Published 1998

3 Overview of Heterosis
and Heterotic Groups
in Agronomic Crops
Albrecht E. Melchinger and Ramesh K. Gumber
Institute of Plant Breeding, Seed Science
and Population Genetics
University of Hohenheim
Stuttgart, Germany

ABSTRACT

Heterotic groups and patterns are of fundamental importance in hybrid breeding.

We start with definitions of these tenns. Theoretical and experimental arguments are given
demonstrating the superiority of inter-group compared with intra-group crosses under two
aspects: (i) a higher mean heterosis and hybrid perfonnance and (ii) a reduced specific
combining ability (SCA) variance and lower ratio of SCA to general combining ability
(GCA) variance, which implies that identification of superior hybrids can be based main-
lyon testing for GCA. We review the degree of heterosis, history and current status of
hybrid breeding, and development of heterotic groups in five major crops with different
pollination systems: allogamous-maize (Zea mays L.) and rye (Secale cereale L.); par-
tially allogamous-faba bean (Vicia/aha L.) and oilseed rape (Brassica napus L.); autog-
amous-rice (Oryza sativa L.). Fundamental principles and systematic approaches for
identification of heterotic groups and patterns and enlarging their genetic base are sug-
gested with special consideration of the use of molecular markers for grouping of
gennplasm. Adapted populations, isolated either by time and/or space are most suitable
candidates for promising heterotic patterns. The potential of heterotic groups in clone and
population breeding also is discussed.

Schnell (1982) pointed out that heterosis is a major yield factor in all breeding
categories except line breeding; however, only in hybrid and clone breeding it is
possible to have maximum exploitation of heterosis. High heterosis can be
expected from a hybrid if the source populations have (i) a high frequency of par-
tially or completely dominant genes, and/or (ii) maximum difference in gene fre-
quencies for overdominant loci. Consequently, for optimum exploitation of het-
erosis in hybrid breeding, the seed and pollen parents should be derived from
genetically unrelated germplasm pools, commonly referred to as heterotic
groups.
Copyright © 1998 Crop Science Society of America, 677 S. Segoe Rd., Madison, WI 53711, USA.
Concepts and Breeding of Heterosis in Crop Plants. CSSA Special Publication no. 25.

29
30 MEICHINGER & GUMBER

Table 3-1. Estimates of midparent heterosis (in percentage of mid-parent performance) for grain yield
in different crops.

Heterosis

Crop/mating system Mean Minimum Maximum Reference

%
Allogamous
Maize-U.S. 121 92 240 Dudley et al.. 1991
-Europe 129 112 143 Melchinger et aI., 1986
Rye 178 86 301 Geiger & Schnell, 1975
207 117 329 Geiger & Wahle, 1978
Partiall~ ailogamous
Faba bean 45 22 69 Kittlitz, 1986
74 55 95 Link et ai., 1996
Oilseed rape-spring 30 20 50 Grant & Beversdorf, 1985
-winter 50 20 80 Lefort-Buson & Dattee, 1982
Autogamous
Rice 36 3 106 Saghai Maroof et ai., 1997
55 31 73 Virmani et aI., 1982
Wheat 9 -14 106 Martin et aI., 1995

In our terminology, a heterotic group denotes a group of related or unrelat-

ed genotypes from the same or different popUlations, which display similar com-
bining ability and heterotic response when crossed with genotypes from other
genetically distinct germplasm groups. By comparison, the term heterotic pattern
used herein refers to a specific pair of two heterotic groups, which express high
heterosis and consequently high hybrid performance in their cross.
Heterotic groups have a strong impact in crop improvement because they
predetermine to a large extent the type of germplasm used in a hybrid breeding
program over a long period of time. The question of suitable heterotic groups and
patterns is becoming relevant in many crops, where hybrid breeding has or will
become feasible with the development of new genetically-engineered systems of
pollination control (Williams, 1995).
The objectives of this review are to describe (i) the importance of heterot-
ic groups in hybrid breeding, (ii) degree of heterosis, (iii) history and current sta-
tus of hybrid breeding, and (iv) development of heterotic groups in five selected
crops. We also suggest fundamental principles and systematic approaches that
might be used for the identification of heterotic groups and patterns as well as
the enlargement of their genetic base.

REVIEW OF HETEROSIS IN DIFFERENT CROPS

Throughout this chapter, the term heterosis refers to the deviation of the F 1
hybrid from the mean (P) of two homozygous parent lines. Analogously, the term
heterotic deviation denotes the difference of the population cross from the means
of two parental popUlations. Table 3-1 summarizes estimates of heterosis for
grain or seed yield, expressed as percentage of P, for selected crops from vari-
HETEROSIS AND HETEROTIC GROUPS 31

ous studies in the literature. In crosses between elite lines of maize (Zea mays L.;
Dudley et al., 1991; Melchinger et aI., 1986), mean heterosis for grain yield was
of similar magnitude (-125%) in germplasm from the U.S. Com Belt and Europe.
In a study with rye (Seeale eereale L.; Geiger & Schnell, 1975; Geiger & Wahle,
1978) using first-cycle inbred lines derived from open-pollinated varieties, mean
heterosis was extremely large (-200%) due to the low yield potential of the par-
ents. Among partially allogamous crops, mean heterosis for yield was similar in
faba bean (50-60%; Kittlitz, 1986; Link et aI., 1996) and oilseed rape (30-50%;
Grant & Beversdorf, 1985). Despite its autogamous nature, a substantial mean
heterosis for yield (35-55%) was reported in rice (Saghai Maroof et al., 1997;
Virmani et aI., 1982). However, heterosis for seed yield was only 9% in wheat
(Triteum aestivum L.; Martin et aI., 1995). These studies clearly demonstrate that
heterosis for grain or seed yield is usually highest for allogamous crops and con-
siderably lower for partially allogamous and autogamous species.

