Comparative geometric morphometrics of male genitalia in Xiphocentron subgenera (Trichoptera: Xiphocentronidae): new species, revision and phylogenetic systematics of the subgenus Sphagocentron

Geometric morphometric statistics have been employed to reduce the subjectivity of visual evaluations in taxonomy. Taxonomy in most insect groups relies strongly on male genitalia morphology which is often the structure with most data available, which is also true to caddisfly taxonomy. Here we revise the caddisfly subgenus Xiphocentron ( Sphagocentron ) adding five new species after 40 years: X. dactylum sp. nov. , X. eurybrachium sp. nov. , X. tapanti sp. nov. , and X. tuxtla sp. nov. Additionally, we describe a new X. ( Antillotrichia ): X. drepanum sp. nov. from French Guiana and provide new species records of Xiphocentronidae from Bolivia, Costa Rica and Ecuador. We performed exploratory geometric morphometric analysis on the male genitalia’s preanal appendage to characterize the shape differences among the species, and to investigate its utility to classify species to subgenera. In order to infer species relationship and assess if shape congruences are due to phylogenetic signal or convergence data from 100 landmarks and semilandmarks, and 30 discrete characters were used to generate a phylogenetic hypothesis. The morphometry partially supports the subgenera delimitations, but the Antillotrichia subgenus greatly overlapped with other subgenera. The discriminant analysis overall classification correctness was 64%. Some suggested phenotypic groups were due to convergence. According to the preanal appendage morphometry, X. ( Antillotrichia ) fuscum is a Sphagocentron species. The phylogenetic analysis recovered Sphagocentron as monophyletic, but not Antillotrichia . Sphagocentron subgenus was placed within a clade of several Antillotrichia species, with X. ( A. ) fuscum as the sister of the other Sphagocentron species, although support values were low.


