Rice Root Nutrient 1994

1994

Los Baños, Laguna, Philippines Mailing address: P.O. Box 933, Manila 1099, Philippines

The International Rice Research Institute (IRRI) was established in 1960 by the Ford and Rockefeller Foundations with the help and approval of the Government of the Philippines. Today IRRI is one of 18 nonprofit international research centers supported by the Consultative Group on International Agricultural Research (CGIAR). The CGIAR is sponsored by the Food and Agriculture Organization of the United Nations (FAO), the International Bank for Reconstruction and Development (World Bank), and the United Nations Development Programme (UNDP). Its membership comprises donor countries, international and regional organizations, and private foundations. IRRI receives support, through the CGIAR, from a number of donors including FAO, UNDP, World Bank, European Economic Community, Asian Development Bank, Rockefeller Foundation, Ford Foundation, and the international aid agencies of the following governments: Australia, Belgium, Canada, People’s Republic of China, Denmark, Finland, France, Germany, India, Islamic Republic of Iran, Italy, Japan, Republic of Korea, The Netherlands, Norway, Philippines, Spain, Sweden, Switzerland, United Kingdom, and United States. The responsibility for this publication rests with the International Rice Research Institute. The designations employed in the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of IRRI concerning the legal status of any country, territory, city, or area, or of its authorities, or the delimitation of its frontiers or boundaries. ©International Rice Research Institute 1994 Los Baños, Philippines Mailing address: P.O. Box 933, Manila 1099, Philippines Phone: (63-2) 818-1926,812-7686 Fax: (63-2) 818-2087 Email: IN%”[email protected]” Telex: (ITT) 40890 Rice PM; (CWI) 14519 IRILB PS; (RCA) 22456 IRI PH; (CWI) 14861 IRI PS

Suggested citation: Kirk, G J D, ed. (1994) Rice roots: nutrient and water use. Selected papers from the International Rice Research Conference. International Rice Research Institute, P.O. Box 933, Manila 1099, Philippines.

ISBN 971-22-0050-7

Contents

Foreword The rice root-soil interface

1

G.J.D. KIRK, J.L. SOLIVAS, AND C.B.M. BEGG

Root growth and nitrogen uptake in rice: concepts for modeling

11

H.F.M. TEN BERGE, T.M. THIYAGARAIAN, B. MISHRA, K.S. RAO, AND R.N. DASH

Genetic variation in nitrogen uptake by rice and the effects of management and soil fertility 29 P.C. STA. CRUZ AND G. WADA

Use of molecular markers to evaluate rice genetic variation in associative N2 fixation, N uptake, and N use efficiency 43 P. WU, J.K. LADHA, S.R. McCOUCH, AND S.B. TENG

Rainfed lowland rice roots: soil and hydrological effects 55 PRADEEP K. SHARMA, G. PANTUWAN, K.T. INGRAM, AND S.K. DE DATTA

Rice root traits for drought resistance and their genetic variation 67 K.T. INGRAM, F.D. BUENO, O.S. NAMUCO, E.B. YAMBAO, AND C.A. BEYROUTY

Use of molecular markers to exploit rice root traits for drought tolerance H.T. NGUYEN, J.D. RAY, AND LONG-XI YU

79

Foreword

The rice plant invests up to 60% of its energy as carbon in its root system. However, the proportion of research effort devoted to the root system has been much smaller than this, and consequently our understanding of the rice roots and their function in the capture of nutrients and water lags behind our understanding of the rest of the plant. This is particularly so for rice, compared with other cereals, because the rice plant's ability to grow under waterlogged soil conditions arises from morphological and physiological adaptations in its roots. These adaptations have important consequences for nutrient and water uptake. A greater understanding of the root system is now needed to support efforts to increase rice yield potential and improve the productivity of resources. As part of the 1992 International Rice Research Conference, a symposium on rice roots and the uptake of nutrients and water was held to review present knowledge and to develop recommendations for future research. The papers included here were reviewed and revised on the basis of the discussions during the symposium. It is hoped that they will serve as a benchmark on the topic for some time to come.

Klaus Lampe Director General

The rice root-soil interface G.J.D. Kirk, J.L. Solivas, and C.B.M. Begg

Models and experimental studies of the rhizosphere of rice plants growing in anaerobic soil show that two major processes lead to considerable acidification (1-2 pH units) over a wide range of root and soil conditions. One process is the generation of H+ in the oxidation of Fe 2+ by O2 released from the roots. The other is the release of H+ from the roots to balance the excess intake of cations over anions, N being taken up chiefly as NH 4+. CO2 exchange between the roots and soil has a much smaller effect. The zone of root influence extends a few millimeters from the root surface. Substantial differences are found along the root length and over time. The acidification and oxidation cause increased sorption of NH4+ by soil solids, thereby impeding the movement of N to absorbing root surfaces but may also cause solubilization and enhanced uptake of soil P. Implications for nutrient management and the develop ment of nutrient-efficient rice germplasm are discussed.

Rice roots growing in anaerobic soil can greatly modify the soil near them. As a result, the soil as “seen” by the root is entirely different from the bulk soil, although it is the latter whose properties we normally measure. Some important processes influenced by conditions in the rice rhizosphere are listed in Table 1. This paper examines the extent and dynamics of root-induced changes and their consequences for nutrient and toxin dynamics. Three main processes operate. First, release of O2 from roots and its reaction with Fe2+ , generating acidity: 4 Fe2+ + O 2 + 10 H2O = 4 Fe(OH)3 + 8H+. Secondly, because the roots take up a considerable excess of cations over anions (N being taken up from anaerobic soil chiefly as NH 4+ ), they release H+ into the soil to

Table 1. Effects of rice rhizosphere conditions. Variable or process affected

Example

Concentration of toxic solute Concentration of major nutrient Concentration of minor nutrient N2 fixation Nitrification Invasion by pathogens Methane oxidation

Fe2+ NH4+ , H2 PO4 - , K + Zn2+ Azospirillum Nitrosomonas

maintain electrical neutrality across the root-soil interface. Thirdly, because high pressures of CO2 arise in anaerobic soil and within the roots, the roots may either release CO2 or take it up from the soil, with corresponding changes in soil pH. The root-influenced zone extends just a millimeter or so into the soil, but existing experimental methods cannot analyze the soil with resolutions much finer than a millimeter. Because of this and because the system involves the simultaneous operation of a series of complex, linked processes, we use simulation models in addition to experiments.

Gas transport through rice roots Gas transport through flooded soil is very slow, and therefore rice roots must supply O2 to respiring tissues via internal gas channels, which are called aerenchyma. Some of this internally transported O2 is lost to the surrounding soil. There is disagreement about the extent of this loss and about whether or not it is beneficial to the plant. Direct measurements of loss vary from several nmol (O2 )/dm2 (root surface) per s near the root tips as measured with cylindrical polarigraphic electrodes (Armstrong et al 1991) to one-hundredth of this for O 2 appearing in continuously replenished, O2 -free solutions bathing whole root systems (Bedford et al 1991). The range is in part due to differences in leakiness between different parts of the root system and in part due to differences in the external O2 sink. In soil, the O2 sink is enhanced by diffusion of Fe2+ toward the root and its reaction with O2 . The net O2 loss depends on the rates of diffusion and reaction.

Model of rice rhizosphere conditions Ahmad and Nye (1990) have measured the kinetics of iron oxidation in reduced soil, and Kirk et al (1990) have used this information to develop a model of the coupled diffusion and reaction of O2 , Fe 2+, and acidity in reduced soil. The kinetic measurements showed that, over periods of a few days, the main reaction is between Fe2+ and O2 with formation of Fe(OH)3 and acidity. The rates are comparable to or exceed O2 consumption by microbes. The model predicts the diffusion of O2 into the soil; the diffusion of Fe2+ toward the oxidation zone; the rate of formation and concentration profile of the Fe(OH)3 formed; and the diffusion by acid-base transfer of the acidity produced in the reaction. 2

Kirk et al

1. Simulated reactant profiles in the rhizosphere after different times. Parameter values given in Kirk (1993).

The rice root-soil interface

3

Kirk (1993) has coupled this model, converted to the cylindrical geometry of a root and its surrounding rhizosphere, to a model of gas transport through the root. The model allows for axial diffusion of O2 and CO2 through the root and for differences with root length and time in cortical porosity, respiration rates in different tissues, and root wall gas permeability. It also allows for the development of short, fine laterals along the primary roots; by the end of the vegetative phase, these may account for the majority of the total root length though only a small percent of root mass. Having high surface area to volume ratios, they tend to be very O2 -leaky. Figure 1 shows the model’s predictions for a particular depth in the soil as a root grows through it. The O2 pressure in the root cortex is held constant and the wall permeability decreases as the root elongates. Thus the soil at the root surface is exposed to a declining O 2 concentration with time. The mean O2 efflux from the rod, averaged over the 10 d of simulation, is about 0.2 nmol (O2 )/dm2 (root) per s. The model shows that substantial amounts of Fe are transferred toward the root surface and accumulate as Fe(OH)3 . A zone of Fe 2+ depletion arises when oxidation is intense, but is rapidly filled in by Fe 2+ continuing to diffuse into the depletion zone after the O2 supply has decreased. Thus Fe accumulation continues after oxidation declines. Of interest in the present context is the very large drop in pH which persists throughout the period simulated. This is due to the combined effects of H+ generated

2. Simulated pH profiles in the rhizosphere after 10 d. Root wall permeability and cortical O 2 concentration held constant giving mean O2 efflux from root = 0.2 nmol/dm 2 per s; H+ efflux from root = 1 nmol/dm 2 per s. Other parameter values as in Figure 1.

4

Kirk et al

in Fe2+ oxidation and H+ released from the root to balance ion intake. The model predicts that CO2 is lost from the root to the soil near the root tip, but taken up by the root near the base; the resultant CO2 concentration profiles in the soil are small. There is an important interaction between the two sources of acidity which results in their combined effect being greater than the sum of their individual effects (Fig. 2). This is explained as follows. pH changes are propagated through soil by the movement of conjugate acid-base pairs: acids from regions of low pH to high and bases in the opposite direction (Nye 1986). The most important pairs are generally H 3O+-H2O and H2CO3 (derived from CO2)-HCO3- . The relative importance of the pair H3O+-H2O is greater at low pH and that of H2CO3-HCO3 at high pH. Therefore the net rate of soil acidity diffusion is minimal in the pH range in which H3O+ and HCO3- concentrations are both low. In this pH range, a flux of acid or base through the soil results in steep pH gradients. The predictions in Figures 1 and 2 are for conditions around aerenchymatous primary roots, but the model predicts similar changes in the rhizospheres of fine lateral roots. Laterals are presumably responsible for the bulk of nutrient uptake in lowland rice because they comprise the bulk of the total root length. (Indeed, the reason they comprise the bulk of the root length is the slow rate of movement of N and other nutrients to root surfaces in anaerobic soil and the consequent need for a large surface area per unit root mass.) Thus, it is conditions in the rhizospheres of laterals that we are most concerned with.

Experimental measurements of rice rhizosphere conditions To test these predictions, we developed an experimental method with which to measure reactant concentration profiles near rice root surfaces (IRRI 1991). We grew rice seedlings in nutrient solutions with their roots contained in 7.5-cm-diameter, l-mmthick nylon mesh bags so that the roots formed a continuous planar layer. Each root plane was then sandwiched between cylinders of anaerobic soil connected to water reservoirs, and the system sealed so that the only means of O2 transfer between the atmosphere and soil was through the roots. The plants grew healthily in this system for more than 2 wk without showing signs of stress. At intervals, the cylinders were sliced into 0.2-mm sections parallel to the root plane, and each section analyzed for pH, Fe(II) extractable with 1 M NH4OAc at pH 2.8, and Fe(III) extractable with 2 M H2SO4. Figure 3 shows profiles of pH and Fe near the root plane after 10 d of root-soil contact. A large quantity of Fe has been transferred toward the root plane and oxidized, producing a zone of Fe(III) accumulation; there has been a very large acidification of the soil within 4 mm of the root plane; the pH gradient is very steep at pH levels below 6. Control experiments with soil cylinders sealed together without roots indicated that the seals were good and that the oxidation in the cylinders with roots was indeed largely due to root-released O2. The total quantity of acidity generated (estimated from the pH profile using the independently measured pH buffer power of the reduced soil while undergoing oxidation = 0.09 mmol/g per pH unit) was of the order of 3 mmol/cylinder, and that The rice root-soil interface

5

3. Profiles of Fe(II), Fe(III), and pH in flooded soil exposed to a planar layer of rice roots for 10 d. Plant and soil characteristics given in IRRI (1991).

generated in Fe oxidation (the quantity of Fe(III) formed multiplied by 2—see reaction stoichiometry given earlier) was of the order of 2 mmol/cylinder. Thus, by difference, that due to cation-anion intake imbalance is 1 mmol/cylinder. The mean flux of O2 across the root plane required to oxidize this quantity of Fe is about 0.6 nmol/dm2 (root plane) per s. In experiments in which we exposed cylinders of the same anaerobic soil directly to air, greater oxidation took place, but the concentration of Fe(III) at the air-exposed surface was much smaller. This is because in the root plane experiments, the O2 concentration maintained by the roots at their surfaces was less than that in air, and therefore the diffusion of O 2 into the soil was smaller and the movement of Fe2+ toward the roots greater. Movement of the soil solution due to water uptake by the roots will enhance this effect.

Consequences for nutrient uptake Nitrogen The ability of the soil-root system to meet the plant’s N demand depends both on the ability of the roots to absorb N from the soil at their surfaces and on the rate of delivery 6

Kirk et al

of N to the root surfaces. For arable crops growing in fertile aerobic soil, the dominant form of N taken up is NO3 - and it is sufficiently mobile in the soil that root morphology and uptake properties do not limit uptake until soil N levels become very low (Drew 1990). But for lowland rice, the situation is entirely different: the dominant form of N taken up is NH 4+, which is far less mobile. The rate of NH4+ transport through the soil to root surfaces depends on its distribution between the soil solid and solution, and this is largely determined by the concentration of anions in solution (in accordance with electroneutrality), though modified by the concentrations of other exchangeable cations. In most flooded soils, HCO3- is the dominant anion (in aerobic soils, it is NO 3 - ) but its concentration will decrease as the rhizosphere pH decreases, becoming negligible below about pH 5.5. NH4 + mobility will be correspondingly decreased. This will be offset, to some extent, by the effect of acidity on cation sorption, tending to decrease the NH4+ concentration in the solid relative to that in solution, but exacerbated by the depletion of exchangeable Fe 2+.

4. Effect of Fe 2+ oxidation in the rhizosphere on the distribution of exchangeable NH4+ between the soil solid and solution. Numbers on curves are concentrations of noncarbonate anions in solution. Other parameter values given in IRRI (1992).

The rice root-soil interface

7

Figure 4 gives predictions of a model of these effects. In the model, cation exchange is governed by electroneutrality in the soil solution and solid and by equations for the preferential sorption of divalent cations over monovalent ones (IRRI 1992). For each mole of Fe2+ oxidized, 2 moles of H+ are produced, and these react with HCO3- and the soil solid. The increase in acidity on the soil solid causes a proportionate decrease in pH and an equal decrease in the equivalents of cations on the solid. The set of parameter values used for Figure 4 represent a realistic mean. It will be seen that rhizosphere oxidation may greatly decrease the fraction of mobile NH4+ in the soil solution, particularly at low concentrations of anions in the soil, and thereby greatly decrease N uptake. In addition, acidification may decrease the rates of microbially mediated processes affecting the N supply, such as mineralization of organic N.

Phosphorus

In contrast to the effect on N uptake, rhizosphere oxidation and acidification may enhance solubilization and uptake of P. This is important because, in many rice soils, flooding results in a decrease in P mobility following an initial release, and rice plants consequently depend on root-induced solubilization for the bulk of their P. We measured concentration profiles of extractable P fractions in the soil near roots, using the same root-plane technique. The fractions, measured sequentially, were anion exchange resin, alkali (subdivided into inorganic and organic), acid, and residual. The results showed a large depletion of acid-soluble and residual P in the zone of Fe oxidation and acidification, 0-5 mm from the root plane (Fig. 5). Part of the solubilized P appeared closer to the root plane in the alkali-soluble P pool, but the bulk was absorbed by the roots. The quantity of P taken up by the plants during the experiment, calculated from the increase in plant dry matter multiplied by the mean % P content, was about 25 µmol. This was roughly matched by the net depletion of acid-soluble and residual soil P. The quantity contained in the freely available resin pool was negligible in comparison with plant demand. In flooded soils, a large part of the P is often acid-soluble, being associated with metal carbonates and hydroxides formed following soil reduction and with other negatively charged surfaces through exchangeable cations. The alkali-soluble pool would include that associated with Fe(OH)3 formed in the oxidation reaction and with root-derived organic material. Evidently, the reimmobilization of acid-solubilized P on Fe(OH)3 and root-derived organic matter did not prevent access of P to the root surfaces. By contrast, we found that the bulk of P taken up by rice growing in aerobic soil was drawn from an alkali-soluble pool (Hedley et al 1994). To solubilize this P, the roots would need an alternative mechanism. Current evidence suggests that this mechanism involves the release of low molecular weight organic acid anions such as citrate. These must leave the root cells in the dissociated salt form, not the acid form, because the cytoplasmic pH (6-7) is well above the pKs of the acids in question. Therefore, from charge-balance considerations, in anaerobic soil where N is taken up as NH4+ , export of organic acid anions would be difficult. 8

Kirk et al

5. Profiles of extractable P fractions in flooded soil exposed to a planar layer of rice roots for 10 d. Numbers of curves are means of values in the soil bulk.

The potential for manipulating root-induced P solubilization to develop more Pefficient rice cultivars is discussed by Kirk et al (1993).

Discussion For reasonable rates of nutrient uptake and O2 loss from rice roots, acidification and Fe 2+ oxidation in the rhizosphere can result in impeded N uptake but enhanced P uptake. These findings have interesting consequences for the relation between root traits and nutrient uptake efficiency under different conditions. In conditions in which N is the limiting nutrient, root porosity and wall permeability should be just sufficient to The rice root-soil interface

9

meet the root’s respirsatory O2 requirements and perhaps to oxidize some of the toxic products of anaerobic respiration in the soil. But in P-limited conditions, rhizosphere oxidation and acidification would be desirable and thus a more aerenchymatous and O2-leaky root would be wanted. Similar arguments should apply to K, which should be immobilized like NH4+, and Zn, which should be mobilized like P.

References cited Ahmad A R, Nye P H (1990) Coupled diffusion and oxidation of ferrous iron in soils. I. Kinetics of oxygenation of ferrous iron in soil suspension. J. Soil Sci. 44:395-409. Armstrong W, Justin SHFW, Beckett PM, Lythe S (1991) Root adaptation to soil waterlogging. Aquatic Bot. 39:57-73. Bedford B L, Bouldin D R, Beliveau B D (1991) Net oxygen and carbon dioxide balances in solutions bathing roots of wetland plants. J. Ecol. 79:943-959. Drew M C (1990) Root function, development, growth and mineral nutrition. Pages 35-57 in The rhizosphere. J.M. Lynch, ed. Wiley-Interscience, Chichester. Hedley M J, Kirk G J D, Santos M B (1994) Phosphorus efficiency and the forms of soil phosphorus utilized by upland rice cultivars. Plant Soil 158: 53-62. IRRI—International Rice Research Institute (1991) Program report for 1990. P.O. Box 933, Manila, Philippines. p. 203-205. IRRI—International Rice Research Institute (1992) Program report for 1991. P.O. Box 933, Manila, Philippines. p. 198-199. Kirk G J D (1993) Root ventilation, rhizosphere modification, and nutrient uptake by rice. Pages 221-232 in Systems approaches for agricultural development. F.W.T. Penning de Vries, P.S. Teng, and K. Metselaar, eds. Kluwer Academic Publishers, Dordrecht, The Netherlands. Kirk G J D, Ahmad A R, Nye P H (1990) Coupled diffusion and oxidation of ferrous iron in soils. II. A model of the diffusion and reaction of Fe 2+, H+ and HCO 3- in soils and a sensitivity analysis of the model. J. Soil Sci. 44:411-431. Kirk G J D, Hedley M J, Bouldin D R (1993) Phosphorus efficiency in upland rice cultivars. Pages 279-295 in Papers and reports on the management of acid soils (IBSRAM/ASIALAND). Network Document No. 6. International Board for Soil Research and Management, Bangkok. Nye P H (1986) Acid-base changes in the rhizosphere. Pages 129-153 in Advances in plant nutrition. Vol. 2. P.B. Tinker and A. Lauchli, eds. Praeger, New York.

