Bulk milk could have a high concentration of MAP antibodies because a dairy herd has a high prevalence of MAP infected cows or because a few infected cows produce large quantities of MAP antibodies. Variation in milk MAP antibody concentration within groups of fecal-positive and groups of fecal-negative cows has been observed due to a small number of cows producing a high level of antibodies and changes in antibody production within individual cows over time . A study by van Weering et al.  showed that a 100-fold dilution of a positive individual milk ELISA sample may still result in a positive ELISA result, indicating that a few positive cows may impact the bulk milk ELISA result. Given the relatively small herd sizes in our study, it is clear that a few cows producing high concentrations of antibodies may influence bulk milk ELISA titers. The precise impact of a single ELISA positive cow would depend on the herd size, milk production, and the difference between the individual milk antibody concentration of a particular cow and the average antibody concentration of the rest of the herd. Therefore, some farms may have higher ELISA values than average because they have higher than average MAP infection rates or because they have some cows producing high concentrations of MAP antibodies. Careful comparison of individual cow milk ELISA tests and cow milk production level to a bulk tank test would be required to differentiate between farms that have several infected, average antibody-producing cows, and those that have a limited number of animals with high antibody titers.
In our model, the location of a farm (NY, OR, or WI) was not significantly associated with corrected OD. However, other studies have found that herds in the Midwest are more likely to be positive for Johne’s disease  and observe clinical signs of Johne’s disease in their herd . Herd size, another study design variable, despite not being significantly associated with ELISA results in this study has previously been found to be positively associated with herd infection status, with larger herds having a greater risk [14, 18]. Additionally, Wells and Wagner  observed a positive association between group housing for calves and the herd infection status whereas our model did not find a significant association between types of calf housing and ELISA result. Spreading manure on forage fields  and open water sources  have also been shown to be associated with a higher risk of MAP infection, but were not significantly associated with corrected OD in our model. Finally, other studies have demonstrated an increased risk of Johne’s disease with high parity and Jersey breed cows , although those variables did not remain in our multivariate model.
The significant seasonal effect, with MAP antibodies highest in the summer and lowest in the winter (Figure 1), represents a change in antibody secretion into bulk milk across the seasons. Seasonal calving could account for this periodic change in antibodies because milk antibodies are, on average, greatest at the beginning and end of lactation . A farm that uses seasonal calving would see a herd-level increase in bulk milk antibodies, including antibodies against MAP, during or shortly after the calving season when most of their cows are just starting to lactate. This could result in high optical densities from the detection of MAP-specific and non-specific antibodies. The bulk milk ELISA value of seasonal calving farms, the majority of which indicated that they attempt to calve their cows in the spring, was greatest during the spring. This supports a days-in-milk dependent change in antibody secretion as a potential explanation for the seasonal trend in MAP antibodies in milk (Figure 1). Thus, seasonal calving could explain some of the seasonal variation but not all since only 46 of the 233 farms in the multivariate model used seasonal calving and non-seasonal calving farms still show a seasonal trend in OD (Figure 1). The seasonal variation also exists in non-grazing conventional farms (Figure 1), suggesting that the seasonal variation in MAP antibodies is not limited to grazing farms, which may have inherent seasonality in their calvings due to seasonal nutritional differences regardless of planned seasonal calving practices. Additionally, the seasonality variable remains significant (P < 0.0001) after removing all grazing farms from the final multivariate model.
Another possible explanation for the seasonal change in antibodies is a seasonal fluctuation in MAP load. It is unlikely that MAP prevalence changes seasonally on any given farm but the MAP load in the environment or in the cows could potentially change with the seasons. A previous study found a higher prevalence of MAP-positive carcasses, as determined by ileum and lymph node cultures and PCR, in the spring than in other seasons . Additionally, an increase in viable MAP isolated from retail milk in the summer has been shown . Humoral responses are known to occur in subclinical MAP infections, which results in activated B cells producing antibodies . It has previously been suggested that MAP exposure can trigger antibody production in infection-resistant adult cattle, possibly resulting in an increased MAP ELISA titer . Thus it is possible that the increase in MAP load of individual animals during the spring and summer could result in an increased humoral immune response in herdmates and therefore increase antibody levels in milk.
If bulk milk MAP ELISA is used as an indicator of MAP infection, it needs to be corrected for the seasonal changes in MAP milk antibodies. Consistently sampling during only one season may need to be recommended in order to compare ELISA results across time at one farm or among farms. Our results would need to be confirmed by similar studies to make such recommendations with more confidence.
Both culling and keeping cows known to be MAP-infected are associated with an increase in bulk milk ELISA value compared to herds without a history of clinical Johne’s disease, which suggests that farms with protocols for MAP-positive cows in place are more likely infected with MAP and possibly have a higher MAP prevalence. The larger increase in bulk milk OD associated with keeping MAP-positive cows or culling them after calving compared to the small increase in OD associated with culling MAP-positive cows immediately was not statistically significant (P = 0.25). However, it has previously been suggested that culling cows immediately is the best method for controlling MAP antibodies in bulk milk. Lu et al.  demonstrated the importance of culling positive animals immediately after detection in order to control MAP transmission. MAP-positive cows that are kept in the herd for any period of time could increase the bulk milk MAP antibody titer by producing MAP antibodies in their milk and by re-exposing other cows, which may then begin secreting MAP antibodies into milk. This would be particularly true for cows with a progressive course of disease, in which antibodies rise rapidly as the cow progresses from moderate to heavy fecal shedding .