If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Instituto de Ciências Biomédicas Abel Salazar (ICBAS), Rua Jorge Viterbo Ferreira, nº 228, 4050-313 Porto, PortugalEPIUnit – Instituto de Saúde Pública da Universidade do Porto (ISPUP), Rua das Taipas, nº 135, 4050-600 Porto, PortugalDepartement of Veterinary Sciences, Escola Universitária Vasco da Gama (EUVG), Av. José R. Sousa Fernandes, 3020-210 Coimbra, Portugal
Instituto de Ciências Biomédicas Abel Salazar (ICBAS), Rua Jorge Viterbo Ferreira, nº 228, 4050-313 Porto, PortugalEPIUnit – Instituto de Saúde Pública da Universidade do Porto (ISPUP), Rua das Taipas, nº 135, 4050-600 Porto, PortugalLaboratório para a Investigação Integrativa e Translacional em Saúde Populacional (ITR), Porto, Portugal
Instituto de Ciências Biomédicas Abel Salazar (ICBAS), Rua Jorge Viterbo Ferreira, nº 228, 4050-313 Porto, PortugalEPIUnit – Instituto de Saúde Pública da Universidade do Porto (ISPUP), Rua das Taipas, nº 135, 4050-600 Porto, PortugalLaboratório para a Investigação Integrativa e Translacional em Saúde Populacional (ITR), Porto, Portugal
Instituto de Ciências Biomédicas Abel Salazar (ICBAS), Rua Jorge Viterbo Ferreira, nº 228, 4050-313 Porto, PortugalEPIUnit – Instituto de Saúde Pública da Universidade do Porto (ISPUP), Rua das Taipas, nº 135, 4050-600 Porto, PortugalLaboratório para a Investigação Integrativa e Translacional em Saúde Populacional (ITR), Porto, Portugal
Mycobacterium avium ssp. paratuberculosis (MAP) is the causative agent of bovine paratuberculosis, also known as Johne's disease. This infection is responsible for negative effects, ranging from reduction of milk production to reproductive compromise and increased susceptibility to other diseases such as mastitis. Contradictory information on the association between this infection and reproductive performance has been reported in dairy cows. The aim of this work was to investigate associations between individual cow MAP seropositivity and lifetime reproduction and production performance. MAP serum ELISA (IDEXX MAP Ac) results from all the 13,071 adult cows present on 191 farms and corresponding birth- and calving-date records obtained from the National Association for Genetic Improvement of Dairy Cattle were used. Cows and farms were classified as positive or negative, based on ELISA results. Outcomes assessed, for all cows and all calvings from first to fifth, were age at first calving (AFC), intercalving intervals (ICI) from first to fourth interval, and average milk production per day of productive cycle (Milk-305/ICI - a ratio between 305 d corrected milk production, for each lactation, and the number of days of the respective ICI). Multilevel mixed models were used to investigate the association of cows' MAP status with AFC, ICI and Milk-305/ICI. Three levels were considered in the models: “measurement occasion,” the first level, was nested within cows and cows were nested within farms. The “measurement occasion” is the time point where all the observed measures (between 2 successive parturitions, such as milk production and SCC) were referred to. Our results indicate that MAP positive Cows have a significant 14-d lower mean AFC than MAP negative ones. The overall average ICI in our study was 432.5 d (s.d. 94,6). The average ICI, from 1st to 4th, was not significantly affected by MAP seropositivity. No significant effect of MAP positivity was found on the overall ICI. In relation to Milk-305/ICI, MAP positive cows did not produce significantly less milk than negative cows, across their productive lifetime. We observed higher but non-significant Milk-305/ICI (Kg/day) in MAP positive cows. In our study, the proportion of MAP positive Cows within lactations remained similar across all lactations suggesting that seropositivity did not increased drop-off rate.
Interpretive Summary Mycobacterium avium ssp. paratuberculosis (MAP), the causal agent of Johne's disease, is a threat to the dairy economy. MAP has been associated with Crohn's and other diseases in humans. This work provides insight on key performance indicators of MAP seropositive and seronegative dairy cows, evaluated in data throughout four consecutive lactation periods. Birth to calving interval, intercalving intervals and milk production per day of production cycle were determined and its variation according to MAP status of cows and farms was investigated.
