Environmental and genetic effects on growth in Timahdite and crossbred lambs in Morocco

The Timahdite breed is the most important native breed of Morocco in number (about 17% of the total ewe population) and geographical distribution. This native breed is well adapted to a wide range of pastoral and mixed farming environments, where it is used mainly in purebreeding, but has poor prolificacy, less than 1.2 lamb per ewe lambing (7). Prolificacy determines the number of lambs available for sale, directly affecting productivity and profitability of the Timahdite flocks. To enhance lambing rate and productivity, a crossbreeding program with D’man prolific rams was carried out at El Koudia experimental station. Previous results reported by El Fadili et al. (8) showed that productivity of D’man x Timahdite ewes and growth of their progeny sired by meat breed were higher than in both parental purebreds, and that D’man x Timahdite first cross lambs showed similar growth rates as Timahdite ones. From a national perspective utilization, D’man x Timahdite crossbred dams might lead to an increase in productivity and profitability in sheep production as a whole. Therefore, interest arises in breeding D’man x Timahdite crossbred as a viable economic alternative to improve both ewe reproductive performances and lamb production, while maintaining their adaptability to grazing. However, development of efficient breeding programs that take into account the different structures of Moroccan breeds requires the knowledge of the genetic variability and genetic correlations between lamb weights and daily gains, which are important components of market lamb production.


s INTRODUCTION
The Timahdite breed is the most important native breed of Morocco in number (about 17% of the total ewe population) and geographical distribution.This native breed is well adapted to a wide range of pastoral and mixed farming environments, where it is used mainly in purebreeding, but has poor prolificacy, less than 1.2 lamb per ewe lambing (7).Prolificacy determines the number of lambs available for sale, directly affecting productivity and profitability of the Timahdite flocks.To enhance lambing rate and productivity, a crossbreeding program with D'man prolific rams was carried out at El Koudia experimental station.Previous results reported by El Fadili et al. (8) showed that productivity of D'man x Timahdite ewes and growth of their progeny sired by meat breed were higher than in both parental purebreds, and that D'man x Timahdite first cross lambs showed similar growth rates as Timahdite ones.From a national perspective utilization, D'man x Timahdite crossbred dams might lead to an increase in productivity and profitability in sheep production as a whole.Therefore, interest arises in breeding D'man x Timahdite crossbred as a viable economic alternative to improve both ewe reproductive performances and lamb production, while maintaining their adaptability to grazing.However, development of efficient breeding programs that take into account the different structures of Moroccan breeds requires the knowledge of the genetic variability and genetic correlations between lamb weights and daily gains, which are important components of market lamb production.
Only Tijani and Boujenane (21) reported estimates of paternal half-sib heritabilities for growth traits in Timahdite lambs, but no information on genetic parameters are available for the D'man x Timahdite lambs.Furthermore, there is no report on genetic parameters in Moroccan breeds estimated using animal model methodology and accounting for maternal additive genetic effect.However, studies showed that both direct and maternal genetic influences are of importance for lamb growth and that ignoring the maternal genetic effect in the models leads to larger estimates of direct genetic heritabilities (5,19,22).Therefore, the genetic correlation between the additive direct and additive maternal components are important, particularly in situations where selection programs include traits affected by both direct and maternal effects.
The purpose of this paper was to quantify the effects of some environmental factors on body weights and daily gains of Timahdite and D'man x Timahdite lambs from birth to 90 days, and to estimate additive direct and maternal genetic heritabilities and genetic correlations for the traits of interest.

Animals
Data came from Timahdite (T) and D'man (D) x Timahdite (DT) lambs born to ewes 1.5 to 6 years old between 1992 and 1998 at the National Institute and Agricultural Research experimental station of El Koudia located 30 km south of Rabat, on the Atlantic Coast, at an altitude of 150 m.The T flock was established in El Koudia station at the beginning of 1980.All D sires and some T sires were purchased outside from the national association of sheep and goat flocks.Data on the T breed included 544 records on the progeny of 322 ewes and 15 sires.Data on the DT (D male x T female) crossbreed included 756 records on the progeny of 467 ewes and 24 sires.In DT genetic groups, 201 crossed lambs were F2 (DT x DT) progeny of 116 ewes and 5 sires.Seven sires in T and 12 sires in DT genetic groups had progeny in various years.A total of 102 T and 161 DT dams lambed in various years.All lambs had complete records for all traits from birth to weaning.The available pedigree was used to form the numerator relationship matrix.

