Addison, KB; Rickert, KG; Cooksley, DG; Robbins, GB; Bushell, JJ; McKeon, GM; Ebersohn, JP
Abstract:
Animal Production in Australia Vol. I5 MODELS OF FEED SYSTEMS FOR GROWING CATTLE IN SUB-COASTAL SOUTHERN QUEENSLAND INTRODUCTION K.B. ADDISON* and K.G. RICKERT* Traditional grazing experiments compare different forages or management options, often on a small scale. Results are usually expressed as production per animal or per unit area and it is left to producers or extension staff to apply the results commercially. Introducing new technology, such as improved pastures, into a property may cause several problems: technology has to be integrated with other forages or property operations, i) often on a much larger scale of operation than the original experiment in which the technology was devised; ii) technology has to contribute to a specific marketable product desired by the producer, which, was probably not an aim of the original experiment; and iii) the long term reliability and profitability of the technology needs to be assessed. Attempts were made to address these problems in the southern spear grass region of Queensland. Technology was available that suggested the traditional system of production could be modified to produce more and younger sale cattle. Traditionally, steers on native pasture reach 500 kg LW when 3.5 to 4.5 years old. This weight should be approached at 2 to 2.5 years in two new production systems which were derived from past results. The systems were demonstrated on a relatively large scale on 'Brian Pastures' Research Station, Gayndah as physical models of improved production systems. In this contract, we note the performance from these demonstrations (i.e. physical models) of new technology. A is also described and used to illustrate some implications and native pasture on a property. Finally, we examine the computer models in enhancing the credibility of innovations ional research. two large-scale field simple computer model of integrating improved roles for physical and arising from tradit- A PHYSICAL MODEL OF BEEF CATTLE PRODUCTION USING INPUTS OF NATIVE PASTURE AND LEUCAENA LEUCOCEPHALA (LEUCAENA) D.G. COOKSLEY* Cattle usually lose weight during the cool months of the year on sub-coastal native pastures of southern Queensland (Alexander and Beattie 1968). These pastures are dominated by spear grass (Heteropogon contortus) and forest blue grass (Bothriochloa bladhii). Grazing trials at 'Brian Pastures' have shown that weight losses on native pasture may be avoided by supplementing cattle during the cool season with an area of leucaena (Leucaena leucocephala cv. Peru) which was left ungrazed during the warm season (Foster and Blight 1982, Addison et al. 1984a). Results varied between 25 to 85 kg/head advantage. These encouraging results, together with results from other supplementary feeding trials (Addison 1970, Addison et al. 1984a) and a stocking rate trial on native pasture (Cooksley, unpublished data) gave rise to a production system using leucaena strategically to supplement animals for two years after weaning. The aim was to produce a liveweight gain of 140 kg/hd/yr, 60 kg in winter/spring plus 80 kg in summer/autumn. Thus, after weaning at 6 months and 180 kg/hd, the target liveweight at 2.5 years was 460 kg/hd. Overall, a liveweight gain of 78 kg/ha/yr Qld. Dept. of Primary Industries, Brian Pastures Research Station, Gayndah, 4625 11 Animal Production in Australia Vol. I5 was expected which was approximately twice the productivity of unimproved pasture. METHOD Each year, after weaning in late May, 52 crossbred first and second generation cattle (3/8 Sahiwal, 5/8 Hereford) entered two paddocks each of which contained 1 ha/hdof native pasture plus 0.2 ha/hd of leucaena. In their second year, animals used double the area of both forages in another two paddocks. In addition to native pasture, five plots of leucaena, sown in rows 3 m apart, in each paddock were grazed during winter in a rotational system of 3 weeks on and 12 weeks off. A more rapid rotation was used in spring to graze the leucaena regrowth. The leucaena was not grazed during summer and autumn. The system commenced in 1978 and four grazing cycles have been completed. Yield, composition, and quality of Animal liveweights were recorded every 6 when the cattle were slaughtered after 2 used in the first two cycles but kept in heifers and steers were run together. all pastures were recorded regularly. weeks and some carcase data was collected years. Steers and unspeyed heifers were separate paddocks. Subsequently, speyed RESULTS AND DISCUSSION This system illustrated the benefits of applying technology from experimental grazing trials on a large scale before recommending its commercial adoption. Several problems became evident. The target liveweight gain of 280 kgfhd was not achieved (Table 1). Heifers averaged a LWG of 231 kg/hd while steers gained 267 kg/hd. Heifers were better finished than steers with 8 mm of fat over the 12/13th rib compared with 5 mm in steers. The application of leucaena seems to depend on the class of animal to be turned off. TABLE 1 Liveweight changes in the production system (kg/hd) Liveweight gain in summer/autumn usually exceeded expectations. Hereford steers were used in