INTRODUCTION
Mycobacterium avium ssp. paratuberculosis (MAP) is the causative agent of bovine paratuberculosis (PTB), also known as Johne's disease. In its clinical presentation the infection is responsible for severe weight loss, diarrhea and death. The effects of subclinical infections, range from the reduction of milk production, to reproductive compromise and increased susceptibility to other diseases such as mastitis (
Association of paratuberculosis sero-status with milk production and somatic cell counts across 5 lactations, using multilevel mixed models, in dairy cows.
) after grouping cows in different risk levels based on 4 ELISA tests, found no significant differences in means of predicted calving intervals between cows in different MAP infection risk groups. However, they found significant associations between different subclinical PTB risk groups, for the number of inseminations per conception, days to first service and non-return rate at 56 d after insemination. These authors concluded that the differences among fertility traits were inconsistent and with no clear trend. The presence of MAP in reproductive related tissues and secretions was documented by
, but those authors concluded that reproductive performance was not impaired by MAP seroconversion. MAP was detected by quantitative real-time IS900 PCR in the follicular fluid from the reproductive tracts of cows originating from one infected farm (
). Vaginal and uterine flush fluids, were also PCR-positive for MAP in samples taken from cattle currently shedding MAP in their feces. The presence of MAP in different parts of the reproductive tract was seen in clinically as well as subclinical infected cows. These findings extended currently scant and contradictory knowledge about the dissemination of MAP in the reproductive tract of female cattle (
MAP infection causes a malabsorption and maldigestion syndrome, as a consequence of the chronic progressive enteropathy. This syndrome reduces the efficiency of nutrient digestion, absorption and usage, eventually leading to negative energetic balance. Regardless of its cause, negative energetic imbalance is a common cause of altered reproductive health and fertility in dairy cows/farms. Interference in reproduction performance associated to this imbalance could be attributed to MAP, as well as in production performance of dairy cows (
). Likewise, if a negative reproductive impact caused by MAP infection occurs, milk production level will be compromised, as optimum milk production is linked to optimum reproduction performance. Therefore, the aim of this study was to investigate the reproductive performance and the associated productivity, of Portuguese dairy cows and farms, according to their MAP infection status. Thus, this investigation might help to clarify the importance of MAP status in dairy production. The objectives of this study were to investigate the association of MAP seropositivity with (1) age at first calving, (2) inter-calving intervals from first to fifth calving and (3) milk production per day of productive cycle (Milk-305/ICI), calculated as the coefficient of the 305-corrected milk yield and the length (in days) of the correspondent calving interval.
MATERIALS AND METHODS
This is a cohort study of retrospective nature. Animal and farm status, as well as productive and reproductive data will be described below.
Farm selection and categorization
Data from a sample of dairy farms and corresponding adult cow population was gathered as previously described (
Association of paratuberculosis sero-status with milk production and somatic cell counts across 5 lactations, using multilevel mixed models, in dairy cows.
). Records from all the cows with birth date and birth farm records, as well as the calving dates and milk production per lactation, were used.
Farms included in this study were those enrolled in a Johne's disease voluntary control program in Portugal, designated Bovicontrol run by SEGALAB animal health laboratories, which is a dairy farmer's sourced company operating in Portugal. Data were collected from August 11, 1997, to March 19, 2013.
The farm infection status was defined, based on all individual ELISA results available. All cows present in each farm, aged over 30 mo, were tested and the farms were categorized as follows: Strongly Negative Farms (SNEG) when all cows tested in the farm, minimum 60 test results (i.e., all adult animals present, minimum 20 animals tested), had negative results; Negative Farms (NEG) when all cows tested negative, but less than 60 tests were available; and Non-Negative Farms (NNEG) when among all cows tested there was at least one dubious result, or no more than one positive result. Positive Farms (POS) when all eligible cows present at the farm were tested and, at least 2 were POS cows.
In the statistical analysis, NNEG farms and the respective cows/lactations were excluded, to preclude classification bias and confusion associated with results. For the purpose of the analysis in this study, Farm Status (FS) was considered as NEG for the NEG and SNEG farms or POS otherwise. Criteria for data editing and inclusion were similar to the ones used in the previously published study (
Association of paratuberculosis sero-status with milk production and somatic cell counts across 5 lactations, using multilevel mixed models, in dairy cows.
