Species delimitation, environmental cline and phylogeny for a new Neotropical genus of Cryptinae (Ichneumonidae)

A morphologically unusual Cryptini, Cryptoxenodon gen. nov. Supeleto, Santos & Aguiar, is described and illustrated, with a single species, C. metamorphus sp. nov. Supeleto, Santos & Aguiar, apparently occurring in two disjunct populations in northern and southeastern South America. The highly dimorphic female and male are described and illustrated. The phylogenetic relationships of the new genus are investigated using a matrix with 308 other species of Cryptini in 182 genera, based on 109 morphological characters and molecular data from seven loci. The analyses clearly support Cryptoxenodon gen. nov. as a distinct genus, closest to Debilos Townes and Diapetimorpha Viereck. Species limits and definition are investigated, but despite much morphological variation the analyses at the specimen level do not warrant the division of the studied populations into separate species. The considerable morphological variation is explored with principal component analyses of mixed features, and a new procedure is proposed for objective analysis of colors. The relationship of color and structural variation with altitude and latitude is demonstrated and discussed, representing an important case study for Ichneumonidae. Externally, Cryptoxenodon gen. nov. can be recognized mainly by its unusually large mandibles, but other diagnostic features include clypeus wide; sternaulus complete, distinct and crenulate throughout; areolet closed, about as long as pterostigma width; petiole anteriorly with distinct triangular projection on each side, spiracle near posterior 0.25; propodeum without posterior transverse carina; and propodeal apophyses conspicuously projected.


Introduction
Ichneumonidae are one of the most diverse and ubiquitous groups of insects, with over 100 thousand estimated species [1]. Contrary to the expectation for most major animal lineages, many groups of ichneumonids seem to be more diverse in temperate regions than in the tropics [2][3][4]; but see Sääksjärvi [5] and Quicke [6] and references therein. The extremely diverse tribe Cryptini, however, thrives in tropical regions, where they are "the most conspicuous of all ichneumonids" [7]. The group is particularly diverse in the Neotropics, with 1150 species listed in Yu et al. [8]. This is about 70% more than in the Oriental (668 spp.) and 80% more than in the Palaearctic (639) regions, and nearly four times more than in the Nearctic region (292). The main framework for the taxonomy of Cryptini, as with most Ichneumonidae, is still the four-volume set of monographs by Townes [9,7,10,11]. In the second and largest volume, Townes describes 40 new genera for Cryptini alone, of which 20 are from the Neotropics. Even after this extensive work, new Cryptini genera are still regularly proposed (Fig 1), having increased from 159 to more than 251 since 1970 (compiled from Yu et al. [8]). Notably, most new genera do not derive from splitting groups of previously known species into separate genera, but from the outright discovery of new species that do not fit into any of the previously known genera. However, these new taxa are often difficult to establish with precision; in part, because most new discoveries fall into large and less characteristic groups of Cryptini, such as Townes' Goryphina (= Cryptus group + Mesostenus group, in part, of Santos [12]). Furthermore, the high level of morphological homoplasy observed in Cryptini often obfuscates generic limits. In fact, the most thorough and up-to-date phylogenetic interpretation of Cryptinae [12] has shown that its diversity is as much fascinating as it is morphologically counter-intuitive, leading to a supra-generic classification now based only on informal, albeit monophyletic, genus-groups.
The challenging nature of supra-specific diversity in Cryptini calls for the combination of multiple sources of data and analytical tools to provide a sound assessment of the validity and relationships of proposed new taxa. Accordingly, this work uses an integrative taxonomy  [7]. Compiled from Yu et al. [8].

Taxonomy and morphology
Morphological terminology and taxonomic conventions follow Santos and Aguiar [13], except that the second trochanter is referred to herein as its more anatomically accurate term, trochantellus; the "posterior transverse carina of mesothoracic venter" is referred to as postpectal carina, for simplicity; and the cell 1+2Rs is called the areolet, as it is standard in ichneumonid systematics literature. The first and subsequent flagellomeres are referred to as f1, f2, f3, etc; the tarsomeres of each leg are referred to, from base to apex, as t1, t2, t3, etc, while first and subsequent metasomal tergites are referred to as T1, T2, T3, etc. When potentially ambiguous, color names are followed by their respective RGB formula, as determined from digital pictures of the studied specimens, according to procedures described by Aguiar [14].

