Effects of land use changes on soil chemical properties of sandy soils
from tropical Hainan, China

Wu, W.¹; M. Chen¹and B. Sun²

Keywords: soil chemical properties, sandy soils, Hainan, changed land use

Introduction

Hainan Island is the southernmost part of China, rising from the vast South China Sea, with an area of approximately 34,000 km². It occupies 42% of the tropical area of China with an annual average temperature of 22º-26ºC and an annual rainfall of 1,500-2,000 mm. Highly weathered ferralsols and ferrisols are the predominant soils on the island. Land use patterns have dramatically changed in the last decades due to deforestation and exploitation for agricultural and forestry production. Natural tropical rainforests covered 35% of the island in the early 1950s. However, it has declined to 18% at present as a result of the “rubber plantation fever” in the 1950s, the “land reclamation fever” in the 1960s, the “crop breeding fever” in the 1970s and the “opening to the outside world fever” in the 1980s. Changes in land use patterns have disturbed natural nutrient cycles and the balance within native ecosystems that have significantly modified soil quality (Lin and Zhong, 2002; Ma et al., 2000; Zhao et al., 2005).

Soils derived from coarse texture materials including granite, sandstone, shallow-sea and coastal deposits, river alluvium, and purple sandstone, amount to more than 84% (granite accounts for 46.7%, sand­stone for 20%, and sea deposit 12.1%, respectively) of the area of Hainan Island, resulting in these soil being predominantly of a sandy texture. Data on soil nutrient content, organic matter contents, cation exchange capacity (CEC), base saturation (BS) and pH of such sandy soils were analysed with the goal of under­standing the consequences of the changes in land use patterns.

Material and method

Data collection

In Hainan Island, after the primary forest (PF) is cleared, the land becomes a secondary forest (SF), an artificial forest (AF), rubber tree plantations (RP), orchards (OC), dry land (DL) or rice fields (RF). Soil chemical properties in the topsoil of sandy soils developed from granite and gneiss from different land use patterns were collected from our databases and published materials, including the Second National Soil Survey, which was conducted in the early 80s at county, provincial and national level (Gong et al., 2004; Soil and Fertilizer Experimental Station, Agricultural Department, Hainan Province, 1993). The number of samples collected for each land use pattern and the main characteristics of each land use pattern are shown in Table 1.

Table 1. Data sample collected and description

Data analysis

The initial stage of the analysis was to compare the soil chemical properties using the Tukey Test. The results of the statistical comparisons were further examined using boxplots for the range of values from the individual samples. These plots are used in a manner analogous to the control charts discussed by Larson and Pierce (1994); basically the range of values for the primary forest provides a measure of the undisturbed or natural range of variability for the local soil quality indicators. The range of values of land use patterns can then be compared to this natural range to assess the possible severity of the observed differences (Pennock and Kessel, 1996).

Results and discussion

The first stage of analysis was to compare the soil chemical properties under the different land use patterns in the studied area. The second stage was to compare the soil quality.

Soil organic matter and total N

Soil organic matter and total N decreased after primary forest was cleared for other land uses. There were significant differences in soil organic matter and total N contents between the primary forest and the other land use types, SF, AF, RP, OC, DL and RF (Table 2).

Table 2. Comparison of soil organic matter and total N contents of different land use patterns

The crude statistical assessment of changes in soil organic matter and total N contents for land use patterns can be placed in a larger context by examining the observed changes relative to the range of values of the primary forest. This was done by comparing the boxplots for different artificial land use types with those for primary forest. For soil organic matter contents, the median levels for the land use patterns SF, OR and RF are clearly falling into the range defined by the highest and lowest median levels of the primary forest, hence these land use type levels are clearly within the natural or ecological range defined by the local primary forest (Figure 1). However the other three land use types, AF, RP DL are below the minimum of primary forest, indicating that they are outside of the range of primary forest. When total soil N was assessed by using the boxplot method, it can be seen in Figure 2 that OR and RF are well inside of the range defined by primary forest, while SF, AF, RP and DL are below the lowered limit of the range.

Figure 1. Boxplots of soil organic matter contents of different land use pattern

Figure 2. Boxplots of soil total N contents of different land use patterns

Soil P and K

Results of statistical analysis shows that there are no remarkable changes in total soil P and K content after primary forests were changed to SF, AF, RP, OC, DL and RF (Table 3). When they were compared with the defined range of primary forest, the median values of them are not outside its maximum and minimum limits (Figure 3 and Figure 4).

Table 3. Comparison of soil total P and K contents of different land use patterns

Tukey t tests indicated no significant difference in available soil P contents between SF, AF, RP, OC and PF, but show an increase for RF and DL (p <1%) (Table 4). However, the boxplots method in Figure 5 indicates no median values of available P for SF, AF, RP, OC, DL and RF exceed the defined range for PF (Table 5). For available soil K, no significant differences appeared when all other land use patterns were compared with PF (Table 4) and the medians of them are also limited within the defined limits by PF (Figure 6).

Figure 3. Boxplots of total soil P for different land use patterns

Figure 4. Boxplots of total soil K for different land use patterns

Table 4. Comparison of soil available P and K contents of different land use patterns

Table 5. Comparison of soil CEC of different land use patterns (cmol_c/kg)

Figure 5. Boxplots of available soil P contents of different land use patterns

Figure 6. Boxplots of available soil K contents of different land use patterns

Soil chemical quality properties

Soil cation exchange capacities for SF, AF, RP, OC, DL and RF are significantly lower (p <1%) than that for PF (Table 5). The former are all below the minimum range defined by PF (Figure 7). There were no obvious differences both in soil pH and base saturation between SF, AF, RP, OC, DL and PF, except for RF, which is significantly higher (p <1%) than PF in them (Table 6). Figure 8 and 9 also show that the medians for pH and base saturation for SF, AF, RP, OC, DL are plotted within the maximum and minimum range for PF, while that for RF are above the maximum limits of the ranges.

