Mastitis-Causing Streptococci Are Important Contributors to Bacterial Counts in Raw Bulk Tank Milk

2644 Journal of Food Protection, Vol. 67, No. 12, 2004, Pages 2644 2650 Copyright, International Association for Food Protection Mastitis-Causing Streptococci Are Important Contributors to Bacterial Counts in Raw Bulk Tank Milk R. N. ZADOKS, 1 * R. N. GONZÁLEZ, 2 K. J. BOOR, 1 AND Y. H. SCHUKKEN 2 1 Department of Food Science and 2 Quality Milk Production Services, Cornell University, Ithaca, New York 14853, USA MS 04-101: Received 12 March 2004/Accepted 5 July 2004 ABSTRACT The objective of this study was to probe the contribution of streptococci to the microbial quality of raw milk. Over a 5- month period, bulk tank milk samples from 48 New York State dairy farms were analyzed qualitatively for bacterial ecology and quantitatively for total bacterial, streptococcal, staphylococcal, and gram-negative bacterial counts. Linear regression analysis was used to determine the contribution of differential counts to total bacterial counts. Streptococci, staphylococci, and gram-negative bacteria accounted for 69, 3, and 3% of total bacterial count variability, respectively. Randomly selected Streptococcus isolates from each bulk tank milk sample were identified to species by means of the API 20 STREP identification system. The most commonly identified streptococcal species were Streptococcus uberis, Aerococcus viridans, and Streptococcus agalactiae, which were detected in 81, 50, and 31% of 48 bulk tank samples, respectively. For five herds, S. uberis isolates from bulk tank milk and individual cows were characterized by PvuII ribotyping. A farm-specific dominant ribotype was found in each bulk tank sample, and that ribotype was isolated from at least one cow within each herd of origin. Bacteriological and strain typing data indicate that control of streptococci, specifically mastitis-causing species, is important for improvement of the microbial quality of raw milk in New York State. The microbial quality of raw milk is routinely evaluated in accordance with guidelines from the Pasteurized Milk Ordinance (PMO) (20). Compliance with PMO standards for grade A milk is mandatory for interstate shipment of milk and recommended for intrastate shipments to ensure the uniformity and high sanitary quality of milk and milk products. The PMO states that to qualify as grade A milk, bacterial counts of individual producer milk shall not exceed 100,000 CFU/ml. This quality criterion is frequently met: mean standard plate counts (SPCs) were 11,400 CFU/ ml in New York State in 1993 through 1996 (1), 19,000 CFU/ml in Vermont in 1990 (8), and ranged from 4,700 to 17,000 CFU/ml among 11 states representing the major milk-producing areas of the United States (16). More recently, the average plate loop count in New York State was reported to be 24,000 CFU/ml (1999 to 2000 data provided by the five biggest milk plants in the state) (21). In addition to SPCs and plate loop counts, which are standard methods for postharvest assessment of the microbial quality of raw milk, other methods are also routinely applied within the dairy industry to assess numbers and types of bacteria present in milk (1, 8, 10, 11, 15, 16). For example, total bacterial count (TBC) as measured on blood-esculin agar is the current milk production industry standard for analyzing bulk tank milk (BTM), with the goal of improving herd udder health and raw milk quality (8, 11, 15). Although mean bacterial counts in these studies are well below PMO limits, many milk shipments have bacterial counts in excess of regulatory standards: the proportion * Author for correspondence. Tel: 607-254-4967; Fax: 607-254-4868; E-mail: rz26@cornell.edu. of shipments with SPCs higher than 100,000 CFU/ml ranged from 0 to 10.5% for 11 dairy processing plants in New York State (1) and from 0 to 12.1% for the states surveyed by Peeler et al. (16). Factors that may contribute to high bacterial numbers in raw BTM include mastitis, poor sanitation practices, and improper milk handling or cooling (7, 11, 13, 17). In a pilot study of farms with highquality milk production, short-lived elevations in bacterial counts were often associated with the presence of streptococci (10). To assess the importance of streptococci as contributors to