CORRELATION ANALYSIS iple (orelalou oelleiat)

76
SEGTTDN three types of correlation coc fficient, viz., genotypic, phenotypic and environmental
(see later) are commonly used.

DETERMINATION OFYIELD COMPONENTS 5. Applications: In plant brecding, estinates of correlation coefficient are use ful in
determining yield components whicl1 can be used for genetic improvement of yíeld.
TYPES OF CORRELATIONS
Correlations are of three typcs, viz., simple, partial and multiple. Simple comelation coefficients
again are of three types, viz., genotypie. phenotvpic and environmental. Abrief classification
divediou o MKelelioiwup
otoal
of correlations is presented in Table 9.1.
TABLE 9.1. Classificatior of Correlations.
SNo:ype of Correlation| Denoted as melfpitiou
Sinple corelation Y Association between two variables.
Genotypic Inherent association between two variables.
Phenotypic Tph 'ph Observed association between two variables.

Correlaiion Anaiysis 2.
Environmental Ye,
Partial correlation Y.3 Ti123
Association: between two variables due to
environmental influence.
The correlation between two variables eliminating
icieeut o covelaliouw ay dit the effect of third variable.
Fitst order TË23 Calculated from simple correlations.
INTRODUCTION Caloulated from first order partial correlations.
Second order I234
The term coefficient of correlation was first used by Karl Pearson an 1902) The statistics 3. Multiple correlation Joint influence of two or more independent
which neasure the degree and direction of association between two or more variables is variables on a dependent variable.
known as correlation. It 1s denoted by rA positive value of)indicates that the changes in
two variables are in the same direction í.e. high values of one variable are associated with
high values of other and vIce 'ersa. When r is negative, the changes are in opposite directions SMPLE CORRELATION T6tal Conelaiou Oordu tornelaiu toe.
i.e. higi values of onc variable are associated witlh low values of othe. The important Simple Correlation refers to theassociationbetween two variables. It is also known as total
coITclation or zerU oider corelation cocficient. Main features_of simple correlation are
propertics cf correlation coefficient are presented as follow's. presented as follows.
1. Unit and Values : It i<independent)of the unit of measurement. Its value lies 1. It involves two variables.
between 1 and1
2. It is denoted a ri2)
2. Infornation Obtained : It ncasures the degrec and dircction of association bctween 3. Itgnores effecs of othet independent variables.
IWo or more variables. 4 I: is esiuuatcd from variances and co-variances (2')
3. Analysis : Analysis of correlations coefficients involves second order statistics. It .5. Its valuc is always lower than multiple correlatjon.
is basecd on estimatespf variances and co-variances.
GAt is of three types, viz., genotypic, phenotypic and environmental.
4. Types: Correlations are of three types. viz., simple, partialand multiple (sce later). 7. It can be calculated from un-replicated data also.
Sinple. partial and multiple correlations can be estimated both replicated und
(un-replicatedlata. Simple correlation coefficients are of the types. v. genolypie, (boly kapliçaad t Uu-replicated )
phenoyDiC and environmental. Estimation of these three types of correlation PARTIAL CcORRELATION NeCoreai)
coe fficients is possible from replicated clata only. In plant brceding, estimates of Partial Correlation refers to the correlation between two variabies eliminating the eflect ol
Ku'rd vouiable
 761
BIOMETRICAL TECHNIQUES IN PLANT BREEDING cORRELATION ANALYSIS 77
thìrd variable. It is denoted as rn It is also
partial correlation are presented below. known ag nct coclation Main features of Genotypic Correlation
L. It involves threc or four The inherent or heritáble association betweenfwo variables is referred to as genotypic
2. It is denoted as23 0r variables( 3Y Vauuabla)(1.2 or genetic correlation. Genotypic correlation ismore stable and is of paramount importance
to the plant breeder to bring about gencticimprovement in one character by selecting ofher
3. F234 3
It(does not ignor effects of other independent variables. character of pair that is genetically corrclated. Main features of genotypic corelation are
4. It is estimated from simple correlations. presented as follows.
5. Its value is always lower than multiple 1. It involves twO Variables.
corelation.
A is of two types, viz, first order and second order. 2. It is denoted as TR12
7. It can be calculated from un-replicated data also. 3. I! is cstimated from genotypic variances and ço-variances.
Its valuc is always lower than nultiple correlation.
(Bota) 4.
MULTIPLE CORRELAToN Ri-a) S. t is of one type only.
Muliiple Correlation refcrs tojomt influençe of two or mor independent variables on a 6. Its estimation is possible fromreplicated dat{ only.
deendent riable. It is deroted by Ri23. Main features of multipe coclationarc presented It is due to(linkagd or pleiotropy_or(both)
parental population as well
If the association betwech two variablc, rermaisthe sane in
segregating population, it is due to plciotropy If is changes in segregating population, it
as
I. It involves several variables.
