Directory UMM :Data Elmu:jurnal:A:Agronomy Journal:Vol93.Issue1.2001:
8
AGRONOMY JOURNAL, VOL. 93, JANUARY–FEBRUARY 2001
Weed suppressing rice cultivars—Does allelopathy play a role?
Weed Res. 39:441–454.
Olofsdotter, M., M. Rebulanan, A. Madrid, W. Dali, D. Navarez, and
D.C. Olk. 2001. Why phenolic acids are unlikely allelochemicals
in rice. J. Chem. Ecol. (in press).
Pheng, S., S. Adkins, M. Olofsdotter, and G. Jahn. 1999. Allelopathic
effects of rice (Oryza sativa L.) on the growth of awnless barnyard
grass [Echinochloa colona (L.) Link]: A new form for weed management. Cambodian J. Agric. 2(l):42–49.
Rice, E.L. 1995. Biological control of weeds and plant diseases: Advances in applied allelopathy. Univ. of Oklahoma Press, Norman.
Rimando, A.M., M. Olofsdotter, and S.O. Duke. 2001. Searching for
rice allelochemicals. Agron. J. 93:16–20 (this issue).
Tanaka, F., S. Ono, and T. Hayasaka. 1990. Identification and evaluation of toxicity of rice root elongation inhibitors in flooded soils
with added wheat straw. Soil Sci. Plant Nutr. 36:97–103.
Wu, H., J. Pratley, D. Lemerle, and T. Haig. 1999. Crop cultivars with
allelopathic capability. Weed Res. 39:171–180.
Barnyardgrass Growth Inhibition with Rice Using High-Performance Liquid
Chromatography to Identify Rice Accession Activity
John D. Mattice,* Robert H. Dilday, Edward E. Gbur, and Briggs W. Skulman
ABSTRACT
Some accessions of rice (Oryza sativa L.) have been shown to
inhibit the growth of barnyardgrass (Echinochloa crus-galli (L.)
Beauv.). Our objective was to determine if high-performance liquid
chromatography (HPLC) chromatograms from leaf extracts of different accessions of rice correlated with weed control activity. Chromatograms of extracts consisting of 10 mg of fresh leaf tissue per milliliter
of methanol (CH3OH) were obtained from 40 accessions of rice.
Cluster analysis was performed using 20 peaks from the chromatograms. Three clusters were found, with one cluster being distinctly
separated from the other two. Although weed control data are not
available for all the accessions, the isolated cluster contains all of the
accessions that have been shown to inhibit growth of barnyardgrass
and none that do not. This indicates that the assay could be used
year-round to screen accessions of rice for weed control potential to
determine which accessions should be further tested in the field. This
could be done in a relatively short time using a small amount of space
in the greenhouse. Because the assay requires only 10 mg of tissue
per milliliter of methanol, it may potentially be used to test individual
plants within an accession for weed control potential in a nondestructive manner.
D
ilday et al. (1989, 1991) first observed the interference of rice on the growth of ducksalad [Heteranthera limosa (Sw.) Willd.] in field tests evaluating
accessions of rice for tolerance to alachlor [2-chloro29,69-diethyl-N-(methoxymethyl)acetanilide]. Since 1987,
laboratory and field tests have been performed to identify accessions that inhibit the growth of several weed
species, including barnyardgrass. Growth inhibition of
barnyardgrass has also been reported by Navarez and
Olofsdotter (1996), Hasan et al. (1998), and Kim and
Shin (1998). We have also observed it routinely in greenhouse bioassays.
Although the interference may be due to allelopathy,
there is also the possibility that it may be due to competition or a mixture of competition and allelopathy. Either
way, if the trait can be incorporated into agronomically
useful varieties, fewer hours may be required for manual
J.D. Mattice, E.E. Gbur, and B.W. Skulman, Dep. of Crop, Soil, and
Environ. Sci., Univ. of Arkansas, Fayetteville, AR 72704. R.H. Dilday,
USDA-ARS, Dale Bumpers Natl. Rice Res. Cent., Stuttgart, AR
72160. E.E. Gbur, Agric. Statistics Lab., Univ. of Arkansas, Fayetteville, AR 72701. Received 29 Nov. 1999. *Corresponding author
(jmattice@uark.edu).
Published in Agron. J. 93:8–11 (2001).
weeding, and reduced rates or fewer applications of
herbicides may be required for weed control.
A useful tool for breeders would be an assay to screen
accessions and individual plants within accessions for
weed control activity. The assay would ideally be accomplished in a relatively short period of time, require a
minimum amount of space, be relatively inexpensive,
and could be done year-round in a greenhouse. We
report here an HPLC procedure that is showing promise
toward meeting most of these criteria.
