«Taxonomic Revision, Molecular Phylogeny and Zoogeography of the huntsman spider genus Eusparassus (Araneae: Sparassidae) Dissertation for attaining ...»
3.1. Comparison of different alignment strategies and phylogenetic analyses Different taxon combinations, alignment strategies and reconstruction methods were used to evaluate how they influenced comparable phylogenetic results. In total 20 combinations were investigated. For comparison corresponding data were mapped upon our preferred topologies of the All-outgroups-8-Sparassidae set (Fig. 2) and the Full-ingroup set (Fig. 3), both resulting from ML reconstructions and applying MAFFT alignment.
Including taxa within and outside the RTA-clade significantly increased the number of gapped sites within the 28S and 16S partitions of our alignments, while Gblocks gradually excluded more of these heterogeneous positions. This caused the initial MAFFT alignments to increase in length, but Gblocks to remove larger portions, such that final alignments incorporating more heterogeneous taxa were significantly shorter than before (Table 4). Against expectations, removing outgroup taxa to reduce the amount of alignment uncertainty in the RTAclade for a clearer phylogenetic signal failed to achieve such results (RTA-out-8-Sparassidae set).
While support in the basal nodes of the RTA-clade was generally weak, analyses using multiple outgroups recovered comparable topologies and yielded higher bootstrap values and posterior probabilities (Fig. 2). Only in a few cases and in shallow divergent clades (e.g. within Eusparassus walckenaeri (Audouin, 1826) and E. dufouri Simon, 1932), support values were found to significantly increase when reducing outgroup taxa (Fig. 3).
Other manipulations more fundamentally influenced the analyses. Removing large portions of 28S and 16S apparently caused significant loss of information on ingroup relationships, as seen in the frequent lower support of Gblocked analyses, especially for the gap-rich All-taxa set (Fig. 3).
Results: chapter 3.3: Molecular phylogenyThis cause is also likely for lower support seen in some outgroup taxa relationships (Fig. 2), but to a much lower extent than seen in the ingroup. In terms of methods we frequently found MrBayes to better support clades than RAxML. MrBayes also showed high support in some reconstructions that did not after processing with Gblocks. No case of high MrBayes support recovered a different phylogenetic constellation than RAxML, as for which both methods remained comparable in supporting the same topology. However, interpretation of results in the absence of, at least sufficient, RAxML support was cautious.
Some results also indicate the need to add more data, e.g. as seen in the values displayed throughout the Clubionoidea+Gnaphosoidea-clade (Fig. 2). Here, variations of the number of outgroup taxa or exclusion of one Clubionidae taxon, as well as cleaning alignments via Gblocks, significantly changed the support for this clade.
In summary no general preference towards a certain combination of taxa, alignment criterion or phylogenetic method is obvious. Our strategy thus proved to be reasonable in recovering differences between single reconstructions, in testing the robustness of our data and in allowing a comparison of phylogenetic results and interpretation of alternatives.
To state a mere observation, ML and BI results showed different performance in terms of supporting comparable phylogenetic splits. While ML was perceived to gradually increase support with increasing strength of the phylogenetic signal, rarely attaining maximum support, BI often quickly and fully supported respective splits. This “all or nothing” behaviour caused BI to support a considerable proportion of splits that received much lower support, often below significance, for ML. However, no case of high MrBayes support recovered a different phylogenetic constellation than RAxML, as for which both methods remained comparable in supporting the same topology.
Results: chapter 3.3: Molecular phylogeny
FIGURE 2. Maximum Likelihood tree with all outgroup taxa and a reduced Sparassidae set using MAFFT alignment (All-outgroups-8-Sparassidae set).
Results of other analyses are mapped upon this tree, summarizing the overall support for each branch. Bold, black or thin, dashed branches without additional
Results: chapter 3.3: Molecular phylogenyFig. 2.continued: table indicate respective support in all compared analyses, with bold, black indicating high support (87bt, 1.00pp), and thin, dashed no considerable support (69bt, 0.94pp). For other branches support is given in detail by the associated tables, branch appearance (bold, or shaded) summarises support by at least 2/3 of the analyses and grey indicates sufficiently support (69bt,
0.94pp). Spider pictures indicate the position of laterigrade groups within the tree and letter “D” indicate the position of Dionycha members; abbreviations Phil. (Philodromidae), Sele. (Selenopidae), Salt.
