Record-Keeping Systems and Control of Data Flow
and information Retrieval to Manage Large
High Producing Herds
MICHAEL A. TOMASZEWSKI
Deparbnent of Animal Science
Texas A&M University
College Station 77843
ABSTRACT US population continues to grow and as health
Record-keeping systems have pro- conscious consumers continue to desire a low
vided an essential link that significantly fat diet, the dairy industry must produce its
increases milk production. As new tech- products more efficiently and market them
nologies are introduced, they are in- more imaginatively. Increased production per
tegrated into total management programs cow and reduced per capita consumption of
that provide for proactive management. dairy products require highly efficient produc-
Maintenance of data flow, not only for tion units that are fine tuned for profitable
the producer but also for other users, production. The widely cyclical prices have
requires increased cooperation among been determined by market force in recent
the various sectors. Larger production years, not by federal dairy price support pro-
units demand products that integrate grams (4). Producers will have to adopt effi-
production and economic parameters to cient proactive management to compete in the
plan strategically for maximum profita- global dairy economy.
(Key words: record keeping, microcom- Historical Record Programs
puters) Successful record programs that have been
incorporated by large herds are those that have
INTRODUCTION adopted appropriate new technologies. No
other livestock enterprise has made the ad-
The need for refined integrated management vances in performance that the dairy industry
programs has escalated because of larger has made in the last 50 yr (27). These improve-
production units that require greater database ments can be mainly attributed to use of
accessibility and increased data analysis (18). records for management and the development
In 1952, >62% of US dairy cows were on and maintenance of a national database.
farms of c30 cows, but, by 1990, only 7% of As in-parlor milk recording meters became
the dairy cows were in herds of this size (1, available, milking machine manufacturing
26). Conversely, the percentage of US cows on companies developed procedures to capture
f m s with >lo0 cows per farm increased from cow side data and to develop on-farm data-
c1 to >42% in the same period. Accordingly, bases. producers, now with access to
milk production has increased as herd size microcomputer programs, began managing by
increased. The 9000-kg production record for a exception (i.e., cows that had a low milk
herd was surpassed several years ago. Accord- weight), and new in-line recording devices
ing to the USDA (l), one state now averages started to replace DHI supervisors in the
>9000 kg of milk per yr. The production goal parlors. The need to develop new systems to
of 13,000 kg will soon be realized. utilize this information became necessary. As
Record-keeping systems provide an oppor- herds continued to increase in size and as
tunity for the dajr producer to remain com- he
milking t r e times daily became more preva-
petitive in the present unstable market. As the lent, gathering data through historical methods
became more costly. Not only was more tech-
nician time needed to collect data, but also, in
Received June 22. 1992. some herds, cow through-put was slow on test
Accepted April 29, 1993. day. The DHI, through their innovative testing
1993 J Dairy Sci 76:3188-3194 3188
SYMPOSIUM: MANAGEMENT IN THE 21ST CENTURY 3189
program, developed Labor Efficient Records how production variables are calculated limits
programs that provided a procedure for large the use of monitoring programs that tend to be
herds, interfaced with on-farm collection com- used by managers of large, high producing
puters, and allowed for component sample on herds (25).
an abbreviated yearly schedule. For example, the Dairy Records Processing
Historical mainframe-based programs, Center at Raleigh has been an early provider of
available through professionally maintained electronic access to a central database (1 1). The
databases, have resulted in a highly reliable Raleigh center maintains four major databases
database. The meticulous maintenance of t h i s that are easily accessed by producers or others
database has resulted in a valuable resource that have access authorization. Managers of
that allows the development and refinement of large herds have spent disproportionately more
genetic evaluation procedures by which time accessing that data than those of small
producers can capitalize on superior genetic herds, and the trend toward increased access is
material (30). As herd size continues to in- expected to continue to increase in the future.
crease, provisions need to be incorporated to
provide for data flow, not only for genetic Microcomputer-Bared Programs
evaluation but also for maintenance of and
calculation of performance goals. As microcomputer availability increased
Historical extension education efforts have and as hardware cost decreased, additional data
been tied to data produced from the DHI became integrated with the on-farm databases.
