Plant Breeding: Breeding Techniques

A. Floral Biology

The oil palm is cross-pollinated and monecious bearing a staminate or pistillate inflorescence in each subtending leaf or frond (Plates 1D, 1E). Inflorescences, which are compound spikes, begin to appear on young palms when they are about 27 months old from seed sowing, which is also the same length of time from inflorescence initiation to its emergence from the sheath in mature palms (Henson 1998). Sex differentiation occurs at the 14^th month from inflorescence initiation. Stress conditions at this time will induce staminate inflorescence while at the 23^rd month will result in inflorescence abortion and non-appearance of inflorescence on the subtending frond. There should be potentially two to three inflorescences produced per month corresponding to the frond production rate. Pollen viability and stigma receptivity last about 5-6 days and 3-4 days in the field respectively. The main pollination agent is the pollinating weevil, Elaedobius camerounicus, which occurs naturally in West Africa and Latin America but was only introduced to the Far East in the early 1980s. Controlled-pollination for breeding and commercial hybrid seed production involves the following protocol:

1. Pistillate and staminate inflorescences are isolated 7-10 days before receptivity/anthesis using terelene/paper/canvas bags with clear plastic windows.
2. Pollen can be stored as oven-dried pollen in the freezer for about 6 months or as freeze-dried pollen in vacuum-sealed ampoules in the freezer for about 24 months.
3. Controlled-pollination is effected by puffing a 1 pollen: 10 to 20 talc mixture through a perforation in the plastic window (Plate 1F).
4. The bag is removed after about a month and the fruits left to mature and ripen in about 5 months.
5. The controlled-pollinated seed from the depulped ripe fruit needs to undergo 40-60 days heat treatment at 37º-38ºC and 17-19% moisture to break its dormancy before it can germinate when remoistened to 21-22% (Plate IIA). These are the requirements for germinating dura seeds. For tenera seeds the corresponding moisture requirements are 20-21% and 27-28% respectively. Germination of the shell-less pisifera and thin-shelled tenera seeds is erratic (Arasu 1970) and in vitro (embryo rescue) germination is sometimes needed.

Plate I. D) Staminate inflorescence at anthesis, stage bagged inflorescence harvested for pollen collection.	Plate I. E) Receptive pistillate inflorescence.
Plate I. F) Pollination of bagged pistillate inflorescence.	Plate II. A) Germinated hybrid seedlings.

B. Breeding Plans and Selection Methods

The oil palm being a cross-pollinated crop borrows a number of breeding methodologies developed in maize breeding such as recurrent selection and topcross testing, but being a perennial tree crop also shares many similarities with animal breeding in the selection techniques such as the importance of sire or pisifera testing, simultaneous multiple trait or index selection, and the close temporal and genetic correspondence of commercial hybrid production with each cycle of breeding.

1. Recurrent Selection. Major oil palm breeding programs adopt one of two basic schemes; the modified recurrent selection scheme and the modified reciprocal recurrent selection scheme.

Modified Recurrent Selection Scheme. This scheme is practiced by most programs in the Far East (Fig. 2). It evolved from the initial use of the Deli dura as the commercial planting material to its subsequent use as a maternal parent in commercial tenera hybrid seed production. Pisiferas, introduced or bred, and being female sterile, are initially selected based on their tenera-sib performance in the tenera × tenera/pisifera cross or based on proven lineage. They are then progeny-tested (top-crossed) with a sample of phenotypically selected Deli dura mother palms. Proven pisiferas and the phenotypically selected duras are then used in the commercial mixed tenera hybrid seed production. Parents used for further breeding in dura × dura and tenera × tenera/pisifera crosses are also phenotypically selected although progeny-tested parents are also sometimes included. As parent selection often involves family with individual palm selection, this scheme is sometimes referred to as the family and individual palm selection scheme (Rosenquist 1990). But as selected parents are interbred in each cycle, Soh (1990) preferred to refer to it as a form of recurrent selection (Allard 1960). This scheme emphasizes and exploits general combining ability (GCA) effects.

The main advantages of this scheme are that more recombinant crosses can be made and tested and in saving time, space and effort from need of extensive progeny-testings. The main disadvantage is that the parents chosen for further breeding and the dura mother palms for hybrid seed production usually have not been hybrid progeny-tested. The key assumption that the additive or GCA effects expressed within population crosses (determined by family and individual means or combining ability analyses) will be reflected in the inter-population hybrid crosses, may be untenable (Soh 1987b, 1999; Soh and Hor 2000). The scheme may also lead to undue prolonged reliance on the Deli as the source of dura mother palms.

Modified Reciprocal Recurrent Selection. This scheme (Fig. 3) practiced mainly in West Africa e.g. Ivory Coast, Nigeria (Sparnaaij, 1969; Meunier and Gascon, 1972) and more recently in Indonesia (Lubis et al 1990) is adapted from the reciprocal recurrent selection method developed in maize breeding (Comstock et al.1949). It exploits both GCA and specific combining ability (SCA) effects. The choice of this scheme presumably arose from the observations and results that crosses of the Deli duras with African teneras or pisiferas tended to give heterotic yields.