MATERIALS AND METHODS
Rice Extraction
Approximately 15 seeds were placed in 100 g of soil sieved
through a 2-mm mesh in the rice growing region of Stuttgart,
AR. The samples were grown in 474-mL (16 oz) plastic cups
and thinned to 10 plants cup21, with three replications per
accession. After 10 d, the leaves from each replication were
removed, cut into approximately 1-cm lengths, and placed in
Erlenmeyer flasks. A volume of HPLC grade methanol was
added such that the ratio of fresh plant tissue/methanol was 10
mg mL21. The samples were placed in a refrigerator overnight.
Then equal parts of the methanol extract and deionized water
were combined and analyzed by HPLC.
High-Performance Liquid Chromatography
Conditions
Analyses were performed using a 25-cm by 4.6-mm Phenomenex Prodigy C18 column. The HPLC system consisted of a
Hitachi L-7450A diode array detector, L-7200 autosampler,
L-7100 pump, and the Hitachi HSM software for data processing. Solvent was degassed with an ERC model 3415a degasser, and the column was held at 358C with an Eppendorf
TC-45 heater. The gradient used 1% acetic acid (vol./vol.)
and HPLC grade acetonitrile (acet). The program was 10%
acet (vol./vol.) at 1.5 mL min21 for 3 min, increased to 50% acet
(vol./vol.) over 27 min at 1.5 mL min21, increased to 80% acet
(vol./vol.) at 2 mL min21 over 0.1 min and held for 1.9 min,
decreased to 10% acet (vol./vol.) over 0.1 min and held for 7.9
min, and decreased to 1.5 mL min21 over 0.1 min. The total
run time was 40 min, and data were collected for the first 30
min. The first and last portions of the chromatogram contained
only peaks that were essentially background. The injection
volume was 30 mL and quantitation was at 320 nm.
Abbreviations: acet, acetonitrile; HPLC, high-performance liquid
chromatography.
MATTICE ET AL.: INHIBITION OF BARNYARDGRASS GROWTH WITH RICE USING HPLC
Cluster Analysis
The peaks that were considered to be above background
were used for data analysis. This resulted in 20 peaks being
used. The chromatograms from some accessions contained all
20 peaks; for other accessions, some peaks were absent.
The set of peak heights from each sample was considered
as a point in 20-dimensional space. The peak height data were
subjected to K-means clustering (Hand, 1981, p. 174) for K 5
2 to 7 clusters. K-means clustering is a nonhierarchical iterative
clustering method in which the centroids of the K initial clusters are determined. If any point within a cluster is determined
to be closer to the centroid of a different cluster, then that point
is reassigned to the different cluster. The cluster centroids are
then recalculated, and the procedure is repeated until there
are no changes in the clusters. The K-means procedure minimizes the sum of squared distances of the observations from
their assigned cluster centroids and is analogous to the minimiTable 1. The accessions studied, weed control activity (if known),
cluster containing accession, and pedigree as listed in the
USDA-ARS Germplasm Resource Information Network
(GRIN).
Accession
Activity†
Cluster
1053RU960153
Alan
Cocodric
Cypress
Dellmont
Delrose
Dixiebelle
U
U
U
U
U
N
N
1
1
1
1
1
1
1
Drew
GuiChao
Jackson
Jasmine
U
Y
U
Y
1
3
1
3
Jefferson
Katy
U
N
1
1
Kaybonnet
Koshihikari
L204
N
U
U
2
1
2
Lacassine
Laffite
Lagrue
U
U
N
2
2
2
Lemont
Litton
M204
N
U
U
1
2
1
Mars
Millie
Newbonnet
Orion
PI 373026
PI 350468
PI 366150
PI 338046
PI 312777
RA-73
Rexmont
RICO1
STG94L42-130
STG95L05-032
Teqing
N
U
N
U
Y
Y
Y
Y
Y
U
N
U
U
U
Y
1(2)§
2(1)
2
2
3
3
3
3
3
2(1)
1
1
1
1
3
TN-1
Y
3
Truebonnet
ZHE733
U
Y
2
3
Pedigree
NA‡
Labelle/L-201
NA
L-202/Lemont
Della-X2/5*Lemont
NA
(Newrex/Bellemont)RU830
3181/CB801
Newbonnet/Katy
developed. From China
RU7603015/L-201
developed 1998. From the
United States
Rosemont/B82-761
Bonnet 73/CI9722//Starbonnet/Tetep/3/Lebonnet
Katy/Newbonnet
Norin 22/Norin 1
Lemont//Tainung-sen-yu 2414/
L-201
Newbonnet/Lemont
NA
Bonnet 73/Nova 76//Bonnet
73/3/Newrex
Lebonnet/CIor 9881/PI 331581
NA
M201/M7/3/M7//ESD7-3/
Kokuhorose
CI9580/Saturn
Lebonnet/L-201
Dawn/Bonnet 73
Brazos/Mars
IR8/NM 54
2*IR8//Yuhkara/TN 1
developed. From Taiwan
2*IR8//2*B5894a4-18-1/TN 1
2*T65/TN 1
NA
Newrex/Bellemont
Nortai//CI 9545/Nova
NA
NA
collected from Zhejiang,
China
Tie-cha-oo-chien/Tsai-yuanchung
NA
developed. From China
† Y, yes; N, no; U, unknown.
