C S I R O
P U B L I S H I N G
Australian Journal of Soil Research Volume 37, 1999 © CSIRO Australia 1999
A journal for the publication of original research in all branches of soil science
w w w. p u b l i s h . c s i r o . a u / j o u r n a l s / a j s r All enquiries and manuscripts should be directed to Australian Journal of Soil Research CSIRO PUBLISHING PO Box 1139 (150 Oxford St) Collingwood Telephone: 61 3 9662 7628 Vic. 3066 Facsimile: 61 3 9662 7611 Australia Email:
[email protected]
Published by CSIRO PUBLISHING for CSIRO Australia and the Australian Academy of Science
Aust. J. Soil Res., 1999, 37, 527–44
.
Nitrogen dynamics in a eucalypt plantation irrigated with sewage effluent or bore water V. O. SnowAD , C. J. SmithA , P. J. PolglaseB , and M. E. ProbertC A B C D
CSIRO Land and Water, GPO Box 1666, Canberra, ACT 2601, Australia. CSIRO Forestry and Forest Products, PO Box E4008, Kingston, ACT 2604, Australia. CSIRO Tropical Agriculture and Agricultural Production Systems Research Unit, 306 Carmody Rd, St Lucia, Qld 4067, Australia. Corresponding author; email:
[email protected]
Abstract Irrigation of treated effluent onto tree plantations is a popular method of land treatment but can lead to unacceptable levels of groundwater degradation. Knowledge of nitrogen transformations and balances is essential to design and operate plantations so as to keep groundwater degradation within acceptable limits. APSIM for Effluent, a model of water, salt, and nitrogen in effluent-irrigated plantations was tested against data from a plantation of Eucalyptus grandis (flooded gum) irrigated with either secondary-treated sewage effluent or bore water. APSIM was then used for quantifying nitrogen transformations, leaching, and balance, within the plantation. Summed over 5 years, the predicted nitrogen balance of the effluent-irrigated treatment showed that the accumulation of nitrogen in the aboveground biomass and litter (335 and 19 kg/ha) was significantly less than the amount of nitrogen added in effluent (508 kg/ha). Denitrification at this site was low, about 52 kg/ha over 5 years, because the soil was permeable and unlikely to become anaerobic for substantial periods of time. After 5 years, organic nitrogen decreased by 167 kg N/ha, and 269 kg N/ha was leached. In the trees irrigated with bore water, accumulation of nitrogen in the biomass and litter (301 and 34 kg/ha) was not much less than for the effluent-irrigated treatment and was considerably greater than the nitrogen added in the bore water irrigation (14 kg/ha). After 5 years, the predicted fluxes were 10 kg/ha denitrified, 389 kg/ha depleted from soil organic matter, and about 58 kg/ha leached. About 75% of the nitrogen leaching occurred in the first year of the experiment. Additional keywords: modelling, leaching, net mineralisation, nitrogen balance.
Introduction Disposal of treated sewage effluent to rivers and coastal waters can degrade water quality. Of particular concern are eutrophication and the occurrence of toxic blue-green algal blooms in the River Murray and other surface water bodies in the Murray–Darling Basin. GH&D (1991) found that discharge of sewage effluent was the major point source of nitrogen and phosphorus in the Murray. As a result there is increasing community and regulatory pressure to reuse treated sewage effluent by irrigation. Irrigated tree plantations are a popular method of land treatment of effluent (Stewart and Boardman 1991; Myers et al . 1995) because of reported fast tree growth and high rates of water use (Stewart and Flinn 1984; Stewart et al . 1988), and for socio-economic reasons (Myers et al . 1994). Rapidly growing plantations initially accumulate nitrogen at a relatively high rate, 75–100 kg q CSIRO 1999
