Phenotypic and genotypic changes in rapeseed after 18 years of storage and regeneration - Wu, X.M., Wu, N.F., Qian, X.Z., Li, R.G., Huang, F.H. & Zhu, L.

Xiao-Ming Wu¹*, Ning-Feng Wu², Xiu-Zhen Qian¹, Ru-Gang Li², Feng-Hong Huang¹ and Li Zhu²

¹Institute of Oil Crops Research, Chinese Academy of Agricultural Sciences, The Second Xu Dong Road No. 2, Wuhan, 430062, China

²Biotechnology Research Center, Chinese Academy of Agricultural Sciences, Beijing, 100081, China

*Correspondence

The viability, vigour and genetic variation of rapeseed (Brassica napus L.) plants grown from seeds stored for 11 and 18 years under ultra-dry conditions were compared with growth properties of plants grown from fresher seeds produced by continuous regeneration. There was a inverse relationship between the height and productivity of plants and the duration of seed storage. The strength of this relationship was lessened somewhat by regrowing all the seed lots. The differences in growth and yield that persisted in the F1 generations were believed to result from genetic changes to populations that occurred either during storage or during regeneration. A preliminary PCR-RAPD analysis of plants arising from seeds of different ages confirmed that there were genetic differences among 18-year-old seed (no regeneration), 11-year-old seed (two regeneration cycles) and fresh seed (five regeneration cycles). While the causes for genetic change cannot be ascertained from this experiment, the proportional differences in DNA fragments and the changes in yield are of a magnitude expected for processes such as selection and genetic drift occurring in the regenerated population.

Keywords: Brassica napus, conservation strategies, germplasm, genetic integrity, genomic DNA, mutation, RAPD, seed storage, seed germination, ultra-dry.

Introduction

Ex situ seed storage is a major strategy for maintaining the genetic diversity of economically important plants. In germplasm banks, the moisture content and temperature of seeds are adjusted to prolong the shelf life. In spite of efforts to maintain seed viability, seeds deteriorate during storage and eventually die. Samples must be regenerated before viability becomes critically low. The challenge for genebank curators is to minimize genetic shift in samples over time through appropriate storage and regeneration procedures.

Chromosomal aberrations occur during seed storage (reviewed by Priestley, 1986; Roberts, 1988). The frequency of these mutations increases as the percentage germination of the seed lot decreases (Murata et al., 1981; Roberts, 1988). It is generally accepted that these gross abnormalities do not persist in samples when they are regenerated, and so they do not contribute to genetic shift (Roos and Rincker, 1982; Murata et al., 1984; Roberts, 1988). Nonetheless, major breaks in chromatids imply that more subtle point mutations also occur, and these may become fixed in the population, eventually causing a change (usually deleterious) in the genetic make-up of the sample (Roberts, 1988). Chromosomal aberrations may be more prevalent in seeds stored under dry conditions (Roberts, 1988), although the evidence for this is scarce and compounded by temperature effects.

The risk of mutations during seed storage may be circumvented by continually regenerating the seed. The genetic shift which occurs during regeneration is governed by population genetic processes such as mutation, selection and genetic drift (Hartl, 1988). Mutations usually occur in populations at a rate of 10^-5 to 10^-9 mutations per phenotype per generation (Antolin, 1998). Selection of more adapted or more fecund genotypes can lead to more rapid changes in genetic frequencies. Genetic drift results from random loss of genes and is predominantly influenced by population size. Rare individuals are often lost from small populations which tend toward homozygosity (Hartl, 1988). In a simulation study, Roos (1984b) demonstrated that shifts toward homogeneity occurred in simulated bean populations containing 64 individuals even without obvious selection forces. A more rapid attrition of genetic variation occurred in seed samples that were severely deteriorated by high temperature and high humidity storage, presumably because of the selective advantage of seeds with greater longevities (Roos, 1984a,b).

Past investigations of various factors which contribute to genetic shift in conserved germplasm were simulations. In particular, the seed aging treatments used previously were extreme, often reducing the germination to 50% within 2 months or less. Very little work, if any, has been done to test the risks of genetic shift in stored seeds compared with fresh seeds obtained by frequent regenerations under storage and regrowth conditions used by genebank curators. In this paper, we report results from an actual genebank scenario. Since the 1970s, samples of rapeseed (Brassica napus L.) germplasm have been packaged in hermetically sealed desiccators and maintained in a storeroom in Wuhan, China. The seed moisture content was held between 3 and 3.5% and the storage temperature averaged about 17°C but ranged from -5 to 37°C. Germination tests indicated that the seeds of some cultivars were still viable in 1993, after 18 years of storage. Seed samples were also regenerated every 3-4 years. In this paper we evaluate phenotypic and genetic differences among populations of rapeseed that have been stored as seed for 18 years or have been regenerated up to five times within that period.

