Groundwater Amphipods in Iceland: Population Structure and Phylogenetics

Groundwater amphipods in Iceland: Population structure and phylogenetics

Etienne Kornobis

Dissertation submitted in partial fulﬁllment of a Philosophiae Doctor degree in Biology

Advisor: Dr. Snæbjörn Pálsson

PhD Committee: Pr. Bjarni K. Kristjánsson Pr. Jörundur Svavarsson

Opponents: Pr. Christophe Douady Dr. Guðmundur Guðmundsson

Faculty of Science School of Engineering and Natural Sciences University of Iceland Reykjavik, october 2011 Groundwater amphipods in Iceland: Population structure and phylogenetics

Dissertation submitted in partial fulﬁllment of a Philosophiae Doctor degree in Biology

Faculty of Science School of Engineering and Natural Sciences University of Iceland Sæmundargötu 2 101 Reykjavik Iceland

Telephone: 525 4000

Bibliographic information: Etienne Kornobis, 2011, Groundwater amphipods in Iceland: Population structure and phylogenetics, PhD dissertation, Faculty of Science, University of Ice- land, 166 pp.

ISBN 978-9935-9064-1-0

Printing: Háskólaprent Reykjavik, Iceland, October 2011 Abstract

Crymostygius thingvallensis and Crangonyx islandicus are two endemic species of groundwater amphipods which were recently discovered in Iceland. C. thingvallensis is uncommon, but represents a new monotypic family Crymostygidae. C. islandicus is widespread in the geologically youngest parts of the island and belongs to a genus which is widely distributed in North America and Eurasia. Both species belong to the superfamily Crangonyctoidea, exclusively composed of freshwater species. Iceland is geographically isolated and was fully covered by an ice sheet during the last glacial maximum, 21 000 years ago, hence the Icelandic biota is characterized by extremely low endemism and low species diversity. Thus, the discovery of these two endemic freshwater species, on an island isolated in the midst of the Atlantic ocean raised questions about their colonization and their survival during the glacial periods of the Ice age. Sub- glacial refugia have been hypothesized to explain the occurrence of these two amphipods in Iceland which might have colonized the country via an ancient land bridge, now submerged, connecting Iceland and Greenland. In this thesis, molecular and morphological markers were used to test this hypothesis. In addition, cryptic diversity, common in groundwater species, was assessed with mitochondrial and nuclear markers. Finally, the variation in secondary structure of the nuclear ITS regions has been characterized and its influence on the phylogeny reconstruction evaluated. The mitochondrial markers (COI and 16S) support that C. islandicus populations diverged within the island for up to five million years and consequently they survived repeated glaciations in Iceland in subglacial refugia. The nuclear marker ITS1 is supporting, though weakly, the phylogeographic pattern observed with the mitochondrial markers. Though the populations of C. islandicus showed high genetic divergence, different species delimitation methods led to different conclusions about the potential species complex status of C. islandicus. The taxonomic status of both Icelandic species is supported by phylogenies based on nuclear markers (18S, 28S, 5.8S, ITS1 and ITS2). These phylogenies show high divergence between Crangonyx species from Eurasia and from North America. C. islandicus is more closely related genetically to Crangonyx species from

i North America, supporting the hypothesis of a colonization via Greenland. Morphological data led to a diﬀerent pattern, C. islandicus clustering with species both from North America and Europe which might be due to convergent evolution in morphological traits. In addition, nuclear phylogenies show that Crangonyx, Synurella and Stygobromus genera are polyphyletic and might need taxonomic revision to restore their monophyly. The secondary structures of the ITS2 for Crangonyctoidea species are highly divergent from the common structure described for Metazoans and appeared to have little impact on the phylogenetic reconstruction.

ii Útdráttur

Crymostygius thingvallensis og Crangonyx islandicus eru tvær einlendar tegundir grunnvatnsmarflóa sem fundust nýlega á Íslandi. C. thingvallensis er sjaldgæf og myndar hún nýja ætt marflóa. C. islandicus er útbreidd á gos- belti Íslands, og tilheyrir ættkvísl sem er útbreidd í N-Ameríku, Evrópu og Asíu. Báðar tegundirnar tilheyra yfirættinni Crangonyctoidea sem finnst ein- göngu í ferskvatni. Lífríki Íslands einkennist af lítilli tegundafjölbreytni og fáum einlendum tegundum. Hefur það verið rakið til landfræðilegrar einangr- unar Íslands og þess að landið var hulið jökli fyrir um 21000 árum. Fundur tveggja einlendra grunnvatnsmarflóa vakti því athygli og spurningar vöknuðu um hvernig þær höfðu borist til landsins. Tilgáta hefur verið sett fram um að þær hafi lifað af á Íslandi á ísöld og að þær hafi í raun fylgt Íslandi allt frá myndun þess eða mögulega numið land um landbrú milli Grænlands og Íslands. Í ritgerðinni er þessi tilgáta prófuð með aðferðum sem byggja á greiningu DNA sameinda. Annað meginviðfangsefni ritgerðarinnar er að greina flokkun tegundanna út frá DNA sameindum og útlitseiginleikum marflónna. Auk þess var athugað hvort dulinn breytileiki greindist í hvatbera og kjarna DNA, en slíkt er algengt meðal grunnvatnstegunda. Breytileiki í annars stigs byggingu svokallaðra ITS DNA raða var einnig lýst og áhrif þess á flokkunartré tegundanna athuguð. Breytileiki í hvatberagenunum COI og 16S RNA styðja að C. islandicus lifði af endurtekin kuldaskeið ísaldar undir jökli. Athugun á ITS1 innröðum í kjarna DNA gefur stuðning við þessa ályktun en sýnir mun minni aðgreiningu milli svæða en hvarberagenin. Erfðafræðileg aðgreining milli stofna C. islandicus gefur vísbendingu um ólíkar tegundir sé að ræða innan Íslands. Flokkunarleg staða beggja tegundanna byggð á RNA genum í kjarna DNA (18S, 28S, 5.8S, ITS1 og ITS2) styður fyrri flokkun byggða á útlitseinkennum. Flokkunartréið sýnir hinsvegar skýra aðgreiningu milli Crangnyx tegunda milli Evrasíu og N-Ameríku. C. islandicus er skyldari Crangonyx tegundum í Amer- íku, það styður landnám frá Grænlandi. Flokkunartré byggt á útlitseinkennum var þó öðruvísi, C. islandicus flokkaðist samkvæmt því bæði með tegundum frá N-Ameríku og Evrópu. Slíkt frávik gæti mögulega stafað af samhliða þróunar meðal útlitseiginleikanna. Flokkunartré byggð á kjarna DNA sýndi einnig að

iii ættkvíslirnar Crangonyx, Synurella og Stygobromus eru fjölstofna og að þörf sé að athuga frekar ﬂokkun þeirra byggða á útlitseiginleikum. Annars stigs bygg- ing ITS2 meðal Crangonyctoidea tegunda víkur frá hinni almennu byggingu sem hefur verið lýst fyrir vefdýr (metazoa) og hafði hún lítil áhrif á byggingu ættartrésins.

iv List of Papers

The present thesis is based on three accepted papers and two currently under review. The papers will be referred in the text by their respective numbers as following:

• Paper 1: Kornobis E, Pálsson S, Kristjánsson BK, Svavarsson J (2010). Molecular evidence of the survival of subterranean amphipods (Arthro- poda) during Ice Age underneath glaciers in Iceland. Molecular Ecology 19: 2516-2530.

• Paper 2: Kornobis E, Pálsson S (2011). Discordance in variation of the ITS region and the mitochondrial COI gene in the subterranean amphipod Crangonyx islandicus. Journal of Molecular Evolution (in press), DOI: 10.1007/s00239-011-9455-2.

• Paper 3: Kornobis E, Pálsson S, Sidorov DA, Holsinger JR, Kristjáns- son BK (2011). Molecular taxonomy and phylogenetic aﬃnities of two groundwater amphipods, Crangonyx islandicus and Crymostygius thingvallensis, endemic to Iceland. Molecular Phylogenetics and Evolution 58: 527-539.

• Paper 4: Kornobis E, Pálsson S. (under review). Phylogenies of Cran- gonyctoidea species based on the ITS regions (ITS1 and ITS2).

• Paper 5: Kornobis E, Pálsson S, Svavarsson J. (under review). Classi- ﬁcation of Crangonyx islandicus based on morphological characters and comparison with molecular phylogenies.

Acknowledgments

It is my great pleasure to acknowledge all the persons which played a role, directly or indirectly, in the realization of this thesis. I had the opportunity during these four years to work in an outstanding environment and especially to be advised by a remarkable doctoral committee. I would like to greatly thank my advisor, Snæbjörn Pálsson, for his constant care, his bonhomie and his contagious interest in Science. I have been truly lucky to be supervised by such mentor and I wish to be able to keep on working with him. Thanks a lot also to Bjarni K. Kristjánsson for the awesome sampling trips, always pertinent suggestions and for creating the highest latitude beer club that I know. I have been delighted as well to work with Jörundur Svavarsson and I thank him for guiding me through the mysteries of Crustacean morphology. Thanks to Christophe Douady and Guðmundur Guðmundson for accepting to be my opponents for this thesis. I am grateful to the University of Iceland Research Fund and the Icelandic Research Council (Rannís) for their financial support. I am indebted to all the contributors of free softwares (GNU/Linux, Emacs, R, LATEX, Python, Inkscape, Gimp, Phylip ...) which were of great use during the realization of this thesis. Thanks also to all my lab mates which made this stay so cheerful. I would particularly like to thank Ubaldo, Eduardo, Dileepa, Sindri, Hlynur, Ehsan, Laurène, Baldur and many more. Thanks to Oli Patrick to have revived the cake club. I still feel sorry to have missed one of his french speeches. Finally, Kalina, my wingman who always was there to cheer me up, even when I was in “zombie mode”. It has been a pleasure to work with my first students in the lab, Virginia, Sonia, Daniel, Jaume and Cedric. I wish them the best for their ongoing career. Thanks to Niklas Zennström and Janus Friis for founding Skype, I think I owe them part of my mental sanity. I am deeply grateful to my parents, Lydie and Michel, for their constant wish to help and their support at so many levels. Their visits, calls and “Mon- day’s pdfs” were of great value during the most difficult times of this thesis. Merci pour tout ! I’m not even able to recall the exact number of times that my

vii brother came to Iceland. His visits and traveling tales always cheered me up. Pierrot, on choisit pas sa famille, çà tombe bien j’aurais pas fait mieux. Ben- jamin (aka Dr. Mathon), thank you my friend for the endless nights working and chilling out together through skype. You particularly animated the last moments of my thesis by your riddles and other puns. J’ai ri à tes blagues! Many thanks to Aurane, Déborah, David, Séverine, Fanny, Nanou, Nanette and to all my friends and family which were always present, one way or the other, during my short trips to France. Finally, mi Cari Mari, who faced once more the Icelandic winter to support me and patiently tried to convince me that everything I was doing was awesome. Celia, you are my most precious ﬁnding in Iceland. Takk Fyrir! I wish a joyful life to all of you people !

viii To my family,

Contents

1 Introduction 1 1.1 The discovery of two endemic amphipod species in Iceland . . .1 1.2 A short Natural History of Iceland ...... 2 1.2.1 Subterranean fauna ...... 6 1.2.2 Glacial survival hypothesis ...... 6 1.3 The superfamily Crangonyctoidea ...... 8 1.4 Groundwater biodiversity ...... 8 1.5 Aims of the thesis ...... 11

2 Materials and Methods 13 2.1 Sampling ...... 13 2.2 Test of the subglacial refugia hypothesis and cryptic diversity . 16 2.3 Taxonomy and colonization of the two Icelandic amphipods . . 17 2.3.1 Background on phylogenetic reconstruction methods . . 18 2.3.2 Phylogeny based on morphological data ...... 20 2.4 Secondary structure of the ITS regions ...... 21 2.4.1 Ribosomal RNA secondary structure ...... 23 2.4.2 Characterization of the ITS regions, structures and phylogenetic reconstruction ...... 23 2.4.3 ITS2 secondary structure and cryptic diversity . . . . . 24

3 Results and Discussion 25 3.1 The subglacial refugia hypothesis ...... 25 3.2 Cryptic diversity within C. islandicus ...... 27

xi 3.3 Phylogenetic status and colonization of the Icelandic amphipods 28 3.4 Taxonomy of the Crangonyctoidea and convergence in morphology 29 3.5 Evolution of the ITS regions ...... 31

4 Conclusion and Perspectives 33

5 Paper 1: Molecular evidence of the survival of subterranean amphipods (Arthropoda) during Ice Age underneath glaciers in Iceland 49

6 Paper 2: Discordance in variation of the ITS region and the mitochondrial COI gene in the subterranean amphipod Cran- gonyx islandicus 67

7 Paper 3: Molecular taxonomy and phylogenetic aﬃnities of two groundwater amphipods, Crangonyx islandicus and Cry- mostygius thingvallensis, endemic to Iceland 81

8 Paper 4: Phylogenies of Crangonyctoidea species based on the ITS regions (ITS1 and ITS2) 99

9 Paper 5: Classiﬁcation of Crangonyx islandicus based on morphological characters and comparison with molecular phylogenies 121

xii List of Figures

1.1 Microscopic photography of the two Icelandic amphipods . . . .2 1.2 The opening of the North Atlantic and the formation of Iceland3 1.3 Geology and volcanism of Iceland ...... 4 1.4 Distribution of the Crangonyctoidea ...... 9

2.1 Sampling localities of C. thingvallensis and C. islandicus in Ice- land ...... 14 2.2 Sampling localities for Crangonyctoidea and Niphargoidea species included in this study ...... 15 2.3 Crangonyx islandicus morphology ...... 21 2.4 The telson of two Crangonyx species...... 22 2.5 Location and structure of the ITS2 ...... 22 2.6 Description of the two main types of secondary structure . . . . 23

3.1 Genetic and geographic structure of C. islandicus populations in Iceland ...... 27 3.2 Phylogenetic relationships among Crangonyctoidea species. . . 29

7.1 Bayesian phylogenetic tree of crangonyctid species based on COI sequences ...... 97

xiii

List of Tables

1.1 Comparison in number of species and endemism between two oceanic islands: Iceland and Hawaï ...... 5

Introduction

1.1 The discovery of two endemic amphipod species in Iceland

Recently, in years 2004 and 2006, two endemic species of groundwater amphipods (Crustacea), Crymostygius thingvallensis and Crangonyx islandicus, were described from springs in lava ﬁelds in Iceland (Fig 1.1). The former represents a new monotypic family Crymostygidae (Kristjánsson and Svavars- son, 2004) and the latter a new species of the genus Crangonyx (Svavarsson and Kristjánsson, 2006). Both species belong to the superfamily Cran- gonyctoidea which occurs worldwide (Väinölä et al., 2008). These ﬁndings are unique to Iceland as the Icelandic biota is characterized by low species diversity and low endemism (Sadler, 1999; Brochmann et al., 2003; Coope et al., 1986). As Iceland has been repeatedly covered by glaciers during the last Ice Age, most of the species currently present are considered to have colonized the island after the last glacial maximum around 21 000 years ago (e.g. Buck- land et al., 1986; Coope et al., 1986). As groundwater amphipods are known to have poor dispersal capacities (e.g. Lefébure et al., 2007; Trontelj et al., 2009), questions have been raised about their origin and the colonization of Iceland by these two endemic amphipod species. Iceland is an island geographically isolated in the midst of the Atlantic Ocean and considering that the superfamily Crangonyctoidea is exclusively composed of freshwater species, largely restricted to subterranean habitats (Väinölä et al., 2008), it is unlikely that these two species recently evolved from marine species. Kristjánsson

1 2 Introduction

Figure 1.1: Microscopic photography of the two Icelandic amphipods: Cry- mostygius thingvallensis on the left and Crangonyx islandicus on the right. and Svavarsson (2007) raised the hypothesis that Crymostygius thingvallensis and Crangonyx islandicus colonized Iceland via a past land connection between Greenland and Iceland and survived repeated glaciations on the island in subglacial refugia. The present thesis evaluates this hypothesis and the taxonomic aﬃnities of the two species using population genetics, phylogeography and phylogenetic methods based on various molecular and morphological characters.

1.2 A short Natural History of Iceland

In the midst of the North Atlantic Ocean, Iceland is a geologically young island which has been formed since the Miocene (i.e. between 23 and 5 million years ago) until now (Einarsson, 1994, p. 231). Its formation has been commonly explained by the presence of a volcanic hotspot i.e. a strong upwelling of magma in a mantle plume, which is now approximately situated beneath the Vatna- jökull glacier in Iceland (see Mihalffy et al., 2008). As the North American and Eurasian tectonic plates started to move apart around 55 million years ago, the mantel plume alimented important volcanism forming a land connection between Greenland and Scotland, known as the Greenland-Scotland transverse ridge (Fig 1.2, see Denk et al., 2011). Due to the continuous spreading of the plates, this ridge is now submerged and Iceland is the only part which re- mains above sea level. The subsistence of Iceland above sea level is due to the A short Natural History of Iceland 3

Scandinavia Spreading axis

Extinct spreading Greenland axis

Plateau basalts of continental origin Oceanic crust formed above Faeroes sea level Oceanic crust formed below sea level Miocene plateau basalts of Iceland

55 Ma 24 Ma 4 Ma

Figure 1.2: The opening of the North Atlantic and the formation of Iceland. Ma: Million years ago. Modified from Denk et al. (2011). combined volcanic production of the mantel plume and the mid-ocean ridge magmatism (see Geirsdóttir et al., 2007). A recent study suggested that the Greenland-Iceland part of the ridge was emergent at least until 10 million years ago while the Iceland-Scotland part submerged earlier (Poore, 2008). Fossil records indicate that plants migrated along this land bridge between Scotland and Iceland until 24 million years ago (Mya) and until 15 Mya or even 6 Mya between Greenland and Iceland (Grimsson et al., 2007; Denk et al., 2010). The geologically youngest part of the island is located at the boundaries of the tectonic plates in the middle of the volcanic active zone and the age of the geological formations are increasing with the distance from these boundaries (Fig 1.3), the oldest being around 16 million years old (see Grimsson et al., 2007). The basaltic bedrocks originating before the Pleistocene i.e. before the onset of the last Ice Age, are now fairly impermeable and thus groundwater is scarce in the geologically oldest parts of Iceland (see Fig 1.3, Einarsson, 1994). Conversely, lava deposits occurring after the glaciations are highly porous and groundwater is plentiful in the youngest parts of the island (Einarsson, 1994). These areas are as well rich in fissures zones, which are mainly located at the tectonic plates boundaries (Fig 1.3) and are often characterized by high geothermal activity (Einarsson, 1994). Most of the early flora in Iceland was widespread in the northern hemisphere and no predominant colonization routes – either from North America 4 Introduction

Figure 1.3: Geology and volcanism of Iceland. 1: eruption fissures, 2: volca- noes. Age of the geological formations, 3: Tertiary, 4: late Pliocene and early Pleistocene, 5: late Pleistocene, 6: Holocene before settlement, 7: Holocene after settlement. The four red circles correspond to: a, Snæfellsness; and three main fissure zones: b, Þingvallavatn; c, Lakagígar; d, Mývatn. Modified from Einarsson (1994). A short Natural History of Iceland 5

Table 1.1: Comparison in number of species and endemism between two oceanic islands: Iceland and Hawaï. N species: number of species. After Sadler (1999).

