- 1 -

Module 3
Overview of Plant Breeding



I. Introduction
II. Setting Breeding Goals
III. Mega Environments
IV. Choosing Parents and Types of Crosses
a. Single cross
b. Top and double cross
c. Backcross
d. Single backcross
e. Synthetics and Wide Crosses
f. Recording Pedigrees
V. Making the Cross
VI. Selection Methods Following Hybridization
a. Breeding populations
b. Heritability
c. Selection procedures
d. Genetic advance
VII. Genotype by Environment Interaction




- 2 -
I. Introduction

Plant Breeding is the science, art, and business of improving plants for human
benefit (Bernardo, 2002; Breeding for Quantitative Traits in Plants). The science of
plant breeding is based on genetics and the understanding of genes and heredity; but
knowledge of other sciences surrounding plants (i.e. anatomy, physiology, biochemistry,
soil science, pathology, and statistics) is necessary in order to be a successful plant
breeder. The art of plant breeding revolves around the breeders savvy and keenness in
observation, their intuition and judgment. The business of plant breeding is the
management of resources (money, people, land, and time) in an effort to maximize
returns. These three facets of plant breeding: science, art and business; must be used with
an integrative approach to create and recognize that which forms the basis of plant
breeding: variation.




Variation, among individuals of a population, is due to genetic and environmental
effects. The genetic component of variation is provided by hybridization and
recombination. The genetic variation observed in a population throughout a breeding
program will depend on the parents used for hybridization, the selection procedure and
the selection intensity, and on the interaction with the environment. In some cases, it will
be desirable to limit the environmental effects on variation, but in some cases it will be
beneficial to exaggerate an environmental effect in order to better select for superior
genotypes. The goals of the breeding program will dictate the degree to which genetic
and environmental effects should be manipulated. It is the job of the plant breeder, in

The diversity of the phenomena of nature is so great,
and the treasures hidden in the heavens so rich, precisely
in order that the human mind shall never be lacking in
fresh nourishment.
Johannes Kepler

The secret of improved plant breeding, apart from
scientific knowledge, is love.
Luther Burbank

Successful plant breeding depends on recognition of the natural diversity that can be
found within the population. A plant breeder must attain an intimate knowledge of
her crop. Without love, one wheat is the same as the next; with love, the nuance
among individuals is made obvious.
- 3 -
order to reach the goal, to explain and exploit these genetic and environmental effects
that account for observable variation.


II. Setting Breeding Goals

Before a single seed is sown and before a single cross is made a set of objectives
must be established. In order to properly develop a set of breeding goals the breeder
must understand the needs of his clients: the farmers, the millers, and the bakers. The
breeder must understand the target environment, and the levels of biotic and abiotic stress
that occur in that environment over each year and over multiple years. The breeder must
understand the production practices of the farmers. The breeder must have knowledge of
the economic infrastructure; are inputs such as fertilizers and pesticides readily available
to the producer, and what product the grain will be processed into. The breeder must
understand all these things and more if he is to properly develop an improved cultivar
that will be adopted.

The CIMMYT Wheat Program currently distributes advanced bread wheat lines
to more than 60 countries. Primary clients are the national agricultural research systems,
both private and public. CIMMYT breeding objectives attempt to address specific
problems and limitations associated with wheat production in these countries.

The primary goal of CIMMYT is to develop broadly adapted, high yielding
germplasm that has high yield stability, durable disease resistance, and acceptable end-
use quality. Soil type, temperature, and moisture levels greatly influence crop stability
and productivity. In order to accomplish this primary objective over many diverse
environments, CIMMYT has divided the wheat growing areas of the world into 12
distinct mega-environments (ME), delineated based on water availability, soil type,
temperature regime, production system, and associated biotic and abiotic stresses.
Breeding objectives have been identified for each mega-environment:



ME1: Irrigated, low rainfall environment (Spring Wheat) ME1 represents the
optimally irrigated, low rainfall areas of the world. The climate during the
wheat growing period ranges from temperate to conditions of late heat stress.
This ME encompasses about 36 million hectares spread primarily over Asia

Spring Wheat - No vernalization, no photoperiod requirements (except ME6)

Facultative Wheat - Some vernalization/photoperiod requirements for optimal
growth

Winter Wheat - Vernalization and photoperiod requirements + cold tolerance

- 4 -
and Africa between 35S - 35N latitudes. Breeding Objectives include high
yield potential, yield stability, semi-dwarf stature, input responsiveness,
lodging tolerance, improved industrial and end-use quality, and durable
resistance to the three rusts (leaf rust, stripe rust, and stem rust). Certain areas
require resistance to Karnal Bunt, powdery mildew, and fusarium head blight.
Most areas require photoperiod insensitivity, but with a range of flowering
dates. Some tolerance for late heat is also needed for certain locations.
Greater emphasis will be given to tolerance to saline soils. White (amber)-
grained types predominate in the vast majority of areas.


ME2: High rainfall environment (Spring Wheat) ME2 is defined as those areas
with average rainfall in excess of 500 mm during the cropping cycle. Total
area exceeds 8 million hectares. Stripe rust, leaf rust, Septoria tritici, and pre-
harvest sprouting are major production constraints. Fusarium head blight is
becoming more widespread and is a serious problem in many areas.
Resistance to barley yellow dwarf virus, bacteria, powdery mildew and the root
rots must also be considered in certain regions of ME2. Tolerance to lodging,
shattering, and soil micronutrient imbalances are becoming more important.
For high yield potential, semi-dwarf stature is essential. Photoperiod
insensitivity is preferred but with a wide range of flowering dates. Red-
grained wheat provides better sprouting tolerance than white grained wheat,
and is therefore generally preferred. Demands for better industrial quality are
increasing, and emphasis is given to the development of germplasm with high
quality.
ME1:
- 5 -


ME3: High rainfall, acid soil environment (Spring Wheat) The total estimated
area is close to 2 million hectares. Disease and stress problems are similar to
ME2 along with rice blast. Also, aluminum and manganese toxicities, plus
phosphorus deficiency, are major constraints to production. Red grain is
generally preferred, except in the Himalayas. High-level quality is demanded
especially in Latin America.


ME4: Semi-arid, drought environment (Spring Wheat) Three distinct types of
drought have been identified based on the stage of plant development at which
drought is most severe. These are:
ME4A: Winter rain or Mediterranean-type drought associated with post-
flowering moisture and heat stress typical of the West Asia and
North Africa region. Representative locations include Aleppo
(Syria) and Settat (Morocco). Total estimated area: 6 million
hectares. Stripe Rust, leaf rust, stem rust, Septoria blotch, root rots,
ME2:
ME3:
- 6 -
nematodes, and bunts are the key biotic constraints. Also late frosts
may occur. Photoperiod insensitivity is preferred but with a wide
range in flowering dates. White-grained wheat with good quality is
preferred.
ME4B: Winter drought or Southern Cone-type rainfall associated with pre-
flowering moisture stress. Marcos Juarez (Argentina) is a
representative location. Total estimated area is 3 million hectares.
Resistance to leaf rust, stem rust, stripe rust, Septoria spp, fusarium
spp. and tan spot are requirements. Pre-harvest sprouting is also a
common problem, due to late rains; hence red seeded varieties are
preferred. Photoperiod insensitivity is preferred but with a wide
range in flowering dates. Good bread-making quality is demanded.
ME4C: Stored moisture after monsoon rains results in continuous or
Subcontinent-type drought under receding moisture conditions. A
representative location is Dharwar (India). Total estimated area is 2-
3 million hectares, and probably decreasing, as irrigation facilities
spread and/or other crop options are explored. Leaf rust occurs
occasionally. Photoperiod insensitivity is preferred but with a wide
range in flowering dates. Seed must be large, bold and amber in
color to fetch a premium at the market place. Medium dough
strength is required.

