Systematics of the Neotropical genus Catharylla Zeller (Lepidoptera, Pyralidae s. l., Crambinae)

Abstract The Neotropical genus Catharylla Zeller, 1863 (type species: Crambus tenellus Zeller, 1839) is redescribed. Catharylla contiguella Zeller, 1872, C. interrupta Zeller, 1866 and Myelois sericina Zeller, 1881, included by Munroe (1995) in Catharylla, are moved to Argyria Hübner. Catharylla paulella Schaus, 1922 and C. tenellus (Zeller, 1839) are redescribed. Six new species are described by Léger and Landry: C. bijuga, C. chelicerata, C. coronata, C. gigantea, C. mayrabonillae and C. serrabonita. The phylogenetic relationships were investigated using morphological as well as molecular data (COI, wingless, EF-1α genes). The median and subterminal transverse lines of the forewing as well as the short anterior and posterior apophyses of the female genitalia are characteristic of the genus. The monophyly of Catharylla was recovered in all phylogenetic analyses of the molecular and the combined datasets, with three morphological apomorphies highlighted. Phylogenetic analyses of the morphology of the two sexes recovered three separate species groups within Catharylla: the chelicerata, the mayrabonillae, and the tenellus species groups. The possible position of Micrelephas Schaus, 1922 as sister to Catharylla, based on both morphological and molecular data, and the status of tribe Argyriini are discussed. The biogeographical data indicate that the chelicerata species group is restricted to the Guyanas and the Amazonian regions whereas the tenellus group is restricted to the Atlantic Forest in the South-Eastern part of Brazil. The mayrabonillae group is widespread from Costa Rica to South Bolivia with an allopatric distribution of the two species. COI barcode sequences indicate relatively strong divergence within C. bijuga, C. mayrabonillae, C. serrabonita and C. tenellus.


Introduction
Pyraloidea is one of the largest superfamilies of the order Lepidoptera. The monophyly of the group and that of its two main lineages, the Pyraliformes and Crambiformes (or Pyralidae and Crambidae, depending on authors), are well supported by morphological characters (Börner 1925, Hasenfuß 1960, Minet 1982 and molecular investigations (Regier et al. 2009(Regier et al. , 2012Mutanen et al. 2010). The abdominal tympanal organs represent a distinctive autapomorphy for the superfamily (Munroe and Solis 1998). The phallus attached medially to the juxta, the crambine-type tympanal organs, as well as the presence of a hair tuft on the dorsal hindwing cubital stem support the monophyly of subfamily Crambinae (Landry 1995). This is corroborated by Regier et al. (2012) based on sequences of several genes, but only three taxa and two of them Crambini. The only available, although partial, phylogenetic analysis of the Crambinae Latreille, 1810 is that of Landry (1995), which, however, did not include Catharylla. The relationships of Catharylla with other genera of Crambinae are not known, except for the placement of the genus in the Argyriini by Munroe (1995) together with Argyria Hübner, 1818, Vaxi Bleszynski, 1962 andUrola Walker, 1863. This placement was based on a suggestion by Bleszynski (1962: 12) that Catharylla may be closely related to Vaxi (Munroe 1995:161). Tribe Argyriini was thought to be monophyletic on the basis of the snow white color of the wings, the broad forewings, and the short labial palpi (Landry 1995, Munroe 1995, but its phylogenetic relationships remain unresolved. Stanislas Bleszynski investigated Catharylla, but he never published his findings, probably due to his accidental death in 1969. Catharylla moths are of medium size, with snow white to cream-colored wings with two ochreous to brown transverse lines on the forewings. Species of Catharylla occur in tropical Central and South America. Nothing is known on the immature stages and biology. The original description of Zeller (1863) gives little information and was restricted to external features. Munroe (1995) included five species in Catharylla (C. contiguella, C. interrupta, C. paulella, C. sericina, C. tenellus), but these share no evident common characteristics other than the snow white color of the wings.
In this work, Catharylla is revised using both morphological and molecular data. Phylogenetic relationships within Catharylla and with putatively related taxa as well as the distribution of each species along with biogeographical hypotheses are analysed.

Taxonomy
The types of the two species C. contiguella Zeller, 1872, andC. interrupta Zeller, 1866 could not be found. With the help of the descriptions and illustrations, they were excluded from Catharylla because of the forewing pattern, which is like that of Argyria Hübner, 1818, with only one median transverse line. For C. sericina Zeller, 1881, based on the description and a photograph of the type in the BMNH, the species was rejected from Catharylla based on the elongated forewing shape and the silvery white pattern without transverse lines. For C. paulella Schaus, 1922, a photograph of the habitus and the genitalia of the female type from the USNM allowed to find other specimens of the same species; the male and female were then associated based on wing pattern. For C. tenellus Zeller, 1839, a photograph of the habitus and the genitalia of the female type were available and the male of the species was associated based on wing pattern. For the descriptions, we followed the nomenclature and terminology used by Landry (1995), except for the use of the term phallus (see Kristensen 2003). New species were recognized based on major differences in male and female genitalia.
The following measurements were made with the use of an ocular micrometer: length of labial palpus (base of segment I to apex of segment III), diameter of eye (greatest vertical width), length of forewing (from base to apex), length of uncus (from tegumen-uncus junction to apex of uncus), length of tegumen connection (from tegumen arms connection medially to tegumen-uncus junction), length of papillae anales (dorso-ventral length), lengths of anterior and posterior apophyses (from base to apex).
Regarding the holotype data, the information was copied exactly as found on the labels with vertical slashes to express changes of lines. Abbreviations are spelled out in square brackets. We assume that the labels are rectangular and white, and that the text is in black ink, otherwise differences are indicated in brackets. Paratype data are reported by country in alphabetical order and the information is recorded without indication of line change. Collecting localities are reported as written on labels, with a question-mark when the locality could not be recovered. Dates and collectors' information were standardized and the latter placed in parentheses. The specimen depositories are reported with the use of the corresponding acronyms.

Biogeographical investigations
The coordinates of the localities were found using Google Earth (2011). The localities that were not registered in Google Earth were localized more or less precisely with the help of internet search engines or with gazeeters from the GEOnet Names Server (GNS) of the National Geospatial Intelligence Agency (http://earth-info.nga.mil/gns/ html/). The localities were reported on a text file (*.txt) and loaded on a map using DIVA-GIS 7.4.0.1 (Hijmans et al. 2011). Distances between localities were calculated with Google Earth. The localities and their coordinates are reported in Appendix I - Table 1. We refer to the provinces of Morrone (2006) in the biogeography section.

