Remediation of soil acidification by form of nitrogen fertilizer on grass swards of Australia and Thailand

Armour, J.D.¹; S. Berthelsen²; S. Ruaysoongnern³; P.W. Moody⁴ and A.D. Noble⁵

Keywords: acid soils, nitrate leaching, alkalinity generation, grass species, legume species

Introduction

Acidification of soil profiles from legume and N fertilized crops is a serious sustainability threat. It is associated with specific limitations to plant growth such as toxicity of aluminium and manganese and deficiency of calcium and molybdenum as well as a reduction in cation retention capacity. Under tropical conditions of Northeast Thailand and Northern Australia, acidification to >90 cm has been recorded in Stylosanthes and Leucaena pastures (Noble et al. 1997). Acidification has also been measured in other Australian cropping systems fertilized with urea or ammonium forms of N (Moody and Aitken 1997). The major processes contributing to anthropogenic acidification are removal of base cations in the harvested product and leaching below the root zone of nitrate from ammonium and urea N fertiliser or legumes that result in a net accumulation of protons in surface horizons. Subsoil acidity is often not corrected by conventional application of liming products to the soil surface (Moody and Aitken, 1997), although cost prohibitive in most cropping systems and invasive, practices such as deep placement of lime can be effective (Farina and Channon 1988). The processes of soil acidification from C and N cycles are additions of NH₄⁺ and NO₃^- (alkalising), leaching of NO₃^-, export of organic anions (calculated from the amount of crop removed and its ash alkalinity) and changes in soil organic matter (Helyar and Porter 1989).

The supply of nitrate to plants has been shown to alkalise soil, in contrast to acidification when supplied with ammonium in pot experimentation (e.g. Jarvis and Robson, 1983). The current study examines the rate, quantity and extent of alkalisation by grasses supplied with nitrate under field conditions.

Materials and methods

Field experiments with pastures on Acrisols in Northeast Australia and Northeast Thailand compared the rates of alkalisation or acidification from N applied as nitrate or as urea (Australia) or ammonium sulphate (Thailand) (Table 1). The plots were cultivated, sown with seed and allowed to establish before N treatments started in October 2000, July 2002 and September 2001 at sites in Mareeba and Tully (Australia) and Chiang Yuen (Thailand), respectively. Rates and forms of N in each application are detailed in Table 1.

Mareeba

Nitrogen treatments were applied during 4 growing seasons (October to May) and irrigation applied as required. Plots were harvested by hand cutting to 75 mm above ground with a 0.25 × 0.25 m quadrant. The remaining plot area was mown and the clippings removed to simulate a cut-and-carry system. N fertilizer at a rate of 50 kg N ha^-1 was applied to the soil surface after each harvest on 7 occasions per year, as either urea (N_urea) or a combination of Ca(NO₃)₂, KNO₃ and NaNO₃ (N_nit) (Table 1). Irrigation equivalent to 25 mm of rain was immediately applied after fertilizer application. Potassium and Ca applications were the same regardless of N source but SO₄-S varied from 100 to 355 kg S ha^-1. Other fertilizer applied at planting supplied P 30, Mg 70, Cu 2, Zn 1.5, B 0.4 and Mo 1.7 kg ha^-1, respectively. Additional P was applied in October 2000 (30), October 2001 (30) and December 2003 (40 kg P ha^-1) to correct low extractable soil P concentrations. Rainfall varied from 443 to 1,248 mm (long term average 840 mm, with 88% falling between December and April).

Soil water samples were collected from 3 replicates of each N treatment with barrel lysimeters (diameter of 31 cm). They contained 3 ceramic cups and were installed at a depth of 1 m before sowing. Water was extracted from the soil profile via the ceramic cup at a negative pressure equivalent to natural drainage. The pressure was generated by a falling head of water that was maintained between fixed heights by a pump. The cups were connected to individual 5 L water traps also located at a depth of 1 m. Samples were collected at a minimum of weekly intervals, after being stored at soil temperature in the light-proof water traps, and more frequently as drainage events dictated. The samples were chilled after extraction and then frozen on delivery to the laboratory until the time of analysis. Volumes of water extracted were calculated from the height of the water in the calibrated external sampler. Water samples were analysed for ammonium-and nitrate-N (Methods 7C2 and G3b, Rayment and Higginson, 1992). The amount of N leached on 136 occasions between November 2000 and June 2004 was calculated from the concentration and volume for each sample.

Table 1. Selected site and agronomic practices implemented at the three experiment sites

The harvested grass was dried at 65ºC, weighed and a sub-sample ground for chemical analysis. Samples were digested by a micro-Kjeldahl procedure and N determined by a continuous flow analyser procedure. Composite soil samples from 3 cores per plot were collected in June-July (mid dry season) at depths of 0-10, 10-20, 20-30, 30-50, 50-70, 70-90 and 90-120 cm (2004 only) and bulked. Samples were dried at 40ºC, sieved to <2 mm and analysed for pH (1:5 in water and in 0.01 M CaCl₂, Methods 4A1 and 4B2 in Rayment and Higginson 1992), mineral N (ammonium- and nitrate-N, Method 7C2) Colwell (bicarbonate extractable) P (Method 9B2) and total organic C (2003 only, Method 6B2). pH in CaCl₂ (pH_Ca) data are presented because of their more consistent trends from year to year.

Similar harvest and analytical procedures were used at the other sites although lysimeters were only installed at Mareeba.

