The circulation of data – “full exchange of information among national, regional and international agricultural research centers” – ranked high among the objectives adopted by representatives to the Consultative Group on International Agricultural Research (CGIAR) at its first meeting in 1971.Footnote ¹ It was considered essential to CGIAR’s most important goals, from identifying the needs of individual countries or regions, to ensuring the coordination of research among different institutions, to allocating funds. In this chapter, we look at how agricultural experts attempted to realize this “full exchange of information” among scientists working at geographically distant sites, in different languages and cultural contexts, and with different organisms and research interests, in the four decades after the founding of CGIAR. Our focus is the historical development of crop descriptors, which CGIAR today defines as providing an “international format and a universally understood language for plant genetic resources data.”Footnote ² We examine crop descriptors as a critical component of CGIAR’s earliest efforts to create “system-wide” research tools and agendas, emphasizing the scientific and political agendas that shaped centralizing, systematizing work orchestrated as a top-down enterprise.

Developers of descriptors aspire to agree on specific characteristics of crops, such as plant height or fruit shape, and exact terms for describing these – for example, “height of plant at maturity, measured in centimeters from ground to top of spike, excluding awns” or “plum-shaped” (Figures 10.1 and 10.2). Historically this work has been motivated by the idea that widely agreed descriptors will allow diverse and globally dispersed users to share plant materials and information. It has sought especially to make it easy to manage and communicate the data associated with samples of plant genetic material tested in field trials or stored in research institutions and gene banks. Finding common labels and formats for such data has long been a challenge for agronomists, plant scientists, and curators, not least because the characteristics of interest to these researchers extend beyond those deemed to be relevant in traditional taxonomy. Today they may include everything from genomic data to breeders’ assessments to ethnobotanical context to market uses.

Figure 10.1 This list of possible fruit shapes was intended to guide researchers working with papaya in systematic description of this trait in their collections and field trials. From IBPGR, Descriptors for Papaya (Rome: IBPGR, 1988), p. 17.

Reprinted by permission of Alliance Bioversity–CIAT.

Figure 10.2 Fruit shape, skin color, flesh color, and productivity were just a few of the several dozen traits and other identifying data that papaya researchers were encouraged to track in standardized form. From IBPGR, Descriptors for Papaya (Rome: IBPGR, 1988), pp. 16–18.

By permission of Alliance Bioversity–CIAT.

Our historical reconstruction of the technically and culturally complex project of descriptor creation shows how, in addition to bridging expert domains, including botany, agronomy, genetics, breeding, and farming, it provided an opportunity for CGIAR to instantiate and consolidate its central position in a larger web of international agricultural research initiatives. Providing descriptors served to advance CGIAR’s identity as an essential resource for globalized development. As we show, descriptors acquired increasing strategic importance within CGIAR over time, serving as evidence of the organization’s role in enabling global agricultural research and as instruments for shaping related policies and strategic objectives. Descriptors fulfilled a politically significant social function, establishing CGIAR as a necessary passage point in coordinating the exchange of data and expertise about plant genetic resources and constraining alternative approaches.

We argue that the project of producing descriptors both defined and embodied CGIAR institutional identity and objectives as these evolved from the 1970s to the 2000s. On the one hand, descriptors were intended as generalizable tools for agricultural development. Well-defined and widely used descriptors would not only enable CGIAR institutions to work together by pooling data and related materials and methods, but also allow CGIAR to respond to – and to some extent shape – the key institutions and regulations, both national and international, of global agricultural development. On the other hand, for universalized descriptors to be adopted and effective in research, they needed to be locally meaningful. This meant identifying descriptors able to encompass different crops, users, research agendas, and even diverging agricultural strategies. This was (and continues to be) a complex challenge, especially given the enduring tensions between a regulatory and scientific sphere dominated by Euro-American interests and expertise and the heterogeneous demands for and understandings of agricultural development emerging from the Global South. In reconstructing this history, our analysis complements studies that approach the history of CGIAR via local experiences of top-down research agendas (see the contributions to Parts I and II of this volume). We show how “the center” – and not “the Centers” – responded to changing circumstances, including frictions felt at the local level.

The agricultural technologies that precipitated the creation of CGIAR, and which remain central to the work of many of its centers, are the seeds of novel crop varieties and the systems of production that sustain their cultivation. As Marianna Fenzi (Chapter 11, this volume) describes, CGIAR’s position as the steward of some of the world’s most extensive collections of “plant genetic resources” – that is, seeds and other crop genetic materials held in gene banks and related facilities – has placed it at the heart of international controversies regarding ownership of and control over these resources.Footnote ³ As a result, seeds and the power associated with the possession and dissemination of these have been central to historical investigations of CGIAR and the research institutions associated with it.Footnote ⁴ By comparison, the history of data management, sharing, and reuse within CGIAR has mostly escaped close observation, although it will come as no surprise to historians of science and agriculture that information about seeds has been as important as the seeds themselves. The infrastructures needed to shuttle seeds from site to site without losing the identifiers and data attached to these have been crucial to CGIAR gaining and retaining power in international agriculture. They are becoming ever more influential within digitalized, data-intensive, and increasingly automated approaches to biology and breeding, in contexts where ownership of seeds and data remains hotly contested (see David J. Jefferson, Chapter 12, this volume).Footnote ⁵ A historical understanding of the role played by CGIAR in this domain is therefore essential to understanding the present and possible futures of global agricultural development.

