Aerenchyma and an Inducible Barrier to Radial Oxygen Loss Facilitate Root Aeration in Upland, Paddy and Deep‐water Rice (Oryza sativa L.)

Abstract

The present study evaluated waterlogging tolerance, root porosity and radial O₂ loss (ROL) from the adventitious roots, of seven upland, three paddy, and two deep‐water genotypes of rice (Oryza sativa L.). Upland types, with the exception of one genotype, were as tolerant of 30 d soil waterlogging as the paddy and deep‐water types. In all but one of the 12 genotypes, the number of adventitious roots per stem increased for plants grown in waterlogged, compared with drained, soil. When grown in stagnant deoxygenated nutrient solution, genotypic variation was evident for root porosity and rates of ROL, but there was no overall difference between plants from the three cultural types. Adventitious root porosity increased from 20–26 % for plants grown in aerated solution to 29–41 % for plants grown in stagnant solution. Growth in stagnant solution also induced a ‘tight’ barrier to ROL in the basal regions of adventitious roots of five of the seven upland types, all three paddy types, and the two deep‐water types. The enhanced porosity provided a low resistance pathway for O₂ movement to the root tip, and the barrier to ROL in basal zones would have further enhanced longitudinal O₂ diffusion towards the apex, by diminishing losses to the rhizosphere. The plasticity in root physiology, as described above, presumably contributes to the ability of rice to grow in diverse environments that differ markedly in soil waterlogging, such as drained upland soils as well as waterlogged paddy fields.

INTRODUCTION

Rice (Oryza sativa L.) is cultivated in environments characterized by marked differences in depth (several metres to absence) and duration (entire growing season, transient, to absence) of soil waterlogging or flooding (Moormann and van Breemen, 1978; O’Toole and Chang, 1979; Grist, 1986). Cultivars adapted to these broad environments have been bred, resulting in deep‐water, paddy and upland cultural types (O’Toole and Chang, 1979; Khush, 1997). Breeding for conditions at more regional and local scales has also occurred (Maurya et al., 1988; Khush, 1997), so that tens of thousands of varieties of rice exist throughout the world. This diversity may provide a resource for ecophysiological studies of plant adaptation to selected environments, e.g. submergence tolerance (Setter and Laureles, 1996; Sauter, 2000) and drought resistance (O’Toole and Chang, 1979; Puckridge and O’Toole, 1981; Champoux et al., 1995). Comparisons of the physiology of upland and lowland rice may also provide a model system for understanding mechanisms of waterlogging tolerance in plants (Colmer et al., 1998).

Genotypes for upland culture are postulated to have been selected from lowland ancestors soon after rice was first cultivated some 4700 years ago (Chang, 1976a, b). Although the term ‘upland rice’ is used widely, it encompasses wide cultivation practices ranging from rain‐fed crops in non‐bunded fields, to dry‐seeded crops that mature in flooded fields (IRRI, 1975), and therefore may not necessarily reflect the hydrological situation determining the availability of water to the crop (Moormann and van Breemen, 1978; Grist, 1986). In fact, while restricted water supply may limit yields of upland rice in many areas (IRRI, 1975, 1982), episodes of waterlogging are also common in upland soils (Moormann and van Breemen, 1978; Grist, 1986). Similarly, rice cultivated in some deep‐water and non‐irrigated paddy systems can encounter periods of low soil water availability, depending on geographic location and seasonal rainfall patterns (Moormann and van Breemen, 1978; Grist, 1986). Therefore, although cultivars with improved adaptation to the broadly defined upland, paddy and deep‐water environments are recognized (O’Toole and Chang, 1979; Khush, 1997), many characteristics vary continuously, with considerable overlap, between lowland and upland cultivars of rice (Chang et al., 1972).

Waterlogged soils are usually anaerobic, and can be chemically reduced (Ponnamperuma, 1984). The roots of rice (Barber et al., 1962; Armstrong, 1971; Clark and Harris, 1981) like those of many other wetland species (Smirnoff and Crawford, 1983; Justin and Armstrong, 1987), contain aerenchyma; the amount of which in rice is enhanced upon exposure to soil waterlogging (Armstrong, 1971; Pradhan et al., 1973; Das and Jat, 1977). The aerenchyma provides a low resistance internal pathway for movement of O₂ (and other gases) within the roots. In addition to extensive aerenchyma, the roots of many wetland species also contain a barrier to radial O₂ loss (ROL) in the basal zones (Armstrong, 1964, 1979; Jackson and Armstrong, 1999; Visser et al., 2000; McDonald et al., 2002). These traits act synergistically to enhance O₂ diffusion to the root tip, and thus root elongation into anaerobic substrates (Armstrong, 1979) and enable an aerobic rhizosphere around the root tip. These features presumably contribute to the waterlogging tolerance of wetland species (Armstrong, 1979; Jackson and Drew, 1984; Justin and Armstrong, 1987), including rice (Armstrong 1971; Green and Etherington, 1977; Colmer et al., 1998).