Table 6. Comparison of soil pH and BS of land use patterns

Figure 7. Boxplots of soil cation exchange capacities of different land use patterns

Figure 8. Boxplots of soil pH of different land use patterns

Figure 9. Boxplots of soil base saturations of different land use patterns

Conclusions

Change in land use pattern from primary forest to others leads to remarkable decrease (p <1%) in soil organic matter, total N, and cation exchange capacity in sandy soils developed from granite in tropical Hainan. Except for the decrease in soil organic matter for the secondly forest, orchard and rice field, and the decrease in soil total N for orchard and rice field that do not go out of the defined range for primary forest, the decreases in soil organic matter, total N and cation exchange capacity for all the other land use forms bring them out of their natural range defined by primary forest.

Succession in land use pattern from primary forest to human influenced land use types does not lead soil total P and K, available P and K, pH and base saturation being outside the defined range for primary forest, except for dry land that its soil available P is above the upper limit of the range and for rice field that its pH and base saturation are higher than the maximum for primary forest.

References

Pennock, D.J. and Van Kessel, C. 1997. Clear-cut forest harvest impacts on soil quality indicators in the mixed wood forest of Saskatchewan, Canada Geoderma 75 13~32.

Lin, M.Z., and Zhong, Y.L. 2002. Dynamic change of tropical forest in Hainan Island (in Chinese). Geographical Research Vol. (6): 703~712.

Ma, R.H., Hu, M.C., Mao, D.Q., and Huang, X.Y. 2000. Study on the dynamic change of land sand desertion and land degradation in the west of Hainan sland based on RS and GIS (in Chinese). Ecologic Science, Vol. 19(2): 19~23.

Soil and Fertilizer Experimental Station, Agricultural Department, Hainan Province, 1993. Soils in Hainan (in Chinese). Hainan Publisher.

Zhao, Y.G., Zhong, G.L., Gong, Z.T. and Deng, W.G. 2005. Soil type, soil quality and crop suitability of soil developed from different geology environments in Hainan Island (in Chinese). Quaternary Science. Vol. 25(3): 389~39

Gong, Z.T., Zhong, G.L., and Qi, X.P. 2004. Soil series in Hainan Island (in Chinese). Science Publisher, Beijing.

Larson, W.E., and Pierce, F.J. 1994. The dynamics of soil quality as a measure of sustainable management. In: J.W. Doran, D.C. Coleman, D.F. Bexdicek and B.A. Stewart (Editors), Defining Soil Quality for a 32 D.J. Pennock. C. NUI Kessel/Grodermu 75 (19971 13-32 Sustainable Environment. SSSA Special Publication Number 35. Soil Sci. Sot. Of America Inc. and Am. Sot. of Agron. Inc., Madison, WI, pp. 37-51.

¹ College of Life Science and Agriculture, Hainan University.
² Institute of Soil Science, Chinese Academy of Science.

Improvement of the saline sandy soil in Northeast Thailand
using polyvinyl alcohol (PVA)

Dejbhimon, K.¹ and H. Wada²

Keywords: aggregate, desalinization, Northeast Thailand, polyvinyl alcohol (PVA), salt-affected soil

Introduction

In Thailand, the Northeastern region is less successful in terms of economy. This is because the economy of this region is supported by agriculture which has been kept undeveloped by both hostile climate and very poor soils. Climate of this region is tropical monsoon with clear alternation of rainy and dry seasons. Surrounding mountain ranges are obstacles of the rain-laden cloud. Rainfall in this region is low and erratic, accordingly. Arable soils are sandy, infertile and acid. A rather large part of the soils are salt-affected, leading to bare lands as shown in Figure 1.

Improvement of those salt-affected soils has been regarded as a key issue for the development of agriculture and economy of this region. Accordingly, Thai scientists alone or in cooperation with foreign scientists have worked for many years in surveying and characterizing the salt-affected soils. This has resulted in several ameliorative technologies (McGowan International Pty. Ltd. 1983, Arunin 1984, Takai et al. 1987, Puengpan 1992, Subhasaram 1994). However, appropriate ameliorative technologies are still lacking for the most degraded sites characterized by 1) an elevated groundwater table, during the rainy season (where the desalinization with percolating ran water is thus suppressed) and 2) accumulation of salt by capillary rise from saline groundwater in the dry season.

Figure 1. Salt-affected soil in Northeast Thailand

The current study aims at establishing a new technology for ameliorating the salt-affected infertile sandy soil with elevated groundwater tables by conducting both laboratory experiments and field experiments. The cardinal point of the new technology is to aggregate the soil with polyvinyl alcohol (PVA) for promoting the leaching out the salt from the soil with rainwater and for suppressing the salt-accumulation by capillary rise. P VA is a non-ionic water-soluble synthetic polymer with high ability of aggregating clay soils irrespective of ambient salt concentrations (Carr and Greenland 1975): We expected PVA could aggregate even the salt-affected sandy soil in Northeast Thailand.

Materials and methods

1. Laboratory experiment

At first, three kinds of PVA were compared in their ability to aggregate two kinds of sandy soils collected in Northeast Thailand for selecting a suitable PVA and for clarifying the effective way of its application. Then, the selected PVA was examined by several methods in its effects on capillary rise of saline water and on hydraulic conductivity through the column of the aggregate.

