Cocoa production is crucial to the economies and rural populations of several countries in West and Central Africa. Côte d’Ivoire alone produces over 40% of the world's cocoa, making it the world's leading producer (ICCO 2023). However, production in this country is currently threatened by an expanding lethal disease, the Cocoa Swollen Shoot Virus (CSSV). The virus, from the family Caulimoviridae and genus Badnavirus (Muller and Sackey 2005), causes different growth disorders and can be detected through specific symptoms, including red banding on leaf veins, leaf chlorosis, swollen shoots and roots and abnormally small and round pods (Oro et al. 2012, Gyamera et al. 2023). Damage to cocoa leads to a quick decay of farms, with the yield dropping from the first year of infection and, finally, the death of trees within a few years, depending on the virus virulence and cocoa growing conditions (Gyamera et al. 2023). Statistics about the disease outbreaks are scarce in Côte d’Ivoire. A survey conducted at a national scale revealed that 19.5% of prospected cocoa farms (n ≈ 440,000 farms) were CSSV-infected in 2016 (Aka et al. 2020). Today, the percentage of infected farms is probably much higher.

CSSV is transmitted to cacao by mealybugs (Hemiptera, Pseudococcidae), one of the most important families of scale insects (Coccomorpha). The vectoring of CSSV by mealybugs is a non-circulative semi-persistent transmission, which means that the virus is located on the stylets and that a mealybug remains infectious for a relatively short period of two days (Roivainen 1976). At least 64 mealybug species live on the cocoa tree, Theobroma cacao L. (Malvaceae) in tropical regions, amongst which 22 have been reported in West Africa (García Morales et al. 2016). Half of these species are known to transmit CSSV to T. cacao in West Africa (Roivainen 1976). In Côte d’Ivoire, 12 species have been reported on T. cacao (García Morales et al. 2016), amongst which seven are known as CSSV vectors (Roivainen 1976), namely Dysmicoccus brevipes (Cockerell), Ferrisia virgata (Cockerell), Formicococcus njalensis (Laing), Planococcus citri (Risso), Planococcus kenyae (Le Pelley), Phenacoccus hargreavesi (Laing) and Pseudococcus longispinus (Targioni Tozzetti). Fo. njalensis and Pl. citri are by far the most common species on cocoa in Côte d’Ivoire (Obodji et al. 2015, N'Guessan et al. 2019, N’guettia et al. 2021).

Around 70 ant species have been found to interact with CSSV vector mealybugs on cocoa farms in Ghana, most of which belong to genera Camponotus, Crematogaster, Oecophylla, Pheidole and Tetramorium (Strickland 1951a, Campbell 1975). Tending ants protect them from rain or natural enemies by enclosing them in carton or soil tents. Ants also help mealybugs by consuming the honeydew they produce which helps colonies by limiting fungal growth (Strickland 1951b, Dufour 1991). They may transport mealybugs between their mandibles, but only on short distances (Strickland 1951a). These ants, therefore, contribute to the establishment of large colonies of mealybugs on cocoa trees and possibly to CSSV outbreaks. Depending on the mealybug species, the mealybug-ant associations can vary from facultative to strict. Species with a facultative association can project honeydew away from the colony by contracting the rectum (Way 1963). On the other hand, association with ants is strict for some species due to the quality of the honeydew produced and the inability to project it away from the colony (Way 1963, Jahn and Beardsley 2000, Jahn et al. 2003). Fo. njalensis, the most abundant species in West African cocoa plantations, belongs to the latter category. In contrast, Pl. citri and other species have a facultative association with the tending ants (Strickland 1951b).