Growth in stagnant deoxygenated nutrient solution enhanced both root porosity and development of a barrier to ROL in adventitious roots of rice (Colmer et al., 1998). The barrier to ROL was most pronounced in the deep‐water and paddy cultivars and least in the two upland cultivars studied. However, we cautioned against drawing general conclusions on the possible differences between cultural types before more genotypes were tested. Nevertheless, some cultivars of upland rice may have a lower capacity for internal gas transport than lowland cultivars (Lee et al., 1981).

The experiments reported in the present paper were designed to address the question, how widespread is the capacity for the induction of increased porosity and a barrier to ROL in adventitious roots of upland, paddy and deep‐water genotypes of rice? Seven upland, three paddy and two deep‐water genotypes were evaluated for root porosity, patterns of ROL, root penetration into an O₂‐free medium, and growth in drained or waterlogged soil. Improved knowledge on this topic should aid understanding of the functioning of root systems of rice cultured under different flooding regimes. In addition, traits determining root aeration may also influence nutrient dynamics in the rhizosphere (Reddy and Patrick, 1986) and escape of methane from the soil via rice plants (Kludze et al., 1993; Jespersen et al., 1998), in flooded soils.

MATERIALS AND METHODS

Plant material

Twelve genotypes (seven upland, three paddy and two deep‐water types) of rice (Oryza sativa L.) were studied (Table 1).

Table 1.

Rice genotypes (Oryza sativa L.) used in the present study

Genotype	Cultivation type	Origin
Omirt 39	Upland	Hungary
Omirt 168	Upland	Hungary
63‐83	Upland	Senegal
Sensho	Upland	Japan
IAC 1131	Upland	Brazil
Fukerupagai II	Upland	Taiwan
IR3880‐5	Upland	Philippines (IRRI)
Calrose	Paddy	USA
Basmati 370	Paddy	Pakistan
Basmati (Kurnool)	Paddy	India
IR442‐57	Deep‐water	Philippines (IRRI)
Leb Mue Nahng III	Deep‐water	Thailand

Selection of the upland and deep‐water types was determined by those available in the rice germplasm collection at Yanco Agricultural Institute (NSW Agriculture), Australia.

All seeds (except Calrose) were originally obtained from the International Rice Research Institute (IRRI), the Philippines.

Experiment in pots of soil

Seeds were surface‐sterilized with 0·4 % (w/v) sodium hypochlorite for 30 s and rinsed thoroughly with deionized water. They were then placed on a plastic mesh floating on 0·5 mol m^–3 CaSO₄ in a container covered with aluminium foil so that they were in darkness. The experiment was conducted in a 25/20 °C (day/night) phytotron. The flux of photosynthetically active radiation at shoot height in the phytotron at midday during the experimental period was 700–900 µmol m^–2 s^–1.

After 24 h, seeds were sown at a depth of 7·5 mm into soil in PVC pots (450 mm high; 150 mm diameter) each containing 7 kg of soil placed over a 100‐mm layer of gravel. Six seeds of the same genotype were sown into each pot, and six pots of each genotype were established. The soil was Waroona sandy clay loam (McArthur et al., 1959) with the following nutrients pre‐mixed through the soil (mg kg^–1): KH₂PO₄, 334; K₂SO₄, 143; NH₄NO₃, 200; CaCl_2·2H₂O, 143; MgSO_4·7H₂O, 22; CuSO_4·5H₂O, 5; ZnSO_4·7H₂O, 5; Na₂MoO_4·2H₂O, 0·4; H₃BO₃, 0·4; MnSO₄.H₂O, 0·4. The same soil type and similar nutrient additions have been used by others for experiments on rice (Setter et al., 1989). The soil was maintained at field capacity (32 %, w/w) by adding deionized water each day. The pots were completely randomized on a bench and re‐randomized to new positions after each watering.

Five days after imbibition, plants were thinned to three per pot and the treatments were imposed. Three pots of each genotype were waterlogged by partially submerging these in tanks so that deionized water entered the bottom of each pot and rose to about 10 mm above the soil surface. A transparent plastic tube attached to the drainage hole and vertically along the side of each pot prevented drainage of water from the waterlogged pots, and the water level was maintained at 10 mm above the soil surface. The soil in the other three pots of each genotype was maintained at field capacity by daily additions of deionized water.

Thirty‐five days after imbibition, the plants were harvested. Shoots were excised at the soil surface and the roots were washed free of soil. The numbers of tillers, numbers of adventitious roots, and length of the longest adventitious root (waterlogged only) were recorded for each plant prior to the tissues being oven dried at 60 °C for at least 3 d before dry mass was determined. The mean of the plants in each pot was used as the value for one replicate in the data analyses.

