Factors Affecting Distribution of Cattle in India

Factors Affecting
Distribution of
Cattle in India
Dr. E. Dayal
The problem of surplus cattle
population in India has been widely
discussed. The low yield of milk
and the roaming cattle everywhere,
even in the cities, give an idea of
surplus cattle population, and have
led to the general izations made
about the irrational wastefulness of
the large cattle numbers. However,
the usefulness of cattle in India
should not be judged only in terms
of milk yield, because cattle
contribute to the prosperity of the
farmer in more important ways than
by providing milk. The basic
question of why the large cattle
population is felt to be necessary
must be examined in the light of
various uses to which cattle are
put, and other factors that affect
the size of the cattle population.
Some .such factors have been
mentioned by earlier writers on the
topic,', but no attempt has been
made to test their ability to explain
the size of the cattle population and
their influence on the distribution
pattern . Also, the problem has been
dealt with only at the national level,
even though the problem of cattle
numbers is more a regional
problem in that many decisions are
affecting economic matters are
taken at the regional state level. In
this paper an attempt has been
made to identify the factors that
seem to affect the size and
distribution
of
the
cattle
population, and through the
application of simple and stepwise
multiple
regression
of
the
statistical significance of selected
factors has been tested.
The Variables
Dr. Dayal is a visiting professor of the
University of Minnesota . His normal position
is Lecturer in Geography in the University of
Wollongang, New South Wales, Australia.
The selection of variables that
appear related to the distribution
of cattle in India has had to be
based upon evidence drawn from
7
existing literature on the subject
and some intuitive reasoning. Nine
independent
variables
were
selected to explain the spatial
variations in cattle density, which is
the dependent variable. The data for
all variables was collected for the
20 states of Indi'a from the
Statistical Abstracts of the Indian
Union.' a A brief description of
each variable is given below:
Y Density of Cattle Units =
Number of Cattle units/Net Sown
Area
For the numerator in this index all
bovine population, i.e., zebu cattle,
buffaloes, and youngstock of both
were converted into comparable
standard units on the basis of food
requirements, using a conversion
factor
developed
by
the
Government of India (133.3 zebu
cattle = 100 buffaloes, 33.3 zebu
cattle
= 100 zebu cattle
youngstock, 50 zebu cattle
100
buffalo youngstock).3 This approach
gives comparable density figures
and provides a fair idea of the
pressure of cattle population on the
agricultural land resources.
=
X1 Density of Net Sown Area
Sown ArealGeograph ical Area
= Net
This index indicates the importance
of agricultural activities in a region.
If cattle are mainly used as draft
animals, one would expect a higher
cattle density in the areas where
the density of net sown area is
high.
areas of India, in which man
benefits more from the presence of
bovines, and therefore there is a
good deal of accord in the
distribution patters of human and
cattle populations .' The bovines
besides being used mainly as draft
animals serve men in several small
ways, and to that extent perform
multiple functions in the rural
economy. The spatial relationship
between the cattle and rural
population has been mentioned by
Spate and Learmonth , who said
"By and large the cattle being
essentially working rather than food
animals are where men are .... '"
X3 Density of Rice Cultivation
Rice Acreage/Net Sown Area
=
Of all the field crops grown in India,
field preparation is most elaborate
for rice. As most of the work is
done through draught animals, it
appears quite sound to assume that
there might be direct spatial
relationship between cattle density
and density of rice cultivation.
X4 Density of Foodgrain Acreage =
Area Under all Foodgrain Crops
Net Sown Area
Raj has mentioned that foodgrain
acreage is spatially related to cattle
density because draft requirement
for foodgrain crops are more than
for non-food crops." It is important
to include this index in the analysis
as foodgrain crops dominate the
agricultural scene in India.
Intensity of Agriculture
Area Under Multiple Cropping
Net Sown Area
There appears to be a symbiotic
relationship between bovine and
human populations in the rural
This index may be spatially related
to cattle denSity, because in the
area where agricultural intensity is
8
X5
=
X2 Rural Population Density =
Rural Population/Net Sown Area
high obviously more cattle will be
in demand throughout the year for
various agricultural and other
operations. It is generally believed
that intensity of cultivation is very
high in rice areas , and that
therefore there is considerable
double cropping. This is not quite
true in India. The intensive use of
labour in rice farming is quite
different from agricultural intensity
as defined here. Thus, there is little
spatial accord in the distribution of
X3 and X5 in India.7
X6 Density of Pasture and Grazing
Land
Area Under Pasture and Grazing
Geographical Area
=
This index was intuitively selected
as a variable, because common
sense reasoning suggests that
there will be higher density of
cattle in the regions where there is
more pasture and grazing land.
