Licuala ferruginea Becc., a tropical forest understorey palm, is observed to have fruits that appear red in colour when unripe, turning pink, then white, purple and finally black in colour as they ripen. We monitored 13 fruiting palms in rainforest fragments and recorded the consumption of fruits by animals via camera traps. We also documented the fruiting phenology of two palms in the nursery.

In the rainforest fragments, a Cream-vented Bulbul (Pycnonotus simplex) was observed plucking a mature purple fruit from a L. ferruginea palm, before flying away with the fruit in its beak. This was the only bird that was observed feeding on the mature fruit. A range of mammals, dominated by edge species such as the Long-tailed Macaque and Wild Boar, were observed to consume L. ferruginea fruits indiscriminately across all five colour stages, thereby limiting the dispersal of the fruits. Forest bulbul gape sizes also matched the fruit size, suggesting that forest bulbuls are the likely dispersers of the palm in the original forest where edge species are not in high densities.

We further posit that the initial phase of red fruits, with high contrasting red reflectance against a green foliage background, might be a form of early advertisement to birds. The fruit then turns pink and white, which have high green reflectance and is less contrasting, thereby reducing the conspicuity of the fruit. This allows the fruit to ripen with high fructose and glucose content, and turn purple and black, which are known visual cues for birds. This study provides indicative support for the dispersal syndrome hypothesis and highlights the potential effects of forest fragmentation on plant-frugivore interactions.

Study overview

This study was conducted in the field and nursery as follows: (a) natural lowland primary and secondary rainforest fragments of Singapore and (b) growing plant house in the nursery with understorey-lighting conditions (Figure 1). Within the lowland rainforest fragments at Bukit Timah Nature Reserve (BTNR; between 1.3463 and 1.3627 latitude and 103.7695 and 103.7876 longitude), Central Catchment Nature Reserve (CCNR; between 1.3355 and 1.4157 latitude and 103.7741 and 103.8380 longitude) and Singapore Botanic Gardens Rainforest (SBG; between 1.3111 and 1.3149 latitude and 103.8151 and 103.8170 longitude), clusters of L. ferruginea palms were identified (Figures 2 I-II). A total of 17 fruiting L. ferruginea palms were identified and monitored over two years, from 2021 to 2022, using camera traps. This sought to identify animals that were visiting and feeding on the fruits, and hence understand the frugivores that were dispersing the fruits of L. ferruginea palms.

At the same time, two individual L. ferruginea palms collected from the wild were grown at the Pasir Panjang Nursery (PPN) in Singapore under partial shade to mimic understorey lighting conditions of natural habitat (Figure 2 I and III). The two palms were monitored continuously and their fruiting phenology were documented to gain insight into fruit colour and biochemical changes (i.e. sugar content) associated with the different stages of fruit development. This was necessary as it was difficult to monitor the complete fruiting phenology of the L. ferruginea palms in the wild due to intense predation pressure on the fruits. In the nursery, 16-16-16 NPK fertilizer was applied once every two weeks to the palms, thereby providing sufficient required nutrients to enable fruiting all year round, as opposed to one or fewer fruiting cycles per year as observed in the wild.

Capturing animal visitation and feeding of fruiting Licula ferruginea palms via camera traps

A total of 17 camera traps (Bushnell Trophy Cam HD Aggressor and Reconyx Hyperfire HC500) were set up between 5 May 2021 and 20 Jun 2022, one camera trap for each of the identified fruiting L. ferruginea palms in the BTNR, CCNR and SBG. Each camera trap was mounted 30-50cm above ground level and facing the infructescences of the fruiting palm, and left in the field until there were no longer any fruits. This resulted in a total accumulated effort of 1270 camera trap nights. For each animal appearance that triggered the camera traps, a 10-to-15-second video was captured and later analysed to identify the animal. Observed animal behaviours in each captured video were categorised into two activities: (1) feeding, and (2) visiting, which refers to animals that were captured in the camera trap, but were not observed to be feeding on the fruit. Captured videos of the same animal with interval less than 15 minutes were considered as replicates and counted as one independent capture. In addition to the anmal visitation and feeding data collected from 17 L. ferruginea palms in the wild, additional opportunistic observations were also made on the two L. ferruginea palms cultivated from seed in PPN.

Documentation of fruiting phenology, spectral study and chemical testing of two Licuala ferruginea palms grown in Pasir Panjang nursery

Photo documentation

The two L. ferruginea palms in the PPN were checked closely for any indication of flowering. As soon as the flower buds emerged, the palms were monitored weekly. RGB photos were taken weekly for the same palms from the same angle to document flowering and fruit development. When fruits were set, Bugdorm-2120 insect rearing tents were used to protect developing fruit bunches from being predated by frugivores (Figure 3).

Sampling of fruits

One bunch of fruits from each of the L. ferruginea palms was randomly selected on the 4th, 5th, 8th, 9th, 13th, 14th, 15th, 16th and 17th week after fruit set, for quantitative measurement of fruit sizes, spectral reflectance and sugar content. Due to limited number of fruit bunches produced by two L. ferruginea palms, only up to a total of three fruit bunches were sampled from each palm per fruiting cycle. The data over several fruiting cycles were pooled to obtain full dataset for complete fruit development with nine time points as indicated formerly.

Fruit size measurement

Collected fruits were measured to obtain width and length of fruits, which were then plotted using ggplot and geom_smooth in RStudio 2021.09.0 build 351 to study the changes of fruit length and width over time.

Colour of fruits

Relative reflectance UV-VIS range (250-700 nm). Relative reflectance spectra in 250-700 nm range of collected fruits were measured using Ocean Insight Flame Miniature Spectrometer with attached Ocean Insight PX-2 Pulsed Xenon Light Source positioned at 45o of fruit surface. Average relative reflectance spectra for all fruits were calculated using Rstudio 2021.09.0 build 351.

Relative reflectance in VIS-NIR range (400-900 nm). Reflectance spectra in 400-900 nm range of the fruits attached on the individual palms were measured weekly using the handheld SPECIM IQ device from the 4th week until 20th week after fruit set. Pixel of fruits in collected his images were labelled using Scyven (Scyllarus Visualisation Environment) developed by National ICT Australia Limited (NICTA). Relative reflectance spectra of all labelled fruit pixels were extracted. Average relative reflectance spectra for all fruits were calculated using Rstudio 2021.09.0 build 351.

Qualitative colour change in response to pH

Anthocyanin has been reported to be the main colour pigment responsible for colour of many palm fruit species (Hazir et al., 2012, Martin et al., 2017). To further confirm the nature of colour pigments of L. ferruginea, pigments of collected ripe fruits were extracted using acidified ethanol (400 mL 92% ethanol : 75 mL acetic acid), with incubation time of 4 hours (Silva et al., 2019). Glacial acetic acid and 10% KOH solution were added separately to sample solutions of extracted pigments. RGB photos of colour of pigment solutions in different pH conditions were taken.

Fruit nutrient content

For each event of fruit sampling, flesh of collected fruits were separated from seeds and combined to make three 5-gram composite samples. The samples were sent to BV-AQ (Singapore) laboratory for testing of hexoses (fructose, galactose, glucose), disaccharides (lactose anhydrous, maltose, sucrose) and total sugar using gas chromatography.