GENETIC ROUTES TO MODULATE RATE OF DRY-MATTER
DISAPPEARANCE OF BARLEY GRAIN IN
THE RUMEN OF CATTLE
by
Jeremy Burke Jewell
A thesis submitted in partial fulfillment
of the requirements for the degree
of
Master of Science
in
Plant Sciences
MONTANA STATE UNIVERSITY
Bozeman, Montana
April 2007
© COPYRIGHT
by
Jeremy Burke Jewell
2007
All Rights Reserved
ii
APPROVAL
of a thesis submitted by
Jeremy Burke Jewell
This thesis has been read by each member of the thesis
committee and has been found to be satisfactory regarding content,
English usage, format, citations, bibliographic style, and consistency,
and is ready for submission to to the Division of Graduate Education.
Dr. Tom Blake
Approved for the Department of Plant Sciences & Plant Pathology
Dr. John Sherwood
Approved for the Division of Graduate Education
Dr. Carl A. Fox
iii
STATEMENT OF PERMISSION TO USE
In presenting this thesis in partial fulfillment of the requirements
for a master’s degree at Montana State University, I agree that the
Library shall make it available to borrowers under rules of the Library.
Copying is allowable only for scholarly purposes, consistent with
“fair use” as prescribed in the U.S. Copyright Law. Requests for
permission for extended quotation from or reproduction of this thesis
in whole or in parts may be granted only by the copyright holder.
Jeremy Burke Jewell
April 2006
iv
ACKNOWLEDGEMENTS
Thanks and respect go to the members of my graduate
committee: Dr. Tom Blake, Dr. Jack Martin, Dr. Jan Bowman, and Dr.
Mike Giroux.
Gratitude to the members of the Oscar Thomas Nutrition Center:
Liz, Brenda, Nicole, and Allison. Gratitude to Pat Hensleigh, Stan
Bates, and MacKenzie Ellison.
Much respect for Dr. Tom Blake and Hope Talbert. They have
been the two greatest influences on my intellectual development
during this time.
Much love and respect for my wife Angie Matule Jewell. She has
been the greatest influence on my emotional development during this
time.
Much love and respect for my parents Wallace Jewell and Pearl
Jewell and my brother Greg Jewell for their continued love and
support.
Love and respect to Dr. Vladimir Kanazin.
v
TABLE OF CONTENTS
1. LITERATURE REVIEW .............................................................. 1
Introduction ........................................................................... 1
Cattle and Feed Barley for Montana Agriculture ........................... 2
Starch Digestion by Ruminants ................................................. 3
Methods to Modulate Rates of Digestion ..................................... 6
Manipulation of Grain for Improved Animal Performance............. 10
References........................................................................... 13
2. EVALUATION OF THE ANT18 MUTATION AS A POTENTIAL TOOL TO
MODULATE RATE OF BARLEY RUMINAL DRY-MATTER
DISAPPEARANCE .................................................................. 18
Abstract .............................................................................. 18
Introduction ......................................................................... 19
Materials and Methods........................................................... 22
Plant Materials ................................................................. 22
Particle size, starch content, and DMD Analysis ..................... 23
Statistical Analysis ............................................................ 25
Results and Discussion .......................................................... 26
Time course experiment .................................................... 26
Similar particle size experiment .......................................... 28
Conclusions.......................................................................... 31
Acknowledgements ............................................................... 32
References........................................................................... 33
3. MAPPING OF QUANTITATIVE TRAIT LOCI FOR FEED-QUALITY
RELATED TRAITS IN A TWO-ROWED INBRED BARLEY
POPULATION ........................................................................ 35
Abstract .............................................................................. 35
Introduction ......................................................................... 36
Materials and Methods........................................................... 38
Plant Material ................................................................... 38
Phenotypic Data ............................................................... 39
Anchor Markers ................................................................ 41
Amplified Fragment Length Polymorphism Markers ................ 43
Illumina GoldenGate Assay ................................................ 46
vi
TABLE OF CONTENTS CONTINUED
Map Construction and QTL Analysis – F5 .............................. 47
Map Construction and QTL Analysis – F7 .............................. 48
Results and Discussion .......................................................... 50
Phenotypic Data ............................................................... 50
Map Construction.............................................................. 52
QTL Analysis .................................................................... 55
Further Analysis of the Haxby/Baku Population ..................... 59
Map Construction: SNPs .................................................... 60
Comparison of the F5 and F7 maps ..................................... 72
QTL Analysis: SNPs ........................................................... 73
QTL Validation.................................................................. 82
References........................................................................... 86
vii
LIST OF TABLES
Table
Page
2-1
Estimated differences and standard error of difference
in particle size of wildtype and ant18 lines*
after cracking with disc mill spacing of 1.25 mm ................... 28
2-2
Sources of variation in DMD between Klages and ant18 lines
after cracking with disc mill spacing of 1.25 mm ................... 29
3-1
Selective primers used for AFLP® analysis ............................ 45
3-2
Average values of dry-matter digestibility (DMD), particle
size (PS), starch content (SC), 500-kernel weight (KW),
height at maturity (Ht), heading date (HD) and standard
deviations for ‘Haxby’ and ‘Baku’ and the RILs ...................... 51
3-3
Pearson correlation coefficients among field measurements
and feed-quality characteristics of the Haxby/Baku F5
population ....................................................................... 52
3-4
Locations, LOD scores, and effects of QTL detected by
simple Composite Interval Mapping in the
Haxby/Baku F5 population ................................................. 56
3-5
Comparison of HarvEST consensus and Haxby/Baku
map lengths..................................................................... 66
3-6
Comparison of F5 and F7 Haxby/Baku map lengths. ...................... 73
3-7
Locations, LOD scores, and effects of QTL detected by
simple Composite Interval Mapping in the Haxby/Baku
F7 population ................................................................... 75
3-8
Type III tests of qDMD-6H and qDMD-7H effects on starch
content ........................................................................... 82
3-9
Type III tests of Bmag0009 effects on DMD and mean particle
size in an F5 validation population ....................................... 83
viii
LIST OF TABLES CONTINUED
Table
Page
3-10 Type III tests of head type effects on DMD and mean
particle size (PS) in an F5 validation population ..................... 84
3-11 Type III tests of Bmag0009 within head type........................ 85
ix
LIST OF FIGURES
Figure
Page
2-1
In situ dry-matter disappearance of Klages and
ant18.623-(Klages) at 0.5, 1.0, 1.5, 2.0, 3.0,
6.0, 9.0, and 12.0 hours. ................................................... 27
2-2
DMD of wildtype lines and their ant18 mutants. .................... 30
2-3
Starch content of wildtype lines and their ant18 mutants. ...... 30
3-1
Linkage map of Haxby/Baku RIL population.......................... 54
3-2
Scans of barley chromosomes 6H and 7H for trait DMD.. ........ 57
3-3
Comparison of Haxby/Baku RIL population linkage map and
HarvEST:Barley consensus linkage map. .............................. 62
3-4
Allele frequency in the Haxby/Baku mapping population......... 68
3-5
Comparisons of QTL for mean particle size (PS) and DMD
detected in the 123-member F5 population and in the 86member F7 population. ..................................................... 76
3-6
7H LOD scores for kernel weight and for DMD....................... 79
3-7
Comparison of the barley 7H QTL region and the
orthologous rice chromosome 6 .......................................... 80
x
ABSTRACT
Recent research has identified important characteristics of barley
grain as feed for cattle. Of these, low ruminal dry-matter digestibility
(DMD) is of particular importance as it is highly correlated with animal
performance and with animal health. This research attempts to
identify genetic loci that contribute to the ruminal DMD of barley grain.
The utility of the barley ant18 mutation for decreasing ruminal
DMD was investigated. The DMD of several barley cultivars and their
ant18 mutations was investigated in a randomized complete block
design in two environments. Genotype by environment interaction
was present: in the greenhouse the DMD of ant18 mutants was less
than that of the wildtype, and in a dryland field the reverse was true.
Because of this interaction, ant18 is not likely to be a reliable method
of modulating DMD.
With the aim of identifying markers for marker-assisted selection
(MAS), a 123-member inbred population was developed from a cross
of Haxby and PI 28624. PI 28624 is a low DMD accession from the
USDA barley collection. The grain of this population was evaluated for
DMD at the F6 generation and F5 DNA genotyped using SSR and AFLP
markers, allowing genetic map construction and quantitative trait locus
(QTL) analysis. Two QTL were detected on chromosome 6H and 7H
explaining 19 and 17% of phenotypic variation, respectively. Due to
the low estimated genome coverage of this map (50 to 65%), 86 F7
lines were genotyped using the GoldenGate SNP genotyping
technology. Use of this technology allowed accurate assessment of
genome coverage, which is quite complete with the exception of the
extreme short arms of 2H, 5H, and 6H. 3H and 6H, though apparently
quite complete, are of much shorter centiMorgan length than the
consensus maps of these chromosomes. Possible causes of this
phenomenon are discussed. In addition to the previously detected
QTL, two new QTL for DMD were detected in this F7 sub-population, on
chromosomes 1H and 7H. These QTL may be useful for MAS if they
can be validated in other populations. This population will be useful
for other genetic studies in barley.
1
CHAPTER 1
LITERATURE REVIEW
Introduction
The modern feedlot is a value-added system for converting
inexpensive commodities (corn, barley, etc.) into a more expensive
commodity, beef. The digestive system of the cow (Bos taurus) is a
system capable of fermenting cellulose due to its complex organization
and microbial ecology, and has evolved for this purpose. In a feedlot
environment, the microbial fauna of this system are presented with a
novel substrate, starch. This perturbation can lead to digestive upset
and reduced animal performance. Rapidly digestible starch sources
such as wheat and barley have been particularly implicated as
causative factors in these disorders. This review will detail these
disorders and common methods of prevention. Further, it will
highlight aspects in which barley may conceivably be improved to
reduce these effects and to improve animal performance.
2
Cattle and Feed Barley for Montana Agriculture
Agriculture continues to be the dominant industry in Montana,
accounting for an average of 34% of the economic production by
Montana’s five major industries from 2002-2005 (USDA National
Agricultural Statistics Service, 2007). Agricultural output in 2005 was
3.1 billion dollars (ibid.). Cattle operations and barley production are
important components of Montana’s agricultural sector, accounting for
50.3% and 4.9% of agricultural receipts from 2001-2005(ibid.). From
2003 to 2005, 47% of harvested barley was sold as malting barley
(ibid.); the remainder was sold as feed barley or used on the farm
where it was produced. Of beef cattle produced in Montana from
2001-2005, only 0.8% were slaughtered commercially in-state (this
figure does not include beef cattle slaughtered non-commercially on
farms) (ibid.). The remaining cattle were “backgrounded,” i.e. weaned
and shipped out of state to be fattened and slaughtered elsewhere.
This beef production system where calves are weaned and sent
directly to a feedlot is termed an intensive system, and is favored by
many large producers due to its greater feed efficiency and
corresponding economic advantage (Lewis et al., 1990a, b). The
export of these cattle represents a sizable loss of revenue in the state
3
of Montana due to the high cost of transport, and mortality and
morbidity of transported cattle (Dr. Tom Blake, pers. comm.).
Starch Digestion by Ruminants
In cattle, the rumen is the primary site of starch digestion; 50 to
90 percent of fed dietary starch is digested ruminally, depending on
grain source and grain processing (Huntington, 1997). For barley, this
value is approximately 80 % based on a review of 5 publications
(Huntington, 1997). Starch digested ruminally is fermented by
microbes to volatile fatty acids (VFAs), methane, and carbon dioxide
(Russell and Rychlik, 2001). The VFAs thus produced are absorbed
ruminally and used as an energy source for the animal (Van Soest,
1994).
Starch not digested in the rumen may be digested postruminally in the small intestine or the large intestine, or escape
digestion by elimination. The fraction of fed starch digested postruminally is between 5 and 20%, depending on source and processing
of grain; for barley this value is approximately 13% (Huntington,
1997). The extent of digestion of starch digested post-ruminally
varies between 46 and 93%, again depending on source and
4
processing of grain; for barley this value is approximately 75% (ibid.).
Starch digestion in the small intestine is accomplished by endogenous
α-amylase, and the glucose produced by this digestion is absorbed into
the epithelium of the small intestine primarily by Na+ co-transport
(Wright, 1993). Ruminal fermentative starch digestion is only 70 to 75
percent as energetically efficient as enzymatic starch digestion (Owens
et al, 1986; Harmon and McLeod, 2001). Energy lost by fermentative
digestion is due to methane and heat production (Black, 1971).
Despite the increased energetic efficiency of small intestine enzymatic
digestion, there may be limits to the amount of starch that can be
digested in this manner (Kreikemeier K.K. et al, 1991; Owens et al,
1986).
Fermentative starch digestion in the large intestine is
accompanied by the same energetic losses as ruminal digestion. Also,
though VFAs may be absorbed in the large intestine, excessive
passage and large-intestinal fermentation of starch produces high
levels of organic acids (VFAs) that the large intestine is not able to
completely absorb. This hindgut acidosis has been shown to be
positively correlated with residual feed intake. Channon et al (2004)
reported that low fecal pH was associated with lower animal efficiency.
5
Diets high in irregularly fed rapidly fermentable carbohydrates
i.e. starch or sugars may result in a situation where the rate of
accumulation of VFAs is greater than the rate of absorption of VFAs.
As a result, the pH of the rumen will decrease, a condition called
acidosis. Ruminal acidosis results in inflammation of the rumen lining,
accumulation of lactate producing bacteria which further reduce
ruminal pH, reduction in numbers of protozoa, and reduced efficiency
of cellulose fermentation (Owens et al, 1998; Goad et al, 1998).
Further, lactate accumulation promotes the accumulation of
Fusobacterium necrophorum, a toxin producing bacterium that, if it
escapes the rumen, may colonize the liver and form abscesses
(Nagaraja and Chengappa, 1998). Prolonged periods of ruminal
acidosis may lead to metabolic acidosis; a condition in which the pH of
the blood itself is reduced below normal, reducing the blood’s ability to
carry oxygen and even resulting in death (Owens et al, 1998). Also,
metabolic and ruminal acidosis are theorized to result in or exacerbate
laminitis, a painful and economically important condition of the bovine
foot (Nocek, 1997).
A further complication of too-rapid ruminal grain digestion is a
condition called grain bloat. In grain bloat, there is excessive
production of bacterial polysaccharides which trap fermentation gases
6
(Cheng et al, 1976). These trapped fermentation gases increase the
intra-ruminal pressure, even to the point that the expanding rumen
will compress the lungs, killing the animal.
Methods to Modulate Rates of Digestion
Because of the general acceptance that too rapid ruminal
digestion of starch adversely affects animal health and performance
(Huntington, 1997; Owens et al, 1998; Nocek, 1997; Channon and
Rowe, 2004; Cheng et al, 1998), numerous methods have been
evaluated to modulate starch rate of digestion and/or alleviate its
adverse effects. These methods include varying the extent of grain
processing, the addition of ionophores and/or antibiotics, or buffers to
the diet, and treatment of the grain with formaldehyde, sodium
hydroxide, or polyphenolics. A final method relies on differences
between grain varieties.
Lykos and Varga (1995) observed that rumen starch digestibility
decreased with increasing grain particle size. Similarly, in an
experiment using corn processed by three different methods, Rémond
et al (2004) observed, both in situ and in vivo, that increased particle
size decreased ruminal starch digestibility. Reduced ruminal
7
digestibility led to reduced small intestinal digestibility, although the
amount of starch digested in the small intestine increased. As might
be expected, they also noted a decrease in total tract digestibility.
