section 6: agroforestry and animals

Competition for water, light and nutrients in agroforestry associations of Pinus radiata and pasture

D.J. Connor, R. Sands and M. Strandgard

School of Agriculture and Forestry, University of Melbourne
 Parkville 3052, Australia

Abstract

An experiment to study the co-productivity of pine and sheep was established in 1983 when pine previously planted into improved pasture in 1981 was thinned to provide a range of pine-pasture combinations. The experimental site is 150 km northwest of Melbourne on an extensive basaltic plain at an elevation of 400 m. The annual rainfall is 650 mm and the annual evaporation 1100 mm. The soil is acidic, duplex with a clay loam surface soil 15 cm deep overlying a medium clay. The original vegetation of the area was a savanna woodland dominated by Eucalyptus camaldulensis but previously converted to improved pasture of Lolium perenne and Trifolium subterraneum but now with considerable invasion of Holcus lanatus.

There are five treatments of which the two boundary treatments are treeless pasture and a plantation of 1650/ha (3 m rows x 2 m tree spacing). The three agroforestry combinations are 100/ha (8x 12 m), 277/ha (4x9m) and a treatment in which 5 rows spaced 3 m apart (trees 4 m within rows) are separated by a 10-row gap. In this case the tree density in the planted area is 830/ha but on a total area basis it is also 277/ha. Plot size is 1.85 ha of which the inner 0.5 ha is separately fenced. Plots are grazed by individual flocks of sheep. Their performance is recorded as wool production and live weight gain.

The large treatment plots afford an excellent demonstration of agroforestry alternatives and the inclusion of grazing adds realism to the microlimate, productivity and nutrient flows in the systems. The comprehensive annual measurements that are being made of tree growth and animal production will, in the long term, allow a complete economic appraisal of the systems. However, it is the interfaces between trees and pasture that provide the most generally useful data. This is so because the co-productivity relationships to microclimatic modification across these margins can be generalized to assess the performance of otherwise unstudied combinations of pine and pasture.

Detailed measurements of the growth and hence of the interaction of trees and pasture are concentrated along transects across the margins between treed plots and the pasture controls. These include various attributes of tree size, pasture production and soil moisture content to 13 m. Measurements within the plots include light, wind run and the redistribution of rainfall by canopy interception and stem flow. Tree-pasture competition is also being studied in the 100/ha treatment using combinations of trenching and herbicide application.

Results so far are limited. Differences in tree growth are evident. The trees in the highest density are tallest but the stem volume of those evenly spaced at 277/ha is greatest. The low density trees (100/ha) have grown least, probably a response to exposure at this windy site but perhaps also to water stress because the lowest density treatment has the lowest subsoil water content of the treed treatments. This may reflect the aerodynamic roughness of the arrangement but it is evident that the presence of trees improves the infiltration of water into the clay subsoil. There has frequently been free water in surface irregularities of pasture and low density tree plots during the winter whereas in the plantation and the more-densely treed plots, free surface water has been conspicuously absent.

Analysis of pasture productivity shows that even in the open pasture the productivity is below potential with maximum spring growth rates over a 90-day period averaging only 60-70 kg/ha-day. So far there is little evidence of a decline in pasture productivity except in the plantation in which grazing was suspended during winter 1986 (5 years after planting). Except in the plantation, the tree cover in the agroforestry combinations, especially following their pruning in winter 1986, is insufficient to dominate pasture productivity.

Trenching of root systems and removal of pasture by the application of herbicide (glyphosate) in the vicinity of trees in the 100/ha treatment has demonstrated the significant effect of tree-pasture competition on the water relationships and growth of trees. To date it does not appear that the effects are any different in the other agroforestry combinations. However, measurements of soil water profiles during summer 1985 show that there was, for the first time, a marked drying under the treed treatments compared with pasture. Competition between the now larger trees and pasture appears to be intensifying after 5 years of relatively independent growth.