1.1 Ability of PVA to form stable aggregate

Soil: Two sandy soil samples were collected from the plow layers of 2 upland fields identified as Korat series (Kt) (fine-loamy, siliseous, isohyperthermic Oxic Paleustults) and Yasothon series (Yt) (fine-loamy, siliseous, isohyperthermic Oxic Paleustults) which were end members of a common catena in the undulating region in Northeast Thailand. They were air-dried and passed through a sieve with 2 mm mesh.

PVA: Three kinds of PVA of different molecular weight (MW) were used. They were a chemical reagent (MW: 88,000, Wako Pure Chemical Industries Ltd. Japan) and 2 industrial products (Gohsenol, Nippon Gohsei Industry Co. Ltd.), GH-20 (MW: 100,000) and GL-30 (MW: 15,000).

Preparation of aggregate: Ten grams of each soil sample was added with 2 ml of the aqueous solution of each PVA of varying concentrations (0, 0.1, 0.25, 0.5, 1.0%). They were manually handled to make aggregates of about 2-5 mm diameter. A part of the aggregates were kept moist, the remaining part being air-dried.

Stability of aggregate: These aggregates were measured for stability against both slaking and shaking in water. A few air-dried aggregates were placed in a Petri dish and were added with a few drops of distilled water. Stability of the aggregates against slaking was judged by the immediate degree of disintegration estimated by naked eye. Ten grams of each aggregated soil were place in a 50 ml beaker and added with 50 ml of distilled water. The beaker was shaken by a rotary shaker for 30 min. and kept at a standstill on a laboratory table for 2 hrs. The amount of the dispersed clay was estimated by measuring turbidity of the solution pipetted from the supernatant in the beaker. The amount of the stable aggregates against the shaking was estimated by measuring the weight of the aggregates of 3 fractions (>500, 500-250, 250-1,501 µm) separated by wet sieving.

1.2 Properties of PVA-amended aggregate formed from the salt-affected soil

Soil: Two soil samples were collected from the surface sandy cover (sand 87.8%, silt 10.6%, clay 1.6%) and the underlying dark colored somewhat clayey semi-impermeable layer (sand 58.3%, silt 34.2%, clay 7.5%) at a salt-affected area identified as Satuk series saline variant, a member of the above mentioned catena (Puengpan 1992). Both soil samples were air-dried and passed through the sieve.

PVA: GH-20 was used.

Preparation of aggregate column: Aggregates of each soil sample were prepared by the above-mentioned way. Each of the aggregates was packed in a respective stainless core (height: 5 cm, diameter: 5 cm, capacity: 100 ml) to prepare a column of the aggregate.

Measurement: The aggregate-columns were saturated with 1 N NaCl solution by capillary rise. The rate of the capillary rise was recorded. Then the aggregate-columns were percolated at first with the saline water and then with fresh water for measuring 2 kinds of hydraulic conductivity.

2. Preliminary field experiment

Before doing a full-scale field experiment, a small-scale field experiment was conducted.

Experimental site: The site was located at a place of an elevated groundwater table in a depressed part of a salt-affected area. The site was divided into 2 sections for 2 Series of experiment at slightly different groundwater table: the groundwater table at Series 1 was about a few 10 cm lower than that for Series 2.

Plot: Eight beds (50 cm high, 1 × 6 m) were prepared for each Series with 4 different plots (T1, T2, T3 and T4) for avoiding flooding. T1 was the control plot. T2 was the plot applied with cow dung (1%) and PVA (0.2%). T3 was the plot applied with cow dung (1%). T4 was the plot applied with cow dung (1%) and PVA (0.05%). All the plots were mulched with rice straw and halved into 2 subplots. One subplot was cultivated with Panicum repens, a perennial grass, and the other with Sesbania rostrata, an annual legume.

Monitoring: The experiment started at the end of a rainy season and terminated at the middle of the next dry season. During this period, the surface soil (5 cm), the groundwater tables and the plants were monitored.

3. Field experiment

To confirm and to supplement the results of the preliminary field experiment, a full-scale field experiment with the 4 plots, which were the same as those in the preliminary field experiment, was conducted near the site of the preliminary field experiment. The field experiment differed from the preliminary field experiment in: (1) Every 4 plot (1 × 1.5 m) was replicated 4 times. (2) Three kinds of vegetables (baby corn, Ipomoea aquatica var. reptans, and tomato) were successively cultivated at every plot in every rainy season for 3 years. (3) Low beds (15 cm high) instead of the tall beds were prepared so that the beds were temporarily flooded every year. (4) The experiment started at the beginning of a rainy season.

Figure 2. Slaking decrease with increasing amount of PVA  

Results and discussion

1.1 Ability of PVA to form stable aggregate

The stability tests demonstrated the following facts: (1) All the PVA formed stable aggregates from the 2 sandy soil samples. (2) In stabilizing the aggregates, the chemical reagent was similar to GH-20 and GH-20 was more effective than GL-30. (3) The air-dried aggregate was more stable than the moist aggregate. (4) GH-20 at an application rate as low as 0.02% (w/w) was enough to generate fully stable aggregate (Figure 2 and Tables 1, 2).

Table 1. Dispersion of clay particles (Turbidity, % Absorbance)

 Table 2. Size distribution of aggregates (Yt soil)

These results implies, (1) almost all the sandy soils in Northeast Thailand can be fully aggregated with PVA, (2) the stability was unaffected by purity of PVA, (3) high molecular weight of PVA was desirable for stabilizing the aggregate and (4) air-drying the aggregates enhances the effect of PVA. Accordingly, GH-20 was selected for further study.

