The Grass was Greener: Repeated Evolution of Specialized Morphologies and Habitat Shifts in Ghost Spiders Following Grassland Expansion in South America

The Grass was Greener: Repeated Evolution of Specialized Morphologies and Habitat Shifts in Ghost... Abstract While grasslands, one of Earth’s major biomes, are known for their close evolutionary ties with ungulate grazers, these habitats are also paramount to the origins and diversification of other animals. Within the primarily South American spider subfamily Amaurobioidinae (Anyphaenidae), several species are found living in the continent’s grasslands, with some displaying putative morphological adaptations to dwelling unnoticed in the grass blades. Herein, a dated molecular phylogeny provides the backbone for analyses revealing the ecological and morphological processes behind these spiders’ grassland adaptations. The multiple switches from Patagonian forests to open habitats coincide with the expansion of South America’s grasslands during the Miocene, while the specialized morphology of several grass-dwelling spiders originated at least three independent times and is best described as the result of different selective regimes operating on macroevolutionary timescales. Although grass-adapted lineages evolved towards different peaks in adaptive landscape, they all share one characteristic: an anterior narrowing of the prosoma allowing spiders to extend the first two pairs of legs, thus maintaining a slender resting posture in the grass blade. By combining phylogenetic, morphological, and biogeographic perspectives we disentangle multiple factors determining the evolution of a clade of terrestrial invertebrate predators alongside their biomes. Amaurobioidinae, anyphaenidae, biogeography, convergence, macroevolution, morphometrics To fully appreciate and understand the evolutionary history of a group of organisms, a holistic approach, considering intrinsic and extrinsic components, is crucial. Changes in the genetic material, morphological traits as well as biotic and abiotic selective pressures interact during the process of diversification of the world’s biota, in an intricate web of adaptation and co-adaptation. The adaptation of organisms to their specific habitats is one of the major driving forces of diversification, often reflected in specific physiological or morphological traits, especially when the association is strict (Wainwright and Reilly 1994). Habitats themselves are spatio-temporally dynamic, responding to changes in the planet’s geomorphology and climate. For example, the grassland biome currently covers an estimated 40% of the Earth’s terrestrial surface (Gibson 2009), and although grasses (Poaceae) are estimated to have existed since the early Cretaceous (Bouchenak-Khelladi et al. 2014), the expansion of the grassland biome did not occur until the Neogene, at different times depending on the continent (Edwards et al. 2010; Strömberg 2011), and probably due to a combination of shifts in climatic CO$$_{\mathrm{2}}$$ levels and fire regimes (Osborne 2008). The spread of grasslands has been postulated as the driver for the origin and diversification of numerous phytophagous insects (e.g., Dietrich 1999; Toussaint et al. 2012) and mammalian groups, including our own genus (e.g., Bobe and Behrensmeyer 2004). Furthermore, well-known cases of co-adaptation and even co-evolution have occurred between grasslands and ungulate grazers (e.g., Stebbins 1981; Jacobs et al. 1999; Bouchenak-Khelladi et al. 2009), although strict co-speciation has not always been the case (Jardine et al. 2012; Palazzesi and Barreda 2012; Strömberg et al. 2013). Besides trophic relations, a well-known driver of morphological evolution is the need to deceive the perception of predators or prey by crypsis or mimicry (see Stevens and Merilaita 2009), of which the melanic moths are a textbook example. Many invertebrates match the background color of the leaves or sticks on which they stand, and hide by precisely controlling the posture of the body and legs (Edmunds 1990), thus several spiders visually cryptic on grasses have thin, long bodies (Comstock 1913). For example, in the family Philodromidae, Tibellus oblongus evolved an elongated body and dwell within the blades of grasses extending their first two pairs of legs forwards and the rear two backwards (Mikulska 1967); similar characteristics can also be found among species of the crab spider genus Monaeses (Dippenaar-Schoeman 1984). These traits have been interpreted as adaptations for living on vegetation, including grasses. Grasslands are found on all major landmasses; for example, South America harbors large areas of grass-dominated ecosystems such as the Brazilian cerrados and campos, and the Argentine pampas (Soriano 1991) or lowland plains. Based on studies of fossil pollen records, the Patagonian region of South America, at present characterized by extensive areas of steppe, was covered mainly by forests (Barreda and Palazzesi 2007) until the Miocene (estimates range between 18.5 and 10 million years ago (Ma); Palazzesi and Barreda 2012; Strömberg et al. 2013), when grass-dominated open habitats began expanding. These changes in the floral composition, therefore, altered the availability of food sources and habitats for a large number of vertebrates and invertebrates in the region. The spider subfamily Amaurobioidinae (Anyphaenidae), currently with 180 species in 23 genera (World Spider Catalog 2017) is found almost exclusively in South America, with many of its genera restricted to the Patagonian region (Ramírez 2003). These spiders are active hunters that live inside silken retreats. Several species can be found living in grass-dominated open habitats, and a few remarkable morphological traits facilitate the spiders living and concealing themselves within the folded inner edge of the grass leaves. More specifically, this grass ecomorph has a more slender and elongated body configuration compared with that of all other members of the subfamily (Fig. 1), and adopts a resting posture along the leaf blade with the first two pairs of legs towards the front and two hind pairs towards the back (Fig. 1f); or, in some species, the third pair of legs is very short, and rests tightly folded besides the body (Fig. 1a). Their cryptic posture resembles that of other spiders and insects that live on elongate leaves, complementing their posture with a pattern of dots or lines matching the background (e.g., Mikulska 1967; Dippenaar-Schoeman 1984; Landry 1994). Figure 1. View largeDownload slide Spiders of the tribe Gayennini. The grass-dwelling species (top) have a slender constitution and a morphology that allows the legs to be tightly extended forwards and backwards, or folded at the side. Other species have a stockier constitution (bottom) and different resting positions. a) Monapia guenoana. b) Monapia charrua. c) Monapia angusta. d) Sanogasta tenuis. e) Sanogasta puma. f) Arachosia bergi. g) Oxysoma saccatum. h) Tomopisthes horrendus. i) Sanogasta maculosa. j) Arachosia proseni. k) Arachosia praesignis. Figure 1. View largeDownload slide Spiders of the tribe Gayennini. The grass-dwelling species (top) have a slender constitution and a morphology that allows the legs to be tightly extended forwards and backwards, or folded at the side. Other species have a stockier constitution (bottom) and different resting positions. a) Monapia guenoana. b) Monapia charrua. c) Monapia angusta. d) Sanogasta tenuis. e) Sanogasta puma. f) Arachosia bergi. g) Oxysoma saccatum. h) Tomopisthes horrendus. i) Sanogasta maculosa. j) Arachosia proseni. k) Arachosia praesignis. The amaurobioidine species known to display these specific morphological characteristics belong to three different genera, Arachosia, Monapia, and Sanogasta (Ramírez 1999, 2003; Rubio and Ramírez 2015), and the phylogenetic trees both from morphological (Ramírez 2003) and DNA data (Labarque et al. 2015) indicate that these putative adaptations have most likely arisen independently. The repeated evolution of similar morphologies as a response to similar environmental conditions, such as seen among several well-studied cases of morphological convergence (e.g., Hawaiian Tetragnatha spiders: Gillespie 2004; African cichlids: Muschick et al. 2012; Caribbean Anolis lizards: Mahler et al. 2013), is an impressive result of evolution under natural selection. Such cases also represent unique natural experiments in which the relative impact of contingency and determinism in evolutionary history can be assessed (Gould 1989; Conway Morris 2003; Turner 2011; Orgogozo 2015). Amaurobioidinae has been the focus of recent molecular phylogenetic studies in the form of a general, total-evidence study (Labarque et al. 2015) providing insight and confirmation of previous, morphology-based classifications (Ramírez 2003), as well as dated phylogenetic studies focusing on specific genera of the subfamily (Ceccarelli et al. 2016a; Soto et al. 2017). Herein, we present the most extensive sampling of Amaurobioidinae for molecular phylogenetic work to date, as well as the first instance of node age estimates for the whole subfamily. Furthermore, we use the dated molecular phylogeny for character mapping and morphospace analyses, to assess the impact of grassland expansions on morphological adaptations in the subfamily in terms of macroevolutionary mechanisms. In particular, we aim to answer the following general questions. 1) Was the origin of grass-dwelling clades in Amaurobioidinae concurrent with the habitat’s expansion in South America? 2) How frequent were the ecological shifts to and from open habitats, including grasslands, throughout the subfamily’s evolutionary history?, and 3) did the independent colonization of grasslands result in the repeated origin of similar morphologies? Through the outcomes of this study, it is possible to evaluate the plasticity in terms of adaptation to habitat changes at a macroevolutionary scale in this group of spiders. Materials and Methods Taxon Sampling For the molecular phylogenetic analyses 119 ingroup and 14 outgroup taxa were used. The ingroup taxa belong to 22 described genera and 1 undescribed genus of the subfamily Amaurobioidinae, while the outgroup taxa comprised nine genera belonging to the anyphaenid subfamily Anyphaeninae and one genus (Elaver), belonging to the family Clubionidae, to root the trees, following Wheeler et al. (2017). Of these taxa, 38 belonging to the subfamily Amaurobioidinae (plus one outgroup) were new to this study, while the remainder had been used before (Labarque et al. 2015; Ceccarelli et al. 2016a; Soto et al. 2017). For five of the previously used taxa, part of the DNA data that had been missing was obtained for this study as well. A list of the taxa used here can be found in Supplementary Table S1 of SI File 1 available on Dryad at http://dx.doi.org/10.5061/dryad.20257. For details on sequencing and phylogenetic inferences including node age estimates, readers are referred to the SI File 2 available on Dryad. Habitat and Area Mapping Geographic range, habitat type, and ecomorph were assigned as traits for each taxon in the dated phylogeny for further analyses. Because the most notable change in grassland expansions in southern South America occurred in what is today known as the Patagonian region (Palazzesi and Barreda 2012; Strömberg et al. 2013), the geographic ranges chosen for the analyses were limited to four categories: the first two, Patagonia (Pg) and non-Patagonian Neotropical (NP) cover the bulk of the diversity (see Fig. 2); Juan Fernández Archipelago (JF) covers the radiation of Philisca species (Soto et al. 2017) and the occurrence of Sanogasta maculosa “skottsbergi” (Ramírez 2003), and the trans-oceanic distribution in South Africa, Australia, and New Zealand (TO) covers in summary the species of the intertidal genus Amaurobioides that were already studied in detail (Ceccarelli et al. 2016a). For Pg, we used the Patagonian Andean region excluding the South American Transition zone as delimited by Morrone (2006), composed by the Argentinean Pg plus Southern and Central Chile (Fig. 2c). This area corresponds well with the distributional limits of many anyphaenid taxa (Ramírez 2003) and other organisms (Morrone 2015). Figure 2. View largeDownload slide Analysis of area and habitat occupancy through time in amaurobioidine spiders. a) Representatives of the three clades exhibiting a grass ecomorph. b) Dated phylogeny with ancestral areas and habitats mapped on nodes (black circles denote nodes with posterior probability $$\geqslant $$0.95). Yellow/light grey circles show the inferred transitions to grass ecomorph with highest posterior probability. c) Regionalization of the main areas of distribution of Amaurobioidinae analyzed in this study. d) Stacked profile of lineages through time, classified by habitat. NP $$=$$ Non-Patagonian Neotropical; Pg $$=$$ Patagonian; PgNP $$=$$ both Pg and NP; JF $$=$$ Juan Fernandez Archipelago; TO $$=$$ transoceanic (South Africa, Australia and New Zealand); Fo $$=$$ broad-leaf forest; Op $$=$$ open; OpFo $$=$$ both Op and Fo. Figure 2. View largeDownload slide Analysis of area and habitat occupancy through time in amaurobioidine spiders. a) Representatives of the three clades exhibiting a grass ecomorph. b) Dated phylogeny with ancestral areas and habitats mapped on nodes (black circles denote nodes with posterior probability $$\geqslant $$0.95). Yellow/light grey circles show the inferred transitions to grass ecomorph with highest posterior probability. c) Regionalization of the main areas of distribution of Amaurobioidinae analyzed in this study. d) Stacked profile of lineages through time, classified by habitat. NP $$=$$ Non-Patagonian Neotropical; Pg $$=$$ Patagonian; PgNP $$=$$ both Pg and NP; JF $$=$$ Juan Fernandez Archipelago; TO $$=$$ transoceanic (South Africa, Australia and New Zealand); Fo $$=$$ broad-leaf forest; Op $$=$$ open; OpFo $$=$$ both Op and Fo. Similarly, the general habitat type was either chosen as open (Op), referring mainly to grasslands, but also including other open habitats with few or no trees, such as scrublands, marshes, and rocky shores; or forest (Fo), mostly representing broad-leaf forests including foliage of shrubs. Because the two habitat types (Op and Fo) frequently co-exist as patches in many localities, and because some species use both habitats, the two variables (geographic range and habitat type) were treated as independent traits in separate analyses. Each terminal taxon in the maximum clade credibility (mcc) tree from BEAST was assigned the code or codes which best defined the species’ geographical range and general habitat type as documented in the taxonomic revisions, the MACN-Ar database (http://www.gbif.org/dataset/34e892b0-d9c7-11de-b793-b8a03c50a862) and field notes of author MJR. A detailed habitat analysis is circumscribed to the densely-sampled tribes Amaurobioidini $$+$$ Gayennini (A$$+$$G Clade); the outgroups, including the subfamily Anyphaeninae and the genus Josa (sister to A$$+$$G Clade) are all non-Patagonian and were sampled more superficially. Finally, species were coded as being of the grass ecomorph, or lumped into a “generalized morphology” category. As explained in the introduction, all species of the grass ecomorph have pale, elongated bodies, and in the case of Monapia and Sanogasta are collected exclusively on grasses such as Coleataenia prionitis, Cortaderia selloana, Paspalum spp., Panicum spp. (Supplementary Table S4 of SI File 1 available on Dryad). Grass inhabiting Arachosia dwell in the same grasses listed above, but are also found in other monocot plants with elongate leaves, such as Typha angustifolia, Eryngium species, sedges (Cyperaceae), and rushes (Juncaceae). To test whether the traits are clustered or randomly distributed on the phylogeny (i.e., show phylogenetic signal) Bayesian Tip-Significance testing was carried out on the last 5000 trees from each BEAST run in the program Befi-BaTS v.0.1.1 (Parker et al. 2008). This program uses three statistics, the Association Index (AI; Wang et al. 2001) the Fitch parsimony score (PS), and the Monophyletic Clade (MC) size statistic, to test the significance of phylogenetic signal in the traits against a null distribution in which the characters at the tips are reshuffled 1000 times. The ancestral trait estimation for areas and habitats were run on the BEAST mcc tree with BioGeoBEARS package v. 0.2.1 (Matzke 2012, 2013) for R v. 3.2.3 (R_Core_Team 2015) under three models: Dispersal-Extinction-Cladogenesis (DEC; Ree et al. 2005), DIVALIKE, modified from the DIVA program of Ronquist (1997), and BAYAREALIKE, modified from the BayArea program of Landis et al. (2013) algorithms. These three models factored to six after the addition of the founder-event speciation parameter J, and the best fit model was selected based on the Akaike information criterion (AIC). The BioGeoBEARS analyses were run using a script provided by N. Matzke (http://phylo.wikidot.com/biogeobears). The area analysis was stratified into two time periods: before and after 4 Ma, reflecting the emergence of the JF area (Stuessy et al. 1984). Consequently, the “areas allowed” and “dispersal multiplier” input matrices for the BioGeoBEARS run factored in a low likelihood for any occurrence in Juan Fernández before 4 Ma. The input matrices for “distance” and “dispersal multiplier” considered short distances within America, long distance to Juan Fernández, and longer distances to South Africa, Australia and New Zealand (see Supplementary Matrices S1 and S2 of SI File 1 available on Dryad). The ancestral state estimations corresponding to the model with the best fit for the two runs (area and habitat) were analyzed to obtain a profile of area and habitat occupancy through time for the ingroup (A$$+$$G Clade); the transitions in habitats and areas were also counted and analyzed for general trends. To obtain an estimate of the uncertainty of the results, we produced area and habitat analyses for each of 100 trees randomly drawn from the post-burnin posterior sample of the Bayesian analysis in BEAST, using the optimal model selected for the mcc tree in BioGeoBEARS. The time of transitions and ancestral state estimations for given time slices were calculated by linearly interpolating between the probabilities in dominant states in ancestor and descendant nodes (Supplementary scripts provided in SI File 3 available on Dryad). The expected number of lineages in each state for these given time slices was also calculated, under the assumption that the lineages in each state increased proportionally through time, and compared with the observed (average) numbers from the 100 trees by means of $$\chi^{\mathrm{2}}$$ and paired $$t$$-tests. For the evolution of ecomorphs, Bayesian stochastic mapping was carried out on 100 randomly selected post-burnin chronograms from BEAST, using the simmap algorithm (Bollback 2006) implemented in the R package phytools v. 0.6 (Revell 2012). Ten thousand Markov Chain Monte Carlo (MCMC) generations were run for 100 replicates on each tree under equal rates and the state changes over all stochastic maps summarized. A single stochastic map was also obtained by running 100 replicates, each with 10,000 MCMC generations on the A$$+$$G clade of the mcc tree obtained in BEAST, to get a graphical representation of the results. Morphological Macroevolution Analyses To explore the origin of the putative morphological adaptations associated with the exploitation of grasslands, four morphometric measurements were taken from museum specimens: total carapace length and width (a proxy for slim vs. stocky body), distance between the condyles of the chelicerae (a proxy for the ability to maintain a slender profile while extending the fore legs forward and close to each other), and length of the tibia of the third pair of legs (a proxy for the ability to tightly fold and hide the third legs at the sides of the body). Specimens were measured using a binocular microscope with an ocular micrometer. Given that putative grass-adapted species are restricted to the tribe Gayennini, morphometric data were only gathered for this clade. With the aim to minimize the effect of sexual dimorphism, only adult females were measured; as is common in spiders, immature anyphaenids throughout all free instars are similar to females in habitus and resting posture, while males have a slender constitution and sometimes cheliceral modifications (Soto and Ramírez 2012). In this way, the measured individuals had similar morphologies to those of the whole life cycle, while bearing the genital characters necessary for taxonomic identification. Whenever possible, multiple female specimens (maximum 5, mean 4.1) per species were measured, and trait values were expressed as the logarithm of the average of individual observations. The historical pattern of morphological change of these characters was studied using a version of the mcc tree pruned down to the 73 members of the tribe Gayennini for which morphological data could be gathered (Supplementary Table S5 of SI File 1 available on Dryad). First, character selection validation was carried out by testing for differences in the allometric scaling of these between grass-dwelling spiders and the rest, a common method to explore the existence of morphological adaptations in a comparative phylogenetic framework (Garland and Ives 2000). A phylogenetic generalized least squares regression (PGLS; Grafen 1989; Martins and Hansen 1997) was performed for all characters against carapace length, which was selected as a proxy for body size. Regressions were performed using R package caper (Orme et al. 2013) excluding species showing putative grassland adaptations, and simultaneously estimating the $$\lambda $$ parameter (Pagel 1999) to account for deviations from a pure Brownian motion (BM), as recommended by Revell (2010). A 95% prediction interval was then calculated for each regression using R scripts provided in Smaers and Rohlf (2016), and data points for grass-dwelling spiders were plotted to see whether they lie within the estimated intervals. Although this procedure does not represent, in a strict sense, a statistical test of deviations from allometric predictions (Smaers and Rohlf 2016), it was simply used here to determine whether the chosen characters were indeed capturing information relevant to the study of niche-associated morphological differentiation. Furthermore, the interactions among these four characters were explored by visualizing morphological variability and evolution using phylomorphospaces (Sidlauskas 2008) on phylogenetic principal component axes (pPCA; Revell 2009). Using the same dataset, the fit of multiple process-based models of macroevolution were explored in a multivariate framework using R package mvMORPH (Clavel et al. 2015). Such an approach to model phenotypic evolution outperforms commonly employed univariate methods given its capability to take into account the existence of correlations among traits. Furthermore, it allows to circumvent any pre-processing of the data, such as the use of principal component analysis or its phylogenetic variant, transformations that are known—under certain conditions—to introduce biases when used upstream of methods of evolutionary inference (Revell 2009; Monteiro 2013; Uyeda et al. 2015). However, recent simulation studies have shown that this method suffers from problems of model misspecification as the number of traits ($$p)$$ increases relative to the number of species ($$N)$$ (Adams 2014; Adams and Collyer 2018). To test whether the $$N$$:$$p$$ ratio of ~ 18:1 is enough to guarantee accurate model selection, 200 replicates of 4 dimensional trait matrices were simulated under BM using the empirical tree and phylogenetic variance-covariance matrix. Following Adams and Collyer (2018), a comparison of model fit between BM and Ornstein–Uhlenbeck (OU, see below) models using AIC scores revealed a percentage of model misspecification of only about 2%. Eight different models were tested, including three models in which morphological change was described using a single process for the entire tree, and five models with different processes dictating change in different regions of the tree. Among the first class, the fit of the following models were analyzed 1) a BM model, the most common null model of evolution in comparative studies, designed to represent a process of random walk with constant variance (described in detail in Freckleton et al. 2002; O’Meara et al. 2006); 2) an OU model, which adds to the random walk a tendency to evolve towards an “optimum value”, generally considered to represent a biological scenario of stabilizing selection (Hansen 1997; Butler and King 2004); and 3) an early burst (EB) model, in which the Brownian rate of morphological change decays exponentially with time (Blomberg et al. 2003). Furthermore, to explore the degree of morphological convergence shown by clades that independently colonized grasslands, five different multi-OU models were also tested. Such models constitute a convenient representation of evolution towards peaks in a macroevolutionary adaptive landscape (Ingram and Mahler 2013), and have become increasingly popular to describe cases of adaptive radiations and convergent evolution (Cressler et al. 2015; O’Meara and Beaulieu 2014; Cooper et al. 2016). All models included a single peak for all species with a general morphology, but differed in the number of peaks used to model the evolution among species of the grass ecomorph, including: 1) a single selective regime for all grass-dwelling species (OU2), 2) two different regimes, one exploited by a single genus and the other by the remaining two (OU3; all three different combinations were explored), and 3) a scenario in which all three grass-dwelling clades exploit different peaks (OU4). These models were specifically designed to test different levels of morphological convergence triggered by the expansion of grasslands, ranging from complete to non-existent (Supplementary Fig. S1 of SI File 4 available on Dryad). In all cases, measurement errors were incorporated in the estimation process (Supplementary Table S5 of SI File 1 available on Dryad), calculated as squared standard errors of the mean values (Clavel et al. 2015). For the small number of species coded from single specimens, errors corresponded to the averaged measurement error for each character. All runs were parameterized to assume a stationary distribution for the selective optimum at the root (option vcv $$=$$ “randomRoot” in function mvOU), corresponding to a scenario of evolution at a steady state (Clavel et al. 2015). This assumption is biologically reasonable in our case study, and has been found to stabilize estimates of optimum values (Cooper et al. 2016). In all cases, regimes were painted on the mcc tree using the make.simmap function in phytools, and the five multi-OU models were fitted to 100 different optimizations derived from stochastic character mappings under equal rates. Furthermore, one simulation was also explored in each of 100 trees randomly drawn from the posterior sample of the Bayesian analysis. In such a way, uncertainty in species trait values (Silvestro et al. 2015), timing and number of origins of the grass ecomorph (Huelsenbeck et al. 2003), divergence times and topology (Huelsenbeck et al. 2000) were all contemplated. To avoid problems of overfitting, which are known to impact multi-OU model selection (Boettiger et al. 2012; Ho and Ané 2014), models were compared using weights derived from sample-corrected ACC scores (wAICc; Burnham and Anderson 2002). In case of nested hypotheses, model support was further explored using log-likelihood ratio tests (LRTs). The tempo of the adaptive process was analyzed using phylogenetic half-lives, representing the time needed to move halfway from an ancestral state to an adaptive optimum, which in a multivariate framework are calculated on the eigenvalues of the selection ($$\alpha )$$ matrix (Bartoszek et al. 2012). Results Phylogenetic Inference and Node Age Estimates The characteristics of the four DNA data matrices used for phylogenetic analyses can be found in Supplementary Table S6 of SI File 1 available on Dryad. The percentage of missing data was 2.25% for the Cytochrome Oxidase I gene fragment (COI), 9.77% for 16S, 5.26% for H3, and 7.52% for 28S. The individual gene trees for each marker (Supplementary Figs. S2–S4 of SI file 4 available on Dryad) vary in their level of resolution, since, as would be expected, the nuclear markers have lower rates of nucleotide substitution (mean rates of substitutions per site per million years of 28S: 0.00243, H3: 0.00498, and mitochondrial loci: 0.0358), therefore resolving only the deeper divergences. The topologies obtained with the concatenated dataset for the Amaurobioidinae taxa in this study (Fig. 2, Supplementary Figs. S2–S8 of SI File 4 available on Dryad) largely agree with the concatenated molecular phylogenetic analyses of previous studies (Labarque et al. 2015; Ceccarelli et al. 2016a; Soto et al. 2017). Relationships for genera vary in the placement of terminals such as the monotypic genera Selknamia and Axyracrus, always with low support. Most genera are recovered as monophyletic in both Bayesian and maximum likelihood (ML) trees; although certain taxonomic inconsistencies remain, mostly due to complications of previous assignations of species to specific genera (see Soto et al. 2017). Species of Arachosia and Monapia displaying the grass ecomorph (A. avalosi, A. bergi, A. kapiipeoi and A. magna; M. angusta, M. charrua and M. guenoana) form reciprocally monophyletic groups in all estimations. The two species of grass-dwelling Sanogasta (S. puma and S. tenuis) are monophyletic only in the ML tree, although with low bootstrap support, while in the Bayesian analyses S. x-signata falls in between, with high support. Furthermore, Sanogasta as a whole is probably paraphyletic with regards to Arachosia, as already suggested by the morphological evidence (Ramírez 2003). While the lack of resolution for some groups may impede strong systematic conclusions (e.g., resolving the relationships among certain genera), the three clades of grass ecomorphs are separated from each other by several well supported nodes, thus confirming a scenario of convergent evolution. In order to account for the statistical uncertainty reflected by the low support of some nodes, all further analyses average over a subset of post-burnin trees (see below). The node age estimates obtained using fossil calibrations are similar to those based on the COI rate prior (tree not shown), as well as node ages from the previous studies focused on Philisca (Soto et al. 2017) and Amaurobioides (Ceccarelli et al. 2016a) (Supplementary Table S7 of SI File 1 available on Dryad), corroborating the node age estimates from the previous studies with the dates obtained here (Supplementary Fig. S4 of SI File 4 available on Dryad). In particular, the estimated dates of dispersal and speciation for the endemic species of the volcanic JF postdate the origin of the islands. We added two Philisca species not studied by Soto et al. (2017), as well as specimens of Sanogasta maculosa from Robinson Crusoe and Serlkirk islands; these last are denoted Sanogasta maculosa “skottsbergi”, because Sanogasta skottsbergi described from the archipelago is currently considered a synonym of S. maculosa (Ramírez 2003). Comparing with the previous studies, our more extensive sampling provides better estimates for the whole subfamily and most of its genera (Fig. 2). Greater accuracy in node age estimates is also expected for the genera with putative grass-adapted species, i.e., Arachosia, Monapia and Sanogasta, as these were also sampled more intensively. With regards to whether the three traits (geographic range, habitat and ecomorph) were clustered or randomly distributed throughout the phylogeny, both the AI and PS metrics detected significant phylogenetic signal in all cases (all $$P < 0.001$$). Accordingly, all states showed non-random distributions according to the MC statistic. However, phylogenetic signal for geographic range was found to be mostly driven by the non-Patagonian areas (MC NP: $$P< 0.001$$; MC Pg: $$P< 0.05$$), while grass-dwelling spiders contributed most of the clustering in the case of ecomorph (MC grass: $$P< 0.001$$; MC other: $$P < 0.05$$). Contributions were more balanced for habitat type (MC open: $$P < 0.01$$; MC forest: $$P< 0.05$$). Therefore, all the traits studied here are to some extent maintained through common descent, with the strongest phylogenetic signal found in species from non-Patagonian areas and those exhibiting a grass ecomorph. Habitat and Area Mapping The model selected by AIC for the area and habitat analyses with BioGeoBEARS was DEC$$+$$J, concentrating 96% and 90% of the AIC weights respectively (Supplementary Table S8 of SI File 1 available on Dryad). The ancestral state estimations (Fig. 2, Supplementary Figs. S9 and S10 of SI File 4 available on Dryad) and lineage profiles through time (Fig. 3) were thus calculated for the selected model. Figure 3. View largeDownload slide Analyses of ancestral habitats and areas for the A$$+$$G clade of amaurobioidine spiders, summarizing runs of 100 randomly selected post-burnin trees from the Bayesian analysis in BEAST. a) Count of lineages through time, by habitat (mean $$\pm$$ standard deviation). b) Count of lineages through time, by area (mean $$\pm$$ standard deviation). c) Habitat shifts grouped in time periods. d) Area shifts grouped in time periods, focusing on Patagonia, and the Neotropics. e) Shifts by combination of habitat and geographic area in South America (average over 100 Bayesian trees) summarized on a vegetation index map; arrow thickness is proportional to estimated number of shifts (image by Reto Stockli, NASA’s Earth Observatory Group, using data provided by the MODIS Land Science Team). Figure 3. View largeDownload slide Analyses of ancestral habitats and areas for the A$$+$$G clade of amaurobioidine spiders, summarizing runs of 100 randomly selected post-burnin trees from the Bayesian analysis in BEAST. a) Count of lineages through time, by habitat (mean $$\pm$$ standard deviation). b) Count of lineages through time, by area (mean $$\pm$$ standard deviation). c) Habitat shifts grouped in time periods. d) Area shifts grouped in time periods, focusing on Patagonia, and the Neotropics. e) Shifts by combination of habitat and geographic area in South America (average over 100 Bayesian trees) summarized on a vegetation index map; arrow thickness is proportional to estimated number of shifts (image by Reto Stockli, NASA’s Earth Observatory Group, using data provided by the MODIS Land Science Team). The ancestor of the subfamily Amaurobioidinae is likely to have arisen from a forest-dwelling Neotropical non-Patagonian lineage; although the outgroup sampling is sparse, no species of Josa or Anyphaeninae live in Pg, and most live in forest habitats. The most recent common ancestor of the A$$+$$G Clade colonized the Patagonian forests about 27 Ma ago (22.01–32 Ma 95% Highest Posterior Density, HPD), where it diversified. Up to approximately 14 Ma ago the increase in diversity was accounted for exclusively by forest-dwelling lineages in Pg (Fig. 3a and b). From that point on starts the increase of lineages in the open habitats (Figs. 2d, 3a) and in Neotropical regions outside Pg (Fig. 3b). Comparing the observed lineage state occupancies (based on 100 randomly-selected post-burnin chronograms from the BEAST analysis) with lineage state occupancies expected if the proportions had been the same throughout history, the mean overall state occupancy in open habitat and forest-dwellers is significantly different ($$\chi ^{\mathrm{2}}_{\mathrm{(66)}} =$$ 114.672, $$P< 0.001$$; see Supplementary Fig. S11a of SI File 4 available on Dryad), driven by the differences observed before 10 Ma ($$\chi^{\mathrm{2}}_{\mathrm{(44)}} =$$ 87.67, $$P< 0.001$$), while there was no significant difference from 10 Ma to the present ($$\chi ^{\mathrm{2}}_{\mathrm{(20)}} =$$ 27.0, $$P =$$ 0.135). This result reflects the occupancy of the open habitat, which is only significantly lower than expected before 10 Ma ($$t$$-test $$t =$$ 4.996; $$P< 0.001$$; Supplementary Fig. S11b of SI File 4 available on Dryad), while the forest occupancy is higher than expected by chance throughout the history of the spiders ($$t = -$$5.407; $$P< 0.001$$; Supplementary Fig. S11b of SI File 4 available on Dryad). Similarly, there is a significant difference between observed and expected state occupancies through time in Pg ($$t =$$$$-$$5.627, $$P< 0.001$$), whereas state occupancy in non-Patagonian areas was significantly lower than expected before 10 Ma ($$t =$$ 3.907, $$P< 0.001$$; Supplementary Fig. S12 of SI File 4 available on Dryad). The overall temporal pattern is that of an initial source of lineages coming from the Patagonian forests, as shown in the earlier start of shifts from forest to open habitats (Fig. 3c), and from Patagonian to NP areas (Fig. 3d). The process becomes dynamic shortly after, with a gradual increase in the reverse transitions near recent times, especially in habitat shifts. The relatively high number of backward shifts in recent times from open to forest habitats is mainly due to generalist species that have come to use both habitats (Supplementary Table S9 of SI File 1 available on Dryad). In comparison, the backward shifts to Pg are very few (Supplementary Table S10 of SI File 1 available on Dryad), and were found to be significantly fewer than the number of movements out of this region (mean $$\pm$$ standard deviation of 100 random posterior BEAST trees, Pg$$\to $$NP and Pg $$\to $$ PgNP $$=$$ 7.88 $$\pm$$ 0.72, NP$$\to $$Pg and NP$$\to $$PgNP $$=$$ 4.49 $$\pm$$ 1.00, $$t$$$$= -$$20.707, $$P< 0.001$$). A dissection of the number of transitions by combinations of habitat and geographic area (averaged over 100 Bayesian trees; Fig. 3e, Supplementary Tables S11 and S12 of SI File 1 available on Dryad) reveals a strong asymmetry in habitat changes within Pg (10 shifts Op $$\to $$ Fo, vs. 5 Fo $$\to $$ Op) compared with the ones that occurred in the non-Patagonian Neotropics (1 vs. 5, respectively). It is also evident that more events of dispersal between these biogeographic regions occurred in open habitats, rather than to/from forests. Origin of Grass-Adapted Species Based on the results of this study, clades showing putative adaptations to grasslands arose from ancestors already living in open habitats in the NP region. Bayesian stochastic mapping on 100 random BEAST trees and the mcc tree (100 replicates in both cases) detected an average of 4.44 and 4.38 state changes across the phylogenies, respectively. These changes were divided as an average of 3.98 (100 trees) and 3.99 (mcc tree) transitions from a general morphology to a grass ecomorph and 0.46 (100 trees) and 0.39 (mcc tree) transitions in the opposite direction. In terms of integers this translates as four transitions to a specialized morphology and zero reversals (see Supplementary Fig. S13 of SI File 4 available on Dryad), although in a few simulations one or more transitions from grass ecomorph to general morphology were detected, most of them apparently accounted for by the ambiguity introduced by the inclusion of S. x-signata inside the clade containing the grass-dwelling Sanogasta. In the case of Arachosia, the genus’ most recent common ancestor (MRCA) is estimated to have diverged 10.75 Ma (8.14–13.6 Ma, 95% HPD) and probably inhabited open habitats, while the grass ecomorph species diversified approximately 1.98 Ma (1.16–2.92, 95% HPD). At ca. 4.07 Ma (2.58–5.8, 95% HPD), the MRCA of these species and their sister-group, Arachosia proseni (Fig. 1j), diverged; A. proseni being a rather flat and greenish species that inhabits bromeliad-like Eryngium species. The second case of putative grass-adaptation is found in Sanogasta; the clade containing the two grass ecomorph species Sanogasta tenuis and S. puma diverged approximately 5 Ma (3.4–6.84, 95% HPD), while their ancestors inhabited open areas from approximately 8.51 Ma (6.36–10.89, 95% HPD). Lastly, while the group of Monapia using open habitats traces back to ca. 9.49 Ma (7.12–12.01, 95% HPD), the clade containing the grass ecomorph Monapia species diversified at about the same time as the grass ecomorph Sanogasta species, around 6.82 Ma (4.70–9.09, 95% HPD). Morphological Macroevolution Analyses Several aspects of the morphology of grass ecomorph spiders were found to diverge from the allometric expectations based on data from the remaining amaurobioidines (Fig. 4, top). Most notably perhaps is the reduced distance separating the chelicerae, the only trait that is found to have arisen in at least some members of all three grass-dwelling clades. On the other hand, members