Some of the trainees eventually did adopt them in their regular research work

As importantly, cotton breeding looks beyond the farm to the post-harvest scales of industrial processing and global trade – the so-called technological or industrial parameters, such as ginning out turn or quality of fiber . According to my interlocutors, cotton is not that complicated a plant to breed. In fact, the commercial varieties found on both sides of the Southern Atlantic are “kin”, ultimately coming from a common genetic poll. Although both South America and the African continent are origin centers of cotton species, the ancestors of most cotton grown in both places today came from the United States.Like so many other elements in the cotton production systems in both Brazil and West Africa, this common ancestry comes from a shared scale in the industrial and global origins of modern cotton remarked above. In fact, the scientific breeding of cotton varieties predated by a few decades the mid-twentieth century Green Revolution and its focus on germplasm exchange and variety improvement for high yields. U.S. upland cotton was introduced in Africa during colonial times for a practical reason: by the late nineteenth century, American cotton was the world standard according to which European spinning machinery was normally calibrated . Much of the work undertaken in the colonial research institutes and botanic gardens consisted in adapting these foreign crop varieties to the environmental conditions found in Sub-Saharan Africa, on the one hand, and to the demands and standards required by the global textile industry, on the other.

After independence, breeding remained a privileged research field in the national institutes,bato bucket which today develop and distribute their own improved cotton seeds to farmers. But in colonial times as today, to develop new varieties “drawing on the best of imported and local cotton” was not enough. Improved seeds required that rural extension programs be simultaneously implemented in order to change farmers’ traditional practices into “‘modern’ plowing, planting, and harvesting skills” . This is because, failing to transform the context into which improved varieties were to be introduced, these seeds would not be able to actualize their full productive potential: the more technified the seed, the tighter and more specialized has to be the socio-technical network sustaining it. This project component therefore faced the same basic challenges of the other two, concerning the availability of proper agricultural inputs, labor, and rains in peasant fields. But while the others, especially no-till, could be transformed more rapidly according to this new context, conventional improved seeds can only be changed through cross-breeding, which is an extremely slow, statistically oriented process. As with the other project components, in breeding the enrollment of human and nonhuman actants also went beyond cotton itself. Its annual capacity-building cycle also targeted breeders working with other plants commonly intercropped with cotton locally, such as sorghum, maize or rice. They were generally well trained and had good basic technical knowledge; several of them, especially the senior ones, had received college and graduate training abroad, in France and other European countries, the United States, or the former Soviet Union.

More than the other two project components, the breeding workshops dwelled largely on abstract technical topics such as genetic resource conservation, genic flow, and quantitative genetics – the more advanced stages of the latter, I must say, I had difficulty following up, even with the rudiments of this field I had learned during a cassava project with Ghana the previous year. Even if, as the Brazilian researchers themselves underscored, breeding is both a science and an art which requires an “eye” that can only be cultivated through experience, technical knowledge in this field required, more than in the others, increasingly specialized and sophisticated material apparatuses in order to be put to work. This was notably the case of statistical software: it was the only workshop where trainees were required to bring along an artifactual attachment, a laptop computer . In 2011, much of the trainings involved teaching how to use the new software: two open-source statistical programs that had been reverse engineered at a Brazilian university. Similarly, breeding was the component for which the African partners made most demands for material infrastructure, especially in biotechnological fields such as marker-assisted breeding. Many of the Brazilian researchers, however, were of the opinion that the high financial investment required for setting up and maintaining state-of-the-art biotechnology facilities generally did not yield worthy returns in terms of actual applications; in their own experience, it was cheaper and easier to outsource certain tasks to other labs.

Still, the project did accede to some of these demands: as Phase I was coming to an end in 2013, a biotechnology lab was being finalized in Sotuba, and a local researcher was to be sent to Brazil for training on how to run it. The rationale for the project’s decision-making in this case, as I understood it, was that in relatively resource-poor institutes such as the ones in the C-4 countries, the point of a lab like this was not necessarily to produce immediate, cost-beneficial results. It would, above all, fulfill a pedagogic function of supporting training and learning of junior researchers. For senior researchers, it would enable the practical enactment of a kind of advanced knowledge that would remain otherwise idle. In this sense, it would also have a sort of “prosthetic” effect: after being trained overseas, the knowledge acquired at the “centers” would go to waste without an appropriate material base through which it could be put work in their home institutes. More than the other project components, thus, breeding brought to the fore, at the scale of research institutes, an asymmetry similar to the one observed between research recommendations and peasant practices: that is, between available knowledge and the artifactual network required to put it in practice appropriately. Besides biotechnology, another sub-field where this appeared quite clearly in the project was genetic resources conservation.African institutes were generally limited to forms of conservation that require less “de-contextualization” , such as in farm, in situ, or ex situ. The cotton germplasm bank in Farako-ba was of this latter type, and included a range of creole cotton plants found across the country: some were as tall and lean as trees, others were low and bushy; some had light yellow flowers, others had dark pink or purple ones . After many breeding cycles over the course of decades, diversity of traits becomes narrower and related to hardly perceptible variables such as pelosity of the stem, length of fiber, relative capacity to absorb certain nutrients, or degree of yellowness. Even among breeders, increasing mediation by artifacts becomes necessary so that the precise difference in performance between the varieties can be ascertained: scales to weight harvest samples and statistically estimate productivity, soil and leaf analysis to calculate nutrient absorption,dutch bucket hydroponic or highly specialized equipment to measure the various technological parameters for quality of fiber. The African institutes had some of these artifacts, but not others, and rarely did they have their latest versions according to global standards. Moreover, none of them had fully functioning facilities for the more de-contextualizing forms of germplasm conservation such as cold chamber or in vitro tissue culture. Not surprisingly, this was part of the demands posed by the project partners, and again, they were partly fulfilled: a cold chamber was being built as part of the project’s central facilities in Sotuba, but not in the other institutes. Finally, the African partners showed significant interest in the possibilities for germplasm exchange opened up by the project.