The breeding approach of CIMMYT attempts to combine high yield potential
with drought resistance for ME4. The combination of water-use efficiency
and water responsive traits plus yield potential is important in drought
environments where rainfall is frequently erratic across years. When rains are
significantly above average in certain years, the crop must respond
appropriately (water responsive) with higher yields, while expressing
resistance to the wider suite of diseases that appear under more favorable
conditions.


ME5: Warm humid tropical environment (Spring Wheat) These are generally
located at low altitude (<1000 m), with a mean minimum temperature during
the coolest month greater than 17C. As a result relative humidity tends to be
ME4:
- 7 -
high. The estimated area is about 9 million hectares. In these warm, humid
locations, resistance to common root rot, and leaf rust, and Helminthosporium
blight, along with tolerance to sprouting are major objectives. Photoperiod
insensitivity is preferred with an emphasis on medium to early maturity. Input
responsiveness is also an objective. Quality demands vary from moderate to
high primarily using white grains.


ME6: High latitude environment (Spring Wheat) The higher latitude (>45N or
45S) requires materials that have a degree of photoperiod sensitivity, which is
different to other spring wheat MEs. Wheat is spring-sown in this ME because
winters are too severe for survival. Typically this ME experiences pre-anthesis
drought followed by rainfall during flowering and grain-filling. Resistance to
Fusarium spp., tan spot, leaf rust, stem rust, and stripe rust along with
tolerance to sprouting are breeding objectives in this environment. Very dry
representatives of ME6 also occur in central and northern Kazakhstan (10
million hectares) and the southern Siberian wheat belt (8 million hectares).
The major diseases are leaf rust and root rots. In addition, drought tolerance
specific to the rainfall pattern in the region is needed. Taller wheat generally
does better under these conditions. The total estimated ME6 area in
developing countries is 20 million hectares. Wheat with high protein and
strong dough are required.


ME7: Irrigated, moderately cold environment (Facultative Wheat) In these
irrigated regions temperatures vary from 0C to 5C. Yield potential, semi-
ME5:
ME6:
- 8 -
dwarf stature, input responsiveness, and disease resistance are the main
breeding objectives.
ME7A: Fully irrigated. The major diseases are stripe rust, leaf rust, and
powdery mildew. A representative location is Zhenzhou, Henan
(China).
ME7B: Supplementary irrigation. Representative countries are certain regions
in Turkey, and Iran with stripe rust, and bunt as the major diseases;
and in the Central Asian Republics with leaf rust, common bunts,
stripe rust and loose smut as the main diseases.


ME8: High rainfall, moderately cold environment (Facultative Wheat) Rainfall
on average exceeds 600 mm per crop cycle, with temperatures varying
between 0C and 5C. Objectives include input responsiveness and semi-
dwarf stature.
ME8A: Photoperiod-sensitive varieties are cultivated. The major diseases
are stripe rust, leaf rust Septoria spp., powdery mildew, Fusarium
spp., and root rot. A representative location is Temuco (Chile).
ME8B: Photoperiod insensitive varieties are cultivated. Representative
regions are the transitional spring/winter wheat zones in Turkey with
stripe rust and common bunt being the major diseases; and Thrace
(Turkey) with leaf rust, root rot, powdery mildew ad common bunts
as the main diseases. Other representative areas include certain
regions on the Central Asian Republics where stripe rust, leaf rust,
root rot, powdery mildew and common bunt are the main diseases.
ME7:
- 9 -


ME9: Semi-arid, moderately cold environment (Facultative Wheat) Rainfall is
moderate to low (<400 mm), and temperatures are in the range of 0C to 5C.
Drought tolerance is required and input efficiency an objective.
ME9A: Heat stress at grain-filling. Mainly non-semidwarf varieties
cultivated. Areas in West Asia and North Africa and the Central
Asian Republics are representative with stripe rust; common bunt
and leaf rust the major diseases.
ME9B: Mainly semidwarf varieties cultivated. The major disease is leaf rust.
Representative locations are certain regions in China.
ME9C: Less cold tolerance required than in 9A and 9B. Mainly semidwarf
varieties cultivated. Areas in South America and South Africa are
representative with leaf rust and Russian wheat aphid the major
biotic stresses.


ME10: Irrigated, severely cold environment (Winter Wheat) These irrigated areas
experience very cold conditions with temperatures dropping to between 0C
ME8:
ME9:
- 10 -
and -10C. High yield potential, semi-dwarf stature, cold tolerance, input
responsiveness and disease resistance are requirements.
ME10A: Fully irrigated. The major diseases are stripe rust, leaf rust,
powdery mildew, barley yellow dwarf virus and nematodes. A
representative location is Beijing (China).
ME10B: Supplemental irrigation. Specific locations in Turkey and Iran are
representative. Stripe rust, common bunt, root rot, and nematodes
are the major diseases. Areas in Western and North-Western Iran
where stripe rust and common bunts predominate; and in the
Central Asian Republics where leaf rust, common bunt,
nematodes, and powdery mildew predominate, are also
representative.

ME11: High rainfall, severely cold environment (Winter Wheat) Rainfall on
average in excess of 600 mm is common, while temperatures are low, between
0C and -10C. Semi-dwarf stature and cold tolerance are desirable traits.
ME11A: Photoperiod-sensitive varieties cultivated. The major disease are
stripe rust, leaf rust, powdery mildew, Septoria spp., Fusarium
spp., barley yellow dwarf virus, and root rot. No representative
locations occur among the developing countries. However, other
example locations are Martonvasar (Hungary), Krasnodar (Russia),
and Odessa (Ukraine).
ME11B: Photoperiod-insensitive varieties cultivated. The major diseases are
leaf rust, Fusarium spp., Septoria spp., root rot, Barley yellow
dwarf virus, and stripe rust. A representative country is North
Korea.


ME12: Semi-arid, severely cold environment (Winter wheat) Conditions are
harsh, with little rainfall (300-450 mm); temperatures vary from 0C to -10C.
Drought tolerance is needed, cold tolerance, and zinc deficiency tolerance are
objectives.
ME10:
ME11:
- 11 -
ME12A: Heat stress at grain-filling. Mainly non-semidwarf varieties are
cultivated in certain representative regions in Turkey, Iran, and
Afghanistan. In these areas stripe rust, common bunt, root rot,
nematodes, and zinc deficiency are the major biotic and abiotic
stresses. In the Central Asian Republics stripe rust, common bunt,
and leaf rust are the main diseases. In representative areas of
China, semidwarf varieties predominate, with stripe rust and
powdery mildew the key diseases.
ME12B: Medium heat stress at grain-filling. Mainly non-semidwarf
varieties are cultivated in certain representative regions in Turkey,
and Iran with stripe rust and common bunt the major diseases. In
representative regions of China semidwarf varieties are generally
cultivated, and stripe rust and common bunt are the main diseases.


Breeding goals will almost always revolve around yield and yield stability,
resistance to one or more environmental stresses (winter hardiness, drought tolerance,
nutrient deficiency or toxicity, disease), and quality. As every wheat breeding program
has its financial limits, it is important to stay within the means of the program when
deciding on the objectives. The numbers of individuals of a segregating population
necessary to select from in order to improve a population can quickly get out of hand if
too many traits are placed in the list of breeding objectives.

Consider a trait controlled by a single locus with alleles A
1
and A
2
, and the
desired phenotype has the genotype A
1
A
1
. The genotypic ratio among inbreds will be
50% A
1
A
1
and 50%A
2
A
2
. The probability of an inbred with the ideal genotype for this
simple trait will be:
1
( )
1
0.50
2
Ideal Genotype
P
| |
= =
|
\
, and the population size (N) required to
have the ideal genotype will be
1
2 2 N = = . However, in considering a trait such as yield,
that is controlled by multiple traits, let us say 20, the probability of having the ideal
genotype for this complex trait will be:
20
( )
1
0.00000095
2
Ideal Genotype
P
| |
=
|
\
, and the
population size required to have the ideal genotype will be
20
2 1, 048, 576 N = = ; not an
easy number of individuals to either maintain or screen.



