This invention relates to a process for obtaining breeding lines of oil palm, and to lines and hybrids obtainable thereby Oil palm is a major world crop. It provided one-quarter of the world's production of vegetable oils in 2007. As well as providing edible oil, it has many other uses;

including in industrial lubricants and cosmetics. It is of growing interest for the production of biodiesel. Two main oil palm plant species are grown commercially: Elaeis oleifera Kunth and Elaeis guineensis Jacq. The historic origin of the oil palm (Elaeis guineensis) is understood to be West Africa, where it has been cultivated for many years: the species was introduced from West Africa to the Pacific region in the first half of the last century, since when it has been widely cultivated throughout tropical regions. Two countries in which it is currently widely grown are Malaysia and Indonesia.

The global production of palm oil has increased 6.4 fold between 1980 and 2005 from 4.5 to 29 million tonnes per year (according to figures from the US Department of Agriculture). This increase in production has been largely due to planting more land rather than increases in yield per hectare. Planting more land means felling rainforests, raising concerns over ecological impact. There is a pressing need to increase yield per hectare. The genetic improvement of oil palm is therefore a target of the first economic importance. In other crops, productivity has improved vastly over the last century due to selective breeding, and in particular the use of V _\ hybrids. V _\ hybrids are now the preferred planting material for numerous crops and remain the ultimate goal for many crop species. This is because of the spectacular yield increases that have been achieved by plant breeders and farmers growing commercial quantities of V _\ hybrid seed (see for example the figures for maize in USA given in the next paragraph). V _\ hybrids typically exhibit heterosis (hybrid vigour) and are genetically uniform, allowing optimization of agronomic practices (precision farming). Fi hybrid varieties are created by large scale crossing between two complementary and homozygous parents to generate Fi hybrid offspring that are heterozygous, genetically uniform and elite. For example, in 1933, only about 1% of maize grown in the USA was hybrid, a value that has now increased to 95% due to the greatly increased yields achieved; during the period from 1920 to 2000, the yield of hybrid corn rose from about 25 to 140 bushels per acre. The essential requirement for such hybrids is an adequate supply of homozygous parents: enough to identify pair-combinations that produce elite Fi hybrids. In monoecious maize this was achieved by recurrent self-fertilization; but this is impractical for many out-breeding, long-lived species. Oil palm is particularly problematic because of the long generation time (minimum of four to five years). The ten years needed to develop a new homozygous parent from an annual crop such as maize through 6-8 cycles of inbreeding would become, for oil palm, nearly half a century. Furthermore, individual palms normally produce exclusively either male or female flowers at any one time, requiring pollen to be stored for pollination onto a female inflorescence when one develops. Despite all this, attempts have been made to produce homozygous oil palm by inbreeding, but without any success.

In most species the inbred parents of commercial Fi hybrids have traditionally been highly homozygous rather than completely homozygous parents: e.g. the parents of maize hybrids were derived by inbreeding and therefore were not completely homozygous and not from doubled haploids. The use of highly homozygous genotypes of oil palm will give substantially uniform Fi hybrids, with similar improvements in uniformity or vigour to those found in other species. A recently published patent application (WO2008/114000) provides for the first time an accessible source of homozygous parents for oil palm Fi hybrids. It has the potential to revolutionise oil palm breeding. In this application, individuals of atypical phenotype are selected from large populations of oil palm, and analysed for ploidy and

heterozygosity. The ploidy analysis turns up a small proportion of haploids: these are converted by conventional means (for example, colchicine treatment) to homozygous doubled haploids, suitable for use as homozygous parents in producing genetically uniform hybrids. Occasionally haploid palms will double spontaneously. In addition zygosity analysis indicates that a small proportion of the seeds from self or open pollinations are highly homozygous (HH), even completely homozygous. These (if necessary after further zygosity testing) are taken to be doubled haploids (DHs) or classed as highly homozygous depending on the extent of homozygosity (see definitions below). Both HH and DH lines are valuable for breeding purposes.

