Background: Farm trials in general

Agricultural research in lower-income countries has seen a marked move away from research stations towards trials on-farms – which has generally been justified on the basis that:

• much technology developed purely on research stations has never been adopted;
• conditions and practices on research stations seldom represent farm situations;
• farmers cannot assess applicability of technology on stations as well as on farms;
• working (even partly) on farms makes scientists more sensitive to farmers' needs; and
• co-researching with farmers can lead to higher rates of innovation than more traditional research approaches (e.g., Norman and Modiakgotla 1990).

Farm trials can never replace research station experiments for complex studies involving several treatments, where control of variables is essential. However, all the points listed above may apply in certain situations in Australia, and the last two points appear to be gaining particular importance - as reasons for initiating programs of on-farm trials in Australian R&D programs.

The commitment of farmers to a process involving design and conduct of trials on farms could be considered as the epitome of participatory RD&Extension, and offers great potential toward developing sustainable farming and agro-forestry systems. While most of the literature on on-farm trials is about crops and animal enterprises, the concepts involved are just as relevant to work with farmers on trees on farms.

Despite the growing acceptance of concepts of farmer participation in research, the topic of on-farm trials has received little attention in the Australian research literature - partly because it is seen as "unscientific" by agricultural and forestry journal editors. Clarifying the differences in aims between various types of trials and experiments may help to improve understanding and importance of the concepts of on-farm research.

On-farm trials in small-scale farming have been classified (Ashby, 1987; Okali et al 1994) in terms of the roles played by researchers and farmers in trial design and implementation (see Box 1. Classes 1-3). The fourth class listed in Box 1 is a variant commonly used in Australia, particularly by large farmer groups such as Southern Farming Systems (1999) and Birchip Cropping Group - where the aims are often set by farmers but the trials run mainly by scientists in relatively controlled situations on farms or demonstration sites.

Box 1: A typology of “On-farm trials”

1. Researcher designed, researcher run 2. Researcher designed, farmer run 3. Farmer designed, farmer run 4. Farmer designed, researcher run 5. Co-design by farmers with researchers

The five categories in Box 1 involve different levels and types of participation by farmers in the design and conduct of farm trials. Higher levels of farmer involvement in trials has been correlated with higher "ownership" and also higher adoption of technologies of ideas tested (Ashby 1987). In Australia, farm trials have been classified mainly in terms of their level of statistical precision (Patabendige et al, 1999), as depicted in Figure 1.

Figure 1: A broad classification of types of on-farm trials

* Farm trials designed at various levels of statistical sophistication are discussed by Patabendige et al (1999) in the “Test as You Grow” manual, and Hunter & Hayes (1996) for plant nurseries.

In Australia now, a number of standard designs (and even manuals) have been published for on-farm tests and more sophisticated trials on farms that are planned to maximise statistical validity. Some R&D programs offer advice on design of farm trials by fax (Blake et al 2000). These designs are allowing farmer groups to test their ideas in their own conditions, on farms or demonstration sites - usually run by technicians (e.g., Southern Farming Systems 2000). Some enterprises such as nurseries and other forms of horticulture, lend themselves to running multiple treatments trials on farms - but most broadacre enterprises do not.

Adaptation trials on farms

In reality, there are many ideas that farmers have for testing that do not lend themselves to statistical design at all. Here the aims of trials are "adaptation" or initial testing, or broad initial comparison, that requires innovation and adaptation by the farmers during the testing period. The "data" from such trials are often on observations and/or opinions of farmers - based on their practical experience in assessing or handling the subject of the trial (e.g., implement, animal feed, chemical, crop, tree, spacing, planting method).

Adaptation trials on farms may fall into class 3 in Box 1 above, but where research/extension agents instigate the trial and recruit farmers, it may fall into class 2. Increasingly, farmer groups are designing trials collaboratively with researchers (class 5). From both the farmer's and the scientist's point of view the objectives of adaptation trials on farms may be to:

• test and adapt an idea to suit local soil, implements or other farm conditions
• evaluate practicality of ideas from research stations, on a small scale before wider use
• elicit (farmer) opinion on the viability and potential of new ideas for improvement
• gain participation and ownership of problems by a farmer or community group.

