Research Article |
Corresponding author: Jure Jugovic ( jure.jugovic@zrs.upr.si ) Academic editor: Zdenek Fric
© 2014 Jure Jugovic, Toni Koren.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Jugovic J, Koren T (2014) Wing pattern morphology of three closely related Melitaea (Lepidoptera, Nymphalidae) species reveals highly inaccurate external morphology-based species identification. Nota Lepidopterologica 37(1): 75-90. https://doi.org/10.3897/nl.37.7966
|
Wing morphology of the three closely related species of Melitaea – M. athalia (Rottemburg, 1775), M. aurelia (Nickerl, 1850) and M. britomartis Assmann, 1847 – co-occurring in the Balkans (SE Europe) was investigated in detail through visual inspection, morphometric analysis and multivariate statistical analysis. Results are compared to recent phylogenetic studies, searching for concordant patterns and discrepancies between the two approaches. The morphology of the genitalic structures is also compared with the results of the other two approaches. The main conclusions are as follows: (1) small albeit significant differences in wing morphology exist among the three species and (2) while the structure of male genitalia and phylogenetic position of the three species are concordant, they are (3) in discordance with the wing morphology. The present study represents another example where identification based on external morphology would lead to highly unreliable determinations, hence identification based on phylogenetic studies and/or genitalia is strongly recommended not only for the three studied species but also more broadly within the genus. Furthermore, we show that some of the characters generally used in the identification of these three Melitaea species should be avoided in future.
The genus Melitaea Fabricius, 1807 (Lepidoptera: Nymphalidae) as it is known today (
The external morphology of the three species is highly similar, rendering the accurate identification based on external morphology questionable (
Melitaea athalia is a trans-European species (except for south-western Europe, where Melitaea celadussa Fruhstorfer, 1910, a former subspecies of M. athalia occurs (
The ecology of the three species in the area of NW Balkans (Slovenia) is similar in terms of flight period, altitudinal distribution and habitat requirements including foodplant (
Morphological traits can reflect either historical isolation and/or local adaptation despite recurrent gene flow (
We wanted to check the concordance in the amount of morphological and genetic differences among the three closely related species. For this purpose a morphometric approach with subsequent multivariate statistical analysis was used in the M. athalia complex for the first time. We evaluated the following hypotheses: (1) the genetic divergence of the three species is accompanied by corresponding morphological differences exceeding the intra-species variability and (2) the level of accompanying morphological differences is concordant with the level of molecular distance between the taxa, meaning that M. aurelia should be the most distant in morphological space from the other two species.
Samples of the three species of Melitaea were collected over the last 40 years all over the NW Balkans (see
Separation of these three species is possible using male genitalia as follows (see also
Specimens used in this study were mounted and photographed on a millimeter grid from the same angle (90°) using a DSLR camera (Canon 450D). Subsequently, 25 metric characters were measured from the photographs using freeware ImageJ (
A photograph of Melitaea sp. with measured structures. Distances are denoted with full lines, surface measurements with dashed lines and angles with Greek alphabet. Left: upperside; right: underside. For explanation of characters’ abbreviations see
List of structures (with morphological characters, their explanation with measuring units) measured in three species of Melitaea.
