Summary
The International Rice Genebank at IRRI maintains a collection of more than 86 750 accessions of cultivated and wild rice species, plus approximately 20 300 recently acquired samples that are being multiplied for long-term conservation. A core collection should facilitate security back-up storage in other genebanks, and will be essential for exploiting germplasm through genome research. Passport, characterization and evaluation data are available for many if not most of the accessions, but no single data set is complete. The existing isozyme classification of rice varieties is useful for validating molecular analyses of rice diversity, but measurements of diversity depend on the choice of molecular markers. Accessions for a rice core collection could be identified using isozyme group frequencies by country, which will reflect much of the existing diversity. These samples can be supplemented by accessions representing all the characterization categories. However, the core collection should change in structure over time, and have variable content and size.
Introduction
The research centres of the Consultative Group on International Agricultural Research (CGIAR) hold in trust more than 500 000 samples of crop germplasm of their mandate species, including the world's staple crops (Fuccillo et al. 1997). The germplasm collection maintained by the International Rice Research Institute (IRRI) in its International Rice Genebank (IRG) represents almost 18% of the CGIAR germplasm holdings. With more than 86 750 registered accessions, plus almost 20 300 samples sent to the genebank for long-term conservation in the past four years (Fig. 1), this germplasm is the world's most genetically diverse collection of rice with more than 110 countries represented (Jackson 1995, 1997). Most of the samples are Asian rice or Oryza sativa L. and a high proportion are farmers' varieties or landraces collected in all the rice-growing ecosystems (irrigated, rain-fed lowland, upland and deepwater), as well as landrace varieties of the West African indigenous cultigen O. glaberrima Steud., and all 22 wild species in the genus Oryza. The genebank collection also contains some released rice varieties, elite breeding lines and genetic stocks. This genebank collection underpins food security in Asia, where rice is the staple food of more than half the world's population (Jackson and Huggan 1993).
In addition to the International Rice Genebank Collection (IRGC), there are large rice collections in China and India. Some countries in Asia, such as Thailand, Philippines and Malaysia, have genebanks exclusively for rice, in addition to national genebanks for other crop species and wild relatives. Many countries have duplicated almost their entire collections of rice germplasm in the IRGC. Two other CGIAR centres - the International Institute for Tropical Agriculture (IITA) in Nigeria, and the West African Rice Development Association (WARDA) in Côte d'Ivoire - also maintain smaller collections of rice than the IRGC. However, most of the germplasm from these two centres is also included in the IRGC (Jackson et al. 1997).
In 1984, Sir Otto Frankel expounded his ideas on core collections, a concept aimed at: (1) enhancing the management of germplasm collections, especially large ones such as that at IRRI, and (2) facilitating the study and use of the conserved germplasm (Frankel 1984). In particular, he was concerned about the apparent lack of evaluation of conserved germplasm, which otherwise should lead to increased use. But, Frankel insisted that this must be based on a deeper understanding of the genetic mechanisms underlying the observed diversity, and that genebank curators should attempt to understand quantitative characters that are polygenic and subject to environmental variation. In proposing the development of core collections, Frankel believed that germplasm evaluation and use were best achieved by studying small and genetically representative subsamples from large collections. In this context, it is therefore appropriate to assess how the core collection concept can be applied to the management and use of the IRGC.
The needs for a core collection of rice
It is easier and more economical to study and evaluate some characters rather than others. Description of highly heritable and easily observable characters is feasible for all accessions, even in a large collection such as the IRGC. For other characters, the existence of a core collection would facilitate their study. The same applies to specific approaches to germplasm conservation and use. A core collection of rice might serve several purposes: (1) for duplicate conservation, (2) to promote use of the IRGC, (3) to study diversity per se, and (4) to exploit synteny among grass genomes.
Duplicate conservation
The IRG meets all the conservation conditions and standards (FAO and IPGRI 1994) approved by the Food and Agriculture Organization of the United Nations (FAO) that were developed by the International Plant Genetic Resources Institute (IPGRI). An external review through the CGIAR's System-wide Genetic Resources Program (SGRP) confirmed the IRG's adherence to these standards (SGRP 1996, 1997). Nevertheless, Vaughan and Jackson (1995) indicated that back-up security conservation of the IRGC in more than one genebank would be desirable.
