Genetic resistance to gastro-intestinal nematode parasites in Galla and Small East African goats in the sub-humid tropics

Animal Science 2001, 73: 61-70 1357-7298/01/09280061$20 00 2001 British Society of Animal Science Genetic resistance to gastro-intestinal nematode parasites in Galla and Small East African goats in the sub-humid tropics R. L. Baker, J. O. Audho, E. O. Aduda and W. Thorpe International Livestock Research Institute (ILRI), PO Box 30709, Nairobi, Kenya E-mail L. BAKER@CGIAR. ORG Abstract A study was carried out from 1992 to 1996 to compare the resistance to naturally acquired gastro-intestinal (GI) nematode parasite infections (predominantly Haemonchus contortus) of Galla and Small East African goats in the sub-humid coastal region of Kenya. A total of 204 Galla and 349 Small East African (SEA) kids were born from five kiddings. These were the progeny of 18 Galla and 17 SEA bucks. Live weights (LWT), blood packed-cell volume (PCV) and faecal egg count (FEC) were recorded at 1- to 2-month intervals from birth to about 14 months of age. The SEA kids were more resistant to GI nematode parasites than Galla kids as shown by their significantly lower FEC (P < 0 001) in the post-weaning period (8- to 14-month-old kids) and lower mortality from birth to 14 months of age (P < 0 05). There was no significant (P > 0 05) breed effect on PCV, but Galla kids were significantly heavier (P < 0 001) at all measurement times between birth and 14 months of age. Heritability estimates for LWT, PCV and FEC at the different sampling times were characterized by high standard errors. Heritability estimates for records taken at 4 5 and 8 months of age from a repeated measures analysis were 0 18 (s.e. 0 08) for PCV and 0 13 (s.e. 0 07) for logarithm-transformed FEC. The phenotypic and genetic correlation estimates between PCV and LFEC were moderately to highly negative and averaged 0 36 and 0 53, respectively. The results are discussed in relation to the limited evidence for resistance to GI nematode infections in goats and compared with the much stronger evidence for resistance in sheep. Keywords: goats, genetic resistance, Haemonchus contortus, tropical Africa. Introduction Gastro-intestinal (GI) nematode parasites constitute one of the most important disease constraints to sheep and goat production in the tropics (Fabiyi, 1987). Widespread infection with internal parasites in grazing animals, associated production losses, anthelminthic drug costs and death of infected animals are some of the major concerns. Current control methods focus on reducing the contamination of pastures through anthelmintic drug treatment and/or controlled grazing. In the tropics, the application and effectiveness of these methods are limited by the high costs of anthelmintic drugs, their uncertain availability, increasing frequency of drug resistance in the parasites (Waller, 1997) and limited scope in many communal farming systems for controlled grazing. An attractive, sustainable solution is to utilize host genetic resistance for disease control. There is now a substantial body of evidence showing that genetic variation for resistance or tolerance (resilience) to GI nematode parasites occurs in sheep (Gray et al., 1995; Woolaston and Baker, 1996). There are also reports of genetic variation for resistance to GI parasites among goat breeds (Preston and Allonby, 1978; Cabaret and Anjorand, 1984; Shavulimo et al., 1988; Richard et al., 1990; Pralomkarn et al., 1997; Baker et al., 1998a; Costa et al., 2000) but for many of these studies the experimental design was inadequate for a valid breed comparison. The first evidence for genetic variation for resistance within goat breeds was discouraging with heritability estimates that were not significantly different from zero (Woolaston et al., 1992). However, some more recent studies indicate that there is significant variation for resistance within goat breeds (Patterson et al., 1996a and b; Morris et al., 1997; Mandonnet et al., 2001; Vagenas et al., 2000). 61

62 Baker, Audho, Aduda and Thorpe Table 1 Number of kids born by breed and year of birth Year of birth Breed 1992 1993 1994 1995 1996 Total Galla 43 48 30 41 42 204 Small East African 48 55 46 89 111 349 Total 91 103 76 130 153 553 This paper reports a study that investigated genetic variation for resistance to GI nematode parasites (predominantly Haemonchus contortus) of kids within and between the Galla and Small East African (SEA) breeds in the sub-humid region of coastal Kenya. Evidence from the same study that SEA does were more resistant to GI nematode parasite infections than Galla does was reported by Baker et al. (1998a). Material and methods Experimental site This study was carried out at the Diani Estate farm located 20 km south of Mombasa in the sub-humid coastal region of Kenya. Rainfall is bimodally distributed with rainy seasons in March-June and October-December. Rainfall data were collected daily at Diani Estate; over the 5-year study period (1992-96) the average annual rainfall was 819 mm. The monthly mean maximum temperatures range between 28 and 33ºC, while relative humidity ranges between 0 60 to 0 90. The soils on the farm are sandy, well drained and are characterized by very low levels of nitrogen and phosphorus, organic matter, cation exchange and water-holding capacity. The vegetation is composed of natural pasture and bush, with Heteropogon contortus and Hyparrhenia rufa the predominant perennial grass species. Experimental design and data recorded The Galla goat has its origin in the arid and semi-arid regions of northern Kenya, southern Somalia and parts of southern and South-eastern Ethiopia. It is a relatively large (long-legged) white goat. The SEA goat, which is found throughout Kenya and in Northern Tanzania, is a relatively small goat of variable colour (white, brown, black). The SEA goats used in this study were all acquired from the subhumid coastal region of Kenya, while the Galla goats were purchased from pastoralists in the semi-arid zone of Kenya. Bucks and does were purchased from as wide a range of flocks and owners as possible to ensure they were unrelated. Usually no more than one buck and two or three does were purchased from each flock. Each year, the does were mated with four or five bucks of their own breed in single-sire groups over a 6-week mating period. At least three or four new bucks of each breed were used at each mating so that over the entire study (1992-1996), 18 Galla bucks and 17 SEA bucks were used. A total of 204 Galla and 349 SEA kids were born in the study (Table 1) and these were the progeny of 175 Galla does and 270 SEA does. The does had the opportunity to produce kids in different years of the study so that over the entire study 94 Galla does and 111 SEA does were