**3.1 Growth characters**

In 2001, the precipitation was 14% lower, the average mean temperature was 0.8 degree higher, and the sunshine hours was 13% longer than the normal year, and it was characterized by low rainfall, high temperature and much sunshine. In 2002, the precipitation was 56% lower, the average mean temperature was 0.9 degree higher, and the sunshine hours was 7% longer than the normal year, and it was characterized by drought, high temperature and much sunshine though lower than in 2001. The field was hit by a typhoon on Aug. 21 in 2001. There was no typhoon damage in 2002.

In both years, the number (per square meter) of nodes on the main stem, racemes with compound leaves and in total was higher, but in the number of branches was lower than in sparse plots (Table 1). The node number on the branches and in total was larger in wide plots than in narrow plots except that in sparse plots in 2001, and also that of racemes with compound leaf in 2001. The main stem length in dense plots was 2-12 cm longer than in sparse plots, and that in narrow plots was 7-16 cm shorter than in wide plots. The weight, diameter and section area of stem were larger than in sparse and narrow plots than dense and wide plots, respectively. The seed/stem weight ratio in dense plots was smaller than in sparse plots among the narrow plots, but not among the wide plots. The ratio in narrow plots was larger than in wide plots among the sparse plots, but not among the dense plots.


Values are means of twelve plants. 'ns' means no siginificant difference at 5% level.

Table 1. Growth characteristics (2001,2002).

#### **3.2 Seed yield and yield components**

In both years, seed yields in dense plots and narrow plots were larger than sparse plots and wide plots, respectively, and those in 2001 were higher than in 2002 because of the much sunshine hours (Fig. 2, Table 2). The highest yield, 668 g m-2, was obtained in narrow/dense plots in 2001. A close correlation (r=0.934, P<0.01) was observed between seed yield and pod

Fig. 2. Effects of planting pattern on seed yield of soybean.

 Wide/Sparse 150 316 137 602 63.4 18.3 66 9.4 53.0 2.20 Wide/ Dense 290 192 183 665 69.6 8.7 74 6.9 30.4 2.51 Narrow/Sparse 141 239 211 591 47.4 18.1 60 9.2 56.4 2.58 Narrow/Dense 296 342 307 944 55.7 12.2 102 7.8 37.1 2.49 LSD(0.05) 9 ns 33 54 3.3 1.4 9 0.4 4.4 ns

Node number (m-2)

with Main

Bran - ch

stem

Table 1. Growth characteristics (2001,2002).

<sup>100</sup> <sup>108</sup> <sup>115</sup>

Fig. 2. Effects of planting pattern on seed yield of soybean.

**3.2 Seed yield and yield components** 

0

100

200

300

400

Seed yield (g

m


500

600

700

800

leaf

 Wide/Sparse 159 272 71 502 61.2 12.5 60 8.5 43.5 1.98 Wide/ Dense 301 248 121 670 63.5 7.7 89 7.0 27.8 2.06 Narrow/Sparse 162 324 89 576 53.6 13.6 70 9.1 49.4 3.39 Narrow/Dense 318 347 122 787 65.3 10.2 111 7.8 38.2 2.25 LSD(0.05) 7 53 24 67 2.5 1.3 14 3.4 0.50 Values are means of twelve plants. 'ns' means no siginificant difference at 5% level.

In both years, seed yields in dense plots and narrow plots were larger than sparse plots and wide plots, respectively, and those in 2001 were higher than in 2002 because of the much sunshine hours (Fig. 2, Table 2). The highest yield, 668 g m-2, was obtained in narrow/dense plots in 2001. A close correlation (r=0.934, P<0.01) was observed between seed yield and pod

150

2001 2002

Sparse Dense Sparse Dense Sparse Dense Sparse Dense

100

129

<sup>170</sup> <sup>183</sup>

Wide row Narrow row Wide row Narrow row

2001

Year / Plot

2002

Main Stem Bran- Stem Stem Seed /

) (mm) (mm<sup>2</sup>

0.1

) ratio

Rac. stem weight ch dia- section stem Total length no. meter area weight

(cm) (g) (m-2

number, indicating that seed yield was determined by the pod number. Seed number per pod and seed setting ratio were not significantly different among plots, and 100 seeds weight in narrow plots tended to be slightly heavier than in wide plots, but the difference was not significant.