microbial counts in raw bovine BTM in New York State, a second study was conducted on milk samples from 48 dairy farms representing a variety of milk quality levels. The objectives of this study were (i) to determine the contribution of streptococcal counts to TBCs in raw BTM and (ii) to explore which streptococcal species are most commonly identified in raw BTM. MATERIALS AND METHODS Sample collection. From October 2000 through March 2001, samples of raw BTM and milk from all lactating cows were collected on 200 dairy farms in New York State during whole-herd mastitis screening surveys performed by the Ithaca Laboratory of Cornell University s Quality Milk Production Services (QMPS). Cow milk and BTM samples were aseptically collected by QMPS personnel as recommended (15). Samples were kept in coolers with cold packs, transported to the laboratory, and stored in a refrigerator at 2 to 4 C. Bacteriological culture was performed within 18 h of sample collection, and BTM samples were subsequently frozen at 80 C. For 21 herds, bedding samples were collected. For each herd, 10% of stalls in the lactating cow house area (tie stall or free stall facilities) were sampled. Composite bedding samples were obtained from the back one third of each

J. Food Prot., Vol. 67, No. 12 STREPTOCOCCI IN BULK TANK MILK 2645 selected stall. Sawdust, straw, hay, recycled manure, and sand were materials used as bedding in the dairies that participated in the study. Bedding samples were collected in plastic storage zipper-type bags and brought to the laboratory for bacteriological analyses. Bedding samples could not be processed for all herds because of the resource- and labor-intensive nature of the analyses. Bacteriology of raw BTM. BTM samples from the 200 farms were qualitatively analyzed for bacterial ecology by the QMPS Central Laboratory (Ithaca, N.Y.). Approximately 0.025 ml of each BTM sample was spread with a sterile cotton swab applicator onto a Trypticase soy agar II plate containing 5% sheep blood and 0.1% esculin (BBL, Becton Dickinson, Sparks, Md.). Plates were incubated aerobically at 37 C for 48 h. Colony morphology and hemolytic patterns on blood agar were observed, and isolates were examined further by means of Gram staining and 3% KOH, catalase, and oxidase testing. Gram-positive, catalasenegative cocci that were CAMP positive and esculin negative were considered Streptococcus agalactiae. All other gram-positive, catalase-negative cocci were designated as Streptococcus spp. Staphylococcus aureus was identified based on typical zones of complete and/or incomplete hemolysis and a positive tube test for free coagulase. Coagulase-negative staphylococci were classified as Staphylococcus spp. Gram-negative organisms were identified based on colony morphology on MacConkey agar, oxidase, indole, and a miniaturized commercial biochemical system (Crystal System, BBL, Becton Dickinson). Other mastitis pathogens were identified as described elsewhere (23). For each BTM sample, up to seven bacterial species or groups (e.g., Streptococcus spp., Staphylococcus spp.) were recorded. Total and differential counts. Quantitative analysis was performed for a convenience sample of the 48 BTM samples. Total bacterial and differential counts were determined as described in the National Mastitis Council Laboratory Handbook on Bovine Mastitis, the reference standard for diagnosing bovine mastitis (15). For determination of TBCs, streptococcal, staphylococcal, and gram-negative bacterial counts, respectively, Trypticase soy agar II plates containing 5% sheep blood and 0.1% esculin, Edwards modified medium (Unipath Co., Oxoid Division, Ogdensburg, N.Y.), Vogel-Johnson medium (Oxoid), and Mac- Conkey medium (Oxoid) were used. Media were inoculated with the National Mastitis Council recommended inoculum of 0.01 ml of thawed BTM sample or, if limited bacterial growth had been observed during qualitative analysis of the BTM sample, with 0.05 ml of thawed BTM sample. Plates were incubated aerobically at 37 C. After 24 h of initial incubation, plates were evaluated for growth. If colonies were too numerous to count, 10-fold dilutions of the BTM samples, which had been stored refrigerated at 2 to 4 C, were made in sterile distilled water and plated. BTM samples were frozen at 80 C again, and