Quantitative trait loci for resistance to Haemonchus contortus artificial challenge in Red Maasai and Dorper sheep of East Africa

doi: 10.1111/j.1365-2052.2012.02401.x Quantitative trait loci for resistance to Haemonchus contortus artificial challenge in Red Maasai and Dorper sheep of East Africa K. Marshall*, J. M. Mugambi, S. Nagda*, T. S. Sonstegard, C. P. Van Tassell, R. L. Baker and J. P. Gibson *The International Livestock Research Institute, P.O. Box 30709-00100, Nairobi, Kenya. Veterinary Research Centre KARI, Muguga, P.O. Box 32-00902, Kikuyu, Kenya. Bovine Functional Genomics Laboratory, Agricultural Research Service, USDA, Beltsville, MD, 20705, USA. P. O. Box 238, Whangamata, 3643, New Zealand. Centre for Genetic Analysis and Applications, C.J. Hawkins Homestead University of New England, Armidale, NSW, 2351, Australia. Summary A genome-wide scan was performed to detect quantitative trait loci (QTL) for resistance to the gastrointestinal nematode Haemonchus contortus in a double backcross population of Red Maasai and Dorper sheep. The mapping population comprised six sire families, with 1026 lambs in total. The lambs were artificially challenged with H. contortus at about 6.5 months of age, and nine phenotypes were measured: fecal egg count, packed cell volume decline, two weight traits and five worm traits. A subset of the population (342 lambs) was selectively genotyped for 172 microsatellite loci covering 25 of the 26 autosomes. QTL mapping was performed for models which assumed that the QTL alleles were either fixed or segregating within each breed, combined with models with only an additive QTL effect fitted or both additive and dominance QTL effects fitted. Overall, QTL significant at the 1% chromosome-wide level were identified for 22 combinations of trait and chromosome. Of particular interest are a region of chromosome 26 with putative QTL for all nine traits and a region of chromosome 2 with putative QTL for three traits. Favorable QTL alleles for disease resistance originated in both the Red Maasai and Dorper breeds, were not always fixed within breed and had significant dominance effects in some cases. We anticipate that this study, in combination with follow-up work and other relevant studies, will help elucidate the biology of disease resistance. Keywords disease resistance, double backcross population, East African sheep, fecal egg count, gastrointestinal nematodes, whole-genome linkage analysis Introduction Gastrointestinal nematode parasites are well recognized as a major animal health constraint to sheep productivity globally, including in developing countries (Over et al. 1992; Perry et al. 2002). It follows that many sheepproducing countries have undertaken studies to better understand the genetic variation of resistance, both between breeds and between animals within a breed (Bishop & Morris 2007). These have included a number of QTL mapping studies for traits indicative of resistance (Crawford Address for correspondence K. Marshall, The International Livestock Research Institute, P.O. Box 30709-00100, Nairobi, Kenya. E-mail: k.marshall@cgiar.org Accepted for publication 03 July 2012 et al. 2006; Davies et al. 2006; Moreno et al. 2006; Beraldi et al. 2007; Gutiérrez-Gil et al. 2009; Marshall et al. 2009; Dominik et al. 2010; Silva et al. 2012). The main justification of this research was to find new means of reducing the impact of the nematodes on sheep productivity in the face of their widespread resistance to anthelmintics (Mwamachi et al. 1995; Waller 1997; Besier 2007; Bishop & Morris 2007). The Red Maasai and Dorper breeds of sheep, as well as their crosses, are commonly found in the East African countries of Kenya and Tanzania, where they make an important contribution to the livelihoods of their keepers, especially the poor. The Red Maasai sheep are resistant to and tolerant of gastrointestinal nematode infection, whereas the Dorper sheep are relatively more susceptible (Preston & Allonby 1978, 1979; Mugambi et al. 1996, 1997; Wanyangu et al. 1997; Baker et al. 2003; Baker & 2012 The Authors, Animal Genetics 2012 Stichting International Foundation for Animal Genetics 1

2 Marshall et al. Gray 2004). This breed difference presumably relates to the Red Maasai being native to specific areas of East Africa which are of high parasite challenge, whereas the Dorper was developed in South Africa in the 1940s from a cross between the Dorset Horn and the Black Head Persian breeds (Wilson 1991; Milne 2000). The increased gastrointestinal nematode resistance and tolerance of the Red Maasai in comparison with the Dorper has been demonstrated to translate into improved performance, more so in an environment with a strong nematode challenge in comparison with an environment with a low-level challenge, suggestive of a genotype by environment interaction (Baker et al. 2004; Bishop 2012). An experiment was initiated in Kenya to identify QTL affecting resistance to gastrointestinal tract parasites, particularly H. contortus, in a double backcross population of Red Maasai and Dorper sheep. This experiment comprised two challenge phases: a field challenge followed by an artificial (indoor trickle) challenge (Mugambi et al. 2005a,b). The QTL mapping results relating to the field challenge were presented in the study by Silva et al. (2012), whereas this paper presents the QTL mapping results from the artificial challenge. In this work, the mapping analysis was performed for a number of indicator traits for disease resistance and for models which assume that the QTL alleles are either fixed or segregating within a breed. The models in which QTL alleles are assumed segregating within a breed were included because there is no prior evidence or population genetics theory which suggests that the QTL alleles should be fixed within these breeds, particularly for the Dorper which is a recent cross. This study and that of Silva et al. (2012) are the first reports of QTL for disease resistance relative to East African sheep. Materials and methods Resource population The resource population has been described previously (Mugambi et al. 2005a,b; Silva et al. 2012). Briefly, a double backcross mapping population was produced by mating Red Maasai and Dorper ewes to six F1 rams (of which two were half-sibs). A total of 1026 lambs with data relevant to this work were produced, with half-sib progeny group sizes from 151 to 