This attractive scheme has three apparent advantages. Firstly, the selfed parents selected for commercial hybrid production and for further breeding are both based on progeny-tests (from the dura and tenera grandparents) of the prospective commercial inter-population hybrids. Secondly, the scheme has two distinct phases. The “within hybrid improvement phase” allows shorter term commercial exploitation of the best test-cross and its improvement by recurrent selection within the selfs/sibs of the selected parents. The “recombinant phase” involving outcrosses, which is run in parallel similarly, allows accumulation of favorable alleles (additive and non-additive) and maintenance of genetic variability for sustained longer term improvement. Thirdly, the commercial hybrid can be reproduced using the duras and pisiferas from the selfs/sibs of the parents (which have been planted concurrently as the progeny- tests) once the inter-population hybrid test results are known (Jacqmard et al. 1981). Its main disadvantage is the large program size requirement. To produce three to four million hybrid seeds from the reproduction of the top 15% of the crosses, about 500 crosses and 180 parental selfs have to be planted over 600 ha and evaluated over 15 to 25 years (Soh 1999).

Breeding Progress. In the reciprocal recurrent selection scheme, the original cross from which subsequent breeding improvements are made can be reproduced to be the standard control cross for measurement of breeding progress achieved. For the first cycle of the reciprocal selection scheme, Gascon et al. (1988) reported a 18% oil yield improvement by selecting the top 2.8% of the 529 hybrid crosses tested for reproduction as commercial hybrid seeds. From early trial results of the second cycle (within hybrid improvement) hybrids, Nouy et al. (1990) predicted a 10-15% oil yield increase which was subsequently confirmed (Cochard et al. 1993). Lubis et al. (1990) suggested a 25% oil yield increase by selecting the top 8% of the 500 hybrids tested in the first cycle of the Indonesian Oil Palm Research Institutes’s (IOPRI) reciprocal recurrent selection program. Progeny-test results for the recombinant phase i.e. recombinant parents or wide crosses, are still unavailable. An important consideration for this phase is whether to use the selfs of the recombinant parents for the production of commercial hybrids which would result in higher within hybrid variability or to await another generation of selfs in order to produce more uniform hybrids. In the study of Durand-Gasselin et al. (1999) on 2, 3 and 4-way hybrid crosses, the increased within cross variability for production traits was minimal but these crosses involved related parents rather than outbred or recombinant parents. Also in this study, as reported earlier by Nouy et al. (1990), breeding progress was achieved mainly through exploitation of GCA effects although SCA effects might have played a small part.

It is difficult to estimate the breeding progress made by the modified recurrent selection scheme with confidence due to the lack of common link crosses across experiments and common base reference populations. Nevertheless, some indications can be obtained by piecing together some experimental results reported.

Hardon et al. (1987) estimated 15% oil yield progress per generation from breeding within the Deli dura but did not translate this in terms of improvement in the corresponding tenera hybrids. Lee et al. (1990) attempted this by comparing the mean performance of the tenera hybrid progenies of third and fourth generation Deli duras when crossed to the same pisiferas but the trials were planted at different periods although at similar localities. A 6.4% increase in oil yield was obtained mainly through improved oil content. Similarly, Rajanaidu et al. (1990) reported a 7% oil yield difference with tenera hybrid progenies of their first and second generation Elmina Deli duras crossed to similar pisiferas planted in the same trial. Based on the progeny-test results, selecting the best pisifera (top 15%) would give a 12% improvement (Lee and Yeow 1985; Hardon et al. 1987). Owing to the limited number (<10) of pisiferas usually available and tested, such stringent selection would seriously restrict commercial hybrid seed production. The top 30-50% selection would have been more likely in which case an improvement of around 5% would be achievable (Soh 1986). By combining the expected progress made in both the dura and pisifera parents, a gross estimate of the 10-15% breeding progress made using the modified recurrent selection scheme would not be unreasonable.

As evident, it is difficult if not futile to compare objectively the relative efficiencies of the two selection schemes. The choice depends very much on one’s bias, objectives (i.e. short or long term gains), and stage of the breeding program. We prefer a combined approach, modified recurrent selection to develop relatively unimproved populations to reduce the final size of recombinant families to go into the modified reciprocal recurrent selection program. Most mature oil palm breeding programs in the Far East are beginning to adopt the latter technique.

2. Other Breeding Methods

Backcross Breeding. This has been used in introgression work to improve a particular genotype or population. The Dumpy trait has been introgressed into the AVROS population to reduce height (Soh et al. 1981; Lee and Toh 1992). The backcross approach is also used to introgress the better oil quality traits from the Nigerian guineensis and the American oleifera materials and to restore the yielding ability and fertility of the recurrent advanced breeding parents respectively (Sterling et al. 1988; Le Guen et al. 1991; Soh et al. 1999; Sharma 2000a).

Recombinant Inbred Lines. The development of recombinant inbred lines have been proposed and attempted in oil palm (Lawrence 1983; Pooni et al. 1989). This would involve the production of dura inbred line cultivars (as the tenera is a heterozygote and the pisifera homozygote is female sterile) with oil yields superior to the best thin-shelled tenera hybrid. This is unlikely. Other disadvantages would be the possible loss of desirable genes linked to yield in the infertile gametes/zygotes in the inbreeding process, and longer time to produce a commercial variety (Soh 1987b). However, the recombinant inbred line approach combined with single seed descent may be considered for development of inbred parents for hybrid production. This will put oil palm hybrid seed production on similar footing as hybrid maize production in the generation of hybrid combinations for testing. The availability of inbred lines particularly isogenic lines will be a boon to basic studies on oil palm genetics (quantitative, molecular, physiological, biochemical) and a host of other fundamental studies and applications.

