Ecological fitting is a sufficient driver of tight interactions between sunbirds and ornithophilous plants

Abstract Plant–bird pollination interactions evolved independently on different continents. Specific adaptations can lead to their restriction when potential partners from distant evolutionary trajectories come into contact. Alternatively, these interactions can be enabled by convergent evolution and subsequent ecological fitting. We studied the interactions between New World plants from the genus Heliconia, Asian plants of genus Etlingera and African sunbirds on a local farm in Cameroon. Heliconia spp. evolved together with hummingbirds and Etlingera spp. with spiderhunters —an oriental subgroup of the sunbird family. Sunbirds fed on all studied plants and individual plant species were visited by a different sunbird spectrum. We experimentally documented a higher number of germinated pollen grains in sunbird‐visited flowers of Etlingera spp. For Heliconia spp., this experiment was not successful and pollen tubes were rarely observed, even in hand‐pollinated flowers, where enough pollen was deposited. The analyses of contacts with plant reproductive organs nevertheless confirmed that sunbirds are good pollen vectors for both Heliconia and Etlingera species. Our study demonstrated a high ecological fit between actors of distinct evolutionary history and the general validity of bird‐pollination syndrome. We moreover show that trait matching and niche differentiation are important ecological processes also in semi‐artificial plant‐pollinator systems.

. Similarly, bird-visited plants can be found in many families. Flores, Ornelas, Wethington, and Arizmendi (2019) reported 105 plant families that contain plant species visited by hummingbirds, 54 of which they classified as ornithophilous or partly ornithophilous. In Australia, Ford, Paton, and Forde (1979) reported 31 plant families visited by birds.
In 15 of these families, not only bird-visited but also bird-pollinated species were found. In sunbirds, Cheke et al. (2001) documented the occurrence of food plants in 94 families. For South Africa, Rebelo (1987) reported 30 plant families that contained at least one ornithophilous species.
Many authors have highlighted differences among adaptations in independently evolved bird-pollination systems as well as similarities which are the consequences of convergent evolution. One of the most famous convergent adaptations of specialized nectarivorous birds is thin bills and tubular tongues which enable them to drink nectar from flowers, even though exact bill and tongue parameters differ among individual bird groups (Paton & Collins, 1989).
The common flower properties of ornithophilous plants are defined by the bird-pollination syndrome. Bird-pollinated flowers are usually red or orange, without scent and produce a lot of nectar (Cronk & Ojeda, 2008). The most often reported example of specific adaptations is related to the fact that New World hummingbirds usually hover whereas Old World sunbirds perch when feeding (Cronk & Ojeda, 2008;Fleming & Muchhala, 2008;Pyke, 1980). As a consequence, we can find many hummingbird-pollinated plant species which have their flowers oriented into free space in the New World (Westerkamp, 1990) and different plant adaptations which enable perching of passerine birds in the Old World (Frost & Frost, 1981; de Waal, Anderson, & Barrett, 2012). Nevertheless, this dichotomy is not without exceptions and there are known Old World plants adapted to sunbird hovering (Janeček et al., 2011;Janeček, Bartoš, & Njabo, 2015;Padyšáková & Janeček, 2016). Similarly, as demonstrated by the pollination systems of two Heliconia spp., foraging behavior, hovering versus perching, can be species-specific in hummingbirds (Taylor & White, 2007). Other discussed differences were assumed to be in nectar properties (Baker, Baker, & Hodges, 1998). Johnson and Nicolson (2008) Abrahamczyk et al. (2017), who, however, showed that sunbird-pollinated plants have more diverse nectar compositions than hummingbird-pollinated ones.
The differences and similarities discussed above lead to the question: how do different evolutionary trajectories and convergent evolution processes effect the compatibility of individual pollination systems and what is the ecological fit (Janzen, 1985)? In simple terms: can plants be effectively pollinated by birds and can birds feed on plants when they have different evolutionary histories? The answer to this question is important for understanding the consequences of divergent and convergent evolution processes. Moreover, with the recent acceleration of global environmental changes, it can help us to predict scenarios when native partners are lost thanks to biodiversity degradation and/or new invasive partners occur (Cox & Elmqvist, 2000).
The feeding of nectarivorous birds on non-native plants has been reported in America (Maruyama et al., 2016), Asia (Ghadiriani, Qashqaei, & Dadras, 2007), Australia (Ford et al., 1979), and Africa (Geerts & Pauw, 2009a). Records of bird pollination of alien plants are much less common. Sunbirds were recorded as pollinators of invasive tobacco Nicotiana glauca, which is naturally pollinated by hummingbirds in America (Ollerton et al., 2012). The sunbirds even pollinate it in a similar way to hummingbirds using hovering flight (Geerts & Pauw, 2009a). Also, in the New World it was shown that the ornamental plant Strelitzia reginae, which is native to South Africa, can be pollinated by the local common yellowthroat warbler Geothlypis trichas (Hoffmann, Fortier, & Hoffmann-Tsay, 2011).
There are, nevertheless, no comparative studies which document how these non-native interactions are affected by evolutionary history and how they function in more complex systems, where other ecological processes, like niche differentiation, might play a role.
In our study, we focus on interactions among African sunbirds and alien ornamental plants on a local farm in Cameroon. These farms produce ornamental plants of the spiderhunter-pollinated genus Etlingera from Asia (Sakai, Kato, & Inoue, 1999;Sakai, Kawakita, Ooi, & Inoue, 2013) and of the hummingbird-pollinated genus Heliconia from America (Linhart, 1973;Temeles & Kress, 2003). Spiderhunters are passerine birds of the genus Arachnothera, part of the sunbird family Nectariniidae (Cheke et al., 2001), whereas hummingbirds (family Trochilidae) represent a very distant phylogenetic lineage (Prum et al., 2015). Thus, the farms represent a unique place to study the consequences of specific and convergent adaptations, as well as possible ecological community processes such as niche differentiation. Using this system, we tested the following three scenarios: Under partial and full ecological fitting scenarios, we expect that sunbirds will show niche differentiation and plants are likely to differ in the spectrum of sunbird visitors. We also decided to determine nectar production and nectar concentration of individual plants to see whether observed patterns can be simply explained by offered rewards.

