Tải bản đầy đủ - 0trang
V. Improvements in Methods of Breeding Wheat
HALF CENTURY OF WHEAT IMPROVEMENT IN UNITED STATE^
ment of special techniques for determining varietal resistance to weather
hazards and pests of various kinds. Methods for measuring quality have
been greatly improved and simplified. More complete knowledge of the
genetic constitution of the wheat plant has greatly facilitated the choice
of parental varieties, the selection of progenics, and breeding procedures.
1. Earl9 Methods of Breeding
Continuous mass selection of the largest and best-appearing heads of
the heaviest and plumpest grain separated by wind or sieves was generally recommended and used at the turn of the century, as it had been
since the time of the Romans and even before. Publication of The
Origirz of Species by Darwin in 1859 provided a theory that greatly stimulated the belief that important improvements could be achieved by continuous selection.
lie Couteur and Shirreff initiated the practice of selecting single
heads and plants early in the nineteenth century. Shirreff emphasized
the importance of the initial selection and believed that little or nothing
was gained by later selection. Hallett (1861) proposed a “pedigree
method” which consisted essentially of selecting the largest grain from
the longest and largest head from the best plant each year and continuing this year after year. Four years of selection doubled the length of
the head, trebled the number of grains per head, and increased the tillering fivefold. He believed that yields per acre were also greatly increased.
Some breeders followed the method developed by Shirreff, but Hallett’s
philosophy predominated until the close of the century and did not disappear until the end of the first decade o r later of the present century.
Willet M. Hays of the Minnesota Station in about 1895 applied to
wheat breeding the concept of the progeny test to determine the relative
superiority of selected strains. He developed the centgener method of
planting for spacing the plants and thereby facilitated the selection of
the best individual plants each year. The method was widely adopted
throughout the United States.
This, in brief, was the prevailing theory and practice until the publication of Johannsen’s pure-line theory in 1901. This publication plus
accumulated general experience, including carefully controlled experimental trials at a number of experiment stations, turned the attention
of wheat breeders to the selection of pure lines and the determination
of their relative value. Most of the improved varieties of the first three
decades of the twentieth century were the product of this method. They
include such varieties as KANRED, NEBRASKA NO. 60, CHEYENNE, IOBRED,
S. C. SALMON, 0. R. MATHEWS, AND R. W. LEUKEL
36, FULHIO, TRUMBULL, NITTANY, GASTA,
Hybridization, it is true, was practiced by a number of breeders before 1900, notably Farrer in Australia; Saunders in Canada; and Blount,
Pringle, Jones, and Spillman in the United States, but it received relatively little attention. The principal purpose of hybridization was to
induce variation o r to “break the type” and thereby afford greater opportunity for selection. Consequently, parents were chosen more or less
at random. A few breeders had clearly in mind the possibility of combining in a single variety the desirable characteristics of two or more.
Saunders recognized the need f o r early maturity when he crossed Hard
Red CALCUTTA with RED FIFE to produce MARQUIS, and Spillman set out
to combine resistance to shattering and winter hardiness in the Washington hybrids. Not until several years after the rediscovery of Mendel’s
laws, however, was there general recognition of the value of hybridization and the need for a careful choice of parents based on their known
A development traceable directly to a, better understanding of genetics is the backcrossing technique, first proposed by Harlan and Pope
(1922) and used especially by Briggs (1938) and associates in California. It has also been used extensively in the breeding of rust-resistant
durums but sparingly elsewhere. Percival (1921) and Clark (1936) reviewed the methods generally used during the last century and in the
early part of the 20th century.
IOWIN, KARMONT, MONTANA NO.
MINDUM, NODAK, KOTO, PROGRESS,
2. Objectives in Breeding
Another development of first importance is the clearly defined objectives of most modern wheat-breeding programs. Early efforts were directed mostly to increasing yields but without any clear concept of what
determined yield. It was assumed that varieties differed in “yielding
capacity,” as, without doubt, they do and also that “yielding capacity”
is synonomous with actual yield, which usually is not the case. Plant
breeding often was regarded as an ast; this implied that breeders acquired a special skill that enabled them to choose heads or plants that
would produce superior yields. Undoubtedly, there is some truth in
most of these assumptions, but there is also a great deal of error. This
concept again was not accepted universally. Carleton (1900, p. 54), for
example, stated that “yield after all is not a distinct quality in itself but
is the combined result of a number of qualities acting independently and
not thought of a t all.”
