Objectives We studied the nitrous oxide (N₂O) emission at the key growth stages of maize in red soils under different fertilizer applications, and the main factors relative to the emission, so as to provide theoretical base for reduction of N₂O emission in the red soil area.

Methods The research based on a long-term fertilization experiment in the typical red soil region, started since 1990. Three of the treatments were selected for the research: no fertilization (CK), N and K fertilizer combined application (NK), and N, P and K fertilizer combined application (NPK). At seedling, big trumpet, grain filling and maturing stage of maize, the temperature, rainfall and N₂O emission were monitored, the top soil chemical indexes (pH, SOC, NH₄⁺-N and NO₃⁻-N content), and the activities of relative enzymes were analyzed.

Results The N₂O emission was pulsed throughout maize season, and the NK and NPK treatment led to significantly higher soil cumulative N₂O emission than CK. NPK treatment significantly increased the N₂O emission at seedling stage, while CK treatment increased that at big trumpet and grain filling stage, the two treatments had similar total N₂O emission. From seedling to maturing stage, soil pH in NK treatment kept decreasing, with decrease range of 17.8%, while the pH in CK and NPK treatment did not changed significantly. Soil organic carbon kept an overall upward trend, with an increase range of 15.2%, 16.4% and 16.2% in CK, NK, and NPK treatment, respectively. The NH₄⁺-N in all treatment soils decreased while NO₃⁻-N increased gradually. The NO₃⁻-N in NPK treatment was higher than NK treatment at seedling stage but lower at big trumpet and grain filling stage, and the NH₄⁺-N in NPK treatment was significantly lower than in NK treatment at all the stages. The activity of soil denitrification enzyme (DE) decreased by 59.1%, 66.9%, and 29.1% from seedling to maturing stage in CK, NK and NPK treatment, respectively. The nitrate reductase (NR) activities in all the three treatment soils decreased to the lowest at the big trumpet stage and then increased, the activities in NK and NPK treatment soils were still higher than in CK soil at each growth stage. Nitrite reductase activities decreased to the lowest at grain filling stage, with the lower range 33.4% and 76.2% in CK and NK treatment, and 23.5% NPK treatment than the maximum. The Pearson correlation analysis showed the cumulative N₂O emission was positively correlated with mean temperature (TEM), NO₃⁻-N, SOC and NR, negatively correlated with cumulative precipitation (PCP), soil pH and DE. Random forest model analysis showed the most important factors affecting cumulative N₂O emissions were soil pH at seedling stage, PCP at big trumpet stage, NR at filling stage, and SOC at harvest stage.

Conclusions In the red soil area, the important factors related to N₂O emission are soil pH at seedling stage, cumulative rainfall at big trumpet stage, NR at filling stage, and SOC at harvest stage. Nitrogen fertilizer application decreases soil pH, increases nitrate and organic carbon content, thereby increases N₂O emission. NK and NPK combined application would not influence the total N₂O emission, but NPK application would increase N₂O emission at maize seedling stage, while NK application would increase that at big trumpet and grain filling stage.