ME12:
- 12 -
III. Choosing Parents and Types of Crosses
After the goals of the program are established, the next step will be to create the
variability for the traits that are to be improved. Sources of germplasm for the crosses
will need to be identified but locally adapted material, CIMMYT developed cultivars for
the target ME, and exotic sources should be considered. Also, a decision on the best type
of cross to carry out must be made:

Single Crosses (Figure 1)
Single crosses are the cross of a cultivar or line with another cultivar or line. In some
cases, a promising segregating line may be used as one or both parents.

Procedure
Select the female parents utilizing your objectives and your knowledge of the materials
for the various crosses. It is necessary to maintain and enhance the variability to have an
effective crossing program. Therefore, select several different lines as females for each
one of your projected crosses, i.e. several with good rust resistance, others with short
straw, others with excellent yield, etc. After the females have been selected choose the
male parents. When selecting male parents you should try to complement the female
parent, i.e. the male parent should have the good characteristics that are lacking in the
female parent. For example, a line has been chosen for its resistance to Septoria but it is
tall, susceptible to stem rust, and has poor yield potential. The male parent can be
susceptible to Septoria but, it must be a dwarf, must have resistance to stem rust, and
should have excellent yield potential. Making crosses in this manner will increase the
chance of obtaining segregates with all of the desirable agronomic characteristics. Do not
cross close sisters!
Every cross should try to obtain lines which are better than the existing cultivars.
To accomplish this, the parents of each cross must be analyzed (mentally) for the
complete group of desirable agronomic characteristics. Never make a cross with only one
objective, for example:
To obtain a line resistant to Septoria regardless of the other characteristics as this
may in fact produce progeny that are resistant to Septoria, but the progeny have a
poor chance of being agronomically acceptable. A cross may be made for
Septoria resistance but the selected progeny must also have resistance to the other
prevalent diseases and good yield potential etc. A line with Septoria resistance is
of no benefit to the farmer if it will not yield.

- 13 -


Top and Double Crosses (Figure 2)
Top or three-way cross is the cross of an F
1
to a variety or line. Double cross is the
mating of two F
l
's.

Procedures
In addition to the knowledge of the various varieties and lines, it is also necessary
to have agronomic notes on the various F
l
s. A general idea of the maturity, height, straw
strength and head characteristics must be noted before crossing begins.
The purpose of using top and double crosses is to increase the chance of obtaining
a desirable gene or genes from exotic or difficult materials. Exotic refers to lines from
other countries which are generally poorly adapted to local conditions. Difficult material
refers to varieties or lines which are tall, poor combiners, or dominant susceptibles, etc.
i.e. lines which have given poor results (progeny) from single crosses in previous
crossing cycles. As in the case of single crosses, parents should be chosen to increase
variability and to complement agronomic traits.
The F
1
which will be the female parent should be chosen first, again following
your crossing objectives. After emasculation, the male must be chosen. In making top and
double crosses, only single cross F
l
's are utilized because they are uniform. The top and
double cross F
l
's will be segregating and it is impossible to identify superior plants at
crossing; therefore, they are not used. The F
l
's are selected for desirable agronomic
characteristics or for desirable parentage. For example:
P1 P2
Parent 1:

Good Yield
Resistance to Fusarium
Moderately Susceptible to Root Rot
Moderate Gluten Strength
Parent 2: Excellent Yield
Susceptible to Fusarium
Resistance to Root Rot
Strong Gluten Strength
Breeding Goal: Excellent Yield, Resistance to Fusarium and Root Rot, and Strong Gluten Strength
X
self
self
self
{
Segregating Population
IC
Improved Cultivar:
Excellent Yield
Resistance to Fusarium
Resistance to Root Rot
Strong Gluten Strength
Fig. 1 - Single Cross
- 14 -
An F
1
made between a tall, poor yielding line with excellent leaf rust resistance
and a semi-dwarf line with excellent yield but poor resistance is chosen. To
enhance the probability of obtaining all of the desirable characteristics, top or
double cross this F
1
to a line or F
1
with the desirable characteristics but of
different parentage to increase variability.
Another example could be the objective of broadening the base of stem rust
resistance:
Line A, tall, poor yield, poor quality and excellent stem rust resistance was
crossed to line B short, good yield, good quality and good stem rust resistance.
Since line A had many negative values, it probably would be desirable to top or
double cross this F
1
. Therefore, select a line of F
1
which is short, has good yield,
good quality and good resistance.




The Choosing for Top or Double Crosses
The question of whether or not to top or double cross an F
1
can frequently be
answered by the amount of labor available. Top crosses require 5 heads per cross and
double crosses require 8-10 heads per cross. This number is necessary because these
crosses will segregate in the next F
1
generation and at least 80 and 150 plants are required
from top and double crosses respectively, to facilitate the selection of desirable plants in
the F
1
. Therefore, if labor is expensive top crosses should be utilized more than double
crosses.
Another criterion in solving the question of whether to top or double cross is the
amount of variability that you think will be necessary to obtain desirable phenotypes.
Double crosses segregate more widely than the top cross in the F
2
. Double crosses
PA PB PC PD
X
F
1
AB
X
F
1
CD
X
self
self
self

(In the case of a
top cross this
circle would
represent parent
C)
Parent A:
Tall
Poor Yield
LR ResistanceB
Parent B:
Semidwarf
Excellent Yield
Poor LR Resistance
Parent C:

Tall
Poor Yield
LR resistanceA
Parent D:
Semidwarf
Excellent Yield
Moderate LR
ResistanceC
IC
Improved Cultivar:
Semidwarf
Excellent Yield
Durable LR ResistanceABC
Fig. 2 - Double Cross
- 15 -
involve 4 different parents, generally 3 desirable and one less desirable, while the top
cross involves 3 parents, 2 desirable and one less desirable parent. Therefore, the decision
to top or double cross depends on the number of poor features in the exotic lines.
Frequently, an exotic line has only one desirable trait. If this is the case, single crosses
will be ineffective and top or double crosses must be utilized to increase the proportion of
desirable segregates.

Backcross (Figure 3)
It may occur that a certain cultivar is both adapted to and productive in a certain
region, but lacks a desirable allele for a single trait. This desired trait is found in other
cultivars, but these cultivars perform very poorly overall in the target environment. The
goal of the back cross is to introduce the exotic allele or alleles into the adapted varieties
background. Beginning in the F
1
and continuing for several generations, hybrid plants
containing the desired allele are selected and successively crossed back to the adapted
parent cultivar. The adapted cultivar, to which the successive crosses are being made to,
is known as the recurrent parent. The other parent, containing the desired allele that is
used in the initial cross is called the donor parent. The number of backcrosses used
depends on how completely the breeder wishes to recover the recurrent parents genetic
background. This varies from between two to five, or more; but in theory, after the fifth
cross 96.875% of the recurrent parents genes will be recovered in the backcross progeny.
The backcross procedure is most easily carried out if the trait being added is
simply inherited, dominant, and easily recognized in the hybrid plants (Pohlmen and
Sleper, 1995). If the desired phenotype occurs with the homozygous recessive genotype
(aa), the trait will not be identifiable after the first cross as the progeny will be
segregating with genotypes (AA) and (Aa). It is necessary to self the progeny one
generation in order to identify the desired genotype (aa) before making the next
backcross.
Backcross procedures are well established and straight forward, and results are
predictable. It can be very efficient as small numbers of plants can be grown in the first
generations of backcrossing, so it can be the least expensive way of improving a cultivar,
or developing improved parent material. Recurrent parents are often identified in locally
adapted material, where a single weakness is frequently recognized. This method is a
rapid and easy response to the weakness. In a dynamic and progressive breeding
program backcross methods may not be a way to realize the breeding objectives as by the
time the backcross generations have been completed (perhaps 5 or 6 years), the recurrent
parent of the backcross may no longer be looked at as an elite cultivar, and the product of
the backcross procedure may likely perform worse than the cultivars that have been
produced using the single cross or double cross methods. Figure 3 details the
backcrossing method to introduce a dominant allele (RR) into a recurrent parents genetic
background.
- 16 -
.
75% genes
from A
50% genes
from A
Selfing
Cultivar
B
(RR)
Cultivar
A
(rr)


