(Reprints available from the author)
Monographs ETSIA, Univ. Politécnica de Madrid 163, 1-10 (2002)

Long term seed preservation:
the risk of selecting inadequate containers is very high

César Gómez-Campo

Dept. Biología Vegetal, Escuela T. S. Ing. Agrónomos. Universidad Politécnica de Madrid. 28040 - Madrid. Spain.


Forty different types of containers, obtained from collaborating seed banks or commercial sources, were tested for their ability to exclude water vapour, using silica gel with a cobalt indicator. Only sealed brass cans, ‘Kilner’ jars with rubber seals, laboratory bottles normally used for liquid chemicals, or flame-sealed glass ampoules prevented moisture intake sufficiently to be safely considered for use in long-term seed preservation. Other types of plastic, glass, metal or foil containers (thirty-six or 90%) all allowed moisture to enter within 2 to 3 years or less. Many of the containers in this second group are often routinely used in seed banks. Therefore, a previous evaluation of storage containers for moisture exclusion either with the simple test used here or by detecting deviations from the original seed moisture is strongly recommended. Whenever possible, subsequent monitoring with a chemical indicator during storage is highly desirable to avoid irreparable losses of valuable genetic material and to reduce the frequency of time-, labour- and seed-consuming germination tests and/or seed regeneration.

Keywords: seeds, seed banks, seed containers, genebanks, germplasm.


Usando gel de sílice con indicador de cobalto se evaluó la hermeticidad de cuarenta tipos distintos de envases, unos comprados y otros cedidos por varios bancos de semillas. Sólo las latas de latón con tapa sellada, los tarros de appertización con junta de goma, los frascos de laboratorio para productos líquidos y las ampollas de vidrio cerradas a la llama, se mostraron suficientemente herméticos o impermeables como para contar con ellos en la conservación de semillas a largo plazo. Todos los demás envases (36, o sea el 90%) fueran de plástico, vidrio, metal o lámina de aluminio/plástico permitieron la entrada de humedad en su interior en menos de 2-3 años. Muchos de los pertenecientes a este segundo grupo se usan con bastante frecuencia en los bancos de semillas. Por ello, es altamente recomendable hacer una evaluación de cualquier envase utilizado o por utilizar, ya sea por el simple procedimiento aquí explicado, o bien por detección de cambios en la humedad de las semillas. En lo posible, resulta muy útil el seguimiento continuo de las muestras almacenadas con un indicador químico, pues así pueden evitarse pérdidas irreparables de valioso material genético y reducirse mucho la frecuencia de las pruebas de germinación y/o regeneraciones que tanto tiempo, trabajo y material consumen.

Palabras clave: semillas, envases, bancos de semillas, bancos de germoplasma.


Harrington (1972) estimated that a 5ºC decrease in storage temperature or a 1% decrease in moisture content can independently increase the life span of orthodox seeds by approx. two-fold. Today, most specialists generally agree that preserving orthodox seeds for one century or two should not be a difficult task. However, the need to regenerate seed bank accessions is frequently mentioned in the literature (Breese, 1989; Anonymous, 1994; Sackville-Hamilton and Chorlton, 1997), this being meaningful for samples with poor initial seed quality or in cases where good quality samples become exhausted through exchange. Otherwise, the need for frequent regeneration of orthodox seeds is clearly a consequence of defficient seed storage conditions.

Since regeneration has deleterious effects on the genetic variability of stored seeds (Stoyanova, 1992; Anonymous, 1994; Diaz et al., 1997) and is labour and seed consuming, it should be reduced to a minimum by placing much more emphasis on the efficiency of seed preservation systems (Anonymous, 1994).

The aim of the present study was to evaluate one possible weakness of current seed preservation practices – the reliability of the storage containers to exclude moisture.

Materials and methods

Containers used for testing were either received from seed banks or obtained from commerce. Each type of container was given a number (1-40) and its main characteristics, such as material, colour, volume, dimensions, type of cap, aperture, supplier, etc. were recorded. These data are only partially listed in Table 1. Photographs of most of the containers evaluated are shown in figures 1 and 2.

