A field trip was undertaken to the Monts de Cristal National Park in Gabon during the first two weeks of November 2023. We visited the rain forests near the Kinguélé hydraulic station, a highly visited spot in the region for botanists.

We sampled pollen from male flowers at full anthesis and conserved them in 70% alcohol in the field. Pollen was prepared following the Couvreur et al. (2024) protocol using a hexane washing method. Individual pollen grains were viewed, photographed, and measured using a Zeiss Axio light microscope.

We inferred the phylogenetic position of the new species by reconstructing the phylogenomic relationships within the genus Uvariopsis. We extracted the DNA of the new species following the protocol of Soulé et al. (2023). A hybrid sequence capture approach was used to sequence nuclear markers: the Annonaceae specific baiting kit (Couvreur et al. 2019), which targets 469 Annonaceae specific exonic regions, and the universal Angiosperms 353 bait kit (Johnson et al. 2018). Libraries and hybridization protocols followed Soulé et al. (2024). We downloaded already available paired fastq sequence data (Dagallier et al. 2023) of all other Uvariopsis species except one (U. dicaprio Cheek & Gosline) from GenBank SRA under Bioproject number PRJNA508895 (https://dataview.ncbi.nlm.nih.gov/object/PRJNA508895). Xylopia hypolampra Mildbr., Monodora myristica (Gaertn.) Dunal, Dennettia tripetala Baker f., Monocylanthus vignei Keay, and three species of the genus Uvariodendron (Engl. & Diels) R.E.Fr. were selected as outgroups based on prior results (Dagallier et al. 2023; Nge et al. 2024) and were also downloaded from Bioproject number PRJNA508895.

Using the newly generated data and the downloaded sequences, we used Hybpiper v.2 (Johnson et al. 2016) to process paired reads and retrieve cleaned alignments. Paralogs identified by Hybpiper were discarded from downstream analyses. We filtered remaining supercontigs to select only those that were present in 75% of specimens for over 75% of their length (Couvreur et al. 2019; Helmstetter et al. 2024). Based on the concatenated alignments, we reconstructed the phylogenetic tree of Uvariopsis under a maximum likelihood method using RAxML “HPC-PTHREADS” (Stamatakis 2014) under a GTRGAMMA model and 100 bootstrap replicates to estimate branch support.

Preliminary information about the pollination biology of the new species was summarised from observations recorded using one PIC camera (Droissart et al. 2021). The camera was placed at ca 5 cm from one female flower at anthesis (emitting a scent) growing on the main trunk, about 50 cm from the ground. We filmed continuously for 30 hours: from 10 am (day 1) to 4 pm the next day (day 2). Recordings were reviewed and all animals that touched the flower were documented. We noted two types of behaviour: 1) touching: animals touching the reproductive organs (carpels); 2) visiting: animals only moving along the non-reproductive parts of the flower (petals) but never touching the reproductive parts. We also recorded the behaviour of the large Staphylinidae (see results) when they visited, fought, and mated. We were not able to film a male flower, but surveyed and photographed one flower at anthesis.

After video recordings, flower visitors were collected, stored in 96% ethanol, and sent to the Centre de Biologie pour la Gestion des Population (CBGP, Montpellier, France) for identification. Each sample was pre-sorted to morphospecies and a leg of one specimen per morphospecies was taken for molecular identification (DNA barcoding). For details on DNA extraction, amplification, sequencing, and post-sequencing treatment, see protocol used in Nève de Mévergnies et al. (2024).

Uvariopsis niangadoumae has tetragonal tetrad pollen grains (Fig. 4), 66–73 µm (± 6 µm) in diameter, with constitutive monads measuring ca 46 µm. The exine had a rugulate ornamentation.

Figure 4. 

Uvariopsis niangadoumae, tetrad pollen morphology. A, B. Tetrad pollen under a light microscope (scale bar A = 20 μm, B = 40 μm). C–F. Pollen grains under a scanning electron microscope. C. Apical view of two tetrads. D. Side view of two tetrads. E. Close up of a single tetrad. F. Close up of side view of tetrad showing exine structure. From Couvreur 1803.

Sequence data of the new species is available on GenBank SRA under Bioproject number PRJNA508895 (https://dataview.ncbi.nlm.nih.gov/object/PRJNA508895). A total of 319 supercontigs were selected after sequencing (75/75, no paralogs) leading to a concatenated alignment of 802,870 bp. Overall, the phylogenetic relationships were supported by 100% bootstrap values (Fig. 5). A few nodes within Uvariopsis were weakly supported, in particular the relationship of U. korupensis (TC-S0735) as sister to the U. bakeriana and U. citrata clade (BS 38). The new species U. niangadoumae clusters within the “large-leaved clade” of Uvariopsis. It is inferred as sister with maximal support to one specimen of U. korupensis (TC-S1630) sampled from Gabon (Louis 1863).