Materials and methods

Plant materials, storage conditions and regeneration procedures

Two cultivars of rapeseed (Brassica napus), Ganyou No. 1 and Oro, were used in this study. The cultivar Ganyou No. 1 is a highly inbred line released in China in 1965. In 1975, samples of both cultivars were placed in long-term storage in the seed genebank at the Institute of Oil Crops Research, Wuhan. Since then, samples of about 100 seeds were regenerated every 3-4 years in isolated fibre-net houses and used as genebank seed stocks. In 1982, samples of the F₂ generation of the original seed stocks of both cultivars were produced and placed in long-term storage. In 1990 and 1993, samples of the F₄ generation of Oro and the F₅ generation of Ganyou No. 1 were produced and placed in medium-term storage. These samples were used as the basis of comparison for growth and genetic changes. In 1994, samples from seeds stored for 0.4, 11 and 18 years (Ganyou No. 1) and 3, 11 and 18 years (Oro) were regenerated to produce F₁ populations. These regenerated populations were actually the F₆, F₃ and F₁, generations of the 1975 Ganyou No. 1 population and F₅, F₃ and F₁ generations of the 1975 Oro population.

Seeds that were put in medium-term storage (i.e. 1990,1993 and 1994 harvests) were dried to about 7% under ambient conditions. Seeds placed in long-term storage (i.e. 1975 and 1982 harvest and intervening generations) were dried to moisture contents of about 3-3.5% by placing them in air-tight desiccators over the desiccant CaCl₂. These seeds remained sealed in the desiccators throughout the storage period. Moisture contents were evaluated according to International Seed Testing Association rules (ISTA, 1985). All seeds were stored at ambient temperature. For storage longer than 1 year, temperatures averaged 17°C and ranged from -5 to 37°C. Seeds from the 1993 and 1994 crops were stored from May to September, when temperatures averaged 25°C and ranged from 12 to 37°C.

Germination assays and field performance

Germination tests were conducted according to the International Rules for Seed Testing (ISTA, 1985). Seeds were germinated on filter paper at 25°C for 5 days. Samples of 100 seeds were used in each assay. In addition to germination percentage, the vigour of seeds was quantified by measuring the fresh mass of the seedlings.

Timing of developmental changes, growth rate and yield of plants grown from seeds stored for 5 months (Ganyou No. 1) or 3 years (Oro), 11 years and 18 years were evaluated in 1993/94. A sample of 200 seeds from each storage treatment was sown in rows 2 m long with 0.33 m between rows. At the five-leaf stage, plants were thinned to 15-20 per row. A randomized block design was used, with each treatment replicated three times. Once the plants began to flower, a net-house was temporarily constructed over them to isolate plants from different treatments during pollination. Seeds produced from these plants were sown in the field in September 1994, using the same design in order to evaluate the field performance of the F₁ generation of each storage treatment.

Genetic analysis

Identification of genetic changes during storage or regeneration can be inferred by comparing DNA fragments of plants grown from seeds stored for different periods. Genetic analyses were performed on Ganyou No. 1 plants only. Total DNA was extracted from the leaf tissue of three plants grown from seed from each storage treatment according to standard protocols (Hillis et al., 1990; Colosi and Schaal, 1993). PCR was carried out on 25 ng DNA in a total volume of 25 ml, containing 10 mM Tris-HCl buffer (pH 8.0), 2.5 mM MgCl₂, 0.2 mM dNTP, 0.64 mm of primer (purchased from the University of British Columbia, Canada) and 0.625 units of Taq polymerase (Pharmacia, LKB). Thirteen primers were chosen arbitrarily (Table 1) and these generated a total of 113 amplified DNA fragments. Amplification of fragments was carried out using a PTC-100 machine (M.J. Research Inc., USA). Temperature regimes of 1 min at 94°C, 5 min at 38°C and 3 min at 72°C were cycled 45 times. Amplification products were separated on 1.5% agarose gels.

Table 1. Arbitrary primers used in PCR-RAPD study to analyse genetic differences among lots of Ganyou No. 1 seed and the number of randomly amplified DNA fragments generated in the experiment

			Distribution of fragments in stored see
			0.4 years		11 years		18 years
Primers	Sequences (5' ® 3')	Number of fragments	P*	A	P	A	P	A
UBC218	CTCAGCCCAG	10	0	0	0	0	0	0
UBC220	CCCGTCAATA	9	0	0	0	0	0	0
UBC213	CAGCGAACTA	12	0	2	0	1	1	0
UBC214	CATGTGCTTG	12	1	1	1	1	2	1
UBC241	GCCCGACGCG	4	0	0	0	0	0	0
UBC243	GGGTGAACCG	8	0	0	0	0	0	0
UBC235	CTGAGGCAAA	9	2	0	2	0	2	0
UBC237	CGACCAGAGC	11	1	0	1	0	1	1
UBC300	GGCTAGGGCG	6	0	0	0	0	0	0
UBC281	GAGAGTGGAA	6	0	0	0	0	0	1
UBC282	GGGAAAGCAG	4	0	0	0	0	0	0
UBC280	CTGGGAGTGG	15	3	0	3	1	3	0
UBC101	GCGGCTGGAG	7	1	0	1	0	1	0
Total		113	8	3	8	3	9	3

* Number of fragments from the indicated primer that were polymorphic (P) or completely absent (A) from the sampled population.