Nomenclatural acts and data availability
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub: B4F31941-37D5-4D48-B007-0A388C08063B. The electronic edition of this work was published in a journal with an ISSN and has been archived and is available from the following digital repository: "https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0237233". All original data used to generate this work is provided as (S1-S3 Files) and as molecular data both registered in GenBank (Accession numbers MT083994, MT084596, MT084597, MT084598, MT084599, MT085819, MT089926, MT089929, MT085820, MT085821) and provided directly in the body of the work (Fig 2). All used images are available as a Morphobank project, under the permalink http://morphobank.org/permalink/?P3669.

Phylogenetic placement
In order to test the validity and phylogenetic placement of the proposed new genus, one specimen ("DNA voucher" in Paratypes list) was included in the phylogenetic matrix of Santos [12], comprising 370 species of Ichneumonidae, including 308 species in 182 genera of Cryptini. Seven molecular loci were sequenced using the exact same primers and protocols as in Santos [12]. All morphological characters used by Santos (op. cit.) were scored for the new taxon, checking all available females.
Regarding the morphological matrix, a warning (item 1 below) and a few corrections (2-5) for the character-set and matrix of Santos [12] must be considered to allow its precise interpretation and usage herein: 1. The scoring in the TNT data matrix provided as supplementary material S4 in Santos [12] presents character-states starting from 1, but in the character-set (Appendix S3 of that work) all character-states start with 0.
2. Character 9 in the data matrix is supernumerary, and therefore does not correspond to character 9 in the character-set. It appears in the data matrix scored as [?] (= unknown) for all taxa.
5. Comments for character 98 in the character-set mention "State 3" but this is a typo; it should read "State 2".
Maximum-likelihood analyses were conducted with RAxML v8.2 [15], with the dataset partitioned by locus and using the GTR+ Γ+I model (as indicated by model testing in the original work). Clade support was assessed with 100 rapid bootstrap replicates.

Species delimitation
To objectively test if the observed intra-specific morphological variation would support more than one species, 24 varying features recognized among examined specimens (Table 1) were coded into a matrix and evaluated with a parsimony analysis using TNT v1.5 [16,17]. All specimens with some variation for the selected characters were included in the analyses. A species of Diapetimorpha Viereck, generally similar and from the same type locality as the new taxon, was used as outgroup. The key objective was to provide a precise and reproducible interpretation of the data, based on the most parsimonious solution for complex, overlapping morphological variations of a considerable number of specimens, which would otherwise be highly subjective to judge. The proposed ordering for character-states was hypothesized from the character analyses [18] of the tree obtained with all states first ran as non-additive. Based on that result, characters 0, 2, 3, 5, 6, 11-15, 17 and 21-23 were run as additive, and all others as non-additive (Table 1). Searches were performed using implied weighting [16]. The constant of concavity (K) was defined by the algorithm setk.run, written by Salvador Arias (Instituto Miguel Lillo, San Miguel de Tucuman, Argentina) according to the rationale of Goloboff et al. [19]. Searches were performed with Ratchet (1000 iterations), sect:slack 40, random seed 11, and hold 20000; all other variables remained set to default values.
To associate the putative males of C. metamorphus sp. nov. with the respective females, in light of an unusually high degree of sexual dimorphism for Cryptini, we sequenced a fragment of the mitochondrial gene cytochrome oxidase I using primers for the "barcoding" region (using LCO and HCO primers from Folmer et al. [20]) for one female (FAS333) and one male (FAS2144) collected in the same area (BRAZIL, ES, Cariacica, Duas Bocas, Alto Alegre, Primary Forest, 501 m, 20˚16'54.4"S, 40˚31'20.7"W) (female on 05-20.X.2016, male on 10-26. IV.2017). DNA extraction, amplification and sequencing were performed using the same protocols as in Scherrer [21]. The two sequences were aligned using the Geneious alignment algorithm (Geneious Prime v.2019.0.4, Biomatters), and compared in their pairwise distance to check for specific association.

