Niche partitioning via host plants and altitude among fruit flies following the invasion of Bactrocera dorsalis

Invasions by exotic species in areas already occupied by related species may lead to extinction or niche partitioning. Bactrocera dorsalis has invaded the Comoros archipelago in 2005 where other tephritids were already present. The host ranges of fruit flies in the Comoros were studied by conducting a seven‐year survey on the three islands of the archipelago with a substantial sampling effort comprising 91 plant species from 37 families. The results showed that 45 fruit species in 16 families were infested by tephritid species. Eight fruit fly species were detected in the sampled fruits, but 78% of the individuals were identified as B. dorsalis, confirming its dominance and polyphagous status. More than ten years after its invasion, B. dorsalis has replaced Ceratitis capitata as the dominant fruit fly species on several species such as mango and citrus. The coexistence between B. dorsalis and C. capitata seems possible due to the capacity of the latter to exploit different niches, i.e. higher altitudinal ranges and a few host fruit species (among which, chillies and coffee berries) constituting a refuge niche. Information about coexistence between B. dorsalis and other tephritids is useful for the management and better estimates of invasion risks and associated biosecurity measures.


INTRODUCTION
Tephritid fruit flies (Diptera: Tephritidae) are important, worldwide pests because their larval stages consume a wide range of fruit and vegetable crops (White & Elson-Harris, 1992). The extent of losses caused by these pests depends on the tephritid species, the species of fruit or vegetable, and the location (Mwatawala et al., 2009). In the last century in Africa, tephritid damages to fleshy fruits were mainly caused by a limited number of highly polyphagous species, most of them belonging to the genus Ceratitis Macleay, such as Ceratitis capitata Wiedemann that was reported from 100 host plant species in 30 families in Africa (Copeland et al., 2002).
A number of Tephritidae are well-known for being invasive and have expanded their distribution beyond their native ranges, and many species in the genus Bactrocera have invaded areas occupied by native Ceratitis spp. (Duyck et al., 2004(Duyck et al., , 2022. Because of their wide host range and invasive potential, several Bactrocera spp. are considered a serious threat to horticultural crops (Clarke et al., 2005). In Africa, only a few indigenous Bactrocera spp. are known, and none is of great economic importance except for the olive fruit fly, Bactrocera oleae Gmelin, which is a notorious pest of cultivated olives in the Mediterranean region (Mwatawala et al., 2007;White & Elson-Harris, 1992).
In the last two decades, some species of Bactrocera have invaded many African countries and islands (Duyck et al., 2004;Ekesi et al., 2016;Moquet et al., 2021;Rasolofoarivao et al., 2021). Based on phylogenetic and biogeographic data, Zeugodacus cucurbitae Coquillett, formerly known as Bactrocera cucurbitae (De Meyer et al., 2015), is the oldest case of an Asian species being introduced into Africa ( Mwatawala et al., 2007). Bactrocera zonata Saunders has probably been in Egypt since the early or mid-twentieth century but was initially misidentified as B. pallida (Abuel-Ela et al., 1998). Two other Bactrocera spp., Bactrocera latifrons Hendel and Bactrocera dorsalis Hendel, have recently invaded Africa. B. latifrons, which is associated with hosts in the Solanaceae, was recorded in Tanzania in 2006 (Mwatawala et al., 2007). B. dorsalis, which was initially described as B. invadens (Drew et al., 2005), was first detected in Kenya in 2003 (Lux et al., 2003) and has rapidly spread across the African continent (Manrakhan et al., 2015). Based on similarities in morphology, molecular structure, and chemoecology, as well as on sexual compatibility, B. invadens was synonymized with B. dorsalis (Schutze et al., 2015).
In Africa, local studies have indicated that B. dorsalis is the dominant tephritid on a wide variety of hosts Mwatawala et al., 2006;Rwomushana et al., 2008) and has high infestation rates on mango and other commercial fruits (Vayssières et al., 2005). In tropical Asia, B. dorsalis attacks 124 host plant species in 42 families (Drew, 1989). B. dorsalis has been recorded on more than 40 cultivated and wild host plants and causes substantial damage to mango and guava fruits in Africa and Indian Ocean islands Goergen et al., 2011;Rwomushana et al., 2008;Moquet et al., 2021;Rasolofoarivao et al., 2021). In these areas, B. dorsalis has displaced several species of Ceratitis, such as C. rosa Karsch, C. quilicii De Meyer et al., C. cosyra (Walker), and C. capitata, on a number of hosts (Ekesi et al., 2009(Ekesi et al., , 2016Moquet et al., 2021).
Polyphagous Bactrocera have usually invaded areas where other polyphagous tephritids already occur (Duyck et al., 2004). Interspecific competition, via exploitative competition in fruits, interference between adults (Duyck, David, Junod, et al., 2006) is usually observed with a reduction of niche and abundance of the already present species (Charlery de la Masselière, Ravigné, et al., 2017;Moquet et al., 2021;Vargas et al., 1995). Niche partitioning, i.e. the process driving competing species into different patterns of resource use or different niches, and by which resident and introduced species may coexist (Denno et al., 1995;Reitz & Trumble, 2002) has been observed among Tephritidae to be driven by abiotic factors such as temperature and rainfall  and by host range, the main resource for which competition occurs being the host fruit (Prokopy & Roitberg, 1984). Complete exclusion after tephritid invasion seems rare but niche partitioning can be asymmetric in favour of the invasive species that the resident species are only able to use a few species, that is defined as a refuge niche (David et al., 2017). A well-documented case is the reduction in the host range of C. capitata after invasion by B. dorsalis in Hawaii where in lowlands, C. capitata persists only on coffee which is considered a refuge host (Reitz & Trumble, 2002 (De Meyer et al., 2012). In Comoros, a wide variety of fruits are important for local consumption and subsistence farming. This variety of fruits also represent a substantial food source for fruit flies (De Meyer et al., 2012) Although fruit flies are a major problem in Comoros, their host range is poorly known. Mangoes and citrus, however, were known to be highly infested by C. capitata before the invasion of B. dorsalis (Hassani et al., 2016). The knowledge of the host range and economic significance of invasive fruit fly species in Comoros are also limited.
Neoceratitis cyanescens was mentioned on Solanaceae in Mayotte and on three other islands in Comoros (Kassim & Soilihi, 2000). According to Wuster (2005), soursop fruit (Annona muricata) was attacked by such as temperature and humidity . This is also the case between B. dorsalis and C. capitata in Comoros where Hassani et al. (2016) showed that B. dorsalis was more abundant in low altitude areas, while C. capitata was more abundant in medium altitude areas, suggesting niche partitioning regarding altitude between the two species.
The main objectives of this study were to determine the host ranges of the different tephritid species in Comoros and to understand their distribution according to host and altitude.

