Effects of light quality on leaf morphogenesis of a heterophyllous amphibious plant, Rotala hippuris

Abstract

Background and Aims

For heterophyllous amphibious plants that experience fluctuating water levels, it is critical to control leaf development precisely in response to environmental cues that can serve as a quantitative index of water depth. Light quality can serve as such a cue because the ratio of red light relative to far-red light (R/FR) increases and blue-light intensity decreases with increasing water depth. Growth experiments were conducted to examine how R/FR and blue-light intensity alter leaf morphology of a heterophyllous amphibious plant, Rotala hippuris.

Methods

Using combinations of far red (730 nm), red (660 nm) and blue (470 nm) light-emitting diodes (LEDs), growth experiments were used to quantitatively evaluate the effects of the R/FR ratio and blue-light intensity on leaf morphology.

Key Results

Under the natural light regime in an outside growth garden, R. hippuris produced distinct leaves under submerged and aerial conditions. R/FR and blue-light intensity were found to markedly affect heterophyllous leaf formation. Higher and lower R/FR caused leaf characters more typical of submerged and aerial leaves, respectively, in both aerial and submerged conditions, in accordance with natural distribution of leaf types and light under water. High blue light caused a shift of trait values toward those of typical aerial leaves, and the response was most prominent under conditions of R/FR that were expected near the water surface.

Conclusions

R/FR and blue-light intensity provides quantitative cues for R. hippuris to detect water depth and determine the developmental fates of leaves, especially near the water surface. The utilization of these quantitative cues is expected to be important in habitats where plants experience water-level fluctuation.

INTRODUCTION

The water surface provides a clear boundary between two contrasting conditions in which plants require different leaf morphology to maintain their photosynthetic activities, i.e. aerial and submerged conditions. Although the majority of plants that live in terrestrial and aquatic habitats utilize either an aerial or a submerged space to expand their leaves, heterophyllous amphibious plants produce two types of leaves, i.e. aerial and submerged leaves, and utilize both spaces across the water–surface boundary (Allsopp, 1965; Wells and Pigliucci, 2000). Aerial leaves are typically wide and thick and have high stomata density. In contrast, submerged leaves are slender and thin and have low stomata density or no stomata. Heterophyllous plants utilize diverse external and internal cues, such as osmotic stress, temperature, photoperiod, light, CO₂ concentration, abscisic acid, gibberellic acid and ethylene, for determining the developmental fates of their leaves (Johnson, 1967; Bristow, 1969; Anderson, 1982; Deschamp and Cooke, 1984; Goliber and Feldman, 1989; Kuwabara et al., 2003; for other references, see Wells and Pigliucci, 2000).

For heterophyllous amphibious plants that experience fluctuating water levels, it is critical to control leaf development precisely in response to environmental cues that can serve as a quantitative index of depth at a scale of a few to tens of centimetres. Light quality, such as the ratio of red-light intensity to far-red-light intensity (hereafter R/FR) and blue-light intensity, may serve as such cues because wavelength-specific light absorption of water produces a strong environmental gradient along with water depth. The absorption rate of water is very high in the infrared region of the light spectrum, decreases rapidly as wavelength decreases and reaches a minimum absorption in the blue region, and then increases again in the violet and UV wavelength regions (Spence, 1981; Smith, 1982; Wetzel, 2001; Ragni and D'Alcalà, 2004). Consequently, the increase in R/FR with water depth can be very large (Chambers and Spence, 1984), and it has been proposed that plants perceive R/FR as an index of depth (Spence, 1981; Smith, 1982, 1994).

