Breeding Science 59: 87–93 (2009)

Note

Variation in root morphology and anatomy among accessions of cultivated rice

(Oryza sativa L.) with different genetic backgrounds

Yusaku Uga*1), Kaworu Ebana1), Jun Abe2), Shigenori Morita3), Kazutoshi Okuno1,4) and Masahiro Yano1)
1) National Institute of Agrobiological Sciences, 2-1-2 Kannondai, Tsukuba, Ibaraki 305-8602, Japan
2) AE-Bio, Graduate School of Agricultural and Life Sciences, The University of Tokyo, Bunkyo, Tokyo 113-8657, Japan
3) Field Production Science Center, Graduate School of Agricultural and Life Sciences, The University of Tokyo, Nishitokyo, Tokyo 188-
0002, Japan
4) Present address: University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8572, Japan

Asian cultivated rice (Oryza sativa L.) has genetic diversification of root characteristics, but this variation
has not been elucidated fully with reference to the genetic background. To clarify the differences in root
anatomical and morphological traits among different varietal groups of cultivated rice, we analyzed four
anatomical traits (root thickness, stele transversal area, total transversal area and number of late metaxylem
vessels) and two morphological traits (root length index and ratio of deep rooting) in 59 accessions. A previous
principal-coordinate analysis study using data on 179 restriction fragment length polymorphisms classified
these accessions into three varietal groups: 13 japonica, 21 indica-I, and 25 indica-II. Based on a principal-
components analysis of the six traits, the japonica group had wide variation in root anatomy compared to
the two indica groups. In particular, japonica upland rice was characterized by a larger stele and xylem struc-
tures. By contrast, the two indica groups had wide variation in root morphology compared to the japonica
group. Of the two indica groups, on average, the indica-I accessions had deeper, thicker roots than the indica-
II accessions. Our results demonstrate that the japonica and indica groups contain different genetic diversity
with respect to their root characteristics.

Key Words: genetic variation, indica, japonica, root anatomy, root morphology.

Plant roots are crucial for the absorption and translocation
of water and nutrients. For this reason, root anatomical and
morphological traits have been well studied in rice (re-
viewed by Morita and Nemoto 1995). Varietal differences in
root morphological characteristics such as length and thick-
ness have been reported in cultivated rice (Oryza sativa L.)
in previous studies (reviewed by O’Toole and Bland 1987).
In general, the roots of upland rice cultivars are thicker and
penetrate more deeply into the soil than those of lowland
cultivars (O’Toole and Bland 1987). Root distribution has
also been quantitatively characterized by using several traits,
including root length, volume, and density in the soil at dif-
ferent depths, and these characteristics differed among culti-
vars (Morita et al. 1995, Nemoto et al. 1998, Hirayama et al.
2007, Kato et al. 2007). However, variations in root charac-
teristics have not been fully elucidated with reference to
their genetic background in cultivated rice, because only a
few cultivars or accessions with unidentified genetic back-
ground were used in the abovementioned studies. Under-
Communicated by H. Kato
Received September 3, 2008. Accepted December 12, 2008.
*Corresponding author (e-mail: yuga@affrc.go.jp)
standing the variations of root characteristics in rice germ-
plasm collections will be essential for genetic improvement
of their root systems. In particular, to facilitate the selection
of materials for breeding or genetic analysis, it is important
to know whether the different varietal groups of rice differ in
their root characteristics.
Asian cultivated rice has been classified into indica and
japonica types based on traits for discrimination, such as
KClO3 resistance, cold or drought sensitivity and apiculus
hair (Oka 1953), and isozyme markers (Glaszmann 1987, Li
and Rutger 2000) and various molecular markers (Wang and
Tanksley 1989, Dally and Second 1990, Zhang et al. 1992).
