Niche partitioning among two Ceratitis rosa morphotypes and other Ceratitis pest species (Diptera, Tephritidae) along an altitudinal transect in Central Tanzania

Abstract Two standard parapheromones, trimedlure (routinely used for monitoring Ceratitis rosa and Ceratitis capitata) and terpinyl acetate (routinely used for monitoring Ceratitis cosyra) were compared with enriched ginger root oil (EGO) lure for detecting and monitoring the presence and relative population abundance of these particular pest species. Standard yellow fruit fly traps were used for the comparison, which was conducted at 10 sites along an altitudinal transect ranging from 540 to 1650 masl on the Uluguru mountains, in Morogoro Region (Central Tanzania). A gradual change of relative occurrence of the two Ceratitis rosa morphotypes was clear from the EGO lure trapping. The morphotype R1 was predominant at lower altitudes while morphotype R2 was predominant at higher altitudes. Further experiments are needed to confirm the consistency of the observed pattern across regions, seasons and years as well as possible differences in the developmental physiology of both morphotypes. The mango fruit fly, Ceratitis cosyra, showed a distinct predominance at altitudes below 800 masl as shown in both the EGO lure and the terpinyl acetate trapping. The catches of all three target species were higher in traps with the EGO lure compared to the conventional lures trimedlure and terpinyl acetate. It is argued that for these species EGO lure can act as a suitable and more effective alternative for trimedlure and terpinyl acetate parapheromones. In addition, EGO lure has the added advantage that it combines the taxon spectrum for the two latter substances, thus requiring the use of only a single attractant.


Introduction
The Natal fruit fly, Ceratitis rosa Karsch, is an indigenous pest of significant importance to horticultural production in Africa. It is a member of the Ceratitis FAR complex, that is comprised of this and two other polyphagous, and morphologically similar species: C. fasciventris (Bezzi) and C. anonae Graham (Barr andMcPheron 2006, Virgilio et al. 2013). The distribution of C. rosa in Africa ranges from South (from Western Cape in South Africa onwards) to eastern Africa, with the northernmost records from the Central Highlands in Kenya (De Meyer 2001). Ceratitis rosa can survive in a wide range of climates, but with less preference for drier areas (De Meyer et al. 2008, De Villiers et al. 2013. The pest can impact production of both tropical and temperate fruits because its population is relatively stable across altitudes (Geurts et al. 2012).
The climatic requirements and potential distribution of C. rosa have been subjects of controversy. This became more evident in studies that compare climatic niche of C. rosa and other Ceratitis species. De Meyer et al. (2008) reported that C. rosa and the Mediterranean fruit fly C. capitata (Wiedemann) appear to have broadly similar potential ranges in Africa and southern Europe, but the latter may be more tolerant to a wider range of climatic conditions. However, there have been contrasting reports about thermo-tolerance of C. rosa. A minimum thermal developmental threshold reported by Duyck and Quilici (2002) is substantially lower than what was reported by Grout and Stoltz (2007). The ensuing confusion is whether the species is more adapted to cooler or warmer climates. In another study, Nyamukondiwa et al. (2010) reported that C. capitata and C. rosa have similar levels of survival to acute high and low temperature exposures under common rearing conditions. However, the time to extinction is greater for C. capitata than for C. rosa, especially in habitats where temperatures frequently drop below 10 °C.
The contrasting observations suggested the existence of two C. rosa biotypes with different climate requirements (Grout and Stoltz 2007). Recently, Virgilio et al. (2013) distinguished two C. rosa genotypes, designated as R1 and R2, that may occur in sympatry. The genotypes conform to two C. rosa morphotypes described by De Meyer et al. (2015). These new insights suggest revisions of current models of ecological niche requirements and invasion risk of C. rosa (Virgilio et al. 2013). Generally, R1 is abundant in the low land warm areas, while R2 is abundant at higher altitude cold areas. But the actual distribution of the two morphotypes is not well known and it is the focus of this study.
