Taxon Sample

Analyses were conducted using 25 species from the eastern North American clade of leiobunine harvestmen, and four outgroup species from a closely related clade occurring in Mexico and the western United States [27]. The sample included all genera and all but six described species from the eastern clade plus four undescribed species. Because discrete genital morphology does not vary within species and monophyly of species groups is well supported [26] (Fig. 4), we conducted our analyses using one population (i.e., one tip) from each multiply represented species examined by [26] in both the phylogenetic and morphological analyses (Fig. 4). Table S1 in the supplementary materials includes additional details regarding taxon sampling for molecular and morphological assignment.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Phylogenetic hypotheses and distribution of reproductive characters.

The maximum clade credibility Bayesian tree (left) was assembled using the TreeAnnotator program [31], visualized with FigTree v1.3.1 [73], and depicts relationships recovered in BEAST v1.7.1 [31] for trees that passed the backbone constraint tree (right). Values above branches indicate the posterior probabilities per node for filtered trees (n = 431). Values below braches are the posterior probabilities of the maximum clade credibility tree for a subset of 1000 random trees resampled from the original posterior probability distribution. Scale is in substitutions per site for the filtered subset maximum clade credibility tree. The most parsimonious distribution of reproductive characters (assuming no parallel gains in penile sacs) is mapped to the maximum clade credibility tree. Geographic codes are given for undescribed species: IL = Illinois, MO = Missouri, NE = Nebraska, TN = Tennessee. The backbone constraint tree (right) depicts relationships that were well supported (>95% posterior probability) in the [26] tree and that were used to generate sets of trees for the present study.

https://doi.org/10.1371/journal.pone.0066767.g004

Phylogenetic Trees

All analyses were conducted using as a template the phylogenetic tree recovered by Bayesian analysis of mitochondrial and nuclear genes in [26]. However, because the branch lengths of the original topology reflect rates of molecular rather than morphological evolution, we used the same molecular data (see Table S1 for GenBank accession numbers) to generate a set of ultrametric trees in which internodal lengths reflect time and lengths of all root-to-tip pathways were equal [30]. Ultrametric trees were constructed using BEAST v1.7.1 [31] assuming a Yule speciation process prior. The data matrix was divided into three partitions—mitochondrial DNA, 28 S rDNA and elongation factor 1-α—analyzed simultaneously using separate GTR+I+Γ models. Ultrametric branch lengths were calculated using unlinked and uncorrelated log-normal relaxed clocks separated by partition [32]. Two independent tree-searching analyses each ran for 100 million iterations, where one configuration was sampled per 1000 generations with the default 10% burn-in (Data deposited in the Dryad on-line repository: (http://dx.doi.org/10.5061/dryad.79d15).

The program TRACER v1.5 [33] was used to ensure that effective sample sizes of the posterior distribution were greater than 1000 for each independent analysis. To achieve a more conservative burn-in of 30%, we discarded an additional 20% of sampled trees using LogCombiner v1.7.1 [31]. The posterior distributions of the two analyses were pooled to yield 1000 trees. Multiply represented taxa were pruned to one population per species (see Table S1 for localities) by list-applying (command ‘lapply’) the “drop.tip” function to the entire set of trees in the ape package [34] available through the R statistical computing language [35]. To ensure consistency with the branching pattern of the original Bayesian tree [26], the posterior distribution was filtered using a rooted backbone constraint tree (Fig. 4) in PAUP* v4.0b [36] which preserved well-supported clades (i.e., posterior probabilities >0.95) while allowing for variation in the placement of poorly supported clades and species. This resulted in a distribution of 431 trees that was used in all analyses of character evolution.

Evolution of the Penis and Male Pedipalps

Males of each species were assigned one of three combinations of penile-sac (S) and pedipalpal (P) features. Species with bilateral penile cuticular sacs that convey a nuptial gift [37] and simple “female-like” pedipalps were coded as S⁺P⁻; species that lack penile sacs and have simple pedipalps were coded as S⁻P⁻; and species that lack sacs but have pedipalps heavily modified for clasping (Fig. 3) were coded as S⁻P⁺ (Figs. 5A, 5B). States were determined for all species by original observations of anatomy. No species is known to have both penile sacs and modified pedipalps, so this combination of traits was not coded. That this combination is unobserved gives strength to our alternative model of male reproductive evolution, so we chose to ignore it, although alternative approaches might include the combination [38].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Transition models used to test hypotheses for the evolution of reproductive characters with Bayesian analysis.