‡ NA, not available.
§ Two replications were in the group without parentheses, and one replication was in the group in parentheses.
9
zation of the sum of squared errors in an analysis of variance
by the least-squares estimators.
Using the clusters defined by the K-means procedure, the
first two canonical variables were calculated and plotted to
show the separation among the clusters as clearly as possible
in two dimensions (Krzanowski and Marriott, 1994, p. 91). All
analyses were carried out using SAS (Version 7, SAS Inst.,
Cary, NC).
Rice Accessions Used
The rice accessions that were used, and information regarding the pedigree, clustering, and weed control activity, when
known, are listed in Table 1.
RESULTS AND DISCUSSION
The chromatograms were of two main types, as shown
in Fig. 1. The chromatogram for PI 312777 is representative of those from accessions showing activity while the
chromatogram for Rexmont is representative of those
showing little or no activity. The peaks in the PI 312777
chromatogram are substantially higher for compounds
whose retention times are 12.2, 12.6, and 13.8 min. PI
312777 also contains compounds producing peaks at
14.1, 14.6, and 15.15 min, which are essentially absent
in the extracts from Rexmont. Additionally, there are
peaks at 13.6, 14.8, and 15.3 min that are larger in the
chromatogram of the Rexmont extract. Figure 2 shows
the expanded section of the chromatograms from 15.0
to 15.7 min. Rexmont, and to a lesser extent PI 312777,
both contain a compound whose retention time is approximately 15.3 min. However, only the chromatogram
from PI 312777 contains a peak at 15.15 min. Most of
the chromatograms from the 40 accessions investigated
were similar to either the Rexmont or PI 312777 chromatograms.
Allelopathy is commonly thought to be a result of
the action of several compounds rather than just one.
This creates a problem when comparing chromatograms
from a set of samples that show activity with a set that
does not show activity. One or more of the compounds
may be unique to the set showing activity, but it is more
likely that the same compounds are present in both sets
Fig. 1. High-performance liquid chromatography (HPLC) chromatograms of methanol extracts of rice leaf tissue from PI 312777 and
Rexmont from 11.8 to 15.8 min.
10
AGRONOMY JOURNAL, VOL. 93, JANUARY–FEBRUARY 2001
Fig. 2. High-performance liquid chromatography (HPLC) chromatograms of methanol extracts of rice leaf tissue from PI 312777 and
Rexmont from 15.0 to 15.8 min.
in differing amounts. The chromatograms in Fig. 1 and
2 show that there may be up to nine peaks that appear
to differentiate the chromatograms of the PI 312777
extracts from those of Rexmont. However, there is some
variability in the peak heights from sample to sample,
and the peaks that are responsible for the clustering
may not be obvious from simple inspection.
One analytical approach is to compare each peak
from the set showing activity with the same peak from
the inactive set. If a peak were found to be significantly
higher in the allelopathic set, it might be related to the
effect. The problem with this approach is that there may
be numerous peaks in the chromatograms that need to
be compared. At a level of significance of P 5 0.05, the
risk of falsely finding a significant difference when there
is none is approximately one minus (0.95)n, where n is
the number of peaks being compared. If, as in our case,
20 peaks were being compared, we would falsely find
significant differences 64% of the time even if there
were no difference in the size of any of the pairs of
peaks. To avoid this problem, our approach has been
to use all 20 peaks in the chromatogram to determine
a point in 20-dimensional space, and then use cluster
analysis to see if the points are in different clusters.
K-means clustering for two clusters did not separate
accessions showing weed control activity from those that
did not. The results for K 5 3 clusters are shown in Fig.