Material
The genes investigated are the mitochondrial COI gene (1474 bp) and the two nuclear genes wingless (353 bp) and EF-1α (679 bp). These genes show different rates of substitution through time, with COI >> wingless > EF-1α (Wahlberg and Wheat 2008). COI performs well at the species level and the two nuclear genes recover accurate phylogenies in deeper nodes of Lepidoptera (Wahlberg and Wheat 2008). Specimens used for molecular investigations are listed in Table 3, with the sequencing success for the different investigated genes. Specimens were preserved in 95% ethanol under cool conditions until molecular investigations, or were pinned and dried. Sequences for the additional species were obtained from GenBank (see Table 3).
For Catharylla specimens collected no more than twenty years ago, a leg was sent to the Canadian Centre for DNA Barcoding (CCDB) at the Biodiversity Institute of Ontario (BIO) in Guelph. The barcode sequences of Catharylla are reported in each species description. The protocol for DNA extraction is found in the supplementary material of Ivanova et al. (2006). The primers LepF1, LepR1 (Hebert et al. 2004) and MLepF1, MLepR1 (Hajibabaei et al. 2006) were used by the CCDB. The protocol for COI amplification is found at http://dev.ccdb.ca/docs/ CCDB_Amplification.pdf

DNA sequencing
DNA extractions were performed following the method of Knölke et al. (2005) and with the NucleoSpin Tissue kit by Magerey-Nagel according to the manufacturer's protocol. For alcohol preserved specimens, only tissue from the thorax was used for DNA extraction. For dried specimens, DNA was extracted from the abdomen or a leg, when the abdomen was missing.
The primers used are recorded in Table 4. Table 3.
List of the material used in the molecular work with voucher numbers, database of origin, collecting depository, and sequencing results for each gene. LEP references refer to the Lepidoptera DNA database of M. Nuss at the SMTD. BC MTD references refer to the barcode sequence voucher of the Barcoding Of Life Database (BOLD). HG references refer to the European Nucleotide Archive. Amplicon length (in basepairs) is given in brackets. The sequences with an asterisk were those used to build the datasets used in the phylogenetic analyses.  (1-353)*

HG793002
(1-626)* PCRs were performed using peqGOLD Taq DNA polymerase (PeqLab). In cases of weak or absent PCR result, a re-examination PCR was done using BIO-X-ACT Short Taq polymerase (Bioline). PCR protocols are given in Appendix II -Tables 1 and 2. Potential contamination was tested along with PCRs by control sample without DNA.
Success of gene amplification was evaluated by an electrophoresis with 1% agarose gel, subsequent gel dying with GelRed and analysis under ultraviolet light. PCR products were purified using ExoSAP-IT (USB Corporation).
Sequence PCR was done with the BigDye Terminator-Kit of Applied Biosystems. The amount of each product is reported in Appendix II - Table 3. The PCR programme is reported in the BigDye Terminator Sequencing Kit protocol. The sequences were obtained from the sample analysis on a 3130 Genetic Analyzer (Applied Biosystems).

Sequence analyses
Sequence alignment was done with BIOEDIT 7.1.3 (Hall 1999) and PhyDE (Phylogenetic Data Editor) 0.9971 (Müller et al. 2011). Nucleotide positions that proved ambiguous in the pherogram were coded according to the IUPAC nucleotide ambiguity code. Depending on the quality of the sequences, the 5' and 3' ends of the sequences were deleted in cases of high error rates for the nucleotide assignment from the phenogram. We used the programme DAMBE (Xia and Xie 2001) to generate the molecular data with third codon position deleted. The alignement was exported as fasta, nexus and phylip files. Sequences were deposited on European Nucleotide Archive, and can be accessed via the following links (for the COI and the nuclear sequences respectively): http://www.ebi.ac.uk/ena/data/view/HG793012-HG793020; http://www.ebi.ac.uk/ ena / data/view / HG793012 -HG793020

Genetic distances
Genetic distances between barcoding sequences of COI are given in Table 5. Distances were calculated using DAMBE. The GTR model was used as the substitution model.

Morphology-based analysis
The 21 characters are listed in Table 6. Characters 12, 16 and 17 were polarized into two sets of continuous values of ratios, with the limit selected subjectively where the gap between two groups of values appeared to be the largest. Phylogenetic analyses were run under PAUP 4.0b10 (Swofford 2002). A 12-taxa dataset was analysed, with Crambus pascuella and Crambus uliginosellus set as outgroups. Maximum parsimony analyses were performed using the Branch-and-bound method as searching algorithm with parameters left unchanged. The bootstrap resampling method with 1000 replicates was used.
The resulting tree was a 50% majority-rule consensus tree with bootstrap (BS) values assigned to each node. BS supports are reported on the tree of Figure 42.

Molecular-based analysis
Because of the age of some of the material used, only the COI barcode sequences were available for C. bijuga and C. coronata, and the whole COI gene for C. paulella. For C. bijuga, we used the sequence BC MTD 1840 to build the datasets as this sequence performed better than BC MTD 1839 in phylogenetic analyses. The COI sequence of LEP 972 was combined with that of wingless for LEP 973 given that the two samples come from the same population and are genetically very similar as attest the barcode sequences. We generated four different datasets from the sequence data: a complete dataset with all three genes sequences available for the 12 taxa (mol_1), a 12-taxa dataset with the 3 rd codon position of COI deleted (mol_2), a dataset restricted to the 7 taxa for which the sequence of the COI gene and at least of one nuclear gene were available (nucl_1), and the same data as nucl_1 but with the 3 rd codon position of the COI deleted (nucl_2). We used the programme RAxML (Stamatakis 2006) to perform phylogenetic analyses of these four datasets under the Maximum Likelihood algorithm. The analyses were run on the online platform CIPRES (Miller et al. 2010) with the web-server adapted interface of RAxML (Stamatakis et al. 2008). The version 7.2.8 of RAxML with the Black Box were used. As in the morphology-based analysis, Crambus pascuella and Crambus uliginosellus were set as outgroups in the analyses of mol_1 and mol_2, and Crambus uliginosellus was set as outgroup in the analyses of nucl_1 and nucl_2. For each dataset, three analyses were performed: one without partition, one with the data partitioned into the three genes, and one with the data partitioned into the three genes and the three codon positions in each gene. According to Mr-AIC (Nylander 2004), the model which best fits our data is GTR+I+G, hence this model was used as evolution model, with parameters estimated for each partition, and the proportion of invariable sites estimated. Bootstrapping was automatically halted by RAxML when a sufficient number of bootstrap replicates was reached. The number of bootstrap replicates is indicated for each analysis. The resulting trees were 50% majority-rule consensus trees with bootstrap values assigned to each node. BS supports of each analysis are reported on the tree (Fig. 42).