184. The lambs were born in five crops at approximately 6-month intervals, beginning November 1997. The animals were subject to a natural challenge followed by an artificial challenge, as described below. Lambs were kept on a ranch in Eastern Kenya until about 3 months of age, at which point they were treated with an anthelmintic and moved to the ILRI campus on the outskirts of Nairobi. Here, the lambs were subjected to a natural endoparasite challenge (mainly H. contortus and Trichostronglus spp.) by grazing infected pastures for 6 to 8 weeks, before being re-treated and moved indoors. After 2 to 3 weeks [during which time the packed cell volume (PCV) returned to near non-infected levels], the start of the artificial (indoor trickle) challenge began. The challenge comprised bi-weekly doses of 1250 H. contortus L3 until the PCV dropped to a group mean of about 21%, which was between 5 and 7 weeks after the start of the challenge, depending on the lamb crop. The average age of animals at the start of the artificial challenge was 194 days or 6.5 months (with a range of 160 223 days), and the average age of animals at the end of the artificial challenge was 236 days or 8 months (with a range of 199 263 days). Further details of the challenge are given in the study by Mugambi et al. (2005a). Phenotypes Individuals live weight (LWT), fecal egg count (FEC) and PCV were recorded at both the start and end of the artificial challenge period. Measurements at the end of the artificial challenge period were taken on two consecutive days and averaged for further analysis. FEC was determined using the modified McMaster method, and PCV by a micro-hematocrit method. Necropsies were performed immediately after the end of the artificial challenge and lasted 3 to 4 weeks per lamb crop. The abomasums were removed and washed, and these samples were used to determine total, adult and immature worm counts (WC_total, WC_adult and WC_imm respectively), adult female worm length (AFWL) and number of eggs per adult female worm (EPW). See Mugambi et al. (2005a,b) for more detail. Nine traits were analyzed in the QTL mapping experiment, as given in Table 1. Data for FEC, WC_total, WC_imm and EPW were not normally distributed, and thus, a Box Cox transformation was applied, as per Silva et al. (2007, 2012). Data for all traits were adjusted by fitting a fixed-effects model with sire, lamb crop, dam-breed, sex, birth type (single vs. multiple), age of the animal at measurement [as a linear covariate, and excepting average daily gain (ADG)], where significant, as well as any significant interactions between these fixed effects. Solutions were estimated using the GLM procedure of SAS (version 9.2). A few individuals (numbering zero to 13, depending on the trait) were removed from a particular trait analysis because their corrected phenotype did not fall within 3.3 standard deviations of the mean. The final number of records per trait ranged from 937 to 1023, with the number of records per family per trait ranging from 140 to 184. Summary statistics are given in Table 2. Differences between the trait means for the two backcross types were small, although significant (at P < 0.001) with the exception of WC_imm. Interestingly, although FEC, the worm counts, AFWL and EPW were lower for the Red Maasai backcross in comparison with the Dorper backcross, PCV decline (PCVD) was not. For the QTL mapping analysis, standardized residuals were used.

QTL for nematode resistance in East African sheep 3 Table 1 Description of traits analyzed in the QTL mapping experiment. Abbreviation Name in full (units) Description Comment 3 LWT Live weight (kg) Live weight of animals at the end of the indoor challenge 2 ADG Average daily gain (g) Last LWT less the live weight of animals at the start of the indoor challenge, divided by the number of days between the weight measurements, multiplied by 1000 PCVD Packed cell volume decline Packed cell volume (%) at the start of the indoor challenge, less packed cell volume (%) at the end of the indoor challenge 2 FEC Fecal egg count Fecal egg count at the end of the indoor challenge 2, Box Cox-transformed WC_total Worm count total Total worm count at necropsy, Box Cox-transformed WC_adult Worm count adult Count of adult worms, at necropsy WC_imm Worm count immature Count of immature worms at necropsy, Box Cox-transformed AFWL Adult female worm length Length of adult female worms at necropsy (mm) EPW Eggs per worm Eggs per female worm at necropsy, Box Cox-transformed Indicators of resilience: the lesser the effect of the challenge on LWT and ADG, the greater the resilience An indicator of the impact of infection (anemia): the lower the PCVD in response to challenge, the lower the impact An indicator of resistance: the lower the FEC in response to challenge, the greater the resistance Indicators of resistance: the lower the values in response to challenge, the greater the resistance 1 QTL, quantitative trait loci. 2 The average of two measurements taken over two consecutive days. 3 See Bishop (1999, 2012); Bishop & Stear (2003); Stear et al. (1997). Table 2 Summary statistics of transformed (where required) and corrected phenotypes by backcross type and for the combined population 1. Backcross to Dorper Backcross to Red Maasai Combined population Trait (units) Number of records Mean (SD) Number of records Mean (SD) Number of records Mean (SD) LWT (kg) 486 20.84 (2.61) 537 17.48 (1.65) 1023 19.05 (4.11) ADG (g) 484 19.73 (7.92) 532 11.70 (7.01) 1016 15.52 (27.94) PCVD 488 10.04 (2.82) 535 10.94 (2.64) 1023 10.51 (4.75) FEC 2 489 40.78 (4.86) 533 39.80 (4.47) 1022 40.27 (8.89) WC_total 2 477 234.03 (26.12) 525 219.01 (34.94) 1022 226.16 (72.41) WC_adult 477 3506.78 (490.33) 525 3114.43 (485.14) 1002 3301.20 (1549.06) WC_imm 2 441 13.66 (2.00) 496 13.62 (1.87) 937 13.64 (4.73) AFWL (mm) 481 23.79 (0.64) 524 23.39 (0.63) 1005 23.58 (1.66) EPW 2 485 82.84 (9.85) 534 80.44 (9.35) 1019 81.58 (19.22) 1 Trait means for the Dorper and Red Maasai backcrosses were significantly different at P < 0.001, with the exception of WC_imm for which P = 0.79 (comparisons made using the t-test procedure of SAS version 9.2). ADG, average daily gain; AFWL, adult female worm length; EPW, eggs per worm; FEC, fecal egg count; LWT, live weight; PCVD, packed cell volume decline; WC, worm count. 2 Box Cox-transformed. Genotyping and linkage map Animals were selectively genotyped based on being classified as the 10% most resistant or susceptible according to distributions of FEC, PCV and PCVD from the natural pasture challenge, as fully described in the study by Silva et al. (2012). A total of 342 genotyped animals were included in this analysis, with 46 to 72 genotyped animals per sire. The correlations between the natural challenge traits used to select animals for genotyping and the artificial challenge traits analyzed here ranged from 0.22 to 0.29. These relatively low correlations mean that the selectively genotyped animals do not necessarily fall into the extremes of the artificial challenge trait distributions. For example, of the 103 animals each falling into the low and high 10% of the FEC distribution in this study (the artificial challenge), 39 and 47 animals respectively were genotyped. This translates to less power for QTL mapping for the artificial

