Corn earworm, Helicoverpa zea Boddie (Lepidoptera: Noctuidae), is a common herbivore that causes economic damage to agronomic and specialty crops across North America. The interannual abundance of H. zea is closely linked to climactic variables that influence overwintering survival, as well as within-season host plant availability that drives generational population increases. Although the abiotic and biotic drivers of H. zea populations have been well documented, prior temporal H. zea modeling studies have largely focused on mechanistic/simulation approaches, long term distribution characterization, or degree day-based phenology within the growing season. While these modeling approaches provide insight into H. zea population ecology, growers remain interested in approaches that forecast the interannual magnitude of moth flights which is a key knowledge gap limiting early warning before crops are planted. Our study used trap data from 48 site-by-year combinations distributed across North Carolina between 2008 and 2021 to forecast H. zea abundance in advance of the growing season. To do this, meteorological data from weather stations were combined with crop and soil data to create predictor variables for a random forest H. zea forecasting model. Overall model performance was strong (R2 = 0.92, RMSE = 350) and demonstrates a first step toward development of contemporary model-based forecasting tools that enable proactive approaches in support of integrated pest management plans. Similar methods could be applied at a larger spatial extent by leveraging national gridded climate and crop data paired with trap counts to expand forecasting models throughout the H. zea overwintering range.

Tropical savannah landscapes are faced with high soil degradation due to climate change and variability coupled with anthropogenic factors. However, the spatiotemporal dynamics of this is not sufficiently understood particularly, in the tropical savannah contexts. Using the Wa municipality of Ghana as a case, we applied the Revised Universal Soil Loss Equation (RUSLE) model to predict the potential and actual soil erosion risk for 1990 and 2020. Rainfall, soil, topography and land cover data were used as the input parameters. The rate of predicted potential erosion was in the range of 0-111 t ha 1yr 1 and 0-83 t ha 1yr 1 for the years 1990 and 2020, respectively. The prediction for the rate of potential soil erosion risk was generally higher than the actual estimated soil erosion risk which ranges from 0 to 59 t ha 1yr 1 in 1990 and 0 to 58 t ha 1yr 1 in 2020. The open savannah areas accounted for 75.8 % and 73.2 % of the total soil loss in 1990 and 2020, respectively. The validity of the result was tested using in situ data from a 2 km2 each of closed savannah, open savannah and settlement area. By statistical correlation, the predicted soil erosion risk by the model corresponds to the spatial extent of erosion damages measured in the selected area for the validation. Primarily, areas with steep slopes, particularly within settlement, were identified to have the highest erosion risk. These findings underscore the importance of vegetation cover and effective management practices in preventing soil erosion. The results are useful for inferences towards the development and implementation of sustainable soil conservation practice in landscapes with similar attributes.