To create water vapour-saturated atmospheres, humidity chambers were prepared by adding demineralized water to plastic boxes (60 x 40 x 40 cm) to a depth of 5-6 cm and closing the box with a flat plastic lid. The seed storage containers to be tested were placed on plastic trays and inserted into these humidity chambers at room temperature (20-30ºC).

Before its placement in humidity chambers, each individual container was loaded with deep blue dehydrated silica gel (Pancreac, 0.5-1.2 mm mesh), at an approximate proportion of 1-2 g of silica gel per 100 ml of the container volume. In a previous experiment, this amount of silica gel absorbed any pre-existing water vapour in the container without changing colour and was a sufficiently small quantity to avoid significant delays in the evaluation of colour changes derived from the intake of external water vapour. These changes in silica gel colour from deep blue to pink were used as an indication that the container failed to exclude moisture.

Table 1. Main characteristics and moisture-tightness performance of forty containers in view to their use for long-term seed preservation. Origin: c = from commerce; sb = from seed bank. Performance: was measured as % of units in which the silica gel kept its blue colour after one, two or three years.
Plastic containers
Number Vol. ml. Origin Description Performance after
1 yr. 2 yr. 3 yr.
1 315 c Polyethylene (PE) bottles; screw cap of the same material with festooned border 48 40 4
2 315 c PE bottles with a star-shaped screw cap of the same material 0 0 0
3 1100 sb PE bottles with star-shaped screw cap; like number 2 but larger in size 100 8 0
4 520 c PE bottles with double top (screw cap/pressure stopper) of the same material, much smaller diameter 100 80 44
5 290 c PE bottles similar to 4; with narrow neck and screw cap of the same material 100 68 32
6 270 c PE bottles; with a flip green top of the same material 10 0 0
7 615 c Cylindrical PE jars with double closing: a black PC screw cap with equal diameter +PE pressure stopper 12 0 0
8 800 sb Cylindrical PE jars with double closing; very similar to number 7 but larger in size 0 0 0
9 555 c Cylindrical transparent poly-ethylene-tereftalate jars, similar in shape to 7 and 8; with a double closing 0 0 0
10 55 sb Truncate-cone shaped PE jars with red polycarbonate (PC) screw cap; currently used for urine samples 0 0 0
11 320 c Cylindrical PE containers; with a cap of the same material which is placed under pressure 12 0 0
12 32 c Polycarbonate (PC) transparent tubes; with a green screw cap of the same material; conical bottom 0 0 0
13 14 c Transparent PC tubes with a red tight cap; currently used for blood samples 0 0 0
14 6 c Special transparent polyolefin tubes; with red screw cap; used for cryo-preservation (cryotubes) 0 0 0
15 -- sb Transparent 10 x 6 cm low density polyethylene (LDPE) bags; sealed all-around by heat 0 0 0
16 -- c Transparent 11 x 7 cm LDPE bags: sealed by heat but closed by pressure on one side 0 0 0
17 2000 c Polyethylene-tereftalate (PET) 2 litre soft-drink bottles; metal screw cap with plastic seal 0 0 0
18 400 c Polyethylene-tereftalate (PET) mineral water bottles; with PE screw cap 0 0 0
Metal containers
Number Vol. ml. Origin Description Performance after
1 yr. 2 yr. 3 yr.
19 500 sb Cylindrical brass cans; with top sealed on a plastic seal by a heating machine 100 100 100
20 100 sb Cylindrical aluminium canisters; with a screw cap of the same material and an adhered plastic seal 52 44 36
Foil (metal+plastic) bags
Number Vol. ml. Origin Description Performance after
1 yr. 2 yr. 3 yr.
21 -- sb 10 x 5 cm three layered laminated polyvinyl / aluminium foil bags (see text) 80 72 60
22 -- sb 12 x 9 cm two layered laminated polyvinyl / aluminium foil bags 0 0 0
23 -- sb 18 x 11 cm laminated polyvinyl-aluminium-paper complex bags; two sets from different banks 0 0 0
Glass containers
Number Vol. ml. Origin Description Performance after
1 yr. 2 yr. 3 yr.
24 1240 sb Kilner jars, with glass lids fastened on a red rubber seal; metallic joint and fastening device 100 100 100
25 375 sb Jars (marmalade type); with twist-off metallic caps of same diameter treated with plastic on its lower side 92 92 88
26 140 sb Same as 25 but smaller in size 100 96 76
27 750 c Similar to 25 and 26 but of larger size 100 100 92
28 250 c Bottles with plastic screw cap with a polyethylene seal; of the typeused for liquid chemicals; three sets from different manufacturers 100 100 100
29 250 c Similar to 28 with a dripping removable device 100 920 84
30 190 c Bottles with narrow neck and plastic screw cap with a thin flat plastic seal; otherwise similar to 28 0 0 0
31 8 sb Flame sealed glass ampoules; made from standard 20 ml glass laboratory test tubes 100 100 100
32 25 c Small glass vials with white flip plastic top; normally used for fixing biological material 0 0 0
33 25 c Small glass vials with white screw cap; also normally used for fixing biological material 12 0 0
34 20 c Glass tubes with a white polypropylene screw cap 100 100 92
35 14 c Glass tubes with a rubber stopper (see text) -- -- --
36 20 sb Glass tubes with a cork stopper (see text) -- -- --
37 6 sb Small glass tubes tightly covered above with a lamina of paraffin film ('parafilm') 0 0 0
38 400 c Glass bottles used for fruit juices; with twist-off metallic caps 56 52 48
39 42 sb Glass cylindrical vials with striated pressure cap 36 0 0
40 250 c Laboratory flasks with long tubular neck; emery polished glass stopper; without paraffin sealing 48 0 0
Only these four containers conserved their silica gel blue after 3 years