Much information as to yield factors is still lacking, but few probably
will doubt that wheat breeders are on solid ground in breeding for re-
HALF CENTURY OF WHEAT IMPROVEMENT IN UNITED STATES
sistance to stem rust in spring wheat for the northern Great Plains, in
breeding for appropriate degrees of winter hardiness in winter wheats,
and in breeding for resistance to all diseases and insect pests and weather
hazards so f a r as is feasible and for areas where they are known to be
important. The important difference is not that the modern wheat
breeder ignores yield, but that he uses what information he has more
effectively in producing higher-yielding varieties. His efforts are more
effective because it is much easier, less expensive, and less time-consuming to test selections for reaction to individual factors that affect yield
than i t is to determine relative yields. If a modern wheat breeder uses
the term “yielding capacity’’ at all, he means yielding capacity under
a specified set of environmental conditions.
A closely related achievement also of importance is the relatively
more precise knowledge of the kind of wheat needed for each wheatgrowing area. The wheat breeder has learned, mostly the hard way,
that nowhere in the United States are varieties wanted that mature as
late as those of northern Europe or as early as those grown in India:
that varieties even earlier than TURKEY are needed for the southern
Great Plains; that in much of the northern Great Plains varieties of
spring wheat susceptible to rust have little chance of successful competition with similar varieties that are resistant to rust; that short stiffstrawed varieties are needed for the Pacific Northwest. There is now
a clearer concept of the degree of winter hardiness needed for each principal wheat-growing area and also general recognition of the fact that
winter hardiness in one area does not necessarily mean winter hardiness
elsewhere. If it is necessary, as seems probable, to accept a compromise
between extremes of winter hardiness and of early maturity in hard
winter wheat, current knowledge as to the needs with respect to each
should make i t easier to attain a suitable compromise.
3. Resistance to Disease, Insect, and Weather Haaards
Certainly among the most important advances in techniques are those
for testing the resistance o r tolerance of varieties and selections to specific disease, insect, and weather hazards. Wheat breeders no longer
wait for natural epidemics but produce what is needed where and when
it is wanted. This method, first proposed by Bolley (1905) soon after
1900 for breeding varieties of flax resistant to wilt, has been all but universally adopted for other crops and diseases wherever feasible. Seed
is inoculated with specific races of the bunt organism and seeded late to
insure suitable temperatures for infection or, in the case of dwarf bunt,
the seed is planted in infected soil. Plantings for leaf and stem rust
resistance are provided with border o r so-called spreader rows of a sus-
6. C. SALMON, 0. R. MATHEWS, AND R. W. LEUKEL
ceptible variety which, in turn, are artificially inoculated and watered
with overhead sprinklers to provide suitable atmospheric humidity for
the germination of the rust spores. Cherewick (1946) has recently described the methods used i n Canada for establishing rust epidemics in
experimental plots. Cartwright and LaHue (1944) have developed a
technique for testing varieties and selections for resistance to Hessian
fly by means of which 20,000 or more may be tested in a single season.
Platt and Farstad (1946) have described a method for insuring epidemics of sawfly that has proved most useful in breeding for resistance
to this insect in Canada and the United States. It consists essentially
of seeding short rows of the material to be tested on summer fallow and
adjacent to infected stubble of the preceding crop. The rows are seeded
at right angles to the stubble in order to insure like infestation of all
rows, since infestation depends materially on the distance the flies must
migrate. An important feature, emphasized by Platt and Farstad in
the interests of economy of labor, is to estimate infestation rather than
make actual counts. The latter is very laborious and time-consuming,
and estimates were found by them to be sufficiently accurate for most
purposes. Similar methods incidentally are widely used and with satisfactory results i n determining relative resistance to rust, bunt, Hessian
fly, winterkilling, lodging, shattering, and other diseases and hazards.