In studying the distribution of the two C. rosa morphotypes, it was desirable to understand niche partitioning between C. rosa and two other economically important Ceratitis species, C. capitata and marula fly C. cosyra (Walker). Male specimens of the three Ceratitis species are attracted to different lures. Ceratitis capitata and C. rosa are attracted to trimedlure, while C. cosyra is attracted to terpinyl acetate (White and Elson-Harris 1994). Recently the Enriched Ginger Oil (EGO) Lure was found to be more effective than trimedlure for C. rosa (Mwatawala et al. 2012). The limited comparisons, which were done in low land warm areas, showed that C. rosa, C. capitata and C. cosyra can be attracted to EGO lure, making it a better, single substitute for multiple lures. However, the results contrast reports from Hawaii, where trimedlure was more attractive to C. capitata than EGO lure (Shelly and Pahio 2013), warranting further investigations. In this experiment we studied the ecological niche partitioning among three Ceratitis species across an altitudinal range while at the same time comparing effectiveness of three lures: EGO lure, trimedlure and terpinyl acetate.

Methods
Ten locations, spaced at similar altitudinal intervals along a transect extending from 550 to 1650 masl were selected in the Morogoro region, Tanzania, (Table 1a, b; see also Geurts et al. (2012) for altitudinal profile of the sampling area except for the lowest sampling point) and sampled for three times in June 2013 (1 Jun, 15 Jun, 29 Jun). The average difference in temperature between the highest and lowest sampling point was previously reported to range between 7-8 °C (June average temperatures 15-22.5 °C, see Geurts et al. 2012). Modified McPhail® traps (Scentry Co, Bilings, MT, USA) were hung on fruit trees, usually mango, except at the high-altitude sites where traps were hung either on peach, plum or apple. Traps were baited with one of three different parapheromones: terpinyl acetate (TA), trimedlure (TM) (both purchased from IPS, Elsmere Port, UK) and EGO lure (EGO) (purchased from Insect Science, Tzaneen, South Africa). In addition to the different lures, a killing agent DDVP (containing 20% W/W dichlorovos; purchased from IPS) was placed in each trap. Sticky glue "tangle foot" was applied on the branches on which traps were hung to prevent predatory ants from accessing insects caught in traps.
Three replicate traps for each lure were placed at each altitude (for a total of 90 traps). Traps were activated for a single week and fresh lures and killing agents were used at each sampling instance. To guarantee replicate interspersion, traps where randomly re-positioned on different tree branches before each sampling. Flies collected from each trap were placed in uniquely marked vials, and brought to the lab for identification, counting and preservation in 70% ethanol. Trapping followed guidelines given by the International Atomic Energy Agency and FAO (IAEA 2013). The identification of flies was done using keys and characters presented by White and Elson-Harris (1994). The two C. rosa morphotypes were sorted following characters given by De Meyer et al. (2015). Only males C. rosa R1 and R2 were sorted as there are no discriminating morphological characters known for females.
The R package GAD (Sandrini and Carmago 2012) was used for analysis of variance (ANOVA) of cumulative abundances of flies collected in each trap. ANOVAs allowed testing differences between (a) abundances of male C. rosa, C. cosyra or C. capitata (with lure as fixed and altitude as random orthogonal factors) and (b) abundances of the two C. rosa morphotypes (R1 and R2) (with type as fixed and altitude as random orthogonal factors). Before analyses, data were fourth root transformed and homogeneity of variances were verified through Cochran's C test (Mair and Eye 2014). Student-Neuman-Keuls (SNK) tests were used for posteriori comparisons of means (Hochberg 2014).

