Sagary Nokoe
8.0 Performance objective
8.1 Introduction
8.2 Examples of experimental designs
8.3 References
Technical paper 8 is intended to enable you to:
· Describe examples of experimental designs being followed in ongoing alley farming field trials.
This paper provides the interested reader with a set of examples of experimental designs for various types of alley farming trials. The examples are drawn from actual ongoing or proposed field trials. They cover basic designs which appear to have universal acceptability for alley farming experiments.
The basic principles for the design and layout of alley farming trials are covered in the previous paper (Technical Paper 7). The standard design recommendations for AFNETA collaborative research projects are available from the network coordination unit.
Example 1: Fallow management in alley farming.
Example 2: Screening of multipurpose trees in different intrarow spacings for alley farming.
Example 3: Screening of Gliricidia collections across a range of environmental and edaphic conditions in West Africa.
Example 4: Assessment of effects of feed supplementation from different legume trees.
Example 5: Comparison of the effectiveness of 3 leguminous tree crops in soil fertility maintenance and in sustainability of crop production.
Example 6: Efficiency of selected multipurpose tree species in alley farming on soil fertility regeneration and agricultural crop yield.
Example 7: Alley farming trials concerning both soil fertility and animal production.
Example 8: Evaluation of species mixture in alley trials.
Example 9: Effect of tree density on water relations of trees in alley farming systems in the dry areas.
Example 10: Evaluation of the negative effects of fodder uptake on agricultural crop yield.
Example 11: Light interception and its effect on crop yield in alley farms.
Example 12: Effect of lime and manure application on the growth of hedgerow species in strongly acidic (pH 3.55) soil.
Example 13: Integration of short grazed fallows in rotation within Leucaena alleys and their effects on soil fertility and crop yield.
Example 14: Pattern of N buildup in the pens of sheep receiving different feed supplementations of alley shrubs.
Example 15: Manurial value of manure dug out from pens receiving known levels of leguminous fodder supplements (Extension of example 14).
Example 16: Effect of hedgerow species on surface soil physical properties.
Example 17: Growth of alley shrubs in farmers' fields
Example 18: Evaluation of an alley cropping species, Calliandra calothyrsus (Meissn.) on an Oxic Paleustalf.
Example 19: Effect of alley crop combinations on sequentially cropped maize and cowpea.
Treatment:
A  4 year cropping/2 year unmanaged fallow
B  4 year cropping/2 year managed fallow
C  4 year alley cropping/2 year unmanaged fallow
D  4 year alley cropping/2 year managed fallow
Intercrop: Maize
Notes:
(i) Treatment combinations will allow comparison of normal cropping with alley cropping at the end of or at any time during the 4year period, as well as monitoring of the effect of fallow management and its interaction with cropping.
(ii) Design can be Randomized Complete Block (RCB) with a minimum of 3 replications.
(iii) Assessment can be in terms of changes in soil fertility status, crop yield/economic returns, etc.
(iv) Possible layout
A 
C 
D 
B 
A 
D 
C 
B 
B 
A 
C 
D 
(v) On farmers' plots, the design can be modified slightly as indicated in the layout below:
A 
B 
Normal Cropping 
C 
D 
Alley Cropping 
With each farmer as a replicate, this can be considered a splitplot arrangement, with type of cropping as the mainplot and management practice in the subplot.
Treatments:
Two factors (Tree Species and Spacing) are involved:
Factor A Tree Species
A1: Acacia albida
A2: A. manginum
A3: Azadirachta indica
A4: Albizia lebbeck
AS: Leucaena leucocephala
A6: Gliricidia sepium
Factor B Intrarow spacing
B1: 50 cm
B2: 100 cm
B3: 150 cm
B4: 200 cm
Notes:
(i) Possible design: m a split plot (with species in main plots and intrarow espacement in subplots) in an RCB design with 3 replications. If the levels of Factor B differ from species to species (which is a possibility), this would lead to a nested design with intrarow spacing nested in the plots containing the trees.
(ii) A possible layout for one replicate could be:
REP 1:
Factor 





Factor A:
B1 





B2 





B4 





B3 





A. manginum 
Treatments: 12 different accessions (ILG50  ILG61)
Design: 3 x 4 rectangular lattice in three replicates, for 5 locations (triple rectangular lattice).
Possible layout at each location:
Block 
Group X 
Group Y 
Group Z 
(i) 
5 4 6 
2 5 12 
4 3 11 
(ii) 
1 2 3 
9 3 6 
2 9 10 
(iii) 
10 11 12 
4 7 10 
6 8 12 
(iv) 
7 8 9 
1 8 11 
5 7 1 
Treatment: Four treatments corresponding to 4 feed rations formulated as follows:
1. 
Normal ration (ALP) 

