237 Euphytica 92: 237–246, 1997. c 1997 Kluwer Academic Publishers. Printed in the Netherlands. Interspecific Oryza Sativa L. X O. Glaberrima Steud. progenies in upland rice improvement Monty P. Jones, Michael Dingkuhn, Gabriel K. Aluko & Mandé Semon West Africa Rice Development Association, 01 BP 2551, Bouaké, Côte d’Ivoire Received 5 March 1996; accepted 14 November 1996 Key words: Wide crosses, doubled haploid breeding, weed suppression, specific leaf area, low input systems, water limited systems, Oryza spp., rice Summary Fertile interspecific progenies between Oryza sativa L. and O. glaberrima Steud. were produced through backcrossing and doubled haploid breeding (DHB). Backcrossing with the O. sativa parents increased fertility and helped combine the O. sativa and O. glaberrima features. The use of DHB to generate a large proportion of doubled haploids from interspecific F2 hybrids, helped overcome constraints associated with the conventional breeding of these species, such as: (1) slow fixation of the lines, (2) frequent partial sterility of the progenies and (3) low recovery of useful recombinants. Although true interspecific progenies between O. sativa and O. glaberrima were generally rare, their occurrence in some populations was as high as 30%. Some progenies combined the high yield potential of O. sativa, a result of high spikelet number caused by secondary branches on the panicle, with useful traits of O. glaberrima such as rapid leaf canopy establishment and high N responsiveness. The progenies partly inherited the O. glaberrima parents’ high specific leaf area (SLA) during early growth, theoretically improving competitiveness with weeds, and from the O. sativa parents the rapid decrease in SLA towards the reproductive stage, theoretically allowing for high leaf photosynthetic rates and high grain yield. Research is in progress to develop new interspecific plant type concepts for resource poor, weed prone upland rice environments in West Africa. Introduction Crop improvement scientists at the West Africa Rice Development Association (WARDA) systematically evaluate germplasm from both in and outside Africa, generate breeding materials, select superior lines and test early and advanced breeding materials on-station and on-farm. WARDA’s strategy for upland rice improvement is to combine specific agro-ecological adaptation of local upland varieties with the yield potential of introductions. Targeting the numerous constraints limiting yields of upland rice (drought, insects and diseases, weeds and low input cultural practices), conventional breeding programs have been active over more than three decades to improve the performance of upland rice varieties in West Africa (Arraudeau, 1992; Bidaux, 1978; Koffi, 1980; Takeoka, 1965). The gains from this research have been limited, in part because Oryza sativa L., the most widely cultivated rice species in West Africa, has limited resistance to many of the stresses that affect upland rice in the region (WARDA, 1992). Although advanced selections from these intra-specific breeding programs mostly outperform farmers’ traditional varieties on-station under relatively high input conditions, they perform poorly when cultivated under the low input systems which dominate extensive upland farming in West Africa. In an effort to break this pattern, WARDA initiatd interspecific hybridization program in 1992 to combine important traits between O. sativa and O. glaberrima, thereby increasing the genetic variability within each type. Interspecific hybridization for plant improvement is complicated by incompatibility barriers. These include 238 hybrid inviability which hinder heterogenetic recombination (Brar & Khush, 1986; Jena & Khush, 1990). Past efforts to transfer useful genes from O. glaberrima into O. sativa were also constrained by sterility in the early progenies of crosses (Second, 1984). Despite these barriers, natural gene flows among O. glaberrima, O. sativa and O. longistaminata have been reported and analysed by a number of researchers (De Kochko, 1987; Second, 1984; Takeoka, 1965) who recommended the exploration of O. glaberrima as a source of new variability for upland rice improvement. Rice scientists have been aware of O. glaberrima for many decades (Brar, 1986; Carpenter, 1978; De Kochko, 1987). Among the eight other species indigenous to Africa, O. glaberrima is known to have been selected and cultivated in parts of West Africa for more than 3500 years (Bidaux, 1978; Carpenter, 1978; Jacqout, 1977). Because O. glaberrima has survived without the help