Central hypotheses for the third agroforestry paradigm within a common definition.
In the four decades of its existence, agroforestry as a concept has been understood and defined in multiple ways, often referring to a specific system scale of interest. This has ranged from trees, via tree-soil-crop interactions at plot level to the interactions between land, labour, knowledge and investment at farm level. These in turn shape human livelihoods at landscape scale, dynamics of tree cover change in space and time, socio-ecological systems at landscape scale, the multiple value chains that start with tree, crop and livestock production in landscapes, and the policy domains of forestry and agriculture in the context of sustainable development goals, globalizing markets and global climate change. We propose a new, simple definition: "Agroforestry, a contraction of the terms agriculture and forestry, is land use that combines aspects of both, including the agricultural use of trees". This includes trees on farms and in agricultural landscapes, farming in forests and at forest margins and tree-crop production, including cocoa, coffee, rubber and oilpalm. It includes interactions between agriculture and forestry as policy domains. Interactions between trees and other components of agriculture may be important at a range of scales: in fields (where trees and crops are grown together), on farms (where trees may provide fodder for livestock, fuel, food, shelter or income from products including timber) and landscapes (where agricultural and forest land uses combine in determining the provision of ecosystem services). At national and global scales, forestry and agriculture interact ecologically and through policies relating to land use and trade and are important with respect to climate change and other environmental concerns. Agroforestry embraces an agroecological approach emphasising multifunctionality and the management of complex systems and polycultures rather than focusing exclusively on monoculture. At each scale we need clarity on what non-agroforestry comparisons can be made, before any hypotheses on properties of agroforestry can be tested. We discuss current use of agroforestry in terms of "theory of place", describing context, "theory of change", understanding options and choices, and "theory of induced change", articulating how interventions can be evaluated a priori on likely effectiveness, in socio-ecological systems with multiple feedbacks. Effectiveness and tradeoffs are to be evaluated across the full set of sustainable development goals. One overarching hypothesis for each of these three theory domains is proposed, with an indication of the type of evidence that could lead to rejection or at least refinement of the statement. Answers to a coherent set of ten questions can increase the system level understanding of emerging issues, available and new options, and ways to evaluate scenarios and form platforms for change, all in local context. As umbrella term for bringing agriculture and forestry aspects together, agroforestry has an ambitious but important role in current debates.