The idea being tested (e.g., a new implement) may allow a design with some replication across farms, but the statistical precision is usually very low. Where a number of farms are involved, the trial is laid out according to a fairly standard format on each farm. But farmers are encouraged to adapt the idea to "make it work" in their situation. Performance may be compared generally with standard practice. However, the main interest is in farmers' opinion of the potential of the idea, and in the adaptations made (to make the idea work), rather than in statistical differences.

The methods and materials used (e.g., implements, stocking rate) in adaptation trials are kept as close to the local convention as possible. Recording forms are prepared to document all information / measurement / opinions at agreed intervals. A camera can be valuable means of recording events, practical problems and results, to share with the participating group.

The concept of “adaptation”

Adaptation trials on farms can be a powerful means for farmers to harness the ability within a participating group to adapt ideas or technology (new or from other areas) to suit their local farming conditions. This is closer to what farmers traditionally do on their farms than normal experimentation - but with more planning of the process, measurements, observations and reflection by outsiders, than usually happens on farms.

Because conditions on each farm will differ and farmers are encouraged to adapt the idea (e.g., a new implement, sowing method, silvicultural practice) to their particular needs, statistically valid comparison may not be feasible. However, the objectives of adaptation trials differ from those of other on-farm trials. While measurements of inputs and performance are often made, the main interest is in farmer opinion of the practicality of the ideas (and adaptations made by farmers), rather than on statistical differences between treatments.

There is considerable scope for farmer groups to undertake adaptation trials on their farms, using PIRD (producer initiated R&D) grants and other funding aimed at promoting farmer innovation. The most interesting results of such farm trials are often in terms of general progress of a group's learning about a topic, and can be quite unexpected - as illustrated in the example outlined below.

Difficulties in conduct of farm trials

While farm trials of all sorts have great potential for co-learning by scientists and farmers in many farming systems and areas, difficulties are often experienced in conducting on-farm trials initially in any one area or group. These most commonly arise from differing perceptions by those involved, of the purpose and nature of the activity. Scientists often wish to see farm trials conducted with great precision and control, as on research stations. Farmers are often used to seeing researchers making the decisions and also supervising all the work of (conventional) trials on farms - as in variety testing programs. The most essential need in any farm trail is for the aims, process and responsibilities to be clearly discussed and agreed before the trial starts, and reviewed regularly during the trial process.

Adaptation trial on tree establishment – an example

A group of Wimmera farmers obtained Landcare funding for a farm trial aimed at improving the success of their efforts to re-establish native Buloke trees on their properties. Three group members each volunteered to fence two sites on their farms - one under Buloke woodland and one on land that had been cropped for many years.

The aims were to compare the success of three methods of tree establishment - direct seeding, standard seedlings and "speedlings". In addition, each treatment was split into two sub treatments - watered and non-watered. A standard layout was used at each site (with three replicates) but soil conditions and weed intensity varied between sites, as did rainfall after planting. Numbers and height of surviving trees were recorded (and photos taken) at various dates, and group meetings were held to discuss these results.

In addition to natural variations between the sites, there was variation due to farmer adaptations. The three farmers were asked to use their own judgement as to when to weed and water the trees - according to local conditions and their past experience. All these differences in treatment meant that no rigorous statistical comparison could be made. Yet, by closely observing the trial and the different responses on the three different farms, the group members were confident enough after one year to make some general conclusions - that helped in their own farm situations (Box 2).

Box 2: Some preliminary results of a trial on farm tree establishment - in a dry year

standard seedlings survived and grew much better than trees planted by other methods weed control was the most critical factor in survival of Buloke seedlings the weed seed bank on cropped land was much greater than under Buloke woodland. there was no inhibition or beneficial effect of planting under mature Buloke trees extra watering in the dry season greatly improved tree survival direct seeding did not work at all in a dry year

These five conclusions seem nothing new (and certainly lack statistical certainty), but farmers often place higher value on visual observations from their own (or a group's) trials than on information from other sources. Although there is danger that farmers will misinterpret the results of such an "uncontrolled" trial, farmers tend to base their conclusions less on the numerical data from the trial - but more on their observations of what transpired under the practical conditions faced on each farm, and on discussions with the farmers involved. One very important outcome of the trial was not planned at all (see Box 3).

