Abundance of Bemisia tabaci Gennadius (Hemiptera: Aleyrodidae) and its parasitoids on vegetables and cassava plants in Burkina Faso (West Africa)

Abstract The whitefly Bemisia tabaci is a pest of many agricultural and ornamental crops worldwide and particularly in Africa. It is a complex of cryptic species, which is extremely polyphagous with hundreds of host plants identified around the world. Previous surveys in western Africa indicated the presence of two biotypes of the invasive MED species (MED‐Q1 and MED‐Q3) living in sympatry with the African species SSA and ASL. This situation constitutes one of the rare cases of local coexistence of various genetic entities within the B. tabaci complex. In order to study the dynamics of the distribution and abundance of genetic entities within this community and to identify potential factors that could contribute to coexistence, we sampled B. tabaci populations in Burkina Faso in 2015 and 2016 on various plants, and also their parasitoids. All four genetic entities were still recorded, indicating no exclusion of local species by the MED species. While B. tabaci individuals were found on 55 plant species belonging to eighteen (18) families showing the high polyphagy of this pest, some species/biotypes exhibited higher specificity. Two parasitoid species (Eretmocerus mundus and Encarsia vandrieschei) were also recorded with E. mundus being predominant in most localities and on most plants. Our data indicated that whitefly abundance, diversity, and rate of parasitism varied according to areas, plants, and years, but that parasitism rate was globally highly correlated with whitefly abundance suggesting density dependence. Our results also suggest dynamic variation in the local diversity of B. tabaci species/biotypes from 1 year to the other, specifically with MED‐Q1 and ASL species. This work provides relevant information on the nature of plant–B. tabaci‐parasitoid interactions in West Africa and identifies that coexistence might be stabilized by niche differentiation for some genetic entities. However, MED‐Q1 and ASL show extensive niche overlap, which could ultimately lead to competitive exclusion.


| INTRODUC TI ON
Vegetable plants are cultivated in many places for their socioeconomic and nutritional importance. In Africa, the main vegetable plants belong to families of Asteraceae, Compositae, Cruciferae, Cucurbitaceae, Euphorbiaceae, Fabaceae, Leguminosae, Malvaceae, and Solanaceae. In sub-Saharan Africa, cassava (Manihot esculenta Crantz, Euphorbiaceae) is also a major food staple crop (AVRDC, 2004) due to its high caloric content. Unfortunately, the production of cassava and vegetables is currently constrained by viral diseases and pest attacks, especially due to the whitefly Bemisia tabaci (Gennadius). This insect affects agricultural crops as well as ornamentals. More than 600 species of plants can be attacked by this pest in both field and glasshouse settings (Gelman, Gerling, Blackburn, & Hu, 2005). It causes damage directly by feeding on sap and, to an even greater extent, indirectly by transmitting several hundreds of plant viruses (Begomovirus, Crinivirus, Ipomovirus, Torradovirus), leading to high yield losses (Brown & Czosnek, 2002;Jones, 2003;Lapidot & Polston, 2010).
In order to understand outbreaks and better control B. tabaci populations, a key element to keep in mind is the biodiversity encountered in this species complex. Indeed, phylogenetic studies based on the mitochondrial gene COI revealed that B. tabaci is a complex of at least 41 cryptic species that belong to 11 major genetic groups (Ahmed, De Barro, Ren, Greeff, & Qiu, 2013;Bing, Ruan, Rao, Wang, & Liu, 2013;De Barro, Liu, Boykin, & Dinsdale, 2011;Dinsdale, Cook, Riginos, Buckley, & De Barro, 2010;Hu et al., 2011;Liu, Colvin, & De Barro, 2012;Sun, Xu, Luan, & Liu, 2011;Wang, Sun, Qiu, & Liu, 2011). Among these putative species, the Middle East-Asia Minor 1 (MEAM1), commonly known as biotypes B and B2, and the Mediterranean (MED), which regroups the biotypes MED-Q, MED-J, and MED-L, are the most invasive ones and predominate in many areas (Dinsdale et al., 2010). According to De Barro et al. (2011), ASL also belongs to the MED species. However, population genetic analyses support that ASL (Africa Silver Leafing) is a distinct species (Mouton et al., 2015), and thus, it will be considered as such in this study. Within the MED-Q biotype, genetic diversity has been observed leading to the recognition of three mtCOI haplotypes, MED-Q1, MED-Q2, and MED-Q3 (Chu et al., 2012;Gueguen et al., 2010).
Five putative species of B. tabaci, namely the sub-Saharan African SSA1 to SSA5 , have also attracted particular attention in the last 10 years because they have been recognized as important pest vectors of several key African cassava viruses inducing severe yield losses and food scarcity, such as the Cassava mosaic diseases (CMD) and the Cassava brown streak disease (CBSV; Legg et al., 2011). These species, within which biotypes can be distinguished, show diverse behaviors related to host plant preference, oviposition, ecological adaptation, and capacity of virus dissemination (De Barro, Trueman, & Frohlich, 2005;Perring, 2001). They also differ in the composition of the community of endosymbiotic bacteria they harbor (De Barro et al., 2011;Dinsdale et al., 2010;Gnankiné, Mouton, Henri et al., 2013;Gueguen et al., 2010).
A remarkable situation has however been described recently in Burkina Faso, where the local ASL species was found in sympatry with two biotypes belonging to the MED species, MED-Q1 and MED-Q3, a biotype that has yet only been found in Burkina Faso (Gnankiné, Ketoh, & Martin, 2013;Mouton et al., 2015). Although data about their host range remain limited, MED-Q1 and ASL seem much more generalists than MED-Q3 that appears mostly restricted to Lantana camara , where neither MED-Q1 nor ASL was found. In addition, the degree of resistance to insecticides also varies between them (Gnankiné, Mouton, Savadogo et al., 2013;Houndété et al., 2010). While still limited, these different elements could play a role in favoring or, on the opposite, in limiting coexistence.
In order to follow the dynamics of the B. tabaci community present in Burkina Faso and to identify potential factors involved in exclusion or coexistence patterns, we extended our sampling to two new years and encompassing a larger diversity of sites and plants to define more precisely the realized niche of the different genetic entities. We monitored not only presence, but also abundance, in order to assess potential competition among individuals. Because herbivores communities may also be affected by top-down effects, we also monitored the presence of parasitoids and the rate of parasitism.
Overall, these new data provide a sound basis to determine the stability of the B. tabaci community at the spatial and temporal scales.