Notes Authors’ address: G.J.D. Kirk, J.L.Solivas, and C.B.M. Begg, International Rice Research Institute, P.O. Box 933, Manila, Philippines Citation information: Kirk G J D, ed. (1994) Rice roots: nutrient and water use. International Rice Research Institute, P.O. Box 933, Manila, Philippines.

10

Kirk et al

Root growth and nitrogen uptake in rice: concepts for modeling H.F.M. ten Berge, T.M. Thiyagarajan, B. Mishra, K.S. Rao, and R.N. Dash

Data from field experiments are used to examine the rate-limiting step in nitrogen (N) uptake by lowland rice. For the vegetative phase over a wide range of fertilizer levels, N uptake per ha (N c) and uptake rates per ha (dNc/dt) increased with the amount of N applied (A). dNc /dt did not reach an upper limit. Increases in dNc /dt with A were more than in proportion to increases in root mass (Mr). After panicle initiation (PI), there was no consistent relation between A, dNc /dt, and Mr. The root-shoot ratio decreased with development stages and was independent of A at any stage. Up to PI, Mr and the rate of N absorption per unit Mr both increased with A to the extent that fertilizer N recovery increased as A increased. An approximate calculation showed that changes in cumulative recovery with A could be explained by differences in Mr. Two hypotheses to explain dNc /dt during the pre-PI stage were consistent with the experimental data: 1) the uptake rate was diffusion-limited and proportional to the concentration of NH4+ in the root zone bulk soil solution [NH4+ ] and to root length density (RLD), and, 2) the uptake rate was independent of [NH4+ ] and governed by the plant internal N content through a mechanism superimposed on the effect of Mr. Total N uptake over the season was a linear function of A. This linearity was the result of a positive feedback between Nc and dN c/dt during vegetative growth, followed by a negative or neutral feedback after PI.

Much work on N in rice has focused on loss processes in relation to floodwater and soil chemistry and biology (e.g., De Datta and Patrick 1986), and detailed physicochemical models of these processes have recently become available (Rachhpal-Singh and Kirk 1993 a,b). Also, the utilization of N by the crop once absorbed and its effects on dry matter production are now reasonably well-understood (Yoshida 1981, Kropff et al 1993). However, the process of N uptake by the root system has received relatively little attention. An understanding of this is essential for the development of a complete model of the fate of N in rice production systems.

The supply of N to a single root depends on the rate of transport from the bulk soil solution to the root surface. For lowland rice, this transport is principally by diffusion (Makarim et al 1991). The maximum rate of diffusion is reached for “zero-sink” conditions in which the root maintains a negligibly low concentration at its surface, thus maximizing the concentration gradient for a given bulk concentration (de Willigen and van Noordwijk 1987). For an entire root system, “supply” is a more complex group of processes. Roots are unevenly distributed and the entire root system is not necessarily equally active. Good mathematical models exist to integrate knowledge on the various physiological, physical, and physicochemical processes affecting uptake. Taking single root behavior as a starting point, the uptake model of de Willigen and van Noordwijk (1987, 1991) assesses nutrient uptake rate as a function of soil conditions (water content, bulk nutrient concentration, and nutrient diffusivity), root length density (RLD), and parameters specifying “crop demand” and maximum flux density over the root surface. But parameterization of such mechanistic models is difficult. The measurement of RLD in rice under field conditions is virtually impossible due to very fine branching (Yoshida 1981, Drenth et al 1991) and strong spatial RLD gradients near the soil surface. Moreover, the roles of different root fractions in uptake are unknown (Makarim et a1 1991), and we have little information about maximum nutrient flux densities across rice root surfaces under either controlled or field conditions, or about the dependence of root nutrient uptake characteristics on soil and crop conditions. Also, the concept of “demand” is not well-defined, and if the crop’s nutritional status affects root uptake characteristics, demand turns out to directly affect supply itself. The purpose of this study was to determine what measurable variables must be considered to describe N uptake at different stages of lowland rice crop development, as a first step toward a mechanistic description of the system.

Materials and methods Three data sets were collected in experiments at three locations in India, with which to test hypotheses on the relation between N uptake, RLD, and [NH4+]. 1. The TNAU-TNRRI set (Thiyagarajan et a1 1991) was collected at the Tamil Nadu Rice Research Institute, Aduthurai, Tamil Nadu, during the late wet season (WS) of 1988-89. The 39-d-old seedlings of ADT39 were transplanted on 10 Dec at 20× 10-cm spacing, at two seedlings per hill, in a randomized block design with four replications. Nitrogen was applied as urea at 0, 100, 200, 300, and 400 kg N/ha in four splits: 50% basal (i.e., incorporated at transplanting), the rest in equal splits at active tillering (26 d after transplanting [DAT]), at panicle initiation (PI) (48 DAT), and at 62 DAT. Flowering varied from 53 DAT (0 kg N treatment) to 63 DAT (400 kg N), and maturity from 89 DAT (0 kg N) to 97 DAT (400 kg N). Plant samples were collected at 18, 25, 32, 39, 46, 53, 63, 81, 89, and 97 DAT. All samples were split; dry weight and N content of leaves, stems plus sheaths, panicles, and roots were determined.

12

ten Berge et al

2. The PUAT set was collected at G.B. Pant University of Agriculture and Technology, Pantnagar, Uttar Pradesh, during the 1987 WS. Twenty-nine-d-old seedlings of Pant Dhan 4 were transplanted on 9 Jul at 20- x 100-cm spacing, at two seedlings per hill, in a randomized block design with four replications. Nitrogen was applied as urea at 0, 60, 120, 180, and 240 kg N/ha in three splits: 50% basal, 25% topdressed at 27 DAT (active tillering), and 25% topdressed at 51 DAT (PI). The 60 kg N/ha treatment received no second topdressing, but received 50% of the total dose at 27 DAT. Flowering occurred at 72 DAT and maturity at 108 DAT. Plant samples were collected at 25, 49, 72, and 108 DAT. Dry mass and N contents of leaves, stems plus sheaths, grains, and roots were measured for all harvests, except root mass which was omitted at maturity. 3. The CRRI set was collected at the Central Rice Research Institute, Cuttack, Orissa, during the 1990 dry season (DS). Thirty-eight-d-old seedlings of IR36 were transplanted on 25 Jan at 15- x 15-cm spacing, at two seedlings per hill, in a randomized block design with four replications. Nitrogen was applied as urea at 0, 50, 100, and 150 kg N/ha in three splits: 50% basal, 25% at 20 DAT, and 25% at 53 DAT (1-2 wk after PI). The crop flowered at 70 DAT and was harvested at 92 DAT. Plant samples were collected at 20, 30, 40, 50, 70, and 92 DAT and analyzed as above. Nitrogen in plant samples was determined only at 20, 50, and 70 DAT.

Results N uptake as a function of N application

Tables 1-3 give the results for TNAU-TNFUU, PUAT, and CRRI, respectively. The maximum amount of N accumulated in the crop Nc (kg/ha), as a function of the amount

1. Relation between maximum total crop (leaves, stems, roots, and panicles) N uptake (N,) and N application in the three experiments. Maximum Nc occurred on 63 DAT at TNAU-TNRRI, on 108 DAT at PUAT, and on 70 DAT at CRRI.

Root growth and nitrogen uptake in rice: concepts for modeling

13

Table 1. Total crop (leaves, stems, roots, and panicles) N uptake and root and shoot mass on different days after transplanting (DAT). TNAU-TNRRI, Aduthurai, 1988-89 WS. N application level (kg N/ha)

DAT

0

100

200

300

400

2.2 38.2 91.5 128.2 155.9 171.1 182.6 240.4 218.0

2.2 49.6 110.6 159.9 201.4 212.7 246.2 303.8 278.0

Crop N uptake (kg/ha) 2.2 13.9 17.0 28.1 37.9 41.7 43.5 41.8 52.5

0 18 25 32 39 46 53 63 81

2.2 22.9 38.5 53.7 60.7 71.1 88.0 96.0 108.0

2.2 27.8 64.8 81.9 97.0 108.9 130.8 165.6 167.0 Root mass (kg/ha)

0 18 25 32 39 46 53 63 81

19 260 326 370 421 463 482 537 530

19 377 494 590 672 734 848 836 836

19 491 620 764 880 930 1,120 1,183 997

19 50 710 925 1,150 1,324 1,502 1,790 1,507

19 521 775 1,030 1,250 1,525 1,750 2,003 1,968

85 1,211 2,738 4,465 5,952 7,371 8,986 12,514 13,660

85 1,370 3,050 5,150 7,184 8,636 10,443 13,669 14,671

Shoot mass (kg/ha) 0 18 25 32 39 46 53 63 81

85 706 1,184 1,880 2,501 3,058 3,566 4,303 5,478

85 916 1,745 2,757 3,593 4,422 5,532 7,806 9,891

85 984 2,341 3,853 5,011 6,132 7,390 9,809 11,856

of N applied A (kg/ha), is shown in Figure 1 for the different sites. At TNAU-TNRRI, the point of maximum N accumulation coincided with flowering, but at PUAT, it occurred at harvest. (For CRRI, Nc is not known after 70 DAT, but the small difference in Nc between 50 and 70 DAT suggests that uptake ceased at around 50 DAT.) Nitrogen uptake in the reference treatments (A = 0) varied from 50 to 60 kg N/ha. If these values represent the amount of soil-supplied N in the other treatments (A > 0), the fertilizer N recovery over the season was around 55% at TNAU-TNRRI, 40-55% at PUAT, and 60-65% at CRRI. Only at PUAT did overall N recovery decrease with increasing A. The linearity of Nc(A) found at the other two sites is consistent with findings of other studies. For numerous experiments in Indonesia, Van Keulen (1977) showed that Nc(A) is usually linear up to high N application levels. 14

ten Berge et al

Table 2. Total crop (leaves, stems, roots, and panicles) N uptake and root and shoot mass on different days after transplanting (DAT). PUAT, Pantnagar, 1987 WS. N application level (kg N/ha) DAT

0

60

120

180

240

0.9 32.6 98.4 119.3 123.7

0.9 41.1 117.4 131.1 136.3

Crop N uptake (kg/ha) 0 25 49 72 108a

0.9 12.8 25.5 43.3 51.7

0.9 19.1 51.8 76.1 77.7

0.9 25.3 70.4 100.1 106.1 Root mass (kg/ha)

0 25 49 72 108

13 317 638 1,016 ndb

13 349 976 1,434 nd

13 414 1,034 1,566 nd

13 479 1,276 1,806 nd

13 520 1,500 2,041 nd

32 1,797 6,813 11,722 9,035

32 2,043 7,774 12,485 9,978

Shoot mass (kg/ha) 0 25 49 72 108

32 853 2,599 6,502 4,910

32 1,187 4,295 9,570 6,737

32 1,481 5,393 10,887 7,741

a Values for 108 DAT do not include root N. b nd = not determined.

2. Relation between maximum root mass M, and N application in the three experiments. Maximum M, occurred on 63 DAT at TNAU-TNRRI, 72 DAT at PUAT, and 50 DAT at CRRI.

Root growth and nitrogen uptake in rice: concepts for modeling

15

Table 3. Total crop (leaves, stems, roots, and panicles) N uptake and root and shoot mass on different days after transplanting (DAT). CRRI, Cuttack, 1990 DS. N application level (kg/ha)

DAT

0

50

100

150

Crop N uptake (kg/ha) 22.0 57.8 60.5

20 50 70

20.3 85.5 90.2

26.0 119.9 120.2

30.1 152.5 160.8

Root mass (kg/ha) 20 30 40 50 70 92

223 323 605 602 500 500

237 523 743 905 861 820

286 539 979 1,116 952 858

333 561 1,078 1,317 932 899

Shoot mass (kg/ha) 539 1,185 2,120 3,292 4,976 5,645

20 30 40 50 70 92

559 1,337 2,654 4,290 6,421 7,960

671 1,768 3,723 5,522 8,117 10,020

707 1,895 4,161 6,100 9,958 10,085

Root biomass and root-shoot ratio as a function of N application The measured values of root biomass, Mr (kg/ha), are included in Tables 1-3. Mr is given in Figure 2 as a function of A, at the time when root mass reached its maximum. In all three experiments, root biomass generally increased linearly with A up to the highest application levels. Maximum root biomass values of 2000 kg/ha were found at TNAU-TNRRI and PUAT. The root-shoot ratio at all stages was independent of A at TNAU and CRRI, but not for young plants at PUAT (Fig. 3). Although the pattern of decreasing ratio with time was found in all cases, values of this ratio for given development stages differed across sites.

Discussion We need to know what the rate-limiting step or steps in N uptake are in order to decide what process or processes must be modeled in detail. At one extreme, the root system might be so efficient at extracting N from the soil that the rate of uptake is essentially independent of root and soil characteristics until the soil N concentration falls to a low level. At the other extreme, the rate of transport of N to root surfaces might be so slow that uptake is solely determined by the rate of transport-asgoverned by the soil transport characteristics and the N concentration-and the root surface area or RLD. 16

ten Berge et al

I

3. Changes in root-shoot ratios with time at different N application levels in the three experiments.

In between these extremes, root characteristics would have a greater or lesser effect on uptake depending on how well the plant is able to regulate the root system’s characteristics to match uptake with the rate of N transport through the soil. Clearly, as the crop grows and soil N becomes depleted, the rate-controlling step may change, and a complex model would be required to describe all eventualities. However, our purpose here is to identify the most critical process or processes and thereby simplify modeling as far as possible. In the following sections, we use our experimental results to test various hypotheses about rate-limiting steps. In the absence of detailed measurements of the changes in Root growth and nitrogen uptake in rice: concepts for modeling

17

[NH4+] and RLD with soil depth and time (z,t), we test the hypotheses with our data on Nc(t) and total root mass, Mr (t), by examining dNc/dt vs t, (dNc/dt)/Mr vs t, and fertilizer N recovery. Two assumptions are made: a) that N is taken up exclusively as NH4+; and (b) that within each experiment, RLD is proportional to M r, i.e., root length per unit root mass is independent of N application. Hypothesis 1: If [NH4+] is above a critical value, the uptake rate is independent of the amount of N applied. This hypothesis addresses the first extreme previously described. In this case, increases in Nc with A would be due to N application extending the period over which [NH4+] is above the critical value. To model this situation, we would only need information on the period when [NH4+] exceeds the critical value in some part of the root zone. We test the hypothesis by comparing rates of N uptake. Ideally, we need instantaneous uptake rates, but we necessarily only have average values over the time between sampling events (sampling intervals). This is awkward because differences in time averaged dNc/dt with A could be due to different durations of uptake at the level that is independent of [NH4+], (dNc/dt)max , within the sampling interval, as opposed to differences in (dNc/dt)max per se. The problem is avoided by comparing two consecutive sampling intervals that are not interrupted by N application. If dNc/dt depends on A in both such intervals, a variable duration of (dNc/dt)max is not responsible and Hypothesis 1 can be rejected. Only the TNAU-TNRRI data set satisfies the requirements for such an analysis. In the TNAU-TNRRI data, dNc/dt increases with A, both during the 0-18 and 1825 DAT periods (from data in Table 1). This finding is supported by other reported data—e.g., Schnier et al (1990) for both direct seeded and transplanted rice and Makarim et al(1991) for transplanted rice. We therefore reject Hypothesis 1. Hypothesis 2: If [NH4+] is above a critical value, the uptake rate is proportional to RLD but is otherwise independent of [NH4+]. This hypothesis addresses a situation in which uptake is governed by A only in so far as it influences plant growth and RLD through the plant internal N content, A model would then need to contain information on the relation between RLD and the plant internal N content, on the volume of soil in which [NH4+] exceeds the critical value at a particular time, and on RLD in these zones. Figure 4 shows uptake rates per unit root mass, (dNc/dt)/Mr, at different times for the period from transplanting to PI in the experiments at TNAU-TNRRI. (dNc/dt is averaged over the sampling interval; M r is estimated for the midpoint of sampling intervals by interpolation.) The figure shows that (dNc/dt)/Mr increased with A in young plants but decreased with crop age. A similar pattern was found for the other data sets. Values of (dN c/dt)/Mr of up to 13 (TNAU-TNRRI), 3 (PUAT), and 5 (CRRI) mg/ kg per d were found in young plants. These are probably not maximum values, as they are likely to occur immediately after N application. The decrease in (dNc/dt)/Mr with 18

ten Berge et al

4. N uptake rates per unit root mass at TNAU-TNRFU. Numbers on curves are N application levels (kg N/ha).

time is attributed to three possible causes: the decreasing [NH4+ ] as N is absorbed by the crop or otherwise removed from the system; a decrease in the fraction of the root system that is active or in the maximum uptake rate per unit root length as the crop ages; and the shallow penetration of N from topdressed urea, which leaves most of the root mass unexposed to fertilizer N. Since Hypothesis 1 is invalid, the dependence of (dNc /dt)/Mr on A found in all three data sets for the period from transplanting to PI must be due to differences in instantaneous uptake rates, as opposed to differences in maximal uptake duration. Therefore Hypothesis 2 also appears to be invalid. But as a result of continuous root growth during the intervals between sampling times, the exact quantity of root mass responsible for the difference in Nc measured over a particular sampling interval is unknown. Therefore, we need to confirm that (dNc /dt)/Mr did indeed increase with A, and we do this as follows. We define coefficients RN and RM such that: (1) and

(2) where D Nc is the change in N, over a particular sampling interval and subscript i indicates the fertilizer treatment. The ratios RM,i and R N,i thus compare treatment i with the next lower N level, i-1. Unlike absolute root mass, Mr, the relative root mass, RM, Root growth and nitrogen uptake in rice: concepts for modeling

19

is fairly constant over time, so its value at the end of the sampling interval can be used for the entire interval. Now, if R N,i > RM,i , the root system of the crop under treatment i takes up more N per unit root mass per time interval than under i-1. This appears to be the case for virtually all i’s in all the data sets, until after PI (Fig. 5). For the sampling events after PI, RN is often smaller than RM, implying that uptake did not increase in proportion to root mass with an increment in A. In summary, we conclude that, irrespective of the effect of N application on root growth, uptake per unit root mass increases with N application level. This is found to be true in all three experiments and at all A(i). We therefore reject Hypothesis 2. Hypothesis 3: Uptake rates are linearly related to both [NH4+ ] and RLD. This hypothesis addresses a situation in which uptake is solely limited by N transport to root surfaces by diffusion. We use analyses of fertilizer recovery to test this hypothesis. Fertilizer recovery depends on the rate of uptake compared with the rate of loss processes and is expected to be closely linked to RLD and uptake rate per unit root length. Nc (A) is plotted for different sampling times in Figure 6. Since urea was applied in different splits, the total amount A(t,i) applied by time t in treatment i is a cumulative value. The leftmost points on the curves apply to treatment i=0 (no fertilizer); the next point on each curve applies to i=l; and so forth. (The notation At(i) is used when evaluating a response across i levels, at fixed t.) Three types of recovery can be evaluated from these graphs. Cumulative recovery at A in Figure 6a is the slope of the line A-E. Incremental recovery—which is also a cumulative value—is found from the slope of the line A-B. Momentary recovery at t applies to the sampling interval preceding t. If the beginning of the time interval (t 1) and

5. Observed ratios of N uptake at two consecutive N application levels (R N ), compared with the corresponding ratios of root mass (RM ). 20

ten Berge et al

6. Relation between total crop (leaves, stems, roots, and panicles) N uptake (Nc ) and N application in the three experiments at different times.

the end of the time interval (t 2) are 25 and 46 DAT in Figure 6a, respectively, the momentary recovery for A(t 2 ,i) is It is the extra (over the i=0 treatment) uptake in treatment i as a fraction of the latest N split applied preceding t2 . For the first sampling event, momentary and cumulative recoveries are identical. The values for each of these recoveries are given in Table 4. In the early tillering stage (18 and 20 DAT for TNAU-TNRRI and CRRI, respectively), recoveries were in the 0-20% range for all treatments, but part of the applied N was still in the soil. No clear trend with A t (i) was found at this stage for any of the recovery types. Later during the tillering stage (25 DAT at TNAU-TNRRI and PUAT), recoveries of all three types increased slightly with At (i). Because constant recovery across i is equivalent to Root growth and nitrogen uptake in rice: concepts for modeling

21

Table 4. Cumulative, momentary, and incremental recovery of fertilizer N observed at TNAUTNRRI (Aduthurai, 1988-89), PUAT (Pantnagar, 1987), and CRRl (Cuttack, 1990).

