Untersuchungen zur Phylogeographie und Systematik von Asselspinnen (Arthropoda: Pycnogonida) des Südpolarmeeres

Studies on the phylogeography and systematics of sea spiders (Arthropoda: Pycnogonida) of the Southern Ocean

Dissertation zur Erlangung des Grades eines Doktors der Naturwissenschaften der Fakultät für Biologie und Biotechnologie an der Internationalen Graduiertenschule Biowissenschaften der Ruhr-Universität Bochum

angefertigt im Lehrstuhl für Evolutionsökologie und Biodiversität der Tiere

vorgelegt von Lars Christian Dietz

aus Duisburg

Referent: Prof. Dr. Ralph Tollrian

Korreferent: Prof. Dr. Dominik Begerow Dissertation Lars Dietz

ERKLÄRUNG

Hiermit erkläre ich, dass ich die Arbeit selbständig verfasst und bei keiner anderen Fakultät eingereicht und dass ich keine anderen als die angegebenen Hilfsmittel verwendet habe. Es handelt sich bei der heute von mir eingereichten Dissertation um sechs in Wort und Bild völlig übereinstimmende Exemplare.

Weiterhin erkläre ich, dass digitale Abbildungen nur die originalen Daten enthalten und in keinem Fall inhaltsverändernde Bildbearbeitung vorgenommen wurde.

Bochum, den

______

1 Dissertation Lars Dietz

Inhaltsverzeichnis

Curriculum Vitae 3

1) General Introduction 4

2) Publikation I: Morphological and genetic data clarify the 21 taxonomic status of Colossendeis robusta and C. glacialis (Pycnogonida) and reveal overlooked diversity

3) Publikation II: Evidence from morphological and genetic data 60 confirms that Colossendeis tenera Hilton, 1943 (Arthropoda: Pycnogonida), does not belong to the Colossendeis megalonyx Hoek, 1881 complex

4) Publikation III: Regional differentiation and extensive 82 hybridisation between mitochondrial clades of the Southern Ocean giant sea spider Colossendeis megalonyx

5) Publikation IV: Assessing demographic responses to past 125 climate change in Southern Ocean sea spiders

6) Publikation V: Pallenopsis patagonica (Hoek, 1881) – a species 151 complex revealed morphology and DNA barcoding, with description of a new species of Pallenopsis by Wilson, 1881

7) Publikation VI: Exploring Pandora’s box: Potential and pitfalls of 177 low coverage genome surveys for evolutionary biology

8) General Discussion 201

9) Summary 217

10) Zusammenfassung 220

11) Acknowledgements 223

2 Dissertation Lars Dietz

Curriculum Vitae von Lars Christian Dietz

Geburt am 22.1.1987 in Duisburg

Schulische Ausbildung: 1993-1996: Grundschule Am Knappert, Duisburg 1996-2005: Reinhard-und-Max-Mannesmann-Gymnasium, Duisburg Abitur am 9.6.2005

Studium: Studium der Biologie und Biotechnologie in Bochum seit Wintersemester 2005 Erwerb des Diploms am 17.11.2010 am Lehrstuhl für Evolutionsökologie und Biodiversität der Tiere (Thema: „Multi-Gen-Analyse zur Phylogenie ausgewählter Pantopoda“) Promotionsstudium seit 18.10.2011 Beschäftigung als wissenschaftlicher Mitarbeiter 1.2.2011 – 31.9.2014 (Ruhr-Universität Bochum) Beschäftigung als wissenschaftliche Hilfskraft 1.10.2014 – 31.1.2015 (Ruhr-Universität Bochum)

3 Dissertation Lars Dietz

1) General Introduction

1.1 The Southern Ocean ecosystem The Southern Ocean is the ocean around the continent of Antarctica (Fig. 1). While there are contrasting definitions of the borders of the Southern Ocean, it is - for biological purposes - most meaningful to define the Antarctic Convergence (also called the Southern Polar Front) as the northern border of the Southern Ocean (e.g. Dell 1972). The region south of the convergence, including land and ocean, is known as the Antarctic and the region to the north of it is called the Subantarctic. The Antarctic convergence, where the warm water masses from the north and the cold ones from the south meet, isolates the Southern Ocean ecosystem from those of other oceans. In the convergence zone there is a strong eastward current, the Antarctic Circumpolar Current (ACC) driven by westerly winds. Most of the Southern Ocean is deep sea, as the Antarctic continental shelf is rather narrow. However, the shelf extends to greater depths than those of other continents (up to 1000 m; Domack 2007). The shelf fauna is considerably better known than that of the deep sea, and most of the samples investigated in this dissertation originated on the shelf. The Antarctic continent, and therefore also its shelf area, is subdivided into the smaller West Antarctica and the larger East Antarctica, which are separated by the Ross and Weddell Seas. The Antarctic Peninsula, with adjacent islands, extends from West Antarctica to the north, and is the best-sampled region of Antarctica (Griffiths et al. 2011). The islands of the Scotia Arc (South Georgia, South Sandwich, South Orkneys, South Shetlands) are situated in the Atlantic sector of the Southern Ocean between South America and the Antarctic Peninsula. Other islands in the Antarctic zone include Bouvet Island in the South Atlantic and the islands of the Kerguelen Plateau in the Indian Ocean. The Antarctic benthic fauna is strongly differentiated from that of other oceans (e.g. Aronson 2007). Particularly notable is the rarity of taxa with planktotrophic larval stages such as decapods, bivalves and teleost fish, while groups without such stages (pycnogonids, isopods, amphipods, some echinoderms) are much more diverse than in other oceans (Clarke & Johnston 2003). Within these groups, a large percentage of species is endemic to the Southern Ocean (35-90% in different taxa according to Arntz et al. 1997, Aronson et al. 2007, see also Fig. 2). There is also strong regional differentiation between various geographical regions of the Southern Ocean, especially between the Antarctic shelf, the surrounding islands, and the deep sea (e.g. Griffiths et al. 2011).

4 Dissertation Lars Dietz

Fig. 1: The Southern Ocean, with the Polar Front and the main island groups marked. From Moles et al. (in press)

Fig. 2. Percentage of species endemic to the Southern Ocean within different benthic invertebrate groups, from Aronson et al. (2007). Bars show Antarctic species number as a percentage of worldwide species number. Numbers are absolute Antarctic species numbers.

5 Dissertation Lars Dietz

In the past, the diversity of Antarctic invertebrate species could only be assessed by the morphological determination of specimens, whose number was often limited due due the difficulties of conducting fieldwork in the Antarctic. Yet, in many groups, the presence of morphologically similar taxa (Lecointre et al. 2013) together with a high within-species variability for some taxa (e.g. the pycnogonid Colossendeis megalonyx; Child 1995 has led to confusion and disagreements concerning the synonymy of species. Molecular data as another line of taxonomic evidence may contribute to address these questions.

1.2 Molecular taxonomy and DNA barcoding In recent years, molecular data has increasingly been used for species-level taxonomy. It has been suggested that a single gene could be used as a “universal barcode” to identify species. In animals, the mitochondrial gene Cytochrome C Oxidase subunit 1 (CO1) has been proposed as the standard barcode fragment (Hebert et al. 2003) due to the fact that in most cases there is a significant gap (“barcoding gap”) between intra- and interspecific genetic distances in CO1. As a mitochondrial DNA (mtDNA) marker which is abundant within cells, relatively simple to amplify and for which several primer pairs have been developed that bind at conserved sites (e.g. Folmer et al. 1994) it is now commonly used throughout the animal kingdom. However, mtDNA is inherited only in the maternal line in almost all animals, meaning that it reflects only a single ancestral lineage. Therefore, mtDNA can be misleading in some cases: First, mtDNA from one species may cross into other related species by hybridization and will be inherited in all descendants of the hybrid in the maternal line (introgressive hybridization, see Toews & Brelsford 2012 for a review). The introgressed mtDNA may also become fixed within the species, especially if it has an adaptive advantage, leading to incongruence between phylogenetic trees based on mitochondrial and nuclear data. Secondly, sex-biased dispersal (i.e. one sex has a greater capability of dispersal than the other) also leads to incongruence between mitochondrial and nuclear data. Thirdly, due to its mode of inheritance and haploidy, mtDNA also generally has a lower effective population size than the nuclear genome (1/4 if the sex ratio is 1:1; Zink & Barrowclough 2008), unless females are much more numerous than males and therefore mutations become fixed more quickly. Because of these problems, mtDNA sometimes gives only an incomplete or misleading picture of population history (Ballard and Whitlock 2004). Therefore, for a detailed investigation of taxonomic questions, an integrative taxonomic approach should be used, incorporating several different datasets such as mtDNA, nuclear DNA and morphological characters. Next-generation sequencing methods such as 454 pyrosequencing (Metzker 2010) increasingly make it possible to generate large amounts of sequence data even for non-

6 Dissertation Lars Dietz model species. These techniques may be useful for phylogenetic and population genetic purposes. As an example, it is possible to generate complete mitochondrial genome sequences from next-generation sequencing data (Dietz et al. 2011, 2015). However, it would be useful to assess how many mitochondrial data, microsatellite markers and other types of data are to be expected in next-generation data from a given taxon. Also, large amount of sequence data can be expected to contain some sequences not belonging to the target organism but represent bacteria or other organisms associated with the target, which can be a problem for bioinformatics analyses. Previous studies have generally not assessed the amount of non-target sequences found in next-generation data. Many studies use fast-evolving genetic markers such as microsatellites, however, for taxa not closely related to any others that have been studied with this method, new suitable markers must be established at first, which can be a tedious task. Alternatively, sequence data from nuclear coding genes can be analyzed. This is often problematic as many protein- coding or rRNA gene regions that are conserved enough for given primers to work in a wide set of taxa are simply not variable enough to be informative for population-level analyses (see Raupach et al. 2010 for an example from Antarctic crustaceans). The Internal Transcribed Spacer (ITS) region of the nuclear ribosomal operon has been shown to be useful for purposes of phylogeography and population genetics. This region consists of the non-coding regions ITS1 and ITS2, between which the 5.8S rRNA gene is situated. As the ITS is flanked by the conserved 18S and 28S rRNA genes, universal PCR primers binding in those regions that successfully amplify ITS in most animal taxa have been developed (Wörheide 1998). The ribosomal operon including those rRNA genes exists in multiple tandemly repeated copies, often on multiple chromosomes. This can lead to intragenomic variability, which has been observed in several taxa (e.g. Harris & Crandall 2000). Despite this problem, ITS has been successfully used for species-level genetic investigations in many different animal taxa (Schlötterer et al. 1994; Odorico & Miller 1997; Chen et al. 2002; Cheng et al. 2006; Arango & Brenneis 2013). Both mitochondrial and nuclear data can be used for species delimitation. Methods of species delimitation are either distance, tree or character-based. Distance-based methods rely on the detection of a barcoding gap, which is taken to be the gap between intra- and interspecific distances. The gap can be detected either manually or with the aid of software such as ABGD (Puillandre et al. 2012). Tree-based species delimitation tries to determine which clades in a phylogenetic tree should be considered as distinct species. A popular model for this is the General Mixed Yule Coalescent (GMYC, e.g. Fujisawa & Barraclough 2013), in which a transition between intraspecific (coalescent) and interspecific (speciation) models of branching on an ultrametric tree is assumed to have occurred at a certain point in time, which is determined using maximum likelihood estimates. Character-based methods try

7 Dissertation Lars Dietz to detect diagnostic molecular characters for species, which is usually practicable only after the species have already been distinguished by other approaches (e.g. Jörger & Schrödl 2014).

1.3 Phylogeography of Southern Ocean taxa While many species of benthic Southern Ocean animals were traditionally thought to demonstrate a circumpolar distribution, molecular studies have revealed the presence of several overlooked species - or cryptic species complexes (e.g. Leese & Held 2008, Brandao et al. 2010, Janosik & Halanych 2010, Wilson et al. 2013, see Thatje 2012 for a review). They are termed “cryptic” as they are thought to be indistinguishable by morphological means. However, in many cases a detailed a posteriori investigation reveals that supposed cryptic species are not morphologically identical (e.g. Brandao 2010). At the same time, those species often have no circumpolar but rather narrow distribution ranges. If the distribution is broad, prominent patterns of genetic differentiation between different regions of the Antarctic shelf (e.g. West vs. East Antarctica) exist at the intraspecific level (e.g. Arango et al. 2011). Differentiation is especially strong between the Antarctic continental shelf and the Subantarctic regions to the north of the Polar Front such as southern South America (Poulin et al. 2014). Differentiation between habitats in different water depths (bathymetric differentiation) has also been detected (Schüller 2011). In some cases, different species within a species complex occur sympatrically, suggesting the presence of mechanisms of reproductive isolation and/or of ecological differences. As an example, in the mobile crinoid Promachocrinus kerguelensis, which has a planktonic larval stage, Hemery et al. (2012) found that the constituent taxa of the species complex are themselves circumpolarly distributed. In most cases, cryptic species complexes have been postulated based on data from a single gene, mostly mitochondrial genes such as the standard barcoding gene CO1. A discordance between the phylogenetic histories of different genes and especially between mitochondrial and nuclear genes has been found in many cases (Toews & Brelsford 2012), indicating that drawing conclusions based on data from a single gene is problematic. However, finding suitable nuclear genes for phylogeographical purposes is difficult, and genes that might be suitable for one taxon may not be of any value for other taxa due to different rates of molecular evolution and other factors. According to Allcock & Strugnell (2012), different genetic structures found in Southern Ocean organisms are caused by differences in life history and by patterns of refugia during the Pleistocene ice ages. Species with a planktonic dispersal stage should show a “star-like” pattern with one common widespread haplotype if they dispersed from a single glacial refugium on the shelf, while a more diffuse network with several frequent and widespread haplotypes is expected if the deep sea acted as a refugium. In species lacking a planktonic

8 Dissertation Lars Dietz dispersal stage, isolation in shelf refugia should lead to strong differentiation and cryptic speciation while dispersal from the deep sea leads to geographically differentiated populations within a single species (Fig. 3). However, it can be remarked that within a cryptic species complex, all of the above mentioned patterns may occur in different species.

Fig. 3: Different patterns of sequence haplotype networks caused by different population histories of Antarctic benthic taxa, from Allcock & Strugnell (2012).

During the Pleistocene glaciations, the Antarctic shelf was almost entirely covered by grounded ice. This implies that the extant benthic shelf fauna must have recolonized the shelf after the Last Glacial Maximum (LGM, about 30,000 years ago) from ice-free refugia. Three possible locations for these refugia have been suggested: 1) Antarctica could have been recolonized from the shelf regions of South America or the Subantarctic islands. In this case, the haplotype diversity found in species inhabiting the Antarctic shelf should show signals of sudden population expansion, i.e. a “star-like” pattern (Fig. 3), and would be a subset of the variation found in the source regions. This has been found by González-Wevar et al. (2013) for the gastropod Nacella concinna, with South Georgia as the potential refugium. 2) The shelf could have been recolonized from the deep sea. In this case, haplotype diversity should show a more diffuse pattern without signals of sudden expansion. This would be expected in eurybathic species that today occur in both shelf and deep

9 Dissertation Lars Dietz

sea regions, and is known for some taxa such as the shrimp Nematocarcinus lanceopes (Raupach et al. 2010, see Fig. 3). 3) The organisms could have survived in temporary ice-free regions on the shelf. This is the most likely explanation for species that are limited to the Antarctic shelf, i.e. not known from either the Subantarctic islands or the deep sea. There would likely be signals of sudden expansions as well as a strong geographic differentiation between the descendants of different refugial populations, at least in species without a planktonic dispersal stage (see Fig. 3).

1.4 Southern Ocean sea spiders (Pycnogonida or Pantopoda) The sea spiders (Pycnogonida, often called Pantopoda in German literature) are a group of arthropods with about 1400 currently accepted species (Bamber & El Nagar 2014). Their position within the Arthropoda has been controversial (see Dunlop & Arango 2004 for a review), but recent studies mostly support their sister-group relationship to all other Chelicerata (Regier et al. 2010). Pycnogonids are known since the Cambrian (Waloszek & Dunlop 2002), although little is understood about their evolutionary history. They are exclusively marine, benthic organisms which occur in all oceans (see King 1973 and Arnaud & Bamber 1987 for detailed reviews of their biology). Most pycnogonids are carnivores or parasites eating sessile organisms such as hydroids, sponges, corals or algae. Characteristically, pycnogonids have seven pairs of appendages: one pair of chelifores homologous to the chelicerae of euchelicerates (Brenneis et al. 2008), one pair of palps, one pair of ovigera, and four pairs of walking legs (see Fig. 4). Each of the three cephalic appendage pairs is missing in certain taxa at least in adults, and the number of walking leg pairs can be increased to five or six. The unsegmented cephalon, which carries the three pairs of cephalic appendages and the first pair of legs, also carries a triradially symmetric proboscis with a terminal mouth. The trunk carries the other pairs of legs and can be segmented or unsegmented. The abdomen has a terminal anus and is highly reduced. The walking legs contain gonads and gut diverticula, and there are no special structures for respiration. In most species, the eggs are carried by the male on the ovigera. Larvae usually hatch as a protonymphon with three pairs of appendages corresponding to the chelifores, palps and ovigera of the adult. In most families, they are ecto- or endoparasitic on hydroids or other benthic organisms, but in the Nymphonidae and Callipallenidae they are – like the eggs – carried by the male. There is no known planktonic larval stage in any pycnogonid. In the Colossendeidae, Austrodecidae and Rhynchothoracidae no ovigerous males have been observed, and their mode of reproduction is unknown.

10 Dissertation Lars Dietz

Fig. 4. Body plan of a typical pycnogonid, Nymphon rubrum, from Brusca & Brusca (2003).

Pycnogonids are particularly species-rich in polar seas, especially the Southern Ocean, where about 20% of the known species occur (Munilla & Soler Membrives 2008). Of all major animal taxa, pycnogonids have the largest percentage of species occurring in the Southern Ocean (Aronson et al. 2007, see Fig. 2). More than half of the known Antarctic species are endemic (Munilla & Soler Membrives 2009) and within the Antarctic region, there are regional differences in the pycnogonid fauna (Griffiths et al. 2011). Species-level diversity appears to be higher south of the Polar Front than north of it, and distinct faunal regions corresponding to the shelf regions of South America, South Africa, the Indian Ocean Subantarctic islands, Australia, and New Zealand can be recognized. Within the Antarctic, there is also clear evidence of depth zonation. Molecular data for most groups of pycnogonids have been generated for phylogenetic studies (Arango 2003, Arango & Wheeler 2007, Nakamura et al. 2007, Arabi et al. 2011, Dietz et al. 2011). However, only few phylogeographic or population genetic studies on pycnogonids have been published. Barreto & Avise (2008, 2010, 2011) used microsatellite markers for paternity analysis in Ammothea hilgendorfi, Pycnogonum stearnsi and

11 Dissertation Lars Dietz

Ammothella biunguiculata. Mahon et al. (2008) and Arango et al. (2011) investigated population structure in the Antarctic species Nymphon australe based on the mitochondrial markers CO1 and 16S. They concluded that there is no evidence of cryptic speciation and little differentiation within the Antarctic Peninsula region. While there is clear genetic differentiation between Antarctic regions, the evidence suggests that N. australe is a single species with circumpolar distribution. Weis & Melzer (2012) showed evidence for strong regional differentiation in Achelia assimilis with CO1 data. Arango & Brenneis (2013) used CO1 and the nuclear marker ITS to differentiate between Australian species of Pseudopallene. The results agree well with morphological data such as live coloration. Within the Pycnogonida, the family Colossendeidae is particularly well represented in the Antarctic (38 species according to Munilla & Soler Membrives 2009). The Colossendeidae are characterized by a relatively large body size and an extremely long proboscis, often longer than the trunk. The adults usually lack chelifores, but palpi and ovigera are well developed. As eggs or larvae are not carried by the male in the Colossendeidae, the mode of larval development is unknown. Southern Ocean colossendeids have been known since the 19th century (Eights 1834, Hoek 1881), and the taxonomy of some species has been controversial for a long time. One of the most common Southern Ocean pycnogonid species is Colossendeis megalonyx Hoek 1881. This species is reported from depths of 7-4900 m and from the whole Southern Ocean, and supposedly has a circumpolar distribution in both the Antarctic and Subantarctic. Additionally it has been reported from South America, South Africa and Madagascar (Munilla & Soler Membrives 2009). The type locality is off the Patagonian east coast. The species is highly variable, and several synonyms have been named, including C. rugosa Hodgson 1907, C. frigida Hodgson 1907, C. orcadense Hodgson 1909, and C. megalonyx arundorostris Fry & Hedgpeth 1969. Several authors (e.g. Child 1995) have suggested that the “C. megalonyx complex” may contain more than one species, but this could be tested only in a detailed morphological study. However, no such study has yet been published. Turpaeva (1973) suggested that the East Pacific species C. tenera Hilton 1943 is synonymous with C. megalonyx, which would be a possible example of dispersal from the Antarctic to other oceans. This proposal has not been generally accepted, but it has not been tested with detailed morphological or molecular data. Krabbe et al. (2010) analyzed CO1 sequences of 96 C. megalonyx specimens and found them to group into six clearly distinct clades with interclade genetic distances of more than 5%, suggesting the presence of cryptic species (Fig. 5). Specimens of one of those clades (Clade C) could be morphologically distinguished by the absence of pigmented eyes. These clades have a geographically restricted distribution, although some of them occur sympatrically. However, their sampling was limited to only a few regions (Falklands, Bouvet Island, South Sandwich Is., South

12 Dissertation Lars Dietz

Shetlands), not including any part of the East Antarctic shelf. Also, the results have not yet been independently tested with nuclear DNA sequence data.

Fig. 5: Haplotype network of CO1 sequences of Colossendeis megalonyx from Krabbe et al. (2010), showing differentiation into six major clades.

Another species generally thought to be very common in the Southern Ocean is Colossendeis robusta Hoek 1881. This species has been recorded from depths of 0-3610 m, with a circumpolar distribution in the Antarctic as well as records from the Subantarctic Bouvet and Kerguelen islands. However, there have been some disagreements about its synonymy with other species. The species Colossendeis glacialis Hodgson 1907 has been synonymized with C. robusta Hoek 1881 (Fry & Hedgpeth 1969). The differences between

13 Dissertation Lars Dietz these species were claimed to be ontogenetic, with C. glacialis representing younger specimens. However, this synonymy has been questioned (Pushkin 1988, Cano & Lopez Gonzalez 2007). Also, the species C. lilliei Calman 1915 has been claimed by Child (1995) to be synonymous with C. robusta, which has been disputed by Cano & Lopez Gonzalez (2007). The species Pallenopsis patagonica (Hoek 1881) is reported from depths of 3-4540 m and occurs circumpolarly on the Antarctic shelf as well as in southern South America. Several other putative species from the Southern Ocean such as Pallenopsis glabra Möbius 1902 and P. hiemalis Hodgson 1907 are currently considered synonyms. So far, the taxonomy of P. patagonica and related species has not been investigated with genetic methods.

1.4 Aims of this thesis The central aim of my thesis is to clarify the taxonomy and phylogeography of Southern Ocean pycnogonids using evidence from mitochondrial data coupled with nuclear gene markers and morphological characters.

In Paper I, the taxonomy of Colossendeis robusta and similar or possibly synonymous species is investigated using molecular (mitochondrial and nuclear) data as well as statistical analysis of morphometric measurements and investigation of the morphology of the ovigeral strigilis using SEM photomicrographs. I address the question whether C. robusta and C. glacialis are synonymous and whether other previously unrecognized species may exist within that group. Comparisons of historical taxonomic descriptions out of the literature with our data are used to test the validity of characters that were proposed to segregate species in the literature. Furthermore, using the molecular data the genetic structure within species is investigated, especially with regard to past population history.

In Paper II, morphological and molecular data as well as morphometric measurements and ovigeral SEM data are used to investigate the position of Colossendeis tenera within the Colossendeidae, especially to test whether its suggested position as part of the C. megalonyx species complex can be supported by the evidence. A possible relationship of C. tenera to the widespread deep-sea species C. angusta Sars 1877 and its possible synonym C. gracilis Hoek 1881 is also investigated, as these species seem to have some morphological similarities. Literature descriptions are compared with our data to evaluate the characters used to distinguish C. angusta and C. gracilis, and to examine how those characters apply to C. tenera.

14 Dissertation Lars Dietz

In Paper III, the CO1 dataset for C. megalonyx is expanded to include a large number of samples from both West and East Antarctica as well as the Scotia Arc islands and other Subantarctic regions. One aim is to test how many species can be recognized in that complex using different species delimitation methods such as GMYC and ABGD. Also, the genetic structure within clades is investigated, particularly regarding possible geographic differentiation and inferences on past population history such as the presence of possible refugia. Additionally, the CO1 data are independently tested by nuclear sequence data (ITS). It is tested whether ITS provides enough resolution to be useful for phylogeography within the C. megalonyx complex and whether the results agree with those of the CO1 analysis, supporting species status for the clades, or whether there are discordances suggesting hybridization between CO1 lineages.

In Paper IV, the large CO1 dataset from Paper III is used to test more detailed hypotheses about the geographic structuring and population history of the four largest clades within the C. megalonyx complex. Standard population genetic metrics are used to test for differentiation between geographically separated populations and for signals of past population expansion. Bayesian analysis is further used to test the presence of any significant geographic differentiation, and Bayesian skyline plots are used to infer the approximate change in population size. Approximate Bayesian Computations (ABCs) are used to test between various models of population history differing in the order of population divergence and the presence and timing of population size change and to estimate current and past population sizes.

In Paper V, mitochondrial data and detailed morphological investigations are used to test whether Pallenopsis patagonica should be regarded as one species or as a complex including several distinct species. This includes investigations of the holotype of P. patagonica as well as material from other closely related species such as Pallenopsis macneilli Clark 1963. The study focuses especially on newly collected material from the Chilean coast.

In Paper VI, next-generation sequencing libraries of different animal taxa including the pycnogonids C. megalonyx, P. patagonica and Austropallene cornigera are systematically searched for several types of data, including mitochondrial sequences, certain nuclear markers, repetitive DNA (micro- and minisatellites) and DNA not belonging to the target organism. The aim was to test the usefulness of next-generation sequencing data for various population genetic and phylogenetic methods but also the risk of contamination.

15 Dissertation Lars Dietz

1.5 References Allcock AL, Strugnell JM (2012): Southern Ocean diversity: new paradigms from molecular ecology. Trends in Ecology and Evolution 27 (9): 520-528 Arabi J, Cruaud C, Couloux A, Hassanin A (2011): Studying sources of incongruence in arthropod molecular phylogenies: sea spiders (Pycnogonida) as a case study. Comptes Rendus Biologies 333: 438-453 Arango CP (2003): Molecular approach to the phylogenetics of sea spiders (Arthropoda: Pycnogonida) using partial sequences of nuclear ribosomal DNA. Molecular Phylogenetics and Evolution 28: 588-600 Arango CP, Brenneis G (2013): New species of Australian Pseudopallene (Pycnogonida: Callipallenidae) based on live colouration, morphology and DNA. Zootaxa 3616: 401-436 Arango CP, Soler-Membrives A, Miller KJ (2011): Genetic differentiation in the circum— Antarctic sea spider Nymphon australe (Pycnogonida; Nymphonidae). Deep Sea Research Part II: Topical Studies in Oceanography (2011 Mar 1) 58: 212-219 Arango CP, Wheeler WC (2007): Phylogeny of the sea spiders (Arthropoda, Pycnogonida) based on direct optimization of six loci and morphology. Cladistics 23: 255–293 Arnaud F, Bamber R (1987): The biology of Pycnogonida. Advances in Marine Biology 24: 1- 96 Arntz WE, Gutt J, Klages M (1997): Antarctic marine biodiversity: an overview. In: Battaglia B, Valencia J, Walton DWH (eds.): Antarctic Communities: Species, Structure, and Survival. Cambridge University Press, Cambridge. Aronson RB, Thatje S, Clarke A, Peck LS, Blake DB, Wilga CD, Seibel BA (2007): Climate change and invasibility of the Antarctic benthos. Annual Review of Ecology Evolution and Systematics 38: 129-154 Ballard JW, Whitlock MC (2004): The incomplete natural history of mitochondria. Molecular Ecology 13: 729-744 Bamber RN, El Nagar A (2014): Pycnobase: World Pycnogonida Database. http://www.marinespecies.org/pycnobase accessed on March 11th, 2015 Barreto FS, Avise JC (2008): Polygynandry and sexual size dimorphism in the sea spider Ammothea hilgendorfi (Pycnogonida: Ammotheidae), a marine arthropod with brood- carrying males. Molecular Ecology 17: 4164-75 Barreto FS, Avise JC (2010): Quantitative measures of sexual selection reveal no evidence for sex-role reversal in a sea spider with prolonged paternal care. Proceedings of the Royal Society B 277: 2951-6 Barreto FS, Avise JC (2011): The genetic mating system of a sea spider with male-biased sexual size dimorphism: evidence for paternity skew despite random mating success. Behavioral Ecology and Sociobiology 65: 1595-1604

16 Dissertation Lars Dietz

Brandao SN (2010): Macrocyprididae (Ostracoda) from the Southern Ocean: taxonomic revision, macroecological patterns, and biogeographical implications. Zoological Journal of the Linnean Society 159: 567-672 Brandao SN, Sauer J, Schön I (2010): Circumantarctic distribution in Southern Ocean benthos? A genetic test using the genus Macroscapha (Crustacea, Ostracoda) as a model. Molecular Phylogenetics and Evolution 55: 1055–1069 Brenneis G, Ungerer P, Scholtz G (2008): The chelifores of sea spiders (Arthropoda, Pycnogonida) are the appendages of the deutocerebral segment. Evolution & Development 10: 717-724 Brusca RC, Brusca GJ (2003): Invertebrates. Sinauer Associates, Sunderland. Calman WT (1915): Pycnogonida. British Antarctic ("Terra Nova") Expedition, 1910 3: 1-74 Cano E, Lopez-Gonzalez PJ (2007): Colossendeis species (Pycnogonida: Colossendeidae) collected during the Italica XIX cruise to Victoria Land (Antarctica), with remarks on some taxonomic characters of the ovigers. Scientia Marina 71: 661-681 Chen CH, Chen CP, Fan TY, Yu JK, Hsieh HL (2002): Nucleotide sequences of ribosomal internal transcribed spacers and their utility in distinguishing closely related Perinereis polychaets (Annelida; Polychaeta; Nereididae). Marine Biotechnology 4: 17–29 Cheng HL, Xia DQ, Wu TT, Meng XP, Ji HJ, Dong ZG (2006): Study on sequences of ribosomal DNA internal transcribed spacers of clams belonging to the Veneridae family (Mollusca: Bivalvia). Acta Genetica Sinica 33: 702–710 Child CA (1995): Antarctic and Subantartic Pycnogonida: The Families Nymphonidae, Colossendeidae, Rhynchothoraxidae, Pycnogonidae, Endeididae, and Callipallenidae. Antarctic Research Series 69: 1-165 Clark WC (1963): Australian Pycnogonida. Records of the Australian Museum 26: 1-81 Clarke A, Johnston NM (2003): Antarctic marine benthic diversity. Oceanography and Marine Biology: an Annual Review 41: 47-114 Dell RK (1972): Antarctic benthos. Advances in Marine Biology 10: 1-216 Dietz L, Brand P, Eschner LM, Leese F (2015): The mitochondrial genomes of the caddisflies Sericostoma personatum and Thremma gallicum (Insecta: Trichoptera). Mitochondrial DNA, advance online publication. Dietz L, Mayer C, Arango CP, Leese F (2011): The mitochondrial genome of Colossendeis megalonyx supports a basal position of Colossendeidae within the Pycnogonida. Molecular Phylogenetics and Evolution 58: 553-558 Domack EW (2007): Continental shelves and slopes. In: Riffenburgh B (ed.): Encyclopedia of the Antarctic. Routledge, New York. Dunlop JA, Arango CP (2005): Pycnogonid affinities: a review. Journal of Zoological Systematics and Evolutionary Research 43: 8-21

17 Dissertation Lars Dietz

Eights J (1834): Description of a new animal belonging to the Arachnides of Latreille; discovered in the sea along the shores of the New South Shetland Islands. Boston Journal of Natural History 1: 203-206 Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R (1994): DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Molecular Marine Biology and Biotechnology 3: 294-299 Fry WG, Hedgpeth JW (1969): The fauna of the Ross Sea. Part 7. Pycnogonida, 1: Colossendeidae, Pycnogonidae, Endeidae, Ammotheidae. New Zealand Department of Science and Industrial Research Bulletin (1969) 198: 1-139 Fujisawa T, Barraclough TG (2013): Delimiting species using single-locus data and the generalized mixed Yule coalescent (GMYC) approach: a revised method and evaluation on simulated datasets. Systematic Biology 62: 707-724 González-Wevar CA, Saucède T, Morley SA, Chown SL, Poulin E (2013): Extinction and recolonization of maritime Antarctica in the limpet Nacella concinna (Strebel, 1908) during the last glacial cycle: toward a model of Quaternary biogeography in shallow Antarctic invertebrates. Molecular Ecology 22: 5221-5236 Griffiths HJ, Arango CJ, Munilla T, McInnes SJ (2011): Biodiversity and biogeography of Southern Ocean pycnogonids. Ecography 34: 616-627 Harris DJ, Crandall KA (2000): Intragenomic variation within ITS1 and ITS2 of crayfish (Decapoda: Cambaridae): implications for phylogenetic and microsatellite studies. Molecular Biology and Evolution 17: 284-291 Hebert PD, Cywinska A, Ball SL, deWaard JR (2003): Biological identifications through DNA barcodes. Proceedings of the Royal Society B 270: 313-21 Hemery LG, Eléaume M, Roussel V, Améziane N, Gallut C, Steinke D, Cruaud C, Couloux A, Wilson NG (2012): Comprehensive sampling reveals circumpolarity and sympatry in seven mitochondrial lineages of the Southern Ocean crinoid species Promachocrinus kerguelensis (Echinodermata). Molecular Ecology 21: 2502-2518 Hilton WA (1943): Pycnogonida from the Pacific. Family Colossendeidae. Journal of Entomology and Zoology, Pomona College, Claremont 35: 2-4 Hodgson TV (1907): Pycnogonida. Natural History Collections of the "Discovery" National Antarctic Expedition, Zoology 3: 1-72 Hodgson TV (1909): The Pycnogonida of the Scottish National Antarctic Expedition. Transactions of the Royal Society Edinburgh 46: 159-188 Hoek PPC (1881): Report on the Pycnogonida, dredged by HMS Challenger during the Years 1873-76. Report on the Scientific Results of the Voyage of the HMS Challenger, Zoology 3: 1-167

18 Dissertation Lars Dietz

Janosik AM, Halanych KM (2010): Unrecognized Antarctic biodiversity: A case study of the genus Odontaster (Odontasteridae; Asteroidea). Integrative and Comparative Biology (2010) 50: 981-992 Jörger KM, Schrödl M (2013): How to describe a cryptic species? Practical challenges of molecular taxonomy. Frontiers in Zoology (2013) 10: 59 King PE (1973): Pycnogonids. Hutchinson, London. Krabbe K, Leese F, Mayer C, Tollrian R, Held C (2010): Cryptic mitochondrial lineages in the widespread pycnogonid Colossendeis megalonyx Hoek, 1881 from Antarctic and Subantarctic waters. Polar Biology 33: 281-292 Leese F, Held C (2008): Identification and characterization of microsatellites from the Antarctic isopod Ceratoserolis trilobitoides: nuclear evidence for cryptic species. Conservation Genetics (2008) 9 ): 1369–1372 Mahon AR, Arango CP, Halanych KM (2008): Genetic diversity of Nymphon (Arthropoda: Pycnogonida: Nymphonidae) along the Antarctic Peninsula with a focus on Nymphon australe Hodgson 1902. Marine Biology 55: 315–323 Metzker ML (2010): Sequencing technologies: the next generation. Nature Reviews Genetics 11: 31-46 Möbius K (1902): Die Pantopoden der deutschen Tiefsee-Expedition 1898-1899. Wissenschaftliche Ergebnisse der Deutschen Tiefsee-Expedition auf "Valdivia" 1898- 1899 3: 175-196 Moles J, Figuerola B, Campanyà-Llovet N, Monleón-Getino T, Taboada S, Avila C (in press): Distribution patterns in Antarctic and Subantarctic echinoderms. Polar Biology (published online 2015 Jan 23) Munilla T, Soler-Membrives A (2009): Check-list of the pycnogonids from Antarctic and sub- Antarctic waters: zoogeographic implications. Antarctic Science 21: 99-111 Nakamura K, Kano Y, Suzuki N, Namatame T, Kosaku A (2007): 18S rRNA phylogeny of sea spiders with emphasis on the position of Rhynchothoracidae. Marine Biology 153: 213- 223 Odorico DM, Miller DJ (1997): Variation in the ribosomal internal transcribed spacers and 5.8S rDNA among five species of Acropora (Cnidaria; Scleractinia): patterns of variation consistent with reticulate evolution. Molecular Biology and Evolution 14: 465-473 Poulin E, González-Wevar C, Díaz A, Gérard K, Hüne M (2014): Divergence between Antarctic and South American marine invertebrates: What molecular biology tells us about Scotia Arc geodynamics and the intensification of the Antarctic Circumpolar Current. Global and Planetary Change 123: 392-399. Puillandre N, Lambert A, Brouillet S, Achaz G (2012): ABGD, Automatic Barcode Gap Discovery for primary species delimitation. Molecular Ecology 21: 1864-77

19 Dissertation Lars Dietz

Pushkin AF (1988): K revisii vida Colossendeis robusta (Pantopoda) iz yuzhnogo okeana. Zoologicesky Zhurnal (Moscow) 67: 953-956 Raupach MJ, Thatje S, Dambach J, Rehm P, Misof B, Leese F (2010): Genetic homogeneity and circum-Antarctic distribution of two benthic shrimp species of the Southern Ocean, Chorismus antarcticus and Nematocarcinus lanceopes. Marine Biology 157: 1783-1797. Regier JC, Shultz JF, Zwick A, Hussey A, Ball B, Wetzer R, Martin JW, Cunningham WC (2010): Arthropod relationships revealed by phylogenomic analysis of nuclear protein- coding sequences. Nature 463: 1079–1084 Sars GO (1877): Prodromus descriptionis crustaceorum et pycnogonidarum, quae in expeditione norvegica anno 1876, observavit. Archiv for Mathematik og Naturvidenskab 2: 237-271 Schlötterer C, Hauser MT, von Haeseler A, Tautz D (1994): Comparative evolutionary analysis of rDNA ITS regions in Drosophila. Molecular Biology and Evolution 11: 513-522 Schüller M (2011): Evidence for a role of bathymetry and emergence in speciation in the genus Glycera (Glyceridae, Polychaeta) from the deep Eastern Weddell Sea. Polar Biology 34: 549-564 Thatje S (2012): Effects of capability for dispersal on the evolution of diversity in Antarctic benthos. Integrative and Comparative Biology 52: 470-482 Toews DP, Brelsford A (2012): The biogeography of mitochondrial and nuclear discordance in animals. Molecular Ecology 21: 3907-30 Turpaeva EP (1973): Mnogokolencatye (Pantopoda) iz severo-zapadnoy casti Tikhogo Okeana. Trudy Instituta Okeanologii "P. P. Shirshova" Akademy Nauk SSSR 91: 178- 191 Waloszek D, Dunlop JA (2002): A larval sea spider (Arthropoda: Pycnogonida) from the Upper Cambrian ’Orsten’ of Sweden, and the phylogenetic position of pycnogonids. Palaeontology 45: 421-446 Weis A, Melzer RR (2012): How did sea spiders recolonize the Chilean fjords after glaciation? DNA barcoding of Pycnogonida, with remarks on phylogeography of Achelia assimilis (Haswell, 1885). Systematics and Biodiversity 10: 361-374 Wilson NG, Maschek JA, Baker BJ (2013): A species flock driven by predation? Secondary metabolites support diversification of slugs in Antarctica. PLoS One 8: e80277 Wörheide G (1998): The reef cave dwelling ultraconservative coralline demosponge Astrosclera willeyana Lister 1900 from the Indo-Pacific. Facies 38: 1-88 Zink RM, Barrowclough GF (2008): Mitochondrial DNA under siege in avian phylogeography. Molecular Ecology 17: 2107-2121

20 Dissertation Lars Dietz

2) Publikation I

Titel: Morphological and genetic data clarify the taxonomic status of Colossendeis robusta and C. glacialis (Pycnogonida) and reveal overlooked diversity

Arthropod Systematics and Phylogeny, akzeptiert am 5. Februar 2015

Hinweise zu Publikation I

• Anteil Planung: 80%

• Anteil experimentelle Durchführung: 75%

• Verfassen des Manuskripts: 85%

Abbildungen, die nicht ausschließlich von mir erstellt wurden

• Fig. 5: Photographien von Meike Seefeldt

• Fig. 6: Photographien von Meike Seefeldt

• Fig. 7: Photographien von Meike Seefeldt

21 Proof _ 01 Proof _ 01

739 (1): 8 – 110 xx.yy.2015

© Senckenberg Gesellschaft für Naturforschung, 2014.

Morphological and genetic data clarify the taxonomic status of Colossendeis robusta and C. glacialis (Pycno- gonida) and reveal overlooked diversity

Lars Dietz, Sven Pieper, Meike A. Seefeldt & Florian Leese *

Ruhr-Universität Bochum, Lehrstuhl für Evolutionsökologie und Biodiversität der Tiere, Universitätsstraße 150, 44801 Bochum, Germany; Lars Dietz [[email protected]]; Sven Pieper [[email protected]]; Meike A. Seefeldt [[email protected]]; Florian Leese [florian.leese@ rub.de] — * Corresponding author

Accepted 05.ii.2015. Published online at www.senckenberg.de/arthropod-systematics on xx.yy.2015.

Abstract Colossendeis robusta Hoek, 1881, originally described from the Kerguelen shelf, is considered as one of the most widespread Antarctic pycnogonids. However, the taxonomic status of this and similar species has long been unclear, as synonymy of C. glacialis Hodgson, 1907 and several other species with C. robusta has been proposed. Here we test the synonymy of C. robusta and C. glacialis with two independent molecular markers as well as comprehensive morphometric measurements and SEM data of the ovigeral spine configuration. We show that C. robusta and C. glacialis are clearly distinct species, and our results also indicate the existence of another previously unrecognized Antarctic species, C. bouvetensis sp.n., as well as an Antarctic lineage closely related to the endemic Kerguelen group C. robusta s.str. We find evidence for strong regional differentiation within each species. Our results suggest that diversity of Antarctic pycnogonids is still underestimated.

Key words Colossendeis robusta, Colossendeis glacialis, Colossendeis bouvetensis, Southern Ocean, Pycnogonida, phylogeography, integrative taxo­ nomy.

1. Introduction

The Southern Ocean hosts a remarkably high diversity of nogonida or Pantopoda; Munilla & Soler Membrives benthic invertebrate species (Clarke & Johnston 2003; 2009; Griffiths et al. 2011). Pycnogonids are exclusively Gutt et al. 2004; Griffiths 2010), of which a large pro­ marine arthropods, which occur in all oceans (Arnaud & portion has radiated in situ after the thermal isolation by Bamber 1987). The systematic position of the Pycnogo­ the Antarctic Circumpolar Current (ACC) (Held 2000; nida in the arthropod tree of life is still unresolved (Dun- Convey et al. 2009; Lecointre et al. 2013). Many of these lop & Arango 2005; Regier et al. 2010; Giribet & Edge- show a remarkably high degree of endemism (Gutt et al. combe 2012). However, their internal phylogeny has been 2004; Peat et al. 2006; Brandt et al. 2007). A great num­ studied, but not completely resolved, using genes and ber of species has only been recognized in the context of morphology (Munilla León 1999; Arango & Wheeler recent molecular studies (see Janosik & Halanych 2010; 2007; Bamber 2007; Nakamura et al. 2007; Arabi et al. Allcock & Strugnell 2012 for reviews). 2010; Dietz et al. 2011). Within the Pycnogonida, there One taxon that shows a uniquely high degree of en­ are over 1,300 species in 10 families (Arnaud & Bam- demism in the Southern Ocean are the sea spiders (Pyc­ ber 1987; Bamber & El Nagar 2014). Almost 20% of

ISSN 1863-7221 (print) | eISSN 1864-8312 (online) 89 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

these species (264 species) occur in the Southern Ocean but they are probably present in the larval development (Arango & Wheeler 2007; Munilla & Soler Mem- and sometimes still seen in subadults (e.g. Hoek 1881). brives 2009), of which 50% are endemic to this region. Within Colossendeis there are several taxonomic con­ The family Colossendeidae Jarzynsky, 1870, in particu­ troversies resulting from different authors’ views on the lar of the genus Colossendeis Jarzynsky, 1870, is espe­ synonymy of certain species. One of these debates con­ cially abundant in the Southern Ocean. 36 of the known cerns Colossendeis robusta Hoek, 1881, and Colossende­ 75 species of Colossendeis occur in the Southern Ocean, is glacialis Hodgson, 1907, two morphologically similar 26 of which are endemic, rendering the region a hotspot species with reported circumpolar distribution. C. robus­ of endemism (Munilla & Soler Membrives 2009). For ta was originally described from Kerguelen but later also 55 species of Antarctic and Subantarctic pycnogonids a reported to be geographically widespread in Antarctic circumpolar distribution is currently assumed (Munilla waters (Fry & Hedgpeth 1969; Munilla & Soler Mem- León 2001; Munilla & Soler Membrives 2009; Grif- brives 2009), whereas C. glacialis was first documented fiths et al. 2011). in the Ross Sea but subsequently also in other regions of However, in view of the occurrence of several cryp­ the Southern Ocean (e.g. Pushkin 1993). Claimed differ­ tic or overlooked species reported for this group (Mahon ences between the two species are summarized in Table et al. 2008; Krabbe et al. 2010; Weis & Melzer 2012), 1. However, some authors regard C. glacialis as a junior consistent with reports from other groups (e.g. Held & synonym of C. robusta (Hodgson 1927; Fry & Hedgpeth Wägele 2005; Wilson et al. 2007; Leese & Held 2008; 1969; Child 1995) and explain the observed morphologi­ Thornhill et al. 2008; Brandao et al. 2010), both the cal differences by intraspecific, mostly ontogenetic vari­ actual number of species as well as their distribution ation. The (mostly) larger species C. robusta is argued ranges have to be regarded with caution. Reports of much to be described based on more fully grown individuals. smaller horizontal distribution ranges for some species Other authors disagree with the synonymization of C. (e.g. Held & Wägele 2005; Krabbe et al. 2010) as well glacialis and C. robusta and continue to treat both spe­ as much narrower vertical distribution ranges (Schüller cies as being distinct (Pushkin 1988; Stiboy-Risch 1993; 2011) challenge the traditional biogeographic concepts Cano & Lopez-Gonzalez 2007). Claimed differences of circumpolarity and eurybathy (but see Hemery et al. between the two species are summarized in Table 1. In 2012). An important point when testing biodiversity es­ both species, according to the literature (Hodgson 1907a; timates as well as biogeographic concepts is that species Calman 1915; Stiboy-Risch 1993; Cano & Lopez-Gon- are correctly delimited. A combination of mitochondrial zalez 2007), the lateral processes of the trunk are sepa­ (DNA barcoding) and nuclear DNA markers coupled rated from each other by about half their diameter. The with morphological approaches has shown strong po­ ocular tubercle is a broad and large cone with four eyes, tential when studying pycnogonids (Arango & Brenneis one pair anterior and posterior eyes, which are placed 2013; Dietz et al. 2013). very closely together. The palps consist of the usual 10 The morphology of Colossendeis species mostly re­ articles, of which the distalmost five are of similar length, sembles that of other Pycnogonida (King 1973; Arnaud but the seventh article is slightly longer than the more & Bamber 1987; Child 1995; Arango 2002; Arango & distal ones. The oviger is also divided into 10 articles, the Wheeler 2007; Cano & Lopez-Gonzalez 2007) except four distal articles are approximately of the same size and that adults lack chelifores. The trunk, comprising most of combine to form the strigilis. The femur and the tibia 1 the body, consists of cephalic and leg-bearing segments. are approximately of the same length (Hoek 1881; Hodg- Four pairs of lateral processes carry the walking legs son 1907a; Calman 1915; Pushkin 1993; Stiboy-Risch (each consisting of eight articles and a terminal claw), 1993; Child 1995; Cano & Lopez-Gonzalez 2007). and at the posterior end of the trunk there is a short abdo­ Although various differences between C. glacialis men. The anterior part of the trunk, the cephalosoma, car­ and C. robusta have been claimed (Table 1), no system­ ries a pair of palpi, a pair of ovigera, the first pair of walk­ atic measurements of numerous specimens testing for ing legs, the proboscis and the ocular tubercle carrying significance of the differences have yet been performed. the eyes. In the genus Colossendeis there are both eye- Also, no molecular study has yet been performed that bearing and eyeless species. In other pycnogonids, the would enable to test the possible distinctiveness of the ovigera are used by the male to carry the eggs, however, morphological characters with an independent source of in Colossendeis no egg-carrying males have been found, data. Besides, it has never been tested whether the Ant­ and they probably serve as cleaning appendages (King arctic specimens assigned to C. robusta really represent 1973; Bain 2003). The last four ovigeral articles bear a the same species as the holotype from Kerguelen. In a complex arrangement of spines and together form the recent work, Dietz et al. (2013) have shown the poten­ strigilis, the morphology of which can be diagnostic for tial of combining molecular and morphological analyses particular species (Cano & Lopez-Gonzalez 2007). The to resolve controversial taxonomic and phylogeographic three post-cephalic leg-bearing segments are fused. The questions and identify unrecognized species inside the individuals of Colossendeis feed on bryozoans, cnidar­ genus Colossendeis. In particular the mitochondrial gene ians, sponges, small mollusks and small polychaetes cytochrome c oxidase subunit I (COI), which is often re­ (Arnaud & Bamber 1987; Braby et al. 2009). Chelifores ferred to as the “standard barcoding gene”, evolves faster are completely absent in adult specimens of Colossendeis than most nuclear genes in many animal phyla and is of­

90 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

Table 1. Claimed differences in morphological characters between C. robusta and C. glacialis.

Body part Colossendeis robusta Colossendeis glacialis Reference Legs Femur, tibia 1 < tibia 2 Femur, tibia 1 > tibia 2 Calman (1915), Stiboy-Risch (1993) Palps Last 5 articles elongate Last 5 articles rounded Hodgson (1927) Proboscis, palps, legs Not spinous Spinous Fry & Hedgpeth (1969), Stiboy-Risch (1993) Palps Article 6 longest of the last distal 5 Article 7 longest of the last distal 5 Pushkin (1988) Legs Claw > ½ propodus Claw < ½ propodus Pushkin (1993) Body Larger, robust Smaller, gracile Stiboy-Risch (1993) Proboscis Shape D’’’:2 (Fry & Hedgpeth 1969) Shape B’’’:1 (Fry & Hedgpeth 1969) Stiboy-Risch (1993) Palps Article 9 shorter than 8 and 10 Last 3 articles equally long Stiboy-Risch (1993)

ten easy to amplify, which makes it an ideal candidate on the research vessel Nathaniel B. Palmer (http://www. for distinguishing species and disentangling their phylo­ icefish.neu.edu) come from South Georgia and the South geographic history (Hebert et al. 2003, 2004; Frézal & Sandwich Islands (n = 12) and Bouvet Island (n = 14). Leblois 2008). However, as data based on a single gene Samples from the Antarctic Peninsula (n = 8) and the may be misleading, independent data, i.e. nuclear genes, Eastern Weddell Sea (n = 1) are from RV Polarstern expe­ are required. One fragment of the nuclear genome widely dition ANT XIV/2 (Kattner 1998). Samples from Terre used for phylogenetic analyses in animals is the Inter­ Adélie (n = 3) are from the REVOLTA expeditions and nal Transcribed Spacer (ITS) region (Fritz et al. 1994; those from Kerguelen (n = 13) are from the POKER II Schlötterer et al. 1994; Odorico & Miller 1997; Chen expedition. For the catch different bottom trawls (Blake, et al. 2002; Cheng et al. 2006). This consists of the 5.8S Otter and Agassiz trawl) were used. Samples were taken rRNA and the non-coding regions ITS1 and ITS2, which from depths between 100 and 648 meters. The captured separate the nuclear rRNA genes 18S, 5.8 S and 28S and animals were fixed directly in 96% ethanol. An over­ are cut off during splicing. For pycnogonids there is only view of the sequenced specimens with their geographical one recent study, which has clearly demonstrated the origin is given in Table 2. Specimens from the POKER utility of this marker to address species-level questions and REVOLTA expeditions are located in the Muséum (Arango & Brenneis 2013). National de l’Histoire Naturelle (MNHN), Paris, France, In this study, we analyzed sequence data from the under the catalog numbers IU-2007-4795 to 5069 and COI and ITS genes as well as measurements to test the IU-2013-15805 to 15812, respectively. distinction between C. glacialis, C. robusta, and other possible species similar to them. The aims of the inte­ grative taxonomic study were threefold: First, we wanted 2.2. Species determination to test whether the two morphologically similar C. ro­ busta and C. glacialis are conspecific F( ry & Hedgpeth 1969; Child 1995) or two distinct species (Stiboy-Risch Specimens were determined using a light microscope 1993; Cano & Lopez-Gonzalez 2007). As a second aim, and according to the identification keys of Child (1995), we tested for further evidence of unrecognized or cryptic Pushkin (1993), the original descriptions of the species, species within C. robusta and C. glacialis. In particular and the results of the study of Stiboy-Risch (1993). we tested whether Antarctic specimens assigned to C. ro­ busta belong to that species by studying specimens from the type locality around the Kerguelen Islands. Thirdly, 2.3. Molecular genetic analysis we investigated whether regional intraspecific variation can be detected in our dataset. DNA was isolated from muscle tissue (extracted from the tibia 1 with sterile tools), using the Qiagen QIAamp DNA mini Kit according to the manufacturer’s instructions. 2. Material and methods Part of the mitochondrial gene region cytochrome C oxidase subunit I (COI) and the nuclear ribosomal DNA region ITS were amplified. For this PCR was performed 2.1. Sampling with a mix containing 2.5 µl of 10 × PCR buffer (Hot­ Master), 2.5 µl dNTP mix (2 mM), 0.125 µl each of the two primers (100 pmol/µl) per gene segment, 0.1 µl Taq The material for the morphological research comes from polymerase (HotMaster; 5 U/µl) and 1 – 3 µl DNA. The the research expeditions ICEFISH 2004, Polarstern expe­ difference to 25 µl was filled up with HPLC water. Prim­ dition ANT XIV/2, REVOLTA I and III, and POKER II ers used for the amplification of the gene segments are “L’Austral”. Samples from the ICEFISH 2004 expedition listed in Table 3. PCR conditions for the COI/ITS gene

91 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

Table 2. Overview of specimens, their sampling locations, COI clade assignment and COI haplotype number in this study. Printed in bold are specimens for which COI sequences for the analysis were obtained from GenBank. For these specimens, no morphological measure­ ments could be obtained.

Specimen Sampling location Latitude Longitude Depth Species Clade Haplotype ITS [m] (COI) (COI) 300-1.4 Eastern Weddell Sea 70°50’28.79’’S 10°35’16.80’’W 268 C. glacialis 1 4 × 257-2.4 Eastern Antarctic Peninsula 64°54’45.00’’S 60°39’0.61’’W 158 C. glacialis 1 2 × AGT42/164 South Shetlands 62°7’59.99’’S 57°40’0.01’’W 555 C. glacialis 1 3 AGT42/175-7 South Shetlands 62°19’0.00’’S 58°42’0.00’’W 496 C. glacialis 1 27 × HM381691 South Shetlands 61°10’12.00’’S 56°0’21.60’’W 148 C. glacialis 1 2 HM381692 South Shetlands 62°29’27.60’’S 61°25’19.20’’W 122 C. glacialis 1 34 HM432370 Eastern Weddell Sea 71°7’8.40’’S 11°26’13.20’’W 228 C. glacialis 1 4 211-5.1 Shag Rocks 53°24’31.79’’S 42°40’41.99’’W 317 C. glacialis 2 9 × 211-6.3.2 Shag Rocks 53°24’48.60’’S 42°40’1.81’’W 315 C. glacialis 2 5 × 29OT27-1 Shag Rocks 53°27’38.88’’S 41°15’40.68’’W 191 C. glacialis 2 6 29OT27-2 Shag Rocks 53°27’38.88’’S 41°15’40.68’’W 191 C. glacialis 2 7 30BT14-2 Shag Rocks 53°20’3.48’’S 41°28’48.00’’W 146 C. glacialis 2 10 45BT24 South Georgia 54°15’0.00’’S 35°32’60.00’’W 100 C. glacialis 2 11 HM426185 South Georgia 53°36’39.60’’S 37°52’40.80’’W 224 C. glacialis 2 12 PA_E006 South Sandwich Islands 56°8’60.00’’S 27°20’24.00’’W 336 C. glacialis 2 12 (GQ386997) PB_E005 South Georgia 54°15’0.00’’S 35°32’60.00’’W 100 C. glacialis 2 14 (GQ386998) PR_E006 Shag Rocks 53°27’40.32’’S 41°15’38.16’’W 193 C. glacialis 2 13 PR_E010 Shag Rocks 53°27’55.44’’S 41°13’27.12’’W 195 C. glacialis 2 5 × PS_E011 Shag Rocks 53°27’40.32’’S 41°15’38.16’’W 193 C. glacialis 2 8 IU-2013-15805 Terre Adélie C. glacialis 6 28 × IU-2013-15808 Terre Adélie C. glacialis 6 29 × IU-2013-15812 Terre Adélie C. glacialis 6 1 HM381674 Ross Sea 74°42’0.00’’S 164°5’60.00’’E 20 C. glacialis 6 1 286-1.1.2 Eastern Weddell Sea 70°51’18.00’’S 10°35’21.01’’W 224 C. glacialis [only ITS] × 226-7.3 Eastern Antarctic Peninsula 64°54’50.40’’S 60°36’37.80’’W 216 C. robusta 3 18 257-2.1 Eastern Antarctic Peninsula 64°54’45.00’’S 60°39’0.61’’W 158 C. robusta 3 18 × Ch126.1 Western Antarctic Peninsula 67°43’36.42’’S 69°18’6.18’’W C. robusta 3 32 Ch231.1 Ross Sea 76°20’28.38’’S 170°51’1.78’’W C. robusta 3 33 HM381689 South Shetlands 61°0’39.60’’S 55°46’30.00’’W 162 C. robusta 3 18 HM381690 South Shetlands 61°0’57.60’’S 55°56’24.00’’W 274 C. robusta 3 17 HM426429 Ross Sea 74°35’25.80’’S 170°16’33.60’’E 283 C. robusta 3 16 HM432386 Eastern Weddell Sea 71°6’18.00’’S 11°32’2.40’’W 175 C. robusta 3 18 HM432414 South Orkneys 61°0’3.60’’S 45°51’54.00’’W 240 C. robusta 3 15 HM432416 South Orkneys 61°0’3.60’’S 45°51’54.00’’W 240 C. robusta 3 18 IU-2007-4795 Kerguelen 48°24’S 70°36’E 119 – 124 C. robusta 10 30 IU-2007-4797 Kerguelen 48°24’S 70°36’E 119 – 124 C. robusta 10 31 IU-2007-4798 Kerguelen 48°24’S 70°36’E 119 – 124 C. robusta 10 30 IU-2007-4800 Kerguelen 48°24’S 70°36’E 119 – 124 C. robusta 10 30 IU-2007-4842 Kerguelen 47°34’S 70°19’E 161 – 162 C. robusta 10 30 IU-2007-4870 Kerguelen 50°27’S 71°42’E 585 – 589 C. robusta 10 31 × IU-2007-4902 Kerguelen 47°32’S 69°42’E 179 – 180 C. robusta 10 31 IU-2007-5039 Kerguelen 48°40’S 70°23’E 117 – 118 C. robusta 10 30 × IU-2007-5043 Kerguelen 48°48’S 70°09’E 103 – 104 C. robusta 10 30 IU-2007-5044 Kerguelen 48°48’S 70°09’E 103 – 104 C. robusta 10 30 IU-2007-5058 Kerguelen 48°11’S 70°09’E 133 – 139 C. robusta 10 30 IU-2007-5063 Kerguelen 48°32’S 70°35’E 114 – 117 C. robusta 10 31 IU-2007-5069 Kerguelen 48°57’S 69°59’E 100 – 103 C. robusta 10 31 257-2.5 Eastern Antarctic Peninsula 64°54’45.00’’S 60°39’0.61’’W 158 C. drakei 4 14 233-3.1.2 Eastern Antarctic Peninsula 65°33’27.61’’S 61°37’17.40’’W 324 C. drakei 4 12 233-3.1.1 Eastern Antarctic Peninsula 65°33’27.61’’S 61°37’17.40’’W 324 C. drakei 4 13 226-7.2 Eastern Antarctic Peninsula 64°54’50.40’’S 60°36’37.80’’W 216 C. bouvetensis 5 2 × 59BT40 Bouvet Island 54°12’36.00’’S 3°6’48.96’’E 465 C. bouvetensis 5 10 × 59OT45 Bouvet Island 54°12’48.96’’S 3°6’10.80’’E 458 C. bouvetensis 5 11 × 66OT97 Bouvet Island 54°14’5.64’’S 3°18’41.04’’E 190 C. bouvetensis 5 11 81OT58 Bouvet Island 54°17’39.12’’S 3°8’34.80’’E 169 C. bouvetensis 5 8 × HM426194 Eastern Weddell Sea 71°4’51.60’’S 11°32’13.20’’W 295 C. bouvetensis 5 2 HM426263 South Sandwich Islands 57°40’15.60’’S 26°27’57.60’’W 630 C. bouvetensis 5 4 HM426326 Eastern Weddell Sea 71°6’18.00’’S 11°32’2.40’’W 175 C. bouvetensis 5 3 HM426374 Ross Sea 76°11’35.16’’S 176°17’45.60’’E 447 C. bouvetensis 5 5

92 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

Table 2 continued.

Specimen Sampling location Latitude Longitude Depth Species Clade Haplotype ITS [m] (COI) (COI) HM426402 Ross Sea 72°20’22.20’’S 175°31’55.20’’E 950 C. bouvetensis 5 5 HM426402 Ross Sea 72°20’22.20’’S 175°31’55.20’’E 950 C. bouvetensis 5 5 HM426434 Ross Sea 74°6’40.32’’S 170°47’45.60’’E 632 C. bouvetensis 5 7 HM432368 Eastern Weddell Sea 71°7’8.40’’S 11°26’13.20’’W 228 C. bouvetensis 5 3 HM432391 South Sandwich Islands 57°40’37.20’’S 26°25’26.40’’W 301 C. bouvetensis 5 4 PA_E003 South Sandwich Islands 56°8’60.00’’S 27°20’24.00’’W 336 C. bouvetensis 5 4 (GQ386999) PF_E008 Bouvet Island 54°20’25.08’’S 3°13’13.08’’E 648 C. bouvetensis 5 11 (GQ387000) PQ_E007 Bouvet Island 54°12’48.96’’S 3°6’10.80’’E 458 C. bouvetensis 5 11 × PQ_E008 Bouvet Island 54°12’48.96’’S 3°6’10.80’’E 458 C. bouvetensis 5 11 × PQ_E010 Bouvet Island 54°12’48.96’’S 3°6’10.80’’E 458 C. bouvetensis 5 11 × PQ_E011 Bouvet Island 54°12’48.96’’S 3°6’10.80’’E 458 C. bouvetensis 5 11 × PQ_E012 Bouvet Island 54°12’48.96’’S 3°6’10.80’’E 458 C. bouvetensis 5 11 × PR_E003 Bouvet Island 54°12’48.96’’S 3°6’8.64’’E 458 C. bouvetensis 5 6 × PR_E004 Bouvet Island 54°12’48.96’’S 3°6’8.64’’E 458 C. bouvetensis 5 11 × PR_E005 Bouvet Island 54°12’48.96’’S 3°6’8.64’’E 458 C. bouvetensis 5 11 × PS_E010 Bouvet Island 54°20’25.08’’S 3°13’13.08’’E 648 C. bouvetensis 5 9 × HM426382 Ross Sea 72°35’25.08’’S 175°20’31.20’’E 475 C. bouvetensis 7 1 HM426381 Ross Sea 72°35’25.08’’S 175°20’31.20’’E 475 C. bouvetensis 8 24 HM426401 Ross Sea 73°14’53.52’’S 178°43’26.40’’E 753 C. bouvetensis 8 26 HM426411 Ross Sea 75°37’27.12’’S 167°19’15.60’’E 474 C. bouvetensis 8 25 HM426433 Ross Sea 72°1’24.60’’S 173°10’48.00’’E 814 C. bouvetensis 8 23 HM426327 Eastern Weddell Sea 71°6’18.00’’S 11°32’2.40’’W 175 C. bouvetensis 9 21 HM426375 Ross Sea 76°12’7.20’’S 176°14’52.80’’E 447 C. bouvetensis 9 19 HM426383 Ross Sea 73°7’28.20’’S 174°19’12.00’’E 321 C. bouvetensis 9 19 HM426389 Ross Sea 76°35’38.40’’S 176°49’40.80’’E 365 C. bouvetensis 9 20 HM432388 Eastern Weddell Sea 71°6’18.00’’S 11°32’2.40’’W 175 C. bouvetensis 9 22 HM432397 Eastern Weddell Sea 70°50’6.00’’S 10°34’44.40’’W 274 C. bouvetensis 9 22

were: denaturation at 94°C for 120 s (both genes) fol­ was identified with BLAST search. The ITS region was lowed by 38 / 37 cycles of 94°C denaturation for 20 s identified by alignment with theC. robusta sequences. (both), annealing for 30 s at 46°C / 55°C, extension at Both COI and ITS sequences were aligned with the 65°C for 60 s / 80 s followed by a final extension at 65°C program MAFFT (Katoh & Standley 2013) as imple­ for 5 min /10 min, respectively. mented in Geneious using algorithm autodetection with PCR products were purified enzymatically: For each a gap opening penalty of 1.53 and offset value of 0.123. 10 µl of the PCR product 0.75 µl ExoI (exonuclease I For the ITS alignment, poorly aligned regions were re­ 20 U/µL) and 1.5 µl FastAP (Fast Alkaline Phosphatase moved with the program Gblocks using less stringent 1 U/µL) were added. The reaction was carried out in a parameters (Talavera & Castresana 2007). In both thermocycler at 37°C for 15 min and inactivated by in­ cases, the program RAxML 7.0.4 (Stamatakis 2006) was cubation at 85°C for 15 min. Purified PCR products were used to calculate a maximum likelihood tree with the sequenced by GATC (Cologne, Germany) bidirectionally model GTR+G. Bootstrap support was calculated using by the Sanger method. 1000 replicates with the approximation GTR+CAT. Mr­ The resulting COI and ITS sequences were edited Bayes 3.2.1 (Ronquist et al. 2012) was used to calculate with Geneious version 5.6.6 (Drummond et al. 2012). a Bayesian tree with 5,000,000 MCMC generations, of The amplified gene fragments were checked by a BLAST which the first 25% were discarded as burn-in. Models of search against the NCBI GenBank to exclude contamina­ evolution for Bayesian analysis were chosen with jMod­ tion. For COI, sequences from all individuals listed in eltest 2.1.2 (Darriba et al. 2012) with the Akaike Infor­ Table 2 (except for 286-1.1.2, for which amplification mation Criterion (AIC). For COI the model GTR+I+G was unsuccessful) and 22 sequences from GenBank, re­ was chosen, and for ITS SYM+G was preferred. For the presenting other colossendeid species, were used. For COI trees, the sequence from Rhopalorhynchus filipes ITS, sequences from all 19 individuals for which PCR Stock, 1991 was used as an outgroup taxon, and the ITS amplification was successful were used (Table 2). Addi­ trees were rooted with sequences of Nymphon australe tionally, an ITS sequence for C. megalonyx clade E was Hodgson, 1902 and Pseudopallene constricta Arango & generated from the 454 data published by Leese et al. Brenneis, 2013 published by Arango & Brenneis (2013). (2012). Contigs were assembled from the reads with Ge­ With the TCS 1.21 program (Clement et al. 2000), par­ neious, and the contig representing the ribosomal operon simony networks of the COI haplotypes were calculated

93 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

Table 3. Primers used for amplification of the mitochondrial (mtDNA) and the nuclear (nDNA) gene regions.

Gene region Primer Sequence Source LCO 1490 5’ GGT CAA CAA ATC ATA AAG ATA TTG G 3‘ Folmer et al. (1994) COI (mtDNA) HCO 2198 5’ TAA ACT TCA GGG TGA CCA AAA AAT CA 3‘ Folmer et al. (1994) ITSRA2 5’ GTC CCT GCC CTT TGT ACA CA 3’ Wörheide (1998) ITS (nDNA) ITS2.2 5’ CCT GGT TAG TTT CTT TTC CTC CG 3’ Wörheide (1998)

Table 4. Morphometric measurements taken for the examined specimens.

Body part Measurement Trunk Length From proximal edge of the trunk to distal edge of the 4th lateral processes Trunk Height Centrally between 2nd and 3rd lateral processes vertical Trunk Widest point Width of the 2nd lateral processes Trunk Width between 2nd and 3rd lateral processes Centrally between 2nd and 3rd lateral processes parallel to the axial plane Proboscis Length Proximal to distal edge Proboscis Diameter Averaged (2 distances) widest point Abdomen Length Proximal to distal edge Abdomen Width Widest point at the proximal end Ocular tubercle Height Base to tip of the ocular tubercle Ocular tubercle Width Width at ocular tubercle base Ocular tubercle Front height Top of eyes to tip of ocular tubercle; both anterior and posterior Ocular tubercle Eye size Vertical longest distance of the eye Palpi Length Proximal to distal edge; all 10 articles; unilaterally Oviger Length Proximal to distal edge; all 10 articles; unilaterally Oviger Width Widest point, all 10 articles; unilaterally Basal element Diameter Diameter widest point, unilaterally Lateral processes Distance Average distance between the neighboring lateral processes at the distal end Lateral processes Width Average width of lateral processes 1 – 4 at the distal end Legs Length Longest distance from proximal to distal edge; coxae I – III, femur, tibia I – II, tarsus, propodus, claw, Averaged length of left and right Legs Width Widest point; coxa I – III, femur, tibia I – II, tarsus, propodus, claw; averaged length of left and right

with a connection limit of 95%. Genetic distances for the were found to be not normally distributed, a non-para­ COI data were calculated with PAUP 4.0b10 (Swofford metric unifactorial Kruskal-Wallis ANOVA was used to 2002) using the Kimura 2-parameter (K2P) correction. test for significant differences between molecular clades. Significance was assessed using a comparison of the paired groups’ mean ranks. A principal component analy­ 2.4. Morphological analyses sis (PCA) with those values available in all individuals was performed to visualize the clustering of specimens based on morphological data. Body measurements were carried out using the digital For SEM analyses the specimens AGT 42/164 (C. caliper “IP54 Water-resistant Digital Caliper” model glacialis clade 1), PA_E006 and 29OT27-2 (C. glacialis “Digimatic”, calibrated to one hundredth of a millimeter. clade 2), 233-3.1.1 (C. drakei), 257-2.1 (C. sp. cf. ro­ All animals were used except for three of the Kerguelen busta), 226-7.2 and PF_E008 (C. bouvetensis sp.n.) were specimens and one specimen from the Antarctic Peninsu­ used. Ovigera of the specimens were dried by adding la (226-7.3) as, despite damage during storage, transport, hexamethyldisilazane (HMDS) in a rising concentration trawling or the preceding genetic analysis all limbs could every 15 min. For electrical conductivity the material be at least partly measured. However, in some cases, dis­ was sputtered with gold for 180 s. SEM pictures were tal leg articles, in particular tibia II, tarsus, propodus and taken with the scanning electron microscope DSM 950 claw were missing. Overall, if all limbs were present, up (ZEISS) and documented with the software program to 124 measurements per specimen were taken. Relative Digital Image Processing System (DIPS). Image quality values expressed as proportions of the trunk length were was increased through the image editing program Adobe used for analyses to avoid biases caused by different ab­ Photoshop CS3. The light microscopic pictures of mu­ solute sizes. All measurements are shown in Table 4. seum material (MNHN IU-2013-15812 and MNHN IU- Using the program Statistica (StatSoft), the morpho­ 2007-5063) were taken under an Olympus SZX16 with logical measurements were first tested for normal distri­ the camera type Olympus SDF Plapo 1.6XPF using the bution using the Lilliefors test. As many measurements software CELLD.

94 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

Table 5. Mean Kimura 2-parameter distances between COI sequences within clades (diagonal) and between different clades (below di­ agonal).

C. glacia­ C. glacia­ C. glacia­ C. robusta C. robusta C. stra- C. drakei C. bouve­- C. bouve­- C. bouve­- C. bouve­- lis 1 lis 2 lis 6 3 10 ­men­ti tensis 5 tensis 7 te­nsis 8 tensis 9 C. glacialis 1 0.36 C. glacialis 2 4.79 0.62 C. glacialis 6 3.17 4.96 1.07 C. robusta 3 16.44 14.85 17.07 1.22 C. robusta 10 17.23 14.35 17.94 4.6 0.09 C. stramenti 17.28 15.43 17.91 10.54 10.9 0 C. drakei 16.83 16.17 17.14 9.17 10.87 9.68 0.25 C. bouvetensis 5 18.26 16.23 17.29 10.25 11.53 12.23 11.27 0.9 C. bouvetensis 7 17.19 15.82 16.67 10.09 11.91 11.16 10.42 5.15 0 C. bouvetensis 8 16.73 15.17 15.29 10.53 11.24 10.99 10.56 4.01 5.28 1.05 C. bouvetensis 9 17.06 16.53 15.83 10.05 12.2 11.87 9.3 5.76 4.87 5.23 0.53

3. Results form one well-supported (posterior probability 1, ML bootstrap 100) and distinct group. Clade 5 (n = 23) is represented by specimens from Bouvet Island, the South 3.1. Molecular analysis Sandwich Islands, the Eastern Weddell Sea and the Ross Sea. Clades 7 (n = 1) and 8 (n = 4) are known only from the Ross Sea, and clade 9 (n = 6) is known from the Ross 3.1.1. Phylogenetic results and Eastern Weddell Seas. Interclade distances between In the phylogenetic trees based on the COI sequences, these four clades range from 4 to 5.8%, while the pair­ the examined sequences fall into ten clearly separated wise distances to C. robusta s.str. (clade 10) range from clades, five of which (clades 1 – 5, Fig. 1) are represent­ 11.2 to 12.2%. Members of clades 5, 7, 8 and 9 are as­ ed in our material, while the remaining five are known signed to C. bouvetensis sp.n. in this study. only from GenBank sequences. Kimura 2-parameter dis­ The phylogenetic trees based on the ITS data (Fig. 2), tances between the clades are given in Table 5. Clades although only present for a subset of specimens, show the 1, 2 and 6 are closely related to each other and consist same topology as the COI data (Fig. 1) in that C. glacia­ of (and include all) specimens identified as C. glacialis, lis (COI clades 1, 2 and 6) groups basally and is clearly with interclade genetic distances ranging from 3.2 to 5%. separated from C. bouvetensis (COI clade 5) and from Clade 1 (n = 7) consists of specimens from the Antarc­ C. robusta and C. sp. cf. robusta (COI clades 3 and 10), tic Peninsula and Eastern Weddell Sea, clade 2 (n = 12) which are closely related to each other. The COI clades 1, of specimens from the Shag Rocks, South Georgia, and 2 and 6 are not recovered within C. glacialis. We did not the South Sandwich Islands, and clade 6 (n = 4) is repre­ succeed in obtaining ITS sequences from C. drakei. sented by specimens from the Ross Sea and Terre Adélie. While support for intraspecific relationships in the tree 3.1.2. Phylogeographic results are generally poor, there is no support for a close rela­ tionship of C. glacialis with C. robusta s.str. The genetic The TCS analysis result in separate networks for all ten distances between C. glacialis and C. robusta and related clades at the 95% connection limit. Within each clade taxa range from 14.4 to 18.3%. with specimens from different regions, geographical Specimens from clade 3 (n = 10) came from very partitioning was obvious (Fig. 3). In clade 1, the Eastern different regions of the Antarctic, i.e. South Orkneys, Weddell Sea specimens group separated from those from Antarctic Peninsula, Eastern Weddell Sea, and Ross the Antarctic Peninsula and South Shetlands. In clade 2, Sea, thus showing a wide distribution. Clade 3 forms a no shared haplotypes are found between Shag Rocks and strongly supported sister-group relationship with clade South Georgia, although haplotypes from both regions 10 (n = 13), which includes the specimens from Ker­ are not separated in the network. The single specimen guelen identified as C. robusta s.str. (4.6% interclade from South Sandwich is widely separated from the oth­ distance). Clade 4 (n = 3) includes specimens from the ers. Clade 3 is divided into two groups, one including Antarctic Peninsula identified here as C. drakei Calman, sequences from the Ross Sea and one sample from the 1915. There is one grouping with good statistical support Western Antarctic Peninsula, the other including se­ (posterior probability 1, ML bootstrap 84) that includes quences from the South Orkneys, South Shetlands, East­ clades 3, 10 and 4 as well as C. stramenti (DQ390078; ern Antarctic Peninsula, and Eastern Weddell Sea. Clade sister to C. drakei). Specimens determined first as C. 5 shows a separation into three groups, one of which is robusta (clade 5) clustered together with sequences of found in the Ross Sea and Bouvet Island, the second in high similarity from GenBank (clades 7, 8 and 9) and the South Sandwich Islands and Eastern Weddell Sea, and

95 PQ_E007 W. Antarctic South PR_E004 GQ387000 Peninsula Shetlands PR_E005 59OT45 PQ_E010 PQ_E008 E. Antarctic Eastern 66OT97 PQ_E012 Peninsula Weddell Sea PQ_E011 HM426374 HM426402 clade 5 HM426434 South Terre Adélie PR_E003 Georgia 59BT40 PS_E010 C. bouve- 81OT58 91 GQ386999 1 Shag HM426263 tensis Ross Sea HM432391 Rocks HM426326 93 HM432368 sp.n. 1 HM426194 226-7.2 61 South 0.99 HM426411 Bouvet Island HM426433 Sandwich 66 HM426381 0.98 clade 8 HM426401 100 HM426382 1 HM426383 clade 7 South HM426375 Kerguelen HM426389 Orkneys 89 HM432397 clade 9 HM432388 HM426327 IU-2007-5044 IU-2007-5043 IU-2007-4798 IU-2007-5039 IU-2007-4795 IU-2007-5058 84 IU-2007-4842 1 IU-2007-4800 100 IU-2007-4797 C. robusta 1 IU-2007-5069 Proof _IU-2007-5063 01 (clade 10) Proof _ 01 IU-2007-4870 IU-2007-4902 Dietz et al.: Southern Ocean sea spider phylogeny and diversity HM432416 100 HM381689 1 HM432386 257-2.1 86 226-7.2 C. sp. cf. robusta 1 HM381690 HM432414 PQ_E007 Fig. 1. Maximum-likelihood phylogenetic tree based on a 545-bp HM426429 W. Antarctic South 60 PR_E004 (clade 3) 73 Ch231.1 alignment of 105 colossendeid COI sequences. Numbers0.5 above GQ387000 Ch126.1 Peninsula Shetlands1 PR_E005 257-2.5 branches correspond to bootstrap support values in the maximum- 59OT45 100 233-3.1.2 65 PQ_E010 1 233-3.1.1 likelihood analysis, those below branches to Bayesian posterior PQ_E008 C. drakei (clade 4) 0.92 DQ390078 Colossendeis stramenti 66OT97 probabilities. BootstrapE. Antarctic supports under 50%Eastern not shown. HQ970327 Colossendeis angusta 90 90 PQ_E012 0.79 1 HQ970328 Colossendeis angusta Peninsula Weddell Sea PQ_E011 94 DQ390061 Colossendeis tenera 1 HM426374 KC462566 Colossendeis tenera HM426402 clade 5 GQ387006 Colossendeis scotti 84 0.67 HM426434 FJ969359 Decolopoda australis South 0.97 99 PR_E003 0.56 71 1 FJ969360 Decolopoda australis Terre Adélie PQ_E007 59BT40 0.96 GQ386992 Dodecolopoda mawsoni W. Antarctic GeorgiaSouth PR_E004 PS_E010 GQ387001 Colossendeis scoresbii C. bouve- GQ387000 81OT58 Peninsula Shetlands GQ38701591 Colossendeis megalonyx 89PR_E00596 GQ386999 0.68 1 1 GQ387016 Colossendeis megalonyx 0.9859OT45 HM426263 tensis Shag PQ_E01089 GQ387026 Colossendeis megalonyx Ross Sea 99 HM432391 PQ_E0080.98 GQ387025 Colossendeis megalonyx 1 HM426326 Rocks 89 GQ387027 Colossendeis megalonyx E. Antarctic Eastern 66OT97 93 HM432368 sp.n. 0.94PQ_E012 GQ387008 Colossendeis megalonyx 1 HM426194 FJ969355 Colossendeis australis Peninsula Weddell Sea PQ_E011 226-7.2 GQ386998HM42637461 91 HM426411 South HM4261850.99 Bouvet Island HM426402 cladeHM426433 5 0.93 PR_E006 Sandwich HM426434 66 HM426381 211-5.1 0.98 clade 8 South PR_E003 HM426401 Terre Adélie 30BT14-2 10059BT40 HM426382 Georgia 45OT24 1 PS_E010HM426383 clade 7C. bouve- PR_E01081OT58 clade 2 South 91 HM426375 211-6.3.2GQ386999 Kerguelen 51 1 HM426389 PS_E011HM426263 Orkneys 0.71 89 HM432397 tensis Shag 29OT27-1 clade 9 Ross Sea HM432391HM432388 29OT27-2 Rocks HM426326HM426327 GQ386997 93 HM432368 IU-2007-5044 sp.n. 257-2.4 1 HM426194 IU-2007-5043 C. glacialis 95 HM381691226-7.2 61 IU-2007-4798 AGT42/164 1 0.99 HM426411 IU-2007-5039 South HM381692 Bouvet Island 87 HM426433 IU-2007-4795 66HM432370 clade 1 Sandwich 1 HM426381 IU-2007-5058 0.98 300-1.4 clade 8 84 HM426401 IU-2007-4842 100 81 HM426382AGT42/175-7 1 IU-2007-4800 1 0.96 HM381674 HM426383 clade100 IU-2007-47977 C. robusta IU-2013-15812 South HM426375 1 IU-2007-5069 Kerguelen HM42638983 IU-2013-15805 0.91 cladeIU-2007-5063 6 Orkneys IU-2013-15808 (clade 10) 89 HM432397 IU-2007-4870 FJ716625 Colossendeisclade minuta 9 HM432388 IU-2007-4902 FJ716624 Colossendeis colossea HM426327 HM432416 0.66 FJ862873 ColossendeisIU-2007-5044 macerrima 100 HM381689 FJ862872 Rhopalorhynchus filipes IU-2007-5043 1 HM432386 IU-2007-4798 257-2.1 IU-2007-5039 0.2 substitutions per site 86 226-7.2 IU-2007-4795 C. sp. cf. robusta 1 HM381690 IU-2007-5058 HM432414 84 IU-2007-4842 HM426429 60 (clade 3) 1 73 IU-2007-4800 Ch231.1 0.5100 1 IU-2007-4797 Ch126.1 C. robusta 1 IU-2007-5069257-2.5 IU-2007-5063100 233-3.1.2 (clade 10) 65 IU-2007-48701 233-3.1.1 C. drakei (clade 4) 0.92 IU-2007-4902 DQ390078 Colossendeis stramenti HM432416 90 90 HQ970327 Colossendeis angusta 100 HM3816890.79 1 HQ970328 Colossendeis angusta 1 94HM432386 DQ390061 Colossendeis tenera 1 257-2.1 KC462566 Colossendeis tenera 226-7.2 84 0.67 86 GQ387006C. Colossendeissp. scotti cf. robusta 1 HM381690 FJ969359 Decolopoda australis 0.97 99 0.56 71 HM4324141 FJ969360 Decolopoda australis HM4264290.96 GQ386992(clade Dodecolopoda 3) mawsoni 73 60 0.5 Ch231.1 GQ387001 Colossendeis scoresbii 1 Ch126.1 89 96 GQ387015 Colossendeis megalonyx 0.68257-2.5 1 GQ387016 Colossendeis megalonyx 100 0.98 233-3.1.2 89 GQ387026 Colossendeis megalonyx 65 1 99 233-3.1.1 0.98 GQ387025C. Colossendeis drakei megalonyx (clade 4) 0.92 1 DQ39007889 Colossendeis stramentiGQ387027 Colossendeis megalonyx 90 90 0.94HQ970327 ColossendeisGQ387008 angusta Colossendeis megalonyx 1 0.79 HQ970328FJ969355 Colossendeis Colossendeis angusta australis 94 GQ386998DQ390061 Colossendeis tenera 91 1 KC462566HM426185 Colossendeis tenera 84 0.93 0.67 GQ387006PR_E006 Colossendeis scotti FJ969359 Decolopoda australis 0.97 99 211-5.1 0.56 71 1 30BT14-2FJ969360 Decolopoda australis 0.96 GQ38699245OT24 Dodecolopoda mawsoni GQ387001PR_E010 Colossendeis scoresbiiclade 2 89 96 GQ387015211-6.3.2 Colossendeis megalonyx 0.68 51 0.98 1 GQ387016 Colossendeis megalonyx 0.71 PS_E011 89 GQ387026 Colossendeis megalonyx 99 29OT27-1 0.98 GQ387025 Colossendeis megalonyx 1 29OT27-2 89 GQ386997GQ387027 Colossendeis megalonyx 0.94 GQ387008257-2.4 Colossendeis megalonyx C. glacialis 95 FJ969355 ColossendeisHM381691 australis GQ386998 91 1 AGT42/164 HM426185 87 HM381692 0.93 PR_E006 clade 1 1 HM432370 211-5.1 300-1.4 30BT14-2 81 AGT42/175-7 45OT24 0.96 HM381674 PR_E010 cladeIU-2013-15812 2 211-6.3.2 83 51 IU-2013-15805 0.91 clade 6 0.71 PS_E011 IU-2013-15808 29OT27-1 FJ716625 Colossendeis minuta 29OT27-2 FJ716624 Colossendeis colossea 0.66 GQ386997 FJ862873 Colossendeis macerrima FJ862872 Rhopalorhynchus257-2.4 filipes C. glacialis 95 HM381691 1 AGT42/164 0.2 substitutions per site 87 HM381692 clade 1 1 HM432370 96 300-1.4 81 AGT42/175-7 0.96 HM381674 IU-2013-15812 83 IU-2013-15805 0.91 clade 6 IU-2013-15808 FJ716625 Colossendeis minuta FJ716624 Colossendeis colossea 0.66 FJ862873 Colossendeis macerrima FJ862872 Rhopalorhynchus filipes

0.2 substitutions per site PQ_E007

PS_E010

PQ_E011

PR_E005

PQ_E010

81OT58 59BT40 C. bouvetensis PR_E004 99 (clade 5) 1 59OT45 Proof _ 01 PQ_E008 Proof _ 01 PQ_E012 100 1 PR_E003 ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015 226-7.2

100 IU-2007-4870 100 1 IU-2007-5039 C. robusta Fig. 2. Bayesian phylogenetic1 tree based on a 1138-bp PQ_E007 257-2.1 alignment of 20 colossendeid ITSC. sequences. sp. cf. Numbers robusta PS_E01PQ_E000 7 C. megalonyx above branches correspond to bootstrap support values in PS_E010 PQ_E01PQ_E001 7 PQ_E007 257-2.4 PQ_E011 100 the maximum-likelihood analysis, those below branches PR_EPS_E01005 0 PQ_E00PS_E0170 1 to BayesianAGT42/175- posterior7 probabilities. Colors as in Fig. 1. PR_E005 PQ_E011 PQ_E01PS_E01PQ_E0100 1 PQ_E010 211-6.3.2 PR_E005 PQ_E01PR_E0015 81OT8158OT58 PQ_E010 IU-2007-15805 PR_PQ_E01E005 0 59B59BT40T40 81OT58 C. bouvetensis 81OT58 C. bouvetensis IU-2007-15808 PQ_E010 PR_E004 <50 P99R_E59B00T404 99 59BT40 C. glacialis 81OT58 C.(clade bouvetensis 5) PR_E010 1 59OT45 C. bouvetensis 0.96 1 PR_E004 (clade 5) 5999 OTPR_45E004 99 59BT40 211-5.1 1 PQ_E008 C.(clade bouvetensis 5) 1 59OT45 (clade 5) PQ_E00PR_59EOT008445 99 PQ_E012 286-1.1.2 100 PQ_E008 (clade 5) 1 59PQ_E00OT45 8 PQ_E01PR_E2003 1 PQ_E012 300-1.4 100 PQ_E012 100 PQ_E008 1 100 PR_E226-7.003 2 1 PR_E003 Nymphon australe (JX196726) 1 PQ_E01PR_E002 3 100 IU-2007-4870 100 226-7.2 226-7.2 100PR_226-7.E0032 Pseudopallene constricta (JX196738)1 1 IU-2007-5039 C. robusta 100 100 IU-2007-4871 IU-2007-4870 0 100 226-7.IU-2007-4872 0 257-2.1 1 100 C. sp. cf. robusta 1 1 100 100 IU-2007-5039 0.1 100 IU-2007-5031 IU-2007-487IU-2007-50390 9 C. robusta 1 1 C. megalonyx C. robustarobusta 257-2.1 100 257-2.1 1 257-2.1IU-2007-5039 C. sp. cf. robusta 257-2.4 C. sp.C. cf. robusta 100 1 C. sp. cf. robusta C. megalonyx 257-2.1C. megalonyx 1 AGT42/175-7 C. megalonyxC. sp. cf. robusta 257-2.4 100 257-2.4 100 C. megalonyx 257-2.4211-6.3.2 100 1 AGT42/175-7 1 257-2.4AGT42/175-7 100 IU-2007-15805 1 AGT42/175-211-6.3.27 1 AGT42/175-211-6.3.27 IU-2007-15808 <50 IU-2007-15805 211-6.3.2211-6.3.2IU-2007-15805 PR_E010 C. glacialis IU-2007-15808 0.96 IU-2007-15808 <50IU-2007-1580IU-2007-15805 5 <50 211-5.1 C. glacialis PR_E010 C. glacialis 0.96 IU-2007-1580PR_E010 8 0.IU-2007-158096 286-1.1.28 <50 211-5.1 <50 PR_211-5.1E010 C. glacialis 300-1.4 C. glacialis 0.96PR_E010 286-1.1.2 0.96 211-5.1286-1.1.2 Nymphon australe (JX196726) 211-5.1300-1.4 286-1.1.300-1.42 Pseudopallene constricta (JX196738) 286-1.1.2 Nymphon australe (JX196726) 300-1.4 Nymphon australe (JX196726) 0.1 Pseudopallene constricta (JX196738) 300-1.4 NymphonPseudopallene australe (JX196726 constrict) a (JX196738)

0.1 Pseudopallene constricta (JX196738) 0.1 Nymphon australe (JX196726)

0.1 Pseudopallene constricta (JX196738)

0.1

the third in the Antarctic Peninsula and Eastern Weddell absolute length of the trunk is much greater, and the tibia Sea. Except for the latter group, no haplotypes are pre­ 2 is longer. The ovigera and palps are in general longer sent in more than one geographical area. Clade 9 shows and the ocular tubercle is lower. Compared to C. bou­ a division between the Ross and Weddell Sea specimens. vetensis, the absolute trunk length is greater, the femur is longer and the distal palp articles are longer. In most of these characters, clade 3 (C. sp. cf. robusta), agrees with clade 10. However, it differs in the proportions of 3.2. Morphological analyses the palps and the ovigera, which are overall longer than in clade 10, in having a higher ocular tubercle with larger eyes, and in having somewhat longer distal leg articles. 3.2.1. Morphometric measurements As clade 3 is represented by only one individual, the sig­ All measurements, as well as a list of significantly dif­ nificance of its differences to the other clades could not ferent measurements, are given in the supplementary be tested. It should be noted that the animal is a juve­ material. The number of significantly different measure­ nile with chelifores and lacking genital pores. However, ments between each pair of clades is given in Table 6. it shows some noticeable differences from other clades. No significant differences were found between clades 1, Clade 5, here named C. bouvetensis, is the most diver­ 2 and 6, all of which were determined as C. glacialis, gent of the groups examined here and significantly dif­ except for the larger anterior front height of the ocular fers from all other clades except for clade 3 in at least tubercle in clade 6 compared to clade 2. Clade 1 has a 25 characters. The animals are absolutely larger and all somewhat shorter proboscis than clade 2, with clade 6 leg articles broader than C. glacialis and C. drakei. The being intermediate. There are some other slight differ­ proboscis is longer, the ocular tubercle broader, the 10th ences but none of them are significant at the p < 0.05 lev­ palp article shorter and all ovigeral articles longer and el. Clade 10 (C. robusta from Kerguelen) differs from C. broader than in all others, except for article 6, which is glacialis in many respects. Compared to C. glacialis, the shorter. Coxa 3 is longer, femur and tibia 1 shorter than in

97 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

C. glacialis C. bouvetensis C. sp. cf. robusta Fig. 3. Statistical parsimony networks of sp. nov COI haplotypes for C. glacialis clades 1 and 2, C. sp. cf. robusta, and C. bou­ vetensis. Diameters of circles represent number of specimens per haplotype, colors represent geographical origin of specimens. Unsampled (i.e. hypothetical) haplotypes are represented by dots. Clade 2

Clade 9 10 substitutions

11 substitutions

Clade 6

Clade 1 Clade 5

Clade 3

South Georgia Bouvet Island S.Shetland Isl. E. Weddell Sea 10 5 Ross Sea South Orkneys Shag Rocks E. Ant. Penins. 3 2 Terre Adèlie W. Ant. Penins. S. Sandwich Isl. 1

Fig. 4. Results of principal component analysis of morphometric measurements. Squares: C. robusta. Star: C. sp. cf. robu­ 6 C. robusta sta. Crosses: C. dra­kei. White circles: C. gla­cialis clade 2. Grey circles: C. gla­ cia­lis clade 1. Black circles: C. glacialis clade 6. White hexagons: C. bouve­tensis, 4 C. drakei Bou­vet Island. Grey hexagons: C. bouve­ ten­sis, South Sandwich Islands. Black he­ xa­gons: C. bouvetensis, Antarctic Penin­ 11.68% 2 sula. C. sp. cf. robusta Factor 2 : Factor 0

-2 C. bouvetensis sp.n.