or interference of man, the species has developed adaptive or protective mechanisms for resisting major biotic and abiotic stresses. O. glaberrima represents a rich reservoir of useful genes for resistance to diseases and pests as well as tolerance to acid soils, iron toxicity, drought, unfavourable temperatures and excess water. This paper presents new procedures for developing fertile interspecific progenies from O. sativa and O. glaberrima crosses, and reports preliminary observations on yield potential and other agronomic traits of the progenies under upland conditions. Materials and methods Morpho-agronomic characterization of O. glaberrima and O. sativa accessions age annual rainfall and temperature in 1991 and 1992 were 1146 mm and 25.15  C respectively (minimum 23.3  C, maximum 27.3  C). The soil is well drained, deep sandy clay to clay with a drain pan between 80 cm and 120 cm. The land was disk plowed and harrowed twice to facilitate levelling with spiketooth or drag harrow in June of each year. The entries were grown in single plots of 4 rows by 5 m long. Seeds were dibbled in the rows at 25 cm apart. The compound fertilizer 10-18-18 NPK was applied at 200 kg/ha, resulting in 20 kg N, 36 kg each of P2 O5 and K2 O. Urea was top-dressed at 40 kg N/ha each at 30 and 60 days after sowing. The pre-emergence herbicide, Ronstar was applied at a rate of 4 litres per ha, and hand weeding was done later as necessary to keep the plots clean. Interspecific hybridization and pedigree selection On the basis of the morpho-agronomic characterization, eight O. glaberrima parents that had the best combinations of traits and the best five O. sativa upland rice varieties, WAB56-50, WAB56-104, WAB181-18, WAB96-1-1, and WAB99-1-1 developed by WARDA, were selected for wide hybridization in 1992. Forty eight crosses between O. sativa and O. glaberrima lines were made at WARDA. Seeds of seven crosses that produced a few fertile grains were collected, and the F1 progenies successively backcrossed to the O. sativa parents. After two backcrosses, individuals from these BC2 F1 populations were subjected to pedigree selection. Seeds from the most promising BC2 F1 individuals were selected and grown to produce the BC2 F2 generation. Selection continued until varietal traits were fixed after an additional six generations. Doubled haploid breeding Drawn from the working collection kept by WARDA, 1,721 accessions of improved (316) and traditional (275) O. sativa and O. glaberrima (1130) were evaluated during 1991 and 1992 for a range of morphological and agronomic traits including seedling vigour, growth duration, plant height, panicles per m2 and grain shape, using the standard evaluation system developed by the International Rice Research Institute (IRRI,1988). The trials were carried out at WARDA’s main research station at M’bé in Cote d’Ivoire, located in the Guinea savanna zone at 5 06’W, 7 52’N. The site has a bimodal rainfall pattern with the wettest period occuring during September and October. The driest months are November to February. The aver- On a parallel track, anther culture was used to obtain fertile plants and shorten the number of generations required for the fixation of particular traits. Some of the BC1 F1 progenies from the seven successful crosses were passed through anther culture, in which tillers at booting stage, were removed and taken into the laboratory. Anthers were removed from the spikelets, and placed to a modified N6 growth medium (Chu et al., 1975) containing 0.5 mg/l of Kinetin, 2 mg/lof 2,4D, 5% maltose and 150 ml/l of coconut milk. Calli emerged from the anthers three to eight weeks after plating. The calli were transferred to an MS medium (Murashige & Skoog, 1962) when they were about 239 2 mm in length. After 3-4 weeks, they developed into small plantlets. The green plantlets were then transferred into an MS medium with half the concentration of chemicals, to which was added multi effective triazole (MET), a hardening chemical. This strengthened the plants allowing them to be transferred to grow in the soil. Field evaluation of interspecific progenies Two field experiments were conducted in 1995 at WARDA’s Mbé experimental station to study the fertility, genetic stability and agronomic traits of newly fixed O. sativa x O. glaberrima progenies. Experiment 1 was located on an exhausted upland soil that had been under short fallow (3 years) following three years of continuous rice cropping. The entries included three O. sativa cultivars, IDSA6, WABC165 and WAB56 104; the O. glaberrima parent, CG14; and 