the original studies (on which the svstem was based) whereas crossbred cattle. which grow faster than Herefords (Robbins and Esdale 1982), were used in this study. Liveweight gain in winter/spring was variable and below expectation particularly in yearling cattle (Table 1). This response probably reflects the amount of supplementation from leucaena. Mean yield of edible leucaena was 954 kg/ha, much less than the average (1972 to 1974) of 3700 kg/ha reported bv Addison et al. (1984b). This difference could be partly explained by variation in rainfall with higher summer rainfall during the early 1970's. 12 Animal Production in Australia Vol. IS A large amount (43%) of leucaena (mainly leaves) which grew during the warmer months, dried and fell to the ground before grazing in June. Addison et al. (1984b) measured yields of leucaena in February and thus avoided some of this loss. This high amount of leaf fall sugrzests that leucaena could be grazed from March to November and spelled in summer for more effective use. There was considerable variation in leucaena yields from year to year and paddock to paddock (average coefficients of variation of 55% and 46%, respectiveBetween-year variation probably reflects the amount and distribution of soil water during the growing season (Cooksley, unpublished data). Supporting studies have shown that the between-paddock variation is not due to nutrient deficiencies and other possible causes are being investigated. No toxic effects have been observed from leucaena as were recorded by Jones et al. (1976). l y ) l The physical model has shown that the initial expectations of animal production have not been fully realised,particularly with heifers. The data collected in monitoring the system has allowed examination of weaknesses with the rainfall-related variation in leucaena yield emerging as one problem. Computer modelling would appear to have a role in simulating such variation through water balance studies and would allow alternative systems of leucaena use to be designed before further field experimentation. A PHYSICAL MODEL OF BEEF CATTLE PRODUCTION us ING INPUTS OF NAT IVE PASTURE, LEUCAENA AN D CROP S G.B. ROBBINS* and J.J. BUSHELL* BACKGROUND A relatively intensive system of beef cattle production was devised to graze the various types of forage available to cattle in sub-coastal southern Queensland. Areas of sown and native pastures, leucaena, and crops were used at different times of the year with the final objective being to produce 450 kg LW steers at less than two years of age. Achieving such an objective in this environment requires faster growthrates than those normally achieved on native pastures, and so requires a high level of nutrition year-round. The approach adopted in combining the various forages was based on the results of several earlier experiments. Shaw and Bisset (1955) suggested that an alternative feed to native pasture was needed to improve growth rates of cattle during winter and spring. Young et al. (1959) and Young and Daly (1967) replaced native pasture with sown grasses and recorded greater liveweight gains from cattle grazing green panic (Panicum maximum var. trichoglume cv. Petrie) compared with native pasture during winter and spring, but recorded similar gains during summer and autumn from both pasture types. Young et al. (1959) suggested that sown pasture be grazed in winter and spring and native pasture in summer and autumn while Scattini (1983) tested different management options for cool-season grazing of green panic. Native pasture quality usually declines in autumn (Addison 1970) indicating that liveweight gains might be improved by autumn supplementation with a pasture legume such as leucaena (Addison et al. 1984b). Different rations are available for fattening steers in this environment but Hendricksen and Myles (1980) indicated the reliability and success of using mixtures of chaffed Lab lab purpureus cv. Rongai (lab lab) and milled sorghum grain. Work such as the above provided the basis for the feed system chosen and a physical model was developed to test the production system. II___-. - * Qld. Dept. of .Primary Industries, Brian Pastures Research Station, Gayndah 4625. 13 Animal Production in Australia Vol. I5 BETHODS Fifty hectares of arable clay soil was subdivided into 10 x 5 ha paddocks used in a lo-course ley rotation of 5 years of nitrogen-fertilized green panic and 5 years of alternating summer crops of lab lab and grain sorghum. Each year, there were five green panic pastures varying in age from one to five years and five paddocks which had been cropped for one to five years. Green panic was grazed by 60 weaner first or second generation 318 Sahiwal 518 Hereford steers during winter and spring at 2.5 weaners/ha. From December to mid-April, the steers grazed 60 ha native pasture (Heteropogon contortus and Bothriochloa bladhii dominant) which contained 6 ha of leucaena freely accessible to grazing during early autumn. After harvesting sorghum for grain and lab lab for hay, the crop residues were grazed late in autumn. The steers were fattened on lab lab chaff and milled sorghum grain fed in equal proportions for about 100 days. RESULTS AND DISCUSSION Steer liveweight gains on the various forages for the