Abstract

Preliminary results of a program to select dura female and pisifera male parents for generating oil palm planting material are reported with phenotypic values obtained for a set of 90 dura lines (out of a set of 225 dura lines) in Experiment 1 and 20 tenera x pisifera families (out of a set of 50 families) in Experiment 2. Experiment 3 provides data of 425 test cross families resulting from crosses involving all 225 dura and all 50 pisifera palms, which were assembled to estimate the General Combining Ability (GCA) values of all the parents of the above-mentioned dura lines and of the pisifera of the above-mentioned tenera x pisifera families. The 225 dura palms are all of the Deli type, derived from the main breeding stations in Southeast Asia; the origins of the pisifera palms are referred to as AVROS, Yangambi, La Me, Dami composite, Ekona, Ghana and Nigeria. Experiments 1 and 2 were planted with four replicates, in January 1996 and April 1996, respectively. Experiment 3 was planted with three replicates, in January 1997. The site of the evaluation is PT. Selapan Jaya's seed garden, named Bina Sawit Makmur (BSM), in South Sumatra. Besides data of the usual target traits, i.e. bunch yield and oil content, records include auxiliary traits which are associated with light interception (leaf area and crown disease incidence) and with the proportion of photosynthetic vegetative parts of the palms (leaf area/leaf weight ratio, palm height and frond dry matter production). For measuring height of, in particular, young palms a novel technique is developed which is revealed in the present report. For all traits the phenotypic values of the dura lines as well as the GCA values of their parents varied considerably. Provided that their heritability is high, this offers ample scope for the selection of dura lines as source of seed palms. Substantial variations were also found among the tenera x pisifera families, and, in particular, between the GCA values obtained from the pisifera palms with different origins. The pisifera of Nigeria origin were superior in terms of high oil yield with short stem height as well as other auxiliary traits associated with high Harvest Index. The report concludes with a detailed account of the current method of individual parent palm selection.

Introduction

The aim of the present paper is to report the progress with the selection of dura female and pisifera male parents for the production of high-grade oil palm planting material. The Selapan Jaya Group in South Sumatra obtained the breeding material from the diverse gene pool of ASD (Costa Rica) between 1995 and 1998. 

The current selection of the dura seed parents and pisifera pollen parents is first of all based on the target traits: bunch yield and the oil and kernel content. Characters indirectly associated with (high) oil yield per ha are used as auxiliary traits. The latter are classified into two groups. One group is related to gross CO₂ assimilation, and thus to dry matter production, and the second group to the proportion of dry matter incorporated in mesocarp oil and kernels (Harvest Index). 

The amount of gross CO₂ assimilation of a palm depends on the size of the light-intercepting leaf surface (the product of the number of green leaves on the palm and the mean area of these leaves). As gross CO₂ assimilation attains its maximum when the canopy is closed, a rapid closure is a primary objective of breeding. The rate of canopy expansion as well as the final size of the leaf surface (together with frond length indicative for optimal planting density) can be derived from the logistic growth curve fitted through the area of the youngest leaf as a function of palm age after planting (Breure, 1985). Due to the young age of the palms under study, only the mean of the available leaf area values (as a measure of extent of light interception) are considered in the present paper.

Evidently, a major condition for optimal light interception is reduced incidence or absence of crown disease. This common disorder appears as the bending of the newly opened leaves. Its incidence becomes manifest during the period until canopy closure. In this period, gross CO₂  assimilation increases proportionally with the amount of intercepted light. As a consequence, crown disease is detrimental for dry matter production and, hence, early yield (Breure and Soebagyo, 1991). Various papers reported that the sole method for improvement is to breed for resistance against the disorder (cf. Breure, 2003). 

Conventionally, breeding for a higher Harvest Index, i.e. the proportion of total dry matter used for the economic products, has been achieved by increasing the performance of the target traits as bunch yield and oil content. Currently, as followed in the present paper, some selection programs also focus on diminishing the production of vegetative dry matter (VDM). In passing it is noted that, since reducing VDM per se may negatively affect the leaf area and thus photosynthetic production, the reduction of VDM production must be restricted to the non-photosynthetic vegetative parts. In terms of standard growth measurements (Corley and Breure, 1982), selection should aim at a high leaf area/leaf weight ratio along with a low leaf dry matter production and a slow height increment (Breure, 2003). Additional advantages of slow (vertical) stem growth are reduction of harvesting costs and extension of the economic live span of a planting. 