2. It is donoted as R3
Ri-23 most likely
resulting in
due to linkagc
recombination
between
between
two
such
genes
genes.
which
The
has broken
pleiotropy or
in segregating population
linkage may involvé two)
The first situation enhances_ the
3. Ioes not ignore) effects of other independent variables. desirable traits or one desirable and one undesirable trait.
the second hinders the progress.
4. It is estimated from simple correlations) genetic improvement, whereas
5. Its value s alwavs higher than simple and partial correlation. Phenotypic Correlation
called phenotypic correlation.
61t is of one type only. The observable correlation between two variables is
includes both genotypic and environmental effects and, therefore, differs under
7. It provides estimate ofcocficient of deierminatie. 1 correlation are presented as
diferent environmental conditions. Main features of phenotypic
8. It is a non-negative estimate. follows.
9. It can be calculated from un-replicated data also. 1. It involves two variables.
Correlations.
TABLE 9.2. Comparison of Simple, Partial and Multiple 2. It is denoted as rphl2
co-variances.
Particulars Simple Partial Müliplef 3. I: is estimated from phenotypic variances and
Coireiau Correlation Correlation 4. 0ts value is always lower than multiple cor:elation.
3ior 4 Severäl S. It is of one type only.
Variables iinvolved
teplicated dath only
Denoted as
6. Its estimation is possible from
2
includes both genotypic and environmental effects.
iwo, first & second Onc only 7. I
8. It is(ess stable th¡n genotypic corelation.
3. Types order 2)
Vaiances and co Simple correlations Simple &ror Vouaue.
4 Estimated from correlations Environmental Correlation
variances is entirely due to environmenta! effects
Not possible) Possible The correlation betwecn two variables which varianceand, therefore, difers under
5 Estimation of coeficjent Not possible x
is called environmental correlation. It is due ta error
of environmental correlation are presentei
different environmental coditions. Main features
of determination Higher than
Lower thanR) Lower than&
6. Value is always Always positive as follows.
-Ve or +ve
-ve or tve
7 Sign of value
(+)
Zo4o -01ceu Cor Net Cor4
do uot! dlo uot
cn- vawob ignore
 CORRELATION ANALYSIS 79
78 BIOMETRICAL TECHNIQUEs IN PLANT BREEDING
variables
1. It involves two variables. DXY= sum of products of products of al! observations on X and Y
2. It is denoted as rel2: IX' =sum of squares ofX variabie
3 It is estimated from environmentai variances and co-varinces. EY = sum of. squarcs of Y variable
4. Its value is always lower than multiple correlation. N= Number 'of observations on X and Y variables
following formula:
5. It is of one type only. (ii) From Replicated Data : It is calculated with the help of
6. Its estimation is possible from eplicated) data only. riy =MSP/[MSr. MStyj/2 MSPE
7. It includes only environinental effects. whcre MSPI = Mean sum of products of genotypes
8. It is not stableo MStr = Mean square of tcatmcnts for the variable a
TABLE 9.3. Comparis on of Genotypic, Phenotypic and Environmental Correlations. MSty = Mean square of treatments for the variable y.
cocficients
Particul:ars The calculation of gcuotypic, phenotypic and environmental correlation
S.No.
GenotypicPhenolypic, Environnental
requires cstimates of corrcsponding variances and
gencral fonnua is as
co-variances. The -Covauibee
correiztiun. correiation correlation
follows (Miller er al., 1958)
1. Variables involved Two Two Two
2. Denoted as Iy = Cov (XI(Var X. Var n
p Cov.(xV)
3. Variances ard co-variances used Genotypic Phenotypic Environmental (a) Phenotypic Correlation = PCov xy/PVx . PVyjl2
for estination
(b) Genotypic Correlation = GCov xy/[GVx . GVyj2 Vay.X . VaY y
Stable Not stable
4. Stability Less stable
(c) Environmental Correlation = ECOVxy/[EVx. EVy]2
where PCOVxy = Phenotypic co-variance between x and y variables Vayuc e
SCALES
PVr = Phenotypic variance of variable x
Searle (1965) has suggested minimum values of correlation cocfficient necessary for indirect PVy =Piienotypic variance of variable y
sclection to be mor eficien than direct selection for yicld. GCOVxy =Genotypic co-variance between x and y variables
TABLE 9.4 Scales for Correlation Coefficients [Searle, 1965]. GVx = Genotypic variance of variable x
SNO. valucs of'Correlation Coeticient SRate or Scale GVy = Genotypic variance of variable y
>0.65 Very strong ECOVy =Environmental co-variancc between x and y variables
0.50 to 0.64 Moderatcly strong EVÊ = Environmental variance of variable x
3. 0.30 to 0.49 Moderatcly weak EVy = Environmental vanance of variable y
4 <0.30 Very weak
2. Partial Correietion
COMPUTATION OF CORRELATION II is calculated from sirple correlation coefficients. It is of two types, iz., first order
and second order. The first order simple correlation is estimated with help of following
1. Simple Correlation formula.
Simple corrclation can be calculated from un-replicated as replicated data as given (1 - ) (l -)}2
below i23 =I2 - 3.
I;2, a andl are estinmates of Simple correlation coefiicients bctween A- 3,
(i) From un-replicated Data : It is calculatcd with the help of following formula yhere
j, respcctively.
R XY (EX . EY)/N/(EX'. (2X)³N. EY?- (EY)N2.