(A) No-precedence model of male morphological evolution versus (B) Penis precedence model, where male morphological transitions are limited to sacculate (S⁺) to nonsacculate (S⁻) penis and simple pedipalps (P⁻) to enhanced pedipalps (P⁺). The root of A was treated as fixed to S⁺P⁻ (Table 1, row 3) or determined empirically (Table 1, row 1). (C) Independent and (D) dependent models of discrete male and female reproductive morphology. Here, the female pregenital barrier is coded as present (B⁺) or absent (B⁻). Both models allow for all possible character transitions. (E) No precedence model was compared to dependent model (F), where character precedence is possible. In this model, penile sac loss (S⁺B⁻ →S⁻B⁻) and barrier acquisition (S⁺B⁻ →S⁺B⁺) are constrained to have equal rates of evolution.

https://doi.org/10.1371/journal.pone.0066767.g005

The ancestral male morphology was determined using parsimony reconstructions with Mesquite v. 2.75 [39] and with BayesTraits Multistate [40]–[41]. The latter was accomplished by comparing marginal likelihoods of two models: a 6-rate model in which all transitions between character states were possible, and a model that differs only in that state 0 (i.e., no penile sacs, simple pedipalps) was assigned to the root. As these models are not nested, they were compared using Bayes factors [42].

To assess the direction of change in male morphology, two potential models of male character evolution were compared: the 6-rate model representing the possibility for transitions between all three character states (Fig. 5A), and a 2-rate model restricting transitions to the loss of penile sacs (S⁺P⁻→S⁻P⁻) followed by the gain of modified pedipalps (S⁻P⁻→S⁻P⁺) (Fig. 5B). The 2-rate model is an evolutionary trajectory wherein each transition is consistent with escalation in intersexual antagonism during mating. Changes from S⁻P⁺→S⁻P⁻→S⁺P⁻, possible in the 6-rate model, suggest decreasing precopulatory antagonism and/or an increase in reliance on courtship (i.e., female appeasement by the male).

Evolution of the Penis and Female Genital Operculum

Each species was assigned one of two discrete states for each character. The penis was coded as having either a bilateral pair of cuticular sacs that convey a nuptial gift (S⁺) or as lacking sacs (S⁻); the female genital operculum was coded as either unarmed (B⁻) (Fig. 2B) or as elaborated to form a pregenital barrier (B⁺) (Figs. 2C, 2D). States were determined for all species by original observations of anatomy. We interpreted the evolutionary changes S⁺→S⁻ and/or B⁻→B⁺ as evidence for an increase in precopulatory antagonism and/or a decrease in female appeasement by the male and change in the opposite direction as a decrease in precopulatory antagonism and/or an increase in female appeasement by the male.

Ancestral states were determined for each character using parsimony [39] and a hierarchical Bayesian method [43] implemented in SIMMAP v. 1.5 [44]. In the Bayesian approach, each character was modeled separately in accordance with standards outlined in [45]; we used either an empirical character-bias prior derived from the frequency of terminal states or a β-distribution prior where the best-fit α-shape value was derived from Markov chain Monte Carlo (MCMC) sampling [44]. The overall evolutionary rate for each character set was modeled using a Γ-tree prior obtained via MCMC sampling for the α-shape parameter and β-rate parameter [44]. Analyses were replicated with and without outgroup taxa to assess outgroup effects on the relative rates of character change. Root states were inferred from the marginal posterior probabilities for each state across all sub-sampled, outgroup-rooted trees (n = 431) with fixed branch lengths for each character.

