Вы находитесь на странице: 1из 8

INTER-CROPPING EXPERIMENTS

Aloke Lahiri
I.A.S.R.I., Library Avenue, New Delhi-110 012.
1.
Introduction
Inter-cropping refers to growing of two (or more) crops simultaneously on the same piece
of land in separate rows and is different from mixed cropping where the mixture of seeds
of two or more crops within the same row are sown. Inter-cropping has several
advantages.
(i)
Increasing cropping intensity
(ii)
Diversification of crops.
(iii) Mitigating risks due to weather aberrations.
(iv)
Optimal use of basic resources, viz., moisture, light and nutrients
(Biological insurance)
(v)
Insect, pest and weed control
In view of the above, the Inter-cropping studies are designed to:
a) Identify crop combinations so that the yield of the base crops is not sacrificed,
b) Identify crop combinations so that the total production and/or revenue is maximized.
c) Identify the proper geometry of planting component crops and to
d) Evaluate the effect singly or in combinations of several factors, such as fertilizers,
geometry, plant population, germ-plasm etc.
1.1
A Brief Review Of Statistical Analysis
In Inter-cropping studies two or more crops are involved and their interdependence poses
serious problems in the appropriate statistical analysis of data. To study in detail the
agronomic aspects mentioned earlier, the statistical problems associated with the
planning and analysis of experiments are many and the opportunities for interesting
statistical work are exciting. In the past, a good number of attempts have been made by
various workers to identity suitable statistical designs suited to various types of Inter-crop
experiments and their methods of analysis. As regards the problem of design, it will not
be out of place to mention that they are not very different from those of sole cropping
[Mead and Riley, 1981]. The problem of designing of experiments has to be viewed
particularly keeping in view the evidence of greater variability in the Inter-cropping
experiments than in sole cropping. This therefore, calls for simple experiments with
small blocks. The number of levels of various factors included should also be minimum
to the extent possible.
In so far as the methodological problems for analysis of Inter-crop data are concerned,
considerable interest has recently been generated by advocating different approaches. It
is generally accepted that more than one analysis should be applied to Inter-crop data
[Mead and Stern 1979]. According to Mead and Riley [1981] in their comprehensive
review of available statistical methods for the analysis of data from Inter-cropping
experiments, there is no single, straightforward method that is universally appropriate.
Each experiment needs careful study.

Inter-Cropping Experiments

1.2
LER Method
According to Willey [1979], the most generally useful single index for expressing the
yield advantage is probably the Land Equivalent Ratio (LER), defined as the relative land
area required as sole crops to produce the same yields as Inter-cropping. Algebraically
LER can be expressed as:
LER = LA + LB = YA / SA + YB / SB
Where LA and LB are the LERs for individual crops, (called partial LERs), YA and YB
are the individual crop yield in Inter-cropping system and SA and SB are their yield as
sole crops (pure stand). The advantages of LER [Mead and Willey, 1980] are that it
provides a standardized basis so that crops can be added to form combined yield.
Various other indices have been suggested which are conversion of yield to money value,
total protein content or calories. Undoubtedly economic indices have advantages but
they have disadvantages that the monetary value, for example, is subject to market
conditions, which are by no means constant. Again calorific value or protein content may
appeal to a dietitian but it does not enter into the consideration of the peasant farmer, who
is the one to be persuaded [Pearce and Gilliver, 1978]. There are two limitations of LER:
1. It is independent of yield levels, being a ratio, large values arise not only when the
sole crop yields are small, but also when the Inter-crop yields are large and
2. The LER values, being the sum of the ratios of two normal variates, follow Cauchys
distribution and hence one of the assumptions of normality underlying the analysis of
variance fails.
However, for the second limitation, Oyejola and Mead [1982] have shown that there is
more than one way to generate LER values from plot wise yield data and that nonnormality is not serious, provided the mean yields of sole crops (over replication) are
taken for the purpose of standardization. Fisher [1977] favored for standardization within
each block, to reduce standard errors and also the skewness of the distribution of LERs.
However, since the LER calculation depends on the choice of Si, the standardization
yield, which can be selected in many ways [Oyejola and Mead 1982].
1.3
Bivariate Analysis of Variance Method
In this exact tests are available for comparison between means. However, unlike the
bivariate method, the LER is easy to interpret, in terms of practical advantage and
comparisons can be made between sole crops and Inter-crop yields. The use of a single
bivariate method of analysis for the yields x1 and x2 of the two crops was proposed by
Pearce and Gilliver [1978] following idea of Steel [1955]. The main idea is that when
two species are interplanted, the yields of the two crops will not in general be
independent. So the two variates cannot be dealt with separately but in conjunction. For
this a bivariate analysis is necessary. The fundamental assumption of the method is that
the correlation between the yields for the two crops is constant for all treatments which
may not be true in the exact sense[Mead and Riley, 1981]. Dear and Mead [1983]
described in detail the bivariate analysis technique for the presentation, analysis and
interpretation of data from Inter-cropping experiments. Mead and Riley [1981] advocated