Second order partial corclation coefticicnts are calculated with help of following fonubn
vhere X sum of all observations on X variable
EY sm of all obscrvations on Y variable
J.sx}(n.sg-(ey] Yn-34
 80
BIOMETRICAL TECHNIQUES IN PLANT BREEDING dutouiaaliou
where 23, y4.3 and rz4.3 are estimates of first order partial correlation coefficients 1,
cORRELATION ANALYSIS Goyelallo
2, 3; 1, 4, 3; and 2, 4, 3 variables Multiple Correlation
In plant breeding and genetics, usuallý first order partial correlations are used. 1. If the value of multiple correlation coefficicnt(R) js highly significant,it con firms
the various independent
that the dependeht variable was highly corretated with
3. Multiple Correlation variables.
Multiple corelation cocfficient is calculated with the help of simple correlation square of multiple correlation
2. The coefficicnt of deternmination which is estimated ascomponents
coeficients using following formula. coefficient is the contribution ofvarious character towards dependent
variable, say yield.
where T12, Ti3 T23 are estimates of simple correlation coeficients between variables CROP IMPROVEMENT
selllio APPLICATIONS IN characters. This information
Correlation studies providinformation phout yield contributing genetic popatigns
from diverse
is usefu! to plant breeder in selcclion of clie ectypos
AERDRETIETATION OF RESULTS
Robinson et al., 1951: Johnson et al., 1955).Mass selectiod has been
indirect selection for highy heritabBe
used to improve grain
(raits ywlhich are associated
Simple Correlaticns yield in several crops through
corelations in pBant breeding (Simnonds,
-1. If the value of()is significant, the association between two characters ishigh. with yield. There are three main itmplicaons of
1979) as discussed below.
2. If the value of(r bears negative () sign, it means that increase in the value of one characters s favorable
character will lead to decrease in second character and vice versa. Similarly, If it EÍfect on Selection : A positive conelation betvreen desirable
improvement of both the characters. A
bears positive ) SIgn, it means that increase in one variable will Jead to increase to the plant breeder because it helps in simaltaneous expression of both characters
simultaneous
in seçond character. negative correlation, on the other hand, will hinder
compromise has to be made.
3. If the value of genotypic correlation coefficient (r) is (higher than phenotypic with high values. In such situation, some economic
be achieved
corelation coefficient (E), it means that there is strong association between_ these A.Correlated Response: The genetic improvement in dependent trait can dependent
two characters genetically ba the phenotyp1c value is lessened by the significant 7 by applying (strong selectionto character which is geneticaiiy correlated with
interaction of environment. character. This is called çorrelated response.
heritability. Under such situation,
4. If the value of phenotypic correlation coefficient (rp) is greater thar genotypic 3. Indirect Selection : Sometimes,a character has low with former trait is chosen to
correiation coefficient (re), it shows that the apparent association of two characters another character having high heritabilii and high correlation
achieved using indirect selection
is not only due to genes, but also favorable influence of environment make selection more effective. Thus genetic improvement is
5. If the value of environmental correlation cocfficient (r) is greater than genotypic through component character with high heritability.
and phenotic correlation coeficients), it means that these two characters are
Ahowing high association due to favorable influcnce of particular environment and LIMITATIONS
coefficientsgive misleeding
this assocíation may change in anothe localiy o wilh change 4eAVÍrOnment. There are some drawbacks of correlations. Sometimes correlation third factor
6. If the value ofDis zero or insignifñcant, it mcans that these two characters are yield of rice and jute is due to
esults. For examplc positive correlation betiveensomc
independent. But if the value ofrnpare are also insignificant, it clearly requirement of both the crons. Jn situation on: may get high association
.e., high water
may happen due to smali sample
indicates the independence natureetwo chraracters. between two varíables when no one exists in practice. This
variables havirg linear relationship. In
Partial Correlation size. In some cases correlation is zero between two
scatter diagram and then correlation
1. If the value of partial correlation coefficient s zero, ymeans the simple correlation such cases the relationship should first be studied through
should be estimated (Singh, et al., 1991).
between x, and x,is due to the effect of another variabl<; but after eliminating
the effect of X, the two variable ntay be found as uncorrerated.
2 If the value of partial correlation coefficient (r;2.3) (is significan) it indicates tråe QUESTIONS
significance in plant breeding.
relationship between x, and X, variables. 1. Define simple correlation and describe its types and
is partial correlation ? Describe its types and method of calculation.
2. What
 DETERMINATION OF YIELD COMPONENTS gLCuoYeu