Oil palm is diploid (2n), with 18 pairs of homologous chromosomes (2n = 2x = 36). The thickness of the lignified endocarp (seed shell) of oil palm fruits is a property of great commercial importance. It is under control of a single gene. This gene has two allelic forms, referred to as Sh and sh. The original cultivated oil palms all possessed thick endocarps and are termed 'dura ' . They are homozygous for the endocarp thickness gene, with genotype Sh/Sh. The original cultivated dura palm varieties exhibited a low oil extraction ratio (OER) in the range of 12-16%. However, a mutant allele of the endocarp thickness gene was later discovered, termed sh. This yielded virtually no lignified endocarp in its homozygous form, sh/sh, giving rise to a type of palm known as pisifera, but in the heterozygous state (Sh/sh) gave rise to an

intermediate shell thickness that allowed higher oil extraction and more than doubling of harvestable yields. Such palms are termed 'tenera'. Modern commercial oil palm varieties are all of the heterozygous 'tenera' type.

Tenera palms are frequently produced on the commercial scale from crosses between dura (Sh/Sh) and pisifera (sh/sh) palms. Pisifera palms are often female-sterile and therefore commercial crossing is done predominantly with dura as the female parent and pisifera as the male parent.

Such crosses would be predicted to produce 100% tenera progeny, since each cross must include both an Sh and an sh allele. In practice, this is not always attained: a few off-types (typically 'dura') can be found in dura x pisifera progeny. These are generally regarded as being the result of erroneous pollination, e.g., by wind-borne pollen from distant plants. However, an alternative explanation could be that they represent haploids or doubled haploids generated from unfertilized dura egg cells. The present invention however derives from the unexpected observation that a small proportion of these atypical progeny are highly, even completely, homozygous. A completely homozygous individual is a doubled haploid. According to the present invention we provide a method for obtaining oil palms useful as breeding stock which comprises selecting non-tenera progeny from a population obtained by crossing dura with pisifera oil palm, and testing the selected progeny for homozygosity, discarding heterozygous oil palms. Preferably the oil palms are also tested for ploidy, for example by flow cytometry, and only diploid palms are retained. Preferably the retained palms are completely homozygous (doubled haploids).

Highly homozygous oil palm plants obtained by the process of the invention may be preserved and propagated as breeding lines by self-fertilization or by clonal propagation. They may be used to produce hybrids of uniform properties by crossing with other highly homozygous lines, which conveniently have complementary properties. Hybrids typically exhibit improved properties, e.g., hybrid vigour. Choosing the most suitable parent lines to cross in order to obtain superior properties is generally initially a matter of trial and error but can take into account complementary characteristics if data is available. Loci contributing to heterosis (hybrid vigour) may be found subsequently and aid parental selection.

Non-tenera progeny may be selected in one of two main ways: by phenotypic or genetic marker tests. At immature stages of palm development, it is generally not possible to distinguish between the phenotypes of tenera, dura and pisifera: it is necessary to grow the population to maturity (fruit production), at which stage the phenotypic differences in the shell thickness of fruit can readily be observed and non-tenera individuals selected. This may take 2-4 years after field planting: but since most of the population is tenera, and is useful directly as such, this does not impose much extra burden or inconvenience on the oil palm producer.

Nevertheless it may sometimes also be convenient to select the non-tenera types at an earlier growth stage, for which purpose genetic marker testing may be used. The invention is further illustrated by reference to the accompanying drawings and the Example. In the drawings:

Figure 1 is a graph showing a homozygosity analysis of 140 palms comparing results obtained with 31 markers and with 111 markers;

Figure 2 is a block diagram of the process of analysing the samples;

Figure 3 shows dura, tenera and pisifera fruit, in section.

Example 1

To produce tenera palms suitable for commercial production of oil palm, flowers of dura palms were pollinated with pisifera pollen, using standard techniques. Both parents were from Indonesia. Tenera trials of dura x pisifera crosses were screened for non-tenera palms. This involved 13 dura x pisifera crosses and a total of 50,899 palms that were grown to maturity. Three years after field planting (five years from pollination) the resulting palms began to mature, flower and set fruit. They were inspected at between 10 and 15 years of age: Figure 3 shows typical characteristics of pisifera, tenera and dura fruit. Nearly all the palms were tenera, the expected progeny of a dura/pisifera cross. However, a small proportion was classified as non-tenera, either dura (if fruit type could be ascertained) or sterile (no fruit produced): no pisifera types were observed in this test. 191 non-tenera palms were found among the 50,899 screened. 153 of these non-tenera palms were then analysed according to the scheme shown in Figure 2. First they were tested for ploidy by flow cytometry, as described below. Flow Cytometry

Plant materials

Leaf samples of about 1 cm were taken from confirmed haploid and diploid plants with a chromosome count of 16 (n) and 32 (2n), respectively. These were used as standards.