Indeed, light quality, such as R/FR and blue-light intensity, has been reported to be a major determinant of heterophyllous leaf development in several amphibious plants. In a classical paper by Gaudet (1963, 1965), it was reported that the most effective means of inducing aerial leaves of an aquatic fern, Marsilea vestila, under water is exposure to far-red light. In another freshwater macrophyte, Hippuris vulgaris, it has been shown that low R/FR light induces aerial leaves, while aerial leaf formation is inhibited at higher R/FR (Bodkin et al., 1980; but see Goliber, 1989). Low R/FR and also higher light intensity induce aerial leaves in Proserpinaca plaustris (McCallum, 1902) and H. vulgaris (Goliber and Feldman, 1990). These responses to high irradiance are probably related to blue-light intensity, as are some leaf development responses of terrestrial species (Franklin et al., 2005), and as is aerial leaf formation in Marsilea quadrifolia (Lin and Yang, 1999). Although the importance of light quality as a determinant of leaf morphogenesis has been suggested in terrestrial plants (Franklin, 2008), studies of heterophylous plants are still limited. We expected that R/FR is a reliable cue of water depth in a range higher than that of direct sunlight (nearly equal to 1), because R/FR sharply increases with water depth. However, under low R/FR, blue light may be more important as a cue to control leaf morphogenesis. Such interactions between R/FR and blue light cues have rarely been tested in studies of heterophylous plants.

Here, we report the results of growth experiments that examined how light quality, including R/FR ratio and blue-light intensity, alters leaf morphogenesis of a heterophyllous amphibious plant, Rotala hippuris. Using combinations of far-red (730 nm), red (660 nm) and blue (470 nm) light-emitting diodes (LEDs), we performed growth experiments that quantitatively analysed the effects of R/FR on leaf morphology and the interactions between R/FR and blue-light intensity. Specifically, we investigated the following questions about the leaf morphogenesis of R. hippuris: (1) Is the response to R/FR qualitative (a threshold response that forms either aerial or submerged leaves) or quantitative (a continuous response from aerial to submerged leaves)? (2) At what R/FR is the response to blue-light intensity most pronounced?

MATERIALS AND METHODS

Plants

Rotala hippuris Makino (Lythraceae) is an aquatic and semi-aquatic perennial that occurs in small lake and irrigation ponds in western Japan, including western Honshu, Shikoku and Kyushyu (Kadono, 1994). The plants form perennial rhizomes in submerged soils, and annual shoots grow from the rhizomes. Submerged stems often branch, especially at the basal parts, and elongate up to 60 cm depending on the depth of water. After the stems reach the water surface, plants form upright aerial shoots that produce flowers at the base of each leaf. The flowering season extends from August to October. The plant is known to be heterophyllus and to produce morphologically distinct leaves on submerged and aerial parts of shoots. The species is threatened and is listed as Endangered in the Red Data Book (Environment Agency of Japan, 2000).

The plant materials used in this study were all derived from a single clone that was originally collected at Matsusaka, Mie Prefecture (34°32′25″N, 136°31′15″E, altitude approx. 40 m) and had been maintained in an experimental garden at Kobe University (34°43′51″N, 135°14′12″E, altitude approx. 140 m). Use of a single clone restricted us from expanding our findings to the extant variation within species. This will need to be studied in future under careful conservation-oriented sampling and propagation programme.

Characterization of aerial and submerged leaves

Aerial and submerged leaves of R. hippuris were collected in August 2003 from plants that had been maintained in large containers placed in the open outside experimental garden. Twelve replicates of each type of fully expanded mature but undamaged leaf were sampled from independent vigorous aerial and submerged shoots, respectively (Fig. 1A, D). Submerged leaves were collected at 5–10 cm depth. Aerial and submerged leaves of R. hippuris have a simple linear shape (Fig. 1B, E). For each leaf, we recorded leaf length and width, leaf tip morphology, length and width of epidermal cells on the leaf upper and lower surfaces, and the stomata densities on the leaf upper and lower surfaces. Six summary variables were calculated, i.e. leaf length/leaf width (leaf L/W), leaf-tip index, length/width of epidermal cells on leaf upper and lower surface (upper and lower cell shapes), and upper and lower stomata densities. Leaf length was measured using a digital caliper (DIGIPA 700-125, Mitutoyo, Kawasaki, Japan). Maximum widths of aerial and submerged leaves were measured using a digital caliper and a microscope with attached micrometer (Scale lupe, PEAK, Japan), respectively. The lateral edges of the leaf tip are toothed; submerged leaves form a U-shaped gap between the two teeth (Kadono, 1994), but the leaf tip is almost flat or even protruding for the aerial leaves (Fig. 1C, F). Leaf-tip index was calculated by dividing either the depth (positive value) of the U-shaped gap or the height (negative value) of the leaf edge relative to the height of the teeth by the width of the structures measured as the distance between the teeth (Fig. 1G). For the upper and lower surfaces of leaves, we measured the length and width of three epidermal cells at the middle part of the leaf using the micrometer set in the microscope. We counted and averaged the numbers of stomata within three 6·25 × 10⁻² mm² areas at the middle part of leaves for each upper or lower surface.

