Summary and Recommendations, June 9, 2010

How completely should we aim to empty cows udders at milking time? G. Mein, with input from R. Dyson, M. Eden (NZ), D. Reinemann (USA), J. Burke and J. Jago (NZ), R. Greenall, J. Penry Summary and Recommendations, June 9, 2010 The traditional recommendation to dairy farmers to milk all cows as completely as possible at every milking can be revised in light of recent research and field experience. Revised recommendations should be along the following lines. Milk most cows as completely as possible, within a reasonable time, at every milking. This implies a maximum ACR threshold setting of 400 ml/min for herds milked once or twice daily. Milk all cows out as evenly as possible. Why? Because uneven milk-out contributes to uneven distribution of milk yield between quarters, leading to less uniform udder conformation which, in turn, reduces the ease and efficiency of machine milking. Don t wait around for slow cows to finish milking. Instead, remove clusters from slow-milking cows based on their expected Maximum Milk-Out Time (see MMOT Quicknote on CowTime website: www.cowtime.com.au). A simplified version of MMOT could be applied at the 10-15% level in herringbone dairies. In a 10-a-side, for example, advise farmers not to wait around for the last cow; in a 20-a-side, don t wait for the last 2 or maybe 3 cows, etc. In rotaries, select a platform rotation time and apply a strict policy that no cow goes around twice unless there is a specific reason. Qualitative and semi-quantitative guidelines for defining and assessing completeness of milkout of individual quarters are given in Table 1. The qualitative guidelines (adapted from A. Joe, 2010) are a practical alternative for use in herds where hand-stripping would cause unacceptable disruption to the operators milking routine - or unacceptable risk from kicking by cows that are unaccustomed to having their teats handled after milking. The semiquantitative assessment provides guidelines for selective hand-stripping, in combination with the qualitative assessment, to improve the reliability of diagnosis. Guidelines for assessment of udder strip yields by machine stripping are also included in the table. Although machine stripping may provide a more reliable and repeatable measurement method, its use is essentially limited to milking systems where permanent milk meters are installed. Table 1. Guidelines for assessing and recording the completeness of milk-out Record milk-out as: G (Good) P (Poor) U (Uneven) Qualitative assessment Quarter is visibly wrinkled Quarter appears slightly plump, possibly indicating unharvested milk One particular quarter appears plumper and less wrinkled, relative to the other quarters Semi-quantitative (hand-stripping of individual quarter) 4 or fewer strips (equating to < 100 ml per quarter) 5 or more easy strips (equating to more than about 100 ml per quarter) Machine-stripping (based on whole udder) Less than 500 ml per udder More than 500 ml per udder One particular quarter appears plumper and less wrinkled, relative to other the quarters

Practical tips for assessing completeness of milking (based on discussions with R. Dyson and other experienced udder health specialists, 2010). 1. To ensure confidence in your results, aim to examine or strip quarters of at least 25 cows, scoring and recording results as suggested in Table 1. If practicable, aim to assess up to 50 cows or 10% of the herd (whichever target is reached first). 2. Most herds can be stripped relatively simply. Use of good cowmanship helps don t just walk up and grab teats, touch the rear of the cow or the udder first. Even so, in some herds, hand-stripping can be a life threatening experience! Remember the motto: No strip yield or teat score is worth getting killed for 3. Visual assessment is a VERY GOOD guide for quarters that are visibly wrinkled. 4. However, visual assessment is a POOR guide for quarters that are visually assessed as slightly plump many such quarters may have unexpectedly low strip yields. 5. Accordingly, a useful compromise is to use both qualitative and semi-quantitative methods but only strip-testing the selected quarters which are judged NOT to have milked out completely. 6. Rather than trying to hand-strip into a measuring jug, or using a vigorous get it all approach to hand-stripping, simply squeeze the teat and squirt successive teat-fulls onto the floor. Allow the teat to refill in the palm of your hand between each squirt, paying attention to the rate and ease of refilling of the teat sinus. With practice, assessment of the ease and rapidity of teat refill offers a quick, repeatable measure of completeness of milk-out. The end-point of stripping is when the teat fails to refill almost immediately. 