Worldwide, mealybugs have a wide range of natural enemies, including parasitoid wasps and predators (Franco et al. 2009). The highly polyphagous and cosmopolitan Pl. citri alone is associated with about 140 chalcidoid species, some of them parasitoids and others hyperparasitoids, mostly belonging to the family Encyrtidae (Hymenoptera, Chalcidoidea), but also Aphelinidae, Eriaporidae, Eulophidae, Pteromalidae and Signiphoridae (Noyes 2019). Fo. njalensis has a more restricted distribution in Central and West Africa and is associated with 40 parasitoid species, mostly in the family Encyrtidae (Noyes 2019). Data on parasitoids of cocoa mealybugs are scarce, with many unidentified species, mainly from studies conducted in Ghana (Bigger 2012). Strickland (1951a) reported 13 encyrtid species on Fo. njalensis, mainly in the genera Aenasius and Anagyrus. More recently, Ackonor and Mordjifa (1999) and Ackonor (2002) identified seven species of encyrtids on both Fo. njalensis and Pl. citri and reported seven other unidentified wasps and a parasitoid fly of the genus Cryptochetum (Diptera, Cryptochetidae). For Côte d’Ivoire, Risbec (1949) and Risbec (1955) identified eight species of encyrtids on Fo. njalensis and three more species of the families Megaspilidae, Platygastridae and Signiphoridae. Mealybugs are also the prey of various generalist predators, including ladybird beetles (Coleoptera, Coccinellidae), lacewings (Neuroptera, Chrysopidae), gall midges (Diptera, Cecidomyiidae), and various spiders (Strickland 1951a, Dufour 1991, Ackonor 2002).

Effectively managing CSSV requires a thorough knowledge of the insects transmitting the virus to cocoa trees. In addition, it is crucial to identify species capable of naturally regulating vector populations as these could be good candidates for biological control. Identification of these insects, typically based on morphological characteristics, presents challenges, especially for non-specialists (Pacheco da Silva et al. 2014, Puig et al. 2021b). Pseudococcidae are highly diversified and different regions harbour different species, sometimes very similar morphologically, making taxonomic identification difficult and posing a major obstacle to the research and management of these species (Puig et al. 2021b). Species identification of scale insects is almost entirely based on the morphological characteristics found in the adult female; adult males and immature stages can, in most cases, not be identified to species and often renders inconclusive results (Pacheco da Silva et al. 2014). In addition, morphological identification usually requires that the specimens be processed, cleared of their internal contents and slide-mounted which is time-consuming and requires special tools and training. To overcome these identification challenges, the DNA Barcoding approach is a powerful alternative and complementary approach that has been applied successfully to mealybugs and their natural enemies (Park et al. 2011, Abd-Rabou et al. 2012, Pacheco da Silva et al. 2014, Malausa et al. 2016, Puig et al. 2021b, Puig et al. 2021a, Pacheco da Silva et al. 2021, Nurbaya et al. 2022). However, this method relies on the availability of curated barcode databases and cross-validation by taxonomic experts to ensure the consistency of species concepts and molecular data.

If DNA barcode data have already been provided for cocoa mealybugs (Wetten et al. 2015, Obok et al. 2021, Puig et al. 2021b), our study is the first to provide a complete DNA barcode database for the entomofauna, including ants and natural enemies, associated with the CSSV disease in Côte d'Ivoire and, more generally, in West Africa. Based on intensive sampling in cocoa fields in Côte d’Ivoire, barcode sequences for 131 species of mealybugs, ants and natural enemies and 190 unique haplotype sequences are reported. This dataset paves the way for the use of high-throughput metabarcoding approaches to identify species in this complex community.

Sample collection

This study was conducted in Côte d’Ivoire, in cocoa plantations of a network implemented within the Cocoa4Future project (see Acknowledgements). The plantation network comprises 15 sites, geographically covering the entire production area of the country. Hence, our study includes the vast majority of environmental, agronomical and historical cocoa growing conditions (Kouassi et al. 2023); (Table 1).

Table 1.

Download as

CSV

XLSX

Location of sampling sites.

Sample site name	Longitude (X)	Latitude (Y)	Mealybug and associated ant sampling	Natural enemy sampling
Aboisso	-3.32031	5.47451	x	x
Azaguié	-4.48282	5.57077	x	x
Soubré	-6.49383	5.74143	x
Guiglo	-7.87404	6.53766	x	x
Man	-7.24257	7.30263	x	x
Bonon	-6.04698	6.91921	x	x
Adzopé	-3.72629	5.97444	x	x
Grand-Bereby	-7.00418	4.84994	x	x
Agnibilekro	-3.20366	7.12785	x
Guibéroua	-6.21471	6.21354	x	x
Meagui	-6.81134	5.59470	x	x
San Pedro	-6.46408	4.89055	x	x
Guéyo	-6.16756	5.80025	x
Blé	-5.17393	5.95151	x
Fresco	-5.57361	5.13914	x	x

Mealybugs and tending ants were sampled in five plantations in each of the 15 sites, i.e. 75 plantations in total (Fig. 1A and B). One hundred cocoa trees were thoroughly searched in each plantation, moving across the plot following a zigzag pattern (Sether et al. 2010) to cover the plantation's structural diversity.