Introduction
Male genitalia features play an important role in insect taxonomy as they contain the principal traits used to delineate genera and delimit species in many insect groups, and often the male genitalia is the structure with most data available (Shapiro and Porter 1989;Hosken and Stockley 2004).Moreover, the genital morphology has the potential to directly contribute to the reproductive isolation and speciation process through divergent sexual selection and mechanical or sensory incompatibility (Eberhard 1992;Arnqvist 1998).Morphometric statistics have been advocated as a tool to reduce the subjectivity of visual evaluations in taxonomy (Mutanen and Pretorius 2007).Geometric morphometrics, for instance, allows detailed analysis and visualization of shape changes in complex structures, removing size variation from the data through superimposition, while capturing all aspects of shape variation (Bookstein 1997;Rohlf and Marcus 1993).This constitutes a sophisticated method for collecting and analyzing data to address questions in anatomy and evolutionary biology (e.g., Klingenberg 2016).Genital morphology is the second most studied insect body part in morphometrics (after wings), with male genitalia accounting for most studies (Tatsuta et al. 2018;Mutanen and Pretorius 2007;Jauset et al. 2017;García-Román et al. 2019).Studies using traditional morphometry on adult caddisflies (Trichoptera) are uncommon, with notable exceptions of Goretti et al. (2005), Svensson (1975), and Bálint et al. (2008) who focused on male genitalia structures.Despite the relevance of genitalia characters in species delimitation, geometric morphometry was so far applied only to caddisfly wings by Sganga et al. (2022).
Since Nielsen (1957), several studies have discussed the morphology of Trichoptera male genitalia (e.g., Holzenthal et al. 2007;Ivanov 2005;Morse 1975; Ross and Unzicker 1977;Schmid 1970Schmid , 1979Schmid , 1989;;Oláh and Johanson 2008).The genitalia is associated with abdominal segments IX, X and XI, and bears several paired appendages: the ventral inferior appendages (gonopods), often bi-articulated, are thought to have a clasping function during copulation; the mesal intermediate appendages (paraprocts) usually perform accessory copulatory and stimulatory functions, and are often fused to other structures (Oláh and Johanson 2008;Nielsen 1957); and the dorsal superior appendages (preanal appendages) are setose and thought to have a sensory function.Some authors infer that these superior appendages are homologous to the cerci (Ross 1938;Ivanov 2005;Oláh and Johanson 2008) while others have stated that Trichoptera do not possess cerci (Holzenthal et al. 2007;Nielsen 1957;Schmid 1998).The preanal appendage is often elongate, and digitate, but can be reduced to a knob-like structure or even be absent or fused to the paraprocts and tergum X (Nielsen 1957;Oláh and Johanson 2008).
The majority of xiphocentronid species are found in small streams in the tropical zone, primarily in the Oriental region (with five genera) and the Neotropical region (with three genera), few species are also distributed in the Afrotropical, East Palearctic and Nearctic regions (Vilarino and Bispo 2020;Vilarino et al. 2021).The adults are small with forewings ranging from 2.5 to 8.5 mm uniformly brown or with one to three white patches on the forewing, and are often active during the day (Flint 1968;Schmid 1982).The family is inferred to have diverged from the sister group, Psychomyiidae, in the early Cretaceous (Thomas et al. 2022;Malm et al. 2013) although the only xiphocentronid fossil known is from Miocene Mexican-amber, Xiphocentron chiapasi Wichard et al. 2006.The male genitalia of the Xiphocentronidae is characterized by: (1) the great horizontal stretching of the genitalia; (2) the reduced tergum IX; (3) the preanal appendages very long and robust; (4) the fused segment X and paraprocts closed ventrally forming a support under the phallus; (5) the phallus reduced to a very short phallotheca and an extremely long and slender cylindrical aedeagus, while the endotheca is completely obliterated (Schmid 1982).The 2nd and 3rd characters are not exclusive to the xiphocentronids, but represent a development taken to the extreme of tendencies already present in certain Psychomyiidae (Schmid 1982).
Currently, the New World fauna comprises 81 species, most of them belonging to the genus Xiphocentron, making it the most species-rich genus within the family with 62 species (Vilarino et al. 2023;Bueno-Soria et al. 2022).Xiphocentron circumscription is based mostly on the absence of characteristics seen in other genera, although the established subgenera in general have defining traits.The genus was classified by Schmid (1982) into five subgenera: Glyphocentron Schmid, 1982 (3 spp.);Rhamphocentron Schmid, 1982 (6 spp.);Sphagocentron Schmid, 1982;Xiphocentron Brauer, 1870 (6 spp., including 1 fossil sp.); and Antillotrichia Banks, 1941 (47 spp.) (senior synonym of Xirocentron Schmid, 1982(Botosaneanu 1988)).The last subgenus Antillotrichia was designated to accommodate species that could not be placed in other subgenera (Schmid 1982), so it might not be monophyletic.The Antillotrichia subgenus is the most heterogeneous among the subgenera established, so Schmid (1982) suggested that it would be subdivided as more species are included.Some characters of each subgenus based on Schmid (1982) descriptions are shown in Table 1.
All the five subgenera occur in Mesoamerica and Central America, while only species placed within Antillotrichia occur in the Caribbean islands and South America (Holzenthal and Calor 2017;Vilarino and Bispo 2020).So far, only two species were placed within the Sphagocentron subgenus: Xiphocentron evandrus Schmid, 1982 from Costa Rica andPanama, andX. julus Schmid, 1982 from Mexico and Panama.Sphagocentron is characterized mainly by the angularly curved inferior appendages, the inner face of the harpago with a large area of fine setae and the simple segment X (including the paraprocts) (Schmid 1982).
In this study we revise the subgenus X. ( Sphagocentron), providing identification key, the synopses of the two previously described, X. jullus and X. evandrus, and the description of five new species, X. (Sphagocentron) dactylum sp.nov.and X. (S.) eurybrachium sp.nov.from Venezuela, X. (S.) tapanti sp.nov.from Costa Rica, and X. (S.) tuxtla sp.nov.from Mexico.In addition, we describe one species, X. (Antillotrichia) drepanum sp.nov., from French Guiana and include new distributional records of Xiphocentronidae from Bolivia, Ecuador, and Costa Rica.
We performed exploratory geometric morphometric analysis on the male genitalia's preanal appendage characterizing the shape differences among the species described, as well as among the five subgenera, in order to investigate the utility of morphometrics to classify species to subgenera.Additionally, we performed a phylogenetic analysis of X. (Sphagocentron) species combining morphometric and discrete characters, to test Sphagocentron subgenus monophyly, new species placement, and to verify if the preanal appendage shape congruences are due to phylogenetic signal or convergence.

