Modern computers combine the capacity not only to carry out calculations very quickly but also to make many calculations simultaneously, to remember what they have done, as well as their instructions, and to do all this using very little space and with great reliability. These abilities to calculate, record and remember far exceed those of people and, where these qualities are all that are required, computers can substitute for people — very often for many people. But, of course, agriculture as a whole does not employ many people and the scope for substitution may not be great: it is, however, likely to be greater in animal rather than in crop production.

Computers can only make decisions where adequate data are available (Blackie and Dent, 1979) and where the weight to be attached to each piece of information is pre-determined. The conceptual framework also has to be adequate, so that decisions are also pre-determined for given values of the variables considered. This is not the same thing as exercising judgement but, in fact, it can come very close to it. A computer landing an aircraft, playing chess or diagnosing a disease behaves very like a human being, and a very skilful, experienced one at that. Computers have considerable advantages in terms of speed of response to feedback and thus, in addition to the power of decision, they possess great power to control and monitor operations.

In looking to the future, therefore, it is probably not necessary to try and predict what computers will be able to do: they are already able to do most things and will be able to do virtually anything. It does not follow, of course, that they should be used just because it is technically possible: mere technical feasibility is not sufficient justification for doing anything. This is true of all technical developments and computers are no different in two important respects.

At the present time, with the high reliability and performance of computer hardware, computer systems applied in any field must be judged more by the quality of the software provided. Thus it is highly relevant in an investigation of the use of computers in a field such as animal production, to concentrate on aspects of the software.

Software provides the computer with the ability to obey instructions and to do as the user wishes. However, before arriving at these ‘machine instructions’ a number of steps have to be covered. First, it is essential to design the software — that is, to establish the requirements to be achieved on the computer. This design stage is followed by the implementation phase, in which the requirements as stated in English are transformed into such instructions as the machine can read and obey. The final phase is testing, in which it must be determined whether the requirements have been met, and to modify the design and iterate until the performance is satisfactory.

Software in general can be divided into three classes — systems, utility and applications software. The systems software drives the machine and its associated peripherals such as a VDU and printer. The systems software also includes a file system for organization of the data on the relevant storage media (floppy disks, cartridges, magnetic tape). Also considered part of the systems software are the assembler, interpreters and compilers for high level languages such as BASIC and FORTRAN and for programming aids such as DEBUGGERS. Systems software is normally supplied with the computer and needs to be evaluated along with the hardware by any prospective purchaser of a computer system.

The achievement of the microelectronics industry in integrating the components of a computer on to a few, or even a single, increasingly complex silicon chip has resulted in a reduction in price of the basic elements of a computer system by more than one hundred times. This technology has brought the cheap calculator which can now provide quite sophisticated programming functions that could only have been obtained with a computer occupying an equipment rack some few yean ago. The microprocessor is the most complex of the family of integrated circuits but its operation is not fundamentally different from that of its much larger predecessors, being a feat of technology rather than a new departure in computer design. It is the dramatic change in price and physical size, coupled with other associated technical advances such as low power consumption and reliability, that has made it possible for the use of computers to be considered in situations which could not reasonably be contemplated previously.

Although computers include those that operate or continuous, or analogue, variables, the predominanl interest is in digital systems. These operate on discrete values of electrical signals that represent, in coded forms, data or the operations to be performed. These operations can be arithmetical or logical and may be combined into a sequence that constitutes a program. The notion of stored program control, where such a sequence is held in the memory of the machine, provides the essential feature of computer operation.

Computers or computing devices are in use, and will be in use in the future to a greater extent, in the interest of animal production. Data collection, in one form or another, is fundamental to that usage. The ready availability of low cost digital processors, data converters and signal amplifiers, where appropriate, has ensured that this is so. The only constraint is with the difficulty of measurement and, in signal recording, whether suitable instrumentation devices or transducers exist or can be devised.

The important considerations are why the data are needed, how many data, when and in what form they are required.

From the answer to these questions will come as many diverse forms of data recording — albeit based on established techniques — as there are requirements to record, all of which may be realized using the same basic hardware components by virtue of the flexibility introduced by the need to program the processing element. In the manner of pure research, the answers to these questions constitute the design of the experiment, which too easily can be ill-considered, resulting in quantities of data useless to its original intent; where automation is considered, they constitute the design of the digital system required.