The development of greenhouse techniques not only for determining
resistance to diseases and insects but also for growing two or more generations per yeas has been most important. Resistance of selected lines
can be determined before the time for seeding in the field the following
spring. The F 2 or segregating generation of a cross is reached a year
earlier by growing the F1 generation in the greenhouse. Closely related
is the practice of seeding a crop in the field in Arizopa, southern California, or Mexico, and shipping the product back to the northern United
States for planting the following spring. Relatively rapid increases in
the quantity of seed of promising varieties have been made in this way.
The relative winter hardiness of new varieties or selections for Central and Southern States is frequently determined by seeding them in
Northern States where partial but not complete killing may be expected.
Artificial freezing tests have been used although not extensively, largely
because equipment is expensive and not readily available. The use of
winter-hardiness nurseries and others for similar purposes has been
greatly facilitated by the co-ordinated co-operative programs previously
described. Relative shattering and lodging are now easily, simply, and
inexpensively determined by permitting border rows to stand for several
weeks after the normal harvesting date and estimating the loss due to
either or both.
HALF CENTURY OF WHEAT IMPROVEMENT IN UNITED STATES
Unfortunately, no simple or even dependable methods for testing
relative drought resistance have been developed other than relative yields
in a dry area for a long period of years.
4. Testilzg for Comparative Yields
Field plots of various sizes seeded with an ordinary grain drill were
used for determining relative yields of varieties long before the end of
the nineteenth century. Usually there were single plots only of each
variety, although occasionally duplicate, triplicate, or even quintuplicate seedings were made. One-tenth acre was a common size plot. It
was recognized early that soil variation of ten seriously vitiated experimental results ; consequently, much attention was devoted in the early
years of the century to the selection of uniform land for experimental
fields, uniform preparation of the land, improvement in uniformity by
drainage, etc., and in extreme cases to the relocation of experimental
fields in order to have reasonably uniform land. The use of replicated
plots in the United States has been almost universal since about 1915,
largely as a result of the application of statistical methods to field experiments, I n recent years the combine has been used to harvest experimental field plots with a material saving in labor.
Probably the most important single advance in methods of comparing
yields was effected by the substitution of rod-row plots for centgeners.
Rod-row plots were first extensively used by Norton (1907). The universal adoption of this and similar methods has resulted in a saving of
labor, and hence in an opportunity to compare more selections, which it
is difficult for those who have not used a centgener machine to appreciate. There can be little doubt also that modern rod-row tests are more
accurate than those made by the centgener method, although there appear to be no data to prove it. An important contribution to the accuracy of rod-row tests was the demonstration by Kiesselbach (1918) a t
the Nebraska Station of border effect, and as a consequence the general
adoption of multiple-row plots.
Kezer (1906) correctly believed that he had made a n important contribution when by substituting a movable for a fixed frame, he and five
men were able to plant the equivalent of 275 centgeners per day as compared with 80 with the machine previously used. But contrast this with
100 rod-rows or 150 head rows per hour per man by modern plant-breeding crews. Many yield nurseries are now seeded with drills powered by
small tractors, harvested and bound into bundles mechanically, and
threshed with improved threshing machines having probably twice the
capacity of those available in 1900. Magee (1951), for example, has
S. C. SALMON, 0. R . MATHEWS, AND R. W. LEUKEL
estimated that three men with a tractor-mounted drill used by him “can
seed more than 5 men using hand seeders.”
One thousand centgeners was a big nursery around 1900. It is not
unusual for a modern wheat-breeding and yield nursery to comprise
10,000 or even 20,000 rows. The improvement, it should be noted, is
due not only to better techniques of planting, harvesting, threshing, etc.,
but also to changes in specific objectives, as noted above, which greatly
reduce the time devoted to each item. For example, many selections are
now discarded without harvesting with perhaps no more than a record
of their defects. With the centgener method and the philosophy accompanying its use, detailed notes were recorded often for each individual
plant in each centgener.
