Commercial oil palm planting materials worldwide are produced from a rather reduced genetic base, limited mainly to the variations present in the Deli dura and AVROS parental lines. This contrasts with the recommendation frequently made in the literature to increase the genetic diversity of planting materials used as industrial crops. ASD de Costa Rica sells four of the most important genetic crosses available in the international market: Deli x BM119 (Ex AVROS), Deli x Ekona (Ex N'Dian), Deli x Calabar (Ex NIFOR) and Deli x La Mé. In addition, ASD is also able to produce other new materials, such as Bamenda (Cameroon highlands) and Kigoma (Tanzania), which can tolerate cool environments. This diversity of materials is unique within oil palm seed suppliers, and makes it possible to grow oil palm in a wider range of environmental conditions. ASD's main goal has been to increase oil palm productivity by increasing oil and kernel contents, while at the same time reducing vegetative growth. With this in mind, several research objectives have been defined:  

1. Use of "compact genotypes" to increase planting density. 
2. Reduction of vegetative growth within conventional lines. 
3. Production of new hybrid lines with novel characteristics, adaptable to marginal ecological conditions. 

Some of the research projects currently underway are related to  

1. low temperature tolerance 
2. drought tolerance 
3. resistance/tolerance to fusarium wilt 
4. increasing oil insaturation level 
5. genotype-environment interactions studies aiming to identify stable genotypes or even some materials with good specific adaptability for very particular environments. 

The following paragraphs and diagrams describe the diversity of origins planted at ASD' s germplasm collection, which make it a leader in the world market of oil palm seeds. 

General aspects 

The first oil palm plants in Central America were planted by ASD (formerly United Fruit Co. -UFCo-) in 1926 and 1929, using seeds brought from Malaysia, Indonesia and Sierra Leone (Richardson 1995). From 1967 onwards, through international germplasm exchange programs, ASD consolidated one of the broadest collections of Elaeis guineensis in the world ( Table 1 ). This gene bank includes all the major breeding populations of restricted origin, as well as wild introductions from specific environments. E. oleifera lines from Brazil, Surinam, Colombia, Panama, Costa Rica and Nicaragua are also present in the collection. 

Breeding populations from restricted origin (BPRO) used as female parents 

Deli dura 

The best known oil palm seed producers in the world use Deli duras as mother palms, since they guarantee an offspring with outstanding bunch and fruit characteristics. However, there are some small research stations in Africa that normally use other dura lines as female parents, mainly because their objectives are to obtain fusarium wilt resistance and drought tolerance. 

Deli dura populations are derived from four palms originally planted in 1848 at the Bogor Botanical Garden in Sumatra. From there, descendants were planted in the Deli district of Sumatra and later introduced into different breeding programs in Indonesia, Malaysia and, subsequently, brought to Costa Rica ( Fig.1a and Fig.1b ). 

Within the Deli dura BPRO there is still apparently room to identify several "breeding subpopulations", as described in detail by Rosenquist (1986, 1992). ASD's collection contains all such relevant breeding populations, such as the Serdang (S), Ulu Remis (UR), Banting (BM8 and BM20) and Johore Labis (JL) series. 

With the cooperation of Dr. H. Hardon in 1969, the Ulu Remis population was introduced to Costa Rica from the Chemara Research Station of the Oil Palm Genetic Laboratory in Malaysia ( Fig 2a ). That same year, the lines BM8 and BM20 were also received from Malaysia through Harrison & Crossfield's Banting Research Station ( Fig. 2b ). Later, in 1970, based on an agreement between UFCo. and the Societé Financière de Caoutchomes (SOCFIN), and with the collaboration of Dr J. C. Knetch, the Johor Labis lines were introduced from Indonesia ( Fig. 2c ). Although we lack detailed information on the selection procedures followed to obtain the JL lines immediately before their introduction to Costa Rica, the F₁ selfings clearly out-yielded the parental lines by 60%, while in the F₂ , the increment was up to 40% when compared to the F₁ (DxP progeny test trials). 

In 1978, with the cooperation of Mr. E. Rosenquist from Harrison & Crossfield (H&C), eleven additional progenies were introduced from the Dami Oil Palm Research Station in Papua New Guinea. Those lines came from the Ulu Remis and Banting breeding populations, and were previously selected at Banting before being introduced to Dami ( Fig. 2d ). 

Another group of four Deli dura lines was introduced in 1978, this time, from the Mardi research station in Malaysia, with the cooperation of Dr. Ooi. Today, ASD makes intensive use of the family MAR559 for seed production, due to the high additive genetic effect presents in this family, in relation to several important economic traits, in particular growth and bunch characteristics .  

