3 PLANT BREEDING31 Breeding strategies for increasing oil yieldOil palm is a

3. PLANT BREEDING3.1 Breeding strategies for increasing oil yieldOil palm is a allogamous highly heterozygous crop in which breeding strategies have been influenced by maize breeding, which relies on development of inbred parental lines to produce homogeneous F1 hybrids. Thanks to the above mentioned, oil palm conventional breeding has adopted reciprocal recurrent selection and family-individual selections methods for the development of parental lines that are used in commercial F1 hybrid seed production (Corley & Tinker, 2003). Since the early 1960’s most oil palm breeding was undertaken by seed producers and research institutions and has been focused on improving tenera (dura x pisifera) hybrids. As the expanding demand for palm oil is affecting directly tropical rain forests, oil palm breeding faces major challenge to increase the global average yield of 3.5 tons of oil per hectare to the full hypothetical yield potential estimated at 11–18t (Corley, 1998; Barcelos et al., 2015) The main methods used in oil palm breeding are Reciprocal recurrent selection (RRS; Fig. 4), Family and individual selection (FIS) and less commonly backcross breeding and pedigree selection. Figure 4: Example of RRS in oil palm (Cochard and Durand-Gasselin, 2018)Oil yield of the oil palm is a composite trait as its expression depends on a number of components, namely fresh fruit bunch yield (FFB), bunch and fruit quality traits. FFB yield is a yield function of the number of bunches produced annually (BN) and the average bunch weight (ABW). The percentage of fruit per bunch (% F/B) is another essential yield factor in this equation. Regarding fruit quality, the most important yield components are the relative proportion of mesocarp to fruit ratio (M/F), shell to fruit ratio (S/F) and kernel to fruit ratio (K/F). Additionally, the range in oil content in mesocarp and kernel is also a determining factor to the yield components and oil to mesocarp (O/M), Oil/Dry mesocarp weight (O/DM), and oil to kernel respectively (O/K) (Hartley, 1988; Oyoke et al., 2009; Sambanthamurthi et al., 2009; Pallas et al., 2013; Teh et al., 2016). Therefore yield improvement in oil palm is governed by a complex polygenic system, which complicates the transfer of single traits into elite cultivars (Sambanthamurthi et al., 2009). Best oil yields (t/ha/yr) have increased from 4.9 in 1962 to 9.6 t/ha/yr in 1988, representing a 93.2% increase (Lee and Toh,1991). However, the narrow genetic base of the Deli dura (originated from four Bogor palms planted in 1848) as well as most common pisiferas (also descendants of a limited number of origins) has halted the improvement of further yield increase since breeders found a low level of additive variation left in the Deli dura after several generations of selection. In order to enhance the genetic base of palm oil parental lines, both ex situ and in situ germplasm collections have been created to address the narrow genetic background by various research institutes and in 2010, FAO reported a survey of 21203 accessions of E. guineensis from 29 participating research institutions (FAO, 2010; Barcelos et al., 2015).Malaysian Palm Oil Board (MPOB) has the one of the largest gene bank for oil palm, with a collection of 100 000 palms from West Africa and Latin America in an area covering nearly 500 ha (Rajanaidu et al., 2018).3.2 Use of MAS to increment yields and speed up breeding Development of genomics has allowed the discovery and tagging of novel alleles and genes that can enhance the efficiency of breeding programs through their use in marker assisted selection (MAS)(Xu & Crouch, 2008). The use of molecular markers has been a mayor strategy to genetically improve the oil palm since the 1990s (Billotte et al., 2010). Rance et al. (2001) used the first genetic map of oil palm published by Mayes et al. (2007) using restriction fragment length polymorphism (RFLP) markers to identify quantitative trait loci (QTL) for fruit yield and its components. The publication of the oil palm genome combined with the development of high-throughput sequencing technologies, namely single marker polymorphism (SNP) and Genome Wide Analysis Studies (GWAS) have contributed to the rise of new opportunities regarding further increases on the yield of oil palm (Singh et al., 2013a and 2013b; Teh et al., 2016).After the identification of the single gene heterosis found in tenera palms carrying the shell gene in the 1940, mayor efforts took place to identify markers linked to the gene. The process of obtaining tenera palms for commercial plantations involves a the controlled pollination of dura and pisifera palms, however this process is far from error free in real world settings, and pollen contamination from wind or self-pollination often leads to the plantation of non-tenera palms for commercial production (Corley, 2005; Ramli et al., 2018). Recently, Singh et al. (2013b) were able to identify the SHELL gene, using a combination of genetic mapping tools as well as the publication of the whole palm genome. The identification of this gene now allows to routinely identify the three different fruit forms of E. guineensis and is found to be a type II MADS-box transcription factor, homologous to the Arabidopsis SEEDSTICK, which controls ovule, seed and lignified endocarp differentiation. Furthermore, the authors were able to proof that expression of SHELL was high in outer layers of the developing kernels in dura fruits, consistent with the proposed function in regulating the formation of heavily lignified shell surrounding the kernel (Singh et al., 2013b). This discovery enabled the development of the first molecular diagnostic assay for the oil palm named SureSawit SHELL which allows the prediction of the palm fruit form