How reliable are the ‘banks’ in securing food for humanity and overall biodiversity?

The recent viral pandemic brutally exposed the vulnerability of humankind and its unpreparedness for a catastrophe. A comparable, albeit functionally different, disaster probably awaits us because of our negligence toward ‘Mother Nature.’ The impending climate crisis poses a significant risk to our existence with the glimpses of recent weather upsets such as wildfires, which were once sporadic, and causing global devastation of the natural resources (Brando et al. 2020; Di Virgilio et al. 2020; Goss et al. 2020).

Seeds, which carry the genetic material forward in time, are also under threat while in storage (Cienska and Schneider 1974; Hong and Ellis 1997; León-Lobos et al. 2012). While some seeds show incredibly long natural survival (e.g., date palm, Phoenix dactylifera L., radiocarbon dated to c. 2000 yr old (Sallon et al. 2008)), these are extremely skewed from the general life-expectancies of seeds stored under recommended controlled conditions. Preserving seed for food security and biodiversity is a critical component of any preparedness plan, serving to mitigate the potentially devastating effects of natural disasters and to protect the diversity of crop genetics (Liu et al. 2018; Mascher et al. 2019). There are over 1750 individual gene banks and approximately 2500 botanical gardens worldwide, collectively storing more than 7.4 million accessions with intentional redundancies across repositories (FAO 2010; Colville and Pritchard 2019).

Long-term seed storage

Numerous large-scale seed preservation facilities, such as the Svalbard Global Seed Vault and the Kew Millennium Seedbank, play a crucial role in conserving genetic resources for the future (O’Donnell and Sharrock 2017; Rivière and Müller 2018). Despite their promise of long-term preservation, it is imperative to underscore the necessity for mandatory and regular monitoring of seed health within these facilities to ensure sustainability (Charles 2006; Ulian et al. 2019). The maintenance of ex situ seed banks is acknowledged as resource-intensive, requiring adherence to strict quality control measures established by international forums to optimize the conservation of genetic resources (FAO 2014).

Storage and conservation during catastrophic events

In times of imminent natural disasters, the role of seed preservation extends beyond long-term storage facilities. It becomes crucial to address short-term storage needs and conservation during catastrophic events to secure immediate access to genetic resources. This involves strategic planning to ensure the availability of viable seeds for rapid recovery and regeneration post-disaster. The resilience of such seed banks to external shocks and their rapid response capabilities are paramount for safeguarding agriculture and ecosystems in the face of unpredictable and adverse events.

The dual focus on long-term seed storage in specialized facilities and the preservation of seeds during catastrophic events underscores the multifaceted approach needed to ensure global food security and biodiversity conservation. Both aspects require continuous monitoring, adherence to quality control measures, and strategic planning to address the diverse challenges posed by natural disasters and other unforeseen events.

Potential technical bottlenecks for seedbank tasks

Apart from sustainable conservation of seed and genetic diversity, there are several integral tasks and goals of ex situ seedbanks to circumvent the consequences of natural catastrophes, such as:

  1. 1.

    Conservation of Crop Diversity and Provide Seed Materials for Research

  2. 2.

    Shield from Climate Change and Natural Disasters

  3. 3.

    Quarantine from Diseases

  4. 4.

    Protection from Man-Made Disasters

All the above-mentioned tasks might face technical bottlenecks for the smooth functioning of ex situ seedbanks.

Task 1: conservation of crop diversity and providing seed materials for research

  1. (a)

    Cataloging Challenges: Efficiently cataloging a diverse range of crop species with different characteristics and traits poses a significant technical bottleneck. Standardizing the cataloging process for various crops and crop wild relatives (CWR), each with unique requirements, can be complex and resource intensive (Maxted et al. 2012). Additionally, progressive ex situ seed banks and gene banks should consider a digital catalog linking functional genetic variants in each accession to external genome information (Sackville Hamilton 2020; Engels and Ebert 2021). This approach enables DNA-based decisions for conservation and informed utilization based on gene function and associated phenotypic data.

  2. (b)

    Optimal Seed Preparation: Preparing seeds for optimal storage involves meticulous procedures to ensure the preservation of genetic material. However, variations in seed size, morphology, and germination requirements across different crops make standardization challenging (Liu et al. 2020a, b).

Task 2: shielding from climate change and natural disasters

  1. (a)

    Storage Conditions Maintenance: Maintaining optimal storage conditions, such as temperature and humidity, is crucial for preserving seeds. Technical challenges arise in creating and sustaining these conditions, especially in the face of unpredictable climate changes that may require continuous adjustments (Li and Pritchard 2009; Hay et al. 2021).

  2. (b)

    Periodic Viability Testing: Ensuring seed viability over time is essential. However, conducting periodic viability tests for a vast collection becomes a technical bottleneck due to the resource-intensive nature of these assessments and the diverse physiological aging trajectories of different species (Hamilton 1994; Breman et al. 2021).

Task 3: quarantine from diseases

  1. (a)

    Disease Detection and Prevention: Detecting and preventing diseases in stored seeds demands advanced techniques and technologies. However, the diversity of pathogens and the need for precision in quarantine procedures make the implementation of effective disease management strategies challenging (Ogwu and Osawaru 2023).

Task 4: protection from man-made disasters

  1. (a)

    Security and Safety Measures: Implementing robust security measures to protect seeds from man-made disasters, such as theft or accidental damage, poses technical challenges. This includes developing advanced security systems that balance accessibility for research purposes with stringent protection requirements.

  2. (b)

    Data Security for Cataloged Information: Safeguarding cataloged information is critical. Ensuring data security and preventing unauthorized access or data loss can be technically challenging, especially when handling extensive databases of diverse seed collections (Breman et al. 2021).

In summary, technical bottlenecks in ex situ seedbanks arise from the inherent complexities of tasks such as viability assessment frequencies and the mode of test procedures.

Challenges in seed viability assessment and conservation strategies

The assessment of seed viability is essential for determining their usability, encompassing factors such as time, materials, qualified technicians, reproducibility, efficiency, and associated hazards, which vary across different test methods (Mohammed et al. 2019; Franks et al. 2019; Porteous et al. 2019; Fig. 1). Seed viability, influenced by physiochemical and biological factors, results in distinct safe storage periods even under optimal conditions for different species (Richards et al. 2015; Nagel et al. 2019; Shaban and Others 2013).

Fig. 1
figure 1

Comparison between traditional destructive- and alternative non-destructive seed viability tests. Challenges in traditional seed bank management encompass seed loss during viability tests, substantial resource allocation (such as employing qualified personnel, maintaining germination conditions over an extended period, and conducting dormancy-release procedures), and the unreliability in interpreting test results. The classifications elucidated in the legend are qualitative and deduced based on the standard operating procedures used to conduct the respective tests. Considering the interconnected nature of cost, duration, and resource allocation, the model took into account both the material and running costs per standard test batch, as well as the quality and quantity of skilled resources needed. Reliability is assessed based on the reproducibility of results and the clarity of decision-making in a test. The hazard status of a test is determined by the nature of instruments and chemicals used, potentially compromising standard laboratory safety measures that must be upheld. While multiple non-destructive seed viability tests present solutions, they may come with increased costs, require advanced instrumentation, or involve occasional safety considerations. The visual representation underscores the positive influence of adopting innovative approaches in seed bank management to ensure sustainable genetic diversity preservation. Consequently, an optimal seed bank management method should ensure consistent accessibility, facilitating easy testing in both small-scale seed repositories and large-scale seed vaults, while maintaining comparable levels of reliability and sustainability

Full size image

To ensure the effectiveness of seed conservation programs, it is imperative to determine the viability of stored seeds periodically and accurately (Hay and Whitehouse 2017; Mohammed et al. 2019). However, determining the frequency of such tests poses a challenge due to the diverse physiological aging trajectories of seeds from various species (Redden and Partington 2019; Colville and Pritchard 2019). Consequently, a general scheduling of seed viability assessments for a diverse ex situ collection may lead to the late detection of rapidly aging collections, reaching a point of no return.