Other sources of dura female parents 

When characteristics of rusticity and adaptability to less-than-optimum conditions are required, Deli dura is not considered a good choice and, instead, Angola dura is commonly used as a female parent source. This line was previously selected by Dr. J. Meunier in the Ivory Coast from CIRAD (formerly IRHO) ( Fig.2e ), and was introduced into the germplasm collection owned by the Kade Research Station in Ghana, from where it was obtained by ASD in 1979. Some of the progenies introduced have no Deli ancestors. 

Duras of wild origin used as female parents  

Other female lines may be used for seed production when oil palm is to be planted in some marginal environments. Two of these materials were collected from wild dura populations in the highlands of Tanzania (Richardson and Chaves 1986) and Cameroon (Blaak 1974), where they grow and produce well in places with rather low temperatures.  

Bamenda (Cameroon). Oil palm grows in the wild in the Bamenda highlands of Cameroon, from where G. Blaak collected 25 accessions, and set aside as part of Lobé Research Station's oil palm gene bank. Nine of these accessions were introduced into Costa Rica in 1970. The original population was evaluated phenotypically at Lobé and Coto (Costa Rica), where it showed a reduced vegetative growth, as well as low bunch yield and extraction, mainly due to a reduced mesocarp to fruit ratio ( Table 2 ). However, at an experimental level, the Bamenda x AVROS progenies (DxP) have shown good bunch composition and a high bunch production potential, as well as very reduced vegetative growth. Other research results in Western Ethiopia at an altitude of 1,000 m (Blaak and Sterling 1996) demonstrated that cold tolerance can be transmitted to DxP descendants. 

Kigoma (Tanzania) . The Kigoma population comes from six plants growing in the wild on the Kigoma hills of Tanzania, near Lake Tanganyika (850 m asl, and temperatures below 12°C). Four accessions were collected from tenera plants in the Kwitanga region, one from a tenera in the Simbo District, and one from a dura plant from the Ilagala District (Richardson and Chaves 1986). The best characteristics of this population are the high bunch production potential, reduced vegetative growth, very thin-shelled kernels and nutshells, and high kernel production ( Table 3 ). 

The Kigoma x AVROS-DxP progenies, have high bunch and oil yield potentials, high kernel contents and vigorous vegetative growth. They perform well at reduced temperatures and have shown a noteworthy performance under conditions of water deficit. 

Sources of males for oil palm seed production  

As it is the case for female progenitors, there are not many genetic sources of males for commercial seed production. The pisifera lines used as male parents have, however, a wider genetic background when compared with the female lines. There are not only more populations available, but a larger number of plants were involved during the development of the original male populations, as compared with the only four plants originating the Deli dura population. 

Derivatives from D'jongo, such as the well-known materials Yangambi, AVROS and BM119, are predominantly used for seed production by ASD. Other populations used are the La Mé (L2T), Nigeria (NIFOR or Calabar) and Ekona. 

Most of the oil palm seed suppliers can sell one type of oil palm seed, but ASD can sell several materials at the same time because it has many pisifera origins.  

D'jongo derivatives 

The Yangambi or D'jongo population originated from a single plant selected at the Eala Botanical Garden of Yangambi in Zaire because of its high oil production. Descendants of this population were brought in to Sungei Panchur in Indonesia, where the Algemene Vereninging van Rubber-planters ten-ostkust van Sumatra (AVROS) selected the famous SP-540. It is generally accepted that this source of pisifera is the most commonly used as a pollen source by seed producers worldwide. 

One of the strains of D'jongo population, the AVROS materials, as they are universally known, were introduced into Costa Rica in 1969 and 1978 from H & C's Banting Research Station. Three other strains derived from AVROS or BM119 families, were introduced from Mardi (PORIM), also in 1978 ( Fig. 3 ). 

Besides the aforementioned AVROS and BM119 lines, there are other Yangambi populations present in ASD's germplasm bank. These introductions are directly derived from Eala's D'jongo population: the Yawenda, N'gasi and Isangi that were brought to the Ivory Coast by IRHO, and subsequently to Costa Rica. A group of selections from this same source was obtained by ASD through an exchange agreement program with SOCFIN, Malaysia. In 1978 five other lines were brought. A complete description of the genealogies of Yangambi or D'jongo population can be found in Hartley (1977), Corley (1992) and Rosenquist et al. (1990). 