Combined bayesian analysis
A nexus file was created for the 12 taxa investigated, with the four following partitions: (1) the morpho-matrix (Table 6), (2) COI, (3) wingless, (4) EF-1α, and was analysed using Mr-Bayes 3.2.1 (Huelsenbeck and Ronquist 2001). The settings were set as follows to fit to the GTR+I+G model: number of states ("nst") = 6, rate variation among sites ("rates") = gamma-shaped rate variation with a proportion of invariable sites ("invgamma"). The relative rates of substitution ("revmat"), the transition/transversion rate ratio ("tratio"), the stationary nucleotide frequencies ("statefreq"), and the alpha shape parameter of the gamma distribution ("shape") were unlinked in order to allow them to vary among partitions. The number of generations was set to 3,000,000. Crambus uliginosellus was chosen as outgroup. The results were considered good when the standard deviation of split frequencies between the two independent runs was below 0.01. Nodes with a posterior probability (PP) over 0.95 are considered well supported. PP values are reported on the tree (Fig. 42). Diagnosis. Catharylla species have snow white to creamy white wings and short labial palpi. They can be separated from other Argyriini by the presence on the forewing of median and subterminal thin transverse lines, slightly curved, convex on costal 1/3. The labial palpi are also shorter in comparison to those of Vaxi. The highly variable male genitalia do not show any synapomorphy or generic diagnostic character. In females, a possible synapomorphy is the strongly reduced anterior and posterior apophyses of abdominal segments VIII and IX, but this is shared with some Crambini and a few other Crambinae (see Landry 1995).

Catharylla
Redescription. Head white, chaetosemata present. Antenna brown, covered with light ochreous to brown scales. Maxillary palpus light ochreous to brown, white tipped. Labial palpus 1.0-1.95 × width of head, curved upward; white basally, light ochreous to ochreous, white tipped, with some brown or dark brown. Thorax white, with ochreous to brown scales at collar. Foreleg coxa white to whitish brown, femur dorsally brown to dark brown, tibia and tarsomeres distally ringed with dark brown. Midleg white to light ochreous with tibia-femur joint ashen brown, with pair of spurs at apex of tibia, tarsomeres II-V dorsally brown to dark brown, with white tips. Hindleg white, with 2 pairs of spurs on tibia, tarsomeres as on midleg. Male frenulum simple, frenulum hook present; female frenulum with 3 or 4 acanthae. Forewing length: 7.5-15 mm in males; 9.5-22 mm in females. Wing venation (of C. chelicerata) ( Fig. 9): R1 present and free, not connected to Sc; R2 free; R3 connected with R4 at 3/4; R5 stalked with R3+R4 at 1/4; M1 from upper corner of cell; cell opened between M1 and M2; M2 and M3 not stalked; CuA1 from lower corner of cell; CuA2 at distal 1/3 of cell; 1A+2A strong. Hindwing Sc+R1 connected to Rs at distal 1/3; M1 connected to Sc+R1 by short narrow vein; M3 connecting to M2 at distal 1/3, CuA1 connecting to M2 at half of length and CuA2 connecting at basal 2/5; 1A unforked; 2A unforked, strong; 3A present, unforked. Forewing (Figs 1-8) background snow white; pattern with costal margin ochreous to brown, sometimes faded; median and subterminal transverse lines thin, ochreous to brown, convex toward costa; outer margin ochreous, sometimes with dark brown spots between veins, or spots forming a continuous line; fringes white; verso light ochreous to ochreous; with marginal spots pronounced. Hindwing snow white to cream-colored; with hairs along 2A and root of M2; with small dark brown spots on outer margin, sometimes in continuous line; sometimes with postmedian transverse line; verso white with marginal spots pronounced.
Distribution. The genus is restricted to the Neotropical Region, from Costa Rica to Santa Catarina, Brazil, from sea level to 1300 m (Figs 43-46).
Biology. The biology of the species remains unknown. In the Serra Bonita Reserve in march 2011, we observed C. serrabonita in its environment, i.e. forested hills up to about 950 m in elevation, surrounded by cacao or coffee plantations in the lowlands. The moths were coming to light, usually very late (after 23:00).
Phylogenetic relationships and monophyly. Presumably, given the reduced labial palpi and the forewing color and pattern, Catharylla has been placed in the Argyriini (Munroe 1995). But our phylogenetic analyses do not support tribe Argyriini with Catharylla included, and the genus seems to be most closely related to Micrelephas.

Catharylla bijuga
Diagnosis. On the forewing (Fig. 1), the seven, thin, marginal dark brown dashes with the most tornal two shaped like spots will separate this species from the others. In male genitalia (Fig. 11), the strongly sclerotized double costal arm of the valva with the ventral arm tubular is a distinctive character. In female genitalia, the best diagnostic character is the sclerotized projection latero-ventrally on sternite VIII (Fig. 34).
Description. Male (n=17) ( Fig. 1): Antenna brown with light ochreous scales; with patch of dark brown scales at base. Maxillary palpus ringed with brown at base and half of length, white tipped. Labial palpus: 1.4-1.6 mm long; ochreous, slightly lighter basally, ringed with dark brown at 2/3, white tipped. Thorax slightly ochreous at collar. Foreleg coxa white; femur ochreous, dark brown dorsally; tibia and tarsomeres ochreous, distally ringed with brown. Midleg and hindleg light ochreous; tibia-femur joint brown on midleg; tarsomeres II-V brown to dark brown on upperside, with white ringed tips. Abdomen dull white to greyish brown. Forewing length: 9.0-10.5 mm; snow white, with yellow ochreous to brown costal margin, partially disrupted when meeting transverse lines; median line yellow ochreous; subterminal line yellow ochreous to brown, forming small triangular spot on costal margin; subapical triangle on costal margin ochreous; outer margin slightly ochreous with five dark brown dashes regularly spaced or sometimes forming faintly continuous line, and one cubital and one anal spots, with cubital spot slightly displaced toward base; fringes brass colored; underside dull white to light ochreous along costal margin, with marginal dashes pro-nounced. Hindwing snow white, veins slightly ochreous, with shiny aspect; marginal line thin, brown, pronounced up to CuA1, then shiny white; fringes white; underside white, with same margin as on recto.   Tympanal organs (n=8): Tympanic pockets extending slightly beyond transverse ridge, rounded. Tympanic drum elongate, more or less oval, postero-laterally extended beyond transverse ridge.
Distribution. The species occurs in lowlands in the three Guianas and Brazil (Fig. 45). Etymology. Bijuga comes from the Latin bijugus, a, um which means "yoked together, double", in reference to the bifid costal arm of the male genitalia.
Notes. In some paratypes from French Guiana the collecting data mention a "pk" (="point kilométrique"). This kilometric marker refers to the distance of the collecting spot on the forest road to the nearest main road. CIRAD (Centre de coopération internationale en recherche agronomique pour le développement) refers to the name of the research institution leading agronomical research on the Combi site. the Combi site. When S. Bleszynski looked into Catharylla, he gave the manuscript name Catharylla ramona to this species, but never published it. The comparison of the tip of the tubular costal arm of the male genitalia and the female lateral projections of sternite VIII shows rather nicely that the male hooks the female genitalia during the mating process. The specimen collected in Parque Nacional do Jaú, Brazil, shows a divergence in COI barcode sequence of 5.05% with that of Roura, French Guiana. In morphology we find no significant difference corroborating this divergence. The relationships of this species to the others remain uncertain in our phylogenetic analyses.

