ILRI’s Institutional Animal Care and Use Committee (IACUC) is mandated to review all experimental procedures and experiments on animals within ILRI’s watch and approve or disapprove. After the committee considered ethical issues and physical procedures associated with the housing, feeding, experimentation, and all other routine matters pertaining to the normal welfare of the animals, it approved the experimental research reported here through written permission.
The field study site was located in Narok District, Southwest Kenya between 10.00’S and 10005’S, and 35005'E and 35015’E (Figure ). The altitude of the area ranges from 1600 to 2130 m above sea level. The study area is classified as semi-arid, with annual rainfall of 750–1250 mm. The rain comes in two seasons: the ‘long’ (main) rains in March to May and the ‘short’ rains in November to December, with two dry periods in January to February and June to October. The study area is primarily savannah grassland, but has four distinct types of vegetation patterns: open grassland, wooded grassland, dense thickets and African acacia genus. There are three tsetse species in the study area: Glossina swynnertoni, G. pallidipes, G. fuscepleuris, but the first two species are predominant. The site is typical of the tsetse infested arid and semi-arid areas of Kenya and was selected on the basis of previous observations of high trypanosomosis prevalence in cattle during field surveys by the Kenya Trypanosomiasis Research Institute (KETRI, Kikuyu, Kenya). The presence of abundant wildlife is an important feature of the study area, being part of the 25,000 sq. km rangelands forming the Mara-Serengeti ecosystem.
A map of Kenya showing Narok field study site (the green area shows the mara-serengeti ecosystem).
Breeding and management of the backcross population
In the course of the Hanotte et al
] QTL mapping study, a frozen bank of semen from 13 F1 (ND x KB) males was established at the International Livestock Research Institute (ILRI, Nairobi, Kenya). The population from which these F1 sires were derived was constructed by single pair matings of 4 ND males (ND7, ND8, ND9 and ND10) and 4 KB females (KB1419, KB1688, KB1801 and KB2094) resulting in 13 F1 (ND ×
males (henceforth: F1 sires) belonging to four full-sib families according to their ND sire and KB dam. In the present study, semen of these 13 F1 males was used to produce 192 backcross (BCB) animals through insemination of about 200 KB females at the ILRI Kapiti ranch (a tsetse free zone), over the period of June 2001 to December 2006. Doses of semen from individual sires were used in rotation so that each successive dose of semen was from a different sire, in order to maximise equality of sire representation over time. As the matings proceeded and information on successful pregnancies became available, semen usage patterns were adjusted to maintain, as much as possible, equal representation of sires among live-born progeny. At the final count, the four F1 sire-families were represented by 42, 33, 65 and 52 BCB progeny, respectively (total: 88 males and 104 females, Table ).
Batch composition by sex and genetic type and duration of tsetse challenge period
In creating the animal population for this experiment, artificial insemination was used. Estrus detection was by experienced herdsmen who in turn reported to the inseminators. Due to cost, estrus synchronization was not implemented. However, possibly due to inadequacies in heat detections, numerous repeat inseminations were required before some of the cows conceived, leading to overall low conception rates. Consequently, the full planned numbers of experimental animals could not be achieved in a single season. This in turn led to exposing the animals to tsetse and trypanosomosis challenge in the field in batches, which encountered different weather conditions and therefore different tsetse fly intensities and different nutritional conditions. This enabled effect of variation in environment to be evaluated by calculation of Batch effects.
Purebred KB calves born contemporaneously with the BCB calves at Kapiti were included in Batches 4 and 5. In addition, frozen semen from two ND bulls (ND164 and ND162) was used to inseminate purebred KB dams at Kapiti to produce F1 calves, born contemporaneously with the BCB calves of Batch 6. ND164 produced 29 F1s while ND162 produced 10 F1s to make the total of 39 F1s. The KB and F1 calves were reared alongside their contemporaneous BCB calves and under the same management regimes throughout the experiment.