| Study site
The study was performed near Bokwango village (4°8′6″N, 9°13′16″E, 940 m a.s.l.), which is administratively part of Buea town (NW Cameroon). The study site was a local farm where the four target species of genus Heliconia and three taxa of genus Etlingera are commercially cultivated for flower production, within an area of approximately 1 ha.

| Plant species
All studied Heliconia spp. (Figure 1) (Andersson, 1981;Berry & Kress, 1991;Govaerts & Kress, 2019) and H. bihai x caribaea cv. Jacquinii is native in Grenada. In contrast to the genus Heliconia, genus Etlingera is native to Indo-Pacific Asia. At the study site, there are two cultivated forms of E. elatior which represent the two extremes of the involucral bract color range occurring in nature: the red form with red involucral bracts and white form with whitish-pink involucral bracts (Sabu & Smisha, 2013).
Both Heliconia and Etlingera spp. produce shoots bearing a maximum of one inflorescence. Heliconia spp. produce inflorescences almost always on the leafy shoots and in Etlingera, there are special shoots bearing either inflorescence or leaves. The individual flowers of both genera flower for a single day (Berry & Kress, 1991;Choon & Ding, 2016). In Heliconia, flowers are usually produced in red and/ or yellow inflorescence bracts. Depending on the Heliconia species up to 50 flowers can be produced on one bract during the flowering period (Berry & Kress, 1991). Whole inflorescence can be either pen- In E. hemisphaerica, the involucral bracts are red whereas in E. elatior, the involucral bracts can range from red (red form) to whitish-pink (grown as white form in our study area) (Sabu & Smisha, 2013).
Most of the Heliconia species seem to be self-compatible (Kress, 1983) as was shown also for H. latispatha (Kress, 1983) and H. bihai Rojas-Sandoval, & Planas, 2008). Considering the genus Etlingera, it was shown that E. elatior is self-compatible but self-pollination is much less effective than cross pollination (Sabu & Smisha, 2013).

