The microbial, chemical, and sensory quality of bulk tank raw milk affects the shelf life and quality of finished dairy products. Traditionally, raw milk quality is determined using total bacteria count (TBC) and SCC. Here, a longitudinal study was conducted over a 15-mo period with 100 conventional dairy farms enrolled, each sampled 6 times. The farms represented a variety of sizes, milking systems, and other farming practices. Samples were evaluated for 24 different quality parameters, covering both traditional (e.g., TBC, SCC, and chemical composition) and novel measures of quality (e.g., sensory defect evaluation, mesophilic spore counts MSC, thermophilic spore counts TSC, psychrotolerant spore counts PSC, and butyric acid bacteria BAB). Overall, we collected 593 bulk tank raw milk samples and conducted microbial, physicochemical, and sensory analyses. Our results showed that New York State raw milk is of exceptional quality, with a mean and SD for total bacteria count of 3.52 ± 0.70 log cfu/mL and geometric mean for SCC of 133,000 cells/mL. Overall sensory scores were also high, with a mean and SD of 8.6 ± 1.4 on a 0.0 to 10.0 scale. The most common attribute was oxidized. For sporeformer levels, mean and SD were 0.61 ± 0.60 log cfu/mL, 0.32 ± 0.60 log cfu/mL, 1.70 ± 0.60 log MPN/L, and 2.22 ± 0.60 log MPN/L for MSC, TSC, PSC, and BAB, respectively. Alongside sample collection, a survey was administered to farm owners and herd managers to capture comprehensive data, including but not limited to farm characteristics, milking characteristics, and parlor practices. Random forest models were developed to identify factors that may be influential in affecting milk quality, specifically through sporeformers and sensory characteristics (overall sensory scores and presence of oxidative defects). For sporeformer levels, the models most commonly identified herd size along with factors associated with udder health and hygiene (predip usage, frequency of udder clipping or flaming, and the vacuum of the milking system) as variables of importance. For sensory parameters, herd size, time spent on pasture, and measures of milk composition, including overall butterfat, percentage of preformed fatty acids, and percentage of de novo fatty acids, were identified as variables of importance. Given the financial and time burden associated with quality testing, parameters must be carefully selected to maximize utility of data. Thus, correlation analysis between test results was performed to identify quality parameters that could signal issues in other parameters. Relationships of note included log TBC and log preliminary incubation count (r = 0.79), log MSC and log TSC (r = 0.71), as well as overall sensory score and proportion of oxidized samples (r = -0.77). Overall, our study provides information that establishes a baseline dataset that can be used by the dairy industry to evaluate incremental improvements in raw milk quality, to identify farming practices that have potential to impact finished products, and to support the selection of parameters that may be used for quality monitoring.
Shaposhnikov et al. (Sun,) studied this question.