1.2. Properties of PVA-amended aggregate formed from the salt-affected soil

These experiments revealed the following facts (1) P VA remarkably increased stability of the aggregates of both soil samples. (2) Capillary rise of the saline water through the aggregate-columns of 2 soil samples was evidently suppressed with PVA. (3) Hydraulic conductivity through the aggregate-columns of 2 soil samples for both the saline and the fresh waters became high by amending with PVA (4) Without PVA, hydraulic conductivity of the dark colored somewhat clayey soil was higher than that of the sandy soil in the saline solution and the reverse was true in the fresh water, most probably due to dispersion of Na-clay in the fresh water (Tables 3, 4).

Table 3. Rate of capillary rise of NaCl solution

* period of wetting of surface soil column divided by height of soil column in NaCl solution.

Table 4. Percolation rate of NaCl solution and of distilled water

* too quick to be measured.

These results confirm P VA is promising for desalinizing the salt-affected sandy soil in Northeast Thailand by suppressing capillary rise of saline water and by promoting leaching of the salt contained in the soil without disruption of the aggregates against the osmotic pressure during the desalinization.

2. Preliminary field experiment

The following facts were recognized:

1. Moisture content in the soil decreased in the dry season, slightly increased in the initial period of rainy season, decreased in the dry spell, then increased rapidly in the late rainy season and decreased again in the next dry season. The groundwater tables of both Series fluctuated in a similar way.
2. In the first dry season, EC increased with decreasing moisture content till at a certain point then decreased in spite of that moisture content still tended to decrease. The increase in EC during the dry season was enhanced with cow dung. The undesirable effect of cow dung was somewhat suppressed with PVA. In the next dry season, the undesirable effect of cow dung was weakened and the effect of PVA was strengthened. These results advance the following suggestions: (1) Small and sporadic rain in the late dry season is responsible for desalinization without increasing moisture content. (2) Plant debris included in the cow dung provides long straight capillaries for capillary rise of saline water from below. (3) P VA generates aggregates containing the plant debris, resulting in weakening of the promoting effect of cow dung on salt-accumulation in the dry season. (4) The effect of cow dung diminishes when the plant debris is comminuted by microbial decomposition in the rainy season.
3. EC of any plot of Series 1 was lower than that of the corresponding plot of Series 2. This should be caused by the difference in the groundwater tables between 2 Series.
4. Low bulk density was accompanied by low soil hardness as expected. They were decreased in the following order: T1 > T3 ≈ T4 > T2. Both low density of cow dung and the ability of PVA to aggregate soil particles may contribute to make the soil more porous and friable. In addition, soil hardness as well as bulk density changed with time. Their changes were associated with changes in moisture content and soil management.
5. Among 4 plots of Series 1 with low EC, growth of 2 plants deceased in the following order: T4 > T3 > T2 > T1. This result suggests: (1) The supply of nutrients from the cow dung, which is decomposed by microorganisms, is a more limiting factor of growth of both plants than salinity and (2) the microbial decomposition of the cow dung is somewhat suppressed with PVA probably because particles of the cow dung is coated with an uneatable film of PVA.
6. Growth of S. rostrata in any plot of Series 2 was markedly worse than that corresponding plot of Series 1. This should be caused by higher EC in Series 2 than in Series 1 and low tolerance of this plant against salinity. In addition, the order of growth of the plant among 4 plots of Series 2 was different from that of Series 1. This suggests both EC and the supply of mineral nutrients from the cow dung control the plant growth in Series 2.
7. Growth of P. repens in Series 2 was not much different from that in Series 1. This may be a reflection of fairly high tolerance of the plant against salinity. However, order of growth of the plant among 4 plots was somewhat different from that in Series 1. Probably, some unknown factors other than EC and the nutrient supply contribute to control growth of the plant under salt stress (Figures 3-6 and Table 5).

3. Field experiment

The field experiment confirmed most of the results of the preliminary field experiment. In addition, the following further information was obtained:

1. The undesirable effect of cow dung to increase EC in the first dry season was less remarkable than that in the preliminary field experiment. This must be resulted from that the capillary of the plant debris included in the cow dung was somewhat destroyed by microbial decomposition of the cow dung during the first rainy season.
2. Effect of P VA to decrease EC became evident with time. PVA may remain rather intact even after 3 years.
3. Order of 4 plots in terms of the plant growth changed year by year: T4 > T3 ≈ T2 > T1 (1^st year), T2 ≥ T4 ≥ T3 > T1 (2^nd year), T2 >> T4 ≈ T3 ≈ T1 (3^rd year). This may be caused by gradual loss of the suppressing effect of PVA on the microbial decomposition of cow dung. In other words, this effect appears to be almost completely lost before the 3^rd year for T4. On the contrary, this effect still remains even in the 3^rd year for T2 (Tables 6-8).

Figure 7 illustrates assumed roles of PVA in the salt-affected infertile sandy soil in Northeast Thailand: to promote the desalinization by generating stable aggregates and to timely supply nutrients from cow dung to the plants by suppressing too rapid microbial decomposition of the cow dung.

Conclusion

On the basis of the present study, application of PVA mixed with cow dung in the beginning of the rainy season is recommended for ameliorating the salt-affected infertile sandy soil at the place with high groundwater table. However, further studies are necessary for extending this recommendation to the farmers in Northeast Thailand. This is because the rate and time of the application of PVA and cow dung should vary according to nature of both the soil and the plant.

In addition, the present study revealed the following new findings: (1) Decrease in EC in the middle to late dry season. (2) The enhancing effect of cow dung on accumulation of salt in the dry season. (3) The suppressing effect of PVA on microbial decomposition of cow dung. These findings should be thoroughly examined to understand their underlying principles. This effort may contribute to the integration of 3 research fields of soil science (soil physics, soil chemistry and soil microbiology).

