Keywords

1 The Challenge of Ex Situ Collections: Maintaining Wildness in Captivity

North America’s rich flora has the potential to contribute genes to make our crops more resilient to disease, pests, and weather extremes. The unique features of many North American taxa also offer the opportunity to find sources for new products and better nutrition or healthier, more sustainable landscapes. Many of the plants described in this book are congeneric with plants commonly used in agriculture. For these, breeders will select the genes of interest from the wild relative and assimilate them into modern cultivars, leaving behind undesirable genes that contribute to a weedy phenotype. The breeder’s job is facilitated by low barriers to interspecific hybridization incumbent with wild species that are closely related to domesticated species (i.e., crop wild relatives, CWR). Similar concepts apply to other wild species that may offer new products or have greater resilience in a changing landscape (Urban, 2015). In these cases, modern-day domestication efforts may be invoked and provide an important reminder of the reservoir of services that wild plants provide, bringing new opportunities for economic growth, ecological sustainability, or aesthetic sensibility.

It makes sense to collect North American genetic resources in genebanks and make them available to agriculture, conservation, engineering, and scientific disciplines that explore biological diversity and the environment. We envision collections of genetic resources as an inventory of “nature’s solutions” to tough environmental problems. Genebank users will sift through these collections for patterns of diversity or traits of interest. Therefore, genebanks must be careful not to “edit” nature’s work in case it masks or removes the very trait that was sought. This creates a real challenge because the highly controlled conditions that genebanks must use can create strong pressure for plastic living systems to adapt and become domesticated. The signature of domestication is particularly recognizable in seed traits (Fig. 10.1) (Schoen and Brown 2001; Meyer et al. 2012). In essence, the crop genebank mission is to maintain “wildness” of the stored germplasm so that genebank users have full responsibility for domestication efforts.

Fig. 10.1
figure 1

Seed of wild progenitors (left) and crops (right) showing domestication traits of lighter color and larger seeds. Top to bottom are Pistacia, Coffea, Glycine, Hordeum, Zizania, and Sorghum. Photo taken by LM Hill

Full size image

The purpose of this chapter is to describe some of the challenges one might experience when capturing and maintaining diversity inherent in collections of CWR – whether originating in North America or elsewhere. While we appreciate that plant (and animal) genebanks are often rationalized in an ethical context of conserving natural diversity (e.g., Soulé 1991; Guerrant et al. 2004, 2014), our premise in this chapter is that all the goals for genebanks – economic sustainability, environmental services, ethical considerations, opportunity for new applications, and aesthetic potential – are realized through scientific approaches. Hence, we view plant genebanks as scientific collections and that the scientists who use these collections need access to well-characterized, high-quality materials that are quality-assured so that the genebanking experience doesn’t affect experimental outcomes (ISBER 2012; Walters et al. 2008; Guerrant et al. 2004, 2014). Overall requirements to gather and document materials that are fit-for-purpose for studies of biological diversity are summarized in Table 10.1.

Table 10.1 Requirements to ensure scientific collections are fit-for-purpose in studies of biological diversity
Full size table

Elements in Table 10.1 interact to contribute to the success (or failure) of genebanks in delivering samples of interest to users. Our focus in this chapter is element one and its interaction with element five. Providing viable samples (i.e., germplasm) differentiates a genebank from other types of plant collections such as DNA banks or herbaria. Arguably, ensuring viability while preventing genetic change is one of the greatest challenges facing genebanks today. Tools to validate how well a sample represents the source population and how it is maintained through time in the genebank are rapidly developing (Kilian and Graner 2012).

Plant genebanks can serve as an important tool for conservation. Ex situ conservation, made possible by genebanks, can complement in situ strategies that operate at habitat and landscape levels (Soulé 1991). Safely preserved at locations protected from social pressures or environmental disasters, genebanks can amass significant genetic diversity from a great range of taxa in a relatively small volume. Numerous land managers and conservation groups within the USA incorporate seed banking to forestall attrition of plant genetic diversity and ensure excellent sources of germplasm for land restoration (Guerrant et al. 2004; Hay and Probert 2013; Haidet and Olwell 2015; PCA 2015). Collections of plants of conservation concern (sensu USFS https://www.fs.usda.gov/Internet/FSE_DOCUMENTS/stelprd3848211.pdf (visited October 3, 2017)) exist already. These can be especially valuable as a source of agronomic traits (Khoury et al. 2013). For example, the highly endangered plant Zizania texana Hitchc., which grows in a 7 km stretch of the San Marcos River near San Antonio, TX, has a desired perennial growth habit and produces seeds with exceptionally high lysine (Kahler et al. 2014).

