«Biotechnology-Assisted Participatory Plant Breeding: Complement or Contradiction? PPB Monograph No. 3 Ann Mane Thro and Charlie Spillane 1 7 ...»
As regards unifonnity standards, double haploid Hnes oClandraces could allow phenotypically uniCorm varieties to be devcloped and maintruned by farroers.
Transgenic approaches to the reduction oC levels oC undesirable compounds may also be possible. Pioneer Hi Bred has developed the use oC genetic modification to reduce mycotoxin contamination of foods by incorporating Cumonisin-metabolizing transgenes ioto the plant's genomc.
SOURCE: J. Duvick (pers. comm.).
Biotechnology·Assisted PPB: Complement or Contradiction?
So far there has been hUl e exploration of wh ether farmers' 'descriptors' can be in tegrated \Vith germpl a sm descriptors or wi th cxisti ng lin ka ge maps as a startin g poin t for enhanci ng far merrc searcher coll a boration in plan t breed ing. Only researchers with a detailed knowledge of farmers' selcction criteria an d practices are likely to be able to relate th ese to cri teria usable by formal breeders or biotechnologists, ancl vice· versa. Ir farmers' seIection criteria change over ti me or vary from place to place, then these relationships, and the process of establishing them, may become complex o Non etheIess, as MAS enters the genomics and phenomics era, it is vital that th is ta s k be addre ssed.
Con ventional pIant breeding has typicaUy used phenotypic observations, someti mes backed by sophisticated statistical analysis, to sclect fo r improved germplasm in brccding popu lations. Althou gh th is a pproach is still valid, there are limitations to what can be achieved by phcnotypic sclection alone. Sorne agronomi cally u scful trBi ts are either very difficu lt to select for (a nd rnaintain) on t he basis of pheno typ c, or can not be selccted for on this basis alon e (e.g., yield). These traits show continuous phenotypic variation because they are controlled by se veral genes, the inclividual effects of which are relatively s mall (Yano a n d Sasaki, 1997 ).
This has made breeding for such traits di.fficult.
The use of molecular markers and genetic maps to select for genes rather th an for phenotype could, in theory, overcome many of the.
limi tations of convcntional breeding (Caetano-Anollés and Trigiano, 1997). These tools are aJready revolution izing breed ing th ro ugh the id entification of th e quantitati ve trait ¡oei (QTLs), the relatively large segments of DNA that underlie many key agronomic trai ts (Sm ith an d Beavis, 1996; Yano and Sasaki, 1997; McCouch et a l, 1997). A wide ran ge of markers and maps are now available (C a etano - Ano l h~s an d Trigiano, 1997; Xiao e t al, 1998; Ayres et a l, 1997; Blair and McCouch, 1997). rn addition, mo lecu lar maps are bei ng integrated with li nkage maps based on observable phenotypes (Yoshimu ra et a l, 1997). This will allow phcnotypic selection to be complemented by MAS fOl" traits of mterest. This approach could prove cost·effective in PPB programs usin g phenotypic selection for traits not casily selectcd for on this basis alone.
Sorne ficId -level practitioners find that farmers are at a disadvantage wh en attempting to identify and sdect effectively for useful genes found at low frequency in populations, particularly when the associated traits are hidden (J. Lenné, pers. comm.). By identifying and mapping molecular markers, formal breeders and biotechnologists can help select such genes.
Finding the loei of these traits in one crop provic1es guidance to whcre they might be in olhcr related c rap species (e.g., Kowalski et al, Biotechnology as a Set 01 Tools lor Fonnal and Infonnal Plant Breedíng 1994; Lin et al, 1995; Ming et al, 1998). The c10se functional and evolutionary relationships between many resistance genes is making it easier to search for them in germplasm collections (e.g., Leister el al, 1996).
A crucial question is whether individual molecular markers can be 'translated' into visual markers or other easily selectable markers, allowing MAS to be applied at field lcvel by formal breeders or farmcrs.