Iceland Hawaï N species Endemics N species Endemics Insects 1245 0 7862 5237 Birds 80 0 274 60 Molluscs 35 0 1656 956 or Europe – can be inferred from fossil records (Grimsson et al., 2007). The warmth-demanding species present at that time became progressively extinct due to climate cooling (Rundgren and Ingólfsson, 1999; Einarsson, 1994) marking the beginning of the glaciation period, between 5 and 10 million years ago in the northern hemisphere (see for review Geirsdóttir et al., 2007). Iceland, located just south of the Arctic circle around latitude 65◦ North, has been repeatedly covered by an ice sheet during the glacial cycles of the last Ice Age which onset occurred around 2.75 million years ago (see for review Geirsdóttir et al., 2007). During the last glacial maximum, around 21 000 years ago, Iceland was almost completely covered by glaciers and only small ice-free areas may have been present along the coastal mountains (Andrews et al., 2000; Geirsdóttir et al., 2007). The last deglaciation started around 15 000 years ago while readvances of glaciers might have occurred in Iceland until 11 000 years ago (see Geirsdóttir et al., 2007). The combination of the short time since the retreat of the glaciers and the geographic isolation of Iceland explains the relatively low terrestrial species diversity and its low endemism (Table 1.1; Sadler, 1999; Brochmann et al., 2003; Coope et al., 1986). During the end of the last glacial period, approximately 12 500 years ago, the shoreline in Iceland was up to 100 meters above the current sea level. Though the melting of the glaciers resulted in an increase in global sea level, it caused also the rise of the island, free from the pressure due to the weight of the ice sheet (see Norðdahl et al., 2008). 6 Introduction

1.2.1 Subterranean fauna

The subterranean fauna in Iceland is poorly known so far and only three stygobiont species – species restricted to aquatic subterranean habitats – have been described. One, is the copepod Parastenocaris glacialis which is also known from groundwater in Europe (Juberthie et al., ress). The two others are the endemic amphipods Crymostygius thingvallensis and Crangonyx islandicus found in groundwater underneath lava fields no older than 10 000 years (Kristjánsson and Svavarsson, 2007; Kornobis et al., 2010). Cran- gonyx islandicus is relatively common and widespread throughout the geologically youngest part of the island, the west and east volcanic zones around the boundaries of the tectonic plates (Svavarsson and Kristjánsson, 2006; Kornobis et al., 2010). Crymostygius thingvallensis is uncommon and has been found to date only at spring inlets in lake Þingvallavatn, Southwest Ice- land (Kristjánsson and Svavarsson, 2004), and in the stomach of an Arctic char (Salvelinus alpinus) in Jokulsargljúfur, East Iceland (Kristjánsson and Svavarsson, 2007; Egilsdóttir and Kristjánsson, 2008). Outside the east and west volcanic zones, in lava fields in the Snæfellsness peninsula (see Fig 1.3), neither of these species was found (see Kornobis et al., 2010). The Snæfellsness peninsula is outside the central volcanic zone and is mainly formed by older (see Fig 1.3) and less porous bedrocks which have been intruded by volcanic magma. Moreover this peninsula lacks fissure zones and its basaltic formations chemistry is different from the ones found in the central volcanic zone (Sigurdsson, 1970).

1.2.2 Glacial survival hypothesis

Most of the European biota, particularly temperate species, have survived during glacial maxima in refugia at lower latitudes, mostly located in the Iberic peninsula, Italy and the Balkans (Taberlet et al., 1998; Hewitt, 2000, 2004). Arctic species tend to exhibit low genetic diversity and shallow genetic clades, as a consequence of population contraction and expansion over the last glacial cycles (Hewitt, 2004). Nonetheless, a growing body of phylogeographic studies supports northern cryptic refugia which might even have occurred within A short Natural History of Iceland 7 the supposed limits of the ice sheet (Stewart and Lister, 2001; Provan and Bennett, 2008). Two theories have been proposed to explain the origin of the Icelandic biota: the tabula rasa, i.e. total extinction of the species present in glaciated areas, and the glacial survival, i.e. the survival in located ice free refugia (see Ægisdóttir and Þórhallsdóttir, 2004; Denk et al., 2011). A vast majority of plants occurring now in Iceland has an European origin (Einarsson, 1963; Rundgren and Ingólfsson, 1999). Plant species have been suggested to have survived the glaciations in ice free areas such as nunataks (e.g. Rund- gren and Ingólfsson, 1999) though the majority is known to have migrated from Europe and northern Eurasia after the last glacial maximum (Denk et al., 2011). According to Wares and Cunningham (2001), the genetic divergence observed between populations of the marine isopod Idotea balthica in Iceland and Europe supports a split older than 200 000 years and therefore the occurrence of a marine refugium in southwestern Iceland. Nonetheless, geological data and the extent of the ice sheet during glaciation in this region are not supporting this hypothesis (Ingólfsson, 2009). These glacial survival depend on the occurrence of unglaciated regions which acted as refugia. Considering the extent of the ice sheet on Iceland during the last glacial maximum, the survival of the two Icelandic freshwater amphipods implies the occurrence of subglacial refugia (Kristjánsson and Svavarsson, 2007). Subglacial refugia have been as well hypothesized to explain geographic distribution of groundwater amphipod species in North America (e.g. Holsinger, 1980; Koenemann and Holsinger, 2001) and Ireland (Hanfling et al., 2008). Nonetheless, the occurrence of unglaciated refugia and postglacial colonization from these adjacent ice free regions to explain current species distributions are commonly diﬃcult to reject (e.g. Lefébure et al., 2007). The occurrence of the endemic species C. islandicus, widespread in a geographically isolated island which was repeatedly covered by ice during glaciations allows for testing the subglacial refugia hypothesis. 8 Introduction

1.3 The superfamily Crangonyctoidea

Crymostygius thingvallensis and Crangonyx islandicus belong to the freshwater superfamily Crangonyctoidea. This superfamily encompasses 11 families and presents an unusual geographic distribution compared to other continental amphipods. Crangonyctoids are distributed worldwide (Fig 1.4) and present more species diversity in the Nearctic region than in the Palearctic (184 and 25 species respectively; Väinölä et al., 2008). Conversely, the Palearctic harbors 70% of the total continental amphipod species diversity (Väinölä et al., 2008). The wide distribution of the exclusively freshwater superfamily Cran- gonyctoidea may be explained by the occurrence of a widespread freshwater ancestor before the breakup of Pangaea i.e. in the late Triassic and early Juras- sic (Holsinger, 1992). This is further supported by the fact that groundwater amphipod species appear to have poor dispersal capabilities (Lefébure et al., 2007; Trontelj et al., 2009). Among Crangonyctoids, the family Crangonyc- tidae is common and widespread in North America but occurs also in Eurasia, Iceland and North Africa (Fig 1.4). The large distribution of the Crangonycti- dae in the northern hemisphere suggests that their ancestor occupied Laurasia before its breakup around 55 Mya (Holsinger, 1992). The only known species from North Africa is Crangonyx africanus, recently described as endemic to Morocco (Messouli, 2006). Unfortunately, the well where the only sample of Crangonyx africanus was found, does not exist anymore. The genus Crangonyx is represented by 47 species, a vast majority (42 species) distributed in North America, three species in Eurasia, one in North Africa and one in Iceland.

1.4 Groundwater biodiversity

The groundwater have been described as “an unseen ocean beneath our feet” and the exhaustive inventory of their biodiversity is still at its beginnings (Danielopol et al., 2000). Groundwater represent 30% of the freshwater on Earth and 99% of the unfrozen continental waters (see Shiklomanov, 1993; Danielopol et al., 2003). Despite the abundance of groundwater habitats and due to their cryptic nature, groundwater species richness surveys have Groundwater biodiversity 9

Figure 1.4: Distribution of the Crangonyctoidea. The different families are: All: Allocrangonyctidae, Aus: Austrocrangonyctidae, Cry: Crymosty- gidae, fal: “falklandiellids”, Neo: Neoniphargidae, Par: Paracrangonyctidae, Per: Perthiidae, Phr: Phreatogammaridae, Pse: Pseudocrangonyctidae, Ste: Sternophysingidae. Modified from Väinölä et al. (2008). 10 Introduction been conducted only recently (e.g. the PASCALIS project in Europe, Gibert et al. 2005). The estimation of the species richness in such particular environments faces major obstacles. Firstly, the nature of the groundwater habitats makes their access and sampling arduous and hydro-biological connections between them are relatively difficult to assess. Secondly, subterranean species are subject to morphological stasis or convergent evolution: due to the similar selective pressures in groundwater habitats (absence of light, low nutrients and low oxygen, see Malard and Hervant 1999), different species often exhibit similar morphologies such as reduction or absence of eyes, elongation of other sensory appendages, loss of pigmentation. The abundance of morphological convergence in subterranean species hinders the clear delimitation of species and leads to an underestimation of the biodiversity when descriptions based solely on morphology are considered. This latter issue can be solved by the use of molecular taxonomy and DNA barcodes for species identification which have been extensively used in the past decades (e.g. Roe and Sperling, 2007; Lefébure et al., 2007; Seidel et al., 2009; Humphreys et al., 2010). Depletion and pollution by human activities such as agriculture, urban- ization, industry and tourism are widely threatening groundwater ecosystems (Danielopol et al., 2003). In addition, the abundance of endemism witnessed in groundwater species is making these species even more subject to extinction. In the six regions of Europe covered by the PASCALIS survey, 41% of the aquatic subterranean species reported could be considered as endemic (Gibert et al., 2009). This highlights the need for a more precise estimation of the biodiversity in subterranean species, particularly using molecular markers, in order to define appropriate conservation areas and units. Aims of the thesis 11

1.5 Aims of the thesis

The main objective of this thesis is to describe the evolution of Icelandic groundwater amphipods. This was achieved through five different studies which addressed more specifically the following aims:

• To test the subglacial refugia hypothesis (Paper 1 and 2).

• To locate putative refugia in Iceland (Paper 1).

• To assess if the widespread populations of Crangonyx islandicus represent a single or a complex of species i.e. to detect cryptic diversity (Paper 1 and 2).

• To evaluate the species status of C. islandicus and the family status of C. thingvallensis by assessing the phylogenetic relationships among various Crangonyctoidea species (Paper 3 and 4).

• To evaluate the phylogeny of the Crangonyx genus based on morphological characters. Due to convergence which occurs in subterranean species the phylogenetic tree is compared with trees based on molecular data (Paper 5).

• To infer how C. thingvallensis and C. islandicus colonized Iceland (Paper 3 and 5).

• To characterize the secondary structure of the internal transcribed regions and evaluate its inﬂuence on the phylogeny of Crangonyctoidea species (Paper 2 and 4).

• To evaluate the behaviors of two commonly used barcode markers to assess variation in groundwater amphipods (Paper 1, 2, 4).

Materials and Methods

2.1 Sampling

To test the subglacial refugia hypothesis, locate their putative locations and detect cryptic diversity, genetic variation found in the widespread populations of C. islandicus was characterized. In order to do so, C. islandicus specimens were sampled from groundwater in 23 localities in Iceland (Fig 2.1). Samples were obtained by electrofishing in springs coming out from lava fields no older than 10 000 years. No specimens were found in springs of lava fields at the Snæfellsness peninsula in West Iceland (see Fig 1.3). The taxonomic status of the two Icelandic species and their colonization was investigated by assessing the phylogenetic relationships of C. islandicus and C. thingvallensis with other species from the superfamily Crangonyctoidea. C. thingvallensis is uncommon and was sampled from the only location known, in Lake Þingvallavatn (Fig 2.1). In addition, samples from eight Crangonyctoidea species were obtained from North America, Europe and Russia (Fig 2.2). DNA sequences from 17 additional species of Crangonyctoidea and Niphargoidea were obtained from Genbank (see Fig 2.2). The phylogenetic status of C. islandicus based on morphological characters was studied by comparing with the classification of the North American Crangonyx species made by Zhang and Holsinger (2003) and by adding specimens of C. subterraneus and C. chlebnikovi from Eurasia (Fig 2.2).

13 14 Materials and Methods 66.5 Latitude (°N) .5 65.5 63.5 64

24 22 20 18 16 14

Longitude (°W)

Figure 2.1: Sampling localities of C. thingvallensis and C. islandicus in Iceland. Triangle: C. thingvallensis, circles: C. islandicus. The volcanic active zone is displayed in red. Glaciers are in blue. Longitudes and latitudes are shown in degrees on the x- and y-axes respectively. Sampling 15

10 12 13 6 11 1 7 8 14 2 3 15 4 9 5

Figure 2.2: Sampling localities for Crangonyctoidea and Niphargoidea species included in this study. Red circles represent samples obtained during this study and blue ones the species obtained from Genbank. Open circles correspond to several sampling locations. 1: Bactrurus brachycaudus, Bactrurus mucronatus, Bactrurus pseudomucronatus, Crangonyx forbesi, Crangonyx sp.; 2: Crangonyx pseudogracilis, Synurella sp.; 3: Crangonyx serratus, Stygobro- mus gracilipes, Stygobromus stegerorum; 4: Synurella dentata, 5: Crangonyx ﬂoridanus; 6: Crangonyx islandicus, Crymostygius thingvallensis; 7: Cran- gonyx subterraneus; 8: Niphargus plateaui; 9: Niphargus frasassianus; 10: Crangonyx subterraneus, Niphargus kochianus, Niphargus fontanus, Niphar- gus schellenbergi; 11: Synurella ambullans; 12: Crangonyx chlebnikovi; 13: Amurocrangonyx arsenjevi; 14: Procrangonyx primoryensis; 15: Pseudocran- gonyx korkishkoorum. 16 Materials and Methods

2.2 Test of the subglacial refugia hypothesis and cryptic diversity

The genetic variation among C. islandicus was characterized using mitochondrial (COI and 16S in Paper 1) and nuclear markers (ITS1 and ITS2 in Paper 2). COI and 16S mitochondrial genes and the nuclear internal transcribed spacers 1 (ITS1) have been commonly used to estimate intraspecific variation and to describe cryptic diversity within species (e.g. in crustaceans Chu et al., 2001; Wattier et al., 2006; Audzijonyte and Väinölä, 2006; Lefébure et al., 2007; Hanfling et al., 2008). Genetic patterns may reflect the past history and distribution of species. Refugia are characterized by high genetic diversity though newly colonized areas usually present low diversity. Correlation between genetic and geographic distances within species can reflect isolation by distance and past vicariance/dispersal events while the lack of correlation might indicate several colonization events from different origins. The genetic variation within Arctic and sub Arctic species are expected to be low due the ice sheet extent during the past glaciation and recent colonization events. Phylogeographic patterns and correlation between genetic and geographic distances are explored in order to infer if the occurrence of C. islandicus in Iceland is due to a single or several colonization events. The COI and 16S mitochondrial (mt) genes substitution rates have been calibrated for crustacean species using dated geological events (Knowlton et al., 1993; Knowlton and Weigt, 1998; Schubart et al., 1998; Cunningham et al., 1992; Stillman and Reeb, 2001). Accord- ing to the molecular clock hypothesis, the rate of amino acid or nucleotide substitution for a gene is approximately constant over evolutionary time (see Zuckerkandl and Pauling, 1965; Nei and Kumar, 2000). The COI and 16S mt genes variation and their calibrated substitution rates were used to provide time estimations for the split of C. islandicus populations which, in the case of a single colonization event, is giving an approximation of the time since the colonization of Iceland (Paper 1). These time estimates were then compared with the date of the last glacial maximum, when Iceland was covered by glaciers. In Paper 1, a Bayesian approach implemented in the software Taxonomy and colonization of the two Icelandic amphipods 17

BEAST (Drummond and Rambaut, 2007) was used to estimate the time of divergence of C. islandicus populations. This software presents the advan- tage to compute simultaneously the phylogeny and the divergence dates for multi-locus data and a specific substitution model can be specified for each locus. The location of the subglacial refugia during the last glaciations was investigated in Paper 1 by testing if COI and 16S genetic diversity was dependent on altitude, distance from the tectonic plate boundaries and age of the lava fields. Considering the weak dispersal capacities of subterranean amphipod species, the current populations with the most genetic diversity are expected to be located nearby the past subglacial refugia. Potential cryptic diversity within C. islandicus was detected by analyzing the variation of the COI and 16S genes and the ITS regions (Papers 1 and 2). The barcoding of life initiative proposed by Hebert et al. (2003) aims at the species identification of unknown specimens and enhancing the discovery of new species (see Moritz and Cicero, 2004). The COI and internal transcribed spacer 2 (ITS2) have been proposed as DNA barcodes and tools for species delimitation, of particular interest for species with high cryptic diversity. Their variation was used in this thesis to infer if C. islandicus represents a species complex. Genetic divergence thresholds for the COI gene (Hebert et al., 2004; Lefébure et al., 2006) and criteria based on the sequence variation and the secondary structure of the ITS2 (Müller et al., 2007) have been developed in order to delimit species and were used to detect putative cryptic species within C. islandicus.

