Modeling growth and composition of perennial tropical forage grasses

In addition to their role as feedstuffs, perennial tropical forage grasses such as bahiagrass (Paspalum notatum Flugge) can play a major role in nutrient management on livestock farms; recycling N from fertilizers and manure to produce feed and reduce the importation of other feeds, while lowering potential N leaching. Balancing feed quality, feed quantity, and nutrient recovery can be difficult. A computer model capable of simulating forage growth, composition, and N-dynamics could be a useful management tool. Farmers and consultants could test management practices virtually, then implement those showing the most promise. Our objective was to develop a tool to predict the growth and composition of bahiagrass that responds to environmental and management inputs.; Bahiagrass sod cores were dug weekly for two 8-week regrowth periods (18 July to 12 September, and 12 September to 7 November). Plants were separated into leaf blades, stem, stolon, and roots. Leaf and canopy photosynthesis were measured in each period. Leaf photosynthetic rate was not different between periods. Leaf and stem growth and rate of development of new leaves was reduced in the second period; however, stolon mass increased dramatically starting in mid-October.; This information aided the development of species-specific parameters required for simulating bahiagrass in the model, CROPGRO. In the process, limitations in the model structure that prevented the prediction of realistic growth patterns were identified. Despite the limitations, prediction of herbage mass was good, having an index of agreement of 0.85, with slightly lower accuracy predicting herbage N concentration.; To address the model's limitations, we modified the CROPGRO source code to include a storage organ (STOR), equivalent to a stolon, and added dormancy functions to increase partitioning of growth to STOR and reduce mobilization from STOR and roots under short daylengths. The freeze-kill function was modified, allowing gradual death of leaves. The Rubisco specificity factor in the leaf-level photosynthesis option was modified for C4 photosynthesis. Model performance was improved, predicting realistic seasonal growth patterns. Excessive N stress was predicted frequently, but the cause was not identified. The forage version of CROPGRO performs realistically but should be tested under cooler temperatures and finer-textured soils.