Color data
Extraction of color features was performed according to an original procedure, as follows. All female specimens were photographed laterally (head + mesosoma) and dorsally (metasoma), with focus stacking (see Brecko et al. [22]). RGB values from uncompressed, lossless (TIFF) images (S1 File) were directly used as color data. Although RGB values are device-dependent (e.g. [23]), any given camera will produce highly comparable results, representing a much more precise proxy for true colors than descriptive or discrete coding based solely on the human eye. Most photographs (84%) used in this procedure were shot with the same equipment; the remaining images were obtained from two other cameras, but their colors were calibrated according to the method developed by Reinhard et al. [24], after adjusted for illuminance, which also interferes with RGB values.
To mitigate the illuminance problem, one of the images was selected as a standard, its background manually removed (i.e., erased to white), and the normalized average exposure of the image of the specimen itself was calculated as The exposure of the specimens depicted on each of the remaining images was then heuristically adjusted, using Bézier curves, to match as precisely as possible that of the standard. In this process, all specimens, except the standard, were roughly isolated from the background through a simplified algorithm-exclusion of the top 55% brightest pixels. The resulting images ended with nearly identical average exposure for the specimens, with an average difference of only 0.05%.
Color samples were then extracted from the images from 26 standardized points on each specimen ( Table 2). It is noteworthy that the key objective of each sampled point was to get a representative color for a given area of the body, such as the lateral portion of the gena, the lateral side of front coxa, etc. Therefore, it is not necessary that the sampled points correspond to precise anatomical equivalents for each specimen, i.e., morphometric landmarks.
To investigate the darkening of specimens (Fig 10), the perceived brightness of each color sample was calculated according to Finley [25], as ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi 0:2989R 2 þ 0:5870G 2 þ 0:1140B 2 p for linear 0-1 RGB values. This value ranges from 0.0 for entirely dark to 1.0 for entirely white.
The selected sampling points on each image were first registered as image coordinates, using the tpsDig software [26]. Actual RGB values were then extracted from pixels around each sampling point, within a radius of 2 pixels, thus equivalent to an area of 13 pixels. The representative RGB values used for each sample was the average value of each set of Rs, Gs and Bs from the corresponding set of 13 pixels. The resulting matrix, of 50 specimens × 78 variables, has 174 missing values, due to damaged or missing parts, and 3726 informative cells (S1 File). Altitude and latitude data were then superimposed on the resulting PCA, on a 3D chart

PLOS ONE
Species delimitation, environmental cline and phylogeny for a new genus of Cryptinae (Fig 9). Exposure adjustments, RGB sampling, and graphics, were all performed or built with Python scripts written for these specific aims.
Since some of the RGB sample points were directly equivalent to color characters coded for the morphological matrix (Tables 1 vs. 2; characters and body areas marked with a plus sign), an alternative character matrix was assembled by replacing all 9 characters marked on Table 1 with data from the 12 equivalent RGB points indicated in Table 2 (that is, 36 RGB characters), as well as by adding all of the remaining 15 RGB points (45 characters) (S2 File); the aim of this matrix was to produce a comparative interpretation (Fig 6).

Intraspecific variation
To investigate for the existence of a latitudinal vs. altitudinal cline in morphological variation, a Multiple Factor Analysis (MFA [27,28]) of the morphological variation matrix (categorical data), including RGB values (continuous data) was performed. An MFA is more accurate than Principal Component Analyses (PCA) when categorical and numeric data are to be analyzed together. The MFA was performed with the R package PCAmixdata [29].
In order to evaluate the degree of association of altitude and latitude with the observed morphology, each of these two variables were compared with the categorical (Table 4) and the continuous (S1 File) characters. The available latitude data, however, is clumped into four distinct subgroups (Fig 5A), preventing its investigation as a continuous variable with correlation indexes (see Aggarwal & Ranganathan [30]). Because of that, latitude data was investigated only as a categorical, ordered variable, with four states: [0] 0 to 5˚N, [1] 0 to -5˚S, [2] -10 to -15˚S, [3] < -15˚S. The correlation between continuous vs. categorical variables was calculated with point-biserial correlation (Sheskin [31], here implemented in Python with the library scipy. stats.pointbiserialr) for binary characters, and with polyserial correlation [32,33] (implemented in R with the package polycor, with ML = TRUE) for characters with three or more states. For assessing the correlation between continuous variables, such as altitude vs. RGBs, the Pearson's product moment correlation was calculated; all prerequisite conditions listed by Aggarwal & Ranganathan [30] were graphically checked in all cases; where doubtful, linearity was further verified by comparing multiple curve fits with the Akaike information criterion [34].
Correlation between sets of categorical variables was measured using the Cramér's V index [35], calculated using a custom Python script extracted from the dython library, by Zychlinski [36]. Since all mentioned indexes are somewhat related (they are all conceptually derived from the Pearson's index) and have equivalent ranges for their absolute values (0.0 to 1.0), comparative plots were presented combining different correlation indexes (Fig 7B-7C).
A separate PCA for color features was performed, using data from nearly all available female specimens, according to the procedures described in the item Color data, above. Missing values were handled through iterative imputation [37].

