Genetic diversity of preserved rice seed samples from the Mikawa area, Japan, stored in the Meiji era.

Preserved rice (Oryza sativa L.) seeds stored for nearly a century as an emergency food stocks from the Mikawa area were investigated for their genetic diversity. Morphologically, the seeds appeared to be typical Japonica. One chloroplast INDEL petN-trnC, two nuclear INDELs Acp1 and Cat1, and three SNP markers in Starch synthase IIa were amplified to characterize the molecular profile. The efficiency of amplification varied among the markers. Most of preserved seeds were classified as Japonica, but some were identified as Indica. The heterozygous genotypes detected suggested a high frequency of outcrossing at that time. On the other hand, 21 SSR markers showed quite a high degree of amplification efficiency. Principal coordinate analysis and STRUCTURE analysis based on the SSR polymorphisms proved that the preserved seeds contained alleles that were not detected among current landraces and breeding varieties, and there were the expected three subpopulations among 96 preserved seeds. These results indicated that these preserved seeds from Mikawa area in Meiji era had high genetic diversity and consisted of some subpopulations including Indica landraces with typical Japonica seed shape. These lines were considered to have been lost from current genetic resources.

Introduction

Rice breeding in Japan began to be carried out on a large scale in the Meiji era (1868–1912). Many historical records indicate that various landraces began to be recognized at that time (Kushibuchi 1990). These landraces became valuable resources for modern rice breeding in Japan. However, newly established varieties were promoted and used instead of landraces, and thus genetic diversity among cultivated rice populations rapidly disappeared (Morishima and Oka 1995).

The materials used in this study were rice stocks for emergency food stored about 100 years go in the Meiji era by a farmer in Inabu-cho, Toyota city, Aichi prefecture (Ezaki 1969, Kondo and Okamura 1932). This component and eating quality of this preserved rice have been reported previously, and it was confirmed that germination ability had been lost (Ezaki 1969, Fukumoto , 2003, Kondo and Okamura 1932). However, the genetic background of these rice has not investigated. Assessment of genetic diversity using preserved rice allows direct insight into the breeding and dispersal of rice varieties in the past. Clarification of varieties that have adapted to the environment at a particular time may be informative for breeding and the conservation of genetic diversity in the present and future.

For phenotypic analysis, seed size variation has been used as a key indicator of phenotypic diversity (Watabe , 1980). Archaeological remains have also been a useful source for clarifying seed diversity or transitional shapes (Fuller and Allaby 2010, Wasano 1993). The grain shape index (ratio of the length/width of a rice grain) was established by Matsuo (1952) and defines three major categories: a-type (short type), b-type (large type), and c-type (long-type). These three types are considered to be typical grain shapes for Temperate Japonica, Tropical Japonica, and Indica, respectively, with a few exceptions, and grain shape is used as a key factor for distinguishing landraces (Oka 1958, Takeda 1990). Generally, Japanese landraces were Temperate Japonica types, with some exceptions (Ishikawa , 1992, Sato 1991, Takeda 1990).

For molecular based genetic analysis, a number of molecular markers have been developed as indicators of genetic diversity. By investigating 40 isozyme loci of 1948 rice strains from Asia and Africa, Second (1982) found that Indica and Japonica had been domesticated independently from their respective ancestors. Glaszmann (1987) found a discriminant marker for Indica-Japonica by polymorphic analysis of 15 isozyme loci and various phenotypic criteria among 1688 Asian rice strains. Ishikawa also found Indica-Japonica discriminant markers among 60 Asian rice strains through polymorphic analysis of isozymes and physiologic and phenotypic criteria based on Morishima and Oka (1981). Chen developed a specific cytoplasmic marker inherited maternally for distinguishing Indica or Japonica. Simple sequence repeat (SSR) markers were also developed and applied to confirm varietal differentiation (Garris ). These reports suggested that Indica and Japonica had maintained a close relationship during the process of domestication.

