6 Conclusions

Farming has a very high potential to exploit multifunctional activities and provide multiple outputs. In particular between farming and rural landscapes a strong joint relation can be found. Any type of farming inevitably has an impact on the landscape. In the past, the skewed remuneration systems in markets made modern farming practices put less emphasis on non-marketable outputs, which were before produced for free (weak disposability). However, because of growing importance of rural areas as consumptive areas, farming should again emphasize this role.

Rather than a bimodal model (Thompson, 2004; Potter and Tilzey, 2005), we see scope for multifunctionality as a unifying concept under which the productive role of agriculture and its role in land management for biodiversity conservation, recreation, water management, climate control and so on can be brought in harmony. By enlarging the definition of competitiveness, production systems efficient in combining several functions can emerge. These can operate either at local markets (both food and non-food), or at international markets. Distinctiveness in markets can be found by unique combinations of resources and values. It is this what is propagated by the term multifunctionality: the fact that specific agricultural systems contribute to distinctive economic, natural, cultural and territorial subsystems. Therefore, we think that by converting the relation and starting the analysis from what makes local rural and farming systems distinctive, we may find clues to build stronger production and food networks which cannot only contribute to rural wealth but bring the existing production system in line with social expectations. In such an approach biodiversity, local identity, cultural heritage and other non-marketable outputs of agriculture become assets with potential value (Marsden et al., 2002*) that should and will be protected.

The above ideas correspond with a lot of thinking on agri-food networks or districts (Goodman, 1999; Murdoch, 2000; Marsden et al., 2002; Bertolini and Giovannetti, 2006; Sonnino, 2007) and also encompasses some of the ideas of neo-institutional economics on micro-institutions (Ménard, 2003, 2004) which argue that in agro-food systems the contractual relations and rules of exchange both horizontally and vertically are important for managing food systems.

To further develop the concept of multifunctionality, we agree therefore with G.A. Wilson (Wilson, 2004; Burton and Wilson, 2006) that there is a need to further theoretically underpin and decompose the concept. But, besides theoretical work (according to the lines sketched in this review) we also think there is a need for empirical research. Four major research lines can hereby be distinguished:

1. More empirical research on evidence about the contributions of agriculture in general, and different farming systems in particular, beyond food and fibre production. As shown in this overview the empirical results are still scarce and not really comprehensive. Such research must start from a clear definition (not confusing with diversification e.g.) and clearly distinguish the contributions of conventional production systems from more specific contributions of alternative production systems (such as more extensive farming systems or organic agriculture). This must allow policy indications on how far a shift to such alternative systems and thus reallocation of support is socially desirable.
2. Research on indicators for multifunctionality: if multifunctionality is accepted to have a normative side, it is important to be able to measure in how far individual farming systems or the farming sector in a region or country is approaching the desired level of multiple outputs. This requires the development of indicators measuring the contributions towards desired outcomes. So far notwithstanding a few exceptions (Wiggering et al., 2006; Mander et al., 2007), such indicators are rare. They may differ from usual sustainability indicators in the sense that they need to emphasize the positive role of farming in society.
3. Research on policy instruments and farm behaviour with respect to multifunctional agriculture. Because the price signal for non-commodity outcomes may be incomplete, it must be corrected by policy instruments to enforce the social desired optimum. Traditional commodity instruments are not adapted to the new outcomes and therefore need to be replaced gradually by newly designed instruments. However, because of the farm and location specific elements of multifunctionality, the individual farm reaction to these instruments becomes a core element as it will be important to know the regional impact of generic but maybe regionally adapted instruments. This requires, as indicated in Buysse et al. (2007) more specific farm modelling and analysis of farm behaviour towards certain policies. Enriching or coupling these models with farm styles research, GIS, multi-agent systems, risk behaviour and so on are promising avenues in this respect.
4. Research on multifunctionality as an asset. As stated before multifunctionality may be regarded as an asset for regional development. Research should show through case studies how such multifunctional farming systems, landscapes or regions can be built on the basis of the assets represented by the joint outputs of these systems. We therefore strongly plea for integrated case study work showing the economic and social value for both individual farms and farming regions of producing non-commodity outputs and to reveal the real value of things such as rural identity and embeddedness both socially and economically. This can evolve in the study of mechanisms to construct, maintain and institutionalise such a multifunctional system.

We conclude with stating that we believe that the conceptual work on multifunctionality in the recent past is only the start for further more theoretical and empirical research to which a lot of disciplines and schools can contribute. Cross fertilisation between the here developed socio-economic frameworks and the insights of e.g. the ecosystem service or landscape based frameworks would certainly further enrich the insights in the concept. We hope that this contribution is a further step in this direction.