22 newly fixed progenies. The entries were planted under high and low input levels of management in 2.5m x 5m plots following a randomized complete block design with four replications. The management and fertilizer level of the high input treatment is as described above in morphoagronomic characterization of the O. glaberrima and O. sativa accessions. For the low input, vegetation was slashed and most of the trees removed. Minimum destumping was practiced and the soil was scarified with a hand hoe. Fertilizer was applied at 20 kg N/ha each at 23 and 43 DAS and hand weeding was done at 21 and 42 days after sowing. Data collected included seedling vigor at 30 DAS, and grain yield and grain moisture content at maturity. Experiment II served a complete growth and yield component analysis under favorable conditions. Four newly fixed lines, WAB450-I-B-P-160-HB, WAB45024-3-2-P18-HB, WAB450-2-3-P33-HB and WAB450I-B-P-31-HB, and their parents, WAB56-104 and CG14, were evaluated in a preliminary yield trial conducted in a fertile upland field that had been under long fallow (> 6 years). The trial was conducted in a two-factorial randomized complete block design with three replications. The factors were variety (six levels) and nitrogen application (0, 40, 80, and 120 kg/ha), half of which was applied basally and the other half, top-dressed at 40 day after seeding (DAS). To all plots, 100 kg/ha as triple super-phosphate and 50 kg/ha as KCL were applied basally. Supplementary sprinkler irrigation was applied to prevent water stress. Dry seed was dibbled at a rate of 3 seeds per hill with a spacing of 25cm x 25cm. Upon seedling establishment, hills were thinned to two plants per hill. At 14-day intervals, specific leaf area (SLA; leaf area / leaf dry weight) of fully emerged, non-senescent leaf blades was determined using in situ area measurement with a LiCor LI-3000 and subsequent sampling for dry matter determination. The SLA is an important genotypic determinant of potential leaf area production, and therefore, weed competitiveness (Dingkuhn et al., 1996). At maturity grain yield and yield components were measured as described by Dingkuhn and Le Gal (1996). Results and discussion Diversity of African rice germplasm Through the morpho-agronomic characterization work in 1991 and 1992 we found wider variation for some traits, such as growth duration, in O. glaberrima than in the traditional and improved O. sativa materials (Table 1). The O. glaberrima landraces also possess many traits adaptative to the traditionally practised shifting cultivation. For example, most O. glaberrima landraces have rapid and profuse vegetative growth with droopy lower leaves (Jennings et al., 1979; Koffi, 1980). These morphological characteristics are at least partly due to a high SLA, resulting in a large leaf area produced with any given amount of assimilates (Dingkuhn et al., 1996). The superior weed competitiveness of upland adapted O. glaberrima landraces has been recently confirmed by Fofana et al. (1995). Weed competitiveness is particularly important for West African rice production since weeding is usually done manually, and effective weed removal may be impossible due to labor shortages or competition with other activities (WARDA, 1993, 1994). Some O. glaberrima varieties have more tillers and panicles than O. sativa. However, the O. glaberrima accessions have far fewer secondary branches on the panicles. For example, in experiment II, the O. glaberrima landrace CG14 had between 11.0 and 12.2 primary branches and between 15.2 and 24.0 secondary branches, depending on N response. By contrast, the O. sativa line WAB56-104 had 12.2 and 12.6 primary and 37.5 and 41.0 secondary branches. In a related study, the O. glaberrima accessions CG14, IG10 and ACC 102257 had between 98 and 134 tillers per hill (mean 120), and nine O. sativa lines of diverse ori- 240 Table 1. Range of variation on some selected characters of O. sativa and O. glaberrima recently characterized for 44 traits at WARDA Mbé station. Character Seedling Vigor (DAS)  Plant height (cm) Growth Duration (days) Number of panicles/m2 Grain lenght to width ratio Number of grain per panicle 1000 Kernel weight(g) O. glaberrima O. sativa (traditional) O. sativa (improved) 7-1 65-138 7-1 80-182 7-1 65-138 75-160 46-470 2-5 75-150 6-38 105-150 6-476 2-5 99-260 6-38 95-160 49-477 2-5 140-280 13-39  Days after sowing (1 to 9, 1 = Extra vigor, 9 = Poor vigor) gin had between 48 and 92 tillers per hill (mean 67) (Fofana et al., 1995). As a result of panicle architecture, most