(Salticidae), Cor. (Corinnidae). Terminal branches of Clubiona japonicola and the Hahniidae sequence mix were artificially shortened to 1/3 of their actual length.
3.2. RTA-clade, ‘Laterigradae’ and the position of Sparassidae
The RTA-clade was recovered as a monophyletic group containing Sparassidae in all of our analyses. The monophyly of the RTA-clade was also supported by the analyses of Spagna and Gillespie (2008) and Miller et al. (2010), although these authors did not include Sparassidae. In analyses with all outgroups (Fig. 2), MrBayes consistently supported Sparassidae as sister to all other RTA-clade members, which was only topologically recovered by RAxML. While not fully supported, this still hints towards a basal position of Sparassidae in the RTA-clade. A similar topology of Sparassidae within RTA-clade was recovered by Agnarsson et al. (2013b) with weak support, probably due to the considerable amount of missing data as mentioned by these authors.
Sparassidae and the rest of the ‘Laterigradae’ are presently considered as “unplaced families” within the higher level clade Dionycha (Coddington, 2005). Yet, in previous molecular studies, Dionycha was not the focus of investigation (Miller et al., 2010; Agnarsson et al., 2013a).
Primarily interested in the position of Sparassidae we included 11 family members of Dionycha in our analyses, to eventually comment on the validity of this group, as well. Thus representatives of all currently “unplaced families” within Dionycha were included (according to Coddington, 2005; Dunlop and Penney, 2011: Anyphaenidae, Clubionidae, Corinnidae, Gnaphosidae, Liocranidae, Chummidae, Homalonychidae, Salticidae and the ‘Laterigradae’ families) (Fig. 2), with the exception of Zoridae. We did not recover Dionycha as a monophyletic entity. The Dionycha members Thomisidae and Liocranidae were firmly nested within a clade otherwise containing only families of the “Higher Lycosoidea”. Other Dionycha families were distributed throughout the tree, but without sufficient support from all analyses to reject Dionycha altogether.
Two clades exclusively included Dionycha families, one strongly supported (CorinnidaeSalticidae-Philodromidae-Selenopidae), and another one weakly supported (AnyphaenidaeClubionidae-Gnaphosidae).
Results: chapter 3.3: Molecular phylogeny
FIGURE 3. Maximum Likelihood tree with all ingroup taxa, using Thelcticopis as outgroup (Full-ingroup set).
Results of other analyses are mapped upon this tree, summarizing the overall support for each branch (compare legend Fig. 2). Branches separating Thelcticopis and all other Sparassidae were artificially shortened to 1/10 of their actual length to fit the layout. The Full-ingroup-2-Eusparassus set analysis
Results: chapter 3.3: Molecular phylogenyFig. 3.continued: excluded several terminal taxa, especially Eusparassus species that are indicated by a diamond. Associated tables feature only results of other analyses (first four columns on the left). Within Eusparassus clade, groups with intermarginal denticles in the chelicerae indicated by letter “D”.
In contrast the families Chummidae and Homalonychidae were nested within groups containing only non-dionychans (Zodarioidea, Agelenoidea and Amaurobioidea) and Sparassidae is recovered in a basal position without clear affiliation to other RTA-clade families. Other studies treated only a few members of Dionycha (Anyphaenidae,Salticidae, Gnaphosidae; Miller et al., 2010; Agnarsson et al., 2013a), and recovered them as sister to Lycosoidea. Dionycha and Lycosoidea are the two most diverse groups of RTA-clade spiders (Platnick, 2013). Despite having a considerable amount of missing data, Agnarsson et al. (2013b) suggested Dionycha may be a polyphyletic group, a result also indicated by our analyses.
More clearly, the results reject the monophyly of ‘Laterigradae’ in the sense of Latreille (1802), comprising Sparassidae, Thomisidae, Philodromidae and Selenopidae. Coddington and Levi (1991) assumed that these four families may eventually be clustered near each other, while also acknowledging that the characteristic leg posture might be homoplastic. This habitus is for example also present in the haplogyne spider Sicarius Walckenaer, 1847, unrelated to the ‘Laterigradae’. Our analyses rather suggest convergent evolution of this habitus, as the groups with crab-like posture appear randomly distributed within the RTA-clade (Fig. 2). Only Philodromidae and Selenopidae were found more closely related, but are separated and sister to the morphologically contrasting families Salticidae and Corinnidae, respectively, with the sister relationships of Philodromidae and Salticidae well supported in all analyses. These four families were also recovered as close relatives in Agnarsson et al. (2013b) but with a different topological arrangement and weak support. Our analyses also nested the family Thomisidae deep within the Lycosoidea with high support and without any relationship towards other ‘Laterigradae’. The potential phylogenetic relationship of Thomisidae with the Lycosoidea was also proposed by Homann (1975) and Bayer and Schönhofer (2013). Here all our analyses strongly support Thomisidae as the sister group of Liocranidae and both firmly nested within Lycosoidea.