2.3 Taxonomy and colonization of the two Icelandic amphipods

The taxonomic status of the two groundwater amphipods and their colonization were assessed by evaluating their phylogenetic relationships with other Crangonyctoidea species. Phylogenies were built based on the mitochondrial genes COI and 16S and the nuclear genes 18S, 28S, 5.8S, ITS1 and ITS2 (Pa- pers 3 and 4). By assessing the genetically closest relatives of the Icelandic 18 Materials and Methods species, it can be possible to infer if they colonized Iceland from the Nearctic or the Palearctic regions. For this purpose, morphological markers were used as well to assess which Crangonyx species was the closest morphologically to C. islandicus (Paper 5).

2.3.1 Background on phylogenetic reconstruction methods

Phylogenetics is the study of evolutionary relationships among species. The most common representation of these relationships is a phylogenetic tree, where the nodes correspond to putative ancestors and the tips of the branches are the organisms studied. These relationships can be inferred from various types of data and the phylogenies presented in this study are based on molecular (Papers 1, 2, 3, 4) and morphological ones (Paper 5). The phylogenetic reconstructions based on molecular data are known to be affected both by the alignment of DNA sequences and the phylogenetic tree building methods (e.g Saitou and Imanishi, 1989; Morrison and Ellis, 1997). Though it has been repeatedly shown that alignment techniques have at least as much effect as the tree building method, publications incorporating different alignment methods to build phylogenies are still scarce (see Morrison, 2008). We consequently used different alignment and tree building methods in this thesis and evaluated their influence on the topology of the trees in Paper 3.

Alignment methods

In order to reconstruct phylogenetic trees based on nucleotide sequences, it is necessary to assess which sites among the available sequences are homologous i.e. derived from a common ancestry. Numerous multiple sequences alignment techniques deal with this issue and several of them were used in Papers 2, 3 and 4. We attempted as well to evaluate the inﬂuence of these alignment techniques on the main conclusions of the Paper 3. The most common approach in multiple alignment is to use a progressive alignment algorithm (see Nei and Kumar, 2000). The principle of this algorithm is to construct an approximate phylogenetic tree and build up the alignment by progressively adding sequences in the order speciﬁed by the tree (see Higgs and Attwood, Taxonomy and colonization of the two Icelandic amphipods 19

2005). This is the algorithm used by the program ClustalW (Larkin et al., 2007). Several improvements have been made in the field of multiple sequences alignment recently (see Morrison, 2009). The program Mafft (Katoh et al., 2002) is incorporating some of them, such as improving the alignment by using an iterative refinement method, and has been shown to perform better than ClustalW (see Morrison, 2009). Another software, RNAsalsa, uses both the secondary structure information (see section below on RNA secondary structure) for adjusting and refining the sequences alignment and the sequence information contained in the alignment to refine the structure predictions in turn (Stocsits et al., 2009). RNAsalsa is particularly suited for alignment of RNA genes such as in Paper 3 and 4, since the divergence between the sequences may depend on the secondary structure of the molecule. The program FSA (Bradley et al., 2009) in Paper 3 is based on a recently developed statistical alignment method. Its principle is to build up the alignment sites per sites according to statistical pairwise estimations of homology (Bradley et al., 2009). Finally, the alignment of the ITS2 sequences was obtained with the program 4Sale (Seibel et al., 2008) which provides a scoring matrix specific to this nuclear region and allows an alignment based on both secondary structures and nucleotide sequences simultaneously.

Phylogenetic methods

The commonly used methods to build phylogenies can be divided in two major groups: the methods based on distances and the ones based on characters (see Yang, 2006). In distance based methods, the phylogenetic tree is constructed with distances computed for all pairs of taxa. The neighbor joining algorithm (Saitou and Nei, 1987) is a distance based method which was used in Papers 4 and 5. The maximum parsimony, maximum likelihood and Bayesian methods are all character based (Yang, 2006), which means that every single site is treated independently. Trees selected according to the maximum parsimony methods are the one which require the smallest number of evolutionary changes among these characters. In the maximum likelihood method, the likelihood value of the tree is used to measure the fit of the tree to the characters states. The tree chosen will be the one with the highest likelihood. Finally, following 20 Materials and Methods the Bayesian method, the tree selected as the “true” tree is the one with the maximum posterior probability i.e. the probability that the tree is true given the data. The substitution model fitting best the data, is used to compute the pairwise distances for distance methods and to calculate the likelihood function for the Bayesian and the likelihood methods (Yang, 2006). In order to assess the influence of the reconstruction method on the tree, a Bayesian (with MrBayes, Ronquist and Huelsenbeck, 2003) and a maximum likelihood method (with Phyml, Guindon and Gascuel, 2003) were applied and are presented in Paper 3. MrBayes software presents the main advantages to allow partition of the data and to implement two different evolution models for paired and unpaired regions according to the secondary structure of rRNA genes. The software ProfdistS (Wolf et al., 2008) used in Paper 4 is based on a profile neighbor joining method which has been shown to provide better estimations of the deepest evolutionary relationships i.e. the basal nodes of the tree (Müller et al., 2004). The particular interest of ProfdistS for this study is the computation of phylogenetic trees based on both the sequence and the secondary structure of the ITS2 region. Trees based on ITS1 and complete data set (ITS1, 5.8S, ITS2) were reconstructed using a simple neighbor joining method for comparison with the tree obtained with the ITS2.

2.3.2 Phylogeny based on morphological data

Phylogenies based on morphological characters were reconstructed to check the taxonomic status of C. islandicus by assessing its morphological affinities with other Crangonyx species and to infer putative colonization routes (Paper 5). In their description of the species, Svavarsson and Kristjánsson (2006) suggested that C. islandicus might represent a new genus according to morphological features unique to the species. The taxonomic status of C. islandicus as a species or a specific genus was evaluated with morphological characters previously defined by Zhang and Holsinger (2003) to assess the phylogenetic relationships among North American Crangonyx species. The phylogeny was reconstructed by adding C. islandicus, C. subterraneus and C. chlebnikovi from Eurasia (Paper 5) and compared with the phylogenies based on molecular markers (Paper 3). The body parts used by Zhang and Holsinger (2003) Secondary structure of the ITS regions 21

Figure 2.3: Crangonyx islandicus morphology. Numbers correspond to anatomical parts of interest for the morphological study in Paper 5. 1: location of the mouthparts, Mandible and Maxilla (not visible on the drawing), 2: Maxilliped, 3: Gnathopod 1, 4: Gnathopod 2, 5: Pleopod 1, 6: Telson. to deﬁne the morphological characters are shown in Figure 2.3 and the Figure 2.4 shows one example of these characters. The phylogeny was reconstructed using both maximum parsimony and neighbor joining methods. Maximum parsimony was designed originally for morphological characters (Hennig, 1966) and was chosen in order to compare with the phylogenies already obtained by Zhang and Holsinger (2003).

2.4 Secondary structure of the ITS regions

The internal transcribed spacers, ITS1 and ITS2, are located respectively between the 18S and 5.8S and between the 5.8S and the 28S ribosomal genes (Fig 2.5). These ITS regions and particularly their secondary structures have a crucial role in the processing of the pre rRNAs (Musters et al., 1990; van Nues et al., 1995). 22 Materials and Methods

Figure 2.4: The telson of two Crangonyx species with varying number of spines on each lobe. a: C. islandicus, less than 4 spines per lobe, b: C. dearolﬁ, more than 4 spines per lobes.

Figure 2.5: Location and structure of the ITS2. The excision site of the ITS2 is highlighted in red. The four main stem structures of the ITS2 are written I, II, III, IV. Modiﬁed from Keller et al. (2009) Secondary structure of the ITS regions 23

Figure 2.6: Description of the two main types of secondary structure. Hydrogen bonds between complementary bases are represented by the blue lines. A compensatory base change between sequence a and b is highlighted in green: the bases are both diﬀerent but the hydrogen bond is conserved.

2.4.1 Ribosomal RNA secondary structure

Ribosomal RNA (rRNA) genes are synthesized in cells as single stranded and fold into a secondary structure which is essential to their function in the ribo- some. Their secondary structure and the ones from their internal transcribed spacers (ITS1 and ITS2) emerge from the presence of complementary bases along the sequence which form hydrogen bonds (Fig 2.6). The two main structures observed are stems (the paired regions) and loops (the unpaired ones). One of the most commonly used method to infer secondary structures of RNAs is based on the minimum free energy (see Mathews and Turner, 2006). Stems stabilize the structure and contribute to negative free energy, which value depends on the pairing of the bases found along the stem. Loops destabilize the structure and contribute to positive free energy. The secondary structure selected according to this method is the one with the minimal free energy. This method was used for secondary structure predictions in Papers 2, 3 and 4.

2.4.2 Characterization of the ITS regions, structures and phylogenetic reconstruction

The variation of the ITS1 and ITS2 regions in terms of sequence and secondary structure was estimated within C. islandicus (Paper 2) and between 24 Materials and Methods

Crangonyctoidea species (Paper 4). The number of stems in the ITS2 secondary structure is variable among Eukaryotes but the stem regions II and III (Fig 2.5) are known to be common to essentially all of them (Coleman, 2007). In addition, a conserved hybridization region between the 5.8S and the 28S and its characteristic secondary structure (see Fig 2.5) have been used to delimit the ITS2 (Keller et al., 2009). This method was employed in Paper 2 and 4. Secondary structures obtained for the ITS2 in Paper 4 have been compared to the previously described model. Due to relatively high rate of evolution of the nucleotide sequences of the ITS1 and ITS2, these markers have been more commonly used to reconstruct phylogenies at the species and genus level. Nonetheless, recent methods have been developed for the ITS2 in order to reconstruct phylogenies at higher taxonomic levels by using both the nucleotide and structure sequence phylogenetic signal (e.g. Tippery and Les, 2008; Buchheim et al., 2011). One such method is implemented in the software ProfdistS (Wolf et al., 2008), which was used to reconstruct the phylogeny of Crangonyctoidea species (Paper 4).

2.4.3 ITS2 secondary structure and cryptic diversity

Compensatory base changes (CBC) are mutations occurring in both nucleotides of a stem structure, retaining the initial paired nucleotide bond (Fig 2.6). The occurrence of these CBC violates the assumption of independent evolution of the bases along the sequence and might aﬀect phylogenetic reconstruction. Thus it is necessary to consider two diﬀerent models of evolution for bases which are in stem and in loop regions. This was done for the phylogenies based on the 18S and 28S rRNA genes in Paper 3 with the software Mrbayes. In addition, CBC present in ITS2 sequences have been proposed as a species delimitation criteria (Müller et al., 2007) and their occurrence have been checked in Paper 2 to assess for potential cryptic species within C. islandicus. Results and Discussion

3.1 The subglacial refugia hypothesis

The occurrence of subglacial refugia in Iceland was tested in Paper 1 and 2. Based on the variation of the mt genes COI and 16S, the populations of C. islandicus are highly structured genetically and geographically (Fig 3.1). Unlike most species at high latitudes and formerly glaciated areas (Hewitt, 2004), C. islandicus presents high genetic diversity and divergent genetic clades. The species is divided in six well supported monophyletic groups, which have diverged from each other from about 430 000 years up to 5 million years ago. In addition, both COI and 16S mt genes show a clear isolation by distance of C. islandicus populations, genetic distances being highly correlated to the geographic distances. This supports a single colonization event of Iceland from ancestral populations of C. islandicus which diverged within the island after the colonization. Such patterns are unlikely to be observed from multiple colonization events or in the case of ancestral polymorphism. As this divergence within C. islandicus is dated to up to 5 million years ago, their ancestors colonized Iceland before the last glacial maximum and even before the onset of the last Ice Age, 2.75 million years ago (see Geirsdóttir et al., 2007). Only small ice free refugia at nunataks were possibly located in northwestern, northern and eastern Iceland during the last glacial maximum (see Geirsdóttir et al., 2007). Survival in such located ice free mountain peaks is unlikely for groundwater species and can hardly explain the phylogeographic structure observed among C. islandicus populations. Thus, the variation observed in mt genes COI and 16S provides a strong evidence for the occurrence of subglacial

25 26 Results and Discussion refugia in Iceland during the last Ice Age. This is the first documented example of a groundwater amphipod surviving underneath glaciers. The nuclear internal transcribed spacers showed conversely much less variation within C. islandicus than the mtDNA sequences (Paper 2) and the ITS1 variation is only weakly supporting the groups defined by mtDNA variation (Fig 3.1). Nonetheless, ITS1 pairwise distances were correlated to COI distances and to geographic distances supporting the isolation by distance in C. islandicus populations observed with the mt genes. As no calibration of the ITS1 and ITS2 substitution rates is available for crustacean species, it was not possible to estimate divergence times from the variation of the ITS region. The putative location of the subglacial refugia was investigated by evaluating the correlation between genetic diversity and altitude, age of the lava fields and distance from the tectonic plates boundaries. The genetic diversity of the COI gene increases significantly with altitude and the less diverse populations were found at an altitude lower than 100 meters. This is reflecting recent colonization of previously submerged areas due to the past increase in sea level in Iceland, which was 100 meters above its current position about 12 500 years ago (Norðdahl et al., 2008). Interestingly, the genetic diversity was correlated with the distance from the tectonic plates boundaries. These boundaries are rich in fissure zones (see Fig 1.3) which are particularly common near the Þingvallavatn, Mývatn and Lakagígar areas (see Fig 1.3). The two most diverse phylogenetic groups (AA’ and D in Paper 1) within C. islandicus were found at Þingvallavatn and Lakagígar. Consequently, the genetic diversity patterns in C. islandicus populations suggest that fissure zones at the tectonic plates boundaries were potential refugia during the glacial periods of the Ice Age. The genetic divergence between C. islandicus populations is supporting the existence of several subglacial refugia. The six main monophyletic groups observed in the mtDNA phylogeny appear to have diverged before the last glacial maximum and hence at least six refuge areas have occurred in the corresponding geographic regions of Iceland during the last glaciations. Cryptic diversity within C. islandicus 27 a b c F A E

B A' C C B D D E A F

Figure 3.1: Genetic and geographic structure of C. islandicus populations in Iceland. a: Sampling locations and their respective mtDNA groups (A, A’, B, C, D, E, F). b: Phylogenetic tree based on COI and 16S mt genes (modified from Paper 1). c: Network of ITS1 haplotypes (modified from Paper 2). Colors correspond in all three figures to the groups AA’, B, C, D, E, F.

3.2 Cryptic diversity within C. islandicus

According to the COI and 16S genes, C. islandicus presents high genetic diversity and is structured in highly divergent clades. One clade from North Iceland (group F in Paper 1) diverged early from the rest of the species and is separated by 0.12 substitutions per site from the other clades for the COI gene. This is lower than the threshold of 0.16 subst/site defined by Lefébure et al. (2006) for the delimitation of Crustacean species, thus it is not supporting C. islandicus as a species complex. The ITS2 variation does not either support C. islandicus as a species complex, since no compensatory base change were observed among C. islandicus populations (Paper 2). Conversely, according to the species screening threshold defined by Hebert et al. (2004), the genetic distances between the F group and the other groups are ten times larger than the average intra-populations distances and therefore, the F group represents a provisional species, different from the other C. islandicus populations (Pa- per 1). No variation, in the morphological characters used in Paper 5, was observed between the A and F group defined by mtDNA variation in Paper 1. Further morphological and morphometric analysis are needed to conclude about the potential species complex status of C. islandicus. 28 Results and Discussion

3.3 Phylogenetic status of the Icelandic amphipods and the colonization of Iceland

Kristjánsson and Svavarsson (2007) suggested that C. thingvallensis and C. islandicus colonized Iceland from Greenland. Molecular phylogenies based on nuclear markers in this thesis are supporting this hypothesis (Paper 3 and 4). According to the phylogenies based on nuclear markers 18S and 28S in Paper 3, C. islandicus is more closely related to North American Crangonyx species. C. subterraneus and C. chlebnikovi (both from Eurasia) are highly divergent genetically from the other Crangonyx species indicating an ancient separation between North American and Eurasian Crangonyx species. Phylogenies in Pa- per 4, based on the ITS2 and the complete dataset (ITS1, 5.8S, ITS2), are as well supporting a ancient separation of C. islandicus from the two Eurasian Crangonyx species. Thus, it is more likely that C. islandicus colonized Iceland from the Nearctic which harbors most of the Crangonyx species, probably via past land contacts with Greenland. C. islandicus have most probably colonized via the Greenland-Iceland transverse ridge which connected Greenland and Iceland until 15 or even 10 Mya, before its submersion (see Grimsson et al., 2007; Denk et al., 2010). This supports further that C. islandicus colonized Iceland before the onset of the last Ice Age and had to survive repeated glaciations in Iceland. The morphological analysis in Paper 5 shows that C. islandicus is closely related to C. subterraneus which is incongruent with the phylogenies based on the nuclear markers. The morphology tends to support a colonization of Iceland from Europe via the Iceland-Scotland part of the North Atlantic transverse ridge. Crangonyx species from Eurasia and from North America are highly divergent genetically and the absence of such split based on morphological data points to potential convergent evolution in morphological traits, particularly common in subterranean species, or to homoplasy in the characters used. Taxonomy of the Crangonyctoidea and convergence in morphology 29 a Crymostygius thingvallensis b Crangonyx islandicus Crymostygius thingvallensis Crangonyx forbesi

Crangonyx sp Crangonyx pseudogracilis

Crangonyx pseudogracilis Crangonyx islandicus Crangonyx floridanus

Synurella dentata Crangonyx forbesi Synurella sp

Amurocrangonyx arsenjevi Stygobromus gracilipes Synurella ambulans a-b Crangonyx chlebnikovi Bactrurus brachycaudus p<0.8 Bactrurus mucrunatus 0.8>p>0.95 Synurella ambulans p>0.95 Bactrurus pseudomucrunatus

Stygobromus mackini Stygobromus stegerorum

Stygobromus stegerorum 0.05 Crangonyx subterraneus Stygobromus gracilipes

Crangonyx subterraneus 0.01 Crangonyx chlebnikovi c

Crymostygius thingvallensis

Crangonyx islandicus

Stygobromus stegerorum Crangonyx subterraneus c

boot<80 Crangonyx chlebnikovi 80

Stygobromus gracilipes

0.01

Figure 3.2: Phylogenetic relationships among Crangonyctoidea species. a: Bayesian tree based on 18S gene, b: Bayesian tree based on 28S gene, c: Neighbor joining tree based on ITS1, 5.8S and ITS2. Posterior probabilities for trees a and b are displayed in box a-b. Bootstrap values for the tree c are displayed in box c. Crangonyx lineages are highlighted in red.