Fig. 1.

Gross morphology of shoots and leaves of Rotala hippuris growing under aerial (A–C) and submerged (D–F) conditions. Photographs of shoots (A, D), leaves (B, E) and leaf tips (C, F). A schematic diagram showing measurement method of leaf-tip morphology of aerial and submerged leaves (G) is also shown. The leaf-tip index is defined as height or depth divided by width. Scale bars in mm.

Growth experiments

Two growth experiments were conducted to investigate the role of light quality in leaf morphogenesis. In the first experiment (Experiment 1), a wide range of R/FR was applied to plants in aerial and submerged conditions. In aerial conditions, we evaluated whether high R/FR alone can be a cue to produce submerged leaves. In the submerged experiment, R/FR responses were evaluated under conditions where other possible cues for submergence are simultaneously operating. In the second experiment (Experiment 2), interactions between blue light and R/FR were studied in the submerged condition. The range of R/FR was set to represent the conditions experienced by plants in natural habitats and was smaller than that in Experiment 1.

For these growth experiments, we used 114 similar-sized plants (60 and 54 plants for Experiments 1 and 2, respectively) selected from approx. 200 plants prepared by the following procedure. Shoot tips were harvested at 5 cm from aerial shoots in August 2004, 2 weeks prior to the start of each of Experiments 1 and 2. Each shoot segment was planted (2 cm from the base was inserted into soil to facilitate rooting) in a plastic pot (5 cm in diameter and 6 cm in depth) filled with soil and was incubated for 2 weeks in a greenhouse prior to the experiments. We used aerial shoots for the growth experiments not only for aerial but also for submerged conditions. Aerial shoots showed vigorous growth even after submergence and newly formed leaves at the shoot tips showed a rapid morphological response. In contrast, it was difficult to obtain rooted shoot cuttings from submerged shoots as they wither when they are exposed to aerial conditions, and new aerial shoots emerged from the basal part of the shoot. Thus using aerial shoots proved to be the easiest procedure to obtain small and even-sized rooted plants. The soil used in the experiments was a mixture of matrix soil (small-sized ‘Akadama’, soil clusters 2–5 mm in diameter), peat moss and culture soil (35 : 15 : 50 by volume). Culture soil contains 0·4, 1·9 and 0·6 g of N, P and K per kg, respectively.

Following greenhouse incubation, we conducted experiments in a growth room that was maintained at 28 ± 3°C air temperature throughout the experimental periods. Plants showed vigorous growth under this temperature regime. We created aerial and submerged conditions using plastic containers 20 × 14 × 15 cm (length × width × depth). The pots (6 cm in depth) were placed in the plastic containers and water of 3–4 or 14–15 cm depth was kept in the containers for aerial and submerged conditions, respectively.