Box 3: Unexpected outcomes of adaptation trial

A most important outcome arose when two farmers in the group were inspired - partly by the group trial activity - to run their own farm observations. They came up with a Buloke establishment method that surpassed all other methods in effectiveness – which is now being widely adopted by members of the farm tree group (meticulous weed control for two years near mature Buloke trees). Thus, the concept of adaptation can stimulate innovation within the participatory group.

Some lessons from past adaptation trials on farms

One problem with the above trial in the first year was that the farmer participants waited for "instructions" and help from a participating student to undertake weed control; they did not feel full ownership. Their respective roles had not been clarified well enough. Farmers did not see themselves as researchers and initially looked for others to “do” the research.

There can also be difficulty amongst scientists not wanting to hand-over the reins to farmers.

Many scientists have difficulty accepting the concept of non-statistically designed trials. It is vital to clarify the (non-statistical) aims of this particular types of trial with farmers and scientists involved (a) before the trial, and (b) in interpreting the results.

Farmers who have seen multi-location trials (e.g., of varieties) run by researchers on local farms (as has been traditional for many years) often expect researchers to supervise all the work of trials on farms. In adaptation trials, farmers' skills are needed to adapt the ideas to best suit their conditions. So with all farm trials, it is critical to clarify the expectations and responsibilities of each farmer and adviser (and/or students or volunteer workers) who may be involved in the trial, from the very start.

Farmers or groups must be well aware of the likely work loads involved in running farm trials from the start. Where there is concern about the work involved, they should be discouraged from taking part - without employing outside labour/help at the critical periods.

Measurements taken should be commensurate with the precision that is feasible - not excessively intensive or so meagre that information is wasted. In the case described above, the measurement of height were not adequate to assess growth rate, so farmers decided to count all branches on seedlings in the second year. It is important to stress that data on farmer opinion of the practicality of the idea being tested can be as valuable as any measurements made in the field. Such data are often collected by means of ranking or scales of approval or disapproval.

Observation and reflection - on all actions and opinions is essential

Regular observation and discussion is vital in the case of group farm trials - so that all members can have input into decisions about management, and hence gain and retain ownership of the trial and its results. Distribution and reflection on the results of trials at each stage is essential part of the co-learning process: What did different farmers think and learn? What did scientists think and learn? Why the differences? Documenting the mistakes made is vital if future years and trials are to run better. Mistakes often include failures in communication as well as technical errors, and these should all be documented and discussed.

References

1. Ashby J. A. (1987) The effects of different types of farmer participation on the management of on-farm trials. Agricultural Administration and Extension. 25, 235-152.

2. Birchip Cropping Group, Birchip, Victoria, Australia.

3. Blake J., Patabendige D. & Pritchard I. (2000) "Test as You Grow Kit": A kit and service for broadscale on-farm testing, In Petheram R.J. (ed.) A Manual of Tools for Participatory R&D in Dryland Cropping Areas. Rural Industries Research and Development Corporation Publication No. 00/132, Canberra, 105-108.

4. Hunter M.N. & Hayes G. W. (eds) (1996) The DOOR Manual for Plant Nurseries. Let's do our own research. Queensland Department of Primary Industries, Brisbane.

5. Norman D. and Modiakgotla E. (1990) Ensuring farmer input into the research process within an institutional setting. Agricultural Administration (Research and Extension) Network. Network Paper 16. Overseas Development Administration, London.

6. Okali, C., Sumberg, J. and Farrington, J. (1994). Farmer Participatory Research. Rhetoric or Reality. Overseas Development Institute, London, U.K..

7. Patabendige D., Selesnew N., Blake J and Pritchard I. (1999) Test as You Grow. A manual for broadacre farm testing. Miscellaneous Publication 2498. TopCrop. Grains Research & Development Corporation.

8. Southern Farming Systems (1999) Trial Results 1999. Southern Farming Systems, 96 Yarra St, Geelong, Victoria. http://www.sfs.org.au/sfs/WhatsNew/trialsbook2000