Zinn (1993) observed an increase in the ruminal digestibility of barley
when it was steam-flaked rather than dry-rolled. In an experiment
attempting to determine the optimal extent of processing of barley
grain, Beauchemin et al (2001) observed that processing barley to a
smaller particle size resulted in a small increase in total tract starch
digestibility, which they warned could lead to problems associated with
acidosis. Although less processed grain had a slower rate of digestion,
total ruminal digestibility was not significantly affected. They noted
that ruminal pH tended to decrease with increased grain processing;
however, they also noted that fecal pH increased with increased grain
processing, probably as a result of decreased starch flow to the large
intestine.
Ionophores are molecules that disrupt the membranes of certain
microbes and have sometimes been used as antibiotics. Ionophores
are generally agreed to have positive effects on the performance of
animals in the feedlot and are widely used as feed additives (McGuffey
et al, 2001). In a diet of 75% corn supplemented with the ionophore
monensin, Zinn and Borques (1993) observed reduced ruminal
8
digestion of organic matter and an increased fraction of organic matter
digested post-ruminally. Also, they noted decreased energetic loss
due to decreased methane production, and increased metabolizable
energy of the diet. Nagaraja et al (1981) noted that the ionophores
monensin and lasalocid were effective in reducing lactic acid
accumulation in steers on a high starch diet. Later (Nagaraja et al,
1987), they showed that a range of other ionophores had the same
effect and that ionophores reduced the population of lactic acid
producing bacteria such as Streptococcus bovis and Lactobacillus
species.
The antibiotic virginiamycin is a fermentation product of
Streptomyces virginiae and is widely used as a poultry, swine, and
cattle feed additive. In 7 experiments over 4 years, Rogers et al
(1995) showed that virginiamycin decreased incidence of liver abscess
in feedlot steers on high energy diets from 30% to less than 20%.
They also observed increased feed efficiency and average daily gain.
Godfrey et al (1992) noted that virginiamycin was effective in
stabilizing the large intestinal pH and lactate levels of sheep fed high
barley diets. Later, in an experiment with sheep fed wheat without an
adjustment period, they observed increased ruminal pH and lower
ruminal L-lactate when virginiamycin was fed (Godfrey et al, 1995).
9
Although it makes intuitive sense that the feeding of buffers to
acidotic animals would stabilize the ruminal pH, in many cases it has
had no effect on ruminal pH. Zinn and Borques (1993) were unable to
detect any effect of buffers on rumen pH, site or extent of starch
digestion, or performance of animals on high grain diets. Similarly, Xu
et al (1994) found no effect of buffers on ruminal pH. In seeming
contrast, Phy and Provenza (1998) found that sheep fed high-wheat
diets drank more water buffered with sodium bicarbonate than unbuffered water. Also, they noted decreased acidosis in animals that
drank buffered water rather than salt water.
Formaldehyde treatment has been shown to decrease ruminal
starch digestion. Michalet-Doreau et al (1997) showed that cereal
grains treated with formaldehyde were less ruminally digestible than
untreated cereal grains and that the effect was dose-dependent, i.e.
treatment with 5% formaldehyde reduced digestibility more than
treatment with 1% formaldehyde. Van Ramshorst and Thomas (1988)
found that formaldehyde treatment increased nitrogen and starch
absorption in the intestines relative to ruminal absorption. After
formaldehyde treatment, starch passed from the rumen to the small
intestine increased from 3.8 to 7.4 % of starch ingested. They found
no effects, however, on rumen pH. Ortega-Cerilla et al (1999), also,
10
found that formaldehyde treatment reduced rumen digestibility of
starch. Schmidt et al (2006) found that either sodium hydroxide
treatment or formaldehyde treatment significantly increased the
amount of starch reaching the small intestine.
Mahmoudzadeh et al (1989) infused starch post-ruminally into
lambs and administered gel capsules containing the phenolic
monomers p-coumaric or ferulic acids. No effect was observed on
starch digestibility. Martínez et al (2005) treated barley grain with 4
concentrations of the polyphenolic compound tannic acid and found
that at the highest level of treatment, the ruminal digestibility of drymatter was reduced. Measurements of starch digestibility were not
reported.
Manipulation of Grain for Improved Animal Performance
By feeding 200 steer calves 10 different barley cultivars, Ramsey
et al (2002) tested the hypothesis that barley ruminal digestibility
affects the health and performance of feedlot steers. The percentage
of animals that bloated at least once was positively correlated with
rate of dry-matter and starch digestion. Incidence of liver abscess,
too, was positively correlated with dry-matter digestibility. In other
words, increased rate of digestion was associated with increased bloat
11
and abscess. Surber et al (2000) reported that ruminal dry-matter
digestibility was negatively correlated with average daily gain and feed
efficiency. In addition, they reported that barley with a slower rate of
digestion had a higher energy value. Particle size was negatively
correlated with both in situ and in vivo digestibility. In contrast to
these results, Boss and Bowman (1996a, b) found that the barley
cultivar showing the greatest weight gain, Harrington, was also the
cultivar of most rapid ruminal digestion. However, the grain producing
greatest average daily gain, corn, was the grain with the slowest rate
of digestion. Also, Harrington did reduce ruminal pH more significantly
than corn (Boss and Bowman, 1996b).
Because proanthocyanidins (or “tannins”) have been shown to
reduce the digestibility of forages, Wang et al (1999) explored whether
naturally occurring barley proanthocyanidins could also affect
digestibility. Their approach was to evaluate the digestibility of
Harrington and three proanthocyanidin-free mutant lines. They found
no significant differences in digestibility between the four lines.
Puroindolines are the primary proteins in determining wheat
grain hardness (Giroux and Morris, 1998). Hordoindolines are the
orthologous genes in barley (Darlington et al, 2001). Beecher et al
(2002) showed that in the cross of the barley cultivars Steptoe and
12
Morex, lines differing in the chromosomal region containing the
hordoindolines also differed in extent of ruminal dry-matter digestion
after 3 hours. It has also been shown that wheat puroindolines can
effect ruminal digestion of wheat dry-matter and starch (Swan et al,
2006).
Based on the results of 18 feedlot trials over 7 years, Surber et
al (2000) used results of laboratory analyses of barley varieties in an
attempt to predict animal performance. They found that barley grain
particle size was negatively correlated with average daily gain (r=0.36, P=0.007), feed efficiency (r=-0.37, P=0.007), energy for
maintenance (r=-0.59, P<0.001), and energy for gain (r=-0.60,
P<0.001). These results indicate that not only is barley ruminal
digestion rate an important factor for ruminal and over-all animal
health, but that it is also an important factor for animal productivity.
13
References
Beauchemin K.A., W.Z. Yang, and L.M. Rode. 2001. Effects of barley
grain processing and extent of digestion of beef feedlot finishing diets.
J. Anim. Sci. 79:1925-1936.
Beecher B., J. Bowman, J.M. Martin, A.D. Bettge, C.F. Morris, T.K.
Blake, and M.J. Giroux. 2002. Hordoindolines are associated with a
major endosperm-texture QTL in barley (Hordeum vulgare). Genome.
45:584-591.
Boss D.L. and J.G.P. Bowman. 1996a. Barley varieties for finishing
steers: I. Feedlot performance, in vivo diet digestion, and carcass
characteristics. J. Anim. Sci. 74:1967-1972.
Boss D.L. and J.G.P. Bowman. 1996b. Barley varieties for finishing
steers: II. Ruminal characteristics and rate, site, and extent of
digestion. J. Anim. Sci. 74:1973-1981.
Bowman J.G.P., T.K. Blake, L.M.M. Surber, D.K. Habernicht, and H.
Bockelman. 2001. Feed-quality variation in the barley core collection
of the USDA National Small Grains Collection. 41:863-870.
Channon, A.F., J.B. Rowe, and R.M. Herd. 2004. Genetic variation in
starch digestion in feedlot cattle and its association with residual feed
intake. Aust. J. Exp. Agr. 44:469-474.
Cheng K.-J., R. Hironaka, G.A. Jones, T. Nicas, and J.W. Costerton.
1976. Frothy feedlot bloat in cattle: production of extracellular
polysaccharides and development of viscosity in cultures of
Streptococcus bovis. Can. J. Microbiol. 22:450-459.
Cheng K.-J., T.A. McAllister, J.D. Popp, A.N. Hristov, Z. Mir, and H.T.
Shin. 1998. A review of bloat in feedlot cattle. J. Anim. Sci. 76:299308.
Darlington H.F., J. Rouster, L. Hoffman, N.G. Halford, P.R. Shewry,
and D.J. Simpson. 2001. Identification and molecular characterization
of hordoindolines from barley grain. Plant Mol. Biol. 47:785-794.
14
Giroux M.J. and C.F. Morris. 1998. Wheat grain hardness results from
highly conserved mutations in the friabilin components puroindoline a
and b. Proc. Natl. Acad. Sci. USA. 95:6262-6266.
Goad D.W., C.L. Goad, and T.G. Nagaraja. 1998. Ruminal microbial
and fermentative changes associated with experimentally induced
subacute acidosis in steers. J. Anim. Sci. 76:234-241.
Godfrey S.I., M.D. Boyce, J.B. Rowe, and E.J. Speijers. 1992.
Changes within the digestive tract of sheep following engorgement
with barley. Aust. J. Agric. Res. 44:1093-1101.
Godfrey S.I., J.B. Rowe, G.R. Thorniley, M.D. Boyce, and E.J. Speijers.
1995. Virginiamycin to protect sheep fed wheat, barley or oats from
grain poisoning under simulated drought feeding conditions. Aust. J.
Agric. Res. 46:393-401.
Harmon, D. L., and K. R. McLeod. 2001. Glucose uptake and
regulation by intestinal tissues: Implications and whole-body
energetics. J. Anim. Sci. 79(E. Suppl.):E59–E72.
Huntington G.B. 1997. Starch utilization by ruminants: from basics to
the bunk. J. Anim. Sci. 75:852-867.
Kreikemeier K.K., D.L. Harmon, R.T. Brandt, Jr., T.B. Avery, and D.E.
Johnson. 1991. Small intestinal starch digestion in steers: effect of
various levels of abomasal glucose, corn starch and corn dextrin
infusion on small intestinal disappearance and net glucose absorption.
J. Anim. Sci. 69:328-338.
Lewis J.M., T.J. Klopfenstein, G.A. Pfeiffer, and R.A. Stock. 1990a. An
economic evaluation of the differences between intensive and
extensive beef production systems. J. Anim. Sci. 68:2506-2516.
Lewis J.M., T.J. Klopfenstein, R.A. Stock, and M.K. Nielsen. 1990b.
Evaluation of intensive vs extensive systems of beef production and
the effect of beef cow milk production on postweaning performance. J.
Anim Sci. 68:2517-2524.
Lykos T. and G.A. Varga. 1995. Effects of processing method on
degradation characteristics of protein and carbohydrate sources in situ.
J. Dairy Sci. 78:1789-1801.
15
Mahmoudzadeh H., E. Karangwa, G.E. Mitchell Jr., R.B. Muntifering,
and R.E. Tucker. 1989. Post-ruminal digestion of starch in the
presence of phenolic monomers. Anim. Feed Sci. Tech. 27:111-116
Martínez T.F., F.J. Moyano, M. Díaz, F.G. Barroso, and F.J. Alarcón.
2005. Use of tannic acid to protect barley meal against ruminal
degradation. J. Sci. Food Agric. 85:1371-1378.
McGuffey R.K., L.F. Richardson, and J.I.D. Wilkinson. 2001.
Ionophores for dairy cattle: current status and future outlook. J.
Dairy. Sci. 84(E. Suppl.):E194-E203.
Michalet-Doreau B., C. Philippeau, M. Doreau. 1997. In situ and in
vitro ruminal starch degradation of untreated and formaldehydetreated wheat and maize. 37:305-312. Nagaraja T.G., T.B. Avery,
Bartley E.E., S.J. Galitzer, and A.D. Dayton. 1981. Prevention of
lactic acidosis in cattle by lasalocid or monensin. J. Anim. Sci. 53:206216.
Nagaraja T.G. and M.M. Chengappa. 1998. Liver abscesses in feedlot
cattle: a review. 76:287-298.
Nagaraja T.G., M.B. Taylor, D.L. Harmon, and J.E. Boyer. 1987. In
vitro lactic acid inhibition and alterations in volatile fatty acid
production by antimicrobial feed additives. J. Anim. Sci. 65:10641076
Nocek, J.E. 1997. Bovine acidosis: implications on laminitis. J. Dairy
Sci. 80:1005-1028
Ortega-Cerilla M.E., H.J. Finlayson, D.G. Armstrong. 1999. Protection
of starch in barley against rumen degradation by glutaraldehyde and
formaldehyde as assessed by the Dacron bag technique. Anim. Feed
Sci. Tech. 77:83-90.
Owens F.N., D.S. Secrist, W. J. Hill, and D.R. Gill. 1997. The effect of
grain source and grain processing on performance of feedlot cattle: a
review. J. Anim. Sci. 75:868-879.
16
Owens F.N., D.S. Secrist, W. J. Hill, and D.R. Gill. 1998. Acidosis in
cattle: a review. J. Anim. Sci. 76:275-286.
Owens F.N., R.A. Zinn, and Y.K. Kim. 1986. Limits to starch digestion
in the ruminant small intestine. J. Anim. Sci. 63:1634-1648.
Phy T.S., and F.D. Provenza. 1998. Sheep fed grain prefer foods and
solutions that attenuate acidosis. J. Anim. Sci. 76:954-960.
Ramsey P.B., G.W. Mathison, and L.A. Goonewardene. 2002. Effect of
rates and extents of ruminal barley grain dry matter and starch
disappearance on bloat, liver abscesses, and performance of feedlot
steers. Anim. Feed Sci. Tech. 13:145-157.
Rémond D., J.I. Cabrera-Estrada, M. Champion, B. Chauveau, R.
Coudure, and C. Poncet. Effect of corn particle size on site and extent
of starch digestion in lactating dairy cow. J. Dairy Sci. 87:1389-1399.
Rogers J.A., M.E. Branine, C.R. Miller, M.I. Wray, S.J. Bartle, R.L.
Preston, D.R. Gill, R.H. Pritchard, R.P. Stilborn, and D.T. Bechtol.
1995. Effects of dietary virginiamycin on performance and liver
abscess incidence in feedlot cattle. J. Anim. Sci. 73:9-20.
Russell J.B., and J.L. Rychlik. 2001. Factors that alter rumen
microbial ecology. Science. 292:1119-1122.
Schmidt J., T. Tóth, and J. Fábián. 2006. Rumen fermentation and
starch degradation by Holstein steers fed sodium-hydroxide or
formaldehyde-treated wheat. Acta. Vet. Hung. 54:1588-2705.
Surber L.M.M., J.G.P. Bowman, T.K. Blake, D.D. Hinman, D.L. Boss,
and T.C. Blackhurst. 2000. Prediction of barley feed quality for beef
cattle from laboratory analyses. Proc. West. Sec. Am. Soc. Anim. Sci.
51:454-457.
Swan C.G., J.G.P. Bowman, J.M. Martin, and M.J. Giroux. 2006.
Increased puroindoline levels slow ruminal digestion of wheat (Triticum
aestivum L.) starch by cattle. J. Anim. Sci. 84:641-650.
USDA National Agricultural Statistics Service. 2007.
(http://www.nass.usda.gov/)
17
van Ramshorst H., and P.C. Thomas. 1988. Digestion in sheep of
diets containing barley chemically treated to reduce its ruminal
digestibility. J. Sci. Food. Agric. 42:1-7.
Van Soest, P.J. 1994. Nutritional ecology of the ruminant. Cornell
University Press.
Wright E.M. 1993. The intestinal Na+/glucose cotransporter. Annu.
Rev. Physiol. 55:575-589.