Seed germination and vigour

Germination percentages and fresh mass of rapeseed seedlings containing 3 to 3.5% water and stored at ambient temperature were progressively lower as storage duration increased (Table 2). Germination percentages of 42 and 4% were recorded for Ganyou No. 1 and Oro seeds, respectively, stored for 18 years (Table 2). Regression analysis of storage time versus seedling mass indicated a loss of about 0.23 g per 100 seed per year under these storage conditions (r² = 0.96 and 0.86 for Ganyou No. 1 and Oro, respectively). High germination percentages were completely restored when the samples were regenerated in 1994, but seedling growth was only partially restored by regeneration (Table 2).

Table 2. Germination and seedling mass of seeds and F₁ generation of seeds from two rapeseed cultivars stored at ambient temperatures for 11 and 18 years

	Storage time (yrs)
	0.4	11	18
Ganyou No. 1
Harvest year	1993	1982	1975
Seed mc (% fwb)	7.5	3.4	3.2
Germination (%)	100	100	42
Seedling mass (g/100 seeds)	7.0	3.5	2.6
F1 of stored Ganyou No. 1
Harvest year	1994	1994	1994
Seed mc (% fwb)	7.5	7.5	7.5
Germination (%)	100	100	100
Seedling mass (g/100 seeds)	7.2	6.9	6.0
	Storage time (yrs)
	3	11	18
Oro
Harvest year	1990	1982	1975
Seed mc (% fwb)	7.7	3.5	3.1
Germination (%)	95	70	4
Seedling mass (g/100 seeds)	3.5	2.2	<0.5
F1 of stored Oro seeds
Harvest year	1994	1994	1994
Seed mc (% fwb)	7.6	7.6	7.6
Germination (%)	100	100	100
Seedling mass (g/100 seeds)	6.9	6.2	4.8

Field performance

The growth characteristics of field-grown plants were examined in the fall/spring of 1993/94 to determine whether plants arising from seeds stored for different periods showed differences in timing of several developmental events (Tables 3 and 4) and morphological traits or yield (Tables 5 and 6). As expected from laboratory germination assays (Table 2), field emergence and time to first true leaf stage took longer in samples from stored seeds - 17, 22 and 25 days for Ganyou No. 1 stored for 0.4, 11 and 18 years, respectively (Table 3), and 21, 21 and 23 days for Oro stored for 3, 11 and 18 years, respectively (Table 4). All populations gave comparable rates of winter survival (data not shown), but the timing of bud emergence and stem elongation in February and March, respectively, was delayed by 1 week or more in plants grown from stored seed (Tables 3 and 4). Flower initiation was slightly delayed in plants from stored seed, and the duration of the flowering and seed-fill period ranged from about 42 to 37 days depending on storage time (Tables 3 and 4). These results indicate that plants from stored seed required more time for field establishment, had an abbreviated period of vegetative growth in the spring, and had a slightly shorter time for reproductive growth (Tables 3 and 4).

Regeneration of samples tended to eliminate the differences in developmental timing observed among seeds of different ages. No consistent differences in time for seedling emergence (6±0 days), vegetative growth in spring (45±2 days) and seed filling (56±3 days) were observed in F₁ generations produced from stored seeds (Tables 3 and 4).

Morphological characteristics and yield were also measured in field-grown plants (Tables 5 and 6). Perhaps owing to the abbreviated growing period (Tables 3 and 4), plants from seeds stored for longer periods were smaller (Tables 5 and 6). Some cotyledons of seeds stored for 18 years were broken and had white or yellow spots. Both fall and spring leaves on plants produced from 18-year seeds were often discoloured and wrinkled, and these abnormalities persisted in the F₁ generation. The size of flowers was smaller in plants from 18-year-old seed, but silique morphology appeared similar. The number of siliques and the size of seeds produced varied according to the age of the seed (i.e. the duration of storage), resulting in almost a two- and three-fold difference in yield between plants from fresh seeds and those from seeds stored for 11 and 18 years, respectively (Tables 5 and 6).