Breeding for Clonal Propagation. This would involve treating the oil palm as a clonally propagated crop such as potato, cassava, rubber, sugar cane. The breeding approach for a clonally propagated crop starts off by generating a segregating hybrid population with high mean and progeny variability. Single plant selection is followed by cycles of increasing scales of testings of the clonal offsprings, in later stages involving multi-location trials (Brown et al. 1988; Kawano et al. 1998; Tan 1987; Simmonds 1979). Currently we do not foresee this happening in oil palm as the feasibility of repeated cycles of recloning ramets (clonal offspring) is still uncertain (Soh 1998). There will be further discussions on this subject in a later section.

3. Index and BLUP Selection

In oil palm breeding, as in animal breeding, the choice of breeding parents is crucial. Firstly the number of parents that can be accommodated in the program is limited because of its perennial tree crop nature. Secondly, the hybrid production cycle is closely linked to the breeding cycle. Knowledge of the breeding values of the parents and multiple trait selection methods will help in the choice of the parents. The selection index and best linear unbiased prediction (BLUP) methods derived from application of linear statistical models which are usually associated with animal breeding (Henderson 1984) are now more commonly used in plant breeding (White and Hodge 1989) including oil palm breeding.

Selection Index. Multi-trait selection index and index selection incorporating plot and family information were found to be useful in choosing candidate palms (ortets) for cloning (Soh and Chow 1989, 1993; Soh et al. 1994a; Baudouin et al. 1994). The same approach can be used in selecting parents for breeding.

BLUP . The combining ability approach has been the usual approach adopted in selecting parents in oil palm breeding (Soh et al. 1999, Breure and Konimor 1992). Many oil palm breeding trials, however, tend to be highly unbalanced in mating as well as experimental designs due to the constraints of the crop. The combining ability approach does not readily allow objective integration and comparison of genetic information from across unbalanced experiments. In this regard, the BLUP method developed for cattle breeding, which have similar constraints, confers the following advantages (Soh 1999):

• it can handle highly unbalanced mating and experimental designs
• it can utilize information from a number of experiments even without standard check varieties
• it can utilize information from other relatives or sources
• it allows a more directed approach in breeding, breeding towards an ideal or optimum genotype and breeding towards a target environment
• it allows refinement as better genetic information becomes available
• it can handle multiple traits simultaneously

Essentially, the BLUP approach removes environment and replicate effects as “nuisance” fixed effects in order to estimate the random genetic effects e.g. breeding values or additive genotype. Soh (1994) pioneered the use of BLUP in oil palm breeding by estimating the breeding values of pisifera parents from a very unbalanced set of tenera hybrid trials and predicted increased usage of this technique. BLUP has now been used not only to estimate breeding values of parents but also in predicting hybrid performances in oil palm (Purba et al. 2000) as have been done in maize (Bernado 1994, 1995, 1996) and sugarcane (Chang and Milligan 1992a,b).

C. In Vitro Methods

1. Commercial clonal propagation. Commercial oil palm planting materials being a mixture of hybrids from non-fully inbred parents have considerable between palm genetic variability arising from between and within family genetic variabilities depending on the relatedness and inbred status among dura and pisifera parents. Individuals could vary by more than 30% of the mean yield of the hybrids (Hardon et al. 1987; Meunier et al. 1988). Although the differences might not be entirely genetic, nevertheless they provided the rationale and impetus in the 1960s to develop the in vitro propagation of the oil palm, which has no natural means of vegetative propagation. The oil palm, being a monocot and a perennial tree, was considered a recalcitrant species for in vitro propagation. However, with concerted research efforts the first in vitro oil palms were achieved by the mid 1970s (Jones 1974; Rabachault and Martin 1976; Lioret and Ollagnier 1981). Rohani et al. (2000) and Krikorian and Kann (1986) have reviewed the details and historical development of the oil palm tissue culture techniques.

However, large-scale commercial propagation and plantings of proven tenera clones have yet to take-off despite more than two decades of research and development (R&D) work involving the setting up of more than 20 tissue culture laboratories worldwide with expenditures running into tens of millions in clonal propagation for trial and pilot commercial test plantings. For successful large-scale commercial propagation of proven tenera clones, a number of critical issues need to be resolved including somaclonal variation, cloning efficiency, ortet (parent palm of clone) selection efficiency, feasibility of recloning and the liquid suspension system, clonal field evaluation requirements and expected genetic progress.

