Published May, 1975

tGenotypic and Phenotypic Variances and Correlations in Peas

2Shivaji Pandey and E. T. Gritton

ABSTRACT (14) found that seed yield/plot was positively corre-

Eight lines of peas (Pisum sativum L.) and all possible lated with no. of pods/plant and yield/plant and
Fl’s amongthem were grown in five environments to study negatively correlated with 1,000-seed weight.
the genotypic and phenotypic variances and correlations Snoad and Arthur (11) observed genotype X envir-
among eight traits. Genotypes and environments had
highly significant effects on all traits. Parents exhibited onment interaction for every trait investigated, which
highly significant genotype X environment interactions included characters relating to time of maturity, uni-
but Fl’s were remarkably stable. formity of maturity, and yield, of six pea (Pisum
Plant height was often positively correlated with pods/ sativum L.) cultivars grown over several environments
plant, seeds/plant, and yield/plant, indicating superior
yielding ability of taller plants. However, indeterminate in eastern England.
growth is unacceptable in peas for canning because of the Krarup and Davis (6) failed to detect a correlation
range in maturity of peas borne at the different nodes among yield components in peas, but observed signi-
and the excessive vine growth which interferes with har-
vesting. Correlations of percentage seed protein with the ficant and positive correlations between the com-
different traits were variable, but generally positively ponents and yield. The magnitudes of the association
correlated with increased yield. There was some indica- decreased in this order; pods/plant vs. seed yield/
tion of lower protein with heavier seeds. Strong positive plant, seeds/pod vs. seed yield/plant, and weight/
genotypic and phenotypic correlation coefficients obtained
seed vs. seed yield/plant. Thus, the order of import-
between pods/plant and seeds/plant, seeds/pod and seeds/
plant, and between those two traits and yield/plant, ance of these components upon yield was pods/plant,
indicate that selection for either or both of these traits peas/pod, and finally weight/seed.
would result in superior yield. The genotypic correlation
between weight/seed and yield was high, but the pheno- The objectives of this study were to estimate geno-
typic association was low. typic and phenotypic variability among several traits
of peas and to determine the degree of association
Additional index words: Pisum sativum L., Pea breed-
among these traits.
ing, Yield, Protein, Yield coinponents.

MATERIALS AND METHODS

K NOWLEDGE of correlations
as yield, maturity, plant height,
among such traits
and percentage We used pea cultivars ’New Line Early Perfection’ (NLEP),
’Tiny,’ ’Tall Alderman’ (T. Ald.), ’Thomas Laxton’ (Th. Lax.),
seed protein under different environments such as ’Alaska,’ and ’A 45,’ and experimental lines ’B 663-1545’ (B663)
years and locations, is useful in designing an effective and ’Line 87-64’ (L 87). Crosses were made in all possible com-
breeding program for any crop. Weiss et al. (13) binations to give the complete set of F~’s. Parental, F~, and
reported significant positive correlations between late- reciprocal F1 plants were grown in fields at Madison and Arling-
ton, Wis., in 1967 and 1968 and at Arlington in 1970, with
ness and low protein content in soybeans, Glycine crosses as whole plots and lines as subplots. Each whole plot,
max (L.) Merr., and found that the associations re- replicated twice, included the two pareuts of a cross, their F~,
mained stable across environments. In the F2 popula- and their reciprocal F~. Twenty seeds/subplot were planted at
tion of soybeans, Weber and Moorthy (12) found approximately 5-cm spacing along trellises spaced 3 m apart.
Data collected included days to bloom, plant height, pods/plant,
that generally, the genotypic correlations were higher seeds/pod, and total seed weight. From these, seeds/plant and
than the phenotypic correlations. They noted positive weight/seed were calculated. A bulk composite of seed from
genotypic correlations between yield and date of all plauts of a subplot was used for Kjeldahl N deternfination.
maturity, yield and plant height, and yield and seed Nitrogen values were converted to.protein percentage (% N
6.25).
weight. Kwon and Torrie (7) observed a negative Each location within each year was considered as an environ-
.relationship between protein content and seed yield meut in all computations. For each of the generations (parent
in two soybean populations. Their estimates of geno- aud F/s) statistical analyses for maturity, pods/plant, seeds/pod,
typic correlations indicated that lines high in seed seeds/plant, weight/seed, and yield/plant, were made using data
from all five environments. Only three environments, both
yield tended to be late, tall, and susceptible to locations of 1968 and Arlington in 1970, were used for plant
lodging. height and percentage seed protein. Reciprocal Fa’s were not
Kheradnam and Niknejad (5) observed significant used for percentage seed protein in 1968, but were included in
positive phenotypic correlations of yield with no. 1970.
of pods/plant, and of seeds/pod and 100-seed weight Combined analyses of variance and covariance were performed
for all traits in the form given in Table 1. Au additioual source
in cowpea (Vigna sinensis L.). They noted signifi- of variation, lines X blocks within environments, was also con-
cant negative association between 100-seed weight sidered in the case of the parental lines, where sampling was in-
and no. of pods/plant and 100-seed weight and no. volved. The geuotypic compoucots were computed by equating
of seed;/pod. In field beans (Vicia faba L.), Yassin the genotypic variances and covariances to their expected mean
squares and cross products, and from these the genotypic corre-
lations were estimated using the following formula:
~ Contribution from project 1368 of the Dep. of Agron., Wis.
Agric. Exp. Stn., Madison, WI 53706. Research supported by r~ =(cov g,~)/~/(o~,)(o=~)
the Coll. of Agric. and Life Sci., U. of Wis., Madison. Received
June 7, 1974. where r~ is the geuotypic correlation coefficient and e=g , e-"g and
~ Presently postdoctoral fellow, CIMMYT,Loudres 40, Mexico
6, D.F., Mexico, and associate professor of agronomy., respec- coy g’,~ are estimates of the line or family components of variance
tively. and covariance, respectively, for traits i and i’