Ecological niche modelling
The potential distribution of the species was modeled from all known occurrence records, using climatic data obtained from the Inpe database (www.dpi.inpe.br), which includes all Worldclim variables and others, with 57 variables total, each one represented by monthly data from 1950 to 2000. Modeling was performed using MaxEnt [38], with 30% random test percentage, 30 replications, the standard convergence limit , maximum number of iterations (5000), and other standard resources. From these results, we further analyzed the top 15 variables which contributed most to the results (as measured by MaxEnt) with a PCA, to measure the degree of correlation between them. Pairs of variables were compared according to the highest values for PC1; if correlation for a given pair was larger than 0.7, the contribution of each variable was recalculated with jackknifing, and that with the lowest contribution was discarded. This was repeated until only 4 variables remained-this number follows a general rule of thumb that sample size (number of unique occurrences; 43 herein) should be 10 times larger than the number of predictors used for modelling [39]. To these, we also added the variable altitude, because of its relevance in this study. PCA and correlation of variables were performed with the R packages raster, rgdal and vegan. The following five variables, with a resolution of 2.5 minutes of arc, were used: alt (altitude), bio3 (isotermality, obtained as bio2/ bio7 � 100), decl (declivity), prec9 (total precipitation of August), and prec12 (total precipitation of December). In order to evaluate distinct ecological scenarios, we calculated two different occurrence thresholds, one considered to be very conservative (Maximum Training Sensitivity Plus Specific) and one more inclusive (Minimum Training Presence).

Phylogeny
Seven molecular loci were sequenced (Fig 2). In addition, the character-coding for the external morphology of Cryptoxenodon gen. nov. is shown in Table 3.
The results strongly support the classification of the new taxon as a separate, independent genus, phylogenetically distinct and morphologically diagnosable from both Debilos and Diapetimorpha, as well as from all other Neotropical taxa.

Species delimitation
Morphological variation among the examined specimens for the new genus proved to be extensive, and occurrence records are clustered in two apparently disjunct population groups. The obvious hypothesis to be tested was that two or more species were present among the examined material. The strict consensus of the most parsimonious tree for the coded characters (Table 4) recovered clades containing representatives of widely separated populations, such as ES + BA + GUI (Fig 4A), or no convincing clades at all (Fig 4B). A PCA analysis of the mixed character matrix (Fig 6) also did not recover discrete groups. These results indicate that there is no reasonable support for splitting, pointing instead to a single, even if variable species.
Comparison of the COI sequences obtained for male and female specimens from the same locality (20˚16'54.4"S, 40˚31'20.7"W) revealed 100% sequence identity (GenBank Accession numbers

Altitudinal and latitudinal clines
The available data for latitude and altitude data were not linearly related, with latitude data showing distinct subgroups ( Fig 5A). This prevents the usage of a correlation coefficient (see [30]), making it difficult to state precisely how much latitude is independent from altitude in this particular study. The categorical morphological variables (Table 4) correlated consistently higher with altitude than with latitude (average correlation with altitude = 0.393207; with latitude = 0.243840; average difference = 0.149367) (Fig 5B), but results considering exclusively the color features (continuous RGBs) do not distinctly favor one variable over the other, with only 0.014824 average difference favoring latitude (average correlation with altitude = 0.356327; with latitude = 0.371151) (Fig 5C). In all cases there is a wide range of correlation values, ranging from nearly 0.0 to about 0.8.
The MFA analysis (Fig 6) shows some stratification for specimens from both different altitudes ( Fig 6A) and latitudes (Fig 6B), albeit this is evident only for the first axis (Dim1). This is perhaps most evident in the plot with altitude + latitude data overlaid on the PCA results for RGB data from all females (Fig 7). Darkening in at least 26% (7/27) of the investigated body areas, however, increases significantly with altitude (P < 0.05) (Fig 8, first seven subplots), but only in one case it shows a clear (P = 0.003) but very weak opposite response (last subplot).
These results seem therefore consistent in supporting a dual cline, with populations being concurrently affected by both altitude and latitude, albeit possibly a little more intensely affected by altitude, as suggested by the differences in Fig 5B.