Xu S., J.H. Harrison, R.E. Riley, and K.A. Loney. 1994. Effect of
buffer addition to high grain total mixed rations on rumen pH, feed
intake, milk production, and milk composition. J. Dairy. Sci. 77:782788
Zinn R.A. 1993. Influence of processing on the comparative feeding
value of barley for feedlot cattle. J. Anim. Sci. 71:3-10
Zinn R.A., and J.L. Borques. 1993. Influence of sodium bicarbonate
and monensin on utilization of a fat-supplemented, high-energy
growing-finishing diet by feedlot steers. J. Anim. Sci. 71:18-25
18
CHAPTER 2
EVALUATION OF THE ANT18 MUTATION AS A POTENTIAL TOOL TO
MODULATE RATE OF BARLEY RUMINAL DRY-MATTER DISAPPEARANCE
Abstract
Rate of ruminal dry-matter digestion (DMD) is an important
parameter of cereal grain as feed for ruminants. Ruminal digestion is
primarily the result of microbial attack. The flavonoid
dihydroquercetin, and flavonoids in general, are antimicrobial agents.
Dihydroquercetin accumulates in the testa layer of mutant ant18
barley grain. Therefore, experiments were conducted to test whether
the ant18 mutation could be used to modulate rate of ruminal
digestion. The barley cultivar Klages and its ant18 mutant were grown
in a dryland field. The barley cultivars Advance and Cougbar and their
ant18 mutants were grown in a greenhouse and a dryland field. The in
situ DMD of Klages and its mutant were evaluated in a time course
experiment, and Klages was found to have a higher rate of digestion
than its ant18 mutant. This result was confounded by the smaller
average particle size and higher starch content of Klages relative to
the mutant. To negate the effects of differing particle size, Advance,
Cougbar, their mutants were cracked to a finer particle size and
19
evaluated for in situ DMD after 3 hours in rumen. Genotype by
environment interaction was important in this second experiment, with
the ant18 mutation reducing DMD in greenhouse grown grain but
increasing DMD in dryland field grown grain. Because of the
inconsistent effects of the ant18 mutation on barley grain DMD across
environments, and because of it may reduce grain starch content,
ant18 mutants are unlikely to be useful as feed for ruminants.
Introduction
Barley and cattle are important segments of Montana’s
agriculture, accounting for an average of 4.9 and 50.3 percent,
respectively, of agricultural receipts between 2001 and 2005 (USDA
National Agricultural Statistics Service). Approximately 850,000
metric tons, or 13.6%, of Montana barley grain sold in market year
2005 was sold as feed barley (USDA National Agricultural Statistics
Service), with the remainder sold as malting barley.
Rate of ruminal digestion is an important parameter of cereal
grain as feed for ruminants. It has been reported that the extent of
barley dry-matter digestibility (DMD) after 3 hours in the rumen was
negatively correlated with the metabolizable energy content of barley
20
grain as a feed for feedlot steers, and that DMD was negatively
correlated with animal performance as measured by average daily gain
and feed efficiency (Surber et al, 2000). Similarly, it was later
reported that average daily gain was positively correlated with the
proportion of rolled barley dry-matter that slowly disappeared in the
rumen (Ramsey et al, 2002). Further, the proportion of slowly
disappearing dry-matter was negatively correlated with incidence of
bloat and liver abscess (ibid.) and it is generally accepted that high
rate of ruminal digestion of cereal starch is a contributing factor to
bloat, acidosis, and liver abscess (for reviews see: Cheng et al, 1998;
Owens et al, 1998; Nagaraja and Chengappa, 1998).
Along with widely-used feed additives such as ionophores that
ameliorate the negative effects of rapid ruminal digestion (Nagaraja et
al, 1987; Rogers et al, 1995), several feed treatments have been
reported to decrease the rate of ruminal digestion of cereal grain drymatter. These treatments include less intensive processing to increase
grain particle size (Beauchemin et al, 2001), treatment of grain with
formaldehyde or sodium hydroxide (Michalet-Doreau et al, 1997;
Schmidt et al, 2006), and treatment with tannic acid (Martínez et al,
2005).
21
Genetic effects have also been explored as a route to altering
the rate of digestion of cereals. In general, six-row barleys digest less
rapidly in the rumen than two-row barleys (Bowman et al, 2001). In
addition to controlling wheat grain hardness, functional puroindolines
are able to decrease rate of wheat dry-matter and starch digestion
(Swan et al, 2006). Further, the genetic locus encoding barley’s
orthologous proteins has been associated with changes in ruminal drymatter digestibility (Beecher et al, 2002). In a digestibility comparison
of Harrington and three proanthocyanidin-free mutants, Wang et al
(1999) reported that the endogenous proanthocyanidins of the barley
testa layer do not affect DMD.
Certain flavonoids possess anti-microbial activity (reviewed by
Cushnie and Lamb, 2005). The flavonoid dihydroquercetin
accumulates in small amounts in the grain testa layer of mutant
anthocyanin-free ant18 barley plants, and dihydroquercetin has been
shown to be an inhibitor of the fungi Fusarium poae, F. culmorum, and
F. graminearum (Skadhauge et al, 1997). Because ruminal digestion
is primarily microbial (Russell and Rychlik, 2001), and because of the
anti-microbial effects of flavonoids and dihydroquercetin in particular,
it is reasonable to hypothesize that accumulation of dihydroquercetin
in ant18 barley grain will result in a slower rate of ruminal digestion.
22
This paper reports the results of experiments intended to test this
hypothesis.
Materials and Methods
Plant Material
Seed of the cultivars Advance (CIho 15804) and Cougbar (PI
496400) were obtained from the USDA National Small Grains
Collection (http://www.ars-grin.gov/npgs/). Seed of Klages was
available here in the stores of the Montana State barley breeding
program. Seed of anthocyaninless-18 mutants was kindly provided by
Dr. Diter von Wettstein. These anthocyaninless-18 (ant18) mutants
were ant18.623 in the Klages background, ant18.660 in the Cougbar
background, and ant18.592 and ant18.621 in the Advance
background.
Cougbar, Advance, and their mutants were planted both in the
greenhouse and in the field. In the greenhouse, they were planted in
a randomized complete block design with two replications. They were
planted in ten-inch pots with five plants per pot. In the field they were
hand-planted in three meter dry-land rows in a randomized complete
block design, twenty seeds of each line in each of three replications.
23
Klages and its mutant were planted in the field in a randomized
complete block design. After harvest, we realized that enough seed
for a time-course experiment could be obtained only by bulking the
seed. Therefore, the time-course experiment was performed using
these bulked grains.
Particle size, starch content, and DMD Analysis
After seed harvest and cleaning, seed of each line was cracked
using a Buehler disc mill (Buehler-Miag, Braunschweig, Germany) with
disc spacing set to 1.5 mm or 1.25 mm, depending on the experiment.
Fifty grams or ten grams, depending on the amount available, of each
cracked sample was placed on the top sieve of a stack of 5
International Standards Organization sieves. Sieves used were 2,360,
1,700, 850, 425, and 90 µm in screen opening diameter. The sieve
stack was shaken for five min using a RoTap shaker (Tyler Co., Mentor,
OH). Geometric mean particle size (dgw) of each line was calculated on
a weight basis of the geometric mean of the diameter openings in 2
adjacent sieves in a stack using the equation (Pfost and Headley,
1976) (dgw) = log–1 [
(Wi log di)/
Wi] in which Wi = weight of
material in sieve i and di = diameter of the sieve i. Dry-matter content
of each line was determined using AOAC method 930.15 (2000) for
24
oven drying and was replicated twice. Before measuring starch
content, grain samples were ground through a 0.5 mm screen using an
UDI Cyclone Sample Mill (UDI Corporation, Boulder, Colorado, USA).
Starch content was determined using the Megazyme starch assay kit
(Megazyme International, Brey, Ireland) and was replicated twice.
The cows used for the DMD analysis were maintained on a diet
of low quality grass hay consumed ad libitum and 3.6 kg per animal
per day dry-rolled barley. The cows had access to fresh water at all
times and were maintained on the diet for 14 days before conducting
the DMD analysis. DMD was determined as in Vanzant et al (1998).
Five g of each cracked grain sample was weighed into each of four 10by 20-cm, 50 μm pore size polyester bags (Ankom Technology,
Fairport, NY) and sealed with a Clamco impulse sealer (Clamco Corp.,
Cleveland, OH). For the time-course experiment, for each time point
two replications of each sample were placed into the rumen of each of
two ruminally cannulated cows. Bags were placed in the rumen at
12.0, 9.0, 6.0, 3.0, 2.0, 1.5, 1.0, and 0.5 hours before the time of
removal and removed at the same time as recommended in Vanzant
et al (1998). For the similar particle size experiment, two replications
of each sample were placed in the rumen of each of two ruminally
cannulated cows at the same time, and removed at the same time
25
after three hours. Also included in each incubation were two empty
bags to correct for DM content from microbial contamination and two
bags of the variety ‘Harrington.’ After removal from the rumen, the
bags were rinsed under cold water until the rinse water ran clear. The
bags were dried at 60° for 48 h and then weighed. Percent DMD was
calculated as (sample weight in * mean DM content) – (dry weight out
– bag weight)/(sample weight in * mean DM content) * 100.
Statistical Analysis
The time-course experiment was analyzed using PROC MIXED of
SAS (SAS Institute, Incorporated, Cary, North Carolina, USA) as
outlined in Littell et al, (1998). Values of DMD of replications of a
sample within a cow at a time were averaged, and cows were treated
as random blocks. Covariance was modeled with an autoregressive
structure.
Data from the similar particle size experiment were analyzed as
a split-plot design using PROC GLM of SAS (SAS Institute,
Incorporated, Cary, North Carolina, USA). Cows and blocks within
environment were treated as random. Ant18 status (mutant versus
wildtype), cow, environment, sample, block within environment, and
genotype by environment interaction were included in the model. One
26
row of Advance in the field produced no seed because of water stress,
and one row of ant18.592 was misplaced.
Results and Discussion
Time course experiment
As an initial test of the effects of the ant18 mutation on DMD,
grain from Klages and grain from ant18.623(Klages) were evaluated in
a DMD time-course experiment. Grain from each line was cracked in a
Buehler disc mill with the disc spacing set to 1.5 mm. After particlesizing, the samples were evaluated for DMD at various times of
incubation in the rumen. Klages had a mean particle size of 1355 μm,
while ant18.623(Klages) had a mean particle size of 1603 μm.
Dry-matter disappearance of Klages and ant18.623(Klages)
increased as time increased (Figure 2-1). Differences between the two
lines were significant at each time point, i.e. line x time interactions
were not significant (P>0.65). Averaged over time points, DMD of
ant18.623(Klages) was 7.9 percentage units less than DMD of Klages
(P=0.03).
27
Klages
Dry-matter disappearance, %
90
ant18.623(Klages)
80
70
60
50
40
30
20
10
0
0
1
2
3
4
5
6
7
8
9
10
11
12
Hours in rumen
Figure 2-1. In situ dry-matter disappearance of Klages and
ant18.623(Klages) at 0.5, 1.0, 1.5, 2.0, 3.0, 6.0, 9.0, and 12.0 hours.
The starch content of Klages was greater than the starch content
of its ant18 mutant (47% and 41%, respectively). It has repeatedly
been emphasized that grain digestibility cannot be evaluated without
taking into account the effects of grain-processing on digestibility, and
the mean particle size of ant18.623(Klages) was greater than the
mean particle size of Klages. Because of the differences in starch
content and particle size, further experimentation was conducted to
reduce any possible effects of particle size on DMD. We also wanted to
determine if the ant18 mutation reduced starch content in other ant18
mutants.
28
Similar particle size experiment
All ground grain of Klages and its ant18 mutant had been
expended in the time course experiment, so grain of two other
cultivars and their ant18 mutants was used instead: Cougbar,
ant18.660(Cougbar), Advance, ant18.592(Advance), and
ant18.621(Advance). All five of these lines were grown in two
replications in the greenhouse, and three replications in the field.
Individual samples were cracked with a disc mill spacing of 1.25
millimeters and particle sized. The decreased disc spacing was an
attempt to minimize any differences in particle size between cultivars
and their ant18 mutants. This attempt was successful. After this grain
processing, ant18 effects were insignificant (Table 2-1).
Table 2-1. Estimated differences and standard error of difference in particle
size of wildtype and ant18 lines* after cracking with disc mill spacing of 1.25
mm
*
Difference
Environment
Estimate (μm)
Standard Error
P Value
wt-ant18
Field
62
47
0.205
wt-ant18
Greenhouse
66
52
0.222
wt-ant18
Both
64
35
0.089
wt-wildtype lines, ant18-ant18 lines
Factorial analysis of variance of the DMD data revealed
significant interaction between ant18 effects and environmental effects
29
(Table 2-2). This result is due to differing significance of ant18
between environments (Figure 2-2). In the greenhouse, ant18 had
significantly reduced DMD (P=0.02), while in the field ant18 lines had
significantly higher DMD (P=0.02).
Table 2-2. Sources of variation in DMD between Klages and ant18 lines after
cracking with disc mill spacing of 1.25 mm
Source
Cow
F Value
P Value
3595.8
149.07
<0.001
0.3
0.01
0.915
Background
306.2
12.70
0.001
Block(Env.)
137.3
1.90
0.147
Environment
729.4
30.24
<0.001
ant18 X Env.
294.5
12.21
0.001
a
ant18
a
Sum of Squares
wildtype versus ant18 averaged over genetic background
A possible reason for the varying digestibility between environments is
revealed by Figures 2-3. Starch is a highly digestible component of
cereal grains (Huntington, 1997). In the greenhouse, wildtype lines
tended toward more starch than their mutants. In the field the
reverse was true. Starch was measured on bulk ground grain of lines
(for example, bulked Advance of blocks 1,2, and 3 in the field), rather
than on individual plantings. It could be that measuring starch on
each individual planting could give enough statistical power to declare
significant differences in starch content
30
75
DMD %
60
Wild-type
ant18
45
30
15
greenhouse
field
Figure 2-2. DMD of wildtype lines and their ant18 mutants. Error bars
indicate standard error of 2.3.
55
50
Starch %
45
40
wildtype
ant18
35
30
25
20
greenhouse
field
Figure 2-3. Starch content of wildtype lines and their ant18 mutants. Error
bars indicate standard error of 1.8.
Starch content of the field-grown ant18 line was decreased in
the time course experiment, while in this experiment the starch
31
content tends toward an increase in the field-grown ant18 lines. If
repeated starch measurements do indicate that the starch content of
the ant18 mutants are increased relative to the wildtype lines, the
question is “Why?” Advance, Cougbar, and their mutants were planted
one-week later than Klages and its mutant. Also, they were planted
by hand at a much shallower depth than Klages and its mutant. It
seems clear that environment does impact the effects of ant18 on
starch content.
Conclusions
The time-course experiment shows that the ant18 mutation can
reduce dry-matter digestibility, though this experiment can not
differentiate whether the mode of action is via reduced particle size,
reduced starch content, or by the hypothesized accumulation of
dihydroquercetin. Further, as the starch content of the ant18 line was
reduced relative to Klages, the ant18 mutation in the Klages
background is likely to be detrimental to animal performance, rather
than beneficial.
The experiment using Advance, Cougbar, and three ant18
mutations of these lines showed a significant genotype by environment
32
interaction, suggesting that this gene may be an unreliable way to
modulate barley grain DMD.
Acknowledgments
I thank Jack Martin for his advice on statistical analyses. I am
thankful to Dr. Diter von Wettstein for providing the seed of the ant18
mutants. Thanks go to MacKenzie Ellison for starch assays.
33
References
AOAC. 2000. Official Methods of Analysis. 17th ed. Assoc. Offic. Anal.
Chem., Gaithersburg, MD.