While the International Seed Testing Association (ISTA) has established clear and standardized protocols for viability tests, including ancillary tests such as the tetrazolium test and embryo cut tests, for major crops and commercial seeds (ISTA 2017), these methods may pose challenges for seedbank management due to their destructive and resource-intensive nature. Additionally, interpretations of results from ancillary methods may be ambiguous. Furthermore, the utilization of seeds in these tests may contribute to a void in genetic diversity. These challenges underscore the need for innovative and non-destructive viability assessment methods that can be efficiently applied in seedbank management.

Drawbacks of traditional destructive seed viability test methods

Germination tests are excellent indicators of seed viability, but they cannot distinguish between dormant and dead seeds. In optimal conditions, both dormant and dead seeds will not germinate unless their dormancy status is identified, and appropriate pre-treatment is applied (Baskin and Baskin 2014). Furthermore, germination tests are lengthy (Sautu et al. 2006), with risks of microbial infections, secondary dormancy (Meyer et al. 2016), and require at least 400 seeds for a statistically robust conclusion. Despite its simplicity, germination tests have certain limitations for its universal applicability, especially for the purpose of large-scale seedbank management. The ancillary tests are either technically demanding, or ambiguous to interpret (Heit 1955; França-Neto and Krzyzanowski 2019; Salazar Mercado et al. 2020), respectively, therefore not universally accepted.

Promising scopes of few non-destructive seed viability tests

With an exponential increase in the number of ex situ conservation of accessions, it is now necessary to optimize management procedures and effective resource investment (Hay and Whitehouse 2017). The goals for ex situ collection should be i) fast and reliable viability assessment, ii) retaining the tested samples (non-destructive), iii) analyzing the physiological age of seed lots, and iv) predicting the seed longevity with regular seed health testing intervals.

Lipid thermal fingerprinting

Lipid thermal fingerprinting is a technique used to analyze the thermal behavior of lipids present in seeds. This method helps identify and characterize lipids, providing insights into the stability of seeds during storage. The process involves the use of techniques such as differential scanning calorimetry (DSC) to study the heat absorption or release associated with lipid transitions. Specifically, DSC measures the heat flow associated with physical and chemical changes in a sample as a function of temperature. As the lipids undergo phase transitions (e.g., melting) at specific temperatures, characteristic peaks or changes in the heat flow curve reveal the lipid thermal fingerprints. This non-destructive test is used as biophysical markers to predict potentially poor performing oily seeds during long-term storage. Differential scanning calorimetry (DSC) of 20 crop wild relatives of Brassicaceae could determine the storage stability of the seeds (Mira et al. 2019).

The infrastructure requirement includes high-quality DSC instruments capable of accurately measuring heat flow changes, controlled temperature environment for precise experiments, and software for analyzing DSC data, identifying peaks, and extracting relevant information about lipid transitions.

Advantages include non-destructiveness and lipid characterization, but challenges include infrastructure needs, limited applicability, and data analysis complexities, warranting extensive validation.

Near infrared (NIR) spectroscopy and hyperspectral/multispectral imaging (NIR-HSI)

NIR spectroscopy involves the interaction of near-infrared light with the molecular vibrations of a material. Different compounds absorb light at specific wavelengths, creating a unique spectral signature. A NIR spectrometer emits a broad spectrum of near-infrared light onto a sample. The reflected or transmitted light is then analyzed to determine the absorption pattern, providing information about the chemical composition of the sample. Unlike traditional photography that captures three color bands (red, green, and blue), hyperspectral imaging (HIS) captures a large number of contiguous narrow bands across the electromagnetic spectrum, including NIR. Therefore, HSI systems capture detailed spectral information for each pixel in an image, providing a wealth of data about the material's chemical composition.

The application of NIR and NIR-HSI in seed phenotyping, quality monitoring, and age prediction requires very high technical and computational resources and uses machine learning and artificial intelligence for seed viability prediction (ElMasry et al. 2019; Yasmin et al. 2019; Venkatesan et al. 2020). Specific infrastructure requirements for NIR and NIR-HIS includes NIR spectrometers, hyperspectral/multispectral imaging camera or sensor, and dedicated image and data processing and analysis software. The monochromatic image/s of infra-red or continuous wavelengths are integrated to project a spatial distribution of relevant compounds influencing seed viability (Ma et al. 2020).