of the genera Monapia and Sanogasta seem to have convergently acquired a reduced body size with a slender carapace morphology, while the length of the third pair of legs seems to be beyond the predicted interval only for grass-dwelling species of Sanogasta. The exploration of this dataset with pPCA reveals a similar pattern (Fig. 4, bottom). The morphology acquired by different clades specialized to life on grasses seems to show some component of convergence, as revealed especially by pPCA axis 3 (Fig. 4, bottom right) which shows the clustering of all three clades and a clear distinction with respect to all other members of the A$$+$$G clade. Species with higher values of pPCA axis 3 tend to have narrower and elongated bodies, with a shorter third pair of legs and a reduced distance between the chelicerae (loadings matrix can be found in Supplementary Table S13 of SI File 1 available on Dryad). This specialized morphology evolved independently in each one of the three lineages from ancestors that exhibited a generalist morphology. However, it is also evident from the remaining axes that grass-dwelling clades have explored partially different regions of morphospace. Figure 4. View largeDownload slide Variability among morphometric traits associated with the colonization of grasslands. Top, allometric scaling of traits, using carapace length as proxy for body size. Values are expressed as the logarithm of the species mean. Solid lines correspond to a PGLS$$_{\mathrm{\lambda }}$$ regression for species of general morphology (black circles), dotted lines correspond to the 95% prediction interval (all $$P< 10^{{-16}}$$; from left to right: $$R^{\mathrm{2}} =$$ 0.98, 0.91, 0.93; $$\lambda =$$ 0.512, 0.836, 0.918). At least one grass-dwelling species (other shapes) lies outside the expected interval in all cases. Bottom, phylomorphospaces on phylogenetic PCA axes, built using the same four traits. Axis pPCA 1 captures differences in size, with low values to the left. The rest represent differences in body proportions (see Supplementary Table S13 of SI File 1 available on Dryad). Figure 4. View largeDownload slide Variability among morphometric traits associated with the colonization of grasslands. Top, allometric scaling of traits, using carapace length as proxy for body size. Values are expressed as the logarithm of the species mean. Solid lines correspond to a PGLS$$_{\mathrm{\lambda }}$$ regression for species of general morphology (black circles), dotted lines correspond to the 95% prediction interval (all $$P< 10^{{-16}}$$; from left to right: $$R^{\mathrm{2}} =$$ 0.98, 0.91, 0.93; $$\lambda =$$ 0.512, 0.836, 0.918). At least one grass-dwelling species (other shapes) lies outside the expected interval in all cases. Bottom, phylomorphospaces on phylogenetic PCA axes, built using the same four traits. Axis pPCA 1 captures differences in size, with low values to the left. The rest represent differences in body proportions (see Supplementary Table S13 of SI File 1 available on Dryad). The relative support for the explored models of macroevolution are shown in Fig. 5. Overall, there was a relatively strong support for a 3 peak multi-OU model, favored across multiple trait histories on the mcc tree (Fig. 5, top) and multiple topologies sampled from the posterior distribution (Fig. 5, bottom). This macroevolutionary scenario (OU3 A–MS) models the history of morphological change in the tribe Gayennini using three multivariate adaptive peaks, with one peak assigned to all species displaying a generalist morphology, one occupied by grass-dwelling species of Arachosia, and a third one convergently encountered by grass-dwelling species of Monapia and Sanogasta. The difference in support between this model and the second most supported one (OU3 S–AM, also a multi-OU model with three peaks), was higher than the threshold value of 4 (Burnham and Anderson 2002) generally considered to represent a strong difference in support (average $$\Delta $$AICc $$=$$ 4.03 for replicates on the mcc tree, 4.22 for those in the sample of posterior topologies). Moreover, across all 200 replicates explored, we found that the addition of a third adaptive regime significantly improved the fit of the model 87% of the times, while the addition of a fourth one only resulted in a significant increase in fit 11% of the times (in both cases, $$P< 0.05$$ for LRT with four degrees of freedom). This further confirms that the repeated colonization of grasslands by spiders of the tribe Gayennini constitutes a scenario of partial convergence, with two different grassland-associated morphologies, one of which has evolved multiple times. Figure 5. View largeDownload slide Histograms of the relative support for each of the eight models of macroevolution tested. Support was measured as AICc weights in 100 replicates of stochastic character mapping histories on the mcc tree (top) and in one replicate in 100 randomly drawn topologies from the posterior sample of trees (bottom). In case of OU3 models, capital letters stand for grass-dwelling genera (A $$=$$Arachosia; M $$=$$Monapia; S $$=$$Sanogasta), and genera on the same side of the dash were modelled using the same selective regime. Values represent as median $$\pm$$ standard error. Figure 5. View largeDownload slide Histograms of the relative support for each of the eight models of macroevolution tested. Support was measured as AICc weights in 100 replicates of stochastic character mapping histories on the mcc tree (top) and in one replicate in 100 randomly drawn topologies from the posterior sample of trees (bottom). In case of OU3 models, capital letters stand for grass-dwelling genera (A $$=$$Arachosia; M $$=$$Monapia; S $$=$$Sanogasta), and genera on the same side of the dash were modelled using the same selective regime. Values represent as median $$\pm$$ standard error. A summary of the inferred optimal values for each trait (i.e., the $$\theta $$ parameters of the multi-OU model) for the three adaptive peaks can be found in Fig. 6. The morphological evolution of grass-dwelling Monapia and Sanogasta is best modelled as having suffered a tendency towards body size reduction, narrowing of the carapace and strong reduction of the distance between chelicerae, as well as a mild decrease in the length of the third pair of legs (although the confidence interval on this trait is very large). On the other hand, the adaptive peak explored by grass-dwelling Arachosia differs much less from the ancestral peak, and seems to entail a slight tendency towards body size increase and, most notably, a strong reduction in the distance between chelicerae. This last attribute therefore characterizes all three clades that have specialized in the exploitation of grassland niches. Figure 6. View largeDownload slide Inferred optimum values ($$\theta $$ parameters) for all traits in 200 replicates of the OU3 A-MS model. Values can also be interpreted as the approximate locations of the peaks in a multivariate adaptive landscape. Violin plots show the probability densities of the estimates after excluding the most extreme 5% of results, rectangles represent median values. Body size optima are shown in mm, other values are expressed as the ratio of each trait to body size, so as to represent inferred differences in proportions. Given the way the replicates were parameterized, uncertainty for the ancestral selective regime is minimal and not shown. Peaks for grass-dwelling clades are identified using the same codes for genera as in Fig. 5. Figure 6. View largeDownload slide Inferred optimum values ($$\theta $$ parameters) for all traits in 200 replicates of the OU3 A-MS model. Values can also be interpreted as the approximate locations of the peaks in a multivariate adaptive landscape. Violin plots show the probability densities of the estimates after excluding the most extreme 5% of results, rectangles represent median values. Body size optima are shown in mm, other values are expressed as the ratio of each trait to body size, so as to represent inferred differences in proportions. Given the way the replicates were parameterized, uncertainty for the ancestral selective regime is minimal and not shown. Peaks for grass-dwelling clades are identified using the same codes for genera as in Fig. 5. Most of the temporal dynamics of this adaptive process seems to have occurred at a relatively fast pace, with average phylogenetic half-lives of 0.71, 1.32, 2.97, and 131.03, for each of the four eigendimensions of the $$\alpha $$ matrix (Supplementary Fig. S10 of the SI File 4 available on Dryad). The first of these dimensions entails the relative reduction of all traits with respect to size (the same pattern as seen in pPCA axis 3), while the second one can be interpreted as changes in overall size (Supplementary Fig. S15 of the SI File 4 and Table S14 of SI File 1 available on Dryad). Discussion The global expansion of grasslands during the Miocene has impacted numerous animal taxa. For grass-specialized herbivores, the increase in their food source presented an ecological and evolutionary benefit; this is especially true for insect groups such as the leafhoppers and stem-borers, many of which have specialized in feeding on grasses (Dietrich 1999; Zahniser and Dietrich 2010; Toussaint et al. 2012) and whose diversification is temporally congruent with grassland expansions. With the increase in grassland herbivores and new habitats to exploit, predators such as falcons also underwent time-congruent diversification in the late Miocene (Fuchs et al. 2015). Similarly, the predators among arthropods such as spiders are expected to exploit the abundant insect biomass found in grasslands (Tscharntke and Greiler 1995). Based on the results of this study, the habitat changes from forest to open habitats in amaurobioidine spiders occurred during the mid- to late Miocene, before which (ca. 10 Ma) there were proportionally fewer lineages than expected in open habitats. These dates are congruent with the increased expansion of the South American grassland biome, especially in Pg (Barreda and Palazzesi 2007; Palazzesi and Barreda 2012) suggesting that, as open habitats started expanding, amaurobiodine spiders readily occupied the newly available niches. The node age estimates obtained in this study are plausible, based on the similar ages obtained from previous studies with different taxon sampling (Ceccarelli et al. 2016a; Soto et al. 2017) as well as geological age corroboration with the volcanic JF (Stuessy et al. 1984) and the age of the MRCAs of the Sanogasta and Philisca species endemic to those islands. These two colonization events of Juan Fernández, plus the trans-oceanic dispersal of Amaurobioides probably by oceanic drift with the Antarctic circumpolar current (Ceccarelli et al. 2016a) are evidence for the long-distance dispersal ability of some amaurobioidine spiders, yet overall, a strong historical component is still preserved in their geographic distribution. While the Amaurobioidinae are probably of NP origin, the A$$+$$G clade originated and diversified in forested Patagonian habitats. South American orogeny must also be considered as an important factor for the diversification and movement of species, and for the evolution of habitats; for example, during the last 10 Ma the Andean uplift also had a profound effect on structuring both Patagonian and other South American fauna (e.g., Sersic et al. 2011; Ceccarelli et al. 2016b). Based on our study, the interchanges of fauna between Pg and the non-Patagonian Neotropics occurred predominantly in open habitats. This is consistent with the prediction that biome shifts are more probable over larger shared boundaries (Donoghue and Edwards 2014), since the boundary between these biogeographic areas is constituted by open habitats, with the forested areas now separated by ample expanses of arid lands (Fig. 3e). It is also noteworthy that the initial flow of species out of Patagonian forests became dynamic shortly after the expansion of grasslands, with a backflow of lineages returning to Pg, or expanding back to forest habitats. Some species of Amaurobioidinae seem especially suited for life on grasses due to morphologies that evolved at least three independent times among lineages already inhabiting open ecosystems in the non-Patagonian Neotropics. Although grasses are common in the understory of forests, grass-dwelling lineages originated exclusively in open habitats, likely a consequence of the difference in prey biomass harbored by grasslands compared with forest grasses. The traits involved in this specialization are all related with the need to fit tightly on a single blade of grass and minimize visual cues to predators, moving and building retreats in the narrow spaces between grass blades, and might therefore be considered morphological adaptations to the peculiar shape and disposition of the graminean leaves. The association between these morphologies and the strict exploitation of grasslands, their departure from the allometric scaling shown by close relatives living in different habitats, as well as their inferred origin only after grass-dominated ecosystems spread all warrant their recognition as adaptations to grasslands. Many shifts from forest to open habitats are inferred to have occurred at about the same time as grasslands were expanding, which in conjunction with the strong phylogenetic signal found in the habitat traits, suggest that there is an intermediate degree of niche conservatism (see Wiens et al. 2010). On the other hand, the strict adaptation to living on grasses is a much more conservative trait, and in at least two cases seems to have originated only long after the colonization of open habitats. The MRCA of grassland-adapted Arachosia and Sanogasta is inferred to have already lived in open habitats some 13 Ma, while adaptations to grasslands are unlikely to have originated any time before 6 Ma; in Monapia the delay is less marked (about 10 Ma for shift to open habitat and 7 Ma for grass-adaptation). This delay in the start of the process of grassland specialization might indicate that the specific microhabitats employed by grass-dwelling ghost spiders—grasslands dominated by large grass species in regions of high humidity and mild temperatures—did not become common until later in the history of the region. The current paleoenvironmental reconstructions in the region (e.g., Barreda et al. 2007) are still too coarse to trace the origin of this specific environment. Nonetheless, based on our results, grassland ecomorphs originated very rapidly once the selective process was underway. The half-lives for the first two eigendimensions, which seem to capture the transition from general morphology to grass ecomorph, as well as changes in overall size, are on the timescale of approximately 1 Ma. This is a relatively short amount of time compared with the results of other studies inferring selective processes on macroevolutionary timescales (Uyeda et al. 2011; Hansen 2012; Toljagić et al. 2018), although the difference in generation times between taxa must also be considered. This rapid adaptation is even more remarkable considering the complexity of the morphological changes involved, which likely involved the accommodation of multiple constraints and trade-offs. For example, it has been shown that the slender abdomen of ant-mimicking spiders increases their resemblance to their model ants at the expense of smaller egg clutches compared with other, stockier salticid spiders (see Cushing 1997). The expansion of grasslands in South America resulted in an increase in complexity of the macroevolutionary adaptive landscape of amaurobioidine spiders, setting them off on previously unexplored evolutionary pathways. The morphology of two of the three genera of grass-adapted spiders is best modelled as the result of evolution under the same selective regime, characterized by body size reduction, narrowing of the carapace, strong reduction of the distance between chelicerae, as well as a mild decrease in the length of the third pair of legs. This can therefore be considered an instance of adaptive morphological convergence driven by habitat, a thoroughly-studied phenomenon in evolutionary biology (e.g., Gillespie 2004; Losos 2011; Muschick et al. 2012; Thiele 2012). Nonetheless, the evolution of a second grassland ecomorph shows that not all evolutionary trajectories associated with grassland specialization have resulted in the same end-product. Grass-adapted species of Arachosia have evolved towards a different adaptive peak, characterized by a longer body and a reduced distance between chelicerae. The fact that replicated events of adaptation to similar environmental conditions fail to converge is considered evidence of the importance of historical contingency in evolutionary processes (Seehausen 2007; Blount et al. 2008; Young et al. 2009). On the other hand, although there is strong statistical support for grassland-adapted amaurobioidine spiders evolving under two different selective regimes, they are both characterized by a common trend: a reduction of the distance between the chelicerae. This slender configuration at the anterior region of the prosoma, which originated at least three independent times, seems to be a necessary condition for the successful colonization of grasslands. At least at this level, adaptation has followed an entirely deterministic route. In conclusion, based on the results of our analyses, it is clear that the evolutionary history of Amaurobioidinae has been shaped by recurrent colonization of open habitats, followed by events of diversification and adaptation. Even considering some of the shortcomings of the phylogenies upon which these conclusions are based, such as poor resolution in some clades, the nodes corresponding to grass-adapted clades are all well-supported. Furthermore, sampling over a subset of posterior trees allowed us to test these hypotheses in light of alternative topologies. We can therefore conclude that these spiders are a group of organisms for which the expansion of grasslands in South America presented an ecological and evolutionary opportunity. The evolution of amaurobioidine spiders shows a traceable history of convergent, yet diverse morphological adaptations in response to changes in the environment, the colonization of new habitats and regions and a dynamic flow of lineages. We have shown here that a combination of phylogenetic, morphological, and biogeographic analyses helps to untangle and understand the multiple factors and levels affecting the evolution of organisms and the biomes they inhabit. Supplementary Material Data available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.20257. Funding This work was supported by the Agencia Nacional de Promoción Científica y Tecnológica Argentina (ANPCyT, grant numbers PICT 2011-1007, PICT 2015-283), and the International Barcode of Life Project (iBOL) Argentina 2012 from the National Scientific and Technical Research Council (CONICET) Argentina. Additional funds were provided by the Spanish Ministerio de Ciencia e Innovación (MA, grant number CGL2009-07639) and from The National Geographic Society (Gustavo Hormiga, MA, grant number 8969-11). Acknowledgments We thank the MACN Barcoding Laboratory for the amplification of DNA and to the Canadian Centre of DNA Barcoding for the sequencing of some of the COI samples used here. Collection permits were provided by Corporación Nacional Forestal (Chile), Administración de Parques Nacionales and Secretaría de Desarrollo Sustentable y Ambiente de Tierra del Fuego (Argentina). Guillermo Martínez Pastur and Vanessa Lencinas of the Centro Austral de Investigaciones Científicas (CADIC) provided logistic support for collecting in Tierra del Fuego, Federico Gerrero and Laura Smith (Quixote Expeditions) generously brought MJR to Isla de los Estados in the Ocean Trump vessel, Jaime Pizarro Araya (Universidad de La Serena) provided support for fieldwork in Chile. Additional specimens were provided by Charles Griswold (California Academy of Sciences) and Ivan Magalhães (MACN). Lorena Zapata helped with the identification and preparation of Sanogasta specimens, and Gonzalo Rubio with those of Arachosia. Ivan Magalhães, Abel Pérez González, Andrés Ojanguren provided useful comments in early versions of this manuscript. Julien Clavel and Thomas F. Hansen provided useful guidance relating to macroevolutionary model fitting and interpretation. References Adams D.C 2014 . Quantifying and comparing phylogenetic evolutionary rates for shape and other high-dimensional phenotypic data . Syst. Biol. 63 2 : 166 – 177 . Google Scholar CrossRef Search ADS PubMed Adams D.C., Collyer M. L 2018 . Multivariate phylogenetic comparative methods: evaluations, comparisons, and recommendations . Syst. Biol. 67 1 : 14 – 31 . Google Scholar CrossRef Search ADS PubMed Barreda V., Anzótegui L.M., Prieto A.R., Aceñolaza P., Bianchi M.M., Borromei A.M., Brea M., Caccavari M., Cuadrado G.A., Garralla S., Grill S., Guerstein G.R., Lutz A.I., Mancini V., Mautino L.R., Ottone E.G., Quattrocchio M.E., Romero E.J., Zamaloa M.C., Zucol A 2007 . Diversificación y cambios de las angiospermas durante el Neógeno en Argentina . Publicación Especial 11 ISSN 0328-347X Ameghiniana 50° aniversario. 