Building a Network

Since its inception, the work of crafting descriptors has been tied up with the management and use of crop genetic resources, especially by breeders. Descriptors initially aimed to identify useful seeds in collections and facilitate their exchange among an ever-growing number of researchers. From the late 1960s, several different crop research communities attempted to coordinate the methods and language they used to document information about breeding materials held in collections or used by researchers. Rice scientists saw standardization in documentation as a way to deal with their own diversity as much as crop diversity.Footnote ⁶ Other researchers were motivated to study standardization by the possibility of using new automated data storage and retrieval systems to coordinate international breeding activities.Footnote ⁷ By the early 1970s, dealing with a surfeit of seeds provided additional impetus. With millions of seeds already in seed banks and more anticipated, only clear and consistent modes of description would enable researchers to navigate these collections.Footnote ⁸

CGIAR’s entry into data standardization initiatives came via its early focus on disseminating new crop varieties, a task that both generated collections of crop diversity and made conservation of these imperative. At their first-ever meeting, in the summer of 1971, members of the CGIAR’s Technical Advisory Committee (TAC) debated what research activities would best ensure that the “promise already shown by the ‘Green Revolution’” could be extended geographically. Towards the end of their deliberations, which mainly focused on what new international research centers would complement the four existing institutes, several participants relayed their concern that the accelerated spread of “modern” crop varieties was causing “the progressive erosion of natural genetic resources.” In other words, they believed that genetically heterogeneous farmers’ varieties were giving way to more uniform breeders’ varieties. As Marianna Fenzi (Chapter 11, this volume) recounts, this concern eventually precipitated a new CGIAR institute, the International Board for Plant Genetic Resources (IBPGR), with the mandate to “promote an international network of genetic resources activities to further the collection, conservation, documentation, evaluation and utilization of plant germplasm.”Footnote ⁹

IBPGR was unusual for a CGIAR center in that it was not a physical facility, but instead a group of geographically dispersed experts who convened at regular intervals and were supported by a secretariat at the United Nations Food and Agriculture Organization (FAO) headquarters in Rome. It was also unusual in that, initially, it didn’t conduct any research itself, but served chiefly to manage funds and – more aspirationally – to coordinate the actions of many widely dispersed and independently motivated researchers and institutions. These two features of IBPGR explain the largest line-item in its budget during its earliest years: investment in the creation of a “Communication, Information and Documentation System.”Footnote ¹⁰ This “integrated system” would, it was hoped, “support all phases of management of genetic resources data,” from collection in a farmer’s field, to filing in seed bank storage, to evaluation by a crop scientist in an experimental plot. In addition to being flexible enough to accommodate these different scientists, it would be adaptable to the different computing capacities found at different institutions managing genetic resources.Footnote ¹¹

Since IBPGR didn’t have its own in-house research and development capacity (beyond desk studies conducted by the small staff of the secretariat at FAO), it contracted out the work of creating its information infrastructure to a research group, the Taximetrics Laboratory, at the University of Colorado, Boulder. A few years earlier, FAO-led efforts to orchestrate collaboration in crop exploration and conservation had prompted an assessment of the Taximetrics Laboratory’s Taxonomic Information Retrieval system (TAXIR), for this purpose.Footnote ¹² With the influx of money from IBPGR, the Taximetrics Laboratory turned its attention to developing an information system to be used in managing genetic materials held at CGIAR centers and national collaborators of IBPGR. TAXIR, which was a product of US National Science Foundation funding, was adapted into a new system for managing genetic resources data, called EXIR.Footnote ¹³

Having an agreed-upon set of identifying information to describe samples in collections – and consistent terms for communicating it – was considered crucial to the operation of this system. Echoing a view already circulating among crop scientists, IBPGR maintained that collections of plant genetic materials were “only as good as the use that can be made of them, and without information they can hardly be used at all.”Footnote ¹⁴ Its planned system aimed to ensure that essential information accompany all samples in its affiliated ex situ collections. This was not simply a matter of creating means for data storage and access. It was also one of dictating the nature of the data stored. As an IBPGR report explained, “Different collections of the same species have been made by different people and for different purposes and so they have been described in dissimilar ways.” Facilitating management of, communication about, and access to collections therefore required an “internationally accepted system” for describing their contents.Footnote ¹⁵

