Introduction

Plant genetic resources for food and agriculture (PGRFA) are the basic building blocks of crop improvement and adaptation and, by extension, of food security. As a result of the history of crop domestication and global dispersal and adaptation, all countries are now highly dependent upon plant genetic resources located (or originally collected from) beyond their borders. Global interdependence on plant genetic resources has been previously discussed (Crosby 1972, 1986; Diamond 1997; Fowler et al. 2001; Halewood et al. 2014; Mann 2011; SGRP 2011), and predictions have been made of increased future interdependence as a result of challenges such as climate change (Lane and Jarvis 2007; Burke et al. 2009; Jarvis et al. 2010; Fujisaka et al. 2011; Ramirez-Villegas et al. 2013) and the evolution of food systems and diets (Khoury et al. 2014). Global recognition of the policy significance of interdependence on PGRFA arguably reached its zenith in 2001 when ‘interdependence’ was explicitly included in Article 11 of the International Treaty on Plant Genetic Resources for Food and Agriculture (ITPGRFA) as one of two criteria—the other being relevance for food security—for including crops or forages in the multilateral system of access and benefit sharing (MLS).Footnote 1

Through the MLS, ITPGRFA parties agree to create a global, virtual pool of genetic resources for 64 crops and forages (these are listed in the Treaty’s Annex 1). In addition to conservation, this germplasm is intended to be utilized for the purposes of training, breeding and research for food and agriculture. Member states agree to provide facilitated access to one another (including natural and legal persons within their borders) on the understanding that monetary benefits will be shared if the recipients incorporate materials in new, commercialized PGRFA products that are not available to others for research, training or breeding. The multilateral architecture of access and benefit sharing under the ITPGRFA was designed to reflect countries’ current and future interdependence on PGRFA. The system was meant to minimize transaction costs that could otherwise multiply beyond acceptable limits, given the magnitude of international exchanges of genetic resources that accompany agricultural research, development and plant breeding.

In recent years, ITPGRFA member states have expressed concerns that the MLS has not been functioning at the anticipated levels, either in terms of generating financial benefits by users to be shared through the international Benefit-Sharing Fund (BSF) or in terms of materials being made available to, and accessed through, the MLS. Based on this concern, the ITPGRFA’s Governing Body created the Ad Hoc Open Ended Working Group to Enhance the Functioning of the MLS. Its mandate is to develop a range of optional measures to both increase user-based payments and contributions to the BSF in a sustainable and predictable long-term manner and enhance the functioning of the Multilateral System by additional measures.

This article focuses on an issue at the heart of the MLS—the state of global interdependence on PGRFA. We hope that the data presented here will be useful within any process aimed at revising or reforming the terms and conditions of the MLS. It is critically important to keep interdependence in mind when developing policies concerning the conditions under which genetic resources can be accessed and used as well as the ways in which benefits derived from their use should be shared. Illustrating the volume, diversity and geographical spread of global flows of plant genetic resources mediated by Consultative Group on International Agricultural Research (CGIAR) centres, the findings highlight the benefits accrued by virtually all countries in the world—namely, being granted access to a rich variety of materials (and associated technology and information) otherwise unavailable within their own borders and difficult to access under bilateral conditions. The resulting conclusions highlight the importance of the system’s non-monetary modalities for sharing benefits, most of which have involved users in developing countries. We hope that such evidence will encourage efforts to maintain and enhance these mechanisms, in addition to improving the mechanisms associated with monetary payments to the BSF.

Data sources and methods

Data on the holdings, acquisitions and distributions of nine CGIAR genebanks was retrieved from the CGIAR’s System-wide Information Network on Genetic Resources (SINGER).Footnote 2 A system-wide database such as SINGER has never been established for the distribution of germplasm from the CGIAR’s breeding programmes, and, therefore, our study focuses on genebank distributions only. We asked each of the genebank curators to validate the accuracy of the data stored in SINGER and/or to provide updates or integrations. In the end, we obtained validated or updated data for seven genebanks, which are those included in this study (Table 1). Given the magnitude of the distributions from the other centres whose data is not included in this research, i.e., CIMMYT, CIAT, IITA, the final conclusions regarding the extent of international interdependence would likely have been even stronger had their data been included.