The palms of the breeding program are still young. Consequently, unconventional recording methods are required to estimate two specific selection traits: (1) the height of the palm and (2) the production of vegetative dry matter. The usual reference point on the stem (being the height of the base of a specified leaf in the spiral of the first frond) measures the height of the growing point one to two years before the date of recording. The duration of this period varies due to considerable variation in the rate of leaf production. Thus, a novel method to estimate the height of the growing point of young palms is described (see Materials and Methods). 

Measurement of vegetative dry matter production is usually confined to the above-ground vegetative parts. In the present report solely the production of leaf dry matter (LDM) is taken, excluding the trunk dry matter. The latter component requires measuring the trunk volume, for which the diameter of the trunk and the annual height increment must be observed. Both records were not available for the palms under study. LDM provides a slight, but reliable, underestimate of VDM since the trunk represents only about 15% of the total above-ground vegetative dry matter.

Data on the primary and secondary selection traits (of above) were obtained from the first set of 90 dura lines (out of a total of 225 lines) and 20 tenera x pisifera families (out of a total of 50 families) in Experiments 1 and 2, respectively. Both experiments were planted within the frame work of the Bina Sawit Makmur seed garden project of the Selapan Jaya Group. Auxiliary traits are known to have a moderate heritability, in particular those associated with the architecture of the palm, such as leaf area/leaf weight ratio. Among the target traits, especially the heritability of bunch yield is known to be low (cf. Breure and Corley, 1983). For this reason, line and family evaluation on the basis of phenotypic values of the first set of 90 dura lines in Experiment 1 and 20 tenera x pisifera families in Experiment 2 (surely) benefits from additional information of the parents, which was obtained in Experiment 3. In the latter experiment, data were analyzed of 425 progenies resulting from crosses involving all 225 dura and all 50 pisifera parent palms in order to estimate the general combining abilities (GCA) or breeding values of their parents (cf. Breure and Bos, 1992). Comparison of the actual with the predicted performances of the test cross families allows estimating the interaction effect of the parents, termed specific combining ability (SCA). 

A comparison is made between the origins of ASD's oil palm material obtained from the Dami, Chemara, Harrisons & Crosfield, MARDI and Socfin breeding programs (female parents); and the origins referred to as AVROS, Yangambi, La Me, Dami composite, Ekona, Ghana and Nigeria (male parents). The main objective is, however, to show the wide variation in the selection traits among the phenotypic values of the individual families and the GCA values of their parents, and how both types of values are exploited in the selection of parent palms for seed production at Bina Sawit Makmur (BSM).

Materials and methods

The dura lines and tenera x pisifera families, together with 425 test cross families resulting from crosses among individual dura and pisifera palms, were introduced from ASD (Costa Rica) and planted at the BSM seed garden project in South Sumatra between 1996 and 1997 (see Breure, 1998 for more details). 

The 225 parents of the dura lines (i.e. selfings of ASD's dura palms) originate from several (locally spread) families which ASD obtained from breeding programs of Dami (117 dura palms), Chemara (68), Harrisons & Crosfield (18), MARDI (7), Socfin (2), along with 13 families due to crosses between palms in ASD's Harrisons & Crosfield and Chemara families. 

The origins of the 50 pisifera male parents used in Experiment 3 (see below) are the widely-used AVROS (15 pisifera palms), Yangambi (4) and La Me (1). New origins for Indonesia, being Dami composite (5), Ekona (10) and two families, GHA 608 (6) and GHA 648 (9), are included. The GHA families were developed at the Nigerian Institute for Oil Palm Research (NIFOR) and introduced to Costa Rica from Kade Oil Palm Research Centre in Ghana. GHA 648, being a selfing, and the female parent of GHA 608 are of Calabar origin, while the male parent of GHA 608 is derived from Ufama and Aba origins. All of the three origins have a long history of selection, starting from grove palms in Eastern Nigeria (Sparnaaij et al., 1963; van der Vossen, 1974). In the present report, pisifera palms of these two African families are referred to as Ghana (GHA 648) and Nigeria (GHA 608) origins, as used by ASD to describe planting material derived from male parents of these two families. 

