Phylogenetic relationships of the genus Mischonyx Bertkau, 1880, with taxonomic changes and three new species description (Opiliones: Gonyleptidae)

The type species of Mischonyx Bertkau 1880, Mischonyx squalidus, was described based on a juvenile. The holotype is lost. Based on a revision of publications, the genus includes 12 species, all in Brazil. The objectives of this research are: to propose a phylogenetic hypothesis for Mischonyx based on Total Evidence (TE); propose taxonomic changes based on the phylogeny; and analyze the phylogenetic hypothesis biogeographically. Using the exemplar approach to taxon selection, we studied 54 specimens, 15 outgroups and 39 ingroup taxa using seven molecular markers (28S, 12S and 16S ribosomal genes, citochrome oxidase subunit I gene, carbamoyl-phosphate synthetase gene, internal transcribed spacer subunit 2 and histone H3 gene), totaling 3,742 bp, and 128 morphological characters. We analyzed the dataset under three optimality criteria: Maximum likelihood (ML), Maximum parsimony (MP) and Bayesian. We discuss the transformation of character states throughout the phylogeny, the different phylogenetic hypotheses using different datasets and the congruence of evidence between the clades obtained by the phylogenetic analysis and the biogeographical hypothesis for the Atlantic Forest areas of endemism. We estimate that Mischonyx clade diverged 50.53 Mya, and inside the genus there are two major clades. One of them cointains species from Paraná, Santa Catarina, South of São Paulo and Serra do Mar Areas of Endemism and the other has species from Espinhaço, Bocaina, South coast of Rio de Janeiro and Serra dos Órgãos Areas of Endemism. The first split inside these two clades occurred at 48.94 and 44.80 Mya, respectively. We describe three new species from Brazil: Mischonyx minimus sp. nov. (type locality: Petrópolis, Rio de Janeiro), Mischonyx intervalensis sp. nov. (type locality: Ribeirão Grande, São Paulo) and Mischonyx tinguaensis sp. nov (type locality: Nova Iguaçu, Rio de Janeiro). The genus Urodiabunus Mello-Leitão, 1935 is considered a junior synonym of Mischonyx. Weyhia spinifrons Mello-Leitão, 1923; Weyhia clavifemur Mello-Leitão, 1927 and Geraeocormobius reitzi Vasconcelos, 2005 were transferred to Mischonyx. Mischonyx cuspidatus (Roewer, 1913) is a junior synonym of M. squalidus Bertkau, 1880. In the results of the phylogenetic analyses, Gonyleptes antiquus Mello-Leitão, 1934 (former Mischonyx antiquus) does not belong in Mischonyx and its original combination is re-established. As it is now defined, Mischonyx comprises 17 species, with seven new combinations.


INTRODUCTION
Laniatores is the most diverse suborder within Opiliones. There are more than 4,200 species in the group (Kury, 2020), of which at least 2,400 are from the Neotropical region (Kury, 2003). The evolution and phylogenetic relationships of most families and genera within the suborder have been poorly studied.
Modern taxonomists base their classifications on cladistics hypotheses (e.g., Bragagnolo & Pinto-da-Rocha, 2009;DaSilva & Gnaspini, 2010;Pinto-da-Rocha, 2002; Pinto-da- Rocha & Bragagnolo, 2010) using a number of markers, including molecular data (e.g., Bragagnolo et al., 2015;Pinto-da-Rocha et al., 2014). This also applies to the taxonomy of Laniatores to a certain extent, but despite recent progress, the classification system devised by Carl F. Roewer (1881Roewer ( -1963 still prevails in this group. Roewer based his nomenclature and groups on a few arbitrary characters. As a result, he created a lot of monotypic genera and placed closely-related species in distinct clades (Pinto-da- .
Gonyleptidae Sundevall, 1833 is one of the families within Laniatores that includes many monotypic genera and artificial groups. According to Kury (1990), there are many species in the family that have been cited only once, suggesting that there may be many synonyms to be established. Recent research on Gonyleptidae subfamilies using phylogenetic systematics has found evidence supporting several groups DaSilva & Gnaspini, 2010;DaSilva & Pinto-da-Rocha, 2010; Pinto-da- Rocha & Bragagnolo, 2010). In addition, with the use of molecular data in phylogenetic inference, Pinto-da- , Benedetti (2017) and Benavides, Pinto-da- Rocha & Giribet (2021), proposed new relationships among most Gonyleptidae subfamilies. One subfamily, however, Gonyleptinae Sundevall, 1833 (39 genera, 140 species in total), remains to be analyzed under a phylogenetic framework (Kury, 2003). The diagnosis of the subfamily, based on the number of areas on the dorsal scutum and the absence of certain features that characterize other subfamilies (Pinto-da- , suggests that Gonyleptinae is a polyphyletic clade and it is possible that several genera will need to be transferred out of it.
In the second half of the 20th century, B. Soares and H. Soares synonymized Ilhaia with Eduardoius (Soares, 1943), Geraecormobiella with Geraeocormobius Holmberg, 1887 (Soares, 1945c) and Ilhaia with Xundarava . Along with that, the authors synonymized some species of these genera and described more species. Kury (2003) synonymized Ilhaia and Giltaya with the almost forgotten genus Mischonyx. Besides that, he transferred G. antiquus (then in Paragonyleptes) to Mischonyx. Since the holotype of Mischonyx squalidus is lost, Kury based his conclusions on Roewer's drawings and description. In his catalog, Kury considers Mischonyx as including 11 species.
Finally, in Vasconcelos (2004Vasconcelos ( , 2005a) the two last Mischonyx species were described: Mischonyx kaisara, from the coast of the state of São Paulo, and Mischonyx poeta, from the northern portion of the state of Rio de Janeiro. He also described Gearaeocormobius reitzi Vasconcelos, 2005b. Besides these publications, there is one unpublished M.Sc. dissertation on the taxonomy of Mischonyx taxonomy (Vasconcelos, 2003).
The main goal of this work is to propose a phylogenetic hypothesis for Mischonyx, based on total evidence combining sequences from seven genes and morphological characters that include the external morphology and genitalia. In addition, we propose taxonomical changes, describe new species and make remarks on biogeography based on the phylogenetic hypothesis.