Tully

After establishment, the plots were fertilized and harvested for 11 months (June 2002 – April 2003) and received N as urea (N_urea) or Ca(NO₃)₂, KNO₃ and NaNO₃ (N_nit) at 3 rates (25, 50 and 75 kg N ha^-1 per application for at total of 9 applications). Applications of K and Ca increased proportionally to rate of applied N with 1.15 kg K/kg N and 0.9 kg Ca/kg N. Soil samples were collected to 90 cm after plant establishment and before the application of N treatments in April 2002 and 1 year later. Rainfall during the experiment was 1,815 mm compared to average annual rainfall of 4,300 mm.

Chiang Yuen

Plots were planted in September 2001 and fertilized shortly after with 41 kg N ha^-1 as potassium nitrate (KNO₃) or ammonium sulphate (N_ammon). These applications were repeated 4 times in 2002 (May to September) and 2 times in 2003 (June, August). K was applied to the ammonium sulphate treatments at each N application and P was applied regularly as triple superphosphate (92 kg P ha^-1). Soil samples were taken at depths of 0-10, 10-20, 20-30, 50-60 and 100-110 cm at 1 year intervals. Harvests were undertaken 6 times over the duration of the study from 50 m². Rainfall was 1,219 mm between September 2001 and August 2002.

Results

Mareeba

There were no differences among treatments in pH_Ca before the N treatments were imposed in 2000 (mean profile pH_Ca 4.84 for 0-90 cm). Mean profile pH_Ca N_nit increased to 5.47 in 2004, while the N_urea pH_Ca remained similar in all 4 years (4.86-4.93). In both 2001 and 2002, pH_Ca for N_nit between 0 and 50 cm was higher by up to 0.8 than the control and N_urea treatments, which were similar (P <0.05). The same trends were measured in 2003 (N_nit up to 0.7 pH_Ca units higher) although the differences extended deeper in the profile to 90 cm. In 2004, pH_Ca for N_nit between 10 and 90 cm was higher by up to 0.5 units than N_urea and control treatments, which were similar for all depths (Figure 1).

Soil mineral N concentrations in 2000-2003 were always low (<2 mg N kg^-1) throughout the profile during the dry season at 68 to 107 days after N application. Mean total OC for 0-30 cm in 2003 was slightly higher in N treatments (0.7-0.9%) than in the bare soil (0.6%).

Figure 1. pH_Ca in soil profile after 4 years of N treatments to grass pasture at Mareeba, Australia. The horizontal lines represent the lsd (P = 0.05) for the depths where there were significant differences

Mean dry matter yields (DM) from 4 years were higher (P <0.01) for N_nit (2.3 t ha^-1 harvest^-1) than for N_urea (1.9 t ha^-1 harvest^-1). The mean annual yield was 13.8 and 11.8 t ha^-1 for N_nit and N_urea treatments, respectively. N uptake was generally higher for N_nit (mean 235 kg N ha^-1 yr^-1) than N_urea (204 kg N ha^-1 yr^-1) but these differences were not significant (P >0.05). Mean N concentration was 1.79% for both N treatments.

Mean inorganic N concentrations in soil water samples at a depth of 1 m were very low during the 3.5 years of measurement. NH₄-N concentration was 0.06 mg L^-1 (range <0.01-2.5 mg L^-1) and NO₃-N concentration was 0.3 mg L^-1 (range <0.01-32 mg L^-1). Losses of inorganic N by leaching were corres­pondingly low. Average annual losses were <1 kg ha^-1 for NH₄-N and 2 to 4 kg ha^-1 of NO₃-N when 350 kg N ha^-1 was applied.

Tully

After N applications for 1 year, pH_Ca was significantly higher (P <0.05) in the N_nit treatment than all other treatments at 0-10 cm depth only (Figure 2). There was a trend for pH_Ca in this treatment to be higher than all other treatments down to the 20-30 cm depth but this was not significant. Soil mineral N concentrations at the final plant harvest during the dry season in 2003 were low (NH₄-N <3 and NO₃-N <1 mg kg^-1) throughout the profile, 35 days after N application.

There were 5 harvests between September 2002 and April 2003. N treatments significantly increased DM compared to the control for 3 harvests, but yields where there were no statistical differences were not obviously constrained (6.1 and 5.1 t/ha for harvests on 15/1/2003 and 29/4/2003, respectively). For both forms, the lowest rate of N (25 kg N ha^-1 application^-1) doubled the yield of the control to an average of 26 t ha^-1. Further increases in N application produced large increases in DM. Applications of 50 and 75 kg N ha^-1 application^-1 produced average DM of 33 and 37 t ha^-1, respectively, but differences among rates of N were significant for only 3 harvests (P <0.05). For all rates of N, the N_nit treatments had slightly higher DM yields (by 6-10%) than the N_urea treatments, although this difference was not significant. N concentrations generally increased with rate of applied N. Mean concentrations for 0, 25, 50 and 75 kg N ha^-1 application^-1 were 0.73, 0.86, 1.13 and 1.27%, respectively. N concentration for the highest rate of N (75 kg N ha^-1 application^-1) was significantly higher than N at 25 kg N ha^-1 application^-1 at each harvest for N_nit and for 3 harvests for N_urea (P <0.05). N uptake for all harvests also generally increased with rate of N application. Mean uptake for 0, 25, 50 and 75 kg N ha^-1 application^-1 were 97, 216, 379 and 471 kg N, respectively. N uptake for the highest rate of N (75 kg N ha^-1 application^-1) was significantly higher than N at 25 kg N ha^-1 application^-1 at each harvest for N_nit and for 3 harvests for N_urea (P <0.05).