2.
Material and methods

Male genitalia morphology analyses
New species propositions are testable hypotheses of the distinction of separately evolving metapopulation lineag-es (Pante et al. 2015).Here, the new species hypotheses were based on distinct male genitalia characters which were used as an indirect indicator of reproductive isolation and evolutionary divergence.In order to visualize internal structures of the adult male genitalia, the entire abdomen of the specimen was removed and diaphonized using 85% lactic acid (Blahnik et al. 2007).The abdomens were then placed in excavated glass slides with a drop of glycerin and examined with a compound microscope at 200-400× magnification.The abdomens of the specimens were stored in microvials with 80% ethanol and kept with the specimen to which they belonged.All species were conserved in 80% ethanol.The species drawings were made by tracing the observed structures in pencil using a camera lucida attached to a compound microscope or taking multiple photographs with a digital camera attached to a microscope.Then the drawings or the photographs were used as a template and digitally traced using Adobe Illustrator CS6 software.The distribution maps were created with the software QGIS v2.8.2.
The morphological terminology for male genitalia followed Vilarino and Bispo (2020).The terminology for wing venation followed the Comstock -Needham system as interpreted for Trichoptera by Mosely and Kimmins (1953).In the descriptions, paired structures are referred to in the singular form.
Types of the species described herein and other material examined are deposited, as indicated in the species descriptions, in the following institutions: BIOUG -Centre for Biodiversity Genomics, University of Guelph, Ontar-

Geometric morphometric analyses
Geometric morphometry allows quantitatively evaluating the shape variation of morphological structures across a sample using standardized images and 'landmarks' (Bookstein 1991;Rohlf and Marcus 1993).For the analysis the datasets included the male genitalia of the species in Table 2 and the genitalia depicted in the original descriptions of X. julus and X. evandrus.Tps files were created from the specimens illustrations using the software TpsUtil v.1.81(Rohlf 2021).The landmarks were digitized using TpsDig2 v.2.31 (Rohlf 2017).
Curves were traced along the preanal appendage in lateral view.We utilized 97 curve points (semi-landmarks) and three landmarks aiming to get most of the shape information (Fig. 1).Curves were appended to landmarks and semi-landmark sliders files were generated with TpsUtil (File S1).All samples were aligned using Generalized Least Squares (GLS) Procrustes superimposition (Bookstein 1991) after importing the data and sliders file into TpsRelw v.1.75(Rohlf 2021).This removed any variations in the scale, position, and orientation of the landmark coordinates.The aligned procrustes were imported into PAST 4.09 software (Hammer et al. 2001), which was used to generate shape thin-plate splines, and perform statistical analyses.To verify the congruence of the preanal appendage shape within genera, possibly misplaced species and suggested phenotypic groups we used three exploratory analyses: The Principal Component Analysis (PCA), was used to characterize the ordination of the specimens irrespective of predetermined grouping, so, it is more conservative about the assumption of shape congruence and distinction between subgenera, it also was used to visualize main contributors of shape change.We also used Discriminant Analysis (DA) in which data is addressed to a priori groups, this analysis maximizes the between-group variance and minimizes the within-group variance, therefore highlighting group differences and being more conservative about the assumption of mis- placed species.Discriminant Analysis is affected by the sample size about the number of variables, so the analysis was made using only the ten first Principal Components (PCs).Jackknifed group assignment was verified to see the correctness of subgenera classification based on the preanal appendage shape.No formal statistical test was performed since the sample for some subgenera is rather small.The shape congruence and distinction among subgenera was observed through the degree of morphospace superimposition in the convex hulls.The overall similarities between the specimen's shape were summarized through a neighbor-joining cluster analysis based on all the PCs, which was utilized to visualize suggested phenotypic groups.