614

Inter-Cropping Experiments

that to get as much information as possible, it seems sensible to use both the LER and the
bivariate method of analysis.
2.
Analytical Procedures
In order to explain the two methods, mentioned earlier, we have considered an
experiment on Inter-cropping conducted at the Hanumangarh (Rajasthan) center under
the All India Coordinated Agronomic Research Project (I.C.A.R.) in the year 1982-83.
The crops were Pigeon Pea (maincrop) and Green gram (Inter-crop). The experiment was
laid out in randomized block design with four replications. The number of treatments
were 15, which consisted of 3 sole crop treatments for the main crop, 3 for the Inter-crop
and 9 composite treatments of main and Inter-crop taken in the mixed stand forming
combination of 3 methods of planting pattern i.e. Normal, Paired row and Skip row
method of planting, 3 levels of fertilizer application to Inter-crop which are at 100%, 50%
and 25% of the recommended dose. The main crop was fertilized at the optimum level. In
all the 15 treatment combinations, plant population of the main crop and the Inter-crop
was kept at the optimum level.
2.1
Analysis By LER Method
As discussed in the previous section, it is clear that the yield of the companion crops are
not additive. The advantages and disadvantages of reducing the two variates to a single
variate such as monetary return, protein content etc., have been pointed out earlier. The
most acceptable single variate analysis of Inter-crop system is the LER.
The following illustration makes the LER calculation clear:
Treatment
Pigeon Pea(Sole)
Greengram (sole)
Pigeon Pea + Green gram
Pigeon Pea PLER(Partial)
Greengram PLER(Partial)

Yield(Q/Ha.)
30
10
22 + 6
22/30 = 0.73
6/ 10 = 0.60

LER =0.73 + 0.60 = 1.33


The LER of the system is 1.33. This means that 33 percent more land would be required
as sole crops to produce the same yields as Inter-cropping.
In this study, two methods which are applicable to the data were adopted. The Inter-crop
yields were standardized against sole crop yields at the same fertility level and methods
of application of fertilizer
(1) Obtained from the respective replication(Method -1) and
(2) Average over all replication(Method - 2)
The LERs through the two methods suggested above were worked out for each plot and
the analysis of variance was carried out.

615

Inter-Cropping Experiments

2.2
Analysis By Bivariate Analysis Of Variance Method
The rational behind the method is that when two crops are planed in the same plot in the
same season, their yield will not, in general, be independent. The main problem arises
from the possible correlation between two variates x1 and x2 (yield of the two component
crops). Therefore, univariate analysis of the two crops separately may not result in correct
inferences about different treatment effects.
Let T11, T22 be the sums of squares and T12 be the sum of product of the component
crops due to treatments, each with t degrees of freedom and E11,
E22 and E12 be the sum of squares and sum of products due to residual, each with e
degrees of freedom. Then, the Wilks criterion is given by
= (E11 E22 - E122 ) / [ ( E11 + T11)( E22 + T22) - ( E12 + T12)2 ]
and F- statistics is related to by
F= ( -1/2 - 1) e / t
= ( -1/2 - 1) (e-1) / t

for t >1
for t=1

This F- statistics is tested for significance at 2t and 2(e-1) degrees of freedom for t>1 and
on 2 and (e-1) degrees of freedom for t = 1.
Bivariate Analysis Of Variance Table

Source

d.f.