Figure 2. pH_Ca in soil profile after 1 year of N treatments (675 kg N/ha) at Tully, Australia. The horizontal line represents the lsd (P = 0.05) for the depths where there were significant differences

Chiang Yuen

At the first soil sampling in December 2001 after one N application, there were no differences in pH _Ca in the profile (mean pH_Ca 4.01) among N treatments. In 2002 after another three N applications (total of 123 kg N ha^-1), pH_Ca for the ammonium treatment was lower in the 0-10 and 10-20 cm layers (3.99 and 3.89, respectively) than all other treatments, which were similar to each other (4.13-4.27 and 3.97-4.06, respectively). In 2003 after a total N application of 289 kg N ha^-1, pH_Ca for unfertilised grass and nitrate fertilised grass was higher (P <0.05) in the 0-10 cm layer than the bare and ammonium fertilized treatments (Figure 3).

The plots were harvested 6 times between December 2001 and November 2003. Mean DM was 4.2 t ha^-1 for unfertilized Gamba grass. N increased DM by 180% to an average of 7.7 t ha^-1 harvest^-1. This difference was significant for 5 of the harvests (P <0.05). Sources of N produced similar DM except in November 2003 when DM N_ammon was higher than N_nit (P <0.05). Plant N concentration and N uptake for the N treatments were usually similar (P >0.05) but higher than the nil N treatment for 5 harvests. N application increased N concentrations in grass from a mean of 0.39% (s.e. ± 0.05) in unfertilized treatment to 0.58% (s.e. ± 0.14) in both N_nit and N_ammon treatments. N uptake over the 6 harvests increased from 102 kg N ha^-1 in the unfertilized treatment to 283 kg N ha^-1 in the N fertilized treatments.

Figure 3. pH_Ca in soil profile after 2 years of N treat­ments at Chiang Yuen, Thailand. The horizontal line represents the lsd (P = 0.05) for the depths where there were significant differences

Discussion

N fertilizer applied in the nitrate form increased the soil pH at each site. The depth and magnitude of alkalisation was greatest at the Mareeba site, presumably due to the high rate of N (1,400 kg ha^-1) applied over 4 years. N_nit increased pH_Ca by as much as 0.5 units to a depth of 90 cm over this time. This is equivalent to 2.7 t ha^-1 of lime but relatively uniformly distributed down the profile. The pH increase was attributed to alkali (OH^- and/or HCO₃^-) extrusion by roots accompanying NO₃^- uptake (Jarvis and Robson, 1983). Leachate measurements with the lysimeter system showed that very little N was leached below 1 m, thus maximising the opportunity for plant uptake. Leaching data were not available at the other sites, but may have been important at Tully because of high rainfall (1,812 mm during the experiment) and some high N application rates. It is not clear why the pH_Ca of unfertilized grass increased to the same extent as the nitrate fertilized grass at Chiang Yuen.

An important result is the ability of nitrate fertilizers to remediate soil acidity to considerable depth in the profile. This is not usually the case with liming products applied to the soil surface. For example, Moody and Aitken (1997) found that regular applications of lime (2.5 t/ha/yr) in a hot, wet environment similar to the Tully site did not prevent acidification in 4 of 5 banana soils to >100 cm. In contrast, Noble and Hurney (2000) found that surface lime applied 18 years earlier increased pH to 100 cm in a Yellow Dermosol. They attributed this to ion pairs of Ca²⁺ and Mg²⁺ with NO₃^- moving through the profile, subsequent uptake of nitrate by roots and secretion of an excess of bases by roots to increase pH.

Nitrate fertilizers are more expensive than urea and there are now restrictions on their purchase in Australia. Calcium nitrate and potassium nitrate cost respectively 8 and 10 times more than urea per unit of N, although these obviously also supply other important plant nutrients. The increased cost of using nitrate-N may be partially offset by significantly higher yields at Mareeba (an increase of 17%) and a trend to higher yields at Tully (an increase of 6-10%).

References

Farina, M.P.W. and Channon, P. 1988. Acid subsoil amelioration: I. A comparison of several mechanical procedures. Soil Science Society of America Journal 52, 169-175.

Helyar, K.R. and Porter, W.M. 1989. Soil acidification, its measurement and the processes involved. In: Robson, A.D., ed., Soil Acidity and Plant Growth. Academic Press Australia.

Isbell, R.F. 1996. The Australian Soil Classification. Collingwood, Australia. CSIRO.

Jarvis, S.C. and Robson, A.D. 1983. The effect of nitrogen nutrition of plants on the development of acidity in West Australian soils II. Effects of differences in cation/anion balance between plants grown under non leaching conditions. Australian Journal of Agricultural Research, 34, 355-365.

Moody, P.W. and Aitken, R.L. 1997 Soil acidification under tropical agricultural systems. 1. Rates of acidification and contributing factors. Australian Journal of Soil Research, 35, 163-173.

Noble, A.D., Cannon, M. and Muller, D. 1997. Evidence of accelerated soil acidification under Stylosanthes-dominated pastures. Australian Journal of Soil Research, 35, 1309-1322.

Noble, A.D. and Hurney, A.P. 2000 Long-term effects of lime additions on sugarcane yield and soil chemical properties in North Queensland. Experimental Agriculture, 36, 397-413.