Phylogenetic analysis
The phylogenetic analysis used the same 29 taxa as the morphometric analyses (Table 2), using 30 discrete morphological characters (five non-informative) and the geometrical morphometric characters from the preanal appendage landmark configurations (Table 3).From the discrete characters, 15 were adapted from Vilarino et al. (2022).Morphological characters and character state constructions were based on Sereno (2007).When a given structure was not present in the analyzed specimens, it was coded as '-'; when the character state was not clear or could not be assessed, it was coded as '?'.All characters were binary.The morphological dataset matrix was built using Winclada 1.89 (Nixon 2002) (File S2).
The TPS file with the aligned procrustes was imported into TNT 1.6 (Goloboff and Morales 2023) and combined with the matrix of discrete characters as detailed in Catalano and Goloboff (2018) (File S3).The analyses were made treating each landmark configuration as a different character (TNT default), landmark optimization, and landmark branch swapping settings also were used in default.The taxon Xiphocentron aureum was set as an outgroup for three rooting.The phylogenetic analysis was performed using implied weighting with the rescaled k, so that the minimum/maximum homoplasy weight ratio is 1 to 10 as implemented in TNT 1.6., with a resulting k = 5.54.Heuristic searches were performed through 'Traditional Search', with 500 replications, five trees saved per replication.The branch support was measured using symmetric resampling (Goloboff et al. 2003) expressed as the difference in the CG (contradicted/present groups) frequency (100 replications).Clades with support above 60 were considered well supported.The trees were visualized using Winclada 1.89 (Nixon, 2002) and edited in Adobe Illustrator CS6.

Preanal appendage morpho metrics
The exploratory PCA resulted in 28 PCs.The distribution of the species in the morphospace was presented over the four first PC axes (90.59% of the total variance).The PC1 (49.74%) and PC2 (17.07%) could partially separate among Sphagocentron, Rhamphocentron and Xiphocentron subgenera, but Antillotrichia mor pho space overlapped most of the species.Some Sphagocentron were highly separated from all other subgenera by the PC2, with a wide preanal appendage bearing an acute apex.The PC3 (13.51%) and PC4 (10.27%) strongly separate Xiphocentron subgenus from all other subgenera, with the species having preanal appendage mostly straight with wide base tapered to a narrow apex.The PC4 also separates Sphagocentron from Rhamphocentron, but with Antillotrichia overlapping with both subgenera.Showing that the preanal appendage morphology allows the sub-generic distinction between Sphagocentron, Rhamphocentron and Xiphocentron species, but not between Antillotrichia and most other subgenera, except Xiphocentron subgenus.The main contributors to the shape changes in each PC are presented in the thin-plate splines at Figure 2.
The divergence of the preanal appendage shape among subgenera with groups defined a priori was visualized through the discriminant analysis over the 10 first PCs (Fig. 4).After the discriminant analysis adjusted the data to maximize between-group differences the subgenera were in general well separated except by the overlapping of X. (S.) julus within Antillotrichia and X. (A.) fuscum within the Sphagocentron subgenus morphospace.The jackknife cross-validation test had an overall correctness 64.29%, and was able to correctly classify 71% (5/7) of the Sphagocentron, 73% (11/15) of the Antillotrichia, 33% (1/3) of the Rhamphocentron, and 33% (1/3) of the Xiphocentron subgenus.