Currently, IRRI has an agreement with the United States Department of Agriculture - Agricultural Research Service (USDA-ARS) for "black box" security back-up of the IRGC at the National Seed Storage Laboratory (NSSL) in Fort Collins, Colorado. Progress in back-up of the IRGC at NSSL is shown in Fig. 2. Few genebanks around the world have the necessary storage facilities (in this case at -18°C) or the capacity to offer back-up conservation on this scale. The safe and durable conservation of the whole IRGC will therefore be ensured through duplicate conservation between the two genebanks.
Given that most germplasm accessions in the IRGC originate in Asia, additional duplicate conservation sites in this region would be appropriate, especially in the light of political concerns that have emerged in recent years about ownership of and access to germplasm collections. Because of limits to storage facilities in most Asian countries, the development of a core collection of the IRGC, in addition to facilitating the search for and use of germplasm in the collection, would permit back-up conservation of a representative sample of the collection, that is, a core, at more than one location (Vaughan and Jackson 1995).
Use of the IRGC
A strength of the CGIAR centres is their close links between conservation and use of genebank collections, and the situation at IRRI is no different. IRRI has very strong links between genetic conservation and rice breeding. In fact, when IRRI plant breeders began to develop the so-called "new plant type" in the late 1980s, they searched the IRGC for rice varieties that they believed would contribute useful traits to the "ideotype" they had designed (Peng et al. 1994; Khush 1995). Unfortunately, many countries have a marked separation between germplasm conservation and use, but that is not the case with the CGIAR centres.
In some ways, a de facto core collection does exist in the IRGC; it encompasses the accessions that rice researchers most frequently request. But, these accessions are unlikely to be representative of the diversity of rice accessions in the collection, as Frankel (1984) had indicated a core should be. Often, the choice of germplasm accessions for breeding or research (particularly nowadays in the areas of biotechnology and molecular biology) is based more on reports of use by other researchers or the existence of a particular useful trait that is expressed by an accession, rather than on demand for a range of diverse germplasm per se. As such, requests for germplasm from the IRGC tend to be targeted toward specific accessions rather than somewhat "random" choices based on origin or assumed pattern of genetic diversity.
In the same way, IRRI breeders (like most breeders) have their own working collections, built from their personal knowledge of rice germplasm and their personal perception of what diversity is useful. Would a core collection built from optimized sampling methodologies promote the use of a larger set of germplasm by breeders? Certainly, it would help breeders who are willing to explore more exotic germplasm choose among thousands of accessions. But a core collection would mostly be useful to evaluate for new traits. On the basis of distribution of frequencies of useful traits among different groups, breeders would be encouraged to explore from the entire IRGC more germplasm belonging to the most valuable group. For example, Glaszmann et al. (1996) used a sample of 261 accessions that represented "the whole array of variation for several parameters, including geographic origin, culture type, and classification based on isozymes". They showed differences between isozyme groups for the distribution of resistant accessions, particularly for blast, but also for rice bacterial leaf blight and tungro virus disease. These results can help in designing screening strategies more focused on particular isozyme groups. As noted by Hamon et al. (1995), "the core is not an entity apart, but is a guide and entry point for the whole collection".
Oftentimes, the definition of use of a germplasm collection is confined to whether accessions have been used in breeding and appear in the pedigrees of released varieties. Under this narrow definition, few germplasm collections probably match this criterion. On the other hand, thousands of germplasm samples are sent to and used by rice researchers worldwide annually to study reaction to biotic and abiotic stresses and to elucidate biochemical pathways and the molecular control of different traits. All this use contributes to rice science, and facilitates the deployment of germplasm accessions that are actually used in rice breeding.
Although a core collection cannot address all the needs of researchers, it would help rationalize resources if a given set of accessions were defined on which all kinds of evaluation would be done. Several researchers define their own representative sample of rice diversity, often through ignorance of the work of others. As these "core collections" differ from each other, cross-comparisons of results are not possible. In the same way that a limited number of mapping populations serve as reference frameworks for the identification of quantitative trait loci (QTL) in rice - that is, the doubled haploid population derived from a cross between indica rice variety IR64 and japonica variety Azucena (Guiderdoni et al. 1992) - and permit the integration of all research efforts, a core collection of rice would help compare the distribution of traits and identify possible correlations. Dr Brigitte Courtois, an upland rice breeder at IRRI, informed us that when she and colleagues started to screen accessions for allelopathic properties (Courtois and Olofsdotter 1998), having these accessions already characterized for their root system would have been extremely useful.