used. Sixty (29%) of the Galla kids were born as twins, while 144 (41%) of the SEA kids were born as twins and nine (3%) were born as triplets. Kids were weighed as close to birth as possible, usually within 24 h and then every 2 weeks up to weaning at 3 months of age. All male kids were castrated within two weeks of birth. Blood packedcell volume (PCV) was recorded on all kids at 1, 2 and 3 months of age and if an individual kid had a PCV less than 16% at these samplings they were treated with an anthelmintic drug. Faecal egg count (FEC) was recorded on all kids at 2 and 3 months of age and if an individual kid had a FEC greater than 3000 eggs per gram (e.p.g.) at these samplings they were treated with an anthelmintic drug. These intervention levels applied to the first two crops of kids (1992 and 1993). The high pre-weaning mortality in 1993 (46%, Table 3) prompted changes of management in the subsequent years: to wean 1 month later and to alter the pre-weaning intervention levels for PCV (level raised to 20%) and FEC (level lowered to 2000 e.p.g. ) for anthelmintic treatment. All the kids were treated with an anthelmintic at weaning. Weaning took place at 3 months of age in the first two crops of kids and at 4 months of age for the last three crops. Post-weaning all the kids of both breeds were grazed together on pasture until a monitor group of about 30 to 40 kids (representing approx. equal numbers of both breeds and both sexes), from which samples were taken every week, reached a FEC averaging about 1000 to 1500 e.p.g. This intervention level for FEC was chosen to ensure that the kids had high enough

Genetics of resistance to nematodes in goat kids in the tropics 63 infection levels to express resistance to endoparasites, but at the same time to keep mortality rates to acceptable levels. All the kids were then weighed and faeces and blood samples were taken on 2 days consecutively. The kids were then all treated with an anthelmintic. This procedure was repeated until the kids reached about 14 months of age, which involved five sampling times. The age of the kids at each post-weaning sampling time varied across years and the sampling ages shown are the averages across years. No kids were individually drenched at any of the post-weaning sampling times. Kids and does were allowed to graze for about 10 h/ day (i.e. from about 07:00 to 17:00 h) and then housed in a simple roofed shed at night for security reasons. FEC were determined using the modified McMaster method (Ministry of Agriculture, Fisheries and Food, 1977) with a lower limit of detection of 50 e.p.g. of faeces. PCV was measured by the microhaematocrit method. At each sampling time bulked faecal samples by breed were cultured and the larval species identified (Hansen and Perry, 1994). The larval differentiation showed that the predominant larvae were Haemonchus contortus (mean 63%, range 28 to 85%), Trichostrongylus spp. (mean 14%, range 2 to 35%) and Oesophagostomum spp. (mean 18%, range 1 to 43%). In the 1st year of the experiment it was ascertained that the GI nematode parasites of both sheep and goats on this farm were resistant to both ivermectin and fenbendazole (Mwamachi et al., 1995). In addition, it was found that this resistance was more severe when goats were treated with the same anthelmintic dose as that recommended for sheep. Therefore all the kids born in 1993 and subsequently were treated with levamisole (Wormacid Plus, Cosmos, Nairobi, Kenya) at 1 5 times the recommended dose rate for sheep (i.e. 22 5 mg per kg body weight based on the heaviest kid at any sampling time) and this anthelmintic treatment remained effective throughout the rest of the experiment. All kids were sprayed with an acaricide (Triatix, Coopers, Kenya) every 2 weeks to control tick infestations and prevent tick-borne diseases. Whenever the PCV fell below 20%, infections with trypanosomes were monitored using the dark ground/phase contrast buffy coat technique (Paris et al., 1982). This was to ensure that the anaemia being measured by PCV was due to Haemonchus contortus rather than to trypanosomosis, which also causes anaemia. While there have been reports of trypanosomosis in cattle in the same farm, no cases were diagnosed in the goats during this study. Statistical analyses The traits analysed at each sampling time were live weights (LWT), PCV and FEC. At all the postweaning sampling times the average of the records taken on 2 days consecutively were analysed. Because of a skewed distribution, FEC was analysed using a logarithmic transformation (LFEC, log 10 (FEC + 25)). The results were back-transformed by taking anti-logarithms of the least squares means and presented as geometric means (GFEC). All statistical tests for FEC were applied to the transformed data. LWT, PCV and LFEC were analysed with a mixed model analysis of variance using ASREML (Gilmour et al., 1999). The model fitted included (when significant) the fixed effects of year of birth (five classes, 1992-96), breed (two classes, Galla and SEA), sex (two classes, male and female), age of dam (four classes, 2 to 5+ years of age), birth type (two classes single and multiple born lambs), lamb age as a linear covariate and any significant (P < 0 05) first-order interactions. An animal model was fitted in the mixed model ASREML univariate analyses and variance components and heritabilities were estimated for the direct additive effect. A maternal effect was also fitted in the REML analyses but because of the limited pedigree structure of the data it was not possible to separate this maternal effect into a genetic component and a permanent environmental component. The likelihood ratio test was used to test the significance of including each additional random effect. The analyses at each sampling time prior to weaning included all kids alive at that sampling time regardless of whether some kids had been treated with an anthelmintic at the previous sampling time or not. The consequence of this was investigated in additional analyses where the effect of treatment at the previous sampling (treated or not treated) was added to the statistical model for analyses of LFEC, PCV and LWT. This effect was only significant occasionally and even when it was significant made little difference to the breed effect. Genetic and phenotypic correlations were estimated using the sire-model option in the REML program (AIREML) of Johnson and Thompson (1995) fitting just the direct additive random effect. Bivariate analyses were undertaken for LWT and PCV, LWT and LFEC and PCV and LFEC at each sampling time. Because it was not possible to fit different fixed effects for different traits in AIREML all the significant fixed effects and any significant interactions for each pair of traits were fitted in these analyses.