Analyte preparation Cell suspensions (analyte) were prepared according to the method of Arumuganathan and Earle (Arumuganathan K, Earle ED, Estimation of DNA contents of plants by Flow Cytometry, Plant Molecular Biology Reporter 1991; 9(3): 229-233, 1991) with the following modifications. Leaf samples (about 1 cm ) were sliced by chopping with a sharp clean razor-blade (20-30 chops), in a plastic 9 cm diameter Petri dish containing 1.5 ml of cold (5 °C) CyStain® UV Ploidy solution modified by addition of 6.48 mM dithiothreitol (DTT) and 1% (v/v) polyvinylpyrrolidone (PVP-40) (Sigma- Aldrich, USA). The addition of DTT and PVP-40 were found to reduce background counts ('noise') in output histograms of particle fluorescence in the analyte. Slicing was done in low light conditions as DAPI is light sensitive. The cell suspensions were then incubated at 5 °C for 1 hour in the dark. Each suspension was filtered through a 40 μιη Cell Tries® (Partec, Germany), sieved into a 3.5 ml x 12 mm Rohren tubes (Sarstedt, Germany) and a further 0.5 ml of cold (5 °C) CyStain® UV Ploidy solution was added. The final suspension was then mixed thoroughly by pumping analyte in and out of a pipette.

The samples were run in a flow cytometer equipped with a UV laser at 375 nm (for DAPI fluorescence). We used either a CyFlow® space flow cytometer or a CyFlow® Ploidy Analyzer (Partec, Germany). Control oil palm samples of known ploidy level (chromosome number: haploid = 18; diploid = 36) were used to adjust and set the fluorescence channels. The diploid signal was set to give a read-out at or about the 200 position on the x-axis; the location of the haploid signal was then confirmed at the 100 position. Samples were run through the flow cytometer, at a speed of 0.5 μΕ/s, for 1 min. The embedded FloMax® (Partec, Germany) software was used to generate histograms, and carry out coefficient of variation calculations.

Ploidy results so obtained are shown in Table 1 below. 8 plants were found to be triploid, the remaining 147 being diploid. Plants identified as diploid were tested for zygosity. The first screen used a cascade of 10 highly polymorphic microsatellite or simple sequence repeat (SSR) markers, which are shown in Table 2 below. Plants were tested with the first SSR marker: those found to be homozygous using this marker were taken forward to be tested with the second SSR marker: and so on until all 10 SSR markers had been used. This set of 10 markers represents genetic loci on different chromosomes. Using this set of 10 markers, 29 out of the 153 non-teneras sampled proved to be homozygous and deemed candidate HH or DH (see Table 1 below). All of these were sent for external testing with an additional 32 markers that cover the entire genome of oil palm and represent a marker on each chromosome arm. A third round of genotyping with more mapped markers may be required to ensure adequate coverage of informative markers. From these data, highly homozygous lines and doubled haploid lines may be identified as defined below.

Table 1

1 2 B 2 1 1 1 2 B B 2n

1 2 B 2 B 1 1 2 B 1 2n

1 2 B 1 1 1 1 2 B B 2n

1 B 1 1 1 1 1 B 2 1 2n

1 B 1 1 1 1 1 B 2 1 2n

1 1 1 1 1 1 1 B 2 1 2n

1 1 1 1 1 1 1 B 2 1 2n

1 1 1 1 1 1 1 B 2 1 2n

1 1 1 1 1 1 2 B 2 1 2n

1 1 1 1 1 1 2 B 2 1 2n

1 1 1 1 1 1 2 B 2 1 2n A 1 1 1 1 B 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 B 1 B 1 1 2n A 1 1 1 1 B 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 B 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n