EXPERIMENTAL ARGUMENTS SUPPORTING THE CONCEPT

OF HETEROTIC PATTERNS

Means of Intra-Group vs. Inter-Group Hybrids

The superiority of inter-group over intra-group crosses in terms of mean

performance and heterosis or heterotic deviation for grain yield and other het-
erotic traits has been well documented in the literature. In maize, with genetical-
ly balanced sets of crosses, inter-group hybrids outyielded the respective intra-
group hybrids by 21 % in Reid Yellow Dent (RYD) x Lancaster Sure Crop (LSC)
crosses (Dudley et aI., 1991) and by 16% in Flint x Dent crosses (Dhillon et aI.,
1993) (Fig. 3-1). In both studies, the percentage of increase in heterosis for yield
of inter-group over intra-group crosses was about twice as large as for hybrid
yield itself. In rice, inter-group hybrids between O. indica and O. japonica
derived lines displayed substantially greater heterosis and FI yield than intra-
group crosses (Xiao et aI., 1996). Compared with intra-group crosses, the parental
genetic distance of inter-group crosses, calculated from RFLP or RAPD data, was
only moderately increased in maize, but substantially higher in rice (Fig. 3-1).
In Fig. 3-2, we compared the performance of inter-population and intra-
population crosses, the latter corresponds to the per se performance of the popu-
lations. In two studies with maize germplasm from the U.S. Com Belt, inter-pop-
ulation crosses outyielded intra-population crosses by 16 to 24% with an aver-
age positive heterotic deviation of about 20% (Kauffmann et al., 1982; Mungoma
& Pollak, 1988). In rye, crosses among popUlations from different germplasm
pools such as Petkus and Carsten had on average 11 % higher yield than intra-pool
population crosses (Hepting, 1978). In spring oilseed rape, two inter-population
crosses between open-pollinated cultivars of Canadian and European origin out-
yielded their parent populations by 50%, the average heterotic deviation (LlH)
across all inter-popUlation crosses amounted to 16% (Grant & Beversdorf, 1985).
These results indicate that heterosis or heterotic deviation increases with an
increase in genetic distance between the parents; however, Moll et al. (1965)
32 MEICHINGER & GUMBER

GO 1r1=:=t.==GO====:2O;:11\:;=;---Ir;AC7H:=='7.42O;:I\;:::=;---Ir A :;: 1 l Mg ha-'

:-OF"",Pc-'=-';21;;;=%
--.J I- 10
1_00

0_90
0_80
A
~:7
0_70
6
0_60
5
0_50
4
0_40
3
0_30
2
0.20

0.10

em RYD _ RYO x lSC : - LSC

GO - Mg ha-'
I I AGO=19% I AH=331\ [ H ,P = 18.21\ J
10
1_00 _
0_90 B 9
0_80
:: 8
0_70 7
::

I
0_60 8

I~
5
..
0.50 [:
0.40 ~~:~
~
0.30 Iffii :: 3
0.20 2
Ii:! ~
0.10
1!lil __-='-----'"",--__-="---+
::
0.00 ~ ~~~ ___-=_
+---"'~ --"hl... 0
Genetic Oistanee MPH (Mg ha') Yield (Mgha" )
_ AlntxDent
i ETI Fint ~ Dent

GO

0_14 IAGO= 1751\ I [ A F ,P=84%

35
0 _12

0.10
C 30

0.08

0_06

0_04

0 .02

0.00
Genetic Ois1ance Yield potential (g pr')

o Indica x indica • japonica x Japonica E3 Indic.o x japonlca

Fig. 3-\. Comparison of genetic distance, midparent heterosis, and yield of intra- and inter-group
hybrids in (A) U.S. maize, (B) European maize, and (C) rice. Data taken from Dudley et aI. (1991),
Dhillon et aI. (1993), and Xiao et al. (1996) , respectively. dH, dGD, and dFIP denote the devia-
tions for heterosis, genetic distance, and FI performance, respectively, in cross Ttl x Tt2 calculated
as: d = [ 1()() * Ttl X ~ / ( Ttl X Ttl + Tt2 x Tt2) / 2 1- 100, where Ttl and Tt2 refers to the two popu-
lations investigated.
HETEROSIS AND HETEROTIC GROUPS 33

8,----------------------------------------;
7 I ~H% 16.4 24.1 10.8 16.3 I

-
' ;'
lIS
.c
5
4

-
aI 3
~
2
1
o --+-__ lL.L...L~
(2)
U.S. Maize Rye Oilseed rape

Intra-population eros... _ Inter-populatlon cross ••

Fig. 3-2. Comparison of intra- and inter-population crosses for yield in U.S. maize, rye, and oilseed
rape. Data taken from (I) Kauffmann et aI. (1982), (2) Mungoma and Pollak (1988) in U.S. maize,
Hepting (1978) in rye, and Grant and Beversdorf (1985) in oilseed rape.

demonstrated that this relationship holds true only up to a certain optimum level
of genetic distance and thereafter, one finds a decrease in hybrid performance and
heterosis as depicted schematically in Fig. 3-3. Link et al. (1996) stressed in this
context that hybrid performance and heterosis are not merely a function of genet-
ic distance but also depend on the adaptation of the parents.