DNA extraction from the endosperm is possible even when germination ability has been lost (Muto ). This extraction method is applicable for not only varietal identification of milled rice grains (Akagi 2000) but also long-preserved seeds like those reported here (Kobayashi ) and carbonized seeds excavated from archaeological sites (Castillo ). Rice straw and chaff are also possible DNA sources (Tanaka , 2010). DNA analysis of ancient botanical specimens found among archaeological remains has been reported well; however, reports of samples from the middle ages to early modern era are very limited in number.

The objective of this study was to understand the genetic background of a preserved seed population. To estimate genetic variation of these preserved seeds, seed size variation, Indica-Japonica classification by polymorphisms among three insertion/deletion (INDEL) markers and three single nucleotide polymorphism (SNP) markers of isozymes, and population structure analyses by using the 21 SSR markers in the preserved seed population and three populations of current landraces and varieties.

Materials and Methods

Plant materials

Ninety-six preserved seeds of rice (Oryza sativa L.) landraces representing part of the seed population preserved in Inahashi-village in the northern Mikawa area (presently Furuhashi-Kaikokan, which is a private museum in Inabu-cho, Toyota-city, Aichi prefecture), Japan, were analyzed in order to estimate past genetic diversity. These preserved seeds were collected from the Nishikamo-district (present Miyoshi-city) in the western Mikawa area about 50 km from Inahashi-village in 1904 and 1906 as food stocks for redistribution in case of emergency. It is unknown how many landraces these consist of or their variety names. These seeds were heat-dried once for 10 hours at 50–58°C after being hulled and then packed in cans to preserve them (Kondo and Okamura 1932). The seeds were pale brown in color (Fig. 1). The seed sizes of the 96 preserved rice seeds were compared with six landraces (‘Shinriki’, ‘Aikoku’, ‘Asahi’, ‘Nagoyashiro’, ‘Houmanshindenine’, and ‘Shirosenbon’) and five modern varieties (‘Koshihikari’, ‘Akitakomachi’, ‘Sasanishiki’, ‘Nipponbare’ and ‘Gohyakumangoku’). The 96 preserved rice seeds and 100 seeds from each of the control varieties (six landraces and five modern varieties) were subjected to measurement of seed width and length to evaluate variance. The seeds were classified into three types based on the grain shape index, as developed by Matsuo (1952): a-type (short), b-type (large), and c-type (long).

Fig. 1.

Phenotypic variation of the preserved rice seeds from the Mikawa area in the Meiji era. Four seeds were randomly chosen from 96 preserved seeds.

To compare genetic diversity, three additional three representative current populations were also analyzed to estimate genetic variation in population structure of the preserved population (Supplemental Table 1). These representative current populations comprised 42 landraces that were used as parental landraces for modern breeding (parental landrace; PL), 42 landraces of non-parental landraces (NP), and 12 modern breeding lines (BL). These cultivars of current populations were selected to include broad genetic and geographic diversity. One individual represented each landrace or variety as one accession. All of these landraces and varieties were conserved at the Hokuriku Research Center of National Agricultural and Food Research Center (NARO), Japan.

DNA extraction

DNA was extracted individually from 96 hulled preserved rice seeds using the CTAB method reported by Muto . DNA from the 96 current rice accessions was extracted using an alternative CTAB method (Doyle 1991) from young leaves of single individual plants. To prevent cross-contamination, DNA experiments using the preserved samples and current samples were performed separately using a dedicated room, tools and reagents. All reagents for experiments were freshly prepared. Negative controls were always performed to check there was no contamination by modern DNAs.