Results
A total of 836 male specimens of the three Ceratitis species were trapped along the transect (Table 2) (female specimens constituted less than 1% of all trappings and were not included in the analyses because of lack of diagnostic morphological features for the two C. rosa morphotypes). More specimens were caught in traps baited with EGO lure, than in traps baited with TA or TM ( Figure 1). Ceratitis cosyra was the most abundant species constituting 61.6% of all trapped specimens, while C. rosa (33.3%) and C. capitata (5%) had lower abundances ( Figure 2). A total of 279 C. rosa R1 and R2 were collected from EGO lure traps with R1 being more abundant (61.2% of. C. rosa morphotypes).
C. rosa also showed significant differences between lures (EGO > TM) and altitudes (Tables 4a and 4b, Figure 3). The distribution of the C. rosa R1 and R2 types along the altitude is shown in Figure 4. Morphotype R1 is present throughout the altitudinal transect, with higher abundances at lower attitudes. Conversely, morphotype R2 was more abundant at higher altitudes, reaching a peak at the Langali -Konrad station (1268 m asl) while being absent at the lower station (SUA, 550 m asl). ANOVA (Table 5a, 5b) showed significantly higher abundances of morphotype R1 at 550 masl and of morphotype R2 at 1170, 1268, 1392 and 1644 masl.
The catches of C. capitata, were remarkably low with only 42 specimens trapped (Table 1, Figure 2).

Discussion
Our results showed a gradual change in the relative abundance of the two C. rosa morphotypes, with R1 being predominant at lower altitudes and R2 being predominant at higher altitudes. Further experiments will have to show if these differences are  consistent across seasons and years and whether the different distributions are related to differences in temperature thresholds and developmental rates of the two morphotypes (Tanga et al. 2015). The results of this study may explain the differences observed between Grout and Stolz (2007) versus Duyck and Quilici (2002). The South African morphotype studied by Grout and Stoltz (2007) may well represent the morphotype R1 that is dominant in lower altitude areas. On the other hand, the population in Réunion could correspond to morphotype R2 predominant in the high altitude areas, as Virgilio et al. (2013) showed that the population studied from Réunion belonged exclusively to R2. In Mpumalanga and Kwa-Zulu Natal regions of South Africa, both types occur, but it is not clear what population was used by Grout and Stolz (2007) for their experiments. The climatic niche partitioning of these two morphotypes is not very clear as both morphotypes were present throughout the altitudinal transect, albeit at contrasting population levels, and it still remains to be explored what biotic and/or abiotic factors exactly determine their distribution. It can be further inferred that the impact of morphotype R2 might be more pronounced on temperate fruits like peach, avocado and apple, while morphotype R1 might have a more important impact on tropical and subtropical fruits. Of course, these hypotheses need further experimental validation including sampling at different fruit phenological states.
Captures of C. cosyra, and possibly of C. capitata, were higher in the lower altitude areas, where tropical fruits are grown, but low at high altitudes. The distributions of these two species in the field conform to the laboratory results by Duyck and Qulici (2002) and Grout and Stoltz (2007), in Réunion and South Africa respectively. According to Geurts et al. (2012) the presence of suitable hosts and the competition between fruit fly species seem decisive for diversity along the altitudinal transect, although climatic suitability cannot be neglected. The competitive ability of Bactrocera dorsalis (Hendel) affects the abundance of Ceratitis species. The presence of B. dorsalis has impacted the abundance of Ceratitis species, notably C. cosyra. Fruits infestation by C. cosyra seems to be negatively affected by B. dorsalis especially in hosts like mango (Mangifera indica L.). In Benin, Vayssières et al. (2005) reported a decrease in density of C. cosyra as the density of B. dorsalis increases. The evidence of competitive displacement of C. cosyra by B. dorsalis was provided by Ekesi et al. (2009) with B. dorsalis having stronger competitive traits than C. cosyra (Salum et al. 2013). The latter is now mostly confined to hosts of the family Annonaceae in this study area (Geurts et al. 2012). On the contrary, the abundance and infestation of C. rosa do not seem to be significantly affected by the abundance of B. dorsalis (Geurts et al. 2012). Bactrocera dorsalis is not yet established in high altitude areas (Geurts et al. 2013), where R2 is dominant. However, the competition between morphotype R1 and B. dorsalis can be expected. So far, data collected from the same region do not suggest the displacement of C. rosa by B. dorsalis.
