/lp/ou_press/to-complicate-or-to-simplify-phylogenetic-tests-of-complexity-trends-yOEc0NjydO

- Publisher site
- See Article on Publisher Site

Abstract Spiders may be good models for studying the evolution of genitalia given their peculiar copulatory mechanism. Gnaphosidae, in particular, have remarkable diversity in copulatory apparatus. The family contains species with bipartite and tripartite palps. Previous studies suggest intermediate palp complexity as the ancestral condition, with an evolutionary trend to simplification. However, this hypothesis has never been tested on a phylogenetic background. The aim of this study was to explore the macroevolutionary patterns and processes related to copulatory organs in Gnaphosidae. The trend to simplicity and predictions about genital evolution hypotheses were tested using phylogenetic comparative methods. A bipartite palp with embolus fused to tegulum and intermediate complexity was found to be the plesiomorphic condition, but there was no trend towards simplification, nor to fusion of sclerites. The same intermediate complexity with no trend was found for females. Additionally, we discovered that the complexity of female and male copulatory organs did not coevolve, and that male genitalia evolved faster than female, favouring the cryptic female choice hypothesis. In Gnaphosidae, there is no general best configuration of male genitalia, and the female choice between a simple or complex palp would be idiosyncratic and not predictable by epigynum complexity. Homology and evolution of genitalic structures are also discussed. comparative method, complexity, cryptic female choice, epigynum homology, evolutionary trend, generalized estimating equations, generalized least square, lock and key, palp homology, sexual antagonistic conflict INTRODUCTION Copulatory organs are characteristic of most animal lineages, present an impressive variety of forms and are thought to evolve relatively faster than other traits (Eberhard, 1985, 2010a; Huber, 2003). Many evolutionary hypotheses have been proposed to explain this great diversity of genitalia, including: (1) pleiotropy, which postulates that genitalia are neutral and evolve with other genetically correlated traits; (2) natural selection, according to which the genitalia would evolve to avoid hybrid formation through a species-specific lock-and-key mechanism; and (3) sexual selection, which predicts that some traits would be favoured by post-copulatory mechanisms, such as cryptic female choice, sperm competition or sexually antagonistic coevolution (Arnqvist, 1997; Hosken & Stockley, 2004; Eberhard, 2010a, b, 2015). Sexual selection by cryptic female choice has recently been indicated as the main general mechanism to explain genital diversity in spiders (Eberhard, 2010a, 2015; Eberhard & Huber, 2010). Nevertheless, few studies test specific hypotheses about genitalia evolution based on phylogenetic comparative methods, and some of them indicate alternative explanations for genital diversity within the group (Kuntner, Coddington & Schneider, 2009). In fact, studies on arthropod genitalia suggest that genital evolution might be more complex, not explainable by a single general mechanism (Simmons, 2014), indicating that much effort is still necessary for the understanding of genital evolution. Spiders may be good models for studying the diversity and evolution of genitalia (Eberhard, 2004a). In Araneae, the transference of sperm evolved in a unique manner among animals through the modification of the male palps into a peculiar secondary copulatory apparatus. The copulatory bulb is a compound of sclerotized structures with reduced sensorial and muscular organs that need to fit precisely into the female genitalia (Eberhard & Huber, 2010; but see Lipke, Hammel & Michalik, 2015; Sentenská et al., 2017 for evidence of neurons in the copulatory bulb). Spider species-level systematics is based mainly on the morphology of male and female copulatory organs. Therefore, the number of described species is a good proxy of genitalia diversity in the order and suggests the existence of > 47000 different forms of copulatory organs in each sex (World Spider Catalog, 2018). Even though the correspondence between the genital morphology and number of species could sometimes be misleading (Jocqué, 2002; Huber, 2004), the great diversity of forms and elements in spiders’ genitalia that are perceived by arachnologists is indubitable and begs for an explanation. The putative plesiomorphic male palp bulb in spiders is a tripartite structure, with proximal (subtegulum), median (tegulum) and terminal (embolus) divisions (Platnick & Gertsch, 1976; Kraus, 1978; Coddington, 1990; Sierwald, 1990). This basic ground plan may have evolved in different directions among higher lineages, resulting in relatively simple copulatory bulbs with fused structures, as in some Mygalomorphae (Bertani, 2000) and Synspermiata spiders (Platnick & Dupérré, 2009), or more complex organs, with additional sclerites and variable shape, as in most Araneoidea spiders (Coddington, 1990). There is less information about the evolution of female genitalia, but it is suggested that these might have evolved from a simple system of glandular sperm storage with no external structure to a more complex system with additional ducts and receptacula, and with external sclerotized structures with folds and projections (Forster, 1980; Sierwald, 1989). Gnaphosidae spiders, besides their great diversity (it is the sixth most species-rich spider family; World Spider Catalog, 2018), are remarkable in having species with simple palps, with a bipartite bulb and few structures (e.g. Litopyllus Chamberlin, 1922; Platnick & Shadab, 1980a), species with intermediate conditions, with tripartite palp and few sclerites (e.g. Gnaphosa Latreille, 1804; Platnick & Shadab, 1975), and species with several additional structures on a tripartite palp (e.g. Zelotes Gistel, 1848; Senglet, 2004). Some studies on palpal structure homology among gnaphosids suggest that an palp of intermediate complexity might be the ancestral condition for the family, from which the more complex and simpler palps would have evolved, but with a major evolutionary trend to fusion of sclerites and copulatory bulb simplification (Zakharov & Ovtcharenko, 2011, 2013). However, those hypotheses have never been tested under a phylogenetic perspective, because no Gnaphosidae phylogeny was available until recently, and the existing studies have been limited to a few taxa. Regarding female genitalia, gnaphosids are considered entelegyne spiders, which means that the internal genitalia (vulva) have two pairs of openings to the exterior, one leading to the copulatory duct and another arising from the fertilization duct, forming a flow-through path for sperm, and have a sclerotized external structure, the epigynum (Coddington & Levi, 1991; Garrison et al., 2016). Both the epigynum and the vulva range from simply structured organs, with short ducts and an undivided external plate (e.g. Cesonia Simon, 1893; Platnick & Shadab, 1980b), to more complex structures, with long, coiled ducts and the epigynum with ventral folds and projections (e.g. Apopyllus Platnick & shadab, 1984; Platnick & Shadab, 1984). However, only some zelotine spiders have been explored in a more detailed comparative morphological study of female genitalia (Senglet, 2004), and little is known about the homology between their components. The patterns of genital evolution and mechanisms involved in copulatory organ diversification in Gnaphosidae were barely known. The aim of this study was to contribute to the understanding of genital evolution through the exploration of macroevolutionary patterns and processes related to the diversity of copulatory organs in Gnaphosidae s.s. (Azevedo, Griswold & Santos, 2017). More specifically, the evolutionary trend in complexity and some predictions of different genital evolution hypotheses (cryptic choice, arms race or lock and key) were tested using phylogenetic comparative methods. We also aimed to discuss the homology and evolution of some genital structures through Gnaphosidae s.s. MATERIAL AND METHODS Data set Our analyses were based on a phylogenetic hypothesis and genitalic characters taken from the morphological matrix of previous work on Gnaphosidae systematics (Azevedo et al., 2017). Only characters that could easily indicate complexity (see below) were extracted from Azevedo et al. (2017) to compile our matrix, resulting in 46 male and 14 female characters, summing 60 characters in total (Supporting Information, Appendix S1). To explore uncertainties in phylogeny, all trees obtained under different implied weights parameters (k value), the Bayesian consensus tree and the trees of the stationary phase of the Markov chain of the Bayesian analysis that were presented by Azevedo et al. (2017) were used (see details below). For graphical and didactic purposes, the working phylogenetic hypothesis presented by Azevedo et al. (2017: fig. 3) was chosen for showing the optimization of characters and complexity (see below). WinClada version 1.00.08 (Nixon, 2002) and Mesquite version 3.04 (Maddison & Maddison, 2015) were used to optimize discrete characters and the continuous complexity traits, under maximum parsimony and squared change parsimony (Maddison, 1991), respectively. The optimization of discrete characters was used to discuss the secondary homology and evolution of some structures. The matrix with all 60 discrete characters was optimized on the tree with all taxa (100 terminals), including the outgroup and taxa with missing data, to avoid ambiguous optimization at the root of the clade of interest, the Gnaphosidae s.s. clade (Azevedo et al., 2017: fig. 3). Complexity Morphological complexity can be hard to define and measure (McShea, 1991; Adami, Ofria & Collier, 2000; Adami, 2002; Ramírez & Michalik, 2014). Here, the structural complexity concept was used, in which complexity is a function of the number of parts in the organism and of the irregularity of their arrangement (McShea, 1991). This concept allows the study of complexity even if little is known about the function of structures (McShea, 1991). Gnaphosidae are a very diverse group. Studies on the mechanical interaction of male and female genitalia during copulation are scarce and, in many genera, such studies are difficult to perform because the spiders can be hard to find in abundance in nature. In contrast, there is already plenty of information on the morphology of the genitalia, and much more could be obtained from museum specimens, which makes the structural complexity concept operationally straightforward. Using this complexity measure, we are exploring the general evolution of the genitalia as a whole unit, assuming that all the structures that contribute to complexity could be affecting the fitness. Therefore, although counting the number of structures is a simple metric, it is the most appropriate, objective and efficient method to test hypotheses of complexity evolution quantitatively, given