We determined whether state changes in the penis and female genital operculum were correlated using the Discrete module in BayesTraits [40]. This was done by comparing the marginal likelihoods of two models: an independent 4-rate model in which state changes in the penis and female genital operculum were estimated separately (Fig. 5C) and a dependent 8-rate model (Fig. 5D) in which single-step changes between the four penis-operculum combinations (S⁺B⁻, S⁺B⁺, S⁻B⁺, S⁻B⁻) were estimated. Thus a comparison of log likelihoods that favors the 4-rate model indicates no association between state changes, and a comparison that favors the dependent model indicates correlated change between the penis and female genital operculum.

SIMMAP was also used to test for correlations between male and female genital morphology across the posterior tree distribution using predictive sampling and stochastic character mapping via a continuous-time Markov chain [46]. The overall evolutionary rate for each character set was modeled with the Γ-distribution prior used in the ancestral state reconstructions, and bias priors for male and female characters were modeled either as β-distributions or empirical priors as in the ancestral state reconstruction analyses. Bayesian parametric bootstrapping was conducted by sampling each tree 10 times with 10 prior draws for a total of 43,100 samples for all model parameters. Results were summarized as M-values (i.e., the correlation between the histories of two characters across the phylogeny) and p-values (i.e., the probability that an association between penis state and female barrier presence/absence as extreme as observed could arise simply by chance).

In contrast to parsimony, likelihood- or Bayesian-based trait-evolution methods can potentially assess whether change in one state is more likely to precede change in another—even along the same branch—by assigning different rates to these changes. Those states with higher rates are more likely to occur before changes in states with lower rates [47]. Assuming character dependence, it is therefore possible to test whether one character state change (e.g., penis loses sacs) promotes a different character state change (e.g., female gains pregenital barrier). We tested whether nuptial sac loss or pregenital barrier gains were significantly different by using the Discrete module in BayesTraits by comparing a dependent, “precedence-possible” 8-rate model in which transitions between the four penis-operculum combinations were estimated simultaneously (Fig. 5F) to a dependent 7-rate model (Fig. 5E) wherein gain of the pregenital barrier (S⁺B⁻→S⁺B⁺) and loss of penile sacs (S⁺B⁻→S⁻B⁻) were assumed to occur at the same rate. A comparison of log-likelihoods that favors the dependent, “no precedence” 7-rate model would indicate that the rates of increased antagonism from the ancestral condition are equivalent between the sexes, whereas a comparison favoring the 8-rate model indicates a difference between the rates of escalation between the sexes. In the event the 8-rate model is favored, the mean and variance of rates of escalation can be further compared between the sexes. The sex that was most likely to have initiated the escalation can then be determined by its significantly higher mean rate of morphological change.

General Procedures for BayesTraits Model Testing

All model comparisons in BayesTraits (Fig. 5) were made after analyzing trait evolution using a Markov-chain Monte Carlo algorithm with standard uniform rate priors, 2.1×10⁸ to 6.0×10⁹ iterations, 30% burn-in, and a rate deviation of 0.001–2.0 in order to reach a target acceptance rate of 20–40% per run. At least four independent analyses were performed for each model (see Table 1). Log files were uploaded to TRACER [32] to determine stabilization of log-likelihoods (standard error of no more than 0.03 and a visual inspection of the harmonic mean traces). Although model harmonic means should theoretically approach model marginal model likelihoods [40]–[41]; [48], this use of harmonic means has been criticized [48]–[49]. Therefore we chose to approximate marginal likelihoods using the "Analysis → Calculate Bayes Factors" function of TRACER [33] summarized in [50] with modifications by [51], calculating 1000 bootstrap replicates of the log-likelihood traces. The mean Bayes factor for each model was calculated and used in model log-likelihood comparisons (Table 1). Where model pairs of interest were nested, marginal likelihood approximations were compared using log-likelihood ratio tests. Except where noted, significance was determined where the test statistic value surpassed the χ² distribution critical value at an α value of 0.05. Degrees of freedom were calculated by solving for the difference in estimated parameters.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Model Bayes factors.

https://doi.org/10.1371/journal.pone.0066767.t001

Evolution of the Penis and Male Pedipalps

The ancestral male reproductive morphology was inferred by considering the likelihood of the evolutionary trajectory of male traits when root state was fixed or not fixed. A comparison of Bayes factors from the 6-rate, fixed-root model (Fig. 5A), where a root state of S⁺P⁻ was constrained, to those from a similar model where no constraint was imposed, showed no appreciable difference in marginal likelihoods of the models (K = 0.538). This result indicates that the co-occurrence of penile sacs and simple male pedipalps is the primitive state for the eastern North American clade of leiobunine harvestmen, which is consistent with the conclusion based on parsimony (Fig. 4).

