True partnerships are vital for linking 'upstream' innovation to 'downstream' uses, says CGIAR's Jean-Marcel Ribaut.
Agricultural research for development spans a broad spectrum of activities - from 'upstream' research, generally at universities or advanced research institutes, to much more 'downstream' research by plant breeders to put better crops in farmers' hands.
As a result of this spread, activities can become fragmented, with little communication between specialised teams along the research & development (R&D) chain. This is often counterproductive, especially when researchers stretch beyond their area of expertise.
In addition, broader and more diverse research portfolios often compromise efficiency and create unhealthy competition for funding. And resulting research projects may never turn into products that improve farm productivity.
True and effective partnerships - connecting the right people from complementary teams - is one obvious way to improve R&D effectiveness.
Linking innovation to application
Linking innovations in upstream research to downstream applications of that research - often referred to as translational science or translational biology - is not a new challenge. But in the field of crop improvement at least, it is easier said than done.
It is important to find the right people and teams, and to have adequate financial and human resources to manage partnerships effectively - keeping everybody moving in the same direction and maximising synergy among teams while keeping an eye on research quality and information sharing.
The CGIAR Challenge Programs are one model of how such partnerships can work effectively. One of these programmes is the Generation Challenge Programme, which I direct: a global consortium of crop research institutions that, within a decade, aimed to demonstrate that applying modern biology and harnessing plant genetic diversity can create crop varieties that meet the needs of resource-poor farmers.
To support collaborative initiatives by independent research teams, the programme's first five-year phase (2004-2008) featured several calls for grant bids. This created healthy competition as winners were selected based on well-defined criteria. A key criterion was 'integrative partnership': each project needed to have at least one partner from an advanced research institute, a CGIAR centre and a developing-country research and/or educational institution.
Through the competitive process, the programme also encouraged the active involvement of developing-country partners, favouring the sustainable adoption of our products and disqualifying what we called 'alibi' partnerships - typically where developing-country participation is reduced to a limited activity and budget. Each project also had to include a capacity-building component.
Refocusing research agendas
In midlife, the programme refocused its research agenda by reducing the number of target crops and countries based on criteria such as the potentially valuable genetic and genomic material available for a given crop, and avoiding duplication with other initiatives.
The second phase (2009-2014) essentially commissioned projects to build on promising outputs from the first phase, translating them into tangible products for our primary users, who are mainly developing-country plant breeders. This was achieved with a progressive shift in budget and leadership from advanced research institutes and CGIAR through direct grants to the local breeders' institutes.
A sorghum initiative that involves scientists from three regions (North and South America and Africa) is an example of such a shift.
First, the researchers, led by a senior scientist from the US Department of Agriculture based at Cornell University in collaboration with a visiting scientist from Brazil's Embrapa research corporation, cloned a gene that confers tolerance to aluminium toxicity in sorghum. Through another Brazil-led competitive project (in the first phase), the researchers then screened a diverse set of genetic material from different families of plants to identify those most favourable for sorghum performance in aluminium-toxic acid soils.
As a third step, building on the successful partnership and outputs generated through the two competitive grants, the programme commissioned a collaborative second-phase project between African breeders and the Brazilian scientists who led the second competitive project. The goal was to transfer the favourable Brazilian genes into African sorghum varieties in Kenya and Niger.
So the decade-long, multi-partner, cross-country relay, starting with a lab experiment in the United States, is expected to lead to products for African farmers within the next two to three years, with a South American stopover in between.
This is, of course, one of our best stories in enabling successful collaboration along the R&D chain. Naturally, there were also some failures - especially when we were too prescriptive in nominating partners we thought were theoretically well-suited for a given activity. We also had projects that generated outputs that turned out to be inappropriate for transfer down the delivery chain.
We learnt that the 'chemistry' between partners - the fact that they appreciate each other, building progressively on previous collaborations - is probably just as important, if not more important, than complementary skills that look promising on paper.
So, overall, the jewel in the crown of our achievements is actually an intangible spirit and culture based on mutual trust, respect and a genuine desire to complement the work - and to tap the skills - of other partners. This was affirmed by a recent external review, which highly rated the programme's partnerships approach. 
There is no perfect recipe for fostering this spirit. Building promising projects into focused, commissioned work is a key ingredient, as is a constant desire to help and, when possible, pass leadership on to the next player in the delivery chain.
I must add a necessary note of caution: this model can work only if it builds on strong and well-established institutions, and as a complement to core activities.
Another key element of success is identifying specific research objectives that can be achieved in a given time frame. Our experience also suggests that the benefits of having an independent management team outstrip the cost it entails.
A key challenge in true partnership is to strike the right balance between management that serves the programme as a whole and creating ownership so all partners can nurture a network spirit.
Jean-Marcel Ribaut is director of the CGIAR Generation Challenge Programme, a plant-breeding partnership network hosted at CIMMYT's headquarters in Mexico. He can be contacted at firstname.lastname@example.org
 Paramjit S. Sachdeva and others Report of the final external review of the Generation Challenge Programme (CGIAR, April 2014)