Chemical and evolutionary analysis of the scent gland secretions of two species of Gonyleptes Kirby, 1819 (Arachnida: Opiliones: Laniatores)

The subfamily Gonyleptinae is the second largest in Gonyleptidae, harboring over 100 species. Gonyleptinae is polyphyletic, nestled in the clade K92, and despite its richness, several species of that subfamily have not had their chemicals of the defensive secretions analyzed. Among these are Gonyleptes curticornis (Mello-Leitão, 1940) and G. horridus Kirby, 1819, the latter being particularly important, because it is the type species of the genus, which in turn names the subfamily Gonyleptinae. Gonyleptes horridus is also used in many phylogenetic analyses, be it using morphological or molecular data. The chemical study of the secretions of these two species by GC–MS and 1H NMR showed the presence of 1-(6-isopropyl-3,4-dihydro-2H-pyran-2-yl)-methylbutanone, 1-(6-isopropyl-3,4-dihydro-2H-pyran-2-yl) isobutanone and 4-methyl-1-hepten-3-one in both species. On the other hand, 4-methyl-1-hexen-3-one was observed only in G. curticornis, and 7-methyl-2-octanol is exclusive of G. horridus. All vinyl-ketones identified have already been described for Gonyleptidae. We ran an Ancestral Character State Reconstruction (ASR) analysis under three different conditions to infer the evolution of the identified compounds (based on modified characters of a previous study) and their chemical nature (multistate character, either as alkylphenol, benzoquinone or vinyl-ketone) on a modified Gonyleptidae phylogeny. Our results corroborate previous studies that alkylphenol is the ancestral most condition, changing to benzoquinone in the ancestor of Gonyleptidae or even earlier in a grassatorean ancestor depending on the analysis. Vinyl-ketones are a synapomorphy of K92. We briefly discuss character codifications and use of weights of ASR analyses of specific compounds, which were inconclusive. 1-(6-isopropyl-3,4-dihydro-2H-pyran-2-yl) isobutanone is shared by both Gonyleptes species and described for the genus Sodreana Mello-Leitão, 1922. 1-(6-(1-methyl-propyl)3,4-dihydro-2H-pyran-2yl)2-methylbutanone and 4-methyl-1-hepten-3-one are also shared by both Gonyleptes species and described for Moreiranula saprophila. From a taxonomic standpoint, combinations of specific compounds might help to diagnose supraspecific groups but given our limited sample, such decision should be taken with care and further tested. Finally, 7-methyl-2-octanol is described for the first time in Gonyleptidae, emphasizing the chemical diverse nature in the K92 clade.


Introduction
With approximately 6900 described species (Kury et al. 2022a), harvestmen (Opiliones) are the third largest group of Arachnida, after Acari (ticks and mites) and Araneae (spiders) (Selden 2007). Opiliones is taxonomically arranged in four monophyletic extant suborders: Cyphophthalmi, Eupnoi, Dyspnoi and Laniatores (Giribet and Kury 2007). Scent glands have been the focus of behavioral (Hara and Gnaspini 2003;Caetano and Machado 2013;Segovia et al. 2015) and chemical studies (Hara et al. 2005;Caetano and Machado 2013;Raspotnig et al. 2014Raspotnig et al. , 2017. The scent glands produce secretions primarily used for defense against predators (Gnaspini and Hara 2007). They have been shown to repel some predators, but not others (Machado et al. 2005;Souza and Willemart 2011;Silva et al 2018), and they can also be used as an alarm pheromone (Machado et al. 2002). Whereas scent gland secretions in harvestmen have traditionally been considered to be products of de novo synthesis, there is evidence for the unusual case of sequestration-derived gland constituents (Raspotnig et al. 2022). Defensive secretions can be emitted in different ways, forming a chemical shield around the body: (i) simply exhaling from the ozopore (with little or no emission of enteric liquid); (ii) as a droplet near the ozopore, either remaining there, being transferred to other parts of the body or by directly rubbing of the droplet on the aggressor using legs, or (iii) emitting the secretion as a jet (Gnaspini and Hara 2007). These forms of emission of the defensive secretion are potential sources of characters for phylogenetic analyses. Caetano and Machado (2013), for example, proposed the non-emission of secretions in droplets or their non-accumulation between the most proximal regions of legs I and II as a synapomorphy of the clade K92 (Kury 1992).