Table 1 Total number of samples sent to national recipients from the seven CGIAR genebanks (1985–2009)

Distribution data followed a standard format gathering information according to the fields shown in Table 2.

Table 2 Fields of information included in the distribution data from CGIAR genebanks

Distribution records were available beginning in 1973 for some of the genebanks included in the study, but there were large gaps in the records until 1985 (due to data storage and reporting systems not being fully in place in all centres). Thereafter, the data were more uniform, which led to the decision to consider only the data from 1985 onwards. Since our focus was the germplasm sent to countries and within-country recipients, intra- and inter-CGIAR centre distributions were removed as well as those from CGIAR genebanks to the Svalbard Global Seed Vault. The total number of distributed samples shown in Table 1 was the basis for our analysis. These centres’ mandate crops (and their wild relatives) include key staples for worldwide food security, such as rice, tropical and dryland legumes and cereals, potatoes and other roots and tubers, bananas and plantains and tropical forages (see Appendix, Tables 6, 7 for details on the collections hosted at all CGIAR centres).

Various ways of measuring international PGRFA movements were explored. We considered the total number of samples distributed [a single sample consisting ideally of between 50 and 100 viable seeds or less vegetative propagules (CGKB 2014)], the number of accessions distributed (excluding the repeated distributions of the same accession) and the number of species distributed. The latter two statistics provide a picture of the diversity, rather than the sheer volume, of the flows.

Further analyses qualified the international germplasm flows facilitated by the genebanks using the number of countries from which the materials distributed were originally collected or improved, the number of recipient countries and types of recipient institutions, the number of genera and species distributed, and the type of materials exchanged. Countries were classified based on their development status according to the United Nations classification system (UN 2012), which helped to analyse the germplasm contributions according to the economy of the donor or recipient country. All data handling and analyses were performed in R (R Development Core Team 2011).

Results and discussion

Global flows of PGRFA, 1985–2009: volumes and diversity

Between 1985 and 2009, germplasm conserved in the selected CGIAR genebanks was distributed to a broad range of users. According to the available data, 999,250 samples of 262,872 accessions belonging to 1470 different plant species were distributed during that period. The average number of samples distributed per year (39,970) is below that of the U.S. National Plant Germplasm System (NPGS), where total annual distributions have increased from around 120,000 (Bretting 2007) to more than 200,000 (Heisey and Day Rubenstein 2015) over the past few years. About 30 % of NPGS yearly distributions are typically to requestors from outside the U.S. However, in making this comparison, our lack of data from three important CGIAR genebanks should be kept in mind. Notwithstanding the missing data, the yearly volumes described are much higher than the average number of distributions of other important germplasm systems, such as the Russian Vavilov Institute (6400) (FAO 2009), the German Institute of Plant Genetics and Crop Plant Resources (4400 of barley only) (Ullrich 2011), the Centre for Genetic Resources in the Netherlands (2500) (Centre for Genetic Resources 2008), the Brazilian Empresa Brasileira de Pesquisa Agropecuária (1800) (Da Silva Mariante et al. 2009), the Institute of Crop Germpasm Resources in China (1550) (ICGR 2015), the Plant Genetic Resources Institute of Canada (1500) (Fowler and Hodgkin 2004). These numbers are useful for providing a general idea of the CGIAR’s relative contribution on the international scene, but they should be considered with caution because of the differences in the reporting periods and the limitations of our data.

Virtually all countries in the world have been involved in the exchange of germplasm. The materials listed in Table 1 were originally collected in, or provided by, at least 189 countries and distributed to at least 191 countries. In addition to distributions from the various genebanks, large amounts of germplasm in different stages of improvement have been sent out by the centres’ breeding programmes, although no system-wide mechanism has ever been set up to document these distributions over time. However, data provided by the centresFootnote 3 for the fourth session of the ITPGRFA’s Governing Body indicate that from August 2008 to December 2009 these breeding programmes sent out over 500,000 samples (SGRP 2011). This amount points to the outstanding contribution that the CGIAR breeders make to international flows of germplasm, in addition to the centres’ genebanks.

