Drying banana seeds for ex situ conservation

Abstract

The ability of seeds to withstand drying is fundamental to ex situ seed conservation but drying responses are not well known for most wild species including crop wild relatives. We look at drying responses of seeds of Musa acuminata and Musa balbisiana, the two primary wild relatives of bananas and plantains, using the following four experimental approaches: (i) We equilibrated seeds to a range of relative humidity (RH) levels using non-saturated lithium chloride solutions and subsequently measured moisture content (MC) and viability. At each humidity level we tested viability using embryo rescue (ER), tetrazolium chloride staining and germination in an incubator. We found that seed viability was not reduced when seeds were dried to 4% equilibrium relative humidity (eRH; equating to 2.5% MC). (ii) We assessed viability of mature and less mature seeds using ER and germination in the soil and tested responses to drying. Findings showed that seeds must be fully mature to germinate and immature seeds had negligible viability. (iii) We dried seeds extracted from ripe/unripe fruit to 35-40% eRH at different rates and tested viability with germination tests in the soil. Seeds from unripe fruit lost viability when dried and especially when dried faster; seeds from ripe fruit only lost viability when fast dried. (iv) Finally, we dried and re-imbibed mature and less mature seeds and measured embryo shrinkage and volume change using X-ray computer tomography. Embryos of less mature seeds shrank significantly when dried to 15% eRH from 0.468 to 0.262 mm³, but embryos of mature seeds did not. Based on our results, mature seeds from ripe fruit are desiccation tolerant to moisture levels required for seed genebanking but embryos from immature seeds are mechanistically less able to withstand desiccation, especially when water potential gradients are high.

Keywords: Crop wild relatives; desiccation tolerance; genebank; genetic resources.

Figure 1

Musa acuminata seed. mi, micropyle cap; en, endosperm; em, embryo; te, testa; ch, chalazal mass.

Figure 2

Desorption isotherm for seeds of three M. acuminata subspecies (bunches 1–3). Seeds were equilibrated in RH-controlled chambers using non-saturated lithium chloride, apart from 4% RH in a desiccator with silica gel and 100% RH with water. The black dashed line is fitted with non-parametric local polynomial regression; the grey shaded area is 0.95 confidence interval; the red dotted horizontal line is mean seed MC directly at extraction without washing. MC calculated on fresh weight basis. No statistically significant differences were found between bunches/subspecies so they are combined in the trend line here.

Figure 3

(a) ER outcomes and (b) tetrazolium chloride outcomes, modelled from MLR based on seeds (bunches 1–3) equilibrated to different RHs.

Figure 4

Results from seeds of two bunches of M. balbisiana at different levels of maturity (bunches 4–5). (a) MC (fresh weight basis) during desiccation in a desiccator with silica gel; ‘fresh’ is without washing, ‘surface’ is after 24 hours surface drying in ambient laboratory conditions after washing, otherwise days are time in a desiccator (n = 3 × 10). (b) ER results as percentage of sample forming shoots within 28 days of initiation samples removed from a desiccator each 24 hours; ‘control’ seeds were left in ambient conditions for 4 days (n = 40). (c) final germination results 55 days after sowing in soil. Whole seeds were sown and seedlings were recorded and removed daily; seeds were covered with 5 mm of soil and exposed to sun in a screen house; seeds were 5 days after extraction maintained at ambient conditions without desiccation (n = 200).

Figure 5

Estimated marginal means of germination based on GLM with quasibinomial error structure; estimates back-transformed to probability scale. P-values (shown above error bars) are contrasts against control seeds that had not been dried. M. balbisiana seeds used (bunch 6), fresh and after three levels of drying and two levels of ripeness. Control seeds were extracted 2 weeks later on return to the laboratory. Seeds were sown in the soil at Plant Resources Center, Ha Noi, Viet Nam, and monitored for 78 days.

Figure 6

Estimated marginal means (and standard errors) of embryo (a) and seed volume excluding embryo and air space (b), based on repeated measures linear mixed model. P-values shown from Dunnett’s test against fresh seeds (‘fresh’), for seeds equilibrated at 15% RH and 100% RH for at least 14 days without imbibition and with imbibition on agar for 7 days (‘imb’) (bunches 7–8).

Figure 7

A 3D visualization of M. balbisiana seeds (bunches 7–8) from CT scans. Displayed seeds were either surface dried only and left in ambient conditions or equilibrated to 15% RH for 14 days and then either scanned immediately after removal from 15% RH conditions or re-imbibed for 7 days on agar at 20°C. Embryos are coloured blue; red shows empty surfaces within the seed where moisture has been desiccated.

Figure 8

Plot of MLR model of ER results of both mature and less mature M. balbisiana seeds (bunches 7–8) used in experiment 4.