Replication
Treatment
Error

Total

SSx1

SPx1x2

SSx2

R1

R12

R2

(v-1) = t

T1

T12

T2

(v-1)(r-1)
=e

E1

E12

(r-1)

E2

Bivariate F

d.f.

(-1/2 -1) e/t 2t,2(e-1)


-

rv-1

2.2
Graphical Representation of Bivariate Data
This approach was advocated by Pearce and Gillivar [1978]. In this approach, the error
variances of the two variates x1 and x2 are V11 and V22 and their error covariance is V12 .
Each pair of observation (x1,x2) are transformed by making use of the transformation:
Y1 = x1 / v111/2 and y2 = (x2 -v12 x1 / v11) / ( v22 -v122 / v11)1/2

616

Inter-Cropping Experiments

These new variates (y1 and y2) have error variance equal to unity and covariance equal to
zero, i.e.; y1 and y2 are independent and can be plotted on the rectangular axes. At each
of the point, a circle of radius (2F/ n)1/2 indicates the confidence region for different
treatment means and a circle of radius 1/n1/2 indicates the standard error of a mean of n
observations, where F is the chosen percentage point of the F distribution on 2 and e
degrees of freedom.
References
Dear, K.B.G. and Mead, R. (1983). The use of bivariate analysis technique for the
presentation, analysis and interpretation of data. Technical Report No. 1, 1983.
University of Reading, UK
Fisher, N.M. (1977). Studies in mixed cropping - I. Experimental Agric. 13 ,177 - 184.
Lahiri, Aloke (1984). Statistical analysis of data from Inter-cropping experiments.
M.Sc. Thesis, P.G.School, I.A.R.I., New Delhi.
Mead, R. and Riley, J (1981). A review of statistical ideas relevant to Inter-cropping
research. Jour. Royal Statistical Soc., A144, 462-509.
Mead, R. and Stern, R.D. (1979). A statistical consideration in experiments to
investigate Inter-cropping - Proc. of Int. workshop on Inter-cropping, ICRISAT,
Hyderabad, 263 - 276.
Mead, R. and Willey, R. W (1980). The concept of a Land Equivalent Ratio and
advantages in yield from Inter-cropping. Experimental Agric.,16, 217- 218.
Oyejola, B.A. and Mead, R. (1982). Statistical assessment of different ways of
calculating Land Equivalent Ratio (LER) - Methodology. Experimental Agric. 18,
125 - 138.
Pearce, S.C. and Gilliver, (1978). The statistical analysis of data from Inter-cropping
Experiments. Jour of Agric. Sc.,91, 625 30.
Steel, R.G.D. (1955). An analysis of perennial crop data. Biometrics, 11, 201-212.
Willey, R.W. (1979 a and b). Inter-cropping, its importance and research needs. Field
crop Abs,32 1-10 and 73 -85.

617

Inter-Cropping Experiments

Center: HANUMANGARH

Year: 1982

TABLE 1. Average grain yield (Q/ha) and LERs for different treatments corresponding
to different system of standardization for Inter-cropping of Arhar and Moong.
Treatment
T1
T2
T3
T4
T5
T6
T7
T8
T9
T10
T11
T12
T13
T14
T15

Grain Yield(Q/Ha.)
LER
Moong
Method-I
9.20
8.55
8.37
6.40
1.70
5.07
1.46
4.46
1.41
4.96
1.57
4.64
1.53
5.05
1.52
6.83
1.61
6.72
1.66
6.50
1.59

Arhar
19.77
18.09
19.04
19.79
17.21
17.17
18.46
17.58
16.59
16.75
16.42
15.58

Method-II
1.69
1.46
1.40
1.56
1.51
1.52
1.62
1.65
1.59

TABLE 2. ANOVA of LER for different methods


Source
variation

of d.f.