Phylogeny
The analysis ran for around four hours and examined 8,641,321 rearrangements, with one most parsimonious topology retained with a score of 3.91330.The tree topology and the distribution of the analyzed species are displayed at Figure 5.Most clades had very low or no statistical support, except X. (Xiphocentron) tarquon and X. (Xiphocentron) polemon (95%), the clade with X. ( Rhamphocentron) species (68%), the clade X. ( Antillotrichia) regulate and X. (Antillotrichia) sturmi (87%) and the clade with X. (Antillotrichia) species from Great Antilles (X.haitiense (X.cubanum, X. nesidion)).Xiphocentron (Sphagocentron) subgenus was placed in a clade including many X.(Antillotrichia) species that do not present a mesal sclerite but instead longer spine-like setae at the same position (character 25( 1)).Xiphocentron (A.) fuscum appears as the sister clade of X. ( Sphagocentron) species (support of 44%), based on the sternum IX apodeme wide and contiguous with sternum margin (char-acter 13(1)), absence of spines at the coxopodite basal region (character 20(0)), and contributions of the preanal appendage morphometry (not shown on the cladogram).The monophyly of X. ( Sphagocentron) subgenus (support of 37%) is recovered based on the character 14(1) (sternum IX, in lateral view 2× as long as high or longer) and the contributions of the preanal appendage morphometry.The new species X. dactylum and X. eurybrachium are shown as first cladogenesis of the subgenus.Clades within X. (Sphagocentron) presented very low support (<30%) (Fig. 5A).Xiphocentron ( Antillotrichia) subgenus is not recovered as a monophyletic group, with the species included in the analysis being split into three different clades.Principal Component Analysis (PCA) of preanal appendage shape, male genitalia.Showing the four major principal components (PCs) (90.59% of the total variance).The percentage of shape variance represented by each PC is presented along the axes.Thin-plate splines show the degree of shape deformation from the mean overall shape.Diagnosis.Xiphocentron dactylum sp.nov. is particularly similar to Xiphocentron (S.) eurybrachium sp.nov.mainly by the shape of the preanal appendage (in lateral view, straight throughout length, apically wide).However, X. dactylum sp.nov.can be differentiated from its congener by the (1) shape of sternum IX in ventral view, elongate and about as wide basally as apically, subrectangular (trapezoid, conspicuously wider apically in X. eury brachium sp.nov.), by ( 2) the posterior margin of sternum IX with wide shallow V-shaped incision (absent in X. eurybrachium sp.nov.) in ventral view.Diagnosis.Xiphocentron eurybrachium sp.nov. is mostly similar to Xiphocentron (S.) dactylum sp.nov..However, X. eurybrachium sp.nov.can be differentiated mainly by the presence of the following characters: (1) the preanal appendage in lateral view conspicuously wide subapically (2× wider than basal section), without narrow digitate projection; (2) inferior appendage basal region (coxopodite) with small spine-like setae in ventral view, apex curved dorsolaterally in dorsal view; (3) paraproct ventroapical lobe wide (as wide as preanal appendage basal section) (narrow or indistinct in the other species); (4) sternum IX trapezoid, posterior margin without incision; ( 5) sternum V with a digitate lateral projection.
Etymology.From the Greek eurus, 'wide', and brakhíōn, 'arm' in reference to the preanal appendage shape in lateral view.
Distribution.Costa Rica, Panama.Remarks.Type was fixed in a permanent slide and displayed in a dorso-lateral view.Diagnosis.The new species is most similar to X. tuxtla sp.nov.by sharing the prominent acute lobes at the posterior margin of sternum IX and the slightly divergent lobes of tergum IX posterior margin.The new species can be diagnosed in lateral view by ( 1) the preanal appendage with about the same width throughout length; (2) the spinelike setae of inferior appendage numerous and very dense (sparser basely in X. tuxtla sp.nov.); and (3) the shape of paraproct apicodorsaly round (truncate in X. tuxtla sp.nov.).In dorsal view (4) the tergum IX is longer than wide (about as long as wide in X. tuxtla sp.nov.).
Etymology.Name in apposition; from the indigenous Nahuatl language: guaitil, 'tree', and nacaztli, 'ear'.It is the popular name of the tree Entelorobium cyclocarpum and also the name of the conservation area where the species was collected.

Diagnosis.
Xiphocentron julus is particularly similar to X. tampati sp.nov.by the posterior margin of sternum IX mostly straight with very small mesal lobes; and can be differentiated from this and other species mainly by the uniform width of the preanal appendage, the posterior margin of sternum IX mostly straight with very small mesal lobes, and the paraproct apex oblique.
Etymology.Name in apposition; named after the type locality, tuxtla comes from the nahuatl language: 'tochtlán', meaning place of rabbits.Subgenus Xiphocentron (Antillotrichia) Banks, 1941: 401 Type species.Antillotrichia cubana Banks, 1941, original designation. -Flint 1964:25 [to  Diagnosis.The new species can be diagnosed mainly by the sickle-shape of preanal appendage in dorsal view, in which the mesal margin has a strong concavity and the apex is wide with a mesal pointed projection.This character is unique to this species.III) < IV < V. Tibial spur formula 2:4:3; spurs unmodified.