Somaclonal Variation. The first issue is the specter of somaclonal variation manifested as abnormal flower and fruit development on otherwise apparently normal palms. In the pistillate flower the vestigial androecium develops into supernumery carpels forming a “mantle” over the developing fruit (Plate IIB) similar to the rare genetic mantled or “poissoni” fruit variant found in nature (Hartley 1988). The gynoecium may be fertilized or develop parthenocarpically. The feminized (carpelloid) flowers of the abnormal staminate inflorescence do not bear pollen. Severely mantled fruit bunches tend to be parthenocarpic leading to bunch abortion and palm sterility (Corley et al. 1986; Paranjothy et al. 1990). Palms bearing slightly to moderately mantled bunches can recover to normal bunch bearing (Ho and Tan 1990; Durand-Gasselin et al 1995; Duval et al. 1997). Susceptibility to the abnormality varies between and within clones and the risk appears to increase with extended culture, the earliest batches of ramets (clonal plants) tended to be mantle free (Paranjothy et al. 1995b). The mantled fruit phenomenon was experienced by all laboratories. As some clones can be severely mantled with susceptibility apparent only at the fruiting stage (2-3 years after field planting), confidence in mass propagation was seriously eroded. Speculations were made on the physiological, epigenetic (gene expression) and genetic bases, including the involvement of cytoplasmic or organelle DNA e.g. mitochondria, of the mantle somaclonal variant (Corley et al. 1986; Soh 1987a; Jones 1995; Paranjothy et al. 1995b). Mantling was found to be apparently sexually transmissible (Rao and Donough, 1990) but this still did not rule out the epigenetic basis. Use of fast growing instead of nodular callus (Duval et al 1988; Marmey et al. 1991) and/or phytohormones such as cytokinins (Agamathu and Ho 1992; Besse et al. 1992; Jones et al. 1995, Ng et al. 1995) in the proliferation stage was suggested as the culprit but this soon proved untenable as laboratories which did not use either still produced mantled ramets.

Plate II. B) Parthenocarpic mantled fruits.

As empirical approaches to research the causal factors through protocol manipulations appear cumbersome, if not imprecise, the alternative molecular approach was initiated. Research on the molecular basis of the mantle somaclonal variation using molecular marker techniques e.g. protein, isozymes, restricted fragment length polymorphism (RFLP), random amplified polymorphic DNA (RAPD), amplified fragment length polymorphism (AFLP), differential display, has been contracted out to the laboratories of six international centers of excellence by MPOB besides in-house efforts (Cheah et al. 1999; Chowdhury 1995; Paranjothy et al. 1995a; Sharifah et al. 1999). The CIRAD (Centre de Cooperation Internationale en Recherché Agronomique pour le Development) group in Montpellier, France has also been involved in similar research (Marmey et al. 1991; Rival et al. 1997; 1998a,b; Jagligot et al. 2000; Morcillo et al. 2000). Preliminary findings implicated an epigenetic basis (also possibly a genetic basis), mediated through methylation perhaps involving retro-transposons, affecting the expression of homeotic (flowering) genes (Rival et al. 1997; Shah et al. 1999; Rajinder et al 2001). Methylation has been implicated in somaclonal variation of other crops (Phillips et al. 1994; Tsaftaris and Polidorus 2000). The development of a diagnostic tool is the ultimate objective although it looks elusive at the moment as putative molecular markers isolated so far appeared to have specific rather than general applicability on clone or culture basis. Meanwhile, an empirical approach would be to look for genotypes which are amenable to cloning and resistant to somaclonal variation and this could be subsequently followed up with the development of molecular markers or genes for these traits. Many laboratories are now reporting significantly reduced mantling rates in packages of clones planted (Table 2) and have attributed them to use of more suitable or improved protocols reinforced by stringent culture selection (Maheran and Abu Zarin 1999; Ho 2000, Soh et al., 2001). Similar experiences have been reported in banana micropropagation (Kikorian 1994).

Table 2. Mantling incidence rates in clones. Source: Soh et al. 2001.

^zA clone with any of its ramets expressing the mantle somaclonal variation is considered mantled.
^yRamets = clonal plants or members of a clone.
^xOrtet = parent tree of clone
^wRepeated cross of superior family

Cloning Efficiency. The second issue is the inefficiency of the cloning process. Although callogenesis rates (100% palm-wise, 17% explant-wise), and percent embryogenic palms (80%) appear not to be process limiting, embryogenesis rates on callus cultures varied from 0.2-36%, averaging 4% per ortet (Table 3). With about 2000 leaf explants per ortet usually sampled, this translates to 12 leaf explants with embryos obtained per ortet on average with some having one or none and others having >120. More importantly is the capability of active proliferation and thus mass propagation of the embryos obtained. Generally about half are capable (Soh et al. 2001; Wooi, 1995), while the remaining will either persist in producing shoots at the detriment of embryo proliferation or fail to develop further. In the final analysis only about 18% of the ortets cultured can provide cultures amenable to mass propagation. Prospects of further improvement in efficiency appear limited although it may be possible by coaxing the cultures through more drastic protocol manipulations. Most laboratories are wary to deviate significantly from their established protocols for fear of increased risk of somaclonal variation and the additional tedious protocol field-tests encumbered.

Table 3. Oil palm cloning and recloning efficiencies, gelled-culture system.
Source: Soh et al. 2001.