records (19). Production data have been used to However, because of the regional differences
evaluate feeding programs and to rank cows on in herd size, discrepancies developed in data-
milk production and other production indices base compatibility and acceptability across the
such as fat-corrected and energy-corrected US. Some of the processing centers, although
milk. Most information was returned to the they had records for large herds, developed
producer in paper form. Unfortunately, re- “bundled programs” that required the purchase
cently the amount of data that was returned to of the software and the hardware at prices
the producers has increased to the extent of close to $lO,OOO plus routine processing
information overload. Some state extension charges.
programs have become proactive by develop- The rigidity of the processing system and
ing herdbook clinics that allow for a periodic the direction that DHI took in the mid 1980s
review of goals and that help identify potential did not advocate database maintenance and
management problems from comparisons with calculation of production parameters at the
these goals (10). Because most extension serv- farms, thus creating a void for information
ices have agents with agricultural backgrounds, management needed by managers of large
strong ties were developed with producers to herds, veterinarians, and consultants. As a re-
assist in development of sound management sult, private sector providers attempted to fill
programs. However, many states are changing this void and became dairy software vendors.
the agricultural training and background of One such product, Dairy Comp 305TM (Val-
their agents, and the link to producers is disap- ley Agricultural Software, Tulare, CA), w s a
pearing in some areas. Large herds are no developed by a group of veterinarians in
longer managed from preprinted forms; their California to meet their needs for increased
management demand access to a database that reproductive information within a record
allows for the development of customized management system (S. Eiker, 1992, personal
reports. communication), Managers of large herds
Electronic access to database information adopted this product because it was used by
systems has become more readily available. In their herd veterinarian, and no monthly DRPC
addition, the number of production variables processing charges were incurred if the data-
maintained in DHI databases has been enlarged base was maintained at the dairy.
since DHI moved from a test period program However, in order to manage herds with on-
to a test day system. Although DHI is a na- farm databases, means are needed to calculate
tional program, the lack of an implemented reference points that can be used to evaluate
national data definition dictionary that defines management goals for multiple herds in a re-
J o d of Dairy Science Vol. 76, No. 10, 1993
gion, thus necessitating the need to allow data generated information. These protocols are the
flow to a central collection point. Data flow embodiment of the Data Information Inter-
from records for large herds to a national change protocol, but they also contain en-
database was obstructed because no defined hancements that will aid the adoption of these
national standard was in place for data transfer. standards.
As recently as 1992, only one Diy ar Agricultural Data Interchange Syntax pro-
Records Processing Center had developed a tocol will provide an international standard for
program that allowed for calculation of the the dairy industry. Managers of large, high
total record by on-farm data collection. How- producing herds will have easy access to soft-
ever, that system did not provide for data ware developed in any part of the world.
transfer to a central database.
Automatic Milk Weighing Devices. All ma-
Because of proposed unification within the jor milk machine companies in the US offer
European community, the Dutch government devices for automatic measurement and record-
undertook a program designated the TAURUS ing of milk production at every milking (23).
project (6), to standardize data entry for the Inherent difficulties in cowside measurement
entire dairy sector. Many on-farm programs of fluid milk have slowed the development of
existed, but no integration existed among the marketable technology. The measurement de-
programs. A dictionary of production variables vice must be accurate over a wide range of
and the methodology to transfer that informa- flow rates, and its accuracy must not be af-
tion into the management system was created. fected by foaming. High initial investment and
In 1989, the National Cooperative DHI Pro- the lack of interfaces to other production soft-
gram Policy Board initiated a joint project with ware systems resulted in a poor adoption rate
the Canadian Milk Recording Board to de- in the US. Many of the early electronic meter-
velop a North American standard for informa- ing systems have been replaced because of
tion interchange among the on-farm collection inability to maintain accuracy under parlor
computers and the DHI systems. The Canadi- conditions.
ans had approached the Policy Board because However, milk measurement technology in-
milking machine manufacturers had indicated itiated changes in areas of dairy management
that returns on their investments would not be in large herds by monitoring other aspects of
sufficient to develop a protocol for Canada that the dairy operation, such as the time needed to
was not compatible with the US market. The milk a cow. The ability to monitor this time
manufacturers cooperated through the Milk and the times between cows became important
Machine Manufacturer’s Council. In the late management criteria in determination of effec-
1980s, because the market share was split tive milking procedures.