Chelicerata species group
Diagnosis. The synapomorphies of the group are the quadrangular valva with a truncated apex and the hook-shaped gnathos in the male genitalia. The chelicerata species group can be separated from the other Catharylla species based on additional diagnostic characters. Externally, the forewing has a clear, dark brown costal band, and its length is usually over 14 mm. In male genitalia, the apex of the uncus is regularly rounded with a short, narrow projection medially and the vesica shows one large, curved, pointed cornutus, preceded by a string of 13-14 smaller cornuti increasing in size toward apex. In female genitalia, the ductus bursae shows a pronounced, tongueshaped projection postero-ventrally.
Notes. This group includes two species. The phylogenetic analyses restricted to the nuclear genes and the combined Bayesian analysis place the group as sister-group of the mayrabonillae species group (Fig. 42).

Catharylla chelicerata T. Léger
Diagnosis. From C. gigantea, C. chelicerata differs in having the male costal arm hook shaped, longer, and thinner than in C. gigantea, and the juxta is strongly downcurved, apically conical whereas it is long, almost straight, without apical conical projection downward in C. gigantea. In female genitalia the sterigma forms a strongly sclerotized symmetrical structure made of two asymmetrical bell-shaped cavities, opened anterad in C. chelicerata whereas it forms a pair of shallow pockets opened posterad in C. gigantea.
Description. Male (n=21) (Figs 2,9): Head with ochreous to brown chaetosemata. Antenna greyish brown with light brown scales, with patch of brown scales at base. Maxillary palpus ochreous to dark brown, lightly ringed with dark brown at 2/3, white tipped. Labial palpus: 1.3-2.0 mm long; ochreous, basally white, tip of segment II light greyish-brown; white tipped. Thorax with dark brown patch at collar. Foreleg coxa white; femur white, ashen brown dorsally, tibia and tarsomeres ochreous, distally ringed with dark brown. Midleg femur white, tibia ashen brown basally, tarsomeres ochreous, brown to ashen brown on upperside, with white ringed tips. Hindleg white, tarsomere I ochreous; II-V brown on upperside, with white ringed tips. Abdomen dull white to light ochreous. Forewing length: 10.5-15.0 mm; costal band wide, brown from base to apex; median and subterminal transverse lines faded brown, sometimes completely faded; dark brown spots on apical margin forming more or less continuous line; fringes brass colored; underside white with costal margin brown; outer margin with somewhat triangular spots. Hindwing snow white, with marginal spots between veins; fringes white; underside silvery white with marginal spots pronounced.
Distribution. The species was found in French Guiana and Brazil (Amazonas) (Fig. 43).  Etymology. "Chelicerata" refers to the shape of the costal arms of the male valva, which look like mygalomorph chelicerae.
Notes. Two females included here have been named Catharylla robustella (genitalia on slide GS-5949-SB, CMNH) and Catharylla tenellina (genitalia on slide Pyralidae Brit. Mus. Slide N°17693, BMNH) by S. Bleszynski, as indicated on labels, but these names were never published. These two specimens are probably C. chelicerata, but the bad genitalia preparations do not allow to see details, and therefore they are not included as paratypes. Diagnosis. From C. chelicerata, C. gigantea differs in having the male costal arm shorter, basally wide and tooth shaped while it is long, narrow throughout and hook shaped in C. chelicerata. The juxta is long, tongue shaped, almost straight, and apically rounded, whereas it is downcurved and apically conical in C. chelicerata. In female genitalia, the sterigma forms a pair of shallow pockets opened posterad whereas in C. chelicerata the sterigma forms a strongly sclerotized symmetrical structure made of two asymmetrical bell-shaped cavities opened anterad.

Catharylla gigantea
Description. Male (n=6) (Fig. 3): Head with ochreous chaetosemata. Antenna brown with light brown scales, with patch of dark brown scales at base. Maxillary palpus brown with dark brown spot at half of length, white tipped. Labial palpus: 1.6-2.4 mm long; ochreous to brown ochreous, basally white, with patch of dark brown scales at half of length, white tipped. Thorax with some brown at collar. Foreleg coxa white, femur white, ashen brown dorsally; tibia and tarsomeres brown-ochreous, distally ringed with dark brown. Midleg white with tibia-femur joint and base of tibia ashen; tarsomeres ochreous to brown ochreous with upperside brown to dark brown, white tipped. Hindleg white with tarsomeres II-V ochreous to brown ochreous, up-perside brown, with white tips. Forewing length: 13.5-14.5 mm; snow white with wide brown to dark brown costal line from base to apex; median and subterminal transverse lines faded brown; dark brown spots on termen forming more or less continuous line; fringes brass colored; underside white, with costal margin brown ochreous, outer margin with subtriangular spots. Hindwing snow white; marginal spots dark brown between R5, M1, M2, M3, and CuA1; fringe white; underside snow white, with same spots as on upperside.
Etymology. The name comes from the Latin giganteus, a, um meaning very large. Notes. The name was given to the species on manuscript labels by S. Blezynski, probably in reference to the large size of the female.

