Advertisement

Prospective cohort study of the relationship between milking machine liner slip, milking performance, and cow characteristics

Open AccessPublished:December 29, 2022DOI:https://doi.org/10.3168/jds.2022-22175

      ABSTRACT

      We conducted a prospective cohort study to investigate the associations of machine milking liner slip with (1) milking performance and (2) cow characteristics. Parlor data including milk flow characteristics and data on the occurrence of milking machine liner slips from a 4,000-cow dairy with a thrice-daily milking schedule were obtained with electronic on-farm milk meters over a 2-mo period. We analyzed data from a total of 686,330 milking observations. A multivariable general linear mixed model revealed no association between liner slip and milking unit on time. Least squares means (95% confidence intervals, 95% CI) were 237 (235–238) s for milking observations with and without a liner slip. We observed statistically significant differences in average milk flow rate; however, these were biologically irrelevant. Least squares means were 3.40 (3.37–3.42) kg/min for a milking observation with and 3.42 (3.40–3.44) kg/min without a liner slip. A multivariable generalized linear mixed model showed an association between liner slip and cow characteristics. Compared with late-lactation cows, the odds ratios (OR, 95% CI) of occurrence of a liner slip were 2.03 (1.59–2.59) in early lactation cows and 1.26 (0.97–1.64) in cows from 101 to 200 days in milk. Presence of a nonlactating quarter increased the odds of liner slip occurrence [OR, 95% CI: 10.35 (8.02–13.35)]. Bimodal milking observations had higher odds of occurrence of a liner slip compared with milking observations with a unimodal milk flow curve [OR, 95% CI: 1.05 (1.005–1.09)]. A 1-kg increase in 2-min milk yield increased the odds of a liner slip [OR, 95% CI: 1.26 (1.25–1.28)]. We conclude that, in the study cohort presented herein, the negative effect of liner slips on milking performance can be diminished. The identified cow characteristics could offer unique opportunities to identify and manage cows at increased risk of liner slips.

      Key words

      INTRODUCTION

      A milking machine liner slip, more commonly referred to as liner slip, is defined as “rapid air leakage past the mouthpiece of the milking machine liner” (
      • Spencer S.B.
      • Volz C.
      Measuring Milking Machine Liner Slips.
      ; p. 1,000) and the bovine teat skin during machine milking. Liner slips have been associated with increased risk of new IMI (
      • O'Callaghan E.
      • O'Shea J.
      • Meaney W.J.
      • Crowley C.
      Effect of milking machine vacuum fluctuations and liner slip on bovine mastitis infectivity.
      ;
      • Baxter J.D.
      • Rogers G.W.
      • Spencer S.B.
      • Eberhart R.J.
      The effect of milking machine liner slip on new intramammary infections.
      ). In addition, there is anecdotal evidence that liner slips interfere with the milking routine, increase the labor involved in milking, and create operator dissatisfaction (
      • Rogers G.W.
      • Spencer S.B.
      Relationships among udder and teat morphology and milking characteristics.
      ). Monitoring of liner slips is therefore paramount to achieve good parlor performance and udder health (
      • Stewart S.
      • Eicker S.
      • Rapnicki P.
      Automated collection of parlor performance data: Information needed and proposed standardized definitions.
      ). Ample research has focused on investigating the association between liner slips and udder health. Conversely, knowledge about their effect on milking performance is scarce.
      Multiple research groups have investigated the influence of milking machine equipment and settings on the occurrence of liner slips.
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      reported that the frequency of liner slips increased with lower operating vacuum. They further suggested that the design of the milking liner influences the occurrence of liner slips. Research work from Denmark confirmed the findings by
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      regarding the effect of milking vacuum (
      • Rasmussen M.D.
      • Madsen N.P.
      Effects of milkline vacuum, pulsator airline vacuum, and cluster weight on milk yield, teat condition, and udder health.
      ). In addition, the researchers found that a lighter milking cluster reduced the frequency of liner slips (
      • Rasmussen M.D.
      • Madsen N.P.
      Effects of milkline vacuum, pulsator airline vacuum, and cluster weight on milk yield, teat condition, and udder health.
      ). Although the cow effect has long been recognized as an important source of variation for liner slips (
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      ), little information is available about the association between cow characteristics and occurrence of a liner slip.
      Historically, occurrence of milking machine liner slips has been assessed by measuring vacuum fluctuations through installation of vacuum transducers (
      • Spencer S.B.
      • Volz C.
      Measuring Milking Machine Liner Slips.
      ;
      • Baxter J.D.
      • Rogers G.W.
      • Spencer S.B.
      • Eberhart R.J.
      The effect of milking machine liner slip on new intramammary infections.
      ;
      • Davis M.A.
      • Reinemann D.J.
      • Mein G.A.
      Effect of liner age on milking characteristics.
      ) or by audible assessment (
      • Rasmussen M.D.
      • Madsen N.P.
      Effects of milkline vacuum, pulsator airline vacuum, and cluster weight on milk yield, teat condition, and udder health.
      ). With the advancement in electronic on-farm milk meter technology, the occurrence of a liner slip during individual cow milking observations can be documented. In a recent study, our group evaluated the performance of the on-farm milk meter technology and compared it with that of audible detection and found moderate agreement between the 2 techniques (unpublished data). This offers unique opportunities to efficiently monitor liner slips in real-time in an objective manner during individual cow milking observations. The objectives of this study, therefore, were to investigate the associations of machine milking liner slip with (1) milking performance and (2) cow characteristics using electronic on-farm milk meters. Our hypothesis was that milking machine liner slips were associated with a longer milking unit on time and a lower average milk flow rate. We further hypothesized that cow characteristics were associated with the occurrence of liner slips.

      MATERIALS AND METHODS

      Animals and Housing

      This study was conducted between January and February 2020 at a commercial dairy farm located near Ithaca, New York. During the study period, approximately 4,100 lactating Holstein cows were housed in freestall pens, bedded with manure solids, and fed a TMR consistent with requirements outlined by the National Research Council (
      • NRC
      Nutrient Requirements of Dairy Cattle.
      ). Herd data were maintained in a dairy management software program (Dairy Comp 305, Valley Agricultural Software). The farm used DHIA services including the individual cow SCC option. The rolling herd key performance indicators were average milk production, 13,713 kg; mean test day SCC, 229,000 cells/mL; monthly clinical mastitis incidence, 9.5%; and 21-d pregnancy rate, 27.0%. The farm management used individual quarter dryoff to manage cows with recurrent clinical mastitis and chronic subclinical mastitis. Cows subjected to individual quarter dryoff were marked with a colored leg band (2 different colors, one indicating the front quarter and one indicating the hind quarter), that was put on the hind leg on the side of the affected quarter. The colored leg band directed the milking technicians to not attach the milking unit to the respective teat and thus, milking of the quarter was omitted. Additionally, an event and item were created on the cow card in DC305 using a custom-made protocol indicating the date and the respective quarter. Nonlactating quarters were monitored by means of the fresh cow monitoring protocol including an examination of the integrity of all 4 quarters on the day of calving and during daily premilking udder preparation through manual forestripping by milking technicians and subsequent examination by dedicated farm personnel. Identification and documentation of cows with a nonlactating quarter that were detected via one of the monitoring protocols were as described above.