DAT

18 25 46 63 81

18 25 46 63 81

TNAU-TNRRI N application (kg/ha) 100 0.18 0.43 0.43 0.54 0.55

0.18 0.43 0.44 0.54 0.56

200

TNAU-TNRRI N application (kg/ha)

300

400

100

200

Cumulative 0.14 0.48 0.50 0.62 0.57

recovery 0.16 0.50 0.65 0.66 0.55

0.18 0.47 0.64 0.66 0.56

0.18 0.25 0.46 0.61 0.69

Momentary 0.14 0.34 0.59 1.11 0.82

Incremental

recovery

0.10 0.53 0.57 0.70 0.59

0.21 0.53 0.93 0.75 0.51

25 49 72

60

0.21 0.44 0.55

120 Cumulative 0.21 0.50 0.47 Momentary

25 49 72

0.21 0.67 nda

0.21 1.08 0.40 Incremental

25 49 72

0.21 0.44 0.55

0.21 0.62 0.40

180 recovery 0.22 0.54 0.42

240

0.24 0.51 0.37

DAT

20 50 70

0.24 1.06 –0.07

20 50 70

0.18 0.29 1.16 0.91 0.35

50 100 150 Cumulative recovery –0.07 0.70 0.59

0.08 0.83 0.60

–0.07 2.22 0.29

0.08 2.32 –0.09

Incremental

recovery 0.24 0.62 0.32

recovery 0.16 0.33 1.10 1.19 0.53

CRRl N application (kg/ha)

Momentary

recovery 0.22 1.18 0.07

400

0.23 0.38 0.62 0.63 0.60

PUAT N application (kg/ha) DAT

300

0.28 0.42 0.20

20 50 70

-0.07 0.70 0.59

0.22 0.96 0.60

0.11 0.84 0.67 recovery 0.11 2.31 0.15 recovery 0.17 0.87 0.81

a nd = not determined since no third split was applied.

increasing absolute uptake rates, this response is consistent with the above observation that dNc/dt increased with At (i). The generally low recoveries during the early development stages are attributed to the limited size of the root system and possible transplanting damage; N demand is unlikely to limit uptake because in that case recovery would decrease with At (i). Recoveries at PI (46, 49, and 50 DAT for TNAU-TNRRI, PUAT, and CRRI, respectively) were generally high. Cumulative, momentary, and incremental recoveries all increased substantially with At (i) in each data set, except at the highest i. The 22

ten Berge et al

latter lack of response shows that Hypothesis 3 is not valid at these high application levels. That does not contradict the earlier observation that Mr and (dNc /dt)/Mr continued to respond to At(i) up to the highest i, but it does imply that these responses were not strong enough to cause a sustained increase in recovery. Momentary recoveries at PI exceeded 1.0 in all data sets, indicating that some N was still available from the basal dressing, even after 3-4 wk. Even so, such high values suggest that practically all N applied around early tillering had been absorbed by the crop by PI. Incremental recoveries were near 1.0 for the two highest N application levels in CRRI, and for i=3 in TNAU-TNRRI. The values of this variable were lower at PUAT, but still reached 0.62 for i=2 and i=3. Results were less consistent for the sampling intervals from PI to flowering. Momentary recoveries were poor in all treatments at PUAT and CRRI. All recovery types decreased from PI to flowering at PUAT, except for i=1. Moreover, recoveries decreased here with increasing i. At CRRI, too, recoveries decreased with time. At TNAU-TNRRI, recoveries remained fairly high and decreased only after flowering (63 DAT). Then, momentary recovery also decreased across i levels. The decline in recovery after PI (or later in the TNAU-TNRRI case) leads us to conclude that either Hypothesis 3 is invalid at this stage, or that the fraction of roots active in N uptake decreases as time proceeds. For the latter to be a valid explanation, however, we have to accept that this sencscence is more pronounced at higher i levels, which is unlikely. The larger root system resulting from N application makes the plant better able to compete for N against loss processes, and hence a larger fraction of N in the root zone is recovered. This was indeed observed during the vegetative phase, which supports Hypothesis 3. The phenomenon of increasing recovery with A, however, may also be caused by other mechanisms involving positive feedback between Nc and dNc /dt. In the following section, we try to explain the observed trend in recovery vs A(i) on the basis of observed root masses. Assuming that the rates of N uptake and loss are both governed by first-order processes, we obtain for the recovery fraction in treatment j as compared with that in treatment i (Appendix 1)

(3) Figure 7 shows cumulative recoveries during the vegetative period calculated with Equation 3 from observed M rj/Mri and observed recovery at the lowest application level. We conclude that cumulative recoveries at the higher A(i) levels can be reasonably well-explained by differences in observed root mass, without invoking other mechanisms. Therefore, Hypothesis 3 cannot be disproved on the basis of these observations, except in the highest i of each experiment. The material presented earlier in disproving Hypothesis 2 provides additional support for Hypothesis 3; this is essentially independent of the recovery analysis. Root growth and nitrogen uptake in rice: concepts for modeling

23

7. Observed cumulative fertilizer N recovery compared with recovery calculated from Eqn (3). The first datum of each series is the observed ri=1, and is used as the starting point in Eqn (3) to calculate recoveries at higher N application levels.

Hypothesis 4: If [NH4 +] is above a critical value, the uptake rate is regulated by a feedback mechanism in the plant other than via RLD. This hypothesis addresses a situation in which the uptake rate per unit root length, or the fraction of the total root length that is active in uptake, is regulated by the plant in accordance with its internal N content. A high crop growth rate induced by Nc is likely to be associated with a high carbohydrate supply to the roots, which would tend to increase N uptake through a positive feedback loop. At high growth rates, possibly the rate of detoxification of ammoniacal N via amino acid synthesis results in another positive feedback loop between Nc and N uptake rate. The present experimental results do not allow differential testing of Hypotheses 3 and 4.

Root-shoot ratio

The lack of sensitivity of the root-shoot ratio to N application observed in these experiments is unusual in crops. Numerous studies, including the early work by Brouwer (1962a,b, 1963), have shown that higher nutrient supply results in lower rootshoot ratio. The concept of functional equilibrium evolved from such observations, and although the mechanisms are often questioned (Lambers 1983, van Andel et al 1983), the occurrence of shifts in dry matter partitioning induced by nutrient supply is widely accepted (Wilson 1988). Partitioning models describe root and shoot growth in terms of C and N balances, applying a resistance coefficient for transport of these compounds within the plant (Thornley 1972) or assuming an optimal allocation strategy by the plant (Johnson and Thornley 1987). On the basis of such concepts, there is no obvious reason why rice should behave differently from other crops. Peculiar characteristics of the lowland rice system include the low mechanical resistance to root growth; the root’s O2 demand being met by diffusion through internal root gas channels; the uptake of N is as NH4+ ions; and the particular rhizosphere chemistry of anaerobic soils. Kirk et al 24

ten Berge et al

(1993) state that NH4 + uptake may be hampered by rhizosphere acidification arising from Fe2+ oxidation and NH 4+ uptake, resulting in increased NH4+ sorption by the soil complex. A phenomenon possibly important in view of O2 supply is the formation of surface root mats (Kirk and Bouldin 1991), the development of which was shown to be enhanced by NH4 + as opposed to NO 3 – nutrition (Soezima and Kawata 1969). Linearity of uptake as a function of N application Field studies generally show that Nc (A) is linear up to high N application levels. Assuming that Hypothesis 3 is valid and that root length per unit root mass is independent of A, this observation conflicts with the observation that the absolute root mass increases with A. Under these conditions, the slope of N c (A)—i.e., cumulative recovery—should increase with A. The apparent contradiction arises because in most studies, Nc (A) is evaluated on a full-season basis and the positive response of cumulative recovery to N application at any time between crop establishment and PI is then masked by the neutral or negative response after PI, as demonstrated in this study.

Conclusions Hypothesis 3 seems to be the easiest starting point for the development of an uptake model. The most important variables would be [NH4+ ] as a function of (z,t), effective NH4+ diffusivity, and RLD. Assessing the active fraction of RLD and its dependence on (z,t) remains problematic. This research showed no consistent relationship after PI among uptake, application level, and root mass. The root-shoot ratio decreased with development stage and was independent of A t (i) at every stage. The possibility that this may be related to root O 2 supply bears further investigation. Root mass and the rate of N absorption per unit root mass both increased with application At (i), resulting in high recovery levels for N. Instantaneous N uptake rate per unit root mass is likely to be roughly proportional to [NH4 +] up to high [NH4 + ] values. Cumulative N uptake over a finite time interval is higher than the first-order function of At (i) during vegetative growth. An approximate calculation shows that the response of cumulative recovery to At (i) can be explained by differences in root mass. Nitrogen absorption by the crop over the entire season was shown to be a linear function of A. This linearity is the result of a positive feedback between cumulative uptake and uptake rate during vegetative growth, followed by a negative or neutral feedback after PI. Nitrogen uptake in the first 3 wk after transplanting was low, but proportional to At (i), implying that demand did not limit N uptake. Ten to twenty percent of N applied basally was absorbed, and momentary recoveries exceeding 1.0 found for the next sampling interval indicate that this N was still available for absorption at the time of the first topdressing.

Root growth and nitrogen uptake in rice: concepts for modeling

25

Uptake between PI and flowering varied among sites, with low values at PUAT and CRRI and high values at TNAU-TNRRI. The highest momentary recoveries at this stage were found for low i at CRRI and PUAT. The generally poor recovery of this later split cannot be entirely due to soil or management conditions, but to the crop’s N status as well. A similar effect was observed for application at flowering in TNAU-TNRRI. Further work is needed to explain decreases in momentary and cumulative recoveries as a function of time in the later stages of plant development, and the decrease in recovery with At(i) at these later stages.

References cited Andel O M van, Soekarjo R, Verkaar H J PA, eds. (1983) Functional equilibrium between shoots and roots. Neth. J. Agric. Sci. 31(4): Brouwer R (1962a) Distribution of dry matter in the plant. Neth. J. Agric. Sci. 10:361-376. Brouwer R (1962b) Nutritive influences on the distribution of dry matter in the plant. Neth. J. Agric. Sci. 10:399-408. Brouwer R (1963) Some aspects of the equilibrium between overground and underground plant parts. Mededeling 213 van het IBS, 31-39. De Datta S K, Patrick W H Jr, eds. (1986) Nitrogen economy of flooded rice soils. Developments in plant and soil sciences. Kluwer Academic Publishers, Dordrecht, The Netherlands. 186 p. Drenth H, ten Berge H F M, Meijboom F W (1991) Effects of growth medium on porosity and branching of rice roots. Pages 162-175 in Simulation and systems analysis for rice production (SARP). F.W.T. Penning de Vries, H.H. van Laar, and M.J. Kropff, eds. PUDOC, Wageningen, The Netherlands. Johnson I R, Thornley J H M (1987) A model of shoot:root partitioning with optimal growth. Ann. Bot. 60: 133- 142. Keulen H van (1977) Nitrogen requirements of rice with special reference to Java. Contrib. Cent. Res. Inst. Agric. Bogor 30: 1-67. Kirk G J D, Bouldin D R (1991) Speculations on the operation of the rice root system in relation to nutrient uptake. Pages 195-203 in Simulation and systems analysis for rice production (SARP). F.W.T. Penning de Vries, H.H. van Laar, and M.J. Kropff, eds. PUDOC, Wageningen, The Netherlands. Kirk G J D, Solivas J L, Begg C B M (1993) The rice root-soil interface. Pages in Rice roots and nutrient and water use. G.J.D. Kirk, ed. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Kropff M J, Cassman K G, van Laar H H (1993) Quantitative understanding of the irrigated rice ecosystem for achieving high yield potential in (hybrid) rice. Pages in Hybrid rice. S.S. Virmani, ed. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Lambers H (1 983) ‘The functional equilibrium,’ nibbling on the edges of a paradigm. Neth. J. Agric. Sci. 31:305-311. Makarim A K, Hidayat A, ten Berge H F M (1991) Dynamics of soil ammonium, crop nitrogen uptake, and dry matter production in lowland rice. Pages 214-228 in Simulation and systems analysis for rice production (SARP). F.W.T. Penning de Vries, H.H. van Laar and M.J. Kropff, eds. PUDOC, Wageningen, The Netherlands. Rachhpal-Singh, Kirk G J D (1993a) A model for predicting the fate of nitrogen fertilizer in lowland ricefields. I. Theory. J. Soil Sci. (in press)

26

ten Berge et al

Rachhpal-Singh, Kirk G J D (1993b) A model for predicting the fate of nitrogen fertilizer in lowland ricefields. II. Predicted dynamics of inorganic carbon, nitrogen and acidity in the soil and floodwater. J. Soil Sci. (in press) Schnier H F, Dingkuhn M, De Datta S K, Mengel K, Faronilo J E (1990) Nitrogen fertilization of direct-seeded flooded vs transplanted rice. I. Nitrogen uptake, photosynthesis, growth and yield. Crop Sci. 30:1276-1284. Soezima M, Kawata S (1969) “Lion tail like root” formation in rice plant and soil conditions. Proc. Crop Sci. Soc. Jpn. 38:442-446. Thiyagarajan T M, Mohandass S, Palanisamy S, Kareem A A (1991) Effect of nitrogen on growth and carbohydrate partitioning in rice. Pages 132-136 in Simulation and systems analysis for rice production (SARP). F.W.T. Penning de Vries, H.H. van Laar, and M.J. Kropff, eds. PUDOC, Wageningen, The Netherlands. Thornley J H M (1972) A model to describe the partitioning of photosynthate during vegetative plant growth. Ann. Bot. 36:419-430. Willigen P de, van Noordwijk M (1987) Roots, plant production and nutrient use efficiency. Ph D thesis, Wageningen Agricultural University, Wageningen, The Netherlands. 282 p. Willigen P de, van Noordwijk M (1991) Modelling nutrient uptake: from single roots to complete root systems. Pages 277-295 in Simulation and systems analysis for rice production (SARP). F.W.T. Penning de Vries, H.H. van Laar, and M.J. Kropff, eds. PUDOC, Wageningen, The Netherlands. Wilson J B (1988) A review of evidence on the control of shoot:root ratio, in relation to models. Ann. Bot. 61:433-449. Yoshida S (1981) Fundamentals of rice crop science. International Rice Research Institute, P.O. Box 933, Manila, Philippines. 269 p.

Notes Authors’ addresses: H.F.M. ten Berge, Centre for Agrobiological Research, P.O. Box 14,6700 AA Wageningen, The Netherlands; T.M. Thiyagarajan, Tamil Nadu Agricultural University, Aduthurai, Tamil Nadu 612101, India; B. Mishra, G.B. Pant University of Agriculture and Technology, Pantnagar, District Nainital, Uttar Pradesh 263145, India; K.S. Rao and R.N. Dash, Central Rice Research Institute, Cuttack, Orissa 753006, India. Citation information: Kirk G J D, ed. (1994) Rice roots: nutrient and water use. International Rice Research Institute, P.O. Box 933, Manila, Philippines.

Root growth and nitrogen uptake in rice: concepts for modeling

27

Appendix 1 Let k1(z,t ) and k2(z,t ) be the first-order rate constants for uptake and loss, respectively (NB k2 accounts for all losses of added fertilizer), and let c( z,t ) be [NH4+] at time t and depth z in the active root zone. The ratio X of uptake to loss over time interval t l -t2 and depth interval z1-z 2 is then

(A1)

Under Hypothesis 3, k1 (Id) is proportional to root mass density mr (kg/m3), specific root length Lr (m/kg), and a coefficient D (m2 /d) accounting for diffusion from the bulk soil solution to the root surface: (A2) The ratio in Equation A1 is equal to r/(l-r), where r is the recovery fraction. Writing this ratio Xi for treatment i and X j for treatment j, we have (A3) Now, assuming that a) ci (z,t ) is proportional at any ( z,t ) to the applied amount A( t,i); b) Lr is not affected by A; and c) mr (z,t ) varies across treatments by a factor Mri(t) /Mrj (t) which is independent of t , then Equation A3 reduces to

(A4) or

(A5)

28

ten Berge et al

Genetic variation in nitrogen uptake by rice and the effects of management and soil fertility P.C. Sta Cruz and G. Wada

Variation in nitrogen (N) uptake among rice genotype groupings (ecotype, hybrid, inbred, modern, and traditional) and between plant types and growth types are investigated for low to moderate fertilizer levels. Rice genotypes of varying growth durations (GD) were evaluated in terms of N uptake and N use efficiency (NUE). Total N in the plant at late spikelet initiation, flowering, and maturity increased with GD. There were substantial differences in total N in the plant up to 30 d after transplanting (DT), but not after maximum tillering (MT). Total N in the plant at 30 DT (NE) was positively correlated with sink size, grain yield, and N harvest index (NHI), but not with percent spikelet degeneration or percent filled spikelets in short- and mediumduration genotypes. This suggests that NE is an important determinant of sink size and grain yield in short- and medium-duration genotypes. Narrow spacing and high basal N application shortened the duration of exponential N uptake and increased the rate of uptake during this phase. A slow-release fertilizer and high soil fertility increased the rate of N uptake during the middle and late growth stages. In general, genotypes with low N uptake during the dry season at high N levels have relatively low N uptake during the wet season at low N levels. The effects of cultural practices and soil fertility status, shown in altered durations and rates of N absorption, were reflected in changes in N uptake patterns.

One strategy for improving the utilization of soil and fertilizer N in rice production systems is to exploit genotypic differences in N absorption ability. Such differences can be sought among a) different genotype groupings: ecotypes, hybrids, inbreds, and modem vs traditional cultivars; and b) different morphological and physiological features: growth type, panicle type, and growth duration (GD). Genotypic differences can also interact with cultural practices and environmental variables. This paper examines the sources of variation in N uptake in various genotype groupings and their interaction with cultural practices and environmental variables. The relationship

between N uptake efficiency and yield-determining parameters in various genotypes is also considered.

Variation in N uptake among rice ecotypes, plant types, and hybrids Ecotypes

Variation in N uptake between indica and japonica ecotypes has been observed (Tanaca et al 1964). Typical indicas, such as Peta and BPI76, show high N uptake at the early growth stages, reflected by increased leaf area. N uptake increases rapidly until panicle initiation (PI), but thereafter tends to plateau and then decline as N and other nutrient elements are lost from the plants. By contrast, the japonica Tainan 3 has low N uptake in the early growth stages with a gradual increase in the amount of N in the plant sustained until maturity. The faster N uptake by the indica types in the early growth stages results in higher total N uptake. Ehara et al (1990) found that at the seedling stage, indica and japonica types showed no distinct differences in N uptake. Semidwarf indicas showed an increase in leaf area at high N levels, suggesting high N uptake. Japonica types had longer culms under high N levels (Maruyama and Tajima 1988). Seedling growth and photosynthesis in improved indicas and japonicas are not any more adversely affected by low N levels than those of primitive cultivars or wild species. Cook and Evans (1983) found, however, that improved cultivars showed a proportional increase in seedling dry weight with increasing N concentration. Although variations in total N uptake between ecotypes do exist, generalizations cannot be made because other variables within ecotypes, such as plant type and GD, may dominate N uptake.