Experiment in nutrient solution

Seeds were surface sterilized with 0·4 % (w/v) sodium hypochlorite for 30 s and rinsed thoroughly with deionized water. Seeds were then imbibed in aerated 0·5 mol m^–3 CaSO₄ for 3 h before being placed on plastic mesh floating on aerated 0·1 strength nutrient solution in a container covered with aluminium foil. The experiment was conducted in the same 25/20 °C (day/night) phytotron as described above for the experiment with plants grown in pots of soil.

The composition of the nutrient solution at full concentration (mol m^–3) was: K⁺, 3·95; Ca²⁺, 1·50; Mg²⁺, 0·40; NH₄⁺, 0·625; NO₃^–, 4·375; SO₄^2–, 1·90; H₂PO₄^–, 0·20; Na⁺, 0·20; H₄SiO₄, 0·10; and the micronutrients (mmol m^–3) Cl, 50; B, 25; Mn, 2; Zn, 2; Ni, 1; Cu, 0·5; Mo, 0·5; Fe‐EDTA, 50. The solution also contained 2·5 mol m^–3 MES and the pH was adjusted to 6·5 using KOH, increasing the final K⁺ concentration to 5·6 mol m^–3. All chemicals used were analytical grade.

Five days after imbibition the seedlings were exposed to light and 0·25 strength nutrient solution. Seven days after imbibition the nutrient solution was increased to full strength. The next day, seedlings were transplanted into 4·5‐l pots (five seedlings of each genotype per pot; six pots of each genotype) containing aerated nutrient solution. Ten days after imbibition, an initial harvest of one plant was taken from each pot, and the treatments were imposed. Half of the pots continued to be well aerated and solution in these was renewed, while the solution in the other pots was replaced with deoxygenated nutrient solution containing 0·1 % (w/v) agar. The dilute agar prevents convective movements in the solution (‘stagnant’ treatment) so that it may mimic the changes in gas composition found in waterlogged soils (e.g. decreased O₂, increased ethylene) better than other methods used to impose root‐zone O₂ deficiency in solution culture (Wiengweera et al., 1997). Pots were arranged in a completely randomized design and re‐randomized every 7 d when all the solutions were renewed.

Twenty‐eight to thirty‐eight days after imbibition, plants were used for measurements of rates of ROL from intact adventitious roots (see below). A first block of measurements was taken for one plant from each genotype × treatment combination, followed by a second block, and finally a third block. The plant used from each pot was chosen randomly, and the order of the genotype × treatment combinations measured in each block was also random.

Thirty‐three days after imbibition, one adventitious root on one plant in each pot was marked near the base with xylene‐free ink and the length of this root was measured then, and again 24 h later, so that rates of root extension were determined. Roots selected for these measurements were of similar lengths to those used in the ROL measurements. Thirty‐five days after imbibition, the length of the longest adventitious root on each plant was recorded, and the porosity of selected adventitious roots was measured (see below).

Root porosity measurements

Porosity (% gas spaces per unit tissue volume) was measured on samples of adventitious roots from each genotype by determining root buoyancy before and after vacuum infiltration of the gas spaces in the roots with water (Raskin, 1983), using the equations as modified by Thomson et al. (1990). For two 35‐d‐old plants taken from each pot, the adventitious roots were excised and those 100–150 mm in length were selected, cut into 50‐mm segments, and samples of about 1·5 g fresh mass were used in the measurements.

ROL measurements

Rates of ROL at selected distances behind the apex of intact adventitious roots when in an O₂‐free root medium were measured using cylindrical root‐sleeving O₂ electrodes (Armstrong and Wright, 1975; Armstrong, 1994). Root systems of intact 28–38‐d‐old plants were immersed in transparent chambers (50 × 50 × 250 mm; breadth × width × height) containing deoxygenated solution of composition: 0·1 % (w/v) agar and (in mol m^–3) K⁺, 5·0; Cl^–, 5·0; Ca²⁺, 0·5; SO₄^2–, 0·5. The shoot base was held with wet cotton wool in a rubber lid sealed on to the top of each chamber, so that the shoots were in air. The chambers and plants were located in a 25 °C constant temperature room with PAR (photosynthetically active radiation) at shoot height of 150 µmol m^–2 s^–1.

For each plant transferred into the experimental system, one adventitious root (100–131 mm in length) was inserted through a cylindrical O₂ electrode (i.d. 2·25 mm, height 5·0 mm) fitted with guides to keep the root near the centre of the electrode. Plants were left for at least 2 h prior to the first ROL recordings; measurements were then taken along each root with the centre of the electrode positioned at 5, 10, 20, 30, 40 or 50 mm behind the root apex. Root diameters at these positions along the roots were then measured using a vernier microscope. One intact root on three individual plants from each treatment (from different replicate pots) was measured to provide three replicates of each genotype and treatment combination.