      Milking System

      Cows were milked 3 times daily at 0100, 0900, and 1700 h in a 100-stall parallel rotary parlor (RP3100HD, DeLaval International AB). The vacuum pump (22.4 kW; 30 HP) was regulated by a variable frequency drive and set to supply a receiver operator vacuum of 44 kPa (13.0 inHg). The milking unit was composed of the cluster MC70 (DeLaval International AB) and a milking liner with a round barrel shape (LS-01 NC, DeLaval International AB). The pulsators (EP100, DeLaval International AB) were set to a pulsation rate of 60 cycles/min, a ratio of 65:35, and a side-to-side alternating pulsation. Measurements of pulsation phases under load as assessed with a digital vacuum recorder (VaDia, Biocontrol) were: 143; b-phase, 497; c-phase, 113; and d-phase, 247 ms. The average claw vacuum during the peak milk flow period was calculated from 20 milking observations using a digital vacuum recorder (VPR100, DeLaval International AB) according to the guidelines outlined by the National Mastitis Council (
      • NMC
      Procedures for Evaluating Vacuum Levels and Air Flow in Milking Systems.
      ) and was 37.6 kPa (11.1 inHg). The automatic cluster removers were set to a cluster remover milk flow threshold of 1.6 kg/min, a 0-s delay, and a vacuum decay time of 1.5 s. The milk sweep was initiated 3 s after unit retraction and lasted for 2 s. The milk line was installed 115 cm below the cow standing level. The length of the hosing system from the cluster to the milk line was 325 cm and partitioned into (1) the milk hose between cluster and stall inlet, 90 cm; (2) the hosing from stall inlet to milk meter, 145 cm; and (3) the milk hose between the milk meter and the milk line, 90 cm. The milking unit alignment device consists of a movable support arm, a chain that is adjustable in length, and a metal shell suspending the milk hose and the paired pulsation hose to support the milking cluster. Milking system settings and milking characteristics were monitored with a dairy farm management software program (DelPro, DeLaval International AB).

      Milking Routine

      The rotational speed of the milking parlor was 5.3 s/stall (530 s/complete rotation) resulting in a theoretical throughput of 679 cows/h. Two teat spray robots (TSR, DeLaval International AB) were installed at the parlor exit for postmilking teat dip application. The parlor was operated by 5 milking technicians who were assigned to 5 different positions including the following tasks: position 1 was to manually forestrip 2 teats and apply teat dip solution to all 4 teats; position 2 was to dry and clean the teat barrel of all teats from lactating quarters with an individual clean cloth towel; position 3 was to dry and clean the teat end with an individual clean cloth towel; position 4 was to attach and align the milking unit; position 5 was to monitor milking liner slips, unit falloffs, unit kickoffs, and realign or reattach the milking unit. The protocol for cows with a nonlactating quarter was to not attach the respective teat cup and to insert a liner plug into the teat cup to avoid air leakage. This task was executed by the milking technician at position 4. The positioning of milking technicians were (assuming that cow entrance = stall 1): position 1, stall 3; position 2, stall 10; position 3, stall 11; and position 4, stall 19 for pens with early- and mid-lactation cows and stall 24 for groups with late-lactation animals. The milking technician at position 5 operated within a wider range around the halfway point of the rotary to ensure timely adjustment of the milking units. This set up resulted in a dip contact time of 37 s, duration of first tactile stimulation (i.e., duration of forestripping) of approximately 3 s, and a preparation lag time (i.e., time from first tactile stimulus to milking unit attachment) of approximately 85 s for early- and mid-lactation animals and 111 s for late-lactation cows.

      Data Acquisition

      Cow Characteristics.

      Cow items (i.e., parity, DIM, SCC, and presence or absence of a nonlactating quarter) were obtained from the dairy management software program (DairyComp 305, Valley Agricultural Software). We used the results from DHIA testing in January 2020 for most cows to obtain the SCC data; for cows that had no test day information this month, we used the February test day data if available.

      Milk Flow Characteristics.

      Milk flow characteristics [milk yield (i.e., yield of milk harvested from start of milking to detachment of the milking unit, kg), milking unit on time (i.e., time recorded from start of milking to detachment of the milking unit, s), average milk flow rate (calculated as milk yield/milking unit on time, kg/min), first 15 s milk flow rate (i.e., average milk flow rate recorded in the first 15 s of milking, kg/min, 15 s), 15–30 s milk flow rate (i.e., average milk flow rate recorded between the first 15 to 30 s of milking, kg/min, 30 s), 30–60 s milk flow rate (i.e., average milk flow rate recorded between the first 30 to 60 s of milking, kg/min, 60 s), 60–120 s milk flow rate (i.e., average milk flow rate recorded between the first 60 to 120 s of milking, kg/min, 120 s), peak milk flow rate (i.e., calculated as the maximum 60-s average milk flow, kg/min), 2-min milk yield (i.e., amount of milk harvested within the first 2 min of milking, kg)], and occurrence or nonoccurrence of a milking machine liner slip were assessed at each milking with electronic on-farm milk meters using near-infrared technology (MM27BC, DeLaval International AB). The operating principle of the milk flow meter for the detection of a milking liner slip is based on the amount of air in the milk. The sensitivity of the detection can be adjusted by means of a nondimensional threshold value, referred to as the slip limit, that ranges from 0 to 255. The value 0 provides the highest sensitivity, and the default threshold value is 140. A liner slip is recorded if the measured air bleed value remains above the threshold value for a minimum of 3 s. The recorded value is a binary value, whereas the duration of the milking liner slip or the number of liner slips that occur during a single milking observation are not recorded. During this study, the slip limit was set to 175. A milking liner slip was therefore defined present if an air leakage with an intensity above the threshold value of 175 was registered for a minimum of 3 s. A comparison study to assess the agreement beyond that of chance of the on-farm milk meter for the detection of machine milking liner slips compared with the detection by means of audible signs indicated moderate agreement and has been reported elsewhere (unpublished data). We created a report in DelPro to automatically record milk flow characteristics for each milking session and export them to a comma-separated values file once daily. For subsequent analyses, a new binary variable to capture bimodality was created and defined as previously described by
      • Wieland M.
      • Virkler P.D.
      • Weld A.
      • Melvin J.M.
      • Wettstein M.R.
      • Oswald M.F.
      • Geary C.M.
      • Watters R.D.
      • Lynch R.
      • Nydam D.V.
      The effect of 2 different premilking stimulation regimens, with and without manual forestripping, on teat tissue condition and milking performance in Holstein dairy cows milked 3 times daily.
      . A bimodal milk flow curve occurred if any of the incremental milk flow rates 15–30 s milk flow rate, 30–60 s milk flow rate, or 60–120 s milk flow rate were lower than any of the previous ones (first 15 s milk flow rate, 15–30 s milk flow rate, 30–60 s milk flow rate), whereas bimodality was absent otherwise.

