3.1 Mean adaptation inclination results
While the coefficients for the pathways predicting mean adaptation inclination are slightly different across all 5 models due to the focal behavior not being included in the mean inclination metric, they are very similar to one another. As a result, we just present the model used as a control for cover crops as it is archetypical of the other models (Fig. 3). This model includes, but does not show, pathways from all variables to cover crop intentions.
Overall the model fits extremely well (X2(13) = 14.401, p = 0.346; RMSEA = 0.012, p = 0.941 [0.000-0.039], CFI = 0.998, n = 767). The coefficients for the pathways suggest that both experience with climate change and anthropogenic climate belief are positively associated with concern about future weather conditions. Consistent with hypothesis 1.1, the pathway from weather concern to mean adaptation inclination is positive and significant. Consistent with hypothesis 1.2, the indirect effect of both experience with climate change and anthropogenic climate belief are positive and significant. This suggests that mean adaptation inclination is a function of concern about future weather conditions, and that the impact of negative experiences and climate beliefs on concern carries though in creating mean adaptation inclinations. With respect to Hypothesis 1.2 it is worth noting that the total effect of belief in anthropogenic climate change is not significant. This suggests that while we can trace the impact of belief in anthropogenic climate change on mean adaptation inclination through its positive impact on concern, it is also having a negative impact through some other unobserved mechanism, rendering the overall effect essentially 0. Further, it is also worth noting that there is a positive, significant direct effect of negative experience with climate change on mean adaptation inclination. This suggests that concern about future weather events is not entirely sufficient to explain the impact of negative experiences from climate change on intentions to adapt. As such, while we have some support for hypothesis 1.2, it is tempered by these additional findings.
We also see positive effects from conservation identity (B = 0.159, p < 0.0005), education (B = 0.162, p < 0.0005), planned succession to a family member (B = 0.062, p < 0.05), farm size (B = 0.251, p < 0.0005), and extent of insurance coverage (B = 0.127, p < 0.0005), as well a negative effect of age (B=-0.151, p < 0.0005). This suggests that, in general, younger, more educated farmers with stronger conservation identities and larger, more extensively insured farms that they plan to pass on to a family member, tend to have stronger adaptation inclinations when considering all the adaptations together.
3.2 Specific adaptation inclination results
For all subsequent models in this section, the coefficients for the relationships predicting weather concern will be identical. However, because the impact of concern about future weather conditions has different impacts on each specific adaptation inclination, the coefficients for the direct and total effects of both negative experiences with climate change and belief that climate change in human caused will be different between models. Also note that the total effects for experience with climate change and anthropogenic climate belief do not include the effect that these variables have through mean adaptation inclination. This portion of the total effect was omitted to highlight when the climate and weather variables have an impact on a specific adaptation inclination that cannot be accounted for by its impact on mean inclinations.
Cover crops. The model fit statistics show excellent fit and were provided previously. As noted earlier, while the model of cover crop adaptation inclination (Fig. 4a) includes pathways from all the other variables to the mean adaptation inclination, they are not shown in this figure (for reference to the values of those pathways see Fig. 3 above). In contrast to the pathways predicting mean adaptation inclination, here we see no significant impact of the weather and climate beliefs or demographics on the inclination to use cover crops, partially supporting hypothesis 1.3. Rather, it is the identity variables that appear to distinguish cover crops from the other adaptations. Specifically, we see a strong positive effect of conservation identity (B = 0.274, p < 0.0005), supporting hypothesis 2.1a. Similarly, we see a strong negative effect of productivist identity (B=-0.152, p < 0.01), supporting hypothesis 2.2c. In addition, we see a further negative effect of age (B=-0.103, p < 0.01) beyond the influence that age exerts on adaptation in general. Finally, we see significant negative effects for the Sugar (B=-0.161, p < 0.0005), Macoupin (B=-0.194, p < 0.0005) and Upper Fox (B=-0.177, p < 0.0005). This model suggests that, on average, a farmer is more likely to choose cover crops as an adaptation strategy if they identify more strongly as a conservationist and less strongly as a productivist, if they are younger, and if they live in the Lower Maumee (reflecting something unique about that local context).
Filter strips. In the model explaining the inclination to use filter strips (Fig. 4b), the fit remains excellent (X2(13) = 15.120, p = 0.300; RMSEA = 0.015, p = 0.993 [0.000-0.040], CFI = 0.998, n = 0.767). The significant pathways broadly mirror the relationships observed in the model of cover crops, with the identity variables carrying most of the influence beyond the impact of the mean adaptation inclination. Specifically, both conservation identity (B = 0.225, p < 0.0005) and productivist identity (B=-0.098, p < 0.01) significantly impact inclinations to use filter strips, supporting hypothesis 2.1b and 2.2d. There are also no significant effects of the climate and weather beliefs beyond the influence on mean adaptation inclinations, further supporting hypothesis H1.3. We also see a significant positive effect of succession (B = 0.063, p < 0.05). However, unlike cover crops, we see a significant negative effect of education (b=-0.069, p < 0.05). Finally, we see significant effects for all the watershed variables suggesting that those in the Sugar (B=-0.133, p < 0.0005), Maple (B=-0.079, p < 0.05), Macoupin (B=-0.118, p < 0.0005) and Upper Fox (B=-0.188, p < 0.0005) are all less inclined than those in the Lower Maumee to install filter strips as a form of adaptation. This model suggests that, on average, a farmer is more likely to choose filter strips as an adaptation strategy if they identify more strongly as a conservationist and less strongly as a productivist, if they are less educated, if they plan on passing their farm on to someone in their family when they retire, and if they live in the Lower Maumee (reflecting something unique about that local context).