Species distribution and areas of endemism
To build an updated map of the geographical distribution of Mischonyx species, we used DIVA-GIS to plot the geographical coordinates of the specimens available in the collection of Museu de Zoologia da Universidade de São Paulo (MZSP) and the Arachnology Lab (IB-USP) tissue collection. We also included the type localities and records extracted from Kury (2003).
The nomenclature used for the areas of endemism of the Atlantic Rainforest and their delimitation follows DaSilva, Pinto-da- Rocha & Morrone (2017).

Type specimens and ingroup selection
We analyzed (see Table 1) at least one type specimen from each valid Mischonyx species listed in Kury (2003), except the holotype of Mischonyx squalidus, which has been lost. Type specimens were compared with the harvestmen tissue collection of the Arachnology Lab (Instituto de Biociências -Universidade de São Paulo). Additionally, we collected fresh specimens for DNA extraction. Individuals that resembled Mischonyx species but did not match described species were also included in the analysis. The ingroup used in the phylogenetic analysis is listed in Table 2.

Molecular data acquisition
Specimens for the molecular analysis were kept at 92-98% ethanol and at −20 C. Our lab has a database with gene sequences originated from different projects. We used sequences from that source and sequenced the DNA from additional species using muscular tissue from coxa IV (Pinto-da- . Alternatively, when the individual to be sequenced was small, we used tissues from the chelicerae and pedipalps. We used the kit AgencourtÒ DNAdvance System (EUA; Beckman Coulter, Brea, California, USA) for extractions and modified the protocols according to Pinto-da- . From the extracted DNA, we amplified seven molecular loci: the ribosomal nuclear gene 28S rRNA; the ribosomal mitochondrial genes 12S rRNA and 16S rRNA; the nuclear sequences of the internal transcribed spacer subunit 2 (ITS2), carbamoylphosphate synthetase 2 gene (CAD) and the histone H3 gene (H3); and the mitochondrial cytochrome oxidase subunit I gene (COI). For polymerase chain reactions (PCRs), we used Thermo-fisher Taq kit, following the concentration present in Pinto-da- .
-We conducted PCR reactions in an Eppendorf MastercyclerÒ gradient thermal cycler and the cycles and temperature used in this work are the same as in Pinto-da- . Afterwards, we inspected the PCR products using agarose gel electrophoresis (2% agarose), purified the products using Agencourt Ampure XP (Beckman Coulter, Brea, CA, USA) and quantified the products using a Thermo Scientific NanoDrop spectrophotometer. In order to prepare the products for sequencing, we used the BigDyeÒ Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Waltham, MA, USA). The precipitation was with sodium acetate and the sequencing process was in an ABI PRISMÒ 3100 Genetic Analyser/HITACHI (Applied Biosystems, Waltham, MA, USA).
first codon position. All sequences are at GenBank and their respective access codes are in Tables 1 and 2.

Morphological data acquisition, terminology and new species drawings
We coded the external morphological characters after analyzing the type material and other individuals of the species when available under a Zeiss Stemi DV4 stereomicroscope. Analyzis of the male genitalia characters was conducted under a Scanning Electron Microscopy (SEM). We followed the protocol of Pinto-da- Rocha (1997) to dissect and prepare the genitalia for Scanning Electron Microscope (Zeiss DSM940, from Instituto de Biociências, Universidade de São Paulo) and built the character matrix using Mesquite 3.51 (Maddison & Maddison, 2017). We coded most characters as binary to avoid redundancy and tried to ensure that all characters were independent from each other (Strong & Lipscomb, 1999). Nonetheless, to avoid building non-comparable characters, in some cases, we used multistate characters and treated them as unordered. The character descriptions follow Sereno (2007). The complete character matrix is available online, at MorphoBank (http://morphobank.org/permalink/?P3599). The general terminology follows DaSilva & Gnaspini (2010). Granules refer to minute elevations, concentrated on a particular region or article. Tubercles are elevations that are clearly distinguishable from granules by their height and width and can have blunt or acuminated apex. Spines are acuminated elevations present on the ocularium. Apophyses, which have different shapes, are the armatures present on coxa IV, free tergites, anterior and posterior margins. The terminology for the shape of the dorsal scutum follows . The terminology for the penial macrosetae follows Kury & Villareal (2015).
We used a stereomicroscope coupled with a camara lucida to make our drawings. After that, we digitalized them and made corrections on the background using Adobe Photoshop Lightroom 6.0Ò.

Nomenclatural acts and collecting license
The electronic version of this article in Portable Document Format (PDF) will represent a publication according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone. 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 can be viewed using any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSIDs for this publication are: urn:lsid: zoobank.org:act:A6F34641-1AF1-4BE2-A16A-4A4497ECA1FC; urn:lsid:zoobank.org: act:3DDE0A87-E9F6-4504-9C54-6DC37D202A0E; urn:lsid:zoobank.org:act:5FA4CC13-EC27-4E3A-AB19-81A97FE74177. The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central and CLOCKSS.

Molecular dating
First, we used only the COI to estimate how long ago Mishconyx diverged from its ancestor. We did this because there are more Gonyleptidae sequences of this gene than any other on GenBank. Only one sequence from each species was included, totaling 122 terminal sequences. To set the priors for the BEAST 2.5 analysis (Bouckaert et al., 2019), we employed the program BEAUti. We used the Beast Model Test to set the site model, a lognormal relaxed clock with substitution rate of 0.005 (according to Bragagnolo et al., 2015 andPeres et al., 2019) with Yule tree and constrained the root using a normal distribution. In this initial analysis three clades were dated: Gonyleptidae, with T MRCA 140 ± 40 Mya, based on Sharma & Giribet (2011);Sodreaninae Kury, 2003clade (sensu Peres et al., 2019, with T MRCA 31.5 ± 10 Mya, based on Peres et al. (2019);Promitobates Rower, 1913, with T MRCA 58.5 ± 3.9 Mya, based on Bragagnolo et al. (2015). We then ran two independent analyses, with 10 million generations each, sampling trees every 10,000 generations. Both analyses were verified in TRACER 1.7 (Rambaut et al., 2018) and checked for EES > 200. The results were combined in LOGCOMBINER 2.5.
Next, we applied the T MRCA estimated for Mischonyx to calibrate the multilocus species tree using Ã BEAST, with the seven genes cited above and the terminals from Table 2, also using BEAST 2.5. We pruned the dataset to one sequence per haplotype per species, used all the priors from the first step and performed two independent analyses with 100 million generations, sampling trees each 5,000 generations. The output from the analyses was checked using Tracer 1.7 and combined trees using LOGCOMBINER 2.5. The maximum clade credibility was annotated and the first 10% was discarded, using TREEANNOTATOR 2.5. The final tree was analyzed using FigTree 1.4.4 (Rambaut, 2010).