According to data available through the GENESYS portal, which gathers information on numerous national and international genebanks, the international ex situ collections hosted by the CGIAR centres currently include 712,834 accessions of their mandate crops and related gene pools, originally collected from a vast number of countries (Appendix, Tables 6, 7, 8). The genebanks that were analysed in this study, currently host 445,785 accessions of 2848 species.Footnote 4 Our data suggest that samples of roughly half the diversity held have been distributed at least once by these genebanks.

During the period analysed, there appears to be have been a slight downward trend in the overall number of samples distributed, as already highlighted elsewhere (Halewood et al. 2013). A similar decline was observed in the diversity of the materials distributed, which was measured according to the number of accessions distributed and the number of species represented (Table 3). This trend may be attributed to the fact that the requests became more targeted as more characterization and evaluation data became available, which led to breeders and researchers making requests for smaller sets of materials (Halewood et al. 2013; López Noriega et al. 2013a). For those CGIAR genebanks actively distributing sets of materials for international adaptation trials, the decline could also be due to decreases in the funding made available for these multi-location field operations. It could be that some of the requests that were traditionally made to the CGIAR are now being directed to other genebanks. In addition to institutions that have always been at the forefront of international distributions, alongside the CGIAR, such as the US Department of Agriculture (USDA), a number of national institutions in other countries have been increasing their collections and may be receiving more germplasm requests (FAO 2010). In addition, some private sector users—those most likely to apply some form of intellectual property rights to the final PGRFA products—may have refrained in recent years from requesting germplasm from the CGIAR because of their reluctance to accept the benefit-sharing clauses of the MLS (Halewood and Nnadozie 2008). It is important to note that traditionally these companies have been an extremely small portion of the users of CGIAR materials, as described later.

Table 3 Results of the models used for analysing trends in the overall flows over time (1979–2009)

Types of materials and frequency of distribution

According to GENESYS, over 50 % of the total germplasm distributed by the CGIAR genebanks over the 25 years analysed are landraces or traditional cultivars, which are predominant within these collections (Fowler et al. 2001; Genesys 2014). Breeding and research lines constitute less than 20 % of the materials distributed, while advanced or improved cultivars comprise only 7 % of the distributions. Wild and weedy relatives amount to 12 % of the samples sent out by the analysed genebanks, not only suggesting their importance as sources of useful traits but also reflecting the greater difficulty of using them in breeding compared to other materials (Fig. 1). The decision about which materials to conserve in the long term is made by each centre independently, often following the outcomes of economic analyses on the costs and benefits of conserving materials in genebanks or breeding programmes (Koo et al. 2004). The data in this study reveal that most centres give priority for long-term storage in their genebanks to materials that belong to the primary genepools – that is, the landraces and wild relatives of their mandate crops. This strategy also reflects the fact that all centres with genebanks also have breeding programmes that actively exchange research, breeding and improved lines with partners worldwide, making the conservation of these sets by the genebank neither necessary nor efficient. However, research, breeding and advanced lines are sometimes included in long-term collections, when the properties, or the use of the material, justify it. For instance, this may be the case with materials that have accumulated unique genetic properties (for example, allele combinations), those that are laborious to reproduce (for example, inter-specific hybrids) or those that are commonly used as benchmark varieties in evaluation trials.

