Abstract

The oil palm is an extensive commercial crop demanding large tracts of land for its exploitation. In addition, vigorous plant growth (rapid stem elongation) is also an important financial limitation, since tall palms are difficult to harvest and excessive height of the palms reduces the economic life of a commercial plantation. High yielding compact palms with slow stem elongation and short leaves become a good alternative for intensifying the crop and boosting its productivity, by increasing planting density and prolonging commercial exploitation. This is particularly important for countries with limited suitable land for oil palm cultivation.

The original compact palm (OCP) was discovered in observation plots planted in Coto, Costa Rica, which were planted with open-pollinated seeds from an E. oleifera x E. guineensis inter-specific hybrid (O x G) with outstanding characteristics of slow growth and short leaves, identified in 1966 in Quepos, Costa Rica. The OCP had short trunks and short leaves, but unfortunately its bunches were substandard. Aiming at improving the bunch composition of the OCP, three cycles of backcrossing to E. guineensis parents have been performed since 1978. At the same time and after each backcross cycle, F₁ palms with outstanding characteristics were selected and intercrossed to produce superior F₂ recombinants, in attempts to produce commercial seeds. Successive back crossing somehow caused a gradual loss of the compact characteristics of short trunk and short leaves due to the introgression of guineensis genes, particularly when backcrossed to AVROS parents. Nevertheless, the second and third backcross cycle produced compact recombinants with short trunks, short leaves, good bunch characteristics and commercial FFB productivity, comparable to traditional commercial seed varieties.

Elite palms with special characteristics, such as high bunch oil content, slow stem growth and short leaves were identified as ortets for clonal production. Selected progenies from selfings of back cross two (BC₂ ) palms were also chosen to produce compact seeds.

The final result of this research, is the possibility of increasing the oil palm yield to new limits, since it is now possible to boost productivity by using high yielding compact clones or seeds planted at densities of up to 160-200 palms per hectare. Planting at higher densities will depend on the conditions affecting growth of each site, particularly rainfall and soil fertility and physical conditions.

Introduction

Selection of palms with slow trunk growth was always in the mind of oil palm breeders, seeking to prolong the economic life span of commercial plantations. Finding special small and productive palms with potential commercial value is not an easy task, and only a few examples of such palms have been documented. Jagoe (1952) discovered the ‘Dumpy' palm, which had large girth and slow height increase, and this constituted the first effort of genetic introgression of dwarf genes into other oil palm populations. By backcrossing E. oleifera x E. guineensis (OxG) hybrids to guineensis parents, Obasola et al. (1976) illustrated the possibility of maintaining the slow trunk growth characteristics of the oleifera parent in the resulting recombinants. More recently, Adon et al. (2001) demonstrated that Dumpy (Serdang) and Pobe origins definitely transmitted slow vertical growth to their descendants in combination with other origins. Also, Rajanaidu et al. (1999) identified the PS1 lines out of population 12 derived from prospections in Africa. The same author (Rajanaidu et al. 2000) indicated that Dumpy genes promoted slow growth in combination with AVROS parents.

In Costa Rica, considerable effort has been devoted to fixing ‘compact' genes, originating from a special palm identified in a backcross progeny of an OxG hybrid open-pollinated with guineensis, which not only transmits slow trunk growth but short leaf characteristics.

The main objective of the compact palm program has been to produce seed varieties and clones capable of yielding commercially at higher densities than the current industry standard of 143 palms per hectare. This paper describes the main events of the compact breeding program by ASD de Costa Rica, S. A. (Agricultural Services and Development) using the backcross method.

Fresh fruit bunch yields (kg) per ha were recorded over five years from the third year after planting in all trials described in Table 1 . Stem height to frond 41, and leaf length (petiole + raquis) in cm, were measured at 55 months after the date of planting.

Four to six bunches were analyzed to characterize each palm during the first 4-6 years of age, according to the method described by Blaak, et al. 1963 and revised by Rao et al. 1983.

In Honduras a 25 ha semi-commercial trial was established in 1995, at 160 palms per hectare, to test 44 compact BC₃ DxP crosses and two DxP control crosses. The trial had 7 replications with 12 palms per plot (84 palms per cross). Yield was recorded from 1988 to 2001.

A clonal trial (CB9702) planted in 1997, included seven compact backcross two (BC₂ ) clones and one DxP Deli x AVROS control. Palms were planted at 170 palms/ha in a Complete Randomized Block Design with 4 replications and 12 palms per plot. The FFB (fresh fruit bunch) production was recorded for three years from the third year after planting. The trunk height (cm) to frond 41, and leaf length (cm), were recorded at 51 months after planting. Bunch analyses were carried out from 43 to 60 months after planting.

