Introduction

African wild bananas, sometimes called “false bananas,” Ensete ventricosum (Welw.) Cheesman (Musaceae), are monocarpic single-stemmed herbaceous plants (Cheesman, 1947; Neumann & Hildebrand, 2009; Simmonds, 1962). These plants occur where there is sufficient moisture and light, including forest gaps and clearings, shrublands, and riverine areas (Baker & Simmonds, 1953; Birmeta, 2004). Of the seven Ensete species, only E. ventricosum occurs in Uganda (Baker & Simmonds, 1953; Fuller & Madella, 2009; Lejju, Robertshaw, & Taylor, 2006). E. ventricosum has been domesticated in Ethiopia where it is widely cultivated and the processed stem, root, and flowers (but not the fruits) are eaten (Bizuayehu, 2008; Garedew, Ayiza, Haile, & Kasaye, 2017). We know little about the ecology of wild E. ventricosum.

Little is known about the ecology of most tropical species (Ghazoul & Sheil, 2010)—so in that sense E. ventricosum is typical. The challenge for anyone confronting such a vast scale of unknowns and seeking to inform management and conservation is to know where to focus and what to assess (Meijaard & Sheil, 2012). One suggestion is to identify those taxa and characteristics which influence, or may determine, the persistence of others. Such taxa are often labeled as ecological keystone species (hereafter “keystone species”; see Diaz-Martin, Swamy, Terborgh, Alvarez-Loayza, & Cornejo, 2014; Kattan & Valenzuela, 2013; Peres, 2000). Identifying such taxa and characterizing the nature and significance of their ecological contribution remain challenging. The aim is not to simply classify species into two classes (i.e., keystone and nonkeystone species) but to better recognize their role and contribution. Camera traps offer a simple option.

E. ventricosum is uncommon in the East African mountain forests but can be locally common in disturbed and open sites and at forest edges. It produces a persistent inflorescence with large amounts of nectar and protein-rich pollen and numerous fleshy fruits (Katende, Birnie, & Tengnäs, 1995; Nayar, 2010; F. S. and D. S., personal observation). We postulated that E. ventricosum is a potential keystone species based on our informal observations over 2 years that it produces flowers and fruits all year-round thus potentially supporting nectivores and frugivores when other food resources are scarce. Unfortunately, we cannot readily assess whether a species has such attributes. We propose a three-step process: (a) identify a promising species (i.e., E. ventricosum), (b) make an initial assessment, and (c) address more specific questions if justified. The second step is what we present in this article.

Here, we clarify which animals feed on E. ventricosum and when this occurs. We used camera traps to record animals that visited flowers and fruits and show that these simple and relatively low-cost methods offer novel insights that can inform conservation practice.

Methods

Site: The study was conducted in the Bwindi Impenetrable National Park (“Bwindi”) in Uganda (0°53'–1°08' S, 29°35'–29°50' E), a UNESCO World Heritage site since 1994 (Howard, 1991; International Union for Conservation of Nature, 1994). Bwindi’s climate is equatorial with two dry seasons from December to January and June to August (Kasangaki, Bitariho, Shaw, Robbins, & McNeilage, 2012). Annual rainfall ranges from 1,130 to 2,390 mm and the mean temperature ranges between 7°C and 29°C depending on elevation (ITFC unpublished). Bwindi has rugged topography and is relatively rich in plant and animal species, including several large mammals of international conservation interest such as the Mountain gorillas (Gorilla beringei beringei Matschie), Eastern chimpanzees (Pan troglodytes schweinfurthii Giglioli), and African bush elephant (Loxodonta africana Blumenbach). The forest has been influenced by prolonged human activity and much of the area has persistent herbaceous and secondary vegetation—which is known to be decades old and believed to result primarily from human activities including fires, timber extraction, and cultivation as well as from natural processes including landslides and feeding by elephants (Babaasa et al., 2004; Butynski, 1984; Ssali, Moe, & Sheil, 2017).

Camera-trap observations were conducted between November 2011 and February 2013 spanning rainy and dry seasons. We selected 40 E. ventricosum plants located in a valley containing secondary vegetation near the Institute of Tropical Forest Conservation (ITFC)—a research station at Ruhija (2,355 m asl). We deployed between five and six cameras at any one time and moved them after 30 to 40 days for seven trapping periods, facing different plants each time (see Supplemental Material for details).