Fig. 1
figure 1

Proportion of the different types of germplasm distributed by the selected CGIAR genebanks based on accession data (1985–2009)

Based on the number of samples per accession sent to recipients, there appears to be enormous variation in the popularity of any single accession. Almost 60 % of the accessions in the dataset have been distributed between two and ten times, while only 5.7 % (150 accessions) have been distributed more than 100 times. Most of the latter come from ILRI, CIP and ICRISAT and have been distributed to an average of over 38 countries (SD 20.5) (see Appendix, Table 9 for details on the top 50 most ‘popular’ accessions of our dataset). More than half of these frequently distributed materials are improved lines, whereas landraces, wild relatives and, to a lesser extent, breeding materials constitute the bulk of the accessions transferred less frequently. Among the possible reasons for the ‘popular’ materials to be more frequently requested (that is, by many institutions worldwide) is the fact that the characterization and/or evaluation data already accumulated on them increases their value for breeding and research. This information, in turn, facilitates their use including in institutions and countries with limited capacity or infrastructure for conducting lengthy and costly pre-breeding research using non-adapted populations and wild relatives (FAO 2010).

Providers and recipients

Of the total 189 countries from which material distributed by the seven CGIAR genebanks was obtained, 112 are developing countries, 54 are developed countries and 23 have economies in transition. Of the total 191 recipients, 116 are developing countries, 19 are economies in transition and 56 are developed countries. Data for developing countries and countries with economies in transition has been combined in our analyses. Both developed and developing countries are net recipients—that is, they receive more diversity than they contribute to international gene banks. While this ‘sink’ behaviour is more evident for developed countries, which tend to harbour comparatively less indigenous genetic diversity in their territories, the majority of global exchanges of germplasm mediated by the CGIAR genebanks is distributed South to South—that is, between developing countries (Fig. 2).

Fig. 2
figure 2

Number of accessions exchanged between developed (the ‘North’) and developing and transition countries (the ‘South’)

In their analysis of the flows from six of the CGIAR genebanks and from the USDA’s National Plant Germplasm System (NPGS) between 1990 and 1999, Smale and Kelly Day Rubenstein (2002) also observed that a predominance of developing countries and transition economies were providers and recipients. So too did the CGIAR’s System-wide Genetic Resources Programme (2011) in its biannual reports to the Governing Body of the ITPGRFA. Tables 4a, b provide more detail on the amount, diversity and geographical coverage of the distributions facilitated by the international genebanks for the top 25 provider countries and the top 25 recipient countries.

Table 4 Top 25 provider countries (including total number of samples, genera and accessions originally sourced in these countries and circulated by the CGIAR genebanks analysed in this study as well as the number of recipient countries) and top 25 recipient countries (total number of samples, genera and accessions received as well as the number of countries where these materials were originally sourced) (1985–2009)

Almost all of the top providers listed in Table 4 are developing countries. Many of them are important centres of origin, domestication or diversification of the crops curated by the genebanks considered in this study, including India (rice, millet), Peru (potatoes), Syria and Turkey (wheat and barley), China (rice) and a number of African countries (particularly for tropical forages). Many of the top recipients are also developing countries, and, again, many of them are centres of origin or diversity of crops or forages that they have requested, underscoring the fact that even diversity-rich countries are not self-sufficient in terms of their PGRFA needs. As an example, the difference in the amount of germplasm flowing in and out of India, compared to other countries, stands out as very significant. India has provided and received massive quantities of germplasm. Interestingly, a significant percentage of the materials originally collected in, or obtained from, India ends up going back to Indian recipients (59 % of the samples and over 70 % of the accessions), which makes it the largest recipient of CGIAR-hosted materials originally obtained from within its own borders. A high percentage of ‘reabsorption’ of their own materials through CGIAR-mediated flows are also recorded for Tunisia and Morocco (48 and 42 % respectively), the Philippines (37 %), Iran and Jordan (30 and 25 %) and others to lesser extents. These observations highlight the additional benefit of germplasm deposited in international collections since it provides long-term secure conservation and availability of quality material (and often value-added characterization and evaluation data) originating from one’s own territory, in addition to access to diversity from hundreds of other countries. The latter benefit is particularly relevant for those countries with limited capacity to establish and maintain national conservation programmes for their own local materials.