Introduction

Although much of the grazing during the past 150 years in southern Australia has been carried out on what was originally grassland or savannah woodland, the management of these grazing lands has been directed solely to the maximization of returns from animal products. Gradually many areas have been completely cleared of trees and replaced, often in a cropping rotation, with improved pastures. In other areas, trees have been thinned and the growth of improved pasture species encouraged by the application of phosphatic fertilizers and trace elements, generally with pasture seed broadcast or lightly scratched into the surface soil. Grazing and fire have prevented the regeneration of replacement trees so that, as the remaining trees die, most land is being converted to treeless pasture. There are a number of possible causes of the death of the remaining trees in what is commonly referred to as 'tree decline' and which is causing considerable community concern. The reasons include increased exposure, insect attack, rising water tables and salinity. However, given the lack of recruitment by tree seedlings to the tree layer under grazing, the result is inevitable as the remaining trees age, whether or not their death has been hastened.

The loss of tree cover is associated in many places with rising water tables and erosion of the surface soil. Over most of the area, the annual potential evaporation far exceeds rainfall but the rainfall/evaporation balance provides excess water which infiltrates into the subsoil during winter.

Whereas the native, evergreen trees and summer-active native grasses removed this water by active transpiration in summer, the winter-active species of improved pastures are summer-dormant annuals under which the subsoils become increasingly wet, leading ultimately to rising water tables in the lower parts of the topography. Where water tables become saline and encroach into the root zone, the productivity of what is often the best land in the topographic sequence is seriously impaired.

Agroforestry offers itself as one way to re-establish a hydrological balance that is compatible with continuing agricultural productivity. Alternatives are to concentrate on summer active agricultural species such as phalaris and lucerne, or to return some or all of the landscape to tree production. In agroforestry options there is an emphasis of Pinus radiata because there is a strong demand for softwood which cannot be met from the extensive, managed native hardwood forests. The combination of P. radiata and pasture has received much attention in Australia and elsewhere, e.g., New Zealand and Chile, but the refinement of the system for Victorian conditions, including its demonstration to farmers, requires further local experimentation. The experiment at Carngham was established for this purpose.

Methods

Experimental Site

The experiment is located at Carngham 150 km WNW Melbourne (Figure 1) on former farmland which was purchased by the government in 1981 for the establishment of

Figure 1 Design of the agroforestry trial at Crangham, Victoria.

P.radiata. The site is at the NW edge of a broad basaltic plain on which the original vegetation was a savannah woodland dominated by Eucalyptus camaldulensis (red gum) with a herbaceous storey of Themeda australis (kangaroo grass) and species of Stipa (spear grass) and Danthonia (wallaby grass). During 100 + years of grazing, mostly by sheep, and by sporadic application of superphosphate fertilizer, the vegetation has changed markedly. Tree seedlings have not regenerated so that trees are now virtually absent from the site. The herbaceous layer is dominated by introduced species of european and mediterranean origin. The preferred pasture contains Lolium perenne, (perennial ryegrass), Dactylis glomerata (cocksfoot) and Trifolium subterraneum (subterranean clover) but at this site there has been substantial invasion by Holcus lanatus (Yorkshire fog grass).

The soil is duplex with a clay loam A horizon sharply overlying a clay B horizon at 15 cm depth. The infiltration rate of the subsoil is low and the seasonal waterlogging that occurs at the junction is demonstrated by the layer of ironstone nodules that have formed there. The surface is slightly undulating but does not have the marked 'gilgai' micro-relief that is characteristic of many grassland and savannah woodland sites on heavy soils in southern Australia.

The seasonal waterlogging of the surface soil restricts cropping to areas of better drainage. The principal crop is oats grown for grazing and/or grain to support animal production but cropping is also a valuable part of the process of pasture renovation. The site of this experiment shows no evidence of previous cultivation.

