Long-lived and large flowers signify high floral maintenance costs. Species of arid/semiarid climates with large flowers are expected to have short flower lifespans and pollination-induced flower longevity in order to curb high floral water and other maintenance costs. We explored the context-dependent large flower/short flower longevity hypothesis in Eriosyce curvispina (Cactaceae), a large-flowered species of the semi-arid central Chilean Andes. We determined breeding system, flower visitation rates, and open-pollination fruit set and quantified floral water content. In a temperature-controlled field manipulative experiment, we measured potential flower longevity and tested for the presence of pollination-induced floral senescence. We measured the timespan of the complete flower cycle defined as from when the flowers began to open to when they were totally closed, and the timespan of fully open flowers defined as from when they were totally open until they began to close. The potential flower lifespan averaged 2.8 days (complete flower cycle) to 2.3 days (fully open flower). The complete flower cycle lasted 21.5 hrs, but flowers were fully open for only 10.1 hrs across days. Flower longevity in days was far shorter than reported for a large sample of species in the area. No evidence was found for pollination-induced flower senescence as a complementary means for reducing floral maintenance costs. Eriosyce curvispina is self-incompatible and abundantly pollinated by two megachilid bees. The level of pollen limitation (L=0.36) was lower than the average reported for self-incompatible angiosperms. Thus, the short flower lifespan in E. curvispina is not an impediment to high fruit set. Flowers contain >5 g of water of which >2 g resides in >40 petaloid tepals. The amount of water is far higher than in another large-flowered, non-cactus species in the area, but only about 15% of that reported in the giant sahuaro cactus, which has larger flowers than E. curvispina.

We determined breeding system, flower visitation rates, and open-pollination fruit set and quantified floral water content. In a temperature-controlled field manipulative experiment, we measured potential flower longevity and tested for the presence of pollination-induced floral senescence. We measured the timespan of the complete flower cycle defined as from when the flowers began to open to when they were totally closed, and the timespan of fully open flowers defined as from when they were totally open until they began to close. For details see the Methods section in the main text.