Introduction

Cacao (Theobroma cacao L., Malvaceae) is one of the world’s most important cash crop species. Yet, despite its economic importance, surprisingly little is known about the pollination ecology of the species. Cacao is generally considered to be mainly pollinated by tiny midges (Ceratopogonidae) but knowledge on the sexual reproduction is based almost exclusively on limited evidence from cultivated cacao, often from parts of the non-native range of the species (Wellensiek 1932; Soria 1973; Soria and Wirth 1974; Soria et al. 1980), and is nonexistent for wild populations.

Theobroma cacao is generally assumed to be an outbreeding species, because of its floral morphology and the occurrence of a self-incompatibility system (e.g. Knight and Rogers 1955; Falque et al. 1995). However, while self-pollination of a single flower is basically impossible in Theobroma due to the flower morphology (Lieberei and Reisdorff 2007), it has been suggested that geitonogamy might be more common in the wild than commonly assumed (Lanaud et al. 1987). Indeed, Chumacero de Schawe et al. (2013) found selfing in wild and cultivated cacao trees. According to Winder and Silva (1972), the natural fertilization rate of cacao reaches just 4 % at the peak of the flowering season.

Which floral rewards attract pollinators to T. cacao is unclear (Wellensiek 1932; Soria et al. 1980; Young et al. 1984) but the unique floral structures suggest highly specialized pollination adaptations for pollination by flying insects (Young et al. 1987). Theobroma cacao flowers have nectaries on the pedicels, sepals and ovaries, “guide lines” of petals and staminodia (Stejskal 1969), and are fragrant (Young et al. 1984). The brightly coloured spatule ligule of the flowers could also play a role in pollinator attraction (Young et al. 1984). Finally, the flowers have ultraviolet light reflectance/absorbance, which probably attract pollinators (Young et al. 1987).

Theobroma cacao is pollinated by insects and has no wind pollination (Harland 1925). It is commonly considered that the tiny ceratopogonid midges Forcipomyia blantoni Soria and Bystrak (Soria et al. 1980), Forcipomyia youngi Wirth and F. quatei Wirth (Young 1983) are the most effective cacao pollinators (e.g., Billes 1941; Posnette 1950; Brew and Borman 1993). Thus, Saunders and Bowman (1956) and Young (1983) pointed out that natural pollination seems to be a limiting factor in the production of cacao due to scarcity of Forcipomyia midges. Soria (1974) indeed found a positive correlation between midge population levels and the intensity of flowering, fruit set, and yield. However, the quantity of flowers that become a ripe fruit in cacao is known to be lower than the pollination rate, because a large number of young fruits wilt (Wellensiek 1932; Groeneveld et al. 2010). Presumably for this reason, Young (1983) found no correlation between pollinator abundance and fruit set. Lately, Groeneveld et al. (2010) obtained experimental evidence for stronger yield limitation by pollination than by plant resources.

Despite this evidence for the role of Ceratopogonid midges as pollinators, cacao pollinator surveys have often found low abundances of these midges (Entwistle 1972; Young 1986b; JL Groeneveld pers. comm.). Indeed, many other potentially pollinating insects are attracted to cacao flowers, possibly seeking nectar or pollen. Thus, in addition to ceratopogonid midges, members of other Dipterian families such as Phoridae, Sciaridae and Drosophilidae have been captured on cacao flowers (Young 1986b). Young (1987) found mainly thrips (70 %) and aphids on cacao flowers, and only one ceratopogonid individual with pollen. Cecidomyiid midges have also been found carrying cacao pollen grains (Lucas 1981). Stingless bees are further occasional floral visitors of cacao (Soria 1975; Young 1981). Thus, although ceratopogonid midges are considered the main pollinator agents, evidence in this regard is actually rather contradictory and one major open question regards the flower visitors of non-cultivated cacao plants in their natural habitat.

In view of the numerous uncertainties surrounding the pollination ecology of T. cacao, in our study we set out to document flower visitors in wild and cultivated plants of T. cacao in the northeast lowlands of Bolivia where presumed wild forms of cacao occur patchily in the natural forest (Chumacero de Schawe et al. 2013). In addition, farmers cultivate commercial hybrids of both the Forastero and Criollo cultivars. The objectives of our study were: (1) to identify floral visitors and (2) to compare species composition and relative abundance of Diptera and Hymenoptera as main visitor groups on wild and cultivated cacao flowers. The hypothesis of the study was that the pollinator assemblages differ between wild and cultivated populations of T. cacao, both with respect to taxonomic composition and abundance of floral visitors and that many flower visitor species, in addition to ceratopogonid midges, function as potential pollinators.