The dura palms were used as the parents of selfings and the pisifera palms as the parents of tenera x pisifera families and as clones. In January 1996, the first set of 90 dura lines were planted at BSM with four replicates and 16 palms per plot (Experiment 1). These lines trace back to palms of the breeding programs at Dami (45 lines), Chemara (24), Harrisons & Crosfield (8), MARDI (5), Socfin (1), and families due to selfings and crosses between palms selected in the Chemara and Harrisons & Crosfield origins (7). The set of 20 full-sib tenera x pisifera families, obtained by pollinating 20 elite tenera palms by one pisifera palm belonging to the same family (sib-mating), were planted with four replicates and plots of 16 palms in April 1996 (Experiment 2). The origins of the pisifera male parents were AVROS (3), Dami composite (5), Ekona (5), Ghana (4), and Nigeria (3), all derived from a total of 12 tenera x tenera families at ASD (from some families more than one pisifera was obtained and two of the pisifera palms pollinated each two different tenera palms). 

To create a larger set of female and male parents for selection than as used in the present experiments (1 and 2), a surplus of palms of the dura lines and tenera x pisifera families in Experiments 1 and 2 were planted in separate blocks of various sizes. In this way, each dura line was represented by about 100 palms and seven tenera x pisifera families by more than 64 palms. 

The group of 425 families in Experiment 3, resulting from 425 crosses involving all 225 dura female parents and all 50 pisifera male parents (incomplete North Carolina mating design II), were planted with three replicates and plots of 16 palms in January 1997 (Experiment 3). The crossing scheme to generate the families was according to an alpha design (see Patterson et al., 1978), an incomplete block design: each dura palm (considered the treatments) was pollinated by two different pisifera palms and each pisifera palm (replacing the incomplete blocks) pollinated nine different dura palms. For the rationale of this design, using the following parameters (v=225, b=50, r=2, and k=9), reference is made to Breure and Verdooren (1995). 

Ablation, the removal of inflorescences, was carried out in 12 monthly rounds (Experiments 1 and 3) or in 9 monthly rounds (Experiment 2).

Observations 

Bunch yield was recorded in Experiments 1 and 2 from July 1998 to June 2003 (5 years), and in Experiment 3 from July 1999 to June 2003 (4 years). 

In all experiments, a sample of bunches was analyzed for the components of oil and kernel content, following the modified method of Blaak et al. (1963). The modifications were mainly that percentage oil/mesocarp was determined by the cold extraction method (Blaak, 1970). Novel is further that, instead of sampling a fixed weight or a fixed number of fruits, as described by Rao et al. (1983), a fixed-volume method was followed. In the latter method the total amount of fertile fruit from one spikelet sample is divided into eight lots, while from one (randomly drawn) lot a fixed volume is taken. The mean fruit weight obtained from the fixed-volume method approximates the actual mean fruit weight in a more reliable way than the conventional sampling methods. 

Standard leaf measurements (Corley et al., 1971; Hardon et al., 1969) were done in Experiments 1 and 2 on leaves marked at opening in June 1998, 1999, 2000, 2001 and 2002, and in Experiment 3 on leaves opening in February 1998, 1999, 2000, 2001, 2002 and 2003. 

The annual number of leaves produced per palm was recorded in Experiments 1 and 2 from June 1998 to June 2002 (4 years) and in Experiment 3 from February 1999 to February 2003 (4 years). 

Novel in the present report is the determination of the height of the growing point, here measured as the height of the point of insertion of the first leaflets on leaves which opened in December 2001 (Experiment 1) or in February 2002 (Experiment 3) or in May 2002 (Experiment 2). These height values incorporate the length of the petiole and should be corrected for it. Therefore, measurements took place about two months after leaf opening when the petiole had reached its final size. Height values were later on corrected for the length of the petiole, being estimated as 0.30 times the rachis length; the plot values of the petiole length, estimated in this way, were found to be highly correlated with the actual values (r = 0.82). The rachis length was measured as part of the standard leaf measurements.

Incidence of crown disease was recorded as the percentage affected palms, while the severity was scored on a total set of the eight youngest opened leaves (the youngest leaf in each of the eight spirals) at four-month intervals, using score 0 (absence of symptoms) to 3 (severe symptoms). Recording continued until no new symptoms were observed anymore.

