|Home | About | Journals | Submit | Contact Us | Français|
The stem growth habit, determinate or indeterminate, of soybean, Glycine max, varieties affects various plant morphological and developmental traits. The objective of this study is to identify the effect of stem growth habit in soybean on the stomatal conductance of single leaves in relation to their leaf morphology in order to better understand the ecological and agronomic significance of this plant trait.
The stomatal conductance of leaves on the main stem was measured periodically under favourable field conditions to evaluate gmax, defined as the maximum stomatal conductance at full leaf expansion, for four varieties of soybean and their respective determinate or indeterminate near isogenic lines (NILs). Leaf morphological traits including stomatal density, guard cell length and vein density were also measured.
The value of gmax ranged from 0·383 to 0·754 mol H2O m−2 s−1 across all the genotypes for both years. For the four pairs of varieties, the indeterminate lines exhibited significantly greater gmax, stomatal density, numbers of epidermal cells per unit area and total vein length per unit area than their respective determinate NILs in both years. The guard cell length, leaf mass per area and single leaf size all tended to be greater in the determinate types. The variation of gmax across genotypes and years was well explained by the product of stomatal density and guard cell length (r = 0·86, P < 0·01).
The indeterminate stem growth habit resulted in a greater maximum stomatal conductance for soybean than the determinate habit, and this was attributed to the differences in leaf structure. This raises the further hypothesis that the difference in stem growth habit results in different water use characteristics of soybean plants in the field. Stomatal conductance under favourable conditions can be modified by leaf morphological traits.
Soybean, Glycine max, is an important crop with seeds that contain abundant protein and lipids. The stem growth habit, governed primarily by the Dt1 locus, has a great impact on soybean plant stature (Heatherly and Elmore, 2004). In the determinate types (D-types), the stem apical meristem stops the differentiation of new leaves after floral induction in the plant. The D-type varieties can be recognized by the appearance of the terminal leaf, similar in size to the earlier leaves. In the indeterminate type (I-type) varieties, leaf production continues until the latter part of the growth season. The I-type varieties can be recognized by the gradual reduction in leaf size and smaller leaves in the upper part of the leaf canopy. These properties generally result in an earlier completion of leaf canopy expansion, a shorter main stem length and a bushier canopy in D-types compared with I-types (Hay and Porter, 2006).
There are clear differences in the regional distribution of the I-type and D-type varieties. In the USA, for instance, most of the soybean varieties in the north are I-types that mature earlier, and in the south most of the varieties are classified as D-types that mature later (Heatherly and Elmore, 2004). The agronomic significance of such a regional distribution, however, has not been fully documented in terms of its physiological and ecological basis, which is necessary to realize the full utilization of this plant trait in soybean crop production.
Stomatal conductance (gs) is one of the most important physiological traits that governs dry matter accumulation and plant water balance. It regulates the diffusion of both carbon dioxide and water vapour through stomatal pores. Indeed, drought can lead to stomatal closure and a reduction of soybean yield that more or less depends on the duration and severity of the drought (Frederick et al., 1990, 1991; Kokubun and Shimada, 1994; De Costa and Shanmugathasan, 2002; Hufstetler et al., 2007). Thus, optimizing the leaf gas exchange characteristics for stable production under water deficit and/or unstable soil moisture conditions can be an important issue for soybean breeding programmes. A strong correlation between leaf photosynthetic rate (Pn) and gs is found in general (Wong et al., 1979; Ohsumi et al., 2007). In soybean, however, it has been said that gs is not necessarily a major factor contributing to significant differences in Pn between cultivars (Dornhoff and Shibles, 1970; Jiang and Xu, 2001; Sinclair, 2004). In our observation of recombinant inbred lines (RILs) derived from a cross between US and Japanese soybean varieties (Tanaka et al., 2008), a close correlation was found between Pn and gs. In that study, the property of leaf gas exchange was suggested to have been associated with leaf anatomy, especially with stomatal density. Such a morphological determination of leaf gas exchange should be studied further because it may help with the optimization of leaf gas exchange characteristics for the adaptation of soybean to given climatic and agronomic environments.