10.1071/SR98093
0004-9573/99/030527$10.00
528
V. O. Snow et al.
N/ha.year before canopy closure (Myers et al . 1995). However, the rate of accumulation decreases when canopy closure is reached (Polglase et al . 1995) leading to increased potential for leaching of nitrogen and contamination of groundwater. Sewage effluent is relatively high in salt (Feigin et al . 1991; Bond et al . 1995). In many climates, acceptable concentrations of salt in the root-zone can only be maintained by deliberate leaching, which can lead to an unacceptable accession of salt to groundwater. Nitrogen leaching can be minimised by controlling effluent additions so that addition of nitrogen in effluent plus net mineralisation approximately equals net uptake by the trees and gaseous losses. In order to control the nitrogen input to appropriate levels, an understanding of various biological and chemical components of the nitrogen cycle and their interactions under effluent irrigation is necessary to calculate appropriate loading rates. A simulation modelling approach based on sound principles, and validated against field data, is one method to guide nitrogen management of effluent-irrigated plantations and so protect groundwater from unacceptable degradation. In 1991 the Wagga Wagga Effluent Plantation Project (WWEPP) was established to examine fluxes of water, salt, nitrogen, and phosphorus in an effluent-irrigated plantation. A component of this project was the development, validation, and application of a process-based model, ASPIM for Effluent, to guide design and management of effluent-irrigated plantations. Comparison of APSIM simulations against data of water and chloride in a eucalypt plantation irrigated with effluent has been presented by Snow et al . (1999). This paper extends those comparisons by examining the ability of APSIM to simulate nitrogen dynamics in the plantation. Model description and data Simulation software All computer modelling was done with APSIM version 1 · 4, the Agricultural Production Systems Simulator (McCown et al . 1996). APSIM is a flexible simulation environment that can be configured with modules suitable for the simulation of various terrestrial production systems. APSIM for Effluent (Snow et al . 1999) is a particular configuration of APSIM which includes the modules required for simulation of effluent-irrigated plantations. The modules comprising APSIM for Effluent are outlined in Table 1. Tree and weed nitrogen dynamics The module GRANDIS simulates tree growth using a simple functional representation of seasonal (litter fall and new leaf growth) and long-term canopy and wood dynamics (Snow et al . 1998, 1999). Nitrogen demand by trees is calculated as the sum of individual growth components (leaves, branches and stems, roots) multiplied by their appropriate nitrogen concentrations. The distribution of nitrogen uptake with depth is proportional to water extraction as determined by the SWIM module (see below). Allowance was made for translocation of nitrogen from senescent leaves before litter fall. Root growth was assumed to be proportional to canopy development. Root length density at the soil surface was assumed to be 0 · 05 mm/mm3 and rooting
Nitrogen dynamics in an irrigated plantation
529
depth was determined from observations (B. J. Myers, unpublished data). An exponential decrease of root length density to the maximum rooting depth was assumed (Snow et al . 1999). Using a specific root length of 20 g/km, simulated root mass was about 20% of the aboveground biomass of the effluent-irrigated trees at maturity. The combination of deeper rooting and lower aboveground biomass in the water-irrigated trees resulted in root mass being closer to 30% of the aboveground biomass. These values are consistent with observations at WWEPP (B. J. Myers, unpublished data) and with information in Attiwill and Leeper (1987). Based on summary information in Attiwill and Leeper (1987), root turnover was assumed to be a quarter of the roots each year. Root nitrogen content was 1% (P. J. Polglase unpublished data). Table 1. APSIM modules in APSIM for Effluent and brief descriptions of their purpose References: 1, McCown et al . (1996); 2, Snow et al . (1999); 3, Verburg et al . (1996); 4, Huth et al . (1996); 5, Probert et al . (1998) Module REPORT INPUT MANAGER MET OPERATNS FERTIGAT EVAPORAT INTERCEP CANOPY SWIM SOLUTE SOILN RESIDUE GRANDIS WEED MAP
Ref.
Purpose
1 1 1
Control of output of simulation variables to files Input of meteorological information Simulation control, manager-like decision making during simulation Ancillary calculations involving meteorological information Pre-set applications and activities during simulation Irrigation of effluent Calculation of plantation evaporative demand by the Penman equation Interception of rainfall and irrigation, evaporation during irrigation Interception of radiation by trees and weeds Numerical simulation of water movement by Richards’ equation and solute movement by the convection–dispersion equation Tracking of salt, chloride, and tracers Nitrogen transformations Leaf and weed litter decomposition Tree growth, nutrient uptake, and water demand Weed growth, nutrient uptake, and water demand Utility module for reformatting layered output
1 2 2 1 3, 4 4 5 5 2 2
For the simulations shown here, the WEED module (Snow et al . 1999) was used in a mode that allowed growth to be interpolated from a supplied time-series of biomass information. As with trees, weed nitrogen demand was calculated as the product of growth and nitrogen concentration, and the distribution of uptake was determined by simulated water extraction. When weeds were sprayed with herbicide or slashed, the resulting nitrogen and carbon residues were added instantaneously to the litter layer. Soil nitrogen transformations Organic N and mineral N transformations in soil (net mineralisation, nitrification, denitrification, and decomposition of plant residues incorporated into soil) were simulated using the module SOILN version 2. An earlier version of SOILN has been described in Probert et al . (1998). The principal change in version 2 is that simulation of nitrification was altered to use Michaelis–Menton kinetics (Godwin and Jones 1991):