The stakes for setting descriptors and the challenges to agreeing on these were evident in IBPGR’s attempt to enlist researchers in setting “a minimum list of taxonomic, morphological, physiological, resistance, and quality characteristics” for wheat and its relatives, and an “inventory of descriptors” to capture these, in the mid-1970s.Footnote ¹⁶ In Boulder, where researchers were central in forging the very concept of descriptor, a group of maize and wheat experts gathered in 1975 to develop initial lists of types of descriptors.Footnote ¹⁷ These proposals informed discussions at an international symposium on wheat later that year in Leningrad (St Petersburg), where attendees agreed on the minimum information that should be attached to every item accessioned into a collection.Footnote ¹⁸ This list went through further refinement in 1977 when Japanese, West German, Soviet, and US wheat scientists, in consultation with the team of Boulder-based data scientists, drew on the Leningrad recommendations, data collated from world wheat collections, a glossary of wheat characteristics, and other data to propose “a list of minimum descriptors.” The purpose of this list was to facilitate an international evaluation of wheat varieties.Footnote ¹⁹ If every collector, curator, and breeder tracked the same thirty-three essential items of information (“descriptors”), using the same scales and standardized responses (“descriptor states”), then it would be irrelevant where and by whom evaluations were done, at least with respect to interpreting the data. Plant height would always appear in centimeters and be calculated without including the awns. Kernel plumpness would be rated from 1 to 9, with 1 indicated “shrivelled” and 9 “plump.” The number of spikelets per spike would be decided by averaging five spikes. And so on.Footnote ²⁰

Expert-developed and community-agreed descriptors like those created for wheat were meant to make collections “useful to workers other than those who have assembled them,” overcoming institutional divisions of labor, as well as cultural divides.Footnote ²¹ Heretofore uncoordinated approaches to assessing phenotypic traits, which made it difficult to share and compare plant materials, would be aligned to a single standard – or, rather, a list of them. The first attempt to implement the new wheat descriptors in an international evaluation program revealed just how ambitious this goal was. Sets of 400 wheat samples from collections were sent to several sites, with instructions to grow and characterize them according to the agreed descriptors. Of the four institutions that returned results by 1980, none returned data for every descriptor. Very few descriptors were recorded across all institutions and one, drought resistance, was not recorded by any.Footnote ²² Researchers on the ground evidently lacked the time to assemble complete datasets. They may also have disagreed with top-level coordinators about which data were useful and which were not.

The aims of IBPGR’s nascent descriptor program, and the obstacles to its realization, reflected the ambitions of the still-young CGIAR and the realities of international agricultural research in the 1970s. These were the heady years when a rapid extension of the Green Revolution through institution-building and technological innovation – especially innovation in crop varieties promising higher yields – seemed possible to funders like the Rockefeller Foundation and the World Bank. A system for shuttling the genetic resources considered essential to crop development from one site to another and one researcher to another would be a critical component of CGIAR’s growing institutional network. If heterogeneous descriptions were an obstacle to the efficient transfer of material and information, then a group of experts could convene to decide standard ones, and all other researchers, whether at CGIAR centers or in national research institutions, would conform to this universal norm. The problem was that local circumstances – institutional, environmental, cultural, political – often resisted this centralizing, universalizing ambition to identify high-yield crops.

Descriptors’ Ascendancy

CGIAR, through the new IBPGR, could and did build on the knowledge and expert communities that FAO had fostered since the 1940s as it began to coordinate the transit of breeding materials and information in the 1970s. Moreover, because it had no research capacity of its own, IBPGR depended on other institutions to achieve its objectives. These included CGIAR centers, national research institutes, and universities. This institutional positioning created circumstances in which the development of descriptor lists became the defining work of IBPGR through the 1980s with respect to its mission of facilitating cross-institutional exchange of genetic resources.

The intensified focus on descriptor lists as one of the defining contributions, if not the defining contribution, of IBPGR to international agricultural research in the 1980s followed a major disruption to the organization’s communication and information program. The research group in Boulder that IBPGR had funded to develop its information management system, TAXIR/EXIR, was recruited in 1978 to lead the development of an information system for the US National Plant Germplasm System. On review, it looked as though IBPGR’s significant financial investment – about $1.5 million between 1975 and 1978 – had mainly gone towards developing technical systems that did not serve most CGIAR centers’ needs and expert knowledge that was now contracted to a different institution. Indeed, a panel assembled to evaluate these efforts deemed EXIR to be cost-ineffective, and IBPGR quickly abandoned its development of a centralized, universal computer system for genetic resources management across CGIAR. “Practical ad hoc adaptation” of any computerized data management system to existing local equipment was thought to promise faster, more sustainable results.Footnote ²³

Although IBPGR abandoned its aspirations for a unified approach to hardware and software, it intensified its goal of developing a universal language for recording and communicating information about breeding materials and crop varieties. The 1982 IBPGR Annual Report reiterated that “the biggest and most difficult problem to solve” with respect to genetic resources remained accurate documentation. This was essential to nearly every task, from planning collecting expeditions to curating to sharing materials.Footnote ²⁴ This in turn justified the accelerated production of lists of standardized descriptors and descriptor states for crops.