degreeoud
dindioua aiocialiou
kelaliguslu
Motoal

()=1to 1)
Correlation Añaiysis
coYYelaliowoay dit
INTRODUCTION esed by KaulUeaww ia 1902
The term coefficient of correlation was first used by Karl Pearson ín 1902) The statistics
which measure the degree and direction of association between two or more variables is
known as correlation. It 1s denoted b rA positive value oindicates that the changes in
two variables are in the same direction i.e. high valucs of one variable are associated with
high values of other and vice versa. When r is negative, the changes are in opposite directions
i.e. mgii values of onc variable are associated with low values of other. The important
propertics of correlation coefficient are puesented as follows.
J. Unit and Values : It is independent )of the unit of mcasurenent. Its value lies
botween -l and I.
2. Inforn1ation Obtained: It mcasures the degree and direction of association between
tvo or more variables.
2)
3. Analysis : Analysis of corelations coe fficients involves secont order statistics. It
is based on cstimates pf variances and co-variances.
4. Types : Correlations ae of three types. viz., simple, partial and multiple (see later).
Simple. partialand multiplecorelations can be estimated fgm both replicated and kof
(un-replicateddata. Simple correlation coefficients are of thte ypes, vi. genotypic,
phenowDic and environmental. Estimation of these threc types of correlation
cocficients is possible Irom replicated clata only. In plant breeding, estimates of

feutial
Fkgliatddala
 cORRELATION ANALYSIS
Siuple (orelaiou eofliat)
three types of correlation coefficient, viz., genotypic, phenotypic and environmental
(see later) are commonly used.
5. Applications: In plant breeding, estimates of correlation coefficient are useful in
determining yield components which can be used for genetic improvement of yield.
TYPES OF CORRELATIONS

Correlations are of three types, viz. sinmple, partial andmultiple. Simple comelationclassification
coficients
again are of three types, viz., genotypic, phenotvpic and environmental. A brief
of correlations is presented in Table 9.1.
aliowlup TABLE 9.1. Classificatiorn of Correlations.

S.No:ype of CorrelationDenolcd as Definitiou

Sinsple corclation Y2 Association between two variables.