Beauchemin K.A., W.Z. Yang, and L.M. Rode. 2001. Effects of barley
grain processing and extent of digestion of beef feedlot finishing diets.
J. Anim. Sci. 79:1925-1936.
Beecher B., J. Bowman, J.M. Martin, A.D. Bettge, C.F. Morris, T.K.
Blake, and M.J. Giroux. 2002. Hordoindolines are associated with a
major endosperm-texture QTL in barley (Hordeum vulgare). Genome.
45:584-591.
Bowman J.G.P., T.K. Blake, L.M.M. Surber, D.K. Habernicht, and H.
Bockelman. 2001. Feed-quality variation in the barley core collection
of the USDA National Small Grains Collection. 41:863-870.
Cheng K.-J., T.A. McAllister, J.D. Popp, A.N. Hristov, Z. Mir, and H.T.
Shin. 1998. A review of bloat in feedlot cattle. J. Anim. Sci. 76:299308.
Cushnie T.P., and A.J. Lamb. 2005. Antimicrobial activity of
flavonoids. Int. J. Antimicrob. Agents. 26:343-356.
Martínez T.F., F.J. Moyano, M. Díaz, F.G. Barroso, and F.J. Alarcón.
2005. Use of tannic acid to protect barley meal against ruminal
degradation. J. Sci. Food Agric. 85:1371-1378.
Michalet-Doreau B., C. Philippeau, M. Doreau. 1997. In situ and in
vitro ruminal starch degradation of untreated and formaldehydetreated wheat and maize. 37:305-312.
Nagaraja T.G. and M.M. Chengappa. 1998. Liver abscesses in feedlot
cattle: a review. 76:287-298.
Owens F.N., D.S. Secrist, W. J. Hill, and D.R. Gill. 1998. Acidosis in
cattle: a review. J. Anim. Sci. 76:275-286.
34
Pfost, H., and V. Headley. 1976. Methods of determining and
expressing particle size. Page 517 in Feed Manufacturing Technology
II. H. Pfost, ed. Anim. Feed Manuf. Assoc., Arlington, VA
Ramsey P.B., G.W. Mathison, and L.A. Goonewardene. 2002. Effect of
rates and extents of ruminal barley grain dry matter and starch
disappearance on bloat, liver abscesses, and performance of feedlot
steers. Anim. Feed Sci. Tech. 13:145-C157.
Rogers J.A., M.E. Branine, C.R. Miller, M.I.Wray, S.J. Bartle, R.L.
Preston, D.R. Gill, R.H. Pritchard, R.P. Stilborn, and D.T. Bechtol.
1995. Effects of dietary virginiamycin on performance and liver
abscess incidence in feedlot cattle. J. Anim. Sci. 73:9-20.
Skadhauge B., K.K. Thomsen, and D. von Wettstein. 1997. The role
of the barley testa layer and its flavonoid content in resistance to
Fusarium infections. Hereditas. 126:147-160.
Surber L.M.M., J.G.P. Bowman, T.K. Blake, D.D. Hinman, D.L. Boss,
and T.C. Blackhurst. 2000. Prediction of barley feed quality for beef
cattle from laboratory analyses. Proc. West. Sec. Am. Soc. Anim. Sci.
51:454-457.
Swan C.G., J.G.P. Bowman, J.M. Martin, and M.J. Giroux. 2006.
Increased puroindoline levels slow ruminal digestion of wheat (Triticum
aestivum L.) starch by cattle. J. Anim. Sci. 84:641-650.
Vanzant E.S., R.C. Cochran, and E.C. Titgemeyer. 1998.
Standardization of in situ techniques for ruminant feedstuff evaluation.
J. Anim. Sci. 76:2717-2729
Wang Y, T.A. McAllister, Z.J. Xu, M.Y. Gruber, B. Skadhauge, B. JendeStrid, and K.-J. Cheng. 1999. Effects of proanthocyanidins, dehulling
and removal of pericarp on digestion of barley grain by ruminal microorganisms. J. Sci. Food Agric. 79:929-938.
35
CHAPTER 3
MAPPING OF QUANTITATIVE TRAIT LOCI FOR FEED-QUALITY RELATED
TRAITS IN A TWO-ROWED INBRED BARLEY POPULATION
Abstract
Barley and cattle are important components of Montana’s
economy. Recent research has identified important characteristics of
barley grain as feed for cattle; these characteristics include high starch
content, large particle size after dry-rolling, and low extent of drymatter digestibility (DMD) after 3 hours incubation in the rumen.
Evaluation of particle size and DMD is time-consuming and relatively
expensive, therefore it would be desirable to identify molecular
markers useful for marker-assisted selection. With this aim, grain
from 123 recombinant inbred lines (RILs) derived from a cross of the
cultivars Haxby and Baku were evaluated for DMD, starch, and particle
size and genotyped with 218 AFLP, SSR, and STS markers. By
composite interval quantitative trait locus (QTL) analysis, two loci were
detected on barley chromosomes 6H and 7H that together explain
43% of phenotypic variation for DMD. An 86-member F7 subset of the
original population was also genotyped using the Illumina GoldenGate
36
SNP genotyping assay. The loci on 6H and 7H were again detected.
The effects of the Nud locus were also detected as was a new, but
minor QTL for DMD on 1H. One marker with significant effects on
DMD and particle size was genotyped in a RIL population derived from
a cross of the cultivars Drummond and Baku. The locus was not found
to be significant in this population, though this result may be due to
masking by the Vrs1 locus. It was concluded that the very low DMD of
Baku is due to many genes of minor effect.
Introduction
Barley and cattle are important segments of Montana’s
agriculture, accounting for an average of 4.9 and 50.3 percent,
respectively, of agricultural receipts between 2001 and 2005 (USDA
National Agricultural Statistics Service). Approximately 850,000
metric tons, or 13.6%, of Montana barley grain sold in market year
2005 was sold as feed barley (USDA National Agricultural Statistics
Service), with the remainder being sold as malting barley.
Despite the importance of barley grain as feed for ruminants,
relatively little work has been done to determine characteristics of a
quality feed barley. However, in an experiment utilizing the results of
37
18 feedlot experiments over the course of 7 years, Surber et al (2000)
reported that high starch content and low acid-detergent fiber (ADF)
content are correlated with feedlot steer performance. Also, it was
noted that the extent of barley dry-matter digestibility (DMD) after 3
hours in the rumen was negatively correlated with the metabolizable
energy content of barley grain as a feed for feedlot steers and that
DMD was negatively correlated with animal performance as measured
by average daily gain and feed efficiency (Surber et al, 2000).
Similarly, it was later reported that average daily gain was positively
correlated with the proportion of rolled barley dry-matter that slowly
disappeared in the rumen (Ramsey et al, 2002). Further, the
proportion of rolled barley dry-matter that slowly disappeared in the
rumen was negatively correlated with incidence of bloat and liver
abscess (ibid.). Grain particle size after cracking or dry-rolling is
negatively correlated with ruminal DMD (Lykos and Varga, 1995;
Surber et al, 2000; Rémond et al, 2004), therefore particle size may
be a useful indicator of the DMD of a potential barley line.
As the assays (in situ or in vitro) for DMD are relatively time
consuming and expensive, it would be preferable to use molecular
markers for early selection of lines in barley breeding programs.
Previously, markers have been identified associated with variation in
38
DMD on barley chromosomes 1H, 3H, and 4H in the Steptoe/Morex
population (Bowman et al, 1996), 1H and 3H in the Lewis/Baronesse
population (Abdel-Haleem, 2004), and 2H in the Valier/PI 370970
population (ibid.).
The objective of this research is to identify molecular markers
linked to loci that decrease extent of barley DMD after three hours in
the rumen or that increase cracked grain mean particle size, with the
ultimate aim of deploying these markers for marker-assisted selection.
Materials and Methods
Plant Material
A 123-member F5 derived recombinant inbred line (RIL)
population was developed by single seed descent from a cross
between the 2-rowed feed cultivar ‘Haxby’ (PI 646160) and the 2rowed USDA barley collection accession ‘Baku’ (PI 28624). Seed from
greenhouse grown F5 plants was bulked and planted at the Arthur H.
Post Research Farm near Bozeman, MT in the spring of 2006. These
F5 derived F6 seed were planted in rain-fed un-replicated 3 meter
rows, with a check variety planted as every twelfth row. Another 96member F5 derived F6 RIL population was developed by the same
39
means from the same parents to be used as a validation population.
An additional 94-member F5 derived validation population was derived
by the same means from a simple cross of ‘Baku’ and the 6-rowed
variety ‘Drummond’. The two validation populations were also planted
in the spring of 2006 at the Arthur Post Research Farm in rainfed unreplicated 3 meter rows with a check variety every twelfth row.
Harvested seed was cleaned and de-awned.
Phenotypic Data
Flowering date for each line was defined as the Julian day when
50% of heads had emerged from the sheath. Plant height was
measured on three individuals of each line at physiological maturity
and averaged. The weight of 500 kernels of each RIL were counted in
triplicate using an electronic seed counter (The Old Mill Company,
Savage, MD) and weighed. 500 kernel weight rather than the more
frequently reported 1000 kernel weight was reported because several
entries had fewer than 1000 kernels, but all entries had at least 500
kernels. Seed of each line was cracked using a Buehler disc mill
(Buehler-Miag, Braunschweig, Germany) with disc spacing set to 0.05
inches. The geometric mean particle size of the cracked seed was
determined in duplicate as in Swan et al (2006). Fifty grams of each
40
cracked sample was placed on the top sieve of a stack of 5
International Standards Organization sieves. Sieves used were 2,360,
1,700, 850, 425, and 90 µm in screen opening diameter. The sieve
stack was shaken for 5 min using a RoTap shaker (Tyler Co., Mentor,
OH). Geometric mean particle size (dgw) of each line was calculated on
a weight basis of the geometric mean of the diameter openings in 2
adjacent sieves in a stack using the equation (Pfost and Headley,
1976) (dgw) = log–1 [
(Wi log di)/
Wi] in which Wi = weight of
material in sieve i and di = diameter of the sieve i. Dry matter (DM)
content of thirty random lines was determined using AOAC method
930.15 (2000) for oven drying and replicated twice. The mean DM
content of the thirty lines was considered to be representative of the
DM content for all lines.
The cows used for the DMD analysis were maintained on a diet
of low quality grass hay consumed ad libitum and 3.6 kg per animal
per day dry-rolled barley. The cows had access to fresh water at all
times and were maintained on the diet for 14 days before conducting
the DMD analysis. DMD was determined as in Vanzant et al (1998).
Five g of each cracked grain sample was weighed into each of four 10by 20-cm, 50 μm pore size polyester bags (Ankom Technology,
Fairport, NY) and sealed with a Clamco impulse sealer (Clamco Corp.,
41
Cleveland, OH). Twenty eight polyester bags (representing 14 RILs)
were placed in the rumen of each of two ruminally cannulated cows at
the same time. Also included in each incubation were two empty bags
to correct for DM content from microbial contamination and two bags
of the variety ‘Harrington.’ After removal from the rumen, the bags
were rinsed under cold water until the rinse water ran clear. The bags
were dried at 60° for 48 h and then weighed. Percent DMD was
calculated as (sample weight in * mean DM content) – (dry weight out
– bag weight)/(sample weight in * mean DM content) * 100.
Anchor Markers
DNA was extracted from individual two-week-old F5 plants using
the DNeasy Plant Mini kit (Qiagen Inc., Valencia, CA). PCR was
performed in a 15 μL reaction mix consisting of 0.3 μM of forward and
reverse primers, 0.5 units of Taq DNA polymerase (Promega
Corporation, Madison, WI), 0.2 mM of each dNTPs, 1 X PCR buffer (50
mM KCL, 10 mM Tris-HCl, 1 g L1 Triton X-1000), 2.5 mM MgCl2, 25 ng
of template DNA, and distilled H2O to a volume of 15 μL. The PCR
amplification consisted of an initial denaturation at 94°C for 3 minutes,
followed by 40 cycles of three steps: denaturation at 94°C for 30
seconds, annealing at 52°C for 30 seconds, and elongation at 72°C for
42
40 seconds. The final step was an elongation step at 72°C for 5
minutes.
Simple sequence repeat (SSR) (Li et al, 2003; Ramsay et al,
1997; Struss et al, 1998) and sequence-tagged site (STS) molecular
markers (Blake et al, 1996) were screened against DNA from ‘Baku’
and ‘Haxby’, those markers showing polymorphism between the
parents were then used to genotype the entire population. Markers
amplifying fragments that differed by more than 10 nucleotides
between the parents were amplified in the population and separated
by ethidium bromide stained 6% polyacrylamide gels in 0.5 X TBE
buffer. Markers that produced bands differing by less than 10
nucleotides were amplified in the population using a three-primer
amplification method (Schuelke, 2000). The three primers are: 1) the
standard reverse primer, 2) the M13 primer (of sequence 5’ CAC GAC
GTT GTA AAA CGA C) labeled with one of the WellRED fluorescent dyes
(Sigma-Aldrich, St. Louis, MO) and 3) the forward primer to which the
M13 primer sequence has been added at the 5’ end. Fifty ng of total
DNA was amplified in a volume of 15 μL of 1 X Taq buffer, 0.1 mM
each DNTP, 2.5 mM MgCl2, 0.5 units Taq DNA polymerase (Promega),
distilled water to volume, 0.3 μM of the M13 oligo and the reverse
oligo, and 0.02 μM of the M13/forward concatenation oligo. In the
43
case of two primer amplification, both forward and reverse primers
were at 0.3 μM. The thermocycler program was 94°C for 3 min, 40
cycles of three steps: 94°C for 30 sec, 52°C for 30 sec, 72°C for 40
sec, and a final extension at 72°C for 5 min. The amplified products
were separated using a CEQ 8800 capillary electrophoresis system
(Beckman Coulter, Inc., Fullerton, CA) and scored using Genographer
(Benham, 1999).
In addition to DNA markers, the state of the Nud locus was
scored by visually assessing the adherence of the glumes to the
caryopsis.
Amplified Fragment Length Polymorphism Markers
Amplified fragment length polymorphism (AFLP®, Vos et al.,
1995) markers were employed essentially as in See et al (2002). One
hundred nanograms of total DNA from the parents, the RILs, and a
simulated heterozygote (i.e. a sample containing equal amounts of
both parental DNA) were digested in a 15 μL reaction volume for 3 h
at 37°C with three units of EcoRI and three units of HpaII. After
digestion, the reaction was terminated by incubation at 65° for 20
min. A 15 μL mix of ligating adapters was added to the digested DNA,
containing
44
HpaII adapters: 5’ GAC GAT GAG TCC TGA G, 150 ng
5’ CGC TCA GGA CTC AT, 132 ng
EcoRI adapters: 5’ CTC GTA GAC TGC GTA CC, 16.8 ng
5’ AAT TGG TAC GCA GTC TAC, 17.4 ng
in 1 X ligation buffer and 1 unit of T4 DNA ligase (New England
Biolabs, Inc., Ipswich, MA). This ligation mix was then incubated at
4°C for 24 h.
1 μL of ligation reaction product was preamplified in 30 μL of 0.1
mM each DNTP, 1 X PCR buffer (Promega), 2.5 mM MgCl2, 0.5 units
Taq DNA polymerase (Promega), and 30 ng of both EcoRI (5’ GAC TGC
GTA CCA ATT CA) and HpaII (5’ GAT GAG TCC TGA GCG GC)
preamplification primers. Pre-amplification conditions were as follows:
94°C for 2 min, 30 cycles of three steps: 94°C for 30 sec, 56°C for 1
min, 72°C for 1 min, and a final extension incubation at 72°C for 5
min. These amplification products were diluted 1:2 and frozen to be
used later as templates for selective amplification.