Plant types

Generally, genotypes that perform well at low N levels are tall, have a low percentage of productive tillers, have high dry matter production, are susceptible to lodging, and have low panicle to straw weight ratios. They grow relatively fast in the early growth stages and cover the fields even without N addition. At high N levels, higher growth rates in the early growth stages result in mutual shading during the middle and late growth stages. This reduces N uptake because net photosynthesis is decreased, resulting in fewer tillers and weaker roots. By contrast, genotypes that respond well to high N have short stature, high panicle to straw weight ratios, high percentage of productive tillers, and slow growth in the early growth stages. This plant type shows little mutual shading (Tanaka et al 1964). Current improved lowland varieties are short- to medium-maturing and intermediate in panicle number type. Although they produce many tillers and have large leaf area indices (LAI), they are short and have erect leaves. Mutual shading is avoided by improved leaf canopy architecture. Yoshida (1981) identified short and stiff culms, erect leaves, and high tillering as the most important characters for lodging resistance; increased photosynthesis as a function of LAI; and yield increment as related to number of panicles. 30

Sta Cruz and Wada

Miyagawa (1981) reported that in indicas, total N uptake of the panicle-number type (14-16 panicles/hill) is generally higher than that of the panicle-weight type (10-11 panicles/hill), even at different fertilizer N levels. Hybrids High total N uptake, dry weight, and leaf area have been observed in F1 hybrids (Blanco and Akita 1988). This was more evident in japonica-indica crosses than in japonicajaponica crosses. Similar results were obtained with F1 hybrids of IR varieties and lines (Sta Cruz et al 1988). The F1 hybrids have higher N uptake compared with their inbred parents, particularly at 3 wk after transplanting (WT). At 5 WT, differential N uptake diminished except for the hybrid with medium GD (125 d). The medium-GD hybrids had consistently higher N uptake until maturity. Likewise, we found that F1 hybrids generally had higher N harvest index (NHI).

Effect of growth duration on N uptake To determine the effect of GD on total N uptake at different growth stages, we planted 60 IR varieties and advanced lines with varying GDs at 20- × 20-cm spacing under 0 and 90 kg N/ha levels (Wada and Sta Cruz 1989).

1. Total N in the plant at flowering in relation to growth duration of different rice genotypes under two N levels. IRRI Farm, 1987 DS.

Genetic variation in nitrogen uptake

31

2. A model for N absorption by rice plants and the occurrence of early panicle initiation in relation to accumulated effective thermal index (Matsushima 1966, Wada 1971, Shoji and Mae 1984).

Total N uptake at critical plant growth stages was positively correlated with GD. At zero and 90 kg N/ha fertilizer rates, the increase in GD was accompanied by a linear increase in N uptake at flowering (Fig. 1), at the late stage of spikelet initiation (Y = 1.94 + 0.78x, r = 0.815**, 90 kg N/ha) and at maturity (Y = 2.07 + 0.069x, r = 0.789**, 90 kg N/ha). The short-duration genotypes generally had lower total N uptake than the long-duration genotypes at flowering, late stage of spikelet initiation, and maturity. The variation in N uptake among genotypes with different GD can be explained by differences in the vegetative lag phase (VLP). The VLP is the difference in days between the occurrences of panicle initiation (PI) and maximum tillering (MT). The PI of short-duration genotypes occurred before or at the same time as MT when total N in the plant was still low. By contrast, the PI of long-duration genotypes occurred after MT when total N in the plant was already high (Fig. 2). Variation in N uptake at particular growth stages Wada and Sta Cruz (1989) evaluated differences in total N uptake of rice genotypes at different growth stages using 10 rice genotypes with the same GD (111 d). Genotypes with the same GD were used to correct for the possible variation in VLP duration. 32

Sta Cruz and Wada

Table 1. Differences in total N uptake of short-duration (111 d) genotypes at different growth stages. IRRl Farm, 1986 DS. Genotype

IR36 lR2588-7-3-1 lR25261-135-1-1 lR29658-69-2-1-2 lR29692-94-2-1-3 IR29725-109-1-2-1 lR31868-64-2-3-3-3 lR32419-81-2-3-3 lR32429-47-3-2-2 lR32429-122-3-1-2

Total N uptake (g/m 2 )a Active tillering (N1) 1.4 bc 1.6 b 2.0 a 1.5 b 1.4 bc 1.3 c 1.3 c 1.2 c 1.3 c 1.5 b

Maximum dN1 tillering (N2-N1) (N2)

flowering (N3)

2.6 c 1.2 c 9.9 bc 3.2 b 1.6 b 10.4 b 4.4 a 2.4 a 11.9 a 2.8 bc 1.3 bc 10.0 bc 2.5 cd 1.1 cd 9.7 c 2.6 c 1.3 bc 9.9 c 2.5 cd 1.2 c 9.8 c 2.1 e 0.9 d 9.2 d 2.2 de 0.9 d 9.8 c 2.6 c 1.1 cd 10.0 bc

dN2 (N3-N2) 7.3 7.2 7.5 7.2 7.2 7.3 7.3 7.1 7.6 7.4

ab ab a ab ab ab ab b a ab

Maturity (N4)

dN3 (N4N3)

12.6 bc 13.2 b 14.4 a 12.4 c 12.3 c 12.6 bc 12.5 c 11.6 c 12.0 c 12.4 c

2.7 2.8 2.5 2.4 2.6 2.7 2.7 2.4 2.2 2.4

a a a ab a a a ab b ab

a In a column, means followed by a common letter are not significantly different at the 5% level by DMRT.

Table 2. Correlation coefficients for N uptake between 0 and 90 kg N/ha and between wet (WS) and dry seasons (DS). a IRRl Farm, 1986 WS and 1987 DS. Parameter 0 N vs 90 N DS WS DS vs WS 0N 90 N

Total N uptake 30 DT

Flowering

Maturity

0.835 0.850

0.905 0.889

0.887 0.781

0.830 0.815

0.880 0.771

0.883 0.841

a All values are significant at the 1% level. n = 60, mean of two replications.

Variation in N uptake among genotypes was apparent during the early growth stages (30 DT) and MT (30-35 DT). Little variation in N uptake was noted from MT to flowering and from flowering to maturity (Table 1). This observation can be explained by the availability of NH4 -N in the plow layer as reported by Wada et al(l989). The NH4 -N in the plow layer decreased exponentially from 7 DT until 35 DT and remained constant at a very low level after 35 DT (Fig. 3). During the early growth stages, NH4-N in the plow layer was not limiting, and variation in N uptake could only have been due to differences in N absorption abilities between the genotypes. The availability of NH4 -N in the plow layer did limit N uptake after MT. Sta Cruz (1990) found that there were positive correlations between 0 and 90 N treatments and between the wet season (WS) and dry season (DS) for total N uptake at 30 DT, flowering, and maturity (Table 2). The genotypes with low total N uptake under low N level also had low total N uptake under high N level (Fig. 4). The same relationship was obtained for N uptake between WS and DS (Fig. 5).

Genetic variation in nitrogen uptake

33

3. Behavior of NH4 -N in the plow layer (Wada et al 1989).

Effect of early N uptake on yield-determining parameters Because there were significant differences in N uptake at the early growth stage, a followup study was made using 60IR varieties and lines with varying GDs to determine the relationship of the amount of N in the plant at the early growth stage (NE) to sink size, grain yield, and NHI (Wada and Sta Cruz 1990). NE was positively correlated with sink size, grain yield, and NHI, but not with degenerated spikelets and percent unfilled spikelets in short- and medium-duration 34

Sta Cruz and Wada

4. Relationship between total N uptake at flowering under 0 and 90 kg N/ha in rice genotypes with different growth durations. IRRI Farm, 1987 DS.

5. Relationship between total N uptake at maturity in dry-season (DS) and wet-season (WS) croppings in rice genotypes with different growth durations. IRRI Farm, 1986 WS and 1987 DS.

Genetic variation in nitrogen uptake

35

Table 3. Correlation coefficients between N uptake at 30 DAT (NE) and yield-determining factors in rice genotypes of different growth durations. a IRRI Farm, 1986 WS and 1987 DS. Growth duration (d)

Y

S

DEG S

R

NHI

1987 DS 102-104 108-109 113-116 119-122 124-127 130 <

0.551** 0.580** 0.600** 0.708 ns 0.660 ns 0.528 ns

0.858** 0.723** 0.897** 0.920** 0.838** 0.534 ns

0.696** 0.299 ns 0.593 ns 0.669 ns 0.936** 0.885**

–0.264 –0.449 –0.394 –0.529 –0.170 –0.538

ns ns ns ns ns ns

0.841** 0.831** 0.938** 0.972** –0.040 ns –0.041 ns

–0.319 –0.665 –0.483 –0.361 –0.073

ns ns ns ns ns

0.796** 0.894** 0.839** 0.532 ns –0.283 ns

1986 WS 104 111-118 117-122 125-127 130 <

0.7,26** 0.685 * 0.948** 0.499 ns 0.269 ns

0.817** 0.894** 0.849** 0.623 ns 0.291 ns

0.339 ns 0.428 ns 0.288 ns 0.725* 0.849**

a *, ** = significant at 5 and 1% level, respectively. ns = not statistically significant at 5% levet. Y = grain yield, S = sink size, DEG S = degenerated spikelets, R = percent filled spikelets, and NHI = nitrogen harvest index.

genotypes (Table 3). This result suggests the importance of NE as a parameter for increasing both sink size and yield of short- and medium-duration genotypes, without increasing the percent unfilled spikelets and degenerated spikelets. The NHI increased with NE in short- and medium-duration genotypes. This further suggests the importance of NE as a parameter in increasing the yield and NUE of shortand medium-duration genotypes. In long-duration genotypes, NE was not correlated with sink size, grain yield, and NHI. Although the proportion of filled spikelets was not affected by NE, degenerated spikelets increased with NE (Table 3). This result suggests that N uptake during the middle and late growth stages is an important yield determinant in long-duration genotypes.

Effect of cultural practices and soil fertility on the pattern of N uptake Takahashi et al (1976) and Wada et al (1989) described the pattern of N absorption by rice plants as biphasic. The amount of N in the plant increases exponentially during the early growth stages and linearly during the middle and late growth stages. The transition point between the two phases coincided with MT for 20- × 20-cm plant spacing (Fig. 2). However, the uptake pattern may be altered by both native soil fertility and cultural practices (i.e., plant spacing and the amount, type, and timing of fertilizer). The effect is usually manifested by a shift in transition time. This shift implies a change in the timing of MT which may alter the amount of N in the plant during the middle and late growth stages. 36

Sta Cruz and Wada

Plant spacing and rate of basal fertilizer Wada et al (1989) found that narrow spacing (20 × 10 cm vs 20 × 20 cm and 20 × 30 cm) and a high rate of basal N (120 vs 60 kg N/ha) shortened the duration of the exponential phase (Figs. 6 and 7). Although the duration of N uptake was shortened, the rate of N uptake in the exponential phase was higher. Total N uptake was higher with narrow spacing (area basis) and with basal N (plant basis). On the other hand, the rate of N uptake in the linear phase was not affected by narrow spacing and high basal N. However, total N uptake was higher due to the relatively high amount of N absorbed during the exponential phase. Use of slow-release fertilizer Wada et al (1989) found that the use of a slow-release fertilizer (urea coated with plastic resin), as compared with ammonium sulfate, resulted in longer duration of the exponential phase (Fig. 8). Although the rate of N uptake using Meister was relatively lower than with ammonium sulfate, the lengthening of the exponential phase increased total N uptake with Meister at MT; Meister increased the rate of N uptake during the linear phase.

6. N absorption patterns of rice plants grown at different plant spacings (Wada et al 1989).

Genetic variation in nitrogen uptake

37

7. N absorption patterns of rice plants grown with two basal N rates (Wada et al 1989).

8. N absorption of rice plants grown with two N fertilizer sources (Wada et al 1989).

38

Sta Cruz and Wada

9. N absorption patterns of rice plants grown under two different soil fertility levels (Wada et al 1989).

Soil fertility level N uptake patterns in two experimental sites with varying soil fertilities were compared by Wada et al (1989). One field (Field E) had higher percentages of C and total N, a higher C-N ratio, and higher cation exchange capacity and ammonification rate than the other field (Field M). The rate of N uptake during the exponential phase was higher at Field E, although duration was shorter because of the fertility level. The effect of higher soil fertility level was apparent during the linear phase where the rate of N uptake was higher (Fig. 9).

Conclusions There is substantial variation in N uptake between rice ecotypes, plant types, hybrids, and plants with different GD. Plant type (based on leaf erectness and panicle number) and GD are the most important factors. Nitrogen uptake during late spikelet initiation, flowering, and maturity increases with GD. The variation in N uptake among genotypes with different GD is due to differences in the duration of VLP. Short-duration genotypes generally have shorter VLP than long-duration ones and consequently have lower total N uptake. Variation in N uptake among genotypes of the same GD is apparent during the early growth stages (30 DT). During these stages, NH 4 -N in the plow layer is likely to be high, so N uptake is controlled by the N absorption ability of the genotype. During the middle and late growth stages, the availability of NH4-N in the plow layer is low and therefore may limit N uptake. Hence, differences in N uptake among rice genotypes during the middle and late growth stages were not observed.

Genetic variation in nitrogen uptake

39

The amount of N taken up during the early growth stages is positively correlated with sink size, yield, and NHI. Therefore, N uptake at these stages is an important parameter for increasing yield and N utilization of short- and medium-duration genotypes. Generally, native soil fertility and cultural practices (such as plant spacing and type and management of fertilizer) affect the N absorption pattern of various genotypes differently. The effect is manifested by changes in the duration and rate of N absorption, which in turn alter the amount of N in plants at critical growth stages, leading to profound effects in NUE.

References cited Blanco L C, Akita S (1988) Physiological mechanism of heterosis in seedling growth of japonica-indica F1 rice hybrids. Jpn. J. Crop Sci. 57 (extra 1):99-100. Cook M G, Evans L T (1983) Nutrient responses of seedlings of wild and cultivated Oryza species. Field Crops Res. 6:205-218. Ehara H, Tsuchiya M, Ogo T (1990) Fundamental growth response to fertilizer in rice plants. Varietal difference in the growth rate at seedling stage. Jpn. J. Crop Sci. 59(3):426-434. Maruyama S, Tajima K (1988) Growth response to nitrogen in Japonica and Indica rice varieties. II. Differences in the rate of increase in culm length and leaf area due to nitrogen fertilization. Jpn. J. Crop Sci. 57(4):692-698. Miyagawa S (198 1) Nitrogen balance in paddy fields planted with different Indica varieties. Jpn. J. Trop. Agric. 25:107-114. Sta Cruz, P C (1990) Sink development in rice genotypes of different growth durations as affected by nitrogen nutrition, plant spacing, and shading. Ph D dissertation, University of the Philippines at Los Baños, Laguna, Philippines. 135 p. Sta Cruz P C, Wada G, Virmani S S (1988) Nitrogen response of F1 hybrid of rice plants. Jpn. J. Crop Sci. 57 (extra 1):247-248. Tanaka A, Navasero S A, Garcia C V, Parao F T, Ramirez E (1964) Growth habit of the rice plant in the tropics and its effect on nitrogen response. IRRI Technical Bulletin 3. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Takahashi J, Wada G, Shoji S (1976) The fate of fertilizer nitrogen applied to the paddy field and its absorption by rice plants. VI. Influence of a thermal factor on the soil ammonium nitrogen and the absorption of nitrogen by rice plants. Proc. Crop Sci. Jpn. 45:213-219. Wada G, Aragones D V, Aragones R C (1989) Nitrogen absorption pattern of rice plants in the tropics. Jpn. J. Crop Sci. 58(2):225-231. Wada G, Sta Cruz PC (1989) Varietal differences in nitrogen response of rice plants with special reference to growth duration. Jpn. J. Crop Sci. 58(4):732-739. Wada G, Sta Cruz P C (1990) Nitrogen response of rice varieties with reference to nitrogen absorption at early growth stage. Jpn. J. Crop Sci. 59(3):540-547. Yoshida S (1981) Fundamentals of rice crop science. International Rice Research Institute, P.O. Box 933, Manila, Philippines. 269 p.

40

Sta Cruz and Wada

Notes Authors' addresses: P.C. Sta Cruz, Agronomy, Soils, and Plant Physiology Division, PhilRice, Maligaya, Muñoz, Nueva Ecija, Philippines; G. Wada, 8-7 Miagino 1, Chome Niyaginoko Sendai 983 Japan. Citation information: Kirk G J D, ed. (1994) Rice roots: nutrient and water use. International Rice Research Institute, P.O. Box 933, Manila, Philippines.

Genetic variation in nitrogen uptake

41

Use of molecular markers to evaluate rice genetic variation in associative N2 fixation, N uptake, and N use efficiency P. Wu, J.K. Ladha, S.R. McCouch, and S.B. Teng

Significant differences among rice varieties in associative N2 fixation, N uptake, and N use efficiency (NUE) have been found over the last 10 yr. Three methods were used: N balance (a measure of the plant’s ability to exploit soil N), acetylene reduction, and 15N dilution (both measures of N2 fixation). The advantages and limitations of these methods for varietal screening are outlined below together with a summary of our work over the past decade. It was found that IR42 consistently had high N 2 fixation and N balance and performed well in soils with low N fertility. Palawan, a local variety, had low N2 fixation and N balance. In 15N dilution experiments in a greenhouse and the field, IR42 and Palawan showed significant differences in associative biological N, fixation, N uptake, and NUE. There was roughly a 20% difference in atom % 15N excess between IR42 and Palawan. There were strong correlations between atom % 15 N excess and the chlorophyll content of the flag leaf, between N uptake and yield, and between NUE and yield. In a primary restriction fragment length polymorphism (RFLP) survey comprising 11 genotypes with 4 enzymes and 44 probes, we found 67% polymorphism between IR42 and Palawan. Based on the survey results, we have selected 120 polymorphic probes throughout the 12 chromosomes for tagging genes underlying the traits identified in a parental survey with 180 probes with 6 enzymes ( Eco RV, Sc al, Hindlll, Xba l, and Dra l). The enzymes of X bal and Dra l performed well in detecting polymorphisms between the test genotypes; the enzymes of Hindlll, Sc a l, and EcoRV were superior to EcoRI.

The components of rice genetic variation in N efficiency include differences in ability to foster associative biological N, fixation (BNF) in the rhizosphere (Ladha et a1 1987, 1988; Watanabe et a1 1987), differences in N uptake efficiency (App et a1 1986), and differences in internal N use efficiency (NUE) (Ladha et a1 1987, 1988; De Datta and Broadbent 1988; Broadbent et al 1987). The plant and soil variables controlling associative BNF have been described (Balandreau and Knowles 1978, Diem et a1 1978,

Asanuma et al 1979, Nishizawa et al 1983, Murty and Ladha 1987) and diallel cross analysis has shown that the ability to foster associative BNF is heritable and is a quantitative character (Iyama et al 1983). There is currently little information on the heritability of N uptake and NUE in rice. The advent of molecular marker technology makes it possible to identify the genetic loci responsible for particular phenotypic characters, even when the products of those genes are unknown. The availability of a well-populated molecular genetic map of rice based on restriction fragment length polymorphism (RFLP) allows genetic evaluation of quantitative characters associated with plant tolerance for a wide range of stress conditions (McCouch and Tanksley 1991). The study reported here aims to localize quantitative trait loci (QTL) governing BNF, N uptake, and NUE in a segregating population of rice. It is a first attempt to bring the tools of molecular genetics to bear on the complex issue of nutrition in the irrigated rice system. We hope that such an approach will prove useful for soil microbiologists, plant nutrition physiologists, and plant breeders alike. Studies in tomato have already demonstrated its effectiveness (Paterson et al, 1988, Martin et al 1989), and identification of QTLs underlying tolerance for low P stress in maize using RFLP was recently reported (Reiter et al 1991). In this paper, we review the state of current knowledge on the screening methods available for measuring associative BNF, N uptake, and NUE. We also report preliminary results of a study to evaluate genotypic variation in associative BNF, N uptake, and NUE in rice to provide a basis for tagging the genes underlying the traits.