3 where the isolated Cluster 3 contains those accessions
that so far have shown activity. The other two clusters
represent more of a division of a cloud of data points
rather than two well-separated groups. The results for
K 5 4, 5, and 6 clusters showed further division of the
latter into smaller, relatively nondistinct groups. The
isolated Cluster 3 containing the accessions showing
activity remained intact through K 5 6 clusters. For
K 5 7 clusters, the cluster split, but the two newly
formed clusters were not well-separated. Hence, three
Fig. 3. Two-dimensional representation of 3 clusters where each point is determined by 20 chromatographic peaks. Cluster 3 contains the
accessions showing weed control activity.
MATTICE ET AL.: INHIBITION OF BARNYARDGRASS GROWTH WITH RICE USING HPLC
clusters appear to be sufficient to separate these 40
accessions into those showing activity and those that
do not.
For our accessions, we have determined that those
with known activity are in a cluster by themselves. However, for any given set of accessions, there is no guarantee that the separation would be so clearly defined because the range of activity may not be as great as was
found in our group. Moreover, if there is more than one
mechanism for growth inhibition, the set of defining
chromatographic peaks would not necessarily be the
same as ours.
The next step in our research is to determine which
peaks, and ultimately which compounds, are primarily
responsible for the clustering. This step must necessarily
consider peak heights as well as missing peaks.
It is important to remember that correlation does not
imply causality, and we do not imply that the compounds
producing the larger peaks in the PI 312777 extract
would be allelochemicals; they would, however, be candidates for identification and testing. Differences in the
peak size, regardless of which chromatogram it is in,
may be useful in differentiating accessions according to
their ability to inhibit barnyardgrass growth. Whether
or not the compounds are allelochemicals, and whether
or not the observed effect is allelopathy or competition,
the procedure shows promise for predicting which accessions are likely to show a weed control effect toward
barnyardgrass and perhaps other weed species. The procedure allows assaying of 7- to 10-d-old samples, so
screening can be done on a series of samples during
late summer through early spring to identify promising
accessions to take to the field for further testing. This
meets the objectives of being accomplished in a relatively short period of time (#10 d), using a minimum
amount of space (≈1 m2 for 30 accessions), and can be
done year-round. High-performance liquid chromatog-
11
raphy is not an inexpensive technique but is widely
available.
Because the procedure requires only 10 mg of tissue
per milliliter of methanol, it can be done in a nondestructive manner on a rice plant. It remains to be seen if the
procedure could be used to identify which plants within
a cross between two accessions would be most likely to
have the highest weed control activity, and would thus
be the most useful to breeders.
ACKNOWLEDGMENTS
We acknowledge the contribution of the USDA, Arkansas
Rice Research and Promotion Board, and IRRI for support
of this project.
REFERENCES
Dilday, R.H., P. Nastasi, J. Lin, and R.J. Smith Jr. 1991. Allelopathic
activity in rice (Oryza sativa L.) against ducksalad [Heteranthera
limosa (Sw) Willd.]. p. 193–201. In J.N. Hanson et al. (ed.) Proc.
Sustainable Agric. for the Great Plains, Beltsville, MD. USDAARS.
Dilday, R.H., P. Nastasi, and R.J. Smith Jr. 1989. Allelopathic observation in rice (Oryza sativa L.) to ducksalad (Heteranthera limosa ).
Proc. Arkansas Acad. Sci. 43:21–22.
Hand, D.J. 1981. Discrimination and classification. John Wiley &
Sons, New York.
Hasan, S.M., I.R. Aidy, A.O. Bastawisi, and A.E. Draz. 1998. Weed
management using allelopathic rice varieties in Egypt. p. 27–38.
In M. Olofsdotter (ed.) Allelopathy in rice. Proc. Workshop on
Allelopathy in Rice, Manila, Philippines. 25–27 Nov. 1996. IRRI,
Makati City, Philippines.
Kim, K.U., and D.H. Shin. 1998. Rice allelopathy research in Korea.
p. 39–44. In M. Olofsdotter (ed.) Allelopathy in rice. Proc. Workshop on Allelopathy in Rice, Manila, Philippines. 25–27 Nov. 1996.
IRRI, Makati City, Philippines.
Krzanowski, W.J., and F.H.C. Marriott. 1995. Multivariate analysis.
Part 1. Distributions, ordination, and inference. Edward Arnold,
London.
Navarez, D.C., and M. Olofsdotter. 1996. Relay seeding technique
for screening allelopathic rice (Oryza sativa ). p. 1285–1290. In Proc.