4 Marshall et al. challenge traits in comparison with the natural challenge traits. Animals were genotyped for 172 ovine microsatellite markers selected from the International Mapping Flock (IMF) sheep genetic map (http://rubens.its.unimelb.edu.au/ ~jillm/jill.htm). Procedures relating to marker genotyping, checking and linkage map construction are given in the study by Silva et al. (2012). For QTL mapping analysis, the marker order and sex-averaged positions according to the IMF sheep genetic map V4.7 were utilized, as shown in Fig. S1 of Silva et al. (2012). Markers covered 25 of the 26 autosomes (there were no markers on chromosome 24), with an average spacing of 13.6 cm, although there was a large (56 cm) gap on chromosome 7. Marker information content was variable across the genome. QTL mapping models Four models were analyzed for each linkage group (chromosome). These were QTL alleles assumed fixed within breed, either with additive (Fixed Add ) or both additive and dominant (Fixed Add,Dom ) QTL effects fitted; and QTL alleles assumed segregating within breed, again either with additive (Segregating Add ) or both additive and dominant (Segregating Add, Dom) QTL effects fitted. Fitting these models is possible because of the design of the resource population. A backcross design allows for the testing of dominance values of QTL. The presence of more than one family allows QTL 9 sire interactions to be fitted; a significant interaction term indicates heterogeneity of QTL effects between sires for which the most likely explanation is that not all F1 sires are identical heterozygous, and hence, QTL alleles are segregating in one or both parent breeds. The reasons for including the Segregating models are given in the discussion. Whole-genome linkage analysis Whole-genome linkage analysis was undertaken using the Backcross/F2 (BCF2) analysis module of GridQTL (Seaton et al. 2006). This module performs interval mapping (Lander & Botstein 1989) based on the Haley Knott regression approach (Haley & Knott 1992; Haley et al. 1994) and uses the F-value as the test statistic. Parameterization of the BCF2 module was as follows. A single QTL (with additive only or additive plus dominance effects) was fitted for each chromosome, with QTL by sire interactions additionally fitted for the models Segregating Add and Segregating Add,Dom. To ensure comparable degrees of freedom between the Segregating and Fixed models, a sire effect (but no interaction) was fitted in the Fixed models. A backcross mean was fitted in all cases. The regression analysis was performed at 1 cm intervals. Empirical significance thresholds were derived from chromosome-wide (CW) permutation testing utilizing 1000 permutations, and 95% confidence intervals of the QTL locations were determined using the bootstrap method with 1000 iterations. It should be noted that the determination of confidence intervals via bootstrapping will result in broader intervals in comparison with the use of the logarithm of odds drop-off method (Visscher et al. 1996). The false discovery rate (FDR) was calculated for each of the four QTL mapping models as the number of QTL expected to be identified by chance divided by the number of significant QTL. Additive and dominance effects of the QTL are reported in phenotypic standard deviation units (r P ). A negative additive effect indicates that the Red Maasai allele decreases the trait value, whereas a positive additive effect indicates that the Dorper allele decreases the trait value. If both the dominance and additive effects are of the same sign, the Red Maasai allele is dominant over the Dorper allele; however, if the dominance and additive effects are of different signs, the Dorper allele is dominant over the Red Maasai allele. The number of animals in the final analysis (i.e. with both genotypes and phenotypes) was 342 for LWT, 336 for ADG, 341 for PCVD, 340 for FEC, 336 for WC_total, 337 for WC_adult, 312 for WC_imm, 336 for AFWL and 341 for EPW. Significance testing of the different models Fixed Add vs. Fixed Add,Dom For the models Fixed Add and Fixed Add,Dom, the reported QTL were identified as follows. First, for each trait by chromosome combination, it was determined whether one of the Fixed Add or Fixed Add,Dom analyses resulted in a test statistic (F-value) exceeding that for the 1% CW threshold as determined by permutation testing. When one or both tests exceeded the threshold, an F-test (see below) was then used to determine whether the model Fixed Add,Dom was statistically more significant than the model Fixed Add, and the appropriate model reported (i.e. Fixed Add unless Fixed Add, Dom was statistically more significant). QTL identified in an analogous manner but for the 5% CW level are reported in Table S1. The best-fitting model (most likely QTL position) for Fixed Add,Dom was tested against the best-fitting model (most likely QTL position) for Fixed Add using an F-test. Denoting the residual sum of squares and degrees of freedom of Fixed Add as RSS1 and DF1 respectively, and that of Fixed Add,Dom as RSS2 and DF2 respectively, an F-ratio was calculated as [(RSS1 RSS2)/(DF1 DF2)]/[RSS2/ DF2] with DF1 DF2 degrees of freedom in the numerator and DF2 degrees of freedom in the denominator. The Fixed Add,Dom model was accepted if it was significantly better than Fixed Add at P < 0.05. Note that permutation testing is not required because this test is performed only once for each QTL which has already been deemed statistically significant.