Figure 1.  

Cocoa plantations in Côte d’Ivoire and mealybug colonies with examples of associated ants and natural enemies. A Healthy cocoa plantation; B Plantation infected by the cocoa swollen shoot virus; C Colony of Dysmicoccus neobrevipes associated with ants of the genus Camponotus; D Colony of Formicococcus njalensis associated with Oecophylla longinoda; E Colony of parasitised mealybugs showing mummies and emerged parasitoids (arrows); F Cecidomyiid larva feeding on a mealybug (arrow).

The mealybugs were searched for on cocoa trees from the ground up to a height of 2 metres, focusing on green parts, where mealybug colonies are usually found due to the easy access to the cambium, i.e. suckers, cracks in the bark of trunks, flower cushions, flowers, pods, leaves, buds and shoots on branches. Cocoa canopy was also prospected by collecting branches using a pruning saw. However, this method was not systematically applied to all cocoa trees because it is destructive and not always accepted by famers. Nevertheless, it allowed us to sample 214 mealybug colonies from cocoa canopy. When detected on trees, mealybug colonies (Fig. 1C and D) were recorded and approximately ten individuals were collected from each colony along with the tending ants. Insects were collected using soft forceps and preserved in collection tubes containing 90% ethanol pending their transfer to the entomology laboratory of Centre d’Excellence Africain sur le Changement Climatique, la Biodiversité et l’Agriculture Durable/West African Science Service Centre on Climate Change and Adapted Land Use/Université Félix Houphouët Boigny (CEA-CCBAD/WASCAL/UFHB), Bingerville. A total of 1,650 mealybug colonies were collected, including the tending ants when present. Samples were sorted, identified to morpho-species level and stored in 96% ethanol for further morphological and molecular identification.

Sampling of mealybug natural enemies was conducted in 10 cocoa plantations in 11 out of the 15 sites, i.e. 110 plantations in total. For each plantation, natural enemies were obtained through two consecutive methods. First, three cocoa pods with mealybug colonies that covered at least one-third of the pod surface (Fig. 1C) were carefully collected with secateurs. The pods were stored individually in two-litre plastic bottles cut in half and closed with muslin held in place with a rubber band to allow aeration. Second, ten other mealybug colonies were gently collected from trees using a camel-hair brush and introduced into 4.5 ml (75 x 12 mm) dry collection tubes (without ethanol), closed with muslin. In all, 1,430 mealybug colonies (three infested pods plus ten mealybug colonies in tubes x 110 plantations) were collected. These samples were stored in the laboratory at room temperature and the emergence of the natural enemies was monitored every five days for 20 days. During observation, insects emerging from the tubes and pods were captured, classified into morpho-species and preserved in microtubes containing 96% ethanol. For each mealybug colony, the tending ants were also separately collected in tubes containing 90% alcohol. The species occurrence was assessed from the field and laboratory records for mealybug colonies, tending ants and natural enemies according to the following categories: (-) rare occurrence (= 1-10 occurrences), (+) moderate occurrence (= 10-20 occurrences), (++) regular occurrence (= 20-40 occurrences) and (+++) very frequent occurrence (= 40 and over occurrences).

Morphological identification

DNA extraction, amplification and sequencing

Sequence analysis

Data resources

A total of 305 COI sequences, with coding for a 385 bp fragment, were generated from 17 species of mealybugs (Pseudococcidae) collected from cocoa plantations in Côte d’Ivoire. Pseudococcidae specimens associated with these sequences were identified morphologically to species level (Table 2). The most abundant mealybug species on cocoa were Formicococcus njalensis (Laing), Planococcus citri (Risso) and Dysmicoccus neobrevipes Beardsley. Intraspecific distances ranged from 0% to 6.57%, with distances higher than 2% within Fo. njalensis (4.57%), Pl. citri (2.93%), Pseudococcus longispinus (Targioni Tozzetti) (3.73%) and Pseudococcus occiduus De Lotto (6.57%; Table 2). Inter-specific distances were consistent with species concepts based on morphology, with divergences ranging from 4% to 19% (Suppl. materials 1, 2).