Light was supplied by 360 LEDs (18 LEDs per plate × 20 plates) for each container, and light quality was controlled by simply altering the number of red, far-red and blue LED plates (MIL-R18, -IF18 and -B18, Sanyo, Moriguchi, Japan) (Table 1). Peaks of the wavelength distribution of blue, red, and far-red LEDs were 470, 660 and 730 nm, respectively. R/FR and blue-light intensity were measured at the top of the containers using SKR110 (Skye Instruments Ltd, Llandrindod Wells, UK) and LI-190SA (LI-COR, Lincoln, NE, USA) sensors with a LI-1400 data logger (LI-COR). In Experiment 1, five treatments with R/FR at 0·0013, 0·095, 0·69, 4·6 and 791·3 were performed for aerial (T1–T5 in Table 1) and submerged conditions (T6–T10). The intensity of blue light was 9·9 µmol s⁻¹ m⁻² for all treatments (Table 1). In Experiment 2, nine treatments were performed by the combinations of three R/FR, i.e. 0·2, 0·9 and 1·7, and three blue-light intensities, i.e. 0, 5·44 and 13·4 µmol s⁻¹ m⁻² (T11–T19, Table 1). We applied these conditions by altering the number of far-red and blue LEDs but keeping the number of red LEDs constant (Table 1). Total photosynthetically active radiation (PAR) was not controlled, and therefore the effects of blue light tested in the present study represent both those via blue-light receptors and those via other mechanisms that responded to PAR. Day/night lengths were set as 14/10 h using a controller (MIL- C1000T, Sanyo).

Table 1.

Treatment conditions (T1–T19) in Experiments 1 and 2

Treatment	Growth conditions (aerial/submerged)	Supplied R/FR and expected R/FR at 10 cm water depth*		Supplied blue light (μmol s−1 m−2 )	Number of LED plates
		Supplied (0 cm)	10 cm		R	FR	B
Experiment 1
T1	Aerial	0·0013	–	9·9	0	16	4
T2	Aerial	0·095	–	9·9	2	14	4
T3	Aerial	0·69	–	9·9	8	8	4
T4	Aerial	4·60	–	9·9	14	2	4
T5	Aerial	791·3	–	9·9	16	0	4
T6	Submerged	0·0013	0·0014	9·9	0	16	4
T7	Submerged	0·095	0·11	9·9	2	14	4
T8	Submerged	0·69	0·76	9·9	8	8	4
T9	Submerged	4·6	5·1	9·9	14	2	4
T10	Submerged	791·3	876·3	9·9	16	0	4
Experiment 2
T11	Submerged	0·2	0·22	0	3	11	0
T12	Submerged	0·9	1·0	0	3	3	0
T13	Submerged	1·7	1·9	0	3	1	0
T14	Submerged	0·2	0·22	5·4	3	11	2
T15	Submerged	0·9	1·0	5·4	3	3	2
T16	Submerged	1·7	1·9	5·4	3	1	2
T17	Submerged	0·2	0·22	13·4	3	11	6
T18	Submerged	0·9	1·0	13·4	3	3	6
T19	Submerged	1·7	1·9	13·4	3	1	6

The growth condition of plants (aerial/submerged), R/FR ratio (supplied ratio and expected values at 10 cm water depth), and intensity of blue light are listed. R/FR ratio and blue-light intensity were controlled by altering the number of red (R), far-red (FR) and blue (B) LED plates (18 LEDs per plate) in the source light supply. The peaks of the wavelength distribution of blue, red and far-red LEDs were 470, 660 and 730 nm, respectively. Altered conditions in each experiment are shown by bold letters.

* Expected R/FR at 10 cm water depth was calculated based on extinction coefficients of water at 470, 660 and 730 nm light estimated from the values presented in Wetzel (2001, table 5-2, p. 57).

For each treatment, six replicate plants were grown in a single container. Growth conditions were precisely controlled by the use of LEDs and by placing them in a single temperature-controlled growth room to minimize micro-environmental variation that might exist between containers. After the plants were grown for 2 weeks, leaves newly formed during the treatment were used to measure the above-described six summary variables (leaf L/W, leaf-tip index, upper and lower cell shapes, upper and lower stomata densities). After application of the experimental conditions, the gradual changes in gross morphology of newly or newly expanded leaves were observed acropetally at the shoot tips until more or less constant, treatment-specific shaped leaves were continually produced. All plants under different treatments reached this stage, and we judged that these treatment-specific shaped leaves were newly formed during the treatment.

