Acid sulphate soils in the Mekong delta cover 1.6 million hectares, of which 400 000 ha are located in the Plain of Reeds. Due to the presence of pyrite that yields acid when oxidised, all acid sulphate soils are (potentially) strongly acidic. Reclamation of the 150 000 ha of severely acid sulphate soils still uncultivated in 1990 became a national priority and now attracts local farmers and migrants. However, these soils present important agronomic problems and farmers urgently need advice to reclaim them. However, the development of recommendations and the cultivation of acid sulphate soils on a large scale are made difficult by their very high variability.
The objectives of this thesis are to:
- characterise and explain spatio-temporal variability, at various scales, of acid sulphate soils and water in the Plain of Reeds, Vietnam;
- assess the impact of soil and water variability on rice cultivation;
- develop a simple model of rice yield build-up under conditions following land reclamation of severely acid sulphate soils in the Plain of Reeds;
- apply this model to identify and rank limiting factors, to precisely identify the optimal time window for cultivation and develop optimal agricultural practices (in particular water management and fertilisation) for the main cropping conditions found in the Plain of Reeds.
Spatio-temporal variability in soil and water is very high, at all scales, as explained by the relative soil/water level which influences oxidation, and consequently the soil chemical status in the short term, and soil development in the long term. Soil microtopography is a key factor as slight differences in altitude induce important differences in intensity and length of soil oxidation and mineralisation. This resulted in the differentiation of two very different but closely intertwined soil types, separated by soils with intermediate characteristics. In the study area, located in the central part of the Plain of Reeds, organic Hydraquentic Sulfaquepts occupy locations below 75 cm above mean sea level where soil development is slow because of waterlogged conditions. Upon drainage, these soils are expected to develop into clayey Typic Sulfaquepts with jarosite and goethite mottles as found above 85 cm above mean sea level.
Because of the high sensitivity of plants to soil chemical characteristics, water management which determines redox conditions and Fe and Al concentrations, is a key to the cultivation on acid sulphate soils. Unfortunately, the very high soil permeability makes water control very difficult, especially in the years immediately following reclamation. This makes it difficult to maintain good cropping conditions through irrigation or drainage. Consequently, the time window during which optimal cropping conditions are met is very short. Its starting date and duration are also spatially variable, in relation to microtopography. To extend this time window, farmers in the Plain of Reeds start cultivation as soon as possible, broadcasting pregerminated rice seeds in flood water, before it has completely receded. In these cropping conditions, farmer-managed trials were conducted for four years, after precise site characterisation (using geostatistical methods). Detailed studies of crop phenology and yield led to the development of a semi-quantitative model of rice yield build-up. Yield is mainly determined by panicle density and plant growth, reflected in the weight of one grain.
Within fields, plant, tiller and panicle densities are linearly correlated to microtopography. This has been explained by higher plant mortality and poorer tillering in low positions due to deeper and longer submersion which reduces light intensity, but also to action of sulphate-reducing bacteria in the deeply reduced conditions of the lowest positions. Plant growth can be affected by the deep reduction inducing iron toxicity in the lowest locations. It is, however, mainly limited by aluminium toxicity linked to acidification of the high positions in oxidised conditions at the end of the growing season. This results in plant growth and yield correlated to microtopography in a quadratic trend, with maximum growth at medium topographic level, and with a strong decrease at high topographic levels.
Between fields, similar correlations are observed in the first year after reclamation. With cultivation, improvement of water control results in yield increase every year. This increase is faster on high fields, in which better plant growth can be obtained together with high densities. Thus, after 3 years, average yields of the fields become linearly correlated with the average topographic level, with maximum values at high topographic levels.
Application of this model allows the improvement of water management strategies, based on field characteristics: it determines the proper timing for sowing pregerminated seeds and optimal water management practices, as a function of field topography and age, and flood characteristics.
High fields (higher than 85 cm above mean sea level), and to a certain extend fields at medium topographic level, mainly suffer from acidification at the end of the growth cycle. Water management on these soils should aim at maintaining wet soil to avoid oxidation. In this respect, early sowing, in deep water (30 to 35 cm) is required in the first year after reclamation. Although this leads to lower densities, it is the only way to maintain wet conditions until the end of the cycle and to allow acceptable plant growth. With improvement of water control, sowing can be progressively delayed, until sowing on wet soil becomes possible. In contrast to this situation, low fields (lower than 75 cm above mean sea level) suffer from submersion and deep reduction. Water management should aim at creating a slight oxidation of the top soil as soon as possible. The best practice consists of sowing on wet soil after pumping water out of the field, which is not always possible in the first year after reclamation because of the high permeability.
The model also allows identification of sources of variability in fertiliser experiments conducted on these soils. The advantages of thermophosphate fertiliser over Di-Ammonium phosphate are shown and explained.
Tools and methods to control variability and to use it as information are also presented, such as the use of correlations between microtopography, soil types and natural vegetation for mapping, the proper design and set up of experiments upon precise characterisation of fields and the use of co-variance analysis with microtopography as a covariate.
The results of the study are meaningful to farmers as well as policy makers, and provide a semi-quantitative picture of the dynamics, risks and opportunities for reclamation and agricultural use of acid sulphate soils in the Plain of Reeds and beyond.