X7 Density of Well Irrigated Area =
Well Irrigated Area/Net Sown Area
About nine million hectares of
agricultural land is irrigated by
wells in India, and cattle are the
main source of power for drawing
water from the wells. This index,
therefore, appears a suitable
independent variable for inclusion
in the model.
X8 Density of Hindus and Jains =
Number of Hindus and Jains
Total Population
There is widespread bel ief that the
large cattle population in India is
the result o f Hindu doctrine of
ahimsa.' The Hindus, who form
the vast majority of the total
population in most states , and
Jains have strong objection to the
killing of the cow and to eating
beef. Therefore, it is assumed that
the states in which the proportion
of Hindus and Jains to the total
population is high, will also have
high cattle densities.
=
X9 Availability of Domestic Fuel
Firewood
Production/Rural
Population
There are about three million cattle
in India which are decrepit and
good for neither milk nor work,
and which may be regarded
useless. However, their existence
has been justified on the ground
that they provide dung which is the
only source of fuel in some parts of
India. According to Majumdar about
300,000,000 tons of dung is used
annually as fuel.' The peasants like
to keep even these useless animals
if they have no other source of fuel ,
which is often the case in several
areas in the Northern Plains of
India. Such a line of argument
suggests some negative correlation
between
fuel
supply
from
alternative sources and cattle
density. As firewood is the only
other fuel used in villages ,
domestic fuel supply has been
measured as firewood production
per 100 rural population .
THE ANALYSIS
it is hypothesized that the
distribution of cattle density in
India (Y) is positively related to the
distribution patterns of the density
of net sown area (X1) , rural
population (X2), ric.e acreage (X3),
all foodgrain acreage (X4), intensity
of agriculture (X5), pasture and
grazing land (X6), well irrigated area
(X7), Hindus and Jains (X8), and
negatively to firewood production
(X9). The data on all these variables
was squareroot transformed .
9
TABLE 1: CORRELATION MATRIX (Imput Variables)
y
X1
X2
X3
y
X1
X2
X3
1.000
-0.535 * *
0 .433* *
-0.482 * •
1.000
0.715 "
0.355
0.324
-0.251
1.000
0.337
0.027
-0.220
1.000
0.510' "
0.553"
1.000
0.309
1.000
1.000
1.000
X4
X5
0.333
0.730*
0.629 *
-0.609 "
X4
X5
X6
X7
X8
X9
X8
X9
X6
X7
0.542* *
0.332
-0.201
-0.032
-0301
-0.009
-0.236
0.408
-0.318
0.593"
0.001
-0.566 "
-0.487 " "
-0.497* "
0.051
-0.200
-0.336
0.017
0.208
-0.450 " "
-0.222
0.339
0.033
0.154
-0.239
1.000
0.175
-0.498 " "
-0.103
1.000
*Significant at 99 percent level of confidence
** Significant at 95 percent level of confiden·ce
o
.-
TABLE 2
STEPS
1. X4
2. X5
3. X1
INDEPENDENT
VARIABLES
Percentage of
Foodgrain Area
Intensity of
Agriculture
Percentage of
Net Sown Area
4. X8 Hindus and Jains as
Percentage of Total
Population
5. X6 Percentage of Area
Under Pasture
6. X2 Density of Rural
Population
7. X3 Percentage of
Rice Acreage
8. X7 Well Irrigated
Area as Percentage
of Net Sown Area
9. X9 Firewood Production
per 100 of Rural
Population
INCREASE PARTIAL OVERALL
in R2
F-value
F-value
R
R2
0.730
0.533
0.5335
20.586* *
20.586* *
0.756
0.572
0.0387
3.675*
11 .369 **
0.789
0.622
0.0500
4.233 *
8.783 * *
0.813
0.661
0.0390
4.712 *
7.318 * *
0.835
0.697
0.0363
2.661
4.234*
0.843
0.712
0.0141
1.269
4.135*
0.856
0.733
0.0215
1.931
4.007 *
0.863
0.745
0.0117
1.008
3.272
0.865
0.748
0.0031
0.332
2.214
** Significant at 99% level
*Significant at 95% level
I
i
11
CATTLE DENSITY
(By States, J Data)
Per 100 Acre of
Net Sown Area
I~tttl
>50 '00
Fj}}]
38 ·10 - 49 ·99
Ed
27·10- 38·09
o
Okms
!
