Resource capture and use in semi-arid overstorey agroforestry systems
The work reported here aimed to provide a comprehensive database of core information to support the development and validation of process-based models of resource capture and growth in semi-arid overstorey agroforestry systems. Intensive field studies were carried out in Kenya over a 30 month period and the results obtained were combined with data from a previous project to produce a dataset spanning a 4.5 year period. This dataset was then used to verify output from the HyPAR model. Allometric procedures developed from the pipe model theory (Lott et al., 1998) were used to estimate tree growth non-destructively throughout the observation period. Significant differences in tree size between the sole (Td) and dispersed agroforestry (CTd) treatments were established during the first 130 days after planting, probably because of competition with the associated crops. The above-ground biomass and trunk length and taper characteristics of the CTd trees remained inferior to those of Td trees throughout the observation period, seriously undermining the economic potential of this agroforestry system. The biomass and grain yield of CTd understorey crops were similar to the corresponding sole crops during the first three seasons, but were negligible in three of the final four seasons, with maize yields reaching 50 % of the equivalent sole crop values only when seasonal rainfall was well above average. This observation suggests that water availability was the primary limitation for CTd maize during the final seasons of the trial, a conclusion supported by the superior performance of maize grown under net enclosures which simulated tree shade in the absence of below-ground competition. Cowpea and maize were grown concurrently in two seasons to examine the impact of grevillea on C4 and C3 crops with contrasting responses to shade; biomass and grain yield were less affected in cowpea than in maize. The tree canopy in the dispersed agroforestry (CTd) treatment reduced the daily mean quantity of radiation incident upon the understorey crops by c.30 % during the final four growing seasons, although the discontinuous nature of the tree canopy caused substantial local variation in shading intensity. Seasonal mean fractional interception was greater for the combined canopies of the CTd treatment when soil moisture status was relatively high than for either of the sole canopies, suggesting the occurrence of spatial complementarity. Tree shade had a substantial moderating influence on meristem temperature since the mean diurnal temperature range was reduced from a maximum of 20°C in sole maize to 13 °C under the trees, and maximum meristem temperature was decreased by up to 6 °C relative to sole maize. However, the non-uniform shading provided by the trees caused substantial spatial variation in thermal time accumulation and hence crop development. Grevillea continued to grow during dry seasons and was therefore able to capture off-season rainfall which might otherwise have been lost from productive use. In addition, adaptation of heat balance gauges for use on grevillea roots (Lott et al., 1996) showed that substantial quantities of water could be extracted from deep-seated reserves below the crop rooting zone during dry periods, indicating the potential for spatial and temporal complementarity. However, transpiration by grevillea greatly exceeded rainfall during the dry season, rapidly depleting residual water supplies which might otherwise have been available for crop growth. In addition, approximately two thirds of the water used by the trees during cropping seasons was extracted from the soil surface horizons by lateral roots at distances of up to 2 m from the trunk. Thus, the potential for above and below-ground complementarity may be seriously undermined by the extensive capture of water by tree roots from the crop rooting zone. Comparison of output from the HyPAR model against the observed results provided information pertinent to future model development. The model proved to be insufficiently flexible for end-users wishing to simulate the growth of different crops during the same simulation cycle, or to use model output to aid management decisions such as the timing of pruning. The allometric procedures used by the model to estimate canopy size from trunk diameter at breast height also proved incapable of accounting for reductions in canopy size resulting from pruning. Estimates of tree height are rounded to the nearest metre within the model, representing a potentially serious loss of resolution when annual increments often do not exceed 2 m. In addition, the numerous parameters required by the model would force most end-users to rely heavily on published information, potentially undermining the reliability of simulations.