The pod number on the main stem relative to the total was higher in dense plots than in sparse plots, and that on the branches was higher in narrow plots than in wide plots (Table 3). The percentage share of basal raceme was higher in 2002 than in 2001. The percentage share of racemes with compound leaves was higher in dense plots than in sparse plots, and was also higher in narrow plots than in wide plots, especially in 2001.


Values are means of twelve plants.

'ns' means no siginificant difference at 5% level.

Table 2. Seed yield and yield components (2001, 2002).


Values are means of twelve plants. 'ns' means no siginificant difference at 5% level. Values in parentheses are relative to total (100).

Table 3. Pod number on main stem or branch and raceme order (2001, 2002).

#### **3.3 Dry weight and leaf area index**

At each growth stage, the dry-weight tended to be heavier in dense plots than in sparse plots, but the difference was not significant (Fig. 3). The dry-weight tended to be heavier in narrow plots than in wide plots except that in sparse plots at 44 days after sowing (DAS) and in dense plots at 65 DAS. At 107 DAS, the dry-weight was heaviest in narrow/dense plots and became lighter in the order of wide/dense plots > narrow/sparse plots > wide/sparse plots.

Fig. 3. Changes in cumulative dry-weight of different plant parts during growth (2001).

The leaf area index (LAI) tended to be larger in dense plots than in sparse plots, and in narrow plots than in wide plots especially at 65 DAS, when LAI in dense plots exceeded 8 (Fig. 4).

#### **3.4 Canopy structure**

At the flowering stage, the higher the canopy layer, the larger the leaf area from 20 to 100 cm above the ground in wide/dense plots, and the larger leaf area was distributed at a 40-100 cm height in narrow/sparse plots (Fig. 5). In dense plots, leaf area was concentrated in the 80-100 cm layer above the ground especially in narrow plots. The total dry-weight of non-assimilative organ was heavier in narrow plots than in wide plots. The light extinction coefficients (k), the lower value indicates that the canopy has a good light-intercepting characteristic, was in the order of narrow/dense (0.60) < wide/dense (0.68) < narrow/sparse (0.73) < wide/sparse (0.81). It was clear that the light penetrated into a deeper layer of the canopy when planted dense and narrow row-spacing. The order of k at the seed growth stage coincided with that at the flowering stage (data not shown).

Fig. 4. Changes in LAI during growth (2001).

At each growth stage, the dry-weight tended to be heavier in dense plots than in sparse plots, but the difference was not significant (Fig. 3). The dry-weight tended to be heavier in narrow plots than in wide plots except that in sparse plots at 44 days after sowing (DAS) and in dense plots at 65 DAS. At 107 DAS, the dry-weight was heaviest in narrow/dense plots and became lighter in the order of wide/dense plots > narrow/sparse plots >

**3.3 Dry weight and leaf area index** 

wide/sparse plots.

0

Sparse

**3.4 Canopy structure** 

Dense

the flowering stage (data not shown).

Sparse

44 DAS

Dense

Dead leaf Flower & pod

Stem & petiole

Leaf

Sparse

Dense

Sparse

Dense

Fig. 3. Changes in cumulative dry-weight of different plant parts during growth (2001).

The leaf area index (LAI) tended to be larger in dense plots than in sparse plots, and in narrow plots than in wide plots especially at 65 DAS, when LAI in dense plots exceeded 8

At the flowering stage, the higher the canopy layer, the larger the leaf area from 20 to 100 cm above the ground in wide/dense plots, and the larger leaf area was distributed at a 40-100 cm height in narrow/sparse plots (Fig. 5). In dense plots, leaf area was concentrated in the 80-100 cm layer above the ground especially in narrow plots. The total dry-weight of non-assimilative organ was heavier in narrow plots than in wide plots. The light extinction coefficients (k), the lower value indicates that the canopy has a good light-intercepting characteristic, was in the order of narrow/dense (0.60) < wide/dense (0.68) < narrow/sparse (0.73) < wide/sparse (0.81). It was clear that the light penetrated into a deeper layer of the canopy when planted dense and narrow row-spacing. The order of k at the seed growth stage coincided with that at

Sparse

Wide Narrow Wide Narrow Wide Narrow Wide Narrow

Dense

Sparse

65 DAS 86 DAS 107 DAS

Dense

Sparse

Dense

Sparse

Dense

200

400

600

Dry weight (g

(Fig. 4).

m


800

1000

Fig. 5. Canopy structures at the full-flowering stage (2001).