from leaves at nursery stage (Singh et al., 2014b). Ultimately, this tool works as a quality control tool to screen the populations and avoid plantation of unwanted planting material, therefore indirectly increasing yields per unit of land area (Ramli et al. 2018). Teh et al. (2016) showed how genomic tools can be used successfully to increase oil palm yield. After seven years of analyzing oil/dry mesocarp weight (O/DM) content of oil palm fruits of individual palms from Deli x AVROS and semi wild Nigerian x AVROS pisifera (59 breeding origins), the authors used GWAS and identified eight loci which are significantly associated to O/DM. Three positively correlating loci were found in Deli x AVROS groups as well as Nigerian x AVROS. Further analysis of the homologues of this candidate genes in or close to the significant SNP’s led to identification of a pyruvate kinase (pk) gene which could be associated to the O/DM. Based on this finding and further breeding trials, Teh et al. (2016) proved the value of these SNP’s, particularly SD_SNP_000019529, finding that when two parents carrying homozygous G/G SD_SNP_000019529 contributed the positive G to form a a 100% homozygous G/G tenera progeny which yielded an approximately 1% more O/DM contents. Furthermore, Teh et al. (2016) confirmed that it was possible combine (pyramid) the three important SNP’s in both groups to produce lines yielding up to 4% extra O/DM. A significant finding, as it has been calculated that an increase of 1% in mesocarp oil content could result in an increase of approximately 500,000 tons of oil annually in Malaysia (Lim, 1998). Future research into a further characterization of the genes in the vicinity to these SNP’s might help to identify the precise genes responsible for these increases of yield so they can be included into elite commercial cultivars.In another GWAS study (Bai et al., 2017), targeted for finding QTL associated with traits for oil yield, namely oil/bunch (O/B) and O/DM ratios. A linkage map was constructed for a tenera population, with a high density (average marker space of 1.02 cM). By QTL mapping in this population, they were able to identify one significant QTL associated with oil yield, and 3 suggestive ones, respectively Qoil_bunch_1.1, Qoil_bunch_8.1, Qoil_mesocarp_8.1, and Qoil_mesocarp_10.1. They stated that these are likely to be novel loci, but they might be specific to their breeding population. When identified from a MAS-perspective, it is interesting that an increased amount of the beneficial QTL (stacked) resulted in increased values for O/B and O/DM (Fig. 5). Thus, QTL pyramiding appears to be effective for the described QTL and may offer options for breeding and MAS. As a strategy for breeding for increased oil yield, stacking of these QTL in a commercial oil palm cultivar using MAS would be suggested. However, since the respective QTL have been identified and characterized in a particular breeding population of oil palm, verifying these QTL in another population would be important, as well as further characterization and annotation of the genes corresponding to these QTL.Oil palm is perennial crop which requires 2-3 years after germination of the seeds to produce flowers. When breeding efforts are made based only on phenotyping selection a breeding cycle can require from 10-19 years (Wong & Bernardo, 2008). Given the rise of molecular tools to understand plant genomes, molecular breeding uses genetic markers linked to the traits of choice to preselect desired phenotypes beforehand and has a potential to greatly hasten the breeding cycle, reducing costs (Barcelos et al., 2015). Based on data by Low et al. (2014), Ting et al. (2014) generated a set of 4,451 SNPs that were selected for the development of a customized oil palm Specific SNP array which allowed genotype across 199 Palms from two separate mapping F1 hybrid populations from all fruit forms of E. guineensis (dura, pisifera and tenera) as well as an interspecific cross from a Colombian E. oleifera X E. guineensis in less than 3 months. This experience also greatly improved marker density and genome coverage in comparison to previous reference maps based on amplified fragment length polymorphism and single sequence repeat markers (Ting et al., 2014). Figure 5: Oil to bunch content (A; O/B) and oil to dry mesocarp content (B; O/DM) in E. guineensis Jacq. tenera with combinations of beneficial marker genotypes at for QTL for oil content. For combinations of QTL, all possible combinations of Qoil_bunch_1.1, Qoil_bunch_8.1, Qoil_mesocarp_8.1, and Qoil_mesocarp_10.1 are used in the quantification of combined QTL (Bai et al., 2017). 3.3 Palm sizeDwarf oil palms are desirable as this trait makes bunch harvesting easier and more viable for mechanization, thus reducing production costs (Cochard and Durand-Gasselin, 2018). Furthermore, achieving slower/shorter growing palms will allow to expand the economic life of the plantation, as height complicates harvest so much that it is an important parameter to renew plantations (Singh et al., 2018). QTL’s have been found associated to palm height (Pootakham et al., 2015; Lee et al., 2015). Lee et al. (2015) identified a gene for asparagine synthetase within this QTL’s for palm height which has been associated to the dwarf phenotype of Arabidopsis while Pootakham et al., 2015 identified two separate candidate genes associated the gibberellin signaling path way to palm height, namely DELLA protein GAI11 and a putative gibberellin 2 oxidase 2. Dumpy Deli population is an important genetic asset as it is known for its dwarfness (Adon et al., 2001). An example of introgression of Dwarfness of Dumpy Deli can be seen in figure 6.Frond length is yet another limiting aspect that influences palm yield, therefore an important breeding objective. As the palm crop develops it