Detailed chemical insights and high-resolution data in seed viability testing are advantages of NIR spectroscopy and NIR-HIS. Hyperspectral imaging provides spatial compound distribution. However, these methods are resource-intensive, demanding high technical and computational resources. Machine learning dependence and specific infrastructure requirements, including NIR spectrometers and imaging systems, pose challenges.

Raman spectroscopy (RS)

Raman spectroscopy is a non-destructive analytical technique used for studying vibrational, rotational, and other low-frequency modes in a system. It provides detailed information about the molecular composition of a sample by measuring the inelastic scattering of monochromatic light characterized by Raman shift and spectra. By analyzing the Raman spectra, one can identify specific vibrational frequencies corresponding to chemical bonds and molecular structures, which is used to generate fingerprints of chemical compositions inside stored seeds. While Fourier transform near-infrared (FT-NIR) could successfully distinguish between heat-killed and viable corn seeds, RS results heavily depended on the mode of spectral data analysis (Ambrose et al. 2016). The different application areas underscore the technical bottleneck for such methodologies to apply in large-scale seedbank viability assessments successfully.

Specific infrastructure requirements for RS includes high-quality Raman spectrometer with appropriate laser sources and detectors to capture Raman-scattered light, monochromatic laser source with the appropriate wavelength for Raman excitation, software for processing and analyzing Raman spectra.

Raman spectroscopy (RS) provides detailed, non-destructive insights into seed molecular composition, aiding chemical analysis. However, RS results depend on spectral data analysis and face technical bottlenecks for large-scale seedbank viability assessments. Infrastructure requirements include a high-quality Raman spectrometer, suitable laser sources, and specialized software.

Infrared thermography (IRT)

Infrared thermography is a non-destructive imaging technique that captures the infrared radiation emitted by objects. It measures the surface temperature of an object and creates a thermogram, which is a visual representation of temperature variations. In the context of seed viability assessment, the working principle involves three aspects: 1) Infrared Radiation Emission: All objects with a temperature above absolute zero emit infrared radiation. The amount and wavelength distribution of this radiation depend on the object's temperature. 2) Thermal Imaging: Infrared cameras detect and convert the emitted infrared radiation into an electrical signal. The resulting thermal image displays temperature variations across the object's surface. 3) Viability Inference: Viable and non-viable seeds may exhibit different thermal profiles due to variations in metabolic activity, water content, and other factors. Integration of thermal and optical imaging to infer the viability of seeds was used to differentiate between the viable and non-viable seeds of several species including garden pea (Pisum sativum), wheat (Triticum aestivum), lettuce (Lactuca sativa), and rape (Brassica napus) (Kranner et al. 2010; Men et al. 2017; Fleming et al. 2019; Liu et al. 2020a, b).

Specific infrastructure requirements for IRT includes high-resolution and high-frame-rate infrared cameras with sensitivity in the infrared spectrum and capable of capturing temperature variations, concurrent optical imaging (visible or near-infrared) to capture additional information about seed characteristics, and algorithms for extracting temperature information and correlating it with seed viability.

Infrared thermography (IRT) offers non-destructive assessment and temperature visualization, aiding viability inference in seeds. Therefore, it is possible to test the same seed sample multiple times for its viability and be restored back to the ex situ seedbank. However, its accuracy may be influenced by external factors, and the integration of concurrent optical imaging adds complexity. Infrastructure requirements include high-resolution infrared cameras, optical imaging equipment, and algorithms for temperature extraction.

X-ray computed tomography (CT)

X-ray computed tomography (CT) is a non-destructive imaging technique that uses X-rays to create cross-sectional images or 'slices' of objects. In the context of seed analysis, X-ray CT is utilized to reveal the internal structures of seeds, providing detailed information about their morphology and quality.