#### **3.5 Diurnal change in canopy light extinction coefficient (k)**

The k-value measured under direct sunlight was higher in the morning and evening, and decreased during the daytime (Fig. 6). The k-values in the morning and evening were similar to those measured under diffuse light (Fig. 5), which were lower in dense and narrow row plots. At midday, k showed the lowest value in wide plots, which suggested that the direct sunlight reached the furrow surface in the non-closed canopy in wide row plots. The extent of variation during the daytime was small in narrow plots due to the closed canopy.

Fig. 6. Diurnal change in canopy light extinction coefficient at the beginning of the flower stage (2001).

#### **3.6 Distribution of cumulative solar radiation at each height within canopy**

The cumulative solar radiation at every height was lower in dense plots than in sparse plots, and was lower near the row (plant) and higher at the furrow in a direction perpendicular to the row (Fig. 7). In narrow row plots, the cumulative solar radiation was lower in dense plots than in sparse plots, and the difference between that on the row and furrow was small.

#### **3.7 Changes in lodging score**

In 2002, lodging did not occur in any plot. In 2001, the lodging score increased in narrow/sparse plots at 34 DAS due to a rainstorm, followed by the gradual increase in wide/sparse plots, and was larger in narrow row plots than in wide row plots (Fig. 8). At 71 DAS, when a typhoon hit, the lodging score increased markedly in dense plots, and was slightly larger in narrow/dense plots than in wide/dense plots. After lodging, plants could not recover during the later growth period.

The k-value measured under direct sunlight was higher in the morning and evening, and decreased during the daytime (Fig. 6). The k-values in the morning and evening were similar to those measured under diffuse light (Fig. 5), which were lower in dense and narrow row plots. At midday, k showed the lowest value in wide plots, which suggested that the direct sunlight reached the furrow surface in the non-closed canopy in wide row plots. The extent of variation during the daytime was small in narrow plots due to the

> 7:00 9:30 12:00 14:30 17:00 Japanese standard time (JST)

Fig. 6. Diurnal change in canopy light extinction coefficient at the beginning of the flower

The cumulative solar radiation at every height was lower in dense plots than in sparse plots, and was lower near the row (plant) and higher at the furrow in a direction perpendicular to the row (Fig. 7). In narrow row plots, the cumulative solar radiation was lower in dense plots than in sparse plots, and the difference between that on the row and furrow was

In 2002, lodging did not occur in any plot. In 2001, the lodging score increased in narrow/sparse plots at 34 DAS due to a rainstorm, followed by the gradual increase in wide/sparse plots, and was larger in narrow row plots than in wide row plots (Fig. 8). At 71 DAS, when a typhoon hit, the lodging score increased markedly in dense plots, and was slightly larger in narrow/dense plots than in wide/dense plots. After lodging, plants could

**3.6 Distribution of cumulative solar radiation at each height within canopy** 

Wide/Sparse Wide/Dense Narrow/Sparse Narrow/Dense

**3.5 Diurnal change in canopy light extinction coefficient (k)** 

closed canopy.

0

**3.7 Changes in lodging score** 

not recover during the later growth period.

0.2

0.4

Canopy light extinction coefficient (k)

stage (2001).

small.

0.6

0.8

Fig. 7. Distribution of cumulative solar radiation at each height within canopy in a direction perpendicular to the row at the beginning flower stage (2001).

Fig. 8. Changes in lodging score (2001).

#### **3.8 Weed emergence**

More weed plants appeared in 2002 than in 2001. *Portulaca* and *Cyperus* species were dominant in 2001, and *Digitaria* and *Galinsoga* in 2002. In both years, there were fewer emerged weeds in narrow plots than in wide plots.


Values indicate the number of weed plants. Average of three quadrats (80cm \* 60cm) .

Table 4. Emergence of weeds at the beginning of flowering of soybean.