^zOrtets = parent palms of clones; ramets = clonal plants

Ortet Selection Efficiency. Selecting superior palms or ortets for cloning is an inefficient process. In the breeding of many crops (Allard 1960; Simmonds 1979) and in clonally propagated crops e.g. potato, cassava, rubber (Brown et al. 1988; Kawano et al. 1998; Tan 1987), single plant selection especially for yield from a mixed seedling-derived population has been inefficient if not futile. Soh (1986) using data from trials of an advanced tenera hybrid (i.e. Deli × AVROS) population estimated that selecting the top 5% of the palms for cloning would give clones yielding 10-15% more than the mean of the hybrids and that clones yielding more than 30% were unlikely because of the low heritability and genetic variability. Also the better clones could only be identified after clonal testing. If the clones were to be compared to the improved hybrids, which would be available by the time the clonal test results are available, the advantage would be much reduced. Corley et al. (1999) was partial towards the higher estimate citing the higher heritability values obtained by Baudouin and Durand-Gasselin (1991) which were subsequently found, when the palms were older, to correspond to Soh’s estimates (Cochard et al. 1999). The elucidation of the low heritability for oil yield in these advanced tenera hybrid populations implied that the same experiences in other crops mentioned earlier would apply in oil palm (Soh 1998). This has been further supported by the clonal trial results of Donough and Lee (1995), Cochard et al. (1999), and Soh et al. (2001). In our first trial (Table 4), although four out of 12 clones derived mainly from highly selected ortets of Deli × AVROS lineage exceeded the Deli × AVROS hybrid control in mean oil yield (1992-2000), only two clones exceeded by more than 10% (111% and 118%). In our second trial (Table 5), which was a trial of clones derived from the embryos of a reproduced superior Deli × (Yangambi × AVROS) hybrid i.e. each clone represented a random ortet in the superior cross, the mean oil yield (1994-2000) was exceeded only substantially by three clones out of 10 (108%, 108%, 113%). Also from the second trial, for oil yield, the genetic variability between palms within the cross, estimated from the between clone variance (genetic coefficient of variation, CVg = 6%), was much smaller than the environmental variability between palms (CVe = 20%), estimated from the within clone variance. This implied that the genetic differences between palms would have been masked by the environmental effects. These results reaffirmed the inefficiency of ortet (individual palm) selection based on its phenotype and the dependence of good clonal tests to pick out the infrequent outstanding clones especially when clones tended to exhibit GxE effects (Lee and Donough 1993; Corley et al. 1995; Soh et al. 1995).

Table 4. Trial BCT4-89; yield performance of ortet clones^z. Source: Soh et al. 2001.

^zOrtet clones = clones derived from selected adult palms
^yO/B = oil in fresh bunch by weight
^xOil yield = FFB × O/B, mean for period
^wDxP (AVROS) = a dura × pisifera hybrid of superior lineage

Table 5. Trial BCT9-91. Yield performance of embryo-derived clones of a reproduced superior dura × pisifera (Yangambi-AVROS) cross. Source: Soh et al. 2001.

^zO/B = oil in fresh bunch by weight
^yOil yield = FFB x O/B, mean for period

Alternative Approaches in Clonal Propagation. A number of analyses incorporating secondary trait, plot and family information in selection indexes had been carried out to improve ortet selection efficiency. Family information was found to be most important (Soh and Chow 1993; Baudouin et al. 1994, Soh et al. 1994a), lending support to two other alternative approaches to oil palm cultivar development using clones proposed earlier by Soh (1986, 1987a). Instead of cloning elite tenera palms which appeared inefficient if not elusive, we could clone a sample of the reproduced seedlings/palms of the best proven cross, or clone one or both parents of the best proven cross which could then be used to produce semiclonal (one clonal parent and one sexual parent) or biclonal (both clonal parents) hybrid seeds.

The rationale for the two other alternative approaches were as follows:

• Both approaches would capture the best family oil yield performance, which would be about 10-15% above the mean of the families, as the best family would have been selected from statistically designed trials with replicated plots as compared to selecting individual palms in the original elite tenera palm cloning approach.
• The cloning best family approach would provide a larger clone and explant base to circumvent the ortet cloning inefficiency. In the case of seed embryos and seedlings, the early and the relative ease in cloning would be advantageous.

In the clonal hybrid seed approach, risk of clonal abnormality would be minimized as fewer ramets would be needed to serve as clonal parents and could be obtained from the early cultures that are usually less susceptible to mantling. Also deleterious recessive mutations could be masked or self-selected out in the sexual seed production process.

Commercial-scale plantings from cloning a sample of the best family approach or from mass selected ortets are increasing because of improved confidence in achieving low mantling rates with the refined laboratory protocols and good yields achievable (Khaw and Ng 1997; Maheran and Othman 1999; Wong et al. 1997,1999a). Trials of biclonal and semi-clonal hybrids are giving encouraging results in terms of negligible mantling and good potential yield (Sharma 2000; Veerappan et al. 2000; C. W. Chin pers.comm. 2001) and commercial clonal hybrid seeds should be available soon.

Commercial production of proven elite clones (Plates IIC, IID), the “holy grail” of oil palm clonal propagation effort, however, has yet to be achieved. The two alternative approaches can only recapture the best family not the best individual performance. Secondly with the still inefficient cloning technique, to achieve an annual ramet production capacity of half a million, we would need to clone 100 ortets per year. The availability of sufficient elite ortets would be limiting unless a program to create a constantly available pool of elite ortets from the reproductions of the best families has been planned earlier (Soh, 1998).

Plate II. C) A precocious high yielding clonal palm or ramet.

Plate II. D). Thick mesocarp, high oil bearing fruits of a clonal palm.