Variances Among Intra-Group vs. Inter-Group Hybrids

The variances due to general (a~CA) and specific combining ability (a§CA)
and their ratio are very important for predicting hybrid performance (Melchinger
et aI., 1987). With predominance of abCAover a~CA' early testing becomes more
effective and superior hybrids can be identified and selected mainly based on
their prediction from GCA effects. We reanalyzed the data of a diallel study with

I
~
c:
o
:;
'> Q)
"C

e
o

Q)
Qi
I

Genetic diversity
Fig. 3-3. Schematic relationship between heterosis in varietal crosses and genetic diversity of their
parents based on data from (1) Moll et al. (1962) and (2) Moll et aI. (1965) in maize.
34 MEICHINGER & GUMBER

Table 3-1. Estimates of variance components due to general combining ability (GCA) and specific
combining ability (SCA) in inter-group crosses (IGC) and in diallel crosses without considering
heterotic groups (WCO) for different forage traits in maize (Melchinger, unpublished results, orig-
inal data from Dhillon et aI., 1990).

IOC WGC

Trait a&A a§CA a§cA:abcA a&A a§CA a§CA:a&A

kg ha- 1
Total dry matter yield 77.68 38.94 0.50 49.93 194.03 3.89
Ear dry matter yield 17.51 5.34 0.30 5.24 67.89 12.94
Stover dry matter yield 33.95 19.43 0.57 31.50 36.57 1.16
Mean ratio 0.49 3.45

six Flint and six Dent inbreds published by Dhillon et al. (1990). Estimates of
a~CA and ratio a~CA : akA were much lower for grain and forage yield in the fac-
torial of inter-group crosses (Flint x Dent) than in the complete diallel, treating
the Flint and Dent inbreds as a sample from one common parent population
(Table 3-2). These findings are in agreement with theoretical results indicating
that inter-group crosses have smaller a~CA and a~CA : a~CA ratios than intra-
group crosses (Melchinger, 1996, unpublished data).

STATUS OF HYBRID BREEDING AND HETEROTIC

GROUPS IN DIFFERENT CROPS

The experimental results clearly demonstrated the superiority of inter-

group over intra-group hybrids and the necessity of heterotic groups in hybrid
breeding. Despite their great importance, most of the primary heterotic groups
were not developed in a systematic manner. They were determined empirically by
relating the observed heterosis and hybrid performance with the origin of parents
included in the crosses. In this section, we review the development of primary
heterotic groups and patterns in five selected crops.

Maize
U.S. Corn Belt and Canada
Richey (1922) provided the first important information that eventually led
to the development of heterotic patterns in maize. He compared heterosis for
yield in a large number of inter-varietal crosses and found higher performance for
hybrids between varieties of different endosperm type than those among varieties
possessing the same endosperm type. His investigation strongly suggested that
crosses between geographically or genetically distant parents express higher per-
formance and consequently increased heterosis.
Based on this principle, the most widely used heterotic pattern in the U.S.
Corn Belt, RYD x LSC, was evolved. The two parent populations of this heterotic
pattern were adapted to different geographic regions and also had different genet-
ic backgrounds (Hallauer et al., 1988). The variety RYD was originated in
Western Illinois by a chance hybridization between strains 'Gordon Hopkins' and
HETEROSIS AND HETEROTIC GROUPS 35

'Little Yellow', while the open-pollinated variety LSC was developed from the
cross of an early Flint population with a late local variety in Lancaster County,
Pennsylvania.
A first systematic attempt to identify alternative heterotic patterns to RYD
x LSC was reported by Kauffmann et al. (1982). They evaluated diallel crosses
among 10 u.s. Com Belt open-pollinated varieties and identified two promising
alternative heterotic groups: Leaming and Midland. These groups were originat-
ed in different regions of the U.s. Com Belt and have distinct morphological
characteristics: Learning developed in southern Ohio has a slender ear, while
Midland from southern Kansas has a wider leaf plant type, short ear, and more
kernel rows. They combine well with each other and with LSC (Hallauer et al.,
1988). These two new heterotic groups were identified in an advanced breeding
stage and thus they were not competitive with germplasm derived from the well-
established RYD and LSC heterotic groups, which had undergone numerous
selection cycles for agronomic traits. More recently, Mungoma and Pollak
(1988) evaluated diallel crosses among seven yellow and three white endosperm
type populations to identify new heterotic patterns. They suggested that a
Mexican Dent population can be used as an alternative to LSC in the U.S. Com
Belt heterotic pattern.

Central and Western Europe

When hybrid breeding was started in Europe after World War II, attempts
were made to introduce early-maturing Dent hybrids from Canada and the U.S.
Com Belt; however, these were not sufficiently adapted to the cooler climatic
conditions of Central Europe (Schnell, 1992). In the 1950s, indigenous Flint pop-
ulations such as Lizagaraute, Lacaune (both from France) as well as Gelber
Badischer Landmais and Rheintaler (from Germany) became the ancestors of the
current Flint germplasm. The productivity of Flint lines like F2, F7, EP1, and DK
105 derived from these ancestral populations was generally low. Therefore, the
Flint lines were crossed to early-maturing highly productive Dent lines of North
American origin to develop high yielding adapted hybrids. This established the
complementary Flint x Dent heterotic pattern.
The source populations of the Dent lines were derived from the cross of
'early slender-stalked' Flint com of the north-eastern USA with the 'late heavy
stalked gourd seed' of the south-central USA. The Flint populations can be traced
back to tropical Flints originating from the West Indies and the Caribbean Islands
(Wallace & Brown, 1956) and had been isolated from Dent populations for at
least 500 yr.