Over the next five years, IBPGR published dozens of descriptor lists in rapid succession. Its schedule ostensibly prioritized economically and socially important crops. In practice, priorities depended on the availability of existing information and relevant expertise, which in turn derived from previous research investments. Crops such as wheat or rice that had long histories in CGIAR centers, linked to their crucial role in industrial economies, provided obvious starting points for systematic information-gathering and discussion. The development of each descriptor list was supported by an advisory body that included biologists and agronomists with expertise in the crop at hand.Footnote ²⁵ The twenty-one lists newly published or revised in 1985 alone included several staple grains (e.g., wheat, rye, oats, millets), fiber crops (cotton), oil plants (sunflower), pulses (lentil, chickpea, mung bean), beans (faba, tepary), fruits (apricot, cherry, peach, plum), multiple forages, and still others.Footnote ²⁶ By 1991, IBPGR had published descriptors for seventy crops.Footnote ²⁷

The development of standard descriptor lists was accompanied by efforts to standardize across crops as well as within them. IBPGR introduced a new list format in 1982, identifying minimum information to be gathered by collectors and to be kept by curators on the status of samples maintained in a gene bank, as well as “standard numbering” and “standard descriptor states.” This revised format also aimed to guide the production of more and better information at various points in the trajectory of a seed from farmer’s field to gene bank to experimental site and back to the bank, not least by clearly demarcating collectors’ responsibilities from those of curators. Collectors were further aided by the creation of standard collectors’ forms, an intervention that was seen as resolving concerns about missing data, as well as inconsistencies in language and content.Footnote ²⁸

In attending to the publication of descriptors as its key contribution to research, IBPGR strove to streamline and standardize the characterization of seeds and other materials in the interest of efficient exchange and use. Its “minimum” lists sought to achieve maximum compliance by limiting the quantity of information required of hurried collectors, harried curators, and financially stressed research institutions. Yet these materials were scattered across institutions that deployed esoteric cataloguing systems and different computer software and hardware, and where researchers and curators spoke different languages. Institutions bore responsibilities for diverse crop species and responded to divergent cultural expectations for those crops, not all of which could be adequately captured in the standard descriptor list. The diversity of research made adherence to the minimalist ideal difficult.

This tension was apparent even in 1980, when external reviewers first formally advised IBPGR to drop the development of hardware and software and focus instead on descriptors themselves. The advisory panel urged against “over-elaborate descriptor lists,” calling these “self-defeating” and recommending instead that lists be kept “as short as possible” by focusing on the institutional identifiers and “basic botanical characters.”Footnote ²⁹ IBPGR’s ostensible emphasis on minimal descriptor lists would suggest that it acceded to the panel’s admonitions as it reoriented its activities in the early 1980s – except that this emphasis was short-lived. By 1992, IBPGR descriptor lists were viewed not as minimal but as maximal and celebrated as providing “the widest number of descriptors that will assist with the characterization of the crop.”Footnote ³⁰ Consider the 1991 descriptor list for sweet potato, an early product of the new comprehensive approach. It included four categories of descriptors: passport (collectors’ data), characterization (highly heritable, highly visible traits), preliminary evaluation (a limited number of traits “thought desirable” by many users consulted during list development), and further evaluation (basically, anything else considered useful in breeding). Users could record, in standardized form, collection data (e.g., site, collector, institution, environmental qualities), basic characteristics of the plant (vine color, leaf shape), more fine-grained details (root surface defects, flesh flavor), and a breathtaking array of evaluation data (data and location, soil taxonomy, root cracking, crude fiber content, keeping quality, drought tolerance, pest resistance).Footnote ³¹

The curator at the CGIAR’s International Potato Center (CIP) in Peru who oversaw this publication insisted that the newly expanded list was essential for improving management and use of sweet potato collections. The minimal list of sweet potato descriptors that had been agreed in 1981 by a small group of experts convened in South Carolina, USA and published by IBPGR had been revised and expanded almost immediately, after researchers attempted to apply it to collections in Fiji and Papua New Guinea. In 1986, when CIP launched an assessment of its 1,500 sweet potato accessions, the curator had expanded this already expanded minimal list still further. Yet, as he later reported, “even this expanded list was not adequate enough to describe all the morphologic variation shown in CIP’s collection.”Footnote ³²

The “single-language” vision of descriptors was abandoned, much as the single computer system had been. What had happened? It is tempting to suggest that the diversity of crops and crop researchers was just too great to be accommodated in universal standardized minimal lists. This is what the example of the sweet potato seems to indicate. However, looking outwards to the political debates and institutional wrangles in which IBPGR was involved in the 1980s suggests that these tussles were at least as important as the technical, biological, and cultural constraints encountered at the coalface of descriptor production. Throughout the decade, IBPGR was embroiled in a fight over the ownership of plant genetic resources that played out with particular fury within FAO (see Marianna Fenzi, Chapter 11, this volume). In response to accusations of its pirating seeds from farmers of the Global South to gene banks of the Global North and grossly mismanaging collections, IBPGR scrambled to show its commitment to maintaining open access to seeds and to rectifying perceived management issues.Footnote ³³ Efforts at centralization and control were pushed aside in favor of inclusivity and inviting broader expertise, and IBPGR renewed its emphasis on data production and circulation.