Management
Both T and DT ewes were raised under similar management conditions and an annual breeding cycle.Ewes were allocated to individual rams at random with an average mating ratio of 25 to 27 ewes per ram.Starting in July, young ewes were exposed to rams at an average age of 1.5 year.No culling was performed for the dams except for infertility, old age, and health problems.During lambing and suckling periods (December-May), ewes grazed on green pasture.Otherwise, they grazed on dry pasture and cereal crop residues.During the mating period of about 45 days, and for 5 to 10 days after lambing, ewes were kept indoors, with a ration composed of cereals (barley, triticale), molasses, sunflower meal, straw, hay, minerals and vitamins.Ewes were supplemented, depending upon available resources, pasture conditions and ewes requirements (maintenance, pregnancy, lactation).Lambs were kept indoors during the day and had free access to hay and commercial creep feed that contained about 16% of crude proteins, 0.8 forage units, minerals and vitamins.An annual program of vaccinations, deworming and dipping was carried out for all animals.
Lambs were born from December to January.The identities of newborn lambs and their dam, the date of birth, sex, birth type, birth weight within 24 hours and rearing type were all recorded.

Environmental effect analysis
Fixed linear models were applied to the data using GLM procedure in SAS (20).Preliminary analyses were used to assess all first order interactions among all fixed effects (genotype of lamb, birth year, sex of lamb, birth type, rearing types and dam age).The interaction between year of birth and sex of lamb was the only significant interaction and was fitted in final models, both in T and DT genetic groups.
In DT lambs, final models for BW, W30 and ADG10-30 included the same fixed effects as in the T models for the same traits, but the birth type contained one more subclass (lambs born triplet and greater), and the genotype of the lamb (F1, F2) was added to the models.Final models for W70, W90, ADG30-70 and ADG30-90 included the same fixed effects as for BW, W30 and ADG10-30, but the birth type was replaced by the rearing type of lambs (single, twin, triplet).In T and DT lambs, the rearing type was equal to the birth type if the lamb was alive at 40 days of age.But, if the lamb died before 40 days, the rearing type was different from the birth type and equal to the number of lambs still alive.

Genetic parameters
To estimate genetic parameters, data for each trait and for T and DT groups were analyzed separately.The same fixed effects as the ones used in fixed analyses were included in mixed models, except in DT lambs where the genotype effect was replaced by the heterozygosity in individual DT lambs (1.0 in F1 and 0.5 in F2 lambs), which was fitted as a covariate.The maternal heterosis and recombination loss effects in the second generation (F2) were omitted, because the available genetic group did not allow their estimation.Furthermore, Boujenane et al. (2) reported negligible and nonsignificant effects of these genetic components in D'man x Sardi crossbred.
The random effects were the animal effects for the additive direct genetic effect and the maternal genetic effect, and the residual effect.The linear mixed models used to analyze growth traits were in matrix notation: where: y is a vector of observations, ß is a vector of fixed effects, b is a vector of the partial regression coefficients of observation y on level of heterozygosity in the lambs, H is the vector of heterozygosity coefficients in DT lambs, d~N(0, Aσ 2 d ) is a vector Retour au menu Retour au menu of direct additive genetic values, m~N(0, Aσ 2 m ) is a vector of maternal additive genetic values, e~N(0, Iσ 2 e ) is a vector of residual effects including both random environmental and non additive genetic effects.The X, Z and M are incidence matrices that assign the appropriate effects to the vector of observations, A is the matrix of additive numerator genetic relationship matrix among animals and I is the identity matrix of order equal to the number of records.The σ 2 d and σ 2 m are the additive direct and additive maternal genetic variances, and σ 2 e is the residual variance.
The REML estimates of (co)variance components were obtained using the program MTDFREML (1).Direct and maternal heritabilities, and correlation between direct and maternal genetic effects were first estimated using single-trait analyses.Next, genetic correlations between all traits were estimated using twotrait analyses with only the direct genetic effect in the model.The starting values for additive genetic and residual variances in the two-trait analyses were those estimated under single-trait analyses.A minimum of two cold restarts were performed to check for global maxima.A variance of 10 -8 of simplex function values was chosen as the convergence criterion.Standard errors of genetic correlations were calculated using the approximate formula given by Falconer and Mackay (9).