For instancc, a single gene that provides a visible morphologica1 marker such as red pigment color (¡.e., a more penetrant version of the currently available anthocyanin Le marker) could conceivably be linked as a reporter, via transgenic techniques (T-DNA tagging) and/or molecular marker-assisted backcrossing, to a major allele for a hard-to-see trmt such as drought tolerance or resistance to a cyclic pest. This could be particularly useful in open-pollinated populations. Even in ayear when the stress is absent, the red pigrnent from the marker would help the farmer identify stress-tolerant plants and save enough seed from them to maintain the trait in the population at a level sufficient to stabilize yearto-year performance. However, while reporter genes such as the GUS and GFP are routinely used to great effect in laboratory rescarch, very few such genes are yet available for use at fieId level.
If markers can be linked to major agronornic alIeles, the allele itseIf does necessarily have to be visually selectable. Use of selectable rnarkcrs (such as herbicide resistance genes) could allow farmers to select for the allele. Howevcr, at least at the current level of technology development it is questionabIe whelher the cost of such an approach would be justified by the bcnefits (M. Gale, pers. cornrn.).
Developing molecular markers for QTLs is important in improving selection for phenotypic traits. QTL analysis looks at the underlying genetic basis of such traits (Ribaut and Hoisington, 1998). Consequent1y, there is likely to be room for considerable interaction between researchers and farmcrs, who will need both to identify desirable traits and to test gcrrnplasm enhanced by this means. Sorne commentators believe that, in breeding for quantitative traits, farrner participatory selection, eithcr among finished varieties or,within segregating populations, could replace MAS, since both end up with the same thing-a product in which you can 'see' or otherwise experience the desired rcsults. However, this seems unlikely, since quantitative traits have traditionally been difficult for breeders to select for on the basis of phenotype, even with the support of complex biometrica1 and genetic analyscs. The reality may tie somewhere in between, with farmer selection criteria proving a useful cornplemen tary source of information for DNA marker-based se1ection. and vice-versa.
The development of suitable populations for mapping. as a prelude to the development of markers, is best done through collaboration between
farmers a nd locally basecl plant brccclers (S. Hu ghes, pers. comm.l. with regional or international inputs where necessary. Fregene (pers.
cornm.) suggests that a team of brceders, molecular gen eticists, and farmcrs could handle perh a ps four breeding populations a t a time.
In the n ear tenn. mo lecular mar kers might facilitate PPB through th e gene rat ion of trait-enriched populations a t an early stage of th e selcction process. Molecular markers can be uscd to increase the frequ ency of certain tra its, s uch as QTLs for drou ght tolerance (Ri bau t et al, 1996, 1997), or of de s irable individu als in an otherwise variable popu lation, creating an 'cnriched' popula tion for furth cr sclcctio n by farmers (S. Bee be, pers. comm.). MAS can enhance total gene tic gain and th e choices available to farmers for difficu lt-to-select tra its, particularly tolcrance or resistance to biotic or a biotic stresses th at may require s pecial stress environ ments to be fully expresscd, and traits that require slow a nd /or costly sam pling method s, such as cookin g q uality or photosynthe tic rate (M. Lee, 1998).
'For crops in which molecular mapping is at an advanced stage, where the underlying gene tics of important agrono mic traits are becoming in creasi ngly c1ear, it m ay be possible to devclop sets of markers that could act as 'sieves' to enrich germplasm population s for linked agronom ic traits. The use of these molecular sieves would help reduce breedin g popu lations to a manageable level (M. Fregene, pers.
comm.). The chanccs of a [armer crea ting desirable material by crossin g two interesting parents would be increased, since the amount of 'j unk' or apparently useless d iversity (M. Loevinsohn, pers. comm.) would have becn reduced by 10 times or more (S. Beebe, pcrs. cornm.). This could, it is though t, change fann ers' perceptions of the costs and benefits of bccoming ¡nvolved in early generation selection efforts in PPB. As Witcombe et al (1996) found in the Chitwan Valley of Nepal, farmers' lack of interest in selecting for early segregating popula tion s is a barrier to lheir participation in the early stages of crop improvement.
In s uch situations th ey find themselves being asked to dca! with too wide a ran ge of prototypes of too low a quality.
Farmers participating in research want to see re sults fast (B.
Visscr, pers. co mm.) and often express a sense of urgency (e.g., Thro et al, 1997). The use of MAS requires additional time early in the research process, when the m arkers are first developed (this takes 2 to 4 years, dependi ng on the co mplexity of the trait and previous knowledge). This time · lag is 'anathe ma' to many farmers involved in pa rticipatory re search (J.K. Lynam, pers. comm.). Yet one of the main a ttraction s of biotechnology lO conventional breeders is that, once the tool dcvelopment stage i5 over, it can greatly speed up the breeding cycle. As more markers become available over time as a result of genome mapping a nd sequencing efforls, the 'tool development' time-lag is likely to s horten.