Statistical analyses

Data on dry masses, tiller numbers, adventitious root numbers, maximum root lengths and root porosity were analysed using two‐way analysis of variance (ANOVA) to determine the main effects and interactions among genotype (or in separate analyses cultural type) and treatments on the variables measured. Data on ROL were analysed using three‐way ANOVA to determine the main effects and interactions among genotype (or in separate analyses cultural type), treatments, and position measured along roots. Means were compared using least significant differences (l.s.d.s) at the P = 0·05 level. The complete sets of statistics from these analyses are not shown in order to maintain brevity; key summary statistics (i.e. P values and l.s.d.s) are presented in appropriate places in the Results section.

RESULTS

Growth of rice in drained or waterlogged soil

Shoot dry mass achieved by the 35th day after germination differed among several of the genotypes (Table 2; P < 0·005). Waterlogging for the final 30 d had no effect on, or slightly increased, shoot dry mass in five of the seven upland, all three paddy, and one of the two deep‐water genotypes, when compared with plants in drained soil (P < 0·05 for genotype × treatment interaction) (Table 2). The largest increases in shoot dry mass in waterlogged plants were the 31 % increase in one of the upland types (Omirt 39) and the 22 % increase in one of the paddy types (Basmati 370). The largest reductions caused by waterlogging were the 34 % decrease in the upland type Sensho, but shoot dry mass was also decreased by 22 % in the deep‐water type Leb Mue Nahng III. Thus, there was no significant difference in the effect of waterlogging on shoot dry mass for the collective responses of genotypes from the three different cultural types (P = 0·38).

Table 2.

Shoot and root dry mass of 12 genotypes of rice grown in drained or waterlogged soil

	Shoot dry mass (g)		Root dry mass (g)
Genotype	Drained	Waterlogged	Drained	Waterlogged
Omirt 39	1·24 ± 0·06	1·61 ± 0·19	0·32 ± 0·03	0·52 ± 0·05
Omirt 168	1·40 ± 0·09	1·70 ± 0·09	0·33 ± 0·06	0·40 ± 0·02
63‐83	1·30 ± 0·01	1·27 ± 0·12	0·41 ± 0·04	0·37 ± 0·07
Sensho	1·32 ± 0·18	0·87 ± 0·15	0·35 ± 0·02	0·22 ± 0·03
IAC 1131	1·08 ± 0·06	1·04 ± 0·06	0·31 ± 0·03	0·26 ± 0·01
Fukerupagai II	1·22 ± 0·26	1·05 ± 0·12	0·42 ± 0·06	0·29 ± 0·03
IR3880‐5	1·36 ± 0·14	1·20 ± 0·02	0·31 ± 0·05	0·37 ± 0·04
Calrose	0·89 ± 0·07	0·80 ± 0·04	0·19 ± 0·02	0·22 ± 0·01
Basmati 370	1·06 ± 0·14	1·29 ± 0·06	0·33 ± 0·04	0·38 ± 0·15
Basmati (Kurnool)	1·08 ± 0·01	1·26 ± 0·13	0·38 ± 0·15	0·36 ± 0·06
IR442‐57	1·28 ± 0·10	1·24 ± 0·07	0·42 ± 0·05	0·35 ± 0·02
Leb Mue Nahng III	1·25 ± 0·11	0·97 ± 0·12	0·50 ± 0·07	0·31 ± 0·07

Plants grown were raised for 5 d in soil at field capacity prior to the treatments being imposed for the final 30 d. Values given are per plant and are means of three replicates ± standard errors.

l.s.d._{(P < 0·05)} for genotype comparisons of: shoot dry mass, 0·244; root dry mass, 0·099.

Root dry mass was maintained, or even increased relative to plants in drained soil, in five of the seven upland, all three paddy, and one of the deep‐water genotypes when waterlogged for 30 d (P < 0·05 for genotype × treatment interaction) (Table 2). Differences in the effects of waterlogging on root dry mass for the collective responses of genotypes from the three different cultural types were only statistically significant at the 10 % level.

Waterlogging increased the numbers of adventitious roots per plant 1·1–2·5‐fold in the upland types, 1·8–2·5‐fold in the paddy types and 1·4–1·5‐fold in the deep‐water types, relative to plants in drained soil (P < 0·001 for effects of genotype, treatment and their interaction; Table 3). The increased numbers of adventitious roots would have contributed to the maintenance of, or in some cases increase in, root dry mass in several of the genotypes when grown in waterlogged, compared with drained soil (Table 2). Tillering was not affected by waterlogging in most genotypes (Table 3); however, tiller numbers were increased by at most one additional stem in some genotypes (e.g. the upland type Omirt 39, and the paddy types Calrose and Basmati 370), whereas it was decreased in others (e.g. the upland type Sensho, and the deep‐water type Leb Mue Nahng III). Comparison of the collective responses of the three cultural types showed that waterlogging did not effect tiller numbers in the upland or paddy types, but it reduced tillering in the deep‐water types (P < 0·05 for cultural type × treatment interaction). Thus, the numbers of adventitious roots per stem (main stem plus tillers) generally increased when plants were grown in waterlogged, compared with drained, soil.