Study sites
This research was conducted on the three islands of the Comoros: Grande Comore, Anjouan and Moheli from June 2013 to July 2020.
Grande Comore is the main island for agricultural activities with Moroni as the main port of entry, and is situated at about 80 km from Anjouan, while Mohéli is about 40 km from the other two islands ( Figure 1). The maximum altitudes are 2361, 1575, and 860 m for Grande Comore, Anjouan, and Mohéli, respectively. Whereas, altitudes go up to 860 m in Moheli, we have not been able to sample at more than 300 m, due to a scarcity of fruit resources at higher altitudes. Although these three volcanic islands have slightly different climatic conditions, they have a hot and rainy season from November to April (28 C-32 C and 376-1018 mm in lowlands) and a cool and dry season from May to October (24 C-27 C and 38-942 mm in lowlands).

Sampling
Cultivated and wild fruits were collected in cultivated fields, backyard gardens, and roadsides covering most parts of these islands from sea level to 887 m asl from June 2013 to July 2020 ( Figure 1, Table S1).
The methods of fruit collection, transport, and incubation in the laboratory were as previously described (Copeland et al., 2002). Fruits at all stages of development were randomly sampled from the plants, and very recently fallen fruits without decomposition or attack by soil organisms were occasionally collected from the ground. Host plants were identified in the field using the manuals of Quilici and Jeuffrault (2001). A sample is defined as a fruit collection from a given place at a  The bottom of each boxes was covered with a layer of sterilized volcanic sand sifted at 2 mm, to allow pupation of mature larvae (Woods et al., 2005;Rwomushana et al., 2008), with dimensions of 12 Â 7 Â 9 cm, 11 Â 9.5 Â 8 cm, or 7 Â 7 Â 4 cm, depending on fruit size.