3.4 Taxonomy of the Crangonyctoidea and convergence in morphology

The taxonomic status of the two Icelandic groundwater amphipods based on morphology (Kristjánsson and Svavarsson, 2004; Svavarsson and Kristjánsson, 2006) are relatively supported by the phylogenies based on 30 Results and Discussion the nuclear markers in Paper 3 and 4. C. thingvallensis is clearly divergent from the other Crangonyctoidea species included in this study for phylogenies based on the 18S, 28S, and the ITS regions. Though the family Crymosty- gidae appears to be a sister family of the Crangonyctidae, larger phylogenies encompassing the 11 families of the superfamily Crangonyctoidea are needed in order to reveal the phylogenetic relationships among these families. Phylo- genies based on 18S, 28S and 16S (Paper 3) show C. islandicus to be closely related to the Crangonyx species, supporting its taxonomic status. However, the distances observed for the 18S and 28S between C. islandicus and the other Crangonyx species are more similar to distances between genera than within other amphipod genera. This tends to indicate that C. islandicus ﬁts poorly within the genus Crangonyx and should possibly be removed from that genus. Nonetheless, the tree based on morphological data in Paper 5 is not giving support to this taxonomic status. Based on morphology, C. islandicus clusters within Crangonyx in a separate branch with C. subterraneus, C. longicarpus, C. baculispina and C. packardi. Sequencing of these species could allow to test their relationships and provide an evaluation of the unique taxonomic status of C. islandicus.

In all phylogenies based on nuclear markers (Paper 3 and 4), the species from the Crangonyctidae family clustered in a monophyletic group. This supports the occurrence of a common ancestor of the family which was widespread in Laurasia before its breakup around 55 Mya, as hypothesized by Holsinger (1992). Based on 18S, 28S and ITS2 phylogenies, the Crangonyx genus is clearly polyphyletic, species from Eurasia being highly divergent from species in North America and in Iceland. Conversely, the morphological data is not supporting a clear split between species from these two geographic regions (Pa- per 5). The most likely hypothesis to explain this discrepancy is convergent evolution of the morphological traits, common among subterranean species. This might aﬀect the characters used to reconstruct the phylogeny based on morphology and leads to an underestimation of the evolutionary distances between North American and Eurasian species. Thus, the genus Crangonyx needs to be redescribed using a more integrative taxonomy (see Dayrat, 2005) i.e. by a combination of morphological, molecular and phylogeographic data. The Evolution of the ITS regions 31 genus Synurella is as well polyphyletic according to the phylogenies based on 18S and COI (Paper 3). Interestingly, Synurella species from North America are clustering with the North American Crangonyx and reciprocally, Synurella ambulans from Eurasia is more related to Eurasian Crangonyx than to other Synurella species. Finally, the monophyly of the genus Stygobromus is poorly supported by phylogenies based on 18S and 28S (depending on alignment methods) and is not supported by the variation in ITS regions. Consequently, three genera among the family Crangonyctidae i.e. Crangonyx, Synurella and Stygo- bromus might need a taxonomic revision, combining morphological and genetic data in order to restore their monophyly.

3.5 Evolution of the ITS regions

The variation of the ITS regions was investigated in this thesis within the species C. islandicus (Paper 2) and between species of the superfamily Cran- gonyctoidea (Paper 4). The ITS1 and ITS2 are two members of the rRNA gene family, which other members are 18S, 5.8S, 28S, ETS1, ETS2 and IGS (see Fig 2.5). The evolution of the RNA gene family is commonly explained by a model of concerted evolution. According to this model, each member of a gene family are evolving in a concerted manner and a mutation occurring in one member will either be purged or occasionally spread through all the members of the family by repeated unequal crossing-overs or gene conversions (Nei and Rooney, 2005). Concerted evolution leads to homogeneous copies within species but large variations between species (e.g. Brown et al., 1972). Paper 4 reports high variations in sequence and secondary structure among Crangonyc- tidae species whereas in Paper 2, low nucleotide variation was observed for the ITS1 and ITS2 regions within C. islandicus. Nonetheless, interesting patterns of evolution were found among C. islandicus populations. A duplication event occurred in the ITS1 region solely in the southern populations. The most likely hypothesis is a recent duplication, after the split of the northern and southern populations around 1 million year ago according to time estimates based on COI and 16S variations in Paper 1. The high divergence observed between the two duplicated regions is therefore pointing at a rapid divergent evolution 32 Results and Discussion where the duplicated gene may acquire a new function (Nei and Rooney, 2005), opposite to the homogenization effect of concerted evolution. The occurrence of the duplication in all individuals from the southern populations might be due to the effect of such divergent evolution. In paper 4, the ITS2 sequences, lengths and secondary structures are found to be highly variable between amphipod species. This appears incongruent with previous descriptions of the ITS2 secondary structure as highly conserved and presenting structures nearly universal among Eukaryotes (Schultz et al., 2005; Coleman, 2007). The number of helices in the secondary structure, seen as relatively constant within different groups of organisms (see Cole- man, 2007), was as well variable among the amphipod species studied. These discrepancies do not appear to result from pseudogenes as the high conservation of the 5.8S and of the ITS2 excision site among the Crangonyctoidea species tend to support the functionality of the ITS2 region amplified. Further investigations of the ITS2 secondary structure variation among amphipods are needed to identify potential homologies and conserved motifs specific to the group. Conclusion and Perspectives

The occurrence of the endemic groundwater amphipod C. islandicus in Iceland can be explained by an ancient colonization, which predates the onset of the last Ice Age, around 2.75 million years ago. Molecular data supports that the ancestor of C. islandicus colonized via past freshwater, or even groundwater, contacts between Greenland and Iceland occurring in the Greenland-Iceland land bridge which was still subaerial around 10 million years ago. C. islandicus thus survived repeated glacial cycles in Iceland. The genetic variation observed within C. islandicus populations is highly supporting the occurrence of several subglacial refugia located in fissure zones near the tectonic plate boundaries, probably maintained by the important geothermal activity of these areas. Though the hypothesis could not be tested for C. thingvallensis due to its restricted distribution, its only location in one of the main fissure area of Ice- land, lake Þingvallavatn, indicates that this species might have survived in the same refugia. This thesis demonstrates that the occurrence of subglacial refugia can explain phylogeographic patterns in groundwater amphipod species. Subglacial refugia have been hypothesized to explain biogeographic patterns in amphipod species in North America and Ireland. Considering the locations along the volcanic active zone in Iceland where the groundwater amphipods are found and the observed genetic patterns, fissure zones with groundwater possibly maintained by geothermal activity underneath glaciers may be required to sustain such refugia. Thus their occurrence could be limited to volcanic active regions. The survival of amphipod species in subglacial refugia indicates that a specific community has thrived in these refugia during glaciations. The ground-

33 34 Conclusion and Perspectives water fauna in Iceland is still poorly described, only three species being known so far (Juberthie et al., ress). Further phylogeographic studies on groundwater species in Iceland might give a better estimate of the biota which survived on the island. For example, several epibionts living on the cuticule of C. islandicus have been observed during this study. Preliminary molecular analysis showed that some of these species might belong to the Ciliophora phylum. These organisms, tightly linked to C. islandicus, could have found the amphipod after the Ice Age or might as well have survived glaciations in Iceland. C. islandicus were found in some places very near to the surface and hence can probably receive epibionts from the limnic environment. The identification of these epibionts by morphological and molecular analysis could shed light on the time of their colonization. In addition, bacterial communities are known to live in subglacial lake underneath the Vatnajökull glacier in Iceland (Gaidos et al., 2004, 2009). This community appears to be dominated by chemotrophic bacteria (Gaidos et al., 2009). Such bacterias could have been primary producers in the ecosystems which survived glaciations in the fissure zones. The species of Crangonyctidae included in this thesis clustered in a monophyletic group in all nuclear phylogenies produced for this thesis and as observed earlier by Englisch et al. (2003). The occurrence of an ancestor of this family, widespread on the Laurasia continent is therefore supported. Nonethe- less, the family Crangonyctidae is composed of 7 extant genera i.e. Crangonyx, Synurella, Bactrurus, Stygobromus, Stygonyx, Amurocrangonyx and Lyurella (see Messouli, 2006; Sidorov and Holsinger, 2007), and adding more taxa to the phylogeny is needed in order to conclude about the monophyly of the family. Due to genetic linkage of the 18S, 28S and ITS2, other nuclear markers should be integrated in future studies. Nuclear phylogenies revealed three genera of the Crangonyctidae as polyphyletic. Two of them, the genera Crangonyx and Synurella were separated into two highly divergent groups corresponding to Palearctic and Nearctic species. Consequently, these genera might need a taxonomic revision based on further molecular and morphological analysis. The morphological markers used did not lead to a distinction between Eurasian and North American Cran- Conclusion and Perspectives 35 gonyx species, while molecular phylogenies show an ancient divergence between the species of these two regions. This could be explained by convergent evolution, morphological stasis or homoplasy in the characters used, which lead to an underestimation of the divergence between species from Nearctic and Palearctic regions. The abundance of convergence or morphological stasis in morphological traits in subterranean species and the discordance between morphological and molecular affinities of Crangonyctidae species observed in this thesis highlight the need for a more integrative taxonomy (see Dayrat, 2005) for groundwater species. The split observed between Nearctic and Palearctic species of the Crangonyx and Synurella genera indicates that phylogeography is of particular interest in this integrative taxonomy approach. The phylogeographic pattern observed in these two genera might correspond to the opening of the North Atlantic ocean which separated Nearctic and Palearctic regions around 55 million years ago. This could be tested by producing phylogenies of the Crangonyx and Synurella genera based on molecular markers with calibrated substitution rates. Other genes than COI and 16S might have to be considered, since the variation of these genes observed in this thesis among Crangonyctidae species showed signs of saturation and gave poor phylogenetic resolution.

In this study, various molecular markers were used at different taxonomic levels. Depending on the taxonomic level, molecular markers with a too low evolution rate will lead to low phylogenetic signal whereas a too high evolution rate will cause a saturation of the signal which both influence the robustness and accuracy of the gene trees reconstructed. The mt genes COI and 16S and the nuclear ITS1 are shown to be informative at the species level, while they gave poor resolution when used to construct phylogenies encompassing different families. Conversely, almost no variation was found with the 18S, 28S and ITS2 at the species level, but these markers were suited to build phylogenies at the family level. Interestingly, the variation pattern among C. islandicus populations is highly different for the two DNA barcodes COI and ITS2. COI sequences were highly variable and the three species delimitation thresholds used led to different conclusions on the occurrence of cryptic species within C. islandicus. Conversely, the ITS2 showed almost no variation. 36 Conclusion and Perspectives

In order to test further the potential species complex of C. islandicus, morphometric analysis could be conducted. Such analysis have been shown to be potentially informative for species delimitation as well as for sex recognition (Wilhelm and Venarsky, 2009). This represents a particular interest for C. islandicus since only female individuals have been described so far, probably due to subtle morphological differences, biased sex ratios or different behaviors between sexes. The estimation of the groundwater biodiversity represents a challenge, both due to sampling difficulties and the abundance of cryptic diversity. As seen in this thesis, species considered congeneric might be highly divergent genetically and the use of integrative taxonomy is necessary to assess subterranean species diversity. A recent study showed that the use of morphospecies – species delim- ited solely by morphology – to assess the impact of the global climate change leads to an underestimation of the biodiversity loss and cryptic diversity should be considered in these estimations (Bálint et al., 2011). The increasing human water consumption, especially for agriculture, and the global climate change will have an certain affect on groundwater ecosystems (Danielopol et al., 2003) and better estimations of the groundwater biodiversity using molecular taxonomy are though needed. Furthermore, the identification of past refugia and the description of distribution changes of the genetic diversity over time are of primary concern to define future conservation units. Bibliography

Andrews, J. T., A. E. Jennings, A. E. Sveinbjörnsdóttir, J. Hardardóttir, G. B. Kristjánsdóttir, G. Helgadóttir, J. P. M. Syvitski, A. Geirsdóttir, S. Schoolfield, L. Smith and K. Thors, 2000. The N and W Iceland Shelf: insights into Last Glacial Maximum ice extent and deglaciation based on acoustic stratigraphy and basal radiocarbon AMS dates. Quaternary Science Reviews 19: 619–631.

Audzijonyte, A. and R. Väinölä, 2006. Phylogeographic analyses of a circumarctic coastal and a boreal lacustrine mysid crustacean, and evidence of fast postglacial mtDNA rates. Molecular Ecology 15: 3287–301.

Bálint, M., S. Domisch, C. H. M. Engelhardt, P. Haase, S. Lehrian, J. Sauer, K. Theissinger, S. U. Pauls and C. Nowak, 2011. Cryptic biodiversity loss linked to global climate change. Nature Climate Change 1: 313–318.

Bradley, R. K., A. Roberts, M. Smoot, S. Juvekar, J. Do, C. Dewey, I. Holmes and L. Pachter, 2009. Fast statistical alignment. PLoS Com- putational Biology 5: 1–15.

Brochmann, C., T. M. Gabrielsen, I. Nordal, J. Y. Landvik and R. Elven, 2003. Glacial Survival or tabula rasa? The History of North Atlantic Biota Revisited. Taxon 52: 417–450.

Brown, D., P. Wensink and E. Jordan, 1972. A comparison of the ribosomal DNA’s of and : the evolution of tandem genes. Journal of Molecular Biology 63: 57–73.

37 38 BIBLIOGRAPHY

Buchheim, M. A., A. Keller, C. Koetschan, F. Förster, B. Mer- get and M. Wolf, 2011. Internal Transcribed Spacer 2 (nu ITS2 rRNA) Sequence-Structure Phylogenetics: Towards an Automated Reconstruction of the Green Algal Tree of Life. PLoS ONE 6: 1–10.

Buckland, P. C., D. W. Perry, G. M. Gíslason and A. J. Dugmore, 1986. The pre-Landnám fauna of Iceland: a palaeontological contribution. Boreas 15: 173–184.

Chu, K. H., C. P. Li and H. Y. Ho, 2001. The ﬁrst internal transcribed spacer (ITS-1) of ribosomal DNA as a molecular marker for phylogenetic and population analyses in crustacea. Marine Biotechnology 3: 355–61.

Coleman, A. W., 2007. Pan-eukaryote ITS2 homologies revealed by RNA secondary structure. Nucleic Acids Research 35: 3322–9.

Coope, G., P. Moore and A. Gibbs, 1986. The Invasion and Colonization of the North Atlantic Islands: A Palaeoecological Solution to a Biogeographic Problem [and Discussion]. Philosophical Transactions of the Royal Society of London. B, Biological Sciences 314: 619–635.

Cunningham, C. W., N. W. Blackstone and L. W. Buss, 1992. Evolution of king crabs from hermit crab ancestors. Nature 355: 539–542.

Danielopol, D., C. Griebler, A. Gunatilaka and A. Notenboom, 2003. Present state and future prospects for groundwater ecosystems. En- vironmental Conservation 30: 104–130.

Danielopol, D. L., P. Pospisil and R. Rouch, 2000. Biodiversity in groundwater: a large-scale view. Trends in Ecology & Evolution 15: 223– 224.

Dayrat, B., 2005. Towards integrative taxonomy. Biological Journal of the Linnean Society 85: 407–415.

Denk, T., F. Grimsson and R. Zetter, 2010. Episodic migration of oaks to Iceland: Evidence for a North Atlantic "land bridge" in the latest Miocene. American Journal of Botany 97: 276–287. BIBLIOGRAPHY 39

Denk, T., F. Grimsson, R. Zetter and L. A. Símonarson, 2011. Intro- duction to the nature and geology of Iceland. In Late Cainozoic Floras of Iceland, volume 35 of Topics in Geobiology, pp. 1–29. Springer Netherlands, Dordrecht.

Drummond, A. J. and A. Rambaut, 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evolutionary Biology 7: 214–222.

Egilsdóttir, H. and B. Kristjánsson, 2008. Dvergbleikja í grennd við Jökulsá á Fjöllum. Náttúrufræðingurinn pp. 109–114.

Einarsson, E., 1963. The elements and aﬃnities of the Icelandic ﬂora. In North Atlantic biota and their history, edited by Löve, A. and D. Löve, pp. 297–302. Pergamon, Oxford.

Einarsson, T., 1994. Geology of Iceland. Rocks and Landscape. Mál & Men- ning, Reykjavik.

Englisch, U., C. O. Coleman and J. W. Wägele, 2003. First observations on the phylogeny of the families Gammaridae, Crangonyctidae, Melitidae, Niphargidae, Megaluropidae and Oedicerotidae (Amphipoda, Crustacea), using small subunit rDNA gene sequences. Journal of Natural History 37: 2461–2486.

