Assessing the storage potential of seed collections to inform the management of wild species seed banks

Introduction

The storage of seeds in seed banks is a primary strategy for the ex situ conservation of plant species and associated genetic diversity (Walters and Pence 2021). Wild plant species may be threatened by many factors including habitat loss, changing climate, pests and diseases. Ex situ conservation programs aim to prevent species extinctions and support species survival in the natural habitat by ensuring the availability of genetic materials in specialised repositories such as seed banks, and providing propagation material for plant translocations and broader ecosystem restoration (Martyn Yenson et al. 2024). Monitoring the viability of seed collections can be a significant challenge for any seed bank (Fu et al. 2015; Hay and Whitehouse 2017), and for seed banks conserving wild species, the diversity of species creates some challenges with respect to establishing effective procedures and workflows (Hay and Probert 2013; Walters 2015). Seeds of wild species may be banked with little direct prior knowledge of seed storage behaviour or germination requirements, meaning that evidence-based curation is an ever-increasing challenge as the number and diversity of the collections grow. The increasing scale and pace of the threats to plant biodiversity mean that the long-standing efforts to upscale seed banking programs are vital for the capture and conservation of plant genetic diversity before this is lost (Liu et al. 2018). Management of collections must be informed by knowledge of seed biology and storage behaviour for seed banks to achieve the goal of preserving and providing high quality germplasm when required.

Periodic testing of the viability of seed collections is a core component of seed bank management. The goal of viability monitoring is to identify declining seed quality before the utility of the accession is compromised, without testing too frequently so as to consume valuable seed stocks unnecessarily (Hay and Whitehouse 2017). International genebank standards recommended viability testing intervals be set at 5 years for short-lived seeds or 10 years for long-lived seeds, or if there is knowledge of the anticipated seed longevity, at one-third of the time for viability to decrease to 85% of the initial viability (FAO 2014). However, the storage behaviour of wild seeds is diverse and seeds of different species vary in the degree of desiccation tolerance, survival of sub-zero storage temperatures and moisture/temperature relationships with longevity (Walters 2015). Effective viability monitoring can also be challenging due to uncertainties surrounding seed germination requirements and the resources required to test large numbers of accessions (Fu et al. 2015). Together with large, diverse collections, many seed banks have seed lots that were banked decades before, some of which may have variable storage histories and limited information regarding initial seed quality or viability. The seed bank at Kings Park and Botanic Garden holds ~14,000 accessions of 4000 species native to Western Australia (WA), with collections dating to the early 1960s. Originally established to provide seeds for the development of plant displays within the Western Australian Botanic Garden, the seed bank program has grown to include a greater focus on the conservation of WA’s flora and supporting the restoration of urban bushland under the management of the Botanic Gardens and Parks Authority. The storage conditions of many of the older collections have changed over time due to the improved understanding of seed banking practices and updates to infrastructure. For many collections, storage history is imperfect and there is a lack of data on the current and/or initial viability to conduct analyses of seed viability in relation to time in storage. Germination testing can provide data on current quality and utility but provides little information on future utility, i.e. longevity. Studies of seed longevity through seed storage experiments can provide additional data regarding the physiological seed quality and can be used to compare the longevity of different species and seed accessions (Hay and Whitehouse 2017; Merritt et al. 2021). In this regard, conducting seed storage experiments could be a useful additional seed bank curation tool to better delineate the quality of seed accessions of similar germination and provide information on the future storage potential to inform scheduling of future viability monitoring and recollection intervals (i.e. for longer vs shorter-lived accessions).

The longevity of desiccation tolerant seeds varies by four orders of magnitude under high temperature and relative humidity storage conditions that have been widely used to rapidly age seeds for experimental purposes (45°C and 60% relative humidity (RH); Newton et al. (2014) and Hay et al. (2022)). The time for seed viability of Australian species to decline to 50% (p₅₀) under such storage conditions has been documented to range between <1 day and >770 days (Probert et al. 2009; Merritt et al. 2014; Satyanti et al. 2018; Sommerville et al. 2023). These studies and others examining seed longevity across a range of storage conditions, including genebank conditions (Walters et al. 2005; Chau et al. 2019; Colville and Pritchard 2019), have identified longer- or shorter-lived seeds to be associated with certain plant families. Shorter-lived seeds have been found within, for example, the Apiaceae, Araliaceae, Asteraceae, Brassicaceae, Poaceae and Orchidaceae. Such associations identify families of interest for research into storage behaviour and longevity but wide variation in seed longevity is also observed within families and longevity can even be specific to different seed lots of a single species (Lee et al. 2019), making generalising storage behaviour and potential longevity in the absence of any prior data difficult.