Biotechnology as a Se! of Tools for Fonnal and Informal Plant Breeding
In addition, discoveries made in comparative mapping have shown that markers from closely related (e.g., rice and wheat) or.e ven distantiy related (e.g., dicot and monocot) species can be successfully used across species (Paterson et al, 1996). This has greatly increased the diversity and genomc coverage oC the markers now available, reducing both their costs and the time required to apply them. Costs will probably continue to decrease as molecular marker assays become cheaper per unit of information gained (Xie and Xu, 1998). In the longer term, technology spillovers from human genetics (notably the human genome project) should further increase the potential of DNA technology for crop improvement, leading to even more favorable cost:benefit ratios. However, this depends on sufficient public-sector funding being made available for technology adaptation and dissemination (Smith and Beavis, 1996).
DFID's Plant Sciences Research Programme is establishing a project in the semi-arid regions of India and Nepal that will combine PPB with the use ofmolecular marker techniques in rice (J.R. Witcombe, pers. comm.).
The project will evaluate the participatory approach. which will be applied to a range of crosses mostly involving the popular variety Kalinga III as one parent. The end products from the crosses wiU be tested using molecular markers to identify linkage blocks representing genomic regions preferred by farmer~ or producing the best results in specific environments. Progeny from a wide cross between the Asian and African rice species Oryza sativa and O. glaberrima will also be evaluated, so that useful genomic regions of O. glaberrima can be introgressed into the sativa varieties preferred by fanners. QTLs for root growth and drought resistance are being introduced into Kalinga III through MAS. The results of this project should shed more light on the usefulness of molecular markers in PPB projects.
Optimizing local genotype x environment interactions
Sorne PPB programs promote the use of a decentralized farmer selectionbased approach to the development of germplasm specifically adapted to different micro-environments (Ceccarelli and Granda, 1996; Ceccarelli et al, 1991, 1994; Simmonds, 1991). These practitioners believe that selection for specific adaptation to local conditions will result in varieties that require reduced levels of inputs and are more Tobust in the stress-prone environments typicalLy used by resource-poor farmers.
This renects a long-standing debate among plant breeders as to whether or not high genotype x environment interactions can be usefully exploited to develop germplasm adaptation to marginal or heterogeneous environments (Gauch and Zobel, 1997). The specific adaptation approach is considered by sorne to stand in opposition to the centralized development of varieties exhibiting brcad adaptation to a wide range of environments (Ceccarelli, 1989; Link et al, 1996). For cost-benefit reasans, most centralized breeding has successfully concentrated on developing varieties adapted to large geographic areas.
Biorechnology·Assisted PPB: Complement o,. Conlradiction?
Many widely adapted varieties have been bred to exh ibit low G x E intcractions for agronornic traits a nd are very successful in homogeneous high -potential cnvironrne nts in whi eh fertilizers and irrigation are used. It has, however, been suggestcd that the suecess of widcly adapted eommercial1y bred varietics is du e less to the inputs they receive than to the amoun t of breeding and testing invested in th eir development (D. Duvick, pers. cornrn.). Sorne widely adapted varieties have becn developed for srnall-scalc farrncrs' conditions, where they perform well despite the absence of buffcring inputs. Experience with rice breeding in South America suggests that rice varicties bred for \Vide geographic adaptatíon are lised by resource-poor farmcrs becausc th ese varieties adapt as well to the extremes occurring under farmcrs ' managcment regime s as they do to the variability found across geographicallocations. For example, the varieties yield weU even when sown too late because of com peting requiremcnts for labor (L. Sanint, pers. comm.) One of the problcms in breeding foc stressful and unpredictable envlronments is me reduced heritability of complex traits su eh as yield in such environments (Cecearelli et al, 1991). MAS has become a factor in the high versus low G x E debate (Kang, 1990). It now allows breedecs to distinguish between low QTL x E a nd high QTL x E loci, QTL x E bcing analogous to G x E inte ractions (Hoisington et al, 1996; Fry et al, 1998; Stratton, 1998; Palerson et al, 1991; Stuber el al, 1992;