INDEL and SNP polymorphisms identification

To characterize the molecular profile, one chloroplast INDEL, petN-trnC, two nuclear INDELs, Acp1 and Cat1, and three SNP markers in starch synthase IIa (SSIIa) were amplified. These markers were designed previously for the identification of Indica- or Japonica-specific alleles at each locus. All primer information is shown in Supplemental Table 2. Maternal DNA variation was estimated using a maternally inherited chloroplast INDEL marker, petN-trnC (Castillo ), which should be amplify a 75 bp product in Japonica and 127 bp in Indica. This marker was confirmed to be amplified from ancient DNA by Castillo . Nuclear DNA variation was estimated using two INDELs, Acp1-INDEL and Cat1-INDEL, as reported by Muto . The Acp1-INDEL was amplified as a 106 bp product in Japonica and 64 bp in Indica, whereas the Cat1-INDEL generated 146 bp and 122 bp products, respectively. PCRs were performed with a 10 μL reaction mixture containing 0.5 ng of template DNA, 0.025 U LA-Taq (Takara Co., Japan) per single reaction, 1× buffer, 0.1 mM dNTPs, 0.25 μM each forward and reverse primers. The amplification conditions were: 95°C preheating for 5 min, followed by 35 cycles of 96°C for 30 sec, 72°C for 1 min, and 75°C for 5 min. The PCR products were separated and detected using 3% agarose gels. Three SNPs (S2, S3, and S4) in the SSIIa gene, which were reported by Umemoto and Umemoto and Aoki (2005), were also genotyped. Among these three SNPs, SSIIa-S2 was used for Indica and Japonica differentiation. SNP-A of SSIIa-S2 was predominantly detected in Japonica and the other SNP-G was present in Indica. These SNP types were confirmed by their digestibility as CAPS markers with a particular restriction enzyme, Msp1, recognizing the SNP site. The Japonica type was detected as a single 90 bp fragment and the Indica type as two fragments (42 bp and 48 bp). SSIIa-S3 was composed of an alternative nucleotide, G or A type. SSIIa-S4 was composed of another alternative nucleotide, T or C. The composition of the PCR mixture was same as INDELs, and the reaction conditions were: 95°C preheating for 3 min, followed by 40 cycles of 95°C for 30 sec, 68°C for 45 sec, 72°C for 60 sec, and 72°C for 3 min. The PCR products were detected using 3% agarose gels. On the other hand, SSIIa-S3 and SSIIa-S4 were confirmed using alternative primers. The PCR reaction mixture was the same composition as SSIIa-S2, and the conditions were: 95°C preheating for 3 min, followed by 35 cycles of 95°C for 30 sec, 60°C for 30 sec, 72°C for 30 sec, and 72°C for 3 min. The PCR products were detected using 3% agarose gels. For the preserved population, the SNP patterns of SSIIa-S2, S3, and S4 were further confirmed by sequencing.

All the fragment patterns of these INDEL and SNP primer sets were preliminarily confirmed with typical Indica and Japonica landraces, comprising 40 accessions collected from Japan, China, and South-east Asian countries provided by the National Institute of Genetics, Mishima, Japan (Oka 1958, Supplemental Table 3).

SSR fragment analysis

To elucidate the genetic diversity of preserved seeds from the Mikawa area, a total of 192 accessions consisting of 96 preserved rice seeds and 96 current rice accessions, were analyzed for 21 SSR loci, which were selected from 12 chromosomes to clarify the general genetic characteristics (Supplemental Table 2). PCR was performed with a 10 μL reaction mixture containing 0.5 ng of template DNA, 0.2 U Ex-Taq (Takara Co., Japan), 1× buffer, 0.25 mM dNTP, 0.5 μL DMSO, 1 μM of each forward and reverse primer, and conditions were: 95°C preheating for 3 min, followed by 40 cycles of 96°C for 30 sec, 55–57°C for 45 sec, 72°C for 60 sec, and 72°C for 3 min. All PCR products were subjected to electrophoresis, and fragment sizes were determined using a QIAxcel system (QIAGEN).

Data analysis

SSR fragment data were analyzed using GenAlEx 6.5 (Peakall and Smouse 2012) to determine the number of alleles and allele frequencies, expected heterozygosity (He), observed heterozygosity (Ho), unbiased expected heterozygosity (uHe), and also subjected to principal coordinates analysis (PCA). In particular, uHe is used as a parameter to evaluate the genetic diversity among populations of different sizes. To infer the population structure of the 192 accessions, a model-biased program, STRUCTURE 2.3 (Pritchard ), was used. The optimum number of populations (K) was selected after 20 independent runs of a burn-in of 10,000 iterations, followed by length of 10,000 for each value of K (testing from K = 1 to K = 5) with the admixture model. ΔK was calculated using the software, Structure Harvester (Earl and vonHoldt 2012).