), i.e., all cows from farms with less than 20 test results available were excluded from the analysis, as well as cows with non-valid lactation records, according to
All the cows aged over 30 mo, present in the farms, were MAP tested by indirect monophasic ELISA (IDEXX MAP Ab, IDEXX Laboratories, Inc. Westbrook, Maine USA), according to manufacturer instructions. The wells are coated with a protoplasmic extract of M. paratuberculosis; before the assay, the sera samples are incubated with an extract of Mycobacterium phlei to minimize possible cross-reactions with atypical mycobacteria. The individual test results were assigned as follows: S/P ≤ 45% were negative, S/P > 55% were positive and dubious if results fall between the 2. Cows' status (CS) was defined as positive (POS), negative (NEG), and dubious (DUB), based on all cows ELISA results. The tests were performed in a Laboratory operating under ISO 17025 accreditation system.
Individual cow records
Records relating to the number of lactation, respective milk production and cow's birth dates were obtained from National Association for Genetic Improvement of Dairy Cattle (NAGIDC) database, for all the cows present in farms selected to this study. Lactations were included if the production records were at least 305 d long, or corrected according to
if they were higher than 210 and less than 305 d. Cows birth date were validated from the Veterinary Authority individual identification database. All data from the different databases were merged using a double index key, farm number and ear tag national official number of the cow.
Calculation of metrics
Birth to Calving Intervals and Inter-Calving Intervals from first to fourth (ICI-1 to -4) were calculated, based on birth date and first to fifth calving dates. For each cow a maximum of 4 ICI were calculated as calving dates higher than the fifth calving were not used, to preclude bias resulting from the statistical analysis of classes with a small number of observations. Birth to first calving (in days) corresponds to the age at first calving (AFC). Extreme values of AFC and ICI, greater than the 99% percentile, were excluded from the analysis, as they could correspond to record errors. Milk per day of production cycle (Milk-305/ICI) was calculated for each cow's lactation, using the 305-corrected milk production from lactation n divided by the ICI number of days from calving n to calving n +1.
Collected data have a natural hierarchical structure and were analyzed using multilevel statistical models (MLM) considering the “measurement occasion” (first level) nested within cows (second level), and cows nested in farms (third level). It should be noted that MLM take into account different numbers of observations per cow, given that for some cows we may have observations for just one ICI/Lactation and for others, 2, 3 or more.
Lactation weighted arithmetic average SCC within the lactation, which presented a right skewed distribution, was log-transformed (lnSCC) so that its distribution became approximately normal. Multicollinearity was investigated through the Variable Inflation Factors.
Several multivariable multilevel models, having as dependent variable ICI or Milk305/ICI and as explanatory variables Lactation number, lnSCC and Milk305 (just for the ICI model) at first level, Cow status at second level, and interactions, were developed. Lactation squared was also explored mirroring the results of the previous study (
Association of paratuberculosis sero-status with milk production and somatic cell counts across 5 lactations, using multilevel mixed models, in dairy cows.
) . The quadratic term was included in the model Milk305/ICI to account for the effect of lactation number in the amount of milk produced, increasing from first to third and decaying afterward.
The categorical variable Cow status was coded as NEG and POS and NEG Cow status was considered the reference category.
Models were fitted using Maximum Likelihood Estimation. Variables ln SCC and Milk305 were centered on the respective means. As these predictor variables were on very different scales, rescaling was considered so that the range of values of the centered variables was reduced and also because it facilitates the interpretation of the response variable in terms of the model coefficients.
All models had at least 3 variance components: a residual variance at level 1, random intercept variances at level 2 and level 3, allowing for the 3-level data structure. To decide which variables, interactions and random terms should be included in the final models, the significance of the fixed effects was assessed by F-tests and p-values using Satterthwaite approximation to the denominator degrees of freedom, whereas random effects were evaluated using Likelihood Ratio Tests. The final models included only the significant random effects. The significance level was set at 0.05. Models were developed using package “Imer Test” (
), in R Core Team (2013). Graphs were developed using R software.