Gaidos, E., B. Lanoil and T. Thorsteinsson, 2004. A viable microbial community in a subglacial volcanic crater lake, Iceland. Astrobiology 4: 327–345.

Gaidos, E., V. Marteinsson, T. Thorsteinsson, T. Jóhannesson, A. R. Rúnarsson, A. Stefansson, B. Glazer, B. Lanoil, M. Skid- more, S. Han, M. Miller, A. Rusch and W. Foo, 2009. An oligarchic microbial assemblage in the anoxic bottom waters of a volcanic subglacial lake. The ISME journal 3: 486–97.

Geirsdóttir, A., G. Miller and J. Andrews, 2007. Glaciation, erosion, and landscape evolution of Iceland. Journal of Geodynamics 43: 170–186. 40 BIBLIOGRAPHY

Gibert, J., A. Brancelj, A. Camacho, F. Castellarini, C. De Broyer, L. Deharveng, M. Dole-Olivier, C. Douady, D. Galassi, F. Malard and Others, 2005. Groundwater biodiversity. Protocols for the assessment and conservation of aquatic life in the subsurface (PASCALIS): overview and main results. In Proceedings of an International Symposium on World Subterranean Biodiversity, December, pp. 39–51.

Gibert, J., D. C. Culver, M.-J. Dole-Olivier, F. Malard, M. C. Christman and L. Deharveng, 2009. Assessing and conserving groundwater biodiversity: synthesis and perspectives. Freshwater Biology 54: 930– 941.

Grimsson, F., T. Denk and L. Simonarson, 2007. Middle Miocene ﬂoras of Iceland - the early colonization of an island? Review of Palaeobotany and Palynology 144: 181–219.

Guindon, S. and O. Gascuel, 2003. A Simple, Fast, and Accurate Algorithm to Estimate Large Phylogenies by Maximum Likelihood. Systematic Biology 52: 696–704.

Hanfling, B., I. Douerello-Soler, L. Knight and G. Proudlove, 2008. Molecular studies on the Niphargus kochianus group (Crustacea: Am- phipoda: Niphargidae) in Great Britain and Ireland. Cave and Karst Science 35: 35–40.

Hebert, P. D. N., A. Cywinska, S. L. Ball and J. R. DeWaard, 2003. Biological identiﬁcations through DNA barcodes. Proceedings of the Royal Society of London B: Biological Sciences 270: 313–21.

Hebert, P. D. N., M. Y. Stoeckle, T. S. Zemlak and C. M. Francis, 2004. Identiﬁcation of Birds through DNA Barcodes. PLoS Biology 2: 1657– 1663.

Hennig, W., 1966. Phylogenetic Systematics. University of Illinois Press, Urbana.

Hewitt, G., 2000. The genetic legacy of the Quaternary ice ages. Nature 405: 907–913. BIBLIOGRAPHY 41

Hewitt, G., 2004. Genetic consequences of climatic oscillations in the Qua- ternary. Philosophical Transactions of the Royal Society of London B 359: 183–195.

Higgs, P. G. and T. K. Attwood, 2005. Bioinformatics and molecular evolution. Wiley-Blackwell Publishing.

Holsinger, J., 1992. Sternophysingidae, a new family of Subterranean amphipods (Gammaridea: Crangonyctoidea) from South Africa, with description of sternophysinx calceola, new species, and comments on phylogenetic and biogeographic relationships. Journal of Crustacean Biology 12: 111–124.

Holsinger, J. R., 1980. Stygobromus canadensis, a new subterranean amphipod crustacean (Crangonyctidae) from Canada, with remarks on Wisconsin refugia. Canadian Journal of Zoology 58: 290–297.

Humphreys, W. F., T. Bradford, A. D. Austin, M. Adams and S. J. B. Cooper, 2010. DNA barcoding of stygofauna uncovers cryptic amphipod diversity in a calcrete aquifer in Western Australia’s arid zone. Molecular Ecology Resources 10: 41–50.

Ingólfsson, A., 2009. A marine refugium in Iceland during the last glacial maximum: fact or ﬁction? Zoologica Scripta 38: 663–665.

Juberthie, C., E. Kornobis and S. Pálsson, in press. Iceland. In ENCY- CLOPÆDIA BIOSPEOLOGICA, edited by Juberthie, C. and V. Decu, chapter Vol 1A, Eu, pp. 333–336. International Society for Subterranean Biology.

Katoh, K., K. Misawa, K.-i. Kuma and T. Miyata, 2002. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Research 30: 3059–66.

Keller, A., T. Schleicher, J. Schultz, T. Müller, T. Dandekar and M. Wolf, 2009. 5.8S-28S rRNA interaction and HMM-based ITS2 annotation. Gene 430: 50–7. 42 BIBLIOGRAPHY

Knowlton, N. and L. A. Weigt, 1998. New dates and new rates for divergence across the Isthmus of Panama. Proceedings of the Royal Society B: Biological Sciences 265: 2257–2263.

Knowlton, N., L. A. Weigt, L. A. Solorzano, D. K. Mills and E. Bermingham, 1993. Divergence in proteins, mitochondrial DNA, and reproductive compatibility across the Isthmus of Panama. Science 260: 1629–32.

Koenemann, S. and J. R. Holsinger, 2001. Systematics of the north american subterranean amphipo genus Bactrurus (Crangonyctidae). Beaufortia 51: 1–56.

Kornobis, E., S. Pálsson, B. K. Kristjánsson and J. Svavarsson, 2010. Molecular evidence of the survival of subterranean amphipods (Arthropoda) during Ice Age underneath glaciers in Iceland. Molecular Ecology 19: 2516– 2530.

Kristjánsson, B. K. and J. Svavarsson, 2004. Crymostygidae, a new family of subterranean freshwater gammaridean amphipods (Crustacea) recorded from subarctic Europe. Journal of Natural History 38: 1881–1894.

Kristjánsson, B. K. and J. Svavarsson, 2007. Subglacial refugia in Ice- land enabled groundwater amphipods to survive glaciations. The American Naturalist 170: 292–6.

Larkin, M. A., G. Blackshields, N. P. Brown, R. Chenna, P. A. McGettigan, H. McWilliam, F. Valentin, I. M. Wallace, A. Wilm, R. Lopez, J. D. Thompson, T. J. Gibson and D. G. Higgins, 2007. Clustal W and Clustal X version 2.0. Bioinformatics 23: 2947–8.

Lefébure, T., C. J. Douady, M. Gouy and J. Gibert, 2006. Rela- tionship between morphological taxonomy and molecular divergence within Crustacea: Proposal of a molecular threshold to help species delimitation. Molecular Phylogenetics and Evolution 40: 435–447.

Lefébure, T., C. J. Douady, F. Malard and J. Gibert, 2007. Testing dispersal and cryptic diversity in a widely distributed groundwater amphipod BIBLIOGRAPHY 43

(Niphargus rhenorhodanensis). Molecular Phylogenetics and Evolution 42: 676–86.

Malard, F. and F. Hervant, 1999. Oxygen supply and the adaptations of animals in groundwater. Freshwater Biology 41: 1–30.

Mathews, D. H. and D. H. Turner, 2006. Prediction of RNA secondary structure by free energy minimization. Current opinion in structural biology 16: 270–8.

Messouli, M., 2006. The ﬁrst African member of the Family Crangonyctidae, Crangonyx africanus, sp. nov.(Crustacea, Amphipoda) in the groundwaters of Western Morocco: systematics and biogeographical implications. Subter- ranean Biology 4: 67–78.

Mihalffy, P., B. Steinberger and H. Schmeling, 2008. The effect of the large-scale mantle flow field on the Iceland hotspot track. Tectonophysics 447: 5–18.

Moritz, C. and C. Cicero, 2004. DNA barcoding: promise and pitfalls. PLoS biology 2: 1529–1531.

Morrison, D. and J. Ellis, 1997. Eﬀects of nucleotide sequence alignment on phylogeny estimation: A case study of 18S rDNAs of Apicomplexa. Molecular Biology and Evolution 14: 428–441.

Morrison, D. a., 2008. A framework for phylogenetic sequence alignment. Plant Systematics and Evolution 282: 127–149.

Morrison, D. A., 2009. Why Would Phylogeneticists Ignore Computerized Sequence Alignment? Systematic Biology 58: 150–158.

Müller, T., N. Philippi, T. Dandekar, J. Schultz and M. Wolf, 2007. Distinguishing species. RNA 13: 1469–1472.

Müller, T., S. Rahmann, T. Dandekar and M. Wolf, 2004. Accurate and robust phylogeny estimation based on proﬁle distances: a study of the Chlorophyceae (Chlorophyta). BMC evolutionary biology 4: 20–26. 44 BIBLIOGRAPHY

Musters, W., K. Boon, C. van der Sande, H. V. Heerikhuizen and R. J. Planta, 1990. Functional analysis of transcribed ribosomal DNA spacers of yeast. EMBO Journal 9: 3989–3996.

Nei, M. and S. Kumar, 2000. Molecular evolution and phylogenetics. Oxford University Press, New York.

Nei, M. and A. Rooney, 2005. Concerted and Birth-and-Death Evolution of Multigene Families. Annual Review of Genetics 39: 121–152.

Norðdahl, H., O. Ingólfsson, H. G. Pétursson and M. Hallsdót- tir, 2008. Late Weichselian and Holocene environmental history of Iceland. Jökull 58: 343–364.

Poore, R. H., 2008. Neogene epeirogeny and the Iceland plume. Ph.D. thesis, University of Cambridge, Cambridge, UK.

Provan, J. and K. D. Bennett, 2008. Phylogeographic insights into cryptic glacial refugia. Trends in Ecology & Evolution 23: 564–571.

Roe, A. D. and F. A. H. Sperling, 2007. Population structure and species boundary delimitation of cryptic Dioryctria moths: an integrative approach. Molecular ecology 16: 3617–33.

Ronquist, F. and J. P. Huelsenbeck, 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574.

Rundgren, M. and O. Ingólfsson, 1999. Plant survival in Iceland during periods of glaciation ? Journal of Biogeography 26: 387–396.

Sadler, J., 1999. Biodiversity on oceanic islands: a palaeoecological assessment. Journal of Biogeography pp. 75–87.

Saitou, N. and T. Imanishi, 1989. Relative eﬃciencies of the Fitch- Margoliash, maximum-likelihood, minimum-evolution, and neighbor-joining methods of the phylogenetic tree construction. Molecular Biology and Evo- lution 6: 514–525. BIBLIOGRAPHY 45

Saitou, N. and M. Nei, 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Molecular Biology and Evolution 4: 406–25.

Schubart, C. D., R. Diesel and S. B. Hedges, 1998. Rapid evolution to terrestrial life in Jamaican crabs. Nature 393: 363–365.

Schultz, J., S. Maisel, D. Gerlach, T. Müller and M. Wolf, 2005. A common core of secondary structure of the internal transcribed spacer 2 (ITS2) throughout the Eukaryota. RNA 11: 361–4.

Seibel, P. N., T. Müller, T. Dandekar and M. Wolf, 2008. Synchronous visual analysis and editing of RNA sequence and secondary structure alignments using 4SALE. BMC research notes 1: 91.

Seidel, R. A., B. K. Lang and D. J. Berg, 2009. Phylogeographic analysis reveals multiple cryptic species of amphipods (Crustacea: Amphipoda) in Chihuahuan Desert springs. Biological Conservation 142: 2303–2313.

Shiklomanov, I., 1993. World fresh water resources. In Water in Crisis. A Guide to the World’s Fresh Water Resources, edited by Gleick, P., pp. 13–24. Oxford University Press, New York, USA.

Sidorov, D. A. and J. R. Holsinger, 2007. Amurocrangonyx, a New Genus of Subterranean Amphipod (Crangonyctidae) From the Russian Far East, With a Redescription of the Poorly Known Crangonyx Arsenjevi and Com- ments on Biogeographic Relationships. Journal of Crustacean Biology 27: 660–669.

Sigurdsson, H., 1970. Structural origin and plate tectonics of the snaefellsnes volcanic zone, Western Iceland. Earth and Planetary Science Letters 10: 129–135.

Stewart, J. and A. Lister, 2001. Cryptic northern refugia and the origins of the modern biota. Trends in Ecology \& Evolution 16: 608–613.

Stillman, J. H. and C. A. Reeb, 2001. Molecular phylogeny of eastern Paciﬁc porcelain crabs, genera Petrolisthes and Pachycheles, based on the 46 BIBLIOGRAPHY

mtDNA 16S rDNA sequence: phylogeographic and systematic implications. Molecular Phylogenetics and Evolution 19: 236–245.

Stocsits, R. R., H. Letsch, J. Hertel, B. Misof and P. F. Stadler, 2009. Accurate and eﬃcient reconstruction of deep phylogenies from structured RNAs. Nucleic Acids Research 37: 6184–93.

Svavarsson, J. and B. Kristjánsson, 2006. Crangonyx islandicus sp. nov., a subterranean freshwater amphipod (Crustacea, Amphipoda, Crangonycti- dae) from springs in lava ﬁelds in Iceland. Zootaxa 1365: 1–17.

Taberlet, P., L. Fumagalli and A. Wust-Saucy, 1998. Comparative phylogeography and postglacial colonization routes in Europe. Molecular Ecology 7: 453–464.

Tippery, N. P. and D. H. Les, 2008. Phylogenetic analysis of the internal transcribed spacer (ITS) region in Menyanthaceae using predicted secondary structure. Molecular phylogenetics and evolution 49: 526–37.

Trontelj, P., C. J. Douady, C. Fišer, J. Gibert, v. Gorički, T. Lefébure, B. Sket and V. Zakšek, 2009. A molecular test for cryptic diversity in ground water: how large are the ranges of macro-stygobionts? Freshwater Biology 54: 727–744.

Väinölä, R., J. D. S. Witt, J. H. Bradbury, K. Jazdzewski, M. Grabowski and B. Sket, 2008. Global diversity of amphipods (Am- phipoda; Crustacea) in freshwater. Hydrobiologia 595: 241–255. van Nues, R. W., J. M. Rientjes, S. A. Morré, E. Mollee, R. J. Planta, J. Venema and H. A. Raué, 1995. Evolutionarily conserved structural elements are critical for processing of Internal Transcribed Spacer 2 from Saccharomyces cerevisiae precursor ribosomal RNA. Journal of Molecular Biology 250: 24–36.

Wares, J. P. and C. W. Cunningham, 2001. Phylogeography and historical ecology of the North Atlantic intertidal. Evolution 55: 2455–69. Wattier, R. A., J. Beguet, M. Gaillard, J. C. Muller, L. Bollache and M. J. Perrot-Minnot, 2006. Molecular markers for systematic iden- tiﬁcation and population genetics of the invasive Ponto-Caspian freshwater gammarid Dikerogammarus villosus (Crustacea, Amphipoda). Molecular Ecology Notes 6: 487–489.

Wilhelm, F. M. and M. P. Venarsky, 2009. Variation in Gnathopod Mor- phology of Cave Amphipods and Its Use in Determination of Sex. Journal of Crustacean Biology 29: 26–33.

Wolf, M., B. Ruderisch, T. Dandekar, J. Schultz and T. Müller, 2008. ProfDistS: (proﬁle-) distance based phylogeny on sequence–structure alignments. Bioinformatics 24: 2401–2.

Yang, Z., 2006. Computational Molecular Evolution. Oxford University Press, New York.

Zhang, J. and J. R. Holsinger, 2003. Systematics of the Freshwater Amphi- pod Genus Crangonyx (Crangonyctidae) in North America. Virginia Museum of Natural History Memoirs.

Zuckerkandl, E. and L. Pauling, 1965. Evolutionary divergence and convergence in proteins. In Evolving genes and proteins, edited by Bryson, V. and H. J. Vogel, pp. 97–166. Academic Press, New York.

Ægisdóttir, H. and Þ. Þórhallsdóttir, 2004. Theories on migration and history of the North-Atlantic ﬂora: a review. Jökull 54: 1–16.

Paper 1

Molecular evidence of the survival of subterranean amphipods (Arthropoda) during Ice Age underneath glaciers in Iceland

Molecular Ecology (2010) 19, 2516–2530 doi: 10.1111/j.1365-294X.2010.04663.x

Molecular evidence of the survival of subterranean amphipods (Arthropoda) during Ice Age underneath glaciers in Iceland

ETIENNE KORNOBIS,* SNÆBJO¨ RN PA´ LSSON,* BJARNI K. KRISTJA´ NSSON† and JO¨ RUNDUR SVAVARSSON* *Department of Biology, University of Iceland, Askja, Sturlugata 7, 101 Reykjavik, Iceland, †Department of Aquaculture and Fish Biology, Ho´lar University College, Ha´eyri 1, 550 Sauða´rkro´kur, Iceland

Abstract Two endemic groundwater arthropod crustacean species, Crangonyx islandicus and Crymostygius thingvallensis, were recently discovered on the mid-Atlantic volcanic island of Iceland. The extent of morphological differences from closest relatives, endemism, along with the geographic isolation of Iceland and its complete coverage by glaciers 21 000 years ago, suggests that these two species have survived glaciation periods in sub-glacial refugia. Here we provide strong support for this hypothesis by an analysis of mitochondrial genetic variation within Crangonyx islandicus. Our results show that the species is divided into several distinct monophyletic groups that are found along the volcanic zone in Iceland, which have been separated by 0.5 to around 5 million years. The genetic divergence between groups reflects geographic distances between sampling sites, indicating that divergence occurred after the colonization of Iceland. The genetic patterns, as well as the dependency of genetic variation on distances from the tectonic plate boundary and altitude, points to recent expansion from several refugia within Iceland. This presents the first genetic evidence of multicellular organisms as complex as crustacean amphipods which have survived glaciations beneath an ice sheet. This survival may be explained by geothermal heat linked to volcanic activities, which may have maintained favourable habitats in fissures along the tectonic plate boundary in Iceland during glaciations.

Keywords: Crustacea, glaciation, phylogeography, sub-glacial refugia, subterranean, volcanism Received 25 January 2010; revision received 29 March 2010; accepted 3 April 2010

preceding expansion of populations after the Weichs- Introduction elian glaciation period. Conversely, the refugia are char- Pleistocene glaciations are known to have shaped both acterized by high taxonomic and genetic diversity the distribution and the genetic diversity of species at among and within regions (Hewitt 2004). high latitudes (Hewitt 2004). During glacial expansions, During the last glacial maximum (LGM) around populations either became extinct or managed to sur- 21 000 BP, Iceland, as well as most of northern Europe vive in ice-free refugia. High-latitude areas and for- and North America (Ehlers & Gibbard 2007), was merly glaciated northern parts of Europe and North almost completely covered by glaciers which extended America are, thus, characterized by low endemism and far offshore so that only small ice-free areas may have species diversity, shallow genetic clades and little diver- been present along coastal mountains, mainly in eastern sity within species (Sadler 1999; Hewitt 2004). This is and northern regions (Geirsdo´ttir et al. 2007). The low expected because of recent colonization and bottlenecks species diversity and lack of endemism in the current Icelandic terrestrial biota (e.g. Buckland et al. 1986) can Correspondence: Etienne Kornobis; E-mail: [email protected] be explained by the short time since the retreat of the

2010 Blackwell Publishing Ltd SUB-GLACIAL SURVIVAL OF CRANGONYX ISLANDICUS 2517 glaciers and the geographical isolation of the island. ments. The biogeographical argument is based on the Recent discoveries of two endemic groundwater amphi- distribution of subterranean amphipods, which are not pod species (Arthropoda, Crustacea) in Iceland (Kris- found at such high latitudes elsewhere in Northeast tjańsson & Svavarsson 2007)—Crangonyx islandicus, America, Greenland or Scandinavia (see Kristjańsson & belonging to the family of groundwater amphipods Svavarsson 2007). This could be explained by a more Crangonyctidae, and Crymostygius thingvallensis, falling widespread occurrence of the group before glaciations within the monotypic family Crymostygiidae—has with few species surviving at high latitudes. According raised questions concerning the survival of multicellular to the taxonomical argument, the two endemic spe- species under a permanent ice sheet. cies—one representing a monotypic family—are found Subterranean amphipod species are characterized by in an area with extremely low endemism due to recent relatively small geographic ranges and high levels of colonizations after the last glaciation. Finally, the lim- endemism (Porter 2007), a pattern which may have ited ability of groundwater amphipods to colonize new resulted from evolution in extreme habitats and led to areas, especially oceanic islands, supports a colonization poor dispersal capabilities, specific adaptations and event during an ancient contact between Icelandic and regressive traits in stygobiont species (e.g. Darwin 1860; continental groundwater. Vandel 1964; Culver et al. 1995). Recent genetic studies The aim of this study was to verify whether have revealed a considerable cryptic diversity and the C. islandicus existed in Iceland during the Pleistocene or genetic patterns have been explained by a combination whether it colonized Iceland after the last glaciation. To of vicariance and dispersal events (Verovnik et al. 2004; test these hypotheses we analyse mtDNA sequence var- Lefe´bure et al. 2006a; Finston et al. 2007; Porter 2007; iation within C. islandicus, with respect to geographical Culver et al. 2009). Under the vicariance model, a prim- and geological features and the climatological history of itively cosmopolitan species is fragmented by a barrier Iceland. Firstly we estimate the age of the different and evolutionary divergence ensues, whereas the dis- cladogenetic events within the species and secondly persal model involves a barrier outside the range of the whether they follow geographic patterns within Iceland. species which is crossed by some of its members, A relationship of genetic distance with geographic dis- founding a new population in a previously unoccupied tance indicates that the divergence occurred within Ice- area (cf. Thornton 1983). The existence of endemic sub- land. The time to the most recent ancestor would then terranean species in previously glaciated areas can give a minimum age for the colonization of Iceland and reflect post-glacial colonization, dispersal from ice-free can be calculated using the global molecular clock regions or survival in sub-glacial refugia (Holsinger approach (Zuckerkandl & Pauling 1965). Thirdly, in 1980; Proudlove et al. 2003; Lefe´bure et al. 2007). order to detect cryptic refugia within Iceland we ana- Located on the Mid-Atlantic Ridge, Iceland has been lyse the genetic variation and examine if it is dependent shaped by plate tectonics and a mantle plume (Sigm- on age of the sampling site (lava field), altitude and dis- undsson & Sæmundsson 2008). The youngest part of tance from the plate boundaries. If the refugia are Iceland consists of two major zones, the western and found within fissures along the volcanic zone we expect the eastern volcanic zones, which are characterized by diversity to decrease with distance from the boundary fissures, volcanism and recent lava fields, rich in of the tectonic plates. In addition, we expect low diver- groundwater. The fissures may have provided a habit- sity below 100 m due to higher sea level during the end able environment for groundwater organisms even dur- of the last glaciation. Finally, we evaluate whether the ing the glaciation periods, due to geothermal activities. mtDNA lineages define any cryptic species by consider- C. islandicus has been found in groundwater springs ing the reciprocal monophyly and the extent of diver- throughout the active volcanic zone in Iceland, whereas gence among groups, applying two recent criteria only five specimens of C. thingvallensis have been found developed by Lefe´bure et al. (2006b) and Witt et al. (Kristjańsson & Svavarsson 2007). Kristjańsson & Sva- (2006). varsson (2007) hypothesized that these two species could be found in Iceland during the Pleistocene or Methods even since Iceland separated from Greenland, up to 40 million years ago (Torsvik et al. 2001; Lundin & Dore´ Data collection 2002). Considering that Iceland was repeatedly covered by glaciers during the Ice Age (around 2.5 million years Samples were obtained in 23 groundwater springs ago until 15 000 years ago, see Geirsdo´ttir et al. 2007), (Fig 1), throughout the active volcanic zone of Iceland, the two species had to survive in sub-glacial refugia. in lava fields less than 10 000 years old (Holocene) and Kristjańsson & Svavarsson (2007) supported their claim at different altitudes (Table 1). No specimens were by biogeographical, taxonomical and biological argu- found in the separate volcanic zone of the Snæfellsnes

2010 Blackwell Publishing Ltd 2518 E. KORNOBIS ET AL.

Fig. 1 Sampling locations of groundwa- F 20 23 ter amphipods in Iceland. The dotted 21 22 E squares represent the different groups A, A¢, B, C, D, E and F. The volcanic 16 active zone is displayed in grey. Gla-

18 ciers are in light grey. Longitudes and latitudes are shown in degrees on the x- 17 and y-axes respectively. 19

A’ C

8 1 Latitude (°N)

5 B 2 D 3 11 7 10 6 9 4 15 13 A 14

12 63.5 64.5 65.5 66.5

24 22 20 18 16 14 Longitude (°W)

Table 1 Geological data for each sampling site of C. islandicus DNA was extracted using 6% Chelex 100 (Bio-Rad) populations in Iceland. Locations (Loc.) are shown in Fig. 1. from whole specimens. Two regions of the mitochon- Altitude (Alt.) is given in metres, distance from the plate drial genome were amplified, 658 bp of cytochrome oxi- boundaries (PB) in kilometres and age of the lava fields (Lava) dase subunit I gene (COI) using primers LCO1490 and in years. GPS coordinates for latitude and longitude are given HCO2198 (Folmer et al. 1994), and 414 bp of the large in decimal degrees. * Data not available. ** Approximate val- subunit 16S RNA gene using primers 16Stf (MacDonald ues et al. 2005) and 16Sbr (Palumbi et al. 1991). PCR ampli- Loc. Latitude Longitude Alt. (m) PB (km) Lava (y) fications for both genes were performed in 10 lL, containing 0.15 mM dNTPs, 0.1% Tween, 1 Taq buffer, ) 1 64.702 20.897 120 45.4 8465 0.35 lM of each primer, 0.5 units of Taq and 10–100 ng ) 2 64.241 21.053 110 13.6 9000 of template. The PCR program used for the COI gene ) 3 64.142 21.042 110 8.1 9680 amplification was: 94 C 4 min, followed by 39 cycles of 4 63.919 )21.322 10 15.2 4880 94 C30s,45C45s,72C 1 min, with a final elonga- 5 64.300 )20.517 200 17.1 3850 6 63.958 )20.266 90 63.2 8000 tion step at 72 C for 6 min. The annealing temperature 7 64.008 )19.919 130 80.6 8000 for the 16S gene amplification was increased to 55 C 8 64.744 )19.432 560 7.1 7200 and the number of cycles reduced to 34. ExoSAP puri- 9 63.886 )19.483 640 25.6 * fied PCR products were sequenced in both directions ) 10 63.999 18.415 480 17.7 685 using ABI BigDye Terminator v3.1 (Applied Biosys- ) 11 64.002 18.403 480 17.7 685 tems). Sequences were then carried out on ABI 12 63.617 )18.500 70 50.7 226 PRISMTM 3100 Genetic Analyser and raw sequences 13 63.782 )18.050 30 56.5 226 14 63.749 )17.945 20 65.2 226 were edited and aligned by eye using BioEdit 7.0.9.0 15 63.883 )17.750 30 61.9 226 (Hall 1999). Sequences have been submitted to GenBank 16 65.954 )17.545 10 100.7 2300 under accession numbers from HM015145 to HM015196 17 65.347 )17.233 380 63.0 8500 (Table S3, Supporting Information). 18 65.556 )16.972 270 36.6 2300 19 65.193 )16.226 480 46.4 2800 20 66.261 )16.400 180 25** 6000 Phylogenetic analysis and time estimation 21 66.269 )16.400 10 25** 6000 22 66.267 )16.400 10 25** 6000 The potential loss of phylogenetic signal due to substi- 23 66.367 )15.917 120 35** 6000 tution saturation (Lopez et al. 1999; Philippe & Forterre 1999) was investigated using the index proposed by Xia et al. (Xia et al. 2003) and implemented in DAMBE peninsula in western Iceland. The amphipods were col- (Xia & Xie 2003). The optimal model of substitution was lected in dip nets after applying electricity with electric selected independently for COI and for 16S with fishing gear. Specimens were stored in 96% ethanol. MrAIC.pl v1.4.3 (Nylander 2004), according to the

2010 Blackwell Publishing Ltd SUB-GLACIAL SURVIVAL OF CRANGONYX ISLANDICUS 2519

Akaike and Bayesian information criterion. The molecu- 1.1 4.1 lar clock hypothesis was tested by a maximum likeli- 3.1 0.24● 2.1 A PAML hood ratio test using 4 (Yang 2007) while both the ● 0.7< P < 0.8 6.1 5.1 ● P > 0.95 COI and 16S trees tested were rooted with the Crangon- 0.43● 10.2 yx pseudogracilis outgroup. For the time estimations, the 9.2 2.2 ● interval of divergence rate was assumed to be 1.40– 0.23 5.2 % 6.2 A' 2.60 per million years for the COI (Knowlton et al. 3.2 0.66● 1.1 1993; Knowlton & Weigt 1998; Schubart et al. 1998), but 7.1 a broader interval (0.53–2.20%) was used for the diver- 8.1 1.2 gence rate of 16S (Cunningham et al. 1992; Stillman & 1.1 1.03 2.1 Reeb 2001). ● 3.1 B ● The phylogeny and divergence dates within C. is- 0.49 4.3 5.3 landicus were simultaneously estimated using BEAST 1.1 C 1.1 v1.4.7 (Drummond & Rambaut 2007) based on separate 0.35● 2.2 ● 3.4 partition for each locus in order to take into account the 1.34● 6.3 D different substitution models and substitution rates for 3.3 5.3 the two loci. The time estimations were based on varia- 4.3 2.1 tion within the COI and 16S genes and were calculated 1.1 4.87 0.18● 3.1 E according to the strict molecular clock model with the 1.2 two divergence rate intervals described above. The pop- 4.2 1.1 ulation growth model of C. islandicus was studied by 0.22● 2.3 2.2 F comparing constant size, exponential and logistic 0.5 3.3 growth models. The simple constant size model was used as no significant differences were detected in the Fig. 2 Bayesian phylogenetic tree of C. islandicus populations in Iceland. The tree was reconstructed from unique COI and likelihood values of the trees or in the time estimations 16S haplotypes found in 97 individuals of C. islandicus. The between the different models. The Jeffrey prior was number at each node represents the time to the most recent used for the construction of the tree, since no informa- common ancestor (TMRCA) of the monophyletic groups in tive prior distribution for the constant population size millions of years based on the COI and 16S gene fragments was available (Drummond & Rambaut 2007). After using a clock rooted tree in BEAST. Posterior probability val- operators tuning, the effective sample sizes were ues are displayed in the box. Branches are drawn to scale, with checked using Tracer (Rambaut & Drummond 2003); the bar indicating 0.5 nucleotide substitution per hundred sites. Haplotypes numbers are displayed in grey and numbering are for each parameter they largely exceeded 200, indicating dependent of the monophyletic groups (First number = COI, a good sampling of the posterior distribution of those second number = 16S haplotype plus see Tables S2 and S3, parameters (see Drummond & Rambaut 2007). Follow- Supporting Information). ing the preliminary operator tuning, the analyses were re-run for 10 million generations, sampled every 1000 generations. A 10% burn-in was discarded from the between sample size and estimated genetic diversity. beginning of each run after checking the parameter Pairwise distances corrected with the TN93 model traces for equilibrium. Independent runs gave similar (Tamura & Nei 1993) were used to characterize the results. The tree in Fig. 2 was obtained using the maxi- genetic structure and diversity within and among C. is- mum clade credibility option in BEAST and was rooted landicus populations. with the direct molecular clock rooting method (Huel- To characterize the structure among C. islandicus pop- senbeck et al. 2002). ulations, F and F statistics were computed using Arle- quin 3.1 (Excoffier et al. 2005). The observed variation was partitioned among the main monophyletic groups Genetic diversity and population structure defined by the genealogy (see below) among sampling Genetic diversity indices (i.e. number of haplotypes, sites within a monophyletic group and within sampling segregating sites, nucleotide diversity and standard sites. The AMOVA was based both on pairwise differences error) were calculated with the APE package (Paradis among haplotypes and on haplotype frequencies. et al. 2004) in R 2.9.0 (R Core Development Team 2005) The relationship of the COI gene haplotypes and their and with MEGA 4 (Tamura et al. 2007). Considering geographic distribution was summarized with the most the small number of individuals which could be col- parsimonious haplotype network, using the program lected in most locations, the effect of sample size on TCS (Clement et al. 2000) and redrawn in R. The associa- diversity was evaluated by calculating the correlation tion of genetic divergence between populations and

2010 Blackwell Publishing Ltd 2520 E. KORNOBIS ET AL. geographic distances was investigated further on two R and were compared to the species delimitation different scales. Firstly, net genetic distances (Nei & Li threshold of 0.16 subst. ⁄ site proposed by Lefe´bure et al. 1979) and geographical distances among sampling sites (2006b), based on patristic distances between Crusta- were compared. Secondly, in order to reduce the risk of cean species. 16S patristic distances will not be analysed genealogical autocorrelation, the mean genetic distances here due to the weak power of 16S sequences variation between monophyletic groups, as defined by the phylo- to infer species delimitation (Lefe´bure et al. 2006b). genetic tree, were compared, while the average geographical coordinates were obtained by taking the Results mean of the coordinates of the sampling localities within each monophyletic group. Pearson’s correlations Sequence variation (r) between geographic and genetic distances were calculated and tested with the non-parametric Mantel test In total, 128 sequences of 658 bp were obtained for the (Mantel 1967) with 1000 permutations using the ade4 COI gene and 103 of 414 bp for the 16S gene; 97 indi- package in R (Dray et al. 2007). viduals were sequenced for both genes, representing an The Tajima D (Tajima 1989) and Fu’s FS (Fu 1997) average of four individuals per sampling location. Dif- tests of selective neutrality, along with mismatch distri- ferent diversity indices within each location and for all butions (Rogers & Harpending 1992), were calculated to samples are summarized in Table 2. Thirty-five distinct study the recent demographic pattern for each mono- haplotypes were found at the COI gene and eighteen phyletic group using Arlequin 3.1. The observed mis- haplotypes at the 16S gene (Table 3). In total, 101 segre- match distribution was fitted both with demographic gating sites are observed (12% of the sites at the COI and spatial expansion models. The time since recent and 24% of the sites at the 16S). expansion was calculated from the s values with substi- Based on the Bayesian information criterion, the best- tution rates given per year. The probability values of fit model was the HKY+I model for the COI gene and the independent tests were adjusted by the Bonferroni the HKY+G for the 16S (Hasegawa et al. 1985). The method (e.g. Cabin & Mitchell 2000). TN93 model (Tamura & Nei 1993), the most similar In order to identify putative refugia in Iceland, multi- substitution model to HKY and implemented in R, was ple regression was used to analyse whether genetic used for pairwise distance calculations. The total diversity was dependent on altitude, distance from the genetic diversity is more than twice as high for COI tectonic plate boundary or age of the lava field. Genetic than 16S (Table 3), reflecting the difference of substitu- diversity at each location was scaled by the maximal tion rate between the two genes within crustaceans diversity observed within the respective monophyletic (Lefe´bure et al. 2006b). Genetic diversity within sam- group. The tectonic plate boundaries were estimated by pling localities is generally low for both genes com- taking the mean between the coordinates that delimit pared to the overall diversity (Tables 2 and 3). The the active volcanic zone (Sigmundsson & Sæmundsson highest values for the COI are observed in locations 1, 2008). The age of the lava fields was obtained from Ice- 2, 3, 10 and 11 (Fig. 1). The 16S nucleotide diversity at landic Geosurvey data (ISOR) and from Sinton et al. each location was uncorrelated (r = 0.26, P > 0.25) to (2005). Altitudes of the sampling points were obtained the sampling effort, whereas COI nucleotide diversity is from the National Land Survey of Iceland. dependent on the sample size (r = 0.51, P = 0.018). The correlation disappears, though, when location 2 is excluded from the analysis (r = )0.02, P > 0.92). Cryptic diversity According to the index developed by Xia et al. (2003), Inter-clade and intra-clade divergences were calculated neither gene showed substantial saturation (P < 0.001), based on unique COI haplotypes. Corrected distances even at the third position of the codon for the COI (TN93 model) were used to assess if C. islandicus popu- gene, indicating that sequences obtained contain sub- lations are composed of several provisional species as stantial phylogenetic information. Amino-acid sequence defined by Hebert et al. (2004) and Witt et al. (2006). translation (with invertebrate mitochondrial code) was The genetic distance between provisional species was unambiguous since no gaps and no nonsense codons defined to be 10 larger than the average intra-popula- have been observed. As noticed earlier by Lefe´bure tion distances (Witt et al. 2006). The species-screening et al. (2006b) for crustaceans, the amino-acid variation threshold was calculated from the average intra-mono- of the CO1 gene is low at the intra- and interspecies phyletic group divergence among unique haplotypes comparison level. In our dataset, only 6 of 219 amino and compared to the divergence among groups. acids sites are variable, two of them being singletons. Patristic distances based on the Bayesian tree of the The DNA sequences evolve according to the strict COI gene obtained with BEAST were calculated using molecular clock null hypothesis (P > 0.65 for both