Method
I

Replication
Treatment
M
F
MxF
Error

3
8
2
2
4
24

S.S.
0.10973
1.27686
1.06904
0.09016
0.11766
0.47657

Total

35

0.86316

M.S.S.
0.03658
0.03461
0.03452
0.04508
0.02941
0.01986

Method
II
F
1.84
1.74
1.74
2.27
1.48

S.S.
0.04495
0.28379
0.07740
0.09174
0.11464
0.29114
0.61989

618

M.S.S.
0.01498
0.03547
0.03870
0.04587
0.02866
0.01213

F
1.23
2.92*
3.19
3.78*
2.36

Inter-Cropping Experiments

Center: HANUMANGARH Year:1982


TABLE 3. Bivariate analysis of variance for yield (kg/plot) of Arhar (X1) and Moong
(X2)
Source
of
variation

d.f.

Replication
Treatment
M
F
MxF
Error

3
8
2
2
4
24

Total

35

S.S (X1)

0.0412
17.5472**
7.4835**
7.9805**
2.0832
6.8177
24.7661

S.P.(X1, X2)

S.S. (X2)

Bivariate

d.f.

0.0455
-2.3714
-6.6205
3.1540
1.0951
0.1769

1.0334
10.6939**
7.6431**
1.2741
1.7767
6.1089

6.35**
11.17**
6.43**
1.70

16,46
4,46
4,46
8,46

- 2.1490

17.8363

Residual Correlation = +0.027

TABLE 4. Bivariate analysis of variance for transformed yields (kg/plot) for Arhar (Y1)
and Moong (Y2)
Y1 = 1.8762 X 1

and

Y2 = 1.9828 X2 - 0.05145 X1

Source of
Variation

d.f

S.S(Y1)

S.P. (Y1,Y2)

S.S. (Y2)

Replication
Treatment
M
F
MxF
Error

3
8
2
2
4
24

1.4060
61.7760**
26.3440**
28.0730**
7.3590**
23.9880

0.1295
-10.5234
-25.3713
10.9649
3.8830
0.0383

4.0687
42.5652**
31.4249**
4.3889*
6.7514**
24.0262

Total

35

87.1700

-10.3556

70.6601

619

Inter-Cropping Experiments

SAS Program For Bivariate Analysis of Data


data incrop;
input rep method level y1 y2;
cards;
1 1 1 11.6 3.4
1 1 2 9.9 2.35
1 1 3 10.0 2.15
1 2 1 11.7 3.5
1 2 2 11.3 3.0
1 2 3 10.12 3.25
1 3 1 10.5 4.09
1 3 2 10.10 3.02
1 3 3 9.2 4.35
2 1 1 12.3 3.81
2 1 2 9.8 2.65
2 1 3 10.1 2.75
2 2 1 10.9 2.95
2 2 2 10.7 2.15
2 2 3 9.9 2.78
2 3 1 10.30 3.40
2 3 2 10.40 4.55
2 3 3 8.7 3.40
3 1 1 11.7 3.75
3 1 2 11.3 3.0
3 1 3 10.7 2.8
3 2 1 11.6 2.65
3 2 2 10.1 2.9
3 2 3 9.4 3.4
3 3 1 9.9 4.8
3 3 2 9.5 4.5
3 3 3 9.7 4.8
4 1 1 11.9 4.4
4 1 2 10.3 3.56
4 1 3 10.4 3.0
4 2 1 10.1 2.8
4 2 2 10.1 3.1
4 2 3 10.4 2.7
4 3 1 9.5 4.1
4 3 2 9.4 4.06
4 3 3 9.8 3.06

;
proc print;
proc glm;
class rep method level;
model y1 y2= rep method level method*level/ss2;
manova h = rep method level method*level/printe printh;
means method level/lsd;
run;

620

Вам также может понравиться