Selection traits 

The above records were used to obtain the following selection traits (cj. Breure and Verdooren, 1995): 

• Bunch yield (kg/palm/year) 
• The percentages Mesocarp/fruit, Oil/mesocarp, and Oil/bunch 
• Oil yield (kg/palm/year) - the product of bunch yield and oil-to-bunch ratio 
• Leaf dry matter (kg/palm/year)- the product of the annual number of leaves produced and the mean weight of the leaves opening at the start and opening at the end of the year of recording 
• Bunch Index - for our purpose calculated as the proportion of bunch plus leaf dry matter (LDM) production used for fruit bunches. This parameter thus excludes trunk dry matter production. 
• Mean leaf area (m²) 
• Leaf area/leaf weight ratio 
• Height (cm) - at the level of the growing point, and used as a measure of the rate of height increment 
• Incidence and severity of crown disease. Only the incidence is reported in the present paper; the severity is taken into account for individual parent selection.

Strategy of the selection of dura and pisifera palms used for seed production 

First of all, the selection of dura seed parents and pisifera pollen parents focuses on the selection of their parental sources, being the dura lines (Experiment 1) and the tenera x pisifera families (Experiment 2), respectively. In addition to their phenotypic values, selection of these parental sources takes into account the General Combining Ability (GCA) values or breeding values of the parents, as estimated from the records of their progenies in Experiment 3. 

Then, within these identified sources of female (Experiment 1) and male (Experiment 2) parents, the search is for individual palms for generating (desirable) dura x pisifera planting material. 

By about 2004, when the recording program of the test cross families (Experiment 3) is completed and the final records of all dura lines and all tenera x pisifera families are available, the estimates of the specific combining ability (SCA) effects can be exploited by reproducing families which give substantially higher yield than the sum of the GCA values of their parents.

Results and discussion

Dura female parents 

Dura lines 

Table 1 presents, in the upper panel, the overall results obtained from the 90 dura lines in Experiment 1, along with the data patterns of the four main origins. 

As indicated by the minima, maxima and coefficients of variation, there is clearly a wide variation in the performance of the entire group of 90 dura lines, and among the dura lines within the main origins (see the standard deviations within the four origins). The variation is manifest in the phenotypic values of the target traits as well as in the values of the auxiliary traits: bunch index, height, leaf area, leaf area/leaf weight ratio and crown disease. 

Conventionally, dura selection in the Southeast Asian programs was mainly based on target traits as bunch yield and oil content. This policy can be traced back in the results. For example, all four origins have by and large equal values of bunch yield and percentage mesocarp/fruit. An exception is, however, the oil/mesocarp percentage which was clearly lower for the Dami origin, with a value of 45.7 (Dami) versus 50.4 (Chemara), 47.6 (H&C) and 51.5 (MARDI). This discrepancy between Dami and the other origins is in line with results of other studies comparing dura lines at ASD and is indicative for the emphasis on selection for oil content, especially in the Chemara and MARDI programs at that time. 

Partly in contrast with the selection program of the three other main origins, the Dami program also focussed on auxiliary traits at that time, such as (high) bunch index and (slow) height increment (cf. Breure et al., 1987). Relatively favourable values are obtained for these two traits in the Dami origin, with a Bunch Index of 0.408 and height of 136 cm, compared with 0.391 and 147 cm (the latter two values being means of the other three origins). The results prove that some progress has been achieved in the Dami origin.

Besides the relatively high GCA values for oil content and its main components (percentage mesocarp/fruit and percentage oil/mesocarp) of the MARDI origin (cf. the GCA data patterns per subgroup in Table 2), the variation in phenotypic values among the three other main origins, as presented in Table 1, is not reflected in the GCA values obtained from Experiment 3. Two factors may be responsible for the lower variation among the GCA values of the origins. One is the effect of selfing which tends to generate larger diversity among the dura lines in Experiment 1. The second factor is related to the mating design of our study, according to which each dura parent was pollinated by two different pisifera parents and each pisifera parent was crossed with nine dura parents. The latter factor renders parental selection based on GCA values obtained from progeny testing in Experiment 3 less reliable for the dura than for the pisifera parents. As a consequence, the breeding values of the dura palms (versus those of the pisifera palms) may to a great extent represent the interaction between the parents, termed specific combining ability (SCA). 

The selection among dura palms may, therefore, be more appropriate when based on the phenotypic values of the dura lines. This phenotypic selection approach, which is followed in the present program, is also effective when the focus is on eliminating deleterious recessive genes, since these are exposed in the selfings. 

Selection of seed palms 

By early 2003, recording will be completed for an additional 75 dura lines, which extends the total number of dura lines to (90 + 75) =165 lines. The larger number of lines allows a stricter selection of sources of female parent palms for seed production. 

The procedure to select individual dura palms within the selected dura lines is as follows. 

