The structure of multivariate relationships influencing crop yields during the transition to organic management

Background/Question/Methods

Crop yields are influenced by multiple, interacting factors, making it challenging to determine how specific management practices affect agroecosystem productivity.  This is especially true in cropping systems studies, which have the benefit of testing the net effect of realistic management systems, but the weakness of not being able to determine which variable interactions contribute to yield variation. We conducted a feed and forage cropping systems experiment in central Pennsylvania to analyze the effects of initial cover crop and tillage intensity on crop productivity during the transition to organic production. We hypothesized that effects of tillage and cover crops on yield would be mediated by labile soil carbon concentration, weed density, and predatory arthropod activity-density.  We used structural equation modeling (SEM) to attribute yield variation to management effects on these three factors. The 3-year rotation was cover crop/soybean (Glycine max)/corn (Zea mays). Initial cover crop treatments were annual rye (Secale cereale) or timothy/clover (Phleum pratense/Trifolium pratense). Tillage treatments were full tillage (moldboard plow) or reduced tillage (chisel plow). 

Results/Conclusions

Through the application of SEM, we found that tillage intensity and initial cover crop did not have direct effects on crop yields. Instead, their effects on crop yields were mediated by perennial weed density and predatory arthropod activity-density. Perennial weed density and labile soil carbon were both sensitive to tillage practices, but only perennial weed density affected crop yields. In addition, the initial cover crop used in the rotation influenced corn yields in the third year indirectly through a legacy effect on relative predatory arthropod activity-density. Combined, these indirect effects of management practices explained 66% of corn yield variability. Our results highlight the challenges of simultaneously managing for variables that influence yields in the short-term, such as weed density, and potentially at longer time scales, such as labile soil carbon.