Host‐race formation in Chaetostomella cylindrica (Diptera: Tephritidae): Morphological and morphometric evidence

Chaetostomella cylindrica is a highly oligophagous tephritid infesting the flower heads of six genera and 10 species of thistles in Lebanon, and is predominant on two hosts in sympatry, Notobasis syriaca and Onopordum illyricum. Adult flies emerging from N. syriaca fit more closely the description of the species with respect to the colour and pattern on the mesonotum. This study compares morphometrically and morphologically the host races associated with N. syriaca and O. illyricum. Immatures of both races were similar, but all stages of the Onopordum‐associated race were significantly larger. Morphometric studies, based on two head and five wing measurements, using canonical discriminant analysis, allowed for the differentiation of the host races with more than 70% accuracy. The aculeus shape and length differed significantly between females of both races. The holotype of Trypeta lurida Loew 1844 was examined and appeared closer to the Onopordum host race. Chaetostomella cylindrica appears to be a complex of cryptic and reproductively isolated species. Present address: Mohammad S. Al‐Zein, Old Dominion University, Norfolk, VA, USA.


Introduction
Host races are populations that are isolated from each other by preferred habitat or host (Diehl and Bush 1984). Rapid host-race formation and sympatric speciation are common among tephritid fruit flies. Phytophagous tephritid populations often show different host preferences but they are morphologically similar or even indistinguishable from one another (Zwö lfer and Harris 1971). Host races sometimes represent unrecognized reproductively isolated sibling species that retain distinct host preferences in the absence of other barriers to gene flow (Bush 1975). Biological attributes such as mating on the host plant and positive correlation between host and mate selection, in addition to genetic control of host selection, allow the establishment and formation of new host races in areas of sympatry (Bush 1975).
Immature stages, namely eggs, third instars, and pupae, of the Onopordum-associated host race were obtained from dissections of flower heads of O. illyricum (Mount Lebanon: Bhamdoun, Aley Co.) while those of the Notobasis-associated host race were obtained from heads of N. syriaca (Mount Lebanon: Khaldeh, Baabda Co. and Bhamdoun, Aley Co.; Beqaa Valley: Chtaura, Zahleh Co.; North Lebanon: Koura Co.) The length and width of the immatures were measured under a stereomicroscope (Leica, Zoom 2000) using a calibrated micrometer. The cephalopharyngeal skeletons, anterior spiracles, and posterior spiracles of the third instar larvae of both species were prepared following the method described by Phillips (1946).

Comparative morphological studies of adult flies
External morphology of adults. The following morphological characteristics of the Notobasisassociated host-race adults (n530) and the Onopordum-associated host-race adults (n530) were recorded: height, width, and colour of the head; length of the antennae (1st, 2nd, and 3rd segments except the arista); colour of the abdomen as well as the colour of the bristles on it; the colour of the thorax; colour of the spots at the base of the dorsocentral and prescutellar setae; colour of the spot on the scutellum as well as the pattern on the mesonotum. For female flies, length and width at the base of the oviscape were also measured.
Comparative morphometric studies of the adults. Five wing measurements (Figure 1), namely the width of the wing at the level of the stigma, the length of the discal medial (dm) cell, the length of the third radial (R 4+5 ) cell at the boundary with the medial and discal cells, and the length of the preapical crossband (as one entity and in parts), as well as the width and height of the head were determined for 30 randomly selected males and 30 selected females of the two races.
Multivariate analyses, namely principal component analysis and canonical discriminant analysis, were performed based on these measurements.
Morphology and morphometry of male terminalia. Dissection and slide preparation of the male terminalia of the Notobasis-associated host race (n513) and the Onopordum-associated host race (n513) were performed according to the method described by White (1988). Measurements of the length and width of the glans (sclerotized part and vesica) of the distiphallus were recorded.
Morphology and morphometry of the aculeus. Dissection of the ovipositors of the Notobasisassociated host-race (n558) and the Onopordum-associated host-race (n541) females was performed according to the method described by White (1988). The length of the oviscape and the aculeus were measured under a stereo microscope and a compound microscope, respectively. A total of seven measurements of the aculeus were recorded: length and maximum width of the dissected aculeus; the location of the ventro-lateral groove on the aculeus tip as represented by two ratios (ratio 1 and ratio 2); the width of the aculeus tip at the level of the ventro-lateral groove (width 1 and width 2); the width of the pointed tip of the aculeus (Figure 2). A photograph of the aculeus tip was taken using a digital camera (Nikon E990) and a protractor was used to take two angle measurements of the aculeus tip. The first measurement is the measure of the angle formed by two imaginary lines joining the apex of the aculeus to the point where the aculeus starts to taper on each side. The second one is the measure of the aculeus apex angle, ''the angle between two imaginary lines placed tangentially across each side of the tapering aculeus'' (White and Marquardt 1989).