X-ray imaging was adopted to reveal the internal quality of the seed to reveal an empty, broken, or irregularly shaped endosperm with 98% accuracy in muskmelon (Ahmed et al. 2018). Seed viability prediction by X-ray of eight wild Saudi Arabian species was comparable to the germination tests (Al-Turki and Baskin 2017). However, the potential impact of the radiation exposure to the seeds and stochastic X-ray damage is not thoroughly studied, therefore cannot be excluded (Lang et al. 2014).

Specific infrastructure requirements for CT includes high-resolution X-ray CT scanner capable of producing detailed images of seeds, adjustable settings to control X-ray intensity and imaging parameters, storage infrastructure for the large CT image datasets, and image processing software for reconstruction and analysis of CT images.

X-ray CT offers non-destructive imaging, revealing internal seed structures with 98% accuracy in assessing seed quality. This test method is regularly employed in Millenium Seedbank to determine the proportion of full, potentially viable seeds (Breman et al. 2021) and can be used in multiple cycles of re-testing the same seed batch for viability loss kinetic studies. However, concerns about potential radiation exposure and stochastic X-ray damage to seeds persist (Okuda et al. 1998). Infrastructure requirements include a high-resolution CT scanner, adjustable settings, storage, and image processing software.

Redox-based resazurin test

The redox-based resazurin test serves as a bioassay for assessing seed deterioration by evaluating the decline in seed-coat integrity, ranging from healthy to aged to damaged or dead seeds. This test involves the conversion of Resazurin to Resorufin (pink) and subsequently to dihydroresorufin (colorless), reflecting the extent of leachates from seeds over a 3–5 h co-incubation period (Chen et al. 2018).

The redox-based resazurin test reveals crucial insights into seed viability with key characteristics. This includes its ability to determine the kinetics of seed aging, as demonstrated in both naturally and artificially aged Brassicaceae seeds (Mohammed et al. 2019; Bhattacharya et al. 2020). Moreover, the test exhibits minimal potential for physiological damage or accelerated aging during repeated cycles, involving a 5-h test, desiccation (to 15% Seed Moisture Content), and preservation (at − 20 °C for 1 month), particularly noted in several Brassicaceae species (Bhattacharya et al. 2020).

Regarding repeated testing, viable seeds identified through the resazurin test can safely return to seed storage, contributing to seed bank reserves. Studies on Brassicaceae species indicate that the potential effects of repeated resazurin testing, encompassing multiple cycles, on seed viability, quality, and future assay results are minimal. This suggests that seeds can undergo testing, desiccation, and preservation without significant negative consequences (Bhattacharya et al. 2020).

The benefits of the redox-based resazurin test extend to its straightforward data interpretation, making it particularly advantageous for large-scale seedbank management. The visual color changes from blue to pink and then colorless provides a simple and easily interpretable indication of seed viability. Disadvantages include limited validation across plant species, potentially restricting applicability, and dependency on subjective color changes for interpretation, requiring careful standardization.

Forecasting seed aging and scheduling viability tests

While seed conservation’s ‘holy grail’ remains the pursuit of a non-destructive method for estimating viability, unlocking the added predictive potential of such tests has the potential to revolutionize traditional management practices (Fig. 2). However, simple extrapolation of artificial seed-aging may differ from the actual kinetics of seed deterioration in long-term storage, hence predicting an accurate seed aging trajectory for several species is a daunting task (Buitink et al. 2000; Fu et al. 2015; Mohammed et al. 2019; Ballesteros et al. 2020). Therefore, one of the largest seedbank in the world (Svalbard Global seed vault, Norway, nicknamed as Doomsday vault), started a 100-year-long experiment to monitor the actual longevity of deposited seed by evaluating an aliquot of 400 seeds once in a decade until 2120 [see: https://www.nordgen.org/en/svalbard-global-seed-vault-commences-seed-experiment-that-will-last-for-100-years]. Moreover, the Millennium Seedbank routinely monitors the viability of all collections (now c. 40,000 species) since its inception and accumulates real-time longevity data for thousands of species.