Phylogenetic inferences
Three separate analyses were carried as follows: (1) morphological data alone, (2) molecular data alone; and (3) combined molecular and morphological matrixes (Total Evidence Analysis). Each matrix was analyzed using Maximum parsimony (MP) and maximum likelihood (ML). In all analyses, we used Promitobates ornatus Mello-Leitão, 1922 to root our trees because its is consistent with Pinto-da-  phylogeny, in which this species is the furthest from Mischonyx clade, when compared to the other species used as outgroups in our research.

Bayesian inference
In the morphological analysis (B1), we activated the morph-models package on BEAUti 2.5 and imported the matrix, with the option "add MK morphological data" while importing. The Lewis MK was chosen as the substitution model, and the relaxed log normal clock and fossilized birth and death model were chosen as tree priors.
For analysis using strictly molecular data (B2), the trees for all genes were linked. The Beast Model Test was selected for calculations of the best model for each gene, estimating the mutation rate. The relaxed log normal clock, with the estimates of clock rate for each gene, followed Bragagnolo et al. (2015) and Peres et al. (2019). The selected tree model was the Birth and Death model.
The same parameters used for the molecular data analysis were used in the total evidence (TE) (B3). We chose Fossilized Birth and Death Model as the tree prior, with 0.05 as the starting value for the tree diversification rate, with estimation of Rho parameter. To estimate the morphological and molecular clock rates we chose the LogNormal distribution.
All Bayesian analyses were carried out on BEAST 2.5, performing two independent analyses, with 100 million generations each, sampling trees every 10,000 generations. We checked the output from the analyses, using Tracer 1.7, checked for EES > 200 and combined trees using LOGCOMBINER 2.5. The maximum clade credibility was annotated and the first 10% was discarded, using TREEANNOTATOR 2.5. The final tree was analyzed using FigTree 1.4.4 (Rambaut, 2010).
Maximum likelihood. For morphological analysis (ML1), we inserted the dataset as input in the IQ-TREE version 1.6.10 (Nguyen et al., 2015), using the best model found by the program, which uses BIC (Bayesian information criterion) (Schwarz, 1978) to analyze which model is the best for that specific dataset. The analysis displayed by the program is the same described for the molecular data below. To analyze character changes, we inserted the phylogeny output from IQ-TREE on YBIRÁ (Machado, 2015).
The DNA sequences were aligned in MAFFT and analyzed with Aliview. The FASTA file contained all the sequences concatenated using SequenceMatrix 1.8 (Vaidya, Lohman & Meier, 2011). The analysis was carried out in IQ-TREE version 1.6.10 (Nguyen et al., 2015). All the partitions coming from the seven different genes present in the concatenated FASTA file (and the morphological dataset for TE) were first analyzed on IQ-TREE through the partition model (Chernomor, von Haeseler & Minh, 2016), using the "-spp" command. The program selected the best substitution model for each gene partition under the BIC (Schwarz, 1978), using the program ModelFinder (Kalyaanamoorthy et al., 2017), through the command "-m TESTNEWMERGE". Maximum Likelihood analysis was based on 10,000 search iterations, using the command "-s -n 10000". Confidence was measured using bootstrap analysis based on 1,000 iterations of ultrafast bootstrap using the command "-bb 1000" (Minh, Nguyen & von Haesler, 2013). The output was analyzed using FigTree 1.4.4 (Rambaut, 2010). We used the parsimony method to analyze character changes because, as pointed by Cheng & Kuntner (2014), the aim is to "understand the evolutionary changes of characters rather than the probability of particular ancestral states on the phylogeny".
Maximum parsimony. The morphological analysis (MP1) was carried out using TNT (Goloboff, Farris & Nixon, 2008). The search was heuristic with TBR branch-swapping (10,000 replicates) while retaining 100 trees per replicate. The command "collapse branches after search" was used to eliminate non-supported nodes, and searches using Ratchet (Nixon, 1999) and Tree Fusing (Goloboff, 1999). The characters were treated as unordered and unweighted. To analyze character changes throughout the phylogeny, we used Winclada 1.61.
The molecular (MP2) and TE (MP3) analyses were implemented using the program POY 5.1.1 (Varón, Vinh & Wheeler, 2010), which searches using direct optimization (hereafter DO) of unaligned sequences (Wheeler, 1996), a strategy referred as Dynamic Homology (Wheeler, 2001a(Wheeler, , 2001b. This strategy differs from the traditional static homology search in that the former integrates both alignment and tree searches, while the last treats them as two separated searches. DO is able to test dynamically, in a static matrix, the hypotheses of homology among unaligned nucleotides, optimizing these sequences directly on the available trees and, concomitantly, converting the transformation series of pre-aligned sequences (Kluge & Grant, 2006;Grant & Kluge, 2009;Sánchez-Pacheco et al., 2017).
An exploratory DO analysis was carried out five times, specifying search time (from two to ten hours, totaling 30 hours of search), to check which one yielded the lowest tree scores as outputs and, consequently, the optimal search time for DO ("max_time" parameter). The best tree scores for our dataset were obtained with a maximum search time of 2 h. After that the dataset was analyzed treating H3, COI and CAD sequences as pre-aligned, because they are coding genes, and 28S, 12S, 16S and ITS to be aligned using dynamic homology methods ("transform" command in POY). The program performed five rounds of searches using the "max_time" (with "search" command). In POY each "search" round implements Tree Bisection and Reconnection (TBR), Wagner tree building, Subtree Pruning and Regrafting (SPR), Branch Swapping (RAS+swapping, as in Goloboff, 1999), Tree fusing (Goloboff, 1999) and Parsimony Ratchet (Nixon, 1999). We used the final trees from this previous analysis in an exact iterative pass (IP) analysis (Wheeler, 2003). Costs for all the previous optimal trees were calculated and POY generated the implied alignment of this final analysis (Wheeler, 2003). TNT 1.5 (Goloboff & Catalano, 2016) was used to calculate Bootstrap values and Bremer support, with "hold" command of 10,000,000 trees, "mult" command of 1,000 replicates, holding 10 trees per replicate. Finally, we analyzed the character changes over the optimal tree using parsimony on YBIRÁ (Machado, 2015).