All Deli x Yangambi materials show very high yields and precocity, as well as high extraction rates and vigorous vegetative growth. However, to achieve their genetic potential they require good environmental growing conditions and cultural practices. 

Ulu Remis tenera 

The pisifera source known as Ulu Remis tenera (URT, Chemara) comes from a bulk pollen mixture of 52 pisiferas selected from the Yangambi population during the first selection cycle of the Institut Nationale pour l'Etude Agronomique du Congo Belge. In 1947 this pollen was sent to Layang Layang, Malaysia and was used on Deli dura plants of the Ulu Remis breeding population. The selfings derived from those crosses originated the first pisiferas used for DxP seed production in Malaysia, and are known as URT. 

This population was introduced into Costa Rica in 1969 ( Fig. 4 ). When compared in progeny test trials with other male sources, the DxP progenies derived from URT, showed high bunch yields combined with very high kernel extraction rates, but were below average in terms of oil extraction when compared with tenera crosses of AVROS origin. 

Ekona 

The Ekona male line originated at the Lobé Research Station in Cameroon, and has been used by Unilever in several breeding programs and seed production units in West Africa and Malaysia (Rosenquist 1986, Rao et al. 1999). The genetic background of this line can be traced to 18 dura plants and 24 tenera, descended from 362 palms, selected in 1948 out of 35,000 individuals due to their high bunch yields. This selection took place on a commercial plantation in the Cowan and N'dian districts of Cameroon ( Fig. 5 ). 

With the cooperation of G. Blaak, a good part of the Ekona population was introduced into Costa Rica from Lobé in 1967. DxP descendants have become an important part of ASD's seed production, not only because they are well represented in ASD's collection, but also due to their very high oil extraction and bunch production and their moderate vegetative growth. 

Some of the stations that are recipients of Ekona planting materials include the Sabah Breeding Program, Department of Agriculture of Malaysia, Harrison & Crossfield, Guthrie, SOCFIN, NIFOR, and Pamol. In 1970 ASD received 14 crosses from the Lobe Program. 

In the Ekona ancestry, the foundation palms were obtained as a result of pressure to select for high yield oil/bunch and fusarium wilt tolerance. In commercial and research trials conducted in Coto and Palmar in Costa Rica, Ekona DxP plants showed a low incidence of crown disease, while fresh fruit bunch production was similar to AVROS. Deli x Ekona, however, exhibited greater variations, and the former tenera source had higher oil/bunch content from superior mesocarp/fruit and oil. 

Nigeria (Calabar) 

This population was introduced into Costa Rica in 1977 from the Oil Palm Research Center of Kade in Ghana, thanks to the cooperation of Dr J. B. Wonky-Appiah. The population originated from 17 plants (11 tenera and 6 dura) with no Deli ancestors selected at the Nigerian Institute for Oil Palm Research (NIFOR, formerly WIFOR) from 1960. It is important to note the participation of the well-known plants CA-256 and 32.3005T, which are considered to be the most remarkable genotypes of this population (Green 1973, Okwuagwu 1986). The 17 plants were crossed in every possible combination according to the "reciprocal recurrent selection" breeding method ( Fig. 6 ). 

The palm 851.253T, and its selfings-descendants are the source of Calabar germplasm present in Costa Rica. This palm originated from the selfings of the palm 32.3005T in Kafir, Ghana.  

Other sources of Nigerian material were used by CIRAD at the La Mé Research Station in the Ivory Coast, where they were named "WA" Series. The WA10 and WA12 lines represent this population in Costa Rica. Progenies from this origin have some degree of variation due to their broader genetic base, compared with Kafir Nigerian germplasm. Descendants of families WA10 and 12 (IRH628, IRH629, IRH636 and GHA608) are characterized by long leaves, short stems, high bunch production and virescens fruit color. 

Other Calabar derivatives present in Costa Rica are GHA647 and GHA648. These families show moderate and very uniform growth, very short leaves, vigorous stems and nigrescens fruit color. 

The Indonesian Oil Palm Research Institute (IOPRI, formerly Marihat), the Saba Breeding Program in Malaysia, NIFOR, the Kade Research Station and ASD de Costa Rica use pisiferas of Calabar origin for seed production. 

La Mé 

La Mé male parents were developed between 1955 and 1973 by IRHO at the La Mé Research Station in Ivory Coast. This population originated from 21 tenera plants (especially from the tenera genotype BrT10). The first selection was made at the Bret Plantation in Benin, from plants that were initially collected from the wild in the Ivory Coast ( Fig. 7 ). 