Comparison of the marginal likelihood approximations of the 6-rate “no precedence” model, where change between any of the three discrete male reproductive characters is possible, to a 2-rate “penis-precedence” model, where only two transitions are allowed, indicated no significant difference between the two models (log-likelihood ratio test: χ² = 2.996, D.F. = 4, p>0.1). Given this result, the simpler 2-rate “penis-precedence” model is preferred, and we conclude that there may have been a tendency for penile sacs to be lost before the male pedipalps were enhanced for clasping the female. The likelihood of this model is further supported by the lack of species with both enhanced pedipalps and penile sacs.

Evolution of the Penis and Female Genital Operculum

In order to assess the ancestral states of male and female reproductive morphology with hierarchical Bayesian analysis, the probability distribution priors were estimated for each character set using an MCMC-sampling method [43]. In all SIMMAP analyses, Γ-tree priors for the overall evolutionary rate of each trait were applied. The overall evolutionary rate best-fit shape (α) and rate (β) parameters for penis morphology were α = 3.515, β = 0.038, and for barrier presence, α = 3.108, β = 0.036. Character-bias priors were modeled with either an empirical approach based on the frequency of tip states, or with a β-distribution prior. The best-fit α values for each character-bias distribution were α = 5.888 for penis morphology and α = 5.906 for morphology of the female genital operculum.

Character mapping under parsimony (Fig. 4) supported the parallel loss of penile sacs from a sacculate ancestor (S⁺→S⁻) and gain of female pregenital barricade from ancestors with an unarmed genital operculum (B⁻→B⁺). At least four such transitions are necessary for each character (Fig. 4), although this number is dependent on topology, which we varied in our analyses due to species paraphyly [26]. These results were consistent with those obtained from the Bayesian approach implemented in SIMMAP. The presence of penile sacs (S⁺) was recovered as the ancestral male character state with marginal posterior probabilities ranging from 78% to 80% (with probability being dependent on use of either the two-state empirical or β-bias prior and inclusion/exclusion of the outgroup). Absence of a pregenital barrier (B⁻) was the most likely ancestral female character state, with marginal posterior probabilities of 77% to 96% (with the probability being dependent solely on inclusion or exclusion of outgroup character states in the analysis). Results from two methods thus support an ancestral taxon wherein males had sacculate penes and females lacked a pregenital barrier.

To assess the hypothesis that there are two syndromes of coevolved morphological features, we first needed to determine whether state changes in the penis and female barrier were correlated. We used the BayesTraits Discrete module to compare marginal likelihoods of two alternative hypotheses modeling either independent or dependent change in traits (Figs. 5C vs 5D). Log-likelihood ratio tests of the marginal likelihood approximations of these models favored the dependent, 8-rate model (χ² = 9.672, D.F. = 4, p<0.05). We conclude that the evolution of male and female reproductive structures is correlated across the phylogeny.

Using SIMMAP we also demonstrated a correlation between male and female reproductive morphology. Bayesian predictive distributions were generated using stochastic mapping of male and female reproductive traits to the filtered posterior tree distribution. When compared to the actual trait states by species, a mean correlation between penis morphology and female pregenital barrier presence of 0.147 (p<0.01) was found under the empirical prior, and a correlation of 0.151 (p<0.05) was derived using the β-bias prior. Individual state covariation between sacculate penis type and absent pregenital barrier (Empirical: m₀₀ = 0.063, p<0.01, β: m₀₀ = 0.065, p<0.01) and non-sacculate penis type and present pregenital barrier (Empirical: m₁₁ = 0.068, p<0.01, β: m₁₁ = 0.07, p<0.05) was found to be positive and significantly distinct from the predictive distribution. The relationships of sacculate penis type with presence of pregenital barrier (Empirical: m₀₁ = −0.054, p<0.01, β: m₀₁ = −0.055, p<0.01) and non-sacculate penis type with a lack of female pregenital barrier (Empirical: m₁₀ = −0.052, p<0.01, β: m₁₀ = −0.053, p<0.01)—both trait combinations seen in a small but non-zero number of species in the phylogeny—co-varied negatively, yet remained significantly different from the predictive distribution.