Materials and methods

Study species

The flowers of T. cacao L. are pentamerous, white with red nectar guides, with five separate or fused petals ca. 10 mm long, consist of a proximal petal hood or pouch, each of which encloses an anther, have five rigid staminodes, an outer whorl of the androecium, surround the pistil like a fence and grow in clusters on the tree trunks. The fruits are ellipsoid to elongated, segmented (10 segments), 15–19 cm in length and greenish, orange or red when ripe, and are dispersed by gravity and mammals. Anthesis occurs at dawn and anther dehiscence takes place throughout the morning hours (Young et al. 1987). Each flower produces about 14.000 pollen grains (Massaux et al. 1976). Flowering phenology of T. cacao is seasonal, varying between genotypes and in response to local climatic conditions, but usually being highest at the onset of the rainy season (Paulin et al. 1983). In Bolivia, the main flowering season is in September–October for wild trees and October–November for the cultivated ones, although some flowering occurs throughout the year among the cultivated trees (C. Chumacero de Schawe, unpubl. data). The flowers are receptive for 1 day; stigma and style receptivity is high at morning and early afternoon (Young et al. 1987). Unpollinated flowers drop after about 2 days. Flowers pollinated with sparse pollen grains fall without setting fruit (Lachenaud 1995). Less than 5 % of the flowers set fruit in cacao plantations (Entwistle 1972).

Study sites and sampling

We studied flowers of cultivated (Fig. 1) and wild T. cacao trees (Fig. 2) on the lands of the native Tacana community Napashi (67°52′17.4″S–14°5′20.6″W) in the Northeast lowlands of Bolivia (Chumacero de Schawe et al. 2013). For our study, we selected 59 trees (26 wild and 33 cultivated trees) from four cultivated plantations and two patches of wild cacao trees. We sampled 2237 flowers (1077 from wild and 1160 from cultivated trees) during 13 days between 14.10. and 16.11.2008, using “Insect Tangle Trap Glue” (Tanglefoot©) to catch pollinators of cacao. The weather conditions during the fieldwork were mostly partially cloudy to sunny; and the temperature varied between 28 and 34 °C. We put some glue around the staminodes using a syringe needle (Fig. 3). The marking time of the flowers was around 07:30 am; they were checked 24 h later. We sampled a mean of seven trees and 125 flowers per day; the number of flowers marked depended on the availability, as wild cacao trees produce much fewer flowers than the cultivars. Flowers with arthropods (N = 448) were stored in ethanol (70 %) and the arthropods were rinsed with a solvent (Terpentin) to remove the glue. The insects were observed under the stereomicroscope to detect the presence of pollen, before and after the cleaning. All the specimens were determined to insect order level, Diptera and Hymenoptera were determined to morphospecies level, and Ceratopogonidae specimens were determined to species level. The Ceratopogonidae specimens were prepared in Canada balsam for determination. The samples were stored in the “Ceratopogonidae of the Neotropics Collection” at Universidad Nacional de la Plata in Argentina. The other Diptera specimens were stored in the “Colección Boliviana de Fauna”, Mayor de San Andrés University, Bolivia. All the other insect samples are currently at the Agroecology Department of Georg August University of Göttingen, Germany.

Fig. 1
figure 1

Cultivated Theobroma cacao tree. Photo credits: C. Chumacero de Schawe

Full size image
Fig. 2
figure 2

View of the forest where wild Theobroma cacao trees occurred. Wild T. cacao trees are 12 m height in average. Photo credits: C. Chumacero de Schawe

Full size image
Fig. 3
figure 3

Hymenoptera trapped on cacao flower using tangle trap glue. Photo credits: C. Chumacero de Schawe

Full size image

Data analysis

We assessed the relative abundance of the floral visitors trapped on cacao flowers per order. Dominance was calculated as the relative abundance of the most abundant insect orders. In particular, species richness and relative abundance of Diptera and Hymenoptera were calculated.

We applied a rarefaction approach prior to assess a G-test to compare species richness of Diptera and Hymenoptera between wild and cultivated cacao. Relative abundance of Diptera and Hymenoptera between wild and cultivated cacao was compared performing a G-test. The Sorensen Index was calculated to compare the similarity in species composition of Diptera and Hymenoptera between both wild and cultivated cacao plants. Rank-abundance curves of Hymenoptera and Diptera species were plotted to compare the species richness in wild and cultivated cacao. We used χ2 tests to compare the proportion of species composition between wild and cultivated cacao.

Results

The mean success capture of insects on wild cacao flowers was 42 insects/day and 27 insects/day on cultivated cacao flowers. The mean captured rate of insects per flower on both wild and cultivated cacao was 0.3. A total of 631 insects were collected belonging to seven orders, 331 insects were found on wild and 300 on cultivated cacao trees.