11 : 173 – 191 . Barreda V., Palazzesi L 2007 . Patagonian vegetation turnovers during the Paleogene-early Neogene: origin of arid-adapted floras . Bot. Rev. 73 1 : 31 – 50 . Google Scholar CrossRef Search ADS Bartoszek K., Pienaar J., Mostad P., Andersson S., Hansen T.F 2012 . A phylogenetic comparative method for studying multivariate adaptation . J. Theor. Biol. 314 : 204 – 215 . Google Scholar CrossRef Search ADS PubMed Blomberg S.P., Garland T Jr. , Ives A.R . 2003 . Testing for phylogenetic signal in comparative data: behavioral traits are more labile . Evolution 57 4 : 717 – 745 . Google Scholar CrossRef Search ADS PubMed Blount Z.D., Borland C.Z., Lenski R.E 2008 . Historical contingency and the evolution of a key innovation in an experimental population of Escherichia coli . Proc. Natl Acad. Sci. USA 105 23 : 7899 – 7906 . Google Scholar CrossRef Search ADS Bobe R., Behrensmeyer A.K 2004 . The expansion of grassland ecosystems in Africa in relation to mammalian evolution and the origin of the genus Homo . Palaeogeogr. Palaeoclimatol. Palaeoecol. 207 : 399 – 420 . Google Scholar CrossRef Search ADS Boettiger C., Coop G., Ralph P 2012 . Is your phylogeny informative? Measuring the power of comparative methods . Evolution 66 7 : 2240 – 2251 . Google Scholar CrossRef Search ADS PubMed Bollback J.P 2006 . SIMMAP: stochastic character mapping of discrete traits on phylogenies . BMC Bioinformatics 7 : 88 . Google Scholar CrossRef Search ADS PubMed Bouchenak-Khelladi Y., Muasya A.M., Linder H.P 2014 . A revised evolutionary history of Poales: origins and diversification . Bot. J. Linn. Soc. 175 1 : 4 – 16 . Google Scholar CrossRef Search ADS Bouchenak-Khelladi Y., Verboom G.A., Hodkinson T.R., Salamin N., Francois O., Ni Chonghaile G., Savolainen V 2009 . The origins and diversification of C4 grasses and savanna-adapted ungulates . Glob. Change Biol. 15 10 : 2397 – 2417 . Google Scholar CrossRef Search ADS Burnham K.P., Anderson D.R 2002 . Model selection and multimodel Inference . New York : Springer . Butler M.A., King A.A 2004 . Phylogenetic comparative analysis: a modeling approach for adaptive evolution . Am. Nat. 164 6 : 683 – 695 . Google Scholar CrossRef Search ADS PubMed Ceccarelli F.S., Ojanguren-Affilastro A.A., Ramírez M.J., Ochoa J.A., Mattoni C.I., Prendini L 2016b . Andean uplift drives diversification of the bothriurid scorpion genus Brachistosternus . J. Biogeogr. 43 10 : 1942 – 1954 . Google Scholar CrossRef Search ADS Ceccarelli F.S., Opell B.D., Haddad C.R., Raven R.J., Soto E.M., Ramírez M.J 2016a . Around the world in eight million years: historical biogeography and evolution of the spray zone spider Amaurobioides (Araneae: Anyphaenidae) . PloS One 11 10 : e0163740 . Google Scholar CrossRef Search ADS Clavel J., Escarguel G., Merceron G 2015 . mvMORPH: An R package for fitting multivariate evolutionary models to morphometric data . Method Ecol. Evol. 6 11 : 1311 – 1319 . Google Scholar CrossRef Search ADS Comstock J.H 1913 . The spider book. Doubleday, Page , New York, NY, USA : Company . Conway Morris S 2003 . Life’s solution: inevitable humans in a lonely universe . Cambridge : Cambridge University Press . Google Scholar CrossRef Search ADS Cooper N., Thomas G.H., Venditti C., Meade A., Freckleton R.P 2016 . A cautionary note on the use of Ornstein–Uhlenbeck models in macroevolutionary studies . Biol. J. Linn. Soc. 118 1 : 64 – 77 . Google Scholar CrossRef Search ADS Cressler C.E., Butler M.A., King A.A 2015 . Detecting adaptive evolution in phylogenetic comparative analysis using the Ornstein–Uhlenbeck model . Syst. Biol. 64 6 : 953 – 968 . Google Scholar CrossRef Search ADS PubMed Cushing P.E 1997 . Myrmecomorphy and myrmecophily in spiders: a review . Fla Entomol. 80 : 165 – 193 . Google Scholar CrossRef Search ADS Dietrich C.H 1999 . The role of grasslands in the diversification of leafhoppers (Homoptera: Cicadellidae): a phylogenetic perspective. In Proceedings of the Fifteenth North American Prairie Conference (pp. 44 – 49 ). Bend, OR : Natural Areas Assoc . Dippenaar-Schoeman A.S 1984 . The crab-spiders of southern Africa (Araneae: Thomisidae) . 4. The genus Monaeses Thorell, 1869. Phytophylactica 16 : 101 – 116 . Donoghue M.J., Edwards E.J 2014 . Biome shifts and niche evolution in plants . Annu. Rev. Ecol. Evol. Syst. 45 : 547 – 572 . Google Scholar CrossRef Search ADS Edmunds M 1990 . The evolution of cryptic coloration. In: Insect defenses: adaptive mechanisms and strategies of prey and predators . Evans D.L., Schmidt J.O Eds. Albany, N.Y., USA : State University of New York Press . p. 3 – 21 . Edwards E.J., Osborne C.P., Strömberg C.A., Smith S.A ; C4 Grasses Consortium . 2010 . The origins of C4 grasslands: integrating evolutionary and ecosystem science . Science 328 5978 : 587 – 591 . Google Scholar CrossRef Search ADS PubMed Freckleton R.P., Harvey P.H., Pagel M 2002 . Phylogenetic analysis and comparative data: a test and review of evidence . Am. Nat. 160 6 : 712 – 726 . Google Scholar CrossRef Search ADS PubMed Fuchs J., Johnson J.A., Mindell D.P 2015 . Rapid diversification of falcons (Aves: Falconidae) due to expansion of open habitats in the Late Miocene . Mol. Phylogenet. Evol. 82 : 166 – 182 . Google Scholar CrossRef Search ADS PubMed Garland T Jr. , Ives A.R . 2000 . Using the past to predict the present: confidence intervals for regression equations in phylogenetic comparative methods . Am. Nat. 155 3 : 346 – 364 . Google Scholar CrossRef Search ADS PubMed Gibson D.J 2009 . Grasses and grassland ecology . New York : Oxford University Press p. 305 . Gillespie R 2004 . Community assembly through adaptive radiation in Hawaiian spiders . Science 303 : 356 – 359 . Google Scholar CrossRef Search ADS PubMed Gould S.J 1989 . Wonderful life: the burgess shale and the nature of life . New York : W.W. Norton . Grafen A 1989 . The phylogenetic regression . Philos. Trans. R. Soc. Lond. B Biol. Sci. 326 1233 : 119 – 157 . Google Scholar CrossRef Search ADS PubMed Hansen T.F 1997 . Stabilizing selection and the comparative analysis of adaptation . Evolution 51 : 1341 – 1351 . Google Scholar CrossRef Search ADS PubMed Hansen T.F 2012 . Adaptive landscapes and macroevolutionary dynamics . In: Svensson E., Calsbeek R , editors. The adaptive landscape in evolutionary biology . Oxford, UK : Oxford University Press . p. 205 – 226 . Google Scholar CrossRef Search ADS Ho L.S.T., Ané C 2014 . Intrinsic inference difficulties for trait evolution with Ornstein–Uhlenbeck models . Method Ecol. Evol. 5 11 : 1133 – 1146 . Google Scholar CrossRef Search ADS Huelsenbeck J.P., Nielsen R., Bollback J.P 2003 . Stochastic mapping of morphological characters . Syst. Biol. 52 2 : 131 – 158 . Google Scholar CrossRef Search ADS PubMed Huelsenbeck J.P., Rannala B., Masly J.P 2000 . Accommodating phylogenetic uncertainty in evolutionary studies . Science 288 5475 : 2349 – 2350 . Google Scholar CrossRef Search ADS PubMed Ingram T., Mahler D.L 2013 . SURFACE: detecting convergent evolution from comparative data by fitting Ornstein–Uhlenbeck models with stepwise Akaike Information Criterion . Method Ecol. Evol. 4 5 : 416 – 425 . Google Scholar CrossRef Search ADS Jacobs B.F., Kingston J.D., Jacobs L.L 1999 . The origin of grass-dominated ecosystems . Ann. Missouri Bot. Garden 86 : 590 – 643 . Google Scholar CrossRef Search ADS Jardine P.E., Janis C.M., Sahney S., Benton M.J 2012 . Grit not grass: concordant patterns of early origin of hypsodonty in Great Plains ungulates and Glires . Palaeogeogr. Palaeoecol. 365 : 1 – 10 . Google Scholar CrossRef Search ADS Labarque F.M., Soto E.M., Ramírez M.J., Arnedo M.A 2015 . Chasing ghosts: the phylogeny of Amaurobioidinae ghost spiders (Araneae, Anyphaenidae) . Zool. Scr. 44 : 550 – 561 . Google Scholar CrossRef Search ADS Landis M.J., Matzke N.J., Moore B.R 2013 . Bayesian analysis of biogeography when the number of areas is large . Syst. Biol. 62 : 789 – 804 . Google Scholar CrossRef Search ADS PubMed Landry J.F 1994 . Resource partitioning in a guild of marsh-dwelling Agonum (Coleoptera: Carabidae) in central Alberta . Can. Entomol. 126 03 : 709 – 728 . Google Scholar CrossRef Search ADS Losos J.B 2011 . Convergence, adaptation, and constraint . Evolution 65 7 : 1827 – 1840 . Google Scholar CrossRef Search ADS PubMed Maddison W.P., Maddison D.R 2015 . Mesquite: a modular system for evolutionary analysis . Version 3.04. Available from: URL http://mesquiteproject.org . Mahler D.L., Ingram T., Revell L.J., Losos J.B 2013 . Exceptional convergence on the macroevolutionary landscape in island lizard radiations . Science 341 6143 : 292 – 295 . Google Scholar CrossRef Search ADS PubMed Martins E.P., Hansen T.F 1997 . Phylogenies and the comparative method: a general approach to incorporating phylogenetic information into the analysis of interspecific data . Am. Nat. 149 4 : 646 – 667 . Google Scholar CrossRef Search ADS Matzke N.J 2012 . Founder-event speciation in BioGeoBEARS package dramatically improves likelihoods and alters parameter inference in Dispersal-Extinction-Cladogenesis (DEC) analyses . Frontiers Biogeogr. 4 : 210 . Matzke N.J 2013 . BioGeoBEARS: BioGeography with Bayesian (and likelihood) evolutionary analysis in R scripts . Berkeley (CA) : University of California . Mikulska I 1967 . Some observations on the biology of the spider Tibellus oblongus (Walck.). Przegląd Zoologiczny 11 : 388 – 391 . Monteiro L.R 2013 . Morphometrics and the comparative method: studying the evolution of biological shape. Hystrix It. J. Mamm. 24 1 : 25 – 32 . Morrone J.J 2006 . Biogeographic areas and transition zones of latin america and the caribbean islands based on panbiogeographic and cladistic analyses of the entomofauna. Annu. Rev. Entomol. 51 : 467 – 494 . Google Scholar CrossRef Search ADS PubMed Morrone J.J 2015 . Biogeographical regionalisation of the Andean region. Zootaxa 3936 : 207 – 236 . Google Scholar CrossRef Search ADS PubMed Muschick M , Indermaur A , Salzburger W 2012 . Convergent evolution within an adaptive radiation of cichlid fishes. Curr. Biol. 22 24 : 2362 – 2368 . Google Scholar CrossRef Search ADS PubMed O’Meara B.C , Ané C , Sanderson M.J , Wainwright P.C 2006 . Testing for different rates of continuous trait evolution using likelihood. Evolution 60 5 : 922 – 933 . Google Scholar CrossRef Search ADS PubMed O’Meara B.C , Beaulieu J.M 2014 . Modelling stabilizing selection: The attraction of Ornstein–Uhlenbeck models. In: Garamszegi L.Z , editor. Modern phylogenetic comparative methods and their application in evolutionary biology . Berlin Heidelberg : Springer , p. 381 – 393 . Google Scholar CrossRef Search ADS Orgogozo V 2015 . Replaying the tape of life in the twenty-first century. Interface Focus 5 6 : 20150057 . Google Scholar CrossRef Search ADS PubMed Orme D , Freckleton R , Thomas G , Petzoldt T , Fritz S , Isaac N , Pearce W 2013 . caper: comparative analyses of phylogenetics and evolution in R. R package version 0.5.2. Available from: URL https://CRAN.R-project.org/package=caper . Osborne C.P 2008 . Atmosphere, ecology and evolution: what drove the Miocene expansion of C4 grasslands? J. Ecol. 96 : 35 – 45 . Google Scholar PubMed Pagel M 1999 . Inferring the historical patterns of biological evolution. Nature 401 6756 : 877 – 884 . Google Scholar CrossRef Search ADS PubMed Palazzesi L , Barreda V 2012 . Fossil pollen records reveal a late rise of open-habitat ecosystems in Patagonia. Nat. Commun. 3 : 1294 . Google Scholar CrossRef Search ADS PubMed Parker J , Rambaut A , Pybus O.G 2008 . Correlating viral phenotypes with phylogeny: accounting for phylogenetic uncertainty. Infect. Genet. Evol. 8 : 239 – 46 . Google Scholar CrossRef Search ADS PubMed R Core Team . 2015 . R: a language and environment for statistical computing . Vienna (Austria) : R Foundation for Statistical Computing . Available from: URL https://www.R-project.org . Ramírez M.J 1999 . New species and cladistic reanalysis of the spider genus Monapia (Araneae, Anyphaenidae, Amaurobioidinae). J. Arachnol. 27 : 415 – 431 . Ramírez M.J 2003 . The spider subfamily Amaurobioidinae (Araneae, Anyphaenidae): a phylogenetic revision at the generic level. Bull. Am. Mus. Nat. Hist. 277 : 1 – 262 . Google Scholar CrossRef Search ADS Ree R.H , Moore B.R , Webb C.O , Donoghue M.J 2005 . A likelihood framework for inferring the evolution of geographic range on phylogenetic trees. Evolution 59 : 2299 – 2311 . Google Scholar CrossRef Search ADS PubMed Revell L.J 2009 . Size-correction and principal components for interspecific comparative studies. Evolution 63 12 : 3258 – 3268 . Google Scholar CrossRef Search ADS PubMed Revell L.J 2010 . Phylogenetic signal and linear regression on species data. Method. Ecol. Evol. 1 4 : 319 – 329 . Google Scholar CrossRef Search ADS Revell L.J 2012 . phytools: an R package for phylogenetic comparative biology (and other things). Method. Ecol. Evol. 3 : 217 – 223 . Google Scholar CrossRef Search ADS Ronquist F 1997 . Dispersal-vicariance analysis: a new approach to the quantification of historical biogeography. Syst. Biol. 46 : 195 – 203 . Google Scholar CrossRef Search ADS Rubio G.D , Ramírez M.J 2015 . Taxonomic revision of the American spider genus Arachosia (Araneae: Anyphaenidae). Zootaxa 3932 : 1 – 105 . Google Scholar CrossRef Search ADS PubMed Seehausen O 2007 . Chance, historical contingency and ecological determinism jointly determine the rate of adaptive radiation. Heredity 99 : 361 – 363 . Google Scholar CrossRef Search ADS PubMed Sersic A.N , Cosacov A , Cocucci A.A , Johnson L.A , Pozner R , Avila L.J , Sites J Jr. , Morando M . 2011 . Emerging phylogeographical patterns of plants and terrestrial vertebrates from Patagonia. Biol. J. Linn. Soc. 103 2 : 475 – 494 . Google Scholar CrossRef Search ADS Sidlauskas B 2008 . Continuous and arrested morphological diversification in sister clades of characiform fishes: a phylomorphospace approach. Evolution 62 12 : 3135 – 3156 . Google Scholar CrossRef Search ADS PubMed Silvestro D , Kostikova A , Litsios G , Pearman P.B , Salamin N. 2015 . Measurement errors should always be incorporated in phylogenetic comparative analysis. Method. Ecol. Evol. 6 3 : 340 – 346 . Google Scholar CrossRef Search ADS Smaers J.B , Rohlf F.J 2016 . Testing species’ deviation from allometric predictions using the phylogenetic regression. Evolution 70 5 : 1145 – 1149 . Google Scholar CrossRef Search ADS PubMed Soriano A 1991 . Río de la Plata grasslands. In: Coupland R.T , editor, Natural grasslands, introduction and western hemisphere . Amsterdam : Elsevier . p. 367 – 407 . Soto E.M , Labarque F.M , Ceccarelli F.S , Arnedo M.A , Pizarro-Araya J , Ramírez M.J 2017 . The life and adventures of an eight-legged castaway: Colonization and diversification of Philisca ghost spiders on Robinson Crusoe Island (Araneae, Anyphaenidae). Mol. Phylogenet. Evol. 107 : 132 – 141 . Google Scholar CrossRef Search ADS PubMed Soto E.M , Ramírez M.J 2012 . Revision and phylogenetic analysis of the spider genus Philisca Simon (Araneae: Anyphaenidae, Amaurobioidinae). Zootaxa 3443 : 1 – 65 . Stebbins G.L 1981 . Coevolution of grasses and herbivores. Ann. Mo. Bot. Gard. 68 : 75 – 86 . Google Scholar CrossRef Search ADS Stevens M , Merilaita S 2009 . Animal camouflage: current issues and new perspectives. Philos. Trans. R. Soc. B Biol. Sci. 364 : 423 – 427 . Google Scholar CrossRef Search ADS Strömberg C.A.E 2011 . Evolution of grasses and grassland ecosystems. Annu. Rev. Earth Planet. Sci. 39 : 517 – 544 . Google Scholar CrossRef Search ADS Strömberg C.A.E , Dunn R.E , Madden R.H , Kohn M.J , Carlini A.A 2013 . Decoupling the spread of grasslands from the evolution of grazer-type herbivores in South America. Nat. Commun. 4 : 1478 . Google Scholar CrossRef Search ADS PubMed Stuessy T.F , Foland K.A , Sutter J.F , Sanders R.W , Silva O.M 1984 . Botanical and geological significance of potassium-argon dates from the Juan Fernández Islands. Science, New Series 225 4657 : 49 – 51 . Thiele H.U 2012 . Carabid beetles in their environments: a study on habitat selection by adaptations in physiology and behavior , Vol. 10 . Berlin, Germany : Springer Science & Business Media . Toljagić O , Voje K.L , Matschiner M , Liow L.H , Hansen T.F 2018 . Millions of years behind: slow adaptation of ruminants to grasslands. Syst. Biol. 67 1 : 145 – 157 . Google Scholar CrossRef Search ADS PubMed Toussaint E.F.A , Condamine F.L , Kergoat G.J , Capdevielle-Dulac C , Barbut J , Silvain J.F , Le Ru B.P 2012 . Palaeoenvironmental Shifts Drove the Adaptive Radiation of a Noctuid Stemborer Tribe (Lepidoptera, Noctuidae, Apameini) in the Miocene. PLoS One 7 7 : e41377 . Google Scholar CrossRef Search ADS PubMed Tscharntke T , Greiler H.-J 1995 . Insect communities, grasses, and grasslands. Ann. Rev. Entomol. 40 : 535 – 558 . Google Scholar CrossRef Search ADS Turner D.D 2011 . Gould’s replay revisited. Biol. Philos. 26 1 : 65 – 79 . Google Scholar CrossRef Search ADS Uyeda J.C , Caetano D.S , Pennell M.W 2015 . Comparative analysis of principal components can be misleading. Syst. Biol. 64 4 : 677 – 689 . Google Scholar CrossRef Search ADS PubMed Uyeda J.C , Hansen T.F , Arnold S.J , Pienaar J 2011 . The million-year wait for macroevolutionary bursts. Proc. Natl Acad. Sci. USA 108 38 : 15908 – 15913 . Google Scholar CrossRef Search ADS Wainwright P.C , Reilly S.M 1994 . Ecological morphology: integrative organismal biology . Chicago, Il, USA : University of Chicago Press . Wang T.H , Donaldson Y.K , Brettle R.P , Bell J.E , Simmonds P 2001 . Identification of shared populations of Human immunodeficiency Virus Type 1 infecting microglia and tissue macrophages outside the central nervous system. J. Virol . 75 23 : 11686 – 11699 . Google Scholar CrossRef Search ADS PubMed Wheeler W.C , Coddington J.A , Crowley L.M , Dimitrov D , Goloboff P.A , Griswold C.E , Hormiga G , Prendini L , Ramírez M.J , Sierwald P , Almeida-Silva L , Alvarez-Padilla F , Arnedo M.A , Benavides Silva L.R , Benjamin S.P , Bond J.E , Grismado C.J , Hasan E , Hedin M , Izquierdo M.A , Labarque F.M , Ledford J , Lopardo L , Maddison W.P , Miller J.A , Piacentini L.N , Platnick N.I , Polotow D , Silva-Dávila D , Scharff N , Szuts T , Ubick D , Vink C.J , Wood H.M , Zhang X 2017 . The spider tree of life: phylogeny of Araneae based on target-gene analyses from an extensive taxon sampling. Cladistics 33 6 : 574 – 616 . Google Scholar CrossRef Search ADS Wiens J.J , Ackerly D.D , Allen A.P , Anacker B.L , Buckley L.B , Cornell H.V , Damschen E.I , Jonathan Davies T , Grytnes J.A , Harrison S.P , Hawkins B.A , Holt R.D , McCain C.M , Stephens P.R . 2010 . Niche conservatism as an emerging principle in ecology and conservation biology. Ecol. Lett. 13 10 : 1310 – 1324 . Google Scholar CrossRef Search ADS PubMed World Spider Catalog . 2017 . World spider catalog. Natural History Museum Bern. Available from: URL http://wsc.nmbe.ch, version 18.0, accessed on 12 June 2017 . Young K.A , Snoeks J , Seehausen O 2009 . Morphological diversity and the roles of contingency, chance and determinism in African cichlid radiations. PLoS One 4 3 : e4740 . Google Scholar CrossRef Search ADS PubMed Zahniser J.N , Dietrich C.H 2010 . Phylogeny of the leafhopper subfamily Deltocephalinae (Hemiptera: Cicadellidae) based on molecular and morphological data with a revised family—group classification. Syst. Entomol. 35 3 : 489 – 511 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press, on behalf of the Society of Systematic Biologists. All rights reserved. For Permissions, please email: journals.permissions@oup.com 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) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Systematic Biology Oxford University Press

