No abstract

Volcanic ash soils are generally recognized as soils with excellent and stable physical properties. Here we characterized the porosity and water properties of volcanic ash Andosols and Nitisols from Guadeloupe in contrasting banana systems: (1) perennial crop without mechanization, (2) mechanized and regularly replanted crop. Desiccation from1 kPa to1550 kPa moisture tension leads to significant shrinkage in the Andosol, representing a 50% reduction of the void space. The clayey Nitisol exhibited limited shrinkage. Soil clods from the mechanized plots had a significantly smaller macroporosity than that from perennial plots. The soil hydraulic conductivity was also drastically reduced in the compacted layers of the mechanized plots. However, Nitisols appeared to be less affected than Andosols. Laboratory compression tests showed that both soils were susceptible to compaction at soil moisture close to field capacity.The shrinkage properties of the Andosol were due to microaggregation of non-crystalline components upon drying.The relative stability of the macroporosity in the Nitisol was probably related to the presence of stable microaggregates made of halloysite and iron oxide.Two major processes promote soil structure degradation in the Andosol under mechanized banana cropping, surface desiccation and soil compaction.They are both induced by repeated tillage after clearing.

The optimization of tillage practices requires the evaluation of the long-term effects of cropping on changes in soil structure. A model is presented that simulates changes in soil structure in the tilled horizons of cultivated ®elds. The indicator of soil structure that we have modelled is based on a morphological description of the structure and the assessment of the percentage of severely compacted zones and clods showing no visible porosity in the tilled layer of a soil pro®le. The model takes into account the spatial variation of the structure induced by traf®c and tillage. Severely compacted zones are created under wheel tracks, they are cut and transferred during ploughing, and they are fragmented into ®ne soil during secondary tillage. The model was evaluated over 7 years in a ®eld trial on loamy soil in three experimental plots, where the crop sequences and ®eld operations induced contrasted changes in the percentage of severely compacted zones and clods. The model satisfactorily reproduced the changes with time of the indicator, and the correspondence between simulated and observed values was fairly good (r = 0.82, model ef®ciency = 0.61). However, some variations from one year to the next were poorly predicted. The sensitivity analysis did not suggest modi®cation to the parameter values but showed that the performance of the model could be improved if the fragmentation (due to weather and tillage) were more precisely described in the model. Nevertheless the present version of the model and the corresponding simulation program (SISOL) can be used to compare the effects of different technical choices on the evolution of soil structure.Correspondence: J. Roger-Estrade.