The scrutiny of IBPGR in FAO forums and beyond prompted significant institutional change. The existing international system for collecting and conserving crop genetic materials, ostensibly overseen by IBPGR and therefore reporting to CGIAR, was heavily reliant on CGIAR centers and well-funded agricultural research institutions in a handful of industrialized countries.Footnote ³⁴ Critics wanted to see FAO placed in charge of such a system. FAO offered equal representation and voice to all member nations, whereas CGIAR in the 1980s was still chiefly an organization of donor countries and their scientist advisors. As part of their bid to undermine IBPGR, advocates of change pointed out that it had no clear legal standing: unlike other CGIAR centers, it had not been founded as an independent international institution via an agreement with a host country.Footnote ³⁵ An initial attempt to resolve these concerns ultimately resulted in an institutional break between IBPGR and FAO and the establishment of IBPGR as an independent entity.

IBPGR’s emphasis on decentralization and inclusivity in the creation of crop descriptor lists came during this period of institutional crisis. It took shape as part of a response to complaints about CGIAR’s largely self-assumed – and to some critics unauthorized – management of global crop genetic resources. This suggests that maximal description was a political solution as much as a technical one. It attempted to improve the quality and usability of descriptors while also shoring up the perceived legitimacy of IBPGR.

Going Global

As the form of crop descriptors expanded, so too did their functions. The elaboration of new international frameworks for managing crop genetic resources, beginning with the International Undertaking on Plant Genetic Resources for Food and Agriculture agreed at FAO in 1983, made data generation and data norms and standards more important than ever before.Footnote ³⁶ First conceived as a tool for the exchange of information about accessions to collections, and therefore the exchange of accessioned materials, descriptors were integrated into new international regimes for tracking and governing plant genetic resources. In the run-up to the 1992 Convention on Biological Diversity (CBD), for example, descriptor lists produced by IBPGR were portrayed as a tool for promoting information exchange as part of the technical and scientific cooperation mandated by the convention.Footnote ³⁷

Positioning descriptor lists as key tools to support international cooperation, thereby highlighting the technical contributions of CGIAR to global agricultural development, was of special strategic significance at the start of the 1990s. CGIAR had grown to encompass eighteen centers and was taxed by the complexity of managing this institutionally diverse and geographically dispersed network while also negotiating an expanded research remit.Footnote ³⁸ CGIAR administrators grappled with pressing financial concerns, including both the extent of resources required to orchestrate work across various locations and the need to comply with the demands of funders while respecting the autonomy of each center. In addition, the United States Agency for International Development (USAID), which had provided most of the financial support for CGIAR since 1971, was increasingly reluctant to do so.Footnote ³⁹ Retrospective accounts have characterized the 1990s as a period of “crisis” for CGIAR, during which it faced criticism for its inconsistent and uncoordinated portfolio, its inability to address emerging challenges as a result of cumbersome managerial and financial structures, and its exclusion of representatives from the Global South.Footnote ⁴⁰

During this period of institutional crisis, CGIAR took steps to shore up its central role in the international flow and management of genetic resources. IBPGR transitioned into a new, independent, and legally authorized CGIAR center, the International Plant Genetic Resources Institute (IPGRI) in 1994. IPGRI was tasked with serving the genetic resources needs of the other CGIAR centers, making its operation the first cross-institute initiative specifically focused on standards for general use. Among the functions that IPGRI assumed – in this case from both IBPGR and FAO – was that of maintaining an authoritative, comprehensive list of internationally accessible gene banks.Footnote ⁴¹

The standards developed by IBPGR/IPGRI were seen as means to connect and coordinate the sprawling network of CGIAR centers, and to clarify their relations to other international initiatives, as well as to address concerns about a lack of inclusivity within CGIAR. At the technical level, one way to show support for a more diverse user base was to emphasize the broad relevance of the standards produced and their inclusivity compared with other systems. In 1992, IBPGR had confirmed its crop descriptor lists as allowing the “widest number of descriptors.” However, when it became apparent that comprehensive descriptors were cumbersome for breeders with fewer resources to deploy – implemented primarily by those at well-resourced institutions with the effect of excluding others, especially those working in the Global South – the pendulum swung back.Footnote ⁴² Around 1993, IBPGR/IPGRI began to resimplify descriptors. Comprehensive descriptors were not abandoned, but instead accompanied by a reduced, general list of “minimal,” “highly discriminating” descriptors that could be applied across species and locations. This new format, first trialed with barley in 1994, was thought to “reduce redundancy” and again make descriptors more user-friendly.Footnote ⁴³ IPGRI acknowledged that some descriptor lists were long but encouraged researchers “to utilize those that are important in their own situations.”Footnote ⁴⁴