Figure 1. Only these four containers (out of forty types studied) conserved their silica gel blue after 3 years (numbers as in Table 1).

Doubtful intermediate changes in silica gel colour were rarely detected. This is in concordance with the relatively quick hydration occurring (Table 1) when a via for moisture intake – even minimal - exists.

Whenever applicable, fastening of any cap or lid was done as in the routine seed bank practice: as firmly as possible without damaging the screw, twist-off or similar mechanism. Also, when applicable, sealing was done according to the common practice for each particular case.

Twenty-five units were tested for each type of transparent or translucent container, where changes in silica gel colour could be directly observed. With few exceptions, observations were made every six months for three years. For opaque containers, a larger number of units (50 to 100) was stored and successive sets were opened yearly.

No statistical analysis was performed because trying to elucidate which containers are simply bad, which are worse or which is the worst, was considered meaningless. For those few keeping their gel blue for three years, the real problem is whether or not they will maintain this behaviour in the future.


The percentage of storage units in which the silica gel remained blue was scored after 1-3 years (Table 1). Brass sealed cans, ‘Kilner’ jars with a rubber seal, laboratory bottles of the type used for liquid chemicals and flame-sealed glass ampoules (numbers 19, 24, 28 and 31, respectively) were the only containers in which all tested units retained blue gel coloration (Fig. 1).

All of the other containers (Fig. 2) performed poorly, and moisture entered relatively quickly in most cases. Though only yearly data are shown, a few weeks or months were often sufficient time for the silica gel inside to turn pink. All plastic-based containers were poor performers.

Containers 35/36 also performed poorly though they are not comparable to others because they had previously been stored under other conditions.


The importance of using hermetic containers was already emphasized by Cromarty et al. (1985). In a survey by Freire and Mumford (1986) containers were evaluated by including seeds and checking their subsequent germination. Instead, we have opted for a more straightforward physical approach avoiding the use of any intermediate complex biological system.