Review of screening methods for rice genotypic variation in associative N2 fixation, N uptake, and NUE Before initiating a gene tagging experiment based on RFLPs, it is necessary to develop simple and reliable screening techniques for large-scale screening and to select appropriate plant material. Diverse methods have shown significant differences in BNF, N uptake, and NUE among rice genotypes. The methods for measuring associative N2 fixation include N balance, acetylene reduction (AR) assay, and 15N isotope techniques ( 15N dilution). N balance method N balance is the terminal N content of the soil minus the initial soil N content following a given treatment. An N balance sheet gives the sum of N gains and losses. An experiment by Ventura and Watanabe (1983) with 15N-labeled soil showed that gains in N balance were largely due to differences in N 2 fixation. In an N balance study with 76 rice genotypes and 8 wild rices grown in pots, App et al (1986) found wide variation in ability to stimulate N gain (Fig. 1). Palawan, a traditional variety, ranked low; IR42, an improved variety, ranked high among the long-duration varieties tested. Extrapolated to a per hectare basis, N gain differences ranged from 16 to 70 kg N/ha per crop. High positive correlation coefficients were obtained between N gain and total N uptake, N gain and total dry matter production, 44

Wu et al

1. Differences in N gains of several rice genotypes (App et a1 1986).

and daily dry matter production. However, because N balance experiments are timeconsuming, they are not suitable for large-scale screening. Acetylene reduction assay method Due to its simplicity, sensitivity, and low cost, the AR assay has been the method most frequently used in plant BNF studies. Several studies using AR assay have demonstrated genotypic variation in associative N2 fixation in rice (Lee et al 1977; Sano et a1 1981; Barraquio et al 1986; Tirol-Padre et a1 1988; Ladha et a1 1986, 1987). To measure a large number of field-grown plants, a short-term laboratory AR assay involving a cut plant-soil system incubated in the dark has been developed (Barraquio et al 1986). Using this assay, it was found that the acetylene-reducing activity (ARA) per plant of several rice genotypes increased with rice development and was maximal at heading (Ladha et al 1986). The coefficient of variation among replicated plants decreased toward heading (Ladha et al 1986, 1987; Tirol-Padre et a1 1988). Acetylene-reducing activity associated with rice genotypes and the relationship between ARA and N harvest index (NHI = N content of grain/total N absorbed), ARA and harvest index (HI), and ARA and N remobilization efficiency (NRE = maximum vegetative N - final vegetative N/maximum vegetative N) were studied. The experiments with 16 rice varieties or lines planted in the wet (WS) and dry seasons (DS) demonstrated that: significant differences in ARA/plant and ARA/g dry weight exist between longand short-duration varieties; long-duration varieties had higher ARA than short-duration ones except IR50 which exhibited the highest ARA among short-duration varieties, IR42 exhibited the highest ARA expressed on a per plant or per g dry weight basis, line IR2191256-3-1-2-2 ranked low in the two seasons; and

Use of molecular markers to evaluate rice genetic variation

45

NHI was significantly different between long- and short-duration varieties and NRE differed among the short-duration varieties; no correlations between ARA per plant and HI, NHI, and NRE were found. Trials with 31 genotypes with different growth durations were conducted during both WS and DS. The ability of a genotype to support ARA per plant or per unit plant dry weight was shown more clearly by plotting ARA/plant vs ARA/g plant dry weight at heading (Fig. 2) (Ladha, unpubl.). Genotypes with higher ARA were on the upper side of the graph. Long-duration IR42 consistently showed the highest ARA. Root, shoot, total plant dry weight, and N uptake positively correlated with ARA/ plant at heading (Table 1). The correlations were significant but lower between ARA and N uptake at maturity. Acetylene-reducing activity, however, was not correlated with grain yield or grain yield per unit of N uptake. The ARA values in the different seasons fluctuated considerably and the wide range of varietal growth duration made comparison difficult. To circumvent these problems, relative ARA, based on those of genotypes exhibiting the highest ARA, was determined from the relation

2. Differences in ARA/plant and ARA/g total plant dry weight of 37 short-, medium-, and longduration genotypes (Ladha et al 1988).

46

Wu et al

Table 1. Correlations of plant traits with plant-associated ARA and N uptake at maturity in 37 rice genotypes of different growth durations (Ladha et ai 1988). a Plant traits vs ARA Root + submerged portion dry weight (heading) Shoot dry weight (heading) N uptake (heading) Total dry weight (maturity) N uptake (maturity) Grain yield Grain N Grain yield/N a

r ARA 0.842** 0.735** 0.735** 0.617** 0.425** 0.266 0.385* 0.108

N uptake at maturity 0.425** 0.294 0.413* 0.496** 0.871** 0.707** 0.822** 0.350*

*, ** = significant at the 5 and 1% level, respectively. ARA = acetylene reduction assay.

Although ARA has many advantages, a large plant-to-plant variation and the need to assay the plant several times during the growth cycle limited its wide use. 15 N

dilution method The 15N dilution method has recently gained acceptance among researchers because it is relatively simple and provides integrated values over the whole growth duration (Chalk 1985). The importance of using the right nonfix reference plants to obtain reliable N 2 fixation estimates and the difficulties in finding suitable ones have been extensively reviewed (Chalk et a1 1983, Hauck and Weaver 1986, Witty 1983, Danso 1986). However, the need for a perfect match of nonfixing and fixing plants becomes less important if the soil organic N is labeled with 15N and, on mineralization, a constant l5N- 14 N ratio in plant available N is produced (Pareek et a1 1990, Rarivoson and Ladha 1992). Labeling and maintaining the stability of 15 N are easier in flooded soils than in upland soils. This is because the plow sole delineates an Ap horizon where most roots are located and which is relatively easy to homogenize. The heterogeneity of added or natural 15N abundance in field soil and the differences in rooting pattern between genotypes can be major sources of error in this kind of research. Therefore, six rice genotypes of different growth durations were grown in well-mixed soil in pots and their 15N contents in grain and straw were measured (Watanabe et a1 1987). Differences in 15N values were small but significant. The difference between the maximum and minimum 15N values in the grains was 22% of the mean value (Table 2).

Selection criteria and ranking system Broadbent et al (1987) considered several parameters in screening rice varietal differences in NUE, including dry matter production (DM), total N uptake (Nt), uptake of soil N (Ns), uptake of fertilizer N (Nf), grain yield (Y), weight of panicle (WP), WP/ Use of molecular markers to evaluate rice genetic variation

47

Table 2. d15 N values of grain and straw from plants grown in pots (Watanabe et al 1987). Variety

d 15N

Growth duration (d)

lR18349-135-2-3-2-1 Hua-chou-chi-mo-mor IR50 IR42 OS4 Rodjolele S. D.

Grain

Straw

(8)a

122 101 101 150 122 150

5.2 6.0 (6) 6.0 (8) 6.1 (7) 6.5 (7) 6.5 (7) 0.38

5.6 (8) 5.8 (6) 5.5 (8) 5.5 (7) 6.2 (6) 5.9 (7) 0.50

a Values in parentheses indicate number of replications.

Table 3. Group of parameters for screening rice ARA and NUE (Ladha et al 1988). Parametera

Group 1 2 3

ARA TDW-M GY/N

RDW-H GY

Stage SDW-H GN-M

N-H

Heading Maturity Maturity

a RDW = root dry weight, SDW = straw dry weight, TDW = total dry weight, GY = grain yield, GN = grain nitrogen,

H = heading, M = maturity.

Nf, WP/Ns, Y/Nt, DM/Ns, WP/Nt, and DM/Nt. Significant variations in several parameters among 24 genotypes were found. Parameters WP/Nt and Y/Nt have been suggested as the most useful for assessing NUE. Three groups of variables were used to rank 21 genotypes with different growth durations during the 1986 and 1987 DS (Table 3). Rankings based on group 1 variables did not correlate significantly with rankings based on variables of groups 2 and 3. Rankings based on groups 2 and 3 significantly and positively correlated, suggesting that a genotype with high N2 fixation may not always be high in N uptake and efficient utilization, but that a genotype with higher N uptake may have higher utilization efficiency.

Preliminary evaluation for RFLP mapping of associative N2 fixation, N uptake, and NUE to select among genotypes for experiments Eleven genotypes were selected from previous field screening trials based on their divergent BNF, N uptake, and NUE (the yield/Nt absorbed) characteristics (Watanabe et al 1978; App et al 1986; Ladha et al 1987, 1988). These genotypes were subjected to preliminary RFLP analysis for evaluating the frequency of polymorphism detected among them. Four enzymes were used: EcoRV, HindIII, ScaI, and Xba I. Forty-six mapped DNA from both a rice genomic library (RG) and an oat cDNA library (CDO) were selected based on their distribution throughout the 12 chromosomes of rice. The results showed that the combination of Cigalon and IR42 had the highest percentage of polymorphism (78%), followed by IR42 and OS4 (73%), Cigalon and IR13429-15048

Wu et al

Table 4. Percent polymorphism of possible combinations differing in associative N 2 -flxing ability

and NUE (46 probes surveyed using 4 enzymes).a Variety Cigalon OS4 ClCA C66-2803 Palawan DR33 BR40-300-2-1-E

BR51-46-3-2-1-2

lR13429-150-3-2-1-2

BG367-4

IR42

67 56 58 52 67 54 50

69 58 60 41 43 65 54

67 58 52 43 58 54 41

78 73 45 45 67 45 37

aEnzymes used: Eco RV, Hind lll, Sca l, Xba l.

3-2-1-2-(69%), IR42 and Palawan (67%), BG367-4 and Cigalon (67%), and BR5146-CI and Cigalon (67%) (Table 4). Considering the good level of divergence measuring in phenotypic values and the percentage of polymorphism detected, IR42 and Palawan were selected for further study. They are both long-duration varieties (125-130 d). IR42 consistently demonstrated superior performance. BG367-4 and IR13427-40-2-3-3-3-3 (100 d) were also selected as test materials. BG367-4 exhibited higher N uptake and higher NUE than IR13427-40-2-3-3-3-3 (Broadbent et al 1987, Ladha et al 1988). Development of mapping population Two sets of crosses were made: Palawan (japonica) and IR42 (indica), and IR1342740-2-3-3-3-3 (indica) and BG367-4 (indica). Following the evaluations of BNF, N uptake, and NUE in the first cropping experiments, only F1 plants from Palawan and IR42 crosses were selected to produce an F2 population of 200 plants. Pot and field experiments The soil used in the pot and field experiments was labeled with 15N-labeled fertilizer as (NH4)2SO4 with 75% atom excess at the rate of 1kg 15 N/ha in February 1990. Open pot and modified Leonard's assembly designs were employed in a pot experiment in the greenhouse in the 1991 WS. Twelve replications for each genotype were planted with 1.5 kg soil/pot and one plant/pot. For the Leonard's assembly experiment, water in the jar was changed every 3 d to keep it fresh. The field experiment was conducted simultaneously with the pot experiments in microplots (7 m x 8 m) with four replications in a randomized block design. Forty kilograms P/ha as K2 HPO4 was incorporated into the soil before transplanting 15-d-old seedlings. Rice-associated BNF was measured by the 15 N dilution method. Measurement of BNF, N uptake, NHI, NUE, leaf chlorophyll, plant biomass, and grain yield. The percentage of N2 derived from the air (% Ndfa), Nt, NHI, NUE, chlorophyll content of flag leaf at heading stage, plant biomass, and grain yield were measured in first cropping in the 1991 WS. In the pot experiments, the plants were sampled at maturity. In the field experiment, plants were sampled at maximum tillering, heading, and maturity.

Use of molecular markers to evaluate rice genetic variation

49

Isotope 15N abundance in the samples was measured by a mass spectrometer (2HT03, Beijing, China). Total N uptakes of the plant shoot, root, and yield were measured by a Perking-Elmer 2400 CHN elemental analyzer (Norwalk, Connecticut, USA, 1988). Chlorophyll content of the flag leaf was measured by a chlorophyll meter SAPD-502 (Minolta, Japan). The chlorophyll content was calculated based on readings from the SAPD-502 with 12 replications. (Formula: Y = 0.0996X - 0.152, where Y = chlorophyll content (mg/100 cm2), X = SAPD-502 reading). Yield and dry biomass of the shoot and the root were measured by weighing the sample after it was dried at 75 °C for 3 d. Atom % 15N excess of selected parents. The ANOVA and Duncan's Multiple Range Test of atom % 15N excess of plants grown in pots showed significant genotypic variation between both pairs of parental genotypes. In the field experiment, significant variations were found at maximum tillering and heading stages, but at maturity, only IR42 and Palawan showed significant differences. Considerable differences in % Ndfa between IR42 and Palawan were found in the field and pot experiments. IR42 exhibited higher ability to stimulate associative N, fixation. Ndfa values for IR42 as compared with those for Palawan were 22% higher in the field experiments and 19% higher in the pot experiments. Yield, Nt, NHI, and NUE of selected parents. IR42 and Palawan showed significant differences in all four parameters tested (yield, Nt, NHI, and NUE) in both field and pot experiments. Between BG367-4 and IR13427-40-2-3-3-3-3, only Nt and NUE differed significantly. The correlations between yield and Nt, yield and NUE, and Nt and NUE are shown in Table 5. The results indicate that a close correlation exists between yield and Nt and yield and NUE, but no clear correlation exists between Nt and NUE. This indicates that Nt and NUE may be used as independent parameters for screening. Table 5. Correlation among parameters, 1991 WS. Experiment a Field Pot an

Correlation coefficient Yield/Nt

Yield/NUE

0.854 0.909

0.780 0.708

Nt/NUE 0.423 0.424

= 8 in the field experiment; n = 12 in the pot experiment.

Table 6. Correlation between chlorophyll content of flag leaf at heading stage and atm % 15N excess of straw samples. Experiment a

Correlation equationb

r

Field Open pot Leonard's pot

Y = –0.017X + 0.54 Y = – 0.009X + 0.42 Y = –0.011X + 0.42

-0.75 -0.65 -0.81

an = 8 for field experiment, n = 12 for open and Leonard's pot experiments. b Y = atm % 15 N excess of straw samples. X = chlorophyll content of flag leaf (mg/100 cm 2 ).

50

Wu et al

Chlorophyll content of flag leaf. A close relationship between associative N2 fixation activity and photosynthesis has previously been established for wheat (M.I. Chumakov, Russian Academy of Sciences, pers. commun.). When the chlorophyll content of the flag leaf was measured in our experiments, significant variation existed among varieties, with similar results found using an open pot and a modified Leonard’s system. Correlations between atom % 15 N excess and chlorophyll content were found to be negative (Table 6), suggesting that chlorophyll content and associative N2 fixation should be positively correlated. Thus, high chlorophyll content appears to be an indication of the ability of the rice genotype to stimulate associative N2 fixation.

RFLP survey of selected genotypes Six enzymes (Eco RI, Eco RV, Sca I, Hind III, Xba I, and Dra I) were used for identifying polymorphic markers between test genotypes. The 120 probes (RG, CDO, and RZ) showed 72% polymorphism between IR42 and Palawan and 62% between BG367-4 and IR13427-40-2-3-3-3-3. The degree of polymorphism detected by each of the six enzymes is listed in Table 7. The results indicate that Dra I, Xba I, and Hind 11 are more effective with respect to detecting polymorphisms between IR42 and Palawan than are Sca I and Eco RV. Eco RI gave the lowest level of polymorphism.

Conclusions and perspectives Palawan and IR42 consistently demonstrated superiority for all parameters measured both in the pot and in the field experiments, with IR42 exhibiting a better performance than Palawan. Accordingly, the cross of Palawan and IR42 will be the focus of future gene tagging. IR42 demonstrated about 20% higher Ndfa compared with Palawan. This difference is important for low-input and sustainable rice production systems needing supplementary sources of N. The second set of experiments must confirm these findings, however, before any conclusions are drawn. The relationship between atom % 15N excess and the chlorophyll content of the flag leaf indicates that the N2 fixed by associative fixation may be available as an N source in rice systems and play an important role in photosynthesis. NUE and Nt can be considered independent parameters and the correlation between them and yield may have profound economic implications. Table 7. Percent polymorphism of six enzymes between IR42 and Palawan. Enzyme

Percent polymorphism (total of 180 probes)

Eco Rl Eco RV Sca l Hind III Xba l Dra l

11 25 26 31 33 33

Use of molecular markers to evaluate rice genetic variation

51

The Palawan/IR42 cross showed good fertility in the F1. The F2 population is growing well and demonstrates good segregation for leaf color, height, and tiller number, suggesting ample recombination of characters. Further study will evaluate associative N 2 fixation, N uptake, and NUE using both the F2 and F7 populations, with the goal of identifying the main genetic loci governing these characters. The F7 recombinant inbred (RI) lines will be derived by single seed descent and will facilitate replication of phenotypic evaluation on identical genotypes. This RI population will provide reliable data for the genetic analysis of physiological and biological parameters related to the complex N nutritional processes in rice.

References cited App A A, Watanabe I, Ventura T S, Bravo M, Jurey C D (1986) The effect of cultivated, wild rice varieties on the nitrogen balance of flooded soil. Soil Sci. 141:448-452. Asanuma S, Tanaka H, Yatazawa M (1979) Rhizosplane microorganisms of rice seedling as examined by scanning electron microscopy. Soil Sci. Plant Nutr. 25:539-551. Balandreau J, Knowles R (1978) The rhizosphere. Pages 234-268 in Interaction between nonpathogenic soil microorganisms and plants. Y. Dommergues and S.V. Krupa, eds. Elsevier, Amsterdam. Barraquio W L, Daroy M L G, Tirol A C, Lahda J K, Watanabe I (1986) Laboratory acetylene reduction assay for relative measurement of N2-fixing activities associated with field-grown wetland rice plants. Plant Soil 90:359-372. Broadbent F E, De Datta S K, Laureles E V (1987) Measurement of nitrogen use efficiency in rice genotypes. Agron. J. 79:786-791. Chalk P (1985) Estimation of N2 fixation by isotope dilution: an appraisal of techniques involving 15N enrichment and their application (review). Soil Biol. Biochem. 17:389-410. Chalk P M, Douglas LA, Buchanan S A (1983) Use of 15N enrichment of soil mineralizable fixed nitrogen. Can. J. Microbiol. 29:1046-1052. Danso S K A (1986) Review: estimation of N2-fixation by isotope dilution. An appraisal of techniques involving 15N enrichment and their application (comments). Soil Biol. Biochem. 18:243-244. De Datta S K, Broadbent F E (1988) Methodology for evaluating nitrogen use efficiency by rice genotypes. Agron. J. 80:793-798. Diem G, Rougier M, Hamad-Fares I, Balandreau J P, Dommergues Y R (1978) Colonization of rice roots by diazotroph bacteria. Environmental role of nitrogen-fixing blue-green algae and asymbiotic bacteria. U. Granhall. ed. Ecol. Bull. (Stockholm) 26:305-311. Hauck R D, Weaver R D (1986) Field measurement of nitrogen fixation and denitrification. Soil Science Society of America, Inc., Madison, Wisconsin, USA. 115 p. Iyama S, Sano Y, Fujii T (1983) Diallel analysis of nitrogen fixation in the rhizosphere of rice. Plant Sci. Lett. 30:427-135. Ladha J K, Padre A T, Punzalan G C, Watanabe I, De Datta S K (1988) Ability of wetland rice to stimulate biological nitrogen fixation, utilize soil nitrogen. Nitrogen fixation: a hundred years after Gustav Fischer. Bothe Bruijin/Newton, eds. Stuttgart, New York. Ladha J K, Tirol-Padre A, Daroy M G, Punzadan G, Ventura W, Watanabe I (1986) Plantassociated N2 fixation (C2H2-reduction) by five rice varieties, and relationship with plant growth characters as affected by straw incorporation. Soil Sci. Plant Nutr. 32:91-106.