1 2 1 1 1 1 1 B 1 1 2n

1 2 1 1 1 1 1 B 1 1 2n

1 2 B 1 1 1 1 B 1 1 2n

1 2 1 1 1 1 1 B 1 1 2n

1 2 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n

1 2 1 1 1 1 1 B 1 1 2n

1 2 1 1 1 1 1 B 1 1 2n

1 2 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n A 1 1 1 1 1 1 1 B 1 1 2n

1 1 1 2 2 2 1 B 1 1 2n

1 2 1 1 1 1 1 B 1 1 2n 75 1 2 1 1 1 1 1 1 1 1 2n

76 A 1 1 1 1 1 1 1 1 1 1 2n

77 1 1 1 1 1 2 1 1 1 1 2n

78 1 1 1 1 1 2 1 1 1 1 2n

79 1 2 1 1 1 2 1 1 1 1 2n

80 1 2 1 1 1 2 1 1 1 1 2n

81 1 2 1 1 1 1 1 1 1 1 2n

82 1 2 1 1 1 1 1 1 1 1 2n

83 1 2 1 2 2 2 1 1 2 1 2n

84 1 2 1 1 2 2 1 1 2 1 2n

85 1 2 1 2 2 2 1 2 2 1 2n

86 1 2 1 2 2 2 1 2 2 1 2n

87 1 2 1 1 2 2 1 1 2 1 2n

88 1 2 1 1 2 2 1 1 2 1 2n

89 1 1 1 2 1 1 1 2 2 1 2n

90 1 1 1 2 1 1 1 2 2 1 2n

91 1 1 1 1 2 1 1 1 2 1 2n

92 1 1 1 1 2 1 1 1 B 1 2n

93 1 2 1 1 2 1 1 2 B 2n

94 1 2 1 1 2 1 1 2 B B 2n

95 A 1 1 1 1 1 1 1 1 B 1 2n

96 A 1 1 1 1 1 1 1 1 B 1 2n

97 1 1 1 1 1 1 1 2 B 1 2n

98 1 2 1 1 1 1 1 2 B B 2n

99 A 1 B 1 1 1 1 B 1 B B 2n

100 1 2 1 1 1 1 1 1 B B 2n

101 1 1 1 1 1 1 1 2 B B 2n

102 1 1 1 1 1 1 1 2 B B 2n

103 1 2 1 1 1 1 1 1 B B 2n

104 1 2 1 1 1 1 1 1 B B 2n

105 1 2 1 2 1 1 1 B B 2n

106 1 2 1 2 1 1 1 B 1 2n

107 1 2 1 2 1 1 1 1 B 1 2n

108 1 2 1 2 1 1 1 1 B 1 2n

109 1 2 1 2 1 2 1 1 B 1 2n

110 1 2 1 2 1 2 1 1 B 1 2n

111 1 2 1 1 1 2 1 1 B 1 2n

112 1 2 1 1 1 2 1 1 B 1 2n

113 B 2 1 2 1 1 2 2 1 1 2n

114 2 2 1 1 1 2 2 2 1 2 2n

115 1 2 1 1 2 2 2 1 2 3n

116 1 2 1 2 1 2 2 2 2 2 3n

117 1 2 1 1 2 2 1 2 1 2n

118 1 2 1 2 1 2 1 2 2 2 2n

119 1 2 1 2 1 1 2 2 2 1 2n

120 1 2 1 2 1 2 2 2 2 2 2n 121 1 1 1 1 1 2 2 2 1 1 2n

122 1 2 1 1 1 2 2 2 1 1 2n

123 2 2 1 1 2 2 2 2 1 1 2n

124 1 2 1 1 2 2 2 2 1 1 2n

125 1 2 1 2 2 1 2 1 B 1 2n

126 1 2 1 2 1 2 2 2 1 1 2n

127 1 2 1 2 2 2 2 2 1 1 2n

128 1 2 1 2 1 2 1 2 1 2n

129 1 2 1 2 2 2 2 2 1 2n

130 1 2 1 2 1 2 2 1 1 1 2n

131 2 2 1 2 1 2 2 1 2n

132 B 2 1 2 2 1 2 1 B 1 3n

133 2 2 1 1 2 2 2 B 1 2n

134 2 2 1 2 1 1 1 1 2 B 3n

135 2 2 1 2 1 1 1 B 2 B 2n

136 1 2 2 2 2 2 1 1 2 1 3n

137 1 B 1 2 B 1 1 1 2 1 2n

138 1 2 2 2 2 2 1 1 2 1 2n

139 1 2 2 2 2 2 1 1 2 1 2n

140 1 2 1 1 1 2 1 1 2 B 2n

141 1 2 2 2 1 2 1 1 2 1 2n

142 1 2 1 2 1 2 1 1 2 1 2n

143 1 2 2 2 2 2 1 1 2 1 3n

144 1 2 2 2 B 1 1 1 2 1 2n

145 1 1 1 2 2 2 1 1 2 1 2n

146 1 1 2 2 2 2 1 1 2 1 2n

147 1 2 1 2 1 2 1 1 2 1 2n

148 1 2 1 2 1 1 1 1 2 1 2n

149 1 2 2 1 1 2 1 1 2 1 2n

150 1 2 1 1 2 2 1 1 1 1 2n

151 1 1 2 2 2 2 1 1 2 1 3n

152 1 2 1 2 1 2 1 1 2 1 2n

153 2 2 1 2 1 2 1 1 2 1 2n

1 = Homozygous 2 = Heterozygous

B = Blank result 2n = Diploid

3n = Triploid

A = plant Advanced to second stage of zygosity screening Table 2

