Sheep milk production is very important for the economy of Mediterranean countries. Dairy sheep also help to prevent environmental degradation and land abandonment in marginal rural areas. In such difficult environmental conditions, that require a particular adaptation, sheep are able to provide high quality milk that is mainly used to produce traditional cheeses (Selvaggi et al. Reference Selvaggi, Laudadio, Dario and Tufarelli2014). Following tradition, lambs are typically allowed free suckling for about 1 month and the milking period starts after weaning. As a consequence, milk production is the most important factor affecting lambs’ growth rate. It is well known that milk production is significantly influenced by both genetic and environmental factors. The knowledge of the relative weight of the variance components and the estimation of heritability for milk yield and quality traits may support animal breeding strategies. Genetic parameters are characteristics of the populations and environmental conditions and may change over time due to management decisions.
Local sheep breeds show many important characteristics such as rusticity, resistance and adaptability to difficult environmental conditions. Leccese is an Italian dairy sheep breed. Most researchers accept the theory of a common ancestor with the Zackel sheep, an Asiatic breed that penetrated also into South East Europe. For centuries, Leccese sheep have lived in a traditional breeding site called Salento, a narrow, flat sea level peninsula, in the southern part of Apulia, characterised by a semi-arid climate with long, dry and hot summers and mild winters. Nowadays, the Leccese sheep breed is included in the list of Italian endangered breeds drawn up by the Department for Environment, Food, and Rural Affairs. The Comisana sheep breed originated from the Maltese and Sicilian breeds in the late 19th and early 20th century in the Southeast region of Sicily. Also known as ‘red head’ because of its characteristic red face, the Comisana is a medium-large breed with all the morphological characteristics of a good dairy breed. This breed is well adapted to the semi-arid Mediterranean environment representing an important resource for the marginal areas of Southern Italy. The Sarda breed is a native breed of Sardinia off the coast of Italy. It originated from the local lowland sheep which were large, polled, and with white wool. Merino and Barbary sheep were also used in developing the breed. This breed is reared for milk production. Sarda and Comisana, are the most consistent dairy sheep breeds in Italy.
Estimation of heritabilities for production traits in local breeds is important both for conservation purposes and to establish appropriate breeding objectives and schemes. Moreover, the environmental effects such as year of lambing, parity number of the ewe and type of lambing play a key role in the phenotypic variation of the milk production traits. With this approach in mind, the aims of the present study were to determine the level of influence of some environmental factors on milk yield and quality traits and lactation length in ewes belonging to three different Italian breeds reared in Apulia region, and to estimate the heritability for the same traits as important and useful genetic parameters that should be considered to design breeding schemes for genetic improvement.
Material and methods
Animals
A total of 2138 lactation records obtained from 535 ewes belonging to three different breeds (Comisana, n = 207; Leccese, n = 181 and Sarda, n = 147), progeny of 77 unrelated rams (Comisana, n = 30; Leccese, n = 26 and Sarda, n = 21) were used in the present study. Data were collected over a 5-year period in three different farms (one per breed) located in the same province of Apulia region, in southern Italy. In order to reduce the variability, the dataset involved only the ewes lambing during February. Furthermore, information concerning lambing type (single or twins) and parity number (1st to 4th) were also available. The animals were reared following the traditional management practices of the area: animals are left to graze in daylight hours and return to the folds at sunset leaving the ewes with their lambs over the night. Lambs were allowed free suckling until 30 ± 3 d postpartum. Milking in these breeds is carried out twice a day, in the morning and in the afternoon. Milk test started after weaning and the milk production was recorded until individual daily milk yield dropped below 0·1 kg. Lactation records were obtained from a set of test-day records taken at fortnightly intervals and reflect the production of milk after weaning. Milk fat, protein, and lactose were analysed by near-infrared analysis (Milkoscan, Foss Electric, Hillerød, Denmark). Total fat, protein and lactose yields were obtained from milk yield and percentages for each test-day. Lactation length was calculated from weaning to drying off.