Rayment, G.E. and Higginson, F.R. 1992. Australian Laboratory Handbook of Soil and Water Chemical Methods. Sydney. Inkata Press, 330 p.

Ritchie, G.S.P. 1989. The chemical behaviour of aluminium, hydrogen and manganese in acid soils. In: Robson, A.D., ed., Soil Acidity and Plant Growth. Academic Press Australia.

¹ Department of Natural Resources & Mines, Mareeba, Australia. [email protected]
² CSIRO Land and Water, Townsville, Australia. [email protected]
³ Department of Land Resources and Environment, Khon Kaen University, Thailand. [email protected]
⁴ Department of Natural Resources & Mines, Brisbane, Australia. [email protected]
⁵ International Water Management Institute, Bangkok, Thailand. [email protected]

Influence of afforestation with eucalypts in Congolese savannas on long-term nutrient availability in the soils

Laclau, J.-P.¹; P. Deleporte ²; J.-P. Bouillet ¹and J. Ranger ³

Keywords: Eucalyptus, savanna, biogeochemical cycles, nutrient budget, sustainability, Congo

Introduction

Large areas of native vegetation have been replaced by monospecific Eucalyptus stands for several decades, and this genus is nowadays the most represented in tropical plantation forests. The ecological impact of Eucalyptus plantations has been widely discussed around the world (Cossalter and Pye-Smith, 2003). This issue has been studied in littoral plains of Congo, after afforestation with Eucalyptus clones in native savannas (Loumeto and Huttel, 1997; Mboukou-Kimbatsa et al., 1998; Bernhard-Reversat et al., 2001; Ranger et al., 2004).

Paired comparisons of soil chemical properties in the top soil under savannas and adjacent Eucalyptus stands showed that 20 years of short-rotation silviculture did not modify carbon stocks but decreased N and Ca contents (Bouillet et al., 2001). To gain insight into the processes governing the mineral functioning of these ecosystems, the biogeochemical cycles of nutrients have been compared in a clonal Eucalyptus stand and an adjacent savanna over three years: modifications of the chemical composition of solutions throughout their transfer were investigated in the two ecosystems (Laclau et al., 2003a, b), as well as the dynamics of biomass and nutrient accumulation in the plants (Laclau et al., 2002; Laclau et al., 2003c). The flux dynamics is an important parameter to consider, because nutrient availability and demand should coincide both in time and space under conditions of high rainfall and permeable soils.

The present study aimed at establishing input-output nutrient budgets for the whole Eucalyptus rotation in order to assess the long-term effects of silvicultural practices on nutrient availability and to elaborate sustainable management practices.

Material and methods

1. Site characteristics

The study site was located on a plateau at an elevation of about 80 m and a distance from the sea of 10 km (4ºS 12ºE). The mean annual rainfall over the last 50 years was 1,200 mm, with a dry season between May and September, and the mean temperature was 25ºC with seasonal variations of approximately 5ºC. The geological bedrock is composed of thick detritic formations of continental origin, dated from plio-pleistocene. An experimental design monitoring the biogeochemical cycles was installed in 1997 in a Eucalyptus stand and a native savanna. The area was flat and the distance between the two designs was about 500 m. Soils were ferralic arenosols (FAO classification), acidic (pH_H2O ≈ 5) and characterized by a sandy texture (sand content >85% down to a depth >12 m) and a low amount of available nutrients (CEC <0.8 cmol_c kg^-1, even in the upper layers). The organic matter content of soils decreased in both stands from about 1.3% in the surface layer (0-5 cm) to about 0.15% below a depth of 2 m. Concentrations of exchangeable elements were of the same order of magnitude in the two stands. Nevertheless, in the topsoil (0-5 cm layer), the concentration of Ca in the eucalyptus stand was lower than in the savanna and the concentrations of available P and Al were higher (Laclau et al., 2005).

2. Plant material

The Eucalyptus clone studied here comes from natural crosses in Congo between a few individuals of Eucalyptus alba Reinw. ex Blume (mother tree) and a group of poorly identified Eucalyptus hybrids (father tree). The stand was planted in January 1992 on savanna, at a stocking of 530 trees per hectare, and a starter fertilization (150 g per cutting of NPK 13:13:21) was applied. Manual weeding in the planting row and chemical weeding in the inter-row were made in the early stages of stand development. The stand was 6 years old at the onset of the study, with a mean height of 26 m and a mean over-bark volume of 158 m³ ha^-1. Other studies dealing with the dynamics of nutrient fluxes throughout a 7-year rotation were performed in this area, using a chronosequence approach for the same clone (Laclau et al., 2003c).

The grass Loudetia arundinacea (Hochst.) Stend represented 80% of the total aerial biomass of the savanna, which reached about 5 t ha^-1 of dry matter at the end of the rainy season. This savanna was burnt every dry season (August) like most savannas in Congo (Laclau et al., 2002).

3. Methodology

Climatic data, soil moisture and solution chemistry were measured in both ecosystems from January 1998 to December 2000. Methods used to assess input-output fluxes were described by Laclau et al. (2005). In brief:

Atmospheric depositions. Lack of reliable measurement methods for dry deposition made it necessary to use a calculation approach based on the comparison of nutrient fluxes in an open area and beneath the canopy. We considered Na⁺ as a tracer, assuming that canopy exchange was negligible for this element compared with dry deposition of marine aerosols in this coastal area (Parker, 1983). Dry deposition of nutrients was considered proportional to that of Na⁺.