Results

Seed size classification

Seed mixtures from various sources are expected to show variations in seed size due to environmental effects. The preserved population appeared to be a mixture from various sources; therefore, we compared the variance of seed size through measurements of length, width, and their ratio (Table 1). Seed size was compared among the 96 preserved seeds, six landraces, and five modern accessions. Mean seed length and width were 5.1 and 3.0 mm in the preserved population, 4.8 and 2.9 mm in the landrace population, and 4.9 and 2.9 mm in the modern population, respectively. The preserved population had the largest length and width among the three populations, and the ratio of hull length and width (L/W) ranged from 1.3 to 2.0, with a mean of 1.7. The L/W ratio was 1.6 in the landrace population and 1.7 in the modern population. No slender-type seeds were found in the preserved population or the current populations. The results of classification by grain shape index are shown in Supplemental Fig. 1. All of the 96 (100%) seeds in the preserved population were classified as a-type, and no b-type or c-type seeds were evident. Additionally, all of the landrace and modern populations were classified as a-type. This result indicated that the shape of all the seeds in the preserved population was classified into the typical Japonica type, as for the landrace and modern populations.

Table 1.

Average seed size, length/width ratio and CV among the preserved population, six landraces and five modern varieties

Population	Variety name	n	Average seed size (mm)			L/W			CV (%)
			Length	Width		Min	Max	Average
Preserved	Unknown	96	5.1	3.0		1.3	2.0	1.7	7.04
Landraces	Shinriki	100	4.8	2.9		1.5	2.2	1.7	6.52
	Aikoku	100	4.7	3.0		1.4	1.7	1.6	4.07
	Asahi	100	5.0	3.0		1.6	1.9	1.7	4.27
	Nagoyashiro	100	4.7	2.9		1.5	2.0	1.6	5.71
	Houman shinden ine	100	5.1	3		1.5	2.1	1.7	4.75
	Shirosenbon	100	4.4	2.9		1.4	1.7	1.5	4.00
	Sub total	600	4.8	2.9		1.5	1.9	1.6	6.45
Modern varieties	Koshihikari	100	4.8	2.8		1.5	2.1	1.7	4.81
	Akitakomachi	100	4.9	2.9		1.6	2.1	1.7	4.94
	Sasanishiki	100	4.7	2.7		1.6	2.1	1.7	4.94
	Nipponbare	100	4.7	2.7		1.6	1.9	1.7	3.61
	Gohyakumangoku	100	5.0	3.1		1.5	1.8	1.6	3.95
	Sub total	500	4.9	2.9		1.6	2.0	1.7	5.19

The coefficient of variance (CV) of the seed size ratio was 7.04% in the preserved population, 6.45% in the landrace population, and 5.19% in the modern population. The CV was clearly highest in the preserved population, followed in order by the landrace and modern populations.

Indica-Japonica classification in terms of INDEL and SNP markers

The result of Indica-Japonica classification of genotype patterns among the preserved population is shown in Table 2. Many samples from the preserved seed population showed low PCR amplification efficiency. The amplification success rate was 24.0% for petN-trnC-INDEL, 84.3% for Cat1-INDEL, 29.1% for Acp1-NDEL, 56.3% for SSIIa-S2, 28.1% for SSIIa-S3, and 38.5% for SSIIa-S4. The chloroplast INDEL was expected to show a high success rate because of its high copy numbers. However, the highest amplification rate was found for Cat1-INDEL, and SSIIa-S2 also showed a high rate of amplification. The 26 presumed Indica samples detected by Cat1-INDEL were not amplified by the marker. Only 12 of 96 samples (12.5%) showed amplification of all six markers.

Table 2.