This population was introduced into Costa Rica around 1980. Particularly important are the selfings of the exceptional plant L2T, which has resulted in exceptionally good pisiferas (Gascon and de Berchoux 1964). 

The commercial Deli x La Mé progenies have elongated fruits, thick shells, and high bunch production (with a large number of low-weight bunches). The oil has high iodine values (over 0.55) and the plants have very long leaves, short stems and long inflorescence peduncles. 

The IOPRI in Indonesia, the Institut Des Forests of Ivory Coast (IDEFOR, the former La Mé Station) and ASD de Costa Rica use this population for commercial seed production. 

Dami composite lines 

The Dami composites were developed by Harrison and Crossfield, aiming at improving the bunch index. To develop these lines, several populations were used, such as BM29, BM31, Dumpy E206, BM119 and Deli ( Fig. 8 ). The first generation was planted in Papua New Guinea. 

The best advantages of these lines are their bunch characteristics, particularly in the DAM858 and DAM586 families, which produce DXP descendants with outstanding oil to bunch ratios, especially because of their exceptional oil to mesocarp content. The DxP progenies of DAM588 have shown excellent performance in different progeny tests in Costa Rica (Sterling et al. 1994). 

Parent selection based on progeny testing 

Oil palm breeding has focused mainly on improving the oil and kernel yields and on extending the commercial life span of commercial plantations. The selection process is based on: 

1. Phenotypic evaluation of the families and siblings of individuals through direct measurement of the parameters of interest. 
2. Progenitor evaluation through progeny tests, where the genotypic additive effect is measured (general combining ability) as well as specific combining ability. 

Female lines (duras) 

Prior to its introduction in Costa Rica, the Deli dura had been undergoing a continuous selection and improvement process since the early 1900s, first in Indonesia and later in Malaysia. Thus, the Deli dura population brought to Costa Rica was already improved when compared with the original Deli Serdang population. This is particularly true for the introductions that came from Dami, as these had undergone two previous selection cycles over the original Banting and Ulu Remis breeding populations. This whole process guarantees the genetic stability and production potential of the female population and their descendants during the process of commercial seed production. 

Since 1969, the Deli dura populations in Costa Rica have been subjected to a strong selection pressure. Emphasis has been placed on oil bunch yield, taking into account characteristics such as high bunch weight, oil quality and limited vegetative growth. 

An important goal for most oil palm breeders is to broaden the gene pool of the female parent used for seed production. This may be achieved by expanding the genetic base of the Deli dura population or using other dura lines. To this end, the Angola population and the Bamenda and Kigoma wild populations have been tested. 

Angola. The main advantage offered by the Angola population when used as a female progenitor is its reduced stem growth rate; however, bunch production is less than in the Deli dura lines. These materials combine well with Ekona and Mardi pollen ( Table 4 ). 

Bamenda. Although it has not been extensively tested, the Bamenda population has exhibited great potential in relation to bunch production. Annual bunch yields for Bamenda x AVROS progenies are of 200 kg plant-1 yr-1 with a reduced stem growth and acceptable bunch composition ( Table 5 ), which means a high commercial potential of over one ton of oil per ha per year over the control. 

Kigoma. DxP progenies obtained from Kigoma female progenitors crossed with several male parent sources showed a high bunch and oil yield potential, very similar to tenera materials derived from Deli duras. This fact is particularly true when the male source is of Mardi or AVROS origins ( Table 6) . In addition, vegetative growth is lower in Tanzanian derivatives than in those of Deli origin. 

Other uncommon sources of dura parents progeny tested. An experiment planted in 1990 in Santo Domingo de los Colorados, Ecuador, showed that the Deli and Deli x Angola female lines, and the Ekona male line were very precocious (highest GCA values), while the best specific combinations were the Deli x Yangambi and Kigoma x Ekona progenies ( Table 7 ).  

Male lines (pisiferas) 

Given that pisifera plants are normally female sterile, this makes it impossible to evaluate the production performance of the individuals selected. The only direct measurements that can be obtained on the pisifera plants are of morphometric or leaf mineral contents. The most commonly recorded traits are leaf area index (LAI), stem growth rate, leaf length, leaf emission rate and magnesium content. The strategy, then, is to evaluate the performance and production of the tenera siblings coming from the same family or, even better, the performance of the DxP descendants directly. 

Thus, selection based on progeny testing appears to be the only reliable criteria to decide the genetic potential of an individual pisifera, with the aim of selecting male parents for commercial oil palm seed production. ASD has been progeny testing its original introduced D'jongo derived sources of male parents since 1969. More recently, from 1987 onwards, advanced breeding generations of the above mentioned origin, as well as new pollen sources, started to be evaluated on terms of progeny performance. 