Agnarsson et al. (2013b) gained the same result, yet without recovering Liocranidae as sister to Thomisidae. Coddington and Levi (1991) mentioned Liocranidae to be a polyphyletic family and as we used a taxon (Itatsina praticola) different from Agnarsson et al. (2013b; Agroeca sp.) these
Results: chapter 3.3: Molecular phylogeny
results require further investigation. Finally, the recovered basal placement of the fourth ‘Laterigradae’ family Sparassidae does not entirely rule out the possibility of the crab-like posture being a plesiomorphic trait that was lost in most of the divergent RTA families. We believe this a rather unlikely scenario, as the radical morphological changes in the sister families of Selenopidae (Corinnidae) and Philodromidae (Salticidae), the fact that not all Sparassidae exhibit laterigrade legs (e.g., Micrommata spp.) and known homoplasy in Haplogynae (Sicarius) suggest that leg posture may not necessarily be a good phylogenetic character. The long (laterigrade) legs associated with the presence of scopula and claw tufts may facilitate handling and overpowering dangerous and large preys (Wolff et al., 2013).
Sparassidae was recovered as an independent lineage within the RTA-clade in this study. Its proposed relationship as sister group of Philodromidae (Homann, 1975) can be clearly rejected.
Other phylogenetic proposals for Sparassidae are also controversial, regarding the basal and isolated placement of the family. Simon (1897, 1903) treated Sparassidae as a subfamily of Clubionidae. Lehtinen (1967) tentatively placed Sparassidae and Clubionidae in the superfamily Sparassoidea, considering them closely related. Our analyses revealed Clubionidae to be more close to Anyphaenidae and Gnaphosidae, as partly suggested by Coddington and Levi (1991), and Amaurobiidae not contained in the large clade uniting these families.
3.3. Monophyly and Systematics of Sparassidae
The monophyly of Sparassidae was recovered in all datasets, but with different supports (Fig. 2), affirming the proposed synapomorphic characters of the family. The type species of the type genus for Sparassidae, Micrommata virescens (Clerck, 1757), is also well nested within the inclusive family clade. Micrommata Latreille, 1804 was doubted to belong in the same family as Heteropoda Latreille, 1804 by Lehtinen (in Croeser, 1996), but both are clearly recovered in Sparassidae while being in different subfamilies. Our data set also recovered and thus supported the monophyly of most Sparassidae subfamilies and their diagnostic characters as suggested by many previous authors (Simon, 1897, 1903; Hogg, 1903; Järvi, 1912, 1914; Croeser, 1996; Jäger, 1998). Further results for the Sparassidae subfamilies are discussed below, following an order moving from deeper to shallower nodes of the tree shown in Fig. 3.
The subfamily is represented here by the genus Thelcticopis only, yet a genus featuring all typical synapomorphic characters of Sparianthinae (type genus Sparianthis Simon, 1880). The Sparianthinae currently consists of eight genera from tropical South America, Africa and Australasia (Jäger, 2001; Jäger and Kunz, 2005), and members are clearly separable from the rest of Sparassidae by a number of autapomorphic characters. They have much smaller teeth at the posterior chelicera margin than on the anterior margin (both of equal size in other Sparassidae) and the lateral projections of the trilobate membranes always extend beyond the median hook (Jäger, 1998). Our result thus recovered Sparianthinae as sister to the remaining Sparassidae (here termed Non-Sparianthinae) and suggest a very early divergence during the evolution of the family (Figs 2-3).
Heteropodinae comprises mostly Australasian genera but also one African genus (Jäger, 2001, 2002; Jäger and Kunz, 2005). With six genera included, our sampling is representative and recovered a well supported monophyletic Heteropodinae sensu stricto. All analyses also strongly supported the Heteropodinae in a basal position within the Non-Sparianthinae clade.