O. glaberrima materials produce only 75 to 150 grains per panicle, as compared to O. sativa, which responds to improved fertility and frequently produces 250 or more grains per panicle. Hence, O. glaberrima has many undesirable traits which result in low yield potential. Lodging, grain shattering, and long seed dormancy are additional constraints to their productivity (WARDA, 1993). In large part, it is due to these agronomically undesirable traits that farmers are rapidly replacing O. glaberrima with O. sativa varieties. Interspecific hybrids Conventional breeding Out of 48 wide crosses, only seven produced more than five percent fertile seeds in the F1 generation. Similar results were obtained by Jena and Khush (1990) who reported 1.3% seed set in a cross between O. officinalis and O. sativa. However, up to 5% fertility was obtained in our F1 hybrids which had the following O. glaberrima parents: CG14, CG20, T2, YG230 and YG170. The F1 plants showed a number of traits similar to their O. glaberrima parents, including infrequent secondary branching of the panicle, and early seedling vigor. Fertility was improved to 30-65% after two backcrossings to the O. sativa parents. The progeny plants were variable morphologically. Traits appeared that were absent in the parents and earlier generations, such as purple leaf sheath, awns and apiculus, presumably as a result of new gene combinations. Backcrossing with the O. sativa parents also helped combine the O. sativa and O. glaberrima features. The BC2 F1 population consisted of 2 or 3 distinct morphological groups. The majority of the plants in the population resembled either the O. sativa or O. glaberrima parent, and with spikelet fertility higher than 70%. In these progenies only a limited number of features of the other parent had been incorporated. Although true interspecific progenies with combined features of O. sativa and O. glaberrima were generally rare, their occurence in some populations, such as WAB449 (WAB56-104 x T2) and WAB450 (WAB56 -104 x CG14), was as high as 30%. The distribution of panicle types of the BC2 F1 individuals was strongly skewed towards the parental types (Figure 1). This shift was considerably accelerated towards the O. sativa panicle type with primary and secondary branches when backcrossing was continued with the O. sativa recurrent parent. Some of the progenies had a ligule length of 15–25 mm, compared to about 6 mm in O. glaberrima and 45 mm in O. sativa. Various degrees of awning and shades of purple apiculus were also observed in the interspecific progenies. Most importantly, some progenies combined high spikelet number of O. sativa, with many useful vegetative traits of O. glaberrima. The latter included rapid vegetative growth with droopy lower leaves, high tillering, short duration between 75 to 100 days, and superior grain quality. Seed fertility of the progenies was between 30 and 65% in the BC2 F1 populations, suggesting a gradual elimination of heterozygotes. However, in the two crosses mentioned above, fertility did increase with each generation. In the repeated selection process within the progenies, desirable traits were retained and fertility rates 241 Figure 1. Frequency distribution of panicle and ligule types in 300 BC2 F1 individuals from the cross, WAB56-104 (O. sativa) and CG14 (O. glaberrima). 1. Progenies with primary and secondary branches on the panicles and with ligule length of 40-45 mm as in the O. sativa parent. 2. Progenies with primary and infrequent secondary pranches on the panicle and with ligule length of 15-25 mm. 3. Progenies with mainly primary branches on the panicle and ligule length of 4-6 mm as in the O. glaberrima parent. increased in each subsequent generation, and reached 94 to 100% in the BC2 F6 generation. Seed shattering was greatly reduced, and the selection of plants with thick culms solved the problem of lodging. BC1 F1 anther explants had 96 to 100% fertility. The rapid genetic fixation through haploid doubling has the additional advantage of retaining genes which would otherwise be lost through conventional selection. Doubled haploid lines Field performance of interspecific progenies In most cases, several plantlets were regenerated from different sectors of a single anther based callus. Twenty-two percent of the plantlets were green, and were used in the next stage of the process. The green plants fell into three distinct groups, depending on their morphological features and the number of chromosomes. Fifty-two percent of the plants were haploids with 12 chromozomes, 41% doubled