Abundance

The order Hymenoptera was the most abundant group found on wild (n = 143) and cultivated (n = 118) cacao flowers. Diptera was the second group in abundance on wild (n = 79) and on cultivated cacao (n = 45) (Fig. 4). The abundance of Hymenoptera did not differ significantly between wild trees (54 % of visitors) and cultivars (43 %) (G test; df = 1; P = 0.31; Fig. 4). The most abundant family of Hymenoptera on wild cacao was Eulophidae (59 % of individuals) and on cultivated cacao Platygastridae (47 %) (Table 1).

Fig. 4
figure 4

Relative abundance of the insect orders visiting wild and cultivated Theobroma cacao (*P < 0.05)

Full size image
Table 1 Hymenoptera families trapped on flowers of wild and cultivated Theobroma cacao
Full size table

The abundance of Diptera was not significantly different between wild (30 % of visitors) and cultivated cacao (17 % of visitors, G test; df = 1; P = 0.078; Fig. 4). The most abundant families of Diptera were Chloropidae, Phoridae and Ceratopogonidae on both wild and cultivated cacao (Table 2). The abundance of Sciaridae in both wild and cultivated cacao was similarly low (G test; df = 1; P = 0.615).

Table 2 Diptera families trapped on flowers of wild and cultivated Theobroma cacao
Full size table

The relative abundance of the orders Thysanoptera (4 and 15 % on wild and cultivated cacao, respectively; G test; df = 1; P = 0.019) and Hemiptera (4 and 19 % on wild and cultivated cacao, respectively; G test; df = 1; P = 0.0023) were significantly higher on flowers of cultivated cacao.

Diversity and species richness

We found 59 species of Hymenoptera belonging to 11 families and 49 species of Diptera belonging to 15 families. Species richness of Hymenoptera was similar on cultivated (29 species) and on wild cacao trees (24) (G test; df = 1; P = 0.58). Species composition of Hymenoptera between both cacao types was not very similar (Sorensen similarity = 0.45). The proportions of Hymenoptera species did not differ significantly between wild and cultivated cacao (χ2 test, df = 1, P = 0.085; Fig. 5A). The rank-abundance curve of Hymenoptera on wild and cultivated cacao were quite similar, denoting similar species evenness in both cases (Fig. 5a). We found the presence of pollen only on a single specimen of Encyrtidae.

Fig. 5
figure 5

Rank-abundance curves of A) Hymenoptera and B) Diptera on wild and cultivated Theobroma cacao

Full size image

Species richness of Diptera did not differ between wild (24 species) and cultivated cacao (19) (G test; df = 1; P = 0.542). However, the species composition of Diptera in wild and cultivated cacao differed considerably (Sorensen similarity = 0.33). The proportion of Diptera species differed significantly between wild and cultivated cacao, accordingly the rank-abundance curves have different shape (χ2 test, df = 1, P = 0.002; Fig. 5b). Additionally, the slope on wild cacao is steeper than the slope on cultivated cacao, denoting lower species evenness. Ceratopogonidae represented 12 % of the Diptera found on wild cacao flowers, including five species, and 13 % of the Diptera trapped on cultivated cacao flowers, including three species (Table 3). Only Forcipomyia (Thyridomyia) nana (Macfie) was found on both wild and cultivated cacao flowers. No pollen grains were found on Ceratopogonidae midge bodies. Except for a single specimen of Culicoides sp. 2, all the midges were females. Only one specimen of Cecidomyiidae was found on wild cacao flowers. Hemiptera was represented mainly by aphids and Thysanoptera by thrips. Only three spiders were captured on cacao flowers.

Table 3 Ceratopogonidae species trapped on wild and cultivated Theobroma cacao flowers
Full size table

Discussion

The relatively low capture rate of insects on cultivated cacao flowers (27 insects/day) evidenced a lower abundance of floral visitors in comparison to wild cacao flowers (42 insects/day). Although wild cacao trees had fewer flowers than cultivated cacao (37 vs. 56 per tree; C. Chumacero de Schawe, unpublished data). Young (1986a) found also more midges on abandoned cacao plantations. He observed lower flowering in abandoned cacao plantations as well; probably because of intense shade as in our wild cacao populations.

We found cacao pollen only on a single specimen of the hymenopteran family Encyrtidae. Partly, this may be due to the collecting process that involves trapping with glue followed by rinsing and cleaning, even though this method has previously been used to identify cacao pollinators (Groeneveld 2008). One possibility would be that most insects were caught during their first visit to a flower, so that no pollen load was present from a previous visit. On the other hand, previous studies have also reported low pollen load on flower visitors (Winder 1977) as well as pollen limitation in fruit production on cacao trees (Bos et al. 2007). In any case, regardless of the reason for the extremely low number of insects with pollen, we were unfortunately unable to directly identify the actual pollinators and were only able to compare the visitors.