Differences exist not only in the amount, but also in the type of materials provided by developed and developing countries. While developed countries provide an overall lower quantity of materials compared to developing countries, they contribute a proportionally higher share of materials for which some formal research, pre-breeding or other form of improvement has been conducted. In total, 27 % of the samples ‘distributed’ by our seven CGIAR genebanks from developed countries were research materials and improved/elite lines (with the United States supplying as much as 80 % of this category); only 14 % of the samples distributed from developing and transition countries belonged to these categories. On the recipient side, the share of germplasm that carried some degree of research and improvement flowing into developing countries and transition economies is 30 % of the overall incoming samples, while it is 14 % for developed countries.

In both developed and developing nations, public institutions (including the National Agricultural Research System (NARS), universities and genebanks) are by far the predominant recipients of CGIAR materials (Table 5; Fig. 3). These public sector recipients are located in developing countries in over 75 % of the cases. The share of samples sent to commercial companies is only around 3 % of the total, and the recipients are primarily (77 %) in developing countries.

Table 5 Type of recipients, samples and accessions and percentages over the total
Fig. 3
figure 3

Share of accessions received by different recipient categories (1985–2009)

These findings are also consistent with those of Smale and Day Rubenstein (2002) who found that most recipients of germplasm from the US NPGS, another important worldwide facilitator of PGRFA for research and breeding, were in the public sector. The volume and diversity of the PGRFA flows described in this study, albeit only a small sample of worldwide exchanges, demonstrate the extent of countries’ interdependence on PGRFA for crop improvement and, ultimately, food security. While acknowledging the limits of our dataset, we believe that the conclusions regarding the extent of international interdependence would likely have been even stronger had the data from important genebanks such as those at CIMMYT, CIAT and IITA been included. The emerging picture confirms an established description of modern agriculture as an interdependent network of seed and germplasm sources, in which very few countries or farming systems in the world do not rely to some degree on the international system that moves crop germplasm, breeding lines and improved varieties across international borders (Duvick 1984).

Analyses by other authors confirm these patterns, describing how crop improvement has benefited from access to a wide range of materials with different origins. Fowler, Smale and Gaiji (2001) undertook an analysis of CGIAR data focusing on a different time frame and different measures than those presented here. Smale et al. (2002) used the case of spring bread wheat released by national programmes in developing countries. Warburton et al. (2006) and Dreisigacker et al. (2005) looked at synthetic hexaploids to illustrate the significance of access to wild relatives from centres of diversity in wheat improvement. Voysest et al. (2003) took the case of beans in Latin America (Fowler et al. 2001; Smale et al. 2002; Voysest et al. 2003; Dreisigacker et al. 2005; Warburton et al. 2006). Additional studies have focused on those countries that are the centres of crop diversity. Rejesus et al. (1996) reported that 45.6 % of the germplasm used by wheat breeders in Western Asia, the Vavilov centre for the species, comes from international sources. Evenson and Gollin (1997) documented the dependence of Asian countries, including the Vavilov-centre countries such as India, Burma, Bangladesh, Nepal and Vietnam, on IRRI for rice germplasm of different provenance (65.0 % in India and 98.1 % in Vietnam) (Rejesus et al. 1996; Evenson and Gollin 1997). All of this evidence points to the ‘international public good’ nature of the materials held and made available by the CGIAR as well as by other actors who make such materials available. It highlights the importance of supporting the continuation and enhancement of conservation as well as the internationally facilitated sharing of germplasm within the framework of the ITPGRFA.

Conclusions

It is clear that access to globally pooled genetic resources is a fundamentally important benefit that all countries have historically exploited when systems were set up to facilitate such access. Any effort to improve the MLS must be guided by the necessity of supporting and improving countries’ ability to further capitalize on this benefit. This is particularly true considering the contemporary challenges associated with climate change (Fujisaka et al. 2011), population growth and the harmonization of diets across the world (Khoury et al. 2014). While acknowledging the importance of improving the monetary benefit-sharing mechanisms, we believe that one should not lose sight of the need to maintain the non-monetary benefit-sharing mechanisms when evaluating the effectiveness of the MLS and considering options for its reform. Significant knowledge and opportunities for crop improvement accompany the materials distributed by the CGIAR genebanks, so focusing exclusively on the monetary benefits that can potentially result from germplasm flows represent too narrow a view of its overall impact. Indeed, it has been argued that non-monetary benefits from the MLS (as outlined in Articles 13.1 and 13.2(a)–(c) of the ITPGRFA) can generate much greater economic return than developing countries would ever gain through the BSF.

