1. INTRODUCTION

Guidelines for field operations in connection with establishment of the Neem provenance trials were originally prepared by a working group during the first international consultation on Neem genetical improvement held in Bangkok January 1993, (10) & (1).

The second international consultation recommended a review of the initially prepared guidelines, and a second working group was nominated and has revised and elaborated on the guidelines. A proposal was presented and discussed at a meeting of the coordinating panel held in Bangkok during 14-15 March 1996. The following paper presents the revised guidelines including important considerations when planning the establishment of field trials. Recommendations for data collection and assessment procedures will be formulated during 1996 by a working group on trial assessment.

Participants who are uncertain about some statistical or practical problems are welcome to write to the Coordinator of the Net-Work.

2. FIELD TRIAL DESIGN

2.1 Objectives of scheme

The objectives are to evaluate performance in respect of adaptation and growth of seedlots of Neem of different provenancelorigin on a number sites and determine whether there are any seedlot by site interactions. See also annex B.

2.2 Summary of recommendations for field design

This section gives only summary recommendations. Arguments and further details are specified in Annex B.

2.2.1 Choice of general design

A proper design. i.e. shape size and arrangements of plots and block, can only be determined when the trial site has been selected and the plants are ready

Examine the chosen site for environmental variation before it is cleared of its vegetation cultivated and otherwise treated. Determine the magnitude and pattern of environmental variation, decide what degree of variation should separate blocks, and what should be the maximum variation to tolerate within blocks. On a map outline the locations of uniform conditions according to the chosen criteria of maximum variation. Then lay-out the blocks and prepare a map before clearing and cultivation are done.

The design to use must be optimal for the local and actual conditions of site and resources; this consideration must be given the highest priority!

Choose as simple a design as possible with regard to site conditions. For a single trial, the layout, administration, and assessment are easier, and the analysis and interpretation of data is simpler and generally more effective, the simpler and more balanced the design.

When it is certain that - site conditions are not very variable over the whole area, use a simple and robust randomized complete blocks design (RCB) including 4-6 blocks with 25 tree arranged in 5 rows by 5 columns.

• If you have limited land or resources, chose 4 blocks, but if you have more freedom choose 6 or more replicates (see "Number of blocks or replicates" below).

Where site conditions seem to be quite variable, the first thing is to ensure that these conditions are uniform (same gradient or patchy variation) within each block, so major differences occur between blocks. This is a most critical and important consideration.

• This is done by modifying size and shape of blocks as necessary. Size is modified either by i) modifying size of plots (see "Plot size" below), or by ii) modifying the number of treatments in blocks.
• Blocks do not need to be of regular or the same shape; a block should, for administrative reasons, preferably be contiguous; different blocks do not need to be neighbouring.

If there is a strong wish to have large plots on a variable site with the obvious result of gradient or large variation within complete blocks, then an incomplete block design should be chosen where the large plot size can be maintained, while block size is made according to the existing variation. Use preferably a balanced type of incomplete blocks design.

2.2.2 Plot shape and size

A plot must sample the existing within-block environmental variation as completely as possible. This would include variation of topography, of soil, water, exposure to wind, exposure to shadow, and competition from surrounding trees, cattle tracks, rides, tracks, etc.

• Compact, e.g. square, plots are useful specifically where the environmental variation is of a patchy type, for example termite mounds distributed regularly over the area.
• In a design with many treatments, square or rectangular plots of 25 or more trees are suitable only when larger blocks of reasonably uniform environmental variation can be made. They are generally not suitable on a gradient type of variation.
• When the environmental variation for a specified plot size obviously changes from plot to plot within blocks, then a reduction of plot size will have to be considered. Use as small a plot size as is necessary. One-tree plots should be used only if the skill of staff at all levels is high.

Use line plots on sites with a gradient type of variation. Line plots are oriented along (parallel to) the gradient.

Plot size may also be influenced by the number of treatments (here seedlots) to include in the trial. Let the provenance with the smallest number of plants to some extent direct the choice of plot size, so all provenances are included. For example: one provenance has 81 usable plants, i.e. less than all other provenances. Make a trial with 16-tree plots in 3 blocks, or g-tree plots in 9 blocks, or even 5-tree plots in line and up to 16 blocks. An example is presented on page 14 in annex B.

Plot size is also influenced by spacing between plants. The trial host organization is to determine spacing according to local condition and experience. Spacing should vary between 3 x 3 metres and 4 x 4 metres. The better the growth conditions, the wider the spacing should be used, and vice-versa.