Statistical analysis

Leaf L/W was log transformed and the other five traits (leaf-tip index, upper and lower cell shapes, and upper and lower stomata densities) were square-root transformed. Student's t-tests (parametric) were used to compare mean values between aerial and submerged leaves sampled from the outside garden. For Experiment 1, we first conducted two-way ANOVAs and highly significant main terms (aerial–submerged treatment and R/FR terms) were detected for all analysed traits even when they were tested against interaction terms (data not shown). We present separate analyses for aerial and submerged conditions, because submerged treatment itself was expected to alter R/FR. One-way ANOVAs were conducted across the five R/FR treatments. Multiple comparisons were conducted by Scheffe's test (P < 0·05). For Experiment 2, two-way ANOVAs were conducted and the two main (R/FR and blue light intensity) and interaction terms were tested. For the traits for which significant interaction terms were detected, main factors were tested against the interaction term. Because we do not have replicates of containers within each treatment, we applied conservative criteria to test the main factors. Statistical analyses were conducted using the software Statview for Windows ver. 5·0 (SAS Institute Inc., Cary, NC, USA) and R ver. 2·7·2 (R Development Core Team).

RESULTS

Aerial and submerged leaves under natural conditions

Aerial and submerged leaves of R. hippuris showed marked differences in their morphology (Fig. 1). The differences between measured and derived traits were highly significant (P < 0·001). Submerged leaves showed an elongated shape, and their average leaf L/W was 4·6 times larger than that of aerial leaves (Table 2). Leaf-tip indices were positive (U-shaped) and very slightly negative (slightly protruding) on average for submerged and aerial leaves, respectively. Epidermal cells were also elongated in submerged leaves compared with those in aerial leaves, and cell shapes were 3·0 and 3·8 times elongated on average for upper and lower surfaces, respectively, in submerged leaves. Stomata were completely absent in submerged leaves, but high densities of stomata were found for both the upper (143 mm⁻²) and lower (125 mm⁻²) surfaces of aerial leaves.

Table 2.

Leaf morphology of Rotala hippuris growing under aerial and submerged conditions under natural light regimes in outside growth garden

Leaf morphological trait	Abbreviation	Aerial (n = 12)	Submerged (n = 12)	t
Leaf length (mm)		5·9  ±  1·7	15·4  ±  2·2	–12·1***
Leaf width (mm)		0·65  ±  0·14	0·38  ±  0·087	5·8***
Leaf length/leaf width	Leaf L/W	9·5  ±  3·9	44  ±  14	–8·2***
Width of leaf-tip structure (mm)		0·17  ±  0·020	0·12  ±  0·021	6·1***
Depth (+)/height (–) of leaf-tip structure (mm)		–0·002  ±  0·017	0·14  ±  0·021	–17·7***
Index of leaf-tip shape	Leaf-tip index	–0·01  ±  0·10	1·2  ±  0·2	–16·7***
Length of epidermal cells on leaf upper surface (μm)		0·059  ±  0·011	0·100  ±  0·013	–8·3***
Width of epidermal cells on leaf upper surface (μm)		0·029  ±  0·007	0·014  ±  0·002	7·1***
L/W of epidermal cells on leaf upper surface	Upper cell shape	2·2  ±  0·7	7·1  ±  1·7	–9·3***
Length of epidermal cells on leaf lower surface (μm)		0·052  ±  0·009	0·103  ±  0·024	–7·0***
Width of epidermal cells on leaf lower surface (μm)		0·026  ±  0·008	0·014  ±  0·004	5·0***
L/W of epidermal cells on leaf lower surface	Lower cell shape	2·2  ±  1·0	8·4  ±  3·2	–6·3***
Stomata density on leaf upper surface (mm−2)	Upper stomata density	142·7  ±  30·1	0  ±  0	16·4***
Stomata density on leaf lower surface (mm−2)	Lower stomata density	125·3  ±  20·3	0  ±  0	21·4***

Values shown are mean ± s.d. The t values for the t-test are also shown and asterisks indicate significant differences in mean values between aerial and submerged leaves (***P < 0·001). The six summary traits shown in bold were used to evaluate leaf responses to R/FR in the experiments in this study.