incubation of plates at 37 C was continued or initiated as applicable. All platings were performed in duplicate. After 48 h of incubation, counts from duplicate plates inoculated with 0.01 or 0.05 ml of undiluted BTM were recorded and averaged for each medium, or when counts were performed on plates with diluted BTM samples, two plates each from two consecutive dilutions, e.g., 10 2 and 10 3, were counted, and the mean colony count expressed in CFU per milliliter of undiluted BTM was used. Only those plates containing approximately 25 to 250 colonies were counted. Characterization of Streptococcus spp. After BTM samples had been used for bacterial counts, they were stored at 80 C for up to five working days. Samples were subsequently thawed and restreaked on Edwards modified medium and incubated for 48 h at 37 C. For each herd, three to five colonies were randomly selected from the Edwards modified medium plate and were identified to species with the API 20 STREP system (biomérieux, Inc., Hazelwood, Mo.) as recommended by the manufacturer. To probe whether specific cows could be sources of Streptococcus uberis in BTM, strain typing of S. uberis isolates from BTM and from cow milk samples collected on the same day as the BTM sample was performed for five herds. For those five herds, aseptically collected cow milk samples from all lactating cows were cultured as described for BTM, and isolates from all Streptococcus-positive cows were identified to species by API typing. Isolates were further typed with restriction enzyme PvuII at Cornell University s Laboratory for Molecular Typing (http://www.riboprinter.cornell. edu/) using the RiboPrinter Microbial Characterization System (Qualicon, Wilmington, Del.). For the remaining herds, cow milk samples were aseptically collected from all lactating animals using recommended procedures (15), and culture was performed as described for BTM, but complete species and subtyping data were not generated. Analysis of bedding samples. Bedding samples were analyzed for the first 21 of the 48 herds for which quantitative and streptococcal species data were generated. Straw and hay were chopped to ease handling, and 10 g of bedding sample was added to 90 ml of sterile phosphate-buffered saline (ph 7.0) in a 400- ml plastic bag. Samples were homogenized in a stomacher (Mix- 1, AES Laboratoire, Microbiology International, Frederic, Md.) for 90 s and left to settle for 5 min. Ten-fold dilutions of the liquid phase in buffered saline were plated onto Edwards modified medium. Plates were incubated aerobically for 24 h at 37 C. Streptococcal isolates were identified following the same procedures as described for BTM samples. Statistical analysis. Results from the 48 herds with bacterial count data were screened for abnormal or missing values, and counts were normalized by log transformation. Associations between count data were examined in scatter plots to determine whether nonlinear associations between variables existed. Nonlinear associations were not detected, and PROC REG (SAS release 8.02, SAS Institute, Inc., Cary, N.C.) was used to model the association of TBC with streptococcal, staphylococcal, and gramnegative bacterial counts. Standardized residuals were evaluated for significant associations with independent variables or predicted values to detect possible bias in the model. Results were considered significant at P 0.05. RESULTS AND DISCUSSION Qualitative bacteriology of BTM. Figure 1 shows the microorganisms that were identified by routine bacteriology of 200 raw BTM samples collected on farms in New York State. Of those 200 BTM samples, 68 (34%) contained S. agalactiae and Streptococcus spp., 1 (0.5%) contained S. agalactiae but not other Streptococcus spp., and 126 (63%) contained Streptococcus spp. but not S. agalactiae. Four BTM samples (2%) tested negative for the presence of streptococci. Similar results were reported by Goldberg et al. (8), who identified S. agalactiae in 32% and other streptococci in 98% of BTM samples from 1,971 Vermont dairy herds. By contrast, S. agalactiae was identified in only 11.4% of 114 BTM samples from Pennsylvania (18) and in none of the 172 BTM samples from Ohio (12), indicating regional differences in microbial quality of raw milk, particularly with respect to S. agalactiae.