The Grass was Greener: Repeated Evolution of Specialized Morphologies and Habitat Shifts in Ghost Spiders Following Grassland Expansion in South America

Loading next page...
 
/lp/ou_press/the-grass-was-greener-repeated-evolution-of-specialized-morphologies-92HyVXg59N
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press, on behalf of the Society of Systematic Biologists. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
1063-5157
eISSN
1076-836X
D.O.I.
10.1093/sysbio/syy028
Publisher site
See Article on Publisher Site

Abstract

Abstract While grasslands, one of Earth’s major biomes, are known for their close evolutionary ties with ungulate grazers, these habitats are also paramount to the origins and diversification of other animals. Within the primarily South American spider subfamily Amaurobioidinae (Anyphaenidae), several species are found living in the continent’s grasslands, with some displaying putative morphological adaptations to dwelling unnoticed in the grass blades. Herein, a dated molecular phylogeny provides the backbone for analyses revealing the ecological and morphological processes behind these spiders’ grassland adaptations. The multiple switches from Patagonian forests to open habitats coincide with the expansion of South America’s grasslands during the Miocene, while the specialized morphology of several grass-dwelling spiders originated at least three independent times and is best described as the result of different selective regimes operating on macroevolutionary timescales. Although grass-adapted lineages evolved towards different peaks in adaptive landscape, they all share one characteristic: an anterior narrowing of the prosoma allowing spiders to extend the first two pairs of legs, thus maintaining a slender resting posture in the grass blade. By combining phylogenetic, morphological, and biogeographic perspectives we disentangle multiple factors determining the evolution of a clade of terrestrial invertebrate predators alongside their biomes. Amaurobioidinae, anyphaenidae, biogeography, convergence, macroevolution, morphometrics To fully appreciate and understand the evolutionary history of a group of organisms, a holistic approach, considering intrinsic and extrinsic components, is crucial. Changes in the genetic material, morphological traits as well as biotic and abiotic selective pressures interact during the process of diversification of the world’s biota, in an intricate web of adaptation and co-adaptation. The adaptation of organisms to their specific habitats is one of the major driving forces of diversification, often reflected in specific physiological or morphological traits, especially when the association is strict (Wainwright and Reilly 1994). Habitats themselves are spatio-temporally dynamic, responding to changes in the planet’s geomorphology and climate. For example, the grassland biome currently covers an estimated 40% of the Earth’s terrestrial surface (Gibson 2009), and although grasses (Poaceae) are estimated to have existed since the early Cretaceous (Bouchenak-Khelladi et al. 2014), the expansion of the grassland biome did not occur until the Neogene, at different times depending on the continent (Edwards et al. 2010; Strömberg 2011), and probably due to a combination of shifts in climatic CO$$_{\mathrm{2}}$$ levels and fire regimes (Osborne 2008). The spread of grasslands has been postulated as the driver for the origin and diversification of numerous phytophagous insects (e.g., Dietrich 1999; Toussaint et al. 2012) and mammalian groups, including our own genus (e.g., Bobe and Behrensmeyer 2004). Furthermore, well-known cases of co-adaptation and even co-evolution have occurred between grasslands and ungulate grazers (e.g., Stebbins 1981; Jacobs et al. 1999; Bouchenak-Khelladi et al. 2009), although strict co-speciation has not always been the case (Jardine et al. 2012; Palazzesi and Barreda 2012; Strömberg et al. 2013). Besides trophic relations, a well-known driver of morphological evolution is the need to deceive the perception of predators or prey by crypsis or mimicry (see Stevens and Merilaita 2009), of which the melanic moths are a textbook example. Many invertebrates match the background color of the leaves or sticks on which they stand, and hide by precisely controlling the posture of the body and legs (Edmunds 1990), thus several spiders visually cryptic on grasses have thin, long bodies (Comstock 1913). For example, in the family Philodromidae, Tibellus oblongus evolved an elongated body and dwell within the blades of grasses extending their first two pairs of legs forwards and the rear two backwards (Mikulska 1967); similar characteristics can also be found among species of the crab spider genus Monaeses (Dippenaar-Schoeman 1984). These traits have been interpreted as adaptations for living on vegetation, including grasses. Grasslands are found on all major landmasses; for example, South America harbors large areas of grass-dominated ecosystems such as the Brazilian cerrados and campos, and the Argentine pampas (Soriano 1991) or lowland plains. Based on studies of fossil pollen records, the Patagonian region of South America, at present characterized by extensive areas of steppe, was covered mainly by forests (Barreda and Palazzesi 2007) until the Miocene (estimates range between 18.5 and 10 million years ago (Ma); Palazzesi and Barreda 2012; Strömberg et al. 2013), when grass-dominated open habitats began expanding. These changes in the floral composition, therefore, altered the availability of food sources and habitats for a large number of vertebrates and invertebrates in the region. The spider subfamily Amaurobioidinae (Anyphaenidae), currently with 180 species in 23 genera (World Spider Catalog 2017) is found almost exclusively in South America, with many of its genera restricted to the Patagonian region (Ramírez 2003). These spiders are active hunters that live inside silken retreats. Several species can be found living in grass-dominated open habitats, and a few remarkable morphological traits facilitate the spiders living and concealing themselves within the folded inner edge of the grass leaves. More specifically, this grass ecomorph has a more slender and elongated body configuration compared with that of all other members of the subfamily (Fig. 1), and adopts a resting posture along the leaf blade with the first two pairs of legs towards the front and two hind pairs towards the back (Fig. 1f); or, in some species, the third pair of legs is very short, and rests tightly folded besides the body (Fig. 1a). Their cryptic posture resembles that of other spiders and insects that live on elongate leaves, complementing their posture with a pattern of dots or lines matching the background (e.g., Mikulska 1967; Dippenaar-Schoeman 1984; Landry 1994). Figure 1. View largeDownload slide Spiders of the tribe Gayennini. The grass-dwelling species (top) have a slender constitution and a morphology that allows the legs to be tightly extended forwards and backwards, or folded at the side. Other species have a stockier constitution (bottom) and different resting positions. a) Monapia guenoana. b) Monapia charrua. c) Monapia angusta. d) Sanogasta tenuis. e) Sanogasta puma. f) Arachosia bergi. g) Oxysoma saccatum. h) Tomopisthes horrendus. i) Sanogasta maculosa. j) Arachosia proseni. k) Arachosia praesignis. Figure 1. View largeDownload slide Spiders of the tribe Gayennini. The grass-dwelling species (top) have a slender constitution and a morphology that allows the legs to be tightly extended forwards and backwards, or folded at the side. Other species have a stockier constitution (bottom) and different resting positions. a) Monapia guenoana. b) Monapia charrua. c) Monapia angusta. d) Sanogasta tenuis. e) Sanogasta puma. f) Arachosia bergi. g) Oxysoma saccatum. h) Tomopisthes horrendus. i) Sanogasta maculosa. j) Arachosia proseni. k) Arachosia praesignis. The amaurobioidine species known to display these specific morphological characteristics belong to three different genera, Arachosia, Monapia, and Sanogasta (Ramírez 1999, 2003; Rubio and Ramírez 2015), and the phylogenetic trees both from morphological (Ramírez 2003) and DNA data (Labarque et al. 2015) indicate that these putative adaptations have most likely arisen independently. The repeated evolution of similar morphologies as a response to similar environmental conditions, such as seen among several well-studied cases of morphological convergence (e.g., Hawaiian Tetragnatha spiders: Gillespie 2004; African cichlids: Muschick et al. 2012; Caribbean Anolis lizards: Mahler et al. 2013), is an impressive result of evolution under natural selection. Such cases also represent unique natural experiments in which the relative impact of contingency and determinism in evolutionary history can be assessed (Gould 1989; Conway Morris 2003; Turner 2011; Orgogozo 2015). Amaurobioidinae has been the focus of recent molecular phylogenetic studies in the form of a general, total-evidence study (Labarque et al. 2015) providing insight and confirmation of previous, morphology-based classifications (Ramírez 2003), as well as dated phylogenetic studies focusing on specific genera of the subfamily (Ceccarelli et al. 2016a; Soto et al. 2017). Herein, we present the most extensive sampling of Amaurobioidinae for molecular phylogenetic work to date, as well as the first instance of node age estimates for the whole subfamily. Furthermore, we use the dated molecular phylogeny for character mapping and morphospace analyses, to assess the impact of grassland expansions on morphological adaptations in the subfamily in terms of macroevolutionary mechanisms. In particular, we aim to answer the following general questions. 1) Was the origin of grass-dwelling clades in Amaurobioidinae concurrent with the habitat’s expansion in South America? 2) How frequent were the ecological shifts to and from open habitats, including grasslands, throughout the subfamily’s evolutionary history?, and 3) did the independent colonization of grasslands result in the repeated origin of similar morphologies? Through the outcomes of this study, it is possible to evaluate the plasticity in terms of adaptation to habitat changes at a macroevolutionary scale in this group of spiders. Materials and Methods Taxon Sampling For the molecular phylogenetic analyses 119 ingroup and 14 outgroup taxa were used. The ingroup taxa belong to 22 described genera and 1 undescribed genus of the subfamily Amaurobioidinae, while the outgroup taxa comprised nine genera belonging to the anyphaenid subfamily Anyphaeninae and one genus (Elaver), belonging to the family Clubionidae, to root the trees, following Wheeler et al. (2017). Of these taxa, 38 belonging to the subfamily Amaurobioidinae (plus one outgroup) were new to this study, while the remainder had been used before (Labarque et al. 2015; Ceccarelli et al. 2016a; Soto et al. 2017). For five of the previously used taxa, part of the DNA data that had been missing was obtained for this study as well. A list of the taxa used here can be found in Supplementary Table S1 of SI File 1 available on Dryad at http://dx.doi.org/10.5061/dryad.20257. For details on sequencing and phylogenetic inferences including node age estimates, readers are referred to the SI File 2 available on Dryad. Habitat and Area Mapping Geographic range, habitat type, and ecomorph were assigned as traits for each taxon in the dated phylogeny for further analyses. Because the most notable change in grassland expansions in southern South America occurred in what is today known as the Patagonian region (Palazzesi and Barreda 2012; Strömberg et al. 2013), the geographic ranges chosen for the analyses were limited to four categories: the first two, Patagonia (Pg) and non-Patagonian Neotropical (NP) cover the bulk of the diversity (see Fig. 2); Juan Fernández Archipelago (JF) covers the radiation of Philisca species (Soto et al. 2017) and the occurrence of Sanogasta maculosa “skottsbergi” (Ramírez 2003), and the trans-oceanic distribution in South Africa, Australia, and New Zealand (TO) covers in summary the species of the intertidal genus Amaurobioides that were already studied in detail (Ceccarelli et al. 2016a). For Pg, we used the Patagonian Andean region excluding the South American Transition zone as delimited by Morrone (2006), composed by the Argentinean Pg plus Southern and Central Chile (Fig. 2c). This area corresponds well with the distributional limits of many anyphaenid taxa (Ramírez 2003) and other organisms (Morrone 2015). Figure 2. View largeDownload slide Analysis of area and habitat occupancy through time in amaurobioidine spiders. a) Representatives of the three clades exhibiting a grass ecomorph. b) Dated phylogeny with ancestral areas and habitats mapped on nodes (black circles denote nodes with posterior probability $$\geqslant $$0.95). Yellow/light grey circles show the inferred transitions to grass ecomorph with highest posterior probability. c) Regionalization of the main areas of distribution of Amaurobioidinae analyzed in this study. d) Stacked profile of lineages through time, classified by habitat. NP $$=$$ Non-Patagonian Neotropical; Pg $$=$$ Patagonian; PgNP $$=$$ both Pg and NP; JF $$=$$ Juan Fernandez Archipelago; TO $$=$$ transoceanic (South Africa, Australia and New Zealand); Fo $$=$$ broad-leaf forest; Op $$=$$ open; OpFo $$=$$ both Op and Fo. Figure 2. View largeDownload slide Analysis of area and habitat occupancy through time in amaurobioidine spiders. a) Representatives of the three clades exhibiting a grass ecomorph. b) Dated phylogeny with ancestral areas and habitats mapped on nodes (black circles denote nodes with posterior probability $$\geqslant $$0.95). Yellow/light grey circles show the inferred transitions to grass ecomorph with highest posterior probability. c) Regionalization of the main areas of distribution of Amaurobioidinae analyzed in this study. d) Stacked profile of lineages through time, classified by habitat. NP $$=$$ Non-Patagonian Neotropical; Pg $$=$$ Patagonian; PgNP $$=$$ both Pg and NP; JF $$=$$ Juan Fernandez Archipelago; TO $$=$$ transoceanic (South Africa, Australia and New Zealand); Fo $$=$$ broad-leaf forest; Op $$=$$ open; OpFo $$=$$ both Op and Fo. Similarly, the general habitat type was either chosen as open (Op), referring mainly to grasslands, but also including other open habitats with few or no trees, such as scrublands, marshes, and rocky shores; or forest (Fo), mostly representing broad-leaf forests including foliage of shrubs. Because the two habitat types (Op and Fo) frequently co-exist as patches in many localities, and because some species use both habitats, the two variables (geographic range and habitat type) were treated as independent traits in separate analyses. Each terminal taxon in the maximum clade credibility (mcc) tree from BEAST was assigned the code or codes which best defined the species’ geographical range and general habitat type as documented in the taxonomic revisions, the MACN-Ar database (http://www.gbif.org/dataset/34e892b0-d9c7-11de-b793-b8a03c50a862) and field notes of author MJR. A detailed habitat analysis is circumscribed to the densely-sampled tribes Amaurobioidini $$+$$ Gayennini (A$$+$$G Clade); the outgroups, including the subfamily Anyphaeninae and the genus Josa (sister to A$$+$$G Clade) are all non-Patagonian and were sampled more superficially. Finally, species were coded as being of the grass ecomorph, or lumped into a “generalized morphology” category. As explained in the introduction, all species of the grass ecomorph have pale, elongated bodies, and in the case of Monapia and Sanogasta are collected exclusively on grasses such as Coleataenia prionitis, Cortaderia selloana, Paspalum spp., Panicum spp. (Supplementary Table S4 of SI File 1 available on Dryad). Grass inhabiting Arachosia dwell in the same grasses listed above, but are also found in other monocot plants with elongate leaves, such as Typha angustifolia, Eryngium species, sedges (Cyperaceae), and rushes (Juncaceae). To test whether the traits are clustered or randomly distributed on the phylogeny (i.e., show phylogenetic signal) Bayesian Tip-Significance testing was carried out on the last 5000 trees from each BEAST run in the program Befi-BaTS v.0.1.1 (Parker et al. 2008). This program uses three statistics, the Association Index (AI; Wang et al. 2001) the Fitch parsimony score (PS), and the Monophyletic Clade (MC) size statistic, to test the significance of phylogenetic signal in the traits against a null distribution in which the characters at the tips are reshuffled 1000 times. The ancestral trait estimation for areas and habitats were run on the BEAST mcc tree with BioGeoBEARS package v. 0.2.1 (Matzke 2012, 2013) for R v. 3.2.3 (R_Core_Team 2015) under three models: Dispersal-Extinction-Cladogenesis (DEC; Ree et al. 2005), DIVALIKE, modified from the DIVA program of Ronquist (1997), and BAYAREALIKE, modified from the BayArea program of Landis et al. (2013) algorithms. These three models factored to six after the addition of the founder-event speciation parameter J, and the best fit model was selected based on the Akaike information criterion (AIC). The BioGeoBEARS analyses were run using a script provided by N. Matzke (http://phylo.wikidot.com/biogeobears). The area analysis was stratified into two time periods: before and after 4 Ma, reflecting the emergence of the JF area (Stuessy et al. 1984). Consequently, the “areas allowed” and “dispersal multiplier” input matrices for the BioGeoBEARS run factored in a low likelihood for any occurrence in Juan Fernández before 4 Ma. The input matrices for “distance” and “dispersal multiplier” considered short distances within America, long distance to Juan Fernández, and longer distances to South Africa, Australia and New Zealand (see Supplementary Matrices S1 and S2 of SI File 1 available on Dryad). The ancestral state estimations corresponding to the model with the best fit for the two runs (area and habitat) were analyzed to obtain a profile of area and habitat occupancy through time for the ingroup (A$$+$$G Clade); the transitions in habitats and areas were also counted and analyzed for general trends. To obtain an estimate of the uncertainty of the results, we produced area and habitat analyses for each of 100 trees randomly drawn from the post-burnin posterior sample of the Bayesian analysis in BEAST, using the optimal model selected for the mcc tree in BioGeoBEARS. The time of transitions and ancestral state estimations for given time slices were calculated by linearly interpolating between the probabilities in dominant states in ancestor and descendant nodes (Supplementary scripts provided in SI File 3 available on Dryad). The expected number of lineages in each state for these given time slices was also calculated, under the assumption that the lineages in each state increased proportionally through time, and compared with the observed (average) numbers from the 100 trees by means of $$\chi^{\mathrm{2}}$$ and paired $$t$$-tests. For the evolution of ecomorphs, Bayesian stochastic mapping was carried out on 100 randomly selected post-burnin chronograms from BEAST, using the simmap algorithm (Bollback 2006) implemented in the R package phytools v. 0.6 (Revell 2012). Ten thousand Markov Chain Monte Carlo (MCMC) generations were run for 100 replicates on each tree under equal rates and the state changes over all stochastic maps summarized. A single stochastic map was also obtained by running 100 replicates, each with 10,000 MCMC generations on the A$$+$$G clade of the mcc tree obtained in BEAST, to get a graphical representation of the results. Morphological Macroevolution Analyses To explore the origin of the putative morphological adaptations associated with the exploitation of grasslands, four morphometric measurements were taken from museum specimens: total carapace length and width (a proxy for slim vs. stocky body), distance between the condyles of the chelicerae (a proxy for the ability to maintain a slender profile while extending the fore legs forward and close to each other), and length of the tibia of the third pair of legs (a proxy for the ability to tightly fold and hide the third legs at the sides of the body). Specimens were measured using a binocular microscope with an ocular micrometer. Given that putative grass-adapted species are restricted to the tribe Gayennini, morphometric data were only gathered for this clade. With the aim to minimize the effect of sexual dimorphism, only adult females were measured; as is common in spiders, immature anyphaenids throughout all free instars are similar to females in