Tenellus species group
Diagnosis. The synapomorphies of the group are the dorsal furrow on the uncus, the uncus apex slightly bifid, the presence of a transtilla in male genitalia, and the absence of a ventral connection of sternite VIII in female genitalia. The tenellus species group can also be separated from the other Catharylla species based on the following additional diagnostic characters: the hindwings are creamy-white, and in female genitalia, the papillae anales are not produced.
Notes. This group includes three species, including two new ones. Catharylla serrabonita and C. tenellus form a monophyletic group within the tenellus species group (Fig. 42). Relationships to other species groups are unknown.
Diagnosis. From C. serrabonita and C. coronata, C. tenellus can be separated by the median transverse line, which is faintly convex towards costa, whereas it is more strongly convex in C. coronata and C. serrabonita. The male genitalia provide the best diagnostic characters. The most obvious refers to the transtilla, which forms a pair of short, narrow sclerotized arms with pointed tips, projecting posterad, with, in between, a pair of brushes directed medio-ventrally, whereas it forms a pair of arms pointing posterad with a string of spines ventrally in C. serrabonita and C. coronata. In female genitalia, the anterior angle of sternite VIII is directed downward into a more or less rounded projection covered with short spinules of same length, whereas it is projected anterad in C. serrabonita, and it is not projected in C. coronata.
Male genitalia (n=13) (Figs 17,18,(27)(28)(29): Uncus about half of length of tegumen arms, broadly downcurved; uncus arms connecting at base, with ventro-lateral tuft of setae; dorsal furrow with few short setae on each side, tip rounded, slightly indented medially, slightly convex in apical 1/3. Gnathos short and thick, arms joining at half of length, laterally compressed toward apex, almost straight, slightly downcurved, with apex pointing upward. Tegumen arms slightly enlarging toward apex; connecting at 3/4, slightly projected dorsally at connection. Costa of valva at 2/3 with arm directed posterodorsally with rounded tip, without basal projection on dorsal edge or narrow with low basal projection, or wide with basal projection; cucculus upcurved in apical 1/4. Juxta ogival, posteriorly directed downward, with pair of thumb-like lobes ventrally reaching about 2/3 of length. Transtilla strongly sclerotized, with pair of narrow arms on each side of middle, pointing posterad, with 3 spines apically, also with pair of shorter brushes of tightly set spines medio-ventrally; in some specimens triangular median projection dorsally with few tiny setae. Phallus S-shaped, subapically with dorsal bump, apically lightly sclerotized, truncated, covered with microspicules barely visible; vesica without cornuti.
Notes. The species was described from "one female collected in Brazil, near Rio de Janeiro". Hence, the lectotype designated by a label by S. Bleszynski is not warranted. This designation is presumably based on the fact that Zeller (1863:50) mentions a pair deposited in the Vienna Museum. The association of sexes in this species is not 100% certain.
Specimens from Porto Seguro, Brazil show a divergence of 3.34% in COI barcode sequences with the specimen from Ubatuba, Brazil. In morphology, differences in male genitalia are also observed: in the specimens from Bertioga, Caraça, São Paulo and Ubatuba the costal arm of the valva is wide and 1/3 of the length of the cucullus, almost reaching its tip, and the dorsal edge at base is slightly produced (Fig. 27). In the specimens from Porto Seguro, the costal arm is about 1/5 the length of the cucullus, relatively narrow, and the dorsal edge is slightly produced at base (Fig. 29). Another form, from Caraça, Minas Gerais (Fig. 28) was also found. No differences were found in the female genitalia. We feel that specimens and data are currently lacking to conclude that possibly more than one taxon should be recognized under C. tenellus, or that there is indeed a deep divergence in the COI barcode between populations of this species. Diagnosis. From C. serrabonita and C. tenellus, Catharylla coronata can be separated with characters of the male genitalia: the uncus is apically bifid and grooved on distal 1/5 in C. coronata whereas it is only indented medially at apex in C. serrabonita and C. tenellus; the costal arm of the valva is short and the apex is curved inward in C. coronata whereas the costal arm is longer and points postero-dorsally in the other two species; the transtilla forms a pair of sclerotized arms slightly bent inward distally, ventrally with a row of short spines increasing in size from base to apex whereas it forms a pair of short, narrow sclerotized arms with pointed tips, projecting posterad, and with a pair of brushes directed medio-ventrally in C. tenellus and a pair of sclerotized arms strongly bent inward on distal 1/4 and with a string of long spines of same length medially along it in C. serrabonita; the juxta is shorter than in C. tenellus, and regularly narrowing toward apex whereas it is strongly narrowing on distal 1/4 in C. serrabonita; the ventral projections of the juxta form a pair of shallow pockets whereas they are bell-shaped in C. serrabonita and thumb-like in C. tenellus; the vesica has a row of 6-7 cornuti in C. coronata whereas it does not show any cornuti in C. serrabonita and C. tenellus. In the female genitalia of C. coronata, the anterior angle of sternite VIII is not projected whereas it is rounded, projected anterad and covered with short spinules in C. serrabonita, and projected downward in C. tenellus. The anterior apophyses are quadrangular, anvil shaped whereas they are spine like in the other two species.
Etymology. The name comes from the latin coronatus, a, um: crowned, referring to the longitudinal string of short spines of the transtilla in the male genitalia.
Notes. Based on our combined phylogenetic analysis, C. coronata is the sister species of the C. tenellus + C. serrabonita pair (Fig. 42). Diagnosis. From Catharylla coronata and C. tenellus, C. serrabonita can be separated by the zigzagging median transverse line with the short triangular dent at CuA2 and the pronounced creamy color of the hindwing. The male genitalia provide the best discriminant characters: in C. serrabonita, the transtilla forms a pair of sclerotized arms bent inward in distal 1/4 and with a string of long spines of same length medially along it, whereas it forms a pair of short, narrow sclerotized arms with pointed tips projecting posterad, and with a pair of brushes directed medio-ventrally in C. tenellus, and two sclerotized arms slightly bent inward distally, with a row of short spines increasing in size from base to apex in C. coronata, and the juxta is apically narrow and pointed whereas it is triangular and regularly narrowed in C. coronata and C. tenellus. In female genitalia, the anterior angle of sternite VIII is projected anterad into a rounded protrusion covered with short spinules in C. serrabonita, whereas it is projected downward in C. tenellus and it is not projected in C. coronata.
Distribution. The species occurs in Brazil (Bahia, Espirito Santo) (Figs 45 & 46). Etymology. The name comes from that of the Serra Bonita Reserve founded by Vitor O. Becker and Clemira de Souza. It is managed by Instituto Uiraçu in the State of Bahia, Brazil.
Notes. Serra Bonita Reserve is located in the Atlantic Forest, in a hilly region of cacao plantations and scattered forest. Adults came late to light, usually after 23:00. Our molecular analysis of the COI barcode sequences highlighted that specimens from Serra Bonita respectively show 3.24 and 2.21 % base differences with those of Porto Seguro and Linhares. This divergence is possibly associated with slight morphological differences in male genitalia as shown in Figs 23-26. No females were found at Serra Bonita.

Mayrabonillae species group
Diagnosis. We have not recovered any obvious synapomorphy for this group. It can be separated from the other Catharylla species based on the shorter forewing length, usually between 7.5 and 9.0 mm (maximum 10.5 mm). In male genitalia, the tegumen connection is more than two times longer than the uncus, the uncus is beak-shaped, with the apex narrowing to a point, the gnathos is bent at an angle of about 90°. In female genitalia, the basal line along the anal papillae is ventrally expanded onto a triangle, and the sterigma forms a pair of sclerotized pockets on each side of the middle, covered with short spines or spicules. The sterigma does not bear tiny setae on the ventral membrane of segment VIII.
Notes. This group includes two species. The phylogenetic analyses restricted to the nuclear genes and the combined Bayesian analysis place the mayrabonillae group as sister to the chelicerata group (Fig. 42).