In order to verify the repeatability of the traits of selected ortets in their respective clones, linear correlations (r²) were calculated using the information of trial CB9702.

With the discovery of an outstanding Elaeis oleifera x Elaeis guineensis (OxG) hybrid in 1966 and the identification of the original compact palm (OCP) in the 1970s (Sterling et al. 1987), different breeding strategies (paths) have been adopted to fix compact genes into different recombinant populations, which would finally lead to commercial seed and clonal production. Table 1 is a summary of the main breeding events of more than 30 years of research in the compact program, and includes details of how the successive backcross cycles were conducted. This summary serves as reference for the discussion of the results of each phase.

Sterling et al. (1987) described the original compact palm (OCP) as having a short stem and short leaves, but unfortunately poor bunch quality; hence, the need to introgress compact genes into other guineensis germplasm was imperative to improve yield and bunch quality of the compact materials. The adoption of the backcross method had the purpose of finding recombinations having the compact traits and improved yield and bunch quality. The challenge was to fix the compact genes in order to produce a uniform variety, since backcrossing produces an array of different phenotypes, showing extreme and intermediate characteristics. Finding the correct combinations was therefore the key to the whole program.

Another important aspect was the selection of the guineensis materials as gene donors for oil content and yield in the compact lines. From the beginning of the program, it was believed that compact genes came from the oleifera ancestor, hence, the dilution of these genes from successive backcrossing to guineensis would probably cause the loss of the compact trait. However, the probability of finding segregates with all desired traits was also great, despite the diminution of the oleifera genes. Indeed, the production of compact commercial seeds was initiated using the BC₂ F₁ population, despite having only 6.25% of oleifera genes compared to the OCP with 25% ( Table 1 ).

The rather low yield values per palm reported in this paper do not necessarily represent the commercial potential of the genetic materials studied. They are mainly the result of differences in the environmental conditions (low solar radiation) where the different trials were planted, and inter-palm competition caused by the type of the field experimental design used. This aspect will be discussed later, but now the focus is on the potential of the new compact materials when compared with a commercial variety like Deli x AVROS growing under similar conditions.

Table 2 is a summary of the whole compact program, showing how the different populations were gradually improved, reaching comparable levels of the DxP control, in terms of FFB and oil yield. Details on how this progress has been achieved are presented during the subsequent discussion on each of the phases of the program.

The results of the first backcross (BC₁ ) trial showed that the poor bunch quality of the OCP could be improved substantially (Sterling et al. 1987), but remained low when compared with the AVROS DxP control. However, the reduced trunk height and short leaves of the OCP were preserved ( Table 2 ), even though the oleifera genes were reduced from 25% to 12.5% ( Table 1 ). The average difference in trunk height between the compact palms and the control was 36 cm and the leaves were 73 cm shorter (P<0.05). However, the compact palms yielded less FFB, which along with a lower oil to bunch ratio (O/B), resulted in significantly reduced oil production per area (O/ha) (P<0.05).

Two compact palms were selected in the first backcross cycle BC₁ , one of them with 50% AVROS genes and the second one with 50% La Me genes. These palms showed a lower production of FFB than the DxP control AVROS, but their trunks and leaves were considerably shorter. The ancestor La Me of palm 122T conferred a shorter trunk (Table 3).

The first attempt to produce compact seeds was using the segregates of the two selected BC₁ compact palms ( Table 3 ) which were then intercrossed and selfed to produce the BC₁ F₁ population. Within this BC₁ F₁ population, five dura and seven tenera compact palms were selected ( Table 4 ). These selected palms were then crossed to guineensis parents (BC₁ F₁ x Eg) to perform the first compact progeny testing trial ( Table 5 ). The two BC₁ selected palms in Table 3 , one with 50% AVROS genes and the other with 50% La Me genes, did not show FFB yields as high as the DxP control, but had shorter stems and leaves. The La Me ancestor of palm 122T conferred an even shorter trunk.

