Effect of tree density on competition between Leucaena leucocephala and Chloris gayana using a Nelder Wheel trial. I. Aboveground interactions

Silvopastoral systems with the tree legume leucaena (Leucaena leucocephala (Lam.) de Wit) and grass pastures are widely used for ruminant feeding in subtropical and tropical regions. Different densities and planting configurations of leucaena will influence relative yields of both species because of intra-and interspecific competition. With the aim to describe the effects of competition between leucaena and Rhodes grass (Chloris gayana Kunth), a Nelder Wheel trial with 10 different leucaena tree densities (100-80 000 trees ha(-1)) growing with and without Rhodes grass was established in a subtropical environment at Gatton, south-east Queensland, in November 2013. From 2014 to 2016, the biomass of leucaena (six harvests) and Rhodes grass (seven harvests) was measured by using allometric equations and the BOTANAL sampling procedure over 742 and 721 days, respectively. No complementary or facilitative aboveground interactions were observed between the leucaena and Rhodes grass components of the pasture system. Increasing leucaena tree density resulted in greater aboveground intra- and interspecific competition. Average maximum individual tree yield (38.9 kg DM tree(-1) year(-1)) was reached at 100 trees ha(-1) without grass competition and was reduced by 60% with grass competition. Rhodes grass biomass yield was negatively affected by shading from the leucaena canopy, with negligible grass yield at tree densities >= 8618 trees ha(-1). Therefore, there was effectively no grass competition on individual tree yield at higher leucaena densities. Accordingly, edible leucaena biomass per unit area was positively related to log(10) leucaena density (R-2 = 0.99) regardless of grass competition, reaching 21.7 t DM ha(-1) year(-1) (2014-15) and 27 t DM ha(-1) year(-1) (2015-16) at the highest leucaena density of 80 000 trees ha(-1). By contrast, the yield of Rhodes grass was linearly and inversely correlated with log(10) tree density (R-2 = 0.99). Practical implications for the design and management of commercial leucaena-grass pastures are discussed.