Strategies for Incorporating Robotic Milking into North American Herd Management

• Undoubtedly robotic milking technology will evolve further and new management strategies will emerge. In larger herds, automatic attachment devices may have future application in parlor milking without voluntary attendance.

Introduction

In Europe, robotic milking systems, available commercially since 1992, were in use on an estimated 750 farms at the end of 2000. (Hogeveen 2001). In North American, the first commercial robotic milking system was installed in Ontario, Canada in March 1999. Ontario now has 17 dairy farms with robotic systems consisting of 27 Lely Astronaut stalls on 16 farms. and 2, 4-box  AMS Liberty systems on 1 farm. In addition, 1 farm in Nova Scotia milks with 2 AMS Liberty systems and there are 23 Lely boxes on 13 farms and 5 DeLaval system on 4 farms in Quebec. USA robotic milking farms include 3 herds in Pennsylvania and 3 in Wisconsin. While the number of herds in Ontario, is too few to permit formal research, early observations gained from working with them, may be helpful in identifying issues relevant to the adoption of robotic milking in the United States and Canada.

The first 15 herds with robots in Ontario were surveyed in December 2000. Owners listed “expansion without hiring non-family labour”, “more frequent milking”, “flexible hours” and “lifestyle” among the main reasons for choosing robotic milking. A common complaint in the survey was that 10-15% of the herd did not voluntarily attend the milking box. Moving these cows with long milking intervals to the holding area or milking box was the major labor component of robotic milking on these farms. 7 Ontario robotic milking farms are operated by recent Dutch immigrants, who have exposure to robotic milking in Europe. These producers claim that this problem is less severe under Dutch management regimes. The more wide spread use of total mixed rations (TMR) and the potential for stray voltage have been proposed as possible factors that could contribute to lower voluntary attendance. (Rodenburg 2001a).Other management issues specific to North American management styles include performance in cold environment barns, and application to larger herds.

Feeding Management, TMR Rations and Robotic Milking

Robotic milking systems rely on the use of feed offered in the milking box, .to motivate cows to visit the milking box. “Forced cow traffic, in which cows coming from the freestalls must pass the milking box before being permitted to access feed in the manger also encourages cows to attend to the milking box. European dairy herds often make extensive use of pasture and computer feeders, and other means of feeding grain separate from forage remain popular. Dairy herds in North America have largely embraced TMR as the feeding system of choice. Benefits of this feeding system include:

When more than 10% of feed dry matter is fed separately from the mixture, the benefits of TMR feeding will be seriously jeopardized. Although TMR feeding conflicts with the “individual management” concept facilitated by robotic milking (Maltz 2000), North American dairy producers have a great loyalty to this feeding system. Of the 17 AMS herds in Ontario, 16 feed a mixed ration in the bunk, but all of these herds also feed pelleted grain ration in the milking box.

In January 2002, 11 Ontario Lely robotic milking herds cooperated in a field study examining feeding management factors contributing to voluntary attendance at the milking box. These producers provided daily records for three consecutive days, of all events in which the operator moved  cows to the robot for milking. Data collected included time of day, cow ID and the reason for the action. The analyses and composition of the pelleted concentrate and of the bunk fed ration were also recorded. Records of all visits to the milking boxes were downloaded from the backup files on the farm computer. The authors gratefully acknowledge the assistance of Lely Canada in formatting this information. One herd was eliminated due to errors in the computer records. For 10 herds the records were examined for the frequency with which cows were moved to the robot by the operator, defined as “involuntary milking” by “involuntary” cows vs. the frequency of voluntary milking.

Calculated daily averages across the 10 farms showed that 19% ± 12.5 of cows were involved in involuntary milkings, and 12.6% ± 8.6 of all milkings were involuntary. The maximum and minimum three day averages for individual farms ranged from 7.2 % of cows and 3.7% of milkings on the lowest farm to 50.8 % of cows and 34.1 % of milkings on the farm with the highest frequency of involuntary milking. Table 1 summarizes the reasons for involuntary milking and their frequency.