For tending ants, 211 COI sequences, coding for a 418 bp fragment, were generated from 54 species collected with mealybug colonies in cocoa plantations. Amongst ant species, 24 were identified at the species level and 30 at the genus level (Table 3). The most abundant ant species tending cocoa mealybugs were Camponotus acvapimensis Mayr, Crematogaster africana Mayr, Lepisiota cocazela Santschi, Paratrechina longicornis (Latreille), Pheidole megacephala (Fabricius) and two other unidentified species of Pheidole. Intraspecific distances ranged from 0% to 6.29%, with distances higher than 2% within Crematogaster sp.1 (2.69%), Ph. megacephala (2.70%), Pheidole sp.1 (3.96%), Pheidole punctulata Mayr (4.98%) and Pa. longicornis (6.29%). A large interspecific divergence (ranging from 6.29% to 36%) was observed amongst major neighbouring species groups (Suppl. materials 3, 4).

Natural enemies of mealybugs on cocoa trees include hymenopteran parasitoids, predatory beetles and predatory dipterans (Table 4). In total, 143 COI sequences of the 418 bp fragment were generated from 65 species of natural enemies. The hymenopteran parasitoids include 22 species of primary parasitoids and eight hyperparasitoid species (Table 4). The most abundant species of parasitoids were Aenasius abengouroui (Risbec), Anagyrus kivuensis Compere, Coccidoctonus pseudococci (Risbec), Leptomastix dactylopii Howard and one hyperparasitoid, Cheiloneurus carinatus (Compere). Substantial interspecific divergences, ranging from 9.57% to 22.15%, were observed between neighbouring species groups (Suppl. materials 5, 6). Genetic distances also revealed a significant intraspecific variation, ranging up to 8.22%, with higher divergences observed within two primary parasitoids: Anagyrus aff. subproximus (8.22%) and L. dactylopii (5.85%). Due to sequencing failure, no COI sequences were obtained for Gyranusoidea sp. and Coccidoxenoides sp.

The predators of cocoa mealybugs are represented by 11 species of ladybugs beetles (Coccinellidae) spread across four genera, Hyperaspis, Platynaspis, Scymnus and Nephus and seven species of gall midges (Diptera), from the family Cecidomyiidae, whose genera and species could not be determined due to the lack of specialists for these groups and the poor condition of specimens after DNA extraction. Although the barcode sequences allowed for the differentiation of seven species of gall midges, these sequences could not be matched to any sequences previously published in online molecular databases. For this group, we thus only used morpho-species concepts. The most abundant predators were two unidentified species of Nephus and Scymnus and two unidentified species of Cecidomyiidae (Table 4). For coccinellids, all intraspecific variations were ≤ 2%. The morphospecies Cecidomyiidae sp2 shows a higher intraspecific divergence of 2.69%. Interspecific divergences amongst coccinellids and cecidomyiids ranged from 11% to 28% (Suppl. materials 7, 8, 9, 10). Four morpho-species of Lepidoptera from the genera Niditinea Petersen (Tineidae), Spalgis Moore (Lycaenidae) and Syringoseca Meyrick (Oecophoridae) and four morpho-species of spiders from the genera Psammitis Menge (Thomisidea), Myrmarachne De Geer (Salticidae) and Theridion Walckenaer (Theridiidae) were collected from mealybug colonies (Table 4).

Some parasitoids associated with Cecidomyiidae, Coccinellidae and lepidopterans were identified: three species of the genus Aphanogmus and one of Platygaster (parasitoids of Cecidomyiidae), two species of Homalotylus (parasitoids of Coccinellidae) and one species each from the genera Antrocephalus, Apanteles and Ooencyrtus (parasitoids of lepidopterans) (Table 4). No sequence was obtained for Ooencyrtus Mayr due to a sequencing failure. Therefore, the species was identified solely based on morphological characteristics (Suppl. materials 11, 12).