      Study Designs

      The study was carried out over 60 d from January 1 to February 29, 2020. The conceptual design of the study was dynamic. Thus, the study population consisted of all lactating cows that were enrolled into the study as they became available. Data from animals that were dried or culled during the study period were included in the analyses until the day of removal from the lactating herd.

      Analytical Approach

      We maintained data in Microsoft Excel (2019 version, Microsoft Corp.) and JMP (version 14, SAS Institute Inc.). Before statistical analyses, we investigated the data for missing (i.e., no data recorded during milking session due to detection failure in milking parlor) and erroneous (i.e., value of 0 for milk yield, milking unit on time, average milk flow rate, and peak milk flow rate) values using JMP. All subsequent analyses were performed with R (
      • R Core Team
      R: A Language and Environment for Statistical Computing.
      ).

      Liner Slip and Milking Characteristics.

      To study our hypothesis that milking machine liner slips were associated with milking unit on time and average milk flow rate, we generated 2 separate general linear mixed models with the ‘nlme' package (

      Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Core Team. 2016. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1–128.

      ). To account for temporal clustering of observations between milking sessions and days within a cow, we included cow and day nested within cow as random effects. We used the first order autoregressive covariance structure to model the covariance of repeated observations within cow. The binary variable occurrence or nonoccurrence of a milking machine liner slip was forced into each model. In a first step, we considered the following covariates for inclusion into each model and screened them through univariable analysis: parity (first, second, and ≥ third lactation), stage of lactation (≤100, 101–200, >200 DIM), presence or absence of a nonlactating quarter, the log10-transformed SCC (logSCC), and milk yield (kg/milking session). We considered all covariates with a P-value <0.20 in this step in the initial models. We used Spearman correlation coefficients to assess collinearity among eligible variables and chose a threshold of |0.50| to be indicative of collinearity. To establish the final model, we performed manual backward elimination until each of the variables had a P-value <0.05. Confounding effects were monitored by observing regression coefficient changes. Variables that modified regression coefficients by >20% were considered confounding factors. We applied Tukey-Kramer's post hoc test to control for the familywise error rate when comparing a family of estimates. Finally, we inspected residual plots versus corresponding predicted values and examined quantile-quantile residual plots to assess if the assumptions of homoscedasticity and normality of residuals were met.

      Cow Characteristics and Liner Slip.

      To test our hypothesis that cow characteristics were associated with the occurrence of a milking machine liner slip, we built a generalized liner mixed model with a logit link and a binomial distribution with the ‘lme4' package (
      • Bates D.
      • Mächler M.
      • Bolker B.
      • Walker S.
      Fitting linear mixed-effects models using lme4.
      ). To account for the correlated structure of the data, we included cow and day nested within cow as random effects. In an initial step, we screened the independent variables parity (first, second, and ≥ third lactation), stage of lactation (≤100, 101–200, >200 DIM), presence or absence of a nonlactating quarter, logSCC, occurrence or nonoccurrence of bimodality, milk yield (kg/milking session), and 2-min milk yield (kg/milking session) through univariable analysis. We considered all variables with a P-value <0.20 in this step as covariates in the initial model. To assess collinearity among eligible covariates, Pearson (for continuous variables) and Spearman (for categorical variables) correlation coefficients were used. We considered that a coefficient of >|0.50| indicated collinearity. Manual backward elimination was performed until each of the variables had a P-value <0.05. We assessed the final model's predictive ability by calculating a receiver operating characteristic curve with the pROC package (
      • Robin X.
      • Turck N.
      • Hainard A.
      • Tiberti N.
      • Lisacek F.
      • Sanchez J.-C.
      • Müller M.
      pROC: An open-source package for R and S+ to analyze and compare ROC curves.
      ). Finally, we calculated the adjusted probabilities (95% CI) for a hypothetical milking observation of a second lactation cow with a milk yield of 13.5 kg/milking session, a logSCC of 4.84, with and without a nonlactating quarter, respectively, using the AICcmodavg package (

      Mazerolle, M. J. 2016. AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). R package version 2.1–0.

      ).

      RESULTS

      Description of Study Population

      We obtained 766,494 milking observations over 60 d from 4,948 cows including 39,559/766,494 (5.2%) observations with a milking machine liner slip. A total of 80,164 (10.5%) observations were excluded due to missing or erroneous values (multiple numeration possible) of milk yield (missing, n = 44,931; erroneous, n = 536), milking unit on time (missing, n = 583; erroneous, n = 44,348), average milk flow rate (missing, n = 44,938; erroneous, n = 27,429), and peak milk flow rate (missing, 44,984; erroneous, n = 25,990) data resulting in 686,330 milking observations that were available for analyses. Cows were in their first (1,876; 37.9%), second (1,332; 26.9%), and third or greater lactation (1,740; 35.2%) and between 1 and 792 DIM (mean ± SD, 134 ± 116). The mean (±SD) logSCC was 4.84 ± 0.59. A total of 813 (16.4%) cows had a nonlactating quarter. The mean values (±SD; range) of milking characteristics were 13.5 ± 4.0 (0.2–47.2) kg/milking session for milk yield, 7.4 ± 2.3 (0.2–17.5) kg for 2-min milk, 3.5 ± 0.9 (0.1–7.6) kg/min for average milk flow rate, and 232 ± 57 (36–685) s for milking unit on time. A total of 220,581 (32.1%) milking observations were bimodal. Occurrence of a milking machine liner slip was documented in 37,931 (5.5%) observations. Table 1 shows milking characteristics stratified by occurrence and nonoccurrence of a milking liner slip.
      Table 1Milking characteristics from 686,330 milking observations from 4,948 cows obtained over a 60-d period stratified by occurrence (n = 37,931) and nonoccurrence (n = 648,399) of a milking liner slip
      Results presented as means and standard deviation unless otherwise stated.
      ItemLiner slip presentLiner slip absent
      Milk yield (kg/milking session)14.7 ± 3.913.4 ± 4.0
      Milking unit on time (s)235 ± 55231 ± 57
      Average milk flow rate (kg/min)3.7 ± 0.83.4 ± 0.9
      2-min milk yield (kg)8.3 ± 2.27.4 ± 2.3
      Bimodality
      Results presented as means and standard deviation unless otherwise stated.
      (n, %)
      12,395 (32.7)208,186 (32.1)
      1 Results presented as means and standard deviation unless otherwise stated.