Environmental effects
Overall means (and standard deviations), test of significance (F) and the proportion of variation explained by the fixed model (R 2 ) are given in table I.The fixed models explained 42 to 64% of the phenotypic variances in all traits in DT lambs, and 47 to 51% in T lambs.Birth year, sex of lambs, birth type, rearing types were all important environmental sources of variation on all growth traits (p < 0.01), both in T and DT lambs.Ewe age was an important effect (p < 0.01) for BW, W30, ADG10-30, but its influence decreased for W70 and ADG30-70 and became non significant for W90 and ADG30-90, both in T and DT lambs.The year by sex of lamb interaction was significant only for W70, W90, ADG30-70 and ADG30-90 in T lambs (p < 0.05) and in DT lambs (p < 0.01).The genotype influence in DT lambs was not significant for BW, W30 and ADG10-30, but highly significant for the other traits (p < 0.01).In general, these results are in agreement with those reported for growth traits, particularly for sex of lamb, year of birth and for the type of birth in previous studies on Moroccan breeds such as the Timahdite ( 21), the D'man (3), and the Beni Guil breeds (4).The number of observations and least-square means (± standard errors) are shown for T lambs (table II) and DT lambs (table III).The effect of the year of birth on lamb weights and daily gains both in T and DT are drawn in figures 1 and 2.
Differences between male and female lambs were 0.23 and 0.12 kg at birth and 2.09 and 3.30 kg at 90 days of age in T and DT lambs, respectively.For ADGs, the differences between male and female lambs were 12 and 13 g/d for ADG10-30 and 24 and 38 g/d for ADG30-90 in T and DT lambs, respectively.Observed weight differences between male and female were consistent with other results (3,4,6,11).
In T breed, single-born lambs were 0.82 kg heavier than twins were at birth.Similarly, single-born lambs were 2.95 kg heavier at 30 days, 3.86 kg at 70 days, and 4.18 kg at 90 days than twins.
In DT, single-born lambs were heaviest by 1.12, 3.67, 4.30, and 6.01 kg, respectively at birth, 30, 70, and 90 days (table III).However, the ADGs between each subclass decreased with the age of lambs, both in T and DT.The least-square means showed that the effects of rearing types on the lamb body development tend to decrease as the lamb became older.This declining age trend can be attributed to a decreasing maternal effect including nursing and milk feeding of the lambs by their mothers.The differences between single and multiple lambs in weights and daily gains at birth and at 90 days have been reported in Barbarine (11), D'man (3) and Beni Guil (4) breeds.
The weights and ADGs of lambs born from dams in first parity were lower (p < 0.05) than those born in subsequent parities, but the differences in ADGs were significant only for ADG10-30 in T lambs and for ADG30-70 in DT lambs.In general, lambs were heavier and grew faster with increased age of dam until an optimum age of 4.5-5.5 years, then decreased both in T and DT lambs (tables II and III).Similar results were reported in the literature (6,11).
In DT lambs, the differences between F1 and F2 lambs were not significant for BW, W30, and ADG10-30 (p > 0.05), but for advanced ages the differences were significant (p < 0.01).The F2 were lighter and showed lower daily gains than F1 lambs: -2.51 kg for W90 and -38 g/d for ADG30-90.The superiority of F1 crossbred lambs can be explained by the heterosis effect for growth traits.Nitter (15), reviewing heterosis for growth in sheep, reported average estimates of individual heterosis to be about 3.2 and 5% for birth and weaning weights, respectively.In F2 lambs, the expected individual heterosis is halved, but the difference in the number of lambs born from F1 dams (prolificacy of 185%) compared to that of F1 lambs born from T dams (prolificacy of 120%) may explain the superiority of F1 lambs.Indeed, singleborn lambs represented 53.17 vs. 40%, twins 20.24 vs. 15.34% and triplets and greater 0 vs. 45%, respectively in F1 and F2 lambs.

