Abstract
Agricultural soils are major sources of nitrous oxide (N₂O) and important regulators of methane (CH₄), but field interpretation remains difficult because observed fluxes depend on the coincidence of system type, hydrothermal conditions, substrate availability, and microbial context. This challenge is especially relevant in tropical production landscapes where integrated crop-livestock systems are increasingly promoted, yet side-by-side field datasets that jointly measure greenhouse-gas fluxes, mineral nitrogen, and microbial indicators remain limited. We compared an integrated crop-livestock system, a long-term monoculture pasture, and native vegetation in the Atlantic Forest biome of Brazil over two years. We quantified N₂O and CH₄ with static chambers, measured soil temperature, water-filled pore space, ammonium, nitrate, and qPCR abundance of nitrogen-cycle marker genes, and summarized patterns with phase-wise comparisons, correlation analysis, and principal component analysis. N₂O varied strongly through time, with seasonal totals ranging from 0.13 to 1.72 kg N₂O-N ha⁻¹, but cumulative values did not differ among systems. CH₄ showed clearer system separation: annual balance was negative in the monoculture pasture and positive in native vegetation and the integrated crop-livestock system. Marker-gene abundances, particularly AOA and nirK, more consistently separated native vegetation from managed systems than did mineral N alone, while ordination summarized a dominant contrast between native vegetation and managed systems and a second pattern associated with hydrothermal variation. These results show that, in the studied landscape, CH₄ balance differentiated systems more strongly than cumulative N₂O and that flux variability was best interpreted from the joint context of hydrothermal conditions, mineral nitrogen, and selected microbial indicators.