7. There s no need to continue stripping once it becomes clear that strip yield exceeds your estimate of 100 ml. Stop stripping, record the result as Poor, and move on to the next teat. 8. It often helps to have some idea what a volume of about 100 ml looks like when squirted onto the concrete floor under a cow. A useful training tip is to calibrate your individual stripping technique by: catching 2 or 3 squirts in a small measuring jug; calculating the approx volume per squirt, then squirting 100 ml onto the floor to provide a mental picture of the size of a 100 ml puddle of strippings milk. 9. Think of a high strip yield not so much as undermilking but rather as a failure of the milking machine to completely remove the available milk from the udder. This may not necessarily indicate a higher risk of mastitis. 10. The core message of point 9 is that estimates of quarter strip yields can provide valuable information to help answer the question 'why' in the event of a failure to completely remove the available milk from an udder. For example, is the failure to evacuate the udders likely to result from poor operator technique, inappropriate machine settings or cluster characteristics, poor udder conformation, failure of letdown or poor cow behaviour, or (commonly) some combination of these factors? Thus, estimates of strip yield are a useful indicator to be interpreted along with all the other measures and observations being assessed. 11. Look for patterns in the recorded data. Consistent differences between strip yields in rear versus front quarters, or between right versus left quarters, usually indicate a problem of poor cluster alignment and/or uneven distribution of cluster weight between teatcups. 12. In herringbone sheds, where clusters are attached from the near side of each udder, diagnosis of uneven milk-out is facilitated if individual quarters are recorded as near hind or far hind, etc, rather than as left hind or right hind etc. 2

13. Cows whose milkings were cut short by a pre-set Maximum Milk-Out Time-limit, should not be included in a herd evaluation of strip yields for at least two reasons. Firstly, test stripping of these slow-milking cows would be disruptive to the milking routine and would, therefore, prolong the herd milking time unnecessarily. Secondly, it is better to concentrate on stripping only those cows on which the clusters have "finished". This strategy simplifies interpretation of the results to answer the question why in terms of a "failure to completely remove the available milk from an udder". Guidelines for interpreting the data Hand-stripping Assume that a problem of incomplete milk-out exists if more that 20% of all quarters produce strip yields of about 100 ml or more. Assume that a problem of uneven milk-out exists if more than 20% of the quarters in one particular quadrant of the udder produce strip yields of 100 ml or more, and the percentage figure is noticeably higher than that for quarters in other quadrants (as in the example sheet below for the near hind quarters). An example of a simple recording sheet is given in the table below. Recording and analysis could be simplified further by writing down only the results for quarters assessed as Poor. Cow ID Hand strip result for individual quarters Far front Near front Far hind Near hind 1 G G G P 2 G G P P 3 G P G P 24 G G P P 25 G P G G Total listed as P 0 3 5 11 Percent 0% 12% 20% 44% Percent of all quarters 19/100 = 19% Machine stripping Assume that a problem of incomplete milk-out exists if more that 20% of cows produced strip yields of about 500 ml or more. 3

Background Since the dawn of mechanical milking in the late 1800s, up until about 1970, dairy farmers in all countries were encouraged and advised to milk all cows as completely as possible. Most cows were either hand-stripped or machine-stripped to ensure that all of the available milk was extracted from udders at any given milking. Effects of incomplete milking on milk yield Experiments cited by Dodd and Griffin (1979) dating back to 1936 indicated that lactational yields were reduced by about 3% when 0.5 kg of the available milk was left in an udder after milking. According to Ebendorff et al. (1987), leaving an average of 0.55 kg of strippings milk per half udder per day reduced milk yield in the non-stripped half udders by about 10% over four lactations. Other studies on incomplete