Table 5. Fresh and dry weights (g) of S. rostrata and P. repens

References

Arunin, S. 1984. Characteristics and management of salt-affected soils in Northeast Thailand. In: “Ecology and Management of Problem Soils in Asia” FFTC Book Series No. 27. pp. 336-351.

Carr, C.E. and Greenland, D.G. 1975. Potential application of polyvinyl acetate and plyvinyl alcohol in the structure improvement of sodic soils. In: Moldenhauer W.C. (ed.), “Soil Conditioners”. SSSA Special Publications Series No. 7. Soil Science of America Inc. Publisher, Madison, USA. pp. 47-63.

Dejbhimon, K. 2004. Utilization of Polyvinyl Alcohol (PVA) for Ameliorating the Salt-affected soil in

Northeast Thailand. PhD thesis, Tokyo University of Agriculture, Tokyo, Japan. 127 p.

McGowan International Pty. Ltd. 1983. Thai-Australia Tung Kula Ronghai Project. Tung Kula Ronghai Salinity Study. 1981-1982. 145 p.

Puengpan, N. 1992. Salt-affected soils in Northeast Thailand and Strategies of their Amelioration. PhD thesis, Tokyo University of Agriculture, Tokyo, Japan.

Subhasaram, T. 1994. Simple and Effective Techniques to Ameliorate Salt-affected Area in Northeast Thailand. PhD thesis, The University of Tokyo, Tokyo, Japan. 476 p.

Takai, Y., Nagano, T., Kimura, M., Sugi, J. and Vacharotayan, S. (eds.) 1987. Coastal and Inland Salt-affected Soils in Thailand. Their Characteristics and Improvement. Nodai Research Institute, Tokyo University of Agriculture, Tokyo, Japan. pp. 37-43.

¹ Research Annex, Faculty of Agriculture, Khon Kaen University, Khon Kaen, Thailand.
² 1-53-12, Umegaoka, Setagaya-ku, Tokyo, Japan.

Impact of agricultural practices on the biogeochemical functioning of sandy salt-affected paddy soils in Northeastern Thailand

Quantin, C.¹; O. Grunberger²; N. Suvannang³ and E. Bourdon⁴

Keywords: Soil salinity, land management, organic matter, paddy systems

Introduction

Most lowlands in Northeastern Thailand are cultivated with rice, but among them, 8.5% have been classified as severely salt-affected (Arunin, 1984) due to the rising of the water table since land clearing. Water rises to the surface by capillary action and evaporates, so salts accumulate at the soil surface and a saline crust can be observed during the dry season. Therefore, salinity drastically affects soil fertility and rice productivity, already affected by the acidity of these sandy soils. Then, farmers focus their efforts on cultivating less affected soils, so the result is that the salt-affected soils become more damaged. However, in Isaan, efficient water management and organic matter addition or green manuring are low cost solutions used by farmers to supply nutrients to these poorly fertile soils.

In paddy soils, incorporating rice straw or green manure can be an useful way of adding organic matter and thus increasing carbon storage and providing nitrogen, phosphorus, potassium and other nutrients to soils (Vityakon et al., 2000). However, a poorly controlled incorporation of OM can be responsible of the appearance of strong reducing conditions that may have adverse effects on rice cropping, as for instance the production of sulphide and the subsequent formation of black roots (Gao et al., 2004) or organic acid toxicity (Saenjan, 1999). The addition of organic matter or the incorporation of crop residues increases the organic matter availability and thus the anaerobic bacterial activity. This can lead to a high transfer of electrons from organic matter to oxides, especially amorphous or poorly crystallised ones, leading to the reduction of both manganese and iron, and also to the establishment of strongly reducing conditions. Solubilisation of these elements is a function redox conditions driven by bacterial activity, organic matter availability, soil moisture and thus agricultural management. Moreover, reduction processes control the pH and the ionic composition of the soil solution, both acting on soil fertility (Ponnamperuma, 1976). Thus, Mn and Fe are key indicators for understanding the reduction processes and thus the biogeochemical functioning of rice paddy soils.

In order to quantify the impact of realistic low cost agricultural practices on the biogeochemical functioning of paddy fields in N.E. Thailand, we have studied the interactions between agricultural management (i.e. organic matter addition, water level control), salinity and redox processes, during the 2003 rainy season. Field measurements included pH, Eh, EC and major elements in the soil solution, with a particular focus on Fe dynamics.

Materials and methods

The field investigation was carried out in 2003, from July to November. Two rainfed lowland paddy fields of a rice cropped watershed in Isaan, close to Phra Yun in the Khon Kaen district (E 102º38′-N 16º22′) were selected, as representative of two types of farming practices: an intensively managed plot (L25 plot, 599 m²) and a poorly managed one (L14, 718 m²). The first plot was characterized by organic matter addition (buffalo, poultry and pig manure mixed with sawdust), of around 170 kg plot^-1 year^-1, corresponding to 2.4 t ha^-1 year ^-1 wet weight, and efficient water control with a high bund system that allows an almost continuous flooding. The second one did not receive any particular treatment, i.e. neither organic matter addition nor water management.

Saline patches were observed during the dry season inside each plot, and monitoring points were selected to reach the maximum contrast of salinity over a short distance (i.e. 8 meters). In each plot, one monitoring point was located inside an area were the production of rice was affected by high salt contents (soil conductivityobtained by EM38 higher than 250 mS.m^-1, labelled S for saline) while the other monitoring point was located in an area where rice yield was non affected by salinity (soil conductivity around 150 mS.m^-1, labelled NS for non saline).

Soils were sandy loam from the soil series Kula Ronghai (Natraqualf), which predominates in salt affected zones of Northeast Thailand lowlands. Main characteristics are summarised in Table 1.

