After the collection of seeds, current protocols for curation in seed genebanks (i. e. Genebank Standards, 1994) roughly consist of the following steps:
Although theoretical predictions (Harrington, 1972) strongly suggest the possibility of preserving orthodox seeds for centuries, many managers of seed genebanks have become increasingly doubtful of this over the years. In fact, none of the oldest seed banks has ever published encouraging germination results after only 40 years. It is not surprising that there are widespread worries about the need to face the burdensome work and the practical problems posed by regeneration. This pessimistic approach is even reflected in some of the presently available protocols and recommendations, such as those referring to the duration of regeneration cycles or the frequency of periodic germination tests.
In a recent article involving seed material from the UPM genebank in Madrid, Pérez-García et al. (2006) report average germination rates of 97.8 % for Brassicaceae after almost 40 years of storage. The only methodological difference applied in this bank simply consists of an extreme care in the control of low seed moisture by using silica-gel (Gómez-Campo, 1972, 2002). This brings back renewed hopes for an efficient long term seed preservation and for more optimistic approaches in the preservation protocols. If orthodox seeds can really stay alive for centuries, several aspects in the current recommendations and protocols obviously become obsolete. Below are some views on how they could be updated.
For well preserved seed accessions with a germination rate close to 100 % after forty years, the initial germination tests become largely irrelevant because the initial viability can never be lower than the final value. For wild species -and for some crop species- they may also be misleading because lower values might be due to dormancy and the dormancy levels may change significantly during seed storage (Ellis et al. 1993; Pérez-García et al. 2006).
An initial value of 65 %, for instance, does not mean that 35 % of the seeds are dead since a large proportion of this 35 % might simply be dormant. This is important because germination tests -initial or periodic- as they are currently performed, do not measure "viability" but only "germinability". The margin of error is smaller in crop species because they normally exhibit no dormancy or low levels of it. However, it is prudent to take this into account.
It is, therefore, important to remove any dormancy. For a high number of plant species from temperate regions, treatments based on scarification or on gibberelic acid are very effective at removing dormancy. A single test including a previous anti-dormancy treatment should lead to an accurate estimation of real viability.
Under these circumstances, initial indicative tests with only 10-15 seeds might be sufficient to rule out the need for further attention to a high number of well germinating accessions (unless special reasons to obtain more accurate data exist). If the germination is lower, these simplified tests can be repeated with some anti-dormancy treatment which could equally serve to make it unnecessary to study many other accessions as well as to detect cases where dormancy is present.
The time and effort saved may be considerable and can be used, for instance, to explore other possible present types of dormancy or other dormancy-removing agents, to distinguish between orthodox and recalcitrant or semirecalcitrant species, to study especially rebellious cases in more detail, etc.
Seeds are commonly desiccated to moisture levels of 4-6 % while the possibility of desiccation to ultra-dry levels below 3 % has been largely ignored. In fact, the concerns about possible deleterious effect of too low moisture contents or "over-drying" have almost persisted until the present time (Walters & Engels 1998). This has brought about a chronic renouncement to extend the expected longevity of seed samples by approximately four to eight times (Harrington, 1972). In the meantime, physiologists from Reading University (Ellis, 1998; Ellis et al. 1993, 2006) have defended the opinion that ultra-dry levels are not harmful. In practice, the latter views are fully endorsed by all successive germination results from the UPM bank (Ellis et al. 1993; Ramiro et al. 1995; Masselli et al. 1999; Pérez-García et al., 2006) where low moistures obtained with silica-gel are used. Engels & Visser (2003) have finally admitted ultra-drying as a valid possibility although no literature source for it is provided.
For orthodox seeds, the use of ultra-dry levels and, particularly, the use of silica-gel within the seed containers merits an explicit recommendation. The advantages of using silica gel are multiple: a) it dries the seeds to ultra-dry levels, b) it efficiently maintains those moisture levels within waterproof containers, c) it warns by changing its colour if anything goes wrong and d) it provides a second important seed preserving mechanism by absorbing toxic gases produced during seed aging.
