Description of third instar larvae of Ceratitis fasciventris, C. anonae, C. rosa (FAR complex) and C. capitata (Diptera, Tephritidae)

Abstract Third instar larvae of members of the Ceratitis FAR complex, including Ceratitis fasciventris (Bezzi), Ceratitis anonae Graham, and Ceratitis rosa Karsch are described and compared with those of Ceratitis capitata (Wiedemann). Diagnostic characters, such as presence vs. absence of a secondary tooth on the mandibles, previously used to separate Ceratitis capitata from Ceratitis rosa, are shown to vary in each species. Significant variation in diagnostic morphological characters among populations of Ceratitis rosa from east and south Africa is documented; however, the differences are not simply congruent with the R1 and R2 designations based on other studies. Quantitative measures of numerous morphological characters are consistently smaller in the larvae of Ceratitis fasciventris and distinguish them from other species of the FAR complex. Larvae of Ceratitis capitata can be distinguished from those of the FAR complex by characters such as absence of accessory plates of the oral ridges, the shape of the anterior spiracle, and the pattern of dorsal spinules. Previous studies indicated that absence of accessory lobes separate the genus Ceratitis from Bactrocera, but this is shown to be incorrect, as accessory lobes are in fact present in several species of Ceratitis.


Introduction
Members of the Ceratitis FAR complex, including Ceratitis fasciventris (Bezzi), Ceratitis anonae Graham, and Ceratitis rosa Karsch, are serious agricultural pests in large parts of Africa. Understanding of the species taxonomy is important to determining host plant relationships, pest management practices, knowledge of geographical distribution, and quarantine and related plant protection issues. In recent years, the taxonomy of species of the Ceratitis FAR complex has been clarified by careful study of adult morphology, and on-going morphological, genetic, and physiological studies suggest that additional, previously unrecognized species may be present (De Meyer 2001, De Meyer and Freidberg 2006, Virgilio et al. 2008, Virgilio et al. 2013). The present morphological study of the immature stages is intended to augment and possibly corroborate the differences seen among adult and various other biological parameters. In particular, this study was originally prompted by a problem in identifying larvae that had been intercepted alive infesting fresh peppers (Capsicum chinense Jacq. 'Habanero') that had been shipped from the Netherlands and intercepted in Miami, Florida in August, 2004. It wasn't clear whether they were C. rosa or Ceratitis capitata (Wiedemann) because of poorly documented overlapping morphological variation in these two species that confounded the identification process. Years later, in a telling of this dilemma during general discussion at the first meeting of the International Atomic Energy Agency's Coordinated Research Project on fruit fly cryptic species complexes in Vienna in 2010, the senior author learned that there existed research colonies of various Ceratitis species at ICIPE from which immature stages could be obtained, and a collaboration to do a morphological study was developed. It was quickly discovered that larvae from the Kenyan colony of C. rosa differed in a significant feature from those of a South African colony as described by Carroll (1998). This prompted a broader study of immature stages of all three species of the FAR complex and their various populations as described here.

Materials and methods
One goal of this paper is to describe morphological variation among geographic populations of C. rosa. To that end and for ease of comparison, Carroll's (1998) well-written description of mature 3 rd instar larvae is repeated verbatim below (in italics) and any differences observed in the populations in the present study are given in parentheses in bold, normal font. Descriptions of other instars, pupae and eggs of most fruit fly species, including important pests, are generally unavailable for comparison. Full data is given in corresponding tables. Some features detailed by Carroll (1998) were not examined in this study and those portions of her description are not repeated. These include most of the sensillae (on the stomal organ, maxillary palp, cephalic segment, Keilin's organ, and other body parts), and vestigial spiracular openings. Most of these sensillae are only seen with a scanning electron microscope (SEM), they are often difficult to find, and may be variably expressed. They are generally highly conserved among cyclorrhaphan Diptera and have not been found useful in alpha taxonomy. Carroll did not provide measurement data on the cephalopharygneal skeleton, but these data are provided here because they do show differences among species, and they reliably separate the instars. The descriptions and measurements follow the terminology of Teskey (1981), Steck and Malavasi (1988), Wharton (1988), andCarroll (1998). New figures are provided for most of the same features illustrated in Carroll (1998).
Corresponding descriptions were made of larvae of other species of the FAR complex, namely C. anonae (minimally described by Silvestri (1914) and included in the key of White and Elson-Harris (1992)) and C. fasciventris (not described previously). A further goal is to provide means of identifying larvae of these species morphologically, if possible. Additionally, equivalent observations were made of C. capitata larvae, as this important pest is likely to be encountered in the same host plants and geographic regions as members of the FAR complex. Voucher specimens of all African FAR colonies were verified morphologically based on adult males (De Meyer et al. 2015) and the Kenyan colonies were also genotyped for both sexes (De Meyer, personal communication 31 March 2015).