This solution was envisaged as cost saving, in that it would reduce the resources dedicated to implementing crop descriptors within each center. It also chimed with, and was subsumed into, a larger quest to develop common computational tools and infrastructure to support system-wide coordination within CGIAR. The early 1990s saw efforts to “solidify a network of computer systems” across the centers, under the guidance of the data communications firm CGNET International, as well as the installation of equipment and software for managing large databases at IBPGR/IPGRI.Footnote ⁴⁵ In 1994, CGIAR launched the System-Wide Information Network for Genetic Resources, which aimed to facilitate data sharing by linking the independent genetic resources databases of twelve CGIAR centers. The quest for internal, system-wide compatibility of the data used to document and manage crop genetic resources sought to make these available both within and – crucially, given the controversies about accessibility of breeding materials to all users – outside the CGIAR system.Footnote ⁴⁶

The changing circumstances in funding to and governance of CGIAR in the 1990s included other efforts to redress the perceived imbalance of power in determining the direction of international agricultural research and development. Responding to concerns that national research institutions, though crucial to the success of most agricultural development objectives, had little voice in setting priorities, CGIAR and other international institutions such as the UN International Fund for Agricultural Development tried to create mechanisms that would amplify the voice and role of national agricultural research systems.Footnote ⁴⁷ Consultative processes that engaged state-level organizations bore witness to their demands for better venues for transnational dialogue and cooperation. These processes led to the convening in 1996 of a Global Forum for Agricultural Research (GFAR) that was to encompass all stakeholders, from farmer organizations to national research systems to the World Bank, FAO, and other international actors. GFAR was charged with, among many things, reassessing the mandate of CGIAR.Footnote ⁴⁸ By dint of the breadth of institutions included, GFAR convenings highlighted the disparity between CGIAR’s central political and strategic influence on global agriculture and its relatively minor economic role. Aggregating across the many and varied institutions engaged in agricultural development, CGIAR represented “only 3% of the annual investment in research geared to agriculture in developing countries,” and yet it played a crucial role in providing the means and standards for effective cooperation among agricultural organizations.Footnote ⁴⁹

This role included coordinating information about genetic resources, an area that GFAR had not actively targeted but was nonetheless of pressing concern for many participants. The 1992 CBD had made obvious the need for a binding international agreement on plant genetic resources, which eventually emerged as the 2001 International Treaty on Plant Genetic Resources for Food and Agriculture, or Seed Treaty. The Seed Treaty’s power to shape global seed exchange depended on international strategy and consensus, but also on local organizations’ willingness to adopt standards and monitor the movement of plant materials. In consultations over the Seed Treaty, which included the formulation of a Global Plan of Action for the Conservation and Sustainable Utilization of Plant Genetic Resources, CGIAR confirmed its position as a key provider of scientific and technical solutions for genetic resources management. It achieved this, in part, through its promotion of descriptor lists. The lists were already tethered to CGIAR centers’ crop germplasm collections, held “in trust for humanity” by CGIAR on behalf of the FAO Plant Genetic Resources Commission. In 1996, IPGRI extended its crop-specific descriptor work to “multi-crop passport descriptors.”Footnote ⁵⁰ Working in collaboration with FAO, IPGRI sought to produce “consistent coding schemes for a number of key passport descriptors that can be used for all crops,” which it imagined – rightly, as it turned out – would facilitate data exchange across national borders.Footnote ⁵¹

These continued efforts to make descriptors as useful as possible, and as widely used as possible, paid off. A 1997 CGIAR survey of seed and gene bank curators revealed that 80 percent relied on standardized descriptors – and more than two-thirds used IPGRI-produced descriptors.Footnote ⁵² The survey underscored the usefulness of such work on the ground, across dispersed sites and diverse crops.Footnote ⁵³ It also illustrated the willingness of national bodies and regional breeder organizations to adopt IPGRI-produced descriptors as guidelines for crop research management and related trade. IPGRI trumpeted this contribution to international collaboration on plant genetic resources, describing in 1999 that “IPGRI is making it easier for genetic resources workers to document and explore collections as well as to identify promising accessions, through development of crop descriptors.” Descriptors also helped CGIAR carry out its mandated responsibilities for stewarding genetic resources, as they were used in the System-Wide Information Network for Genetic Resources to standardize databases across eleven CGIAR gene banks.Footnote ⁵⁴

