The PNC present in aseptically collected milk samples of cattle with IMI comprise a large group of Streptococcus spp., Lactococcus spp., Enterococcus spp., and Aerococcus spp. . In spite of this large spectrum, Strep. uberis, Strep. dysgalactiae, and Strep. agalactiae are important mastitis pathogens whereas the remaining PNC are less frequently observed or nonpathogenic. Thus, it appears reasonable, at least for the situation observed in Swiss farms, to focus the diagnostics on these 3 major PNC pathogens .
Traditionally, bacterial identification has been performed by conventional phenotypic methods [4, 10]. The results of the present study indicate that the power of these methods to discriminate PNC is limited. In the case of Strep. uberis, the combination of the results of 9 classical tests resulted in a probability of 92% that a strain showing the pattern defined as typical for this organism was actually a member of this taxon. These findings are in good agreement with the results of Odierno et al. who achieved correct identification in 94% of the cases by use of 11 biochemical tests. Khan et al. described the cultural, biochemical, serological and molecular properties of 132 strains of Strep. uberis. While all Strep. uberis strains in the present study were also ESC positive, which is in good agreement with the findings of the cited publications, other PNC (i.e. L. garvieae and A. viridans) were 100% ESC positive as well. These results confirm that correct phenotypic differentiation of PNC requires the use of multiple tests. The present study also demonstrates the phenotypic variability within the Strep. uberis species as the present strains differed from those of Khan et al. with respect to the frequency of alpha-hemolysis and of positive reactions for INU, HIP and PYR. The probability of correct identification was lower for Strep. dysgalactiae (90%) than for Strep. uberis, but high for Strep. agalactiae (100%). Performing 9 phenotypic tests requires considerable material and manpower so that only a few of them are carried out routinely by diagnostic laboratories. In particular, final identification of Strep. agalactiae is usually based on the CAMP reaction alone. Indeed, all Strep. agalactiae strains of the present study showed a positive reaction. Only a few strains of Strep. uberis also showed positive results. However, using the probability calculation of Willcox (1973)  for the CAMP reaction alone, the probability was only 87% that an unknown PNC with a positive reaction actually belonged to the Strep. agalactiae taxon. Similar results have been described previously . Nevertheless, the CAMP reaction as a single test seems to possess the best discriminatory power in the present study.
Some strains of Strep. uberis and L. garvieae showed very similar phenotypic properties. These findings indicate that discrimination of these bacteria is difficult if only one or a few of the evaluated phenotypic assays including ESC are used, therefore false identification may happen. As a consequence, the occurrence and importance of L. garvieae as a mastitis pathogen might have been underestimated in the past.
The multiplex PCR used in this study showed a high analytical sensitivity and specificity. All strains of the 3 major pathogens were correctly identified and no false positive results were noticed, which is in good agreement with the results of Phuektes et al. (2001)  or Gillespie et al. (2005) . Koskinen et al. (2010) showed similar results for a commercially available multiplex PCR (Patho Proof®) . Unfortunately, as some primers applied in this particular commercial PCR test are not published, the analysis of milk samples under monopoly conditions is considerably more expensive than the classical culture method. Phenotypic methods are therefore favored by many practitioners. The results of the present study show that reliable identification of the clinically most relevant PNC (Strep. uberis, Strep. agalactiae and Strep. dysgalactiae) can be obtained by use of a combination of colony morphology, hemolysis type, and catalase reaction as a first identification for triage and a multiplex PCR with specific primers restricted to these 3 pathogens.
The newer method MALDI-TOF MS appears to be a reliable tool to identify the entire spectrum of relevant bovine PNC associated with IMI. Indeed, a high level of agreement between this method and 16S rRNA sequencing (used here as the reference method for all isolates in the study) was observed for most of the PNC including Strep. uberis, Strep. agalactiae, and L. garvieae. Unsatisfying results, however, were observed for Strep. dysgalactiae. In comparison with 16S rRNA sequencing, only 2 analyzed strains were identified as Strep. dysgalactiae subsp. dysgalactiae. These findings are in agreement with those of Bizzini et al. and Van Veen et al. who suspected that the discordant results were due either to difficulties to differentiate among closely related organisms, such as e.g. Strep. dysgalactiae and Streptococcus pyogenes, or to insufficient reference spectra in the MALDI-TOF MS database. In the present study, all analyzed strains of Strep. dysgalactiae were allocated to the same cluster by the MALDI-TOF MS method. Although there were difficulties at the subspecies level, identification by MALDI-TOF MS was always correct at the species level when the broad cluster variation observed in the present study was considered. Correct identification at least at the species level, as achieved in the present study, should thus be sufficient for clinical purposes. The MALDI-TOF MS is fast and allows the identification of a broad spectrum of bacteria aside of PNC. These properties support the use of this methodology in diagnostic laboratories . The authors determine a major drawback of MALDI-TOF MS, at least at present, in the fact that the necessary devices are very expensive, which limits their use in routine diagnostic. In contrast, PCR is established for routine milk analyses in diagnostic laboratories. However, compared to MALDI-TOF MS, the spectrum of bacteria covered by one analysis is limited as it depends on the number of primer pairs included in one PCR. The spectrum can be enlarged by running several PCR in parallel (i.e. multiplex PCR). Using milk culturing on BA, the combination of colony morphology, hemolysis type, and catalase reaction allows a first identification for triage. Based on these easy and fast steps, final identification of Strep. uberis, Strep. dysgalactiae, and Strep. agalactiae by the presented PCR approach is efficient and highly specific. Indeed, all strains of these 3 pathogens were correctly identified and differentiated from the other PNC. Final results can be obtained within 12 hours.
Despite the low numbers of some strains, the results of the present study suggest that the combination of limited conventional microbiology and PCR with a fast DNA extraction protocol allow rapid identification of major mastitis pathogens belonging to the PNC group. This approach is expected, therefore, to be affordable for routine diagnostic laboratories, with the benefit of a considerably more accurate identification of PNC than can be achieved using even a combination of multiple conventional tests.