Table 3.

Number of tillers (not including main stem) and numbers of adventitious roots per plant for 12 genotypes of rice grown in drained or waterlogged soil

	Tillers per plant		Adventitious roots per plant
Genotype	Drained	Waterlogged	Drained	Waterlogged
Omirt 39	3·3 ± 0·33	4·3 ± 0·19	42 ± 6·2	102 ± 4·7
Omirt 168	3·3 ± 0·69	4·0 ± 0·69	38 ± 2·3	71 ± 5·0
63‐83	3·2 ± 0·29	3·4 ± 0·44	33 ± 7·6	35 ± 2·9
Sensho	3·5 ± 0·29	2·8 ± 0·44	29 ± 2·3	38 ± 6·1
IAC 1131	2·9 ± 0·11	3·1 ± 0·48	18 ± 0·7	33 ± 2·2
Fukerupagai II	3·9 ± 0·80	3·9 ± 0·22	20 ± 2·6	51 ± 1·3
IR3880‐5	3·9 ± 0·40	4·0 ± 0·51	40 ± 4·1	53 ± 9·5
Calrose	2·1 ± 0·29	3·0 ± 0·19	18 ± 1·7	46 ± 1·7
Basmati 370	3·8 ± 0·29	4·9 ± 0·29	32 ± 2·2	57 ± 3·0
Basmati (Kurnool)	3·5 ± 0·17	3·9 ± 0·29	23 ± 0·3	47 ± 3·3
IR442‐57	3·6 ± 0·40	2·9 ± 0·11	35 ± 6·8	53 ± 5·1
Leb Mue Nahng III	3·2 ± 0·40	2·1 ± 0·29	32 ± 1·5	44 ± 6·1

Plants were raised for 5 d in soil at field capacity prior to the treatments being imposed for the final 30 d.

Values given are per plant and are means of three replicates ± standard errors.

l.s.d._{(P < 0·05)} for genotype comparisons of: tiller numbers, 0·81; adventitious root numbers, 9·4.

In summary, with the exception of Sensho, the other upland types were generally as tolerant of the 30 d of soil waterlogging as the paddy and deep‐water types evaluated in this study.

Penetration of adventitious roots of rice into waterlogged soil or stagnant deoxygenated nutrient solution

In waterlogged soil the average length of the longest adventitious roots ranged from 180 to 249 mm for the upland types, 168 to 237 mm for the paddy types and 182 to 212 mm for the deep‐water types (Table 4). With the exception of two genotypes [upland type IAC 1131 and paddy type Basmati (Kurnool)], the longest lengths of adventitious roots were 31–121 mm longer in the stagnant solution than in the waterlogged soil (Table 4). Nevertheless, adventitious root lengths for plants in the stagnant solution were 38–285 mm shorter than the lengths of the longest roots for the same genotypes when grown in aerated solution (P < 0·001; Table 4). Comparison of the collective responses of maximum root lengths for the three cultural types showed that the cultural type × treatment interaction for plants grown in the solution culture experiment was only significant at the 10 % level. Thus, root penetration into O₂‐deficient media (waterlogged soil or stagnant deoxygenated nutrient solution) was in general no worse in the upland types, when compared with the paddy or deep‐water types.

Table 4.

Lengths of the longest adventitious roots for 12 genotypes of rice grown in waterlogged soil or stagnant deoxygenated nutrient solution

	Length of the longest adventitious root (mm)
	Soil		Nutrient solution
Genotype	Drained	Waterlogged	Aerated	Stagnant
Omirt 39	n.d.	225 ± 5·5	331 ± 16·5	268 ± 19·1
Omirt 168	n.d.	249 ± 9·5	359 ± 15·9	314 ± 10·2
63‐83	n.d.	219 ± 6·2	490 ± 2·7	263 ± 0·8
Sensho	n.d.	191 ± 14·7	480 ± 27·1	255 ± 20·8
IAC 1131	n.d.	208 ± 9·6	502 ± 27·3	217 ± 4·4
Fukerupagai II	n.d.	200 ± 9·2	549 ± 27·4	302 ± 5·6
IR3880‐5	n.d.	180 ± 30·8	418 ± 25·2	256 ± 14·4
Calrose	n.d.	168 ± 29·0	327 ± 12·6	289 ± 16·0
Basmati 370	n.d.	206 ± 9·4	299 ± 40·6	237 ± 10·5
Basmati (Kurnool)	n.d.	237 ± 11·4	458 ± 29·6	233 ± 15·3
IR442‐57	n.d.	212 ± 2·4	409 ± 21·2	283 ± 7·9
Leb Mue Nahng III	n.d.	182 ± 15·2	415 ± 2·3	266 ± 5·5

Plants in soil were raised for 5 d at field capacity prior to the treatments being imposed for the final 30 d.

