AUTOMATIC MILKING SYSTEMS AND MASTITIS

Transcription

1 AUTOMATIC MILKING SYSTEMS AND MASTITIS Kees de Koning Manager Dairy Campus, Wageningen University & Research Centre, Boksumerdyk 11, 9084 AA Leeuwarden, the Netherlands, Internet: Contact: Mastitis is associated with two main milk quality aspects that are used in many dairy countries: somatic cell count (SCC) and with clinical mastitis, visible abnormal milk (watery, clots). The farmer has the job to deliver milk of sufficient quality and healthy cows. Therefore in order to deliver high quality milk, attention should be given to both adequate detection and prevention of mastitis (Hogeveen et al, 2010). This is the same for all dairy farms including farms with automatic milking systems. Udder health on farms milking with an automatic milking system (AMS) can be a problem. The percentage of cows with a high somatic cell count (SCC) seems to be higher after the introduction of an AMS and on two out of three farms the incidence of clinical mastitis was also higher (Hillerton et al 2004). Besides a decrease in udder health short after the introduction of an AMS, the bulk milk SCC (BMSCC) is shown to be, on average, higher in the long term (Van der Vorst et al 2002, De Koning et al, 2004). The basic mechanisms leading to intra-mammary infections will be the same on farms with an AMS as on farms with conventional milking. However there are essential differences between AMS and conventional milking, such as detection of clinical mastitis, use of sensor information and the absence of a milker. USE OF SENSOR TECHNOLOGY Following the introduction of electronic animal identification, applications have been sought for. Automatic feeding dispensers were introduced, electronic milk meters, pedometers and so on. Due to the high costs related with mastitis, from the beginning of the eighties, much attention was paid to the development of sensors to measure electrical conductivity of milk as a measure for sub clinical mastitis. Although such sensors became commercially available in the eighties and early nineties, they were never applied in a large scale. Due to the introduction of automatic milking, the milker is no longer present at time of milking, the need for sensors to detect clinical mastitis and abnormal

2 142-2 nd International Symposium of Dairy Cattle milk became apparent. So interest in the application of sensors to detect mastitis and abnormal milk has been gaining considerably in the past decade (Hogeveen et al, 2010). AM-systems are equipped with sensor technology and integrated data management systems to observe and to control all relevant processes. With their sensors AM systems collect enormous amounts of data, that have to be processed with appropriate software (Hogeveen & Ouweltjes, 2003, De Koning & Ouweltjes, 2004). The challenge for both manufacturers as end users is to detect in the data the abnormalities, so actions can be taken. Because abnormalities are rare this is called management by exception. One of the main challenges is clinical mastitis, especially in relation to abnormal milk. By definition, milk of cow suffering from mastitis has an abnormal visual appearance. It is also one of the most frequently occurring diseases in dairy cattle, and is responsible for the majority of abnormal milk. Despite this, milking a cow with abnormal milk is a rather exceptional event on most dairy farms. As an example, assume that abnormal milk is always caused by mastitis, that 25% of all cows have one case of mastitis each year, and that each mastitis case causes 10 milkings with abnormal milk. For a 100 cow herd with 310 days in milk per cow per year and 2.5 milkings per cow per day, only 0.32% of all milkings will be abnormal. This figure clearly indicates that, even with a high mastitis frequency, the percentage of abnormalities is very low. Modern milking systems are equipped with various sensors ranging from sensors to control the milking process till sensors that analyze the milk quality in several ways, like milk composition, cell counts, blood detection, conductivity, progesterone, and so on (table 1). All these sensors require smart data handling solutions in order to help the farmer to make the right decision. In the last decade special guide lines were developed and approved for automatic milking systems and involved sensor technology within the framework of the International Standards Organization (ISO20966, 2007). These guide lines include an Annex, dealing with methods of detecting abnormal milk and the interpretation of test results. The Annex describes a minimum sensitivity of 80%, combined with a specificity larger than 99%, stating a false alert rate smaller than 10 per 1,000 milkings (Hogeveen et al, 2010).

3 IV Simpósio Nacional de Bovinocultura de Leite Table 1 - Overview of sensors in dairy production (De Koning et al, 2012, ICAR Cork) Measurement Indications Management issue Hormones Heat Reproduction Urea Ketosis Feeding Proteins Inflammation Health Pathogens Mastitis/diseases Health / Product quality Conductivity Mastitis Health Residues Milk quality Product quality Yield, components Feed quality Feeding Body condition Condition, feed intake Feeding Locomotion score, activitiy Claw health, heat Health, fertility Location (GPS) Diseases, welfare Health Rumination Feed intake Feeding, health Sensors to detect Mastitis and abnormal milk Intra-mammary infections will lead to changes in the composition of milk. Clinical mastitis is per definition an intra-mammary infection where visible changes in the milk (watery, clots), the udder (red, swollen), or both occur as an effect of the inflammation process. Also an increase of somatic cell counts (SCC) will be seen, therefore SCC is the basis of most all milk quality programs. Besides increase in SCC, other compositional changes in milk will occur. So different sensors have been developed that can detect such changes on-line (table 2). All these sensors produce data, some even large amounts of data. However without data analysis to generate information and possibilities to transfer this information in to action, the data in itself is not useful. Many algorithms have been proposed, developed and described to alert the herdsman that a cow has mastitis during a milking. A very simple example is the use of a threshold: if the measured value (say EC) is above that threshold, an alert is generated. Because of the low prevalence of the event that has to be detected (clinical, subclinical, abnormal milk), high demands are necessary on the processing of these sensor data (Hoogeveen et al, 2010). The algorithm is essential in transferring the sensor data into an easily interpretable value that can be translated into an action. Without action no success!.