1000
!
Fig. 1. Distribution of Cattle Un its in India.
12
*
15 '10 - 27·09
Data not available
RESIDUALS FROM
MULTIPLE REGRESSION
(Y-Yc)/Syc
Standardized Values
~ 1·00 to 1·85
§
o
[3]
Okms
1000
L . . '_ _ _ _ _ _ _,
*
0 to 0 ·99
-0,99 to -0·01
-1 ,00 to -1 ,76
Data not available
Fig . 2. Pattern of residuals from regression
13
5
The results of simple correlation
analyses show that the density of
area under foodgrains is the most
important variable in explaining the
variation of cattle density. (Table 1).
Next in order are the intensity of
agriculture, density of net sown
area, and density of rural
population ,
which
are
all
statistically significantly related to
the distribution of cattle density.
However, the association between
net sown area and cattle density is
inverse rather than direct as was
postulated.
A more careful
examination seems to support the
results of the correlation analysis.
In regions where the area sown to
crops relative to total area is large,
the methods of farming are often
extensive, because the pressure of
rural population on cultivated land
is relatively low, as for example in
the southern section of the Punjab
Plains , Gujarat , west central
Deccan Plateau and Rajasthan . In
such areas the demand for cattle
for agricultural work will be less per
acre of cultivated land because of
the extensive nature of farming .
This is also supported by the
negative correlation between
density of net sown area on the one
hand and density of rural population and intensity of agriculture
on the other (Table 1). Also, the
overall
negative
relationship
between cattle density and net
sown area may be affected by some
states where the proportion of the
net sown area is small, because of
the large extent of hilly and
forrested tracts, leading to a higher
density of cattle on net sown area
although the total cattle population
is small , as for example in
Kashmir,
Himachal
Pradesh ,
Assam , Nagaland, Manipur, Tirpura,
and Madhya Pradesh.
14
Of the remaining five in dependent variables the direction
of association for three , rice
acreage, area under well irrigation,
and firewood supply, are the same
as were postulated, but the
correlation coefficients are weak
and statistically not significant. The
relationship between cattle density
and the percentage of Hindus and
Jains
is
negat i ve
and
not
significant. This is an important
result and shows clearly that the
large cattle population in India does
not reflect the Hindu attitude
against cattle slaughter and the
eating of beef, thus supporting the
views of Harris'o and Raj. " For
example the largest concentrations
of non-Hindu population are in the
states of Kashmir, West Bengal,
Assam, Tirpura, Manipur, and
Kerala, but it is also in these states
that
there
is
a
marked
concentration of cattle population
with the exception of the last. The
relationship between cattle density
and density of pasture and grazing
land is also negative. This may be
because in India animal raising is
not an organized commercial
enterprise. Cattle are not kept for
milk or meat but for work, and are
largely fed on crop residues. Also,
the area which is cultivatable but
not used for cultivation for some
reason is classified as pasture and
grazing land . Greater extents of
such land are found in the areas
which are less important for
agriculture. Thus, the negative
relationship brought out by the
correlation analysis appears to be
valid.
But
the
Correlation
coefficient is weak and not
sign ificant.
All the nine variables have been
listed, and the results of simple
correlation analyses substantiate
three hypotheses, modify one, and
do not support five (Table 1). The
proportion of the variation in cattle
density accounted for by individual
independent variables is generally
small. Only the density of foodgrain
acreage (X4) accounts for more
than fifty percent of the variation in
Y. It is however, inevitable that
higher values of R' might be
obtained if the variables were
treated simultaneously in relation
to cattle density. This appears
closer to the real world situation,
where several factors operate
together to influence a situation.
The five independent variables
proved insignificant in the simple
correlation analyses are retained in
the multiple model, assuming that
they might contribute significantly
to the value of R2 when treated with
other variables.
satisfactory. At each step the
significance of the last variable
added is tested by a partial F-test,
and an overall F-test indicates the
significance of the multiple
regression equation formed at that
step." After the fourth step the new
variable added ceases to make any
significant contribution to the
explanation of the variation in Y,
but overall regression equation
continues to be significant up to
the seventh step (Table 2). Logically
only those independent variables
which
make
a
significant
contribution to the explanation of
the variation in Y should be
included in the refined multiple
regression equation . Thus, the
stepwise regression model yields
four independent variables for
inclusion in the refined multiple
regression equation :
Multiple Correlation and
Regression Analysis
14.6771 - 2.9140x1 +
Yc
1.6772X4 + 2.2495X5 - 1.7589X8
The coefficient of multiple
correlation based on all the nine
independent variable (R = 0.865) is
statistically highly significant, and
accounts for 75 percent of the
variation in cattle density. However,
it is desirable to find out the
minimum number of variables that
can account for maximum variation .