Kilner (Scotch)-type hermetic jars with silica gel inside (either in the bottom or in a transparent bag) could provide the best solution for large sized crop seeds and also to stop rapid aging of samples which had previously been preserved under deficient conditions. It is no exaggeration that the number of such samples can be estimated as several millions worldwide.
On the contrary, ultra-dry storage is particularly damaging and should be avoided for the so-called recalcitrant seeds. These correspond to many species from humid habitats (or from some mature arboreal ecosystems from temperate zones) and may not even tolerate normal drying to 4-6 %. In any case, their life span cannot be extended further than 2-3 years. An easy-to-design %test would help to recognise recalcitrant seeds in advance, since if they are kept for a time %(suggested: one month) in ultra-dry conditions they would certainly die. Semi-recalcitrant %seeds might not die but their germination rate would significantly drop.
Perhaps one of the major causes of failure in the performance of many seed genebanks consists in the use of inadequate containers (Gómez-Campo, 2002). Many containers are not fully hermetic in the long term, so that seeds become equilibrated with the high humidity prevailing in the cold room. In RH uncontrolled cold rooms, what is gained by low temperature is lost by an increased moisture content. Only four types of container - out of forty types tested - proved to be water-vapour-proof. In the long term, perhaps none can fully prevent water uptake, thus the use of silica gel with moisture indicator is an excellent system to monitor internal moisture conditions. In this case, container walls should be transparent.
For seeds that can stay almost fully alive for 40 years, it is obvious that intermediate controls are largely useless. An enormous amount of effort can be saved if periodic germination tests are performed over longer periods (perhaps beyond 60 /75 years) than those currently recommended (5/10 years). When silica gel is placed with the seeds inside transparent containers, an efficient extra monitoring of seed storage conditions is added to the preservation system.
Again, the time and effort saved may then be used to study the evolution of seed dormancy during storage, or for any other purpose related to seed physiology and preservation such as those mentioned above.
Emphasis has often been placed on low temperatures. FAO/IPGRI standards (Geenebank Standards, 1994) recommend -18ºC but also warn against the tendency to over-emphasise the benefits of reducing temperature. UPM evaluation of a group of sealed glass vials which had been left at room temperature for almost 40 years yielded a mean germination rate of 91 % compared to the 97.8 % of those stored in the cold room for the same period (Pérez-García et al. 2006). Perhaps temperature is less important than it is normally considered to be and power-saving temperatures of -5º to 0ºC might be sufficient when the moisture is well controlled.
It is a reasonable recommendation to regenerate any seed accession whose germination rate drops below 85 % of its original value. However, it is firmly believed that an efficient seed preservation procedure could in most cases delay such situation for at least 100-200 years. This is an extremely important reason for using preservation procedures of demonstrated efficiency such as the silica-gel method. Regeneration is a source of unwanted crosses, unwanted selection, reduced genetic variability (thus, genetic erosion), mislabeling and other mistakes. It requires large amounts of time, labour, land space, money, etc. For wild material it may be better to recollect than to regenerate since, difficulties in reproducing the conditions prevailing in the natural habitats where the plants grow (i. e. photoperiod, substrate, rainfall, seasonality, presence of pollinators, etc.) must be added to the aforementioned inconveniences.
Emphasis on efficient seed preservation could extend the lifespan of seed samples for centuries. This should modify our views on the standards of seed bank curation activities. On the one hand, they should be adapted to the existence of more efficient preservation procedures and, on the other, they should be simplified to a large extent to save considerable unnecessary effort.
The contents of this article were presented within the side events of the FAO International Treaty on Plant Genetic Resources for Food and Agriculture in Madrid, June 2006. The author wishes to thank the organisers and also M.P. Ballesteros and M.E. Gómez-Tortosa for their valuable comments to the manuscript.