F1
(Rr)

Cultivar
A
(rr)

Cultivar
A
(rr)

Cultivar
A
(rr)

Cultivar
A
(rr)

X
X
X
X
X
Discard (rr)
Discard (rr)
Discard (rr)
Discard (rr)
(Rr)



(rr
)
(Rr)



(rr
)
(Rr)



(rr
)
(Rr)



(rr
)
1
st
Backcross
2
nd
Backcross
3
rd
Backcross
4
th
Backcross
F1 of BC4
87.5 % genes
from A
93.75 % genes
from A
96.875 % genes
from A
Discard (Rr) on basis of progeny test
F2 of BC4
Discard (rr)


(RR)


(Rr)


(rr)
Fig. 3 - Diagram of the backcross method. The average proportion of alleles
from cultivar B (the donor parent) in the hybrid population is reduced by one-
half with each backcross. Because R is dominant to r, plants of genotype Rr can
be recognized after each backcross and used for the next backcross. Plants of
genotype rr are discarded. At the end of the process the selected progeny will be
over 96% genetically similar to cultivar A (the recurrent parent) and carry also
the donor parents RR alleles
- 17 -
CIMMYT and the Single Backcross Strategy
CIMMYT breeders typically use a single backcross strategy. This strategy allows
the breeder to create a breeding population that maintains most characteristics of certain
cultivar (the recurrent parent), but still allows selection for several new genes. In effect
the single backcross strategy enables three things to occur:
1. Increase the probability of maintaining and reselecting desirable genes of the
recurrent parent.
2. Allows for the transfer of multiple genes or characters simultaneously.
3. Allows additional genes or characters from the donor parent to be selected for.
The basic method of a single backcross at CIMMYT is to cross an adapted cultivar with
6-10 donor parents that are selected based on the needs identified in the breeding
objective. F
1
plants are, from each of the 6-10 crosses, then backcrossed to the adapted
cultivar with the goal of developing 400-500 BC
1
seeds. It has been shown that this
crossing method has been successful in creating breeding populations with a higher mean
value for yield than through traditional simple or three way crosses (Fig. 4).

0
5
10
15
20
25
<
6
0
6
0
-
6
5
6
5
-
7
0
7
0
-
7
5
7
5
-
8
0
8
0
-
8
5
8
5
-
9
0
9
0
-
9
5
9
5
-
1
0
0
1
0
0
-
1
0
5
1
0
5
-
1
1
0
1
1
0
-
1
1
5
Grain yield (% Kambara )
%

L
i
n
e
s
Single back cross Traditional
N = 4088
N = 726
0.8% > Check
10.7% > Check
Fig. 4 - Grain yields of wheat lines developed through traditional (Simple
and 3-way crosses) and single-backcross approach
Cd. Obregon 2004-2005
- 18 -
Synthetics and wide crosses
CIMMYT scientists have been crossing durum wheat with wild relatives of
wheat, such as goat grass (Aegelops tauschii), since the early 1990s. This cross creates
synthetic wheats, which can easily be crossed with improved varieties to incorporate new,
useful genes. The resulting wheats are improved varieties that also have desirable traits
from the wild parents. CIMMYT has produced synthetic wheats and their derivatives
with traits such as resistance to Septoria spp. and Fusarium spp. head blight and also
tolerance to drought, heat, salt, or waterlogging.
Researchers in China recently crossed CIMMYT synthetic wheats with local
wheats and released the results to farmers in 2003. Breeders in Sichuan province have
been using the CIMMYT-developed synthetic hexaploid wheat since 1995 to improve
quality, yield potential, and disease resistance. After crossing and backcrossing this wheat
with high-yielding local varieties, they have developed several lines and are currently
testing five more.
The synthetic wheats pass on beneficial traits such as large kernels, heavy spikes,
and resistance to new races of Chinese stripe rust. During two years of yield trials, the
two varieties derived from synthetic wheats had 20% to 35% higher yields than the
commercial check variety. One of these varieties, named Chuanmai42, had the highest
average yields more than six tons per hectare in the trials. Since it was released in
Sichuan in 2003, Chuanmai42 has been recommended by the government to farmers and
has been delivered to most wheat breeding research programs in China.
In 2003, Spain registered a CIMMYT synthetic wheat derivative under the name
Carmona. This fast-growing variety matures and provides seed in a shorter period than
most commercial cultivars, which is valuable for wheat growers who often plant late in
the year in southern Spain. Carmona has better grain quality and is suited to zero-tillage
systems, where it resists foliar diseases and produces higher yields.
Wide crosses involve transferring alien genes into wheat. An alien gene can be
considered to be any gene transferred to wheat from a related species (Knott D., Wheat
and Wheat Improvement). The genus Triticum contains a broad range of species some of
which cross readily and some of which cross with great difficulty to bread and durum
wheats. Wheat will also cross with species in a number if different genera including:
Agropyron, Elymus, Elytrigia, Haynaldia, Hordeum, Leymus, and Secale. These relatives
of wheat are adapted to a large range of environments and are potential sources for useful
genes such as: disease resistance, cold tolerance, salt tolerance, drought tolerance,
lodging resistance, early maturity, and yield. Wheat 1B/1R translocation, a common
alien transfer, is the wheat translocation line in which the short arm of 1B chromosome in
wheat is replaced by the short arm of 1R chromosome in rye. Because of the
characteristic of disease resistance, high and stable yield and broad adaptation, it has been
widely planted around the world. In China the proportion of 1B/1R translocation lines has
reached 45% in major extended cultivars of main wheat growing areas in the past thirty
years and partial recently bred lines (Chai J et al (2006) Plant Genomics in China VII).
Synthetics and Wide Crosses along with Hybrid Wheat are discussed further in the
module: Biotechnology and Wheat Breeding.
- 19 -

Recording Pedigrees
Pedigrees provide the parentage or the sequence through which a cultivar was
obtained and are important sources of information. The pedigree for each line is
generally entered into field books, crossing books, and on crossing tags. At present,
CIMMYT uses the United States Department of Agriculture (USDA) system of
designating pedigrees. This system is widely used around the world, making it easier to
trade data.

This system consists of just a few basic rules:
1. The female parent is written first and is separated from the male parent by:
a. A single slash / in the first cross
Example: Logan / Heart
b. A double slash // in the second cross
Example: Logan / Heart // 3270-A / Rusalka
c. Two slashes with the number of the cross between them, /3/, /4/, /5/, etc
in the subsequent crosses.
Example: Logan / Heart // 3270-A / Rusalka /3/ Tn-1685 / IA-22 // 6767 / 216-6-3
2. The parental material involved in any particular cross includes all that is listed on
either side of the highest number of crosses in the pedigree.
Parent 1 pedigree: Logan / Heart // 3270-A / Ruskalka
Parent 2 pedigree: Tn-1685 / IA-22 // 6767 / 216-6-3


3. Backcrosses are indicated with an asterisk (*) and a number indicating the number
of times the recurrent parent was used. The asterisk and the number are placed
next to the crossing symbol that divides the recurrent parent and donor parents.
Examples:
LR 64 / 3* Son64
Son 64 / *2 TZPP

IV. Making the Crosses
There are some important things to consider before carrying out the crossing of
two cultivars. The plants to be used as the females should be emasculated slightly (3 to 4
days) before dehiscence of the male plants. This can become complicated if the cultivars
to be used have different heading dates. If the wheat to be crossed is a winter wheat, it
will need to vernalized for usually 8 weeks. Emasculation is a time consuming process
that takes patience, and concentration; if too many emasculations need to be done in a
single day, anthers will be left behind and self fertilization will occur. A way to avoid
HOPEWELL
LOGAN HART 3270-A RUSALKA TN-1685 IA-22 6767 216-6-3
- 20 -
some of these potential problems is to plant more than you will need of the selected
parents over a number of days, many of these plants will then not be used but it will
ensure that both the female parent and the male parent will be at the proper flowering
stage when you are ready. Figure 5 shows a wheat spikelet and may be useful to those
unfamiliar with emasculation.