milking reviewed by Hamann & Dodd (1992) showed similar, high losses. For example: leaving 0.2-0.9 kg of strippings after each milking in the udders of heifers reduced lactational yields by 0.5-7% leaving 0.5-0.9 kg strippings in the udders of mature cows reduced lactational yields by 6%-10%. It is possible that the yield loss reported by Ebendorff et al. was magnified by compensatory effects between stripped and non-stripped udder halves. The same subtle but powerful effect of compensatory shifts in milk yield between quarters within an udder might have magnified the apparent differences in at least some of the other studies reviewed by Hamann & Dodd. If, for example, milk yield in non-stripped quarters falls by 3% but the yield of well-stripped quarters increases by 2% due to a compensatory shift between quarters, the penalty of incomplete milking will appear to be 5% rather than the actual loss of 1% per udder. Such a compensatory shift would help to explain (to me, at least) why the apparent losses in lactational yield were higher in mature cows compared with heifers. Despite the widely-reported average loss (of about 3% when 0.5 kg of available milk is left unharvested), routine machine stripping has not been advocated for at least the last 40 years in Australia. Australian dairy farmers have been willing to sacrifice small losses in lactational yield to obtain the great benefits of improved labour productivity. Effects of incomplete milking on mastitis Published evidence on the relationship between completeness of milking and new mastitis infection rates is conflicting. Most of the older publications reviewed by O'Shea (1987) show that mastitis increased when machine stripping was omitted. In contrast, at least nine studies indicated that small quantities of milk left in the udder did not increase new infection rate or clinical mastitis, and at least three studies found higher levels of infection associated with machine stripping. The latter findings are not surprising. It is likely that the new mastitis infection rate would be increased by vigorous machine stripping which causes sudden air admission into one or more teatcups just before the teatcups are removed. Furthermore, extra weights placed on claws affect their balance and may increase cup slippage which increases the risk of mastitis, especially when cup slips occur at or near the end of milking. 4

New knowledge from research on Shorter Milking Times (SMT) The results of recent research on raising the threshold setting for automatic cluster removers (Rasmussen, 1993), in conjunction with setting a maximum time limit for milking slow cows (Clarke et al. 2004, 2006; Jago et al. 2010, Jago et al. in press; Burke & Jago, in press), have opened up new possibilities for milking herds more quickly with no apparent adverse effects. In Rasmussen s pioneering study (Rasmussen, 1993), milking time was reduced by 0.5 min per cow with no loss of milk yield when the end-of-milking setting for automatic cluster removers (ACRs) was raised from 0.2 kg/min to a flow-rate threshold of 0.4 kg/min. Milking equipment companies in the USA soon realised the potential for shortening machine-on time for individual cows and for reducing herd milking times. Threshold flowrate settings for ACRs were raised from default settings of about 0.3 kg/min up to 0.5 kg/min for herds milked twice per day, and to levels as high as 0.9 kg per min for some herds milked thrice daily. At the same time, the typical settings of 10-20 sec time delay for ACR activation were shortened to 0-5 sec. Under typical conditions of milking management in large commercial herds in the USA, these changes reduced milking times by up to 1 min per cow with no reported loss of milk yield, no change in SCC or mastitis levels. These excellent results were obtained in high-producing, North American herds milked three times per day with (typically) good pre-milking teat preparation, calm consistent milking routines, narrow-bore liners, and milking units that were positioned carefully on the udder by the operator(s) at the start of milking. Because these milking conditions are rare in Australia, a series of experiments was conducted under more typical Australian milking conditions - including twice-daily milking, minimal udder pre-milking preparation, use of relatively widebore liners, and no particular effort made to ensure good cluster balance or alignment. Higher threshold settings combined with fixed time removal of clusters The first results from Clarke et al. (2004) showed that the use of timed maximum milking durations