Tanaka et al. (2008) have also suggested the existence of an unexpected effect of the type of stem growth habit on leaf gs and Pn among the RILs, but the anatomical and physiological basis of this phenomenon remains unclear. There is no other information about the effect of soybean stem growth habit on the anatomy and gas exchange characteristics of the single leaf. Considering the differences in agronomic characteristics and the regional distribution of I-types and D-types, it may be important to elucidate how the type of stem growth habit affects gas exchange characteristics in soybean for the efficient utilization of Dt1 under various cultivation conditions.
In the present study, we hypothesized that the stem growth habit directly affects the leaf anatomy and gs. To test this, four pairs of near-isogenic lines (NILs) of stem growth habit (Bernard et al., 1991), each having a common genetic background, were planted in the field for 2 years, and gs was measured along with leaf morphological traits. The second objective of the study is to elucidate the relationship between gs and leaf morphological traits across different genotypes.
The Clark (I-type), Harosoy (I-type), Elf (D-type) and Williams (I-type) soybean [Glycine max (L.) Merr.] varieties and the NILs of each cultivar for stem growth habit (in total eight lines, Bernard et al., 1991) were used (Table 1). Field experiments were conducted during two seasons, in 2007 and 2008, on an alluvial sandy loam soil (Fluvic Endoaquepts) in Kyoto, Japan. The dates of sowing were 21 June 2007 and 24 June 2008. Plant spacing was 0·8 × 0·15 m, and N, P2O5 and K2O fertilizers were applied at 3, 10 and 10 g m−2, respectively, before sowing in both years. Two plot replications were established for each genotype in a completely randomized design. Each experimental plot consisted of 45 plants in three rows. In 2007, there was a severe dry period from the end of July to the middle of August. Furrow irrigation was conducted before gs measurements several times in both seasons to avoid drought stress.
An SC-1 leaf porometer (Decagon, Pullman, WA) was used for gs measurements from the abaxial sides of central leaflets of three plants in each plot. We had no information for the adaxial sides of leaflets, but the importance of this side is considered to be less than that of the abaxial side in this study. All of the measurements were made between 1000 h and 1300 h on clear sunny days during the flowering to seed filling stage of the soybean plants. To evaluate the maximum value of gs, we used the linear decline model of gs (Tanaka et al., 2008). Briefly, the gs of each genotype is represented as a linear function of the days after full leaf expansion (t). Measurements of gs were made at least three times for each central leaflet at several positions of the main stem. The dates of the full expansion of all the measured leaflets were recorded to calculate t for each gs measurement date. The maximum of gs (gmax) was defined as gs at full leaf expansion, which was estimated by the linear regression of gs measurements and t [eqn (1)]:
where a is a constant for each genotype (Fig. 1).
To observe the leaf anatomy, replicas of the leaf epidermis of the abaxial sides of fully expanded central leaflets were made using Suzuki's Universal Method of Printing (SUMP, Tokyo, Japan). In short, the abaxial sides of leaves were pressed on to a SUMP plate treated with 10 µL of SUMP liquid (amyl acetate) and held in place until the liquid became solid. These replicas were observed at 400-fold (10 × 40) magnification with a light microscope (BH-323, Olympus, Tokyo, Japan). Digital images were obtained with a digital camera (COOLPIX P5100, Nikon, Tokyo, Japan) set on the microscope. The stomatal density (Nstoma), the number of epidermal cells per unit area (Nepi) and the guard cell length (Lguard) were determined by analysing these digital images on ImageJ (NIH, USA). The stomatal index (SI) was calculated by following eqn (2):
In 2008, fresh leaves were put directly on the microscope slide to obtain digital images of leaf venation at 100-fold (10 × 10) magnification (Fig. (Fig.2).2). These images were used to determine the total length of leaf venation per unit area (Lvein) with the segmented line tool on ImageJ.
The area of central leaflets (LA) was measured with a portable area meter (LI-1700, Li-COR, Lincoln, NE). The leaf dry mass was determined after drying at 80 °C for 48 h. The leaf mass per area (LMA) was derived from the LA and leaf dry mass measurements.