The composition of flooding water, groundwater and soil solution was monitored during the entire rainy season. Flooding water was sampled close to each monitoring point and groundwater in piezometers. The free soil solution was sampled every week or every two days, inside and outside the saline patches, at 10, 25 and 45 cm depth in polypropylene pierced boxes buried in soil, as described by Boivin et al. (2004). These devices allowed the soil solution to enter by free drainage and the sampling was carried out under a N2 atmosphere.

Table 1. Bulk soil analysis

a: Electrical conductivity of saturated paste, b: Organic C by Walkey and Black method, c: Kjeldahl method, d: Ammonium acetate method, e: Pipette method, f: 100 cm³ cylinder method (average of 5 replications), g: acid attack (HF + HClO₄; HCl + HNO₃), h: Tamm extraction

pH and EC were measured in the field as soon as possible after filtration of the soil solution, in order to prevent strong re-oxidation. Another aliquot was acidified for cation analysis by ICP-OES.

Eh was measured in situ during the entire cropping season, inside and outside the saline patches in the two plots, at 10 and 25 cm depth. In L25 plot, Eh was measured continuously every hour, whereas in L14, measurement was performed manually once a week.

Results

Flooding water did not vary in composition during the cropping period, except a slight increase in EC at the end of October, mainly due to increasing Na concentration (Table 2). The groundwater salinity was very high and the aquifer was not chemically homogenous (Table 2).

Table 2. Flooding water and groundwater analysis

In the L25 plot, strong reducing conditions prevailed. In L25NS and L25S, Eh rapidly decreased after transplanting and stabilised around -200 to -250 mV at 10 cm depth. At 25 cm depth, Eh also decreased quickly in L25S, reaching -180 to -200 mV before increasing and stabilising at around -100 mV. In L25NS, this decrease was slower, and Eh stabilised at around -100 m V. Oxidation peaks occurred at different times, corresponding to rainfall events. Eh rapidly increased to oxidised values when plots were drained in November.

In L14NS, the Eh pattern was close to that observed in L25NS. In L14S, Eh was significantly higher than in L14NS, and higher at 25 cm than at 10 cm depth. Eh values were practically always above +100 mV, apart from an occasional rapid decrease to -45 mV at 10 cm depth.

In all locations except in L14S, EC increased with depth. Outside the saline patches, EC remained almost constant with time with slight fluctuations, around 6.4 ± 1.5 dS.m^-1, 10.7 ± 3.2 dS.m^-1 and 11.6 ± 0.8 dS.m^-1 in L25NS at 10, 25 and 45 cm depth, respectively, and around 3.5 ± 1.9 dS.m^-1, 8.1 ± 1.5 dS.m^-1 and 8.3 ± 0.6 dS.m^-1 in L14NS at 10, 25 and 45 cm depth, respectively. Electrical conductivity values and variations were larger inside the saline plots. In L14S, EC at 10 cm depth was higher than EC at 25 and 45 cm depth.

pH changes in the soil solution differed depending on the management practices and on the depth. After transplanting, pH at 10 cm depth in L25NS increased from 5 to 6.5 in a few days A similar increase, with a time lag, occurred at 25 cm (Figure 1). At 45 cm depth, pH remained low at 4.1 ± 0.5 during the entire cropping period. In L25S, pH at 10 cm depth showed the same trend as in L25NS, stabilising around 6.5. At 25 and 45 cm depth, pH remained very low, around 4.8 ± 0.5 and 3.8 ± 0.1, respectively (Figure 1). In L14NS, pH increased from 4.8 to 6.7 at 10 cm depth, much slower than in L25 and remained quite high until harvesting. At all other depths and also at 10 cm in L14S, pH remained constant around 4.

The main cation in the soil solution was Na, ranging from 14-80 mmol.l^-1 at 10 cm, 60-10 mmol.l^-1 at 25 cm and 50-135 mmol.l^-1 at 45 cm depth in L25NS and L14NS plots. Inside the saline patches, Na concentrations were significantly higher and more variable with time with most of the values ranging from 150 to 450 mmol.l^-1.

As for pH, Fe and Mn concentrations in the soil solution differed depending on the management practices and on the depth. Fe solubilisation was significantly higher in L25 than in L14 (Figure 2), particularly outside the saline patch. Fe reduction increased with rice growth and Fe concentrations reached 2.74 and 2.5 mmol.l^-1 in L25NS and 1.72 and 0.85 mmol.l^-1 in L25S at 10 and 25 cm depth, respectively. At 45 cm depth, Fe solubilisation was also high in L25NS, with the same trend as at the other depths, while it remained low in L25S, less than 0.2 mmol.l^-1. After 55 days, Fe concentrations decreased dramatically until the rice harvesting. In L14, Fe reduction was low (Fe concentration 0 to around 1 mmol.l^-1), or even nil, inside the saline patch at all depths.

Figure 1. Changes in soil solution pH at 10, 25 and 45 cm depth, in the four monitoring points

Figure 2. Fe content of the soil solution at 10, 25 and 45 cm depth, in the four sampling points

Discussion

Flooding leads to major chemical changes in the soil that affect element mobility. Within a few hours to days after submergence, O₂ is consumed by aerobes, which is reflected by the rapid decrease in Eh after flooding. Moreover, the controlled flooding limited, even decreased, the salinity in the top-soil layer, especially at 10 cm depth, i.e. in the root zone. The salt concentrations inside the saline patches decreased with flooding, mainly due to dilution.

The almost continuous submerged conditions induced anaerobic conditions, which were maintained during the cultural cycle. At 10 cm depth, i.e. in the rooted soil in well managed plot (L25NS and S) and also in L14 outside the saline patch, Eh reached very low values, lower than in deeper horizons, and topsoil remained highly reduced during almost the entire growth period.