The samples in this study include
Larvae were killed in hot water and preserved in 70% ethanol or isopropanol. Specimens intended for SEM examination were sonicated for 30 seconds, then dehydrated in an ethanol series, followed by ethyl acetate, then air-dried, mounted on stubs, sputter-coated with gold-palladium, and examined in a JEOL JSM-5510LV SEM at FDACS/DPI, Gainesville, FL. Measurements derived from stub-mounted specimens were made from SEM photographs and calibrated using the embedded scale bar.
Specimens intended for examination under dissecting and compound optical microscopes were macerated overnight in 10% NaOH at room temperature. Once cleared, they were temporarily slide mounted in glycerin and positioned to allow measurements as needed. Most measurements were made manually using an eyepiece reticle calibrated for conversion to mm. Some measurements were made using a Zeiss AxioCam ICc 5 digital camera and ZEN 2 software (Blue edition, 2011). The optical microscopes used were a Zeiss Discovery V8 dissecting microscope, Nikon Labophot compound microscope, and Olympus BX51 compound microscope. Some imagery was obtained using a Leica Z16 APO lens, JVC KY-F75U digital camera, and Synchroscopy Auto-Montage v. 5.01.0005 software. Finally, specimens were mounted in Euparol or Canada balsam as permanent vouchers deposited at the Florida State Collection of Arthropods (FSCA) in Gainesville.
Sample sizes from which data were derived are provided with each species description. Not all character states could be observed nor was it possible to make all measurements on each specimen, as some were prepared for SEM examination and others were slide-mounted with varying success.
The length and width data presented for spiracles are based on cleared, slidemounted specimens only. The range of measurement made on the same structures as seen under the SEM often do not overlap. We consider the measurements made on slide-mounted specimens to be more accurate, as it is much easier using this preparation to determine whether or not the structure is lying flat when measured.
The data from these descriptions will be incorporated into an interactive system based on that of Carroll et al. (2004) to improve the accuracy of fruit fly larval identifications.
Head (Figure 1c-g): with cephalic lobes moderately developed, in lateral view more rounded and protuberant than in C. capitata (true for some C. capitata samples, but in others the difference is very subtle at best; observed differences may be due, at least in part, to the method used to kill and preserve larvae); antenna 2-segmented, both segments with sclerotized walls, the distal segment apically thin-walled and conical; maxillary palp with ... sensilla ... visible by SEM as 3 papilla sensilla and 2 knob sensilla, the remainder as pits; dorsolateral group of sensilla with 2 papilla sensilla and a pit sensillum, adjacent to but distinct from palp; stomal organ (Figure 1c-g) with primary lobe small, bearing 3-4 unbranched peg sensilla, ...; 6 secondary lobes present: a broad, flat subtending lobe and a lobe medial to it; usually 2 additional lobes immediately surrounding the primary lobe anteriorly, and usually 2 lobes anteromedial to these, all with edges entire; none of these secondary lobes is strikingly similar to oral ridges; sclerotized stomal guards absent; labium short, triangular, with narrow lateral lobes; .... Oral ridges (Figure 2c-h) usually 9-11 (but two had 8 and three had 12 ridges on one side only) (same range); well developed with margins scalloped to 1/4-1/5 of their depth (visible by SEM), located on a semicircular region laterad of mandible; accessory plates (supernumerary ridges) and other reticulation absent (accessory plates present or absent, variably developed in different populations) (Figure 2c-h).
Variation among populations. There is considerable variation among populations in various quantitative characters, and some ranges do not overlap. See accompanying tables of comparative data for various measures and counts taken on the oral ridges, accessory plates, stomal organ, and anterior spiracles (Table 1), cephalopharyngeal skeleton (Table 2; mandible abbreviated as MH), posterior spiracles and anal lobes (Table  3), and mandible secondary tooth (Table 4). Some notable differences among populations include: Dorsal spinules-Carroll (1998) described a Stellenbosch-derived colony as having dorsal spinules present on segments T1-T3 and usually A1. In all populations observed in this study, however, including a newly derived colony from Stellenbosch, larvae have dorsal spinules present on T1 and T2 only, but none on T3, A1 or beyond.
Oral ridge accessory plates-Carroll (1998) described a Stellenbosch-derived colony as lacking accessory plates. However, among populations observed here, larvae of R1-Kenya, R2-Kenya, and R1-Nelspruit have accessory plates present, numerous and well-developed to the point of having serrate edges. Alternatively, larvae of R2-Pretoria have accessory plates lacking or minimally present as a few thin ridges or nubs, without serrate edges, and on R2-Stellenbosch they range from a single well-developed serrate accessory plate plus a few additional nubs, to numerous nubs, to completely lacking.
Anal lobes-Carroll (1998) described a Stellenbosch-derived colony as having anal lobes entire (rarely grooved). In all populations observed in this study, however, including a newly derived colony from Stellenbosch, larvae have grooved anal lobes.
Quantitative measures that do not (or minimally) overlap among samples include the trachea diameter at base of anterior spiracle (larger in Kenyan populations than South African populations regardless of hot or cold type); length from tip of mandible to notch in CPS, length from tip of mandible to tip of ventral prominence, and height of dorsal arch (larger in R1-and R2-Kenya + R2-Stellenbosch vs. smaller in R1-Nelspruit and R2-Pretoria). Various other individual pairings of samples do not overlap for some measures and counts especially those associated with the posterior spiracles (See Table 3). These differences in quantitative characters may merely reflect relatively small sample sizes or the result of artificial selection in laboratory colonies.     Table 3. Posterior spiracle quantitative measures and anal lobes (length in mm).    Figures 1b, 2b, 3b, 4b, 5b, 6b, 7b, 8b, 9b Diagnosis of third instar. Medium-sized muscidiform larvae with mandibular tooth ventrally grooved, with minute subapical mandibular tooth; with 10-11 oral ridges; accessory plates present; petal-like secondary stomal lobes present, sclerotized stomal guards absent; dorsal spinules present on segments T1-T2; anterior spiracles with 10-13 tubules in a single straight to slightly curved or sinuous row; base of anterior spiracle cylindrical, ca. half as wide as apical width; posterior spiracles with rimae ca. 3 times longer than wide; spiracular processes mostly unbranched to mostly branched with narrow bases; caudal ridge present; anal lobes grooved, posterior portion often larger than anterior portion.