six cycles 1976177 to 1981/82 are given in Table 2. Mean final liveweight over the six cycles was 426 kg/hd, with a range of 384 to 453 kg/hd. Several problems were identified which strongly influenced weight gains. Rainfall variation over the six cycles was regarded as the main factor responsible. Production from green panic averaged 53 kg/hd and was usually much more variable in spring than in winter. Spring growth reflected rainfall at that time, whereas winter production from stand-over green panic was more reliable. Weight gains from five-year-old green panic were only half those from one-year-old pasture and this run-down in productivity was studied within the production system. Seasonal variation also had a marked effect on the sowing of crops and their subsequent yields. Lab lab hay yields varied from 1 to 7 t/ha and sorghum yielded from 1 to 5 t/ha of grain. TABLE 2 Cattle production using various forages from 1976/77 to 1981/82 Average weight gains from native pasture over summer were 0.5 kg/hd/d suggesting that an added sward legume could prove beneficial. Gains on crop residues were generally poor but feedlot fattening on chaff and grain was usually successful. Annual lab lab production exceeded requirements for the feedlot by an average of 17 t/y but an annual shortfall of 1.3 t/y in sorghum grain had to be made up from other sources. The system has also provided a useful framework for investigating other issues such as the effect of withholding food on -subsequent feedlot performance (Robbins et al. 1982), the merits of wet or dry 14 Animal Production in Australia Vol. 15 curfews prior to slaughter (Wythes et al. 1982), and the benefit of altering the proportion of grain in the feedlot ration to satisfy market premiums. The physical model showed that 60 head of cattle could be reliably turned off every year from 110 ha of land. Overall average production was 134 kg LWG/ ha/year which was approximately a three fold increase on the 30-50 kg LWG/ha/year that could be expected from native pasture. Furthermore the ley farming system in operation on the arable soil in this system is likely to be stable in the long term and could be particularly suited for marginal cropping land. A COMPUTER MODEL OF THE INTEGRATION OF FORAGE OPTIONS FOR BEEF PRODUCTION G.M. McKEON# and KG. RICKERTA Computer models of integrated forage systems havethe potential to avoid the outlay of time, space and resources required in physical models. They can consider a much greater range of management options (e.g. stocking rate, season of use, fertilizer, burning, conservation) and evaluate these alternatives against the variability of weather and prices. The model presented here examines the integration of native and sown pastures as tested in the physical models of Robbins and Bushel1 (above), Scattini (J983)andAddison (1970). The management options considered are stocking rate, season of use, proportion of sown pasture in the property and direct supplementation of animals grazing native pasture in winter and spring. Future model development will include economic evaluation and analysis of year-to-year variation in pasture and animal production. MODEL DESCRIPTION Our aim is to develop a model of a beef enterprise with the following specifications: 1. capable of considering many combinations of management alternatives; enough to extrapolate outside the bounds of experimental data; 2. robust 3. capable of considering year-to-year variation in plant production; and 4. suitable for use in micro-computers. In this paper, we present the basis of the biological model by considering a simple example involving the integration of only two options. The aim of the model is to predict annual liveweight gain of a weaner steer for a given grazing system. Seasonal (91 days) liveweight gain (LWG) was assumed to be related to dry matter intake (DMI). The equations of Siebert and Hunter (1977) relating intake and LWG to diet nitrogen for tropical pastures were reworked to give: (kghdd) LWG = 0.304 DM1 - 1.058; (kg/hd/d) (equation 1, Fig. la) Different management options (stocking rate, season of use) were modelled by describing their effects on dry matter intake. However, we have found (for these pastures) little relation between pasture on offer and LWG except at low pasture yields (500 kg/ha, Hendricksen et al. 1982). Hence, the usual pathway of modelling animal production by relating intake to pasture yield has not been followed. Instead we set a potential intake for each season, from equation 1 using observed LWG at low stocking rates. This potential reflects the overriding seasonal effect on diet quality. Potential intake was then reduced by * Qld. Dept. of Primary Industries, Brian Pas tures Res arch Station, Gayndah ,4625 /,, Qld. Dept. Primary Industries, P.O. Box 46, Brisbane Qld. 4001 15 Animal Production in Australia Vol. 15 an 'intake modifier' which was calculated from the proportion of accumulated growth which had been eaten (Fig. lb) i.e. proportion eaten since start of summer: proportion eaten = amount eaten since December 1 (equation 2) accumulated growth since December 1 (equation 3) intake modifier = a - b x proportion eaten The intake modifier is constrained between 0 and 1. Fig. la Relationship between intake and liveweight gain (LWG) Fig. lb Relationship between intake modifier and proportion of pasture eaten Because of the variation in seasonal production in tropical pastures in southern Queensland, pasture yields generally exceed feed requirements. This allows increased opportunity for diet selection between pasture components (e.g. leaf and stem) at low stocking rates. The 'intake modifier' attempts to describe the reduction in intake that is likely to occur due to reduced opportunity for selection with increased stocking rates or long durations of grazing e.g. year round. An analytical solution of equations (2) and (3) was .used to calculate the average intake restriction for a season. Sequence of calculations in the model: 1. 