Currently, approximately 140 species from all suborders have had their chemical defenses studied, mainly in the suborders Laniatores and Eupnoi (approximately 50 species from each one) (Gnaspini and Hara 2007;Raspotnig 2012;Raspotnig et al. 2012Raspotnig et al. , 2014Raspotnig et al. , 2017Caetano and Machado 2013;Rocha et al. 2013;Wouters et al. 2013;Segovia et al. 2015). Since scent glands are autapomorphic for the order (Shultz 1990), Opiliones are suitable for phylogenetic chemosystematics, i.e. making evolutionary inferences (Hara et al. 2005;Raspotnig 2012;Caetano and Machado 2013;Wouters et al. 2013;Raspotnig et al. 2014Raspotnig et al. , 2017) based on secretion from homologous glands (Raspotnig et al. 2017). In Laniatores, the infraorder Grassatores features phenols and benzoquinones, as well as acyclic compounds that superficially resemble some compounds found in Eupnoi (Hara et al. 2005;Raspotnig 2012). As more compounds have been identified in the defensive secretion, gradually more precise mapping, and inferences of the evolution of such trait became possible for the whole order (Raspotnig 2012;Caetano and Machado 2013;Wouters et al. 2013;Raspotnig et al. 2010Raspotnig et al. , 2017. Despite recent efforts to identify the compounds in the defensive secretions in an increasing number of species, a few, taxonomic-wise key species have been overlooked so far. Gonyleptidae is one of the largest families of Grassatores, and it has been the focus of many phylogenetic analyses (Pinto-da- Rocha et al. 2014;Benavides et al. 2021). One of the most used species in such analyses is Gonyleptes horridus Kirby, 1819 (for instance, Pinto-da- Rocha et al. 2014;Benedetti and Pinto-da-Rocha 2019;Ázara et al. 2020), because it is the type-species of Gonyleptes Kirby, 1819, that names both the polyphyletic subfamily Gonyleptinae (second largest one; Pinto-da- Rocha et al. 2014) and the family. However, despite its pivotal taxonomic importance, the defensive secretion of G. horridus is still unknown. In this study, we identify the compounds in the defensive secretion of this key species. Gonyleptes is composed of 26 species that inhabit mostly the Atlantic Rain Forest (Kury 2003;Kury et al. 2022a). Gonyleptes curticornis (Mello-Leitão, 1940), a species phylogenetically related to G. horridus, and abundant in the northern coast of São Paulo State, is also studied here for comparison. Both species are included in the clade K92, a repeatedly retrieved clade in phylogenetic analyses, composed of the subfamilies Gonyleptinae Sundevall, 1833, Sodreaninae Soares & Soares, 1985, Caelopyginae Simon, 1879, Hernandariinae Sørensen, 1884and Progonyleptoidellinae Soares & Soares, 1985(Kury 1992. Thus, we hope to provide further data that combined with those already available, hopefully will shed some light to understand the evolution of the scent gland secretion of Gonyleptinae focusing on the true Gonyleptinae. Hitherto no species of Gonyleptes had its defensive secretion accessed, as former Gonyleptes saprophylus Mello-Leitão, 1922 is now placed in Moreiranula Roewer, 1930.

Collection and maintenance
We collected 4 females and 9 males (totalizing 13 adult specimens) of Gonyleptes horridus at Tijuca National Park, Rio de Janeiro city, Rio de Janeiro State, Brazil, from 2 to 4 of December 2019. We collected 10 males and 10 females (totalizing 20 adult specimens) of Gonyleptes curticornis at the Cachoeira da Fazenda in the city of Ubatuba (State of São Paulo, Brazil) on February 20 th , 2018. We collected all the animals manually at night. Once in the laboratory, we maintained specimens in plastic containers with paper towels on the bottom and a wet cotton ball to provide humidity.