Genetic parameters estimation Single-trait analyses
Direct (h 2 d ), maternal (h 2 m ) heritabilities and direct-maternal additive correlation (r dm ) for lamb weights and ADGs obtained from the single-trait analyses are given in table IV.These estimates should be taken with caution due to the small data set used both in T and DT lambs.
Heritability estimates fluctuated across traits and within the T and DT lambs, taking a wide range of values.For daily gains, h 2 d estimates were slightly larger in T than in DT lambs, except for ADG10-30.On the other hand, h 2 m estimates were larger in DT than in T lambs, except for BW.
For both T and DT, h 2 d estimates for lamb weights fall within the range of values reported in the literature.In the Barbarine breed, Khaldi and Boichard (11) found an h 2 d of 0.02 to 0.04 for weight at birth to 90 days.Using an animal model, Maria et al. (13) reported in the Romanov breed an h 2 d of 0.04 to 0.34 for weights at birth to 90 days, and Tosh and Kemp (22) found values of 0.05 to 0.39 for weight at birth to 100 days in three breeds of sheep.Estimates of heritability in models, which did not consider maternal effects, as reported by Boujenane and Kerfal (3) in D'man breed, were larger and ranged from 0.23 to 0.56.In the T breed, our estimates of h 2 d were higher than those observed by Tijani and Boujenane (21) for the same breed.Their estimates were of 0.02 for BW and 0.06 for W90.These discrepancies could be due to differences in animals, models and computational methods.These authors used data from a T flock raised in the medium Atlas area of Morocco, a paternal half-sib model, and a computational method III of Henderson (10).
Estimates of h 2 m were low in T to moderate in DT lambs, but high for BW in both genotypes (table IV).Burfening and Kress (5) also found a large maternal effect on BW that ranged from 0.30 to 0.65.In DT, the estimates of h 2 m (0.20 to 0.36) were larger than those for h 2 d (0.02 to 0.18).Poivey et al. (17) reported h 2 m estimates of 0.30 for W30 and W90 in Ile de France lambs.In DT lambs, the estimates of h 2 m of the present study were higher than those observed by Khaldi and Boichard (11) for Barbarine lambs, with h 2 m of 0.04 (W30) and 0.03 (W90).They were also higher than those reported by Maria et al. (13) in Romanov lambs for BW (h 2 m = 0.22) and W90 (h 2 m = 0.01), by Tosh and Kemp (22) in Hampshire for BW (h 2 m = 0.22) and W90 (h 2 m = 0.19), in Polled Dorset for BW (h 2 m = 0.31) and W90 (h 2 m , = 0.09), and in Romanov lambs for BW (h 2 m = 0.13) and W90 (h 2 m = 0.02).On the other hand, the estimates in T lambs are similar to those observed by Maria et al. (13) and by Khaldi and Boichard (11).The results for body weights show that within breed groups, estimates tend to increase for h 2 d and to decline for h 2 m from birth to 90 days of age.This was expected because maternal influence, expressed during gestation and lactation, decreases in importance as lambs become independent of their dams (increasing expression of genes with direct additive effects on body development).This tendency was also observed in other studies (5,13,14,16,22).
For average daily gains, the h 2 d estimates in DT and T lambs were lower than those observed by Djemali et al. (6) in Barbarine lambs with h 2 d = 0.24 (ADG30-70) and h 2 d = 0.31 (ADG30-90), and by Boujenane and Kerfal (3) in D'man lambs with h 2 d = 0.56 (ADG30-90), and they were higher than those reported for the Barbarine breed (11) and T lambs (21).
High h 2 m estimates of 0.29 (ADG10-30) and 0.27 (ADG30-90) were reported by Poivey et al. (17) in Ile de France lambs.The h 2 m estimates for ADG30-90 in this study were higher than those observed by Khaldi and Boichard (11) and by Maria et al. (13), who reported low estimates of h 2 m ranging from 0.01 to 0.06 for growth traits in Barbarine and Romanov sheep.Maternal heritability decreases with age, which confirms the finding of Notter and Hough ( 16), Tosh and Kemp (22), and Yazidi et al. (23), who observed that maternal effects are substantial in young animals but diminish with age.Small estimates of h 2 m in T breed indicate a much lower proportion of genetic variation in T than in DT crossbred lambs.The greater maternal genetic effect on weights and daily gains in DT lambs is expressed through the variation in the uterine environment, litter size and milk production of prolific DT dams.Estimated genetic correlations (r dm ) between direct additive and additive maternal effects were negative, but unreasonably high, both in T and DT lambs, ranging from -0.80 to -1.0 for body weights and from -0.90 to -1.0 for daily gains.It is difficult to find a biological explanation for this high correlation.It may be due to the small data size and to the structure of this data set (i.e., the number of generations for which animals were measured both directly and as dams was limited).In general, to avoid this type of