Catharylla mayrabonillae T. Léger
Diagnosis. The best discriminant characters externally between the two species of the mayrabonillae group are the shape of the forewing outer margin, which is slightly produced apically in C. mayrabonillae and not produced in C. paulella, and the forewing median transverse line with two strongly pronounced spots at 1/3 and 2/3 in C. paulella, whereas these spots are lacking in C. mayrabonillae. The hindwing of C. mayrabonillae has a faded subterminal transverse line on costal half whereas the hindwing of C. paulella lacks this marking. In male genitalia, the heavily sclerotized sacculus bears a dorso-lateral sclerotized string of short spines on distal 1/4 whereas the two processes of the costa are S-shaped in C. paulella, and the apex of the phallus is trifid, rounded medially, shortly triangular laterally, whereas it is simply rounded in C. paulella. In female genitalia, the sterigma forms double rounded cavities with a mustachio-shape arrangement of short spines in ventral view, and the ductus bursae is wide, progressively widening toward corpus in C. mayrabonillae, whereas the sterigma forms a pair of shallow rounded pockets on each side of middle and the ductus bursae is narrow, with the rounded corpus bursae clearly differentiated from it in C. paulella.
Description. Male (n=17) (Fig. 7): Head with light ochreous chaetosemata. Antenna brown, with white scales dorsally and patch of dark brown scales at base. Maxillary palpus light ochreous, ringed with dark brown at 2/3; white tipped. Labial palpus: 1-1.4 mm; white, with patch of dark brown scales at 1/3 and 2/3 laterally. Thorax with patch of light ochreous scales at collar. Foreleg coxa whitish brown, femur white, dorsally ashen brown, tibia and tarsomeres ochreous, distally ringed with dark brown. Midleg femur white, tibia light ochreous, basally brown, tarsomeres II-V ochreous with tips ringed white. Hindleg white, except tarsomeres, as in midleg. Abdomen dull white. Forewing length: 7.5-8.5 mm; with apex slightly produced; costal line thin, ochreous or white in basal half, white in apical half; median transverse line ochreous, slightly undulated; subterminal transverse line ochreous; transverse lines enlarging into brown spot on costal margin with ochreous bar on costa following subterminal transverse line; terminal sector with light ochreous between veins, margin with thin, dark brown line from apex to CuA1, with two dark brown spots in cubital sector, with spot between CuA1 and CuA2 slightly displaced toward base; fringes brass colored; under-side light ochreous with some brownish scales, with thin brown margin. Hindwing white with thin transverse subterminal line faded ochreous, in continuity with forewing median transverse line; outer margin line pronounced, dark brown; underside dull white with thin faded brown margin; fringes white.
Female genitalia (n=16) (Fig. 40): Papillae anales strongly curved in lateral view; sclerotized line along papillae expanding ventrally into triangle. Posterior apophyses 0.35-0.45 × length of papillae anales. Tergite VIII about 1/3 length of sternite VIII; postero-dorsal margin with few setae of moderate length; anterior apophyses 0.03-0.1 × papillae anales; sternite VIII with patches of minute setae antero-ventrally on each side of bare median band. Sterigma forming double rounded cavities with mustachioshaped arrangement of short spines (in ventral view); remaining cavity wall with tiny spines. Ductus bursae short and wide, enlarged near middle; partly sclerotized on right side of enlargement and posterior section. Corpus bursae circular to elongate, about as long as tergite VII; single signum faintly pronounced.
Distribution. The species has been found so far in Panama, Costa Rica, Colombia, Venezuela, Guyana, Suriname, French Guiana, Ecuador, Peru and Brazil (Acre, Amazonas, Distritò Federal, Pará, Rondônia) (Fig. 44). It is the most widespread species of Catharylla and the only one so far found in Central America and in Venezuela, Columbia, Ecuador and Peru.
Etymology. Catharylla mayrabonillae is named in honor of Ms. Mayra Bonilla of San Jose, Costa Rica, in recognition of her artistic portrayal of the biodiversity and ecosystems of Costa Rica and her many years of support for the existence of the rain forest in Area de Conservacion Guanacaste.
Notes. The relatively strong COI barcode divergence of 4.34% between samples LEP 1126 from Peru and 07-SRNP-113921 from Costa Rica (Table 5) is notable but it is not associated with morphological variation. Schaus, 1922 http://species-id.net/wiki/Catharylla_paulella Figs 8, 10, 32, 33, 41, 44 Catharylla paulella Schaus, 1922: 131;Bleszynski and Collins 1962: 226;Bleszynski 1967: 97;Munroe 1995: 35;Nuss et al. 2013. Diagnosis. This species can be easily separated from the other Catharylla species by the forewing median transverse line with two strongly pronounced spots at 1/3 and 2/3. The forewing is also sparkled with dark brown scales, which is unique in the genus. In male genitalia, the two S-like projections of the costal arm of the valva discriminate this species from the other species of Catharylla. In female genitalia, the sterigma forms a pair of shallow rounded pockets on each side of middle, and the ductus bursae is narrow, with the rounded corpus bursae clearly differentiated from it in C. paulella, whereas it forms double rounded cavities with a mustachio-shape arrangement of short spines in ventral view, and the ductus bursae is wide, progressively widening toward corpus in C. mayrabonillae.
Male genitalia (n=2) (Figs 32, 33): Uncus almost straight, densely setose, about 1/4 length of tegumen arms. Gnathos with arms joining at 3/5, then directed upward at slightly less than 90° angle; apically narrowly rounded. Tegumen arms regularly widening toward uncus, connecting at about half their length. Cucculus of medium width, slightly curved upward in distal 1/4; costal arm of valva divided with short spatula at base and Sshaped projection with rounded apex apically. Juxta triangular with distal third narrower, apically rounded, with baso-lateral narrow, triangular projections pointing anterad. Phallus with apex more thickly sclerotized, with blunt apical margin, with short triangular ventral projection; vesica covered on basal 1/4 with tiny spicules, with barely visible microspicules all along, with one wide and curved cornutus at about 1/4 length of phallus.
Female genitalia (n=5) (Fig. 41): Papillae anales ventrally slightly projected; sclerotized line at base enlarging medially to triangular shape covered by minute punctuation. Posterior apophyses 0.4-0.5 × length of papillae anales, narrow, tubular, with rounded tips. Tergite VIII short, about half of length of greatly enlarged sternite VIII. Anterior apophyses 0.05-0.1 × length of papillae anales. Lamella antevaginalis of sterigma dorsally covered with minute spicules; pair of shallow rounded pockets on each side of middle opened posterad. Base of ductus bursae sclerotized, forming circular membranous and narrow pocket. Corpus bursae regularly rounded, without signum.
Notes. The original description doesn't mention the original number or sex of the specimens but it is assumed that there was only one. S. Bleszynski gave the new name of Catharylla hibisca to specimens that appear to be C. paulella. The BMNH São Paulo specimen is associated with slide n° 17692, but the genitalia on this slide seem to be wrongly associated, given the inscription "wrong abdomen?" on the label, as well as the size of the abdomen, which is much bigger than those of C. paulella. Therefore, this specimen cannot be identified with certainty. An error is possible in the association of the sexes of this species as there are no series of both sexes from the same locality or other means of associating them with 100% confidence. Forewing costa with thick, brown to greyish brown stripe (Figs 2, 3); forewing length usually > 14 mm. Hindwing white. In male genitalia, gnathos regularly curved, juxta without latero-ventral projections (Figs 13, 15) .......

COI barcode sequences and genetic distances within Catharylla
The number of bases obtained for each barcode sequence is given in Table 3 and the genetic distances between specimens in Table 5. As most of the data were restricted to few sequence samples per species, and some of the sequences came from the same populations, we don't have the definitive picture of the intraspecific variation in the COI barcode sequences of Catharylla species. We observe a relatively high divergence between different barcode haplotypes in C. bijuga (5.05%), C. mayrabonillae (4.34%), C. serrabonita (3.24%) and C. tenellus (3.34%), sometimes possibly associated with differences in morphological characters (see Notes of the descriptions of C. bijuga, C. tenellus) and with geographical distances. The divergence between C. chelicerata samples LEP 1703 and LEP 1704 is of 0.15% (1 base) because they are issued from the same population. Sample LEP 1290 differs from sam- ples LEP 1703 and LEP 1704 respectively by 0.62 and 0.46%. We observe no variation between samples LEP 1708, LEP 1709, LEP 1710 and LEP 1888 because they are all issued from the same population. The inter-specific variation in COI barcode sequences (6.29-16.84%) is always higher than the intraspecific variation (0.15-5.05%).