Study of diversity
The study of diversity per se is important for developing or refining conservation strategies that should permit better targeting of germplasm-collecting sites or the identification of populations for in situ conservation. Our understanding of genetic diversity constantly progresses. Isozymes were a revolution as they provided an insight into selectively neutral polymorphism. Since then, other molecular markers have been used to assess genetic polymorphism. Molecular markers have been used in rice to measure diversity (Resurreccion et al. 1994; Yang et al. 1994; Ghareyazie et al. 1995; Mackill 1995; Virk et al. 1995a), predict quantitative traits (Virk et al. 1996), and understand evolutionary patterns (Second 1991). Virk et al. (1999) have argued strongly that anonymous markers such as RAPD and AFLP are particularly useful for assessing rice diversity because they are distributed throughout the genome. The next significant move in the description of genetic diversity is likely to be the assessment of functional diversity at the DNA level, that is, the description of allelic diversity at QTL.
For O. sativa, the study of tens of thousands of accessions using molecular markers is clearly prohibitively expensive now, although we might expect costs to fall as technology develops in the future. Evaluation of a smaller subset or core would be a more appropriate course to follow, based on the most relevant premise for undertaking such a molecular marker study of diversity in the first place. Nevertheless, molecular marker data can be used to refine a core collection selected on the basis of other morphological and agronomic characters that are in the main more easily observable.
A possible application is provided by the approximately 10 000 accessions that were collected in the Lao People's Democratic Republic (Appa Rao et al. 1997). This wealth of diversity cannot, because of its size, be used optimally or easily. Development of a core collection of this Lao collection is feasible, based on passport data that were, in this case, carefully recorded. The diversity of a core collection could then be assessed with isozymes and DNA markers to derive an estimate of the genetic diversity in the various subgroups generated from passport data. This evaluation would give information on the most interesting subgroups, permitting a more efficient screening process and the possible building of a more complete core of Lao accessions.
Synteny among grass genomes
An exciting area of comparative genetics is the observed synteny among grass genomes (Devos and Gale 1997; Gale and Devos 1998). Molecular genetics is now changing our understanding and manipulation of diversity within germplasm collections. Considerable data also link molecular markers to gene loci for pest and disease resistance, and QTL. Recent developments in comparative genetics have demonstrated the validity of exploiting genome information between species. For example, rice is regarded as a model for cereal genomics because of its small genome. The extensive synteny of cereal genomes means that the genetic and physical maps of rice can be used as reference points for exploring the much larger and more difficult genomes of the other major cereal crops. Conversely, decades of breeding work and molecular analysis of maize, wheat and barley can now find direct application in rice improvement. Clearly, such studies cannot be carried out on a large number of accessions, and the existence of a core could be advantageous in this respect.
Opportunities
There is no need to develop a rice core collection in a vacuum, so to speak, because several necessary data sets already exist. The IRGC has two important characteristics, however, which should be emphasized. First, more than 90% of the O. sativa accessions have been scored for 44 morphological and agronomic traits in characterization plots at Los Baños (14°N, 50 m above sea level) in the Philippines. Furthermore, many accessions have been evaluated for resistance to or tolerance for a wide range of biotic and abiotic stresses, such as diseases, insects and cold (Jackson 1995). Second, all data on the collection are managed in a comprehensive database, the International Rice Genebank Collection Information System (IRGCIS). Despite the wide availability of these data, however, we must caution that they are not complete for any one character, and this includes the passport data as well.
As indicated earlier, the isozyme classification developed by Glaszmann (1987) for O. sativa has two important properties: it is both robust and biologically meaningful. As noted by Glaszmann et al. (1996), this classification has been largely confirmed by several studies of the diversity of O. sativa using various molecular markers such as RAPD (Mackill 1995; Virk et al. 1995a) and RFLP (Wang and Tanksley 1989; Second and Ghesquière 1994). This classification reflects biological differences, for example, for cold tolerance (Glaszmann et al. 1990), reproductive barriers (Pham 1991) or disease resistance (Glaszmann et al. 1996). Thus, it is clear to us that the development of a rice core collection must take into account this isozyme variation, even though that would require the generation of considerable data.
In addition to the more standard morphological characterization of rice germplasm, we have developed several molecular protocols to generate useful diversity data for rice, which can be used consistently across different accessions (Virk et al. 1995a, 1996) and between laboratories.