Incubation of fruit samples
The lids were perforated and covered with muslin. The boxes with F I G U R E 1 Map showing the 147 sites in Comoros Islands where fruits were sampled to determine the host range of the different Tephritidae species. The inset shows the location of the Union of Comoros Islands in the Indian Ocean fruits were placed in a room at 25 C AE 3 C with a relative humidity of 71% AE 10% and a photoperiod of 12:12 AE 1 h (L:D), which allow the development of all studied species. Pupae were collected by sieving the sand after 1 and 2 weeks of incubation, respectively. All of the pupae collected from an individual fruit were placed in one transparent box with a perforated lid. After adult emergence, tephritids were placed after one week in 95% alcohol, sexed, and identified. An identification key of Comoros fruit flies was initially prepared based on the study of Comoros Archipelago tephritid diversity (De Meyer et al., 2012). The field guide to the management of economically important tephritid fruit flies in Africa (Ekesi & Billah, 2006) was also used. All identified species were sent to Cirad La Réunion for morphological and genetic confirmation by barcoding.

Statistical analysis
To understand the interaction between host specialization and altitude, we choose to study Grande-Comore island in detail because of the presence of higher number of host fruit species, the greater number of samples collected and the higher gradient regarding altitude compared to the other islands. In order to ensure that infestation rates were sufficiently representative of the field, we only used samples > = 10 (98 samples were not used in the analysis) and calculated the number of infested fruits / total number of fruits collected for each host plant species.
Modularity is a good proxy of interaction niches in ecological networks for coexisting species or populations and simplify the description and understanding of an ecological system, by representing not each and every species, but aggregating their interactions (Dormann & Strauss, 2014). The modules represent interacting groups with withinmodule interactions more prevalent than between-module interactions (Dormann & Strauss, 2014). In order to understand how the interactions between Tephritid species and host are partitioned in the community, we measured modularity, using Beckett (2016) algorithm, as implemented in the function computeModules from the package bipartite (Dormann et al., 2008) in R version 4.1.0 (R Core Team, 2021). Five modules were identified for all interactions between Tephritidae and their host fruit species in Grande-Comore.
We then analysed infestation rates (number of infested samples over total samples collected for each fruit species) by generalized linear mixed-effects models (GLMMs) with a binomial error (Bolker et al., 2009). Fruit fly species, group of host plants identified for each module, altitude, and interactions between these variables were defined as fixed effects, while host plant species and samples were defined as random effects.

Tephritid species and host plant use
A total of eight species of fruit flies were obtained from the fruits sampled in Grande Comore (Table S2)

