Molecular phylogeny of the genus Lythria and description of the male genitalia of L . venustata ( Lepidoptera : Geometridae : Sterrhinae )

We present a molecular phylogeny incorporating all five species of the Palaearctic geometrid genus Lythria, based on a 2810-bp combined data matrix comprising the full sequence of the mitochondrial gene cytochrome oxidase subunit one (COI) and fragments of the nuclear genes elongation factor 1 alpha (EF-1 ) and wingless (wgl). L. venustata, which was recently rediscovered from Kazakhstan, is shown to be sister taxon to all other members of the genus. The remaining species within the genus form two pairs of sister species: L. purpuraria groups together with L. plumularia, and L. cruentaria with L. sanguinaria. The phylogeny is well supported by characters of the male genitalia of all Lythria species. In addition to the molecular phylogeny of the genus Lythria, we illustrate the external appearance of L. venustata for the first time and describe the anatomy of its male genitalia.

Lythria comprises five species, all of which occur in the Palaearctic (Scoble, 1999;Viidalepp, in press).Two of them, L. purpuraria (Linnaeus, 1758) and L. cruentaria (Hufnagel, 1767), are widely distributed, from Western Europe to Central Asia (Häuser, 2001;Viidalepp, in press).Historically, there has been a lot of confusion between L. purpuraria and L. cruentaria, since both species exhibit considerable variation in size and wing pattern.Separation of these species on the basis of external morphology is therefore often difficult.Consequently, these taxa were usually treated as a single species, L. purpuraria (but see, e.g., Borkhausen, 1794;Laspeyres, 1803;Duponchel, 1830), until Prout (1914) noted that the taxon in fact comprises two species that are easily distinguishable on the basis of genital characteristics.Subsequently, the genitalia of L. purpuraria and L. cruentaria [the synonym L. purpurata (Linnaeus, 1761) being widely used in the older literature] have been illustrated repeatedly (e.g., Zerny, 1916;B eszy ski, 1965;Häuser, 2001), and differences in the distribution and life-history traits of these species have been well documented (e.g., Koch, 1984;Müller, 1996;Häuser, 2001;Viidalepp, 1996).
The third species, L. sanguinaria (Duponchel, 1842), occurring only in the Iberian Peninsula and in southern France (Prout, 1914;Viidalepp, in press) was treated as a separate species for a long time (see, e.g., Staudinger & Rebel, 1901;Spuler, 1903-10).However, both Prout (1914) and Zerny (1916) found that the differences between the genitalia of L. cruentaria and L. sanguinaria were insufficient to regard them as distinct species.Since their studies, the latter has been treated as a subspecies of L. cruentaria (Prout, 1937;Herbulot, 1962;Leraut, 1997;Scoble, 1999).Recently, Viidalepp (in press) showed that there are limited but consistent differences between the taxa in the wing pattern and both in male and female genitalia.He therefore proposed reinstating species status to L. sanguinaria.A subsequent study by Õunap et al. (2008) revealed that the level of genetic divergence between L. sanguinaria and L. cruentaria is almost as large as that between L. cruentaria and L. purpuraria, thus implicitly supporting the view of Viidalepp (in press) that L. sanguinaria is a distinct species.
The fourth species, L. plumularia (Freyer, 1831), is an endemic of the European High Alps (Prout, 1914;Viidalepp, in press).Though L. plumularia is also extremely variable in appearance (Gradl, 1938), it nevertheless is morphologically clearly distinct from the taxa discussed above.Confusion between L. plumularia and other European taxa is therefore highly unlikely.The genitalia of this species have recently been illustrated for the first time (Vasilenko, 2009).