habitus and resting posture, while males have a slender constitution and sometimes cheliceral modifications (Soto and Ramírez 2012). In this way, the measured individuals had similar morphologies to those of the whole life cycle, while bearing the genital characters necessary for taxonomic identification. Whenever possible, multiple female specimens (maximum 5, mean 4.1) per species were measured, and trait values were expressed as the logarithm of the average of individual observations. The historical pattern of morphological change of these characters was studied using a version of the mcc tree pruned down to the 73 members of the tribe Gayennini for which morphological data could be gathered (Supplementary Table S5 of SI File 1 available on Dryad). First, character selection validation was carried out by testing for differences in the allometric scaling of these between grass-dwelling spiders and the rest, a common method to explore the existence of morphological adaptations in a comparative phylogenetic framework (Garland and Ives 2000). A phylogenetic generalized least squares regression (PGLS; Grafen 1989; Martins and Hansen 1997) was performed for all characters against carapace length, which was selected as a proxy for body size. Regressions were performed using R package caper (Orme et al. 2013) excluding species showing putative grassland adaptations, and simultaneously estimating the $$\lambda $$ parameter (Pagel 1999) to account for deviations from a pure Brownian motion (BM), as recommended by Revell (2010). A 95% prediction interval was then calculated for each regression using R scripts provided in Smaers and Rohlf (2016), and data points for grass-dwelling spiders were plotted to see whether they lie within the estimated intervals. Although this procedure does not represent, in a strict sense, a statistical test of deviations from allometric predictions (Smaers and Rohlf 2016), it was simply used here to determine whether the chosen characters were indeed capturing information relevant to the study of niche-associated morphological differentiation. Furthermore, the interactions among these four characters were explored by visualizing morphological variability and evolution using phylomorphospaces (Sidlauskas 2008) on phylogenetic principal component axes (pPCA; Revell 2009). Using the same dataset, the fit of multiple process-based models of macroevolution were explored in a multivariate framework using R package mvMORPH (Clavel et al. 2015). Such an approach to model phenotypic evolution outperforms commonly employed univariate methods given its capability to take into account the existence of correlations among traits. Furthermore, it allows to circumvent any pre-processing of the data, such as the use of principal component analysis or its phylogenetic variant, transformations that are known—under certain conditions—to introduce biases when used upstream of methods of evolutionary inference (Revell 2009; Monteiro 2013; Uyeda et al. 2015). However, recent simulation studies have shown that this method suffers from problems of model misspecification as the number of traits ($$p)$$ increases relative to the number of species ($$N)$$ (Adams 2014; Adams and Collyer 2018). To test whether the $$N$$:$$p$$ ratio of ~ 18:1 is enough to guarantee accurate model selection, 200 replicates of 4 dimensional trait matrices were simulated under BM using the empirical tree and phylogenetic variance-covariance matrix. Following Adams and Collyer (2018), a comparison of model fit between BM and Ornstein–Uhlenbeck (OU, see below) models using AIC scores revealed a percentage of model misspecification of only about 2%. Eight different models were tested, including three models in which morphological change was described using a single process for the entire tree, and five models with different processes dictating change in different regions of the tree. Among the first class, the fit of the following models were analyzed 1) a BM model, the most common null model of evolution in comparative studies, designed to represent a process of random walk with constant variance (described in detail in Freckleton et al. 2002; O’Meara et al. 2006); 2) an OU model, which adds to the random walk a tendency to evolve towards an “optimum value”, generally considered to represent a biological scenario of stabilizing selection (Hansen 1997; Butler and King 2004); and 3) an early burst (EB) model, in which the Brownian rate of morphological change decays exponentially with time (Blomberg et al. 2003). Furthermore, to explore the degree of morphological convergence shown by clades that independently colonized grasslands, five different multi-OU models were also tested. Such models constitute a convenient representation of evolution towards peaks in a macroevolutionary adaptive landscape (Ingram and Mahler 2013), and have become increasingly popular to describe cases of adaptive radiations and convergent evolution (Cressler et al. 2015; O’Meara and Beaulieu 2014; Cooper et al. 2016). All models included a single peak for all species with a general morphology, but differed in the number of peaks used to model the evolution among species of the grass ecomorph, including: 1) a single selective regime for all grass-dwelling species (OU2), 2) two different regimes, one exploited by a single genus and the other by the remaining two (OU3; all three different combinations were explored), and 3) a scenario in which all three grass-dwelling clades exploit different peaks (OU4). These models were specifically designed to test different levels of morphological convergence triggered by the expansion of grasslands, ranging from complete to non-existent (Supplementary Fig. S1 of SI File 4 available on Dryad). In all cases, measurement errors were incorporated in the estimation process (Supplementary Table S5 of SI File 1 available on Dryad), calculated as squared standard errors of the mean values (Clavel et al. 2015). For the small number of species coded from single specimens, errors corresponded to the averaged measurement error for each character. All runs were parameterized to assume a stationary distribution for the selective optimum at the root (option vcv $$=$$ “randomRoot” in function mvOU), corresponding to a scenario of evolution at a steady state (Clavel et al. 2015). This assumption is biologically reasonable in our case study, and has been found to stabilize estimates of optimum values (Cooper et al. 2016). In all cases, regimes were painted on the mcc tree using the make.simmap function in phytools, and the five multi-OU models were fitted to 100 different optimizations derived from stochastic character mappings under equal rates. Furthermore, one simulation was also explored in each of 100 trees randomly drawn from the posterior sample of the Bayesian analysis. In such a way, uncertainty in species trait values (Silvestro et al. 2015), timing and number of origins of the grass ecomorph (Huelsenbeck et al. 2003), divergence times and topology (Huelsenbeck et al. 2000) were all contemplated. To avoid problems of overfitting, which are known to impact multi-OU model selection (Boettiger et al. 2012; Ho and Ané 2014), models were compared using weights derived from sample-corrected ACC scores (wAICc; Burnham and Anderson 2002). In case of nested hypotheses, model support was further explored using log-likelihood ratio tests (LRTs). The tempo of the adaptive process was analyzed using phylogenetic half-lives, representing the time needed to move halfway from an ancestral state to an adaptive optimum, which in a multivariate framework are calculated on the eigenvalues of the selection ($$\alpha )$$ matrix (Bartoszek et al. 2012). Results Phylogenetic Inference and Node Age Estimates The characteristics of the four DNA data matrices used for phylogenetic analyses can be found in Supplementary Table S6 of SI File 1 available on Dryad. The percentage of missing data was 2.25% for the Cytochrome Oxidase I gene fragment (COI), 9.77% for 16S, 5.26% for H3, and 7.52% for 28S. The individual gene trees for each marker (Supplementary Figs. S2–S4 of SI file 4 available on Dryad) vary in their level of resolution, since, as would be expected, the nuclear markers have lower rates of nucleotide substitution (mean rates of substitutions per site per million years of 28S: 0.00243, H3: 0.00498, and mitochondrial loci: 0.0358), therefore resolving only the deeper divergences. The topologies obtained with the concatenated dataset for the Amaurobioidinae taxa in this study (Fig. 2, Supplementary Figs. S2–S8 of SI File 4 available on Dryad) largely agree with the concatenated molecular phylogenetic analyses of previous studies (Labarque et al. 2015; Ceccarelli et al. 2016a; Soto et al. 2017). Relationships for genera vary in the placement of terminals such as the monotypic genera Selknamia and Axyracrus, always with low support. Most genera are recovered as monophyletic in both Bayesian and maximum likelihood (ML) trees; although certain taxonomic inconsistencies remain, mostly due to complications of previous assignations of species to specific genera (see Soto et al. 2017). Species of Arachosia and Monapia displaying the grass ecomorph (A. avalosi, A. bergi, A. kapiipeoi and A. magna; M. angusta, M. charrua and M. guenoana) form reciprocally monophyletic groups in all estimations. The two species of grass-dwelling Sanogasta (S. puma and S. tenuis) are monophyletic only in the ML tree, although with low bootstrap support, while in the Bayesian analyses S. x-signata falls in between, with high support. Furthermore, Sanogasta as a whole is probably paraphyletic with regards to Arachosia, as already suggested by the morphological evidence (Ramírez 2003). While the lack of resolution for some groups may impede strong systematic conclusions (e.g., resolving the relationships among certain genera), the three clades of grass ecomorphs are separated from each other by several well supported nodes, thus confirming a scenario of convergent evolution. In order to account for the statistical uncertainty reflected by the low support of some nodes, all further analyses average over a subset of post-burnin trees (see below). The node age estimates obtained using fossil calibrations are similar to those based on the COI rate prior (tree not shown), as well as node ages from the previous studies focused on Philisca (Soto et al. 2017) and Amaurobioides (Ceccarelli et al. 2016a) (Supplementary Table S7 of SI File 1 available on Dryad), corroborating the node age estimates from the previous studies with the dates obtained here (Supplementary Fig. S4 of SI File 4 available on Dryad). In particular, the estimated dates of dispersal and speciation for the endemic species of the volcanic JF postdate the origin of the islands. We added two Philisca species not studied by Soto et al. (2017), as well as specimens of Sanogasta maculosa from Robinson Crusoe and Serlkirk islands; these last are denoted Sanogasta maculosa “skottsbergi”, because Sanogasta skottsbergi described from the archipelago is currently considered a synonym of S. maculosa (Ramírez 2003). Comparing with the previous studies, our more extensive sampling provides better estimates for the whole subfamily and most of its genera (Fig. 2). Greater accuracy in node age estimates is also expected for the genera with putative grass-adapted species, i.e., Arachosia, Monapia and Sanogasta, as these were also sampled more intensively. With regards to whether the three traits (geographic range, habitat and ecomorph) were clustered or randomly distributed throughout the phylogeny, both the AI and PS metrics detected significant phylogenetic signal in all cases (all $$P < 0.001$$). Accordingly, all states showed non-random distributions according to the MC statistic. However, phylogenetic signal for geographic range was found to be mostly driven by the non-Patagonian areas (MC NP: $$P< 0.001$$; MC Pg: $$P< 0.05$$), while grass-dwelling spiders contributed most of the clustering in the case of ecomorph (MC grass: $$P< 0.001$$; MC other: $$P < 0.05$$). Contributions were more balanced for habitat type (MC open: $$P < 0.01$$; MC forest: $$P< 0.05$$). Therefore, all the traits studied here are to some extent maintained through common descent, with the strongest phylogenetic signal found in species from non-Patagonian areas and those exhibiting a grass ecomorph. Habitat and Area Mapping The model selected by AIC for the area and habitat analyses with BioGeoBEARS was DEC$$+$$J, concentrating 96% and 90% of the AIC weights respectively (Supplementary Table S8 of SI File 1 available on Dryad). The ancestral state estimations (Fig. 2, Supplementary Figs. S9 and S10 of SI File 4 available on Dryad) and lineage profiles through time (Fig. 3) were thus calculated for the selected model. Figure 3. View largeDownload slide Analyses of ancestral habitats and areas for the A$$+$$G clade of amaurobioidine spiders, summarizing runs of 100 randomly selected post-burnin trees from the Bayesian analysis in BEAST. a) Count of lineages through time, by habitat (mean $$\pm$$ standard deviation). b) Count of lineages through time, by area (mean $$\pm$$ standard deviation). c) Habitat shifts grouped in time periods. d) Area shifts grouped in time periods, focusing on Patagonia, and the Neotropics. e) Shifts by combination of habitat and geographic area in South America (average over 100 Bayesian trees) summarized on a vegetation index map; arrow thickness is proportional to estimated number of shifts (image by Reto Stockli, NASA’s Earth Observatory Group, using data provided by the MODIS Land Science Team). Figure 3. View largeDownload slide Analyses of ancestral habitats and areas for the A$$+$$G clade of amaurobioidine spiders, summarizing runs of 100 randomly selected post-burnin trees from the Bayesian analysis in BEAST. a) Count of lineages through time, by habitat (mean $$\pm$$ standard deviation). b) Count of lineages through time, by area (mean $$\pm$$ standard deviation). c) Habitat shifts grouped in time periods. d) Area shifts grouped in time periods, focusing on Patagonia, and the Neotropics. e) Shifts by combination of habitat and geographic area in South America (average over 100 Bayesian trees) summarized on a vegetation index map; arrow thickness is proportional to estimated number of shifts (image by Reto Stockli, NASA’s Earth Observatory Group, using data provided by the MODIS Land Science Team). The ancestor of the subfamily Amaurobioidinae is likely to have arisen from a forest-dwelling Neotropical non-Patagonian lineage; although the outgroup sampling is sparse, no species of Josa or Anyphaeninae live in Pg, and most live in forest habitats. The most recent common ancestor of the A$$+$$G Clade colonized the Patagonian forests about 27 Ma ago (22.01–32 Ma 95% Highest Posterior Density, HPD), where it diversified. Up to approximately 14 Ma ago the increase in diversity was accounted for exclusively by forest-dwelling lineages in Pg (Fig. 3a and b). From that point on starts the increase of lineages in the open habitats (Figs. 2d, 3a) and in Neotropical regions outside Pg (Fig. 3b). Comparing the observed lineage state occupancies (based on 100 randomly-selected post-burnin chronograms from the BEAST analysis) with lineage state occupancies expected if the proportions had been the same throughout history, the mean overall state occupancy in open habitat and forest-dwellers is significantly different ($$\chi ^{\mathrm{2}}_{\mathrm{(66)}} =$$ 114.672, $$P< 0.001$$; see Supplementary Fig. S11a of SI File 4 available on Dryad), driven by the differences observed before 10 Ma ($$\chi^{\mathrm{2}}_{\mathrm{(44)}} =$$ 87.67, $$P< 0.001$$), while there was no significant difference from 10 Ma to the present ($$\chi ^{\mathrm{2}}_{\mathrm{(20)}} =$$ 27.0, $$P =$$ 0.135). This result reflects the occupancy of the open habitat, which is only significantly lower than expected before 10 Ma ($$t$$-test $$t =$$ 4.996; $$P< 0.001$$; Supplementary Fig. S11b of SI File 4 available on Dryad), while the forest occupancy is higher than expected by chance throughout the history of the spiders ($$t = -$$5.407; $$P< 0.001$$; Supplementary Fig. S11b of SI File 4 available on Dryad). Similarly, there is a significant difference between observed and expected state occupancies through time in Pg ($$t =$$$$-$$5.627, $$P< 0.001$$), whereas state occupancy in non-Patagonian areas was significantly lower than expected before 10 Ma ($$t =$$ 3.907, $$P< 0.001$$; Supplementary Fig. S12 of SI File 4 available on Dryad). The overall temporal pattern is that of an initial source of lineages coming from the Patagonian forests, as shown in the earlier start of shifts from forest to open habitats (Fig. 3c), and from Patagonian to NP areas (Fig. 3d). The process becomes dynamic shortly after, with a gradual increase in the reverse transitions near recent times, especially in habitat shifts. The relatively high number of backward shifts in recent times from open to forest habitats is mainly due to generalist species that have come to use both habitats (Supplementary Table S9 of SI File 1 available on Dryad). In comparison, the backward shifts to Pg are very few (Supplementary Table S10 of SI File 1 available on Dryad), and were found to be significantly fewer than the number of movements out of this region (mean $$\pm$$ standard deviation of 100 random posterior BEAST trees, Pg$$\to $$NP and Pg $$\to $$ PgNP $$=$$ 7.88 $$\pm$$ 0.72, NP$$\to $$Pg and NP$$\to $$PgNP $$=$$ 4.49 $$\pm$$ 1.00, $$t$$$$= -$$20.707, $$P< 0.001$$). A dissection of the number of transitions by combinations of habitat and geographic area (averaged over 100 Bayesian trees; Fig. 3e, Supplementary Tables S11 and S12 of SI File 1 available on Dryad) reveals a strong asymmetry in habitat changes within Pg (10 shifts Op $$\to $$ Fo, vs. 5 Fo $$\to $$ Op) compared with the ones that occurred in the non-Patagonian Neotropics (1 vs. 5, respectively). It is also evident that more events of dispersal between these biogeographic regions occurred in open habitats, rather than to/from forests. Origin of Grass-Adapted Species Based on the results of this study, clades showing putative adaptations to grasslands arose from ancestors already living in open habitats in the NP region. Bayesian stochastic mapping on 100 random BEAST trees and the mcc tree (100 replicates in both cases) detected an average of 4.44 and 4.38 state changes across the phylogenies, respectively. These changes were divided as an average of 3.98 (100 trees) and 3.99 (mcc tree) transitions from a general morphology to a grass ecomorph and 0.46 (100 trees) and 0.39 (mcc tree) transitions in the opposite direction. In terms of integers this translates as four transitions to a specialized morphology and zero reversals (see Supplementary Fig. S13 of SI File 4 available on Dryad), although in a few simulations one or more transitions from grass ecomorph to general morphology were detected, most of them apparently accounted for by the ambiguity introduced by the inclusion of S. x-signata inside the clade containing the grass-dwelling Sanogasta. In the case of Arachosia, the genus’ most recent common ancestor (MRCA) is estimated to have diverged 10.75 Ma (8.14–13.6 Ma, 95% HPD) and probably inhabited open habitats, while the grass ecomorph species diversified approximately 1.98 Ma (1.16–2.92, 95% HPD). At ca. 4.07 Ma (2.58–5.8, 95% HPD), the MRCA of these species and their sister-group, Arachosia proseni (Fig. 1j), diverged; A. proseni being a rather flat and greenish species that inhabits bromeliad-like Eryngium species. The second case of putative grass-adaptation is found in Sanogasta; the clade containing the two grass ecomorph species Sanogasta tenuis and S. puma diverged approximately 5 Ma (3.4–6.84, 95% HPD), while their ancestors inhabited open areas from approximately 8.51 Ma (6.36–10.89, 95% HPD). Lastly, while the group of Monapia using open habitats traces back to ca. 9.49 Ma (7.12–12.01, 95% HPD), the clade containing the grass ecomorph Monapia species diversified at about the same time as the grass ecomorph Sanogasta species, around 6.82 Ma (4.70–9.09, 95% HPD). Morphological Macroevolution Analyses Several aspects of the morphology of grass ecomorph spiders were found to diverge from the allometric expectations based on data from the remaining amaurobioidines (Fig. 4, top). Most notably perhaps is the reduced distance separating the chelicerae, the only trait that is found to have arisen in at least some members of all three grass-dwelling clades. On the other hand, members of the genera Monapia and Sanogasta seem to have convergently acquired a