64 Baker, Audho, Aduda and Thorpe Kid mortality was analysed by AIREML as a binomial trait (dead or alive) using a logit transformation. The fixed effects fitted in the mixed model analyses were year of birth, breed, sex and birth type. Mortality was assessed pre-weaning (birth to 3 (or 4) months of age), post-weaning (3 (or 4) to 14 months of age) and for total mortality (birth to 14 months of age). Results Breed effect Breed effects for LWT, PCV and FEC are shown in Table 2. Galla kids were significantly heavier (P < 0 001) than SEA kids at birth and at all subsequent recording times between one and 14 months of age. At some sampling times there was a year of birth breed interaction, which in all cases was due to a greater difference in weight between the breeds in the years with the more favourable seasons for growth. There was no significant breed effect for PCV (P > 0 05), except for the sampling at 10 months of age when SEA kids had a significantly higher (P < 0 05) PCV than the Galla kids. There was no significant breed difference (P > 0 05) in FEC in young kids between 2 and 4 5 months of age. However, at all subsequent sampling times (i.e. for kids between 8 and 14 months of age), the SEA kids were more resistant to endoparasites than Galla kids Table 2 Least-squares means (standard errors in parentheses) by breed, residual standard deviation and direct additive heritability (h 2 a) and maternal effect (m) restricted maximum likelihood estimates (standard errors in parentheses) for live weight (LWT, kg), packed cell volume (PCV, percent), logarithm transformed faecal egg count (LFEC) and the geometric mean of faecal egg count (GFEC, eggs per gram) at the different sampling ages No. No. Significance Residual Trait Galla Galla SEA SEA for breed s.d. h 2 a m Birth WT 204 2 79 (0 06) 349 1 95 (0 05) *** 0 39 0 39 (0 16) 0 23 (0 06) 1 month LWT 156 5 8 (0 14) 293 4 6 (0 13) *** 0 96 0 15 (0 12) 0 25 (0 06) PCV 156 24 5 (0 50) 293 23 7 (0 46) 3 92 0 24 (0 14) 2 months LWT 152 8 0 (0 20) 287 6 3 (0 18) *** 1 51 0 16 (0 13) 0 14 (0 06) PCV 152 27 7 (0 42) 287 27 2 (0 36) 3 50 0 07 (0 10) 0 12 (0 06) LFEC 152 2 01 (0 07) 287 1 97 (0 07) 0 50 0 08 (0 07) GFEC 152 102 287 93 3 months LWT 145 9 8 (0 23) 283 7 6 (0 20) *** 1 82 0 04 (0 10) 0 16 (0 06) PCV 145 27 3 (0 43) 283 27 1 (0 37) 3 23 0 16 (0 13) 0 14 (0 06) LFEC 125 2 58 (0 07) 242 2 43 (0 05) 0 64 0 13 (0 11) GFEC 125 380 242 269 4 5 months LWT 140 11 9 (0 19) 274 9 7 (0 18) *** 2 01 0 05 (0 10) 0 18 (0 06) PCV 140 25 5 (0 20) 274 25 5 (0 19) 2 82 0 36 (0 17) 0 17 (0 06) LFEC 140 2 76 (0 17) 274 2 65 (0 16) 0 52 0 26 (0 15) GFEC 140 575 274 447 8 months LWT 126 14 4 (0 35) 248 11 5 (0 28) *** 1 97 0 24 (0 15) 0 17 (0 06) PCV 126 26 9 (0 27) 248 26 7 (0 21) 2 89 0 05 (0 11) LFEC 126 3 16 (0 04) 248 2 91 (0 03) *** 0 35 0 20 (0 14) GFEC 126 1445 248 812 10 months LWT 121 16 0 (0 38) 246 12 9 (0 30) *** 1 98 0 39 (0 19) 0 15 (0 06) PCV 121 24 2 (0 40) 246 25 2 (0 34) * 3 22 0 30 (0 18) LFEC 121 3 20 (0 03) 246 3 12 (0 03) * 0 28 0 24 (0 15) GFEC 121 1585 246 1318 12 months LWT 118 17 0 (0 20) 243 14 3 (0 16) *** 1 94 0 20 (0 16) 0 17 (0 06) PCV 118 25 9 (0 54) 243 26 4 (0 41) 2 82 0 02 (0 08) 0 23 (0 07) LFEC 118 3 28 (0 03) 243 3 14 (0 02) *** 0 23 0 13 (0 14) GFEC 118 1905 243 1380 14 months LWT 113 19 3 (0 35) 236 16 1 (0 26) *** 2 21 0 11 (0 14) 0 18 (0 07) PCV 113 25 0 (0 31) 236 25 6 (0 22) 3 11 0 00 (0 09) LFEC 113 3 30 (0 03) 236 3 18 (0 02) *** 0 24 0 08 (0 12) GFEC 113 1995 236 1514

Genetics of resistance to nematodes in goat kids in the tropics 65 as shown by their significantly lower FEC (P < 0 05). At the samplings when the kids were 3 months of age and 12 months of age there was a significant year of birth by breed interaction. For the 3 month sampling this interaction was due to the Galla kids having a significantly higher FEC than the SEA kids in the 3 years with the highest level of challenge (i.e. when the mean GFEC ranged between 400 e.p.g. and 1100 e.p.g. ), while in the other 2 years which had low challenge levels (mean GFEC of less than 200 e.p.g.), there was no significant breed difference for FEC. For the 12-month sampling the Galla had a higher FEC than SEA in 4 years while in the other year (1995) the Galla had a significantly lower FEC than the SEA. Mortality was consistently higher in the Galla than the SEA kids both pre-weaning and post weaning, but this breed difference was only significant (P < 0 05) for the total mortality from birth to 14 months of age (Table 3). Mortality was significantly higher in males than females both pre-weaning and post weaning, while twin-born kids had a significantly higher total mortality than single-born kids (Table 3). Year of birth had a highly significant effect (P < 0 001) on both pre-weaning and postweaning mortality (Table 3). Mortality was highest in the first 2 years of the experiment. For those kids for which reasons for death or loss were ascertained (85% of the total number of kids that died or disappeared from the experiment), the predominant causes of pre-weaning mortality were the mismothering complex (20%), poisoned or predation (19%), lost or stolen (11%), abortions and stillbirths (8%) and pneumonia (7%). The predominant causes Table 3 Least-squares means (standard errors in parentheses) by breed, sex, birth type and year of birth for pre-weaning, postweaning and total mortality (%) from