Anaerobiosis leads to the establishment of reducing conditions, particularly strongly when organic matter is incorporated into the soils. Thus, other electron acceptors than O₂ are sequentially reduced: NO₃^-/N₂, Mn(III, IV)/Mn(II), Fe(III)/Fe(II), SO₄^2-/HS^-, and CO₂/CH₄ (Sposito, 1989), by facultative anaerobes followed by strictly anaerobes in order to oxidise organic matter (Ehrlich, 1996, Madigan et al., 2000). As a result, the concentration of reduced compounds like Fe²⁺ increases in the soil solution, and the changes in the soil solution composition are closely related to the metabolic activity of the microbial communities involved. Ferric iron is commonly the dominant electron acceptor in anoxic systems and may contribute considerably to organic matter biodegradation (Thamdrup, 2000, van Bodegom and Stams, 1999).

The reduction of oxidised compounds depends both on their availability and the presence of microbial communities in soils. Iron release increased with rice growth and was stimulated by organic matter addition, i.e. with high organic carbon availability, as observed by other authors (Tanji et al., 2003, van Asten et al., 2004). Both organic matter addition and high reducible Fe contents may increase the intensity of reduction. High concentrations of dissolved organic carbon, originating from the organic matter added, from root exudation and from the anaerobic biodegradation of root materials, were measured in L25, higher than in L14 at 10 and 25 cm depth (data not shown). These high dissolved organic carbon concentrations suggest there is no limitation in electron donors to sustain Fe reduction, particularly in the well-managed plot. Nevertheless, Fe release was highly limited under saline conditions. Salinity drastically affects microbial biomass and specific metabolic activities (Rietz and Haynes, 2003) and it can be argued that Fe reduction capabilities are also reduced. The number of active anaerobic iron-reducing bacteria is probably lower inside the saline patches and, even if organic substrates are added, their activity is probably strongly inhibited. Thus, by affecting the reducing activity inside the saline patches, salinity drastically affects the pH of the soil solution, even if organic matter is incorporated (see below).

In flooded anaerobic soils, Fe is highly available for plants and Fe toxicity may occur (Marschner, 1995; Becker and Asch, 2005). Even if lowland rice is adapted to such anaerobic conditions, Fe²⁺ can become toxic at high concentrations or when other nutrient deficiencies occur. Fe toxicity seems to appear when Fe concentration is high in the rhizosphere (Montas Ramirez et al., 2002) and when multiple deficiencies occur. In the four studied profiles, Fe concentrations were well below the concentration mentioned by Vizier (1978), so Fe toxicity seems not to be a major risk in these fields. Organic matter addition may also enhance sulfide toxicity resulting in black roots symptom (Tanji et al., 2003), but this can be counteracted by the precipitation of sulfide minerals like FeS when Fe²⁺ concentration is high (Gao et al., 2004). Even though we did not observe retarded plants having such symptoms, the conditions favouring both Fe solubilisation and sulfide accumulation need further investigations in amended plots especially if we want to propose alternative farming practices to farmers.

It is commonly observed that pH of submerged acid soils increases with time to fairly stable values close to neutrality (Ponnamperuma, 1976, Genon et al., 1994, Scott et al., 2003, Kirk et al., 2003, Kirk, 2004). pH greatly influences soil fertility as it controls several parameters like chemical equilibria and surface charges, microbial processes, plant absorption of nutrients, organic matter biodegradation, and also the availability of toxic substances (Sposito, 1989). The pH increase was quicker when organic carbon was highly available as in the L25 plot. Where organic matter was not added to soil, i.e. in low management plot, high salinity prevented high reduction in the top layer. Therefore, no pH increase occurred and pH remained unfavourable for rice growth. The two-stage pH vs time profile observed in paddy soils is mainly explained by the Fe systems. The initial pH rise is explained by the reduction of Fe oxides and the Fe(OH)₃/Fe²⁺ couple. The further pH stabilisation around 6.5, when observed, may be explained by the formation of a mixed Fe(II)-Fe(III) hydroxide. Indeed, ion speciation computations revealed that the soil solutions were oversaturated with respect to an hydroxy-green rust in the intensely managed plot particularly (data not shown).

The alternation of oxic and anoxic conditions produce redoximorphic features, which were observed in the soil profile. Fe- and Mn oxide nodules were observed at 30-40 cm depth, particularly in non saline profiles. During the dry season, red coatings and iron sheaths could be observed around rice roots. They provide evidence of the biogeochemical redox processes occurring in the soil profile. Iron sheaths usually limit phosphorus availability for rice (Berthelin and de Giudici, 1993). The reduction process also may affect the CEC of the soil, especially because of the reduction of coatings that may neutralise negatively charged sites on clay minerals. Such poorly crystallised or amorphous Fe oxides, which represented a large proportion of iron in the studied soils, are preferentially reduced in hydromorphic soils (Munch and Ottow, 1980, Francis and Dodge, 1988). Moreover, structural Fe can also be reduced, thus increasing CEC (Stucki et al., 1987, Favre et al., 2002).

Conclusion

Intensive management of rainfed lowland rice soils in Northeastern Thailand has great impact on the biogeochemical functioning of these soils. It leads to the establishment of strong reducing conditions in the topsoil and to pH increase to near neutrality. Organic matter input seems to counteract salinity effects on soil chemistry and soil microbial activity. Moreover, the proportion of Fe released from the reductive dissolution of amorphous or poorly crystallized Fe oxides depends on such organic management. Thus, water management and availability of organic carbon, which maintain saturation and control the extent of the reduction, are processes of major importance for pH regulation and rice production. While strong reducing conditions appear to be favourable for pH, Fe toxicity could occur for sensitive rice cultivars as the Fe concentration in the root zone reaches high values. Repeated incorporation of organic matter could enhance the risk of Fe toxicity, particularly in such degraded and low buffered soils.