reaches its full canopy width and therefore has a impact on palm spacing or plantation density which in turn affects harvest potential. Frond length in most contemporary cultivars ranges from 5 to 9 m which determines the optimum planting density around 143 palm/ha. Further increases of this parameter cause stress in the crop leading to abortion in female inflorescences and reduced yield (Breure et al. 1990; Forster et al. 2017). To round up, selecting for short and compact palms will allow to increment the palm yield productive cycle as well as reduce costs and increment yields. It is suggested to use the QTL’s for palm height (asparagine synthetase, DELLA protein GAI11 and a putative gibberellin 2 oxidase 2) to aid conventional breeding with recurrent selection. As there are no markers identified for frond length, it is suggested to phenotype and create markers for this traits in the most promising palms of our breeding program in order to be able to incorporate them in future cultivars. Backcross breeding can be used as an alternative breeding strategy as the introgression of this traits can bring negative yield impacts (Soh et al., 2006). Figure 6: Example of introgression of Dwarfness of Dumpy Deli and PO452T in improved populations (Cochard & Durand-Gasselin, 2018).3.4 Importance of tissue culture to breeding oil palm (yield)There are great expectations of the contribution of tissue culture to the improvement of oil palm yields through the multiplication of superior palm cultivars with improved characters (Ramli et al., 2018). Due to the oil palms botanical characteristics, vegetative propagation can be only achieved through tissue culture (Jones, 1974). Palm clones achieved through this technique are regarded to have greater uniformity and up to 30% higher yields compared to conventional seedling material which in turn has raised expectations in the palm industry for “next wave” of yield improvement through clonal planting material (Ting et al., 2013). However, tissue culture in oil palm has shown to be a complicated process with various pitfalls that have delayed its implementation: Low callogenesis rate of 19% (Corley & Tinker, 2003), average embryogenesis rate of between 3-6% (Ho et al., 2008) and somaclonal abnormality which leads to the production of clonal palms with abnormal fruit set and flower characteristics (Rohani et al., 2000). Recently, a major breakthrough was achieved by Ong- Abdullah et al., (2015) by identifying the retrotransposon Karma which lies in the EgDEF1 gene responsible for epigenetic variation that led to the “mantled” phenotype. This achievement has opened the possibility of developing diagnostic assays that will allow to select and eliminate the abnormal mantled palms prior to cultivation in the fields; this would help improving yields by avoiding the plantation of unwanted elite clones form tissue culture which carry the mantled phenotype. 3.5 Importance of GM approachesTransgenic crops offer the possibility of creating new varieties for breeders that can be either difficult or impossible to create through conventional breeding methods (Murphy, 2018). One of the most desired traits to modify in the oil palm has been its oil composition, in order to create niche novel oil varieties, especially since “tropical oils” used to be perceived to have nutritional disadvantages given the relatively high levels of saturated oils. By manipulating the acyl content at will, GM technologies are believed to enable production of high stereate varieties for solid fat markets, or ultra-high laureate for cosmetic and cleaning products (Barcelos et al., 2015; Murphy, 2014; Sambanthamurthi et al., 2009). Another important target for GM palms would be to significantly increase the oleic acid content in the mesocarp, to further increase actual average contents of 40-45% (Murphy, 2018). Other genes of interest that have been researched for the oil Palm are the expression of Bt insecticidal proteins (Lee et al., 2006) and the possibility of modifying lignin biosynthesis to achieve higher resistance to the fungal pathogen Ganoderma boninense (Paterson et al., 2009). The progress of addressing any of these traits have been slow and limited by various technical difficulties regarding tissue cultures techniques required as well as GM approaches, high costs, scarce resources as well as concerns to add further controversy surrounding oil palm crop, which already suffers from an image problem specially in European countries (Murphy, 2018). 3.6 Breeding strategyFull use of the oil palm genetic diversity has to be exploited to improve yields, thus using the broad gene pool present in MPOB should be a starting point. Altogether, it is recommendable to combine reciprocal recurrent selection with marker assisted breeding in order to target increased yields, and enhance the speed of the breeding cycle. Backcross breeding is an complementary method that should aid the introgression of important traits that potentially carry negative consequences. Tissue culture should be used to propagate the elite materials, using Karma retrotransposon diagnostic assays to avoid mantled phenotypes. SureSawit SHELL method should be used to asses the plantlets to check avoid illegitimate accidental crosses in the plantations. 4. Further considerations: physiology and breedingChanging key enzymes, alleles or genes is not without consequences. By engineering the oil palm the proportions of intermediates and metabolites are changed. Therefore, since the majority of palm oil is produced in the mesocarp, the genes to be engineered should be targeted in the mesocarp using a tissue specific mesocarp promoter. By making use of the Mesocarp-specific promoter (MT3-A) (Akmar & Zubaidah, 2008) the genes will only be expressed spatially and at the time of oil synthesis and thereby reducing risks.