AbstractStudies were conducted to assess the numerical response of ground-dwelling arthropods to a habitat management system ('push–pull') developed to control maize stemborers using spiders (Araneae) as an indicator group. In this cropping system, maize is intercropped with a stemborer moth-repellent (push) plant while an attractant trap crop (pull) is planted around this intercrop. Two study sites in western Kenya and one site at the Grain Crops Institute of the Agricultural Research Council in Potchefstroom, South Africa, were sampled. Treatments comprised a maize monocrop and an intercrop of maize and desmodium, Desmodium uncinatum Jacq., with Napier grass, Pennisetum purpureum (Schumach), as a trap crop around the field ('push–pull') in each site. Experiments were laid out in a completely randomized design with four replications at each site. Ground-dwelling spiders were sampled using a combination of pitfall traps and soil samples. A total of 2175 spiders, 78 species in 18 families, were recovered in Kenya and 284 spiders, 34 species in nine families, were recovered in South Africa. Lycosidae was the most abundant family, accounting for >50% of all individual spiders and 27.6% by species richness. Spiders were significantly more abundant at the Kenyan sites than in South Africa while species diversity was significantly higher in South Africa than at the Kenyan sites. At all sites, spider abundance was significantly higher in the 'push–pull' than in the maize monocrop plots. However, the overall spider diversity was only significantly higher in the 'push–pull' than in the maize monocrop plots in South Africa. Moreover, species dominance did not differ between the two cropping systems at all sites. The results showed that the 'push–pull' system evidently enhances overall abundance of spiders, illustrating its potential in further pest control in the maize agroecosystems where spiders may often be one of the most important predatory groups.