Selective amplification included 1 μL of the diluted
preamplification products in 15 μL of 0.1 mM each DNTP, 1 X PCR
buffer (Promega Corporation, Madison, WI), 2.5 mM MgCl2, 0.5 units
Taq DNA polymerase (Promega Corporation, Madison, WI), distilled
water, and 5 ng of fluorescently labelled EcoRI primer (5’ ACT GCG
45
TAC CAA TTC + 3 selective bases) and 30 ng of HpaII primer (5’ GAT
GAG TCC TGA GCG GC + 2 selective bases). Five EcoRI and six HpaII
primers were used in combination, for a total of thirty primer pairs
used (Table 3-1). The EcoRI primers were labeled with WellRED
fluorescent dyes (Sigma-Aldrich, St. Louis, MO) to enable AFLP®
product visualization on a CEQ 8800 automated capillary DNA
sequencer. AFLP® segregation was scored using Genographer version
1.6.0 (Benham, 2001).
Table 3-1. Selective primers used for AFLP® analysisa
P
Name
Sequence
Name
Sequence
e
5’GACTGCGTACCAATTCA
H
5’GATGAGTCCTGAGCGGC
e33
5’e + AG
h49
5’h + AG
e40
5’e + GC
h50
5’h + AT
e41
5’e + GG
h55
5’h + GA
e42
5’e + GT
h56
5’h + GC
e43
5’e + TA
h58
5’h + GT
h60
5’h + TC
a. Bold letters indicate selective nucleotides
46
Illumina GoldenGate Assay
Further genotyping of the population was performed with the
Illumina GoldenGateTM SNP genotyping platform. The GoldenGateTM
assay is a single-nucleotide extension assay making use of universal
primers and fiber-optic arrays (Shen et al, 2005). An oligonucleotide
pool assay (OPA) denotes the collection of all SNPs to be genotyped in
the assay. The OPA used for GoldenGateTM genotyping of the
Haxby/Baku population was PilotOPA1 (Close, 2006). As SNP allelic
state is detected using a 96-array configuration (Shen et al, 2005), a
subset of the original Haxby/Baku mapping population was chosen for
further characterization. Ninty-four RILs were chosen at random to be
genotyped along with Haxby and Baku. DNA was extracted from
individual two-week-old F7 plants using the DNeasy Plant Mini kit
(Qiagen Inc., Valencia, CA). DNA quantity and quality was evaluated
using a NanoDrop ND-1000 Spectrophotometer (NanoDrop
Technologies, Inc., Wilmington, DE). After quantitation, DNA was
diluted to 50 ng/uL with TE buffer (10 mM Tris, pH 7.5; 1 mM EDTA).
The SNP genotyping was performed by Dr. Tom Blake in the laboratory
of Dr. Shiaoman Chao in Fargo, North Dakota. The SNP assay was
performed as recommended by the manufacturer.
47
Map Construction and QTL Analysis – F5
Heterozygous data points were scored as missing data. JoinMap
3.0 was used for linkage grouping and map construction (Van Ooijen
and Voorrips, 2001). The complete data set was first analyzed to
separate linkage groups. After this initial analysis, anchor markers
were used to assign linkage groups to chromosomes.
After the initial mapping, data quality and consistency was
evaluated in the following manner. Within linkage groups, loci with
mean chi-square contributions greater than 3.0 were removed. Also,
genotype probabilities at loci were examined. For a specified map
order, JoinMap calculates the probability of each genotypic data point
for an individual, conditional on the genotypes of neighboring loci.
Loci in which the average of these probabilities was less than 0.05
were excluded.
After removing suspect loci based on the above
criteria, map orders and distances were recalculated. This process was
repeated until no further markers could be removed. Each marker
included in the map was tested for segregation distortion using the χ2
test implemented in JoinMap. Map distances were calculated as
Kosambi centiMorgans (Kosambi 1944).
QTL analysis was performed with the composite interval mapping
module of Windows QTL Cartographer Version 2.5 (Wang et al, 2006).
48
Cofactors to be used in the analysis were selected using forward and
backward regression. The walk speed was 2 cM and the window size
was 10 cM. The mean value of DMD from the four replications was
used for the analysis. 1000 permutations were used to estimate a
LOD score for which genome-wide significance was P=0.05 (Churchill
and Doerge 1994).
Results of map construction and QTL analysis were displayed
graphically using MapChart 2.1 (Voorrips 2002).
Map Construction and QTL Analysis– F7
During the SNP assay, 8 DNA samples were lost because of a
malfunction in the centrifuge. Therefore, data were available for 88
lines; 86 RILs, Haxby, and Baku. Before mapping, data were manually
checked for quality. The complete dataset contains the allelic state of
1,481 loci for all genotyped lines, regardless of polymorphism in the
population. First, loci missing data for more than eight individuals
were removed, leaving 1,205 high quality marker loci. Next, 825
obviously monomorphic markers were removed, leaving the final
dataset of 380 polymorphic loci and 86 RILs.
JoinMap 3.0 was used for linkage grouping and map construction
(Van Ooijen and Voorrips, 2001). Linkage groups were formed using a
49
maximum recombination percentage of 35 and a minimum LOD of 4.
The 380 markers were separated into linkage groups at a logarithm of
odds (LOD) score of 8.0. HarvEST:Barley, Version 1.55 (available at
http://harvest.ucr.edu/) contains consensus map positions for SNP loci
from PilotOPA1 that have been genetically mapped. This resource was
used to assign linkage groups to chromosomes. Based on
chromosome assignment, linkage groups from the same chromosome
were combined and marker order was recalculated. The stringent LOD
score of 8.0 used as the criteria for initial linkage grouping means that
the marker order within groups is highly self-consistent, therefore
these linkage group orders were used as fixed orders in the final map
calculation.
Data quality and consistency were evaluated as described for the
F5 population.
QTL analysis was performed as described for the F5 population
with the exception that window size for QTL analysis was 2 cM.
50
Results and Discussion
Phenotypic Data
The mean values and standard deviations for all measured traits
are summarized in Table 3-2. Haxby had higher DMD and smaller
mean particle size than Baku (46.4 versus 36.6 % and 1101 versus
1284 μm, respectively). Haxby had lower average 500-kernel weight
than Baku (18.77 g versus 20.09 g). All parental differences except
starch content and height were significant (P<0.03) over six random
independent replications. There was considerable variation for many
traits measured in the RILs; DMD, starch, 500-kernel weight, and
plant height all had minimum and maximum values greater than two
phenotypic standard deviations from the population mean. All traits
were normally distributed except particle size, which was skewed
strongly to the right. An inverse square transformation of the particle
size data resulted in a normal distribution and this transformed data
was used for QTL analysis.
51
Table 3-2. Average values (ave)of dry-matter digestibility (DMD), particle
size (PS), starch content (SC), 500-kernel weight (KW), height at maturity
(Ht), heading date (HD) and standard deviations (s) for ‘Haxby’ and ‘Baku’
and the RILs
‘Haxby’
‘Baku’
ave (s)a
ave (s)a
ave (s)
Min
Max
DMD (%)
46.4 (1.9)
36.6 (3.8)
41.2 (6.6)
18.9
55.2
PS (μm)
1101 (40)
1284 (53)
1183 (150)
945
1884
SC (%)
62.7 (6.2)
57.5 (2.8)
60.4 (5.2)
46.0
76.0
KW (g)
18.77 (0.97)
20.09 (0.78)
19.80 (2.42)
14.19
25.64
Ht (cm)
81 (4)
86 (5)
80 (8)
60
107
HD (day)
182 (1)
181 (1)
184 (3)
179
194
Trait
a
RILs
N=6
Significant phenotypic correlation existed between many of the
traits (Table 3-3). DMD was positively correlated with starch content
and average plant height, and negatively correlated with heading date.
As in previous studies of barley DMD (Lykos and Varga, 1995; Surber
et al, 2000; Rémond et al, 2004), particle size was negatively
correlated with DMD. Particle size was negatively correlated with
kernel weight and starch content, and positively correlated with
heading date.
52
Table 3-3. Pearson correlation coefficients among field measurements and
feed-quality characteristics of the Haxby/Baku RIL population
HD
Ht
KW
SC
DMD
HD
1
―
―
―
―
Ht
0.04
1
―
―
―
KW
-0.20*
1
―
―
SC
-0.01
0.34***
-0.05
1
―
DMD
-0.41****
0.30***
0.17
0.32***
1
PS
0.33***
*
-0.10
-0.15
indicates significance at P<0.05,
significance at P<0.0001
***
-0.43****
-0.23*
indicates significance at P<0.001,
-0.63****
****
indicates
Map Construction
Of 240 STS and SSR primer pairs tested, 97 revealed
polymorphism between ‘Haxby’ and ‘Baku.’ Thirty five of these,
chosen at random, were used to genotype the population and act as
anchor markers. An additional 183 markers were generated using 30
AFLP primer pairs. Two AFLP markers, e33h55(514/517) and
e42h60(475/484), were scored as co-dominant markers. In the initial
analysis of the complete segregation data, 11 linkage groups were
formed. Forty markers remained un-linked to linkage groups
containing anchor markers. The presence of previously mapped
markers in the linkage groups allowed the groups to be assigned to a
chromosome. In this way, a linkage map was constructed covering
53
680 Kosambi centiMorgans (Figure 3-1). This map length indicates
approximate genome coverage of between 49 and 64% (Costa et al,
2001; Qi et al, 1998). Chromosomes 6H and 7H each have regions
containing no markers: the two linkage groups comprising 6H are
separated by a recombination fraction of approximately 0.4, while the
two linkage groups comprising 7H are separated by a recombination
fraction of approximately 0.5. The estimate of genome coverage is
likely an underestimate as there are several regions of segregation
distortion in this genetic map, and segregation distortion at multiple
loci usually results in reduced apparent genetic distance (Zhu et al,
2007). These markers that show distortion occur in blocks in this
map. Sixty-three markers showed segregation distortion at the level of
P ≤ 0.05. Thirty-two markers showed segregation distortion at the
level of P ≤ 0.01 (Figure 3-1). If it is assumed that intervals between
distorted markers are also distorted, 5% (35.3 cM) of this map is
distorted at the level of P ≤ 0.01, and 14% (96.0 cM) is distorted at
the level of P ≤ 0.05. Distorted markers were included in the QTL
analysis.
54
Figure 3-1. Linkage map of Haxby/Baku RIL population. Distances are in
Kosambi centimorgans. Anchor markers are underlined. Blue intervals are
distorted (P≤0.05) in favor of Baku alleles, yellow toward Haxby.
55
Figure 3-1. continued.
QTL Analysis
Using simple Composite Interval Mapping, 2 putative QTL of
significance greater than LOD 2.8 were detected for ruminal drymatter digestibility (Table 3-4). These loci also reached the critical
value by single-marker analysis and simple interval mapping. Twoway analysis of variance detects no epistasis between these loci.
These two QTL were on chromosomes 6H and 7H and explained 19%
and 17% of phenotypic variation, respectively. Total phenotypic
variation explained by the QTL model was 43%. The additive effect of
56
a Haxby allele at the 6H locus is 3.3 percentage units, while the
additive effect of a Haxby allele at the 7H locus is -2.8 percentage
units. The LOD peak of the 6H locus is at 65 cM in the present map,
nearest to the SSR marker Bmag0009, and the 2-LOD support interval
is 5 cM in length. The LOD peak of the 7H locus is at 16.5 cM, nearest
the AFLP marker e43h49(479). The 2-LOD support interval is 10 cM in
length, although the portion of the LOD profile above the critical value
of 2.8 is very broad, extending from 5 to 28 cM on the present map.
Table 3-4. Locations, LOD scores, and effects of QTL detected by simple
Composite Interval Mapping in the Haxby/Baku RIL population
Trait
Positiona
Intervalb
LOD
Effectc
R2 d
DMD (%)
6H(64.7)
64.2-69.4
7.39
3.33
19
7H(16.5)
14.8-25.4
6.63
-2.83
17
5H(4.2)
0-15.4
6.01
96
15
6H(33.1)
25-41.4
2.80
80
7
6H(64.7)
58.2-70.4
2.83
-64
7
7H(66.3)
55.2-71.8
4.34
-80
11
PSf (μm)
a
TR2 e
43
45
Chromosome and cM position of maximum LOD score, b2-LOD support interval,
average effect of one Haxby allele, dPercent phenotypic variation explained by the
indicated QTL, ePercent phenotypic variation explained by QTL model, fParticle Size
c
57
Although only two regions of the genome exceeded the critical
value of 2.8 for DMD, a region extending from 42 to 54 on 6H is near
the critical value, reaching a maximum LOD of 2.5 and a maximum
additive effect of -2.0 (Figure 3-2). This locus is suggestive of a QTL
for DMD.
6H(6)
7
6
84
5
e40h55(81)
4
75
78
3
e43h58(89)
e40h55(83)
2
64
66
1
e40h58(276)
e40h58(139) e40h58(270.5)
0
11
14
16
21
22
23
28
31
32
38
40
44
45
46
47
52
54
55
57
qDMD-7H
Bmac0577B
e40h60(257)
GBMS230A e43h49(479)
e33h50(339)
e42h60(115.5)
e33h50(97) e33h50(340)
Bmag0507
e40h58(110)
Nud
e40h50(306) e42h50(307)
e40h50(234) e43h56(111)
e33h56(103)
Bmag0120
e40h49(139)
e40h56(139)
e33h56(192)
e40h56(384.5)
e42h60(475/484) e33h60(206)
e40h56(381)
-1
0
3
5
-2
77
80
e43h49(293)
e33h50(335)
e43h58(76)
-3
e40h58(157.5)
e40h50(160) e40h58(159.5)
8
65
66
69
71
7
Bmag0009
Bmac0018 GMS6
e43h49(406)
e40h55(475)
6
53
54
5
47
e40h60(132)
e42h60(133)
4
e40h58(126) e40h50(126)
3
37
39
42
2
31
e42h55(116)
e42h50(434)
Bmag0500
1
26
qDMD-6H
Bmac0316
e43h50(415.5)
0
20
22
-1
0
e43h56(257) e43h55(427)
e43h55(415)
e33h55(103)
-2
e43h55(582)
7H(1)
100
Figure 3-2. Scans of barley chromosomes 6H and 7H for trait DMD. Dashed
line indicates allelic effect at loci. Solid line indicates LOD scores at loci.
Dotted line is LOD of 2.8. Bars and lines to right of chromosomes are 1- and
2-LOD support intervals, respectively.
58
Previously, QTL for DMD have been identified on barley
chromosomes 1H, 3H, and 4H in the Steptoe/Morex population
(Bowman et al, 1996), 1H in the Lewis/Baronesse population (AbdelHaleem, 2004), and 2H in the Valier/PI 370970 population (ibid.).
Therefore, the putative QTL detected in the Haxby/Baku population
seem to be previously unidentified.
Four QTL were detected for mean particle-size (Table 3-4). In
two QTL, the Haxby allele increases mean particle-size. One QTL in
which the Haxby allele decreases particle size is coincident with the 6H
QTL for DMD in which the Haxby allele increases DMD, thus it is likely
that this locus impacts DMD indirectly by its effects on particle size.
Although the DMD 7H locus does not reach the critical value of 2.8 by
composite interval mapping, by F-test the interval e43h58(76)Bmag0507 is significant at P<0.02, and the Haxby allele that
decreases DMD is also associated with increased particle size. Fortyfive percent of total phenotypic variation was explained by the four
detected QTL. No epistasis was detected between these loci.
QTL for particle size have previously been identified on 2H, 3H,
4H, and 7H in the Steptoe/Morex map, 5H and 7H in the
Lewis/Baronesse map, and 2H and 7H in the Valier/PI 370970 map.
