The second stage of the study was carried out in two steps: validation of compiled information by Centres, and analysis of incoming data.
In mid-July a note was sent by the TAC Secretariat to the 14 Centres currently involved in S&W research, requesting their cooperation in the critical evaluation of our compilation of activities in soils and water-related research from MTPs and P&Bs of their respective Centres; as well as to estimate the proportion of project funds allocated to such fields of research. They were also asked to identify both partner institutions that collaborate with them in carrying out the S&W research agenda and possible trends in soils expertise at their Centres since 1985.
All Centres responded to our request. As expected, the main point of misunderstanding regarded the selection of S&W activities specifically aimed at the maintenance and improvement of the resource base (vis-à-vis those of a productivity-enhancing nature). For example, many Centres proposed germplasm adaptation to soil conditions as a relevant activity for that purpose. However, it was finally decided not to incorporate it, on the grounds that research focused on plant mechanisms and not on those of the soils and water complex. There was a need then to discuss these issues with the majority of Centres, which were willing to review their information and correct estimated allocations, when appropriate. This desk study would have not been possible without their cooperation, which is indeed most appreciated.
In all cases, the total budget figures used in the analysis are those supplied by Centres for activities considered in the study. Their figures are the result of our heuristic judgments on which activities corresponded with the study focus, and how much of the funding was specifically related to S&W research.
Although total project figures were taken as given, the proportion of funds allocated to activities within "mega" projects (e.g., in allocating funds of transnational projects to land types) had to be "guesstimated". For lack of a more rational criterion, in most of the cases we used simple arithmetics (i.e., dividing the total budget by number of activities in the mega project). We do realize that this can lead to over- and underestimates, and hope that they compensate across Centres.
Other "analytical liberties" we had to take concern the allocation of CGIAR research activities to both scales in the systems hierarchy and research approaches, i.e., applied and strategic (see Table 2). We believe allocations on the scale dimension to be straightforward. The debatable point could be the placing of 2.3, about maintenance and improvement of the resource base, at the "field" level. Although it is recognized that it could also be placed at higher levels in the hierarchy, the main reason for putting it there was that cropping systems prevail as a focus for such activities in the Centres.
Two points should be made about the approach dimension. First, in slotting 1.2.1 under the strategic approach, emphasis was given to its more dynamic component in the CGIAR description, that of "monitoring sustainability developments". As a consequence of this emphasis, projects focusing on the more descriptive part of characterization studies were shifted to 1.2.2, farming systems. The second point refers to placing 4.3, public management systems, under strategic research. It is based on the role of such institutions in understanding technology adoption processes, a central issue in the "management" of natural resources. Policies certainly influence such processes, but they are not seen as contributing as much to research on the prediction of behaviourial patterns as local institutions do.
Finally, a note on the land types. As it was not possible to accommodate all the types used by the different Centres, land types sharing relevant characteristics were combined within one of the eight selected. Relevant cases are those of the savannas and the rainfed lowlands. The former type includes both the less-acid savannas of western and central Africa, and the acid savannas of Latin America. It also includes half of the resources allocated by Centres to inland valley projects. The rainfed lowlands include all three types of "lowlands" (flood prone, lowlands and uplands), as well as the other half of the inland valleys.