Weathering of soil minerals. Soil minerals were considered to be external to the available soil nutrient reservoir where trees take up their nutrients. Mineralogy of the different particle size fractions and mineral bearing nutrients were quantified using identification of mineral by X ray diffraction, total and selective chemical analysis, thermogravimetric analysis and normative calculation. The geochemical Profile model developed by Sverdrup and Warvfinge (1988) was calibrated for the site to estimate the magnitude of this flux, overlooking weathering processes for the very stable accessory minerals.

Accumulation in the aerial ligneous biomass. Twelve trees were sampled in the Eucalyptus stand at age 6.5 years and 10 trees at age 8 years. Selected trees were cut down, and the major components were isolated: stemwood, living and dead branches, stembark, and leaves. The fresh stem weight was measured and disks of wood and bark of constant thickness were taken every 3 meters. Samples were dried (65ºC) and the dry biomass of the components in each tree was calculated proportionally. All the branches and leaves were collected and samples of these components were dried. One composite sample for each component of each tree was ground, homogenized and sent to the laboratory for chemical analyses. Tables were established and applied to the stand inventories to quantify the stand biomass and nutrient content per hectare.

Nutrient losses by drainage. Nutrient fluxes were calculated multiplying the water fluxes at each soil depth (assessed from hydrological model) by the mean concentration of nutrients in gravitational solutions. Run off solutions were collected from 4 replicates of 1 m² collectors in the two ecosystems. Soil solutions were sampled in both ecosystems from 4 replicates of ceramic cup lysimeters installed at various depths between 15 cm and 6 m and connected to a vacuum pump maintained manually (daily checking) at a constant suction of about - 60 kPa. A complete description of the lysimetry design was presented by Laclau et al. (2003b). Three to five replicates of TDR probe (Trase System I) were installed at the depths of 15, 50, 100, 200, 300, 400 and 500 cm in the Eucalyptus stand, and at the same depths down to 3 m in the savanna. Volumetric water content was measured automatically every 3 hours from July 1998 to December 2000 in the Eucalyptus stand and once or twice a week in the savanna. A model based on the Richard’s equation for simulating one-dimensional water flows (Hydrus 1D) was calibrated in the Eucalyptus ecosystem and in the experimental savanna to quantify water flows at the depths where lysimeters were installed (Laclau et al., 2005). All the equations used in the model were established from measurements performed in 1998 and 1999. The ability of the model to predict the fluxes was checked during the year 2000.

Chemical analyses

Once a week, after volume measurements and sampling, the solutions were collected and carried to the laboratory where they were kept at +4°C. Pooled volume weighted samples were made every 4 weeks for chemical analyses. Each replicate of soil solution collected was analyzed separately. The solutions were filtered (0.45 |j,m) in the Congo and measurements of pH (HI 9321) and SO₄²- by colorimetry (ANA 8 Prolabo) were performed as quickly as possible. The samples were then acidified with H₂SO₄, sent to the CIRAD laboratory in France where nitrate and ammonium were measured by colorimetry (INTEGRAL PLUS - Alliance instruments). Total P, K, Ca, Mg were determined by ICP emission spectroscopy (JY 50).

In plant samples, N was determined by thermal conductivity after combustion (FP-428) and P, K, Ca, Mg, by a sequential spectrometer ICP (JY 24) after digestion by hydrofluoric acid and double calcination.

Input-output budgets

Current annual and seasonal input-output budgets were established between 1998 and 2000 from the measurements performed, considering losses by deep drainage at a depth of 4 m under savanna and 6 m under Eucalyptus. Indeed, previous studies showed the absence of roots beyond the depth of 3 m under savanna, and extremely low densities in this Eucalyptus stand (Laclau et al., 2002). The nutrient uptake by plants considered in the current budgets was the permanent uptake: immobilization in stemwood for the Eucalyptus stand and transfers to the atmosphere during burning in the savanna.

The calculation of nutrient budgets over the whole rotation required assessing the nutrient fluxes during the first years after afforestation. To measure losses by drainage during the early growth of the stand, the experimental savanna was planted with the same Eucalyptus clone in May 2001. Soil solutions were collected and analyzed for 2 years with the same methodology. Losses were assessed every 4 weeks, multiplying the average concentration in soil solutions at a depth of 4 meters by the simulated water flow (Unpublished data). Dry deposition was adjusted proportionally to the foliar biomass of the stand throughout stand rotation (Laclau et al., 2000). From 2 years onward, annual fluxes in the Eucalyptus ecosystem were considered identical to values measured at the end of the rotation. Indeed, a chronosequence approach showed that the main fluxes of the biological cycle are roughly constant (Laclau et al., 2003c).

Results and discussion

1. Main changes in nutrient cycling after afforestation

Planting Eucalyptus in the native savanna modified the inputs of nutrients to soil through the following processes:

A filter effect of the canopy at the end of the Eucalyptus rotation led to dry depositions of nutrients of the same order of magnitude as wet deposition, whereas the pattern of Na⁺ concentration in wet depositions and throughfall suggested that this flux was negligible in the savanna (Laclau et al., 2003a). Atmospheric deposition was in the range of values given in the literature for forest stands. However, this flux was determined with high degree of uncertainty which influence the accuracy of determination of input-output budgets at the ecosystem level (Laclau et al., 2005).

Fertilizers are usually applied in Eucalyptus plantations. This flux was considered in the budgets calculated for the whole rotation (Figure 1) but not in Table 1, because fertilizers were applied only at planting in the stand studied.