Discussion
Although larval stages of numerous fruit fly species have been described, very few are based on wide geographic sampling, and often they are based on colony material. The extent to which these descriptions reflect actual variation in nature is generally unknown. Specimens from laboratory colonies are probably more homogenous than those collected directly from the wild. Sample sizes used in this study are small, so we have to expect that the range of measurements presented here is less than that in nature.
There are consistent morphological differences in larval character states among some of the C. rosa populations studied here. Some of these would be considered key diagnostic characters to recognize different species in other genera, e.g. Anastrepha (Steck et al. 1990). Whether they represent intra-specific variation or diagnostic differences among biologically distinct taxa of Ceratitis cannot be answered using these data alone. The larval morphological differences observed among populations examined here are not congruent in any simple way with the R1 and R2 designations. The diagnosis and description of C. rosa s.l. given here incorporates data from all of the populations observed. If some of these eventually are determined to represent different taxa, then the diagnosis and description of C. rosa s.s. may require alteration, as some character states do not overlap among populations. Additional data of other types (e.g., genetic, behavioral, etc.) are required to determine the taxonomic interpretation of larval morphology data.
Even beyond the question of possible cryptic species among C. rosa s.l., it is maddeningly difficult to find reliable diagnostic differences in larval morphology among species of the FAR complex. Many of the quantitative larval characters seem to be little constrained and their ranges vary wildly. However, Ceratitis fasciventris can generally be distinguished from C. rosa s.l. and C. anonae by its smaller dimensions of the CPS and anterior spiracle apical width, and lower counts of spiracular processes and narrowness of their bases.
C. capitata larvae can be separated from most individuals of the FAR complex by the absence of oral ridge accessory plates and the presence of dorsal spinules on T3. Also the shape of the anterior spiracle seems consistently different (smoothly expanded from the base to the tubules), as described and illustrated by Carroll (1998). Presence vs. absence of a small secondary tooth on the mandible is not a reliable character state to separate them. While the secondary tooth is typically present and easy to see on FAR larvae, it may be absent or poorly developed as seen in the C. rosa R2 colony at ICIPE. Conversely, the secondary tooth is often present on larvae of C. capitata, although it frequently is poorly developed and not likely to be noticed except by use of SEM.
It should be noted that presence vs. absence of oral ridge accessory plates has been used as a key character to separate larvae of the genus Ceratitis from those of Bactrocera (White and Elson-Harris 1992). It is now clear that this distinction was falsely based on too limited taxon sampling within Ceratitis. Further studies are needed to determine reliable characters to separate these genera.