Response to a wide range of R/FR (Experiment 1)

All six analysed traits showed statistically significant responses to changes of R/FR and aerial and submerged treatments (two-way ANOVAs, data not shown). In the separate analyses for each of the submerged and aerial conditions, all traits showed statistically significant responses to R/FR in either of the conditions (Table 3). Under submerged conditions, R/FR lower than 0·1 caused responses resulting in trait values more typical of aerial leaves, namely non-elongated leaf shape (low leaf L/W), low leaf-tip index, non-elongated cell shapes and formation of stomata on both surfaces of leaves (Fig. 2). In contrast, under aerial conditions, R/FR higher than 0·1 caused responses resulting in traits more typical of submerged leaves, namely elongated leaf shape (high leaf L/W), high leaf-tip index, elongated cell shapes and reduction in stomata densities on both surfaces of leaves (Fig. 2). Under aerial conditions, the number of stomata also increased in response to extremely low R/FR, and stomata that exceeded naturally observed densities were formed on both surfaces of leaves at R/FR of 0·013 (Fig. 2E, F).

Table 3.

One-way ANOVA tables for the six leaf traits across five R/FR treatments in Experiment 1

Trait	Growth condition	Factor	d.f.	SS	MS	F	P
Leaf L/W	Aerial	R/FR	4	6·12	1·53	135·2	<0·0001
		Residuals	25	0·29	0·011
	Submerged	R/FR	4	14·3	3·57	346·9	<0·0001
		Residuals	25	0·26	0·01
Leaf-tip index	Aerial	R/FR	4	0·32	0·08	20·6	<0·0001
		Residuals	25	0·097	0·004
	Submerged	R/FR	4	2·02	0·51	27·3	<0·0001
		Residuals	25	0·46	0·019
Upper cell shape	Aerial	R/FR	4	3·38	0·85	18·9	<0·0001
		Residuals	25	1·12	0·045
	Submerged	R/FR	4	4·95	1·24	27·6	<0·0001
		Residuals	25	1·12	0·045
Lower cell shape	Aerial	R/FR	4	5·91	1·48	52·3	<0·0001
		Residuals	25	0·71	0·028
	Submerged	R/FR	4	7·91	1·98	23·5	<0·0001
		Residuals	25	2·1	0·084
Upper stomata density	Aerial	R/FR	4	523·2	130·8	74·9	<0·0001
		Residuals	25	43·6	1·75
	Submerged	R/FR	4	272·6	68·2	60	<0·0001
		Residuals	25	28·4	1·14
Lower stomata density	Aerial	R/FR	4	652	163	78·9	<0·0001
		Residuals	25	51·6	2·07
	Submerged	R/FR	4	200·1	50	21·1	<0·0001
		Residuals	25	59·4	2·38

The data under aerial and submerged conditions were analysed separately. Leaf L/W was log transformed and the other five traits were square-root transformed in these analyses.

Fig. 2.

Responses of six leaf traits of Rotala hippuris grown in submerged and aerial conditions (as indicated in the key) with a wide range of R/FR (0·0013, 0·095, 0·69, 4·6 and 791; note the log scale) in Experiment 1. The responses of leaf L/W (A), leaf-tip index (B), upper cell shape (C), lower cell shape (D), upper stomata density (E) and lower stomata density (F) are shown. Bars represent the s.d. Typical values for submerged and aerial leaves (means in Table 2) are shown by closed and open arrowheads, respectively. Different letters next to symbols indicate significant differences (P < 0·05) in mean values between the R/FR treatments in Scheffe's multiple comparison tests conducted separately for the aerial and submerged conditions (a–d and α–δ, respectively).