Figure 5. 

Uvariopsis niangadoumae, phylogenetic relationships within Uvariopsis. Maximum likelihood tree inferred by RAxML from 319 nuclear markers. Red circles indicate nodes with bootstrap values below 100. Outgroups Xylopia hypolampra and Monodora myristica removed for better visualization. The large-leaved clade is highlighted in red. Uvariopsis korupensis is not illustrated. TC-S numbers are the DNA accession numbers available in the SRA database.

We detected a strong unpleasant/putrid scent emitted by the flowers during the day. The scent resembled “old mushrooms”. We were able to sample four visitors collected on the female flower and identified them using DNA barcoding and morphology as: one fly (Diptera, family Phoridae), one ant (Formicidae, genus Crematogaster Lund, 1831), one leaf beetle (Chrysomelidae, subfamily Galerucinae), and one rove beetle (Staphylinidae, subfamily Staphylininae). We were not able to identify any of the samples to species level with available literature and barcode reference databases. Sequences are provided in Supplementary material 1 as a baseline for future improvement of knowledge of the identity of species visiting flowers of Uvariopsis.

The video recording (Couvreur 2025: https://youtu.be/wANN0FYkJzw) revealed a very active life around the female flower during the day (Fig. 6, Suppl. materials 2, 3). During the night (6 pm to 6 am) no activity was recorded (Fig. 6), except for one grasshopper that seemed to be licking the petals around 4 am (Suppl. material 3F). The main visitor was the large rove beetle identified as belonging to subfamily Staphylininae (Fig. 3, Suppl. materials 2, 3). An individual entered the female flower and waited (for several minutes) until another individual arrived. If it was a male, the two rove beetles fought until one of the individuals was ejected from the flower. In one case, two individuals entered the flower and waited, but after a few minutes they fought after which only one remained. If a female beetle arrived, the male resident would initiate mating (Suppl. material 3D). They would first meet face to face. The female would then turn and the male would grab the female’s abdomen with his hind legs. The male would then proceed to caress the abdomen of the female. This would last anywhere between 7 and 18 minutes, after which both would detach and pause for a while. The female then normally moved away from the inner part of the flower, eventually flying away. For the two days of observation, the mating phase was observed at midday and was generally preceded by a peak in fighting between males (Fig. 6D). Observation of the male flowers showed that the large rove beetles also visited the male flowers and emerged covered in pollen (Fig. 3F, Suppl. material 2).

Figure 6. 

Uvariopsis niangadoumae, pollination activity. Insect activity recorded around one female flower from 10 am to 4 pm the next day. Scripts for the graphs can be found here: https://forge.ird.fr/diade/global_erc/pollination_annonaceae.

Besides the behaviour of the large rove beetle, a number of other insects were observed during the day and less so during the night (Fig. 6). Small Diptera (probably several species, but just one sampled) were seen mainly just after sunrise resting on the petals (Suppl. material 3C), but sometimes moving towards the inner side of the petals and even touching the carpels (Fig. 6A). The presence of flies attracted a spider (not identified) that preyed on them (Suppl. material 3C, D). In addition, we observed another smaller Staphylininae rove beetle species (we did not sample this species, but see Suppl. material 3A, C) that visited the flower without showing the behaviour described above, although it did spend time touching the carpels. Finally, at night the flower is deserted, and only a single curious grasshopper was seen (Suppl. material 3F).

The male flower also emits a strong mushroom-like scent. Male flowers are slightly smaller than female ones (Fig. 2A), with a highly convex receptacle (torus). At anthesis, the male flower accommodates only one of these larger Staphylinidae at a time (Suppl. material 2). We observed them moving within the flower and emerging covered in pollen (Fig. 3F). We did not investigate whether flower scents vary throughout the day.

Pollination ecology has been well studied across Annonaceae, and Staphylinid beetles are generally cited as frequent flower visitors (Gottsberger 2012; Saunders 2012). However, none of the studies report the type of interactions and very specific diurnal pollination mode observed here. Indeed, thanks to the use of a PICT camera (Droissart et al. 2021), we were not only able to continuously observe what insects visited the flowers but also to document the behaviour of these visitors. Given recent technical developments and the increasing availability of new cameras capable of filming tiny and/or furtive plant-insect interactions (i.e. Darras et al. 2024; Sittinger et al. 2024), it is likely that other pollination mechanisms and/or animal behaviour will be documented in the near future.

