P L O T S T R U C T U R E — L E C T U R E N O T E Plot structure Plot structure—R. Coe Introduction Even though the basic rules of experimental design remain the same, agronomists, foresters and agroforesters face different problems when designing a plot structure for an experiment. Agroforestry trials are often more complex because of the presence of different components and their long-term nature. An agronomist will basically have to choose plot length and width and will use standard agronomic practices for an agricultural experiment (e.g. variety screening, fertilizer experiment). There is a lot of experience available in standard agricultural experimental design and plot structure. The consequences of getting it wrong are mostly limited since quite often these experiments only last one growing season. An agroforester will also have to choose plot length and plot width but in addition to this will have to consider tree and crop numbers and position in an agroforestry experiment. Being a relatively new approach to land use and a rather complex field, there is much less experience available in agroforestry experimental design and plot structure. Getting it wrong at the start of an experiment can have far reaching consequences on its results and validity because such experiments often last several years. Plot design and types of experiments Different types of trials may require different types of plot structure. For example, screening trials of tree species and provenances can use a line or a block plot structure. Gross plot Net plot Tree Line plot Block plot It must be noted that only a limited number of characteristics of interest to agroforestry can be measured in such plots. P L O T S T R U C T U R E — L E C T U R E N O T E In a process trial, the plot structure will be defined by the objective of the trial, e.g. studying rooting depth, and such trials do not necessarily have to consider farmer practices or particular agroforestry systems or technologies. Such trials are implemented to address some very specific research questions. In a systems trial, the experimental plots do have to consider existing farming systems, practices and conditions. Experimental plots and treatments should be designed and put together in such a way that the experiment resembles a farmers field. In the following drawing, Plot A has hedgerows on the edges of the plot. If new plots are added adjacent to this one, their hedgerows will be too close too each other and the complete experiment will not resemble a field planted under this technology. The ‘systems’ plot will reflect this much better. Hedge Crop row Plot A ‘Systems’ plot Other examples of “systems” plots and experiments are those resembling “trees in cropland” or “boundary planting”. In the case of a boundary planting experiment, crops planted adjacent to the tree boundary should be grown on an area that is sufficiently large to represent a real farming situation. Data taken from plots with narrow crop areas do not adequately represent a farming situation. Plot design features Size of plots: In theory, larger plots give a higher precision for a fixed number of plots but a lower precision for a fixed total area. In practice, leaving border areas will modify theoretical considerations. Plots must be large enough to apply the treatments and to collect the necessary measurements. They should also be small enough to be cost effective, to manage the experiment and to fit in the available land area. Shape of plots: Long thin plots are more precise than square plots of the same area but in practice, the shape of the plots will be determined by the system under consideration, management requirements and other practical considerations regarding the available area. P L O T S T R U C T U R E — L E C T U R E N O T E Guard rows and border areas: Theoretically, the measured area of a plot should behave as a whole field that received a specific treatment and one plot should not influence another one in any way. Interference between plots is one of the largest sources of errors in agroforestry experiments and will seriously affect the results and thus the validity of the experiment. The following steps can be taken to avoid certain types of interference: 1. leave large borders (costly since they will need to be managed and will not produce experimental data). 2. spatially separate the plots (requires more land and may decrease precision due to site heterogeneity). 3. physically prevent interference (root pruning, wire mesh barriers,...). This can be expensive and is not always effective. The following table illustrates the types of interference one may experience in an improved fallow experiment and ways of controlling this: Above ground At ground level Below ground light shading water moves between removal of water and plots nutrients through tree root invasion windbreak effect litter moves between plots rain shadow Amelioration by: leaving a border using small bunds avoiding edges pruning replacing residues root pruning avoiding edge areas barriers The main problems with interference between plots are that: 1. the effects below ground are often not noticed in good time 2. the bias tends to be in favour of the “agroforestry” plot 3. the problems get worse with time as the trees grow and develop 4. remedial steps taken when the experiment is already established may not be effective since the problem may already have influenced the experiment..
Pages to are hidden for
"Plot structure"Please download to view full document