2646 ZADOKS ET AL. J. Food Prot., Vol. 67, No. 12 FIGURE 1. Number of raw bulk tank milk (BTM) samples out of 200 total samples showing growth of specified microorganisms after 48 h of aerobic culture at 37 C on Trypticase soy blood esculin agar. Others category includes contaminated samples (n 3), Escherichia coli (n 2), Klebsiella (n 1), Proteus (n 4), Prototheca (n 5), and Pseudomonas (n 5). Bacterial counts. Using differential media to analyze BTM from the 48 selected farms, 48, 45, and 27 samples were positive for streptococci, staphylococci, and gramnegative bacteria, respectively. Bacterial counts (CFU/ml) were as follows: streptococci, 1,000 to 400,000 (median, 11,750); staphylococci, 100 to 38,000 (median, 1,550); gram-negative organisms, 50 to 21,500 (median, 400). The distribution of log-transformed streptococcal, staphylococcal, and gram-negative bacterial counts in tanks positive for them is shown in Figure 2. Goldberg et al. (8) also reported that streptococcal counts from BTM were higher than coliform counts, and staphylococcal counts were between streptococcal and coliform counts. There is conflicting information about the effect of freezing of milk samples on viability of bacteria. Schukken et al. (19) showed that freezing and increased length of storage resulted in a decrease in the number of samples that were culture positive for E. coli or Arcanobacterium pyogenes and an increase in the number of samples that were culture positive for coagulase-negative staphylococci. FIGURE 2. Distribution of differential bacterial counts among 48 raw bulk tank milk samples. Streptococci were detected in 48 samples using Edwards modified medium. Staphylococci were detected in 45 samples using Vogel-Johnson medium. Gram-negative bacteria were detected in 27 samples using MacConkey agar. Categories on the x axis indicate the natural logarithm of differential bacterial counts (3 indicates 3.0 up to not including 4.0; 4 indicates 4.0 up to but not including 5, etc.). Freezing had no effect on streptococci and Staphylococcus aureus when samples were stored frozen for a minimum of 4 weeks (19). By contrast, Murdough et al. (14) reported that freezing of milk samples for less than 6 weeks did not affect the viability of E. coli. Recovery of S. aureus, Staphylococcus hyicus, Staphylococcus chromogenes, Staphylococcus xylosus, S. agalactiae, Streptococcus dysgalactiae, S. uberis, and Corynebacterium bovis also was not affected. In a study by Dinsmore et al. (4), overnight freezing of milk samples before culture yielded a significantly higher positive culture rate overall but did not affect the positive culture rate of any individual bacterial species, including environmental streptococci and coliform bacteria. According to Villanueva et al. (22), freezing of milk samples for 23 days before culture increased the number of samples positive for S. agalactiae and S. aureus, but data on other bacterial species were not provided. Results from these studies (4, 14, 19, 22) suggest that prolonged frozen storage may affect culture results but short-term freezing has a very limited impact on culture results. Thus, because BTM samples were stored frozen for only a limited number of days, it seems unlikely that freezing had a significant effect on our results. If anything, coliform numbers may have been slightly underestimated after freezing, but this effect would not explain the several-log difference observed between coliform and streptococcal counts. Staphylococcal numbers may have been slightly overestimated after freezing. Nevertheless, staphylococcal numbers detected in our samples were lower than streptococcal numbers, which supports our conclusion that streptococci were the most important contributors to bacterial numbers in BTM samples. Streptococcal, staphylococcal, and gram-negative bacterial counts were all significantly and positively associated with TBC (P 0.0001, P 0.03, and P 0.04, respectively; parameter estimates of 0.53, 0.08, and 0.05, respectively). Based on partial R 2 values, streptococcal counts accounted for 69% of model variability, with staphylococcal and gram-negative bacterial counts each explaining 3% of model variability. The strength of the correlation of streptococcal counts with TBC in our study is between values reported previously from the United Kingdom for the cor-