For both years, the values of gs of three plants in each plot were averaged and then used for the calculation of gmax. For the epidermal observations, two central leaflets were sampled from each plot to make the SUMP replicas. Another two leaves were sampled to measure the LA and LMA. In 2008, these leaves were also used to determine the Lvein. All of these values were then averaged across the two plots of each genotype and the standard error (s.e.; n = 2) was calculated for each trait. We followed Gomez and Gomez (1984) to evaluate the effect of year, background genotypes and stem growth habit on gmax variation. In each year, analysis of variance (ANOVA) was conducted in a split plot design with cultivar as the main plot and stem growth habit as the sub-plot. Then we conducted combined ANOVA over 2 years. Year, cultivar and growth habit were considered as fixed factors. Replication was considered as a random factor and nested in years. For leaf morphological traits, the variation between genotypes was evaluated by Tukey's test (P = 0·05). All of these statistical analyses were conducted using Microsoft Excel (Microsoft, Redmond, WA).
In 2007, gmax ranged from 0·518 ± 0·035 mol m−2 s−1 in Williams-D to 0·754 ± 0·020 mol m−2 s−1 in Clark (Fig. 3). In the comparison between the stem growth habits, gmax tended to be greater in I-types in each background genotype. This tendency was common to both years, though gmax was relatively lower in 2008; it ranged from 0·383 ± 0·064 mol m−2 s−1 in Harosoy-D to 0·556 ± 0·008 mol m−2 s−1 in Clark in 2008. Based on the ANOVA (Table 2), the effects of year (P < 0·05) and stem growth habit (P < 0·01) on the variation of gmax were significant, but the effect of the cultivar was not significant. None of the interaction terms for the three factors was significant.
Tables 3 and and44 show the variation in leaf morphological traits between genotypes and stem growth habits in 2007 and 2008, respectively. Over the two years, the Nstoma ranged from 181 mm−2 for Harosoy-D and Williams-D in 2008 to 378 mm−2 for Elf-I in 2007. The Nepi ranged from 784 mm−2 for Clark-D in 2007 to 1478 mm−2 for Elf-I in 2008. The background genotype of Elf had a relatively greater Nstoma and Nepi in each group of growth habit in both years. The I-types exhibited constantly greater Nstoma and Nepi for any cultivar pair for both years. Averaged across all the genotypes, Nstoma in 2008 was lower by 11 % than that of 2007, and Nepi in 2008 was greater by 12 % than that of 2007.
There was no difference in the SI calculated from Nstoma and Nepi between genotypes in either year. The SI ranged from 21·4 to 22·3 in 2007 and from 17·0 to 19·1 in 2008. In contrast, the average value of SI across all the genotypes was lower in 2008 by 17 % than in 2007.
The Lguard was greater in the D-types of any cultivar pair in both years. The Lguard varied from 17·4 µm for Elf-I in 2008 to 26·2 µm for Clark-D in 2007. Unlike Nstoma and SI, a difference between the two years was not observed. The background genotype of Elf had a relatively lower Lguard than the others.
The LA varied from 46·9 cm2 for Harosoy in 2007 to 94·0 cm2 for Williams-D in 2008. It tended to be greater in the D-types in both years with few exceptions. When all the genotypes were averaged, the LA in 2008 was greater by 39 % than that in 2007. For LMA, the values varied from 38·8 g m−2 for Harosoy in 2008 to 67·8 g m−2 for Williams-D in 2007. In some genetic backgrounds, D-types exhibited a greater LMA than I-types. The difference in LMA between years was not significant.
The Lvein was only measured in 2008, and it ranged from 3·54 ± 0·079 mm mm−2 in Harosoy-D to 5·25 ± 0·134 mm mm−2 in Elf-I (Fig. 4). The I-types clearly had a greater Lvein than the D-types. The background genotype of Elf again exhibited relatively greater Lvein than the others. The variation of Lvein was very closely correlated with that of Nstoma (Fig. 5, r = 0·98, P < 0·01).
The Nstoma was also significantly correlated with gmax, r = 0·687 (P < 0·01). No significant relationship was detected between Lguard and gmax when all the data were pooled. The value of Nstoma × Lguard varied from 4·29 mm mm−2 for Harosoy-D in 2007 to 7·41 mm mm−2 for Elf-I in 2008. The variation in Nstoma × Lguard was highly correlated with gmax across all genotypes and years (Fig. 6, r = 0·86, P < 0·01).