Statistical analysis
Data were analysed using the restricted maximum-likelihood method from Mixed Procedure of SAS software (1999) to identify the environmental factors affecting milk yield (MY); lactation length (LL); protein, fat and lactose yielded (PY, FY and LY, respectively), and for protein, fat and lactose content (PC, FC and LC, respectively). The model accounted for the variation due to the breed (1,…,3); the year of lambing (1,…,5); the parity number of the ewe (1,…,4) and the type of lambing (1, 2). All these effects were considered as fixed. Sires of ewes and ewe were considered as random effects. In a preliminary data analysis, the fixed effect of the flock was considered in the model; anyway, it was finally excluded since it was not found to be significant. The significant differences among sub-classes were detected using Duncan's multiple range test (Duncan, Reference Duncan1955). Then, a mixed model was used to estimate variance components for the same traits and one analysis per breed was made. The model accounted for the variation due to the year of lambing (1,…,5); the parity number of the ewe (1,…,4) and the type of lambing (1, 2) as fixed effects. The sire was considered as random. The error was assumed to be randomly and independently distributed, with mean of zero and a variance of σ 2e.
In matrix notation, the model can be written as:
$$y = Xb + Zs + e$$
where y is the vector of phenotypes for the considered traits; X and Z are known incidence matrices relating records to fixed and random effects, respectively; b is the vector of fixed effects; s is the vector of sire effects, and e is the vector of residual errors.
Heritability estimates were based on sire component of variance (σ 2s) as follows:
$$h^2 = 4\sigma ^2 {\rm s}/\left( {\sigma ^2 {\rm s} + \sigma ^2 {\rm e}} \right)$$
The standard error of heritability was approximated using the method described by Becker (Reference Becker1968).
Results and discussion
Environmental factors
Mean squares and test of significance of factors affecting milk yield and lactation length are given in Table 1. Breed of ewes showed effect (P < 0·001) on both MY and LL; Leccese breed produced less milk compared with Comisana and Sarda, with a shorter duration of lactation. Differences in MY and LL among breeds were commonly reported in literature. Year of lambing was a significant effect for both MY and LL (P < 0·001). This may be due to variation in climate, food nutritional quality and composition from 1 year to another. Moreover, MY was affected by parity and type of birth (P < 0·001) with milk production increasing until the third lactation and with a greater production in ewes with twins in comparison with ewes delivering single lambs. The effect of parity on milk yield was reported in other investigations (Macciotta et al. Reference Macciotta, Cappio-Borlino and Pulina1999; Pollott & Gootwine, Reference Pollott and Gootwine2001), and the positive influence of birth type on MY was in accordance with a previous report (Pollott & Gootwine, Reference Pollott and Gootwine2001). Beside the positive effect of suckling stimulus on milk yield, it is well known that litter size modifies the endocrine profiles in prepartum ewes and, as a consequence, mammogenesis and lactogenesis may be positively affected by an increased fetal and placental mass. On the other hand, some investigations reported no effect of litter size on milk yield and lactation length (Hassan, Reference Hassan1995; Macciotta et al. Reference Macciotta, Cappio-Borlino and Pulina1999). In such cases the effect of litter size may be hidden by undernutrition of the dam before or after parturition. In the present study, lactation length was not influenced by parity and type of birth. These results agree with those reported by Hassan (Reference Hassan1995) for Egyptian ewes. Similarly, Pollott & Gootwine (Reference Pollott and Gootwine2001) reported no influence of litter size on lactation length, although an effect of lactation number was observed.
Table 1. Least-square means and tests of significance of factors affecting milk yield (MY), lactation length (LL) and milk quality of ewes
***P < 0·001; n.s. not significant.
Means within a trait and factor followed by different letters differ significantly a, b = P < 0·05; A, B = P < 0·01.
FC, fat content; PC, protein content; LC, lactose content.
Least-square means for fat, protein and lactose content and the tests of significance of non-genetic factors affecting these traits are shown in Table 1. All these parameters were significantly affected by breed in accordance with results previously reported (Oravcová et al. Reference Oravcová, Margetín, Peskovicova, Dano, Milerski, Hetényi and Polák2007). The year and the parity of the ewes did not affect milk quality as reported by Hassan (Reference Hassan1995) in Saidi and Ossimi breeds. Conversely, a significant effect of year on milk composition was reported by Matutinovic et al. (Reference Matutinovic, Kalit, Salajpal and Vrdoljak2011). In accordance with our results, Oravcová et al. (Reference Oravcová, Margetín, Peskovicova, Dano, Milerski, Hetényi and Polák2007) found no significant effect of lactation number on fat and protein content in Tsigai, Improved Valachian and Lacaune sheep breeds. Furthermore, Sevi et al. (Reference Sevi, Taibi, Albenzio, Muscio and Annicchiarico2000) reported that parity affected the quality of milk in Comisana breed, and in particular its content of fat, protein, and casein which increased with increasing parity; on the contrary, lactose content significantly decreased with increasing number of lactation.