2. 
ALP + Gliricidia sepium 
(800 gm DM/animal/day) 
3. 
ALP + Leucaena sp. 
(800 gm DM/animal/day) 
4. 
ALP + Flemingia sp. 
(800 gm DM/animal/day) 
Procedure:
(a) Select 20 animals; usually they will be of different weights or ages. Divide animals into, for example, 5 weight groups of equal sizes, and ensure that each weight group receives all treatments. The design is thus identical to a block design with 5 replicates per treatment.(b) Alternatively, allocate the 4 treatments randomly to the 20 animals, ensuring 5 replicates per treatment. This can then be considered as unrestricted randomized design. However, considering differences in weights and the possibility of unbalanced mean weight among treatment groups, the method of Covariance Analysis with initial animal weights as covariate should be used. The analysis of covariance is an extremely valuable statistical technique for increasing precision.
Treatments:
Two factors are involved in this example, namely, tree species and fertilization:
Tree species  Azadirachta indica, Leucaena sp., Gliricidia sp.
Fertilization  NPK 45 kg/ha, No fertilization
Design possibilities:
(i) For onfarm, farmermanaged trials, the ideal design is splitplot with fertilization in the main plot.(ii) For researchplot trials, factorial arrangement in a block design allows equal evaluation of both factors. However, if interest lies more on one factor than the other, then the splitplot arrangement is recommended, with the more important factor in the split or smallest plot.
(iii) The factorial arrangement in a Randomized Complete Block (RCB) is used to demonstrate the analysis of variance outline.
Treatments:
A. Alley farming with Species A
B. Alley farming with Species B
C. Alley farming with Species C
D. Nonalley farming
Notes:
(i) A LatinSquare arrangement is most suitable
(ii) Possible layout could be as follows:
A 50 * 
B 12 
C 18 
D 92 
B 63 
C 4 
D 52 
A 8 
C 12 
D 72 
A 35 
B 83 
D 21 
A 25 
B 32 
C 62 
* Numerical figures represent hypothetical data after 4 years of trial.
The data concerning treatment combinations could be observed from a known variable.
Treatments: Three pruning regimes are planned, as follows:
1. Zero mulching/3 prunings for feeding.
2. 1 mulching/2 prunings for feeding.
3. 2 mulching/1 pruning for feeding.
4. 3 mulching/0 pruning for feeding
Assessment considerations:
(i) Dependent variables cover both the effects of mulching on soil and the effect of feeding animals with prunings, assuming the same amount of prunings are made on each occasion.
(ii) If soil fertility is assessed indirectly on the basis of agricultural crop yield, then an economic analysis (with revenue, for example, as the dependent variable) that combines both crop and animal productivity will be desirable. The analysis can also be performed separately for crop and livestock components. In the case of crop analysis, such separation will allow assessment of the effects on crop yield of removing prunings.
(iii) A block design is appropriate especially if trials are carried out on farmers plots, in which case a farmer's plot may be considered as a replicate. If several species are being evaluated a splitplot arrangement should be considered.
Treatment: This type of experiment could involve 3 species and 23 factorial arrangement in a block design.
Factors: 
Species A at 0, 1 levels 

Species B at 0, 1 levels 

Species C at 0, 1 levels 
Treatment Combinations:
A0B0C0 
= 
(1) 
control 
A1B0C0 
= 
(a) 

A1B0C0 
= 
(ac) 

A1B1C0 
= 
(ab) 

A0B0C1 
= 
(c) 

A0B1C1 
= 
(bc) 

A0B1C0 
= 
(b) 

A1B1C1 
= 
(abc) 