This has been achieved through the
development of a stepwise multiple
regression model. The computer
programme (BMD02R) used in this
analysis first picks up the
independent
variable
most
correlated with Y, and finds the first
order linear regression equation.
The next variable added , out of the
remaining independent variables, is
the one whose partial correlation
coefficient with Y is the highest.
This process continues until the
regression equation is most
Standard Errors of
Coefficients (0.743) (0.644) (0.690) (0.669)
=
R = 0.813
R2
0.661
=
The regression model provides a
statistical description of the spatial
difference between cattle density
and a set of variables which
describe the place characteristics.
Density of foodgrain acreage (X4),
intensity of agriculture (X5), density
of net sown (X1) are most strongly
associated with the density of
cattle (Y). The variable added at the
fourth step is the proportion of
Hindus and Jains to the total
population of Hindus and Jains to
the total population (X8), which has
a very low negative correlation with
Y (Table 1), but it has a higher
partial correlation with Y than any
15
of the remaining variables at the
fourth step. This four variable
regression model, based on the
density of net sown area, density of
foodgrain acreage, intensity of
agriculture and proportion Hindus
and Jains, explains two-thirds of
the variation in cattle density,
which is quite satisfactory when
compared with the results generally
achieved in spatial regressions.
The only variable which is
significantly correlated with Y and
is not included in the refined
multiple regression equation is the
density of rural population . This is
because rural population is also
sign ificantly correlated with the
density of foodgrain acreage (X4)
and therefore its explanatory power
has been considerably reduced by
the inclusion of X4 at the first step.
The remaining four variables,
namely, density of rice acreage (X3),
density of pasture and grazing land
(X6), density of well irrigated area
(X7), and firewood production (X9),
which were proved insignificant in
the simple correlation analysis also
remain insignificant in the multiple
regression model.
A
re-examination
of
the
hypotheses associated with these
variables
suggests
some
explanations for the weak or
insignificant relationship between
them and cattle density. The weak
relationship between firewood
production and cattle density may
be because cow dung is not used
as domestic fuel all over India,
particularly in the southern states
and in the hilly districts even in
north India. Harris appears to have
made an over statement when he
said that dung was the main source
of domestic fuel in India. 13 The
relationship between pastures and
cattle is low because pastures and
16
grazing lands are found more in the
areas less important to agriculture.
In such areas there is no need for
keeping large number of cattle as
they are mainly required for
agricultural work. The cattle
in
India live on scrub, organic waste,
like stalks of rice, wheat, maize,
and other crops, and even consume
large quantity of waste paper." The
poor relationship between rice
density and cattle is because rice
alone perhaps does not adequately
represent the cropping pattern even
in some main rice areas. The
relationship improves when total
foodgrain acreage is taken. This is
also
supported
by
low
intercorrelation between density of
rice acreage and density of total
foodgrain acreage.
The Analysis of Residuals
A comparison of the residual
map with that showing the density
of cattle in terms of standardized
values shows that high negative
residuals from regression occur in
Punjab, Maharashtra, and Andhra
Pradesh states, where the density
of cattle units is low (Figures 1 and
2). Therefore, in these three states
there is an over prediction, which
means that the model over
estimates the importance of
selected independent variables in
explaining cattle density in these
areas. The high positive residuals
occur in Kashmir, Mysore, and
Madras states, where the cattle
density is high to medium. In these
three states the variables included
in the model explain only a part of
the actual density. In the two
groups of states involved in over
prediction or under prediction the
cropping pattern is not adequately
expressed by the density of all
foodgrain acreage. In these states
TABLE 3
WORKING CAPACITY OF DRAFT ANIMALS IN SELECTED AREAS
Regions
(District)
Punjab (Karnal)
Area ploughed
per animal per
day (8 hours
working day)
(Acres)
1.41
Dominant
_Crops
wheat , gram
Percentage of
farmers who felt
a shortage of
draft animals in
the peak season
56.6
1.27
wheat, bajra
gram
87.5
Eastern U.P.
(Varanasi and Allahabad)
0.54
wheat , rice
and millets
86.6
Bihar (Chapra)
0.32
rice, maize
62.7
Assam (Nowgong)
0.25
rice, jute
53.3
Western U.P.