| Sampled localities
This study was conducted in Burkina Faso, located in the west part of the African continent ( Figure 1). The climate is tropical with two seasons: the dry (from October to April) and the rainy (from May to September) seasons. This study was conducted during the dry season, the main season for vegetable production, from Identification of the insect larvae was carried out using the taxonomic keys of Martin (1987) and Martin, Mifsud, and Rapisarda (2000). The plant species were identified according to the key of Thiombiano (2012) and Akoégninou, Vander Burg, Vander, and Maessen (2006).  Legg, Ndunguru, and Thresh (2004). For Lantana Camara, the design sampling was different and thirty plants were selected randomly on a surface of about the quarter of hectare. As adults feed and oviposit preferentially on immature leaves, whitefly abundance was determined by the average number of adults on the five youngest apical leaves of each plant. Each leaf was held by the petiole and gently inverted in order to count the number of adults present on the lower surface (Avidov & Harpaz, 1969;Fargette, 1985). These adults were collected and preserved in 1.5-ml tubes containing 70% ethanol.

| Parasitism rate
The rate of parasitism was evaluated on the same plants by dividing the number of B. tabaci nymphs that were parasitized by the total number of nymphs present on the five leaves. Parasitized nymphs can be identified by eyes through their color, black or dark yellow.
The leaves with parasitized whitefly nymphs or pupae were collected, put into Petri dishes, and brought back to the laboratory. On the same plants, adult parasitoids of B. tabaci were also captured and Biotype/species

| Determination of whitefly species/biotypes by PCR-RFLP
The species/biotype of 630 adult whiteflies was determined using a PCR-RFLP method as described in Henri, Terraz, Gnankiné, Fleury, and Mouton (2013). This method proved to allow the identification of all the species/biotypes of B. tabaci present in western Africa. Briefly, DNA was extracted from each individual using the NucleoSpin Tissue kit (MACHEREY-NAGEL). Elution was performed in 100 μl of BE buffer.
Extracts were then stored at −20°C until use. The primers C1-J-2195 and L2-N-3014 were used to amplify a fragment of the mtCOI gene of B. tabaci with an approximate size of 800 bp (Simon et al., 1994). PCR amplifications were performed in a final volume of 25 μl containing 200 mmol/L dNTPs, 200 nmol/L of each primer, 1 mmol/L of MgCl 2, 0.5 U of DreamTaq, and 2 μl of DNA template. The cycling profile consisted of an initial denaturing phase at 94°C for 2 min, followed by 34 cycles consisting of 94°C for 30 s (denaturing), 58°C for 30 s (annealing), and 72°C for 1 min (extension), followed by a final extension phase at 72°C for 5 min. PCR products were digested with XapI and BfmI (10 UI) at 37°C for 3 hr, separated by electrophoresis on a 1.5% agarose gel at 100 V for 1 hr, and visualized by ethidium bromide staining.

| Statistical analyses
Statistical analyses were performed with the R software version 3.2.2 (http://www.r-project.org/). The effects of the locality, the host plant species, and the year on the abundance of whiteflies and on the rate of parasitism were tested globally by ANOVA and also using Tukey's pairwise mean comparison tests. The distribution of species/biotype on host plants and sites was tested using Fisher's exact tests. A linear regression was conducted to test the relationship between the rate of parasitism and the total number of nymphs, and between the whitefly abundance in 2015 and that in 2016.

| Host plants of B. tabaci
Our field surveys showed that the diversity of host plants of B. tabaci in Burkina Faso is high. In total, B. tabaci was found on 55 plant species belonging to 18 families were recorded (Table 1). Ten of the 55 species were ornamental, seventeen species were vegetables, and nineteen, six, and two species were weeds, strictly economic crops as defined by Li et al. (2011), such as cassava, tobacco, potato, and fruits, respectively.