With respect to monetary benefit sharing, it is important to underscore the fact that the primary users of germplasm from the CGIAR and the MLS have been public sector organizations (in developing countries) rather than private sector entities. Indeed, it has been pointed out that a crucial factor that determines the demand for genetic resources in the seed and crop protection industries is the effort required to turn them into usable materials. Genetic resources that widen a company’s gene pool, but without the identified properties of interest, are typically considered to have little commercial value since they require considerable investment and the return on investment is often risky (Smolders 2005). Although new technology can assist in the search for a specific trait, the expense of doing so is generally prohibitive, particularly for smaller companies (Laird and Wynberg 2006). Larger companies that would most likely trigger the mandatory financial benefit-sharing provisions associated with the MLS tend to opt out of receiving materials from the system (Halewood and Nnadozie 2008). These kinds of reasons likely underlie the failure of efforts to ‘privatize’ monetary benefit sharing through the adoption of mechanisms for mandatory payments from companies based on sales of products that incorporate materials from the MLS.

We believe that some other approach to monetary benefit sharing, linked to the operation of the MLS, is necessary. Such an approach should more closely reflect the public goods nature of PGRFA as well as the historical development of the international and national collections that host most of the materials that do, and will, constitute the MLS. It should also be as simple as possible, and less administratively burdensome on both the providers and users of PGRFA, to encourage, rather than discourage, participation. In particular, it could be useful not to link the collection of financial benefits to the privatization of products incorporating materials from the MLS. Rather, it could be governments or public authorities which devise means to assume the costs of the MLS’ proper functioning, in a more familiar form of state assumed responsibility on publically valuable assets. Governments could then decide if and how they would need to recoup some of those costs; one option, which was actually discussed in early Treaty days, could be some sort of contribution from the commercial sector based on their annual seed sales. This approach would also be in line with the way public organizations have historically supported the collections.

Of course, there are other ways to improve and enhance the functioning of the MLS and to acknowledge countries’ increasing interdependence on PGRFA, beyond adopting a new approach to monetary benefit sharing. No matter how well the system is designed or reformulated, there are practical, institutional and capacity limitations for all countries and all potential beneficiaries (from farmers to breeders and researchers) to take advantage of the MLS, even once their legal ability to do so has been established. This may be particularly true in some developing countries. Capacities and strong partnerships need to be established among the broadest possible range of stakeholders, enabling them to recognize specific trait-based needs, identify where the potentially useful materials could be within the MLS, and request, receive and use the materials concerned. A more proactive and widespread participation would contribute to a greater willingness to voluntarily introduce materials into the MLS, increasing the diversity available to agricultural research and development and giving rise to additional monetary and non-monetary benefits to be shared.

It has been argued that capacity building, technology transfer and information exchange in the context of the MLS should take place in close relation to other ITPGRFA objectives, particularly the recognition and protection of farmers’ rights (Article 9). Indeed, a number of countries have flagged their concern about the MLS having too narrow a focus to the detriment of issues that are more closely related to farmers and their role in on-farm conservation (López Noriega et al. 2013b). After all, most of the ex situ materials that are being, or will be, circulated globally thanks to the MLS are landraces or naturally adapted resources developed and conserved by small farmers, often from developing countries. Their role today is ever more crucial for allowing the continued conservation, evolution and development of genetic resources with the potential to adapt to changing climates. Greater synergy between the architecture of the MLS and the implementation of farmers’ rights would also contribute to moving the ITPGRFA forward as a package of integrated measures, building confidence among a wider range of key stakeholders and truly reflecting global interdependence.