AKST and current agricultural development models have contributed to increasing inequality and the exclusion of indigenous and tribal peoples.
     In LAC and parts of ESAP the selective perception of production requirements and exclusion of or limited atten­tion given to certain agroecosystems, such as dryland agri­culture, coastal fisheries, mountain ecosystems, and pastoral systems, worsens the inequality already compounded by local exploitation, rent seeking and corruption, appropria­tion of resources of the poor—especially common pool re­sources—and social prejudices like caste and gender biases. The challenge for development policy and AKST is to de­velop agricultural and food systems that can reduce income inequalities and ensure fair access to production inputs and knowledge to all. Governments and international donors are now beginning to invest in long-term commitments to AKST integrated into pro-poor development policies.

AKST—Current constraints, challenges and opportunities
More than five decades after formal AKST made its entry into almost all countries, the explicit economic and political legitimization of investments in AKST remains food security, livelihoods and poverty reduction in developing countries, and trade and environmental sustainability in industrialized countries. While the development models-poverty-environ­mental degradation nexus is evident in different forms in different countries, the formal AKST apparatus available to address these variations is the same in structure, content and the conduct of science in almost all countries. The AKST apparatus tends to focus on mainstream, input-intensive, irrigated monocropping systems—mainly cereals, livestock and other trade-oriented commodities, to the relative ne­glect of arid/dryland agriculture, mountain ecosystems, and other non-mainstream production systems that have been discussed above. It is important to recognize that this con­straint, more or less universal in formal AKST is not inci­dental, but part of an overall development model in which scientific knowledge is institutionalized in its utilitarian role. Resources are allocated to production systems that can show the highest economic returns to crop/commodity pro­ductivity. The capacity of AKST to address the challenges of poverty, livelihoods, health and nutrition, and environmen­tal quality is conditioned by its capacity to address its own internal constraints and challenges.
     Organized AKST in the form of public sector R&D, extension and agricultural education across world regions, are based upon a linear top-down flow of technologies and information from scientific research to adopters. Despite increasing polarization of the debate on new technologies, especially biotechnology and transgenics, and years of well-published knowledge on differential access to technologies and appropriate institutional arrangements, formal AKST has yet to address the question of democratic technology choice. AKST as currently organized in public and private sector does little to interact with academic initiatives in ba­sic biological, ecological and social sciences to design rules, norms and legal systems for market-oriented innovation and demand-led technology generation, access and use appro­priate for meeting development and sustainability goals.
     There is a significant volume of literature from all the

regions on the high rates of return per unit of investment in agricultural R&D, especially in crops and in farming sys­tems that have been the focus of the AKST apparatus. Some of the conditioning factors for high rates of return lie out­side agriculture and AKST in complementary investments such as rural infrastructure or microcredit units that reduce market transaction costs or provide appropriate institutions or norms. A rate of return analysis is insufficient for captur­ing returns to investment that meet development and sus­tainability goals; other economic and social science methods are needed for this task.
     Declining investments in formal AKST by international donors and a number of national governments is causing concern among the developed and developing countries. Public investments in agricultural R&D continue to grow although rates have declined during the 1990s. In many industrialized countries investment has stalled or declined, while in ESAP countries investments have grown relative to other regions (annual growth rate of 3.9% in the 1990s). As a result, ESAP accounts for an increasing share of global public R&D investment, from 20% in 1981 to 33% in 2000. In contrast to the 1980s, the annual growth rate of total spending in SSA decreased in the 1990s from 1.3 to 0.8%. A disturbing trend in 26 SSA countries for which time series data are available is that the public sector spent less on agricultural R&D in 2000 than a decade earlier. Globally public sector R&D is becoming increasingly concentrated in a handful of countries. Among the rich countries, just two, the USA and Japan, accounted for 54% of public spending in 2000, and three developing countries, China, India and Brazil, accounted for 47% of the developing world's public agricultural research expenditures. Meanwhile, only 6% of the agricultural R&D investments worldwide were spent in 80 mostly low-income countries whose combined popula­tion in 2000 was more than 600 million people.
     In the industrialized countries investment by the private sector has increased and is now higher than total public sector investments. In contrast, private sector investment in developing countries is small and will likely remain so given weak funding incentives for private research. In 2000, private firms invested only 6% of total spending in the de­veloping world, of which more than half was invested in ESAP. Private investment in AKST is, and is likely to re­main, largely confined to appropriable technologies, with intellectual property protection, which can earn significant revenues in the market.
     Currently AKST actors and organizations are not suf­ficiently able to deal with the challenges ahead because of the focus on too narrow a set of output goals. The current knowledge infrastructure, which is oriented toward these goals, historically has largely excluded ecological, environ­mental, local and traditional knowledges and the social sciences. AKST infrastructure will need to encompass and work with this much broader set of understanding and data if AKST challenges are to be met. The knowledge infrastruc­ture of AKST is closely allied with particular branches of economics appropriate for meeting production goals, but to the relative neglect of other capacities in the economic sci­ences that are needed to meet AKST challenges.
     Meeting the challenges will require a different organi­zational framework than currently exists in fundamental