353
354 CROP SCIENCE, VOL. 15, MAY-JUNE1975

Phenotypic correlations were. calculated using the following Phenotypic Con:elations

equation:
rr
) = M,~/x/(M,,)(MII The highly significant phenotypic correlations (rp)
where rp is the phenotypic correlation coefficient, Mil is the between plant height and yield (Table 3) indicate
mean cross product for lines, and Mlt and Mll are line mean that superior yielding ability is associated with taller
squares for traits i and j. plants. This association may be due to an increase
in general vigor of the plants. Generally speaking,
RESULTS AND DISCUSSION taller plants have more pod-hearing nodes, so it is
not surprising to find significant correlations between
Both environments and lines had important influ- pods/plant and plant height. The no. of pods/plant
ences in the parental and Fx generations for the traits was more closely associated with no. of seeds/plant
studied, Table 2. The highly significant line X
(rp -~ 0.91 to 0.92) than was seeds/pod (r~ -- 0.33 to
environment interactions encountered in the parents 0.47), and showed a somewhat stronger phenotypic
show that the genotypes were not equally affected by association with yield. Weight/seed was significantly
environmental factors for the traits investigated. Ffs
correlated with yield only for parents. Increased
were remarkably stable, though, with only p.ercentage numbers of pods/plant were associated with lighter
seed protein having a significant line X environment
seeds (rp -- --0.50 to --0.55), but there was little
interaction effect. Significant genotype X environ-
association between seeds/pod and weight/seed. The
ment interaction for protein in peas has previously
best indicator of yield among the traits studied was
been observed by All-Khan and Youngs (1) and
seeds/plant (r v z 0.60 to 0.85).
the present authors (unpublished). Plants which bloomed later tended to be taller,
and in the parents, had more pods/plant, seeds/pod,
Table 1. Form of variance and covariance analyses. seeds/plant, yield/plant, and lighter seeds. The only
Expected
other significant association with days to bloom was
Source of variation dr* Mean squares Cross products
a low r~ value for weight/seed in the F1.
Environments 1-1 .... Percentage seed protein showed a positive associa-
LlnesB1°cks
in earl ......
to
g-1(b-11) tion with plant height in the parents and Ffs, and
~e + b~ + bl~g cove + bc~I + blCOVg
Linesx locations (g-l) (I-l) 0"~ + b~
with pods/plant and seeds/plant in parents. The
cove + bcov
I
Error 1 (b-l)(g-I) cov only significant negative rp value involving protein
e
was with weight/seed for the parents (rp ~ --0.51).
* 1, b, and g refer to numbers of locations~ replteations/location, and lines, respective-
ly. Subscripts e, I, and g on variance and covariance components refer to environ- The insignificant rp values between yield/plant and
mental, genotype x environment, and genotypte effecte, respectively. percentage protein indicate that, unlike many other

Table 2. Meansquares from the combinedanalysis of variance for traits of peas.