Germplasm may be maintained in genebanks as samples actively growing, under field greenhouse or tissue culture conditions, or as alive-but-not-growing samples maintained by highly controlled conditions. Collections of the former are often referred to as living, and the latter can be referred to as stored, suspended (for suspended animation), or preserved. Often stored germplasm and seed banks are synonymous, although technologies have rapidly advanced to preserve many other germplasm forms in addition to seeds (see next section). Living collections allow curators to observe traits and regenerate samples, but they also increase the risk of losing samples to inclement weather, pests, pathogens, social unrest, and old age. Genetic erosion through drift, inadvertent selection, or introgression with neighboring related plants can also occur while growing or regenerating a sample. And, regeneration is especially expensive in terms of land and labor for large plants that may take years to sexually mature. Risks to field collections can be partially mitigated by maintaining plants in vitro, but labor and suitable space to maintain these collections can be cost-prohibitive (Pence 2011). Preserved collections are less expensive and more space efficient, carry lower risk from natural or anthropogenic disasters, and make genetic resources readily available regardless of season, year, or location (Li and Pritchard 2009; Volk et al. 2009; Pence 2011). Moreover, risks of genetic erosion during regeneration are mitigated when storage conditions are exceptional and maintain high viability with no mortality over extended periods (Richards et al. 2010; Walters et al. 2015a).

Plant genebanks are proliferating worldwide, and currently, about 1750 exist to serve agriculture, conservation, and studies of ecology, evolution, and diversity (Hay and Probert 2013; FAO 2014). These germplasm collections focus on a wide array of plant genetic resources and usually invoke a combination of living and stored approaches to maintain and evaluate samples. Major questions challenging genebank operations include forms of germplasm that can be preserved, propagated, and utilized, indicators for when a collection is “complete,” life expectancy during storage, and assessment of genetic quality and potential uses of genebanked samples.

2 The “Conservation Target,” Germplasm that Is “Fit-For-Purpose,” and Genebank Management Plan

Germplasm samples must align with genebank mission, which is defined by the genebank’s specific objectives or rationale. For example, the purpose of the USDA National Plant Germplasm System (NPGS) is to provide diversity that benefits research and education about agriculture. For this reason, NPGS collections focus on the subset of the world’s approximately 300,000 plant species that have economic potential. Currently the NPGS collection contains only about 16,000 species, but this is represented by nearly 600,000 accessions (an accession is a sample with unique identifying information, such as taxon, location, and harvest details (this information is also called passport data)) – one of the world’s largest plant germplasm collections globally (https://npgsweb.ars-grin.gov/gringlobal/query/summary.aspx visited October 3, 2017). NPGS accessions are roughly divided into named cultivars (50%), genetic stocks (20%), and wild relatives or landraces of crops (30%). About 250,000 accessions are distributed each year to users. This strongly suggests that interest in genebank collections among the scientific community resides in questions at the sub-taxonomic level.

Twenty five years ago, conservation targets for genebanks hovered at the taxonomic level (Soulé 1991). For example, botanical gardens used the living genebank strategy and broadly collected species but just had a few exemplars for each. This strategy provides support for phylogenetic distinctions but is unlikely to reveal variation within a species (Marshall and Brown 1975; Hokanson et al. 1998; Lawrence et al. 1995). Proliferation of genebanks that take advantage of advancing storage technologies and data management offer the opportunity for collections to explore a finer scale of genetic variation (Charlesworth et al. 2001; Lockwood et al. 2007a, b; Franks et al. 2008; Walters et al. 2008; Engelmann 2011). Conservation targets at these finer scales include populations, ecotypes, families (e.g., maternal lines), and individuals with exceptional characteristics, traits, or even particular gene expression patterns (Khoury et al. (2015). At the writing of this chapter, we feel that the conservation target(s) for CWR is/are mostly undefined for most crop collections. Conservation targets might range from samples that provide phylogenetic representation to samples that confer particular traits or ecotypes (e.g., drought tolerance). In many instances, collections of CWR are sought to provide a general representation of population diversity of the species, as a contingency against outbreak of disease or pests. When the conservation target is defined below the taxonomic level, stringency for maintaining genetic identity of the sample tightens (Table 10.1, element 1) and requires metrics to demonstrate the proficiency by which a genebank delivers samples that reflect the finer-scaled conservation target (Van de Wouw et al. 2010).

2.1 Germplasm

What part of the plant should be sampled for genebanking purposes? For DNA analyses, a fresh leaf or other non-senescent tissue is usually sufficient (Walters and Hanner 2006). However, for genebanks with the additional responsibility of providing live material (Table 10.1, element 5), the choice of propagule is a critical decision (Volk and Walters 2004). Within the genebank, ability to provide samples of high structural integrity that faithfully represent the conservation target usually rests on whether the material can be stored and easily distributed. Costs of processing and storage should figure significantly into the genebank’s business model to determine the volume of material that can be managed effectively. Additional criteria for the collector might be timing of the collection trip in relation to plant phenology, remoteness of populations, permitting allowances, amenability to harvest, impact to the population and site, and potential for opportunistic collection of other species. Users may have additional preferences for the ease and required time for growing or propagating the material as well as the immediate availability of germplasm. Luckily, many plants are fairly plastic in their reproductive behavior and offer numerous propagule types to meet a range of requirements, preferences, and constraints (Table 10.2). Genebanks frequently distinguish between propagules that are sexually-derived (i.e., seeds and pollen) and those that arise from vegetative cuttings (i.e., clonally propagated). In agriculture, this distinction usually occurs because the conservation target is a specific genotype and the plant is highly heterozygous and outcrossing, for example, fruit crops (Volk and Walters 2004). Clonal propagation may be necessary for plants of conservation concern if there is reproductive failure in the wild (e.g., inbreeding, no pollinators) or if population sizes are inviable (Pence 2013). Since the conservation target of CWR is usually at either the population or gene level, stringent control of the genotype may be unnecessary and may actually impede broader representation of diversity within the population or incorporation of useful genes into a cultivar (Volk and Walters 2004).