Genotype patterns among the preserved population using a three INDEL markers and three SNP markers. The genotypes of four markers, petN-trnC I-32, CatB 5ʹ UTL, Acp1 intron 2 and SSIIa-S2 were especially indicated Indica and Japonica, respectively

Gene name	Genome	Marker type	Product size (bp)	No. of seeds (%)
				Japonica	Indica	Heterozygote	NA	Total
petN-trnC I-32	Chloroplast	INDEL	75 or 127	23 (24.0)	0  (0.0)	0  (0.0)	73 (76.0)	96 (100.0)
CatB 5ʹ UTR	Nuclear	INDEL	106 or 64	37 (38.5)	26 (27.1)	18 (18.8)	15 (15.6)	96 (100.0)
Acp1 intron 2	Nuclear	INDEL	146 or 122	20 (20.8)	7  (7.3)	1  (1.0)	68 (70.8)	96 (100.0)
SSIIa-S2	Nuclear	SNP	90a	54 (56.3)	0  (0.0)	0  (0.0)	42 (43.8)	96 (100.0)
				S3/G	S3/A	Heterozygote	NA	Total
				S4/T	S4/C
SSIIa-S3	Nuclear	SNP	130	10 (10.4)	3  (3.1)	15 (15.6)	68 (70.8)	96 (100.0)
SSIIa-S4	Nuclear	SNP	116	16 (16.7)	13 (13.5)	8  (8.3)	59 (61.5)	96 (100.0)

NA: PCR amplification was not successful.

Product size before digestion with enzyme.

Among the 96 samples from the preserved population, 37 (38.5%) were of the Japonica type and 26 (27.1%) of the Indica type in terms of Cat1-INDEL, and 20 (20.8%) were of the Japonica type and 7 (7.3%) of the Indica type in terms of Acp1-INDEL. The Indica type was detected at high frequency. Heterozygotes were detected in 18 (18.8%) of Cat1-INDEL, 1 (1.0%) of Acp1-INDEL, 15 (15.6%) of SSIIa-S3, and 8 (8.3%) of SSIIa-S4. A total of 24 samples showing different genotypes among the markers were also detected. These results indicated that the preserved population included Indica and Japonica individuals, and also contained recombinant individuals between Indica and Japonica. There was an apparent tendency for smaller sized products to show higher amplification efficiency, suggesting that differences in amplification efficiency among markers were due to product size. However, this was not proven statistically, probably because of the small sample population size.

Genetic diversity of the four populations in terms of SSR polymorphisms

To determine the genetic diversity and structure of the preserved population, polymorphism analysis was conducted using 21 SSR markers among the preserved population, and three representative current populations, including PL, NP, and BL. PCR amplification was successful for 88.7 individuals (92.4%) in the preserved population for an average of 21 markers, being much more efficient than for INDEL markers. The expected product sizes ranged from 71 bp to 314 bp, with an average of 175 bp overall for the 21 loci examined. The genetic diversity of each population in terms of SSR polymorphism is shown in Table 3. The Ho in Table 3 is shown 0.00 for four populations. In fact, one heterozygote was observed at RM8208 in the preserved population and Ho was 0.001, but this is not shown in Table 3 because it was rounded to two decimal places. The index of genetic diversity within the population uHe is also shown in Table 3. uHe was adopted to evaluate populations of different sizes. The value was 0.40 for the preserved population, 0.48 for PL, 0.44 for NP, and 0.50 for BL. The preserved population showed genetic diversity equivalent to that of the other representative current populations.

Table 3.

Genetic diversity of preserved rice population and present accessions in Japan based on polymorphism of 21 SSR loci

Population		N	Na	He	Ho	uHe
Preserved	Mean	88.70	3.35	0.40	0.00a	0.40
	SE	0.76	0.37	0.06	0.00a	0.06
PL	Mean	39.40	3.95	0.47	0.00	0.48
	SE	0.44	0.36	0.05	0.00	0.05
NP	Mean	38.45	3.40	0.44	0.00	0.44
	SE	0.84	0.32	0.06	0.00	0.06
BL	Mean	11.05	3.05	0.48	0.00	0.50
	SE	0.27	0.26	0.04	0.00	0.04
Total	Mean	44.4	3.44	0.45	0.00	0.46
	SE	3.16	0.17	0.03	0.00	0.03

PL: Parental line of modern breeding landrace, NP: Non-parental landrace, BL: Modern breeding line, Na = No. of different alleles, He = Expected heterozygosity = 1 – Sum pi2, Ho = Observed Heterozygosity = No. of Hets/N, uHe = Unbiased expected heterozygosity = (2N/(2N – 1)) * He.

