Time lapse imagery was recorded in three places on St Helena, Commidendrum robustum capitula in Millennium Forest and Peak Dale and C. rugosum capitula in Blue Point (Fig. 2). All capitula recorded were 30-50 cm from the ground level, due to the height of the camera and tripod used. The Millennium Forest capitulum was situated at 430 m of elevation, approximately 20 m from the edge of a dense C. robustum planting, on a low branch of a young tree of 2 m high. Millennium Forest is the largest individual ecological restoration site on St Helena, with varied plantings of dryland endemic species. C. robustum has not generally reached over 3 m high in this seasonally very dry and exposed site. Peak Dale capitula were situated in 560 m of elevation, approximately 15 m from the edge of the C. robustum forest, on a low branch of a tree. Peak Dale is the only remaining relatively sizeable natural stand of C. robustum forest, with trees reaching to 10-15 m in height and thus most of the capitula higher up than the ones recorded with the time-lapse camera. Blue Point capitula were situated in 560 m of elevation on the windward side of the ridge and 580 m of elevation on the leeward side of the ridge, on C. rugosum shrubs less than 1 m high. The windward side of the Blue Point ridge is highly exposed, with low and sparse vegetation and a relatively large C. rugosum population. The leeward side is dominated by introduced thicket, up to 5 m high, with a small C. rugosum population situated in a few openings in the higher vegetation. Blue Point is one of the strongholds of endemic dryland vegetation on St Helena.

According to the time-lapse data, the main diurnal visitors are Lepidoptera and Diptera (Table 1) with Lepidoptera tending to occur more often on Commidendrum robustum and Diptera characteristic of C. rugosum, but also visiting C. robustum frequently. The main Diptera pollinators are the endemic Loveridge’s hoverfly Sphaerophoria beattiei, the non-endemic hoverfly Syritta stigmatica and the “small black fly” morphotaxon (possibly more than one species of very small Diptera: Muscidae/Anthomyiidae and possibly including the endemic muscid Limnophora helenae). (Unfortunately this morphotaxon is too small for critical identification features to be resolved by our camera system. Further studies will be needed to identify this/these species). The main Lepidoptera pollinators are the introduced moths Spoladea recurvalis and Herpetogramma licarsisalis and the non-endemic long-tailed blue butterfly Lampides boeticus. Florets in the capitulum open sequentially over several days (Table 2;Suppl. material 1). The peak visits occurred when the majority of florets were open during mid-flowering of the capitulum (Figs 3, 4). Our data do not include nocturnal visitors, but general field observations of gumwood indicate that nocturnal visits by several moth species occurs. Night-vision equipment was not available to record this, but full 24 hour recordings would be of interest for future work. Combined data on pollinator preferences are shown graphically in Fig. 5.

Figure 5.  

Pie charts to illustrate pollinator preference, in visits (A, C) and residence time (B, D) for Commidendrum rugosum (A, B) and C. robustum (C, D). Proportions for moths, flies, hymenoptera and butterflies (Lycaenidae).

Overall, there is considerable pollinator activity in both the species (Fig. 6), whether measured as insect visitation rates (mean/median visits per capitulum per day, V) or insect residence time (mean/median pollinator kiloseconds per capitulum per day, R). Differences in pollinator type (moth vs. fly) affect patterns of pollinator activity. For C. rugosum (mainly fly), the measures are: V = 16.4 and R = 3.101 (medians V = 7, R = 0.864) and for C. robustum (mainly moth), they are: V = 34.0 and R = 8.274 (medians V = 20, R = 6.488). It is evident that moth visitors tend to be on C. robustum and spend longer on a visit than flies (mean visit duration: moth 505 seconds; fly 189 seconds).

Figure 6.  

Daily totals of pollinator activity per capitulum, with total number of visits per day plotted against total pollinator time spent on a capitulum. Each dot represents a capitulum on a particular day. Pollinator activity varies between good days and poor days depending mainly on weather (wind and rain). Grey triangles = Commidendrum rugosum, mainly flies. Black dots = C. robustum), moths and flies.

Finally, we noted that moths and syrphid flies interact differently with capitulum morphology. Commidendrum robustum has longer projecting anther tubes, ca. 1.8 mm versus 0.9 mm in C. rugosum. Moths alight on the tops of the projecting styles and anther tubes of C. robustum and may be seen to probe the corolla tubes from above with their probosces.