Phylogeny
Except by the exceptionally long mandibles, C. metamorphus sp. nov. is externally remarkably similar to species of Diapetimorpha. However, the combined morphological and molecular phylogenetics analysis was unequivocal that the new taxon falls out of the Diapetimorpha clade, which was represented by four species. In fact, C. metamorphus sp. nov. was shown to be most closely related to Debilos, providing further support for its treatment as a distinct genus.
In the analyses of Santos [12], the position of the clade Diapetimorpha+Debilos was variable within the Cryptini tree, in some analyses recovered as sister to the Lymeon group (and therefore arguably part of it), in others as sister to a lineage containing three major clades (Osprynchotus group + (Trachysphyrus group + Agrothereutes group)). Herein a third result was recovered, with the clade (Diapetimorpha + (Cryptoxenodon + Debilos)) recovered as sister to an even broader clade (S3 File). This instability, combined with the relatively low bootstrap values at higher levels of the phylogeny, highlights the challenge in reconstructing higher-level cryptine phylogeny even after sequencing multiple genes. In phylogenomic analyses using ultraconserved elements [42], the clade (Diapetimorpha + Debilos) was recovered as sister to the Lymeon group, but those analyses had much smaller taxon sampling (70 cryptine species).

PLOS ONE
Species delimitation, environmental cline and phylogeny for a new genus of Cryptinae

Species delimitation
The combination of multiple sources of evidence and analytical frameworks did not support the recognition of more than one species of Cryptoxenodon gen. nov. Although the specimens examined display a considerable amount of morphological variation, no method could point out to a discrete division among the observed populations. The pairwise genetic distance between the sequenced specimens, from French Guyana and southeastern Brazil (5.7%), was higher than the 3% threshold often used to separate species by DNA barcoding initiatives (e.g. [43]), but high levels of intraspecific divergence for COI sequences are not uncommon among insects, including studies recording 30.8% for cockroaches [44], 31.15% for thrips [45], 21.8% for mosquitoes [46], 18.3% for praying mantises [47], and 17.5% for Drosophila [48]. Sequencing of only two specimens does not allow for further inferences of population structure in C. metamorphus, but further sequencing efforts were not part of the original aim of this work, and would not be possible for most specimens due to previous processing with ethyl acetate, that destroys DNA [49]. Compared to other cryptine genera, C. metamorphus sp. nov. shows an unusual level of sexual dimorphism (see Taxonomy section for a detailed account), complicating female-male associations. The 100% identity observed in COI sequences represents unequivocal corroboration of the morphology-based association and will help in future morphological interpretation of similar character systems in other variable species.

Morphological clines
Results presented above make a strong case for a dual cline, altitudinal and latitudinal, with a slightly more intense influence of altitude. The underlying reasons for this, however, are not obvious. Water loss, temperature, and UV light, for example, might be involved.
Altitude. Higher altitudes are not necessarily directly linked to reduced humidity, but elevated areas, such as hills or cliffs within a forest, can sometimes be drier than the plateau (e.g.  [50]). This is significant because ichneumonids are especially vulnerable to desiccation [51], and need to drink water daily [51,52]. Shapiro and Pickering [53] also noted that parasitoids of larval Lepidoptera, such as many Ichneumonidae, may be further affected by desiccation by having to search for hosts up on the foliage, where it is dryer than near the soil.
It has been suggested that increased melanization may reduce cuticular permeability and allow insects to better resist desiccation [54], and such an effect was shown to appear rapidly from selection in laboratory experiments [55][56][57][58], suggesting that melanism and desiccation resistance are linked through differences in cuticular permeability [54].