The fifth species, L. venustata Staudinger, 1882, is the least well known.For more than a century, only the holotype (male) was known.The specimen was collected from Zaisan in eastern Kazakhstan and is currently housed at the Museum für Naturkunde, Humboldt-Universität, Berlin, Germany (Fig. 1).Although L. venustata was included in the major monographs of Palaearctic Geometridae by Staudinger & Rebel (1901) and Prout (1914), these did not include illustrations of the species; neither have illustrations appeared elsewhere.Recently, two additional L. venustata males were collected, one from a lowland semidesert in western Kazakhstan in 2006 (Fig. 2), another from Tarbagatai mountains in northeastern Kazakhstan in 2007 (Vasilenko, 2009).On the basis of the latter, Vasilenko (2009) illustrated the male genitalia of L. venustata for the first time.Unfortunately, he did not spread the heavily sclerotized valvae.This resulted in misinterpretations regarding the development and position of some genitalic structures.
The complete phylogeny of Lythria has hitherto not been studied.However, on the basis of external morphological characters, the moths fall into two groups.First, L. plumularia and L. venustata, which have ochreousyellow forewings with three heavily contrasting dark transverse fasciae, are assumed to be closely related (Staudinger, 1882;Prout, 1914;Vasilenko, 2009).Second, L. purpuraria, L. cruentaria and L. sanguinaria, which typically have greenish-yellow forewings with two (L.purpuraria) or three (L.cruentaria, L. sanguinaria) purple red transverse lines, are believed to form another group.Within the latter group, both external characters and the male and female genitalia suggest that L. cruentaria and L. sanguinaria represent sister taxa (Viidalepp, in press).Surprisingly, Õunap et al. (2008) found L. sanguinaria and L. purpuraria to be very closely related, with L. cruentaria appearing as sister taxon to them.However, the authors of this study noted that their intrageneric phylogeny may have been influenced by homoplasy in their data matrix (Õunap et al., 2008).
Thus, the phylogeny of Lythria including all known species (neither L. plumularia nor L. venustata have been incorporated into previous molecular phylogenetic analyses) still remains to be determined.For this purpose, the full sequence of mitochondrial gene cytochrome oxidase subunit 1 (COI) and partial sequences of two nuclear genes [elongation factor 1 alpha (EF-1 ) and wingless (wgl)] were sequenced for all five Lythria species and used to construct the phylogeny of the genus.The male  genitalia of L. venustata were illustrated for the second time and their anatomy discussed and compared to the male genitalia of the other Lythria species.

Moths
In total, eight species were analyzed (Table 1).In addition to the five Lythria species, Rhodometra sacraria (Linnaeus, 1767) from the tribe Rhodometrini, which is currently believed to be the closest relative to the Lythriini (Õunap et al., 2008), was also included in the analysis, while Timandra griseata Petersen, 1902 andT. comae Schmidt, 1931 were used as outgroups.Details of collecting data and the depositories of voucher specimens are presented in Table 1.

DNA extraction, amplification and sequencing of gene fragments
Total genomic DNA was extracted from the anterior abdominal segments (for details, see Õunap et al., 2005) or from two or three legs of individual specimens.Purification of total genomic DNA was performed using a High Pure PCR Template Preparation Kit (Roche Diagnostics GmbH, Mannheim, Germany) following the manufacturer's instructions for isolating nucleic acids from mammalian tissue, except that the first incubation step was 55°C for 2-3 h rather than for 1 h.
Two partially overlapping fragments were amplified using two main primer pairs (cov1f+nan and v1+4r2, Table 2) to obtain the full sequence of the COI gene.However, PCR amplification using these primers failed in some taxa.Therefore, additional primer pairs, cov1f+cov1r, ron+nan, v1+v2 and 4f2+4r2 (Table 2), were used to amplify COI in four shorter partially overlapping fragments.A partial sequence of the wgl gene was amplified using primers LepWG1+LepWG3, while a fragment of EF-1 was amplified using primers ef44 and efrcM4 (Table 2).As was the case for COI, the latter primers did not perform well in some taxa, so additional primer combinations, LepEF-1f+LepEF-1r, EF51.9+Niina2 and LepEF-2f+Niina2 (Table 2), were used to amplify EF-1 in two shorter partially overlapping fragments.PCR was performed in a total volume of 20 µl, with the reaction mixture containing 1X BD Advantage 2 PCR buffer, 1U BD Advantage 2 Polymerase mix (BD Biosci-ences, San Jose, USA), 0.2 mM dNTP (Fermentas, Vilnius, Lithuania), 4 pmol of primers and 20-80 ng of purified genomic DNA.