Several trials began in 1991 to measure the genotypic by environmental effect, and it was shown that all seven pollen sources tested ( Table 8 ) were very similar in bunch production (Alvarado et al. 1997). However, differences in height, oil extraction and environmental adaptability could be attributed mainly to the pisifera source and, to a lesser extent, to the dura source effect. The La Mé and IR1039 showed the lowest stem growth. Progenies from Calabar and Ekona pisiferas had intermediate heights, while the highest stem growth came from AVROS lines. 

Bunch characteristics were also influenced by heredity, particularly the oil content in the mesocarp. This variable showed the highest values in DxP progenies obtained of Ekona and Yangambi pisiferas. 

In a similar type of test at the Surat Thani Horticultural Research Center in Southern Thailand, a rather dry region, Nakorn (1997) it was found that nine of the 60 progenies tested yielded from 22 to 25% more oil per hectare than the experiment average ( Table 9 ). Those results confirm the high yield potential of ASD materials, as all nine progenies come from parental lines used for seed production in Costa Rica. 

Conclusions 

ASD de Costa Rica's breeding program has been producing commercial oil palm planting materials since 1974. Initially, the program was oriented towards the exploitation of the genetic potential present within Deli and AVROS BPRO's, as the basis for seed production. More recently however, the inclusion of new sources of germplasm has allowed ASD to produce new alternatives for D'jongo derivatives. These options include the well known La Me, Nigeria, (Calabar) and Ekona origins, as well as some pisifera sources of wild origin (e.g. Kigoma, Bamenda). This broad genetic resource allows the Costa Rican breeding program to provide oil palm planting materials to meet the global expansion of the crop. This includes supplying materials for new environments that were previously ruled out for oil palm cultivation due to adverse conditions, such as drought and low temperatures.  

References 

Alvarado, A.; Sterling, F.; Montoya, C; Angulo, V. 1997. Interacción genotipo x ambiente y análisis de estabilidad genotípica en cuatro localidades. In. Conferencia Internacional en Avances Agronómicos de la Palma de Aceite. ISOPA/CENIPALMA. Cartagena, Colombia. Pp. 206-226. 

Blaak, G. 1974. Personal communication.  

Blaak, G. Sterling F. 1996. The Prospects of Extending Oil Palm Cultivation to Higher Elevations Through Using Cold-Tolerant Plant Material. The Planter Kuala Lumpur. 72, 645-652. 

Gascon, J.P.; de Berchoux, C. 1964. Caracteristiques de la production d' Elaeis guineensis Jacq. de diverses origines et de leurs croisements. Oleagineaux. 19(2):75-84. 

Green, A.H. 1973. Annual review of research 1971. Internal Report. Unilever Plantations Group. London. 

Hartley, C.W.S. 1977. The Oil Palm. Second Ed. Longman. London. 806pp 

Nakorn. 1997. Surat Thani Horticultural Research Center, Thailand. Personal communication.. 

Okwuagwu, C.0. 1986. The genetic base of the NIFOR oil palm breeding program. In. Proc of Int Workshop on Oil Palm germplasm and utilization. Kuala Lumpur. PORIM. 1986. 10:228-237. 

Rao, V.; Law. I. H.; Shaharudin, Z.; Chia, C.C. 1999. Ekona and AVROS- a tale of two pisiferas. PORIM Oil Palm International Conference. Kuala Lumpur, Malaysia. 

Richardson, D. L. 1995 . The History of Oil Palm Breeding in the United Fruit Company. ASD Oil Palm Papers. 11:1-23.  

Richardson, D. L.; Chaves, C. 1986. Oil palm germplasm of Tanzanian origin. Turrialba. 36:493-498. 

Rosenquist, E. A. 1986. The genetic base of oil palm breeding populations. In. Proceedings of Int. Workshop on oil palm germplasm and utilization. ISOPB/PORIM. Pp. 27-56. 

Rosenquist, E. A. 1992. Some ancestral palm and their descendants. Int. Symposium on the Science of Oil Palm Breeding. ISOPB/IRHO/PORIM/BUROTROP. Montpellier, France. 28 p. 

Rosenquist, E. A., Corley, R.H.; De Greef, W. 1990. Improvement of tenera populations using germplasm for breeding program in Cameroon and Zaire. In. Proceedings of Int. Workshop on progress of oil palm breeding population. Kuala Lumpur. PORIM. 1990. 16