QTL for nematode resistance in East African sheep 5 Segregating Add vs. Segregating Add,Dom The QTL reported here for the models Segregating Add and Segregating Add,Dom were identified in an analogous manner to the above. Table S2 gives QTL identified using a 5% CW significance level. Fixed Add vs. Segregating Add For each QTL significant at the 1% CW level for Fixed Add and Segregating Add,anF-test was performed to determine whether the Segregating Add model was statistically more significant than the Fixed Add model. This was performed in an analogous manner to the above. If QTL from the Fixed and Segregating models were detected on the same chromosome at a distance of more than 5 cm apart, they were considered separate QTL and two F-tests performed. However, if QTL from the two models were detected on the same chromosome within 5 cm of each other, they were considered the same QTL and a single F-test performed (using the location of the QTL from the Fixed model: little difference was obtained if the QTL location from the Segregating model was used). Results Trait correlations The correlations between the standardized residuals of the traits included in the QTL mapping analysis are given in Table 3. Note that FEC and PCVD were positively correlated (0.64), and both were negatively correlated with LWT and ADG and positively correlated with the worm-related traits. The correlation between FEC and LWT of 0.23 is similar to that reported in a number of other studies (reviewed in Bishop & Stear 2003). In relation to the QTL mapping analysis, these correlations suggest that some pleiotropic QTL effects (i.e. those with an effect on more than one trait) could be expected, some of which may be biologically real, whereas others may arise as statistical artifacts of the correlation between phenotypes. Whole-genome linkage analysis: QTL alleles assumed fixed within breeds Putative QTL, significant at the 1% CW level, for the models Fixed Add and Fixed Add,Dom are presented in Table 4. Eleven putative QTL are reported in total, over three chromosomes and for all traits bar ADG and WC_imm. The trait with the highest number of putative QTL (n = 3) was FEC. In three cases (27% of the total), the model with additive and dominance QTL effects (Fixed Add,Dom ) was more significant than the model with additive effects only (Fixed Add ). Absolute QTL allelic substitution effects (a) ranged from 0.29 to 0.50 (average 0.43) r P and absolute dominance effects (d) from 0.37 to 0.67 (average 0.50) r P. The use of Haley Knott regression with selective genotyping (as was done here) tends to overestimate effect sizes for selected traits and their correlates (Feenstra et al. 2006), but the correlation of all traits with the natural challenge traits used for selective genotyping was low, so any bias should be small. Favorable alleles (considered as that which result in lower FEC, PCVD and values for the worm traits, but higher LWT, following infection in comparison with the alternate allele) were contributed from the Red Maasai in three cases (the three chromosome 23 QTL) and from the Dorper in eight cases (the two chromosome 2 QTL and the six chromosome 26 QTL). The Red Maasai allele (which was unfavorable) was dominant over the Dorper allele for the three putative QTL with dominance effects, and overdominance (where d is greater than a) was suggested in two cases. Considering the two models individually, the FDR for Fixed Add and Fixed Add,Dom was 0.28 and 0.32 respectively (as there were eight putative QTL at the 1% CW level for Fixed Add and 7 for Fixed Add,Dom ; only 11 QTL are reported in Table 4 because both models were significant in four cases). Whole-genome linkage analysis: QTL alleles assumed segregating within breeds For the models Segregating Add and Segregating Add,Dom,19 putative QTL were identified over nine chromosomes and for all traits but LWT, AFWL and EPW (Table 5). The trait Table 3 Correlation coefficients between the standardized residuals of transformed (where required) and corrected trait values. LWT ADG PCVD FEC WC_total WC_adult WC_imm AFWL ADG 0.30 PCVD 0.40 0.25 FEC 0.23 0.13 0.64 WC_total 0.20 0.11 0.49 0.55 WC_adult 0.11 0.07 0.44 0.50 0.95 WC_imm 0.32 0.16 0.38 0.36 0.57 0.35 AFWL 0.12 0.06 0.47 0.51 0.46 0.44 0.27 EPW 0.07 0.03 0.25 0.30 0.32 0.30 0.20 0.49 ADG, average daily gain; AFWL, adult female worm length; EPW, eggs per worm; FEC, fecal egg count; LWT, live weight; PCVD, packed cell volume decline; WC, worm count.

6 Marshall et al. Table 4 Putative QTL, significant at the 1% chromosome-wide level, from models in which the QTL alleles were assumed to be fixed within breed 1. Chr cm (CI) Flanking markers 2 Trait Model F F 1% CW a (SE) d (SE) 2 263 (66 289) INHA BM2113 FEC Fixed Add,Dom 7.55 7.44 0.49 (0.18) 0.47 (0.18) 2 261 (57 289) INHA BM2113 PCVD Fixed Add,Dom 9.87 7.63 0.37 (0.17) 0.67 (0.17) 23 18 (8 50) BMS2526 BMS2270 FEC Fixed Add 11.84 10.24 0.50 (0.15) 23 48 (20 50) MAF35 UW69A WC_total Fixed Add 12.70 9.08 0.44 (0.12) 23 46 (8 49) MAF35 UW69A WC_adult Fixed Add 10.96 8.24 0.39 (0.12) 26 39 (35 51) CSRD163 LWT Fixed Add 13.77 10.05 0.46 (0.12) 26 39 (0 54) CSRD163 PCVD Fixed Add 12.33 10.10 0.44 (0.13) 26 39 (29 40) CSRD163 FEC Fixed Add 15.49 9.76 0.49 (0.12) 26 39 (29 58) CSRD163 WC_total Fixed Add 11.01 10.26 0.39 (0.12) 26 39 (29 58) CSRD163 AFWL Fixed Add 11.74 8.96 0.43 (0.12) 26 36 (21 71) LS41 CSRD163 EPW Fixed Add,Dom 7.38 6.19 0.28 (0.12) 0.37 (0.12) 1 Given is the chromosome (chr), most likely position in cm and the confidence interval surrounding that position (CI); F is the F-statistic at the most likely QTL position and F 1% CW is the F-statistic for the 1% chromosome-wide threshold; a and d are the allelic substitution effect and dominance effect respectively, given with their standard errors (SE). AFWL, adult female worm length; CW, chromosome-wide; EPW, eggs per worm; FEC, fecal egg count; LWT, live weight; PCVD, packed cell volume decline; QTL, quantitative trait loci; WC, worm count. 