The most abundant and diverse insect order on both wild and cultivated cacao trees was Hymenoptera. The taxonomic composition of the hymenopteran assemblages was largely similar between wild and cultivated cacao, although the family Eulophidae was the most abundant family on wild cacao, whereas on cultivated cacao it was the Platygastridae. Most species of Eulophidae are parasitoids whose larvae feed on insects or spiders while some species are known to parasitize Thysanoptera (Hanson and Gauld 1995). Species of Platygastridae are exclusively parasitoids. Adult parasitoids need sugar for their survival. In this sense, they might pollinate flowers of cacao while they suck nectar. Diptera was the second most commonly collected insect order on cacao flowers in our study but only a low proportion corresponded to midges of the family Ceratopogonidae that is widely considered to be the main pollinators of cacao (e.g., Billes 1941; Posnette 1950; Brew and Borman 1993). In our pollinator survey, we trapped 13 ceratopogonid specimens belonging to seven species. This low number of Ceratopogonidae found on cacao flowers supports previous observations by Groeneveld (2008) who caught only 28 ceratopogonids during extensive sampling in cultivated cacao plantations in Indonesia. Among the seven species recorded by us, only Forcipomyia (Thyridomyia) nana was present on both wild and cultivated cacao flowers. A larger sample size would be necessary to test if the species composition of Ceratopogonidae midges differs between wild and cultivated cacao. According to our observations, species composition of ceratopogonid midge assemblages in Bolivia differs from those in other studies. We thus did not find any of the ceratopogonid species reported by Winder (1977) in Brazil or by Sánchez et al. (2001) in Venezuela. In contrast to other studies (e.g., Winder 1977), we only trapped one Cecidomyiidae midge on wild cacao.

Among the remaining Diptera, Chloropidae and Phoridae flies were abundant. Chloropidae larvae are mainly phytophagous, while Phoridae has scavenger, herbivore, parasitoids and predatory species (Brown 1992). The adults of these families might carry cacao pollen when they feed on flower nectar. Furthermore, Phoridae are known to pollinate species of the genus Herrania (Young et al. 1984), a genus that is closely related to Theobroma (Cuatrecasas 1964). Accordingly, Young et al. (1984) suggested that Phoridae might play a role in cacao pollination.

The rank-abundance curves of Hymenoptera on wild and cultivated cacao showed similar shapes denoting a similar proportion of species, where the high-ranking species were more abundant than the low ranking species. The same morphospecies of Eulophidae was the most abundant species on both wild cacao and cultivated cacao. In contrast, the proportion of Diptera differed between both wild and cultivated cacao. Accordingly, the slopes of the rank- abundance curve were different; it was steeper on wild cacao denoting lower species evenness than on cultivated cacao.

The relatively high abundance of aphids (Hemiptera) found on cacao flowers evidenced economically important pest infestation on cacao cultivars (also found by Maas et al. 2013), whereas aphids were very rare on wild cacao. It has been discussed whether aphids may be cacao pollinators, but Billes (1941) found only evidence that they are not involved in cacao pollination. In our study, aphids were found on the outer part of the flowers (mainly on the pedicels) and thus we consider that aphids do not pollinate cacao flowers. Thrips on the other hand, could contribute to pollinate cacao flowers, as they were found on the petal pouch and inside the flowers. Some authors already considered thrips as pollinator agents (Billes 1941; Entwistle 1972; Soria et al. 1980), although they transport very few pollen grains.

In conclusion, although we were unable to identify the actual pollinators of the cacao plants in our study systems, we documented a wide range of flower visitors as well as considerable differences between the visitors of wild and cultivated cacao. Ceratopogonid midges were so infrequent that it is unlikely that they were the main or even sole pollinators in the study system. The assemblage of floral visitors of a plant species consists of a variety of insects that vary in time and space (Herrera 1988; Cane and Payne 1993; Kandori 2002; Klein et al. 2008) and we thus cannot exclude the possibility that the low abundances of ceratopogonid midges on cacao flowers were only a temporal phenomenon. However, our study was conducted at the peak flowering period of both wild and cultivated cacao trees in the study region (C. Chumacero de Schawe et al. unpublished data) so that the pollinator assemblages observed by us presumably correspond to those responsible for a large part of the pollination events in cacao in the study region. In this respect, among the insect groups other than Ceratopogonidae, there are a number of potential additional pollinators, including small Diptera and Hymenoptera that may carry pollen when they feed on flower tissue or suck nectar. The size of the visiting insect is known to be important for efficient pollination of cacao flowers and Soria et al. (1980) considered medium sized midges (wing size: 0.9 mm length–0.3 mm width) as the most effective pollinators. Most of the Diptera and Hymenoptera found by us fall in this size range and may thus fit the bill as pollinators of cacao in the study region.