      Liner Slip and Milking Characteristics

      Milking Unit on Time.

      All covariates tested through univariable analyses were associated with milking unit on time (P < 0.0001). Spearman correlation coefficients indicated no collinearity (r ≤ |0.37|) and, thus, all covariates were entered into the initial model. The final multivariable model included occurrence or nonoccurrence of a milking machine liner slip (P = 0.17), parity (P < 0.0001), stage of lactation (P < 0.0001), presence or absence of a nonlactating quarter (P < 0.0001), logSCC (P = 0.045), and milk yield (P < 0.0001, Table 2). Least squares means (95% CI) were 237 (235–238) s for both, milking observations with and without occurrence of a milking machine liner slip.
      Table 2Results of multivariable general linear mixed model showing the association of milking machine liner slip, parity, stage of lactation, presence or absence of a nonlactating quarter, SCC, and milk yield with milking unit on time (s)
      Item
      Intercept omitted for clarity.
      β
      Linear regression coefficient.
      (SE)
      P-valueLSM (95% CI)
      Liner slip0.17
       Occurrence0 (0.2)237 (235–238)
       NonoccurrenceReferent237 (235–238)
      Parity<0.0001
       First8 (1.3)
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      243 (240–245)
       Second−2 (1.5)
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      233 (230–235)
       ThirdReferent
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      235 (233–237)
      Stage of lactation (DIM)<0.0001
       ≤10013 (1.4)
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      245 (243–247)
       101–2002 (1.5)
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      233 (231–236)
       >200Referent
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      232 (230–234)
      Nonlactating quarter<0.0001
       Present11 (1.5)242 (240–245)
       AbsentReferent231 (230–232)
      LogSCC
      LogSCC = SCC, log10 transformed.
      2 (1.0)0.046
      Milk yield
      Milk yield: 1-unit increase = 1 kg.
      (kg/milking session)
      10 (0.0)<0.0001
      a,b Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      1 Intercept omitted for clarity.
      2 Linear regression coefficient.
      3 LogSCC = SCC, log10 transformed.
      4 Milk yield: 1-unit increase = 1 kg.

      Average Milk Flow Rate.

      Univariable analyses revealed that all tested covariates were associated with average milk flow rate (P < 0.0001) and offered into the initial model. In the initial model, logSCC yielded a P-value = 0.12; it was not considered a confounder based on the definition outlined above and, thus, was eliminated from the model. The final multivariable model included occurrence or nonoccurrence of a milking machine liner slip (P < 0.0001), parity (P = 0.01), stage of lactation (P < 0.0001), presence or absence of a nonlactating quarter (P < 0.0001), and milk yield (P < 0.0001, Table 3). Least squares means (95% CI) for a milking observation with and without occurrence of a milking machine liner slip, respectively, were 3.40 (3.37–3.42) and 3.42 (3.40–3.44) kg/min.
      Table 3Results of multivariable general linear mixed model showing the association of milking machine liner slip, parity, stage of lactation, presence or absence of a nonlactating quarter, and milk yield with average milk flow rate (kg/min)
      Item
      Intercept omitted for clarity.
      β
      Linear regression coefficient.
      (SE)
      P-valueLSM (95% CI)
      Liner slip<0.0001
       Occurrence−0.0 (0.0)3.40 (3.37–3.42)
       NonoccurrenceReferent3.42 (3.40–3.44)
      Parity0.01
       First−0.0 (0.0)
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      3.38 (3.35–3.42)
       Second0.1 (0.0)
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      3.45 (3.41–3.48)
       ThirdReferent
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      3.40 (3.36–3.43)
      Stage of lactation (DIM)<0.0001
       ≤100−0.2 (0.0)
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      3.29 (3.26–3.32)
       101–2000.0 (0.0)
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      3.49 (3.45–3.52)
       >200Referent
      Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      3.45 (3.41–3.48)
      Nonlactating quarter<0.0001
       Present−0.2 (0.0)3.32 (3.28–3.36)
       AbsentReferent3.50 (3.48–3.52)
      Milk yield
      Milk yield: 1-unit increase = 1 kg.
      (kg/milking session)
      0.1 (0.0)<0.0001
      a,b Levels marked with different letters differ at a level of P < 0.05 in Tukey-Kramer's post hoc test.
      1 Intercept omitted for clarity.
      2 Linear regression coefficient.
      3 Milk yield: 1-unit increase = 1 kg.
      For both models, the assumptions of homoscedasticity and normality of residuals were met.

      Cow Characteristics and Liner Slip

      All covariates that were tested in univariable analyses were associated with occurrence or nonoccurrence of a machine milking liner slip (P-values ≤ 0.05). Collinearity was detected between milk yield and 2-min milk yield (r = 0.55, P < 0.0001). We considered 2-min milk yield as the biologically more plausible variable and, therefore, elected to include it in the initial multivariable model. Parity and logSCC were eliminated from the initial multivariable model as P-values ≥ 0.08, and no confounding effects were observed. The final multivariable model included stage of lactation (P < 0.0001), presence or absence of a nonlactating quarter (P < 0.0001), bimodality (P = 0.03), and 2-min milk yield (P < 0.0001, Figure 1). The adjusted probabilities (95% CI) for a hypothetical milking observation of a mid-lactation (101–200 DIM) cow with a 2-min milk yield of 7.4 kg/milking session, a bimodal milk flow, with and without a nonlactating quarter, respectively, were 1.56 (1.19–2.03) and 0.15 (0.12–0.19) %. The area under the receiver operator characteristic curve was 0.96.
      Figure thumbnail gr1
      Figure 1Results of a multivariable generalized linear mixed model showing the association of stage of lactation, presence of a nonlactating quarter, presence of bimodality, and 2-min milk yield with occurrence or nonoccurrence of a milking machine liner slip. Adjusted odds ratio and 95% CI (error bars) are presented. Bimodality: bimodal milk flow curve was present if any of the incremental milk flow rates (15–30 s milk flow rate, 30–60 s milk flow rate, and 60–120 s milk flow rate) were lower than any of the previous ones (first 15 s milk flow rate, 15–30 s milk flow rate, and 30–60 s milk flow rate). Two-minute milk yield (kg/milking session): 1-unit increase = 1 kg.