RESULTS
In our working data set, there were 13,071 cows with first lactation records and 35,993 calvings, with the respective lactation records, from 191 farms from the Portuguese continental territory. From the 35,993 calvings/lactations available, 36.3% (n = 13,071) were first parity, 29.5% (n = 10,613) were second, 19% (n = 6,829) third, 10.2% (n = 3,687) fourth and 5% (n = 1,793) fifth parity. Average lactation number was 2.18 (s.d. = 1.17). The Milk-305 production followed an approximately normal distribution and was on average 9,458 kg (s.d. = 1,908) (Figure 1.). The distribution of the lnSCC of each lactation was slightly skewed to the right, with a mean value of 4.92 (s.d. = 1.10; Figure 2.). From total lactations available, 4.7% (n = 1,683) belonged to MAP POS cows. A subset of 20,826 lactations from this pool was used after excluding all cows for which there was incomplete information regarding age at first calving, ICI and lnSCC, which permitted a proper analysis with multilevel models. In this subset there were 1003 MAP POS lactations (Table 1).
Figure 1Histogram of Milk-305 (Milk 305 = 305 d corrected milk production of the lactation, Kg).
The AFC was analyzed according to cow or farm MAP status. AFC (measured in days) was calculated first for a data set of all cows with available birthdate and first calving records. In this data set there were 11,543 cows, of which 523 POS. Second, AFC was analyzed in the subset of data (20,826 lactations) resulting after the application of exclusion criteria necessary for multilevel analyses; there were just 8,247 AFC records of which 403 POS. Of these, 5,209 AFC were in MAP POS farms and 3,038 in MAP NEG farms.
A Students' t-test for independent samples was used to assess differences between the average AFC of cows from MAP-POS and NEG farms (regardless of cow individual MAP status), and no significant difference was detected in both the data sets: AFC was 835.6 (s.d. = 112.4) days and 831.5 d (s.d. = 113.8) in NEG and POS farms, respectively (n = 11,543, mean difference of 4.1 d, t = 1.870, P = 0.061); in the subset, average AFC was 831.7 (s.d. = 110.0) days and 829.4 (s.d. = 114.8) days in NEG and POS farms, respectively (n = 8,247, mean difference of 2.2 d, t = 0.869, P = 0.385).
When comparing AFC for NEG and POS cows (regardless of farm calving MAP status), different results from both data sets were found. In the data set (n = 11,543) the mean AFC was 833.6 (s.d. = 113.5) days in MAP-NEG cows and 819.6 (109.8) days in MAP-POS cows, which corresponds to a significant mean difference of 14 d (t = 2.759, P = 0.006). However, the difference in AFC from the subset (n = 8,247) decreased, and lost significance: the mean values were 830.7 (s.d. = 113.1) days and 821.3 (s.d. = 112.8) days for NEG and POS cows, respectively (mean difference of 9.4 d, t = 1.621, P = 0.105) (Table 2).
Table 2Students' t-test for age at first calving, in days. Mean differences according to cow or farm MAP status, calculated for both the full data set (all cows) and the subset
Number of AFC in analysis
Mean Difference
Std. error difference
p
Differences in average age at first calving according to farm MAP status
subset 8,247
2.243
2.581
0.385
full data set 11,543
4.136
2.211
0.061
Differences in average age at first calving according to cow MAP status
Table 1 shows the number of lactations selected with complete information to calculate the ICI and to compute the models, distributed per Lactation number and Cow status. From the 20,826 lactations, 1,003 were MAP POS (4.8%) and the remaining 19,823 were MAP NEG. From this latter group, 7,713 were from MAP NEG farms. The proportion of lactations from the original data set and the ratio MAP POS/MAP NEG was approximately the same. Table 1 shows that the proportion of MAP POS lactations across the 4 Lactations, does not change significantly (χ2 = 0.514; P = 0.92), varying from 4.9% to 4.5%.
The MAP status of cows was assigned to their respective ICI (totalizing 1,003 MAP POS and 19,823 NEG ICI). ICIs higher than 829 d (corresponding to the 99% percentile) were censored to prevent the effect of extreme and unusual observations. The ICI-1, ICI-2, ICI-3 and ICI-4 had a relative frequency of 45.63% (n = 9,503), 30.10% (n = 6,268), 16.30% (n = 3,396) and 7.97% (n = 1,659) (Table 1.). The ICI average length was 432 d (s.d. = 94.6), distribution skewed to the left.