could save up to 35% of normal milking time of slow milking cows with no adverse effect on their daily milk yield (averaging up to 26 L/d), milk composition, teat condition or cow behaviour. Subsequent studies (Clarke et al., 2007) indicated that early termination of milking had no significant effects on incidence of clinical mastitis, sub-clinical mastitis or average SCC in healthy quarters or in quarters sub-clinically infected with either S. aureus or Str. uberis mastitis pathogens. These relationships have not been examined in Str agalactiae herds. The major practical outcome of these studies by Clark et al. has been a marked reduction in the time required to milk herds in which the SMT guidelines are implemented. The combination of a pre-set maximum milking time and an end-point determined by ACR threshold (whichever comes first) has great potential to shorten milking times per herd by reducing or eliminating the bottlenecks caused by slow-milking cows. The initial goal, set for Australian conditions, has been to remove clusters from about 80% of cows at a flow-rate threshold of 0.4 kg/min while truncating the milking time of the slowest 20% of cows (CowTime Quicknote, 2007) and, thereby, inducing some undermilking in these cows. Two new studies in NZ (Jago et al. 2010; Jago et al. in press) have confirmed and extended the results of these Australian studies in two major ways: firstly, the time-saving strategy of truncating the milking of slow cows can be started before cows reach the peak of their 5

lactation. Secondly, further time savings can be achieved when the SMT strategy is applied more aggressively. On average, 30% of cow-milkings were truncated in the NZ study compared with a less aggressive target of 20% in the studies by Clarke et al. The NZ field study by Jago et al. (2010) evaluated four criteria for determining the end-point of milking for their effects on production, udder health, SCC, teat skin condition and mean milking time per cow. Milking end-point treatments were: clusters removed at a milk flow-rate of 0.2 kg/min (ACR200); clusters removed at a milk flow-rate of 0.4 kg/min (ACR400); clusters removed at a milk flow-rate of 0.2 kg/min or at a pre-determined maximum cups-on time, applied from start to end of lactation (MaxT early); clusters removed at a milk flow-rate of 0.2 kg/min until peak lactation, then at a predetermined maximum cups-on time or the pre-set flow-rate from the peak to the end of lactation (MaxT peak). Maximum milking times were set at 7.5 min for am milkings and 5.4 min for pm milkings. Mean strip yield was significantly greater for cows in the MaxT early group at week 12 of lactation (average strip yield/cow: 0.20 kg for ACR200; 0.61 kg for ACR400; 1.24 kg for MaxT early; 0.22 kg for MaxT peak) but did not differ between groups in week 27. No significant differences were found for lactational milk yield, composition, teat condition or the proportion of cows with clinical mastitis throughout the lactation length study. Although mean SCC was slightly higher in the ACR400 group (123,000 vs 80,000 cells/ml), this small difference was thought to be due to the inadvertent allocation of several high SCC cows to the ACR400 group. Subsequently, no significant difference in SCC was found between ACR threshold settings of 0.2 and 0.4 kg/min in a follow-up, short-duration crossover study conducted in mid-lactation cows (J. Burke, pers comm., 2010). Discussion It is important to keep the goal of fast milking per cow in perspective relative to the goal of quicker milking of the herd. The average cups-on time per cow is not always (in fact not usually) the limiting factor in cows milked/hour. As long as the average cups-on time is not the limiting factor in determining cow throughput (the number of cows milked/hour) there may be advantages in extending the cups-on time to improve the completeness of milking for some cows, especially for old cows with high cell counts. According to one experienced udder health adviser, such cows often drop from 400,000 to less than 300,000 cells/ml simply by milking out otherwise under-milked udders. Bearing these points in mind, the Australian and NZ studies suggest the following preferred strategy for implementing SMT in herds milked twice daily (and, perhaps, once daily too). ACR threshold settings can be raised to 0.4 kg/min for milking typical Australian and NZ herds. However, despite the benefits claimed for even higher ACR threshold settings in herds milked in the USA (quicker milking, better teat condition, calmer cows), it is most unlikely that Australian or NZ farmers would gain any further benefits from raising ACR thresholds above 0.4 kg/min. Applying a pre-set time limit for early cluster removal, based on the average yield per milking of the highest-producing (eg, the top 25%) cows in any herd, is a far more effective way to ensure quicker milking of typical herds in Australia or NZ compared 6