      DISCUSSION

      Liner Slip and Milking Characteristics

      Our primary objective of this study was to investigate the association between milking machine liner slips and milking characteristics. We conducted the study on a single high-producing commercial dairy farm in New York State with a thrice-daily milking schedule to reflect milking practices in modern dairy herds in the Northeast of the United States. Our results show that there was no association between milking machine liner slip and milking unit on time. They further indicate that the observed differences in average milk flow rate between milking observations with and without occurrence of a liner slip, respectively, were unlikely due to chance but not biologically relevant.
      We believe that the absence of a meaningful effect of liner slips on milking unit on time and average milk flow rate can be attributed to the milking routine of the study herd. One milking technician was devoted to monitor liner slips, unit falloffs, and unit kickoffs, as well as realigning or reattaching the milking unit. In most cases, this likely led to an immediate intervention that decreased the duration of a liner slip, prevented possible sequelae such as a unit falloff or kickoff, and reduced its possible effect on milking efficiency. To a smaller extent, the lack of a meaningful association could have been due to a consistent use of the milking unit alignment device. This might have diminished the negative effect of a liner slip on milking characteristics by reducing its duration and mitigating possible sequelae. However, because we had no means to measure the duration of a liner slip or compare milking routines from different dairy farms and parlor systems, these possible explanations remain speculative. It is also possible that the absence of associations of milking machine liner slips with milking unit on time and average milk flow rate has its root in the detection method including the chosen threshold value.
      The frequency distribution of 5.5% of liner slips documented in this study compares to 12.4% in a previous study from our group using the identical detection method and the same threshold value (
      • Wieland M.
      • Virkler P.D.
      • Weld A.
      • Melvin J.M.
      • Wettstein M.R.
      • Oswald M.F.
      • Geary C.M.
      • Watters R.D.
      • Lynch R.
      • Nydam D.V.
      The effect of 2 different premilking stimulation regimens, with and without manual forestripping, on teat tissue condition and milking performance in Holstein dairy cows milked 3 times daily.
      ). Other researchers used different techniques including measuring vacuum fluctuations with vacuum transducers (
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      ;
      • Baxter J.D.
      • Rogers G.W.
      • Spencer S.B.
      • Eberhart R.J.
      The effect of milking machine liner slip on new intramammary infections.
      ) or by means of audible assessment (
      • Rasmussen M.D.
      • Madsen N.P.
      Effects of milkline vacuum, pulsator airline vacuum, and cluster weight on milk yield, teat condition, and udder health.
      ).
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      used a slip detection and counting device described by
      • Spencer S.B.
      • Volz C.
      Measuring Milking Machine Liner Slips.
      and defined a liner slip as the abrupt loss of milking vacuum >8 kPa from the average milking vacuum in <0.25 s. The mean number of liner slips per milking observation was related to the operating vacuum and ranged from 3.9 to 8.8 (
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      ).
      • Baxter J.D.
      • Rogers G.W.
      • Spencer S.B.
      • Eberhart R.J.
      The effect of milking machine liner slip on new intramammary infections.
      employed the system used by
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      to study the relationship between milking machine liner slips and new IMI. They defined a liner slip as a drop in claw vacuum of >10 kPa in <0.25 s and reported an average number of liner slips per milking session of 3.6 and 6.1 for 2 different milking liners (
      • Baxter J.D.
      • Rogers G.W.
      • Spencer S.B.
      • Eberhart R.J.
      The effect of milking machine liner slip on new intramammary infections.
      ).
      • Rasmussen M.D.
      • Madsen N.P.
      Effects of milkline vacuum, pulsator airline vacuum, and cluster weight on milk yield, teat condition, and udder health.
      assessed the frequency of audible liner slips to assess differences among different milkline vacuum settings, cluster weights, and pulsator airline vacuums. The frequency distribution of liner slips ranged between 2.4% and 30.0%, again depending on vacuum setting or milking equipment (i.e., cluster weight).