2 A single marker is given when the most likely QTL position is at a marker location. with the highest number of putative QTL was WC_adult (n = 7), followed by FEC (n = 5). In seven cases (37% of the total), the model with additive and dominance QTL effects (Segregating Add,Dom ) was more significant than was the model with additive effects only (Segregating Add ). Over all sires, the absolute values of a ranged from 0.01 to 1.04 (average = 0.39), and the absolute values of d ranged from 0 to 1.80 (average = 0.52) r P. As mentioned above, however, there may be some overestimation of effect size because of the selective genotyping. In 40% of the cases (46/114), the Red Maasai allele was favorable for disease resistance (i.e. decreased the values of FEC, PCVD and the worm traits in comparison with the alternate allele), and in 43% of the cases (18/42), the Red Maasai allele was dominant over the Dorper allele. Further, over-/underdominance (where the magnitude of d is greater than a) was suggested in 62% of the cases (26 of 42). However, it should be noted that if dominance is present, a high frequency of apparent over-/underdominance is expected owing to the random sampling terms of the effect estimates (see SE in Table 5). Considering the models independently, the FDR was 17% for Segregating Add and 28% for Segregating Add, Dom (as there were 13 putative QTL at the 1% CW level for Segregating Add and 8 for Segregating Add,Dom ; only 19 QTL are reported in Table 5 because both models were significant in two cases). Comparison of fixed and segregating QTL mapping models Statistical comparisons were made to determine how often the Segregating Add model was more significant than the Fixed Add model. For the Segregating Add and Fixed Add models respectively, there were 13 and 9 QTL significant at the 1% CW level; however, in six cases, both models detected QTL that were considered the same (same trait and chromosome and within 5 cm of each other), and thus, 16 comparisons were made in total. The Segregating Add model was accepted over the Fixed Add model in eight of the 16 cases (50.0%). These cases were for WC_adult on chromosomes 12, 18, 22, 25 and 26; WC_total on chromosomes 25 and 26; and ADG on chromosome 26. The Segregating Add model was not statistically more significant than Fixed Add for the remaining cases, which were LWT on chromosome 26; PCVD on chromosome 26; FEC on chromosomes 10, 23 and 26; WC_adult and WC_total on chromosome 23; and PCVD on chromosome 26. Overview and chromosomes of particular interest Overall, putative QTL for the weight traits (LWT and ADG) were found on chromosome 26 only; for FEC on chromosomes 2, 4, 10, 23 and 26; for PCVD on chromosomes 2 and 26; for worm counts (WC_total, WC_adult and WC_imm) on chromosomes 2, 12, 18, 22, 23, 25 and 26; and for AFWL and EPG on chromosome 26. The single most significant QTL, as that with the highest value of F/F 1% CW, was for FEC on chromosome 26. Of particular interest is chromosome 26, which had putative QTL for all traits, significant at the 1% CW level, under at least one of the models (see Fig. 1 for QTL mapping profiles for traits significant under Fixed Add ). Also of interest is chromosome 2, which had putative QTL significant at the 1% CW level for PCVD, FEC and WC_adult (see Fig. 2 for QTL mapping profiles for traits significant under Segregating Add,Dom ) and, in addition, putative QTL significant at the 5% CW level for LWT, ADG and WC_total (see Tables S1 & S2). Other regions of note when considering putative QTL significant at the 5% CW level are chromosome 23 (with QTL for LWT, PCVD, FEC, WC_total, WC_adult, AFWL and EPW) and chromosome 22 (with QTL for LWT, ADG, FEC, WC_total, WC_adult and WC_imm), although in both of these cases the most likely QTL locations vary somewhat for the different traits.

QTL for nematode resistance in East African sheep 7 Table 5 Putative QTL, significant at the 1% chromosome-wide level, from models where the QTL alleles were assumed to be segregating within breed 1. Chr cm (CI) Flanking markers 2 Trait Model F F 1% CW a (SE)/d (SE) Sire 1 Sire 2 Sire 3 Sire 4 Sire 5 Sire 6 2 259 (40 288) 2 260 (53 289) 2 259 (53 289) 4 14 (11 112) 10 10 (10 77) 12 95 (8 95) 18 44 (22 76) 22 14 (14 77) 23 18 (8 50) 23 48 (6 50) 23 47 (3 50) 25 28 (28 68) 25 28 (28 68) 26 21 (0 71) 26 38 (0 66) 26 39 (21 40) 26 39 (21 62) 26 38 (17 62) 26 48 (0 71) INHA- BM2113 INHA- BM2113 INHA- BM2113 BMS1788- MCM2 WC_adult Segregating Add,Dom 3.05 2.69 0.63 (0.24)/ 0.80 (0.35) FEC Segregating Add,Dom 3.08 2.80 0.45 (0.27)/ 0.26 (0.40) PCVD Segregating Add,Dom 2.8 2.50 0.30 (0.27)/ 0.69 (0.40) FEC Segregating Add,Dom 2.8 2.35 0.11 (0.25)/ 0.61 (0.36) 0.57 (0.22)/ 0.41 (0.37) 0.70 (0.25)/ 0.51 (0.42) 0.45 (0.25)/ 0.76 (0.41) 0.02 (0.28)/ 0.98 (0.42) 0.31 (0.25)/ 0.29 (0.35) 0.50 (0.28)/ 0.06 (0.38) 0.44 (0.28)/ 0.38 (0.39) 0.11 (0.26)/ 0.07 (0.35) 0.11 (0.25)/ 0.85 (0.42) 0.12 (0.28)/ 1.51 (0.46) 0.05 (0.28)/ 1.80 (0.46) 0.68 (0.30)/ 1.32 (0.56) 0.18 (0.23)/ 0.43 (0.35) 0.75 (0.25)/ 0.73 (0.39) 0.51 (0.25)/ 0.61 (0.39) 0.60 (0.27)/ 0.05 (0.34) 0.58 (0.28)/ 0.66 (0.41) 0.04 (0.32)/ 0.30 (0.46) 0.11 (0.31)/ 0.23 (0.46) 0.11 (0.29)/ 0.98 (0.42) BL1022 FEC Segregating Add 3.12 2.98 0.52 (0.27) 0.10 (0.36) 0.36 (0.27) 1.01 (0.27) 0.22 (0.28) 0.70 (0.35) INRA35 WC_adult Segregating Add 3.03 2.90 0.67 (0.29) 0.28 (0.23) 0.03 (0.23) 0.03 (0.22) 0.21 (0.26) 0.70 (0.27) BP33- BM7243 WC_adult Segregating Add 3.47 2.74 0.48 (0.20) 0.41 (0.24) 0.03 (0.21) 0.15 (0.21) 0.05 (0.22) 0.82 (0.28) BMS907 WC_adult Segregating Add 3.06 2.95 0.48 (0.26) 0.07 (0.20) 0.32 (0.24) 0.30 (0.26) 0.34 (0.24) 0.67 (0.27) BMS2526- BMS2770 MAF35- UW65A MAF35- UW65A FEC Segregating Add 3.27 3.26 0.22 (0.27) 0.82 (0.26) 0.66 (0.25) 0.17 (0.24) 0.73 (0.23) 0.34 (0.34) WC_total Segregating Add 3.59 3.20 0.70 (0.22) 0.53 (0.23) 0.60 (0.21) 0.25 (0.20) 0.41 (0.21) 0.21 (0.31) WC_adult Segregating Add 3.57 3.35 0.68 (0.21) 0.45 (0.22) 0.55 (0.20) 0.25 (0.19) 0.36 (0.20) 0.33 (0.30) BL25 WC_total Segregating Add 2.84 2.74 0.41 (0.26) 0.36 (0.30) 0.21 (0.25) 0.23 (0.29) 0.24 (0.25) 0.91 (0.34) BL25 WC_adult Segregating Add 3.60 3.15 0.59 (0.25) 0.45 (0.29) 0.20 (0.24) 0.09 (0.27) 0.19 (0.24) 0.86 (0.32) LS41- CSRD163 LS41- CSRD163 ADG Segregating Add,Dom 3.09 2.38 0.51 (0.23)/ 1.04 (0.36) PCVD SegregatingAdd,Dom 2.94 2.38 0.20 (0.22)/ 0.71 (0.32) 0.23 (1.04)/ 0.36 ( 0.21) 0.22 ( 0.71)/ 0.32 (0.26) 1.04 (0.36)/ 0.21 (0.22) 0.71 (0.32)/ 0.26 (0.23) 0.36 ( 0.21)/ 0.22 (0.07) 0.32 (0.26)/ 0.23 (0.50) 0.21 (0.22)/ 0.07 (0.30) 0.26 (0.23)/ 0.50 (0.31) 0.22 (0.07)/ 0.30 ( 0.48) 0.23 (0.50)/ 0.31 (0.47) CSRD163 FEC Segregating Add 4.71 3.49 0.18 (0.22) 0.28 (0.23) 0.41 (0.22) 1.01 (0.20) 0.38 (0.21) 0.59 (0.26) CSRD163 WC_total Segregating Add 4.53 3.54 0.11 (0.21) 0.22 (0.21) 0.39 (0.21) 0.83 (0.19) 0.36 (0.20) 0.61 (0.24) LS41- CSRD163 CSSM43- JMP58 WC_adult Segregating Add 4.18 3.06 0.16 (0.20) 0.13 (0.21) 0.41 (0.21) 0.71 (0.19) 0.38 (0.20) 0.62 (0.23) WC_imm Segregating Add,Dom 2.17 2.14 0.08 (0.22)/ 0.40 (0.31) 0.36 (0.25)/ 0 (0.31) 0.24 (0.24)/ 0.67 (0.31) 0.70 (0.22)/ 0.78 (0.31) 0.01 (0.23)/ 0.33 (0.32) 0.21 (0.27)/ 0.13 (0.40) 1 Given is the chromosome (chr), most likely position in cm and the confidence interval surrounding that position (CI); F is the F-statistics at the most likely QTL position and F 1% CW is the F-statistic for the 1% chromosome-wide threshold; a and d are the allelic substitution effect and dominance effect respectively, given with their standard errors (SE). ADG, average daily gain; FEC, fecal egg count; PCVD, packed cell volume decline; QTL, quantitative trait loci; WC, worm count. 2 A single marker is given when the most likely QTL position is at a marker location.

8 Marshall et al. F-statistic 20 15 10 5 0 0 10 20 30 40 50 60 70 cm Discussion LWT PCVD FEC WC_total AFWL EPW 1%CW LWT 1%CW PCVD 1%CW FEC 1%CW WC_total 1%CW AFWL 1%CW EPW Figure 1 Quantitative trait loci (QTL) mapping profiles for chromosome 26 under the model Fixed Add. traits with QTL significant at the 1% chromosome-wide (CW) level are shown. F-statistic 4 3 2 1 0 50 100 150 200 250 300 cm PCVD FEC WC_adult 1%CW PCVD 1%CW FEC 1%CW WC_adult Figure 2 Quantitative trait loci (QTL) mapping profiles for a region of chromosome 2 under the model Segregating Add,Dom. traits with QTL significant at the 1% chromosome-wide (CW) level are shown. Quantitative trait loci for indicators of resistance to H. contortus were identified in a double backcross of Red Maasai by Dorper sheep under various QTL mapping models. We hypothesized that some QTL might be segregating in one or both parent breeds given the small difference in trait means between the two backcross types, because population genetics theory does not predict that all QTL will be at fixation in a given population and because the Dorper is a synthetic of two breeds with widely different origin (Dorset and Persian). The results supported this hypothesis with 11 of 19 QTL detected by the Segregating model but not detected by the Fixed model. Further, the Segregating model was accepted as being statistically significantly better than the Fixed model in half of the tested cases. This indicates that previous published studies involving crosses, and for which the possibility of QTL segregating within breeds was not tested, may have missed some QTL. Additionally, we compared models fitting additive and dominant QTL effects to models fitting additive QTL effects only, with the models including dominant effects accepted over models without in about one-third of the cases. Favorable alleles for disease resistance appeared to be a mixture of both Red Maasai and Dorper origin. This contrasts to the results of the pasture challenge of the same animal population where favorable QTL alleles for the disease resistance indicator traits were always contributed from the Red Maasai (Silva et al. 2012). At least one other study has found favorable alleles for disease resistance originating from both the susceptible and resistant breeds (Hanotte et al. 2002). The main result of this study is the detection of two chromosomal regions, located on chromosomes 2 and 26, where there is strong evidence for one or more QTL with effects on multiple disease resistance indicator traits. This outcome may be due to a single pleiotropic QTL or due to several linked QTL. No QTL identified in this work overlapped in terms of location to that reported for the primary pasture challenge (Silva et al. 2012) barring QTL for FEC on chromosomes 6 and 22 (significant at the 5% CW level) where confidence intervals overlap. This may be attributable to age and/or immune status specificity of the QTL, the different parasite exposure, or biological differences between field and artificial challenges. This result is similar to those of other studies where QTL for FEC following primary and secondary challenges of sheep with gastrointestinal nematodes generally did not correspond (Marshall et al. 2009; Dominik et al. 2010). Across all studies on QTL for resistance to gastrointestinal nematodes in sheep (Schwaiger et al. 1995; Paterson et al. 1999; Coltman et al. 2001; Beh et al. 2002; Charon et al. 2002; Janssen et al. 2002; Janben et al. 2004; Crawford et al. 2006; Davies et al. 2006; Moreno et al. 2006; Beraldi et al. 2007; Benavides et al. 2009; Gutiérrez-Gil et al. 2009; Marshall et al. 2009; Dominik et al. 2010; Silva et al. 2012; this work), evidence for QTL has been found on all autosomes as well as on the X chromosome, including specifically for artificial challenge with H. contortus. It is of note that the confidence interval of the chromosome 2 region identified in this study (40 289 cm) overlaps with a FEC QTL at 225 cm reported by Moreno et al. (2006), as well as an 80- to 180-cM region reported in several other studies (Crawford et al. 2006; Davies et al. 2006; Beraldi et al. 2007; Marshall et al. 2009). Further, the chromosome 26 region identified in this study (with a most likely position at 39 cm between markers LS41 and CSRD163) matches closely with a FEC QTL for secondary artificial challenge (at marker LS41) reported in the study by Marshall et al. (2009). This result is encouraging and suggests that these regions are promising candidates for further fine mapping.