The range of requirements for data recording extends from manual keyboard entry to small computers engaged in farm management, through the systems designed to gather data automatically for the management of animal enterprises, to the specialized requirements for data to substantiate mathematical models of particular physiological processes in the animal. Within this range, the variation of requirement is fundamentally that of its dependancy upon the dimension of time.

The Unit's work on herd monitoring was motivated by the need for a system to record data for a Study of Methods of Improving Oestrus Detection (Esslemont, 1973). These interests were shared with Dr R. S. Morris, then at Melbourne University, who made available a suite of computer programs for analyzing dairy herd health and fertility data which suited the immediate needs (Morris, Blood, Cannon, Cannon and Williamson, 1978). Out of this generous co-operation has evolved the ‘MELBREAD’ Herd Health and Fertility Reporting Scheme, the more comprehensive ‘DANDAIR’ system, the ‘YOUNGSTOCK’ system for recording heifers, and now ‘DAISY’ with which data processing is being decentralized to farms and veterinary practices. We are also starting to develop similar systems for intensive pig enterprises.

This paper presents background information on the development of herd monitoring and a description of computerized record systems for the dairy herd developed by VEERU, especially the ‘DAISY’ software package for small computer systems.

An estimate of the breakdown of fixed and variable costs on a per cow basis for a UK dairy herd is shown in Table 2.2.1a and typical good performance levels are shown in Table 2.2.1b. These figures show the need for very high technical performance in order to produce truly economic results. They also show that costs per cow for improved management information systems (at £5/cow/year) can be kept below 1% of turnover.

Farm management is the meeting-point of science, economics, man-management, marketing, planning and decision-making on the farm. From such a diverse background and because of the variable nature of farming itself, the supply of management information to the farmer is a subject which merits considerable attention.

To identify the importance of management, it is worth considering the range in performance between farms, as measured by dairy herd gross margin per hectare. The average for 1979/80 of Farm Management Services (FMS) fully costed herds (MMB, 1980) was £546/ha, while the result for the ‘top 25%” was £809/ha, a difference of £263/ha. It is exceedingly difficult to define precisely the part which management plays in this difference. However, through its effect on feeding, breeding, grassland management, fertility and health, it is probably the dominant factor. If the range in FMS results were to be taken as the normal distribution for the national herd, one could postulate that bringing up the national average gross margin to that of the “top 25%” would add a further £127m to the national dairy herd gross margin. This would be equivalent to slightly in excess of £4 000 in the average 60 cow herd. Whilst admitting that not every farmer nor every farm can achieve the level of performance of the “top 25%”, such calculations do indicate the extent over which management has an influence.

Increasing importance is being put on quality control of livestock both from performance and marketing aspects. The breeding of superior stock relies heavily on progeny testing, first in small numbers and later in larger scale field trials. With commercial stock, emphasis is placed on the food conversion efficiency; health of the individual, flock or herd; the market requirement and the likely date of attaining market readiness. All these factors rely heavily on close monitoring of the performance of the stock. However, with increasing pressures for shorter working hours and greater responsibilities per employee, close monitoring is likely to diminish rather than increase, unless modern technology comes to the aid of the producer. Already, electronics have been making inroads in this area. The advantages of using electronics for animal weighing were first demonstrated in the early 70's (Smith and Turner, 1974). Interest has been slow to build up but now most farmers and manufacturers recognize the potential. However, despite using electronic weight indication, the process of animal weighing remains a manual operation involving at least one man, more often two or three. Despite advances in animal handling procedures, in most manual weighing exercises there is always a risk of injury to both stock and men. A system of weighing which removes this risk and reduces stress on the animal would be welcomed by the producer. Electronic aids for milk yield recording and egg production have also been under development for some years (Burgess, 1980; Anon., 1980) but will not be dealt with here. This paper will deal solely with the prospects for fully automatic weight recording of live animals and describe some of the results of work conducted by the National Institute of Agricultural Engineering at Silsoe.

Computers appropriate for small-to-medium sized businesses (such as large farms, agri-businesses, and management and production consultants) can be broadly classified into three groups: pocket sized programmable calculators (£50-£500); micro-computers (£500-£5000); mini-computers (£5000-£50000). There are also the main-frame computers which take up considerably more resources of both money and space. Main-frame computers are likely to be restricted for use in specialist functions such as the handling of very large programs, complex statistical packages and so on.