530
V. O. Snow et al.
+ + M = k × [NH+ 4 ]/([NH4 ] + [NH4 ]half )
where M is the potential nitrification rate (g N/Mg soil.day), k is the potential nitrification rate constant (g N/Mg soil.day), [NH+ 4 ] is the concentration of ammonium in the soil (g N/Mg soil), and [NH+ 4 ]half is the half-saturation concentration of ammonium (g N/Mg soil). The potential nitrification rate was modified for soil moisture, soil temperature, and soil pH as described in Probert et al . (1998). Nitrate and ammonium in irrigation water were added to the appropriate pools in the soil profile. The vertical distribution was controlled by transport of water and solutes in soil as simulated by the water module, SWIM (see below). The module SOILN v.2 was modified to allow addition of the organic matter in effluent. Organic N and C in the irrigated effluent were assumed to be primarily dead algae with a fast decomposition rate, and so they were added to the carbohydrate-like fraction of fresh organic matter. All parameters are as given in Table 1 of Probert et al . (1998) with the following exceptions. For the new nitrification description, k and [NH+ 4 ]half were 40 g N/Mg soil.day and 90 g N/Mg soil (Godwin and Jones 1991). The value of the parameter controlling potential mineralisation rate from the humus pool, R hum , was the lowest value that supplied sufficient nitrogen to trees and weeds to match observed uptake in the treatment irrigated with bore water. The value was 0 · 00025/day, higher than the value of 0 · 00015/day found by Probert et al . (1998). Litter decomposition Decomposition of slashed weeds and of fallen leaves was simulated using the module RESIDUE v.2. As described for RESIDUE v.1 (Probert et al . 1998), potential decomposition was calculated as a potential rate modified for moisture, temperature, C : N ratio of the residues, and for contact between the residues and the soil surface. Actual decomposition also depended on the availability of nitrogen for immobilisation during the decomposition process. Based on observations of the mixed soil–litter layer, mineral N within the top 25 mm of soil was assumed available to the decomposers and decomposition products were returned to the top 25 mm of soil. In order to match observations of the amount of nitrogen in the litter layer, the potential rate coefficient for residue decomposition, R decomp , was required to be considerably less than the value of 0 · 1/day proposed by Probert et al . (1998). For the simulations here, R decomp was adjusted to a value of 0 · 025/day. Soil water and solute movement Water and solutes (nitrate, ammonium, organic nitrogen and carbon, and total dissolved salts) were simulated using SWIM (Huth et al . 1996), an adaptation of SWIM v.2 (Ross et al . 1992; Verburg et al . 1996). SWIM employs a numerical solution to the Richards’ equation for water movement and the convection– dispersion equation for solute movement. The ability of APSIM to simulate water and non-reactive solute movement has been demonstrated in Snow et al . (1999).
Nitrogen dynamics in an irrigated plantation
531
Site, soil, and experimental design Experimental work was carried out on an effluent-irrigated plantation near Wagga Wagga, New South Wales. During the 5 years of the experiment, rainfall averaged 675 mm/year and was slightly winter-dominant. Mean annual pan evaporation was 1585 mm, mean daily minimum temperature in midwinter was 3◦ C, and mean daily maximum temperature was 31◦ C in midsummer. Myers et al . (1996) and Falkiner and Smith (1997) have described in detail the site and experimental design of the WWEPP. Either effluent or bore water was irrigated onto 16 plots (each 0 · 3 ha) of Eucalyptus grandis or Pinus radiata. The plots were arranged in a single row oriented approximately south–north. In total there were 4 treatments arranged in 2 replicate blocks, designated ‘south’ and ‘north’. The 2 treatments considered here are designated ‘effluent’ and ‘water’. In both treatments the amount of irrigation applied was only that sufficient to replace estimated water use (Myers et al . 1996) and keep drainage to a minimum. Water use was assessed individually for each treatment but there was little difference between the treatments. The annual irrigation application ranged between about 600 and 1300 mm/year depending on stage of development and weather. The effluent treatment was irrigated with effluent and the water treatment with bore water. Average nutrient concentrations in effluent were 4 · 5 g/m3 of nitrate-N, 2 · 5 g/m3 of ammonium-N, and 4 · 3 g/m3 of organic N, and the C : N ratio of the organic matter was 4. The bore water contained 0 · 05 g/m3 of ammonium-N, and 0 · 3 g/m3 of organic N. The soil type was a red kandosol (Isbell 1996). Although soil bulk density was high, 1 · 66–1 · 78 Mg/m3 , it was quite permeable having a saturated hydraulic conductivity of 850 mm/day at 0 · 3 m and 35 mm/day at 1 · 0 m. Bond et al . (1998) gives more information about measurement of the soil’s hydraulic properties. Tree and weed biomass nitrogen Tree biomass was assessed at approximately 6-monthly intervals by destructive harvesting. In August each year, about 25 trees per treatment were selected to represent the range of size classes in each treatment. A smaller harvest was taken in February when about 5 trees per treatment were measured. Selected trees were cut at ground level and separated into several components (Myers et al . 