Contributions to Zoology, 84 (4) – 2015Zorica Nedeljković; Jelena Ačanski; Mihajla Đan; Dragana Obreht-Vidaković; Antonio Ricarte; Ante Vujić: An integrated approach to delimiting species borders in the genus Chrysotoxum Meigen, 1803 (Diptera: Syrphidae), with description of two new species
Results

To refer to this article use this url: http://ctoz.nl/vol84/nr04/a02

Geometric morphometric evidence

Wing size and shape

The ANOVA of wing centroid size showed significant differences between C. vernale and C. montanum sp. nov. (males: F1,117 = 8.99; P < 0.00331, Tukey’s post hoc test p < 0.01227; females: F1,117 = 81.12; p < 0.00000, Tukey’s post hoc test p < 0.00010). ANOVA test and Tukey’s post hoc test showed significant differences in wing size between C. orthostylum sp. nov. and C. vernale (F1,182 = 6.69; p < 0.01045, Tukey’s post hoc test p < 0.00969) and C. orthostylum sp. nov. and C. montanum sp. nov. (F1,166 = 30.12; p < 0.00000, Tukey’s post hoc test p < 0.00000). Chrysotoxum vernale has larger wings than C. montanum sp. nov. and C. orthostylum sp. nov. (Fig. 4B).

FIG2

Fig. 4. Wing shape differences among analysed species. A, scatter plot of individual scores of the two first canonical axes; B, boxplot of mean centroid size.

Differences in wing shape between C. vernale and C. montanum sp. nov. were highly significant using MANOVA (males: Wilks’ Lambda = 0.25314; F28,95 = 10.01020; p < 0.00000; females: Wilks’ Lambda = 04.40454; F28,90 = 4.73120; p < 0.00000). Highly significant differences in wing shape were also shown between C. orthostylum sp. nov. and C. vernale (Wilks’ Lambda = 0.15640; F28,155 = 29.85972; p < 0.00000), and C. orthostylum sp. nov. and C. montanum sp. nov. (Wilks’ Lambda = 0.16331; F28,139 = 25.43285; p < 0.00000).

DA correctly classified the material studied in accordance with the three defined species. The overall classification success of the DA was 92.95% which indicates wing shape is a reliable predictor of interspecific discrimination. Of the 249 digitalisations, only 21 were misclassified. Canonical variate analysis conducted on shape variables gave two highly significant axes. The first canonical axis (CV1), with 65.4 % of total variation, separated C. vernale from C. orthostylum sp. nov. (Wilks’ Lambda = 0.09515; χ2 = 662.18; p < 0.00000). The second canonical axis (CV2), with 34.6% of total variation, separated C. vernale and C. orthostylum sp. nov. from C. montanum sp. nov. (Wilks’ Lambda = 0.38388; χ2 = 269.52; p < 0.00000) (Fig. 4A). The superimposed outline drawings showing major wing deformations between C. vernale and C. montanum sp. nov. occur in the central and distal part of wing and are associated with landmarks 2, 4, 5 and 9-11 (Fig. 5A). Major differences in wing shape were found in the central part of the wing (landmarks 2, 7-11) of C. orthostylum sp. nov. and C. vernale (Fig. 5C), and in the central (landmarks 7, 8, 12 and 13) and apical parts in C. orthostylum sp. nov. and C. montanum sp. nov. (landmarks 4 and 5) (Fig. 5B). Chrysotoxum vernale has wider wings, while C. orthostylum sp. nov. longer wings.

FIG2

Fig. 5. Superimposed outline drawings showing wing shape differences between analysed species. A, Chrysotoxum vernale and Chrysotoxum montanum sp. nov.; B, Chrysotoxum orthostylum sp. nov. and Chrysotoxum montanum sp. nov.; C, Chrysotoxum orthostylum sp. nov. and Chrysotoxum vernale.

Surstylus shape

next section

Significant differences in surstylus shape between C. vernale and C. montanum sp. nov. were found with MANOVA (F1,91 = 32.3; p < 0.00099) and canonical analyses (axis 1: Wilks’ Lambda = 0.0425; χ2 = 199.02; p < 0.00099). Discriminant analysis of surstylus shape classified correctly C. vernale and C. montanum sp. nov. with 100% overall classification success. This high rate of overall classification success indicates that surstylus shape is a reliable predictor of interspecific discrimination. Representation of the thin-plate spline deformation of surstyli showed differences in surstylus shape between C. vernale and C. montanum sp. nov. (Fig. 6).

FIG2

Fig. 6. Superimposed outline drawings showing surstylus shape differences between Chrysotoxum montanum sp. nov. and Chrysotoxum vernale.

Ecological evidence

The PCA reduces the Bioclim variables (Table 1) to three PCs with eigenvalue ≥ 1.88% of the variation. Separation among species were statistically significant along PC1 and PC2 (ANOVA, PC1: F1,302 = 141.26; p < 0.00000, Fisher LSD post hoc test p < 0.00000; PC2: F1,302 = 155.27; p < 0.00000, Fisher LSD post hoc test p < 0.00000), but not along PC3 (F1,302 = 1.90; p < 0.168397). Because of this, PC3 was excluded in the interpretation of results. The precipitation-related PC1 explained 61% of the total variation, while the temperature- and altitude-related PC2 explained 17% of the total variation (Table 1). The scatter plot of PC1 against PC2 illustrated a clear niche separation between C. vernale and C. montanum sp. nov. Chrysotoxum vernale occurs across a wide temperature range, precipitation and altitude (Fig. 7), while C. montanum sp. nov. occurs in a much narrower range of temperatures. Chrysotoxum montanum sp. nov. occurs only at high altitudes with high precipitation, and its distribution is strongly influenced by the precipitation in the coldest and wettest quarters of the year (BIO19, BIO12, BIO16 and BIO13).

FIG2

Table 1. Principal component analysis (PCA) of 19 Bioclim variables plus altitude for C. montanum sp. nov. and C. vernale. Significant factor loadings are printed in bold.

FIG2

Fig. 7. Scatter plot of factor loadings of two PC scores showing the positions of Chrysotoxum montanum sp. nov. and Chrysotoxum vernale in the environmental space.