Inherent association between two variabls.
Genotypic Nq Observed association between two variables.
Phenotypic Tph Tph
Association: between two variables due to
Environmental Ye, environmental influence.
2. Partial correlation Yi23 23 The correlation betweentwo variables eliminating
the effect of third variable.
First order
Calculated from simple correlations.
Second order T12.34 Caloulated from first order partial correlations.
Multiple correlation Ri23 Joint influence of two or more independent
3. väriables on a dependeDt variabie.

SH#PLE COPRELATIONTolal Connealiou0ordue tornelalou toel

the(associatiobetween two variables. It is also known as total
Simple Correlation refers to
Main features of simple correlation are
coITClation or zero order correlation cocfficient.
prescnted as follows.
1. It involves two variables.
2. It is denoted ar12:
variables.
3. Itgnores effecfs of other indepencient
4. It is estiiatcd from varianccs
and co-variances. (2)
multiple correlation.
.5. Its valuc is always lower than
environmental.
1t is of three types, viz., genotypic, phenotypic and
data also.
7. It can be calculated from un-rep}licated
koe
(boty, Kapliasd t Ua-replicatael )
PARTIAL cORRELATION Net-(otelaio
n
the effeet of
Partial Correlation refers to the corelalion betwveen two variables climinating
a'vd vauable
 BIOMETRICAL TECHNIQUESIN PLANT BREEDING
third variable. It is denoted as r i t is also known af nct cortelation Main fcatures of G
partial correlation are presented below.
1. tinvolves thrcc or four variables 8-4
2. It is denoted as [23 Or i234.
n3
vauabla)(in2 to
C
3. Itdoes not ignord effects of other independent variables.
4. It is estimated from simple correlations.
5. Its value is always lower than multiplc correlation.
Ais of two types, viz, first order and second order.
7. It can be calculated from un-replicated data also.

MULTIPLE cORRELATIONRi)
(hota)
Muiiple Correlation refers tooint influenç of two or mor independen) variables on a
dependent) nriable. It is deroted by R23: Main features of rmultiple comelation are prcsented
bcto.
1. It involves several variables.
2. It is dnoted as R3 R23
3. Idoes not ignore) effects of other independent variabies. d
4. It is estimated from simple correlations)
5. Its value s alavs higher th¡n simple and partial correlation. P

61t is of one type only.

7. Ii provides estimate ofcocficient of deierminatien
8. It is a non-negative estimate.
9. It can be calculated from un-replicated data also.
TABLE 9.2. Comparison of Simple, Partial and Multiple Correlations.
SNo. Particulars Simple Partial Mültiple
Curreidudn Correlation Correlation
Variables iinvolved 3 or 4 Several t
2. Denoted 'as riz3Y)2-3
3.. Types Three, ra, and re 1wo, first & second One only
(3) order ()
4 Estimatecd from Variances and c0 Simple corelations Simple
variances correlations
5 Estination of coeficjent Not possible X x Not possibleX Possible
of determination
6 Value is always Lower than(R) Lower than (R Higher than r
Sign of value Ve or tve V Or +ve Always positive
+)
Ze40 -0vceu lor Net Corm 4
ötal (önel: clo ust
do uot
NG cORRELATION ANALYSIS
of Genotypic Correlation
The inherent or heritáble association betweenfwo variables is referred to as genotypic
or genetic correlation. Genotypic correlation i< more stable and is of paramount importance
to the plant breeder to bring about genctic improvement in one character by selecting ofher
character of pair that is genetically correlated. Main features of genotypic correlation are
presented as follows.
1. It involves two variables.
2. It is denoted as Tgi2:
3. I! is estimated from genotypic variances and co-Varianccs.
4. Its value is always lower than nultiple correlation.
5. It is of one type only.
n a
6. Its estimation is ypossible from replicated dat only.
nted ( is duc to linkagd or pleiotropy orfboth)
population as well
If the association betwecn two variablc renains the same in parental
population, it
as segregating populaticn, it is due t plciotropy) If is changes in segregating population
most Iikely due to linkage between two genes which has broken in segregating
linkage may involvé two
resulting in recombination between such genes. The pleiotropy or
situation enhances the
desirable traits or one desirable and one undesirable trait. The first
genetic improvement, whereas the second hinders the progress.
Phenotypic Correlation
phenotypic correlation.
The observable corelation between two variables is called
therefore, differs under
includes both genotypic and environmental effects and,
correlation are presented as
different environmental conditions. Main features of phenotypic
follows.
1. It involves two variables.
2. It is denoted as Tphl2:
co0-variances.
3. I: is estimated from phenotypic variances and
4. 0ts value is always lower than multiple cor:slation.
5. lt is of one type only. ( )
234 6. Its estimation is possible from teplicated dath only
effects.
7. It includes both genotypic and environmental
8. It is less stable ttan genotypic corTelation.
Environmental Correlation &Yor Vouauc.
entirely due to environmental effects
The corTelation betwecn two variables which is
ta error varianceand, therefore, differs under
is called environmental correlation. It is due env1ronmental correlation are presentei
nrT different environmental coditions. Main features of
sitive as follows.
 BIOMETRICAL TECHNIQUES IN PLANT BREEDING
1. lt involves two variables.
2. It is denoted as rel2: (Yee
3. It is estimated from environmentai variances and
co-variances.
4. Its value is always lower than multiple correlation.
S. It is of one type only.
6. Its estimation is possible from eplicate) dat· only.
7. It ineludes only environnental efects.
8. It is mot stable
TABLE 9.3. Comparison of Genolypic, Phenotypic and Environmental Correlations.
S.No. Partieul:ars Genofypic Phenotypic,
PË Environnental
corre2ui. Correiation Correlaiioi)
Variables involved Two Two Two
2. Denoted as
3 Variances and co-variances used
for estimation
Genotypic Phenotypic Environmental