Recloning. As discussed earlier, good clonal tests are required to pick out the outstanding clones. To reproduce proven clones in reasonably large numbers, the feasibility of recloning (from ramets) must be in place as resampling (from ortets) will still be constrained in ramet production by the inefficient technique and limited explant source. The feasibility of clonal reproductions from cryopreserved somatic embryos, the alternative approach, has yet to be proven (Engelman and Duval 1986; Paranjothy et al. 1986; Dumet et al. 1993, 1994). Initial experiences with recloning were discouraging because of the higher susceptibility to mantling encountered (T. Durand-Gasselin pers.comm.1999; Wong et al 1999a), although it appeared to be more amenable giving higher embryogenesis rates (Table 3). Subsequent reclonings produced lower mantling rates (Table 6). Our latest results from 24 reclones averaged 2% mantling with only two clones having relatively higher mantling rates (about 10%). A possible explanation of the decreasing mantling susceptibility with the later reclones could be the “wearing-off” of the carry-over epigenetic effects in the older ramets sampled. Recloning now appears feasible.

Table 6 . Mantling incidence rates in reclones. Source: Soh et al . 2001.

^zRamets = clonal plants
^yReclones = clones obtained from cloning ramets
^xClones with any of its secondary ramets (from recloning) mantled.
^wRamets with any of its secondary ramets (from recloning) mantled.

Liquid Suspension Culture. With the feasibility to reclone, the prospect of mass propagation has improved. However, the conventional gelled-culture protocol involving proliferating polyembryogenic cultures with asynchronous development on which recloning has been based is still very inefficient in terms of ramet production rates and labor and space utilization (Plate IIE). It also offers limited scope for improvement and automation thus constraining the level of mass propagation. The latest development in the liquid suspension culture technique may solve this problem (De Touche et al. 1991; Duval et al. 1995; Texeira et al. 1995; Wong et al. 1999b). High efficiency has been achieved in our liquid culture system. About 80-90% of the embryogenic palm (ortet, ramet) cultures and 60-90% of their embryogenic callus cultures proliferated in the liquid system (Table 7) and most importantly, at much higher rates (6-7 times per monthly subculture) than in the gelled system (two times per bimonthly subculture). Very high success rates have also been achieved at shoot regeneration of the embryos and subsequent root induction on gelled medium (Plate IIF). Synchronized culture development is the hallmark of the liquid system and this has been largely achieved although further improvement in synchronization of embryo germination is still possible.

Plate II. E) Polyembryogenic cultures with asynchronous development in gelled medium.

Plate II. F) Plant development from somatic embryos in gelled medium.

^z Ortets = parent palms of clones; ramets = clonal plants

Predisposition to higher risk of somaclonal variation has been attributed to the liquid culture technique in other crops (George 1993) and also in oil palm (Y. Duval, pers.comm. 1993). The latest results from our liquid-cultured ramet plantings, for both clones and reclones, did not indicate increased risk of mantling. The few mantling incidences observed occurred mostly at much higher levels of ramet production (possible with the liquid culture method) as compared to the experiences with the gelled-culture technique (Tables 8, 9, 10, 11). CIRAD has apparently observed similar results on smaller sample test plantings (A. Rival pers. comm.. 2001). With the latest advances in the recloning and the liquid suspension culture techniques, the feasibility of commercial mass production of proven elite tenera clones is at hand.

Table 8. Mantling incidence rates in clones, liquid suspension system. Source: Soh et al. 2001.

^yMixed = clones/reclones having mixture of normal and mantled embryo lines
^zA clone/reclone with any of its clonal plants mantled is considered mantled.

Table 9. Mantling incidence rates in embryo lines, liquid suspension system. Source: Soh et al . 2001.

^zA clone/reclone with any of its clonal plants mantled is considered mantled

Table 10. Mantling incidence rates of ramets, liquid suspension system. Source: Soh et al. 2001.

^zRamets = clonal plants
^yPrimary ramets = ramets from cloning
^xSecondary ramets = ramets from recloning

Table 11. Ramet production and mantling rates of clones in gelled vs. liquid culture systems at various subculture levels. Source: Soh et al. 2001.

^zAbout 40 ramets (clonal plants) were sample-tested per subculture.
^yExtrapolated figures based on proliferation and germination rates of cultures

Field Evaluation. The oil palm industry is about to enter a new era of large-scale plantings of more genetically uniform materials in the form of clonal hybrids and clones. Nevertheless, there are still issues that need to be addressed and investigated now to ensure successful exploitation of these technological developments and minimize risk of catastrophic setbacks.

On the propagation side there is need to demonstrate that clonal hybrids do not perform poorer than sexual hybrids because of reduction in fitness of the clonal parents (Bregitzer and Poulson 1995; Phillips et al. 1994). This should be repeated with hybrids of recloned parents as there may be need to perpetuate the parents of outstanding hybrids. There is also need to know whether mantled ramet parent or normal ramet parent of mantled clones would transmit the abnormality to their hybrid progenies.