birth to 14 months of age Pre-weaning Post-weaning Total Effect mortality mortality mortality Breed * Galla 25 (3 6) 20 (3 5) 43 (3 9) Small East African 18 (2 6) 16 (2 4) 32 (3 0) Sex * ** *** Female 18 (2 6) 13 (2 4) 32 (3 1) Male 25 (3 0) 23 (3 3) 45 (3 4) Birth type * Single 19 (2 4) 14 (2 3) 33 (2 8) Twin 24 (3 4) 22 (3 7) 43 (4 0) Year of birth *** *** *** 1992 13 (3 4) 42 (6 0) 49 (5 6) 1993 46 (5 6) 23 (5 3) 60 (5 1) 1994 25 (5 2) 20 (4 9) 40 (5 7) 1995 25 (4 3) 23 (4 2) 41 (4 5) 1996 8 (2 2) 3 (1 1) 11 (2 4) of post-weaning mortality were pneumonia (28%) haemonchosis (22%) and lost or stolen (10%). Genetic parameter estimates Univariate heritability estimates for all traits analysed are characterized by high standard errors (Table 2). The direct additive heritability estimate for LWT was high at birth (0 39, s.e. 0 16), declined to about 0 05 (s. e 0 10) at 3 to 4 months of age and then increased again post weaning with a maximum heritability of 0 39 (s.e. 0 19) in 10-month-old kids. The maternal effect was significant for LWT at all ages from birth to 14 months of age. The univariate direct additive heritability estimates for PCV and LFEC were not significant at any of the pre-weaning measurement times (1, 2 and 3 months of age). The maternal effect estimates were significant at some measurement ages for PCV but were not significant and were very small at all measurement ages for LFEC. All analyses for LFEC in Table 2 just included the direct additive genetic effect in the model. There was an indication that there may be significant direct additive genetic variation for both PCV and LFEC post weaning and particularly at 4 5, 8 and 10 months of age (Table 2). To investigate this further multivariate REML analyses were undertaken for PCV and LFEC at these ages (Table 4) fitting just a direct additive genetic effect. For PCV the pair-wise genetic Table 4 Genetic and phenotypic correlation estimates and direct additive heritability estimates for packed cell volume (PCV, %) and logarithm transformed faecal egg count (LFEC) at 4 5, 8 and 10 months of age from multivariate animal model REML analyses and direct additive heritability estimates from repeated measures analyses PCV LFEC Estimate Estimate Parameter estimate (standard error) (standard error) Genetic correlation 4 5 and 8 months 0 79 (0 25) 0 47 (0 48) 4 5 and 10 months 0 86 (0 18) 0 52 (0 55) 8 and 10 months 0 99 (0 32) 0 74 (0 50) Phenotypic correlation 4 5 and 8 months 0 29 (0 05) 0 04 (0 05) 4 5 and 10 months 0 31 (0 05) 0 00 (0 05) 8 and 10 months 0 38 (0 05) 0 07 (0 05) Heritability (multivariate REML) 4 5 months 0 25 (0 11) 0 15 (0 10) 8 months 0 11 (0 10) 0 16 (0 10) 10 months 0 29 (0 11) 0 12 (0 10) Heritability (repeated measures) 4 5 and 8 months 0 18 (0 08) 0 13 (0 07) 8 and 10 months 0 11 (0 06) 0 14 (0 08) 4 5, 8 and 10 months 0 16 (0 06) 0 09 (0 05)

66 Baker, Audho, Aduda and Thorpe Table 5 Genetic and phenotypic correlation estimates among live weight (LWT, kg) packed cell volume (PCV, %) and logarithm transformed faecal egg count (LFEC) at each measurement age from bivariate sire model REML analysis Sampling Genetic correlations (standard errors) Phenotypic correlations age (months) LWT - PCV LWT - LFEC PCV - LFEC LWT - PCV LWT - LFEC PCV - LFEC 2 0 62 (0 20) 0 43 (0 21) 0 44 (0 23) 0 48 0 08 0 28 3 0 82 (0 12) 0 03 (0 31) 0 12 (0 32) 0 51 0 18 0 35 4 5 0 83 (0 12) 0 78 (0 46) 0 95 (0 34) 0 40 0 19 0 47 8 0 25 (0 47) 0 28 (0 37) 0 32 (0 64) 0 19 0 05 0 27 10 0 46 (0 25) 0 18 (0 35) 0 59 (0 29) 0 38 0 18 0 47 12 0 48 (0 28) 0 25 (0 39) 0 73 (0 38) 0 16 0 01 0 41 14 n.e. n.e. n.e. 0 19 0 03 0 30 n.e. = not estimable analysis did not converge. correlations among these three measurement times were high and significant and averaged 0 88. It therefore seems reasonable to consider these three measurements as predictors of the same trait and treat them as repeated measures. The repeated measures REML analysis included both the direct additive genetic effect and a permanent environmental effect as random effects in the model. When this was done the highest heritability was 0 18 (s.e. 0 08) for records taken at 4 5 and 8 months of age. When the same approach was taken for analysis of LFEC the pair-wise genetic correlations among measurement times were positive (average of 0 58) but not significantly different from either zero or unity. If we assume that these three records are predictors of the same trait and treat them as repeated records very similar heritability estimates were obtained from repeating records at 4 5 and 8 months of age (0 13, s.e. 0 07) and repeating records at 8 and 10 months of age (0 14, s.e. 0 08). There was no increase in heritability for either PCV or LFEC by fitting records at all three measurement times in a repeated measures analysis (Table 4), nor was there any significant change in heritability from including the records taken at 12 and 14 months of age in repeated measures analyses. The permanent environmental effect among these measurement ages was not significant for PCV or LFEC in the repeated measures analyses. The increase in the heritability of PCV and LFEC in the post-weaning measurements versus the estimates in 1, 2 and 3 month old kids is at least partly due to the estimates for kids from 4 5 to 14 months of age being based on the average of records taken on 2 days consecutively. The average repeatability