among several companies, this organization Electronic Identificution. Electronic iden-
endorsed the need for a common protocol to tification has been used in dairy management
interface with on-farm collection computers. for many years and is the key to data flow in
This effort resulted in the development of the large herd management. In 1973, electronic
Data Information Interchange format, which animal identification became operational and
provided for the collection of information from was used to automatically record individual
on-farm collection computers and the transfer milk production (21). In 1975, the idea was
to a centrally maintained database. further developed to supply concentrates to
Because of industry problems in Europe, individual cows. In 1976, the first cow identifi-
TAURUS standards were not accepted as a cation systems were on the market.
universal standard, and International Standards Recently, interest has been renewed in de-
Organization established a working group to velopment of a transponder that is implantable,
develop Agricultural Data Interchange Syntax permanent, and unique. Most current systems
protocols (P. Dukas, 1992, personal communi- are not battery operated. When a responder
cation). The committee was to develop a com- (transmitter) is in the vicinity of a receiver, it is
mon basis for the processing of agriculturally activated by the magnetic field from the re-
Journal of Dairy Science Vol. 76, No. 10, 1993
SYMPOSIUM: MANAGEMENT IN THE 21ST CENTURY 3191
ceiver. The responder then transmits its coded Eddy (13) reported that activity monitoring
number. Managers of large herds need such appeared to be a potential aid in detection of
systems to provide easy and inexpensive estrus. When steps made by the cows were
methods for data collection. counted, activity appeared to have risen 30 to
Proactive systems require the capability to 200% (14, 15, 22). Activity meters could detect
capture real-time production data. These sys- approximately 72% of the cows in estrus.
tems can be coupled with appropriate software, When Rossing and Ipema (21) combined tem-
which may be used to detect deviant milk perature and activity measurements, as shown
production. Integration of the electronic iden- in Table 1, detection rate was 93%. Walton
tification into a dairy data sensor collection and King (28) thought that additional variables
system provides the opportunity to record should be considered to predict estrus. Decline
other measures (Le., conductivity, activity, and in morning milk production appeared to be one
temperature). of the most reliable indicators of estrus.
Integrated collection systems require the use Mastitis Detection. Historical monitoring
of signals that can be measured quickly and and detection systems are inadequate (2) as
are automatically analyzed by appropriate soft- health regulations on interstate shipment of
ware algorithms to detect differences. This milk require lower somatic cell and bacterial
reference is used to prepare a prognosis for the counts. Handlers are unable to commingle milk
next measurement, which is then compared from herds with high counts to lower the com-
with the actual measurement (21). posite score. Historical conductivity measure-
Temperature, Conductivity, and Estrus. ments have been refined, and preliminary work
Sick cows and cows in estrus can be detected from The Netherlands indicates that quarter
through deviations from the cow’s normal monitoring will effectively integrate into most
state. Body temperature is an indicator of the systems (21). Routine monitoring of milk al-
overall status of the cow, but daily monitoring lows a producer to identify potential problems
in a conventional parlor is time-consuming and as they develop in the cow and, in turn, to
labor intensive. However, research (15) has isolate the problem and its cause: cow, ma-
shown that milk temperature, measured by sen- chine, or human. Udder infection is the leading
sors in the milking equipment, is a good indi- cause for loss of milk. Historical procedures,
cator of body temperature. Thus, milk temper- such as forestripping and cowside mastitis de-
ature could be recorded and used to monitor tection programs, have attempted to alert
the cow’s general status. producers to potential problems. However,
Activity is an indicator of estrus. In 1977, these tests depended on labor and, in general,
TABLE 1. Results of different methods of detection of estrus (separate and combined).
Method 1 2 1 2
Progesterone test] 82 31 100 100
Visual observation2 71 16 a7 52
Milk temperature3 61 23 74 74
Physical activity4 59 19 72 61
Visual observation, temperature, or both 78 26 95 84
Visual observation, physical activity, or both 78 26 95 84
Temperature, physical activity, or both 76 27 93 87
Visual observation, temperature,
physical activity, or all 80 30 98 97
‘Twice weekly reference method.
ZThree times daily.
