Generalized food-deceptive pollination in four Cattleya (Orchidaceae: Laeliinae) species from Southern Brazil

- The breeding system and pollination of four Brazilian Cattleya spp. were studied.
- All studied species are self-compatible but pollinator-dependent.
- Flowers are nectarless and pollinated by native Apidae bees or hummingbirds.
- Visitation rates, fruit-set and pollination efficiency are consistently low.
- The features above suggest a generalized food-deceptive pollination strategy.

We studied the breeding system and pollination biology of Cattleya coccinea, C. intermedia, C. purpurata and C. tigrina in native populations from Rio Grande do Sul, Southern Brazil. At the State level, the three latter species are considered under risk of extinction and this work is intended to be a contribution towards their conservation and management. To date, their pollination needs and breeding system remain unknown.Breeding system was elucidated through controlled pollinations applied to isolated plants. Pollination was studied in the field, by means of photographic and filmic records. Cattleya purpurata, C. intermedia and C. tigrina are pollinated by large native Apidae bees of the genera Bombus, Xylocopa and Epicharis that bear the pollinaria on the scutellum and mesonotum. In spite of their differences in colour and perianth size, bee-pollinated species share a number of flower features which are absent in C. coccinea. The latter species is pollinated by the hummingbird Chlorostilbon lucidus (Trochilidae). In agreement, flowers of C. coccinea present a set of flower features normally associated to ornithophily. All studied species lack rewards and deceive their food-seeking pollinators. All species are self-compatible but pollinator-dependent. Fruiting success following manual self-pollination and cross-pollination did not differ, except in C. coccinea. However, in all species fruit from cross-pollinations yielded significantly higher numbers of viable seed. All studied species displayed low visitation rates, low natural fruit set (ranging from 0.5 to 14%) and Nilsson's male efficiency factor of less than one, indicating that less than one flower was pollinated per pollinarium removed.All these features are consistent with those of other orchid groups pollinated through generalized food-deception.