      Cow Characteristics and Liner Slip

      The second objective was to study the association between cow characteristics and liner slips. Our data show that cow characteristics are associated with liner slips and explain some of the variability of their occurrence between cows, which is in accordance with findings by previous authors (
      • Rogers G.W.
      • Spencer S.B.
      Relationships among udder and teat morphology and milking characteristics.
      ;
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      ). Liner slips have been associated with increased mastitis risk (
      • O'Callaghan E.
      • O'Shea J.
      • Meaney W.J.
      • Crowley C.
      Effect of milking machine vacuum fluctuations and liner slip on bovine mastitis infectivity.
      ;
      • Baxter J.D.
      • Rogers G.W.
      • Spencer S.B.
      • Eberhart R.J.
      The effect of milking machine liner slip on new intramammary infections.
      ). Our findings, therefore, could help identify cows at increased risk of liner slips which would allow dairy producers to implement specific management strategies that mitigate the frequency of liner slips and possibly improve udder health. The odds of a liner slip decreased with increasing stage of lactation. A possible explanation could be that late-lactation cows are more familiar with the milking routine compared with early lactation animals. This may have led to less adverse movements in late-lactation cows that could have hampered the ability to attach and properly align the milking unit and, thus, decreased the risk of liner slips. It is also possible that udder edema, which has been reported to affect >50% of cows shortly after calving (
      • Morrison E.I.
      • DeVries T.J.
      • LeBlanc S.J.
      Short communication: Associations of udder edema with health, milk yield, and reproduction in dairy cows in early lactation.
      ), may have had an influence on the observed differences in liner slips among cows at different stages of lactation. As outlined previously, the swelling and firmness associated with an edematous teat can make milking unit attachment difficult (
      • Melendez P.
      • Hofer C.C.
      • Donovan G.A.
      Risk factors for udder edema and its association with lactation performance on primiparous Holstein cows in a large Florida herd.
      ). This could be aggravated by increased defensive movements from early lactation cows with udder edema and ultimately leading to more liner slips. Unlike
      • Rogers G.W.
      • Spencer S.B.
      Relationships among udder and teat morphology and milking characteristics.
      , our multivariable model indicated no association between parity and the occurrence of liner slips. Discrepancies in study population, milking machine equipment and settings, as well as detection technique could be variables that help explain the differences between the previous study and ours.
      Furthermore, we found that the presence of a nonlactating quarter increased the odds of a liner slip by >900%. We hypothesize that this was related to the following 3 facts: First, due to inherent limitations, milking unit alignment was likely worse in cows with a nonlactating quarter. This could have led to insufficient support of the milking cluster and increased the risk of a liner slip. Insufficient unit alignment also contributes to unequal quarter emptying. This may have increased the risk of overmilking of individual quarters, which is characterized by an increased mouthpiece chamber vacuum (
      • Moore-Foster R.
      • Norby B.
      • Schewe R.L.
      • Thomson R.
      • Bartlett P.C.
      • Erskine R.J.
      Short communication: Herd-level variables associated with overmilking in Michigan dairy herds.
      ). If the mouthpiece chamber vacuum exceeds the inertia of the mouthpiece lip, an increased mouthpiece chamber vacuum can result in a liner slip. Second, we speculate that differences in quarter-level milk production are larger in cows with a nonlactating quarter; specifically, the ipsilateral quarter likely produces more milk compared with the ones of the intact udder half. This could have led to unequal emptying of quarters, even in the presence of optimal unit alignment, resulted in overmilking of individual quarters, and yielded to increased risk of liner slips. Third, although the milking protocol for cows with a nonlactating quarter included insertion of a liner plug into the respective teat cup before attachment of the milking unit, compliance to this standard might have been variable. Instead, it is common practice on this dairy farm that milking technicians kink the milking liner at the aspect of the short milk tube after unit attachment. This practice may have increased the risk of liner slips in cows with a nonlactating quarter.
      Our results indicate that bimodality increased the odds of a liner slip by 5%. A possible explanation could be the interrelationship between bimodality, mouthpiece chamber vacuum, and liner slips. As outlined by previous researchers (
      • Borkhus M.
      • Rønningen O.
      Factors affecting mouthpiece chamber vacuum in machine milking.
      ;
      • Erskine R.J.
      • Norby B.
      • Neuder L.M.
      • Thomson R.S.
      Decreased milk yield is associated with delayed milk ejection.
      ), when milk flow is low or nonexistent, as for example during a bimodal milk flow curve, the teat barrel diameter decreases as a consequence of lower positive intracisternal pressure. This results in a poor seal between the teat and the liner barrel wall, which in turn increases leakage of milking vacuum from the teat end into the mouthpiece chamber. If the friction between the mouthpiece lip and the teat fails to withstand the forces created by the vacuum in the mouthpiece chamber, a subsequent sudden air leakage through the mouthpiece opening may result in a liner slip (
      • Mein G.A.
      • Thiel C.C.
      • Westgarth D.R.
      • Fulford R.J.
      Friction between the teat and teatcup liner during milking.
      ;
      • Borkhus M.
      • Rønningen O.
      Factors affecting mouthpiece chamber vacuum in machine milking.
      ). It is also possible that the observed association is due to a correlation between the case definitions for bimodality and the detection of a liner slips of the milk flow meter system.
      Last, a 1-kg increase in 2-min milk increased the odds of liner slip by 26%. We attribute this phenomenon to the milk flow-dependent vacuum drop affecting the force that retains the teat cup assembly in place when attached to a teat. According to
      • Mein G.A.
      • Thiel C.C.
      • Westgarth D.R.
      • Fulford R.J.
      Friction between the teat and teatcup liner during milking.
      , opposing forces act both to remove the liner from the teat and to cause the teat to move further into the liner during machine milking. The weight of the milking cluster creates the force that tends to separate the milking liner from the teat. The opposing force that tends to pull the teat further into the milking liner is a function of the vacuum level in the milking liner and the cross-sectional area of the teat exposed to the vacuum. The frictional force maintains the teat cup stable on the teat during machine milking. The sources of friction are derived from the force between the teat barrel and the wall of the milking liner barrel, as well as between the teat base and the mouthpiece lip (
      • Mein G.A.
      • Thiel C.C.
      • Westgarth D.R.
      • Fulford R.J.
      Friction between the teat and teatcup liner during milking.
      ). Due to an inverse relationship between milk flow and teat end vacuum (
      • Bade R.D.
      • Reinemann D.J.
      • Zucali M.
      • Ruegg P.L.
      • Thompson P.D.
      Interactions of vacuum, b-phase duration, and liner compression on milk flow rates in dairy cows.
      ;
      • Ambord S.
      • Bruckmaier R.M.
      Milk flow-dependent vacuum loss in high-line milking systems: Effects on milking characteristics and teat tissue condition.
      ), the force that retains the milking liner at the teat decreases during high milk flow rates. If the frictional forces between the teat and the milking liner then fail to keep the milking liner in place, the seal between the teat and milking liner barrel may break and a liner slip can occur. This possible explanation is supported by previous research indicating that the frequency of liner slips increased with lower milking vacuum (
      • Spencer S.B.
      • Rogers G.W.
      Effect of vacuum and milking machine liners on liner slip.
      ;
      • Rasmussen M.D.
      • Madsen N.P.
      Effects of milkline vacuum, pulsator airline vacuum, and cluster weight on milk yield, teat condition, and udder health.
      ). Differences in teat characteristics that lead to higher 2-min milk yield (
      • Wieland M.
      • Nydam D.V.
      • Virkler P.D.
      A longitudinal field study investigating the association between teat-end shape and two minute milk yield, milking unit-on time, and time in low flow rate.
      ) and possibly poorer liner fit could also play a role in the proposed mechanism as described by previous researchers (
      • Rogers G.W.
      • Spencer S.B.
      Relationships among udder and teat morphology and milking characteristics.
      ).
      Based on the results of our data analyses, possible measures to mitigate the frequency of milking machine liner slips in the current study population are different management strategies for (1) cows with a nonlactating quarter, (2) cows at risk of bimodality, and (3) animals with high 2-min milk yield. Specific changes in the milking routine for cows with a nonlactating quarter could be insertion of a milking liner plug before attachment of the milking unit on a consistent basis. Possible technological solutions include modifications of the milking unit alignment device and a flow-controlled milking cluster with quarter-level vacuum regulation.
      Bimodality occurs when pre-milking udder stimulation fails to accommodate the physiological requirements of the cow (
      • Kaskous S.
      • Bruckmaier R.M.
      Best combination of pre-stimulation and latency period duration before cluster attachment for efficient oxytocin release and milk ejection in cows with low to high udder-filling levels.
      ). It can be a consequence of insufficient tactile stimulation, inadequate preparation lag time (i.e., time from first tactile stimulation to milking unit attachment) or a combination of both (
      • Bruckmaier R.M.
      • Blum J.W.
      Oxytocin release and milk removal in ruminants.
      ). A preparation lag time of 85 s for early- and mid-lactation animals and 111 s for late-lactation animals is consistent with current industry recommendations and recent research (
      • Kaskous S.
      • Bruckmaier R.M.
      Best combination of pre-stimulation and latency period duration before cluster attachment for efficient oxytocin release and milk ejection in cows with low to high udder-filling levels.
      ;
      • Watters R.D.
      • Schuring N.
      • Erb H.N.
      • Schukken Y.H.
      • Galton D.M.
      The effect of premilking udder preparation on Holstein cows milked 3 times daily.
      ). By contrast, a stimulation time of approximately 3 s likely fails to elicit the cow's maximum milk ejection capacity. It does not meet current recommendations (
      • Reneau J.K.
      • Chastain J.P.
      Premilking cow prep: Adapting to your system.
      ;
      • Durst P.T.
      • Chojnacki R.
      How is variability in your farm's milking procedure affecting milk flow?.
      ;
      • Wieland M.
      • Nydam D.V.
      You have to go slow to go fast when milking cows.
      ) and contrasts recent research on pre-milking teat stimulation using a minimum stimulation time of 15 s (
      • Kaskous S.
      • Bruckmaier R.M.
      Best combination of pre-stimulation and latency period duration before cluster attachment for efficient oxytocin release and milk ejection in cows with low to high udder-filling levels.
      ;
      • Edwards J.P.
      • Jago J.G.
      • Lopez-Villalobos N.
      Short-term application of prestimulation and increased automatic cluster remover threshold affect milking characteristics of grazing dairy cows in late lactation.
      ;
      • Vetter A.
      • van Dorland H.A.
      • Youssef M.
      • Bruckmaier R.M.
      Effects of a latency period between pre-stimulation and teat cup attachment and periodic vacuum reduction on milking characteristics and teat condition in dairy cows.
      ). In a recent study, we showed that the odds of bimodality were lower in cows who received a pre-milking stimulation regimen that included forestripping and resulted in a stimulation time of 16 s compared with a regimen that consisted of 7 s of wiping of teats (
      • Wieland M.
      • Virkler P.D.
      • Weld A.
      • Melvin J.M.
      • Wettstein M.R.
      • Oswald M.F.
      • Geary C.M.
      • Watters R.D.
      • Lynch R.
      • Nydam D.V.
      The effect of 2 different premilking stimulation regimens, with and without manual forestripping, on teat tissue condition and milking performance in Holstein dairy cows milked 3 times daily.
      ). We therefore believe that increasing the stimulation time has the greatest potential to reduce bimodality and possibly decrease the odds of liner slips in the study herd presented herein.
      To mitigate a possible effect of high milk flow (i.e., 2-min milk yield) on the frequency of liner slips, a flow-dependent vacuum regulation as reported by
      • Feierabend M.
      • Stauffer C.
      • Bruckmaier R.M.
      Milking performance and teat condition at high vacuum milking with or without vacuum reduction during low milk flow and at different detachment levels.
      could be a viable option. Such a system could facilitate milk flow-based adjustment of the milking vacuum and potentially decrease the frequency of liner slips without adversely affecting teat tissue condition. Because a causal relationship between the described risk factors and liner slips cannot be made from this study, the proposed measures to mitigate the frequency of liner slips on our study herd remain speculative.