	% of cows involved in one or more involuntary milking	% of milkings involuntary
Fresh or New (training)	2.3 ± 2.2	1.8 ± 1.9
Udder Conformation	5.6 ± 2.5	4.2 ± 2.0
Clinical Mastitis	0.2 ± 0.4	0.1 ± 0.2
Clinical Lameness	0.6 ± 0.9	0.3 ± 0.4
“Lazy”	10.3 ±10.1	6.2 ± 6.4
Total all reasons	19.2 ± 12.5	12.6 ± 8.6

 Table 1. Summary of involuntary milkings.

The reason, “udder conformation”, includes cows moved by the operator because they were programmed for manual milking as well as cows with long intervals due to repeated failure to attach. Cooperators indicated they had more of these cows than normal due to current high replacement costs. Many of these cows had close rear teats, which were tolerated as a temporary problem in late lactation. Three herds provided detailed descriptions of the udder conformation problems. Descriptions given in order of frequency were unbalance side to side, high rear udder, close rear teats, teats pointing out and entire udder too high.  

Cows were described as “lazy”, if they did not attend for voluntary milking , but their appearance and behaviour were otherwise normal. On some farms these cows were only moved when the interval since the last milking was more than 12 to 14 hours at the time barn chores were being done, with the result that these cows were milked once daily. On the majority of farms these cows were moved twice daily based on their established history of never attending voluntarily.

When involuntary milkings are included, data on milking frequency is influenced by the intervention of the operator and does not reflect the extent to which the management program achieves voluntary milking. To provide real data on voluntary attendance the following records were calculated for each herd for each of three days.

Cows on these farms visited the robotic milking box 3.34 ± 0.38 times per day, and were milked 2.34 ± 0.18 times. On the farms with forced cow traffic, average number of daily visits per cow was 3.40 ± 0.44 and milkings was 2.36 ± 0.22. This is fewer visits and milkings than reported in European studies (Van’tLand 2000, Wendl 2000). In the latter group, “visits” coincides with the number of meals of TMR. This is many meals fewer than the 12.1 (Vasilatos 1980) per day reported in a trial with free access and parlor milking. If fewer meals are related to lower dry matter intake (Dado 1994) there may be a cost in terms of lower milk production, associated with forced cow traffic.

The importance of feeding palatable concentrate in the milking box, to attract cows to voluntary milking, is illustrated by a case study on one of the farms. Prior to January 2002 a pellet formulated to help balance the needs of higher producing cows and to be low in cost was used in two Lely milking boxes. This concentrate suffered from poor pellet strength and a build up of fines in the bottom of the feeders was noted. In January 2002 a product formulated for high palatability, and manufactured with good pellet strength was substituted. Along with a major reduction in the amount of fines, the new pellet was lower (22 vs 24%) in protein, higher (1.96 vs 1.56 Mcal/Kg) in net energy lactation, higher in molasses content (3 vs 0 %) and higher in ingredients rated high in palatability (96 vs 65 %). Lower palatability ingredients in the former mix included small amounts of added fat, corn gluten meal and canola. Measures of voluntary milking were calculated three consecutive days 2 weeks prior and 2 weeks after the change was made. Results are summarized in table 2.

	Low Quality Pelleted Concentrate	High Quality Pelleted Concentrate
Voluntary visits/cow/day	3.40	4.04
Voluntary milkings/cow/day	1.72	2.06
% “lazy” cows	27.3	12.7
% “Lazy”milkings	16.0	7.1
Production/cow /day (liters)	25.8	26.3

Table 2. Voluntary attendance in a case study herd switching from low to high quality pelleted concentrate.

The benefit of higher pellet quality in this case included and increase in the frequency of voluntary visits, but the main impact on herd management was a dramatic decrease in the number of lazy cows that did not attend voluntarily.