In the present paper, birth type affected milk volume and hence yield of fat, protein and lactose but had no effect on the content of these components (P > 0·05, Table 1). The number of lambs suckling did not significantly affect fat, total solids and solids-not-fat of milk obtained from Ossimi and Saidi sheep breeds (Hassan, Reference Hassan1995).
Heritability estimates
Estimates of sire's and environmental variance components and heritability values are given in Table 2. Heritabilities for MY were moderate, being 0·25 for Comisana and Sarda, and 0·28 for Leccese breed. These estimates fall within the range of values already published for other dairy breeds (0·10–0·38) being closer to the upper limit of the reference range (Sanna et al. Reference Sanna, Carta and Casu1997; Pollott & Gootwine, Reference Pollott and Gootwine2001; Gutiérrez et al. Reference Gutiérrez, Legaz and Goyache2007). With regards to lactation length, heritability was particularly low (0·07, 0·11 and, 0·08 for Comisana, Leccese and Sarda, respectively); previous studies reported values ranging from 0·01 to 0·14 (Sanna et al. Reference Sanna, Carta and Casu1997; Gutiérrez et al. Reference Gutiérrez, Legaz and Goyache2007). Heritability estimates for fat and protein content and for fat and protein yield were moderate to high, ranging from 0·20 to 0·42, and lower for lactose, ranging from 0·13 to 0·18 (Table 2). The Sarda exhibited the lowest heritabilities for fat related traits, but the lowest protein trait heritabilities were for the Comisana breed. In the literature, heritabilities for FC and PC range from 0·06 to 0·55 (Sanna et al. Reference Sanna, Carta and Casu1997; Hamann et al. Reference Hamann, Horstick, Wessels and Distl2004; Riggio et al. Reference Riggio, Finocchiaro, Van Kaam, Portolano and Bovenhuis2007). Some authors reported values of heritabilities for FY and PY slightly lower than those obtained in our breeds (Hamann et al. Reference Hamann, Horstick, Wessels and Distl2004; Riggio et al. Reference Riggio, Finocchiaro, Van Kaam, Portolano and Bovenhuis2007), whereas a previous investigation on Sarda breed reported slightly higher values (Sanna et al. Reference Sanna, Carta and Casu1997). To the best of our knowledge, only one heritability value for lactose content has been published, which was 0·28 in Merino sheep (Izquierdo et al. Reference Izquierdo, Corral, Padilla and Hernández2010), higher than we report here. In dairy cows, early studies have shown heritability estimates for lactose percentage and yield to be less or intermediate to those for fat and protein percentages; later references reported values in the range of 0·17–0·64 (Stoop et al. Reference Stoop, Bovenhuis and Van Arendonk2007; Ptak et al. Reference Ptak, Brzozowski and Bieniek2012; Sneddon et al. Reference Sneddon, Lopez-Villalobos, Davis, Hickson and Shalloo2015). In goat species, Brito et al. (Reference Brito, Silva, Melo, Caetano, Torres, Rodrigues and Menezes2011) reported heritability estimates for lactose yield and percentage of 0·15 and 0·17, respectively. These values are similar to those obtained in the present investigation in sheep.
Table 2. Estimates of sire (σ 2 s) and environmental (σ 2 e) variance components, heritability (h 2) and standard error (se h 2) for ewe lactation traits of Comisana, Leccese and Sarda breeds
MY, milk yield; LL, lactation length.
FC, fat content; FY, fat yield.
PC, protein content; PY, protein yield.
LC, lactose content; LY, lactose yield.
Conclusion
The wide variability of heritabilities for a certain trait reported in different studies conducted on different populations may be due to many reasons, such as differences among breeds and management factors, sampling errors, and differences in the estimation procedures adopted. Nevertheless, estimation of the heritabilities for milk yield and quality traits is necessary in order to design appropriate breeding programs. On the whole, the presently reported estimates are within the range of those already obtained in other dairy breeds by other authors, with values for lactation length being very low in all the investigated populations. The low heritabilities for lactation length may suggest low genetic gain, whereas improvements in environmental factors affecting this parameter may lead to greater gains for this phenotypic trait. On the other hand, the results obtained in the present study for milk traits other than lactation length suggest that their genetic variability is adequate for selection indicating that a good response to selection could be expected by adopting the correct breeding strategies.