ICI model
In the MLM model for ICI, all the independent variables were highly significant, except cow status. The MLM model reveals that lactation number has a negative significant (P < 0.0001) effect on ICI length, whereas Milk305 production and lnSCC both show positive significant effects on ICI (P < 0.0001). After adjusting for the remaining variables, no significant effect of cow MAP positivity was found in ICI (P = 0.45) (Table 3). Interactions between the independent variables were assessed but, as they showed no significant effects, were not included in the final model.
Table 3Multilevel Model for the variation of ICI (in number of days)
Model Variables
Est. (SE)
95% Confidence Interval
P value
Intercept
444.354 (2.163)
440.112 – 448.621
<0.0001
Lactation
−5.294 (0.734)
−6.734 – −3.854
<0.0001
Milk 305 (centered)
4.547 (0.431)
3.6978 – 5.397
<0.0001
lnSCC (centered)
10.973 (0.681)
9.638 – 12.307
<0.0001
Cow Status NEG
reference
—
—
POS
−2.482 (3.298)
−8.948 – 3.986
0.452
Variance/Covariance
Level 3
Sigma (Intercept)
20.03
17.471 – 22.988
Level 2
Sigma (Intercept)
33.38
30.832 – 35.796
Level 1
Sigma (Residual)
85.84
84.706 – 86.995
−2Log-likelihood
123,773.5
Farms
191
Cows
10,780
Lactations
20,826
1ICI = Intercalving interval, in days; lnCCS = log-transformed lactation weighted arithmetic average somatic cell count; Milk 305 = 305 d corrected milk production of the lactation.
The ratio between Milk-305 from lactation “n” and the corresponding ICI (Milk 305/ICI, kg /day) is the outcome variable of the second model. The independent variables were Lactation and Lactation2, lnSCC, Cow Status and an interaction between lnSCC and Lactation.
Except for Cow Status (P = 0.052) all other variables had highly significant (P < 0.0001) effects on the ratio. The coefficients of the variables Lactation and Cow Status were positive while those of Lactation2 and lnSCC were negative. The interaction between lnSCC and Lactation showed also a negative effect (P < 0.0001). The impact of the interaction became more evident as lactations evolved from first to fourth. The model shows that MAP POS cows do not produce significantly less milk than MAP NEG cows across their productive lifetime (Table 4; Figure 3).
Table 4Multilevel Model for Milk305/ICI (kg/day) variation
Model Variables
Est. (SE)
95% Confidence Interval
P value
Intercept
16.92 (0.240)
16.446 – 17.388
<0.0001
Lactation
4.126 (0.181)
3.771 – 4.480
<0.0001
Lactation2
−0.588 (0.004)
−0.664 – −0.511
<0.0001
lnSCC (centered)
−0.498 (0.008)
−0.654 – −0.342
<0.0001
Cow Status NEG
reference
—
—
POS
0.369 (0.190)
−0.003 – 0.741
0.0519
lnSCC * Lactation
−0.146 (0.004)
−0.219 – −0.073
<0.0001
Variance/Covariance
Level 3
Sigma (Intercept)
2.234
1.966 – 2.5466
Slope (Lactation)
0.431
0.335 – 0.539
corr
0.130
−0.130 – 0.404
Level 2
Sigma (Intercept)
2.242
2.129 – 2.352
Level 1
Sigma (Residual)
4.601
4.541 – 4.662
−2Log-likelihood
127,239.5
Farms
191
Cows
10,780
Lactations
20,826
1ICI = Intercalving interval, in days; lnCCS = log-transformed lactation weighted arithmetic average somatic cell count; Milk 305 = 305 d corrected milk production of the lactation.
Figure 3 presents the expected curves built from the model parameters for Milk305/ICI (kg/day) where is possible to see a permanent higher, non-significant, level of Milk305/ICI for MAP POS cows. Figure 4 shows a lower significant Milk305/ICI (Kg/day) for the groups of cows above or below lnSCC 5.3 (200 000 SCC/ml) across consecutive lactations, regardless of MAP status.