In this study, we tested the germination of banked seed collections of species of Asteraceae and Araliaceae, as plant families putatively having species with shorter-lived seeds. We quantified the seed germination of each accession to identify any trends in viability decline amongst the accessions and in relation to time in storage. For species for which several accessions were determined to be of equal current viability, we conducted a seed storage (rapid ageing) experiment to differentiate the longevities of these accessions, hypothesising that seed longevity will be lower in older accessions due to prior time in storage, often under variable conditions. Our goal was to examine the utility of experimental seed ageing for identifying the storage potential of seed collections of unknown quality to inform the scheduling of future viability monitoring and/or use of such collections.

Materials and methods

Results

Seed viability testing

Final percentage seed germination varied significantly between accessions for all species except H. cotula, where all three accessions germinated to 95–96% (P = 0.93), P. angustifolia (100% germination in all four accessions (P = 1.0)), and T. pilosa (98–100% germination in all six accessions (P = 1.0; Fig. 1)). Germination of the four accessions of B. iberidifolia collected between 1999 and 2016 was also relatively high, ranging between 71 and 82%, with no significant difference between these accessions (P ≥ 0.41) and only the seeds collected in 1982 recording zero germination.

Fig. 1.

Cumulative germination curves for the 10 study species, with multiple accessions from different collection years. Seeds of Olearia pimeleoides collected in 1987 did not germinate – the line is obscured by data for the 1989 collection.

In the other species, seeds of at least one accession germinated to a high percentage (>75–80%) but there was also significant variation in germinability between different accessions. In most of these species (M. gueriniae, O. axillaris, O. pimeleoides and P. cordata), the lowest germination was associated with the oldest accession tested but a sequential decline in germination with accession age was not observed, and only in P. cordata and O. pimeleoides did the youngest accession germinate to the highest percentage. Seeds of M. gueriniae collected in 1995 germinated to 78%, significantly higher (all P < 0.014) than seeds collected in 1990, 1996, 2003 and 2016 (between 0 and 18% germination). Seeds of O. axillaris collected in 2007, 2010 and 2014 germinated to 84–88%, significantly higher (P < 0.001) than seeds collected in 1992 (0%) and 2015 (39%). Only seeds of O. pimeleoides collected in 2003 germinated (84%), with no germination of seeds collected in 1987 and 1989. P. cordata seeds collected in 2011 germinated to 94%, significantly higher (all P < 0.05) than in the four other accessions that ranged from 25% (1992 accession) to 83% (2005 accession). Seeds of R. sterilescens had relatively high germination across all accessions but there was a significant difference (all P < 0.004) in germination between seeds collected in 1992 (94%) and seeds collected in 1988, 1998 and 2016 (68–76%). Seeds of one of the accessions of P. lessonii collected in 2000 germinated to 77%, but those of the other accession collected in 2000, and those collected in 2003 and 2004 had significantly lower (all P < 0.003) germination (51–54%).

Seed longevity in experimental storage

Slope values of transformed survival curves could be constrained to a common value (i.e. common 1/σ) for seeds of H. cotula, O. axillaris, P. angustifolia and T. pilosa (Table 2) but could not be constrained for B. iberidifolia or P. lessonii (Table 2). K_i ranged from 0.1 for one of the P. lessonii seed lots collected in 2003 (20040681) to 4.38 for P. angustifolia seeds collected in in 2007 (Table 2). K_i was <1 for several collections of B. iberidifolia and P. lessonii. K_i could be constrained to a common intercept for H. cotula and O. axillaris. The p₅₀ values for all species and accessions varied between <1 day and 58.78 days, with seeds of P. lessonii (0.56 days ≤ p₅₀ ≥ 4.98 days), O. axillaris (5.57 days) and B. iberidifolia (4.67 days ≤ p₅₀ ≥ 15.62 days) being much shorter-lived than seeds of H. cotula, P. angustifolia and T. pilosa, in which p₅₀ values generally ranged between 40 and 50 days (Fig. 2, Table 2).

Table 2.Information for study species rapidly aged at 45°C and 60% RH, including results from probit analysis.