#### **4. Discussion**

In soybean, dense planting has been reported to increase the node number, pod number and therefore seed yield without the consideration of lodging (Nakaseko and Goto 1975, Costa et al. 1980, Miura et al. 1987, Saitoh et al. 1998a). The square- or triangular-shape planting increased the space occupied by plants than rectangular-shape planting, and promoted the development of branches, thus increasing the seed yield (Cooper 1977, Costa et al. 1980, Duncan 1986, Miura and Gemma 1986, Miura et al. 1987, Board et al. 1990b, Ikeda 2000). Nakano et al. (2001) also reported that planting pattern affected the light environment within the canopy, which determined the branch node number, pod number and seed yield. In the present study, the seed yield was in the order of narrow/dense > narrow/sparse > wide/dense > wide/sparse (Table 2, Fig. 2), and the yield increase in narrow row planting was due to the yield increase on the branches especially on the raceme with compound leaves (Table 3).

The raceme with compound leaves is morphologically the same as a branch. The branch differentiates on the leaf axil just above the petiole on the main stem, and the raceme with compound leaves differentiates on the left and right axils of the basal raceme in the upper node of the main stem and branches, and develops a stem with one to four leaves. In a previous study, the differentiated racemes developed compound leaves when assimilates were supplied to the raceme (Saitoh et al. 2001). In the present two- year study, seed yield was positively correlated with total pod number (r=0.934, P<0.001) and pod number on racemes with compound leaves (r=0.864, P<0.01). Thus the increase in the pod number on the raceme with compound leaves contributed to the increase in seed yield.

The longer sunshine hours accelerated the source activity and increased assimilates were supplied to the axil of each node. Our three-year planting density experiment showed that the number of floral buds on racemes with compound leaves increased markedly in the year with longer sunshine hours (Saitoh et al. 1998a), and the pod number on racemes with compound leaves increased especially when the twelfth node was isolated by pruning the

More weed plants appeared in 2002 than in 2001. *Portulaca* and *Cyperus* species were dominant in 2001, and *Digitaria* and *Galinsoga* in 2002. In both years, there were fewer

> *Amaranthus Portulaca Digitaria Cyperus Rorippa Galinsoga Setaria Chenopodium Euphorbia Mollugo viridis oleracea ciliaris indica ciliata viridi album supina pentaphylla*

Year/Plot Total

 Wide/Sparse 7.6 9.7 - 8.3 2.1 2.1 2.1 2.1 - - 34.0 Wide/ Dense 5.2 16.0 - 7.6 2.1 2.1 2.1 9.0 - 3.1 47.2 Narrow/Sparse - 2.1 - - - - - - - - 2.1 Narrow/Dense - 14.6 - - - 2.1 2.1 2.1 - - 20.8

 Wide/Sparse - 52.8 11.1 - - 60.2 - - 28.7 - 124.1 Wide/ Dense - - 23.1 - - 94.4 - - - - 117.6 Narrow/Sparse - 63.9 9.3 - - 11.1 3.7 - 25.0 - 78.7 Narrow/Dense - - 18.5 - - 45.4 4.6 - - - 68.5

In soybean, dense planting has been reported to increase the node number, pod number and therefore seed yield without the consideration of lodging (Nakaseko and Goto 1975, Costa et al. 1980, Miura et al. 1987, Saitoh et al. 1998a). The square- or triangular-shape planting increased the space occupied by plants than rectangular-shape planting, and promoted the development of branches, thus increasing the seed yield (Cooper 1977, Costa et al. 1980, Duncan 1986, Miura and Gemma 1986, Miura et al. 1987, Board et al. 1990b, Ikeda 2000). Nakano et al. (2001) also reported that planting pattern affected the light environment within the canopy, which determined the branch node number, pod number and seed yield. In the present study, the seed yield was in the order of narrow/dense > narrow/sparse > wide/dense > wide/sparse (Table 2, Fig. 2), and the yield increase in narrow row planting was due to the yield increase on the branches especially on the raceme with compound

The raceme with compound leaves is morphologically the same as a branch. The branch differentiates on the leaf axil just above the petiole on the main stem, and the raceme with compound leaves differentiates on the left and right axils of the basal raceme in the upper node of the main stem and branches, and develops a stem with one to four leaves. In a previous study, the differentiated racemes developed compound leaves when assimilates were supplied to the raceme (Saitoh et al. 2001). In the present two- year study, seed yield was positively correlated with total pod number (r=0.934, P<0.001) and pod number on racemes with compound leaves (r=0.864, P<0.01). Thus the increase in the pod number on

The longer sunshine hours accelerated the source activity and increased assimilates were supplied to the axil of each node. Our three-year planting density experiment showed that the number of floral buds on racemes with compound leaves increased markedly in the year with longer sunshine hours (Saitoh et al. 1998a), and the pod number on racemes with compound leaves increased especially when the twelfth node was isolated by pruning the

the raceme with compound leaves contributed to the increase in seed yield.