Of these, the simplest type to operate allows specific programs to be written onto magnetic cards which are fed in as required. Such programs may calculate: nutrient requirements for given levels of production; nutritional quality of a combination of ingredients in a farm mix; average herd performances to date; a register of input and output for the unit; and the like. The capacity of these machines is limited, both as to the extent and complexity of the calculations allowed and to the volume of data which can be stored and executed. The alphabetic printing and layout capability is rather poor and the user must be familiar with the input format and with the configuration for the output.

The number of computer “packages” now being offered to farmers is growing rapidly and this paper does not present an exhaustive or critical account of these. Rather, it concentrates on packages offered by the feed trade to the intensive animal production sector of British agriculture. The paper will deal with the cattle, pig and poultry sectors, in that order, and then examine computer applications as they have developed, are now, and how they may become. Finally, the important criteria when considering the potential success of new computer packages will be described.

Though some may consider the cattle sector to be less progressive than the pig and poultry sectors, as far as the application of the computer is concerned, cattle led the way. The earliest application was in the late 50's and early 60's, when the feed trade abandoned hand formulation of cattle feeds and provided one of the first commercial applications of linear programming—least-cost feed formulation. Though in those days, diet specifications were heavily constrained by raw material limits (for example, 5% fixed linseed cake and 5% copra cake in a well known high performance dairy feed), the effects were to reduce formulation costs by about 5% (£1 per ton in those days!). Cattle feeds were soon universally computer formulated, closely followed by pig feeds. Broiler and turkey foods remained as fixed formulae for some time but, as nutritional knowledge increased, these too fell into line. Currently, only a few specialized diets, such as baby pig foods, remain as fixed formulae. The remaining products now have very few raw material constraints left but many are tightly constrained on amino acids, undegradable protein, fatty acid levels and rations, various descriptions of energy components, etc.

For the purpose of this paper ‘hill resources’ will be interpreted to mean the land resources of the hills and uplands. ‘Management’ will be assumed to include strategic and tactical decision making. It will further be assumed that the invitation is to discuss computers in animal production in relation to farming practice and not to research aimed at improving farming practice. In fact, computers will hardly be discussed at all. Computers are tools in this context and are aids to decision making. It seems much more likely that they will be more effectively used if we can be clear about decision making and its nature in relation to current knowledge and understanding and to the needs of farming systems based on the land resources of the hills and uplands.

The official statistical deposition of hill sheep farms in Scotland includes reference to land resources; these farms have more than 90% of the land in rough grazings and less than 15% in crop. The definition includes reference to the allocation of man-days among sheep and cattle. The Scottish upland farm description also includes reference to land resources and there are, for example, constraints in the relationship of crop and rough grazing areas. Hill and upland farms in England and Wales come into the category known as Livestock Rearing farms, where distinction between hill sheep and upland farms is made on the distribution of man-days among cattle and sheep.

Since the pioneer work of the Cambridge group into the physiology and biochemistry of ruminant digestion (Barcroft, McAnally and Phillipson, 1944), research has firmly established that the processes of digestion, synthesis and passage of nutrients within, and absorption of the end products from, the ruminant alimentary tract are extremely complex, highly interrelated and considerably influenced by the nature and quantity of the diet consumed (Baldwin and Koong, 1980). Consequently, the nutrients absorbed from the digestive tract have only limited quantitative and qualitative similarity to those consumed, mainly due to the processes of anaerobic fermentation and microbial synthesis occurring within the reticulo-rumen. The development of techniques to measure nutrient flow and microbial synthesis within the ruminant alimentary tract (Beever, Harrison, Thomson, Cammell and Osbourn, 1974; MacRae and Armstrong, 1969a; Smith and McAllan, 1970), has led to numerous studies on nutrient digestion and flow, mainly with sheep fed forage and cereal based diets (Beever, Thomson and Cammell, 1976; Beever, Thomson, Cammell and Harrison, 1977; Beever, Terry, Cammell and Wallace, 1978; Coehlo da Silva, Seeley, Beever, Prescott and Armstrong, 1972; Coehlo da Silva, Seeley, Thomson, Beever and Armstrong, 1972; Egan, Walker, Nader and Ulyatt, 1975; Faichney and Weston, 1971; Faichney and White, 1977; Harrison, Beever, Thomson and Osbourn, 1973, 1975; Hogan and Weston, 1970; MacRae and Armstrong, 1969b; MacRae and Ulyatt, 1974; Siddons, Evans and Beever, 1979; Ulyatt and MacRae, 1974) and, from some of these studies, empirical relationships have been established (Beever et al., 1976; Chalupa. 1975; Ulyatt and Egan, 1979; Weston and Hogan, 1973) in an attempt to predict, from a knowledge of feed input and composition, the likely end products of digestion. However, considering the complexities of ruminant digestion, the limited applicability of such relationships was not surprising and the need for a more fundamental consideration and integration of the processes of digestion and synthesis was established. To this end, the work of Baldwin and his colleagues (Baldwin, Lucas and Cabrera, 1970; Baldwin, Koong and Ulyatt, 1977; Reichl and Baldwin, 1975) stimulated interest in the area of rumen modelling.

At the Hill Farming Research Organization (HFRO), we have been modelling various components of sheep systems and we have also attempted to model, where a need was seen, whole sheep systems at different levels of organization. The sheep systems involved are of course pastoral grazing systems on hill or upland farms. The purpose of this paper is to briefly describe these models, to identify reasons for their existence and to outline some of the problems encountered in their construction.

Models have been constructed at HFRO to satisfy particular needs. They cover a spectrum of interests. There are those which are concerned with resource allocation and economic evaluation, those which are built in an attempt to develop more precise operational management procedures and those which describe biological mechanisms and processes which may be conceptual but which provide a better understanding of particular components of an animal production system.

The first model (Maxwell, Eadie and Sibbald, 1973) was built to provide answers to the economic questions “Is a particular level of investment in a hill sheep system a viable proposition in both the short and the long term?” and “How does the economic performance of this strategy compare with those of alternatives?”, alternatives being, for example, different levels and rates of flock build-up, with variations in the timing of investments. The model makes year-by-year calculations of cash flow and balance based on Gross Margin, taking account of capital invested, changes in variable costs, tax and interest rates. The Gross Margin is determined from price and cost data and from biological performance. The model also calculates overall economic performance, for a pre-selected project length, as Net Present Value and Internal Rate of Return, taking account of increased stock valuation. These results are used to compare alternative schemes with different rates of investment.

Whittemore and Fawcett (1976) and Whittemore (1976) showed how ideas and data from different disciplines could be brought together to simulate, using a computer program, the growth of the pig on controlled feeding. Controlled feeding, as shown in Figure 3.3.1, is one of three types of feeding system (Emmans, 1981). Parks (1970) dealt with systems of the second type, in which animals are given free access to single feed. In such a system, the composition of the diet, but not its amount, is known in advance. Parks (1970) suggested that the rate of feed intake in growing animals increased at a diminishing rate with age towards an asymptotic value; Figure 3.3.2. This function has two parameters — the asymptote and a time constant. The values of these were seen as functions of feed composition, genotype and environment. He went on to predict growth as a function of feed intake and a conversion efficiency. While a description of how ad libitum feed intake changes with age is useful, it is not as powerful as an understanding of what governs feed intake would be. While there are many theories about the mechanisms that animals use to control their feed intake at a given level, there is none that can be used to predict what this level will be. One is proposed in this paper.

An algebraic description of the lactation curve is a useful component of any model of the day-to-day production of lactating animals. Observation and common sense suggest that such a function should rise to a peak early in lactation and decline thereafter but simpler models have been used. Ostergaard (1979), for example, used a linear model to study strategies for concentrate feeding to obtain optimum feeding levels in high yielding dairy cows. His model was

where y (t) was the yield in week t and a and b were the usual linear regression coefficients. The error term is omitted here and elsewhere for clarity. Gaines (1927) used the decay function

where A and k are the constants fitted to log y (t).

These models, one linear, the other exponential, peak in Week 1 and require only two parameters. In this paper, more sophisticated functions are described and compared.

A lactation curve is assumed to contain two components, one of which is the intrinsic biological drive to produce milk and the other is an environmental constraint upon it. The algebra may be justified by biological argument according to the skill and the leanings of the modeller.

There is a belief among research workers that their experiments have two possible outcomes, either they are successful or else the results will have to be taken to a statistician. Although the belief is not universal, it does account for a certain amount of brooding by statisticians on their place in scientific research. This may be why I have made the title I was given, the excuse for examining some of my own activities and attitudes in bringing agricultural research data into juxtaposition with mathematical models. I apologize in advance for the egocentricity.

Statisticians in agricultural research find considerable employment in enabling their customers to adorn papers with a sufficiency of standard errors and significance tests for the satisfaction of editorial boards. Underlying this apparently cosmetic activity is the important function of testing the logic of experimental conclusions and preventing their too easy or too general acceptance. The statistical approach to model building is therefore that of a strength tester. Theoretical assumptions may be vital for the construction of the model but statistical appraisal must be empirical, testing the correspondence between the model and these aspects of experience which it is intended to describe. On the basis of these tests, some parts or some uses of the model may be rejected as unsound and other parts or uses may be accepted, with varying degrees of confidence, as being reliable. Warning notices against undue dependence may be posted on these parts which cannot be tested for lack of data.

This paper concerns a look at the future use of computers in the farm office, particularly in relation to animal production. It would perhaps be appropriate to start by examining the current position and look briefly at the types of farms involved, the possible computer applications and the computer systems currently available. On the basis of what is known about future developments in hardware and software, a view of the future use of computers in animal production in the farm office will be attempted.

The information largely comes from the ADAS On Farm Computing Team, which was set up in April 1979 to examine the implications of microcomputers in the farm office. The team has made contact with a large cross section of the industry. It has been in touch with farmers who own microcomputers, firms actively selling machines and programs, manufacturers, software houses, bureaux, consultants, accountants, secretarial agencies and others offering alternative services. The team has also had several machines and programs under evaluation.

Our aim has been to brief the ADAS field advisers on the new technology, the possible applications, advantages, disadvantages and the criteria to be considered when a farmer is faced with the decision of whether or not to buy a microcomputer.

The remit of this paper is to discuss “…provision of appropriate information to write suitable software for the animal production industry” and … how much material has actually flowed from Universities and Research Institutes into useable computer programs and what are seen as being the main stumbling blocks to the rate of flow and effectiveness of take-up?”. Canvassing opinion from several people, including some of those who have contributed at this Meeting, gave many useful ideas but some of them would not wish these ideas to be associated with their names, for a variety of reasons. Therefore, the author is reluctantly unable to thank those who helped by name but is very pleased to do so collectively. The remit has also been broadened slightly to include the feedback from the use of models in the field to the research worker and uses examples of the author's own attempts to model voluntary food intake in ruminants to illustrate some shortcomings.

Three obvious statements but well worth repeating: “Computers are only an aid to summarizing information for use in assisting management decisions”; “Records are only of any use if they are used — they are not an end in themselves”; and “No point having sophisticated computerized management aids if the basic information is not available and if only decisions at a simple level are to be made”.

Amongst dairy and pig producers, the need for keeping records is well established and probably over 40% of all dairy cows and pigs in Britain are involved in an enterprise recording scheme. By contrast, all too few beef and sheep producers currently keep any detailed computerized records in an organized way. It is estimated that only about 2% of all commercial beef and sheep are involved in an enterprise recording scheme. Good basic records are an essential ingredient to any more sophisticated management aid because these need to be based on relevant reliable data for individual situations. The need to use computers and their value on the individual farm is dependent upon having sufficient information to make it worthwhile employing them. The sophistication of management aids required for enterprises will be determined by the level of recording already involved. It is obvious that the requirements for beef and sheep producers are at a much more basic level than for many pig and dairy producers.

In addition to data and record handling and the completion of complex and arduous mathematical calculations, the computer may serve a wide range of purposes for the agricultural industry; these include the provision of day-to-day management information, the mechanical enactment of management decisions, business forecasting, interpretation of real life and prediction of future response by use of simulation models, analysis of cost effectiveness of various tactics and stratagems, the transfer of information, scrutinization of existing knowledge and the formulation of experimental programmes. The computer is seen as a major linking medium between research, development and production practice; being both the preferred route for information flow and an ideal way of packaging dispersed pieces of knowledge into practical, usable, systems advice.

The concern of practical producers is not with discrete little problems but with systems. To help, the extension worker must bring forward systems solutions. Often research and development workers try to get across to producers potential benefits in small bits (3 times daily milking gives a yield lift of 15%; flat rate feeding gives better margins over concentrates; high density diets improve feed efficiency), whereas producer benefits come from the cost effectiveness of whole integrated systems.