Table 10.2 Some common propagules used in plant genebanks
Full size table

Seeds are the most commonly used propagule for plant genebanks. Usually compact, plentiful, storable, growable, and representative of maternal and pollen-donor lines, seeds might just be the ideal medium for plant genebanking. Indeed, over 95% of the USDA National Plant Germplasm System collection uses seeds as the propagule form of choice. Seed-related traits such as fertility, fecundity, uniformity, germination speed, harvesting ability, and longevity – traits that facilitate genebanking – reflect traits selected during domestication (Meyer et al. 2012) (Fig. 10.1). We should expect disparities in these seed traits between untamed wild progenitors and their derivative modern cultivars. Consequently, we should also expect contrasting response to genebanking from domesticated and wild-collected germplasm. Moreover, we can expect wild-collected germplasm to be more prone to genetic erosion arising from the highly artificial conditions implicit within a genebank. Anecdotal accounts of greater difficulties genebanking seeds from the wild are increasing (Hay and Probert 2013; Walters 2015a). Despite these challenges, conservation groups and land managers have demonstrated the feasibility of genebanking wild seeds and the utility of this germplasm in restoration work (Maschinski and Haskins 2012; Guerrant et al. 2004; Haidet and Olwell 2015). As collaborators in these efforts, we have gained experience working with the seeds of truly wild species and can describe some of the pitfalls we’ve encountered that reduced the efficiency and accountability of genebanking efforts. These are not insurmountable problems; they simply indicate the need for adjustments in procedures, expectations, and anticipated costs for properly genebanking populations from natural populations compared to established methods using crop seeds.

Depending on several factors, seeds may be a less preferred germplasm form for sampling CWR in natural populations. Some plant species produce seeds that are less suited for genebanking because viability is lost quickly when standard genebanking conditions (sensu FAO 2014) are used. Seeds exhibiting low survival under standard genebanking conditions are nominally classified as “recalcitrant” or “intermediate” (and collectively as “non-orthodox”) compared to counterparts that are considered “orthodox” (Walters 2015b; RBG 2017). In interspecies hybridization zones, cuttings may provide more certain taxonomic identification than progeny from uncertain pollen sources. At sites where there are few individuals or low fecundity, cuttings may provide a means to collect germplasm with lower potential impact to the natural population; these can then be grown-out in field collections to facilitate characterization and regeneration through seeds. Pollen is under-appreciated as a germplasm form in plants, which contrasts with animal genebanks in which semen, the counterpart to pollen, is the most commonly used germplasm form (Mazur et al. 2008). Pollen might be an effective alternative germplasm form that can capture genes of interest and deliver them to a breeding population when seeds are unavailable or have poor storage characteristics or when maintaining cuttings is cost-prohibitive. For example, pollen from oak trees is desiccation tolerant, while oak seeds tend to be recalcitrant (Franchi et al. 2011). Pollen is storable (Hoekstra 1995; Volk 2011), but it lacks the longevity traits exhibited in seeds of the most common agronomic species (Dafni and Firmage 2000). The requirement for rapid processing of pollen samples makes it a less-ideal germplasm form when collecting from remote natural populations.

Quality of seed set and phenology are also important factors; a plant collector should sample seeds in the fruiting period, vegetative tissues when plants show active growth flushes or cold-adapted twigs are available, and pollen if plants are flowering. Alternative germplasm forms, such as pollen or cuttings, may augment genetic diversity lost by high mortality during banking of non-orthodox seeds.

2.2 Sampling Strategies and Management Plans.

In addition to the type of germplasm collected, the conservation target also defines the sampling strategy (Guerrant et al. 2014; Hoban et al. Chap. 8, this volume) as well as the genebank management plan. For germplasm banks, management plans must be suitable to deliver viable germplasm (Table 10.1, element 5); however management plans can vary depending on the conservation target. When the conservation target is an exemplar of phylogenetic representation, sampling probably occurred at one or a few convenient locations, and there probably wasn’t great effort expended to get an accurate genetic representation of the species or particular populations. In this case, management at the genebank should complement the sampling effort to ensure sufficient viability for representatives of the taxon. Conservation targets for agricultural-based genebanks are usually at the sub-taxonomic level, for example, diversity is sought for a specific trait (e.g., aluminum tolerance, salt tolerance, and disease resistance) or for broad population representation needed for contingencies in the future. For conservation targets at trait or population levels, sampling usually occurred across diverse locations, and the sites and number of sites were selected carefully to maximize the sought diversity with fewest possible samples (so as to not overwhelm genebanking operations). Stringent genebank management plans are needed to ensure that the sample remains genetically representative of the source population. These are discussed further in Sect. 10.4 (Standards and Best Practices).