Patterns of visits to individual C. robustum capitula are shown in Fig. 3. At the Millennium Forest site, the main visitors recorded were two introduced species of day-flying moths, Spoladea recurvalis and Herpetogramma licarsisalis (Fig. 3). Hoverflies were also recorded from the capitulum, Syritta stigmatica being the main species. The endemic hoverfly Sphaerophoria beattiei was not recorded from the Millennium Forest capitulum. Peak Dale capitula were visited mainly by Diptera, especially Sphaerophoria beattiei. Interestingly, day-flying moths were rare on Peak Dale capitula, the main day-time Lepidoptera visitor being the non-endemic butterfly Lampides boeticus. Although no quantitative night-time observations were made, general field observations revealed a number of moths visiting C. robustum capitula after dark at both sites. Identified nocturnal species at Millennium forest included numerous introduced moths: Herpetogramma licarsisalis, Hypocala rostrata (Fabricius, 1794), Spodoptera littoralis (Boisduval, 1833), Spoladea recurvalis and Trichoplusia ni (Hübner, 1803). At Peak Dale (21 February 2014), the endemic moth Helenoscoparia scintillulalis (Wollaston, 1879) was observed, as well as the introduced Hypocala rostrata (Suppl. material 2).

Patterns of visits to individual C. rugosum capitula are shown in Fig. 4A–F. In general, it seems that scrubwood capitula are mainly visited by Diptera, although a non-endemic butterfly, Lampides boeticus, is also a relatively common visitor. On capitulum 6, it is noticeable how on day 5 very few flies visited the flower while there is a high level of visitation on other capitula. We attribute this to the presence of a spider (marked as “other” in the figures) that occupied the capitulum for most of the daylight hours. Interference between arthropod visitors is thus an important factor in pollination. Weather conditions also contribute considerably to the abundance of flower visitors, especially on exposed sites. This can be seen on capitulum 3 on day 4 and on capitulum 4 on days 6 and 7, all of which were very windy and/or rainy with no visitors to the capitulum. Field observations at night failed to find moths or any other nocturnal pollinating insects on C. rugosum capitula.

The absence of insects on a capitulum occupied by a spider (noted above) is an example of an arthropod interaction affecting pollination. However, we were also interested in whether there were interactions between pollinator insects (particularly between moths and flies) that could affect pollination mode. To gain evidence about this, we first looked at patterns of “double occupancy”, cases in which two insects shared the same capitulum. We found this to be relatively rare, including 77 instances of moth-moth double occupancy (all on C. robustum), resulting from a total number of moth visits recorded in our study of 215 (35.8%). This compared with only 22 cases of fly-fly double occupancy (20 in C. rugosum, two in C. robustum), resulting from a total of 997 fly visits (2.2%). Most of these cases (14) involved the “small black fly” morphotaxon alone, rather than syrphids. In only two instances did we find two syrphids sharing a capitulum. There was only one case of moth-fly double occupancy (involving the “small black fly”).

To further investigate the possible sensitivity of flies to other insects, of larger or smaller size, we examined instances of “replacement”, where one insect is immediately replaced by another in a timelapse series, indicating the possibility of an interaction between the two. We found 19 replacements involving flies. There were 12 instances of fly replacing fly, six cases of which were of a larger fly replacing a smaller, two cases of a smaller fly replacing a larger and four same size replacements. There were seven instances of flies being replaced by non-fly flower visitors, in all cases the other visitors were larger in size. Conversely, we found no instances of flies replacing non-flies. Although the numbers are small, the patterns are consistent with flies (especially syrphids) not being attracted to capitula occupied by other organisms and, if on a capitulum, being readily disturbed by other organisms (particularly if larger than themselves).

Commidendrum robustum (St Helena gumwood) and C. rugosum (St Helena scrubwood) are clearly distinct species with distinct morphology and ecologies (Ashmole and Ashmole 2000). The species are known to hybridise, forming morphotypes intermediate to the parental species (Eastwood et al. 2004). Differing capitulum morphologies suggest that pollination might have a role in enforcing the species boundary. Furthermore, although there is overlap and differences across sites, there seems to be some differentiation between the two species in their main insect visitors, with moths often abundant on C. robustum capitula (at least at Millennium Forest), while C. rugosum capitula appear to attract mainly flies.