Description
Forewing forks II and IV present; fork II sessile at discoidal cell; discoidal slightly shorter than thyridial cell.

Discussion
In this study new species of Xiphocentron ( Sphagocentron) are described after more than 40 years.Now, the subgenus includes seven species from Central and South America.However, the preanal appendage morphometric analysis and the phylogenetic analysis suggest the inclusion of some Antillean species currently in Antillotrichia subgenus.Additionally, a species of X. (Antillotrichia) was also described sharing similarities with other species from northern South America.The new records expand the distribution of the genus with the first records of the family in Bolivia, which have a similar fauna to that of northern Argentina and the Peruvian Amazon.Also, the first species of Xiphocentron are identified in Ecuador showing a wider distribution range of X. mnesteus and X. sturmi along the northern Andean foothills, and with X. sturmi species occurring even in Central America.
The phenotypic groups defined by the cluster analysis (Fig. 3) only partially recovered the phylogenetic groups (Fig. 5A).The principal components of preanal appendage shape (Fig. 2) could mostly differentiate between the subgenera Xiphocentron, Sphagocentron and Rhamphocentron but not the Antillotrichia subgenus which overlapped with most of other subgenera.In the discriminant analysis (Fig. 4) the Antillotrichia subgenus only partially overlapped with Sphagocentron subgenus.The jackknife cross-validation test was around 70% to Sphagocentron and Antillotrichia, and 33% to genera with a very small sample (n = 3).Therefore, despite the overlapping, the preanal appendage shape shows a good accuracy to classify species to subgenera when an adequate sample size is used.The shape overlapping is in accordance with the initial expectations that X. ( Antillotrichia) is possibly paraphyletic (or polyphyletic), which is also corroborated by the phylogenetic analysis (Fig. 5A).Only a single Glyphocentron was introduced in the analyses and its shape does not show any conspicuous feature, being placed closer to the X. (Sphagocentron) cluster (Fig. 3), although it was placed near X. ( Rhamphocentron) species in the phylogenetic results (Fig. 5A).
The generalized shape patterns that could be recognized in the majority of the analyzed species within each subgenus are as follows: In the Xiphocentron subgenus species exhibit a straight preanal appendage that is wide at the base and progressively tapers to a narrow apex (indicated by the positive values of PC4).In the Rhamphocentron subgenus, most species have the preanal appendage conspicuously wide at the basal third, and becomes slightly wavy and equally subnarrow at the apical 2/3 with a round apex (indicated by the positive values of PC1).In the Sphagocentron subgenus, most species share an overall wide preanal appendage, which becomes enlarged subapically and narrows towards apex (indicated by the negative values of PC2) (Fig. 2).
The Antillotrichia subgenus exhibited great morphological variation, with the cluster analysis (Fig. 3) revealing three main species groups: A group of several South American species that share preanal appendage strong-ly curved, narrow mesally with an enlarged, round apex (seen in the negative values of the PC4).The second group includes Antillean species, that present appendage slightly wavy and subequal narrow width throughout their length (seen in the positive values of the PC2).The third are species from Lesser Antilles with overall wide preanal appendage similar to Sphagocentron.However, the Antilliotrichia subgenus groups inferred from the preanal appendage phenotype differed from the phylogenetic results (Fig. 5A).The species with wavy appendages narrow at midlenght and wide, round apex were placed at distinct clades defined more by characters of the inferior appendage setose ornaments rather than preanal appendage shape, suggesting convergent evolution of the preanal appendage.
Some species with less modified preanal appendage, such as X. (S.) julus and X. (R.) numanus were not grouped with other congeners in the morphometric analysis (Fig. 3), however, they have other characters that support their recovering within their subgenera in the phylogenetic analysis (Fig. 5A).The placement of X. (A.) fuscum as shown in the cluster, phylogenetic and discriminant analyses, suggests that this species belongs to X. ( Sphagocentron).However, the phylogeny indicated that other species also could be included in a broader sense of the Sphagocentron subgenus (Fig. 5A).Nevertheless, the support for this clade was very low (18%), and many included groups had no support at all.Additionally, this clade was based only on a single discrete character and the contributions of landmarks.Therefore, we consider the evidence too weak to make conclusive taxonomic changes.
The morphofunctional aspects of the xiphocentronid genitalia are mostly unknown.Questions underlying the preanal appendage function during the copula and the selective forces related to its shape evolution have never been investigated.The copula was described to the sister group Psychomyiidae, in Tinodes the slender preanal appendages are loosely placed across segment IX of the female while the grasping is primarily provided by a spiny harpago of the male inferior appendages, holding onto membranous pits of the female segment VIII, and additional grasping is provided by the spiny paraprocts that are inserted with the phallus (Fisher 1977).In Xiphocentron the only spiny structure is often the mesal region of male inferior appendages.The long and robust preanal appendage frequently crosses the inferior appendage harpago, which should help to lock the inferior appendages in position around the female segment IX, preventing lateral movements.The paraprocts seem to have no grasping role.
Therefore, we hypothesize that the long preanal appendage in Xiphocentronidae and its different shapes should be associated with evolutionary strategies to help with additional grasping and shoring of the inferior appendage around the female, while groups in which a strong grasping is provided by structures in the phallus, paraproct or inferior appendage may have slender or reduced preanal appendages.Further research is needed to clarify the morphofunctional aspects of the preanal appendage, and test this hypothesis.