The I-types exhibited greater gmax than the D-types across all the background genotypes (Fig. 3). The effect of stem growth habit on gmax had no interaction with that of the year or the background genotype (Table 2). These results confirmed that the indeterminate stem growth habit independently and directly has a positive effect on gmax. The variation of gmax as related to stem growth habit seemed to be primarily due to a greater Nstoma in I-types (Tables 3 and and4).4). The I-types also had a greater Nepi. The values of SI were almost constant across the different growth habits, indicating that I-types had relatively smaller cells and finer structure on their leaf epidermis. The smaller Lguard of I-types seems to be consistent with these results. For now, this is the first finding that shows the direct influence of stem growth habit on leaf morphological traits and stomatal conductance in the field.
The I-types are known to have smaller leaflets on the main stem compared with the D-types, especially in the later stages of vegetative growth. It is likely that the smaller LA is related to the smaller cells and the finer structure of the epidermis in I-types. The correspondence between LA and Nepi, however, was not clear. The correlation between LA and Nepi was not significant in either 2007 (P = 0·38) or 2008 (P = 0·22), or for the pooled data of the two years (P = 0·78). Clark (I-type) had larger leaflets than Clark-D in 2007, but neither the Nepi nor the Nstoma of Clark was greater than that of Clark-D. These results indicate that the finer structure of the leaves of I-types may not simply be attributed to the smaller size of I-type leaves.
The Elf background genotype tended to have a greater Nstoma and Nepi, and a smaller Lguard for both stem growth habits and years. This suggests that Elf has some genetic factor that is related to its fine structure of leaves other than stem growth habit. Further experiments should be conducted to explore this point.
We observed large differences in leaf morphological traits between the two years. Leaf epidermal anatomy in various plant species is generally known to be responsive to many environmental factors, such as CO2 concentration (Woodward, 1987; Woodward and Kelly, 1995), light intensity (Coupe et al., 2006), UV-B treatment (Gitz et al., 2005) and water conditions (Mehri et al., 2009). Since our experiment was conducted in the same field in both years, it seems that the differences observed across years were primarily caused by the difference in the water conditions. In the case of the experimental site, the soil is more or less dry during the period from the end of July to the middle of the August. While the drought of this period was particularly evident in 2007, we had frequent precipitation events during the corresponding period in 2008.
Although the response of stomatal aperture to water deficit has been well studied, little is known about the effect of water conditions on stomatal development in terms of density and size. The information about this subject is somewhat contradictory among studies or crop species, suggesting that there is a complex regulation of stomatal development in response to the environment (Casson and Gray, 2008). For soybean, Buttery et al. (1993) have indicated that Nstoma increases under drought conditions compared with well-watered conditions. They have also observed that the leaf area per plant is decreased by drought stress due to the inhibition of leaf expansion.
These trends seem to be similar to that of this study in 2007, because we observed an increased Nstoma with a smaller LA. It is possible that the soybean plants were exposed to a water deficit in 2007 during flowering when new leaves were still expanded in both types. Although we conducted furrow irrigation several times before the gs measurements, it may not have been sufficient to avoid the water stress during the dry period due to the severe drought conditions in 2007.
The influence of stem growth habit on Nstoma and Lguard was evident in our experiments, but the effects of these morphological differences on gmax have yet to be evaluated. The gas exchange through the stomata is physically determined by the length, width, and depth of single stomata and the stomatal density. Using our abbreviations for this study, the relationship is represented as follows (Parlange and Waggoner, 1970):
where w, d and k are the stomatal width, stomatal depth and a constant, respectively. This equation clearly shows the importance of the effect of Nstoma and Lguard on gs. A short-term response of gs to the environment, such as the water conditions or light intensity, is primarily caused by variation in stomatal width. Then we rearranged eqn (3) as follows:
where wmax is the maximum width of stomata under favourable conditions. It is likely that the shapes of the stomatal pores are similar across the different soybean genotypes with various stomatal sizes. Thus, we assumed that the ratios d/wmax and Lguard/wmax are both constant. By this assumption, eqn (4) indicates that the variation of gmax is simply proportional to the product of Nstoma and Lguard. A clear relationship between gmax and Nstoma × Lguard across all the background genotypes, growth habits and years (Fig. 6) is consistent with this expectation. Although much of the variation shown in this figure arises from the difference between years, the correlation between gmax and Nstoma × Lguard is still significant when values are averaged across the two years (r = 0·89, P < 0·01). This analysis gives us a more complete understanding of the relationship between stem growth habit, leaf morphology and stomatal conductance, and it suggests that Nstoma × Lguard is a useful index for the evaluation and screening of gmax among large plant populations.