Recently, Kojima et al. (2005) made a collection consisting
of 69 cultivated rice accessions called rice diversity research
set of germplasm (RDRS) from 332 accessions maintained
at Genebank of the National Institute of Agrobiological Sci-
ences (NIAS) based on genome-wide RFLP polymorphism
survey. The RDRS were classified into three varietal groups
(japonica, and two indica groups designated as indica-I and
indica-II in this paper), based on principal coordinate analy-
sis using 179 RFLP data (Kojima et al. 2005). The RDRS
holds 91% of the alleles identified in the original 332 acces-
sions. In addition, most of the range of variation in several
 88 Uga, Ebana, Abe, Morita, Okuno and Yano
agro-morphological traits from the original 332 accessions is
covered by the RDRS (Kojima et al. 2005), making the
RDRS a powerful research tool for discovering potential
variation in root characteristics and other traits of rice. Our
objective was to detect differences in root anatomical and
morphological traits among the three varietal groups of cul-
tivated rice. To do so, we quantitatively analyzed 59 acces-
sions, including 13 japonica, 21 indica-I, and 25 indica-II,
with respect to six anatomical and morphological traits
(Table 1). Data could not be obtained for all the root traits of
10 accessions from the RDRS collection because of disease
or insect injury, so these accessions were omitted from the
statistical analyses. All experiments were conducted under
rainfed conditions in an upland field at NIAS (36°1′N,
140°6′E). The soil at the experimental site is a volcanic ash
soil of the Kanto loam type (Humic Andosol). The topsoil
(0–30 cm) is a dark humic silty loam (pH = 6.2). The subsoil
(below 30 cm) is a red-brown silty clay loam (pH = 5.8).
Chemical fertilizers were applied at a sowing rate of N, P, K
= 26 kg, 36 kg, 28 kg ha−1 in all experiments. Weeds were
controlled by hand weeding and herbicide application.
Root thickness and structures of the stele and xylem
We examined the anatomical traits of roots from the main
culm of the plant according to the procedure of Uga et al.
(2008). We used three plants of each accession, grown at a
spacing of 20 × 40 cm for 6 weeks in the summer of 2004, to
obtain cross-sectional images from the third-highest rooting
node of the main stem. Terashima et al. (1987) reported that
roots selected from the third-highest node could properly
evaluate variations of root thickness and vascular system in
different varieties, and were therefore used in this study. We
stained the sections with Toluidine Blue O to facilitate mea-
surement of the following anatomical traits—root thickness
(RTH), stele transversal area (STA), and total transversal
area and number of late metaxylem vessel (MXVII) (MXA
and MXN, respectively)—under a digital microscope
(VHX-200; Keyence Co., Ltd., Osaka, Japan). We used 9
images (3 plants × 3 nodal roots) to calculate the mean value
for each accession.
Phenotypic variations in four anatomical traits of roots in
RDRS accessions are shown in Table 1 and Fig. 1. Keiboba
had the thickest roots (RTH = 1155.7 μm), and this value was
1.8 times the value for Rambhog, which had the thinnest
roots (647.3 μm). Despite this wide range of RTH values, we
found no statistically significant differences in mean RTH
among the three varietal groups. Variation in the RTH of the
japonica accessions was thus similar to that in the two indica
groups, indicating that variation in RTH is not associated
with the differentiation of cultivated rice into these varietal
groups. On the other hand, we found significantly higher
mean STA and mean MXA in the japonica accessions than
in the indica-II accessions. Mean MXN, however, showed
no significant differences among the three varietal groups.
The japonica accession Khao Nok exhibited the largest STA
(88 155.4 μm2), which was 5.4 times the corresponding value
for Rambhog, the indica-II accession with the smallest STA
(16 342.9 μm2). Khao Nok also showed the largest MXA
(21 245.1 μm2) and MXN (6.4), and Rambhog had the smallest
values (2591 μm2 and 2.6, respectively) among the acces-
sions. Only three japonica accessions (Khao Nok, Khau
Mac Kho, and Phulba) had STA greater than 70 000 μm2.
These cultivars were upland rice according to passport data
in the NIAS Genebank. Kondo et al. (2000) compared the
anatomy of nodal roots of five japonica and seven indica
cultivars classified by Glaszmann (1987). Among them,
three traditional japonica upland cultivars (Azucena,
Kinandang Patong, and Moroberekan) had the largest stele
and MXVII diameters. We used the results for Azucena and
Kinandang Patong as reference values here. The abovemen-
tioned three japonica accessions had larger anatomical val-
ues than those of the two reference cultivars. Our results
demonstrated that japonica upland rice was characterized by
a larger stele and xylem, and would thus be useful breeding
material for anatomical improvement of rice roots.