Table 3. Timing of developmental events of Ganyou No. 1 rapeseed grown from stored seeds and the F₁ generation of stored seeds. Seeds were stored at ambient temperatures and about 3.2% water (11 and 18 years) and 7% water (0.4 years)

	Storage time (years)
	0.4		11		18
Stage	Date	DAP*	Date	DAP	Date	DAP
Stored seed
Sowing	25 Sep 93	0	25 Sep 93	0	25 Sep 93	0
Seedling	01 Oct 93	6	03 Oct 93	8	04 Oct 93	9
Single leaf	12 Oct 93	17	17 Oct 93	22	20 Oct 93	25
Bud emergence	12 Feb 94	137	21 Feb 94	146	28 Feb 94	153
Stem elongation	20 Feb 94	145	04 Mar 94	159	12 Mar 94	167
Flower initiation	21 Mar 94	176	24 Mar 94	179	26 Mar 94	181
Flowering ends	12 Apr 94	197	14 Apr 94	199	16 Apr 94	201
Seed maturation	02 May 94	217	03 May 94	218	03 May 94	218
Total days
Spring vegetative growth		39		33		28
Flower and seed fill		41		39		37
F1 of stored seed
Sowing	24 Sep 94	0	24 Sep 94	0	24 Sep 94	0
Seedling	30 Sep 94	6	30 Sep 94	6	30 Sep 94	6
Single leaf	-		-		-
Bud emergence	27 Jan 95	123	28 Jan 95	124	23 Jan 95	119
Stem elongation	09 Feb 95	135	11 Feb 95	137	12 Feb 95	138
Flower initiation	10 Mar 95	166	14 Mar 95	170	08 Mar 95	164
Flowering ends	01 Apr 95	187	31 Mar 95	186	28 Mar 95	184
Seed maturation	05 May 95	221	06 May 94	222	06 May 95	222
Total days
Spring vegetative growth		43		46		45
Flower and seed fill		55		52		58

*DAP, days after planting.

	Storage time (years)
	3		11		18
Stage	Date	DAP*	Date	DAP	Date	DAP
Stored seed
Sowing	25 Sep 93	0	25 Sep 93	0	25 Sep 93	0
Seedling	05 Oct 93	10	03 Oct 93	8	03 Oct 93	8
Single leaf	16 Oct 93	21	16 Oct 93	21	18 Oct 93	23
Bud emergence	09 Feb 94	134	12 Feb 94	137	18 Feb 94	143
Stem elongation	20 Feb 94	145	28 Feb 94	153	12 Mar 94	167
Flower initiation	16 Mar 94	171	24 Mar 94	179	28 Mar 94	183
Flowering ends	13 Apr 94	198	15 Apr 94	200	17 Apr 94	202
Seed maturation	03 May 94	218	03 May 94	218	05 May 94	220
Total days
Spring vegetative growth		37		42		40
Flower and seed fill		47		39		37
F1 of stored seed
Sowing	24 Sep 94	0	24 Sep 94	0	24 Sep 94	0
Seedling	30 Sep 94	6	30 Sep 94	6	30 Sep 94	6
Single leaf	-		-		-
Bud emergence	18 Jan 95	114	25 Jan 95	121	15 Jan 95	111
Stem elongation	10 Feb 95	136	12 Feb 95	138	12 Feb 95	138
Flower initiation	06 Mar 95	162	11 Mar 95	167	27 Feb 95	153
Flowering ends	31 Mar 95	186	01 Apr 95	187	02 Apr 95	188
Seed maturation	06 May 95	222	06 May 94	222	07 May 95	223
Total days
Spring vegetative growth		48		46		42
Flower and seed fill		60		55		70

*DAP, days after planting.

Many of the differences in morphological and yield characteristics of plants grown from seeds of different ages persisted even when the seeds were regenerated. The F₁ populations from fresh seeds were slightly taller and more branched than the F₁ populations derived from seeds stored for 11 and 18 years (Tables 5 and 6), in spite of the similar developmental patterns (Tables 3 and 4). Although seed mass was similar among F₁ populations, the number of siliques per plant and the number of seeds per silique was progressively less in F₁ populations derived from older seed, and this resulted in about a 1.5- and 2.9-fold difference in yield between F₁ plants from fresh seeds and those from seeds stored for 11 and 18 years, respectively (Tables 5 and 6). Thus, regenerating seeds restored seed vigour and plant developmental patterns, but growth and yield characteristics of plants produced from seeds stored for long periods remained relatively low in the F₁ generation.

Table 5. Morphological traits and yield factors of Ganyou No. 1 rapeseed grown from stored seeds and the F₁ generation of stored seeds. Seeds were stored at ambient temperatures and about 3.2% water (11 and 18 years) and 7% water (0.4 years). Values in parentheses represent standard deviation of the mean of three replicates