Climate

Mean monthly climatic data for the nearby (20 km) city of Ballarat are summarized in Table 1. These data are generally applicable to the experimental site. Months of effective rainfall are defined as those in which rainfall exceeds one third of pan evaporation, a form of analysis that has wide applicability in southern Australia. In this case it shows that the five months from May to September inclusive have a supply of moisture adequate to sustain plant growth and that there is a high probability that conditions will remain effective in April and October also. On average the period of effective rainfall is nine months.

Mean daily minimum temperatures fall below 10 °C in the period April to November and are below 5 °C in the three months June to August. During these three months the growth of temperate species (pine and pasture) are restricted by low temperature. Mean monthly maximum temperatures do not exceed 28 °C and although a few hot days (35 °C) do occur, high temperature does not exercise an important direct controlling restriction on plant growth. During the summer months, water shortage caused by high temperatures and evaporative demand restricts growth of trees and pasture.

Strong winds are a feature of the climate of the basalt plains and one which can be successfully ameliorated by trees. The modification of wind, particularly of the 'chilling factor' during winter will receive attention in this experiment because of its significance to stock survival, particularly of ewes and lambs.

An automatically recording climate station was installed at the site. It records hourly integrals of global radiation, rainfall and wind run together with measurements on the hour (5-min average) of wet- and dry-bulb temperature and soil temperatures at two depths. Complete data are now available for 1985 and 1986. Two key parameters are wind run and global radiation; the former because modifications to the wind regime is an important feature of agroforestry systems and the latter because the potential productivity of pasture under trees depends ultimately upon the penetration of solar radiation to the grass layer.

Experimental design

The experiment was established in 1983 when pine previously planted in 1981 at plantation spacing (3 x 2 m = 1650/ha) was thinned to provide three replicates of the five treatments described in Figure 2. Plot size is 111 x 167 m (1.85 ha). Within each plot there is an internal plot of 48 x 104 m (0.50 ha) that is fenced, supplied with water and separately stocked with sheep for the measurement of grazing production. The external areas of all plots are, with a few exceptions where additional measurements are being made, bulk-grazed by a single flock of sheep. The pasture was slashed in the first year of the experiment but has been grazed since December 1984. The number of sheep grazing individual plots depends upon pasture availability. When feed is inadequate the sheep are removed.

Figure 2 Measurements of (a) tree height (cm), (b) canopy diameter (cm) and (c) stem conic volume (m x 10" ) over a two-year period.

Measurement

Regular measurements of soil water status, rainfall redistribution by the trees, and tree and pasture growth are made in each treatment. Additional measurements have been made of tree water potential and stomatal diffusive conductance in the 100/ha treatment in response to the removal of grass in the vicinity of the trees and to trenching designed to separate the root system of pasture from that of the trees.

Annual measurements of tree and animal production will provide the economic analysis of the agroforestry alternatives included in the experiment. Measurements of water redistribution, water balance and the competition studies will assist in the interpretation of the results. However, in order that the results obtained in this experiment can have the widest application, regular, detailed measurements of tree and pasture growth and water balance are concentrated along transects that traverse the boundaries between treed treatments and pasture controls. The establishment of tree and pasture growth response functions (Figure 3) across the interfaces and within the treed area will enable the competitive relationships included in this experiment to be extrapolated to tree-pasture combinations that are not included in this experiment.

Figure 3 Soil volumetric water content over a two year period for soil depths (a) 0-80cm, (b) 80-140 cm and (c) 160-220cm.

Soil water content is measured by neutron moisture meter (NMM) from the 20 cm depth at 20 cm-depth intervals in 3 m deep aluminium access tubes (90 in all) located along the transects. Pasture growth is measured at 3-monthly intervals (summer, autumn, winter and spring) in grazing exclusion cages of 1 m radius (200 in all) also located along the transects. Soil water is measured monthly whenever possible although the interval between readings can be extended to two months in winter with little loss of information when monthly changes are small. Canopy interception and stem flow are currently being measured in the 1650, 277 (even) and 100/ha treatments. Linear rain gauges (4 m) are deployed under tree canopies and between tree rows. Six trees per plot have been fitted with stem collars to collect stem flow. Additional NMM access tubes have been located in the vicinity of the interception equipment.

