Abstract Orb web spiders face a dilemma: forage in open habitats and risk predation or forage in closed habitats to minimize risk but at reduced foraging profitability. We tested whether Argiope keyserlingi opts for safer habitats at the expense of foraging success by (i) determining habitat selection indices in open and closed habitats; (ii) marking and releasing individual juvenile, subadult and adults over two 4-week periods to determine if life-history stage influences habitat selection; and (iii) determining the biotic and abiotic environmental parameters that relate to A. keyserlingi abundance. We found that A. keyserlingi selected closed habitats. Sedge and anthropogenic structures were selected and trees were avoided. Juveniles were never found in open habitats, most likely because of high postdispersal mortality. Subadults and adults may shift from closed to open habitats while juveniles never shifted habitat. Foliage density, plant height, potential prey abundance, and mantid and bird abundance were correlated with A. keyserlingi abundance, with only bird abundance explaining habitat selection. We measured web capture area, spiral distance (distance between spiral threads) and the number of decoration arms (0, 1, 2, 3 or 4) in the field and did laboratory experiments to test the influence of (i) space and vegetation; (ii) prey abundance; and (iii) web damage, on web architecture. Argiope keyserlingi webs exhibited geometric plasticity by having larger prey capture areas and spiral distances in open habitats. Decoration design did not differ between habitats however. Variation in space availability, air temperature, prey abundance and web damage explained the variations in web architecture. Potential prey size and diversity differed between habitats but prey abundance did not. As large prey may be important for spider survivorship, foraging success appears to be compromised by occupying closed habitats.