Uvariopsis species generally lack a pollination chamber (Saunders 2010; Mertens et al. 2018), leaving the carpels fully exposed at all times, which makes the videos exceptionally clear and revealing. We show that the female flowers of Uvariopsis niangadoumae act as a “nuptial bed” for a species of rove beetle (Staphylinidae, subfamily Staphylininae), but also as a “battle arena”, where males fight for control over the flower and thus mating privileges. Flowers attract the rove beetles via a strong unpleasant fungi-like scent during the day (7 am–4 pm, Fig. 6B), and visual cues on the inner side of the four petals mimicking what resembles fungal fruiting bodies (Fig. 2C). Overall, this pollination mode is linked to saprocantharophily, a specialized form of pollination involving beetles that are attracted to decaying organic matter and mushroom-like visual cues for feeding and/or reproduction (Teichert et al. 2012). Such a pollination mode was described in Annonaceae for the first time in the Neotropical species Duguetia cadaverica Huber (Teichert et al. 2012). Annonaceae have a wide variety of flower scents ranging from sweet to rotten, and play a major role in pollination especially deceptive strategies (Goodrich 2012). More detailed studies of the floral scent should be undertaken in order to characterize the chemical composition and better understand how these rove beetles are attracted.

Our results add to the diversity of pollination strategies described for Uvariopsis. To date, sapromyiophily (pollination by flies) has been reported in U. congolana (De Wild.) R.E.Fr. and U. bakeriana (Gottsberger et al. 2011), while species of Orthoptera and Blattodea Wattenwyl, 1882 were suggested as the main pollinators in U. dioica (Diels) Robyns & Ghesq. (Mertens et al. 2018). In this latter species, visitors were frequent during the night and the day, while in the former, visitors were only reported during the day, similar to U. niangadoumae.

Besides the main rove beetle pollinator, the female flowers of U. niangadoumae were also visited by several other species (a second smaller species of rove beetle, flies, Chrysomelidae, one spider, ants) that appear to play little or no role in pollination (Suppl. material 3A). This is common in Annonaceae where scented flowers can attract a large non-specialized cohort of insects (Gottsberger 1999). At night, the flower is deserted, and we only observed a visiting grasshopper. The flower also acted as a “food court” for an opportunistic spider (indet.) that spent significant time catching the flies that were attracted by the putrid scent. The flies (Diptera, family Phoridae; plus, potentially several other species) regularly touched the carpels, especially before the arrival of the rove beetles (Fig. 6C). They were also observed on the male flowers (Fig. 3C). However, Uvariopsis pollen forms large tetrads (Couvreur et al. 2008: fig. 3), which are generally adapted for beetle pollination (Fig. 4). These flies are unlikely pollinators, as we did not observe any with pollen on their bodies. Ants, identified as a species of the genus Crematogaster, were seen forming small colonies around the sepals of male flowers (Fig. 3E), but not always, and were never seen within the flower or touching the reproductive structures. Ants have been reported to be frequent visitors of Uvariopsis dioica, but never showed much interest (Mertens et al. 2018), similar to our observations here.

The mating system of the rove beetles reported here appears to follow one already known from other Staphylinidae (Jepson née Walker 1984), who suggested eight stages of copulation: (1) examination of the female by the male, (2) examination of the male by the female, (3) prolonged examination between male and female (Suppl. material 3B), (4) male mounting, (5) extrusion of male genitalia, (6) intromission (that is the insertion of the male copulatory organ into the female during mating), (7–8) aggression of female and/or male (separation). Overall, we did observe these different stages, but based on our video it is hard to precisely see the transition between phases 5 and 6. The copulation time, interpreted from the start of abdomen caresses till separation, is quite variable in length, ranging between 7 to 18 min. This phase has been reported to last different lengths and might be species specific (Jepson née Walker 1984). In parallel, the female flower is also an area where two males will fight to retain control of the nuptial bed. Fights happen often, in many cases with the newly arrived Staphylinid winning and becoming the new resident (Fig. 6D).

In addition to re-emphasizing the importance of Monts de Cristal National Park as a centre of African biodiversity, this study also underlines a potential hotspot for pollination mode diversity with Annonaceae. Indeed, another recently discovered species is Sirdavidia solannona, the floral morphology of which suggests buzz pollination by bees (Couvreur et al. 2015). Other species known to the area such as Uvariodendron molundense (Diels) R.E.Fr. or Xylopia aethiopica (Dunal) A.Rich. (Haran et al. submitted) are likely pollinated by specialized weevils of the genus Endaeus Schoenherr, 1826 acting as brood-site pollinators, the larvae developing in the large petals (Dao et al. 2023). More research could be undertaken to document other species pollination syndromes such as Uvariopsis citrata or Piptostigma macranthum Mildbr. & Diels which probably also differ. Finally, with five species described to date, the large-leaved clade could serve as a model to understand diversification of plants along the Central African Atlantic Swell mountain range. Indeed, the combination of unique traits such as strong scented leaves, diverse petal morphology, and specialized pollination systems could be suggested as potential drivers of diversity in this heterogeneous rain forest region.