First of all, low-yielding palms as well as palms with severe incidence of crown disease, and/or excessive height, and/or undesirable visual characteristics (sterile fruit, boron deficiency etc.) are excluded. For the remaining palms, individual records of fruit components are listed. Palms with records of high mesocarp/fruit percentage and high oil/mesocarp percentage are marked for additional bunch analysis. Eventually, palms to be used as female parents are selected on the basis of bunch yield and mean analysis results of at least six bunches per palm. Selection will also be directed to favourable values for auxiliary traits, in particular for high leaf area/leaf weight ratio and small stem height. 

Pisifera pollen parents 

Tenera x pisifera families

The lower panel of  Table 1 presents the phenotypic values for the 20 tenera x pisifera families of Experiment 2, along with the specific data patterns for their five origins. The phenotypic values of most traits show a still greater variation than those of the dura lines (cf. the coefficients of variation of Experiment 1 versus Experiment 2). Indeed, selfing generates diversity among the dura lines, but the pronounced variation in performance among the families of Experiment 2 is not surprising in view of the distinct origins underlying the tenera x pisifera families. Further, the four new origins introduced to Indonesia (Dami composite, Ekona, Ghana and Nigeria) are superior to the widely-used AVROS origin, both in terms of the target traits and the auxiliary traits. Clearly, the palms of the AVROS origin are by far the tallest (246 cm), with the most frequent incidence of crown disease (42.2 %), along with the lowest values of Bunch Index (BI) and of the target traits as bunch yield and oil content. 

The superiority of the newly-introduced origins over the AVROS pisifera palms is confirmed by the GCA values obtained from Experiment 3, which are presented in the lower panel of Table 2. It should be noted that the results of Experiment 3 are more valid than those of Experiment 2 for evaluating the origins of the pisifera palms. Firstly, the higher number of pisifera palms tested in Experiment 3 as compared to the number of pisifera palms used as parents of the tenera x pisifera families in Experiment 2 (50 versus 20 palms) renders the data of Experiment 3 more meaningful than those of Experiment 2. Secondly, the GCA values of Experiment 3 more validly assess the relative performances of the origins than the tenera x pisifera families in Experiment 2. The reason is that in the tenera x pisifera families of Experiment 2, the performance is for 50% controlled by their tenera parent. 

When looking at the GCA values of oil yield per origin, one must conclude that the new introductions originating from Kade Oil Palm Research Centre in Ghana (the Ghana and Nigeria origins) are superior. When comparing the latter two, it is interesting to notice that the Nigeria origin combines high oil yield with two favourable auxiliary traits, namely, the height (being 147 versus 136 cm for pisifera palms of the Ghana and Nigeria origin, respectively) and high leaf area/leaf weight ratio (2.76 versus 2.54 for the Ghana origin). On the other hand, the GCA values of leaf area and rachis length are smaller for the Ghana origin (4.67 m² and 375 cm, respectively), which implies that a slightly higher planting density can be adopted for Ghana-derived planting material. Hence, the oil yield per ha of Ghana material may be relatively higher than estimated on an individual palm basis.

Elite pisifera 

The potential of an origin can be greatly improved by selecting the best pisifera, say, elite palm of an origin. Since the main interest is oil production, the highest GCA value for oil yield within the subgroup representing some origin is used as selection criterion. To obtain a general profile for each elite pisifera, GCA values of all the remaining target traits and auxiliary traits are also taken into account, as presented in the (lower panel) of Table 2. 

Clearly, the elite pisifera of the Nigeria origin combines the highest values of oil yield (31.5 kg/palm/year), Bunch Index (0.547) and leaf area/leaf weight ratio (2.78) with the lowest value for stem height (114 cm). All these traits are associated with a high Harvest Index. Note also its low value of leaf area of 4.79 m² compared with the average of the whole Nigeria group (5.06 m2) and also its relatively low value of rachis length (368 cm versus 379 for the average of the group). Interestingly, this elite pisifera palm (BSM 33) is the male parent of family 13. Family 13 (see the lowest line of Table 1) is characterized by the highest bunch yield, above average extraction rate, and the lowest palm height among all the 20 families in Experiment 2. Both on account of the GCA value of the pisifera male parent and the phenotypic values of its tenera x pisifera full-sib family, Nigeria family 13 is considered the most desirable source of pisifera male parents for seed production.

