Lantana camara L. (Verbenaceae) is a shrub of Neotropical origin that has spread to at least 60 countries (Day et al. Reference DAY, WILEY, PLAYFORD and ZALUCKI2003), and its ability to rapidly occupy disturbed habitat has led to it being named one of the ten worst weeds on the planet (Cronk & Fuller Reference CRONK and FULLER1995, Sharma et al. Reference SHARMA, RAGHUBANSHI and SINGH2005). A number of traits may have contributed to its success as an invasive species. For example, L. camara increases the available soil nitrogen in otherwise nitrogen-poor ecosystems (Sharma & Raghubanshi Reference SHARMA and RAGHUBANSHI2009), it is extremely resilient to disturbance (Gentle & Guggin Reference GENTLE and DUGGIN1997a, Hiremath & Sundram Reference HIREMATH and SUNDRAM2005), it is allelopathic to native plants (Achhireddy & Singh Reference ACHHIREDDY and SINGH1984, Gentle & Duggin Reference GENTLE and DUGGIN1997b) and its fruits are dispersed by a variety of birds (Mandon-Dalger et al. Reference MANDON-DALGER, CLERGEAU, TASSIN, RIVIERA and GATTI2004, Swarbrick et al. Reference SWARBRICK, WILSON, HANNAN-JONES, Panetta, Groves and Shepherd1998). A potentially unappreciated aspect of L. camara biology that may also contribute to its invasiveness relates to pollination ecology.

There are conflicting reports as to whether L. camara can self-fertilize (either autonomously or geitonogamously) and there have been no conclusive tests that demonstrated animals were required for pollination (Hilje Reference HILJE1985, Mohan Ram & Mather Reference MOHAN RAM and MATHUR1984). Autonomous self-pollination refers to fertilization via the transfer of pollen within a flower, while geitonogamy refers to fertilization in which pollen was derived from a different flower on the same plant (Hessing Reference HESSING1988, Lloyd & Schoen Reference LLOYD and SCHOEN1992). For this study, animal pollination is defined as the transfer of pollen to stigma by a biological entity (Faegri & van der Pijl Reference FAEGRI and VAN DER PIJL1971). Autonomous self-pollination is considered advantageous to plants in newly colonized habitats (Cox Reference COX1991) and could significantly contribute to the invasiveness of L. camara.

Alternatively, if L. camara requires animal pollination, its success may be enhanced by its ability to utilize local pollinators wherever it was introduced. Indeed, a variety of pollinators have been identified. In India, butterflies (Nymphalidae, Papilionidae, Pieridae), skippers (Hesperiidae) (Dronamraju Reference DRONAMRAJU1960) and thrips (Thripidae, Phlaeothripiae) (Mather & Mohan Ram Reference MATHER and MOHAN RAM1978) were reported as pollinators, while in lowland Costa Rica only butterflies (Lycaenidae, Nymphalidae, Papilionidae, Pieridae) and skippers (Hesperiidae) were reported as pollinators (Schemske Reference SCHEMSKE1976). Reports from Australia indicated honeybees (Apidae) pollinated L. camara (Goulson & Derwent Reference GOULSON and DERWENT2004) and in Brazil hummingbirds (Trochilidae) were identified (Winder Reference WINDER1980).

This study addressed two fundamental questions regarding the pollination ecology of L. camara within its native range: (1) is L. camara capable of autonomous self-pollination and, if not, (2) which animals provide pollination services to L. camara within part of its native range? Research was conducted from 13–21 August 2010 at the Las Cruces Biological Station in Costa Rica, located at 1200 m asl in the Fila Cruces mountain range in Puntarenas province. The station contained approximately 300 ha of wet forest surrounded by a matrix of agricultural farm and ranch lands, and received an average rainfall of 4 m y⁻¹ (Janzen Reference JANZEN1983).

To evaluate autonomous self-pollination individual L. camara plants were located at the interface between forest and ranchlands. Two umbels were selected for study from each plant. Umbels were comprised entirely of unopened flowers, which reflected stage II of the Lantana floral development scheme (Mohan Ram & Mathur Reference MOHAN RAM and MATHUR1984). Flowers are able to receive pollen for 1 d and mature centripetally (Barrows Reference BARROWS1976). One umbel from each pair was enclosed in a fine-mesh cloth to prevent animal access to the flowers; in total 22 pairs of umbels were utilized. After 7 d the umbels were removed for examination under a stereomicroscope. Corolla tubes were removed from the calyx and the reproductive organs of each flower were examined for signs of fertilization. Fertilization was confirmed if both a withered style and swollen ovary were observed (Leduc et al. Reference LEDUC, DOUGLAS and MONNIER1992, Webster Reference WEBSTER1887). All statistical analyses were conducted in R (R version 2.12, R Core Development Team, Vienna Austria) against a significance value of α = 0.05. All data and statistical codes generated for this project are freely available on Dryad, the scientific database repository (www.datadryad.org; doi: 10.5061/dryad/p90js585).