As male and female morphology was demonstrated to covary across the phylogeny, we additionally tested whether the evolutionary rate at which penile sacs were lost was equal or unequal to the rate at which females acquired the pregenital barrier, all relative to the sacculate, barrier-free ancestor. We compared the likelihood of a 7-rate “No Precedence” model where the rates of pregenital barrier acquisition and loss of penile sacs were forced to be equal, to an 8-rate "Precedence-Possible" model (Fig. 5E). Comparisons of marginal likelihoods revealed a significant difference between models and favored the 8-rate scheme (Fig. 5F) (χ² = 9.936, D.F. = 1, p<0.01). Thus, the rates of change of the penis and female genital operculum cannot be assumed to be equal, and the precedence of one sex's trait change over the other is supported. However when comparing rates of character change and accounting for rate variance, the “Precedence-Possible” model does not appear, on average, to estimate a higher rate for either transition (μ_{q12 vs.q13} = 10.94±18.27, D.F. = 3, t = 1.905, p = 0.0765). An increased number of simulations or alternative priors on μ might change the significance of this difference. Ultimately, there is evidence that changes in the penis and female genital operculum are correlated and that rates of state changes are unequal, which suggests that change in one may precede change in the other. Parsimony on the backbone constraint tree (Fig. 4) suggests pregenital barrier development may have preceded the loss of sacs, but this result is subject to tree topology. As uncertainty in topology was included in the model testing procedure, no conclusion regarding character evolution precedence may be made by parsimony alone, although improved sampling and resolution of the Hadrobunus species group might alter this. We found no evidence that change in one sex strongly tended to precede change in the other.

Patterns in the Evolution of Reproductive Structures in Leiobunine Harvestmen

The results from our analysis indicate that the leiobunine harvestmen of eastern North America are descended from an ancestor with reproductive structures that are consistent with a mating system dominated by courtship where males entice or appease females to obtain copulation. Results from both Bayesian and parsimony-based methods of character reconstruction showed that ancestral males had a subterminal pair of penile cuticular sacs used in conveying a nuptial secretion to the female during the precopulatory phase of mating (Figs. 1B, 3). The male pedipalps were used to clasp the female at the base of her second leg pair but were morphologically similar to those of females. The pregenital openings of female lacked sclerites that might serve as a barrier to forced intromission by the male. This syndrome of reproductive features has persisted in several diverse lineages, and these offer opportunities to explore further the details of the ancestral system. It is reasonable to suppose that the evolutionary mechanism of female mate choice has played a predominant role in shaping the ancestral reproductive syndrome.

There have also been at least four phylogenetically independent transitions from the ancestral system toward morphologies and behaviors consistent with precopulatory antagonistic behavior. This assumes no parallel gains of the nuptial gift sacs, a reasonable assumption given the complexity of the structure and its function in mating [52]. We suspect that the number of independent transitions from sacculate to non-sacculate conditions will increase as phylogenetic relationships within the Hadrobunus group are clarified. In each case, penile sacs have been lost and females have evolved sclerotized pregenital barriers. In addition, the male pedipalps of species within the calcar and vittatum species-groups are enhanced for clasping the female. Each transition has resulted in a different construction of the penis, the female pregenital barrier (Fig. 2) and male pedipalps (compare Figs. 1A, 12).