As noted in the Results, moths tend to alight on top of the long-projecting anther tubes on the C. robustum capitulum. In this position they are likely to accumulate pollen on their undersides and upper legs and thus effect pollination. By contrast, flies visiting C. robustum capitula, because of their short proboscis (ca. 1–2 mm), have to probe with their heads at the mouth of the corolla tube (i.e. below the anther tubes) in order to access nectar. In this position, they may potentially avoid the pollen and styles presented above. The C. rugosum capitula with their shorter anther tubes may be better suited to fly pollination, as the head of the fly (covered with short hairs and thus suitable for pollen transfer) remains at a similar level as the presented pollen and styles at the top of the short anther tubes. We therefore hypothesise that the difference in anther-tube length may be a key feature in differentiating the pollination niche.

It is possible that inter-insect behaviour reinforces the difference in fly vs. moth visitation rates, as hoverflies seem to be sensitive to interference and avoid visiting capitula occupied by moths and other larger insects. A study of wild roses visited by bumblebees and syrphids showed that the presence of a bumblebee on flowers deterred syrphids (Morse 1981). This was not due to any physical interaction (the bees ignored the flies completely), merely that the presence of a bumblebee rendered the flower unattractive to flies. We noted the same phenomenon between moths and flies with the presence of a moth on a capitulum rendering the capitulum unattractive to flies. The implication of this is that, given the high populations of moths, a capitulum attractive to moths could be wholly moth-pollinated, not because it is unattractive to flies, but because flies are excluded by the other visitors. Moths exhibit a tendency to spend relatively long times on the flowers, thus effectively stopping hoverflies from visiting the flowers by this behavioural exclusion mechanism. There is an important corollary to this, in that most native moths are nocturnal and flies are diurnal, allowing an avoidance of interference. The introduction of alien day-flying moths has potentially had a significant effect on pollination, reducing the amount of fly pollination in the mixed fly/moth pollination system of C. rugosum.

It should be noted that our limited study cannot rule out the possibility of moths being more common visitors to C. rugosum capitula at other times. Nevertheless, the almost complete failure of C. rugosum to attract abundant and nearly ubiquitous species of day flying moths in our study does indicate that scrubwood capitula may be less attractive to moths and, in general field observations, we failed to observe nocturnal moth activity on C. rugosum while finding it frequently on C. robustum capitula. Further studies, including nocturnal observations and observations at all seasons, would be valuable for further confirmation.

Many fly groups are anthophilous and contribute to pollination, but of these, the hoverflies or flower flies (Diptera: Syrphidae) are generally thought to be especially important (Larson et al. 2001), although their contribution may vary (Orford et al. 2015). Many St Helena endemic plants have small white flowers in clusters that are regularly visited by the syrphid flies, especially the endemic Sphaerophoria beattiei and the non-endemic, but probably native Syritta stigmatica, apparently as a convergent myophilous pollination syndrome. Fly pollination also extends to the St Helena Asteraceae, despite Hymenoptera being a dominant group in the pollination of Continental Asteraceae (Lane 1996). A previously-studied example of a St Helena endemic plant pollinated by these flies is Nesohedyotis arborea (Roxb.) Bremek. (Percy and Cronk 1997). In this, it was shown that syrphid pollination is highly effective as a pollen vector up to ca. 30 m but that this falls off rapidly to be minimal at 500 m (Suppl. material 3). While the myophilous pollination syndrome is highly effective at effecting pollination within populations and effecting gene transfer in continuous habitat, it is a relatively ineffective in fragmented anthropogenic landscapes, that would benefit from long-distance gene transfer. One exception to this is the strong-flying Eristalis tenax (see Conclusions section below).

There are currently almost 140 moth species known in St Helena, out of which some 60 are endemic to the island (Ashmole and Ashmole 2000, Roger Key pers. comm.). Most of the moth species endemic to St Helena are in the superfamily Tineoidea, family Tineidae (currently 40 species, 28 endemic) with smaller number of endemics in the superfamilies Noctuoidea (30 species, five endemic) and Pyraloidea (20 species, nine endemic). Most moth species on the island are nocturnal, with the exception of endemic Glyphipterix semilunaris and introduced species Spoladea recurvalis and Herpetogramma licarsisalis. Karisch (2018) reports that Herpetogramma licarsisalis seems to have a clear pattern in occurrence with more individuals being observed from the end of March to mid-April compared to February and early March. Variation in emergence times could therefore be one explanation for the scarcity of moths visiting the Peak Dale C. robustum capitula in January 2017 when the time-lapse footage was taken.