No. | Character abbreviation | Description | Unit |
---|---|---|---|
Forewing | |||
1 | FWS | forewing surface | mm2 |
2 | FWCS | forewing cell surface | mm2 |
3 | FWL1 | forewing length 1 | mm |
4 | FWL2 | forewing length 2 | mm |
5 | FWH | forewing height | mm |
6 | α | angle at forewing apex | ° |
7 | A1 | distance A1 | mm |
8 | A2 | distance A2 | mm |
9 | A3 | distance A3 | mm |
10 | A4 | distance A4 | mm |
11 | FWDS | forewing dot surface | mm2 |
12 | FWLS | forewing lunule surface | mm2 |
13 | FWLL | forewing lunule length | mm |
14 | FWLH | forewing lunule height | mm |
Hindwing | |||
15 | HWS | hindwing surface | mm2 |
16 | HWCS | hindwing cell surface | mm2 |
17 | HWL1 | hindwing length 1 | mm |
18 | HWL2 | hindwing length 2 | mm |
19 | HWL3 | hindwing length 3 | mm |
20 | ß | angle at hindwing apex | ° |
21 | B1 | distance B1 | mm |
22 | B2 | distance B2 | mm |
23 | HWLS | hindwing lunule surface | mm2 |
24 | HWLL | hindwing lunule length | mm |
25 | HWLH | hindwing lunule height | mm |
Categorical character | |||
26 | Marginal line colour in contrast to lunular colour | (1 = equal; 1.5 = slightly darker; 2 = darker; 3 = much darker) | |
Ratios | |||
27 | FWCS/FWS | ||
28 | FWL2/FWL1 | ||
29 | FWH/FWL1 | ||
30 | A2/A1 | ||
31 | A3/A1 | ||
32 | FWDS/FWCS | ||
33 | FWDS/FWS | ||
34 | FWLH/FWLL | ||
35 | FWLS/FWS | ||
36 | FWLS/FWCS | ||
37 | ß / α | ||
38 | HWCS/HWS | ||
39 | HWL1/HWL2 | ||
40 | HWL3/HWL1 | ||
41 | B2/B1 | ||
42 | HWLS/HWS | ||
43 | HWLS/HWCS | ||
44 | HWLH/HWLL |
Since in our samples males were prevalent and not all females could be reliably identified to a species, females were excluded from subsequent statistical analyses. With this approach, we also avoided the influence of possible sexual dimorphism. Only for the analysis of the marginal line colour were a few unambiguously identified females added to the sample (no sexual dimorphism was noticed in this character). We also had to remove from the analysis three aberrant individuals which lacked measuring points for some characters, hence they are commented on separately.
For each species, the Kolmogorov-Smirnov test (at p = 0.05) and Normal Q-Q plots were used to examine the normality of the data distribution, and the homogeneity of variances was evaluated visually through scatterplots. The multivariate analysis of variance (MANOVA, at p = 0.001) was used to test for significant differences between species. One-way analysis of variance (ANOVA) was used to assess the variation within a species for each character, significant variation in a character being accepted if the between species variation was significant at p < 0.001. Pearson’s correlation coefficients (r) were computed to evaluate the extent to which each character contributes unique information; only one character was chosen to represent a pair or a group of characters where |r| > 0.9. The Durbin-Watson test (at p = 0.05 and p = 0.01) was applied to test for possible spatial (latitude, longitude, altitude) and temporal (year of collection) autocorrelations of morphometric data (
Multivariate Principal Component Analysis (PCA) was used to identify the structure of our data i.e. to detect the possible influence of the species specific characteristics. Also, Discriminant Function Analysis (DFA) was carried out to examine possible separation of the three species. In DFA, the contribution of each species was weighted according to its sample size (number of specimens). Post-hoc Games-Howell and Bonferonni tests were performed to assess the rates of morphological divergences between pairs of species in details. The analyses were performed using Microsoft Excell (2010), SPSS 14.0 for Windows (2005) (
The multivariate analysis of variance (MANOVA) showed significant differences between species (p < 0.001). Out of 43 morphometric characters, 12 characters (without a single ratio) were selected after Kolmogorov-Smirnov (p < 0.05), ANOVA (p < 0.001) and Pearson correlation (|r| < 0.9) tests. No spatial or temporal autocorrelation was detected in these characters neither at p = 0.01 nor at p = 0.05 (Durbin-Watson test). Statistic description of selected 12 characters (mean, standard error, 12th percentile, 88th percentile and extreme values) is presented in
The principal result of the PCA (
Statistical description of twelve metric characters in males of three species of the genus Melitaea that were used in subsequent Principal component and Discriminant Function analyses. Upper row: average ± std. error; lower row: (min), 12th percentile–88th percentile (max).