Traits
Days to Plant Pods/ Seeds/ Seeds/ Wl/ Yield/ Seed
Source df~" Generation bloom ht plant pod plant seed plant protein,
P
Environments 4(2
HP _~_ P 2,484.30** 8,756.25** 264. 38** 8.42** 9,633.52** 1.45** 196.46** 50. 90**
4(2~ F 472.42** 7,438. ~10"* 173. 26** 2. 74** 5,512.62** 0. 85** 196.43** 18. 30**
HP 1
Blocks within 5(3) P 6. 30"* 141.42 2.~1 0.82 197.84"* 0.07’ 6. ~7"* 2.74
HP
environments 5i3~ F 1,274.78** 15,860. 30** 395. 92** 6.39** 14,990. 62** 3.00"* 321.95** 3. 69*
1
Lines 7 !v 2,411. 18"* 28,576.00’* 420. 26** 31.52** I0,600. 65** 23.27** 372.53** 182. 04**
P 109. 39** 46.02** 2.76** I, 250. 45** I. 22** 39. 41"* 8. 32**
55(27) F 3,073.21’*
HP I
Lines x 28(14) P 32.62** 759. 98** 14. 81"* 1.50** 386.64** 0. 63** 22.29** 6.56’*
p 96.92 I.59*
environments 220(110)H(54) F 3.49 88.21 3.03 0.47 0.05 3.95
I
Error p 64.00 I. 95 0.45 64.48 0.02 2. 05 I. 44
480(288
r275(165)n,~,(81)~ P I. 80
F 3.56 111.57 3.85 0. 54 125. 97 0.06 5. 61 0.94
1
*, ** Significant at the 0. 05 or 0. 01 level, respectively. ~" Bracketed degrees of freedom refer only to height (H), protein (P), or both (HP).

Table ~. Phcnotypic and genotypic correlations in parental and F1 generations (phenotypic above and gcnotypic below in each line).
Traits~"
Plant ht Pods/plant Seeds/pod Seeds/plant Wt/seed Yield/plant Protein, %
P F P F P F P F P F P F P F
I I l 1 1 1 I
Days to bloom 0. 23** 0. 12" 0. 34** -0. 07 0. 25** 0.08 0. 38** -0. 06 -0. 23** 0. 13"* 0.25** 0. 06 0. 10 -0. 02
0. 36 0. 39 0.74 0. 43 0, 67 0.67 0. 90 0. 64 --t --~ 0, 61 0.76 0. 43 0. 56
Plant ht 0.53** 0.60** 0.11" 0.01 0.55** 0.05 0.03 0.71’* 0.51"* 0.71’* 0.37** 0.22**
0, 54 0.48 0. 01 0.18 0. 52 0. 56 0.75 --~: 0.48 0.77 0. 53 0. 35
Pods/plant 0. 11" -0.03 0. 91"* 0. 92** -0.55** -0. 50** 0. 44** 0.76** 0.44** 0. 06
0. 09 -0. 06 0. 94 0. 92 --$ --t 0. 18 0, 55 0. 83 -0. I1
*’~ 0. 34** -0. 06
Seeds/pod 0. 47** 0. 53** 0. 01 -0. 09* 0. 53 0. 03
0.41 0.38 0.84 0.71 0.58 0.50 0.09 0.20
Seeds/plant -0, 47"* -0. 50"* 0. 60** 0. 85** 0. 40** 0. 06
--~t --~ 0.38 0.72 0.80 -0.00
Wt/seed 0. 34** -0. 03 -0. 51"* -0. 00
--§ 0.99 --~C --§
Yield/plant -0. 10 0. 04
-0.04 0.07