Plants in nutrient solution were aerated for the first 10 d and then roots were exposed to either aerated or stagnant deoxygenated solution for the final 25 d.

Values given are means of three replicates ± standard errors.

n.d. = not determined.

l.s.d._{(P < 0·05)} for genotype comparisons of the lengths of longest roots, 38 (solution experiment only).

Porosity of adventitious roots of rice grown in aerated or stagnant deoxygenated nutrient solution

Porosity in adventitious roots of rice grown in aerated solution ranged from 19·7 to 25·9 % (Table 5). Growth in stagnant solution increased root porosity in all genotypes to values of 28·9–40·8 % (P < 0·001 for treatment effect). Therefore, although porosity differed among several genotypes (P < 0·001), the genotype × treatment interaction was only significant at the 10 % level (P = 0·08). The 1·2–1·8‐fold increases in porosity for adventitious roots of individual genotypes when grown in stagnant solution (Table 5) were confirmed in a second experiment (data not shown). In the second experiment plants were grown under the same conditions (except 2 months later) and adventitious root porosity was 20·5–27·7 % for aerated plants and it increased to 27·0–39·8 % for plants grown in stagnant solution (data not shown). In both experiments, the upland types IAC 1131 and IR3880‐5 had the lowest adventitious root porosity and the upland type Fukerupagai II and deep‐water type IR442‐57 had the highest, when grown in stagnant solution. Overall, porosity in the adventitious roots of most of the upland types was similar to the values in the paddy types (Table 5), so that there was no overall difference in porosity between the three cultural types of rice (P = 0·29).

Table 5.

Porosity of adventitious roots of 12 genotypes of rice grown in aerated or stagnant deoxygenated nutrient solution

	Porosity (% gas spaces per unit tissue volume)
Genotype	Aerated	Stagnant
Omirt 39	23·1 ± 1·2	34·6 ± 3·1
Omirt 168	25·9 ± 1·2	36·3 ± 3·4
63‐83	23·1 ± 2·3	35·0 ± 1·8
Sensho	20·8 ± 1·7	39·2 ± 1·7
IAC 1131	25·2 ± 1·1	31·3 ± 1·7
Fukerupagai II	27·9 ± 2·0	40·8 ± 0·8
IR3880‐5	19·7 ± 1·0	31·2 ± 1·2
Calrose	22·4 ± 2·4	35·9 ± 0·8
Basmati 370	21·7 ± 1·2	28·9 ± 2·1
Basmati (Kurnool)	23·9 ± 2·3	36·7 ± 1·1
IR442‐57	22·1 ± 0·6	39·0 ± 2·7
Leb Mue Nahng III	24·6 ± 0·5	36·0 ± 1·0

Plants were raised in aerated solution for the first 10 d, and then in either aerated or stagnant deoxygenated solution for the final 25 d.

Measurements were taken on 100–150 mm adventitious roots.

Values given are means of three replicates ± standard errors.

l.s.d._{(P < 0·05)} for genotype comparisons of root porosity, 3·6.

ROL from intact adventitious roots of rice when in an O₂‐free medium

Growth in aerated or stagnant solution had a marked effect on the spatial pattern of ROL from the adventitious roots of 11 of the 12 genotypes studied (Fig. 1, P < 0·05). ROL from adventitious roots of plants grown in aerated solution prior to the measurements was lowest near the root apex and it increased as the electrode was moved to more basal positions (i.e. closer to the root/shoot junction) in all genotypes, except the paddy type Basmati 370. In contrast to this result for roots of aerated plants, rates of ROL from adventitious roots of plants grown in stagnant solution prior to the measurements were highest just behind the apex and decreased to very low rates, or even zero, in the most basal positions measured in ten of the 12 genotypes (except two upland types; Sensho and IAC 1131). The distances behind the apex that ROL first became very low for adventitious roots of plants grown in stagnant solution were: 50 mm in one deep‐water type, ≥40 mm in five of the seven upland types, ≥30 mm in two of the paddy types and one deep‐water type, and ≥20 mm in one paddy type. The marked differences in spatial patterns of ROL along roots of rice grown in aerated or stagnant solution resulted in a significant (P < 0·001) treatment × position interaction in the statistical analysis; however, the position × cultural type and position × cultural type × treatment interactions were not statistically significant (P = 0·18 and P = 0·36, res pectively).

Fig. 1. Rates of radial O₂ loss (ROL) along intact adventitious roots of 12 rice genotypes when in an O₂‐free root medium with shoots in air. Plants were raised in either aerated (open symbols) or stagnant deoxygenated nutrient solution (closed symbols) prior to the measurements taken along one 100–131 mm adventitious root of each 28–38‐d‐old plant, with the treatments imposed on day 10. Measurements were taken at 25 °C and the O₂‐free root medium contained some basal electrolytes as well as 0·1 % agar to prevent convection. Data given are means of three replicates ± standard errors.