1994) before drying and subsampling for dry matter and nutrient analysis. Concentrations of total nitrogen in plant tissues were determined by automated colorimetry after digestion of subsamples in H2 SO4 /H2 O2 at 340◦ C (Heffernan 1985). Weeds in the eucalypt plots died out soon after canopy closure, approximately 18 months after the spring 1991 planting. Before canopy closure, weeds were managed by periodic slashing and spraying. A record was kept of all weed control measures. Weed biomass was measured occasionally and included estimates of the mass of dry matter slashed and nutrient content (N. D. O’Brien, unpublished data). Standing biomass between the measurement dates was estimated by linear interpolation. These data were converted into the necessary time series of leaf area index and height. Root depth was assumed proportional to simulated weed height. Litter nitrogen The mass of leaves falling from the trees was measured using 8 litter traps per plot. Before January 1993, when the plantation canopy was open, small traps of 0 · 1 m2 were placed directly beneath trees. As the plantation approached canopy closure, larger traps of 0 · 5 m2 were used and placed randomly throughout each plot. Leaf litter was removed from the traps each month, the dry weight measured, and a sample analysed for nutrient content as described above for biomass measurement. Nitrogen stored in the litter layer on the soil surface was measured in July 1995 and August 1996. On each occasion, 10 litter samples were collected randomly within each plot using a 0 · 16-m2 ring. The entire litter layer was collected and was easily distinguishable from the underlying soil. Samples were dried, weighed, ground finely, and analysed for nitrogen using a LECO CHN combustion analyser. Soil cores Mineral N in the soil was measured on 5 occasions, all during the fifth year of the experiment. Six cores, 44 mm diameter, per plot were taken to a depth of 0 · 9 m and
532
V. O. Snow et al.
subdivided into 0 · 1 m depth intervals. At each sampling, the total amount of soil in each section was collected in polyethylene bags and weighed. After mixing all fresh soil in each sample, mineral N was extracted by shaking the equivalent of 10 g of oven-dry soil with 100 cm3 of 2 M KCl for 1 h (Keeney and Nelson 1982). The suspensions were filtered and the filtrate was stored frozen until analysed. Ammonium was measured by reacting with hypochlorite liberated from dichloroisocyanurate under alkaline conditions and measuring the absorbance at 660 nm (Alpkem 1992). Nitrate was determined by quantitative reduction to nitrite by a cadmium column using a segmented flow analyser (Alpkem 1992). The absorbance of light was measured at 540 nm.
Soil solution sampling During winter 1994, porous ceramic soil solution samplers were installed at 0 · 5 and 1 · 0 m deep, replicated at 6 positions in each plot. The ceramic tip of each sampler was connected to a PVC pipe (50 mm diameter) which extended to the soil surface. Samplers were installed by augering a slightly over-sized hole for the length of the sampler except for the deepest 150 mm. Soil for this last section was extracted to ensure direct and tight contact between the ceramic tip and soil. After inserting the sampler, the hole was back-filled with B-horizon clay. A bentonite plug, about 200 mm long, was inserted in the back-fill at about the middle of the sampler. Soil solution samples were collected approximately every 28 days. The samplers were sealed under vacuum of about 50 kPa and left for about 48 h for the sample to accumulate before collection. Samples were transported immediately back to the laboratory and stored at 4◦ C until analysis, usually within a few days of collection. Samples were analysed for nitrate and ammonium by automated colorimetry and for total nitrogen after digestion in H2 SO4 /H2 O2 at 340◦ C (Heffernan 1985).
Mineralisation rate measurements Mineralisation rate was measured using a sequential coring technique replicated 8 times in each plot (Polglase et al . 1995). PVC cores, 50 mm internal diameter, were driven 0 · 2 m into the soil and the top was loosely covered to exclude rainfall and irrigation. Matching samples of soil near the newly installed cores were taken to provide measurement of the mineral N concentration at the beginning of the period. After about 28 days the cores were removed and a new set installed. Soil was sectioned into 0–0 · 1 and 0 · 1–0 · 2 m increments, sieved to
