The encroachment of velvet mesquite (Prosopis velutina) has resulted in increased N inputs to Southwestern rangeland soils. Because woody plant proliferation threatens the sustainability of livestock grazing, however, control of mesquite has been advocated as a means to increase rangeland forage production. The influences of mesquite control on the fluxes of CO₂, N₂O, and CH₄ have been poorly characterized to date. This study reports soil trace gas fluxes under established mesquite (11.4 mg C cm^-3), open grassland (6.1 mg C cm^-3), and 40-year old mesquite stumps (7.9 mg C cm^-3) in southeastern Arizona over 2 monsoon seasons. During the 2003 monsoon (170 mm total rainfall), differences in soil C content did not correlate with variations in CO₂ production, and efflux averaged 180.4 mg CO₂ m^-2 h^-1 over all sites. In 2004, lower rainfall (111 mm) corresponded with increased (30%) CO₂ emissions beneath mesquite, while efflux in the open and stump plots decreased by 40%. N₂O emissions averaged 13.6 (mesquite), 5.7 (open), and 6.8 μg N₂O m^-2 h^-1 (stump) during the 2003 monsoon, but increased by more than 40% across all sites in 2004. Mesquite soils showed a steady (-17.4 μg CH₄ m^-2 h^-1) CH₄ sink across both seasons, while the open and stump sites showed early-monsoon CH₄ production, possibly due to belowground termite activity. As the monsoon progressed, these CH₄ sources turned into small (-6.2 μg CH₄ m^-2 h^-1) sinks. Global warming potentials (GWPs) of all sites were strongly dominated by CO₂ efflux. Given adequate rainfall, mesquite control measures did not impact the GWP of semiarid soils. Under limited rainfall, however, grassland and stump sites had GWPs 40% lower than that of the mesquite, indicating that climate change shifts resulting in reduced monsoon precipitation may contribute to negative feedbacks for future global warming.