the nature of our data. Neomorphic characters, i.e. the ones that indicate the appearance or loss of a trait (Sereno, 2007), were used to count the presence of structures on male and female genitalia. Given that some transformational characters, which represent modification of an existent trait (Sereno, 2007), could also be useful for individualizing structural elements (Ramírez & Michalik, 2014), they were also used to calculate complexity. For example, some species have a very long and coiled embolus; others have an embolus that is approximately as long as the tegulum, whereas others have an embolus shorter than half the tegulum. A value of two, one and zero was given to those structures, respectively. Although the short embolus is coded as zero, the presence of an embolus is coded as one, therefore contributing to the overall complexity of the palp. Transformational characters representing shape of structures were not used. Taxa with > 15% of entries missing (for either male or female) were excluded from complexity analysis. For the remaining taxa, missing characters were conservatively treated as absent, because there were only a few missing entries and the inclusion of these taxa could be informative (although conservative) for understanding genital evolution. Only taxa with both male and female data available were used for complexity analyses. The final data set for testing genital evolution hypotheses and trends in complexity (see below) consists of 60 species of the Gnaphosidae s.s. clade (Supporting Information, Table S1). Testing trends in complexity Generalized estimating equations (GEEs) for comparative data were used to test for evolutionary trends in complexity. This method allows parameter estimation of a generalized linear model (GLM), taking into account the species autocorrelation, which is determined by a phylogenetic hypothesis, and there is no need to estimate ancestral states (Paradis & Claude, 2002). However, the method does not account for uncertainties of a phylogeny, and the tree used is assumed to represent the true relationships. To circumvent this limitation, two GEE analyses were performed using two different phylogenetic hypotheses considered by Azevedo et al. (2017): the Bayesian consensus tree (Azevedo et al., 2017: fig. 2) and the most parsimonious tree under implied weighting with a concavity value of 19 (Azevedo et al., 2017: fig. 3). The effects of taxa depth (distance to the root) on the complexity measurement were analysed to test models of complexity evolution. If there is an evolutionary trend of increasing complexity, a better fit of the data is expected to a normal linear regression model, with a positive relationship of taxa depth and complexity (positive regression coefficients). If the tendency is to simplify genitalia, the normal regression coefficients would be estimated to be negative. If there is no relationship, a reduced model with the regression coefficients fixed to zero would have the best fit, and/or the regression coefficient estimated on the full model would not be statistically different from zero. The model choice was based on the quasi-likelihood information criterion (QIC), a modification of the traditional Akaike information criterion (AIC) for GEE (Pan, 2001). Analyses were carried out using the APE package (Paradis, Claude & Strimmer, 2004) for R version 3.2.1 (R Core Development Team, 2015). Given that the parsimony tree used provides no information on branch lengths, the taxa depths were calculated as the number of nodes that separates the terminal from the root of the tree. This is the same as considering all branch lengths equal to unity, and is equivalent to assuming a punctuational view of evolution (Pagel, 1994, 1997, 1999). In this way, changes expected in a trait would be proportional to the number of speciation events (nodes) estimated from the root to the species, independent of the length of the branches. We believe this is a conservative approach, more appropriate when branch lengths are not available. Alternatively, the Bayesian consensus tree was also used in GEE analysis, because in the Bayesian inference, the branch lengths are estimated together with the topology. Using estimated branch lengths with no transformation is equivalent to assuming a gradualism model of evolution, in which branch lengths are a proxy of expected changes in a trait (Pagel, 1994, 1997, 1999). Analysing both phylogenetic hypotheses permits the exploration of models with different assumptions on trait evolution. Trends in complexity were also tested using generalized least squares (GLS) under a Bayesian approach (Pagel, 1997, 1999). An advantage of this method is that it allows the testing of models taking into account uncertanties about phylogeny and about parameter estimations. The GLS was performed with two sets of trees from Azevedo et al. (2017). One set consists of all trees (with branch lengths equal to one) obtained in parsimony analyses (Azevedo et al. 2017) under different concavity values (k = 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27), and the other consists of the trees (with estimated branch lengths) obtained in the stationary phase of the Markov chain of the Bayesian phylogenetic analysis by Azevedo et al. (2017). Bayes factors (Kass & Raftery, 1995) were used to choose between a model that estimates the correlation of the complexity and taxa depth (directional trend model) and a model in which this correlation is not assumed (traditional random walk model). The analyses were run in BayesTraits version 2 (Pagel & Meade, 2014) and consisted of 100000000 Markov chain Monte Carlo (MCMC) chains with a sample period of 10000 after a burn-in phase of 1000000 generations. All priors were set to uniform from −100 to 100. Stepping stones were used to estimate marginal likelihoods, using 1000 ‘stones’ with 100000 iterations each. Testing trend in embolus fusion Some Gnaphosidae have a tubular distal membrane that separates the terminal part from the tegulum, giving some movability to the embolus, whereas in others this membrane is absent, and the embolus and tegulum are fused into a single structure (Zakharov & Ovtcharenko, 2011, 2013). If there is a trend towards fusion, it is expected that the rate of loss of the distal tubular membrane (character 16) would be higher than the rate of gain in this character [i.e. the changes from state 0 (absence) to state 1 (presence) of the character would be less frequent than the changes from state 1 to state 0]. This hypothesis was tested with a Bayesian approach to fit continuous-time Markov models to discrete characters (Pagel, 1997, 1999) in BayesTraits, taking into account the uncertanties in phylogeny. Bayes factor was used to choose between a model that assumes that the transition rates of gain (q01) and loss (q10) are different and are estimated separately, and a model in which the rates are constrained to be equal. Analyses were done using two sets of trees (implied weighting and Bayesian trees) and consisted of 100000000 MCMC chains with a sample period of 10000 after a burn-in phase of 1000000 generations. All priors were set to uniform from zero to 100. Stepping stones were used to estimate marginal likelihoods, using 1000 ‘stones’ with 100000 iterations each. Testing genital evolution hypotheses Occasionally, it may be hard to distinguish between hypotheses of genital evolution, because they are not always mutually exclusive (Hosken & Stockley, 2004; Eberhard, 2010a, 2015). The sexually antagonistic coevolution (SAC) and lock-and-key (LK) hypotheses predict a correlation between the morphology of male and female genitalia, but a correlation may or may not be present if genitalia are evolving under the cryptic female choice (CFC) mechanism (Arnqvist, 1997; Eberhard, 2010b, 2015). If females are cryptically choosing males based mainly on stimuli that male genitalia produce regardless of female genital shape, the coevolution of genital morphology would not be expected under CFC. Therefore, although a significant correlation would not reject CFC, a lack of correlation of genital morphological complexity would favour CFC over SAC and LK. Generalized estimating equations for comparative data were used to test the correlation between female and male genital complexity using both the Bayesian consensus tree and the parsimony implied weighting (k = 19) tree as phylogenetic hypotheses. A model that estimates the coefficient (slope) of a normal regression (assuming correlation) was compared with one in which the coefficient is zero (i.e. a model with no correlation). Model choice was based on the QIC. If the best model included the estimation of the coefficient, it was also checked to determine whether the value was significantly different from zero. Analyses were carried out using the APE package. Generalized least squares was also used to test the correlation between male and female complexity in BayesTraits using all the implied weighting trees and the Bayesian stationary MCMC trees. Bayes factors were used to check whether a model that assumes the correlation fits the data better than a model in which there is no correlation between traits. Parameters were the same as for the trends in complexity analyses. If morphological changes that stimulate the female better are undergoing strong positive selective pressure by CFC (Eberhard, 2010b), one could expect that the male copulatory organs could have a faster evolutionary rate of morphological changes than female genitalia. An equal rate of evolution of male and female genitalia would not necessarily reject CFC, because the selection could be acting on the male behaviour. Nevertheless, our study focuses on the processes that moulded diversity of genitalia, and the lack of difference in the rate of evolution could indicate that the CFC has not acted strongly on the complexity of genitalia. To test the difference in evolutionary rate, the variance of evolutionary change (rate of change; sigmaparameter of the Brownian motion; Pagel, 1997, 1999) of male and female complexity was estimated using the GLS model through Bayesian methods in BayesTraits, with the same parameters mentioned above. The female and male mean rates estimated on the stationary phase of the MCMC (9910 samples) were compared with a t-test with a significance level of 0.05. RESULTS Ancestral state reconstruction and character evolution The reconstruction of ancestral character states suggests that the male palp of the most recent Gnaphosidae s.s. ancestor would have had the following structures: petiole (character 34), subtegulum, tegulum, embolus fused to tegulum (character 15), embolus about the same size or smaller (ambiguous) than the copulatory bulb (character 21), conductor (character 14), a median apophysis (character 27), which is shaped like a hook (character 30), and a retrolateral tibial apophysis (character 51). These traits do not necessarily constitute synapomorphies; they are merely the conditions found on the common ancestor of the clade (see character optimization in