Tropics and Subtropics

Promising heterotic patterns used in the tropics are ETO x Tuxpeno, Suwan
I x Tuxpeno, Cuban Flints x Tuxpeno, and Caribbean Flints x Tuxpeno (M.
Bjamason, 1996, personal communication). In addition, Caribbean Flints x
Tuxpeno in Brazil, Mexico, and Colombia, and ETO x Caribbean Flints in
Venezuela, and American early Dent x Tuxpeno in Egypt also are used in hybrid
breeding (Eberhart et aI., 1995; Pandey & Gardner, 1992).
36 MEICHINGER & GUMBER

A systematic search by Goodman (1985) resulted in the identification of 10

heterotic patterns based on testcross evaluation of tropical races. He suggested
the use of only those promising heterotic patterns that involved populations of
Tuxpeno, Tuson, Cuban Flint, Coastal tropical Flint, and Chandelle. To identify
heterotic patterns among CIMMYT quality protein maize (QPM) germplasm,
Vasal et al. (1993) evaluated diallel crosses among 10 parents, six QPM gene
pools and four QPM populations. They suggested Pool 32 QPM, Population 68
QPM, and Population 69 QPM as ideal candidates for promising heterotic pat-
terns due to their high grain yield and good combining ability.

Rye

Hybrid breeding in rye was started with the detection of cytoplasmic-genic

male sterility (Pampa CMS) from the Argentinean population 'Pampa rye' by
Geiger and Schnell (1970) more than two decades ago. First commercial hybrids
were released in 1985. They had 15 to 20% higher grain yield but slightly lower
plant height than open-pollinated population varieties. Presently >50% of the rye
area in Germany is planted with hybrids.
The systematic search for suitable heterotic patterns by Hepting (1978) laid
the foundation of hybrid breeding in rye. Based on the hybrid performance and
heterotic deviation in a complete diallel among seven open-pollinated popula-
tions, he found that cross combinations involving populations from the two most
widely used germplasm groups, Petkus and Carsten, were most promising for
grain yield. In fact, all rye hybrids released in Germany since 1985 are based on
the Pampa CMS system and are of the Petkus x Carsten type (H.H. Geiger, 1996,
personal communication).
The Petkus and Carsten germplasm groups originated from two open-pol-
linated populations and were developed from different landraces. Both popula-
tions were well adapted to temperate continental climate but different soil condi-
tions. The use of different source materials and independent selection for differ-
ent plant ideotypes resulted in genetic differences for various traits. Besides their
different history of evolution, they were separated from each other for at least 100
yr. Allozyme data confirmed a greater genetic distance between the Petkus and
Carsten germplasm groups than within Petkus and Carsten (Adam et al., 1987).
Due to its wide adaptation and broad genetic base, the Petkus material has been
used either directly or as an important germplasm source in most rye breeding
programs in the world. Carsten germplasm has high lodging resistance and
greater number of kernels per ear but a smaller lOO-kernel weight than Petkus
(H.H. Geiger, 1996, personal communication).

Oilseed Rape

The presence of substantial heterosis for seed yield (up to 72%) and effec-
tive pollination control systems (Ogura, 1968; Williams, 1995) stimulated the
research on hybrid development in oilseed rape. Recently, commercial hybrids
have been introduced in Canada, USA, France, and Germany. Hybrids have 15 to
22% higher seed yield and improved yield stability than the best line varieties.
HETEROSIS AND HETEROTIC GROUPS 37

No distinct heterotic groups comparable to maize have been established so

far in winter and spring oilseed rape. Winter and spring cultivars have been cul-
tivated in different geographical regions and represent distinct germplasm groups
with little inter-crossing in the past (Diers & Osborn, 1994). Spring cultivars are
grown extensively in Canada, Australia, and Scandinavia, while winter cultivars
are grown in the rest of Europe and Japan.
In a search for suitable heterotic patterns, Grant and Beversdorf (1985)
evaluated diallel crosses among six open-pollinated cultivars of European and
Canadian spring rapeseed. Based on hybrid performance and heterotic deviation,
they suggested European spring x Canadian spring rapeseed as a promising het-
erotic pattern. Similarly, Lefort-Buson et al. (1987) found crosses between win-
ter rapeseed of European x Asian origin to be most productive. The results of
diallel crosses among seven spring rapeseed cultivars of European, Canadian, and
Asian origin further confIrmed that crosses between European x Asian and
Canadian x Asian spring oilseed rape exhibit higher heterosis than crosses
between parents originating from the same region (Brandle & McVetty, 1990).
More recently, Diers and Osborn (1994) and Becker et al. (1995) classified win-
ter, spring, and Asian (both spring and winter) oilseed rape into three different
groups using RFLP assays.
The potential of heterosis between winter and spring oilseed rape crosses
has not been investigated so far. A recent proposal (T. Osborn, 1993, personal
communication) has suggested the development of separate complementary het-
erotic groups for winter and spring oilseed rape by crossing genetically diverse
lines of winter and spring rapeseed. Selection should be performed in subsequent
backcross generations for plants having most of their genome from winter rape-
seed but with spring growth habit. A reciprocal procedure could be applied to
select for the genetic background of spring oilseed rape but with vernalization
requirements and frost resistance; however, the latter may be more difficult
because frost resistance and winter hardiness are quantitative traits.
Krahling (1987) suggested the use of resynthesized lines to develop new
genetically divergent groups and to widen the genetic base of spring and winter
oilseed rape. Engqvist and Becker (1994) recommended first improving the
resynthesized lines by recurrent selection before using them in hybrid breeding
with winter and spring oilseed rape. Moreover, the genetic distance of resynthe-
sized lines to current rapeseed germplasm needs to be checked as many of the
resynthesized lines showed a rather small genetic distance to winter and spring
rapeseed lines measured by RFLPs (Becker et aI., 1995).

