Influence of parity and thermal stress (thi) on the average milk yield of pasture-based braunvieh cows raised in mountain environments: a population-level analysis using machine learning models

Summary

This population-level analysis examined how thermal stress (measured by THI) and parity interact to influence milk yield in pasture-based Braunvieh dairy cows using machine learning and classical statistics on 17,396 records. Across all parity classes, each unit increase in THI was associated with significant milk yield reductions (0.18–0.26 kg/day), with multiparous cows showing intermediate sensitivity. The study demonstrates that non-linear machine learning approaches better capture the accelerating production losses under high thermal stress, suggesting combined use of classical inference and predictive modelling for understanding complex environmental–demographic interactions in mountain dairy systems.

Regional applicability

This study was conducted on Brazilian mountain pasture systems with Braunvieh cattle and may have limited direct applicability to United Kingdom dairy conditions, which have substantially different ambient temperature ranges and humidity profiles. However, the methodological approach—combining machine learning with classical inference to model non-linear thermal–production relationships—could inform UK research into heat stress impacts on pasture-based dairy systems, particularly as climate variability increases.

Key measures

Average daily milk yield (kg/day); Temperature–Humidity Index (THI); parity class (primiparous, second-parity, multiparous); R² values for model fit; polynomial regression and Random Forest predictive models

Outcomes reported

The study measured average daily milk yield in relation to Temperature–Humidity Index (THI) and parity class, using machine learning and statistical modelling on 17,396 cow records. It quantified milk production losses associated with thermal stress across different parity groups in mountain-based pasture systems.

Topic tags