of PCV taken on 2 days consecutively was 0 60 and the average repeatability for LFEC was 0 70. The heritability of a single LFEC measurement (single PCV measurement) at 4 5, 8 and 10 months of age was 0 04, s.e. 0 07 (0 13, s.e. 0 09), 0 14, s.e. 0 09 (0 06, s.e. 0 08) and 0 11, s.e. 0 09 (0 22, s.e. 0 12), respectively. Genetic and phenotypic correlations among LWT, PCV and LFEC at each measurement time are presented in Table 5. The phenotypic and genetic correlations between LWT and PCV were all positive and averaged 0 33 and 0 58, respectively. All the phenotypic correlations were significantly different from zero, but only the genetic correlations at 2, 3 and 4 5 months of age were significantly different from zero and these averaged 0 76. All but one of the seven phenotypic correlation estimates between LWT and LFEC were negative and the seven estimates averaged 0 10. None of the genetic correlation estimates between LWT and LFEC were significantly different from zero and there was no consistent trend with three estimates being negative and the other three estimates being positive. All the phenotypic and genetic correlation estimates between PCV and LFEC were moderately to highly negative and averaged 0 36 (all estimates significantly different from zero) and 0 53 (three estimates significantly different from zero), respectively. Discussion Breed differences for resistance to endoparasites In many of the published reports on breed differences for resistance to GI nematodes in goats, the sample sizes per breed or genotype were small (Preston and Allonby, 1978; Shavulimo et al., 1988; Pralomkarn et al., 1997; Costa et al., 2000), limiting the possibility of detecting significant breed differences or of making correct inferences about the breeds being evaluated. However, a series of larger studies in France have shown that Alpine dairy goats are more resistant than Saanen goats (Cabaret and Anjorand, 1984; Richard et al., 1990).

Genetics of resistance to nematodes in goat kids in the tropics 67 In the present study significant breed differences in resistance as assessed by FEC was not found until the kids reached about 8 months of age and then differences in resistance persisted through to 14 months of age. The only significant breed difference in PCV was at the 10 month of age sampling when SEA had a higher PCV than Galla kids (P < 0 05). When the GI nematode infection is predominantly H. contortus, as it was in this study, then PCV can be considered as an indicator of resilience. Previous studies have made the distinction between resistance and resilience to endoparasites (Woolaston and Baker, 1996). Resistance is defined as the initiation and maintenance of responses provoked in the host to suppress the establishment of parasites and/or eliminate parasite burdens. Resilience (or tolerance) is defined as the ability of the host to survive and be productive in the face of parasite challenge. In the does which were the parents of these kids, there were significant breed differences for both FEC and PCV with the SEA having lower FEC (more resistant) and higher PCV (more resilient) than the Galla (Baker et al., 1998a). This can be contrasted with the findings from a sheep study undertaken at the same farm and at the same time as the goat study (Baker, 1998). The resistant breed (the Red Maasai) manifested both increased resistance (lower FEC) and resilience (higher PCV) to GI nematode parasites (predominantly H. contortus) compared with the susceptible breed (the Dorper). The breed difference for PCV was first significant in 2-monthold lambs, while for FEC it was first significant in 3-month-old lambs and then both the resistance and resilience of the Red Maasai persisted in the lambs up to 12 months of age (Baker, 1998) and in Red Maasai ewes (Baker et al., 1999). These results suggest that in these breeds in this environment the resistant sheep breed (the Red Maasai) developed resistance and resilience at an earlier age than the resistant goat breed (the SEA). A possible explanation for this finding is that goats are predominantly browsers while sheep are predominantly grazers and therefore natural selection for resistance to GI nematode infections has been stronger in sheep than goats. Another finding to support this explanation is that the breed differences in mortality were much more marked in sheep (both lambs and ewes) than in the goats (kids and does). In both cases the resistant breeds had significantly lower mortality than the susceptible breed (Baker, 1998) and mortality can be considered as another manifestation of resilience, as well as an indicator of the strength of natural selection. Genetic parameter estimates The first published estimates of heritability of resistance to GI nematode infections (predominantly H. contortus and T. colubriformis) in goats were from an experiment in Fiji and were not at all encouraging (Woolaston et al., 1992). Heritability estimates for LFEC were not significantly different from zero in weaner kids (0 04, s.e. 0 03) or in adult goats (0 08, s.e. 0 06) and repeatability estimates were also nonsignificant. The heritabilities for LFEC and PCV in the present study are more encouraging and from the repeated measures analyses (Table 4) are in the range from 0 10 to 0 20 for kids between 4 5 and 10 months of age. There have been some indications from sheep studies that the