Collection of secretions
We collected the defensive secretion only from adult specimens in 1-2 days following the field collection. We used each specimen only once, and we deposited voucher specimens in the Museum of Zoology of the University of São Paulo, in Brazil (MZUSP). To collect secretions, we held the specimen between our fingers and pressed the abdomen dorso-ventrally. We placed a microscopy glass slide on the dorsal scutum between legs III and IV and used it as a screen to retain the defensive secretion. We used micropipettes (capillary glass tube of known volume) to collect the released defensive secretion. Their content was dissolved either in vials with 3.5 µL of methanol (HPLC grade-TEDIA) for CG-MS analysis or 1 μl of DMSO-d 6 for the 1 H NMR analysis, always separating the sexes. We kept vials for both types of analyses in the freezer at -10ºC.

Chemical analysis
We ran the CG-MS analysis at the Department of Botany, Institute of Biosciences of University of São Paulo. We injected 1 µL of samples dissolved in 100 µL of methanol samples into a gas chromatograph coupled to a mass spectrometer (GC-MS, Agilent 6890N/5975) with a HP-5MS capillary column (30 m × 0.25 mm × 0.25 µm) using splitless mode. We adjusted the initial temperature at 30 °C for 5 min, increasing at 5 °C·min −1 until 280 °C, remaining at this temperature for 5 min. Helium was the carrier gas at a constant flow rate of 1 mL·min −1 . The injector, ions source, and quadrupole temperatures were 280, 230 and 150 °C, respectively. MS detection was achieved with electron multiplier voltage (EM) at 70 eV, operating in the full scan acquisition mode with a range of 30-500 units of atomic mass and 3.1 scans.s −1 . We calculated the retention index for each compound detected in the output identified with 90% or more of similarity to spectra available in the NIST library. To perform this calculation, we injected a standard C 8 -C 20 hydrocarbon column (Sigma-Aldrich), and used the following formula: IR = 100 *((tc-tn / tn + 1-tn) + n) (Viegas and Bassoli 2007). Afterward, we compared the calculated values with the retention index described in the literature, thus allowing the identification of the compounds.
We used spectral data such as 1 H NMR and mass spectra to elucidate the structure of the compounds. We obtained the NMR spectra on a Bruker Ascend™ III 600 14.1 T instrument operating at 600 MHz. We used DMSO-d 6 as solvent and tetramethyl silane (TMS) as internal reference. We recorded chemical shifts δ in ppm and coupling constants J in Hz.

Mapping the compounds
We used modified characters of Caetano and Machado (2013) (Tables 3 and 4) in the present study. The correspondence between our set of characters to those of Caetano and Machado (2013) are provided in Table 3, indicated between parentheses and the # symbol. We included a character (#34) based on the compound chemical class (either alkylphenol, benzoquinone or vinyl-ketone) of the compounds that is multistate instead of the majority of those of Caetano and Machado (2013) that are coded in absent/present fashion. This character might be an oversimplification considering the available data nowadays, but it allowed for the proposition of a transformation series that is otherwise hindered. This approach is similar to that of Raspotnig et al. (2014). The proposition of a transformation series using the identified compounds per se is considerably hampered because of the difficulty to propose homologue compounds given their vast amount (so far at least 32 in Gonyleptidae, see Tables 3  and 4). Given this limitation, we used character coding as those of Caetano and Machado (2013) for the identified compounds per se and we attempted a further approach to analyze those in an evolutionary perspective mapping them in a Gonyleptidae phylogeny. For that, we used the Ancestral Character State Reconstruction (ASR) approach, which relies on a model of evolution to recover ancestral states applied to a given phylogeny of the studied group (see Joy et al. 2016). Considering that there are over 30 compounds identified in Gonyleptidae, we opted to restrict a more detailed analysis to those identified in the defensive secretion of Gonyleptes that is the focus of this study.
The tree we chose to apply the ASR was a modified phylogeny of Gonyleptidae from Pinto-da- Rocha et al. (2014). Our choice is based on the following arguments: (i) it is more comprehensive than others available (such as Caetano and Machado 2013;Kury et al. 2022b), and (ii) it included more taxa with identified defensive secretion. The phylogeny proposed by Caetano and Machado (2013), despite having the merit of being the first phylogeny of Gonyleptidae that is based on non-morphological data, presents incongruences from the taxonomic point of view (Kury 2014;Pinto-da-Rocha et al. 2014;Carvalho and Kury 2018) that may lead to mistaken inferences. Gonyleptidae systematics have greatly advanced in the past decade, and most of the studies (for instance Kury 2014, Bragagnolo et al. 2015, and Carvalho et al. 2021) are congruent with Pinto-da- Rocha et al. (2014) phylogeny. Many of the polyphyletic groupings detected in Pinto-da- Rocha et al. (2014) were dismembered and placed into monophyletic units, such as Gonyassamiinae Soares & Soares, 1988, Pachylinae sensu stricto, Roeweriinae Carvalho and Kury 2018, and so on, just to name a few (Kury et al. 2020). We used the taxa names based on Kury et al. (2022b), because it clarifies many of 1 3 classification incongruences, such as Moreiranula saprophila, placed formerly in the genus Gonyleptes. We deleted taxa which defensive secretions were not studied yet except for Ampheres leucopheus (Gonyleptidae, Caelopyginae), to avoid an oversimplification of the phylogeny. Keeping those taxa which defensive secretions are unknown would hamper the analysis due the large amount of missing data, thus causing more noise than desirable in the reconstruction of the ancestral states. To facilitate the indication of the character in Results and discussion section, we used the "#" annotation to refer to a character. Thus, #34, for instance, means character 34 of Tables 3 and 4. Gonyleptidae sensu Pinto-da- Rocha et al. (2014) refers to Gonyleptidae plus Ampycinae and Manaosbiinae.
For the ASR, we used a similar routine as Raspotnig et al. (2014). We used the ASR Package (Trace > Character History Source > Reconstruct Ancestral States) for Mesquite ver. 3.2 (Maddison and Maddison 2021). We used Parsimony models as follows: (i) unordered, assuming equal weights for gains and losses of compounds; (ii) step matrix with factors of 3 and 2 to favor loss against gains (the higher the factor, the stronger the bias), as hypothesized by Raspotnig et al. (2014); (iii) ordered in case of the multistate character (#34). This paper is a first approach to infer the evolution of the identified compounds in the defensive secretion of true species of Gonyleptes rather than running a total evidence type of analysis or a reanalysis of Gonyleptidae semiochemicals like Raspotnig et al. (2014) did.

Results and discussion
Specimens of G. horridus and G. curticornis emitted the defensive secretion in the form of a jet posteriorly (Gnaspini and Hara 2007). Eventually the emitted secretion spread over the carapace, where it evaporated. According to GC-MS and 1 H NMR analysis, there were five compounds in total in the defensive secretion of G. curticornis and G. horridus. According to the chromatographic conditions established for the separation of the compounds present in the G. curticornis secretion, we identified 1, 2, 4 and 5 in the chromatogram (Table 1, Fig. 1). On the other hand, we found compounds 2, 3, 4 and 5 in G. horridus secretion (Table 1, Fig. 2). The retention index values calculated and those observed in literature of all the identified compounds showed similar values, except for 4-methyl-1-hepten-3-one (2). However, its identification can be confirmed by the 1 H NMR data together with the fragments obtained by mass spectrometry.
However, analysis of the 1 H NMR spectral data of the mixtures of molecules present in these two secretions, allows only the identification of the major compounds, being 4-methyl-1-hexen-3-one (1) in G. curticornis and 4-methyl-1-hepten-3-one (2) in G. horridus.
The compound with the retention peak at 12.47 min (compound 2) in the defensive secretion of G. horridus (Fig. 1) and G. curticornis (Fig. 2) showed the molecular ion with m/z 126 and fragments of m/z 111, 84 and 55 (Table 1). The fragment of m/z 55 corresponds to an α-carbonyl cleavage, while the fragment of m/z 84 results from McLafferty rearrangement (Silverstein et al. 2005). According to the 1 H NMR analysis, the presence of a terminal double bond is evidenced by the presence of three double doublets at δ 6.25 (J = 18.0 and 1.2 Hz), δ 5.85 (J = 10.5 and 1.4 Hz) and δ 6.47 (J = 18 and 10.4 Hz) close to a ketone. The methyl vicinal to the methine carbon can be evidenced by the signals of a multiplet at δ 2.83 (1H) and the doublet at δ 0.98 (J = 7.5 Hz, 3H). The triplet at δ 0.79 (J = 7.8 Hz, 3H) indicates a terminal methyl coupling the methyl hydrogens. The multiplet at δ 1.59 (2H) refers to the methylene at position 5 of the molecule, which is unprotected as a function of the carbonyl, while the signal at δ 1.34 (m, 2H) refers to the methylene hydrogens of carbon 6. All these data set confirmed this compound as 4-methyl-1-hepten-3-one (2) ( Table 2).