Where pi is the frequency of the i th allele for the population & Sum pi2 is the sum of the squared population allele frequencies.

Value was 0.001 before rounded off two decimal places.

Population structure by PCA analysis and STRUCTURE analysis

Fig. 2 shows the results of PCA for the four populations in terms of SSR polymorphism. The dispersal did not show a trend corresponding to the Indica-Japonica classification. The PL, NP, and BL populations were dispersed continuously and overlapped with each other. This was because these populations were selected to cover a wide range of current genetic resources. On the other hand, the dispersal of the preserved population was independent of the other populations. This indicated that the preserved population had a genetic structure different from the PL, NP, and BL populations. The preserved population seemed to divide into three subpopulations: I, II, and III (Fig. 2).

Fig. 2.

PCA among the preserved population and three representative current populations, based on the genotypes of 21 SSR loci. X-axes and Y-axes indicate the first and second coordinates, which explained 10.2% and 9.4% of the variation, respectively. Pres: Preserved population, PL: Parental landraces of modern breeding line, NP: Non-parental landraces, BL: Modern breeding line.

The LnP (D) value derived from STRUCTURE analysis increased with the assumed cluster number (K). ΔK calculated using the approach of Evanno and Earl and vonHoldt (2012) was maximal at K = 3. From these results, the most appropriate cluster number was considered to be K = 3. In the case of K = 3, populations were classified into three clusters indicated in green, blue, and red. The green and blue clusters were dominant in the preserved population, whereas the red cluster was dominant in the PL, NP, and BL populations (Fig. 3). These results confirmed that the preserved population had a structure different from that of the three representative current populations.

Fig. 3.

Representative estimate of population structure at K = 3 for the preserved population, parental line (PL), non-parental line (NP), and modern breeding line (BL).

Discussion

DNA extraction and PCR amplification efficiency with preserved seeds

In a previous study, it was confirmed that the germination ability of the preserved seeds had been lost already, but the nutritional and eating quality had been preserved quite well (Fukumoto , Kondo and Okamura 1932). Therefore, it was inferred that the DNA condition had also preserved well. As expected, amplification of the SSR markers in this study showed quite high efficiency. On the other hand, amplification of INDEL markers showed low efficiency. This was possibly because the INDEL analysis was performed several years after the SSR analysis, and the DNA quality might have deteriorated in the interval. Because of the multicopy number of genomes in the plastid, the efficiency of the chloroplast marker amplification was expected to be higher than that of other markers that had originated from the nuclear genome, but this was not the case. A negative correlation was observed between product size and amplification efficiency. Amplification efficiency is considered to be strongly related to the fragment size digested by DNase. These results suggested that analysis of preserved DNA should be done immediately after extraction, as is the case for ancient DNA.

Genetic diversity of the preserved seed population

Average seed size was larger in the preserved population than in the other two populations. The L/W ratio of 1.7 was similar to the others; however, co-variance was clearly highest in the preserved population, and even in the landrace and modern populations consisting of five or six different varieties. All of the current landrace and modern populations in this study were grown in an experimental paddy field. It was considered that their stable cultivation environment reduced their seed size variance. Genetic uniformity as a consequence of breeding may also be an important factor. If the preserved population consisted of more than one landrace, the large variance of seed size indicated that it had been grown in various environmental conditions. Among the preserved population, SSR analysis showed one heterozygote (<0.1%). In addition, INDEL and SNP analysis indicated 3.1% and 27.1% of Indica type and 1% and 18.8% of Indica-Japonica heterozygotes, respectively. From these results, it was revealed that Indica and Japonica landraces were grown mixed or sufficiently close to each other to allow outcrossing in the Mikawa area in the Meiji era. Ootsuka reported that most of the Japanese fragrant landraces shared the same functional SNP as a recessive allele involving fragrance, but they had been outcrossed with other surrounding landraces and gradually differentiated from the original populations. Upland rice in Laos also tends to show a mixture of Indica and Japonica types accompanied by heterozygotes and chloroplast-substituted landraces (Ishikawa , Muto ). Weedy rice is another derivative originated from such outcrossing (Kawasaki , Olsen , Tang and Morishima 1997, Ushiki ). These examples support our conclusion that heterozygotes resulting from natural outcrossing had occurred frequently in the cultivation environment of the preserved population.

