The aim of this work was to fit and compare three non-linear models (Wood, Milkbot and diphasic) to model lactation curves from two approaches: with and without cow random effect. Knowing the behaviour of lactation curves is critical for decision-making in a dairy farm. Knowledge of the model of milk production progress along each lactation is necessary not only at the mean population level (dairy farm), but also at individual level (cow-lactation). The fits were made in a group of high production and reproduction dairy farms; in first and third lactations in cool seasons. A total of 2167 complete lactations were involved, of which 984 were first-lactations and the remaining ones, third lactations (19 382 milk yield tests). PROC NLMIXED in SAS was used to make the fits and estimate the model parameters. The diphasic model resulted to be computationally complex and barely practical. Regarding the classical Wood and MilkBot models, although the information criteria suggest the selection of MilkBot, the differences in the estimation of production indicators did not show a significant improvement. The Wood model was found to be a good option for fitting the expected value of lactation curves. Furthermore, the three models fitted better when the subject (cow) random effect was considered, which is related to magnitude of production. The random effect improved the predictive potential of the models, but it did not have a significant effect on the production indicators derived from the lactation curves, such as milk yield and days in milk to peak.

The objective of the study was twofold: (i) to quantify the differences in daily milk yield (DMY) and total milk yield (TMY) between lame and non-lame dairy ewes and (ii) to determine the shape of lactation curves around the lameness incident. The overall study was a prospective study of lameness for the surveyed sheep population, with a nested study including the selection of matching controls for each lame ewe separately. Two intensively reared flocks of purebred Chios ewes and a total of 283 ewes were used. Data, including gait assessment and DMY records, were collected on a weekly basis during on-farm visits across the milking period. A general linear model was developed for the calculation of lactation curves of lame and non-lame ewes, whereas one-way ANOVA was used for the comparisons between lame ewes and their controls. Lameness incidence was 12·4 and 16·8% on Farms A and B, respectively. Average DMY in lame ewes was significantly lower (213·8 g, P < 0·001) compared with the rest of the flock, where DMY averaged 1·340 g. The highest DMY reduction in lame ewes was observed during the week 16 of the milking period (P < 0·001), whereas the reduction of DMY, for lame ewes, remained significant at P < 0·001 level from week 8 to week 28 of milking. Comparisons between lame and controls revealed that at the week of lameness diagnosis a significant DMY reduction (P ≤ 0·001) was observed in lame ewes (about 32·5%), which was maximised 1 week later (35·8%, P ≤ 0·001) and continued for several weeks after recovery, resulting in 19·3% lower TMY for lame ewes for the first 210 d of the milking period (P < 0·01). Moreover, at flock level, TMY for non-lame and lame ewes, as calculated by the general linear model, was 318·9 and 268·0 kg, respectively. The results of this study demonstrate evidence of significant financial losses in dairy sheep due to lameness which, however, need to be accurately estimated in further, more detailed, analyses.

Bayesian theory is applied to compare the characteristics of the estimated lactation curves of two groups of 5-year-old Merino ewes. The diets of the two groups were supplemented respectively by DL-methionine and maleyl-DLmethionine. The purpose is to illustrate the Bayesian approach when analysing for the effect of supplement on the lactation pattern of the sheep. Using Wood's model, the posterior distributions of the model parameters are determined for the two groups. This is achieved by assuming a hierarchical Bayes model and applying the Gibbs sampler, a sampling based computer intensive algorithm that is very efficient in obtaining marginal distributions of functions of parameters. The Gibbs sampler enables us to obtain marginal posterior distribution of characteristics of the lactation curve such as peak yield, time of peak yield, persistency and total milk yield. The results are notable differences in the marginal posterior distributions of mean peak milk yield and mean total yield. The posterior probability that the mean peak milk yield of the group supplemented by maleyl-DL-methionine is higher than that of the group with DL-methionine supplement is 0·98, while the same probability for mean total yield is 0·83.