Morphological data
The morphological matrix totals 128 characters, some of which were taken from the literature and are distributed as follows: 45 characters from the dorsal scutum, 44 characters from the appendages, 6 characters from free tergites, 27 characters from the male genitalia and two characters from the general habitus.  79. Leg IV, femur, dorso-basal apophysis, apex width: 0, Base more than 4 times wider than apex (

Morphological analyses
In all analyses using morphological data alone, under the maximum likelihood (hereon ML1, Fig. S3), under Bayesian (hereon B1, Fig. S4) and under maximum parsimony (heron MP1, Fig. S5) criteria, the first lineage branching off inside the Mischonyx clade is composed of M. arlei comb. nov., M. minimus sp. nov. and M. intermedius, followed by the divergence of G. antiquus (former Mischonyx antiquus, before this work). The only difference is that, in B1, Multumbo species are in a clade with M. intermedius, M. minimus and M. arlei. Moreover, all analyses recover the clade formed by M. anomalus, M. clavifemur comb. nov. and M. reitzi comb. nov., consistent with the results of the molecular and TE analyses (Figs. 16-18 and S6-S12). B1 is the only analysis that places both Multumbo species within the Mischonyx clade. The results of ML1 and MP1 agree with our TE results (see below). All analyses were weakly supported by Bootstrap.
The bootstrap values obtained for the Mischonyx clade is 25 in ML1 and 7 in MP1. All internal branches inside the genus have values below 50 in both analyses (Figs. S3 and S5). In B1, the posterior probability of the Mischonyx clade was 0.872 and most nodes inside the genus have posterior probabilities lower than 0.6.

Molecular analyses
In all analyses using molecular data alone, Mischonyx is monophyletic if G. antiquus (former Mischonyx antiquus) is removed from the genus: under maximum likelihood  with species from SMSP, SSP, PR and SC AoE sister to the lineage with species from Boc, Esp, LSRJ and Org AoE.
The support values were high in all three analyses: Bootstrap (in ML2 and MP2), Bremer (in MP2) and posterior probability (in B2). The bootstrap value for the Mischonyx clade in ML2 was 92 and in MP2 it was 100. In MP2, the node with the lowest bootstrap value is the one holding Deltaspidium, Multumbo and some Mischonyx species (cited above). In ML2, the lowest value inside the Mischonyx clade is 67 (Figs. S6 and S8). All posterior probabilities inside the genus are higher than 0.9, except for two nodes, which have values above 0.6.

Molecular dating
The Bayesian analysis (henceforward abbreviated as BM, fig. 16) generally corroborates the topologies obtained from the other molecular analyses, except for the position of M. poeta.

Total evidence analyses
All TE analyses, under maximum likelihood (hereon ML3, Fig. 17 18A). Posterior probabilities inside the genus are higher than 0.6 and Mischonyx clade posterior probability is 0.971 (Fig. 18B). Henceforward, we are going to consider ML3 as our working phylogeny to present the further results regarding character state changes and to discuss relationships and character evolution.
Females have dorsal scutum outline a, with coda long and clearly separated from mid-bulge. Anterior margin with lateral armature, normally two or three tubercles on each side. Frontal hump high and narrow, with a pair of median tubercles (except in M. processigerus, which has two pairs). Lateral margin of prosoma with several granules, posterior to ozopore. Ocularium narrow and not very high, armed with median spines or tubercles. Some species have small tubercles anterior or posterior to the eye (or both). Posterior margin of prosoma with a pair of tubercles. Dorsal scutum with three areas. Mesotergal area I is divided by a longitudinal groove. Areas I and II armed with median tubercles (which are large and whitish in M. arlei comb. nov. and M. minimus sp. nov.). Area III with a pair of median elliptic tubercles (except in M. arlei comb. nov. and M. minimus sp. nov.), which can vary in size and lateral compression. Some species have other elliptic tubercles besides the median ones (e.g., M. spinifrons comb. nov.). Lateral margin of dorsal scutum (mid-bulge) with rounded tubercles, which are fused in some species (e.g., M. spinifrons comb. nov.). Distitarsi of all legs with three segments. Basitarsus of leg I with three or four segments. Basitarsus II with foureight segments. Basitarsi III and IV with four or five segments. Ventral surface of coxae I generally with more developed tubercles than the ones on the other coxa. Coxa IV with apical prolateral apophysis, generally robust, in some speciemens with ventral process and a basal tubercle. Trochanter IV short and robust, with a blunt prolateral apophysis and at least one retrolateral armature. Femur IV with DBA, which can be small (as in M. arlei comb. nov. and M. minimus sp. nov.), or large (most species). DBA can be branched or not and varies in shape and size in every species. Retrolateral row of tubercles generally with some large apophysis. Penis with ventral plate trapezoidal with an apical parabolic groove; three pairs of MS A and one pair of MS B on lateral projections; three pairs of helicoidal MS C, two pairs of reduced MS E, one pair of MS D, venter of ventral plate with microsetae type T1 covering its whole extension or the basal half. Glans with ventral process, flabellum can be serrated or smooth. Stylus with microsetae, inclined in relation to axis of penis and with ventral groove.