Statistical analysis
Statistical differences in morphometric characteristics between males and females within the same race, between males and females of the two races, and/or between the two races regardless of gender, were tested by performing two-tailed t tests for independent samples for variables that have a normal distribution, as well as Mann-Whitney tests for variables whose distribution is different from normal. The frequency distribution of measurements for each variable was tested for normality using one-sample Kolmogorov-Smirnov tests. All statistical tests were performed at 95% confidence level using SPSS (version 13.0).
To determine the dispersal of the 120 analysed specimens (considered as operational taxonomic units or OTUs), principal component analysis was performed using SPSS. A correlation matrix of the seven measured characters was generated, the values of the characters were standardized and the eigen values determined. Only the first two components were considered to determine the level of variation among the specimens. The same analysis was repeated for female specimens only (n560 OTUs), and then for male specimens only (n560 OTUs).
To determine whether the two races could be separated morphometrically, based on the two head and five wing measurements, canonical discriminant analysis (canonical variate analysis) was performed using SPSS. This analysis considers within-and between-group variations. A total of 120 specimens (30 females and 30 males reared from O. illyricum; 30 females and 30 males reared from N. syriaca) for which measurements had been taken were analysed; prior to the analysis, specimens were given codes to ensure that they belonged to either of the races.
Holotype (female) examined. Turkey, south coast of Asia Minor, ruins of Patara (ZMHU). Head yellow, 1.1 mm in width and 1.33 mm in height. Antennae yellow, 0.48 mm in length. Thorax yellow and partly obscured by pin. No spots at the base of the prescutellar setae. Shiny black spots on scutellum. Abdomen yellow with black bristles and brown spots. Aculeus 1.73 mm long. Aculeus broken at very tip. Ratio 2 ca 4.2. Wing 4.8 mm long and 1.8 mm wide.
Males. Head yellow to yellow brown, 1¡0.02 mm in width (n530), 1.28¡0.02 mm in height (n530). Antennae yellow, yellow brown or yellow orange, 0.47¡0.00 mm (n56) long. Thorax yellow with no spots at base of dorsocentral setae. Black spots at base of prescutellar setae, but sometimes faint brown or absent. Shiny black spots on scutellum. Usually faint brown, but sometimes brown or black pattern on mesonotum. Abdomen yellow with black bristles. Usually with two black or faint brown spots on each abdominal tergum.
Females. Head yellow, yellow brown or sometimes yellow orange, 1.1¡0.02 mm in width (n530), 1.42¡0.03 mm in height (n530). Antennae yellow or yellow orange, 0.49¡0.03 mm in length. Thorax yellow or yellow brown with no spots at the base of the dorsocentral setae. Black, sometimes very small faint brown or absent spots at the base of prescutellar setae. Shiny black spots on scutellum. Faint brown but sometimes dark brown, grey, or no pattern on mesonotum. Abdomen yellow to yellow brown with black bristles. In most cases, two black to faint brown spots on each abdominal tergum.