Using additional fertilizer. In the model explaining the inclination to use additional fertilizer (Fig. 5), the fit remains excellent (X2(13) = 15.029, p = 0.306; RMSEA = 0.014, p = 0.994 [0.000-0.040], CFI = 0.998, n = 767). Similar to the previous models, we see significant effects from the identity variables and not from the climate and weather beliefs, only in this instance, the signs are reversed with a positive effect from productivist identity (B = 0.161, p < 0.0005) and a negative effect of conservation identity (B=-0.176, p < 0.0005), supporting hypotheses 1.3, 2.1d and 2.2b. We also see a significant positive effect for age (B = 0.076, p < 0.05) and a marginal positive effect for farm size (B = 0.068, p < 0.01) and off-farm income (B = 0.068, p < 0.01). We also see significant negative effects for the number of acres in predominantly loam soil (B = 0.101, p < 0.05) and the extent of insurance coverage (B=-0.091, p < 0.05). Finally, we see significant positive effects of all 4 watershed variables suggesting that those in the Sugar (B = 0.228, p < 0.0005), Maple (B = 0.197, p < 0.0005), Macoupin (B = 0.265, p < 0.005) and Upper Fox (B = 0.236, p < 0.0005) watersheds are all more likely to use additional fertilizer than the Lower Maumee. This model suggests that, on average, a farmer is more likely to choose additional fertilizer as an adaptation strategy if they identify more strongly as a productivist and less strongly as a conservationist, are older, have larger farms, have off-farm income, have less acres in loam soil, have less insurance coverage and do not live in the Lower Maumee (reflecting something unique about that local context).
Installing additional tile drainage. In the model explaining the inclination to install additional tile drainage (Fig. 5), the fit remains excellent (X2(13) = 15.552, p = 0.274; RMSEA = 0.016, p = 0.992, CFI = 0.997, n = 770). In contrast to the previous models, for tile drainage we see no significant effect of identity (failing to support H2.1c and H2.2a, though there are small positive indirect effects of both through weather concern). However, we do see a significant effect of weather concern (B = 0.108, p < 0.01) on inclination to install additional tile drainage beyond the impact of mean adaptation inclinations (failing to support H1.3). We also see significant indirect effects of both negative experience with climate change (B = 0.061, p < 0.05) and belief in anthropogenic climate change (B = 0.012, p < 0.05), though neither total effect is significant suggesting that both beliefs may also be contributing to reducing intentions through an unobserved mechanism. We also see significant negative effects of age (B=-0.086, p < 0.05) and the proportion of income derived from off-farm sources (B=-0.063, p < 0.01), as well as a positive effect of farm size (B = 0.128, p < 0.05). Finally, we see significant negative effects for the Macoupin (B=-0.175, p < 0.0005) and Upper Fox (B = 0.150, p < 0.0005) watersheds. This model suggests that, on average, a farmer is more likely to choose tile drainage as an adaptation strategy if they have greater concern about future weather conditions (driven by beliefs in climate change and past negative experiences), larger farms, and when they are younger, have less off-farm income, and live in the Lower Maumee (versus the Macoupin or Upper Fox).
Renting out the farm. In the model explaining the inclination to rent out the farm (Fig. 6), the fit remains excellent (X2(13) = 15.872, p = 0.256; RMSEA = 0.017, p = 0.991 [0.000-0.042], CFI = 0.997, n = 764). Here we see the weakest effect of mean adaptation inclination (B = 0.181, p < 0.001) suggesting that renting out the farm may be the most dissimilar to the other adaptations. We do not see a significant effect of concern or identity, supporting H1.3, but failing to support H2.1e or H2.2e. We do see a significant positive effect of age (B = 0.254, p < 0.001) and belief in anthropogenic climate change (B = 0.083, p < 0.05), as well as negative effects of succession (B=-0.135, p < 0.001), farm size (B=-0.143, p < 0.001) and the extent of insurance coverage (B=-0.104, p < 0.01). We also see significant positive effects for the Sugar (B = 0.134, p < 0.001) and Maple (B = 0.150, p < 0.001) watersheds. This model suggests that, on average, a farmer is more likely to rent out the farm as an adaptation strategy when they are older, believe in anthropogenic climate change, do not have a succession plan, and live in the Sugar or Maple.