M. athalia | M. aurelia | M. britomartis | ||
---|---|---|---|---|
forewing height | FWH | 11.62 ± 0.15 | 10.25 ± 0.16 | 10.73 ± 0.17 |
(9.27) 10.50–12.79 (13.56) | (8.93) 9.15–11.46 (12.23) | (9.19) 9.63–11.51 (12.78) | ||
forewing lunule surface | FWLS | 3.82 ± 0.10 | 2.82 ± 0.10 | 3.09 ± 0.18 |
(2.01) 3.07–4.79 (4.98) | (1.82) 1.95–3.55 (4.21) | (1.74) 2.19–4.29 (5.28) | ||
forewing lunule height | FWLH | 1.69 ± 0.03 | 1.45 ± 0.03 | 1.56 ± 0.03 |
(1.21) 1.50–1.99 (2.09) | (1.11) 1.24–1.61 (1.66) | (1.33) 1.36–1.76 (1.79) | ||
hindwing surface | HWS | 138.70 ± 4.60 (96.40) 106.64–171.47 (181.71) |
109.55 ± 20.50 (81.09) 89.78–144.80 (153.48) |
120.20 ± 4.40 (93.94) 103.09–161.06 (170.21) |
distance B1 | B1 | 3.86 ± 0.09 | 3.21± 0.08 | 3.50 ± 0.09 |
(2.76) 3.09–4.67 (5.10) | (2.39) 2.63–3.76 (4.06) | (2.63) 3.02–4.09 (4.57) | ||
hindwing cell surface | HWCS | 9.23 ± 0.24 | 7.61 ± 0.26 | 7.94 ± 0.26 |
(6.39) 7.35–11.42 (12.66) | (5.17) 5.75–9.42 (10.85) | (6.30) 6.60–9.44 (12.16) | ||
hindwing length 1 | HWL1 | 8.53 ± 0.14 | 7.54 ± 0.18 | 7.90 ± 0.16 |
(6.48) 7.52–9.88 (10.15) | (5.58) 6.45–8.78 (10.20) | (6.44) 7.05–8.96 (9.40) | ||
hindwing lunule surface | HWLS | 4.29 ± 0.15 | 3.39 ± 0.14 | 3.58 ± 0.20 |
(2.52) 3.14–5.32 (6.75) | (2.08) 2.65–4.57 (5.00) | (2.44) 2.53–4.82 (6.26) | ||
hindwing lunule length | HWLL | 2.48 ± 0.05 | 2.27 ± 0.05 | 2.20 ± 0.06 |
(1.73) 2.02–2.85 (3.24) | (1.68) 1.95–2.62 (2.79) | (1.73) 1.81–2.49 (2.82) | ||
hindwing lunule height | HWLH | 1.95 ± 0.04 | 1.71 ± 0.03 | 1.80 ± 0.05 |
(1.44) 1.69–2.34 (2.70) | (1.28) 1.48–1.93 (2.16) | (1.46) 1.55–2.08 (2.33) | ||
distance A4 | A4 | 1.84 ± 0.05 | 1.59 ± 0.05 | 1.59 ± 0.05 |
(1.09) 1.39–2.14 (2.49) | (1.11) 1.32–1.92 (2.18) | (1.13) 1.30–1.83 (2.01) | ||
forewing lunule length | FWLL | 2.46 ± 0.05 | 2.13 ± 0.05 | 2.19 ± 0.09 |
(1.63) 1.98–2.80 (3.04) | (1.52) 1.83–2.40 (2.56) | (1.56) 1.59–2.75 (3.32) |
In the DFA (
Variation in four morphometric characters in M. athalia, M. aurelia and M. britomartis: forewing height (FWH), forewing lunule surface (FWLS), forewing lunule height (FWLH) and hindwing surface (HWS). Statistically significant differences between pairs of species (Bonferroni test) are shown with capital (p < 0.001) and small (0.001 < p < 0.05) letters. When only a trend in differences between species is shown (0.05 < p < 0.1), small letters with apostrophe are used.
Contributions of 12 metric characters (MC) to discrimination along both discriminant functions (DF). Discriminant function analysis run on 95 males of three species of Melitaea. Asterisk (*) denotes the highest absolute correlation between the character and DF.