Plastics have variable permeability to gases depending upon their thickness, the size of their pores and the size of the gas molecule. Therefore, diffusion of water vapour through them is a possibility (Vieth, 1991; Noegi, 1996). However, during our experimental period, diffusion of water vapour was probably significant only in cases such as 15, 16, 17 and 18, which have particularly thin walls. In all the other cases, poor performance is likely because of a lack of complete tightness of the cap (cap and container often have different dilatation coefficients). The numerous inefficient glass containers identified support this idea.

The response of laminated aluminium foil bags is important, since they have often been recommended and are widely used by many crop seed banks all over the world. Three-layered bags (number 21) are currently considered effective (Cromarty et al., 1985; Freire and Mumford, 1986; Anonymous, 1996), but the present results do not confirm this view. It has been suggested (Engels, personal communication), that the sealing might not have been entirely satisfactory in this particular case. Improved new brands might probably work but they should convincingly be tested and subsequently monitored. Further research on this topic seems important. Two-layered foil bags or those including paper (see numbers 22 and 23 in Table 1) proved to be especially inefficient. General inconveniences of foil bags are: a) irregularities of the seed coat have been shown to damage the internal lamination (Tao, 1992) and b) being opaque, they do not permit a subsequent monitoring with a coloured chemical indicator.

Marmalade twist off jars (numbers 25 and 27) did not behave satisfactorily in the present experiment; however, moisture intake was relatively slow and n. 27 suggests that other brands or perhaps n. 27 itself might eventually work. Again, a test before adopting any new container is important.

Many types of containers were unable to exclude moisture in the present experiment. According to these results, fourteen out of seventeen containers currently in use by twelve collaborating institutions belonging to four countries, did not exclude moisture effectively. Only three containers (19, 24 and 31) belonged to the high performance group. This apparently extended and continued use of inappropiate containers in seed banks – often within RH-uncontrolled cold rooms – is to create concern because large amounts of genetic material might be ageing too rapidly. The enormous amount of money and effort invested in germplasm collection and storage in the past decades might be considered at risk or partially lost.

The good performance of ‘Kilner’ jars (num. 24) suggests many useful applications. With a proper desiccant in its bottom, most models have enough volume to hold several large seed samples in paper bags. The rubber seal might perhaps need replacement every ten years or so. Number 28 corresponds to glass bottles of the type used for laboratory chemicals. These have well designed caps with efficient plastic seals; cap size and shape are standardized for bottles of different volumes. They could be used in a similar way although their narrow aperture could be an inconveniency. Glass ampoules (n. 31) have a small volume and cannot be opened unless they are broken, but are ideal for base or "black box" collections of species with small seeds (Cruciferae, Solanaceae, many Compositae, Amaranthaceae, etc.).

Containers with proven efficiency after three years might not necessarily maintain it after 10, 20 or more years. Only flamesealed glass ampoules (n. 31) have demonstrated moisture-tightness after 25-30 (35) years (Ellis et al., 1993; Ramiro et al., 1995; Maselli et al., 1999). Successful use of sealed glass after 110 years has also been reported (Steiner and Ruckenbauer, 1995).

Conditions favouring seed longevity (low temperature and low moisture) need not only to be obtained but also maintained. The above results suggest that the need for premature ‘regeneration’ may largely be due to failure to maintain low moisture levels as a consequence of using inappropriate seed storage containers. Seeds stored in non-hermetic containers will eventually become equilibrated with the high relative humidity (RH) values prevailing in any RH-uncontrolled cold room and will absorb moisture to dangerous levels. Thus, the beneficial effect of low temperature is counterbalanced to a large extent by the harmful effect of an increase in moisture content.

Seed bank managers tend to be optimistic with regards the containers they use (see Anonymous, 1996). The statement "my seeds germinate, thus my containers are good" is substantially false because under inadequate conditions seeds may age significantly before reaching the ‘shoulder’ of the survival curve where germination starts rapidly to decline. Simply drying seed samples and storing them in cold conditions is not an efficient tactic unless one is 100% sure about the integrity of the container. If not so, conducting germination tests for all the accessions in a large collection is not an attractive solution. As a matter of fact, monitoring based on periodic germination tests followed by regeneration when the germination rate drops to 75-85% (Anonymous, 1994) is only meaningful when there is a reasonably expectation for a long conservation period and such tests can be spaced accordingly.

