Maintenance of soil carbon stocks is vital for the environment at large and for maintenance of soil chemical, physical and biological fertility. Tonga represents a country in agricultural transition from subsistence to commercial production and whilst this is good for the national economy the impact on soil resources is less clear. The major cropped soils, fallow vegetation types and forest systems of Tonga were identified in each island group and samples of representative soils (0.15 m depth) from each land use unit were taken. Total carbon (C T ) and d 13 C were measured and labile carbon (C L ) determined by oxidation with 333 mM KMnO 4 . These data were used to determine the carbon management index (CMI) and the proportion of carbon from C4 species in the C T pool. Relative to primary forest, the soil C T and C L generally declined with changes in vegetation and more intense mechanical tillage. The contribution of C4 plants to soil C increased with intensity of mechanical tillage and the prevalence of C4 guinea grass (Panicum maximum Jacquin) fallow. The changes in soil C were reflected in the CMI, and C L was a more sensitive indicator of change than C T . These data indicates that all land use systems have experienced a large net loss of soil C relative to the forest systems. Soil mean weight diameter (MWD) decreased significantly with increased intensity of mechanical tillage and to a lesser extent with the intensity and length of cropping. The relationship between soil MWD and soil C was similar with soil C T and C L . Grass fallow was as effective as permanent vegetation systems in improving soil MWD and lowering the micro-aggregate (<125 lm) fraction.

Intensification of crop production in Tonga has resulted in a move to mechanical soil preparation, often with a rotary hoe, and a consequent loss in both chemical physical fertility. An experiment was conducted on a clay loam soil (Typic Argiudoll) on the Forestry Vaini Research Station, Tongatapu Island, Tonga, to investigate the effect of a once-off application of mulch on yield and quality of watermelon, maize and capsicum grown in rotation over a 1-yr period. The treatments applied were a nonmulched control, transparent plastic and 200-mm-thick applications of locally available coconut sawdust, guinea grass and mature coconut fronds. The fresh fruit yield of watermelon in all the mulch treatments was 7.3-18.1% higher than in the nonmulched control. There was no significant effect of thick vegetative mulch on maize grain dry matter yield; however, the maize yield was significantly lower in the transparent plastic mulch than in the other treatments. The capsicum marketable fruit yield in the following crop was increased by 49-73% in all the vegetative mulch treatments compared to the non-mulched control. The higher crop yields with the vegetative mulch were attributed to the measured lower soil temperatures and higher soil moisture in these treatments. There was no effect of mulch on soil total N (TN), but soil total C (TC), soil labile C (LC) and the carbon management index (CMI) were increased.

As in many areas of the developing world, intensification of agriculture in Tonga, and other Pacific Islands, has put increased pressure on the soil resource. Two experiments were conducted to evaluate the effect of mulch on the growth and yield of two important food and fibre crops. The first was conducted on sloping land to evaluate the effect of guinea grass (Megathyrsus maximus) mulch and hedgerows on taro [Colocasia esculenta (L.) Schott] yield, and in controlling soil erosion. The second compared the response of paper mulberry [Broussonetia papyrifera (L) Ventenot] to different management regimes of a grass fallow. Thick vegetative mulch increased taro corm yield by 81% and reduced soil loss by 50% compared to local farmer practice, and the soil loss from taro with mulch was comparable to the perennial cash hedgerow treatment. Mulch increased paper mulberry bark yield by 30% compared to the non-mulch control. Comparative economic analysis showed that increased net profit in the mulched treatments compared to the non-mulched control was T$2660/ha for taro and T$12 108/ha for paper mulberry. Considering that mulch is readily available to many farmers throughout the Pacific Islands and elsewhere in the tropics, it is recommended as a sustainable practice for crop production.

Accession of SO42- S from rain in Tonga has been assessed as approximately 6 kg S ha-1 year-1 from analysis of rainwater collected at three sites on two islands over a period of 365 days. The accession is of similar magnitude to crop removal of soil S. This accounts for the absence of S deficiencies in crops grown on andesitic soils which, on the basis of low to very low levels of adsorbed sulfate and low plant yields in glasshouse trials, were considered S-deficient. From consideration of ion ratios, it was established that sea spray/marine aerosol is the major source of SO42- S in Tongan rainwater.