Growing degree days and cover crop type explain weed biomass in winter cover crops

Abstract

Cover crops are increasingly being adopted to provide multiple ecosystem services, including weed suppression. Understanding what drives weed biomass in cover crops can help growers make the appropriate management decisions to effectively limit weed pressure. In this paper, we use a unique dataset of 1764 measurements from seven cover crop research experiments in Pennsylvania (USA) to predict, for the first time, weed biomass in winter cover crops in the fall and spring. We assessed the following predictors: cover crop biomass in the fall and spring, fall and spring growing degree days between planting and cover crop termination, cover crop type (grass, brassica, legume monocultures, and mixtures), system management (organic, conventional), and tillage before cover crop seeding (no-till, tillage). We used random forests to develop the predictive models and identify the most important variables explaining weed biomass in cover crops. Growing degree days, cover crop type, and cover crop biomass were the most important predictor variables in both the fall (r2 = 0.65) and spring (r2 = 0.47). In the fall, weed biomass increased as accumulated growing degree days increased, which was mainly related to early planting dates. Fall weed biomass was greater in legume and brassica monocultures compared to grass monocultures and mixtures. Cover crop and weed biomass were positively correlated in the fall, as early planting of cover crops led to high cover crop biomass but also to high weed biomass. In contrast, high spring cover crop biomass suppressed weeds, especially as spring growing degree days increased. Grass and brassica monocultures and mixtures were more weed-suppressive than legumes. This study is the first to be able to predict weed biomass in winter cover crops using a random forest approach. Results show that weed suppression by winter cover crops can be enhanced with optimal cover crop species selection and seeding time.