Supporting Information, Appendix S2. The conductor was lost unambiguously six times and regained three times during the evolution of Gnaphosidae s.s. (Fig. 1). An articulation between the embolus and tegulum (the distal tubular membrane) arose four times and was lost again at least five times (Fig. 2). A long (longer than tegulum), coiled embolus appears eight times and is reversed two times (on one occasion it is reversed to an embolus as long as the tegulum and on the other it is reduced to about half the length of the tegulum; Fig. 3). The median apophysis was lost seven times, with no regain (Fig. 4). The terminal apophysis (a sclerite on the terminal division of the bulb, proximally and ventrally positioned in relationship to the embolus) arises six times on gnaphosids (Fig. 5). Figure 1. View largeDownload slide A, the evolution of the conductor in Gnaphosidae s.s. B, C, scanning electron micrographs of the male palp, in ventral view, of Sosticus insularis (Banks, 1895) (B) and Drassodes saccatus (Emerton, 1890) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: c, conductor; e, embolus; ma, median apophysis. Scale bar: 100 μm. Figure 1. View largeDownload slide A, the evolution of the conductor in Gnaphosidae s.s. B, C, scanning electron micrographs of the male palp, in ventral view, of Sosticus insularis (Banks, 1895) (B) and Drassodes saccatus (Emerton, 1890) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: c, conductor; e, embolus; ma, median apophysis. Scale bar: 100 μm. Figure 2. View largeDownload slide A, the evolution of the distal tubular membrane in Gnaphosidae s.s. B, C, scanning electron micrographs of the male palp of Litopyllus temporarius Chamberlin, 1922 in ventral view (B) and Apopyllus silvestrii (Simon, 1905) in retrolateral view (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: dm, distal tubular membrane; e, embolus; t, tegulum. Scale bar: 100 μm. Figure 2. View largeDownload slide A, the evolution of the distal tubular membrane in Gnaphosidae s.s. B, C, scanning electron micrographs of the male palp of Litopyllus temporarius Chamberlin, 1922 in ventral view (B) and Apopyllus silvestrii (Simon, 1905) in retrolateral view (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: dm, distal tubular membrane; e, embolus; t, tegulum. Scale bar: 100 μm. Figure 3. View largeDownload slide A, the evolution of embolus length in Gnaphosidae s.s. B–D, scanning electron micrographs of the male palp, in ventral view, of Hypodrassodes maoricus (Dalmas, 1917) (B), Cryptoerithus occultus Rainbow, 1915 (C) and Apopyllus silvestrii (Simon, 1905) (D), exemplifying the states of the character. Only unambiguous changes are shown. Abbreviation: e, embolus. Scale bar: 100 μm. Figure 3. View largeDownload slide A, the evolution of embolus length in Gnaphosidae s.s. B–D, scanning electron micrographs of the male palp, in ventral view, of Hypodrassodes maoricus (Dalmas, 1917) (B), Cryptoerithus occultus Rainbow, 1915 (C) and Apopyllus silvestrii (Simon, 1905) (D), exemplifying the states of the character. Only unambiguous changes are shown. Abbreviation: e, embolus. Scale bar: 100 μm. Figure 4. View largeDownload slide A, the evolution of the median apophysis in Gnaphosidae s.s. B, C, scanning electron micrographs of the male palp, in ventral view, of Orodrassus coloradensis (Emerton, 1877) (B) and Haplodrassus hiemalis (Emerton, 1909) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: c, conductor; e, embolus; ma, median apophysis; ta, terminal apophysis. Scale bar: 100 μm. Figure 4. View largeDownload slide A, the evolution of the median apophysis in Gnaphosidae s.s. B, C, scanning electron micrographs of the male palp, in ventral view, of Orodrassus coloradensis (Emerton, 1877) (B) and Haplodrassus hiemalis (Emerton, 1909) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: c, conductor; e, embolus; ma, median apophysis; ta, terminal apophysis. Scale bar: 100 μm. Figure 5. View largeDownload slide A, the evolution of the terminal apophysis in Gnaphosidae s.s. B, C, scanning electron micrographs of the male palp, in ventral view, of Lygromma chamberlini Gertsch, 1941 (B) and Camillina cordifera (Tullgren, 1910) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: c, conductor; e, embolus; ma, median apophysis; ta, terminal apophysis. Scale bar: 50 μm. Figure 5. View largeDownload slide A, the evolution of the terminal apophysis in Gnaphosidae s.s. B, C, scanning electron micrographs of the male palp, in ventral view, of Lygromma chamberlini Gertsch, 1941 (B) and Camillina cordifera (Tullgren, 1910) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: c, conductor; e, embolus; ma, median apophysis; ta, terminal apophysis. Scale bar: 50 μm. The ancestral gnaphosid epigynum would have had an anterior fold (character 0), lateral folds (character 4), which are composed of furrows (character 5), primary spermathecae (character 9), and secondary spermathecae (character 10) that have a defined lumen (character 11), are smaller than the primary spermathecae (character 12) and have short stalk (character 12). The anterior fold was lost nine times in Gnaphosidae s.s. (Fig. 6). The furrowed lateral folds became a suture ten times (or seven times with three reversions; Fig. 7). The primary spermathecae were lost three times with one regain (Fig. 8). The secondary spermathecae disappeared three times in the evolution of Gnaphosidae s.s. (Fig. 9). Figure 6. View largeDownload slide A, the evolution of the anterior fold of the epigynum in Gnaphosidae s.s. B, C, scanning electron micrographs of the epigynum, in ventral view, of Litopyllus temporarius Chamberlin, 1922 (B) and Eilica bicolor Banks, 1896 (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: af, anterior fold; co, copulatory openings; lf, lateral fold. Scale bar: 40 μm. Figure 6. View largeDownload slide A, the evolution of the anterior fold of the epigynum in Gnaphosidae s.s. B, C, scanning electron micrographs of the epigynum, in ventral view, of Litopyllus temporarius Chamberlin, 1922 (B) and Eilica bicolor Banks, 1896 (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: af, anterior fold; co, copulatory openings; lf, lateral fold. Scale bar: 40 μm. Figure 7. View largeDownload slide A, the evolution of the lateral fold of the epigynum in Gnaphosidae s.s. B–D, scanning electron micrographs of the epigynum, in ventral view, of Chilongius palmas Platnick, Shadab & Sorkin, 2005 (B), Odontodrassus aphanes (Thorell, 1897) (C) and Zelanda erebus (L. Koch, 1873) (D), exemplifying the absence, the presence as a furrow and the presence as a suture, respectively. Only unambiguous changes are shown. Abbreviations: af, anterior fold; co, copulatory openings; lff, lateral fold as a furrow; lfs, lateral fold as a suture; s, scape. Scale bar: 40 μm. Figure 7. View largeDownload slide A, the evolution of the lateral fold of the epigynum in Gnaphosidae s.s. B–D, scanning electron micrographs of the epigynum, in ventral view, of Chilongius palmas Platnick, Shadab & Sorkin, 2005 (B), Odontodrassus aphanes (Thorell, 1897) (C) and Zelanda erebus (L. Koch, 1873) (D), exemplifying the absence, the presence as a furrow and the presence as a suture, respectively. Only unambiguous changes are shown. Abbreviations: af, anterior fold; co, copulatory openings; lff, lateral fold as a furrow; lfs, lateral fold as a suture; s, scape. Scale bar: 40 μm. Figure 8. View largeDownload slide A, the evolution of the primary spermathecae in Gnaphosidae s.s. B, C, scanning electron micrographs of the vulva, in dorsal view, of Lygromma chamberlini Gertsch, 1941 (B) and Setaphis subtilis (Simon, 1897) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: cd, copulatory ducts; ps, primary spermathecae; ss, secondary spermathecae. Scale bar: 40 μm. Figure 8. View largeDownload slide A, the evolution of the primary spermathecae in Gnaphosidae s.s. B, C, scanning electron micrographs of the vulva, in dorsal view, of Lygromma chamberlini Gertsch, 1941 (B) and Setaphis subtilis (Simon, 1897) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: cd, copulatory ducts; ps, primary spermathecae; ss, secondary spermathecae. Scale bar: 40 μm. Figure 9. View largeDownload slide A, the evolution of the secondary spermathecae in Gnaphosidae s.s. B, C, scanning electron micrographs of the vulva, in dorsal view, of Sosticus insularis (Banks, 1895) (B) and Zelanda erebus (L. Koch, 1873) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: cd, copulatory ducts; ps, primary spermathecae; ss, secondary spermathecae. Scale bar: 40 μm. Figure 9. View largeDownload slide A, the evolution of the secondary spermathecae in Gnaphosidae s.s. B, C, scanning electron micrographs of the vulva, in dorsal view, of Sosticus insularis (Banks, 1895) (B) and Zelanda erebus (L. Koch, 1873) (C), exemplifying the absence and presence of the character, respectively. Only unambiguous changes are shown. Abbreviations: cd, copulatory ducts; ps, primary spermathecae; ss, secondary spermathecae. Scale bar: 40 μm. Complexity evolution and trends The palp complexity index in Gnaphosidae s.s. ranges from 2.0, in Gertschosa Platnick & Shadab, 1981, to 12.0, in Zelanda Özdikmen, 2009 (Supporting Information, Table S1). The mean palp complexity index of the data set is 6.9, and the ancestral palp complexity in Gnaphosidae s.s. is 6.4. Therefore, an intermediate complexity is the ancestral condition, from which more complex and simpler palps evolved (Fig. 10). The epigynum complexity index in Gnaphosidae s.s. ranges from 2.0, in Chilongius Platnick, Shadab & Sorkin, 2005 and Zelotibia Russell-Smith & Murphy, 2005, to 10.0, in Zimiromus Banks, 1914 (Supporting Information, Table S1). The mean epigynum complexity index of the data set is 6.1, and the ancestral epigynum complexity in Gnaphosidae s.s. is 6.8. As in males, an epigynum with intermediate complexity seems to be the ancestral condition (Fig. 10). The palpal complexity was not correlated with the taxa depth in any of the analyses used (Tables 1, 2). Therefore, there is no evidence of trends towards either increasing or decreasing genitalic complexity in Gnaphosidae s.s. Table 1. Results of the generalized estimating equations of the test of trends in the palp complexity using the implied weighting and Bayesian consensus trees Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No trend 363.08 7.71 3.27 × 10−6 – – Trend 372.43 6.16 0.001 0.18 0.15 Bayesian consensus tree No trend 375.73 8.14 0.00 – – Trend 382.08 6.12 0.00 0.21 0.04 Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No trend 363.08 7.71 3.27 × 10−6 – – Trend 372.43 6.16 0.001 0.18 0.15 Bayesian consensus tree No trend 375.73 8.14 0.00 – – Trend 382.08 6.12 0.00 0.21 0.04 The ‘no trend’ model assumes no relationship between palp complexity and taxa depth (regression coefficient fixed to zero). In the ‘trend’ model, the intercept and coefficient of the linear regression between these two variables are estimated. Abbreviation: QIC, quasi-likelihood information criterion of the model. Parameters with P-values < 0.05 were