heritabilty of LFEC in lambs may be higher in the more susceptible breeds v. the resistant breeds (Baker, 1998; Baker et al., 1998b). There were not enough sires used in the present study to investigate this issue in any meaningful way. For example, the heritability estimate for LFEC from the repeated measures analysis for records taken at 4 5, 8 and 10 months of age in the complete data set (i.e. pooling variance components within breeds) was 0 09 (s.e. 0 05) (Table 4). The heritability estimate for the same measurements from repeated measures analyses was 0 17 (s.e. 0 16) for the Galla kids and 0 17 (s.e. 0 10) for SEA kids. In support of the results of the present study, some large, ongoing experiments in Scotland and Guadeloupe are also indicating that there may indeed be significant genetic variation for resistance within goat breeds. Patterson et al. (1996a and b) reported the segregation of male and female adult Cashmere goats into responders (resistant goats) and non-responders (susceptible goats) following natural pasture challenge and indoor trickle challenge with Teladorsagia (Ostertagia) circumcincta and Trichostrongylus vitrinus. Based on these results an experiment was initiated at the Macaulay Land Use Research Institute in Scotland in 1993 with the objective of estimating genetic variance for resistance to GI nematode parasites in a crossbred population of cashmere goats and genetic and phenotypic correlations with production traits (Vagenas et al., 2000). Within this population of goats one line was selected for reduced FEC and a randomly selected control line was also maintained. After 5 years of selection (1997) a significant selection response for reduced FEC had been achieved. FEC measurements were recorded on average five times each year when the goats were between 12 and 17 months of age. The cube root of FEC (CFEC) was analysed using AIREML to estimate the genetic parameters (830 goats with FEC measurements and 3100 goats with cashmere

68 Baker, Audho, Aduda and Thorpe and LWT data). Consistent with the results of the present study maternal effects were only important for LWT and the direct additive heritability for a single CFEC measurement was 0 17 (s.e. 0 02) and for the mean of the five measurements 0 31 (s.e. 0 08). None of the genetic or phenotypic correlations between CFEC and the production traits (LWT and the cashmere-fibre traits) were significantly different from zero. Genetic resistance of Creole goats to GI nematode parasites (predominantly H. contortus and T. colubriformis) has been studied at Guadeloupe in the French West Indies since 1993 with the objective of estimating genetic parameters for use in breeding schemes. Mandonnet et al. (2001) estimated genetic parameters for FEC (fourth root transformed), PCV and LWT using a REML animal model for kids at weaning (1202 kids from 49 sires) and during a postweaning fattening period up to 10 months of age (979 kids from 55 sires). For FEC at weaning at 82 days of age the direct additive heritability estimate was 0 20 and the maternal genetic heritability was 0 26. After weaning the maternal genetic effect was not significant for FEC and the direct additive heritability estimates increased from 0 14 (s.e. 0 05) at 4 months of age to 0 33 (s. e 0 06) in 10-month-old kids. Direct additive heritability estimates for PCV were 0 20, 0 22, 0 10 and 0 33 in kids recorded at 4, 6, 8 and 10 months of age respectively. It has been suggested that the heritability of FEC in goats may be somewhat lower than that found in sheep (Morris et al., 1997). However, as more data for goats are analysed from larger experiments it appears that the heritability estimates for both FEC and PCV in kids are very similar to those found in lambs (i.e. for kids and lambs up to about one year of age). The heritability of FEC in Merino lambs in Australia is in the range of 0 2 to 0 3 under typical pasture challenge conditions, although it may be higher when artificial indoor challenge is used under highly controlled conditions or if the average of several FEC measurements is used (Woolaston and Eady, 1995). The weighted heritability of FEC in lambs from nine estimates from New Zealand (all pasture challenge) involving Romney or Romney derived breeds was 0 23 (s.e. 0 02) (Morris et al., 1995). As is also found in lambs, the heritability of FEC in kids increases with age. The highest estimates are commonly found when kids are between about 10 and 18 months of age. However, significant heritability estimates for FEC in lambs are commonly found when the lambs are 4 to 6 months of age (Morris et al., 1995; Bishop et al., 1996; Baker, 1998), suggesting that genetic variation for resistance manifests itself at a slightly older age in goats than sheep. Heritability estimates for FEC in ewes are similar in magnitude to those in lambs and there is a positive genetic correlation between these traits (Morris et al., 1998). To date the published heritability estimates for FEC in adult goats have all been less than 0 10 and not significantly different from zero (Woolaston et al., 1992; Morris et al., 1997) but further data are required to confirm these results. Most of the genetic correlation estimates among FEC, PCV and LWT in goats published to date (Vagenas et al., 2000; Mandonnet et al., 2001), including those in this study, are from relatively small data sets and are therefore