Root length index
Because it is necessary to develop an efficient method for
measuring the root length of rice plants grown in the field (a
laborious and time-consuming task), we established a root
length index as a criterion to evaluate root length in refer-
ence to a root depth index developed by Oyanagi (1998). To
calculate this index, polyvinyl pipes (50 cm long by 7 cm in
diameter) were buried at a spacing of 40 × 40 cm in the study
field before sowing. The pipes were inserted vertically into
the soil, with care taken not to compact the soil contained
within the pipe, and a rice seed was sown in each pipe. Three
rice plants in each accession were grown for 7 weeks in the
summer of 2005. Rice plants were pulled out along with the
pipe in which they were planted, and the root system was
washed out from the pipe. The roots of each plant were then
divided into horizontal sections at intervals of 10 cm from
the basal part of root (0 to 10 cm, 10 to 20 cm, 20 to 30 cm,
30 to 40 cm, and 40 to 50 cm), and the root samples in each
horizon distance category were oven-dried at 80°C for 3
days. The proportions of total root dry weight in each 10-cm
interval were then calculated. These percentage values were
multiplied by the midpoint depth for each sampling interval
(i.e., 5, 15, 25, 35, and 45 cm). The resulting five values were
then added to obtain the root length index (RLI); the larger
the RLI value, the longer the root system was in total. This
index calculated based on root dry weight may be affected to
varying degrees by differences in the ratio of root compo-
nents such as seminal, nodal and lateral roots in each sam-
pling interval;. however, we preliminarily investigated the
reliability of RLI in several accessions and those with long
root systems tended to show large RLI (data not shown);
therefore, we decided that RLI could identify differences in
root length among different varietal groups.
The mean RLI differed significantly between the indica-I
and indica-II groups. On average, the indica-I accessions from
South Asia had longer roots than the indica-II accessions
 Genetic variation of root traits in cultivated rice 89

Table 1. Values of the six root traits for the 59 rice cultivars

Variety RTH (μm) STA (μm2) MXA (μm2) MXN RLI (cm) RDR (%)
Name Origin
group Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D.
Tupa729 Bangladesh japonica 652.2±145.9 16646.5±5478.0 2956.7±925.8 3.8±0.2 6.7±0.5 14.9±1.1
Tima Bhutan japonica 851.6±31.2 22361.2±2040.7 4302.8±334.3 4.1±0.2 7.5±2.4 13.9±2.6
Jaguary Brazil japonica 858.9±62.0 36750.7±3959.9 9555.6±756.9 4.3±0.3 7.3±0.7 24.4±2.5
Dianyu 1 China japonica 759.2±167.1 24767.7±8374.5 5089.5±1429.1 3.2±0.2 6.2±1.0 31.8±5.7
Phulba India japonica 1010.2±86.8 72387.3±9124.1 14774.7±1974.8 5.9±0.4 14.0±1.3 22.3±4.8
Padi Perak Indonesia japonica 986.4±62.9 52387.7±4543.7 11976.4±1107.2 5.9±0.5 9.3±3.0 32.6±9.2
Nipponbare Japan japonica 671.1±66.3 19567.4±4076.1 4235.9±802.7 3.2±0.2 7.4±1.9 15.3±4.1
Urasan 1 Japan japonica 856.6±58.4 66926.0±9692.0 15246.7±2917.5 5.9±0.5 9.9±1.0 46.3±2.3