A total of three flowers (out of 442) from covered umbels were fertilized (~0.5%), while a total of 129 (out of 425) flowers from exposed umbels were fertilized (~30%). There was a statistically significant difference in the number of fertilized ovules between treatment groups (paired t-test, t = 6.79, P < 0.001). While three covered ovaries appeared to be fertilized, these data strongly suggest that L. camara is not capable of autonomous self-pollination. Other plants that are incapable of producing viable seed through autonomous self-pollination do occasionally produce swollen ovaries at a similar rate (Sahli & Conner Reference SAHLI and CONNER2007) and these ovaries are not viable. However, further experimentation is required to conclusively demonstrate that L. camara is incapable of producing viable seed through autonomous self-pollination.

Surveys were conducted for 7 d (from 07h00 to 15h00) to identify animals associated with L. camara. Afternoon thunderstorms often forced cessation of activities, though on three occasions observations continued through the rain and into the night. Nighttime observations were aided by a Night Owl 5× Night Vision™ scope mounted on a tripod. All flower visitors were noted and voucher specimens of all insects were collected and deposited in the Instituto Nacional de la Biodiversidad (INBio).

To determine which animals were moving pollen among L. camara plants, fluorescent dye particles were placed on the anthers of yellow flowers (stage III of Mohan Ram & Mathur Reference MOHAN RAM and MATHUR1984) using a single nylon filament. Half of the yellow flowers on each umbel were dusted with powder and then observed for 30 min from a distance of 5 m with the aid of binoculars. Fluorescent dye particles have been shown to act as good analogues for pollen in similar studies (Adler & Irwin Reference ADLER and IRWIN2006, Fenster et al. Reference FENSTER, HASSLER and DUDASH1996). A subset of insects visiting L. camara flowers were netted and examined for fluorescent powder with a UV light. At the conclusion of each observation period the unpainted flowers were examined for transferred fluorescent powder. A total of 36 observation periods were conducted, each on a different plant, during which 228 butterflies visited the study plants. No other animals were observed nectaring on flowers, though numerous moleponine bees (Apidae: Meloponinae) did crawl between flowers, apparently to drill a hole in the floral tube and rob nectar, which has been previously reported (Barrows Reference BARROWS1976). Of the visiting butterflies (Anartia fatima, A. jatrophe, Cissia renata, C. polyphemus, C. metaleuca, Chloreuptychia arnaea, Eresia poecilina and Heliconius hecale), 81 were closely examined for fluorescent powder and none was observed. A total of two unpainted flowers were observed with fluorescent powder (out of 315 possible), which was likely transferred by butterflies that were not inspected.

These data did not suggest that butterflies were providing significant pollination services to L. camara, and a replicate experiment with fluorescent powder was established with the observation distance increased to 10 m. For this experiment, 360 flowers had fluorescent powder applied to them, leaving 361 unpainted. Hummingbird (Trochilidae) activity was immediately observed (to the exclusion of butterfly visitation), which suggested that the 5-m observation distance had biased the floral visitors previously observed. While permit restrictions did not allow vertebrates to be detained for inspection, fluorescent powder was observed on the bills of some hummingbirds (Amazilia decora and A. tzacatl) with the aid of binoculars. Additionally, 86 unpainted flowers from this experiment had fluorescent powder transferred onto them. A comparison of the butterfly and hummingbird treatment groups resulted in a statistically significant difference (Fisher's exact test, P < 0.001).

Taken as a whole, these data strongly suggest that butterflies did not transfer significant pollen among L. camara and likely did not provide significant pollination services during the study period. Butterflies have previously been shown to be frequent visitors of flowers without providing significant pollination services (Fishbein & Venable Reference FISHBEIN and VENABLE1996, Levin & Berube Reference LEVIN and BERUBE1972, Schemske & Horvitz Reference SCHEMSKE and HORVITZ1984). In this study, hummingbirds were implicated as pollinators for L. camara, although not conclusively. Hummingbirds have often been observed as major pollinators at high-altitude sites, apparently taking over as primary pollinators as insect diversity declines (Kay & Schemske Reference KAY and SCHEMSKE2003, Schemske & Bradshaw Reference SCHEMSKE and BRADSHAW1999). These data support the postulation that L. camara is capable of utilizing different pollinators throughout its range, which may contribute to its success as an invasive species.