PCR was performed using a T1 thermocycler (Biometra, Göttingen, Germany) with the following cycling parameters: A 2 min denaturing step at 94°C followed by 35-40 cycles of 30 s at 94°C, 30 s at various temperatures, depending on primers (Table 2) and 60-75 s at 68°C with a subsequent 7-min final extension at 68°C PCR products were visualised on a 1.6% agarose gel, and 10 µl of the PCR solution was treated with shrimp alkaline phosphatase and exonuclease I (Fermentas, Vilnius, Lithuania).One unit of each enzyme was added to the PCR solution, which was incubated for 27 min at 37°C followed by 15 min inactivation at 80°C.
DNA cycle sequencing was performed in a total volume of 10 µl using the DYEnamic ET Terminator Cycle Sequencing Kit (GE Healthcare, Chalfont St Giles, United Kingdom).Cycling conditions were: 33 cycles of 20 s at 95°C, 20 s at various temperatures, depending on the primer (Table 2), and 60 s at 60°C.Both DNA strands were sequenced using 5 pmol of primers, and sequences were resolved using an ABI PRISM 377 automated sequencer (Applied Biosystems, Foster City, USA).

Phylogenetic analysis
Consensus sequences were created in Consed (Gordon et al., 1998), using sequence data from both DNA strands.Sequences were double-checked by eye, edited in BioEdit (Hall, 1999) and aligned in ClustalW using the default settings (Thompson et al., 1994).Homogeneities between all gene sequences were calculated using the partition homogeneity test in PAUP*4.0b10(Swofford, 1998).This test was performed with the random addition heuristic search option, using 1,000 replicates.As no incongruence between different genes was found, the whole data matrix was subsequently analysed as a single entity.Modeltest 3.06 (Posada & Crandall, 1998) was used to search for the optimal model of DNA substitution.Bayesian phylogenetic inference, maximum likelihood (ML), maximum parsimony (MP) and neighbour-joining (NJ) approaches were all used to evaluate the robustness of the phylogenetic analysis.
The GTR+I+ model, selected by Modeltest, was used for NJ and ML analysis in PAUP.Branch supports were assessed using 1,000 bootstrap replicates.MP analysis with simple addition of  1. Information on the specimens used in molecular analyses.Collecting site (ESP -Spain, EST -Estonia, KAZ -Kazakhstan, SUI -Switzerland) and date, collector's name, GenBank accession numbers for COI, EF-1 and wingless sequences and depositories of the studied specimens are indicated.Authorships of the sequences downloaded from GenBank are indicated as follows: ¤ - Knölke et al., 2005;# -Õunap et al., 2008.taxa was also performed in PAUP and resulted in a single most parsimonious tree.Branch supports for this tree were assessed using 1,000 bootstrap replicates, with 10 heuristic searches and simple addition of taxa used for each replicate.ML, NJ and MP trees were visualised in TreeView 1.6.6 (Page, 1996).
Bayesian phylogenetic analysis implementing the GTR+I+ model was performed using MrBayes 3.1 (Ronquist & Huelsenbeck, 2003).Four simultaneous Markov chains (one cold and three heated) were run for 2,000,000 generations, with trees sampled every 1,000 generations.Likelihood values were inspected, and the first 500 sampled trees were discarded as "burn-in".To estimate posterior probabilities of recovered branches, a 50% majority rule was applied.Phylograms were created as average-branch-length consensus trees and visualised in TreeView 1.6.6.