59
The QTL on 7H in the Valier/PI 370970 is towards the short arm
relative to the nud locus, while in the Haxby/Baku map it is towards
the long arm. In the Steptoe/Morex map the 7H QTL is again on the
short-arm. In the Lewis/Baronesse map there are limited points of
reference so it is difficult to make any comparisons.
Further Analysis of the Haxby/Baku Population
Two QTL were identified in the initial mapping of this
population. They are on chromosomes 6H and 7H and are estimated
to explain 19 and 17 percent of phenotypic variation, respectively.
Because these loci are approximately equal in additive effect but
opposite in sign, we presume that there must be more loci conferring
the extremely low DMD to Baku. Further, as the map used to detect
these QTL is estimated to cover 50-60% of the barley genome, it is
quite possible that important DMD loci remain undetected in this
population. With the recent development of an extremely rapid and
reliable single-nucleotide polymorphism (SNP) genotyping method in
barley (Rostoks et al, 2006), it has become feasible to genotype up to
1,536 SNPs in a single experiment. Due to the apparently limited
coverage of the initial mapping, a subset of the Haxby/Baku mapping
population was genotyped using this new system, Illumina
60
GoldenGate, in an attempt to extend genome coverage and identify
more loci with important effects on DMD.
Map Construction:SNPs
At a LOD threshold of 8.0, the SNP markers were grouped into
26 linkage groups. The 26 linkage groups were assigned to
chromosomes based on chromosomal assignment of markers in
HarvEST:Barley, Version 1.55 (available at http://harvest.ucr.edu/).
On further inspection, one linkage group was comprised of SNP
markers previously mapped to chromosomes 2H and 5H. This group
was split at a LOD of 9.0. The 27 linkage groups were combined into
their respective chromosomes at a LOD of from 2.0 to 7.0. Several
properties of the resulting genetic maps are interesting to note and are
summarized in Figure 3-3.
Gross marker order is in most cases consistent with the HarvEST
consensus map (Figure 3-3). The most obvious areas of marker order
discrepancy are located in regions of high marker density in either of
the maps (for example: 1H:130 cM, 2H:147 cM). Differences in
marker order between the two maps are probably the result of
sampling error or lack of recombination information.
61
Comparison to the HarvEST consensus map allows direct
estimation of genome coverage. Genomic coverage is apparently quite
complete with several exceptions (Figure 3-3). Approximately
eighteen cM of the extreme short arm of chromosome 2H contains no
markers. Also, twenty cM of the extreme short arm of chromosome
5H is not covered by markers. For the purposes of this mapping
project, it is unfortunate that the 5HS region contains no markers. It
has been reported that the genes encoding the grain softness proteins
that reside in this area, the barley hordoindolines and the wheat
puroindolines, are important contributors to DMD (Beecher et al, 2002;
Swan et al, 2006). This problem could easily be rectified by genetic
mapping of either of the hordoindolines or of GSP Seventeen
centiMorgans of the short arm of 6H contains no markers. A final area
of incomplete genome coverage is the short arm of 7H, where 29
centiMorgans in the Haxby/Baku map contains no markers, though this
region in the consensus map is only 6.7 centiMorgans in length.
62
1H-HarvEST
1H-Haxby/Baku
2H-HarvEST
0.0
Tel5P
0.0
ABC13023-1-10-344
5.8
ABC13023-1-10-344
6.1
8.2
4943-571
3822-1180
13.1
2609-350
18.3
ABC05684-pHv2534-05
23.5
10922-503
23.5
ABC01004-sfp18-05
29.6
30.1
6792-1945
5318-436
29.2
1865-396
42.5
7144-973
50.3
7623-818
57.1
60.3
ConsensusGBS0524-2
ABC12560-1-1-421
74.8
77.2
2284-1738
4956-1444
82.7
6600-453
90.1
3469-1152
12.4
4943-571
26.8
10922-503
34.5
5318-436
55.9
2510-1464
63.7
2577-1122
68.0
1968-1263
78.4
4665-882
88.3
3204-811
93.5
95.8
ABC06571-1-2-356
3752-175
101.5
2711-234
110.7
112.8
116.1
2121-1519
5690-1045
7389-555
129.0
129.7
131.0
ConsensusGBS0361-5
5878-1810
472-1376
139.1
141.9
142.3
ConsensusGBS0554-4
4057-2114
1078-170
47.1
48.2
50.8
51.1
51.2
51.3
53.6
53.8
54.3
54.8
55.3
63.6
66.4
73.8
73.9
87.0
88.5
91.6
92.7
93.4
93.8
101.9
103.2
106.0
111.3
117.8
2510-1464
3561-892
1749-948
3710-852
2141-2425
ABC12550-1-3-276
ABC15349-1-1-162
ABC07427-1-1-329
3689-1101
7800-594 5402-929
2577-1122
3845-1089
1968-1263
4665-882
1770-1477
3204-811
ABC06571-1-2-356
3752-175
3201-603
9279-368
2935-1634
2711-234
1497-628
2121-1519
5690-1045
7389-555
131.1
132.2
133.0
134.1
134.9
136.0
141.6
146.6
147.7
472-1376
5878-1810
ConsensusGBS0361-5
4027-1814
4978-1030
199-393
1217-546
ConsensusGBS0554-4
4057-2114
1.0
Tel2S
106.2
107.0
682-767
ABC13569-1-1-107
114.7
116.3
5088-59
3763-595
128.8
3271-1422
133.2
2822-739
148.9
152.4
154.4
155.4
160.9
3536-89
7808-727
570-1376
1283-332
1344-930
165.3
ABC16814-1-3-297
2H-Haxby/Baku
0.0
0.7
14.3
20.2
29.7
36.0
51.1
51.4
52.5
54.0
55.4
55.7
56.0
56.8
57.0
57.1
57.6
59.5
59.9
66.3
71.0
71.9
80.3
84.7
85.3
88.9
90.9
91.5
92.7
94.3
94.4
94.7
97.3
103.9
104.4
105.4
105.8
106.5
112.6
113.1
113.2
117.7
119.2
119.6
119.8
128.7
128.9
129.2
129.8
131.1
146.9
148.1
149.3
153.9
ABC05684-pHv2534-05
3184-791
ABC01004-sfp18-05
1865-396
7144-973
6338-682
2128-874
ConsensusGBS0524-2
5113-624
3122-909
796-1148
4717-386
ABC17685-1-4-365 5233-1070
2719-672 ABC05800-1-1-142
7489-442
ConsensusGBS0008-1 ABC06091-1-1-187
ABC11345-1-2-363
ABC03181-1-1-164
3806-486
ABC12560-1-1-421
7735-657
2284-1738
4956-1444
4944-2118
6600-453
2017-635 3469-1152
2371-950
2020-539
4419-1392
1635-691
ABC11853-1-2-343
5347-1010
ABC10472-1-2-247
ABC14531-1-2-91 682-767
ABC13569-1-1-107
3763-595
4833-420
3000-1074
5161-1809
7576-818
3271-1422
ConsensusGBS0379-1
1207-1186
2688-1022
8501-449
ABC05640-1-1-248
868-675
6157-1233
ABC07356-1-1-109
252-556
3292-418
ABC16528-pHv407 3536-89
570-1376
1283-332
7808-727
1344-930
Figure 3-3. Comparison of Haxby/Baku RIL population and HarvEST:Barley
consensus linkage maps
63
Figure 3-3. Continued. All SNPs mapped in the Haxby/Baku population are
shown. For clarity, only a subset of consensus SNPs are shown.
3H-HarvEST
0.0
5.1
6.0
9.0
12.9
3H-Haxby/Baku
ABG070
Tel3S
ConsensusGBS0194-1
918-928
7458-1163
0.0
3.3
0.0
4.8
7.3
8.7
9.5
13.8
13.9
15.1
18.7
18.8
25.8
26.4
26.5
26.6
62.3
1746-1527
70.9
1827-958
92.0
2236-773
117.5
118.5
ABC13753-1-2-167
76-1059
130.0
135.4
137.6
138.6
6402-691
2847-485
ABC05919-1-2-157
ABC13678-1-2-369
165.8
166.6
168.3
2346-318
ConsensusGBS0632-3
265-1229
4H-HarvEST
26.8
30.1
30.9
31.4
34.8
35.0
35.3
36.1
36.4
40.5
41.5
44.5
48.8
50.6
51.5
57.4
58.1
59.4
60.2
67.5
70.7
71.2
71.8
72.9
73.1
73.6
ConsensusGBS0194-1
7458-1163 918-928
6299-529
3906-558
3886-313
1074-992
972-505
3177-1362
6530-670
8387-187
1746-1527
5038-1035 ABC10667-1-1-288
3354-121 ABC21245-1-2-275
ABC04214-1-2-360 1630-1150
ABC04826-1-1-174
ConsensusGBS0471-1 ABC13089-1-2-478
ABC08184-2-1-35
ABC18717-1-3-215 6364-645
ABC19175-1-2-375
1827-958
ABC13107-1-3-299
5960-1302
4647-248
2315-702
ConsensusGBS0284-1
963-386 2236-773
1898-580
ABC10632-1-4-309
6402-691
ABC08260-1-1-108
6069-304
11609-524
3808-1763 76-1059
ABC13753-1-2-167
ABC13678-1-2-369
ConsensusGBS0038-2
ABC38781-pHv2346-01
ABC36454-pHv2499-01
4643-867
7169-713
4737-368
ConsensusGBS0632-3
4H-Haxby/Baku
MWG634
1996-652
22.8
22.9
1513-514
2065-3135
31.0
1094-801
38.1
2670-1431
42.9
1230-523
52.2
3917-1365
0.0
9.7
12.8
25.1
32.9
36.0
36.6
37.8
38.0
41.5
41.8
42.1
42.9
43.0
44.4
44.5
44.7
45.2
80.4
82.5
ConsensusGBS0666-1
ABC20090-1-1-275
97.4
98.5
41-695
3652-872
108.2
ABC09877-1-1-108
118.9
2297-1250
129.9
130.7
2878-574
954-1377
138.3
Tel4M
45.4
45.6
45.8
47.9
48.4
52.1
53.5
53.7
54.3
54.8
63.6
63.8
81.2
81.8
85.4
86.5
90.7
90.9
95.8
95.9
106.1
106.2
1996-652
1513-514
2065-3135
1094-801
2670-1431
1230-523
1180-70
ABC09216-1-4-392
5726-414
424-423
6464-1115
1194-234 ABC09432-1-1-160
ConsensusGBS0010-2
ABC09662-1-3-352 4139-888
1169-944
3042-1225
3489-854
3917-1365 5889-1154
ABC10254-1-2-250
3716-910
ABC08788-1-1-329
3644-1483
ABC05369-1-3-231
ConsensusGBS0589-1
1375-2534
ConsensusGBS0461-3
9149-1316
3704-1947
ConsensusGBS0666-1
ABC20090-1-1-275
5245-304
41-695
3652-872
ABC09877-1-1-108
1561-1053
4160-1365
4773-1009
10956-366
2297-1250
954-1377
2878-574
64
Figure 3-3. Continued.
5H-HarvEST
0.0
5H-Haxby/Baku
Gsp
20.6
ABC01741-1-4-299
31.1
35.1
421-528
6672-803
49.9
51.2
52.8
ABC10045-1-1-164
ABC17741-1-1-203
ABC11529-1-1-295
63.7
8107-154
78.1
82.5
83.3
89.7
89.8
ABC06144-pHv86-02
370-443
ABC11984-1-2-158
3685-894
ConsensusGBS0304-1
96.9
ABC14689-1-9-399
105.0
105.9
3641-828
ABC11221-1-3-410
122.6
124.4
126.1
129.0
ConsensusGBS0531-1
ConsensusGBS0234-1
211-1181
1697-636
142.5
145.6
2395-2083
3477-1774
157.6
158.2
160.7
161.7
168.4
7337-388
5163-896
2617-1234
6803-442
ABC09278-1-4-69
173.7
5413-2541
181.4
5145-1355
191.3
194.5
2978-938
Tel7L
6H-HarvEST
0.0
0.0
8.0
12.1
13.8
14.8
14.9
16.7
16.9
17.0
19.5
27.1
27.2
30.1
34.3
36.5
38.2
42.0
42.4
53.4
54.4
55.2
57.9
58.8
64.8
65.7
71.3
71.5
72.6
74.7
77.1
85.1
85.4
88.0
93.2
95.6
97.4
97.5
97.8
98.0
100.0
101.4
115.9
120.2
120.3
123.8
126.4
126.8
127.2
132.1
ABC01741-1-4-299
421-528
6672-803
2267-1173
ABC05926-1-1-51
ConsensusGBS0527-5
ABC07010-1-2-150
ABC09365-1-3-378
ABC10045-1-1-164 ABC17741-1-1-203
ABC11529-1-1-295
ABC11984-1-2-158
370-443
8107-154
3685-894
ConsensusGBS0304-1
ABC14689-1-9-399
ABC11221-1-3-410
3641-828
ConsensusGBS0531-1
ConsensusGBS0234-1
211-1181
ConsensusGBS0712-1
ABC17073-1-1-298
1697-636
ABC04322-1-3-208
6450-755 ConsensusGBS0704-2
ABC03113-1-1-251
6050-1625
2395-2083
ABC04352-pHv108-01
3883-616
3477-1774
2746-1501
ConsensusGBS0451-1
7337-388
3720-52 3759-1385
7342-535
2223-1688
5163-896
2617-1234
6803-442
ABC09278-1-4-69
ConsensusGBS0390-11 5413-2541
6735-754
2258-980
5145-1355
6736-452
3007-1337
2978-938
Tel6S
16.9
ConsensusGBS0346-1
24.4
1066-2110
30.8
6719-1166
6H-Haxby/Baku
0.0
2.1
2.3
8.5
9.6
15.9
16.2
19.2
20.2
21.4
21.8
22.1
22.5
22.7
52.9
3580-331
66.4
2298-1526
83.1
ConsensusGBS0239-5
100.6
8504-785
130.3
130.6
135.7
136.0
1852-509
1473-1115
2377-522
ABC05241-1-5-271
23.2
23.7
24.0
25.2
25.5
27.4
28.7
28.9
29.0
30.2
31.7
32.6
32.7
32.9
33.4
39.6
39.7
40.1
40.5
42.9
50.6
50.8
51.4
ConsensusGBS0346-1
1066-2110
1628-410
885-104
4611-178
3580-331
1588-537
2176-891
5909-859
ABC18140-1-1-43
ABC10536-1-2-232
ABC08769-1-1-205
ABC00917-1-1-70
828-545 2559-1796
ABC06493-1-1-45 4071-1565
210-450
ABC13045-1-1-226
ABC14849-1-1-533
3596-1091
4313-482
2298-1526
ABC05969-1-1-360
4077-76
1140-1508
1565-514 ABC13717-1-1-328
3436-354
4146-1154
1899-739
ABC10265-sfp25-01
2047-850
ConsensusGBS0239-5
ABC08038-1-3-160
1689-919
1473-1115
1852-509
2377-522
5124-1707
10425-725
8504-785
65
Figure 3-3. Continued.