The hydraulic architecture of leaves is a key determinant of water transport through the leaves (Cochard and Coll, 2004). Brodribb et al. (2007) have found a clear relationship between the length of the hydraulic pathway from the ends of the vein system and the leaf hydraulic conductance or maximum photosynthetic rate across species. Based on these results, they have also predicted that thicker leaves with lower vein densities will have longer hydraulic pathways and, hence, a lower leaf hydraulic conductance. In 2008, we found that the D-types had a lower Lvein (Fig. 4) and a greater LMA (Table 4). A strong correlation between Lvein and Nstoma (Fig. 5) as well as a lower gmax in the D-types suggests that the D-types have a lower leaf hydraulic conductance in their leaves due to their lower Lvein and thicker leaves compared with the I-types. In soybean, a low leaf hydraulic conductance is known to be associated with ‘slow wilting’ characteristics for some cultivars in a drying soil (Sinclair et al., 2008). Fletcher et al. (2007) have pointed out that the whole plant transpiration rate of the slow wilting genotypes is lower than that of commercial cultivars when the vapour pressure deficit (VPD) is increased. This ‘water conservation’ characteristic of slow wilting genotypes may be useful for breeding drought-tolerant soybean varieties with efficient water use under water deficit conditions (Sloane et al., 1990; King et al., 2009).
It is possible to hypothesize that D-types are relatively slow wilting genotypes compared with the I-types, but the effect of stem growth habit on dry matter productivity is not yet clear. Using the NILs for stem growth habit under free air CO2 enrichment (FACE) in the field, Ainsworth et al. (2004) have shown that, in contrast to the case of the D-type, the I-type NIL had no downregulation of photosynthesis under high [CO2]. They concluded that the greater accumulation of non-structural carbohydrates in D-type leaves due to a lower sink–source ratio resulted in the severe downregulation of photosynthesis under high [CO2]; however, they did not observe any difference in Pn between NILs under ambient air conditions. Beaver et al. (1985) have suggested that there is an effect of stem growth habit on dry matter productivity, but their results were not consistent across years. In a comparison of the crop radiation use efficiency (RUE) of cultivars, Shiraiwa et al. (1994) did not find a consistent difference in RUE between I-type and D-type cultivars. Also, Robinson and Wilcox (1998) found no difference between D-type and I-type varieties in seed yield. Based on our discussion, however, the influence of stem growth habit on dry matter productivity should be evaluated not only under favourable conditions but also in relation to water use characteristics or drought tolerance. In support of our hypothesis above, Mahmood et al. (1999) have reported that an I-type cultivar ‘Touzan 66’ has more drastic yield reduction under drought stress compared with the D-type cultivar ‘Enrei’. Further experiments concerning this relationship should be conducted.
The specific mechanisms by which growth habit affects leaf anatomy and gmax still remain unresolved. In Arabidopsis, the ERECTA gene affects the plant stature, inflorescence development and stem elongation (Torii et al., 1996). This gene encodes a putative leucine-rich repeat receptor-like kinase, which mediates cell to cell communication and has an important role in plant morphogenesis (Yokoyama et al., 1998). Interestingly, Masle et al. (2005) have reported a remarkable effect of erecta mutation on transpiration efficiency and gs. They have also observed that a large difference in stomatal density is caused by this mutation, but no difference in stomatal index was detected. These results are similar to our results in soybean. Molecular studies on the function and regulation of the Dt1 gene could be expected to reveal the link between stem growth habit and leaf anatomy.
In summary, there was a clear effect of stem growth habit on stomatal conductance under favourable conditions. The greater gmax in I-types was primarily attributable to their finer leaf structure and greater Nstoma, while the Lguard is relatively smaller in I-types. The SI was fairly constant across stem growth habits. The value of Nstoma × Lguard explained the variation of gmax across all the genotypes and both years very well, suggesting that the stomatal conductance of single leaves under favourable conditions can be modified by these morphological traits. Stem growth habit also affected the Lvein and LMA. These findings suggest that I-types and D-types have quite different strategies for water use at the field level and that D-types might be helpful for breeding ‘slower-wilting’ cultivars. Further experiments should be conducted to examine the relationship between stem growth habit and the response of dry matter accumulation or water use to various water conditions at the field level.