Int. Weed Control Congr., 2nd, Copenhagen, Denmark. 25–28
June 1996.
AGRONOMY JOURNAL, VOL. 93, JANUARY–FEBRUARY 2001
Weed suppressing rice cultivars—Does allelopathy play a role?
Weed Res. 39:441–454.
Olofsdotter, M., M. Rebulanan, A. Madrid, W. Dali, D. Navarez, and
D.C. Olk. 2001. Why phenolic acids are unlikely allelochemicals
in rice. J. Chem. Ecol. (in press).
Pheng, S., S. Adkins, M. Olofsdotter, and G. Jahn. 1999. Allelopathic
effects of rice (Oryza sativa L.) on the growth of awnless barnyard
grass [Echinochloa colona (L.) Link]: A new form for weed management. Cambodian J. Agric. 2(l):42–49.
Rice, E.L. 1995. Biological control of weeds and plant diseases: Advances in applied allelopathy. Univ. of Oklahoma Press, Norman.
Rimando, A.M., M. Olofsdotter, and S.O. Duke. 2001. Searching for
rice allelochemicals. Agron. J. 93:16–20 (this issue).
Tanaka, F., S. Ono, and T. Hayasaka. 1990. Identification and evaluation of toxicity of rice root elongation inhibitors in flooded soils
with added wheat straw. Soil Sci. Plant Nutr. 36:97–103.
Wu, H., J. Pratley, D. Lemerle, and T. Haig. 1999. Crop cultivars with
allelopathic capability. Weed Res. 39:171–180.
Barnyardgrass Growth Inhibition with Rice Using High-Performance Liquid
Chromatography to Identify Rice Accession Activity
John D. Mattice,* Robert H. Dilday, Edward E. Gbur, and Briggs W. Skulman
ABSTRACT
Some accessions of rice (Oryza sativa L.) have been shown to
inhibit the growth of barnyardgrass (Echinochloa crus-galli (L.)
Beauv.). Our objective was to determine if high-performance liquid
chromatography (HPLC) chromatograms from leaf extracts of different accessions of rice correlated with weed control activity. Chromatograms of extracts consisting of 10 mg of fresh leaf tissue per milliliter
of methanol (CH3OH) were obtained from 40 accessions of rice.
Cluster analysis was performed using 20 peaks from the chromatograms. Three clusters were found, with one cluster being distinctly
separated from the other two. Although weed control data are not
available for all the accessions, the isolated cluster contains all of the
accessions that have been shown to inhibit growth of barnyardgrass
and none that do not. This indicates that the assay could be used
year-round to screen accessions of rice for weed control potential to
determine which accessions should be further tested in the field. This
could be done in a relatively short time using a small amount of space
in the greenhouse. Because the assay requires only 10 mg of tissue
per milliliter of methanol, it may potentially be used to test individual
plants within an accession for weed control potential in a nondestructive manner.
D
ilday et al. (1989, 1991) first observed the interference of rice on the growth of ducksalad [Heteranthera limosa (Sw.) Willd.] in field tests evaluating
accessions of rice for tolerance to alachlor [2-chloro29,69-diethyl-N-(methoxymethyl)acetanilide]. Since 1987,
laboratory and field tests have been performed to identify accessions that inhibit the growth of several weed
species, including barnyardgrass. Growth inhibition of
barnyardgrass has also been reported by Navarez and
Olofsdotter (1996), Hasan et al. (1998), and Kim and
Shin (1998). We have also observed it routinely in greenhouse bioassays.
Although the interference may be due to allelopathy,
there is also the possibility that it may be due to competition or a mixture of competition and allelopathy. Either
way, if the trait can be incorporated into agronomically
useful varieties, fewer hours may be required for manual
J.D. Mattice, E.E. Gbur, and B.W. Skulman, Dep. of Crop, Soil, and
Environ. Sci., Univ. of Arkansas, Fayetteville, AR 72704. R.H. Dilday,
USDA-ARS, Dale Bumpers Natl. Rice Res. Cent., Stuttgart, AR
72160. E.E. Gbur, Agric. Statistics Lab., Univ. of Arkansas, Fayetteville, AR 72701. Received 29 Nov. 1999. *Corresponding author
(jmattice@uark.edu).
Published in Agron. J. 93:8–11 (2001).
weeding, and reduced rates or fewer applications of
herbicides may be required for weed control.
A useful tool for breeders would be an assay to screen
accessions and individual plants within accessions for
weed control activity. The assay would ideally be accomplished in a relatively short period of time, require a
minimum amount of space, be relatively inexpensive,
and could be done year-round in a greenhouse. We
report here an HPLC procedure that is showing promise
toward meeting most of these criteria.