imprecisely estimated. However it appears that the genetic correlation estimates between FEC and LWT in kids are mostly close to zero or not significantly different from zero and therefore not antagonistic to index selection. This is consistent with most of the published estimates for the same genetic correlation in lambs (Morris et al., 2000), with the notable exception of a high negative correlation between FEC and LWT in Scottish Blackface lambs between 3 and 6 months of age (Bishop et al., 1996). It is possible that the genetic correlation between FEC and LWT may be affected by the nematode parasite species which is infecting the lambs. Most of the strong antagonistic genetic correlations have resulted from Teladorsagia (Ostertagia) species infections compared with the more neutral estimates from studies where H. contortus or Trichostrongylus species are the predominant infections. Genetic and phenotypic correlation estimates between FEC and PCV are consistently negative (and therefore favourable) in both sheep (Albers et al., 1987; Baker et al., 1999) and goats when they are infected with H. contortus. Conclusion These results, in conjunction with the results of the production and reproduction of the two breeds in this environment (Baker, 1998; Baker et al., 1998a and b) have important implications for smallholder farmers in coastal Kenya and other humid/subhumid regions of East Africa. SEA goats are clearly more productive than Galla goats in this environment, not only in being more resistant and resilient to GI nematode parasites (particularly H. contortus), but also in terms of lower kid and doe mortality and having a higher overall reproductive and maternal performance. The previous evaluations of SEA and Galla goats have been undertaken in the semi-arid highlands of Kenya (Ruvuna et al., 1988). There is no evidence for any breed by environment interaction for resistance (e.g. Shavulimo et al., 1998) as the SEA goats are consistently the more resistant breed in both humid and semi-arid environments. However, it is likely that the overall reproductive

Genetics of resistance to nematodes in goat kids in the tropics 69 performance of Galla goats would be better in a drier climate from where they originated but further studies are required to document this. The present study is barely large enough to estimate variance components very precisely but the estimates of heritability for both PCV and LFEC in kids between about 4 and 10 months of age are encouraging enough to suggest that selection for these traits should be feasible. Acknowledgements We are extremely grateful to Dr Rene Haller, Managing Director, Baobab Farm for provision of land, animals and facilities at Diani Estate for this research. This is ILRI publication no. 200079. References Albers, G. A. A., Gray, G. D., Piper, L. R., Barker, J. S. F., Le Jambre, L. F. and Barger, I. A. 1987. The genetics of resistance and resilience to Haemonchus contortus in young Merino sheep. International Journal for Parasitology 17: 1355-1363. Baker, R. L. 1998. A review of genetic resistance to gastrointestinal nematode parasites in sheep and goats in the tropics and evidence for resistance in some sheep and goat breeds in sub-humid coastal Kenya. Animal Genetics Resources Information Bulletin 24: 13-30. Baker, R. L., Mwamachi, D. M., Audho, J. O., Aduda, E. O. and Thorpe, W. 1998a. Resistance of Galla and Small East African goats in the sub-humid tropics to gastrointestinal nematode infections and the peri-parturient rise in faecal egg counts. Veterinary Parasitology 79: 53-64 Baker, R. L., Mwamachi, D. M., Audho, J. O., Aduda, E. O. and Thorpe, W. 1999. Genetic resistance to gastro-intestinal nematode parasites in Red Maasai, Dorper and Maasai Dorper ewes in the sub-humid tropics. Animal Science 69: 335-344. Baker, R. L., Rege, J. E. O., Tembely, S., Mukasa-Mugerwa, E., Anindo, D., Mwamachi, D. M., Thorpe, W. and Lahlou- Kassi, A. 1998b. Genetic resistance to gastrointestinal nematode parasites in some indigenous breeds of sheep and goats in East Africa. Proceedings of the sixth world congress on genetics applied to livestock production, Armidale, vol. 25, pp. 269-272. Bishop, S. C., Bairden, K., McKellar, Q. A., Park, M. and Stear, M. J. 1996. Genetic parameters for faecal egg count following mixed, natural, predominantly Ostertagia circumcincta infection and relationships with live weight in young lambs. Animal Science 63: 423-428. Cabaret, J. and Anjorand, N. 1984. Comparaison de l'infestation naturelle par les strongyles et Moniezia sp. chez les races caprines Alpine et Saanen. Bulletin de la Societe Francaise de Parasitologie 2: 49-52. Costa, C. A. F., Vieira, L. da S., Berne, M. E. A., Silva, M. U. D., Guidoni, A. L. and Figueiredo, E. A. P. 2000. Variability of resistance in goats infected with Haemonchus contortus in Brazil. Veterinary Parasitology 88: 153-158. Fabiyi, J. P. 1987. Production losses and control of helminths in ruminants of tropical regions. International Journal for Parasitology 17: 435-442. Gilmour, A. R., Cullis, B. R., Welham, S. J. and Thompson, R. 1999. ASREML reference manual. New South Wales (NSW) Agriculture biometrics bulletin no. 3. NSW Agriculture, Orange, NSW, Australia. Gray, G. D., Woolaston, R. R. and Eaton, B. T. 1995. Breeding for resistance to infectious diseases of small ruminants. Australian Centre for International