RESULTS AND DISCUSSION
All three gene fragments were successfully sequenced for all analysed specimens.The total length of the COI gene was 1536 bp for all Lythria species and R. sacraria, while both Timandra species had an 8-bp insertion (AAAAATAT) between the COI positions 1531 and 1532.The total length of the COI gene was therefore 1533 bp for the Timandra species, as the 8-bp insertion resulted in the formation of a TAA stop codon in COI positions 1531-1533.The length of the successfully sequenced fragment of EF-1 gene was 883 bp, while the length of the successfully sequenced wgl fragment was 383 bp.No insertions or deletions were identified in either of these nuclear genes.The total length of the combined molecular data matrix was 2810 bp.
All phylogenetic analyses resulted in well-resolved trees that exhibited identical topology and maximal or near-maximal indices of support for all nodes (Fig. 3).R. sacraria was placed as the closest relative to the genus Lythria, while L. venustata appeared as sister taxon to all other Lythria species.Four remaining Lythria species formed two clades of sister taxa: L. purpuraria was placed together with L. plumularia and L. cruentaria with L. sanguinaria (Fig. 3).
Lythriini were postulated as the sister group to the Rhodometrini by Õunap et al. (2008) on the basis of molecular data.In addition, they listed a few morphological and ecological characters that link Lythriini with Sterrhinae in general, and support the tribe's position as sister to the Rhodometrini.Similarly, Sihvonen & Kaila (2004) found a close association between Lythria and Rhodometra Meyrick, 1892.This confirms the finding of Õunap et al. (2008) that the genus belongs to the Sterrhinae.A synapomorphy for the Timandrini lineage sensu Sihvonen & Kaila (2004), the absence of sensilla on the ventral surface of the male flagellomere, is also characteristic of Lythria.
Additionally the following four synapomorphies, from a total of seven found to be characteristic of the Rhodometrini by Sihvonen & Kaila (2004), are also found in Viidalepp et al., 2007 57°C 58-61°C Reverse EF-1 , 3' half  the Lythriini: terminal line of the forewings absent; hindwing veins Sc and R1 fused for a relatively long distance (also noted by Õunap et al., 2008); uncus naked; and arms of transtilla do not meet dorsally.The remaining three characters (Sihvonen & Kaila, 2004), however, are absent in Lythriini: reddish straight line from the forewing apex; forewing ground colour bright yellow; and spinose apex of sacculus.The distinctive shape of ansa, which was listed as an additional character of Rhodometrini by Sihvonen & Kaila (2004), is not found in Lythriini.The central flap of the ansa is strongly dilated unilaterally in Rhodometrini but only slightly broadened in Lythriini.
Additional common characteristics that support the proposed sister relationship between the Rhodometrini and Lythriini are the following: Large vinculum (resembling that of Scopulini; see Sihvonen & Kaila, 2004), weak tegumen, weak juxta, absence of saccus and the presence of a pair of pad-like socii on the base of the uncus in Casilda Agenjo, 1952 and Lythria.Socii in Rhodometra are, however, as strongly sclerotized as tegumen is.
