section 5 : results of agroforestry experiments

Tree/crop interface investigations Preliminary results with Cassia siamea and maize

P. Huxley, E. Akunda, T. Darnhofer, D. Gatama and A. Pinney

International Council for Research in Agroforestry (ICRAF)
P. O. Box 30677, Nairobi, Kenya

Abstract

The basic unit of all agroforestry systems is the 'ree/crop interface' (TCI). By studying these in small plots an understanding of the interactions involved can be developed. Such investigations are best thought of as initial, observational trials to help design more formal experiments. One such is described in which Cassia siamea and maize were grown adjacently in a' Y' design which enabled orientation to be investigated. The planting arrangements and assessment methods are described and the preliminary results of the first three seasons with the maize crop are given.

Although unreplicated, the data indicate the extent to which trees, crop and trees, and crops at the interface behaved in this association, on this site. Assessment criteria for the crop, the yield profiles from the interface to the plot edge, are shown by season for both the tilled and unfilled parts of the plots. The effects of orientation need further study.

The work so far has opened up many considerations of methodology and design not only for TCI experiments but agroforestry experimentation in general.

Introduction

Agroforestry land use systems can vary considerably in complexity in terms of numbers and kinds of plant components, their spatial arrangements, and the various temporal sequences of natural plant growth and development patterns which occur naturally, or are managed. For research purposes, systems can be grouped into 'zonal' or 'mixed' spatial arrangements where the intimacy of associated plant components is, respectively, restricted or encouraged; or 'rotational' arrangements, where the woody perennials are completely separated from the non-woody plant components in time. Spatially-arranged systems can also be manipulated temporally, but within a seasonal time frame.

The complexities of studying spatially-arranged systems can be greatly reduced if they are considered, in the first instance, to be formed from pairs of all the associated plant components; i.e., from an appropriate number of 'tree/crop interfaces' (Huxley 1986). Here 'tree' and 'crop' mean, respectively, woody and non-woody plant components, or, in some cases components. Indeed, the choice of zonal or mixed systems, and their actual design and management options, can not be logically made until an awareness of plant-environment effects and interactions at the tree/crop interfaces are known, or better still, understood.

Because the outcome of even an apparently simple tree/crop interface (TCI) experiment is the result of all short and long-term biophysical and environmental factors at the experimental site, it still presents an extremely complicated situation (Figure 1). The ultimate goal is that of estimating the size of the resource pools (for light, water and nutrients), and investigating the magnitude and rates of the environmental resource-sharing processes that produce — to a greater or a lesser degree — plant stress on each component. Measuring (or estimating) the results and relative importance of these stresses on plant growth and development can then, hopefully, provide sound guidelines for the manipulation of the system, both in order to improve it, and/or to provide grounds for extrapolation to other environmentally different sites. Even a more limited understanding of the resource-sharing processes can add considerably to the predictive outcome of an experiment.

Figure 1 Tree/crop interactions will result in and be a consequence of changes in environmental resource pools, shelter effects, etc. (Source: Huxley 1986).

Whatever the precise objectives may be, experience to date has shown that a careful set of observations of plant behaviour throughout the changes of season can expose a whole range of ways in which the plant components can be seen to be responding to primary or plant-induced environmental changes at the various stages of plant growth and development, e.g., germination, emergence, planting-out and so on. Such trials can also, by producing circumstantial evidence indicative of particular environmental stresses, help define precisely the experiments needed to investigate the environmental factors concerned.

Field research with woody/non-woody mixtures is a relatively new experience for most researchers with an agricultural or forestry background. It was, therefore, considered essential to carry out the kinds of detailed observations reported here before progressing to field trials. Because this investigation was to explore a wide range of responses of the plant materials, and to develop assessment methodology, the plots were not replicated as they would be under experimental conditions. Satisfactory designs for such experiments can only be achieved after sufficient has been learned about the way plants behave and interact in such associations, exactly what assessments need to be made, and when.

Tree/crop interface investigations are, therefore, as described here, probably best thought of as initial, observational trials which are a cost-effective way of 'opening-up' the problem. They will be followed by either more extensive, but well-focussed field trials, or by more precise resource-rich experiments in which selected environmental factors are investigated in depth. In their initial stages, tree/crop interface investigations can be carried out wherever they occur (as long as the conditions are fully understood and described). Alternatively, they can be conducted in small plots using either single trees or some relevant arrangement of the plant components, e.g., hedgerows (Huxley 1987).

It is important, if a regular, linear plant arrangement is used, to consider whether the effects of row orientation are to be investigated. Sun angles, wind that affects rainfall distribution (see next paper, this Section) and 'shelter' may influence tree/crop interface responses. If orientation is important, a 'geometric' design can be used, such as the ' Y' design used in the experiment described here. The present investigation is part of a series involving different tree/crop combinations.