| Plant traits
To compare the nectar value of individual species, we measured nectar production over 12 hr. The day before the nectar measurements, we marked flower buds and covered them with dense nets to prevent visitors from consuming the nectar after the flower had opened. If flowers opened on the day of nectar measurement, we marked them again as experimental flowers in the morning (around 6 a.m.). After 12 hr, we collected and measured nectar volume and concentration. Nectar volume was measured by Hamilton syringe

| Sunbird visitors
Observations of visitors were performed on 11 days between 14 May 2018 and 2 July 2018. During each day, we observed one plant (flowering shoot) of each species. Observation was performed using AEE MagiCam 70S (AEE Technology Co., Ltd) sport cameras.
Cameras were permanently connected to power banks. The duration of observations of individual plants on individual days differed due to logistical issues (charging of power banks, downloading data etc.) and some technical errors. In total, each species was observed approximately for 90 hr (Table S1). From the acquired video material, we extracted information on frequencies of visits per inflorescence, per flower and frequency of flower visits when the visitor was in contact with plant reproductive organs.

| Experiment on sunbird pollination
To assess the significance of sunbirds in plant pollination, we set up a manipulative experiment. In this experiment, we covered randomly selected inflorescences with a sparse net with mesh size of 1 cm to exclude bird visitors. As a control, we marked nonmanipulated inflorescences and allowed them to be naturally exposed to visitors.
The day after in the morning, when flowering was over, we collected gynoecia and fixed them in 96% ethanol for future analyses.
In the laboratory, we soaked the gynoecia in distilled water for 24 hr. and then put them into 10 M NaOH for 24 hr. Thereafter, we carefully washed the gynoecia in distilled water and left them overnight in aniline blue dye (Dafni, Kevan, & Husband, 2005). Pollen tubes were counted under a fluorescence microscope. Styles from one inflorescence collected on the same day represented one sample. The experiment took place from 18 till 24 May 2018.
To test whether there are any problems with pollen tube germination and/or staining (e.g., incompatibility of local pollens, problems with pollen tube coloring etc.), we also set up a hand-pollinated control. In this subexperiment, we supplied stigmas of all species with pollen from distant flowering shoots in the morning. These gynoecia were then processed in the same way as those from the bird exclusion experiment.

| Statistical analyses
Data from most of the datasets were not normally distributed, and therefore, we used analogical nonparametric permutation methods included in the PERMANOVA program which is an extension of the software PRIMER (Anderson, Gorley, & Clarke, 2008). Data on nectar volumes, visitation frequencies, number of pollen tubes were log (x+1) transformed to decrease the effect of extreme values of dependent variables. In the analyses on differences between genera, we treated species ID nested in Genus and Date (except for analysis on flower length) as random factors. When comparing species, we used Date as a random factor. In analyses on nectar production and concentration, when more samples were taken from one plant, individual plant ID was also treated as a random factor nested in plant species ID.

| Plant traits
The studied plants differ in total flower length ( Figure S1). Heliconia spp. have longer flowers than Etlingera spp., but this difference was only marginally significant (perm. ANOVA, Genus: F 1,63 = 7.86, p = .0557). Nevertheless, according to the video analysis the total

| Sunbird visitors
Studied plant species were visited by three sunbird species: Cinnyris chloropygius, Cyanomitra olivacea, and Cyanomitra oritis. Although the cameras were set up to record sunbirds (i.e., cameras were as far as possible from the plants to avoid scaring bird visitors, and each inflorescence was recorded from just one side), we also observed some insect visitors. The most common were honeybees but also butterflies and ants were observed on both Heliconia and Etlingera spp.
The most frequently visited species were E. elatior white form  (Table 1).
Etlingera species were visited mainly by C. oritis., H. latispatha was visited mainly by C. chloropygius followed by C. olivacea and C. oritis. A much higher frequency of visits to H. latispatha was recorded per inflorescence than per flower, and this difference was more obvious than for Etlingera species (Figure 3a,b). When we tested the opposite, that is how bird species differed in the spectrum of visited plants, there were significant differences between bird species when considering frequency of visits per inflorescence but only a marginally significant difference when considering frequency per flower (Table 1; Figure S2).  (Table 1; Figure   S2c).