5 Lines which are homozygous for most, but not necessarily all, loci may be classed as highly homozygous (HH). The following definitions are used:

1. An individual palm is classified as highly homozygous (HH)when it is scored as having 20 unlinked homozygous loci where the parent with the same shell thickness phenotype is heterozygous at each locus analyzed. Since dura DHs 10 are expected to arise from a dura haploid set of chromosomes, only markers heterozygous in the dura parent are informative. Likewise, pisifera DHs require 20 heterozygous unlinked markers in the pisifera parent to be homozygous.

2. An individual palm is classified as doubled haploid (DH) when it has 28 unlinked homozygous loci regardless of the parental genotypes.

3. 'Unlinked' refers to loci that are on separate linkage groups (chromosomes) or where the minimum distance between any two loci in the same linkage group is

50 centi-Morgans.

The class of palms that are highly homozygous (HH) includes the class of palms that are completely homozygous (DH, doubled haploid), doubled haploids being a special case of HH.

To date, one palm (no. 70 in Table 1) has been tested with a total of 29 markers without showing any heterozygosity. It qualifies as doubled haploid (DH) according to the definition given above.

Table 3 gives probabilities of homozygous lines occurring by normal sexual processes (meiosis and fertilization). An individual is classed as HH when the probability of 20 or more unlinked markers, heterozygous in parental lines, all being homozygous is less than 1 in 1 million.

Table 3

10 1,024

11 2,048

12 4,096

13 8.192

14 16,384

15 32,768

16 65,536

17 131,072

18 262,144

19 524,288

20 1,048,576 Highly homozygous

21 2,097,152 Highly homozygous

22 4,194,304 Highly homozygous

23 8,388,608 Highly homozygous

24 16,777,216 Highly homozygous

25 33,554,432 Highly homozygous

26 67,108,864 Highly homozygous

27 134,217,728 Highly homozygous

28 268,435,456 Doubled haploid

29 536,870,912 Doubled haploid

30 1,073,741,824 Doubled haploid

31 2,147,483,648 Doubled haploid

32 4,294,967,296 Doubled haploid

In addition, we show in Figure 1 that using a relatively small number of markers (31) provides a good estimate of homozygosity: results from this set are highly correlated with data from a set of over 100 markers. Figure 1 is a homozygosity analysis of 140 palms. The results for a set of 31 markers are plotted along the x axis while those for a set of 111 markers are plotted along the y axis. The best straight line through the points is represented by the equation y = 0.8798x + 0.0827; the correlation obtained (R ) is 0.8759.