Data from the 10 cooperating herds were used to calculate these same four measures of voluntary attendance. Herds were ranked for the following management and feeding related parameters:

Herds were ranked for each of the above parameters. Because individual herd data was quite variable, indicators of voluntary attendance were averaged across three herds with adjacent rankings and plotted against the management and feeding parameters. Plots with a slope near zero represent parameters which had no association with voluntary milking  in these herds.

More cows per milking box and a larger volume of milk produced per milk box were associated with a decrease in measures of voluntary attendance. As illustrated in Fig. 1 this field data suggests there is a marked increase in the % lazy cows when there are more than 60 cows per milking box. The trend line for daily milk production per milking box is similar. (Fig. 2) Although this field study is based on a small number of herds, results for these parameters agree with the more formal research of others. (Van’t Land 2000)

As shown in Fig. 3, herds with less than a year of experience with robotic milking had higher numbers of lazy cows and lower frequency of visits and milkings. Widespread confidence in the other trends and associations reported here may be ill advised due to the small sample size, but fact that the above results follow expected trends does inspire some confidence. Seven of the 10 herds used forced cow traffic, but in these comparisons, this was not associated with a difference in measures of voluntary use.

Frequent feeding and frequent trips through the barn to push up remaining feed are often recommended as a way of encouraging more frequent meals and higher feed intake from TMRs.

Feeding times per day	5	2	1
No. Herds	1	3	6
% lazy cows	9.6	9.7	10.7
% lazy milkings	4.2	5.8	6.7
Visits per cow per day	4.21	4.16	4.17
Milkings/cow/day	2.21	2.29	2.17

Table 3. TMR feding reguency and measures of voluntary milking

Table 3 suggests that in these herds more frequent feeding of TMR led to a slight increase in visits to the milking box and fewer lazy cows. When only the 7 forced traffic herds were included, feeding twice per day vs once per day resulted in 7.2 vs 11.6 % lazy cows, 4.2 vs 7.6 % lazy milkings, 4.29 vs 4.21 voluntary visits per cow and 2.39 vs 2.15 voluntary milkings per cow. The number of times stale feed was pushed up and whether or not the farm had reported orts did not affect measures of voluntary milking. Based on these results feeding fresh TMR several times per day, especially when combined with forced cow traffic, may be an effective way to encourage voluntary milking in TMR fed herds.

When characteristics of the concentrate used in the milking box were plotted, the use of molasses, the use of flavouring agents and inclusion rates of higher palatability ingredients were not associated with more voluntary milking. This result is surprising since others (Arave 1989) reported a preference for flavoured concentrates. In one trial (Murphy 1997) a sweetener had no affect on intake of TMR , but in others there were increases in intake of sweet flavoured diets vs control (Weller 1989, Nombekala 1994). Palatability ratings in our study were based on literature (Amaral-Phillips 1993, Maiga 1997) in which highest palatability was assigned to brewers grains, distillers grains, hominy, molasses and beet pulp. Soybean meal, roasted soybeans, corn, barley and wheat midlings were ranked intermediate, raw soybeans, and canola meal were ranked low, and corn gluten meal, blood, meat and fish meals, tallow, bypass fats, mineral mixes, buffers and niacin were ranked very low. In our study, the variation in real palatability may have been small since 7 of the 10 concentrates included more than 90% palatable ingredients.

The other 3 were rated in the 60% range because they contained corn gluten feed, which was assigned a low palatability score it may not deserve. Molasses level ranged from 3% on 2 farms to 0% on 3 farms with intermediate levels on the remaining 5. Flavouring agents were used in 6 of the 10 concentrates. Lower protein, was associated with more frequent visits and milkings and fewer lazy cows as illustrated in Fig. 4.  Higher levels of mineral and higher levels of Nel showed no clear association with measures of voluntary milking, despite the fact that high levels of mineral would be expected to be unpalatable.

