Much of the early literature on predicting adoption of conservation approaches focused on three models: farmer demographics, farm structure, and the diffusion of information process (McCann et al., 1997). Farm structural dimensions and profit maximization frameworks, particularly, have received much attention. As well, traditional diffusion frameworks suggested that influencing attitudes will lead to changes in behaviour and that communicating properly will lead to a rational adoption.
The classic innovation-diffusion models of extension, used to communicate adoption messages for many of these agri-environmental programme designs, have been useful to understand adoption of practices or technologies with a clear commercial benefit (Vanclay and Lawrence, 1994). Unfortunately, they have not worked well for explaining adoption of conservation practices and systems, especially for “preventive” practices and systems for which adoption rates are historically low (Padel, 2001; Deffuant et al., 2002). As well, voluntary adoption is typically low for practices or systems with many off-site benefits that equal or outweigh on-site benefits, unless other incentives are available or farmers already display strong stewardship attitudes (Reimer et al., 2012).
This slow uptake has caused some rethinking amongst adoption theorists interested in sustainability, dating back some 20 years (Buttel et al., 1990; Vanclay and Lawrence, 1994). In fact, in some literatures – e.g., Agricultural Knowledge and Information Systems, Farmer-first and Bottom-up Approaches - adoption and barriers to adoption are considered irrelevant concepts, since change is driven by farmers themselves and their priorities (Vanclay and Lawrence, 1994). In our modelling scenarios, however, adoption is deliberately driven by external positive incentives and is not fully a farmer-led change process, so the barriers-to-adoption approach is pertinent for our framing.
Most earlier normative modelling studies on the transition to organic farming, limited as they were, used static linear programming approaches (Acs et al., 2005). In some work, threshold models have been used, on the presumption that once a threshold number of adopters in a network undertake the innovation, many others will follow (for a review, see Kaufman et al., 2009). Finding this threshold approach wanting, Deffuant et al. (2002) used hybrid computer science / cognitive agent approaches with cellular automata network methods to examine adoption in one French Department. To some extent building on this work, Kaufman et al. (2009) used the Theory of Planned Behaviour, which presumes, in this case, that the degree of intention to convert to organic farming is a reliable indicator of whether they will actually convert. However, the literature on adoption barriers indicates that such a presumption may be overly optimistic and it is necessary to broaden the range of applicable forces and indicators. Lopez-Ridaura et al. (2012) also used ABM (among several modeling techniques) to identify trajectories of conversion for rice production and livestock breeders in the Camargue region of southern France. For the large rice producers, a conversion subsidy of 150 Euros/ha was insufficient to maintain a favourable gross margin in the short term, but this improved in later years. For beef farmers, gross revenues remained relatively constant through conversion.
In a review of studies examining factors influencing farmer adoption of conservation practices (including organic farming), Knowler and Bradshaw (2007) found no consistently compelling relationships but did conclude that social capital could be a significant, and insufficiently examined, variable. Certainly, the role of social networks has received considerable attention recently (Jussaume and Glenna, 2009; Morgan, 2011). Building on other models, within this framework, adoption of new systems is likely to be a product of the relationships and interactions between different actors, human and non-human, institutional, group and individual (Coughenor, 2003).
In this view, innovations of a systemic nature come from both technical knowledge, and the construction of new kinds of networks to develop, guide and support them. The key variables in such an approach are the farm agroecological and socioeconomic setting, the managerial and technical abilities of the farmer, the support network and the policy environment. This newer approach lends itself well to packages of technologies and systems rather than single technologies.
But our understanding of the interplay between many of these variables, and how they influence adoption remains limited, with only a few studies exploring them extensively. According to Knowler and Bradshaw (2007), “consideration of the many factors within an agricultural system that are exogenous to a farm and its operator is arguably still in its infancy in the adoption literature. This is less true with respect to the role of information in the adoption process, but is certainly true with respect to the role of policy and, more significantly, social capital.” This reality has affected our identification of factors to include in the model and explains our later attempts to quantify these dimensions.