J. Food Prot., Vol. 67, No. 12 STREPTOCOCCI IN BULK TANK MILK 2647 FIGURE 3. Streptococcal species identified by API 20 STREP analysis of three to five randomly selected gram-positive, catalase-negative isolates each from 48 raw bulk tank milk samples. relation of streptococcal counts on streptococcal selective agar with total colony counts on Yeastrel milk agar (r 0.98) (13) and from Vermont for correlations between S. agalactiae counts or non S. agalactiae counts on Trypticase blood esculin agar and SPC (r 0.28 and 0.43, respectively) (8). Differences in culture media used in the various studies and differences in the predominant flora in raw BTM are likely contributors to differences in correlation coefficients. In all reports, correlations between streptococcal counts and total counts were positive, significant, and for non S. agalactiae stronger than correlations between total counts and other reported differential counts. TBC and streptococcal counts were significantly higher (P 0.01) in BTM samples from herds with somatic cell counts exceeding PMO limits of 750,000 cells per ml than in BTM samples from herds that did not exceed this limit. For staphylococcal and gram-negative bacterial counts, the difference was not significant (P 0.05 and 0.78, respectively). Seven of 48 BTM samples had TBCs in excess of 100,000 CFU/ml. In four samples, streptococci alone were present in numbers exceeding this threshold, and streptococcal counts were high (93,000 and 75,000 CFU/ml) in two additional samples. The seventh sample contained elevated numbers of streptococci (33,000 CFU/ml) and staphylococci (24,800 CFU/ml). In milk that is cooled according to PMO guidelines, S. agalactiae and S. uberis do not multiply (5, 9). Hence, high numbers of these streptococci in properly cooled BTM are likely to be the result of high numbers of streptococci entering the bulk tank at the time of milking. Streptococcal species identification. Figure 3 shows the species identified among gram-positive, catalase-negative isolates from 48 BTM samples. One, two, or three Streptococcus, Enterococcus, or Aerococcus species were identified among three to five randomly picked isolates from 12, 21, and 15 BTM samples, respectively. S. uberis was the most frequently identified streptococcal species (n 39 of 48 BTM samples), followed by Aerococcus viridans types 2 (n 15) and 3 (n 12), and S. agalactiae (n 11). In four of the seven samples exceeding PMO limits for BTM, S. uberis was detected (twice as the only identified streptococcal species, once in combination with A. viridans types 2 and 3, and once in combination with A. viridans type 2 and S. dysgalactiae), whereas in the other three samples S. agalactiae was identified (once as the only species, once in combination with A. viridans type 2, and once in combination with A. viridans type 2 and Streptococcus acidominimus). Aerococci have been described as saprophytic bacteria, as common airborne organisms in human hospitals, and as marine pathogens affecting lobster (6). They have also been isolated from mammary secretions (3) and BTM (18). Little is known about the origin and impact of Aerococcus in dairy herds or milk. Among 85 gram-positive, catalase-negative isolates from bedding material in 21 herds, the majority (n 62) were A. viridans (5, 29, and 28 isolates of types 1, 2, and 3, respectively). By contrast, S. uberis was only identified once among the bedding isolates. Enterococcus faecium, Enterococcus faecalis, S. acidominimus, and Streptococcus bovis group isolates were found multiple times (n 5, 3, 8, and 4, respectively, among 21 bedding samples), but most of these species were rarely encountered in BTM samples (n 1, 2, 1, and 6, respectively, among 48 BTM samples). Although these data are limited, results suggest that the presence of bacteria in bedding is