52

Wu et al

Ladha J K, Tirol-Padre A, Punzalan G C, Watanabe I (1987) Nitrogen-fixing (C 2H2 -reducing) activity and plant growth characters of 16 wetland rice varieties. Soil Sci. Plant Nutr. 33(2):187-200. Lee K K, Castro T, Yoshida T (1977) Nitrogen fixation throughout growth and varietal differences in nitrogen fixation by the rhizosphere of rice planted in pots. Plant Soil 48:613619. Martin B, Nienhuis J, King G, Schaefer A (1989) Restriction fragment length polymorphisms associated with water use efficiency in tomato. Science 243:1725-1728. McCouch S R, Tanksley S D (1991) Development and use of restriction fragment length polymorphism in rice breeding and genetics. Pages 109-125 in Rice biotechnology. G.S. Khush and G.H. Toenniessen, eds. CAB International, Wallingford, UK. Murty M G, Ladha J K (1987) Differential colonization of Azospirillum lipoferum on roots of two varieties of rice ( Oryza sativa L.). Biol. Fert. Soils 4:3-7. Nishizawa N, Yoshida T, Arima Y (1983) Electron microscopic study of associative N2 -fixing bacteria in roots of rice seedling. Soil Sci. Plant Nutr. 29:261-270. Pareek R P, Ladha J K, Watanabe I (1990) Estimation of N2 fixation by Sesbania rostrata and S. cannabina in lowland rice soil by 15N dilution method. Biol. Fert. Soils 10:77-88. Paterson A H, Lander E S, Hewitt J D, Peterson S, Lincoln S E, Tanksley S D (1988) Resolution of quantitative traits into Mendelian factors by using a complete RFLP linkage map. Nature 335:721-726. Rarivoson C, Ladha J K (1992) Estimation of N2 fixed by stem and root inoculated Sesbania rostrata using 15N enrichment of soil ammonium-N as a reference. Biol. Fert. Soils. (in press) Reiter R S, Coors J G, Sussman M R, Gabelman W H ( 1991) Genetic analysis of tolerance to lowphosphorus stress in maize using restriction fragment length polymorphisms. Theor. Appl. Genet. 82:561-568. Sano Y, Fujii T, Iyama S, Hirota Y, Komagata K (1981) Nitrogen fixation in the rhizosphere of cultivated and wild rice strains. Crop. Sci. 21:785-761. Tirol-Padre A, Ladha J K, Gloria C P, Watanabe I (1988) A plant sampling procedure for acetylene reduction assay to detect rice varietal differences in ability to stimulate N2 fixation. Soil Biol. Biochem. 20 (2):175-183. Ventura W, Watanabe I (1983) 15 N dilution technique of assessing the contribution of nitrogen fixation to rice plant. Soil Sci. Plant Nutr. 29(2):123-131. Watanabe I, Kee, Alimano B V (1978) Seasonal change of N2-fixing rate in rice field assayed by in situ acetylene reduction technique. Soil Sci. Plant Nutr. 24:1-13. Watanabe I, Yoneyama T, Padre B, Ladha J K (1987) Difference in natural abundance of 15N in several rice ( Oryza Sativa L.) varieties:application for evaluating N2 fixation. Soil Sci. Plant Nutr. 33(3):407-415. Witty J F (1983) Estimating N2 fixation in the field using 15 N-labeled fertilizer: some problems and solutions. Soil Biol. Biochem. 15:631-639.

Use of molecular markers to evaluate rice genetic variation

53

Notes Authors’ addresses: P. Wu and J.K. Ladha, Soil and Water Sciences Division; S.R. McCouch, and S.B. Teng, Plant Breeding, Genetics, and Biochemistry Division, International Rice Research Institute, P.O. Box 933, Manila, Philippines. Acknowledgments: It is a pleasure to acknowledge the financial support of the United Nations Development Programme, Division of Global and International Projects. We thank Ms. Agnes T. Padre and Genalin A. van Coppenolle for their excellent assistance. Citation information: Kirk G J D, ed. (1994) Rice roots: nutrient and water use. International Rice Research Institute, P.O. Box 933, Manila, Philippines.

54

Wu et al

Rainfed lowland rice roots: soil and hydrological effects Pradeep K. Sharma, G. Pantuwan, K.T. Ingram, and S.K. De Datta

Rice root systems differ greatly among cultivars, soil conditions, and hydraulic regimes under rainfed lowland conditions. Unsaturated, light-textured soils should be used for evaluating varietal differences in root systems for drought tolerance. Future research to increase drought resistance through improved root systems must combine germplasm improvement and soil physical management.

Rainfed rice environments are diverse and unpredictable and insufficient or excessive water is often the major factor limiting productivity. Because root characteristics are closely associated with drought resistance, they are particularly important under rainfed conditions. Root characteristics are basically genetically controlled but they are also strongly affected by soil conditions and crop management practices (Ghildyal and Tomar 1982, Cruz et a1 1986, Sharma et al 1987, Mambani et al 1990, Thangaraj et a1 1990). The occurrence of drought or flood depends on the frequency and amount of rainfall, daily evapotranspiration, and soil physical properties governing infiltration, water retention capacity, and drainage. These factors in turn affect rice root growth. Soil nutrient deficiencies and toxicities also affect root growth. In this paper, we focus on the effects of soil texture, hydrology, and mechanical impedance on rice root growth.

The rice root system The characteristics of the rice root system are described by Yoshida (1981). Root growth, in terms of length, dry weight or volume, generally follows a sigmoidal curve (Fig. 1). Maximum root growth is observed near flowering. Rooting depth depends on the cultivar and the nature of the rooting medium. Roots of upland rice cultivars in light-

1. Root growth of rice (IR54) with time in a puddled and flooded clay soil (Thangaraj et al 1990). Table 1. Depthwise root distribution of some lowland cultivars in puddled soils.

Soil texture

Cultivar

Clay loam Sandy loam Clay Clay Clay loam Sandy loam Clay

IR36 IR36 IR20 IR20 IR20 IR20 IR54

Root length (% of total) 0-10 cm 75 72 89 92 93 88 75

10-20cm 16 16 8 3 4 7

20-30cm

30-40cm

6 8 2 1 3 4

3 4 1 4 1 -

Source Sharma et al (1987) Mambani et al (1989)

Thangaraj et al (1990)

textured upland soils may grow as deep as 1 m or more. But roots of rice cultivars in lowland soils rarely grow deeper than 40 cm; about 90% of the total root system is restricted to the top 20 cm of the soil layer (Table 1).

Hydrological effects We classify ricefield hydrology into three groups depending on topographic location and water supply: i) pluvial, also called upper paddy, where rainfall is the only source of water and surface water accumulation is minimal; ii) phreatic, also called middle paddy, where rainfall, seepage, water table, and runoff provide water to the root zone with moderate surface water accumulation; and iii) fluxial, also called the lower paddy, where rainfall, seepage, water table, and runoff provide water to the root zone with reliable surface water accumulation. Most rainfed lowland rice areas fall in the fluxial and phreatic toposequence positions, and only some are in the pluvial category (O’Toole and Chang 1978, Greenland and Bhuiyan 1982). Drought is the major 56

Sharma et al

Table 2. Influence of soil moisture condition on oxygen diffusion rate (ODR) of soil, and root porosity, root length, and root mass of IR8 (Pradhan et al 1973). ODR (10-8 g /cm2 per min)

Root porosity (%)

Root length (cm)

Root mass (g)

Flooding, 6±1 cm Flooding, 3±1 cm Saturation 2 kPa 6 kPa 35 kPa 50 kPa 100 kPa

5.14 6.32 10.17 22.36 37.28 61.33 75.59 83.07

22.78 21.06 18.04 14.11 10.23 5.94 5.27 4.65

25.37 26.62 30.95 30.17 37.75 43.32 45.10 44.02

7.25 8.12 7.25 6.25 5.62 4.87 3.50 2.25

SEM ± LSD (0.05) LSD (0.01)

4.71 9.97 13.98

0.22 0.65 0.89

0.49 1.93 2.65

0.42 1.22 1.66

Soil moisture treatment

constraint to rice productivity in the pluvial and phreatic toposequence positions; flooding is the main constraint in the fluxial toposequence position. The rice root system is greatly. influenced by water regime. Under submerged, anaerobic conditions, rice roots are generally short and thick with high porosity as a result of aerenchyma formation (Pradhan et al 1973, Kar et al 1974, Maurya and Ghildyal 1975, Ogunremi 1991). Das and Jat (1977) observed an increase in total root length when the moisture regime changed from unsaturated to saturated. Effects of moisture regime on some root growth parameters are shown in Table 2. Root porosity is inversely associated with soil oxygen content. With increasing soil moisture, the flux of oxygen through the soil to the roots decreases. In response to this, cells in the root cortex degrade, forming aerenchyma through which oxygen is transported to respiring tissues. As a consequence, root porosity and soil aeration near the roots are increased. In a submerged soil, the oxygen needs of rice roots are largely met through oxygen transport via the aerenchyma. We transplanted three rice cultivars (KDML 105, RD9, and IR46) recommended for cultivation in rainfed lowland areas at three locations representing two toposequence positions: high (at the border between pluvial and phreatic) and low (fluxial, both with loamy sand soil texture) and two soil textures: loamy sand and clay, both in the low toposequence position. The hydraulic regimes of the three locations are shown in Figure 2. Roots were sampled at anthesis with 20- x 20- x 50-cm metal monolith samplers, centered over a hill. Samples were divided into soil layers, soil was washed from the roots, and roots were dried and weighed to measure root mass density. A strong interaction existed between the effects of rice cultivar and hydraulic regime on root mass density (RMD). KDML 105 had greater RMD in the 0- 10 cm soil layer at the high toposequence position (low-moisture regime) than it had at the low toposequence position (high-moisture regime). RD9 and IR46 had greater RMD in the 10-30 cm soil layer at the low toposequence than they had at the high toposequence position (Fig. 3). In other soil layers, RMD was not significantly different between the two toposequence positions for any cultivar. Among cultivars, however, at the high Rainfed lowland rice roots: soil and hydrological effects

57

2. Hydraulic regimes at three experimental sites.

toposequence position, KDML 105 produced the greatest RMD in the 0-30 cm soil layer, but at the low toposequence position, RD9 had the greatest RMD in the 10-30 cm soil layer (Fig. 4). The differences in RMD among cultivars in other soil layers were not significant. RD9 and IR46 produced relatively more root mass in the 10-40 cm soil layer under the high-moisture regime; no such differences were detected for KDML 105 in either toposequence position (Table 3). In general, root mass decreased exponentially with soil depth (Fig. 5). Maximum depth of root penetration was similar for all cultivars at both toposequence positions.

Soil textural effects Soil texture affects the root system partly through its influence on pore size distribution. Larger pore radii in sandy soils and greater mobility of particles under the pressure of growing roots as compared with clayey soils favor rice root growth in coarse-textured soils (Kar et al 1979). Kar et al concluded that high content of silt and sand with a moderate clay content (20-35%) provided the most favorable physical environment for rice root growth. In a field experiment at a low toposequence position, the effects of soil texture on root growth differed among cultivars. In the 0- 10 cm soil layer, KDML 105 had greater RMD in the loamy sand than in the clay soil. Similarly, IR46 had greater RMD in the 10-20 cm layer in the loamy sand than in the clay soil. On the other hand, RD9 had the greater RMD in the 10-30 cm layer of the clay than in that of the loamy sand soil (Fig. 6). Furthermore, in the clay soil, RMD did not differ significantly among cultivars in any soil layer, but in loamy sand, RD9 had greater RMD than the other two cultivars 58

Sharma et al

3. Effect of moisture regime on root mass density of three rice cultivars at two toposequence positions in loamy sand soil. Table 3. Hydrological effects on percent distribution of root mass of three rice cultivars in loamy sand soil. Soil depth (cm) 0-10 10-20 20-30 30-40 40-50

Root mass (% of total) KDML 105

RD9

Higha

Low b

77.0 20.1 2.0 0.5 0.4

77.4 19.4 2.8 0.3 0.1

High 84.0 15.0 0.7 0.2 0.1

IR46 Low 66.7 26.4 6.3 0.5 0.1

High 84.1 13.8 1.6 0.3 0.2

Low 79.4 16.3 3.8 0.4 0.1

a High in toposequence (low-moisture regime). b Low in toposequence (high-moisture regime).

Rainfed lowland rice roots: soil and hydrological effects

59

4. Varietal differences in root mass density of rice at two toposequence positions in loamy sand soil.

5. Depthwise distribution of root mass of three rice cultivars in loamy sand soil at two toposequence positions. 60

Sharma et al

in the 10-30 cm layer (Fig. 7). Therefore, soil texture must be considered when evaluating differences in root systems among rice cultivars.

Soil mechanical impedance effects Root growth is closely associated with soil mechanical impedance in the root zone. To elongate, roots must exert pressure greater than the soil mechanical impedance. Soil mechanical impedance depends greatly on bulk density and soil moisture content. Impedance increases with bulk density and decreases with moisture content. Drought, therefore, would increase the mechanical impedance of soil by decreasing soil moisture content and increasing bulkdensity, especially in swelling and cracking soils. Thangaraj et al (1990) studied the effect of mechanical impedance on the root growth of IR54 under moisture deficit in a Maahas clay soil. Rice seedlings were grown

6. Effect of soil texture on root mass density of three rice cultivars grown at low toposequence positions. Rainfed lowland rice roots: soil and hydrological effects

61

7. Varietal effects on root mass density of three rice cultivars grown at low toposequence positions in loamy sand and clay soils.

uniformly under flooded conditions until 42 days after transplanting (DT). From 42 to 61 DT, different moisture regimes were created by using a line-source sprinkler irrigation system. The crop was reflooded uniformly after that. Soil mechanical impedance increased from 0 to 1.0 MPa as gravimetric moisture content declined from 160 to 65%. Root length density decreased as a power function of soil mechanical impedance (Fig. 8). Mechanical impedance as little as 0.05 MPa reduced root length density by 60% as compared with the continuously flooded control. Mechanical impedance greater than 0.3-0.5 MPa decreased root growth and root extension by 75%. Although root mass and root length densities responded similarly to mechanical impedance, effects on root length were of greater magnitude than those on root mass density. Thus, roots became shorter and thicker in response to increased mechanical impedance. The effects of bulk density and mechanical impedance on root length density of IR36 in a clay soil under flooded conditions were studied (Sharma and De Datta 1986). Soil bulk density and mechanical impedance were modified by applying different tillage treatments. Root length density decreased as a power function of bulk density or mechanical impedance (Fig. 9). Similar observations were made by Mambani et al (1989) for IR20 in soils of different texture using different tillage practices. They concluded that mechanical impedance of 0.57 MPa was the critical limit for root growth of rainfed lowland rice. 62

Sharma et al

8. Relation between root length density of IR54 relative to control and soil mechanical impedance in the drying puddled lowland soil at 0-10 cm soil depth (Thangaraj et a1 1990). Table 4. Effect of interaction between submerged soil temperature regime and bulk density of a lateritic sandy loam soil on root growth of rice (Kar et al 1976). Bulk density (Mg/m3)

Soil temperature regime (°C) 27-15

32-20

37-25

LSD 42-30

(0.05)

(0.01)

1.22 1.25 1.41 1.49

0.25

0.33

Root dry weight (g) 1.5 1.6 1.7 1.8

1.46 1.53 1.60 1.63

3.18 3.05 2.38 1.80

3.44 3.62 2.46 1.92

In flooded soils, adverse effects of increased mechanical impedance on root growth may be compounded by anoxia (Wiersum 1957). Roots lose solutes under anaerobiosis. Reduced osmotic potential and turgor may reduce the ability of the roots to overcome soil resistance. Furthermore, a strong interaction exists between mechanical impedance and soil temperature for their effects on rice root growth. Kar et a1 (1976) found that root growth decreased drastically when soil temperature exceeded 37 °C and bulk density was greater than 1.6 Mg/m 3 (Table 4). Rice roots degenerate in submerged conditions; higher soil temperatures favor this process. Rainfed lowland rice roots: soil and hydrological effects

63

9. Relationship between bulk density and soil strength at transplanting and root length density of rice (IR36) at harvest in the 0-10 cm layer of a flooded clay soil (Sharma and De Datta 1988).

According to a survey of 35 rice-growing locations in South and Southeast Asia, mechanical impedance in the surface 10 cm soil layer averaged 0.64 MPa under flooded conditions and 1.7 MPa in puddled fields without standing water. Lower soil horizons had impedance >2.8 MPa, high enough to inhibit root elongation (Hasegawa et al 1985). Therefore, research on the interactive effects of bulk density, soil mechanical impedance, soil temperature, and hydrology on rice root growth have great implications in boosting rice production in rainfed areas of South and Southeast Asia.

Conclusions Growing rice cultivars having extensive deep root systems that are capable of extracting water from deep soil layers is a means of achieving more efficient use of rainwater under drought-prone rainfed situations. The rice root system differs strongly between cultivars, soil conditions, and hydraulic regimes. Hence, it is difficult to make generalizations about the effects of individual variables. We found that differences between cultivars in root growth generally decreases with a change from unsaturated to saturated conditions and with texture from loamy sand to clay. Unsaturated, lighttextured soils rather than saturated clay soils should be used for evaluating differences in root systems among rice cultivars. Effects of mechanical impedance on root growth, however, appeared to be similar under saturated and unsaturated conditions; root length and root mass densities decreased as a power function of soil mechanical 64

Sharma et al

impedance for different rice cultivars. KDML 105 appeared relatively more responsive to hydrology and soil texture changes than RD9 and IR46. Thus, KDML 105 may be more widely adapted to rainfed lowland situations than the other cultivars. Future research should combine germplasm improvement and soil physical management to increase drought resistance through an improved rice root system.

References cited Cruz R T, O’Toole J C, Dingkuhn M, Yambao E B, Thangaraj M, De Datta S K (1986) Shoot and root responses to water deficits in rainfed lowland rice. Aust. J. Plant Physiol. 13:567-575. Das D K, Jat R L (1977) Influence of three soil water regimes on root porosity and growth of four rice varieties. Agron. J. 69:197-200. Ghildyal B P, Tomar V S (1982) Soil physical properties that affect rice root systems under drought. Pages 83-96 in Drought resistance in crops wih emphasis on rice. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Greenland D J, Bhuiyan S I (1982) Rice research strategies in selected areas: environment management and utilization. Pages 239-262 in Rice research strategies for the future. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Hasegawa S, Thangaraj M, O’Toole J C (1985) Root behavior: field and laboratory studies. Pages 383-395 in Soil physics and rice. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Kar S, Varade S B, Ghildyal B P (1979) Pore size distribution and root growth relations of rice in artificially synthesized soils. Soil Sci. 128:364-368. Kar S, Varade S B, Subramanyam T K, Ghildyal B P (1974) Nature and growth pattern of rice root system under submerged and unsaturated conditions. I1 Riso 23:173-179. Kar S, Varade S B, Subramanyam T K, Ghildyal BP (1976) Soil physical conditions affecting rice root growth: bulk density and submerged soil temperature regime effects. Agron. J. 68:23-26. Mambani B, De Datta S K, Redulla C A (1989) Land preparation requirements for rainfed rice as affected by climatic water balance and tillage properties of lowland soils. Soil Till. Res. 14:219-230. Mambani B, De Datta S K, Redulla C A (1990) Soil physical behaviour and crop responses to tillage in lowland rice soils of varying clay content. Plant Soil 126:227-235. Maurya P R, Ghildyal B P (1975) Root distribution pattern of rice varieties evaluated under upland and flooded soil conditions. I1 Riso 24:239-244. Ogunrerni L T (1991) Influence of bulk density and moisture regime of a permeable soil on the performance of a lowland rice, Oryza sativa L. Trop. Agric. 68:129-134. O’Toole J C, Chang T T (1978) Drought and rice improvement in perspective. IRRI Res. Pap. Ser. 14. Pradhan S K, Varade S B, Kar S (1973) Influence of soil water conditions on growth and root porosity of rice. Plant Soil 38:501-507. Sharma P K, De Datta S K (1986) Physical properties and processes of puddled rice soils. Adv. Soil Sci. 5:139-178. Sharma P K, De Datta S K, Redulla C A (1987) Root growth and yield response of rainfed lowland rice to planting method. Exp. Agric. 23:305-313. Thangaraj M, O’Toole J C, De Datta S K (1990) Root response to water stress in rainfed lowland rice. Exp. Agric. 26:287-296. Rainfed lowland rice roots: soil and hydrological effects

65

Wiersum L K (1957) The relationship of the size and structural rigidity of pores to their penetration by roots. Plant Soil 9:75-85. Yoshida S (1981) Fundamentals of rice crop science. International Rice Research Institute, P.O. Box 933, Manila, Philippines. 269 p.