Special factors that affect development of a rice core collection
While the needs for a rice core collection can be easily articulated, and some opportunities could be exploited for the IRGC, we must address several obstacles that will constrain its development. Most nations adopted the Convention on Biological Diversity (CBD) at the Earth Summit (UNCED) in Rio de Janeiro in July 1992. Since its coming into force as an international legal instrument in December 1993, it has changed forever the policy environment that governs access to and management of genebank collections, as well as the sharing of benefits that result from the use of genebank accessions. The legal status of genebank collections acquired prior to the CBD has still not been resolved. The IRGC comprises two distinct sets of material - pre- and post-CBD accessions - that must be managed legally in different ways. Under the CBD, nations have sovereign rights over their genetic resources and can therefore place certain restrictions on access to and distribution and use of accessions in the IRGC. These conditions of ownership and duplication in the IRGC may therefore preclude their inclusion in any future rice core collection.
A further complicating factor relates to the whole issue of intellectual property rights (IPR). The controversy that continues to surround IPR and their effect on conserved germplasm has made nations much more conscious of the status of duplicated germplasm in international collections, and wary of possible misuse of their germplasm. Under the current situation, developments to broaden and facilitate access to rice germplasm accessions as part of a core collection may not be universally welcomed. We can only hope that discussions in the FAO Commission on Genetic Resources for Food and Agriculture will resolve the outstanding issues on access to germplasm, and the important role of the international genebank collections in underpinning world food security.
Several species
Given the current structure of the IRGC, we strongly believe that the development of a single core for the entire collection is neither necessary nor particularly feasible. On the contrary, we advocate the development of separate cores for O. sativa, O. glaberrima, and the wild species. We emphasize this because of different conservation needs and different types and availability of data and their use in rice breeding. Access to information on the approximately 100 000 samples of O. sativa rice varieties should have top priority, simply because of the size and complexity of this germplasm. For example, most data in the collection are available for O. sativa accessions and, furthermore, the classification based on isozymes has been developed only for O. sativa (Glaszmann 1987).
Passport data
Selection of germplasm accessions on the basis of passport data is usually the primary criterion for inclusion in any core collection, but this approach is limited for the germplasm in the IRGC. Regrettably, most O. sativa accessions in the IRGC have inadequate passport data. In fact, comprehensive data are available for less than a quarter of the collection. Most O. sativa germplasm was donated to the IRGC without the necessary passport data, and a tremendous effort would be needed to trace its origin unequivocally. The diversity of rice-growing conditions is such that the country of origin data are not sufficient. Data for the wild species are much better because IRRI scientists have been more active directly in recent years in collecting this germplasm. On the other hand, the availability of characterization and evaluation data may offset the deficiency in passport data.
Molecular markers
Although the use of molecular markers has undoubtedly opened up new perspectives on the structure of genetic diversity in germplasm collections, several points should be considered. The first relates to the cost of such studies relative to the benefit gained because the cost of generating molecular data is rather high. As stated earlier, generating such data without posing specific questions does not seem worthwhile. Second, from our own studies we do know that different molecular marker systems give different assessments of genetic diversity (Virk et al. 1999), as does the position of markers in the genome (Parsons et al. 1997).
Duplicate accessions
A rice core collection should contain, with the minimum of repetitiveness, a good representation of the diversity of O. sativa. But we know that the IRGC has duplicate accessions, although we cannot determine how many with accuracy. To make a core collection useful, we should try to avoid as far as possible having any duplicate accessions.
How easily can duplicate accessions be identified and by which methods? An obvious first start is the use of variety names or similarities of names to identify possible duplicates, although even with identical names this is by no means unequivocal, as our field work with farmers' varieties in the Cagayan Valley of northern Luzon, Philippines, has proven. Morphological comparisons may provide further insight but again cannot be used with certainty. The use of molecular markers has often been advocated in this respect, and we have shown that duplicate and suspected duplicate accessions can be identified using RAPD analysis (Virk et al. 1995b). What makes this approach less feasible is the number of primer combinations required to achieve a high degree of confidence in designating duplicate accessions, and also the magnitude of the overall exercise that could amount to analysis of several thousand accessions. Therefore, we must consider carefully whether the costs of duplicate accession identification outweigh the benefits that would accrue from such an operation.