Species new combinations
Besides the combinations and synonyms present in Kury (2003) and Pinto-da-Rocha et al. Xundarava anomala Mello-Leitão, 1936: 13, fig. 10; Soares, 1945b: 192;1945c: 366;Soares, 1945d: 210;  Diagnosis. Mischonyx arlei comb. nov. resembles M. minimus sp. nov. by the following combinations of characters: mesotergal area I with a pair of well-developed median tubercles, which are paler (whitish) than the rest of the dark brown body; median armatures on mesotergal area III are spines; lateral margin of dorsal scutum with several small tubercles; Free tergite II with a well-developed median apophysis; prolateral apophysis on coxa IV small and pointing posteriorly; retrolateral side of trochanter IV with two armatures; femur IV with several small apophyses on dorsal and retrolateral row of tubercles; femur IV with a well-developed terminal tubercle on pro and retrolateral rows of tubercles; ventral plate with three subdistal MS C on each side; MS B smaller than MS A; flabellum with serrated ends. It differs from M. minimus sp. nov. in the following: size (7-8 mm) (3-3.5 mm in M. minimus sp. nov.); mesotergal area II with median tubercles small and darker than the rest of the body (median tubercles whitish and as large as the median tubercles on mesotergal area I in M. minimus sp. nov); basitarsus II with seven segments (four in M. minimus sp. nov); leg IV curved in dorsal view (straight in M. minimus sp. nov); MS D reduced (well-developed in M. minimus sp. nov).  As M. kaisara was recently described and there is no new combination for the species, Vasconcelos (2004) diagnosis for the species remains unaltered and with no necessity to add information.
Mischonyx parvus (Roewer, 1917) comb. nov. (Figs. 5B, 5D, 11D-11F  Taxonomic remarks: Kury (2003) synonymized this species with M. squalidus. However, the distribution of M. parvus does not match with the original location of the described individual in Bertkau (1880). In the latter work, the location of the specimen is "Copacabana, Rio de Janeiro". By the distribution map in the Figs. 15, S1 and S2, the registers from this species are from Mangaratiba and Angra dos Reis, which are to the south of Rio de Janeiro state. For this reason, we removed this species from the synonymy created by Kury (2003). As M. poeta was recently described and there is no new combination for the species, Vasconcelos (2005a) diagnosis for the species remains unaltered and with no necessity to add information.
Ilhaia lutescens: Soares, 1943: 56. Taxonomic remarks: Vasconcelos (2003, proposed this new combination in his dissertation. We analyzed Bertkau's original drawing (Bertkau, 1880, fig. 38) and the original description of M. squalidus, but were not able to lay hands on the holotype because it is lost. It was deposited at the Institut Royal des Sciences Naturelles de Belgique. Part of the description translated from German is presented below: "… The first abdominal dorsal segment is almost fused with the thorax, and in general the articulation skin between each segment is not very flexible. The first three [abdominal] segments have in their superior part a line of "dots", of which the median ones stand out in height, like little spines." (Bertkau, 1880, pp. 107) The only species that has one median armature on each free tergite in females and juveniles in the region Bertkau collected the specimen (Copacabana, Rio de Janeiro) is the traditionally called M. cuspidatus. Therefore, we propose that Ilhaia cuspidata is a junior synonym of M. squalidus. We know the holotype is a juvenile, based on the image in Bertkau (1880), and Roewer's (1923) and Kury's (2003) statements.
Diagnosis. M. squalidus resembles M. spinifrons comb. nov. in the following: lateral margin of dorsal scutum with whitish tubercles (in ethanol); posterior tubercles on lateral margin of dorsal scutum fused; retrolateral apophysis of coxa IV visible in dorsal view; DBA with apex directed anteriorly; dorsal row on femur IV with three tubercles after DBA, on distal half; retrolateral row on femur IV with median apophysis more developed than the others in this row; ventral side of ventral plate without microsetae on distal half; lateral projections of ventral plate projected dorsally and behind ventral projection of glans; MS A forming a triangle; MS B reduced; apical groove of ventral plate reaching the line of the most basal MS C. It differs from M. spinifrons comb. nov. in the following: median tubercles on mesotergal area III strongly compressed and large (small and elliptic but not strongly compressed laterally in M. spinifrons comb. nov.); prolateral apophysis on coxa IV approximately same length as trochanter IV (smaller in M. spinifrons comb. nov.); Free Tergites I-III with median apophysis (without median apophysis in M. spinifrons comb. nov.); prolateral row with median tubercles larger than the others in this row (all tubercles subequal in size in M. spinifrons comb. nov.); retrolateral row on femur IV with several (7)(8) large tubercles basal to median apophysis (three tubercles basal, followed by a gap and one tubercle after this gap in M. spinifrons comb. nov.). Etymology. From the Latin adjective minimus, a, um meaning small, little. This is due to its reduced size when compared to other Mischonyx species, specially Mischonyx arlei comb. nov., sister species of M. minimus sp. nov.. Diagnosis. Mischonyx minimus sp. nov. resembles M. arlei comb. nov. in the following: mesotergal area I with pair of well-developed median tubercles, paler (whitish) than rest of body (dark brown); median armatures on mesotergal area III are spines; lateral margin of dorsal scutum with several small tubercles; free tergite II with a well-developed median apophysis; prolateral apophysis on coxa IV small and pointing posteriorly; retrolateral side of trochanter IV with two tubercles; femur IV with several small apophyses on dorsal and retrolateral row of tubercles; femur IV with a well-developed apical Lateral margin of prosoma with numerous small tubercles. Posterior portion of prosoma with a pair of tubercles. Besides these tubercles, prosoma has a low density of granules. Dorsal scutum divided into three mesotergal areas, with low density of granules (DaSilva & Pinto-da- . Areas: Area I divided by a median longitudinal groove, with a pair of whitish large median tubercles and no granules; area II with a pair of large whitish median tubercles, same size as the tubercles on Area I without granules; Area III with a pair of dark median sharp spines, smaller than the other armatures on other mesotergal areas, a pair of tubercles posterior to median spines. Lateral margins of dorsal scutum with a row of small tubercles, approximately the same size, extending from the middle of area I until the posterior margin of Area III; no fusion of tubercles. Posterior margin of dorsal scutum with a line of small tubercles. Free tergite I with a line of small tubercles approximately the same size. Free tergite II with a large sharp median apophysis and two large tubercles, lateral to the median apophysis; free tergite III with a line of small tubercles. Dorsal anal operculum with small sparse tubercles. Venter. Coxa I with several sparse tubercles, larger than the ones on other coxa. Coxa II with sparse numerous granules. Coxa III with an anterior and a posterior basal-apical row of tubercles; coxa IV with sparse numerous granules. Ventral anal operculum with granules. Chelicerae. Segment II with several setae, mainly apical. Fix and movable fingers with seven teeth each. Pedipalps. Venter of trochanter with few sparse tubercles; tibia setation: prolateral IIi, retrolateral IiIi. Tarsal setation: prolateral IiI, retrolateral III, ventral side with two baso-apical lines of setae. Legs. Leg I: trochanter with several ventral tubercles, femur, patella and tibia with granules. Leg II: Trochanter II with several ventral tubercles; femur, patella and tibia with granules. Leg III: trochanter with several ventral tubercles; femur, patella and tibia with granules; Leg IV: Coxa IV: robust apical oblique prolateral apophysis, smaller than the trochanter size; large retrolateral apophysis, visible in dorsal view. Trochanter IV: prolateral small blunt apophysis; retrolateral side with a line of three large tubercles, two slightly more ventral. Femur IV: long, thin and straight; all tubercles on prolateral row approximately the same size; DBA small, unbranched, conic, sharp, pointing upwards; dorsal row with several small tubercles after DBA; retrolateral row of with several small tubercles and two more developed tubercles on the apical half; all tubercles on the ventral row small. Tarsal formula: 6(3)-6(3)-4-5. Male genitalia (Figs. 14A-14C). Ventral plate: Ventral surface covered with microsetae; pronounced apical groove (reaching the line of the first basal MS C); lateral lobes basal when compared to other species (e.g., Mischonyx intervalensis sp. nov.); three sub-apical helicoidal MS C on each side; two MS E, ventral and in the same baso-apical orientation of MS C; long MS D when compared to other species (e.g., Mischonyx intervalensis sp. nov.), basal relative to MS C and in the same dorso-ventral orientation of MS C; three spatular MS A, forming a diagonal baso-apical line; one reduced MS B, much smaller than MS A. Glans: Small dorsal process; flabelum triangular, with serrated apex; stylus with subapical microsetae, with the apex inclined relative to the penis axis and keeled. Color. Dark brown; pedipalps and trochanters I-III yellow.