Species	Accession	Collection year	Pre-ageing germination (2022, %)	Model	Ki ± s.e. (NED)	1/σ ± s.e. (days−1)	σ (days)	p50 ± s.e. (days)
Brachyscome iberidifolia	20000426	1999	74	Independent models	0.70 (0.117)	0.13 (0.015)	7.69	5.34 (0.728)
	20020768	2000	82		0.86 (0.112)	0.09 (0.008)	11.11	9.90 (0.966)
	20040734	2003	71		0.50 (0.111)	0.11 (0.012)	9.09	4.67 (0.861)
	20171454	2016	76		1.04 (0.129)	0.07 (0.006)	14.29	15.62 (1.23)
Hyalosperma cotula	20060615	2005	95	Ki and σ constrained	2.31 (0.126)	0.05 (0.003)	20.00	42.48 (0.995)
	20080060	2007	96
	20100160	2009	95
Olearia axillaris	20061736	2007	86	Ki and σ constrained	0.84 (0.065)	0.15 (0.01))	6.67	5.586 (0.349)
	20100811	2010	88
	20140749	2014	84
Panaetia lessonii	20010288	2000(a)	77	Independent models	0.70 (0.104)	0.14 (0.016)	7.14	4.98 (0.623)
	20010648	2000(b)	52		0.17 (0.105)	0.17 (0.027)	5.88	1.01 (0.602)
	20030441	2003(a)	51		0.10 (0.111)	0.11 (0.016)	9.09	0.90 (1.048)
	20040681	2003(b)	54		0.14 (0.114)	0.26 (0.046)	3.85	0.56 (0.448)
Podotheca angustifolia	20040253	2003	100	σ constrained	3.65 (0.201)	0.08 (0.004))	12.50	49.40 (1.478)
	20061247	2005	100		3.01 (0.172)			40.30 (1.320)
	20080478	2007	100		4.38 (0.239)			58.78 (1.672)
	20120326	2011	100		3.91 (0.213)			52.36 (1.537)
Trachymene pilosa	20020589	2001	100	σ constrained	3.83 (0.186)	0.08 (0.003)	12.50	49.88 (1.476)
	20040574	2003	98		3.48 (0.167)			45.35 (1.361)
	20061202	2004	100		2.12 (0.124)			27.64 (1.170)
	20061215	2005	100		3.31 (0.162)			43.13 (1.325)
	20080458	2007	100		3.82 (0.180)			49.76 (1.443)
	20150067	2014	100		3.85 (0.181)			50.17 (1.452)

Where K_i or σ could be constrained across accessions within a species, results of the constrained model are provided. Pre-ageing germination data are summarised from data for relevant accessions presented in Fig. 1.

Fig. 2.

Survival curves of rapidly aged (at 45°C and 60% RH) seeds fitted by probit analysis for different collections of six study species.

In most species except H. cotula and O. axillaris, p₅₀ varied between the accessions tested and there was no consistent relationship between p₅₀ and collection year. For example, for P. lessonii the oldest collection from 2000 had the greatest p₅₀ and the greatest remaining longevity.

Discussion

Seed bank managers often contend with historical seed collections with incomplete records, changes to facilities, infrastructure and storage practices over long (e.g. decadal) timeframes, and resource constraints – all while new collections continue to accumulate (Hay et al. 2021). The Kings Park seed bank comprises many accessions for which there are no prior data or knowledge on seed viability or longevity. We tested the viability of multiple seed accessions, spanning time in storage of many decades, of a range of species hypothesised to be shorter-lived in storage. In accessions determined to have similar levels of viability, we applied a rapid ageing protocol to better discern differences in seed quality and longevity between the accessions. Viability testing of seeds retrieved from storage showed that for three species (Hyalosperma cotula, Podotheca angustifolia and Trachymene pilosa), germination was ≥95% across all accessions regardless of age, with storage periods ranging between 17 and 21 years. In the remaining species, there were significant differences in germination between accessions but no relationship was evident between collection year and seed viability. In some species (e.g. Brachyscome iberidifolia, Olearia axillaris and O. pimeleoides), accessions collected and banked in the 1980s or 1990s had poorer viability (0–20%) than more recent accessions, as expected. However, there were also examples of species for which older accessions retained high viability – in some cases, more so than those collected more recently. For example, Myriocephalus gueriniae seeds collected in 1995 germinated to 80%, far higher than all more recent accessions (all <20%). Similarly, for Rhodanthe sterilescens, seeds collected in 1992 had greater germination than the other accessions tested, including two accessions collected more recently in 1998 and 2016. This is despite seeds of the 1992 collection of R. sterilescens being subjected to poorer storage conditions than the more recently collected seeds stored continuously at −18°C. Notably, the range of conditions (temperature and RH) under which seeds were stored varies substantially amongst accessions, and notwithstanding the poor seed germination of many of the oldest accessions (i.e. those stored in an air-conditioned room at 23°C, up until moving to −18°C storage in 1997), there was no consistent relationship between storage conditions and current germination. Based on the eRH determined at the time of seed retrieval from storage for viability testing (Table 1), many accessions have been stored at relatively high moisture contents but these accessions did not have lower germination than seeds stored following drying to 15–20% RH. For example, seeds from all collections of Podotheca angustifolia and Trachymene pilosa germinated to ~100% despite RH (measured at the time of retrieval) varying substantially between accessions, ranging from 15 to 50% (Table 2). Such results demonstrate that storage history is not necessarily an accurate indicator of seed quality and is not instructive for decision-making on seed curation of these collections.

Seed longevity determined via rapid ageing experiments (p₅₀) varied between accessions of each species and older accessions did not have lower longevity, as originally hypothesised. There were incidences of p₅₀ of older accessions being significantly greater than more recently banked accessions (e.g. Panaetia lessonii and Podotheca angustifolia) and for most species, longevity varied in a manner unrelated to prior time in storage. The differences in longevity between accessions of a species may be attributed to differences in both the initial viability (K_i) at the commencement of the ageing experiment and the rate of viability loss (1/σ), or both (Table 2). The rapid ageing experiment thus distinguished differences in initial viability amongst accessions that were not evident through the germination testing conducted when seeds were first removed from storage (i.e. the pre-ageing germination data presented in Table 2). For example, seeds of Podotheca angustifolia collected in 2005 had a K_i of 3.01, while the K_i of other accessions ranged between 3.65 and 4.38 despite germination testing showing all accessions to have 100% viability prior to the rapid ageing experiment. These data point to factors such as seed maturity at the point of collection, and post-harvest handling practices including cleaning, drying and banking procedures, and the influence of these on seed quality at the time of banking likely being more influential to the future longevity than prior time in storage.

Seeds of all accessions were determined to have p₅₀ values < 50 days. These values are consistent with those of other species from seasonally dry regions of Australian with shorter-lived seeds, and with p₅₀ data for other Australian Araliaceae and Asteraceae (Probert et al. 2009; Merritt et al. 2014, 2021). Based on p₅₀ data from similar comparative longevity studies for a broad range of species from across the globe (Colville and Pritchard 2019), these data characterise the study species as having seeds with ‘short’ to ‘medium’ longevity (Mondoni et al. 2011). It should be noted, however, that because all accessions were stored for many years prior to the ageing experiments the p₅₀ values do not account for the potential decline in longevity during storage. Whilst most species we selected for the ageing experiments comprised multiple accessions of relatively high (>85%) initial germination (except Brachyscome iberidifolia and Panaetia lessonii), meaning that the p₅₀ values are broadly comparable to those presented in the studies of Probert et al. (2009) and Merritt et al. (2014) calculating p₅₀ for freshly collected seeds would give additional weight to these species-level comparisons of seed longevity. All species had some accessions that retained high viability after ~15–20 years of storage, even with storage under sub-optimal conditions, providing confidence that seed banking is practical for germplasm conservation of these species over at least this timeframe.

We found many accessions to have a high RH at the time of retrieval from the seed bank for study. Many accessions were stored prior to the installation of the controlled environment drying room at Kings Park in December 2005, and the wide variation in RH between these accessions likely reflects the post-harvest processing practices of the time and the drying of seeds under ambient environmental conditions. One accession (Myriocephalus gueriniae collected in 1991) was measured at 72% eRH, indicating a likely failure of the seal or puncture to the laminated foil bag. Also, despite the installation of the seed drying room in December 2005, we identified 13 collections with an eRH above the target range (i.e. 15–20% RH), indicating incomplete drying of seeds and lack of procedural rigour by seed bank operational staff at the time of banking or poor dry-chain management procedures for accessions that were periodically removed to retrieve seeds.

Conclusions

Experimental ageing and the resultant p₅₀ data provide insights into the initial viability and longevity of freshly collected seeds at the time of banking, and accessions that lose viability more rapidly can be noted for more frequent monitoring (Hay and Whitehouse 2017; White et al. 2023). We used this approach to distinguish seed quality factors for previously banked accessions of wild species postulated to be short-lived and subject to differing storage environments. Storage history was not an indicator of the likelihood of higher or lower germination of the accessions upon retrieval from the seed bank and based on the p₅₀ data, is not informative regarding future storage potential. The rapid ageing experiment revealed differences in seed quality amongst accessions that were not evident in a single germination test. These data can be used to schedule future germination re-tests, and prioritise accessions for utilisation and species for recollection.