An important question arises about collection completeness, “completeness” being defined as how well the samples in the collection represent the diversity within the conservation target. There are few specific metrics to determine completeness (though see Hoban et al. Chap. 8, this volume for fuller discussion). It is important to note that metrics will differ among collections that are focused on different conservation targets such as a species representative, a specific trait or a contingency collection. International policy and legislation (e.g., International Treaty on Plant Genetic Resources for Food and Agriculture (ITPGRFA-FAO 2001), the Convention on Biological Diversity Strategic Plan (SCBD 2010), and the Global Strategy for Plant Conservation (SCBD 2014)) encourage strategies for ex situ conservation of CWR but provide few recommendations for the conservation target or metrics for effectiveness of sampling or management plans. Using species distribution models along with validating genetic data (Hoban et al. Chap. 8, this volume) may provide collectors more sophisticated tools to locate and monitor genetic variation and estimate uncertainty about collection “completeness.”

3 Genebanking Wild-Collected Germplasm

Maintaining germplasm collections within allotted resources is a top priority for most genebanks. As mentioned earlier, living genebanks are limited in the amount of genetic diversity they can represent, and living germplasm is more vulnerable to stressful conditions. Preserved germplasm is maintained under highly controlled conditions. There is an expectation that it will remain viable into the future, but that duration is usually poorly defined. Often genebanking duration is defined as short-, medium-, or long-term, which most commonly define the storage conditions (ambient, refrigerated, and freezer, respectively) (FAO 2014), rather than the needed longevity, such as over a breeder’s career, until an imminent restoration project is implemented, or forever. At NLGRP, we target a 100-year lifespan for most of our seed accessions (Walters et al. 2004).

The apparent stasis imposed by preserving germplasm often lures the naïve into a perception of simplicity; however the complexity of the effort is revealed by considering the timescale in which genebanks operate – usually decades. It is not generally appreciated that the impacts of seemingly minor deviations today won’t be evident until sometime in the future; hence, an unsuspecting genebank manager may inherit a “ticking time-bomb.” Genebank failures can go unexplained without standardized methods or stringent documentation. Herein lies a paradox: how do genebanks use standardized treatments for diverse materials and not encounter highly variable responses? And, do the variable responses impact how well the sample can represent the conservation target? A better understanding of time-scales is required to address these questions.

3.1 Stopping the Clock: A Primer in Preservation Technologies.

Understanding how to stabilize biological materials and predict the effects of time is a highly practical science needed for everyday problems. The food industry needs to provide expiration dates for product quality. Effectiveness of drugs and dosage response must consider the variable conditions that occur in household medicine cabinets. Plastic products, coverings, and packaging lose form and function over time. Everyone experiences the yellowing of old paper, the brittleness of aged rubber bands, and the failure of worn tape to hold documents together. Like all materials, the fundamental process for structural stability (i.e., preservation) is solidification (Menard 2008; Walters et al. 2010) – in the case of germplasm, this involves solidifying, or vitrifying, cytoplasm (i.e., forming a “glass”) without too much disruption to the cell structure.

Most cytoplasm vitrifies at room temperature when samples dry to between 30% and 50% RH. If this level of drying can be accomplished without too much cell shrinkage, the cell survives; loss of more than 50% cell volume is considered lethal (Walters 2015b). During embryogenesis, food reserves (starch, protein, or lipid) are deposited into cells, displacing water which consumed as much as 80–90% of the cell volume in an immature embryo and 60–80% of cell volume in a mature recalcitrant (i.e., desiccation-sensitive) seed. The cell volume of a mature orthodox seed changes very little during desiccation (less than 30%), and we believe this explains their extreme tolerance to desiccation as well as subsequent longevity (Walters 2015b). The process of drying without dying makes desiccation tolerant organisms, including orthodox seeds, the original material engineers. This process is highly regulated during embryogenesis (Righetti et al. 2015), and incompletion or disruption of the established program, like any material, can have dire consequences to the functionality and stability of the end product – in this case, seed survival through time (Walters et al. 2010; Walters 2015b).

Once cells are in the glassy state (at temperatures below the glass transition temperature or Tg), they are relatively stable, meaning that change occurs, but over a much longer time scale than reactions occurring in fluid systems. In other words, preservation doesn’t stop the clock, it just slows it down. In solids, such as vitrified cytoplasm, structure and mobility become two sides of the same coin. The “structure” is defined by how the compressed molecules impede movement of neighboring molecules. Pores formed during glass formation and molecules in the glass now shift to fill those pores and pack more efficiently. This rearrangement defines the “mobility” as well as the rate of change within the glass (Menard 2008). The movement brings molecules slowly into closer proximity, where they interact and oxidize; the material becomes brittle. The time scale is often experimentally intractable, which is one of the reasons why seed longevity is difficult to predict or measure. Water is a “plasticizer” of biological glasses, meaning it promotes larger pore space, hence greater mobility and faster aging. If seeds are not sufficiently dried, molecules in the cytoplasm move faster, causing more rapid deterioration. Anti-plasticizers (e.g., cryoprotectants) stabilize structure by a number of mechanisms. There is some speculation that anti-plasticizers of unknown identity accumulate in long-lived seeds (Walters 2015b).