This study aimed to explore the diversity of mealybugs as potential vectors of CSSV disease in Côte d’Ivoire, as well as that of their tending ants and natural enemies. The results show that the COI gene fragments enable species-level identification in this taxon-rich community. In all functional groups, however, some species showed significant intraspecific divergences suggesting possible cases of complexes of cryptic or closely-related species. Ps. occiduus, for example, with a maximum divergence of 6.57%, indicates the possible presence of cryptic species, as no morphological differences were observed between divergent lineages. In contrast, the 4.57% of divergence between specimens of Fo. njalensis was associated with a slight morphological divergence, already reported by Hall (1946) and Padi and den Hollander (1996), supporting the hypothesis that Fo. njalensis may be a complex of cryptic species. The problem of delimiting complexes of cryptic species in most cases was resolved using an integration of morphology and molecular information, which was especially true in some genera, such as the parasitic wasp Anagyrus and the mealybug Planococcus. After analysing their DNA barcodes, we revisited their morphological identification and the careful examination allowed the observation of subtle morphological differences and the attribution of correct species names to all individuals. In the genus Planococcus, for instance, the complex Pl. citri/kenyae/minor is formed by three cryptic species (Dufour (1991) observed this for citri and kenyae) and are easily separated by their barcodes. Detailed morphological examination of voucher specimens revealed useful diagnostic characters, based on criteria defined by Cox (1989) for Pl. citri/minor. In tending ants and for most parasitoids, the high intraspecific genetic distance could not be correlated to morphological divergence. Slight morphological divergences were observed between individuals bearing distant haplotypes in Clausenia, Zaplatycerus and Acerophagus only. The clarification of the taxonomic status of these lineages is beyond the scope of this study. These cases deserve further morphological investigation as well as the complementary use of longer fragments of COI and nuclear genes to verify the validity of the species concepts used here.

This study expands the previous inventories of cocoa-associated fauna of mealybugs and their tending ants and natural enemies in Côte d’Ivoire and in West Africa in general. Our findings are comparable to those obtained in Togo, where Dufour (1991) recorded ten species of mealybugs on cocoa and in Ghana, where Strickland (1947) and Campbell (1983) reported ten species and Strickland (1951a) identified eight species plus nine unnamed species of mealybug in cocoa. However, mealybug communities significantly differ amongst studies. Seven species, namely Paraputo anomalus (Newstead), Delococcus tafoensis (Strickland), Paraputo loranthi (Matile-Ferraro), Formicococcus celtis (Strickland), Tylococcus westwoodi Strickland, Nipaecoccus masakensis (James) and Pseudococcus calceolariae (Maskell) were reported from Togo and/or Ghana, but not in the present study. On the other hand, we identified three species previously unreported on cacao in Africa (García Morales et al. 2016): D. neobrevipes, Fe. dasylirii and Ph. solenopsis. In Côte d'Ivoire, recent surveys conducted by Obodji et al. (2015), N'Guessan et al. (2019) and N’guettia et al. (2021) recorded eight, nine and eleven species of mealybugs on cacao trees, respectively. Amongst these, Pseudococcus viburni (Signoret), identified by N’guettia et al. (2021), is the only species absent from the present report. Through the combined use of morphological identification and molecular analysis, based on DNA barcoding, eight species of mealybugs were identified from cocoa in Côte d’Ivoire for the first time, namely D. neobrevipes, Pa. marginatus, Ph. solenopsis, Pl. minor, Ps. concavocerarii, Ps. occiduus, M. ugandae and Fe. dasylirii. Pa. marginatus, Ph. solenopsis and D. neobrevipes are significant pests of papaya, cotton and pineapple, respectively (Germain et al. 2014, Dey et al. 2018, Rostami et al. 2024). These three species are polyphagous and can, therefore, live on various host plants (Joshi et al. 2010, Germain et al. 2014, Rostami et al. 2024). Most of the mealybug species identified in this study are amongst the 14 species known to be vectors of CSSV in cocoa (Entwistle and Longworth 1963, Roivainen 1976, Dufour 1991). However, six species, namely D. neobrevipes, Pa. marginatus, Ph. solenopsis, Pl. minor, Ps. occiduus and Fe. dasylirii have not been reported as CSSV vectors. Nevertheless, their presence in cocoa plantations could increase the risk of CSSV spread, as mealybugs are globally considered as potential plant virus vectors (Entwistle and Longworth 1963). Additionally, Fe. dasylirii and Fe. virgata are in the same genus and are morphologically similar, with morphological and molecular variations within some populations (Kaydan and Gullan 2012), suggesting that Fe. dasylirii could have similar CSSV transmission capabilities as Fe. virgata. The same applies to D. neobrevipes, the same genus and a species morphologically and genetically similar to D. brevipes, which has been identified as a vector of CSSV on cocoa (Roivainen 1976). These two Dysmicoccus species are recognised as good vectors of the Pineapple Mealybug Wilt-associated Virus (PMWaV), an Ampelovirus from the family Closteroviridae, transmitted in a semi-persistent manner, similar to the mode of transmission of CSSV (Jahn et al. 2003, Dey et al. 2018). Pa. marginatus is an invasive mealybug known to attack at least 54 plant families, particularly papaya, causing significant economic losses. Originally from the Americas, it has rapidly spread to Asia and Africa (Germain et al. 2014, Rostami et al. 2024). Pa. marginatus, along with M. hirsutus, are vectors of the Mulberry Mosaic Virus (MMV) in mulberries, a tospovirus from the family Bunyaviridae, transmitted persistently in a non-propagative manner (Naik et al. 2013). However, while M. ugandae has been confirmed as a vector of CSSV in cacao (Roivainen 1976, N’Guessan et al. 2019), there is currently no direct evidence demonstrating the ability of M. hirsutus to transmit this virus. Nevertheless, M. hirsutus may possess similar vectoring capacities to M. ugandae, given their morphological similarity and shared taxonomic classification within the same genus. Consequently, the presence of Pa. marginatus on cacao trees represents a potential risk, as this species may adapt, become a major cacao pest and possibly transmit diseases, including CSSV. Ph. solenopsis, known as the cotton mealybug, is a notorious polyphagous pest capable of infesting many other plants (Joshi et al. 2010, Germain et al. 2014). However, no information about its ability to transmit viruses to plants is available. Similarly, the status of Pseudococcus occiduus as a virus vector, particularly for CSSV, has never been demonstrated, as this study is the first to report the presence of this mealybug species on cacao trees in West Africa. The only previous mention of Ps. occiduus on cacao dates back to De Lotto (1964), who recorded it in Uganda. Additionally, Couturier et al. (1985) reported its presence in Côte d'Ivoire, in the dense Taï Forest in the south-western region of the country, during a mealybug survey, but not on cacao trees. Further research is needed to assess whether this species could be a vector of CSSV, as many colonies are now established in cocoa orchards in Côte d'Ivoire. In terms of occurrences, the species most frequently found in cocoa plantations in Côte d'Ivoire are Fo. njalensis, the Pl. citri/kenyae/minor group and D. neobrevipes. Except for D. neobrevipes, which is starting to become well-established in Ivorian orchards, our observations confirm those of Obodji et al. (2015), N'Guessan et al. (2019) and N’guettia et al. (2021), who reported that Fo. njalensis and Pl. citri are the dominant species in Côte d'Ivoire's cocoa orchards. These two species are also recognised as the primary vectors of several forms of CSSV in cocoa trees (Entwistle and Longworth 1963, Roivainen 1976, Dufour 1991). Their abundance in cocoa plantations significantly increases the risk of spread of the disease.