In order to make a cross the breeder will need some basic materials: Small, sharp,
fine quality scissors; forceps or pincers with somewhat blunted ends; 3x5 cm tags with an
attached string; 5x15 cm glassine bags; a wax pencil; a permanent marker; staples; paper
clips; rubber bands; and a crossing block book.





- 21 -
Preparing Crossing Tags

Figure 6
1. Assembly and Preparation of Tags
a) Ready tags before crossing starts
b) On side 1, identify year and location (MV-07)
c) On side 1, write or stamp date of emasculation (VII-31), and date of pollination
(VIII-2)
d) The initials of the individual doing the crossing are sometimes added as well.
2. Identify female Parent
a) On the top half of side 2, write the nursery plot number (CB HARI, 1), the name
and pedigree of the female parent (Glennson).
3. Identify the male parent
a) Usually it is easier to prepare all of the needed tags for each female first and then
to add the name and pedigree of the male. Select the male parents from the
nurseries for each cross (CB HARI 15). Place the name and pedigree of the male
parent on the lower half of the tag (BOW S; CM 33203).
b) The complete pedigree must be made on at least one tag. At CIMMYT, one
complete tag is made for single crosses, two for top crosses and three for double
crosses. You can abbreviate the name on additional tags, but be sure you can
positively identify the parents.

Steps of Emasculation:

1. Select Lines and Varieties to Be Emasculate
a) The number of spikes to be emasculated will depend upon the genetic stability,
(more for segregating materials in top and double crosses) seed set, and the
desired size of the F
1
population.
b) Specify the number of spikes to be emasculated for each cross
i) For each single cross: 3-6
ii) For each top or double cross: 8-13
2. Select Individual Plants and Spikes for Emasculation
a) Individual plants should be healthy and representative of the variety or line.
b) Do not emasculate more than one spike per plant.
c) Select spikes which are almost fully exerted from the flag leaf.
d) Do not select spikes which have begun anthesis. Emasculation should be done 3
to 4 days before expected dehiscence.
- 22 -
e) Carefully separate the selected spike from those surrounding it in order to make
work easier, being careful not to break the stem.
3. Remove Upper and Lower Spikelets
a) With your scissors or forceps, remove one or two spikelets from both the bottom
and the top of the spike. These spikelets are removed because they mature more
slowly than the middle spikelets and very often aborted or produce smaller seeds.
b) Keep 15 to 20 spikelets per plant.
4. Remove the Center Florets
a) With your forceps pull out the center florets of each spikelet, leaving only florets
1 and 2.
b) This is done for two reasons:
i) It makes the emasculation much easier.
ii) The ovaries of these florets do not mature at the same time.
5. Cut Back the Lemma and Palea
a) With your scissors cut the glumes, lemma, and palea just above the top of the
stigma
b) Do not injure the stigma or the anthers by cutting too low.
c) If the cut is too high, emasculation is more difficult and pollen may not reach the
stigma as the palea and lemma will remain folded over the stigma.
** Steps 4 and 5 may be reversed if it is more convenient for the emasculator **
6. Remove the Anthers
a) Carefully remove the three anthers from each floret with the forceps.
b) With practice, you will be able to remove all three florets at one time.
c) Remove the anthers carefully. Do not break them, as pollen may be shed causing
self pollination.
d) The extracted anthers may stick to the forceps. An easy clean way of removing
the anthers from the forceps is to eat them.
e) Do not injure the stigma with forceps.
f) At first, you will want to check each floret after you have completed the spike;
with experience, double checking may not be necessary.
7. Cover the Spike with Glassine Bag
a) Mark the date and your initials on the glassine bag
b) Place bag over spike.
c) Fold bottom edge over and staple or place a paper clip on fold so that it is tight
against the peduncle or stem to prevent wind from blowing off the bag.
8. Keep record of which plants have been emasculated in your crossing block book.
This way you will know when it is time for pollination.

Pollination by the Twirl Method

1. When to Begin Pollination
a) Depending on temperature and humidity, the female parent will be ready for
pollination two to four days after emasculation. The female is receptive when the
stigma is enlarged and feathery.
b) A crossing tag, representing the cross that is to be made, should be made the day
before pollination will take place.
- 23 -
2. Gather Pollen
a) Take one spike, as a pollen source, for each tag.
b) Select spikes from healthy plants.
c) Spikes should just be ready to flower one or two anthers may show.
d) Pull the spike from the stem; if it is ready to flower, it will separate easily.
3. Tie the Spikes Together
a) Group all spikes of the same male parent, and fasten with a rubber band.
b) Attach appropriate tags to each group of spikes.
c) Clip peduncles to uniform length if desired.
4. Place Spikes in Water
a) Place the spikes in water and put the container in the shade.
b) Spikes may be kept for four to five hours.
c) Be sure to maintain the identity of each pollen parent.
5. Clip the lemma and palea
a) Clip the lemma and palea just above the anthers. This cut will be somewhat
higher than the cut for emasculation.
b) Clip the number of spikes needed for each cross to a particular female.
6. Place the Spike in the Sun
a) Place the peduncles in the ground in a sunny place protected from the wind; or if
it is cold, hold the spike in your cupped hands and blow on it gently until anthers
emerge.
b) While the other spikes are clipped the anthers of the first spikes will extrude and
are then ready to be used for pollination.
7. Cut Open the Top of the Glassine Bag
a) With the scissors cut open the top of the glassine bag of the selected female
parent.
b) Blow into the bag to open it.
c) Keep open only long enough to pollinate the enclosed spike.

8. Pollinate the Emasculated Spike
a) Carefully place the pollen spike in the bag. Avoid dusting the pollen outside the
bag.
b) Twirl the spike around the female several times, and then discard the pollen spike.
c) One male spike is enough to pollinate the female with just two or three good
anthers extruded
9. Close the Bag
a) Fold the top of the glassine bag over.
b) Place a paper clip on top, or staple to hold in place.
10. Attach Crossing Tag
a) Attach the tag specific to each cross to peduncle.
b) Place the tag under the paper clip, at the bottom of the glassine bag with side
showing dates of emasculation and pollination out.




- 24 -
V. Selection Methods Following Hybridization

Following hybridization, the breeder must establish and maintain the segregating
population (that ideally has a high mean and a large genetic variance for the desired
traits) from which he will make selections. Selection occurs when some individuals
produce more progeny than others due to a superior genotype. In nature, this is due to
either superior fertility or superior viability; but in a crop, such as wheat, the decision of
which individuals will or will not produce progeny is largely in the hands of the plant
breeder. The job of the breeder is to, with the breeding objectives in mind, select
individual or groups of genotypes such that the population contains all the alleles that
confer the superior phenotype, and none of the alleles that confer the inferior phenotype.

Before selection can be further discussed, it will be important to understand some
things about population genetics, gene action and interaction and heritability. The
presentations, of the above concepts, that are to follow are based on mathematical models
of probability developed by Godfrey Hardy and Wilhelm Weinberg. Although, the
assumptions that are made in order to justify these mathematical models are rarely seen in
the practice of plant breeding (random mating, large population size, no selection, no
mutation, and no immigration or emigration) the concepts are useful in understanding
sources of variation within the crop and will aid the breeder in making sound decisions
during the selection process.