A glass is stabilized by lowering the temperature. For orthodox seeds, which formed glasses during drying at ambient temperatures, molecular rearrangements to form lethal ice crystals at sub-zero (°C) temperatures are improbable, and so longevity increases progressively with lower temperature (to a point) (Walters, 2004). Most genebanks use freezers at −18 °C for conventional storage, because it is highly accessible technology, being easily achieved using a single-stage compressor. Freezer storage of plant germplasm began in the 1970s, and there was strong debate about its benefits until the early 2000s (Zheng et al. 1998; Walters 1998).

Cryogenic storage for plant germplasm became accepted in the mid-1980s and routine in the mid-1990s. Cryogenic storage may occur through a number of platforms, and the appropriate treatment and storage temperature for plant germplasm relies on the vitrification temperature, Tg. Technically, all storage below 0 °C is cryogenic; however, here, we consider it as storage below temperatures achieved by conventional freezers. Mechanical freezers with a dual-stage compressor cool to −80 °C and may be effective for germplasm with a relatively low Tg or unstable glass or when lipid transitions are important to survival. Most frequently, cryogenic storage is associated with the use of Dewar flasks or cryovats that are cooled by liquid nitrogen. Germplasm is either immersed into the fluid and stored at −196 °C or stored in the vapor above liquid nitrogen (between −150 and −190 °C, depending on distance from fluid surface and convection within the tank). A few status reports on longevity of cryopreserved germplasm are available (Towill et al. 2004; Walters et al. 2004; Volk et al. 2008; Ballesteros and Pence 2017; Pence et al. 2017).

Many plant propagules do not survive the desiccation stress required to form glasses at ambient temperatures (e.g., vegetative propagules and some non-orthodox seeds) (Table 10.2). Lowering the temperature of these non-vitrified systems poses high risk of lethal ice formation, which can only be avoided by forming a glass during the cooling process and maintaining it below Tg so that glasses don’t melt and ice doesn’t form during storage. Inhibiting ice formation at sub-zero (°C) temperatures, while maintaining cell viability, requires optimization of interacting treatments for moisture adjustments, additions of cryoprotectants and rapid cooling (Walters et al. 2013; Wesley-Smith et al. 2014).

3.2 Conservation Targets, Sample Quality, and Preservation Success

We preface this section by the infamous story of the Pará rubber tree, Hevea brasiliensis (Willd. ex A. Juss.) Müll. Arg., which produces a non-orthodox seed. Mostly told for political intrigue, the story describes how less than 1% of over 50,000 seeds survived to domesticate the species and initiate rubber industries in Malaysia and Singapore (Brockway 1979). Here, the conservation target for H. brasiliensis was primarily its rubber-producing trait, a species characteristic that can be captured by exemplars. Thus, loss of 99% of the collected seeds, as a result of poor shelf-life, was considered acceptable. We would likely find those losses unacceptable by today’s standards because our conservation targets tend to be at finer scales: genetic diversity representative for the whole species or for key populations or individuals. The salient point, for the context of this chapter, is that preservation success is defined by whether loss of genetic diversity during genebanking is acceptable, which is largely dependent on the stated conservation target.

The Hevea brasiliensis story also illustrates that it is possible to genebank even when seeds are not orthodox. Usually a species producing non-orthodox seeds will be harvested as a cutting or fresh seed and immediately grown out to form a living collection. As described in the previous section, it is now possible to preserve non-orthodox seeds and other germplasm forms that do not survive cytoplasmic solidification at ambient temperatures. However, this effort requires exacting coordination between collector and curator.

An initiative to bank wild species requires background information on how propagules respond in storage. To that end, we cross-referenced prioritized CWR from the USA (Khoury et al. 2013) with information available from Royal Botanic Kew’s Seed Information Database (SID) (RBG 2017) (Table 10.3). A limited number of US species were included in the SID, emphasizing the lack of information available for storage behavior of germplasm from CWR native to the USA. Extrapolating from the genus level, we believe that 75% of the 179 genera from the Inventory will produce orthodox seeds and that at least 8% will not produce orthodox seeds. High variability in seed storage response is exhibited in about 8% of congeners, and no records are given for 8% of the genera.

Table 10.3 Probable seed storage behavior for taxa native to the USA that are congeneric to domesticated species
Full size table

Even orthodox seeds present challenges for storage, especially with conservation targets at the population level, which is where most agronomic- and conservation-based collections are poised. The increasing number of anecdotal accounts that seeds collected from the wild are harder to store are not surprising (Hay and Probert 2013; Walters 2015a; Balleseros and Pence 2017). We know that embryo development is critical to longevity, and metabolic pathways expressed during embryogenesis are keys (Righetti et al. 2015; Walters 2015b). Seed quality is dependent on processes that are uncontrolled in the wild during the growing season, such as moisture availability, nutrition, competition, and pathogens, and it will decline if developmental programs are not completed (Probert et al. 2007) or extended toward germination (Tarquis and Bradford 1992). Seed quality is also under genetic control (Clerkx et al. 2004; Schwember and Bradford 2010; Nagel et al. 2011; Righetti et al. 2015), with ecotypes within a species having contrasting storage behavior (Tweddle et al. 2003; Clerkx et al. 2004; Daws et al. 2004; Walters et al. 2005; Probert et al. 2009; Kochanek et al. 2009; Mondoni et al. 2014). Phenology, fecundity, carbon partitioning, composition, seed coverings, resistance to pests, and drought tolerance are all inherited traits that affect seed longevity. These traits are more uniform in domesticated plants but vary considerably in seeds from natural populations; hence, an accession of seeds collected from the wild will be heterogeneous, and this will result in differences on how individual seeds within the sample respond to genebanking conditions.