considered significantly different from zero. View Large Table 1. Results of the generalized estimating equations of the test of trends in the palp complexity using the implied weighting and Bayesian consensus trees Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No trend 363.08 7.71 3.27 × 10−6 – – Trend 372.43 6.16 0.001 0.18 0.15 Bayesian consensus tree No trend 375.73 8.14 0.00 – – Trend 382.08 6.12 0.00 0.21 0.04 Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No trend 363.08 7.71 3.27 × 10−6 – – Trend 372.43 6.16 0.001 0.18 0.15 Bayesian consensus tree No trend 375.73 8.14 0.00 – – Trend 382.08 6.12 0.00 0.21 0.04 The ‘no trend’ model assumes no relationship between palp complexity and taxa depth (regression coefficient fixed to zero). In the ‘trend’ model, the intercept and coefficient of the linear regression between these two variables are estimated. Abbreviation: QIC, quasi-likelihood information criterion of the model. Parameters with P-values < 0.05 were considered significantly different from zero. View Large Table 2. Log-likelihood and Bayes factor of models to test trends in palp complexity using the Bayesian approach of the generalized least square method with all implied weighting and all Bayesian stationary chain trees Model Log-likelihood Implied weighting trees No trend (M0) −126.78 Trend (M1) −132.11 Bayes factor 10.66 Bayesian stationary chain trees No trend (M0) −123.84 Trend (M1) −125.62 Bayes factor 3.55 Model Log-likelihood Implied weighting trees No trend (M0) −126.78 Trend (M1) −132.11 Bayes factor 10.66 Bayesian stationary chain trees No trend (M0) −123.84 Trend (M1) −125.62 Bayes factor 3.55 Bayes factors > 2.0 were considered positive evidence against M1. View Large Table 2. Log-likelihood and Bayes factor of models to test trends in palp complexity using the Bayesian approach of the generalized least square method with all implied weighting and all Bayesian stationary chain trees Model Log-likelihood Implied weighting trees No trend (M0) −126.78 Trend (M1) −132.11 Bayes factor 10.66 Bayesian stationary chain trees No trend (M0) −123.84 Trend (M1) −125.62 Bayes factor 3.55 Model Log-likelihood Implied weighting trees No trend (M0) −126.78 Trend (M1) −132.11 Bayes factor 10.66 Bayesian stationary chain trees No trend (M0) −123.84 Trend (M1) −125.62 Bayes factor 3.55 Bayes factors > 2.0 were considered positive evidence against M1. View Large Figure 10. View largeDownload slide Optimization of palp (A) and epigynum (B) complexity values on the phylogenetic hypothesis of Gnaphosidae s.s. Figure 10. View largeDownload slide Optimization of palp (A) and epigynum (B) complexity values on the phylogenetic hypothesis of Gnaphosidae s.s. The GEE of the epigynum trend in complexity revealed that the best models suggested by QIC are the ones that estimate the regression coefficients either using the implied weighting or the Bayesian consensus trees. However, in both analyses the regression coefficients estimated are very small and not significantly different from zero (Table 3). Regarding the GLS using the Bayesian approach, both tests using the different sets of trees showed support for the model of traditional random walk with no trend in complexity (Table 4). Therefore, there is not enough evidence in the data to support either an increasing or a decreasing genital complexity in Gnaphosidae s.s. Table 3. Results of the generalized estimating equations of the test of trends in epigynum complexity using the implied weighting with k = 19 and the Bayesian consensus trees Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No trend 163.2 6.27 2.25 × 10−7 – – Trend 161.6 5.7 3.79 × 10−5 0.07 3.91 × 10−1 Bayesian consensus tree No trend 165.36 6.62 5.61 × 10−12 – – Trend 161.08 5.91 1.12 × 10−7 0.07 2.41 × 10−1 Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No trend 163.2 6.27 2.25 × 10−7 – – Trend 161.6 5.7 3.79 × 10−5 0.07 3.91 × 10−1 Bayesian consensus tree No trend 165.36 6.62 5.61 × 10−12 – – Trend 161.08 5.91 1.12 × 10−7 0.07 2.41 × 10−1 ‘No trend’ model assumes no relationship between palp complexity and taxa depth (regression coefficient fixed to zero). In the ‘trend’ model, the intercept and coefficient of the linear regression between these two variables are estimated. Abbreviation: QIC, quasi-likelihood information criterion of the model. Parameters with P-values < 0.05 were considered significantly different from zero. View Large Table 3. Results of the generalized estimating equations of the test of trends in epigynum complexity using the implied weighting with k = 19 and the Bayesian consensus trees Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No trend 163.2 6.27 2.25 × 10−7 – – Trend 161.6 5.7 3.79 × 10−5 0.07 3.91 × 10−1 Bayesian consensus tree No trend 165.36 6.62 5.61 × 10−12 – – Trend 161.08 5.91 1.12 × 10−7 0.07 2.41 × 10−1 Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No trend 163.2 6.27 2.25 × 10−7 – – Trend 161.6 5.7 3.79 × 10−5 0.07 3.91 × 10−1 Bayesian consensus tree No trend 165.36 6.62 5.61 × 10−12 – – Trend 161.08 5.91 1.12 × 10−7 0.07 2.41 × 10−1 ‘No trend’ model assumes no relationship between palp complexity and taxa depth (regression coefficient fixed to zero). In the ‘trend’ model, the intercept and coefficient of the linear regression between these two variables are estimated. Abbreviation: QIC, quasi-likelihood information criterion of the model. Parameters with P-values < 0.05 were considered significantly different from zero. View Large Table 4. Log-likelihood and Bayes factor of models to test trends in epigynum complexity using the Bayesian approach of the generalized least square method with all implied weightings and all Bayesian stationary chain trees Model Log-likelihood Implied weighting trees No trend (M0) −124.51 Trend (M1) −130.56 Bayes factor 12.1 Bayesian stationary chain trees No trend (M0) −119.95 Trend (M1) −123.03 Bayes factor 6.16 Model Log-likelihood Implied weighting trees No trend (M0) −124.51 Trend (M1) −130.56 Bayes factor 12.1 Bayesian stationary chain trees No trend (M0) −119.95 Trend (M1) −123.03 Bayes factor 6.16 Bayes factors > 2.0 were considered positive evidence against M1. View Large Table 4. Log-likelihood and Bayes factor of models to test trends in epigynum complexity using the Bayesian approach of the generalized least square method with all implied weightings and all Bayesian stationary chain trees Model Log-likelihood Implied weighting trees No trend (M0) −124.51 Trend (M1) −130.56 Bayes factor 12.1 Bayesian stationary chain trees No trend (M0) −119.95 Trend (M1) −123.03 Bayes factor 6.16 Model Log-likelihood Implied weighting trees No trend (M0) −124.51 Trend (M1) −130.56 Bayes factor 12.1 Bayesian stationary chain trees No trend (M0) −119.95 Trend (M1) −123.03 Bayes factor 6.16 Bayes factors > 2.0 were considered positive evidence against M1. View Large Embolus fusion evolution The results of the estimated rates of change in discrete character states suggest that the best model supported by the data is the one with equal rates of gain and loss of the terminal tubular membrane, using both sets of trees (implied weighting trees and stationary Bayesian trees; Table 5). It is possible to conclude, therefore, that there is no trend towards fusion of the embolus with the tegulum, nor to detachment of these structures. Table 5. Log-likelihood and Bayes factor of models to test trends in embolus fusion with tegulum using the Bayesian approach of the generalized least squares method with all implied weighting and all Bayesian stationary chain trees Model Log-likelihood Implied weighting trees Equal rates model (M0) −48.91 Different rates model (M1) −53.74 Bayes factor 9.67 Bayesian stationary chain trees Equal rates model (M0) −46.59 Different rates model (M1) −49.82 Bayes factor 6.46 Model Log-likelihood Implied weighting trees Equal rates model (M0) −48.91 Different rates model (M1) −53.74 Bayes factor 9.67 Bayesian stationary chain trees Equal rates model (M0) −46.59 Different rates model (M1) −49.82 Bayes factor 6.46 In the equal rates model, the rate of fusion is forced to be equal to the rate of detachment. The model with different rates suggests a trend. Bayes factors > 2.0 were considered positive evidence against M1. View Large Table 5. Log-likelihood and Bayes factor of models to test trends in embolus fusion with tegulum using the Bayesian approach of the generalized least squares method with all implied weighting and all Bayesian stationary chain trees Model Log-likelihood Implied weighting trees Equal rates model (M0) −48.91 Different rates model (M1) −53.74 Bayes factor 9.67 Bayesian stationary chain trees Equal rates model (M0) −46.59 Different rates model (M1) −49.82 Bayes factor 6.46 Model Log-likelihood Implied weighting trees Equal rates model (M0) −48.91 Different rates model (M1) −53.74 Bayes factor 9.67 Bayesian stationary chain trees Equal rates model (M0) −46.59 Different rates model (M1) −49.82 Bayes factor 6.46 In the equal rates model, the rate of fusion is forced to be equal to the rate of detachment. The model with different rates suggests a trend. Bayes factors > 2.0 were considered positive evidence against M1. View Large Genital evolution hypothesis The GEE of the relationship between palp and epigynum complexity, using the implied weighting tree as the hypothesis, suggests that the best model is the one that estimates the regression coefficient. However, this coefficient is very small and not significantly different from zero, showing no support for the correlation between these two traits (Table 6). The GEE using the Bayesian consensus tree revealed that the best model is the one in which the regression coefficient between complexity measures is fixed at zero, showing no correlation between traits (Table 6). All analyses using the Bayesian approach of GLS suggest that the best model is the one with no correlation between palp and epigynum complexity (Table 7). The Brownian variance parameter estimated in BayesTraits using MCMC shows a significant difference in the evolutionary rates of palp and epigynum, either using the implied weighting set of trees [female mean rate (SD) = 1.20 (0.03), male mean rate (SD) = 1.26 (0.04), P < 0.0001, t = 128.2, d.f. = 19818] or the Bayesian stationary phase set of trees [female mean rate (SD) = 20.96 (1.99), male mean rate (SD) = 24.18 (2.27), P < 0.0001, t = 105.9, d.f. = 19818]. The results indicate that the male gnaphosid palp evolved ~1.05 times faster than the epigynum for the analysis using implied weighting trees and 1.15 times faster for the analyses using the Bayesian set of trees. Therefore, the data analysed here support the CFC hypothesis of genital evolution. Table 6. Results of the generalized estimating equations of the test of correlation between palp and epigynum complexity using the implied weighting with k = 19 and the Bayesian consensus trees Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No relationship 363.08 7.71 3.27 × 10−6 – – Linear 362.02 7.5 3.51 × 10−5 0.03 7.35 × 10−1 Bayesian consensus tree No relationship 375.72 