Khao Nok Laos japonica 1101.8±113.1 88155.4±22297.8 21245.1±5469.5 6.4±0.8 11.6±2.5 39.3±3.4
Ma sho Myanmar japonica 774.2±76.6 50130.2±8715.5 11280.1±2222.9 5.5±0.2 12.2±3.2 25.4±2.5
Rexmont USA japonica 915.4±33.6 47711.7±4744.1 10725.1±1656.0 5.3±0.3 9.1±1.0 23.3±4.8
Khau Mac Kho Vietnam japonica 1080.1±64.8 87082.9±7041.8 18013.9±1484.8 5.6±0.3 10.6±1.3 26.1±7.2
Khau Tan Chiem Vietnam japonica 753.1±73.3 25100.8±3962.3 4786.0±661.0 4.4±0.4 7.7±1.1 22.5±4.4
Shoni Bangladesh indica-I 888.4±51.6 54013.4±4706.0 11220.1±951.6 5.9±0.2 15.1±1.6 64.2±16.9
Tupa 121-3 Bangladesh indica-I 957.3±49.0 44896.6±4779.2 10248.0±1190.7 5.6±0.8 9.7±1.4 52.8±3.8
Jarjan Bhutan indica-I 873.2±30.5 38607.9±1287.4 8677.5±381.1 5.3±0.3 10.8±1.9 27.4±1.2
ARC 5955 India indica-I 866.3±90.3 27623.3±1971.0 5271.6±606.9 4.4±0.3 8.3±1.8 11.6±1.1
ARC 7047 India indica-I 910.7±52.2 30427.2±2212.3 5836.1±241.4 4.7±0.3 9.7±0.2 37.5±5.2
ARC 7291 India indica-I 882.3±10.6 39643.1±3921.3 10331.1±1489.4 5.3±0.9 9.6±2.2 55.7±8.6
Jhona 2 India indica-I 971.2±67.2 42536.7±4927.2 9199.2±1090.2 5.2±0.3 11.7±2.9 29.0±3.5
Kasalath India indica-I 924.0±27.3 30919.5±3846.9 6953.0±728.8 4.7±0.0 8.3±3.0 20.0±6.9
Local Basmati India indica-I 772.7±80.6 24215.6±4941.3 4672.3±901.2 4.6±0.7 10.9±1.9 32.6±5.8
Nepal 555 India indica-I 940.6±90.0 41954.0±6207.2 7883.4±849.2 4.6±0.1 10.8±0.8 43.3±8.2
Ratul India indica-I 882.7±111.3 37030.2±9756.8 7824.0±1801.4 5.2±0.2 8.3±1.1 37.8±1.5
Surjamukhi India indica-I 759.1±93.4 35235.8±8793.8 8134.7±1777.2 5.2±0.4 13.3±0.8 57.2±5.2
Muha Indonesia indica-I 985.3±130.7 41418.9±12562.1 8932.7±2074.8 4.7±0.7 10.4±1.2 29.8±11.3
Anjana Dhan Nepal indica-I 962.1±106.6 39692.0±10231.9 9243.4±2700.1 5.3±0.3 8.7±0.2 36.1±8.5
Badari Dhan Nepal indica-I 951.3±81.3 37592.1±6591.9 7800.0±1121.0 4.4±0.2 8.6±1.7 32.0±6.9
Jena 035 Nepal indica-I 1061.1±67.5 50958.9±5910.6 11027.1±1402.0 5.3±0.8 9.9±0.7 15.9±4.7
Kalo Dhan Nepal indica-I 940.2±43.5 45963.4±4468.2 9989.4±833.9 5.3±0.6 12.7±2.2 17.4±0.7
Nepal 8 Nepal indica-I 1038.0±48.1 55043.9±7132.2 12365.9±2206.5 6.0±0.6 11.3±1.1 29.1±5.6
Basilanon Philippines indica-I 772.7±41.5 22533.2±2413.8 3907.2±412.2 3.3±0.0 12.1±2.5 22.9±5.8
Calotoc Philippines indica-I 787.7±42.0 30889.1±2146.9 7583.1±847.3 4.0±0.5 7.6±1.2 12.3±2.3
Kaluheenati Sri Lanka indica-I 841.1±96.5 35274.9±4875.8 8352.9±1160.4 4.8±0.8 11.2±1.7 48.9±11.5
Asu Bhutan indica-II 819.3±94.3 26738.7±1355.8 5757.0±462.4 4.7±0.0 10.9±0.2 33.6±4.9
Bei Khe Cambodia indica-II 655.8±54.2 19437.3±2259.5 4520.4±590.3 4.7±0.6 6.3±0.6 25.9±3.9
Neang Menh Cambodia indica-II 788.0±44.3 32215.3±2501.7 8056.2±22.9 5.5±0.7 8.6±2.0 37.2±11.0
Neang Phtong Cambodia indica-II 1133.7±64.6 49455.2±5862.2 11349.5±1032.1 5.0±0.0 12.0±1.1 43.3±6.1
Deng Pao Zhai China indica-II 856.9±89.3 26288.6±5224.6 5039.8±1088.7 3.8±0.4 7.4±1.3 15.5±3.9
Jinguoin China indica-II 830.8±48.7 29588.6±2609.5 6101.4±672.8 4.1±0.2 8.9±1.3 38.7±9.6
Keiboba China indica-II 1155.7±101.2 54647.5±4792.2 11749.5±612.0 5.7±0.3 7.6±0.8 14.7±4.1
Qingyu (Seiyu) China indica-II 870.4±104.9 30479.4±5762.2 6375.1±858.9 5.1±0.7 8.5±2.4 10.9±4.6