FabaBean

Populations or open-pollinated varieties are presently under cultivation in

faba bean. Due to the partially allogamous propagation of the crop, these varieties
use only about half of the available heterosis. Hybrid breeding in faba bean would
be very promising because it allows maximum exploitation of heterosis and com-
bines high yield potential with improved yield stability, a major problem in this
crop (Stelling et aI., 1994); however, no hybrid has been released on a commer-
cial scale due to complications associated with the existing hybridization sys-
38 MEICHINGER & GUMBER

terns. The discovery of two new CMS systems and effective restorers have
opened new avenues for hybrid breeding ( Link et al., 1997).
If hybrid breeding becomes feasible in faba bean, the following three well-
established germp1asm pools are available: Minor, Major, and Mediterranean.
Minor germplasm are tall (up to 180 cm), small-seeded spring beans, mainly used
for animal feed. Major germplasm are generally shorter (up to 120 cm), large-
seeded spring beans, cultivated primarily for direct human consumption.
Mediterranean germplasm has a short stature (up to 110 cm) and comprises small,
medium-, and large-seeded winter beans. The Minor and Major groups are grown
in Central and North-western Europe, while the Mediterranean group is cultivat-
ed in southern Europe and northern Mrica.
Link et aI. (1996) conducted a systematic search for the most promising
heterotic pattern among these three heterotic groups for two mega-environments:
Central Europe and the Mediterranean region. Based on the F 1 performance and
mid-parent heterosis between balanced sets of crosses, the authors suggested the
Minor x Major heterotic pattern for Central European environments. The Minor
x Mediterranean heterotic pattern also has great promise and could be an alter-
native, if lodging is prevented (Link et al., 1996). Based on RAPD assays (Link
et aI., 1995), the large genetic distance between the Minor and Mediterranean
groups is responsible for the expression of high heterosis and hybrid perfor-
mance. For the Mediterranean environments, intra-group crosses of
Mediterranean germplasm was recommended as a short-term solution for hybrid
breeding. The Minor x Mediterranean heterotic pattern seems to be a suitable
alternative but the extremely poor adaptation of Minor germplasm to
Mediterranean environments may hamper their direct use (Kittlitz et aI., 1993).
Because the genetic distance, measured with RAPD markers, within the
Mediterranean group is comparatively high (Link et aI., 1995), it might be possi-
ble to identify sufficiently divergent heterotic groups within the Mediterranean
germplasm.

Rice

Substantial heterosis for grain yield, first reported by Jones (1926), and an
effective cytoplasmic male sterility system ('WA' CMS; Lin & Yuan, 1980) pro-
moted hybrid breeding in rice. Hybrid rice has contributed significantly to the
dramatic increase of rice production in the world. In China alone, hybrid rice is
grown annually on about 17 million ha, which corresponds to 55% of the total
area under rice cultivation (Xiao et al., 1995). New hybrid rice varieties have a
30 to 45% yield advantage over conventional line cultivars (Yuan, 1992).
Encouraged by the Chinese success in hybrid rice, several other countries in Asia
as well as Brazil, Mexico, and the USA have initiated hybrid rice breeding pro-
grams. High yielding hybrids have been produced from crosses between O. indi-
ca and O. japonica (Xiao et aI., 1995); however, hybrids between indica and
japonica rices show a variable degree of sterility. Moreover, fertility restorer fre-
quency is lower in japonica rice than indica rice. Presently, hybrids involving
indica x indica crosses are under cultivation and express as much as 70% het-
erosis.
HETEROSIS AND HETEROTIC GROUPS 39

The three subspecies of O. sativa (i.e., indica,japonica, andjavanica) have

different morphological and physiological characteristics and ecogeographical
distribution and, therefore, may serve as distinct heterotic groups. Genotypes
belonging to indica and japonica are adapted to tropical and temperate climates,
respectively, whereas javanica rice has intermediate characteristics. The average
genetic distance measured by RFLPs between indica lines was three to four times
higher than that betweenjaponica lines (Zhang et al., 1992). Yuan (1992) point-
ed out that heterosis for grain yield in crosses among the three subspecies has the
following trend: indica x japonica > indica x javanica > javanica x japonica >
indica x indica > japonica x japonica. This is in harmony with the increased
genetic distance between parents of indica x japonica crosses compared with
indica x indica and japonica x japonica crosses found in a recent study with
RAPD markers (Xiao et al., 1996).
Zhang et al. (1995) studied molecular divergence and hybrid performance
in diallel crosses among eight O. indica lines representing the parents of the best
performing commercial rice hybrids grown in China. Their results suggest the
existence of two heterotic groups within O. indica, one comprising rice strains
from southeast China and the other comprising strains from Southeast Asia.

CRITERIA FOR THE IDENTIFICATION OF NEW HETEROTIC

GROUPS AND PATTERNS

The above review indicates that heterotic groups and patterns in many
crops were identified by relating the heterosis observed in crosses with the ori-
gin of the parents included in the crosses. We recommend the following criteria
for the identification of heterotic groups and patterns in descending order of
importance:
1. high mean performance and large genetic variance in the hybrid popula-
tion to ascertain future selection response by adopting the usefulness
concept of Schnell (1978);
2. high per se performance and good adaptation of both or at least one of
the parental heterotic groups;
3. low inbreeding depression in the source materials for the development
of inbreds; and
4. a stable CMS system without deleterious side effects, as well as effec-
tive restorers and maintainers, if hybrid breeding is based on cytoplas-
mic male sterility.
While the mean hybrid performance of a large number of crosses can be
determined rather accurately by replicated field evaluation, estimation of the
genetic variation within parent or hybrid populations by field evaluation of a
large number of S], half-sib, or full-sib progenies per population is very labori-
ous and resource demanding. Alternatively, molecular markers such as RFLPs,
AFLPs, or RAPDs could be employed, which have proved very effective in mea-
suring the genetic diversity at the DNA level (Melchinger, 1993).
40 MEICHINGER & GUMBER

APPROACHES FOR THE IDENTIFICATION OF HETEROTIC

GROUPS AND PATTERNS

The fundamental principle underlying the approaches for the identification

of heterotic patterns is to select, from a large number of germplasm sources, par-
ent populations of crosses with highest hybrid performance. Diallel and test-
crosses are generally used for this purpose. The various steps involved in the
identification of heterotic groups and patterns depend on the source materials and
the availability of established heterotic patterns. In principle, these approaches
could also be used for enlarging the genetic base of existing heterotic groups.