8.14 8.53 × 10−10 – – Linear 385.03 8.68 5.64 × 10−7 0.08 5.80 × 10−1 Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No relationship 363.08 7.71 3.27 × 10−6 – – Linear 362.02 7.5 3.51 × 10−5 0.03 7.35 × 10−1 Bayesian consensus tree No relationship 375.72 8.14 8.53 × 10−10 – – Linear 385.03 8.68 5.64 × 10−7 0.08 5.80 × 10−1 In the ‘no relationship’ model, the regression coefficient is fixed to zero. In the ‘linear’ model, the intercept and coefficient of the linear regression between these two variables are estimated. Abbreviation: QIC, quasi-likelihood information criterion of the model. Parameters with P-values < 0.05 were considered significantly different from zero. View Large Table 6. Results of the generalized estimating equations of the test of correlation between palp and epigynum complexity using the implied weighting with k = 19 and the Bayesian consensus trees Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No relationship 363.08 7.71 3.27 × 10−6 – – Linear 362.02 7.5 3.51 × 10−5 0.03 7.35 × 10−1 Bayesian consensus tree No relationship 375.72 8.14 8.53 × 10−10 – – Linear 385.03 8.68 5.64 × 10−7 0.08 5.80 × 10−1 Model QIC Intercept P-intercept Coefficient P-coefficient Implied weighting tree No relationship 363.08 7.71 3.27 × 10−6 – – Linear 362.02 7.5 3.51 × 10−5 0.03 7.35 × 10−1 Bayesian consensus tree No relationship 375.72 8.14 8.53 × 10−10 – – Linear 385.03 8.68 5.64 × 10−7 0.08 5.80 × 10−1 In the ‘no relationship’ model, the regression coefficient is fixed to zero. In the ‘linear’ model, the intercept and coefficient of the linear regression between these two variables are estimated. Abbreviation: QIC, quasi-likelihood information criterion of the model. Parameters with P-values < 0.05 were considered significantly different from zero. View Large Table 7. Log-likelihood and Bayes factor of models to test correlation between epigynum and palp complexity using the Bayesian approach of the generalized least square method with all implied weighting and all Bayesian stationary chain trees Model Log-likelihood Implied weighting trees Correlation (M0) −251.6 No correlation (M1) −251.62 Bayes factor 0.02 Bayesian stationary chain trees Correlation (M0) −243.83 No correlation (M1) −244.04 Bayes factor 0.41 Model Log-likelihood Implied weighting trees Correlation (M0) −251.6 No correlation (M1) −251.62 Bayes factor 0.02 Bayesian stationary chain trees Correlation (M0) −243.83 No correlation (M1) −244.04 Bayes factor 0.41 Bayes factors > 2.0 were considered positive evidence against M1. View Large Table 7. Log-likelihood and Bayes factor of models to test correlation between epigynum and palp complexity using the Bayesian approach of the generalized least square method with all implied weighting and all Bayesian stationary chain trees Model Log-likelihood Implied weighting trees Correlation (M0) −251.6 No correlation (M1) −251.62 Bayes factor 0.02 Bayesian stationary chain trees Correlation (M0) −243.83 No correlation (M1) −244.04 Bayes factor 0.41 Model Log-likelihood Implied weighting trees Correlation (M0) −251.6 No correlation (M1) −251.62 Bayes factor 0.02 Bayesian stationary chain trees Correlation (M0) −243.83 No correlation (M1) −244.04 Bayes factor 0.41 Bayes factors > 2.0 were considered positive evidence against M1. View Large DISCUSSION Genitalia structure homology Homology between male palp structures may be hard to establish, especially when the number of structures varies among species being analysed (Ramírez, 2007). It is not always clear which genital structures are homologous among spiders, and more comparative morphological and phylogenetic studies are necessary to unravel this issue at a broad scale (Coddington, 1990; Sierwald, 1990). Also, the supposed rapid evolution of genitalia may have caused several gains and losses. Some important studies on gnaphosid spiders have contributed to hypotheses of primary homology among palp characters in the family (Senglet, 2004; Zakharov & Ovtcharenko, 2011, 2013). However, testing the congruence of these characters against others is an important step in the process of establishing secondary homology (de Pinna, 1991; Richter, 2005). Here, the results of a previous phylogenetic analysis (Azevedo et al., 2017), putting to test primary homology propositions, were used to explore the evolution of some genitalic characters to help with understanding their homology in Gnaphosidae. The median apophysis, a projection that arises from the membranous median area of the tegulum (Azevedo et al., 2017), is homologous among all Gnaphosidae and probably originated more deeply in spider phylogeny. This data set suggests that it would be present in the common ancestor of the more inclusive Oblique Median Tapetum clade (Ramírez, 2014). However, although this structure was lost several times in Gnaphosidae with no regains, which leaves little doubt about its homology, there was one reappearance of the median apophysis after a loss in the outgroup. Therefore, broad taxonomic sampling is needed to understand median apophysis homology in the deep and broad context of spider evolution. The conductor, an outgrowth of the proximal part of the tegular wall associated with the embolus tip (Zakharov & Ovtcharenko, 2011), is absent in about half of the gnaphosid genera. In many cases, the supposed function of this structure (to help conduct the embolus during copulation) may be taken over by other parts of the palp. In some Zelotinae (Gnaphosidae), for example, the embolus passes on the back side of the median apophysis during the intromission phase (Senglet, 2004). In some taxa, such as Leptodrassinae, a secondary accessory median apophysis surrounds the embolus and could act as a conductor (Azevedo et al., 2017). Although the conductor is present in the most recent common ancestor of Gnaphosidae s.s., its homology throughout the family is not supported, because several independent reappearances suggest convergence in many clades. A tripartite bulb is supposed to be the plesiomorphic condition for spiders and for Gnaphosidae (Platnick & Gertsch, 1976; Kraus, 1978; Coddington, 1990; Sierwald, 1990; Zakharov & Ovtcharenko, 2011, 2013). However, our results suggest that a palpal bulb with three divisions is a derived condition within Gnaphosidae s.s. The ancestral gnaphosid probably had no tubular distal membrane separating the embolus from the tegulum, resulting in a bipartite bulb. In fact, the results suggest that the fusion of tegulum and embolus might have arisen before the origin of Gnaphosidae s.s. The membrane that separates the terminal and median bulb divisions appears four times in the family, and a return to the ancestral condition (fused) occurs five times. The rate of gain of the tubular membrane was estimated to be the same as the rate of loss of this structure. Therefore, there is no evidence to support the hypothesis of a trend to fusion of the terminal and median divisions of the embolus. Detailed studies on the comparative morphology and evolution of female genitalia are rare in spiders, but it has been suggested that the primary spermatheca are homologous among all entelegyne spiders (Forster, 1980; Sierwald, 1989; ‘BS’). Here, the primary (Fig. 8; character 111, expansion of the spermathecal lumen) and the secondary (Fig. 9; character 114; expansion of a pore-plate; Sierwald, 1989; ‘HS’) spermathecae were found to be homologous among all gnaphosids and evolved long before the most recent common ancestor of the family, which may support homology throughout Entelegynae. The primary spermathecae have been lost at least three times in Gnaphosidae s.s., becoming an undifferentiated part, usually less sclerotized, of the terminal tract of the copulatory duct. The lateral folds are also plesiomorphic for gnaphosids and probably arose long before the family’s ancestor. However, these structures may have been completely lost or can be present as sutures. It has been shown that precursors of the lateral folds are present in the early stage of epigynum ontogeny and that they invaginate to form the internal ducts (Sierwald, 1989). Given that the plesiomorphic feature for the family is the lateral fold present as furrows in the epigynum, the apomorphic conditions of lateral folds as sutures or complete absence of the structure may be regarded as paedomorphic conditions. In those gnaphosid taxa with an epigynal suture, the formation of the internal genitalia and sexual organs may have reached maturation before the complete development of the precursor of lateral folds. An ontogenetic study could help to elucidate this hypothesis of a gnaphosid paedomorphic epigynum, as suggested by the phylogeny. However, the understanding of lateral fold evolution among spiders in general will require much study effort on the distribution of these characters through spider phylogeny. Complexity The putative plesiomorphic condition of a tripartite palp in Araneae, with many taxa showing an apomorphic fusion of sclerites, may support the hypothesis of a trend towards simplification of male genitalia (Platnick & Gertsch, 1976; Kraus, 1978; Coddington, 1990; Sierwald, 1990). In Gnaphosidae, this simplification trend was also proposed as an explanation for the presence of genera with a bipartite palp in the family (Zakharov & Ovtcharenko, 2013). Here, we show that the ancestral condition is, in fact, a palp with the embolus fused to the tegulum, which suggests complex palps originating from more simple ones. However, as mentioned above, an articulated palp appeared and disappeared a few times, with equal transition rates, without a trend. Besides, the complexity is not measured only by the fusion of two sclerites, but by the number of structures in the male palp. Some genera with a non-flexible embolus can have other sclerites, increasing complexity. The complexity measure shows no evolutionary trend for male genitalia. The plesiomorphic condition in Gnaphosidae s.s. is a palp with moderate complexity giving rise to both complex and simple palps. The complexity would be evolving according to Brownian motion, in which the trait would evolve independently in each branch of a cladogenesis event, in a random direction towards simplification or increasing complexity, leaving no macroevolutionary trend. Regarding the females, the plesiomorphic condition of spider genitalia with few structures, with possible intermediate conditions in some clades (Mygalomorphae, Synspermiata and Leptonetidae), and the presence of many taxa with genitalia bearing many structures, could give rise to the idea of increasing complexity (Forster, 1980). However, the same pattern that was observed for males was also found for complexity evolution of the epigynum, with plesiomorphic moderate complexity evolving in Brownian motion, with no trends. The method applied here uses a simple complexity measure and includes no information on the absolute time involved in Gnaphosidae s.s. diversification. Some modern methods allow the use of techniques of geometric morphometrics to calculate a more elaborate measure of complexity or to