The selected BC₁ F₁ compact recombinants used for the first progeny test, all showed short stems and leaves, but relatively low FFB and oil/ha ( Table 4 ). When crossed to guineensis parents, the results were not satisfactory, since both yield parameters remained lower than the DxP control (P<0.05). However, the compact trait was retained in all combinations ( Table 5 ). Unfortunately, the use of the honeycomb design, Fasoulas, A. (1976), which allows planting single palms representing each cross in a hexagonal design, caused the more vigorous Deli x AVROS controls to overshade the compact palms in the hexagon. Using this design is probably not suitable for this type of testing unless all palms are planted in a condition without inter-palm competition. Even though the yield results were not encouraging, the stability of the compact trait gave hope to continue with the program ( Table 5 ).

A second attempt to produce commercial seeds was by using the recombinants of BC₁ F₂ population, which resulted from intercrossing the five selected BC₁ F₁ dura and seven tenera, which were also out-crossed to compact BC 1 F 1 pisiferas ( Table 6 ). In these trials, no comparison with a DxP control was performed. However, the FFB/palm and oil/ha of the T x D compact progenies were still not promising.

A second backcross population (BC₂) was generated using the two selected BC 1 palms described in Table 3 , which were crossed to Deli x AVROS, Bamenda, Ekona, Nigeria and pure AVROS parents. The rationale for advancing to a second cycle of backcrossing was the fact that the compact trait remained consistent, despite the dilution of the oleifera genes and the type of cross and germplasm combination. Hence, the introgression of more and different guineensis genes was a good option for improving the low production of FFB and oil observed in the previous compact generations ( Tables 5 and Table 6 ).

With the objective of introgressing compact genes into a uniform and stable guineensis commercial variety, selected commercial Deli x AVROS teneras with 50% Deli genes and pure Ekona and AVROS teneras were considered as good parent options, aiming always at increasing the frequency of guineensis genes in compact variants of the first backcross cycle. The exception was the use of Bamenda parents as a new source of guineensis genes, because this germplasm was not as widely manipulated as the standard commercial varieties and could generate new interesting combinations ( Table 7 ).

FFB yield and bunch quality were substantially improved in the second backcross cycle. The compact progenies of this cycle with only 6.25% oleifera genes performed at almost the same level as the DxP control. The combinations with Bamenda and Ekona germplasm were particularly superior ( Table 7 ).

The trunk height and leaf length of the BC₂ compacts remained lower than the DxP control in all compact BC₂ progenies ( Table 7 ). These results led to the creation of a BC₂ F₁ population by selfing selected BC₂ dura and tenera compacts and crossing them to BC₂ pisifera siblings, to attempt seed production for the third time ( Table 8 ).

Bunch quality was improved in compact teneras BC₂ palms with Ekona genes, particularly for mesocarp to fruit and oil to mesocarp ratios, both higher than in the D x P control. These compact BC₂ palms yielded better combinations in BC₂ F₁ , and some of them were even better than the DxP control, such as the combinations with the compact palm 494D and 173D ( Table 9 ). On the other hand, the compact trait remained intact in this population; the trunk height was on average 36 cm shorter and the frond length 131 cm shorter than the DxP control ( Table 9 ).

It is important to point out that the trunk height differences between individual compact palms and the control can be greater with age, as described in the following sections. Similarly, compact palms can show considerably shorter leaves than the control.

Based on the satisfactory results of the second backcross cycle, seed production using compact BC₂ F₁ compact population was finally started in 2002, aimed at planting them at a density of 180 palms per ha in sites with high solar radiation of more than 400 langleys/day (cal/cm²/day) and 160 palms per ha in places with less solar radiation.

The encouraging results with BC₂ F₁ population, led to a third backcross cycle (BC₃ ) by crossing selected BC₂ compact palms to various guineensis origins ( Table 10 ). In terms of FFB and oil production, the BC₃ compact palms reached the desired levels, showing no differences with the DxP AVROS control (P<0.05). The mesocarp to fruit and the oil to mesocarp ratios were remarkably improved Table 10 ).

Even though the compact palms in this third cycle had only 3.125 % of oleifera genes; they still showed the compact traits: trunks and leaves were 22 cm and 98 cm shorter respectively, when compared with the DxP control at 55 months after planting. This seemingly unimportant reduction of 22 cm in stem height becomes relevant as the palm grows, achieving differences as great as 1.5 m at the age of 8 years. On the other hand, the difference in leaf length of compact palms compared with the control was more consistent with age, being around 90 to 120 cm. Logically, individual compact palms within different populations will show even greater differences in leaf length when compared to the control ( Table 11 ).