The size of a core collection
With a rice collection of more than 105 000 samples, what should be the size of the core? Even at 10%, as suggested by Brown (1989a), the size would be larger than the whole collections of some crop species. Its relatively large size would still present difficulties for management and evaluation. A core of 10 000-15 000 accessions would unlikely bring the benefits of a core commensurate with the effort required to identify those accessions in the first place. Even 3000 accessions as suggested by Brown (1989b) for large collections would still be a large amount for evaluation with DNA markers or other sophisticated approaches. Lawrence et al. (1995a, 1995b) have argued that the actual numbers of individuals needed for genetic conservation purposes are far lower, by an order of magnitude, than we might assume. They do not, however, advocate having a germplasm collection of just 172 individuals (the number of individuals they calculated to represent the diversity of a species under specified circumstances). The implication, however, is that a core collection for rice may not require several thousand accessions. This means that selection of the actual accessions should be based on several criteria and, regrettably, some of the essential data are not available.
The size of the core collection should also clearly reflect users' needs. We believe that the size should be variable and depend on need. Several levels of the core could therefore be defined, for example, 250, 500, 1000, 5000, 10 000. This is true for the two basic functions we have defined for a core collection of the IRGC: (1) to facilitate security back-up of the collection (national systems with different storage facilities could be given the core collection of the most appropriate size), and (2) to provide an entry point to the whole collection.
For example, in searching for resistance to or tolerance for a particular stress, we might need to screen thousands of accessions. In contrast, for synteny studies, it would not be possible because of cost or logistics to handle more than a few hundred samples. Nevertheless, the working collection of Dr Courtois has about 300 accessions (for upland rice only), while the representative sets of accessions developed by Glaszmann et al. (1996) and Resurreccion et al. (1994) have 261 and 212 accessions, respectively. We therefore suggest that developing a core collection of 250 to 500 accessions should have top priority.
Strategies for developing a core collection
The accessions chosen for the core collection should represent the genetic spectrum in the whole collection (Brown 1989b). To construct a core collection for rice, we believe that several approaches could be used. The characterization data available for many registered accessions could be used to select samples fitting each of the characterization categories. Multivariate analysis has also been widely used to select individual accessions. Virk et al. (1996) used this approach successfully on morphological trait data from field trials to select a subset of highly diverse accessions with which to predict quantitative characters using RAPD markers. Again, the principal constraint is the incomplete data set across the germplasm collection, even though high for some individual characters. Such an analysis should provide an interesting picture of the collection's morphological diversity.
Van Hintum (this volume) has outlined different ways of constructing a core collection, and other examples of how this has been achieved for several species are reported in this volume. As we emphasized earlier, the isozyme classification of O. sativa varieties is an important basis for using germplasm in rice breeding. We believe that these data are very important for helping build a core collection, even though the data set is incomplete. We know that the geographical distributions of isozyme groups is uneven (Glaszmann 1988). One approach worth investigating would be to estimate the isozyme group frequencies by country, by studying a random sample from each country. The advantage of this approach is that (1) the information is already available for a few countries, and (2) the country of origin is the most frequently available passport data, although exactness cannot be guaranteed. The countries represented in the genebank collection would then be re-sampled according to these frequencies and the weight given to each isozyme group by logarithmic or proportional sampling (Brown 1989b) would be used to form the core collection. By "weight", we mean the size of the sample per category, and the possibility of taking into account a second evaluation of the genetic diversity with a DNA marker, such as microsatellites. Nevertheless, because of data incompleteness and some inconsistencies, an approach with mixed strategies is most likely to have the best outcome.
Conclusions
The genebank data management system, the IRGCIS, permits identification and selection of germplasm by single or multiple criteria to suit users' needs, but not physically distinct from the whole collection. This list (or lists) would obviously be available to anybody as any other germplasm-related information. We should remember that the IRGC is held in trust under the auspices of FAO. IRRI has an obligation to provide back-up storage for national germplasm collections, although disposition of the germplasm accessions is not entirely at the discretion of the Institute. We have also reiterated the importance of a core for conservation purposes, as well as for genome research, where the sheer size of the collection makes study of all but a handful of the accessions, relatively speaking, unthinkable. This may change, however, as technological progress in genomics advances and handling larger samples becomes feasible. A core collection developed today will reflect our current understanding of the diversity of rice. Of necessity, a core collection will change in structure, and from the outset we should think of the core as having both a variable structure and variable content and size.