      Limitations and Future Research

      This study was conducted on a single commercial dairy farm in New York State. Our results are therefore likely to represent the situation of commercial operations in this region. However, the external validity of our study may be limited to dairy farms with a similar population, parlor system, and milking routine. Exclusion of 80,164 (10.5%) milking observations including 1,628 (2.0%) observations with a liner slip may have resulted in a selection bias. However, we believe that if such a selection bias had occurred, it would have been minor. Milking machine liner slips were assessed with an electronic on-farm milk meter as a binary outcome; no information of slip duration or time of occurrence (i.e., start versus end of milking) were recorded. Knowledge about for example the time of liner slip occurrence would allow us to further discriminate potential risk factors and possible effects on milking performance and udder health. Milking machine liner slips have been associated with increased risk of new IMI; however, contradictory conclusions exist regarding its effect size.
      • Baxter J.D.
      • Rogers G.W.
      • Spencer S.B.
      • Eberhart R.J.
      The effect of milking machine liner slip on new intramammary infections.
      suggested that more than 2,000 liner slips are needed to cause 1 new mastitis case. Conversely, researchers from Ireland (
      • O'Callaghan E.
      • O'Shea J.
      • Meaney W.J.
      • Crowley C.
      Effect of milking machine vacuum fluctuations and liner slip on bovine mastitis infectivity.
      ) conducted a series of experiments after which they suggested that milking machine liner slip “is the single most important factor affecting mastitis infectivity” (p. 417). Both works have in common that they date back several decades and no longer represent modern dairy operations. Future research should therefore focus on the influence of milking machine liner slips on udder health.

      CONCLUSIONS

      We detected no association between milking machine liner slips and milking unit on time in the current study cohort. The statistically significant association between liner slip and average milk flow rate was biologically irrelevant. We attribute the lack of meaningful differences between milking observations with and without a liner slip to the milking routine including alignment of the milking unit and immediate intervention by a dedicated milking technician at the half waypoint of the rotary parlor. This likely decreased the duration of a liner slip, prevented possible sequelae, and reduced its possible effect on milking efficiency. We conclude that the negative effect of liner slips on milking performance can be diminished if certain conditions are met. Stage of lactation, presence of a nonlactating quarter, bimodal milk letdown, and 2-min milk yield were associated with the occurrence of liner slip. These cow characteristics could offer unique opportunities to identify and manage cows at increased risk of liner slips. Future research is needed to determine the economic effect of liner slips to justify implementation of cost-intensive measures.

      ACKNOWLEDGMENTS

      This study received no external funding. The authors thank the farm owners for providing their data. The authors have not stated any conflicts of interest.