In Honduras, a semi-commercial trial on 25 ha comprising 44 compact BC₃ crosses and two DxP control crosses showed that the yield potential of compact BC₃ can be quite high after three years of production, when planting at 160 palms per hectare ( Table 12 ). The best BC₃ combinations came from Tanzania and Calabar pisiferas, which out-yielded the Deli x Ekona and Deli x Lame DxP commercial controls, after the first three years of accumulated production. The production difference between BC₃ Tanzania with the Deli x Ekona DxP control was significant (P<0.05). It is expected that the BC₃ compact palms will gradually produce more FFB/ha than the testers when inter-palm competition will become more severe with age for the DxP controls with longer leaves.

In general, the compact BC₃ varieties produce smaller but more numerous bunches than the DxP controls. This characteristic is advantageous according to the current oil palm industry trend, and it is estimated that varieties with small, numerous bunches are more productive.

An important advantage of the backcross method, used to fix compact genes, is the possibility of selecting individual high yielding palms for cloning among the different segregating compact populations. The main advantage of cloning, bearing in mind the massive production of exact copies of special palms as the compacts, is the possibility of fixing genotypes more efficiently than with the classic breeding techniques using seeds. On the other hand, cloning allows the reproduction of superior palms without paying too much attention to their origin, which substantially increases the possibilities of consolidating special high-yielding commercial clones.

Besides the compact trait, selection of compact ortets was based on high FFB yield of more than 150kg/palm/year and good bunch quality, mainly M/F above 85% and O/M above 50%. However, substandard ortets were also included to check for repeatability of the different traits.

Evaluation of compact clones will not be discussed in detail in this paper, since this subject will be covered in a separate document. As a general illustration of the potential of compact clones, the results of trial CB9702 established to evaluate seven compact clones at 170 palms per hectare are presented in Table 13 .

Two compact clones were outstanding; 217T and 514T, which outyielded the DxP control in terms of both FFB and oil/ha/year (P<0.05). Clone 217 showed high O/M above 50%; this characteristic made this clone yield more oil/ha than clone 514T (9.7 versus 9.3 t) despite having lower FFB production (179.8 versus 210.2 kg/palm/year). Clone 273T showed similar oil/ha yield to the DxP control (P<0.05), again due to its higher O/M of 56.9%, rather than its FFB production ( Table 13 ). The rest of the compact clones did not perform as expected, and were below the DxP control. These results were in agreement with Soh (1986) and Soh et al. (2001) who found that despite selecting high yielding ortets from advanced breeding populations, there is a need to carry out clonal field testing to prove their real commercial value.

Most compact clones preserved the compact traits, having on average 20 cm shorter trunks and 80 cm shorter leaves compared with the DxP control at 51 months after planting. As mentioned before, the difference in trunk height will become more dramatic with age, when inter-palm competition will certainly affect the DxP planting materials at 170 palms/ha more than the compact palms.

Stem height, frond length and M/F, are repeatable traits in the clones (r² , P<0.05) ( Table 14 ). However, FFB, F/B, O/M, O/B and O/ha, were not consistent with the same traits observed in the ortets. The most productive 217T clone showed one of the shortest trunks (65 vs. 95 cm in the control) and short fronds (573 vs. 646 cm in the control); this result confirms that the compact trait was successfully fixed through the backcross method. However, not all clones showed the compact trait; for instance, clone 465T had the same trunk height as the control (97 vs. 95 cm), and consistently this clone had the longest trunk in trial CB9702 ( Table 13 ). This result confirms that trunk height was a highly repeatable trait in the clones.

The fact that no abnormalities were observed in trial CB9702 makes the production of compact clones a reality for commercial plantations. More than 80,000 compact ramets were produced in 2003 and it is expected that in 2004 production will be increased to 350,000 ramets. ASD de Costa Rica (Agricultural Services & Development) produces clones using inflorescence tissue since the onset of the program back in 1990 (Guzman, 1995). So far this strategy has proved to be highly satisfactory and safe.

1. After more than 30 years of continued research and three breeding cycles, it was possible to successfully fix the compact trait (short trunk + short fronds) in different breeding lines using the backcross method.
2. Although, the number of genes that determine the compact trait is not yet known, the results achieved to reproduce the compact phenotype indicate that probably only a few genes are involved.
3. Since the compact palm originated from a hybrid, it is believed that the related genes came from the oleifera ancestor, which led to the selection of special recombinants, despite their frequency reduction in the genome, from 25% in the original compact palm to 3.125 % in the third backcross cycle.
4. Undoubtedly, the tremendous effort put into managing the compact-related field trials and the years of research, could be eased if gene marker technology were available. For future research in the compact program, marker-assisted breeding needs to be implement for more accurate selection of parents and ortets and to reduce the research costs involved.
5. Compact seeds and clones will allow planting oil palms at higher densities, thus, boosting the productivity per hectare. The expectation that the compact trait will be consistently reproduced in clones, will assure planting at higher densities between 160 to 200 palms per hectare, depending on characteristics of each site.