4
Stability Stable Less stable Not stable

SCALES
Searle (1065) has suggested minimum values of correlation cocficient necessary for indirect
selection to be moré eficien? than direct selection for yield.
TABLE 9.4 Scales for Correlation Coefficients (Searie, 1965].
SNO. Values of Correlation Coefiient Rate or Scale
> 0.65
2
Very strong
0.50 to 0.64
3
0.30 to 0.49
Moderately strong
4
Moderately weak
<0.30 Very wcak
COMPUTATION OF CORRELATION
1. Simple Correlation
Simple correlation can be calculated from un-replicated as
below : replicated data as given
() From un-replicated Data : IL is calculatcd with the help of
following formula
R- EXY - (EX.EY)/N/[(2X? - (2X)N. EY?-
where XX su) of all observations on X variable
(EY`NU2,
EY Sum of all obscrvations on Y variable n y $24
G
CORRELATION ANALYSIS 79
EXY=sum of products of products of all observations on X and Y variables
sum of squares of X variable
EY = sum of squarcs ofY variable
N= Number'of observations on X and Y variables
(ii) Fron Replicated Data : It is calculated with the help of following formula:
Tiy = MSP/[MSLx . MStyj2 MSIE
where MSPt = Mean sum of products of genotypes
MStr = Mean squarc of icatmcnts for the variable X
MSty = Mean square of trcatnents for the variable y.
environmental correlation cocfficients
The calculation of gouotypic, phenotypic and forinuia is as
co-varianccs. The gencral lovauibucL
requires estimates of corrcsponding variances and
follows (Miller et al., 1958)
ry = Cov (XYY[Var X.Var Yi
Cov-(xv)
al (a) Phenotypic Corelation = PCov xy/[PVx. PVy]e
(b) Genotypic Correlation = GCov xy/[GVx. GVy]2
EVy]2
(c) Environmental Correlation = ECOVxy/[EVx . +
between x and y variables
where PCOVxy = Phenotypic co-variance
PVx = Phenotypic variance of variable x
eçt PVy = Pinenotypic variance of variable y
y variables
GCOVxy = Genotypic co-variance between x and
GVx =Genotypic variance of variable x
GVy = Genotypic variance of variable y
x and yvariables
ECOVxy = Environmental co-variance between
variable x
EVx = Environmnental variance of
EVy = Environmental vanance of
variable y

2. Partial Correialion
coefficients. It is of two types, viz., first order
II is calculated from simple correlation of following
simple correlation is estimated with help
and second order. The first order
formula.
iven 'i23 =i2 - 3. x/[(l - ) (l- ) }l2
correlation cocficients between a -),
I2, Ia andIr3 are estimates of simple
ala
X} and xz - X, respcctively.
calculatcd with help of following fonnuh.
Second order partial conclation coefficicnts are
80 BIOMETRICAL TECHNIQUES IN PLANT BREEDING
where
Ti23, 43and r243 are estimates of first order partial correlation coefficients 1,
2, 3; 1, 4, 3; and 2, 4, 3 variables
In plant breeding and genetics, usuall first order partia correlations are used.
3. Multiple Correlation
Multiple corelation coeficient is calculated with the hclp of simple correlation
coefticients using following formula.