with raising ACR threshold settings above 0.4 kg/min. Calculation of the MMOT based on the top 25% of the herd ensures that such cows, which may take longer to milk (only because they have above-average daily milk production) are not penalised. Although the NZ studies indicated no adverse effects of time-based truncation of milking for 30% or more cows from early lactation onwards, it might be worth applying a less aggressive formula for calculating the maximum milking time in the period between the start and peak of lactation, especially in higher yielding herds. {Note: mean milk yield per cow in the herd used in the NZ study was ~ 3000 kg for the 35-week trial period]. For example, a conservative starting point would be to set a max. cups-on time that would truncate the milking times of 10-15% of the herd. SMT should not be recommended in herds where Str agalactiae is present because, as far as we are aware, it has not been trialled in any herds infected with this pathogen. According to some early publications, it is likely that SCC and incidence of clinical mastitis will be increased by incomplete milking of cows infected with this pathogen. Completeness of milking in herds milked more than twice per day Completeness of milking becomes less important as milking frequency increases. Herds that are milked 3 or more times per day may benefit from a higher ACR threshold setting compared with that recommended for twice daily milking. The practical upper limit for ACR threshold settings is likely to be the peak flow-rate of the slowest-milking quarter - ie, about 1 kg/min. According to Stewart et al. (2002), ACR flow-rate thresholds can be increased to reduce the average cups-on time until a noticeable drop in daily production is detected. Until better data are available, a reasonable guideline for evaluating the completeness of milking, in herds milked three or more times per day, would be: < 0.75 kg average strip-yield per udder, when a representative group of cows is machine-stripped. Problems with hand-stripping The main problem with recommending hand-stripping as a quantitative assessment method is the great variation in stripping technique - and therefore in the amount of milk harvested - by different individuals. The range of techniques between people whom I have observed varies between a timid young veterinary student, who squeezed teats ineffectually and obtained almost no strippings, to an experienced adviser in Denmark. This vigorous Dane consistently obtained about three times the volume of my own strip-yields in the same herd. Furthermore, hand-stripping is a difficult, messy and potentially dangerous measurement procedure in herds where cows are not used to having their teats handled after milking. For better or worse, this includes most Australian and NZ herds. Also, strip yields obtained by hand-stripping often are significantly lower than quarter or udder strip yields obtained by machine-stripping (Davis & Reinemann, 2001). Given these potential problems and sources of variation, it would be sensible for Australia to adapt the practical method of assessment described Adrian Joe (2010). According to Joe, completeness of milk-out is subjectively based mostly on visual assessment of udder fill, feel of milk in the udder, and ease of hand-stripping (of 50 cows per herd) immediately after machine milking. The amount of milk that can be stripped is correlated to this visual assessment as: Very good (udder visibly wrinkled and very little milk stripped, <100 ml); Good (udder with some wrinkles and some milk able to be stripped, 100-500 ml); or Poor (obvious filling of the udder and very easy to strip milk, >500 ml). 7