*, ** Phenotypic value significant at 0. 05 or 0. 01 level, respectively. ~f Phenotypte valae is the first number in each colunm; genotypic, the second. :~ Genotyptc correlation coeffi-
cient eq~s or is less than -1.0. § Genotypte correlation coefficient eqtmla or exceeds 1.0.
 PANDEY & GRITTON: STATISTICAL ANALYSIS OF PEAS
crops, selection of pea plants higher in percentage
seed protein should not result in genotypes producing
lower yields. These results are consistent with the
findings of Furedi (3) and Ali-Khan and Youngs (1),
who observed low and insignificant correlations be-
tween yield and percentage protein in peas.
Genotypic Correlations
Genotypic correlation coefficients (rg) provide a
measure of genetic association between traits, and
thus help identify the more important as well as the
lesser important traits to be considered in a breeding
program. Since a suitable test for significance of the
genotypic correlations is not available, their main
usefulness is in strengthening the interpretations
based on phenotypic correlations.
Out of 28 possible combinations in each of the two
generations, five were > 1 or < —1 in each the
parents and F1's; these result when the rg error vari-
ances are higher than line genotypic variances, leading
to a small denominator in the formula for rg, or if
the signs of the line and error mean products are
different, leading to a relatively large genotypic mean
product (numerator in the formula for rg).
The association of increased yield with increased
plant height, noted in the rp values, is further indi-
cated in the very similar rg values. The implication
is that one should select taller plants if trying to
improve yield in peas. This may be possible if the
plants are grown to maturity and dry seeds harvested,
as was the case in this study, but where peas are
grown for once-over harvest of seeds in the green
succulent state, the indeterminate type of growth is
unacceptable because of the range in maturity of peas
borne at the different nodes and the excessive vine
growth which interfers with harvesting. In this con-
nection, it is gratifying to see the strong rg values
between seeds/pod and yield. Selection pressure to
increase no. of seeds/pod should thus result in in-
creased yield; this would be desirable as it would not
be accompanied by increased range in maturity of
the shelled peas at the processing stage.
The large positive rg values between yield/plant
and days to bloom and yield/plant and weight/seed
indicate that selection for late maturing plants with
heavier seeds would produce plants higher in yield.
However, the situation is complicated by the strong
negative genotypic association between days to bloom
and weight/seed. The genotypic association may be
due to pleiotropy or linkage. Lush (9) and Lerner
(8) have suggested that if the genotypic correlation
is due to pleiotropic genes, selection for two traits in
the same direction will cause a negative change;
simultaneous selection for two traits in opposite di-
rections will cause a positive change, in the genotypic
correlations. If the negative association is due to
pleiotropy, it is unlikely a new combination of traits
can be attained. On the other hand, if the genotypic
correlation is due to linkage, it may be possible to
break the linkage by breeding techniques such as
mutagenesis or intercrossing segregating populations.
Unlike the rp values, rg values indicated a strong
positive relationship between lateness and percentage
seed protein. There was strong association too of per-
355
centage seed protein with plant height, pods/plant,
and seeds/plant, especially for the parents. The asso-
ciation of protein and seeds/pod was not consistent,
but there would appear to be no sacrifice of protein
with selection for increased seeds/pod. The only
strong negative rg values connected with percentage
protein occurred with weight/seed in the parents.
However, the rg value was positive for the FI'S. It thus
appears, that selection of plants higher in protein
would not result in genotypes producing lower yields.
It should be recognized that improvement in one
trait as a result of selection for another depends not
only upon the genotypic and phenotypic correlations
between the two traUs, but also upon the genotypic
and phenotypic variances associated with them, as
suggested by Johnson, Robinson, and Comstock (4).
The masking effects of environment on genotype and
involvement of a large number of loci further reduce
the probability of obtaining desirable combinations.
Recurrent selection involving intercrossing of selected
progenies has been advised by Al-Jibouri, Miller, and
Robinson (2) under such situations. Undesirable as-
sociations between important traits, for example plant
height and yield, in the case of peas grown for har-
vest in the green succulent state, are expected to re-
sult in a reduction in the rate of progress from selec-
tion of some traits, and selection for one may lead to
cultivars with increased levels of other undesirable
traits. Rendel (10) has discussed the genotypic corre-
lations in terms of developmental processes and sug-
gested that if the two traits shared the same substrate
resources, the genotypic correlations will be influenced
by the total resources available and by the distribu-
tion of these resources between the traits. The posi-
tive correlation between yield and plant height may
prove to be a serious limitation in breeding canning
peas for high yield.
ACKNOWLEDGMENT
The authors acknowledge with appreciation the assistance
of J. N. Senturia in certain phases of the statistical analysis.
356 CROP SCIENCE, VOL. 15, MAY-JUNE 1975