A selection of poorly performing containers. Silica gel turned pink

Figure 2. A selection of poorly performing containers. In a variable proportion of each, silica gel turned pink (see their identification numbers and relative performances in Table 1)

An attractive alternative consists of drying the seeds to equilibrium with silica gel and placing an amount of this substance with the seeds inside the container, separated by means of some permeable material (cotton, filter paper, styro-foam, etc.). This method – first used in 1966 with ampoules of type 31 in the author’s laboratory (Gómez-Campo, 1972, 1987) has shown excellent results (Ellis et al., 1993; Ramiro et al., 1995; Maselli et al., 1999) and is successfully used today in a variety of ways by an increasing number of banks. Silica gel may also be placed in a small transparent bag, visible from outside. Only opaque containers cannot be monitored in this way. As a desiccant, silica gel produces seed moisture levels between 2.5 and 4.5%, thus 2-3% lower than other conventional methods. It maintains these levels indefinitely in moisture proof containers and, most importantly, warns for possible leaks by changing its colour. Another important, but little known, property resides in its ability to absorb toxic gases produced during seed ageing (Esashi, personal communication). For years, a controversy on whether or not an overdrying might be damaging to seeds (Vertucci and Roos, 1990; Walters and Engels, 1998) has referred to moisture levels just below those produced by the presence of silica gel or when desiccation is combined with very low temperatures (below -18º C). More recently, Buitink et al. (2000) predictions suggest that silica gel might originate sub-optimal moisture levels; this would mean that, after all, moisture might not be so important. If such predictions are practically confirmed, it should be pondered whether or not they are actually compensated by the advantages of preservation with silica gel. For the moment, the last method has a demonstrated efficiency.

In any case, when a cobalt or iron chloride moisture indicator is used, a new reasoning arises: "the indicator has the right colour, my seeds are in good condition". This will never be enough to eliminate germination tests but it could reduce their frequency from every 5-10 years to perhaps every 25-50.

In conclusion, any seed container in actual or prospective use should always be checked for its ability to exclude water vapour. Since similar brands may radically differ in their behaviour, suggestions on specific containers – including those from this article – should only be taken as an orientation.


This work was jointly supported by IPGRI (International Plant Genetic Resources Institute, Rome) and INIA (Instituto Nacional de Investigación y Tecnología Agrarias, Madrid). The final report was presented in Rome in February 9, 1997. The author is indebted to those seed banks providing their containers for testing; each manager has been directly thanked and informed on the performance of his own container. Also F. Pérez-García, J. Pérez, J. Chávez, G. Montáñez and J. Fagúndez provided valuable help in different stages of this work. Mrs. C. Coope, E. Gz. Tortosa and Prof. M. A. Cohn efficiently contributed to edit the manuscript. Thanks are also due to the UPM (Universidad Politécnica de Madrid) for providing a prompt method of publication after some referees persistently delayed it for a long time. Unpleasant or not, the above results can easily be reproduced by anybody, and it is important and urgent to save large amounts of germplasm from rapid ageing or loss.


