Test day models were used to estimate the lactation curves for Valle del Belice ewes and to study the main environmental effects on milk yield and on percentage of fat and protein. Environmental effects were treated as fixed. A random effect was associated with each lactation to evaluate the mean correlation among all test day records of an individual ewe. Lactation curves were constructed by adding solutions for classes of either days in milk nested within parity or days in milk nested within season of lambing to appropriate general means. Parity primarily affected the lactation curve for milk yield, which was lower and flatter for first lambing ewes; effects on fat and protein were smaller. Season of lambing affected all traits. Seasonal productivity had the greatest effect on milk composition, resulting in an imbalance between fat and protein percentages. Flock and feed supplementation affected only the lactation curve for milk yield. The lactation curve of Valle del Belice ewes stood at a relatively high level. However, the presence of notable, perturbative effects (environmental and random variation) on milk yield and composition suggests that management is unable to meet the requirements of ewes consistently.

The interpretation of the several variables that contribute to defining milk quality is difficult due to the high degree of correlation among them. In this case, one of the best methods of statistical processing is factor analysis, which belongs to the multivariate groups; for our study this particular statistical approach was employed. A total of 1485 individual goat milk samples from 117 Girgentana goats, were collected fortnightly from January to July, and analysed for physical and chemical composition, and clotting properties. Milk pH and tritable acidity were within the normal range for fresh goat milk. Morning milk yield resulted 704 ± 323 g with 3.93 ± 1.23% and 3.48±0.38% for fat and protein percentages, respectively. The milk urea content was 43.70 ± 8.28 mg/dl. The clotting ability of Girgentana milk was quite good, with a renneting time equal to 16.96 ± 3.08 minutes, a rate of curd formation of 2.01 ± 1.63 min- utes and a curd firmness of 25.08 ± 7.67 millimetres. Factor analysis was performed by applying axis orthogonal rotation (rotation type VARIMAX); the analysis grouped the milk components into three latent or common factors. The first, which explained 51.2% of the total covariance, was defined as “slow milks”, because it was linked to r and pH. The second latent factor, which explained 36.2% of the total covariance, was defined as “milk yield”, because it is positively correlated to the morning milk yield and to the urea con- tent, whilst negatively correlated to the fat percentage. The third latent factor, which explained 12.6% of the total covari- ance, was defined as “curd firmness,” because it is linked to protein percentage, a30 and titatrable acidity. With the aim of evaluating the influence of environmental effects (stage of kidding, parity and type of kidding), factor scores were anal- ysed with the mixed linear model. Results showed significant effects of the season of kidding and parity on common fac- tors, while no differences were found between goats with one or more kids. The multivariate factor analysis technique was effective in describing the quality of Girgentana milk with a low number of new latent variables. These new variables have been useful in the study of the effect of some technical factors such as parity and season of kidding on the quan- titative and qualitative aspects of milk production in this goat breed.