Most of the mealybug species reported in the present study were associated with ants. In total, 54 tending ant species were identified in this inventory. In Ghana, Strickland (1951b) reported around 70 ant species associated with Pseudococcidae vectors of CSSV in cocoa trees, whereas in Togo, Dufour (1991) recorded 39 ant species tending mealybugs in cocoa plantations. The tending ant community, reported in the present study, is similar to those recorded in Ghana and Togo. However, 15 additional ant species were recorded tending mealybugs in Côte d’Ivoire, namely Ca. maculatus, Ca. solon, Ca. aff. solon, Cr. solenopsides, L. cocazela and M. invidium, M. pharaonis, N. angulatus, Para. longicornis, Ph. punctulata, Pl. intermedia, Ta. lugubre, Ta. melanocephalum and Tec. aff. pallipes. Before this study, data on ants tending mealybugs on cocoa were scarce for Côte d'Ivoire. However, some information on certain species, including Camponotus sp., Oe. longinoda, Cr. africana, Crematogaster sp. and Ph. megacephala, can be found in the studies conducted by Alibert (1951), Risbec (1955) and Babacauh (1982). A study of ant diversity conducted by Yeo et al. (2011) near tropical forests and in cocoa plantations in the Oumé Region recorded 155 species belonging to 43 genera. Most of the species recorded in our study were listed by these authors. Association between ants and mealybugs is generally considered as an adaptive strategy for ants to easily access a regular food source, namely the honeydew produced by mealybugs (Dufour 1991). In return, tending ants protect the mealybugs from natural enemies and climatic threats like heavy rains by enclosing them in tents constructed with plant debris or soil (Dufour 1991). Cornwell (1955) also noted that the interaction between ants and mealybugs stimulates reproduction in the latter. In the present study, ant species that were found tending mealybugs the most frequently were Ca. acvapimensis, Cr. africana, L. cocazela, Pa. longicornis and Ph. megacephala. Oe. longinoda, Od. troglodytes and some species of Crematogaster seem to become increasingly dominant in their relationship with mealybugs (ADKK pers. obs.). Many other species have been occasionally found with mealybugs. Though less common, the sum of these associations represents a significant part of the mealybug-ant interactions and contribute to the complexity of the system.