Small Number of Populations

We suggest producing a complete diallel with a small number of popula-

tions. The diallel crosses together with their parent populations are evaluated in
replicated field trials for hybrid performance and heterotic deviation. Using the
identification criteria given above, parent populations of cross combinations with
high performance are selected as potential heterotic groups and patterns. If well-
established heterotic patterns are available, their performance is compared with
the new heterotic patterns.
New heterotic patterns in the U.S. Com Belt such as Midland x Leaming
and Midland x LSC (Kauffmann et al., 1982) or BSSS(R)ClO x Mexican Dent
(Mungoma & Pollak, 1988) were identified by this approach. Similarly, Hepting
(1978) in rye and Grant and Beversdorf (1985), Lefort-Buson et al. (1987) and
Brandle and McVetty (1990) in oilseed rape identified promising heterotic pat-
terns using diallel crosses of open-pollinated populations.

Large Number of Germplasm Accessions

Case 1: Heterotic Patterns Are Not Available

With large number of inbred lines or populations available, it is not feasi-
ble in most crops to make diallel crosses and produce sufficient FI seed for multi-
environment field testing. Therefore, we suggest a multi-stage procedure to iden-
tify heterotic groups, which consists of the following steps:
1. grouping the germplasm based on genetic similarity;
2. selection of representatives genotypes (e.g., two to four lines or one
population) from each subgroup for producing diallel crosses;
3. evaluation of diallel crosses among the subgroups together with the par-
ents in replicated field trials; and
4. selection of the most promising cross combinations as potential heterot-
ic patterns using the identification criteria.
A preliminary classification of germplasm into subgroups of genetically
similar lines could be based on the geographic origin, morphological data, pedi-
gree information, and breeding history of the crop. An extremely powerful tool
for grouping of germplasm are molecular markers such as RFLPs, AFLPs, and
HETEROSIS AND HETEROTIC GROUPS 41

RAPDs (Melchinger, 1993). Using genetic similarities determined from molecu-

lar data, relationships between genotypes can be represented graphically by clus-
ter analysis or principal coordinate analysis as demonstrated in Fig. 3-4.

Case 2: Established Heterotic Patterns Are Available

If established heterotic patterns are available, we recommend using select-
ed elite genotypes from them as testers for the production and evaluation of the
germplasm to be classified. Based on the testcross performance, populations or
lines having similar combining ability and heterotic response could be merged to
constitute a new independent heterotic group, if they behave differently from the

0.4r,Ir-~~==~~I.=~~~----1
I· Dent 1 I I", Flint 1 III
I
A I
I
0.2 I
•• I
• •• I
• • I
• •• I D. 6

PC2 0.0 -----~~ --~--- --- ~---

..
-0.2

..•
-0 . 4~I=I ______________~____________~I~
- 0.4 -0.2 0.0 0.2 0.4
PC1

B XQ
CT
,..-----IR
O. indica ,..----CE
'-----R4
' - - - - - - - - MH
r----CB
'------',---Jl
NP
O.japonica L-----RG
,
0.12 0.09 0.06 0:03 0.00
Genetic distance

Fig. 3-4. Grouping of germplasm in maize and rice. (A) Principal coordinate analysis of genetic sim-
ilarities determined from RFLPs in European maize (Messmer et al., 1992): principle coordinate
I (PCI) separates the Dent lines (squares in Quadrants I and II) from the Flint lines (triangles in
Quadrants III and IV); (B) Cluster analysis of genetic distances determined from RAPDs in rice
(Xiao et aI., 1996): XQ, CT, IR, CE, R4, and MH represent lines from O. indica, and CB, JL, NP,
and RG represent lines from 0. japonica.
42 MEICHINGER & GUMBER

existing heterotic groups; however, if their behavior is similar to an existing het-

erotic group, they could be merged with it to enlarge its genetic base. The genet-
ic diversity of lines or populations selected either for constituting a new heterot-
ic group or broadening existing ones should be confirmed by molecular analysis.
For instance, Kauffmann et al. (1982) broadened the genetic base of the
newly identified maize heterotic groups Midland and Learning by using them as
testers in crosses with a large number of cultivars. Lines showing a heterotic
response similar to Midland in crosses with Learning were grouped into the
Midland heterotic group and vice versa.

CONCLUSIONS

Heterotic groups are the backbone of successful hybrid breeding. The deci-
sion regarding the choice of heterotic groups is of fundamental importance and
must be made at the beginning of a crop improvement program. Once the het-
erotic groups are established and improved over a number of selection cycles, it
is extremely difficult to develop new competitive ones, because at an advanced
stage, the gap in performance between improved breeding materials and unim-
proved source materials is usually too large. Only if the breeding goals change, is
there a good chance to change or develop new heterotic groups with reasonable
effectiveness. Moreover, if hybrid breeding is based on eMS, the selection of a
heterotic group representing the seed parents (male steriles) and pollen parents
(restorers) must be made at an early stage, because considerable efforts are need-
ed for the transfer of eMS and restorer genes. It is evident from the review of
various studies that genetic diversity caused by isolation in space and/or time is
essential for the establishment of promising heterotic patterns. Heterotic groups
should not be considered as closed populations, but should be broadened contin-
uously by introgressing unique germplasm to warrant medium and long-term
gains from selection. While characterization of germplasm is possible with mol-
ecular markers, heterosis and hybrid performance among unrelated germplasm is
generally not predictable and requires field evaluation of crosses arnong them
(Melchinger, 1993).