quantify other morphological aspects, such as size, shape and volume (Rowe & Arnqvist, 2012). However, the techniques of geometric morphometrics may be difficult to apply to a large gnaphosid genitalia data set because of: (1) the great variety of morphologies, resulting in no clearly correspondent structures and, consequently, precluding landmark establishment; and (2) the tridimensionality of structures with few landmarks, requiring surface analysis methods of three-dimensional images (Adams, Rohlf & Slice, 2004; Zelditch et al., 2004). Finally, dated phylogenies would permit incorporation of other models of trait evolution, including the variation in rate among branches (Baker et al., 2015). An analysis of molecular data would provide another perspective on Gnaphosidae evolution and a test of the conclusions presented here. However, a good sample of gnaphosid genera for molecular analysis would require a worldwide effort to obtain fresh material, because the family is globally distributed, or an effort to sample old museum specimens. Until those data are available, the analysis presented here (based upon morphological data and using a simple complexity measure) is a good test and overview of the macroevolutionary evolution of complexity in genitalia of Gnaphosidae. Sexual selection Much evidence suggests sexual evolution by cryptic female choice as the most likely general mechanism generating genital diversity in spiders. Some morphological and sexual behaviour studies on species from five spider families show that females may be able to control copulation and manipulate sperm storage (Huber, 1999; Burger, Nentwig & Kropf, 2003; Burger et al., 2006; Welke & Schneider, 2009; Aisenberg, Barrantes & Eberhard, 2015; Calbacho-Rosa & Peretti, 2015; Schneider, Uhl & Herberstein, 2015). Additionally, Eberhard (2004b) found no support for coevolution of male structures and respective contact area in female genitalia among several arthropod groups (including spiders), favouring CFC over SAC. However, a correlation between male and female genital complexity was demonstrated for nephilid spiders, suggesting the occurrence of SAC, possibly together with CFC (Kuntner et al., 2009). The analysis presented here, using a simple measure of complexity, shows no support for SAC as the main mechanism generating genital diversity at a macroevolutionary scale in the ground spiders. In Gnaphosidae s.s., the female may be cryptically choosing male genitalia that best stimulate her sensory system during copulation, regardless (or with little influence) of the palpal complexity and mechanical fit to the epigynum. As there is no trend in complexity, there is no general best condition of male genitalia, and the female choice between simple or complex palps would be idiosyncratic of each species and not predictable by epigynum complexity. It has been argued that the bias towards the study of male genitalia could hamper the understanding of genital evolution (Ah-King, Barron & Herberstein, 2014). Indeed, morphological and comparative studies of female genitalia are scarcer for spiders. Detailed studies of haplogyne spiders with simple external genitalia revealed that the internal structure may be complex (Burger et al., 2003, 2006). However, the internal structures found in those studies of haplogyne female genitalia are related to sperm control, once again suggesting female choice. There is little evidence for sperm competition (SC) in spiders (Schneider et al., 2000). For instance, some Pholcidae use palpal structures to remove the sperm of rival males present in female genitalia (Calbacho-Rosa et al., 2013; Calbacho-Rosa & Peretti, 2015). The presence of genital plugs may also indicate sperm competition (Uhl, Nessler & Schneider, 2010; Eberhard, 2015). Genital plugs have been reported for gnaphosid species, but the frequency of their use, how they are formed, and the ability of males to remove plugs is barely known. It was shown that the removal of a genital plug in Micaria sociabilis Kulczyński, 1897 is rare and that the plug may be an effective way of preventing sperm competition (Sentenská et al., 2015). In some zelotines, there is a hectic brushing with the tibial apophysis on the epigynum during copulation (Senglet, 2004). Although this movement could help to remove the genital plug, the purpose and consequences of this behaviour are unknown. Evidence of sperm competition, nevertheless, may not exclude the occurrence of CFC, and females may have some influence on sperm competition (Eberhard, 2015). In Micaria sociabilis, males that couple genitalia faster are allowed by females to copulate longer and to complete genital plug production (Sentenská et al., 2015). The hypotheses of sexual selection are not mutually exclusive, and different processes may together affect the evolution of genital complexity (Simmons, 2014). It has been shown that different parts, or even a single structure, in the genitalia of insects may be under different selective pressures and could evolve through distinct mechanisms of sexual selection (Rowe & Arnqvist, 2012; Frazee & Masly, 2015). Some gnaphosid species have a very long, coiled embolus, which seems to be, at least in part, correlated with long, coiled copulatory ducts in the female. These two structures may have evolved in an ‘arms race’ in order to control copulation, whereas other structures in the genitalia may have evolved under selection by cryptic female choice. The fact that the ecological environment may influence the context of sexual selection also illustrates how complex the processes acting on genital diversity can be (Anderson & Langerhans, 2015). Here, evidence is provided for CFC acting on the genitalia of Gnaphosidae s.s. at a macroevolutionary scale using a simple measure of complexity. However, the exploration of different aspects of genitalia besides complexity, such as size, shape and mechanical function, may reveal other processes and yield a more complete view of the evolution of copulatory organs. Therefore, detailed studies of descriptive and comparative morphology, copulatory mechanisms and sexual behaviour in populations and groups of species and genera are desirable in order to contribute to a better understanding of the mechanism generating the genital diversity of Gnaphosidae. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Appendix S1. Data matrix. Appendix S2. Character optimizations. Table S1. Epigynum and palp complexity values and respective proportion of missing data for the taxa used in complexity analysis. ACKNOWLEDGEMENTS We would like to thank Almir R. Pepato (Universidade Federal de Minas Gerais), Antonio D. Brescovit (Instituto Butantan, São Paulo), Fernando A. Silveira (Universidade Federal de Minas Gerais), Lina M. A. Silva (Instituto Butantan, São Paulo), Martín J. Ramírez (Museo Argentino de Ciencias Naturales), B. A. Huber and an anonymous reviewer for valuable comments on a previous version of the manuscript. This work was supported by grants from the Exline-Frizzell Fund of California Academy of Sciences, a Coordenação de Aperfeiçoamento de Pessoal de Nível Superior scholarship to G.H.F.A., and grants from Conselho Nacional de Desenvolvimento Científico e Tecnológico (407288/2013-9, 308072/2012-0), Fundação de Amparo à Pesquisa do Estado de Minas Gerais (PPM 00335-13, PPM 00651-15) and Instituto Nacional de Ciência e Tecnologia dos Hymenoptera Parasitóides da Região Sudeste Brasileira (http://www.hympar.ufscar.br/) to A.J.S. The authors declare no conflict of interest. REFERENCES Adami C . 2002 . What is complexity ? BioEssays: News and Reviews in Molecular, Cellular and Developmental Biology 24 : 1085 – 1094 . Google Scholar CrossRef Search ADS Adami C , Ofria C , Collier TC . 2000 . Evolution of biological complexity . Proceedings of the National Academy of Sciences of the United States of America 97 : 4463 – 4468 . Google Scholar CrossRef Search ADS Adams D , Rohlf F , Slice DE . 2004 . Geometric morphometrics: ten years of progress following the ‘revolution’ . Italian Journal of Zoology 71 : 5 – 16 . Google Scholar CrossRef Search ADS Ah-King M , Barron AB , Herberstein ME . 2014 . Genital evolution: why are females still understudied ? PLoS Biology 12 : e1001851 . Google Scholar CrossRef Search ADS Aisenberg A , Barrantes G , Eberhard WG . 2015 . Post-copulatory sexual selection in two tropical orb-weaving Leucauge spiders . In: Peretti AV , Aisenberg A , eds. Cryptic female choice in arthropods: patterns, mechanisms and prospects . Heidelberg : Springer , 79 – 108 . Google Scholar CrossRef Search ADS Anderson CM , Langerhans RB . 2015 . Origins of female genital diversity: predation risk and lock-and-key explain rapid divergence during an adaptive radiation . Evolution; International Journal of Organic Evolution 69 : 2452 – 2467 . Google Scholar CrossRef Search ADS Arnqvist G . 1997 . The evolution of animal genitalia: distinguishing between hypotheses by single species studies . Evolution in Health and Disease 60 : 365 – 379 . Azevedo GHF , Griswold CE , Santos AJ . 2017 . Systematicsand evolution of ground spiders revisited (Araneae, Dionycha, Gnaphosidae) . Cladistics : 1 – 48 . doi: 10.1111/cla.12226 . Baker J , Meade A , Pagel M , Venditti C . 2015 . Adaptive evolution toward larger size in mammals . Proceedings of the National Academy of Sciences of the United States of America 112 : 5093 – 5098 . Google Scholar CrossRef Search ADS Bertani R . 2000 . Male palpal bulbs and homologous features in theraphosinae (Araneae, Theraphosidae) . Journal of Arachnology 28 : 29 – 42 . Google Scholar CrossRef Search ADS Burger M , Michalik P , Graber W , Jacob A , Nentwig W , Kropf C . 2006 . Complex genital system of a haplogyne spider (Arachnida, Araneae, Tetrablemmidae) indicates internal fertilization and full female control over transferred sperm . Journal of Morphology 267 : 166 – 186 . Google Scholar CrossRef Search ADS Burger M , Nentwig W , Kropf C . 2003 . Complex genital structures indicate cryptic female choice in a haplogyne spider (Arachnida, Araneae, Oonopidae, Gamasomorphinae) . Journal of Morphology 255 : 80 – 93 . Google Scholar CrossRef Search ADS Calbacho-Rosa L , Galicia-Mendoza I , Dutto MS , Córdoba-Aguilar A , Peretti AV . 2013 . Copulatory behavior in a pholcid spider: males use specialized genitalic movements for sperm removal and copulatory courtship . Die Naturwissenschaften 100 : 407 – 416 . Google Scholar CrossRef Search ADS Calbacho-Rosa L , Peretti AV . 2015 . Copulatory and post-copulatory sexual selection in haplogyne spiders, with emphasis on Pholcidae and Oonopidae . In: Peretti AV , Aisenberg A , eds. Cryptic female choice in arthropods . Heidelberg : Springer , 109 – 144 . Google Scholar CrossRef Search ADS Coddington JA . 1990 . Ontogeny and homology in the male palpus of orb-weaving spiders and their relatives, with comments on phylogeny (Araneoclada: Araneoidea, Deinopoidea) . Smithsonian Contributions to Zoology 496 : 1 – 52 . Google Scholar CrossRef Search ADS Coddington JA , Levi HW . 