Joumal of Dairy Science Vol. 76, No. 10, 1993
3 192 TOMASZEWSKI
were not included as a routine part of the 1OOO-cow facilities install milk metering
milking procedure. Research (9, 12) has shown equipment.
that mastitis changes the composition of the
milk. Fernando and Spahr (8) found that the Declrlon Modeling
length of milking interval and the type of
sample are significant factors affecting concen- Managers of large, well-managed herds
trations of somatic cells and osmolar compo- have access to greater amounts of data and a
nents of milk. Fernando (7) studied the effi- need for software that integrates production
ciency of on-line quarter milk electrical and financial information. Unfortunately, little
conductivity measures for mastitis detection or no financial incentive has been available
and showed that the mean of the 10 highest through funding channels to develop and to
electrical conductivities collected at 6-s inter- incorporate new techniques for handling infor-
vals throughout the milking was marginally mation for use in the large dairy operation.
better than an index formed by the 5 highest However, experiment stations in New York,
values. Fernando (7) also found that the most Minnesota, and Texas have been developing
reliable data for distinguishing between value-added products. One example is a herd
healthy and infected quarters was during the evaluation program developed at Cornel1 (24).
last third of milking, compared with data from This program allows for ready access to a
the first- or second-third of milking. Accuracy database on a mainframe or on a personal
of identification of infected quarters was im- computer and the maintenance of that informa-
proved 10% by consideration of data from tion on the personal computer. In Minnesota,
multiple milkings. The accuracy of the detec- software has been developed that accepts infor-
tion achieved by Fernando (7) with data from mation generated by the DHI, and herd diag-
multiple nonconsecutive milkings was 94.7% nostics have been developed using expert sys-
in identifying infected quarters and 90.3% for tem technology (s. J. Conlin, 1992, personal
uninfected quarters. communication). The Texas group has devel-
Recently, Rossing and Ipema (21) devel- oped software to assist in interpretation of
oped procedures that sample each quarter ev- production changes for each test day in order
ery 6 s with a sensor that can be mounted in to identify potential production problem areas
the claw piece. Comparison of the conductivi- (3).
ties for the quarters and comparison of those Technological Opportunities. From neural
data with previously collected data provided a networking to interactive video, the potential
method to identify deviant quarters. In this for database access is increasing. New technol-
way, 75 to 80% of mastitis cases could be ogies allow for increased access to data.
detected ( 0 . Managers of large, well-managed herds will
However, information that can be collected utilize these new technologies to enhance
through sensors is most useful when it is in- profit through monitoring their information
cluded as a part of a management system. sources to optimize their management opportu-
Nielen et al. (17) concluded that on-line sys- nities. New tools will increase information
tems that combine multiple data and perform awareness.
multifactorial analyses will be of interest to the
dairy industry. Within the past 2 to 3 yr, integrated Programs
computer-related sales to dairy producers have
been relatively stable. According to an industry Historical production records have been
representative (M. Juett, 1992, personal com- only concerned with production-related varia-
munication), 15% of US dairy operations are bles. Although variables such as income over
currently equipped with some type of com- feed cost and feed cost per kilogram of milk
puter system; 15 to 20% have some type of have been in the systems, historical systems
controlled feeding; and 10 to 20% have some have not responded to the need for a totally
type of herd monitoring software. During this integrated program. Production and financial
time, use of computer feeding programs gener- programs were not integrated. Recently, de-
ally decreased, and use of milk metering velopments by DeLorenzo et al. (5) provided
equipment increased. More than 50% of new an economic analysis to maximize the total
Journal of Dairy Science Vol. 76, No. IO. 1993
SYMPOSIUM: MANAGEMENT IN THE 21ST CENTURY 3 193
cash output of the drury operation. Their pro- CONCLUSIONS
gram provides historical information and a
forecast and goal. The presentation also The information age offers new technolo-
graphically supplies details to help the gies and opportunities for the dairy industry.
producers make decisions on optimal breeding, Change is always challenging; however, data
culling, and replacement. Receipts are com- collection and proactive management will pro-
pared with those fiom the previous 12 mo. vide great benefits. As microcomputers be-
Past receipts provide a running total that come more heavily used in dairy operations,
changes each month. When the forecast is the amount of data collected will also increase.