Evidence from other crops, particularly minor species or ones with quite small collections, has shown that developing a core has stimulated further research on and use of the germplasm. In comparison, the IRGC is huge and demands for its management are somewhat different. Because of the strategic importance of rice as a staple food for half the world's population, access to and use of germplasm in the IRGC are widely appreciated. IRRI's first obligation is to ensure the long-term availability of the germplasm in the collection. After all, the development of a core collection is not a substitute for sound conservation practices. Without those, the germplasm would no longer exist to underpin rice breeding efforts and remain a major component of food security strategies.
References
Appa Rao, S., C. Bounphanouxay, V. Phetpaseut, J.M. Schiller, V. Phannourath and M.T. Jackson. 1997. Collection and preservation of rice germplasm from southern and central regions of the Lao PDR. Lao J. Agric. For. 1:43-56.
Brown, A.H.D. 1989a. The case for core collections. Pp. 136-156 in The Use of Plant Genetic Resources (A.H.D. Brown, O.H. Frankel, D.R. Marshall and J.T. Williams, eds.). Cambridge University Press, Cambridge, UK.
Brown, A.H.D. 1989b. Core collections: a practical approach to genetic resources management. Genome 31:818-824.
Courtois, B. and M. Olofsdotter. 1998. Incorporating the allelopathy trait in upland rice. Pp. 57-67 in Allelopathy in Rice (M. Olofsdotter, ed.). Proceedings of the workshop on allelopathy in rice, 25-27 November 1996. International Rice Research Institute, Manila, Philippines.
Devos, K.M. and M.D. Gale. 1997. Comparative genetics in the grasses. Plant Mol. Biol. 35:3-15.
FAO and IPGRI. 1994. Genebank standards. Food and Agriculture Organization of the United Nations, Rome, and International Plant Genetic Resources Institute, Rome.
Frankel, O.H. 1984. Genetic perspectives of germplasm conservation. Pp. 161-170 in Genetic Manipulation: Impact on Man and Society (W.K. Arber, K. Llimensee, W.J. Peacock and P. Starlinger, eds.). Cambridge University Press, Cambridge, UK.
Fuccillo, D., L. Sears and P. Stapleton (eds.). 1997. Biodiversity in Trust: Conservation and Use of Plant Genetic Resources in CGIAR Centres. Cambridge University Press, Cambridge, UK.
Gale, M.D. and K.M. Devos. 1998. Plant comparative genetics after 10 years. Science 282:656-659.
Ghareyazie, B., N. Huang, G. Second, J. Bennett and G.S. Khush. 1995. Classification of rice germplasm. I. Analysis using ALP and PCR-based RFLP. Theor. Appl. Genet. 91:218-227.
Glaszmann, J.C. 1987. Isozymes and Asian rice varieties. Theor. Appl. Genet. 74:21-30.
Glaszmann, J.C. 1988. Geographic pattern of variation among Asian native cultivars (Oryza sativa L.) based on fifteen isozyme loci. Genome 30:782-792.
Glaszmann, J.C., R.N. Kaw and G.S. Khush. 1990. Genetic divergence among cold tolerant rices (Oryza sativa L.). Euphytica 45:95-104.
Glaszmann, J.C., T. Mew, H. Hibino, C.K. Kim, T.I. Vergel de Dios-Mew, C.M. Vera Cruz, J.L. Notteghem and J.M. Bonman. 1996. Molecular variation as a diverse source of disease resistance in cultivated rice. Pp. 460-466 in Rice genetics III (G.S. Khush, ed.). Proceedings of the Third International Rice Genetics Symposium. International Rice Research Institute, Manila, Philippines.
Guiderdoni, E., E. Galinato, J. Luistro and G. Vergara. 1992. Anther culture of tropical japonica × indica hybrids of rice (Oryza sativa L.). Euphytica 62:219-224.
Hamon, S., S. Dussert, M. Noirot, E. Anthony and T. Hodgkin. 1995. Core collections - accomplishments and challenges. Plant Breed. Abst. 65:1125-1133.
Jackson, M.T. 1995. Protecting the heritage of rice biodiversity. GeoJournal 35:267-274.
Jackson, M.T. 1997. Conservation of rice genetic resources - the role of the International Rice Genebank at IRRI. Plant Mol. Biol. 35:61-67.
Jackson, M.T. and R.D. Huggan. 1993. Sharing the diversity of rice to feed the world. Diversity 9:22-25.
Jackson, M.T., G.C. Loresto, S. Appa Rao, M. Jones, E.P. Guimaraes and N.Q. Ng. 1997. Rice. Pp. 273-291 in Biodiversity in Trust: Conservation and Use of Plant Genetic Resources in CGIAR Centres (D. Fuccillo, L. Sears and P. Stapleton, eds.). Cambridge University Press, Cambridge, UK.