reduced body size with a slender carapace morphology, while the length of the third pair of legs seems to be beyond the predicted interval only for grass-dwelling species of Sanogasta. The exploration of this dataset with pPCA reveals a similar pattern (Fig. 4, bottom). The morphology acquired by different clades specialized to life on grasses seems to show some component of convergence, as revealed especially by pPCA axis 3 (Fig. 4, bottom right) which shows the clustering of all three clades and a clear distinction with respect to all other members of the A$$+$$G clade. Species with higher values of pPCA axis 3 tend to have narrower and elongated bodies, with a shorter third pair of legs and a reduced distance between the chelicerae (loadings matrix can be found in Supplementary Table S13 of SI File 1 available on Dryad). This specialized morphology evolved independently in each one of the three lineages from ancestors that exhibited a generalist morphology. However, it is also evident from the remaining axes that grass-dwelling clades have explored partially different regions of morphospace. Figure 4. View largeDownload slide Variability among morphometric traits associated with the colonization of grasslands. Top, allometric scaling of traits, using carapace length as proxy for body size. Values are expressed as the logarithm of the species mean. Solid lines correspond to a PGLS$$_{\mathrm{\lambda }}$$ regression for species of general morphology (black circles), dotted lines correspond to the 95% prediction interval (all $$P< 10^{{-16}}$$; from left to right: $$R^{\mathrm{2}} =$$ 0.98, 0.91, 0.93; $$\lambda =$$ 0.512, 0.836, 0.918). At least one grass-dwelling species (other shapes) lies outside the expected interval in all cases. Bottom, phylomorphospaces on phylogenetic PCA axes, built using the same four traits. Axis pPCA 1 captures differences in size, with low values to the left. The rest represent differences in body proportions (see Supplementary Table S13 of SI File 1 available on Dryad). Figure 4. View largeDownload slide Variability among morphometric traits associated with the colonization of grasslands. Top, allometric scaling of traits, using carapace length as proxy for body size. Values are expressed as the logarithm of the species mean. Solid lines correspond to a PGLS$$_{\mathrm{\lambda }}$$ regression for species of general morphology (black circles), dotted lines correspond to the 95% prediction interval (all $$P< 10^{{-16}}$$; from left to right: $$R^{\mathrm{2}} =$$ 0.98, 0.91, 0.93; $$\lambda =$$ 0.512, 0.836, 0.918). At least one grass-dwelling species (other shapes) lies outside the expected interval in all cases. Bottom, phylomorphospaces on phylogenetic PCA axes, built using the same four traits. Axis pPCA 1 captures differences in size, with low values to the left. The rest represent differences in body proportions (see Supplementary Table S13 of SI File 1 available on Dryad). The relative support for the explored models of macroevolution are shown in Fig. 5. Overall, there was a relatively strong support for a 3 peak multi-OU model, favored across multiple trait histories on the mcc tree (Fig. 5, top) and multiple topologies sampled from the posterior distribution (Fig. 5, bottom). This macroevolutionary scenario (OU3 A–MS) models the history of morphological change in the tribe Gayennini using three multivariate adaptive peaks, with one peak assigned to all species displaying a generalist morphology, one occupied by grass-dwelling species of Arachosia, and a third one convergently encountered by grass-dwelling species of Monapia and Sanogasta. The difference in support between this model and the second most supported one (OU3 S–AM, also a multi-OU model with three peaks), was higher than the threshold value of 4 (Burnham and Anderson 2002) generally considered to represent a strong difference in support (average $$\Delta $$AICc $$=$$ 4.03 for replicates on the mcc tree, 4.22 for those in the sample of posterior topologies). Moreover, across all 200 replicates explored, we found that the addition of a third adaptive regime significantly improved the fit of the model 87% of the times, while the addition of a fourth one only resulted in a significant increase in fit 11% of the times (in both cases, $$P< 0.05$$ for LRT with four degrees of freedom). This further confirms that the repeated colonization of grasslands by spiders of the tribe Gayennini constitutes a scenario of partial convergence, with two different grassland-associated morphologies, one of which has evolved multiple times. Figure 5. View largeDownload slide Histograms of the relative support for each of the eight models of macroevolution tested. Support was measured as AICc weights in 100 replicates of stochastic character mapping histories on the mcc tree (top) and in one replicate in 100 randomly drawn topologies from the posterior sample of trees (bottom). In case of OU3 models, capital letters stand for grass-dwelling genera (A $$=$$Arachosia; M $$=$$Monapia; S $$=$$Sanogasta), and genera on the same side of the dash were modelled using the same selective regime. Values represent as median $$\pm$$ standard error. Figure 5. View largeDownload slide Histograms of the relative support for each of the eight models of macroevolution tested. Support was measured as AICc weights in 100 replicates of stochastic character mapping histories on the mcc tree (top) and in one replicate in 100 randomly drawn topologies from the posterior sample of trees (bottom). In case of OU3 models, capital letters stand for grass-dwelling genera (A $$=$$Arachosia; M $$=$$Monapia; S $$=$$Sanogasta), and genera on the same side of the dash were modelled using the same selective regime. Values represent as median $$\pm$$ standard error. A summary of the inferred optimal values for each trait (i.e., the $$\theta $$ parameters of the multi-OU model) for the three adaptive peaks can be found in Fig. 6. The morphological evolution of grass-dwelling Monapia and Sanogasta is best modelled as having suffered a tendency towards body size reduction, narrowing of the carapace and strong reduction of the distance between chelicerae, as well as a mild decrease in the length of the third pair of legs (although the confidence interval on this trait is very large). On the other hand, the adaptive peak explored by grass-dwelling Arachosia differs much less from the ancestral peak, and seems to entail a slight tendency towards body size increase and, most notably, a strong reduction in the distance between chelicerae. This last attribute therefore characterizes all three clades that have specialized in the exploitation of grassland niches. Figure 6. View largeDownload slide Inferred optimum values ($$\theta $$ parameters) for all traits in 200 replicates of the OU3 A-MS model. Values can also be interpreted as the approximate locations of the peaks in a multivariate adaptive landscape. Violin plots show the probability densities of the estimates after excluding the most extreme 5% of results, rectangles represent median values. Body size optima are shown in mm, other values are expressed as the ratio of each trait to body size, so as to represent inferred differences in proportions. Given the way the replicates were parameterized, uncertainty for the ancestral selective regime is minimal and not shown. Peaks for grass-dwelling clades are identified using the same codes for genera as in Fig. 5. Figure 6. View largeDownload slide Inferred optimum values ($$\theta $$ parameters) for all traits in 200 replicates of the OU3 A-MS model. Values can also be interpreted as the approximate locations of the peaks in a multivariate adaptive landscape. Violin plots show the probability densities of the estimates after excluding the most extreme 5% of results, rectangles represent median values. Body size optima are shown in mm, other values are expressed as the ratio of each trait to body size, so as to represent inferred differences in proportions. Given the way the replicates were parameterized, uncertainty for the ancestral selective regime is minimal and not shown. Peaks for grass-dwelling clades are identified using the same codes for genera as in Fig. 5. Most of the temporal dynamics of this adaptive process seems to have occurred at a relatively fast pace, with average phylogenetic half-lives of 0.71, 1.32, 2.97, and 131.03, for each of the four eigendimensions of the $$\alpha $$ matrix (Supplementary Fig. S10 of the SI File 4 available on Dryad). The first of these dimensions entails the relative reduction of all traits with respect to size (the same pattern as seen in pPCA axis 3), while the second one can be interpreted as changes in overall size (Supplementary Fig. S15 of the SI File 4 and Table S14 of SI File 1 available on Dryad). Discussion The global expansion of grasslands during the Miocene has impacted numerous animal taxa. For grass-specialized herbivores, the increase in their food source presented an ecological and evolutionary benefit; this is especially true for insect groups such as the leafhoppers and stem-borers, many of which have specialized in feeding on grasses (Dietrich 1999; Zahniser and Dietrich 2010; Toussaint et al. 2012) and whose diversification is temporally congruent with grassland expansions. With the increase in grassland herbivores and new habitats to exploit, predators such as falcons also underwent time-congruent diversification in the late Miocene (Fuchs et al. 2015). Similarly, the predators among arthropods such as spiders are expected to exploit the abundant insect biomass found in grasslands (Tscharntke and Greiler 1995). Based on the results of this study, the habitat changes from forest to open habitats in amaurobioidine spiders occurred during the mid- to late Miocene, before which (ca. 10 Ma) there were proportionally fewer lineages than expected in open habitats. These dates are congruent with the increased expansion of the South American grassland biome, especially in Pg (Barreda and Palazzesi 2007; Palazzesi and Barreda 2012) suggesting that, as open habitats started expanding, amaurobiodine spiders readily occupied the newly available niches. The node age estimates obtained in this study are plausible, based on the similar ages obtained from previous studies with different taxon sampling (Ceccarelli et al. 2016a; Soto et al. 2017) as well as geological age corroboration with the volcanic JF (Stuessy et al. 1984) and the age of the MRCAs of the Sanogasta and Philisca species endemic to those islands. These two colonization events of Juan Fernández, plus the trans-oceanic dispersal of Amaurobioides probably by oceanic drift with the Antarctic circumpolar current (Ceccarelli et al. 2016a) are evidence for the long-distance dispersal ability of some amaurobioidine spiders, yet overall, a strong historical component is still preserved in their geographic distribution. While the Amaurobioidinae are probably of NP origin, the A$$+$$G clade originated and diversified in forested Patagonian habitats. South American orogeny must also be considered as an important factor for the diversification and movement of species, and for the evolution of habitats; for example, during the last 10 Ma the Andean uplift also had a profound effect on structuring both Patagonian and other South American fauna (e.g., Sersic et al. 2011; Ceccarelli et al. 2016b). Based on our study, the interchanges of fauna between Pg and the non-Patagonian Neotropics occurred predominantly in open habitats. This is consistent with the prediction that biome shifts are more probable over larger shared boundaries (Donoghue and Edwards 2014), since the boundary between these biogeographic areas is constituted by open habitats, with the forested areas now separated by ample expanses of arid lands (Fig. 3e). It is also noteworthy that the initial flow of species out of Patagonian forests became dynamic shortly after the expansion of grasslands, with a backflow of lineages returning to Pg, or expanding back to forest habitats. Some species of Amaurobioidinae seem especially suited for life on grasses due to morphologies that evolved at least three independent times among lineages already inhabiting open ecosystems in the non-Patagonian Neotropics. Although grasses are common in the understory of forests, grass-dwelling lineages originated exclusively in open habitats, likely a consequence of the difference in prey biomass harbored by grasslands compared with forest grasses. The traits involved in this specialization are all related with the need to fit tightly on a single blade of grass and minimize visual cues to predators, moving and building retreats in the narrow spaces between grass blades, and might therefore be considered morphological adaptations to the peculiar shape and disposition of the graminean leaves. The association between these morphologies and the strict exploitation of grasslands, their departure from the allometric scaling shown by close relatives living in different habitats, as well as their inferred origin only after grass-dominated ecosystems spread all warrant their recognition as adaptations to grasslands. Many shifts from forest to open habitats are inferred to have occurred at about the same time as grasslands were expanding, which in conjunction with the strong phylogenetic signal found in the habitat traits, suggest that there is an intermediate degree of niche conservatism (see Wiens et al. 2010). On the other hand, the strict adaptation to living on grasses is a much more conservative trait, and in at least two cases seems to have originated only long after the colonization of open habitats. The MRCA of grassland-adapted Arachosia and Sanogasta is inferred to have already lived in open habitats some 13 Ma, while adaptations to grasslands are unlikely to have originated any time before 6 Ma; in Monapia the delay is less marked (about 10 Ma for shift to open habitat and 7 Ma for grass-adaptation). This delay in the start of the process of grassland specialization might indicate that the specific microhabitats employed by grass-dwelling ghost spiders—grasslands dominated by large grass species in regions of high humidity and mild temperatures—did not become common until later in the history of the region. The current paleoenvironmental reconstructions in the region (e.g., Barreda et al. 2007) are still too coarse to trace the origin of this specific environment. Nonetheless, based on our results, grassland ecomorphs originated very rapidly once the selective process was underway. The half-lives for the first two eigendimensions, which seem to capture the transition from general morphology to grass ecomorph, as well as changes in overall size, are on the timescale of approximately 1 Ma. This is a relatively short amount of time compared with the results of other studies inferring selective processes on macroevolutionary timescales (Uyeda et al. 2011; Hansen 2012; Toljagić et al. 2018), although the difference in generation times between taxa must also be considered. This rapid adaptation is even more remarkable considering the complexity of the morphological changes involved, which likely involved the accommodation of multiple constraints and trade-offs. For example, it has been shown that the slender abdomen of ant-mimicking spiders increases their resemblance to their model ants at the expense of smaller egg clutches compared with other, stockier salticid spiders (see Cushing 1997). The expansion of grasslands in South America resulted in an increase in complexity of the macroevolutionary adaptive landscape of amaurobioidine spiders, setting them off on previously unexplored evolutionary pathways. The morphology of two of the three genera of grass-adapted spiders is best modelled as the result of evolution under the same selective regime, characterized by body size reduction, narrowing of the carapace, strong reduction of the distance between chelicerae, as well as a mild decrease in the length of the third pair of legs. This can therefore be considered an instance of adaptive morphological convergence driven by habitat, a thoroughly-studied phenomenon in evolutionary biology (e.g., Gillespie 2004; Losos 2011; Muschick et al. 2012; Thiele 2012). Nonetheless, the evolution of a second grassland ecomorph shows that not all evolutionary trajectories associated with grassland specialization have resulted in the same end-product. Grass-adapted species of Arachosia have evolved towards a different adaptive peak, characterized by a longer body and a reduced distance between chelicerae. The fact that replicated events of adaptation to similar environmental conditions fail to converge is considered evidence of the importance of historical contingency in evolutionary processes (Seehausen 2007; Blount et al. 2008; Young et al. 2009). On the other hand, although there is strong statistical support for grassland-adapted amaurobioidine spiders evolving under two different selective regimes, they are both characterized by a common trend: a reduction of the distance between the chelicerae. This slender configuration at the anterior region of the prosoma, which originated at least three independent times, seems to be a necessary condition for the successful colonization of grasslands. At least at this level, adaptation has followed an entirely deterministic route. In conclusion, based on the results of our analyses, it is clear that the evolutionary history of Amaurobioidinae has been shaped by recurrent colonization of open habitats, followed by events of diversification and adaptation. Even considering some of the shortcomings of the phylogenies upon which these conclusions are based, such as poor resolution in some clades, the nodes corresponding to grass-adapted clades are all well-supported. Furthermore, sampling over a subset of posterior trees allowed us to test these hypotheses in light of alternative topologies. We can therefore conclude that these spiders are a group of organisms for which the expansion of grasslands in South America presented an ecological and evolutionary opportunity. The evolution of amaurobioidine spiders shows a traceable history of convergent, yet diverse morphological adaptations in response to changes in the environment, the colonization of new habitats and regions and a dynamic flow of lineages. We have shown here that a combination of phylogenetic, morphological, and biogeographic analyses helps to untangle and understand the multiple factors and levels affecting the evolution of organisms and the biomes they inhabit. Supplementary Material Data available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.20257. Funding This work was supported by the Agencia Nacional de Promoción Científica y Tecnológica Argentina (ANPCyT, grant numbers PICT 2011-1007, PICT 2015-283), and the International Barcode of Life Project (iBOL) Argentina 2012 from the National Scientific and Technical Research Council (CONICET) Argentina. Additional funds were provided by the Spanish Ministerio de Ciencia e Innovación (MA, grant number CGL2009-07639) and from The National Geographic Society (Gustavo Hormiga, MA, grant number 8969-11). Acknowledgments We thank the MACN Barcoding Laboratory for the amplification of DNA and to the Canadian Centre of DNA Barcoding for the sequencing of some of the COI samples used here. Collection permits were provided by Corporación Nacional Forestal (Chile), Administración de Parques Nacionales and Secretaría de Desarrollo Sustentable y Ambiente de Tierra del Fuego (Argentina). Guillermo Martínez Pastur and Vanessa Lencinas of the Centro Austral de Investigaciones Científicas (CADIC) provided logistic support for collecting in Tierra del Fuego, Federico Gerrero and Laura Smith (Quixote Expeditions) generously brought MJR to Isla de los Estados in the Ocean Trump vessel, Jaime Pizarro Araya (Universidad de La Serena) provided support for fieldwork in Chile. Additional specimens were provided by Charles Griswold (California Academy of Sciences) and Ivan Magalhães (MACN). Lorena Zapata helped with the identification and preparation of Sanogasta specimens, and Gonzalo Rubio with those of Arachosia. Ivan Magalhães, Abel Pérez González, Andrés Ojanguren provided useful comments in early versions of this manuscript. Julien Clavel and Thomas F. Hansen provided useful guidance relating to macroevolutionary model fitting and interpretation. References Adams D.C 2014 . Quantifying and comparing phylogenetic evolutionary rates for shape and other high-dimensional phenotypic data . Syst. Biol. 63 2 : 166 – 177 . Google Scholar CrossRef Search ADS PubMed Adams D.C., Collyer M. L 2018 . Multivariate phylogenetic comparative methods: evaluations, comparisons, and recommendations . Syst. Biol. 67 1 : 14 – 31 . Google Scholar CrossRef Search ADS PubMed Barreda V., Anzótegui L.M., Prieto A.R., Aceñolaza P., Bianchi M.M., Borromei A.M., Brea M., Caccavari M., Cuadrado G.A., Garralla S., Grill S., Guerstein G.R., Lutz A.I., Mancini V., Mautino L.R., Ottone E.G., Quattrocchio M.E., Romero E.J., Zamaloa M.C., Zucol A 2007 . Diversificación y cambios de las angiospermas durante el Neógeno en Argentina . Publicación Especial 11 ISSN 0328-347X Ameghiniana 50° aniversario. 