Global waterbodies are experiencing increased risk of eutrophication and harmful algal blooms due to excess nutrients including phosphorus and nitrogen discharged from human activity on the landscape and as a result of climate change. Despite modeling that suggests the efficacy of best management practices in agricultural systems to be sufficient to address the problem, adoption by farmers remains far below the levels needed to achieve significant water quality improvements and new approaches to encourage and sustain adoption are urgently needed. In this work, we apply a modified transtheoretical model (TTM) of behavior change to a longitudinal dataset (N = 584) of farmers' adoption decisions and stated intentions to use cover crops, collected in the Maumee Basin of Lake Erie, USA in 2016 and 2018. The TTM posits that behavior changes over time and is influenced by different social-psychological processes at each stage of change. Our findings confirm past research into the importance of many of the factors investigated, while providing new insight into their role in specific stages of the change process with potential implications for the design of interventions for farmers in different stages. Several factors investigated (mean environmental concern, education, information from conservation groups and off-farm income) were uniquely important to a particular stage. Other factors (response efficacy at the field level, total farm size and risks of spring planting interference) were important at both an earlier and later stage, but less important in predicting middle stages of change. A third set of factors (self-efficacy, proportion rented, no-till adoption and uncertain long-term paybacks) were statistically important across each stage of the TTM model. In applying the TTM longitudinally, we found evidence that farmers in a more advanced stage of cover crop adoption, in the first wave of data collection (2016) were more likely to have adopted cover crops in the second wave (2018), a result not predicted by individual factors alone. We report findings for cover crops but see the potential for the transtheoretical model of behavior change to be applied to other best management practice adoption decisions and to diverse populations of farmers to generate similarly novel insight and utility for intervention design and targeting.