What they sought in the Brazilian varieties was not the plants as such, but certain traits that could potentially ameliorate the varieties created by them and their predecessors in the national institutes. The introduction of Brazilian cotton varieties would provide for an always-welcome enlargement of the genetic pool available to the institutes’ cotton programs and their breeders. “Our cotton has good quality fiber, better even than the Brazilian, and it’s longer”, one of the C-4 breeders explained to me. “It’s also better adapted to our environment. But the Brazilian varieties have very good productivity”, and some had whiter fiber. For the Brazilians, quality of fiber and resistance to drought were traits of potential interest in the African varieties. This was an exchange between peer researchers; in spite of the narratives on resource plundering and biopiracy often associated with the global flow of genetic resources, as in bioprospection they do not necessarily lead to commercially viable products: “it’s always good to have this kind of germplasm around, just in case”, one of the Embrapa breeders put it plainly. Having being de-contextualized from its original assemblage in Brazil, relations between germplasm and context had to be remade anew at the arrival point: this was precisely what the breeding component’s adaptation experiments were about, as will be seen in the next chapter. But to take validated Brazilian varieties or cross-bred hybrids outside of the C-4 research institutes will be a whole other story. Rarely does one single variety carry all desirable traits, and systematic cross-breeding is a lengthy process: ten or more years may pass before a stable variety can be ready to be transferred to farmers. Aside from such technical issues, there are also explicitly political ones: the direct transfer of Brazilian varieties to African farmers as such will not necessarily be of interest to all local actors, and would call for an altogether different level of legal discussion in terms of cultivar protection and licensing procedures. This is one of the few areas where a potential remains for economic gain by public agronomic research institutes, rendered possible by the worldwide expansion of legal frameworks enforcing intellectual property rights . This makes breeding a more sensitive field than the other two, something that also concerns the ethnographer: indeed, at points some of my interlocutors seemed more worried about the identity of the plants than their human counterparts. As others have also remarked , in the age of IPR, anonymity seems to be as much an ethical matter for non-humans as for humans.Early on during fieldwork, in the CECAT course on seed production in 2010, I had learned that the seed is one of the most effective forms of technology transfer. That one small grain contains years, often decades, of research efforts by breeders and other experts that can, in such condensed and disembedded form, be taken to farmers with relative ease. A year later, in the C-4 project, I learned that to unravel all that knowledge and technology again out of the tiny seed once it gets to its new destination implies launching a whole other – reverse, if you will – translation chain in a socio-technical assemblage that will necessarily differ from the original. This difference between contexts, which seemed minor at the level of official discourse because based on analogies that took them for granted as shared preexisting backgrounds for relations, came to the fore at the front line where the researches operated; in fact, it became the very “stuff” on which they worked. How the new context differs from the original, and with what consequences for technology transfer, is not something that is given previously to the relation, but that is actively made by those who work across the new interface. Cooperation front liners from both sides of the Southern Atlantic were, in this sense, brokers; but rather than brokering flows of material and symbolic resources through a multi-layered social network, as normally emphasized by the literature on development , they brokered the making of social and natural contexts for the network itself. Technology transfer in this case turned out to be less about “rendering technical” through planned intervention than about demonstration, collaborative context-making and scaling operations, co-production between technology and context, and, as will be seen in greater detail in the next chapter, attention to the controls available to actors at various scales. These characteristics stem not just from South-South principles of horizontality or demand-drivenness, but from the organizational and practical conditions of Brazilian cooperation remarked in Chapter 1. Here, the technologies being transferred did not appear as “a blueprint for an ongoing reorganization of farming so that the latter corresponds with the assumptions and requirements built into the technological design” .