The rapid development of high throughput genotyping techniques makes them very cost effective for assaying large numbers of markers, which is useful in genetic mapping. In this study, the generation of SNP and SSR marker data was greatly accelerated (compared with conventional gel-based methods) using high throughput genotyping platforms. The Illumina Infinium customized OPSNP3 genotyping array and ABI3100 capillary-based fragment analysis allowed us to genotype the same set of SNP and SSR markers in two mapping populations simultaneously and data generation was completed within a short time. Genotyping of 252 SSR and 4,451 SNP markers across 199 palms (including parental palms) of P2 and OxG took less than three months. Furthermore both systems detect co-dominant alleles with differences as small as one base thus minimizing errors in scoring the genotype of markers. The high quality marker data helped in generation of marker linkage groups without obvious inflation of map distances that can be caused by false linkages . The genotyping cost was considered reasonable for the generation of data from a large set of samples and markers.
The genetic maps constructed in this study have thus successfully extended the coverage of the previous tenera map  and the dura and pisifera maps . A large set of SNP and SSR markers was used to replace the existing AFLP and RFLP markers to simplify the process of marker ordering in the genetic maps. The average gap between markers in the new maps (estimated as map length over the number of mapped markers) was greatly reduced from one marker every 7.2 cM to one marker every 1.6 cM in tenera (T), 5.4 cM to 2.4 cM in dura (D) and 5.5 cM to 2.1 cM in pisifera (P) maps. The gap was down to 1.4 cM in the DP intraspecific integrated genetic map from the previous 1.8 cM  for an integrated tenera x dura map. To the best of our knowledge, the parental and integrated maps presented in this study are the most comprehensive genetic linkage maps published for oil palm to date.
For the OxG interspecific cross, a previous effort failed to construct a map for the female parent, E. oleifera, mainly due to the limited number of AFLP and RFLP markers used . In this study, 10 small LGs with a majority derived from SNP markers were constructed for the E. oleifera parent. Although a large number of SNP markers and a significant number of E. oleifera-derived SSR markers were used, we still could not detect sufficient polymorphisms to saturate the map. This strongly suggests that the Colombian E. oleifera (used in creating the OxG hybrid) is very homozygous. Based upon this finding, future studies should include crossing Colombian E. oleifera with a distantly related E. oleifera in order to increase the level of heterozygosity in E. oleifera. Several rounds of such crossing with different E. oleifera may have to be done first, before generating the OxG hybrid. Nevertheless, this study has, for the first time, produced an E. oleifera map, albeit with only limited markers (65 markers/471 cM). Furthermore, the E. oleifera map was successfully integrated with the tenera parental map.
The established high density genetic maps proved useful for comparative mapping analyses. Previously, it was difficult to compare the published genetic maps of oil palm, as different labs used different markers. The markers used in this study also included the published SSRs thus enabling the current maps to be linked to the previously published maps [8, 9, 17, 23, 24]. Comparisons based on these anchoring markers is presented in Additional file 4. More importantly, SSR markers linked to QTLs for traits such as bunch number, bunch weight, fresh fruit bunch yield, fruit to bunch ratio, kernel to fruit ratio  and C16:0 and C18:1 , were successfully localized in the maps reported here. The mapping populations in this study also segregate for similar traits and the SSR markers will therefore be useful to indicate the LG locations of the QTLs and the nearby SNPs will help to reveal and validate similar QTL regions for yield components and FAC in P2 and OxG, respectively.
All genetic maps constructed in this study can be directly compared because of the high frequency of transferable markers. Approximately 75% of the mapped markers were cross-mapped in at least two parental maps of dura, pisifera, E. oleifera and tenera, and serve as good anchors in connecting the maps. The transferable markers will be good starting points to analyze other mapping populations and germplasm. The known map positions will be an added advantage, particularly for estimation of linkage disequilibrium (LD), assessment of genetic diversity and association mapping in oil palm. In oilseed rape for example, over 4,000 SNP markers derived from the genetic map were analyzed on different germplasm collections . Combination of the map information and genotyping data generated from the germplasm analysis allowed the determination of genetic diversity and the extent of LD. From the results, the authors also identified a smaller set of markers suitable for genome-wide association studies. This avoids the use of markers with an uneven or narrow distribution which can severely affect the association of markers to traits of interest. Furthermore, inclusion of transferable markers from specific map locations (such as targeted QTL regions) is useful in identifying and validating marker (s) linked to similar traits in germplasm collections. For example, in wheat, SSR markers associated with QTLs for plant height, spike length, grain yield and other traits on chromosome 4A, were also associated with the traits in germplasm collections .
This study reports the first comparative mapping analysis between intra- and interspecific hybrids of oil palm. The consistency of marker order and the limited map expansion observed made comparative mapping of the P2 and OxG crosses relatively simple. A high level of congruence between the genetic maps was observed, revealing 57 – 61% transferable markers between the two hybrid genomes. The transferable markers identified in each LG can help in identifying the corresponding genomic regions in other fruit forms of oil palm, leading to further fine mapping and structural analyses. In Brassica, common QTLs for plant height, flowering time and seed traits were identified in B. rapa, B. napus and B. juncea through synteny analysis . However, the study also examined the expression levels of genes between the species segregating for the QTLs and the counterparts that did not segregate for the same trait. This expression level information, when compared with the whole genome sequencing data of Brassica ssp and A. thaliana led to the discovery of several potentially useful candidate markers for explaining this variation. With the availability of the oil palm genome sequence , the anchor markers can be linked to the oil palm physical map. This will facilitate further saturation of the LGs where specific SSRs or SNPs can be designed in areas with large genetic gaps. In fact, the P2 genetic maps constructed in this study showed excellent co-linearity in marker order consistent with the oil palm genome sequence assemblies . Comparison with the whole genome sequence will be very helpful in identifying markers more tightly linked to traits than the ones discovered so far.
Comparative mapping also revealed that the distorted markers mainly occurred in clusters, possibly in the segregation distortion regions (SDRs) [47–50]. However, there is no intimation that SDRs are found at the same map regions or involve the same groups of markers across P2 and OxG. Instead, the SDRs are largely unique to each population and parental palm reflecting the respective polymorphisms and recombination events as has also been observed in potato  and barley . The SDRs also may indicate genomic regions involved in selection, e.g. of gametes in the maternal and paternal meioses. A greater tendency for segregation distortion has always been reported in interspecific, or wide crosses in plants [50–52]. Similar results were obtained in this study with the distortion rate higher in the OxG interspecific cross than in the P2 intraspecific cross. The higher segregation distortion in OxG is not surprising as it reflects the substantial genetic divergence and unequal levels of heterozygosity in the two parental palm species. The distortion segregation rate in P2 was similar to that observed in the E. guineensis intraspecific map .
In oil palm, SDRs have not been extensively studied compared to the situation in rice, wheat, maize and soybean where they are often associated with genes causing gametic competition , gametophytic selection (ga) [53–56] and sterility [54, 57, 58]. In maize, preferential fertilization has been frequently reported in pollen with the ga1 allele that mediates pollen – pistil interactions . This results in male gamete competition during transmission which might cause segregation distortion at the ga region. In rice, it was postulated that other chromosomal regions involved in pistil interaction might also interact with ga alleles and cause segregation distortion . However, the results on SDRs in P2 and OxG indicate that in many regions of the genome, there is a preference for one of the grandparental alleles due to some form of selection (either gametic or zygotic). Regions with segregation ratios close to 30% or 70% may contain incompatibility genes. These regions are of interest as they may give insight into the selection and evolution of a species .