Biological N₂ fixation is necessary to explain the sustainability of savannas in littoral areas of Congo for about 3,000 years showed by isotopic studies (Trouvé, 1992), despite large losses of N during annual fires. The input of N by symbiotic fixation, assessed roughly to balance the N budget in savanna, amounted to about 22 kg ha^-1 year^-1 (Table 1). The legume species Eriosema erici-rosenii R.E. Fries (Papilionoideae) was found in all the savannas of the region. Chemical weeding during the early growth of the Eucalyptus stand led to the disappearance of this flux.

Calculations performed with the Profile model suggested that the amounts of P, Ca and Mg released by the weathering processes were negligible in this Ferralic Arenosol. The amount of K released on a soil depth of 6 m in the Eucalyptus ecosystem was about 0.3 kg ha^-1 year^-1 and half on a depth of 3 m under savanna. This flux was consistent with the extremely low nutrient concentrations in soil solutions sampled by tension lysimeters below the rooting zone (Laclau et al., 2003b).

Figure 1. Input-output budgets (kg ha^-1) of N, P, K, Ca and Mg for the whole Eucalyptus rotation, and for different harvesting scenarios
Scenario 1: de-barked pulpwood harvest
Scenario 2: de-barked pulpwood and firewood harvest
Scenario 3: pulpwood with bark harvest
Scenario 4: whole tree harvest

Outputs of nutrients also were modified by afforestation:

Transfers of plant nutrients to the atmosphere during burning were a major output of nutrients in thi savanna. They amounted to 23.4, 1.5, 2.4, 2.6 and 2.9 kg ha^-1 year^-1 for N, P, K, Ca and Mg, respectively (Laclau et al., 2002). They represented, respectively, 85%, 25%, 39%, 21%, and 28% of the N, P, K, Ca and Mg contents in the above-ground biomass before the fire. This output of nutrients was avoided after afforestation.

Table 1. Mean input-output fluxes of nutrients in the soil under the Eucalyptus stand at the end of the rotation, and under the savanna, from 1998 to 2000 (kg ha^-1 year^-1)

¹ Calculated to balance the nitrogen budget in the savanna,
² Nutrient immobilization in stemwood.

Nutrients immobilized in stemwood were exported from the ecosystem at the harvest. They amounted to about 30 kg ha^-1 year^-1 of N, and from 3 to 5 kg ha^-1 year^-1 of P, K, Ca and Mg, at the end of stand rotation (Table 1). Nutrient exportation with biomass removal does not occur in most of Congolese savannas owing to the absence of cattle.

Losses of nutrients by surface runoff were negligible in savanna and represented very low amounts in the Eucalyptus stand (<0.5 kg ha^-1 year^-1). At the end of the rotation, the Eucalyptus stand was as efficient as the native ecosystem of savanna to prevent losses of nutrients by deep drainage. The losses of N, P, K, Ca and Mg amounted to 4.3, 0.3, 2.1, 1.1 and 1.2 kg ha^-1 year^-1, in the Eucalyptus stand, respectively, and 3.0, 0.1, 0.6, 0.4 and 0.2 kg ha^-1 year^-1 in savanna (Table 1). Losses of nutrients by deep drainage observed over the first two years after afforestation of the experimental savanna remained low. Although the mineralization of savanna residues occurred when the root system of the trees was not completely established, the fluxes of N, P, K, Ca and Mg at a depth of 4 m were lower than 3 kg ha^-1 year^-1 on average (Unpublished data). Rainfall amounts were much lower during this period than from 1998 to 2000 which may explain the very low losses by drainage observed. Therefore, it cannot be excluded that nutrient outputs by drainage would be higher, and consequently the N budget more unbalanced, if afforestation occurred during a period with a more normal precipitation distribution.

Moreover, afforestation led to great changes in the internal nutrient cycling within the ecosystem (Laclau et al., 2003a, c). Foliar leaching of cations observed in savanna decreased after planting eucalypts. Plant uptake of nutrients from the soil increased sharply after afforestation. Internal retranslocations of nutrients occurred in the savanna but were not quantified. This process supplied about 30% of the annual requirements of N and P in the Eucalyptus plantation from 2 years of age onwards, and about 50% of K requirements. Litter fall and litter decomposition were negligible in the savanna ecosystem and became important nutrient fluxes from age 2 years onwards, in the Eucalyptus stands.

The biogeochemical cycle of N was the most affected by afforestation. In particular, both mineralisation and nitrification rates permanently increased after planting eucalypts (Ranger et al., 2004). The immediate increase in nitrification rate observed after the destruction of the savanna was interpreted as a drastic change in the control of the nitrifying populations. Savanna vegetation was known to inhibit the activity of nitrifiers by allelopathic processes not yet completely elucidated. Investigations made in Ivory Coast showed the role of grass ecotypes on the control of nitrifiers (Abbadie et al., 2000). Allelochemicals responsible for the inhibition of nitrifiers were not yet identified in the savannas, contrarily to other specific forest situations (Paavolainen et al., 1998).