The mealybug’s natural enemies identified in this study included 22 primary parasitoids, eight hyperparasitoids and 23 predators. The species community we have described in our study is richer than those reported in the past in Togo, Ghana and Côte d’Ivoire. For instance, Strickland (1951a) reported 13 parasitoid species (Encyrtidae) of Fo. njalensis in Ghana. More recently, Ackonor and Mordjifa (1999) and Ackonor (2002), identified seven species of parasitoids of Fo. njalensis and Pl. citri in Ghana. In Togo, Dufour (1991) also identified eight primary parasitoid species, seven secondary parasitoids and eight predators (Cecidomyiidae, Coccinellidae, Chrysopidae and Lycaenidae) of Fo. njalensis, Pl. citri/kenyae, Fe. virgata and M. ugandae. In Côte d’Ivoire, Risbec (1955) identified six primary parasitoid species and two hyperparasitoid species in colonies of Fo. njalensis. The present inventory lists nine parasitoid species of cocoa mealybugs unreported in Africa until now, namely Ac. aff. dysmicocci, Ae. advena, Aloencyrtus sp, An. kamali, An. aff. pseudococci, Cocco. pulvinariae, Cl. aff. corrugata, G. aff. tebygi and Za. aff. natalensis and one hyperparasitoid namely Pa. muscarum. In addition to these species, two primary parasitoids, Bl. insularis and Cocci. pseudococci and one hyperparasitoid, Ch. cyanonotus, have not been reported on cocoa mealybugs in Côte d’Ivoire. Some of these species like Ae. abengouroui, An. kivuensis, An. pseudococci and L. dactylopii are well-known parasitoids of mealybugs on different crops. Three of them, An. pseudococci, An. kivuensis and Le. dactylopii, in fact, are exotic species introduced to Ghana between 1951 and 1955 to control mealybug vectors of CSSV in cocoa plantations (Entwistle 1972). However, Risbec (1949) reported its presence in Côte d'Ivoire in the same year, stating that "this species, already described from the Belgian Congo, is undoubtedly widespread throughout tropical Africa". Similarly, Risbec (1949) also recorded Leptomastix longipennis Merc., which is now recognised as a junior synonym of L. dactylopii Howard. This observation, therefore, predates its presumed first introduction to Ghana. Today, these species are very common in Côte d'Ivoire, where they are the primary parasitoids of Fo. njalensis and Pl. citri. Additionally, An. kamali, a parasitoid used in Egypt to control M. hirsutus, a major pest of Hibiscus (Moore 1988), is reported here for the first time on cocoa mealybugs in West Africa, likely parasitising these same species in cocoa plantations. Prinsloo and Annecke (1979) indicate that Cocci. pseudococci is an ectoparasite of Coccodiplosis coffeae (Cecidomyiidae), a predator of mealybugs. However, it is worth noting that Cocci. pseudococci has also been observed parasitising Fo. njalensis and Pl. citri as a primary parasitoid, which is confirmed by our observations of Cocci. pseudococci emerging from mealybug mummies in the laboratory. In our study, Ch. carinatus is the dominant hyperparasitoid, which is in line with the report by Dufour (1991), who noted that Ch. carinatus primarily parasitised L. bifasciata, Ae. abengouroui and several species of Anagyrus spp. Ch. carinatus was also reported as the most abundant hyperparasitoid in Nigeria (Donald 1956) and Ghana (Ackonor and Mordjifa 1999). This study also highlights the presence in cocoa mealybug colonies of 11 species of predatory ladybugs, seven morpho-species of predatory gall midges, four species of predatory spiders, one species of predatory Lycaenidae and three other species of Lepidoptera, whose functional status has yet to be established. Although reliable species level identification could not be obtained in this study, our results are in line with findings of Dufour 1991 and Ackonor (2002), who indicated in their respective studies conducted in Ghana and Togo that the main predators of CSSV mealybugs were Scymnus kibonotensis, Scymnus sp., Platynaspis solieri, Platynaspis hingginsi, Hyperaspis quadrilla, Hyperaspis egregia, the larvae of gall midges of the species Coccodiplosis coffeae and Lepidoptera from the family Lycaenidae. Donald (1956) also reported Nephus ornatulatus as a predator of Ferrisia virgata, Phenacoccus madeirensis and Pseudococcus longispinus on cacao in Nigeria. Additionally, Majer (1975) recorded two other species on cacao by pyrethrum knockdown. Our study also reports the presence of large numbers of four species of ladybugs from the genus Nephus. Gall midges are the most frequently found predators within colonies of Fo. njalensis and Pl. citri, amongst many other mealybug species. Only the larvae are capable of preying on mealybugs at all developmental stages (Dufour 1991, Ackonor 2002).

This study provides a first curated DNA barcode database to identify mealybugs and associated arthropods involved in the transmission of CSSV disease to cocoa. Through intensive sampling, we report a total of 17 mealybug species, including eight species new for cocoa in Côte d’Ivoire, a significant diversity of ants (54 species) tending mealybugs and a notable diversity of natural enemies of cocoa mealybugs, including 14 unreported species of primary parasitoids, two hyperparasitoids, 11 species of predatory ladybirds and seven species of predatory Cecidomyiidae. The fragments of COI used in this study allow for effective species identification, even between closely-related species. In all, of the 192 haplotype sequences (beyond 2% of divergence) obtained for mealybugs, ants and their natural enemies, 151 are newly provided here and made available on GenBank. This database provides valuable references for the rapid and accurate identification of entomofauna associated with CSSV disease on cocoa in Côte d’Ivoire. It provides a solid foundation for developing integrated pest management strategies based on metabarcoding in cocoa plantations and promoting a biological control approach, based on the conservation and promotion of natural biodiversity.

We express our gratitude to all the cocoa producers involved in the Cocoa4Future project, who generously provided access to their cocoa plantations, enabling us to carry out our research activities. We especially thank the project's funders (European Union and Agence Française de Développement) for providing the necessary resources for this study. Our thanks also go to the Cocoa4Future project coordination team for their assistance with logistical support during the most challenging moments of this study. We are grateful to Audrey Weber (AGAP laboratory) for the MiSeq sequencing and the Genotoul bioinformatics platform Toulouse Midi-Pyrénées (Bioinfo Genotoul). Finally, we acknowledge all individuals who contributed directly or indirectly to this study, whether in the field for data collection, in the lab for sorting or during the final data analysis.

This study was conducted within the framework of the Cocoa4Future (C4F) project, which is funded by the European DeSIRA Initiative under grant agreement No. FOOD/2019/412-132 and by the French Development Agency. The C4F project pools a broad range of skills and expertise to meet West African cocoa production development challenges. It brings together many partners jointly striving to place people and the environment at the core of tomorrow's cocoa production.

This project (ID 2202-218) was funded through Labex AGRO ANR-10-LABX-0001-01 under the University of Montpellier I-Site framework, coordinated by Agropolis Fondation.

Cocoa4Future (C4F) project

Centre d’Excellence Africain sur le Changement Climatique, la Biodiversité et l’Agriculture Durable CEA-CCBAD/WASCAL, of Université Félix Houphouët Boigny, Abidjan, Côte d'Ivoire

Centre de Coopération Internationale en Recherche Agronomique pour le Développement (Cirad), Montpellier, France

Centre de Biologie et de Gestion des Populations (CBGP), Montpellier, France

Study design: A.D.K. Koffi, R. Babin, S-W.M. Ouali N'goran, J. Haran

Specimen sampling: A.D.K. Koffi, R. Babin, J. Haran, K.J.H. Koigny, S.K. Kpangui

Sequence analyses: A.D.K. Koffi, J. Haran, M. Galan, L. Benoit

Taxonomic expertise, results, validation: G. Delvare, C.V. Yodé, S. Chérasse, D. Ouvrard, J. Haran, R. Babin, M.S-W. Ouali N'goran, E.M. Shimbori

Writing of manuscript: A.D.K. Koffi, R. Babin, S-W.M. Ouali N'goran, J. Haran

All authors edited and commented on the manuscript.