Indica rice was introduced into Japan in the Middle Ages, when it was called “Daito-mai” and “Karabousi”, and was cultivated widely until recent times (Miyagawa 1987). Most of these landraces were regarded as red rice with long grains (c-type), with some exceptions showing a white pericarp (Arashi 1969, 1974, Hamada 1986, Itani and Ogawa 2004). The preserved seeds in this study were stored as dehulled grains, and no red rice or long grains were found among them. All of seeds seemed to be typical Japonica (a-type); however, 3.1% to 27.1% Indica genotypes were found among the preserved population by INDEL and SNP analysis. It was inferred that the Indica seeds found in the preserved population were not “Daito-mai”. After the Meiji era, “Daito-mai” was dismissed because of its red grain color and stem lodging tendency. However, Indica landraces in the preserved population were retained, because they had a typical Japonica-like shape and white-colored grains. Rice seeds do not retain their germination ability for long. At ordinary temperature and after natural drying, germination ability degrades significantly within 1 year (Oka and Tsai 1955). Indica had been regarded as an emergency stock in the Kansai area because of its higher seed longevity. Later research confirmed that the seed longevity of Indica was longer than that of Japonica (Chang 1991, Ellis ). It is possible that the Indica type seeds among the preserved population from the Meiji-era were some of those Indica landraces grown on purpose for their seed longevity.

Genetic diversity (uHe) reveled by SSR analysis showed a relatively lower value (0.40) in the preserved population relative to the PL (0.48), NP (0.44), and BL (0.50) populations (Table 3). However, this value appears high if they were grown within the confined Mikawa area. PCA revealed that there were three subpopulations in the preserved population (Fig. 2). This confirmed the results of the STRUCTURE analysis: the three clusters in green, blue, and red, and K = 3 almost corresponded to the subpopulations I, II, and III in PCA, respectively (Supplemental Fig. 2). Indica and recombinants were contained in all three subpopulations, i.e. there were no subpopulations that consisted of only Japonica or Indica. From these results, it was assumed that the preserved population was grown in a rough and diversified cultivation environment. On the other hand, the dispersal of the preserved population did not overlap with the PL, NP, and BL populations in the PCA. The preserved population also consisted of different clusters—PL, NP, and BL populations in the STRUCTURE analysis (Fig. 3). These results revealed that the preserved population had a degree of genetic diversity that did not exist among the current rice varieties.

In conclusion, the preserved population was found to consist of rice landraces with a diverse genetic background. This diversity had been generated by frequent outcrossing among genetically diverged populations in the rough cultivation environment existing in Mikawa in the Meiji era. The genetic variation present in the preserved population was not detected among landraces and breeding varieties in current used. This result clearly revealed that there genetic resources have been lost in the process of modern breeding. Analysis of not only archaeological but also preserved remains with a historical background are important for understanding the cultivation environment and genetic diversity at the time. Proactive investigations of preserved remains such as emergency rice stocks are justified, because this may provide valuable information for improving the conservation of genetic resources such as landraces and wild species for future breeding.

Author Contribution Statement

CM, YS, NK and RI designed the project. CM, KT and HT performed the molecular work and analyzed the data. HT, NK and RI analyzed and discussed the data. CM and RI wrote the manuscript. All authors have read and approved the manuscript.

Supplemental Figures

Supplemental Tables