Ling Shui Chan Hong Mi China indica-II 800.0±62.4 24567.9±3970.4 5378.7±773.0 4.1±0.4 8.4±1.5 45.2±3.4
Qiu Zao Zhong China indica-II 818.6±66.4 26501.2±4509.1 5357.7±699.5 4.5±0.2 8.7±1.6 20.6±4.7
Co 13 India indica-II 774.4±135.8 25063.3±7357.1 4557.9±1462.1 4.4±0.2 10.8±3.1 42.3±5.0
Naba India indica-II 860.1±55.3 29554.7±2096.9 6637.5±1008.6 4.6±0.4 7.6±0.3 41.7±6.3
Rambhog India indica-II 647.3±169.2 16342.9±7986.2 2591.0±659.7 2.6±0.2 8.4±0.6 48.3±10.7
Puluik Arang Indonesia indica-II 990.7±132.0 37262.9±7203.2 7775.2±1605.3 4.4±0.8 7.6±0.4 24.5±4.3
Milyang 23 Korea indica-II 886.4±167.9 28890.8±7074.4 5563.4±904.9 4.3±0.3 9.0±2.4 48.1±4.4
Vary Futsi Madagascar indica-II 737.9±77.2 26732.7±6845.9 5638.8±1591.4 3.8±0.8 7.7±1.3 20.1±2.2
Kemasin Malaysia indica-II 1060.8±29.7 47872.5±2947.4 11985.6±969.2 5.9±0.4 7.9±0.7 11.5±2.4
Radin Goi Sesat Malaysia indica-II 978.4±153.1 37550.8±5876.9 8506.6±1509.1 5.4±0.5 8.8±2.1 12.8±4.0
Bingala Myanmar indica-II 879.1±5.3 35892.7±922.2 7947.8±459.9 5.7±0.6 9.4±1.5 26.3±4.3
Shwe Nang Gyi Myanmar indica-II 875.3±14.2 41734.2±883.6 8908.6±402.2 5.2±0.3 12.0±2.5 19.4±0.9
Davao 1 Philippines indica-II 774.7±43.4 23605.7±2798.7 4988.8±642.8 4.0±0.0 9.4±0.8 31.1±2.7
IR 58 Philippines indica-II 732.2±80.8 21455.3±4322.5 4466.0±809.4 4.3±0.3 7.2±1.6 12.5±0.1
Pinulupot 1 Philippines indica-II 736.7±93.6 25947.2±1246.2 5456.4±131.1 4.7±0.6 6.1±0.5 12.9±2.6
Tadukan Philippines indica-II 806.4±62.5 23381.4±3247.0 5247.2±607.7 4.1±0.4 10.4±0.5 9.7±3.7
Bleiyo Thailand indica-II 1030.1±75.1 56060.7±5003.9 14142.0±1754.8 5.9±0.4 11.1±0.9 11.1±4.6
RTH, root thickness; STA, stele transversal area; MXA, MXVII transversal area; MXN, Number of MXVII; RLI, root length index; RDR, ratio of
deep rooting.
 90 Uga, Ebana, Abe, Morita, Okuno and Yano

from East and Southeast Asia. Shoni, the indica-I acces-
sion with the highest RLI value (15.1 cm), had an RLI value
2.5 times that of Pinulupot 1, the indica-II accession with
the lowest value (6.1 cm). Phulba, a japonica cultivar that
originated in India, had the largest RLI value (14.0) among
the japonica varieties. These results suggest that many
accessions from South Asia have longer roots than acces-
sions from other regions.
Ratio of deep rooting
We measured the average growth angles of the nodal
roots by using the basket method (Oyanagi et al. 1993). Kato
et al. (2006) reported that root growth angle was associated
with deep root development on upland fields; therefore, this
method was used as an index of deep rooting. In this ap-
proach, a plastic mesh basket (with open top and bottom di-
ameters of 15 and 8.5 cm, respectively, and a height of 6 cm)
was filled with upland-field topsoil mixed evenly with fertil-
izer. The baskets were buried at a spacing of 40 × 40 cm in

Fig. 1. Frequency distributions for the six root traits among the three varietal groups in the RDRS. A, japonica (n = 13); B, indica-I (n = 21);
C, indica-II (n = 25). White inverted triangles indicate the mean value of each varietal group. Triangles labeled with the same letter do not differ
significantly (5% level, Tukey’s multiple-comparison test).
