Viability monitoring

regarding Viability Monitoring

I would be interested to hear about the common practices regarding viability monitoring: how often and with how many seeds.

This contribution is especially about the latter. I know that some gene banks reduce the number of seed used in a test if they are short of seeds. Makes sense, but they maintain the thresholds, this is not correct! Let me explain.

Assume that the threshold for regeneration is a germination of 85% or below, and the usual number of seeds used in a germination test is 200, this could also be expressed as the chance that the germination is below 80% is lower than 2.8% (based on the binomial distribution). When only 50 seeds are used, and we would want to know that the chance of the germination being below 80% is less than 2.8%, we need to use a threshold of 90%. (The ‘sequential germination test’ as proposed by Ellis et al way back in 1985 is based on this principle.)

This implies that if the seed lot is very good (>90%) of if it is bad (<70%), you will find out with 50 seeds. However if it is in the critical zone, the sampling error caused by the small sample can cause false decisions, i.e., regeneration if it is not needed or the opposite. Adjusting the threshold can reduce the chance of erroneously not regenerating, but will obviously increase the chance of regenerating without need. Additional testing can solve this problem.

Conclusion, if you are short of seeds, use fewer seed for your germination test, but be aware that the threshold might need to be adjusted accordingly.


PS From the CGN Quality Manual (we are NEN-EN-ISO 9001:2008 certified)
– Seeds are tested in intervals between 20 and 30 year (depending on capacity). If the last germination was below 90% or if it concerns specific crops (such as potato, lettuce, pepper and onion), this interval is 10 years.
– If the germination drops with 15% or more as compared to the previous test, or if the germination is below 80% (or 60% for wild species), the material needs to be regenerated as soon as possible, but always within 10 years. If in specific cases these standards are not followed, this is only allowed with approval of all curators.
– All activities and decisions are logged and checked by external auditors.


  1. Theo van Hintum said

    this message was not meant to be anonymous 😉
    Theo van Hintum from the Centre for Genetic Resources, The Netherlands (CGN)

  2. Jan Engels said

    Dearb Theo,

    You made a very valid observation regarding the number of seeds to use within the “grey zone” and under conditions when seed numbers are low. I have no solution to offer other than to support the importance of understanding statistics. In this respect I would like to stress the importance of the sequencial seed testing that Richard Ellis et al. proposed and that is intended to allow statiscally sound tests to be consucted with FEWER seeds. Is this not the best answer on your question? (Sorry for only stressing this, you had already identified this “solution”!

    With kind regards,


  3. Robin Probert said

    Theo has made a good point and I agree with Jan that genebank managers should be familiar with the opportunity that sequential testing offers when dealing with small sample sizes. However, I thought I should share the approach we are planning to implement at the Millennium Seed Bank where we are also stuck with having to use small samples sizes in our germination testing. We will use our understanding of species differences in seed longevity and the results from recent extensive analysis of retest data over the last 30-40 years at Kew (to be published next year) to identify two categories of seed longevity: short lived and long lived (this might be extended to three categories in time). We will set a retest interval of 5 years for known short lived species and 10 years for long lived species. Retest data will be automatically analysed statistically. When there are just two tests (initial and first retest) we will use a Chi squared test and when there are three or more data points we will use probit analysis. The latter will be used to predict the storage time when viability will fall to 85% (new viability standard). At the first indication of a significant decline in viability subsequent retest intervals with be cut in half (10 to 5 or 5 to 2.5 yrs). If there is no detectable decline in viability after three or more consecutive tests, subsequent retest intervals will be extended (5 to 10 or 10 to 20 yrs). This strategy embraces the idea of active management that we discussed at the workshop.
    A management decision (regenerate or recollect) will be triggered when it is predicted that viability will reach 85% before the next scheduled retest.

RSS feed for comments on this post · TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: