Two decades of climate change alters seed longevity in an alpine herb: implications for ex situ seed conservation

Climate warming in mountain areas is increasing faster than the global average, threatening alpine plants. Climate affects many traits including seeds, the longevity of which is important for conservation, facilitating genebank storage. Seeds of alpine species are considered short-lived in storage, but their longevity increases when produced under a warmer parental environment. Consequently, with climate warming, seeds of alpine species may have increased fitness and be longer-lived in genebank storage. We assessed seed longevity under artificial ageing in 10 accessions of the arctic-alpine species Viscaria alpina stored under genebank conditions for different time periods over the last 20 years. The seed collection site was in the northern Apennines, where above average warming and variable precipitation has been recorded. The time taken for viability to fall to 50% (p50) was estimated using probit analysis; correlation and general linear regression were used to investigate the effects of length of time in storage, seed mass and climate under which seeds were produced on seed longevity. p50 varied between 7.77 and 18.49 days. There was no relationship between length of time in storage or seed mass on seed longevity. p50 was higher in years with increased temperature and lower precipitation during the growing season, with precipitation having more impact than temperature. The results suggest that seeds of alpine species are suitable for genebank storage, and inter-annual variation in precipitation induces a plastic response in seed longevity. Using genebank stored seeds provides insights into how alpine species may respond to future climate changes and could have implications for genebank storage.


Introduction
Anthropogenic-induced climate change is occurring at different rates around the world with some regions more vulnerable than others, including alpine areas which are regarded as climate change hotspots (IPCC 2018). Climatic changes in these areas have already been observed from long-term records (e.g. Marty and Meister 2012), and models predict that over the next decades the temperature in high mountain regions will increase by 0.3 °C per decade, higher than the expected global average increase (Hock et al. 2019).
There is evidence that temperatures at high elevations are already increasing faster than the global rate, with a variety of contributory, interconnected factors (Pepin et al. 2015). Precipitation trends are more difficult to quantify and vary depending on area, time period and season (Gobiet et al. 2014). Despite this, warming has consistently decreased the depth and duration of snow cover, especially in spring (Gobiet et al. 2014), leading to longer and drier growing seasons (Taylor and Seastedt 1994;Calanca 2007). Climate change in mountain regions is likely to be heterogeneous, with some areas experiencing larger changes than others (Hock et al. 2019). Of particular concern are Mediterranean mountains where, in addition to increasing temperature, a large decrease in rainfall is predicted, unlike in other European mountain ranges (Nogués-Bravo et al. 2008). A large body of literature shows that the impacts of these climatic changes are already being seen with alteration in distribution, phenology and physiology of several plant species (see Winkler et al. 2019 for review). Species at the edge of their distributional ranges are more likely to experience climatic 1 3 extremes related to climate change, meaning these populations may be better adapted to the climate under future scenarios but are also vulnerable to extirpation .
In alpine environments, seeds play an important role as the main vehicle for plant regeneration (Walck et al. 2011), migration (Parolo and Rossi 2008) and persistence (Schwienbacher and Erschbamer 2001). Understanding the impact of climatic changes on seed traits is important because the early stages of plant development are highly sensitive and could represent a recruitment bottleneck (Walck et al. 2011). For example, despite their limited capacity for long-distance seed dispersal (Morgan and Venn 2017), differences in diaspore traits helped distinguish weak from strong colonisers in alpine plants (Vittoz et al. 2009), which may play a crucial role in their response to climate warming. Climate warming has already been shown to affect other seed traits such as soil seed bank persistence by increasing the number of species emerging from the seed bank but decreasing the total number of germinants (Hoyle et al. 2013). Several studies have shown that warming either increased or had no effect on seedling recruitment in alpine species (e.g. Meineri et al. 2013;Angers-Blondin et al. 2017;Klanderud et al. 2017), whilst a minority reported a decrease (Shevtsova et al. 2009;Hoyle et al. 2013) or a shift in time of emergence . Despite this, our understanding of the effects of climate change on alpine seeds relies on a relatively small number of studies, mostly focussed on seed germination, whilst other seed functions, such as longevity, have received less attention (see Mondoni et al. 2022 for review).
Seed longevity, i.e. the germinability after time in a state of desiccation (Rajjou and Debeaujon 2008), is a fundamental trait for plant survival, contributing to persistence in the soil seed bank until conditions are suitable for germination (Saatkamp et al. 2014). Many species produce orthodox seeds which are desiccation tolerant, enabling them to enter into a quiescent state and survive long-term in the soil, in some cases for many years (Long et al. 2015). The natural ability of seeds to survive for long periods has been exploited to facilitate long-term ex situ storage in genebank facilities. Seeds can be stored for many years under genebank conditions (typically at -18 °C, after drying to 15% eRH (equilibrium relative humidity), FAO 2014), up to decades or centuries (Walters et al. 2005;Hay et al. 2021). Despite this, seed ageing during storage is an inevitable process and an important consideration for genebanks when monitoring and refreshing their collections (FAO 2014). The loss of viability during storage varies between and within species, displaying characteristic patterns (by family or geographical area) that have been related to taxonomy (Walters et al. 2005), seed physiology (Tausch et al. 2019) and/ or climate (Probert et al. 2009). In this context, by testing seeds from 63 species collected from either an alpine or lowland environment, Mondoni et al. (2011) demonstrated the alpine seeds were comparatively shorter-lived than those from lowland habitats. This work suggests that seeds of alpine species may be less suitable for conventional genebank storage. Nevertheless, Bernareggi et al. (2015) demonstrated using open top chamber (OTC) experiments, that alpine plants exposed to artificially high temperatures produced longerlived seeds than plants under natural conditions, suggesting a rapid and plastic response to a warmer environment. These responses may underline adaptive transgenerational strategies to cope with more stressful environments (Herman and Sultan 2011) and indicate that alpine seeds might be more suited to genebank storage with climate warming. However, artificial manipulation of climate such as the use of OTCs, greenhouse or common garden experiments, do not reflect the complexities in nature of changes in temperature, precipitation and CO 2 concentration that occur with climate change (Kreyling and Beier 2013). OTCs can also produce unwanted side effects including altered light, moisture, wind and pollination (Marion et al. 1997). Moreover, there have been no studies on the effects of precipitation patterns on seed longevity. The effects of climate change on seed longevity of alpine plants therefore requires further investigation.
In this study, we explored the climate change effects on seed longevity of an alpine herb, Viscaria alpina (L.) G. Don (Caryophyllaceae), occurring at a site in the northern Apennines that has experienced considerable warming (0.6 °C) over the last two decades. V. alpina is an ideal model species because it shares many traits with other alpine plants including that it is a long-lived perennial. The population is at the southern edge of the range distribution so is already at its ecological limits, located in the Mediterranean-influenced northern Apennines, an area that is predicted to experience strong impacts from climate change (Nogués-Bravo et al. 2008;Hock et al. 2019). Using 10 seed accessions collected from the same population between 2001 and 2019 and stored in a seed bank, we first estimated the temporal trend in seed longevity changes over the study period. Specifically, we hypothesised that seed longevity under experimental ageing conditions will be lower in older accessions compared to more recent accessions, due to the negative effect of time in storage (Walters 1998;Rajjou and Debeaujon 2008). Second, we estimated the effects of climatic conditions at the collection site for each year on seed longevity. Seeds produced in warmer and drier environments are generally longer-lived, and this is hypothesised to promote persistence in hot, dry environments where there is irregular precipitation (Probert et al. 2009). Given the high sensitivity of alpine plant reproduction to temperature and precipitation regimes (Inouye 2008;Alatalo et al. 2021), we anticipated that seed longevity should be higher for seeds produced during warmer and drier growing seasons.

Study species and site
Viscaria alpina is an arctic-alpine plant with a global distribution in the northern hemisphere in Europe, Greenland and North America, predominantly in alpine and sub-arctic sites (Nagy 2013; Fig. 1). The species is a perennial herb with a life span of up to 10 years, composed of a semi-rosette of leaves and a single flowering stem that dies back during winter; it has a strong taproot that allows the plant to resist frost and snow cover (Nagy 2013). It is generally insect pollinated but self-pollination may occur (Nagy 2013). The seeds of V. alpina are 0.5-0.8 mm long and are dispersed passively once the seed capsules have matured in autumn (Nagy 2013). The soil seed bank of V. alpina at a site in Scotland was estimated to exist to a depth of 45 mm and contain between 300 and 1000 seeds m −2 (Nagy and Proctor 1996). Although V. alpina is reasonably widespread globally, in Italy the species is classed as vulnerable according to the IUCN criteria (Abeli et al. 2009).
The seeds for genebank storage were collected at the time of natural dispersal (August) in 2001-2019 from a population located on the summit of Monte Prado in the Northern Apennines, Italy (44°14'N 10°24'E, 2054 m above sea level; Fig. 1). The species here is at the southern limit of its range and is affected by the recent climate change: in years with higher temperatures plants produced more inflorescences (Abeli et al. 2012). Population numbers at this site are stable but extreme temperatures during heatwaves reduced the reproductive performance (Abeli et al. 2012). The site is within the Appennino Tosco-Emiliano National Park and the population is located on the northern slopes that consequently has longer lasting snow cover, but a rocky soil substrate allows good drainage. The climate of the study site has an annual mean temperature and precipitation of 6.2 °C and 1931 mm, respectively. Due to the influence of the Mediterranean Sea, the site experiences high snowfall in winter, with snow cover remaining into June (Abeli et al. 2012;Tomaselli et al. 2018). Seeds were collected from approximately 30 individuals located randomly within the whole population. Collected seeds were cleaned, dried and stored in accordance with FAO genebank standards (FAO 2014), i.e. in airtight glass containers with a rubber seal at -18 °C; after drying to 15% eRH at 15 °C (measured with a hygrometer housed in an AW-DI0 water activity probe in conjunction with a HygroPalm 3 display unit (Rotronic Instruments UK Ltd, Crawley, UK)), at the University of Pavia Seed Bank in Pavia, Italy. Accessions 2001 and 2005 were stored at the Millennium Seed Bank, Royal Botanic Gardens, Kew in the UK until 2019 when they were transferred to the University of Pavia in sealed aluminium packets and placed directly in the seed bank freezer.

Climate data/ climate change at the study site
The climate at the collection site was extrapolated from an online database provided by the Emilia Romagna Region (https:// arpae prv. datamb. it/ datas et/ erg5-eracl ito; Antolini et al. 2016). In the database, minimum and maximum daily temperatures were available, and the mean of these values 1 3 was calculated to produce the daily mean temperature. Daily total precipitation data were also available. For the period approximately covering our seed collection  the climate at the growing site has warmed by 0.03 °C per year (i.e. + 0.6 °C over 20 years), despite large inter-annual temperature variation (Fig. 2a)  To evaluate the long-term climatic change at the study site, we compared the temperature and precipitation data from the study period  with the corresponding means from 1961 to 1990 to allow an historical comparison of the climate. This time period is recommended by the World Meteorological Organization to be the baseline for the pre-warming 'climate normal' period (WMO 2015). Since the year 2000, the mean, annual temperature has always been above the historical mean and the precipitation has been lower than the mean for 13 of the last 20 years, meaning there has been double the number of comparatively dry years compared to wet years (Fig. 2c, d). The growing period (duration between snowmelt and seed dispersal) was estimated through snow cover observations at the site (from Abeli et al. 2012) and directly from a soil temperature logger (Tinytag, Gemini Data Loggers Ltd, UK) placed at the site since 2014. We selected June as the beginning of the growing season and the end of August as the latest time for seed dispersal.

Seed longevity in experimental storage
Seed longevity was determined using the standard rapid ageing protocol of Newton et al. (2014). Seeds were removed from storage and rehydrated at 47% RH at 20 °C in open glass vials; the vials were placed over a non-saturated solution of LiCl (anhydrous, Laboratory Reagent Grade; Fisher Scientific UK Ltd, Leicester, UK) in distilled water held in a sealed 300 × 300 × 130 mm electrical enclosure box (Ensto UK Ltd, Southampton, UK). At the end of the rehydration period (14 d), seed equilibrium relative humidity (eRH) was checked using a sample of the equilibrating seeds. To induce Fig. 2 Climate at growing site from 1961 to 2020. a Mean annual temperature, the dark grey area is the 95% confidence interval and the dashed line is the regression line, b mean annual precipitation, c temperature difference from the 1961-1990 mean annual temperature, with red bars indicating years warmer than the mean and blue bars indicating years colder than the mean; d precipitation difference from the 1961-1990 mean total annual precipitation, with red bars indicating years drier than the mean and blue bars indicating years wetter than the mean. Data from the climate dataset of the Emilia Romagna Region (https:// arpae prv. datamb. it/ datas et/ erg5-eracl ito) 1 3 ageing, seeds were transferred to a second electrical enclosure box, over a non-saturated solution of LiCl at 60% RH and placed in an oven without light at 45 ± 2 °C (Binder FD53, Binder GmbH, Tuttlingen, Germany). Specifically, 150 seeds for seed lots 2001 and 2005 and 200 seeds for the other seed lots were used. There were fewer seeds used in the 2001 and 2005 accessions due to the limited number available and using only 150 seeds has previously been demonstrated to be an effective method (Davies et al. 2016). For seed lots 2001 and 2005, 25 seeds were removed from the ageing conditions after 2 and 10 days and 50 seeds were removed after 20 and 30 days for germination testing. For the other seed lots, 50 seeds were removed after the same intervals described above (i.e. on days 2, 10, 20, 30). Sampling interval at four time points was found to produce seed survival curves and longevity parameters comparable with the standard protocol (Davies et al. 2016).
To test the viability of the removed seeds, seeds were sown on 1% distilled water agar held in 50 mm diameter Petri dishes. Dishes were placed in a temperature and light controlled incubator (LMS 250A, LMS Ltd, Sevenoaks, UK) at conditions previously found to be optimal for germination (25 °C with addition of 250 mg L −1 GA 3 , Mondoni et al. 2018) and 12 h daily photoperiod (photosynthetically active radiation 40-50 μmol m −2 s −1 ). Plates were checked weekly for germination and seeds were scored as germinated once the radicle had reached approximately 2 mm in length. At the completion of each germination test (4 weeks after sowing), non-germinated seeds were cut-tested to confirm whether there were empty seeds. Non-germinated full seeds were assumed to have died during the ageing treatment, since there were no other constraints to germination and preliminary germination test showed high initial germination (i.e. 96-100%). Conversely, non-germinated empty seeds were excluded from the following analyses. The experiments were carried out between October and December 2020.

Seed mass
The mass of 1000 seeds was determined in ten replicates of 100 seeds. For accessions 2001, 2005 and 2007 there were fewer than 1000 seeds available, so all seeds in these accessions were weighed to the nearest 100 seeds.

Statistical analysis
The germination data from the seed storage (ageing) experiments were analysed by probit analysis, estimating the time for viability to fall to 50% under the storage conditions used (p 50 ), as well as the intercept (K i ) and slope (1/σ) of the seed survival curve, according to the viability equation (Ellis and Roberts 1980): where v is the viability in probits after p days in experimental storage.
The p 50 values were then used in a Spearman's (nonparametric) correlation analysis against the year, seed mass and the climatic variables of annual and growing season mean temperature and total precipitation. The effect of the different variables on p 50 was also explored using stepwise general linear regression modelling and Wald test, weighting the p 50 values by the inverse of the square of the standard error of the estimate of p 50 .
We then explored how well the longevity equation of Probert et al. (2009): predicted the p 50 estimates for our seed lots. In this equation, E has the value 1 for endospermic seeds and the value 0 for non-endospermic seeds; AT is the mean annual temperature; and AR is the total annual precipitation. Lastly, we adapted this equation to allow parameter variables to vary and substituting AT and AR with mean temperature and total precipitation, respectively, during the growing season only. This was done by fixing K i for each seed lot and allowing the slope to vary based on the relationship between p 50 and the climate variables shown in Eq. 2, i.e. fitting the equation: in which c (constant), t (response to temperature) and r (response to precipitation) are the parameters for estimation.
Analysis was performed using Genstat for Windows 20 th edition (VSN International Ltd. 2019). Graphs were created in the R statistical environment version 4.1.3 (R Core Team 2022) and Origin (v 6.1).

Seed longevity over twenty years
Seed longevity at 60% RH and 45 °C varied between accessions (collection year), with the shortest-lived seeds collected in 2014 (p 50 of 7.77 days) and the longestlived collected in 2007 (18.49 days; Table S1). Fitting the observed p 50 values to the predicted values based on climate variables and Eq. 2 (Probert et al. 2009), the observed values showed a similar trend but were consistently lower than predicted, indicating that the seeds of V. alpina are shorter-lived than the global average (Fig. 3).

Seed longevity and climate
The Spearman's correlation coefficients between longevity (p 50 ), seed mass and the various climate variables considered were not significant except in the case of p 50 and precipitation during the growing season (P = 0.015; Table 1). Similarly, the multiple regression analysis indicated that the most important term for explaining p 50 was precipitation during the growing season (F-probability in the accumulated analysis of variance = 0.02 cf. 0.11-0.46 for the other variables). Therefore, seed mass and length of time in seed bank storage did not have a significant impact on seed longevity. Seeds were longer-lived when collected during years when there was higher than average mean temperature and lower than average precipitation (Table 1 and Fig. 4). Seeds collected in 2007 had the greatest seed longevity, but was an outlier when p 50 was plotted against the climatic variables, so it was excluded from the analysis.

years of genebank storage does not affect seed longevity
Contrary to what we assumed, seeds of the older accessions of the alpine species V. alpina were not shorter-lived under the experimental ageing conditions than the most recent accessions. Our data do not account for possible longevity loss over the last 20 years of genebank storage and, if this was the case, the older accessions would have been initially (i.e. at the time of collection) longer-lived than observed. Therefore, we can reject our first hypothesis that due to the negative effect of genebank storage seeds from the older accessions would be shorter-lived under experimental ageing conditions, but the results observed here need to be treated with caution. Our results are however consistent with previous research indicating that seeds of alpine plants are shorter-lived than the global average (Fig. 3 Eq. 2). Results from comparative longevity protocols are expected to be meaningful in understanding seed longevity in genebanks (Hay et al. 2022), but still, little is known about the relationship between longevity after controlled deterioration tests and genebank storage due to the lack of comparative studies between the two environments (but see Gianella et al. 2022). In our study, longevity differences between accessions estimated through experimental ageing conditions did not depend on the timing of genebank storage, suggesting that seed viability of several (short-lived) alpine species can be safety retained for at least 20 years. This is an important and  novel observation, highlighting that alpine species may be suitable for genebank storage and, for the first time, sets a first timeframe for ex situ conservation of these vulnerable species. Although we cannot rely on initial germination data, the overall high initial viability (K i ) observed across accessions (Table S1) indicates that seed viability must have been very high at the time of collection, further supporting that long-term genebank storage is suitable for this species.
The reason for the differing longevities between accessions was due to differences in the initial viability (K i ) at the start of the artificial ageing treatment, the probit rate of viability loss ( −1 ), or both (Table S1). There are various factors that influence these parameters including climatic conditions during seed production, age of seeds, species, genetic factors, and post-harvesting processes (Hay et al. 2019). Considering these seeds are from the same species collected at the same site and month, and have undergone the same post-harvest treatment, the most likely explanation for differences in the p 50 we observed is the climate experienced by the parent plant during each year. Fig. 4 The effect of temperature and precipitation at the population site on the subsequent longevity of seed lots of Viscaria alpina. a Survival curves determined by probit analysis in which estimates of K i were pre-determined (fixed) and the slope was allowed to vary based on temperature and precipitation (Eq. 3). b Relationships between p 50 (log scale) and mean annual and growth season temperature. c Relationships between p 50 (log scale) and total annual and growth season precipitation. d The same relationships as in c, but with precipitation transformed according to Eq. 2 from Probert et al. (2009). Note that in b-d, the p 50 values are the same, plotted against the different temperature or precipitation parameters. The points indicated by the arrows are for the seed lot collected in 2007, which appears to be an outlier. It was therefore not included in the correlation or regression analyses, or in the model fitting shown in a

Higher temperature and lower precipitation increase seed longevity
In the correlation analyses, we tested the length of time in storage, seed mass, and growing season and annual temperatures and precipitations with p 50 . There was a weakly positive correlation between the growing season temperature and seed longevity, and a strongly negative correlation between precipitation and p 50 (Fig. 4, Table 1). This result echoes previous findings that plants from warmer and drier climates produce longer-lived seeds (Probert et al. 2009;Mondoni et al. 2011;Bernareggi et al. 2015). There was a stronger correlation with precipitation and longevity than with temperature and longevity. Studies that had previously used OTCs to raise the surrounding temperature (e.g. Bernareggi et al. 2015) did not take into account that the humidity is also reduced inside the OTC, and the size of the chamber opening can alter the precipitation received by the plant (Marion et al. 1997). Therefore, previous results that a raised temperature increased seed longevity may in fact have been due to reduced water availability. Supporting this view, seeds collected in 2007 were an outlier, with a p 50 1.5 times higher than the average p 50 value. 2007 was the driest year of the 10 accessions tested, and the second driest year on record since 1961, with 30% less precipitation than the average. In the last 20 years, the number of years in which total precipitation was below the average was almost two-fold higher than the years with above average precipitation (i.e. 13 vs. 7, see Fig. 2), indicating that drought conditions have been more dominant in last two decades. Most importantly, considering that annual temperature was always above the average, our results highlight, for the first time, that the net effects of warming on seed longevity will likely depend on water availability.
We looked at climatic parameters during the growing season because this is a particularly critical time for alpine plants: they experience a short growing season that can be affected by various factors, such as temperature, precipitation, snow cover, wind and UV radiation (Körner 1999). Alpine species have evolved over millennia to survive low temperatures, but with climate warming they can be released from this temperature constraint (Gottfried et al. 2012). Therefore, if temperature is no longer a limitation to growth, precipitation during the growing season may become a critical factor for growth and survival. This has been demonstrated in the alpine forb Rumex alpinus where increased summer temperatures have released plants from temperature impediments, but growth has become limited by water availability (Dolezal et al. 2020). From our results, it appears that in alpine habitats, when conditions are wet, seeds of lower quality are produced, as indicated by the reduced longevity. Our result mirrors the global finding that reduced precipitation increases seed longevity (Probert et al. 2009) with this hypothesised to be a survival mechanism allowing seeds to persist in dry periods. However, there are exceptions to this global pattern. For example, some studies have found the opposite, with an increase in seed longevity with higher precipitation in Australian species (Merritt et al. 2014) and under cool and wet experimental conditions (Kochanek et al. 2010). The reasons for the relationship between precipitation and seed longevity are not well understood and need to be investigated.

Conclusion
From our results, it appears that seeds of V. alpina are suitable for long-term genebank storage for much more than 20 years, although still shorter-lived than the global average. Seed longevity has been affected by the climatic changes experienced at the collection site over the last 20 years, with precipitation having a larger impact on longevity than temperature. This may be an important survival strategy for alpine plants where, especially in Mediterranean mountains, precipitation is expected to decrease with climate change. The link between increased precipitation and reduced seed longevity requires further study. Currently, laboratory experiments are underway to understand the molecular basis behind the changes in longevity we observed. Our approach of using temporally separated populations from ex situ genebank storage is an example of the usefulness of stored seeds to provide information about the effects of climate change on seed traits and gives insight into how plants may adapt to a warmer and drier climate. We suggest that the climatic conditions during the growing season, in particular precipitation, should be considered when collecting seeds of alpine species for genebank storage.

Conflicts of interest
The authors have no relevant financial or nonfinancial interests to disclose.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.