Genebanking wild-collected seeds carries inherent risks for genetic erosion: when the shorter-lived seeds in an accession die, seed traits will tend toward greater uniformity, and when seeds are regenerated, germination, flowering, and maturity are likely to become more synchronized. Unless extraordinary measures prevent these tendencies, genebanking seeds collected from the wild can be an exercise in domestication as a result of inadvertent selection of traits controlling preservability, growth habit, morphology/physiology, and reproductive capacity (e.g., Burton and Burton 2002; Gilligan and Frankham 2003; Harding 2004; Aubry et al. 2005; Falk et al. 2006). Even though sample quality may remain high during genebanking, there may be an attrition of sought-after genetic diversity and an increasing tendency for genetic representation of the sample to veer away from the source population (Table 10.1, element 1).

3.3 Predicting Longevity and Detecting Aging

The intention to use genebanking to conserve genetic diversity extant in the wild underpins genebanking practices designed to prolong seed shelf-life. Storage treatments, viability monitoring frequencies, and regeneration are all based on assumptions about longevity. However, longevity varies tremendously among seed lots within a species due to uncontrolled and unknown factors of seed quality. Therefore, the actual longevity of a seed lot is only known after-the-fact. Genebanks need reliable assessments of longevity before and during storage.

Longevity might be predicted using an empirical model that is parameterized by constants for moisture and temperature effects for different species. The Seed Viability Equations (Ellis and Roberts 1980) [http://data.kew.org/sid/viability/ (visited October 3, 2017)] provide “ball-park” estimates of survival with time when conditions are not-too-cold or not-too-dry; this model becomes unreliable beyond the limits of inference of the data used to parameterize it (Walters 1998; Walters et al. 2004). Since freezer storage is a relatively new practice (it was first used routinely at NLGRP in 1978), there are few data sets that actually demonstrate longevity in the freezer, let alone predict it reliably.

Models such as the Viability Equations allow us to “standardize” different laboratory experimental conditions, so diverse species can be ranked for longevity in a similar context (Hay et al. 2003). Not surprisingly, information for CWR species from the USA (Khoury et al. 2013) are not available. The information might be gleaned from behavior of congeners, but only 15 of the 135 genera believed to be orthodox had species coefficients listed in the SID (Table 10.3). More extensive comparisons of seed longevity within the genebank, and estimates of within-species variation for some CWR, come from early genebank results of seeds stored initially at 5 °C (Walters et al. 2005; Nagel and Börner 2010). Additional insights come from seed aging experiments conducted at warmer temperatures under high humidity challenges (Probert et al. 2009) or drier conditions (e.g., Fig. 10.2). There is general agreement that seeds from Apiaceae tend to be short-lived and seeds from Chenopodiaceae tend to be long-lived. Seeds from Asteraceae and Poaceae exhibit a wide range of longevities. In the future, we hope to adjust species-level information with data about habitat, location, and weather data associated with the sample.

Fig. 10.2
figure 2

Survival of a seed sample of Malus sieversii, progenitor species of domesticated apple. Seed moisture was adjusted at 25 °C at indicated relative humidity for a 3-week period; seeds were then sealed in foil laminate packages and placed at 35 °C for indicated time and then germinated. The increasing longevity with decreasing moisture treatment (75–13% RH) is indicative of orthodox behavior, and the faster aging at very low RH (1%) demonstrates a limited benefit of drying. This seed lot survived for about 3 years (+1000 days) at 35 °C. Extrapolating longevity to −15 °C using a general rule for temperature effects (doubling for every 10 °C; Q10 = 2), we can predict this seed lot would survive for about 100 years in a genebank (original data)

Full size image

Wide variation of longevity within a species makes it difficult to accurately predict how a particular sample will behave in the genebank (e.g., Walters et al. 2005). These differences are accommodated in the Viability Equations using an initial quality factor, which is dependent on a highly accurate measure of initial germination as well as an assumed high correlation between initial germination and longevity. Despite model predictions, there is a poor correlation between initial germination and longevity (Walters et al. 2005; Nagel and Börner 2010; Ballesteros and Pence 2017), which we attribute to quality factors that have initial but not long-term effects (Mead and Gray 1999) and factors that have long-term but not initial effects (Hay and Probert 1995; Tarquis and Bradford 1992; Walters et al. 2004).

Monitoring viability is currently the genebank’s only tool for assessing whether quality is maintained. For seeds, this involves a germination assay which, like the initial test, is a snapshot having little predictive power of future change. Statistical considerations related to sample size also influence how well change can be detected (Guerrant and Fiedler 2004; Richards et al. 2010). Accumulation of deaths in time, marked by changes in germination potential, is a poignant demonstration that responses of individuals within a sample vary during storage.