PLOS ONE
or latitudes, because they heat up faster at a given level of solar radiation. This efficient use of radiation for heating allows an increased period of activity for dark individuals in cold environments, which may in turn increase success in feeding, escaping predators, and finding mates [60,62], and even increasing fertility [63].
Dark coloration at higher altitudes also may relate with protection against UV radiation, which increases with altitude [64], and intense ultra-violet may be a contributing factor in the production of dense pigments [65], which serve to protect the deeper and delicate tissues from injury by this radiation.
While the melanism-desiccation hypothesis or the thermal melanism hypothesis seem inadequate to explain variations such as cuticular microsculpturing or wing venation features, they are both consistent with our data, which shows darker individuals of C. metamorphus sp. nov. in higher altitudes and colder areas. Similar altitude-related color variation is also reported for other insect groups, such as butterflies [60,66] and grasshoppers [67,68]. For C. metamorphus sp. nov., however, the average yearly temperature for the areas occupied by the northern populations is 24.8˚C, versus 21.0˚C for the southeastern populations, a much smaller variation (3.8˚C) than that investigated in previous studies, e.g. 15.0˚C in [59].
Latitude. Latitudinal clines, on their turn, are also reported for multiple aspects of the life of Hymenoptera, such as diversity of parasitoid bees and ichneumonids [69,70], guild composition of parasitoid wasps [71], diapause in Braconidae [72], nest architecture in Vespidae [73], rates of predation in ants [74], and even dramatic variations in sex-ratio, for Pelecinidae [75][76][77]. Published investigations on the relationship between morphological traits and latitude, however, are usually limited to testing Bergmann's rule, i.e., that size increases with latitude [78]. For the Neotropical region, it seems that only Oliveira et al. [79], working with leafhoppers (Cicadellidae), went somewhat further by showing that individuals from higher latitudes were not only larger and heavier, but also darker than those from lower latitudes. Increasing latitudes generally imply lower temperatures, which are also characteristic of higher altitudes. This could suggest that the effect of latitude and altitude are similar, but temperature apparently is not a key factor for C. metamorphus sp. nov., as discussed above, and higher altitudes also affect a series of other variables to which insects are quite sensitive, such as humidity, intensity of UV light, atmospheric pressure, oxygen concentration, wind, and others.

PLOS ONE
Species delimitation, environmental cline and phylogeny for a new genus of Cryptinae (Figs 11A and 14), about as long as second mid tarsomere. Ovipositor distinctly compressed, dorsal valve with nodus and pre-apical notch, ventral valve apex with few minute serrations (Fig 15D and 15E).
Description. Body surface matte. HEAD. Clypeus outline in lateral view distinctly convex on basal half, apical portion truncate; clypeal margin medially convex, even if slightly, without teeth, laterally without projections. Mandible ventral margin not projected as a flange or crest; ventral tooth longer than dorsal tooth. Flagellum subapical width enlarged, distinctly greater than the rest of flagellum; apical flagellomere apex regular, uniformly tapered. Supra-antennal area outline strongly concave throughout, medially with a low, suture-like median longitudinal line; supra-antennal horns or tubercles absent. Gena ventrally from about as wide as at eye midlength to somewhat swollen, distinctly wider than at eye midlength. Vertex dorsomedial THORAX. Pronotum dorsal margin outline regular, not or only slightly swollen; central portion flat or slightly concave, without a transverse sulcus; pronotal collar posterior margin not bordered by a carina; epomia short or moderately long, very delicate to moderately strong, usually  WINGS. Forewing hyaline to moderately infuscate, sometimes with 1-2 dark spots, even if weak; short vein projection (ramellus) at junction of forewing veins 1m-cu and 1-Rs+M from completely absent to present, long; crossvein 1cu-a from basad to slightly apicad base of

PLOS ONE
Species delimitation, environmental cline and phylogeny for a new genus of Cryptinae 1M+Rs; cell 2Cu (first subdiscal) approximately trapezoidal, apically distinctly longer than basally; vein 2Cua distinctly longer than crossvein 2cu-a, sometimes nearly the same length of 2cu-a; veins 2Cua and 2cu-a angled, even if slightly; crossvein 2m-cu distinctly convex, with one bulla, placed centrally; crossvein 3r-m entirely spectral or nebulous, almost indistinct; crossveins 2r-m and 3r-m parallel or nearly so; crossvein 2r-m about as long as crossvein 3rm; vein 2-M distinctly longer than 3-M.
Hind wing vein M+Cu subapically distinctly convex; vein Cua distinctly longer than crossvein cu-a; veins Cua and 1-M forming approximately right angle; vein 2-Rs complete, reaching wing margin; vein 1-R1 undifferentiated; vein Cub complete, reaching wing margin or almost so; vein Cub apical half distinctly convex; vein 2-1A distinct, reaching more than half the way to wing margin.