1991 . Systematics and evolution of spiders (Araneae) . Annual Review of Ecology and Systematics 22 : 565 – 592 . Google Scholar CrossRef Search ADS Eberhard WG . 1985 . Sexual selection and animal genitalia . Cambridge, MA : Harvard University Press . Google Scholar CrossRef Search ADS Eberhard WG . 2004a . Why study spider sex: special traits of spiders facilitate studies of sperm competition and cryptic female choice . Journal of Arachnology 32 : 545 – 556 . Google Scholar CrossRef Search ADS Eberhard WG . 2004b . Rapid divergent evolution of sexual morphology: comparative tests of antagonistic coevolution and traditional female choice . Evolution 58 : 1947 – 1970 . Google Scholar CrossRef Search ADS Eberhard WG . 2010a . Evolution of genitalia: theories, evidence, and new directions . Genetica 138 : 5 – 18 . Google Scholar CrossRef Search ADS Eberhard WG . 2010b . Rapid divergent evolution of genitalia . In: Leonard J , Córdoba-Aguilar A , eds. The evolution of primary sexual characters in animals . Oxford : Oxford University Press , 40 – 78 . Eberhard WG . 2015 . Cryptic female choice and other types of post-copulatory sexual selection . In: Peretti AV , Aisenberg A , eds. Cryptic female choice in arthropods . Heidelberg : Springer , 1 – 26 . Google Scholar CrossRef Search ADS Eberhard WG , Huber BA . 2010 . Spider genitalia – precise maneuvers with a numb structure in a complex lock . In: Leonard J , Córdoba-Aguilar A , eds. The evolution of primary sexual characters in animals . Oxford : Oxford University Press , 249 – 284 . Forster RR . 1980 . Evolution of the tarsal organ, the respiratory system and the female genitalia in spiders . In: Gruber J , ed. Verhandlungen der 8 Internationaler Arachnologen-Kongress, Wien . Vienna : Verlag H. Engermann , 269 – 284 . Frazee SR , Masly JP . 2015 . Multiple sexual selection pressures drive the rapid evolution of complex morphology in a male secondary genital structure . Ecology and Evolution 5 : 4437 – 4450 . Google Scholar CrossRef Search ADS Garrison NL , Rodriguez J , Agnarsson I , Coddington JA , Griswold CE , Hamilton CA , Hedin M , Kocot KM , Ledford JM , Bond JE . 2016 . Spider phylogenomics: untangling the Spider Tree of Life . PeerJ 4 : e1719 . Google Scholar CrossRef Search ADS Hosken DJ , Stockley P . 2004 . Sexual selection and genital evolution . Trends in Ecology & Evolution 19 : 87 – 93 . Google Scholar CrossRef Search ADS Huber BA . 1999 . Sexual selection in pholcid spiders (Araneae, Pholcidae) . Journal of Arachnology 27 : 135 – 141 . Huber BA . 2003 . Rapid evolution and species-specificity of arthropod genitalia: fact or artifact ? Organisms Diversity & Evolution 3 : 63 – 71 . Google Scholar CrossRef Search ADS Huber BA . 2004 . The significance of copulatory structures in spider systematics . In: Schult J , ed. Studien zur Theorie der Biologie, Band 5, Biosemiotik - Praktische Anwendung und Konsequenzen für die Einzeldisziplinen . Berlin : VWB-Verlag für Wissenschaft und Bildung , 89 – 100 . Jocqué R . 2002 . Genitalic polymorphism—a challenge for taxonomy . Journal of Arachnology 30 : 298 – 306 . Google Scholar CrossRef Search ADS Kass RE , Raftery AE . 1995 . Bayes factors . Journal of American Statistical Association 90 : 773 – 795 . Google Scholar CrossRef Search ADS Kraus O . 1978 . Liphistius and the evolution of spider genitalia . Symposia of the Zoological Society of London 42 : 235 – 254 . Kuntner M , Coddington JA , Schneider JM . 2009 . Intersexual arms race? Genital coevolution in nephilid spiders (Araneae, Nephilidae) . Evolution; International Journal of Organic Evolution 63 : 1451 – 1463 . Google Scholar CrossRef Search ADS Lipke E , Hammel JU , Michalik P . 2015 . First evidence of neurons in the male copulatory organ of a spider . Biology Letters 11 : 20150465 . Google Scholar CrossRef Search ADS Maddison WP . 1991 . Squared-change parsimony reconstruction of ancestral states for continuous-valued characters on a phylogenetic tree . Systematic Zoology 40 : 304 – 314 . Google Scholar CrossRef Search ADS Maddison WP , Maddison DR . 2015 . Mesquite: a modular system for evolutionary analysis. Version 3.04 . Available at: mesquiteproject.org/mesquite/download/download.html McShea DW . 1991 . Complexity and evolution: what everybody knows . Biology and Philosophy 6 : 303 – 324 . Google Scholar CrossRef Search ADS Nixon KC . 2002 . WinClada, version 1.00.08 . Ithaca, NY : published by the author . Pagel M . 1994 . Detecting correlated evolution on phylogenies: a general method for the comparative analysis of discrete characters . Proceedings of the Royal Society of London B: Biological Sciences 255 : 37 – 45 . Google Scholar CrossRef Search ADS Pagel M . 1997 . Inferring evolutionary processes from phylogenies . Zoologica Scripta 26 : 331 – 348 . Google Scholar CrossRef Search ADS Pagel M . 1999 . Inferring the historical patterns of biological evolution . Nature 401 : 877 – 884 . Google Scholar CrossRef Search ADS Pagel M , Meade A . 2014 . Bayes Traits V2. Computer program and documentation . Available at: http://www. evolution. rdg. ac. uk/BayesTraits. html. Pan W . 2001 . Akaike’s information criterion in generalized estimating equations . Biometrics 57 : 120 – 125 . Google Scholar CrossRef Search ADS Paradis E , Claude J . 2002 . Analysis of comparative data using generalized estimating equations . Journal of Theoretical Biology 218 : 175 – 185 . Google Scholar CrossRef Search ADS Paradis E , Claude J , Strimmer K . 2004 . APE: analyses of phylogenetics and evolution in R language . Bioinformatics (Oxford, England) 20 : 289 – 290 . Google Scholar CrossRef Search ADS de Pinna MGG . 1991 . Concepts and tests of homology in the cladistic paradigm . Cladistics 7 : 367 – 394 . Google Scholar CrossRef Search ADS Platnick NI , Dupérré N . 2009 . The American goblin spiders of the new genus Escaphiella (Araneae, Oonopidae) . Bulletin of the American Museum of Natural History 328 : 1 – 151 . Google Scholar CrossRef Search ADS Platnick N , Gertsch WJ . 1976 . The suborders of spiders: a cladistic analysis (Arachnida, Araneae) . American Museum Novitates 2607 : 1 – 15 . Platnick NI , Shadab MU . 1975 . A revision of the spider genus Gnaphosa (Araneae, Gnaphosidae) in America . Bulletin of the American Museum of Natural History 155 : 1 – 66 . Platnick NI , Shadab MU . 1980a . A Revision of the North American Spider Genera Nodocion, Litopyllus, and Synaphosus (Araneae, Gnaphosidae) . American Museum Novitates 2691 : 1 – 26 . Platnick NI , Shadab MU . 1980b . A revision of the spider genus Cesonia (Araneae, Gnaphosidae) . Bulletin of the American Museum of Natural History 165 : 337 – 385 . Platnick NI , Shadab MU . 1984 . A revision of the neotropical spiders of the genus Apopyllus (Araneae, Gnaphosidae) . American Museum Novitates 2788 : 1 – 9 . R Core Development Team . 2015 . R: a language and environment for statistical computing, 3.2.1 . Available at: http://www. r-project. org Ramírez MJ . 2007 . Homology as a parsimony problem: a dynamic homology approach for morphological data . Cladistics 23 : 588 – 612 . Ramírez MJ . 2014 . The morphology and phylogeny of dionychan spiders (Araneae: Araneomorphae) . Bulletin of the American Museum of Natural History 390 : 1 – 374 . Google Scholar CrossRef Search ADS Ramírez MJ , Michalik P . 2014 . Calculating structural complexity in phylogenies using ancestral ontologies . Cladistics 30 : 635 – 649 . Google Scholar CrossRef Search ADS Richter S . 2005 . Homologies in phylogenetic analyses–concept and tests . Theory in Biosciences = Theorie in den Biowissenschaften 124 : 105 – 120 . Rowe L , Arnqvist G . 2012 . Sexual selection and the evolution of genital shape and complexity in water striders . Evolution; International Journal of Organic Evolution 66 : 40 – 54 . Google Scholar CrossRef Search ADS Schneider JM , Herberstein ME , De Crespigny FC , Ramamurthy S . 2000 . Sperm competition and small size advantage for males of the golden orb-web spider Nephila edulis . Journal of Evolutionary Biology 13 : 939 – 946 . Google Scholar CrossRef Search ADS Schneider J , Uhl G , Herberstein ME . 2015 . Cryptic female choice within the genus Argiope: a comparative approach . In: Peretti AV , Aisenberg A , eds. Cryptic female choice in arthropods: patterns, mechanisms and prospects . Heidelberg : Springer , 55 – 77 . Google Scholar CrossRef Search ADS Senglet A . 2004 . Copulatory mechanisms in Zelotes, Drasillus and Trachyzelotes (Araneae, Gnaphosidae) with additional faunistic and taxonomy data on species from Southwest Europe . Mitteilungen der schweizerischen entomologischen Gesellschaft 77 : 87 – 119 . Sentenská L , Pekár S , Lipke E , Michalik P , Uhl G . 2015 . Female control of mate plugging in a female-cannibalistic spider (Micaria sociabilis) . BMC Evolutionary Biology 15 : 18 . Google Scholar CrossRef Search ADS Sentenská L , Müller CHG , Pekár S , Uhl G . 2017 . Neurons and a sensory organ in the pedipalps of male spiders reveal that it is not a numb structure . Scientific Reports 7 : 12209 . Google Scholar CrossRef Search ADS Sereno PC . 2007 . Logical basis for morphological characters in phylogenetics . Cladistics 23 : 565 – 587 . Sierwald P . 1989 . Morphology and ontogeny of female copulatory organs in American Pisauridae, with special reference to homologous features (Arachnida, Araneae) . Smithsonian Contributions to Zoology : 1 – 24 . Sierwald P . 1990 . Morphology and homologous features in the male palpal organ in Pisauridae and other spider families, with notes on the taxonomy of Pisauridae (Arachnidae: Araneae) . Nemouria 35 : 1 – 59 . Simmons LW . 2014 . Sexual selection and genital evolution . Austral Entomology 53 : 1 – 17 . Google Scholar CrossRef Search ADS Uhl G , Nessler SH , Schneider JM . 2010 . Securing paternity in spiders? A review on occurrence and effects of mating plugs and male genital mutilation . Genetica 138 : 75 – 104 . Google Scholar CrossRef Search ADS Welke K , Schneider JM . 2009 . Inbreeding avoidance through cryptic female choice in the cannibalistic orb-web spider Argiope lobata . Behavioral Ecology 20 : 1056 – 1062 . Google Scholar CrossRef Search ADS World Spider Catalog . 2018 . World spider catalog. Version 19.0 . Bern : Natural History Museum Bern . Available at: http://wsc.nmbe.ch. Accessed 20 February 2018. Zakharov BP , Ovtcharenko VI . 2011 . Morphological organization of the male palpal organ in Australian ground spiders of the genera Anzacia, Intruda, Zelanda, and Encoptarthria (Araneae: Gnaphosidae) . Journal of Arachnology 39 : 327 – 336 . Google Scholar CrossRef Search ADS Zakharov B , Ovtcharenko V . 2013 . Male palp organ morphology of three species of ground spiders (Araneae, Gnaphosidae) . Arachnologische Mitteilungen 45 : 15 – 20 . Google Scholar CrossRef Search ADS Zelditch ML , Swiderski DL , Sheets HD , Fink WL . 2004 . Geometric morphometrics for biologists: a primer . London : Elsevier . © 2018 The Linnean Society of London, Zoological Journal of the Linnean Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Zoological Journal of the Linnean Society – Oxford University Press