References

Arunin, S., 1984. Characteristics and management of salt affected soils in the Northeast of Thailand. In: Food and Fertilizer Technology Center for the Asian and Pacific Region, ed., Ecology and management of problem soils in Asia, Taiwan, Republic of China, 336-350.

Becker, M. and Asch, F., 2005. Iron toxicity in rice -conditions and management concepts. J. Plant Nutr. Soil Sci., 168, 558-573.

Berthelin, J., and de Giudici, P., 1993. Processus microbiens et physicochimiques liés à la biodisponibilité des nutriments dans la rhizosphére du riz: cas des riziéres de bas-fond à Madagascar. In: CIRAD, ed., Bas-fonds et riziculture, Antananarivo, 273-282.

Boivin, P., Saejiew, A., Grunberger, O., and S., A., 2004. Formation of soils with contrasting textures by translocation of clays rather than ferrolysis in flooded rice fields in Northeast Thailand. European Journal of Soil Science, 55, 713-724.

Ehrlich, H.L., 1996. Geomicrobiology, 3^rd ed., New York, 719 p.

Favre, F., Tessier, D., Abdelmoula, M., Génin, J.M., Gates, W.P., and Boivin, P., 2002. Iron reduction and changes in cation exchange capacity in intermittently waterlogged soil. European J. Soil Science, 53, 175-183.

Gao, S.D., Tanji, K.K., and Scardaci, S.C., 2004. Impact of rice straw incorporation on soil redox status and sulfide toxicity. Agronomy Journal, 96, 70-76.

Genon, J.G., de Hepcée, N., Delvaux, B., Dufey, J.E., and Hennebert, P.A., 1994. Redox conditions and iron chemistry in highland swamps of Burundi. Plant Soil, 166, 165-171.

Kirk, G., 2004, The biogeochemistry of submerged soils. Wiley, 291 p.

Marschner, H., 1995. Mineral nutrition of higher plants. 2^nd edition, Academic Press, London, UK.

Montas Ramirez, L., Claassen, N., Amilcar Ubiera, A., Werner, H., and Moawad, A.M., 2002. Effect of phosphorus, potassium and zinc fertilizers on iron toxicity in wetland rice (Oryza sativa L.). Plant Soil, 239, 197-206.

Munch, J.C., and Ottow, J.C.G., 1980. Preferential reduction of amorphous to crystalline iron oxides by bacterial activity. Soil Sci., 129, 15-21.

Ponnamperuma, F.N., 1976. Physicochemical properties of submerged soils in relation to fertility. The fertility of paddy soils and fertilizer applications for rice. Taipei, Taiwan, Rep. of China, Food and fertilizer technology center for Asian and Pacific Region, 1-27.

Rietz, D.N., and Haynes, R.J., 2003. Effects of irrigation-induced salinity and sodicity on soil microbial activity. Soil Biol. Biochem., 35, 845-854.

Saenjan, P., 1999. Biochemical properties of Roi Et and Phimai soil receiving organic debris. Kasetsart J. (Nat. Sci.), 33, 317-329.

Scott, H.D., Miller, D.M., and Renaud, F.G., 2003. Rice soils: Physical and chemical characteristics and behaviour. In Smith, C.W., and Dilday, R.H., eds., Rice: Origin, history, technology and production, Wiley and Sons, Inc., 297-329.

Sposito, G., 1989. The chemistry of soils, Oxford University Press, 277 p.

Stucki, J.W., Komadel, P., and Wilkinson, H.T., 1987. Microbial reduction of structural iron(III) in smectites. Soil Sci. Soc. Am. J., 51, 1663-1665.

Tanji, K.K., Gao, S., Scardaci, S.C., and Chow, A.T., 2003. Characterizing redox status of paddy soils with incorporated rice straw. Geoderma, 114, 333-353.

Thamdrup, B., 2000. Bacterial manganese and iron reduction in aquatic sediments. Advances Microbial Ecology, 16, 41-84.

van Asten, P.J.A., van’t Zelfde, J.A., van der Zee, S.E.A.T.M., and Hammecker, C, 2004. The effects of irrigated rice cropping on the alkalinity of two alkaline rice soils in the Sahel. Geoderma, 119, 233-247.

van Bodegom, P.M., and Stams, A.J.M., 1999. Influence of alternative electron acceptors on methanogenesis in rice paddy soils. Chemosphere, 39, 167-182.

Vityakon, P., Meepech, S., Cadisch, G., and Toomsan, B., 2000. Soil organic matter and nitrogen transformation mediated by plant residues of different qualities in sandy acid upland and paddy soils. Netherlands Journal of Agricultural Science, 48, 75-90.

Vizier, J.F., 1978. Etude de la dynamique du fer dans des sols évoluant sous l’effet d’un excès d’eau. Etude expérimental sur des sols de rizières de Madagascar. Cah. ORSTOM, sér. Pédol, XVI, 23-41.

¹ UMR 8148 IDES, CNRS-Paris-Sud XI University, F-91405 Orsay Cedex, France, [email protected]
² UR SOLUTIONS, IRD – Land Departement Developement, Office of Science for Land Developement, Phaholyothin Road, Chatuchak, Bangkok 10900, Thailand.
³ Land Departement Developement, Office of Science for Land Developement, Phaholyothin Road, Chatuchak, Bangkok 10900, Thailand.
⁴ UR SOLUTIONS, IRD, 34095 Montpellier cedex, France.