Figure 4Milk305/ICI (Kg/day) variation for the group of cows above or below lnSCC 5.3 (200 000 SCC/ml), across consecutive lactations, built from the model parameters.
The variance/covariance structure of both models identifies that all the levels contributed significantly to variability of the association of independent variables and the outcomes.
DISCUSSION
Paratuberculosis is an untreatable infection that affects dairy cows, with a highly heterogenic progression dynamic at individual and farm level. The disease begins as a localized infection that may become systemic, although this does not always happen (
). Our findings corroborate the difficulties reported in different research approaches when attempting to provide clear evidence of the effects of MAP infection in dairy cows but provide new insights.
Overall findings
In this work we found that AFC of heifers from POS farms happens few days earlier than from NEG farms, although no statistical significance was found in both data sets. The AFC of MAP POS heifers is lower than AFC from MAP NEG mates, irrespective of farm status, and this difference was significant in one of the 2 data sets.
The average ICI length was not significantly different between MAP POS and NEG cows, suggesting the absence of effect from MAP status in this reproductive parameter. MAP POS cows did not produce significantly less milk than MAP NEG cows, across their productive lifetime. However, we did observe a higher, non-significant, Milk305/ICI (kg/day) in MAP POS cows compared with MAP NEG mates of the same lactation number. Results of our model (Figure 4) confirm the influence of lnSCC on milk production: as expected, using the cutoff value of 200,000 SCC/ml (lnSCC = 5.3), the Milk305/ICI was significantly lower for the lactations with SCC above the cutoff value, highlighting the effect that SCC has in milk production.
Relevant to emphasize is that the proportion of MAP POS cows remained similar across all lactations, from first to fourth, suggesting that MAP positivity did not influence cow drop-off from farm.
Design of the study
This study combines a cross-sectional design, regarding the MAP status characterization of cows and farms, with a longitudinal component in which the complete productive life of every cow was available. Consequently, the number of lactations and calving dates, from first lactation up to 4 used represents the productive lifespan of each cow. The assessment of the potential impact of MAP infection in reproduction and production efficiency was carried out during the entire lifetime of each cow, through the investigation of the effect of MAP status over AFC, ICI and Milk-305/ICI across 4 consecutive lactations. By assessing the productive life of each cow over consecutive and dependent 4 lactations, the lifelong pathogeny of MAP infection (
) was taken into consideration. Our models, using a 3-level structure, considered in the first level “measurement occasion,” nested within the respective cow, in turn nested within a farm. Lactation number, as independent variable, was included in the models, given its effect over the productive life of the cow (
The animal structure of our sample was composed by 10,780 cows and 20,826 lactations in 191 farms. The average Milk-305 production (9,458 kg), the average lactation of cows (1.93) and the distribution of observations across parity, closely follows the structure of the source NAGID population. A large number of farms belonging to a voluntary MAP monitoring program were included, independently of their a priori positivity, with the only inclusion criteria of having more than 20 cows tested.
Our study design was not meant to dissect the impact of different MAP infection progression stages in reproductive performance of individual cows, but rather, the lifetime impact of MAP on the combined reproduction and production efficiency of cows within farms. The different types of patterns for seroconversion or shedding described (
) were not specified in our models but may be present and most certainly influence the variance/covariance structure of the MLM.
Premature culling due to MAP
We observed that the relative proportion of MAP POS within the data set remained constant across lactations, from first to fourth (min. = 4.5%; max. = 4.9%). This finding suggests that the probability of cows being kept in the farm for the next production cycle, was not influenced by MAP positivity.
Age at first calving
In the assessment of AFC of POS versus NEG cows, irrespective of farm status a reduction of 14 d in AFC was seen in the data set with larger number of animals. A second data set was obtained after inclusion criteria imposed to develop multilevel models resulting in a smaller subset. In this second data set the average difference in AFC was 9.4 d, still lower for POS cows. No significant differences were observed in AFC of cows from POS and NEG farms, irrespectively of cows' status. These findings could suggest that the difference in AFC is not due to farm effects, but rather, most probably related to the same factors that render these cows apparently more susceptible to MAP infection.