Notes Authors’ addresses: Pradeep K. Sharma, Department of Soil Science, HPKV, Palampur 176062 H.P., India; G. Pantuwan, Ubon Rice Research Center, P.O. Box 65, Ubon Ratchathani 34000, Thailand; K.T. Ingram, Agronomy, Plant Physiology, and Agroecology Division, International Rice Research Institute, P.O. Box 933, Manila, Philippines; S.K. De Datta, College of Agriculture and Life Sciences, Virginia Polytechnic Institute and State University, Blacksburg, VA 2401-0334, USA. Citation information: Kirk G J D, ed. (1994) Rice roots: nutrient and water use. International Rice Research Institute, P.O. Box 933, Manila, Philippines.

66

Sharma et al

Rice root traits for drought resistance and their genetic variation K.T. lngram, F.D. Bueno, O.S. Namuco, E.B. Yambao, and C.A. Beyrouty

The root traits that confer drought resistance are not well-defined. Glasshouse research has shown that the root xylem vessel cross-sectional area does not explain water relation differences across genotypes. In field experiments under upland conditions, root length density (RLD) was negatively correlated with grain yield of three rice genotypes grown under different levels of water deficit stress. On the other hand, in minirhizotron experiments under rainfed lowland conditions, only RLD in the 10-20 and 20-30 cm soil layers were related to relative yields (30-d vegetative phase stress yield/continuously flooded control yield) in 12 rainfed lowland genotypes and only on 2 sampling days (86 and 89 d after sowing). Two root characteristics were common to drought-resistant rainfed lowland genotypes: 1) rapid root responses to changing soil moisture level in the 10-20 and 20-30 cm soil layers, and 2) greater absolute RLD below 20 cm.

Capacity for water uptake may limit rice productivity even in flooded soils. Thus, midday stomatal closure (a symptom of leaf water deficit) has been observed in rice growing in flooded soils (Dingkuhn et al 1990). However, water uptake and transport by rice roots are most important and have been primarily studied in nonflooded soils, especially as they affect yields under water-limited conditions, i.e., drought resistance. Drought is not a single stress. Rather, as suggested by O’Toole and De Datta (1986), it is a syndrome. The three major dimensions of the drought syndrome are the timing, duration, and intensity of water deficit within the crop life cycle. Stress effects in each of these dimensions result from a complex interaction between crop, soil, and atmospheric factors. Just as drought is a syndrome rather than a single stress, plant adaptations that maximize yields under water-limited conditions are numerous and diverse. O’Toole and Chang (1979) identified four components of drought resistance: tolerance,

avoidance, escape, and recovery. The second component, avoidance, can be subdivided by its mechanisms—those that allow the plant to maintain a high internal water status when available water is less than evaporated demand—into a) water conservation and b) water collection. Plant mechanisms for water conservation include leaf rolling and stomatal closure; which tend to reduce growth and yield. Mechanisms for water collection result in increased capacity for water uptake (e.g., through deeper rooting or more efficient water extraction from soils); which should reduce yield losses. This paper focuses on water collection mechanisms in roots. We will describe recent research on genetic variation and heritability of rice root traits as well as the interrelationships among root characters, drought resistance, and water use.

Variation and heritability of root traits Research on the genetic variation and heritability of rice root traits was recently reviewed by O’Toole and Bland (1987). The findings are summarized in Tables 1 and 2. It has been long thought that the best root system for drought resistance has deep and thick roots (Armenta-Soto et al 1983, Chang and Loresto 1986), so most studies of root traits to date have emphasized those traits directly or indirectly. Table 1. Root traits showing genetic variation in rice. Diameter/thickness Water extraction Maximum depth Deep root-shoot ratio Xylem vessel diameter

Japonica > indica

Root dry weight Thick root number

0-10 cm, indica >japonica below 30 cm, japonica > indica

Branching Number

Indica > japonica

Table 2. Narrow sense heritability (NSH) of root traits in rice. NSH (%) Maximum root length Basal diameter Acropetal diameter Root dry weight Root-shoot ratio Root number Number of thick roots Root volume Root length density Uprooting force

35-61 60-80 49-62 43-92 53 44 33-53 18-55 44-77 39-47

Sources: Armenta-Soto et al 1983, Chang et al 1982, Ekanayake et al 1985 a,b.

68

lngram et al

Geneticists have estimated narrow sense heritability from observations in F1, F2, or F3 populations growing in either soil or aeroponic media (Table 2). For most traits, inheritance was shown to be additive and polygenic. Overall, the root traits studied show strong potential for improvement through breeding, with the greatest breeding limitation being the large quantity of labor required for screening most root traits. For much of the above research, genetic variation and heritability were estimated from indica-japonica comparisons. Such a comparison ignores the fact that drought occurs over widely varying rice environments, since indica rices are grown primarily in the rainfed lowland ecosystem and japonica rices in the uplands. Thus, differences in root traits between rice ecotypes may not be of practical use in breeding for drought resistance. Focusing on the genetic variation and heritability within a single rice ecotype instead might more quickly yield progress toward drought resistance. Similarly, as we consider whether root traits can be transferred from wild Oryza species or from one rice ecotype to another, we need to recognize that a trait that confers drought resistance in one soil and rainfall environment may not be useful in another.

Root growth and development Field research with four cultivars under continuously flooded conditions showed that wet seeded rice (WSR) had about 1.5 times as much total root length (TRL) as transplanted rice (TPR) by 50 d after seeding (DAS) (Fig. 1). Binato, a traditional cultivar, had greater TRL than the other three cultivars studied, which were all improved indica types (Fig. 2). The greater TRL of WSR was accompanied by greater maximum root depth (IRRI 1991). This difference was maintained through maturity and through a ratoon crop. The peak and decline of TRL shortly after panicle initiation was probably related to increased competition for assimilates by the developing

1. Total root length of wet seeded and transplanted rice through a plant and ratoon crop cycle. Each point is the average of four cultivars. Vertical bars show significant differences by LSD at P < 0.05. PI = panicle initiation, F = flowering, H = harvest.

Rice root traits for drought resistance and their genetic variation

69

2. Total root length of four rice cultivars through a plant and ratoon crop cycle. Each point is the average of two establishment methods. Vertical bars show significant differences by LSD at P < 0.05. Table 3. Grain yields (av of 24 rainfed lowland rice genotypes) in response to vegetative water deficit under transplanted and wet seeded crop establishment. Transplanted Well-watered 3.1 a a

Wet seeded Deficit 2.2

b

Well-watered 3.1 a

Deficit 3.0 a

a Means followed by the same letter are not significantly different at P < 0.05 by LSD.

panicle. Similarly, the increase in TRL beginning about 15 d after panicle emergence was probably associated with decreased competition for assimilates as grains reached physiological maturity. Although TRL changed greatly through the season, once roots had reached their maximum depth, there were no significant changes in root distribution through the profile (IRRI 199 1). Changes in TRL probably result mostly from death and initiation of branch roots rather than from changes in primary root axes. These results have important implications for both nutrient uptake and drought resistance. Greater TRL and deeper roots may allow greater or more efficient nutrient uptake in WSR than in TPR. These root system differences also explain differences in drought resistance observed when TPR and WSR are subjected to water deficit. For 24 rainfed lowland rices, water deficit significantly reduced yields of TPR but not of WSR (Table 3, IRRI 1988).

70

lngram et al

Root relationships with drought resistance and water use When upland rice is grown in a range of soil moisture conditions, grain yields are negatively related to RLD (Fig. 3). In contrast, Mambani and Lal (1983) reported that for upland rice growing across a toposequence, grain yields are related to RLD. Several researchers have reported that rice root growth under aeroponic or hydroponic culture is well-correlated with root growth under upland field conditions (Ahmadi 1983, Chang and Loresto 1986). Unfortunately, root traits observed in aeroponic culture were not at all related to drought resistance under rainfed lowland conditions (Ingram et al 1990). Also, neither total RLD nor RLD by soil layer were significantly correlated with yields of 30 rainfed lowland rices grown in a farmer’s field in Tarlac, Philippines (data not shown). The lack of correlation between aeroponic and rainfed lowland field root growth underscores the need to consider the soil environment for which germplasm improvement is targeted. One hypothesis is that thick roots confer drought resistance because they have greater xylem vessel radii, lower axial resistance to water flux, and hence, greater capacity for water uptake from deeper soil layers (Yambao et al 1992). This hypothesis is the opposite of that proposed for wheat grown on residual soil moisture—i.e., that smaller root vessels confer drought resistance (Richards and Passioura 1989). In rice, the axial resistance to water flux can be estimated by summing the individual resistances for different root vessels, assuming that the vessels act as capillaries and obey the Poiseuille-Hagan equation (Yambao et al 1992). Root box and hydroponic research showed that differences in axial resistance to water flux explained shoot-water

3. Relationship between root length density and grain yield of upland rice.

Rice root traits for drought resistance and their genetic variation

71

4. For IR36, the response surfaces for relationships between equivalent root vessel radius, transpiration, and leaf water potential as estimated by a) Poiseuille-Hagan equation or b) the best-fitting curve. Variation in root vessel radius was achieved through root pruning in hydroponic culture.

relations for a single cultivar (Fig. 4) but did not explain differences across cultivars (Fig. 5). Clearly, factors other than axial resistance explain differences among cultivar responses to water deficit. Thick roots may also be hypothesized to confer drought resistance because branching is directly related to root thickness (Fitter 1991). Thick roots persist longer, produce more and larger branch roots, thereby increasing RLD and water uptake capacity. 72

lngram et al

5. For 5 cultivars, the response surfaces for relationships between equivalent root vessel radius, transpiration, and leaf water potential as estimated by a) Poiseuille-Hagan equation or b) the bestfitting curve. Variation in root vessel radius was achieved through different cultivars and through root pruning in hydroponic culture.

Root growth under water deficit stress The minirhizotron has enabled us to make observations of the dynamic changes in root distribution in response to water deficit. In field experiments, we grew 12 rice genotypes under either continuously flooded conditions or drained 25-55 DAS. Total root length for two of the genotypes throughout the season is shown in Figure 6. We grouped the genotypes into three classes based on RLD under stress and nonstress Rice root traits for drought resistance and their genetic variation

73

6. Total root length (0-70 cm soil layer) of IR46 and IR20 when continuously flooded or nonflooded at 25-55 DAS throughout the full crop season.

conditions in the 10-30 cm soil layer, grading them high, medium, and low. We further considered yield potential and effect of stress on maturity (Table 4). The most important finding was that there was great genetic variation in root system plasticity. In all genotypes, root length of stressed plants rapidly dropped below that of continuously flooded plants when the plots were initially drained. There was also a sharp dip in root length when stressed plants were reflooded (Fig. 6). After each dip, genotypes with good root systems (e.g., IR46) produced greater root length in the stressed than in the well-watered plots and had maximum RLDs greater than 0.5 cm cm3 in the 20-30 cm soil layer. Genotypes with poor root systems (e.g., IR20) had greater root length in the control than in the stressed plots and had RLDs less than 0.5 cm/cm 3. The dynamic changes in root length are shown as the ratios of root length in stressed and control treatments (Fig. 7). We hypothesize that these dynamic root responses to 74

lngram et al

Table 4. Root length and other characteristics related to drought resistance of 12 rice genotypes under rainfed lowland conditions. a Relative nonstressed RLD

Maximum stressed RLD

Yield potential

Good root characters S >C

High

Good

+/0

S>C

High

Medium

0

S>C

High

Low (blight)

0

Poor root characters C>S

Low

Good

+

C>S

Medium

Good

0

Intermediate C S

root

characters High

Excellent/Poor

Maturity delay by stress

++

Cultivar/line

IR46 lR49612-45-6-2-2 IR46331-PMI-53-2-1-3 lR47310-117-1-3-3 lR13146-45-2-3 lR43526-523-1-1-1

IR20 IR19431-72-2 IR43449-SKN-522-3-1-3 IR28626-33-3252-3

IR72 Binato

a C = continuously flooded control: S = nonflooded 2555 DAS: +, ++ = less than 5 d and greater than 8 d delay

in maturity by stress, respectively.

7. Idealized relative root length density (root length stress divided by root length control) for drought-resistant and drought-susceptible rice cultivars through a full crop season.

Rice root traits for drought resistance and their genetic variation

75

soil moisture are mostly a result of the shedding of old branch roots and the production of new ones. According to our hypothesis, branch roots produced during stress are adapted to aerobic conditions and have little or no aerenchyma; those produced after reflooding are adapted to anaerobic conditions, with aerenchyma.

Summary and conclusions Among the root traits studied, TRL is the most strongly related to drought resistance under upland conditions. Root traits found to confer drought resistance under rainfed lowland conditions were RLD in the 10-30 cm soil layer and dynamic shedding of roots and production of root length in response to changing moisture conditions. There are still large gaps in our knowledge of the overall function and specific genetic traits of roots in relation to water uptake. The potential for transferring desirable root traits from upland japonicas to lowland indicas is uncertain, although it may be possible to accomplish these transfers with molecular biological tools. The sensing mechanisms that determine root responses to changes in soil moisture or to soil strength are little understood. Many of the root traits that have been proposed as contributing to drought resistance have not yet been fully evaluated. These include maximum depth of longest roots, number of vertically oriented root axes, root osmotic adjustment, and hardpan penetration.

References cited Ahmadi N (1983) Variabilite genetique at heredite do mechanismes de tolerance a la secheresse chez le riz Oryza sativa L.: I. Development du system racinaire. Agron. Trop. 38:110-117. Armenta-Soto J, Chang T T, Loresto G C, O’Toole J C (1983) Genetic analysis of root characters in rice. SABRAO J. 15:103-116. Chang T T, Loresto G C (1986) Genetic analysis of rice root systems by the aeroponic technique and applications in plant breeding. Pages 123-220 in New frontiers in breeding research. B. Napompeth and S.S. Abandhu, eds. Kasetsart University, Bangkok. Chang T T, Loresto G C, O’Toole J C, Armenta-Soto J L (1982) Strategy and methodology of breeding rice for drought-prone areas. Pages 217-244 in Drought resistance in crops with emphasis on rice. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Dingkuhn M, Schnier H F, De Datta S K, Wijangco E J, Dorffling K (1990) Diurnal and developmental changes in canopy gas exchanges as related to growth in transplanted and direct seeded lowland rice. Aust. J. Plant Physiol. 17:119-134. Ekanayake I J, Ganity D P, Masajo T M, O’Toole J C (1985a) Root pulling resistance in rice: inheritance and association with drought resistance. Euphytica 34:905-913. Ekanayake I J, O’Toole J C, Ganity D P, Masajo T M (1985b) Inheritance of root characters and their relations to drought resistance in rice. Crop Sci. 25:927-933. Fitter A H (1991) Characteristics and functions of root systems. Pages 3-25 in Plant roots, the hidden half. Y. Waisel, A. Eshel, and U. Kafkafi, eds. M. Dekker, Inc., New York. Ingram KT, Real J G, Maguling M A, Obien M A, Loresto G C (1990) Comparison of selection indices to screen lowland rice for drought resistance. Euphytica 48:253-260. IRRI—International Rice Research Institute (1988) Annual report for 1987. P.O. Box 933, Manila, Philippines. 640 p.

76

lngram et al

IRRI—International Rice Research Institute (1991) Program report for 1990. P.O. Box 933, Manila, Philippines. 317 p. Mambani B, Lal R (1983) Response of upland rice varieties to drought stress. II. Screening rice varieties by means of variable moisture along a toposequence. Plant Soil 73:73-94. O’Toole J C, Bland W L (1987) Genotypic variation in crop plant root systems. Adv. Agron. 41:91-145. O’Toole J C, Chang T T (1979) Drought resistance in cereals rice: a case study. Pages 373-405 in Stress physiology in crop plants. H. Mussel andR.C. Stables, eds. John Wiley & Sons, New York. O’Toole J C, De Datta S K (1986) Drought resistance in rainfed lowland rice. Pages 145-158 in Progress in rainfed lowland rice research. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Richards R A, Passioura J B (1989) A breeding program to decrease the diameter of the major xylem vessel in the seminal roots of wheat and its effect on grain yield in rainfed environments. Aust. J. Agric. Res. 40:943-950. Yambao E B, Ingram K T, Real J G (1992) Root xylem influence on the water relations and drought resistance of rice. J. Exp. Bot. 43(252):925-932.

Notes Authors’ addresses: K.T. Ingram, F.D. Bueno, O.S. Namuco, and E.B. Yambao, Agronomy, Plant Physiology, and Agroecology Division, International Rice Research Institute, P.O. Box 933, Manila, Philippines; C.A. Beyrouty, Depatment of Agronomy, University of Arkansas, Fayetteville, Arkansas 72701, USA. Citation information: Kirk G J D, ed. (1994) Rice roots: nutrient and water use. International Rice Research Institute, P.O. Box 933, Manila, Philippines.

Rice root traits for drought resistance and their genetic variation

77

Use of molecular markers to exploit rice root traits for drought tolerance H.T. Nguyen, J.D. Ray, and Long-Xi Yu

Current progress in the application of molecular markers and root-specific gene expression to exploit root traits for drought tolerance in rice is reviewed. The root system is central to the mechanisms by which plants resist or escape drought. Research indicates significant genetic variability in the associated root traits, especially root morphology and the ability to penetrate compacted soil layers. However, genetic improvement for root traits is difficult using conventional phenotypic selection criteria. Recent advances in molecular biology and genome mapping may provide new strategies for incorporating root traits into breeding programs.

Plants have various means by which they escape or resist periods of limited water availability (drought) (Ludlow and Muchow 1990). In this paper, we consider the role of the root system in the response of rice to drought and the use of molecular markers as selection tools for incorporating root traits in improved rice cultivars.

Drought and root systems Roots can moderate the effects of drought by growing deeper or thicker to reach more water (and thus increase water supply) or by changing the rate at which water becomes available (Gregory 1989). Individual root characteristics such as thickness, depth of rooting, and penetration ability have been associated with drought avoidance (O’Toole and Chang 1979, Yoshida and Hasegawa 1982, Ekanayake et al 1985). O’Toole and De Datta (1986) suggest that increased root depth and root density in rice increase the capacity to extract available soil water and may be responsible for increased drought avoidance in some rice genotypes. Thangaraj et al (1990) suggest that during water stress, water use is primarily determined by root system density and depth. While

exploitation of a deeper soil volume may be beneficial in avoiding drought, water in these deeper regions is often unavailable due to the presence of compacted soil layers. Compacted soil layers are often found in lowland ricefields, formed as a result of the cultural practices used in preparing land for rice (O’Toole and De Datta 1986). Mechanical disruption of such layers is expensive and short-lived, as the compacted layer often reforms in a few years (Threadgill 1982, Busscher et al 1986). Under these conditions, the ability of roots to penetrate compacted soil layers is an important characteristic associated with drought avoidance.