It would be useful to demonstrate the heritability of clonability and susceptibility/resistance to clonal abnormality and their repeatability in clones and reclones. It would also be essential to identify the predisposing environmental factors. Solving these issues, perhaps with the development and assistance of molecular markers, would facilitate the selection (perhaps also breeding) of amenable and resistant genotypes for cloning. We also need to establish, if feasible, safety limits for mantling risk and reduced fitness for cycles of recloning and for levels of production especially for the liquid culture technique. Molecular markers would have an important role here. Prospects for optimization and automation of the liquid culture protocol are still good and the development of synthetic seed might be more efficient than dealing with regenerated plantlets.

On the field testing and planting side, clonal hybrids and clones being genetically more homogeneous and discrete than mixed hybrids, would respond differently to different environmental factors e.g. location, planting density, fertilizer requirement, and abiotic (soil, water, mineral) and biotic (pest, disease) stresses i.e. exhibit GxE effects (Corley and Donough 1992; Donough et al. 1996; Smith et al. 1996). Adaptability trials are thus mandatory. For clonal hybrids, the choice of the hybrids should be based on GxE tests of the original sexual parental hybrids. In the case of clones which more prone to GxE effects and where ortet selection is inefficient, repeat GxE tests may be necessary to pick out the superior clones. In both cases to reduce the time and effort in GxE testing for a perennial tree crop, the selection of a minimum sample of environments that can discriminate the differential responses of the genotypes yet representative of the major environments where the genotypes are to be grown needs urgent consideration and investigation. This would be especially so for clones to reduce the need for repeat cycle tests and the number of candidate clones in each subsequent cycle. Use of molecular marker and quantitative trait loci (QTL) assisted selection has often been suggested but with a quantitative trait susceptible to GxE effects such as yield, the task would be difficult if not daunting.

As is common for successful cultivar development in most crops, there is always the need to produce or reproduce superior families preferably of different genetic origins and in larger numbers to provide the ortets for cloning to feed the subsequent clonal testing and selection program. With successful cloning, large plantings of genetically homogeneous materials of restricted genotypes can be foreseen and this would lead to genetic vulnerability of the crop to epiphytotics, pest outbreaks, inadequate pollination and moisture and mineral deficiencies. Thus, research is required for the packaging of clones and their field planting arrangements to reduce the genetic vulnerability and to also synchronize the cropping pattern with the future mechanized harvesting, and new crop management systems.

Genetic Progress. There have been reports of markedly superior yields of clones over tenera hybrids. Comparisons have sometimes been made against unspecified commercial hybrid materials with the attending errors due to between and within material variabilities and sampling. Even when individual crosses were used, the relative performances of the standard crosses within the respective proprietary commercial hybrid materials were not known. The comparisons of Cochard et al. (1999) were more objectively based having link crosses representing the first generation commercial hybrid material from their reciprocal recurrent selection program. The relative performance of the second generation hybrids with respect to the first generation hybrids was also known. The advantage of cloning improvement over breeding improvement should be based on the comparison between those of the clones and the second generation hybrids as discussed earlier. Soh (1998) suggested use of the reproduced hybrid family of the ortet as the control since the relative performance of the family in the hybrid progeny test, which represents the commercial hybrids, is usually known. This will also enable studies on selection response and selection efficiency to be made.

In terms of the relative genetic progress achievable by breeding and cloning, Jacquemard and Durand-Gasselin (1999) basing on their reciprocal recurrent selection scheme and their higher heritability estimate projected a 10-15% oil yield advantage of clones over seeds (conventional, clonal) for 7-8 years. The projection by Soh et al. (2000) was more conservative at 5-10% advantage for about 5 years with their modified recurrent selection based program (Fig. 4).

Fig. 4. Expected Breeding & Cloning Progress For Oil Yield (OY) .

DxP1 = current commercial tenera hybrids
DxP2 = reproduction of top DxP1 hybrids
DxP3 = reproduction of top families of improved second breeding cycle hybrids
Clones1, 2 = clones from selected palms of the best families of the first and second cycle hybrids
Clonal Seeds1,2 = reproductions of the best families of the first and second cycle hybrids using clonal parents
Reclones1 = reproduction of the best Clones1 clones
Assumptions (discussed in text):
Breeding progress = 15% per generation of 8-10 years;
Progress from reproduction of top families = additional 5%, obtainable 1-2 years after progeny-test results;
Progress from reproduction of best families by clones/clonal seeds = additional 15%, obtainable 5-6 years after clonal test results;
Progress reproduction of best clones i.e. reclones = additional 10-15%, obtainable 3 years after clonal test results.

Selection and reproduction of the top tenera hybrids i.e. DxP2 would give about 5% oil yield increase from the current hybrids (DxP1) within one to two years of obtaining the DxP Progeny Test1 results. Reproduction of the top hybrids (DxP3) from DxP Progeny Test2 results of the next breeding cycle hybrids eight to ten years later would give another 15% progress, 10% from improved second generation hybrids and 5% more from the top hybrids. Reproduction of the best hybrid family by cloning a sample of the family or from clonal seeds (hybrids from clonal parents) i.e Clones1 / Clonal Seeds1 available in five to six years’ time would give 15% progress. Similarly the next generation of clones (Clones2) and clonal seeds (Clonal Seeds2) available by year 12-14 would give an additional 15% progress. Reproduction of the best clone (Reclones1) in year 13-16 from the Clonal Test1 results would give an additional 10-15% progress from Clones1. Thus on average clones would have an advantage of 5-10% over improved hybrids for about five years.