Values indicate the number of weed plants. Average of three quadrats (80cm \* 60cm) .

Table 4. Emergence of weeds at the beginning of flowering of soybean.

**3.8 Weed emergence** 

2001

2002

**4. Discussion** 

leaves (Table 3).

emerged weeds in narrow plots than in wide plots.

top above the twelfth node and removing all of the leaves, petioles and floral organs except those on the twelfth node at the flowering stage. Under such conditions, assimilates were concentrated to the twelfth node (Saitoh et al. 1998b), and the number of racemes with compound leaves on the main stem and branches increased when the leaves on branches and main stem were removed, respectively (Saitoh et al. 2001).

The present study revealed that the increase in pod number by narrow row planting was due to the increase in that on the racemes with compound leaves suggesting that the microclimate within canopy affected the development of racemes with compound leaves in narrow row-spacing. The narrow row-spacing canopy had a lower light extinction coefficient, i.e., better light-intercepting characteristics (Fig. 5).

In wide row-spacing, solar radiation was distributed non-uniformly, penetrated a deeper layer of the canopy due to fewer leaves distributed within the furrow, and decreased markedly above the row space (Fig. 7). In narrow row-spacing, solar radiation was distributed uniformly, the difference between the row and furrow was small, so that many racemes developed compound leaves due to the surplus assimilative supply to the raceme from the upper layer of canopy. The raceme with compound leaves is not only a sink organ, but also a source organ.

The canopy light extinction coefficients (k) measured under direct sunlight decreased during the daytime (Fig. 6). The decrease in k-value means that the sunlight penetrated uniformly into a deeper layer of the closed canopy with a higher LAI, however, sunlight reached a deeper layer directly and leaves received the excess light in non-closed canopy with lower LAI like wide row-spacing. This suggests that the k-value during the daytime can not evaluate the light intercepting characteristics in non-uniformly foliage distributed canopy.

The comparison of dry matter production in the plants with different planting patterns revealed that dry-weight was heavier and LAI was larger in dense plots than in sparse plots along as shown by others (Shibles and Weber 1965, Sugiyama et al. 1967, Asanuma et al. 1977, and also in narrow row-spacing than wide row-spacing (Fig. 3, 4) in accordance with the previous studies (Bullokck et al. 1998, Duncan 1986,Shibles and Weber 1965, 1966). In narrow row-spacing, the distance between plants was longer than in wide row-spacing, so that the canopy had a better light-intercepting environment, which accelerated the development of branches and racemes with compound leaves and the expansion of leaf area during the earlier stage, though, LAI in dense planting at 65 DAS exceeded 8, which means over luxuriant growth (Sugiyama et al. 1967).

Next, we should consider the effects of lodging. The lodging score was larger in narrow plots than in wide plots, (Fig. 8). This is because the distance between plants was longer in narrow row-spacing, and there was less mutual support with the neighboring plants. After the full flowering stage, a large amount of foliage was distributed in the upper layer of the canopy in the narrow/sparse plots (Fig. 5), and the higher the center of gravity, the higher the susceptibility to lodging. In narrow row-spacing, the main stem length was 15cm shorter and 0.9mm thicker than in wide row-spacing in 2001 (Table 1) because the competition between plants for elongation growth decreased due to the longer distance between plants. Despite this, the lodging score was larger in narrow row-spacing, meaning that the lodging of soybean was influenced by the above ground weight and center of gravity than the main stem length and stem thickness. Further study is needed to analyze the factors affecting the lodging tolerance in soybean.

Finally, let me consider about the weed management. In narrow row-spacing, we should eradicate weeds by hand if early weed control fails. It is impossible to kill weeds by cultivator after sowing. It was already demonstrated that the narrow row cultivation decreased weeds emergence and the alternative application of herbicide to soil or foliage (Gramineae weeds) could control weeds with labour saving and stability (Ohdan et al. 2005). Present results also showed that the less number of weeds were appeared in narrow row plots than in wide row plots (Table 4), in both plots herbicide was applied to the soil surface after sowing and the soil molding was conducted with a rotary cultivator in wide row plots. The dry-weight of weeds per square-meter was about 2g, which was extremely less than that of soybean, 300-400 g m-2, i.e., weeds could be controlled sufficiently. We considered that weeds could be controlled by one application of herbicide to the soil surface after sowing. If we failed to kill weeds by the soil applied herbicide, the additional application of bentazone, newly registered foliar applied herbicide in Japan, can be used after sowing.