Trait	Storage time (years)
	0.4	11	18
Stored seed
Plant height (cm)	156.7 (4.1)	111.3 (5.5)	81.7 (2.9)
Height of 1st branch (cm)	65.7 (15.8)	23.3 (8.5)	24.0 (2.0)
No. of primary branches	5.7 (1.5)	5.0 (2.0)	3.3 (0.6)
Branches /cm	0.036	0.045	0.040
Leaf colour	Green	Green	Dark green/purple
Leaf thickness	Normal	Normal	Fleshy
Leaf shape	Normal	Normal	Wrinkled
Length of inflorescence	54.7 (6.4)	56.0 (5.2)	37.3 (10.3)
No. of siliques	118.0 (52.0)	78.3 (47.1)	65.7 (6.8)
No. of siliques from main inflorescence	47.7 (12.7)	20.7 (3.8)	23.7 (7.5)
Silique density*	0.87 (0.12)	0.40 (0.10)	0.63 (0.15)
Silique length (mm)	48.0 (6.1)	41.7(1.5)	51.7 (4.7)
Silique width (mm)	4.8 (0.6)	4.5 (0.5)	3.7 (0.6)
No. of seeds/silique	15.3 (4.5)	13.4 (0.4)	17.1 (1.6)
1000 seed weight (g)	3.91 (0.15)	4.20 (0.13)	2.44 (0.05)
Cotyledon width (mm)	6.5 (0.6)	5.0 (0.6)	3.4 (0.7)
Cotyledon (length) (mm)	11.4 (1.1)	7.6 (1.2)	5.1 (1.2)
Yield per plant (g)	7.1	4.4	2.7
F1 of stored seed
Plant height (cm)	162.8 (22.2)	153.4 (19.1)	116.4 (19.0)
Height of 1st branch (cm)	36.8 (9.9)	50.8 (10.0)	31.8 (12.8)
No. of primary branches	10.0 (1.9)	7.8 (1.6)	7.0 (1.9)
Branches /cm	0.061	0.051	0.060
Leaf colour	Green	Green	Dark green /purple
Leaf thickness	Normal	Normal	Fleshy
Leaf shape	Normal	Normal	Wrinkled
Length of inflorescence	64.8 (9.0)	58.6 (16.1)	42.8 (19.9)
No. of siliques	483.8 (161.8)	381.2 (213.1)	254.6 145.5)
No. of siliques from main inflorescence	56.4 (10.7)	69.2 (11.7)	35.6 (116.5
Silique density	1.4 (0.2)	2.1 (0.3)	2.2 (2.4)
Silique length (mm)	49.6 (7.7)	46.4 (4.6)	38.4 (2.1)
Silique width (mm)	4.2 (0.3)	4.0 (0.5)	3.9 (0.1)
No. of seeds/silique	18.4 (5.1)	14.3 (2.5)	11.2 (2.9)
1000 seed weight (g)	3.10 (0.40)	2.70 (0.20)	3.20 (0.20)
Cotyledon width (mm)	6.7 (0.5)	6.4 (0.6)	4.1 (0.7)
Cotyledon (length) (mm)	11.0 (0.8)	10.9 (0.7)	7.7 (0.8)
Yield per plant (g)	27.6	14.8	9.2

*No. of siliques on main inflorescence/length of main inflorescence.

To determine whether the differences in growth and yield among seeds stored for different periods were of genetic origin, an experiment was conducted to compare DNA fragments from populations arising from 18-year-old seed, 11-year-old seed and 5-month-old (fresh) seed of Ganyou No. 1. Since only three plants from each population were sampled, these results can only be considered preliminary. Nonetheless, the experiments introduce a procedure to evaluate genetic changes in stored and regenerated germplasm.

Thirteen primers were used for PCR and a total of 113 DNA fragments were produced (Table 1). Of the 113 fragments, three were absent in plants produced from the 18-year-old seed and three were absent in the plants produced from fresh seed (Tables 1 and 7). Despite the low degree of genetic variation among plants of Ganyou No. 1, some variation existed as indicated by eight or more polymorphic fragments within each population (i.e. indicated by a fragment that was absent in one or two plants) (Tables 1 and 7). As is expected from populations that are closely related, most of the polymorphisms from each population were detected from the same fragments: of the eight fragments giving polymorphic expression in the fresh seed population, seven and five were identical in the 11- and 18-year-old seed populations, respectively (Table 7). Thus in the populations derived from 11- and 18-year-old seed, 0.9 and 3.6% of the fragments studied displayed polymorphisms that were different from fresh seed. The combination of different polymorphisms and different fragments absent from one population and not another give a total difference in the amplification of DNA fragments of 2.7 and 9% in populations from 11- and 18-year-old seed, respectively, compared with the population derived from fresh seed (Table 7).