| Experiment on sunbird pollination
After a controlled hand pollination supplement, we observed germinating pollen grains on the stigmas of Etlingera spp. (Figure 4c).
In contrast, almost no germinating pollen grains were observed on the stigmas of Heliconia spp. ( Figure S3). The same pattern was observed in the experiment (Figure 4). Pollen tubes on Heliconia plants moreover germinated on the stigma and did not grow into the style (Figure 4b). Plant species differed in pollen tube number, sunbird exclusion had a negative effect on the number of pollen tubes, and this treatment was species-specific (perm. mixed-effect model; Plant: F 6,334 = 11.81, p = .0001; Treatment: F 1,334 = 6.90, p = .0092; Plant x Treatment F 6,334 , p = .0020).
Genera differed in the number of pollen tubes and were differently affected by the treatment (perm. mixed-effect model; Genus:

| D ISCUSS I ON
Our study demonstrated a high degree of ecological fitting. We also show that distant evolutionary history of individual actors cannot prevent fundamental ecological processes such as occupation of new niches by alien plants or niche partitioning.
Our study complements the ideas of Janzen (1985) that most of the interactions we see around us are not necessarily the consequences of coevolutionary processes in a given place, but of simple ecological fitting. Janzen´s main arguments were that biological communities are dominated by widespread invasive species, which have extended their range and are not originally adapted to most of their habitats. From this point of view, our study can be considered experimental support of this statement showing that organisms from three different continents can create a complex interacting community on the basis of ecological fitting. However, we have also shown that the Asian Etlingera spp. established tighter interactions with African sunbirds. Etlingera spp.
were visited legitimately more often, and sunbirds seem to be very effective pollinators. From the sunbirds' point of view, it does not seem to be a simple consequence of bigger rewards, because both Etlingera and Heliconia spp. produced similar amounts of nectar. Etlingera spp. even had a slightly lower nectar standing crop (but only marginally significant). We assume the reason for this perfect ecological fit could be a more comfortable perching position when feeding and from the plant's point of view, the compact inflorescence enables only legitimate entering of the flowers. Pollen of Etlingera spp. is then precisely placed in high quantity on the bills of perching sunbirds. Heliconia spp. had weaker but still functional interactions with the local sunbirds which is amazing if we consider that the genus Heliconia is in the oldest known clade of hummingbird-pollinated plants (Iles et al., 2017). We were surprised how H. rostrata, which is pollinated by hovering hummingbirds (Iles et al., 2017), is able to precisely place pollen on the heads of perching C. olivacea and C. oritis (Video S2). Heliconia spp. were not visited with the same frequency indicating that also intragenus trait differences are important. We assume that the high visitation rate of H. latispatha is mainly due to fact that the flowers are not deeply hidden in bracts and can be more easily reached in both legitimate and illegitimate ways. Although we observed frequent contacts of Heliconia reproductive organs and even pollen on the heads of sunbirds, we were not able to evaluate pollinator effectivity directly by counting germinated pollen tubes. There were almost no pollen tubes observed, or they did not grow inside the style. Because this was the case not only for experimental treatments but also for hand pollination, we TA B L E 1 PERMANOVA analyses on differences between visitor communities on plant species (i.e., plant species as an explanatory variable), differences in visitor communities between genera (i.e., plant genera as an explanatory variable), and differences in the spectrum of plants visited by individual bird species (i.e., bird species as an explanatory variable) and comparing visitation community on individual plant species (i.e., plant species as an explanatory variable). Data were log (x+1) transformed. Date was used as a random effect in all analyses. In addition, to test the effect of genus we considered plant species nested in factor genus as a random factor  (Betts, Hadley, & Kress, 2015), it is possible that these negative results are the consequence of a methodological mistake. This can be supported by the fact that Heliconias on the farm produced fruits (including H. rostrata of which we did not observed any pollen tubes).
Using this example of a semiarbitrary plant-bird community from three continents, we demonstrated that potential ornithophilous invasive plants can be easily incorporated into local communities.
In consequence, we are delivering a similar message from Africa and reproductive success of native coflowering species (Morales & Traveset, 2009). In consequence, the effects of these plantations on local flora need to be studied in detail.

ACK N OWLED G M ENTS
We are grateful to the Bokwango village community, which greatly