Results

From a total of 892 camera-trap days over 13 months, 3,480 out of 55,850 flower visitor images and 7,704 out of 55,587 fruit visitor images were useable. The total number of verified vertebrate camera-trap “events” (i.e., an animal seen in one or more of the event images) was 1,129 and 562 for flower and fruit visitors, respectively (Tables 1 and 2). All animal images were of single individuals except for 24 events which included groups of two or three L'hoest's monkeys Allochrocebus l’hoesti P. Sclater feeding on fruit, and one event with one image of a Servaline genet Genetta servalina Pucheran photographed with an African dormouse Graphiurus murinus Desmarest in its mouth (Figure 1).

Figure 1.

Selected images of animals visiting and utilizing E. ventricosum fruits and flowers. The animals are (a) Carruther’s mountain squirrel Funisciurus carruthersi Thomas, (b) Forest giant pouched rat Cricetomys emini Wroughton, (c) L'hoest's monkey A. l’hoesti P. Sclater, (d) Olive baboon Papio anubis Lesson, (e) Nectar bat Megaloglossus woermanni Pagenstecher, (f) African dormouse G. murinus Desmarest, (g) Sunbird (Nectariniidae), (h) African palm civet Nandinia binotata Gray, and (i) Servaline genet G. servalina Pucheran.

Table 1.

Number of Flower Visitor “Events” During a Total of 475 Camera Days (n = 3,480 Useable Images Out of a Total of 55,850 Images).

Table 2.

Number of Fruit Visitor “Events” During a Total of 417 Camera Days (n = 7,704 Clear Images Out of a Total of 55,587 Images).

The most common flower visitors were the African dormouse (53.3%), Nectar bat M. woermanni Pagenstecher (43.8%), and sunbirds (family Nectariniidae; 2.4%) (Table 1 and Figure 1). Fruits were visited by Carruther’s mountain squirrel F. carruthersi Thomas (54.1%), L'hoest’s monkey (18.7%), and Forest giant pouched rat C. emini Wroughton (18.6%) (Table 2 and Figure 1). The plants were also visited by predators including Servaline genet (0.1%), African wood owl Strix woodfordii A. Smith (0.2%), and the African palm civet N. binotata Gray (0.2%) (Tables 1 and 2).

Flower visitor events were mainly recorded during the night with bird visits peaking at dusk (Figure 2). We saw clear indications of animals feeding on floral resources, possibly nectar, and searching for prey in the inflorescences. The Nectar bat and sunbirds were often recorded with their snouts and bills inserted into the perianths, while the African dormouse, African palm civet, and Servaline genet were commonly seen moving in and out of flowers.

Figure 2.

Number of flower visitor “events” with four or more observations (range = 1–597; see Table 1) versus time of day during a total of 11,241 camera hours. G. servalina (n < 4 “events”) was observed between 19:00 and 20:00 hours.

Fruit visitors exhibited both diurnal and nocturnal activity patterns with Carruther’s mountain squirrel, L'hoest’s monkey, and Olive baboon P. anubis Lesson observed during daytime, while the African dormouse, Forest giant pouched rat, and African soft-furred mice Praomys sp were observed at night (Figure 3). Several animals were recorded holding and eating fruits (see Figure 1). Monkeys and baboons also appeared to be touching and smelling fruits to gauge if they were ripe. They often plucked the fruits (sometimes more than one finger at a time), placed them in the mouth, and used their teeth to remove the skins. The larger rodents (i.e., Carruther’s mountain squirrel and Forest giant pouched rat) appeared to first nip off the tip of the fruits and then fill their often distended cheeks with pulp and seeds. The African wood owl and the African palm civet may have been seeking prey (Figure 1 and Table 2).

Figure 3.

Number of fruit visitor “events” with four or more observations (range = 1–300; see Table 2) versus time of day during a total of 6,987 camera hours. Praomys sp, S. woodfordii, and N. binotata (n < 4 “events”) were observed between 21:00 and 01:00, 00:00 and 01:00, and 20:00 and 21:00, respectively.