where T12, T13: 23 are estimates of simple correlation coefficients between variables

RIERPRETATION OF RESULTS
Simple Correlaticns
1. If the valuc of(r)is significant, the association between two characters is high.
2. If the value of(r bears negative ( sign, it means that increase in the value of one
character will lead to decrease in second character and vice versa. Similarly, If it
bears positive () Sign, it means that increase in one variable will Jead to increase
in seçond character.
3. If the value of genotypic correlation coefficient (r) is (higher)than phenotypic
correlation coefficient (Em, it means that there is strong association between these
two characters genetically bat the phenotyp1c value is lessened by the significant
interaction of enVironment.
4. If the value of phenotypic correlation coeficient (rp) is greater thar1 genotypic
coreiation coefficient (r.), it shows that the apparent association of two characters
is not only due to genes, but also favorable influence of environment.
5. If the value of environmental correlation cocfficient (r) is greater than genotypic
and phenotyoic correlation coefficients), it means that these two characters are
Ahowing high as_ociation due to favorable influence of particular environment and
this association may change in another localiy o viti change H-eAV0ronient.
6. Ift'he valueof(is zero or insignificant, it means that these two characters are
independent, But if the valueo ;nphcharacters.
re are also insignifcant, it clearly
indicates the indepcndcnce naturef two
Partial Correlation
1. Ifthe valuc of partial correlation coefficicnts zero, i means the simple corrlation
between x and x, is due to the effect of another variabl3) but after eliminating
the effect of x the two variable may be found as uncorrerated.
2. If the value of partial correlation coeficient (r23) (is significant it indicates tråe
relationship between x and x, variables.
 81
cORRELATION ANALYSISs omdallo
Multiple Correlation
1. If the value of muliple correlation coefficicn/(R) }s highly significant, it confirns
independent
that the dependefht variable was highly corretated with the various
variables.
2. The coefficient of deternination which is cstimated as square of
multiple corrclation
coefficient is the contribution ofvarious character components towards dependent
variable, say yield.
APPLICATIONS IN CROP IMPROVEMENT
sellulio
This information
Correlation studies providinformation gbout yield contributing characters.genetic populatians
from diverse
is usefu! to plant breeder in selcction of cltgeniotypos been used to improve grain
Robinson et al., 1951; Johnson et al., 1955).(Mass selectio has (raits ywlich are associated
highly heritable
yield in several crops through indirect selection for
coTelations in plant brceding (Simionds,
with yield. There are three main implicaions of
1979) as discussed below.
. Effect on Selection: A positiye
conelation betvween desirabie characters s favorable
simultaneous improvement of both the characters. A
to the plant breeder because it helps in
hinder simultaneous expression of both characters
negative correlation, on the other hand, will
economic compromise has to be made.
with high values. In such situation, some
Correlated Response: The genetic improvement in dependent trait can be achieved
A. geneticaiiy corelated yith dependent
(strong selection to character which is
by applying
character. This is called correlated response.A
heritability. Under such situation,
3. Indirect Selection:Sometimes,a character has low
high heritability and high correlation with former trait is chosen to
another character having
effective. Thus genetic improvement is achieved using indirect selection
make selection more
heritability.
through component character with high
LIMITATIÔNS
correlations. Sometimes correlation coeficientsgive misle ading
There aresome drawbacks of between yield of rice and jute is due to thd
factor
positive correlation
results. or example In some situation ons may get
high association
the crons.
Le.,high water requirement of both exists in praçtice. This may happen due to smali sample
betweentwo variables when no one In
zero between two variables having linear relationship.
correlation is
Size. In some cases diagram and then correlation
such casesthe relationshipshould first bestudied through scatter
should be estimated (Singh, et al., 1991).
QUESTIONS
breeding.
correlation and describe its types and significance in plant
1. Define simple calculation.
Describe its types and mothod of
2. What is partial correlation ?