2.2.3 Cost and size of experimental area

The smaller the total experimental area, the cheaper the maintenance and protection. If funds for establishment and maintenance are limited, a reduction in experimental area should be considered. Reduction can be done by reducing the plot size or by changing design to a type of incomplete balanced design. For a completely randomized blocks design do not go below 9-tree plots in 4 blocks, or 5-tree plots in 6 blocks.

3. TRIAL SITE

3.1 Choice of site

The trial site should as far as possible be representative of localities where Neem is to be planted.

Avoid sites that are definitely unsuitable to Neem, i.e. primarily sites without a reasonably deep soil. Neem will definitely suffer where there is a hard crust too close to the surface that can disturb the development of the taproot system of Neem.

3.2 Site description

The description of the site should include a general site description plus maps. Annex A presents suggestions for site description. The following maps would be useful:

1. a location map showing the site's relative position to a town or other well known characteristic landmarks; the scale would be from 1:100,000 to 1:1,000,000;
2. a map showing in detail how to come to the trial, usually in a scale of 1:50,000-1:100,000, and
3. a detailed map of the layout of the trial with blocks and the seedlots marked, in appropriate scale, is essential. This map should clearly show compass direction as well as location of each seedlot, planting space and date of planting, a description of the demarkation used for blocks and plots, border lines, as well as surrounding vegetation and landscape features. Finally,
4. an environmental map showing the layout of blocks in relation to environmental variation would be useful. This could include either contours, or a three-dimensional drawing.

3.3 Site preparation

After having observed and demarcated environmental variation based on vegetation, topographical features, obvious soil characteristics, and surrounding features that may influence the trial (rivers, water holes, older plantations or forests, and the like) the vegetation is to be cleared, preferably stumps should also be removed to avoid coppicing Thereafter a thorough soil preparation is carried out according to soil type and normal practice.

3.4 Demarcation and positioning of blocks and plots

The position of block corners is most critical and is to be observable for several years, preferably for the entire duration of the trial. Block corners must therefore be demarcated using large posts of durable wood or concrete, which are solidly fixed in the ground.

Plot corners are marked with smaller posts that do not need to be very durable.

Planting spots are marked with small pegs, wooden or bamboo stakes.

The arrangement of blocks has been discussed above; if an incomplete block design has been chosen, then the blocks have to be arranged at random within each replicate. The arrangement of treatments, i.e. seedlots must be done randomly within each block. Assign each treatment (seedlot) a number from 1 up to the number corresponding to the number of treatments. Randomize the position of the treatment numbers within each block. Use tables of random permutation, marbles, of paper slips to ensure proper randomization. See also (5) for randomization, and examples of making random permutations.

4. PLANTING

The procedure of planting of potted plants and of stumps is presented in annex C.

The problems of adequate selection of plants and labelling are discussed below.

It is important that usable plants of each respective seedlot are assigned properly to plots. Do not grade plants, for example in order to plant small plants in one block, etc, because this may incur a selection of families (a family is seeds from one tree) within a seedlot. It is necessary to have representative samples of all families of each seedlot in each plot and block in which the seedlot is planted. A practical approach is to divide the nursery bed for a seedlot into a number of plots corresponding to the total number of plants of that seedlot to be planted in the trial. Then within each plot select systematically one usable plant, for example the usable plant in the centre of the plot, and if none there, then the nearest usable plant in one specified compass direction.

Be sure that each plant is properly marked with a label showing seedlot, plot and block identification.

If the planting site is located so far away from the planting site that all plants are lifted and brought to the planting site in one transport operation, then assign seedlots to plots and blocks only at the planting site. If on the other hand nursery and planting sites are close to each other, then lift, remove and transport plants of all seedlots for one block at a time.

Try to complete planting in one block at a time. In this way differences in climate from day to day is absorbed by the blocks. Avoid letting one person plant all plants of one seedlot in all blocks, and so on, because there are commonly substantial differences between persons in planting success. Instead, let all persons plant in one plot at the same time, or if impractical then assign plots in a block randomly to persons (randomize separately for each block).

Immediately after planting, or ideally during planting, be sure to prepare maps as described in section 3.2 above. It would also be useful to prepare a small report describing operations and climatic conditions during trial establishment.

5. POST PLANTING MAINTENANCE

Reference is made to annex D, which presents the procedures recommended by the previous work group (1 and 10). Most importantly, keep plots completely free of weeds as long as is possible with regard to economical possibilities! The same type and intensity of management must be done to all plots in the experiment. If a management operation cannot be done for the whole experiment within a short period, i.e. a few days, then be sure to complete activities by blocks at the time.

6. LITERATURE & REFERENCES