Khush, G.S. 1995. Breaking the yield frontier of rice. GeoJournal 35:329-332.
Lawrence, M.J., D.F. Marshall and P. Davies. 1995a. Genetics of genetic conservation. I. Sample size when collecting germplasm. Euphytica 84:89-99.
Lawrence, M.J., D.F. Marshall and P. Davies. 1995b. Genetics of genetic conservation. II. Sample size when collecting seed of cross-pollinating species and the information that can be obtained from the evaluation of material in gene banks. Euphytica: 84:101-107.
Mackill, D.J. 1995. Classifying japonica rice cultivars with RAPD markers. Crop Sci. 35:889-894.
Parsons, B.J., H.J. Newbury, M.T. Jackson and B.V. Ford-Lloyd. 1997. Contrasting genetic diversity relationships are revealed in rice (Oryza sativa L.) using different marker types. Mol. Breed. 3:115-125.
Peng, S., G.S. Khush and K.G. Cassman. 1994. Evolution of the new plant ideotype for increased yield potential. Pp. 5-20 in Breaking the Yield Barrier (K. Cassman, ed.). International Rice Research Institute, Manila, Philippines.
Pham, J.L. 1991. Genetic diversity and intervarietal relationships in rice (Oryza sativa L.) in Africa. Pp. 55-65 in Rice genetics II. Proceedings of the Second International Rice Genetics Symposium. International Rice Research Institute, Manila, Philippines.
Resurreccion, A.P., C.P. Villareal, A. Parco, G. Second and B.O. Juliano. 1994. Classification of cultivated rices into indica and japonica types by the isozyme, RFLP and two milled-rice methods. Theor. Appl. Genet. 89:14-18.
Second, G. 1991. Molecular markers in rice systematics and the evaluation of genetic resources. Pp. 468-494 in Rice. Biotechnology in Agriculture and Forestry, 14 (Y.P.S. Bajaj, ed.). Springer-Verlag, New York.
Second, G. and A. Ghesquière. 1994. Cartographie des introgressions réciproques entre les sous-espèces indica et japonica de riz cultivé (Oryza sativa L.). Pp. 83-93 in Colloque "Techniques et Utilisations des Marqueurs Moléculaires," Les colloques de l'INRA 72. INRA, Paris, France.
SGRP. 1996. Report of the internally commissioned external review of the CGIAR genebank operations. International Plant Genetic Resources Institute, Rome, Italy.
SGRP. 1997. Report of the internally commissioned external review of the CGIAR genebank operations. Annex: Centres' responses to the external review. International Plant Genetic Resources Institute, Rome, Italy.
Vaughan, D.A. and M.T. Jackson. 1995. The core as a guide to the whole collection. Pp. 229-239 in Core Collections of Plant Genetic Resources (T. Hodgkin, A.H.D. Brown, T.J.L. van Hintum and E.A.V. Morales, eds.). John Wiley and Sons, Chichester, UK.
Virk, P.S., B.V. Ford-Lloyd, M.T. Jackson and H.J. Newbury. 1995a. Use of RAPD for the study of diversity within plant germplasm collections. Heredity 74:170-179.
Virk, P.S., B.V. Ford-Lloyd, M.T. Jackson, H. Pooni, T.P. Clemeno and H.J. Newbury. 1996. Predicting quantitative traits in rice using molecular markers and diverse germplasm. Heredity 76:296-304.
Virk, P.S., H.J. Newbury, M.T. Jackson and B.V. Ford-Lloyd. 1995b. The identification of duplicate accessions within a rice germplasm collection using RAPD analysis. Theor. Appl. Genet. 90:1049-1055.
Virk, P.S., H.J. Newbury, M.T. Jackson and B.V. Ford-Lloyd. 1999. Are mapped markers more useful for assessing genetic diversity? Theor. Appl. Genet. (In press).
Wang, Z.Y. and S.D. Tanksley. 1989. Restriction fragment length polymorphism in Oryza sativa L. Genome 32:1113-1118.
Yang, G.P., M.A. Saghai Maroof, C.G. Xu, Qifa Zhang and R.M. Biyashev. 1994. Comparative analysis of microsatellite DNA polymorphism in landraces and cultivars of rice. Mol. Gen. Genet. 245:187-194.