5. Techniques for Measuring Quality
Striking improvements in methods of measuring or comparing quality have been made since Saunders used the chewing test to determine
the quality of MARQUIS. At least four distinct categories must be recognized: milling quality; quality for bread; quality for cake, cookies, crackers, and similar products collectively called pastries ; and in the case of
durum wheat, quality for macaroni.
a. Milling Quality. A good milling variety in the eyes of the miller
is one that mills easily and produces a high yield of flour, especially of
patent flour. Improvements in techniques for determining or measuring
milling quality have been few. They have consisted mostly of standardization and refinements in the operation of experimental mills, and there
have been some improvements in the mills themselves, especially adjustments to enable satisfactory tests to be made on small samples. Recent
studies indicate that it may be possible to relate poor milling properties
to definite physical or chemical characteristics of the grain and thus
facilitate the identification of poor milling varieties.
b. Quality for Bread. Perhaps the most important advances i n techniques relating to measuring quality are those f o r quality of bread. The
history of the subject is surprisingly confusing in view of the rather
clear picture that has emerged in recent years. The literature is voluminous, and only the barest outline can be presented here. Larmour (1940)
has given an informative account of much of the research in the field u p
to about 1940 as it relates to a comparison of hard red spring and winter
Quality of undamaged or normal wheat and flour for bread has long
been known to depend on quantity and quality of protein. Protein content is easily determined, but until recent years there was no adequate
test for protein quality. The latter was at best inaccurately estimated
HALF CENTURY O F WHEAT IMPROVEMENT. I N UNITED STATES
from the baked loaf and then only when interpreted in terms of the
quantity of protein in the flour; that is, if differences in the quality of
bread could not be explained by differences in protein content, they were
automatically attributed to protein quality. Interpretation was usually
difficult and inaccurate, and especially so if the protein contents of the
varieties being compared were different.
This difficulty stemmed principally from the general experience, as
shown by Thomas (1917), Coleman et al. (1927), and Shollenberger
[cited by Larmour, (1940) 1, that the relation between protein content
and loaf volume by the methods then used is nonlinear. Flours with a
medium protein content would generally produce better bread than
those with a low protein content but would also usually produce as good
bread as flours with very high protein content. Yet it was known that
high protein flours used in blends with low protein flours would improve
the latter in proportion to protein content.
This paradox was cleared up by a long series of investigations, including those by Larmour and MacLeod (1929), Geddes and Larmour
(1933), Larmour and Brockington (1933) , Ofelt and Larmour (1940),
Larmour (1940), and especially Finney and Barmore (1944,1945,1948).
Briefly, it was found that the baking test formulas and procedures then
used were such that the potential value of high protein flour was not
expressed, As a corollary, it was found that when proper formulas and
techniques are used, the relation between protein content and loaf volume
is perfectly or almost perfectly linear, at least within the range ordinarily found.
Another important discovery was that the slope of the regression line
is a varietal characteristic and such that for a large number of varieties
differing in protein quality, they fall into a €an-shaped pattern with
rela.tively little difference between them at low protein levels and wide
differences a t high protein levels. This means that for the first time
cereal chemists have an adequate and reasonably accurate measure of
Very recently it has been found that exposure of the growing wheat
to excessively high temperatures and relatively low humidities during the
latter part of the fruiting period may deleteriously affect protein quality
and distort the usual linear relation between protein content and loaf
volume. Other environmental factors may require consideration but
even so the bread-baking test is far superior to that of fifty years ago.
G. Pastry Quality. The fact, long suspected but not proved until recently that pastry quality depends more on the physical properties of
the flour and on the characteristics of the protein than on the protein
content, has been of great assistance in evaluating varieties for pastry
S. 0. SALMON, 0. R. MATHEWG, AND R. W. LEUKEL
purposes. The most generally acceptable evaluation is by means of the
cooky-baking test. Cake-baking tests are generally less useful because
small differences are obscured, probably because flour constitutes less
than half of the ingredients in a cake formula. Cooky-baking tests apparently were first proposed by Alexander (1933), first used for evaluating varieties by Fifield et al. (1936-1950), and since greatly improved
by Finney and Yamazaki (1946) and Finney et al. (1950).
d. Bread-Baking Tests f o r Soft Wheat. A t the beginning of the century and for many years thereafter, soft wheats for the Eastern States
were evaluated by bread-baking tests. One reason was that soft wheats
were then extensively used for bread and family flours. Another reason,
probably not widely accepted, was the assumption that varieties of poor
quality for bread were suitable for pastries and vice versa. This, as is
now well known, is only partly true. Important here, especially in relation to future breeding, is the discovery that the suitability of certain
varieties of soft wheat for bread or for pastries depends on their protein
content, which is determined mostly by environment. If high in protein
content, they may be used for bread, and if low in protein content, they
may be used for cakes, cookies, crackers, and other pastry products. This
is true, however, only for those varieties that yield flours having the
necessary desirable physical properties.
e. Macaroni Quality. Macaroni tests for quality of durum varieties,
the only tests available during the early years of the century, are especially time-consuming and like the bread-baking tests require considerable grain. Relative quality of durum varieties is now regularly
determined by the disk test first proposed by Fifield e t al. (1937). I n
this method, the wheat is milled to produce semolina in the usual way.
