Assessment of milk yield in hair sheep ewes: a comparative study of two methods Evaluación comparativa de la producción de leche en ovejas peludas utilizando dos métodos

Abstract

This study compared two methods for estimating milk yield (MY) in multiparous Pelibuey and Katahdin ewes accounting for litter size and week of lactation under tropical conditions. MY was assessed using the weight-suckle-weight (WSW) technique and hand milking (HM). Exploratory analysis was performed using box plots, followed by a repeated-measures ANOVA based on a mixed linear model including method, breed, litter size, and week as fixed effects, and ewe as a random effect. Method agreement was evaluated using Bland–Altman analysis. MY increased during the first three weeks of lactation, then stabilized or slightly declined. HM consistently produced higher average values than WSW. Katahdin ewes and those rearing twin litters yielded more milk than Pelibuey and single-litter counterparts. The mixed model showed that estimation method and litter size were the main sources of variation, followed by breed and week. The intraclass correlation coefficient (ICC = 28.11 %) indicated that slightly more than one-quarter of total variability was attributable to individual differences. The means ± SE plot suggested an interaction, with the largest HM–WSW discrepancy observed in Katahdin twin-litter ewes, the highest-producing group. The Bland–Altman plot revealed a systematic bias (-0.78 kg; HM > WSW) and poor agreement (limits of agreement -2.26 to 0.66 kg), indicating potential individual-level discrepancies. HM tends to overestimate MY relative to WSW, particularly in highly productive ewes. The two methods are not interchangeable without adjustment; their selection should consider ewe characteristics and study objectives, and correction for systematic bias is recommended when comparing results.