Table 6. Morphological traits and yield factors of Oro rapeseed grown from stored seeds and the F₁ generation of stored seeds. Seeds were stored at ambient temperatures and about 3.2% water (11 and 18 years) and 7% water (3 years). Values in parentheses represent standard deviation of the mean of three replicates

	Storage time (years)
Trait	3	11	18
Stored seed
Plant height (cm)	146.5 (2.1)	140.8 (18.0)	103.3 (7.6)
Height of 1st branch (cm)	33.5 (2.1)	40.7 (20.6)	41.7 (11.9)
No. of primary branches	7.0 (0.0)	6.7 (1.5)	3.3 (0.6)
Branches /cm	0.048	0.048	0.032
Leaf colour	Green	Green	Dark green/purple
Leaf thickness	Normal	Normal	Fleshy
Leaf shape	Normal	Normal	Wrinkled
Length of main inflorescence	78.0 (7.1)	58.0 (11.3)	59.0 (7.9)
No. of siliques	285.5 (44.5)	224.7 (74.7)	110.0 (59.8)
No. of siliques from main inflorescence	55.0 (4.2)	61.3 (17.2)	31.3 (8.0)
Silique density	0.80 (0.0)	1.03 (0.29)	0.57 (0.21)
Silique length (mm)	47.8 (0.7)	42.3 (5.9)	44.0 (0.0)
Silique width (mm)	4.5 (0.0)	4.2 (0.3)	4.0 (0.0)
No. of seeds /silique	16.8 (0.7)	10.6 (4.4)	15.7 (2.9)
1000 seed weight (g)	4.15 (0.09)	3.94 (0.01)	3.61 (0.07)
Cotyledon width (mm)	7.1 (1.6)	8.1 (1.3)	3.7 (0.5)
Cotyledon (length) (mm)	12.8 (1.4)	12.6 (1.7)	5.2 (0.8)
Yield per plant (g)	19.9	9.4	6.2
F1 of stored seed
Plant height (cm)	140.0 (16.8)	132.2 (8.1)	93.2 (12.5)
Height of 1st branch (cm)	34.2 (16.8)	33.6 (10.5)	8.0 (2.2)
No. of primary branches	9.2 (4.7)	7.6 (1.5)	6.0 (2.9)
Branches /cm	0.066	0.057	0.064
Leaf colour	Green	Green	Dark green /purple
Leaf thickness	Normal	Normal	Fleshy
Leaf shape	Normal	Normal	Wrinkled
Length of inflorescence	58.4 (7.6)	63.8 (4.6)	64.8 (24.1)
No. of siliques	410.4 (207.1)	320.6 (52.2)	175.6 (91.7)
No. of siliques from main inflorescence	67.2 (11.2)	57.0 (9.7)	39.8 (22.8)
Silique density	1.7 (0.5)	1.2 (0.2)	0.9 (0.1)
Silique length (mm)	44.6 (3.7)	43.6 (2.1)	37.4 (3.0)
Silique width (mm)	3.9 (0.2)	3.9 (0.9)	3.5 (0.6)
No. of seeds /silique	14.2 (2.1)	12.4 (6.6)	12.2 (3.0)
1000 seed weight (g)	2.89 (0.38)	3.2 (0.3)	2.91 (0.09)
Cotyledon width (mm)	7.3 (0.5)	6.5 (1.4)	4.2 (0.6)
Cotyledon (length) (mm)	11.8 (0.8)	11.0 (0.9)	7.5 (1.2)
Yield per plant (g)	16.8	12.7	6.2

Table 7. Number and distribution of randomly amplified DNA fragments from Ganyou No. 1 rapeseed grown from seeds stored at ambient temperatures and 7% water for 0.4 years (fresh seeds) and about 3.2% water for 11 and 18 years

		Storage time (years)
		0.4	11	18
Number of fragments
Present in plants from fresh seed
	total	110	109	107
	polymorphic fragments	8	8	7
	polymorphic fragments same in fresh and stored	8	7	5
	fragments absent in stored	0	1	3
Absent in plants from fresh seed but present in plants from stored seeds
	total	-		3
	polymorphic fragments	-	0	2
All fragments
	total amplified	110	110	110
Frequency of differential fragment amplifications
	polymorphisms (%)	7.2	7.2	8.2
	polymorphic fragments
	same as fresh (%)	7.2	6.4	4.6
	differ from fresh (%)	0	0.9	3.6
	fragments absent in stored (%)	0	0.9	2.7
	fragments absent in fresh but
	present in stored (%)	-	0.9	2.7
	total difference from fresh (%)	0	2.7	9.0

Discussion

The objective of seed genebanks is to preserve the biological diversity of plants by maintaining the viability as well as the genetic integrity of seed accessions. The efficacy of these germplasm reserves must be evaluated according to both the time that seeds in storage remain alive, and the genetic changes that occur during that time in the stored accession compared with the same population conserved by continuous regeneration. This study considers the storage life of two cultivars of rapeseed stored at ambient temperatures, and makes inferences about genetic changes that occur during the 18 years of the experiment.

Seeds stored under ambient conditions in temperate climates generally die within 5-10 years (Priestley, 1986). We report P₅₀ values in excess of 11 years for both Ganyou No. 1 and Oro rapeseed (Table 2) despite the relatively warm storage conditions (average yearly temperature 17°C; average summer temperature 28°C). The extended storage life may be attributed to the low moisture content of storage (Ellis et al., 1989). Seeds placed in long-term storage were sealed in the presence of the desiccant CaCl₂ which gives a water content of about 3.2 %. Although data are not provided in this paper, experience at the Institute of Oil Crops Research has shown that seeds stored at ambient relative humidities (90% in summer, 60-70% in winter, corresponding to about 7% water in rapeseed) survive for less than 10 years, and often less than 5 years. Our results indicate that even if refrigeration is not available to germplasm repositories, the practice of storing seeds with activated desiccant will give greater shelf life to seeds compared with open storage in a humid environment (Cheng et al., 1991; IBPGR, 1992, 1993).