Materials and methods

The 'Y' design used is illustrated in Figure 2, in which plot size and plant spacings are given.

Tree establishment and management

Young, nursery-grown, plants of Cassia siamea, were planted in April 1984, using 2 litres of diluted Aldrex (48%) in the planting hole together with two shovels of farmyard manure. No fertilizer was applied until the third crop season (see below). Plants were watered each week until the short rains started in October because it is essential in this type of design to have a complete plant stand. The area was first weeded four months after planting by scraping the surface. Subsequent weeding was done monthly at first, and then as needed under the trees as the canopy began to close.

Trees were pruned ('capped') 11 months from planting in order to regulate the biomass of each plant (larger plants were pruned more heavily than smaller ones) and to train the plants to form multiple stems and so obtain canopy closure sooner.

Figure 2 Shape and dimensions of the 'Y' design plot used in this investigation.

Pest problems started with powdery mildew at 15 months after planting. This was checked by spraying a fungicide ('Bayleton'), applied every two months during the season when the disease was most severe (i.e., at the end of a rainy period). Stalk borers were first noticed after 16 months. They were checked, but not completely eliminated, by spraying Dieldrin on the stems. Lebaycid was also applied in the damage holes, and larvae were killed with wire probes.

Subsequently, the trees were allowed to grow freely, except that the side branches were pruned vertically at the interface with the crop each sowing time. Crops were sown after the trees had grown three seasons.

Crop management

The cropped area in the outer half of each arm was left unfilled and the soil in the inner half dug with a hand hoe prior to sowing (Figure 2). An adapted short-season (90-day) maize composite ('Katumani') was sown in the fourth rainy season (i.e., October 1985) in parallel rows on each side of the three arms of the 'Y' (i.e., in 6 plots). A spacing of 0.5 m x 0.5 m was used, with the first row 0.5 m from the trees. In order to produce a complete stand, three seeds were sown per hole and thinned to one plant within 12 days of emergence, during which time any essential gapping-up was done. All maize residues were returned to the appropriate plots. Plots were weeded as necessary by scraping the soil and/or by pulling. Pesticide (Dipterex) was applied for stalk-borer control when the maize was knee high. The need to control rodents prior to sowing became apparent in the first season.

Maize was planted in every subsequent season (i.e., twice a year), as above. No fertilizer was applied in the first two seasons in order to maximize nutrient stress; but, on these relatively infertile soils (Oxisols), it was considered necessary to fertilize from season 3 on with a uniform application of N and P at the rate of 100 and 40 kg/ha, respectively, prior to sowing. Fertilizer was applied over the whole plot including the area planted with trees. Harvesting was delayed until all the maize was fully mature.

Assessment of tree and crop

Measurements were taken of the diameter of tree-stem collars (at soil level) and height of all trees at the beginning and end of every season. Maize height, date of 50 per cent tasseling, fresh and dry weight of the true stem (i.e., after stripping away the leaf bases), and cob and seed weight were recorded. In the first season this was done row-by-row, and the number of plants remaining after removing discards at the ends was recorded. In the second and third seasons these measurements were taken for every plant. It was not possible to obtain measurements of leaf area index for either the crop or tree, but photographic records were kept throughout the seasons.

Results and comments

We provide here a selection of data from the first three seasons when maize was sown. The data illustrate some of the issues that have emerged from the study so far. The work will continue for at least three more seasons.

Rainfall

Figure 3 shows the rainfall. In this region the rainfall is bimodal, usually with a 'first rains' (March-May) and a 'second rains' (October-December). The pattern over the period of observations taken here was consistent with this but some rain extended into January in both 1986 and 1987.

Cassia siamea

Figure 4 indicates the growth of C. siamea in height and stem collar diameter. As expected, trees at the interface were larger in both respects than those in the middle of the plot, where a fairly uniform stand was apparent. As the standard errors of the row means show, trees were more variable on the outside, i.e., at the interface.

Maize

Assessment criteria

Figures 5a and 5b show, for tilled and unfilled plots respectively, measurements of maize height on three occasions during the second planting season, compared with the yield of maize cobs at harvest. Even in agricultural experiments plant height is an unsatisfactory assessment criterion, since, for any single germplasm selection, it represents the net effect of the stage of plant development, vigour of growth as affected by the plants' immediate environment, and morphogenetic effects of shading (mutual, or otherwise). The last is likely to be a common effect in TCI experimental plots and, as Figures 4a and 4b indicate clearly, there is a very poor relationship between the height of maize plants and their cob yield. In this third season (second rains, 1986), untilled plots produced less maize than did tilled plots, but in neither case was yield affected to any extent past the first or second row. However, there was an indication of a trend for greater yields towards the outside of the plots (i.e., away from the interface). A detailed analysis of different tree/crop interface effects awaits the results from the continuation of this work.