Each calf born in Kapiti was weighed at birth, and subsequently at monthly intervals until transported to the field site. Each calf was assigned a unique ear tag, a tattoo on the ear and a brand burned on the rump that uniquely identified the animal and was associated with pedigree information on the animal.
Progeny were run as suckler calves with their dams on natural pasture without feed supplementation. In accordance with standard ranch management conditions, all cattle were treated with acaricides (Triatix®) every two weeks to control ticks; vaccinated quarterly against Foot and Mouth Disease (FMD); and dewormed both prophylactically and tactically (e.g. in anticipation of a worm outbreak following the rains) until weaning at 8 to 10 months of age. Male calves were not castrated. Weaner groups were transported to the KETRI Muguga ranch before being taken to the KETRI operated challenge pastures at Narok ranch. In Muguga, they were again vaccinated against FMD and also against Contagious Bovine Pleuropneumonia (CBPP) and East Coast Fever (ECF), diseases that are known to be endemic at the Narok ranch, The BCB progeny were moved to the Narok field site in six different batches, with start dates between December 2003 and July 2006, and end dates between March 2005 and June 2007 (Table ). There was considerable overlap between field challenge dates of the different batches. Batches 1, 2, 3 overlapped one another to some extent, as did Batches 3, 4, 5. There was a one-month overlap between Batches 5 and 6. Each Batch was exposed to field tsetse challenge over the course of a year or more, with the exception of Batch 5 that was kept under field challenge conditions for 239 d.
The genetic-type comparison study included 71 BCB cattle (38 males and 33 females) in Batches 4, 5 and 6; 23 purebred KB (15 males and 8 females) in Batches 4 and 5; and 37 F1 (13 males and 24 females) in Batch 6.
Definition of phenotypes and traits
From the raw phenotypes recorded weekly or biweekly for each individual (PCV, PAR, BW and NT) various “traits” were meticulously defined and constructed making a total of 49 phenotypic traits (Table ). Many of the traits included a “time” component, e.g., DF1, days from first exposure to infection. Traits were classified into groups as absolute body weight traits (BW), PCV traits (PC), parasitemia traits (PT), infection cycle traits (IC), treatment-related traits (TT), body-weight-change traits (BWC). Traits thought to be associated with trypanotolerance were classified with respect to direction of effect, i.e., as to whether high trait-value indicated a higher degree of trypanotolerance (H-traits) or low trait-value indicated higher trypanotolerance (L-traits). For example, DF1, days to first infection after transfer to Narok, was classified as an H-trait since a larger number of days to first infection was an indication that the animal was more trypanotolerant than one that became infected after a shorter period. In contrast, MPAR, mean parasitemic score, was classified as an L-trait, since a low mean parasitemic score indicated that the animal was better able to control the parasitemia as compared to an animal with a higher mean parasitemic score; thus indicating higher trypanotolerance. Where neither of these applied, the trait was listed as not relevant to trypanotolerance (NR). NR-traits included: Total weeks sampled (WIC); Initial PCV (PCI) and PCV at first screening (PCSR), which were taken before challenge; and some traits related to body weight (Traits 35–41), such as Initial Body Weight (WT1), which were thought to be primarily determined by loci affecting BW rather than by loci affecting trypanotolerance. For each trait, Table shows the trait number, definition, trait group, acronym, direction of effect (H, L or NR), and mode of calculation.
Traits analyzed and their definitions and mode of calculation
Most of the defined traits were based on data obtained in the first two infection cycles, or on averages taken across all weeks sampled, and hence are more or less independent of the total number of weeks sampled (WIC). However, total observed parasitemia infections (STR), total number of infection cycles (NINF), total number of treatments (NT and NT1), and total weight change across the entire challenge period (WTC) can all be expected to show a linear relationship with WIC. See, e.g., Figure showing scattergram of STR against WIC. A clear linear relationship is apparent (R2= 0.487 and r = 0.695). This is expected, as the longer the animals were exposed the more opportunity they had to be re-infected and thus present positive detection of trypanosomes.