The compound with the retention time of 14.11 min (compound 3) in the defensive secretion of G. horridus showed molecular ion of m/z 144. The loss of 15 units (M-15) + from the molecular ion (m/z 144), indicates the presence of a methyl group in the molecule. The fragment of m/z 45 and m/z 57 refers to [CH3CH = OH] + , and [CH 3 CH 2 CHCH 3 ] + , respectively. The losses of C 4 H 9 from the molecular ion generate the fragment with m/z 87. This dataset indicates that the compound in question is 7-methyl-2-octanol 3. So far, 7-methyl-2-octanol 3 can be found in cape chamomile oil (Mierendorff et al. 2003) but has not been described for other opilionid defensive secretion.
To infer the evolution of the chemical nature (either alkylphenol, benzoquinone or ketone), we proposed it as a multistate character (#34, Tables 3 and 4) and performed the ASR on a modified version of phylogeny proposed by Pinto-da- Rocha et al. (2014). The analyses assuming #34 as unordered with equal weights or as ordered without using step matrix yielded the same result (Fig. 3), as detailed further on. Alkylphenols is the most ancestral state (Cryptogeobiidae, here represented by Camarana flavipalpi Soares, 1945 and Cosmetidae), changing to benzoquinones in the ancestral of Gonyleptidae (sensu Pinto-da- Rocha et al. 2014) and then, to vinyl-ketones in clade K92, with independent reversals to alkylphenols in Progonyleptoidellus striatus (Roewer, 1913) and Khazaddum inerme (Soares & Soares, 1947) (Fig. 3). This evolution pattern is overall congruent with that proposed by Raspotnig et al. (2014) but the different phylogenies used to map hinders further comparisons between the present and the former study. It is noteworthy to mention that Raspotnig et al. (2014) proposed the production of benzoquinone in Grassatores after the split-off of Stygnopsidae (i.e. rather early in grassatorean evolution) while our results suggests it in the ancestral of Gonyleptidae sensu Pinto-da-Rocha et al. (2014) even though we scored the presence of both alkylphenol and benzoquinone in Cosmetidae. Raspotnig and collaborators (2014) represented Cryptogeobiidae by two species, Camarana flavipalpi and Cryptogeobius crassipes Mello-Leitão, 1935 in their analyses, but then they were considered as Gonyleptidae (Fig. 2 of Raspotnig et al. 2014), leading them to an interpretation as a reversal to the ancestral state of alkylphenols inside the Gonyleptidae. In our analyses, C. flavipalpi is represented as the root of the tree, reinforcing alkylphenols as the ancestral state. Recent phylogenies of Gonyleptoidea (Benavides et al. 2021) consider Manaosbiinae and Ampycinae as phylogenetically close to but not Gonyleptidae. In that case, the production of benzoquinones is pushed earlier in Grassatores' evolution, corroborating Raspotnig et al. (2014). Those analyses corroborate that benzoquinones are indeed plesiomorphic for Gonyleptidae. The production of vinyl-ketones as synapomorphic for clade K92 is also corroborated here (Caetano and Machado 2013;Raspotnig et al. 2014).
If the evolution of the chemical nature in grassatorean harvestmen might be roughly settled, on the other hand, inferring the specific ancestral compounds over the phylogeny is an entirely unsettled issue. This is mostly because of  the reasons pointed out in Materials and methods section, namely the difficulty (or rather impossibility at the moment) to propose homology between the identified compounds, thus hampering any series of transformation. Unveiling the biosynthetic pathways is vital to propose such series, but it is known only for Iporangaia pustulosa Mello-Leitão, 1935 ) so far in Gonyleptidae. Considering the available data at hand, we used a modified version of the chemical characters of Caetano and Machado (2013) (Tables 3 and 4). Analyzing compounds 1, 2, 4, 5, and 5-methyl-1-hexen-3-one (#16, #17, #22, #26 and #18, respectively) as unordered neomorphic characters using equal weights resulted in all of them evolving independently (Figs. 4,5,6). This result is similar to that in Caetano and Machado (2013). The analysis assuming a weighting factor of three for gains in an ordered character state resulted in all of those as ancestral to K92 or even earlier, in the ancestor of Gonyleptidae (sensu Pinto-da- Rocha et al. 2014). However, considering that both Ampycinae and Manaosbiinae lack vinyl-ketones, it would be more realistic to propose that those compounds are synapomorphic for K92 with several losses (or deactivation of its biosynthetic pathway or further changed in another compound). Hence, proposing the ancestral specific vinyl-ketone for K92 remains inconclusive. Those results are probably biased by the applied weight since additional analyses using weighting factor of two render many ambiguities (uncertainties) of the ancestral state. Raspotnig et al. (2014) performed the first ancestral state character reconstruction in Gonyleptidae including the knowledge of biosynthetic pathways of harvestmen defensive secretion largely based on Rocha et al. (2013). It is an improvement over the analysis of Caetano and Machado (2013) (that has merit of its own), and we want to stress the importance of including relevant data to perform comparatively more precise evolutionary inferences. However, methodologically wise, an analysis also can be strongly biased by weighting character state changes or ordering them, resulting in an equally biased outcome-it is possible to see the effect by comparing the results of the ASR using a stronger factor in step matrix (three versus two, see Fig. 4, 5, 6). A possible, more suitable way to include the knowledge from biosynthetic pathways might be through the propositions of series of transformations, i.e. proposition of which chemical compound (or precursor) changed to another one. The amount of available data of defensive secretion biosynthesis in Opiliones as a whole is very scarce to our knowledge except for Rocha et al. (2013). For the time being, we agree that critically using weighting factors in inferences as proposed by Raspotnig et al. (2014) is the best alternative. This scenario restrains much of our power to make inferences,  Table 4 Modified chemical character matrix from Caetano and Machado (2013)  but we believe that persisting in identifying the defensive secretion of more taxa and promoting studies of biosynthesis of harvestmen defensive secretion might gradually replace usage of weights and neomorphic characters by series of transformations. Thus, most probably those will contribute to unveil Gonyleptidae chemosystematics, as the past decade has gradually shown. From a taxonomic standpoint, we noted that despite the difficulty to propose the ancestral specific compounds on the phylogeny, certain combinations of compounds are exclusive of supraspecific (genus or subfamily categories, for instance) groups in K92. Analyses using equal weights and unordered character changes further corroborate this, seemingly supporting only one grouping, with few convergences. In such analyses, compounds 2, 4 and 5 are shared by both species of Gonyleptes and compounds 2 and 4 are also present in Moreiranula saprophila (Fig. 5). Whether compounds 2 and 4 are synapomorphic or homoplastic for those species depend on the performed analysis (using a weighting factor of three and equal weights, respectively). Fig. 3 Ancestral character state reconstruction (ASR) parsimony model-based of the class of compounds of the scent glands (character #34) in a selected clade of Grassatores (Arachnida, Opiliones), using a modified version of the phylogeny obtained by Pinto-da- Rocha et al. (2014). This interpretation was obtained both with unordered and ordered modeling. Alkylphenols were coded as "0" (white), benzoquinones as "1" (green) and ketones as "2". In some cases, the species could have compounds of two different classes, and they were coded as such (Mischonyx squalidus, Neosadocus maximus, Iporangaia pustulosa). Although we lack data from Ampheres leucopheus, Promitobates ornatus and Longiperna concolor we kept them in the tree to avoid oversimplification. The alkylphenols seem to be ancestral in Grassatores, with the appearing of benzoquinones in the ancestral of Gonyleptidae sensu Pinto-da- Rocha et al. (2014), which includes