2. Input-output budgets

Harvesting method had a great influence on the nutrient budgets (Figure 1). The range of variation between the most conservative harvesting method (scenario 1) and the most costly in nutrients (scenario 4) was about 180 kg ha^-1 for N, 25 kg ha^-1 for P, 55 kg ha^-1 for K, Ca, and 30 kg ha^-1 for Mg. De-barking the stems on site retained at the soil surface 31, 9, 21, 28 and 16 kg ha^-1 of N, P, K, Ca and Mg, respectively. These values represented about 10% of the amount of N accumulated in the above-ground part of the trees at harvest, 20% of that of P and K, and 35% of that of Ca and Mg. The removal of firewood for surrounding populations in Congo (scenario 2) led to further losses of 50, 8, 20, 9 and 7 kg ha^-1 of N, P, K, Ca and Mg, respectively, relative to the most conservative method where only de-barked pulpwood is harvested. The current silviculture in Congo led to a deficit of 144 kg ha^-1 of N for the first rotation after afforestation. This deficit represented about 7% of the initial amount of total N in the A₁ horizon (0-50 cm) under savanna. In highly-weathered sandy soils, the long-term sustainability of these plantations is therefore greatly dependent on the reliability of fertilization practices.

Even if certain fluxes were assessed with large uncertainty, input-output budgets demonstrate clearly that Eucalyptus plantations take advantage, during the first rotation after afforestation, of a N soil fertility inherited from the previous vegetation of savanna. Unfavourable qualitative changes add further to the quantitative deficit of the N budget: savanna organic matter is progressively replaced by Eucalyptus organic matter poor in N (Trouvé et al., 1994), and whose chemical composition (tannins, lignin, polyphenols) leads to a slower mineralization (Bernhard-Reversat et al., 2001). For the other elements, the budgets for the whole rotation were well balanced relative to the amounts of available elements in the soil. This behaviour is consistent with fertilizer field trials in this area, which show that tree responses to N inputs increase over successive rotations, whereas no response to P and K inputs is observed, even in replanted sites 20 years after savanna conversion (Bouillet et al., 2003).

3. Consequences for sustainable silivicultural practices

Low amounts of nutrients in the soils of this area (P excepted), and the high cost of fertilizer inputs make it essential to strictly limit nutrient losses throughout stand rotation. Several modifications in silvicultural practices were proposed to achieve this goal:

• Field trials of fertilization. The quantitative data from the budgets show that field trials should focus on N fertilization. Future plantations, with much more productive clones of the hybrid E. urophylla × E. grandis, might lead to unbalanced budgets of K, Ca and Mg in the soils. It would then be important to determine whether these elements become limiting after several rotations.
• Soil preparation and weed control. Minimum cultivation is recommended to limit nutrient losses by erosion and planting must occur as quickly as possible after harvesting to reduce nutrient losses by drainage. Moreover, weed control must be planned to take advantage of the temporary fixation of nutrients in the biomass of weeds during the early growth of the stands (Nambiar and Sands, 1993).
• Harvesting method. The effects of various harvesting scenarios on nutrient budgets were quantified. They show that current practices including de-barking on site is fundamental owing to the chemical paucity of the soil. This feature was confirmed by an experiment dealing with organic matter management in Congolese Eucalyptus plantations (Nzila et al., 2002).
• Fire prevention. Nutrient budgets provide new light on the negative effects of fires in these plantations. Large losses of N by volatilization during burning have clearly detrimental consequences on the long-term production in this area where N is the first nutritional limiting factor. Effective fire prevention is therefore crucial for the sustainability of these plantations.
• Introduction of a legume understorey. Numerous studies show that mixed plantations between Eucalyptus and legume species can have beneficial effects on soil N fertility (e.g. Binkley et al., 2003). In the Congo, mixed plantations cannot be established because the growth of N fixing trees is much lower than that of Eucalyptus. In this case, experiments were set up recently introducing an understorey of Acacia in Eucalyptus stands to enhance soil fertility, through inputs of organic matter and atmospheric N.

Conclusion

Quantification of nutrient fluxes throughout a rotation of Eucalyptus in the Congo demonstrated that the influence of silvicultural practices varied greatly according to the elements. Whereas the amounts of P, K, Ca and Mg in the soil were roughly stable throughout stand rotation, current silvicultural practices led to a deficit of about 140 kg ha^-1 of N in the soil. The budgets were strongly dependent on the harvesting method because this period accounted for the major output flux from the system. Input-output budgets suggested that Eucalyptus stands benefit from a N fertility inherited from the previous ecosystem of the savanna. Weeding destroyed a legume species responsible for N input in the savanna ecosystem estimated at around 20 kg ha^-1 year¹.

Therefore, the sustainability of Eucalyptus plantations in this area will require an increase in N fertilizer inputs over successive rotations. Another option to improve the N status in these soils might be to introduce a biological nitrogen fixing species, compensating for the destruction of the native legume species in the savanna. Several experiments have been set up recently in the Congo to assess the influence of various Acacia species introduced as understorey in Eucalyptus stands. Further research is necessary to investigate silvicultural practices providing a positive influence of a leguminous understorey on soil N availability in the long term, without competing significantly with eucalypts during the early growth of the stands.

References

Abbadie, L., Lata, J.C., Tavernier, V. 2000. Impact des graminées pérennes sur une resource rare: l’azote. In: Ch., Floret and R., Pontanier, ed., 2000. La jachère en Afrique tropicale. Paris, John Libbey Eurotext, 189-193.

Bernhard-Reversat, F., Loumeto, J.J., Laclau, J.-P. 2001. Litterfall, litter quality and decomposition changes with eucalypt hybrids and plantation age. In: Bernhard-Reversat, F., ed., 2001. Effect of exotic tree plantations on plant diversity and biological soil fertility in the Congo savanna: with special reference to Eucalypts. Bogor, Center for International Forestry Research, 23-29.