Breeding Populations
The population of plants created by crossing two or more parents is known as a
breeding population, and a population can be described in terms of its genotypic or allele
frequencies. Considering one locus, a population of 900 individuals is a population of
1800 alleles. For example, at a single locus A, an F
1
population derived from two
homozygotes A
1
A
1
A
2
A
2
will have genotype frequencies (P), and allele frequencies (p
the dominant allele and q the recessive allele) of:

Genotype Frequency (P)
A
1
A
1
P
11
= 0.0 p = P
11
+ P
12
= 0.5
A
1
A
2
P
12
= 1.0

A
2
A
2
P
22
= 0.0 q = P
22
+ P
12
= 0.5
The sum of allele frequencies at any locus is p + q = 1 (two alleles); and a locus with
three alleles: (p
1
+ p
2
+ p
3
) = 1. A population, after one generation of random mating will
have the allele frequencies of p
2
+ 2pq + q
2
. As an example if a population with the
following genotypes were to randomly mate:
Genotype Number P
ij

A
1
A
1
20 0.2 p = P
11
+ P
12
= 0.4
A
1
A
2
40 0.4
A
2
A
2
40 0.4 q = P
22
+ P
12
= 0.6
total 100 1.0 0.1
The next generations genotype frequency would be:
( ) ( ) ( )
2 2 2
1 1 1 2 2 2
( ) 2 0.16 0.48 0.36 1.0 p q p pq q A A A A A A + = + + = + + = .
- 25 -
Likewise for the three allele locus:
Genotype Number P
ij

A
1
A
1
20 0.2
A
1
A
2
10 0.1 p
1
= P
11
+ P
12
+ P
13
= 0.275
A
1
A
3
5 0.05
A
2
A
2
15 0.15 p
2
= P
22
+ P
12
+ P
23
= 0.35
A
2
A
3
30 .30
A
3
A
3
20 0.2 p
3
= P
33
+ P
31
+ P
32
= 0.375
Total 100 1.0 1.0

and the next generations genotype frequency will be:
( )
( ) ( ) ( ) ( ) ( ) ( )
2
2 2 2
1 2 3 1 1 2 1 3 2 2 3 3
1 1 1 2 1 3 2 2 2 3 3 3
2 2 2
0.076 0.193 0.206 0.123 0.263 0.141
1.0
p p p p p p p p p p p p
A A A A A A A A A A A A
+ + = + + + + +
= + + + + +
=
.


A superior phenotype is a result of both its genotype and the environment which it
is in. The phenotypic value of individual k with the A
i
A
j
genotype can be modeled as
( ) ( )
ij ij k ij k
P G e = +
and as a deviation from the population mean
( ) ( ) ij ij k ij k
P g e = + +
where
( ) ij k
P is the phenotypic value of individual k with the A
i
A
j
genotype, is the
population mean, G
ij
is the value of the A
i
A
j
genotype (g
ij
being G
ij
-), and
( ) ij k
e
represents the value for the non genetic effects (the environment) that individual k with
the A
i
A
j
genotype is exposed to. Value, refers to the measure of the effect on the
quantitative trait. Genotypic values cannot be measured directly but can be estimated
from phenotypic values.

The value of a genotype, A
i
A
j
, can be modeled as a deviation from the midparent
value, P , the average of the two parental values. The value a is given to the A
1
A
1
genotype, the value a is given to the A
2
A
2
genotype, and the value d is given to the A
1
A
2

genotype (see fig.6). The value of d can very depending on the degree of dominance
exhibited by one allele over another. Partial dominance exists when 0<d<a; no
dominance exists when d/a = 0; whereas d/a = 1 indicates complete dominance. Over
dominance exists when d>a.




- 26 -


Figure 6
At a single locus, the mean () of a population is equal to the sum of genotypic values
multiplied by their allele frequencies. As an example let us consider the following
population of 100 individuals with single locus and no dominance of the A
1
allele:

Genotype Number P
ij
Trait Value
A
1
A
1
63 0.63 20 p = P
11
+ P
12
= 0.705
A
1
A
2
15 0.15 18
A
2
A
2
22 0.22 10 q = P
22
+ P
12
= 0.295
Total 100 1.00

Using the coded genotypic values, at a single locus the mean () of the population
following random mating will be:
( ) 2
15
5
3
5
15 5(0.705 0.295) 2(0.705)(0.295)(3)
18.30
P a p q pqd
P
a
d
a

= + +
=
=
=
=
= + +
=

When the trait is controlled by two or more loci that do not interact with each other, the
population mean is the sum of the contributions from each locus:
( ) 2
l l l l l l l
P a p q p q d = + +


where l refers to the lth locus.

Coded genotypic values are reflections of genotypes, not alleles, and it will be
more useful to know the average effect of an allele that is passed on. Ronald Fisher, a
founder of todays statistical science, formulated the average effect of an allele. The
average effect of allele A
1
expressed as a deviation from the population mean, is denoted
by
1
.
- 27 -
( )
( )
1
( ) 2 P pa qd P a p q pqd
q a d q p
( = + + + +

( = +


Likewise, the average effect of the A
2
expressed as a deviation from the population mean,
can be calculated as:
( ) ( )
[ ]
2
2
(
P pd qa P a p q pqd
p a d q p
( = + + +

= +

These equations show that the average effect of an allele is not a property of the allele
itself, but it is a joint property of the allele and the population in which the allele is found.
During the selection process one allele will be favored over another and this implies an
allele substitution. The average effect of an allele substitution (A
1
for A
2
) is equal to:
1 2
( ) a d q p
=
= +
.
Given the above, the genotypic value of A
i
A
j
can be expanded from:
ij ij
G g = + to
ij i j ij
G = + + + ,
where
ij
is the dominance deviation. Dominance deviations imply that the heterozygote
is more like one parent or the other. Dominance deviations are zero in the absence of
dominance, and are modeled by:
( )
11 11 1 1
2
12 12 1 2
22 22 2 2
2
2( )
2
( )
2
2( )
2
G
a pa qd
q d
G
d pa qd pd qa
pqd
G
a pd qa
p d






=
= +
=
=
= +
=
=
=
=

Epistatic effects are the result of interactions of genes affecting a single trait but are
found at different loci. Epistatic effects exist when the total of the effect is not equal to
the sum of the individual allele effects. With that in mind the two-locus model can be
introduced with individual A
i
A
j
B
k
B
l
, and I
ijkl
representing the epistatic effect:
( ) ( )
ijkl i j ij k l kl ijkl
G I = + + + + + + + .

As stated before, variation is the key to plant breeding; and variation within the
population (assuming equal environments) is a function of alleles. It is the goal of the
plant breeder to shift the populations distribution of the trait value by selecting favorable
alleles, and in a sense altering the variation within the breeding population. As the
previous models showed a phenotype is a function of both genotype and environment;

- 28 -


In the same manner phenotypic variance of a population can be explained by genetic
variance and environmental variance. The variance is defined as the mean of the
squared deviations of a random variable from the population mean. The population
variance (
2
) for variable X is equal to the expectation
2
( )
i X
E X , but would be
difficult if not impossible to calculate so the sample variance is generally used:
2
2
( )
1
X
x x
s
n
| |
=
|
|

.
And just as genotypic values could be broken down into a mean along with additive
effects, dominance effects, and epistatic effects; so too can the genotypic variance be
broken down into different variance components and can be useful in the selection
process:

Variance Component Symbol
Phenotypic V
P
Genetic V
G
Additive V
A
Dominance V
D
Epistatic V
I

Genotype x Environment V
GE
Error V

Further:
Fig. 7

Selection for a
quantitative trait
shifts the
population
distribution.

- 29 -
[ ]
[ ]
( )
( ) ( ) ( ) ( ) ( )
( )
( ) ( ) ( ) ( ) ( )
P G e ij k
G A D I i j ij k l kl Iijkl
GE e ij k
P i j ij k l kl Iijkl GE
V V V
V V V V V V V V V
V V V
therefore
V V V V V V V V

+ +
+ +
= +
= + + = + + + +
= +
= + + + + + +

This states that phenotypic variance is a function of (in a two loci system) variance for
the additive and dominant effects for alleles A
1
,

A
2
,

B
1
,

B
2
, their epistatic relationship with
each other, and how they interact in the environment, along with unexplained variance.
Next we will take the theoretical models introduced in the previous pages and put them
towards a few useful applications (For a more inclusive look at population genetics see
Rex Bernardos text: Breeding for Quantitative Traits in Plants from where much of the
information was gathered.