According to Karisch (2018) there were 51 species of Lepidoptera in C. robustum forest habitat and 19 species in C. rugosum shrubland habitat. The genus Opogona is especially widely diversified on the island and species are present in both Commidendrum habitats. Another important genus that has diversified on St Helena is Helenoscoparia, of which one species H. nigritalis (day and night flying) was described as being the most numerous moth species on the Island by Wollaston (Wollaston 1879) and is still relatively common (Roger Key pers. comm.). Night-time observations with a night-enabled time-lapse camera would be required to further advance knowledge on the role of native moths on the pollination of St Helena’s endemic species, as most of the endemic species seem to be largely nocturnal. Research into nocturnal moth pollination has also been identified as a gap and a research priority in the Galapagos Archipelago (Chamorro et al. 2012).

In the Galapagos, members of the woody Asteraceae genus Scalesia not only have capitula reminiscent of C. robustum (i.e. rather globular capitula on long peduncles), they are also visited by a mixture of flies and moths (McMullen and Naranjo 1994). Partial moth pollination in St Helena's C. robustum may not only serve to recruit a seasonally-abundant pollination service, but also might recruit insects that disperse further than flies and that extend pollination temporally, from diurnal to nocturnal. However, Lepidoptera tend to have wide seasonal variation in abundance, being usually highly voltine, so reinforcing the need for a generalist guild of pollinators in moth-pollinated plants (Pettersson 1991). This may restrict any further specialisation towards obligate moth pollination. We have at present no information as to the relative effectiveness of moth vs. fly flower visitors as pollinators in St Helena, but such information would be useful in this regard. Further studies of endemic plant pollination on St Helena are greatly needed.

In cases where habitats are altered or fragmented, pollinator-plant mutualisms can be altered or destroyed with impacts on plant fitness and survival (Kearns et al. 1998, Fabienne Harris and Johnson 2007). There are three major plant conservation concerns: (1) that pollination failure can occur within populations leading to reproductive failure (Wilcock and Neiland 2002); (2) that habitat fragmentation and/or compromised pollination services prevent gene flow between populations and promote inbreeding and (3) pollinator and/or plant distributional changes break down pollinator-mediated species barriers and lead to interspecific gene flow, which can have both positive and negative evolutionary outcomes (Hamilton and Miller 2015). Taking the first concern, in the species considered here, seed-set within populations does not seem to be a problem. Our data show that insect visits to the capitula are numerous. Suitable native insects are still abundant enough to be effective pollinators. These have been joined by introduced flower-visiting insects, such as introduced flower-visiting moths and flies, which are now abundant.

The second concern, inter-population gene flow, is more serious. The reliance of C. rugosum on small insect pollinators, such as the endemic syrphid Sphaerophoria beattiei, that are generally not considered effective pollinators at distances of over 1 km (Percy and Cronk 1997, de Jong et al. 2005), means that many small isolated populations resulting from habitat destruction are now effectively cut off from gene flow unless supplemental conservation planting is carried out. One possible way in which populations might be connected by occasional gene flow is represented by the drone fly, Eristalis tenax (and to a lesser extent by the introduced Eristalinid, Eristalinus aeneus). Although a much less frequent visitor to the flowers than Sphaerophoria beattiei, E. tenax is a strong flier and more observations on the distribution of this species throughout the island would be useful. Its capacity for flight is considerable: specimens of Eristalis tenax have been taken at sea 74 km from the nearest land (Krčmar et al. 2010).

Finally, there is the issue of interspecific gene flow. A low level of interspecific gene flow has probably always occurred between the two species considered here and some apparent hybrids may be found. The overlap of pollinators, shown in this study, indicates how this may occur and highlights the importance of spatial distance, rather than pollinator specificity, in the isolation of these species. The wholesale destruction of habitat and ecological shifts, occurring over the last 500 years may have made interspecific gene flow less, or more, likely depending on specific locality. A limited amount of interspecific gene flow may have conservation benefits (Hamilton and Miller 2015) by allowing adaptive introgression (Suarez-Gonzalez et al. 2018) and so preventing genetic collapse due to inbreeding. On the other hand, under certain conditions, interspecific hybridisation can have negative impact (Todesco et al. 2016).