Niche partitioning among Tephritidae via host fruit and altitude
Regarding the interaction between host fruit and Tephritidae species in Grande-Comore, the network analysis was significantly modular (Q = 0.61) and composed of five modules (Figure 4). This analysis F I G U R E 2 Network of interaction between fruit fly species and host plant species for the three islands of Comoros, and line thickness is proportional to number of infested fruits by each Tephritidae species. The different colours correspond to the tephritid species most commonly found in that host. For the different host plant species (the right-hand boxes), colours correspond to the tephritid species, the most present in that host Nevertheless, two other frugivorous species C. malgassa and D. vertebratus, not found in the current fruit collection, were described from these islands but in low numbers from specimen collected in 2004(De Meyer et al., 2012. Almost ten years after B. dorsalis invasion in Comoros with a two-year Tephritidae trapping survey, Hassani  (Clarke et al., 2005;Rwomushana et al., 2008;Moquet et al., 2021).
While C. capitata is considered as a polyphagous species able to infest a wide variety of species (Copeland et al., 2002) it has only been found on 15 host species in our study (in low numbers), and being dominant on only five hosts including coffee berry and two species of chilli. Trirhithrum nigerrimum, also considered as polyphagous  was retrieved in our study in very low quantity and only in 5 host plants.
While tephritids are typically non-dispersive, especially when hosts are plentiful, some individuals may travel up to 12 km (Froerer et al., 2010). As our methodology involved collecting fruits, this provides direct estimates of the ability of fruit fly species to grow in a given host plant at a given altitude. Long distance dispersal is more often linked to human translocation of fruits, such as in cases of invasions between islands (Duyck et al., 2004). In different islands, host range of a given tephritid species can be different because of the differences in presence and abundance of host plant species but also linked to the presence and abundance of other fruit fly species that F I G U R E 4 Infestation rate of the different Tephritidae species in the different host plants in Grande-Comore (left panel). Using the same modules, data are detailed for 0-300 m, 300-600 m and > 600 m altitude. Five modules were identified using Beckett (2016) algorithm (Dormann et al., 2008, see methods for details) for all interactions between Tephritidae and their host fruit species in Grande-Comore. Order of rows and columns are defined by modules in red and are kept for the three altitude sections. The modules represent interacting groups with within-module interactions more prevalent than between-module interactions (Dormann & Strauss, 2014). Absence of fruit in a given altitude is represented by light grey interact via exploitative competition but also interference competition (Duyck, David, Junod, et al., 2006). The length of time after the onset of the invasion may also affect host range choice, with newly introduced species being primarily associated with their preferred hosts and the host range gradually expanding as the abundance of the tephritid population increases (Moquet et al., 2021).