Results

Tree growth

Measurements of tree height, canopy width and stem conic volume are presented in Figure 2. Trees in the two most dense treatments (1650 and 277/ha) are the tallest, approaching 4 m by age 5 years. The trees in the low density (100/ha) treatment and the 5-row pattern are shorter by approximately 1 m.

Canopy width is greatest in the 277/ha treatment. The two open- spaced treatments have compensated for less height with greater lateral expansion of their canopies, but as with height growth, canopy width demonstrates their poorer growth. These estimates of mean canopy diameter lead to gross estimates of vertically projected cover of 0.42, 0.08 and 0.02 for 1650,277 and 100/ha respectively. Within the treed area, the cover of the 5-row pattern is 0.13.

Tree productivity is most closely approximated by stem volume. The data in Figure 2 show that the trees at 277/ha, which have the widest canopies, also have the greatest stem volume. The open spaced planting (100/ha) and the 5-row pattern are 50% smaller after 5 years of growth.

Analyses of the growth of individual trees across the pasture- tree interface are presented for two occasions (August 1984 and April 1986) in Table 2. Trees grow best at the margins of the 1650/ha and 5-row treatments but performance across the interfaces of the other more open treatments is variable.

Soil water content

Changes in volumetric water content for three depth intervals 20-60,80-140 and 160-220 cm are shown in Figure 3. Seasonal changes in soil water content have occurred only above 80 cm shown here in the data for the 20-60 cm. In this layer the pasture and the plantation had the highest soil water content for the first two years of measurement but during the last summer the plantation dried the most. The three agroforestry combinations (100/ha, 277/ha and 5-row pattern) were at all times drier than the pasture but during the last summer were wetter than the plantation.

Below 80 cm there is little evidence of seasonal change except in the last summer when water content in the 80-140 cm fell under the plantation. An interesting observation is that the soil water content at depth below the plantation has been wetter than that under all other treatments. There is no seasonal fluctuation and no differences have been established between treatments in the depth range 160-220 cm.

Table 2 Growth of individual trees.

Analyses of the change in soil water content across the pasture- tree margins are incomplete but show that the gradients are steep across the interface between the pasture and the plantation and the 5- row pattern. In the case of the 100/ha and 277/ha treatments, the trees are so sparse that the gradients within the planting are as marked at this stage in the development of the systems as are those across the interface.

Pasture growth

The data presented in Figure 4 describe the 1986 spring pasture growth across the treatment-pasture interface for the plantation (1650/ha) and the 5-row pattern. These treatments now show an effect of treatment on pasture productivity. Within the 277/ha and 100/ha plantings there is still no difference in pasture production from that of open pasture.

In the 5-row pattern the effect is relatively small. The data show no sites of high production within the planted area so that the average production is less than in the pasture. Across the pasture- plantation interface the effect of trees on the productivity of pasture is sharp (Figure 4). Pasture productivity in the open averages 5.1 g m^-2 day^-1 while that under the trees is 2.8. The variation in pasture production within the plantation reflects the distance of samples from individual trees rather than the distance from the edge of the treatment (Figure 5).

Figure 4 Spring pasture productivity (gm-2 day-1 ) across (a) the platation-pasture interface, and (b) the 5-row treatment-pasture interface.

Figure 5 Pasture productivity (gm~ day" ) across (a) the platation-pasture interface, and (b) the 5-row treatment-pasture interface, expressed in terms of distance to the nearest tree.

Tree-pasture competition

In the 100/ha treatment an experiment was undertaken to investigate the effect of pasture on the growth of trees. The 100/ha treatment was chosen because it was considered that there would be no significant competition between trees at this stage. Two square zones, inner (side 1.5 m) and outer (side 3 m), were established around a number of trees. Inner and outer trenches were dug to 1 m, lined with plastic film and refilled. Pasture growth was removed using the herbicide glyphosate. The combinations and further details are shown in Figure 6.