Plants produced from 18- and 11-year-old seed had slower growth and lower yields than plants from 3-year-old seed (only Oro studied) or from 5-month-old seed (only Ganyou No. 1 studied) (Tables 2-6). The reduced vigour of seeds stored for extended periods may contribute to the poorer stand establishment. When this factor was removed by regenerating seeds, seed and seedling growth of the three populations were indistinguishable (Tables 2-4), but the F₁ generation of plants grown from stored seeds remained smaller and yielded less than the F₁ from fresher seeds (Tables 5 and 6). This suggests that some of the differences in growth may be attributed to genetic changes that occurred during storage or regeneration.

Genetic differences between plants grown from fresh, 11- and 18-year-old seed were evaluated by comparing the distribution of randomly amplified DNA fragments. A preliminary study using Ganyou No. 1 revealed 2.7 and 9.0% difference in the presence of DNA fragments in 11- and 18-year-old seed, respectively, compared with fresh seed (Table 7).

The genetic differences suggested between plants from fresh seed and stored seeds may arise from changes in the stored seed, or in the continuously regenerated population, or both. The source of the genetic shift must be understood if we are to develop the most effective conservation strategy. A PCR-RAPD analysis of the parent population in 1975, and of subsequent generations, would have made it possible to discern the source of genetic change. However, the recent development of this technology and its introduction to China only in the late 1980s make the ideal experiment impossible. The confounding effects of storage and regeneration in the experiment presented here make it difficult to distinguish between the different causes of genetic change.

Changes in the genetic composition of the stored seed may arise from mutations that were not repaired (Roberts, 1988) or from inadvertent selection when the viability declined and individuals with poor seed longevity were lost from the gene pool (Roos, 1984a). According to the DNA analysis shown in Table 7, a mutation rate of greater than 10^-3 mutations per nucleotide per year can be hypothesized (i.e. a proportional difference of 0.027 over 11 years and 0.09 over 18 years). These mutations are considered point mutations, and would not contribute to the chromosomal aberrations observed previously (Murata et al., 1984; Roberts, 1988). It seems unlikely that minor mutations affecting about 3 and 10% of the population of 11- and 18-year-old seed, respectively (Table 7), can give rise to the large differences in yield obtained from growing seeds of different ages (Table 5). The yield differences also cast some doubt on the hypothesis that selection of longer-lived individuals caused genetic shift (Roos, 1994a), since this implies an unlikely inverse relationship between seed longevity and yield.

Alternatively, genetic differences in seeds stored for different periods may arise from the continual regeneration required to maintain the fresh seed population. The fresh seed lot of Ganyou No. 1 represents the F₃ and F₅ generations of the seed lots stored for 11 and 18 years, respectively. Changes in the genetic composition of the fresh seed population may result from several processes, including background-level mutations, introduction of genes from other populations (contamination by other pollen sources), genetic drift or inadvertent selection of more fecund genotypes (Hartl, 1988).

It is unlikely that base-level mutations caused genetic shift within the regenerated seed lots. Known rates of mutation in nuclear DNA (10^-5 mutations per nucleotide per generation; Antolin, 1998) are low in comparison with rates of about 10^-2 mutations per nucleotide per generation that can be calculated from the data in Table 7 (0.027 per three generations, or 0.09 per five generations). Also, one expects mutations to have deleterious effects rather than the beneficial effects of increased growth and yield observed in Table 5.

Genetic drift causes erosion of genetic variation in a population. It is completely random and it is unavoidable in small populations (Hartl, 1988). The rate of genetic drift is expressed by the equation

F_t = 1-[1-(1/2N)]^t

where F_t is the proportion of change in generation t, N is the population size and t is the number of generations (Hartl, 1988). In an outbred population, F₀=0 and F_t approaches 1 as the population becomes more inbred. Clearly genetic drift occurs very rapidly in small populations (i.e. about 5% loss in genetic variation per generation in an outbred population of 10 individuals) but substantial change also occurs in larger populations after several generations (i.e. in a population of 100 individuals, a 5% loss of variation occurs in 10-11 generations). In the experiment presented here, where the fresh seed lot was the F₃ and F₅ generation produced from 100 individuals of the 11-and 18-year-old seed, respectively, a change in gene frequency of 0.015 and 0.025 is predicted. The proportional difference between fresh and 11- or 18-year-old seed measured in the RAPD analysis (0.027 or 0.09, respectively; Table 7) is of the magnitude expected from genetic drift in populations of 30-60 individuals. Thus, as pointed out by Roos (1984b), drift in continually regenerated seeds may have an important influence on the genetic composition.