 Genetic variation of root traits in cultivated rice
the upland field. Three plants from each accession were
grown in separate baskets for 7 weeks in the summer of
2007. The growth direction of each nodal root was evaluated
on the basis of the position where the root elongated across
the basket mesh (Oyanagi et al. 1993). These positions were
classified into upper and lower parts of the basket according
to the horizontal angle with respect to the ground surface (0
to 53° and 53 to 90°). This classification was chosen on the
basis of the report by Kato et al. (2006) on the frequency of
higher root-growth angles (50° to 90°). We defined the ratio
of deep rooting (RDR) as the proportion of the total number
of roots penetrating the lower part of the mesh (i.e., the flat
8.5-cm-wide bottom of the basket) to total number of roots
penetrating the whole mesh.
We found no significant differences in mean RDR among
the three varietal groups. Four accessions, however, exhibit-
ed an RDR value greater than 50% in the indica-I group
(Shoni, Tupa 121-3, ARC 7291, and Surjamukhi). This sug-
gests that the indica-I group has wider variation in the pro-
portion of deeper roots than did the other groups. Shoni, the
indica-I accession with the highest RDR (64.2%), had a val-
ue 6.6 times that of Tadukan, the indica-II accession with the
lowest RDR (9.7%). Shoni also had the largest RLI value.
Therefore, we hypothesize that this cultivar can elongate its
roots into deeper soil layers than can most of the other acces-
sions. Kato et al. (2006) reported that 12 lowland and upland
cultivars had RDR values ranging from 9% to 34%. Thus,
the RDRS contains a wider range of variation in this param-
eter.
Correlations among the six root traits
We correlated the root traits of the 59 RDRS accessions
with each other (Table 2). There were several significant
positive correlations and no significant negative correla-
tions, but the highest correlation was between STA and
MXA (r = 0.982**). Uga et al. (2008) previously reported a
significantly high correlation between STA and MXA (r =
0.923**) in F3 lines derived from across between IR64, low-
land rice, and Kinandang Patong, upland rice. Six quantita-
tive trait loci (QTLs) for STA and MXA were detected on
chromosomes 2, 9, and 10. Of these, five were found in the
same regions of chromosomes 2 and 9. Therefore, the corre-
lation between STA and MXA can be attributed either to a
tight linkage of the QTLs for these traits or to a pleiotropic
effect of one QTL (Uga et al. 2008). These genetic behaviors
may be responsible for the high correlation between the two
traits. The root anatomical traits were not significantly cor-
related with RDR, but RDR was significantly but weakly
correlated with RLI (r = 0.383), suggesting that RDR is con-
trolled by genetic factors different from those that control
root anatomical traits. Kato et al. (2006) reported that nodal
root diameter was closely correlated with RDR, possibly be-
cause they used a limited range of accessions that exhibited
this correlation. Our results, however, showed no significant
overall correlation between root diameter and the ratio of
deep rooting.
91
Table 2. Correlation coefficients among the six root traits for the 59
rice cultivars
RTH STA MXA MXN RLI
STA 0.765**
MXA 0.754** 0.982**
MXN 0.676** 0.789** 0.818**
RLI 0.332* 0.513** 0.471** 0.471**
RDR −0.014 0.132 0.130 0.116 0.383**
* and ** indicate significance at the 5% and 1% levels, respectively.
RTH, root thickness, STA, stele transversal area, MXA, MXVII trans-
versal area; MXN, Number of MXVII, RLI, root length index; RDR,
ratio of deep rooting.
Relationships between root characteristics and the three
varietal groups
To elucidate the relationships among the six root traits,
we examined their correlation matrix by using the principal-
components analysis (PCA) module of JMP version 6.0
(SAS Institute, Cary, NC, USA). We calculated eigenvec-
tors, factor loadings, and principal-component (PC) scores
for the six traits of the 59 accessions on the basis of a corre-
lation matrix (Table 3). The contributions of PC1, PC2, and
PC3 to the multivariate variation were 62.2%, 19.8%, and
8.5%, respectively, indicating that the contribution of PC1
was more than three times that of PC2. The sum of these
components explained more than 90% of the total variation
of each trait, except for RTH (78.4%) and MXN (78.9%). In
the factor loading, PC1 was most strongly correlated with
the four anatomical traits, suggesting that PC1 is a function
of root anatomical size. PC2 was positively correlated with
RDR and RLI, suggesting that PC2 is a function of root ar-
chitecture. In particular, RDR was strongly correlated with
PC2 (0.82).