ACKNOWLEDGMENTS

We are grateful to H.H. Geiger and T. Miedaner for fruitful discussions in

preparing this review. The financial support from the Vater und Sohn Eiselen
Stiftung, Ulm, to R.K. Gumber is greatly appreciated.

REFERENCES

Adam, D., V. Simonsen, and V. Loeschke. 1987. Allozyme variation in rye, Secale cereale L: 2.
Commercial varieties. Theor. Appl. Genet. 74:560--565.
Becker, H.C., G.M. Engqvist, and B. Karlsson. 1995. Comparison of rapeseed cultivars and resyn-
thesized lines based on a1lozyme and RFLP markers. Theor. Appl. Genet. 91:62--67.
HETEROSIS AND HETEROTIC GROUPS 43

Brandle, J.E., and P.B.E. McVetty. 1990. Geographical diversity, parental selection and heterosis in
oilseed rape. Can. J. Plant Sci. 70:935-940.
Dhillon, B.S., J. Boppenmaier, w.G. Pollmer, R.G. Herrmann, and A.E. Melchinger. 1993.
Relationship of restriction fragment length polymorphisms among European maize inbreds
with ear dry matter yield of their hybrids. Maydica 38:245-248.
Dhillon, B.S., P.A. Gurrath, E. Zimmer, M. Wernke, w.G. Po\lmer, and D. Klein. 1990. Analysis of
diallel crosses of maize for variation and covariation in agronomic traits at silage and grain
harvest. Maydica 35:297-302.
Diers, B.W., and I.C. Osborn. 1994. Genetic diversity of oilseed Brassica napus germplasm based on
restriction fragment length polymorphisms. Theor. Appl. Genet. 88:662--668.
Dudley, 1.w., M.A. Saghai-Maroof, and G.K. Rufener. 1991. Molecular markers and grouping of par-
ents in a maize breeding program. Crop Sci. 31:718-723.
Eberhart, S.A., W. Salhuana, R. Sevilla, and S. Taba. 1995. Principles of tropical maize breeding.
Maydica 40:339-355.
Engqvist, G.M., and H.c. Becker. 1994. What can resynthesized Brassica napus offer to plant breed-
ing. Sver UtsiidesfOrenings Tidsk 104:87-92.
Geiger, H.H., and EW. Schnell. 1970. Cytoplasmic male sterility in rye (Secale cereale 1.). Crop Sci.
10:590--593.
Geiger, H.H., and EW. Schnell. 1975. Experimental basis for breeding hybrid varieties in rye.
Hodowla Rosl. Aklim. Nasienn. 19:381-385.
Geiger, H.H., and G. Wahle. 1978. Struktur der Heterosis von Komplexmerkmalen bei Winterroggen-
Einfachhybriden. Z. Pflanzenziichtg 80:198-210.
Goodman, M.M. 1985. Exotic maize germplasm: Status, prospects and remedies. Iowa State 1. Res.
59:494-527.
Grant, I., and W.D. Beversdorf. 1985. Heterosis and combining ability estimates in spring oilseed rape
(Brassica napus 1.). Can. J. Genet. Cytol. 27:472-478.
Hallauer, A.R., W.A. Russell, and K.R. Larokey. 1988. Corn breeding. p. 463-564. In G.E Sprague,
and J.w. Dudley (ed.) Corn and corn improvement. 3rd ed. Agron. Monogr. 18. AS A, CSSA,
and SSSA, Madison, WI.
Hepting,1. 1978. Analyse eines 7 x 7 - Sortendiallels zur Ermittlung geeigneten Ausgansmaterials
fur die Hybridziichtung bei Roggen. Z. Pflanzenziichtg. 80: 188-197.
Jones, J.w. 1926. Hybrid vigour in rice. 1. Am. Soc. Agron. 18:423-428.
Kauffmann, K.D., c.w. Crum, and M.E Lindsey. 1982. Exotic germplasm in a corn breeding pro-
gram. III. Corn Breeder's School 18:6-39.
Kittlitz, E. von. 1986. Some observations in reciprocal crosses between Vicia faba major and Vicia
faba minor. BioI. Zentralbl. 105:147-153.
Kittlitz, E. von, K.I.M. Ibrahim, P. Ruckenbauer, and 1.D. Robertson. 1993. Analysis and use of inter-
pool crosses (Mediterranean x Central European) in faba beans (Viciafaba 1.): I. Performance
of Mediterranean and Central European faba beans in Syria and Germany. Plant Breed.
110:307-314.
Krahling, K. 1987. Utilization of genetic variability of resynthesized rapeseed. Plant Breed.
99:209-217.
Lefort-Buson, M., and Y. Dattee. 1982. Genetic study of some agronomic characters in winter oilseed
rape (Brassica napus 1.): I. Heterosis II. Genetic parameters. Agronomie 2:315-332.
Lefort-Buson, M., B. Guillot-Lemoine, and Y. Dattee. 1987. Heterosis and genetic distance in rape
seed (Brassica napus 1.): Crosses between European and Asiatic selfed lines. Genome
29:413-418.
Lin, S.C., and 1.P. Yuan. 1980. Hybrid rice breeding in China. p. 35-51. In Innovative approaches to
rice breeding. IRRI, Manila, Philippines.
Link, w., C. Dixkens, M. Singh, M. Schwall, and A.E. Melchinger. 1995. Genetic diversity in
European and Mediterranean faba bean germplasm revealed by RAPD markers. Theor. Appl.
Genet. 90:27-32.
Link, W, W. Ederer, R.K. Gumber, and A.E. Melchinger. 1997. Detection and characterization of two
new CMS systems in faba bean. Plant Breed. 116:158-162.
Link, w., B. Schill, A.C. Barbera, J.I. Cubero, A. Filippetti, 1. Stringi, E.V. Kittlitz, and A.E.
Melchinger. 1996. Comparison of intra- and inter-pool crosses in fababean (Viciafaba 1.): I.
Hybrid performance and heterosis of crosses in Mediterranean and German environments.
Plant Breed. 115:352-360.
Martin, J.M., 1.E. Talbert, S.P. Lanning, and N.K. Blake. 1995. Hybrid performance in wheat as relat-
ed to parental diversity. Crop Sci. 35:104-108.
44 MEICHINGER & GUMBER