MC | (symbol) | DF 1 | DF 2 |
---|---|---|---|
forewing height | (FWH) | 0.851* | –0.002 |
forewing lunule surface | (FWLS) | 0.810* | 0.110 |
forewing lunule height | (FWLH) | 0.791* | –0.129 |
hindwing surface | (HWS) | 0.787* | –0.022 |
distance B1 | (B1) | 0.680* | –0.120 |
hindwing cell surface | (HWCS) | 0.651* | 0.158 |
hindwing length 1 | (HWL1) | 0.624* | –0.013 |
hindwing lunule surface | (HWLS) | 0.575* | 0.128 |
forewing lunule length | (FWLL) | 0.571* | 0.147 |
hindwing lunule height | (HWLH) | 0.562* | –0.032 |
distance A4 | (A4) | 0.533* | 0.285 |
hindwing lunule length | (HWLL) | 0.455 | 0.476* |
Although the DFA aims to find the differences between the a priori defined groups, the misclassification rates were high, especially when cross-validation process was employed (
In addition to the characters used in the DFA, two characters frequently mentioned in the literature as diagnostic for the recognition of (some of) Melitaea species should be mentioned. For the characters like the spacing of the marginal, submarginal, postdiscal and discal line on the forewing upperside, the colour of the submarginal spots on the hindwing underside, the coloration of hairs on the palps as well as that of the marginal line in contrast to the lunular colour on the underside of the hindwing, no consistency with species attribution was found in our examination. For the latter, slight differences were found between the three species; however, species could not be identified with certainty using the lunular coloration on the hindwing underside due to the high intraspecific variability of this character. The coloration of the marginal line is in general equal to the coloration of the lunules in M. athalia, much darker in M. britomartis and only slightly darker in M. aurelia. Nevertheless, all of these categories were noticed in each species (
The wing pattern and coloration show high intra- and interspecies variability. While in some species dark melanistic forms are common (e.g. in Melitaea britomartis ssp. michielii from Slovenian Karst and surroundings (
The genitalia structures (
All analysis conducted in this study indicates the important differences among the three species in the size of most of the measured characters, however, only in average values. Due to high variability of these characters, the clinal variation among species has been noticed (see the results of multivariate analyses). According to literature (e.g.
The allozyme polymorphism of these three species was studied in samples from the Carpathian basin, and revealed that M. britomartis and M. aurelia are more closely related, while M. athalia appears to be a further relative (
In contrast to the allozyme study, the phylogenetic analysis used as a framework for the explanation of our results (
No external characters were proven to be reliable for species delimitation even though the coloration of the marginal line of the hindwing underside shows the trend towards the correct identification. Although the majority of specimens correspond to the character states given in literature for a particular species, all three presumably species-specific categories of this character (
The high variability of these species is further shown in their qualitative characters, for example in wing pattern and coloration. In M. athalia three out of more than 120 sampled specimens had a very peculiar coloration, with no visible marginal lines. Albeit unrecognized, we believe that differences in (some of) local conditions clearly show the importance of environmental factors for the wing morphology. This further demonstrates the importance of genitalia or phylogeny based identification in the herein investigated Melitaea species. Moreover, the latter two approaches should have a significant advantage also for the identification of (some) other species of the genus that should undergo thorough revision in the future (cf.
The morphometric analysis of three closely related Melitaea species (M. athalia, M. aurelia and M. britomartis) revealed the following: (1) small albeit significant differences in wing morphology exist among the three species; (2) only characters describing the size of the species with no characters describing their wing shapes (ratios) were statistically important for the separation of the species; (3) whereas the structure of the male genitalia and the phylogenetic position of the three species are concordant, the sequence of phylogenetic splits is not reflected in the rate of external morphological differences among them; (4) our study represents another example where external morphology based identification would lead to highly unreliable determinations, hence a use of genitalia based identifications is strongly recommended.
We are grateful to Ivan Jugovic (Kranj, Slovenia) for data from his collection, useful discussion and help in identification. We also thank Barbara Zakšek (Voličina, Slovenia) for providing us with the literature on genitalia preparation, and Elena Varljen Bužan (Koper, Slovenia) for valuable comments regarding the molecular background. We are thankful to Roman Luštrik (Ljubljana, Slovenia) for valuable help with statistical analysis. The referees and the editor contributed greatly to the improvement of the manuscript.
Three categories of marginal line colour in contrast to lunular colour: 1=equal (a); 2=darker (b); 3=much darker (c) were used. In some cases when decision between categories 1 and 2 was hard (i.e. when the two authors disagree on whether the category should be assigned to category 1 or 2), value 1.5 was used.