The authors wish to acknowledge the contributions of Dr. D. L. Richardson and Mr. F. Sterling, for the design of the breeding strategy and selection of the parental palms, which made possible the development of today's compact varieties and clones. To Mrs. N. Guzman, who made the production of compact clones from inflorescences a reality, to Mr. F. Peralta, for his agronomy studies aimed at facilitating the use of compact clones by the growers, and to Dr. C. Chinchilla for reviewing this paper.

ADON B, Cochard B, Flory A, Potier F, Quencez P and Durand-Gasselin T. Introgression of slow vertical growth in improved oil palm (Elaeis guineensis Jacq.) populations. In: Proc. 2001 International. Palm Oil Congress. - Agriculture, pp. 210-217, Malaysian Palm Oil Board, Kuala Lumpur

BLAAK, G., SPARNAAIJ, L.D. AND MENEDEZ, T. (1963) Breeding and inheritance in the oil palm (Elaeis guineensis Jacq.) II. Methods of bunch quality analysis. J. W. Afr. Inst. Oil Palm Res. 4: 146-155.

FASOULAS, A. (1976). Principles and methods of plant breeding. Department of Genetics and Plant Breeding. Aristotelian University of Thessaloniki - Greece. Publication No. 6, 55 pp.

GUZMAN, N. (1995). Present status of clonal propagation of oil palm Elaeis guineensis Jacq. in Costa Rica by culture of immature inflorescences. ISOPOB Conference, Barranquilla, Colombia, 7-9 June, 1995.

JAGOE, R.B. (1952) The "Dumpy" oil palm. Malay. Agric. J. 35: 12-21

SOH, A. C. (1986). Expected yield increase with selected oil palm clones from current DxP seedling materials and its implication on clonal propagation, breeding and ortet selection. Oleagineux, 41, p. 51-56.

SOH A.C., WONG G., TANG C.C., CHEW P.S., HOR T.Y., CHONG S.P. & GOPAL K. (2001). Recent advances towards commercial production of elite oil palm clones. In: Proc. 2001 Int. Palm Oil Congr. - Agriculture, pp. 33-44, Malaysian Palm Oil Board, Kuala Lumpur

STERLING, F., RICHARDSON, D. L., CHAVES, C. (1991). Honeycomb and randomized block designs for selection among oil palm progenies. In: International Oil Palm Conference. Progress, Prospects & Challenges Toward 21st Century. Palm Oil Research Institute of Malaysia. Kuala Lumpur. 48-57 pp.

STERLING, F., RICHARDSON, D.L., CHAVES, C. 1987. Some phenotypic characteristics of the descendants of QB049, an exceptional hybrid of oil palm. Proceedings Oil Palm/Palm Oil Conference. Progress and Prospects. Palm Oil Research Institute of Malaysia. 135-146 pp.

OBASOLA, C. O., OBESESAN, I. O. and OPUTE, F. I. 1976 Breeding of short stemmed oil palm in Nigeria. Int. Agric. Oil Pam Conference, Kuala Lumpur, 1976.

RAJANAIDU N., JALANI B. S., KUSHAIRI D., RAFII M.Y., MOHD DIN A., MAIZURA I., and Ariffin Daurus (1999). Breeding strategies for the oil palm materials PS1 and PS2 and future PS series. Proceedings of 1999 Seminar on PS1 and PS2 planting materials, Kuala Lumpur. Pp 76-90.

RAJANAIDU N., KUSHAIRI D., RAFII M.Y., MOHD DIN A., MAIZURA I., ISA Z.A. and JALANI B.S. (2000). Oil palm genetic resources and their utilization - a review. International Symposium on Oil Palm Genetic Resources and utilization. Eds RAJANAIDU N. and ARIFFIN D. 8-10 June 2000. Kuala Lumpur. Malaysia.

RAO, V., SOH A.C., CORLEY, R.H.V., LEE C.H., RAJANAIDU, N., TAN Y.P., CHIN C.W., LIM K.C., TAN S.T., LEE T.P. and NGUI, M. (1983). A critical reexamination of the method of bunch quality analysis in oil palm breeding. Palm Oil Res. Inst. Malaysia. PORIM OCC. Paper 9: 1-28.