Mischonyx intervalensis
Diagnosis. It resembles Mischonyx anomalus in the following: Anterior margin of dorsal scutum with two tubercles on each side; Areas I and II with small median tubercles; area III with well-developed and elliptic median tubercles; other tubercles on area III rounded; all free tergites with small tubercles; retrolateral row of leg IV with large median apophysis; retrolateral row of leg IV with several well-developed tubercles. It differs from M. anomalus in the following: prolateral apophysis of coxa IV with ventral process and basal tubercle (not present in M. anomalus); retrolateral side of trochanter IV with three tubercles (one in M. anomalus); DBA of leg IV branched and dorsal branch is the largest (not branched in M. anomalus); one apophysis on the dorsal row of tubercles of leg IV after DBA (three in M. anomalus); tubercles on prolateral row of tubercles on leg IV small and subequal in size (median tubercles larger in M. anomalus); ventral plate with the same approximate height and width (square-shaped) (higher than wider in M. anomalus); lateral processes of the ventral plate medial (basal in M. anomalus).
Description density of granules. Free tergites I-II with a line of small tubercles of the same approximate size. Free tergite III with a row of tubercles larger than the ones on the other free tergites and central tubercle slightly larger than the others. Dorsal anal operculum with small sparse tubercles. Venter. Coxa I with several sparse tubercles, larger than the ones on other coxae. Coxae II-IV with sparse numerous granules. Ventral anal operculum with granules. Chelicerae. segment II with several setae, mainly apical. Fixed finger with eight and movable finger with 12 teeth. Pedipalps. Ventral side of trochanter with few sparse tubercles; tibia setation: prolateral IiIi, retrolateral IiI. Tarsal setation: prolateral IiI, retrolateral II, ventral side with two baso-apical lines of setae. Legs. Leg I: trochanter, femur, patellae and tibia with granules. Leg II: Trochanter II with two retrolateral tubercles; femur, patella and tibia with granules. Leg III: trochanter, femur, patella and tibia with granules. Leg IV: coxa IV: robust apical prolateral apophysis, slightly inclined relative to the axis of the base of coxa IV, with ventral process and basal tubercle, with the approximate trochanter size; retrolateral apophysis small, not visible in dorsal view. Trochanter IV: prolateral small blunt apophysis; retrolateral side with a line of three large tubercles, two slightly more ventral. Femur IV: short and robust; all tubercles on prolateral row with approximately the same size; dorsal row of tubercles with a large tubercle before the DBA, DBA branched with the largest branch pointing upwards, one large tubercle after DBA; retrolateral row of with a large median apophysis, eight large tubercles before, three large (yet smaller than the ones anterior to the median apophysis) and three small tubercles posterior to the median apophysis, intercalated; all tubercles on the ventral row small. Tarsal   Etymology. Species name derives from "Tinguá", due to its first collecting locality, Reserva Biológica Tinguá, type and only locality registered for this species + the suffix -ēnsis, -ēnse, in order to form an adjective. Anterior surface of the ocularium with one pair of tubercles, one pair of median tubercles (as tall as the ocularium height). Lateral margin of prosoma with numerous small tubercles. Posterior part of prosoma with a pair of tubercles. Besides these tubercles, prosoma has a low density of granules (DaSilva & Pinto-da-Rocha, 2010   Anoploleptes Piza, 1940: 56;Soares, 1943: 53;Kury, 2003: 133 [= Mischonyx Bertkau, 1818 (type species Anoploleptes dubium Piza, 1940, by original designation).