The prominence that descriptors acquired during the 1990s resulted from internal and external policies developed at CGIAR as it sought to maintain relevance in a changing institutional and political landscape, as well as the novel technical demands that emerged from new cross-institution programs and international agreements. Initially set up as tools to enable the circulation of crop materials, descriptor lists became a concrete mechanism through which to foster cooperation and exchange among locations and an instrument for the international governance of plant genetic resources. As a largely autonomous entity, IPGRI could act as a reference point for institutions seeking technical standards to ground new forms of cooperation, regulation, and monitoring. By its nature, the scope of descriptor development extended well beyond the CGIAR network, connecting users such as breeders and crop scientists worldwide. It therefore enhanced the visibility and impact of CGIAR, and to some extent made outside researchers dependent on its continued activities. At the turn of the twenty-first century, descriptor lists were central to the global system of germplasm exchange, and CGIAR accrued prominence and legitimacy as their principal creator.

Expanding Scope

In the years leading up to the 2001 Seed Treaty, descriptor lists were established as key tools for the legal and institutional governance of plant genetic resources. At FAO and IPGRI, staff focused on ensuring that descriptor lists would retain their international credibility and be ready to facilitate compliance with the Seed Treaty. Among other things, this meant expanding existing descriptor lists and prioritizing crops included in Annex 1 – that is, the sixty-four crops for which genetic resources would be made available through the less restricted multilateral system.Footnote ⁵⁵ The five years leading up to the Seed Treaty saw the second-highest number of lists ever produced, with a marked drop after 2001 (Table 10.1).

This work intersected with preparations for the formal launch of the multicrop passport descriptors list, also in 2001, which was coordinated by FAO with CGIAR.Footnote ⁵⁶ The technical labor of developing these standards involved their alignment not only with existing and forthcoming descriptor lists but also, in some cases, with regional data management systems. For example, the European Plant Genetic Resources Search Catalogue (EURISCO), which stored passport information on ex situ collections maintained in forty European countries, was developed on the basis of the multicrop passport descriptor standard.

The passport standards came to play a central role in agricultural research, in part thanks to the prominence acquired by genetic technologies and genome sequencing by the turn of the millennium. The promise of precision agriculture, which included a focus on innovation driven by genomic manipulation, directed new attention to transnational information systems. Genomic information could be readily digitized and shared, especially in comparison with the highly diverse and often intractable data linked to plant morphology. Meanwhile, efforts to expedite computerized data exchange were frustrated by the limits of information and communication technologies. The technological focus was therefore less on the general opportunities offered by comprehensive data collection and more on how to exploit new genetic technologies. This arguably led to a shift in the very concept of what constituted a descriptor, with novel descriptor types accepted as significant and complementary to the global circulation of crop germplasm – and, increasingly, the availability of genomic information about such germplasm. In 2004 the Genetic Marker Technologies list was launched, establishing genetic descriptors as important tools alongside those focused on plant morphology.Footnote ⁵⁷

At the same time, the entrenchment of descriptor lists, marker technologies, and related passport standards into global agricultural research and international trade made it ever more evident that decisions about whether and how to include crops in such systems would shape the recognition (or not) of those plants as socially, scientifically, or economically significant. This facet of international standard-setting was heightened by continued lack of agreement over intellectual property rights in plants. Many questions centered on so-called traditional or Indigenous knowledge about plants: whether such knowledge should be captured in databases, and to what extent this was possible given a system centered on traits of relevance to “modern” agriculture and reliant on English as a lingua franca. The 1990s saw ethnobotany rise to new prominence, and ethnobotanical knowledge increasingly featured among potential sources of data for crop scientists.Footnote ⁵⁸ IPGRI in turn developed standards to facilitate communication of contextual information about plants’ lifecycles and uses.Footnote ⁵⁹

This aligned with a larger CGIAR agenda. In 1996, the CGIAR Chairman Ismail Serageldin’s vision for future research emphasized local knowledge: “the CGIAR’s research programs need to be guided … by the need for greater stakeholder participation in the research process. … Indigenous knowledge must be integrated with new science.”Footnote ⁶⁰ Within the realm of descriptor development, this meant new recognition for previously overlooked information. It ultimately led to the 2009 Descriptors for Farmers’ Knowledge of Plants list, which set standards for integrating traditional knowledge into descriptor lists. Here characteristics such as “seed supply system,” “plant uses,” and “market traits” appeared alongside morphological, functional, and environmental ones.Footnote ⁶¹

Meanwhile, IPGRI devoted increased attention to developing descriptor lists in languages other than English. Scarce funding, and the resulting need to focus on the widest possible audiences, meant that additional languages were nonetheless limited to Spanish, French, and Portuguese, thus producing descriptor lists that mapped onto each crop’s colonial heritage (Table 10.2).