Manaosbiinae and Ampycinae. The ketones appear in the ancestral of clade K92, which includes Caelopyginae, Gonyleptinae, Hernandariinae, Progonyleptoidellinae and Sodreaninae. These results corroborate Caetano and Machado (2013) and Raspotnig et al. (2014) Anyway, the combination of compounds 2, 4 and 5 is unique for Gonyleptes and hence, could be employed to diagnose it. Similarly, compound 5 is shared by both species of Gonyleptes and Sodreaninae (Fig. 6). However, the combination of compound 5 with 5-methyl-1-hexen-3-one is exclusive to Sodreaninae species. We are aware that the limited sampling in an analysis might be prone to biased inferences or diagnosis. However, as Raspotnig et al. (2014) already pointed out, it is realistically impossible to include all existing taxa in an analysis. Furthermore, the indication of an evolutive novelty (autapomorphies) even with reduced sampling might reveal to be more comprehensive in the future (hence, becoming synapomorphies), as commonly seen for morphological traits in systematics (Mendes 2011;Gueratto et al. 2017;Hara et al. 2012). Considering that clade K92 has been focus of many phylogenetic studies recently (Pinto-da- Rocha et al. 2014;Kury et al. 2022b), and the boundaries of its supraspecific taxa are still relatively uncertain, we believe that the semiochemicals might contribute taxonomically as diagnostic features. Gonyleptes is composed of 22 species (Kury et al. 2022b), but none have been studied so far. This is the Fig. 4 Ancestral character state reconstruction (ASR) parsimony model-based of ketones in clade (Manaosbiinae + Ampycinae) + K92 (Opiliones, Gonyleptidae) of characters #11 to #16, using a modified version of the phylogeny obtained by Pinto-da- Rocha et al. (2014). The characters were considered neomorphic and we coded "0" for absences and "1" for presence. The trees represent the interpretation with stepmatrix favoring losses of compounds against gains with a factor of three (as in Raspotnig et al. 2014). However, we represented also in "Navajo-rugs" where there was divergence in the ASR using stepmatrix with a factor of two, and with unordered modeling, where losses and gains are equally considered. Although we lack data from Ampheres leucopheus we kept it in the tree to avoid oversimplification first step toward a chemical characterization of the genus. Future studies, including more taxa from Gonyleptes as well as other genera and subfamilies will corroborate or not the usefulness of semiochemicals for this purpose. Of course, it also is prone to modifications like those based on morphological characters, as more data are gathered and evaluated.
Due to shortage of behavioral studies on the effectiveness of defensive chemicals and natural history in Opiliones, it is unfortunately not possible to comment on the use of such chemicals against predators or in other contexts. Finally, we report 7-methyl-2-octanol (3) that is present in G. horridus defensive secretion for the first time in Opiliones. It is autapomorphic for this species and the first alcohol in (Opiliones, Gonyleptidae) of characters #17 to #22, using a modified version of the phylogeny obtained by Pinto-da- Rocha et al. (2014). The characters were considered neomorphic and we coded "0" for absences and "1" for presence. The trees represent the interpretation with stepmatrix favoring losses of compounds against gains with a factor of three (as in Raspotnig et al. 2014). However, we represented also in "Navajo-rugs" where there was divergence in the ASR using stepmatrix with a factor of two, and with unordered modeling, where losses and gains are equally considered. Although we lack data from Ampheres leucopheus we kept it in the tree to avoid oversimplification a gonyleptid secretion to our knowledge, emphasizing that clade K92 is indeed diverse in semiochemical aspect. in collecting harvestmen. We also thank the editor and anonymous referees for the comments and suggestions that greatly improved the manuscript.
Author contributions All authors contributed to the conceptualization, formal analyses, investigation, methodology, supervision and writing of the manuscript. All authors have read and agreed to the published version of the manuscript. Data availability All data generated or analysed during this study are included in this published article.

Conflict of interest
The authors declare that they have no conflicts of interest.

Fig. 6
Ancestral character state reconstruction (ASR) parsimony model-based of ketones in clade (Manaosbiinae + Ampycinae) + K92 (Opiliones, Gonyleptidae) of characters #23 to #27, using a modified version of the phylogeny obtained by Pinto-da- Rocha et al. (2014). The characters were considered neomorphic and we coded "0" for absences and "1" for presence. The trees represent the interpretation with stepmatrix favoring losses of compounds against gains with a factor of three (as in Raspotnig et al. 2014). However, we represented also in "Navajo-rugs" where there was divergence in the ASR using stepmatrix with a factor of two, and with unordered modeling, where losses and gains are equally considered. Although we lack data from Ampheres leucopheus we kept it in the tree to avoid oversimplification