Long-term productivity of a Grevillea robusta-based overstorey agroforestry system in semi-arid Kenya - II. Crop growth and system performance

Maize and cowpea were grown as sole stands or in agroforestry systems conta ining grevillea trees (Grevillea robusta A. Cunn.). Crop and system perform ance were examined over a 4.5-year-period (nine growing seasons) commencing in October 1991; failure of the rains caused the loss of one cropping seas on. A rotation of cowpea (Vigna unguiculata L. Walp.) and maize (Zea mays L .) was grown during the first five seasons after planting the trees, while maize was grown continuously during the final four seasons. Sole maize was also grown under spectrally neutral shade netting which reduced incident ra diation by 25, 50 or 75% to establish the relative importance of shade and below-ground competition for water and nutrients in determining the perform ance of understorey crops. The above-ground biomass and,grain yield of understorey crops were not sign ificantly affected by the presence of grevillea during the first four seaso ns, but were greatly reduced in subsequent seasons as the trees became incr easingly dominant; maize yields reached 50% of the sole crop values only on ce during the final four seasons, when rainfall was unusually high. The hyp othesis that competition for water was the primary limiting factor for unde rstorey crops was supported by the observation that above-ground biomass an d grain yield were greater in the shade net treatments than in agroforestry maize, demonstrating that shade was not solely responsible for the substan tial yield losses in the latter treatment. Performance ratios (ratio of val ues for the agroforestry system relative to sole stands) for total above-gr ound and trunk biomass in grevillea were initially low, reflecting the impa ct of competition with associated crops during tree establishment, but incr eased to unity within 2.5 years. Performance ratios for the understorey cro ps exhibited the reverse trend, initially being close to unity but approach ing zero for three of the final four seasons. Performance ratios were never close to unity for both trees and crops during the same season, indicating that there was always competition for available resources irrespective of crop species or tree size. The implications for agroforestry system design and future research are discussed. (C) 2000 Elsevier Science B.V. All right s reserved.