Intercalving intervals
The sum of all productive days along the lifetime of each cow is a key performance indicator of major importance to the economic success of the farm. For dairy cows, the sum of nonproductive days, from birth to calving, is the result of the age at first calving (in days), the number of dry days, and from disease episodes occurring during lactation, usually with negligible contribution to the total sums. To keep average non-productive days as low as possible is vital for reproductive and productive efficiency of dairy operations. To assess the importance of dry days in the ICI of MAP POS and NEG cows, ICI length, from 1st to 4th lactations, were analyzed. The number of ICI days in the study data set decreased as parity increased, following the general observed culling rate of the cow census from dairy operations present in the NAGDIC (data not shown).
Our model shows that ICI length increases with lnSCC and Milk-305, making evident the impact of udder health and production level on reproductive performance. Inversely, ICI length decreases as lactation number increases, suggesting that short ICIs augments the odds of a cow being kept in the farm for next lactation. MAP status of the cows did not significantly influence the ICI, in spite of positive cows exhibiting a slightly lower ICI. AFC was not kept in the model.
Results from our models seem to be not biologically supported by previous reported results. Negative impacts of MAP infection in different aspects of the reproductive biology and performance of the cow have been reported by
, analyzed time to calving using a proportional rates model and found that high- shedding animals had lower calving rates in comparison with low-shedding or ELISA positive animals. Surprisingly, the same study reported that ELISA-positive animals tended to have higher calving rates than animals tested-negative. A recent study of calving interval was performed by
, in 3 large commercial farms, using one calving interval only, for each cow; in the study, the calving interval of MAP POS cows (based on a milk IDEXX Paratuberculosis Screening Ab Test, IDEXX Laboratories, Inc., Westbrook, ME, USA) was 33.8 d longer [95% CI: 13.2 54.4 d, P = 0.0013, (+9.7%)], when compared with the calving interval of negative cows, on average. Additionally, milk ELISA positive cows were less likely to conceive to first insemination (OR: 0.49) and required on average 0.42 more inseminations to conceive. Those findings are easily supported by biologically plausibility. Based on the contradictory evidence it is difficult to support a definitive conclusion about the effects of MAP in reproductive performance.
Productive efficiency of the cow
The average milk production per day of productive cycle Milk-305/ICI was calculated and analyzed in the attempt to better capture the impact of MAP infection on the production of dairy operations. Average daily milk production per productive ICI cycle is a primary key performance indicator contributing to diluting the fixed cost necessary to produce a heifer and keep a cow for its lifetime through productive and non-productive days. MAP POS cows show a slightly higher production, although not statistically significant. Milk-305/ICI was negatively affected by lnSCC which showed a significant interaction with lactation number. As in the previous model of ICI the AFC was not retained in the model. The fact that MAP POS cows were not producing less than the negative ones, in our model, is very relevant. The results from this study do not collide with our previous results, were we concluded that MAP POS cows produced less milk throughout 5 consecutive lactations, because in that first analysis (
Association of paratuberculosis sero-status with milk production and somatic cell counts across 5 lactations, using multilevel mixed models, in dairy cows.
), the average length of ICI was not accounted for.
Also relevant is the finding that there was no significant difference in the proportion of MAP POS cows that survive for consecutive calvings /lactations, when compared with the proportion of NEG cows, suggesting that the selection pressure for a cow to pass over to next lactation applies in the same manner regardless of MAP status of the cows.
Sources and control of bias
There was an effort to increase predictive positive and negative values of results used to assign cow status. Only one test was available for the majority of the cows, but overall specificity of the study design is considered high. The assignment of cow and farm MAP status is offered by the highly specific ELISA test used, reinforced by the criteria of assigning positive status only to the farms with 2 or more seropositive cows. The farm sensitivity was increased by including only farms with more than 20 cows tested, reinforcing predictive value of farm and cow negative status. Farms with only one POS cow were excluded from our analysis, to enhance the farm's predictive positive value. However, in our study, false negative animals could be present in MAP POS farms, resulting in an underestimation of associated effect.
The different types of patterns described for seroconversion or shedding (
) that are the result of MAP-host dynamics, were not specified in our models, but should certainly be present. We consider that it is expectable that different seroconversion and shedding patterns occur in our sample of cows, given the large sample of cows and farms used and the lifelong design adopted, even though they were not specified into the model. These would reflect the observed variability at lactation and at cow level.