Discussion

Our study is the first to describe the animals visiting African wild bananas. Our data show that bats and sunbirds form the bulk of flower visitors, while primates and rodents are the most common fruit visitors. Flower visitors are likely involved in pollination, while fruit visitors likely help deposit seeds away from the parent crown thereby reducing seed mortality (Janzen, 1970; Liu, Li, Wang, & Kress, 2002; Nyiramana, Mendoza, Kaplin, & Forget, 2011). We also found that flower visitors were mainly nocturnal, while fruit visitors exhibited both diurnal and nocturnal activity patterns. This indicates that E. ventricosum can sustain a diverse faunal community with some temporal segregation. Given our observations it seems reasonable to suppose that these species help maintain pollinators (especially bats and sunbirds) and seed dispersers that can themselves support other species. For example, the Nectar bat and sunbirds are recognized pollinators for a wide range of plants including Kigelia africana, Nuxia congesta, Albizia gummifera, Musa sp., Dalbergia sp., and Anthocleista vogelii (Fleming & Muchhala, 2008; Fujita & Tuttle, 1991; Nsor, 2015; Weber, Kalko, & Fahr, 2009). Few studies have determined what limits the abundance of these animals, but one multisite evaluation of banana (Musa spp.) pollinating bat species in tropical Brazil has shown that their abundance is positively correlated to local food resources (Luz, Costa, & Esbérard, 2015). We suspect that dormice too may play a role in pollen transfer and pollination—and noting that the role of such rodents in pollination has been widely overlooked (Ratto et al., 2018)—this indicates the need for further study (see e.g., Biccard & Midgley, 2009).

Since E. ventricosum produce flowers and fruits year-round (like most true bananas), and are visited when other flowers and fruits are scarce, we infer that these plants can support species dependent on sustained access to such resources. Our observations suggest that more animals tend to visit the plants during the dry season compared with the wet season—which likely reflects the lower abundance of other similar food resources in this period. Trees bearing fleshy fruits tend to be scarce during or at the beginning of the dry season—a trend observed for selected fruiting species in both Bwindi and the nearby Nyungwe National Park (Adamescu et al., 2018; Polansky & Robbins 2013; Sun et al., 1996). Like fig trees (Ficus spp.) in many lowland forests, E. ventricosum has passed the initial evaluation given its use by a number of other species and thus has the potential to be viewed as a “keystone species” (see Diaz-Martin et al., 2014; Gautier-Hion & Michaloud, 1989; Lambert & Marshall, 1991; Mills, Soulé, & Doak, 1993; Peres, 2000; Terborgh, 1986). In Bwindi, there are some other tree species, including Ficus densistipulata De Wild., Psychotria mahonii C. H. Wright, and Myrianthus holstii Engl. that exhibit low seasonal fluctuation in fruit availability (none are common, e.g., M. holstii occurs at a density of just a few stems ha⁻¹, see Kissa & Sheil, 2012) and may together with E. ventricosum support frugivores during periods of scarcity (see the online supplementary material of Polansky & Robbins, 2013). E. ventricosum has thus passed our initial screening that aimed to assess if it might have some properties expected for a keystone species. Further more comprehensive evaluations would be needed to formalize this result and demonstrate that other species would be reduced or lost if E. ventricosum were absent (see e.g., Peres, 2000). Managers, however, may be more interested in simply using these plants to bolster food for valued pollinators and frugivores and to perhaps support forest recovery (see later and Supplemental Material).

E. ventricosum plants may benefit a broader range of species than we observed—for example, observations elsewhere have shown that primates will often avoid some food species for long periods but still rely on them when other preferred foods are unavailable (Marshall & Wrangham, 2007). While neither gorillas nor chimpanzees nor elephants visited our selected plants during the study, we believe all three would find them attractive food resources as all these animals occasionally seek out and damage banana plants (Musa spp.) in neighboring fields (various local people’s personal communication and F. S. and D. S., personal observation).

Implications for Conservation

Our study and others in tropical and temperate regions (e.g., Burton et al., 2015; Krauss, Roberts, Phillips, & Edwards, 2018; Steen, 2017) show how camera traps can help characterize ecological processes and species. While such data are insufficient to rigorously determine “keystone functions,” they provide useful indications. Indeed, given the improving affordability, reliability, and accessibility of camera traps along with our profoundly limited knowledge of most species and interactions, it is important to examine the opportunities that are now possible through the simple application of such tools (Burton et al., 2015; Sheil, Mugerwa, & Fegraus, 2013). We encourage efforts to explore these opportunities.

Our observations indicate that E. ventricosum benefits a range of species, even when flowers and fruits are otherwise scarce. We believe that establishing E. ventricosum in disturbed landscapes through field-grown suckers or seedlings (see e.g., Katende et al., 1995; Tripathi, Matheka, Merga, Gebre, & Tripathi, 2017) would bolster ecological activity and conservation value and may facilitate forest recovery. Furthermore, other herbaceous species that sustain flower and fruit resources in other sites (e.g., many Musaceae and other Zingiberales) are prime contenders for providing similar valuable ecological roles and thus for protection and restoration (see e.g., Liu et al., 2002; Marod, Pinyo, Duengkae, & Hiroshi, 2010).

Supplemental Material