In TMR herds, the use of pelleted concentrates in the milking box as a supplement to a TMR provides an opportunity to enhance the diet of higher producing cows in the group with additional protein, commonly from bypass sources, additional energy, commonly from fat, and additional mineral or feed additives. Findings here are unclear, but based on published data showing that many of these ingredients are less palatable, this approach to formulating concentrates for use in milking boxes of robotic herds may be ill advised.

Management recommendations for computer feeders should be equally valid for concentrate in robotic milking boxes. Daily monitoring for uninterrupted feed flow, daily monitoring for presence of fines, weekly calibration of the feeding rate, provision of small portions with minimal carryover, precalving training or lead feeding, regular cleaning of the feed bowl and drop pipes, and good fly and odor control are advised (Pritchard 1999).

As shown in Fig. 5, the amount of concentrate fed in the milking box has no clear relationship with measures of voluntary milking in these 10 herds. Where there is a commitment to the principles of TMR feeding it is desirable to feed the minimum amount of grain necessary to attract cows for milking. Fig 5 suggests 1.5 to 2.0 kg per day was as effective in attracting cows as larger amounts in these herds.

The concentration of protein, energy and proportion forage in the total diet were also evaluated and plotted against measures of voluntary milking. As illustrated in Fig. 6, and 7, in these herds there were strong associations between high energy, low forage diets and  a decline in measures of voluntary milking. These diets resulted in both lower frequency of visits and milkings by cows attending voluntarily and in a much greater number of lazy cows. Among these herds, measures of voluntary milking appear to be impaired in diets with more than 1.66 Mcal per kilogram dry matter Nel or more than 48% concentrate. In Fig. 8 higher protein in the diet is associated with more frequent voluntary milking and fewer lazy cows. The association between high energy, low forage diets and a greater number of lazy cows on these farms confirms the anecdotal claims of Ontario owners of robotic milking systems. 

It is well known that high grain diets are associated with laminitis. (Manson 1988a). Perhaps the farms using using high grain diets in this study suffer from a level of “subclinical” laminitis, which is decreasing the mobility of cows. Attention to carbohydrate level and fermentation rates, matching rumen availability of protein, and attention to level and form of dietary fibre are the key factors which influence rumen acidosis and laminitis. Limits of 25 to 35 % NDF, with 75% from forage, 35 to 40% non structural carbohydrate, and 30 to 40% starch in the diet dry matter, and a ratio of forage neutral detergent fibre to ruminally degradable starch of > 1:1 have been recommended. (Nocek 1997) Accurate ration formulation with attention to conservative levels of starch etc. may be critical to achieving high milking frequency and fewer lazy cows in robotic milking herds. Since hoof trimming reduces severity of lameness on high grain diets (Manson 1989) regular hoof trimming may also be more important in robotic herds on TMR rations. The association of high protein diets with fewer lazy cows in this study is contrary to some work suggesting more lameness with high protein (Manson 1988b). In the case of forced cow traffic, it also disagrees with a study reporting fewer meals on high protein diets (Tolkamp 2000) It may be that higher protein resulted in better utilization of ruminal starch, thereby decreasing laminitis problems in study.

Alternatively it may be possible that cows on high grain diets are less aggressive due to a direct metabollic effect on behaviour. It has been reported that cows on high grain diets spend less time eating and ruminating and more time resting (Robinson 1997), and consume fewer TMR meals (Friggens 1998). If this is specific to TMR diets, forced cow traffic may be less effective in increasing voluntary milking with high grain TMRs.  Results of this study suggest robotic milking herds may need to consider lower grain feeding levels in the TMR, despite the fact that they may scarifice some production in the process.

The type of diets described as high grain and high energy in this paper are typical of feeding programs commonly used with higher producing TMR herds throughout North America. If these diets result in poorer voluntary attendance for milking and lower milking frequency, this may have very important implications for the adoption of this technology. Understanding this relationship better will be an important area of future research. For the time being, robotic milking herds should choose to formulate diets with slightly less grain, and accept slightly lower milk production in the interest of higher milking frequency and fewer lazy cows.  

Robotic Milking and Cold Housing

Although cold environments for dairy cattle exist in Europe, published reports have not addressed operation of robotic milkers in freezing conditions. Freestall barns in Ontario are constructed with a wide range in the amount of insulation, and operated at a wide range of winter temperatures. Among the 17 robotic milking facilities 11 are located in insulated barns with a ceiling liner under a “scissors truss”, covered with a vapor barrier and R15 to R25 blown in insulation. One is a two storey barn where hay storage above the housing area provides ample insulation. These barns are operated at above freezing temperatures and report no cold related problems with AMS. Four of the barns are constructed with a minimal amount of insulation as a drip barrier under the roof, and one barn is uninsulated. These producers report that robotic milkers cannot operate in freezing conditions. Reports of ice on lasers and rollers and frozen tracks and mechanical components demonstrate that the immediate environment of the robot must be maintained frost-free. In these barns, producers have alleviated freezing problems by enclosing the working area on the clean side of the robot with walls and a ceiling, and adding a small space heater and positive ventilation in this “room.”    

Stray Voltage

The most common source of stray voltage is neutral current generated by normal power consumption in the grounded neutral electrical distribution system used throughout Canada and the USA. Electrical distribution in Europe is phase to phase and with the rare exception of electric shock from ground faults, stray voltage does not occur. Although most recent research suggests that the practical significance of low levels of stray voltage is minimal, a behavioral, “avoidance” response is recognized to be the most likely first effect of exposure (Southwick 1995). This effect has been observed on farm as refusal to use computer feeders in which cows were exposed to shocks of 2 to 3 milliamps, or 1 to 1.5 volts in a mouth to hooves pathway. Robotic milking is highly dependant on voluntary visits by cows to the milking box (Lind 2000). If cows experience electric shocks when visiting the milking box, they will reduce their voluntary visits. Measurements taken from cow contact points on both types of systems in Ontario indicate the metal equipment is case grounded and provides a potential cow contact for stray voltage (Rodenburg 2001b). Since the metal floor is an integral part of the milking box, the cow is on an equi-potential plane while in the stall and therefore protected from stray voltage during milking. Cows are exposed to a “step potential” when entering and leaving the box. We have recently examined records from one Ontario herd where corrective action was taken to eliminate stray voltage measured at 0.38 volts at the entrance to the milking box. Computer records showed no change in frequency of visits or milkings and no change in milk production. In all probability this level is too low to be of concern.

On several of the farms the area beside the milking stall is slatted, providing minimal grounding and thereby alleviating any risk of stray voltage. This may be a practical solution in new construction of robotic milking barns. In solid floor barns, installation of transition gradients (Gustafson 1984) at the time of construction as illustrated in Fig. 9, may be a worthwhile preventative measure. As shown in Fig 9A. providing a gridded area near the AMS and positioning the gradient in a narrow traffic lane is effective where a holding area is used. A third option for mitigating stray voltage is the installation of corrective devices such as the Hammond Tingle Voltage Filter®, the Ronk Blocker®, or DEI Neutral Isolator®.  In North America, control of stray voltage may be an important consideration in robotic milking installations.

3.6 m ground rods 15 –30 cm apart		Rods bonded to milking stall Ground rods at 45° angle

Fig. 9 cross-section view (left) and top view (right) of an equipotential gradient.

 A welded wire mesh grid, bonded to the robot milking stall and extending into the concrete area on which the cow approaches and departs from the robot, prevents her from getting a shock when she steps onto the metal floor. If this “plane” ends abruptly the cow is exposed to a front to rear hoof shock at the point she steps onto and off of the plane, so it is necessary to create a “gradient” that ensures a gradual change in the exposure. This is done with 3.6 meter long ground rods driven at a 45 degree angle at the edge of the grid and bonded to it. Rods are spaced 15 to 30 centimeters apart. This gradient reduces the level of voltage between two points over the length of the cow by approximately 50%.

Gradient ground rods	Grid made of 10 x 10 cm mesh	Gradient ground rods

Fig.9A.  Top view of an equipotential plane and gradient for a robotic milking stall with a holding area and exit lane.

Robotic Milking and Larger Herds
In parlor milking, it is recommended that cows be housed in groups that can be milked in 1 hour. As milking parlors became larger and more automated, cow groups have increased in size to several hundred cows. Larger groups result in fewer gates, simpler manure and feed handling and simpler barn design. With robotic milking the potential for working with larger groups is largely unexplored. Average herd size of the 10 cooperating herds in this study was 98 ± 30 cows, with 57 ± 8 cows per milking box. With on exception cows were grouped so that each group of 67 cows or less had access to only one milking box. In one herd a group of 92 cows had access to both a right and left entry box. With few exceptions, individual cows used only the box they were originally trained to use. Our experience to date shows that considerable retraining is needed if cows are switched to the “opposite entry” milking box. Multi box systems are theoretically able to handle up to 150 cows in a group, but a trend to fewer boxes per robot and smaller cow groups with multi box systems is apparent in Europe today. Queing for milking adds a new dimension to interaction between cows in robotic milking herds. For group sizes greater than 100 cows, the ability to recognize all group mates may diminish (Grant 1997) resulting in altered behaviour at the milking box. Until further research defines this better, group sizes of 100 cows or less may be preferred for robotic milking.

In parlor milked herds it is also customary to group cows by stage of lactation, or milk production, to create groups with more uniform nutritional requirements and milking times. Efficient use of robotic milking systems depends on high utilization of the milking box. Groups with a uniform stage of lactation would be inefficient since the total milking time will be much longer when production is high than in later lactation. Further research is needed to define ideal grouping strategies for large herds using robotic milking. Until such studies are undertaken, single box systems based on groups of 60 cows with varied calving dates and multi box systems with groups up to 100 cows per group may be the only practical choice. Where the herd size results in many groups, grouping strategies that create groups with uniform age, (Phelps 1992) size or temperament may be beneficial .

Undoubtedly robotic milking technology will evolve further and new management strategies will emerge. In larger herds, automatic attachment devices may have future application in parlor milking without voluntary attendance. Speculation about “robot arms on rotary platforms” has likely been the topic of conversation in more than one farm office or university lunch room. But predicting the speed or direction in which new technology evolves is ill advised and will not be attempted here.    

References:

Amaral-Phillips, D. M., and R. W. Hemken. Using by-products to feed dairy cattle. U. of Kentucky. On-line publ. ASC-136.

Arave, C. W., D. Purcell, and M. Engstrom. (1989) Effects of feed flavors on improving choice of ten percent meat and bone meal dairy concentrate. J. Dairy Sci. 72:563.

Dado, R. G. and M. S. Allen. (1994) Variation in and relationships among feeding, chewing, and drinking variables for lactating dairy cows. J. Dairy Sci. 77:132.

Friggens, N. C, G. C. Emmans, I. Kyriazakis, J. D. Oldham, and M. Lewis. (1998) Feed intake relative to stage of lactation for dairy cows consuming total mixed diets with a high or low ratio of concentrate to forage. J. Dairy Sci. 81:2228.

Grant, R. and J. Albright, 1997,Dry matter intake influenced by cow grouping, behaviour Feedstuffs Dec. 8, pp12

Gustafson, R.J. and D.A. Folen. 1984. Transition Designs for Equipotential Planes in Dairy Facilities, Am. Soc. of Agricultural Engineers paper No. 84-4063, ASAE, St. Joseph, Michigan

Hogeveen, H., Y. van der Vorst, W. Ouweltjes, and B.A. Slaghuis. 2001. Automatic Milking and Milk Quality: A European Perspective, Proceedings of the 40th Annual Meeting of the National Mastitis Council, pp 152-162.

Lind, O., A.H. Ipema, C. de Koning, T.T. Motram and H.J. Hermann. 2000. Automatic Milking: Realities, Challenges and Opportunities, In Robotic Milking: Proc. of the International Symposium held in Lelystad, the Netherlands 17-19 August 2000, Wageningen Press. pp 19-31

Maltz, E. 2000,Precision Agriculture in Dairying: Individual Management by Automatic Milking Systems In Robotic Milking: Proc. of the International Symposium held in Lelystad, the Netherlands 17-19 August 2000, Wageningen Press. pp 132-142

Manson, F.J. and J.D. Leaver, 1988a,The influence of concentrate amount on locomotion and clinical lameness in dairy cattle. Anim. Prod. 47:185

Manson, F.J. and J.D. Leaver, 1988b,The influence of dietary protein intake and of hoof trimming on lameness in dairy cattle. Anim. Prod. 47:191

Manson, F.J. and J.D. Leaver, 1989,The effect of concentrates: silage ratio and of hoof trimming on lameness in dairy cattle. Anim. Prod. 49:15

Murphy, M.R., A.W.P. Geijsel, E.C. Hall, and R.D. Shanks, 1997,Dietary variety via sweetining and voluntary intake of lactating cows J. Dairy Sc. 80:894-897

Nocek, J. E., 1997.Bovine Acidosis: Implications on Laminitis, J. Dairy Sc. 80:1005-1028.

Nombekela, S. W., M. R. Murphy, H .W. Gonyou, and J. I. Marden. (1994) Dietary preferences in early lactation cows as affected by primary tastes and some common feed flavors. J. Dairy Sci. 77:2393.

Phelps, A, 1992,Vastly superior first lactations when heifers fed separately, Feedstuffs May 11, pp 11

Pritchard, D. E., M. L. Eastridge, and J. E. Winkler. (1999) Computerized concentrate feeders for dairy cows. Ohio State University Fact Sheet No. AS-0004-99.

Robinson, P. H. and R. E. McQueen. (1994) Influence of supplemental protein source and feeding frequency on rumen fermentation and performance in dairy cows. J. Dairy Sci. 77:1340.

Rodenburg, J. and D.F. Kelton, 2001a, Automatic Milking Systems in North America: Issues and Challenges Unique to Ontario, Proceedings of the 40th Annual Meeting of the National Mastitis Council, pp 162-169.

Rodenburg, J. D.F. Kelton, and K. Hand, 2001b, Canadian Experience with Automatic Milking Systems, Commission on Animal Management and Health, Session 1- Paper # 6, 52nd Annual Meeting of the European Association of Animal Production,Budapest, Hungary

Southwick, L.H. 1995. Testing for Stray Voltage - Was it Done Properly, Proceedings of the 34th Annual Meeting of the National Mastitis Council, pp 89-99

Tolkamp, B. J., D. P. Schweitzer, and I. Kyriazakis. (2000) The biologically relevant unit for the analysis of short-term feeding behavior of dairy cows. J. Dairy Sci. 83:2057.

Van’tLand, A., A.C. Van Lenteren, E. Van Scooten, C. Bouwmans, D.J. Gravesteyn and P. Hink, 2000, In Robotic Milking: Proc. of the International Symposium held in Lelystad, the Netherlands 17-19 August 2000, Wageningen Press. pp 167-176

Vasilatos, R. and P. J. Wangsness. (1980) Feeding behavior of lactating dairy cows as measured by time-lapse photography. J. Dairy Sci. 63:412.

Weller, R.F., and R.H. Phipps, 1989,Preliminary studies on the affect of flavouring agents on the dry matter intake of silage by lactating dairy cows” J. agric. Sci. Camb. 112:67-71

Wendl, G., J. Harms and H. Schon, 2000, Analysis of Milking Behaviour on Automatic Milking- In Robotic Milking: Proc. of the International Symposium held in Lelystad, the Netherlands 17-19 August 2000, Wageningen Press. pp 143-151

Related Links:

Alternative Feeds for Dairy Cattle in Northwest Minnesota: An Update. Dairy Update Issue 126.
Maiga, H.A., G.D. Marx, V. W. Crary and J. G. Linn.