QTL for nematode resistance in East African sheep 9 Several studies have suggested that a mutation in or near the interferon gamma (IFNG) and DRB genes is associated with gastrointestinal nematode resistance in sheep (Schwaiger et al. 1995; Crawford & McEwen 1998; Coltman et al. 2001; Beh et al. 2002; Charon et al. 2002; Davies et al. 2006). Both the IFNG (a cytokine) and DRB gene (belonging to class II of the major histocompatibility complex) are known for their role in the regulation of the immune response to parasite infection (Urban et al. 1996; Wakelin, 1996). This study found no evidence of significant linkage to either of these, in accordance with the study by Silva et al. (2012). This result may be due to the lack of QTL segregating in these regions or insufficient power of the current sample size. Despite the large number of QTL mapping studies for disease resistance in sheep to date, there have been few individual genes identified (with IFNG and DRB being exceptions). In recent work, however, Sayre & Harris (2012) used a systems genetics approach to detect common pathways associated with genetic resistance to internal parasites. A list of 31 genes of interest was reported, for which we ascertained the likely locations of 25 on the map used in this study (via identifying adjacent markers using the sheep genome browser http://www.livestockgenomics. csiro.au/gi-bin/gbrowse/oarv2.0/). Overlap between these locations and QTL identified in this study (when considering confidence intervals) was limited to the chemokine (C X C motif) receptor 4 gene (at 214 222 cm of chromosome 2) and the mannan-binding lectin serine peptidase 2 gene (at 57 63 cm of chromosome 12). Confirmation that mutations in these genes are contributing to variation in disease resistance would require sequencing of both Red Maasai and Dorper animals for these genes, and this may be warranted. Control measures used by smallholders in East Africa for nematode infections in sheep include the use of commercial anthelmintics (for which resistance is common), plant preparations such as dewormers, pasture rotation and genetically resistant breeds and their crosses (Mwamachi et al. 1995; Wanyangu et al. 1996; Maingi et al. 1997; Waller 1997; Gatongi et al. 1998; Keyyu et al. 2002; Athanasiadou et al. 2007; Gakuya et al. 2007). However, these strategies are not universally applied, and productivity losses to nematode infections remain high. Given the difficulty of implementing sustainable breeding programmes within smallholder production systems in developing countries, particularly those using high-end technologies such as marker-assisted or genomic selection (Marshall et al. 2011), we do not envisage using the results of this study for this purpose in the near future. Instead, we anticipate that the results of this study and planned future work, including single-nucleotide polymorphism based genome-wide association studies on the same animal population, will help to elucidate the biology of disease resistance. If this leads to the development of new animal healthcare products, such as a vaccine against gastrointestinal nematodes, a major constraint to sheep productivity in both the developed and developing world could be overcome. Acknowledgements This project was supported by the 2004 USAID linkages program and by project 1265-31000-084D (BFGL) from the USDA Agricultural Research Service. We thank Alicia Beavers (USDA, BFGL) for processing of genotyping reactions. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. The USDA is an equal opportunity provider and employer. References Athanasiadou S., Githori J. & Kyriazakis I. (2007) Medicinal plants for helminth parasite control: facts and fiction. Animal 1, 1392 400. Baker R.L. & Gray G.D. (2004) Appropriate breeds and breeding schemes for sheep and goats in the tropics. In: Worm Control for Small Ruminants in Tropical Asia (Ed. by R.A. Sani, G.D. Gray & R. L. Baker), pp. 63 95. ACIAR Monograph 113, Canberra, Australia. Baker R.L., Nagda S., Rodriguez-Zas S.L., Southey B.R., Audho J.O., Aduda E.O. & Thorpe W. (2003) Resistance and resilience to gastro-intestinal nematode parasites and relationships with productivity of Red Maasai, Dorper and Red Maasai 9 Dorper crossbred lambs in the sub-humid tropics. Animal Science 76, 119 36. Baker R.L., Mugambi J.M., Audho J.O., Carles A.B. & Thorpe W. (2004) Genotype by environment interactions for productivity and resistance to gastro-intestinal nematode parasites in Red Maasai and Dorper sheep. Animal Science 79, 343 53. Beh K.J., Hulme D.J., Callaghan M.J., Leish Z., Lenane I., Windon R. G. & Maddox J.F. (2002) A genome scan for quantitative trait loci affecting resistance to Trichostrongylus colubriformis in sheep. Animal Genetics 33, 97 106. Benavides M.V., Weimer T.A., Borba M.F.S., Berne M.E.A. & Sacco A.M.S. (2009) Genetic analysis of polymorphisms on ovine chromosomes 5 and 20 and their effect on resistance to internal parasites. Small Ruminant Research 83, 67 73. Beraldi D., McRae A.F., Gratten J., Pilkington J.G., Slate J., Visscher P.M. & Pemberton J.M. (2007) Quantitative trait loci (QTL) mapping of resistance to strongyles and coccidia in the free-living Soay sheep (Ovis aries). International Journal for Parasitology 37, 121 9. Besier B. (2007) New anthelmintics for livestock: the time is right. Trends in Parasitology 23, 21 4. Bishop S.C. (1999) Interpretation and prediction in disease genetics. Proceedings of the Association for the Advancement of Animal Breeding and Genetics 18, 255 62. Bishop S.C. (2012) Possibilities to breed for resistance to nematode parasite infections in small ruminants in tropical production systems. Animal 6, 741 7. Bishop S.C. & Morris C.A. (2007) Genetics of disease resistance in sheep and goats. Small Ruminant Research 70, 48 59.