To analyze the relationship between the root traits and the
three varietal groups, we constructed a scatterplot of all the
accessions based on their PC1 and PC2 scores (Fig. 2). The
PC1 scores of the japonica accessions were scattered more
widely than those of the indica groups, whereas the PC2
scores of indica groups were scattered more widely than
those of the japonica accessions. This suggested that the
japonica and indica groups contain different genetic diversi-
ty with respect to their root characteristics. On the basis of
the abovementioned PCA results, the japonica group had
wide variation in root anatomical traits, whereas the two
indica groups had wide variation in root system architecture.
Among the indica cultivars, the PC1 and PC2 scores of
many accessions belonging to indica-I were found at the
right and upper sides of the ordination, respectively, some-
what separated from the indica-II accessions. This indicates
that, on average, indica-I accessions had deeper and thicker
roots than the indica-II accessions. Most of the indica-I cul-
tivars are distributed over areas geographically different
from those of the indica-II cultivars (Table 1). It is possible
that the differences in root characteristics observed between
 92 Uga, Ebana, Abe, Morita, Okuno and Yano

Table 3. Eigenvectors and factor loadings for the principal components obtained from the six root traits observed in the 59 rice cultivars
Eigenvector Factor loading
Trait υ1 (%)a
PC1 PC2 PC3 PC1 PC2 PC3
Root thickness 0.43 −0.27 0.17 0.83 −0.30 0.12 78.4
Stele transversal area 0.50 −0.09 0.08 0.96 −0.10 0.05 93.0
MXVII transversal area 0.49 −0.11 0.14 0.95 −0.12 0.10 93.5
Number of MXVII 0.46 −0.08 0.04 0.88 −0.08 0.03 78.9
Root length index 0.33 0.48 −0.80 0.63 0.52 −0.57 99.2
Ratio of deep rooting 0.11 0.82 0.56 0.21 0.89 0.40 99.7
Eigenvalue 3.73 1.19 0.51
Contribution (%) 62.2 19.8 8.5
PC1, PC2, and PC3: first, second, and third principal components, respectively.
a contribution of the sum of the three principal components.

the two groups are associated with geographical differentia-
tion. However, we could not explain from our results why
such differentiation of root characteristics occurred in the
indica groups. Further analysis will be needed to clarify
varietal differentiation of root characteristics among the
indica groups.
The japonica accession Nipponbare and the indica-I ac-
cession Kasalath, which were used as reference accessions
for discrimination of japonica and indica accessions in the
RFLP survey (Kojima et al. 2005), were located in the third
quadrant (with small negative values of PC1 and PC2) in
Fig. 2. When both cultivars were used as reference values for
the genetic variations in root characteristics, we found that
the RDRS contained wide genetic diversity with respect to
the root characteristics of these accessions. Such wide varia-
tion may be controlled by many genetic factors. Utilization
of accessions selected from the RDRS on the basis of the re-
sults of this study will thus be useful for genetic analysis or
breeding to improve root characteristics. Many root traits
were affected by environmental effects such as water regime
and soil condition (Kondo et al. 2003, Kato et al. 2006), al-
though Kato et al. (2006) mentioned that environmental ef-
fects on the nodal root growth angle were relatively small.
Our experiments were performed only under rainfed con-
ditions in an upland field; therefore, further study will be
needed to clarify genetic variations in root traits under other
environmental conditions, such as paddy fields.
Acknowledgments
We thank Dr. M. Shimamura (National Agriculture and
Food Research Organization) for his technical advice re-
garding anatomical measurements of rice roots. We also
thank Dr. Y. Kato (University of Tokyo) for his technical ad-
vice regarding the basket method. This work was supported
by a grant from the Ministry of Agriculture, Forestry, and
Fisheries of Japan (Genomics for Agricultural Innovation,
QTL-4003).

Literature Cited
Dally, A.M. and G. Second (1990) Chloroplast DNA diversity in wild
and cultivated species of rice (Genus Oryza, section Oryza).
