Development of water use models for dryland Pinus radiata agroforestry systems

Abstract

This study investigated the water use and water relations of a 5-6 year old Pinus radiata agroforestry plantation at Lincoln, New Zealand. There were two tree stocking densities: 1) 400 stems/ha, during the 1995-96 growing season; and 2) 200 stems/ha during the 1996-97 growing season. The two understories investigated were: a) a mixture of 'Wana' cocksfoot (Dactytis glomerata) with clovers; and b) 'Yatsyn' ryegrass (a cultivar of Lotium perenne cv). A second trial was located in Eyrewell Forest, in Canterbury, New Zealand. Here the effects of irrigation and understorey type on tree canopy conductance (Gt) and tree transpiration on a projected crown area basis (Fdʸ) were investigated during the 1995-96 growing season. Eyrewell understorey treatments consisted of 1) a grass mixture, where Browntop "Massey Basyn" (Agrostis capillaries L.) and Yorkshire fog (Holcus lanatus L.) dominated; 2) a legume mixture where "Grasslands Maku" (Lotus pendunculatus) dominated; and 3) bare ground. At this trial some treatments were irrigated. At both sites tree transpiration was estimated for experimental trees by measuring sapflow using the heat pulse method. Soil moisture storage was measured using a TDR at both sites and a neutron probe at the Lincoln trial only.

At the Lincoln site, the cocksfoot understorey was found to be more competitive than ryegrass for soil moisture and used 78.3% and 81.9% of the total plot water use, compared with the ryegrass understorey at 58.7% and 64.4% of the total plot water use, during the 1995-96 and 1996-97 seasons, respectively. Differences in tree transpiration were not reflected in saw log growth, but rather as differences in the size of branches. An advantage of using a cocksfoot understorey was the improvement of wood quality due to smaller branches without significant reduction in saw log production. In addition, greater understorey dry matter production was recorded in the cocksfoot treatment.

Tree xylem sap flux density (Fd' mol m⁻² S⁻¹) treatment differences during the 1995-96 growing season included: a) at Lincoln cocksfoot treatment trees had lower Fd' than ryegrass treatment trees; and b) at Eyrewell, irrigation increased Fd' only in the absence of an understorey. Total tree canopy conductance to diffusion of water vapour (Gt ,mol m⁻² S⁻¹) was calculated by inverting a simplified Penman-Monteith model. At the Lincoln site, trees with an understorey of cocksfoot had a lower Gt than those with ryegrass, although the sensitivity of Gt to increasing air saturation deficit (Da) did not differ between treatments. At the Eyrewell trial, understorey vegetation and irrigation had significant effects on Gt. In all cases the presence of an understorey, or lack of irrigation, or both, reduced Gt and increased its sensitivity to Da.

At the Lincoln trial, average tree stomatal conductance (gs) was derived by inverting the Penman-Monteith equation. The sensitivity of gs to decreasing soil moisture (θ) was greater for trees growing over cocksfoot, and also increased for the higher stocking density in both understorey treatments. A model of tree transpiration (Te) performed well overall at the Lincoln trial during both seasons, but required a model of gs specific to each treatment to account for understorey effects on tree transpiration.

To predict understorey evapotranspiration, the Penman-Monteith model was used and performed well, even with difficulties in determining an aerodynamic resistance (rₐ) function. The aerodynamically rough tree canopy produced an enhanced degree of coupling of the understorey, as shown by the coupling parameter Ω when using the Penman-Monteith formula. Ω was nearer to values typical of forest than of short pasture.

The spatial variability of rainfall receipt to the understorey was large due to a rain shadow effect to the NNE or lee-side of the crown. The redistribution of gross precipitation (PG) has potentially the largest effect on the production of both the pasture and trees, rather than the magnitude of interception loss (I), which was found to be only on average 8.2% of PG for a tree at 200 stems/ha.

A growing season water balance model was tested for the Lincoln trial and performed well for both seasons with the large changes in both tree leaf area (L) and θ. When the model was run with an "average tree gs" model, it appeared to be transportable, with the modelled soil moisture θ tracking measured θ well at both 200 and 400 stem/ha stand densities.

This study indicated that the selection of an appropriate understorey species is important in designing agroforestry systems for low rainfall areas.