not necessarily correlated with the microbial flora of raw BTM. Theoretically, the fact that A. viridans is airborne (6) may account for the difference between the prevalence of A. viridans and that of other bacterial species in BTM, but further study is necessary to elucidate the route of contamination of raw BTM with aerococci. Strain typing. S. agalactiae in BTM samples is assumed to have originated from infected cows (7, 9, 11). S. uberis, however, can be found in the cow s environment (2). It is therefore commonly assumed that the presence of S. uberis in BTM usually results from contamination of teat skin or milking equipment rather than from intramammary infection (7, 11). S. uberis was rarely found in bedding samples in our study, but it was almost invariably identified from cows in herds with S. uberis positive BTM: 36 of 39 herds with S. uberis in BTM had S. uberis positive cows, and for the remaining 3 herds complete bacterial species data at the cow level were not available. Thus, we evaluated the hypothesis that cows are the source of S. uberis in raw BTM. The low numbers of coliform bacteria and fecal enterococci encountered in BTM supported the hypothesis that contamination of teat skin or equipment was not the only or main source of elevated bacterial counts in our study. Many different strains of S. uberis exist (24). If cows are the source of S. uberis in BTM, BTM isolates and cow isolates must belong to the same strain of S. uberis. To our knowledge, comparisons of S. uberis strain types from cow milk and BTM have not been published. In our BTM study, strain typing was done for a limited number of herds meeting the following conditions: (i) 100% of lactating cows had been sampled during the herd survey, (ii) isolates from all Streptococcus-infected cows were used for species iden-

2648 ZADOKS ET AL. J. Food Prot., Vol. 67, No. 12 TABLE 1. Streptococcus uberis ribotypes detected in raw bulk tank milk and composite cow milk samples from five dairy herds where all cows were surveyed and all Streptococcus-positive cows were tested for presence of S. uberis

J. Food Prot., Vol. 67, No. 12 STREPTOCOCCI IN BULK TANK MILK 2649 tification, (iii) S. uberis was detected in BTM and in at least one cow, and (iv) multiple S. uberis isolates from BTM were available. Thus, each S. uberis infected cow in the herd had a chance of being detected, comparison of cow and BTM isolates was possible, and strain diversity of BTM isolates could be explored. In total, 13 BTM isolates and 13 cow isolates from five herds were used for strain typing by means of ribotyping with PvuII (Table 1). In each herd, a dominant ribotype was identified among BTM S. uberis isolates; four of five herds had only one ribotype among two or three isolates. In each of the five herds, at least one cow shed the dominant ribotype found in BTM. In two herds (Table 1, herds A and B), streptococcal counts were more than 100,000 CFU/ml. Thus, strain typing data support the hypothesis that infected cows are the main source of S. uberis in raw BTM and can contribute to bacterial counts in excess of PMO limits. Whether this hypothesis is true only for short-term elevations of BTM counts, such as described by Hayes et al. (10), whereas consistently high bacterial counts are more likely to be caused by contaminated equipment (17), remains to be established. Ribo- Printer patterns for all isolates can be accessed on-line in Pathogen Tracker 2.0 (http://www.pathogentracker.net) and are identified by a code (e.g., 116-799-5) consisting of instrument identification (i.e., 116) and pattern identification (e.g., 799-5). In four herds (Table 1, herds A, B, C, and D), additional S. uberis ribotypes were found in cow milk samples but not in BTM samples. Failure to detect in the BTM samples all ribotypes that were shed by cows is likely the result of differences among cows in shedding of mastitis pathogens (CFU per milliliter of milk) and to a lesser extent the result of differences among cows in amounts of milk contributed to the BTM. Infected udders can shed as many as 10 9 CFU/ml (17) or