(Delhi and Meerut)
Source: Compiled by the author, and is based on field survey.
17
l
commercial crops like tobacco,
cotton , groundnuts, and coconuts
are of much importance in the
cropping pattern . In the remaining
fourteen states the density of cattle
is well predicted by the model as
the values of the standardized
residuals for these states are less
than one (Figure 2). They form a
large contiguous area, spread over
almost the entire Northern Plains
the Northern Slopes of the Decca~
Plateau , and the Assam Valley.
These areas are the most important
agricultural areas of the country,
where there is a high concentration
of foodgrains acreage, double
cropping, and net sown area. The
model has taken good account of
the influence of these independent
variables in explaining cattle
density over this large area. The
pattern of residuals indicates no
obvious factor, in addition to those
already included in the model that
could improve predictablility. '
It is important to recognize the
limitations of the regression model
which operates in a deterministi~
framework,
and is therefore
descriptive and not analytic .
However, it does give some idea of
the relative importance of some of
the principal factors that seem to
affect the distribution of cattle
density in India.
CONCLUSION
The spatial association between
cattle density and a set of factors
has been tested using multiple
correlation and regression analysis
at the state scale. The analysis
seems to suggest that the
distribution of cattle in India is
fairly strongly related to four
environmental
factors ,
the
proportion of net sown area
agricultural intensity, proportion of
18
crop area sown to all foodgrains
and proportion of Hindus and Jain~
to total population , which were
distilled frorll an original nine.
The pattern of cattle density
shows a high concentration of
cattle in agriculturally important
areas, where a majority of the above
four variables is also well
represented, as for example the
Northern plains of India, the Assam
Valley, and some coastal areas.
Low densities of cattle occur in the
areas that are less important for
agriculture such as Rajasthan,
Gujarat, and much of western
Deccan Plateau, where the values
of the independent variable are also
generally low. The more important
agricultural areas in India largely
produce foodgrains to feed the
large local populations. Since
foodgrain production requires more
elaborate and more frequent
cultivation , there is greater demand
for draft animals, and farmers have
to keep more work animals than
their counterparts in less important
agricultural areas and in areas
where commercial crops dominate.
Information collected in a field
survey of some selected villages in
North India in 1975, for the purpose
of checking the resulta of this
analysis, shows that over much of
the Northern plains , where there is
largest concentration of cattle ,
there is often a shortage of cattle in
the peak farming season (Table 3).
Thus, it may be concluded that in
the areas where cattle density is
high there is really no surplus , and
that cattle are kept because there is
a need , that is, for economic rather
than any sentimental reasons. The
only way to reduce the cattle
numbers is to bring about a change
in agricultural technology. Perhaps
partial mechanization of some
agricultural operations, such as
ploughing with tractors, which will
not seriously affect the rural
employment situation, might help
reduce the numbers and improve
the quality of cattle.
References
Kumarappa, J. D. (ed.) The Cow in Our Economy.
Bhargava Press, Varanasi , India, 1957.
2. Government of India, Central Statistical Organization,
Statisticat Abstracts of the tndlan Union, Manager of
Publications, New Deihl , 1967 and 1968.
3. Government of India, Cabinet Secretariat, Tables and
Notes on Animal Husbandry, National Sample Survey,
Report No. 65, New Deih l, 1962.
1.
4.
5.
6.
HarriS, M., " Cultural Ecology of India's Sacred Cattle,"
Current Anthropology, 7, 1966, pp. 51-66.
Spate, O. H. K. and Learrnonth , A. T. A., India and
Pakistan, Methuen, London, 1967, p. 254.
Raj , K; N., " Investment In Livestock In Agrarian
Economies," Indian Economic Review , 4 (New Series),
1969, pp. 54-65.
Spate and Learrnonth, op. cit., pp. 235 and 245 (cf.
Figures 8.3 and 8.8).
8. Harris, M., op. cit., p. 51.
9. Majumdar, N. A., "Cow Dung as Manure," Economic
Weekly, 12, 1969, pp. 734·744.
10. Harris, M., op. cit., p. 64.
11 . Raj, K. N. , op. cit., p. 84.
7.
12. Draper, N. R. and Smith, H., Applied Regression
Analysis, John Wiley and Sons, New York, 1967, pp.
169·170.
13. Harris. M., op. cit., p. 53.
14. Perelman, M. A., " Sacred Cows," American Journal 01
Agricultural EconomiCS, 54, 1972, pp. 526-527.
19