Fig. 2
figure 2

Integrated Seedbank Management using Non-destructive and Predictive Tests. The figure highlights the vital role of non-destructive and predictive seed viability tests in safeguarding seedbank samples. Traditional germination tests are time-consuming and deplete tested seeds. Non-destructive tests (e.g., IRT, X-ray-CT, redox-based resazurin test), which differentiate healthy from damaged seeds and predict aging kinetics, offer an alternative without seed loss. Viability assays, based on the relative slope of the viability loss curve (for slow, moderate, and fast-aging seeds derived from their relative germination %), determine the frequency of future tests, saving significant resources compared to regular seed viability assessments. Considering the intricacies of germination decline and variations in aging kinetics among species, typically characterized by sigmoidal patterns, our analysis focused on the rapid change in slope and the inflection point of the saturating hyperbola. This focus was chosen for its relevance to the 'actionable' and simplified linear phase in seed aging kinetics. An optimized redox-based resazurin test acts as the holy grail in seedbank management, distinguishing between healthy and damaged seeds through a color gradient and predicting aging kinetics in three Brassicaceae species. Thanks to its non-destructive nature, identified healthy seeds can be returned to the seedbank without viability loss until the next test, allowing for repeated testing. This approach serves as a crucial component in the holy grail of seedbank management, enabling the recurring replenishment of seed collections while seamlessly integrating into seed biology research, where a non-destructive viability test is essential. These sustainable tests, meeting all viability criteria, practically ensure the enduring longevity of stored seeds by optimizing resource use for regular assessments

Full size image

In the given context, employing predictive tests based on viability loss (Takeya et al. 2013) or analyzing multispectral seed phenotype data (Hansen et al. 2016) provides a viable solution for systematically scheduling the regeneration of selected accessions. These methods also allow the seeds to be restored back to the storage conditions for multiple cycles without the loss of viability. The predictive approach of the redox-based resazurin test, supported by studies by Mohammed et al. (2019), and Bhattacharya et al. (2020), offers a means to anticipate and address the diverse collection and replenishment needs, effectively mitigating the impact of subsequent depletion of stored samples.

The implementation of this updated predictive exercise represents a significant advancement in seedbank management practices, contributing to the efficient utilization of resources. By strategically planning viability tests based on predictive outcomes, seedbanks can optimize their operational efficiency, addressing challenges associated with traditional management approaches. This not only streamlines resource allocation but also enhances the sustainability of seedbanks, ensuring their long-term viability in the face of depletion and collection demands.

However, it is essential to acknowledge that, currently, many non-destructive tests, including the redox-based resazurin test, have primarily been validated for a limited range of species. To unlock the full potential of these predictive tools for both crop and wild species seedbanks, concerted efforts must be made to expand the scope of validation across a broader range of plant species. This necessitates substantial investments in time and effort to validate the applicability and accuracy of such tests for a more extensive array of species within the conservation seedbank. The utility of predictive and non-destructive tests can only be fully realized through extensive validation across a diverse spectrum of species and is crucial for their widespread adoption in conservation seedbanks, ensuring effective and sustainable preservation of both crop and wild plant genetic resources. However, with the assistance of customized automation and predictive modeling through machine learning and artificial intelligence, the forecasting power and accuracy of non-destructive seed viability test methods can be greatly enhanced, ensuring meaningful data acquisition and validation.

Conclusion

Seedbanks, biodiversity conservation centers, and global seed vaults play a crucial role as the ‘backup of backups,’ safeguarding our agricultural legacy and genetic diversity. However, we face the challenge of ensuring the viability of stored seeds without depleting the precious stock. To overcome the predicament of obligatory viability assessment of the stored seeds, we must embrace non-destructive viability test methods, allowing us to assess seed viability while preserving the samples intact. Simple, reliable, and predictive viability test methods, such as infrared thermography (IRT), lipid thermal fingerprinting and redox-based resazurin tests, can safeguard our future food sources. Extensive and inclusive efforts from the scientific community, stakeholders, and policymakers to welcome these new advancements and rigorous testing in diverse species can guarantee the eternity of the ‘banks’ that safeguard our future food resources.