Binkley, D., Senock, R., Bird, S., Cole, T.G. 2003. Twenty years of stand development in pure and mixed stands of Eucalyptus saligna and nitrogen-fixing Falcataria moluccana. Forest Ecology and Management, 182, 93-102.

Bouillet, J.-P., Safou-Matondo, R., Laclau, J.-P., Ranger, J., 2001. Changes in N-P-K fertilisation over successives rotations in Eucalypt plantations in the Congo. In: Developing the eucalypt of the future. Proceedings of an international seminar, Valdivia, 11-15 September 2001, IUFRO, 10 p.

Bouillet, J.-P., Safou-Matondo, R., Laclau, J.-P., Nzila, J.D., Ranger, J., Deleporte, P. 2003. Pour une production des plantations d’eucalyptus au Congo: la fertilisation. Bois et Forêts des Tropiques, 279, 23-35.

Cossalter, C, Pye-Smith, C. 2003. Fast-wood forestry myths and realities. Bogor, Center for International Forestry Research, 50 p.

Laclau, J.-P., Sama Poumba, W., Nzila, J.D., Bouillet, J.-P., Ranger, J. 2002. Biomass and nutrient dynamics in a littoral savanna subjected to annual fires in Congo. Acta Oecologica, 23, 41-50.

Laclau, J.-P., Ranger, J., Bouillet, J.-P., Nzila, J.D., Deleporte P. 2003a. Nutrient cycling in a clonal stand of Eucalyptus and an adjacent savanna ecosystem in Congo. 1. Chemical composition of rainfall, throughfall and stemflow solutions. Forest Ecology and Management, 176, 105-119.

Laclau, J.-P., Ranger, J., Nzila, J.D., Bouillet, J.-P., Deleporte P. 2003b. Nutrient cycling in a clonal stand of Eucalyptus and an adjacent savanna ecosystem in Congo. 2. Chemical composition of soil solutions. Forest Ecology and Management, 180, 527-544.

Laclau, J.-P., Deleporte, P., Ranger, J., Bouillet, J.-P., Kazotti, G. 2003c. Nutrient dynamics throughout the rotation of Eucalyptus clonal stands in Congo. Annals of Botany, 91, 879-892.

Laclau, J.-P., Ranger, J., Deleporte, P., Nouvellon, Y., Saint-André, L., Marlet, S., Bouillet, J.-P. 2005. Nutrient cycling in a clonal stand of Eucalyptus

and an adjacent savanna ecosystem in Congo. 3. Input-output budgets and consequences for the sustainability of the plantations. Forest Ecology and Management, 210, 375-391.

Loumeto, J.J., Huttel, C. 1997. Undestory vegetation in fast-growing tree plantations on savanna soils in Congo. Forest Ecology and Management, 99, 65-81.

Mboukou-Kimbatsa, I.M.C., Bernhard-Reversat, F., Loumeto, J.J. 1998. Change in soil macrofauna and vegetation when fast-growing trees are planted on savanna soils. Forest Ecology and Management, 110, 1-12.

Nambiar, E.K.S., Sands, R. 1993. Competition for water and nutrients in forests. Can. J. For. Res., 23, 1955-1968.

Nzila, J.D. 2001. Caractérisation minéralogique des sols ferrallitiques sableux sous plantation d’Eucalyptus et sous savane naturelle de la région de Pointe-Noire (Congo). Pointe-Noire, UR2PI, 51 p.

Nzila, J.D., Bouillet, J.P., Laclau, J.P., Ranger, J., 2002. The effect of slash management on nutrient cycling and tree growth in Eucalyptus plantations in the Congo. Forest Ecology and Management, 171, 209-221.

Paavolainen, L., Kitunen, V., Somlander, A. 1998. Inhibition of nitrification in forest soil by monoterpenes. Plant and Soil, 205, 147-154.

Parker, G.G. 1983. Throughfall and stemflow in the forest nutrient cycle. Adv. Ecol. Res., 13, 57-120.

Ranger, J., Jussy, J.H., Moukoumi, J., Zeller, B., Bouillet, J.-P., Deleporte, P., Nzila, J.D., Laclau, J.-P. 2004. Control of the N-mineralising activity by vegetation is a key-parameter for forest management. In: Carnus, J.M., ed., 2004. Forests soils under global and local changes: from research to practice. Proceedings of an international seminar, 15-18 september 2004. Bordeaux, IUSS-IUFRO.

Sverdrup, H., Warfvinge, P. 1988.Weathering of primary silicate minerals in the natural soil environment in relation to a chemical weathering model. Water, Air, and Soil Pollution, 38, 387-408.

Trouvé, C. 1992. Apport de la géochimie isotopique (8¹³C) à l'etude du renouvellement des matières organiques et des sucres neutres dans les sols tropicaux soumis à des changements d’écosy stèmes. Cas des aménagements forestiers sur les savanes de Pointe-Noire au Congo. PhD Thesis. University of Orléans, 112 p.

¹ CIRAD, UR80, département Forêts, Programme Arbres et Plantations, TA 10/C, 34398 Montpellier Cedex 5. Fax: (33) 4 67 59 37 33 E-mail: [email protected]; jpbouillet @cirad.fr
² CIRAD/UR2PI, BP 1291, Pointe-Noire, République du Congo. Fax: (242) 94 47 95 E-mail: [email protected]
³ INRA, Biogéochimie des écosystèmes forestiers, 54280 Seichamps, France. Fax: (33) 5 83 39 40 68 E-mail: [email protected]