Seasonal changes in tree--grass complementarity and competition for water in a subhumid tropical silvopastoral system

Previous studies have shown that silvopastoral systems are sometimes unsuitable in the subhumid tropics because strong competition for water between species reduces grass growth and production. The work described in this paper was carried out in order to analyse the microclimate inside a silvopastoral system and to assess how it modifies the evaporative demand and soil water availability. The study was conducted on plots of 11-year leguminous tree (Gliricidia sepium (Jacq.) Walp.)-grass (Dichanthium aristatum (Poir.) C.E. Hubbard) association (SS treatment), and of an open grassland with the grass only (OG treatment), cultivated on a vertisol. Both plots were managed as cut-and-carry systems where residues of tree pruning (every 2-6 months, tree height <2 m) and cut grass (every 42 days) were removed from the site and animals were excluded. A three compartment model of soil porosity was constructed to interpret the results of water dynamics in both grasslands. The three compartments corresponded to the structural water (easily available), the matric water (not easily available), and the cracks (bypass flow); i.e. three levels of soil porosity. The incoming photosynthetically active radiation at the grass level (PARi) varied from 4.5 to 9.5 MJ m-2 per day in OG, and from 2.5 to 7.5 MJ m-2 per day in SS. Although windspeed was lower under the trees, air temperature (average 25 °C) and air humidity (>70% even in the dry season) at the grass level were similar in both treatments. Therefore, evapotranspiration from the grass was always lower in SS because it depended principally on PARi. During the rainy season, soil water content was lower in SS due to a higher water uptake by the association, and the simulations performed with the model indicated that the structural porosity was an important water reserve for the crops. In the dry season, when the matric reserve was the major water source, water uptake was reduced by the low hydraulic conductivity of the soil, and then water content was similar in both treatments. These results suggested that the tree component was relatively less competitive for water in the dry season, and it provided a microclimate which reduced grass evapotranspiration compared with OG. This was supported by the results of grass production in the dry season which was similar in both treatments. Conversely, grass production in the wet season was higher in OG because the lower PARi limited grass growth in SS. The model satisfactorily described the changes in soil water content in both treatments. However, some discrepancies between model and experimental data were observed when the soil was rewatered following heavy rainfall (>90 mm per day). This was due to the assumption concerning bypass flow within the soil cracks, which needs to be improved in future work.