Conclusion
After more than 40 years since the establishment of the subgenus Sphagocentron new species are described, and the monophyly and relationships within the subgenus were accessed through geometric morphometry and a morphological phylogeny.This study reported the following findings: (1) The geometric morphometric analyses of the preanal appendage were consistent with the phylogenetic results regarding the non-monophyly of the Antillotrichia subgenus and the proximity of certain X. (Antillotrichia) species to the subgenus Sphagocentron.
(2) The monophyly of X. (Sphagocentron) was recovered, although with little support.X. (Antillotrichia) species having a narrow zone of dense spine-like setae on the inferior appendages and no mesal sclerite but longer setae may represent a monophyletic group together with X. (Sphagocentron) species.
(3) Phenotypic clusters fail to fully recover the phylogenetic groups, with preanal appendage shapes evolving independently in some species.
(4) The geometric morphometry was able to highlight preanal appendage subgeneric diagnostic traits, and partially distinguishing between each subgenus, but with a greater overlapping of X. (Antillotrichia) subgenus.Despite the shape overlapping, cross-validation test was able to correctly classify around 70% of the species to subgenera in better sampled groups.Therefore, the preanal appendage shape has a good informativeness for subgeneric classification.
(5) The preanal appendage shape in xiphocentronids is hypothesized to be associated to an auxiliary grasping function.

Data sharing and data accessibility
The data underlying this article, including: the discrete characters matrix (.ss), the combined discrete and morphometric characters matrix (.tnt) and the morphometric procrustes (.dat) are available at the Open Science Framework (OSF) repository and can be accessed at https://osf.

Figure 1 .
Figure 1.Location of the landmarks on the preanal appendages.Variations among samples at distinct landmarks are displayed as small dots, dark circles are landmarks and white circles are semi-landmarks.

Figure 2 .
Figure2.Principal Component Analysis (PCA) of preanal appendage shape, male genitalia.Showing the four major principal components (PCs) (90.59% of the total variance).The percentage of shape variance represented by each PC is presented along the axes.Thin-plate splines show the degree of shape deformation from the mean overall shape.

Figure 3 .
Figure 3. Neighbor-joining cluster analysis based on distance coordinates of the principal components (PCs).Representation of the preanal appendage of each species is also depicted.Bootstrap values greater than ten are shown at the nodes.

Figure 4 .
Figure 4. Discriminant analysis of preanal appendage shape over the ten first Principal Components (PCs).The percentage of shape variance is presented along the axes.

Table 2 .
Xiphocentron specimens used the morphometric analysis, country, location and depository.

Table 3 .
List of morphological characters used in the cladistic analysis.Phylogenies that previously used the character and values of Consistency index (CI) of each character are shown.