On the basis of male genital morphology it is possible to distinguish three groups within Lythria.First, L. purpuraria and L. plumularia, though superficially very different, share the presence of two cornuti on the vesica, the presence of a pair of long postero-lateral extensions on the tegumen and a short sack-like, globular valvula attached to the roughly triangular valva (Figs 4-5).In both species, the valvula is approximately as long as it is broad.Second, L. cruentaria and L. sanguinaria have short, roughly rectangular valvae with long membranous valvulae, short postero-lateral extensions on the tegumen and the presence of one cornutus on the vesica.In these species, the valvula is approximately four times longer than it is broad, though it is clearly more slender in L. sanguinaria than in L. cruentaria (Figs 6-7).Third, although L. venustata is externally quite similar to L. plumularia [which obviously led Staudinger (1882), Prout (1914) and Vasilenko (2009) to the conclusion that these species are closely related], its male genitalia are strikingly different from those of the other Lythria species.The valvae of L. venustata are distally bipartite, the valvulae are absent and the remnants of the socii are missing.The postero-lateral extensions of the tegumen are visible but are much shorter than those in other Lythria species.The rounded shape and massive sclerotization of the vinculum gives the genital armature a distinctive appearance (Fig. 8).Vasilenko (2009) described the male genitalia of  L. venustata as having long finger-like socii on the posterior edge of the tegumen.This misinterpretation apparently has happened because he did not spread the tough and strongly sclerotized valvae.These "socii" are actually the projections of the sacculi.Furthermore, in Vasilenko's (2009) interpretation L. venustata also lacks an uncus and the most distal part of the genitalic capsule is instead the anellus.In contrast, we regard the most distal part of the genitalic capsule as a weakly sclerotized uncus, as seen in the lateral view of the male genitalia (Fig. 8).The aedeagus of L. venustata is more slender than that of L. purpuraria and L. plumularia, but resembles the slim aedeagus of L. cruentaria and L. sanguinaria.In common with L. cruentaria and L. sanguinaria, L. venustata has one cornutus on the vesica; however, in contrast to all the other Lythria species, L. venustata has a well developed juxta.
Thus, the results of the molecular phylogenetic analysis are in concordance with those derived from examination of the genital morphology of Lythria males.L. venustata, whose genital morphology is dissimilar to all other Lythria species, was placed phylogenetically as sister taxon to all other species in the genus (Fig. 3).Interestingly, such a position receives additional support from morphological similarities shared by L. venustata and R. sacraria.Specifically, these two taxa both exhibit a rounded vinculum, which contrasts with the strongly emarginated vinculum of the other Lythria species.Moreover, the valva of R. sacraria, which is slightly bilobed in shape, also somewhat resembles that of L. venustata (see Hausmann, 2004), while the very short postero-lateral extensions on the tegumen of L. venustata appear intermediate between Rhodometrini (no such extensions) and other Lythria species (well-developed extensions).Sister-taxa relationships between L. purpuraria and L. plumularia, and L. cruentaria and L. sanguinaria, respectively (Fig. 3), revealed by the phylogenetic analysis, are also supported by the examination of genital morphology (Figs 4-7).On the other hand, the preliminary subdivision of Lythria into two groups on the basis of wing colour (see above) is not supported by the results of the phylogenetic analysis.Though both L. venustata and L. plumularia have ochreous-yellow forewings in contrast to the other three species, they do not pair as sister taxa (Fig. 3).Earlier hypotheses by Staudinger (1882), Prout (1914) and Vasilenko (2009), which assumed close affinities between L. venustata and L. plumularia, are therefore rejected, and their superficial similarity is to be regarded as a plesiomorphic or homoplasic condition.Similarly, the sister-taxa relationship between L. sanguinaria and L. purpuraria proposed by Õunap et al. (2008) must be rejected.
In summary, L. venustata appears to be clearly distinct from other Lythria species on the basis of both molecular phylogenetic and morphological evidence.Considering the extent of the differences between the male genitalia of L. venustata and other Lythria species, it may be appropriate to move the former into a genus of its own, as already pointed out by Vasilenko (2009).However, as a female L. venustata has yet to be described, we prefer not  to take this step.Nonetheless, we highlight this as a point for consideration in future studies.
5'-CCT GGA AGG ACT CCA CRC ACA G-3Details of primers and PCR and cycle sequencing reactions.PCR and CS indicate the temperature during the annealing step of PCR amplification and cycle sequencing, respectively.Primer Verdi4 was used only as an internal primer for sequencing the 5' half of EF-1 .

Fig. 3 .
Fig. 3. Bayesian phylogenetic tree (GTR+I+ model) of the genus Lythria, based on a 2810 bp combined sequence of COI, EF-1 and wgl sequences.Bayesian posterior probabilities are given above the branches, bootstrap support for the ML/NJ/MP trees, which exhibited identical topology, are presented below the branches.