4 144-2 nd International Symposium of Dairy Cattle Table 2 - Different sensors for mastitis detection and the measurement principle Mastitis sensor Electrical conductivity (EC) L-Lactate dehydrogenase (LDH) Color Somatic cell count (SCC) Near Infrared (NIR) Californian mastitis test (CMT) Measurement EC is the measure of the resistance to an electric current. The technology is known in dairy for automatic cluster detachment (detection end of milking) and to measure the changes in conductivity in milk as a result of mastitis. In most automatic milking systems the conductivity is measured per quarter offering the possibility to compare udder quarters, thus increasing test characteristics (Hogeveen et al, 2010) LDH is the result of an enzymatic reaction following mastitis. In fact LDH is a responsive indicator of mastitis as a result of the animal s immune response against infections. Color can be used as a direct measure of the physical characteristics of abnormal milk (mostly due to clinical mastitis). The principle of the sensor is based on the reflection of light. White milk will reflect more light. Such sensors use normally the red, green and blue wavelengths of light (RGB). SCC is the most widely used parameter used for detection of mastitis (SCC). SCC measurement is routinely carried out in laboratories by using rapid and accurate technologies and often used to monitor udder health. NIR has shown to be able to measure SCC in raw milk and commercial available in line analyzers are now introduced. Moreover sensors that measure SCC on-line, based on the CMT principles are now commercially available on several automatic milking systems. Algorithms therefore can make a huge difference in the performance of sensor management systems. Because mastitis is associated with many changes in the cow and the milk, a combination of more than one sensor will be more useful. The most used sensor for EC is often combined with milk yield and milk temperature. Sensitivity varies from 43% to 100%, while specificity varies from 69% to 99.75% (Hogeveen et al, 2010). Although is some studies a sensitivity of 100% was found when using EC to detect clinical mastitis, several studies showed that in general sensitivity is rather low when tested under field conditions (Hogeveen et al, 2010). Table 3

5 IV Simpósio Nacional de Bovinocultura de Leite adopted from Hogeveen et al, 2010 gives an overview of more recent studies with new sensor devices for color, SCC or LDH, even some measurement were included not carried out in the milk. However also in these studies a large variation in sensitivities, specificities and time windows can be found. Table 3 - Overview of performance, including sensitivity (SE), specificity (SP) and used time window of detection models using different sensors (Adapted from Hogeveen et al, 2010) Paper SE SP Time Window Definition of event Kamphuis et al, d Treated cases 2008 of mastitis Kamphuis et al, <1d Clinical 2010 mastitis Kramer et al, d Treated cases 2009 of mastitis Friggens et al, d Treated cases of mastitis Sensors EC, in-line somatic cell count EC, color, milk yield Milk yield, dry matter intake, water intake, activity, previous diseases L-lactate dehydrogenase Decision support systems based on sensor measurement are a valuable tool in mastitis control on farms using an AM-system. Including cow information like lactation stage does predict the risk of clinical mastitis. However adding such non-ams cow information to the data as shown by Steeneveld et al (2010) did not improve the clinical mastitis detection performance as is often expected by farmers. DUTCH MASTITIS RESEARCH PROGRAM PROJECT RISK FACTORS FOR MASTITIS Within the Dutch Mastitis Research Program ( a study (Neijenhuis et al, 2009) was conducted to identify risk factors for mastitis and to translate these risk factors into preventive measures. The first step in the research was to gather expert knowledge on udder health during two sessions. A list of risk factors and management measures was constructed. This list was complemented with knowledge from scientific literature. Not only the specific risk factors for udder health on AMS farms were implemented in this list but also the risk factors