2. 3. 4, 5. 6. Calculate accumulated pasture growth and amount eaten since 1 December. Calculate potential seasonal intake (Table 3 and equation 1). Calculate intake modifier as a function of previous growth and grazing, and current season's growth, potential intake and stocking rate (equation 2 and 3) Calculate actual intake = intake modifier x potential intake. Calculate LWG from equation 1 and update animal live weight. Start again for next season. l Required inputs to formulate the model are given in Table 3: (1) seasonal pasture growth which is calculated by a pasture growth model (McKeon et al. 1982); (2) field measurements of seasonal liveweight gain at low stocking rates; and (3) effect of stocking rate on annual liveweight gain. Potential intake for each season was calculated through equation 1 using seasonal liveweight gain from Table 16 Animal Production in Australia Vol. 15 3. With these average inputs, equation 3 in the model was solved by fitting the model to the observed stocking rate response data for native pastures (from Brian Pastures Research Station) and for sown pastures (from Brigalow Research Station 200 km NW of Gayndah, Walker pers. comm). Solutions were: native pasture a=1.074 and b=0.486; sown pasture a=1.039 and b=0.437. TABLE 3 Input data for model development VALIDATION The model was tested against sixteen independent data sets from Brian Pastures where grazing was restricted to 6 month periods (Table 4). The model agreed well (residual standard deviation: 8 kg/hd) with observed data suggesting information gained from year round stocking rate relationships can be used to extrapolate to shorter grazing periods. The model correctly predicted the results of two other trials which have examined the effect of light grazing in summer/autumn on subsequent winter/spring performance. On native pasture, Addison (1970) measured a 5 kg/hd reduction in winter/spring due to summer grazing. For green panic,Scattini (pers. comm.) found a 9 kg/hd reduction in winter/spring LWG due to autumn grazing. In these two cases the model predicted a reduction of 6 and 8 kg/hd, respectively. SIMULATIONS A hypothetical property was considered with varying proportions of sown and native pasture. A maximum 'drought-safe' stocking rate was calculated assuming that native pasture area would carry 0.5 b/ha year round and sown pasture 1.0 b/ha. Cotton seed supplementation '(0.7 kg/d) on native pasture in winter and spring was simulated by increasing LWG by 9 kg/hd and 30 kg/hd respectively (Addison 1970). No effect on intake (substitution or intake increase) was considered. The various options were compared on a LWG per ha basis. Two techniques were used to conduct simulation experiments: (1) a factorial design with 4 seasons of use and 4 possible stocking rates on one option each season (0, 0.5, 1.0, 1.5 b/ha) giving 256 combinations; and (2) the model was linked to a 'hill climbing' optimization package to maximise LWG/ha. The former approach allows a better description of the overall response surface while the latter approach allows the continuum of stocking rate to be considered. However, in some cases the optimization approach gave slightly different solutions with different starting conditions. The factorial analysis indicated that a variety of combinations gave similar results although the general principle of integration was common to all optimal solutions. 17 Animal Production in Australia Vol. I5 TABLE 4 Comparison of observed and predicted liveweight gain for periods of seasonal grazing + Where required LWG for hereford or Gl crossbred cattle was converted to a G2 crossbred equivalent: ADG(G2) = 0.804 ADG(H) + ,135; ADG(G2) = 0.9 ADG(G1) kg/d * Fed a molasses urea supplement LWG (Pred) = 0.954 LWG (Obs) +0.772, r = 0.96, n = 17, s = 7.9 Advantages from integration reflect the degree in which the two forage options differed in seasonal production. The maximum advantage of sown pasture occurred in winter (30 kg) compared with 18 kg in autumn and spring and 15 kg in summer. As expected, the optimal solutions (Table 5) involved the use of sown pasture mainly in winter with variable use in autumn and spring. However, production from an 'optimal' integrated property did not greatly exceed (<lCkg/ha) the combined production from each pasture independently grazed year round at the safe stocking rates, i.e. non-integrated use (Fig. 2a). The impact of optimalintegrated systems compared to the non-integrated use was greatest on average LWG per head for properties where sown pasture makes up a small proportion of the property (Fig. 2b). Supplementation increased LWG on native pasture in spring above sown pasture LWG and hence the optimal production was achieved by grazing sown pastures in autumn and winter. The model showed that limited input data of pasture growth, animal production and stocking rate response can be used to simulate a wide range of combinations of stocking rate, and season of use, and hence provides a powerful tool for suggesting more refined physical models for testing. 