Agricultural Research (ACIAR) monograph no. 34, Canberra, Australia. Hansen, J. and Perry, B. 1994. The epidemiology, diagnosis and control of helminth parasites of ruminants, second edition. International Laboratory for Research on Animal Diseases (ILRAD), Nairobi, Kenya. Johnson, D. L. and Thompson, R. 1995. Restricted maximum likelihood estimation of variance components for univariate animal models using sparse matrix techniques and average information. Journal of Dairy Science 78: 449-456. Mandonnet, N., Aumont, G., Fleury, J., Arquet, R., Varo, H., Gruner, L., Bouix, J. and Vu Tien Khang, J. 2001. Assessment of genetic variability of resistance to strongyles in Creole goats in the humid tropics. Journal of Animal Science In press. Ministry of Agriculture, Fisheries and Food. 1977. Manual of veterinary parasitology laboratory techniques. Ministry of Agriculture Fisheries and Food technical bulletin no. 18. Her Majesty's Stationery Office, London. Morris, C. A., Bisset, S. A., Vlassoff, A., West, C. J. and Wheeler, M. 1998. Faecal nematode egg counts in lactating ewes from Romney flocks selectively bred for divergence in lamb faecal egg count. Animal Science 67: 283-288. Morris, C. A., Vlassoff, A., Bissett, S. A., Baker, R. L., Watson, T. G., West, C. J. and Wheeler, M. 2000. Continued selection of Romney sheep for resistance or susceptibility to nematode infection: estimates of direct and correlated responses. Animal Science 70: 17-27. Morris, C. A., Watson, T. G., Bisset, S. A., Vlassoff, A. and Douch, P. G. C. 1995. Breeding sheep in New Zealand for resistance or resilience to nematode parasites. In Breeding for resistance to infectious diseases in small ruminants (ed. G. D. Gray, R. R. Woolaston and B. T. Eaton), pp. 77-98. Australian Centre for International Agricultural Research monograph no. 34, Canberra, Australia. Morris, C. A., Wheeler, M., Hosking, B. C., Watson, T. G., Hurford, A. P., Foote, B. J. and Foote, J. F. 1997. Genetic parameters for milk yield and faecal nematode egg count in Saanen does. New Zealand Journal of Agricultural Research 40: 523-528. Mwamachi, D. M., Audho, J. O., Thorpe, W. and Baker, R. L. 1995. Evidence for multiple anthelmintic resistance in sheep and goats reared under the same management in coastal Kenya. Veterinary Parasitology 60: 303-313. Paris, J., Murray, M. and McOdimba, F. A. 1982. An evaluation of the sensitivity of current parasitological techniques for the diagnosis of bovine African trypanosomiasis. Acta Tropica 39: 307-316. Patterson, D. M., Jackson, F., Huntley, J. F., Stevenson, L. M., Jones, D. G., Jackson, E. and Russel, A. J. F. 1996a.

70 Baker, Audho, Aduda and Thorpe Studies on caprine responsiveness to nematodiasis: segregation of male goats into responders and nonresponders. International Journal for Parasitology 26: 187-194. Patterson, D. M., Jackson, F., Huntley, J. F., Stevenson, L. M., Jones, D. G., Jackson, E. and Russel, A. J. F. 1996b. The response of breeding does to nematodiasis: segregation into responders and non-responders. International Journal for Parasitology 26: 1295-1303. Pralomkarn, W., Pandey, V. S., Ngampongsai, W., Choldumrongkul, S., Saithanoo, S., Rattaanachon, L. and Verhulst, A. 1997. Genetic resistance of three genotypes of goats to experimental infection with Haemonchus contortus. Veterinary Parasitology 68: 79-90. Preston, J. M. and Allonby, E. W. 1978. The influence of breed on the susceptibility of sheep and goats to a single experimental infection with Haemonchus contortus. Veterinary Record 103: 509-512. Richard, S., Cabaret, J. and Cabourg, C. 1990. Genetic and environmental factors associated with nematode infection of dairy goats in Northwestern France. Veterinary Parasitology 36: 237-243. Ruvuna, F., Cartwright, T. C. and Blackburn, H. 1988. Gestation length, birth weight and growth rates of purebred indigenous goats and their crosses in Kenya. Journal of Agricultural Science, Cambridge 111: 363-368. Shavulimo, R. S., Rurangirwa, F., Ruvuna, F., James, A. D., Ellis, P. R. and McGuire, T. 1988. Genetic resistance to gastrointestinal nematodes, with special reference to Haemonchus contortus, in three breeds of goats in Kenya. Bulletin of Animal Health and Production in Africa. 36: 233-241. Vagenas, D., Bishop, S. C., Jackson, F., Merchant, M. and Russel, A. J. F. 2000. Heritabilities of, and correlations between, faecal egg counts and cashmere traits in goats. Proceedings of the British Society of Animal Science, 2000, p. 18. Waller, P. J. 1997. Anthelmintic resistance. Veterinary Parasitology 72: 391-412. Woolaston, R. R. and Baker, R. L. 1996. Prospects of breeding small ruminants for resistance to internal parasites. International Journal for Parasitology 26: 845-855. Woolaston, R. R. and Eady, S. J. 1995. Australian research on genetic resistance to nematode parasites. In Breeding for resistance to infectious diseases in small ruminants (ed. G. D. Gray, R. R. Woolaston and B. T. Eaton), pp. 53-75. Australian Centre for International Agricultural Research monograph no. 34, Canberra, Australia. Woolaston, R. R., Singh, R., Tabunakawai, N., Le Jambre, L. F., Banks, D. J. D. and Barger, I. A. 1992. Genetic and environmental influences on worm egg counts of goats in the humid tropics. Proceedings of the Australian Association of Animal Breeding and Genetics 10: 147-150. (Received 19 October 2000 Accepted 20 February 2001)