Selection of pisifera pollen parents 

Dura x pisifera planting material derived from pisifera in family 13 is expected to give a considerable improvement of the important characteristics: high oil yield combined with high Harvest Index and slow height increment. Obviously, the selection of pollen parents is restricted to palms producing male inflorescences. Since as yet no pisifera palms have been progeny tested, all pollen-producing palms among the 52 pisifera belonging to tenera x pisifera family 13 are used as male parent for seed production. At a later stage, GCA values will be obtained from a group of pisifera in this family, as done in Experiment 3. From the difference in GCA values between the mean of the five palms and the elite within the Nigeria origin, as found in Experiment 3 (see Table 2 ), one may infer that there is ample scope for further improvement of the male parents of this origin.

Acknowledgements

The author acknowledges the permission of PT Selapan Jaya to present these preliminary results of the BSM seed garden. The assistance of the staff of the BSM project, in particular Ir. Jati Cahyono (coordinator of the project) and Ir. Y. Puspitaningrum (statistician), mandors and field recorders is highly appreciated. Thanks are due to Dr. P.J.G. Keuss for his editorial assistance and to Dr. I. Bos for his valuable comments during the preparation of the paper.

References

Blaak, G. (1970). L'extraction de l'huile a froid dans l'analyse des regimes de palmier a huile. Oleagineux , 25, 165 - 168. 

Blaak, G., Sparnaaij L.D., and Menendez T. (1963). Breeding and inheritance in the oil palm ( Elaeis guineensis Jacq.) part 2 : Methods of bunch quality analysis. J.W. Afr. Inst. Oil Palm Res. 4, 146 - 155. 

Breure, C. J. (1985). Relevant factors associated with crown expansion in oil palm ( Elaeis guineensis Jacq.). E uphytica 34, 161- 175. 

Breure, C.J. (1998). Origin of ASD's Oil Palm Genetic Material Introduced in Indonesia and Method of Parent Selection. Proceedings 1998 International Oil Palm Conference, Bali, Indonesia, 174-188. 

Breure, C.J. (2003). The search for yield in Oil Palm - Basic Principles. In: The Oil Palm. Management for Large and Sustainable Yields. (Eds. T. Fairhurst and R. Hardter). Potash & Phosphate Institute/Potash Institute of Canada and International Potash Institute. 59-99. 

Breure, C.J. and Bos, I. (1992). Development of elite families in oil palm ( Elaeis guineensis Jacq.). Euphytica 64, 99 - 112. 

Breure, C.J. and Corley, R.H.V. (1983). Selection of oil palm for high density planting. Euphytica 32, 177-186. 

Breure, C.J. and Soebagyo, F.X. (1991). Factors associated with occurrence of crown disease in oil palm ( Elaeis guineensis Jacq.) and its effect on growth and yield. Euphytica 54, 55-64. 

Breure, C.J. and Verdooren, L.R. (1995). Guidelines for testing and selecting parent palms in oil palm. Practical aspects and statistical methods. ASD Oil Palm Papers 9, 1-68. 

Breure, C.J., Rosenquist, E.A.R., Konimor, J. and Powell, M.S. (1987). Oil Palm Introductions to Papua New Guinea and formulation of selection methods at the Dami Oil Palm Research Station. Proceedings of the international workshop on oil palm germplasm and utilisation. PORIM, Malaysia, 189-197 

Corley, R.H.V. and Breure, C.J. (1982). Measurements in Oil Palm Experiments. Internal report. Unilever Plantation Group, London. 35 pp. 

Corley, R.H.V., Hardon, J.J. and Tan, G.Y. (1971). Analysis of growth of the oil palm ( Elaeis guineensis Jacq.) I. Estimation of growth parameters and application in breeding. Euphytica 20, 307-315. 

Hardon, J.J., Williams, C.N. and Watson, I. (1969). Leaf area and yield in the oil palm in Malaysia. Experimental Agriculture 5, 25-52. 

Patterson, H.D., Williams, E.H. and Hunter, E.A. (1978). Block designs for variety trials. Journal of Agricultural Science, Cambridge, 90, 395-400. 

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. PORIM Occasional paper, No. 9 

Sparnaaij, L.D., Menendez, T. and Blaak, G. (1963) Breeding and inheritance in the oil palm ( Elaeis guineensis Jacq.), part 1: The design of a breeding program. J.W. Afr. Inst. Oil Palm Res. 4, 126-145 

Van der Vossen, H.A.M. (1974). Towards more efficient selection for oil palm. ( Elaeis guineensis Jacq.). Ph.D thesis, University of Wageningen, the Netherlands, 108 pp.