The population of C. capitata recorded in this study was very low. This species is more restricted in this study area to hosts like Fortunella margarita (Thunb.) Swingle (Mwatawala et al. 2009), and Capsicum spp. (Mziray et al. 2010). There are no data on distribution and abundance of C. capitata prior to the introduction of B. dorsalis in the study region, hence competitive displacement cannot be ascertained. The distribution of Ceratitis species along the altitude has an implication of management programs. As C. rosa of morphotype R2 is the predominant pest species at higher altitude areas, any fruit flies management program in this particular region should target morphotype R2.
Of the three male lures tested, EGO lure attracted more flies than TM with regard to C. rosa and C. capitata (and higher catches than TA with regard to C. cosyra). In a previous study, the catches of C. rosa and C. capitata by EGO lure were equal or superior to TM . The present study showed that EGO lure is a significantly stronger attractant for the males of C. rosa, C. capitata and C. cosyra.
The findings of this study support the results of Cunningham (1989) who reported that alpha-copaene is 2-5 times more attractive for male Mediterranean fruit flies than TM. This is in contrast to Shelly and Pahio (2002) and Shelly (2013) who observed higher catches of C. capitata in traps baited with TM than EGO lure, especially as time progressed. They went on to suggest that neither capilure (not a subject of the current study) nor EGO lure can be an adequate substitute for TM. According to Shelly (2013) the discrepancy in the results for C. capitata between Hawaii and Africa could reflect differences in the composition of the (ginger) oils used in the two regions. The presence and concentration of sesquiterpenes other than α-copaene may affect Mediterranean fruit fly response to natural oils. Also variation in the chemical composition of ginger root oils from different suppliers could generate different results in trapping studies (Shelly 2013) and should be studied.
Despite the observed discrepancies, EGO lure has an added advantage of attracting a wider spectrum of pest fruit flies, which allows deployment of a single lure trap rather than two different ones. TM is an effective lure for surveying and monitoring activities for male Mediterranean fruit flies (Grout et al. 2011) and members of the Ceratitis FAR complex (Virgilio et al. 2008) including C. rosa. Ceratitis cosyra males are not attracted to TM but to TA (White and Elson-Harris 1994). This study showed that C. cosyra responds more to EGO lure than TA. It is concluded that EGO lure should be considered as a suitable alternative for TM in detection, monitoring and control programs for African fruit flies of the genus Ceratitis. The major drawback at the present moment is, however, the cost of EGO lure which is currently about tenfold of that for either TM or TA, when purchased from commercial suppliers. As such, the purchase of EGO lure by poor farmers is currently a financial restraint if no additional financial aid is provided.
Further studies are currently being carried on across diverse ecologies in Africa (Manrakhan pers. comm.) in order to verify the current observations, before EGO lure can be generally regarded as a better substitute for other attractants. Such studies should include a wide range of attractants for Ceratitis species. Probably, EGO lure from different sources should also be tested within the same framework. More advanced studies like capture-mark-release studies (see also Manrakhan et al. 2014) can be conducted to test the sensitivity of these Ceratitis species to EGO lure. This information is necessary to verity the effectiveness of EGO lure as part of management program for Ceratitis pest species.

Conclusion
This study has presented the distribution of two C. rosa morphotypes across an altitudinal transect. Morphotype R1 is more dominant in lower altitude, warmer areas while morphoptype R2 is prevalent in high altitude, cooler areas. However, both morphotypes occur throughout the transect. EGO lure attracted all the three Ceratitis species, including the two C. rosa morphotypes, more effectively than TA and TM. It is suggested that the use of EGO lure as a single attractant for the combined capture of these important Ceratitis species should be further explored.