| Whitefly abundance
The mean whitefly abundance varied according to both the locality and the year, with a high interaction between these two factors (F = 66.97, df = 8, p value <2.10 −16 ; Figure 2 Figure S1A). On the other hand, a significant linear correlation was detected for host plants (r 2 = .42, p = .01). However, this correlation was driven by two host plants, L. camara and I. batatas, that showed a very high abundance for the 2 years. When these two points were removed, the relationships became nonsignificant (r 2 = .005, p = .85; Figure S1B).
In conclusion, predicting abundance of B. tabaci is very difficult as it is highly variable across years, sites, and host plants.

| Geographic distribution
Among the 630 individuals sampled, three species were identified: ASL, the basal group SSA (sub-Saharan African), and MED ( Figure 4).
Within the MED species, two biotypes were found, MED-Q1 and MED-Q3. Our results suggest that species and biotypes are not randomly distributed between sites (Fisher's exact test, p = .0005). The MED-Q1 biotype was found in all the localities (except one, Tiébélé in 2015) where it predominated (Figure 1). ASL was found in several areas, but at lower frequencies (Table S1). MED-Q3 and SSA2 were only found in a restricted number of localities, which could be explained by their host specialization. The frequency and the composition of the species/biotypes varied significantly from 2015 to 2016, but MED-Q1 remained the most represented these two years (Fisher's exact test, p = 2.08.10 −11 ). In some places, as in Lilbouré and Tiébélé, the proportion of MED-Q1 increased greatly between 2015 and 2016 from 10% to 90% and from 0% to 100%, respectively (Table S1).

| Distribution on host plants
A significant relationship was found between the plant and the species/biotypes distribution (Fisher's exact test, p = .0005). Globally, most of the vegetables were infested by MED-Q1 which was sometimes associated with ASL, MED-Q3, and SSA2 ( Figure 5). MED-Q3 was the only biotype detected on Lantana camara. Interestingly, this

| Larval parasitism of B. tabaci
Only two species of B. tabaci parasitoids were found, which are both hymenopteran wasps from the Aphelinidae family. One is Eretmocerus mundus, while the other, based on mtCOI sequences, is related to Encarsia vandrieschei (92% identity). Encarsia specimens were only found in one locality, Lilbouré. On the contrary, E. mundus was recorded in all the sites except in this locality. The rates of parasitism varied according to both the areas and years with a significant interaction (F = 48.53, df = 8, p value <2.10 −16 ; Figure 6). In 2015, the rate of larval parasitism was higher in Tanghin and Lilbouré with a mean of 32.28% and 25.68% per plant, respectively, and the lowest was recorded in Boulbi and Tiébélé (4.17% and 5.49% per plant). In 2016, Koubri, Tanghin, and Tiébélé areas exhibited the highest rates of larval parasitism (18.19%, 18.58%, and 37.37%), and the lowest rate was recorded in Boulbi, Lilbouré, and Loumbila areas (6.15%, 7.67%, and 8.88% per plant). In Tiébélé, this rate increased greatly between 2015 and 2016: from 5.49% to 37.37% per plant.
There is also a significant interaction between the host plant and the year for the rate of parasitism (F = 35.7, df = 11, p value <2.10 −16 ), even though, for the 2 years, the highest rate was observed on L. camara (72.52% in 2015, 76.13% in 2016). I. batatas presents one of the lowest rates in 2015, 4.60%, but a rate close to the rate observed on L. camara in 2016, 59.10% (Figure 7). continuous increase in the B. tabaci resistance to chemical products that has largely reduced the effectiveness of most available pesticides in Burkina Faso especially on cotton and vegetables plants (Gnankiné, Mouton, Savadogo et al., 2013).
As previous surveys in Burkina Faso did not include cassava, it is probable that SSA2 was already present at that time.      was recorded as those observed 12 years ago by Gnankiné (2005) and Otoidobiga, Vincent, and Stewart (2004). This confirms that parasitoids parasitize B. tabaci in a density-dependent manner. This phenomenon could be explained by the production of allomones by plants. Indeed, it is known that the production of allomones is an indirect defense strategy of the plant that can attract parasitoids and predators (Mumm & Dicke, 2010). Most herbivore-induced plant volatiles can be classified as terpenoids, green leaf volatiles, phenylpropanoids, and sulfur-or nitrogen-containing compounds.
Increased release of allomones by plants due to a higher B. tabaci abundance may thus attract more parasitoids and explain why the parasitism rate depends on whitefly abundance. It is however not clear how this phenomenon may interact with the local dynamics of B. tabaci community.

| CON CLUS ION
The present work has led to a better understanding of the distri- between the 2 years highlight that abundance of B. tabaci is highly fluctuant, which will probably impose regular adjustments in local control strategies.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N
OG, RR, SFD, and FT contributed to the design of the study. VF, HH, and LM critically contributed to the implementation of the study.
OG, RR, SFD, FT, HH, and LM conducted field evaluations and laboratory research. RR drafted the manuscript and OG, VF, HH, and LM revised it. All authors read and approved the final manuscript.