PLOS ONE
Species delimitation, environmental cline and phylogeny for a new genus of Cryptinae dorsolateral carina complete, except weak to indistinct on smallest specimens; median dorsal carina distinct until the spiracle, or absent (smallest specimen); ventrolateral carina absent; postpetiole usualy not distinctly bent, at most forming weak angle with petiole; T1 subapically slightly wider than at posterior margin. Thyridium distinctly longer than wide or distinctly wider than long. T7-8 in lateral view of similar length, shorter than T5-6. Ovipositor in profile moderately slender, overall shape straight or nearly so, distinctly compressed. Dorsal valve without minute punctures throughout, its apical surface smooth; nodus moderately tall and distinct, giving triangular shape to apex; notch distinct; ovipositor tip moderately pointed. Ventral valve apex not projected dorsally as a lobe, smooth; tip with serrations, which are approximately vertical, uniformly arched.

PLOS ONE
Species delimitation, environmental cline and phylogeny for a new genus of Cryptinae transverse carina medially almost always with two strong and irregular longitudinal carinae; propodeal sculpture distinctly coarser than in female. First metasomal segment slender than in female, with T1LW 4.02-5.08 (vs. 1. 70-2.20), and distinctly less widened apically, with T1WW 1.44-1.69 (vs. 2.10-3.00). There are also considerable differences in color patterns, as detailed in the description of the species.
Comments. Cryptoxenodon gen. nov. can be readily distinguished from all other known Cryptinae by the unusually stout, large mandibles (Fig 10), 2.45-2.95 as long as basal width, with the ventral tooth distinctly longer than the dorsal one. A diagnostic key is also provided below. There are several other cryptine genera with long mandibles, including all nine genera of Townes' subtribe Osprynchotina (= Nematopodiina) and Dotocrytpus Brèthes. In these genera, however, the mandibles are almost always slender and distinctly tapered towards the apex, sometimes sickle shaped. Furthermore, in all those genera except for the distinctive Dotocryptus Brèthes, the dorsal tooth is longer than the ventral one.
It remains to be seen whether the large mandibles will hold as diagnostic for other eventual species in the genus. This could represent an isolated specialization, as with the markedly large and bent mandibles of Gabunia flavitarsis Kriechbaumer, the only species to show this trait out of eight known for the genus.
Specimens of Cryptoxenodon gen. nov. are otherwise similar to Diapetimorpha Viereck and will run to that genus in Townes' key [7]. The new genus can be additionally separated from Diapetimorpha by having fore wing vein 2-M much longer than 3-M, i.e. the crossvein 2m-cu arises apicad to the middle of the areolet (vs. 2m-cu always basal or at midlength, 2-M smaller than or as long as 3-M); ventral tooth distinctly longer than dorsal tooth (vs. slightly shorter than dorsal tooth); clypeus nearly flat, confluent with supra-clypeal area (vs. rather evenly convex except somewhat flattened or impressed near apical margin); female flagellum subapical width enlarged, distinctly greater than the rest of flagellum (vs. regular, more or less uniform with rest of flagellum); supra-antennal area outline strongly concave throughout (vs. distinctly concave on ventral half); mesosoma more elongate, about 1.80 × as long as tall (vs. often shorter, around 1.50-1.60 × as long as tall). Cryptoxenodon gen. nov. is also similar to some species of Debilos Townes 1966, from which it can be differentiated by having the first tergite with an anterolateral tooth (vs. absent); ventral tooth distinctly longer than dorsal tooth (vs. about half of dorsal tooth length); and areolet closed, with crossvein 3r-m spectral or sometimes nebulous (vs. areolet open, 3r-m absent).
Biology. Unknown. Etymology. From the Greek xenos, meaning strange, foreign, and the Greek suffix -odon, for tooth. The name is a loose reference to the unusual mandible, apparently unique within the Cryptinae.