Scattergram of number of positive trypanosome cases detected (STR) against weeks in challenge (WIC).
On the assumption that for the time-related traits, the relation of trait value to WIC is approximately linear, they were standardized to the same challenge period by dividing by WIC, giving mean trait value per week. Thus, %PARD = 100*STR/WIC; MNT = NT/WIC; MNTI = NT1/WIC and WTCW = WTC/WIC.
All animals in the field were recorded weekly for packed cell volume (PCV) as a measure of anemia, and for parasitaemia score (PAR). PCV was determined by centrifugal techniques [19
]. Parasitemia species scoring (whether T. congolense
or T. vivax
) was based on morphological features, while the concentration of parasites per ml of blood was estimated according to the parasitemia score [20
] after examination of about 50 fields. Body weight (BW) measures of all animals were obtained in the morning every two weeks, on an electronic scale (Griffith Elder, UK) at ± 1 kg. When PCV was at 18% or less, animals were treated with a trypanocide (Berenil) by intramuscular injection at a dose of 7 mg/kg body weight. The number of treatments given to each animal in any given week (NT) was recorded as a fourth phenotype.
Estimation of gender and batch effects
Fixed effect analysis of BCB animals for batch and gender effects
Batch and gender effects expressed as deviation from a reference group, and their statistical significance were estimated by a two-way ANOVA with batch and gender as main effects, using the GLM module of the SAS statistical package (SAS 9.1). For batch and gender effects, the males of Batch 1 served as the reference group, and batch and gender effects were estimated as deviation of respective batch or gender mean from the reference-group mean. This was done separately for each trait. Individual trait values were then corrected for gender and batch effects, using these estimates of gender and batch effects. For gender correction, only statistically significant (P≤0.05) gender-effect estimates were used. For batch correction, all batch effect estimates were used for trait correction, irrespective of statistical significance. This was done because the overall batch effects were highly significant for most traits, and were generally much more significant than the gender effects. Estimated mean values by batch for each trait were computed by adding respective batch effect to the reference mean value for that trait. In order to obtain a general picture of the overall behavior with respect to infections and treatments of the BCB animals across all batches, a parameter herein referred to as “MeanAll” (Table ) was obtained for each trait, by calculating mean batch effect across all batches and adding this to the value of the reference group. Within-batch coefficients of variation (CV-within) were calculated for each trait, by computing standard deviation (SD) within each batch, and then mean SD pooled across batches; CV(within) = pooled SD/MeanAll. This within-batch CV was a measure of the ability of genetic and environmental factors varying within a given population and herd-year to affect trait value. In order to provide a similar parameter for the effects of between-batch environment on trait value, the between-batch coefficient of variation was calculated as CV (Batch) = SD (Between)/MeanAll, where SD (Between) is the standard deviation of batch effects across all batches.
Gender and Genetic-type effects on trypanotolerance
Female-gender effects as deviation from the male for the BCB animals across all six batches, were obtained from the two-way ANOVA as described above. Female-gender effects as a deviation from the male among the F1 in Batch 6 and among the KB in Batch 4, were separately analysed by batches, using a one-way ANOVA with gender as main effect. Genetic-type effects as a deviation from the BCB, were calculated for KB vs BCB in Batches 4 and 5, and for F1 vs BCB in Batch 6. Here too, the analyses were done separately by batches, using a one-way ANOVA with genetic type as main effect, after correcting for gender effects as obtained from the BCB, KB and F1 analyses.
The estimated effects between BCB and KB with their SE, were provided separately by ANOVA for Batch 4 and Batch 5 and then combined to a single estimate by calculating the simple unweighted mean of the two batch effects. SE of the combined effect was calculated as:
The effect of F1 as a deviation from KB was calculated as: Effect of F1 (as deviation from BC) - effect of KB (as deviation from BC), with
Statistical significance of differences between genders and genetic types were tested using absolute z-values = |D|/SE (D). The corresponding p-value was obtained as twice the area to the right of z in the standard z-tables (two-tail test).