      REFERENCES

        • Ambord S.
        • Bruckmaier R.M.
        Milk flow-dependent vacuum loss in high-line milking systems: Effects on milking characteristics and teat tissue condition.
        J. Dairy Sci. 2010; 93 (20655427): 3588-3594
        • Bade R.D.
        • Reinemann D.J.
        • Zucali M.
        • Ruegg P.L.
        • Thompson P.D.
        Interactions of vacuum, b-phase duration, and liner compression on milk flow rates in dairy cows.
        J. Dairy Sci. 2009; 92 (19233784): 913-921
        • Bates D.
        • Mächler M.
        • Bolker B.
        • Walker S.
        Fitting linear mixed-effects models using lme4.
        J. Stat. Softw. 2015; 67: 1-48
        • Baxter J.D.
        • Rogers G.W.
        • Spencer S.B.
        • Eberhart R.J.
        The effect of milking machine liner slip on new intramammary infections.
        J. Dairy Sci. 1992; 75 (1578015): 1015-1018
        • Borkhus M.
        • Rønningen O.
        Factors affecting mouthpiece chamber vacuum in machine milking.
        J. Dairy Res. 2003; 70 (12916822): 283-288
        • Bruckmaier R.M.
        • Blum J.W.
        Oxytocin release and milk removal in ruminants.
        J. Dairy Sci. 1998; 81 (9594382): 939-949
        • Davis M.A.
        • Reinemann D.J.
        • Mein G.A.
        Effect of liner age on milking characteristics.
        in: Proc. Natl. Mastitis Counc., Ann. Mtg. Natl. Mastitis Counc. Inc, 2000: 186-187
        • Durst P.T.
        • Chojnacki R.
        How is variability in your farm's milking procedure affecting milk flow?.
        Michigan State University Extension, 2017
        • Edwards J.P.
        • Jago J.G.
        • Lopez-Villalobos N.
        Short-term application of prestimulation and increased automatic cluster remover threshold affect milking characteristics of grazing dairy cows in late lactation.
        J. Dairy Sci. 2013; 96 (23332836): 1886-1893
        • Erskine R.J.
        • Norby B.
        • Neuder L.M.
        • Thomson R.S.
        Decreased milk yield is associated with delayed milk ejection.
        J. Dairy Sci. 2019; 102 (31030923): 6477-6484
        • Feierabend M.
        • Stauffer C.
        • Bruckmaier R.M.
        Milking performance and teat condition at high vacuum milking with or without vacuum reduction during low milk flow and at different detachment levels.
        in: Proc. Natl. Mastitis Counc., Ann. Mtg. Natl. Mastitis Counc. Inc, 2020: 140-141
        • Kaskous S.
        • Bruckmaier R.M.
        Best combination of pre-stimulation and latency period duration before cluster attachment for efficient oxytocin release and milk ejection in cows with low to high udder-filling levels.
        J. Dairy Res. 2011; 78 (21118612): 97-104
      1. Mazerolle, M. J. 2016. AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). R package version 2.1–0.

        • Mein G.A.
        • Thiel C.C.
        • Westgarth D.R.
        • Fulford R.J.
        Friction between the teat and teatcup liner during milking.
        J. Dairy Res. 1973; 40: 191-206
        • Melendez P.
        • Hofer C.C.
        • Donovan G.A.
        Risk factors for udder edema and its association with lactation performance on primiparous Holstein cows in a large Florida herd.
        Prev. Vet. Med. 2006; 76 (16774793): 211-221
        • Moore-Foster R.
        • Norby B.
        • Schewe R.L.
        • Thomson R.
        • Bartlett P.C.
        • Erskine R.J.
        Short communication: Herd-level variables associated with overmilking in Michigan dairy herds.
        J. Dairy Sci. 2019; 102 (31279548): 8400-8404
        • Morrison E.I.
        • DeVries T.J.
        • LeBlanc S.J.
        Short communication: Associations of udder edema with health, milk yield, and reproduction in dairy cows in early lactation.
        J. Dairy Sci. 2018; 101 (30100499): 9521-9526
        • NMC
        Procedures for Evaluating Vacuum Levels and Air Flow in Milking Systems.
        National Mastitis Council, 2012
        • NRC
        Nutrient Requirements of Dairy Cattle.
        7th rev. ed. The National Academies Press, 2001: 408
        • O'Callaghan E.
        • O'Shea J.
        • Meaney W.J.
        • Crowley C.
        Effect of milking machine vacuum fluctuations and liner slip on bovine mastitis infectivity.
        Isr. J. Agric. Res. 1976; 15: 401-418
      2. Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Core Team. 2016. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1–128.

        • R Core Team
        R: A Language and Environment for Statistical Computing.
        R Foundation for Statistical Computing, 2017
        • Rasmussen M.D.
        • Madsen N.P.
        Effects of milkline vacuum, pulsator airline vacuum, and cluster weight on milk yield, teat condition, and udder health.
        J. Dairy Sci. 2000; 83 (10659967): 77-84
        • Reneau J.K.
        • Chastain J.P.
        Premilking cow prep: Adapting to your system.
        in: Proc. Nat. Mastitis Counc. Reg. Mtg. Proc. Natl. Mastitis Counc, 1995: 46-63
        • Robin X.
        • Turck N.
        • Hainard A.
        • Tiberti N.
        • Lisacek F.
        • Sanchez J.-C.
        • Müller M.
        pROC: An open-source package for R and S+ to analyze and compare ROC curves.
        BMC Bioinformatics. 2011; 12 (21414208): 77
        • Rogers G.W.
        • Spencer S.B.
        Relationships among udder and teat morphology and milking characteristics.
        J. Dairy Sci. 1991; 74 (1787190): 4189-4194
        • Spencer S.B.
        • Rogers G.W.
        Effect of vacuum and milking machine liners on liner slip.
        J. Dairy Sci. 1991; 74: 429-432
        • Spencer S.B.
        • Volz C.
        Measuring Milking Machine Liner Slips.
        J. Dairy Sci. 1990; 73: 1000-1004
        • Stewart S.
        • Eicker S.
        • Rapnicki P.
        Automated collection of parlor performance data: Information needed and proposed standardized definitions.
        in: Proc. 2nd International Symposium on Mastitis and Milk Quality. National Mastitis Council and American Association of Bovine Practitioners. Natl. Mastitis Counc, 2001: 377-381
        • Vetter A.
        • van Dorland H.A.
        • Youssef M.
        • Bruckmaier R.M.
        Effects of a latency period between pre-stimulation and teat cup attachment and periodic vacuum reduction on milking characteristics and teat condition in dairy cows.
        J. Dairy Res. 2014; 81 (24433587): 107-112
        • Watters R.D.
        • Schuring N.
        • Erb H.N.
        • Schukken Y.H.
        • Galton D.M.
        The effect of premilking udder preparation on Holstein cows milked 3 times daily.
        J. Dairy Sci. 2012; 95 (22365201): 1170-1176
        • Wieland M.
        • Nydam D.V.
        You have to go slow to go fast when milking cows.
        Hoard's Dairyman. 2017; : 747
        • Wieland M.
        • Nydam D.V.
        • Virkler P.D.
        A longitudinal field study investigating the association between teat-end shape and two minute milk yield, milking unit-on time, and time in low flow rate.
        Livest. Sci. 2017; 205: 88-97
        • Wieland M.
        • Virkler P.D.
        • Weld A.
        • Melvin J.M.
        • Wettstein M.R.
        • Oswald M.F.
        • Geary C.M.
        • Watters R.D.
        • Lynch R.
        • Nydam D.V.
        The effect of 2 different premilking stimulation regimens, with and without manual forestripping, on teat tissue condition and milking performance in Holstein dairy cows milked 3 times daily.
        J. Dairy Sci. 2020; 103 (32828498): 9548-9560