MSPD will be published on a phenotypic scale where the average is roughly 7 pounds per minute in U.S. Holsteins. For example, daughters of a Holstein bull with a MSPD Predicted Transmitting Ability (PTA) of 8 lbs/min would be expected to milk faster than the average cow in the population. Conversely, daughters of a Holstein bull with a MSPD PTA of 6 lbs/min would be expected to milk slower than average.

Pre-release analysis indicates active A.I. Holstein sires in December 2024 range from 6.2 pounds per minute to 8.1 pounds per minute, with the average at about 7.1 pounds per minute. MSPD has a standard deviation of 0.3 pounds per minute, meaning it is assumed that 68% of bulls will have an PTA for MSPD between 6.7 and 7.3.

Young genomic bulls are expected to have reliabilities averaging 58% for milking speed, and progeny-tested bulls are expected to have genomic reliabilities averaging 67%. As additional data are accumulated, reliabilities will increase and MSPD evaluations can be published for more breeds. 

These higher reliabilities at launch are possible due to the high heritability (42%) and the amount of data included in each cow record.

MSPD will not be incorporated into the Lifetime Merit Indices at launch. It could be included in the future. MSPD is lowly correlated with Lifetime Net Merit $ (NM$), which suggests it is a trait not already captured in NM$ and could add significant economic value if incorporated.

It is suggested that producers continue to rely primarily on a composite economic index, with the specific choice dependent on the farm’s milk payment situation and management system. In the meantime, producers might consider avoiding bulls with extremely high or low predictions for MSPD.

Milking duration does not account for the amount of milk produced, making it a less useful metric.

Geneticists at CDCB and USDA Animal Genomics and Improvement Laboratory (AGIL) worked with raw milking parlor data from 304 herds provided by Dairy Records Management Systems (DRMS). This data set consisted of milk weights, milking durations, breed, parity, and meter manufacturer (OEM). After data cleaning, the final dataset used in the research and development of MSPD represented 165 herds, nine milk meter manufacturers, and 47,833 complete lactations of 43,788 Holstein and Jersey cows, including 12,237 genotyped cows. All available phenotype data used in the computation of MSPD lactation records was from complete lactations only. While the initial dataset included automated milking system (AMS) data, it was not included in the final dataset. Inclusion of this data will be investigated as research continues.

The largest significant (P < 0.05) correlation with milking speed PTA was with somatic cell score (SCS) PTA at 0.43 and mastitis (MAST) PTA at -0.28, while no phenotypic correlations were identified. This may be explained by the moderate-strong correlation between milking speed and milk yield and the well-known antagonistic correlation between milk yield and SCS. No significant correlations were estimated between longevity or fertility traits. A low correlation of MSPD with Lifetime Net Merit $ (NM$), < 0.10, is a desirable outcome as it suggests that MSPD is not a trait already captured in the national index and could add significant economic value if included.

Like all data used to calculate U.S. genetic evaluations, information moves from the farm to the National Cooperator Database through a certified collaborator network. For MSPD data, dairy records processing centers (DRPCs) will utilize a new data format (Format 8) to manage and move this sensor-derived data. More information about the Format 8 is available here.

Within the Format 8 data file, a distinction is made to identify what type of system was used to collect the data.  As each manufacturer has a unique way to measure the factors included in MSPD, the data must be standardized for use in genetic evaluations.

At the time of introduction, MSPD does not include milking speed data from Automatic milking systems. Data utilized to calculate the trait is collected from in-line sensors in traditional parlor equipment.