Procedure: Randomize treatment combinations in blocks, and analyze results with 7 degrees of freedom (df) per treatment. The treatment source of variation is further split into:
Factors A, B. C, each with 1 df
Interactions AB, AC, BC, each with 1 df
Interaction ABC with 1 df
Treatments: A single tree species is investigated at different densities.
Tree density A
Tree density B
Tree density C
Tree density D
Tree density E
Design considerations are as for Example 6. Given more than one species, a splitplot may be considered.
Treatments:
A  No fodder removed
B  20% fodder removed
C  40% fodder removed
D  60% fodder removed
E  80% fodder removed
F  100%. fodder removed
Design: Any of the basic designs may be used, depending on land availability.
Notes:
(i) Observe crop yield periodically and analyze according to design used. A block design or an ordinary randomized design could be appropriate.
(ii) Alternatively, the trend in the effect of fodder reduction on crop yield can be investigated using single degree of freedom orthogonal polynomials. The relevant data would be percentage losses in yield from the previous cropping season, during which no fodder was removed from any plot.
(iii) During analysis, check whether the trend in percentage losses over increasing removal of fodder is linear, quadratic, etc.
Treatments: Three factors are involved:
Factor H Hedgerow shrubs at 5 levels
H1: Acioa
H2: Alchornea
H3: Gliricidia
H4: Leucaena
H5: NIL (as control)
Factor S Interhedgerow spacing at 2 levels
S1: Spacing 2m
S2: Spacing 4m
Factor F Fertilization at 2 levels
F1: 452020 NPK kg/ha
F2: 904040 NPK kg/ha
Intercrop: Maize
Observations: Incident solar radiation of maize leaves at known height
(i) Height at which solar radiation values are taken (Hc)
(ii) Corresponding shrub height at time of solar observation (Hh)
(iii) Height/distance index ({Hh  Hc}/S)
(iv) Crop (maize) yield.
Design: Splitplot with Factor H in main plot, and a crossed (factorial) combination of Factor S and Factor F in the 4 subplots.
Analyses:
(i) Analysis of variance with the splitplot breakdown
(ii) Linear or nonlinear regression relating:
 % incident light to height/distance index
 Crop yield to % incident light
 Crop yield to dry pruning biomas
Treatments: Three factors are involved:
Factor S Shrub species at 4 levels
S1: Sesbania
S2: Calliandra
S3: Leucaena
S4: Markhamia
Factor L Liming at 2 levels
L1: 0 t/ha
L2: 10/ha
Factor M Manure at 3 levels
M 1: 0 t/ha
M2: 5 t/ha and M3: 10t/ha
Design: Splitsplit plot arrangement. Factor S is the main plot, with each plots divided into 2 subplots. Three months after establishment the sub plots received the 2 levels of Factor L respectively. Each subplot is further divided into 3(subsub) plots to which levels of Factor M are allocated randomly.
Observations:
(i) Height growth of shrubs at predetermined intervals
(ii) A derived variable known as the Lime Response Index (LRI)
Definition: _{}Where HML (8) = the mean height of shrubs receiving liming at 8 months after planting (MAP)
HWL (8) = the mean height of shrubs without liming at 8 MAP
HWL (3) = the mean height of shrubs without liming at 3 MAP.
(Source: Yamoah, Grosz and Nizeyimana, 1989)
Treatments: Cropping systems at 5 levels:
C1: Continuous cropping without trees (control)
C2: Continuous cropping in Leucaena alleys
C3: Grazed tallow/cropping rotation in Leucaena alleys
C4: Cropping/grazed fallow rotation in Leucaena alleys
C5: Continuous alley grazing in Leucaena alleys
Duration of experiment is 4 years, with rotation in C3 and C4 effected every 2 years
Intercrop: Maize
Design: Randomized block design
Observations:
(i) Chemical analyses of soil samples at beginning of trial and before each first season crop(ii) Dry matter and nitrogen content values from prunings of Leucaena hedgerows
(iii) Maize crop yield
(Source: AttaKrah, 1990)
Treatments: Diet supplement at 4 levels
D1: 200 g dry matter (DM)/head/day of mixed (1: 1 w/w ratio) of Leucaena and Gliricidia forageD2: 400 g DM/head/day
D3: 800 g DM/head/day
D4: 1200 g DM/head/day
Design: Completely randomized, with 40 pregnant West African Dwarf sheep randomly allocated to the 4 diet supplementations (10 pens per diet treatment). All animals receive ad libitum chopped Panicum maximum grass plus 50 g of sundried cassava peel as basal diet Each pen also has 5 kg of wood shavings spread on top of litter at 4 weeks.
Observation: Random samples of wood shavings, analyzed for N at 2, 4, and 5 weeks after the sheep have been placed in the pens
(Source: Cobbina, AttaKrah, and Kang, 1989)