Notobasis-associated host race of C. cylindrica
Males. Head yellow to yellow brown, sometimes yellow green, 0.89¡0.02 mm wide (n530), 1.1¡0.02 mm long (n530). Antennae yellow to yellow orange, 0.44¡0.01 mm (n511) long. Thorax yellow or yellow brown; usually no spots or rarely very small faint brown spots at the base of dorsocentral setae. Black spots at base of prescutellar setae, but sometimes brown or faint. Shiny black spots on scutellum. Grey black or black pattern on mesonotum. Abdomen similar to the Onopordum-associated host race.
Females. Head yellow to yellow green, 0.97¡0.02 mm wide (n530), 1.23¡0.003 mm long (n530). Antennae yellow, yellow orange, or sometimes orange, 0.46¡0.02 mm long. Thorax yellow, sometimes yellow brown or yellow green, with no spots at base of dorsocentral setae. Black or sometimes brown spots at base of prescutellar setae. Shiny black or dark brown spots on scutellum. Grey black or black pattern on mesonotum. Abdomen similar to the Onopordum-associated host race.

Comparative morphometry of the immatures
The size of all the immature stages of the Onopordum-associated host race was significantly greater than those of the Notobasis-associated host race. The egg length and diameter of the two host races were statistically different (t57.4, df579, P,0.001 and t55.97, df579, P,0.001, respectively). The mean diameter and mean length of the third instar larvae were found to be significantly different between the two races (t55.5, df573, P,0.001 and t55.85, df573, P50.027, respectively). The length and width of the puparia also showed significant differences between the two races (t54.19, df543, P,0.001 and t52.51, df543, P50.016, respectively).

Comparative morphology and morphometry of the adults
Adult head and wings. Seven morphometric characteristics (five wing and two head measurements) were measured for a random sample of 60 flies of each race, 30 males and 30 females. Statistical analysis using two-tailed t test for independent samples showed that the means for the seven characters studied were significantly different (P(0.05) for males and females within and between the two races, as well as between the 60 Notobasisassociated and 60 Onopordum-associated flies regardless of the gender (Table I). The variation within species reflects sexual dimorphism in the two races as males were usually smaller than females. Given the intraspecific variation, principal component analysis and canonical discriminant analysis were performed to determine whether the observed interspecific variation could be useful in differentiating between these two races.
Principal component analysis revealed that the first principal component which accounted for most of the observed variation between the specimens was the length of the preapical cross band measured as one entity (component loading (CL)50.965), followed by the wing width (CL50.947) ( Table II). The first principal component also accounted almost equally for the variation due to any of the wing measurements. This suggests that the five wing measurements were highly correlated and including any one of them in any further analysis is sufficient. The second PC showed high loadings for the head width (CL50.579) followed by the head length (CL50.305). Similar results were obtained when PC analysis was done for only female specimens and for only male specimens. Plots of PC1 versus PC2 done for male specimens only and for female specimens only showed that there are two separate clusters reflecting the two races and that there is some overlap between the two groups ( Figure 5). However, there was less overlap and better clustering for male specimens compared to female specimens.
To separate between the two races based on within-and between-group variations, two canonical discriminant analyses (CD) were performed, one using all the seven morphometric characters and the other using two characters, namely the head height and the wing width, as the canonical discriminant function based on these two variables Table I. Mean values of the seven morphometric characteristics measured for a sample of 30 adult females and 30 adult males, for each of the Notobasis-associated and the Onopordum-associated host races of Chaetostomella cylindrica.