**Published: ** May 2, 2018

Loading...

personal research library

It’s your single place to instantly

**discover** and **read** the research

that matters to you.

Enjoy **affordable access** to

over 18 million articles from more than

**15,000 peer-reviewed journals**.

All for just $49/month

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Read from thousands of the leading scholarly journals from *SpringerNature*, *Elsevier*, *Wiley-Blackwell*, *Oxford University Press* and more.

All the latest content is available, no embargo periods.

## “Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”

Daniel C.

## “Whoa! It’s like Spotify but for academic articles.”

@Phil_Robichaud

## “I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”

@deepthiw

## “My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”

@JoseServera

DeepDyve ## Freelancer | DeepDyve ## Pro | |
---|---|---|

Price | FREE | $49/month |

Save searches from | ||

Create folders to | ||

Export folders, citations | ||

Read DeepDyve articles | Abstract access only | Unlimited access to over |

20 pages / month | ||

PDF Discount | 20% off | |

Read and print from thousands of top scholarly journals.

System error. Please try again!

Already have an account? Log in

Bookmark this article. You can see your Bookmarks on your DeepDyve Library.

To save an article, **log in** first, or **sign up** for a DeepDyve account if you don’t already have one.

Copy and paste the desired citation format or use the link below to download a file formatted for EndNote

**EndNote**

All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.

ok to continue