Removal of pasture from around the tree bole (no pasture and outer pasture treatments in Table 3) caused an improvement in tree water status as indicated by a significantly higher midday needle water potential measured in January 1986. Pasture removal treatments, made in June 1985, were associated with an increased wood production over the period to January 1986 (Table 3). The effect was more marked by January 1987.

Figure 6 Trenching and pasture removal combinations.

The increased wood production was probably due in part to the improved tree water status. However, even though the 'no pasture' and 'outer pasture' treatments were equally effective in raising needle water potential, they were quite different in their patterns of water use. The 'no pasture' treatment used least and the 'outer pasture' treatment used most soil water over the period June to January (Table 3).

Table 3 Growth and water relations of pine in response to the removal of pasture.

Discussion

The agroforestry trial at Carngham is a traditional randomized block design. Although statistically robust, such trials occupy large areas and are expensive to establish and maintain and consequently can compare a limited number of tree spacings. By contrast designs such as the Nelder Wheel and a variety of parallel row designs can examine tree spacing as a continuous variable over a wide range. These designs are, however, upset by silvicultural treatments such as thinning and are sensitive to tree mortality. In the experiments described here an attempt has been made to maintain the robustness of the traditional design while at the same time gaining some of the advantages of the systematic designs. This has been done by making additional measurements along transects across the interfaces between trees over a range of spacings and pasture.

It is envisaged that the examination of these interfaces will provide valuable information about the competition between trees and pasture and that the production functions of trees and pasture across the interfaces will be appropriate relationships to include in simulation models of the productivity of a wider range of agroforestry systems.

It will certainly give a more comprehensive understanding of an agroforestry system than could be gained from measurements taken only within the individual treatments. The approach should also assist the design of windbreaks and of a wide range of discontinuous agroforestry systems. At this early stage of the investigation only preliminary results have been obtained.

For the first five years the effect of pasture on tree growth has not been significantly different between treatments. Now, differences in tree growth are emerging and gradients are being established at the interfaces between treatments. The effect of tree growth on pasture production is evident in the plantation, becoming established in the 5-row treatment (effective density 825/ha) but at the lower tree densities, the growth of pasture is still unaffected.

The subsoil under the plantation (1650/ha) is wetter than under pasture. Considering the relative rooting depth of pasture and trees this is a surprising result. It is likely that the wetter subsoil reflects higher infiltration of water into the soil under the trees rather than a lower rate of water extraction. Certainly, while surface water has been relatively common during winter in the surface irregularities of the pasture treatment, it was infrequent and short- lived in the plantation and denser tree treatments. It is expected that this trend will be reversed as the trees develop larger canopies and denser, deeper root systems. The beginnings of this are perhaps evident in the drying of the subsoil which was observed for the first time in summer 1986.

Agroforestry experiments typically test a range of tree spacings against pasture (or crops). Results from these experiments tell more about the effects of trees on pasture than of the effects of pasture on trees. The reciprocal experiment, in which a range of pasture densities (or treatments) is tested against a constant tree spacing is more informative about the effect of pastures on trees but few such experiments have been carried out except in studies of weed control in plantation forestry.

The pasture removal experiment reported in Table 3 is an example of such a reciprocal experiment. It is well known that clearing herbaceous competition from around the stem of pine can substantially improve growth, particularly in the early establishment phase when tree roots are shallow and actively competing with roots of herbaceous species for soil water and nutrients. The use of herbicides in the establishment of pine is now standard practice, and this may explain that while the effect of trees planted into pasture may appear to have minimal differential effect on total pasture productivity there is nonetheless significant competition which restricts the early growth of trees.

Measurements are being continued to measure the co-productivity and water balance of trees and pasture. Examination of their performance across the treatment interfaces will assist in the interpretation of the treatment responses and provide production functions that will be useful in the design of a wider range of agroforestry systems.

Acknowledgement

This work is supported by a grant from the Reserve Bank of Australia.