Treatments: Two factors involved
Factor D 5 levels of diet supplement (as in example 15) for 5 weekperiod.
D1: Pen with 200 g/DM/Head/day
D2: Pen with 400 g/DM/head/day
D3: Pen with 800 g/DM/head/day
D4: Pen with 1200 g/DM/head/day
D5: Control (raw wood shavings)
Factor S Manure rates at 3 levels
S1: 4 g/kg soil
S2: 8 g/kg soil
S3: 12 g/kg soil
Design: Completely Randomized (CRD) or Randomized Complete Block (RCB) design with 5 x 3 factorial arrangment of treatments. Each treatment is replicated 3 times. The choice as to CRD or RCB depends on the arrangement of pots in the greenhouse. In each pot, maize is planted.
Observations:
(i) Maize shoot dry matter yield
(ii) Soil chemical analyses (pre and posttrial)
(Source: Cobbina, AttaKrah and Kang, 1989)
Treatment: Hedgerow species at 5 levels
S1: Leucaena leucocephala
S2: Gliricidia sepium
S3: Alchorneaa cordifolia
S4: Acioa barteri
S5: Control (no hedgerow species)
Intercrop: Sequential cropping of maize (main season crop) and cowpea (minor season crop)
Design: Randomised complete block with 3 replications (blocks). Inter hedgerow spacing is 4 m.
Observations:
(i) Crop yield
(ii) Soil physical properties (bulk density, pore size, water infiltration etc.) at predetermined intervals.
(Source: Hulugalle and Kang, 1990)
Treatments: Alley shrubs at 3 levels
S1: Leucaena leucocephala var K28
S2: Gliricidia sepium local variety
S3: No alley shrub (control)
Design: Eight farmers plots are selected within a known area. Each farm constitutes a block, with 3 plots receiving randomly either S1, S2, or S3 This results in a completely randomized design with 8 replications (the replicates being the farmers)
Observations
(i) Soil chemical characters based on sample from each plot (before and during trial)
(ii) Shrub height growth at predetermined periods.
(Source: Cobbina, Kang and AttaKrah, 1989)
Treatments: Treatments comprise combinations of two factors as follows:
Factor N  Rate of N applications at 3 levels
N1: 0 N
N2: 45 N in kg/ha
N3: 90 N in kg/ha
Factor P  Prunings management at 2 levels
P1: Prunings removed (PR)
P2: Prunings retained (+PR)
Intercrop: Maize
Design:
(i) In Layout 1, randomization is such that all N levels are in each row. The +PR or PR factor is randomly allocated such that in the third row no combination from the first row is repeated! This arrangement is not recommended.(ii) In Layout 2, randomization of the N levels is made on row plots which have either +PR or PR. This is a splitplot arrangement with PR in main plots and N in the subplots. This design is preferred to Layout 1.
(iii) In Layout 3, treatment arrangement is factorial (3 x 2). The design is simply the Randomized Block, and is much preferred toLayout 1 and Layout 3 Its preference over Layout 2 is due to the fact that both PR and N levels can be evaluated equally.(V = the middle row which is planted with same maize crop as in the alleys but receives no treatment)
Observations:
(i) Crop yield,
(ii) Soil chemical content
(Source: Gichuru and Kang, 1989)
Treatment: Species combinations as follows:
Acioa barteri 
Leucaena leucocephala  
T1: 
r 
0% 
T2: 
0% 
100% 
T3: 
25% 
75% 
T4: 
50% 
50% 
T5: 
75% 
25% 
T6: 
100% 
0% 
Design: Randomized Complete Block plots are split at the cropping stage into 2 equal parts and receive 0 and 60 kg N/ha. This later modification changes the design to a splitplot.
(Source: Siaw, Kang, and Okali  In press).
AttaKrah A.N. 1990. Alley farming with Leucaena: Effect of short grazed fallows on soil fertility and crop yields. Exp. Agric 26, 110.
Cobbina J., AttaKrah A.N. and Kang B.T . 1989. Leguminous browse manure. supplementation effect on the agronomic value of sheep and goat Biological Agriculture and Horticulture, 6, 115121.
Cobbina J., Kang B.T. and AttaKrah A.N. 1989. Effect of soil fertility on early growth of Leucaena and Gliricidia in alley farms. Agroforestry Systems 8, 157164.
Hulugale N.R. and Kang B.T. 1990. Effect of hedgerow species in alley cropping systems on surface soil physical properties of an Oxic Paleustalf in southwestern Nigeria. 301307.
Gichuru M.P. and Kang B.T. 1989. Calliandra calothrysus (Meissn.) in an alley cropping system with sequentially cropped maize and cowpea in southwestern Nigeria. Agroforestry Systems 9, 19 1 203
Siaw D.E.K.A., Kang B.T. and Okali D.U.U. 1991. Alley cropping with Leucaena leucocephala and Acioa barteri. (In press).
Yamoah C., Grosz R. & Nizeyimana E. 1989. Early growth of alley shrubs in the highland region of Rwanda. Agroforestry Systems, 9, 171184.