Diagnosis. It resembles
REMARKS: We reestablished Anoploleptes as a subjective junior synonym of Gonyleptes as first established by Soares (1943).
12. Retrolateral branch of DBA evidently larger than other branch; two apophysis on the leg IV dorsal row of tubercles, after DBA; prolateral apophysis of coxa IV with a prominent ventral process (Fig. 4A)  Both branches of DBA of the same size; two well-developed apophyses on leg IV retrolateral row of tubercles (Fig. 7A)  One extra row of tubercles between dorsal and prolateral rows; median tubercles prolateral row of tubercles of Leg IV more developed; one apophysis on the leg IV terminal third of the retrolateral row of tubercles (Fig. 7B)

Biogeographical remarks
In general, harvestmen in the Atlantic Forest have a high degree of endemism (Pinto-da- Rocha, DaSilva & Bragagnolo, 2005). Throughout the order, species distributions are restricted to specific areas of few thousands of square kilometers, with a few exceptions (e.g., Pinto-da- Rocha, DaSilva & Bragagnolo, 2005). The distribution of most species of Mischonyx are consistent with this pattern. One exception is M. squalidus. There are records of this species from the southeastern state of Espirito Santo to the southern state of Rio Grande do Sul. It occurs not only in Atlantic Rainforest but also in cerrado areas (Figs. 15,S1 and S2), where the climate is drier (Resende, Pinto-da-Rocha & Bragagnolo, 2012). Mestre & Pinto-da-Rocha (2004) demonstrated that this species is synanthropic. It is able to thrive in environments like residential areas and agricultural areas. This characteristic may explain its wide distribution, since it helps these hasvestmen to disperse and colonize new areas more efficiently than most other species.
The distribution area of most Mischonyx species is restricted to only one or few records that are in close proximity to each other. This is consistent with the hypothesis that harvestmen have a high degree of endemism (DaSilva, Pinto-da- Rocha & Morrone, 2017). Serra do Órgãos, Mantiqueira, south coast of Rio de Janeiro and Serra do Mar areas of endemism hold 11 from the 16 species of the genus. According to Pinto-da- Rocha, DaSilva & Bragagnolo (2005) and DaSilva, Pinto-da- Rocha & Morrone (2017), the southern coast of Rio de Janeiro and Serra dos Órgãos areas are the most species rich. This is supported by our findings and is an important piece of information for conservation, since the few remaining harvestmen habitats are under the impact of anthropic changes (Morellato & Haddad, 2000). To maintain the diversity of the entire group, these endemic areas need to be better protected (DaSilva, Pinto-da- Rocha & Morrone, 2017;Nogueira et al., 2019aNogueira et al., , 2019b.

Divergence time of Mischonyx clade
We are going to work with the Bayesian hypothesis to discuss divergence time and biogeography. BM is the preferred optimality criteria for estimating divergence time and there were no significant differences in the relationships among the internal branches of the topologies recovered using BM and TE (MP3 and ML3).
Two previous publications on two gonyleptid genera of the Atlantic Forest dated the divergence time of clades: Bragagnolo et al. (2015), using Promitobates, and Peres et al. (2019), using Sodreana. The divergence time of Mischonyx (~50 Mya) is consistent with the estimates obtained for Promitobates. Sodreana diverged more recently (~35.5 Mya) and occurs in a more restricted area than the other two genera (from the southern state of Paraná to the southern limit of Serra do Mar in the state of São Paulo). Promitobates occurs from the state of Santa Catarina to the northern edge of the state of São Paulo and Mischonyx occurs from Santa Catarina to the northern portion of the state of Rio de Janeiro (excluding M. squalidus, which is more widely distributed). The wider distribution of the last two genera may be a function of their older diversification times.
As stated by DaSilva, Pinto-da-Rocha & Morrone (2017), "The main geographical barriers associated with the general historical patterns are the Valleys of the Doce, Paraıba do Sul, and Ribeira do Iguape rivers and the Todos os Santos Bay". Within Mischonyx, the split between the two major lineages occurred at~45 Mya, which is consistent with the formation of Valley of Ribeira do Iguape River, 50-56 Mya (Almeida & Carneiro, 1998;Pinto-da-Rocha, DaSilva & Bragagnolo, 2005;DaSilva, Pinto-da-Rocha & Morrone, 2017).
In one of the lineages (Fig. 16), the split dividing species from SMSP from the species from SSP, PR and SC occurred at~48 Mya. This could be the result of the rise of Serra do Mar   (Almeida & Carneiro, 1998;Pinto-da-Rocha, DaSilva & Bragagnolo, 2005). Still inside this lineage, the split between M. intervalensis sp. nov., a species occurring at the northern portion of Ribeira do Iguape River (SSP AoE), from the species from the southern portion of this river (PR and SC AoE) occurred at~28 Mya. The timing of this split is consistent with the results of Inside the other lineage (Fig. 16), the first split occurred at~45 Mya, when M. intermedius diverged from the remaining species. This species is the only one from Esp AoE. It is very likely that the distensive tectonic activity from the tertiary period, which separated the Rio Doce, Paraíba do Sul and São Francisco basins (Cherem et al., 2012;Morais et al., 2005), isolated it from the sister species from Org, LSRJ and Mnt AoE. Many other studies with different taxa corroborate the relevance of the Doce River disjunction in shaping biogeographical patterns (Müller, 1973;Prance, 1982;Amorim & Pires (1996); Pellegrino et al., 2005;Sigrist & Carvalho, 2009;Brunes et al., 2010;Thomé et al., 2010;Silva et al., 2012;Cabanne et al., 2014;DaSilva, Pinto-da-Rocha & Morrone, 2017). The split of M. processigerus (Mnt AoE) from species from LSRJ and Org occurred at~29 Mya, agreeing with the formation of the Paraíba do Sul Valley and its river change of course, during the Oligocene-Miocene (Almeida & Carneiro, 1998;Pinto-da-Rocha, DaSilva & Bragagnolo, 2005;Cherem et al., 2012) In general, the divergence times of Mischonyx species are older than 5 Mya (except for M. clavifemur comb. nov. diverging from M. reitzi comb. nov. and M. parvus comb. nov. diverging from M. squalidus). This is consistent with the speciation events in Promitobates (Bragagnolo et al., 2015). Authors who support the Pleistocene refugia hypothesis have proposed that it happened beginning~5 Mya (Ravelo et al., 2004, Carnaval & Moritz, 2008Carnaval et al., 2009;Holbourn et al., 2014). Therefore, the ancient cooling of the Miocene/Pliocene probably shaped most of the divergences between species inside the genus and the Pleistocene refugia contributed to the most recent speciation events to shape the extant population diversity.
Finally, it is important to stress that M. squalidus appears in all analyses using molecular and TE as sister to M. parvus comb. nov., inside the clade with species from LSRJ. Based on that we conclude that it probably diverged at this AoE in the past and, later, spread all over the Atlantic Forest and Cerrado areas, as discussed in the biogeographical session. Therefore, from now on, in discussions regarding the AoE and the relationship among clades, we will consider M. squalidus as belonging to LSRJ AoE.
The hypothesis of TE under maximum likelihood as the optimality criteria (ML3) We choose ML3 grounded in the following arguments.
In the results of MP3, M. tinguaensis sp. nov. has more than 30 autapomorphies. This long branch encompasses almost one third of all morphological characters coded in the analysis. Additionally, comparing this situation with the number of morphological changes in other harvestmen phylogenies DaSilva & Gnaspini, 2010;DaSilva & Pinto-da-Rocha, 2010;Pinto-da-Rocha & Bragagnolo, 2010), we believe that it it is unlikely that this single species has accumulated so many changes and that the results of ML3 are more likely.
Another reason to choose ML3 is the position of M. tinguaensis sp. nov. in MP3 (Fig. 18A), inside the clade formed strictly by M. spinifrons comb. nov.. It separates the seven sequenced specimens into two polyphyletic lineages. The polyphyly of M. spinifrons comb. nov. seems odd, since the individuals analyzed by us are morphologically identical and there are few site changes in their sequences. In contrast, ML3 (Fig. 17) places M. tinguaensis sp. nov. as the lineage diverging after M. processigerus. This odd placement of M. tinguaensis sp. nov. inside the clade formed by another species' clade in MP3 seems to contribute to the fact that this species has 30 autapomorphies, as discussed in the last paragraph. Similarly, MP3 and B3 (Fig. 18B) place M. insulanus inside the clade formed by M. kaisara, splitting this last species into two polyphyletic lineages. To match this hypothesis, the character changes in this clade containing M. insulanus and M. kaisara, in B3, has several homoplastic changes, as in M. tinguaensis sp. nov. In ML3, instead of this split, M. kaisara is monophyletic and sister to M. insulanus. Therefore, we prefer ML3, since it does not separate exemplars from the same species into polyphyletic lineages.