Genebanks must monitor seed viability, but without a priori information on longevity and germination, there is no guidance for monitoring interval or seed counts. When a priori knowledge is insufficient, frequent viability monitoring is recommended (FAO 2014); this demands extensive resources and can quickly deplete the sample, rendering the entire exercise useless. Better processing and storage conditions to prolong longevity, better methods to relate aging at higher temperatures to freezer conditions, and new tools to monitor nondestructively (Colville et al. 2012; Mira et al. 2016; Fleming et al. 2017) will aid the genebanking operation.

4 Standards and Best Practices

Standards or best practices communicate how state-of-art science should be implemented in order to maintain quality samples for a desired timeframe (ISBER 2012). They also communicate to future users how samples were treated (ISBER 2012). Intentions of standards are quality control, predictability, and metrics for how the genebanking experience affected the sample relative to its conservation target. This is particularly important if methods to accomplish bio-banking goals (sensu Table 10.1) are not yet established. Future users deserve to know the history of the sample and how protocols might affect the sample’s usefulness to them. Therefore, best practices and standard operating procedures (SOPs) must convey information to users so that they can access whether the sample is fit for their purpose.

Standards and best practices for genebanking must align with the sampling strategy, which ultimately must align with the stated conservation target. In our opinion, the conservation target is often not sufficiently defined for samples collected from wild populations, and this can cause a mismatch between sampling and genebanking protocols. In the example of Hevea brasiliensis, ability to produce latex for rubber is expressed at the species level, which meant collectors needn’t search for this useful trait among certain ecotypes or individuals. Collectors harvested an excessively large number of seeds because they had prior failures due to the difficult physiology – not because they wanted to capture genetic diversity (Brockway 1979). In other words, the primary conservation target was a few specimens of the species, and extraordinary measures to keep all 50,000 of the harvested seeds alive would have been initially unappreciated. Therefore, in this context, stringent management of the collection wasn’t that necessary. Eventually, though, the new rubber industries learned that greater genetic diversity from higher seed survival would have been beneficial, and collections for genetic diversity within the species would have required more stringent genebanking protocols. Linking this analogy to the case of CWR within the USA, we need to know whether the agronomic traits we seek (e.g., Khoury et al. 2013) are expressed by all individuals of a species, by specific ecotypes, or by some rare individuals. We believe the answer to this question can help to define the conservation target as well as the technical investment required to maintain US collections.

Developing a set of agreed-upon standards or “best practices” (sensu ISBER 2012) for collecting and maintaining diversity invariably develops into discussions of practicalities, impossibilities, and available resources. Experienced genebanks know that working with wild-collected materials usually involves choosing among less-than-optimum options. Often, quality control of samples coming into the genebank is limited by the biology of the organism. Standards, such as FAO’s Genebank Standards (FAO 2014), can guide genebanks established for agricultural purposes (domesticated plants and CWR), and modifications for wild plant species collected for conservation purposes have been suggested (Hay and Probert 2013). Currently genebanking standards for samples collected in the wild require initially high viability and frequent viability monitoring as well as management decisions to regenerate or recollect when viability degrades to 85% of initial viability (FAO 2014, Seed Conservation Standards for “MSB Partnership Collections” at http://www.kew.org/sites/default/files/MSBP%20Seed%20Conservation%20Standards_Final%2005-02-15.pdf (visited on October 3, 2017)). These stringent guidelines are to ensure the sample remains genetically representative of the wild population from which it is harvested (Table 10.1, element 1) (FAO 2014). If the conservation target is not specific to a population, is it necessary to follow these technically stringent standards? If the conservation target is specific to a population, does stringency of SOPs increase risks of genetic erosion by premature consumption of the sample through too much testing or too frequent or infrequent regeneration (Richards et al. 2010)?

FAO’s standards for orthodox seeds (Chap. 4 in FAO 2014) serve as the foundation for many national and international seed banks around the globe, including the USA. These standards were developed about 7 years ago through consensus of a large group of experts having different opinions on the intent of standards and even the interpretation of existing storage data to guide standards. Remaining ambiguities and inaccuracies were place-marked for future research and understanding, and sufficient scientific knowledge has accumulated to call for an update, or at least modification, to meet the needs of seed banks. Improved methods are especially needed to address uncertainty associated with genebanking seeds when there is little knowledge about the species or contingencies when samples are heterogeneous and prone to genetic erosion. Current standards may also be too stringent for some conservation targets. Therefore, we take this opportunity to examine these standards for orthodox seeds and suggest areas for better alignment with conservation targets for germplasm collected from wild populations. To encourage conversation among genebanks, we have listed some standards we feel need adjustment and have provided alternative language (Table 10.4). Standards for non-orthodox seeds in Chap. 6 of the Genebanking Standards (FAO 2014) can also be modified to reflect rapidly developing technologies.