MATERIALS AND METHODS
Rice Extraction
Approximately 15 seeds were placed in 100 g of soil sieved
through a 2-mm mesh in the rice growing region of Stuttgart,
AR. The samples were grown in 474-mL (16 oz) plastic cups
and thinned to 10 plants cup21, with three replications per
accession. After 10 d, the leaves from each replication were
removed, cut into approximately 1-cm lengths, and placed in
Erlenmeyer flasks. A volume of HPLC grade methanol was
added such that the ratio of fresh plant tissue/methanol was 10
mg mL21. The samples were placed in a refrigerator overnight.
Then equal parts of the methanol extract and deionized water
were combined and analyzed by HPLC.
High-Performance Liquid Chromatography
Conditions
Analyses were performed using a 25-cm by 4.6-mm Phenomenex Prodigy C18 column. The HPLC system consisted of a
Hitachi L-7450A diode array detector, L-7200 autosampler,
L-7100 pump, and the Hitachi HSM software for data processing. Solvent was degassed with an ERC model 3415a degasser, and the column was held at 358C with an Eppendorf
TC-45 heater. The gradient used 1% acetic acid (vol./vol.)
and HPLC grade acetonitrile (acet). The program was 10%
acet (vol./vol.) at 1.5 mL min21 for 3 min, increased to 50% acet
(vol./vol.) over 27 min at 1.5 mL min21, increased to 80% acet
(vol./vol.) at 2 mL min21 over 0.1 min and held for 1.9 min,
decreased to 10% acet (vol./vol.) over 0.1 min and held for 7.9
min, and decreased to 1.5 mL min21 over 0.1 min. The total
run time was 40 min, and data were collected for the first 30
min. The first and last portions of the chromatogram contained
only peaks that were essentially background. The injection
volume was 30 mL and quantitation was at 320 nm.
Abbreviations: acet, acetonitrile; HPLC, high-performance liquid
chromatography.
MATTICE ET AL.: INHIBITION OF BARNYARDGRASS GROWTH WITH RICE USING HPLC
Cluster Analysis
The peaks that were considered to be above background
were used for data analysis. This resulted in 20 peaks being
used. The chromatograms from some accessions contained all
20 peaks; for other accessions, some peaks were absent.
The set of peak heights from each sample was considered
as a point in 20-dimensional space. The peak height data were
subjected to K-means clustering (Hand, 1981, p. 174) for K 5
2 to 7 clusters. K-means clustering is a nonhierarchical iterative
clustering method in which the centroids of the K initial clusters are determined. If any point within a cluster is determined
to be closer to the centroid of a different cluster, then that point
is reassigned to the different cluster. The cluster centroids are
then recalculated, and the procedure is repeated until there
are no changes in the clusters. The K-means procedure minimizes the sum of squared distances of the observations from
their assigned cluster centroids and is analogous to the minimiTable 1. The accessions studied, weed control activity (if known),
cluster containing accession, and pedigree as listed in the
USDA-ARS Germplasm Resource Information Network
(GRIN).
Accession
Activity†
Cluster
1053RU960153
Alan
Cocodric
Cypress
Dellmont
Delrose
Dixiebelle
U
U
U
U
U
N
N
1
1
1
1
1
1
1
Drew
GuiChao
Jackson
Jasmine
U
Y
U
Y
1
3
1
3
Jefferson
Katy
U
N
1
1
Kaybonnet
Koshihikari
L204
N
U
U
2
1
2
Lacassine
Laffite
Lagrue
U
U
N
2
2
2
Lemont
Litton
M204
N
U
U
1
2
1
Mars
Millie
Newbonnet
Orion
PI 373026
PI 350468
PI 366150
PI 338046
PI 312777
RA-73
Rexmont
RICO1
STG94L42-130
STG95L05-032
Teqing
N
U
N
U
Y
Y
Y
Y
Y
U
N
U
U
U
Y
1(2)§
2(1)
2
2
3
3
3
3
3
2(1)
1
1
1
1
3
TN-1
Y
3
Truebonnet
ZHE733
U
Y
2
3
Pedigree
NA‡
Labelle/L-201
NA
L-202/Lemont
Della-X2/5*Lemont
NA
(Newrex/Bellemont)RU830
3181/CB801
Newbonnet/Katy
developed. From China
RU7603015/L-201
developed 1998. From the
United States
Rosemont/B82-761
Bonnet 73/CI9722//Starbonnet/Tetep/3/Lebonnet
Katy/Newbonnet
Norin 22/Norin 1
Lemont//Tainung-sen-yu 2414/
L-201
Newbonnet/Lemont
NA
Bonnet 73/Nova 76//Bonnet
73/3/Newrex
Lebonnet/CIor 9881/PI 331581
NA
M201/M7/3/M7//ESD7-3/
Kokuhorose
CI9580/Saturn
Lebonnet/L-201
Dawn/Bonnet 73
Brazos/Mars
IR8/NM 54
2*IR8//Yuhkara/TN 1
developed. From Taiwan
2*IR8//2*B5894a4-18-1/TN 1
2*T65/TN 1
NA
Newrex/Bellemont
Nortai//CI 9545/Nova
NA
NA
collected from Zhejiang,
China
Tie-cha-oo-chien/Tsai-yuanchung
NA
developed. From China
† Y, yes; N, no; U, unknown.