Table IV
Estimates of direct (h 2 d ) and maternal (h 2 m ) heritability and additive-maternal genetic correlation (r dm ) for growth traits in D'man x Timahdite and Timahdite lambs -Single character model BW = weight at birth; W30 = weight at 30 days of age; W70 = weight at 70 days of age; W90 = weight at 90 days of age ADG10-30 = average daily gains from 10-30 days of age; ADG30-70 = average daily gains from 30-70 days of age; ADG30-90 = average daily gains from 30-90 days of age Retour au menu Retour au menu problem, three generations and large grandparent offspring relationships in the data are needed (12).However, our estimates were obtained after rerunning the program with different starting values and with two or three cold restarts, and they still converged to the same estimates.Negative estimates for direct-maternal covariance for early growth traits are numerous in the literature for sheep.Maria et al. (13) reported strong negative correlations (r dm ) between direct and maternal additive effects for lamb body weights of -0.99 (BW), -0.98 (W30), -0.97 (W90), and -0.99 (ADG).Poivey et al. (17) also reported high negative estimates of r dm ranging from -0.61 to -0.70 for growth traits to weaning in Ile de France lambs.Other authors (5,11,22) found important correlations between additive direct and additive maternal effects for body weight traits, ranging from -0.74 to -0.18.
In a simulation study Robinson (18) found that a large proportion of the negative correlation between direct and maternal for weaning weight could be caused by a sire by year interaction, which is especially important when a large proportion of sires are introduced into the population each year.This was the case in the present study as some T and D sires were bought out of the flock.
Other studies found that correlation between direct and maternal genetic effects could not be important or favorable.Indeed, Saatci et al. (19) observed no correlation between direct and maternal additive effects for 12-week weights of Welsh Mountain lambs.Nasholm and Danell (14) and Yazidi et al. (23) observed positive direct-maternal additive correlations for BW (0.11 and 0.18) and the weaning weight (0.47 and 0.50).

Two-trait analyses
The estimates of the genetic (r G ) and phenotypic (r P ) correlations between traits in DT and T lambs are presented in tables V and VI, respectively.These estimates were obtained using only an additive direct animal model because, in the complete model, the genetic correlation between direct and maternal additive effects was high.Medium to high genetic correlations in DT (r G = 0.61 to 0.81) and T (r G = 0.52 to 0.81) were found between BW and other growth traits.The corresponding phenotypic correlations ranged from 0.12 to 0.55 in T, and from 0.27 to 0.58 in DT.All genetic correlation estimates were positive and high among W70, W90, ADG10-30, ADG30-70 and ADG30-90 traits, and exceeded 0.62 and 0.75 in T and DT lambs, respectively.The phenotypic correlation estimates among these traits were low to high ranging from 0.26 to 0.92 in DT, and from 0.13 to 0.94 in T lambs.In general, the largest genetic and phenotypic correlations were between chronologically adjacent weights rather than non-adjacent ones.As was reported in other sheep populations (Beni Guil, D'man, and Barbarine), there was a close association between postnatal weights and ADGs, as reflected by the high genetic and phenotypic correlations of W30 and ADG10-30, W70 and ADG30-70, and W90 and ADG30-90.

s CONCLUSION
The environmental factors investigated in this study had a high influence on T and DT lamb weights and on growth rates from birth to 90 days of age.Therefore, it is necessary to consider these environmental factors to obtain accurate estimates of breeding values.The direct heritability estimates obtained in this study were medium, except for the low values of BW both in T and DT lambs.The genetic and phenotypic correlations among growth traits of T and DT lambs were positive and high indicating that selection for one of those traits could result in genetic improvement in all other traits.Estimates of additive direct-maternal correlations were high and negative but the accuracy of such estimates is low due to the small size of the data set.In general, negative direct-maternal genetic correlation may complicate selection programs in sheep when farmers want to improve both an animal's own performances for growth and its dam's maternal genetic contribution to that growth.This is the case in the local Moroccan sheep where lamb growth depends closely on the milk production of their dams in the first 3-4 months.
BW = weight at birth; W30 = weight at 30 days of age; W70 = weight at 70 days of age; W90 = weight at 90 days of age ADG10-30 = average daily gains from 10-30 days of age; ADG30-70 = average daily gains from 30-70 days of age; ADG30-90 = average daily gains from 30-90 days of age Means within a column that do not have a common subscript differ (p < 0.05)

Table II
Number of observations and least square means (± standard errors) for growth traits in Timahdite lambs BW = weight at birth; W30 = weight at 30 days of age; W70 = weight at 70 days of age; W90 = weight at 90 days of age ADG10-30 = average daily gains from 10-30 days of age; ADG30-70 = average daily gains from 30-70 days of age; ADG30-90 = average daily gains from 30-90 days of age Means within a column that do not have a common subscript differ (p < 0.05) Figure 2: Effect of the birth year on weights and daily gains in D'man x Timahdite lambs.BW = weight at birth; W30 = weight at 30 days of age; W70 = weight at 70 days of age; W90 = weight at 90 days of age; ADG10-30 = average daily gains from 10-30 days of age; ADG30-70 = average daily gains from 30-70 days of age; ADG30-90 = average daily gains from 30-90 days of age.