PLOS ONE
PROPODEUM (Figs 11A, 12A-12C and 14). Somewhat short, as long as wide, glabrous, coarsely and richly sculptured. Anterior transverse carina (ATC) with two diverging carinae arising centrally and reaching propodeal furrow; area anterior to ATC from coarsely rugulose to microareolate near spiracle; area behind ATC from transversely to somewhat concentrically, coarsely, areolate-rugose; median longitudinal carina weakly distinct from ATC to level of apophyses, partially confluent with surrounding rugosities; lateral longitudinal carinae partially distinct, confluent with surrounding rugosities; propodeal area sloped at about 90 degrees at the level of this carina. Propodeal spiracle elliptic, SWL 1.30. Propodeal tubercles large, subtriangular, about same length as second mid tarsomere. WINGS (Figs 9 and 11B-11C). Fore wing 1-Rs+M with bulla placed distinctly before its midlength; ramellus quite short but distinct; crossvein 1cu-a distinctly basad (right wing) or opposite origin of 1M+Rs; vein 2Cua 1.25 × as long as 2cu-a; bulla at crossvein 2m-cu short, about 1/5 of the crossvein length; cell 1+2Rs (areolet) moderately small, APH 0.95, transverse pentagonal, distinctly wider than long, AWH 0. 75 . The morphological variation is extensive and related to both altitudinal and latitudinal clines. In general, low altitude specimens (below 500 m) tend to show lighter tones, with orange on legs and metasoma, and T5-7 mid-apically sometimes with whitish spots, while high altitudes lead to nearly fully dark specimens, but the full range of variation is complex (Table 4; see also item Discussion).
In nearly all specimens the anterior mid-longitudinal carinae of propodeum is V-shaped, but in specimens from Santa Maria de Jetibá (ES) it is distinctly Y-shaped. The posterior transverse carina is sometimes present only medially as a strong wrinkle, which may also be connected with the short mid-longitudinal carina ( Fig 14A); more rarely, the posterior transverse carina occurs as a weak wrinkle extending all the way between the apophyses.
Variation (male) (Figs 18A-18B, 19A-19B, 20). Antenna with 26-32 flagellomeres; posterior transverse carina of propodeum medially sometimes briefly interrupted; longitudinal carinae of propodeum arising from ATC towards propodeal sulcus varying in shape as in female; in about 65% of the specimens propodeum with a short mid-longitudinal carina between anterior and posterior transverse carinae ( Fig 20C); scutellum sometimes with a posterior or central whistish spot, or more rarely entirely whitish ( Fig 20A); scutelar carina, postscutellum and propodeal apophyses sometimes whitish; three specimens from PA and three from BA with T4 extensively yellow; posterior yellow stripes of T2 and T3 sometimes narrow and almost indistinct; T4 and T5 rarely with a yellow stripe on anterior margin, or with yellow marks posteriorly; colour pattern of hind coxa varying from light orange to blackish (Figs 17, 18A, 19A).
Comments. Males of C. metamorphus sp. nov. can be easily confused with males of a number of other Cryptinae taxa that have generally similar color pattern, in particular species of Diapetimorpha, which also show an anterolateral tooth on the T1; males of that genus also often show different color parttern from the respective females and remain unassociated in most cryptine samples. Males of C. metamorphus sp. nov. can also be mistaken for males of Basileucus Townes, for which a diagnostic feature is having the T1 anteriorly whitish. Nonetheless, males of C. metamorphus sp. nov. can be differentiated from these other taxa by having the ventral tooth of the mandible distinctly longer than the dorsal one (vs. slightly to distinctly shorter); mandible long and generally similar in shape and structure to that of the female, MLW > 2.20 (vs. short to moderately long, MLW < 2.00); and the areolet usually distinctly wider than long (vs. about as long as wide) (see Figs 18 and 19).
Etymology. From the Greek meta, meaning over, beyond, and the Greek morphus, meaning form, shape; a reference to the extreme morphological variation of the species.
Occurrence. Forested areas in Amazonia (Colombia, Brazil, French Guiana) and the Atlantic Forest (Fig 21). Apparently restricted to tropical areas: no specimens have been found in samples from southern, subtropical portions of Brazil. Ecological niche modelling (Fig 22), using either a conservative (Fig 22B) or inclusive threshold (Fig 22C), shows that the species appears to be mostly concentrated in two main areas, one in central + northern Amazonia (themselves somewhat isolated by a gap), and another ranging through most of the Eastern coast, in the Atlantic Forest, and reaching Paraguay. The seasonal records of all known females (Fig 23) indicate that flying adults in the southern hemisphere occur with similar frequency in all seasons except winter (Fig 23A), with an occurrence peak apparently in mid spring. The abundance of the species also seems quite uniform in different altitudes (Fig 23A) or latitudes (Fig 23B).