At the end of the same century, jungle rubber is left as the main reservoir of lowland forest biodiversity in the plains of Sumatra and probably West Kalimantan. Because of its nature as a complex agroforestry system, biodiversity and structure of these rubber gardens are similar to old secondary forest. The impact of rubber garden selective cleaning on biodiversity and composition is assessed by comparing them to undisturbed secondary vegetation. Results show that regularly cleaned plots have a distinctly lower biodiversity than not cleaned ones. The management intensity decides upon diversity and species composition of agroforestry systems. Several types of Rubber Agroforestry Systems (RAS) are currently being experimented in an on-farm trial network by the SRAP projects. RAS integrates biodiversity with improved cropping system through adaptation to farmers' strategies for land use and ressource management. It is now recognized that local farmers did not cause the majority of the fires, because properly used, fire in traditional shifting cultivation does not spread. The farmers actually, are interested in decreasing shifting cultivation (for the purpose of upland rice production) and are putting emphasis on more intensified and more productive systems based on tree-crops that require less land, new altematives such as RAS which produce a far more interesting return-to-labour and reduce risks. Improved. RAS increase farm income while requiring less land per household and support the development from shifting agriculture to a permanent tree crop based agriculture. The role of rubber agroforestry as an interesting alternative to other systems are also discussed in this paper, as well as the potential for RAS systems to rehabilitate Imperata grassland