Response to blue-light intensities under different R/FR (Experiment 2)

Under high R/FR, leaves showed no response to blue light, and the six measured traits showed values that were similar to those of submerged leaves irrespective of blue-light intensity. At intermediate and low R/FR, responses to blue light were seen, and statistically significant interactions of blue light and R/FR were detected for all of the six analysed traits except for leaf-tip index (Fig. 3, Table 4). Leaf-tip index showed a significant response to R/FR, but terms including blue light effect were not significant in the two-way ANOVA (Table 4). Marked effects of blue light were seen at low and intermediate R/FR for upper and lower cell shapes and stomata densities, and higher intensity resulted in less-elongated cell shapes and increased production of stomata for the leaves grown under submerged conditions (Fig. 3). At intermediate R/FR, quantitative responses to blue-light intensity were observed in upper cell shapes and upper stomata densities. At low R/FR, intermediate and high intensities of blue light caused similar responses in upper and lower cell shapes and stomata densities.

Fig. 3.

Responses of six leaf traits of Rotala hippuris to three R/FR (0·2, 0·9, and 1·7) when grown under three different levels of blue-light intensity (as indicated in the key) in Experiment 2. The plants were grown under submerged conditions. The responses of leaf L/W (A), leaf-tip index (B), upper cell shape (C), lower cell shape (D), upper stomata density (E) and lower stomata density (F) are shown. Bars represent the s.d.

Table 4.

Two-way ANOVA tables for the responses of six leaf traits in Experiment 2

Trait	Factor	d.f.	SS	MS	F	P
Leaf L/W	R/FR	2	4·78	2·39	10·3	0·026
	Blue light	2	0·037	0·019	0·082	0·92
	R/FR × blue light	4	0·92	0·23	3·33	0·018
	Residuals	45	3·1	0·069
Leaf-tip index	R/FR	2	0·19	0·094	29·6	< 0·0001
	Blue light	2	0·007	0·003	1·03	0·36
	R/FR × blue light	4	0·049	0·012	3·8	0·091
	Residuals	45	0·14	0·003
Upper cell shape	R/FR	2	6·62	3·31	8·58	0·036
	Blue light	2	2·47	1·23	3·19	0·15
	R/FR × blue light	4	1·54	0·39	8·2	< 0·0001
	Residuals	45	2·11	0·47
Lower cell shape	R/FR	2	5·56	2·78	7·92	0·041
	Blue light	2	2·81	1·4	4·00	0·11
	R/FR × blue light	4	1·41	0·35	5·99	0·0006
	Residuals	45	2·64	0·059
Upper stomata density	R/FR	2	300	150	10·5	0·025
	Blue light	2	219·5	109·7	7·71	0·042
	R/FR × blue light	4	56·9	14·2	5·05	0·0019
	Residuals	45	126·7	2·82
Lower stomata density	R/FR	2	263·8	132	4·04	0·11
	Blue light	2	186·1	93	2·85	0·17
	R/FR × blue light	4	130·5	32·6	7·37	0·0001
	Residuals	45	199·1	4·43

The effects of R/FR, blue light and their interaction were examined. Leaf L/W was log transformed and the other five traits were square-root transformed in these analyses. For the traits for which significant interaction terms were detected, main factors were tested against the interaction term. Probability levels <0·05 are shown in bold.

DISCUSSION

Under the natural light regime in the outside growth garden, the clone of R. hippuris used in this study produced distinct submerged and aerial leaves. The measured values for leaf size are equivalent with those reported for specimens collected in natural habitats of the study species (Kadono, 1994). Therefore, we judged that leaf characteristics of the clone were more or less equivalent to those found in natural habitats, and we used these values as those typical of submerged and aerial leaves of the study species when interpreting the results of the growth experiments.

The results clearly showed that R/FR is one of the major determinants of heterophyllous leaf formation of R. hippuris. Higher and lower R/FR caused responses of the leaf characters more typical of those of submerged and aerial leaves, respectively, under both aerial and submerged conditions. Similar responses of heterophyllous amphibious plants have been reported in submerged leaves of Marsilea vestila (Gaudet, 1963, 1965) and Hippuris vulgaris (Bodkin et al., 1980). Our results showed that the responses occur in leaves not only under water but also under aerial conditions. The observed strong responses to R/FR are in agreement with the idea that heterophyllous plants perceive R/FR as an index of depth (Spence, 1981; Smith, 1982, 1994), although underwater light quality in natural conditions can be highly variable (Smith, 1994).