6 146-2 nd International Symposium of Dairy Cattle known on farms with a conventional milking system. The list was used as input for constructing a research data collection protocol. The second step was the setup of a research protocol to be used on the farms. The research protocol consisted of observations on cows (like hygiene and locomotion), AMS (like teat cup attachment, teat spraying and hygiene of the AMS parts) and stable (like the presence of a waiting area). Furthermore, an extensive survey was conducted to collect data on the farm structure, the information sources used by farmers, handling of the AMS, mastitis management, housing and feeding. Farmers were asked to fill in a list with propositions (to get an idea of the attitudes and interests of the farmer). Finally the data from the AMS (robot performances) and the milk production registration (MPR) were collected. Udder health was described in terms of clinical mastitis incidence (given by the farmer) and subclinical mastitis in terms of mean cow somatic cell count (SCC) and fraction of new high SCC animals from the MPR data from November 2007 until November AMS farms with at least one year experience with automatic milking, were asked to participate through their dairy cooperative. Out of the more than 1,100 farms with automatic milking systems, 150 farms were visited by pairs of trained master students in veterinary science. All data were analyzed by univariate and multivariate modeling. Using principal component analysis, for each of the following clusters, the most important variables were selected: observations on cows, on AMS and in the stable, survey and characterization of the farmer based on combination of propositions. These variables were modeled in a multivariate analysis with udder health marks as dependent variables. In this model farm size, milk yield and milking intervals were tested, after which variables on bulk milk SCC and mean cow SCC around the period of switching to AMS were added. Finally, remaining data from the AMS were tested. RESULTS & DISCUSSION The average size of the farms was 84 dairy cows (with a range from 30 to 420 cows), with an average 305 day milk production of 9,020 kg (range 5,500 11,000). The weighed average cow SCC of the MPR was 262,000 cells/ml, on a log scale, the average cow SCC was 4.6 (range ). The average percentage of cows with an increased SCC was 24 % (range 6-42). Average new high SCC was 10 % (range

7 IV Simpósio Nacional de Bovinocultura de Leite ). Average incidence of clinical mastitis was 27 cases per 100 cows per year (range 1-135). During statistical analyses, it became apparent that the variable clinical mastitis incidence (representing farmer s reported incidence of clinical mastitis) was a difficult variable to work with (Neijenhuis et al, 2009, Dohmen et al,2010). Many of the found associations seemed, from a biological point of view, to be the result of a high percentage clinical mastitis rather than being the cause of a high percentage of clinical mastitis. Results indicated that smaller farms with an AMS had, on average, a better udder health than larger dairy farms. Also the milk production level of cows was associated with udder health on farms with an AMS. A higher milk production was related to a better udder health. Overall farm management may play a role in this association. The main areas related with good udder health were hygiene of cows, and especially teats, proper milking technique of the AMS to ensure proper teat health (ringing of teat base and hemorrhages), use of a waiting area, preventive health care (e.g., controlling BVD) and time spent on health control of dairy cattle. The udder health status before introduction of AMS was an important factor explaining the udder health status when milking with an AMS. For advising an AMS farmer to improve udder health, no full blueprint can be given. Clear attentions and measures based on inspections on cows (hygiene, milking intervals and health) should be tailor-made per farm. The research project provided a number of clear areas for attention for good udder health on AMS farms. Moreover, the results showed that professional skills of the farmer are an important contributor to good udder health on AMS farms (Dohmen et al, 2010). CONCLUDING REMARKS As with conventional milking also with automatic milking mastitis will occur. Automatic milking systems do have a lot of sensors available. The most applied sensor measures electrical conductivity. New sensor developments are driven by the introduction of automatic milking and are using NIR, in line measurement of SCC and LDH. Using smart algorithms to analyze the sensor data into information that can be applied by the farm manager is the key challenge. Moreover this information has to be transferred into real actions, otherwise this information is not useful. In general it can be stated that the professional

8 148-2 nd International Symposium of Dairy Cattle skills of the involved herdsman are a key-factor in achieving a good udder health, both for conventional farms as for farms with automatic milking. REFERENCES De Koning, C.J.A.M., B. Ipema, P. Hogewerf, P. Huijsmans, 2012, The role of new on-farm technologies in dairy herd improvement (DHI) and farm management, ICAR conference Cork 2012, Dohmen, W., F. Neijenhuis, H. Hogeveen, 2010, Relationship between udder health and hygiene on farms with an automatic milking system, Journal of Dairy Science 93 : Hillerton JE, Dearing J, Dale J, Poelarends JJ, Neijenhuis F, Sampimon OC, Miltenburg JDHM, Fossing C. 2004, Impact of automatic milking on animal health. In: Meijering A, Hogeveen H, De Koning CJAM (eds). Automatic milking A better understanding. Pp Wageningen Academic Publishers, Wageningen, the Netherlands, Hogeveen, H., C. Kamphuis, W. Steeneveld, H. Mollenhorst, 2010, Sensor and Milk Quality the Quest for the perfect alert. In: The first North American Conference on Precision Dairy Management www. Precisiondairy2010.com. Neijenhuis, F.N., J. Heinen, H. Hogeveen, 2009, Automatic milking: Risk factors for udder health, Report 257 (in Dutch), Wageningen UR Livestock Research. Neijenhuis, F., H. Hogeveen, K. de Koning, 2010, Automatic milking systems: a Dutch study on risk factors for udder health. In: The first North American Conference on Precision Dairy Management www. Precisiondairy2010.com. Steeneveld, W., C. Kamphuis, E. Mollenhorst, H. Hogeveen, 2010, In: Decision support for mastitis on farms with an automatic milking system / Steeneveld, W., Utrecht : Faculty of Veterinary Medicine, Utrecht University, ISBN Van der Vorst Y, Knappstein K, Rasmussen MD Milk quality on farms with an automatic milking system; Effects of automatic milking on the quality of produced milk. Research Institute for Animal Husbandry, Lelystad, The Netherlands.