Notobasis associated
Onopordum associated  gave good separation between the two races. The characteristics of the two CD functions used are summarized in Table III. Based on the two CD functions, the flies were classified into two groups, in such a way that each fly has an equal probability of belonging to any of the two groups. Based on the first CD function, 79.2% of the flies were predicted to belong to the race they actually belonged to. However, 18.3% of the flies belonging to the Notobasis-associated host race were classified as Onopordum-associated and 23.3% of the flies belonging to the Onopordum-associated host race were classified as Notobasis-associated (Table IV). Based on the second CD function, in which only two variables were considered, 72.5% of the flies were predicted to belong to the race they originally belonged to. However, 26.7% of the flies belonging to the Notobasis-associated host race were classified as Onopordum-associated and 28.3% of the flies belonging to Onopordum-associated host race were classified as Notobasis-associated (Table IV). This shows that CD analysis can be useful in differentiating between these cryptic races with more than 70% accuracy (or 30% error).
Using the standardized CDF coefficients of the two CD functions (Table III), a score for each of the 120 flies analysed was calculated, and three scatter plots of these scores were constructed ( Figure 6). In the first plot, the scores obtained for each fly using the second standardized CD function were plotted as a function of the scores obtained using the first CD function. The same was done in the second and third plots, except that only male scores were used in the second plot and only female scores were used in the third. Analysis of the plots shows that there is, to some extent, clustering of the flies into two groups. The presence of some flies of the first species in the vicinity of the second race can be attributed  to the fact that some small or less mature flower heads had been picked up and adult flies had been reared from these flower heads. These flies might have been smaller than normal due to the fact that their larvae did not have enough food and had to pupate ahead of time, before storing enough nutrients to be used in the pupal stages (Tsitsipis 1989). Male terminalia. The morphological and morphometric comparisons undertaken proved that the terminalia of adult males belonging to the two races are similar, with the exception of fine differences in the glans, which was also longer in the Onopordum host race (Figure 7). The length of the epandrium was 0.48¡0.01 mm (n56) and 0.5¡0.01 mm (n59) while its width was 0.22¡0.02 mm (n56) and 0.23¡0.01 mm (n59), for the Notobasis-and Onopordumassociated host races, respectively. No significant differences were detected in the length (t50.99, df513) and width of the epandrium (t50.1, df513) between the two host races (P.0.05). On the other hand, the glans (the sclerotized part of the aedeagus together with the vesica) was significantly larger in the Onopordum host race. Its length was 0.58¡0.01 mm (n58; range 0.54-0.62) and 0.64¡0.01 mm (n58; range 0.57-0.68) (t53.23, df514, P,0.001) while its greatest width was 0.19¡0.004 mm (n59; range 0.15-0.2) and 0.20¡0.006 mm (n59; range 0.19-0.24) (t52.2, df516, P,0.05), for the Notobasis-and Onopordum-associated host races, respectively. The similarity in morphology of the male terminalia, however, does not necessarily mean that the two races can interbreed; taking into consideration that successful mating does not necessarily produce viable offspring.
The aculeus of the Onopordum-associated host race seems to be longer, a little wider and slightly blunter than the more pointed aculeus of the Notobasis-associated host race (Figures 8, 9), as is reflected by its mean length and width and the two angle measurements taken and as is cited in the literature (Knio et al. 2002). The aculeus apex angle differed by 1.2u between the two races; however, no statistical difference could be detected in this regard. The longer aculeus in the Onopordum-associated host race can be attributed to the difference in size between the hosts of the two races and can be considered a morphological adaptation of the females to their oviposition substrates. Since the flower heads of Onopordum illyricum are larger than flower heads of Notobasis syriaca, the Onopordum-associated females might need a longer aculeus to be able to penetrate deeper into the flower heads of their host and deposit their eggs between the bracts. Zwölfer (1972) suggested that a narrower and sharper aculeus was needed by a female fly in order to be able to deliver its eggs deeper into the larger flower heads it infests. However, the aculeus of the Onopordum-associated host race turned   out to be a little wider and less pointed than that of the Notobasis-associated host races, although the flower heads of Onopordum illyricum are larger in size. Another significant difference in the aculeus between the two races was in the location of the ventro-lateral grooves bearing three pairs of elongated sensilla as measured by ratio 1 and ratio 2 (Table V). In the Onopordum-associated females, the sensory ventro-lateral groove was located closer to the aculeus tip although the aculeus was longer (Figures 8, 9). This could also reflect the larger size of the host exploited by this race. The sensilla on the ventro-lateral grooves of the aculeus tip have been identified as mechano-chemosensilla and these are used by female tephritids to locate suitable hosts and to access the host quality and suitability for oviposition (Stoffolano 1989;Stoffolano and Yin 1987;Zacharuk et al. 1986).

Key to host races
This key allows the correct identification of at least 70% of specimens. The best way for identification is to check the plant host, examine sampled populations, and check the sample means for the diagnostic measurements mentioned in the following key.
Identifications (with 70% accuracy) could also be made by calculating the position of sampled specimens with respect to CV functions (or axes) I and II: