General arthropod phylogeny discussion

an artifact that creates bias in several clades in the time-homogeneous tree and has a major impact on the reconstructed phylogeny. In the time-homogenous approach non-monophyletic Hexapoda and Entognatha are recovered. However, comparing the results for Hexapoda and Crustacea it can be shown that most major hexapod clades are robustly revealed in the non-stationary approach contrary to most crustacean clades, despite this time-heterogeneous approach. That eventually indicates the limitation of rRNA data for crustacean phylogeny while hexapods are in comparison reconstructed with relatively robust support values.

In ML and Bayesian approaches of the phylogenomic data (analysis [C]), Entognatha (Protura, Diplura and Collembola) are recovered albeit weakly supported. Since the phylogenomic analysis [C] included all critical members of primarily wingless hexapods and showed that hexapods are monophyletic, the present study supports that hexapods are most likely monophyletic and not paraphyletic in respect to crustaceans as revealed in above cited studies. Relationships among hexapods and pterygote insects are still disputed but not topic of this thesis.

AL. 2004) is generally not supported by morphological data (see e.g. BÄCKER ET AL. 2008).

STOLLEWERK & CHIPMAN (2006) find a support from nervous system pattern but this is conflicting with extensive anatomical data that affirm Mandibulata (HARZSCH ET AL. 2005;

BÄCKER ET AL. 2008).

The results of this thesis support a “modified” clade Mandibulata with a grouping of Pancrustacea + Myriapoda (see figure 1.4, B).

Myriapoda problematics: The present analysis [B] provides no final conclusion with respect to the conflict Mandibulata versus Myriochelata.

The position of Pauropoda and Symphyla causes Myriapoda beeing polyphyletic. To evaluate the impact on the topology of the very likely incorrect positions of Symphyla and Pauropoda, the time-heterogeneous analysis was repeated using a reduced dataset excluding these taxa. The analysis was limited to ten chains with 7,000,000 generations each (2, 000, 000 burn-in). Differences occurring in the inferred consensus topology (not shown) of the final three chains (15, 000, 000 generations) show that some nodes are still sensitive to taxon sampling, since e.g. Pycnogonida cluster with (Chilopoda + Diplopoda) after exclusion of pauropod and symphylan sequences. Also the crustacean topology changes. The remaining long branch taxa Hutchinsoniella and Speleonectes cluster together in the reduced dataset, forming a clade with (Branchiura + Cirripedia).

Despite the endeavor to break down long branches by a dense taxon sampling, some long-branch problems persist. The reason cannot be clearly addressed but, due to the symptoms, it can be assumed that saturation by multiple substitutions caused signal erosion in the rRNA data (class II effect, WÄGELE & MAYER 2007). The exact reconstruction of the position of Myriapoda within the Euarthropoda and a final conclusion supporting Myriochelata or Mandibulata is not possible. But it seems that a clade Myriochelata might be based on a systematic error in molecular analyses. Recent studies demonstrated for example a high sensitivity with respect to gene choice, taxon sampling and out-group choice (BOURLAT ET AL. 2008; PHILIPPE ET AL 2009).

Myriapoda emerged with weak support in the Bayesian trees in the phylogenomic data (analysis [C]) as sister group to chelicerates. In the ML tree reconstructions the relationships of myriapods, sea spiders, euchelicerates and pancrustaceans is not resolved. Applying the matrix reduction heuristics the previous robust support value obtained for this clade in the unreduced dataset vanishes and confirms above cited morphological studies. Thus the application of new markers and suitable phylogenetic strategies like those applied in the phylogenomic analysis (C) have to be applied and developed further.

Chelicerata are in general accepted as a SG clade to Myriapoda + Crustacea + Hexapoda based on morphological, developmental and paleontological data (HARZSCH 2004; HARZSCH ET AL. 2005; RICHTER 2001; SCHOLTZ & EDGECOMBE 2006;

WEYGOLD 1998). The Schizoramia concept (TCC) is to date obsolete and was mainly based on fossil data (CISNE 1974; see e.g. SCHOLTZ & EDGCOMBE 2005). This TCC hypothesis is rejected

in all trees of the present thesis. Pycnogonida are placed by morphological and recent neuroanatomical studies mostly in two competing positions (see DUNLOP 2005; DUNLOP &

ARANGO 2005), either as SG to Euchelicerata (BRENNEIS ET AL. 2008) or as basal euarthropods (MAXMEN ET AL. 2005, but see contrary SCHOLTZ & EDGECOMBE 2006).

In molecular studies the Chelicerata are more ambiguous and mostly revealed in a Myriochelata clade (see previous section). The Pycnogonida (=Pantopoda) are in most studies the SG to Euarthropoda (GIRIBET ET AL. 2001; REGIER & SHULTZ 2001; SHULTZ & REGIER

2001), thus supporting the “Cormogonida” hypothesis (ZRZAV ET AL. 1998A). Yet, a frequent finding is also a SG relation to Euchelicerata (MALLATT ET AL. 2004; DUNN ET AL. 2008;

WHEELER & HAYASHI 1998; REGIER ET AL. 2005) which is supported by developmental data (BRENNEIS ET AL. 2008). Only TELFORD ET AL. (2008) AND BOURLAT ET AL. (2008) published until today a Euarthropoda clade sensu SNODGRASS (1935) with the Pycnogonida as SG to Euchelicerata, relying on molecular data (see figure 1.4, B).

The rRNA data of analysis [B] supports in both approaches a monophyletic Chelicerata with Pycnogonida as SG to Euchelicerata, rejecting the Cormogonida hypothesis. Both approaches reveal a clade “Myriapoda partim”, the unusual position of symphylans and pauropods creates a polyphyletic Myriapoda which is obviously an artifact.

While the position of sea spiders is not resolved in the ML tree (fig. 3.13) with the phylogenomic data (analysis [C]), the Bayesian tree (fig. 3.12) shows monophyletic chelicerates with high support (posterior probability, pP 0.99), including sea spiders. This result corroborates recent molecular analyses (DUNN ET AL. 2008) and neuroanatomical studies, which demonstrate the homology of deuterocerebral appendages of Pycnogonida and Euchelicerata (BRENNEIS ET AL. 2008). It further implies that chelicerae and pedipalpi as head appendages evolved only once and are a diagnostic character of chelicerates. The strongly supported clades Ixodida (ticks) and Astigmata (mites) belonging to Acari, Chelicerata, Decapoda, Copepoda (crustaceans) corroborate results of studies based on single nuclear genes (KJER 2004; LUAN 2005; MISOF ET AL. 2007). This is contrary to studies based on (as previously discussed) problematic mitochondrial protein coding genes (CARAPELLI ET AL. 2007; NARDI ET AL. 2003).

Euarthropoda: Most authors confirm on morphological (BUDD 2001; HUGHES ET AL. 2008;

WALOSZEK ET AL. 2007), molecular (TELFORD 2005; ROEDING ET AL. 2007; BLEIDORN ET AL. 2009) and developmental data (HARZSCH ET AL. 2005; HARZSCH 2006; LOESEL 2005; SCHOLTZ &

EDGECOMBE 2005) the commonly accepted Euarthropoda including Chelicerata, Crustacea, Hexapoda and Myriapoda. For internal relationships see previous sections. The Euarthropoda are a group of the widely accepted Ecdysozoa (TELFORD 2005; TELFORD 2006; TELFORD ET AL.

phylogenetic positions of tardigrades (water bears) and onychophorans to the Euarthropoda can be resolved.

Onychophora strongly resemble arthropod-like animals. They are characterized by lobopods with pads and claws, one set of antennae, reminiscents of primary segmentation and a ladder-like central nervous system. An extensive fossil record exists of Onychophora-like taxa, the Lobopodia. Morphologically the classification of these taxa within the Arthropoda is still discussed, but it seems that evidence for a close relation to other Lobopodia and Euarthropoda outbalances other interpretations (RAMSKÖLD & JUNYUAN 1998). But some authors conclude an ambiguous position of the Onychophora constituting polyphyletic, unresolved relationships of the taxa Onychophora, Lobopodia and Tardigrada as SG to Euarthropods (WALOSZEK ET AL. 2005; BUDD 2009; BUDD

2001).