Niche partitioning regarding host range and altitude
While niches of the eight species overlap, there is a clear niche partitioning regarding host range for most of the species. Ceratitis capitata is known to be able to infest a wide variety of host species around the world, nevertheless, in the present study, it was only found dominant on a few small or toxic fruit species shared with T. nigerrimum. The usual described main hosts in the tropics of C. capitata (from the literature) are from Myrtaceae, Anacardiaceae, and Rutaceae families, which are occupied and dominated by B. dorsalis in Comoros. Niche partitioning between B. dorsalis and C. capitata in Comoros is also linked to altitude as already suggested by Hassani et al. (2016). Previous studies showed that B. dorsalis prefers warmer environment compared to C. capitata (De Meyer et al., 2010;De Villiers et al., 2015;Vera et al., 2002). The distribution and population dynamics of tephritids are closely related to influence of temperature and rainfall, and have a direct impact on tephritids life history traits . This dominance of B. dorsalis and potential refuge niche of C. capitata in higher altitude or on fruits such as coffee berry has been observed in Hawaii after B. dorsalis invasion (Vargas et al., 1995). A very similar pattern was recently documented in La Réunion and in numerous African countries where B. dorsalis significantly reduced host range and climatic niche of species already present such as C. capitata, C. quilici, C. cosyra (Ekesi et al., 2009;Rwomushana et al., 2008;Mwatawala et al., 2009;Moquet et al., 2021).
In its refuge niche, partly on several Solanaceae, C. capitata may also compete for resources with N. cyanescens, specialist of this family (Brévault et al., 2008). However, these two species have different distributions regarding to altitudes, N. cyanescens being more present in lowland areas.
While the three cucurbit fruit flies recorded in this study may interact, there is an important and significant niche partitioning linked to altitude. Indeed D. bivittatus was found more in lowlands and D.
ciliatus in higher altitudes. Dacus punctatifrons, while widespread in Africa (De Meyer et al., 2012) was found in very low number in Comoros. The latter species is dominated by the two other cucurbit infesting species and seems to have no private niche regarding to host plants, nor altitudes, and it could be in the process of extinction. A similar low level of the indigenous fruit fly D. etiennelus has been observed, which seems in a process of extreme population decrease that might also lead this species to extinction. F I G U R E 5 Relationship between infestation rate of each Tephritidae species and altitude predicted from model for each group of host plant (see Figure 4) Competitive replacement by B. dorsalis Nutrient composition of host fruits of Tephritidae explains their suitability for larvae as it greatly affects larval growth, developmental time, and survival (Krainacker et al., 1987;Hafsi et al., 2016). Larval development is correlated with female preference (Charlery de la Masselière, Facon, et al., 2017) and explains mainly host range in the field (Facon et al., 2021). Host fruit specialization is therefore partly explained by their fundamental niche (Charlery de la Masselière, Facon, et al., 2017;Hafsi et al., 2016) and by interaction among species such as interspecific competition (Charlery de la Masselière, Ravigné, et al., 2017;Facon et al., 2021;Moquet et al., 2021). Ceratitis capitata was probably present on a wide range of host plants in Comoros before the invasion by B. dorsalis. While we have no direct data on that from the present study, we were able to trace different mentions of C. capitata on major host fruits in Comoros from the literature (see the Introduction section). Furthermore, a recent study in laboratory showed that larvae of C. capitata were able to exploit a wide range of host fruits and had a good fitness on most infested fruits by B. dorsalis found in the present study such as Indian almond, mango, guava, or strawberry guava (Hafsi et al., 2016). These hosts are also among the preferred egg-laying hosts of C. capitata in laboratory conditions (Charlery de la Masselière, Facon, et al., 2017). This wide fundamental niche is however counterbalanced by the fact that C. capitata is a weak competitor in terms of exploitative competition in fruits and interference between females (Duyck, David, Junod, et al., 2006). This competitive disadvantage is probably linked to the small body size of this species compared to other polyphagous invasive Tephritidae (Duyck et al., 2007).
The potential competitive displacement by B. dorsalis has led to an apparent specialization of C. capitata on a refuge niche on a few host fruits where it is still dominant. Some of these fruits such as coffee berry and the two species of chilli are known to contain toxic compounds and are particularly of small size. Larvae of C. capitata, may need less resource and accomplish quickly their development compared to larvae of bigger size polyphagous species (Duyck et al., 2007). While the small size body of C. capitata appears a disadvantage in term of competition with B. dorsalis in fruits containing a lot of resource, it is probably an advantage in its refuge niche.
C. capitata is also know have a refuge niche in coffee berry and chilli in Hawaii and La Réunion (Moquet et al., 2021;Vargas et al., 1995).
While T. nigerrimum is also considered as polyphagous , it is only present on a few species, some shared with C. capitata. The other species are known to be specialized on Cucurbitaceae and Solanaceae, being rarely present on hosts of other families.
Description of the detailed host ranges of Tephritidae in Comoros is important for pest management. For instance, trapping or releases of parasitoid to manage populations of B. dorsalis would need to be done close to its most highly infested hosts. Moreover, some general useful principles for biosecurity can be drawn from our study. Compared to C. capitata, the risk of introduction of B. dorsalis is probably lower in cooler climate as its infestation rate decreased with altitude (Hassani et al., 2016) but global potential geographical distribution of B. dorsalis may be impacted by climate change (Qin et al., 2019). For countries where B. dorsalis is already present, the risk of invasion by C. capitata is probably weaker as fewer host fruit species would be available, or climate would be less favourable to start new populations where B. dorsalis is less present. The habitat with the most abundant and accessible resource may act as an invasion filter in which potential invasive species need to be competitively superior to already establish species, before spreading to other habitats (David et al., 2017).

ACKNOWLEDGMENTS
We thank the farmers and all those who allowed us to collect fruits and vegetables in their fields and gardens. We thank the Institut National de Recherche pour l'Agriculture, la Pêche et l'Environnement

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

SUPPORTING INFORMATION
Additional supporting information can be found online in the Supporting Information section at the end of this article.
Appendix Table S1. Sampling sites in the three studied islands Table S2. List of plant species collected with number of samples, fruits and abundance of each Tephritidae species in the island of Grande-Comore (2013-2018. A sample is defined as a fruit collection from a given place at a given date Table S3. List of plant species collected with number of samples, fruits and abundance of each Tephritidae species in the island of Anjouan (2013)(2014)(2015)(2016)(2017)(2018). A sample is defined as a fruit collection from a given place at a given date