7H-HarvEST
7H-Haxby/Baku
0.5
0.9
1.1
2.5
5.0
12.7
4644-1363
8444-948
7172-1536
2811-81
984-583
6517-602
29.2
32.0
2124-984
ABC03024-1-3-368
60.6
4475-340
73.6
67.2
3186-1560
74.8
74.9
79.1
79.2
79.7
79.9
80.0
0.0
2.8
3.8
5.4
6.2
33.7
38.1
52.2
53.0
54.9
57.4
60.9
63.9
64.1
66.1
81.2
102.4
3900-611
122.2
122.5
125.0
129.4
13008-352
5764-430
8469-1036
6433-124
147.2
148.3
1847-1745
ABC03674-1-1-226
81.3
88.1
88.7
92.7
93.1
93.7
107.1
118.6
121.7
121.8
124.2
128.6
129.7
130.7
136.2
137.5
144.8
150.3
7172-1536
4644-1363
8444-948
2811-81
984-583
6517-602
1073-916
8142-676
2124-984
ABC03024-1-3-368
2781-821
4767-1374
4475-340
1333-554
ABC10361-1-5-380
9820-455 1735-1424
3186-1560 ABC04803-1-1-392
ABC11018-1-1-216
779-2258
2429-1929 ABC10546-1-2-488
2251-643
ABC28974-pHv78-02
7712-674
2924-1189 4779-1563
2060-422
3372-751 1997-77
2792-749 2015-562
ABC10040-1-1-238 1212-890
486-1812
ABC17088-1-1-247 478-1291
ConsensusGBS0250-2 3731-103
3232-201
Nud
ABC14397-1-2-208
3089-1605
3334-2072
2174-1211 ABC17091-1-2-109
3900-611
5764-430
8469-1036
13008-352
4671-856
382-2624
6433-124
ConsensusGBS0084-1
ABC11252-1-2-254
710-1699
3818-1123
1847-1745
Despite apparently complete coverage of the genome (based on
comparison with the consensus maps obtained in HarvEST), two of the
chromosomal maps are of dramatically shorter centiMorgan length
(Figure 3-3). 3H and 6H are only 46% and 43% of the centiMorgan
length of the consensus maps, respectively (Table 3-5). From the
standpoint of QTL mapping, this is unfortunate because any QTL
mapped on these short chromosomes will likely be mapped to larger
66
intervals due to less recombination information, and the effects of
linked genes may not be separated (Noor et al, 2001).
Table 3-5. Comparison of HarvEST consensus and Haxby/Baku map lengths.
Map length (cM)
%a
Chromosome
HarvEST
Haxby/Baku
1H
136.1
147.7
109
2H
142.6
153.9
108
3H
160.6
73.6
46
4H
127.4
106.2
83
5H
170.7
132.1
77
6H
119.1
51.4
43
7H
147.2
150.2
102
all
1003.7
815.1
81
a
Haxby/Baku chromosomal map lengths expressed as % of HarvEST chromosomal
map lengths
An appealing hypothesis to explain the reduction in map length
is selection against heterozygotes (or selection for a homozygous
class) in early generations. This population is an F7 recombinant
inbred population. Calculation of map length in F7 recombinant
inbreds assumes that residual heterozygosity beyond the first
generation allows increased opportunity for informative recombination.
JoinMap corrects for this expected increase in observed pair-wise
67
recombination by reducing recombination frequencies used in the
calculation of map length (Stam, 1993). Therefore, selection resulting
in less than expected heterozygosity would reduce calculated map
length. However, because informative recombination frequency in an
RIL is expected to be twice that of a single-meiosis population,
reduction in map length of greater than 50% cannot be explained even
by complete lack of heterozygotes in early generations. Also, selection
for a homozygous class (such as could be used to help explain map
compression) would be expected to result in non-random segregation
of alleles, i.e. segregation distortion. Though chromosome 3H does
show extensive and significant segregation distortion, the chromosome
that shows the greatest map compression, 6H, shows very little
segregation distortion (Figure 3-4). For completeness, all
chromosomes are depicted in Figure 3-4.
68
Haxby allele frequency (%)
80
70
60
50
40
30
20
0
25
50
75
100
125
150
Map distance (cM)
Chromosome 1H
Haxby allele frequency (%)
80
70
60
50
40
30
20
0
25
Chromosome 2H
50
75
100
125
150
Map distance (cM)
Figure 3-4. Allele frequency in the Haxby/Baku mapping population. Dashed
line indicates different from expectation at P=0.05.
69
Figure 3-4. Continued.
Haxby allele frequency (%)
80
70
60
50
40
30
20
0
10
20
30
40
50
60
70
Map distance (cM)
Chromosome 3H
Haxby allele frequency (cM)
80
70
60
50
40
30
20
0
Chromosome 4H
25
50
Map distance (cM)
75
100
70
Figure 3-4. Continued.
Haxby allele frequency (%)
80
70
60
50
40
30
20
0
25
Chromosome 5H
50
75
100
Map distance (cM)
Haxby allele frequency (%)
80
70
60
50
40
30
20
0
Chromosome 6H.
20
Map distance (cM)
40
125
71
Figure 3-4. Continued.
Haxby allele frequency (%)
80
70
60
50
40
30
20
0
25
Chromosome 7H
50
75
100
125
150
Map distance (cM)
Although the cause of the reduced Haxby/Baku map length
relative to the consensus map is unclear, differences in meiotic
recombination frequency and therefore map length have been
observed by other investigators, and some of this variability in
recombination frequency has been attributed to genetic effects rather
than sampling or environmental effects (Cornu et al, 1989; Hadad et
al, 2006). Further, Barth et al (2001) used fourteen mapped antibiotic
resistance gene insertions to estimate recombination frequencies in
Arabidopsis and found that recombination frequencies differed
depending both on the parental ecotypes used in the cross and on the
chromosome or chromosomal region being evaluated.
72
Comparison of the F5 and F7 maps
Comparisons of F5 and F7 maps length are summarized in Table
3-6. Based on map length, the AFLP mapping of the full F5 population
covered approximately 83% of the F7 SNP genetic map length. The
discrepancy between the map lengths of 3H and 6H between
generations is likely due to AFLP marker error. The main effect of
genotyping errors on genetic mapping is to inflate the map length
(Hackett and Broadfoot, 2003). Based on duplicate genotyping of
samples, the error rate of AFLP genotype calls is estimated to be
between 2 and 5% (Bonin et al, 2004). The estimated error rate of
GoldenGate genotype calls is estimated to be as low as 0.01% (Fradin
and Bougneres, 2007).
73
Table 3-6. Comparison of F5 and F7 Haxby/Baku map lengths.
Map length (cM)
%a
Chromosome
a
F7
F5
1H
147.7
110.4
75
2H
153.9
132.4
86
3H
73.6
99.6
135
4H
106.2
72.2
68
5H
132.1
73.1
55
6H
51.4
99.5
194
7H
150.2
91.2
61
all
815.1
678.4
83
F5 chromosomal map lengths expressed as % of F7 chromosomal map lengths
QTL Analysis: SNPs
A number of phenotypic traits were measured in the population.
All QTL detected will be presented (Table 3-7), but only those QTL for
DMD and PS will be discussed in detail.
A QTL for particle size was detected on the short arm of
chromosome 5H (Table 3-7). This QTL explained fourteen percent of
phenotypic variation and, on average, increased particle size by
seventy micrometers. This QTL for particle size has been detected
previously. In the F5-based mapping population, four QTL for particle
size were detected, and the QTL with the greatest R2 was located on
the short arm of 5H centered approximately at the AFLP marker
74
e33h55(382). By integrating markers from the AFLP map into this
map, it can be shown that the two QTL are coincident (Figure 3-6). In
Figure 4-3, the integrated map is used as a bridge between the F5 and
F7 maps rather than simply using an integrated map for QTL analysis
because broad-scale integration of the two maps resulted in
unacceptable changes in SNP marker order, changes in distance
between markers, and exclusion of SNP markers (results not shown).
By composite-interval mapping, four QTL explaining between 11
and 14 % of DMD phenotypic variation were detected on chromosomes
1H, 6H, and 7H (Table 3-7). The 1H QTL was not previously detected,
though this is not due to incomplete coverage in this region (Figure 35), it may be that the different QTL detection algorithm is the cause.
The 6H QTL and the 7H QTL at 66 cM were detected in the F5 mapping
population (Figure 3-5). Because three of the four QTL detected are
QTL in which Haxby decreases DMD, it is concluded that the very low
DMD of Baku is due to the combined effects of several minor genes
rather than the effects of a small number of genes with major effects.
75
Table 3-7. Locations, LOD scores, and effects of QTL detected by simple
Composite Interval Mapping in the Haxby/Baku RIL population
Positiona
Intervalb
LOD
Effectc
R2d
TR2e
6H(37.5)
33.3-41.7
5.2
2.6
22
22
2H(97.4)
94.7-101.5
3.1
0.7
11
7H(72.2)
67.3-74.3
8.9
-1.5
27
7H(88.2)
85.1-88.5
16.4
2.3
57
2H(22.6)
14.1-29.2
3.1
2.4
7
2H(56.9)
55.4-58.1
11.4
4.9
29
7H(73.8)
67.4-79.0
4.5
-3.0
10
1H(88.5)
79.0-91.3
5.7
-1.4
11
1H(145.6)
142.9-end
7.4
-2.0
24
2H(63.9)
59.7-70.3
3.8
1.2
12
7H(52.2)
43.4-53.4
7.4
1.7
26
54
Particle Size (μm)
5H(17.1)
15.0-19.3
4.3
65.0
14
14
DMD (%)
1H(47.1)
39.1-48.1
3.4
-3.0
13
6H(23.0)
20.2-25.3
3.6
2.7
11
7H(66.4)
64.1-72.5
3.3
-3.0
11
7H(85.5)
81.5-88.6
3.5
-3.4
14
Trait
Starch (%)
f
Kernel Wt. (g)
Height (cm)
Head Date (day)
a
65
62
48
Chromosome and cM position of maximum LOD score, b2-LOD support interval,
average effect of one Haxby allele, dPercent phenotypic variation explained by the
indicated QTL, ePercent phenotypic variation explained by QTL model, fweight of 500
kernels
c
76
0.0
ABC01741-1-4-299
8.0
421-528
5HS-F5
0.0
ABC01741-1-4-299
8.7
9.8
e42h50(230)
421-528
6672-803
Bmac0096
e33h55(382)
12.1
6672-803
16.7
ABC07010-1-2-150
14.5
16.0
18.2
19.5
ABC11529-1-1-295
22.6
ABC11529-1-1-295
27.4
Bmag0337
30.5
370-443
370-443
27.2
1H-F7
13.1
DMD
5HS-integrated
1H-integrated
2609-350
Bmac0213
23.5
10922-503
24.4
10922-503
29.6
30.1
6792-1945
5318-436
30.4
30.9
6792-1945
5318-436
47.1
48.2
50.8
51.1
51.2
2510-1464
3561-892
1749-948
3710-852
2141-2425
46.8
48.4
49.4
51.9
Bmag0876
2510-1464
3561-892
1749-948
0.0
e42h50(230)
8.6
Bmac0096
13.9
e33h55(382)
18.1
Bmag0337
1H-F5
4.6
18.3
PS
PS
5HS-F7
e43h58(363)
10.7
Bmac0213
34.4
37.1
38.7
40.4
e42h49(393)
Bmag0876
e33h56(424)
e42h55(190)
Figure 3-5. Comparisons of QTL for mean particle size (PS) and DMD
detected in the 123-member F5 population and in the 86-member F7
population. Marker positions (cM) are indicated to the left of the maps. QTL
regions are to the left of the F5 and F7 genetic maps and the 1-LOD QTL
support interval is indicated by shaded bars, the 2-LOD interval is indicated
by lines.
77
6H-integrated
e40h60(132)
885-104
8.5
885-104
6.9
8.3
15.9
3580-331
15.8
3580-331
22.1
22.9
23.2
23.7
24.5
25.3
27.3
28.4
ABC18140-1-1-43
GMS6
828-545
Bmac0018
3596-1091
Bmag0009
e40h55(475)
ABC05969-1-1-360
ABC18140-1-1-43
828-545
3596-1091
27.4
ABC05969-1-1-360
4767-1374
60.4
61.9
e43h49(293)
4767-1374
66.1
ABC10361-1-5-380
66.2
ABC10361-1-5-380
69.3
Bmac0577B
73.6
9820-455 1735-1424
3186-1560 ABC04803-1-1-392
73.8
74.8
ABC04803-1-1-392
GBMS230A
81.5
3232-201
88.4
89.0
Nud
ABC14397-1-2-208
3232-201
88.1
88.7
Nud
ABC14397-1-2-208
e40h60(132)
64.7
65.7
66.3
Bmag0009
Bmac0018
GMS6
70.6
e40h55(475)
7H-integrated
60.9
81.3
53.5
7H-F5
DMD
21.4
22.7
24.0
7H-F7
DMD
6H-F5
DMD
DMD
Figure 3-5. Continued.
6H-F7
0.0
e43h49(293)
11.0
Bmac0577B
15.8
GBMS230A
32.3
Nud
One possible candidate gene of the 7H QTL for DMD (qDMD-7H)
at 66 cM is that it is simply the Nud locus. Covered barley grains may
be expected to disappear more slowly in the rumen due to a protective
effect of the hull. While the Haxby allele of the QTL reduces DMD and
78
Haxby grain is covered, which is consistent with the hypothesis that
qDMD-7H is Nud, I find it unlikely for several reasons. By singlemarker QTL analysis, the most significant marker identified as
impacting DMD is ABC10361-1-5-380. This marker is more than
twenty centiMorgans away from the Nud locus (Figure 3-5). Further,
by composite-interval mapping, the Nud locus rises above the
significance threshold, but the LOD profile between the two peaks
decreases to zero. In other words, both QTL (qDMD-7H and Nud)
seem to have been detected as individual loci.
Further evidence exists to counter the claim that qDMD-7H is
Nud in the form of consecutive QTL mapped for kernel weight. Two
QTL were detected for kernel weight on 7H at 72 and 88 cM (Figure 36). The peak of the 88 cM kernel weight QTL LOD profile is at the
precise position of Nud and the Haxby allele increases kernel weight,
which is to be expected. The Haxby allele at 72 cM decreases average
kernel weight and is coincident with the QTL in which the Haxby allele
decreases average DMD. It seems that two loci impacting both DMD
and kernel weight exist within twenty centiMorgans of each other.
Both Haxby alleles at both QTL on 7H reduced DMD, but Haxby Nud
allele increased kernel weight, while the Haxby allele at ~ 70 cM
decreased kernel weight. The reduction in kernel weight by qDMD-7H
79
may not deleterious to animal performance as long as it does not
reduce test-weight. Barley test-weight has been shown to be an
important indicator of animal performance (Mathison et al, 1991;
Grimson et al, 1987). These plant materials will be grown again this
year at the Post Farm in Bozeman, and test-weight will be measured.
Kernel Weight
18
DMD
16
14
LOD
12
10
8
6
4
2
0
0
10
20
30
40
50
60
70
80
90 100 110 120 130 140 150
Map distance (cM)
Figure 3-6. 7H LOD scores for kernel weight and for DMD. LOD threshold for
significance is shown as a dotted line at LOD 2.9. Nud is at 88 cM.
It is interesting to note that a significant QTL explaining 10 % of
height variation was also detected at approximately 70 cM on
chromosome 7H (Table 3-7, Figure 3-7). It may be that a
developmental gene or gene complex with pleiotropic effects on DMD,
80
kernel weight, and height exists in this region of 7H. The homologous
region in rice consists of nearly four Megabases of DNA and nearly 600
predicted genes, therefore identification of candidate genes by
comparison to the rice genome is daunting if not impossible, especially
given that the physiological basis of this locus (or loci) is unknown.
Haxby/Baku-7H
Os-Chromosome6
1-ABC10361-1-5-380
DMD
1-Os06g0178600
2.9
7.4
Ht
KW
0.8
0.1
0.1
1.2
0.1
5-Os06g0231300
2-Os06g0245800
3-Os06g0247500 4-Os06g0247800
2-9820-455 3-1735-1424
4-3186-1560 5-ABC04803-1-1-392
ABC11018-1-1-216
779-2258
2429-1929 ABC10546-1-2-488
4.2
0.1
2251-643
ABC28974-pHv78-02
Figure 3-7. Comparison of the barley 7H QTL region and the orthologous rice
chromosome 6. 1- and 2-LOD QTL support intervals are indicated by bars
and lines, respectively, to the left of the barley 7H QTL region. Distance
between loci on 7H are in cM, distances between loci on rice 6 are in
Megabases. Orthologous rice and barley loci are preceded by the same
number.
81
It is conceivable that a locus impacting DMD may do so indirectly
by its effect on starch content, as starch is highly digestible in the
rumen. Further, starch is the principle source of energy in barley grain
(Huntington, 1997). Therefore, a QTL that reduces DMD by reducing
starch content may be of limited use: grain energy content would
likely be reduced. In order to assess pleiotropic or linked effects of
DMD QTL loci on starch content, interval analysis was conducted using
ANOVA. The starch content of RILs was compared at contrasting
alleles in the detected QTL intervals GMS6-Bmag0009 and
e43h58(76)-Bmag0507 (Table 3-8). The DMD QTL on 6H is nearly
significant (P=0.08) for starch content and, as may be expected, the
Haxby allele that increases DMD is also associated with increased
starch content. This QTL requires further research to ensure that it
does not have deleterious effects on barley feed quality. The DMD QTL
on 7H is not associated with changes in starch content (P=0.97).
82
Table 3-8. Type III tests of qDMD-6H and qDMD-7H effects on starch
content
Mean SCa
F Value
P Value
qDMD-6H, Haxby allele
61.4
4.08
0.08
qDMD-6H, Baku allele
58.0
qDMD-7H, Haxby allele
59.6
0.01
0.97
qDMD-7H, Baku allele
59.8
a
Least squares means starch content (%)
QTL Validation
Ninety-four RILs of a cross between Drummond and Baku were
evaluated for digestibility and particle size and genotyped with the
marker Bmag0009 in an attempt to validate the putative QTL on 6H.
Genotyping this population for markers at the 7H locus would not
make sense, as low DMD at that locus is associated with the Haxby
allele. By F-test, the 6H locus did not significantly affect DMD or mean
particle size in the validation population (Table 3-9). In fact, contrary
to expectation, the mean DMD of lines with the Baku allele is greater
than the mean DMD of those lines with the Drummond allele.
83
Table 3-9. Type III tests of Bmag0009 effects on DMD and mean particle
size in an F5 validation population
Trait
Allele
Meana
F Value
P Value
DMD (%)
Drummond
40.4
2.55
0.114
Baku
43.5
Drummond
1249
1.56
0.216
Baku
1197
Particle Size (μm)
a
Least squares means
The inability to validate this locus in a second population is
disappointing. There are several possible causes for this lack of
validation. DMD variation in this population may be a result of
multiple loci with minor effects. It has repeatedly been shown by
computer simulations and by QTL validation experiments that QTL
effects are often overestimated in QTL detection populations
(Melchinger et al, 1998; Utz et al, 2000). It may be that the QTL
detected on 6H is a true QTL, but its effect is smaller than estimated,
small enough even that its effect is undetectable in another
population. A second possibility is that the detected QTL is actually
false and has no effect on DMD whatsoever. This possibility seems
unlikely given that the LOD score of 7.4 obtained by composite interval
analysis equates to a P value of approximately 1e-8. With single
marker regression, the LOD of 3.37 equates to a P value of less than
84
0.001. It must be admitted that this value is greater than the value of
0.0001 that was suggested by Bernardo (2004) for use in marker-trait
regression analysis.
The final possibility is that the 2-row/6-row gene, Vrs1, is
masking the effects of the 6H DMD QTL. Drummond is a 6-row
cultivar and Baku is a 2-row cultivar and it has been shown that 6-row
cultivars generally have lower DMD than 2-rows (Bowman et al, 2001).
In populations segregating for Vrs1, head type accounts for 30-60% of
variation in DMD. It has been suggested that the reduced DMD is the
result of lateral kernels escaping extensive cracking due to their
reduced size and that less damaged kernels allow less microbial
colonization. The Drummond/Baku population again shows that the 6row head-type is associated with reduced DMD (Table 3-10).
Table 3-10. Type III tests of head type effects on DMD and mean particle
size (PS) in an F5 validation population
Trait
Head Type
Meana
F value
P Value
DMD (%)
2-row
46.8
41.89
<0.0001
35
6-row
35.5
2-row
1125
63.26
<0.0001
47
6-row
1372
PS (μm)
a
Least squares means
R2 (%)
85
Contrasts between Bmag0009 alleles within head type were not
significant (Table 3-11). Because of the small numbers involved in the
contrasts, however, it is difficult to say whether the lack of significance
is meaningful.
Table 3-11. Type III tests of Bmag0009 within head type
Head Type
Bmag0009
Mean DMD (%)a
F Value
P Value
6-rown=52
Drummond
35.6
0.01
0.93
Baku
35.4
Drummond
47.0
0.02
0.88
Baku
46.7
2-rown=42
a
Least squares means
We have evaluated the DMD of a 96 member population derived
from the same cross of Haxby and Baku. This population may be more
useful for validation of the DMD loci.
86
References
Abdel-Haleem H.A. 2004. Genetics and mapping of Quantitative Trait
Loci of feed quality-related traits in barley (Hordeum vulgare L.). PhD
dissertation. Montana State University-Bozeman.
(www.montana.edu/etd/available/abdel-haleem_0505.html).
AOAC. 2000. Official Methods of Analysis. 17th ed. Assoc. Offic. Anal.
Chem., Gaithersburg, MD.
Barth S., A.E. Melchinger, B. Devezi-Savula, and T. Lϋbberstedt.
2001. Influence of genetic background and heterozygosity on meiotic
recombination in Arabidopsis thaliana. Genome. 44:971-978.
Beecher B., J. Bowman, J.M. Martin, A.D. Bettge, C.F. Morris, T.K.
Blake, and M.J. Giroux. 2002. Hordoindolines are associated with a
major endosperm-texture QTL in barley (Hordeum vulgare). Genome.
45:584-591.
Benham J.J. 2001. Genographer, Version 1.6.0. Montana State
University, Bozeman.
(http://hordeum.oscs.montana.edu/software/genographer/)
Becker J., and M. Heun. 1995. Barley microsatellites: allele variation
and mapping. Plant Mol. Biol. 27:835-845.
Bernardo R. 2004. What proportion of declared QTL in plants are
false? Theor. Appl. Genet. 109:419-424.
BlakeT.K., D. Kadyrzhanova, K.W. Shepard, A.K.M.R Islam, P.L
Langridge, C.L. McDonald, J. Erpelding, S. Larson, N.K. Blake and L.E.
Talbert. 1996. STS-PCR markers appropriate for wheat-barley
introgression. Theor. Appl. Genet. 93:826-832.
Bonin A., E. Bellemain, P. Bronken Eidesen, F. Pompanon, C.
Brochmann, and P.Taberlet. 2004. How to track and assess
genotyping errors in population genetics studies. Mol. Ecol. 13:32613273.
Bowman J.G.P., T.K. Blake, L.M.M. Surber, D.K. Habernicht, and H.
Bockelman. 2001. Feed-quality variation in the barley core collection
of the USDA National Small Grains Collection. Crop Sci. 41:863-870.
87
Bowman J.G.P., T.K. Blake, L.M.M. Surber, D.K. Habernicht, T.K.
Daniels, and J.T. Daniels. 1996. Genetic factors controlling
digestibility of barley for ruminants. Proc. West. Sec. Am. Soc. Anim.
Sci. 47 (http://hordeum.oscs.montana.edu/cowdocs/cow596.htm)
Churchill, G. A. and R. W. Doerge (1994) Empirical threshold values
for quantitative trait mapping. Genetics 138:963-971.
Close T. 2006. OPAs. Centerpiece of BarleyCAP genotyping activities.
In: BarleyCAP Update. December 2006: 8
(http://barleycap.coafes.umn.edu/pdfs/Newsletter%20December%202
006.pdf)
Close T., and R. Waugh. 2006. Recent history of artificial outcrossing
facilitates whole-genome association mapping in elite inbred crop
varieties. PNAS. 103:18656-18661.
Cornu A., E. Farcy, and C. Mousset. 1989. A genetic basis for the
variations in meiotic recombination in Petunia hybrida. Genome.
32:46-53.
Costa J.M., A. Corey, P.M. Hayes, C. Jobet, A. Kleinhofs, A. KopischObusch, S.F. Kramer, D. Kudrna, M. Li, O. Riera-Lizarazu, K. Sato, P.
Szucs, T. Toojinda, M.I. Vales, and R.I. Wolfe 2001. Molecular
mapping of thee Oregon Wolfe Barleys: a phenotypically polymorphic
doubled-haploid population. Theor. Appl. Genet. 103:415-424.
Fradin D. and P. Bougneres. 2007. Three common intronic variants in
the maternal and fetal thiamine pyrophosphokinase gene (TPK1) are
associated with birth weight. Ann. Hum. Genet. 71:1-8.
Grimson, R.E., R.D. Weisenburger, J.A. Basarab, and R.P. Stilborn.
1987. Effects of barley volume-weight and processing method on
feedlot performance of finishing steers. Can. J. Anim. Sci. 67:43-53.
Hackett C.A. and L.B. Broadfoot. 2003. Effects of genotyping errors,
missing values and segregation distortion in molecular marker data on
the construction of linkage maps. Heredity. 90:33-38.
Hadad R.G., T.W. Pfeiffer, and C.G. Poneleit. 1996. Repeatability and
heritability of divergent recombination frequencies in the Iowa Stiff
Stalk Synthetic. Theor. Appl. Genet. 93:990-996.
88
Kosambi D.D. 1944. The estimation of map distance from
recombination values. Ann Eugen 12 : 172Ð175
Li J.Z., T.G. Sjakste, M.S. Roder, and M.W. Ganal. 2003.
Development and mapping of 127 new microsatellite markers in
barley. Theor. Appl. Genet. 107:1021-1027.
Liu Z.-W., R.M. Biyashev, and M.A. Saghai Maroof. 1996.
Development of simple sequence repeat DNA markers and their
integration into a barley linkage map. Theor. Appl. Genet. 93:869-876.
Mathison G.W., R. Hironaka, B.K. Kerrigan, I. Vlach, I.P. Milligan, and
R.D. Weisengurger. 1991. Rate of starch degradation, apparent
digestibility, and rate and efficiency of steer gain as influenced by
barley grain volume-weight and processing method. Can. J. Anim. Sci.
71:867-878.
Melchinger A.E., H.F. Utz, and C.C. Schön. 1998. Quantitative trait
locus (QTL) mapping using different testers and independent
population samples in maize reveals low power of QTL detection and
large bias in estimates of QTL effects. Genetics. 149:383-403
Noor M.A.F., A.L. Cunningham, and J.C. Larkin. 2001. Consequences
of recombination rate variation on quantitative trait locus mapping
studies: simulations based on the Drosophila melanogaster genome.
Genetics. 159: 581-588.
Pfost, H., and V. Headley. 1976. Methods of determining and
expressing particle size. Page 517 in Feed Manufacturing Technology
II. H. Pfost, ed. Anim. Feed Manuf. Assoc., Arlington, VA
Qi X., P. Stam, P. Lindhout. 1998. Use of locus-specific AFLP markers
to construct a high-density molecular map in barley. Theor. Appl.
Genet. 96:376-384.
Ramsay L., M. Macaulay, S. degli Ivanissevich, K. MacLean, L. Cardle,
J. Fuller, K. J. Edwards, S. Tuvesson, M. Morgante, A. Massari, E.
Maestri, N. Marmiroli, T. Sjakste, M. Ganal, W. Powell, and R. Waugh.
2000. A simple sequence repeat-based linkage map of barley.
Genetics 156:1997-2005.
89
Ramsey P.B., G.W. Mathison, and L.A. Goonewardene. 2002. Effect of
rates and extents of ruminal barley grain dry matter and starch
disappearance on bloat, liver abscesses, and performance of feedlot
steers. Anim. Feed Sci. Tech. 13:145-157.
Rostoks N., L. Ramsay, K. MacKenzie, L. Cardle, P.R. Bhat, M.L. Roose,
J.T. Svensson, N. Stein, R.K. Varshney, D.F. Marshall, A. Graner, T.J.
Säll T. 1990. Genetic control of recombination in barley. II.
Variation in linkage between marker genes. Hereditas. 112:171-178.
Schuelke, M., 2000. An economic method for the fluorescent labeling
of PCR products. Nat. Biotechnol. 18: 233–234.
See D., V. Kanazin, K. Kephart, and T. Blake. 2002. Mapping genes
controlling variation in barley grain protein concentration. Crop Sci.
42: 680-685
Shen R., J.-B. Fan, D. Campbell, W. Chang, J. Chen, D. Doucet, J.
Yeakley, M. Bibikova, E.W. Garcia, C. McBride, F. Steemers, F. Garcia,
B.G Kermani, K. Gunderson, and A. Oliphant. 2005. High-throughput
SNP genotyping on universal bead arrays. Mutat. Res. 573:70-82.
Stam P. 1993. Construction of integrated genetic linkage maps by
means of a new computer package: JoinMap. Plant J. 3:739-744
Struss D., J. Plieske. 1998. The use of microsatellite markers for
detection of genetic diversity in barley populations. Theor. Appl.
Genet. 97:308–315
Surber L.M.M., J.G.P. Bowman, T.K. Blake, D.D. Hinman, D.L. Boss,
and T.C. Blackhurst. 2000. Prediction of barley feed quality for beef
cattle from laboratory analyses. Proc. West. Sec. Am. Soc. Anim. Sci.
51:454-457.
Swan C.G., J.G.P. Bowman, J.M. Martin, M.J. Giroux. 2006. Increased
puroindoline levels slow ruminal digestion of wheat (Triticum aestivum
L.) starch by cattle. J. Anim Sci. 84: 641-650.
USDA National Agricultural Statistics Service. 2007.
(http://www.nass.usda.gov/)
90
Utz H.F., A.E. Melchinger, and C.C. Schön. 2000. Bias and sampling
error of the estimated proportion of genotypic variance explained by
quantitative trait loci determined from experimental data in maize
using cross validation and validation with independent samples.
Genetics. 154:1839-1849.
Van Ooijen, J.W. & R.E. Voorrips, 2001. JoinMap® 3.0, Software for
the calculation of genetic linkage maps. Plant Research International,
Wageningen, the Netherlands.
Vanzant, E.S., R.C. Cochran, and E.C. Titgemeyer. 1998.
Standardization of in situ techniques for ruminant feedstuff evaluation.
J. Anim. Sci. 76:2717-2729
Voorrips, R.E., 2002. MapChart: Software for the graphical
presentation of linkage maps and QTLs. The Journal of Heredity 93
(1): 77-78.
Vos P., J. Hogers, M. Blaeker, M. Reijans, T. Lee, M. Hornes, A.
Frijters, J. Pot, J. Peleman, M. Kuiper, and M. Zabeau. 1995. AFLP: a
new technique for DNA fingerprinting. Nucleic Acids Res. 23: 4407–
4414.
Wang S., C. J. Basten, and Z.-B. Zeng 2006. Windows QTL
Cartographer 2.5. Department of Statistics, North Carolina State
University, Raleigh, NC.
(http://statgen.ncsu.edu/qtlcart/WQTLCart.htm)
Zhu C., C. Wang, and Y. Zhang. 2007. Modeling segregation
distortion for viability selection I. Reconstruction of linkage maps with
distorted markers. Theor Appl Genet. 114:295-305.
Zieliński H. and H. Kozlowska. 2000. Antioxidant activity and total
phenolics in selected cereal grains and their different morphological
fractions. J. Agric. Food Chem. 48:2008-2016.