‡ NA, not available.
§ Two replications were in the group without parentheses, and one replication was in the group in parentheses.
9
zation of the sum of squared errors in an analysis of variance
by the least-squares estimators.
Using the clusters defined by the K-means procedure, the
first two canonical variables were calculated and plotted to
show the separation among the clusters as clearly as possible
in two dimensions (Krzanowski and Marriott, 1994, p. 91). All
analyses were carried out using SAS (Version 7, SAS Inst.,
Cary, NC).
Rice Accessions Used
The rice accessions that were used, and information regarding the pedigree, clustering, and weed control activity, when
known, are listed in Table 1.
RESULTS AND DISCUSSION
The chromatograms were of two main types, as shown
in Fig. 1. The chromatogram for PI 312777 is representative of those from accessions showing activity while the
chromatogram for Rexmont is representative of those
showing little or no activity. The peaks in the PI 312777
chromatogram are substantially higher for compounds
whose retention times are 12.2, 12.6, and 13.8 min. PI
312777 also contains compounds producing peaks at
14.1, 14.6, and 15.15 min, which are essentially absent
in the extracts from Rexmont. Additionally, there are
peaks at 13.6, 14.8, and 15.3 min that are larger in the
chromatogram of the Rexmont extract. Figure 2 shows
the expanded section of the chromatograms from 15.0
to 15.7 min. Rexmont, and to a lesser extent PI 312777,
both contain a compound whose retention time is approximately 15.3 min. However, only the chromatogram
from PI 312777 contains a peak at 15.15 min. Most of
the chromatograms from the 40 accessions investigated
were similar to either the Rexmont or PI 312777 chromatograms.
Allelopathy is commonly thought to be a result of
the action of several compounds rather than just one.
This creates a problem when comparing chromatograms
from a set of samples that show activity with a set that
does not show activity. One or more of the compounds
may be unique to the set showing activity, but it is more
likely that the same compounds are present in both sets
Fig. 1. High-performance liquid chromatography (HPLC) chromatograms of methanol extracts of rice leaf tissue from PI 312777 and
Rexmont from 11.8 to 15.8 min.
10
AGRONOMY JOURNAL, VOL. 93, JANUARY–FEBRUARY 2001
Fig. 2. High-performance liquid chromatography (HPLC) chromatograms of methanol extracts of rice leaf tissue from PI 312777 and
Rexmont from 15.0 to 15.8 min.
in differing amounts. The chromatograms in Fig. 1 and
2 show that there may be up to nine peaks that appear
to differentiate the chromatograms of the PI 312777
extracts from those of Rexmont. However, there is some
variability in the peak heights from sample to sample,
and the peaks that are responsible for the clustering
may not be obvious from simple inspection.
One analytical approach is to compare each peak
from the set showing activity with the same peak from
the inactive set. If a peak were found to be significantly
higher in the allelopathic set, it might be related to the
effect. The problem with this approach is that there may
be numerous peaks in the chromatograms that need to
be compared. At a level of significance of P 5 0.05, the
risk of falsely finding a significant difference when there
is none is approximately one minus (0.95)n, where n is
the number of peaks being compared. If, as in our case,
20 peaks were being compared, we would falsely find
significant differences 64% of the time even if there
were no difference in the size of any of the pairs of
peaks. To avoid this problem, our approach has been
to use all 20 peaks in the chromatogram to determine
a point in 20-dimensional space, and then use cluster
analysis to see if the points are in different clusters.
K-means clustering for two clusters did not separate
accessions showing weed control activity from those that
did not. The results for K 5 3 clusters are shown in Fig.