Effect of compacted soil layers on root systems Compacted soil layers—variously known as hardpans, plowpans, tillage pans, plow soles, and tillage soles (Taylor 1974)—impede root growth. They are one of the most important factors determining root elongation and proliferation (Bengough and Mullins 1990, Tu and Tan 1991). They cause changes in both root morphology and physiology, including a decrease in cell length and a corresponding increase in cell thickness and a decrease in the rate of cell division. Other physiological changes such as osmotic adjustment and changes in plant growth regulator activity are also involved (Bengough and Mullins 1990, Glinski and Lipiec 1990). The effects of this mechanical impedance to roots on plant growth and development include decreased transpiration rates, decreased leaf area expansion, and decreased dry matter accumulation (Masle and Passioura 1987, Ludlow et al 1989, Assaeed et al 1990, Masle 1992). These effects may be direct consequences of reduced root access to water and nutrients. Masle and Passioura (1987), however, found that the effects of soil compaction on shoot growth were manifest before the supply of water and nutrients to the shoot was affected and concluded that plant growth regulators produced in the roots triggered the shoot response (Ludlow et al 1989).

Genetic variability in root systems O’Toole and Bland (1987) have reviewed the genetic variability of root systems in crop plants. Most examples of genetic variation in root system traits indicate quantitative control (Schiefelbein and Benfey 1991). In rice, significant genetic variation in several root characteristics associated with drought avoidance has been reported (O’Toole and Chang 1979; Kandasamy 1981; O’Toole 1982; Chang et al 1982; Ekanayake et al 1985, 1986). The inheritance of root system characteristics in rice, including root length density, root thickness, and root mass, was reported by Chang et al (1982) and Ekanayake et al (1985). These studies indicate that genetic improvement for several root morphological traits in rice is possible through early generation selection. Genotypic variation in root penetration ability has been reported for cotton (Kasperbauer and Busscher 199l), wheat and barley (Masle 1992), and rice (Kandasamy 1981, O’Toole 1982). We have evaluated root penetration by upland and lowland rice cultivars using a wax layer system (Taylor and Gardner 1960) to simulate a compacted soil layer. We found that root penetration ability varied with genotype and that upland 80

Nguyen et al

rice genotypes had a greater penetration ability than lowland rice genotypes. Moreover, root penetration ability was positively correlated with root thickness (unpubl. data).

Use of molecular genetic markers in tagging root characteristics Incorporation of root traits in breeding programs is hampered by the difficulty of measuring root traits in the field and the consequent lack of reliable and efficient screening techniques (Gregory 1989, O’Toole 1989). The quantitative nature of many root characteristics is a further complication. Attempts made to correlate root characteristics with more readily measurable shoot characteristics have met with limited success. For example, deep root systems have been positively correlated with plant height in rice (Armenta-Soto et al 1983, Ekanayake et al 1985, Chang et al 1986). However, while selection for plant height could result in selection for deeper root systems, the ideotype upland rice plant is of short or intermediate height with a deep root system (Ekanayake et al 1985). Ekanayake et al (1985) did obtain recombinant plants with short height and deep root systems, but obviously selection based on plant height alone would not have identified these plants. This illustrates the need for more effective selection criteria for root characteristics. One solution may be the identification of molecular genetic markers associated with specific root traits, such as penetration ability. RFLP molecular genetic markers Plant breeders and geneticists have long made use of genetic markers in selection programs, inheritance evaluations, phylogenic studies, genotype identification, and construction of linkage maps. Classical linkage maps are constructed on the basis of linked morphological characters and have existed since the early 1900s. Other genetic markers have made use of biochemical differences, the most common of which are allozyme (isozyme) markers developed in the 1950s. Since the 1970s, the development of techniques to analyze DNA directly has led to a new class of genetic markers called molecular markers, the most common being restriction fragment length polymorphism (RFLP). RFLP molecular markers have several advantages over more conventional morphological genetic markers. They are usually inherited codominately, exhibit phenotypic neutrality, rarely have epistatic or pleiotropic interactions, and are typically tissue- and phenologically independent (Beckmann and Soller 1986, Tanksley et al 1989). These characteristics and the potentially unlimited number of RFLP loci in a population speed the construction and increase the resolution of linkage maps. Other potential uses of RFLP markers are described elsewhere (Helentjaris et al 1985, Beckmann and Seller 1986, Landry and Michelmore 1987, Tanksley et al 1989). Restriction fragment length polymorphisms are the result of differences (polymorphisms) in DNA fragments obtained by cleaving the DNA with restriction endonucleases. Such polymorphisms are detected as changes in length (molecular weight) of otherwise homologous DNA fragments relative to the restriction site of a specific endonuclease. The changes in the length of the fragments are due to base pair Use of molecular markers to exploit rice root traits for drought tolerance

81

changes in the highly specific recognition site of the endonuclease; DNA remangements resulting in the relocation of the site; or to insertion/deletion events within one of the fragments (Botstein et al 1980; Beckmann and Soller 1983, 1986). Botstein et al (1980) first proposed the use of RFLPs as genetic markers in humans. By 1983, the potential applications of RFLPs in plants were being suggested (Burr et al 1983, Soller and Beckmann 1983, Tanksley 1983), and by the mid-l980s, RFLP linkage maps were published for maize and tomato (Bernatzky and Tanksley 1986, Helentjaris et al 1986). In varying degrees of resolution, RFLP linkage maps are now available for a large number of species including rice (McCouch et al 1988, McCouch and Tanksley 1991). High-resolution linkage maps increase the possibility of identifying markers linked to traits of interest, including quantitative trait loci (QTL) (McCouch et al 1988, Knapp et al 1990, McCouch and Tanksley 1991). The rice RFLP linkage map is presently of sufficient resolution (about 1 cM, McCouch and Tanksley 1991) to permit mapping of QTLs. Identification of RFLP markers linked to desirable quantitative root characteristics would greatly facilitate their selection in breeding programs (Hanson et al 1990). Identification of RFLP markers would also minimize the requirement for exhaustive experiments to characterize root systems since REPS analyze the DNA directly and are tissue-independent. RAPD molecular genetic markers A new class of molecular genetic markers has recently been developed based on randomly amplified polymorphic DNA (RAPD, pronounced “rapid”) (Williams et al 1990, Welsh and McClelland 1990). The RAPD technique is based on the use of single arbitrary sequence oligodeoxynucleotides as primers for polymerase chain reactions (PCR). When primers bind to opposite strands of DNA within an amplifiable distance, the intervening DNA segment can be amplified using PCR. The length of the amplified DNA segments depends on the sequence and length of the primer in conjunction with the reaction conditions (reaction buffer, annealing temperature, etc.). Comparison of the amplified DNA segments by electrophoresis in agarose gels allows the detection of differences (polymorphisms) between DNA from different sources. Polymorphisms may be the result of changes in the primer binding sites, DNA insertions resulting in the binding sites being too far apart for amplification, or DNA insertions or deletions that change the length of the amplified segment (Williams et al 1990). The RAPD polymorphisms are molecular genetic markers and have been used as diagnostic tools in the same manner as RFLP molecular genetic markers (Hu and Quiros 1991, Crowhurst et al 1991, Klein-Lankhorst et al 1991, Martin et al 1991, Paran et al 1991). They have advantages over RFLP markers because they are easier to use, less costly, and faster (Anderson and Fairbanks 1990). An additional advantage is that they do not use radioisotopes, although nonradioactive hybridization techniques have recently been developed for RFLP markers. It has been suggested that RAPD markers could be used to develop linkage maps with greater efficiency and higher resolution than do RFLP linkage maps (Williams et al 1990). However, there are some disadvantages and difficulties in using arbitrary 82

Nguyen et al

primers in constructing linkage maps, such as the dominant nature of RAPD markers which prevents direct determination of heterozygosity (Devos and Gale 1992, Halward et al 1992).

Root-specific gene expression Numerous genes that are expressed in specific tissues have been isolated in plants. However, only a limited number of genes that are specifically expressed in roots have been isolated. These include genes from tobacco (Conkling et al 1990, Keller and Lamb 1989), barley (Lerner and Raikhel 1989), and rice (Pater and Schilperoort 1992). The function of the protein coded by the gene isolated from rice is unknown. Gene expression in roots remains little explored, probably due to the difficulty inherent in studying roots and the limited understanding of root biochemical processes. However, isolation of root genes and characterization of the regulatory elements governing their expression during drought or in response to other environmental stimuli (such as a compacted soil layer) could lead to an understanding of the mechanisms involved in drought avoidance. Recently, a review by Schiefelbien and Benfey (1991) emphasized the lack of genetic and molecular studies on roots compared with shoots. Research using molecular approaches is needed to characterize the patterns of gene expression and isolate genes associated with root penetration ability in rice. Such genes are expected to play a role in allowing the root tip region to grow and penetrate through the compacted soil layers. Understanding the genetic organization and expression of these genes is essential to establish the molecular basis of root penetration traits in plants. In addition, the isolated cDNAs can be used as molecular markers for breeding applications.

Conclusions Incorporation of selection for root characteristics such as root penetration ability into breeding programs offers the possibility of developing genotypes better able to exploit soil water. Better exploitation of soil water reserves would allow plants to avoid the adverse effects of periods of low rainfall. Phenotypic selection of root characteristics in large breeding programs is difficult and requires too much time and labor, thus limiting progress in developing improved genotypes. One solution is to use indirect selection criteria based on the correlation between root characteristics and more readily identifiable shoot phenotypes. However, this may not always be feasible or reliable. A more direct approach is through the use of molecular genetic markers tightly linked to the root characteristic of interest. The first step in using molecular markers in breeding programs is to identify markers associated with desirable root characteristics. We have begun this process in our laboratory with the tentative identification of four RFLP molecular genetic markers associated with root penetration ability in rice (Texas Technical University, unpubl. data). The identification of such markers will provide breeders with better selection criteria for incorporation of desirable root characteristics into improved cultivars. Use of molecular markers to exploit rice root traits for drought tolerance

83

References cited Anderson W R, Fairbanks D J (1990) Molecular markers: important tools for plant genetic resource characterization. Diversity 651-53. Armenta-Soto J, Chang TT, Loresto G C, O’Toole J C (1983) Genetic analysis of root characters in rice. SABRAO J. 15:103-116. Assaeed A M, McGowan M, Hebblethwaite P D, Brereton J C (1990) Effect of soil compaction on growth, yield and light interception of selected crops. Ann. Appl. Biol. 117:653-666. Beckmann J S, Soller M (1983) Restriction fragment length polymorphisms in genetic improvement: methodologies, mapping and costs. Theor. Appl. Genet. 67:35-43. Beckmann J S, Soller M (1986) Restriction fragment length polymorphisms in genetic improvement of agricultural species. Euphytica 35: 111-124. Bengough A G, Mullins C E (1990) Mechanical impedance to root growth: a review of experimental techniques and root growth responses. J. Soil Sci. 41:341-358. Bernatzky R, Tanksley S D (1986) Towards a saturated linkage map in tomato based on isozymes and random cDNA sequences. Genetics 112:887-898. Botstein D, White R L, Skolnick M, Davis R W (1980) Construction of a genetic linkage map in man using restriction fragment length polymorphisms. Am. J. Hum. Genet. 32:314-331. Burr B S, Evola V, Burr F A, Beckman J S (1983) The application of restriction fragment length polymorphisms to plant breeding. Pages 45-59 in Genetic engineering: principles and methods. J.K. Setlow and A. Hollaender, eds. Plenum Press, New York. Busscher W J, Sojka R E, Doty C W (1986) Residual effects of tillage on coastal plain soil strength. Soil Sci. 141:144-148. Chang T T, Armenta-Soto J L, Mao C X, Peiris R, Loresto G C (1986) Genetic studies on the components of drought resistance in rice ( Oryza sativa L.). Pages 387-398 in Rice genetics. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Chang T T, Loresto G C, O’Toole J C, Armenta-Soto J L (1982) Strategy and methodology of breeding rice for drought-prone areas. Pages 217-244 in Drought resistance in crops with emphasis on rice. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Conkling M A, Cheng C L, Yamamoto Y T, Goodman H M (1990) Isolation of transcriptionally regulated root-specific genes from tobacco. Plant Physiol. 93: 1203-1211. Crowhurst R N, Hawthorne B T, Rikkerink E H A, Templeton M D (1991) Differentation of Fusarium solani f. sp. cucurbitae races 1 and 2 by random amplification of polymorphic DNA. Curr. Genet. 20:391-396. Devos K M, Gale M D ( 1992) The use of random amplified polymorphic DNA markers in wheat. Theor. Appl. Genet. 84:567-572. Ekanayake I J, Garrity D P, O’Toole J C (1986) Influence of deep root density on root pulling resistance in rice. Crop Sci. 26: 1181-1186. Ekanayake I J, O’Toole J C, Garrity D P, Masajo T M (1 985) Inheritance of root characters and their relations to drought resistance in rice. Crop Sci. 25:927-933. Glinski J, Lipiec J (1990) Soil physical conditions and plant roots. CRC Press Inc., Boca Raton. Gregory P J (1989) The role of root characteristics in moderating the effects of drought. Pages 141-150 in Drought resistance in cereals. F.W.G. Baker, ed. CAB International, Wallingford, UK. Halward T, Stalker T, LaRue E, Kochert G (1992) Use of single-primer DNA amplifications in genetic studies of peanut ( Arachis hypogaea L.). Plant Mol. Biol. 18:315-325. Hanson A D, Peacock W J, Evans L T, Arntzen C J, Khush G S (1990) Drought resistance in rice. Nature 345:26-27.

84

Nguyen et al

Helentjaris T G, King G, Slocum M, Siedenstrang C, Wegman S (1985) Restriction fragment polymorphisms as probes for plant diversity and their development as tools for applied plant breeding. Plant Mol. Biol. 5:109-118. Helentjaris T G, Slocum M, Wright S, Schaefer A, Nienhuis J (1986) Construction of genetic linkage maps in maize and tomato using restriction fragment length polymorphisms. Theor. Appl. Genet. 72:761-769. Hu J, Quiros C F (1991) Identification of broccoli and cauliflower cultivars with RAPD markers. Plant Cell Rep. 10:505-511. Kandasamy S (1981) Response of rice roots to simulated soil compaction with emphasis on water uptake. MS thesis, University of the Philippines at Los Baños, Laguna, Philippines. Kasperbauer M J, Busscher W J (1991) Genotypic differences in cotton root penetration of a compacted subsoil layer. Crop Sci. 31:1376-1378. Keller B, Lamb C J (1989) Specific expression of a novel cell wall hydroxyproline glycoprotein gene in lateral root initiation. Genes Dev. 3:1639-1646. Klein-Lankhorst R M, Vermunt A, Weide R, Liharska T, Zabel P (1991) Isolation of molecular markers for tomato (L. esculentum ) using random amplified polymorphic DNA (RAPD). Theor. Appl. Genet. 83:108-114. Knapp S J, Bridges W C, Birkes D (1990) Mapping quantitative trait loci using molecular marker linkage maps. Theor. Appl. Genet. 79:583-592. Landry B S, Michelmore R W (1987) Methods and applications of restriction fragment length polymorphism analysis to plants. Pages 25-44 in Tailoring genes for crop improvement: an agricultural perspective. B. Bruning, T. Kosuge, J. Harada, and A. Hollaender, eds. Plenum Press, New York. Lerner D R, Raikhel N Y (1989) Cloning andcharacterization ofroot-specific barley lectin. Plant Physiol. 91:124-129. Ludlow M M, Muchow M C (1990) A critical evaluation of traits for improving crop yields in water-limiting environments. Adv. Agron. 43:107-153. Ludlow M M, Sommer K J, Flower D J, Ferraris R, So H B (1989) Influence of root signals resulting from soil dehydration and high soil strength on the growth of crop plants. Curr. Topics Plant Biochem. Physiol. 8:81-99. Martin G B, Williams J G K, Tanksley S D (1991) Rapid identification of markers linked to Pseudomonas resistance gene in tomato by using random primers and near isogenic lines. Proc. Natl. Acad. Sci. USA 88:2336-2340. Masle J (1992) Genetic variation in the effects of root impedance on the growth and transpiration rates of wheat and barley. Aust. J. Plant Physiol. 19:109-125. Masle J, Passioura J B (1987) The effect of soil strength on the growth of young wheat plants. Aust. J. Plant Physiol. 14:643-656. McCouch S R, Kochert G, Yu Z H, Wang Z Y, Khush G S, Coffman W R, Tanksley S D (1988) Molecular mapping of rice chromosomes. Theor. Appl. Genet. 76:815-829. McCouch S R, Tanksley S D (1991) Development and use of restriction fragment length polymorphism in rice breeding and genetics. Pages 109-133 in Rice biotechnology. G.S. Khush and G.H. Toenniessen,eds. CAB International, Wallingford, UK. O’Toole J C (1982) Adaptation of rice to drought-prone environments. Pages 195-213 in Drought resistance in crops with emphasis on rice. International Rice Research Institute, P.O. Box 933, Manila, Philippines. O’Toole J C (1989) Breeding for drought resistance in cereals: emerging technologies. Pages 81-94 in Drought resistance in cereals. F.W.G. Baker, ed. CAB International, Wallingford, UK.

Use of molecular markers to exploit rice root traits for drought tolerance

85

O’Toole J C, Bland W L (1987) Genotypic variation in crop plant root systems. Adv. Agron. 41:91-145. O’Toole J C, Chang T T (1979) Drought resistance in cereals - rice: a case study. Pages 374-405 in Stress physiology of crop plants. H. Mussell and R.C. Staples, eds. Wiley-Interscience, New York. O’Toole J C, De Datta S K (1986) Drought resistance in rainfed lowland rice. Pages 145-158 in Progress in rainfed lowland rice research. International Rice Research Institute, P.O. Box 933, Manila, Philippines. Paran I, Kesseli R, Michelmore R S (1991) Identification of restriction fragment length polymorphism and random amplified polymorphic DNA markers linked to downy mildew resistance genes in lettuce, using near-isogenic lines. Genome 34:1021-1027. Pater B S, Schilperoort R A (1992) Structure and expression of a root-specific rice gene. Plant Mol. Biol. 18:161-164. Schiefelbein J W, Benfey P N (1991) The development of plant roots: new approaches to underground problems. Plant Cell 3:1147-1154. Soller M, Beckmann J (1983) Genetic polymorphism in varietal identification and genetic improvement. Theor. Appl. Genet. 67:25-33. Tanksley S D (1983) Molecular markers in plant breeding. Plant Mol. Biol. Rep. 1:3-8. Tanksley S D, Young N D, Paterson A H, Bonierbale M W (1989) RFLP mapping in plant breeding: new tools for an old science. Biotechnology 7:257-264. Taylor H M (1974) Root behavior as affected by soil structure and strength. Pages 271-291 in The plant root and its environment. E.W. Carson, ed. University Press of Virginia, Charlottesville. Taylor H M, Gardner H R (1960) Use of wax substrates in root penetration studies. Soil Sci. SOC. Am. Proc. 24:79-81. Thangaraj M, O’Toole JC, De Datta S K (1990) Root response to water stress in rainfed lowland rice. Exp. Agric. 26:287-296. Threadgill E D (1982) Residual tillage effects as determined by cone index. Trans. ASAE 25:859-863. Tu J C, Tan C S (1991) Effect of soil compaction on growth yield and root rots of white beans in clay loam and sandy loam soil. Soil Biol. Biochem. 23:233-238. Welsh J, McClelland M (1990) Fingerprinting genomes using PCR with arbitrary primers. Nucleic Acids Res. 18:7213-7218. Williams J G K, Kubelik A R, Livak K J, Raflaski J A, Tingey S V (1990) DNA polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic Acids Res. 18:65316535. Yoshida S, Hasegawa S (1982) The rice root system: its development and function. Pages 97-114 in Drought resistance in crops with emphasis on rice. International Rice Research Institute, P.O. Box 933, Manila, Philippines.

Notes Authors’ addresses: H.T. Nguyen, J.D. Ray, and Long-Xi Yu, Plant Molecular Genetics Laboratory, Institute for Biotechnology and Department of Agronomy, Horticulture, and Entomology, Texas Technological University, Box 42122, Lubbock, Texas 79409, USA. Citation information: Kirk G J D, ed. (1994) Rice roots: nutrient and water use. International Rice Research Institute, P.O. Box 933, Manila, Philippines.

86

Nguyen et al