Although R. hiipuris responded to R/FR under aerial and submerged conditions, differences of character values between the conditions at a given R/FR value were often great. The responses were greater under submerged conditions for leaf, leaf-tip and cell shapes than those under aerial conditions, and vice versa for stomata density. These observations suggested that not only R/FR but also other cues provided by submergence, such as osmotic stress and CO₂ concentration, were operating in determining leaf morphology (Wells and Pigliucci, 2000).

Four traits, i.e. leaf L/W, leaf-tip index, and upper and lower cell shapes, showed either quantitative or threshold-type responses depending on R/FR: steep responses between R/FR of 0·95 and 4·6 and a weak or no response to more extreme R/FR, such as 0·0013 and 791·3. The R/FR of daylight is approx. 1–1·1, but can decrease to 0·1 under vegetation, depending on canopy density (Smith, 1982). Assuming R/FR is 1 at the water surface, the expected depth in pure water at which R/FR = 4·6 is about 1·5 m. Measurements of R/FR in natural lakes (Smith, 1982; Chambers and Spence, 1984) have suggested that an R/FR of 4·6 corresponded to that at a water depth of 1–1·5 m. The range of R/FR at which R. hippuris showed high sensitivity of the four traits corresponds to that which is experienced by the plants in natural conditions. Stomata densities showed quantitative responses to R/FR of 4·6 and lower, and they exceeded the natural range of stomata densities at R/FR of 0·0013 under aerial conditions. The response to the extremely low R/FR may be a by-product of the underlying regulatory mechanism of stomata density.

Underwater leaves of R. hippuris responded to blue-light intensity in all of the six analysed traits except for leaf-tip index. High blue light caused a shift of trait values toward those of typical aerial leaves. The results suggested that the response to blue-light intensity is important in determining leaf developmental fate near the water surface. A quantitative response to the three levels of blue-light intensity investigated is observed at R/FR below 0·9, which is close to the value expected at the water surface in natural conditions. The responses to blue light were weak or absent at higher R/FR (1·7; expected at about 50 cm depth under pure water with an R/FR of 1·0 at the surface), suggesting a dominance of R/FR over blue light as a developmental cue below the water surface. At low R/FR (0·5), the value expected under aerial conditions, even weak blue light was effective for producing aerial-type leaves under submerged conditions, and blue light was found to be necessary for this response, especially regarding cell shapes and stomata densities. In the aquatic fern Marsilea quadrifolia, it has been reported that application of blue light induced the development of aerial-type leaves (Lin and Yang, 1999). Our study indicated that blue-light intensity can be a critical cue of heterophyllous leaf development also for aquatic seed plants.

In conclusion, our experiments clearly showed that R/FR and blue-light intensity provide quantitative cues for R. hippuris to determine water depth, especially near the water surface. The utilization of these quantitative cues is likely to make it possible for R. hippuris to prepare aerial leaves at the shoot apical meristem just beneath the water surface prior to its emergence from the water. Dosage-dependent regulation of leaf morphogenesis is expected to be important, especially in the habitats of many amphibious plants where plants experience water-level fluctuation. Further study will be required to determine whether the response to light quality found in this study is a general one among other heterophyllous plants. Furthermore, we need to compare the response to light quality between plants that grow in terrestrial and aquatic habitats. Blue light and R/FR are known to be major cues of the morphogenesis of many terrestrial plants, because they inform about the degree of shade by surrounding vegetation (Franklin, 2008). Interestingly, the correlation of R/FR with the level of shade is reversed between terrestrial and aquatic conditions, i.e. low R/FR and high R/FR are indicative of low PAR levels in aerial and underwater conditions, respectively. It has been reported recently that stomata density of Arabidopsis thaliana decreased in response to low R/FR (Boccalandro et al., 2009), and the direction of the response was opposite of what we found here. Comparative studies on the mechanisms underlying these responses should help to explain the evolution of these opposite responses to the same environmental cue.