developed, the program takes into account the The challenge is in management of the in-
farm's most recent DHI test day and projects creased data. The industry, the technologies,
what the cash flow will be in the next 12 mo. and the management are changing. Extension
The forecast considers the seasonal milk price, fact sheets do not provide sufficient manage-
milk flow at the dairy, and the herd's ment guidelines. Herd-specific problems must
reproductive performance. This information be managed. Programs that provide decision
provides a way for producers to illustrate their analysis are expensive to construct and to
cash flow trend and allows them to identify maintain. Development of the needed informa-
where and when profits will be made. The goal tion delivery systems to allow, large, well-
identifies the best way to manage that specific managed dairy operations to remain profitable
herd to make the most profit. in the competitive global economy is a major
Standardized Pet$ormance Analysis. Other challenge for the dairy industry.
livestock species have integrated financial and
production information (16). Standardized per- REFERENCES
formance analysis provides a method to stan-
1 Agricultural Statistics Board. 1992. Dairy situation
dardize production and economic parameters. and outlook report. DS-434. USDA E o n . Res. Serv.
The analysis can also demonstrate to the user a Commodity Bono. Div., Washington, DC.
set of production parameters that relate to past .
ZBoeckman, S , and K. R. Carlson. 1991. Ml and ik
events in that herd. Dairy Beef Residue Prevention Protocol. Agri-
Economics of Information Management. Education Inc., Stratmore, [A.
Data are only useful when they provide usable 3Cannon. T. J., M. A. Tomaszewski, and R. H.
Fourdraine. 1992. Using current test day analysis to
information. Data collection is costly and is evaluate changes in dairy herd management practices.
not easy to price. The value of data depends on J. Dairy Sci. 75(Suppl. 1):165.(Abstr.)
how they are used. 4Cropp, R. 1992. Dairy outlook. Page 9 in American
Data can be categorized (29) as those that Farm Bureau Federation 1991 Dairy Summary. Am.
fulfill some requirement or constraint imposed Farm Bur. Fed., Park Ridge, IL.
5DtLorenz0, M.A.. T. H. Spreen, G. R. Bryan, D. K.
from outside the business or those that in- Beede. and J.A.M. Van Arendonk. 1992. Optimizing
fluence a decision or decisions within the busi- model: insemination, replacement, seasonal produc-
ness. Through data evaluation, the user can tion and cash flow. J. Dairy Sci. 75885.
choose between alternatives, those alternatives 6 Engelbw, F.W.G.A.. and J. Doeksen. 1990. Develop-
can be identified, and their outcome can be ments to standardize the use of information in Dutch
daiq farming. Page 158 in h 3rd Int. Congr.
evaluated. Because decisions deal with future Computer Technol., Deutsche Landwirtschafts-
outcomes, some degree of uncertainty is al- Gescllschaft e. V., Frankhut an Main, Bad Soden,
ways involved. GelIlWlY.
As dairies continue to increase in size, 7Femando, R. S. 1983. Diagnosis of intramammary
managers will spend more time on evaluation ik
infection and the bacterial conductivity of m l .Ph.D.
of records. Farmers and managers spent almost Diss., Univ. IUiois. Urbana.
8 Fernando. R. S.. and S. Spahr. 1983. Effects of m l -
half (45.3%) of their time on activities classi- ing interval on selected milk constituents from normal
fied as farm business management (29), onbut, and infected quarters. J. Dairy Sci. 66:1155.
large farms, proportionately less time was 9 Fernando, R. S.,S.L. Spahr, and E. H. Jaster. 1985.
spent on physical work. More time was spent i k ih
Comparison of electrical conductivity of m l w t
on management decision-making. As drury other indirect methods for detection of subclinical
mastitis. J. Dairy Sci. 68:449.
production units continue to increase in size, 10Gilmore. J. A., T. J. Cannon, and M. A. Tomaszewski.
the time spent on information analysis and 1991. Using annual, split annual and current sample
decision-making will increase. day herd lactation curves to evaluate dairy herd
Journal of Dairy Science Vol. 76, No. IO. 1993
management. J. Dairy Sci. 7qSuppl. 1):235.(Abstr.) 21 Rossing. W.. and B. Ipema. 1990. From cow identifi-
11 Hansen. T., J. Clay, and K. Butcher. 1989. State cation to fully automatic milking. Page 59 in Proc.
Management DART. Dairy Records Processing Ccn- 20th Int. Stockmen’s School. Int. Stockmen’s Wuc.
ter, Raleigh, NC. Found., Houston, TX.