The rates of ROL at 5 mm behind the apex of intact adventitious roots of plants raised in stagnant solution, when compared with those of the same genotypes when raised in aerated solution, were: 6–85‐fold higher for upland types, 3–39‐fold higher for paddy types, and 258‐ and 266‐fold higher in the deep‐water types (Fig. 1). The enhanced internal O₂ diffusion to the apex was presumably the result of the increased porosity in the roots of plants grown in stagnant, compared with aerated solution (Table 5). In marked contrast to the higher rates of ROL just behind the apex, ROL decreased from basal positions of roots of plants grown in stagnant solution; when compared with roots of plants raised in aerated solution, rates at 50 mm behind the apex were only 0–60 % for upland types, 0–19 % for paddy types, and 0–0·4 % for deep‐water types. Thus, the resistance to radial O₂ movement from the aerenchyma to the medium had increased markedly in the more basal zones of the adventitious roots of ten of the 12 genotypes of rice when grown in stagnant, compared with aerated, solution (Fig. 1).

Rates of extension were measured for intact adventitious roots of lengths similar to those used in the ROL measurements. If the extension rates of roots had differed between the two treatments, this would have caused tissues at equivalent distances behind the apex to differ in age, and such differences would have complicated interpretations of direct treatment affects on the spatial patterns of ROL along the roots. Extension rates of adventitious roots, initially 84–134 mm in length, were not statistically different between plants grown in aerated or stagnant solution for nine of the 12 genotypes (Table 6). In three of the genotypes [IAC 1131, IR3880‐5, Basmati (Kurnool)], however, root extension rates declined in stagnant solution (Table 6). The slower root extension in stagnant solution may have influenced, to some degree, the differences in spatial patterns of ROL between roots of aerated or stagnant plants for these three genotypes. The cause of the slower root extension in these genotypes remains to be determined; however, it presumably was not due to O₂ deficiency since rates of ROL were relatively high near the root tip (Fig. 1).

Table 6.

Rates of extension of adventitious roots of 12 genotypes of rice grown in aerated or stagnant deoxygenated nutrient solution.

	Root extension rate (mm d–1)
Genotype	Aerated	Stagnant
Omirt 39	19 ± 1·8	20 ± 1·5
Omirt 168	20 ± 0·6	19 ± 2·6
63‐83	25 ± 2·2	26 ± 2·1
Sensho	24 ± 1·7	22 ± 2·6
IAC 1131	28 ± 0·9	20 ± 4·0
Fukerupagai II	22 ± 2·7	23 ± 2·9
IR3880‐5	19 ± 1·8	13 ± 1·5
Calrose	24 ± 2·7	20 ± 2·3
Basmati 370	19 ± 3·2	15 ± 1·2
Basmati (Kurnool)	29 ± 2·4	18 ± 6·5
IR442‐57	19 ± 1·3	19 ± 4·7
Leb Mue Nahng III	21 ± 1·5	19 ± 3·2

Plants were raised in aerated solution for the first 10 d, and then in either aerated or stagnant deoxygenated solution for the final 25 d. Measurements were taken for one root (84–134 mm initial lengths) on each of two plants in each replicate pot, between the 23rd and 24th day of treatments (i.e. 33 and 34 d after imbibition).

Values given are means of three replicates ± standard errors. l.s.d. (P < 0·05) for genotype comparisons of root extension rates, 5·2.

DISCUSSION

The present study showed roots of rice (Oryza sativa) acclimate to growth in stagnant deoxygenated solution by (a) increasing porosity above the already constitutively high levels and (b) induction of a ‘tight’ barrier to ROL in the basal zones. These changes were evident in the adventitious roots of selected genotypes of upland, paddy and deep‐water rice. Furthermore, in all but one of the 12 genotypes, the number of adventitious roots per stem increased when plants were grown in waterlogged, compared with drained, soil. These traits and acclimations presumably contribute to waterlogging tolerance (Armstrong, 1971; Jackson and Drew, 1984; Colmer et al., 1998) in rice.

Variation in response of dry mass production to soil waterlogging was observed among the 12 genotypes; however, the upland types (except Sensho) were generally as tolerant as the paddy and deep‐water types (Tables 2 and 3). The high degree of waterlogging tolerance even in the upland types may reflect that episodes of waterlogging can be common in many upland soils used to cultivate rice (IRRI, 1975; Moormann and van Breemen, 1978; Grist, 1986), hence traits associated with waterlogging tolerance have been retained. Root porosity increased substantially in all 12 genotypes when grown in stagnant solution, with no significant difference among the three cultural types (Table 5). Furthermore, a ‘tight’ barrier to ROL was induced in roots of ten of the 12 genotypes when grown in stagnant solution (Fig. 1). The two exceptions (Sensho and IAC 1131), like one other genotype in an earlier study (Colmer et al., 1998), were upland types. Further insight into the diversity of root aeration traits in upland rice would be gained by studies of genotypes taken from well‐defined environments, especially soils known to be well drained.