-4 C. glacialis

-15 -10 -5 0 5 10 Factor 1: 53.19%

98 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

Table 6. Number of measurements that are significantly different between the COI clades.

C. glacialis 1 C. glacialis 2 C. glacialis 6 C. robusta 3 C. robusta 10 C. drakei C. bouvetensis 5 C. glacialis 1 — C. glacialis 2 0 — C. glacialis 6 0 1 — C. robusta 3 0 0 0 — C. robusta 10 13 28 10 0 — C. drakei 1 3 2 0 9 — C. bouvetensis 5 18 77 25 0 26 40 —

others. Propodus and claw are longer than in C. glacialis. trast, the configuration in C. bouvetensis strongly differs Clade 4 (C. drakei), is most similar to C. glacialis, with (Fig. 5B,E,I,L). The long spines are much less closely only 1 – 3 significantly different measurements, the only spaced and less pointed than in C. glacialis, a distinct consistent one being the longer fifth ovigeral article, but row of medium spines is absent, and there are about two 40 are significantly different to clade 5. In general, all rows of irregularly placed short spines which are some­ leg articles are more slender than in all other examined what sparser than in C. glacialis. In the specimen 257- specimens, the trunk is shorter, the femur is longer, the 7th 2.1 (C. sp. cf. robusta Fig. 5C,F), all spines insert on palp article is shorter, the 5th oviger article is longer than a ridge running along the oviger article. Endally of the in C. glacialis and C. sp., the 10th palp article is longer row of long spines, there are irregularly spaced spines of than in C. glacialis and C. bouvetensis, and the propodus decreasing length, and the most endal ones point endally and claw are longer than in C. glacialis. much more strongly than in other specimens. A simi­ In the principal component analysis (Fig. 4), C. robus­ lar condition appears to be present in C. robusta s.str. ta, C. sp. cf. robusta and C. bouvetensis clearly formed (Fig. 7O,P), where however the long spines are oriented separate clusters. C. glacialis and C. drakei group apart parallel to the length of the article, not perpendicular as from the other groups but the space occupied by them in 257-2.1. The ovigeral claw is rather short compared overlaps. The PCA did not clearly separate the northern, to the terminal article in C. glacialis, much longer and southern and East Antarctic groups of C. glacialis (clades more slender in C. bouvetensis, and intermediate in 1, 2 and 6). Interestingly, the two measured C. bouveten­ C. ro­busta s.str. and 257-2.1. In C. drakei it appears to sis specimens from the Antarctic Peninsula (226-7.2) and be bi­furcated. South Sandwich Islands (PA_E003) grouped apart from a large cluster formed by all measured Bouvet Island speci­ mens. Factor 1 of the PCA, explaining 53.2% of the vari­ ance, was correlated most strongly with measurements of the width of leg and oviger articles but also e.g. the length 4. Taxonomy of the femur and tibia 1, and factor 2, explaining 11.7%, was correlated most strongly with the absolute length of the trunk and the height of the ocular tubercle. Colossendeis bouvetensis sp.n., Dietz & Leese, 2014

3.2.2. SEM data The examined specimens of C. glacialis show three The new species is attributed to the family Colossendei­ clearly distinct types of ovigeral spines (Fig. 5D,K), in­ dae because of its long proboscis, absence of chelifores cluding from ectal to endal one row of long spines, one (in adults), presence of 10-articled palps and ovigera, and row of medium spines, and irregularly placed short spines absence of auxiliary claws. Within the Colossendeidae, placed in two not clearly distinct rows, agreeing with the it belongs to the genus Colossendeis because of its lack illustrations of Cano & Lopez-Gonzalez (2007). This of visible segmentation, non-reduced abdomen, and pres­ condition is similar to C. tenera and C. megalonyx (Cano ence of only four pairs of legs. & Lopez-Gonzalez 2007; Dietz et al. 2013). In the speci­ men 29OT27-2 (C. glacialis clade 2; Fig. 5A,D) the long Description. Trunk length about 9 – 11 cm. Body and ex­ spines show a constriction in the middle, which is not tremities slightly setose. Ocular tubercle slightly broader apparent in the specimen AGT42/164 (C. glacialis clade than high, conical with a rounded tip, with well developed 1; Fig. 5H,K) and IU-2007-15812 (C. glacialis clade 6; and pigmented eyes. Lateral processes separated by about Fig. 7N). A similar configuration is also shown in the 2/3 their own diameter. Proboscis cylindrical with slight C. drakei specimen 233-3.1.1 (Fig. 5G,J), which howev­ dilations in the middle and at the distal end, straight, di­ er differs by the orientation of the medium spines being rected slightly downwards (type B’’’: 1 according­ to Fry perpendicular to the length of the oviger article. In con­ & Hedgpeth 1969), about the same length as the trunk.

99 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

Fig. 5. Ovigeral claw (top) and spine configuration (bottom) of examinedColossendeis specimens. A,D: 29OT27-2 (C. glacialis clade 1). B,E: 226-7.2 (C. bouvetensis, probably juvenile, Antarctic Peninsula). C,F: 257-2.1 (C. sp. cf. robusta, probable juvenile, Antarctic Peninsula). G,J: 233-3.1.1 (C. drakei). H,K: AGT42/164 (C. glacialis clade 2). I, L: PF_E008 (holotype of C. bouvetensis, Bouvet Island).

100 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

Table 7. Morphological differences between the species examined in this study. Characters unique for Colossendeis bouvetensis are marked in bold.

C. glacialis C. drakei C. robusta s.str. C. sp. cf. robusta C. bouvetensis Trunk length (mm) 8 – 10.5 6.5 – 9 14 – 20 >12 8 – 11 Ocular tubercle Higher than broad Higher than broad Broader than high Higher than broad Broader than high Proboscis : trunk length 0.7 – 0.95 0.8 – 0.85 0.9 – 1.05 0.85 0.95 – 1.05 ratio B’:2 (distally much Proboscis shape A’:1 to B’’’:1 B’’’:1 more dilated than in B’’’:1 B’’’:1 other species) Spinousness of body Slightly to very spinous Not spinous Not spinous Not spinous Slightly spinous Distance between lateral ~1.1 diameters ~0.7 diameters ~0.75 diameters ~0.7 diameters ~0.7 diameters processes Ratio 3rd : 5th palp article 1.2 – 1.6 1.45 – 1.65 1.3 – 1.5 1.15 1.7 – 2.2 7th palp article Longest of distal 5 Not longest of distal 5 Longest of distal 5 Longest of distal 5 Longest of distal 5 Ratio 10th : 9th palp article 0.75 – 1.25 1 – 1.15 0.8 – 1 0.95 0.3 – 0.7 3 types of spines, ~2 3 types of spines, ~2 2 types of spines, 3 – 4 2 types of spines, ~3 2 types of spines, ~2 Ovigeral strigilis rows of short spines rows of short spines rows of short spines rows of short spines rows of short spines Ovigeral claw Short, robust Short, bifurcated? intermediate intermediate Long, thin Femur length : width ratio 8.5 – 14 13 – 17 ~9 ~9 ~7 Tibia 1 : femur ratio ~1 ~0.85 ~0.93 ~0.9 ~1 Tibia 2 : femur ratio ~0.8 ~0.75 ~1 ~1 ~1.2 Propodus : tarsus ratio 0.6 – 0.8 0.8 – 0.95 ~0.75 ~0.8 ~0.9 Claw : propodus ratio 0.3 – 0.6 0.7 0.65 – 0.85 0.75 0.7 – 0.85

No chelifores. Abdomen straight or bent slightly down­ cis is absent. The ratio of 3rd to 5th palp article length is wards, length about 25 – 30% of trunk. higher and the tenth palp article is shorter. The ovigeral Palps ten-segmented. Third palp article longest, ap­ strigilis has more rows of short spines and the ovigeral proximately twice as long as fifth one. Last five palp arti­ claw is more slender and somewhat longer. The leg and cles short and of approximately similar length, somewhat oviger articles are relatively broader. The tibia 2 / femur longer than broad, 7th the longest, 10th the shortest. and propodus / tarsus length ratios are higher. Ovigera ten-segmented, fourth and sixth article long­ Compared to C. glacialis, the proboscis is longer and est. Ovigeral strigilis formed by last four articles, which the ocular tubercle is lower. The lateral processes are less bear one ectal row of longer spines and about two irregu­ widely separated. The ratio of third to fifth palp article lar endal rows of shorter spines. Ovigeral claw long and length is higher and the tenth palp article is shorter. The thin, curved, about the length of the last article. ovigeral strigilis lacks a distinct row of medium spines Legs: All leg articles tubular without special thicken­ and the ovigeral claw is much longer and more slender. ed regions. Three coxae about equally long, together The leg and oviger articles are relatively much broader. about 40% as long as femur. Tibia 1 about as long as The tibia 2 / femur and propodus / tarsus length ratios are femur, tibia 2 15 – 20% longer than either. Femur about higher, and the claw is proportionally longer. 7 times as long as broad. Last three leg articles together Compared to the otherwise very similar species C. lil­ as long as femur. Propodus about 90% as long as tarsus, liei Calman, 1915, the lateral processes are more widely claw about 75% as long as propodus, slightly curved. separated, the proboscis is relatively shorter, and the in­ Measurements of holotype (in mm): length of trunk tersegmental suture lines are not well visible. 9.86, proboscis 10.38, abdomen 2.82. Length of palp ar­ The differences between the species examined here ticles: 1: 0.46, 2: 0.47, 3: 5.81, 4: 0.75, 5: 2.83, 6: 1.19, are given in Table 7. 7:1.49, 8: 1.09, 9: 1.32, 10: 0.82. Length of oviger arti­ cles: 1: 0.88, 2: 0.98, 3: 0.81, 4: 8.03, 5: 3.32, 6: 6.87, Derivatio nominis. After Bouvet Island, the location 7: 2.81, 8: 2.44, 9: 2.16, 10: 1.77. Length of articles of where the holotype and most other specimens we exam­ third leg: coxa 1: 1.45, coxa 2: 2.08, coxa 3: 1.99, femur: ined were found. 13.98, tibia 1: 14.09, tibia 2: 16.56, tarsus: 5.32, propo­ dus: 4.8, claw: 3.26. Distribution. So far known from the Eastern Antarctic Peninsula, South Sandwich Islands, and Bouvet Island. Differential diagnosis. Compared to C. robusta, the Specimens from the Ross and Eastern Weddell Seas not animal is smaller and the body and extremities are more examined by us also seem to belong to C. bouvetensis setose. The wide dilation of the distal part of the probos­ based on COI sequence data.

101 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

Material. Holotype: specimen PF_E008, female. Type locality: near available, although only C. glacialis seems to occur on Bouvet Island, 54°20′25.08″S, 3°13′13.08″E, 648 m depth. Label: ’76 Shag Rocks / South Georgia and only C. bouvetensis is OT [Otter Trawl] 50 | 28.06.04 | Bouvet | PF_E008‘. This specimen has known from Bouvet Island. Within C. glacialis, three been deposited in the collection of the Zoologische Staatssammlung München (voucher ID XXXXX). Illustrations: Fig. 5I: ovigeral claw, different geographically separated clades are present, as 5L: ovigeral spine configuration, 7D: proboscis and trunk, 7E: ocular clade 2 is present on Northern Scotia Arc islands, clade 1 tubercle, 7F: palps. – Paratypes: 226-7.2, 59BT40, 59OT45, 66OT97, on the Weddell Sea continental shelf, and clade 6 in East 81OT58, PA_E003, PQ_E007, PQ_E008, PQ_E010, PQ_E011, PQ_ Antarctica. It can be assumed that these clades originated E012, PR_E003, PR_E004, PR_E005, PS_E010. Localities where the in different glacial refugia, the one of clade 2 located in specimens were found are listed in Table 2. Illustrations of specimen 226-7.2: Fig. 5B: ovigeral claw, 5E: ovigeral spine configuration, 6D: the Scotia Arc, and the other two in different places on proboscis and trunk, 6E: ocular tubercle, 6F: palps. the Antarctic shelf. Hence, it appears that the Antarctic Peninsula shelf was probably not recolonized from the Subantarctic islands in this case, although we lack data from the more southern Scotia Arc islands (South Ork­ neys and South Shetlands). The strong differentiation of 5. Discussion the one sequenced South Sandwich Islands individual also suggests that the South Sandwich Islands might have been a separate refugium from South Georgia and the 5.1. Molecular phylogenies Shag Rocks, while there appears to have been exchange between populations from the two latter locations based on the COI haplotype networks. In C. bouvetensis, the Our phylogenetic analysis of the COI gene supports the highest diversity was found in the Ross Sea, where in­ presence of several distinct species groups. One such dividuals from four different clades (5, 7, 8, 9), possibly monophyletic group includes C. robusta, C. drakei, C. stra­- interpretable as separate species, were found, in one case menti Fry & Hedgpeth, 1969, and several other species. even occurring sympatrically. Clade 9 is found only in Clearly, C. glacialis is not part of that group, but instead the Ross and Eastern Weddell Sea, possibly supporting appears to be basal to all other “longitarsal” Colossend­ the hypothesis of an ice-free connection between these eis species examined here (Fig. 1). Hence, it is concluded seas during previous interglacials (Barnes & Hillen- that C. robusta and C. glacialis are not particularly closely brand 2010). Clade 5, the most widespread clade, shows related within the genus Colossendeis. The fact that ITS a remarkable distribution. It can be separated into three sequence data show the same species-level groupings as subgroups, the first of which is found in Bouvet Island COI leads to the conclusion that there is no detectable and the Ross Sea. This distribution is puzzling, as Bouvet hybridization between C. glacialis and C. robusta in our is geographically very remote from the Ross Sea and the data, supporting the view that they are distinct species. clade is not found at intermediate locations such as the However, the lack of resolution between the different C. Eastern Weddell Sea. However, we lack data from large glacialis clades that were well-resolved with COI suggests parts of East Antarctica as well as some Subantarctic is­ that ITS data have less resolution, and therefore ITS may lands in the Indian and Pacific Ocean. The second group not be as useful on the intraspecific level as COI. is found in the South Sandwich Islands and Eastern Wed­ We find that the clade 10 C. robusta specimens from dell Sea, which can be explained by north-south dispersal the type locality (Kerguelen) are related to an Antarctic of unknown direction. The third group is found on both group (clade 3) with apparently circumpolar distribution. sides of the Eastern Weddell Sea, similar to C. glacialis Both groups can be regarded as belonging to the species clade 1. Obviously, more specimens from more sampling C. robusta. However, they are clearly distinct from an­ points are needed to test explicitly for competing popula­ other widespread species that has until now not been dis­ tion genetic and phylogeographic hypotheses. tinguished from C. robusta. This previously overlooked species is here described as a new species, Colossendeis bouvetensis sp.n. 5.3. Morphological data

The specimens identified here as C. bouvetensis differ 5.2. Population structure from C. robusta both according to our own measure­ ments and published descriptions of the holotype and Similar to the Southern Ocean pycnogonids C. megal­ other specimens of C. robusta. The absolute size of the onyx (Krabbe et al. 2010) and Nymphon australe Hodg­ specimens is much larger in C. robusta (trunk length son, 1902 (Mahon et al. 2008; Arango et al. 2011), C. 15.57 – 19.99 mm in C. robusta, 8.22 – 11.14 mm in C. robusta and C. glacialis show a strongly geographically bouvetensis). The proboscis in C. bouvetensis is cylindri­ partitioned population structure. This is expected in ben­ cal with only slight dilations in the middle and the end thic animals without a planktonic dispersal stage (Thatje (Fig. 6D, 7D), not bottle-shaped as in C. robusta (Fig. 2012). C. bouvetensis, C. robusta and C. glacialis co- 7M). The femur / tibia 2 ratio is 0.97 – 1 in C. robusta, occur in most of the regions from which sequences were while it is only 0.83 – 0.88 in C. bouvetensis. The rela­

102 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

tive width of the legs is greater in C. bouvetensis (length- bouvetensis in this respect. The specimens also have a width ratio of the 3rd femur 6.46 – 7.41 in C. bouvetensis, high and pointed ocular tubercle (Fig. 6H), while it is 8.63 – 9.46 in C. robusta). The ratio between the lengths lower in C. robusta from Kerguelen (Fig. 7K, agreeing of the 3rd and 5th palp articles is higher in C. bouvetensis with Stiboy-Risch 1993), similar to C. bouvetensis. Be­ (1.57 – 2.26) than in C. robusta (1.3 – 1.56). The 10th palp sides, the specimens are covered in fine spines, which article is relatively shorter in C. bouvetensis than in C. also disagrees with C. robusta. In general, the morpho­ robusta. These and other significant differences corrobo­ logical data agree with the molecular data that clade 3 rate the molecular data and support the classification of is most closely related to, but distinct from, C. robusta. C. bouvetensis as a separate species. However, as we lack detailed measurements from adult The specimens from Kerguelen agree well with pub­ specimens, we refrain from naming it as a new species, lished descriptions of C. robusta from that location, in­ although we suggest that many previous records of C. cluding the holotype (Hoek 1881; Möbius 1902; Calman robusta from Antarctica, such as those of Calman (1915) 1915; Stiboy-Risch 1993). The proboscis has a similar and Bouvier (1913) as well as those specimens measured shape, and both absolute measurements of body size and by Fry & Hedgpeth (1969) not referable to C. glacialis, the relative measurements of the leg and palp articles belong to this taxon. agree more with C. robusta from Kerguelen than with any Our specimens of C. glacialis appear to be consistent other species examined here. The leg is relatively shorter with previous descriptions of that species. The significant in our specimens (6.07 – 6.98 times trunk length, 7.52 in differences found between clades 1 and 2 raise the ques­ the holotype according to Calman 1915, 8 in the specimen tion whether these clades can be recognized as separate reported by Möbius 1902), but this may be explained if species. For those significantly different measurements these authors measured trunk length from the insertion of that were also recorded by Calman (1915) and Stiboy- the proboscis to that of the abdomen, instead of to the end Risch (1993) for the holotype and by Cano & Lopez- of the 4th lateral processes as we did. According to Stiboy- Gonzalez (2007), whose specimen also came from the Risch (1993), the ratio is 6.62 in the holotype. However, Ross Sea, their values agree with those of clade 2 and according to the measurements of Stiboy-Risch (1993), clade 6 specimens. As the holotype is from East Antarc­ the palp in the holotype is much shorter (1.26 times trunk tica, where so far only clade 6 is known, the latter might length excluding the basal article) than in our specimens retain the name C. glacialis. In this case, as discussed (1.47 – 1.57). In all our specimens, as well as those re­ below, the name C. gracilipes Bouvier, 1913 would prob­ ported by Calman (1915) and Bouvier (1913) the propo­ ably be available for clade 1, while a new species name dus / claw ratio is 1.18 – 1.47, while in the holotype it is 2 would be required for clade 2. However, as the morpho­ (Calman 1915), more similar to C. glacialis. The drawing logical and molecular differences may not be enough for given by Möbius (1902) agrees with our specimens in this recognizing a new species and can also be explained as respect. Palp article 9 is significantly shorter than articles intraspecific geographic variation, we do not erect a new 8 or 10 in the holotype, which is not the case in our speci­ species here. mens, in which article 10 is always slightly shorter than The specimens here identified asC. drakei differ from 8 and 9. Despite these differences, our specimens from the original description (Calman 1915) in their smaller Kerguelen are closer to the holotype of C. robusta than relative lengths of proboscis, abdomen, and palps. How­ any other specimens we examined, and are here identified ever, this could be due to a different way of measuring as C. robusta. Child (1995) noted that C. robusta speci­ the trunk, as already discussed for C. robusta. As in one mens from Heard Island (part of the Kerguelen Plateau) of our specimens (257-2.5) the legs are relatively much are larger and have a distally more inflated proboscis than shorter than in the others, the leg lengths of the types fall those from Antarctica, which is consistent with them be­ into the range of variation of our specimens. Identifica­ longing to C. robusta s.str. This appears to be an endemic tion of our specimens as C. drakei was based primarily species from the Kerguelen Plateau. on the proportions of the distal palp articles, which agree The two specimens 257-2.1 and 226-7.3 group in a well with the original description but differ from all other separate clade (clade 3) together with other sequences similar Antarctic species in that article 7 is not the longest from specimens not examined by us. Specimen 257-2.1 of the distal 5 articles. is unique in having chelifores, and the lack of genital pores indicates that it is an immature specimen despite its large size. Specimen 226-7.3, which could not be com­ 5.4. Comparison with previous work pletely measured, is even larger, agreeing with the Ant­ arctic C. robusta specimens reported by Calman (1915) and Bouvier (1913). In some measurements, such as the According to Fry & Hedgpeth (1969) and Stiboy-Risch femur / tibia 2 ratio, clade 3 resembles C. robusta, includ­ (1993), C. glacialis shows much more spiny legs than C. ing the type, more than any other Antarctic specimens robusta. We can confirm this, but the degree of spinous­ we measured. The legs are also relatively longer than in ness appears to be variable within both C. glacialis and all specimens of C. bouvetensis, but fall into the range of C. bouvetensis so that it cannot be used to distinguish variation of C. robusta. However, the proboscis is shorter between those species. The observation of Stiboy-Risch than in C. robusta, which agrees more closely with C. (1993) that the tibia 2 is the longest leg article in C. ro­

103 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

Fig. 6. Proboscis and trunk (left column), ocular tubercle (middle column) and palps (right column) of examined Colossendeis specimens. A – C: 29OT27-2 (C. glacialis clade 1). D – F: 226-7.2 (C. bouvetensis, probable juvenile, Antarctic Peninsula). G – I: 257-2.1 (C. sp. cf. ro­ busta, probable juvenile, Antarctic Peninsula). J – L: AGT42/164 (C. glacialis clade 2).

busta also agrees with our data, but the tibia 1 is not al­ in our C. glacialis clade 1 (Fig. 6C) and 6 (Fig. 7L) speci­ ways the longest in our specimens of C. glacialis, as the mens, they are elongated in the clade 2 specimen (Fig. femur is about equally long. Also agreeing with Stiboy- 6L). Risch (1993), the three last palp articles in C. glacialis Cano & Lopez-Gonzalez (2007) described the ovi­ are about equally long, although we cannot confirm that geral spine configuration of C. glacialis, which agrees the second to last palp article is significantly shorter in well with our observations of that species. They also C. robusta; instead, the last article is the shortest in our agree with the original description by Hodgson (1907a). specimens. No specimens among our material could be The shape of the long spines in their Ross Sea specimen assigned to the species C. acuta described by Stiboy- agrees more with our clade 1 than clade 2 specimen. The Risch (1993). configuration with three distinct types of spines also oc­ According to the key of Hodgson (1927) the last three curs in C. drakei (this study), C. megalonyx, C. tenera palp articles are elongated in C. robusta but rounded in (Dietz et al. 2013), C. australis and C. scotti (Cano & C. glacialis. While they are indeed not longer than wide Lopez-Gonzalez 2007). In C. bouvetensis (this study),

104 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

Fig. 7. Proboscis and trunk (left column), ocular tubercle (middle column) and palps (right column) of examined Colossendeis specimens. M: Pro­ boscis. N,P: Last oviger article and claw in side view. O: Last oviger article in endal view. A – C: 233-3.1.1 (C. drakei). D – F: PF_E008 (holo­type of C. bouvetensis, Bouvet Island). G – I,M,O – P: IU-2007-5063 (C. robusta, Kerguelen). J – L,N: IU-2103-15812 (C. glacialis clade 6).

C. angusta (Dietz et al. 2013), C. wilsoni and C. lilliei the holotype of C. robusta as illustrated by Hoek (1881) (Cano & Lopez-Gonzalez 2007) there is no distinction are extremely worn, they show no evidence of a separate between medium and short spines. Although the spines in row of medium spines. This also agrees with the con­

105 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

dition in our clade 3 specimen, which is closely related long to C. glacialis, and that C. bouvetensis was probably to C. robusta. According to our phylogenetic analysis, not represented in their samples. a configuration similar to that in C. glacialis appears to be ancestral for this group of Colossendeis species and has been reduced multiple times. Interestingly, C. wil­ 5.5. Comparison with other similar soni and C. lilliei, both species rather similar in overall species appearance to C. bouvetensis, show a similar spine con­ figuration and shape of the ovigeral claw, suggesting that they are related more closely to it than C. glacialis. Bouvier (1913) described the species Colossendeis gra­ In the key of Pushkin (1993) C. glacialis is differ­ cilipes based on two specimens from the South Shetlands entiated from C. robusta and other species by the claw and Antarctic Peninsula. The species was synonymized of the leg being less than half as long as the propodus. with C. glacialis by Calman (1915), as the only differ­ Although this ratio is indeed consistently lower in C. gla­ ences to that species are a narrowing of the proboscis at cialis than in C. robusta and other species in our data, the base and a longer last palp article. The former charac­ there are some specimens of C. glacialis in which it is ter is not apparent as a consistent difference in our speci­ slightly larger than 0.5. In the holotype of C. robusta, it mens, but in the latter C. gracilipes agrees well with clade is ca. 0.52 (Stiboy-Risch 1993), lower than in any of our 1 specimens, which is consistent with its geographical specimens of that species. Our results also disagree with origin. The holotype of C. glacialis from the Ross Sea the claim (Pushkin 1988) that the 6th palp article (5th in is more similar to clades 2 and 6 in that character. In the Pushkin’s terminology) is the longest of the distal five other measurements given by Bouvier (1913), C. gracili­ articles in C. robusta. The 7th article is the longest in all of pes also agrees with C. glacialis and the synonymization our specimens except for those assigned to C. drakei, and appears to be correct. this is also the case in the holotype of C. robusta (Stiboy- hodgson (1907b) described the species Colossendeis Risch 1993). patagonica from the Patagonian Atlantic coast based on C. robusta and C. glacialis have been synonymized an incomplete specimen. It is sometimes synonymized by Fry & Hedgpeth (1969), which has been accepted with C. glacialis (e.g. Bamber & El Nagar 2014) which by Child (1995). The synonymization was based on however is not otherwise known from Patagonia (Push- the claim that the recorded differences show too much kin 1993; Munilla & Soler Membrives 2009). The rela­ variability to be useful distinguishing characters and tive lengths of palp articles and the presence of only are partially explained by ontogeny. However, the data three rows of ovigeral spines differ from C. glacialis. As presented by Fry & Hedgpeth (1969) actually support the specimen was incomplete and no illustrations were the presence of two distinct species in their dataset. The published, it is difficult to identify with another species. femur – tibia 2 ratio shows a bimodal distribution con­ However, synonymy with C. glacialis appears to be un­ sistent with our data, with a smaller ratio in C. robusta. likely. Specimens assigned to C. bouvetensis have an even low­ Child (1995) synonymized C. lilliei Calman, 1915 er femur / tibia 2 ratio than any of those given by Fry & with C. robusta. The species was originally distinguished Hedgpeth (1969) and elsewhere in the literature (Calman from C. robusta by Calman by the lateral processes being 1915; Cano & Lopez-Gonzalez 2007). Proboscis length placed much closer together, the presence of visible in­ compared to leg length is significantly larger inC. robus­ tersegmental suture lines, a proportionally longer probos­ ta. According to Fry & Hedgpeth (1969), the data is con­ cis, and a smaller femur / tibia 2 ratio. In the latter charac­ sistent with a single-species explanation if one assumes ter, it agrees with C. bouvetensis, but it differs in the other a sudden increase in proboscis length at a certain point mentioned characters. None of the specimens examined during ontogeny, as specimens with a relatively shorter by us could be determined as C. lilliei. As the differences proboscis (here interpreted as C. glacialis) are typically to C. robusta appear to be consistent, we regard the syn­ smaller. However, their graph actually shows two larger onymy as unlikely. However, C. lilliei appears to be re­ specimens with a proboscis / leg ratio more typical of C. lated to C. bouvetensis based on the ovigeral characters, glacialis. Our data in general agree with those of Fry & which are virtually identical in the two species. Also, the Hedgpeth (1969) in finding that the ratio is typically larg­ proportions of the palp articles, especially the length of er for C. robusta as well as C. bouvetensis, and we also the 3rd relative to the 5th article and the shortness of the identify a smaller specimen (226-7.2) as C. bouvetensis 10th article, are similar to C. bouvetensis but differ from showing that the ratio does not change significantly with the other examined species. The leg proportions are also increasing size. However, the range of variation for those most similar to C. bouvetensis. In conclusion, while C. characters is so great in C. glacialis that it would be dif­ lilliei appears to be distinct from C. bouvetensis, these ficult to differentiate those species based on this charac­ species are probably closer to each other than to C. ro­ ter alone. Notably, Fry and Hedgpeth’s data data include busta or C. glacialis. holo- and paratypes of both species, which agree in their The Patagonian species Colossendeis smirnovi Push­ measurements with the specimens examined by us. From kin, 1988 was synonymized with C. drakei by Child their data, we conclude that most of the C. robusta speci­ (1995), but the proportions of the palp segments do not mens examined by Fry & Hedgpeth (1969) actually be­ agree with it and appear to be more similar to C. gla­

106 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

cialis. C. smirnovi was differentiated from C. glacialis species concept is used. Within C. glacialis, C. robusta by Pushkin (1988) by a lower width / length ratio of the and C. bouvetensis, we find regional intraspecific vari­ femur, different relative lengths of the last five palp arti­ ation suggesting survival in multiple refugia during the cles and the claw being longer than half of the propodus. Pleistocene glaciations. However, his measurements of the femur for both species fall into the range of variation for our C. glacialis clade 2 specimens. The claw is longer than half of the propodus in some of our C. glacialis specimens as well as in one 6. Acknowledgments specimen described by Calman (1915) and the relative lengths of the palp articles are variable in C. glacialis. Despite the similarity, however, we regard synonymy of We would like to thank Franz Krapp and Claudia P. Arango for C. smirnovi with C. glacialis as doubtful as the latter spe­ helpful discussions on the taxonomy of the Colossendeidae. We cies has not to our knowledge been recorded from Pa­ thank Laure Corbari (Muséum National del’Histoire Naturelle) for tagonia, although some specimens may have been misi­ providing specimens for analysis. This project was supported by dentified as C. drakei. DFG grant LE 2323/2 to FL. Some other Southern Ocean Colossendeis species with proportions similar to C. robusta and C. glacialis have been described. C. wilsoni Calman, 1915 is similar especially to C. lilliei, but differs from all other Colos­ 7. References sendeis species by having only 9 palp articles. Notoende­ is germanica Hodgson, 1915 (later placed into the genus Colossendeis by Hodgson 1927) appears to be similar to Allcock A., Strugnell J. 2012. Southern Ocean diversity: new C. wilsoni but was described as having a fully segmented paradigms from molecular ecology. – Trends in Ecology and body, which would be unique within the Colossendei­ Evolution 27: 520 – 528. nae. C. stramenti Fry & Hedgpeth, 1969 from the South Arabi J., Cruaud C., Couloux A., Hassanin A. 2010. Studying American shelf differs from all other species in having sources of incongruence in arthropod molecular phylogenies: a highly reduced ocular tubercle without eyes. C. gras­ sea spiders (Pycnogonida) as a case study. – Comptes Rendus sus Pushkin, 1993 resembles C. lilliei in having the lat­ Biologie 333: 438 – 453. eral processes closely placed together but differs in the Arango C. 2002. Morphological phylogenetics of the sea spiders form of the ocular tubercle. C. avidus Pushkin, 1970 has (Arthropoda: Pycnogonida). – Organisms, Diversity and Evo­ leg proportions similar to C. glacialis but a significantly lution 2: 107 – 125. longer and differently shaped proboscis. It appears to be Arango C., Brenneis G. 2013. New species of Australian Pseudo­ very similar to C. acuta, and we regard these species as pallene (Pycnogonida: Callipallenidae) based on live coloura­ probably synonymous. C. bruuni Fage, 1956 from the tion, morphology and DNA. – Zootaxa 3616: 401 – 436. Kermadec Trench and C. curtirostris Stock, 1963 from Arango C., Soler Membrives A, Miller K.J. 2011. Genetic dif­ South Africa have a short proboscis similar to C. glacia­ ferentiation in the circum-Antarctic sea spider Nymphon aus­ lis, but a claw longer than the propodus, which differs trale (Pycnogonida; Nymphonidae). – Deep Sea Research II from all similar Antarctic species and especially from C. 58: 212 – 219. glacialis. As none of those species appears to be present Arango C., Wheeler W. 2006. Phylogeny of the sea spiders (Ar­ in our material, we cannot assess their validity in the pre­ thropoda, Pycnogonida) based on direct optimization of six sent study. loci and morphology. – Cladistics 23: 255 – 293. Arnaud F., Bamber R. 1987. The biology of Pycnogonida. – Ad­ vances in Marine Biology 24: 1 – 96. Bamber R. 2007. A holistic re-interpretation of the phylogeny 5.6. Conclusions of the Pycnogonida Latreille, 1810 (Arthropoda). – Zootaxa 1668: 295 – 312. The results of this study strongly support the hypothesis Bamber R., El Nagar A. 2014. Pycnobase: World Pycnogonida that C. robusta and C. glacialis are separate species and Database. – http://www.marinespecies.org/pycnobase. not conspecific as suggested by several authors; they Barnes D., Hillenbrand C. 2010. Faunal evidence for a late qua­ are not even closely related within Colossendeis. The ternary trans-Antarctic seaway. – Global Change Biology 16: integrative taxonomic approach identified a new spe­ 3297 – 3303. cies, C. bouvetensis, previously not distinguished from Bouvier E. 1913. Pycnogonides du «Pourquoi-Pas»? Deuxième ex­ C. robusta, which has a broad geographic distribution. pédition antarctique française (1908 – 1910) commandée par le There might be further overlooked species in the C. ro­ Dr. Jean Charcot: 1 – 169. busta group, yet, more data are needed to show this. We Braby C., Pearse V., Bain B., Vrijenhoek R. 2009. Pycnogonid- also show that, while there is an endemic grouping in the cni­darian trophic interactions in the deep Monterey Submarine Kerguelen region that appears to include the holotype of Can­yon. – Invertebrate Biology 128: 359 – 363. C. robusta, there is a related widespread grouping in the Brandao S., Sauer J., Schon I. 2010. Circumantarctic distribution Antarctic that may be included in that species if a broad in Southern Ocean benthos? A genetic test using the genus

107 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

Macroscapha (Crustacea, Ostracoda) as a model. – Molecular Frézal L., Leblois R. 2008. Four years of DNA barcoding: current Phylogenetics and Evolution 55: 1055 – 1069. advances and prospects. – Infection, Genetics and Evolution Brandt A., De Broyer C., De Mesel I., Ellingsen K., Gooday A., 8: 727 – 736. Hilbig B., et al. 2007. The biodiversity of the deep Southern Fritz G., Conn J., Cockburn A., Seawright J. 1994. Sequence ana­ Ocean benthos. – Proceedings of the Royal Society of London lysis of the ribosomal DNA internal transcribed spacer 2 from B 362: 39 – 66. populations of Anopheles nuneztovari (Diptera: Culicidae). – Calman W. 1915. Pycnogonida. – British Antarctic (“Terra Nova”) Molecular Biology and Evolution 11: 406 – 416. Expedition, 1910 3: 1. Fry W., Hedgpeth J. 1969. The fauna of the Ross Sea. Part 7. Pycno­ Cano E., Lopez-Gonzalez P. 2007. Colossendeis species (Pycnogo­ gonida, 1: Colossendeidae, Pycnogonidae, Endeidae, Ammo­ nida: Colossendeidae) collected during the Italica XIX cruise theidae. – New Zealand Department of Science and Industrial to Victoria Land (Antarctica), with remarks on some taxo­­- Research Bulletin 198: 1 – 139. no­ mic­­ characters of the ovigers. – Scientia Marina 71: 661 – Giribet G., Edgecombe G. 2012. Reevaluating the arthropod tree of 681. life. – Annual Review of Entomology 57: 167 – 186. Chen C., Chen C.-P., Fan T.-Y., Yu J.-K., Hsieh H.-L. 2002. Nu­ Griffiths H. 2010. Antarctic marine biodiversity – what do we cleotide sequences of ribosomal internal transcribed spacers know about the distribution of life in the Southern Ocean? – and their utility in distinguishing closely related Perinereis PLoS ONE 5: e11683. polychaets (Annelida; Polychaeta; Nereididae). – Marine Bio­ Griffiths H.J., Arango C.P., Munilla T., Mcinnes S.J. 2011. Bio­ technology 4: 17 – 29. diversity and biogeography of Southern Ocean pycnogonids. Cheng H.-L., Xia D.-Q., Wu T.-T., Meng X.-P., Ji H.-J., Dong – Ecography 34: 616 – 627. Z.-G. 2006. Study on sequences of ribosomal DNA internal Gutt J., Sirenko B., Smirnov I., Arntz W. 2004. How many mac­ transcribed spacers of clams belonging to the Veneridae fam­ rozoobenthic species might inhabit the Antarctic shelf? – Ant­ ily (Mollusca: Bivalvia). – Acta Genetica Sinica 33(8): 702 – arctic Science 16: 11 – 16. 710. Hebert P., Cywinska A., Ball S., deWaard J. 2003. Biological Child C. 1995. Antarctic and Subantartic Pycnogonida: The Fami­ identifications through DNA barcodes. – Proceedings of the lies Nymphonidae, Colossendeidae, Rhynchothoraxidae, Pyc­ Royal Society of London B 270: 313 – 321. nogonidae, Endeididae, and Callipallenidae. – Antarctic Re­ Hebert P., Stoeckle M., Zemlak T., Francis C. 2004. Identifica­ search Series 69: 1 – 165. tion of birds through DNA barcodes. – PLoS Biology 2: 1657 – Child C. 1998. Pycnogonida from Prydz Bay, East Antarctica. – 1663. Records of the South Australian Museum 31: 1 – 19. Held C. 2000. Phylogeny and biogeography of serolid isopods Clarke A., Johnston N. 2003. Antarctic marine benthic diversity. – (Cru­stacea, Isopoda, Serolidae) and the use of ribosomal ex­ Oceanography and Marine Biology: an Annual Review 41: pan­sion segments in molecular systematics. – Molecular Phy­ 47 – 114. lo­genetics and Evolution 15: 165 – 178. Clement M., Posada D., Crandall K. 2000. TCS: a computer pro­ Held C., Wägele J.-W. 2005. Cryptic speciation in the giant Antarc­ gram to estimate gene genealogies. – Molecular Ecology 9: tic isopod Glyptonotus antarcticus (Isopoda: Valvifera: Chaeti­­ 1657 – 1659. liidae). – Scientia Marina 69 (Suppl. 2): 175 – 181. Convey P., Stevens M., Hodgson D., Smellie J., Hillenbrand C., Hemery L., Eléaume M., Roussel V., Améziane N., Gallut C., Barnes D., et al. 2009. Exploring biological constraints on the Steinke D., et al. 2012. Comprehensive sampling reveals cir­ glacial history of Antarctica. – Quaternary Science Reviews cumpolarity and sympatry in seven mitochondrial lineages of 28: 3035 – 3048. the Southern Ocean crinoid species Promachocrinus kerguelen­ Darriba D., Taboada G., Doallo R., Posada D. 2012. jModelTest sis (Echinodermata). – Molecular Ecology 21: 2502 – 2518. 2: more models, new heuristics and parallel computing. – Na­ Hodgson T. 1902. Crustacea. Pp. 228 – 269 in: Report on the Col­ ture Methods 9: 772. lections of Natural History, made in the Antarctic Regions dur­ Dietz L., Krapp F., Hendrickx M.E., Arango C.P., Krabbe K., ing the Voyage of the “Southern Cross”. – W. Cloves and Sons, Spaak J.M., Leese F. 2013. Evidence from morphological and London. genetic data confirms that Colossendeis tenera Hilton, 1943 Hodgson T. 1907a. Pycnogonida. – Natural History Collections of the (Arthropoda: Pycnogonida), does not belong to the Colossend­ “Discovery” National Antarctic Expedition, Zoology 3: 1 – 72. eis megalonyx Hoek, 1881 complex. – Organisms Diversity & Hodgson T. 1907b. Pycnogoniden. – Ergebnisse der Hamburger Evolution 13: 151 – 162. Ma­galhaensischen Sammelreisen 2: 1 – 20. Dietz L., Mayer C., Arango C., Leese F. 2011. The mitochondrial Hodgson T. 1927. Die Pycnogoniden der Deutschen Südpolar-Ex­ genome of Colossendeis megalonyx supports a basal position pedition 1901/03. – Deutsche Südpolar-Expedition 19: 305 – of Colossendeidae within the Pycnogonida. – Molecular Phy­ 358. logenetics and Evolution 58: 553 – 558. Hoek P.P.C. 1881. Report on the Pycnogonida, dredged by HMS Dunlop J., Arango C. 2005. Pycnogonid affinities: a review. – Challenger during the Years 1873 – 76. – Report of the Sci­ Journal of Zoological Systematics and Evolutionary Research entific Results of the Voyage of HMS Challenger, Zoology 3: 43: 8 – 21. 1 – 167. Folmer O., Black M., Hoeh W., Lutz R., Vrijenhoek R. 1994. Hughes J., Hillyer M., Bunn S. 2003. Small-scale patterns of ge­ DNA primers for amplification of mitochondrial cytochrome c netic variation in the mayflyBungona narilla (Ephemeroptera: oxidase subunit I from diverse metazoan invertebrates. – Mo­ Baetidae) in rainforest streams, south-east Queensland. – lecular Marine Biology and Biotechnology 3: 294 – 299. Freshwater Biology 48: 709 – 717.

108 Proof _ 01 Proof _ 01

ARTHROPOD SYSTEMATICS & PHYLOGENY — 73 (1) 2015

Janosik A., Halanych K. 2010. Unrecognized Antarctic biodiver­ Regier J., Shultz J., Zwick A., Hussey A., Ball B., Wetzer R., sity: a case study of the genus Odontaster (Odontasteridae; As­ et al. 2010. Arthropod relationships revealed by phylogenomic teroidea). – Integrative and Comparative Biology 50: 981 – 992. analysis of nuclear protein-coding sequences. – Nature 463: Katoh K., Standley D.M. 2010. MAFFT Multiple Sequence Align­ 1079 – 1084. ment Software Version 7: Improvements in Performance and Ronquist F., Teslenko M., van der Mark P., Ayres D., Darling Usability. – Molecular Biology and Evolution 30: 772 – 780. A., Höhna S., et al. 2012. MrBayes 3.2: efficient Bayesian Kattner G. 1998. The expedition ANTARKTIS XIV/2 of RV “Po­ phylo­ genetic­ inference and model choice across a large model larstern” in 1996/97: 1 – 87. space. – Systematic Biology 61: 539 – 542. King P. 1973. Pycnogonids. – Hutchinson, London. Schlötterer C., Hauser M., von Haeseler A., Tautz D. 1994. Com­ Krabbe K., Leese F., Mayer C., Tollrian R., Held C. 2010. Cryp­ parative evolutionary analysis of rDNA ITS regions in Dro­ tic mitochondrial lineages in the widespread pycnogonid sophila. – Molecular Biology and Evolution 11: 513 – 522. Colossendeis megalonyx Hoek, 1881 from Antarctic and Sub­ Schüller M. 2011. Evidence for a role of bathymetry and emer­ antarctic waters. – Polar Biology 33: 281 – 292. gence in speciation in the genus Glycera (Glyceridae, Poly­ Lecointre G., Ameziane N., Boisselier M.-C., Bonillo C., Busson chaeta) from the deep Eastern Weddell Sea. – Polar Biology F., Causse R., et al. 2013. Is the species flock concept opera­ 34: 549 – 564. tional? The Antarctic shelf case. – PLoS ONE 8: e68787. Stamatakis A. 2006. RAxML-VI-HPC: maximum likelihood-bas­ Leese F., Brand P., Rozenberg A., Mayer C., Agrawal S., Dam- ed phylogenetic analyses with thousands of taxa and mixed bach J., et al. 2012. Exploring Pandora’s box: potential and models. – Bioinformatics 22: 2688 – 2690. pitfalls of low coverage genome surveys for evolutionary bio­ Stiboy-Risch C. 1993. Funde antarktischer und subantarktischer logy. – PLoS ONE 7: e49202. Pantopoden einschließlich Colossendeis acuta sp. n. – sowie Leese F., Held C. 2008. Identification and characterization of mi­ ein Beitrag zur Artbestimmung von Colossendeis glacialis crosatellites from the Antarctic isopod Ceratoserolis trilobi­ Hodgson, 1907 und Colossendeis robusta Hoek, 1881. – Mit­ toides: nuclear evidence for cryptic species. – Conservation teilungen aus dem Hamburgischen Zoologischen Museum und Genetics 9(5): 1369 – 1372. Institut 90: 251 – 264. Mahon A., Arango C., Halanych K. 2008. Genetic diversity of Swofford D. 2002. PAUP*. Phylogenetic Analysis Using Parsi­ Nymphon (Arthropoda: Pycnogonida: Nymphonidae) along the mony (*and Other Methods). Version 4. – Sinauer Associates, Antarctic Peninsula with a focus on Nymphon australe Hodg­ Sunderland, Massachusetts: ii + 142 pp. son 1902. – Marine Biology 155: 315 – 323. Talavera G., Castresana J. 2007. Improvement of phylogenies af­ Möbius K. 1902. Die Pantopoden der deutschen Tiefsee-Expedition ter removing divergent and ambiguously aligned blocks from 1898 – 1899. – Wissenschaftliche Ergebnisse der Deutschen protein sequence alignments. – Systematic Biology 56: 564 – Tiefsee-Expedition auf “Valdivia” 1898 – 1899 3: 175 – 196. 577. Munilla León T. 1999. Evolución y filogenia de los picnogóni­ Thatje S. 2012. Effects of capability for dispersal on the evolution dos. – Boletín de la Sociedad Entomológica Aragonesa 26: of diversity in Antarctic benthos. – Integrative and Compara­ 273 – 279. tive Biology 52: 470 – 482. Munilla León T. 2001. A new species of Ammothea (Pycnogonida) Thornhill D., Mahon A., Norenburg J., Halanych K. 2008. Open- and other pycnogonids from Livingston Island and surrounding ocean barriers to dispersal: a test case with the Antarctic Polar waters (South Shetland Islands, Antarctica). – Antarctic Sci­ Front and the ribbon worm Parborlasia corrugatus (Nemertea: ence 13: 144 – 149. Lineidae). – Molecular Ecology 17: 5104 – 5117. Munilla T., Soler Membrives A. 2009. Check-list of the pycnogo­ Weis A., Melzer R. 2012. How did sea spiders recolonize the Chil­ nids from Antarctic and sub-Antarctic waters: zoogeographic ean fjords after glaciation? DNA barcoding of Pycnogonida, implications. – Antarctic Science 21: 99. with remarks on phylogeography of Achelia assimilis (Has­ Nakamura K., Kano Y., Suzuki N., Namatame T., Kosaku A. 2007. well, 1885). – Systematics and Biodiversity 10: 361 – 374. 18S rRNA phylogeny of sea spiders with emphasis on the posi­ Wilson N., Hunter R., Lockhart S., Halanych K. 2007. Multiple tion of Rhynchothoracidae. – Marine Biology 153: 213 – 223. lineages and absence of panmixia in the “circumpolar” crinoid Odorico D., Miller D. 1997. Variation in the ribosomal internal Promachocrinus kerguelensis from the Atlantic sector of Ant­ transcribed spacers and 5.8S rDNA among five species of arctica. – Marine Biology 152: 895 – 904. Acropora (Cnidaria; Scleractinia): patterns of variation con­ Wörheide G. 1998. The reef cave dwelling ultraconservative coral­ sistent with reticulate evolution. – Molecular Biology and Evo­ line demosponge Astrosclera willeyana Lister 1900 from the lution 14: 465 – 473. Indo-Pacific. – Facies38 : 1 – 88. Peat H., Clarke A., Convey P. 2006. Diversity and biogeography of the Antarctic flora. – Journal of Biogeography34 : 132 – 146. Pushkin A.F. 1988. K revisii vida Colossendeis robusta (Pantopo­ da) iz juzhnogo okeana. – Zoologichesky Zhurnal (Moscow) 67: 953. Pushkin A.F. 1993. The Pycnogonida fauna of the South Ocean (Biological­ results of the Soviet Antarctic Expeditions) – Fauna mnogokolenchatyx (Pycnogonida) juzhnogo okeana (Re­zultaty biologcheskix issledovaniy sovetskix antarkticesix Eks­pediziy). – Samperi, Messina.

109 Proof _ 01 Proof _ 01

Dietz et al.: Southern Ocean sea spider phylogeny and diversity

Electronic Supplement Files at http://www.senckenberg.de/arthropod-systematics (“Contents”)

File 1: dietz&al-colossendeis-asp2015-electronicsupplement-1.xls — Table ES1. Morphometric measurements of examined Colos­ sen­deis specimens in mm. P = Palp, O = Oviger, L1 – 4 = Walking leg 1 – 4, C1 – 3 = Coxae 1 – 3, F = Femur, T1 – 2 = Tibiae 1 – 2, TA = Tarsus, PR = Propodus, CL = Claw. L = length, H = height, W = width. Lateral 12 = distance between first and second lateral processes. a = anterior, p = posterior.

File 2: dietz&al-colossendeis-asp2015-electronicsupplement-2.xls — Table ES2. Morphometric measurements of examined Colos­ sendeis specimens, expressed as proportion of trunk length. P = Palp, O = Oviger, L1 – 4 = Walking leg 1 – 4, C1 – 3 = Coxae 1 – 3, F = Femur, T1 – 2 = Tibiae 1 – 2, TA = Tarsus, PR = Propodus, CL = Claw. L = length, H = height, W = width. Lateral 12 = dis­ tance between first and second lateral processes. a = anterior, p = posterior.

File 3: dietz&al-colossendeis-asp2015-electronicsupplement-3.xls — Table ES3. Measurements that are significantly different (p < 0.05) between different clades. P = Palp, O = Oviger, L1 – 4 = Walking leg 1 – 4, C1 – 3 = Coxae 1 – 3, F = Femur, T1 – 2 = Tibiae 1 – 2, TA = Tarsus, PR = Propodus, CL = Claw. L = length, H = height, W = width. Lateral 12 = distance between first and second lateral processes. a = anterior, p = posterior. < and > signs refer to clade listed second compared to first.

110 Electronic supplement 1: absolute measurements in mm Trunk L Trunk H Trunk W Trunk W23 Proboscis W Proboscis L Abdomen L Abdomen W P1L P2L P3L P4L P5L P6L P7L P8L P9L P10L 257-2.4 8.040 1.570 4.110 1.630 1.255 5.800 1.760 0.400 0.270 0.300 2.770 0.470 1.930 0.760 1.030 0.750 0.620 0.790 300-1.4 8.530 1.760 4.450 1.680 1.495 6.650 1.630 0.400 0.270 0.290 2.860 0.550 2.360 0.820 1.170 0.750 0.780 0.880 AGT42/164 8.670 1.820 4.760 1.660 1.435 6.630 1.840 0.420 0.340 0.360 2.970 0.430 2.130 0.770 1.160 0.830 0.690 0.870 211-5.1 9.290 1.940 4.970 1.750 1.485 7.260 2.180 0.640 0.230 0.280 3.950 0.490 2.810 1.000 1.430 0.900 0.810 0.920 211-6.3.2 8.580 1.790 5.240 1.860 1.605 7.550 1.810 0.500 0.260 0.320 2.900 0.590 2.500 0.920 1.190 0.770 0.830 0.870 29OT27-1 9.750 1.950 5.160 1.580 1.615 8.530 1.960 0.380 0.340 0.350 4.450 0.580 2.870 0.980 1.640 1.070 0.900 0.830 29OT27-2 8.060 1.580 4.550 1.310 1.400 6.690 1.770 0.310 0.240 0.250 2.970 0.410 2.530 0.910 1.390 0.850 0.930 0.880 30BT14-2 10.530 2.140 5.780 1.880 1.900 9.020 2.310 0.670 0.300 0.340 4.510 0.600 3.330 1.120 1.620 1.130 1.010 1.140 45BT24 9.150 1.810 5.170 1.650 1.495 8.660 2.030 0.580 0.270 0.280 3.980 0.540 2.940 0.920 1.490 0.920 0.860 1.080 PA_E006 8.910 1.950 5.040 1.740 1.735 7.440 2.120 0.440 0.310 0.280 3.750 0.710 2.670 1.030 1.360 0.980 1.080 0.820 PB_E005 9.750 1.800 5.260 1.720 1.615 8.450 2.110 0.490 0.330 0.310 4.140 1.050 2.800 0.830 1.560 0.920 1.070 0.770 PR_E006 10.230 1.980 5.410 1.680 1.700 9.270 2.090 0.400 0.350 0.360 4.290 0.670 3.000 0.890 1.400 0.900 0.870 0.830 PR_E010 10.400 2.060 5.260 1.790 1.860 9.020 2.490 0.450 0.570 0.400 4.550 0.790 2.820 0.990 1.640 1.010 1.160 0.920 PS_E011 9.850 1.950 4.930 1.750 1.660 9.180 2.230 0.520 0.320 0.340 4.390 0.690 2.850 0.960 1.530 1.060 1.010 0.950 257-2.1 12.180 2.590 6.910 2.410 2.070 10.400 2.620 0.630 0.580 0.630 4.430 1.220 3.850 1.290 1.770 1.260 1.290 1.240 257-2.5 6.640 1.180 3.370 1.080 1.070 5.400 1.020 0.510 0.270 0.260 2.450 0.590 1.760 0.640 0.520 0.620 0.870 0.900 233-3.1.2 8.720 1.450 5.060 1.290 1.200 7.370 1.270 0.550 0.360 0.410 4.090 0.830 2.490 0.860 0.780 0.670 0.950 1.040 233-3.1.1 8.530 1.520 4.520 1.190 1.150 7.290 0.000 0.000 0.370 0.410 3.950 0.780 2.730 0.840 0.740 0.780 1.030 1.180 226-7.2 7.190 2.290 4.760 2.070 1.835 7.320 1.780 0.640 0.330 0.380 3.690 0.610 2.340 0.710 1.080 0.670 0.710 0.390 59BT40 9.490 2.450 5.590 2.390 1.915 9.140 2.410 0.640 0.380 0.360 4.780 0.840 2.840 1.080 1.370 1.030 1.090 0.710 59OT45 10.260 2.820 6.050 2.550 2.195 10.620 2.680 0.780 0.430 0.550 5.580 1.230 3.060 1.080 1.590 1.010 1.200 0.350 66OT97 9.780 2.680 5.470 2.440 2.085 9.890 2.690 0.610 0.380 0.410 5.260 0.760 2.780 0.890 1.450 0.980 1.010 0.610 81OT58 9.270 2.250 5.410 2.340 1.980 9.270 2.470 0.660 0.480 0.380 5.050 0.660 2.550 0.910 0.940 0.820 0.870 0.270 PA_E003 8.220 2.480 5.120 2.360 1.935 8.320 2.530 0.590 0.340 0.380 4.280 0.640 2.720 1.130 1.520 1.040 0.960 0.440 PF_E008 9.860 2.890 6.440 2.710 2.110 10.380 2.820 0.770 0.460 0.470 5.810 0.750 2.830 1.190 1.490 1.090 1.320 0.820 PQ_E007 11.140 3.180 6.630 2.810 2.305 10.950 2.690 0.770 0.430 0.450 6.240 1.070 2.750 1.170 1.540 0.940 1.110 0.740 PQ_E008 9.330 2.610 5.480 2.370 1.960 9.270 2.370 0.710 0.310 0.280 4.970 0.940 2.340 0.940 1.270 0.790 0.960 0.620 PQ_E010 10.570 2.980 6.550 2.820 2.245 10.480 2.920 0.790 0.350 0.390 5.960 1.090 3.180 1.100 1.580 0.890 1.020 0.690 PQ_E011 9.900 2.820 5.720 2.420 2.085 9.420 2.570 0.710 0.370 0.440 5.580 1.040 2.680 1.030 1.370 0.830 1.010 0.640 PQ_E012 9.940 2.780 6.010 2.530 2.120 10.040 2.640 0.740 0.340 0.370 5.160 1.040 2.660 1.060 1.560 0.960 1.110 0.670 PR_E003 9.120 2.370 5.370 2.320 1.965 9.080 2.490 0.710 0.360 0.380 4.390 0.600 2.620 0.960 1.210 0.900 0.830 0.560 PR_E004 9.980 2.710 5.870 2.560 1.905 9.850 2.730 0.740 0.470 0.540 5.750 0.980 3.020 0.930 1.320 0.970 1.030 0.710 PR_E005 10.250 2.850 6.410 2.650 2.165 10.230 2.660 0.880 0.530 0.480 5.590 1.020 2.860 1.000 1.420 0.860 1.160 0.710 PS_E010 9.940 2.670 5.870 2.480 2.120 9.740 2.540 0.780 0.310 0.300 5.170 0.820 2.480 0.870 1.270 0.820 1.060 0.640 IU-2007-4842 16.480 3.860 9.190 3.130 3.415 14.740 4.730 0.870 0.760 0.650 7.190 1.390 5.170 2.120 3.120 1.950 2.070 1.740 IU-2007-5063 16.880 3.930 9.390 3.190 3.630 16.310 3.960 1.010 0.740 0.560 6.960 1.310 5.020 2.010 2.940 2.220 2.080 1.840 IU-2007-5039 15.710 3.580 9.210 2.890 3.710 16.540 3.580 0.790 0.670 0.580 6.890 1.170 5.310 1.830 2.970 1.990 2.020 1.720 IU-2007-5069 15.570 3.570 8.940 3.100 3.460 15.330 3.290 0.750 0.680 0.570 7.000 1.190 5.020 1.910 2.980 1.890 2.060 1.690 IU-2007-5043 16.830 4.050 9.860 3.260 3.700 16.760 3.690 0.870 0.670 0.560 7.200 1.250 4.990 1.980 3.080 1.960 1.980 1.750 IU-2007-4798 16.310 3.590 8.990 3.150 3.360 15.550 3.660 0.860 0.830 0.760 7.030 1.430 5.120 2.230 3.040 2.020 2.040 1.970 IU-2007-4870 19.990 4.440 10.770 3.690 4.015 18.590 4.280 0.990 0.870 0.780 9.330 1.620 5.970 2.350 3.740 2.720 2.410 2.240 IU-2007-4800 14.060 3.170 8.090 2.640 2.805 13.430 2.990 0.690 0.610 0.520 6.710 0.960 4.360 1.710 2.590 1.580 1.760 1.440 IU-2007-4795 16.060 3.870 9.520 3.120 3.515 15.160 3.190 0.850 0.650 0.580 7.250 1.240 5.470 1.920 3.030 2.020 1.980 1.760 IU-2007-5044 15.710 4.100 8.870 2.980 3.385 14.860 3.530 0.840 0.660 0.580 7.160 1.230 5.090 1.850 2.870 1.900 1.940 1.690 IU-2013-15805 9.450 1.980 4.720 1.710 1.430 7.380 2.160 0.540 0.250 0.280 3.610 0.540 2.780 1.080 1.520 0.900 1.060 0.970 IU-2013-15808 8.950 1.850 4.180 1.630 1.395 7.060 2.040 0.480 0.290 0.320 3.390 0.570 2.470 0.930 1.420 1.020 1.110 0.830 IU-2013-15812 9.200 1.930 4.730 1.690 1.430 7.230 2.020 0.490 0.280 0.260 3.280 0.500 2.680 1.090 1.380 0.940 1.020 0.890 O1L O1W O2L O2W O3L O3W O4L O4W O5L O5W O6L O6W O7L O7W O8L O8W O9L O9W 010L O10W 0.480 0.620 0.600 0.680 0.700 0.690 5.100 0.530 1.790 0.680 5.680 0.520 1.610 0.480 1.250 0.630 1.200 0.370 1.040 0.260 0.580 0.770 0.690 0.790 0.950 0.660 7.050 0.640 1.890 0.740 6.980 0.670 1.720 0.610 1.610 0.640 1.540 0.530 0.840 0.380 0.570 0.780 0.630 0.760 0.770 0.720 7.000 0.570 1.890 0.770 7.030 0.600 1.970 0.430 1.410 0.700 1.330 0.390 1.200 0.320 0.670 0.630 0.680 0.680 1.070 0.790 7.590 0.730 1.950 0.850 7.780 0.770 1.880 0.590 1.480 0.620 1.460 0.550 1.270 0.400 0.620 0.730 0.530 0.730 0.690 0.680 7.400 0.710 1.950 0.830 8.210 0.700 1.800 0.640 1.380 0.670 1.330 0.520 0.900 0.240 0.760 0.890 0.690 0.710 0.860 0.850 9.400 0.700 2.560 0.900 9.550 0.660 1.890 0.610 1.460 0.650 1.640 0.570 1.140 0.340 0.590 0.650 0.550 0.570 0.770 0.680 7.130 0.580 1.840 0.690 7.060 0.650 1.540 0.560 1.280 0.570 1.290 0.470 1.120 0.300 0.800 0.960 0.730 0.910 1.080 0.850 8.500 0.880 2.170 0.980 9.010 0.760 1.760 0.620 1.560 0.650 1.680 0.590 1.390 0.410 0.770 0.740 0.670 0.760 0.930 0.800 7.930 0.720 2.080 0.830 9.890 0.830 1.700 0.610 1.310 0.550 1.440 0.490 1.230 0.350 0.610 0.750 0.670 0.650 0.750 0.850 8.180 0.730 1.630 0.970 8.160 0.890 1.840 0.660 1.540 0.640 1.810 0.660 0.900 0.230 0.740 0.760 0.720 0.680 0.860 0.800 8.740 0.660 2.270 0.760 8.850 0.710 1.720 0.650 1.510 0.640 1.260 0.580 1.210 0.290 0.710 0.750 0.750 0.760 0.930 0.860 8.470 0.770 2.480 0.870 9.530 0.760 1.780 0.670 1.490 0.590 1.580 0.550 1.430 0.330 0.720 0.860 0.690 0.780 0.830 0.870 8.720 0.740 2.490 0.940 10.190 0.800 2.030 0.710 1.860 0.640 1.780 0.610 1.290 0.380 0.790 0.840 0.750 0.700 0.930 0.940 8.920 0.720 2.370 0.750 9.350 0.660 2.120 0.550 1.750 0.560 1.700 0.450 1.250 0.260 1.060 1.190 1.240 0.860 1.010 1.200 8.750 0.970 3.680 0.750 9.380 0.940 2.270 0.960 2.310 0.820 2.380 0.700 1.770 0.590 0.560 0.530 0.500 0.450 0.660 0.460 4.850 0.560 2.540 0.550 5.620 0.410 1.330 0.450 0.890 0.430 0.770 0.360 0.820 0.310 0.690 0.670 0.660 0.620 0.760 0.550 9.210 0.720 4.000 0.680 9.960 0.680 1.740 0.560 1.250 0.610 1.150 0.450 1.010 0.340 0.620 0.670 0.550 0.470 0.710 0.490 8.770 0.720 3.530 0.670 9.030 0.640 1.440 0.450 1.330 0.600 1.270 0.550 1.120 0.300 0.620 0.750 0.670 0.720 0.640 0.670 5.320 0.930 2.270 1.020 4.310 0.960 1.630 0.890 1.820 0.740 1.660 0.720 1.380 0.470 0.830 0.970 0.690 0.770 0.870 0.920 6.830 1.010 2.730 1.040 6.270 0.880 2.090 0.820 1.970 0.860 1.770 0.730 1.540 0.390 0.890 1.050 0.860 0.810 1.030 0.960 8.160 1.010 3.120 1.070 7.550 1.110 1.960 1.040 2.040 1.070 1.970 0.940 1.730 0.590 0.750 0.980 0.660 0.800 0.910 1.130 7.550 0.970 2.620 1.190 7.620 0.930 2.000 0.940 2.030 0.860 1.920 0.770 1.630 0.500 0.830 1.030 0.780 0.820 0.930 1.010 6.840 0.920 2.470 1.120 5.720 0.960 1.820 1.030 2.050 0.990 1.650 0.900 1.580 0.510 0.730 1.010 0.840 0.900 0.770 0.910 6.580 1.050 2.580 1.070 6.640 1.030 2.000 0.900 2.060 0.870 2.010 0.730 1.890 0.490 0.880 0.940 0.980 0.920 0.810 0.860 8.030 1.040 3.320 1.340 6.870 1.060 2.810 1.050 2.440 0.930 2.160 0.760 1.770 0.450 0.890 1.220 0.750 0.890 1.120 1.210 7.690 1.340 3.190 1.320 7.380 1.070 2.410 1.030 2.340 1.010 2.180 0.840 1.910 0.620 0.780 1.010 0.670 0.810 0.950 1.090 6.560 1.120 2.870 1.160 6.560 0.940 2.050 0.840 2.010 0.820 1.830 0.700 1.470 0.470 0.860 1.090 0.830 0.970 1.130 1.230 8.420 1.220 3.330 1.270 7.260 1.020 2.430 0.990 2.370 0.930 2.030 0.760 1.600 0.520 0.820 1.050 0.800 0.950 1.050 1.140 6.830 1.130 3.040 1.130 6.990 0.970 2.130 0.930 2.080 0.850 1.930 0.730 1.490 0.490 0.770 0.990 0.860 0.930 1.090 1.100 7.140 1.120 3.040 1.160 6.930 1.010 2.170 0.900 2.110 0.950 1.870 0.780 1.510 0.490 0.720 0.980 0.800 0.860 0.960 1.000 6.780 0.850 2.640 1.120 6.340 0.910 1.890 0.860 1.860 0.780 1.710 0.630 1.440 0.400 0.740 0.890 0.830 0.930 0.980 1.040 7.970 1.150 2.700 1.240 7.370 0.970 2.050 0.960 2.430 0.840 2.060 0.740 1.720 0.440 0.810 1.060 0.830 0.950 0.980 1.110 7.600 1.080 2.960 1.110 7.110 0.990 1.920 0.980 2.180 0.930 1.760 0.790 1.610 0.520 0.830 1.070 0.760 0.920 0.920 1.070 7.000 1.230 3.010 1.130 6.930 1.030 2.350 0.910 1.990 0.940 1.770 0.770 1.510 0.480 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.630 1.890 1.620 1.920 1.540 1.700 13.560 1.980 5.550 1.620 15.520 1.710 3.760 1.610 3.470 1.410 3.160 1.330 3.180 1.030 1.510 1.620 1.310 1.540 1.510 1.620 13.050 1.870 5.510 1.680 14.660 1.620 3.840 1.370 3.530 1.320 3.270 1.240 2.960 0.830 1.410 1.690 1.450 1.550 1.500 1.730 12.750 1.870 4.890 1.580 13.990 1.650 3.770 1.230 3.610 1.360 3.270 1.150 3.210 0.920 1.520 1.800 1.490 1.750 1.470 1.620 13.160 1.970 5.640 1.780 14.620 1.710 3.750 1.460 3.460 1.470 3.340 1.320 3.040 0.940 1.340 1.830 1.370 1.630 1.610 1.680 12.790 1.950 5.260 1.750 14.160 1.650 3.790 1.560 3.240 1.490 3.490 1.340 2.840 0.960 1.650 1.870 1.770 2.110 1.660 1.830 15.500 1.980 5.720 1.890 16.850 1.980 4.050 1.740 4.090 1.770 3.740 1.450 3.380 1.050 1.240 1.550 1.190 1.490 1.210 1.570 11.000 1.510 4.120 1.340 11.360 1.460 3.170 1.260 2.890 1.190 2.860 1.110 2.600 0.760 1.460 1.890 1.390 1.480 1.650 1.700 13.280 1.900 5.170 1.860 14.980 1.810 3.910 1.500 3.790 1.540 3.690 1.360 3.100 0.840 1.140 1.630 1.300 1.680 1.430 1.670 12.660 1.780 5.050 1.650 13.730 1.650 3.720 1.520 3.120 1.470 3.320 1.290 2.860 0.910 0.620 0.710 0.670 0.690 0.790 0.760 7.620 0.760 2.090 0.860 8.550 0.750 1.690 0.570 1.460 0.620 1.530 0.570 1.250 0.370 0.620 0.700 0.700 0.670 0.750 0.690 6.270 0.690 2.000 0.730 6.490 0.490 1.630 0.440 1.440 0.410 1.590 0.450 1.140 0.290 0.570 0.690 0.590 0.630 0.770 0.810 7.200 0.750 2.040 0.720 7.860 0.710 1.640 0.680 1.470 0.650 1.510 0.530 1.230 0.320 Basal element W Lateral 12 Lateral 23 Lateral 34 W Lateral 1 W Lateral 2 W Lateral 3 W Lateral 4 Ocular tubercle W Ocular tubercle H Front height a Eye size a Front height p Eye size p L1 C1 L L1 C1 W L1 C2 L L1 C2 W 0.605 0.930 1.245 0.940 1.055 1.165 1.105 1.100 0.830 1.140 0.480 0.380 0.530 0.290 1.200 1.230 1.480 1.290 0.670 1.035 1.220 1.150 1.050 1.130 1.090 1.060 0.880 1.180 0.530 0.350 0.000 0.000 1.290 1.460 1.720 1.440 0.655 1.295 1.475 1.265 1.040 1.110 1.215 1.015 0.960 1.250 0.460 0.400 0.550 0.280 1.250 1.370 1.650 1.420 0.800 1.040 1.435 1.445 1.255 1.230 1.340 1.245 1.040 1.410 0.490 0.345 0.680 0.270 1.440 1.420 1.720 1.340 0.780 1.105 1.515 1.530 1.230 1.195 1.235 1.230 1.080 1.920 0.480 0.380 0.580 0.405 1.430 1.390 1.320 1.300 0.630 1.070 1.500 1.305 1.160 1.245 1.350 1.145 1.040 1.310 0.510 0.445 0.580 0.330 1.450 1.510 1.700 1.390 0.590 0.920 1.260 1.050 1.070 1.190 1.195 1.040 0.990 1.160 0.680 0.400 0.640 0.340 1.250 1.160 1.480 1.150 0.960 1.170 1.665 1.620 1.290 1.525 1.535 1.460 1.190 1.520 0.580 0.435 0.680 0.310 1.560 1.500 2.120 1.650 0.800 1.195 1.485 1.345 1.140 1.365 1.415 1.270 1.020 1.330 0.520 0.355 0.580 0.265 1.410 1.430 1.950 1.600 0.745 1.175 1.295 1.105 1.370 1.480 1.470 1.350 0.980 1.360 0.520 0.510 0.630 0.425 1.330 1.370 1.800 1.530 0.790 1.260 1.565 1.230 1.260 1.380 1.415 1.215 1.140 1.390 0.640 0.370 0.620 0.270 1.590 1.470 1.870 1.370 0.715 1.240 1.725 1.320 1.340 1.395 1.455 1.305 1.330 1.470 0.740 0.480 0.720 0.450 1.420 1.320 1.820 1.430 0.840 1.255 1.830 1.855 1.390 1.520 1.480 1.360 1.180 1.310 0.460 0.425 0.640 0.460 1.680 1.710 1.880 1.530 0.815 1.255 1.670 1.435 1.305 1.360 1.455 1.230 1.110 1.500 0.550 0.385 0.600 0.405 1.430 1.440 1.900 1.490 1.065 1.095 1.450 1.240 2.105 2.260 2.210 2.015 1.230 1.650 0.850 0.635 0.910 0.545 2.250 2.360 2.780 2.020 0.690 0.585 0.670 0.785 0.935 0.890 0.820 0.930 0.770 0.840 0.340 0.270 0.460 0.235 0.920 1.010 1.170 0.860 0.725 0.725 0.815 1.095 1.170 1.130 1.125 1.190 0.830 1.080 0.510 0.330 0.520 0.340 1.500 1.280 1.630 1.140 0.605 0.745 0.830 1.075 1.135 1.045 1.185 1.105 0.940 1.190 0.370 0.330 0.400 0.360 1.330 1.260 1.610 1.080 0.705 0.510 0.815 0.715 1.255 1.375 1.340 1.210 1.340 1.130 0.330 0.490 0.370 0.385 1.280 1.700 1.550 1.630 0.890 0.985 1.320 1.015 1.545 1.720 1.660 1.520 1.780 1.330 0.430 0.505 0.490 0.495 1.300 1.830 1.860 1.610 0.930 0.760 1.155 1.045 1.500 1.755 1.780 1.545 1.670 1.560 0.510 0.330 0.480 0.610 1.440 1.990 2.080 1.870 0.785 0.925 1.255 1.020 1.610 1.785 1.655 1.555 1.820 1.280 0.460 0.415 0.510 0.395 1.370 1.950 1.870 1.630 0.860 0.630 0.960 0.755 1.500 1.620 1.695 1.520 1.670 1.460 0.360 0.420 0.250 0.405 1.360 1.750 1.890 1.730 0.840 0.605 0.910 0.825 1.380 1.510 1.470 1.360 1.440 1.170 0.360 0.445 0.490 0.460 1.280 1.780 1.850 1.740 0.930 0.670 1.180 1.030 1.510 1.810 1.785 1.410 1.670 1.320 0.380 0.475 0.530 0.405 1.280 1.880 2.180 1.890 0.950 0.960 1.310 1.180 1.780 1.830 1.935 1.715 1.820 1.580 0.450 0.480 0.380 0.380 1.450 1.880 2.120 1.860 0.785 0.820 1.065 0.925 1.480 1.615 1.670 1.480 1.560 1.380 0.390 0.405 0.320 0.335 1.290 1.660 1.990 1.670 0.930 0.930 1.280 1.030 1.575 1.720 1.805 1.625 1.760 1.360 0.460 0.495 0.600 0.385 1.420 1.890 2.080 1.890 0.890 0.895 1.185 0.895 1.460 1.665 1.635 1.450 1.640 1.260 0.400 0.445 0.360 0.390 1.390 1.780 2.000 1.660 0.790 0.865 1.170 0.950 1.590 1.785 1.725 1.625 1.500 1.400 0.460 0.445 0.520 0.345 1.370 1.810 1.950 1.620 0.930 0.690 1.130 0.860 1.455 1.615 1.590 1.425 1.600 1.280 0.500 0.390 0.260 0.460 1.390 1.710 1.800 1.700 0.805 0.735 1.260 1.035 1.665 1.725 1.695 1.615 1.810 1.330 0.350 0.485 0.540 0.480 1.470 1.940 2.010 1.780 0.965 0.860 1.355 1.080 1.665 1.765 1.845 1.610 1.870 1.560 0.410 0.380 0.380 0.305 1.450 1.970 2.020 1.780 0.760 0.835 1.215 0.935 1.525 1.635 1.745 1.510 1.750 1.630 0.410 0.395 0.370 0.340 1.390 2.010 2.080 1.740 1.640 1.820 2.595 2.200 2.145 2.335 2.335 2.495 1.880 1.550 0.640 0.430 0.590 0.405 2.605 2.660 3.240 2.370 1.635 2.160 1.945 2.230 2.450 2.550 2.570 2.490 1.860 1.630 0.740 0.505 0.550 0.475 3.085 3.000 3.315 2.660 1.495 1.805 1.975 2.050 2.245 2.465 2.530 2.300 1.700 1.580 0.710 0.500 0.720 0.435 2.795 2.845 3.050 2.555 1.465 1.765 1.885 2.165 2.305 2.575 2.490 2.325 1.630 1.520 0.620 0.440 0.510 0.380 2.675 2.815 3.070 2.545 1.600 1.955 2.035 2.335 2.420 2.605 2.645 2.445 1.730 1.570 0.650 0.500 0.710 0.460 2.990 2.890 3.300 2.700 1.580 1.815 1.995 2.405 2.345 2.515 2.575 2.445 1.630 1.610 0.830 0.455 0.570 0.405 2.765 2.865 3.045 2.500 1.865 2.885 3.105 3.015 2.650 2.720 2.780 2.630 1.710 2.070 0.740 0.540 0.840 0.475 3.135 2.880 3.805 2.625 1.295 1.555 1.750 1.945 1.985 2.210 2.215 2.110 1.330 1.320 0.680 0.490 0.840 0.340 2.420 2.410 2.780 1.870 1.625 1.945 1.990 2.180 2.165 2.510 2.520 2.305 1.690 1.490 0.630 0.460 0.640 0.405 2.740 2.810 3.150 2.630 1.565 1.850 1.705 2.185 2.235 2.505 2.510 2.365 1.680 1.570 0.590 0.405 0.630 0.380 2.745 2.670 3.175 2.320 0.855 1.015 1.510 1.550 1.240 1.280 1.370 1.280 1.130 1.400 0.390 0.395 0.770 0.310 1.465 1.385 1.720 1.390 0.700 0.970 1.520 1.445 1.225 1.245 1.260 1.240 0.950 1.230 0.330 0.365 0.690 0.295 1.315 1.335 1.485 1.310 0.755 0.990 1.580 1.550 1.225 1.345 1.360 1.290 1.100 1.410 0.350 0.390 0.690 0.290 1.380 1.450 1.570 1.475 L1 C3 L L1 C3 W L1 F L L1 F W L1 T1 L L1 T1 W L1 T2 L L1 T2 W L1 TA L L1 TA W L1 PR L L1 PR W L1 CL L L1 CL W L2 C1 L L2 C1 W L2 C2 L L2 C2 W L2 C3 L L2 C3 W 1.380 1.250 10.510 1.380 10.870 1.150 8.740 1.020 4.910 0.680 3.430 0.510 1.240 0.200 1.190 1.280 1.500 1.230 1.260 1.280 1.480 1.390 14.660 1.520 15.650 1.320 12.830 1.170 6.110 0.710 3.910 0.470 1.710 0.280 1.290 1.370 1.830 1.410 1.650 1.440 1.500 1.390 14.550 1.440 15.320 1.260 11.740 1.090 5.680 0.670 3.960 0.490 2.400 0.280 1.420 1.360 1.620 1.540 1.540 1.400 1.390 1.320 14.300 1.200 14.920 1.230 11.850 1.040 5.350 0.720 3.900 0.580 1.470 0.270 1.360 1.530 1.810 1.470 1.550 1.420 1.340 1.360 14.510 1.180 15.060 1.150 11.610 1.020 5.580 0.700 3.120 0.560 1.390 0.230 1.590 1.550 1.370 1.380 1.400 1.430 1.530 1.570 18.310 1.390 16.260 1.230 12.650 1.030 3.890 0.660 3.200 0.500 1.810 0.200 1.400 1.510 1.840 1.550 1.760 1.600 1.250 1.150 13.440 1.010 13.380 1.040 10.030 0.800 4.810 0.640 3.540 0.540 1.430 0.200 1.300 1.280 1.610 1.280 1.410 1.190 1.760 1.610 16.540 1.610 16.490 1.340 12.880 1.200 5.400 0.750 3.890 0.640 2.020 0.240 1.630 1.720 1.990 1.760 1.660 1.620 1.590 1.470 18.390 1.500 18.730 1.220 13.730 1.070 5.590 0.660 3.600 0.560 1.600 0.220 1.570 1.540 1.810 1.560 1.300 1.410 1.810 1.560 14.240 1.610 14.780 1.270 11.890 1.100 5.490 0.730 4.010 0.620 1.430 0.260 1.430 1.580 1.780 1.530 1.730 1.560 1.710 1.470 15.910 1.350 15.960 1.100 12.170 0.900 5.660 0.660 4.140 0.580 1.920 0.200 1.570 1.540 1.870 1.510 1.640 1.470 1.600 1.640 17.010 1.380 16.610 1.250 11.150 1.070 4.490 0.680 3.690 0.620 1.670 0.300 1.640 1.560 1.930 1.620 1.580 1.480 1.570 1.610 18.890 1.750 18.980 1.330 14.900 1.160 5.770 0.700 4.090 0.560 1.960 0.270 1.740 1.590 2.040 1.570 1.800 1.660 1.680 1.460 17.250 1.510 17.350 1.240 13.100 1.090 5.810 0.670 4.120 0.540 1.810 0.190 1.450 1.510 2.010 1.560 1.770 1.470 2.430 2.100 18.690 2.110 16.730 1.900 18.850 1.710 8.690 1.470 6.710 0.930 4.730 0.540 2.140 2.210 2.670 2.020 2.490 2.180 1.140 0.860 10.360 0.800 8.790 0.660 7.820 0.540 3.760 0.410 3.510 0.370 2.330 0.260 0.890 1.000 1.280 0.920 1.220 0.870 1.540 1.120 18.520 1.100 15.330 0.770 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.400 1.170 1.540 1.170 1.540 1.190 1.480 1.130 17.690 1.030 14.850 0.920 12.990 0.740 6.090 0.520 5.050 0.390 3.320 0.230 1.430 1.310 1.650 1.130 1.450 1.160 1.610 1.650 8.060 1.980 8.360 1.720 9.280 1.520 3.660 1.020 3.580 0.740 2.480 0.380 1.160 1.800 1.540 1.710 1.650 1.690 1.740 1.650 12.180 1.760 12.620 1.610 13.860 1.420 5.000 0.930 4.380 0.750 3.300 0.330 1.390 1.880 1.940 1.610 1.880 1.660 1.990 1.780 14.160 2.040 14.840 1.870 16.640 1.560 5.770 1.150 4.880 0.830 3.480 0.380 1.660 1.950 2.170 1.920 2.030 1.890 1.740 1.720 12.920 1.890 13.670 1.860 14.910 1.520 5.060 1.010 4.500 0.800 3.490 0.420 1.350 2.010 1.970 1.670 1.980 1.830 1.680 1.720 12.160 1.780 12.880 1.800 14.430 1.620 4.720 1.030 4.100 0.740 2.990 0.330 1.500 1.870 1.970 1.760 1.850 1.760 1.780 1.710 11.740 1.910 11.710 1.720 13.040 1.530 4.920 1.040 4.300 0.760 3.060 0.360 1.330 1.670 1.910 1.850 1.910 1.770 1.990 1.920 13.770 2.140 14.200 1.960 16.300 1.670 5.590 1.220 4.770 0.840 3.480 0.420 1.580 1.980 2.140 2.000 1.980 1.880 2.120 2.040 15.760 1.970 16.400 2.010 18.600 1.730 5.970 1.200 5.110 0.900 3.520 0.480 1.710 2.210 2.420 2.110 2.530 1.990 1.930 1.650 12.930 1.750 13.230 1.590 15.090 1.430 5.000 0.980 4.440 0.780 3.510 0.440 1.410 1.830 1.810 1.600 1.800 1.640 2.070 2.000 14.550 2.130 15.130 1.880 17.140 1.680 5.660 1.180 5.030 0.840 3.620 0.370 1.590 2.020 2.190 1.980 2.110 1.980 1.880 1.640 13.010 1.870 13.330 1.800 15.180 1.530 5.180 1.050 4.590 0.780 3.490 0.410 1.450 1.950 2.010 1.810 1.970 1.720 1.950 1.650 13.180 1.850 13.850 1.750 15.260 1.480 5.150 1.000 4.350 0.750 3.380 0.350 1.430 1.930 2.070 1.830 2.030 1.690 1.770 1.640 12.030 1.820 12.430 1.700 14.420 1.440 4.660 1.000 4.190 0.790 3.070 0.400 1.490 1.820 1.940 1.700 1.900 1.620 1.910 1.790 13.980 1.810 14.430 1.790 15.990 1.570 5.570 1.040 4.600 0.760 3.700 0.350 1.540 1.830 2.100 1.880 1.980 1.830 1.880 1.790 13.830 2.110 14.220 1.990 16.650 1.750 5.390 1.130 4.820 0.810 3.460 0.490 1.520 1.900 2.100 1.880 1.960 1.870 2.020 1.720 13.890 1.910 13.640 1.640 16.100 1.440 5.310 1.020 4.670 0.750 3.330 0.420 1.490 1.920 2.230 1.870 2.080 1.720 3.190 2.475 22.460 2.450 20.650 2.295 22.610 2.010 9.620 1.410 7.210 0.970 3.980 0.650 2.485 2.590 3.530 2.480 3.390 2.500 3.210 2.435 25.590 2.500 23.835 2.450 26.175 2.225 11.065 1.465 8.120 1.080 5.830 0.630 2.985 2.825 3.290 2.670 3.250 2.620 3.015 2.435 24.900 2.655 23.550 2.440 26.145 2.200 10.995 1.380 8.175 1.000 5.650 0.635 2.900 2.955 3.215 2.655 3.000 2.545 3.180 2.470 24.195 2.725 22.430 2.500 24.855 2.080 10.270 1.245 7.715 0.915 5.135 0.515 2.765 2.920 3.435 2.700 3.110 2.640 3.255 2.535 25.085 2.725 23.290 2.655 25.485 2.265 11.115 1.525 8.415 1.065 5.280 0.620 3.025 3.135 3.505 2.890 3.450 2.815 3.005 2.450 24.000 2.570 22.400 2.460 24.880 2.220 10.620 1.410 7.640 1.060 5.470 0.550 4.140 2.805 3.065 2.500 2.570 2.430 3.695 2.695 29.870 2.950 29.000 2.800 32.060 2.450 12.775 1.605 9.055 1.225 7.140 0.690 3.195 3.035 4.085 2.890 3.710 2.745 2.420 1.920 18.360 1.980 15.090 1.730 15.790 1.530 6.000 1.180 5.160 0.790 4.650 0.500 2.370 2.515 2.750 2.240 2.805 2.200 3.080 2.580 24.230 2.610 22.210 2.610 25.900 2.460 10.950 1.600 7.800 1.090 5.470 0.620 2.760 2.840 3.210 2.700 3.395 2.600 3.075 2.440 23.180 2.425 21.710 2.405 23.675 2.100 10.190 1.365 7.765 1.115 5.945 0.635 2.515 2.620 3.230 2.395 3.485 2.555 1.485 1.325 17.155 1.545 17.370 1.265 13.880 1.120 7.085 0.700 4.550 0.635 1.475 0.190 1.575 1.520 1.770 1.545 1.705 1.455 1.340 1.295 12.865 1.470 13.450 1.305 10.955 1.065 5.640 0.675 3.825 0.520 1.230 0.185 1.370 1.355 1.505 1.390 1.455 1.275 1.430 1.415 15.730 1.485 16.505 1.335 13.455 1.115 5.620 0.640 4.470 0.570 1.370 0.210 1.415 1.515 1.640 1.540 1.590 1.470 L2 F L L2 F W L2 T1 L L2 T1 W L2 T2 L L2 T2 W L2 TA L L2 TA W L2 PR L L2 PR W L2 CL L L2 CL W L3 C1 L L3 C1 W L3 C2 L L3 C2 W L3 C3 L L3 C3 W L3 F L L3 F W 11.220 1.350 11.550 1.200 9.540 1.010 4.680 0.680 3.600 0.580 1.430 0.240 1.190 1.290 1.430 1.200 1.350 1.340 11.140 1.320 15.780 1.540 16.340 1.410 13.290 1.080 6.260 0.710 4.260 0.500 1.800 0.290 1.240 1.450 1.680 1.410 1.520 1.420 15.120 1.700 16.020 1.520 15.830 1.310 12.270 1.120 5.570 0.680 3.910 0.530 2.790 0.190 1.260 1.340 1.570 1.380 1.540 1.320 15.560 1.500 15.050 1.110 16.360 1.170 12.880 1.090 6.040 0.770 3.740 0.610 1.930 0.490 1.290 1.650 1.910 1.440 1.620 1.440 15.300 1.310 15.860 1.250 15.930 1.130 12.500 1.040 5.490 0.680 3.630 0.510 1.250 0.290 1.510 1.440 1.710 1.340 1.620 1.280 15.700 1.340 19.750 1.410 19.440 1.300 14.190 1.180 5.790 0.600 4.000 0.510 2.090 0.240 1.380 1.380 1.640 1.260 1.460 1.340 15.530 1.280 13.550 1.090 13.150 1.100 9.750 1.030 4.280 0.500 3.200 0.480 1.830 0.200 1.330 1.290 1.700 1.210 1.280 1.180 14.440 1.070 18.190 1.510 18.280 1.390 13.240 1.110 5.930 0.750 4.170 0.640 1.990 0.310 1.400 1.620 2.150 1.610 1.710 1.510 17.870 1.560 20.640 1.480 19.670 1.290 14.390 1.050 5.820 0.600 4.010 0.540 1.510 0.200 1.560 1.470 1.900 1.400 1.650 1.420 19.880 1.410 15.720 1.630 16.410 1.210 12.870 1.190 5.740 0.780 4.070 0.720 1.740 0.270 1.520 1.560 1.720 1.350 1.400 1.440 15.430 1.620 17.120 1.430 17.150 1.110 12.780 0.860 5.660 0.640 4.000 0.450 1.720 0.180 1.320 1.400 1.950 1.350 1.660 1.380 16.720 1.490 18.440 1.690 17.740 1.240 12.920 1.140 5.480 0.650 3.740 0.580 1.920 0.280 1.480 1.550 1.850 1.440 1.520 1.520 18.270 1.560 20.270 1.680 20.580 1.390 15.660 1.210 6.410 0.770 4.170 0.520 1.750 0.260 1.600 1.610 1.790 1.460 1.540 1.670 19.880 1.700 18.370 1.460 18.220 1.260 10.230 1.090 5.120 0.640 3.140 0.510 1.500 0.180 1.520 1.580 2.030 1.620 1.730 1.520 18.090 1.620 20.510 2.140 18.680 1.890 20.540 1.780 8.930 1.410 6.940 0.910 5.030 0.560 2.200 2.340 2.750 2.310 2.550 2.140 19.110 2.120 10.880 0.890 9.270 0.650 8.260 0.630 3.730 0.410 3.490 0.350 2.670 0.180 0.910 1.030 1.280 0.910 1.240 0.900 10.480 0.820 19.920 1.080 16.540 0.780 14.930 0.700 0.000 0.000 0.000 0.000 0.000 0.000 1.580 1.230 1.660 1.150 1.520 1.130 19.030 1.070 19.160 1.030 16.190 0.820 13.820 0.710 6.500 0.510 5.190 0.360 3.720 0.250 1.470 1.320 1.730 1.150 1.500 1.130 18.540 1.090 8.820 1.940 9.080 1.690 10.260 1.510 3.770 1.030 3.680 0.750 2.330 0.370 1.180 1.830 1.600 1.740 1.580 1.720 8.410 1.940 12.680 1.760 12.830 1.630 14.790 1.480 4.870 0.900 4.380 0.750 3.580 0.340 1.380 1.900 1.980 1.620 1.880 1.670 12.300 1.800 14.690 2.170 15.070 1.860 17.470 1.660 5.770 1.240 5.150 0.810 3.600 0.420 1.540 1.980 2.140 1.930 2.060 1.900 14.300 2.120 13.470 1.890 14.080 1.840 15.940 1.570 5.010 1.020 4.510 0.830 3.540 0.370 1.510 1.910 2.010 1.760 1.850 1.760 13.030 2.010 12.870 1.890 13.080 1.890 15.230 1.620 4.940 1.000 4.230 0.740 3.050 0.290 1.470 1.870 1.900 1.710 1.810 1.800 12.320 1.850 12.360 1.920 12.250 1.810 14.020 1.510 4.620 1.120 4.070 0.770 2.970 0.370 1.160 1.850 1.810 1.710 1.790 1.720 11.740 1.710 14.430 2.160 14.720 2.060 17.180 1.800 5.280 1.220 4.910 0.860 3.760 0.370 1.450 2.120 2.080 1.960 1.990 1.940 13.980 2.170 16.640 2.280 16.670 2.070 19.820 1.810 6.040 1.220 5.280 0.890 3.390 0.540 1.600 2.170 2.190 1.940 2.150 2.010 16.030 2.170 13.170 1.730 13.670 1.720 15.340 1.410 5.040 0.950 4.310 0.710 3.230 0.370 1.380 1.850 1.890 1.730 1.770 1.660 12.820 1.720 15.320 2.030 15.620 1.990 18.440 1.770 5.840 1.090 4.990 0.840 3.690 0.450 1.510 2.080 2.200 2.010 2.100 1.920 14.810 2.190 13.700 1.740 13.760 1.720 16.080 1.470 5.320 1.020 4.460 0.820 3.380 0.400 1.420 1.890 2.020 1.840 1.920 1.780 13.370 1.900 13.770 1.850 14.230 1.750 15.730 1.450 5.160 0.980 4.380 0.740 3.410 0.440 1.400 1.960 2.010 1.740 1.960 1.760 13.370 1.860 12.660 1.960 13.100 1.810 14.920 1.510 4.910 1.000 4.470 0.770 3.170 0.350 1.420 1.860 1.990 1.690 1.830 1.680 12.360 1.910 14.470 1.960 14.960 1.740 17.150 1.630 5.560 1.060 4.680 0.770 3.730 0.430 1.570 2.040 2.140 1.790 2.080 1.870 14.150 1.910 14.350 2.070 15.000 1.850 17.630 1.670 5.390 1.030 4.970 0.730 3.570 0.380 1.620 1.960 2.010 1.890 2.000 1.830 14.160 2.090 14.520 1.930 14.230 1.790 17.230 1.540 5.530 1.030 4.740 0.760 3.200 0.470 1.390 1.940 2.210 1.720 2.020 1.740 14.010 1.940 25.045 2.515 23.075 2.290 25.805 2.185 10.150 1.410 7.575 0.965 4.655 0.545 2.400 2.600 3.715 2.485 3.250 2.460 23.255 2.505 28.690 2.740 26.720 2.370 29.730 2.260 11.080 1.580 8.260 1.190 5.930 0.590 2.780 3.130 3.480 2.820 3.285 2.550 26.750 2.825 27.405 2.790 26.750 2.405 28.830 2.235 11.715 1.600 8.265 1.035 5.730 0.520 2.765 3.010 3.185 2.650 2.915 2.505 25.340 2.760 27.580 2.755 25.730 2.515 28.550 2.220 10.610 1.595 8.230 1.110 5.410 0.430 2.750 3.125 3.515 2.715 3.315 2.620 25.630 2.855 28.115 2.795 26.330 2.600 28.680 2.360 11.490 1.545 8.210 1.105 5.395 0.600 2.800 3.190 3.695 2.955 3.290 2.795 26.670 3.090 26.970 2.660 25.320 2.540 28.030 2.350 11.010 1.570 7.770 1.030 5.670 0.590 2.630 2.940 3.325 2.565 2.995 2.545 24.515 2.680 33.700 2.900 32.240 2.790 35.370 2.665 13.720 1.730 9.470 1.200 6.720 0.780 3.260 3.100 4.135 2.830 3.745 2.755 30.745 2.995 21.490 2.260 20.210 2.100 22.805 1.850 9.150 1.300 6.860 0.970 4.980 0.490 2.285 2.470 2.815 2.125 2.640 2.125 19.520 2.125 26.625 2.765 25.130 2.530 28.385 2.230 11.050 1.590 8.220 1.085 5.575 0.585 2.695 3.000 3.260 2.725 3.180 2.595 25.375 2.880 24.965 2.660 23.940 2.335 26.145 2.170 10.345 1.440 7.730 1.100 5.540 0.580 2.580 2.845 3.360 2.545 3.090 2.415 23.335 2.645 18.305 1.475 18.705 1.375 14.975 1.225 7.145 0.745 4.550 0.615 1.500 0.215 1.380 1.575 1.620 1.500 1.385 1.460 17.835 1.535 13.820 1.495 14.120 1.330 11.250 1.065 5.680 0.690 3.825 0.575 1.325 0.190 1.270 1.345 1.505 1.295 1.400 1.300 13.270 1.500 16.970 1.515 17.340 1.355 13.290 1.115 6.600 0.710 4.215 0.610 1.475 0.230 1.315 1.465 1.500 1.460 1.390 1.430 16.380 1.495 L3 T1 L L3 T1 W L3 T2 L L3 T2 W L3 TA L L3 TA W L3 PR L L3 PR W L3 CL L L3 CL W L4 C1 L L4 C1 W L4 C2 L L4 C2 W L4 C3 L L4 C3 W L4 F L L4 F W L4 T1 L L4 T1 W 11.350 1.230 9.360 1.040 4.920 0.680 2.980 0.470 1.350 0.220 1.170 1.260 1.410 1.340 1.230 1.280 10.510 1.250 11.370 1.260 16.020 1.530 12.890 1.220 5.850 0.710 3.880 0.540 1.460 0.230 1.250 1.340 1.480 1.400 1.270 1.340 13.390 1.320 15.550 1.230 15.690 1.420 12.190 1.010 5.860 0.640 3.520 0.480 2.150 0.230 1.280 1.450 1.490 1.470 1.390 1.360 13.840 1.280 14.210 1.230 16.010 1.230 11.720 1.330 5.620 0.680 4.110 0.620 1.770 0.390 1.600 1.580 1.980 1.530 1.690 1.620 14.080 1.310 14.780 1.250 15.980 1.140 12.550 0.930 5.370 0.720 3.760 0.570 1.300 0.260 1.590 1.590 1.750 1.490 1.480 1.390 14.610 1.290 14.880 1.070 15.320 1.140 11.380 0.950 4.320 0.610 3.380 0.450 1.680 0.220 1.650 1.600 2.100 1.640 1.880 1.470 17.670 1.400 17.620 1.270 12.940 0.880 10.320 0.870 4.710 0.640 3.410 0.440 1.740 0.230 1.350 1.320 1.730 1.350 1.410 1.310 12.980 1.090 13.410 1.030 17.720 1.300 12.790 1.210 5.790 0.770 3.940 0.630 1.940 0.300 1.640 1.580 2.070 1.680 1.860 1.680 16.320 1.560 16.370 1.340 19.680 1.240 14.530 1.010 5.910 0.620 4.190 0.500 1.620 0.220 1.570 1.420 1.790 1.470 1.690 1.410 18.310 1.460 18.190 1.140 16.330 1.370 12.870 1.240 5.630 0.810 4.290 0.740 1.540 0.230 1.530 1.550 1.870 1.490 1.600 1.590 14.310 1.470 15.510 1.330 16.500 1.230 12.970 1.140 5.720 0.640 3.940 0.550 1.690 0.230 1.210 1.470 2.030 1.330 1.520 1.340 15.630 1.340 15.850 0.940 18.040 1.290 13.070 1.070 5.410 0.710 3.730 0.540 1.960 0.220 1.590 1.620 1.990 1.510 1.590 1.540 17.040 1.530 16.310 1.350 20.230 1.480 15.510 1.190 6.120 0.760 4.100 0.590 1.940 0.210 1.650 1.540 2.070 1.550 1.960 1.730 18.300 1.610 18.950 1.300 17.870 1.300 13.760 1.160 5.880 0.690 4.240 0.540 1.560 0.190 1.640 1.650 2.070 1.660 1.830 1.600 16.550 1.500 16.920 1.170 17.410 1.780 19.570 1.710 8.310 1.270 6.740 0.880 4.920 0.490 2.010 2.410 2.590 1.980 2.310 2.120 17.800 1.940 15.650 1.950 9.130 0.700 8.100 0.610 3.580 0.410 3.410 0.350 2.410 0.260 0.960 1.020 1.270 0.920 1.120 0.890 9.470 0.780 8.420 0.670 16.400 0.840 14.270 0.740 6.380 0.550 5.130 0.430 0.000 0.000 1.460 1.240 1.600 1.120 1.490 1.070 17.120 0.970 14.920 0.770 15.490 0.900 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.330 1.320 1.630 1.100 1.560 1.170 16.530 1.120 14.330 0.760 8.730 1.750 10.140 1.530 3.850 1.000 3.720 0.780 2.730 0.420 1.180 1.670 1.800 1.720 1.670 1.700 8.220 1.630 8.510 1.710 12.280 1.710 14.330 1.350 4.850 1.010 4.480 0.750 3.610 0.350 1.250 1.770 1.990 1.830 1.910 1.750 11.440 1.730 11.710 1.540 14.730 1.930 16.910 1.650 5.710 1.120 5.100 0.820 3.560 0.450 1.580 1.940 2.070 1.770 1.910 1.800 13.290 2.030 13.490 1.870 13.490 1.810 15.340 1.500 4.600 1.050 4.510 0.820 3.810 0.430 1.290 1.840 1.980 1.760 1.880 1.820 12.120 1.930 12.520 1.730 12.670 1.690 14.690 1.500 4.880 0.950 4.250 0.700 3.200 0.300 1.360 1.890 1.940 1.700 1.840 1.770 11.570 1.960 12.050 1.700 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.320 1.790 1.780 1.830 1.870 1.760 11.190 1.930 11.620 1.760 14.090 1.920 16.560 1.720 5.320 1.200 4.800 0.870 3.260 0.490 1.540 1.970 2.040 1.880 1.920 1.920 13.000 1.990 13.430 1.990 16.330 2.040 18.280 1.750 5.910 1.180 5.160 0.870 3.630 0.510 1.620 2.120 2.200 1.970 2.170 2.030 15.000 2.130 15.480 1.970 13.070 1.710 15.500 1.460 5.040 1.040 4.480 0.750 3.220 0.390 1.350 1.810 1.880 1.610 1.790 1.640 11.810 1.810 12.210 1.610 15.030 1.920 17.980 1.680 5.760 1.100 4.980 0.840 3.800 0.380 1.530 1.980 2.110 1.950 2.070 2.050 13.580 2.130 13.640 1.810 13.220 1.700 15.430 1.510 5.050 1.080 4.490 0.800 3.510 0.380 1.490 1.860 2.170 1.850 1.920 1.870 12.200 1.850 12.540 1.680 13.810 1.710 15.880 1.590 5.170 0.990 4.310 0.720 3.520 0.420 1.350 1.810 2.020 1.920 1.970 1.880 12.530 1.840 13.110 1.740 12.360 1.690 14.730 1.470 4.890 0.950 4.490 0.770 3.780 0.410 1.390 1.800 1.790 1.620 1.770 1.650 11.290 1.800 11.400 1.710 14.440 1.790 16.700 1.590 5.330 0.990 4.780 0.730 3.840 0.370 1.470 1.880 2.010 1.860 1.970 1.880 12.990 1.980 13.490 1.760 14.070 1.770 16.570 1.640 5.450 1.020 4.880 0.800 3.760 0.390 1.490 1.940 1.940 1.820 1.960 1.910 13.270 2.080 13.590 1.840 13.730 1.870 16.370 1.510 5.310 1.020 4.760 0.770 3.430 0.370 1.340 1.940 2.170 1.780 2.040 1.740 12.770 1.840 12.760 1.700 21.765 2.285 23.835 2.130 9.710 1.680 7.330 1.120 4.780 0.760 2.410 2.720 3.620 2.590 3.430 2.660 21.820 2.350 20.110 2.320 24.830 2.595 26.935 2.380 10.395 1.510 7.720 1.035 5.530 0.655 2.710 3.020 3.440 2.875 2.920 2.685 24.655 2.655 21.835 2.480 23.645 2.560 25.790 2.250 10.370 1.530 7.960 0.930 5.170 0.610 2.635 2.970 3.160 2.695 2.735 2.660 23.385 2.790 21.855 2.540 23.855 2.525 26.015 2.265 10.020 1.580 7.970 1.070 5.660 0.560 2.575 2.950 3.340 2.655 2.740 2.670 23.450 2.870 21.630 2.590 24.055 2.770 26.640 2.325 10.660 1.555 7.925 1.155 5.375 0.635 2.745 3.005 3.435 2.840 2.745 2.665 24.745 2.850 21.965 2.610 22.910 2.650 24.820 2.330 9.670 1.570 7.380 1.070 5.810 0.530 2.535 2.770 3.415 2.625 3.020 2.410 23.370 2.490 20.070 2.570 29.090 2.860 31.770 2.550 12.145 1.675 8.940 1.185 7.680 0.720 3.065 2.820 3.840 2.760 3.540 2.710 28.120 2.850 26.920 2.720 18.125 1.990 20.130 1.835 8.140 1.345 6.730 0.920 5.210 0.440 2.295 2.395 2.875 1.955 2.380 1.940 17.790 2.005 16.615 1.800 23.300 2.580 25.900 2.210 9.910 1.570 7.870 1.090 5.020 0.600 2.650 2.890 3.180 2.610 3.100 2.410 22.870 2.560 21.320 2.580 21.725 2.500 23.795 2.125 9.460 1.420 7.450 1.020 5.370 0.575 2.440 2.750 2.980 2.395 2.765 2.450 21.225 2.610 19.625 2.430 17.465 1.310 14.175 1.125 6.930 0.640 4.485 0.535 1.375 0.235 1.420 1.500 1.645 1.475 1.575 1.410 16.980 1.575 17.325 1.325 13.880 1.250 10.995 1.090 4.370 0.655 3.480 0.475 1.290 0.220 1.325 1.390 1.490 1.355 1.420 1.315 12.570 1.430 13.260 1.295 16.815 1.380 13.240 1.230 6.370 0.680 4.350 0.560 1.370 0.230 1.335 1.450 1.520 1.525 1.490 1.445 15.120 1.520 16.020 1.250 L4 T2 L L4 T2 W L4 TA L L4 TA W L4 PR L L4 PR W L4 CL L L4 CL W 9.230 1.010 4.420 0.670 3.420 0.560 1.490 0.220 12.350 1.200 6.130 0.530 3.440 0.420 1.500 0.240 11.440 1.020 5.570 0.580 3.900 0.480 2.640 0.210 11.780 1.070 5.930 0.690 4.060 0.550 1.130 0.360 11.670 1.010 5.190 0.660 3.660 0.480 1.190 0.260 12.320 1.080 4.450 0.580 3.400 0.470 1.700 0.250 10.100 0.910 4.780 0.630 3.540 0.480 1.620 0.170 12.410 1.180 5.390 0.610 3.930 0.620 1.890 0.320 13.610 0.910 5.530 0.650 3.880 0.540 1.610 0.190 12.420 1.160 5.410 0.770 4.050 0.690 1.590 0.250 12.220 0.910 5.400 0.680 3.980 0.490 1.840 0.180 12.520 1.100 5.420 0.650 3.930 0.580 2.030 0.280 14.710 1.170 6.080 0.760 4.230 0.590 1.950 0.260 13.160 1.120 5.700 0.700 4.080 0.550 1.830 0.210 17.560 1.850 7.660 1.310 6.410 0.950 4.880 0.530 8.060 0.560 3.490 0.410 3.270 0.370 2.250 0.180 13.170 0.690 5.660 0.550 5.020 0.430 3.910 0.260 12.730 0.680 5.850 0.550 4.950 0.420 3.250 0.290 9.580 1.480 3.580 1.060 3.650 0.770 2.990 0.390 13.270 1.350 4.690 0.890 4.250 0.730 3.640 0.360 15.960 1.580 5.340 1.100 4.820 0.810 3.850 0.400 14.400 1.420 4.490 1.070 4.420 0.810 3.530 0.380 14.350 1.590 4.540 1.000 4.150 0.730 3.400 0.310 13.010 1.560 3.200 0.940 4.250 0.680 3.150 0.400 15.930 1.780 5.300 1.140 4.740 0.890 3.680 0.460 18.310 1.660 5.730 1.180 5.120 0.840 3.520 0.450 14.510 1.370 4.770 0.900 4.180 0.750 3.180 0.340 16.400 1.550 5.590 1.100 5.160 0.820 3.660 0.420 14.790 1.450 5.140 1.010 4.460 0.800 3.480 0.370 15.090 1.450 4.970 0.990 4.580 0.770 3.450 0.390 13.530 1.450 4.420 0.980 4.130 0.740 3.130 0.350 16.120 1.540 5.220 0.960 4.660 0.730 3.920 0.430 16.100 1.620 5.210 0.990 4.950 0.800 3.770 0.440 15.640 1.410 4.720 0.980 4.670 0.690 3.130 0.340 22.010 1.970 8.930 1.440 6.990 1.090 5.420 0.630 24.940 2.195 9.745 1.595 7.360 1.010 5.770 0.515 24.465 2.240 9.735 1.515 7.480 0.990 4.710 0.440 24.870 2.235 9.455 1.495 7.755 1.025 4.985 0.420 24.890 2.135 9.790 1.555 7.525 1.055 6.005 0.580 23.430 2.050 9.340 1.440 7.280 1.010 5.580 0.540 29.030 2.380 11.270 1.610 8.860 1.210 7.510 0.700 18.780 1.545 7.620 1.135 5.880 0.880 4.600 0.470 23.630 2.110 9.000 1.410 7.480 1.060 5.650 0.550 22.125 1.985 8.920 1.420 7.155 1.085 5.800 0.545 14.090 1.150 6.630 0.670 4.410 0.560 1.520 0.240 10.650 1.095 5.160 0.640 3.615 0.500 1.375 0.225 12.540 1.110 5.820 0.640 4.300 0.560 1.390 0.230 Electronic supplement 2: Relative measurements expressed as proportions of the trunk length Trunk H Trunk W Trunk W23 Proboscis W Proboscis L Abdomen L Abdomen W P1L P2L P3L P4L P5L P6L P7L P8L P9L P10L O1L 257-2.4 0.19527 0.51119 0.20274 0.15609 0.72139 0.21891 0.04975 0.03358 0.03731 0.34453 0.05846 0.24005 0.09453 0.12811 0.09328 0.07711 0.09826 0.05970 300-1.4 0.20633 0.52169 0.19695 0.17526 0.77960 0.19109 0.04689 0.03165 0.03400 0.33529 0.06448 0.27667 0.09613 0.13716 0.08792 0.09144 0.10317 0.06800 AGT42/164 0.20992 0.54902 0.19146 0.16551 0.76471 0.21223 0.04844 0.03922 0.04152 0.34256 0.04960 0.24567 0.08881 0.13379 0.09573 0.07958 0.10035 0.06574 211-5.1 0.20883 0.53498 0.18837 0.15985 0.78149 0.23466 0.06889 0.02476 0.03014 0.42519 0.05274 0.30248 0.10764 0.15393 0.09688 0.08719 0.09903 0.07212 211-6.3.2 0.20862 0.61072 0.21678 0.18706 0.87995 0.21096 0.05828 0.03030 0.03730 0.33800 0.06876 0.29138 0.10723 0.13869 0.08974 0.09674 0.10140 0.07226 29OT27-1 0.20000 0.52923 0.16205 0.16564 0.87487 0.20103 0.03897 0.03487 0.03590 0.45641 0.05949 0.29436 0.10051 0.16821 0.10974 0.09231 0.08513 0.07795 29OT27-2 0.19603 0.56452 0.16253 0.17370 0.83002 0.21960 0.03846 0.02978 0.03102 0.36849 0.05087 0.31390 0.11290 0.17246 0.10546 0.11538 0.10918 0.07320 30BT14-2 0.20323 0.54891 0.17854 0.18044 0.85660 0.21937 0.06363 0.02849 0.03229 0.42830 0.05698 0.31624 0.10636 0.15385 0.10731 0.09592 0.10826 0.07597 45BT24 0.19781 0.56503 0.18033 0.16339 0.94645 0.22186 0.06339 0.02951 0.03060 0.43497 0.05902 0.32131 0.10055 0.16284 0.10055 0.09399 0.11803 0.08415 PA_E006 0.21886 0.56566 0.19529 0.19473 0.83502 0.23793 0.04938 0.03479 0.03143 0.42088 0.07969 0.29966 0.11560 0.15264 0.10999 0.12121 0.09203 0.06846 PB_E005 0.18462 0.53949 0.17641 0.16564 0.86667 0.21641 0.05026 0.03385 0.03179 0.42462 0.10769 0.28718 0.08513 0.16000 0.09436 0.10974 0.07897 0.07590 PR_E006 0.19355 0.52884 0.16422 0.16618 0.90616 0.20430 0.03910 0.03421 0.03519 0.41935 0.06549 0.29326 0.08700 0.13685 0.08798 0.08504 0.08113 0.06940 PR_E010 0.19808 0.50577 0.17212 0.17885 0.86731 0.23942 0.04327 0.05481 0.03846 0.43750 0.07596 0.27115 0.09519 0.15769 0.09712 0.11154 0.08846 0.06923 PS_E011 0.19797 0.50051 0.17766 0.16853 0.93198 0.22640 0.05279 0.03249 0.03452 0.44569 0.07005 0.28934 0.09746 0.15533 0.10761 0.10254 0.09645 0.08020 257-2.1 0.21264 0.56732 0.19787 0.16995 0.85386 0.21511 0.05172 0.04762 0.05172 0.36371 0.10016 0.31609 0.10591 0.14532 0.10345 0.10591 0.10181 0.08703 257-2.5 0.17771 0.50753 0.16265 0.16114 0.81325 0.15361 0.07681 0.04066 0.03916 0.36898 0.08886 0.26506 0.09639 0.07831 0.09337 0.13102 0.13554 0.08434 233-3.1.2 0.16628 0.58028 0.14794 0.13761 0.84518 0.14564 0.06307 0.04128 0.04702 0.46904 0.09518 0.28555 0.09862 0.08945 0.07683 0.10894 0.11927 0.07913 233-3.1.1 0.17819 0.52989 0.13951 0.13482 0.85463 0.04338 0.04807 0.46307 0.09144 0.32005 0.09848 0.08675 0.09144 0.12075 0.13834 0.07268 226-7.2 0.31850 0.66203 0.28790 0.25522 1.01808 0.24757 0.08901 0.04590 0.05285 0.51321 0.08484 0.32545 0.09875 0.15021 0.09318 0.09875 0.05424 0.08623 59BT40 0.25817 0.58904 0.25184 0.20179 0.96312 0.25395 0.06744 0.04004 0.03793 0.50369 0.08851 0.29926 0.11380 0.14436 0.10854 0.11486 0.07482 0.08746 59OT45 0.27485 0.58967 0.24854 0.21394 1.03509 0.26121 0.07602 0.04191 0.05361 0.54386 0.11988 0.29825 0.10526 0.15497 0.09844 0.11696 0.03411 0.08674 66OT97 0.27403 0.55930 0.24949 0.21319 1.01125 0.27505 0.06237 0.03885 0.04192 0.53783 0.07771 0.28425 0.09100 0.14826 0.10020 0.10327 0.06237 0.07669 81OT58 0.24272 0.58360 0.25243 0.21359 1.00000 0.26645 0.07120 0.05178 0.04099 0.54477 0.07120 0.27508 0.09817 0.10140 0.08846 0.09385 0.02913 0.08954 PA_E003 0.30170 0.62287 0.28710 0.23540 1.01217 0.30779 0.07178 0.04136 0.04623 0.52068 0.07786 0.33090 0.13747 0.18491 0.12652 0.11679 0.05353 0.08881 PF_E008 0.29310 0.65314 0.27485 0.21400 1.05274 0.28600 0.07809 0.04665 0.04767 0.58925 0.07606 0.28702 0.12069 0.15112 0.11055 0.13387 0.08316 0.08925 PQ_E007 0.28546 0.59515 0.25224 0.20691 0.98294 0.24147 0.06912 0.03860 0.04039 0.56014 0.09605 0.24686 0.10503 0.13824 0.08438 0.09964 0.06643 0.07989 PQ_E008 0.27974 0.58735 0.25402 0.21008 0.99357 0.25402 0.07610 0.03323 0.03001 0.53269 0.10075 0.25080 0.10075 0.13612 0.08467 0.10289 0.06645 0.08360 PQ_E010 0.28193 0.61968 0.26679 0.21239 0.99149 0.27625 0.07474 0.03311 0.03690 0.56386 0.10312 0.30085 0.10407 0.14948 0.08420 0.09650 0.06528 0.08136 PQ_E011 0.28485 0.57778 0.24444 0.21061 0.95152 0.25960 0.07172 0.03737 0.04444 0.56364 0.10505 0.27071 0.10404 0.13838 0.08384 0.10202 0.06465 0.08283 PQ_E012 0.27968 0.60463 0.25453 0.21328 1.01006 0.26559 0.07445 0.03421 0.03722 0.51911 0.10463 0.26761 0.10664 0.15694 0.09658 0.11167 0.06740 0.07746 PR_E003 0.25987 0.58882 0.25439 0.21546 0.99561 0.27303 0.07785 0.03947 0.04167 0.48136 0.06579 0.28728 0.10526 0.13268 0.09868 0.09101 0.06140 0.07895 PR_E004 0.27154 0.58818 0.25651 0.19088 0.98697 0.27355 0.07415 0.04709 0.05411 0.57615 0.09820 0.30261 0.09319 0.13226 0.09719 0.10321 0.07114 0.07415 PR_E005 0.27805 0.62537 0.25854 0.21122 0.99805 0.25951 0.08585 0.05171 0.04683 0.54537 0.09951 0.27902 0.09756 0.13854 0.08390 0.11317 0.06927 0.07902 PS_E010 0.26861 0.59054 0.24950 0.21328 0.97988 0.25553 0.07847 0.03119 0.03018 0.52012 0.08249 0.24950 0.08753 0.12777 0.08249 0.10664 0.06439 0.08350 IU-2007-4842 0.23422 0.55765 0.18993 0.20722 0.89442 0.28701 0.05279 0.04612 0.03944 0.43629 0.08434 0.31371 0.12864 0.18932 0.11833 0.12561 0.10558 IU-2007-5063 0.23282 0.55628 0.18898 0.21505 0.96623 0.23460 0.05983 0.04384 0.03318 0.41232 0.07761 0.29739 0.11908 0.17417 0.13152 0.12322 0.10900 0.09656 IU-2007-5039 0.22788 0.58625 0.18396 0.23616 1.05283 0.22788 0.05029 0.04265 0.03692 0.43857 0.07447 0.33800 0.11649 0.18905 0.12667 0.12858 0.10948 0.09612 IU-2007-5069 0.22929 0.57418 0.19910 0.22222 0.98459 0.21130 0.04817 0.04367 0.03661 0.44958 0.07643 0.32241 0.12267 0.19139 0.12139 0.13231 0.10854 0.09056 IU-2007-5043 0.24064 0.58586 0.19370 0.21985 0.99584 0.21925 0.05169 0.03981 0.03327 0.42781 0.07427 0.29649 0.11765 0.18301 0.11646 0.11765 0.10398 0.09031 IU-2007-4798 0.22011 0.55120 0.19313 0.20601 0.95340 0.22440 0.05273 0.05089 0.04660 0.43102 0.08768 0.31392 0.13673 0.18639 0.12385 0.12508 0.12078 0.08216 IU-2007-4870 0.22211 0.53877 0.18459 0.20085 0.92996 0.21411 0.04952 0.04352 0.03902 0.46673 0.08104 0.29865 0.11756 0.18709 0.13607 0.12056 0.11206 0.08254 IU-2007-4800 0.22546 0.57539 0.18777 0.19950 0.95519 0.21266 0.04908 0.04339 0.03698 0.47724 0.06828 0.31010 0.12162 0.18421 0.11238 0.12518 0.10242 0.08819 IU-2007-4795 0.24097 0.59278 0.19427 0.21887 0.94396 0.19863 0.05293 0.04047 0.03611 0.45143 0.07721 0.34060 0.11955 0.18867 0.12578 0.12329 0.10959 0.09091 IU-2007-5044 0.26098 0.56461 0.18969 0.21547 0.94589 0.22470 0.05347 0.04201 0.03692 0.45576 0.07829 0.32400 0.11776 0.18269 0.12094 0.12349 0.10757 0.07257 IU-2013-15805 0.20952 0.49947 0.18095 0.15132 0.78095 0.22857 0.05714 0.02646 0.02963 0.38201 0.05714 0.29418 0.11429 0.16085 0.09524 0.11217 0.10265 0.06561 IU-2013-15808 0.20670 0.46704 0.18212 0.15587 0.78883 0.22793 0.05363 0.03240 0.03575 0.37877 0.06369 0.27598 0.10391 0.15866 0.11397 0.12402 0.09274 0.06927 IU-2013-15812 0.20978 0.51413 0.18370 0.15543 0.78587 0.21957 0.05326 0.03043 0.02826 0.35652 0.05435 0.29130 0.11848 0.15000 0.10217 0.11087 0.09674 0.06196 O1W O2L O2W O3L O3W O4L O4W O5L O5W O6L O6W O7L O7W O8L O8W O9L O9W 010L O10W Basal element W 0.07711 0.07463 0.08458 0.08706 0.08582 0.63433 0.06592 0.22264 0.08458 0.70647 0.06468 0.20025 0.05970 0.15547 0.07836 0.14925 0.04602 0.12935 0.03234 0.07525 0.09027 0.08089 0.09261 0.11137 0.07737 0.82649 0.07503 0.22157 0.08675 0.81829 0.07855 0.20164 0.07151 0.18875 0.07503 0.18054 0.06213 0.09848 0.04455 0.07855 0.08997 0.07266 0.08766 0.08881 0.08304 0.80738 0.06574 0.21799 0.08881 0.81084 0.06920 0.22722 0.04960 0.16263 0.08074 0.15340 0.04498 0.13841 0.03691 0.07555 0.06781 0.07320 0.07320 0.11518 0.08504 0.81701 0.07858 0.20990 0.09150 0.83746 0.08288 0.20237 0.06351 0.15931 0.06674 0.15716 0.05920 0.13671 0.04306 0.08611 0.08508 0.06177 0.08508 0.08042 0.07925 0.86247 0.08275 0.22727 0.09674 0.95688 0.08159 0.20979 0.07459 0.16084 0.07809 0.15501 0.06061 0.10490 0.02797 0.09091 0.09128 0.07077 0.07282 0.08821 0.08718 0.96410 0.07179 0.26256 0.09231 0.97949 0.06769 0.19385 0.06256 0.14974 0.06667 0.16821 0.05846 0.11692 0.03487 0.06462 0.08065 0.06824 0.07072 0.09553 0.08437 0.88462 0.07196 0.22829 0.08561 0.87593 0.08065 0.19107 0.06948 0.15881 0.07072 0.16005 0.05831 0.13896 0.03722 0.07320 0.09117 0.06933 0.08642 0.10256 0.08072 0.80722 0.08357 0.20608 0.09307 0.85565 0.07217 0.16714 0.05888 0.14815 0.06173 0.15954 0.05603 0.13200 0.03894 0.09117 0.08087 0.07322 0.08306 0.10164 0.08743 0.86667 0.07869 0.22732 0.09071 1.08087 0.09071 0.18579 0.06667 0.14317 0.06011 0.15738 0.05355 0.13443 0.03825 0.08743 0.08418 0.07520 0.07295 0.08418 0.09540 0.91807 0.08193 0.18294 0.10887 0.91582 0.09989 0.20651 0.07407 0.17284 0.07183 0.20314 0.07407 0.10101 0.02581 0.08361 0.07795 0.07385 0.06974 0.08821 0.08205 0.89641 0.06769 0.23282 0.07795 0.90769 0.07282 0.17641 0.06667 0.15487 0.06564 0.12923 0.05949 0.12410 0.02974 0.08103 0.07331 0.07331 0.07429 0.09091 0.08407 0.82796 0.07527 0.24242 0.08504 0.93157 0.07429 0.17400 0.06549 0.14565 0.05767 0.15445 0.05376 0.13978 0.03226 0.06989 0.08269 0.06635 0.07500 0.07981 0.08365 0.83846 0.07115 0.23942 0.09038 0.97981 0.07692 0.19519 0.06827 0.17885 0.06154 0.17115 0.05865 0.12404 0.03654 0.08077 0.08528 0.07614 0.07107 0.09442 0.09543 0.90558 0.07310 0.24061 0.07614 0.94924 0.06701 0.21523 0.05584 0.17766 0.05685 0.17259 0.04569 0.12690 0.02640 0.08274 0.09770 0.10181 0.07061 0.08292 0.09852 0.71839 0.07964 0.30213 0.06158 0.77011 0.07718 0.18637 0.07882 0.18966 0.06732 0.19540 0.05747 0.14532 0.04844 0.08744 0.07982 0.07530 0.06777 0.09940 0.06928 0.73042 0.08434 0.38253 0.08283 0.84639 0.06175 0.20030 0.06777 0.13404 0.06476 0.11596 0.05422 0.12349 0.04669 0.10392 0.07683 0.07569 0.07110 0.08716 0.06307 1.05619 0.08257 0.45872 0.07798 1.14220 0.07798 0.19954 0.06422 0.14335 0.06995 0.13188 0.05161 0.11583 0.03899 0.08314 0.07855 0.06448 0.05510 0.08324 0.05744 1.02814 0.08441 0.41383 0.07855 1.05862 0.07503 0.16882 0.05275 0.15592 0.07034 0.14889 0.06448 0.13130 0.03517 0.07093 0.10431 0.09318 0.10014 0.08901 0.09318 0.73992 0.12935 0.31572 0.14186 0.59944 0.13352 0.22670 0.12378 0.25313 0.10292 0.23088 0.10014 0.19193 0.06537 0.09805 0.10221 0.07271 0.08114 0.09168 0.09694 0.71970 0.10643 0.28767 0.10959 0.66070 0.09273 0.22023 0.08641 0.20759 0.09062 0.18651 0.07692 0.16228 0.04110 0.09378 0.10234 0.08382 0.07895 0.10039 0.09357 0.79532 0.09844 0.30409 0.10429 0.73587 0.10819 0.19103 0.10136 0.19883 0.10429 0.19201 0.09162 0.16862 0.05750 0.09064 0.10020 0.06748 0.08180 0.09305 0.11554 0.77198 0.09918 0.26789 0.12168 0.77914 0.09509 0.20450 0.09611 0.20757 0.08793 0.19632 0.07873 0.16667 0.05112 0.08027 0.11111 0.08414 0.08846 0.10032 0.10895 0.73786 0.09924 0.26645 0.12082 0.61704 0.10356 0.19633 0.11111 0.22114 0.10680 0.17799 0.09709 0.17044 0.05502 0.09277 0.12287 0.10219 0.10949 0.09367 0.11071 0.80049 0.12774 0.31387 0.13017 0.80779 0.12530 0.24331 0.10949 0.25061 0.10584 0.24453 0.08881 0.22993 0.05961 0.10219 0.09533 0.09939 0.09331 0.08215 0.08722 0.81440 0.10548 0.33671 0.13590 0.69675 0.10751 0.28499 0.10649 0.24746 0.09432 0.21907 0.07708 0.17951 0.04564 0.09432 0.10952 0.06732 0.07989 0.10054 0.10862 0.69031 0.12029 0.28636 0.11849 0.66248 0.09605 0.21634 0.09246 0.21005 0.09066 0.19569 0.07540 0.17145 0.05566 0.08528 0.10825 0.07181 0.08682 0.10182 0.11683 0.70311 0.12004 0.30761 0.12433 0.70311 0.10075 0.21972 0.09003 0.21543 0.08789 0.19614 0.07503 0.15756 0.05038 0.08414 0.10312 0.07852 0.09177 0.10691 0.11637 0.79659 0.11542 0.31504 0.12015 0.68685 0.09650 0.22990 0.09366 0.22422 0.08798 0.19205 0.07190 0.15137 0.04920 0.08798 0.10606 0.08081 0.09596 0.10606 0.11515 0.68990 0.11414 0.30707 0.11414 0.70606 0.09798 0.21515 0.09394 0.21010 0.08586 0.19495 0.07374 0.15051 0.04949 0.08990 0.09960 0.08652 0.09356 0.10966 0.11066 0.71831 0.11268 0.30584 0.11670 0.69718 0.10161 0.21831 0.09054 0.21227 0.09557 0.18813 0.07847 0.15191 0.04930 0.07948 0.10746 0.08772 0.09430 0.10526 0.10965 0.74342 0.09320 0.28947 0.12281 0.69518 0.09978 0.20724 0.09430 0.20395 0.08553 0.18750 0.06908 0.15789 0.04386 0.10197 0.08918 0.08317 0.09319 0.09820 0.10421 0.79860 0.11523 0.27054 0.12425 0.73848 0.09719 0.20541 0.09619 0.24349 0.08417 0.20641 0.07415 0.17234 0.04409 0.08066 0.10341 0.08098 0.09268 0.09561 0.10829 0.74146 0.10537 0.28878 0.10829 0.69366 0.09659 0.18732 0.09561 0.21268 0.09073 0.17171 0.07707 0.15707 0.05073 0.09415 0.10765 0.07646 0.09256 0.09256 0.10765 0.70423 0.12374 0.30282 0.11368 0.69718 0.10362 0.23642 0.09155 0.20020 0.09457 0.17807 0.07746 0.15191 0.04829 0.07646 0.09951 0.11197 0.09597 0.11374 0.09123 0.10071 0.80332 0.11730 0.32879 0.09597 0.91943 0.10130 0.22275 0.09538 0.20557 0.08353 0.18720 0.07879 0.18839 0.06102 0.09686 0.10312 0.08339 0.09803 0.09612 0.10312 0.83068 0.11903 0.35073 0.10694 0.93316 0.10312 0.24443 0.08721 0.22470 0.08402 0.20815 0.07893 0.18842 0.05283 0.09516 0.10854 0.09313 0.09955 0.09634 0.11111 0.81888 0.12010 0.31407 0.10148 0.89852 0.10597 0.24213 0.07900 0.23186 0.08735 0.21002 0.07386 0.20617 0.05909 0.09409 0.10695 0.08853 0.10398 0.08734 0.09626 0.78194 0.11705 0.33512 0.10576 0.86869 0.10160 0.22282 0.08675 0.20559 0.08734 0.19846 0.07843 0.18063 0.05585 0.09507 0.11220 0.08400 0.09994 0.09871 0.10300 0.78418 0.11956 0.32250 0.10730 0.86818 0.10116 0.23237 0.09565 0.19865 0.09135 0.21398 0.08216 0.17413 0.05886 0.09687 0.09355 0.08854 0.10555 0.08304 0.09155 0.77539 0.09905 0.28614 0.09455 0.84292 0.09905 0.20260 0.08704 0.20460 0.08854 0.18709 0.07254 0.16908 0.05253 0.09330 0.11024 0.08464 0.10597 0.08606 0.11166 0.78236 0.10740 0.29303 0.09531 0.80797 0.10384 0.22546 0.08962 0.20555 0.08464 0.20341 0.07895 0.18492 0.05405 0.09211 0.11768 0.08655 0.09215 0.10274 0.10585 0.82690 0.11831 0.32192 0.11582 0.93275 0.11270 0.24346 0.09340 0.23599 0.09589 0.22976 0.08468 0.19303 0.05230 0.10118 0.10376 0.08275 0.10694 0.09102 0.10630 0.80586 0.11330 0.32145 0.10503 0.87397 0.10503 0.23679 0.09675 0.19860 0.09357 0.21133 0.08211 0.18205 0.05792 0.09962 0.07513 0.07090 0.07302 0.08360 0.08042 0.80635 0.08042 0.22116 0.09101 0.90476 0.07937 0.17884 0.06032 0.15450 0.06561 0.16190 0.06032 0.13228 0.03915 0.09048 0.07821 0.07821 0.07486 0.08380 0.07709 0.70056 0.07709 0.22346 0.08156 0.72514 0.05475 0.18212 0.04916 0.16089 0.04581 0.17765 0.05028 0.12737 0.03240 0.07821 0.07500 0.06413 0.06848 0.08370 0.08804 0.78261 0.08152 0.22174 0.07826 0.85435 0.07717 0.17826 0.07391 0.15978 0.07065 0.16413 0.05761 0.13370 0.03478 0.08207 Lateral 12 Lateral 23 Lateral 34 W Lateral 1 W Lateral 2 W Lateral 3 W Lateral 4 Ocular tubercle W Ocular tubercle H Front height a Eye size a Front height p Eye size p L1 C1 L L1 C1 W L1 C2 L L1 C2 W L1 C3 L L1 C3 W 0.11567 0.15485 0.11692 0.13122 0.14490 0.13744 0.13682 0.10323 0.14179 0.05970 0.04726 0.06592 0.03607 0.14925 0.15236 0.18408 0.16045 0.17164 0.15547 0.12134 0.14302 0.13482 0.12309 0.13247 0.12778 0.12427 0.10317 0.13834 0.06213 0.04103 0.15064 0.17057 0.20106 0.16823 0.17292 0.16295 0.14937 0.17013 0.14591 0.11995 0.12803 0.14014 0.11707 0.11073 0.14418 0.05306 0.04614 0.06344 0.03230 0.14360 0.15744 0.19031 0.16378 0.17243 0.16032 0.11195 0.15447 0.15554 0.13509 0.13240 0.14424 0.13402 0.11195 0.15178 0.05274 0.03714 0.07320 0.02906 0.15501 0.15231 0.18515 0.14370 0.14909 0.14155 0.12879 0.17657 0.17832 0.14336 0.13928 0.14394 0.14336 0.12587 0.22378 0.05594 0.04429 0.06760 0.04720 0.16667 0.16200 0.15326 0.15093 0.15559 0.15793 0.10974 0.15385 0.13385 0.11897 0.12769 0.13846 0.11744 0.10667 0.13436 0.05231 0.04564 0.05949 0.03385 0.14872 0.15487 0.17385 0.14205 0.15641 0.16051 0.11414 0.15633 0.13027 0.13275 0.14764 0.14826 0.12903 0.12283 0.14392 0.08437 0.04963 0.07940 0.04218 0.15509 0.14392 0.18300 0.14268 0.15509 0.14206 0.11111 0.15812 0.15385 0.12251 0.14482 0.14577 0.13865 0.11301 0.14435 0.05508 0.04131 0.06458 0.02944 0.14767 0.14245 0.20085 0.15670 0.16714 0.15290 0.13060 0.16230 0.14699 0.12459 0.14918 0.15464 0.13880 0.11148 0.14536 0.05683 0.03880 0.06339 0.02896 0.15410 0.15628 0.21257 0.17486 0.17377 0.16066 0.13187 0.14534 0.12402 0.15376 0.16611 0.16498 0.15152 0.10999 0.15264 0.05836 0.05724 0.07071 0.04770 0.14927 0.15320 0.20146 0.17172 0.20258 0.17452 0.12923 0.16051 0.12615 0.12923 0.14154 0.14513 0.12462 0.11692 0.14256 0.06564 0.03795 0.06359 0.02769 0.16256 0.15026 0.19128 0.14051 0.17487 0.15077 0.12121 0.16862 0.12903 0.13099 0.13636 0.14223 0.12757 0.13001 0.14370 0.07234 0.04692 0.07038 0.04399 0.13832 0.12903 0.17742 0.13978 0.15591 0.15982 0.12067 0.17596 0.17837 0.13365 0.14615 0.14231 0.13077 0.11346 0.12596 0.04423 0.04087 0.06154 0.04423 0.16154 0.16394 0.18029 0.14663 0.15048 0.15433 0.12741 0.16954 0.14569 0.13249 0.13807 0.14772 0.12487 0.11269 0.15228 0.05584 0.03909 0.06091 0.04112 0.14518 0.14569 0.19239 0.15076 0.17056 0.14822 0.08990 0.11905 0.10181 0.17282 0.18555 0.18144 0.16544 0.10099 0.13547 0.06979 0.05213 0.07471 0.04475 0.18473 0.19376 0.22824 0.16544 0.19910 0.17241 0.08810 0.10090 0.11822 0.14081 0.13404 0.12349 0.14006 0.11596 0.12651 0.05120 0.04066 0.06928 0.03539 0.13780 0.15136 0.17545 0.12877 0.17093 0.12877 0.08314 0.09346 0.12557 0.13417 0.12959 0.12901 0.13647 0.09518 0.12385 0.05849 0.03784 0.05963 0.03899 0.17144 0.14622 0.18693 0.13073 0.17603 0.12787 0.08734 0.09730 0.12603 0.13306 0.12251 0.13892 0.12954 0.11020 0.13951 0.04338 0.03869 0.04689 0.04220 0.15533 0.14713 0.18875 0.12661 0.17292 0.13189 0.07093 0.11335 0.09944 0.17455 0.19124 0.18637 0.16829 0.18637 0.15716 0.04590 0.06815 0.05146 0.05355 0.17733 0.23644 0.21558 0.22601 0.22323 0.22949 0.10379 0.13909 0.10695 0.16280 0.18124 0.17492 0.16017 0.18757 0.14015 0.04531 0.05321 0.05163 0.05216 0.13699 0.19231 0.19600 0.16965 0.18282 0.17334 0.07407 0.11257 0.10185 0.14620 0.17105 0.17349 0.15058 0.16277 0.15205 0.04971 0.03216 0.04678 0.05945 0.14035 0.19347 0.20224 0.18177 0.19347 0.17300 0.09458 0.12832 0.10429 0.16462 0.18252 0.16922 0.15900 0.18609 0.13088 0.04703 0.04243 0.05215 0.04039 0.14008 0.19888 0.19070 0.16616 0.17791 0.17536 0.06796 0.10356 0.08145 0.16181 0.17476 0.18285 0.16397 0.18015 0.15750 0.03883 0.04531 0.02697 0.04369 0.14617 0.18878 0.20334 0.18662 0.18123 0.18501 0.07360 0.11071 0.10036 0.16788 0.18370 0.17883 0.16545 0.17518 0.14234 0.04380 0.05414 0.05961 0.05596 0.15511 0.21655 0.22506 0.21107 0.21655 0.20803 0.06795 0.11968 0.10446 0.15314 0.18357 0.18103 0.14300 0.16937 0.13387 0.03854 0.04817 0.05375 0.04108 0.12982 0.19016 0.22110 0.19118 0.20183 0.19473 0.08618 0.11759 0.10592 0.15978 0.16427 0.17370 0.15395 0.16338 0.14183 0.04039 0.04309 0.03411 0.03411 0.12971 0.16831 0.19031 0.16697 0.18986 0.18312 0.08789 0.11415 0.09914 0.15863 0.17310 0.17899 0.15863 0.16720 0.14791 0.04180 0.04341 0.03430 0.03591 0.13773 0.17738 0.21329 0.17899 0.20632 0.17631 0.08798 0.12110 0.09745 0.14901 0.16272 0.17077 0.15374 0.16651 0.12867 0.04352 0.04683 0.05676 0.03642 0.13387 0.17881 0.19631 0.17833 0.19536 0.18921 0.09040 0.11970 0.09040 0.14747 0.16818 0.16515 0.14646 0.16566 0.12727 0.04040 0.04495 0.03636 0.03939 0.13990 0.17929 0.20202 0.16768 0.18990 0.16515 0.08702 0.11771 0.09557 0.15996 0.17958 0.17354 0.16348 0.15091 0.14085 0.04628 0.04477 0.05231 0.03471 0.13732 0.18159 0.19618 0.16298 0.19567 0.16600 0.07566 0.12390 0.09430 0.15954 0.17708 0.17434 0.15625 0.17544 0.14035 0.05482 0.04276 0.02851 0.05044 0.15186 0.18695 0.19682 0.18586 0.19353 0.17982 0.07365 0.12625 0.10371 0.16683 0.17285 0.16984 0.16182 0.18136 0.13327 0.03507 0.04860 0.05411 0.04810 0.14729 0.19389 0.20140 0.17786 0.19088 0.17936 0.08390 0.13220 0.10537 0.16244 0.17220 0.18000 0.15707 0.18244 0.15220 0.04000 0.03707 0.03707 0.02976 0.14098 0.19171 0.19659 0.17366 0.18341 0.17463 0.08400 0.12223 0.09406 0.15342 0.16449 0.17555 0.15191 0.17606 0.16398 0.04125 0.03974 0.03722 0.03421 0.13984 0.20171 0.20926 0.17505 0.20322 0.17254 0.11044 0.15746 0.13350 0.13016 0.14169 0.14169 0.15140 0.11408 0.09405 0.03883 0.02609 0.03580 0.02458 0.15807 0.16141 0.19660 0.14381 0.19357 0.15018 0.12796 0.11523 0.13211 0.14514 0.15107 0.15225 0.14751 0.11019 0.09656 0.04384 0.02992 0.03258 0.02814 0.18276 0.17773 0.19639 0.15758 0.19017 0.14425 0.11489 0.12572 0.13049 0.14290 0.15691 0.16104 0.14640 0.10821 0.10057 0.04519 0.03183 0.04583 0.02769 0.17791 0.18109 0.19414 0.16264 0.19192 0.15500 0.11336 0.12107 0.13905 0.14804 0.16538 0.15992 0.14933 0.10469 0.09762 0.03982 0.02826 0.03276 0.02441 0.17180 0.18080 0.19717 0.16346 0.20424 0.15864 0.11616 0.12092 0.13874 0.14379 0.15478 0.15716 0.14528 0.10279 0.09329 0.03862 0.02971 0.04219 0.02733 0.17766 0.17172 0.19608 0.16043 0.19340 0.15062 0.11128 0.12232 0.14746 0.14378 0.15420 0.15788 0.14991 0.09994 0.09871 0.05089 0.02790 0.03495 0.02483 0.16953 0.17566 0.18670 0.15328 0.18424 0.15021 0.14432 0.15533 0.15083 0.13257 0.13607 0.13907 0.13157 0.08554 0.10355 0.03702 0.02701 0.04202 0.02376 0.15683 0.14407 0.19035 0.13132 0.18484 0.13482 0.11060 0.12447 0.13834 0.14118 0.15718 0.15754 0.15007 0.09459 0.09388 0.04836 0.03485 0.05974 0.02418 0.17212 0.17141 0.19772 0.13300 0.17212 0.13656 0.12111 0.12391 0.13574 0.13481 0.15629 0.15691 0.14352 0.10523 0.09278 0.03923 0.02864 0.03985 0.02522 0.17061 0.17497 0.19614 0.16376 0.19178 0.16065 0.11776 0.10853 0.13908 0.14227 0.15945 0.15977 0.15054 0.10694 0.09994 0.03756 0.02578 0.04010 0.02419 0.17473 0.16996 0.20210 0.14768 0.19574 0.15532 0.10741 0.15979 0.16402 0.13122 0.13545 0.14497 0.13545 0.11958 0.14815 0.04127 0.04180 0.08148 0.03280 0.15503 0.14656 0.18201 0.14709 0.15714 0.14021 0.10838 0.16983 0.16145 0.13687 0.13911 0.14078 0.13855 0.10615 0.13743 0.03687 0.04078 0.07709 0.03296 0.14693 0.14916 0.16592 0.14637 0.14972 0.14469 0.10761 0.17174 0.16848 0.13315 0.14620 0.14783 0.14022 0.11957 0.15326 0.03804 0.04239 0.07500 0.03152 0.15000 0.15761 0.17065 0.16033 0.15543 0.15380 L1 F L L1 F W L1 T1 L L1 T1 W L1 T2 L L1 T2 W L1 TA L L1 TA W L1 PR L L1 PR W L1 CL L L1 CL W L2 C1 L L2 C1 W L2 C2 L L2 C2 W L2 C3 L L2 C3 W L2 F L L2 F W 1.30721 0.17164 1.35199 0.14303 1.08706 0.12687 0.61070 0.08458 0.42662 0.06343 0.15423 0.02488 0.14739 0.15858 0.18595 0.15299 0.15609 0.15920 1.39552 0.16791 1.71805 0.17819 1.83411 0.15416 1.50352 0.13716 0.71571 0.08265 0.45838 0.05510 0.19988 0.03283 0.15064 0.16002 0.21395 0.16530 0.19343 0.16823 1.84994 0.17995 1.67762 0.16551 1.76644 0.14533 1.35352 0.12514 0.65456 0.07670 0.45617 0.05652 0.27624 0.03230 0.16321 0.15629 0.18685 0.17762 0.17705 0.16148 1.84775 0.17474 1.53875 0.12863 1.60549 0.13186 1.27503 0.11195 0.57589 0.07696 0.41927 0.06189 0.15770 0.02906 0.14639 0.16469 0.19483 0.15823 0.16631 0.15231 1.61948 0.11895 1.69114 0.13753 1.75524 0.13403 1.35315 0.11830 0.64977 0.08100 0.36364 0.06469 0.16142 0.02622 0.18473 0.18065 0.15967 0.16026 0.16259 0.16608 1.84848 0.14569 1.87795 0.14205 1.66769 0.12615 1.29692 0.10513 0.39846 0.06769 0.32769 0.05077 0.18513 0.02051 0.14308 0.15436 0.18872 0.15846 0.18051 0.16410 2.02513 0.14462 1.66749 0.12469 1.66005 0.12841 1.24380 0.09926 0.59615 0.07878 0.43921 0.06700 0.17742 0.02481 0.16129 0.15881 0.19975 0.15881 0.17432 0.14764 1.68114 0.13462 1.57075 0.15290 1.56600 0.12726 1.22270 0.11349 0.51282 0.07075 0.36942 0.06030 0.19136 0.02279 0.15432 0.16287 0.18851 0.16714 0.15764 0.15385 1.72745 0.14292 2.00929 0.16339 2.04699 0.13279 1.50055 0.11639 0.61038 0.07158 0.39290 0.06066 0.17486 0.02404 0.17104 0.16831 0.19781 0.17049 0.14208 0.15355 2.25574 0.16175 1.59764 0.18070 1.65825 0.14254 1.33389 0.12290 0.61560 0.08193 0.44949 0.06958 0.16049 0.02918 0.16049 0.17677 0.19921 0.17116 0.19416 0.17508 1.76375 0.18238 1.63128 0.13846 1.63692 0.11282 1.24769 0.09179 0.58051 0.06718 0.42462 0.05897 0.19641 0.02051 0.16103 0.15795 0.19128 0.15487 0.16769 0.15077 1.75590 0.14615 1.66276 0.13490 1.62366 0.12170 1.08993 0.10459 0.43891 0.06647 0.36070 0.06012 0.16325 0.02933 0.16031 0.15249 0.18866 0.15787 0.15445 0.14418 1.80254 0.16471 1.81635 0.16779 1.82452 0.12740 1.43269 0.11154 0.55433 0.06683 0.39279 0.05385 0.18798 0.02548 0.16683 0.15240 0.19615 0.15096 0.17308 0.15962 1.94904 0.16106 1.75127 0.15279 1.76142 0.12589 1.32995 0.11066 0.58985 0.06802 0.41827 0.05482 0.18376 0.01929 0.14670 0.15330 0.20355 0.15787 0.17970 0.14873 1.86497 0.14772 1.53407 0.17282 1.37356 0.15558 1.54762 0.14039 0.71346 0.12028 0.55090 0.07594 0.38834 0.04433 0.17529 0.18103 0.21880 0.16544 0.20443 0.17857 1.68391 0.17570 1.56024 0.11973 1.32304 0.09864 1.17696 0.08057 0.56551 0.06175 0.52786 0.05572 0.35090 0.03916 0.13404 0.15060 0.19277 0.13780 0.18373 0.13102 1.63855 0.13328 2.12328 0.12615 1.75803 0.08830 0.16055 0.13360 0.17603 0.13360 0.17661 0.13647 2.28383 0.12328 2.07386 0.12016 1.74033 0.10727 1.52227 0.08675 0.71395 0.06096 0.59203 0.04572 0.38921 0.02696 0.16706 0.15299 0.19343 0.13247 0.16999 0.13540 2.24560 0.12016 1.12031 0.27538 1.16203 0.23853 1.28999 0.21140 0.50834 0.14186 0.49791 0.10223 0.34423 0.05285 0.16134 0.24965 0.21349 0.23783 0.22879 0.23435 1.22670 0.26912 1.28293 0.18546 1.32929 0.16913 1.45996 0.14910 0.52634 0.09800 0.46101 0.07903 0.34721 0.03477 0.14594 0.19758 0.20443 0.16913 0.19810 0.17492 1.33562 0.18546 1.38012 0.19883 1.44639 0.18226 1.62183 0.15205 0.56238 0.11160 0.47515 0.08041 0.33918 0.03655 0.16179 0.19006 0.21150 0.18665 0.19737 0.18372 1.43129 0.21101 1.32055 0.19274 1.39724 0.19018 1.52454 0.15542 0.51738 0.10327 0.46012 0.08180 0.35634 0.04243 0.13753 0.20501 0.20143 0.17025 0.20194 0.18661 1.37730 0.19274 1.31176 0.19148 1.38889 0.19417 1.55663 0.17476 0.50917 0.11057 0.44175 0.07983 0.32201 0.03506 0.16181 0.20119 0.21197 0.18986 0.19903 0.18932 1.38781 0.20334 1.42762 0.23175 1.42397 0.20864 1.58637 0.18552 0.59854 0.12591 0.52251 0.09185 0.37165 0.04380 0.16119 0.20316 0.23236 0.22445 0.23175 0.21472 1.50365 0.23358 1.39655 0.21653 1.44016 0.19878 1.65264 0.16937 0.56694 0.12373 0.48377 0.08519 0.35294 0.04209 0.15974 0.20081 0.21653 0.20233 0.20030 0.19067 1.46298 0.21907 1.41472 0.17639 1.47172 0.18043 1.66921 0.15530 0.53546 0.10727 0.45826 0.08034 0.31553 0.04309 0.15350 0.19838 0.21724 0.18941 0.22666 0.17819 1.49327 0.20422 1.38585 0.18757 1.41747 0.16988 1.61736 0.15327 0.53591 0.10450 0.47588 0.08360 0.37567 0.04716 0.15113 0.19614 0.19346 0.17149 0.19239 0.17578 1.41104 0.18542 1.37654 0.20104 1.43141 0.17786 1.62110 0.15847 0.53500 0.11164 0.47540 0.07900 0.34201 0.03453 0.14995 0.19111 0.20672 0.18685 0.19962 0.18685 1.44939 0.19205 1.31414 0.18838 1.34646 0.18182 1.53283 0.15404 0.52323 0.10556 0.46313 0.07879 0.35253 0.04091 0.14596 0.19697 0.20303 0.18232 0.19899 0.17323 1.38384 0.17576 1.32596 0.18561 1.39336 0.17555 1.53521 0.14889 0.51811 0.10060 0.43763 0.07495 0.34004 0.03471 0.14336 0.19416 0.20825 0.18410 0.20423 0.17002 1.38481 0.18612 1.31853 0.19901 1.36294 0.18640 1.58114 0.15789 0.51042 0.10965 0.45888 0.08607 0.33607 0.04386 0.16283 0.19901 0.21272 0.18640 0.20833 0.17763 1.38761 0.21436 1.40080 0.18086 1.44589 0.17936 1.60220 0.15681 0.55762 0.10421 0.46092 0.07615 0.37024 0.03457 0.15381 0.18287 0.20992 0.18838 0.19790 0.18337 1.44990 0.19639 1.34878 0.20585 1.38732 0.19415 1.62390 0.17073 0.52537 0.10976 0.47024 0.07854 0.33707 0.04780 0.14780 0.18537 0.20488 0.18341 0.19122 0.18244 1.40000 0.20146 1.39688 0.19165 1.37223 0.16449 1.61922 0.14487 0.53421 0.10211 0.46982 0.07495 0.33451 0.04225 0.14990 0.19316 0.22384 0.18813 0.20926 0.17254 1.46026 0.19366 1.36286 0.14867 1.25303 0.13926 1.37197 0.12197 0.58374 0.08556 0.43750 0.05886 0.24150 0.03944 0.15079 0.15716 0.21420 0.15049 0.20570 0.15170 1.51972 0.15261 1.51600 0.14810 1.41203 0.14514 1.55065 0.13181 0.65551 0.08679 0.48104 0.06398 0.34538 0.03732 0.17684 0.16736 0.19491 0.15818 0.19254 0.15521 1.69964 0.16232 1.58498 0.16900 1.49905 0.15532 1.66423 0.14004 0.69987 0.08784 0.52037 0.06365 0.35964 0.04042 0.18460 0.18810 0.20465 0.16900 0.19096 0.16200 1.74443 0.17759 1.55395 0.17502 1.44059 0.16057 1.59634 0.13359 0.65960 0.07996 0.49550 0.05877 0.32980 0.03308 0.17759 0.18754 0.22062 0.17341 0.19974 0.16956 1.77136 0.17694 1.49049 0.16191 1.38384 0.15775 1.51426 0.13458 0.66043 0.09061 0.50000 0.06328 0.31373 0.03684 0.17974 0.18627 0.20826 0.17172 0.20499 0.16726 1.67053 0.16607 1.47149 0.15757 1.37339 0.15083 1.52544 0.13611 0.65113 0.08645 0.46842 0.06499 0.33538 0.03372 0.25383 0.17198 0.18792 0.15328 0.15757 0.14899 1.65359 0.16309 1.49425 0.14757 1.45073 0.14007 1.60380 0.12256 0.63907 0.08029 0.45298 0.06128 0.35718 0.03452 0.15983 0.15183 0.20435 0.14457 0.18559 0.13732 1.68584 0.14507 1.30583 0.14083 1.07326 0.12304 1.12304 0.10882 0.42674 0.08393 0.36700 0.05619 0.33073 0.03556 0.16856 0.17888 0.19559 0.15932 0.19950 0.15647 1.52845 0.16074 1.50872 0.16252 1.38294 0.16252 1.61270 0.15318 0.68182 0.09963 0.48568 0.06787 0.34060 0.03861 0.17186 0.17684 0.19988 0.16812 0.21139 0.16189 1.65785 0.17217 1.47549 0.15436 1.38192 0.15309 1.50700 0.13367 0.64863 0.08689 0.49427 0.07097 0.37842 0.04042 0.16009 0.16677 0.20560 0.15245 0.22183 0.16264 1.58912 0.16932 1.81534 0.16349 1.83810 0.13386 1.46878 0.11852 0.74974 0.07407 0.48148 0.06720 0.15608 0.02011 0.16667 0.16085 0.18730 0.16349 0.18042 0.15397 1.93704 0.15608 1.43743 0.16425 1.50279 0.14581 1.22402 0.11899 0.63017 0.07542 0.42737 0.05810 0.13743 0.02067 0.15307 0.15140 0.16816 0.15531 0.16257 0.14246 1.54413 0.16704 1.70978 0.16141 1.79402 0.14511 1.46250 0.12120 0.61087 0.06957 0.48587 0.06196 0.14891 0.02283 0.15380 0.16467 0.17826 0.16739 0.17283 0.15978 1.84457 0.16467 L2 T1 L L2 T1 W L2 T2 L L2 T2 W L2 TA L L2 TA W L2 PR L L2 PR W L2 CL L L2 CL W L3 C1 L L3 C1 W L3 C2 L L3 C2 W L3 C3 L L3 C3 W L3 F L L3 F W L3 T1 L L3 T1 W 1.43595 0.14925 1.18657 0.12562 0.58147 0.08458 0.44714 0.07214 0.17786 0.02985 0.14739 0.15983 0.17786 0.14925 0.16729 0.16604 1.38495 0.16356 1.41107 0.15299 1.91559 0.16471 1.55744 0.12661 0.73388 0.08324 0.49941 0.05862 0.21102 0.03341 0.14478 0.16940 0.19637 0.16530 0.17819 0.16589 1.77198 0.19871 1.87808 0.17878 1.82526 0.15052 1.41522 0.12918 0.64187 0.07785 0.45098 0.06055 0.32180 0.02134 0.14533 0.15398 0.18108 0.15859 0.17705 0.15225 1.79412 0.17301 1.80911 0.16321 1.76103 0.12594 1.38590 0.11679 0.65016 0.08288 0.40205 0.06566 0.20721 0.05274 0.13886 0.17761 0.20506 0.15501 0.17438 0.15447 1.64693 0.14047 1.72282 0.13240 1.85664 0.13112 1.45629 0.12121 0.63986 0.07867 0.42308 0.05944 0.14510 0.03380 0.17541 0.16783 0.19930 0.15559 0.18823 0.14860 1.82984 0.15618 1.86247 0.13228 1.99385 0.13282 1.45538 0.12051 0.59385 0.06154 0.40974 0.05179 0.21436 0.02410 0.14154 0.14154 0.16769 0.12872 0.14923 0.13744 1.59282 0.13128 1.57128 0.11641 1.63151 0.13586 1.20968 0.12779 0.53040 0.06203 0.39702 0.05955 0.22705 0.02419 0.16501 0.15943 0.21030 0.14950 0.15819 0.14578 1.79156 0.13213 1.60546 0.10918 1.73599 0.13200 1.25736 0.10541 0.56315 0.07123 0.39601 0.06030 0.18851 0.02896 0.13295 0.15385 0.20418 0.15290 0.16239 0.14340 1.69706 0.14767 1.68281 0.12298 2.14973 0.14098 1.57268 0.11475 0.63552 0.06557 0.43825 0.05847 0.16503 0.02131 0.16995 0.16066 0.20710 0.15246 0.17978 0.15519 2.17213 0.15355 2.15082 0.13497 1.84175 0.13524 1.44444 0.13356 0.64366 0.08698 0.45623 0.08025 0.19529 0.03030 0.17059 0.17508 0.19304 0.15095 0.15713 0.16105 1.73176 0.18126 1.83277 0.15376 1.75846 0.11385 1.31026 0.08821 0.58051 0.06513 0.40974 0.04615 0.17590 0.01846 0.13538 0.14308 0.20000 0.13795 0.17026 0.14103 1.71487 0.15231 1.69231 0.12615 1.73363 0.12072 1.26246 0.11144 0.53519 0.06354 0.36510 0.05621 0.18719 0.02688 0.14418 0.15152 0.18084 0.14076 0.14858 0.14858 1.78543 0.15200 1.76344 0.12561 1.97837 0.13365 1.50577 0.11587 0.61635 0.07404 0.40096 0.04952 0.16827 0.02500 0.15337 0.15433 0.17163 0.13990 0.14808 0.16058 1.91106 0.16298 1.94519 0.14231 1.84924 0.12741 1.03807 0.11015 0.51929 0.06497 0.31827 0.05127 0.15178 0.01777 0.15381 0.15990 0.20558 0.16447 0.17513 0.15381 1.83655 0.16447 1.81371 0.13147 1.53325 0.15517 1.68637 0.14614 0.73276 0.11535 0.56938 0.07471 0.41256 0.04557 0.18021 0.19171 0.22537 0.18966 0.20895 0.17570 1.56897 0.17406 1.42898 0.14614 1.39533 0.09789 1.24322 0.09488 0.56175 0.06175 0.52560 0.05196 0.40211 0.02711 0.13630 0.15437 0.19202 0.13630 0.18599 0.13554 1.57756 0.12349 1.37500 0.10467 1.89679 0.08945 1.71216 0.08028 0.18062 0.14048 0.19037 0.13131 0.17431 0.12959 2.18177 0.12213 1.88073 0.09633 1.89801 0.09613 1.62016 0.08324 0.76202 0.05979 0.60844 0.04162 0.43552 0.02872 0.17233 0.15416 0.20223 0.13482 0.17585 0.13247 2.17351 0.12720 1.81536 0.10492 1.26287 0.23435 1.42629 0.20932 0.52364 0.14256 0.51113 0.10362 0.32406 0.05076 0.16412 0.25382 0.22253 0.24131 0.21905 0.23922 1.16968 0.26982 1.21419 0.24339 1.35142 0.17123 1.55796 0.15543 0.51264 0.09431 0.46154 0.07903 0.37671 0.03530 0.14542 0.19968 0.20811 0.17071 0.19810 0.17597 1.29610 0.18915 1.29399 0.17966 1.46832 0.18129 1.70224 0.16131 0.56238 0.12086 0.50195 0.07895 0.35088 0.04045 0.14961 0.19298 0.20858 0.18762 0.20078 0.18470 1.39327 0.20614 1.43519 0.18762 1.43916 0.18814 1.62986 0.16002 0.51176 0.10429 0.46063 0.08487 0.36196 0.03783 0.15440 0.19530 0.20552 0.17996 0.18916 0.17996 1.33231 0.20552 1.37935 0.18507 1.41046 0.20334 1.64239 0.17476 0.53290 0.10787 0.45631 0.07983 0.32848 0.03128 0.15858 0.20119 0.20442 0.18393 0.19525 0.19417 1.32902 0.19903 1.36677 0.18177 1.49027 0.22019 1.70499 0.18309 0.56144 0.13564 0.49513 0.09307 0.36131 0.04440 0.14051 0.22445 0.21959 0.20803 0.21776 0.20925 1.42822 0.20803 1.49239 0.20842 1.74239 0.18205 0.53550 0.12373 0.49797 0.08722 0.38134 0.03753 0.14706 0.21450 0.21045 0.19828 0.20183 0.19675 1.41734 0.21957 1.42901 0.19473 1.49596 0.18582 1.77873 0.16203 0.54219 0.10952 0.47352 0.07989 0.30431 0.04803 0.14363 0.19479 0.19614 0.17415 0.19300 0.17998 1.43851 0.19479 1.46544 0.18268 1.46463 0.18382 1.64416 0.15113 0.53966 0.10129 0.46141 0.07610 0.34620 0.03912 0.14791 0.19775 0.20204 0.18542 0.18917 0.17738 1.37406 0.18382 1.40086 0.18274 1.47777 0.18780 1.74409 0.16746 0.55251 0.10265 0.47162 0.07947 0.34910 0.04210 0.14238 0.19631 0.20814 0.19016 0.19820 0.18117 1.40066 0.20672 1.42148 0.18117 1.38990 0.17323 1.62374 0.14798 0.53737 0.10303 0.45051 0.08232 0.34141 0.04040 0.14343 0.19040 0.20404 0.18535 0.19343 0.17980 1.35051 0.19141 1.33535 0.17172 1.43109 0.17555 1.58199 0.14588 0.51911 0.09809 0.44014 0.07445 0.34306 0.04376 0.14034 0.19718 0.20221 0.17505 0.19668 0.17656 1.34457 0.18712 1.38934 0.17203 1.43640 0.19792 1.63542 0.16502 0.53783 0.10965 0.49013 0.08443 0.34704 0.03783 0.15515 0.20395 0.21765 0.18476 0.20066 0.18421 1.35471 0.20888 1.35526 0.18531 1.49900 0.17385 1.71794 0.16333 0.55661 0.10571 0.46894 0.07715 0.37375 0.04309 0.15731 0.20391 0.21393 0.17936 0.20842 0.18687 1.41784 0.19138 1.44639 0.17886 1.46293 0.18049 1.72000 0.16293 0.52537 0.10049 0.48439 0.07122 0.34829 0.03659 0.15756 0.19122 0.19610 0.18439 0.19463 0.17805 1.38146 0.20341 1.37268 0.17220 1.43109 0.18008 1.73340 0.15493 0.55584 0.10312 0.47686 0.07646 0.32193 0.04728 0.13984 0.19467 0.22183 0.17254 0.20272 0.17505 1.40895 0.19517 1.38129 0.18763 1.40018 0.13896 1.56584 0.13258 0.61590 0.08556 0.45965 0.05856 0.28246 0.03307 0.14563 0.15777 0.22542 0.15079 0.19721 0.14927 1.41110 0.15200 1.32069 0.13865 1.58294 0.14040 1.76126 0.13389 0.65640 0.09360 0.48934 0.07050 0.35130 0.03495 0.16469 0.18543 0.20616 0.16706 0.19461 0.15107 1.58472 0.16736 1.47097 0.15373 1.70274 0.15309 1.83514 0.14227 0.74570 0.10185 0.52610 0.06588 0.36474 0.03310 0.17600 0.19160 0.20274 0.16868 0.18555 0.15945 1.61299 0.17568 1.50509 0.16295 1.65254 0.16153 1.83365 0.14258 0.68144 0.10244 0.52858 0.07129 0.34746 0.02762 0.17662 0.20071 0.22575 0.17437 0.21291 0.16827 1.64611 0.18337 1.53211 0.16217 1.56447 0.15449 1.70410 0.14023 0.68271 0.09180 0.48782 0.06566 0.32056 0.03565 0.16637 0.18954 0.21955 0.17558 0.19548 0.16607 1.58467 0.18360 1.42929 0.16459 1.55242 0.15573 1.71858 0.14408 0.67505 0.09626 0.47639 0.06315 0.34764 0.03617 0.16125 0.18026 0.20386 0.15727 0.18363 0.15604 1.50307 0.16432 1.40466 0.16248 1.61281 0.13957 1.76938 0.13332 0.68634 0.08654 0.47374 0.06003 0.33617 0.03902 0.16308 0.15508 0.20685 0.14157 0.18734 0.13782 1.53802 0.14982 1.45523 0.14307 1.43741 0.14936 1.62198 0.13158 0.65078 0.09246 0.48791 0.06899 0.35420 0.03485 0.16252 0.17568 0.20021 0.15114 0.18777 0.15114 1.38834 0.15114 1.28912 0.14154 1.56476 0.15753 1.76743 0.13885 0.68804 0.09900 0.51183 0.06756 0.34714 0.03643 0.16781 0.18680 0.20299 0.16968 0.19801 0.16158 1.58001 0.17933 1.45081 0.16065 1.52387 0.14863 1.66423 0.13813 0.65850 0.09166 0.49204 0.07002 0.35264 0.03692 0.16423 0.18109 0.21388 0.16200 0.19669 0.15372 1.48536 0.16836 1.38288 0.15913 1.97937 0.14550 1.58466 0.12963 0.75608 0.07884 0.48148 0.06508 0.15873 0.02275 0.14603 0.16667 0.17143 0.15873 0.14656 0.15450 1.88730 0.16243 1.84815 0.13862 1.57765 0.14860 1.25698 0.11899 0.63464 0.07709 0.42737 0.06425 0.14804 0.02123 0.14190 0.15028 0.16816 0.14469 0.15642 0.14525 1.48268 0.16760 1.55084 0.13966 1.88478 0.14728 1.44457 0.12120 0.71739 0.07717 0.45815 0.06630 0.16033 0.02500 0.14293 0.15924 0.16304 0.15870 0.15109 0.15543 1.78043 0.16250 1.82772 0.15000 L3 T2 L L3 T2 W L3 TA L L3 TA W L3 PR L L3 PR W L3 CL L L3 CL W L4 C1 L L4 C1 W L4 C2 L L4 C2 W L4 C3 L L4 C3 W L4 F L L4 F W L4 T1 L L4 T1 W L4 T2 L L4 T2 W 1.16356 0.12873 0.61132 0.08458 0.37065 0.05846 0.16729 0.02674 0.14552 0.15609 0.17537 0.16667 0.15236 0.15858 1.30721 0.15547 1.41418 0.15672 1.14801 0.12562 1.51055 0.14244 0.68581 0.08324 0.45487 0.06331 0.17057 0.02638 0.14654 0.15709 0.17351 0.16413 0.14889 0.15709 1.56917 0.15475 1.82239 0.14361 1.44783 0.14068 1.40600 0.11592 0.67589 0.07382 0.40600 0.05479 0.24798 0.02595 0.14764 0.16667 0.17186 0.16955 0.15975 0.15686 1.59631 0.14706 1.63841 0.14129 1.31892 0.11765 1.26103 0.14263 0.60495 0.07266 0.44241 0.06674 0.18999 0.04144 0.17223 0.17008 0.21313 0.16469 0.18138 0.17384 1.51561 0.14047 1.59096 0.13455 1.26803 0.11518 1.46270 0.10781 0.62587 0.08333 0.43765 0.06643 0.15093 0.02972 0.18473 0.18531 0.20338 0.17308 0.17249 0.16142 1.70280 0.15035 1.73427 0.12413 1.35956 0.11713 1.16667 0.09692 0.44308 0.06256 0.34667 0.04615 0.17179 0.02256 0.16923 0.16410 0.21538 0.16821 0.19282 0.15077 1.81231 0.14308 1.80718 0.12974 1.26359 0.11077 1.27978 0.10794 0.58375 0.07878 0.42246 0.05459 0.21588 0.02792 0.16749 0.16377 0.21464 0.16749 0.17494 0.16253 1.61042 0.13524 1.66377 0.12779 1.25310 0.11290 1.21462 0.11443 0.54986 0.07265 0.37369 0.05935 0.18376 0.02849 0.15575 0.15005 0.19658 0.15954 0.17616 0.15907 1.54938 0.14767 1.55413 0.12726 1.17854 0.11206 1.58798 0.11038 0.64590 0.06776 0.45792 0.05410 0.17650 0.02350 0.17158 0.15519 0.19508 0.16066 0.18415 0.15355 2.00055 0.15956 1.98743 0.12404 1.48689 0.09891 1.44444 0.13861 0.63187 0.09035 0.48148 0.08305 0.17284 0.02581 0.17172 0.17340 0.20932 0.16723 0.17957 0.17845 1.60550 0.16442 1.74018 0.14871 1.39394 0.12963 1.33026 0.11692 0.58667 0.06564 0.40410 0.05590 0.17282 0.02359 0.12410 0.15077 0.20769 0.13641 0.15538 0.13692 1.60308 0.13744 1.62564 0.09590 1.25282 0.09333 1.27761 0.10459 0.52835 0.06940 0.36461 0.05230 0.19110 0.02102 0.15494 0.15787 0.19453 0.14761 0.15543 0.15005 1.66569 0.14907 1.59384 0.13196 1.22336 0.10704 1.49087 0.11394 0.58798 0.07308 0.39375 0.05625 0.18654 0.02019 0.15817 0.14808 0.19904 0.14856 0.18846 0.16587 1.75962 0.15481 1.82163 0.12452 1.41442 0.11202 1.39645 0.11777 0.59645 0.06954 0.42995 0.05431 0.15787 0.01878 0.16599 0.16751 0.21015 0.16853 0.18528 0.16193 1.68020 0.15178 1.71726 0.11878 1.33553 0.11371 1.60632 0.14039 0.68186 0.10427 0.55296 0.07184 0.40353 0.04023 0.16502 0.19745 0.21264 0.16256 0.18966 0.17365 1.46100 0.15887 1.28448 0.15969 1.44171 0.15189 1.21988 0.09111 0.53840 0.06175 0.51355 0.05196 0.36295 0.03916 0.14458 0.15361 0.19127 0.13855 0.16867 0.13404 1.42620 0.11747 1.26807 0.10090 1.21386 0.08434 1.63647 0.08486 0.73165 0.06307 0.58830 0.04931 0.16686 0.14220 0.18349 0.12844 0.17087 0.12271 1.96330 0.11067 1.71101 0.08830 1.50975 0.07913 0.15592 0.15475 0.19109 0.12896 0.18288 0.13716 1.93787 0.13130 1.67995 0.08910 1.49238 0.07972 1.41029 0.21210 0.53547 0.13908 0.51739 0.10848 0.37900 0.05772 0.16342 0.23157 0.24965 0.23922 0.23227 0.23644 1.14256 0.22670 1.18359 0.23783 1.33241 0.20584 1.50948 0.14226 0.51054 0.10590 0.47155 0.07903 0.37987 0.03688 0.13119 0.18599 0.20969 0.19231 0.20074 0.18388 1.20548 0.18230 1.23340 0.16228 1.39779 0.14173 1.64815 0.16082 0.55653 0.10916 0.49708 0.07992 0.34698 0.04386 0.15400 0.18908 0.20175 0.17251 0.18616 0.17495 1.29483 0.19737 1.31433 0.18226 1.55556 0.15400 1.56851 0.15337 0.47035 0.10736 0.46115 0.08384 0.38957 0.04397 0.13139 0.18814 0.20245 0.17996 0.19223 0.18558 1.23926 0.19683 1.28016 0.17689 1.47239 0.14519 1.58468 0.16127 0.52643 0.10248 0.45793 0.07551 0.34520 0.03236 0.14617 0.20334 0.20874 0.18339 0.19849 0.19040 1.24757 0.21090 1.29935 0.18339 1.54746 0.17152 0.16058 0.21715 0.21655 0.22263 0.22689 0.21350 1.36071 0.23479 1.41363 0.21350 1.58273 0.18917 1.67951 0.17444 0.53955 0.12170 0.48631 0.08773 0.33063 0.04970 0.15619 0.19980 0.20690 0.19016 0.19422 0.19473 1.31846 0.20132 1.36156 0.20183 1.61511 0.18053 1.64093 0.15709 0.53052 0.10592 0.46275 0.07765 0.32585 0.04533 0.14497 0.19031 0.19749 0.17684 0.19434 0.18223 1.34650 0.19120 1.38959 0.17684 1.64363 0.14856 1.66077 0.15648 0.54019 0.11147 0.48017 0.07985 0.34512 0.04180 0.14416 0.19346 0.20096 0.17203 0.19132 0.17524 1.26581 0.19346 1.30868 0.17203 1.55520 0.14630 1.70104 0.15894 0.54494 0.10360 0.47114 0.07947 0.35951 0.03595 0.14428 0.18685 0.19962 0.18401 0.19536 0.19347 1.28430 0.20151 1.29044 0.17077 1.55156 0.14664 1.55808 0.15253 0.51010 0.10859 0.45303 0.08030 0.35455 0.03838 0.15000 0.18788 0.21919 0.18636 0.19343 0.18838 1.23232 0.18687 1.26616 0.16970 1.49394 0.14646 1.59708 0.15996 0.51962 0.09909 0.43310 0.07243 0.35412 0.04175 0.13581 0.18209 0.20272 0.19266 0.19819 0.18913 1.26006 0.18511 1.31891 0.17455 1.51811 0.14588 1.61513 0.16118 0.53618 0.10417 0.49232 0.08443 0.41447 0.04496 0.15241 0.19737 0.19627 0.17763 0.19353 0.18092 1.23794 0.19682 1.24945 0.18695 1.48300 0.15899 1.67285 0.15932 0.53407 0.09920 0.47896 0.07265 0.38477 0.03707 0.14679 0.18838 0.20090 0.18637 0.19689 0.18838 1.30110 0.19790 1.35170 0.17585 1.61523 0.15431 1.61610 0.16000 0.53171 0.09951 0.47561 0.07756 0.36683 0.03805 0.14537 0.18878 0.18927 0.17707 0.19073 0.18585 1.29415 0.20244 1.32585 0.17951 1.57073 0.15805 1.64688 0.15191 0.53421 0.10211 0.47837 0.07746 0.34457 0.03722 0.13481 0.19517 0.21831 0.17907 0.20523 0.17505 1.28471 0.18511 1.28370 0.17103 1.57344 0.14185 1.44630 0.12925 0.58920 0.10194 0.44478 0.06796 0.29005 0.04612 0.14624 0.16505 0.21966 0.15716 0.20813 0.16141 1.32403 0.14260 1.22027 0.14078 1.33556 0.11954 1.59568 0.14100 0.61582 0.08945 0.45735 0.06132 0.32761 0.03880 0.16055 0.17891 0.20379 0.17032 0.17299 0.15906 1.46060 0.15729 1.29354 0.14692 1.47749 0.13004 1.64163 0.14322 0.66009 0.09739 0.50668 0.05920 0.32909 0.03883 0.16773 0.18905 0.20115 0.17155 0.17409 0.16932 1.48854 0.17759 1.39115 0.16168 1.55729 0.14258 1.67084 0.14547 0.64355 0.10148 0.51188 0.06872 0.36352 0.03597 0.16538 0.18947 0.21452 0.17052 0.17598 0.17148 1.50610 0.18433 1.38921 0.16635 1.59730 0.14355 1.58289 0.13815 0.63339 0.09239 0.47089 0.06863 0.31937 0.03773 0.16310 0.17855 0.20410 0.16875 0.16310 0.15835 1.47029 0.16934 1.30511 0.15508 1.47891 0.12686 1.52177 0.14286 0.59289 0.09626 0.45248 0.06560 0.35622 0.03250 0.15543 0.16983 0.20938 0.16094 0.18516 0.14776 1.43286 0.15267 1.23053 0.15757 1.43654 0.12569 1.58929 0.12756 0.60755 0.08379 0.44722 0.05928 0.38419 0.03602 0.15333 0.14107 0.19210 0.13807 0.17709 0.13557 1.40670 0.14257 1.34667 0.13607 1.45223 0.11906 1.43172 0.13051 0.57895 0.09566 0.47866 0.06543 0.37055 0.03129 0.16323 0.17034 0.20448 0.13905 0.16927 0.13798 1.26529 0.14260 1.18172 0.12802 1.33570 0.10989 1.61270 0.13761 0.61706 0.09776 0.49004 0.06787 0.31258 0.03736 0.16501 0.17995 0.19801 0.16252 0.19303 0.15006 1.42403 0.15940 1.32752 0.16065 1.47136 0.13138 1.51464 0.13526 0.60216 0.09039 0.47422 0.06493 0.34182 0.03660 0.15532 0.17505 0.18969 0.15245 0.17600 0.15595 1.35105 0.16614 1.24920 0.15468 1.40834 0.12635 1.50000 0.11905 0.73333 0.06772 0.47460 0.05661 0.14550 0.02487 0.15026 0.15873 0.17407 0.15608 0.16667 0.14921 1.79683 0.16667 1.83333 0.14021 1.49101 0.12169 1.22849 0.12179 0.48827 0.07318 0.38883 0.05307 0.14413 0.02458 0.14804 0.15531 0.16648 0.15140 0.15866 0.14693 1.40447 0.15978 1.48156 0.14469 1.18994 0.12235 1.43913 0.13370 0.69239 0.07391 0.47283 0.06087 0.14891 0.02500 0.14511 0.15761 0.16522 0.16576 0.16196 0.15707 1.64348 0.16522 1.74130 0.13587 1.36304 0.12065 L4 TA L L4 TA W L4 PR L L4 PR W L4 CL L L4 CL W 0.54975 0.08333 0.42537 0.06965 0.18532 0.02736 0.71864 0.06213 0.40328 0.04865 0.17585 0.02755 0.64245 0.06690 0.44983 0.05479 0.30392 0.02364 0.63832 0.07374 0.43703 0.05920 0.12110 0.03821 0.60490 0.07634 0.42599 0.05536 0.13869 0.03030 0.45641 0.05949 0.34872 0.04821 0.17385 0.02564 0.59305 0.07816 0.43921 0.05955 0.20099 0.02109 0.51187 0.05793 0.37274 0.05840 0.17901 0.03039 0.60437 0.07104 0.42404 0.05847 0.17541 0.02022 0.60718 0.08642 0.45398 0.07688 0.17845 0.02806 0.55333 0.06923 0.40821 0.05026 0.18821 0.01795 0.52981 0.06305 0.38416 0.05621 0.19795 0.02688 0.58462 0.07260 0.40673 0.05673 0.18750 0.02500 0.57868 0.07107 0.41421 0.05533 0.18579 0.02081 0.62890 0.10755 0.52627 0.07800 0.40066 0.04351 0.52560 0.06175 0.49247 0.05572 0.33886 0.02711 0.64908 0.06250 0.57569 0.04931 0.44839 0.02982 0.68581 0.06448 0.58030 0.04924 0.38101 0.03400 0.49791 0.14743 0.50765 0.10709 0.41586 0.05424 0.49420 0.09378 0.44731 0.07692 0.38356 0.03741 0.52047 0.10673 0.46979 0.07846 0.37476 0.03899 0.45910 0.10941 0.45194 0.08282 0.36094 0.03885 0.48975 0.10787 0.44714 0.07821 0.36624 0.03290 0.38869 0.11375 0.51703 0.08273 0.38321 0.04866 0.53702 0.11562 0.48022 0.09026 0.37272 0.04665 0.51391 0.10592 0.45961 0.07496 0.31598 0.04039 0.51072 0.09646 0.44748 0.07985 0.34084 0.03591 0.52886 0.10407 0.48770 0.07711 0.34626 0.03926 0.51869 0.10202 0.45051 0.08081 0.35101 0.03737 0.49950 0.09909 0.46026 0.07746 0.34658 0.03924 0.48410 0.10746 0.45230 0.08059 0.34320 0.03838 0.52255 0.09619 0.46693 0.07315 0.39279 0.04259 0.50780 0.09610 0.48244 0.07756 0.36780 0.04244 0.47485 0.09859 0.46982 0.06942 0.31489 0.03421 0.54187 0.08738 0.42415 0.06614 0.32888 0.03823 0.57731 0.09449 0.43602 0.05983 0.34182 0.03051 0.61967 0.09644 0.47613 0.06302 0.29981 0.02801 0.60726 0.09602 0.49807 0.06583 0.32017 0.02697 0.58170 0.09239 0.44712 0.06269 0.35680 0.03446 0.57265 0.08829 0.44635 0.06193 0.34212 0.03311 0.56378 0.08054 0.44322 0.06053 0.37569 0.03502 0.54196 0.08073 0.41821 0.06259 0.32717 0.03343 0.56040 0.08780 0.46575 0.06600 0.35181 0.03425 0.56779 0.09039 0.45544 0.06906 0.36919 0.03469 0.70159 0.07090 0.46667 0.05926 0.16085 0.02540 0.57654 0.07151 0.40391 0.05587 0.15363 0.02514 0.63261 0.06957 0.46739 0.06087 0.15109 0.02500 Electronic supplement 3: Statistically significant differences between clades clade1-clade4 clade1-clade5 clade1-clade10 clade2-clade4 clade2-clade5 clade2-clade10 clade2-clade6 clade4-clade5 clade4-clade10 O 5 L< Ocular tubercle W < Basal element W < O 5 L< Ocular tubercle W < Eye size a > Front height a > L1 C1 W < O 1 W< L12 > L2 T2 L < L23 > L1 C1 W < Ocular tubercle H > L1 C2 W < O 10 L < L2 C1 W < O 1 L< L4 PR L < L1 C2 W < Basal element W < L1 C3 W < O 2 W< L4 C3 L < O 10 L < L1 C2 L < L1 C3 L < L1 F W < O 3 W< L4 TA W < O 4 W< L1 C3 W < L1 T1 L > L1 F L > O 9 L < L4 TA L > O 5 L< L1 C3 L < L2 T2 L < L1 T1 L < P7< Lat1 < O 6 W < L1 F W < L3 C3 L < L2 C1 W < P8< Lat2 < P5< L1 F L > L4 T1 L > L2 C2 W < Proboscis W < Lat3 < P6< L1 T1 L < O 1 W< L2 C3 W < Trunk L < Lat4 < P7< L1 T1 L > O 1 L< L2 F W < O 1 L< P8< L12 > O 10 W < L2 F L > O 4 W< P9< L2 C1 W < O 10 L < L2 T1 W < O 5 W< Trunk L < L2 C2 W < O 2 W< L2 T2 W < O 7 W < L2 C2 L < O 2 L< L3 C1 W < P3< L2 C3 W < O 4 W< L3 C2 W < P4< L2 C3 L < O 5 L< L3 C3 W < Proboscis L < L2 F W < O 6 W < L3 F W < Trunk W < L2 F L > O 7 L < L3 F L > L2 T1 W < O 8 W < L4 C1 W < L2 T1 L > O 8 L < L4 C2 W < L2 T2 W < O 9 W < L4 C3 W < L2 T2 L < O 9 L < L4 F W < L23 > P1< L4 F L > L3 C1 W < P6< L4 F T1 < L3 C2 W < P9< L4 F T2 < L3 C3 W < Proboscis W < L4 PR W < L3 C3 L < Trunk L < L4 TA W < L3 F W < Front height a > Lat2 < L3 F L > Lat3 < L34 > O 2 W< L4 C1 W < O 3 W< L4 C1 L > O 5 W< L4 C2 W < O 6 L> L4 C3 W < O 7 W < L4 C3 L < O 8 L < L4 F W < O 9 L < L4 F L > P10> L4 F T1 < Proboscis W < L4 F T2 < Trunk W23 < L4 CL W < Trunk H < L4 CL L < L4 PR W < L4 PR L < L4 TA W < L4 TA L > L4 T1 L > L4 T2 L < Lat1 < Lat2 < Lat3 < Lat4 < O 1 W< O 10 W < O 10 L < O 2 W< O 3 W< O 4 W< O 4 L> O 5 W< O 5 L< O 6 W < O 6 L> O 7 W < O 8 W < O 8 L < O 9 W < O 9 L < P10> P2< P3< P4< Proboscis W < Proboscis L < Trunk W < Trunk W23 < Trunk H < Front height a > clade4-clade6 clade5-clade10 clade5-clade6 clade10-clade6 L23 < Eye size a > L1 C2 L > Ocular tubercle H < O 5 L> Ocular tubercle W > L1 C3 W > L3 C2 L > Ocular tubercle H > L1 C3 L > O 1 W> L1 C1 L < L2 C1 W > O 1 L> L1 C2 W > L2 C2 L > O 10 W > L1 C3 W > L3 C1 W > O 2 W> L1 F W > L3 C2 L > O 5 L> L12 < L3 C3 L > O 7 L > L2 C1 L < L34 < P1> L2 C2 W > L4 C2 W > Proboscis W > L2 C3 W > L4 C2 L > L2 F W > L4 C3 W > L3 C3 W > L4 C3 L > L34 < L4 CL W > L4 C2 W > L4 CL L > L4 C3 W > L4 TA L < L4 C3 L > O 1 W> L4 TA L < O 5 W> O 6 L< O 7 W > P10< O 8 W > P6< P3> P7< P4> P8< Proboscis W > P9< Proboscis L> Trunk W23> Trunk W > Trunk L < Dissertation Lars Dietz

3) Publikation II

Titel: Evidence from morphological and genetic data confirms that Colossendeis tenera Hilton, 1943 (Arthropoda: Pycnogonida), does not belong to the Colossendeis megalonyx Hoek, 1881 complex

Organisms, Diversity and Evolution, 13: 151-162

Hinweise zu Publikation II

• Anteil Planung: 80%

• Anteil experimentelle Durchführung: 75%

• Verfassen des Manuskripts: 85%

Abbildungen, die nicht ausschließlich von mir erstellt wurden

• Fig. 3: REM-Aufnahmen von Johanna Spaak und Meike Seefeldt

60 Org Divers Evol (2013) 13:151–162 DOI 10.1007/s13127-012-0120-4

ORIGINAL ARTICLE

Evidence from morphological and genetic data confirms that Colossendeis tenera Hilton, 1943 (Arthropoda: Pycnogonida), does not belong to the Colossendeis megalonyx Hoek, 1881 complex

Lars Dietz & Franz Krapp & Michel E. Hendrickx & Claudia P. Arango & Kathrin Krabbe & Johanna M. Spaak & Florian Leese

Received: 16 June 2012 /Accepted: 13 December 2012 /Published online: 25 January 2013 # Gesellschaft für Biologische Systematik 2013

Abstract Within the Pycnogonida, genetic studies have (Trudy Instituta Okeanology "P. P. Shirshova", Akademy Nauk revealed that Colossendeis megalonyx Hoek (Challenger SSSR, 103, 230–246, 1975). Colossendeis tenera occurs pre- Report, Zoology, 3(X), 1–167, 1881), consists of a complex of dominantly along the Pacific Coast of North America from the several cryptic or overlooked species. Colossendeis megalonyx Bering Sea to central California. Prominent differences between is a typical Southern Hemisphere species complex distributed these two currently distinct species are found in body proportions primarily on the continental shelves in the Antarctic and and other characters that were interpreted by Turpaeva as a Subantarctic. However, a different Colossendeis species possible case of pedomorphosis induced by deep-sea conditions. with a completely different geographic distribution range, In this study, we tested the hypothesis that Colossendeis tenera Colossendeis tenera Hilton (Journal of Entomology and belongs to the Colossendeis megalonyx complex by analyzing Zoology, Pomona College, Claremont, 35(1), 2–4, 1943), was available and novel sequence data (CO1 and H3) of both considered a subspecies of Colossendeis megalonyx by Turpaeva Colossendeis megalonyx and Colossendeis tenera as well as a similar, apparently closely related species, Colossendeis angusta – Electronic supplementary material The online version of this article Sars (Archiv for Mathematik og Naturvidenskab, 2, 237 271, (doi:10.1007/s13127-012-0120-4) contains supplementary material, 1877). We compared morphometric data and SEM of the ovigera which is available to authorized users. of these species. Our results clearly indicate that Colossendeis L. Dietz : K. Krabbe : J. M. Spaak : F. Leese (*) tenera and Colossendeis angusta arenotapartofthe Department of Animal Ecology, Evolution and Biodiversity, Colossendeis megalonyx complex. A sister-group relationship Ruhr University Bochum, 44801 Bochum, Germany of Colossendeis tenera and Colossendeis angusta is strongly e-mail: [email protected] supported, but Colossendeis tenera is not clearly resolved as F. Krapp monophyletic with respect to Colossendeis angusta. This work Zoologisches Forschungsmuseum Alexander Koenig, highlights the need for further examination of the variation found Adenauerallee 160, in the tenera-angusta clade. It also gives a first hint of the 53113 Bonn, Germany phylogenetic affinities of species within Colossendeis. M. E. Hendrickx Laboratorio de Invertebrados Bentónicos, Keywords Pycnogonida . Colossendeidae . Colossendeis Unidad Académica Mazatlan, ICML, UNAM, P.O. Box 811, tenera . Colossendeis megalonyx . Colossendeis angusta . Mazatlan, Sinaloa 82000, Mexico Ovigera . Integrative taxonomy . Biogeography C. P. Arango Natural Environments Program, Queensland Museum, PO BOX 3300, South Brisbane, QLD 4101, Australia Introduction

K. Krabbe Labor für Abstammungsbegutachtungen, Marie-Curie-Straße 1, Pycnogonids or sea spiders are a group of marine arthropods 53359 Rheinbach, Germany that occur from shallow to deep-sea habitats and from polar 152 L. Dietz et al. to tropical regions. Despite considerable efforts to resolve determined as C. tenera by Turpaeva 1975) (Fig. 1). C. the arthropod tree of life (Regier et al. 2010; Meusemann et tenera has been recorded in depths from 225 m (this publi- al. 2010), it is not clearly resolved whether pycnogonids are cation) to 5,200 m (Turpaeva 1975) and can therefore be the sister group to other Chelicerata or to all other arthro- considered a eurybathic species. pods. The taxonomy and systematics within the pycnogo- While most authors regarded C. tenera as a distinct species, nids have also changed considerably over the last century Turpaeva (1975) classified it as a subspecies of the otherwise (Hedgpeth 1947;Fry1978; Munilla León 1999; Arango Southern Hemisphere species C. megalonyx Hoek, 1881.This 2002; Arango and Wheeler 2007; Bamber 2007). Within synonymization was based on the similar general appearance of the Pycnogonida, the Colossendeidae are one particularly the two species as noted by Hedgpeth (1943)andthelackof species-rich (more than 100 species according to Bamber sufficient diagnostic characters in the denticulate spines of the and El Nagar 2012) and broadly distributed clade. Their ovigers according to Turpaeva’s examinations (Turpaeva 1975). phylogenetic position within the group was controversial Colossendeis angusta Sars, 1877, is a species of interest but recent data add support for their basal position within in the context of C. tenera affinities given the known gen- the Pycnogonida (Dietz et al. 2011). Most of the species in the eral similarities between the species already reported in family are classified in the genus Colossendeis Jarzynsky, Hedgpeth (1943). Another species worth further examina- 1870, which is not clearly monophyletic since the polymerous tion and briefly discussed herein is C. gracilis Hoek, 1881, genera Decolopoda Eights, 1835,andDodecolopoda Calman which is often synonymized with C. angusta (e.g., Fry and & Gordon, 1933, appear to group within Colossendeis Hedgpeth 1969). (Krabbe et al. 2010). Colossendeis megalonyx is a Southern hemisphere circum- Colossendeis tenera Hilton, 1943, was very briefly de- polar eurybathic species complex distributed in depths of 3– scribed from the Washington and Oregon coasts by Hilton, 4,900 m on the continental shelves of Antarctica, the and a more detailed description was provided by Hedgpeth Subantarctic islands, southern South America, South Africa, (1943). Subsequent descriptions confirmed its presence and Madagascar (Munilla and Soler-Membrives 2009; Griffiths along the Pacific Coast of North America from the Bering et al. 2011)(seeFig.1). Recent molecular analyses (Krabbe et Sea (Bowers Bank, Child 1995) to central California al. 2010) have shown that C. megalonyx is a complex of at least (Farallon Islands, Child 1994) (see Fig. 1). Other records six lineages that can be clearly differentiated by mitochondrial of C. tenera are from Honshu (Japan) and the Kuriles sequences and probably represent overlooked species. (Turpaeva 1975)soC. tenera is best characterized as a Nevertheless, this C. megalonyx complex forms a clearly sup- circum-North Pacific species with only very limited occur- ported monophyletic group within the Colossendeidae. rence in the Southern Hemisphere on the South American Examples of species that spread from the Southern Ocean coast (one specimen reported off the Peruvian coast was to other distant geographical regions are known in benthic

Fig. 1 Reported distribution ranges of C. megalonyx (black) from the ScarMARbin database (access data 25.4.2012), C. tenera (white), and the locations of the C. angusta specimens analyzed in this study (gray). Map: Ocean Data View version 4 Genetic and morphological analyses of Colossendeis megalonyx 153 invertebrates (Strugnell et al. 2008). If the inclusion of C. and Cano and López González 2007 for C. megalonyx, Fage tenera within the megalonyx complex is confirmed, it would 1956 for C. angusta and Minnaard and Zamponi 1984 for C. be another example of wide geographic distribution and gracilis). extend distribution of the megalonyx complextothe To compare relative lengths, all values were also expressed Northern hemisphere. as proportions of the trunk length. The relative length of the The aim of this study was to investigate the relationship proboscis and the length of coxae 1-3, the femur, tibia 1, tibia of C. tenera and the megalonyx complex while also exam- 2, propodus, tarsus, and terminal claw were used as variables ining affinities with C. angusta. We analyzed mitochondrial in a principal component analysis (PCA) with the and nuclear sequence data to assess phylogenetic affinities STATISTICA version 10 software (StatSoft Inc.). The lengths of C. tenera, and we also examined morphological data of the three coxae were summed since measurements of the including new SEM data on the morphology of the ovigera single segments were not available for comparison for all of all three species. material. Further morphological traits were not consistently available for all specimens and therefore were not used.

Material and methods Molecular phylogenetic analyses

Morphology We generated cytochrome oxidase 1 (CO1) sequences for the C. tenera specimen from Mexico and for the nine Ethanol-preserved specimens were analyzed under a stereo- Northeast Atlantic C. angusta (GenBank accession numbers microscope (Olympus BX40) and a Zeiss DSM 950 scanning KC462557-KC462566). Other CO1 sequence data used in electron microscope (SEM). Since the ovigeral spine fields are this article had been produced in a previous study from the of particular taxonomic importance (see, e.g., Fry and Bochum’s laboratory (Krabbe et al. 2010). Additionally, we Hedgpeth 1969; Cano and López González 2007), we ana- retrieved relevant Colossendeis CO1 sequences from lyzed ovigeral spine fields in samples from five of the six Genbank (Online Resource 1). Some sequences labeled as Colossendeis megalonyx clades, in two C. tenera specimens, Colossendeis in Genbank were not included as we could not and one C. angusta specimen (Table 1). For the SEM pictures validate the species identification. we first dried samples of the ovigera by adding hexamethyl- We tested the affinities of C. tenera within Colossendeidae disilazane (HMDS) in a rising concentration every 15 min. analyzing a total of 44 CO1 sequences (545 bp) with Bayesian Second, the samples were air-dried underneath an outlet. and maximum likelihood analyses. Rhopalorhynchus Wood- Third, the samples were carbon-glued on cylindrical alumi- Mason, 1873, which is classified as a separate subfamily num blocks and sputter-coated (180 s) with a 99-nm-thick within Colossendeidae by Bamber and El Nagar (2012), was palladium-gold layer and then analyzed under the SEM. used as an outgroup. Multigene Bayesian and ML analyses To study differences in other morphological traits, we were performed with a subset of ten taxa for which the nuclear measured 59 specimens of Antarctic and Subantarctic C. gene histone 3 (H3) data were also available. We produced megalonyx (8 clade A, 15 B, 10 C, 14 D, 12 E, see Online new H3 sequences from representatives of the six C. mega- Resource 4), 2 C. tenera specimens, 1 from British lonyx clades (Krabbe et al. 2010). GenBank accession numb- Columbia and 1 from the Pacific coast of Mexico, and 10 ers: KC456423-KC456506. In these concatenated analyses, C. angusta specimens, 9 from the Northeast Atlantic and 1 Colossendeis macerrima, which grouped in a basal position from the West African coast. Specimens were determined within Colossendeis in the CO1 tree, was used as an outgroup. using published keys and descriptions (Fry and Hedgpeth An analysis using CO1 and H3 as well as the mitochondrial 1969 and Child 1995 for C. megalonyx, Hedgpeth 1943 for ribosomal genes 12S and 16S was also performed (see Online C. tenera,Sars1891 for C. angusta). C. tenera and C. Resource 4). angusta were differentiated mostly by the presence of eyes For the Bayesian analyses, the program MrBayes version and lack of anaxial insertion of the ninth palp article in the 3.1.2 (Huelsenbeck and Ronquist 2001) was used, and the former species. Most of the C. megalonyx material was part appropriate model of evolution was determined with of the molecular study by Krabbe et al. (2010) representing MrModeltest 2.3 (Nylander 2004) using the Akaike infor- five of the six clades described based on mitochondrial mation criterion (AIC). For the multigene analyses, the sequence data (Table 1). We took body, proboscis, and leg model was partitioned by individual gene. The analysis measurements, which are widely used in the literature, i.e., was run for 10,000,000 generations with two independent length of the proboscis, trunk, abdomen, and the individual runs with four chains each and a sample frequency of every articles of the third leg. We compared our measurement data 100th generation. Results were checked for convergence with those reported previously for C. tenera, C. megalonyx, (splits frequencies of the likelihoods < 0.01) and the con- and C. angusta (Hedgpeth 1943 for C. tenera, Calman 1915 sensus tree calculated, discarding the first 25 % of the trees 154

Table 1 Ranges of morphological measurements of C. tenera, C. angusta, and C. megalonyx specimens, including both our own measurements and those from the literature (lengths ration relative to trunk are shown if not listed otherwise). For a full list of the individual measurements, see Online Resource 4

Species Description Trunk Proboscis Abdomen 1st coxa 2nd coxa 3rd coxa Sum of Femur Tibia 1 Tibia 2 Tarsus Propodus Terminal (in mm) coxae claw

C. tenera Male paratype 6.5 1.308 0.123 0.154 0.231 0.231 0.615 2.769 2.308 1.692 0.731 0.538 0.615 C. tenera Female paratype 7 1.393 0.143 0.143 0.179 0.214 0.536 2.857 2.429 1.643 0.571 0.429 0.536 C. tenera (Mexico) 6.8 1.544 0.147 0.206 0.294 0.294 0.794 2.324 2.132 1.618 0.721 0.691 0.765 C. tenera (British 10.366 0.964 0.115 0.130 0.153 0.135 0.418 1.954 1.742 0.728 0.284 0.260 0.291 Columbia) C. angusta (West Africa) 0.863 0.233 0.137 0.137 0.123 0.397 1.918 1.904 1.233 0.521 0.438 0.575 C. angusta (Norwegian Sea) 7.91–9.04 0.906–0.988 0.177–0.217 0.128–0.15 0.113–0.133 0.102–0.122 0.356–0.403 2.053–2.321 1.861–2.15 1.284–1.507 0.468–0.575 0.411–0.417 0.508–0.565 C. angusta (Iceland) 6.43–8.01 0.905–0.966 0.174–0.25 0.15–0.173 0.125–0.173 0.109–0.15 0.384–0.48 2.223–2.435 2.012–2.199 1.362–1.499 0.494–0.595 0.484–0.569 0.602–0.771 C. megalonyx Holotype 11 1.818 0.255 –––0.636 2.068 1.909 1.614 0.932 0.727 0.636 (Calman 1915) C. megalonyx Clade A 9.6–12.9 1.402–1.534 – 0.190–0.208 0.116–0.188 0.116–0.179 0.434–0.563 1.51–2.009 1.51–1.947 1.354–1.86 0.543–0.893 0.527–0.759 0.279–0.411 C. megalonyx Clade B 11.14–13.83 1.439–1.725 – 0.179–0.219 0.11–0.172 0.11–0.16 0.416–0.542 2.052–2.61 1.835–2.314 1.594–1.994 0.86–1.114 0.638–0.804 0.46–0.638 C. megalonyx Clade C 10.76–13.37 1.504–1.864 – 0.198–0.232 0.142–0.204 0.138–0.196 0.491–0.603 1.992–2.536 1.698–2.105 1.419–1.864 0.768–1.06 0.519–0.693 0.403–0.569 C. megalonyx Clade D 11.4–13.65 1.53–1.878 – 0.184–0.236 0.16–0.195 0.147–0.193 0.509–0.607 2.105–2.77 1.851–2.42 1.64–2.187 0.93–1.307 0.724–1.602 0.44–0.579 C. megalonyx Clade E 11.52–13.41 1.704–1.895 – 0.182–0.221 0.154–0.181 0.127–0.166 0.492–0.563 2.088–2.685 2.02–2.434 1.777–2.16 0.978–1.241 0.712–0.883 0.42–0.552 .Deze al. et Dietz L. Genetic and morphological analyses of Colossendeis megalonyx 155 as burn-in. Posterior probabilities were calculated from the and tibia 1 are longer than in C. megalonyx, and tibia 2, the remaining trees. For ML analyses, the program RAxML tarsus, and propodus are shorter. It should be noted that in version 7.0.3 (Stamatakis 2006) was used with the model this specimen, both third legs were broken off, and the distal GTR+G for finding the best tree, and bootstrap analyses articles from tibia 2 on were measured on the fourth pair of were performed using 1,000 fast replicates with the legs, implying that the rather extreme values for this spec- GTR+CAT approximation. The multigene analysis was par- imen may not necessarily be reliable because the fourth leg titioned by gene. is often shorter than the third leg. In the Mexican specimen, coxae 2 and 3 are relatively longer than in C. megalonyx, tibia 2 and the tarsus are shorter, and the claw is longer. In Results all C. tenera and C. angusta specimens, the claw is longer than the propodus, while it is shorter in C. megalonyx. Morphometric comparison of C. tenera and C. megalonyx The specimens we examined agree in most of the distinc- tive characters mentioned by Hedgpeth (1943) and Child The proportions of the proboscis, trunk, abdomen and leg (1994, 1995), such as a seventh palp article not longer than articles in C. tenera analysed by us and from the literature broad, claws as long or longer than the propodus, and a differ considerably from those in C. megalonyx, both according proportionally short abdomen. However, all of them have a to our own and to published data (Table 1, Online Resource 4). rather low ocular tubercle that does not resemble the figure According to the measurements of Hedgpeth (1943), C. of Hedgpeth (1943). There is no indication that the tubercle tenera differs considerably from C. megalonyx. The abso- is broken off, as claimed by Hedgpeth (1943)forthe lute length of the trunk is much shorter in C. tenera than in Washington specimen. Confirming previous reports, in all C. megalonyx. The ratios of proboscis/trunk and abdomen/ specimens of C. tenera we examined the eyes are unpig- trunk are lower in C. tenera than in C. megalonyx. Coxa 1 is mented. The Washington and Oregon specimens agree with proportionally shorter, coxa 3, the femur and tibia 1 are Hedgpeth’s(1943) description in having front eyes some- longer, the tarsus and propodus are again shorter, and the what larger than back ones, but in the specimen from British claw is again longer. The sixth joint of the oviger is longer Columbia, the size difference is much more extreme. than in C. megalonyx. The C. tenera specimen from British Columbia differs Morphometric comparison of C. tenera and C. angusta from all others in its large absolute size, relatively shorter proboscis, and short legs. The legs are relatively shorter than The West African specimen of C. angusta falls outside the in any other specimen of C. tenera, C. angusta/gracilis,or range of variation of the specimens from the Northeast C. megalonyx known to us. The short proboscis is compa- Atlantic. Relative to trunk length, the proboscis, femur and rable to the West African C. angusta specimens. In general tibia 2, and the total leg length are shorter (Table 1, Online shape, the proboscis agrees with the illustration of Hedgpeth Resource 4). Other measurements fall into the range of (1943). Due to the short legs, the proportions of some leg variation for the North Atlantic specimens. Comparing leg articles are more similar to C. megalonyx, but the short coxa articles relative to total leg length, the femur is slightly 1, tibia 2, tarsus, propodus, and claw fall outside its range of shorter and the claw is longer in the West African specimen. variation, as do the proboscis and abdomen. All measured C. tenera specimens have a shorter abdo- The Mexican C. tenera specimen has proportions some- men than C. angusta, and all except the British Columbia what more similar to C. megalonyx than those described by specimen have a shorter proboscis. In the C. tenera speci- Hedgpeth (1943). Only the short abdomen, the long coxae 2 mens reported by Hedgpeth (1943), all leg articles except for and 3, the short tarsus and the long claw fall outside the coxa 1, the propodus, claw, and tarsus of the female speci- range of variation for C. megalonyx (Table 1). The propor- men are longer than in C. angusta relative to trunk length. tions therefore differ noticeably from those of the northern Relative to total leg length, only coxa 3 is consistently longer C. tenera specimens. The proboscis is longer and has a very in both C. tenera specimens than in C. angusta.Inthe different shape, being strongly inflated distally and more Mexico specimen, all three coxae, tibia 2, the tarsus, and similar to species such as C. scotti Calman, 1915, than to propodus are longer than in C. angusta relative to trunk other specimens of C. tenera or C. megalonyx. length. Relative to total leg length, all three coxae, the tarsus, When length of leg articles are expressed as percentage and propodus are longer, and the femur and tibia 1 are shorter of the total leg length, coxa 1 is shorter in C. tenera than in than in C. angusta. In the British Columbia specimen, all leg C. megalonyx, coxa 3, the femur and tibia 1 are longer, and articles except for the three coxae are shorter than in C. tibia 2, the tarsus, and propodus are again shorter. The claw/ angusta relative to trunk length. Relative to total leg length, propodus ratio is higher in C. tenera than in C. megalonyx. all three coxae, the femur and tibia 1 are longer, and tibia 2, In the British Columbia specimen, coxa 2 and 3, the femur, the tarsus, propodus, and claw are shorter. 156 L. Dietz et al.

In the C. angusta specimen recorded by Fage (1956), the the paratypes cluster together in the upper right sector of the proboscis, abdomen, femur, and tibia 1 and 2 are longer than PCA, whereas the specimen from British Columbia falls with- in the specimens measured by us. Except for the abdomen, in the opposite sector, with no similarities to the other speci- this shows greater agreement with the C. tenera paratypes, mens analyzed. It fits closer to the C. angusta specimens, although tibia 1 and 2 are even longer. In the C. gracilis however, which all form one dense group in the plot. specimen described by Minnaard and Zamponi (1984), the proboscis is much longer than in C. tenera or our C. angusta SEM analyses of ovigera C. tenera and C. megalonyx specimens, but the short abdomen and long leg articles agree more with C. tenera, although coxa 1 is longer but femur Analyses of the SEM pictures showed that the ovigeral and tibia 1 are somewhat shorter. spine configuration of C. tenera (Fig. 3a,b,f)isbroadly similar to C. megalonyx (Fig. 3c-e) in having one row of Principal component analyses (PCA) long spines, one row of medium spines, and several rows of short spines, which are shaped rather similarly in both In a PCA using the proportions of the leg segments relative species. Our SEM data of C. megalonyx also agree well to the trunk (Fig. 2), factor axis 1 represents 46.56 % of the with those of Cano and López González (2007). However, total variation. It was primarily determined by the variables even though ovigera from different C. megalonyx clades tibia 2 and tarsus and with almost equal proportions by the show consistent differences (Spaak et al. unpublished data), variables proboscis, propodus, and first coxa (see Online they are much more similar to each other than to those of C. Resource 3 for factor loadings and eigenvalues). The termi- tenera. In the examined specimens of C. megalonyx, the nal claw had the smallest loading for factor 1. Factor axis 2, short spines of the strigilis are very densely placed in three representing 21.14 % of the total variance, was primarily to nine rows, which are not clearly segregated from each determined by the variables terminal claw, femur and tibia 1. other, while those of C. tenera are much less numerous and Together, factors 1 and 2 explain 67.7 % of the total vari- are placed in only two clearly segregated rows in the ance. Factor 3 adds an additional 15.35 % (based almost Mexican specimen (Fig. 3a,b) or in about two irregular rows completely on coxae 2 and 3) but brings no further resolu- in the British Columbia specimen (Fig. 3f). As far as we tion to the data. The PCA based on factors 1 and 2 shows a could discern the morphology of the Washington and clear distinction between the C. megalonyx specimens and all Oregon specimens from our light-microscopic observations, other specimens. The C. tenera specimen from Mexico and they are more similar to the British Columbia specimen. The original description of C. megalonyx by Hoek (1881) shows only two to three rows of short spines, but the spines are

4 densely spaced, unlike in C. tenera. Oregon Oregon

Mexico SEM analyses of ovigera C. tenera and C. angusta

2 The ovigeral spine fields of North Atlantic C. angusta (Fig. 3g, h) specimens are rather different from those of both C. tenera (Fig. 3b,f) and C. megalonyx (Fig. 3c-e). In the Norwegian Sea specimens, the spines of the medium 0

Brit. Columb. row are much less closely spaced than in the other species,

Factor 2 (21.14%) and the short spines are also rather sparse and placed in about three to four irregular rows, with spines gradually getting smaller endally (Fig. 3g). In contrast to other speci- -2 mens we examined, the spines are oriented orthogonally to C. tenera the length of the oviger article. Medium and short spines are C. megalonyx C. angusta rather similar, but can still be clearly distinguished by their orientation and the arrangement of medium spines in one -6 -4 -2 0 2 4 regular row. In the specimens from the Icelandic coast, there Factor 1 (46.56%) is no clear distinction between medium and short spines. Fig. 2 Results of the principal component analyses for specimens of The spines are rather sparse, become smaller endally, are Colossendeis tenera, C. angusta, and C. megalonyx. PCA scores for oriented regularly, and placed into four to five irregular rows the first two factors with the highest eigenvalues are plotted on both (Fig. 3h). axes. Analysis was based on the different leg segments and the pro- boscis length relative to the length of the trunk. See Online Resource 3 The ovigera of C. gracilis figured by Hoek (1881) agree for further details much better with C. tenera, having a clearly distinguished Genetic and morphological analyses of Colossendeis megalonyx 157

Fig. 3 Scanning electron microscopic pictures of ovigeral spine fields in Colossendeis tenera from Mexico (a,b), of different Colossendeis megalonyx specimens (c–e), of Colossendeis tenera from British Columbia (f), and of Colossendeis angusta from Iceland (g,h)

row of densely placed medium spines and one or two sister to the C. megalonyx complexinanyanalyses. irregular rows of sparsely placed short spines. CO1 and multigene analyses also generally agreed on the phylogeny of Colossendeidae, including the group- Molecular phylogenetic analyses ing of the ten-legged Decolopoda within the genus Colossendeis, as previously found by Krabbe et al. For the CO1 data set, the model HKY+I+G was specified by (2010). The H3 sequences of C. megalonyx generally MrModeltest, while GTR+I+G was specified for the multi- confirmed the phylogeny resulting from CO1 sequences gene data set (Fig. 4). (Krabbe et al. 2010), although they did not distinguish The results of the both ML and Bayesian analyses the sister clades A/F, B/C, and D/E. clearly showed that C. tenera is not a member of the C. The genetic distances between the two C. tenera megalonyx complex, agreeing with the morphological specimens whose CO1 sequences were included in this data presented in this article. Monophyly of the mega- analysis were remarkably high (9 % uncorrected dis- lonyx complex excluding C. tenera was highly sup- tance), possibly indicating that C. tenera is itself part ported in all analyses. A clade consisting of C. tenera of a species complex consisting of at least two species, and the morphologically similar C. angusta was strongly which would also include C. angusta andpossiblyother supported in the analyses of the CO1 data set (BS = 95, species. We also found clear differentiation within C. PP = 1). Remarkably, C. tenera itself was not resolved angusta, with the specimens from near the Icelandic as monophyletic, and the sequence from British coast forming one closely related group with ca. 3 % Columbia grouped closer to C. angusta than the one genetic distance to another group consisting of the from Mexico but with poor support (BS = 75, PP = Norwegian Sea specimens and one of the two from 0.69). While the position of the C. tenera/angusta clade Newfoundland. Both groups appeared to be genetically was poorly resolved in all analyses, it did not group as mostly homogenous, with individual sequences differing 158 L. Dietz et al.

73 / 0.77 GQ387024 Colossendeis megalonyx GQ387025 Colossendeis megalonyx GQ387021 Colossendeis megalonyx GQ387023 Colossendeis megalonyx 60 / 0.94 GQ387020 Colossendeis megalonyx 63 / 0.88 GQ387022 Colossendeis megalonyx 99 85 / GQ387019 Colossendeis megalonyx 0.99 / 1 GQ387018 Colossendeis megalonyx Colossendeis megalonyx D 94 / 1 GQ387026 Colossendeis megalonyx

93 / 0.86 Colossendeis megalonyx E 67 / 0.95 GQ387014 Colossendeis megalonyx 76 / 62 / 0.87 Colossendeis megalonyx C 0.94 GQ387015 Colossendeis megalonyx 100 / 1 GQ387013 Colossendeis megalonyx 100 / 1 Colossendeis megalonyx B GQ387012 Colossendeis megalonyx 96 Colossendeis megalonyx F GQ387010 Colossendeis megalonyx / 1 complex 100 / 1 Colossendeis megalonyx A 96 GQ387011 Colossendeis megalonyx 96 / 1 / 1 100 / 1 GQ387017 Colossendeis megalonyx Colossendeis australis GQ387016 Colossendeis megalonyx

Colossendeis stramenti 65 / 0.53 GQ387009 Colossendeis megalonyx 100 / 1 Colossendeis megalonyx Colossendeis tenera GQ387007 Colossendeis megalonyx GQ387008 Colossendeis megalonyx Colossendeis macerrima 93 / 66 / 0.83 GQ387028 Colossendeis megalonyx 0.03 0.95 100 / 1 FJ969356 Colossendeis megalonyx GQ387027 Colossendeis megalonyx

69 / 0.90 19305−1 Colossendeis angusta 19306−1 Colossendeis angusta 100 19306−2 Colossendeis angusta / 1 19305−3 Colossendeis angusta 98 / 1 19080−1 Colossendeis angusta 78 / HQ970328 Colossendeis angusta 0.98 97 / 75 / 19080−2 Colossendeis angusta 0.69 0.77 HQ970327 Colossendeis angusta 95 58 / / 1 DQ390061 Colossendeis tenera 88 / 0.90 0.67 Colossendeis tenera Mexico Colossendeis tenera / Colossendeis tenera Colossendeis angusta 0.61 GQ387006 Colossendeis scotti 100 / 1 GQ386993 Decolopoda australis DQ390063 Decolopoda australis 94 / 0.99 81 / 0.91 GQ386992 Dodecolopoda mawsoni GQ387001 Colossendeis scoresbii 91 / 0.98 DQ390078 Colossendeis stramenti 54 / GQ387000 Colossendeis robusta 0.58 GQ387002 Colossendeis australis GQ386998 Colossendeis hoeki FJ862873 Colossendeis macerrima FJ862872 Rhopalorhynchus filipes

0.2

Fig. 4 Phylogenetic tree 1: Large tree: Phylogenetic tree based on Phylogenetic tree based on maximum likelihood analysis of 323 bases maximum likelihood analysis of 545 bases of the CO1 gene of 44 of H3 and 560 bases of CO1 of 10 representative specimens of specimens of Colossendeidae. ML bootstrap and Bayesian posterior Colossendeidae. ML bootstrap and Bayesian posterior probability sup- probability support (if >50/0.5) are drawn on branches. Small tree: port (if >50/0.5) are drawn on branches

only by single nucleotide substitutions. The other spec- in preparation. However, the analyzed data raise further imen from Newfoundland was placed basal to both interesting questions on the validity and distribution of groups, with about 5.7–7 % genetic distance to them. C. tenera. The records described by Turpaeva (1975) extending the distribution of C. tenera have not been considered Discussion by other authors who discussed the species (Child 1994, 1995). However, it is interesting to note that C. tenera Morphology and DNA separate C. tenera from C. megalonyx is otherwise not known from Japanese waters (Hedgpeth 1949; Nakamura and Child 1983; Nakamura and Child Data from both morphology and gene markers unequiv- 1991; Y. Takahashi, personal communication), suggest- ocally support the view that C. tenera and C. megalo- ing that the single specimen reported by Turpaeva nyx are distinct taxa and reject the hypothesis that C. (1975), if correctly identified, does not represent a regular tenera is a subspecies of C. megalonyx as suggested by occurrence of the species. Turpaeva (1975). They also provide further evidence for Turpaeva (1996) mentions differences in the length ratio of the distinction of the C. megalonyx clades identified by the terminal claw to the propodus, with specimens from deeper Krabbe et al. (2010), about which more detailed work is waters having proportionally longer claws. Also, the abdomen Genetic and morphological analyses of Colossendeis megalonyx 159 of deep-sea specimens was found to be longer than in shallow- lengths of 1.38 (Fage 1956) and even 1.72 (Minnaard and water specimens. We cannot confirm this, as the specimens Zamponi 1984) have been recorded for C. angusta/gracilis examined by us had long claws and short abdomina irrespec- specimens. The ocular tubercle is variable within C. angusta tive of the depth in which they were found. The Peruvian (Bamber and Thurston 1995), but the original description specimen had the eye tubercle situated further behind than (Sars 1877, 1891) mentions a very high tubercle similar to the North Pacific specimens (Turpaeva 1975). The specimens Hedgpeth’s(1943) illustration of C. tenera. However, we we examined had a much lower ocular tubercle than described found that our C. tenera specimens had a much lower, by Hedgpeth (1943), Turpaeva (1975), and Child (1994, rounded tubercle, apparently not broken off. Our C. angusta 1995), but agreed with the description of Hilton (1943)thatit specimens from the Norwegian Sea show a high, pointed is not pointed. The British Columbia specimen sequenced by ocular tubercle, similar to the original description, while in Arango and Wheeler (2007) also had front eyes much larger those from near the Icelandic coast the tubercle is much than back ones and a relatively shorter proboscis, i.e., about the lower and pointed, and in those from West Africa it is low same length as the trunk (as described by Child 1994), as well and rounded. This difference within North Atlantic C. as much shorter legs. These differing characters led to the angusta is consistent with the genetic data showing differ- separation of the British Columbia C. tenera from the others entiation between the Norwegian Sea and Iceland speci- in the PCA. As our genetic study has revealed the presence of mens. The extremely short eighth palp article and long probable unrecognized lineages within C. tenera, these char- claws, which are characteristic of C. tenera among the acters may distinguish several distinct geographically and/or Colossendeis species from the North American Pacific coast bathymetrically segregated clades when more specimens are (Child 1995), were present in all our specimens of both C. available. The claw and abdomen characters were interpreted tenera and C. angusta. On the other hand, C. angusta/ by Turpaeva (1996) as a result of pedomorphosis, which was gracilis adults are known for their ninth palp articles hypothesized to be part of a supposed general trend of deep-sea inserted anaxially on the eighth ones (e.g., Hoek 1881; pycnogonid forms developing pedomorphic characters. The Sars 1891), which was also observed in the specimens we relatively longer legs and lower number of ovigeral spines of examined. This palp configuration is not present in C. C. tenera compared to C. megalonyx were also attributed to tenera, and apparently not present in juveniles of C. angusta pedomorphism. Our own results cannot confirm a correlation either (Meinert 1899). between a deep-sea habitat and decreasing number of ovigeral Unlike other Colossendeis species from the same region, spine rows, as the C. megalonyx clades known from the deep- C. tenera is mostly found with its legs raised vertically est waters (C and E) do not have the lowest numbers of above the trunk (Child 1994). All specimens of C. angusta ovigeral spine rows (in fact, E has the highest number). Also, examined by us also show this condition, which has not to the total leg length of C. tenera as a proportion of the body our knowledge been described in the literature for that length does not fall outside the range of variation for C. species. It is also present in other Colossendeis species megalonyx (except in the British Columbia specimen, in which such as C. macerrima Wilson, 1881, but only rarely in C. the legs are in fact shorter), but the proportions of the leg megalonyx (Dietz, personal observation). This is a typical articles are different. Although we generally appreciate that position adopted by pycnogonids while swimming, which pedomorphic traits can evolve under certain conditions (see has been recorded in diverse taxonomic groups (e.g., Clark Diz et al. 2012 for an example), we do not regard this as a and Carpenter 1977), but it could also be a post-mortem resonable explanation here. artifact. The presence of well-developed eyes is another differ- C. tenera and C. angusta ence between C. angusta and C. tenera, although Gordon (1944) mentions small eyes present in C. gracilis, a possible Our molecular results indicate that C. tenera is closer to C. synonym of C. angusta (Fry and Hedgpeth 1969). Minnaard angusta than to C. megalonyx. The two species were already and Zamponi (1984) also reported the presence of eyes in C. compared by Hedgpeth (1943), and C. tenera was claimed gracilis. We examined C. angusta specimens from the North to differ from C. angusta by its longer proboscis, a higher Atlantic and West Africa, and eyes appeared to be absent or ocular tubercle, and the presence of eyes. Our own measure- at least vestigial in all of them. Another character that differs ments also showed that the proboscis is shorter in North between C. tenera and most specimens of both C. angusta/ Atlantic and West African C. angusta than in C. tenera gracilis and C. megalonyx examined and described in the (except for the British Columbia specimen), and this is also literature is the very short abdomen in C. tenera, which true for Hoek’s(1881) original description of C. gracilis however agrees with the measurements reported by (relative length 1.08–1.11). However, the relative proboscis Minnaard and Zamponi (1984) for C. angusta/gracilis from lengths of C. tenera and C. angusta seem to overlap con- unspecified locations. Turpaeva (1996) reported that in both siderably (see Results for C. tenera). Relative proboscis C. angusta and C. tenera the abdomen length is variable, 160 L. Dietz et al. with specimens from deeper waters having longer abdom- irregularly in two to three rows. The specimen from Peru ina. Additionally, C. angusta/gracilis specimens retaining pictured by Turpaeva (1975) agrees with the northern speci- fully developed chelifores have often been described (e.g., mens in this. Schimkewitsch (1893) mentioned that the Hoek 1881; Meinert 1899; Fage 1956; Turpaeva 1996, one specimens attributed by him to C. gracilis from the of the West African specimens examined by us), but this is Central and South American Pacific coast had three irregu- not known for C. tenera. lar rows of small spines instead of two in Hoek’s(1881) The position of C. tenera relative to C. angusta is com- specimens, which agrees with C. tenera, but also with the C. plicated by the status of the latter as a very widespread and angusta specimens examined by us. While the spine con- variable deep-water species of the Northern Hemisphere and figuration of C. angusta, especially the Iceland specimens, its possible synonymy with its Southern Hemisphere coun- seems strikingly different from that of C. tenera in the lack terpart C. gracilis. According to Stock (1963), C. gracilis of a distinct row of closely spaced medium spines, such a differs from C. angusta by the presence of a highly conical row also occurs in a number of other Colossendeis species, ocular tubercle (variable in North Atlantic C. angusta such as C. australis or C. scotti. It is also clearly present in according to our own observations and Bamber and Hoek’s(1881) illustration of C. gracilis from the Southern Thurston (1995), and actually present in the type material Indian Ocean. The differing morphology of North Atlantic of C. angusta according to Sars (1891); also variable in C. angusta should therefore probably be considered autapo- specimens currently attributed to C. tenera), shortness of morphic and therefore not evidence for a closer relationship the three distal palp articles, small body size, and slender between C. tenera and C. megalonyx. legs, characters that seem to fit C. tenera. In the C. angusta Schimkewitsch (1893) described a variety of C. gracilis we examined, all characters mentioned by Stock (1963), from the Mexican Pacific coast that he named C. gracilis except for the ocular tubercle of some specimens, agreed var. pallida. This was distinguished from the typical form of with C. gracilis. Also within C. angusta s. str., noticeable the species by its longer proboscis, a very high ocular differences exist between different geographical regions tubercle with unpigmented eyes, a pale yellow color, more (Appellöf 1912), as our own observations of the ocular developed setae on the distal palp articles, and shorter tubercle also indicated. According to Bamber and ovigera. The first two of these characters are the same that Thurston (1995), the variability within specimens attributed Hedgpeth (1943)usedtodistinguishC. tenera from C. to C. angusta/gracilis is so complicated that it is currently angusta. The color is the same as in the specimens from impossible to clearly distinguish multiple species. We con- British Columbia, Oregon, and Mexico examined by us, but sider it would be rather improbable that all records assigned not the specimen from Washington, which is dark red. to C. angusta/gracilis in the taxonomic literature can be However, it also agrees with the North Atlantic C. angusta classified as a single species, as this would mean that C. specimens. The West African C. angusta we examined show angusta has a worldwide distribution from the Arctic to the a much more intense yellow color. Sars (1891) and Möbius Southern Ocean, which is a very rare distribution pattern for (1902) mention that living specimens of North Atlantic C. marine benthic organisms (but see, e.g., Pawlowski et al. angusta have an intense brick red color, which however 2007). vanishes in ethanol. According to Wilson (1881) the color The ovigeral spine configuration would seem to contra- among US East Coast specimens varies “from straw yellow dict the hypothesis that C. tenera (Fig. 3b, f)isclosely to nearly white.” The setae on the palps are also mentioned related to C. angusta (Fig. 3g,h), as it shares more character- and figured in Hedgpeth’s(1943) description of C. tenera. istics with C. megalonyx (Fig. 3c-e). While our C. mega- This similarity indicates that C. gracilis var. pallida might lonyx specimens showed a higher number of spine rows than be a senior synonym of C. tenera. The Mexican C. tenera C. tenera, the original descriptions of C. megalonyx (Hoek specimen sequenced here falls within the distribution range 1881), C. rugosa (Hodgson 1907), and C. m. arundorostris of C. gracilis var. pallida as given by Schimkewitsch (Fry and Hedgpeth 1969), as well as the Scotia Sea speci- (1893). However, Schimkewitsch (1893) did not illustrate mens described by Turpaeva (1975), indicated that there are the specimens except for the distal palp articles or report any also members of the C. megalonyx complex with a lower detailed measurements, and without examining the relevant number of ovigeral spine rows than those sampled either by specimens this suggestion cannot be tested. us or by Cano and López González (2007). However, it The apparent non-monophyly of C. tenera with respect to must be noted that the number of spine rows increases with C. angusta specimens from the North Atlantic, the strong ontogeny (Fry and Hedgpeth 1969). The ovigeral spines of genetic differentiation within C. angusta, and the morpho- the Mexican specimen of C. tenera (Fig. 3a,b) differed from logical diversity observed within both species raises the those of the C. tenera specimens from further north (Fig. 3f) question whether any distinction can be drawn between and from both C. angusta and C. megalonyx in that the the two species, or whether they are both part of a species small spines are placed regularly in two distinct rows, not complex, which is in need of a profound and thorough Genetic and morphological analyses of Colossendeis megalonyx 161 traditional taxonomic review. Examination of more material Calman, W. T., & Gordon, I. (1933). A dodecapodous pycnogonid. – from different geographical areas with both morphological Proceedings of the Royal Society of London (B), 113, 107 115. Cano, E., & López González, P. J. (2007). Colossendeis species and molecular methods would be necessary to test this (Pycnogonida: Colossendeidae) collected during the Italica XIX hypothesis and investigate the geographical and bathymetric cruise to Victoria Land (Antarctica), with remarks on some taxo- distribution of possible undescribed species. nomic characters of the ovigers. Scientia Marina, 71(4), 661–681. In conclusion, our results support the hypothesis that C. Child, C. A. (1994). Deep-sea Pycnogonida from the temperate west coast of the United States. Smithsonian Contributions to Zoology, tenera and C. megalonyx are two morphologically and mo- 556,I–III. 1-23. lecularly distinct species complexes. The two sequences of C. Child, C. A. (1995). Pycnogonida of the Western Pacific Islands, XI: tenera, however, are very different from each other and form a Collections from the Aleutians and other Bering Sea Islands, monophyletic group together with C. angusta, which is also Alaska. Smithsonian Contributions to Zoology, 569,I–IV + 1–30. Clark, W. C., & Carpenter, A. (1977). Swimming behaviour in a similar morphologically. Within both nominal species there pycnogonid. New Zealand Journal of Marine and Freshwater appears to be a large amount of morphological variability. Research, 11(3), 613–615. Hence, future studies with more material will need to analyze Dietz, L., Mayer, C., Arango, C. P., & Leese, F. (2011). The mitochon- C. tenera, C. angusta, and related species in more detail. drial genome of Colossendeis megalonyx supports a basal position of Colossendeidae within the Pycnogonida. Molecular Phylogenetics and Evolution, 58(3), 553–558. Acknowledgements We would like to thank Anja Friederichs (Mu- Diz, A. P., Páez de la Cadena, M., & Rolán-Alvarez, E. (2012) Proteomic seum für Naturkunde Berlin) for taking additional photographs of the evidence of a paedomorphic evolutionary process within a marine Mexican C. tenera specimen, Hieronymus Dastych (Zoologisches snail species: a strategy for adapting to extreme ecological condi- Museum Universität Hamburg) for providing the samples of C. tions? Journal of Evolutionary Biology, 25,2569–2581. angusta, Anna Soler i Membrives for information on the C. angusta Eights, J. (1835). Description of a new animal belonging to the Arachnida sequence data in GenBank, Saskia Brix and the organizers of the of Latreille; discovered in the sea along the shores of the New ICEAGE cruise for C. angusta, and Meike Seefeldt for the SEM Shetland Islands. Boston Journal of Natural History, 1(2), 203–206. photographs of C. angusta. We also thank Yoshie Takahashi for infor- Fage, L. (1956). Pycnogonides (exc. le genre Nymphon). Galathea mation on Colossendeis species from Japan. Furthermore, we would Report, 2, 167–182. like to thank Christoph Held, Christoph Mayer, Markus Gronwald, and Fry, W. G. (1978). A classification within the pycnogonids. Zoological Ralph Tollrian for support and helpful discussions, and three anony- Journal of the Linnean Society (London), 63(1/2), 35–58. mous reviewers for comments that were a great help in improving the Fry, W. G., & Hedgpeth, J. W. (1969). The fauna of the Ross Sea. Part manuscript. This work was funded by DFG grant LE 2323/2 to FL 7. Pycnogonida, 1: Colossendeidae, Pycnogonidae, Endeidae, within the priority programme, SPP 1158. Ammotheidae. New Zealand Department of Science and Industrial Research Bulletin, 198,1–139. Ethical standards The experiments comply with the current laws of Gordon, I. (1944). Pycnogonida. British and New Zealand Antarctic the country in which they were performed. Research Expedition 1929-1931. Reports (B = Zoology & Botany), 5(1), 1–72. Griffiths, H. J., Arango, C. P., Munilla, T., & McInnes, S. J. (2011). Conflict of interest The authors declare that they have no conflict of Biodiversity and biogeography of Southern Ocean pycnogonids. interest. Ecography, 34(4), 616–627. Hedgpeth, J. W. (1943). On a new species of pycnogonid from the North Pacific. Journal of the Washington Academy of Sciences, 33 References (7), 223–224. Hedgpeth, J. W. (1947). On the evolutionary significance of the Pycnogonida. Smithsonian Miscellaneous Collections, 106(18), 1– Appellöf, A. (1912). Invertebrate bottom fauna of the Norwegian Sea 53. Pl. 1. and North Atlantic. Ch. 8. In J. Murray & J. Hjort (Eds.), The Hedgpeth, J. W. (1949). Report on the Pycnogonida collected by the depths of the sea: 821. London: Macmillan. Albatross in Japanese Waters in 1900 and 1906. Proceedings of Arango, C. P. (2002). Morphological phylogenetics of the sea spiders the United States National Museum, 98, 233–321. (Arthropoda: Pycnogonida). Organisms, Diversity and Evolution, Hilton, W. A. (1943). Pycnogonida from the Pacific. Family 2(2), 107–125. Colossendeidae. Journal of Entomology and Zoology, Pomona Arango, C. P., & Wheeler, W. C. (2007). Phylogeny of the sea spiders College, Claremont, 35(1), 2–4. (Arthropoda, Pycnogonida) based on direct optimization of six Hodgson, T. V. (1907). Pycnogonida. Natural History Collections of loci and morphology. Cladistics, 23, 255–293. the "Discovery" National Antarctic Expedition, Zoology. London, Bamber, R. N. (2007). A holistic re-interpretation of the phylogeny of 3,1–72. Pls. 1–10. the Pycnogonida Latreille, 1810 (Arthropoda). Zootaxa, 1668, Hoek, P. P. C. (1881). Report on the Pycnogonida, dredged by H. M. S. 295–312. "Challenger" 1873/76. Challenger Report, Zoology, 3(X), 1–167. Bamber, R.N., & El Nagar, A. (Eds.) (2012). Pycnobase: World Pls I-XXI. Pycnogonida Database. Available online at http://www.marinespecies. Huelsenbeck,J.P.,&Ronquist,F.(2001).MRBAYES:Bayesian org/pycnobase/ accessed on 2012-06-08 inference of phylogenetic trees. Bioinformatics, 17, 754–755. Bamber, R. N., & Thurston, M. H. (1995). The deep-water pycnogo- Jarzynsky, T. (1870). Praemissus catalogus Pycnogonidarum, nids (Arthropoda: Pycnogonida) of the northeastern Atlantic inventarum in mari Glaciali, ad oras Lapponiae rossicae et Ocean. Zoological Journal of the Linnean Society (London), in mari Albo, anno 1869 et 1870. Annales de la Société des 115,117–162. Naturalistes de St. Pétersbourg, 1,319–320. Calman, W. T. (1915) British Antarctic (“Terra Nova”) Expedition, Krabbe, K., Leese, F., Mayer, C., Tollrian, R., & Held, C. (2010). 1910. Natural History Reports. Zoology, 3(1), 1–74. Cryptic mitochondrial lineages in the widespread pycnogonid 162 L. Dietz et al.

Colossendeis megalonyx Hoek, 1881 from Antarctic and Sars, G. O. (1891). Pycnogonida. The Norwegian North-Atlantic Subantarctic waters. Polar Biology, 33, 281–292. Expedition 1876/78. Christiania: Grøndahl & Søns. Meinert, F. (1899). Pycnogonida. Copenhagen. Schimkewitsch, W. (1893). Compte-rendu sur les Pantopodes. In Meusemann, K., von Reumont, B. M., Simon, S., Roeding, F., Strauss, Reports on the dredging operations off the west coast of Central S., Kück, P., et al. (2010). A phylogenomic approach to resolve America to the Galapagos, to the west coast of Mexico, and in the the arthropod tree of life. Molecular Biology and Evolution, 27 Gulf of California, in charge of Alexander Agassiz, carried on by (11), 2451–2464. the U. S. Fish Commission Steamer "Albatross", during 1891, Minnaard, V. A., & Zamponi, M. O. (1984). Estudios sistemáticos de Lieut. Commander Z. L. Tanner, U. S. N., commanding. Bulletin algunos pantopodos de la región subantártica. Historia Natural, 4, of the Museum of Comparative Zoology, Harvard College 25(2), 257–279. 27-43. Pls. 1-2. Möbius, K. (1902). Die Pantopoden der deutschen Tiefsee-Expedition Stamatakis, A. (2006). RAxML–VI–HPC: maximum likelihood–based 1898-1899. Wissenschaftliche Ergebnisse der Deutschen Tiefsee- phylogenetic analyses with thousands of taxa and mixed models. Expedition auf "Valdivia" 1898-1899. Jena, 3(6), 175–196. Bioinformatics, 22, 2688–2690. Tafeln 24-30. Stock, J. H. (1963). South African deep-sea Pycnogonida, with Munilla León, T. (1999). Evolución y filogenia de los picnogónidos. descriptions of new species. Annals of the South African Boletín de la Sociedad Entomológica Aragonesa, 26, 273–279. Museum, 46(12), 321–340. Munilla, T., & Soler-Membrives, A. (2009). Check-list of the pycno- Strugnell, J., Rogers, A. D., Prodöhl, P. A., Collins, M. A., & Allcock, A. gonids from Antarctic and sub-Antarctic waters: zoogeographic L. (2008). The thermohaline expressway: the Southern Ocean as a implications. Antarctic Science, 21,1–13. centre of origin for deep-sea octopuses. Cladistics, 24,853–860. Nakamura, K., & Child, C. A. (1983). Shallow-Water Pycnogonida Turpaeva, E. F. (1975). On some deep-water pantopods (Pycnogonida) from the Izu Peninsula, Japan. Smithsonian Contributions to collected in north-western and south-eastern Pacific. Trudy Zoology, 386,1–71. Instituta Okeanology "P. P. Shirshova", Akademy Nauk SSSR, Nakamura, K., & Child, C. A. (1991). Pycnogonida from Waters adjacent 103, 230–246. In Russian. to Japan. Smithsonian Contributions to Zoology, 512,1–74. Turpaeva, E. F. (1996). Variability of sea spiders morphological char- Nylander, J. A. A. (2004). MrModeltest v2. Program distributed by the acters (Pycnogonida) related to their habitat depth. Okeanologiya, author. Evolutionary Biology Centre, Uppsala University. 36(6), 892–896. In Russian. Pawlowski, J., Fahrni, J., Lecroq, B., Longet, D., Cornelius, N., Wilson, E. B. (1881). XIII. Report on the Pycnogonida. In Reports on Excoffier, L., et al. (2007). Bipolar gene flow in deep-sea benthic the Results of Dredging, under the Supervision of Alexander foraminifera. Molecular Ecology, 16(19), 4089–4096. Agassiz, along the coast of the United States, during the summer Regier, J. C., Shultz, J. W., Zwick, A., Hussey, A., Ball, B., Wetzer, R., et of 1880, by the U. S. Coast Survey Steamer "Blake“, Commander al. (2010). Arthropod relationships revealed by phylogenomic anal- J. R. Bartlett, U. S. N., Commanding. Bulletin of the Museum of ysis of nuclear protein-coding sequences. Nature, 463,1079–1083. Comparative Zoology, 239-256. Pls. I- V. Sars, G. O. (1877). Prodromus descriptionis crustaceorum et pycnogo- Wood-Mason, J. (1873). On Rhopalorhynchus Kröyeri, a new genus nidarum, quae in expeditione norvegica anno 1876, observavit. and species of Pycnogonida. Journal of the Asiatic Society Archiv for Mathematik og Naturvidenskab, 2, 237–271. Bengal, 42(9), 171–175. Tab. 13.

Online Resource 1: Sequences used for phylogenetic analyses

Cytochrome oxidase 1 sequences used for phylogenetic analysis

Species Accession number Location Reference Rhopalorhynchus FJ862872 Arabi et al. (2010) filipes Decolopoda australis DQ390063 Livingston, Arango & Wheeler Antarctica (2007) Decolopoda australis GQ386993 Shag Rocks Krabbe et al. (2010) Dodecolopoda GQ386992 Elephant Island, Krabbe et al. (2010) mawsoni Antarctica Colossendeis HQ970327 Flemish Cap, unpublished angusta Canada Colossendeis HQ970328 Grand Bank, Canada unpublished angusta Colossendeis KC462557 Norwegian Sea This publication angusta Colossendeis KC462560 Norwegian Sea This publication angusta Colossendeis KC462561 Iceland This publication angusta Colossendeis KC462562 Iceland This publication angusta Colossendeis KC462563 Iceland This publication angusta Colossendeis KC462564 Iceland This publication angusta Colossendeis KC462565 Iceland This publication angusta Colossendeis GQ387002 South Sandwich Krabbe et al. (2010) australis1 Islands Colossendeis scotti2 GQ387006 South Sandwich Krabbe et al. (2010) Islands Colossendeis FJ862873 Arabi et al. (2010) macerrima Colossendeis FJ969356 Victoria Land, Nielsen et al. (2009) megalonyx Antarctica Colossendeis GQ387007 South Sandwich/ Krabbe et al. (2010) megalonyx Elephant Island Colossendeis GQ387008 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387009 South Sandwich Krabbe et al. (2010) megalonyx Islands Colossendeis GQ387010 Burdwood Bank Krabbe et al. (2010) megalonyx Colossendeis GQ387011 Burdwood Bank Krabbe et al. (2010) megalonyx Colossendeis GQ387012 Burdwood Bank Krabbe et al. (2010) megalonyx Colossendeis GQ387013 Burdwood Bank Krabbe et al. (2010) megalonyx Colossendeis GQ387014 Burdwood Bank Krabbe et al. (2010) megalonyx Colossendeis GQ387015 Burdwood Bank Krabbe et al. (2010) megalonyx Colossendeis GQ387016 Bouvet Island Krabbe et al. (2010) megalonyx Colossendeis GQ387017 Bouvet Island Krabbe et al. (2010) megalonyx Colossendeis GQ387018 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387019 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387020 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387021 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387022 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387023 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387024 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387025 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387026 South Sandwich/ Krabbe et al. (2010) megalonyx Bouvet Island Colossendeis GQ387027 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis GQ387028 Elephant Island, Krabbe et al. (2010) megalonyx Antarctica Colossendeis robusta GQ387000 Burdwood Bank Krabbe et al. (2010) Colossendeis GQ387001 South Sandwich Krabbe et al. (2010) scoresbii3 Islands Colossendeis hoeki3 GQ386998 South Sandwich Krabbe et al. (2010) Islands Colossendeis DQ390078 Antarctica Arango & Wheeler stramenti (2007) Colossendeis tenera DQ390061 British Columbia, Arango & Wheeler Canada (2007) Colossendeis tenera KC462566 Colima, Mexico This publication Colossendeis FJ969358 Balleny Islands, Nielsen et al. (2009) tortipalpis Antarctica 1 Originally incorrectly identified as C. wilsoni in Krabbe et al. (2010), here redetermined as C. australis. 2 Originally incorrectly identified as C. lilliei in Krabbe et al. (2010), here redetermined as C. scotti. 3 Originally incorrectly identified as C. robusta in Krabbe et al. (2010), here redetermined as C. scoresbii. 4 Originally incorrectly identified as C. scoresbii in Krabbe et al. (2010), here redetermined as C. hoeki.

CO1 and H3 sequences used for two-gene analysis

Species CO1 H3 Reference Colossendeis FJ862873 FJ862884 Arabi et al. (2010) macerrima Colossendeis GQ387007 KC456499 Krabbe et al. (2010) megalonyx A Colossendeis GQ387010 KC456431 Krabbe et al. (2010) megalonyx B Colossendeis GQ387017 KC456434 Krabbe et al. (2010) megalonyx C Colossendeis GQ387019 KC456450 Krabbe et al. (2010) megalonyx D Colossendeis GQ387026 KC456481 Krabbe et al. (2010) megalonyx E Colossendeis GQ387027 KC456500 Krabbe et al. (2010) megalonyx F Colossendeis DQ390078 DQ390187 Arango & Wheeler stramenti (2007) Colossendeis tenera DQ390061 DQ390171 Arango & Wheeler (2007) Decolopoda australis DQ390063 DQ390172 Arango & Wheeler (2007)

Sequences used for multigene analysis

Species 12S 16S CO1 H3 Reference Rhopalorhynchus FJ862832 FJ862842 FJ862872 -1 Arabi et al. filipes (2010) Hedgpethia DQ389952 DQ390002 - DQ390162 Arango & dofleini Wheeler (2007) Decolopoda DQ389961 DQ390012 DQ390063 DQ390172 Arango & australis Wheeler (2007) Colossendeis sp. - GQ328953 GQ328965 - Koenemann et al. (2010) Colossendeis sp. FJ862837 FJ862847 FJ862877 FJ862888 Arabi et al. MNHN-JAD2 (2010) Colossendeis FJ969307 FJ969325 FJ969355 - Nielsen et al. australis (2009) Colossendeis FJ862833 FJ862843 FJ862873 FJ862884 Arabi et al. macerrima (2010) Colossendeis HQ450773 HQ450773 HQ450773 KC456465 Dietz et al. megalonyx (2011) Colossendeis - FJ969326 FJ969356 - Nielsen et al. megalonyx (2009) Colossendeis FJ969308 FJ969327 FJ969357 - Nielsen et al. notialis (2009) Colossendeis DQ389973 DQ390027 DQ390078 DQ390187 Arango & stramenti Wheeler (2007) Colossendeis DQ389959 DQ390010 DQ390061 DQ390171 Arango & tenera Wheeler (2007) Colossendeis - FJ969328 FJ969358 - Nielsen et al. tortipalpis (2009) 1The H3 sequence of R. filipes (FJ862883) reported by Arabi et al. (2010) is almost identical to that of Colossendeis macerrima (FJ862884) published by the same authors. As the sequence groups within C. macerrima in our analyses, we consider it to be misidentified and did not include it in the final analysis.

Additional references Arabi, J., Cruaud, C., Couloux, A., Hassanin, A. (2010). Studying sources of incongruence in arthropod molecular phylogenies: sea spiders (Pycnogonida) as a case study. C. R. Biol. 333, 438–453. Koenemann, S., Jenner, R. A., Hoenemann, M., Stemme, T., von Reumont, B. M. (2010). Arthropod phylogeny revisited, with a focus on crustacean relationships. Arthropod Structure and Development 39, 88-110. Nielsen, J. F., Lavery, S., Lörz, A.-N. (2009). Synopsis of a new collection of sea spiders (Arthropoda: Pycnogonida) from the Ross Sea, Antarctica. Polar Biology, 32, 1147– 1155.

Online Resource 2: Concatenated phylogenetic tree based on CO1, 18S, 28S and H3 Supporting information 3: a) Information on the Eigenvalues of the ten calculated factors and the total variance. b) Factor loadings indicating correlations with the different variables. Factors 1 and 2 were visualized in the PCA. a) Eigenvalues of correlation matrix, and related statistics (PCA_11variables_alltax-averagesCme_STAT) Active variables only Eigenvalue % Total Cumulative Cumulative Value number variance Eigenvalue % 1 4.655691 46.55691 4.65569 46.5569 2 2.114222 21.14222 6.76991 67.6991 3 1.535457 15.35457 8.30537 83.0537 4 0.612813 6.12813 8.91818 89.1818 5 0.400226 4.00226 9.31841 93.1841 6 0.280617 2.80617 9.59903 95.9903 7 0.154686 1.54686 9.75371 97.5371 8 0.126886 1.26886 9.88060 98.8060 9 0.093289 0.93289 9.97389 99.7389 10 0.026113 0.26113 10.00000 100.0000 b) Factor-variable correlations (factor loadings), based on correlations (PCA_11variables_alltax-avera Variable Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 1st coxa 0.720378 -0.536142 0.131491 -0.076468 0.170003 0.314647 0.156810 2nd coxa 0.546568 0.172973 0.777125 0.038139 -0.034801 0.009595 -0.185592 3rd coxa 0.511099 0.211997 0.783958 0.122785 -0.074965 -0.065276 0.137883 femur 0.620554 0.662155 -0.260583 0.217588 -0.030231 0.009312 0.194752 tibia 1 0.580367 0.694163 -0.310077 0.201900 -0.004568 -0.014428 -0.059825 tibia 2 0.906381 -0.087863 -0.262186 0.090551 -0.148566 0.128260 -0.138733 tarsus 0.898615 -0.274087 -0.215995 -0.026210 -0.068022 0.094597 -0.086731 propodus 0.762507 -0.287137 -0.110920 -0.384333 -0.309828 -0.252799 0.082438 terminal claw 0.226851 0.746184 0.019986 -0.577571 0.216278 0.072856 -0.029051 Proboscis 0.770038 -0.330936 -0.086632 0.110179 0.440558 -0.286969 -0.026902

b) (continued) Factor-variable correlations (factor loadings), based on correlations (PCA_11variables_alltax-avera Variable Factor 8 Factor 9 Fact.10 1st coxa 0.127074 -0.042373 -0.005794 2nd coxa 0.140721 0.096161 0.034194 3rd coxa -0.160715 -0.089254 -0.038471 femur 0.001747 0.136360 0.060743 tibia 1 0.148959 -0.109881 -0.079556 tibia 2 -0.111136 -0.132081 0.079033 tarsus -0.127998 0.160230 -0.082889 propodus 0.095602 -0.016318 0.004155 terminal claw -0.068054 -0.012712 0.006253 Proboscis -0.020201 -0.008060 0.015440

Online Resource 4: Morphological measurements of C. tenera, C. angusta and C. megalonyx specimens, including both our own measurements and those from the literature (lengths relative to trunk are shown if not listed otherwise) Collection trunk terminal Species proboscis trunk abdomen 1st coxa 2nd coxa 3rd coxa femur tibia 1 tibia 2 tarsus propodus / ID (in mm) claw

C. tenera USNM (male paratype) 6.5 1.308 1.000 0.123 0.154 0.231 0.231 2.769 2.308 1.692 0.731 0.538 0.615 C. tenera USNM (female paratype) 7 1.393 1.000 0.143 0.143 0.179 0.214 2.857 2.429 1.643 0.571 0.429 0.536 MNH C. tenera Berlin (Mexico) 6.8 1.544 1.000 0.147 0.206 0.294 0.294 2.324 2.132 1.618 0.721 0.691 0.765 Queenslan C. tenera d Museum (British Columbia) 10.366 0.964 1.000 0.115 0.130 0.153 0.135 1.954 1.742 0.728 0.284 0.260 0.291 University C. angusta Hamburg (West Africa) 0.863 1.000 0.233 0.137 0.137 0.123 1.918 1.904 1.233 0.521 0.438 0.575 Holotype (Calman C. megalonyx (1915) 11 1.818 1.000 0.255 not listed not listed not listed 2.068 1.909 1.614 0.932 0.727 0.636 Cano & Lopez- C. megalonyx Gonzalez (2007) 6.5 1.769 1.000 0.215 0.231 0.231 0.169 2.000 2.000 1.846 0.985 0.769 0.477

C. angusta 19.305-1 ICEAGE 6.69 0.940 1.000 0.250 1.01 1.05 0.88 15.26 14.71 10.03 3.7 3.24 4.2 C. angusta 19.305-2 ICEAGE 8.01 0.966 1.000 0.197 1.35 1.28 1.1 18.06 16.8 11.42 3.96 4 4.95 C. angusta 19.305-3 ICEAGE 6.94 0.921 1.000 0.174 1.04 0.87 0.76 15.75 13.96 9.45 3.91 3.88 5.35 C. angusta 19.306-1 ICEAGE 7.09 0.939 1.000 0.196 1.11 1.23 1.06 15.76 14.68 10.14 4.22 3.45 4.27 C. angusta 19.306-2 ICEAGE 6.43 0.905 1.000 0.202 1.11 1 0.87 15.66 13.99 9.34 3.64 3.66 Broken C. angusta 19.080-1 ICEAGE 9.04 0.982 1.000 0.177 1.27 1.02 0.93 19.47 17.74 12.29 4.31 3.85 4.59 C. angusta 19.080-2 ICEAGE 7.91 0.906 1.000 0.217 1.18 1.05 0.96 16.24 14.72 10.16 3.7 3.25 4.47 C. angusta 19.080-3 ICEAGE 8.66 0.985 1.000 0.214 1.11 1.09 1.04 19.97 18.62 12.58 4.57 3.97 4.47 C. angusta 19.080-4 ICEAGE 8.38 0.988 1.000 0.189 1.26 1.04 1.02 19.45 17.85 12.63 4.82 4.08 4.37 C. megalonyx PA_E005 Krabbe et al. (2010) 10.55 1.468 1 not listed 0.190 0.133 0.133 1.953 1.825 1.649 0.758 0.682 0.389 C. megalonyx PL_E004 Krabbe et al. (2010) 11.4 1.491 1 not listed 0.193 0.158 0.158 1.921 1.947 1.860 0.868 0.667 0.377 C. megalonyx PN_E002 Krabbe et al. (2010) 11.1 1.477 1 not listed 0.198 0.162 0.162 1.901 1.748 1.640 0.820 0.685 0.387 C. megalonyx PN_E007 Krabbe et al. (2010) 9.6 1.490 1 not listed 0.208 0.156 0.156 1.510 1.510 1.354 0.615 0.625 0.323 C. megalonyx PS_E003 Krabbe et al. (2010) 11 1.527 1 not listed 0.191 0.136 0.127 2.009 1.882 1.764 0.855 0.691 0.345 C. megalonyx PS_E004 Krabbe et al. (2010) 11.2 1.402 1 not listed 0.196 0.188 0.179 2.027 1.920 1.759 0.893 0.759 0.411 C. megalonyx PS_E005 Krabbe et al. (2010) 12.9 1.446 1 not listed 0.202 0.116 0.116 1.946 1.814 1.496 0.543 0.527 0.279 C. megalonyx PS_E006 Krabbe et al. (2010) 11.6 1.534 1 not listed 0.207 0.164 0.147 2.009 1.897 1.741 0.862 0.664 0.379 C. megalonyx PJ_E006 Krabbe et al. (2010) 12.03 1.675 1 not listed 0.188 broken broken broken broken broken broken broken broken C. megalonyx PJ_E007 Krabbe et al. (2010) 12.81 1.624 1 not listed 0.187 0.133 0.133 2.365 2.114 1.819 0.948 0.700 0.460 C. megalonyx PJ_E009 Krabbe et al. (2010) 13.83 1.558 1 not listed 0.179 0.119 0.140 2.356 2.037 1.727 0.860 0.664 0.473 C. megalonyx PO_E003 Krabbe et al. (2010) 13.2 1.564 1 not listed 0.214 0.135 0.135 2.234 2.025 1.698 0.935 0.716 0.589 C. megalonyx PO_E004 Krabbe et al. (2010) 12.49 1.725 1 not listed 0.198 0.157 0.145 2.610 2.314 1.994 1.114 0.804 0.545 C. megalonyx PO_E005 Krabbe et al. (2010) 12.04 1.537 1 not listed 0.188 0.154 0.150 2.282 2.047 1.735 0.917 0.684 0.565 C. megalonyx PO_E007 Krabbe et al. (2010) 13.36 1.639 1 not listed 0.189 0.132 0.153 2.269 2.045 1.695 0.965 0.711 0.510 C. megalonyx PO_E008 Krabbe et al. (2010) 12.09 1.563 1 not listed 0.191 0.147 0.147 2.108 1.888 1.613 0.922 0.697 broken C. megalonyx PO_E009 Krabbe et al. (2010) 12.83 1.610 1 not listed 0.190 0.118 0.117 2.244 1.998 1.686 0.970 0.705 0.593 C. megalonyx PO_E010 Krabbe et al. (2010) 11.29 1.599 1 not listed 0.219 0.171 0.159 2.444 2.176 1.743 0.881 0.739 0.638 C. megalonyx PO_E011 Krabbe et al. (2010) 13.76 1.439 1 not listed 0.196 0.110 0.110 2.052 1.835 1.594 0.872 0.666 broken C. megalonyx PO_E012 Krabbe et al. (2010) 11.14 1.544 1 not listed 0.209 0.153 0.153 2.326 2.059 1.777 0.975 0.787 0.614 C. megalonyx PP_E001 Krabbe et al. (2010) 12.75 1.580 1 not listed 0.182 0.166 0.160 2.314 2.044 1.737 0.950 0.638 0.540 C. megalonyx PS_E001 Krabbe et al. (2010) 11.96 1.513 1 not listed 0.203 0.157 0.142 2.579 2.201 1.808 0.958 0.732 0.587 C. megalonyx PS_E002 Krabbe et al. (2010) 10.23 1.593 1 not listed 0.214 0.172 0.155 2.277 1.921 1.732 0.999 0.786 broken C. megalonyx PF_E002 Krabbe et al. (2010) 12.3 1.854 1 not listed 0.199 0.179 0.163 2.488 2.114 2.016 1.081 0.683 0.553 C. megalonyx PF_E006 Krabbe et al. (2010) 12.5 1.744 1 not listed 0.220 0.160 0.148 2.272 1.920 1.680 0.960 0.668 0.488 C. megalonyx PG_E009 Krabbe et al. (2010) 12.3 1.756 1 not listed 0.224 0.175 0.138 2.228 1.878 1.610 0.886 broken broken C. megalonyx PH_E007 Krabbe et al. (2010) 11 1.864 1 not listed 0.232 0.182 0.182 2.345 2.105 1.627 0.768 0.582 0.436 C. megalonyx PI_E002 Krabbe et al. (2010) 12.9 1.585 1 not listed 0.198 0.163 0.163 1.992 1.698 1.419 0.775 0.519 0.403 C. megalonyx PP_E005 Krabbe et al. (2010) 12.63 1.504 1 not listed 0.211 0.188 0.188 2.253 1.835 1.625 0.959 0.676 0.481 C. megalonyx PQ_E001 Krabbe et al. (2010) 11.75 1.792 1 not listed 0.217 0.179 0.196 2.536 2.102 1.864 broken broken broken C. megalonyx PQ_E002 Krabbe et al. (2010) 10.9 1.679 1 not listed 0.216 0.165 0.165 2.477 2.073 1.844 1.060 0.693 0.569 C. megalonyx PQ_E003 Krabbe et al. (2010) 13.37 1.638 1 not listed 0.207 0.142 0.142 2.219 1.882 1.634 0.846 0.598 0.455 C. megalonyx PQ_E004 Krabbe et al. (2010) 10.76 1.673 1 not listed 0.226 0.204 0.174 2.299 1.929 1.669 0.943 0.684 0.514 C. megalonyx PJ_E001 Krabbe et al. (2010) 11.6 1.664 1 not listed 0.211 0.190 0.172 2.315 2.030 1.815 1.013 0.845 0.517 C. megalonyx PJ_E002 Krabbe et al. (2010) 11.4 1.697 1 not listed 0.184 0.184 0.158 2.105 1.851 1.640 0.930 0.789 0.504 C. megalonyx PJ_E003 Krabbe et al. (2010) 12.4 1.698 1 not listed 0.202 0.177 0.161 2.403 2.121 1.879 1.024 0.806 0.500 C. megalonyx PJ_E004 Krabbe et al. (2010) 11.75 1.617 1 not listed 0.204 0.170 0.170 2.328 2.034 1.817 broken broken broken C. megalonyx PJ_E005 Krabbe et al. (2010) 11.6 1.530 1 not listed 0.190 0.172 0.147 2.207 1.974 1.741 0.931 0.724 0.440 C. megalonyx PM_E010 Krabbe et al. (2010) 13.15 1.719 1 not listed 0.205 0.160 0.160 2.251 1.954 1.749 0.996 0.791 0.449 C. megalonyx PM_E011 Krabbe et al. (2010) 13.65 1.579 1 not listed 0.220 0.161 0.154 2.168 1.949 1.758 1.018 0.755 broken C. megalonyx PN_E008 Krabbe et al. (2010) 13.6 1.636 1 not listed 0.199 0.169 0.169 2.493 2.206 2.026 broken broken broken C. megalonyx PN_E009 Krabbe et al. (2010) 11.45 1.878 1 not listed 0.236 0.183 0.183 2.664 broken broken broken broken broken C. megalonyx PN_E010 Krabbe et al. (2010) 12.9 1.640 1 not listed 0.209 0.186 0.186 2.628 2.236 2.016 1.132 0.903 0.516 C. megalonyx PN_E011 Krabbe et al. (2010) 13 1.731 1 not listed 0.200 0.177 0.169 2.615 2.277 2.119 1.185 0.931 0.450 C. megalonyx PS_E007 Krabbe et al. (2010) 13.3 1.744 1 not listed 0.203 0.173 0.173 2.278 1.970 1.805 1.045 1.602 0.579 C. megalonyx PS_E008 Krabbe et al. (2010) 12.2 1.770 1 not listed 0.221 0.193 0.193 2.512 2.225 2.012 1.131 broken broken C. megalonyx PS_E009 Krabbe et al. (2010) 12.85 1.782 1 not listed 0.233 0.195 0.171 2.770 2.420 2.187 1.307 0.957 0.545 C. megalonyx PF_E001 Krabbe et al. (2010) 13.41 1.723 1 not listed 0.182 0.179 0.131 2.253 2.046 1.777 0.978 0.764 0.448 C. megalonyx PF_E003 Krabbe et al. (2010) 12.1 1.826 1 not listed 0.221 0.179 0.141 2.622 2.385 2.160 1.241 0.873 broken C. megalonyx PF_E004 Krabbe et al. (2010) 12.8 1.734 1 not listed 0.216 0.166 0.156 2.602 2.315 2.029 1.105 0.793 0.491 C. megalonyx PF_E011 Krabbe et al. (2010) 12.82 1.704 1 not listed 0.197 0.154 0.143 2.381 2.091 1.832 1.034 0.712 0.420 C. megalonyx PG_E003 Krabbe et al. (2010) 12.67 1.760 1 not listed 0.212 0.178 0.142 2.343 2.125 1.784 1.053 0.739 0.477 C. megalonyx PG_E010 Krabbe et al. (2010) 13.09 1.715 1 not listed 0.207 0.176 0.127 2.341 2.110 1.805 1.020 0.782 broken C. megalonyx PH_E005 Krabbe et al. (2010) 12.05 1.755 1 not listed 0.209 0.181 0.151 2.388 2.152 1.907 broken broken broken C. megalonyx PH_E011 Krabbe et al. (2010) 12.69 1.789 1 not listed 0.219 0.178 0.165 2.088 2.020 1.795 1.035 0.812 0.512 C. megalonyx PI_E003 Krabbe et al. (2010) 13.04 1.806 1 not listed 0.210 0.172 0.166 2.519 2.186 1.918 broken broken broken C. megalonyx PI_E005 Krabbe et al. (2010) 12.98 1.895 1 not listed 0.219 0.177 0.149 2.229 2.082 1.827 1.085 0.814 0.552 C. megalonyx PQ_E005 Krabbe et al. (2010) 11.52 1.766 1 not listed 0.194 0.164 0.147 2.685 2.434 2.155 1.227 0.841 0.492 C. megalonyx PQ_E006 Krabbe et al. (2010) 12.18 1.724 1 not listed 0.207 0.157 0.141 2.475 2.208 1.994 1.131 0.883 0.519

Dissertation Lars Dietz

4) Publikation III

Titel: Regional differentiation and extensive hybridisation between mitochondrial clades of the Southern Ocean giant sea spider Colossendeis megalonyx

Royal Society Open Science, in Begutachtung seit 5. November 2014

Hinweise zu Publikation III

• Anteil Planung: 80%

• Anteil experimentelle Durchführung: 85%

• Verfassen des Manuskripts: 85%

Abbildungen, die nicht ausschließlich von mir erstellt wurden

• Fig. 1: Photographie von Claudia P. Arango

• Fig. 2: Visualisierung der Barcode Gap mit Hilfe eines R-Skripts von Vasco

Elbrecht

82 Dissertation Lars Dietz

Regional differentiation and extensive hybridisation between mitochondrial clades of the Southern Ocean giant sea spider

Colossendeis megalonyx

Lars Dietz1, Claudia P. Arango2, Ken Halanych3, Avril M. Harder4, Christoph Held5, Andrew

R. Mahon4, Christoph Mayer6, Roland R. Melzer7,8, Greg Rouse9, Andrea Weis7, Nerida

Wilson9,10 & Florian Leese1*

1) Ruhr University Bochum, Faculty of Biology and Biotechnology, Dept. of Animal Ecology, Evolution and Biodiversity, Universitaetsstrasse 150, 44801 Bochum, Germany 2) Natural Environments Program, Queensland Museum, Australia 3) Auburn University, 101 Rouse Life Sciences Bldg., AL 36849, USA 4) Department of Biology, Institute for Great Lakes Research, Central Michigan University, Mount Pleasant, MI USA 5) Alfred Wegener Institute, Helmholtz Center for Marine and Polar Biology; Am Alten Hafen 26, 25768 Bremerhaven, Germany 6) Zoological Research Museum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany 7) Bavarian State Collection of Zoology – SNSB, Münchhausenstraße 21, 81247 Munich, Germany 8) Ludwig-Maximilians-Universität München, Department Biology II, Großhaderner Straße 2, 82152 Planegg-Martinsried, Germany; GeoBio-Cente, Richard-Wagner-Straße 10, 80333 Munich, Germany 9) Scripps Institution of Oceanography, University of California, San Diego, 9500 Gilman Drive, La Jolla 92093-0202, California, USA 10) Western Australian Museum, Locked Bag 49, Welshpool DC, Western Australia 6986, Australia

*Corresponding author: Florian Leese, Email: [email protected], Phone:

+49.234.3225072; Fax: +49.234.3214114.

Keywords: Speciation, Phylogeography, Antarctic, Gene Flow, Glacial Refugia, Long

Distance Dispersal, Pycnogonida, Benthos

83 Dissertation Lars Dietz

Abstract

The giant sea spider Colossendeis megalonyx is one of the most widespread Southern

Ocean pycnogonid species. Based on mitochondrial data it was recently suggested that C. megalonyx is a complex of at least six cryptic species with mostly small and non-overlapping distribution ranges. Here we expand the sampling to include over 500 sequences of specimens from around the Antarctic for the mitochondrial Cytochrome Oxidase subunit 1 gene (COI). Additionally, we present sequences from the nuclear ITS region for a subset of these specimens.

Using distance-based and general mixed Yule-coalescent species delimitation approaches, the number of distinct mitochondrial OTUs increased from six to 15-19 with our larger data set. In contrast to earlier studies, many of these mitochondrial show a circumpolar distribution. The nuclear data are incongruent with the mitochondrial results in that some specimens from the same region group together in the ITS data although they were assigned to different COI OTUs. These mito-nuclear discordances can be reconciled by inferring that some OTUs characterized by divergent mitochondrial lineages can hybridize and should thus not be interpreted as cryptic species. The results of this study confirm that the presence of cryptic species should not be inferred based on mitochondrial data alone.

84 Dissertation Lars Dietz

Introduction

Species diversity in the marine Antarctic benthos is severely underestimated [1–3]. One of the main reasons for this problem is still the limited sampling of remote regions and habitats such as the continental slope [4]. Another major challenge is the presence of cryptic or overlooked species, i.e. species that are currently not distinguished morphologically but are genetically distinct (see [5] for a review). With the recent use of molecular techniques, in particular a fragment of the mitochondrial cytochrome c oxidase subunit I (COI) or ‘barcoding gene’, many highly divergent clades have been found and interpreted as different species in different phyla (see [6–10] for examples). These molecular data supporting the discovery of new species have not only improved our knowledge about the true magnitude of biodiversity in the Antarctic, they have also challenged central biogeographic dogmas in the Southern

Ocean: Traditionally, it has been assumed that many Southern Ocean marine animal species have a circumpolar [11–13] and eurybathic [14] distribution. The identification of cryptic species with molecular-based tools in a variety of Antarctic invertebrates has challenged these concepts as several taxa show a very strong regional differentiation, particularly in brooders with a holobenthic lifestyle (i.e. no planktonic dispersal stage, see [15] for a review).

However, some brooders with a regionally differentiated population structure were not found to contain cryptic species (e.g. the pycnogonid Nymphon australe [16,17]) while others with a planktonic dispersal stage have widely distributed cryptic species, such as the crinoid

Promachocrinus kerguelensis [18]. In some species groups, several lineages occur in sympatry (e.g. [9,19]), suggesting that ecological differences should exist between them if indeed they are distinct species. The role of bathymetry in speciation has been reported from other Southern Ocean invertebrates [20]. In some groups, morphological investigations support the distinction of unrecognized species identified with molecular data (e.g. [21–25]).

However, most of these molecular studies have been based only on mitochondrial genes. As several cases have been observed where mitochondrial and nuclear data disagree due to phenomena such as introgressive hybridization or sex-biased dispersal (reviewed in [26]),

85 Dissertation Lars Dietz this can be misleading. Therefore, nuclear data should be studied as well before the existence of cryptic species can be established.

In this study, we analyzed the diversity of the giant sea spider species Colossendeis megalonyx Hoek, 1881 using both nuclear and mitochondrial gene data. C. megalonyx is one of the most widespread pycnogonid species in the Southern Ocean [27], with a circumpolar distribution in Antarctic and Subantarctic waters as well as in South America, South Africa, and Madagascar, at depths of 3-4900 m [28]. Although other sea spiders are benthic brooders with paternal care, the reproductive mode of the entire family Colossendeidae family is still unknown [29]. Because of its wide distribution and high morphological variability, it has often been questioned whether C. megalonyx is a single species (see [30]), and several subspecies and putatively synonymous species have been described (e.g. [31,32]).

However, no detailed systematic morphological study has been published yet.

A recent study by Krabbe et al. [33] investigated C. megalonyx from a molecular perspective.

It was shown that COI sequences of C. megalonyx fall into six major clades with limited distribution ranges, whose genetic distances from each other are comparable to those of distinct species. However, the 96 samples included in that study were from only few areas

(South Sandwich Islands, Elephant Island, Bouvet Island, Burdwood Bank). Here we substantially expanded the dataset of Krabbe et al. [33] by adding COI data for over 300 specimens from the same areas as well as from other regions in South America, along the

Scotia Arc, and from the West and East Antarctic shelf. We further included data from an additional locus, the nuclear ribosomal gene region Internal Transcribed Spacer (ITS) for a subset of individuals. This region, which includes the gene for 5.8S rRNA as well as the non- coding ITS1 and ITS2, has been found to be useful to distinguish closely related species in many different animal groups (e.g. [34–36]), including pycnogonids [37]. With the new data set we tested i) whether there are further overlooked mitochondrial clades additional to the six clades found by Krabbe et al. [33], ii) whether the proposed narrow distribution ranges of the clades were supported by the new data from many more regions, iii) whether or not the

86 Dissertation Lars Dietz nuclear data support the pattern revealed by the mitochondrial data, and iv) we analysed whether C. megalonyx colonised the Antarctic from the Subantarctic or vice versa.

Figure 1: Map of the Southern Ocean with sampling sites of the specimens of Colossendeis megalonyx analysed in this study. Colours correspond to those in Figs. 3 and 4. For a detailed overview of samples and sampling sites see supporting information S1 and S2. Photo of Colossendeis megalonyx: Claudia P. Arango.

Materials and Methods

A 658 bp fragment of the mitochondrial COI gene was sequenced for a total of 388 putative

C. megalonyx specimens from different parts of the Southern Ocean (see Fig. 1 for a map of the sampling sites) and for an additional 73 specimens belonging to other colossendeid species (Table 1). Individuals were determined to species level with the keys of Child [30] and Pushkin [38] prior to completing any genetic analyses. DNA extractions were performed

87 Dissertation Lars Dietz using the Qiagen DNeasy Blood & Tissue Kit following the manufacturer’s protocol with the exception of using only 100 µl elution buffer (EB) to increase final DNA concentration. PCR for COI was performed as outlined in Krabbe et al. [33].

An ~1000 bp fragment of the ITS (18S – ITS1 – 5.8S – ITS2 – 28S) was sequenced for a subset of 57 C. megalonyx specimens and 26 other colossendeids. PCR was performed as follows: 94°C for 2 min, followed by 37 cycles of 94°C for 20 s, 55°C for 30 s, and 65°C for

80 s, with a final extension at 65°C for 10 min. Primers used for PCR were ITSRA2 and

ITS2.2 [39].

For both gene regions, the PCR mix consisted of 2 µl 10x HotMaster Taq Buffer (5Prime,

Hilden, Germany), 2 µl of 2 mM dNTPs, 0.1 µl of 100 µM HCO primer, 0.1 µl of 100 µM LCO primer [40], 0.1 µl of 5 U/µl HotMaster Taq (5Prime, Hilden, Germany), 1 µl DNA (~20 ng), filled up to 20 µl with sterile H2O. PCR products were purified with a 1:2 mix of Exo and

FastAP for 15 min at 37°C followed by inactivation for 15 min at 85°C. Sequencing was performed at GATC Biotech (Cologne, Germany).

For COI, the colossendeid sequences from Krabbe et al. [33] (96 C. megalonyx, 19 from other species) and all sequences from GenBank that were identified as members of the

Colossendeidae by BLASTn searches (37 C. megalonyx, 113 from other species) were added to the total resulting dataset for analyses.

For both gene regions, sequences were edited with Geneious 6.1.6 [41]. COI was aligned using MUSCLE [42] with the default parameters as implemented in Geneious, using eight iterations. ITS was aligned with MAFFT 7 [43] using the E-INS-I algorithm with a gap opening penalty of 1.53 and offset value 0. For COI, we verified that all codons could be translated into amino acids without stop codons using the invertebrate mitochondrial genetic code. For

ITS, a version of the alignment with ambiguously aligned regions removed was produced with Gblocks 0.91b [44] using less stringent parameters (smaller blocks, gaps in final alignment allowed, less strict flanking positions). Bayesian phylogenetic analysis was performed with MrBayes v. 3.2.1 [45] using 5,000,000 MCMC generations, of which the first

25% were discarded as burn-in (test for convergence: split divergence < 0.01). The most

88 Dissertation Lars Dietz suitable model of molecular evolution for the analyses (GTR+G) was selected with jModeltest

2.1.2 [46]. Maximum-likelihood analysis was performed with RaxML 7.03 [47] and support was assessed with 1000 rapid bootstrap replicates.

For COI, sequences were collapsed into haplotypes with the online Fabox haplotype collapse tool [48]. A Kimura 2-parameter (K2P) distance matrix was created using PAUP 4.0b10 [49].

For species delimitation, a general mixed Yule coalescent (GMYC) analysis was performed.

For this, a linearized tree of the haplotypes was calculated using BEAST 1.8 [50] with the model specified by jModeltest. Convergence and effective sampling size (ESS >200) of parameter estimates were checked using Tracer 1.5 [51], and a consensus tree was calculated using TreeAnnotator 1.8 [52] and analysed with the SPLITS program available as a package for the statistical software environment R [53]. An additional test for the presence of distinct clades was performed using the program ABGD [54] using K2P distances.

Haplotype networks for the four largest clades (A, D1, E and I) were created with TCS 1.21

[55] using a 95% connection limit.

Results

After removal of poorly represented regions at the 3’ and 5’ end, the COI alignment had a length of 545 sites, of which 265 were variable and 226 parsimony-informative. The 387 C. megalonyx sequences grouped into 145 haplotypes. The ITS alignment had a total length of

1138 sites, of which 329 were variable and 296 parsimony-informative. The 57 ITS sequences of C. megalonyx grouped into 31 haplotypes. After removing ambiguously aligned regions from the ITS alignment with Gblocks, the number of bases was reduced to 984 sites, of which 281 are variable and 256 are parsimony-informative. The ITS alignment also contained several gaps. For both loci, the model GTR+G was chosen by jModeltest.

Species delimitation

COI data: The COI data showed consistency with morphological identifications as specimens determined as C. megalonyx form a clearly delimited monophyletic group. Only two

89 Dissertation Lars Dietz specimens from Kerguelen initially determined as C. megalonyx grouped outside that clade, suggesting that they do not belong to the C. megalonyx complex. The K2P genetic distances show a clear bimodal distribution, with a barcoding gap of approximately 4% (Fig. 2).

GMYC analysis showed that a single-threshold GMYC model was better than a single population model (p=1.83*10-9) and maximum likelihood resulted in a number of 19 ML entities (confidence interval: 18-20), including the six clades already recognized by Krabbe et al. [33]. For 16 of them, material was available, while three clades (J, K, L) were only based on GenBank specimens. The number of samples in each clade ranges from one (clades J,

M) to 156 (clade A). Average intra-clade distances range from 0 to 1.9%, while inter-clade distances range from 2.7 to 12.5%. These are the groupings shown in Table 1.

A multiple-threshold model was preferred to a single-threshold model (p=0.008) and distinguished 29 ML entities (confidence interval 25-32). Clade D1 is split into four, F into three, and D2, E, G, I, and N2 into two entities. The ABGD analysis resulted in only 15 clades, here termed A-O. Clades D and N correspond respectively to three clades in the

GMYC analysis, here named D1/D2/D3 and N1/N2/N3 (see Table 2 for clade delimitations resulting from different approaches).

Bayesian phylogenetic analysis recovered most of the single-threshold GMYC groupings as monophyletic, although B was paraphyletic with respect to M (Fig. 3). While many interclade relationships are poorly resolved, several clades formed strongly supported monophyletic groups, namely A+F+G+(H+I), (B+M)+C, (D+E)+N, and L+O. Interestingly, the clades B

(Falkland Islands), M (Chile), and C (Bouvet) form a Subantarctic group that clusters inside the mostly Antarctic C. megalonyx complex. Similar results are found in the maximum- likelihood analysis, except that clade D1 is found to be paraphyletic with respect to D2 and

D3, and the sister-group relationship between the groups D+E and N1+N2+N3 is not supported.

ITS data: The ITS phylogenetic tree shows a differentiation into five distinct groups named I-

V (Fig. 4), which mostly correspond to larger groupings of different COI clades. The analysis of the dataset cropped with Gblocks results in a slightly different phylogenetic tree, but the

90 Dissertation Lars Dietz differentiation into five groups is not changed. Group IV is paraphyletic with respect to Group

V in the Bayesian and the maximum-likelihood analyses. There is no differentiation between specimens belonging to different COI clades within these groups. As an example, group IV includes individuals belonging to the COI clades A, H and I, but those clades cannot be distinguished by ITS sequences. Group II shows strong intra-group variation, but no division into the COI clades D1 and E can be found. In some cases, there are also discrepancies in assignment to larger groups between COI and ITS. Group II mostly includes individuals from clades D and E, but also one clade C individual. Group III includes individuals from clades

N2 and N3 as well as one each from clade E and G. Individuals from clade I are found both in groups IV and V.

Figure 2: Histogram of uncorrected pairwise genetic distances for the COI fragment within C. megalonyx (all clades).

91 Dissertation Lars Dietz

Figure 3: Bayesian phylogenetic tree of C. megalonyx COI sequences. Clades recognized by GMYC analysis have been collapsed. Numbers above branches are posterior probabilities, numbers below branches are bootstrap percentages for the maximum-likelihood analysis. Numbers beside clade names show number of samples. Colors indicate geographical origin of samples.

92 Dissertation Lars Dietz

Table 1: Number of specimens per COI GMYC clade in each location. The total number may be larger than the sum of numbers for individual regions as some specimens lack locality information. Colours refer to those used in Fig. 1.

S. Sandwich Is. W. Ant. Penins. Amundsen Sea E. Weddell Sea E. Ant. Penins. South Georgia S. Shetland Is. Scott Seamts. Terre Adélie Elephant Is. Falkland Is. S. Orkneys Balleny Is. Bouvet Is. Ross Sea Clade Chile T otal

A 13 21 4 104 1 156 B 21 24 C 11 11 D1 55 14 18 5 5 100 D2 2 2 1 5 D3 6 7 E 47 19 1 24 92 F 2 2 1 1 1 9 G 3 1 1 1 2 5 14 H 4 1 3 1 10 I 6 5 23 1 28 1 68 J 1 1 K 1 1 2 L 2 2 M 1 1 N1 3 3 N2 3 3 2 8 N3 1 2 3 O 2 1 3

93 Dissertation Lars Dietz

Figure 4: Bayesian phylogenetic tree of C. megalonyx ITS sequences. Numbers above branches are posterior probabilities, numbers below branches are bootstrap percentages for the ML analysis. Letters on branches refer to COI clade assignment of the respective specimen. Colours indicate geographical origin of samples.

94 Dissertation Lars Dietz

Phylogeographic analysis

COI data: Most of the clades are geographically widely distributed, and eight of them are found in both East and West Antarctica (Table 1). Of the clades represented by more than three specimens, only clade A (n=156) is restricted to the Scotia Arc, while clades B (n=24) and C (n=11) are restricted to the Falklands/Burdwood Bank and Bouvet Island respectively.

Even representatives of very rare clades, such as K (n=2) and Q (n=3), are found in widely distant regions. None of the clades found in this study seem to show a truly circumpolar distribution as they are notably absent in entire well-sampled regions in our sampling. For example, clade D1, which is known from the South Orkneys/South Shetlands as well as from

Terre Adélie/George V Land, was not found in South Georgia, the South Sandwich Islands or the Eastern Weddell Sea. Specimens of distinct clades often occur sympatrically in the same regions, especially around the South Orkneys (six clades) and the South Shetlands (seven clades). The representatives of different clades often occur in the same place. For instance, at station 11740 in the South Shetlands, 22 individuals were found which belong to five different clades.

The haplotype network for clade A (Fig. 5) shows a “star-like” pattern centred around the common haplotype A-2 (n=104). All other 27 haplotypes from the South Sandwich, South

Orkney and South Shetland Islands differ from it by only 1-3 substitutions and are known from only 1-3 samples (except for A-7 from Elephant Island with n=7). Compared to all other regions there is much more variability in South Georgia, with haplotype A-2 occurring less frequently compared to other regions.

For clade D1, there is a clear differentiation between eastern (Terre Adélie/George V Land, n=6) and western (South Orkneys/South Shetlands/Western Antarctic Peninsula, n=94) specimens, with the latter forming a star-like pattern that is mostly due to the large number of specimens with haplotype D1-2 from the South Orkneys. However, a single specimen from

East Antarctica (haplotype D1-14) groups closer to the western specimens.

In clade E, there is also a strong differentiation between western (South Sandwich/Bouvet, n=62) and eastern (Terre Adélie/George V Land, n=22) samples, with eight steps in

95 Dissertation Lars Dietz between. Interestingly, all specimens from Bouvet (n=43) belong to a single haplotype (E-1), which is also the most common in the South Sandwich Islands. One specimen from the

Eastern Antarctic Peninsula has a haplotype that otherwise occurs in Terre Adélie specimens.

Clade I is divided into two clusters, with four steps in between. One cluster is found only in the South Orkneys/South Shetlands, the other one occurs in the South Shetlands (incl.

Elephant Island) as well as in the Eastern Weddell Sea and in Terre Adélie. However, no haplotypes are shared between the Eastern Weddell Sea and Scotia Arc locations.

Representatives of both clusters were found in the same sampling stations (11719 and

11740).

ITS data: Notably, the ITS sequences of individuals from the same locality often group together even if they belong to different COI clades. As an example, sequences of specimens from Station 260 on the Eastern Antarctic Peninsula, which belong to the mitochondrial clades E, G and P, form a cluster in the ITS phylogenetic tree. One clade C individual from Bouvet has an ITS sequence similar to individuals in clade E from the same location.

Figure 5: Haplotype network for C. megalonyx clades A, D1, E, and I based on the mitochondrial COI gene. Sizes of circles are proportional to number of individuals per haplotype. Colours indicate geographical origin of samples. Grey dots represent hypothetical haplotypes.

96 Dissertation Lars Dietz

Table 2: Differences in COI clade delimitation in C. megalonyx based on single-threshold GMYC and ABGD. Numbers in rightmost column refer to ITS clades to which individuals from the respective COI clade are assigned.

Clade GMYC ABGD ITS group A IV B I C I, II D1 II D2 D3 E II, III F G III, V H IV I IV, V J K L M N1 N2 III N3 III O

Discussion

Number of mitochondrial groupings

Based on the COI data, it appears that the C. megalonyx complex consists of about 15-19 distinct unrecognized species. Different methods (GMYC and ABGD) disagree on the exact delimitation of some clades. These mostly involve those cases in which the distances are at an intermediate level falling into the barcode gap (about 2-5%), i.e. the clades D1+D2+D3 and N1+N2+N3. Interpretation of these as one, two or three distinct groupings should be considered ambiguous. In all other cases, the two methods agree, therefore showing a clear distinction between inter- and intraclade divergence levels, i.e. the presence of a barcoding gap. The multiple-threshold GMYC analysis, however, distinguishes a significantly greater number of clades. We find it hard to accept this result since some of the clades distinguished within clade D1 are separated only by two base substitutions. Our MT-GMYC results agree with the finding of [56] that the multiple-threshold model tends to overestimate the number of

97 Dissertation Lars Dietz clusters. Therefore, only the results obtained from the single-threshold model are further discussed.

Geographic distribution of mitochondrial groups

Specimens of some of the different clades recovered show a narrow distribution range, others are widely distributed and occur in sympatry. This stands in contrast to the findings of

Krabbe et al. [33] who analysed only 96 specimens with less complete geographic sampling than in the present study. The results agree with several other studies from Antarctic benthic species (e.g. [9,16,18] that found circumpolar distributions in some Southern Ocean invertebrate species with molecular data. It should be noted that at least two clades exhibit obvious morphological differences from all others, namely clade C, which lacks eyes, and clade F, which includes animals significantly larger than all others examined. Lack of eyes has previously been reported for the (sub)species C. (megalonyx) orcadense [32,57], but these specimens come from the South Orkneys as well as South Africa and Madagascar, while clade C appears to be restricted to Bouvet Island in our samples. If several species coexist sympatrically, it would be expected that they exhibit ecological differences. Therefore, we expect that, if the mitochondrial clades are indeed distinct species, a detailed study would reveal noticeable morphological or physiological [19] differences.

Implications of the nuclear data

In several instances, the ITS data are incongruent with the COI data. This could be explained by retention of ancestral polymorphisms or by intragenomic polymorphism in ITS, as has been described in other species including crustaceans [58,59]. Moreover, the differences observed between different ITS groups in our study are generally higher than observed in those cases. Besides, we detected no large amounts of conflicting signals in our sequence electropherograms, which would be expected in the case of polymorphic ITS sequences. We also analysed assembled 454 sequence data of the ITS gene region obtained from a preceding project [60]. The assembly had a coverage of 18.6x and showed no evidence for

98 Dissertation Lars Dietz multiple intragenomic variants. Furthermore, the clustering of ITS sequences from individuals found geographically close to each other would also be difficult to explain in the presence of polymorphisms. Therefore, we assume that the best explanation for our results is hybridization between different COI clades. As there seems to be extensive hybridization between related COI clades such as A, H and I, they cannot be considered distinct species.

Between the larger monophyletic groupings such as D+E or A+G+H+I (the latter also including F, for which we lack ITS data), hybridization events appear to be rare, and they may be recognized as distinct species. For the specimens included in the study of Krabbe et al. [33] belonging to clades A-F, the nuclear gene H3 has also been sequenced [61], further supporting the validity of the larger groups A+F, B+C and D+E, but showing no differentiation between the COI clades within those groups.

The occurrence of very similar ITS sequences in specimens from the same site belonging to different COI clades indicates that hybridization is still ongoing, i.e. animals with very divergent mitochondrial genomes seem to belong to the same gene pool. Therefore, the ITS data provide evidence that the number of 15-19 cryptic species inferred from the COI sequences could be a significant overestimation. While the distinct COI clades probably differentiated in isolation from each other, possibly as a consequence of temporary isolation during earlier glaciation periods, apparently there have been no barriers to hybridization after these clades came into contact again. The hybridizing clades have genetic distances for the

COI gene of up to 8%, which according to standard molecular clocks for arthropod taxa

[62,63] would imply a divergence time of more than one million years ago. If that is correct, it appears that in this case a long period of independent evolution did not lead to reproductive isolation in this case. Despite some incongruences between ITS and COI clades, the larger monophyletic groups recognized with COI mostly agree with those recognized with the nuclear gene regions H3 and ITS. Although there are some exceptions, those groupings seem to be largely reproductively isolated and therefore could be regarded as species. The number of known species within the C. megalonyx complex would then be possibly about 5-

99 Dissertation Lars Dietz

7. Limited hybridization between them resulting in mitochondrial-nuclear discordance is similar to that reported from other groups of related species [26].

It is also possible to draw conclusions on the population history in some regions. For instance, within ITS group II, only three out of six examined clade E individuals from Bouvet

Island show ITS sequences highly similar to those of specimen PB_E002 from the South

Sandwich Islands, while the others group more basally within group II. All clade E specimens from Bouvet share a single COI haplotype that is also found on South Sandwich. This might indicate that not all of the Bouvet population originated in a single colonization event from

South Sandwich, as would be inferred from the COI analysis. Instead, there may have been several different colonisations of Bouvet, and the mitochondrial haplotype originating in a recent dispersal from South Sandwich seems to be fixed in the population while the ITS region retains more variability. In general, it is expected that fixation of mitochondrial gene variants occurs faster than in nuclear DNA due to the smaller effective population size [64].

On the one hand, our results contrast with those found in some other marine benthic organisms, including pycnogonids [37], nudibranchs [19], Antarctic isopods [65] and amphipods [9] in which mitochondrial and nuclear data agree on the delimitation of unrecognized species. On the other hand, Hemery et al. [18] found similar results in the

Antarctic crinoid Promachocrinus kerguelensis, in which two major groups were found with both mitochondrial markers and ITS but further differentiation into seven mitochondrial clades was not supported by ITS data. However, the CO1 differentiation between these clades is in most cases lower than in the C. megalonyx complex, and lack of differentiation in ITS may be due to a taxon-specific lower mutation rate in P. kerguelensis. Similar results also occur in species with significantly different life histories, such as the stonefly Dinocras cephalotes

[66], in which two highly divergent COI lineages occur in sympatry but no differentiation was found with nuclear data. In many cases, coexistence of highly divergent mitochondrial lineages within a single species can be explained by introgressive hybridization with other species (e.g. [67]). However, in this study, all mitochondrial haplotypes found within the C.

100 Dissertation Lars Dietz megalonyx complex clearly form a monophyletic group and no introgression from other colossendeid species was found.

Environmental change may lead to the breakdown of ecological barriers between reproductively isolated groups and therefore to “speciation reversal” [68]. Although this has been demonstrated mostly for anthropogenic change, it is possible that glaciations would have the similar effects on the Antarctic shelf by restricting the distribution of benthic organisms to small refugia. This would imply that previously isolated lineages collapsed into a hybrid swarm, which may have lead to strong mitochondrial-nuclear discordance. The question arises why such a pattern is not present in other Antarctic species that have been investigated. Possibly, due to differences in environmental conditions between glacial refugia, selection would have lead to different adaptations [19]. While in some cases these differences were sufficient for reproductive isolation, this was apparently not the case for the

C. megalonyx radiation.

Out of Antarctica hypothesis

We found that there is a monophyletic “Subantarctic“ grouping restricted to South America and Bouvet Island, nested within the Antarctic C. megalonyx complex. This pattern suggests that the Subantarctic was colonized from the Antarctic and not vice versa, as also found e.g. in cephalopods [69]. As the holotype of C. megalonyx is a specimen from the South

American shelf [70], it can be expected to belong to the Subantarctic group, to which the species name should therefore be restricted. C. megalonyx would then lose its status as an

Antarctic pycnogonid, as the species would be restricted to the Subantarctic. Our results provide no support for the hypothesis that C. megalonyx survived the glaciations in refugia in the Subantarctic shelf regions, as no sequences from Antarctic specimens nest within the

Subantarctic clades. However, we lack samples from several non-Antarctic areas where C. megalonyx has been reported, such as South Africa, Kerguelen, and the New Zealand

Subantarctic islands. There is good evidence that South Georgia acted as a refugium for clade A, as shown by the much greater haplotype diversity arguing against a recent

101 Dissertation Lars Dietz expansion, in contrast to the more southern Scotia Arc islands. A similar pattern was recently found by González-Wevar et al. [71] for the limpet Nacella concinna. Geological evidence suggests that South Georgia was not fully glaciated during the Last Glacial Maximum [72], meaning that the South Georgia shelf could plausibly have been a refugium for benthic taxa.

The hypothesis that the shelf was recolonized from the deep sea after the last glacial maximum cannot be rejected by our data, as we lack samples from deeper than about 1300 m. However, we consider it unlikely, as circumpolar survival in the deep sea would lead to greater genetic homogeneity across regions and lack of signatures for recent expansion, as seen in the shrimp Nematocarcinus lanceopes [73], but not in our data for C. megalonyx.

The hypothesis most consistent with our data is the in situ survival in ice-free refugia.

Because of the strong intra-clade regional differentiation seen e.g. within clades D1 and E, it seems likely that these clades survived in more than one refugium during the Last Glacial

Maximum (LGM), spreading from there and in some cases (clade I) coming into secondary contact. Our data support spreading with the ACC at least in the case of clade E, which apparently colonized Bouvet from the South Sandwich Islands, indicating a relatively recent

(only one haplotype known from Bouvet) eastward spreading in latitudes dominated by the

ACC. Survival in multiple refugia would indicate that interclade splits precede the LGM, and probably occurred during earlier Pleistocene glaciations or even earlier.

In a few cases, we observe the same haplotype in geographically widely separated regions, such as a clade E haplotype (E-3) that occurs both in the Antarctic Peninsula and Terre

Adélie. This has also been observed in other invertebrates without a planktonic stage [9,74] and might be explained by rafting on floating material carried by currents, including ice.

Pycnogonids have also been observed swimming [75].

The strong regional differentiation, which apparently persisted since the LGM, is typical of benthic brooding organisms with limited dispersal capability. Adult pycnogonids are almost exclusively benthic, the reproduction mode of colossendeids is unknown and no larvae have been recorded from plankton samples. The distribution of C. megalonyx contrasts with that of

102 Dissertation Lars Dietz benthic invertebrates with planktonic larvae such as the crinoids Promachocrinus kerguelensis, whose lineages mostly show a truly circumpolar and sympatric distribution [18].

Conclusions

Our largely expanded and multi-locus data set supports that Colossendeis megalonyx is a complex of several overlooked species, many with broad geographic distribution ranges.

However, the analysis of a highly variable nuclear data in addition to the mitochondrial COI gene suggests that the number of actually overlooked species is smaller that the number of mitochondrial clades. This highlights the importance of including independent nuclear markers in such analyses. The taxonomy of the C. megalonyx complex may be further clarified by including nuclear data from other genes as well as morphological data. Next- generation sequencing technologies, which have the potential to sequence large numbers of loci at once, could be particularly useful in resolving this and similar questions.

Data accessibility

All new sequence data have been deposited to GenBank (accession numbers: XXX will be updated prior to publication).

Competing interests

We have no competing interests.

Authors’ contributions

FL and LD conceived the study. LD carried out the molecular lab work, performed data analyses and wrote the manuscript together with FL CM and CH participated in the design of the study. CPA, KH, AMH, CH, ARM, RRM, GR, AW and NW provided specimens and/or sequence data. CPA, CH, ARM, CM, RRM and NW helped drafting the MS. All authors gave final approval for publication.

103 Dissertation Lars Dietz

Acknowledgments

We thank Ralph Tollrian for support and the EvoEco Journal Club for helpful comments on the manuscript. We also thank Susie Lockhart for assistance. Furthermore, we thank all participants on the NBP1105 RV/IB Nathaniel B. Palmer and ANT-XXVIII/4 FS Polarstern cruises.

Funding

This work was supported by DFG grant LE 2323/2 to FL, CM and CH and in part by Sea Life

Center Munich (research grant “Biodiversity of the Chilean fjords” to RRM. Collection of material was supported by the US National Science Foundation Office of Polar Programs

Antarctic Organisms and Ecosystems Program (ANT-1043749 to NGW), Scripps Institution of Oceanography, and the US Antarctic Marine Living Resources program.

104 Dissertation Lars Dietz

References

[1] Griffiths H. 2010 Antarctic marine biodiversity--what do we know about the distribution

of life in the Southern Ocean? PLoS ONE, 5, e11683.

[2] Kaiser S, Brandão S, Brix S, Barnes D, Bowden D, Ingels J, Leese F, Schiaparelli S,

Arango C, Badhe R et al. 2013 Patterns, processes and vulnerability of Southern

Ocean benthos: a decadal leap in knowledge and understanding. Mar Biol, 1, 2295-

2317.

[3] De Broyer C and Danis B. 2011 How many species in the Southern Ocean? Towards

a dynamic inventory of the Antarctic marine species. Deep Sea Research Part II

Topical Studies in Oceanography, 58, 5-17.

[4] Gutt J, Sirenko B, Smirnov I and Arntz W. 2004 How many macrozoobenthic species

might inhabit the Antarctic shelf? Antarct Sci, 16, 11–26.

[5] Janosik A and Halanych K. 2010 Unrecognized Antarctic Biodiversity: A Case Study

of the Genus Odontaster (Odontasteridae; Asteroidea). Integrative and Comparative

Biology, 50, 981–992.

[6] Held C. 2003 Molecular evidence for cryptic speciation within the widespread

Antarctic crustacean Ceratoserolis trilobitoides (Crustacea, Isopoda). In: AHL

Huiskes, WWC Gieskes, J Rozema, RML Schorno, SM van Der Vies & WJ Wolff

(eds) “Antarctic Biology in a Global Context,” 135–139.

[7] Held C and Wägele J-W. 2005 Cryptic speciation in the giant Antarctic isopod

Glyptonotus antarcticus (Isopoda: Valvifera: Chaetiliidae). Scientia Marina, 69 (Suppl.

2), 175–181.

[8] Linse K, Cope T, Lorz A-N and Sands C. 2007 Is the Scotia Sea a centre of Antarctic

marine diversification? Some evidence of cryptic speciation in the circum-Antarctic

bivalve Lissarca notorcadensis (Arcoidea: Philobryidae). Polar Biol, 30, 1059–1068.

[9] Baird H, Miller K and Stark J. 2011 Evidence of hidden biodiversity, ongoing

speciation and diverse patterns of genetic structure in giant Antarctic amphipods. Mol

Ecol, 20, 3439–3454.

105 Dissertation Lars Dietz

[10] Allcock A, Barratt I, Eleaume M, Linse K, Norman M, Smith P, Steinke D, Stevens D

and Strugnell J. 2010 Cryptic speciation and the circumpolarity debate: A case study

on endemic Southern Ocean octopuses using the COI barcode of life. Deep-Sea Res

II, 58, 242-249.

[11] Dayton P, Mordida B and Bacon F. 1994 Polar marine communities. American

Zoologist, 34, 90–9.

[12] Clarke A and Johnston N. 2003 Antarctic marine benthic diversity. Oceanography and

Marine Biology: An Annual Review, 41, 47-114.

[13] Knox G. 2007 Biology of the Southern Ocean (Second Edition). CRC press, 531 pp.

[14] Brey T, Dahm C, Gorny M, Klages M, Stiller M and Arntz W. 1996 Do Antarctic

benthic invertebrates show an extended level of eurybathy? Antarct Sci, 8, 3–6.

[15] Thatje S. 2012 Effects of capability for dispersal on the evolution of diversity in

Antarctic benthos. Integrative and Comparative Biology, 52, 470–482.

[16] Arango C, Soler-Membrives A and Miller K. 2011 Genetic differentiation in the

circum—Antarctic sea spider Nymphon australe (Pycnogonida; Nymphonidae). Deep

Sea Research Part II: Topical Studies in Oceanography, 58, 212–219.

[17] Mahon A, Arango C and Halanych K. 2008 Genetic diversity of Nymphon

(Arthropoda: Pycnogonida: Nymphonidae) along the Antarctic Peninsula with a focus

on Nymphon australe Hodgson 1902. Mar Biol, 155, 315–323.

[18] Hemery L, Eléaume M, Roussel V, Améziane N, Gallut C, Steinke D, Cruaud C,

Couloux A and Wilson N. 2012 Comprehensive sampling reveals circumpolarity and

sympatry in seven mitochondrial lineages of the Southern Ocean crinoid species

Promachocrinus kerguelensis (Echinodermata). Mol Ecol, 21, 2502–2518.

[19] Wilson N, Maschek J and Baker B. 2013 A species flock driven by predation?

Secondary metabolites support diversification of slugs in Antarctica. PLoS One, 8,

e80277.

[20] Schüller M. 2011 Evidence for a role of bathymetry and emergence in speciation in

the genus Glycera (Glyceridae, Polychaeta) from the deep Eastern Weddell Sea.

106 Dissertation Lars Dietz

Polar Biol, 34, 549–64.

[21] Brandao S, Sauer J and Schön I. 2010 Circumantarctic distribution in Southern

Ocean benthos? A genetic test using the genus Macroscapha (Crustacea, Ostracoda)

as a model. Mol Phylogenet Evol, 55, 1055–1069.

[22] Lörz A, Maas E, Linse K and Coleman C. 2009 Do circum-Antarctic species exist in

peracarid Amphipoda? A case study in the genus Epimeria Costa, 1851 (Crustacea,

Peracarida, Epimeriidae). ZooKeys, 18, 91.

[23] Lörz A-N, Smith P, Linse K and Steinke D. 2012 High genetic diversity within

Epimeria georgiana (Amphipoda) from the southern Scotia Arc. Mar Biodiv, 42, 137–

159.

[24] Weis A and Melzer R. 2012 How did sea spiders recolonize the Chilean fjords after

glaciation? DNA barcoding of Pycnogonida, with remarks on phylogeography of

Achelia assimilis (Haswell, 1885). Systematics and Biodiversity, 10, 361–374.

[25] Weis A, Meyer R, Dietz L, Dömel J, Leese F and Melzer R. 2014 Pallenopsis

patagonica (Hoek, 1881) – a species complex revealed by morphology and DNA

barcoding, with description of a new species of Pallenopsis Wilson, 1881. Zool J Linn

Soc, 170, 110–131.

[26] Toews D and Brelsford A. 2012 The biogeography of mitochondrial and nuclear

discordance in animals. Mol Ecol, 21, 3907–3930.

[27] Griffiths H, Arango C, Munilla T and McInnes S. 2011 Biodiversity and biogeography

of Southern Ocean pycnogonids. Ecography, 34, 616–27.

[28] Munilla T and Soler Membrives A. 2009 Check-list of the pycnogonids from Antarctic

and sub-Antarctic waters: zoogeographic implications. Antarct Sci, 21, 99.

[29] Arnaud F and Bamber R. 1987 The biology of Pycnogonida. Advances in Marine

Biology, 24, 1.

[30] Child C. 1995 Antarctic and Subantartic Pycnogonida: The Families Nymphonidae,

Colossendeidae, Rhynchothoraxidae, Pycnogonidae, Endeididae, and

Callipallenidae. Antarctic Research Series, 69, 1.

107 Dissertation Lars Dietz

[31] Hodgson T. 1907 Pycnogonida. Natural History Collections of the “Discovery”

National Antarctic Expedition, Zoology, 3, 1.

[32] Fry W and Hedgpeth J. 1969 The fauna of the Ross Sea. Part 7. Pycnogonida, 1:

Colossendeidae, Pycnogonidae, Endeidae, Ammotheidae. New Zealand Department

of Science and Industrial Research Bulletin, 198, 1.

[33] Krabbe K, Leese F, Mayer C, Tollrian R and Held C. 2010 Cryptic mitochondrial

lineages in the widespread pycnogonid Colossendeis megalonyx Hoek, 1881 from

Antarctic and Subantarctic waters. Polar Biol, 33, 281–292.

[34] Odorico D and Miller D. 1997 Variation in the ribosomal internal transcribed spacers

and 5.8S rDNA among five species of Acropora (Cnidaria; Scleractinia): patterns of

variation consistent with reticulate evolution. Mol Biol Evol, 14, 465–473.

[35] Navajas M and Boursot P. 2003 Nuclear ribosomal DNA monophyly versus

mitochondrial DNA polyphyly in two closely related mite species: the influence of life

history and molecular drive. Proc R Soc Lond B, 270, S124–127.

[36] Cheng H-L, Xia D-Q, Wu T-T, Meng X-P, Ji H-J and Dong Z-G. 2006 Study on

sequences of ribosomal DNA internal transcribed spacers of clams belonging to the

Veneridae family (Mollusca: Bivalvia). Acta Genetica Sinica, 33, 702–710.

[37] Arango C and Brenneis G. 2013 New species of Australian Pseudopallene

(Pycnogonida: Callipallenidae) based on live colouration, morphology and DNA.

Zootaxa, 3616, 401–36.

[38] Pushkin A. 1993 The Pycnogonida fauna of the South Ocean (Biological results of the

Soviet Antarctic Expeditions) - Fauna mnogokolencat/x (Pycnogonida) {jnogo okeana

(Resultat/ biologceskix issledovaniy sovetskix antarkticesix Ekspediziy).

[39] Wörheide G. 1998 The reef cave dwelling ultraconservative coralline demosponge

Astrosclera willeyana Lister 1900 from the Indo-Pacific. Facies, 38, 1.

[40] Folmer O, Black M, Hoeh W, Lutz R and Vrijenhoek R. 1994 DNA primers for

amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan

invertebrates. Mol Mar Biol Biotechnol, 3, 294–299.

108 Dissertation Lars Dietz

[41] Drummond A, Ashton B, Buxton S, Cheung M, Cooper A, Duran C, Field M, Heled J,

Kearse M, Markowitz S et al. 2011 Geneious v5.4.6, Available from

http://www.geneious.com.

[42] Edgar R. 2004 MUSCLE: multiple sequence alignment with high accuracy and high

throughput. Nucl Acids Res, 32 (5), 1792–1797.

[43] Katoh K and Standley D. 2013 MAFFT multiple sequence alignment software version

7: improvements in performance and usability. Mol Biol Evol, 30, 772–780.

[44] Talavera G and Castresana J. 2007 Improvement of phylogenies after removing

divergent and ambiguously aligned blocks from protein sequence alignments. Syst

Biol, 56, 564–577.

[45] Ronquist F, Teslenko M, van der Mark P, Ayres D, Darling A, Höhna S, Larget B, Liu

L, Suchard M and Huelsenbeck J. 2012 MrBayes 3.2: efficient Bayesian phylogenetic

inference and model choice across a large model space. Systematic Biology, 61,

539–542.

[46] Darriba D, Taboada G, Doallo R and Posada D. 2012 jModelTest 2: more models,

new heuristics and parallel computing. Nat Methods, 9, 772.

[47] Stamatakis A. 2006 RAxML-VI-HPC: maximum likelihood-based phylogenetic

analyses with thousands of taxa and mixed models. Bioinformatics, 22, 2688–2690.

[48] Villesen P. 2007 FaBox: an online toolbox for fasta sequences. Mol Ecol Notes, 7,

965–968.

[49] Swofford D. 2002 PAUP*. Phylogenetic Analysis Using Parsimony (*and Other

Methods). Version 4. Sinauer Associates, Sunderland, Massachusetts, ii + 142 pp.

[50] Drummond A, Suchard M, Xie D and Rambaut A. 2012 Bayesian phylogenetics with

BEAUti and the BEAST 1.7. Mol Biol Evol, 29, 1969–73.

[51] Rambaut A, Suchard M, Xie D and Drummond A. 2013 Tracer v1.5, Available from

http://beast.bio.ed.ac.uk/Tracer.

[52] Rambaut A and Drummond A. 2013 TreeAnnotator v1. 7.0.

[53] Ezard T, Fujisawa T and Barraclough T. 2009 Splits: species’ limits by threshold

109 Dissertation Lars Dietz

statistics. R Package Version, 1.

[54] Puillandre N, Lambert A, Brouillet S and Achaz G. 2012 ABGD, Automatic Barcode

Gap Discovery for primary species delimitation. Mol Ecol, 21, 1864–1877.

[55] Clement M, Posada D and Crandall K. 2000 TCS: a computer program to estimate

gene genealogies. Mol Ecol, 9, 1657–1659.

[56] Fujisawa T and Barraclough T. 2013 Delimiting species using single-locus data and

the generalized mixed Yule coalescent (GMYC) approach: a revised method and

evaluation on simulated datasets. Syst Biol, 62, 707-724.

[57] Hodgson T. 1908 VI.—The Pycnogonida of the Scottish National Antarctic Expedition.

Transactions of the Royal Society of Edinburgh, 46, 159–188.

[58] Harris D and Crandall K. 2000 Intragenomic variation within ITS1 and ITS2 of crayfish

(Decapoda: Cambaridae): implications for phylogenetic and microsatellite studies.

Mol Biol Evol, 17, 284–291.

[59] Chu K, Li C and Ho H. 2001 The first internal transcribed spacer (ITS-1) of ribosomal

DNA as a molecular marker for phylogenetic and population analyses in Crustacea.

Mar Biotechnol, 3, 355–61.

[60] Leese F, Brand P, Rozenberg A, Mayer C, Agrawal S, Dambach J, Dietz L, Doemel J,

Goodall-Сopstake W, Held C et al. 2012 Exploring Pandora’s box: potential and

pitfalls of low coverage genome surveys for evolutionary biology. PLoS ONE, 7,

e49202.

[61] Dietz L, Krapp F, Hendrickx M, Arango C, Krabbe K, Spaak J and Leese F. 2013

Evidence from morphological and genetic data confirms that Colossendeis tenera

Hilton, 1943 (Arthropoda: Pycnogonida), does not belong to the Colossendeis

megalonyx Hoek, 1881 complex. Organisms Diversity & Evolution, 13, 151–162.

[62] Brower A. 1994 Rapid morphological radiation and convergence among races of the

butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolution. Proc

Natl Acad Sci USA, 91, 6491.

[63] Schubart C, Diesel R and Hedges S. 1998 Rapid evolution to terrestrial life in

110 Dissertation Lars Dietz

Jamaican crabs. Nature, 393, 363–365.

[64] Zink R and Barrowclough G. 2008 Mitochondrial DNA under siege in avian

phylogeography. Mol Ecol, 17, 2107–2121.

[65] Leese F and Held C. 2008 Identification and characterization of microsatellites from

the Antarctic isopod Ceratoserolis trilobitoides: nuclear evidence for cryptic species.

Conserv Genet, 9 (5), 1369–72.

[66] Elbrecht V, Feld C, Gies M, Hering D, Sondermann M, Tollrian R and Leese F. 2014

Genetic diversity and dispersal potential of the stonefly Dinocras cephalotes in a

central European low mountain range. 33, 181–92.

[67] Darling J. 2011 Interspecific hybridization and mitochondrial introgression in invasive

Carcinus shore crabs. PLoS ONE, 6, e17828.

[68] Seehausen O, Takimoto G, Roy D and Jokela J. 2008 Speciation reversal and

biodiversity dynamics with hybridization in changing environments. Mol Ecol, 17, 30–

44.

[69] Strugnell J, Rogers A, Prodöhl P, Collins M and Allcock A. 2008 The thermohaline

expressway: the Southern Ocean as a centre of origin for deep-sea octopuses.

Cladistics, 24, 853–860.

[70] Hoek P. 1881 Report on the Pycnogonida, dredged by HMS Challenger during the

Years 1873-76. Rep Sc Res Voy Hms Challenger Zool, 3, 1.

[71] González-Wevar C, Saucède T, Morley S, Chown S and Poulin E. 2013 Extinction

and recolonization of maritime Antarctica in the limpet Nacella concinna (Strebel,

1908) during the last glacial cycle: toward a model of Quaternary biogeography in

shallow Antarctic invertebrates. Mol Ecol, 22, 5221-5236.

[72] Hodgson D, Graham A, Roberts S, Bentley M, Ó Cofaigh C, Verleyen E, Vyverman

W, Jomelli V, Favier V, Brunstein D et al. 2014 Terrestrial and submarine evidence for

the extent and timing of the Last Glacial Maximum and the onset of deglaciation on

the maritime-Antarctic and sub-Antarctic islands. Quaternary Science Reviews, 100,

137–158.

111 Dissertation Lars Dietz

[73] Raupach M, Thatje S, Dambach J, Rehm P, Misof B and Leese F. 2010 Genetic

homogeneity and circum-Antarctic distribution of two benthic shrimp species of the

Southern Ocean, Chorismus antarcticus and Nematocarcinus lanceopes. Mar Biol,

157, 1783–1797.

[74] Wilson N, Schrödl M and Halanych K. 2009 Ocean barriers and glaciation: evidence

for explosive radiation of mitochondrial lineages in the Antarctic sea slug Doris

kerguelenensis (Mollusca, Nudibranchia). Mol Ecol, 18, 965–84.

[75] Clark W and Carpenter A. 1977 Swimming behaviour in a pycnogonid. New Zealand

Journal of Marine and Freshwater Research, 11, 613–615.

112 Dissertation Lars Dietz

Electronic Supplementary Materials

ESM1: Sampling sites details

ESM2: GenBank sequences included in this study

113 ESM 1: Sampling sites details specimen number region lat long depth cruise clade haplotype 41BT22-1 South Georgia -54,22375 -36,534 230 ICEFISH A A-1 57OT42 South Sandwich -57,0708 -26,7733 120 ICEFISH A A-2 BT32-1 South Sandwich -57,0567 -26,7483 130 ICEFISH A A-2 11341 Elephant Island -61,3387 -55,6248 143-162 A A-2 11342 Elephant Island -61,3387 -55,6248 143-162 A A-2 11343 Elephant Island -61,3387 -55,6248 143-162 A A-26 11357 Elephant Island -61,2066 -54,2522 188-196 A A-2 11367 Elephant Island -61,2066 -54,2522 188-196 A A-23 11368 Elephant Island -61,3387 -55,6248 143-162 A A-2 11371 Elephant Island -61,2066 -54,2522 188-196 A A-2 11376 Elephant Island -61,3387 -55,6248 143-162 A A-2 11382 Elephant Island -61,2066 -54,2522 188-196 A A-13 11385 Elephant Island -61,3387 -55,6248 143-162 A A-2 11388 Elephant Island -61,2066 -54,2522 188-196 A A-2 11389 Elephant Island -61,2066 -54,2522 188-196 A A-2 11396 Elephant Island -61,3387 -55,6248 143-162 A A-2 11399 Elephant Island -61,3387 -55,6248 143-162 A A-2 11463 Elephant Island -61,2065 -54,2522 188-196 A A-2 11470 Elephant Island -61,2065 -54,2522 188-196 A A-7 11482 Elephant Island -61,2065 -54,2522 188-196 A A-2 11485 Elephant Island -61,2065 -54,2522 188-196 A A-24 11486 Elephant Island -61,2065 -54,2522 188-196 A A-2 11488 Elephant Island -61,2065 -54,2522 188-196 A A-2 11503 Elephant Island -61,2065 -54,2522 188-196 A A-2 11514 Elephant Island -61,2065 -54,2522 188-196 A A-2 11535 Elephant Island -61,2065 -54,2522 188-196 A A-2 11554 Elephant Island -61,2065 -54,2522 188-196 A A-2 11370 Elephant Island -61,3387 -55,6248 143-162 A A-2 11416 Elephant Island -61,3387 -55,6248 143-162 A A-2 11417 Elephant Island -61,2065 -54,2522 188-196 A A-2 11440 Elephant Island -61,2065 -54,2522 188-196 A A-22 11441 Elephant Island -61,2065 -54,2522 188-196 A A-16 11728 Elephant Island -60,8693 -55,5017 251-248 A A-36 11731 Elephant Island -60,8693 -55,5017 251-248 A A-2 11719-11 Elephant Island -61,187 -54,5883 278-356 A A-32 11719-13 Elephant Island -61,187 -54,5883 278-356 A A-2 11719-6 Elephant Island -61,187 -54,5883 278-356 A A-2 11716-10 Elephant Island -61,3343 -55,5277 149-160 A A-31 11716-11 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-12 Elephant Island -61,3343 -55,5277 149-160 A A-8 11716-16 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-17 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-18 Elephant Island -61,3343 -55,5277 149-160 A A-35 11716-19 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-2 Elephant Island -61,3343 -55,5277 149-160 A A-9 11716-20 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-21 Elephant Island -61,3343 -55,5277 149-160 A A-7 11716-23 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-24 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-3 Elephant Island -61,3343 -55,5277 149-160 A A-28 11716-4 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-5 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-6 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-7 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-8 Elephant Island -61,3343 -55,5277 149-160 A A-2 11716-9 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-1 Elephant Island -61,3343 -55,5277 149-160 A A-7 11717-11 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-12 Elephant Island -61,3343 -55,5277 149-160 A A-7 11717-13 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-14 Elephant Island -61,3343 -55,5277 149-160 A A-7 11717-15 Elephant Island -61,3343 -55,5277 149-160 A A-20 11717-17 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-18 Elephant Island -61,3343 -55,5277 149-160 A A-10 11717-19 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-2 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-20 Elephant Island -61,3343 -55,5277 149-160 A A-21 11717-21 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-22 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-23 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-25 Elephant Island -61,3343 -55,5277 149-160 A A-7 11717-26 Elephant Island -61,3343 -55,5277 149-160 A A-7 11717-27 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-28 Elephant Island -61,3343 -55,5277 149-160 A A-12 11717-29 Elephant Island -61,3343 -55,5277 149-160 A A-10 11717-3 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-30 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-31 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-32 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-33 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-34 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-35 Elephant Island -61,3343 -55,5277 149-160 A A-8 11717-36 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-37 Elephant Island -61,3343 -55,5277 149-160 A A-9 11717-39 Elephant Island -61,3343 -55,5277 149-160 A A-13 11717-4 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-40 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-41 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-42 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-43 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-44 Elephant Island -61,3343 -55,5277 149-160 A A-11 11717-45 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-47 Elephant Island -61,3343 -55,5277 149-160 A A-2 11717-5 Elephant Island -61,3343 -55,5277 149-160 A A-27 11717-6 Elephant Island -61,3343 -55,5277 149-160 A A-29 11717-7 Elephant Island -61,3343 -55,5277 149-160 A A-30 11344 South Georgia -55,081 -35,1725 253-196 A A-2 11345 South Georgia -55,081 -35,1725 253-196 A A-2 11360 South Georgia -55,081 -35,1725 253-196 A A-18 11369 South Georgia -55,081 -35,1725 253-196 A A-19 11372 South Georgia -55,081 -35,1725 253-196 A A-2 11380 South Georgia -55,081 -35,1725 253-196 A A-19 11381 South Georgia -55,081 -35,1725 253-196 A A-19 11458 South Georgia -55,081 -35,1725 253-196 A A-15 11502 South Georgia -55,081 -35,1725 253-196 A A-17 11505 South Georgia -55,081 -35,1725 253-196 A A-15 11508 South Georgia -55,081 -35,1725 253-196 A A-14 11531 South Georgia -55,081 -35,1725 253-196 A A-2 11468 South Orkneys -60,613 -45,1417 81-96 A A-2 11515 South Orkneys -60,613 -45,1417 81-96 A A-2 11660-1 South Orkneys -60,613 -45,1417 81-96 A A-2 11660-2 South Orkneys -60,613 -45,1417 81-96 A A-2 11335 South Sandwich 103-221 A A-2 11390 South Sandwich -57,0365 -26,7902 128-175 A A-2 11461 South Sandwich -59,3936 -27,3227 110-188 A A-2 11498 South Sandwich -58,468 -26,218 172-164 A A-2 11418 South Sandwich -59,3936 -27,3227 110-188 A A-25 11649-1 South Sandwich -58,3773 -26,2823 153-420 A A-2 11649-10 South Sandwich -58,3773 -26,2823 153-420 A A-2 11649-12 South Sandwich -58,3773 -26,2823 153-420 A A-2 11649-13 South Sandwich -58,3773 -26,2823 153-420 A A-2 11649-2 South Sandwich -58,3773 -26,2823 153-420 A A-34 11649-3 South Sandwich -58,3773 -26,2823 153-420 A A-2 11504 South Shetlands -62,3558 -60,7305 164-166 A A-2 11745 A A-2 11756 A A-7 11758 A A-7 S0933 A A-2 S6735 A A-2 S6790 A A-13 S6797 A A-2 S6798 A A-2 S6799 A A-33 PS61/45-2 Elephant Island -60,9856 -55,1897 196 Polarstern 61 A A-2 PS61/45-3 Elephant Island -60,9856 -55,1897 196 Polarstern 61 A A-2 PS61/45-4 Elephant Island -60,9856 -55,1897 196 Polarstern 61 A A-2 #87-21 A A-2 11473 Burdwood Bank -54,665 -61,2357 302-337 B B-2 11479 Burdwood Bank -54,665 -61,2357 302-337 B B-1 11558 Burdwood Bank -54,665 -61,2357 302-337 B B-1 208-5.2 Burdwood Bank -54,54683 -56,16667 294 Polarstern 77 B B-1 PO_E006 Burdwood Bank -52,6583 -60,2892 202 B B-2 76OT50 Bouvet -54,6371 3,3046 648 ICEFISH C C-1 102,25 Western Antarctic Peninsula -63,6691 -61,16745 D1 D1-29 122,15 Western Antarctic Peninsula -67,7403 -69,28965 D1 D1-2 C5 Western Antarctic Peninsula D1 D1-13 C8 Western Antarctic Peninsula D1 D1-30 C11 Western Antarctic Peninsula D1 D1-7 11719-4 Elephant Island -61,187 -54,5883 278-356 D1 D1-3 11719-7 Elephant Island -61,187 -54,5883 278-356 D1 D1-9 11719-8 Elephant Island -61,187 -54,5883 278-356 D1 D1-3 11719-9 Elephant Island -61,187 -54,5883 278-356 D1 D1-7 11717-16 Elephant Island -61,3343 -55,5277 149-160 D1 D1-3 11717-38 Elephant Island -61,3343 -55,5277 149-160 D1 D1-7 11717-46 Elephant Island -61,3343 -55,5277 149-160 D1 D1-2 11349 South Shetlands -62,3558 -60,7305 164-166 D1 D1-15 11536 South Shetlands -62,3052 -60,7687 259-253 D1 D1-25 11740-10 South Shetlands -62,0918 -60,5322 286-364 D1 D1-2 11740-19 South Shetlands -62,0918 -60,5322 286-364 D1 D1-7 11753 D1 D1-6 S1269 D1 D1-26 PS61/45-5 Elephant Island -60,9856 -55,1897 196 Polarstern 61 D1 D1-27 IU-2007-31b(1) not 316Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC D1 D1-17 IU-2007-31b(6) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC D1 D1-17 IU-2007-50 Terre Adelie/George V Land -65,996851 142,658763 447,35 CEAMARC D1 D1-17 IU-2007-249b Terre Adelie/George V Land -65,700372 140,568145 544,95 CEAMARC D1 D1-23 IU-2007-255 Terre Adelie/George V Land -66,342595 139,994179 567,75 CEAMARC D1 D1-16 YPM48435-2 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-3 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-9 YPM48435-4 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-5 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-12 YPM48435-6 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-7 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-8 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-9 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-6 YPM48435-12 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-13 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-15 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-11 YPM48435-16 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-17 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-18 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-19 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-18 YPM48435-23 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-24 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-27 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-30 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-31 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-19 YPM48435-34 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-35 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-38 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-10 YPM48435-39 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-40 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-41 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-43 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-44 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-21 YPM48435-45 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-46 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-47 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-48 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-12 YPM48435-49 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-22 YPM48435-52 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-54 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-6 YPM48435-55 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-20 YPM48435-56 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-57 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-24 YPM48435-58 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-6 YPM48435-59 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-60 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-61 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-6 YPM48435-63 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-66 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-68 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-69 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-6 YPM48435-70 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-73 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-28 YPM48435-74 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-75 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-76 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-77 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-79 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 YPM48435-80 South Orkneys -60,40005 -46,7562 280 Yale D1 D1-2 AGT42/175-9 South Shetlands -62,31667 -58,7 496 Polarstern 42 D1 D1-1 AGT42/175-11 South Shetlands -62,31667 -58,7 496 Polarstern 42 D1 D1-2 AGT42/164-1 South Shetlands -62,13333 -57,66667 555 Polarstern 42 D1 D1-5 AGT42/164-2 South Shetlands -62,13333 -57,66667 555 Polarstern 42 D1 D1-4 AGT42/164-3 South Shetlands -62,13333 -57,66667 555 Polarstern 42 D1 D1-3 PM_E012 D1 D1-8 CEA035 CEAMARC D1 D1-14 174.2.1 Amundsen Sea -72,7805 -104,5538 496 D2 D2-3 177.20.1 Amundsen Sea -73,7215 -103,6186 D2 D2-2 215.36.2 Ross Sea -75,3296 -176,9859 D2 D2-1 260.4.1 Ross Sea -76,2453 174,5041 D2 D2-4 161,1 Amundsen Sea -71,75245 -102,2349 459 D3 D3-1 161,2 Amundsen Sea -71,75245 -102,2349 459 D3 D3-1 170.1.3 Amundsen Sea -72,7805 -104,5538 496 D3 D3-3 170,1 Amundsen Sea -72,7805 -104,5538 496 D3 D3-2 170,2 Amundsen Sea -72,7805 -104,5538 496 D3 D3-2 170.3.3 Amundsen Sea -72,7805 -104,5538 496 D3 D3-2 59BT40-1 Bouvet -54,3792 3,1292 465 ICEFISH E E-1 76OT50-1 Bouvet -54,6371 3,3046 648 ICEFISH E E-1 76OT50-2 Bouvet -54,6371 3,3046 648 ICEFISH E E-1 11346 South Sandwich -59,3831 -27,3454 926-701 E E-4 11347 South Sandwich -59,3831 -27,3454 926-701 E E-7 11353 South Sandwich -59,3831 -27,3454 926-701 E E-1 11361 South Sandwich -59,3831 -27,3454 926-701 E E-8 11363 South Sandwich -59,3936 -27,3227 110-188 E E-4 11378 South Sandwich -59,3831 -27,3454 926-701 E E-8 11384 South Sandwich -59,3831 -27,3454 926-701 E E-6 11509 South Sandwich -59,3831 -27,3454 926-701 E E-1 11562 South Sandwich -59,3831 -27,3454 926-701 E E-5 11565 South Sandwich -59,3831 -27,3454 926-701 E E-4 11649-15 South Sandwich -58,3773 -26,2823 153-420 E E-1 11649-17 South Sandwich -58,3773 -26,2823 153-420 E E-1 11649-5 South Sandwich -58,3773 -26,2823 153-420 E E-1 11649-6 South Sandwich -58,3773 -26,2823 153-420 E E-1 11649-8 South Sandwich -58,3773 -26,2823 153-420 E E-1 257-2.5.2 Eastern Antarctic Peninsula -64,9125 -60,65017 158 Polarstern 77 E E-3 IU-2007-26 CEAMARC E E-12 IU-2007-31b(2) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC E E-2 IU-2007-31b(3) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC E E-2 IU-2007-31b(4) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC E E-2 IU-2007-31b(5) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC E E-3 IU-2007-31b(7) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC E E-3 IU-2007-31b(8) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC E E-2 IU-2007-31b(9) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC E E-3 IU-2007-31b(10) Terre Adelie/George V Land -65,865805 144,110923 1096,35 CEAMARC E E-10 IU-2007-37d Terre Adelie/George V Land -65,470932 139,355465 799,45 CEAMARC E E-2 IU-2007-39 Terre Adelie/George V Land -65,999508 143,000647 470,05 CEAMARC E E-2 IU-2007-51b Terre Adelie/George V Land -65,996851 142,658763 447,35 CEAMARC E E-3 IU-2007-62(1) Terre Adelie/George V Land -66,325528 143,273265 697,8 CEAMARC E E-9 IU-2007-62(3) Terre Adelie/George V Land -66,325528 143,273265 697,8 CEAMARC E E-2 IU-2007-62(4) Terre Adelie/George V Land -66,325528 143,273265 697,8 CEAMARC E E-2 IU-2007-62(5) Terre Adelie/George V Land -66,325528 143,273265 697,8 CEAMARC E E-11 IU-2007-73 Terre Adelie/George V Land -66,745493 145,001929 648,3 CEAMARC E E-2 IU-2007-248 Terre Adelie/George V Land -65,700372 140,568145 544,95 CEAMARC E E-2 IU-2007-249g Terre Adelie/George V Land -65,700372 140,568145 544,95 CEAMARC E E-2 IU-2007-287 Terre Adelie/George V Land -66,75 145,3333 567-604 CEAMARC E E-2 CEA020 CEAMARC E E-3 CEA021 CEAMARC E E-2 #87-20 E E-2 11348 South Orkneys -60,5508 -45,176 278-222 F F-2 11501 South Orkneys -60,5508 -45,176 278-222 F F-2 AGT42/175-5 South Shetlands -62,31667 -58,7 496 Polarstern 42 F F-1 286-1.5 Eastern Weddell Sea -70,844 -10,60183 248 Polarstern 77 F F-3 IU-2007-41a Terre Adelie/George V Land -66,45 140,5333 855-1204 CEAMARC F F-4 214.1.1 Ross Sea -75,3296 -176,9859 G G-6 214.10.11 Ross Sea -75,3296 -176,9859 G G-8 214.4.1 Ross Sea -75,3296 -176,9859 G G-8 214.7.11 Ross Sea -75,3296 -176,9859 G G-7 215.12.2 Ross Sea -75,3296 -176,9859 G G-7 11471 South Orkneys -60,5508 -45,176 278-222 G G-5 11475 South Orkneys -60,5508 -45,176 278-222 G G-4 11444 South Orkneys -60,5508 -45,176 278-222 G G-4 11740-2 South Shetlands -62,0918 -60,5322 286-364 G G-3 257-2.5.7 Eastern Antarctic Peninsula -64,9125 -60,65017 158 Polarstern 77 G G-2 260-6.3 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 G G-1 260-6.7 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 G G-2 POLA20080517 G G-2 133.3.1 Western Antarctic Peninsula -65,66405 -68,0371 H H-1 232.1.1 Ross Sea -76,3412 -170,8505 H H-4 232.2.1 Ross Sea -76,3412 -170,8505 H H-4 11740-1 South Shetlands -62,0918 -60,5322 286-364 H H-1 11740-11 South Shetlands -62,0918 -60,5322 286-364 H H-2 11740-12 South Shetlands -62,0918 -60,5322 286-364 H H-1 11740-7 South Shetlands -62,0918 -60,5322 286-364 H H-3 133.2.1 Western Antarctic Peninsula -65,66405 -68,0371 I I-7 11453 Bransfield Strait -62,8648 -57,2107 161-163 I I-7 11719-1 Elephant Island -61,187 -54,5883 278-356 I I-7 11719-10 Elephant Island -61,187 -54,5883 278-356 I I-7 11719-12 Elephant Island -61,187 -54,5883 278-356 I I-9 11719-3 Elephant Island -61,187 -54,5883 278-356 I I-12 11513 South Orkneys -60,5508 -45,176 278-222 I I-7 11352 South Shetlands -62,3558 -60,7305 164-166 I I-7 11362 South Shetlands -62,3558 -60,7305 164-166 I I-7 11472 South Shetlands -62,3558 -60,7305 164-166 I I-7 11740-13 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-14 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-15 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-16 South Shetlands -62,0918 -60,5322 286-364 I I-9 11740-17 South Shetlands -62,0918 -60,5322 286-364 I I-10 11740-18 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-20 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-21 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-22 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-3 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-4 South Shetlands -62,0918 -60,5322 286-364 I I-9 11740-5 South Shetlands -62,0918 -60,5322 286-364 I I-8 11740-6 South Shetlands -62,0918 -60,5322 286-364 I I-11 11740-8 South Shetlands -62,0918 -60,5322 286-364 I I-7 11740-9 South Shetlands -62,0918 -60,5322 286-364 I I-7 S5796 I I-7 S5798 I I-7 S5834 I I-11 260-6.1 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-6 260-6.2 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-6 260-6.4 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-3 260-6.5 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-3 260-6.6 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-5 260-6.8 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-3 260-6.9 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-5 260-6.10 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-3 260-6.11 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-5 260-6.14 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-5 260-6.15 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-3 260-6.16 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-1 260-6.17 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-5 260-6.20 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-4 260-6.22 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-6 260-6.23 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 I I-6 291-1.6 Eastern Weddell Sea -70,84167 -10,58733 268 Polarstern 77 I I-6 292-2.4 Eastern Weddell Sea -70,84483 -10,59183 252 Polarstern 77 I I-5 300-1.3 Eastern Weddell Sea -70,84133 -10,588 268 Polarstern 77 I I-5 308-1.7 Eastern Weddell Sea -70,855 -10,58917 224 Polarstern 77 I 0326-2 Polarstern 77 I I-5 YPM48435-1 South Orkneys -60,40005 -46,7562 280 Yale I I-7 YPM48435-26 South Orkneys -60,40005 -46,7562 280 Yale I I-7 YPM48435-33 South Orkneys -60,40005 -46,7562 280 Yale I I-7 YPM48435-36 South Orkneys -60,40005 -46,7562 280 Yale I I-7 YPM48435-51 South Orkneys -60,40005 -46,7562 280 Yale I I-7 YPM48435-53 South Orkneys -60,40005 -46,7562 280 Yale I I-7 YPM48435-65 South Orkneys -60,40005 -46,7562 280 Yale I I-14 AGT42/175-1 South Shetlands -62,31667 -58,7 496 Polarstern 42 I I-1 IU-2013-15811 Terre Adelie REVOLTA I I-13 CEA087 CEAMARC N1 N1-1 CEA088 CEAMARC N1 N1-1 CEA089 CEAMARC N1 N1-2 215.16.2 Ross Sea -75,3296 -176,9859 N2 N2-4 233.1.1 Ross Sea -76,3412 -170,8505 N2 N2-4 286-1.3.2 Eastern Weddell Sea -70,844 -10,60183 248 N2 N2-1 03-132-2 Eastern Weddell Sea -71,105003 -11,534 175 Polarstern 65 N2 N2-2 03-132-5 Eastern Weddell Sea -71,105003 -11,534 175 Polarstern 65 N2 N2-2 IU-2007-36c Terre Adelie/George V Land -66,002909 139,99768 190,8 CEAMARC N2 N2-5 IU-2007-152 Terre Adelie/George V Land -66,558067 142,646945 261,55 CEAMARC N2 N2-3 CEA04 Terre Adelie -66,159166 140,657159 212,15 CEAMARC N2 N2-3 257-2.5.1 Eastern Antarctic Peninsula -64,9125 -60,65017 158 N3 N3-1 257-2.5.3 Eastern Antarctic Peninsula -64,9125 -60,65017 158 N3 N3-1 AGT42/164-4 South Shetlands -62,13333 -57,66667 555 Polarstern 42 N3 N3-1 260.5.1 Ross Sea -76,2453 174,5041 O O-2 AGT42/175-2 South Shetlands -62,31667 -58,7 496 Polarstern 42 O O-1 AGT42/175-8 South Shetlands -62,31667 -58,7 496 Polarstern 42 O O-1 167,1 Amundsen Sea -72,4825 -104,56285 australis C14 Western Antarctic Peninsula australis 248-3.1 Eastern Antarctic Peninsula -65,92817 -60,3345 429 Polarstern 77 australis 260-6.2 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 australis 286-1.1.1 Eastern Weddell Sea -70,844 -10,60183 248 Polarstern 77 australis 292-2.2 Eastern Weddell Sea -70,84483 -10,59183 252 Polarstern 77 australis 300-1.1 Eastern Weddell Sea -70,84133 -10,588 268 Polarstern 77 australis 301-1.1.1 Eastern Weddell Sea -70,84983 -10,58717 227 Polarstern 77 australis 301-1.1.2 Eastern Weddell Sea -70,84983 -10,58717 227 Polarstern 77 australis 308-1.3 Eastern Weddell Sea -70,855 -10,58917 224 Polarstern 77 australis 308-1.4 Eastern Weddell Sea -70,855 -10,58917 224 Polarstern 77 australis 308-1.5 Eastern Weddell Sea -70,855 -10,58917 224 Polarstern 77 australis IU-2007-35f Terre Adelie/George V Land -67,052252 144,667145 197,75 CEAMARC australis IU-2013-15803 Terre Adelie/George V Land REVOLTA australis IU-2013-15804 Terre Adelie/George V Land REVOLTA australis IU-2013-15806 Terre Adelie/George V Land REVOLTA australis IU-2013-15807 Terre Adelie/George V Land REVOLTA australis PM_E002 Elephant Island -60,93333 -55,7 136 ICEFISH australis PM_E004 ICEFISH australis PM_E005 ICEFISH australis WOO0054 australis 301-1.2.1 Eastern Weddell Sea -70,84983 -10,58717 227 bouvetensis IU-2007-4794 Kerguelen -46,9833 70,45 425-445 C. sp. IU-2007-4852 Kerguelen -50,1833 69,8833 225-258 C. sp. CEA010 C. sp. 260-6.1 Eastern Weddell Sea -70,84267 -10,60333 252 Polarstern 77 cf. scotti 291-1.1.1 Eastern Weddell Sea -70,84167 -10,58733 268 Polarstern 77 cf. scotti 291-1.1.2 Eastern Weddell Sea -70,84167 -10,58733 268 Polarstern 77 cf. scotti 291-1.1.3 Eastern Weddell Sea -70,84167 -10,58733 268 Polarstern 77 cf. scotti 291-1.1.4 Eastern Weddell Sea -70,84167 -10,58733 268 Polarstern 77 cf. scotti 292-2.1 Eastern Weddell Sea -70,84483 -10,59183 252 Polarstern 77 cf. scotti POLA20080521 colossea 286-1.2 Eastern Weddell Sea -70,844 -10,60183 248 Polarstern 77 Decolopoda 292-2.3 Eastern Weddell Sea -70,84483 -10,59183 252 Polarstern 77 Decolopoda 301-1.3 Eastern Weddell Sea -70,84983 -10,58717 227 Polarstern 77 Decolopoda 308-1.1 Eastern Weddell Sea -70,855 -10,58917 224 Polarstern 77 Decolopoda IU-2007-5035 Kerguelen -48,5333 70,8833 126-132 Decolopoda 286-1.1.2 Eastern Weddell Sea -70,844 -10,60183 248 Polarstern 77 glacialis POLA20080516 near colossea 230.1.2 Ross Sea -76,3412 -170,8505 near HM381711 230,1 Ross Sea -76,3412 -170,8505 near HM381711 157,1 Bellingshausen Sea -70,842 -95,4112 near tortipalpis 158,1 Bellingshausen Sea -70,842 -95,4112 near tortipalpis 164,1 Amundsen Sea -71,75245 -102,2349 459 near tortipalpis 174,1 Amundsen Sea -72,7805 -104,5538 496 near tortipalpis 178.2.1 Amundsen Sea -73,7215 -103,6186 near tortipalpis 180.1.2 Amundsen Sea -72,2402 -103,5963 near tortipalpis 180,1 Amundsen Sea -72,2402 -103,5963 near tortipalpis 204.1.1 Marie Byrd Land -73,4985 -129,9188 near tortipalpis C12 Western Antarctic Peninsula near tortipalpis 265-2.1 Eastern Weddell Sea -70,789 -10,67317 Polarstern 77 near tortipalpis POLA20080523 near tortipalpis 126,1 Western Antarctic Peninsula -67,7268 -69,3017 robusta 231,1 Ross Sea -76,3412 -170,8505 robusta 102,31 Western Antarctic Peninsula -63,6691 -61,16745 scotti 118,11 Western Antarctic Peninsula -64,1387 -62,76005 scotti 11413 South Georgia -55,0387 -35,4347 124-125 scotti 11544 South Georgia -55,0515 -35,3953 119 scotti 11550 South Georgia -55,0515 -35,3953 119 scotti 11456 South Georgia -55,0515 -35,3953 119 scotti 11430 South Sandwich 226-344 scotti 11433 South Sandwich 235-83 scotti PB_E004 South Sandwich -57,0567 -26,7483 130 scotti PJ_E010 South Georgia -55,0661 -35,2417 117 scotti PL_E001 South Georgia -55,0661 -35,2417 117 scotti PL_E002 South Georgia -55,0661 -35,2417 117 scotti PL_E006 South Georgia -55,0661 -35,2417 117 scotti PL_E007 South Georgia -55,0661 -35,2417 117 scotti PL_E012 South Georgia -55,0661 -35,2417 117 scotti CEA023 CEAMARC tortipalpis PN_E003 South Shetlands -61,76667 -57,53333 343 tortipalpis PN_E005 South Shetlands -61,76667 -57,53333 343 tortipalpis ESMII: GenBank sequences included in this study number region lat long clade haplotypes AF259659 "Wales" A A-37 GQ328965 A A-2 GQ387007 A A-2 GQ387008 A A-10 GQ387009 A A-38 HM381694 Elephant Island -61,009 -55,851 A A-2 HM381696 Elephant Island -61,041 -55,767 A A-2 HM381697 Elephant Island -61,041 -55,767 A A-2 HM381698 Elephant Island -60,967 -55,918 A A-2 HM426357 South Sandwich -59,47 -27,278 A A-2 HQ970454 Elephant Island -61,13 -56,167 A A-2 HQ970456 Elephant Island -61,13 -56,167 A A-2 GQ387010 B B-3 GQ387011 B B-5 GQ387012 B B-6 GQ387013 B B-1 GQ387014 B B-2 GQ387015 B B-4 HM426343 Falklands -54,52 -56,149 B B-1 KF603930 Falklands -51,0967 -61,7333 B B-3 KF603931 Falklands -51,0967 -61,7333 B B-2 KF603933 Falklands -51,0967 -61,7333 B B-1 GQ387016 C C-2 GQ387017 C C-1 GQ387018 D1 D1-33 GQ387019 D1 D1-34 GQ387020 D1 D1-2 GQ387021 D1 D1-3 GQ387022 D1 D1-31 GQ387023 D1 D1-32 GQ387024 D1 D1-7 GQ387025 D1 D1-13 HM381695 Elephant Island -61,009 -55,851 D1 D1-15 FJ969357 Balleny Islands -66,2162 162,4415 D2 D2-5 HQ970336 D3 D3-4 GQ387026 E E-1 HQ970299 Terre Adelie -67,047 145,151 E E-2 FJ969356 Ross Sea -71,5307 171,3005 F F-5 GQ387027 F F-1 GQ387028 F F-3 HQ970455 Elephant Island -61,13 -56,167 G G-9 FJ969358 Balleny Islands -66,3655 162,5748 H H-4 HM426347 South Shetlands -62,525 -61,827 H H-5 HQ970298 Ross Sea -76,594 176,828 H H-6 HM426328 Eastern Weddell Sea -70,942 -10,53 I I-5 HM426330 South Shetlands -62,525 -61,827 I I-7 HM426348 South Shetlands -62,525 -61,827 I I-7 HM426359 Eastern Weddell Sea -71,119 -11,437 I I-5 HM432395 Eastern Weddell Sea -70,835 -10,579 I I-3 HM432407 Eastern Weddell Sea -70,946 -10,519 I I-3 HM432408 Eastern Weddell Sea -70,946 -10,519 I I-5 HM432410 Eastern Weddell Sea -70,946 -10,519 I I-5 HM432411 Eastern Weddell Sea -70,946 -10,519 I I-5 HM426390 Scott Seamounts -67,6243 -178,886 J J-1 HM381710 Terre Adelie -66,739 144,307 K K-1 HM426341 South Orkneys -61,096 -47,088 K K-2 HM432366 Eastern Weddell Sea -71,119 -11,437 L L-1 HM432371 Eastern Weddell Sea -71,119 -11,437 L L-1 KF603932 Chile -36,4002 -73,7179 M M-1 HQ970327 British Columbia angusta HQ970328 British Columbia angusta KC462557 North Atlantic angusta KC462558 North Atlantic angusta KC462559 North Atlantic angusta KC462560 North Atlantic angusta KC462561 North Atlantic angusta KC462562 North Atlantic angusta KC462563 North Atlantic angusta KC462564 North Atlantic angusta KC462565 North Atlantic angusta FJ969355 Ross Sea -71,7658 171,1672 australis GQ387002 australis GQ387003 australis HM381684 Elephant Island -60,881 -55,489 australis HM381701 Terre Adelie -66,389 140,429 australis HM426408 australis HM426436 Ross Sea -74,5817 170,25 australis HM432385 Eastern Weddell Sea -71,105 -11,534 australis HM432387 Eastern Weddell Sea -71,105 -11,534 australis HM432393 Eastern Weddell Sea -70,835 -10,579 australis HM432409 Eastern Weddell Sea -70,946 -10,519 australis HM432427 Ross Sea -74,5905 170,276 australis HM426194 Eastern Weddell Sea -71,081 -11,537 bouvetensis HM426263 South Sandwich -57,671 -26,466 bouvetensis HM426326 Eastern Weddell Sea -71,105 -11,534 bouvetensis HM426327 Eastern Weddell Sea -71,105 -11,534 bouvetensis HM426374 Ross Sea -76,1931 176,296 bouvetensis HM426375 Ross Sea -76,202 176,248 bouvetensis HM426381 Ross Sea -72,5903 175,342 bouvetensis HM426382 Ross Sea -72,5903 175,342 bouvetensis HM426383 Ross Sea -73,1245 174,32 bouvetensis HM426389 Ross Sea -76,594 176,828 bouvetensis HM426401 Ross Sea -73,2482 178,724 bouvetensis HM426402 Ross Sea -72,3395 175,532 bouvetensis HM426411 Ross Sea -75,6242 167,321 bouvetensis HM426433 Ross Sea -72,0235 173,18 bouvetensis HM426434 Ross Sea -74,1112 170,796 bouvetensis HM432368 Eastern Weddell Sea -71,119 -11,437 bouvetensis HM432388 Eastern Weddell Sea -71,105 -11,534 bouvetensis HM432391 South Sandwich -57,677 -26,424 bouvetensis HM432397 Eastern Weddell Sea -70,835 -10,579 bouvetensis FJ862877 C.sp. HM426312 Eastern Weddell Sea -67,521 -0,006 C.sp. HM432428 New Zealand -48,532 164,947 C.sp. HM432466 USA C.sp. HQ970329 British Columbia C.sp. HQ970330 British Columbia C.sp. HM381711 Terre Adelie -65,707 140,597 cf. scotti HM432396 Eastern Weddell Sea -70,835 -10,579 cf. scotti HQ970305 Terre Adelie -65,707 140,597 cf. scotti FJ716624 California 36,61 -122,43 colossea FJ716626 California 36,61 -122,43 colossea HM432376 South Sandwich -58,414 -25,014 colossea HM432377 South Sandwich -58,414 -25,014 colossea HM432380 South Sandwich -58,414 -25,014 colossea DQ390063 South Shetlands Decolopoda FJ969359 Ross Sea -71,3092 170,4503 Decolopoda FJ969360 Ross Sea -71,622 171,923 Decolopoda GQ386993 Decolopoda GQ386994 Decolopoda GQ386995 Decolopoda GQ386996 Decolopoda HM381683 Elephant Island -60,881 -55,489 Decolopoda HM381685 Elephant Island -61,011 -55,775 Decolopoda HM381686 Elephant Island -61,007 -55,886 Decolopoda HM381687 Elephant Island -61,333 -56,007 Decolopoda HM381688 Elephant Island -62,244 -55,309 Decolopoda HM426331 South Orkneys -60,735 -46,5 Decolopoda HM426358 South Shetlands -62,879 -60,999 Decolopoda HM426400 Ross Sea -74,5817 170,25 Decolopoda HM426427 Ross Sea -75,6217 169,805 Decolopoda HM431643 Western Antarctic Peninsula -68,186 -67,595 Decolopoda HM432400 Elephant Island -61,334 -55,195 Decolopoda HM432415 South Orkneys -61,001 -45,865 Decolopoda HQ970302 Terre Adelie -67,047 145,151 Decolopoda GQ386992 Dodecolopoda HM381693 Elephant Island -61,027 -55,94 Dodecolopoda HM381713 Terre Adelie -66,003 142,952 Dodecolopoda HQ970308 Terre Adelie -65,707 140,597 Dodecolopoda GQ386997 glacialis GQ386998 glacialis HM381674 Ross Sea -74,7 164,1 glacialis HM381691 Elephant Island -61,17 -56,006 glacialis HM381692 South Shetlands -62,491 -61,422 glacialis HM426185 South Georgia -53,611 -37,878 glacialis HM432370 Eastern Weddell Sea -71,119 -11,437 glacialis FJ716625 California 36,61 -122,43 japonica FJ716627 California 36,61 -122,43 japonica HM426309 Weddell Sea Basin -64,989 -43,035 japonica FJ862873 macerrima HM432426 macerrima JN018213 macerrima KF603928 Chile -45,90785 -75,60035 macerrima KF603929 Chile -45,90785 -75,60035 macerrima HM426361 Eastern Weddell Sea -69,408 -5,323 near colossea HM432461 Australia -35,5238 117,213 near colossea HM432394 Eastern Weddell Sea -70,835 -10,579 near tortipalpis HM432412 South Sandwich -58,267 -24,896 near tortipalpis FJ862872 Rhopalorhynchus GQ386999 robusta GQ387000 robusta HM381689 Elephant Island -61,011 -55,775 robusta HM381690 Elephant Island -61,016 -55,94 robusta HM426429 Ross Sea -74,5905 170,276 robusta HM432386 Eastern Weddell Sea -71,105 -11,534 robusta HM432414 South Orkneys -61,001 -45,865 robusta HM432416 South Orkneys -61,001 -45,865 robusta GQ387001 scoresbii KF603934 Falklands -51,28 -62,9633 scoresbii KF603935 Falklands -50,675 -62,435 scoresbii KF603936 Falklands -51,0967 -61,7333 scoresbii GQ387004 scotti GQ387005 scotti GQ387006 scotti HM381699 Elephant Island -60,986 -55,786 scotti HM432417 South Orkneys -61,001 -45,865 scotti DQ390078 stramenti DQ390061 tenera KC462566 North Atlantic tenera HM426377 Ross Sea -73,1245 174,32 tortipalpis HM426379 Ross Sea -72,5903 175,342 tortipalpis HM426380 Ross Sea -72,5903 175,342 tortipalpis HM426430 Ross Sea -73,1245 174,32 tortipalpis HM426431 Ross Sea -73,1245 174,32 tortipalpis HM426432 Ross Sea -73,1245 174,32 tortipalpis HQ970301 Terre Adelie -66,321 144,309 tortipalpis HQ970306 Terre Adelie -65,707 140,597 tortipalpis Dissertation Lars Dietz

5) Publikation IV

Titel: Assessing demographic responses to past climate change in Southern Ocean sea spiders

In Vorbereitung, Veröffentlichung bei Polar Biology geplant

Hinweise zu Publikation IV

• Anteil Planung: 85%

• Anteil experimentelle Durchführung: 90%

• Verfassen des Manuskripts: 90%

125 Dissertation Lars Dietz

Assessing demographic responses to past climate change in Southern Ocean sea spiders

Lars Dietz* & Florian Leese

Corresponding author; [email protected]

Ruhr University Bochum, Department of Animal Ecology, Evolution and Biodiversity, Universitaetsstrasse 150, D-44801 Bochum, Germany

Abstract: The phylogeography and population history of Southern Ocean benthic invertebrates has been studied for several species. However, the probability of different demographic scenarios has rarely been explicitly tested. Here, we use a previously published dataset for the widespread Southern Ocean pycnogonid species complex Colossendeis megalonyx. We test for different scenarios of population size change and geographic differentiation in the four largest species-level clades of the complex using both classic population genetic analysis and Bayesian methods. We show that large-scale geographic differentiation is strongly supported, suggesting the existence of several distinct refugia during the Pleistocene glaciations. The results confirm that one of the refugia was probably located on the shelf of South Georgia. There is little differentiation at smaller scales, suggesting a relatively recent expansion from the refugia. We find strong signals for population expansion, but there is no clear support for any specific demographic scenario. The dating of the expansion is consistent with it having occurred after the end of the Last Glacial Maximum.

Keywords: Population Genetics, Ice Age, Bottleneck, Expansion, Genetic Diversity

126 Dissertation Lars Dietz

Introduction The Southern Ocean fauna is strongly isolated from that of other oceans by the strong Antarctic Circumpolar Current (ACC) and the Polar Front (PF), which marks a large change in water temperature. Consequently, the faunal diversity in the Southern Ocean differs substantially from the neighbouring Oceans in that some taxon groups are much more diverse in the Southern Ocean (e.g. pycnogonids, isopods), whereas others are underrepresented (e.g. decapods). During the Pleistocene ice ages, the Antarctic shelf was covered completely or almost completely by grounded ice (Thatje et al. 2005), raising the question where the species-rich Southern Ocean benthic fauna survived. Three possible refuge areas are discussed: 1) Deep-sea refugia: the shelf was recolonized from the deep sea. The eurybathy of many taxa adds support for this scenario (Brey et al. 1996). There is genetic evidence for this e.g. in the shrimp Nematocarcinus lanceopes (Raupach et al. 2010, Dambach et al. 2013). 2) Subantarctic refugia: the Antarctic shelf could have been recolonized from the shelf regions of the Subantarctic islands and southern South America. The finding of high species diversity in particular at the tip of the Antarctic Peninsula (e.g. Griffiths et al. 2011) adds support to this scenario (ref.). 3) Antarctic shelf refugia: the fauna could have survived in temporary ice-free refugia on the Antarctic shelf. These could either have been permanently ice-free sites on the shelf (Thatje et al. 2005, 2008, Wilson et al. 2007) or temporarily ice-free sites e.g. due to time-transgressive expansions and contractions of the shelf (Convey et al. 2009). All three scenarios would have lead to different geographic and demographic signatures that can be detected with genetic markers (Allcock & Strugnell 2012). There is also evidence for each of the different scenarios for different organisms (see Thatje 2012 for a review). However, while in several studies haplotype networks have been described and, often loosely, assigned to a certain scenario based e.g. on mismatch-distribution tests (Wilson et al. 2009, Baird et al. 2011, Hemery et al. 2012), few studies have systematically compared different scenarios of demographic history against another (Raupach et al. 2010, González- Wevar 2013). In this study we used Antarctic sea spiders to test different scenarios. Sea spiders (Pycnogonida) are a significant component of the Antarctic benthos. They are strongly overrepresented in the Southern Ocean compared to other regions (Aronson et al. 2007). Pycnogonids have a holobenthic life cycle and most species are brooders with male parental care, however the larvae are unknown in the family Colossendeidae, to which the species studied here belongs. Several molecular studies on pycnogonids have been published showing strong regional differentiation and in some cases the possible presence of cryptic species (Mahon et al. 2008, Arango et al. 2011, Weis & Melzer 2012, Weis et al. 2014). Most of these studies used only mitochondrial markers, but it has also been shown

127 Dissertation Lars Dietz that nuclear markers such as the ribosomal Internal Transcribed Spacer (ITS) show results agreeing with the mitochondrial markers in species delimitation (Arango & Brenneis 2013, Dietz et al. in press). Colossendeis megalonyx Hoek, 1881 is one of the most common Southern Ocean sea spider species (Griffiths et al. 2011). It has a circumpolar eurybathic distribution, occurring on the Antarctic shelf as well as in South America and the Subantarctic islands, from 3-4900 m water depth (Munilla & Soler Membrives 2009). Krabbe et al. (2010) analyzed CO1 data from C. megalonyx and found six distinct clades with limited geographic distribution which they interpreted as cryptic species. However, Dietz et al. (in review) increased the number of clades to 15-19 and showed that many of them actually have a circumpolar distribution, often with large gaps. Internal Transcribed Spacer 1 (ITS) data showed some incongruence with CO1, indicating that clades that would be considered cryptic species based on mitochondrial data alone can freely hybridize. Therefore, while there seem to be barriers to hybridization between less closely related clades within C. megalonyx, the number of cryptic species is likely much lower than 15-19. In the present study we investigated the population history of C. megalonyx based on a large mitochondrial CO1 dataset and used different statistical approaches to measure population differentiation between different geographical locations and infer changes in population size for the four largest mitochondrial clades (A, D1, E and I, see Figure 1) to test various scenarios of postglacial recolonization. In addition to standard population genetic analyses, we also used several Bayesian approaches that allow to test for different scenarios and to estimate past population sizes. Bayesian geographic clustering assigns sequences to clusters using geographic location as prior information. It is possible to test for the probability of different numbers of clusters and the method also estimates the probability with which unsampled geographic regions belong to which cluster. In Approximate Bayesian Computations (ABC), different scenarios of population history are tested by performing a large number of simulations for each scenario, with parameters such as effective population size or times of population splits drawn at random from a prior distribution. The scenarios are then compared based on how many of the simulated datasets have population genetic summary statistics close to the original datasets, and a posterior distribution of the parameters is calculated based on the best-fitting simulated datasets. In Bayesian skyline plots (Drummond et al. 2005), it is assumed that effective population size changes at each node in a coalescent genealogy, with each population size during each interval depending on that in the previous interval. Using a Bayesian Markov Chain Monte Carlo (MCMC) approach, the posterior distribution of estimated population size is then taken for each point in time.

128 Dissertation Lars Dietz

Materials and Methods The COI data used in this study are from Dietz et al. (in review). For population genetic analyses of the four dominant clades (A, D1, E and I, see Figure 1), populations were defined as groups of samples from individual island groups in the Scotia Arc (South Georgia, South Sandwich, South Orkneys, South Shetlands) and Antarctic regions elsewhere (Eastern Weddell Sea, Terre Adélie/George V Land, Bouvet Island).

Figure 1: Statistical parsimony networks calculated for the C. megalonyx clades A, D1, E and I using a fragment of the mitochondrial COI. Figure adapted from Dietz et al. Royal Society Open Sciences (in review).

Population genetic parameters For the four largest clades, the program Arlequin (Excoffier & Lischer 2010) was used to calculate haplotype and nucleotide diversity (Nei 1987) for each population. FST values were calculated with Arlequin to measure differentiation between different populations within the same clade. We performed neutrality tests (Tajima’s D and Fu’s Fs; Tajima 1989, Fu & Li 1993), in which significantly negative values are a sign either of selection or of recent population expansion. Three-level Analyses of Molecular Variance (AMOVA; Excoffier et al. 1992) were performed with Arlequin to estimate the percentage of total variance contained within populations, between populations within larger groups, and between groups. For clades D1, E and I, we grouped Scotia Arc (also including Bouvet Island for clade E) samples on the one hand and East Antarctic samples on the other hand together as groups for AMOVA. For clade A, which is known only from Scotia Arc, we used two different schemes, one where South Georgia was classified as one group and all other Scotia Arc islands as

129 Dissertation Lars Dietz another, and one with the groups South Georgia/South Sandwich vs. South Orkneys/South Shetlands. The program Geneland 2.0 (Guillot et al. 2005) in the R environment was used to infer clusters of samples based on genetic data with geographical distance being taken into account as prior information. The program was run for 5,000,000 MCMC generations, with the first 50,000 generations discarded as burn-in and a number of clusters ranging from 1 to 5.

Approximate Bayesian Computation simulations The program DIYABC v. 2.0.4 (Cornuet et al. 2014) was used for Approximate Bayesian Computation (ABC) simulations to infer the population history of the four largest clades. The sequences were grouped into populations based on the island groups as above. For all four clades, the number of datasets used was one million times the number of scenarios, and the fit of scenarios was compared using both direct comparison and logistic regression. As little information about present and past effective population sizes in pycnogonids exist, the prior range was set from 10 to 106 using a logarithmic distribution prior. Divergence time was allowed to vary from 10 to 106 generations ago using a uniform distribution prior. A logarithmic distribution was not used for time estimations as we wanted to avoid bias towards smaller values. Mutation rate was modeled with a uniform distribution from 10-9 to 10-7 substitutions per site and year. The HKY model of evolution was used, with a gamma parameter of 2 and 10% invariant sites for simulations. For Clade A, the following scenarios were tested: 1) divergence of all populations at the same time without change of population size, 2) basal divergence of South Georgia population without change of population size, 3) divergence of all populations at the same time with change in population site in all populations after divergence, 4) basal divergence of South Georgia population with change in population site in all populations after divergence, 5) basal divergence of South Georgia population with change in population size in South Georgia and ancestral southern populations before divergence of the latter, 6) as in 5), only without population change in South Georgia, and 7) as in 5) with population change only in S. Georgia. For Clade D1, the following scenarios were tested: 1) divergence of all populations at the same time without change of population size, 2) basal divergence of Terre Adélie population without change of population size, 3) divergence of all populations at the same time with change in population site in all populations after divergence, 4) basal divergence of Terre Adélie population with change in population site in all populations after divergence, 5) basal divergence of Terre Adélie population with change in population size in Terre Adélie and ancestral West Antarctic populations before divergence of the latter, 6) as in 5), only without

130 Dissertation Lars Dietz population change in Terre Adélie, and 7) as in 5) with population change only in Terre Adélie. For Clade E, the following scenarios were tested: 1) divergence of all populations at the same time without change of population size, 2) basal divergence of Terre Adélie population without change of population size, 3) divergence of all populations at the same time with change in population site in all populations after divergence, 4) basal divergence of Terre Adélie population with change in population site in all populations after divergence, 5) basal divergence of Terre Adélie population with change in population size in Terre Adélie and ancestral West Antarctic populations before divergence of the latter, 6) as in 5), only without population change in Terre Adélie, 7) as in 5) with population change only in Terre Adélie, 8) simultaneous divergence of all populations with change of population size only in Bouvet Island, and 9) basal divergence of Terre Adélie population with change of population size only in Bouvet Island. For Clade I, the following scenarios were tested: 1) divergence of all populations at the same time without change of population size, 2) basal divergence of Eastern Weddell Sea population without change of population size, 3) divergence of all populations at the same time with change in population site in all populations after divergence, 4) basal divergence of Eastern Weddell Sea population with change in population site in all populations after divergence, 5) basal divergence of Eastern Weddell Sea population with change in population size in Eastern Weddell Sea and ancestral Scotia Arc populations before divergence of the latter, 6) as in 5), only without population change in Eastern Weddell Sea, and 7) as in 5) with population change only in Eastern Weddell Sea.

131 Dissertation Lars Dietz

Figure 2: Overview of the seven different scenarios tested systematically with DIYABC for the four clades of Colossendeis megalonyx. Different colours represent changes in the parameter (Ne or tn) inferred. For clade E, the additional scenarios 8 and 9 were tested, corresponding to respectively to scenarios 1 and 2 but with population size change assumed for Bouvet Island only.

Bayesian Skyline plots The program BEAST 1.8.0 (Drummond et al. 2012) was used to for Bayesian skyline analyses of the four clades. For clade A, separate analyses were performed for the complete dataset, the South Georgia samples only, and the more southern samples only. For all other clades, separate analyses were performed for the complete dataset and for both the western and the eastern samples only. The analysis was run for 100,000,000 MCMC generations, with the first 50% (clade A, complete dataset) or 10% (all other datasets) discarded as burn- in after checking for convergence of the likelihood values. A HKY model of evolution with a transition/transversion ratio of 2 was used. The program Tracer was used to visualize the data.

132 Dissertation Lars Dietz

Results

Clade A

Significant positive FST values were found for population comparisons of South Georgia against both Elephant Island and South Sandwich populations (Tab. 3). Elephant Island and South Sandwich showed significant negative values of both Tajima’s D and Fu’s Fs (Tab. 1). The Geneland analysis recovered one single cluster to which all samples were assigned with a probability of 1.

Clade D1

Significant positive FST values were found for comparisons of East Antarctic against all three West Antarctic populations and for Elephant Island vs. South Orkneys populations (Tab. 4). The South Orkneys and South Shetland populations showed significant negative Tajima’s D, and Elephant Island and for South Shetlands showed significant negative Fu’s Fs values (Tab. 1). The Geneland analysis found two clusters, one for the Scotia Arc/Antarctic Peninsula and one for the East Antarctic. All individuals were assigned to their clusters with a probability of 1 except for one from the Western Antarctic Peninsula (p=0.99).

Clade E

FST values between all three locations were significantly positive (Tab. 5). The Terre Adélie population showed significant negative Tajima’s D and Fu’s Fs values, and in the South Sandwich population only Fu’s Fs was significantly negative (Tab. 1). The Geneland analyses recovered two clusters, one for the South Sandwich Islands and Bouvet and one for Terre Adélie and the single specimen from the Eastern Antarctic Peninsula. All samples were assigned to their clusters with a probability of 1.

Clade I

FST is significant between the Eastern Weddell Sea and both South Orkneys and South Shetlands populations (Tab. 6). Geneland recovered two clusters for the Eastern Weddell Sea and South Orkneys/South Shetlands, respectively. All samples were assigned to their clusters with a probability of 1.

133 Dissertation Lars Dietz

Table 1: Population genetic parameters calculated with Arlequin for each C. megalonyx population included in this study. *: p<0.05, **: p<0.01, ***: p<0.001. No. of No. of Haplotyp Nucleotid Tajima’s Populations sequence haplotype e e Fu’s Fs D s s diversity diversity Clade A S. Georgia 13 7 0.8718 0.004328 -0.329 -1.794 S. Sandwich 20 4 0.2842 0.00055 -1.723* -2.749** S. Orkneys 4 1 0 0 0 - Elephant Is. 100 23 0.5347 0.001405 -2.476*** - 3.4*1038*** S. Shetlands 1 1 0 0 0 - Clade D1 S. Orkneys 45 14 0.6111 0.030648 -2.23*** 6.225 Elephant Is. 24 11 0.9058 0.005172 -0.66 -3.368* S. Shetlands 9 8 0.9722 0.004077 -1.843** -5.485*** Terre Adélie 6 4 0.8 0.089174 -0.273 6.506 Clade E S. Sandwich 19 6 0.655 0.001738 -1.05 -2.519* Bouvet Is. 43 1 0 0 0 - Terre Adélie 22 6 0.619 0.001509 -1.55* -2.716* Clade I S. Orkneys 7 1 0 0 0 - Elephant Is. 5 3 0.7 0.049174 -1.146 6.287 S. Shetlands 21 6 0.5571 0.023556 -0.633 8.487 E. Weddell 26 9 0.8769 0.054493 -0.17 15.169 Sea

Table 2: Results of the AMOVA for the different clades of Colossendeis megalonyx. Percentages of variance found i) between groups, ii) within groups between populations, and iii) within populations is shown. For Clade A two different partitioning schemes for defining the groups were used: Scheme 1: South Sandwich is grouped with the more southern populations, scheme 2: South Georgia and South Sandwich are grouped together. Percentage Clade A Clade A Clade D1 Clade E Clade I of variance (scheme 1) (scheme 2) Between 46.5 -19.59 84.01 94.29 30.63 groups Within -2.94 38.71 0.62 0.72 -5.07 groups between populations Within 56.44 80.88 15.37 4.99 74.44 populations

134 Dissertation Lars Dietz

Table 3: FST values between all populations of Colossendeis megalonyx Clade A calculated by Arlequin. *: p<0.05, **: p<0.01, ***: p<0.001. Clade A S. Georgia S. Sandwich S. Orkneys Elephant Is. S. n=13 n=20 n=4 n=100 Shetlands n=1 S. Georgia - S. Sandwich 0.36048*** - S. Orkneys 0.16466 -0.13772 - Elephant Is. 0.40578*** -0.00338 -0.13037 - S. -0.33333 -1 0 -0.96374 - Shetlands

Table 4: FST values between all populations of Colossendeis megalonyx Clade D1 calculated by Arlequin. *: p<0.05, **: p<0.01, ***: p<0.001. Clade D1 S. Orkneys Elephant Is. S. Shetlands Terre Adélie n=45 n=24 n=9 n=6 S. Orkneys - Elephant Is. 0.07173* - S. Shetlands 0.0171 0.00509 - Terre Adélie 0.77197*** 0.87874*** 0.79737*** -

Table 5: FST values between all populations of Colossendeis megalonyx Clade E by Arlequin. *: p<0.05, **: p<0.01, ***: p<0.001. Clade E S. Sandwich Bouvet Is. Terre Adélie n=19 n=43 n=22 S. Sandwich - Bouvet Is. 0.19026*** - Terre Adélie 0.90195*** 0.96707*** -

Table 6: FST values between all populations of Colossendeis megalonyx Clade I of the four clades calculated by Arlequin. *: p<0.05, **: p<0.01, ***: p<0.001. Clade I S. Orkneys Elephant Is. S. Shetlands E. Weddell Sea n=7 n=5 n=21 n=26 S. Orkneys - Elephant Is. 0.07285 - S. Shetlands -0.02537 -0.08345 - E. Weddell Sea 0.29956* 0.04324 0.23064** -

135 Dissertation Lars Dietz

For no population, the results of the mismatch distribution analyses showed any significant results, which supports a scenario of sudden demographic expansion of all populations. In all four clades, AMOVA (Tab. 2) showed that less than 1% of variation was due to differentiation among populations (island groups in Scotia Arc) within groups (East vs. West, or Scotia Arc vs. more southern islands in clade A), with differing amounts due to intra- population or inter-group variance. The proportion of intra-group variance ranged from 30.63% (clade I) to 94.29% (clade E). Intra-group inter-population differentiation was significant only for clade E. Using an alternative scheme for clade A in which South Sandwich was grouped with South Georgia instead of South Orkneys/South Shetlands, no differentiation between groups was found (-19.59% of variance), while significant differentiation was found between populations (38.71%).

Table 7: parameters for clade A estimated by ABC. Numbers in first row refer to tested scenarios (see text). Numbers refer to effective population size of each population (“old” means population size before expansion), and in the last three rows to time before present in generations. Clade A 2 3 4 5 6 S. Georgia 98400 298000 318000 224000 209000 S. Sandwich 25500 137000 125000 67300 70900 S. Orkneys 355 467 578 557 448 Elephant Is. 185000 635000 636000 506000 429000 S. Georgia 2250 2430 2800 old Southern 608 548 islands old S. Sandwich 1090 831 old S. Orkneys 2690 2870 old Elephant Is. 394 262 old Time of 8040 8150 56700 83900 expansion Time of split 350000 502000 245000 26500 38400 Time of 685000 696000 528000 552000 SG/southern split

136 Dissertation Lars Dietz

Table 8: parameters for clade D1 estimated by ABC. Numbers in first row refer to tested scenarios (see text). Numbers refer to effective population size of each population (“old” means population size before expansion), and in the last three rows to time before present in generations. Clade D1 3 4 5 6 S. Orkneys 234000 344000 172000 159000 Elephant Is. 178000 97200 178000 132000 S. Shetlands 709000 766000 532000 551000 Terre Adélie 49300 39500 71500 38900 Scotia Arc old 1660 711 S. Orkneys old 366 230 Elephant Is. old 229 1330 S. Shetlands 218 938 old Terre Adélie 1500 687 2490 old Time of 8120 6950 31700 27900 expansion Time of split 607000 298000 2210 1860 Time of 546000 417000 379000 West/East split

Table 9: parameters for clade E estimated by ABC. Numbers in first row refer to tested scenarios (see text). Numbers refer to effective population size of each population (“old” means population size before expansion), and in the last three rows to time before present in generations. Clade E 1 2 3 4 5 6 7 8 9 S. 61100 57200 125000 143000 78700 68200 62800 62300 59100 Sandwich Bouvet Is. 311 277 286 307 323 289 324 297 250 Terre 58400 58600 144000 155000 77600 63600 66000 58500 56100 Adélie Western 2450 2700 pops. old S. 1860 1300 Sandwich old Bouvet Is. 2880 3110 2360 2720 3530 old Terre 1350 1320 2640 Adélie old Time of 82400 51700 416000 478000 477000 305000 208000 expansion Time of 550000 286000 606000 370000 173000 200000 195000 715000 510000 split Time of 707000 745000 763000 795000 796000 798000 West/East split

137 Dissertation Lars Dietz

Table 10: parameters for clade I estimated by ABC. Numbers in first row refer to tested scenarios (see text). Numbers refer to effective population size of each population (“old” means population size before expansion), and in the last three rows to time before present in generations. 1 2 3 4 5 6 7 S. 4710 6130 13100 15500 8300 10200 17900 Orkneys Elephant 41700 58400 339000 35300 80000 81800 68200 Is. S. 144000 35000 35600 54000 47800 97600 124000 Shetlands E. 124000 121000 209000 393000 249000 136000 189000 Weddell Sea Scotia Arc 3630 2000 old S. 25700 333 Orkneys old Elephant 1950 644 Is. old S. 20300 10200 Shetlands old E. 10 1860 1470 1240 Weddell Sea old Time of 74900 15400 181000 361000 332000 expansion Time of 38400 4870 317000 200000 2810 3520 13000 split Time of 471000 871000 717000 784000 820000 West/East split

For clade A (Tab. 7), the DIYABC results show that the only scenarios with significant support in the logistic approach (posterior probability > 0.02) were 3 and 4, which assume independent population changes in all five populations. In the direct comparison these scenarios also have by far the highest support, but scenarios 2, 5 and 6 also have posterior probability higher than 0.05. Both in the direct and logistic approaches scenario 4, with an earlier split of the South Georgia population from the rest, was preferred to scenario 3 (pp=0.61 vs. 0.34 in the logistic approach with n=70000). Estimated population size was consistently highest for Elephant Island and very low for the South Orkneys, for which only three sequences were available. In scenario 2, which does not assume expansion, the estimated current population sizes are lower than in scenarios that assume expansion. In scenarios 3-4, those populations with a large sample size (South Georgia, South Sandwich, Elephant Island) showed signals for strong population growth, which was stronger for Elephant Island (expansion by a factor of 1612-2427) than for the more northern locations (factor 126-150). The median estimated initial population sizes for South Georgia (2250- 2430) are higher than those for South Sandwich (831-1090) or Elephant Island (262-394). 138 Dissertation Lars Dietz

Scenario 5 also shows a higher initial population size for South Georgia than for the ancestral southern population. The time of population expansion was calculated with as 8040 (scenario 3) or 8150 (scenario 4) generations ago. The time of separation of southern populations in scenario 4 has a median of 245000 generations ago. For clade D1 (Tab. 8), both direct and logistic comparisons show that scenarios 1 and 2, assuming no change in population size, as well as scenario 7, assuming population size change only in Terre Adélie, are very unlikely. According to the direct comparison, scenarios 4 (pp=0.4) and 3 (pp=0.3), which assume population expansion after the split between the West Antarctic populations, are more likely than scenarios 5 and 6 (pp=0.09-0.10), which assume expansion before the split. In the logistic approach, the scenarios 5 (pp=0.27) and 6 (pp=0.29) are rated as more likely than 3 (pp=0.14) and 4 (pp=0.24). The estimated present effective population size is highest for the South Shetlands (532,000-719,000) and lowest for Terre Adélie (38,900-71,500). All scenarios support a strong population expansion, which is however stronger for the western populations (factor 73-3252) than for Terre Adélie (factor 29-57). When the split is assumed to have preceded the expansion, the latter is calculated as 6950-8120 generations ago, while if it is assumed to have occurred before the split it is calculated as 27900-31700 generations ago. This assumption also has a strong effect on the dating of the split between the western populations, which is calculated as 1860-2210 generations ago in scenarios 5-6, but 298000 generations ago in scenario 4. The East-West split is calculated as 379000-607000 generations ago. For clade E (Tab. 9), no scenario can be strongly rejected. Except for scenarios 3 and 4, which presume independent expansion in all populations, all have posterior probabilities of about 0.08-0.1 in both the direct and logistic approach. Scenarios 3 and 4 have higher probabilities, and in the direct approach scenario 3 is more probable than 4 when only the closest datasets to the original are taken into account. The more datasets are included in the calculation, the more probable does scenario 4 become. Scenario 4 is also preferred in the logistic approach. The estimated effective population size I similar for the South Sandwich (57,200-143,000) and Terre Adélie (56,100-155,000). For Bouvet, where only one haplotype occurs, the estimated population size is very low (250-324). The estimated original population size for South Sandwich is 1300-1860 and for Terre Adélie 1320-2640. Remarkably, the estimated original population for Bouvet is larger (2720-3530) than the current size. The time of population size change is estimated as 51,700-82,400 generations ago for scenarios where it occurred after the population split, and 416,000-478,000 generations ago for scenarios where it occurred before the split. If a South Sandwich/Bouvet split is assumed, it is estimated as 173,000-510,000 generations ago. For clade I (Tab. 10), no scenario can be strongly rejected. Notably, the results for the direct and logistic approach are very different: scenario 4, which is rated as most likely in the direct

139 Dissertation Lars Dietz approach, is rated as least likely in the logistic approach. In the logistic approach, scenarios with a population expansion before the split between the Scotia Arc populations (scenarios 5- 7) are strongly preferred, while this is not the case in the direct comparison. Estimated population sizes are lower for the South Orkneys (470-17,900) than for other populations, and in most scenarios they are highest for the Eastern Weddell Sea. There is no signal for a strong expansion of the South Shetland population, while the Elephant Island one expands in scenario 4 but not in scenario 3. The Eastern Weddell Sea population expands from an initial size of 10-11,860 to 18,900-39,300 when expansion is assumed. If the split between Scotia Arc populations is assumed to have occurred after or without expansion, it is dated as a relatively recent time of 2810-13,000 generations ago while the expansion is dated as 181000-332000 generations ago. Bayesian skyline analysis for clade A shows an increase from theta=1.28*10-2 to 5.9*10-2 for the complete dataset, with a root age of 5.61*10-4. The South Georgia population showed a slight size increase from 2.9*10-3 to 5.11*10-3, with a root age of 9.07*10-4. The southern population showed a slight increase from 3.08*10-2 to 5.92*10-2, with a root age of 3.8*10-4. Skyline analysis for clade D1 showed a root age of 2.93*10-3. The initial theta is 3.82*10-3, with a strong expansion at around 1-5*10-4 to a current theta value of 5.95*10-2. Taking only the West Antarctic samples into account, there is also a strong signal for expansion from an initial theta of 1.93*10-3 (root age: 1.19*10-3) to a current value of 5.94*10-2. The expansion is strongest at about 3-8*10-4. Clade E also shows a strong expansion from an initial value of 1.96*10-3 (root age: 2.85*10-3) to a current one of 9.75*10-3. The expansion started at about 2.5*10-4. However, no such strong increase can be detected when the analysis is limited to the western (South Sandwich, Bouvet) samples. There is only a slight expansion from 2.89*10-3 to 4.45*10-3 (root age: 6.91*10-5). Similarly, when only the East Antarctic samples are included, the theta value stays virtually the same, from 3.73*10-3 in the beginning (root age: 9.97*10-5) to 4.13*10-3 at present. Like clades D1 and E, clade I also shows a strong increase from 2.56*10-3 to 1.26*10-2 (root age: 2.72*10-3). The increase occurs mainly around 1-3*10-4. This increase is also evident when only the western samples are included (from 2.09*10-3 to 7.18*10-3, root age 2.07*10-3), while the Eastern Weddell Sea samples show only a slight increase from 1.47*10-3 to 2.7*10- 3 (root age 2.63*10-4).

140 Dissertation Lars Dietz

Figure 3: Results of Geneland analyses for clades D1, E and I. Black dots indicate sampling locations. Darker colors indicate greater probabilities of belonging to cluster 1. Note that, as only two clusters were inferred, the probability of belonging to cluster 2 is one minus the probability of belonging to cluster 1. Coordinates are not accurate geographical coordinates.

141 Dissertation Lars Dietz

Figure 4: Result of Bayesian Skyline analyses for clades A and D1. x axis indicates time before present and y axis indicates population size, both as theta values. Note that the x axes are not to the same scale.

142 Dissertation Lars Dietz

Figure 5: Result of Bayesian Skyline analyses for clades E and I. x axis indicates time before present and y axis indicates population size, both as theta values. Note that the x axes are not to the same scale.

Discussion Population genetic investigations of Antarctic marine invertebrates often revealed signals for a strong population expansion in the form of a “star-like” haplotype network, in which most specimens belong to a single haplotype while other haplotypes are present in only a few specimens and differ from the main haplotype in only few mutations. Taxa that show such patterns in their haplotype networks include the nemertean Parbolasia corrugatus (Thornhill et al. 2008), the shrimp Chorismus antarcticus (Raupach et al. 2010), the echinoid Sterchinus neumayeri (Díaz et al. 2011), and the ostracod Macroscapha turbida (Brandao et

143 Dissertation Lars Dietz al. 2010). The gastropod Nacella concinna also fits this pattern, except for the South Georgia population (González-Wevar et al. 2013). According to Allcock & Strugnell (2012), such a pattern is typical for taxa that survived the Pleistocene glaciations in a single refugium on the continental shelf, as survival in the deep sea would produce a more diffuse pattern as seen e.g. in the shrimp Nematocarcinus lanceopes (Raupach et al. 2010), due to the much larger habitat available. As noted by Dietz et al. (in review), some Colossendeis megalonyx clades, especially clade A, show a similar pattern. However, for most clades, the networks are actually composed of several different star-like patterns separated by about 5-10 steps and mostly restricted to different geographic regions. This is not easily classified according to the different categories of Allcock & Strugnell (2012), although it resembles the pattern “c” described therein (i.e., diffuse network), in which the different networks are however not connected. This evidence hints at the survival of populations in several glacial refugia for each larger clade, except for clade A, which is restricted to the Scotia Arc. The geographical restriction of these subgroups, which is also confirmed by our results of strong geographical differentiation as shown by FST values and Geneland results, suggests limited dispersal capabilities of C. megalonyx. In contrast, invertebrates with a planktonic stage such as the crinoid Promachocrinus kerguelensis (Hemery et al. 2012) show much less geographic differentiation. This is in accord with its biology as a benthic organism without a known planktonic dispersal stage. Colossendeid larvae are unknown, but these results suggest that there is no planktonic stage, and one would expect that planktonic larvae would have been discovered if they existed as colossendeids are a common and widespread group in the Southern Ocean.

The results of the both FST and AMOVA calculations show strong evidence for differentiation between the western (Scotia Arc) islands and eastern (Eastern Weddell Sea or Terre Adélie) populations in clades D1, E and I. These results support the view that western and eastern populations were isolated from each other for many generations. As also argued in Dietz et al. (in review), the results are best explained by survival in different refugia during the Last Glacial Maximum (LGM) with only very limited migration between West and East Antarctica. This supports the idea that C. megalonyx survived the LGM in temporary ice-free refugia in the Antarctic shelf region, as a deep sea refugium would probably have led to greater genetic similarity across the Southern Ocean, as seen e.g. in the shrimp Nematocarcinus lanceopes (Raupach et al. 2010). There is only very little genetic differentiation between the different Scotia Arc island groups, meaning that either there is frequent migration between these populations or the region was colonized relatively recently. As the dispersal capabilities of pycnogonids are limited due to their benthic habit and putative lack of a planktonic stage, the hypothesis of frequent migrations seems to be unlikely. A recent colonization from a single source would mean that the Scotia Arc was presumably colonized after the LGM from a

144 Dissertation Lars Dietz single refugium. For clade A, that refugium was probably South Georgia, which has a higher haplotype diversity than the more southern islands and a less strong signature of population expansion. For the other clades, the refugia have yet to be identified, and none of the island groups sampled so far are likely to have been the refugium as they all show strong signatures of population expansion. It is possible that the refugia for these clades were located on the Antarctic continental shelf and the Scotia Arc was colonized only later.

The Geneland results agree with the FST and AMOVA calculations in the presence of an East/West split in clades D1, E and I, but do not show any clear split between South Georgia and the other islands in clade A. This contrasts with the results of González-Wevar et al. (2013) for the gastropod Nacella concinna, which shows a very similar pattern of higher genetic diversity in South Georgia compared to the more southern islands, but for which Geneland analysis separated South Georgia from the other locations. A possible reason for this may be the relatively low sampling for South Georgia compared to the other island groups in our study. The results of the Bayesian skyline analyses in general confirm the hypothesis of strong population expansion as also suggested by the ABC results and significant D and Fs values. Similar results are found in the octopus Pareledone turqueti (Strugnell et al. 2012). For clade A, the South Georgia population is suggested to be older than the more southern populations and also has a higher effective population size, confirming the possible status of South Georgia as a refugium, as also seen in Nacella concinna (González-Wevar et al. 2013). In all cases for which a clear signature for population expansion exists, it happens at around 1- 8*10-4. The root age of the southern population in clade A also falls into this range. If a mutation rate of 10-8 per year (i.e. 1% per million years, as often assumed for CO1) is assumed, this corresponds to a time of 10000-80000 years ago, which is consistent with a population expansion after the LGM. The results are also similar to those of the ABC analysis, where the mutation rate was allowed to vary between 10-7 and 10-9. While the generation time of C. megalonyx is unknown as nothing is known about the reproductive biology of colossendeids, in other pycnogonids breeding is seasonal and larval development can take several months (reviewed by Arnaud & Bamber 1987), which suggests a generation time of at least a year. The ABC calculations often do not definitely favor a specific scenario, but scenarios that assume population growth were always preferred. Also, scenarios that assumed an East- West split were preferred to scenarios where all populations split at the same time. These results agree with those from the FST and AMOVA calculations. Scenarios with expansion after the split into the Scotia Arc subpopulations were generally rated as more likely at least in the direct comparison, suggesting that these habitats were initially colonized by very small populations. For clade A, a larger initial population size is inferred for South Georgia than for

145 Dissertation Lars Dietz the more southern islands, confirming the hypothesis that this location may have acted as a refugium. It should be noted that the uncertainty for the inferred population sizes is often rather high, especially for subpopulations whose sample size is very low (e. g. just five clade A sequences from the South Orkneys), so the inferences in those cases should be treated with caution. Nevertheless the signal for population expansion is clearly present in most populations. Nuclear sequence data from the ribosomal gene region ITS show some disagreements with the CO1 data (Dietz et al. in review) which suggest that the mitochondrial clades discussed here may not be distinct species. Of the clades studied here, A and I as well as D1 and E seem to be interfertile and therefore cannot be considered distinct populations in regions where they are sympatric (the southern Scotia Arc for A and I, Terre Adélie for D1 and E). Nevertheless, the evidence for population growth can be considered valid as it is found in all four clades.

Outlook The present analysis is limited by being restricted to sequences from a single gene. Although it has been possible to infer some conclusions on the population history of C. megalonyx, the statistical power of the methods used here will be greatly improved if more loci are used. Due to recent technological advances, it has become possible to sequence thousands of loci even for non-model species with techniques like restriction site associated DNA sequencing (RADseq; Miller et al. 2007). These techniques have already been shown to be useful for population genetic studies (e.g. Macher et al. 2015).

Acknowledgments We thank Ralph Tollrian for support of this work. We thank Claudia Arango (Queensland Musum/Brisbane), Ken Halanych (Auburn University), Katrin Linse (British Antarctic Survey/Cambridge), Andrew Mahon (Central University of Michigan), Roland Melzer (Zoologische Staatssammlung/Munich) and Nerida Wilson and Greg Rouse (Scripps Institution of Oceanography/San Diego), Samples were provided Furthermore, we thank all participants on the NBP1105 RV/IB Nathaniel B. Palmer and ANT-XXVIII/4 FS Polarstern cruises. This work was supported by DFG grant LE 2323/2 to FL

146 Dissertation Lars Dietz

References Allcock AL, Strugnell JM (2012): Southern Ocean diversity: new paradigms from molecular ecology. Trends in Ecology and Evolution 27: 520-528. Arango CP, Brenneis G (2013): New species of Australian Pseudopallene (Pycnogonida: Callipallenidae) based on live colouration, morphology and DNA. Zootaxa 3616: 401- 436. Arango CP, Soler-Membrives A, Miller KJ (2011): Genetic differentiation in the circum— Antarctic sea spider Nymphon australe (Pycnogonida; Nymphonidae). Deep Sea Research Part II: Topical Studies in Oceanography 58: 212-219. Arnaud F, Bamber R (1987): The biology of Pycnogonida. Advances in Marine Biology 24: 1- 96. Aronson RB, Thatje S, Clarke A, Peck LS, Blake DB, Wilga CD, Seibel BA (2007): Climate change and invasibility of the Antarctic benthos. Annual Review of Ecology Evolution and Systematics 38: 129-154. Baird HP, Miller KJ, Stark JS (2011): Evidence of hidden biodiversity, ongoing speciation and diverse patterns of genetic structure in giant Antarctic amphipods. Molecular Ecology 20: 3439-3454. Brandao SN, Sauer J, Schön I (2010): Circumantarctic distribution in Southern Ocean benthos? A genetic test using the genus Macroscapha (Crustacea, Ostracoda) as a model. Molecular Phylogenetics and Evolution 55: 1055–1069. Brey T, Dahm C, Gorny M, Klages M, Stiller M, Arntz WE (1996): Do Antarctic benthic invertebrates show an extended level of eurybathy? Antarctic Science 8: 3-6. Convey P, Stevens MI, Hodgson DA, Smellie JL, Hillenbrand CD, Barnes DKA, Clarke A, Pugh PJA, Linse K, Cary SC (2009): Exploring biological constraints on the glacial history of Antarctica. Quaternary Science Reviews 28: 3035-3048. Cornuet J, Pudlo P, Veyssier J, Dehne-Garcia A, Gautier M, Leblois R, Marin J, Estoup A (2014): DIYABC v2.0: a software to make approximate Bayesian computation inferences about population history using single nucleotide polymorphism, DNA sequence and microsatellite data. Bioinformatics: advance online publication. Dambach J, Raupach MJ, Mayer C, Schwarzer J, Leese F (2013): Isolation and characterization of nine polymorphic microsatellite markers for the deep-sea shrimp Nematocarcinus lanceopes (Crustacea: Decapoda: Caridea). BMC Research Notes 6: 75. Díaz A, Féral JP, David B, Saucède T, Poulin E (2011): Evolutionary pathways among shallow and deep-sea echinoids of the genus Sterechinus in the Southern Ocean. Deep Sea Research Part II: Topical Studies in Oceanography 58: 205-211.

147 Dissertation Lars Dietz

Dietz L, Arango CP, Halanych K, Harder AM, Held C, Mahon AR, Mayer C, Melzer RR, Rouse G, Weis A, Wilson N, Leese F (in review): Extensive sampling reveals regional differentiation and mitochondrial-nuclear discordances in the Southern Ocean giant sea spider Colossendeis megalonyx Hoek, 1881. Royal Society Open Science. Dietz L, Pieper S, Seefeldt MA, Leese F (in press): Morphological and genetic data clarify the taxonomic status of Colossendeis robusta and C. glacialis (Pycnogonida) and reveal overlooked diversity. Arthropod Systematics & Phylogeny. Drummond AJ, Rambaut A, Shapiro B, Pybus OG (2005): Bayesian coalescent inference of past population dynamics from molecular sequences. Molecular Biology and Evolution 22: 1185-1192. Drummond AJ, Suchard MA, Xie D, Rambaut A (2012): Bayesian phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution 29: 1969-73 Excoffier L, Lischer HEL (2010): Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources 10: 564-567. Excoffier L, Smouse PE, Quattro JM (1992): Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491. Fu YX, Li WH (1993): Statistical test of neutrality of mutations. Genetics (1993) 133: 693- 709. González-Wevar CA, Saucède T, Morley SA, Chown SL, Poulin E (2013): Extinction and recolonization of maritime Antarctica in the limpet Nacella concinna (Strebel, 1908) during the last glacial cycle: toward a model of Quaternary biogeography in shallow Antarctic invertebrates. Molecular Ecology 22: 5221-5236. Guillot G, Mortier F, Estoup A (2005): Geneland: a computer package for landscape genetics. Molecular Ecology Notes 5: 712-715. Griffiths HJ, Arango CJ, Munilla T, McInnes SJ (2011): Biodiversity and biogeography of Southern Ocean pycnogonids. Ecography 34: 616-627. Hemery LG, Eléaume M, Roussel V, Améziane N, Gallut C, Steinke D, Cruaud C, Couloux A, Wilson NG (2012): Comprehensive sampling reveals circumpolarity and sympatry in seven mitochondrial lineages of the Southern Ocean crinoid species Promachocrinus kerguelensis (Echinodermata). Molecular Ecology 21: 2502-2518. Hoek PPC (1881): Report on the Pycnogonida, dredged by HMS Challenger during the Years 1873-76. Report on the Scientific Results of the Voyage of HMS Challenger, Zoology 3: 1-167.

148 Dissertation Lars Dietz

Krabbe K, Leese F, Mayer C, Tollrian R, Held C (2010): Cryptic mitochondrial lineages in the widespread pycnogonid Colossendeis megalonyx Hoek, 1881 from Antarctic and Subantarctic waters. Polar Biology 33: 281-292. Macher JN, Rozenberg A, Pauls SU, Tollrian R, Wagner R, Leese F (2015): Assessing the phylogeographic history of the montane caddisfly Thremma gallicum using mitochondrial and restriction-site associated DNA (RAD) markers. Ecology and Evolution 5: 648-662 Mahon AR, Arango CP, Halanych KM (2008): Genetic diversity of Nymphon (Arthropoda: Pycnogonida: Nymphonidae) along the Antarctic Peninsula with a focus on Nymphon australe Hodgson 1902. Marine Biology 155: 315–323 Miller MR, Dunham JP, Amores A, Cresko WA, Johnson EA (2007): Rapid and cost-effective polymorphism identification and genotyping using restriction site associated DNA (RAD) markers. Genome Research 17: 240-248. Munilla T, Soler Membrives A (2009): Check-list of the pycnogonids from Antarctic and sub- Antarctic waters: zoogeographic implications. Antarctic Science 21: 99-111. Nei M (1987): Molecular Evolutionary Genetics. Columbia University Press, New York. Raupach MJ, Thatje S, Dambach J, Rehm P, Misof B, Leese F (2010): Genetic homogeneity and circum-Antarctic distribution of two benthic shrimp species of the Southern Ocean, Chorismus antarcticus and Nematocarcinus lanceopes. Marine Biology 157: 1783-1797. Strugnell JM, Watts PC, Smith PJ, Allcock AL (2012): Persistent genetic signatures of historic climatic events in an Antarctic octopus. Molecular Ecology 21: 2775-2787. Tajima F (1989): Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123: 585-595. Thatje S (2012): Effects of capability for dispersal on the evolution of diversity in Antarctic benthos. Integrative and Comparative Biology 52: 470-482. Thatje S, Hillenbrand CD, Larter R (2005): On the origin of Antarctic marine benthic community structure. Trends in Ecology and Evolution 20: 534-540. Thatje S, Hillenbrand CD, Mackensen A, Larter R (2008): Life hung by a thread: Endurance of Antarctic fauna in glacial periods. Ecology 89: 682-692. Thornhill DJ, Mahon AR, Norenburg JL, Halanych KM (2008): Open-ocean barriers to dispersal: a test case with the Antarctic Polar Front and the ribbon worm Parborlasia corrugatus (Nemertea: Lineidae). Molecular Ecology 17: 5104-5117. Weis A, Melzer RR (2012): How did sea spiders recolonize the Chilean fjords after glaciation? DNA barcoding of Pycnogonida, with remarks on phylogeography of Achelia assimilis (Haswell, 1885). Systematics and Biodiversity 10: 361-374. Weis A, Meyer R, Dietz L, Dömel JS, Leese F, Melzer RR (2014): Pallenopsis patagonica (Hoek, 1881) – a species complex revealed by morphology and DNA barcoding, with

149 Dissertation Lars Dietz

description of a new species of Pallenopsis Wilson, 1881. Zoological Journal of the Linnean Society 170: 110-131 Wilson NG, Hunter RL, Lockhart SJ, Halanych KM (2007): Multiple lineages and absence of panmixia in the "circumpolar" crinoid Promachocrinus kerguelensis from the Atlantic sector of Antarctica. Marine Biology 152: 895-904. Wilson NG, Schrödl M, Halanych KM (2009): Ocean barriers and glaciation: evidence for explosive radiation of mitochondrial lineages in the Antarctic sea slug Doris kerguelenensis (Mollusca, Nudibranchia). Molecular Ecology 18: 965-984.

150 Dissertation Lars Dietz

6) Publikation V

Titel: Pallenopsis patagonica (Hoek, 1881) – a species complex revealed morphology and DNA barcoding, with description of a new species of Pallenopsis by Wilson, 1881

Zoological Journal of the Linnean Society 170: 110-131

Hinweise zu Publikation V

• Anteil Planung: 5%

• Anteil experimentelle Durchführung: 10%

• Verfassen des Manuskripts: 5%

Abbildungen, die nicht ausschließlich von mir erstellt wurden

• Alle Abbildungen wurden von Andrea Weis erstellt.

151 bs_bs_banner

Zoological Journal of the Linnean Society, 2014, 170, 110–131. With 11 figures

Pallenopsis patagonica (Hoek, 1881) – a species complex revealed by morphology and DNA barcoding, with description of a new species of Pallenopsis Wilson, 1881

ANDREA WEIS1*, ROLAND MEYER1, LARS DIETZ2, JANA S. DÖMEL2, FLORIAN LEESE2 and ROLAND R. MELZER1

1Zoologische Staatssammlung München, Münchhausenstraße 21, 81247 München, Germany 2Ruhr-Universität Bochum, Evolutionsökologie und Biodiversität der Tiere, Universitätsstraße 150, D-44801 Bochum, Germany

Received 7 May 2013; revised 6 October 2013; accepted for publication 7 October 2013

Pallenopsis patagonica (Hoek, 1881) is one of the most taxonomically problematic and variable pycnogonid species, and is distributed around the southern South American coast, and the Subantarctic and Antarctic areas. We conducted a phylogenetic analysis of mitochondrial cytochrome c oxidase subunit I (COI) sequences of 47 Pallenopsis specimens, including 39 morphologically identified as P. patagonica, five Pallenopsis pilosa (Hoek, 1881), one Pallenopsis macneilli Clark, 1963, one Pallenopsis buphtalmus Pushkin, 1993, and one Pallenopsis latefrontalis Pushkin, 1993. Furthermore, we studied morphological differences between the different COI lineages using light and scanning electron microscopy, including also material from Loman’s and Hedgpeth’s classical collections, as well as Hoek’s type material of P. patagonica from 1881. The molecular results unambiguously reveal that P. patagonica is a complex of several divergent clades, which also includes P. macneilli, P. buphtalmus, and P. latefrontalis. Based on the material available, two major clades could be identified, namely a ‘Falkland’ clade, to which we assign the nominal P. patagonica, and a ‘Chilean’ clade, which is distinct from the ‘Falkland’ clade. We describe the ‘Chilean’ clade as new species, Pallenopsis yepayekae sp. nov. Weis, 2013. All molecular results are confirmed by specific morphological characteristics that are discussed in detail and compared with Pallenopsis species closely related to the P. patagonica complex. Our results reveal that P. patagonica is a species-rich complex that is in need for a thorough taxonomic revision, using both morphological and genetic approaches.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131. doi: 10.1111/zoj.12097

ADDITIONAL KEYWORDS: biogeography – Chile – COI – cryptic species – Pallenopsidae – Pantopoda – Subantarctic.

INTRODUCTION this species, viz. Pallenopsis glabra Möbius, 1902, Pallenopsis hiemalis Hodgson, 1907, Pallenopsis Pallenopsis patagonica (Hoek, 1881), from the meridionalis Hodgson, 1915, Pallenopsis moebiusi material of the HMS Challenger expedition, was, as Pushkin, 1975, and Bathypallenopsis meridionalis the name implies, first sampled off southern South (Hodgson, 1927) (Bamber & El Nagar, 2011). In addi- American coasts. It represents one of the most tion, some valid species exist that are morphologically taxonomically problematic and variable pycnogonid very similar to P. patagonica, e.g. Pallenopsis species known to date. The complexity can already buphtalmus Pushkin, 1993. Pallenopsis patagonica is be recognized by the various synonyms that exist for known from Antarctic and Subantarctic regions, mainly the Scotia Sea, Ross Sea, Antarctic Peninsula, and South America, including the Magellan Strait, *Corresponding author. E-mail: [email protected] but is also known from the Falkland Islands, South

110 © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 111

Georgia, and shelf regions from the east Antarctic performed for the genera Colossendeis (Krabbe et al., sector (Hoek, 1881; Möbius, 1902; Hodgson, 1907; 2010; Dietz et al., 2011; Dietz et al., 2013), Nymphon Loman, 1923a, b; Gordon, 1932; Marcus, 1940; (Mahon et al., 2008; Arango, Soler-Membrives & Hedgpeth, 1961; Pushkin, 1975, 1993; Stock, 1975; Miller, 2011) and Pseudopallene (Arango & Brenneis, Müller, 1993; Child, 1995; Munilla & Soler 2013). With Pallenopsis we want to open the field for Membrives, 2009; Weis & Melzer, 2012b). Specimens a further, very complex, variously shaped group, with can be found in depths ranging from 3 down to a focus on southern South American coasts and sur- 4540 m (Munilla & Soler Membrives, 2009). rounding areas. To unscramble the complex taxonomy of P. patagonica, and to test whether all morphologically MATERIAL AND METHODS variable specimens available for our analysis repre- sent a single species, we sequenced a fragment of SPECIMENS AND VOUCHERS the mitochondrial cytochrome c oxidase subunit I Specimens from the Chilean coast were collected by (COI) gene. This gene is variable, and has been SCUBA diving during expeditions organized by the applied successfully for species-level taxonomy in Huinay Scientific Field Station between 2006 and pycnogonids (Mahon, Arango & Halanych, 2008; 2011 (Försterra, 2009). Additionally, we received Krabbe et al., 2010; Weis & Melzer, 2012a). Alto- material from the region of Valparaiso, a more north- gether, 39 P. patagonica specimens were sampled ern area in Chile, from the Falkland Islands, South from 33–72°S and 11–170°W (in a depth range of Georgia, and the Weddell Sea (see Acknowledge- 3–466 m), with a focus on the area around the south- ments). A detailed overview of the different sample ern tip of South America. Furthermore, the morphol- locations of the studied individuals is given in ogy of all available specimens was studied in detail Figure 1. Material was fixed in 96% ethanol to ensure with light and scanning electron microscopy (SEM), high-quality DNA for genetic analysis. Pycnogonids demonstrating differences among samples from differ- were identified based on morphology using a variety ent localities. Morphological analyses include speci- of literature (Hoek, 1881; Möbius, 1902; Hodgson, mens from the type material of Pallenopsis tumidula 1907; Gordon, 1932, 1944; Stock, 1957, 1975; Loman, 1923 (SMNH Type 1293 and syntypes), one Pushkin, 1975, 1993; Child, 1995; Weis & Melzer, specimen of P. patagonica (SMNH-125527) from 2012b). Furthermore, synonyms, depth ranges, and Hedgpeth’s collections from the Swedish Museum distribution patterns were taken from Müller’s (1993) of Natural History (Loman, 1923b; Hedgpeth, 1961), World Catalogue and Bibliography of the recent eight other specimens of P. patagonica (SMNH- Pycnogonida, Munilla & Soler Membrives (2009), and 125445, SMNH-125507, SMNH-125508, SMNH- Pycnobase (Bamber & El Nagar, 2011). All barcoded 125509, SMNH-125510), and one unidentified voucher specimens are kept at ZSM under specific Pallenopsis sp. (SMNH-125514). In addition, we voucher IDs (see Table 1), including PpaE_001–008, also studied/consulted Hoek’s type material of PpaE_010, PpaA_001, and PxxE001-002. The respec- P. patagonica (BMNH 1881.38, three specimens) tive DNA extract aliquots are stored partially at and P. patagonica var. elegans (BMNH 188.38, one the Canadian Center for DNA Barcoding (CCDB), specimen), which are kept in the Natural History the ZSM DNA bank facility, and Ruhr University Museum in London. Furthermore, we analysed three Bochum. Collection data, BOLD or GenBank acces- Pallenopsis notiosa Child, 1992 specimens, which are sion numbers of all 39 pycnogonid sequences exam- housed at the Zoologische Staatssammlung München ined in this study, as well as chosen out-group taxa (ZSM) (Weis & Melzer, 2012b). Our morphological are summarized and listed on Table 1. Some of the data set includes a total of 61 specimens. specimen details can further be accessed in the As mentioned in our previous study (Weis & Melzer, Barcode of Life Data Systems (Ratnasingham & 2012a), the southern Chilean coastline provides Hebert, 2007), under the project Chilean Fjord an interesting opportunity for studying speciation Pycnogonids (CFAP), as part of the Marine Life processes. Given that the last glaciation ended only (MarBOL) campaign. The sequences FJ969367–69 15 000 years ago, and the low dispersal ability of of P. patagonica from the Ross Sea were accessed pycnogonids, haplotypes of cryptic species have only from GenBank (Nielsen, Lavery & Lörz, 2009). a rather limited geographical distribution, as was Furthermore, we used five GenBank sequences the case for Achelia assimilis (Haswell, 1885) (Weis & of Pallenopsis pilosa (Hoek, 1881) (PxxE001, Melzer, 2012a). Whether similar effects can be found PxxE002, KC848052, KC848053, KC848054), one concerning the species P. patagonica is one aim of the sequence of P. buphtalmus Pushkin, 1993 present study. (HM426171), one Pallenopsis latefrontalis Pushkin, As yet, further molecular studies focusing on par- 1993 (HM426218), and Pallenopsis macneilli Clark, ticular groups of pycnogonids have only explicitly been 1963 (DQ390086) as outgroups. Although specimens

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 112 A. WEIS ET AL.

Figure 1. Map of sampling sites of Chilean, Antarctic, and Subantarctic Pallenopsis specimens deposited at the Bavarian State Collection of Zoology. Sequences of specimens from the Ross Sea were downloaded from GenBank.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 04TeLnenSceyo London, of Society Linnean The 2014 © Table 1. Summary of collection data and registration of specimens used in this study

BOLD ID/ Voucher ID Haplotype Species Country/Region Latitude Longitude Depth GenBank ID

ZSMA20111000 HT 7 Pallenopsis yepayekae sp. nov. Chile; Region de Magallanes y de la Antarctica Chilena 48°44′11.4″S 75°24′53.1″W 15 m CFAP013-11 ZSMA20111002 HT 6 Pallenopsis yepayekae sp. nov. Chile; Region de Magallanes y de la Antarctica Chilena 50°50′07.1″S 74°08′20.9″W 25 m CFAP017-11 ZSMA20111003 HT 3 Pallenopsis yepayekae sp. nov. Chile; Region de los Lagos 43°25′03.0″S 74°04′51.2″W 25 m CFAP006-11 ZSMA20111004 HT 1 Pallenopsis yepayekae sp. nov. Chile; Region de los Lagos 43°24′34.5″S 74°05′00.7″W 9 m CFAP005-11 ZSMA20111005 HT 4 Pallenopsis yepayekae sp. nov. Chile; Region de Magallanes y de la Antarctica Chilena 48°44′11.4″S 75°24′53.1″W 23 m CFAP014-11 ZSMA20111006 HT 1 Pallenopsis yepayekae sp. nov. Chile; Region de los Lagos 43°25′03.0″S 74°04′51.2″W 20 m CFAP007-11 ZSMA20111008 HT 28 Pallenopsis patagonica Chile; Region de Magallanes y de la Antarctica Chilena 50°24′52″S 74°33′33″W 15–25 m CFAP026-11 ZSMA20111009 HT 2 Pallenopsis yepayekae sp. nov. Chile; Region de los Lagos 43°23′33.4″S 74°07′56.5″W 26 m CFAP004-11 ZSMA20111012 HT 8 Pallenopsis yepayekae sp. nov. Chile; Region de los Lagos 43°46′28.5″S 073°02′63.2″W 22 m CFAP008-11 ZSMA20111016 HT 9 Pallenopsis yepayekae sp. nov. Chile; Region de Magallanes y de la Antarctica Chilena 48°36′28.7″S 74°53′55.7″W 32 m CFAP012-11 ZSMA20111017 HT 11 Pallenopsis patagonica Chile; Region de Magallanes y de la Antarctica Chilena 48°36′28.7″S 74°53′55.7″W 32 m CFAP025-11 olgclJunlo h ina Society Linnean the of Journal Zoological ZSMA20111024 HT 10 Pallenopsis yepayekae sp. nov. Chile; Region de Magallanes y de la Antarctica Chilena 49°34′38.7″S 74°26′49.3″W 28 m CFAP016-11 ZSMA20111072 HT 29 Pallenopsis patagonica Chile; Region de Valparaiso 33°23′55″S 71°52′78.2″W 339 m CFAP023-11 ZSMA20111339 HT 5 Pallenopsis yepayekae sp. nov. Chile; Anihue Raul Marin Balmaceda 43°46′31.35″S 73°01′44.14″W 19 m CFAP019-11 ZSMA20111340 HT 12 Pallenopsis patagonica Chile; Region de Magallanes y de la Antarctica Chilena 55°00′00.6″S 68°18′88.1″W 24 m CFAP018-11 ZSMA20111348 HT 14 Pallenopsis patagonica Falkland Islands 50°26′4.00″S 62°46′5.00″W 146–148 m CFAP027-11 ′ ″ ′ ″ ZSMA20111349 HT 13 Pallenopsis patagonica Falkland Islands 51°16 8.00 S 62°57 8.00 W 171–174 m CFAP034-11 PATAGONICA PALLENOPSIS ZSMA20111350 HT 15 Pallenopsis patagonica Falkland Islands 51°16′8.00″S 62°57′8.00″W 171–174 m CFAP035-11 ZSMA20111351 HT 20 Pallenopsis patagonica Falkland Islands 51°16′8.00″S 62°57′8.00″W 171–174 m CFAP036-11 ZSMA20111352 HT 27 Pallenopsis patagonica Falkland Islands 51°16′8.00″S 62°57′8.00″W 171–174 m CFAP037-11 ZSMA20111354 HT 17 Pallenopsis patagonica Falkland Islands 51°05′8.00″S 61°44′0.00″ W 174–176 m CFAP028-11 ZSMA20111355 HT 18 Pallenopsis patagonica Falkland Islands 51°05′8.00″S 61°44′0.00″W 174–176 m CFAP029-11 ZSMA20111357 HT 16 Pallenopsis patagonica Falkland Islands 51°05′8.00″S 61°44′0.00″W 174–176 m CFAP030-11 ZSMA20111359 HT 18 Pallenopsis patagonica Falkland Islands 51°05′8.00″S 61°44′0.00″W 174–176 m CFAP031-11 ZSMA20111360 HT 15 Pallenopsis patagonica Falkland Islands 51°05′8.00″S 61°44′0.00″W 174–176 m CFAP032-11 ZSMA20111361 HT 19 Pallenopsis patagonica Falkland Islands 51°05′8.00″S 61°44′0.00″W 174–176 m CFAP033-11 PpaE_004 HT 18 Pallenopsis patagonica Falkland Islands 52°57′42″S 60°08′36″W 378 m KC794961 PpaE_005 HT 15 Pallenopsis patagonica Falkland Islands 52°57′42″S 60°08′36″W 378 m KC794962 PpaE_006 HT 17 Pallenopsis patagonica Falkland Islands 52°57′42″S 60°08′36″W 378 m KC794963 2014, , PpaE_007 HT 15 Pallenopsis patagonica Falkland Islands 52°57′42″S 60°08′36″W 378 m KC794964 PpaE_008 HT 15 Pallenopsis patagonica Falkland Islands 52°57′42″S 60°08′36″W 378 m KC794965 PpaE_010 HT 15 Pallenopsis patagonica Falkland Islands 52°57′42″S 60°08′36″W 378 m KC794966 PCE COMPLEX SPECIES A – 170 PpaE_001 HT 24 Pallenopsis patagonica Subantarctic; West of South Georgia; Shag Rocks 53°46′12″S 41°26′6″W 193 m KC794959 ′ ″ ′ ″ 110–131 , PpaE_002 HT 25 Pallenopsis patagonica Subantarctic; South Georgia 54°00 59 S 37°26 14 W 78 m KC794960 PpaE_003 HT 26 Pallenopsis patagonica Subantarctic; Burdwood Bank 54°33′00″S 58°49′20″W 158 m KC794969 PpaA_001 HT 23 Pallenopsis patagonica Antarctic; Eastern Weddell Sea 71°08′09″S 11°31′37″W 123 m KC794958 NIWA46256 HT 21 Pallenopsis patagonica Antarctic; Ross Sea 71°15′45″S 170°38′08″W 466 m FJ969367 NIWA46257 HT 21 Pallenopsis patagonica Antarctic; Ross Sea 72°00′81″S 170°46′47″W 235.5 m FJ969368 NIWA46258 HT 22 Pallenopsis patagonica Antarctic; Ross Sea 71°37′24″S 170°51′99″W 204.5 m FJ969369 HM426218 Pallenopsis latefrontalis Antarctic; Eastern Weddell Sea 71°5′31.23″S 11°30′28.8″W 302 m HM426218 HM426171 Pallenopsis buphtalmus Antarctic; Eastern Weddell Sea 71°19′1.2″S 13°56′31.2″W 848 m HM426171 DQ390086 Pallenopsis macneilli Australia; Rocky Point, Torquay 38°20′38.07″S 144°19′12.77″E 0.5 m DQ390086 PxxE001 Pallenopsis pilosa Subantarctic; Bouvet Island 54°21′00″S 3°11′36″E 465 m KC794967 PxxE002 Pallenopsis pilosa Subantarctic; Bouvet Island 54°21′30″S 3°26′6″E 200 m KC794968 KC848053 Pallenopsis pilosa Antarctica 66°23′S 140°25′43.87″E 743 m AAC7281 ′ ″ ′ ″

KC848054 Pallenopsis pilosa Antarctica 65°52 11.81 S 143°0 5.57 E 428 m AAC7183 113 KC848052 Pallenopsis pilosa Antarctica 65°51′9.32″S 144°2′23.15″E 1104 m AAC7182 114 A. WEIS ET AL.

PxxE001 and PxxE002 were checked for correct deter- Tris-Borate-EDTA (TBE) agarose gel. 10 μL of the PCR mination, we could not access the outgroup specimens product were purified enzymatically with 0.5 μL KC848052, KC848053, and KC848054 (deposited at Exonuclease I (20 U μL–1) and 1 μL FastAP (1 U μL–1; the British Antarctic Survey in Cambridge), and Thermofisher), by incubating in a thermocycler at HM426171, HM426218, and DQ390086. 37 °C for 15 min, followed by 96 °C for 15 min prior For comparative morphological analyses, in addition to sequencing. Sequencing was conducted at GATC to our specimens used for DNA sequencing, we inves- (Konstanz, Germany) or performed partially at the tigated 18 specimens from historical collections housed CCDB using the standard protocols of IBOL. at the Swedish Museum of Natural History and the British Museum of Natural History, i.e. P. tumidula (SMNH Type 1293 and seven syntypes), P. patagonica PHYLOGENETIC ANALYSIS (SMNH-125445, SMNH-125507, SMNH-125508, A total of 47 pycnogonid sequences were used for the SMNH-125509, SMNH-125510), and one unidentified phylogenetic analyses of the 657-bp fragment of COI. Pallenopsis sp. (SMNH-125514) from the Loman col- All 47 DNA sequences were aligned with MUSCLE lection, as well as one P. patagonica (SMNH-125527) using GENEIOUS PRO 5.5.4 (Drummond et al., 2011). sampled by the Lund University Chile expedition, To check for frameshift mutations or stop codons, the determined by Hedgpeth (label: det. Hedgpeth 1949). COI sequences were translated into amino acids using Beyond that we examined Hoek’s type material from the invertebrate mitochondrial genetic code (transla- the HMS Challenger expedition, which include three tion table 5, available from http://www.ncbi.nlm.nih specimens of P. patagonica and one specimen desig- .gov/Taxonomy/taxonomyhome.html/index.cgi?chapter nated as P. patagonica var. elegans (BMNH 1881.38). =cgencodes). After the calculation of ‘base pair’ fre- Furthermore, we studied a related species, P. notiosa quencies and uncorrected pairwise distances with (ZSMA20111077–79), which is kept at the ZSM, and MEGA 5.05 (Tamura et al., 2011), we tested the align- has been discussed in a previous paper (Weis & Melzer, ment statistically for substitution saturation in 2012b). DAMBE 5.2.69 (Xia et al., 2003; Xia & Lemey, 2009). For morphological documentation we used the fol- Using MEGA 5.05 software we calculated nucleotide lowing specimens: ZSMA20111000, ZSMA20111002, composition, maximum parsimony (MP), and, as we ZSMA20111004, ZSMA20111006, ZSMA20111009, were interested in shallow species-level differences, ZSMA20111016, ZSMA20111348, ZSMA20111350, neighbour-joining (NJ) trees based on the Kimura ZSMA20111357, PpaE007, and PpaE010 for two-parameter (K2P) model (Kimura, 1980; Saitou & light microscopy; ZSMA20111006, ZSMA20111009, Nei, 1987), with bootstrap values. For maximum like- ZSMA20111024, ZSMA20111349, ZSMA20111359, lihood (ML) and Bayesian inference (BI) we first and ZSMA20111360 for SEM studies. identified the most appropriate substitution model using jMODELTEST 2 and the Akaike/Bayesian infor- mation criteria (AIC/BIC; Darriba et al., 2012). For ML DNA EXTRACTION AND SEQUENCING we used the full set of 88 models, for MrBayes we used As all the individuals studied were of a suitable size, the reduced model search scheme (nst = 1, 2, and 6; +I, it was sufficient to take only a piece of leg for DNA +G, and +IG). Just as for MP and NJ, we used 1000 extraction. Here, muscle tissue from the tibia was bootstrap replicates for the ML analysis under extracted using the DNeasy Mini Kit following the RAxML 7.0.4. The 1000 rapid bootstraps were con- manufacturer’s tissue protocol. As a modification from ducted by using the –x option (random seed number). the original protocol, we used only 100 μL of EB buffer Based on jMODELTEST 2 the best model, according to for elution. Amplification of a 657-bp fragment of COI both AIC and BIC, was GTR+I + G, and this was used was performed using standard Folmer primers in RAxML and the Bayesian analyses with MrBayes (Folmer et al., 1994) in 25-μL reactions. Individual 3.2 (Ronquist et al., 2012). Bayesian analysis was reactions consisted of 1× polymerase chain reaction performed using four independent runs with four (PCR) buffer (5Prime HotMaster), 0.2 mM independent chains and 5 million Metropolis-coupled deoxyribonucleotides (dNTPs), 0.5 μM of each primer, Markov chain Monte Carlo (MCMC) generations each. 0.025 U μL–1 Taq (5Prime Hotmaster), 1–3 μL Every 500th tree was saved (giving 10 000 in total). extracted DNA (depending on yield), and was filled The four independent runs reached stationarity after up to 25 μL with molecular biology-grade H2O. Cycle 0.7–0.9 million generations (average standard devia- conditions were: initial denaturation at 94 °C for tion of split frequencies below 0.01), and thus the 2 min, followed by 36 cycles of 94 °C for 20 s, 48 °C for consensus tree was calculated after discarding the first 30 s, and 65 °C for 80 s. After a final extension at 65 °C 25% of the trees as burn-in (1.25 million generations). for 5 min the reactions were stored at 4 °C. Both DNA The figure of the recovered phylogenetic tree was made extraction and PCR success were checked on a 1% using FigTree 1.4.0.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 115

SEARCH FOR SPECIES BOUNDARIES USING as described by Weis & Melzer (2012a). The editing and DNA SEQUENCES composition of both light microscopic and SEM pic- To be independent from morphology, we decided tures was performed with Adobe PHOTOSHOP CS. to perform molecular analyses on the whole data set, including P. macneilli, P. buphtalmus, and P. latefrontalis. To check for species boundaries in our NOMENCLATURAL ACTS P. patagonica complex, we conducted a general mixed The electronic edition of this article conforms to the Yule-coalescent (GMYC) analysis (Pons et al., 2006; requirements of the amended International Code of Monaghan et al., 2009). As identical sequences cannot Zoological Nomenclature (ICZN), and hence the new be considered by GMYC we removed identical name contained herein is available under that code sequences, resulting in a data set of 32 sequences. An from the electronic edition of this article. This pub- ultrametric starting tree was obtained using BEAUTi lished work and the nomenclatural acts it contains and BEAST (both versions 1.6.1; Drummond & have been registered in ZooBank, the online registra- Rambaut, 2007). The chain length for the MCMC tion system for the ICZN. The ZooBank life-science algorithm was set to 10 million generations, sampling identifiers (LSIDs) can be resolved and the associated trees every 1000 generations. Effective sampling sites information viewed through any standard web and convergence of the parameter estimates was browser by appending the LSID to the prefix ‘http:// inspected using TRACER 1.5. A consensus tree was zoobank.org/’. The LSID for this publication is: obtained using TreeAnnotator 1.6.1. The burn-in was urn:lsid:zoobank.org:act:0E39E226-30C7-4853-A6A1- set to 2500, rejecting the first 25% of the trees, and 7DD2336F33FE. The electronic edition of this work the posterior probability threshold was set to 0.5. The was published in a journal with an international resulting ultrametric tree was subsequently imported standard serial number (ISSN), and has been into the statistics software R 2.15.2 (http://www.R- archived and is available from the following digital project.org/). GMYC analysis was conducted with the repositories: PubMed Central and LOCKSS. R package ‘SPLITS’ (Species Limits by Threshold Statistics; http://r-forge.r-project.org/projects/splits). We used the single and multiple threshold model for RESULTS the inference of the number of entities with standard MOLECULAR AND PHYLOGENETIC ANALYSIS parameters [interval = c(0, 10)] and used a likelihood ratio test to select the appropriate model. The 657-bp COI alignment of 47 pycnogonid speci- Furthermore, we used the freely available soft- mens showed no gaps. Base-pair frequencies indi- ware ABGD (Automatic Barcode Gap Discovery; cated an arthropod-typical bias towards adenosine Puillandre et al., 2012) for searching barcoding gaps and thymine: A, 31.31%; C, 19.80%; G, 13.95%; and T, between all 42 sequences (sequences of P. pilosa were 34.68%. The value of substitution saturation (Iss), excluded), and for calculating their intraspecific which was calculated for the whole alignment as well distance/variance. as for the third codon position, was always signifi- cantly lower than the critical value (Iss.c). An Iss value

lower than Iss.c implies only a low level of substitution COI NETWORK saturation for the sequences analysed. The 657 base As networks are better suited to visualize the often pairs consisted of 410 conserved sites and 247 vari- reticulate relationships within, as well as among, able sites, of which 202 were parsimony-informative. closely related species, we constructed a NeighborNet Translating the COI sequences into amino acid of all individual COI sequences, using SPLITSTREE sequences showed neither frame-shift mutations nor 4.12 (Huson & Bryant, 2006) and K2P-corrected stop codons. distances. Phylogenetic trees constructed using different approaches (BI, MP, NJ, ML) showed no major differ- ences, and therefore we present the Bayesian tree MORPHOLOGICAL ANALYSIS (Fig. 2). Support values for the other methods are also Specimens were photographed using a Wild M400 shown on the branches. Minor differences are found photomacroscope, equipped with a digital camera with respect to the position of ZSMA20111008 and (Nikon D700), by taking several shots focused at ZSMA20111072. Both slightly change in position different levels along the z-axis. To constitute a greater within the tree, but never affect any of the other depth of field this series of pictures was then edited well-supported clades. and combined to form a single respective image using Specimens of P. patagonica from Chile the computer software Helicon Focus (http://www (ZSMA20111000, ZSMA20111002–06, ZSMA20111009, .heliconsoft.com/). Specimens were prepared for SEM ZSMA20111012, ZSMA20111016, ZSMA20111024,

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 116 A. WEIS ET AL.

ZSMA20111072 HT29 P. macneilli 1/46/-/67 DQ390086 ZSMA20111008 HT28

ZSMA20111004 HT1

1/25/-/68 ZSMA20111006 HT1

ZSMA20111003 HT3

ZSMA20111009 HT2 1/100/99/100 ZSMA20111005 HT4 Chile ZSMA20111339 HT5 clade ZSMA20111002 HT6

ZSMA20111016 HT9

0.62/48/-/54 1 ZSMA20111012 HT8

ZSMA20111024 HT10

ZSMA20111000 HT7

PpaE003 HT26 1/100/99/100 ZSMA20111352 HT27

ZSMA20111017 HT11 1/100/99/100 ZSMA20111340 HT12

1/100/99/100 ZSMA20111351 HT20

1/25/40/88 ZSMA20111361 HT19

ZSMA20111359 HT18 0.63/91/ 92/100 1 ZSMA20111355 HT18

1 PpaE004 HT18

ZSMA20111360 HT15

PpaE008 HT15 Falkland

PpaE007 HT15 clade

PpaE005 HT15

0.89/61/ ZSMA20111350 HT15 41/75 PpaE010 HT15

ZSMA20111349 HT13

ZSMA20111348 HT14 0.60/-/-/72 ZSMA20111357 HT16

PpaE006 HT17

ZSMA20111354 HT17

HM426218 P. latefrontalis 0.83/67/ 1/99/98/100 56/85 FJ969367 HT21 1/100/ 99/100 FJ969368 H21 1/98/97/100 0.98/76/ FJ969369 HT22 75/96 0.85/-/ 54/78 HM426171 P. buphtalmus

PpaA001 HT23 0.84/93/75/78 PpaE001 HT24

PpaE002 HT25

CEA047 1/100/99/100 PxxE001 Outgroup PxxE002 P. pilosa CEA112

CEA082 0.07 subst./site

Figure 2. Bayesian phylogenetic tree of cytochrome c oxidase subunit I (COI) sequences of 28 Pallenopsis patagonica (Falkland clade and others), 11 Pallenopsis yepayekae sp. nov. (Chile clade), one Pallenopsis macneilli, one Pallenopsis buphtalmus, one Pallenopsis latefrontalis, and five Pallenopsis pilosa, which serve as the outgroup. Posterior probabilities of the Bayesian inference and bootstrap values of neighbor-joining (NJ), maximum-parsimony (MP), and maximum- likelihood (ML) analyses are displayed above or below branches; branch lengths indicate substitutions per site. Different haplotypes of the studied specimens are defined as HT1–HT29.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 117

ZSMA20111339) and the Falkland Islands (ZSMA the Falkland and ‘Antarctic’ clades (although with 20111348–51, ZSMA20111354–55, ZSMA20111357, low support). Pallenopsis macneilli clusters with ZSMA20111359–61, PpaE004–008, PpaE010) cluster ZSMA20111008. The results reveal that specimens within two well-supported, geographically distinct initially identified as P. patagonica are genetically clades (Figs 2, 3). Several specimens cluster outside very heterogeneous, and some show close affinities these two distinct groups, highlighting the complex with specimens identified as different species. nature of P. patagonica: ZSMA20111008, ZSMA Figure 3 shows the NeighborNet of all Pallenopsis 20111072 (both from 33°S), [ZSMA20111017 (48°S) specimens. and ZSMA20111340 (Region de Magallanes)], and The five specimens of P. pilosa selected as the out- [PpaE003 and ZSMA20111352 (Falklands)]. Speci- group cluster apart from all other 42 pycnogonids. In mens from the Ross Sea (FJ969367–69) cluster the phylogenetic trees, the statistical support for the together with one individual from the eastern Weddell in-group is good for the model-based inferences (BI, 1; Sea (PpaA001), one from the Shag Rocks (PpaE001), ML, 88), but is poor for the NJ and MP inferences and two from the Southern Ocean assigned to dif- (25 and 40, respectively). Interestingly, the five ferent species (P. buphtalmus and P. latefrontalis), P. pilosa specimens are genetically highly hetero- forming an ‘almost Antarctic’ clade. The only specimen geneous, hinting at further problems with the tax- from South Georgia (PpaE002) clusters basally with onomy of other Pallenopsis specimens. In general, the

Figure 3. NeighborNet of all individual cytochrome c oxidase subunit I (COI) sequences, using SPLITSTREE and Kimura two-parameter (K2P) corrected distances.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 118 A. WEIS ET AL.

Figure 4. Pairwise genetic distances (Kimura’s two- parameter, K2P) for cytochrome c oxidase subunit I (COI) Figure 5. Automatic Barcode Gap Discovery (ABGD) sequences of Pallenopsis specimens used in the present analysis for 42 Pallenopsis specimens used in the present study (with Pallenopsis pilosa excluded). study (with Pallenopsis pilosa excluded). mitochondrial COI fragment is a suitable marker for uncovering lineages previously undetected by mor- phological analyses (see also Weis & Melzer, 2012a). To test whether these clusters comprise cryptic or overlooked species we calculated and compared uncorrected pairwise distances between the different specimens/clades (see Table S1). Variation between clades or specimens are high, with a maximum of 23.6% uncorrected pairwise genetic distance. Genetic distances between P. patagonica sensu stricto (Falk- land clade) and Pallenopsis yepayekae sp. nov. (Chilean clade) were high (14.9–19.1%), whereas the variation within these clades was low (0–1.1% and 0–3.5%, respectively). In addition, we analysed the distance data for distinct barcode gaps using ABGD. Including all 42 Figure 6. Automatic Barcode Gap Discovery (ABGD) ana- pycnogonids studied (five specimens of P. pilosa were lysis for 27 Pallenopsis specimens: 16 specimens from the excluded), no barcode gap is visible (Fig. 5). When Falkland clade versus 11 specimens from the Chilean clade. ranking the pairwise genetic distances and plotting them there is a large increase at the beginning of the slope, the two horizontal lines are connected by also compared the single-threshold model versus the several dots or small clusters of dots; however, when multiple-threshold model, and found support for the repeating the analyses only including the 11 speci- single-threshold model P = 0.861 (χ2 = 0.751 and three mens from the Chilean clade (Pallenopsis yepayekae degrees of freedom). According to the single-threshold sp. nov.) together with the 16 specimens from the GMYC model, the tree consists of 32 haplotypes split Falkland clade (Pallenopsis patagonica sensu stricto) into three clusters (confidence interval, 3–5) and 15 a barcoding gap becomes more obvious (Fig. 6). The distinct GMYC species (ML entities; confidence inter- two horizontal lines in the distance plot are now val, 11–16). The threshold between Yule speciation clearly separated vertically, without any dots in and coalescence within populations is indicated by a between them. vertical line in the lineage-through-time plot (LTT) of For the tree-based assessment of unrecognized the Bayesian tree in Figure 7. According to this, the species, using the GMYC model with multiple branch- specimens in the Falkland clade and the Chilean clade ing events (indicating the presence of several species) represent two distinct GMYC species. Furthermore, was preferred over the null model (single coalescent [FJ969367 and FJ96968 (HT21)] and [ZSMA20111017 branching model): likelihood ratio test, P < 0.001. (HT11) and ZSMA20111340 (HT12)], and all other 11 This indicates the presence of several species. We specimens represent distinct GMYC species.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 119

Specimens from the Chilean coast seem to be smaller in their body size compared with the speci- mens captured from the Falkland Islands and South Georgia (except ZSMA20111352 and PpaE003, from Burdwood Bank; Figs 8A, 9A). Whereas the shape of the proboscis is cylindrical along its length for most specimens (Fig. 8B), individuals from the Chilean clade show a distinct swelling at the middle of the proboscis (Fig. 9C). Also, specimens ZSMA20111008 (Region de Magallanes), ZSMA20111072 (Region de Valparaiso), ZSMA20111352 (Falkland Islands), and PpaE003 (Burdwood Bank) show a light swelling at about half the length of the proboscis. Almost all specimens studied bear an upward-erected slender abdomen (except PpaE002, horizontal), with some short setae. The abdomen from specimens from the Chilean clade is dorsodistally sloped. At the beginning of the slope a rounded edge is found bearing two very prominent spines (Fig. 9D). In contrast, specimens Figure 7. Lineage-through-time plot of the number of from the Falkland Islands and the Antarctic lack lineages (N) in the linearized Bayesian haplotype tree those spines, but show several randomly distributed (32 unique cytochrome c oxidase subunit I, COI, barcode short setae on the abdomen (Fig. 8C). All individuals sequences). Vertical line represents the single threshold examined show a pointed or slightly pointed ocular identified by the GMYC model between Yule speciation tubercle. Specimen ZSMA20111008 is the only one and coalescence within populations. The number of GMYC with a rounded ocular tubercle. species suggested by this analysis was 15. Furthermore, whereas the length of the cement gland tubes in the Chilean pycnogonids is about three times their diameter (Fig. 10B), specimens from the MORPHOLOGY Falklands and Antarctic area show a very short To check whether the results of our sequence analy- cement gland tube (Fig. 10A), which is sometimes ses, i.e. that P. patagonica might be a complex of only hardly visible. Additionally, females of the several species, are paralleled by previously unde- Chilean clade show a swollen fourth oviger segment, tected morphological differences between these clades which is not noticeable in the females from the Falk- we made a detailed analysis of all available speci- land clade (Figs 8D, 9F, 10C, D). Furthermore, female mens. Table S2 displays the enormous morphologi- ovigers from the Chilean clade are eight- to nine- cal variance of the different clades/specimens with segmented (distal segments often fused), compared respect to their general body size, length of the with females of the Falkland clade that exhibit a cement gland tube, leg setation, and auxiliary claw ‘ten-segmented’ oviger (Fig. 10C, D). length, but they all fit in the traditional definition of The proportion of the length of the different leg P. patagonica. As for most cases only one specimen is segments is similar throughout all specimens studied, available, and because these lack morphological dif- with tibia 2 being the longest. The number of heel ferences that allow us to decide whether they repre- spines on the propodus varies between three and sent variations or putative species-specific features, four (Fig. 8F). Concerning the leg setae, all individ- their analysis will be continued when more specimens uals have setae not longer than the diameter of are available. Thus we focused our analyses on the the segment upon which they are situated (except two biggest clades, initially referred to as the Chilean ZSMA20111017). The 11 specimens from the Chilean clade and the Falkland clade (including 11 and 16 clade show numerous distinct small and stout hairs specimens, respectively). Within each of the two on the distal ventral side of the second and third clades we observed constant morphological features, coxae (Fig. 9E). Although this characteristic is weakly which is in accordance with the molecular results. developed in juveniles, it is already discernable at Light-microscopy pictures of individuals from the that stage. This characteristic is not visible or very Falkland and the Chilean clades are shown in prominent in any of the other specimens studied Figures 8 and 9. Furthermore, Figure 10 displays (Fig. 8E). Furthermore, the setae themselves show detailed SEM studies of the cement gland ducts, remarkable differences. The setae on the second and female ovigers, and hairs of the second and third third coxae of the specimens from the Chilean clade coxae from specimens of both clades. bear several tiny hairs on their surface (Fig. 10F),

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 120 A. WEIS ET AL.

Figure 8. Light microscopy of Pallenopsis patagonica sensu stricto. (Falkland clade). A, dorsal view; scale bar=4mm. B, ventral view of proboscis; scale bar = 2 mm. C, dorsal view of abdomen; scale bar = 500 μm. D, right oviger (female); scale bar = 500 μm. E, detail view of second and third coxae of left fourth walking leg; scale bar = 1 mm. F, tarsus and propodus with claw and auxiliary claws of right third walking leg; scale bar = 500 μm. A, PpaE007; B, PpaE010; C, ZSMA20111357; D, E, ZSMA20111350; F, ZSMA20111348. Abbreviations: ac, auxiliary claws; cf, chelifore; cl, claw; cx, coxa; eg, eggs; fm, femur; os, oviger segment; ov, oviger; pp, propodus; pr, proboscis; ts, tarsus; tb, tibia.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 121

Figure 9. Light microscopy of Pallenopsis yepayekae sp. nov. (Chilean clade). A, dorsal view; scale bar = 2 mm. B, lateral view of trunk; scale bar = 1 mm. C, ventral view of proboscis; scale bar = 500 μm. D, detail view of abdomen, note two prominent spines (arrows); scale bar = 250 μm. E, detail view of second and third coxae of right walking legs, note several short and prominent hairs (arrows); scale bar = 500 μm. F, left oviger (female); scale bar = 250 μm. A, ZSMA20111009; B, ZSMA20111006; C, ZSMA20111000; D, ZSMA20111004; E, ZSMA20111002; F, ZSMA20111016. Abbreviations: ab, abdomen; cf, chelifore; cx, coxa; os, oviger segment; ov, oviger; pr, proboscis; tr, trunk; wl, walking leg.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 122 A. WEIS ET AL.

Figure 10. Scanning electron microscopy (SEM) of Pallenopsis patagonica (A, C, E) and Pallenopsis yepayekae sp. nov. (B, D, F). A, detail view of cement gland tube of left first walking leg; scale bar = 200 μm. B, detail view of cement gland tube of left second walking leg; scale bar = 100 μm. C, right oviger (female); scale bar = 1 mm. D, right oviger (female); scale bar = 200 μm [insert: detail view of distal oviger segments (female); scale bar = 100 μm]. E, detail view of hairs on third coxa of left fourth walking leg; scale bar = 100 μm. F, detail view of hairs on second coxa of left second walking leg; scale bar = 20 μm. A, ZSMA20111360; B, F, ZSMA20111006; C, ZSMA20111349; D, ZSMA20111009, insert: ZSMA20111024; E, ZSMA20111359.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 123 whereas the setae from the other specimens are the second and third coxae. Also, the structure of ‘smooth’ or rather normally developed (Fig. 10E). these hairs accords well with that described for the The length of the auxiliary claws varies between individuals of our ‘Chilean clade’. The abdomen shows one-third and one-half the length of the main claw, the same shape bearing two spines on the rounded without distinction between specimens from different edge of the beginning of the dorsodistal slope. One of areas. Only specimen ZSMA20111008 from the the spines on the dorsal side is broken and the other Chilean fjord region at 50°S bears extremely short is not as prominent as in most of the individuals from auxiliary claws, being one-quarter the length of the our ‘Chilean clade’, but nevertheless is clearly visible. main claw. Moreover, Hoek’s material also contains a specimen Pallenopsis patagonica specimens from the Swedish called P. patagonica var. elegans from station 320 Museum of Natural History, determined by Loman, near the La Plata estuary in Argentina (37°17′S, show similar morphological characteristics to those 53°52′W). As Hoek already mentioned, this individual of our specimens from the Falkland clade. Loman’s resembles a variety of P. patagonica, i.e. our Falkland specimens were collected by the Swedish South Polar clade, with only a more slender appearance. Expedition (1901–1903) at the Graham Region, South The results of our morphological analyses as well as Georgia, and the Falkland Islands. The undetermined our molecular data strongly indicate that the Chilean Pallenopsis sp. (SMNH-125514) was collected at the clade, i.e. the 11 specimens collected at the southern Patagonia Archipelago (Tierra del Fuego) 55°10′S, Chilean coast, represents a new species that is 66°15′W, and is in good accordance in morphology described below. with our Chilean clade. This specimen, an ovigerous male, shows the characteristic hairs on the ventral side of the second and third coxae, has a long cement PALLENOPSIS YEPAYEKAE SP. NOV.WEIS gland tube (more than three times its width), and a URN:LSID:ZOOBANK.ORG:ACT:0E39E226-30C7-4853- proboscis with a light swelling at half of its length. A6A1-7DD2336F33FE The specimen of Hedgpeth (SMNH-125527) appears FIGURES 9A–F, 10B, D, F, 11A–F to be a female and was collected by the Lund Univer- The new species can clearly be attributed to the genus sity Chile Expedition (1948–49) at Canal San Antonio Pallenopsis Wilson, 1881 by its slender segmented 41°47′S, 73°15′W. This is the exact region where body, cylindrical proboscis, rudimentary palps, ten- samples from our Chilean clade are from. Also this segmented ovigers in males, and slender legs with specimen shows the same morphological characteris- claws and auxiliary claws (Wilson, 1881). tics as our specimens from the Chilean fjords: a The species description of P. yepayekae sp. nov. is nine-segmented oviger (with the fourth oviger segment based altogether on 14 specimens: 11 specimens col- swollen), a proboscis with a slight swelling at the lected by the ‘Huinay Fjordos’ expeditions 2006–2011, middle, and prominent brush-like setae on the ventral one specimen (SMNH-125514) that was only deter- side of the second and third coxae. mined to genus level by Loman (1902), and two further specimens that were initially determined as P. patagonica, namely SMNH-125527 from Hedgpeth REINVESTIGATION OF HOEK’S TYPE MATERIAL (1949) and BMNH-1881.38 from Hoek (1881). Hoek’s type material consists of three female speci- mens: one bigger specimen, upon which his type Types determination is based, and two smaller specimens Holotype: Male (ZSMA20111002), Chile, Hanover that he designated as juveniles. The three individuals area, Canal Pitt Chico, 50°50′07.1″S, 74°08′20.9″W, were sampled from three different stations, namely 25 m, 07.03.2006, leg. R. Melzer, M. Schrödl. station 304, 308, and 313 (located at 46°53′S, 75°11′W, 50°10′S, 74°42′W, and 52°20′S, 68°0′W, respectively). Unfortunately it is not known which specimen was Paratypes captured from which sample site, as the specimen Four males: ZSMA20111000, Chile, Western labels don’t contain this information. Whereas the Katalalixar, Canal Castillo, 48°44′11.4″S, 75°24′ bigger specimen and one of the smaller ones are 53.1″W, 15 m, 12.03.2006, leg. R. Melzer, M. Schrödl; morphologically identical with the individuals of our ZSMA20111006, Chile, Fjords of region X, Inio 4, Falkland clade, the other one resembles completely 43°25′03.0″S, 74°04′51.2″W, 20 m, 24.02.2008, leg. the specimens from our Chilean clade. It shows dis- G. Försterra; ZSMA20111339, Chile Anihue Raul tinct prominent features: (1) a proboscis slightly Marin Balmaceda, Islas Tres Hermanas, 43°46′ swollen at the middle; (2) an eight- to nine-segmented 31.35″S, 73°01′44.14″ W, 19 m, 17.01.2011, leg. V. oviger, with the fourth oviger segment thickened; and Häussermann; SMNH-125514, South Atlantic Ocean, (3) several short brush-like setae at the ventral side of Argentina, Patagonia archipelago (Tierra del Fuego),

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 124 A. WEIS ET AL.

Figure 11. Drawings of Pallenopsis yepayekae sp. nov. A, dorsal view. B, lateral view of female and detailed view of abdomen. C, walking leg, with enlargement of setae of coxae 2 and 3. D, propodus with claw and auxiliary claws. E, female oviger. F, male oviger.

55°10′S, 66°15′W (st. no. 60 of Swedish South Polar Channel and Fjords, Paso del Abismo, 49°34′38.7′S, Expedition 1901–03), 100 m, 15.09.1902, leg. J. C. C. 74°26′49.3′W, 28 m, 10.03.2006, leg. R. Melzer, M. Loman. Schrödl; SMNH-125527, South Pacific Ocean, Chile, Seven females: ZSMA20111003, Chile, Fjords of Canal Chacao, Canal San Antonio, 41°47′40″S, region X, Inio 4, 43°25′03.0″S, 74°04′51.2″W, 25 m, 73°15′40″W (st. no. M109 of Lund University Chile 24.02.2008, leg. RF; ZSMA20111004, Chile, Fjords Expedition 1948–49), 36 m, 06.05.1949; BMNH- of region X, Inio 5, 43°24′34.5″S, 74°05′00.7″W, 9 m, 1881.38, either from station 304, 308, or 313 of 24.02.2008, leg. NR; ZSMA20111009, Chile, Fjords of the HMS Challenger expedition 1872–76 between region X, Inio 3, 43°23′33.4″S, 74°07′56.5″W, 26 m, 46°53′S, 75°11′W and 52°20′S, 68°0′W, 82–320 m, 24.02.2008, leg. V. Häussermann; ZSMA20111016, 31.12.1875–20.01.1876. Chile, Western Katalalixar, Canal Adalberto, 48°36′28.7′S, 74°53′55.7′W, 32 m, 12.03.2006, leg. R. Two juveniles: ZSMA20111005, Chile, Western Katala- Melzer, M. Schrödl; ZSMA20111024, Chile, Messier lixar, Canal Castillo, 48°44′11.4″S, 75°24′53.1″W,

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 125

23 m, 12.03.2006, leg. V. Häussermann; Chelifores with movable finger equipped with ZSMA20111012, Chile, Raul Marin, Las Hermanas, setose pad. Tips overlap when closed, inner edges join 22 m, 11.03.2007, leg. R. Meyer, K. Jörger. when closed. Lateral palp buds have the form of short Beside the specimens there are also DNA aliquots knobs (Fig. 11B). (including ten paratypes plus holotype) stored under Oviger ten-segmented, typical for genus (Fig. 11F). specific voucher IDs at ZSM (see also Table 1) and Distal segments more setose than proximal segments, CCDB. with setae pointing in various directions. Legs (Fig. 11C) with several setae not longer than Etymology the diameter of the segment upon which they are In Kawésar language, yepayek is the name of the situated. Coxae 1 and 3 subequal. Second coxa about ciprés de las güaitecas (Pilgerodendron uviferum). If twice the length of third coxa. Second and third coxae one looks at the fine ramification of the branches of a with many conspicuous short brush-like setae on the cypress-like tree, the similarity to the structure of the ventral side (Fig. 11C). Femur and tibia 1 of about setae of the ventral side of the second and third coxae equal size. Tibia 2 longest leg article. Tarsus short, of the new species described here becomes obvious. armed with one bigger spine on the ventral side. The name of the species also refers to the Yepayek,a Propodus (Fig. 11D) slightly curved, with three or ranger boat of the Corporación Nacional Forestal four heel spines. Sole with many shorter spines. Claw (CONAF) named after the tree, which carried the robust, slightly curved, auxiliary claws about one- scientists to the different places in the Chilean fjords third to one-half of main claw length. sampled during ‘Huinay fjordos’ expedition 3. It is Cement gland tube about three times as long as its the Yepayek and its always friendly and cooperative diameter, medioventrally on femur on slightly raised crew to whom we owe the chance to collect this new surface. Sexual pores on ventral side of second coxae species; therefore, we decided to name the species of third and fourth pair of legs. Pallenopsis yepayekae sp. nov. and also to keep in mind the adventurous trip through the labyrinth of Measurements (holotype, in mm): Length of trunk the Chilean fjords. (anterior margin of first trunk segment to distal margin of fourth lateral processes), 4.82; trunk width (across first lateral processes), 2.94; proboscis length, Diagnosis 2.29; abdomen length, 1.81; third leg, coxa 1, 0.85; Compared with P. patagonica, a rather small species coxa 2, 2.58; coxa 3, 1.23; femur, 5.90; tibia 1, 5.49; of smooth habitus, and in a few individuals the legs tibia 2, 7.06; tarsus, 0.27; propodus, 1.44; claw, 0.76; show red stripes. Proboscis (Fig. 11B) with distinct auxiliary claws, 0.50. Different leg segments were swelling at the middle. Abdomen (Fig. 11B) erect measured in natural posture. (about 45°) and dorsodistally sloped. The beginning of the slope shows a rounded edge on which two very Female: General habitus and size similar to male. prominent spines are sited (Fig. 11B). Second and Differences are only in the sexual characters: oviger third coxae with many conspicuous short brush-like (Fig. 11E) eight- to nine-segmented, with fourth setae on the ventral side (Fig. 11C). Oviger of the oviger segment swollen; distal oviger segments fused females eight- to nine-segmented with the fourth and less setose than in the male; all setae pointing oviger segment being swollen (Fig. 11E). Cement distally. Sexual pores on all second coxae on gland duct of males relatively long, measuring about ventrodistal surface. three times the length of its diameter. Distribution: Chilean fjord region 41°47′40″–55°10′S Description and 66°15′–75°24′53.1″W; depth range 9–100 m. Male: Size moderate to small, leg span less than As Hoek’s syntypes series of P. patagonica includes 60 mm. Trunk glabrous with distinct segment one specimen of P. yepayekae sp. nov., a lectotype borders, lateral processes separated by about one- for P. patagonica must be designated. Of the two third their diameter (Fig. 11A, B). Ocular tubercle at specimens from the BMNH-1881.38 material of the anterior portion of cephalic segment, slightly pointed HMS Challenger expedition, the larger specimen, (Fig. 11B). Eyes prominent, pigmented with posterior upon which Hoek’s description is based, shall eyes smaller than anterior eyes. Proboscis slightly be the lectotype, and the smaller specimen the directed downwards, swollen at middle (Fig. 11B). paralectotype. The lectotype of P. patagonica can Abdomen erect, somewhat extending beyond the clearly be distinguished from the new species distal margins of the lateral processes, dorsodistally P. yepayekae sp. nov. by the following characteristics: sloped, with two very prominent spines on the dorsal abdomen without two prominent spines on the dorsal side (Fig. 11B). side, ten-segmented oviger in females, second and

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 126 A. WEIS ET AL. third coxae without conspicuous short brush-like between the single clades, i.e. haplotypes of different setae on the ventral side, and a cylindrical proboscis subnetworks are present at the same location (see without a swelling at the middle. Fig. 3). The same pattern has been observed at several locations for the giant sea spider Colossendeis megalonyx Hoek, 1881 (Krabbe et al., 2010). To DISCUSSION confirm this finding, more sequences from specimens The results of our study indicate great morphological from South Georgia, Antarctica, and more northern as well as genetic variation in the individuals exam- areas of the Chilean coast are required. ined, indicating that P. patagonica sensu lato is a good Again, as in other pycnogonids, in P. patagonica we example for studying species complexes. observe very high interspecific distances compared To avoid circular reasoning by mixing morphology- with other taxa (Hebert et al., 2004; Lefebure et al., based considerations and molecular results, all 2006; Raupach et al., 2010). Either the number of molecular analyses were performed using the whole undescribed species in Pycnogonida is higher than data set, and checked against the morphological in other taxa, or there is a peculiar ‘pycnogonid’ results later. Correspondingly, the morphology of the phenomenon not understood at the moment. specimens was analysed without taking sequence- Furthermore, the tree-based GMYC modelling defined groupings into account. After the first mor- analyses, a recently developed species delimitation phological determinations all specimens studied could method (Pons et al., 2006; Monaghan et al., 2009) be assigned to P. patagonica according to the hitherto that has been used in several groups of organisms existing definitions (Gordon, 1932; Stock, 1957; (Barraclough et al., 2009; Bode et al., 2010; Esselstyn Pushkin, 1975, 1993; Child, 1995). We also decided to et al., 2012; Williams et al., 2012), reveal the presence include the available sequences of P. macneilli, of about 15 distinct GMYC species, of which only two P. buphtalmus, and P. latefrontalis in our studies, are represented by our two bigger clades (Falkland owing to their close relationship with P. patagonica. and Chilean clades). This suggests the presence of Furthermore, as we did not have these three speci- possibly unrecognized species; however, further sam- mens at hand to check whether the determinations pling is needed to test explicitly for this phenomenon. and the genetic data show their affinities to the P. patagonica complex, we treated them as neutrally MORPHOLOGICAL ANALYSIS as possible and considered them also as possible P. patagonica specimens. As for most of the clusters/clades only a few or even only one specimen is available at the moment, more specimens from these scattered clades are needed to MOLECULAR ANALYSIS unravel this complex phenomenon. However, there Regarding the molecular results presented in this are enough specimens in the Falkland and Chilean study, different clades are supported by high boot- clades for making conclusions regarding their species strap or posterior probability values. Regarding all status. As the original description of P. patagonica Pallenopsis specimens studied, two bigger clades can (Hoek, 1881) fits perfectly with the morphology of the be clearly distinguished: the Chilean clade with 11 16 specimens from the Falkland clade, they must specimens and the Falkland clade with 16 specimens. be the P. patagonica sensu stricto. In contrast, speci- This is not surprising, as our morphological data mens from the Chilean clade show several morpho- already grouped the Chilean and Falkland specimens logical and molecular differences, which leads us to separately (see Table S2). the description of a species new to science. Combining all evidence of our results, in particular Specimens described by Hoek have a cylindrical the extremely high intraspecific distances of 23%, proboscis without swelling at half of its length and a and also considering the high ‘intraspecific’ variation ten-segmented oviger in females. The bigger female of 10.4% for P. patagonica reported in our previous Hoek describes has a body length of about 16 mm, study (Weis & Melzer, 2012a), we conclude that which is similar to our specimens from the Falkland P. patagonica might represent a large species Islands, South Georgia and Antarctica. Hoek men- complex, potentially hiding several undescribed new tions some small and stout hairs at the swollen species. extremity of the second, third, and fourth joint of the In contrast to our previous study of Achelia leg (meaning coxa 2, coxa 3, and femur, respectively). assimilis (Haswell, 1885), where we assumed subspe- Perhaps this could be the setae that we describe in cies because of the geographic pattern (possible the specimens from the Falkland clade on the ventral allopatric speciation process), in P. patagonica we side of coxae two and three. However, these hairs are find another case. As seen in the network and not visible in his drawings (see Hoek, 1881: plate XII, the phylogenetic tree, there is geographic overlap figs 6–9), implying that they are not as prominent as,

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 127 for example, in our individuals studied from the To check that there are no other species hidden Chilean coast. Hoek’s specimens were captured by the behind the 39 specimens studied, we chose P. pilosa HMS Challenger at station 304 (46°53′S, 75°11′W), as an out-group, as well as P. buphtalmus, P. station 308 (50°10′S, 74°42′W), and station 313 latefrontalis, and P. macneilli, and examined and (52°20′S, 68°0′W). Fortunately two of our specimens, compared the descriptions of other Pallenopsis species namely ZSMA20111008 and ZSMA20111002, are from found in this area with our individuals. Child (1992) almost exactly the same location as HMS Challenger described two new Pallenopsis species from Chile, station 308. Regrettably Hoek did not mention which namely P. notiosa and P. truncatula. The latter one of the three specimens is from which sample location. has very short auxiliary claws (about 0.15 the length We assume that the only adult female, on which his of the main claw), well-separated lateral processes, a description and drawings are also based, has been glabrous abdomen, a very short cement gland tube in captured east of Chile in the Atlantic at station 313, males, and a ten-segmented oviger in females. None as this description matches much better with our of our individuals show all of these features in com- specimens from the Falkland Islands and surround- bination. For example ZSMA20111008 is the only ing area (see above). specimen bearing such short auxiliary claws, but in If one follows the first description given by Hoek contrast to P. truncatula it has a rounded ocular (1881) under the synonym Phoxichilidium patagoni- tubercle, a setose abdomen, and a femur being as cum, the specimens from the Falkland Islands and long as tibia 1 (femur is shorter than tibia 1 in Antarctica would match better than those from the P. truncatula). Also, P. notiosa can be excluded con- Chilean clade. Hoek focused his description on just the cerning our specimens, as it has a rounded ocular bigger individual, and denominated the smaller ones tubercle, well-separated lateral processes, and a as juveniles, without giving them any more attention. very long second coxa (about three times coxa 3; In our opinion these two specimens are adult females see Weis & Melzer, 2012b). Our specimens have a as well, as both are already carrying eggs inside the slightly conical or pointed ocular tubercle, only little- femur. After specific study, one of the smaller females separated lateral processes, and a second coxa is shown to resemble exactly P. yepayekae sp. nov. being about twice the length of the third coxa. Furthermore, one of Hoek’s sample locations (station ZSMA20111008, for example, has a rounded ocular 308) falls exactly in the area of the sample sites given tubercle, but the other characteristics do not match. for P. yepayekae sp. nov. Hence, we assume that this Furthermore, in neither of the two species Child individual of Hoek’s material derives from station 308. mentions are there prominent hairs on the ventral Unfortunately, we cannot deduce, either from Hoek’s side of the second and third coxae, which occur in our descriptions or from his material that we have at Chilean specimens. Pallenopsis macneilli, which is hand, which specimen was captured at which station. closest to ZSMA20111008 in the tree, does not fit with The bigger specimen and the one that resembles our material because of its horizontal abdomen, rela- P. yepayekae sp. nov. are both kept in the same tube tively long auxiliary claws, and its distribution area, labelled with station 313, which is obviously wrong as which is located in Australia. according to Hoek’s original data these samples come Two other interesting possible species could be from two different locations. Also, the sample site of Pallenopsis tumidula Loman, 1923 and Pallenopsis the third specimen is not well documented. candidoi Mello-Leitao, 1949, as both seem to exhibit Later, Möbius (1902), Hodgson (1907), Hodgson the short hairs on the ventral side of the second (1915), Bouvier (1913), Calman (1915), Loman and third coxa. However, the latter has an eight- (1923a), Gordon (1932), Marcus (1940), Hedgpeth segmented oviger in females, and auxiliary claws (1961), Pushkin (1975, 1993), Stock (1957), and Child clearly longer than half the length of the main claw, (1994) also described several further specimens and which differs from our specimens. Furthermore, synonyms of P. patagonica. The specimens were P. candidoi is only sampled from South Georgia mainly captured from the Southern Ocean, including to South Brazil so far. Pallenopsis tumidula is Bouvet and South Georgia, or from the Falkland characterized and drawn by Stock (1957) with Islands and the Atlantic coast of South America. ‘Fiederdornen’ on the ventral distal side of coxae 2 With every newly added description the species and 3. He mentions that this feature makes P. P. patagonica, with its various existing synonyms, tumidula clearly distinguishable from P. patagonica. became more and more diverse and variable. The Confusingly, if one regards the original description of morphological frame under which one could assign a 1923, Loman neither mentions short hairs on the pycnogonid to this species became broader and more coxae nor shows them in his drawings. Furthermore, ambiguous. Hence, it is not surprising that in a the type material we had at hand from the Swedish broader sense all of our studied specimens match Museum of Natural History didn’t show any promi- with the characterization of P. patagonica. nent hairs on the coxae. Only our specimens from the

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 128 A. WEIS ET AL.

Chilean clade show this kind of ‘Fiederdornen’, and in describes the propodus as more heavy and robust, it is contrast to P. tumidula they have eight- or nine- considered long by Child (1995). segmented ovigers in females, whereas Loman men- The length of the auxiliary claw is given as either tions a ‘ten-segmented’ oviger in his individuals. One one-third the length of the main claw (Stock, 1957), drawing by Loman of a young female shows the last half the length of the main claw (Möbius, 1902; oviger segments to be fused, which could be more Hodgson, 1907; Calman, 1915; Gordon, 1944), or even consistent with our specimens. But this would mean longer (Pushkin, 1975; Pushkin, 1993). Except for one that all our specimens from the Chilean clade would specimen (ZSMA20111008), our studied specimens be just juveniles, which can be excluded for example have auxiliary claws reaching one-third to one-half by the visible eggs inside the femur of females, indi- the length of the main claw. cating an adult state. Furthermore, Loman does not Whereas Stock (1957) remarks that P. patagonica mention any setae on the abdomen. Besides several lacks ‘Fiederdornen’ (stellate setae) on the second and short setae, our specimens also show two very promi- third coxae of the legs, some kind of short hairs are nent larger spines on the distal end of the abdomen. mentioned in Pushkin (1975): ‘. . . The few very small Another fact that should be kept in mind is that spines are located along the ventral surface of the P. tumidula has only been captured from North 2nd and 3rd segments. Similar spines surround the Argentina so far. All this leads us to the decision that genital pore and form a small cluster on the ventral our specimens can not be P. tumidula. dilatation of the distal part of the third segment.’ Concerning our specimens from the Falkland clade, Here, specimens from the Chilean clade are distin- on the first view one possible candidate could be guishable from specimens from the Antarctic region Pallenopsis kupei Clark, 1971; however, the auxiliary or Falkland Islands by their ‘Fiederdornen’. claws, being more than half as long as the main claw Another very variable characteristic affects the (Clark, 1971), and the Macquarie and New Zealand cement gland of the males. Whereas the cement gland Plateau distribution of this species (Child, 1995), tube itself, when present, is always very short, the separate it from P. patagonica. ventral pore can be on a flat surface, on a broad raised Furthermore, analysing Loman’s P. patagonica surface, or something in between (Child, 1995). Our collection and one P. patagonica specimen of specimens show a mixture of everything: sometimes Hedgpeth from the Swedish Museum of Natural the cement gland tube is hardly visible (PpaE_001– History confirms our considerations. Eight specimens 002, PpaA_001), is short (specimens from the Falk- (SMNH-125445, SMNH-125507, SMNH-125508, land Islands), or is three times its own width (which SMNH-125509, SMNH-125510) captured from the is the case for the Chilean clade). Concerning the Graham region, South Georgia and Falkland orientation of setae of the ovigers, we could detect Islands, determined as P. patagonica by Loman, are the same sexual dimorphism as mentioned in perfectly in accordance with the morphology of Bamber (2002). There are no differences between our specimens from the Falkland clade. In contrast, P. yepayekae sp. nov. and P. patagonica. the specimen SMNH-125527, determined as P. Moreover, the abdomen of P. patagonica can be patagonica by Hedgpeth, and collected at 41°47′S, long and erect or be shorter and horizontal (Child, 73°15′W, fits better with the description of the 1995). The only specimen with a straight horizontal specimens of our Chilean clade. This would mean abdomen is PpaE_001 from the Shag Rocks, near that this specimen is not a P. patagonica, but a South Georgia. All other individuals have an upward- P. yepayekae sp. nov. Furthermore, the only specimen erected abdomen. As the morphological differences undetermined by Loman (SMNH-125514), which was among the corresponding specimens lie well within collected at Tierra del Fuego (55°10′S, 66°15′W), the broad variation described in the literature, shows the same characteristics as P. yepayekae we assigned all of our studied specimens (except sp. nov., here described as a new species. This also those assigned to P. yepayekae sp. nov.) tentatively to explains why Loman determined this specimen only P. patagonica; however, in parallel with our molecular to genus level. He seemed to see the differences to results this pronounced morphological variability P. patagonica. in many features indicates that P. patagonica is a For P. patagonica, however, a broad variability con- species complex. cerning different characteristics is discussed. Gordon (1932) notices that the gap between the lateral pro- CONCLUSION cesses ranges from being little separated to separated by about their own diameter. Furthermore, the To summarize, we could not assign our specimens spination of the propodus varies greatly in numbers (except P. yepayekae sp. nov. described in the present and length, bearing two, three, or four spines, for paper) to any of the described/known Pallenopsis example (Gordon, 1944). Whereas Stock (1975) species other than P. patagonica occurring near the

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 129 studied area with sufficient certainty. It seems neces- for processing, coordination, and providing part of sary to include, beyond the morphological description, the sequence data. We thank Dr Claudia Arango for another level/source of information, i.e. a data set access to unreleased GenBank sequences for out- independent of morphology, as is presented here. With group species. Special thanks go to Miranda Lowe our molecular data, this is the first attempt/step to (Natural History Museum London) for making Hoek’s unravel the species complex of P. patagonica with type material available. We are also grateful to the a wider set of techniques. But the molecular data following colleagues who kindly provided pycnogonid also confirm the variability of the species, resulting samples: Günther Försterra and Dr Verena in different clades supported by high bootstrap Häussermann (Huinay Scientific Field Station, values. Chile), Dr Javier Sellanes López (Universidad As already discussed in our previous study (Weis & Católica del Norte, Facultad de Ciencias del Mar, Melzer, 2012a), with a focus on Achelia assimilis, the Coquimbo, Chile), Dr Vladimir Laptikhovsky (Falk- distribution area of P. yepayekae sp. nov. corresponds land Islands Fisheries Department), and Michael well with the area covered by glaciers during the Schrödl (ZSM). We thank Stefan Friedrich (ZSM) for last ice age. However, the Pallenopsis habitat extends expert technical assistance and collection manage- to much deeper waters (down to 3500 m) than for ment. Special thanks also go to the Yepayek crew: Achelia (about 900 m) (Child, 1994). Therefore, the German Coronado, Don Victor Munoz Aguero, and present-day distribution was either achieved by Guillermo Igor Almonacid. In addition, we thank two recolonization from deeper waters or by leading-edge anonymous referees for their helpful comments and recolonization from more northern, ice-free habitats. suggestions. This study was supported by a graduate The diversity of different haplotypes does not imply student stipend (Graduiertenstipendium; BayEFG) to that there was a strong bottleneck; however, further A. Weis, travel funds from GeoBioCenterLMU to specimens are needed to verify this assumption. R. Melzer, and a grant by Sealife Center Munich. The extremely high genetic distances between the F. Leese was supported by DFG grant LE 2323/2 Falkland ‘patagonica’ clade and the Chilean within the DFG priority programme 1158. This article ‘yepayekae’ clade indicate that these do not resemble is no. 79 from the Huinay Scientific Field Station. populations that are geographically isolated. Over a long geographic gradient, genetic distances within P. yepayekae sp. nov. were low. Therefore, an REFERENCES allopatric speciation, possibly influenced by the Arango CP, Brenneis G. 2013. New species of Australian massive glaciations, may be a likely explanation for Pseudopallene (Pycnogonida: Callipallenidae) based on live the speciation. colouration, morphology and DNA. Zootaxa 3616: 401–436. The morphological and molecular results strongly Arango CP, Soler-Membrives A, Miller KJ. 2011. Genetic support the hypothesis that the specimens from the differentiation in the circum-Antarctic sea spider Nymphon Chilean clade represent a species new to science, australe (Pycnogonida: Nymphonidae). Deep-Sea Research described here as P. yepayekae sp. nov. The decision to Part II-Topical Studies in Oceanography 58: 212–219. erect P. yepayekae sp. nov. as a new species is also Bamber RN. 2002. Bathypelagic pycnogonids (Arthropoda, supported by the eleven individuals that do not differ Pycnogonida) from the Discovery deep-sea cruises. Journal strongly, both genetically and morphologically. It is of Natural History 36: 715–727. known from previous works (for example, Hebert Bamber RN, El Nagar A. 2011. Pycnobase. World et al., 2004) that in less extensively studied inver- Pycnogonida Database. Available at: http://marinespecies tebrate taxa (such as pycnogonids) hidden biological .org/pycnobase diversity, in the form of cryptic or overlooked species, Barraclough TG, Hughes M, Ashford-Hodges N, is often the rule rather than the exception. Investi- Fujisawa T. 2009. Inferring evolutionary significant units of bacterial diversity from broad environmental surveys of gating how many further species may be hidden single-locus data. Biology Letters 5: 425–428. behind the Pallenopsis complex remains beyond the Bode SNS, Adolfsson S, Lamatsch DK, Martins MJF, scope of this paper. This will be an interesting ques- Schmit O, Vandekerkhove J, Mezquita F, Namiotko T, tion for further analyses with hopefully more speci- Rossetti G, Schön I, Butlin RK, Martens K. 2010. mens available from the Southern Ocean. Exceptional cryptic diversity and multiple origins of parthe- nogenesis in a freshwater ostracod. Molecular Phylogenetics ACKNOWLEDGEMENTS and Evolution 54: 542–552. Bouvier EL. 1913. Pycnogonides du Pourquoi Pas? Deuxième First we must thank the Canadian Centre for Expédition Antarctique Française (1908–1910), 6, 1–169. Barcoding at the Department of Zoology, University Calman WT. 1915. Pycnogonida. British Antarctic (‘Terra of Guelph, Ontario, Canada (financed by Genome Nova’) expedition, 1910. Natural History Report. Zoology 3: Canada through the Ontario Genomics Institute), 1–74.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 130 A. WEIS ET AL.

Child CA. 1992. Pycnogonida of the Southeast Pacific reveals cryptic species in the neotropical skipper butterfly Biological Oceanographic Project (SEPBOB). Smithsonian Astraptes fulgerator. Proceedings of the National Academy Contributions to Zoology I–IV: 1–43. of Sciences of the United States of America 101: 14812– Child CA. 1994. Antarctic and Subantarctic Pycnogonida: 14817. I, Ammotheidae and II, Austrodecidae. Biology of the Antarc- Hedgpeth JW. 1961. Reports of the Lund University Chile tic Seas XXIII. In: Cairns SD, ed. Antarctic research series 63. Expedition 1948–49, 40: Pycnogonida. Lunds Universitets Washington, DC: American Geophysical Union, 1–99. Årsskrift. Ny Foljd. Avdelningen 2, 57(3). Lund: C.W.K. Child CA. 1995. Antarctic and Subantarctic Pycnogonida. III. Gleerup, 1–18. The family Nymphonidae. Antarctic and Subantarctic Hodgson TV. 1907. Pycnogonida. National Antarctic Expedi- Pycnogonida: Nymphonidae, Colossendeidae, Rhynchothora- tion 1901–1904. Reports of the National Antarctic Expedi- xidae, Pycnogonidae, Endeidae, and Callipallenidae. Biology tion of 1901–1904. Natural History 3: 1–72. of the Antarctic Seas XXIV. Antarctic research series volume Hodgson TV. 1915. The Pycnogonida collected by the Gauss 69. Washington, DC: American Geophysical Union, 1–165. in the Antarctic regions, 1901–03; preliminary report. Clark WC. 1971. Pycnogonida of the Antipodes islands. New Annals and Magazine of Natural History, (ser.8) 15: 141– Zealand Journal of Marine and Freshwater Research 5: 149. 427–452. Hoek PPC. 1881. Report on the Pycnogonida dredged by Darriba D, Taboada GL, Doallo R, Posada D. 2012. HMS Challenger 1873–76. Reports of the Scientific Results jModelTest 2: more models, new heuristics and parallel of the Exploring Voyage of HMS Challenger 3: 1–167. computing. Nature Methods 9: 772. Huson DH, Bryant D. 2006. Application of phylogenetic Dietz L, Krapp F, Hendrickx ME, Arango CP, Krabbe K, networks in evolutionary studies. Molecular Biology and Spaak JM, Leese F. 2013. Evidence from morphological Evolution 23: 254–267. and genetic data confirms that Colossendeis tenera Hilton, Kimura M. 1980. A simple method for estimating evolution- 1943 (Arthropoda: Pycnogonida), does not belong to the ary rates of base substitutions through comparative studies Colossendeis megalonyx Hoek, 1881 complex. Organisms, of nucleotide sequences. Journal of Molecular Evolution 16: Diversity & Evolution 13: 151–162. 111–120. Dietz L, Mayer C, Arango CP, Leese F. 2011. The Krabbe K, Leese F, Mayer C, Tollrian R, Held C. mitochondrial genome of Colossendeis megalonyx supports 2010. Cryptic mitochondrial lineages in the widespread a basal position of Colossendeidae within the Pycnogonida. pycnogonid Colossendeis megalonyx Hoek, 1881 from Ant- Molecular Phylogenetics and Evolution 58: 553–558. arctic and Subantarctic waters. Polar Biology 33: 281–292. Drummond AJ, Ashton B, Buxton S, Cheung M, Lefebure T, Douady CJ, Gouy M, Gibert J. 2006. Cooper A, Duran C, Field M, Heled J, Kearse M, Relationship between morphological taxonomy and molecu- Markowitz S, Moir R, Stones-Havas S, Sturrock S, lar divergence within Crustacea: proposal of a molecular Thierer T, Wilson A. 2011. Geneious v5.4. Available at: threshold to help species delimitation. Molecular Phylo- http://www.geneious.com/ genetics and Evolution 40: 435–447. Drummond AJ, Rambaut A. 2007. BEAST: Bayesian evo- Loman JCC. 1923a. Subantarctic Pycnogonida from the lutionary analysis by sampling trees. BioMed Central Evo- Stockholm Museum. Arkiv Zoologi 15: 1–13. lutionary Biology 7: 214. Loman JCC. 1923b. The Pycnogonida. Further Zoological Esselstyn JA, Evans BJ, Sedlock JL, Khan FAA, Heaney Results of the Swedish Antarctic Expedition 1: 1–41. LR. 2012. Single-locus species delimitation: a test of the Mahon AR, Arango CP, Halanych KM. 2008. Genetic mixed Yule-coalescent model, with an empirical application diversity of Nymphon (Arthropoda: Pycnogonida: Nympho- to Philippine round-leaf-bats. Proceedings of the Biological nidae) along the Antarctic Peninsula with a focus on Society B 279: 3678–3686. Nymphon australe Hodgson 1902. Marine Biology 155: Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R. 1994. 315–323. DNA primers for amplification of mitochondrial cytochrome Marcus E. 1940. Os Pantopoda brasileiros e os demais sul- c oxidase subunit I from diverse metazoan invertebrates. americanos. Boletin da Faculdade de Filosofia, Ciencias e Molecular Marine Biology and Biotechnology 3: 294–299. Letras da Universidade de Sao Paulo, Series 4, Zoology 19: Försterra G. 2009. Ecological and Biogeographical aspects of 3–179. the Chilean fjord region. In: Häussermann V, Försterra G, Möbius K. 1902. Die Pantopoden der Deutschen Tiefsee- eds. Marine Benthic Fauna of Chilean Patagonia: illustrated Expedition, 1898–99. Wissenschaftliche Ergebnisse deutscher identification guide. Puerto Montt, Chile: Nature in Focus, Tiefsee-Expedition. Dampfer ‘Valdivia’, 1898–1899 3: 177– 61–76. 196. Gordon I. 1932. Pycnogonida. In: Discovery reports vol. VI. Monaghan MT, Wild R, Elliot M, Fujisawa T, Balke M, Fetter Lane, London: Cambridge University Press, 1–138. Inward DJ, Lees DC, Ranaivosolo R, Eggleton P, Gordon I. 1944. Pycnogonida. Reports of the British, Austral- Barraclough TG, Vogler AP. 2009. Accelerated species ian and New Zealand Antarctic Research Expedition 5: inventory on Madagascar using coalescent-based models of 1–72. species delineation. Systematic Biology 58: 298–311. Hebert PDN, Penton EH, Burns JM, Janzen DH, Müller HG. 1993. World catalogue and bibliography of the Hallwachs W. 2004. Ten species in one: DNA barcoding recent Pycnogonida. Wetzlar, Germany: Wissenschaftlicher

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 PALLENOPSIS PATAGONICA – A SPECIES COMPLEX 131

Verlag, Laboratory for Tropical Ecosystems Research & Stock JH. 1957. Pantopoden aus dem Zoologischen Information Service, 1–388. Museum Hamburg. Mitteilungen aus dem Hamburgischen Munilla T, Soler Membrives A. 2009. Check-list of the Zoologischen Museum und Institut 55: 81–106. pycnogonids from Antarctic and sub-Antarctic waters: zoo- Stock JH. 1975. Pycnogonida from the continental shelf, geographic implications. Antarctic Science 21: 99–111. slope, and deep sea of the tropical Atlantic and East Pacific. Nielsen JF, Lavery S, Lörz AN. 2009. Synopsis of a new Biological results of the University of Miami deep-sea collection of sea spiders (Arthropoda: Pycnogonida) from the expedition, 108. Bulletin of Marine Science 24: 957–1092. Ross Sea, Antarctica. Polar Biology 32: 1147–1155. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Pons J, Barraclough TG, Gomez-Zurita J, Cardoso A, Kumar S. 2011. MEGA5: molecular Evolutionary Genetics Duran DP, Hazell S, Kamoun S, Sumlin WD, Vogler Analysis Using Maximum Likelihood, Evolutionary Dis- AP. 2006. Sequence-based species delimitation for the DNA tance, and Maximum Parsimony Methods. Molecular Biology taxonomy of undescribed insects. Systematic Biology 55: and Evolution 28: 2731–2739. 595–609. Weis A, Melzer R. 2012a. How did sea spiders recolonize Puillandre N, Lambert A, Brouillet S, Achaz G. the Chilean fjords after glaciation? DNA barcoding of 2012. ABGD, Automatic Barcode Gap Discovery for Pycnogonida, with remarks on phylogeography of Achelia primary species delimitation. Molecular Ecology 21: 1864– assimilis (Haswell, 1885). Systematics and Biodiversity 10: 1877. 1–14. Pushkin AF. 1975. Revision of the pycnogonids (Pantopoda) Weis A, Melzer RR. 2012b. Chilean and Subantarctic of the Pallenopsis patagonica group from Antarctica and Pycnogonida collected by the ‘Huinay Fjordos’ Expeditions adjacent waters. Byulleten Sovietskoi Antarkticheskoi 2005–2011. Zoosystematics and Evoltuion 88: 185–203. Ekspeditsii 90: 72–83. Williams PH, Brown MJF, Carolan JC, An J, Goulson D, Pushkin AF. 1993. The pycnogonids fauna of the South Aytekin AM, Best LR, Byvaltsev AM, Cederberg B, Ocean: biological results of the Soviet Antarctic expeditions. Dawson R, Huang J, Ito M, Monfared A, Raina RH, Russian Academy of Sciences. Exploration of the Fauna of Schmid-Hempel P, Sheffield CS, Sima P, Xie Z. 2012. the Seas. St. Petersburg, Portorosa, Sizilien, 1–397. Unveiling cryptic species of the bumblebee subgenus Ratnasingham S, Hebert PDN. 2007. BOLD: the Barcode Bombus s. str. worldwide with COI barcodes (Hymenoptera: of Life Data System (www.barcodinglife.org). Molecular Apidae). Systematics and Biodiversity 10: 21–56. Ecology Notes 7: 355–364. Wilson EB. 1881. Report on the Pycnogonida. Reports on the Raupach MJ, Astrin JJ, Hannig K, Peters MK, Stoeckle results of dredging, under the supervision of Alexander MY, Wägele JW. 2010. Molecular species identification of Agassiz, along the East Coast of the United States, during Central European ground beetles (Coleoptera: Carabidae) the summer of 1880, by the U.S. Coast Survey Steamer using nuclear rDNA expansion segments and DNA bar- ‘Blake’. Bulletin of the Museum of Comparative Zoology, codes. Frontiers in Zoology 7: 26. Harvard 8: 239–256. Ronquist F, Teslenko M, van der Mark P, Ayres DL, Xia XH, Lemey P. 2009. Assessing substitution saturation Darling A, Höhna S, Larget B, Liu L, Suchard MA, with DAMBE. In: Lemey P, Salemi M, Vandamme A, eds. Huelsenbeck JP. 2012. MrBayes 3.2: efficient Bayesian The phylogenetic handbook: a practical approach to DNA Phylogenetic Inference and Model Choice Across a Large and protein phylogeny. Cambridge: Cambridge University Model Space. Systematic Biology 61: 539–542. Press, 615–630. Saitou N, Nei M. 1987. The neighbor-joining method–anew Xia XH, Xie Z, Salemi M, Chen L, Wang Y. 2003. An index method for reconstructing phylogenetic trees. Molecular of substitution saturation and its application. Molecular Biology and Evolution 4: 406–425. Phylogenetics and Evolution 26: 1–7.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Table S1. Range of uncorrected pairwise distances (min-max) between specimens and clades (below diagonal) and the respective standard error (above diagonal). Table S2. Morphological characteristics of specimens that were available for morphological studies.

© 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 170, 110–131 Supporting information S1: Genetic distances between CO1 sequences PxxE001 PxxE002 CEA082 CEA112 CEA047 DQ390086 HM426218 HM426171 FJ969367 FJ969368 FJ969369 PxxE001 0.000 0.046 0.056 0.000 0.046 0.062 0.060 0.057 0.057 0.059 PxxE002 0.000 0.046 0.056 0.000 0.046 0.062 0.060 0.057 0.057 0.059 CEA082 0.166 0.166 0.045 0.046 0.048 0.054 0.055 0.052 0.052 0.054 CEA112 0.209 0.209 0.151 0.056 0.053 0.042 0.052 0.047 0.047 0.045 CEA047 0.000 0.000 0.166 0.209 0.046 0.062 0.060 0.057 0.057 0.059 DQ390086 0.163 0.163 0.177 0.195 0.163 0.066 0.058 0.065 0.065 0.061 HM426218 0.253 0.253 0.207 0.135 0.253 0.271 0.038 0.023 0.023 0.030 HM426171 0.252 0.252 0.220 0.191 0.252 0.236 0.110 0.040 0.040 0.024 FJ969367 0.220 0.220 0.191 0.162 0.220 0.269 0.047 0.123 0.000 0.036 FJ969368 0.220 0.220 0.191 0.162 0.220 0.269 0.047 0.123 0.000 0.036 FJ969369 0.236 0.236 0.206 0.149 0.236 0.252 0.071 0.047 0.096 0.096 ZSMA20111340 0.268 0.268 0.237 0.177 0.268 0.286 0.096 0.137 0.109 0.109 0.136 ZSMA20111017 0.253 0.253 0.255 0.163 0.253 0.271 0.096 0.136 0.109 0.109 0.135 PpaE003 0.191 0.191 0.192 0.177 0.191 0.177 0.136 0.163 0.122 0.122 0.177 ZSMA20111352 0.191 0.191 0.192 0.177 0.191 0.177 0.136 0.163 0.122 0.122 0.177 ZSMA20111072 0.252 0.252 0.208 0.191 0.252 0.206 0.164 0.163 0.193 0.193 0.164 ZSMA20111008 0.206 0.206 0.179 0.191 0.206 0.149 0.164 0.163 0.193 0.193 0.177 PpaE002 0.222 0.222 0.255 0.135 0.222 0.236 0.123 0.135 0.164 0.164 0.122 PpaA001 0.321 0.321 0.285 0.220 0.321 0.284 0.167 0.097 0.196 0.196 0.085 PpaE001 0.253 0.253 0.206 0.191 0.253 0.236 0.084 0.035 0.096 0.096 0.035 Chile clade 0.222-0.253 0.222-0.253 0.192-0.221 0.149-0.191 0.222-0.253 0.221-0.252 0.206-0.236 0.177-0.206 0.191-0.220 0.191-0.220 0.177-0.206 Falkland clade 0.236-0.252 0.236-0.252 0.254-0.271 0.162-0.177 0.236-0.252 0.236-0.253 0.083-0.096 0.109-0.123 0.096-0.109 0.096-0.109 0.109-0.122

Chile clade = ZSMA20111000, ZSMA20111002-006, ZSMA20111009, ZSMA20111012, ZSMA20111016, ZSMA20111024 and ZSMA20111339.

Falkland clade = PpaE004-008, PpaE010, ZSMA20111348-51, ZSMA20111354-55, ZSMA20111357, ZSMA20111359-61 ZSMA20111340 ZSMA20111017 PpaE003 ZSMA20111352 ZSMA20111072 ZSMA20111008 PpaE002 PpaA001 PpaE001 Chile clade Falkland clade 0.064 0.062 0.053 0.053 0.061 0.053 0.058 0.073 0.061 0.056-0.061 0.059-0.061 0.064 0.062 0.053 0.053 0.061 0.053 0.058 0.073 0.061 0.056-0.061 0.059-0.061 0.060 0.064 0.052 0.052 0.054 0.049 0.064 0.066 0.054 0.050-0.056 0.062-0.066 0.048 0.046 0.048 0.048 0.051 0.052 0.042 0.056 0.052 0.043-0.051 0.046-0.049 0.064 0.062 0.053 0.053 0.061 0.053 0.058 0.073 0.061 0.056-0.061 0.059-0.061 0.069 0.066 0.048 0.048 0.054 0.043 0.057 0.067 0.059 0.056-0.061 0.058-0.061 0.036 0.036 0.043 0.043 0.049 0.049 0.040 0.050 0.033 0.054-0.059 0.033-0.035 0.044 0.042 0.046 0.046 0.046 0.046 0.041 0.036 0.019 0.050-0.055 0.037-0.040 0.038 0.038 0.040 0.040 0.053 0.054 0.047 0.055 0.035 0.051-0.056 0.034-0.037 0.038 0.038 0.040 0.040 0.053 0.054 0.047 0.055 0.035 0.051-0.056 0.034-0.037 0.042 0.042 0.049 0.049 0.048 0.050 0.039 0.033 0.021 0.049-0.054 0.037-0.039 0.016 0.040 0.040 0.046 0.053 0.041 0.053 0.046 0.045-0.051 0.021-0.024 0.023 0.040 0.040 0.045 0.053 0.041 0.052 0.045 0.048-0.053 0.020-0.023 0.123 0.122 0.000 0.040 0.030 0.039 0.057 0.044 0.044-0.053 0.032-0.036 0.123 0.122 0.000 0.040 0.030 0.039 0.057 0.044 0.044-0.053 0.032-0.036 0.151 0.150 0.111 0.111 0.033 0.045 0.057 0.042 0.056-0.064 0.050-0.053 0.181 0.179 0.071 0.071 0.084 0.037 0.057 0.044 0.048-0.056 0.043-0.047 0.123 0.123 0.122 0.122 0.150 0.109 0.041 0.041 0.047-0.055 0.032-0.035 0.193 0.192 0.222 0.222 0.209 0.222 0.135 0.037 0.055-0.062 0.047-0.049 0.150 0.149 0.149 0.149 0.136 0.149 0.135 0.098 0.049-0.055 0.040-0.042 0.149-0.177 0.163-0.191 0.149-0.191 0.149-0.191 0.207-0.252 0.163-0.206 0.162-0.206 0.221-0.267 0.177-0.206 0.0-0.035 0.045-0.053 0.035-0.047 0.035-0.047 0.083-0.096 0.083-0.096 0.166-0.181 0.137-0.151 0.083-0.096 0.163-0.178 0.122-0.136 0.149-0.191 0.0-0.012 Electronic supplement 2: morphological characters of specimens Chile clade (11 specimens) Falkland clade (16 specimens) ZSMA20111072 ZSMA20111008 PpaE003 ZSMA20111352 PpaA001 PpaE001 PpaE002 ZSMA20111017 ZSMA20111340

A: shape of Proboscis 2 0 1 2 2 2 1 0 0 0 0

B: trunk length 0 1 0 0 0 0 0 1 1 0 0

C: orientation of abdomen 0 0 0 0 0 0 0 1 0 0 0

D: spines on abdomen 2 1 0 1 1 0 1 1 1 1 1

E: auxiliary claw length 1 1 1 0 1 1 1 1 1 1 1

F: shape of ocular tubercle 1 1 1 0 2 2 1 1 2 2 2

G: ventral hairs on coxa 2 and coxa 3 2 1 1 0 0 0 0 0 1 1 0

H: shape of 4th oviger segment in females 0 1 1 - - 1 juvenile - - juvenile juvenile

I: number of oviger segments in females 1 0 0 - - 0 - - - - -

J: length of cement gland tube in males 2 1 - 1 2 - - 0 0 - -

K: distance between lateral processes 0 0 1 1 0 0 0 0 1 1 0

A: 0 = oval, 1 = slightly swollen at middle, 2 = swollen at middle B: 0 = ≤ 7,5 mm, 1 = ≥ 8,5 mm C: 0 = erected, 1 = horizontal D: 0 = glabrous, 1 = two rows of lateral spines, two outermost spines not conspicuously larger, 2 = two outermost spines very prominent (about three times larger) E: 0 = ≤ 1/3 main claw lengths, 1 = 1/3-1/2 main claw lengths F: 0 = rounded, 1 = slightly pointed, 2 = pointed G: 0 = almost glabrous, 1 = few hairs, 2 = many prominent hairs H: 0 = swollen, 1 = not swollen/straight I: 0 = 10 oviger segments, 1 = < 10 oviger segments J: 0 = hardly/not visible, 1 = ≤ 2 times its diameter, 2 = > 2 times its diameter K: 0 = separated by less their diameter, 1 = separated ≥ their diameter Dissertation Lars Dietz

7) Publikation VI

Titel: Exploring Pandora’s box: Potential and pitfalls of low coverage genome surveys for evolutionary biology

PLos ONE 7: e49202

Hinweise zu Publikation VI

• Anteil Planung: 10%

• Anteil experimentelle Durchführung: 15%

• Verfassen des Manuskripts: 5%

Abbildungen, die nicht ausschließlich von mir erstellt wurden

• Alle Abbildungen wurden von Andrey Rozenberg erstellt.

Diejenigen Teile des elektronischen Zusatzmaterials, zu denen meine eigene Arbeit nicht beigetragen hat, sind hier nicht beigefügt. Sie sind online unter http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0049202#s6 einsehbar.

177 Exploring Pandora’s Box: Potential and Pitfalls of Low Coverage Genome Surveys for Evolutionary Biology

Florian Leese1,2*, Philipp Brand1., Andrey Rozenberg1., Christoph Mayer4, Shobhit Agrawal3, Johannes Dambach4, Lars Dietz1, Jana S. Doemel1, William P. Goodall-Copstake2, Christoph Held3, Jennifer A. Jackson2, Kathrin P. Lampert1, Katrin Linse2, Jan N. Macher1, Jennifer Nolzen1, Michael J. Raupach5, Nicole T. Rivera6, Christoph D. Schubart6, Sebastian Striewski1, Ralph Tollrian1, Chester J. Sands2 1 Ruhr University Bochum, Department of Animal Ecology, Evolution and Biodiversity, Bochum, Germany, 2 British Antarctic Survey, High Cross, Madingley Road, Cambridge, United Kingdom, 3 Alfred Wegener Institute for Polar and Marine Research, Functional Ecology, Bremerhaven, Germany, 4 Zoologisches Forschungsmuseum Alexander Koenig, Bonn, Germany, 5 Senckenberg am Meer, German Center for Marine Biodiversity Research, Molecular Taxonomy Group, Wilhelmshaven, Germany, 6 University of Regensburg, Biologie 1, Department of Evolution, Behavior and Genetics, Regensburg, Germany

Abstract High throughput sequencing technologies are revolutionizing genetic research. With this ‘‘rise of the machines’’, genomic sequences can be obtained even for unknown genomes within a short time and for reasonable costs. This has enabled evolutionary biologists studying genetically unexplored species to identify molecular markers or genomic regions of interest (e.g. micro- and minisatellites, mitochondrial and nuclear genes) by sequencing only a fraction of the genome. However, when using such datasets from non-model species, it is possible that DNA from non-target contaminant species such as bacteria, viruses, fungi, or other eukaryotic organisms may complicate the interpretation of the results. In this study we analysed 14 genomic pyrosequencing libraries of aquatic non-model taxa from four major evolutionary lineages. We quantified the amount of suitable micro- and minisatellites, mitochondrial genomes, known nuclear genes and transposable elements and searched for contamination from various sources using bioinformatic approaches. Our results show that in all sequence libraries with estimated coverage of about 0.02–25%, many appropriate micro- and minisatellites, mitochondrial gene sequences and nuclear genes from different KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways could be identified and characterized. These can serve as markers for phylogenetic and population genetic analyses. A central finding of our study is that several genomic libraries suffered from different biases owing to non-target DNA or mobile elements. In particular, viruses, bacteria or eukaryote endosymbionts contributed significantly (up to 10%) to some of the libraries analysed. If not identified as such, genetic markers developed from high-throughput sequencing data for non-model organisms may bias evolutionary studies or fail completely in experimental tests. In conclusion, our study demonstrates the enormous potential of low-coverage genome survey sequences and suggests bioinformatic analysis workflows. The results also advise a more sophisticated filtering for problematic sequences and non-target genome sequences prior to developing markers.

Citation: Leese F, Brand P, Rozenberg A, Mayer C, Agrawal S, et al. (2012) Exploring Pandora’s Box: Potential and Pitfalls of Low Coverage Genome Surveys for Evolutionary Biology. PLoS ONE 7(11): e49202. doi:10.1371/journal.pone.0049202 Editor: Ben J. Mans, Onderstepoort Veterinary Institute, South Africa Received July 6, 2012; Accepted October 8, 2012; Published November 21, 2012 Copyright: ß 2012 Leese et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: FL and CM were supported by German Research Foundation (DFG) grants LE 2323/2 and MA 3684/3 within the DFG priority programme (SPP) 1158. FL was furthermore supported by a European Science Foundation ‘‘Frontiers of Speciation Research‘‘ exchange grant to Cambridge, UK. CJS was supported by an Antarctic Science Bursary grant. CDS and NTR were supported by DFG grants 1460/3, 1460/8, by Ju¨rgen Heinze, and by a student scholarship of The Crustacean Society to NTR. KL and JJ were supported by Consortium Grant (NE/DO1249X/1) and the British Antarctic Survey Polar Science for Planet Earth Programme both funded by The Natural Environment Research Council. JD was supported by DFG grant RA 1688/2. SS was supported by a scheme to support specific activitiesof doctoral students of the rectorate of the Ruhr University Bochum. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected] . These authors contributed equally to this work.

Introduction ecological and evolutionary studies [2] and promises to overcome the statistical pitfalls associated with still often-used single marker Recent advances in high throughput sequencing technologies studies (see [3] for discussion). With the ‘‘rise of the machines’’ [4], have caused a paradigm shift in molecular evolutionary biology novel methodological possibilities are provided for addressing [1]. Whereas traditionally the analysis of many markers was a questions at both micro- (e.g. [5,6]) and macroevolutionary levels costly and tedious task and restricted mainly to genetic model (e.g. [7,8]). The basic principle common to both is that the organisms, it is now possible to screen large proportions of genomic regions identified for marker development and analysis previously unexplored genomes with high-throughput sequencing should be informative enough to answer the biological question methods almost as easily as known genomes. This hugely facilitates

PLOS ONE | www.plosone.org 1 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing under study. For microevolutionary questions, genomic regions this study was to analyse the data for possible contamination by with high variability at the population level are important [9], viruses, bacteria or endosymbionts. Our high-throughput libraries whereas in studies addressing old speciation events markers need originate from genetic non-model species and may thus be more to be less variable to avoid problematic homoplasious signals (e.g. representative of the majority of exploratory biological studies. [10]). For microevolutionary studies, microsatellites and minisa- Our results from 14 non-enriched genomic libraries show that low- tellites often represent the marker system of choice due to their coverage genome surveys of non-model organisms can yield many high variability [9,11,12]. Here, with high-throughput sequencing, informative genetic markers besides microsatellites. However, we the straightforward sequencing of enriched and non-enriched also identify significant contributions of intragenomic, intra- and libraries on fractions of 454 plates can provide a good solution extracellular DNA contamination in several libraries that, if not when searching for microsatellite markers [13–16] (for a review see systematically identified and removed using bioinformatic tech- [4,17,18]). niques, can lead to erroneous conclusions about the evolutionary For studies aiming to investigate recent divergence processes processes under study. between species, mitochondrial genes were and still are often the first choice [19]. Most of the mitochondrial genes evolve Materials and Methods comparatively fast and have the advantage of being haploid and abundant in cells. If evolutionary events that date back many Species investigated millions of years are the central theme of a study, the analysis of We analysed 14 genomic libraries of species from four different more conservative (slow evolving) regions is appropriate to avoid animal phyla (1 cnidarian, 9 arthropods, 3 molluscs, 1 echino- too many multiple substitutions overwriting the meaningful signal. derm, see Table 1). Furthermore, we also analysed two libraries Typical regions for phylogenetic questions are the conserved core enriched for microsatellites according to the protocol by Leese et regions of the nuclear multicopy rRNA genes [20]. al. [40] from the mollusc Lissarca notorcadensis and the asteroid With high-throughput sequencing, large sets of expressed echinoderm Odontaster validus (Table 1). sequence tags (ESTs) or specifically targeted nuclear genes can be amplified and compared among taxa [7,8,21]. An interesting DNA extraction and sequencing point in this respect is that with deep sequencing of nuclear or For DNA isolation, specimen tissue was extracted under clean mitochondrial genomes it is not only the sequence variation conditions in the lab to avoid contamination. For each genomic between homologous loci that can be used as phylogenetic library, DNA was extracted (see Supporting information S1) and information, but also the genome morphology, i.e. the order and 5 mg of genomic DNA sent to Macrogen Inc. (Seoul, South Korea) organisation of the mitochondrial genes [22–24]. Mitochondrial for library preparation. Individually-tagged libraries were analysed genome sequencing by traditional methods, such as primer on two full 454 plates on a GS-FLX sequencer (Roche) (Table 1). walking strategies or the use of conserved primers for long-range PCR, are time-consuming and have a limited success rate whereas Assembly high-throughput sequencing approaches can greatly facilitate development of complete or nearly complete mitochondrial From the raw sequence files, FASTA, quality and trace genomes [25–27]. information files were extracted using the sff_extract v. 0.2.8 In many published high-throughput sequencing studies, the python script [41]. Sequence tags of the reads were clipped. The sequence libraries are only partially explored, focussing on a processed raw data were assembled using MIRA version 3.2.1.5 particular set (certain protein coding genes) or type (microsatellites) [42] using the 454 default settings of the ‘‘de novo, genome, of markers and often neglect potential pitfalls of high-throughput accurate, 454’’ mode with two modifications after several tests: data. In particular, contamination of genomic libraries by bacteria, The parameter AL:mrs was set to 85 (default 70), which is the viruses or symbionts, by human material or cross-contamination is minimum percentage similarity of two overlapping sequences to be a well known problem (e.g. [28,29]). Such contamination can bias assembled, The parameter AS:mrpc was set to 2, i.e. at least two subsequent evolutionary analyses leading to erroneous conclusions reads (and not five or more as usual in higher coverage situations) (e.g. [30–32]). Therefore, the detection and removal of contam- were needed to create a contig (see results for full explanation of inant sequences is important prior to downstream analysis. the parameters). The MIRA assembler was chosen since it has Bioinformatic tools that aid in the process of identifying unique features such as chimera clipping, repeat masking and a contamination by heuristic comparisons of query sequences very flexible algorithm that can be adjusted to the specific 454 low- against reference databases, such as BLAST [33], BLAST+ [34] coverage data. The quality of the assemblies was visually inspected and BLAT [35], or programs that map the new sequences against using Geneious 5.4.6 [43]. To aid further analyses the contigs were reference genomes such as BWA [36], BWA-SW [37] or SSAHA uploaded into a custom MySQL database (MySQL-server v. [38], can further speed up and improve the process of identifying 5.1.44) [44]. All of the filtering steps and the final dataset and removing contaminant sequences from the genomic libraries production were performed in the database using SQL-com- (see [29] for a comparison of programs on metagenomic datasets). mands. The current study builds upon the first studies that have The Animal Genome Size Database [45] was used as a primary documented the potential of low coverage genome surveys, which resource to obtain genome sizes to compute approximate genomic analyse only a part of the whole genome, for evolutionary coverages for the libraries. We selected the closest relatives to our inferences (e.g. [25,39]). With the goal of widening the scope of target species from the database for comparison. Especially in low-coverage genome survey data, we explore their use not only cases for which no closely related species were found in the for one marker type, but for 1) micro- and minisatellites, 2) database, this approximation is to be treated with caution. For the mitochondrial genes and genomes, and 3) for nuclear genes genome size estimates of the Antarctic krill (Euphausia superba)we (protein-coding genes, rRNA genes, transposable elements). used the recently published information on genome size ranges Moreover, we demonstrate that several mid- to small budget labs published by Jeffrey [46]. can tap into the potential of high-throughput sequencing by sharing costs and thus maximizing output. A central objective of

PLOS ONE | www.plosone.org 2 November 2012 | Volume 7 | Issue 11 | e49202 LSOE|wwpooeog3Nvme 02|Vlm su 1|e49202 | 11 Issue | 7 Volume | 2012 November 3 www.plosone.org | ONE PLOS Table 1. Species analysed in this study and characteristics of the libraries.

Average Number of Average read Sum of read Number of Sum of contig contig length, Library Taxonomy Plate Library type reads length, bp lengths, bp contigs lengths, bp bp

Cnidaria Favia fragum (Esper) Anthozoa: Scleractinia 1 genomic 96,040 376.6 34,055,442 69,405 27,520,221 396.5 Arthropoda Austropallene cornigera (Mo¨bius) Pycnogonida 1 genomic 73,557 293.0 20,396,151 40,883 13,973,404 341.8 Colossendeis megalonyx Hoek Pycnogonida 1 genomic 100,719 259.4 25,499,956 57,425 17,539,519 305.4 Pallenopsis patagonica (Hoek) Pycnogonida 2 genomic 134,846 325.5 41,741,628 62,753 25,378,904 404.4 Uristes adarei (Walker) Malacostraca: Amphipoda 1 genomic 68,047 211.5 15,482,430 43,336 10,580,572 244.2 Euphausia superba Dana Malacostraca: Euphausiacea 1 genomic 49,802 247.6 12,098,817 42,256 10,868,476 257.2 Nematocarcinus lanceopes Bate Malacostraca: Decapoda 2 genomic 168,267 250.3 43,343,686 79,740 25,343,246 317.8 Hyas araneus (Linnaeus) Malacostraca: Decapoda 1 genomic 175,098 244.8 44,890,134 93,050 28,037,451 301.3 Metopaulias depressus Rathbun Malacostraca: Decapoda 2 genomic 186,890 265.5 55,152,741 63,040 26,186,278 415.4 Sericostoma personatum (Kirby & Spence) Hexapoda: Trichoptera 2 genomic 253,210 336.3 78,747,514 139,237 53,927,755 387.3 Mollusca Lepetodrilus sp. nov. Gastropoda: Vetigastropoda 2 genomic 339,640 330.7 105,577,603 178,788 69,304,964 387.6 Limatula hodgsoni (Smith) Bivalvia: Limoida 2 genomic 168,113 241.1 39,015,159 105,801 28,438,377 268.8 Arctica islandica (Linnaeus) Bivalvia: Veneroida 1 genomic 71,385 308.3 20,577,244 54,266 17,790,671 327.8 Lissarca notorcadensis Melvill & Standen Bivalvia: Arcoida 1 enriched 205,905 194.7 46,847,086 84,498 17,059,967 201.9 Echinodermata Odontaster validus Koehler Asteroidea 1 enriched 183,166 200.5 39,311,172 86,280 17,972,482 208.3 Gorgonocephalus chilensis (Philippi) Ophiuroidea 1 genomic 60,181 330.8 18,452,499 39,809 14,681,395 368.8 oeta iflso ihTruhu Sequencing Throughput High of Pitfalls & Potential The number of reads and number of Mbp refers to the unassembled raw data, whereas the number of contigs refers to the number of unique sequences after assembly with MIRA. doi:10.1371/journal.pone.0049202.t001 Potential & Pitfalls of High Throughput Sequencing

Taxonomic and functional characterization of the contigs were significantly simple. Following the recommendation by In order to classify the contigs we performed a number of Meglecz et al. [52] we made a final stringency filtering retaining BLAST searches with different strategies (see below) [33]. The only single read contigs with appropriate primers. The Phobos results were parsed and uploaded into the MySQL-database. We output data as well as the designed primers were stored in the used the accession numbers from the BLAST hits to obtain the MySQL database. The respective tables were queried to output ‘‘definition’’ and ‘‘description’’ sections of the corresponding total numbers and coverage of tandem repeats and numbers of loci sequences as well as the associated taxonomic information using with potentially suitable primers. an in-house tool that retrieves this information automatically from the NCBI Entrez Utilities Web Service (see http://www.ncbi.nlm. Searching for mtDNA nih.gov/entrez/query/static/esoap_help.html). These data were For the identification of mitochondrial DNA (mtDNA), all stored in the database and queried later for functional and assembled contigs and single-read contigs of individual species taxonomic assignment; summary statistics and inputs for subse- were converted to a BLAST database (BLAST+ package version quent downstream processing were obtained. 2.2.25+, [34]). Mitochondrial genome sequences of closely related For the purpose of taxonomic annotations, BLASTn searches species deposited in GenBank were used as queries for local with our sequences as queries against the whole nucleotide BLASTn and tBLASTx searches against this BLAST database. collection of NCBI sequences were performed on local servers. A Contigs in the database that had BLAST hits with an e-value conservative threshold e-value of #10212 was used. Only the best #10212 for a given query were assembled using Geneious version hits were collected and stored in the MySQL database. These data 5.4.6 [43]. The resulting contigs were inspected manually as were used to obtain information about non-eukaryotic sequences described in [26]. Every scaffold was examined by BLAST and sequences derived from known mobile elements (see section searches against GenBank, and proteins and rRNAs annotated ‘‘transposable elements’’ below) and to produce ‘‘contamination- accordingly. tRNAs were annotated using tRNAscan-SE 1.21 [53] free’’ datasets in which these sequences were removed. and ARWEN 1.2 [54].

Tandem repeat analysis Searching for nuclear genes Micro- and minisatellites (1–6 basepairs (bp) and 7–50 bp motif To obtain functional information on nuclear-encoded proteins, length, respectively) were searched for in all contamination-free we analysed our data (the contamination-free dataset: see below) (see below) contigs and single reads with a minimum length of with aid of the KEGG Automatic Annotation Server (KAAS [55]) 100 bp. This tandem repeat search was performed using the (August 2012). We utilized the online version of KAAS with the software Phobos 3.3.12 [47]. Since different studies used different single-directional best-hit method and default score thresholds. software and search criteria to find tandem repeats (see [48] for The results, i.e. the KEGG-Orthology assignments for individual discussion) we applied three different parameter settings to contigs, were uploaded into our database and the hits were further compare the results with other studies. First, we used the search classified according to the BRITE functional classification [56] parameters used in a recent comparative study on micro- and retrieved via the public services provided by KEGG [57]. Each minisatellites [48] (Phobos parameter settings –searchMode KEGG-Orthology record can potentially map to different BRITE- imperfect -u 1 -U 6 -g -5 -m -5 -s 12). In order to design primers classes and this problem of inherent redundancy was resolved with for only the best loci, the results were filtered for 100% perfect a simple weighting system: each BRITE-class assigned to a contig microsatellites. Second, we applied the search criteria used by was given a score equal to the number of reads for the contig Santana et al. [15] to search for microsatellites (equivalent Phobos divided by the number of pathways for that contig. BRITE-classes parameters –searchMode exact -u 1 -U 1 -s 11 for mononucleotide related to higher-level groups ‘‘Organismal Systems’’ and ‘‘Hu- repeats and –searchMode exact -u 2 -U 6 –minLength_b 5 -s 8 for man Diseases’’ as well as the class ‘‘Enzyme Families’’ were di- to hexanucleotide repeats). Third, we employed the search ignored when creating the frequency charts, since the functional parameters used by Abdelkrim et al. [13] and Gardner et al. [4] annotations were too imprecise for our data. (equivalent Phobos parameters –searchMode exact -u 2 -U 6 – Furthermore, to obtain an independent estimate of the number minLength_b -s 8). With the exception of [48] these studies did not of contigs with high similarities to known proteins, BLASTx explicitly search for minisatellites. In this study we searched for searches against the Swiss-Prot database [58] were performed minisatellites in the range of 7–50 bp motifs with the Phobos (October 5th, 2011) with a threshold e-value of #10212. Only the settings -u 7 -U 50 -R 30 -m -5 -g -5 -s 12 [48]. best hits were collected and stored in the MySQL database. These Since the aim of the study was to detect tandem repeats that data were used to obtain information about non-eukaryotic could be used as genetic markers we performed a search for sequences and sequences derived from known mobile elements appropriate primers annealing to the respective flanking regions (interspersed repeats) (see below) and to produce ‘‘contamination- with Primer3 v. 2.3.4 [49]. The parameters were the default ones free’’ datasets. Functional mapping of the BLASTx hits was with the following modifications: PRIMER_MAX_NS_AC- performed with the aid of the KEGG-database (Kyoto Encyclo- CEPTED = 1, PRIMER_PRODUCT_SIZE_RANGE = 100– pedia of Genes and Genomes, [57]). The database was accessed 300, PRIMER_PAIR_MAX_DIFF_TM = 8, PRIMER_MAX_- with a PHP-written client as follows: A GI-number (NCBI’s POLY_X = 4, PRIMER_NUM_RETURN = 3 and all tandem GenInfo Identifier) of a matched sequence was mapped to the repeats were masked with SEQUENCE_EXCLUDED_RE- KEGG gene identification number with the aid of interface GION. Further, all primer pairs were checked whether the functions (UniProt Mapping web-service) provided by the UniProt respective regions had low complexity (‘‘cryptic simplicity’’). This database [59]. Using the KEGG web-service, the KEGG gene simplicity test was performed with SIMPLE v. 5 [50,51]. The identification number was assigned to its respective KEGG- parameters were as follows: sequence type ‘n’ (DNA/RNA), equal Orthology identifier that was subsequently used to make functional weights for mono- to tetranucleotide motifs, 50 random sequences, annotations according to the BRITE pathways functional classi- shuffle elements method, and (half-) window size of 4. From the fication [56]. The annotation data were added to the same maximum of three primer pairs queried we stored either the pair MySQL database that stored the BLAST hits. This database without signs of simplicity or just the best one if primers in all pairs served as a source for final data analysis, comparison, and the

PLOS ONE | www.plosone.org 4 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing creation of the tables and figures. Each gene could potentially map to different pathways and this problem of inherent redundancy was resolved with a simple weighting system: each pathway assigned to a contig was given a score equal to the number of reads for the contig divided by the number of pathways for that contig. Pathways related to higher-level groups ‘‘Organismal Systems’’ and ‘‘Human Diseases’’ were ignored for the remainder of this study.

Searching for rRNA genes rRNA genes in the contigs were identified by conducting BLASTn searches on local computers against the nr Database of NCBI and extracting the best 20 hits. Definition lines and taxon information for the hits were obtained as outlined above. rRNA genes were detected in the MySQL database with a searching query for NCBI records explicitly containing one or more of the terms ‘‘rRNA; 18S; 28S; 5S; 5,8S; 5.8S; 23S; 25S; 17S; ribosomal RNA; rDNA; SSU; LSU; internal transcribed spacer; ITS1; ITS2; external transcribed spacer’’ in their descriptions.

Searching for transposable elements Figure 1. Workflow showing the methodological approach Similar to the searches for rRNA genes, potential transposable followed in this study. In this study we used a MySQL database (*) elements in the contigs were identified by filtering the best for storing the contigs. Other database formats are possible or reads BLASTn hits (case insensitive) for the terms ‘‘transposon, retro- can also be stored locally without a specific database. transposon, transposable element, interspersed element, inter- doi:10.1371/journal.pone.0049202.g001 spersed repeated mobile element, SINE sequence, SINE Alu, SINE family, LINE family, LINE sequence, Alu repeat’’. The Results terms ‘‘transposon’’ and ‘‘retroposon’’ were searched for in the ‘‘species’’ name field. If one of the terms ‘‘flanking region’’, Sequencing statistics/assembly ‘‘flank_region’’, ‘‘flanking end’’ occurred in the definition line, the Read number per genomic library ranged from 49,802 in hit was excluded from consideration. Antarctic krill Euphausia superba to 339,640 in the vent limpet Lepetodrilus sp. nov. The total number of base pairs for the clipped Searching for contamination reads ranged from 12,098,817 (Euphausia superba) to 105,577,603 (Lepetodrilus sp. nov., see Table 1). Average read lengths after Viruses. To account for possible viral contamination, BLASTx searches against the NCBI RefSeq Virus genomes quality clipping ranged from 211.5 bp (highly repetitive genome of Proteins Database were performed (viral*.protein.faa.gz, access the amphipod Uristes adarei) to 376.6 bp for the genomic library of date 09.09.2011). To avoid possible false positives (i.e. hits against the coral Favia fragum. In the microsatellite-enriched and length- loci similar to viral proteins, but not of viral origin) a very selected libraries the average lengths were shorter (194.7 bp and conservative approach with a maximum e-value of 10260 was 200.5 bp, for the bivalve Lissarca notorcadensis and the asteroid chosen. Odontaster validus, respectively). Even though approximately 5 mgof In addition, we used the web version of the software DeconSeq DNA were used consistently for library preparation, variation in [29] exploring the whole range of parameter combinations read numbers obtained for the tagged libraries on the plates was (coverage from 16 to 1006, identity from 60% to 100%). Both high (Table 1) reflecting both the strong variation inherent in the parameters were incremented by steps of one, resulting in 4,099 technology (mainly library preparation) and differences in DNA tested parameter combinations used to detect hits against viruses quality. in the genomic library of Metopaulias depressus (data available on Prior to producing the final assembly, we tested and compared request). different assembly settings and adjusted parameters for the MIRA Prokaryotic DNA. The data on prokaryotic contamination assembler. To accommodate for the low-coverage situation we were obtained with the same BLASTn searches described in the adjusted parameters and found that increasing the AL:mrs ‘‘Searching for protein-coding nuclear genes’’ section. Taxonomic parameter to 85, while using the accurate de novo genome assemble information was used to find sequences of prokaryotic origin. SQL mode of the MIRA assembler, produced high quality and and custom PHP scripts were utilized to obtain summary statistics conservative results. Increasing the AL:mrs stringency parameter concerning the numbers of reads and contigs assigned to respective reflected a trade-off between the low-coverage situation on the one groups, frequency charts coloured according to respective hand and a known increased percentage of wrong base calls at prokaryotic phyla and lists of highly frequent bacterial species. read ends using a 454 sequencing approach and allelic variability Life-history characterization of bacteria for a chosen library of on the other hand. In addition, the AS:mrpc parameter was set to Austropallene cornigera was performed manually through inspection 2, which means that at least two reads were needed to create a of the relevant literature (see Supporting information S7). contig (see Material and Methods). The assembly resulted in a An overview of the methodological workflow is presented in great number of assembled contigs, but most of the reads Figure 1. The data for this study can be viewed at http://www. remained single-read contigs (Supporting information S2). evoeco.de. We estimated coverage ranges for the genomic libraries by using information on C-values from closely related organisms (deposited in the Animal Genome Size Database). Since in the case of our

PLOS ONE | www.plosone.org 5 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing target species genome-size estimates were only available for one accepting only microsatellites with 100% perfection from the species, i.e. the Antarctic krill Euphausia superba [46], all other imperfect search with Phobos, the number of candidate loci and coverage approximations must be interpreted with caution. their total number decreased (see Supporting information S3), However, even when using the smallest genome size among ranging from 1,239 perfect microsatellites in Austropallene cornigera related taxa for computing coverage values, we always found that to 13,625 in Hyas araneus. After primer design with Primer3 the only a small proportion of the genome has been sequenced. For number of suitable loci decreased further. Considering only single the krill species Euphausia superba, our read library presents just a read contigs (to avoid potential paralogous loci) and rejecting low fraction of 0.03% of the genome (0.02% for contigs, see Table 2). complexity priming regions, the number of candidate loci ranged Similarly low coverage values were estimated for the amphipod from 109 in Uristes adarei to 1,079 in Lepetodrilus sp. nov. (see Uristes adarei based on a comparison to the uristid amphipod Anonyx Figure 2, Table 3). In the highly repetitive genome of the nugax (genomic coverage in the library of only 0.06% for reads and amphipod Uristes adarei most of the many microsatellites discovered 0.04% for contigs). For the species with most reads, the vent initially lacked a second flanking region or primers contained low limpet, coverage estimates range from 5.5% to 20.1% for the reads complexity regions and therefore most (98.73%) microsatellite loci (3.4% to 12.1% for contigs). were discarded from the initially 8,607 microsatellites resulting in the listed 109 (1.27%) candidate loci retained, when using the Genetic markers detected search parameters proposed in Mayer et al. [48]. For the settings Tandem repeats. Non-enriched genomic libraries generally suggested by Santana et al. [15], 232 microsatellites, and 25 for the mirror the microsatellite distribution in the genome [17]. Hence, extremely restrictive search parameters used by Abdelkrim et al. the analysis of a large proportion of non-enriched genomic reads [13] and Gardner et al. [4]. allows estimation of the genomic density of these repeats. By Minisatellites, i.e., repeats with a unit size of 7–50 bp, were analysing the density of microsatellites in the contigs (including found in all libraries (Figure 3, Supporting information S3). The single reads), using the search parameters of Mayer et al. [48], we coverage of minisatellites with a perfection of at least 95% ranged estimated densities for the individual libraries ranging from from 0.35% (3,529 bp/Mbp) in Euphausia superba to 10.34% 2,080 bp/Mbp, i.e., 0.21% of the genome in the bivalve Limatula (103,423 bp/Mbp) in Colossendeis megalonyx (see Supporting infor- hodgsoni to 161,435 bp/Mbp, i.e. 16.1% of the genome in the mation S3). The number of minisatellites in single read contigs amphipod Uristes adarei. With the proportion of tandem repeats with appropriate flanking regions and primers ranged from 101 in recovered from the genome of Uristes adarei, we document the Euphausia superba to 1,730 in Lepetodrilus sp. nov. For the enriched highest genomic microsatellite density reported so far for a libraries, the number of microsatellites retained after strict filtering metazoan genome (see [60] for a heteropteran species with a high was in the range of the other libraries (64 for Odontaster validus, microsatellite density in the unit size range of 2–10 bp, but 4,347 for Lissarca notorcadensis). In two enriched libraries created for detected with less restrictive search parameters). The actual other taxa, the proportion of microsatellites was about 2 orders of numbers of microsatellites identified per library ranged from magnitude higher, even after rigorous filtering (Supporting 1,961 (Austropallene cornigera) to 26,700 (Hyas araneus, see Supporting information S3). information S3). When applying strict filtering criteria, i.e.,

Table 2. Coverage estimations for the sequenced genomic libraries based upon genome size information of closely related taxa found in the Animal Genome Size database.

C-values Coverage estimations for assembled data for raw reads

larger smaller larger smaller Reference taxa Values, pg genome genome genome genome

Favia fragum Anthozoans 0.23 1.14 2.45% 12.12% 3.03% 15.00% Austropallene cornigera Pycnogonids 0.21 0.43 0.76 1.86% 6.74% 2.72% 9.84% Colossendeis megalonyx Pycnogonids 0.21 0.43 0.76 2.34% 8.46% 3.40% 12.30% Pallenopsis patagonica Pycnogonids 0.21 0.43 0.76 3.38% 12.24% 5.56% 20.14% Uristes adarei Uristidae: Anonyx nugax 27.00 0.04% 0.06% Euphausia superba Euphausia superba* 48.50 0.02% 0.03% Nematocarcinus lanceopes Caridea range: 3.30 40.89 0.06% 0.78% 0.11% 1.33% Hyas araneus Majoidea 2.21 2.30 3.88 3.90 4.55 0.62% 1.29% 1.00% 2.06% Metopaulias depressus Sesarmidae 3.99 4.40 0.60% 0.66% 1.27% 1.40% Sericostoma personatum unknown ? ? Lepetodrilus sp.nov. Lepetodrilidae 1.04 1.05 1.80 3.90% 6.75% 5.94% 10.29% Limatula hodgsoni Limidae: Lima 1.20 1.60 1.80% 2.40% 2.47% 3.29% Arctica islandica Close taxa: Veneridae+Corbicularange: 0.96 2.30 0.78% 1.88% 0.91% 2.17% Gorgonocephalus chilensis Ophiuroids 2.20 2.30 2.40 3.00 3.30 0.45% 0.68% 0.57% 0.85%

*Information on genome size of Euphausia superba is based upon the flow-cytometry estimates listed in [46]. doi:10.1371/journal.pone.0049202.t002

PLOS ONE | www.plosone.org 6 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing

Figure 2. Percentage of contigs with candidate microsatellites found in the non-enriched libraries with three different search parameter settings. Search parameter settings were adapted from the three studies [4,15,48] and used in Phobos [47] runs. Numbers on top of the columns represent the total number of perfect microsatellites retained after restrictive filtering for quality criteria. doi:10.1371/journal.pone.0049202.g002

Mitochondrial DNA markers that are available for comparison. As an example, no mitochon- All 14 genomic libraries contained mitochondrial DNA drial hits were initially found for the genomic library of fragments (Figure 4). A significant positive correlation between Gorgonocephalus chilensis. However, after the sequence of Astrospartus the number of contig bp of the assembly and the number of mediterraneus (GenBank Accession Number FN562580.1, [61]) was mitochondrial bp found was detected (Spearman rank correlation: deposited in GenBank, 2,870 bp of mitochondrial contigs were r = 0.6049, P = 0.0219, Figure 5). However, individual library found in the tBLASTx searches against the NCBI database and success varied considerably and the number of recovered genes could therefore be classified as such. and tRNAs differed substantially. For the spider crab Hyas araneus, full or partial sequences of every mitochondrial gene including 22 Nuclear DNA markers tRNAs were found (Figure 4). In the microsatellite-enriched Functional annotations performed with the KAAS pipeline libraries not a single mitochondrial read was found as expected allowed us to identify sequences with similarities to known (see Supporting information S4). proteins. The results showed that up to 2,772 contigs (for From the 454 data complete or nearly complete mitochondrial Lepetodrilus sp. nov.) had hits to known or predicted protein genes genomes can be obtained by linking contigs via Sanger sequencing (Table 4). As expected, the microsatellite-enriched libraries (i.e. (see e.g. [26] for the Colossendeis megalonyx library). from Lissarca and Odontaster) showed the lowest percentage of Interestingly, in the case of the coral Favia fragum it was even identifiable protein-coding genes. Among the 14 genomic libraries possible to isolate not only the almost complete (15,718 bp) the data obtained for the presumably more compact genomes (the mitochondrial genome of the host but also a 1,663 bp fragment of coral Favia fragum, pycnogonids and molluscs (except for Arctica the mitochondrial genome of its dinoflagellate symbiont Symbiodi- islandica)) showed higher values. A less sophisticated analysis nium spp. (Supporting information S4). utilizing BLASTx searches against the Swiss-Prot database showed The success of finding genes in a genetically uncharacterized comparable amounts of protein-coding sequences in our data, but taxon always depends on the availability and similarity of the data

PLOS ONE | www.plosone.org 7 November 2012 | Volume 7 | Issue 11 | e49202 LSOE|wwpooeog8Nvme 02|Vlm su 1|e49202 | 11 Issue | 7 Volume | 2012 November 8 www.plosone.org | ONE PLOS Table 3. Total number and genomic density of microsatellites found in the libraries before and after applying stringent filtering criteria (best primers, single read contigs only, see Supporting information S3 for further information).

Mayer et al. (2010) Santana et al. (2009) Gardner et al. (2011)

After stringent After stringent After stringent Density filtering, with Density filtering, with Density filtering, with Species Total Number (bp/Mbp) primers Total Number (bp/Mbp) primers Total Number (bp/Mbp) primers

Favia fragum 1,837 1,485 395 3,358 2,487 656 610 920 78 Austropallene cornigera 1,239 2,188 209 2,015 2,900 286 271 945 28 Colossendeis megalonyx 3,063 8,346 345 9,418 18,753 601 2,584 12,430 48 Pallenopsis patagonica 3,255 3,391 486 5,821 5,112 785 987 2,406 72 Uristes adarei 4,294 70,628 109 12,632 147,413 232 9,145 140,908 25 Euphausia superba 1,792 11,301 172 4,689 19,187 502 1,972 15,851 91 Nematocarcinus lanceopes 7,144 6,864 639 20,719 19,680 1,155 6,752 10,960 114 Hyas araneus 13,625 29,673 832 35,209 61,375 1,420 17,714 50,778 266 Metopaulias depressus 6,612 10,248 494 14,724 21,824 831 6,406 15,622 155 Sericostoma personatum 6,950 3,644 860 12,609 5,737 1,377 2,981 2,992 97 Lepetodrilus sp. nov. 11,994 5,569 1,079 30,609 12,378 2,019 13,693 8,768 305 Limatula hodgsoni 1,575 1,022 304 3,456 1,796 714 214 285 21 Arctica islandica 1,591 1,937 294 2,563 2,688 459 483 881 62 Lissarca (enriched) 27,768 76,580 464 59,565 138,709 806 37,181 114,629 640 Odontaster (enriched) 44,370 120,505 416 86,309 201,394 701 61,859 176,862 852 Gorgonocephalus chilensis 1,545 2,378 286 3,791 4,604 506 795 1,775 57

Results are given for three different sets of search parameters which correspond to search parameters in: Mayer et al. [48] but filtering for perfect microsatellites only, Santana et al. [15], Gardner et al. [4]. doi:10.1371/journal.pone.0049202.t003 oeta iflso ihTruhu Sequencing Throughput High of Pitfalls & Potential Potential & Pitfalls of High Throughput Sequencing

Figure 3. Percentage of contigs with microsatellites or minisatellites found in the non-enriched genomic libraries. Search parameters were according to Mayer et al. [48] used in Phobos [47]. For the analysis, repeats with a perfection greater or equal to 95% were retained. The numbers on top of the columns represent the total number found per library. doi:10.1371/journal.pone.0049202.g003 overall lower than the values obtained with KAAS due to different transposons. This reflects a proportion of 0.75% of the reads. candidate selection criteria. However, all of the species analysed in this study have a great Functional classes identified by KAAS in our libraries are very phylogenetic distance from classical genetic model species with diverse (Figure 6). For the genomes with large predicted sizes, in well-annotated transposable elements (data not shown). All more particular Euphausia superba and Uristes adarei, few hits to known closely related species are only poorly, if at all, genetically protein-coding genes were found. For the other genomes, up to characterized. Therefore, it is very likely that a major proportion 1,903 hits to genes from the KEGG categories ‘‘Genetic of transposable elements in our genomic libraries went unnoticed. Information Processing’’, ‘‘Cellular Processes’’, ‘‘Environmental Information Processing’’ and ‘‘Metabolism’’ were obtained. This Non-target organism DNA information could be important for a wide range of molecular We systematically searched for traces of DNA not belonging to studies. the organism under study. In particular, we searched for expected Ribosomal RNA genes. Ribosomal RNA (rRNA) genes symbionts and for bacterial and viral contamination. For each were detected with the aid of BLASTn searches against GenBank section we will here highlight cases in which the contamination and various rRNA genes were identified in the genomic libraries was particularly prominent. (Table 5). The number of positive rRNA gene hits ranged from 62 Symbionts. From the coral Favia fragum, tissue was extracted for Uristes adarei to 2,027 sequences for Limatula hodgsoni, adding up that contained a DNA mixture of the host Favia fragum and its to a total contig length of 2,453 bp for Uristes adarei and 58,588 for symbionts belonging to the dinoflagellate genus Symbiodinium. the 633 sequences detected for Gorgonocephalus chilensis. Therefore, the DNA could potentially include DNA of the nuclear Transposable elements. In the libraries of the three and mitochondrial host genome, the nuclear and mitochondrial pycnogonids and the two decapod species Metopaulias depressus symbiont genome, as well as the plastid genome of the symbiont. and Hyas araneus we found 5–81 reads (1–9 contigs) with matches The results of the mitochondrial DNA marker detection revealed to known transposable elements (Table 6). In the genomic library 15,718 assembled bp of mitochondrial reads for the coral and of Sericostoma personatum, however, we found 1,895 reads (assembled 1,663 bp for the symbiont (Supporting information S4). to 243 contigs) with high similarity to insect mariner retro-

PLOS ONE | www.plosone.org 9 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing

To explicitly distinguish between nuclear and plastid DNA of host and symbiont we performed BLASTn searches of all ‘‘Faviinae’’ and ‘‘Dinoflagellata’’ sequences as well as the newly sequenced genome of the cnidarian Nematostella vectensis obtained from GenBank (access date 24.09.2011; for exact search terms see Supporting information S5) against the 77,440 F. fragum tissue contigs (42,696,657 bp) as a database. We counted and assigned the hits with an e-value not exceeding 10212 (Supporting information S5) resulting in 434 contigs with at least one hit. Of all contigs 17 had matches to more than one of the genomes of interest. All these cases indicate erroneous annotations in the database. In addition, ten contigs had only hits against plastid sequences, 14 contigs had exclusive hits against the dinoflagellate genome and 393 contigs had hits against coral DNA only. Together with the results from the mitochondrial DNA these findings indicate that even low-coverage genome surveys may allow the identification of phylogenetically different genomes hidden within one organism. Bacteria. Up to 1.57% of the reads (1.31% of the contigs) in the libraries had highest similarity to bacterial DNA. Most hits were found for the three analysed pycnogonid species Austropallene cornigera (n = 537), Colossendeis megalonyx (n = 170) and Pallenopsis patagonica (n = 54), but bacterial DNA was also recorded in the vent limpet (n = 118, see Figure 7). Analysing the bacterial hits for the pycnogonid libraries showed that most had closest matches to various Gammaproteobacteria, whereas for the vent limpet the bacterial origin was very diverse (Figure 7, Supporting information S6). The diversity of bacterial species reported by the searches was high. For Austropallene cornigera, an Antarctic species, most of the Figure 4. Overview over the different mitochondrial genes hits were assigned to strains of Psychromonas ingrahamii, a cold- found in the non-enriched libraries. The upper section indicates adapted species known from Arctic waters (Supporting informa- full (dark blue) and partial (bright blue) mitochondrial protein-coding or rRNA genes recovered. The pie chart indicates the proportion and total tion S7). Furthermore, our data revealed many reads with hits to number of tRNAs found. In the lower section the total contig lengths (in various species of Shewanella, which are predominantly found in kb) of mitochondrial genes is shown. deep-sea habitats. Interestingly, 89 reads were assembled to one doi:10.1371/journal.pone.0049202.g004 contig that had the best match with Helicobacter pylori, a species

Figure 5. Correlation between genomic library size (y-axis) and total length of mitochondrial genome recovered (x-axis). A significantly positive linear correlation (Pearson r = 0.6049, P = 0.0219) between the number of base pairs sequenced and the proportion of the mitochondrial genome recovered was found. doi:10.1371/journal.pone.0049202.g005

PLOS ONE | www.plosone.org 10 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing

Table 4. Summary of the nuclear gene identification in the genomic and enriched libraries.

Total number of Library Type Taxonomic Group contigs Number of contigs With blastx hits With a KEGG against Swiss-Prot Orthology number

Favia fragum genomic Cnidaria 69,405 1,529 (2.20%) 1,634 (2.35%) Austropallene cornigera genomic Arthropoda 40,883 2,616 (6.40%) 1,717 (4.20%) Colossendeis megalonyx genomic Arthropoda 57,425 1,273 (2.22%) 820 (1.43%) Pallenopsis patagonica genomic Arthropoda 62,753 1,600 (2.55%) 1,057 (1.68%) Uristes adarei genomic Arthropoda 43,336 370 (0.85%) 110 (0.25%) Euphausia superba genomic Arthropoda 42,256 139 (0.33%) 127 (0.30%) Nematocarcinus lanceopes genomic Arthropoda 79,740 737 (0.92%) 209 (0.26%) Hyas araneus genomic Arthropoda 93,050 1,362 (1.46%) 566 (0.61%) Metopaulias depressus genomic Arthropoda 63,040 2,789 (4.42%) 530 (0.84%) Sericostoma personatum genomic Arthropoda 139,237 2,767 (1.99%) 1,639 (1.18%) Lepetodrilus sp. nov. genomic Mollusca 178,788 2,868 (1.60%) 2,772 (1.55%) Limatula hodgsoni genomic Mollusca 105,801 795 (0.75%) 756 (0.71%) Arctica islandica genomic Mollusca 54,266 422 (0.78%) 446 (0.82%) Gorgonocephalus chilensis genomic Echinodermata 39,809 462 (1.16%) 248 (0.62%) Lissarca notorcadensis enriched Mollusca 84,498 45 (0.05%) 127 (0.15%) Odontaster validus enriched Echinodermata 86,280 23 (0.03%) 34 (0.04%)

The number and proportion of contigs that had tBLASTx hits to proteins in the Swiss-Prot database and the number of contigs with a K-number assigned by the KEGG Automated Annotation Server pipeline KAAS is given. A visual representation of the KEGG categories of the hits is given in Figure 6. doi:10.1371/journal.pone.0049202.t004 commonly known from human stomachs where it induces gastritis Discussion [62]. Other abundant bacteria were also free-living, commensalic or pathogen bacteria that have been reported from various For all 14 genomic libraries analysed, the sequence coverage marine, often either deep-sea and/or cold-water environments. was just a minor fraction of the total genome. Estimated coverage Viruses. BLASTx searches against NCBI RefSeq Virus values ranged from 0.1 to 20%. Our results highlight the great genomes Proteins Database yielded hits in five libraries (Table 7). potential of such low-coverage next-generation sequencing data Most exceptional in terms of number of hits against virus for the simultaneous analysis of multiple genetic markers sequences was the bromeliad crab Metopaulias depressus. Here, we supplementing primary results of Rasmussen and Noor [25]. identified 14,131 reads (7.56% of total reads) with hits to the Moreover, for the first time we systematically compare the impact White-Spot-Syndrome-Virus (WSSV) that is well known primarily that different non-target DNA sources may have on analysed from penaeid shrimp aquaculture and repeatedly reported for libraries. The approach we advocate differs in one fundamental other decapods and even for other crustacean groups [63,64]. For aspect from most other studies (e.g., [4,14,15,25]): prior to the a more accurate assignment, we took the WSSV genome (gi: main analyses of the low-coverage data, an assembly was 17016399) as a query and performed tBLASTx searches with an e- performed to reduce redundancy. Although for average coverage value threshold of 10212 against all assembled and one-read values of ,1 it may seem unlikely that overlapping reads exist, it contigs in the genomic libraries of M. depressus. The tBLASTx turns out that several genomic fragments are highly overrepre- approach revealed that 9.23% of the sequenced DNA had hits and sented and form rather long contigs. We found this to be relevant thus resemble WSSV-related viruses. Interestingly, the already for rRNA genes (Table 5), mitochondrial genes (Figure 4, sequenced WSSV consists of only 292,967 bp in 531 ORFs, Supporting information S4), transposable elements (see for whereas we have found 453,318 unique bp in this study. It cannot examples the mariner retrotransposons, Table 6) but also for be excluded that horizontal gene transfer has contributed to the other, possibly single-copy nuclear genes (Table 4). Hence, the pattern observed in Metopaulias depressus. strategy of using a stringent assembly with repeats masked to avoid Using the DeconSeq software with different parameter combi- merging reads that are not from the same physical locus is nations, only between 0.51% (n = 322) and 5.95% (n = 3749) of the important to prepare the data for all subsequent steps. In a few proposed virus contaminant reads were found that had initially cases (1–4% of the contigs) MIRA did not mask terminal repeats, been detected by BLAST searches. All hits found with standard leading to some potentially erroneously assembled contigs. settings (coverage 906, identity $94%) belonged to repetitive Attempts to assemble the data without masking internal repeats regions or even consisted solely of a tandem repeat. No sequence using the assembler available in the commercial software Geneious of the WSSV-related virus was detected with default settings. In led to artifactual results, since several reads ending with the same the least restrictive search (coverage 16, identity $60%), only 30 tandem repeat were assembled. We therefore suggest to assemble of the 3,749 DeconSeq hits were contigs identified using the the reads only with a software capable of masking repeats prior to BLAST approach. All others seem to be false positives (mainly the assembly process to prevent unlinked contigs being joined tandem repeats). artificially by paralogous repeat regions.

PLOS ONE | www.plosone.org 11 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing

Figure 6. Hits of nuclear genes against KEGG BRITE Ontology database using the KAAS pipeline for the 16 genomic libraries. The number of hits is listed below the species name. Colours assigned according to the highest level of KEGG Orthology hierarchy (different organismal/ cellular pathway groups/ecosystem processes). doi:10.1371/journal.pone.0049202.g006

For our study we used 454 pyrosequencing as the sequencing filtering criteria led to a significant dropout of candidate loci for technique. This provides comparatively fewer but longer sequenc- the three different search parameter settings (see Supporting es as compared to most other high-throughput technologies [65], information S3). The extremely strong decrease in the candidate in particular when comparing it to Illumina sequencers. The microsatellites with suitable primers found in the Southern Ocean advantage of Illumina is that a much greater coverage can be amphipod Uristes adarei (only 1.95% of the initially found obtained. The short reads have the drawback that microsatellite candidates retained) was mainly due to microsatellites in this development is more difficult and homology searches are less highly repetitive genome, that were lacking a second flanking informative. It has been demonstrated that the disadvantage of regions because of great repeat length. short reads can be compensated effectively by using paired-end Although the choice of appropriate search parameters still Illumina sequences [66]. remains a subject of controversial discussion, it is obvious that for all search parameter sets, even with very stringent filtering criteria Tandem repeats (i.e. perfect microsatellites filtered from an imperfect Phobos Different studies have used different search criteria for defining search, considering only single reads with appropriate flanking microsatellites (see [48] for discussion). Hence, the computed regions) the total number of reads containing suitable candidate tandem repeat contents are difficult to compare. In this study we loci was sufficient for many candidate microsatellites ranging from used three different published sets of search parameters to detect 109 (2.53% of the microsatellite candidates) in Uristes adarei to microsatellites [13,15,48] and compared the results. Whereas most 1,085 (8.98%) in the vent limpet Lepetodrilus sp. nov. In general, for repeats were reported for the parameters used by Santana et al. molecular ecological or population genetic studies on non-model [15], a much lower number was found when applying the rather organisms, microsatellites have usually been obtained by enriching restrictive criteria used by Abdelkrim et al. [13] and Gardner et al. genomic libraries, cloning and shotgun Sanger sequencing of these [4]. fragments [69,70,40]. However, due to recurrent PCR amplifica- Strict filtering criteria led to a decrease in obtained microsat- tions, the redundancy is often considerable and the number of ellites mainly due to short read/contig lengths, which in turn led to clones that can be sequenced is limited due to the involved costs absent flanking regions (see also [13,14,67,68]). These strict (about 5 USD per plasmid prep and sequence read). For high-

PLOS ONE | www.plosone.org 12 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing

Table 5. rRNA genes found in the different libraries.

Avg reads Total contig Number of number per Number of Library Type Taxonomic Group length, bp reads contig contigs

Favia fragum genomic Cnidaria 17,606 700 31.82 22 Austropallene cornigera genomic Arthropoda 14,290 151 6.04 25 Colossendeis megalonyx genomic Arthropoda 25,166 506 23.00 22 Pallenopsis patagonica genomic Arthropoda 28,756 787 29.15 27 Uristes adarei genomic Arthropoda 9,314 62 3.26 19 Euphausia superba genomic Arthropoda 20,753 181 4.89 37 Nematocarcinus lanceopes genomic Arthropoda 21,651 1380 47.59 29 Hyas araneus genomic Arthropoda 36,624 827 20.17 41 Metopaulias depressus genomic Arthropoda 12,045 124 7.29 17 Sericostoma personatum genomic Arthropoda 55,832 637 8.27 77 Lepetodrilus sp. nov. genomic Mollusca 11,628 186 5.81 32 Limatula hodgsoni genomic Mollusca 42,545 2027 47.14 43 Arctica islandica genomic Mollusca 12,486 97 3.59 27 Gorgonocephalus chilensis genomic Echinodermata 58,588 633 10.91 58 Lissarca notorcadensis enriched Mollusca 2,453 96 7.38 13 Odontaster validus enriched Echinodermata 12,299 246 6.83 36

The total number of reads, the number of assembled contigs with coverage and the total unique rRNA gene bp are listed. doi:10.1371/journal.pone.0049202.t005 throughput sequencing data, the cloning step can be avoided and difference in the nuclear genome size may also impact the sequencing costs have decreased to less than 0.02 USD per 454 proportion of recovered mitochondrial genome fragments. For the read, with an average clipped length of 272.02 bp in our Antarctic krill species Euphausia superba, Jeffery (2010) documented examples. With increasing throughput and sequence lengths, an abnormally large genome size [46]. For this species, a next-generation sequencing platforms such as 454, Illumina or the particularly low proportion of mitochondrial gene fragments was Ion-Torrent, facilitate marker development drastically. In partic- recovered (5,502 bp) which might be a consequence of dilution ular, when applying the strict filtering criteria and when scanning effect due to large nuclear genome size. In other studies on for problematic reads, the polished high-throughput sequencing invertebrates, comparable or slightly higher proportions of datasets are superior to classical approaches. An additional benefit mitochondrial DNA were recovered from 454 libraries [25,72]. of using this methodology is that microsatellites in the vicinity of Completing the mitochondrial genomes by Sanger sequencing on coding genes (in particular within 59 and 39 UTRs) can be the basis of sequences obtained in this study was trivial for designed and compared to putatively neutrally evolving microsat- Colossendeis megalonyx, Sericostoma personatum, Austropallene cornigera, ellites in intergenic regions. Microsatellites in coding regions (i.e. and Pallenopsis patagonica [26]. Compared to primer walking non-neutral markers) reflect the selection regime prevalent in approaches with often unpredictable outcomes (see discussion in populations/species and can be used to identify functional traits [24]), we instead suggest to invest in high-throughput sequencing that explain evolutionary differences. as demonstrated in this study or by Groenenberg et al. [73]. The sequenced libraries that were enriched for microsatellites yielded many more microsatellite loci than the non-enriched Non-target genome DNA genomic libraries. However, all our non-enriched libraries Even though low-coverage genomic surveys represent only a provided sufficient unique and suitable microsatellite loci to work minor fraction of the genome, they offer a great potential for with in subsequent studies (46 unique contigs with suitable evolutionary biologists. Solely extracting markers in a traditional microsatellites for the genome of Uristes adarei), even with extremely way, i.e. picking those that look appropriate without doing a restrictive filter settings. This is in agreement with former sophisticated analysis of the whole large dataset, may result in comparisons of enriched vs. non-enriched 454 libraries in two overlooking interesting and important phenomena, such as DNA case studies of non-model and model organisms ([68] for Acacia of other organisms (viruses, bacteria, symbionts). Furthermore, harpophylla, [16] for Apis mellifera). primers may be designed for microsatellites located in mobile DNA elements in the genome, which leads to genotyping Mitochondrial genes problems. We have demonstrated that with some effort, these For all species, several mitochondrial gene fragments were important elements can be identified in order to maximise the use identified, although the overall yield differed considerably. For the of the polished high-throughput libraries. individual taxa, between 2,870 bp and 16,158 bp of the In the process of developing genetic markers it is commonly mitochondrial genomes were recovered. These differences may assumed that the presence of non-target DNA is negligible and be due to the extraction of different tissue types (Supporting hence requires no sophisticated action. However, contamination is information S1), since the copy number of mitochondrial DNA a severe problem in genetic research [32,74] and many different per cell can vary among different tissues [71]. Furthermore, sources of contamination of the target DNA exist. In this study we

PLOS ONE | www.plosone.org 13 November 2012 | Volume 7 | Issue 11 | e49202 LSOE|wwpooeog1 oebr21 oue7|Ise1 e49202 | 11 Issue | 7 Volume | 2012 November 14 www.plosone.org | ONE PLOS Table 6. Characteristics of contigs with homology to known transposable elements in the libraries.

Total contig Number of Avg read number Number of Total read Library Type Taxonomic Group length (bp) reads per contig contigs length (bp) Main hits

Favia fragum genomic Cnidaria 0 0 0 0 0 Austropallene cornigera genomic Arthropoda 5,955 81 10.1 8 20,220 Rana (frog) Tc1 Colossendeis megalonyx genomic Arthropoda 978 5 5.0 1 1,389 Xiphophorus (fish) Rex3- retrotransposons Pallenopsis patagonica genomic Arthropoda 4,875 26 2.9 9 9,994 Rana (frog) Tc1; Lepeophtheirus (copepod) Tc3 Uristes adarei genomic Arthropoda 0 0 0 0 0 Euphausia superba genomic Arthropoda 0 0 0 0 0 Nematocarcinus lanceopes genomic Arthropoda 0 0 0 0 0 Hyas araneus genomic Arthropoda 2,707 33 16.5 2 11,246 Galatheid (decapods) GalEa transposon Metopaulias depressus genomic Arthropoda 1,765 7 2.3 3 3,291 Litopenaeus (decapod) non-LTR retrotransposon I-type; insect Mariner-transposons Sericostoma personatum genomic Arthropoda 250,720 1,895 7.8 243 642,767 insect Mariner-transposons Lepetodrilus sp.nov. genomic Mollusca 0 0 0 0 0 Limatula hodgsoni genomic Mollusca 0 0 0 0 0 Arctica islandica genomic Mollusca 0 0 0 0 0 Gorgonocephalus chilensis genomic Echinodermata 0 0 0 0 0 Lissarca notorcadensis enriched Mollusca 0 0 0 0 0 Odontaster validus enriched Echinodermata 0 0 0 0 0

doi:10.1371/journal.pone.0049202.t006 oeta iflso ihTruhu Sequencing Throughput High of Pitfalls & Potential Potential & Pitfalls of High Throughput Sequencing

Figure 7. Bacterial hits found in four genomic libraries. Hits for the bacterial species are displayed next to the chart pie for species with $4% of the hits. For the three pycnogonid species, Gammaproteobacteria are predominant, whereas for the vent limpet Lepetodrilus sp. nov. different bacterial groups were detected. The colours of the charts relate to the phyla/classes of Bacteria (see legend). doi:10.1371/journal.pone.0049202.g007 used a BLAST approach to quantify the (minimum) amount of outlined above, can be superior over DeconSeq. Although slower, non-target DNA in the analysed libraries. Other bioinformatic BLAST was able to identify the WSSV-related virus in the approaches to detect contamination had been tested in phyloge- Metopaulias depressus. Our datasets may serve as a source for further nomic studies and were found to be superior to BLAST in terms of benchmark tests, similar to the study of Schmieder and Edwards speed [29]. In particular, approaches that align short reads against [29]. a known reference sequence of the potential contaminants using Evidence for the presence of symbionts was obtained for the Burrows-Wheeler Transform (BWA) are described as powerful hard coral Favia fragum. Here, the tissue extracted from organism [36]. Using the software DeconSeq [29], which utilizes the BWA, can potentially contain five genomes 1) the nuclear genome of the we could only detect a small subset of the virus contaminant in our coral host Favia fragum, 2) the mitochondrial genome of the coral library of Metopaulias depressus, but found a huge number of false host, 3) the nuclear genome of the symbiont Symbiodinium sp., 4) the positive, repetitive hits. In contrast, the BLAST approach mitochondrial genome of the symbiont, and 5) the chloroplast identified 9.23% of the total number of reads as originating from genome of the symbiont. This complex mixture of genomes is WSSV-related viruses and thus should be classified as a usually avoided in coral studies. Researchers use larval tissue (e.g. contamination. The comparatively low success of DeconSeq [75]) or sperm (e.g. [76]) to enrich the amount of host DNA and seems to be due to the low similarity of the virus found in minimize the presence of symbiont and mitochondrial genomes. Metopaulias depressus and the WSSV reference genome. In Our study, however, found that including the holobiont might exploratory studies on non-model organisms from weakly charac- provide a lot of additional data without necessarily reducing the terized habitats, as in our study, reference genomes for potential level of information obtained from the target host species. Even contaminants do not exist. Therefore, slower but more thorough without enriching the extracted tissue for the host nuclear DNA, approaches such as a combination of different BLAST searches, as the majority of obtained reads/contigs belonged to the host

PLOS ONE | www.plosone.org 15 November 2012 | Volume 7 | Issue 11 | e49202 LSOE|wwpooeog1 oebr21 oue7|Ise1 e49202 | 11 Issue | 7 Volume | 2012 November 16 www.plosone.org | ONE PLOS Table 7. Characteristics of contigs with homology to viral protein sequences in the libraries.