Practical application of SMT and MMOT Research conducted by the Department of Primary Industries, Victoria showed clearly that taking the cups off at a pre-determined MMOT saves time without affecting milk production, quality, mastitis or somatic cell count. In addition we monitored about 20 case study farms that implemented the MMOT and most saved about 15 minutes a milking with no effect on production or quality, said Darold Klindworth. A herd s MMOT depends on the average milk production per cow per milking. For example, 80% of cows in a typical Australian herd in which cows are producing an average of 10 L at a single milking will be milked in 6.3 min. IF MMOT was applied at this suggested target level of 80%, then the slowest 20% of cows in that herd would have their milkings truncated by pulling clusters off after 6.3 min. Similarly, and assuming that the milking system is functioning correctly, then herds with an average yield per cow of: 12 L/milking will be milked in 7.2 min 14 L/milking will be milked in 8.0 min 16 L/milking will be milked in 8.8 min 18 L/milking will be milked in 9.5 min 20 L/milking will be milked in 10.2 min CowTime s MMOT calculator is available on-line at: www.cowtime.com.au. CowTime recommends using a basic timer for a few milkings to get a feel for the slow cows. In practice, however, it appears that many farmers struggle to see how they can easily incorporate the MMOT concept and how to implement a fixed time removal (eg, how to measure it, how to set it, how to apply it, and how to use an alarm clock or similar). Furthermore, there are other considerations to milking times, platform speeds, etc. (e.g. ability of cows to eat their grain ration in the allocated time, ability of operators to keep up with a certain platform speed). An alternative simple, practical method of incorporation has been suggested (by R.Dyson, pers.comm). In herringbone dairies, advise farmers not to wait for the last X number of cows on each side where X is a proportion of the number of units on each side. A simplified version of MMOT can be applied at the 10-15% level, meaning in a 10-a-side, advise farmers not to wait for the last cow; and in a 20-a-side, don t wait for the last 2 or maybe 3 cows, etc. In a rotary dairy, it means selecting a platform rotation time (allowing for the time lost across the bridge from cups off to cups on), and then saying no cow goes around twice unless there is a specific reason (eg, kicked the cups off) Another very simple mode of implementation is to advise farmers/milkers that they don t have to wait for that tough old bitch anymore Every farm has one or more tough old bitches and simply not waiting anymore for these cows makes a big difference to some farmers and milkers. 8

References Burke, J. and J. Jago (in press, ADSS). The effect of two milk flow-rate thresholds for automatic cup removal on milking duration, production and somatic cell count of peak-lactation dairy cows. Clarke, T., E.M. Cuthbertson, R.K. Greenall, M.C. Hannah and D. Shoesmith. 2004. Milking regimes to shorten milking duration. J. Dairy Res. 71:419-426. Clarke, T., D. Cole and R.K. Greenall, M.C. 2007. Shorter Milking Times research program: Technical information package for advisers, December 2006. Department of Primary Industries, 1 Spring St., Melbourne, 3000. ISBN 978-1-74199-065-2 CowTime Quicknote. 2007. Maximum Milk Out Times. Website: www.cowtime.com.au Davis, M.A. & D.J. Reinemann. 2001. Methodology of measuring strip yield. In: Proceedings of 2nd International Symposium on Mastitis & Milk Quality, NMC- AABP, Vancouver, Canada, pp387-391. Dodd, F.H. & T.K. Griffin. 1979. Milking Routines. Machine Milking Technical Bulletin 1, Ch 7, National Institute for Research in Dairying, Reading, England, pp179-200. Ebendorff, W., K. Kram, G. Michel & J. Ziesack. 1987. Machine stripping, milk yield and udder health results over 4 lactations. Milchwissenschaft 42:23-25. Hamann, J & FH Dodd. 1992. Milking Routines. In: Machine Milking and Lactation, Eds: AJ Bramley, FH Dodd, GA Mein & JA Bramley, Insight Books, England, pp79-80. Jago, J, J Burke & J Williamson. 2010. Does reducing cups-on time affect milk production, clinical mastitis, somatic cell count or teat condition? Proc. 5 th IDF Mastitis Conference, Ed: JE Hillerton, Christchurch, NZ, p468-472. Jago, J, J McGowan & J Williamson. In press. NZ Vet J. Setting a maximum milking time from peak lactation: effects on production, milking time and udder health Joe, A, SW Cranefield, IC Hodge & TG Clarke. 2010. Proc. 5 th IDF Mastitis Conference, Ed: JE Hillerton, Christchurch, NZ, p363-369. O'Shea J. 1987. Machine milking factors affecting mastitis. Machine milking and mastitis, Bulletin of the International Dairy Federation No. 215, Brussels, Belgium, pp5-32. Rasmussen, M.D. 1993. Influence of switch level of automatic cluster removers on milking performance and udder health. J. Dairy Res. 60:287-297. Stewart, S., S. Godden, P. Rapnicki, D.A. Reid, A. Johnson and S. Eicker. 2002. Effects of automatic cluster remover settings on average milking duration, milk flow, and milk yield. J. Dairy Sci. 85:818-823. 9