#### **5. Conclusion**

The narrow row-spacing (wide distance between plants) and dense planting in soybean increase seed yield than in the wide row-spacing (narrow distance between plants), which was caused by the decrease in competition among plants for elongation growth, the promotion of branch development, the development of racemes with compound leaves, and the increase in pod number due to the uniform light environment within the upper layer of canopy. The improvement of lodging tolerance and perfect weed control will be needed in the narrow row and dense planting of soybean were considered to be needed.

#### **6. References**


cultivator after sowing. It was already demonstrated that the narrow row cultivation decreased weeds emergence and the alternative application of herbicide to soil or foliage (Gramineae weeds) could control weeds with labour saving and stability (Ohdan et al. 2005). Present results also showed that the less number of weeds were appeared in narrow row plots than in wide row plots (Table 4), in both plots herbicide was applied to the soil surface after sowing and the soil molding was conducted with a rotary cultivator in wide row plots. The dry-weight of weeds per square-meter was about 2g, which was extremely less than that of soybean, 300-400 g m-2, i.e., weeds could be controlled sufficiently. We considered that weeds could be controlled by one application of herbicide to the soil surface after sowing. If we failed to kill weeds by the soil applied herbicide, the additional application of bentazone, newly registered foliar applied herbicide in Japan, can be used

The narrow row-spacing (wide distance between plants) and dense planting in soybean increase seed yield than in the wide row-spacing (narrow distance between plants), which was caused by the decrease in competition among plants for elongation growth, the promotion of branch development, the development of racemes with compound leaves, and the increase in pod number due to the uniform light environment within the upper layer of canopy. The improvement of lodging tolerance and perfect weed control will be needed in

Asanuma, K., Naka, J. & Kogure, K. (1977). On the relation between dry matter production

Beatty, K.D., Eldridge, I.L. & Simpson, A.M.Jr. (1982). Soybean response to different planting patterns and dates, *Agronm Journal*, Vol.74, pp. 859-862, ISSN 0002-1962. Bullokck, D., Khan, S. & Rayburn, A. (1998) Soybean yield response to narrow rows is

Board, J.E., Harville, B.G. & Saxton A.M. (1990a). Narrow-row seed-yield enhancement in determinate soybean. *Agronm Journal*, Vol.82, pp. 64-68, ISSN 0002-1962. Board, J.E., Harville, B.G. & Saxton, A.M. (1990b). Branch dry weight in relation to yield

Cooper, R.L. (1971). Influence of soybean production practices on lodging and seed yield in

Cooper, R.L. & Nave, W.R. (1974). Effect of plant population and row width on soybean

Cooper, R.L. (1977). Response of soybean cultivars to narrow rows and planting rates under weed-free conditions, *Agronm Journal*, Vol.69, pp. 89-92, ISSN 0002-1962.

and plant density in autumn type soybeans. *Technical Bulletin Faculty of Agriculture* 

largely due to enhanced early growth, *Crop Science*, Vol.38, pp. 1011-1016, ISSN

increases in narrow-row soybean, *Agronm Journal*, Vol.82, pp. 540-544, ISSN 0002-

highly productive environments, *Agronm Journal*, Vol.63, pp. 490-493, ISSN 0002-

yield and harvesting loss, Transactions of the ASAE, Vol.17, pp. 801-805. ISSN 0001-

the narrow row and dense planting of soybean were considered to be needed.

*Kagawa University*, Vol.28, pp. 11-18, ISSN 0368-5128.

after sowing.

**5. Conclusion** 

**6. References** 

0011-183X.

1962.

1962.


Torigoe Y., Shinji H. & Kurihara H. (1982). Studies on developmental morphology and yield determining process of soybeans. II. Developmental regularity of flower clusters and flowering habit from a view point of gross morphology, *Japanese Journal of Crop Science*, Vol.51, pp. 89-96, ISSN 0011-1848.
