The social wasp nests were quantified in three different plant physiognomies (forested Caatinga, shrubby Caatinga, and agricultural systems) to analyze the effect of environmental seasonality and plant physiognomy on the richness, nest abundance, and species composition of social wasps in the region of tropical dry forest of Brazil. The forested Caatinga physiognomy had the greatest richness of species (S = 16), followed by shrubby Caatinga (S = 13) and by agricultural system (S = 12). The first axis of detrended correspondence analysis (DCA) explained 67.8% of the variability and shows a gradient of the fauna from agricultural system and shrubby Caatinga to forested Caatinga. In the first axis, wet season scores were much higher than those for the dry season in forested Caatinga. The second axis explained 18.7% of the variability and shows a separation of samples collected during the wet or the dry periods in shrubby Caatinga. This separation was less evident in the agricultural system. Variations in nest abundance were more intense in arbustive caatinga (45% decrease in number of active nests in the dry period), moderate in forested Caatinga (24% decrease in number of active nests in the dry period), and low in agricultural systems (8% decrease in the dry period).