as few as 10 2 CFU/ml (24). Even after a 10 4 dilution step, equivalent to one infected quarter in a 2,500-cow herd, one infected quarter shedding 10 9 CFU/ml of milk could result in a TBC of 10 5 CFU/ml, in excess of PMO limits. However, a shedding level of 10 2 CFU/ml would not result in high bacterial counts in BTM. A bacterial strain from a cow shedding such low levels of bacteria would have a low probability of being isolated, especially when only a limited number of isolates per BTM sample is selected for strain typing, as in our study. For S. agalactiae, a correlation between the proportion of shedding cows and counts in BTM exists in many but not all herds, and discrepancies between the proportion of shedding cows and counts in BTM are attributed to variation in levels of shedding among cows (7, 9). For other streptococci, the correlation between proportion of shedding cows and BTM counts has previously been reported as very weak (R 2 0.08) (9), in agreement with results from our study (Fig. 4; R 2 0.02 for proportion of Streptococcus-infected cows versus log-transformed streptococcal count in raw BTM). For non S. agalactiae streptococci, poor correlation between the proportion of Streptococcus-infected cows and streptococcal numbers in BTM is often attributed to the fact that environmental sources may harbor non S. agalactiae streptococci (7, 9). However, our strain typing data indicate FIGURE 4. Relation between proportion of cows testing positive for Streptococcus agalactiae and/or other Streptococcus species and streptococcal counts in raw BTM. R 2 0.02 for log-transformed counts. that infected cows are a likely source of high numbers of S. uberis in BTM and suggest that poor correlation between number of infected cows and streptococcal counts in BTM may at least in part be the result of variability among cows in the level of bacterial shedding. In one herd (Table 1, herd C), a ribotype was detected in BTM but not in a cow milk sample. This mismatch could be due to an infected cow going undetected, a mixed infection in a cow going undetected, or the isolate originating from teat skin or environmental contamination rather than from mastitic milk. Analysis of qualitative, quantitative, and strain typing data from 48 herds in New York State revealed that (i) streptococci are major contributors to total bacterial counts in raw BTM and may be present at levels in excess of the PMO limit of 100,000 CFU/ml; (ii) S. uberis is the most commonly identified streptococcal species in BTM, followed by A. viridans and S. agalactiae; and (iii) bedding material is a likely source of A. viridans, but infected cows are a major source of S. uberis. These results indicate that control of streptococcal mastitis, specifically that caused by S. uberis and S. agalactiae, will contribute to improvement of the microbial quality of raw BTM. ACKNOWLEDGMENTS This work reflects milk producer commitment to consumer product quality, as demonstrated by financial support of Cornell University research by New York State dairy farmers through the NYS Dairy Promotion Order. The authors thank QMPS field and laboratory personnel, specifically Robin Manthei, for sample collection and processing, Katy Windham for assistance with ribotyping, and Martin Wiedmann for critical review of the manuscript. The Pathogen Tracker 2.0 on-line public subtyping database is supported by U.S. Department of Agriculture special research grants 2001-34459-10296 and 2002-34459-11758 (to Martin Wiedmann). REFERENCES 1. Boor, K. J., D. P. Brown, S. C. Murphy, S. M. Kozlowski, and D. K. Bandler. 1998. Microbiological and chemical quality of raw milk in New York State. J. Dairy Sci. 81:1743 1748. 2. Bramley, A. J. 1982. Sources of Streptococcus uberis in the dairy herd. I. Isolation from bovine faeces and from straw bedding of cattle. J. Dairy Res. 49:369 373.