  1. Anonymous (1994). Genebank standards Food and Agriculture Organization. International Board for Plant Genetic Resources. Rome.
  2. Anonymous (1996). Evaluation of seed storage containers used in genebanks. Report of a survey. International Plant Genetic Resources Institute. Rome.
  3. Breese, E.L. (1989). Regeneration and multiplication of germplasm resources in seed genebanks: the scientific background. International Board for Plant Genetic Resources (IBPGR). Rome.
  4. Buitink, J., Leprince, O., Hemminga, M. A. and Hoekstra, F.A. (2000). Molecular mobility in the cytoplasm: an approach to describe and predict lifespan of dry germplasm. Proceedings of the National Academy of Sciences. USA 97, 2385-2390.
  5. Cromarty, A.S., Ellis, R.H. and Roberts, E.H. (1985). The design of seed storage facilities for genetic conservation. Handbooks for genebanks. Number 1. International Board for Plant Genetic Resources (IBPGR), Rome.
  6. Diaz, O., Gustafsson, M. and Astley, D. (1997). Effect of regeneration procedures on genetic diversity in Brassica napus and B. rapa as estimated by isozyme analysis. Genetic Resources and Crop Evolution 44, 523-532.
  7. Ellis, R.H., Hong, T.D., Martin, M.C., Pérez-García, F. and Gómez-Campo, C. (1993). The long-term storage of seeds of seventeen crucifers at very low moisture contents. Plant Varieties and Seeds 6, 75-81.
  8. Freire, M. and Mumford, P.M. (1986). The efficiency of a range of containers in maintaining seed viability during storage. Seed Science and Technology 14, 371-381.
  9. Gómez-Campo, C. (1972). Preservation of West Mediterranean members of the cruciferous tribe Brassiceae. Biological Conservation 4, 355-360.
  10. Gómez-Campo, C. (1987). A strategy for seed banking in botanic gardens: some policy considerations. pp. 151-160 in Bramwell, D., Heywood, V.H. and Synge, H. (Eds.). Botanic gardens and the world conservation strategy. Academic Press.
  11. Harrington, J.F. (1972). Seed storage and longevity. pp. 145-245 in Kozlowsli, T.T. (Ed.). Seed biology. Volume 3. New York and London, Academic Press.
  12. Maselli, S., Pérez-García, F. and Aguinagalde, I. (1999). Evaluation of seed storage conditions and genetic diversity of four crucifers endemic to Spain. Annals of Botany 84, 207-212.
  13. Noegi, P. (1996). Diffusion in polymers. New York, Marcel Dekker.
  14. Ramiro, M.C., Pérez-García, F. and Aguinagalde, I. (1995). Effect of different seed storage conditions on germination and isozyme activity in some Brassica species. Annals of Botany 75, 579-585.
  15. Sackville-Hamilton, N.R. and Chorlton, K.H. (1997). Regeneration of accessions in seed collections: a decision guide. Handbooks for genebanks. Number 5. International Plant Genetic Resources Institute (IPGRI). Rome.
  16. Steiner, A.M. and Ruckenbauer, P. (1995). Germination of 110-year old cereal and weed seeds, the Vienna Sample of 1877. Verification of effective ultradry storage at ambient temperature. Seed Science Research 5, 195-199.
  17. Stoyanova, S.D. (1992). Effect of seed ageing and regeneration on the genetic composition of wheat. Seed Science and Technology 20, 489-496.
  18. Tao, K.L. (1992). Should vacuum packing be used for seed storage in genebanks? Plant Genetic Resources Newsletter 88/ 89, 27-30.
  19. Vertucci, C.W. and Roos, E.E. (1990). Theoretical basis of protocols for seed storage. Plant Physiology 94, 1019-23.
  20. Vieth, W.R. (1991). Diffusion in and through polymers. Munchen, Hanser.
  21. Walters, C. and Engels, J. (1998). Effect of storing seeds under extremely dry conditions. Seed Science Research 8, 3-8.


Containers 19, 24, 28 and 31 have kept their silica gel intact for three additional years (until March 2002) thus totalizing six years of satisfactory behaviour.

Another brand of smaller octogonal 'Kilner' jars behaved very poorly after a single year. This illustrates what is said above in the last paragraph of the Discussion. The case of numbers 28 and 30 (almost identical externally) is also demonstrative.

Experimental conditions used in this research (approx. 100% R.H. and room temperature) were designed to accelerate water intake for an early detection of defective containers. Such intake should be slower for lower humidities and/or temperatures.

However tempting, the use of poorly performing containers with larger amounts of gel (to be substituted whenever necessary) cannot be recommended because silica gel is able to absorb a certain amount of water before colour changes are observable.