Soil compaction, in addition to waterlogging, may also have acted as a selection pressure to retain the capacity to develop root aerenchyma in upland rice. Aerenchyma is induced in roots of some species when challenged with soils of high mechanical impedance (He et al., 1996), and improved internal aeration may enhance root growth through compacted soil layers (Comis, 1997).

Substantial differences in rates and patterns of ROL were documented for individual genotypes; inter‐varietal variation in rates of ROL had been reported previously for paddy types (Armstrong, 1969; Chen et al., 1980), but had only been evaluated for two upland types (Colmer et al., 1998). Growth in stagnant solution increased the resistance to radial movement of O₂ from the aerenchyma to the medium, in the basal root zones in ten of the 12 genotypes (Fig. 1), a finding consistent with the responses of three of the four rice genotypes in an earlier study (Colmer et al., 1998). Growth in stagnant deoxygenated solution also induced a barrier to ROL in adventitious roots of Hordeum marinum (McDonald et al., 2001) and Lolium multiflorum (McDonald et al., 2002). In contrast, roots of several other wetland species had a ‘tight’ barrier to ROL when grown in either aerated or stagnant solution; examples are Juncus effusus (Visser et al., 2000), Echinochloa crus‐galli, Eleocharis acuta and Schoenoplectus validus (McDonald et al., 2002).

Not all wetland plants, however, have roots with a ‘tight’ barrier to ROL; the basal root zones in some species lose substantial amounts of O₂ to the external medium (e.g. Rumex spp.) (Laan et al., 1989; Visser et al., 2000). The importance of a barrier to ROL for effective longitudinal diffusion of O₂ in roots diminishes as the volume of aerenchyma becomes large (Armstrong, 1979), especially if roots are relatively thick. Moreover, differences among wetland species in patterns of ROL may also be related to their habitat; species from sandy substrates low in organic matter (i.e. not highly reduced) may have roots relatively more permeable to ROL, compared with those from more reducing substrates (Smits et al., 1990). Evaluations of ROL from roots of more species from well‐characterized environments that differ in duration and frequency of waterlogging events, as well as degree of O₂ deficiency (cf. Armstrong and Boatman, 1967) and soil redox potential, are required to determine if the type of barrier to ROL (constitutive or inducible, ‘tight’ or ‘weak’) formed in roots is associated with plants from any particular niche wetland environment(s).

Knowledge on the physiological and anatomical basis of barriers to ROL in plant roots is scant. In roots of Phragmites australis, a combination of densely packed cells, suberin deposits, lignification in the outer cell layers and O₂ consumption by cell layers exterior to the aerenchyma, have been suggested as the mechanisms reducing ROL from basal root zones (Armstrong and Armstrong, 1999; Armstrong et al., 2000). The mechanism(s), and the possibility of ‘passage areas’, would determine whether having a barrier to ROL also affects the radial permeability of the root to substances other than O₂ (e.g. reduced soil toxins, water, nutrients, methane, etc.) [suggested by Armstrong (1979) and Armstrong et al. (2000)]. In the case of paddy rice sampled from waterlogged soil, a ‘tight’ barrier to ROL was evident even when respiration was inhibited by cooling the root medium to 3 °C (Armstrong, 1971), indicating a physical barrier to ROL in the adventitious roots of rice.

A physical barrier to ROL may impede water and nutrient uptake by roots of wetland species (Armstrong, 1979; Koncalova, 1990). In roots of rice, hydraulic conductivity in the radial direction was relatively low (Miyamoto et al., 2001) and so were rates of ion net uptake in the basal regions (Colmer and Bloom, 1998), when compared with dryland species, even when the rice plants were grown in aerated solution. It seems reasonable, therefore, to speculate that roots containing a barrier to ROL may be less efficient at water and nutrient uptake, at least when in drained soils, than roots without such a barrier. Moreover, a barrier to ROL would not only inhibit O₂ loss from the root, but also O₂ uptake (T. D. Colmer, unpublished data), so roots with a constitutive barrier to ROL would rely on longitudinal diffusion in the aerenchyma to supply O₂ to internal tissues (e.g. the stele) even when growing in drained soil. Thus, an inducible barrier to ROL may be of adaptive significance to plants inhabiting transiently waterlogged soils and/or contribute to the physiological plasticity that may enable a species (e.g. Oryza sativa) to succeed in diverse environments ranging from drained to flooded soils.

ACKNOWLEDGEMENTS

I thank Mike McDonald, Jeremy English and Digby Short for assistance with plant harvests, Laurie Lewin for advice on rice genotypes and for providing the seeds, and Hank Greenway for comments on a draft of this manuscript.

Received: 6 August 2001; Returned for revision: 23 November 2001; Accepted: 17 December 2001