Melchinger, A.E. 1993. Use of RFLP markers for analysis of genetic relationship among breeding
materials and prediction of hybrid performance. p. 621--628. In International Crop Science I.
CSSA, Madison, WI.
Melchinger, A.E., H.H. Geiger, and F.W Schnell. 1986. Epistasis in maize (Zea mays L.): 2. Genetic
effects in crosses among early flint and dent inbred lines determined by three methods. Theor.
Appl. Genet. 72:231-239.
Melchinger, A.E., H.H. Geiger, G. Seitz, and G.A. Schmidt. 1987. Optimum prediction of three-way
crosses from single crosses in maize (Zea mays L.). Theor. Appl. Genet. 74:339-345.
Messmer, M.M., A.E. Melchinger, J. Boppenmaier, E. BrunkJaus-Jung, and R.G. Herrmann. 1992.
Relationship among early European maize inbreds: I. Genetic diversity among flint and dent
lines revealed by RFLPs. Crop Sci. 32:1301-1309.
Moll, R.H., 1.H. Longquist, J.Y. Fortuna, and E.C. Johnson. 1965. The relation of heterosis and genet-
ic divergence in maize. Genetics 52:139-144.
Moll, R.H., WS. Saihuana, and H.F. Robinson. 1962. Heterosis and genetic diversity in variety cross-
es of maize. Crop Sci. 2:197-198.
Mungoma, C., and L.M. Pollak. 1988. Heterotic patterns among ten Com Belt and exotic maize pop-
ulations. Crop Sci. 28:500-504.
Ogura, H. 1968. Studies on the male sterility in Japanese radish with special reference to the utiliza-
tion of the sterility towards the practical raising of hybrid seed. Mem. Fac. Agric. Kagoshima
Univ. 6:39-48.
Pandey, S., and C.O. Gardner. 1992. Recurrent selection for popUlation, variety and hybrid improve-
ment in tropical maize. Adv. Agron. 48:2-79.
Richey, F.D. 1922. The experimental basis for the present status of com breeding. J. Am. Soc. Agron.
14:1-17.
Saghai Maroof, M.A., G.P. Yang, Qifa Zhang, and K.A. Gravois. 1997. Correlation between molecu-
lar marker distance and hybrid performance in U.S. southern long grain rice. Crop Sci.
37:145-150.
Schnell, F.W 1978. Progress and problems in utilizing quantitative variability in plant breeding. Plant
Res. Dev. 7:32-43.
Schnell, F.W 1982. A synoptic study of the methods and categories of plant breeding. Z.
Pflanzenziichtg. 89: 1-18.
Schnell, FW. 1992. Maisziichtung und die Ziichtung forschung in der Bundesrepub1ik Deutschland.
Vortr. Pflanzenziichtung 22:27-44.
Stelling, D., E. Ebmeyer, and W Link. 1994. Yield stability in faba bean, Viciafaba L: 2. Effects of
heterozygosity and heterogeneity. Plant Breed. 112:30-39.
Vasal, S.K., G. Srinivasan, F Gonzalez c., D.L. Beck, and J. Crossa. 1993. Heterosis and combining
ability of CIMMYTs quality protein maize germplasm: II. Subtropical. Crop Sci. 33:51-57.
Virmani, S.S., R.C. Aquino, and G.S. Khush. 1982. Heterosis breeding in rice COryza sativa L.) Theor.
Appl. Genet. 63:373-380.
Wallace, H.A., and WL. Brown. 1956. Com and its early fathers. Michigan State Univ. Press, East
Lansing.
Williams, M.E., 1995. Genetic engineering for pollination control. Tibtech 13:344-349.
Xiao, J., 1. Li, L. Yuan, S.R. McCouch, and S.D. Tanksley. 1996. Genetic diversity and its relation-
ship to hybrid performance and heterosis in rice as revealed by PCR-based markers. Theor.
Appl. Genet. 92:637--643.
Xiao, J., 1. Li, L. Yuan, and S.D. Tanksley. 1995. Dominance is the major genetic basis of heterosis
in rice as revealed by QTL analysis using molecular markers. Genetics 140:745-754.
Yuan, L.P. 1992. Development and prospects of hybrid rice breeding. p. 97-105. In C.B. You and Z.L.
Chen (ed.) Agricultural biotechnology: Proc. Asian Pacific Conf. on Agricultural
Biotechnology, China Agriculture Press, Beijing.
Zhang, Q., YJ. Gao, M.A. Saghai Maroof, S.H. Yang, and J.X. Li. 1995. Molecular divergence and
hybrid performance in rice. Molec. Breed. 1: 133-142.
Zhang, Q., M.A. Saghai Maroof, T.Y. Lu, and B.Z. Shen. 1992. Genetic diversity and differentiation
of indica andjaponica rice detected by RFLP analysis. Theor. App!. Genet. 83:495-499.