haploids, and 7% were polyploids having chromosome number beyond the normal diploid state. Fifteen percent of the spontaneously doubled haploid lines displayed only partial fertility. Sterility might have been caused by aneuploidy of the regenerated plants or to the action of sterility genes. Partially sterile recombinant inbred lines have been produced from distant hybrids of rice through fixation of allelic recombinations of these genes (Guiderdoni et al. 1992; Oka, 1988). However, several genetically stable The incorporation of genes from O. glaberrima into O. sativa produced new plant prototypes with interesting morphological and agronomic traits. In experiment I, the interspecific progenies responded to input levels in much the same way as the O. sativa cultivars. Under high input conditions, 11 progenies were the top yielders. With average grain yields of 3.4 to 3.8 t/ha, they significantly outyielded the O. glaberrima parent, CG14 and were equivalent in yield to the O. sativa check, IDSSA6 (Table 2). A new progeny, WAB450-24-2-3-P33-HB, produced the highest plot grain yield of 4.7 t/ha. Under low input management, 2 of the newly fixed progeny outyielded the O. sativa check cultivar and had grain yields similar to or higher than the O. glaberrima, indicating their adaptation to low input conditions. Seedling vigor ratings varied between 1 (extra vigorous) in the O. glaberrima and 5 (normal seedlings) 242 Table 2. Grain yield of 22 Interspecific progenies and their parents under high and low input levels of management Variety grain yield (kg/ha) High input Low input WAB450-I-B-P-106-HB WAB450-24-2-3-P33-HB WAB450-I-B-P-38-HB WAB450-24-2-1-P-5-HB WAB450-I-B-P-20-HB WAB450-I-B-P-160-HB WAB450-24-3-2-P18-HB WAB450-11-1-1-P3-HB WAB450-I-B-P-133-HB WAB450-I-B-P 91-HB WAB450-I-B-P-31-HB IDSA 6 WAB450-I-B-P-32-HB WAB450-I-B-P-23-HB WABC 165 WAB450-I-B-P-62-HB WAB56-104 WAB450-I-B-P-142-HB WAB450-24-3-1-P135-HB WAB450-I-B-P-147-HB WAB450-I-B-P-105-HB WAB450-15-3-1-P8-HB WAB450-11-1-8-KB WAB450-I-B-P-137-HB WAB450-I-B-P-28-HB CG14 3692 a 3665 ab 3610 ab 3600 ab 3583 ab 3578 ab 3565 ab 3533 ab 3525 ab 3448 abc 3370 abc 3368 abc 3305 abc 3278 abcd 3270 abcde 3259 abcde 3255 abcde 3220 abcde 3120 bcdef 2933 cdef 2715 defg 2699 efg 2608 gf 2493 g 2373 g 1693 g 1613 bcde 1496 1833 abcde 1511 de 2003 abc 2164 a 1463 de 1395 de 1685 bcde 1533 cde 1703 abcde 1390e 1521de 1498 de 1885 abcd 1714 abcde 1460 1456 de 1739 abcde 1705 abcde 1667 bcde 1640 bcde 1542 cde 1409 de 1668 bcde 2073 ab O. sativa check = IDSA 6 O. glaberrima check = CG14 Means with the same alphabets do not differ significantly from each other. in the O. sativa parent. Ten progenies scored 1 or 2 while 11 scored 2.5 (Figure 2). Early vegetative vigor showed the ability of the new plant types to rapidly establish ground cover filling the space between plants and rows. In experiment II, grain yields of the progenies responded strongly to N inputs and were in all treatments equal or superior to the O. sativa parent, and in all cases superior to the O. glaberrima parent for which grain production did not respond at all to N levels. The progeny WAB450-24-3-2-P18-HB significantly outyielded both parents under high N levels (5.6 t/ha; P < 0.05). Among the test entries, only the O. glaberrima parent showed lodging (100% at heading stage, even in the zero-N treatment) and grain shattering (about 1 t/ha, or 30 to 40% of the grain produced). The high, sometimes transgressive, yields observed in the progenies resulted in part from a high rate of secondary panicle branches (1.5 secondary branches per primary branch in the O. glaberrima parent, 3.2 in the O. sativa parent, and between 3.2 and 3.5 in the progenies). The number of primary branches was identical in the two parents (about 12 per panicle), but significantly higher in the four progenies, indicating that primary and secondary branching were transgressive. Tertiary branches were generally rare. The progenies had comparatively open panicles that were mostly larger than those of both parents (Figure 3). The O. glaberrima parent and the progenies showed superior seedling vigor as compared to the O. sativa parent, resulting in rapid aboveground biomass accumulation during exponential growth. The rapid initial growth of the progenies and theirO. glaberrima parent was associated with faster leaf growth (Figure 4). The O. glaberrima parent developed more than twice the leaf area of the O. sativa parent under zero-N inputs, and 3.5 times its leaf area in the 80 kg N treatment. The leaf area index (LAI) of the progenies was