Diagnosis of previews authors
Although Vasconcelos (2005a) described some characteristics of Mischonyx, and noted two possibly diagnostic characters (yellowish-reddish tubercles on lateral margin of mid-bulge and large median tubercles on area III), Pinto-da-  were the first to propose a diagnosis for the genus, which includes the presence of well-developed median tubercles on mesotergal areas (and add their elliptic form) and the lateral tubercles of mid-bulge paler than the rest of body, in addition to robust spines on the anterior border of dorsal scutum. Pinto-da-  also suggested that Mischonyx is closely-related to Hernandariinae.
In view of our results, we agree that the elliptic median tubercles on area III are diagnostic for Mischonyx. The shape of the tubercle differs in the clade containing M. arlei comb. nov., M. intermedius and M. minimus sp. nov., but is elliptic in all other species of the genus. Along with that, our character "Lateral tubercles on anterior margin of dorsal scutum subequal in size" (#7-0) is roughtly equivalent to "robust spines on the anterior border of dorsal scutum" proposed by Pinto-da- .
In the results of our analyses, Mischonyx is not close to the Hernandariinae species (Piassagera brieni and Pseudotrogulus telluris), even when only morphological characters are considered (Figs. S3-S5). This is in agreement with Pinto-da- , who considered Mischonyx squalidus (Mischonyx cuspidatus in the article) to lie outside of Hernandariinae.

Other taxonomical and topological remarks
Recent publications on the taxonomy and systematics of harvestmen considered G. antiquus as a member of Mischonyx (Kury, 2003;Vasconcelos, 2005a andPinto-da-Rocha et al., 2012). Our morphological analysis also places this species inside the genus. However, these results are not consistent with molecular and TE analyses (Figs. 16-22 and S6-S8). In ML3, it is sister to Ampheres leucopheus, a Caelopyginae. This indicates that the morphological similarities are convergences.
On the other hand, MP2, which does not include morphological characters, places a clade with Multumbo and Deltaspidium species inside Mischonyx, as sister to the clade with species from SMSP, SSP, PR and SC AoE. This group makes no morphological or biogeographical sense, since these species are from Org and LSRJ AoE. However, when we include morphological characters, MP3 does not recover the same clade and excludes Multumbo and Deltaspidium from Mischonyx genus.
The arguments discussed in the last two paragraphs hightlight the importance of combining morphological and molecular data to solve conflicting topologies. Wiens (2004) and Baker & Gatesy (2002) supported the hypothesis that morphological data is important especially when the results from molecular analysis seem problematic. For example, in the research of De Sá et al. (2014), questionable relationships among frog speces became elucidated when morphological and behavioral characters from both larvae and adults were added. Here, we conclude that morphological characters also helped to strengthen the hypotheses and solve some problematic relationships in MP2, consistent with Wipfler et al.
(2016), Lee & Palci (2015) and Giribet (2015) who consider morphological characters fundamental even in the phylogenomics era, since the combination of morphological and molecular data provide independent sources of evidence.
The new composition of the genus after all synonyms, combinations and new species description is as follows: Mischonyx. anomalus (Mello-Leitão, 1936); Mischonyx arlei (Mello-Leitão, 1935b) comb. nov., Mischonyx clavifemur, (Mello-Leitão, 1927a)  We believe that the most plausible phylogenetic hypothesis was recovered using Total Evidence and Maximum Likelihood. Unlike the rival hypotheses, it does not require an unusual number of character changes (apomorphies) leading to M. tinguaensis sp. nov., it has high bootstrap support for Mischonyx and is well supported by morphological synapomorphies. The Mischonyx clade is supported by the following morphological characters: lateral tubercles on anterior margin of dorsal scutum with the same size, elliptic tubercles on area III, absence of prolateral apophysis on females, femur prolaterally curved, three to six apophysis on the apical half of retrolateral row on femur IV and brown as the general body color. There are two major clades inside Mischonyx: one with species from LSRJ, Mnt, Org and Esp AoE, and the other with species from SMSP, SSP, PR and SC AoE. The divergence time of these clades are in agreement with geological events. We estimate that Mischonyx clade diverged 50.53 Mya, and inside the genus there are two major clades. One of them cointains species from Paraná, Santa Catarina, South of São Paulo and Serra do Mar Areas of Endemism and the other has species from Espinhaço, Bocaina, South coast of Rio de Janeiro and Serra dos Órgãos Areas of Endemism. The first split inside these two clades occurred at 48.94 and 44.80 Mya, respectively.