairy Sci. 89,3615–3625.

nant Res

Q

R

R

S

R.C. Bett et al. / Small Rumi

ueiroz, S.A., Holon, P., Ribeiro, A.C., Resende, K.T., Ribeiro, S.D.A., 2000.Milk yield heritability, repetability and genetic trends in Saanen goatsin Brazil. In: Proceedings of the Seventh International Conference onGoats, 15–18 May, 2000, Tours, France, p. 241.

ibeiro, A.C., Lui, J.F., Queiroz, S.A., Ribeiro, S.D.A., Resende, K.T., 2000. Milkyield, genetic trend and genetic and phenotypic parameter estimatesfor milk yield and lactation length of Saanen goats in Brazil. In: Pro-ceedings of the Seventh International Conference on Goats, 15–18 May2000, Tours, France, pp. 242–243.

odriguez-Zas, S.L., Gianola, D., Shook, G.E., 2000. Evaluation of models forsomatic cell score lactation patterns in Holsteins. Livest. Prod. Sci. 67,19–30.

echi, S., Salaris, S., Scala, A., Rupp, R., Moreno, C., Bishop, S.C., Casu, S.,2009. Estimation of (co)variance components of nematode parasites

earch 102 (2012) 135– 141 141

resistance and somatic cell count in dairy sheep. Ital. J. Anim. Sci. 8,156–158.

Sivarajasingam, S., 1995. A method to estimate economic weights fortraits of disease resistance in sheep. In: Proceedings of the AustralianAssociation of Animal Breeding and Genetics, vol. 11, Australian Asso-ciation of Animal Breeding and Genetics, Roseworthy, Australia, pp.65–69.

Sivarajasingam, S., 1998. Economic weights to maximise responses in pro-duction traits. In: Proceedings of the Sixth World Congress on Genetics

Applied to Livestock Production, vol. 25, Armidale, NSW, Australia, pp.339–342.

Stear, M.J., Murray, M., 1994. Genetic resistance to parasite disease: partic-ularly of resistance in ruminants to gastrointestinal nematodes. Vet.Parasitol. 54, 161–176.