3 where the isolated Cluster 3 contains those accessions
that so far have shown activity. The other two clusters
represent more of a division of a cloud of data points
rather than two well-separated groups. The results for
K 5 4, 5, and 6 clusters showed further division of the
latter into smaller, relatively nondistinct groups. The
isolated Cluster 3 containing the accessions showing
activity remained intact through K 5 6 clusters. For
K 5 7 clusters, the cluster split, but the two newly
formed clusters were not well-separated. Hence, three
Fig. 3. Two-dimensional representation of 3 clusters where each point is determined by 20 chromatographic peaks. Cluster 3 contains the
accessions showing weed control activity.
MATTICE ET AL.: INHIBITION OF BARNYARDGRASS GROWTH WITH RICE USING HPLC
clusters appear to be sufficient to separate these 40
accessions into those showing activity and those that
do not.
For our accessions, we have determined that those
with known activity are in a cluster by themselves. However, for any given set of accessions, there is no guarantee that the separation would be so clearly defined because the range of activity may not be as great as was
found in our group. Moreover, if there is more than one
mechanism for growth inhibition, the set of defining
chromatographic peaks would not necessarily be the
same as ours.
The next step in our research is to determine which
peaks, and ultimately which compounds, are primarily
responsible for the clustering. This step must necessarily
consider peak heights as well as missing peaks.
It is important to remember that correlation does not
imply causality, and we do not imply that the compounds
producing the larger peaks in the PI 312777 extract
would be allelochemicals; they would, however, be candidates for identification and testing. Differences in the
peak size, regardless of which chromatogram it is in,
may be useful in differentiating accessions according to
their ability to inhibit barnyardgrass growth. Whether
or not the compounds are allelochemicals, and whether
or not the observed effect is allelopathy or competition,
the procedure shows promise for predicting which accessions are likely to show a weed control effect toward
barnyardgrass and perhaps other weed species. The procedure allows assaying of 7- to 10-d-old samples, so
screening can be done on a series of samples during
late summer through early spring to identify promising
accessions to take to the field for further testing. This
meets the objectives of being accomplished in a relatively short period of time (#10 d), using a minimum
amount of space (≈1 m2 for 30 accessions), and can be
done year-round. High-performance liquid chromatog-
11
raphy is not an inexpensive technique but is widely
available.
Because the procedure requires only 10 mg of tissue
per milliliter of methanol, it can be done in a nondestructive manner on a rice plant. It remains to be seen if the
procedure could be used to identify which plants within
a cross between two accessions would be most likely to
have the highest weed control activity, and would thus
be the most useful to breeders.
ACKNOWLEDGMENTS
We acknowledge the contribution of the USDA, Arkansas
Rice Research and Promotion Board, and IRRI for support
of this project.
REFERENCES
Dilday, R.H., P. Nastasi, J. Lin, and R.J. Smith Jr. 1991. Allelopathic
activity in rice (Oryza sativa L.) against ducksalad [Heteranthera
limosa (Sw) Willd.]. p. 193–201. In J.N. Hanson et al. (ed.) Proc.
Sustainable Agric. for the Great Plains, Beltsville, MD. USDAARS.
Dilday, R.H., P. Nastasi, and R.J. Smith Jr. 1989. Allelopathic observation in rice (Oryza sativa L.) to ducksalad (Heteranthera limosa ).
Proc. Arkansas Acad. Sci. 43:21–22.
Hand, D.J. 1981. Discrimination and classification. John Wiley &
Sons, New York.
Hasan, S.M., I.R. Aidy, A.O. Bastawisi, and A.E. Draz. 1998. Weed
management using allelopathic rice varieties in Egypt. p. 27–38.
In M. Olofsdotter (ed.) Allelopathy in rice. Proc. Workshop on
Allelopathy in Rice, Manila, Philippines. 25–27 Nov. 1996. IRRI,
Makati City, Philippines.
Kim, K.U., and D.H. Shin. 1998. Rice allelopathy research in Korea.
p. 39–44. In M. Olofsdotter (ed.) Allelopathy in rice. Proc. Workshop on Allelopathy in Rice, Manila, Philippines. 25–27 Nov. 1996.
IRRI, Makati City, Philippines.
Krzanowski, W.J., and F.H.C. Marriott. 1995. Multivariate analysis.
Part 1. Distributions, ordination, and inference. Edward Arnold,
London.
Navarez, D.C., and M. Olofsdotter. 1996. Relay seeding technique
for screening allelopathic rice (Oryza sativa ). p. 1285–1290. In Proc.
Int. Weed Control Congr., 2nd, Copenhagen, Denmark. 25–28
June 1996.