Posted on

single seed

To be effective a high-throughput MAS system requires simple and rapid DNA extraction methods [34] to be combined with simple and effective sampling strategies that permit high quality data return and tracking of information. In this study a high-throughput single seed-based sampling method was compared to a conventional leaf-based sampling method using DNA extraction protocols developed at Intertek-AgriTech ( for routine MAS applications in rice. The reported single seed-based sampling strategy for forward breeding applications of MAS should also be compatible with different DNA extraction protocols developed for rice in different studies [16, 24, 26, 31]. The quality and quantity of DNA obtained from these methods though should be validated as outlined in this study in order to ensure suitability for MAS deployment.

When the cost of genotyping is higher than those associated with generating a fixed line, MAS activities to determine homozygous genotypes for high priority traits in fixed lines are more cost-effective than in segregating generations. In this study we demonstrated that a single whole seed in rice can be effectively used for MAS, avoiding the complexity of seed chipping an asymmetrical rice seeds, and ensuring a high probability that the genotypic profile of the sampled seed matches the remnant seed from the same plant, as demonstrated by the high genotypic concordance rates reported in this study.

Additionally, an in situ large-scale comparison between the seed-based sampling strategy and the current routine leaf-based sampling protocol was conducted by sampling single seeds and leaf punches from 1466 F6 plants from the head rows in the line stage testing (LST) breeding class of IRRI’s irrigated lowland rice breeding program that are routinely prioritized for MAS (Additional file 1: Table S1). Lines in the LST class were grown in 2-row plots with 6 plants per row spaced 40 cm from each other at IRRI’s ZES field facilities in Los Baños, Philippines. During the implementation of each sampling activity, the number of workers, time spent on each step, and operational costs were recorded to compare the relative difference in cost-effectiveness between the two protocols.


Plant breeding programs producing inbred lines have two concurrent goals: (i) identifying parents for subsequent breeding cycles, and (ii) identifying new inbreds for varietal release [18]. Reducing the time to complete both activities is an effective way to increase the rate of genetic gain and efficiently deliver new varieties to farmer’s fields [30]. Integrated breeding approaches such as the combination of marker-assisted selection (MAS) and rapid line fixation (either through single seed decent (SSD; [6, 28] or double haploid creation (DH; [22] can be used to both increase selection efficiency and shorten the breeding cycle [35, 36]. In marker-assisted selection (MAS), molecular markers associated with favorable large-effect alleles are used as indirect selection criteria to improve breeding populations by deterministically increasing the frequency of specific high-value haplotypes in the breeding program [2, 29]. Rapid generation advance using SSD is an easy and cost-effective way to quickly attain line homozygosity in rice and effectively reduce the duration of variety development. In particular for self-pollinated crops, SSD methods are often cheaper than doubled haploid technologies since the later not only required high level of technical expertise but sophisticate tissue culture laboratories and facilities to generate large numbers of double haploid lines. The use of simple agronomic interventions can encourage early flowering [14, 35] reducing even further the time required in SSD methodologies to generate fixed lines. It’s common in many public plant breeding programs to impose marker assisted selection in the F2 generation in order to reduce the number of selection candidates handled by the program, however, closer to fixation (e.g. F6 or S6) the frequency of desired homozygous genotypes for two unlinked target loci more than triples (increasing from 0.0625 in the F2 to 0.235 in the F6 population). Thus, to identify 50 lines with the desired homozygous genotypes at two loci, on average 800 F2 individuals would be required compared to only 213 in the F6 population. As a consequence, the cost of genotyping is substantially reduced when MAS is conducted closer to line fixation. This is especially relevant as adding additional MAS targets increases population sizes exponentially. Further, leaf sampling F2 individuals is required as each plant is genetically unique and F2 derived seeds maternal and embryonic tissue DNA profiles are still segregating. However, imposing MAS at the F6 generation permits the use of whole seed sampling for genotyping with minimal risk of failure because each seed on the resulting panicle is nearly genetically identical.

The effect of seed developmental stage on the DNA quantity and quality was also compared between whole seed and leaf-based sampling strategies using a set of 14 accessions (Additional file 1: Table S1) replicated 5 times. For each accession a single panicle per plant was tagged at the day of its exertion and seeds were sampled at 7, 15, 25, and 30 days after panicle initiation (DAPI). These time points represent well described developmental phases in rice, namely: milky, dough, yellow-ripe, and maturity stages. At each time point, five seeds from one panicle per accession were sampled to make up 70 samples on a single deep-well plate. In addition, 5 leaf samples were also collected from each accession to be compared with the whole seed-based sampling results. Plants were grown in pots at IRRI’s Zeigler Experimental Station (ZES) screen house facilities in Los Baños, Philippines.

99%. This allows alternative tissue sampling protocols to be adapted for plants grown under field or greenhouse conditions. Likewise, a concordance rate of

An analysis of variance (ANOVA) was used to test for significant differences in the CT values of DNA extracted from leaf tissue and single seeds with contrasting physical or chemical grain properties. Custom R scripts (2018, R core development team) were used for calculating the call-rate and concordance between single seed and leaf-based genotype data. SNP call rates were calculated as the average proportion of successfully called genotypes for each SNP across all samples from different accessions and seed developmental stages. The SNP genotypic concordance rate was measured as the proportion of exact genotypic matches between identical SNPs genotyped on samples processed using the single seed and leaf-based tissue sampling protocols. ANOVA was calculated using the R function ‘anova’ [9]. Multiple comparisons were estimated using the Tukey’s HSD (honestly significant difference) [5] method using the R function ‘HSD.test’ from the R package ‘agricolae’ [25].

Tax calculation will be finalised during checkout.

The single seed descent (SSD) method of inbreeding minimizes the amount of genetic sampling. The single-pod descent (SPD) and bulk methods (BM) produce redundant inbred lines that are descended from either the same F2 or F3 plant. However, for soybean [Glycine max (L.) Merr.], the SSD method requires more time to process the seed than the SPD or BM. Our experiment is the first to compare the SSD, SPD and BM by sampling the same population in the field and then evaluating the methods using molecular markers. Our objective was to determine the relative efficiency of the SSD, SPD and BM procedures. We defined unique lines as those lines that were not paired with any other line at a coefficient of similarity (Sxy) level ≥ 0.875, which was an alike-in-state criterion. The efficiency was defined as the number of unique lines developed by each procedure. We genotyped 100 F4:5 lines from each of the three genetic sampling methods, using 21 polymorphic simple sequence repeat markers. The number of unique lines was the same for all three sampling methods at the 0.05 level of Type I error. Based on our criterion, the three sampling methods are equally efficient. Our conclusions were the opposite of all other previously published reports. Each breeder will have to determine the best method for generation advancement, based on the amount of resources required to harvest and process the seed.

Access options

Kervella J, Fouilloux G (1992) A theoretical study of the bulk breeding method: I. Importance and consequences of losses due to sampling. Euphytica 60:185–195

Liu B-H (1997) Statistical genomics: linkage, mapping, and QTL analysis. CRC Press, Boca Raton

Hanson WD (1959) Minimum family sizes for the planning of genetic experiments. Agron J 51:711–715