Avg read Total contig Number of number per Number of Total read length Library Type Taxonomic group length (bp) reads contig contigs (bp) Best hits

Favia fragum genomic Cnidaria 0 0 0 0 0 Austropallene cornigera genomic Arthropoda 0 0 0 0 0 Colossendeis megalonyx genomic Arthropoda 449 3 3.0 1 1,073 Enterobacteria phage lambda (tail component) Pallenopsis patagonica genomic Arthropoda 0 0 0 0 0 Uristes adarei genomic Arthropoda 0 0 0 0 0 Euphausia superba genomic Arthropoda 0 0 0 0 0 Nematocarcinus lanceopes genomic Arthropoda 0 0 0 0 0 Hyas araneus genomic Arthropoda 1,927 19 19.0 1 7,747 Cotesia congregata bracovirus (hypothetical protein) Metopaulias depressus genomic Arthropoda 447,712 14,131 147.2 96 5,253,117 White spot syndrome virus Sericostoma personatum genomic Arthropoda 8,674 145 72.5 2 48,174 Strawberry vein banding virus (retrotransposase), Vaccinia virus (ribonucleotide reductase) Lepetodrilus sp.nov. genomic Mollusca 468 1 1.0 1 468 Emiliania huxleyi virus 86 (ribonucleoside-diphosphate reductase) Limatula hodgsoni genomic Mollusca 0 0 0 0 0 Arctica islandica genomic Mollusca 0 0 0 0 0 Gorgonocephalus chilensis genomic Echinodermata 0 0 0 0 0 Lissarca notorcadensis enriched Mollusca 0 0 0 0 0 Odontaster validus enriched Echinodermata 0 0 0 0 0 oeta iflso ihTruhu Sequencing Throughput High of Pitfalls & Potential doi:10.1371/journal.pone.0049202.t007 Potential & Pitfalls of High Throughput Sequencing genome (Favia fragum). In addition, a very large proportion of the living. Although several libraries were not obviously affected by Favia mitochondrial genome could be assembled providing bacterial reads, the contribution of 1.57% in the library of valuable additional markers as well as a very good basis for Austropallene advises caution and highlights the importance of mitogenome completion using conventional Sanger sequencing. testing for contamination prior to subsequent analyses. For the symbiont mitogenome, an important mitochondrial Cases of contamination by other eukaryotic species were rare, marker (CO1) could be identified. Summarizing, we can conclude but present (e.g. the presence of a dragonfly sequence, although that the presence of several different genomes enhances the this template was not extracted from any of the authors’ amount of information that can be obtained from low-coverage laboratories). Clearly, such an unexpected contamination needs genome surveys. to be taken into account by active searching. This issue is further With respect to viral reads, we found several instances in which complicated for eukaryotic symbionts. Here, successfully finding a the amount of non-target DNA was considerably high with a certain non-target DNA depends on a homolog sequence being contribution of up to 10% of the total number of sequenced base deposited in the database that is used for contamination screening. pairs in the library (as in Metopaulias depressus). Viruses are capable Consequently, an unknown proportion of the libraries may of infecting organisms from all evolutionary lineages and actually originate from so called ‘‘dark matter’’ sequences of other species do so very frequently [77]. Hence, genomic traces of viruses, the that are not represented in the public databases. ‘‘virome’’, have been reported from genomic libraries, particularly from sequenced model organisms and revealed a huge diversity Conclusions (e.g. [78]). We found that bacterial reads were present in a non-negligible Using examples from 14 low-coverage genomic 454 libraries, proportion in four of our libraries. Interestingly, three of these genetic markers for population genetic analyses as well as for libraries were from pycnogonids and one from a hydrothermal phylogenetic studies or other biological disciplines were identified vent limpet. Pycnogonids have a special anatomy in that their and characterised. We suggest a series of steps which are critical to organs are shifted mainly into their legs, due to their very small avoid some of the problematic pitfalls of processing low coverage trunk. This, however, enhances the risk of including gut content libraries for evolutionary biology. We recommend an initial within the extracted DNA. Although we used only the tissue from stringent assembly of the reads as a key step for reducing the distal leg parts in Colossendeis, we had to grind whole legs for redundancy and increasing per locus information content, even in Austropallene and Pallenopsis to achieve the necessary amount of low coverage surveys. Masking repeats prior to assembly is DNA. Bacterial contamination, in particular in the latter two important to avoid merging unrelated reads that are united by pycnogonids, very likely stems from ingested marine bacteria. similar repeat motifs. Prior to downstream analyses of sequence Reads identified as bacterial contamination in the three pycno- data, it is important to validate the origin and identity of gonids usually had the closest matches to Gammaproteobacteria, sequences. Although for uncharacterized genomes little informa- which are cold-water adapted prokaryotes. Although studies of the tion on sequence identity is available in public databases, we have molecular diversity of bacterial communities in the Southern demonstrated that a significant proportion of library reads were of Ocean are in their infancy [79], preliminary data show that the non-target origin, using simple BLAST routines. If not excluded bacterial species differ from those in other oceans and have typical from the libraries prior to downstream analyses, such contaminant adaptations to the constantly cold marine environment. Species reads can lead to biased or even strongly misleading inferences of found in pelagic bacterial culture collections from the Southern evolutionary processes from the contaminated data. Ocean frequently belong to Gamma- or Alphaproteobacteria [79]. This view is mostly consistent with the hits observed in our library. Supporting Information Interestingly, the number of different bacteria we found was high and only few redundant reads were found, further highlighting the Supporting information S1 Information on sampling enormous bacterial diversity. One highly redundant contig, sites, tissue and DNA extraction protocols for the composed of 89 single reads, had the best match against the specimens analysed in this study. gram-negative bacterium Helicobacter pylori. Although Helicobacter (PDF) pylori is not only known from human intestines, but from different Supporting information S2 Overview over the assembly aquatic habitats including marine habitats [80], the strong results for the different genomic libraries. The number of overrepresentation of one fragment suggests that it may result contigs (y-axis, log-scale) with the respective number of reads from a contamination of the library. The limpet Lepetodrilus sp. included in the contig (x-axis). In all cases, single-read contigs nov. grazes on bacterial films in the vent habitat. Thus (x = 1) represented the majority of contigs after assembly. contamination by bacteria attached to the tissue processed is a (PDF) likely explanation. Rogers et al. [81] investigated the bacterial communities in the vent habitat by 16S rDNA clone library Supporting information S3 Information on the micro- sequencing. They found a high proportion of Gammaproteoba- satellites found (total number, bp, density, filtered ceria, Alphaproteobacteria, Bacteroidetes and Deferribacterales. candidate loci with primers). For the 95% perfection With the exception of the latter, these groups were also analyses we searched for imprefect microsatellites/minisatellites represented in our identified hits. In addition, Epsilonproteobac- and filtered out only those with a perfection equal or higher than teria, in particular bacteria of the genus Arcobacter, were found 95%. several times. (XLS) In principle, lateral gene transfer between symbiotic bacteria Supporting information S4 Overview over the different and eukaryotic genomes could be a further explanation for the mitochondrial genes found in the different libraries. An data, since it may be more common than expected and may even ‘f’ indicates that the whole gene was found whereas ‘p’ indicates be of functional importance in the course of evolution [82]. A that only a part of the gene was found. major argument against this possible explanation is the fact that most of the closest hits in the bacteria were species that are free- (XLS)

PLOS ONE | www.plosone.org 17 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing

Supporting information S5 Analysis of the contigs from information on habitats. References are found in the second the Favia fragum 454 library. The number of BLASTn hits worksheet (‘‘References’’). against either dinoflagellate (GenBank Taxon ID ‘‘Dinoflagellate’’ (XLS) NOT gene_in_plastid_chloroplast[PROP]), dinoflagellate plas- tome (GenBank Taxon ID ‘‘Dinoflagellate’’ AND gene_in_plas- Acknowledgments tid_chloroplast[PROP]) or coral (GenBank Taxon ID ‘‘Faviinae’’ We thank Anna Eckart and Julia M. Vollmer for laboratory assistance. and the Nematostella vectensis genome). The first 6 contigs had hits Sebastian Chevreux provided valuable hints on how to apply MIRA to the for both, nuclear and plastid dinoflagellate fragments. 11 contigs low-coverage datasets. We greatly acknowledge the efforts of the ICEFISH had hits for nuclear dinoflagellate and coral fragments. For 2004, ANDEEP II, CEAMARC and ChEsSO cruise organisers. We thank information on the mitochondrial genes found for the coral and two anonymous referees for helpful comments and suggestions on the the symbiont see Supporting information S4. manuscript. (XLS) Author Contributions Supporting information S6 Taxonomic classification of bacterial hits found within the three pycnogonids and Conceived and designed the experiments: FL CJS. Performed the the vent limpet Lepetodrilus sp. nov. (see also Figure 7). experiments: CDS CH CJS FL KPL LD JD JJ JN JNM JSD MJR NTR SA SS. Analyzed the data: FL AR CJS CM PB. Contributed reagents/ (PDF) materials/analysis tools: AR CDS CH CJS CM FL KL KPL MJR RT. Supporting information S7 Bacterial hits for Austropal- Wrote the paper: FL AR PB CM. Helped drafting the manuscript and lene cornigera obtained with BLAST and respective discussed the data: CDS CJS JD JJ JSD KL KPL MJR RT SA WGC. Read and approved the final version of the manuscript: All authors.

References 1. Hudson M (2008) Sequencing breakthroughs for genomic ecology and 22. Gissi C, Iannelli F, Pesole G (2008) Evolution of the mitochondrial genome of evolutionary biology. Mol Ecol Resour 8: 3–17. Metazoa as exemplified by comparison of congeneric species. Heredity 101: 2. Ekblom R, Galindo J (2011) Applications of next generation sequencing in 301–320. molecular ecology of non-model organisms. Heredity 107: 1–15. 23. Rota-Stabelli O, Kayal E, Gleeson D, Daub J, Boore J, et al. (2010) Ecdysozoan 3. Brito P, Edwards S (2009) Multilocus phylogeography and phylogenetics using mitogenomics: evidence for a common origin of the legged invertebrates, the sequence-based markers. Genetica 135: 439–455. Panarthropoda. Genome Biol Evol 2: 425–440. 4. Gardner M, Fitch A, Bertozzi T, Lowe A (2011) Rise of the machines - 24. Kilpert F, Held C, Podsiadlowski L (2012) Multiple rearrangements in recommendations for ecologists when using next generation sequencing for mitochondrial genomes of Isopoda and phylogenetic implications. Mol microsatellite development. Mol Ecol Resour 11: 1093–1101. Phylogenet Evol 64: 106–117. 5. Elshire R, Glaubitz J, Sun Q, Poland J, Kawamoto K, et al. (2011) A robust, 25. Rasmussen D, Noor M (2009) What can you do with 0.16genome coverage? A simple genotyping-by-sequencing (GBS) approach for high diversity species. case study based on a genome survey of the scuttle fly Megaselia scalaris (Phoridae). PLoS ONE 6: 19379. BMC Genomics 10: 382. 6. Hohenlohe P, Bassham S, Etter P, Stiffler N, Johnson E, et al. (2010) Population 26. Dietz L, Mayer C, Arango C, Leese F (2010) The mitochondrial genome of genomics of parallel adaptation in threespine stickleback using sequenced RAD Colossendeis megalonyx supports a basal position of Colossendeidae within the tags. PLoS Genet 6: 1000862. Pycnogonida. Mol Phyl Evol 58: 553–558. 7. Meusemann K, von Reumont B, Simon S, Roeding F, Strauss S, et al. (2010) A 27. Timmermans M, Dodsworth S, Culverwell C, Bocak L, Ahrens D, et al. (2010) phylogenomic approach to resolve the arthropod tree of life. Mol Biol Evol 27: Why barcode? High-throughput multiplex sequencing of mitochondrial 2451–2464. genomes for molecular systematics. Nucl Acids Res 38: e197. 8. Bybee S, Bracken-Grissom H, Haynes B, Hermansen R, Byers R, et al. (2011) 28. Nederbragt A, Rounge T, Kausrud K, Jakobsen K (2010) Identification and Targeted Amplicon Sequencing (TAS): a scalable next-gen approach to quantification of genomic repeats and sample contamination in assemblies of multilocus, multitaxa phylogenetics. Genome Biol Evol 3: 1312–1323. 454 pyrosequencing reads. Sequencing 2010: 782465. 9. Sunnucks P (2000) Efficient genetic markers for population biology. Trends Ecol 29. Schmieder R, Edwards R (2011) Fast identification and removal of sequence Evol 15: 199–203. contamination from genomic and metagenomic datasets. PLoS ONE 6: e17288. 10. Wa¨gele J, Mayer C (2007) Visualizing differences in phylogenetic information 30. Bourlat S, Nielsen C, Lockyer A, Littlewood D, Telford M (2003) Xenoturbella content of alignments and distinction of three classes of long-branch effects. is a deuterostome that eats molluscs. Nature 424: 925–928. BMC Evol Biol 7: 147. 31. Philippe H, Brinkmann H, Lavrov D, Littlewood D, Manuel M, et al. (2011) 11. Goldstein D, Pollock D (1997) Launching microsatellites: a review of mutation Resolving difficult phylogenetic questions: why more sequences are not enough. processes and methods of phylogenetic interference. J Hered 88: 335–342. PLoS Biology 9: e1000602. 12. Goldstein D, Schlo¨tterer C (1999) Microsatellites: evolution and applications. 32. Laurin-Lemay S, Brinkmann H, Philippe H (2012) Origin of land plants New York, USA: Oxford University Press Inc. 352 p. revisited in the light of sequence contamination and missing data. Curr Biol 22: 13. Abdelkrim J, Robertson B, Stanton J, Gemmell N (2009) Fast, cost-effective R593–R594. development of species-specific microsatellite markers by genomic sequencing. 33. Altschul S, Madden T, Schaffer A, Zhang J, Zhang Z, et al. (1997) Gapped Biotechniques 46: 185–192. BLAST and PSI-BLAST: a new generation of protein database search 14. Castoe T, Poole A, Gu W, De Koning A, Daza J, et al. (2010) Rapid programs. Nucl Acids Res 25: 3389–3402. identification of thousands of copperhead snake (Agkistrodon contortrix) microsat- ellite loci from modest amounts of 454 shotgun genome sequence. Mol Ecol 34. Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, et al. (2009) Resour 10: 341–347. BLAST+: architecture and applications. BMC Bioinformatics 10: 421. 15. Santana Q, Coetzee M, Steenkamp E, Mlonyeni O, Hammond G, et al. (2009) 35. Kent W (2002) BLAT—the BLAST-like alignment tool. Genome Res 12: 656– Microsatellite discovery by deep sequencing of enriched genomic libraries. 664. Biotechniques 46: 217–223. 36. Li H, Durbin R (2009) Fast and accurate short read alignment with Burrows– 16. Malausa T, Gilles A, Megle´cz E, Blanquart H, Duthoy S, et al. (2011) High- Wheeler transform. Bioinformatics 25: 1754–1760. throughput microsatellite isolation through 454 GS-FLX Titanium pyrose- 37. Li H, Durbin R (2010) Fast and accurate long-read alignment with Burrows- quencing of enriched DNA libraries. Mol Ecol Resour 11: 638–644. Wheeler transform. Bioinformatics 26: 589–595. 17. Martin J, Pech N, Megle´cz E, Ferreira S, Costedoat C, et al. (2010) 38. Ning Z, Cox A, Mullikin J (2001) SSAHA: a fast search method for large DNA Representativeness of microsatellite distributions in genomes, as revealed by databases. Genome Res 11: 1725–1729. 454 GS-FLX titanium pyrosequencing. BMC Genomics 11: 560. 39. Straub S, Fishbein M, Livshultz T, Foster Z, Parks M, et al. (2011) Building a 18. Guichoux E, Lagache L, Wagner S, Chaumeil P, Le´ger P, et al. (2011) Current model: Developing genomic resources for common milkweed (Asclepias syriaca) trends in microsatellite genotyping. Mol Ecol Resour 11: 591–611. with low coverage genome sequencing. BMC Genomics 12: 211. 19. Avise J (2000) Phylogeography: the history and formation of species. Cambridge, 40. Leese F, Mayer C, Held C (2008) Isolation of microsatellites from unknown MA: Harvard University Press. 453 p. genomes using known genomes as enrichment templates. Limnol Oceanogr 6: 20. Hillis D, Dixon M (1991) Ribosomal DNA: molecular evolution and 412–426. phylogenetic inference. Q Rev Biol 66: 411–453. 41. Blanca J, Chevreux B (2011) sff_extract. Website:http://bioinf.comav.upv.es/ 21. Dunn C, Hejnol A, Matus D, Pang K, Browne W, et al. (2008) Broad sff_extract, acceessed 2012 March 3rd. phylogenomic sampling improves resolution of the animal tree of life. Nature 42. Chevreux B (2005) MIRA: an automated genome and EST assembler. PhD 452: 745–749. thesis, Ruprecht-Karls University, Heidelberg, Germany. 161 p.

PLOS ONE | www.plosone.org 18 November 2012 | Volume 7 | Issue 11 | e49202 Potential & Pitfalls of High Throughput Sequencing

43. Drummond A, Ashton B, Buxton S, Cheung M, Cooper A, et al. (2011) 64. Chen L, Lo C, Chiu Y, Chang C, Kou G (2000) Natural and experimental Geneious v5.4.6. Website:http://www.geneious.com. Accessed 2011 October infection of white spot syndrome virus (WSSV) in benthic larvae of mud crab 13th. Scylla serrata. Dis Aquat Organ 40: 157–161. 44. Widenius M, Axmark D, MySQL A (2002) MySQL Reference Manual: 65. Glenn T (2011) Field guide to next-generation DNA sequencers. Mol Ecol Documentation from the Source. O’Reilly Community Press. 802 p. Resour 11: 759–769. 45. Gregory T, Nicol J, Tamm H, Kullman B, Kullman K, et al. (2007) Eukaryotic 66. Castoe T, Poole A, de Koning A, Jones K, Tomback D, et al. (2012) Rapid genome size databases. Nucl Acids Res 35: D332–D338. microsatellite identification from Illumina paired-end genomic sequencing in 46. Jeffery N (2012) The first genome size estimates for six species of krill two birds and a snake. PLoS ONE 7: e30953. (Malacostraca, Euphausiidae): large genomes at the north and south poles. Polar 67. Csencsics D, Brodbeck S, Holderegger R (2010) Cost-effective, species-specific Biol 35: 959–962. microsatellite development for the endangered Dwarf Bulrush (Typha minima) 47. Mayer C (2010) Phobos Version 3.3.12. A tandem repeat search program. 20 p. using next-generation sequencing technology. J Hered 101: 789–793. Available:http://www.rub.de/spezzoo/cm/cm_phobos.htm. Accessed 2011 th 68. Lepais O, Bacles C (2011) Comparison of random and SSR-enriched shotgun October 13 . pyrosequencing for microsatellite discovery and single multiplex PCR optimi- 48. Mayer C, Leese F, Tollrian R (2010) Genome-wide analysis of tandem repeats in zation in Acacia harpophylla F. Muell. ex Benth. Mol Ecol Resour 11: 711–724. Daphnia pulex - a comparative approach. BMC Genomics 11: 277. 69. Zane L, Bargelloni L, Patarnello T (2002) Strategies for microsatellite isolation: a 49. Rozen S, Skaletsky H (2000) Primer3 on the WWW for general users and for review. Mol Ecol 11: 1–16. biologist programmers. Methods Mol Biol 132: 365–386. 50. Hancock J, Armstrong J (1994) SIMPLE34: an improved and enhanced 70. Glenn T, Schable N (2005) Isolating microsatellite DNA loci. Methods Enzymol implementation for VAX and Sun computers of the SIMPLE algorithm for 395: 202–222. analysis of clustered repetitive motifs in nucleotide sequences. Comput Appl 71. Cavelier L, Johannisson A, Gyllensten U (2000) Analysis of mtDNA copy Biosci 10: 67–70. number and composition of single mitochondrial particles using flow cytometry 51. Alba` M, Laskowski R, Hancock J (2002) Detecting cryptically simple protein and PCR. Exp Cell Res 259: 79–85. sequences using the SIMPLE algorithm. Bioinformatics 18: 672–678. 72. Feldmeyer B, Hoffmeier K, Pfenninger M (2010) The complete mitochondrial 52. Meglecz E, Costedoat C, Dubut V, Gilles A, Malausa T, et al. (2010) QDD: a genome of Radix balthica (Pulmonata, Basommatophora), obtained by low user-friendly program to select microsatellite markers and design primers from coverage shot gun next generation sequencing. Mol Phylogenet Evol 57: 1329– large sequencing projects. Bioinformatics 26: 403–404. 1333. 53. Schattner P, Brooks A, Lowe T (2005) The tRNAscan-SE, snoscan and snoGPS 73. Groenenberg D, Pirovano W, Gittenberger E, Schilthuizen M (2012) The web servers for the detection of tRNAs and snoRNAs. Nucl Acids Res 33: complete mitogenome of Cylindrus obtusus (Helicidae, Ariantinae) using Illumina W686–W689. next generation sequencing. BMC Genomics 13: 114. 54. Laslett D, Canba¨ck B (2008) ARWEN: a program to detect tRNA genes in 74. Longo M, O’Neill M, O’Neill R (2011) Abundant human DNA contamination metazoan mitochondrial nucleotide sequences. Bioinformatics 24: 172–175. identified in non-primate genome databases. PLoS ONE 6: 16410. 55. Moriya Y, Itoh M, Okuda S, Yoshizawa A, Kanehisa M (2007) KAAS: an 75. Polato N, Vera J, Baums I (2011) Gene discovery in the threatened elkhorn automatic genome annotation and pathway reconstruction server. Nucl Acids coral: 454 sequencing of the Acropora palmata transcriptome. PLoS ONE 6: Res 35: W182–W185. e28634. 56. Kanehisa M, Goto S, Hattori M, Aoki-Kinoshita K, Itoh M, et al. (2006) From 76. Shinzato C, Shoguchi E, Kawashima T, Hamada M, Hisata K, et al. (2011) genomics to chemical genomics: new developments in KEGG. Nucl Acids Res Using the Acropora digitifera genome to understand coral responses to 34: D354–D357. environmental change. Nature 476: 320–323. 57. Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M (2004) The KEGG 77. Edwards R, Rohwer F (2005) Viral metagenomics. Nat Rev Microbiol 3: 504– resource for deciphering the genome. Nucl Acids Res 32: D277–D280. 510. 58. Boeckmann B, Bairoch A, Apweiler R, Blatter M, Estreicher A, et al. (2003) The 78. Rooks D, Smith D, McDonald J, Woodward M, McCarthy A, et al. (2010) 454- SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003. pyrosequencing: a molecular battiscope for freshwater viral ecology. Genes 1: Nucl Acids Res 31: 365–370. 210–226. 59. Bairoch A, Apweiler R, Wu C, Barker W, Boeckmann B, et al. (2005) The 79. Murray A, Grzymski J (2007) Diversity and genomics of Antarctic marine micro- universal protein resource (UniProt). Nucl Acids Res 33: D154–D159. organisms. Phil Trans R Soc B 362: 2259–2271. 60. Perry J, Rowe L (2011) Rapid microsatellite development for water striders by 80. Twing K, Kirchman D, Campbell B (2011) Temporal study of Helicobacter pylori next-generation sequencing. J Hered 102: 125–129. presence in coastal freshwater, estuary and marine waters. Water Res 45: 1897– 61. Perseke M, Bernhard D, Fritzsch G, Bru¨mmer F, Stadler P, et al. (2010) 1905. Mitochondrial genome evolution in Ophiuroidea, Echinoidea, and Holothur- oidea: Insights in phylogenetic relationships of Echinodermata. Mol Phylogenet 81. Rogers A, Tyler P, Connelly D, Copley J, James R, et al. (2012) The discovery of Evol 56: 201–211. new deep-sea hydrothermal vent communities in the Southern Ocean and 62. Cover T, Blaser M (2009) Helicobacter pylori in health and disease. Gastroenter- implications for biogeography. PLoS Biol 10: e1001234. ology 136: 1863–1873. 82. Hotopp J, Clark M, Oliveira D, Foster J, Fischer P, et al. (2007) Widespread 63. Hossain M, Chakraborty A, Joseph B, Otta S, Karunasagar I, et al. (2001) lateral gene transfer from intracellular bacteria to multicellular eukaryotes. Detection of new hosts for white spot syndrome virus of shrimp using nested Science 317: 1753–1756. polymerase chain reaction. Aquaculture 198: 1–11.

PLOS ONE | www.plosone.org 19 November 2012 | Volume 7 | Issue 11 | e49202 Supporting information 1: Information on sampling sites, tissue and DNA extraction protocols for the specimens analysed in this study.

Libary Description of extraction protocol Originally, the Favia individual came from Curaçao. Tissue was scraped off an area of 3 by 3 mm and DNA was extracted using the Qiagen DNeasy Blood and Tissue kit following the providers instructions. Favia fragum DNA integrity was confirmed on a 1% agarose gel and the concentration was determined using an Implen photometric cell.

Specimens were sampled during the ICEFISH 2004 expedtion from Bouvet Island. Sampling details: Station 59 Otter Trawl 44, 26.06.2004. Tissue was extracted from Tibia1, Tibia2, Tarsus, Propodus of 4 Austropallene walking legs using Qiagen the DNA Blood & Tissue kit according to the manufacturer's protocol.Integrity of DNA was verified on a 1.5% TBE agarose gel and concentration determined on a Implen cornigera photometric cell. Specimen was sampled during the ICEFISH 2004 expedition from Bouvet Island waters. Sampling details: Station 76, Otter Trawl 50, 28.06.2004. Tissue was extracted from Tibia1, Tibia2, Tarsus, Propodus Colossendeis of 2 walking legs using Qiagen the DNA Blood & Tissue kit according to the manufacturer's protocol. Integrity of DNA was verified on a 1.5% TBE agarose gel and concentration determined on a Implen megalonyx photometric cell. Specimen was sampled during the ICEFISH 2004 expedition from the Falkland Islands (Ppa E010). Tissue was extracted from Tibia1, Tibia2, Tarsus, Propodus of 4 walking legs using Qiagen DNA Blood & Pallenopsis Tissue kit according to the manufacturer's protocol. Only 100 ul AE buffer was used for elution. Integrity of DNA was verified on a 1.5% TBE agarose gel and concentration determined on a Implen patagonica photometric cell.

CEMARC expedtion (2008-09) to Terre Adelie. Station 1472. Tissue was extracted from all walking legs and 2 pleopod I using Qiagen DNA Blood & Tissue kit according to the manufacturer's protocol. Only Uristes adarei 100 ul AE buffer was used for elution. Integrity of DNA was verified on a 1.5% TBE agarose gel and concentration determined on a Implen photometric cell.

CEMARC expedition (2008-09). Date: 15.1.08 , Location: Terre Adelie (-66.166858, 139.6211), Antarctic Euphausia superba Tissue was extracted from legs using the Qiagen DNA Blood and Tissue kit according to manufacturer's protocol. Quality and quantity of DNA was checked on a Nanodrop (Thermo Scientific).

Nematocarcinus Tissue was extracted from legs using the Qiagen DNA Blood and Tissue kit according to manufacturer's protocol. Quality and quantity of DNA was checked on a Nanodrop (Thermo Scientific). lanceopes

Specimens used for extraction were collected at Helgoland and Spitzbergen (Ny Ålesund). Tissue was extracted from legs using the Qiagen DNA Blood and Tissue kit according to manufacturer's protocol. Hyas araneus Quality and quantity of DNA was checked on a Nanodrop (Thermo Scientific).

Metopaulias Sampling information: NR-15-7, Limetreegarden (Jamaica, 22.02.2009, Rivera, Schubart, Lopez, Stemmer), 18° 20,138' N - 77° 23,762' W. DNA was extracted from muscle tissue using the Qiagen DNA Blood depressus and Tissue kit according to the manufacturer's recommendations.

Sericostoma Sampling location: Breitenbach, Hessen. Tissue was extracted from muscle tissue in the pro-, meso-, and metathorax using Qiagen DNA Blood & Tissue kit according to the manufacturer's protocol. Only personatum 100 ul AE buffer was used for elution. Integrity of DNA was verified on a 1.5% TBE agarose gel and concentration determined on a Implen photometric cell.

Sampled during JC42 RRS James Cook NERC funded ChEsSO consortium cruise. DNA was extracted both into 200ul TE buffer using Qiagen DNeasy extraction kits, then used Illustra GenomiPhi whole Lepetodrilus sp. nov. genome amplification to increase copy numbers, for a final concentration of 490ng/ul.

Sampled during ANDEEP II ANTXIX/4 RV Polarstern cruise. DNA was extracted from tissue into 200ul TE buffer using Qiagen DNeasy extraction kits, then used Illustra GenomiPhi whole genome Limatula hodgsoni amplification to increase copy numbers, for a final concentration of 190ng/ul.

Specimens Ais F100 from Norway, AisF129 Kattegatt, AisF200 Iceland and AisF244from the WhiteSea were processed. Tissue was extracted from legs using the Qiagen DNA Blood and Tissue kit according Arctica islandica to manufacturer's protocol. Quality and quantity of DNA was checked on a Nanodrop (Thermo Scientific). 1.25 microgram of each specimen was pooled.

Gorgonocephalus The specimen of Gorgonocephalus was caught using Agazzis trawls from the RRS James Clark Ross from either the JR144, 179 or 230 cruises. DNA extraction was 'Salting Out' (Sunnucks and Hales 1996). chilensis

Lissarca Specimens of Lissarca notorcadenis were caught using Agazzis trawls from the RRS James Clark Ross from the JR144 cruise. DNA extraction was 'Salting Out' (Sunnucks and Hales 1996). notorcadensis

Odontaster validus Specimens of Odontaster were hand collected at Rothera station. DNA extraction was 'Salting Out' (Sunnucks and Hales 1996).

Supporting information S2: Overview over the assembly results for the different genomic libraries.

Supporting information 4: Overview over the different mitochondrial genes found in the different libraries. An 'f' indicates that the whole gene was found whereas 'p' indicates that only a part of the gene was found. Gene Favia Symbiodinium Austropallene Colossendeis Pallenopsis Uristes Euphausia Nematocarcinus Hyas Metopaulias nad1 p - p p p p p p f p nad2 p - p p - p p p f - nad3 f - - f p p p p f - nad4 f - p p p p - p p p nad4L f - p f - - p p f - nad5 p - p p p p p p f p nad6 - - - - - p p p f - atp6 f - f f - p p - f p atp8 f - * f - * f - f - cox1 f p p p p p - p p p cox2 f - p f - p - p f - cox3 f - p f p p p p f - cob p - p p p p p p f p rrnS - - p - p - - - f p rrnL - - p p - p p p f p tRNAs 2 0 7 10 3 6 5 12 22 1 total, bp 15718 1663 11083 11711 5906 7427 5532 11300 16158 5045

* = annotated as single pass membrane protein but not exactly as atp8 Supporting information 4 (continued) Sericostoma Lepetodrilus Limatula Arctica Lissarca Odontaster Gorgonocephalus f f f p - - - p f - - - - p f f - p - - - f p p - - - p p f - - - - - p p p p - - - - f - - - - - f f - p - - - f * - - - - - f f p - - - p f p p - - - p f p p - - - - f p p p - - p p p p - - - - f p p p - - - 17 10 5 1 0 0 2 13776 17590 8537 5768 0 0 2870 Dissertation Lars Dietz

8) General Discussion

The aims of this thesis were to study phylogeographic and systematic questions in Antarctic pycnogonids. To study patterns and infer processes, morphological and genetic data were investigated using a variety of methods. In the following sections I will integrate over the different chapters of this thesis to summarize the general conclusions on pycnogonid systematics and the history of Antarctic marine biodiversity that can be drawn from the results. I also describe opportunities for future research that result from the studies presented here.

Morphological systematics Morphological analyses with the objective of resolving large-scale affinities are often performed as cladistic analyses based on a matrix of presence-absence data (e.g. Arango & Wheeler 2007). However, for investigating affinities on a smaller scale, such as testing the distinctness of morphologically similar species within a genus, such an approach may be impracticable as clear presence-absence data are difficult to define or even non-existent. In this case, it may be more useful to do statistical morphometric analyses, i.e. to conduct detailed measurements of as many different characters (e.g. leg articles, body segments) as possible and to test for the presence of significant differences between species. My results show that such an approach, while not suitable for generating a phylogenetic tree, can be effective in testing proposed synonymies between taxa, especially when accompanied by molecular data used to delimit the groupings whose validity is being tested morphologically. In the specific case of the genus Colossendeis, my results show that the North Pacific species C. tenera should not be treated as a subspecies of the Southern Ocean C. megalonyx (contra Turpaeva 1973). This suggestion would have implied that C. tenera is a non-Southern Ocean taxon nested within a Southern Ocean group, which is known in some cases (e.g. Strugnell et al. 2008) and may be explained by the Southern Ocean origin of most of the world oceans’ deep-sea water. However, the data reject the synonymization of C. tenera with C. megalonyx and show it to be more closely related to the widespread deep-sea species C. angusta and C. gracilis. The sequences of C. tenera included in the study were found to be paraphyletic towards C. angusta, which agrees with the more plesiomorphic morphological characters of the former such as the presence of eyes and a less simplified ovigeral spine configuration. While an Antarctic ancestry of C. tenera and related species is still possible as the “longitarsal” group of Colossendeis (Gordon 1932), to which it belongs, is mostly Antarctic, the evidence for this is ambiguous due to the low taxon sampling especially of non-Antarctic species. As the examined specimens of C. angusta are exclusively North

201 Dissertation Lars Dietz

Atlantic, further data from this species and C. gracilis from other oceans would be needed to elucidate their taxonomy and evolutionary history. Another finding of my studies is that Colossendeis robusta and C. glacialis are distinct species, contrary to the suggestions of Fry & Hedgpeth (1969) and Child (1995), and that there is also another widespread species, C. bouvetensis Dietz & Leese 2015, which resembles and thus has been misclassified as C. robusta. All these species have in common that they are rather small for colossendeids (except for C. robusta from the type locality), have a more or less cylindrical proboscis about equally long as the trunk, and the distal five palp articles are approximately equally long. Surprisingly, the molecular data showed that, while most of the species with those characters form a clade, C. glacialis is not closely related to them. The similarity may be explained by adaptation to a similar feeding style, but this cannot yet be tested as very little is known about the differences in feeding ecology between Antarctic colossendeid species. My results show a high degree of congruence between morphological and molecular data suggesting that both data sources may be useful for recognizing cryptic diversity in pycnogonids, as has also been shown by Arango & Brenneis (2013) for Australian Pseudopallene species. These results, as well as those found in Pallenopsis leading to the recognition of the new species P. yepayekae, suggest that many pycnogonid taxa may be “overlumped”, i.e. a substantial amount of species-level diversity may have been missed by previous morphological studies that could not recognize interspecies diagnostic characteristics due to small sample size and no testing of hypotheses by molecular data. Live coloration, which has not yet been studied in detail in colossendeids as it is lost in ethanol-preserved specimens, is a useful taxonomic character in Pseudopallene (Arango & Brenneis 2013). It might also be useful in colossendeids, as e.g. C. glacialis is brightly colored (Gordon 1932). My morphological results also validate the usefulness of the ovigeral strigilis, i.e. the arrangement of spines on the distal four articles of the oviger used for cleaning, as a taxonomic character for colossendeids. This was first described by Hoek (1881) and studied in detail for Antarctic Colossendeis species by Cano & Lopez Gonzalez (2007). My studies agree with these published results in that the spine configuration of the strigilis appears to be diagnostic at the species level and hence useful for distinguishing otherwise quite similar species such as C. robusta and C. glacialis. It appears that in the ground plan of the longitarsal colossendeids, the strigilis consists from ectal to endal of one row of long spines, one of medium spines, and a field of short spines not clearly placed into rows. This arrangement is preserved in C. megalonyx, C. glacialis, C. tenera and other species, but in C. angusta as well as in C. robusta and relatives it was independently simplified by the loss of a distinct row of medium spines.

202 Dissertation Lars Dietz

Morphological differences within C. megalonyx have been studied by Spaak (2010). She found significant differences in the morphometric measurements between each of the examined clades (A, B, C, D1, E). The results seem to contradict those presented here according to which some of those clades, particularly D1 and E, are not actually distinct based on nuclear data. However, all specimens that were examined morphologically by Spaak (2010) originated from West Antarctica, as the East Antarctic individuals were not available yet. The West Antarctic populations of clades D1 and E are geographically isolated from each other, as the former are restricted to South Sandwich and Bouvet Island and the latter to the southern Scotia Arc, so the differences may be explained by intraspecific geographic variation. This could be tested by including Clade D1 and E specimens from Terre Adélie in the morphometric analysis, as these clades occur sympatrically in that region. Nevertheless, the results of Spaak (2010) do show that different distantly related clades occurring sympatrically show significant morphological differences, supporting the idea that more than one species is included within the C. megalonyx complex. There are also morphological differences between C. megalonyx clades that are already evident after superficial inspection, such as the lack of eyes for the Bouvet Island endemic clade C, distinguishing it from the sympatric clade E, and the very large body size in the widespread but rare clade F. Turpaeva & Rajsky (2013) suggested a subspecific classification of C. megalonyx based on morphological data. While this classification also includes some taxa that do not belong to C. megalonyx according to molecular data (C. tenera, C. scoresbii), the validity of the remaining subspecies and the suggested distinguishing characters should be tested with additional data complemented by molecular information.

Mitochondrial vs. nuclear data One central result of my thesis is the compilation of a large CO1 dataset of Southern Ocean Colossendeidae, which will be useful for species identification of further samples by DNA barcoding. The phylogenetic analysis based on that dataset contributes to our knowledge of the phylogenetic history of Colossendeidae, which so far has been little studied. The results suggest that the longitarsal group of the genus Colossendeis, including most shallow-water Antarctic species, is monophyletic, while the brevitarsal group, which includes mostly deep- sea species both inside and outside the Southern Ocean with very short distal leg articles, is not clearly supported as monophyletic. The taxon sampling for the brevitarsal group is rather small, but its paraphyly would be consistent with the fact that in some of the more basal colossendeids (genera Rhopalorhynchus and Hedgpethia) the three distal leg articles are also very short. Within the brevitarsal group, only limited resolution is provided, and the only robustly supported groupings are C. glacialis as sister to all others and C. australis as sister

203 Dissertation Lars Dietz to all species except for C. glacialis. A clearly supported monophyletic grouping is formed of C. robusta and morphologically similar species like C. bouvetensis and C. drakei, and another clade is formed by Decolopoda and Dodecolopoda. The finding of Nakamura et al. (2007) and Krabbe et al. (2010) that the genera Decolopoda and Dodecolopoda, which have an increased number of walking legs, group inside the genus Colossendeis is confirmed. This leads to the nomenclatural problem that the name Decolopoda Eights 1835 is older than Colossendeis Jarzynsky 1870. As the solutions of renaming all Colossendeis species as Decolopoda or breaking up the genus Colossendeis into many poorly distinguishable genera do not appear to be practicable, the best solution would probably be not to change the nomenclature despite the paraphyly of the genus Colossendeis. The reconstructed phylogenetic history based on the nuclear gene region ITS is mostly congruent with the CO1 tree topology, especially in the positions of C. glacialis and C. australis. Within the remaining species, the resolution is only improved if incomplete sequences are excluded, in which case the C. robusta group is sister to the others and C. scotti is sister to the multi-legged forms. As the longitarsal group is mostly Antarctic, the analysis suggests that the non-Antarctic longitarsal forms such as the C. tenera/angusta group dispersed out of Antarctica. However, to test this suggestion it would be necessary to include more samples from non-Antarctic longitarsal colossendeids, which may include the type species of the genus, C. proboscidea. The CO1 data seem to perform well at recognizing major species groups within the Colossendeidae, as the resulting clusters agree with both morphological identification and with the ITS data. The Antarctic taxa which are represented by several specimens in our database and clearly identified by CO1 sequences include Colossendeis australis, C. scotti, C. tortipalpis and Decolopoda australis. In some of these taxa, a detailed integrative study including molecular and morphological data could test the presence of additional unrecognized diversity that is suggested by the CO1 data. In C. scotti, the CO1 sequences fall into three highly distinct clusters. Three clusters also appear in D. australis, which was here taken to include the supposed synonym D. antarctica whose status has often been disputed (Bouvier 1913, Gordon 1932, Fry & Hedgpeth 1969). In the C. tortipalpis group, the situation appears to be more complex, and the distances are higher than it would be expected in a single species. However, it is not possible to clearly give a fixed number of species-level clusters, suggesting a complex species status in this group. The morphologically similar species C. longirostris and C. enigmatica may be part of this group, and their species status should be re-investigated with morphological and molecular methods. The CO1 and ITS data were also successful in identifying species within the C. robusta group, agreeing with the morphological data as discussed above. Within the species C. glacialis, C. robusta and C. bouvetensis, groupings with limited geographical distribution were found with CO1, and these clades also exhibit some significant morphological

204 Dissertation Lars Dietz differences in cases where this could be investigated. If species are narrowly delimited, they may therefore be regarded as distinct species. However, ITS data did not resolve these clusters, which may mean that the resolution provided by ITS is not enough to resolve recent differentiation, but it also means that the possibility of ongoing low-level gene flow cannot be excluded and recognizing these clades as distinct species would be premature. With respect to C. bouvetensis, it is also possible that some of those clades for which no morphological or ITS data were available represent the similar but clearly distinct species C. lilliei or C. wilsoni. However, no molecular data are available for those two species. The C. megalonyx group could also be clearly delimited with CO1 and ITS, agreeing with morphological determinations. The only exceptions were the individuals from Kerguelen, which were morphologically determined as C. megalonyx but did not cluster within that grouping in the CO1 tree. This suggests that supposed C. megalonyx specimens from Kerguelen might represent another species and calls into question the reported occurrence of C. megalonyx in that location, showing the need for detailed restudy of specimens from Kerguelen. Within C. megalonyx, the CO1 and ITS data often show incongruent results. There is support for some larger groupings such as that of clades A+G+H+I, or D+E. Many of these groupings occur sympatrically, e.g. in the southern Scotia Arc without interbreeding, suggesting that they are largely reproductively isolated and can be regarded as distinct species. The same groups are also supported by the nuclear gene Histone 3 (H3), which however provides no resolution within the groups. Therefore, mitochondrial and nuclear data as well as morphological data for a limited number of samples (Spaak 2010, as discussed above) support the idea that C. megalonyx as usually understood is a species complex consisting of at least 5-7 species. Further morphological comparisons are needed to test the validity of species currently considered synonymous such as C. frigida, C. rugosa, C. arundorostris and several species described by Pushkin (1993) as well as the subspecies proposed by Turpaeva & Rajsky (2013). However, limited hybridization between the groups here identified as likely species appears to exist. The clearest example is one station in the Eastern Antarctic Peninsula, where specimens belonging to three distantly related CO1 clades have nearly identical ITS sequences. This shows that limited gene flow between species-level clades is ongoing. According to a strict interpretation of the Biological Species Concept, they should therefore not be designated as distinct species. However, many cases of gene flow between undisputed species in the wild are known (Koblmüller et al. 2007, Alves et al. 2008, Darling 2011; see Toews & Brelsford 2012 for a review). Gene variants that are adaptive for a certain environment can be fixed quickly after the introgression from one species into another (Hedrick 2013), and it is conceivable that this has occurred in the Antarctic marine environment, where the climate changes relatively rapidly on a geological timescale. A

205 Dissertation Lars Dietz possible candidate for adaptive introgression would be the mitochondrial DNA of clade E on Bouvet Island, where all individuals have the same CO1 haplotype, although ITS data suggest that the island shelf was colonized multiple times. The hybridization between distantly related mitochondrial clades (more than 3% genetic distance in CO1) raises the question how the mitochondrial differentiation evolved without mechanisms of reproductive isolation. The most plausible explanation seems to be that the mitochondrial clades represent lineages that evolved in geographical isolation from each other for a long time but were able to hybridize when coming into contact again due to lack of reproductive isolation. This can be explained by survival of populations in different isolated refugia during the earlier Pleistocene glaciations, followed by secondary contact and hybridization of those lineages during interglacials and the current postglacial.

Population history of colossendeid species Population history was investigated here mainly with CO1 data. The results may be incomplete as nuclear genes may have a different history than the mitochondrial gene CO1, and this is the case in C. megalonyx as discussed above. However, events such as sudden changes in population size and the presence or absence of frequent gene flow between populations should be testable using a single gene if the sampling size is large enough and the rate of evolution of the gene is not too slow. My results show that such inferences are possible in C. megalonyx with CO1 in those clades where the sampling size is large enough. There is good evidence for geographic differentiation between East and West Antarctic populations of C. megalonyx. This is obvious from the haplotype networks of clades D1, E and I which show separation into two distinct subgroups separated by 5 to 10 substitutions. The same pattern appears in other clades with a smaller sample size such as clades G or H, and also in C. robusta and C. bouvetensis. For the large clades, it is further corroborated by the AMOVA results showing a significant degree of differentiation between western and eastern populations, by Geneland results showing two distinct geographical clusters, and by support for an early East-West spilt in the ABC results. This suggests that the eastern and western populations survived in different refugia during the Last Glacial Maximum (LGM), and the dates for the East/West split estimated by ABC are consistent with this. Due to the low mobility of pycnogonids in general, gene flow between geographically distant regions would be restricted, as also shown by the results of Arango et al. (2011) for Nymphon australe and Weis & Melzer (2012) for Achelia assimilis. Nevertheless, there is some evidence for recent dispersal over wide geographical distances. One C. megalonyx clade E haplotype is found in Terre Adélie as well as in one specimen from the Eastern Antarctic Peninsula. The CO1 sequences of Bouvet Island specimens of C. bouvetensis are very close to those from the Ross Sea, and within this species haplotypes shared by Ross and Weddell

206 Dissertation Lars Dietz

Sea individuals are also found. This suggests that long-distance dispersal is possible but rather infrequent in colossendeids. Pycnogonids in general could have been transported by the Antarctic Circumpolar Current, either swimming (see King 1973 for a description of swimming in pycnogonids) or attached to drifting material such as algae. Another possibility especially for Ross/Weddell Sea connections is the presence of a Transantarctic seaway during previous interglacials (Barnes & Hillenbrand 2010), which however would not explain recent dispersal events as suggested by identical haplotypes. Identical haplotypes in geographically distant regions are also known from other Antarctic benthic invertebrates, such as Eusirus amphipods (Baird et al. 2011). Within the larger geographic regions, especially Scotia Arc, the best-studied region, only very weak population differentiation between different locations can be found in C. megalonyx.

Tests for differentiation such as FST did not give any significant result. The same also appears to be true of C. robusta according to the limited available data. This suggests that either there is ongoing exchange between populations from these regions or they have been colonized too recently for detectable genetic differentiation to form. The hypothesis of frequent exchange seems to be unlikely due to the low mobility of the animals. It is also contradicted by the very different frequencies of the clades in different island groups. As an example, clade A is by far the most frequent in the South Sandwich Islands and Elephant Island, but is much rarer in the South Orkneys and especially in the South Shetlands. The hypothesis of recent colonization would suggest that the population originated from a single refugium. In this case, there would probably have been a strong increase of the population size, which is supported by my data (discussed below). The best evidence for a single- refugium origin is in clade A, where the CO1 data suggest that all of the Scotia Arc was colonized from South Georgia. This provides a parallel to the very similar pattern in the gastropod Nacella concinna (González-Wevar et al. 2013). There is no evidence that South Georgia also served as a refugium for other clades, as they are not known from this location (except for the poorly sampled clade F). The results also contrast with those for C. glacialis, where strong differentiation between sequences from South Georgia and South Sandwich is found. However, in the light of the ITS results, the possibility must be considered that the clade A situation represents not a colonization of previously uninhabited locations, but an introgression of possibly adaptive genes from the South Georgian population into preexisting southern populations. The same also applies to the population expansions found in other clades. To test these scenarios, additional information, preferably large-scale genomic data, would be needed. Population genetic analyses provide good evidence that populations of different clades within the C. megalonyx complex underwent expansion in the past. The characteristic “star-like” haplotype networks, the ABC and Bayesian skyline analyses, and neutrality tests such as

207 Dissertation Lars Dietz

Tajima’s D and Fu’s Fs all agree on this. These expansions cannot be reliably dated, but a time of expansion of around 10,000-50,000 years ago seems to be consistent with both ABC and Bayesian skyline results. Therefore, the expansion may be dated to the end of the Last Glacial Maximum, as deglaciation of the West Antarctic ice shield started between 14,000 and 15,000 years ago (Clark et al. 2009). This would have opened up many formerly ice- covered habitats for the recolonization by benthic organisms. The results agree with the data on other benthic invertebrates, which also show a strong signature of population expansion, such as nemerteans (Thornhill et al. 2008), shrimps (Raupach et al. 2010), echinoids (Díaz et al. 2011), octopods (Strugnell et al. 2012), crinoids (Hemery et al. 2012), and gastropods (González-Wevar et al. 2013). However, except in the cases of Pareledone turqueti (Strugnell et al. 2012) and Nacella concinna (González-Wevar et al. 2013), the results have generally not been tested with methods that attempt to reconstruct the change in population size in detail. Also, in most cases, the probability of different scenarios of colonization and population expansion has not been tested. This study therefore provides one of the first attempts to investigate the population history of Antarctic invertebrates using Bayesian methods. It is likely that in previous glacial/interglacial cycles, similar scenarios as after the LGM would have occurred. It has been shown that a population bottleneck at the LGM can erase signals of previous population expansion/contraction cycles (Grant et al. 2012). The isolation of populations in refugia for tens of thousands of years during glaciations could have led to the differentiated CO1 lineages that are found in C. megalonyx today, which are shown to probably not be cryptic species according to the ITS data, suggesting that no barriers to hybridization evolved during isolations. As population sizes in the refugia probably were small, they can be assumed to have been genetically rather homogeneous. One of the goals of the thesis was to clarify the location of the refugia in which C. megalonyx survived the Pleistocene glaciations. While exact locations of refugia could not be identified, except for South Georgia in clade A, the evidence supports strong population expansion and geographic differentiation, which argues against survival in the circum-Antarctic deep sea as this would have led to a higher population size during the glaciations and less geographic variation, resulting in “diffuse” haplotype networks (Allcock & Strugnell 2012) as seen e.g. in the shrimp Nematocarcinus lanceopes (Raupach et al. 2010). The hypothesis of Subantarctic refugia would imply that the variation found in the Antarctic is a subset of the variation found in the Subantarctic place of origin. For most Antarctic clades investigated here, this is also not supported, as they are not represented in the Subantarctic. Instead, the Subantarctic may have been colonized from the Antarctic, as the South American C. megalonyx clades are not basal within the species complex but group within the Antarctic clades. This may have happened during earlier periods of global cooling, when cold-adapted Antarctic species

208 Dissertation Lars Dietz were able to colonize more northern regions. As discussed above, this may also apply to non-Antarctic brevitarsal Colossendeis species in general. Similar results are known from other organisms, such as gastropods (Göbbeler & Klussmann-Kolb 2010, González-Wevar 2014) and octopods (Strugnell et al. 2008). The only exception to the pattern, as discussed above, is clade A, for which a clear signature of recolonization of the other Scotia Arc island groups from South Georgia is found. South Georgia was probably partially ice-free during the LGM (Hodgson et al. 2014) and the position of the island in the path of the Antarctic Circumpolar Current may have created temporary ice-free regions east of the island. In other cases, the refugia were probably located in different regions on the Antarctic shelf. Evidence from other sources, such as the survival of terrestrial organisms, also shows that temporary regions temporarily free of grounded ice must have existed on the Antarctic continent and shelf (Convey & Stevens 2007, Convey et al. 2009), including areas that were kept ice-free by geothermal activity (Fraser et al. 2014). This is also supported by the existence of many endemic Antarctic species that do not occur in the deep sea, including pycnogonid species such as Colossendeis scotti (Munilla & Soler Membrives 2009).

Outlook: Opportunities for future research The work in this thesis investigates the taxonomy and population history of Southern Ocean pycnogonids using only two gene loci. While this made it possible to make inferences on the taxonomic status of several species and the history of Antarctic pycnogonids during and after the LGM, the work still leaves several open questions. For instance, the CO1 data suggest the presence of unrecognized diversity in several Colossendeis species not examined in detail here, which could be tested with additional molecular and morphological data. Within C. megalonyx, the data suggest that the different clades went through a population size bottleneck, presumably during the LGM, but it cannot be shown whether any special adaptations allowed them to survive these conditions or to expand after deglaciation. Also, if the non-Antarctic clades originated in the Antarctic, they would probably have developed detectable adaptations to the warmer climate. The sympatric coexistence of several apparently cryptic species is also only possible if they exhibit ecological differences so that competition is reduced. Such adaptative differences may be found in tolerance for abiotic factors such as temperature, depth, oxygen content or pH of the seawater, in preferred types of seafloor or in feeding preferences. Stout & Shabica (1971) investigated the feeding preferences of several species of Antarctic pycnogonids and found that, among colossendeids, Decolopoda australis and Colossendeis robusta were observed feeding primarily in soft sediments while C. megalonyx and C. australis preferred tubicolous polychaetes. Similar differences in food preference could also exist between different clades of C. megalonyx. Colossendeids are also known to feed on sea anemones (Braby et al.

209 Dissertation Lars Dietz

2009) and hydroids (Dayton et al. 1970). In a preliminary study of transcriptomic data for two specimens of C. megalonyx clade D3 from the Amundsen Sea, 18S and CO1 sequences belonging to sponges, echinoids and marine oligochaetes were found by BLAST search in addition to sequences from the target organism. These studies could be extended by targeted next-generation sequencing of gut contents of specimens from different clades to investigate their food preferences, as done e.g. by Leray et al. (2013) for fish. Adaptive differences between species may also be reflected in morphology. Although unrecognized species-level diversity is often called “cryptic”, detailed investigation often reveals morphological differences (e.g. Brandao et al. 2010) rendering the investigated species pseudo-cryptic, as also seen in C. robusta. There may be differences e.g. in the proportions of leg articles as adaptations to walking on different grounds, and such differences within the C. megalonyx complex were already found by Spaak (2010). The different feeding preferences may have led to differences in the form and length of the proboscis, which are well known between colossendeid species, and also its internal anatomy. The mouth opening is surrounded by three setose “lips”, whose morphology is variable within the Pycnogonida (Fahrenbach & Arango 2014), and the proximal part of the proboscis is occupied by a filter apparatus whose morphology and extent also varies (Dohrn 1881, Soler-Membrives et al. 2013). The proboscis musculature also differs between taxa (Fry 1965). Detailed microscopic comparisons should clarify whether there are differences in these characters within the C. megalonyx complex. There may also be physiological or biochemical differences between so-called cryptic species, as shown by Wilson et al. (2013) for the sea slug Doris kerguelenensis. Genomic studies may have the potential to identify adaptive differences within or between species. By sequencing thousands of loci, it is possible to test whether any of them show signatures of selection, such as significantly higher genetic homogeneity (as measured by standard fixation indices such as FST) in the case of directional or stabilizing selection, or lower homogeneity for balancing selection. This has been done to identify gene loci that were under particular selection e.g. in the evolution of humans (Nielsen et al. 2007). If genomic data are available for several closely related species, it is also possible to test for adaptive introgression (Hedrick 2013), i.e. the spreading of advantageous gene variants from one species into another by hybridization. While sequencing a complete genome is still costly and time-intensive, next-generation methods that allow the generation of hundreds of thousands of sequences at once may also be useful for these purposes, even if they cover only a small subset of the genome. The 454- data studied here have already been used to sequence the complete mitochondrial genome of C. megalonyx (Dietz et al. 2011), as well as almost complete mitogenomes for the other two pycnogonids included in the dataset, with only small gaps remaining (L. Dietz,

210 Dissertation Lars Dietz unpublished data). Additionally, nuclear loci such as the complete ribosomal operon and Histone 3 could also be extracted, showing the use of these data in phylogenetics. The microsatellites found in the data could also be useful for future population genetic work if they can be successfully amplified for a sufficient number of specimens. However, the data also show that contamination from bacteria and other non-target organisms is to be expected, and such sequences must be carefully filtered and excluded prior to performing analyses. This is especially problematic in organisms with relatively small genomes such as pycnogonids (Gregory 2015). As sequencing a complete genome is costly and non-targeted next-generation sequencing may not yield large numbers of homologous loci between individuals, it is often more effective to sequence reduced-representation libraries, in which only particular regions of the genome are selected and sequenced with next-generation techniques. One useful method for this is restriction-site associated DNA sequencing (RADseq; Miller et al. 2007). In this approach, only loci that include a specific restriction enzyme cutting site are amplified and sequenced. RADseq has been shown to be useful in inferring phylogeny and phylogeography at scales that could not be resolved by conventional approaches, including postglacial recolonization (Emerson et al. 2010). RADseq has also shown to be useful for identifying signatures of selection in the genome and comparing them across populations, e.g. in stickleback (Hohenlohe et al. 2010). As a modified version of RAD (ddRAD, Peterson et al. 2012) sequencing results for four C. megalonyx specimens already exist, the technique seems to be functioning in principle for pycnogonids. Another useful technique for generating large amounts of data useful for population genetics and identifying genes under selection is transcriptomics, in which RNA is extracted from an organism and complementary DNA (cDNA) is generated for it, which is then sequenced by next-generation techniques. By this process, only genome regions that are actually transcribed are sequenced, which reduces the amount of non-coding DNA sequences. Transcriptomes are already published for two pycnogonid species (Dunn et al. 2008, Meusemann et al. 2010), and, as mentioned above, C. megalonyx transcriptomes also have been sequenced (provided by A. Mahon). One disadvantage of transcriptomics is that RNA usually degrades very quickly, and therefore samples fixed in any of the usual media such as ethanol are unsuitable and only samples that were fixed in a special medium immediately after collection can be used. Colossendeis transcriptomes will also be useful for arthropod phylogeny reconstructions in general, as the Colossendeidae appear to be relatively basal within the Pycnogonida (Arango & Wheeler 2007, Dietz et al. 2011) and pycnogonid position within the Arthropoda is still not completely resolved (see e.g. Regier et al. 2010).

211 Dissertation Lars Dietz

References Allcock AL, Strugnell JM (2012): Southern Ocean diversity: new paradigms from molecular ecology. Trends in Ecology and Evolution 27: 520-528. Alves PC, Melo-Ferreira J, Freitas H, Boursot P (2008): The ubiquitous mountain hare mitochondria: multiple introgressive hybridization in hares, genus Lepus. Philosophical Transactions of the Royal Society of London B 363: 2831-2839. Arango CP, Brenneis G (2013): New species of Australian Pseudopallene (Pycnogonida: Callipallenidae) based on live colouration, morphology and DNA. Zootaxa 3616: 401- 436. Arango CP, Soler-Membrives A, Miller KJ (2011): Genetic differentiation in the circum— Antarctic sea spider Nymphon australe (Pycnogonida; Nymphonidae). Deep Sea Research Part II: Topical Studies in Oceanography 58: 212-219. Arango CP, Wheeler WC (2007): Phylogeny of the sea spiders (Arthropoda, Pycnogonida) based on direct optimization of six loci and morphology. Cladistics 23: 255–293. Baird HP, Miller KJ, Stark JS (2011): Evidence of hidden biodiversity, ongoing speciation and diverse patterns of genetic structure in giant Antarctic amphipods. Molecular Ecology 20: 3439-3454. Barnes DKA, Hillenbrand CD (2010): Faunal evidence for a late quaternary trans-Antarctic seaway. Global Change Biology 16: 3297-3303. Bouvier EL (1913): Pycnogonides du "Pourquoi-Pas?” Deuxième expédition antarctique française (1908-1910) commandée par le Dr. Jean Charcot. 1-169. Braby CE, Pearse VB, Bain BA, Vrijenhoek RC (2009): Pycnogonid-cnidarian trophic interactions in the deep Monterey Submarine Canyon. Invertebrate Biology 128: 359- 363. Brandão SN (2010): Macrocyprididae (Ostracoda) from the Southern Ocean: taxonomic revision, macroecological patterns, and biogeographical implications. Zoological Journal of the Linnean Society 159: 567-672. Cano E, Lopez-Gonzalez PJ (2007): Colossendeis species (Pycnogonida: Colossendeidae) collected during the Italica XIX cruise to Victoria Land (Antarctica), with remarks on some taxonomic characters of the ovigers. Scientia Marina 71: 661-681. Child CA (1995): Antarctic and Subantarctic Pycnogonida: The Families Nymphonidae, Colossendeidae, Rhynchothoraxidae, Pycnogonidae, Endeididae, and Callipallenidae. Antarctic Research Series 69: 1-165. Clark PU, Dyke AS, Shakun JD, Carlson AE, Clark J, Wohlfarth B, Mitrovica JX, Hostetler SW, McCabe AM (2009): The Last Glacial Maximum. Science 325: 710-714. Convey P, Stevens MI (2007): Antarctic biodiversity. Science 317: 1877-1878.

212 Dissertation Lars Dietz

Convey P, Stevens MI, Hodgson DA, Smellie JL, Hillenbrand CD, Barnes DKA, Clarke A, Pugh PJA, Linse K, Cary SC (2009): Exploring biological constraints on the glacial history of Antarctica. Quaternary Science Reviews 28: 3035-3048. Darling JA (2011): Interspecific hybridization and mitochondrial introgression in invasive Carcinus shore crabs. PLoS ONE 6: e17828. Dayton PK, Robilliard GA, Paine RT (1970): Benthic faunal zonation as a result of anchor ice at McMurdo Sound, Antarctica. Antarctic Ecology 1: 244-257. Díaz A, Féral JP, David B, Saucède T, Poulin E (2011): Evolutionary pathways among shallow and deep-sea echinoids of the genus Sterechinus in the Southern Ocean. Deep Sea Research Part II: Topical Studies in Oceanography 58: 205-211. Dietz L, Mayer C, Arango CP, Leese F (2011): The mitochondrial genome of Colossendeis megalonyx supports a basal position of Colossendeidae within the Pycnogonida. Molecular Phylogenetics and Evolution 58: 553-558. Dohrn A (1881): Die Pantopoden des Golfes von Neapel und der angrenzenden Meeresabschnitte. Fauna und Flora des Golfes von Neapel 3: 1-152. Dunn CW, Hejnol A, Matus DQ, Pang K, Browne WE, Smith SA, Seaver E, Rouse GW, Obst M, Edgecombe GD, Sorensen MV, Haddock SH, Schmidt-Rhaesa A, Okusu A, Kristensen RM, Wheeler WC, Martindale MQ, Giribet G (2008): Broad phylogenomic sampling improves resolution of the animal tree of life. Nature 452: 745–749. Eights J (1834): Description of a new animal belonging to the Arachnides of Latreille; discovered in the sea along the shores of the New South Shetland Islands. Boston Journal of Natural History 1: 203-206. Emerson KJ, Merz CR, Catchen JM, Hohenlohe PA, Cresko WA, Bradshaw WE, Holzapfel CM (2010): Resolving postglacial phylogeography using high-throughput sequencing. Proceedings of the National Academy of Sciences 107: 16196-16200. Fahrenbach W, Arango CP (2007): Microscopic anatomy of Pycnogonida: II. Digestive system. III. Excretory system. Journal of Morphology 268: 917-935. Fraser CI, Terauds A, Smellie J, Convey P, Chown SL (2014): Geothermal activity helps life survive glacial cycles. Proceedings of the National Academy of Sciences 111: 5634- 5639. Fry WG (1965): The feeding mechanisms and preferred foods of three species of Pycnogonida. Bulletin of the British Museum (Natural History), Zoology 12: 195-223. Fry WG, Hedgpeth JW (1969): The fauna of the Ross Sea. Part 7. Pycnogonida, 1: Colossendeidae, Pycnogonidae, Endeidae, Ammotheidae. New Zealand Department of Science and Industrial Research Bulletin 198: 1-139.

213 Dissertation Lars Dietz

Göbbeler K, Klussmann-Kolb A (2010): Out of Antarctica? - new insights into the phylogeny and biogeography of the Pleurobranchomorpha (Mollusca, Gastropoda). Molecular Phylogenetics and Evolution 55: 996-1007. González-Wevar CA, Chown SL, Morley S, Coria N, Saucéde T, Poulin E (in press): Out of Antarctica: quaternary colonization of sub-Antarctic Marion Island by the limpet genus Nacella (Patellogastropoda: Nacellidae). Polar Biology (advance online publication) González-Wevar CA, Saucède T, Morley SA, Chown SL, Poulin E (2013): Extinction and recolonization of maritime Antarctica in the limpet Nacella concinna (Strebel, 1908) during the last glacial cycle: toward a model of Quaternary biogeography in shallow Antarctic invertebrates. Molecular Ecology 22: 5221-5236. Gordon I (1932): Pycnogonida. Discovery Report 6: 1-138. Grant WS, Liu M, Gao TX, Yanagimoto T (2012): Limits of Bayesian skyline plot analysis of mtDNA sequences to infer historical demographies in Pacific herring (and other species). Molecular Phylogenetics and Evolution 65: 203-212. Gregory TR (2015) Animal Genome Size Database. www.genomesize.com, accessed 2015/04/09. Hedrick PW (2013): Adaptive introgression in animals: examples and comparison to new mutation and standing variation as sources of adaptive variation. Molecular Ecology 22: 4606-4618. Hemery LG, Eléaume M, Roussel V, Améziane N, Gallut C, Steinke D, Cruaud C, Couloux A, Wilson NG (2012): Comprehensive sampling reveals circumpolarity and sympatry in seven mitochondrial lineages of the Southern Ocean crinoid species Promachocrinus kerguelensis (Echinodermata). Molecular Ecology 21: 2502-2518. Hodgson DA, Graham AGC, Roberts SJ, Bentley MJ, Ó Cofaigh C, Verleyen E, Vyverman W, Jomelli V, Favier V, Brunstein D, Verfaillie D, Colhoun EA, Saunders KM, Selkirk PM, Mackintosh A, Hedding DW, Nel W, Hall K, Mcglone MS, Van Der Putten N, Dickens WA, Smith A (2014): Terrestrial and submarine evidence for the extent and timing of the Last Glacial Maximum and the onset of deglaciation on the maritime-Antarctic and sub- Antarctic islands. Quaternary Science Reviews 100: 137-158. Hoek PPC (1881): Report on the Pycnogonida, dredged by HMS Challenger during the Years 1873-76. Report on the Scientific Results of the Voyage of HMS Challenger, Zoology 3: 1-167. Hohenlohe PA, Bassham S, Etter PD, Stiffler N, Johnson EA, Cresko WA (2010): Population genomics of parallel adaptation in threespine stickleback using sequenced RAD tags. PLoS Genetics 6: e1000862.

214 Dissertation Lars Dietz

Jarzynsky T (1870): Praemissus catalogus Pycnogonidarum, inventarum in mari Glaciali, ad oras Lapponiae rossicae et in mari Albo, anno 1869 et 1870. Annales de la Société des Naturalistes de St. Petersbourg 1: 319–320. King PE (1973): Pycnogonids. Hutchinson, London. Koblmüller S, Duftner N, Sefc KM, Aibara M, Stipacek M, Blanc M, Egger B, Sturmbauer C (2007): Reticulate phylogeny of gastropod-shell-breeding cichlids from Lake Tanganyika - the result of repeated introgressive hybridization. BMC Evolutionary Biology 7: 7. Krabbe K, Leese F, Mayer C, Tollrian R, Held C (2010): Cryptic mitochondrial lineages in the widespread pycnogonid Colossendeis megalonyx Hoek, 1881 from Antarctic and Subantarctic waters. Polar Biology 33: 281-292. Leray M, Yang JY, Meyer CP, Mills SC, Agudelo N, Ranwez V, Boehm JT, Machida RJ (2013): A new versatile primer set targeting a short fragment of the mitochondrial COI region for metabarcoding metazoan diversity: application for characterizing coral reef fish gut contents. Frontiers in Zoology 10: 34. Meusemann K, von Reumont BM, Simon S, Roeding F, Strauss S, Kück P, Ebersberger I, Walzl M, Pass G, Breuers S, Achter V, von Haeseler A, Burmester T, Hadrys H, Wägele JW, Misof B (2010): A phylogenomic approach to resolve the arthropod tree of life. Molecular Biology and Evolution 27: 2451-2464. Miller MR, Dunham JP, Amores A, Cresko WA, Johnson EA (2007): Rapid and cost-effective polymorphism identification and genotyping using restriction site associated DNA (RAD) markers. Genome Research 17: 240-248. Munilla T, Soler Membrives A (2009): Check-list of the pycnogonids from Antarctic and sub- Antarctic waters: zoogeographic implications. Antarctic Science 21: 99-111. Nakamura K, Kano Y, Suzuki N, Namatame T, Kosaku A (2007): 18S rRNA phylogeny of sea spiders with emphasis on the position of Rhynchothoracidae. Marine Biology 153: 213- 223. Nielsen R, Hellmann I, Hubisz M, Bustamante C, Clark AG (2007): Recent and ongoing selection in the human genome. Nature Reviews Genetics 8: 857-868. Peterson BK, Weber JN, Kay EH, Fisher HS, Hoekstra HE (2012): Double digest RADseq: an inexpensive method for de novo SNP discovery and genotyping in model and non- model species. PLoS ONE 7: e37135. Pushkin AF (1993): The Pycnogonida fauna of the South Ocean (Biological results of the Soviet Antarctic Expeditions). Samperi, Messina. Raupach MJ, Thatje S, Dambach J, Rehm P, Misof B, Leese F (2010): Genetic homogeneity and circum-Antarctic distribution of two benthic shrimp species of the Southern Ocean, Chorismus antarcticus and Nematocarcinus lanceopes. Marine Biology 157: 1783-1797.

215 Dissertation Lars Dietz

Regier JC, Shultz JF, Zwick A, Hussey A, Ball B, Wetzer R, Martin JW, Cunningham WC (2010): Arthropod relationships revealed by phylogenomic analysis of nuclear protein- coding sequences. Nature 463: 1079–1084. Soler-Membrives A, Arango CP, Cuadrado M, Munilla T (2013): Feeding biology of carnivore and detritivore Mediterranean pycnogonids. Journal of the Marine Biological Association of the United Kingdom 93: 635-643. Spaak JM (2010): Integrative Taxonomie am Beispiel des Colossendeis megalonyx Hoek 1881 Artkomplexes (Chelicerata; Pycnogonida; Colossendeidae). MSc. Thesis, Ruhr University Bochum. Stout WE, Shabica SV (1970): Marine ecological studies at Palmer Station and vicinity. Antarctic Journal of the United States 5: 134-135. Strugnell JM, Rogers AD, Prodöhl PA, Collins MA, Allcock AL (2008): The thermohaline expressway: the Southern Ocean as a centre of origin for deep-sea octopuses. Cladistics 24: 853-860. Strugnell JM, Watts PC, Smith PJ, Allcock AL (2012): Persistent genetic signatures of historic climatic events in an Antarctic octopus. Molecular Ecology 21: 2775-2787. Thornhill DJ, Mahon AR, Norenburg JL, Halanych KM (2008): Open-ocean barriers to dispersal: a test case with the Antarctic Polar Front and the ribbon worm Parborlasia corrugatus (Nemertea: Lineidae). Molecular Ecology 17: 5104-5117. Toews DP, Brelsford A (2012): The biogeography of mitochondrial and nuclear discordance in animals. Molecular Ecology 21: 3907-30. Turpaeva EP (1973): Mnogokolencatye (Pantopoda) iz severo-zapadnoy casti Tikhogo Okeana. Trudy Instituta Okeanologii "P. P. Shirshova" Akademy Nauk SSSR - 91: 178- 191. Turpaeva EP, Rajsky AC (2013): Morskie pauki roda Colossendeis (Colossendeidae, Pycnogonida) morya Ueddella i prilezhashikh akvatorii. Journal of Siberian Federal University. Biology 2: 130-149. Weis A, Melzer RR (2012): How did sea spiders recolonize the Chilean fjords after glaciation? DNA barcoding of Pycnogonida, with remarks on phylogeography of Achelia assimilis (Haswell, 1885). Systematics and Biodiversity 10: 361-374. Wilson NG, Maschek JA, Baker BJ (2013): A species flock driven by predation? Secondary metabolites support diversification of slugs in Antarctica. PLoS One 8: e80277.

216 Dissertation Lars Dietz

9) Summary

In this thesis, I study the evolution and biodiversity of Southern ocean benthic organisms, mainly with molecular tools but also including morphological investigations. The work focuses on pycnogonids, a widespread group of benthic arthropods which are particularly species- rich in the Southern Ocean. My main focus is the genus Colossendeis, especially the C. robusta and C. megalonyx species groups, which have a circumpolar eurybathic distribution in the Antarctic and Subantarctic. The work attempts to resolve questions on their taxonomy, especially the possible presence of cryptic species, and their population history, especially the location of refugia where the species survived during Pleistocene glaciations. Paper I (Dietz et al., Arthropod Systematics and Evolution in press) investigates the controversial status of the Southern Ocean pycnogonid species Colossendeis robusta and C. glacialis. The results show that the previously suggested synonymy of these species is neither supported by molecular data from both mitochondrial and nuclear genes nor by detailed morphometric measurements and SEM data of the ovigeral strigilis, including data from specimens from the type locality of C. robusta (Kerguelen Island shelf). A third species was identified based on molecular and morphological evidence which has also been misidentified as C. robusta, but appears to be phylogenetically closer related to C. lilliei and C. wilsoni. This new species was formally described as part of this publication and named Colossendeis bouvetensis Dietz & Leese, 2015. The three species C. glacialis, C. robusta and C. bouvetensis all seem to have circumpolar distributions in the Antarctic shelf, and also occur on some Subantarctic islands. Within the different species, regional differentiation is also found, suggesting survival in more than one glacial refugium. In Paper II (Dietz et al., Organisms, Diversity and Evolution 2013) it is demonstrated based on molecular data (CO1 and H3) as well as morphometric measurements that the North Pacific species Colossendeis tenera is distinct from the Antarctic C. megalonyx. Instead, C. tenera is more closely related to the widely distributed deep-sea species C. angusta and its possible synonym C. gracilis. These results show that C. tenera does not represent an example of a non-Southern Ocean form with Antarctic origins, as known from other taxa such as octopuses. The data suggest that C. angusta/gracilis may have originated in the North Pacific. In Paper III (Dietz et al., Royal Society Open Science, under review) the population genetic structure within the C. megalonyx complex is investigated. The molecular sampling for C. megalonyx is increased to 491 CO1 and 57 ITS sequences from different parts of West and East Antarctica as well as surrounding islands. With the newly generated data, the number of COI clades regarded as potentially distinct species increases from six to 15-19 depending on the method used. These clades often have a circumpolar distribution in that they occur in

217 Dissertation Lars Dietz both West and East Antarctica, but most of them seem to have gaps in their distribution, as they are unknown from certain regions. Within the clades, differentiation between geographic regions, especially West and East Antarctica, is often found, suggesting survival in different refugia during the Pleistocene glaciations. In clade A, which is limited to Scotia Arc, the sequence diversity in South Georgia is much higher than in the more southern islands, suggesting that the former might have been a refugium. The ITS data only partially confirm the CO1 results. Most of the larger monophyletic groups formed by the CO1 clades, such as a grouping of clades D+E, are also recovered with ITS, but individual clades are not distinguished within that grouping. In a few cases even specimens belonging to different larger CO1 groupings cluster together in the ITS phylogeny reconstruction. The data show that individuals belonging to different CO1 clades - which normally would be considered distinct species - can freely hybridize with each other, and even hybridization between more distantly related clades appears to be present. We conclude that, while there probably are several distinct species within the C. megalonyx complex between which gene flow is rare but not nonexistent, not all of the clades identified by CO1 can be recognized as species. In Paper IV (Dietz et al., in prep.), hypotheses about population structure and history within the C. megalonyx clades A, D1, E and I are tested. Our data confirm that in all of these clades (except for clade A, which is only known from Scotia Arc) significant differentiation between western (Scotia Arc) and eastern (Eastern Weddell Sea or Terre Adélie) populations is present, while no significant differentiation can be found between the Scotia Arc island groups, except for South Georgia vs. other islands in clade A. This suggests that the southern Scotia Arc islands were colonized only recently, presumably after the Last Glacial Maximum, and from a different refugium than the eastern regions. For groups with a large sampling size, a relatively recent increase in population size was suggested both by Bayesian skyline plot analyses and Approximate Bayesian Computations (ABC). Although the ABC results were not conclusive on which exact scenario should be preferred, they mostly support the hypothesis of East/West differentiation and recent population expansion. Paper V (Weis et al., Zoological Journal of the Linnean Society 2014) shows that the widespread pycnogonid species Pallenopsis patagonica as currently recognized consists of several clades that are highly distinct according to mitochondrial data. All of them have restricted geographical distribution ranges. Some of these clades are more closely related to other recognized species such as P. macneilli. Detailed morphological comparison especially of animals from the Chilean coast with those from the type locality near the Falklands, including type material, shows that both groups are morphologically distinguishable, and the Chilean clade is therefore distinguished as a new species, Pallenopsis yepayekae Weis 2014. The Antarctic individuals also belong to separate mitochondrial clades which may also be distinct species, but this has not yet been tested morphologically.

218 Dissertation Lars Dietz

In Paper VI (Leese et al., PLoS One 2012), 454 pyrosequencing data for different organisms, including three pycnogonids, are analyzed. Depending on genome size, the libraries were found to cover about 0.02 to 20% of the genome. As known pycnogonid genome sizes are rather small, the coverage for pycnogonids is about 2-20% of the genome. Mitochondrial reads covering almost the entire mitochondrial genome could be extracted in pycnogonids. A large number of microsatellite loci potentially useful for population genetics could be found in all examined taxa. However, it was also discovered that each library contained DNA of non- target organisms such as bacteria, showing that 454 data should be checked for contamination before being used e.g. in phylogenetics. Among the identified bacteria there are many cold-adapted forms, as can be expected in the Southern Ocean.

In conclusion, this thesis highlights the potential of integrative studies including molecular data from independent sources as well as morphological and geographical data for studying the taxonomy and population history of Southern Ocean marine biodiversity using pycnogonids as an example taxon. The work clarifies some questions on the evolutionary history of pycnogonids, but also identifies problems that can only be solved by further work such as analyses of large amounts of genomic data.

219 Dissertation Lars Dietz

10) Zusammenfassung

In meiner Dissertation wird die Evolution und Biodiversität benthischer Organismen des Südpolarmeeres hauptsächlich mit molekularen Methoden, aber auch unter Berücksichtigung von morphologischen Daten untersucht. Hauptthema der Arbeit sind die Asselspinnen (Pycnogonida), eine weit verbreitete benthische Arthropodengruppe, die im Südpolarmeer besonders artenreich ist. Der Hauptfokus liegt auf der Gattung Colossendeis, vor allem die Artengruppen C. robusta und C. megalonyx, die zirkumpolar und eurybathisch in der Antarktis und Subantarktis verbreitet sind. Die Arbeit behandelt Fragen zur Taxonomie dieser Arten, insbesondere die mögliche Existenz kryptischer Arten, und zu ihrer Populationsgeschichte. Insbesondere wird untersucht, in welchen Refugien diese Arten die pleistozänen Eiszeiten überlebten. Publikation I (Dietz et al., Arthropod Systematics and Evolution im Druck) behandelt den kontroversen Status der antarktischen Asselspinnenarten Colossendeis robusta und C. glacialis. Die Ergebnisse zeigen, dass die früher vorgeschlagene Synonymie dieser Arten weder durch molekulare Daten von mitochondriellen und nukleären Genen noch durch morphometrische Messungen und REM-Daten der Beborstung des Oviger gestützt wird, wobei auch Exemplare von der Typuslokalität von C. robusta, dem Schelf der Kerguelen- Inseln, berücksichtigt werden. Eine dritte Art, die ebenfalls als C. robusta fehlidentifiziert wurde, aber näher mit C. lilliei und C. wilsoni verwandt zu sein scheint, wurde aufgrund von molekularen und morphologischen Daten identifiziert. Diese neue Art wurde als Teil diese Publikation unter dem Namen Colossendeis bouvetensis Dietz & Leese, 2015 formell beschrieben. Die Arten C. glacialis, C. robusta und C. bouvetensis scheinen alle auf dem antarktischen Schelf zirkumpolar verbreitet zu sein und kommen auch auf einigen subantarktischen Inseln vor. Innerhalb der einzelnen Arten fanden wir auch regionale Differenzierung, die auf Überleben in mehr als einem glazialen Refugium hindeutet. In Publikation II (Dietz et al., Organisms, Diversity and Evolution 2013) wird aufgrund von molekularen Daten (CO1 und H3) sowie morphometrischen Messungen gezeigt, dass die nordpazifische Art Colossendeis tenera nicht als Synonym zur antarktischen C. megalonyx anzusehen ist. Stattdessen ist C. tenera näher verwandt mit der weit verbreiteten Tiefseeart C. angusta und ihrem möglichem Synonym C. gracilis. Diese Ergebnisse zeigen, dass C. tenera kein Beispiel einer nicht-antarktischen Form antarktischen Ursprungs ist, wie dies von anderen Taxa wie z. B. Kraken bekannt ist. Der Schluss liegt nahe, dass C. angusta/gracilis im Nordpazifik entstanden sein könnte. In Publikation III (Dietz et al., Royal Society Open Science, in Begutachtung) wird die populationsgenetische Struktur innerhalb des C. megalonyx-Komplexes untersucht. Die molekulare Datenmenge für C. megalonyx wird auf 491 CO1- und 57 ITS-Sequenzen von

220 Dissertation Lars Dietz verschiedenen Teilen der West- und Ostantarktis und umliegenden Inseln erhöht. Mit den neu generierten Daten erhöht sich die Anzahl der CO1-Kladen, die als mögliche distinkte Arten bewertet werden, von sechs auf 15-19 je nach verwendeter Methode. Diese Kladen haben oft eine zirkumpolare Verbreitung, da sie sowohl in der West- als auch Ostantarktis vorkommen. Die meisten Kladen aber scheinen eine lückenhafte Verbreitung zu haben, da sie aus bestimmten Regionen nicht bekannt sind. Innerhalb der Kladen wird oft eine Differenzierung zwischen verschiedenen geographischen Regionen, insbesondere West- und Ostantarktis, gefunden, was auf Überleben in verschiedenen Refugien während der pleistozänen Eiszeiten hindeutet. Bei Klade A, die auf den Scotia-Bogen beschränkt ist, ist die Sequenzdiversität in Südgeorgien viel höher als bei weiter südlich gelegenen Inseln, was dafür spricht, dass Südgeorgien ein Refugium war. Die ITS-Daten bestätigen die Ergebnisse der CO1-Daten nur teilweise. Die meisten größeren monophyletischen Gruppen, die von den CO1-Kladen gebildet werden, z. B. eine Gruppierung der Kladen D und E, werden auch durch ITS bestätigt. Innerhalb dieser Gruppierungen werden die einzelnen Kladen jedoch nicht unterschieden. In einigen Fällen gruppieren in der ITS-Topologie sogar Individuen zusammen, die mit CO1 zu verschiedenen größeren Gruppen gestellt werden. Die Daten zeigen, dass Tiere, die zu verschiedenen CO1-Kladen gehören, die üblicherweise als Arten angesehen würden, frei miteinander hybridisieren können, und dass sogar Hybridisierung zwischen entfernter verwandten Kladen möglich zu sein scheint. Wir schließen, dass es zwar wahrscheinlich mehrere Arten innerhalb des C. megalonyx-Komplexes gibt, zwischen denen Genfluss selten, aber nicht völlig inexistent ist. Allerdings können nicht alle Kladen, die durch CO1 identifiziert wurden, als Arten anerkannt werden. In Publikation IV (Dietz et al., in Vorbereitung) werden Hypothesen zur Populationsstruktur und –geschichte der C. megalonyx-Kladen A, D1, E und I getestet. Unsere Daten bestätigen, dass alle diese Kladen (außer Klade A, die nur von der Scotia Arc bekannt ist) signifikante Differenzierung zwischen westlichen (Scotia Arc) und östlichen (Östliches Weddellmeer oder Terre Adélie) Populationen aufweisen. Zwischen den Inselgruppen des Scotia-Bogens kann jedoch keine signifikante Differenzierung gefunden werden, außer zwischen Südgeorgien und den anderen Inseln bei Klade A. Dies spricht dafür, dass die Inseln des südlichen Scotia-Bogens erst vor relativ kurzer Zeit kolonisiert wurden, vermutlich nach dem letzten glazialen Maximum, und von einem anderen Refugium aus als die östlichen Regionen. Bei Kladen mit großer Datenmenge wurde ein vor relative kurzer Zeit erfolgter Anstieg der Populationsgröße sowohl durch Bayesian Skyline Plot-Analysen als auch durch Approximate Bayesian Computations (ABC) gestützt. Obwohl die ABC-Ergebnisse nicht definitiv ein bestimmtes Szenario bevorzugen, unterstützen sie meist die Hypothese einer Ost-West- Differenzierung und eine vor relativ kurzer Zeit erfolgte Populationsexpansion.

221 Dissertation Lars Dietz

Publikation V (Weis et al., Zoological Journal of the Linnean Society 2014) zeigt, dass die weit verbreitete Asselspinnenart Pallenopsis patagonica im derzeit anerkannten Umfang nach mitochondriellen Daten aus mehreren deutlich getrennten Kladen besteht, die eine beschränkte geographische Verbreitung haben. Einige dieser Kladen sind näher mit anderen anerkannten Arten wie P. macneilli verwandt. Detaillierte morphologische Vergleichungen insbesondere von Tieren von der chilenischen Küste mit solchen von der Typuslokalität nahe der Falkland-Inseln (darunter auch Typusmaterial) zeigen, dass beide Gruppen morphologisch unterscheidbar sind. Die chilenische Klade wird daher als neue Art Pallenopsis yepayekae Weis 2014 beschrieben. Die antarktischen Tiere gehören ebenfalls zu eigenen mitochondriellen Kladen, die möglicherweise eigene Arten sind, was aber noch einer morphologischen Überprüfung bedarf. In Publikation VI (Leese et al., PLoS One 2012) werden 454-Pyrosequenzierungsdaten von verschiedenen Organismen, darunter drei Asselspinnen, analysiert. Je nach Genomgröße umfassten die 454-Libraries etwa 0.02-20% des Genoms. Da bekannte Genomgrößen bei Asselspinnen relativ klein sind, beträgt die Abdeckung für die Asselpinnen ca. 2-20% des Genoms. Mitochondrielle Sequenzen, die fast das gesamte mitochondrielle Genom umfassen, konnten bei Asselspinnen festgestellt werden. Bei allen Taxa wurde eine große Anzahl an Mikrosatelliten-Loci gefunden, die potenziell für Populationsgenetik nützlich sind. Allerdings wurden auch in allen Libraries Sequenzen von Nicht-Zielorganismen wie Bakterien gefunden, was zeigt, dass 454-Daten auf Kontamination überprüft werden sollten, bevor sie z. B. für Phylogenetik verwendet werden. Unter den identifizierten Bakterien sind viele kälteadaptierte Formen, wie dies im Südpolarmeer zu erwarten ist.

Zusammengefasst verdeutlicht diese Dissertation das Potential integrativer Studien, die molekulare Daten aus unterschiedlichen Quellen, aber auch morphologische und geographische Daten berücksichtigen, für die Untersuchung der Taxonomie und Populationsgeschichte der marinen Biodiversität des Südpolarmeeres und insbesondere der Asselspinnen. Die Arbeit klärt einige Fragen zur Evolutionsgeschichte der Asselspinnen, identifiziert aber auch Probleme, die nur durch weitere Forschungsarbeiten, z.B. durch Analysen großer Mengen genomischer Daten gelöst werden können.

222 Dissertation Lars Dietz

11) Acknowledgements

I thank Prof. Dr. Ralph Tollrian and Prof. Dr. Dominik Begerow for support of this thesis project. It was great to be able to work in Ralph Tollrian’s department and to finish my thesis here. I am especially thankful to Dr. Florian Leese for supporting my work in every way and for his very useful advice and the discussions with him on the subjects of this thesis. I thank Prof. Dr. Roland Melzer and Dr. Andrea Weis for the discussions we had on pycnogonids and other things and also for contributing specimens. I also thank Dr. Claudia Arango for her hospitality during my visit in Brisbane, for the interesting discussions on pycnogonids and also for providing samples. Also, I would like to thank Drs. Greg Rouse, Harim Cha and Nerida Wilson for the support during my visit to San Diego and for allowing me to take samples at the Scripps Institution. I also thank Prof. Dr. Ken Halanych and Prof. Dr. Andrew Mahon for the support during my visit to Auburn/Alabama and for contributing sequences. I thank everyone else who contributed samples or sequences for my studies, including Dr. Katrin Linse at the British Antarctic Survey and Dr. Laure Corbari at the Museum National d’Histoire Naturelle. Dr. Franz Krapp is thanked for his valuable support and advice on pycnogonid morphology. I would also like to thank Dr. Alex Weigand for his great advice and criticism which helped to bring this thesis into its final form. I also would like to thank everyone at the Department for Evolutionary Ecology and Biodiversity of Animals for contributing to the great atmosphere in this department which made we feel welcome. I thank the Deutsche Forschungsgemeinschaft for financial support for this work (Grant LE 2323/2 to Florian Leese, Christoph Mayer and Christoph Held). Finally, I would like to thank my parents for their patience and support during my writing of this thesis and my entire biology studies.

223