10 Marshall et al. Bishop S.C. & Stear M.J. (2003) Modeling of host genetics and resistance to infectious diseases: understanding and controlling nematode infections. Veterinary Parasitology 115, 147 66. Charon K.M., Moskwa B., Rutkowski R., Gruszczynska J. & Swiderek W. (2002) Microsatellite polymorphism in DRB1 gene (MHC class II) and its relation to nematode faecal egg count in Polish Heath Sheep. Journal of Animal and Feed Sciences 11, 47 58. Coltman D.W., Wilson K., Pilkington J.G., Stear M.J. & Pemberton J.M. (2001) A microsatellite polymorphism in the gamma interferon gene is associated withresistance to gastrointestinal nematodes ina naturally parasitized population of Soay sheep. Parasitology 122, 571 82. Crawford A.M. & McEwen J.C. (1998) Identification of Animals Resistant to Nematode Parasite Infection. New Zealand Provisional Patent 330201, New Zealand. Crawford A.M., Paterson K.A., Dodds K.G. et al. (2006) Discovery of quantitative trait loci for resistance to parasitic nematode infection in sheep: I. Analysis of outcross pedigrees. BMC Genomics 7, 178 88. Davies G., Stear M.J., Benothman M., Abuagob O., Kerr A., Mitchell S. & Bishop S.C. (2006) Quantitative trait loci associated with parasitic infection in Scottish blackface sheep. Heredity 96, 252 8. Dominik S., Hunt P.W., McNally J., Murrell A., Hall A. & Purvis I.W. (2010) Detection of quantitative trait loci for internal parasite resistance in sheep. I. Linkage analysis in a Romney 9 Merino sheep backcross population. Parasitology 137, 1275 82. Feenstra B., Skovgaard I.M. & Broman K.W. (2006) Mapping quantitative trait loci by an extension of the Haley Knott regression method using estimating equations. Genetics 173, 2269 82. Gakuya D.W., Nganga C.J., Karanja D.N., Wabacha J.K. & Mutune M.N. (2007) Multiple anthelmintic resistance on a sheep farm in Kenya and its implication for Helminth control. Bulletin of Animal Health and Production in Africa 55, 226 30. Gatongi P.M., Prichard R.K., Ranjan S., Gathuma J.M., Munyua W.K., Cheruiyot H. & Scott M.E. (1998) Hypobiosis of Haemonchus contortus in natural infections of sheep and goats in a semiarid area of Kenya. Veterinary Parasitology 77, 49 61. Gutiérrez-Gil B., Pérez J., Álvarez L., Martínez-Valladares M., de la Fuente L.-F., Bayón Y., Meana A., San Primitivo F., Rojo-Vázquez F.-A. & Arranz J.-J. (2009) Quantitative trait loci for resistance to trichostrongylid infection in Spanish Churra sheep. Genetics, Selection, Evolution 41, 46 56. Haley C.S. & Knott S.A. (1992) A simple regression method for mapping quantitative trait loci in line crosses using flanking markers. Heredity 69, 315 24. Haley C.S., Knott S.A. & Elsen J.M. (1994) Mapping quantitative trait loci in crosses between outbred lines using least squares. Genetics 136, 1195 207. Hanotte O., Ronin Y., Agaba M. et al. (2002) Mapping of quantitative trait loci controlling trypanotolerance in a cross of tolerant West African N dama and susceptible East African Boran cattle. Proceedings of the National Academy of Sciences 100, 7443 8. Janben M., Weimann C., Brandt H., Gauly M. & Erhardt G. (2004) Parasitological parameters after artificial infections with Haemonchus contortus in Merinoland sheep and its association to genetic markers on chromosome 20. Archives of Animal Breeding 47, 36 42. Janssen M., Weimann C., Gauly M. & Erhardt G. (2002) Associations between infections with Haemonchus contortus and genetic markers on ovine chromosome 20. Proceedings of the 7th World Congress on Genetics Applied to Livestock Production: CD-ROM Communication No. 28-09. Montpellier, France. Keyyu J.D., Mahingika H.M., Magwisha H.B. & Kassuku K.A. (2002) Efficacy of albendazole and levamisole against gastrointestinal nematodes of sheep and goats in Morogoro, Tanzania. Tropical Animal Health and Production 34, 115 20. Lander E.S. & Botstein D. (1989) Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121, 185 99. Maingi N., Bjørn H., Thamsborg S.M., Munyua W.K., Gathumat J.M. & Dangolla A. (1997) Worm control practices on sheep farms in Nyandarua District of Kenya. Acta Tropica 68, 1 9. Marshall K., Maddox J.F., Lee S.H., Zhang Y., Kahn L., Graser H.-U., Gondro C., Walkden-Brown S.W. & van der Werf J.H.J. (2009) Genetic mapping of quantitative trait loci for resistance to Haemonchus contortus in sheep. Animal Genetics 40, 262 72. Marshall K., Quiros-Campos C., Van der Werf J.H.J. & Kinghorn B. (2011) Marker-based selection within smallholder production systems in developing countries. Livestock Science 136, 45 54. Milne C. (2000) The history of the Dorper sheep. Small Ruminant Research 36, 99 102. Moreno C.R., Gruner L., Scala A. et al. (2006) QTL for natural resistance to internal parasites in two designs based on natural and experimental conditions of infection. Proceedings of the 8th World Congress of Genetics Applied to Livestock Production, Belo Horizonte, MG, Brazil. Mugambi J.M., Wanyangu S.W., Bain R.K., Owango M.O., Duncan J.L. & Stear M.J. (1996) Response of Dorper and Red Maasai lambs to trickle Haemonchus contortus infections. Research in Veterinary Science 61, 218 21. Mugambi J.M., Bain R.K., Wanyangu S.W., Ihiga M.A., Duncan J.L., Murray M. & Stear M.J. (1997) Resistance of four sheep breeds to natural and subsequent artificial Haemonchus contortus infection. Veterinary Parasitology 69, 265 73. Mugambi J.M., Audho J.O., Njomo S. & Baker R.L. (2005a) Evaluation of phenotypic performance of a Red Maasai and Dorper double backcross resource population: indoor trickle challenge with Haemonchus contortus. Veterinary Parasitology 127, 263 75. Mugambi J.M., Audho J.O., Njomo S. & Baker R.L. (2005b) Evaluation of phenotypic performance of a Red Maasai and Dorper double backcross resource population: natural pasture challenge with gastro-intestinal nematode parasites. Small Ruminant Research 56, 239 51. Mwamachi D.M., Audho J.O., Thorpe W. & 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 13. Over H.J., Jansen J. & Von Olm P.W. (1992) Distribution and Impact of Helminth Diseases of Livestock in Developing Countries. FAO Animal Production and Health Paper 96. FAO (Food and Agriculture Organization of the United Nations), Rome, Italy, 221. Paterson K.A., McEwan J.C., Dodds K.G., Morris C.A. & Crawford A. M. (1999) Fine mapping a locus affecting host resistance to internal parasites in sheep. Proceedings of the Association for the Advancement of Animal Breeding and Genetics 13, 91 4.