Heritability
Individual plants in a breeding population will ideally vary for the traits identified
in the breeding goals. These will include height, heading and maturity dates, disease
resistance, quality parameters, and yield. Two randomly selected plants will show these
variations but it will not be known whether the differences are due to genotype or the
environment. The plant breeder is interested in identifying how much of the observed
phenotypic variability is due to genetics and to do this she must estimate heritability.
Heritability is defined and estimated in two distinct ways: Broad-sense heritability and
Narrow-sense heritability.

Broad-sense heritability (H) estimates the proportion of the phenotypic variance
that is explained on the basis of all genetic effects:
G P
H V V = .
Calculation of broad-sense heritability can be done by using data to estimate the variance
components (ANOVA). These procedures will not be detailed here but can be found in
statistics and some quantitative genetics books. However, in a cross between two pure
lines an estimate of the broad-sense heritability can be calculated by using the variance of
the F
2
population as V
P
. Environmental variance, or error variance (V
E
), can roughly be
estimated by using a uniform population grown in proximity to the segregating F
2
population since:
P G E
G P E
V V V
V V V
= +
=



Narrow-sense heritability (h
2
) estimates the proportion of the phenotypic variance
that is explained by only the additive effects:
- 30 -
2
A P
h V V = .
Narrow-sense heritability can be estimated by finding the slope of a line from regressing
progeny values on parent values:
2 2
( )( ) / ( ) h b x x y y x x = =


with x representing the parent values and y representing the progeny values. The x and y
data are taken from measurements of a quantitative character.

Quantitatively inherited traits differ in heritability. A polygenic trait such as
yield, influenced greatly by the environment, will have low heritability; while monogenic
and oligogenic traits unaffected by the environment will have high heritability. Selection
from a population in the F
2
generation will not be very effective for traits with low
heritabilty; while selection for traits, within the same F
2
population, that have high
heritability estimates can be very effective. Examples of traits with relatively high
heritability are: heading date and kernel size; while traits generally having low
heritability are yield, lodging resistance, and protein content.

Selection procedures that may be used to identify favorable genotypes from
segregating populations following hybridization for wheat will be explained in the
following figures and are: Fig. 8 Pedigree Selection, Fig.9 Bulk-Population Selection,
Fig.10 Single Seed Descent, Fig.11 Modified Pedigree Selection, and CIMMYTs current
preferred selection method called the selected bulk method shown in figure 12. It is
implied that each subsequent generation after the initial hybridization is a product of self
pollination. Bear in mind that the following examples are basic ideas and may vary
according to specific breading goals and resources.




- 31 -
Fig. 8 Pedigree-Selection Method


Modifications of the pedigree-selection method can be made, such as introducing yield
trials as early as the F
3
and F
4
generations. Keep in mind that progeny from a single plant
are 50% more identical than the previous generations progeny. The pedigree method is
labor intensive and requires detailed record keeping. The advantage is that only progeny
with desirable characteristics are carried forward to the next generation. This method
also permits collection of genetic information that can be used as variation among and
within families can be partitioned allowing for a closer examination of narrow sense
heritability and selection efficiency.
Cross cultivar A by cultivar B according to the
breading goals that have been previously identified
PA PB X
F
1
- Grow 50 to 100 plants.

F
2
- Grow 2000 to 3000 plants. Harvest seed from
individually selected plants, keeping seed from
each separate.
F
3
- Grow head rows with seed harvested from
superior plants. Identify superior rows, and
select the best 3 to 5 of the best plants within
rows. Maintain identity of plant and row while
recording phenotypic data. Continue selection
between and within rows through the F
5

generation.
F
4

F
5
- Normally, 25 to 50 families are harvested at the
end of this generation.
F
6
- Grow families of head rows. Rouge off types.
Uniform related families can be harvested
together and the seed bulked. Harvested Seed
lots are designated experimental lines.

F
7
- Experimental lines are grown in preliminary yield
trials in comparison with local checks.
F
8
to F
10
Yield trials of superior experimental lines
are tested at multiple locations with local
checks. Only the lines identified as
having all the traits desired, are
maintained for testing in subsequent
generations.
F
11
- A cultivar found to be superior to local checks is
named, certified, and released for increase and
distribution
- 32 -
Fig. 9 Bulk-Population Selection Method


The bulk-population method of breeding is simple, convenient, requires less labor, and is
less expensive to conduct during the early segregating generations than the pedigree-
selection method. It is necessary to grow a large population of the spaced plants (F
5

above) to have a reasonable chance of finding desirable segregates. During the early
segregating generations there is a good chance that some desirable genotypes will be lost
as they are suppressed by neighboring plants that are taller or have a more spreading
habit.
PA PB X
Cross cultivar A by cultivar B according to the
breading goals that have been previously identified.
F
1
- Grow 50 to 100 plants. Harvest en masse and
bulk seed
F
2
- Grow 2000 to 3000 plants. Harvest en masse
and bulk seed from all plants.
F
3
to F
4
- Grow 100m
2
plots with bulked seed from
previous generation. Plots can be
subjected to biotic and/or abiotic stresses
in order to foster natural selection.
F
5
- Space plant 3000 to 5000 seeds. Select and
harvest 300 to 500 superior plants keeping
seed separate from each plant.
F
6
- Grow head rows of selected plants. Select 30 to
50 head rows that show desirable
characteristics. Bulk seed from selected rows.
F
7
- Grow plots made of selected rows from F
6
in
preliminary yield trials.
F
8
to F
10
Yield trials of superior experimental lines
are tested at multiple locations with local
checks. Only the lines identified as
having all the traits desired, are
maintained for testing in subsequent
generations.
F
11
- A cultivar found to be superior to local checks is
named, certified, and released for increase
and distribution

- 33 -
Fig. 10 Single-Seed-Descent Method

The single-seed-descent method was developed as a way to maintain a maximum of
diversity in the F
2
generation, and reduce the loss of genotypes during the segregating
generations. The single-seed-descent method can also reduce the time necessary to grow
segregating generations. There is also no need for record keeping in the early segregating
generations. As only one seed is harvested from each plant careful development of the F
2

to F
4
generations is not necessary. In a greenhouse; by thickly planting seeds, with low
soil fertility, and using light regimes to force early flowering, two or three generations
can be harvested in a 1-year period. Undesired genotypes, however, will not be
eliminated in the early generations, and as no family structure is maintained, superior
segregate families can not be identified.

PA PB X
Cross cultivar A by cultivar B according to the
breading goals that have been previously identified.

F
1
- Grow 50 to 100 plants. Harvest seed en masse
and bulk.
F
2
- Grow 2000 to 3000 plants. Harvest a single
seed from each plant.
F
3
and F
4
- Grow seed harvested from previous
generation harvest a single seed from
each plant.
F
5
-

Space plants in field with single seeds harvested
from F
4
. Select plants according to breeding
goals and harvest seeds from selected plants.
F
6
- Grow head rows with seed harvested from F
5
.
Harvest rows superior for desired traits.
F
7
- Grow plots made of selected rows from F6 in
preliminary yield trials.

F
8
to F
10
Yield trials of superior experimental lines
are tested at multiple locations with local
checks. Only the lines identified as
having all the traits desired, are
maintained for testing in subsequent
generations.

