Phenomics

Leveraging automatic milking system data to identify resilience indicators

---

Abstract

Automatic milking systems (AMS) relieve farmers from the time-consuming task of manual milking, allowing them to focus on other essential operations in the farm and to monitor each cow's production and wellbeing in real time, improving herd management and overall productivity. AMS increase the amount and resolution of data available, enabling the estimation of more accurate lactation curves. Consequently, they provide a powerful tool for animal breeding programs and for making informed decisions on culling and replacement strategies. Moreover, the analysis of lactation curves from AMS could help to identify more consistent or less resilient cows. The aim of this study was to analyze individual lactations curves to identify resilience indicators in Italian Simmental (IS). A total of 83,596 records (240 cows) were collected from six herds equipped with AMS. Production data and pedigree of the cows were provided by the IS Breeders Association. We considered only one lactation per cow (up to parity five) and at least 305 daily records within the lactation. Individual lactation curves were fitted using the Wood model; daily residuals were computed as the difference between the actual and the predicted daily milk yield (MY). The following nine parameters were computed for daily milk yield: standard deviation (STD_MY), coefficient of variation (CV_MY), variance (VAR_MY) and its natural logarithm (lnVAR_MY), median absolute deviation (MAD_MY), root mean square of successive differences (RMSSD_MY), the number of successive drops greater than 10% (DROP_MY), standard deviation of the ratio between the successive difference in milk yield and the successive difference in DIM (SLOPE_MY), and the number of times SLOPE changed the sign (CHA_MY). For the residuals, five statistics were computed: sum of the absolute values (SUM_RES), variance (VAR_RES) and its natural logarithm (lnVAR_RES), median absolute deviation (MAD_RES), and the autocorrelation for lag 1 (LAG1_RES). Heritability of all traits were estimated using herd as random effect, calving year (2022-2044) as cross-classified fixed effect, and animal as random effect (constructed using the pedigree-relationship matrix A, tracking back three generations of ancestors). Heritabilities for MY indicators were: 0.41±0.16 for STD_MY, 0.49±0.18 for CV_MY, 0.45±0.17 for VAR_MY, 0.35±0.16 for lnVAR_MY, 0.39±0.15 for MAD_MY, 0.14±0.11 for RMSSD_MY, 0.29±0.16 for DROP_MY, 0.14±0.12 for SLOPE_MY, and 0.11±0.12 for CHA_MY. As far as the indicators from residuals were concerned, heritability estimates were 0.20±0.12, 0.16±0.11, 0.19±0.12, 0.22±0.14, and 0.11±0.09 for SUM_RES, VAR_RES, lnVAR_RES, MAD_RES, and LAG1_RES, respectively. Heritabilities were low-to moderate with quite high standard errors, as expected because of the small sample size. These preliminary results suggest the possibility of including some of these parameters as breeding goals to improve resilience in the Italian Simmental cattle breed.

Keywords: 2026

How to Cite:

Degano, L., Cesarani, A., Gaspa, G., Ferrari, C., Vicario, D., Negrini, R., Dimauro, C. & Macciotta, N., (2026) “Leveraging automatic milking system data to identify resilience indicators”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2292938. doi: https://doi.org/10.31274/wcgalp.24310

Rights: 1