To explain very low fruit production in the tropical tree Ipomoea wolcottiana (Convolvulaceae), experiments and observations are presented regarding hypotheses that fecundity is limited by foliar herbivory, low soil phosphorus, inadequate cross-pollination, low rates of pollinator visitors, and parasitism of flower buds. A 2 × 2 factorial experiment was undertaken in two consecutive years which included control of herbivores with contact and systemic insecticides and addition of phosphorus. Phosphorus application in the second year increased the number of inflorescences and flowers by 200%, and the initiation of fruits by 400%. No response was observed for the number of mature fruits. Herbivore control and the addition of phosphorus increased seed weight by 22% in the first year, but could not be measured in the second year. Hand-pollination experiments increased the fruit set by 58% in the first year and by 75% in the second. Observations on visits to the flowers showed a six-fold difference between years in rates of pollinator visits. Parasitism of flower buds was 18% in the first year and 33% in the second. Several of the differences between years probably resulted from lesser and desynchronized flowering, due to heavy unseasonal rains in the second year. The differences in floral visitation and predation of floral buds were reflected in open fruit production: 16% in the first year and 6% in the second. All the factors investigated have important effects on the fecundity of I. wolcottiana. From the evolutionary point of view, the results suggest a complex interaction among selective forces which interact with the reproductive system. The theories of sexual selection and bet-hedging are the most plausible alternatives for explaining high levels of floral abortion in this species.