Cladistic-mutation and genetic-distance analysis. Theor. Appl. Genet.
80: 209–222.
Glaszmann, J.C. (1987) Isozymes and classification of Asian rice vari-
eties. Theor. Appl. Genet. 74: 21–30.
Hirayama, M., H. Nemoto and H. Hirasawa (2007) Relation between
root system and drought resistance in Japanese upland rice (Oryza
sativa L.) varieties with medium to late maturing under field condi-
tions. Jpn. J. Crop Sci. 76: 245–252.
Kato, Y., J. Abe, A. Kamoshita and J. Yamagishi (2006) Genotypic vari-
ation in root growth angle in rice (Oryza sativa L.) and its associa-
tion with deep root development in upland fields with different
water regimes. Plant Soil 287: 117–129.
Fig. 2. Scatterplot of the principal-component scores for the 59 rice Kato, Y., A. Kamoshita, J. Yamagishi, H. Imoto and J. Abe (2007)
accessions in the RDRS. PC1, first principal component; PC2, second Growth of rice (Oryza sativa L.) cultivars under upland conditions
principal component. ×, japonica; , indica-I; , indica-II; 1, with different levels of water supply. Plant Prod. Sci. 10: 3–13.
Nipponbare; 2, Kasalath. Kojima, Y., K. Ebana, S. Fukuoka, T. Nagamine and M. Kawase (2005)
 Genetic variation of root traits in cultivated rice
Development of an RFLP-based rice diversity research set of germ-
plasm. Breed. Sci. 55: 431–440.
Kondo, M., A. Aguilar, J. Abe and S. Morita (2000) Anatomy of nodal
roots in tropical upland and lowland rice varieties. Plant Prod. Sci.
3: 437–445.
Kondo, M., P.P. Pablico, D.V. Aragones, R. Agbisit, J. Abe, S. Morita
and B. Courtois (2003) Genotypic and environmental variations in
root morphology in rice genotypes under upland field conditions.
Plant Soil 255: 189–200.
Li, Z. and J.N. Rutger (2000) Geographic distribution and multilocus
organization of isozyme variation of rice (Oryza sativa L.). Theor.
Appl. Genet. 101: 379–387.
Morita, S. and K. Nemoto (1995) Morphology and anatomy of rice
roots with special reference to coordination in organo- and histo-
genesis. In: Baluška, F. et al. (eds.) Structure and function of roots,
Kluwer Academic Publishers, the Netherlands, pp. 75–86.
Morita, S., S. Yamada and J. Abe (1995) Analysis on root system mor-
phology in rice with reference to varietal differences at ripening
stage. Jpn. J. Crop Sci. 64: 58–65.
Nemoto, H., R. Suga, M. Ishihara and Y. Okutsu (1998) Deep rooted
rice varieties detected through the observation of root characteris-
93
tics using the trench method. Breed. Sci. 48: 321–324.
Oka, H.I. (1953) Phylogenetic differentiation of the cultivated rice
plant. 1. Variation of various characters and character combina-
tions among rice varieties. Jpn. J. Breed. 3: 33–43.
O’Toole, J.C. and W.L. Bland (1987) Genotypic variation in crop plant
root systems. Adv. Agron. 41: 91–143.
Oyanagi, A. (1998) A new approach to crop root systems with quanti-
tative analysis of the root depth. Jpn. J. Crop Sci. 67: 3–10.
Oyanagi, A., T. Nakamoto and M. Wada (1993) Relationship between
root growth angle of seedlings and vertical distribution of roots in
the field in wheat cultivars. Jpn. J. Crop Sci. 62: 565–570.
Terashima, K., H. Hiraoka and I. Nishiyama (1987) Varietal difference
in the root of rice plant I. Varietal difference in the morphology of
crown root. Jpn. J. Crop Sci. 56: 521–529.
Uga, Y., K. Okuno and M. Yano (2008) QTL underlying natural varia-
tion in stele and xylem structures of rice root. Breed. Sci. 58: 7–14.
Wang, Z.Y. and S.D. Tanksley (1989) Restriction fragment length
polymorphism in Oryza sativa L. Genome 32: 1113–1118.
Zhang, Q., M.A.S. Maroof, T.Y. Lu and B.Z. Shen (1992) Genetic di-
versity and differentiation of indica and japonica rice detected by
RFLP analysis. Theor. Appl. Genet. 83: 495–499.