The effect of soil water content on efflux of CO2 from soils has been described by linear, logarithmic, quadratic, and parabolic functions of soil water expressed as matric potential, gravimetric and volumetric water content, water holding capacity, water-filled pore space, precipitation indices, and depth to water table. The effects of temperature and water content are often statistically confounded. The objectives of this study are: (1) to analyze seasonal variation in soil water content and soil respiration in the eastern Amazon Basin where seasonal temperature variation is minor; and (2) to examine differences in soil CO2 emissions among primary forests, secondary forests, active cattle pastures, and degraded cattle pastures. Rates of soil respiration decreased from wet to dry seasons in all land uses. Grasses in the active cattle pasture were productive in the wet season and senescent in the dry season, resulting in the largest seasonal amplitude of CO2 emissions, whereas deep-rooted forests maintained substantial soil respiration during the dry season. Annual emissions were 2.0, 1.8, 1.5, and 1.0 kg C m-2 yr-1 for primary forest, secondary forest, active pasture, and degraded pasture, respectively. Emissions of CO2 were correlated with the logarithm of matric potential and with the cube of volumetric water content, which are mechanistically appropriate functions for relating soil respiration at below-optimal water contents. The parameterization of these empirical functions was not consistent with those for a temperate forest. Relating rates of soil respiration to water and temperature measurements made at some arbitrarily chosen depth of the surface horizons is simplistic. Further progress in defining temperature and moisture functions may require measurements of temperature, water content and CO2 production for each soil horizon.