Environmental concerns provided an additional impetus to expand the remit of descriptors. The potential impact of climate change on agriculture fostered interest in environmental information, such as data on soil and climate. In addition, a major review of CGIAR in 1998 had recommended refocusing on the environmentally sustainable management of natural resources.Footnote ⁶² This led to a restructuring of CGIAR operations around heritage crops and the role of biodiversity in developing resilient sources of food, and created space for interest in medicinal plants.Footnote ⁶³ A drive to include the health of forests and wildlife within CGIAR’s remit further expanded the focus beyond the usual staple crops.Footnote ⁶⁴ The growing focus on measuring and fostering biodiversity within CGIAR included the rebranding of IPGRI as Bioversity International in 2006 and culminated in the launch of the Biodiversity for Food and Nutrition Project at the Convention on Biological Diversity in 2012. The project, which aimed to identify and promote biodiverse, nutrient-rich plant species, was coordinated by Bioversity and funded by the Global Environment Facility, a trust fund administered by the World Bank and financed by forty donor countries.Footnote ⁶⁵

The expertise and resources devoted by CGIAR to developing descriptors and other data standards sat at the technical epicenter of a global shift towards precision agriculture and environmental stewardship driven by diverse but standardized data about crops, cultures, and climates. At the same time, what should count as a descriptor, and how descriptor lists could and should complement genetic data collection, became more contested as technological opportunities grew. The very expertise employed to provide feedback and input into descriptor lists shifted from the 1990s to early 2000s, with the gradual disappearance of the Crop Advisory Groups once selected by IPGRI to develop the lists, and the emergence of ad hoc, crop-specific collectives whose composition shifted depending on the type of crop and related funders and stakeholders.Footnote ⁶⁶

Bioversity signaled its continuing attention to descriptor lists as a core mechanism for facilitating transnational collaboration on plant genetic resources, including via the Seed Treaty, by launching a survey of the lists’ users in 2006. A part of the “External Review” of Bioversity’s Understanding and Managing Biodiversity program, the survey measured the usefulness of descriptors “in facilitating the establishment and development of databases; improving collaboration and information exchange among organizations; and finalizing the ambitious objective of building a Clearing-House Mechanism to assure a full implementation of the Convention on Biological Diversity.”Footnote ⁶⁷ The results of the survey supported a view of Bioversity descriptor lists as the best-known standard for descriptors in the world, relied on well beyond CGIAR and acclaimed by users as an effective tool for crop data collection and sharing. This spurred further work on multiple descriptor lists, which became the backbone of influential regional and global crop databases, including the Global Information System backed by the International Treaty on Plant Genetic Resources and the FAO/Bioversity List of Multicrop Passport Descriptors.Footnote ⁶⁸

Conclusion

From the founding of CGIAR until the early 2000s, descriptor lists occupied a central place within the network of institutions connected via CGIAR and beyond. Descriptors were a technical solution to facilitate the international exchange of breeding materials and information about them. Over time, descriptor lists became standards essential to the implementation of increasingly stringent mechanisms for the international governance of plant genetic resources. As global agriculture extended its focus from the appropriation of seeds and other plant germplasm materials towards the capture of molecular, environmental, and traditional knowledge about germplasm, descriptors proved essential to aggregating and linking disparate sources of data and relevant biological materials. Descriptor lists were therefore a key means for CGIAR, working especially through IBPGR and its successor institutions, IPGRI and Bioversity, to position itself as a central repository of scientific and technical know-how to sustain both agricultural development and global policy. Even as other closely related elements of CGIAR activities came under political fire, such as its management of seed banks and its environmental and social sustainability, descriptors served as a tool for demonstrating responsiveness to those critiques and willingness to reform.

Early ambitions for universalizing the standards and protocols for describing crops, and recording these descriptions so that all researchers could use and benefit from them, were repeatedly derailed. Although the gap between ambition and achievement could sometimes be traced to the limitations of technology or financial resources, the implementation of universal descriptors was more often stymied by the diversity – of crops, humans, institutions, and goals – encompassed in the international agricultural research community that descriptors sought to discipline.

Over the last decade, developments in digital “Big Data” technologies and curatorial standards have promised to finally encompass such diversity and therefore enable the implementation of descriptors in their original, idealized form without incurring losses, discrimination, or exclusions. One of the most significant recent expressions of this expectation is the GARDIAN database, set up in 2017 to power the CGIAR Big Data Platform that would facilitate – and monitor – the sharing of data across CGIAR centers.Footnote ⁶⁹ In 2021, the Big Data Platform became a key element of CGIAR’s restructuring as “One CGIAR,” further highlighting the scale and ambition of the data integration effort envisaged and its perceived role in coordinating across CGIAR institutes. The digital platform of One CGIAR is meant to include all data produced by CGIAR centers and their collaborators, encompassing crops, pathogens, soil composition, climate, socioeconomic information about farming communities, and more.Footnote ⁷⁰ Crop descriptors are essential to this data linkage system.Footnote ⁷¹ Their continued use defies concerns about the potential implications of such an extensive standardization and testifies to the power of naming standards – and by extension the institutions that control these – within an ever more digitalized system of global agricultural governance.Footnote ⁷²