2650 ZADOKS ET AL. J. Food Prot., Vol. 67, No. 12 3. Devriese, L. A., J. Hommez, H. Laevens, B. Pot, P. Vandamme, and F. Haesebrouck. 1999. Identification of aesculin-hydrolyzing streptococci, lactococci, aerococci and enterococci from subclinical intramammary infections in dairy cows. Vet. Microbiol. 70:87 94. 4. Dinsmore, R. P., P. B. English, R. N. Gonzalez, and P. M. Sears. 1992. Use of augmented cultural techniques in the diagnosis of the bacterial cause of clinical bovine mastitis. J. Dairy Sci. 75:2706 2712. 5. Dogan, B., and K. J. Boor. 2004. Short communication: growth characteristics of Streptococcus uberis in UHT-treated milk. J. Dairy Sci. 87:813 815. 6. Evans, J. B. 1985. Genus Aerococcus, p. 1080. In P. H. A. Sneath, N. S. Mair, M. E. Shapre, and J. G. Holt (ed.), Bergey s manual of systematic bacteriology. Williams & Wikins, Baltimore. 7. Farnsworth, R. J. 1993. Microbiologic examination of bulk tank milk. Vet. Clin. North Am. Food Anim. Pract. 9:469 474. 8. Goldberg, J., J. W. Pankey, P. A. Drechsler, P. A. Murdough, and D. B. Howard. 1991. An update survey of bulk tank milk quality in Vermont. J. Food Prot. 54:549 553. 9. Gonzalez, R. N., D. E. Jasper, R. B. Bushnell, and T. B. Farver. 1986. Relationship between mastitis pathogen numbers in bulk tank milk and bovine udder infections in California dairy herds. J. Am. Vet. Med. Assoc. 189:442 445. 10. Hayes, M. C., R. D. Ralyea, S. C. Murphy, N. R. Carey, J. M. Scarlett, and K. J. Boor. 2001. Identification and characterization of elevated microbial counts in bulk tank raw milk. J. Dairy Sci. 84: 292 298. 11. Jayarao, B. M., and D. R. Wolfgang. 2003. Bulk-tank milk analysis a useful tool for improving milk quality and herd udder health. Vet. Clin. North Am. Food Anim. Pract. 19:75 92. 12. Khaitsa, M. L., T. E. Wittum, K. L. Smith, J. L. Henderson, and K. H. Hoblet. 2001. Characteristics and management practices associated with milk production in dairy herds in Ohio enrolled in official Dairy Herd Improvement Association programs. Am. J. Vet. Res. 62: 1262 1265. 13. McKinnon, C. H., G. J. Rowlands, and A. J. Bramley. 1990. The effect of udder preparation before milking and contamination from the milking plant on bacterial numbers in bulk milk of eight dairy herds. J. Dairy Res. 57:307 318. 14. Murdough, P. A., K. E. Deitz, and J. W. Pankey. 1996. Effects of freezing on the viability of nine pathogens from quarters with subclinical mastitis. J. Dairy Sci. 79:334 336. 15. National Mastitis Council. 1999. Laboratory handbook on bovine mastitis. National Mastitis Council, Madison, Wis. 16. Peeler, J. T., J. W. Messer, R. L. Sanders, and H. K. Bachelor. 1989. A Comparison of preliminary incubation counts and standard plate counts of grade a farm bulk tank milk from eleven states. Dairy Food Environ. Sanit. 9:494 497. 17. Saran, A. 1995. Disinfection in the dairy parlour. Rev. Sci. Tech. OIE 14:207 224. 18. Sawant, A. A., S. R. Pillai, and B. M. Jayarao. 2002. Evaluation of five selective media for isolation of catalase-negative Gram-positive cocci from bulk tank milk. J. Dairy Sci. 85:1127 1132. 19. Schukken, Y. H., F. J. Grommers, J. A. Smit, D. Vandegeer, and A. Brand. 1989. Effect of freezing on bacteriologic culturing of mastitis milk samples. J. Dairy Sci. 72:1900 1906. 20. U.S. Department of Health and Human Services. 2001. Grade A pasteurized milk ordinance, 2001 revision, publication no. 229. Public Health Service, Food and Drug Administration, Washington, D.C. 21. Van Schaik, G., M. Lotem, and Y. H. Schukken. 2002. Trends in somatic cell counts, bacterial counts, and antibiotic residue violations in New York State during 1999 2000. J. Dairy Sci. 85:782 789. 22. Villanueva, M. R., J. W. Tyler, and M. C. Thurmond. 1991. Recovery of Streptococcus agalactiae and Staphylococcus aureus from fresh and frozen bovine milk. J. Am. Vet. Med. Assoc. 198:1398 1400. 23. Wilson, D. J., R. N. Gonzalez, and H. H. Das. 1997. Bovine mastitis pathogens in New York and Pennsylvania: prevalence and effects on somatic cell count and milk production. J. Dairy Sci. 80:2592 2598. 24. Zadoks, R. N., B. E. Gillespie, H. W. Barkema, O. C. Sampimon, S. P. Oliver, and Y. H. Schukken. 2003. Clinical, epidemiological and molecular characteristics of Streptococcus uberis infections in dairy herds. Epidemiol. Infect. 130:335 349.