Morphological results
The 21 analysed characters are listed in Table 6.

Phylogenetic analysis of Catharylla and related genera based on molecular data
The monophyly of Catharylla is supported by the analysis of mol_1 and the combined Bayesian analysis (BS support = 90, PP = 0.95) and by the analyses of nucl_1 and nucl_2 (C. chelicerata, C. mayrabonillae, C. serrabonita and C. tenellus represented). Three synapomorphies (11: 1, 17: 1, 19: 1, with one reversal to the ancestral state in the chelicerata group for character 11) and one non-unique apomorphy (16: 1, apomorphy observed in the two outgroups as well) support the group. The mayrabonillae group is well supported in all analyses (BS supports = 100 in mol_1 and mol_2 analyses, PP = 1.00 in the combined Bayesian analysis) except in the morphology-based analysis (BS support = 60), because no clear synapomorphy was found for the group. However, two non-unique synapomorphies and one reversal (observed only once in Catharylla) are observed (9: 1, 12: 1, 19: 0). The tenellus group is well supported by the morphology-based analysis (BS support = 97) with four clear synapomorphies (5: 1, 7: 3, 10: 1, 20: 1) and two reversals (3: 0, 15: 0), but show no support in other analyses, probably because only the barcode sequence was available for C. coronata. The closer relationship between C. serrabonita and C. tenellus within the tenellus group is supported by the combined Bayesian analysis and by one non-unique apomorphy (but unique in Catharylla). The tenellus group as represented in the analyses of nucl_1 and nucl_2 (C. serrabonita + C. tenellus) is well supported (BS supports of 99 and 93, respectively). The chelicerata group is well supported by the morphology-based analysis (BS support = 96) and the combined Bayesian analysis (PP = 1.00). The group shows two synapomorphies (8: 1, 9: 2), one non-unique apomorphy (7: 1) and one reversal (11: 0). Unfortunately, no sequence was available for C. gigantea, thus we cannot compare the morphology with the molecular-based analyses. The closer relationships between the chelicerata and the mayrabonillae groups is strongly supported by the nucl_1 analysis (BS support = 100) and the combined Bayesian analysis (PP = 1.00). Two reversals (4: 0, 6: 0) support the group. The position of C. bijuga as sister group of the other Catharylla species is weakly supported by the combined Bayesian analysis (PP = 0.90), and show no support in other analyses. The position of Micrelephas pictellus as sister

18.
Ventral tongue shaped pronounced sclerotization postero-ventrally on ductus bursae: absent (0); present (1)  group of Catharylla is supported by the analysis of mol_2 and nucl_2 (respective BS supports of 91 and 100). This node is supported by one synapomorphy (6: 1, with one reversal in the chelicerata + mayrabonillae clade). Node 10 (Argyria lacteella + Micrelephas pictellus + Catharylla) is supported by the analysis of mol_1 (BS support = 100) and the combined Bayesian analysis (PP = 1.00). One synapomorphy (2: 1) supports the group. The monophyly of the two Crambus species chosen as outgroups is well supported in the mol_1 analysis (BS support of 100). The settings of MrBayes do not allow to choose two outgroups, therefore the monophyly of the two Crambus species is not supported by the combined Bayesian analysis. Three non-unique synapomorphies support the genus based on these two species (7: 1, 12: 1, 16: 1). show distinct distribution patterns. The chelicerata group is widespread in the Northern Amazonian rainforest of Brazil, and in the three Guyanas (Fig. 43). The tenellus group is restricted to the south-eastern Atlantic coast of Brazil in the Atlantic Forest, which is formed by tropical moist forests (Thomas and Britton 2008) (Fig. 45). These two groups seem to be restricted to moist forest habitats. The sister species C. mayrabonillae and C. paulella show an allopatric distribution: C. mayrabonillae is widespread northwards up to Costa Rica and southwards down to Brasilia (15°47 S), whereas C. paulella is widespread from Feira Nova (07°00 S) down to São Paulo (23°35 S) (Fig. 44). They are both recorded from the Federal District, and from Maranhão (Feira Nova). Both species show wide geographical distribution covering a large range of ecosystems: C. mayrabonillae occurs in the tropical moist forests of the Carribean, the Northwestern South American and the Amazonian subregions, as well as in the Cerrado province of Central Brazil (records from Brasilia, Distrito Federal) (Figs 43, 44), where savannahs predominate with a semi-humid tropical climate and a pronounced dry period during winter (aprilseptember) (Ratter et al. 1997). Catharylla paulella occurs in the Cerrado (Planaltina, Distrito Federal), in the Chaco formation (Urucum, Matto Grosso, Brazil and Provincia del Sara, Bolivia) (Figs 43, 44), which is a hot and dry area with xerothermic deciduous forests (Candollea 1993). Catharylla bijuga is restricted to the Guyana province. The distribution areas of the chelicerata group and that of C. bijuga are allopatric with respect to that of the tenellus group. The dry diagonal (Prado and Gibbs 1993) formed by the Caatinga, the Cerrado and the Gran Chaco, highlighted by the low rainfalls rates (represented in yellow, orange and brown on Fig. 43), constitutes the climatic barrier between the Atlantic and the Amazonian Forests. Both barcode haplotypes of C. bijuga are from the Amazonian Forest, but they are distant from each other by about 1290 km      Roura). Catharylla mayrabonillae barcode haplotype 07-SRNP-113921 is found in the Carribean subregion, whereas haplotype LEP 1126 is found in the Amazonian subregion. The barcode haplotypes of the Catharylla serrabonita populations from the two coastal localities of Linhares and Porto Seguro are more closely related, whereas the haplotype from the population of the forested hills of the Serra Bonita Reserve clearly diverges from the other two (Fig. 46).

Reliability of phylogenetic analyses
The analyses of nucl_1 and nucl_2 datasets brought better BS supports than the analyses of mol_1 and mol_2 on nodes 1, 6, 9, and for the monophyly of Catharylla (nodes 7 & 8). The reduced number of taxa (7 taxa in nucl datasets vs 11 in mol datasets, with 4 Catharylla species in nucl_1 and nucl_2 vs 7 in mol_1 and mol_2), as well as the quality of the datasets (complete sets of genes for nucl_1 and nucl_2 vs partly complete sets of genes in mol_1 and mol_2) explain the better results obtained in nucl_1 and nucl_2, because taxa with incomplete data tend to lower the resolution and the bootstrap supports of the tree (Regier et al. 2009). The barcode sequence evolves rapidly and therefore fails to accurately reconstruct phylogenetic relationships. The analysis of a higher number of genes would increase the support values (Wahlberg and Wheat 2008). The morphology-based analysis yielded good support for two species groups within Catharylla, but failed to reconstruct the phylogenetic relationships between different Catharylla species groups and between Catharylla and other genera. The weak number of characters of the matrix used (21, reported in Table 6) compared to the number of taxa, as well as the homoplastic characters explain the lack of bootstrapsupported groups (Bremer et al. 1999). The combined Bayesian analysis yielded good supports within Catharylla and between Catharylla and other genera.