12 Isaksson, A., A. Philip. E. Goransson, and H. Bjor- 22Rossing. W., A. H. Ipema, and K. Maatje. 1983.
kenfeld. 1987. The electrical conductivity of bovine Actrons for measuring activity of dairy cows. Page
milk in mastitis diagnosis. Acta Vet. Scand. 28:456. 127 in Roc.Symp. Automation Dairying. Inst. Agric.
13 Kiddy, C. A. 1977. Variation in physical activity as an Eng., Wageningen, Neth.
indication of estrus in dairy cows. J. Dairy Sci. 69: 23 Spahr, S. L. 1989. Recent advances in automatic
235. acquisition and analysis of individual cow data. Page
14 Maatje, K.,and W. Rossing. 1976. Detecting estrus by 81 in Roc. Mtg. Natl. Mastitis Counc., Natl. Mastitis
measuring mik temperature of dairy cows during Counc., Arlington, VA.
milking. Livest. Prod. Sci. 3% 24te Braake, M., W. Chang, and L. R. Jones. 1992. A
15 Maatje, K. F., F. Wiersma, and W. Rossing. 1987. model for estimating herd productivity using lactation
Measuring milk temperature during milking and ac- curve and nproductive parameters. J. Dairy Sci.
tivity of the dairy cow for detecting suck cows and 75(Suppl. 1):22O.(Abse.)
cows in estrus. Page 176 in Roc. 3rd Automation in 25Tomaszcwski, M. A. 1990. Dairy decision support
Dairying. Inst. Agric. Eng., Wageningen, Neth. system survey of processing centers. Mimeo. Texas
16 McGrann, J. 1992. SPA-A standardized performance Agric. Ext. Serv., College Station.
analysis for the cow-calf producer. Mimeo, Natl. Cat- 26 United States Bureau of the Census. 1968. Census of
tleman’s Assoc., Denver, CO. Agriculture. Vol. 2. General Report Statistics by Sub-
ject. US Gov’t. Printing Office, Washington, DC.
17 Nielen, M.. H. Deluyker, Y. H. Schukken, and A.
27 Voelker. D. E. 1985. History of dairy record keeping.
Brand. 1992. Electrical conductivity of milk: measure-
National Cooperative Dairy Herd Improvement Hand-
ment. modifiers, and meta analysis of mastitis detec- book. A-2 Natl. DHI Assoc., Columbus, OH.
tion performance. J. Dairy Sci. 75:606. 28 Walton, J. S., and G. J. King. 1986. Indicators of
18Office of Technology Assessment. 1991. US Dairy estrus in Holstein cows housed in tie stalls. J. Dairy
Industry at a C o s o d Biotechnology and Policy Sci. 69:2966.
Choices. OTA-F470. US Congr., Office Technol. As- 29 Webster, J.P.G. 1990. Reflections on the economics of
sessment, US Gov’t Printing office, Washington, DC. decision support systems. Page 307 in Proc. 3rd int.
19 Orth, R. 1986. Searching for S in your DHIA records. Congr. Comput. Technol., Deutsche Landwirtschafts-
Illinois-Iowa Dairy Handbook. Mimeo, Illinois Coop. Gesellschaft e. V.. Frankfurt an Main, Bad soden,
Ext. Serv.. Urbana, I .L GClllUIlY.
20 Rossing, W., and B. Ipema. 1989. From cow identifi- 30 Wiggans, G. R., and P. M. VanRaden. 1989. USDA-
cation to fully automatic milking. Page 68 in Roc. DHIA animal model genetic evaluations. National
28th Mtg. Natl. Msii Counc., Natl. Mastitis Cooperative Dairy Herd Improvement Handbook.H-2.
Counc., Arlington, VA. Natl. DHI Assoc., Columbus, OH.
Journal of Dairy Science Vol. 76, No. 10, 1993