F
11
- A cultivar found to be superior to local checks is
named, certified, and released for increase and
distribution

- 34 -
Fig. 11 Modified Pedigree-Bulk Selection Method



With the globalization of CIMMYTs Bread Wheat Breeding Program in the 1980s and the
evolution of the concept of 12 MEs, the number of crosses made annually increased dramatically from
2000 in the early 1970s to 10 000 in the 1980s. The total number of segregating populations (F2 to F7)
grew from 20 000 lines to 150 000. Similarly, the number of entries in yield trials increased from 1 000 to
5000 annually. The total area in breeding and testing expanded from 30 ha to 100 ha in the same period.
To accommodate this increase in breeding populations, the methodology of selection was changed from a
pedigree system to a modified pedigree-bulk selection approach. The new method allows one experienced
CIMMYT breeder to evaluate all segregating populations, in a timely fashion, except for the F2.
Simultaneously, total mechanization of planting and harvesting and the computerization of field books have
allowed a limited group of support staff and technicians to carry out all responsibilities. These three major
changes introduced in the CIMMYT operation have increased the ability to introgress variability by
significantly increasing the number of crosses directed for specific MEs, while keeping the selection
program highly efficient and without sacrificing population size per cross.
PA PB X
Cross cultivar A by cultivar B according to the
breading goals that have been previously identified

F
1
- Grow 50 to 100 plants. Back-cross or top cross if
desired.
F
2
- Grow 2000 spaced plants. Individual plants are
selected based on agronomic type, disease
resistance, seed health and grain size.
F
3
to F
5
- Selected F
2
are grown at a commercial
seeding rate in 3 rows of 2m in order to
observe competitive ability within the
line. Selection is based on agronomic
performance and disease resistance in
the plot. 10 to 15 heads are harvested
and bulked for each selected plot and
promoted to the next generation.
F
4

F
5

F
6
- Five to ten heads from selected plots are
threshed individually.
F
7
- Heads selected from F
6
are planted individually in
head-rows. Selected rows are harvested in bulk.
F
8
to F
10
- Selections continue based on preliminary
yield trials and subsequent replicated
yield trials, along with industrial quality
testing.
F
11
- Superior lines are included in one of CIMMYTs
international Nurseries or Yield Trials.
- 35 -

Fig. 12 Selected Bulk Method

The main difference between modified pedigree/bulk and the bulk selection method is
how selected plants from the F2 populations are handled. In the modified pedigree-bulk
selection method plants are kept separate in the F2; using the selected bulk method all
seeds are bulked in the F2. The advantage of bulking the F2 is that it requires less land
and much less bookkeeping. The advantage of keeping the F2 plants separate is that
selection is more efficient at an early stage. However, computer simulations have shown
that the genetic gain is similar for the two methods. A more detailed look at how
CIMMYT conducts their Selected Bulk method will be found in the module: Managing a
Breeding Program.
Selected bulk method
F1

F2

F3

F4

F5

F6

F7

F8
to
F10
Parent A x Parent B
Bulk plot

Space planted

Bulk plot with selection

Bulk plot with selection

Bulk plot with selection

Plant rows

Preliminary yield trials

Replicated yield trials
- 36 -
Genetic advance is the expected gain in the mean of a population for a particular
trait by one generation of selection. Genetic advance can be both predicted and
calculated, and therefore can be used to both determine breeding objectives and track
progress while selecting from populations. For predicting genetic advance the following
formula can be used:
( )
( )( )
2
s P
G i V h = .
Where G
s
is the predicted genetic advance, i is a constant based on selection intensity
(table 1), V
P
is the phenotypic variance, and h
2
is the narrow-sense heritability. In order
to use this formula narrow-sense heritability must be known. For calculating genetic
advance during the selection process the following formula can be used:
( )( )( )
2
p P
R k V h = .
Where R is the response to selection, and k
p
is the standardized selection
differential
p
P
S
k
V
| |
= |
|
\
. The selection differential is denoted by S and is equal to the
difference of the selected individuals mean minus the mean of the
population:
selected population
S = . The wheat breeder will need to improve not only
one trait, but several traits at one time. This can be done by establishing minimum
performance for each trait and excluding the others in the following generations. This is
known as independent culling. Independent culling levels, along with amount of
resources available to the breeder will determine the selection intensity. For example,
60% of the population might meet the height requirements, 30% might meet the disease
resistance requirements, and 40% might meet the quality requirements. In this case,
60% 30% 40% 7.2% = of the current population will be advanced to the following
selection cycle. If the breeder has 10,000 head rows and the resources to manage 1000
preliminary yield trial plots, these independent culling levels are appropriate. A more
efficient form of multiple trait selection is by using a selection index with separate traits
being given different weights depending on their level of importance. The Smith-Hazel
index, or optimum index, is a good example of a selection index; however an
understanding of covariance and matrix algebra is needed and cannot be covered here.

Table1 Commonly Practiced Selection Intensities
and Their Corresponding Constants
Selection Intensity (%) i
1 2.665
5 2.063
10 1.755
20 1.400


VI. Genotype X Environment (GxE) Interactions
As stated previously, phenotypic variance can be attributed to both genetics and
genetic by environment interaction. Genetically identical wheat populations grown in
- 37 -
various environments will perform quite differently. Soil fertility, temperature,
precipitation (amount and timing of), tillage practices, crop-rotation practices, day length,
and disease pressure are all environmental variables among many others that can have an
effect on wheat performance. Environmental variability can be accounted for by both
diverse positions (locations, and replications) and time (years, or seasons).

There are three approaches a plant breeder can take when addressing GxE. First,
the breeder can ignore it. With this approach potential cultivars are tested over a wide
range of environments. The performance of cultivars, resulting from selection, is
superior when averaged over all the target environments; but is probably not the best in
each environment. The second approach is to reduce GxE interaction. This can be done
by partitioning the target environments into smaller more homogeneous subgroups. This
can be done on a grand scale such as CIMMYT has done by defining the 12 Mega-
Environments, or it could be done on a much region. Cultivar recommendations are then
made for each distinct subgroup. Cluster analysis and principal component analysis
are useful when partitioning environments. The third approach is to exploit GxE
interaction (reduce the GxE variance component). This approach is used to identify
cultivars best suited for a specific, niche environment in an effort to maximize
productivity in that one environment. This third approach is a way to exploit positive
GxE interaction and might be at first thought the best approach of all, until the amount of
resources necessary to breed cultivars for every niche environment is considered.
Stability analysis and multiplicative models can be useful tools when the third approach
to addressing GXE is desired.

As stated and implied previously the wheat breeders job is to select genotypes
that will outperform the local check or standard cultivar. Also expressed earlier is that a
pure line populations performance is based not only on its genotype but also how the
genotype interacts with the environment:
P G GE
V V V V

= + + . In order to more accurately

select the superior genotypes the breeder must be able to explain a maximum of V
P
by
V
G
. This can be done by increasing the number of environments that the genotypes are
grown in and the number of replications in each environment. By increasing the number
of environments a line is tested in; the unexplained variance V

, or the error variance, is

reduced and more of the variance can be explained by V
G
and V
GE
. In turn an increase in
variance explained by genotype will increase heritability estimates, leading to an increase
in selection response (remember:
( )( )( )
2
p P
R k V h = ).


- 38 -
Questions to Test Understanding:

1. In what way is plant breeding an art?
2. What are the advantages to the CIMMYT breeders of dividing the world
in Mega Environments (ME)?
3. What is a single backcross breeding strategy?
4. What is the purpose of the single-seed-descent method?
5. What is the definition of Genetic Advance?
6. What are the different approaches that a breeder can take to addressing
GxE?
- 39 -

References:
John Milton Poehlman and David Allen Sleper. Breeding Field Crops (4
th
Ed) 1995
Iowa State University Press, Ames Iowa

E.G. Heyne editor. Wheat and Wheat Improvement. 2
nd
edition. American Society of
Agronomy, Inc. Crop Science Society of America, Inc., Soil Science Society of
America, Inc. Publishers Madison, Wisconsin USA 1987

Rex Bernardo. Breeding For Quantitative Traits In Plants. 2002 Stemma Press