Our results indicate that loss of penile sacs and elaboration of male pedipalps are correlated. In fact, modified pedipalps always co-occur with non-sacculate penes, although not all non-sacculate species have enhanced male pedipalps. This evolutionary trajectory supports the hypothesis that precopulatory antagonism has originated or increased several times in leiobunine phylogeny, a hypothesis further supported by the correlation found between male and female morphological states. There is some additional evidence from our Bayesian modeling of the evolution of the penis and male pedipalps that the loss of penile sacs tends to precede the elaboration of the male pedipalps (Fig. 4). Interestingly, there are few morphological specializations in females that appear to be dedicated to resisting clasping by males. The only possible exception occurs in the Hadrobunus group, where females in all species (both sacculate and non-sacculate) have a spike-like process or “coxal spur” on the posterior margin of the basal segment of the second leg, where the base of the male pedipalpal tarsus likely contacts the female [53].

The timing of the loss of penile sacs and gain of pregenital barriers are strongly correlated. The Bayesian analysis of character evolution showed that a difference likely exists in the rates of the two transformations, which may indicate a tendency for one kind of evolutionary change to precede the other [47]. However, additional tests aimed at resolving these rates failed to find significant differences, and it was not possible to determine whether evolution in the structures of one sex tends to lead the coevolutionary change.

The distribution of morphological characters made parsimony-based character mapping similarly uninformative for reconstructing the sequence of change in the penis and pregenital barrier. The ancestral condition (penile sacs present, pregenital barrier absent) and one derived condition (penile sacs absent, pregenital barrier present) were by far the most common, but unambiguous losses of penile sacs and gains of female pregenital barriers mapped to the same branches and were necessarily interpreted as effectively simultaneous events. However, two species, Hadrobunus grandis and an undescribed Hadrobunus (H. n. sp. 3 IL) have both sacs and barriers, and one species, Leiobunum relictum, lacks both sacs and barriers. Depending on their exact phylogenetic positions, these species could represent either an intermediate stage in the transition from courtship to antagonism or a reversal from antagonism back to courtship. Although we regard a reversal to the sacculate condition per se as unlikely [52], the secondary loss of a pregenital barrier is more plausible. In fact, both L. relictum and the undescribed Hadrobunus n. sp. 2 MO show evidence of incipient or vestigial pregenital barrier structures that are fully developed in closely related taxa. The phylogenetic placement of these species requires corroboration by additional molecular data and analyses.

Explaining Evolutionary Change in Reproductive Structures

Our present work on the natural history and morphology of leiobunine harvestmen suggests an association between the type of precopulatory mechanism within a species and the duration of its breeding season. Specifically, species with potentially longer breeding seasons tend to have sacculate penes and other features consistent with female enticement by males, while species with shorter breeding seasons tend to have non-sacculate penes and traits associated with precopulatory antagonism. Tropical leiobunines have potentially long breeding seasons and virtually all species retain the ancestral conditions of sacculate penes, simple male pedipalps and unfortified female pregenital openings (J.W. Shultz, pers. obs.). Furthermore, males of these species are typically much smaller than females and tend to have short, poorly sclerotized penes with relatively large gift-bearing sacs. In contrast, species with features consistent with precopulatory antagonism (non-sacculate penes, enlarged male pedipalps, female pregenital barriers) are limited almost exclusively to north temperate regions (J.W. Shultz, pers. obs.), where breeding seasons are presumably limited by the onset of cold winters. Non-sacculate species overwinter as eggs and reach the final instar in mid-summer or later. Significantly, those populations with the most well-developed male palps and female pregenital barriers tend to occur on mountains (e.g., Leiobunum hoffmani and L. calcar) [54], where breeding seasons are likely to be short. There are also sacculate species in the north temperate region but most overwinter as immatures, attain adulthood in late spring and have potentially long breeding seasons (Fig. 4: "early-season" clade). There are exceptions to these patterns (e.g., L. aldrichi and L. politum are sacculate but mature in summer), and the precise onset of sexual maturity and duration of breeding seasons are unknown for all species. Additional research will be required to define the precise reproductive phenology for all eastern leiobunines, but these differences may be key to identifying the mechanism(s) by which reproductive structures have diversified. In light of these life history traits, multiple coevolutionary scenarios may be invoked to identify the origin and/or maintenance of reproductive morphology in the leiobunine harvestmen. We offer three hypotheses that may explain the association between male and female armaments observed across the phylogeny.