Taxa positions on the phylogenetic tree
Our molecular-based analyses failed to place with certainty two of the four taxa lacking nuclear sequences (C. bijuga and C. coronata) because of the great divergence of these sequences from those of other Catharylla species. However, the clear support brought by the morphological analysis places C. coronata together with C. serrabonita and C. tenellus. The neighbor-joining (NJ) analysis of barcode sequences of Catharylla species (not represented here) places C. coronata together with C. serrabonita (divergence of 7.59% with the sequence BC MTD 1843) and C. tenellus (divergence of 7.3% with the sequence BC MTD 1842), and thus corroborates the findings of the morphology-based analysis. Moreover, the species is morphologically very similar to C. serrabonita especially regarding the transtilla in the male genitalia. The basal position of C. bijuga in Catharylla, as sister taxon of other Catharylla species, is doubtful (PP = 0.90) and may be an artefact due to the great divergence of the barcode sequence (the lowest divergence with other Catharylla species is of 9.7% with sample BC MTD 1843 of C. serrabonita). The supports brought by the combined Bayesian analysis have to be carefully considered since the results are sensitive to small variations of the taxon and character sampling, and the posterior probabilities tend to overestimate the strongness of the nodes (Buckley et al. 2002, Whittingham et al. 2002. For C. bijuga and C. coronata, a complete gene dataset would be needed because the barcode sequence alone is not sufficient to undoubtfully assess their position within Catharylla. The closer relationships of the chelicerata and the mayrabonillae groups (Fig. 42, node 6) is well supported by the analysis of nucl_1 and by the combined Bayesian analysis (BS support = 100, PP = 1.00), but a decrease of the BS support is observed when comparing mol_1 to mol_2, and nucl_1 to nucl_2, showing that the third codon position of the COI gene supports this node. This suggests a rapid evolution of this group where mutations accumulated only at the less constrained third codon position, or this could result from convergent substitutions in each lineage. Micrelephas pictellus is here placed as sister group of Catharylla, an unexpected topology, given that Catharylla has been placed in the Argyriini (type genus: Argyria) along with Argyria, Urola and Vaxi, and Micrelephas in Crambini (Landry 2003). The Argyriini as defined here are not monophyletic and would need to be redescribed, whether to include Micrelephas, or to exclude Catharylla. A more extensive analysis of the tribe with Urola, Vaxi and more genera of Crambini would be desirable to assess the validity and the composition of the Argyriini. The position of Micrelephas as sister group of Catharylla is subject to caution as only three Crambinae genera of the 179 currently recognized (Nuss et al. 2013) are represented in the analyses. The short labial palpi (i. e. the length of labial palpi/eye diameter < 2/1), which constitutes a synapomorphy for node 10 is observed in few other genera of Crambinae (Euchromius, Myelobia and Urola, see Landry 1995), for which long, porrect palpi are the norm.

Species delimitation and possible cryptic species in Catharylla
The intra-specific divergence in barcode sequences within Catharylla (Table 5; distances mapped on Figs 44, 45 and 46) is higher than the threshold of 2-3% predicted to delimit the species level in Lepidoptera according to Hebert et al. (2003), with intraspecific divergence often representing overlooked species (Hebert et al. 2003, Mutanen et al. 2012. The maximum intraspecific COI barcode divergences observed in C. bijuga (5.05%), C. mayrabonillae (4.34%), C. serrabonita (3.24%) and C. tenellus (3.34%) could therefore suggest that these different barcode haplotypes represent different species. Some morphological variation in male genitalia was found in C. serrabonita and C. tenellus (Figs 23-29) haplotypes, but none in C. bijuga and C. mayrabonillae, which is counter-intuitive given that the latter two species have more divergence than the other two. Moreover, none of the intraspecific divergence observed in Catharylla is higher than the interspecific divergence. A greater number of specimens would be required to properly investigate the COI and nuclear gene variation along with morphological and ecological variations to judge if these COI barcode haplotypes and morphological variations represent meaningful differences linked with speciation or incipient speciation or if these species are really older than average. For now, given the available data, we are confident in our decision to recognize the eight species treated here in genus Catharylla.

Biogeography
The different species groups of Catharylla highlighted by the phylogenetic analyses show distinct distribution patterns, with species widely spread, such as C. mayrabonillae and C. paulella (Fig. 44) and other species confined to smaller areas, such as C. bijuga (Fig. 45), and the chelicerata and tenellus species groups (Figs 43 & 45). The humid formations of the Amazonian and Atlantic Forests constituted two different refuges separated by the Caatinga and the Cerrado xerothermic formations, and vicariant evolutionary processes in plants (Rizzini 1997;Perret et al. 2012) and in birds (Eberhard and Bermingham 2005) have been documented. The distribution of C. mayrabonillae and C. paulella (Fig. 44), respectively northwards and southwards of the dry diagonal could also result from the split of a former population between the Amazon and the Atlantic Forests where speciation events occurred during dry spells, with expansion of the species afterwards. The patterns of distribution of the chelicerata group (Fig. 43) suggest that its diversification might have occured in the Amazonian Forest. The two species are sympatric as they are both recorded from Reserva Ducke. Similarly, C. coronata and C. tenellus are sympatric (both collected in Rio de Janeiro and São Paulo), as well as C. serrabonita and C. tenellus (both collected in Porto Seguro), while C. coronata and C. serrabonita are vicariant, C. serrabonita being distributed north of Linhares, whereas C. coronata is found south of this locality. The patterns of distribution suggest that the diversification of the tenellus group might have occurred in the Atlantic Forest, or have occurred in a wider area that reduced afterwards. In C. tenellus, two forms of male valva linked to different COI barcode sequences seem to be geographically separated, with the form from Ubatuba (associated to barcode sequence BC MTD 1842), Bertioga, São Paulo (São Paulo State), and Caraça (Minas Gerais) occurring more to the south than the form from Porto Seguro (associated with barcode sequences BC MTD 1709, 1710, 1711, 1888. The third form, observed in Caraça (Minas Gerais) suggests that this locality could represent a point of contact between the two other forms. Regarding C. serrabonita, the locality of Serra Bonita is a moist forest of middle elevation (800 m) (Fig. 46), while the localities of Porto Seguro and Linhares are part of a drier biome along the Atlantic Coast (V. O. Becker, pers. comm.). Therefore, the genetic distance found between these populations of this species could barely be explained by the distance between the two localities (130 km between Serra Bonita and Porto Seguro), but could be related to different ecological conditions.