18 Animal Production in Australia Vol. 15 Fig 2 TABLE 5 Seasonal stocking rates (beasts/ha) that give optimum integration for different proportions of native and sown pasture with overall animal numbers determined by the safe annual stocking rates (0.5 and 1.0 beasts/ha) for each pasture Native pasture Su Au Wi Sp Sown pasture Au Wi Sp Proportion of sown pasture Su no direct supplementation to animals on native pasture 0.25 0.5 0.75 06 .8 .7 .3 0 0 0 0 0 .4 0 0 .6 .7 .g 1.5 1.5 1.2 2.5 1.5 1.2 1.4 1.5 1.2 direct supplementation in Wi and Sp to native pasture animals ( + 38 kg/head) 0.25 0.5 0.75 .6 .6 0 0 0 0 .8 0 0 1.5 0 3.5 .6 .9 1.2 2.5 1.5 1.2 2.5 1.5 1.2 0 0 0 ENHANCING THE CREDIBILITY OF RESEARCH RESULTS K.G. RICKERT>k and J.P. EBERSOHNQ Concern often is expressed that farmers are not responding readily to opportunities offered by integrating research results with their production systems. In this paper we assess two methods, namely physical and computer models of production systems, of enhancing the credibility of research results. We refer to preceeding papers in this contract as examples that demonstrate how important it is for innovations to be put into production systems so that implications can be observed. However, in exploring these issues it is also useful to consider the reluctance to adopt technology and what can be done to gain more rapid adoption. A Qld. Dept. Primary Industries, Brian Pastures Research Station, Gayndah 4625. 't Qld. Dept. Primary Industries, P.O. Box 83, Nambour, Qld 4560. 19 Animal Production in Australia Vol. I5 DIAGNOSIS Rogers and shoemaker (1971) listed five classes adoption rate of new technology. Only one class of namely, the attributes of innovations as the farmer of the new technology: . is it simple or complex in adoption; . is it compatible with what he is now doing or is . are general advantages obvious; . is it likely to benefit him and to what degree? of factors which influence the these need concern us here, sees them. The farmer asks it revolutionary; The farmer's decision on complexity, compatability and general advantages is largely intuitive. However what needs to be made more explicit is the benefit to him, on his farm, and under his management. Lionberger (1961) calls the latter exercise 'legitimation' of research results or putting the local stamp of approval on it. He also describes five stages through which an individual progresses when accepting an innovation: awareness, interest, evaluation, trial and (finally) adoption. Thus legitimation is concerned mainly with the last stages. To us, legitimation means establishing the credibility of innovations, or how repeatable results are in time and in space. We contend that the farmer wants to know, indeed, needs to be given some prognosis, as to what production and financial advantages and management implications the innovation hold for him, under his unique situation. Failure to meet this need is where the potential adopter% assurance most often becomes unstuck. How do we then face up to the challenge? PHYSICAL MODELS Agricultural research commonly falls into three broad categories that contribute information to each other, to extension services and to farmers along paths illustrated in Fig. 3. Ideally the information flows in both directions but the most dominant flows change as technology is first developed and then adopted by industry. Relationships between these information flows are described by considering the three types of research indicated in Fig. 3 in more detail. Basic research is a relative term but in this context it refers to experiments that test hypotheses or solve problems associated with components or processes within a system. It provides greater insight by defining the biological principles operating within a system. Results from basic research commonly have wide application in understanding responses beyond the site or location of study, but direct application to producers is limited because components are studied in isolation. The role of soil nitrogen in the 'run down' of sown pastures observed by Robbins and Bushel1 (above) could be a subject of basic research. Applied research consists of traditional experiments with clear objectives and often direct application to industry. It is a common form of research with many variations but in our context it includes experiments on stocking rates, fertilizer rates, species evaluations, supplementary feeding, etc. Results from applied research were used to formulate the physical models of production systems described in this contract. Both physical models in this contract illustrate that results from applied research have limitations due to site, time or techn@es, and these assume importance in applying the results to integrated production systems. It is commendable for agricultural scientists to publicise findings from the physical models described here as basic and applied research, but that is not enough. Where they have studied an isolated component of a system they have an 20 Animal Production in Australia Vol. I5 Fig. 3 Possible information pa thways producer nexus within the research-extension- obligation to participate in the integration of their findings back into the system. If extzznsion officers or producers are forced to make this assessment in relative isolation from the researcher, there is a likelihood of a distorted reaction to the innovation. Demonstrations are a traditional met,hod for enhancing the credibility of innovations. They range from small plots that 'show off' a new species to large scale physical models like those discussed in this contract. They may involve both extension and research staff and be conducted on commercial properties to increase producer involvement and awareness. Often statistical interpretation is limited but they may provide a useful framework for supporting studies on various components as illustrated by both models in this contract. How, effective are demonstrations of production systems in enhancing credibility? Large scale physical models, located in an area of likely application and executed for a number of years, sample the innovation's spatial and temporal fit. Variation in results from the physical model described by Cooksley (above)is an excellent example of this feature. Secondly, physical models may indicate the magnitude of problems such as sown pasture'run-down' mentioned by Robbins and Bushell. Thus these two examples support Ebersohn's (1976) conclusion: rather than being a tool for expediting adoption, such first order combinations of forages primarily demonstrate to the researcher the amplitude of his findings. They are a simplification of the real world because of temporal,spatial and management constraints. 21 Animal Production in Australia Vol. I5 In as gross period. likely to addition, results from research can be assessed by economic packages such margins, cash flow and risk analyses, internal rate of return or payback These explore the financial cost and advantages that a novel practice is have when it becomes integrated into an existing system or situation. We contend that such physical models, or demonstrations on a large or small scale, and economic analyses, primarily are feasibility exercises that increase awareness, kindle interest, and reinforce opinions. That is they foster the first three stages of Loinberger's adoption process. The remaining stages, trial and commercial adoption, which flow from these are basically acts of faith; a step into the unknown because of many specific operating constraints. In a harsh economic climate this does not get us very far. Some more logical method is needed whereby insight is enhanced and adoption of worthy innovations expedited. COMPUTER MODELS Several approaches can be adopted in the use of computer modelling depending on the objective of the model (McKeon and Scattini 1980). The reliability, or year to year variation in production of a single option can be assessed with simple models, e.g. water balance for grass/legume pasture (Rickert et al. 1981). Complex models of the energy and nutrient flow in the soil-plant-animal system can be developed to evaluate management alternatives, e.g. burning of native pasture (Scattini and Powell 1982). The optimal allocation of property resources to various forage options can be found using formal mathematical optimizing techniques, such as linear prograrmning (Miller 1982). Recently Kennedy (1981) reviewed the role of dynamic programming as decision-making aids for a farmer's problems. The trend towards better computers and better models will continue with a parallel improvement in ability to mimic a farm situation and farmer decisions, however unique. Innovations can be included in models and their implications assessed at little cost. The computer model in this contract illustrates some implications of integrating two forage options. It is a simple model easily adapted to a micro-computer that an extension officer or farmer might use. When coupled with predictions of growth that account for temporal and space variability, and with price variability we believe it could become a powerful extension x001. Since the potential adopter or farmer must interact and state the rules of operation for both dynamic programming and simulation, these rules meet the adoption criteria set by Emery and Oeser (1958) that 'the agrarian culture dictates that knowledge must be achieved and tested by personal practice and experience'. Adopters fear the unknown, but fear is lessened if the unknown is made qualitative ('by physical models) and quantitative (computer models). ACKNOWLEDGEMENTS We are grateful for the funding support of the Australian Meat and Livestock Corporation and the critical advice of Dr. R.F. Brown REFERENCES ADDISON, K-B. (1970). Proc. XI- Int. Grassld. Congr. p. 789. -P ADDISON, K.B., CAMERON, D.G. and BLIGHT, G.W. (1984a). Trop. Grasslds 18: (in press). ADDISON, K.B., CAMERON, D.G. and BLIGHT, G.W. (1984b). Trop. Grasslds18:- (in press) 22 Anim