Figure 3 Rainfall at the ICRAF field station, Machakos, Kenya, for the period of observations.

Figure 4 Tree diameter and height on 14 January 1987, at the end of the observation period. Means for three tilled and untilled arms. 

Figure 5 Maize height and cob dry weight by row, second rains 1986; a) means of 6 tilled plots; b) means of six unfilled plots.

Effect of season and tillage

Figures 6a and 6b show, for tilled and untilled plots, respectively, the same maize yield trends at and away from the interface for each of the three seasons in the tilled plots. Although general yield levels varied considerably between season, the untilled plots always yielded less than tilled plots. However, there is little indication that the degree and extent (i.e., general shape) of the interface effect was different between seasons for the tilled plots. There was little change in maize yield between mid-plot and the interface for untilled plots in the last season (Figure 6b).

One reason for growing maize with and without tillage was the hypothesis that undisturbed tree plots might extend more rapidly into the untilled area and thus competitively affect adjacent maize yields more adversely than on seasonally tilled plots. A more detailed examination of interaction between tillage and season must await the results from subsequent seasons.

Effect of orientation

The 'Y' (120°) design was chosen as the simplest (and least costly) approach to observe major effects of orientation. Figure 7 shows an example of maize yields from the tilled plots in the second (short rains) season of 1985. So far no clear cut influence of orientation has been observed with this tree/crop combination on this particular site. There is some indication that emergence of maize in rows close to the interface on the 'rain shadow' side (see Section 000, this volume) can be slightly delayed. There may also be shading effects that need further investigation.

If shading and/or shelter by the trees influenced the crop, then one would expect to observe differences b when comparing rows with difference in proximity to the interface and with different orientations. Although the maize yields from the first two rows of plots 1 and 6, and 2 and 3 (refer to Figure 2) were slightly higher than in plots 4 and 5, with the sun traversing from south of east to south of west at this time of the year, and winds being predominantly easterly, there are large (and dissimilar) differences occurring in the middle of the plots. Hence even row-by-row comparisons present conflicting evidence that will require more detailed study to resolve.

Discussion

Even in its early stages, this investigation (together with those on several other layouts and tree/crop combinations which the team is studying) has provided a good deal of useful guidance for the design, assessment and data-analysis methodology for both tree/crop interface experiments/w se and field trials with tree/crop mixtures in general. Because, in some seasons, data have been collected for maize on an individual plant basis, it is possible to use this for assessing suitable plot sizes, the extent of guard areas, etc. Indeed, it has become clear that the outside part of each plot (away from the interface) is not merely a 'discard' area but can, itself provide clues about the growth and yield of the crop growing under a more stress-free (but, perhaps more environmentally exposed) situation.

As well as assessing maize height and cob yields, we examined true stem dry weight (by removing the sheathing leaf base) and seed dry weight. True stem dry weights were not found to be useful and, in any case, required a great deal of labour to obtain. Seed dry weight, as might be expected, is highly correlated with cob dry weight for any particular germplasm source.

Some of the interface results obtained may well depend on below-ground competition and we intend to explore this, either by root sampling and/or by indirectly observing seasonal changes in the status of soil water profiles.

It is becoming clear that effects of orientation will need to be investigated over a larger number of seasons with different rainfall regimes, because orientation is likely to influence different stages of crop growth in different ways.

Finally, the type of crop response at the interface that we have described here is not always found with other tree/crop combinations or on other sites. If future work confirms this, then this has important implications for the design and implementation of agroforestry field experiments because, even if they prove to be imposed by site and/or management. It is essential to be able to control or allow for such effects if experimental variability is to be contained.

Figure 6 Effect of season on row yields of maize (9 rows in the first season, 11 in the other two; a) means of six tilled plots; b) means of six unfilled plots.

.

Figure 7 Effect of orientation. Cob weight by rows for separate tilled plots in the second season 1985.

Acknowledgements

ICRAF's 'tree/crop interface' project on research methodology for conducting experiments involving woody and non-woody plant mixtures has been largely funded by the Federal Republic of Germany through BMZ/GTZ. It has been carried out at the ICRAF field station for which SIDA has provided supporting funds.

References

Huxley, P.A. 1986. The tree/crop interface or simplifying the biological/environmental study of mixed-cropping agroforestry systems. Agrofor. Syst. 3: 251-266. (This is a revised version of ICRAF Working Paper No. 13,1984).

Huxley, P.A. 1987. Working committee recommendations for initial research proposals for the ICAR/ICRAF workshop-cum-training course on agroforestry research, held at C.R.l.D.A., Hyderabad, India, 16 September - 1 October 1986. Nairobi: International Council for Research on Agroforestry (ICRAF).