11 : 173 – 191 . Barreda V., Palazzesi L 2007 . Patagonian vegetation turnovers during the Paleogene-early Neogene: origin of arid-adapted floras . Bot. Rev. 73 1 : 31 – 50 . Google Scholar CrossRef Search ADS Bartoszek K., Pienaar J., Mostad P., Andersson S., Hansen T.F 2012 . A phylogenetic comparative method for studying multivariate adaptation . J. Theor. Biol. 314 : 204 – 215 . Google Scholar CrossRef Search ADS PubMed Blomberg S.P., Garland T Jr. , Ives A.R . 2003 . Testing for phylogenetic signal in comparative data: behavioral traits are more labile . Evolution 57 4 : 717 – 745 . Google Scholar CrossRef Search ADS PubMed Blount Z.D., Borland C.Z., Lenski R.E 2008 . Historical contingency and the evolution of a key innovation in an experimental population of Escherichia coli . Proc. Natl Acad. Sci. USA 105 23 : 7899 – 7906 . Google Scholar CrossRef Search ADS Bobe R., Behrensmeyer A.K 2004 . The expansion of grassland ecosystems in Africa in relation to mammalian evolution and the origin of the genus Homo . Palaeogeogr. Palaeoclimatol. Palaeoecol. 207 : 399 – 420 . Google Scholar CrossRef Search ADS Boettiger C., Coop G., Ralph P 2012 . Is your phylogeny informative? Measuring the power of comparative methods . Evolution 66 7 : 2240 – 2251 . Google Scholar CrossRef Search ADS PubMed Bollback J.P 2006 . SIMMAP: stochastic character mapping of discrete traits on phylogenies . BMC Bioinformatics 7 : 88 . Google Scholar CrossRef Search ADS PubMed Bouchenak-Khelladi Y., Muasya A.M., Linder H.P 2014 . A revised evolutionary history of Poales: origins and diversification . Bot. J. Linn. Soc. 175 1 : 4 – 16 . Google Scholar CrossRef Search ADS Bouchenak-Khelladi Y., Verboom G.A., Hodkinson T.R., Salamin N., Francois O., Ni Chonghaile G., Savolainen V 2009 . The origins and diversification of C4 grasses and savanna-adapted ungulates . Glob. Change Biol. 15 10 : 2397 – 2417 . Google Scholar CrossRef Search ADS Burnham K.P., Anderson D.R 2002 . Model selection and multimodel Inference . New York : Springer . Butler M.A., King A.A 2004 . Phylogenetic comparative analysis: a modeling approach for adaptive evolution . Am. Nat. 164 6 : 683 – 695 . Google Scholar CrossRef Search ADS PubMed Ceccarelli F.S., Ojanguren-Affilastro A.A., Ramírez M.J., Ochoa J.A., Mattoni C.I., Prendini L 2016b . Andean uplift drives diversification of the bothriurid scorpion genus Brachistosternus . J. Biogeogr. 43 10 : 1942 – 1954 . Google Scholar CrossRef Search ADS Ceccarelli F.S., Opell B.D., Haddad C.R., Raven R.J., Soto E.M., Ramírez M.J 2016a . Around the world in eight million years: historical biogeography and evolution of the spray zone spider Amaurobioides (Araneae: Anyphaenidae) . PloS One 11 10 : e0163740 . Google Scholar CrossRef Search ADS Clavel J., Escarguel G., Merceron G 2015 . mvMORPH: An R package for fitting multivariate evolutionary models to morphometric data . Method Ecol. Evol. 6 11 : 1311 – 1319 . Google Scholar CrossRef Search ADS Comstock J.H 1913 . The spider book. Doubleday, Page , New York, NY, USA : Company . Conway Morris S 2003 . Life’s solution: inevitable humans in a lonely universe . Cambridge : Cambridge University Press . Google Scholar CrossRef Search ADS Cooper N., Thomas G.H., Venditti C., Meade A., Freckleton R.P 2016 . A cautionary note on the use of Ornstein–Uhlenbeck models in macroevolutionary studies . Biol. J. Linn. Soc. 118 1 : 64 – 77 . Google Scholar CrossRef Search ADS Cressler C.E., Butler M.A., King A.A 2015 . Detecting adaptive evolution in phylogenetic comparative analysis using the Ornstein–Uhlenbeck model . Syst. Biol. 64 6 : 953 – 968 . Google Scholar CrossRef Search ADS PubMed Cushing P.E 1997 . Myrmecomorphy and myrmecophily in spiders: a review . Fla Entomol. 80 : 165 – 193 . Google Scholar CrossRef Search ADS Dietrich C.H 1999 . The role of grasslands in the diversification of leafhoppers (Homoptera: Cicadellidae): a phylogenetic perspective. In Proceedings of the Fifteenth North American Prairie Conference (pp. 44 – 49 ). Bend, OR : Natural Areas Assoc . Dippenaar-Schoeman A.S 1984 . The crab-spiders of southern Africa (Araneae: Thomisidae) . 4. The genus Monaeses Thorell, 1869. Phytophylactica 16 : 101 – 116 . Donoghue M.J., Edwards E.J 2014 . Biome shifts and niche evolution in plants . Annu. Rev. Ecol. Evol. Syst. 45 : 547 – 572 . Google Scholar CrossRef Search ADS Edmunds M 1990 . The evolution of cryptic coloration. In: Insect defenses: adaptive mechanisms and strategies of prey and predators . Evans D.L., Schmidt J.O Eds. Albany, N.Y., USA : State University of New York Press . p. 3 – 21 . Edwards E.J., Osborne C.P., Strömberg C.A., Smith S.A ; C4 Grasses Consortium . 2010 . The origins of C4 grasslands: integrating evolutionary and ecosystem science . Science 328 5978 : 587 – 591 . Google Scholar CrossRef Search ADS PubMed Freckleton R.P., Harvey P.H., Pagel M 2002 . Phylogenetic analysis and comparative data: a test and review of evidence . Am. Nat. 160 6 : 712 – 726 . Google Scholar CrossRef Search ADS PubMed Fuchs J., Johnson J.A., Mindell D.P 2015 . Rapid diversification of falcons (Aves: Falconidae) due to expansion of open habitats in the Late Miocene . Mol. Phylogenet. Evol. 82 : 166 – 182 . Google Scholar CrossRef Search ADS PubMed Garland T Jr. , Ives A.R . 2000 . Using the past to predict the present: confidence intervals for regression equations in phylogenetic comparative methods . Am. Nat. 155 3 : 346 – 364 . Google Scholar CrossRef Search ADS PubMed Gibson D.J 2009 . Grasses and grassland ecology . New York : Oxford University Press p. 305 . Gillespie R 2004 . Community assembly through adaptive radiation in Hawaiian spiders . Science 303 : 356 – 359 . Google Scholar CrossRef Search ADS PubMed Gould S.J 1989 . Wonderful life: the burgess shale and the nature of life . New York : W.W. Norton . Grafen A 1989 . The phylogenetic regression . Philos. Trans. R. Soc. Lond. B Biol. Sci. 326 1233 : 119 – 157 . Google Scholar CrossRef Search ADS PubMed Hansen T.F 1997 . Stabilizing selection and the comparative analysis of adaptation . Evolution 51 : 1341 – 1351 . Google Scholar CrossRef Search ADS PubMed Hansen T.F 2012 . Adaptive landscapes and macroevolutionary dynamics . In: Svensson E., Calsbeek R , editors. The adaptive landscape in evolutionary biology . Oxford, UK : Oxford University Press . p. 205 – 226 . Google Scholar CrossRef Search ADS Ho L.S.T., Ané C 2014 . Intrinsic inference difficulties for trait evolution with Ornstein–Uhlenbeck models . Method Ecol. Evol. 5 11 : 1133 – 1146 . Google Scholar CrossRef Search ADS Huelsenbeck J.P., Nielsen R., Bollback J.P 2003 . Stochastic mapping of morphological characters . Syst. Biol. 52 2 : 131 – 158 . Google Scholar CrossRef Search ADS PubMed Huelsenbeck J.P., Rannala B., Masly J.P 2000 . Accommodating phylogenetic uncertainty in evolutionary studies . Science 288 5475 : 2349 – 2350 . Google Scholar CrossRef Search ADS PubMed Ingram T., Mahler D.L 2013 . SURFACE: detecting convergent evolution from comparative data by fitting Ornstein–Uhlenbeck models with stepwise Akaike Information Criterion . Method Ecol. Evol. 4 5 : 416 – 425 . Google Scholar CrossRef Search ADS Jacobs B.F., Kingston J.D., Jacobs L.L 1999 . The origin of grass-dominated ecosystems . Ann. Missouri Bot. Garden 86 : 590 – 643 . Google Scholar CrossRef Search ADS Jardine P.E., Janis C.M., Sahney S., Benton M.J 2012 . Grit not grass: concordant patterns of early origin of hypsodonty in Great Plains ungulates and Glires . Palaeogeogr. Palaeoecol. 365 : 1 – 10 . Google Scholar CrossRef Search ADS Labarque F.M., Soto E.M., Ramírez M.J., Arnedo M.A 2015 . Chasing ghosts: the phylogeny of Amaurobioidinae ghost spiders (Araneae, Anyphaenidae) . Zool. Scr. 44 : 550 – 561 . Google Scholar CrossRef Search ADS Landis M.J., Matzke N.J., Moore B.R 2013 . Bayesian analysis of biogeography when the number of areas is large . Syst. Biol. 62 : 789 – 804 . Google Scholar CrossRef Search ADS PubMed Landry J.F 1994 . Resource partitioning in a guild of marsh-dwelling Agonum (Coleoptera: Carabidae) in central Alberta . Can. Entomol. 126 03 : 709 – 728 . Google Scholar CrossRef Search ADS Losos J.B 2011 . Convergence, adaptation, and constraint . Evolution 65 7 : 1827 – 1840 . Google Scholar CrossRef Search ADS PubMed Maddison W.P., Maddison D.R 2015 . Mesquite: a modular system for evolutionary analysis . Version 3.04. Available from: URL http://mesquiteproject.org . Mahler D.L., Ingram T., Revell L.J., Losos J.B 2013 . Exceptional convergence on the macroevolutionary landscape in island lizard radiations . Science 341 6143 : 292 – 295 . Google Scholar CrossRef Search ADS PubMed Martins E.P., Hansen T.F 1997 . Phylogenies and the comparative method: a general approach to incorporating phylogenetic information into the analysis of interspecific data . Am. Nat. 149 4 : 646 – 667 . Google Scholar CrossRef Search ADS Matzke N.J 2012 . Founder-event speciation in BioGeoBEARS package dramatically improves likelihoods and alters parameter inference in Dispersal-Extinction-Cladogenesis (DEC) analyses . Frontiers Biogeogr. 4 : 210 . Matzke N.J 2013 . BioGeoBEARS: BioGeography with Bayesian (and likelihood) evolutionary analysis in R scripts . Berkeley (CA) : University of California . Mikulska I 1967 . Some observations on the biology of the spider Tibellus oblongus (Walck.). Przegląd Zoologiczny 11 : 388 – 391 . Monteiro L.R 2013 . Morphometrics and the comparative method: studying the evolution of biological shape. Hystrix It. J. Mamm. 24 1 : 25 – 32 . Morrone J.J 2006 . Biogeographic areas and transition zones of latin america and the caribbean islands based on panbiogeographic and cladistic analyses of the entomofauna. Annu. Rev. Entomol. 51 : 467 – 494 . Google Scholar CrossRef Search ADS PubMed Morrone J.J 2015 . Biogeographical regionalisation of the Andean region. Zootaxa 3936 : 207 – 236 . Google Scholar CrossRef Search ADS PubMed Muschick M , Indermaur A , Salzburger W 2012 . Convergent evolution within an adaptive radiation of cichlid fishes. Curr. Biol. 22 24 : 2362 – 2368 . Google Scholar CrossRef Search ADS PubMed O’Meara B.C , Ané C , Sanderson M.J , Wainwright P.C 2006 . Testing for different rates of continuous trait evolution using likelihood. Evolution 60 5 : 922 – 933 . Google Scholar CrossRef Search ADS PubMed O’Meara B.C , Beaulieu J.M 2014 . Modelling stabilizing selection: The attraction of Ornstein–Uhlenbeck models. In: Garamszegi L.Z , editor. Modern phylogenetic comparative methods and their application in evolutionary biology . Berlin Heidelberg : Springer , p. 381 – 393 . Google Scholar CrossRef Search ADS Orgogozo V 2015 . Replaying the tape of life in the twenty-first century. Interface Focus 5 6 : 20150057 . Google Scholar CrossRef Search ADS PubMed Orme D , Freckleton R , Thomas G , Petzoldt T , Fritz S , Isaac N , Pearce W 2013 . caper: comparative analyses of phylogenetics and evolution in R. R package version 0.5.2. Available from: URL https://CRAN.R-project.org/package=caper . Osborne C.P 2008 . Atmosphere, ecology and evolution: what drove the Miocene expansion of C4 grasslands? J. Ecol. 96 : 35 – 45 . Google Scholar PubMed Pagel M 1999 . Inferring the historical patterns of biological evolution. Nature 401 6756 : 877 – 884 . Google Scholar CrossRef Search ADS PubMed Palazzesi L , Barreda V 2012 . Fossil pollen records reveal a late rise of open-habitat ecosystems in Patagonia. Nat. Commun. 3 : 1294 . Google Scholar CrossRef Search ADS PubMed Parker J , Rambaut A , Pybus O.G 2008 . Correlating viral phenotypes with phylogeny: accounting for phylogenetic uncertainty. Infect. Genet. Evol. 8 : 239 – 46 . Google Scholar CrossRef Search ADS PubMed R Core Team . 2015 . R: a language and environment for statistical computing . Vienna (Austria) : R Foundation for Statistical Computing . Available from: URL https://www.R-project.org . Ramírez M.J 1999 . New species and cladistic reanalysis of the spider genus Monapia (Araneae, Anyphaenidae, Amaurobioidinae). J. Arachnol. 27 : 415 – 431 . Ramírez M.J 2003 . The spider subfamily Amaurobioidinae (Araneae, Anyphaenidae): a phylogenetic revision at the generic level. Bull. Am. Mus. Nat. Hist. 277 : 1 – 262 . Google Scholar CrossRef Search ADS Ree R.H , Moore B.R , Webb C.O , Donoghue M.J 2005 . A likelihood framework for inferring the evolution of geographic range on phylogenetic trees. Evolution 59 : 2299 – 2311 . Google Scholar CrossRef Search ADS PubMed Revell L.J 2009 . Size-correction and principal components for interspecific comparative studies. Evolution 63 12 : 3258 – 3268 . Google Scholar CrossRef Search ADS PubMed Revell L.J 2010 . Phylogenetic signal and linear regression on species data. Method. Ecol. Evol. 1 4 : 319 – 329 . Google Scholar CrossRef Search ADS Revell L.J 2012 . phytools: an R package for phylogenetic comparative biology (and other things). Method. Ecol. Evol. 3 : 217 – 223 . Google Scholar CrossRef Search ADS Ronquist F 1997 . Dispersal-vicariance analysis: a new approach to the quantification of historical biogeography. Syst. Biol. 46 : 195 – 203 . Google Scholar CrossRef Search ADS Rubio G.D , Ramírez M.J 2015 . Taxonomic revision of the American spider genus Arachosia (Araneae: Anyphaenidae). Zootaxa 3932 : 1 – 105 . Google Scholar CrossRef Search ADS PubMed Seehausen O 2007 . Chance, historical contingency and ecological determinism jointly determine the rate of adaptive radiation. Heredity 99 : 361 – 363 . Google Scholar CrossRef Search ADS PubMed Sersic A.N , Cosacov A , Cocucci A.A , Johnson L.A , Pozner R , Avila L.J , Sites J Jr. , Morando M . 2011 . Emerging phylogeographical patterns of plants and terrestrial vertebrates from Patagonia. Biol. J. Linn. Soc. 103 2 : 475 – 494 . Google Scholar CrossRef Search ADS Sidlauskas B 2008 . Continuous and arrested morphological diversification in sister clades of characiform fishes: a phylomorphospace approach. Evolution 62 12 : 3135 – 3156 . Google Scholar CrossRef Search ADS PubMed Silvestro D , Kostikova A , Litsios G , Pearman P.B , Salamin N. 2015 . Measurement errors should always be incorporated in phylogenetic comparative analysis. Method. Ecol. Evol. 6 3 : 340 – 346 . Google Scholar CrossRef Search ADS Smaers J.B , Rohlf F.J 2016 . Testing species’ deviation from allometric predictions using the phylogenetic regression. Evolution 70 5 : 1145 – 1149 . Google Scholar CrossRef Search ADS PubMed Soriano A 1991 . Río de la Plata grasslands. In: Coupland R.T , editor, Natural grasslands, introduction and western hemisphere . Amsterdam : Elsevier . p. 367 – 407 . Soto E.M , Labarque F.M , Ceccarelli F.S , Arnedo M.A , Pizarro-Araya J , Ramírez M.J 2017 . The life and adventures of an eight-legged castaway: Colonization and diversification of Philisca ghost spiders on Robinson Crusoe Island (Araneae, Anyphaenidae). Mol. Phylogenet. Evol. 107 : 132 – 141 . Google Scholar CrossRef Search ADS PubMed Soto E.M , Ramírez M.J 2012 . Revision and phylogenetic analysis of the spider genus Philisca Simon (Araneae: Anyphaenidae, Amaurobioidinae). Zootaxa 3443 : 1 – 65 . Stebbins G.L 1981 . Coevolution of grasses and herbivores. Ann. Mo. Bot. Gard. 68 : 75 – 86 . Google Scholar CrossRef Search ADS Stevens M , Merilaita S 2009 . Animal camouflage: current issues and new perspectives. Philos. Trans. R. Soc. B Biol. Sci. 364 : 423 – 427 . Google Scholar CrossRef Search ADS Strömberg C.A.E 2011 . Evolution of grasses and grassland ecosystems. Annu. Rev. Earth Planet. Sci. 39 : 517 – 544 . Google Scholar CrossRef Search ADS Strömberg C.A.E , Dunn R.E , Madden R.H , Kohn M.J , Carlini A.A 2013 . Decoupling the spread of grasslands from the evolution of grazer-type herbivores in South America. Nat. Commun. 4 : 1478 . Google Scholar CrossRef Search ADS PubMed Stuessy T.F , Foland K.A , Sutter J.F , Sanders R.W , Silva O.M 1984 . Botanical and geological significance of potassium-argon dates from the Juan Fernández Islands. Science, New Series 225 4657 : 49 – 51 . Thiele H.U 2012 . Carabid beetles in their environments: a study on habitat selection by adaptations in physiology and behavior , Vol. 10 . Berlin, Germany : Springer Science & Business Media . Toljagić O , Voje K.L , Matschiner M , Liow L.H , Hansen T.F 2018 . Millions of years behind: slow adaptation of ruminants to grasslands. Syst. Biol. 67 1 : 145 – 157 . Google Scholar CrossRef Search ADS PubMed Toussaint E.F.A , Condamine F.L , Kergoat G.J , Capdevielle-Dulac C , Barbut J , Silvain J.F , Le Ru B.P 2012 . Palaeoenvironmental Shifts Drove the Adaptive Radiation of a Noctuid Stemborer Tribe (Lepidoptera, Noctuidae, Apameini) in the Miocene. PLoS One 7 7 : e41377 . Google Scholar CrossRef Search ADS PubMed Tscharntke T , Greiler H.-J 1995 . Insect communities, grasses, and grasslands. Ann. Rev. Entomol. 40 : 535 – 558 . Google Scholar CrossRef Search ADS Turner D.D 2011 . Gould’s replay revisited. Biol. Philos. 26 1 : 65 – 79 . Google Scholar CrossRef Search ADS Uyeda J.C , Caetano D.S , Pennell M.W 2015 . Comparative analysis of principal components can be misleading. Syst. Biol. 64 4 : 677 – 689 . Google Scholar CrossRef Search ADS PubMed Uyeda J.C , Hansen T.F , Arnold S.J , Pienaar J 2011 . The million-year wait for macroevolutionary bursts. Proc. Natl Acad. Sci. USA 108 38 : 15908 – 15913 . Google Scholar CrossRef Search ADS Wainwright P.C , Reilly S.M 1994 . Ecological morphology: integrative organismal biology . Chicago, Il, USA : University of Chicago Press . Wang T.H , Donaldson Y.K , Brettle R.P , Bell J.E , Simmonds P 2001 . Identification of shared populations of Human immunodeficiency Virus Type 1 infecting microglia and tissue macrophages outside the central nervous system. J. Virol . 75 23 : 11686 – 11699 . Google Scholar CrossRef Search ADS PubMed Wheeler W.C , Coddington J.A , Crowley L.M , Dimitrov D , Goloboff P.A , Griswold C.E , Hormiga G , Prendini L , Ramírez M.J , Sierwald P , Almeida-Silva L , Alvarez-Padilla F , Arnedo M.A , Benavides Silva L.R , Benjamin S.P , Bond J.E , Grismado C.J , Hasan E , Hedin M , Izquierdo M.A , Labarque F.M , Ledford J , Lopardo L , Maddison W.P , Miller J.A , Piacentini L.N , Platnick N.I , Polotow D , Silva-Dávila D , Scharff N , Szuts T , Ubick D , Vink C.J , Wood H.M , Zhang X 2017 . The spider tree of life: phylogeny of Araneae based on target-gene analyses from an extensive taxon sampling. Cladistics 33 6 : 574 – 616 . Google Scholar CrossRef Search ADS Wiens J.J , Ackerly D.D , Allen A.P , Anacker B.L , Buckley L.B , Cornell H.V , Damschen E.I , Jonathan Davies T , Grytnes J.A , Harrison S.P , Hawkins B.A , Holt R.D , McCain C.M , Stephens P.R . 2010 . Niche conservatism as an emerging principle in ecology and conservation biology. Ecol. Lett. 13 10 : 1310 – 1324 . Google Scholar CrossRef Search ADS PubMed World Spider Catalog . 2017 . World spider catalog. Natural History Museum Bern. Available from: URL http://wsc.nmbe.ch, version 18.0, accessed on 12 June 2017 . Young K.A , Snoeks J , Seehausen O 2009 . Morphological diversity and the roles of contingency, chance and determinism in African cichlid radiations. PLoS One 4 3 : e4740 . Google Scholar CrossRef Search ADS PubMed Zahniser J.N , Dietrich C.H 2010 . Phylogeny of the leafhopper subfamily Deltocephalinae (Hemiptera: Cicadellidae) based on molecular and morphological data with a revised family—group classification. Syst. Entomol. 35 3 : 489 – 511 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press, on behalf of the Society of Systematic Biologists. All rights reserved. For Permissions, please email: journals.permissions@oup.com 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)

Journal

Systematic BiologyOxford University Press

Published: May 19, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
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

Explore the DeepDyve Library

Search

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

Organize

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

Access

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

Your journals are on DeepDyve

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.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off