The progressive increase in yield with increased number of regeneration cycles (Tables 5 and 6) also suggests that seeds from the more productive plants may have been inadvertently selected for the next generation. Care was taken to minimize selective forces and to prevent introduction of genetic variation, but the influence of these factors during regeneration cannot be overlooked.

Summary

Rapeseed may be stored without loss of viability at ambient temperatures in Wuhan, China for over a decade if the water content of the seeds is maintained at <3.5%. Seeds that were stored for 11 years had high germination percentages, but the seedling growth was slow compared with fresh seed. Seedling vigour of slightly and severely deteriorated seeds was restored by regenerating seeds, but growth patterns and yield were different in F₁ populations from stored seeds and fresh seeds produced by continuously regenerating the sample. Preliminary investigations of DNA polymorphisms detected some differences in the genetic make-up of plants from stored seeds and continuously regenerated seeds. It is critical to know the source of the genetic shift; however, the causes of genetic change cannot be determined from this study. If the observed changes were a result of deterioration during seed storage, then researchers must establish better methods to maintain viability or prevent mutations. If the genetic changes were a result of genetic drift and selection during regeneration of germplasm, then this study points out the need for seed banks as an efficient way to preserve the genetic composition of a population.

Acknowledgements

The authors greatly appreciate Drs Michael Antolin (Colorado State University) and Christina Walters (USDA-ARS) for their constructive suggestions and critical editing of the manuscript. This research was financially supported by the International Plant Genetic Resources Institute and the Chinese National Agricultural Research Key Project. The authors also thank Professor Zhou Ming-De, Coordinator for IPGRI Office for East Asia, for her support.

References

Antolin, M.F. (1998) Assessments of strains, varieties and ecotypes. pp 62-70 in Gass, T., Podyma, W., Puchalski, J., Eberhart, S.A. (Eds) Proceedings of the International Conference on Crop Germplasm Conservation Problems with Special Emphasis on Rye, July 1996, Warsaw, Poland. Rome, International Plant Genetic Resources Institute.

Cheng, H.Y., Zheng, G.H. and Tao, K.L. (1991) Effects of ultradrying on aging, cell ultrastructure and vigour of Chinese cabbage. FAO/IBPGR Plant Genetic Resources Newsletter 83/84, 9-14.

Colosi, J.C. and Schaal, B.A. (1993) Tissue grinding with ball bearing and vortex mixer for DNA extraction. Nucleic Acids Research 21,1051-1052.

Ellis, R.H., Hong, T.D. and Roberts, E.H. (1989) A comparison of the low-moisture-content limit to the logarithmic relation between seed moisture and longevity in twelve species. Annals of Botany 63, 601-611.

Hartl, D.L. (1988) A Primer of Population Genetics. (2nd edition). Sunderland, Massachusetts, Sinauer Associates.

Hillis, D.M., Larson, A., Davis, S.K. and Zimmer, E.A. (1990) Nucleic Acids III. Sequencing, pp 318-370 in Hillis, D.M., Mortiz, C. (Eds) Molecular Systematics. Sunderland, Massachusetts, Sinauer Associates.

ISTA (1985) International rules for seed testing. Seed Science and Technology 13, 299-355.

IBPGR (1992) Ultradry seed storage, p. 40 in Annual Report 1991. Rome, International Board for Plant Genetic Resources.

IBPGR (1993) Ultradry seed storage. p. 30 in Annual Report 1992. Rome, International Board for Plant Genetic Resources.

Murata, M., Roos, E.E. and Tsuchiya, T. (1981) Chromosome damage induced by artificial seed aging in barley. I. Germinability and frequency of aberrant anaphases at first mitosis. Canadian Journal of Genetics and Cytology 23, 267-280.

Murata, M., Tsuchiya, T. and Roos, E.E. (1984) Chromosome damage induced by artificial seed aging in barley 3. Behavior of chromosomal aberrations during plant growth. Theoretical and Applied Genetics 67, 161-170.

Priestley, D.A. (1986) Seed Aging. Ithaca, New York, Cornell University Press.

Roberts, E.H. (1988) Seed aging: the genome and its expression, pp 466-198 in Nooden, L.D., Leopold, A.C. (Eds) Senescence and Aging in Plants. New York, London, Academic Press.

Roos, E.E. and Rincker, C.M. (1982) Genetic stability in 'Pennlate' orchardgrass seed following artificial aging. Crop Science 22, 611-613.

Roos, E.E. (1984a) Genetic shifts in mixed bean populations. I. Storage effects. Crop Science 24, 240-244.

Roos, E.E. (1984b) Genetic shifts in mixed bean populations. II. Effects of regeneration. Crop Science 24, 711-715.

© CAB INTERNATIONAL, 1998