These were at best rough projections as sufficient data on GxE effects and somaclonal effects on production traits in clones and reclones (Phillips et al. 1994; Bregitzer and Poulson 1995), although reported in other crops, are as yet unavailable in oil palm. Genetic progress also depends on the relative selection pressures placed on the hybrid families and clones for commercial reproduction and the genetic variabilities of the base hybrid populations.

2. Clones For Basic Studies And Other Applications

Besides commercial propagation, clones are useful for basic studies in genetics and physiology and other applications. Clones which allow replication of individual plant genotypes have useful application in basic studies on quantitative genetics e.g. partitioning between genetic and environmental variability between palms, estimation of heritability, and comparison of selection responses to different selection methods (Soh 1999, 2000). Clones will also be useful for physiological studies on crop yield potential (Corley and Donough 1992; Smith et al. 1996), the physiological bases of GxE responses and for plant developmental studies both at the plant and the in vitro level particularly with the more homogeneous suspension cultures.

Genetic transformation depends on a reliable clonal propagation technique (Parveez et al. 1998). Protoplast cultures were attempted to facilitate genetic transformation but plant regeneration has yet to be achieved (Bass and Hughes 1984; Budiani et al. 1998; Sambanthamurthi et al. 1996). Parveez et al. (1998) successfully used the biolistics approach with polyembryogenic cultures. Anther/microspore culture is useful in initiating haploids for dihaploid hybrid breeding but dihaploids have yet to be obtained in oil palm (Tirtoboma 1998).

D. Molecular Breeding

1. Marker-Assisted Selection. Marker-assisted selection has long been a quest of plant breeders (Allard 1960). With the development of molecular techniques, molecular markers closely linked to quantitative trail loci (QTL) controlling desirable traits are being sought. Such markers enable earlier and more precise selection, which is particularly advantageous for perennial tree crop breeding. The development of genetic linkage maps and markers associated with the desirable traits is a prerequisite. A program to map the oil palm genome using RFLP, AFLP, and microsatellite probes has been initiated (Mayes et al. 1997a; Cheah et al. 1999; Cheah 2000). Mapping the shell gene using bulk segregant analysis is the first obvious objective (Mayes et al. 1997b; Moretzsohn et al. 2000). Rajinder and Cheah (1999) using the bulk segregant and pseudo-test cross analyses on the selfed progenies of a high iodine value Nigerian breeding parent and a tenera clone and the progenies of a oleifera × quineensis interspecific hybrid were able to locate the locus for the virescens fruit and also found putative QTL for carotene content and the clonal mantling abnormality. The generation of high density maps for positional cloning and efforts at rapid gene discovery using expressed sequence tags (EST) are underway. Use of the GISH technique to discriminate the guineensis and oleifera chromosomes in oleifera × guineensis hybrid backcross breeding has been mentioned earlier. Other traits of interest for marker-assisted selection would include dwarfness, disease resistance and amenability to tissue culture.

While molecular marker assisted selection, which is essentially genotypic selection, is an exciting development for breeding, its success to date has been below expectation even in annual crops. Marker-assisted selection has appeared to be effective in early segregating generations, a combination of marker-assisted selection and phenotypic selection better for subsequent generations, and only phenotypic selection effective in later cycles (Stuber et al. 1999). Likewise, Bernado (2000), in a simulation study to evaluate the impact of genomic knowledge (sequence and function) of the genes controlling a quantitative trait, found that the incorporation of gene information with phenotypic selection using BLUP was most useful in selection of traits with few loci but not for traits with many loci e.g yield. Lack of well saturated uniformly spaced markers, QTL with small effects generally well distributed throughout the genome, which may be susceptible to GxE effects, level of linkage disequilibria, sample sizes, breakdown of favorable epistatic complexes, and multicollinearity (lack of independence among the factors whose effects are being measured), are some the factors considered responsible for the limited success of marker-assisted selection (Stuber et al. 1999; Bernado 2000). Undeniably, marker-assisted selection will be most useful in backcrossing and introgression programs. Molecular markers and probes have also been used to discriminate oil palm breeding populations, newly collected semi-wild populations (Shah et al. 1994; Mayes et al. 1997b; Kulratne et al. 2000; Purba et al 2000b; Rajanaidu et al. 2000), and clones (Cheah and Wooi 1995). They have also been used to determine the coancestry of parents for quantitative genetic analyses (Purba et al. 2000a).

2. Transgene Technology

Transgenic oil palms containing the Basta® resistance gene are available at the nursery seedling stage. The transgenes for high oleic acid production, which is the objective, have yet to be inserted (Parveez et al. 1998). High oleic acid was selected as the transgene of choice because a high oleic palm oil would enable it to compete with the liquid cooking oil market in temperate countries. Furthermore, oleic acid is a good industrial chemical feedstock. The approach (Cheah 2000; Shah et al. 2000) adopted involves down regulating the palmitoyl-ACP thioesterase, by using an antisense gene and up regulating the ß-ketoacyl ACP synthase II (KASII) activities, by using a strong promoter, to maximize oleic acid production at the expense of palmitic acid (Fig. 5) The thioesterase, KASII, and desaturase genes and the promoters for expression of the genes in the mesocarp tissues have been obtained and transformation will follow.

Back