generally intermediate. Specific leaf area (SLA) was in part responsible for the extremely different LAI observed among genotypes. The SLA was strongly affected by genotype and phenological stage, but not by N resources (Figure 5). The O. glaberrima parent had a high SLA throughout its development. By contrast, the O. sativa parent had a much lower initial SLA which even decreased significantly in the course of development, indicating that leaves accumulated more biomass per unit leaf area, and, most likely, become thicker. The progenies had generally intermediate SLA during early growth stages, followed by a decrease at least as sharp as that of the O. sativa parent. We observed linear correlations between LAI and SLA across genotypes at any given sampling date (exemplarily shown for 52 DAS in Figure 5). The slope of these relationships, however, depended on N inputs. Genotypes with high SLA achieved a high LAI because less resources were invested per unit area. This, in turn, enabled a higher light harvest and more rapid growth, as confirmed by preliminary modelling studies based on the rice growth model ORYZA-1 (Kropff et al., 1994). At any given sampling date, and across genotypes, SLA was negatively correlated with areal leaf N and chlorophyll content (P < 0.05). This resulted in relatively pale leaves in O. glaberrima and some of the progenies. The weight-based N content of O. 243 Figure 2. Seedling vigor of O. sativa, O. glaberrima and interspecific lines tested under low-input management, (1 = high vigor, 9 = low vigor). Figure 3. Panicle types observed for the O. sativa line WAB56-104. the O.glaberrima cultiva CG14 and an inter-specific progeny WAB45024-3-2-P18-HB. 244 Figure 4. Time courses of leaf area index (LAI) observed in an upland field fertilized with 0 and 80 kg N/ha for an inter-specific rice progeny (WAB 450-24-3-2-P18-HB) and its parents WAB 56-104 (O.sativa) and CG14 (O. glaberrima). Figure 5. Left: Time cources of specific leaf area (SLA) for an inter-specific rice progeny (WAB 450-24-3-2- P18-HB) and its parents WAB 56-104 5 (O.sativa) and cg14 (O. glaberrima). The broken line indicates the ‘ideal’ SLA for a high yielding, weed competitive plant types. Right: Relationship between leaf area index (LAI) and SLA across inter-specific progenies and their parents, as observed during late vegetative stage in an upland field fertilized with two nitrogen levels. 245 Figure 6. Diagram of the basic concept for the development of a weed competitive , inter-specific plant type. glaberrima leaves was high, but spread over a larger area. It follows that the generally pale appearance ofO. glaberrima leaf canopies in this study was not caused by N deficiency (because the weight-based N content was high), but rather, by the thin and translucent leaves. It also follows that, when genotypic variability in SLA is high, selection for dark green leaves does not only result in plants with high N content, but also in thicker leaves (low SLA). This may unintentionally favor plants with poor initial vigor and ground cover, and consequently, poor weed competitiveness. Preliminary studies on other O. glaberrima landraces confirmed that high SLA and pale leaves are common traits in this species, and are associated with high growth vigor during vegetative stages. Conclusion and outlook The present study provided morpho-physiological evidence for trait incorporations from O. glaberrima into an O. sativa background. Most of these incorporations are of potentially high adaptive value in a resource- limited production environment, particularly in weed prone upland fields. They include: (1) rapid vegetative growth and leaf area development, at least in part caused by high initial SLA and partitioning of much assimilate to leaves during early growth; and (2) droopy leaves during early growth stages, theoretically resulting in a high extinction coefficient for solar radiation, and thus, high light use efficiency and suppression of weeds. Expressed in combination with agronomically useful traits derived from O. sativa, such as large panicles, sturdy stems and erect foliage during reproductive stages, these traits are expected to improve yield stability at a high level of potential yield. Preliminary studies indicate that weed competitiveness is not the only adaptation that new plant types can draw from O. glaberrima. Although we know little about the underlying mechanisms, it is already evident that at least one of the O. glaberrima parents in this study, CG14, is highly drought resistant. Many O. glaberrima landraces are