Comparison of results between different studies
As seen from the results of our models, all the levels significantly contribute to variability in the outcomes assessed. Full external validity is difficult to claim given the differences in design compared with other studies and dairy production systems from other countries. When comparing designs from different studies it is difficult to draw definitive conclusions about the effect of MAP at cow and farm level. Furthermore, commercial dairy farms vary a lot across countries of the world in production, genetic composition, feeding and facilities, and management practices. The consequence being in making it difficult to extrapolate our results to other countries. The same applies from the extrapolation of results from other country-based studies to the rest of the world.
Nevertheless, it is useful to explore limitations often present in serological based studies. One is that the poor predicted negative value of diagnostic tests which enables false negative cows in the MAP positive farms, enhancing variance in the negative control group.
Another is the assignment of MAP status to the cows and to the farms. Current knowledge of MAP infection is that MAP infects the calves and affects cows for their lifetime (
), thus, a comprehensive assessment of effects and impact improves when taking into consideration the cow's complete productive lifetime, to encompass the different possible infection progression course, and the age-related effects in different animals within farms. In many studies only one lactation of the same animal is assessed, and the focus is on infected farms, excluding cows from negative farms from comparisons. Additionally, in different studies, infected animals are tested one or several times during one lactation and do not take into consideration the performance of these animals during periods before the test positive event, when even though testing negative, they were most probably already MAP infected, since majority of infections occurs early in life. A strong point in our study is the assessment of the complete life production of each cow and status assignment for the entire lifetime.
Association of paratuberculosis sero-status with milk production and somatic cell counts across 5 lactations, using multilevel mixed models, in dairy cows.
) one question arises: is there something particular in the cows that have the higher efficiency potential (productive and reproductive), that relates to MAP infection susceptibility or MAP infection progression and expression? New prospective studies designed to highlight the association between susceptibility to infection and genomics are needed to clarify the contribution of genetic dimension to MAP in dairy. Follow-up from birth to culling should be granted and all animals periodically tested within few days after each calving and in mid lactation, both with new high performing diagnostic tests and with conventional ELISA. The assessment of MAP status of the animals should start in the early life, and be based on new MAP diagnostic techniques that allow for early detection and differentiation of MAP gastrointestinal passage of even low MAP counts, and to access the presence of subclinical infection at early age (at least before first calving) (e.g., serum biomarker of MAP infection, serum α 2-Macroglobulins (A2M) levels, (
Preparation of immunomagnetic beads coupled with a rhodamine hydrazine immunosensor for the detection of Mycobacterium avium subspecies paratuberculosis in bovine feces, milk, and colostrum.
); metabolomic techniques using direct analysis in real time coupled to high resolution mass spectrometry (DART-HRMS), coupled with a mid-level data fusion approach, (
). Genomics, microbiomics, production and reproduction data, gathered from the same animals, could then be analyzed and correlated to the different MAP-cow test results and the corresponding patterns of MAP control by the immune system.
Our study did not highlight a negative association between MAP seropositivity and overall lifelong productivity and reproductive performance. This information should be integrated in culling decision models. Especially when test-and-cull is implemented independently of a cow individual milk production performance.
ACKNOWLEDGMENTS
SEGALAB, ANABL, BOVINFOR.
Disclosure of Interest: None Declared
REFERENCES
Alonso-Hearn M.
James Genovese K.
Pozzato N.
Prà D.M.
Buck D.J.
Tata A.
Pallante I.
Massaro A.
Miano B.
Bottazzari M.
Fiorini P.
Dal Prà M.
Paganini L.
Stefani A.
De Buck J.
Piro R.
Serum Metabolomic Profiles of Paratuberculosis Infected and Infectious Dairy Cattle by Ambient Mass Spectrometry.
Preparation of immunomagnetic beads coupled with a rhodamine hydrazine immunosensor for the detection of Mycobacterium avium subspecies paratuberculosis in bovine feces, milk, and colostrum.
Association of paratuberculosis sero-status with milk production and somatic cell counts across 5 lactations, using multilevel mixed models, in dairy cows.
00068-7&id=gr1.jpg){kind=link}
00068-7&id=gr2.jpg){kind=link}
00068-7&id=gr3.jpg){kind=link}
00068-7&id=gr4.jpg){kind=link}