Table 10.4 Best practices for acquisition and storage of orthodox seeds having agricultural utility
Full size table

5 Sample Regeneration and the Nexus of Different Genebanking Strategies

Genebanking seeds collected from wild populations will likely impose genetic bottlenecks (Falk et al. 2006) that can be minimized by careful collection (Hoban and Schlarbaum 2014) and curation (Richards et al. 2010) as well as large enough sample sizes and treatments informed by the biology of the sample (Hay and Probert 2013). The extent and direction of genetic shifts can have large impact on the usefulness of the sample, especially if the conservation target is a rare allele or an accounting of the population at a particular site and time (e.g., Franks et al. 2008). Plant genebanks can invoke certain activities to forestall, or at least understand, the extent of genetic erosion in curated samples.

5.1 Recollect from the Same Wild Population Over Time

Complementation of in situ reserves and ex situ collections provides a reservoir for replenishing genetic resources from wild populations (Maschinski and Haskins 2012; Guerrant et al. 2004; PCA 2015) as well as an opportunity for identifying traits of interest and how they are distributed in a natural population (e.g., Franks et al. 2008). Probabilities of mutation and fixation through drift are predicted to be lower in natural populations, compared to preserved samples, according to theoretical models (Schoen and Brown 2001). Resampling natural populations also provides the opportunity to measure the extent of genetic changes that have occurred through natural forces and those imposed by the genebank (Thormann et al. 2016; Greene et al. 2014). Programs to collect and resample after 15 years have been recently instituted and include species that are widely distributed in North America (Franks et al. 2008), some of which are priority species according to Khoury et al. (2013).

5.2 Regenerate Collected Seeds During Early Phases of Genebanking

Major limitations of seed accessions collected directly from wild populations are low seed number and poor seed quality. As the few available seeds age quickly, situations arise where a sample is regenerated from fewer than ten individuals, resulting in a significant bottleneck. An alternative management practice might be to immediately regenerate a newly collected sample. This would maximize the number of parents contributing to the regenerated sample as well as provide more seeds from a better growth environment and so presumably longer-lived. Studies are underway, using wild-collected germplasm of Limnanthes, Humulus, and Artemisia, to test this hypothesis and gain greater understanding of the interaction between storage time, seed degradation, and shifts in allelic richness for original and offspring populations created before storage and after notable degradation (unpublished; Walters, Richards, Hill, Jenderek).

Sometimes there are simply too few individuals to regenerate a population from seed. Under these circumstances, increasing sample size through clonal propagation can be effective (Pence 2013). When possible, it is important to identify the absolute number of propagules used during regeneration, as well as the number of founders from which they came: 100 individual seeds from a single maternal plant will provide a different regenerated population than one seed each from 100 different plants.

5.3 Use Seeds and Pollen to Back Up Wild-Collected Germplasm of Clonally Propagated Crops.

We estimate that over 30,000 accessions are maintained in living collections of the NPGS, collectively called clonal repositories. These field collections are expensive and vulnerable (Volk et al. 2015). Approximately one third of NPGS clonal accessions are CWR (Volk and Walters 2004). With current cryopreservation technologies and resources, we estimate that it will take 50–100 years to backup these accessions in preserved collections, which might not even begin until the higher priority cultivars are backed up – some 100 to 200 years from now. Clearly rapid and effective strategies to preserve these vulnerable collections are needed.

The conservation targets of CWR may be exemplars of species, snapshots of populations, or particular genes for crop improvement, but usually not the specific genotype of the collected plant. In most respects, CWR of clonal crops can be treated analogously to CWR of crops in general. Backing up these collections can be accomplished if there are seeds remaining from wild plant explorations that fit standards for quantity and quality (FAO 2014, Chap. 4 with adjustments as suggested in Table 10.4). When there are not adequate quantities of viable original seed available, trees planted in the field can serve as parents for regenerated populations. The diversity extant in these field collections can be captured through appropriate parental combinations. Feasibility of maximizing diversity with fewest crossing parents using a maximization algorithm was demonstrated (Richards et al. 2004, 2007; Volk et al. 2005), and a detailed genetic analysis shows high efficiency in capturing alleles with only minor introgression from neighboring pollen (Volk et al. 2016). Storing pollen may also be a promising method to capture and backup genetic diversity within collections (Hoekstra 1995; Volk 2011). Some CWR produce seeds that are not as amenable to conventional genebank storage conditions (i.e., they are not orthodox seeds) (Table 10.3). Methods to preserve these materials are available (Walters et al. 2013) but usually labor intensive and associated with some mortality (Wesley-Smith et al. 2014). No shifts in genetic composition were measured in recovering embryos of a high-priority CWR, Zizania texana, after cryoexposure (Richards et al. 2004).

6 Summary

Genebanks are tasked with ensuring safe preservation of genetic resources so they are available for future use. It is often difficult to predict the eventual use or the timeline for use. The challenge for plant genebanks is to provide viable germplasm that is unaffected by the genebanking experience (i.e., no genetic erosion). This mandate is harder to achieve for natural populations collected from the wild. Some wild-collected materials are likely to resist the extremely controlled conditions of the genebank and will die. Others might adapt and through drift or inadvertent selection become more domesticated. By first identifying the causes for changes in quality of germplasm in preserved collections and then offering strategies to slow down, or at least quantify, the effects of the genebanking experience, wild-collected germplasm can successfully be conserved in genebanks.