Molecular analyses place Onychophora either as sister group to Tardigrada + Euarthropoda (REGIER ET AL. 2005; SHULTZ & REGIER 2000) or sistergroup to Euarthropoda (BRUSCA & BRUSCA

2003; DUNN ET AL. 2008; GIRIBET 2008; MALLATT ET AL 2006; TELFORD ET AL. 2008) which is the most common finding.

The rRNA data based analyses ([A] + [B]) cannot enlight the position of Onychophora within arthropods. The Onychophora drop into the euarthropods within these analyses, except the tree based on the hand-optimized dataset (figure 3.5).

Congruent to cited molecular and phylogenomic studies a strong support for a clade Onychophora + Euarthropoda, excluding tardigrades is received with the phylogenomic data (analysis [C]). It appears that onychophorans are the sistergroup of euarthropods. The extensive fossil record of onychophorans (EDGECOMBE 2009) can thus profitably be used to improve our time scale of arthropod evolution.

Tardigrada are tiny animals of which no consensus exists at the moment regarding their phylogenetic position (RAMSKÖLD & JUNYUAN 1998; EDGECOMBE 2009).

BRUSCA & BRUSCA (1990) place Tardigrada in the lineage leading to Euarthropoda above the Onychophora. Anyhow, some morphological characters of Tardigrada are reminiscent of both Arthropoda and Cycloneuralia (BRUSCA & BRUSCA 2003, GIRIBET 2003).

Arthropod-like characters of tardigrades include the segmented body, limbs and ladder-like central nervous system (BRUSCA & BRUSCA 2003, GIRIBET 2003). On the other hand, structures of mouth, pharynx, cuticle and sensory organs of tardigrades resemble those of some Cycloneuralia (GIRIBET 2003). Traditionally, tardigrades are allied with arthropods (BRUSCA &

BRSUCA 1990; BRUSCA & BRUSCA 2003), an arrangement that is recovered also by molecular phylogenetic studies based on ribosomal RNA (MALLATTT ET AL. 2004). Unfortunately some molecular studies are not able to resolve the position of tardigrades revealing a polytomic clade Tardigrada + Onychophora + Euarthropoda (REGIER ET AL. 2005; REGIER ET AL. 2008). A clade of Tardigrada + (Onychophora + Euarthropoda) (DUNN ET AL. 2008; ROEDING ET AL. 2007) would support the evolution of segmentation, segmented appendages and a

ladder-like central nervous system within this group. A sistergroup relationship of tardigrades with Cycloneuralia (LARTILLOT ET AL. 2008, but without Onychophora!; BLEIDORN ET AL. 2009) would imply a loss of these characteristics within Cycloneuralia and a very ancient evolution of a segmented body plan as the most parsimonious explanation.

The not fully understood position of Onychophora and Tardigrada and the possible grouping of Tardigrada as SG to euarthropods or “Tactopoda” (BUDD 2001) makes an out-group choice of Tardigrada at least questionable to infer arthropod phylogeny. However, the rRNA based analyses [A] + [B] are basically focused on the internal relationships of the Euarthropoda. Thus the selection of tardigrades can be justified as chosen out-group. For the phylogenomic analysis the out-group is changed to Mollusca and a massively broadened taxon sampling is applied.

In the phylogenomic analysis tardigrades (Hypsibius and Richtersius) are recovered as sister group of nematodes (BS 100%, pP 1.0), corroborating results of recent phylogenomic studies (ROEDING ET AL. 2007; BLEIDORN ET AL. 2009). In contrast, DUNN ET AL. (2008) found tardigrades as SG of arthropods (including onychophorans) with the CAT model of amino acid evolution (LARTILLOT & PHILIPPE 2004), whereas applying the WAG model the result also suggests an association of tardigrades and nematodes. It might therefore be speculated whether some of the arthropod characteristics are actually plesiomorphic for arthropods and are shared character states of a much more inclusive group. However, this interpretation is still very preliminary since data of several important representatives of ecdysozoan taxa are missing.