Semolina dough is then pressed into disks one-fourth of an inch thick and
with about the diameter of a silver dollar. Translucency and color, on
which quality for macaroni depends, can be determined from them as
easily and as accurately as from macaroni. In terms of labor and time,
the disk test is seven or eight times as efficient as the macaroni test.
Macaroni is still sometimes made for final evaluations of promising new
varieties but otherwise is seldom needed.
f. Ancillary Quality Tests. Baking tests whether of bread or cookies
are time-consuming and relatively expensive. They also require more
flour than is usually available in early-generation selections from a cross.
Consequently, there has been a n urgent need for simple chemical or
physical tests that require less time and less flour. This need has in part
been supplied by various tests including the (1) viscosity test, (2) doughball, Pelshenke, or fermentation time test, (3) pearling test, (4) mixogram curves, (5) water-absorption test, (6) the sedimentation test, and
HALF CENTURY OF WHEAT IMPROVEMENT IN UNITED STATES
others. None of these can be completely substituted for baking tests,
but each provides a certain kind of information that is most useful in
evaluating quality and especially so in early generations, since each test
requires only a small quantity of grain. Finney and Yamazaki (1953)
have recently developed an alkaline viscosity test, and Pamazaki (1953)
has developed a n alkaline water-retention capacity test for soft wheat
flours that correlates very highly with the cooky test. Micro-milling and
micro-baking test procedures, developed by Finney and Pamazaki
(1946) and Finney et al. (1950), are now available that require only a
small quantity of grain and hence are useful for testing quality in early
Little was known about the nature of wheat diseases or how to control them before the beginning of the twentieth century. Loose smut
and stinking smut or bunt were widespread, and the latter especially
threatened to become a limiting factor in wheat production in many
areas. Control methods known a t that time were inconvenient or ineffective or both and were not generally used until severe losses had occurred. Both stem and leaf rust were recognized as important diseases
in many sections. Other diseases were present, but most of them had
not been identified and fully described as to symptoms, causal organisms,
and manner of spread.
1. #tern Rust
a. History and Distribution. Some important facts regarding stem
rust had been established before the twentieth century, such as the general characteristics of the causal organism, the role of the barberry as
an alternate host, and the fact that the variety of stem rust that attacks
oats, for example, does not attack wheat.
Laws to enforce the eradication of the barberry were passed in Connecticut, Rhode Island, and Massachusetts between 1726 and 1766, long
before the relationship between barberry bushes and stem rust infection
was clearly established. The existence of physiologic races that attack
some varieties of wheat but not others was not known, nor had the role
of wind-blown spores in producing stem rust epidemics been discovered.
Carleton (1900) and others had recognized differences in susceptibility
of varieties to stem rust, but resistant varieties, as we know them today,
were not available.
b. Development of Resistant Varieties. The discovery of the outstanding resistance to stem rust in IUMILLO durum wheat and in YAROSLAV
emmer marked the beginning of the first successful attempts to breed
S. C. SALMON, 0. R. MATHEWS, AND R. W. LEU-
resistant varieties. So far as the writers are aware, the details of this
discovery have never been published. The first significant observations
regarding the rust resistance of these varieties were made a t the South
Dakota Agricultural Experiment Station a t Brooking8 in 1902, when
Mr. John 8. Cole, then in charge of the cereal-breeding plots, reported
in part as follows : *
“Rust damage to common spring wheat was so great that no comparisons could be made that were of any value. . . Durum wheats from
Spain, Italy, and Bulgaria . . show but little promise. One, however, NO.