also highly resistant to blast, rice yellow mottle virus and the African rice gall midge. A series of studies has been initiated at WARDA to morpho-physiologically and genetically character- 246 ize the traits mentioned above, trace incorporations through molecular markers, develop models to compose and test environment specific plant type concepts, and identify compatible donors for the traits constituting the new plant types. The new plant type concepts, as well as the screening tools that will contribute to their realization, will take into account the dynamic nature of morpho-physiological traits. For example, plants will be screened for a droopy, profuse foliage with high SLA during early growth stages, and for erect leaves with much lower SLA during reproductive stages. In other words, the new plant types will resemble O. glaberrima during early andO. sativa during later growth stages (Figure 6). Preliminary results indicate that this ambitious objective is indeed feasible. Acknowledgements The authors wish to thank the Rockefeller Foundation for providing funds for the anther culture studies. References Arraudeau, M., 1992. Upland rice breeding concepts, strategies, accomplishments, problems. Briefing paper for brainstorming session in PBGB Division. 11 October 1992. Bidaux, J.M., 1978. Screening for horizontal resistance to rice blast (Pyricularia oryzae) in Africa. In: I.W.Buddenhagen & G.J. Persley (Eds.), Rice in Africa. pp. 159–174. Academic Press, London. Brar, P.S. & G.S. Khush, 1986. Wide hybridizaion and chromosome manipulation in cereals. In: D.A. Evans, W.R.Sharp & P.C. Ammirato (Eds.), Handbook of plant cell culture. Vol 4. pp. 221– 256. Carpenter, A.J., 1978. Rice history. In: I.W. Buddenhagen & G.J.Persley (Eds.), Rice in Africa. pp. 3–10. Academic Press, London. CHU, C.C., C.S. Wang, C.S. Sun, C. Hsu, K.C. Yin, C.Y. & Chu 1 F. Bi, 1975. Establishment of an efficient medium for anther culture of rice through comparative experiments on nitrogen sources. Sci. Sinica 18: 659–668. De Kochko, A., 1987. Isozymic variability of traditional rice, Oryza sativa L. in Africa. Theor. Appl. Genet. 73: 675–826. Dindhuhn, M. & P.Y. Le Gal, 1996. Effect of drainage date on yield and dry matter partitioning in irrigated rice. Field Crop Research (in press). Dingkuhn, M., M. P. Jones & A Sow, 1996. New high-yielding, weed competitive plant types drawing from O. sativa and O. glaberrima genepools. In: WARDA. 1996. Annual Report for 1995. pp. 4–12. Fofana, B., T. Koupeur, M.P. Jones & D.E. Johnson, 1995. The development of rice varieties competitive with weeds. Brighton Crop Protection Conference. Weeds - 1995, Volume 1. pp. 187– 192. 49 Downing street, Farnham, Surrey GU9 7PH, UK. Guiderdoni, E., E. Galinato, J. Luistro, & G. Vergara, 1992. Anther culture of tropical japonica x indica hybrids of rice ( O. sativa L.). Euphytica 62: 219–224. IRRI., 1988. Standard Evaluation System for Rice. International Rice Research Institute Los Banos, Philippines. Jacqout, M., 1977. IRAT and Rice genetic resources. Paper presented at the Rice genetic conservation workshop. IRRI/IBPGR. December 1977. p. 11. Jena, K.K. & G.S. Khush, 1990. Introgression of genes from Oryza officinalis ex Watt. to cultivated rice, O. sativa L. Theor. Appl. Genet. 80: 737–745. Jennings, P.R., W.R. Coffman & H.E. Kauffman, 1979. Rice improvement. Intl. Rice Res. Inst., Los Banos, Laguna, Philippines. Koffi, G., 1980. Etude de la Variabilité des Riz Africains en Vue de leur Utilisation en sélection. Bouaké, Côte d’Ivoire. Kropff, M.J., H.H. van Laar & R.B. Matthews. 1994. ORYZA 1. A basic model for irrigated lowland rice production. Intl. Rice Res. Inst. Los Banos, Philippines. Murashige, T. & F. Skoog, 1962. A revised medium for rapid growth and bio-assay with tobacco tissue culture. Physiologia Plantarum, 15: 473–497. Oka, H.I., 1988. Origin of cultivated rice. Developments in Crop Science. Vol 14, Japan Scientific Society Press okyo and Elsevier. Amsterdam Second, G., 1984. Relations évolutives chez le genre oryza et processus de domestication des riz. Etude set theses, ORSTOM, Paris, France. Takeoka, T., 1965. Taxonomy and chromosome numbers of African representatives of the Oryza officinalis complex. Bot. Mag. (Tokyo) 78: 198–201. WARDA, 1992. Annual Report for 1991. West Africa Rice Development Association, 01 BP 2551, Cote d’Ivoire. WARDA, 1993. Annual Report for 1992. West Africa Rice Development Association, 01 BP 2551, Cote d’Ivoire. WARDA, 1994. Annual Report for 1993. West Africa Rice Development Association, 01 BP 2551, Cote d’Ivoire.