1736, a wheat of peculiar type from Italy, appears to be almost perfectly resistant to black rust and promises to yield well.’’ NO. 1736 is
IUMXLLO, which later provided most of the genes for the resistance of
THATCHER to stem rust. Cole also stated: “Results with emmers were
very striking, especially in the matter of rust resistance. Four varieties
were found to be highly resistant to rust and yielded about 40 bushels
per acre, whereas the yield of other varieties grown under the same conditions, but which rusted severely, went as low as 4 bushels.” Among
these four resistant varieties was YAROSLAV emmer, which McFadden
later crossed with MARQUIS in producing HOPE and H-44 and from which
most of the genes for resistance in common spring wheats other than
THATCHER have been derived.
Cole’s observations were verified during the famous stem rust epidemic of 1904 and again in 1905. By 1904, the varieties mentioned by
Cole were being grown at other stations where observations regarding
their rust resistance served to call attention to the possibilities of using
them i n breeding programs. Fo r further details, the reader is referred
to Ausemus’ (1943) excellent review of breeding for resistance to stem
rust and other diseases in wheat and other small grains.
c. Discovery of Physiologic Races. Another contribution of basic
importance regarding the rust fungus was the discovery in 1917 of the
existence of physiologic races that attack different varieties of wheat.
This outstanding contribution was made by Stakman and his associates
at the Minnesota Experiment Station (Stakman and Piemeisel, 1917 ;
Levine and Stakman, 1918).
d. Barberry Eradication. The disastrous stem rust epidemics in 1904
and 1916 were chiefly responsible for the region-wide barberry eradication laws passed by North Dakota in 1917 ; by South Dakota, Minnesota,
Iowa, Nebraska, Colorado and Michigan in 1918 ; and later by Montana,
* Report of the cooperator’s work on cereals at the South Dakota Agricultural
Experiment Station during the season of 1902. Nov. 7, 1902. (Unpublished.) Filed
with the Division of Cereal Crops and Diseases, Bureau of Plant Industry, Soils, and
E W ~ FCENTURY OF WHEAT IMPROVEMENT
IN UNITED STATES
Wyoming, Wisconsin, Illinois, Indiana, Ohio, Pennsylvania, Virginia,
West Virginia, Missouri, and Washington. By 1950 about 340,000,000
bushes had been destroyed and about four-fifths of the area in eighteen
states had been cleared of barberry. About 200,000 square miles are
still partially infested, mostly in areas that are not easily accessible.
Barberry eradication has been accompanied by a marked reduction
in losses due to stem rust. This period has coincided with the increasing
and widespread use of resistant varieties in the Great Plains. It is this
area i n which the most severe losses occur. Damage to susceptible common wheat varieties in experimental plots in the spring wheat region as
late as 1945 and to d u r n wheat in 1951 and 1952 showed very clearly
that barberry eradication alone is not sufficient to control stem rust in
this area. The reason is now known to be the wind-blown spores coming
from the overwintering areas in south Texas and northern Mexico. Infection from barberry bushes usually occurs earlier in the spring than
that from wind-blown spores, and in some areas these bushes are the
principal source of infection. These facts, plus the production of new
races of stem rust by hybridization on barberry bushes as discovered by
Craigie (1927, 1928), provide sound reasons for continuing the eradication campaign.
2. Leaf Rust
Leaf rust of wheat, although more generally and uniformly distributed throughout the humid wheat-growing areas of the world, causes less
damage than does stem rust. Formerly, it was most severe in the winter
wheat areas of the United States, especially in the East and Southeast.
I n more recent years, damage has been considerable in the central United
States and in Canada on hard red spring wheats. In 1938 it occurred
in epidemic force in this area and caused losses u p to 30 per cent in several states from Texas to Canada.
Leaf rust does not require infection of a n alternate host in its life
history, and unlike stem rust it survives the winter on wheat plants, as
far north as Maryland. Although the rust can resist cold, it cannot endure high summer temperatures, and, consequently, it often dies during
the hot summers of the southern wheat areas. Chester (1939) showed
by me:i.ns of airplane spore traps that viable spores are blown back from
the North by the northerly fall winds to infect the winter wheat in the
South. A t times this infection is so severe that it injures the wheat for
winter pasture and causes a poor survival of the crop. Leaf rust thrives
during periods of damp weather with temperatures ranging from 50" F.
to 60" F. Little or no infection occurs above 80" F.
The most practical method of combating leaf rust is by breeding re-