2010 Blackwell Publishing Ltd SUB-GLACIAL SURVIVAL OF CRANGONYX ISLANDICUS 2521

Table 2 Genetic diversity within sampling locations of C. islandicus populations in Iceland, for COI, 16S and both combined. N: total number of sequences, K: number of different haplotypes, S: number of segregating sites, p: nucleotide diversity (TN93 corrected)

COI 16S COI-16S location NKS p (100) N K S p (100) N p (100) Groups

1 3 3 3 0.31 3 2 1 0.16 3 0.25 A’ 2 35 14 22 0.77 21 4 3 0.18 20 0.50 A&A’ 3 4 4 11 0.90 4 2 1 0.12 4 0.60 A&A’ 4 7 1 0 0.00 4 1 0 0.00 3 0.00 A 5 8 4 5 0.22 8 2 1 0.10 8 0.17 B 6 4 1 0 0.00 5 2 1 0.15 4 0.06 B 7 5 2 1 0.06 5 2 1 0.15 4 0.11 B 8 1 1 0 0.00 1 1 0 0.00 1 0.00 C 9 1 1 0 0.00 1 1 0 0.00 1 0.00 D 10 2 2 3 0.46 2 2 1 0.24 2 0.38 D 11 9 3 4 0.26 8 2 1 0.13 8 0.22 D 12 6 2 1 0.05 5 1 0 0.00 5 0.04 D 13 5 1 0 0.00 4 2 1 0.16 3 0.06 D 14 5 2 1 0.06 4 2 1 0.12 4 0.09 D 15 4 1 0 0.00 4 1 0 0.00 4 0.00 D 16 6 1 0 0.00 4 2 1 0.12 4 0.05 E 17 7 2 2 0.09 5 1 0 0.00 5 0.07 E 18 4 2 1 0.08 4 1 0 0.00 4 0.05 E 19 5 2 1 0.06 5 2 1 0.10 4 0.05 E 20 1 1 0 0.00 1 1 0 0.00 1 0.00 F 21 2 1 0 0.00 2 1 0 0.00 2 0.00 F 22 1 1 0 0.00 1 1 0 0.00 1 0.00 F 23 3 2 2 0.20 3 2 1 0.24 2 0.28 F

Table 3 Diversity indices within groups of C. islandicus populations in Iceland for COI, 16S and both combined within monophyletic groups. N: total number of sequences, K: number of different haplotypes, S: number of segregating sites, p: diversity indice (TN93 corrected), p SE: p standard error, s: time since expansion parameter

COI 16S COI-16S

Groups NKSp (100) nSE s NKSp (100) p SE s NKS p (100) p SE

A 21 6 8 0.29 0.11 6.37 11 2 1 0.04 0.04 0.00 20 6 8 0.24 0.09 A’ 28 10 11 0.35 0.13 3.66 21 4 3 0.15 0.11 0.55 10 9 11 0.27 0.09 B 17 5 6 0.12 0.05 0.00 18 2 2 0.14 0.10 0.81 16 6 8 0.14 0.05 C 1 1 0.00 0.00 0.00 1 1 0.00 0.00 0.00 1 1 0.00 0.00 D 32 6 7 0.22 0.10 3.92 28 3 3 0.18 0.12 0.89 27 7 10 0.22 0.08 E 22 4 4 0.11 0.06 0.94 18 2 1 0.15 0.13 0.81 17 5 5 0.13 0.06 F 7 3 3 0.16 0.09 1.27 6 4 3 0.29 0.18 1.54 6 4 6 0.22 0.09 Total 128 35 79 2.14 0.04 103 18 22 1.03 0.02 97 38 101 1.75 0.04 genes). As this study is based on intra-specific variation and F. With the two genes COI and 16S combined, the and possibly on recent cryptic species divergence, sig- five monophyletic groups are well supported (pp:pos- nificant substitution rate differences between lineages terior probabilities for the main groups are always are unlikely to be observed. ‡0.95 except for group BC) and relatively well differentiated. The F group diverged from the other groups around 5 million years ago, suggesting a new potential Phylogenetic structure and time estimation cryptic species. The F group differs from the other The genotypes clustered in a well-resolved genealogy groups at two (most of A, A¢, B, and E haplotypes) and (Fig. 2), comprising five monophyletic groups: A and three (for D, C) amino-acid sites. Despite large genetic A¢ joined, B and C (only one individual) joined, D, E divergence within C. islandicus, the amphipods are

2010 Blackwell Publishing Ltd 2522 E. KORNOBIS ET AL. clearly more related to each other than to other species distinct area (Figs 1 and 2). Haplotypes are shared only of the Crangonyctidae family which have been among locations which are relatively close to each sequenced for COI and 16S (the mean genetic distance other. A, A¢ and B groups are found in the western vol- with C. pseudogracilis is 36.52% with the TN93 model). canic zone in Southwest Iceland (sampling locations 1– The time to the most recent common ancestor 7). Sampling points 2 and 3 are located in a rift valley (TMRCA) of the other groups, excluding F, ranged from at the large, spring-fed Lake Thingvallavatn. Sampling 0.79 to 2.04 million years (Fig. 2 and Table S1, Support- point 5 is located nearby but outside this valley. C is ing Information). The E group diverged early on from from central Iceland (sampling point 8). D and E are all the other monophyletic groups but the cluster of A– found in the eastern volcanic zone, D in southern Ice- D groups is less well supported (pp = 0.79) than the land (sampling points 9–15) and E in northern Iceland clustering of each group (pp > 0.95). The haplotype (sampling points 16–19). Sampling points 7 and 9 are divergence time observed within each monophyletic separated by a large volcanic ridge, Mount Hekla. The group defined above (A, A¢, BC, D, E, F) ranges from F group is located in Northeast Iceland (sampling 0.18 to 0.49 million years. The highest posterior density points 20–23). The analyses of molecular variance based (HPD) intervals are wide, but the lowest values of the on pairwise differences for the COI and 16S combined HPD for the TMRCA always exceed 21 000 years BP, the (Table 4) show that 84.9% of the variation is observed time to the last maximum glaciation (see Table S1, Sup- among the monophyletic groups which are confined to porting Information). Recently some concerns have been specific areas. The remaining variation is mainly raised on the time dependency of molecular rates explained (10.5%) by the genetic diversity within loca- (Penny 2005; Ho & Larson 2006; Ho et al. 2007). Molec- tions, while the genetic variation within groups between ular rates observed on genealogical time-scales (<1 mil- locations accounts for only 4.6% of the total variation. lion years) are at least an order of magnitude greater A considerably lower genetic differentiation was than those measured over geological time-scales observed among monophyletic groups when the analy- (>1 million years) (Ho & Larson 2006); this has been sis was based on haplotype frequencies (14.6%), as it observed in various species and for different mitochon- does not take into account the large nucleotidic diver- drial genes (e.g. Audzijonyte & Vaïno¨la¨ 2006; Ho et al. gence among the groups. However, the variation among 2007; Millar et al. 2008). The time estimations of intra- locations within groups is large (19.6%), possibly specific divergence provided herein may therefore be reflecting a recent divergence of samples with closely overestimated. related haplotypes. Different clustering was obtained for the COI and 16S Genetic patterns within regions show a certain depen- sequences analysed separately (see Figs. S1 and S2, dency on geological features. Haplotypes from groups Supporting Information). The C group, represented by A and A¢ were both found in springs emerging into only one individual, presents a unique CO1 haplotype Lake Thingvallavatn (sampling locations 2 and 3), different from the haplotypes of the A, A¢ and B showing a high diversity. North of the lake (location 1), groups, whereas its 16S haplotype is identical to an A¢ only haplotypes from the A¢ group were found whereas haplotype. Furthermore, the early divergence of the E the population south of the lake (location 4) has only group from the A, A¢, B, C and D groups is highly sup- shown haplotypes from the A group (Fig. 3 and ported by the 16S phylogeny (pp = 1) but is poorly sup- Tables S2, S3, Supporting Information). A similar pat- ported by the phylogeny based on CO1 (pp < 0.5). tern was observed within the D group where samples (except for location 9) were obtained at the lava fields from the Laki eruption, Skafta´reldar, in 1783. An inter- Population structure esting diversity was seen at locations 10 and 11 close to The genetic variation is geographically structured and Laki, which show the maximum genetic diversity of the each monophyletic group represents a geographically D group. Conversely, samples from springs in two lava

Table 4 Analyses of molecular variances of C. islandicus populations in Iceland based on COI and 16S mitochondrial genes. Groups present monophyletic groups AA¢, BC, D, E, F. ***P < 0.001

Pairwise differences Haplotype frequencies

Source of variation d.f. % of variation F-statistics % of variation F statistics

Among groups 5 84.9 0.849*** 14.6 0.146*** Among locations within groups 17 4.6 0.307*** 19.6 0.230*** Within location 74 10.5 65.8

2010 Blackwell Publishing Ltd SUB-GLACIAL SURVIVAL OF CRANGONYX ISLANDICUS 2523

A & A’ Fig. 3 Unrooted parsimonious network 8 1234 of COI gene haplotypes of C. islandicus populations in Iceland. Monophyletic A 7 groups are denoted with letters A, A¢, 3 5 6 B, C, D, E and F. The size of pies 2 F A’ reﬂects haplotype frequencies. Different 2 1 2 3 grey scales refer to the number of sam- 1 4 pling locations in Fig. 1, as indicated in boxes for each group separately (AA¢,B,

3 C, D, E, F). The number of mutations 1 F between haplotypes is indicated by dots 20 21 22 23 on each branch of the network. 9 6 5 4 B 10 567 B 4 2 E 1 2 1 5 4 1 3 2 E 16 3 17 18 3 D 19 6 9 10 11 4 12 C 13 D 14 1 8 15 C 5

fields running from Laki, a western one (locations 12, diversity at 16S could not be explained by any of the 13, 14) and an eastern one (location 15), are less diverse three variables. When analysing the geographic and (Fig. 3 and Tables S2, S3, Supporting Information). The geological associations, the F group was excluded due samples at these two lava fields harbour only types to its large differentiation from the other groups. The C found at Laki (locations 10 and 11) but do not share group was represented by only one individual and was any haplotypes. The altitudes of E-group locations are thus omitted. all above 250 m except location 16, which is down- The test statistics of selective neutrality, Tajima’s D stream and at an altitude of 10 m. That sample shows and Fu’s F, were all negative but not significant after the lowest diversity within the E group (Tables 1 and Bonferroni adjustment for multiple tests. Mismatch dis- 2). tributions for both 16S and COI did not show convinc- In addition to the clear phylogeographic structure, a ing evidence for expansion within each group (see strong correlation was found between geographic and Fig. S3, Supporting Information). Groups A, A¢ and D genetic distances analysed for COI and 16S, separately show clear bimodal distributions for the COI gene, indi- and combined. The correlation is significant both for cating stable populations as witnessed by larger p val- comparison among sampling locations (P £ 0.001) and ues (Table 2). B and D deviated significantly (P < 0.001) among groups (P < 0.05); in both cases a high propor- from the expansion model. Assuming the mean substi- tion of the genetic diversity was explained by geograph- tution rate of 2% per million years for the COI, expan- ical distance (R2 > 0.8 and see Fig. 4). Diversity at the sion of groups E and F started about 60 000 years ago. COI gene at each sampling location decreased with dis- These time estimates are, however, only suggestive, as tance from the plate boundaries (partial regression coef- the groups are small and formed from small, genetically ) ficient: b = )6.50 10 3, P = 0.04, Fig. 5a) and increases different samples. ) with altitude (b = 1.32 10 3, P = 0.01, Fig. 5b), while locations under 100 m display less than 20% of the Cryptic diversity maximum genetic diversity observed within the same group (R2 adjusted = 0.55). The association with the age Detection of cryptic species was based on the five major of the lava fields was not significant. The nucleotidic clades, with A and A¢ clustered together and B and C

2010 Blackwell Publishing Ltd 2524 E. KORNOBIS ET AL.

a b Genetic distances (subst/sites) Genetic distances (subst/sites) 0.000 0.005 0.010 0.015 0.020 0.006 0.008 0.010 0.012 0.014 0.016 0.018 0 50 100 150 200 250 50 100 150 200 Geographic distances between locations (km) Mean geographic distances between groups (km)

Fig. 4 Genetic difference versus geographic distance of C. islandicus populations in Iceland. The geographic distances are in kilometres. The genetic distances are expressed in number of substitutions per hundred sites, corrected with the TN93 substitution model, and were calculated from the two genes COI and 16S concatenated. The F group has been excluded from the analysis. (a) Compari- son between each samples. (b) Comparison between average geographic distances between groups and net average genetic distances between groups. The lines are drawn with the LOWESS regression method.

a b Scaled diversity Scaled diversity 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

20 40 60 80 100 0 100 200 300 400 Distance from "tectonic plate boundary" (km) Altitude

Fig. 5 The dependency of genetic diversity on geological distances of C. islandicus populations in Iceland. The genetic diversity has been scaled as a proportion of the maximum diversity observed within each monophyletic group. Genetic diversity is shown versus distances from the tectonic plate boundary (a) and versus altitude (b). The lines are drawn with the LOWESS regression method. together (Table 5). Divergence within these two groups separately. Correction of the distances (TN93) led to a is large (0.68, 0.95%) compared to the other groups and slight increase (by a factor of 1.02 to 1.07) in inter-clade decreases to 0.40–0.56% when the groups are analysed divergence and did not change the main result: only

2010 Blackwell Publishing Ltd SUB-GLACIAL SURVIVAL OF CRANGONYX ISLANDICUS 2525

Table 5 Genetic distances within and between the main F, occurred before the beginning of the last deglaciation monophyletic groups of C. islandicus, based on unique COI about 15 000 years ago (Geirsdo´ttir et al. 2007). The haplotypes. Intra-group divergence is shown along the diago- splitting event between the F group and all the other nal, pairwise divergence between groups is above the diagonal groups goes back to around 5 million years ago, even and net pairwise divergence between groups is below before the onset of the Ice Age around 2.5 million years Groups AA’ BC D E F ago (Geirsdo´ttir et al. 2007). Although these estimates may be biased due to the time dependency of the AA’ 0.95 1.58 2.43 2.74 6.55 molecular rates (see above), adjusted values would not BC 0.76 0.68 1.72 2.2 6.43 affect our conclusion, as the time estimates of all inter- D 1.77 1.19 0.39 2.18 6.89 nal nodes are 20 to 230 times larger than the time to the E 2.11 1.71 1.83 0.3 7.27 last glacial maximum. F 5.92 5.94 6.55 6.97 0.3 The correlation between genetic and geographic distances indicates that the populations diverged from each other from a common ancestor within Iceland, the F group passes the species-screening threshold rather than two or more diverged lineages colonized (5.30%, n = 5, SE = 0.13) with net pairwise distances Iceland. The large divergence between F and the other ranging from 6.31 to 7.47%. The maximum net diver- groups may have resulted from two separate coloniza- gence observed between the other groups is much tion events. It is, however, likely that the F group lower (2.19%). diverged from the others within Iceland. The poor dis- Patristic distances were on average 25% larger than persal capacities of groundwater amphipods (Kristjańs- the uncorrected distances and the two distances were son & Svavarsson 2007), the isolated position of Iceland highly correlated (Mantel test, R2 = 0.97,P<0.001). and the genetic proximity between those groups com- However none of the patristic distances passed the 16% pared to other Crangonyx species (Kornobis et al. subst ⁄ sites threshold defined by Lefe´bure et al. (2006b). unpublished) support this hypothesis. Therefore the The largest distance, between F and the others, is 12% average TMRCA of all haplotypes gives us an idea of subst. ⁄ site. The E group differs from A, A¢, B, C and D the minimum time since C. islandicus colonized Iceland. by 1.5% to 2.5% subst. ⁄ site. The evidence strongly suggests that C. islandicus has The genetic distances between and within the mono- been in Iceland for around 5 million years, well before phyletic groups are overlapping for the 16S gene. COI the onset of the Ice Age (Pleistocene), and has survived sequences are particularly informative for intra- and repeated glaciations on the island. Moreover, the AA¢, inter-species diversity characterization whereas 16S may BC, D, E and F present very distinct monophyletic be a more suitable marker for higher taxonomic ranks groups which have survived the last cold periods of the than for intra-specific comparison, as mentioned also by Ice Age in distinct refugia, as reflected by their time of Lefe´bure et al. (2006b). divergence. Such spatial structuring, with centres of diversity as putative refuge areas and possible recolon- ization routes within regions, has been suggested as Discussion further evidence for the occurrence of cryptic glacial refugia (see Provan & Bennett 2008 for a review). Ice Age survivors in Iceland Crangonyx islandicus survived glaciations and diverged Fissure areas as potential refugia into several monophyletic groups during the Pleistocene in Iceland. The extensive genetic diversity observed Ice-free zones at high latitudes, such as nunataks, have within C. islandicus differs from the low diversity been well documented and suggested as refugia, partic- observed in other high latitude species (Sadler 1999; ularly for plant species in Iceland (Rundgren & Ingo´lfs- Hewitt 2004), which are commonly characterized by son 1999). Small ice-free regions during the last shallow genealogies within regions. Such patterns have glaciation maximum in Iceland could have occurred in been explained by repeated population contractions the north, northwest and east of the island (Andrews towards southern refugia during the cold periods of the et al. 2000; Geirsdo´ttir et al. 2007). So far, no population Ice Age and expansion phases towards Arctic regions of C. islandicus nor any suitable habitats have been during warmer periods. The genetic structure found found in northwestern or eastern Iceland. The compact within C. islandicus populations and its endemism in bedrock shaped by the pressure of the glaciers is differ- Iceland can not be explained by a similar process. ent in these regions from porous lava fields, intruded Indeed, the haplotype divergence within each of the by fissures, in the volcanic zone. Furthermore, ice-free C. islandicus monophyletic groups, A¢, A, BC, D, E and refugia restricted to coastal mountains in the northern

2010 Blackwell Publishing Ltd 2526 E. KORNOBIS ET AL. regions can hardly account for the clear phylogeograph- One more argument that supports sub-glacial refugia ic pattern observed within C. islandicus. Except for those in fissure areas is that no groundwater amphipods have constricted ice-free areas, Iceland was completely cov- been found on the Snæfellnes peninsula. The Snaefells- ered by glaciers, hence C. islandicus had to survive in nes peninsula is a volcanic region outside the main vol- sub-glacial refugia during the multiple glaciation peri- canic zone and the tectonic plate boundary in Iceland, ods of the Quaternary. Unlike Lefe´bure et al. (2007), which was revitalized when magma intruded old, less- who argued that the observed genetic patterns in Ni- porous bedrock similar to that found in northwestern phargus rhenorhodanensis in the Alps could have resulted and eastern Iceland (see above). Although several from dispersal from ice-free regions close to or within springs are found in the recent (<10 000 years) lava the region, we can reject that possibility for C. islandi- fields in this region, which may be a suitable habitat for cus. the groundwater amphipods, the lack of fissures con- Fissure areas present in the rift zone of Iceland are nected to the plate boundary in this area (Sigmundsson commonly influenced by geothermal activities, which & Sæmundsson 2008) and the distance from putative could have maintained suitable habitats for C. islandicus source populations in fissures along the tectonic plate during the glaciations. These fissure areas are mainly boundary may explain why we did not find amphipods located around the tectonic plate boundaries. As subter- in this area, after an extensive sampling effort (Korn- ranean amphipods are bad dispersers, apparently only obis, unpublished). flowing along with a water current, we expected that the population with the highest genetic diversity is the Dispersal ⁄ vicariance one which is closest to the source population which survived glaciations. The hypothesis of fissure areas as The highly divergent monophyletic groups within C. potential refugia is supported by the reduced COI gene islandicus represent distinct geographic areas and show diversity with distance from the tectonic plate bound- a high correlation between genetic and geographic dis- aries and with lower altitude. As an example, sampling tances. This can be explained by a more widespread locations near two large fissure areas, Lake Thingvalla- distribution of C. islandicus ancestors throughout Ice- vatn (locations 2 and 3) and Lakagigar (locations 10 and land with extinctions of numerous transitional popula- 11), together with their respective monophyletic groups tions during the Quaternary glaciations. Repeated A, A¢ and D, show the highest COI genetic diversity. vicariance events during periods of glaciations, i.e. These three groups may have been at a stable popula- constriction of populations in distinct refugia due to tion size for a long period and have diverged from an glacier accretion, led to divergence of populations by ancestor within each group over a period of about 0.23– drift and mutations, currently observed among mono- 0.35 million years. Adjacent sampling locations display phyletic groups. This is especially evident when com- less diversity and are composed of a subset of the hapl- paring the F group in North-east Iceland from the otypes found in those interior locations, which are other groups, but is also relevant when the northern probably acting as source populations. The coexistence (E group) and southern Iceland (AA¢, B, D groups) are of the two very distinct clades A and A¢ at two loca- compared. tions in Lake Thingvallavatn may also reflect a second- All the groundwater springs sampled flow through ary admixture of recently separated groups, rather than lava fields younger than 10 000 years. Therefore, the a single large population. The two groups may have occurrence of amphipod populations in those newly diverged in distinct refugia as they have been separated formed habitats can only be explained by dispersal for at least 240 000 years and are found in separate geo- events less than 10 000 years ago. Iceland, located on graphic regions. the Mid-Atlantic Ridge, is spreading by 18.3 mm ⁄ year The positive association of genetic diversity with alti- in the direction N105E and the mantle plume under tude indicates that the colonization and migration Iceland contributes to frequent eruptions (Sigmundsson events from the refugia have been mainly passive and & Sæmundsson 2008), generating new potential habi- oriented by the subterranean water flow, from source tats. It appears that recent habitats were colonized populations at high altitudes to lower-altitude popula- from nearby source populations. The diversity tions. The lower genetic diversity observed at low alti- observed within regions reflects this process well, tudes can be explained by recent dispersal events and especially within AA¢ and D groups (see above). The colonization of previously immersed potential habitats. dependency of genetic diversity with distance from In Iceland around 13 000 years ago, the sea level was plate boundaries can be explained by bottlenecks fol- around 100 m above the current level, due to the pres- lowing colonization, from refugia in the fissure areas sure exerted by the extensive mass of ice on the island to the newly formed lava fields which extend outside (Norðdahl et al. 2008). the volcanic zone.

2010 Blackwell Publishing Ltd SUB-GLACIAL SURVIVAL OF CRANGONYX ISLANDICUS 2527

a surprise. The sub-glacial refugium hypothesis has Cryptic diversity previously been suggested for groundwater amphi- Groundwater domains, compared to other extreme pods (e.g. Koenemann & Holsinger 2001; Proudlove environments, exhibit important species richness (Dani- et al. 2003; Kristjańsson & Svavarsson 2007) but so far elopol et al. 2000). As morphological convergences or all the northern glacial refugia described involve ice- parallelism are frequent in such environments, molecu- free areas (see Provan & Bennett 2008). This hypothe- lar markers, less subject to convergent evolution, have sis is now well supported by the divergence time esti- played an important role and have revealed highly sub- mates provided in this study among the populations divided populations and numerous cryptic species (Ve- of C. islandicus. To our knowledge, this is the first rovnik et al. 2004; Lefe´bure et al. 2006a; Culver et al. proof that terrestrial individuals as complex as meta- 2009). zoans can survive repeated glacial periods under the According to the genealogical species concept (Baum ice sheet and our results enhance the interest of sam- & Shaw 1995), two reciprocally monophyletic groups pling actual sub-glacial water systems in Iceland and should be considered as distinct species (Hudson & in Antarctica. Coyne 2002). Nonetheless, this species recognition This type of refugium may be exclusively restricted method is not straightforward due to differences to areas with combined effects of a tectonic rift zone between gene trees and species trees (see Hudson & and geothermal activities as found in Iceland. During Coyne 2002; Kizirian & Donnelly 2004). The amount of the Holocene, after the retreat of the glaciers, new lava divergence observed between the two most divergent fields created suitable habitats which were colonized monophyletic groups within C. islandicus (F and the from nearby refugia. Moreover, the presence of species other) is 0.12 subst. ⁄ site, corresponding to around 5 mil- such as amphipods, which do not belong to the first lion years, nearly passing the species delimitation crite- trophic level, accounts for an ecosystem which has been rion based on COI variation that was developed by able to survive in these sub-glacial refugia. Lefe´bure et al. (2006b) for Crustaceans. This species threshold has been used recently to reveal potential Acknowledgements cryptic species, especially for groundwater Crustaceans (e.g. Page et al. 2008; Murphy et al. 2009; Zaksˇek et al. We are grateful to the University of Iceland and the Icelandic 2009). Research Council (Rannı´s) for financial support. We are also However, according to the Species-Screening Thresh- grateful to the editor and the anonymous reviewers who provided helpful suggestions on the draft mauscript. old (SST) for the COI gene previously applied to amphipod species (e.g. Witt et al. 2006; Bradford et al. 2010), the F group and other monophyletic groups References clustered together represent two provisional species. ´ ´ et al. Although such criteria can be useful for comparison, Andrews J, Hardardottir J, Helgadottir G (2000) The N and W Iceland Shelf: insights into Last Glacial Maximum ice they can not be expected to provide an unequivocal extent and deglaciation based on acoustic stratigraphy and standard or rule. It is evident that considerable diver- basal radiocarbon AMS dates. Quaternary Science Reviews, 19, gence has occurred among C. islandicus for long peri- 619–631. ods in allopatry which has resulted in evolutionary Audzijonyte A, Vaïno¨la¨ R (2006) Phylogeographic analyses of a independent lineages or even cryptic species. A further circumarctic coastal and a boreal lacustrine mysid inspection of the nuclear genome and morphological crustacean, and evidence of fast postglacial mtDNA rates. Molecular Ecology, 15, 3287–3301. differentiation will help to clarify the taxonomic status Baum D, Shaw K (1995) Genealogical perspectives on the of these groups. species problem. In: Experimental and Molecular Approaches to Plant Biosystematics (ed. Hoch PC, Stephenson AG), pp. 289– Conclusion 303, Missouri Botanical Garden, St. Louis. Bradford T, Adams M, Humphreys WF, Austin AD, Cooper Diverse assemblages of microorganisms have been SJB (2010) DNA barcoding of stygofauna uncovers cryptic found in perennial lakes, maintained by volcanic heat, amphipod diversity in a calcrete aquifer in Western Australia’s arid zone. Molecular Ecology Resources, 10, 41– under the Vatnajo¨kull ice cap in Iceland (Gaidos et al. 50. 2004, 2008). It has been proposed that Antarctic volca- Buckland PC, Perry DW, Gı´slason GM, Dugmore AJ (1986) The nic and geothermal activity could also provide such pre-Landna´m fauna of Iceland: a palaeontological suitable environments, even for multicellular organ- contribution. Boreas, 15, 173–184. isms (Vogel 2008). Nonetheless, the occurrence of Cabin RJ, Mitchell RJ (2000) To Bonferroni or not to groundwater species capable of surviving tens of thou- Bonferroni: when and how are the questions? ESA Bulletin, sands of years in terrestrial sub-glacial refugia is quite 81, 246–248.

2010 Blackwell Publishing Ltd 2528 E. KORNOBIS ET AL.

Clement M, Posada D, Crandall KA (2000) TCS: a computer Ho SYW, Larson G (2006) Molecular clocks: when times are program to estimate gene genealogies. Molecular Ecology, 9, a-changin’. Trends in Genetics, 22, 79–83. 1657–1659. Ho SYW, Shapiro B, Phillips MJ et al. (2007) Evidence for time Culver DC, Kane TC, Fong DW (1995) Adaptation and Natural dependency of molecular rate estimates. Systematic Biology, Selection in Caves: The Evolution of Gammarus minus. Harvard 56, 515–522. University Press, Cambridge. Holsinger JR (1980) Stygobromus canadensis, a new subterranean Culver DC, Pipan T, Schneider K (2009) Vicariance, dispersal amphipod crustacean (Crangonyctidae) from Canada, with and scale in the aquatic subterranean fauna of karst regions. remarks on Wisconsin refugia. Canadian Journal of Zoology, Freshwater Biology, 54, 918–929. 58, 90–97. Cunningham CW, Blackstone NW, Buss LW (1992) Evolution Hudson RR, Coyne JA (2002) Mathematical consequences of of king crabs from hermit crab ancestors. Nature, 355, 539– the genealogical species concept. Evolution, 56, 1557–1565. 542. Huelsenbeck JP, Bollback JP, Levine AM (2002) Inferring the Danielopol DL, Pospisil P, Rouch R (2000) Biodiversity in root of a phylogenetic tree. Systematic Biology, 51, 32–43. groundwater: a large-scale view. Trends in Ecology and Kizirian D, Donnelly MA (2004) The criterion of reciprocal Evolution, 15, 223–224. monophyly and classification of nested diversity at the species Darwin CR (1860) The Origin of Species by Means of Natural level. Molecular Phylogenetics and Evolution, 32, 1072–1076. Selection, or the Preservation of Favoured Races in the Struggle Knowlton N, Weigt LA (1998) New dates and new rates for for Life, revised edn. D. Appleton, New York. divergence across the Isthmus of Panama. Proceeding of the Dray S, Dufour AB, Chessel D (2007) The ade4 package-II: Royal Society London B, 265, 2257–2263. two-table and K-table methods. R News., 7, 47–52. Knowlton N, Weigt LA, Solorzano LA, Mills DK, Bermingham Drummond AJ, Rambaut A (2007) BEAST: Bayesian E (1993) Divergence in proteins, mitochondrial DNA, and evolutionary analysis by sampling trees. BMC Evolutionary reproductive compatibility across the Isthmus of Panama. Biology, 7, 214–222. Science, 260, 1629–1632. Ehlers J, Gibbard PL (2007) The extent and chronology of Koenemann S, Holsinger JR (2001) Systematics of the North Cenozoic global glaciation. Quaternary International, 164-165, American subterranean amphipod genus Bactrurus 6–20. (Crangonyctidae). Beaufortia, 51, 1–56. Excoffier L, Laval G, Schneider S (2005) Arlequin (version 3.0): Kristjańsson BK, Svavarsson J (2007) Subglacial refugia in an integrated software package for population genetics data Iceland enabled groundwater amphipods to survive analysis. Evolutionary Bioinformatics Online, 1, 47–50. glaciations. The American Naturalist, 170, 292–296. Finston TL, Johnson MS, Humphreys WF, Eberhard SM, Halse Lefe´bure T, Douady CJ, Gouy M, Trontelj P, Briolay J, Gibert J SA (2007) Cryptic speciation in two widespread subterranean (2006a) Phylogeography of a subterranean amphipod reveals amphipod genera reflects historical drainage patterns in an cryptic diversity and dynamic evolution in extreme ancient landscape. Molecular Ecology, 16, 355–365. environments. Molecular Ecology, 15, 1797–1806. Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R (1994) DNA Lefe´bure T, Douady CJ, Gouy M, Gibert J (2006b) Relationship primers for amplification of mitochondrial cytochrome c between morphological taxonomy and molecular divergence oxidase subunit I from diverse metazoan invertebrates. within Crustacea: Proposal of a molecular threshold to help Molecular Marine Biology and Biotechnology, 3, 294–299. species delimitation. Molecular Phylogenetics and Evolution, 40, Fu YX (1997) Statistical tests of neutrality of mutations against 435–447. population growth, hitchhiking and background selection. Lefe´bure T, Douady CJ, Malard F et al. (2007) Testing dispersal Genetics, 147, 915–925. and cryptic diversity in a widely distributed groundwater Gaidos E, Lanoil B, Thorsteinsson T et al. (2004) A viable amphipod (Niphargus rhenorhodanensis). Molecular microbial community in a subglacial volcanic crater lake, Phylogenetics and Evolution, 42, 676–686. Iceland. Astrobiology, 4, 327–345. Lopez P, Forterre P, Philippe H (1999) The root of the tree of Gaidos E, Marteinsson V, Thorsteinsson T et al. (2008) An life in the light of the covarion model. Journal of Molecular oligarchic microbial assemblage in the anoxic bottom waters Evolution, 49, 496–508. of a volcanic subglacial lake. The ISME Journal, 1–12. Lundin E, Dore´ AG (2002) Mid-Cenozoic post-breakup Geirsdo´ttir A´ , Miller GH, Andrews JT (2007) Glaciation, deformation in the ‘passive’ margins bordering the erosion, and landscape evolution of Iceland. Journal of Norwegian–Greenland Sea. Marine and Petroleum Geology, 19, Geodynamics, 43, 170–186. 79–93. Hall T (1999) BioEdit: a user-friendly biological sequence MacDonald KS, Yampolsky L, Duffy JE (2005) Molecular and alignment editor and analysis program for Windows morphological evolution of the amphipod radiation of Lake 95 ⁄ 98 ⁄ NT. Nucleic Acids Symposium Series, 41, 95–98. Baikal. Molecular Phylogenetics and Evolution, 35, 323–343. Hasegawa M, Kishino H, Yano T (1985) Dating of the human- Mantel N (1967) The detection of disease clustering and a ape splitting by a molecular clock of mitochondrial DNA. generalized regression approach. Cancer Research, 27, 209– Journal of Molecular Evolution, 22, 160–174. 220. Hebert PDN, Stoeckle MY, Zemlak TS et al. (2004) Millar CD, Dodd A, Anderson J et al. (2008) Mutation and Identification of birds through DNA barcodes. PloS Biology, evolutionary rates in Ade´lie penguins from the Antarctic. 10, 1657–1663. PloS Genetics, 4, 1–9. Hewitt GM (2004) Genetic consequences of climatic oscillations Murphy NP, Adams M, Austin AD (2009) Independent in the Quaternary. Philosophical Transactions of the Royal colonization and extensive cryptic speciation of freshwater Society of London B, 359, 183–195. amphipods in the isolated groundwater springs of

2010 Blackwell Publishing Ltd SUB-GLACIAL SURVIVAL OF CRANGONYX ISLANDICUS 2529

Australia’s Great Artesian Basin. Molecular Ecology, 18, 109– Tajima F (1989) Statistical method for testing the neutral 122. mutation hypothesis by DNA polymorphism. Genetics, 123, Nei M, Li W (1979) Mathematical model for studying genetic 585–595. variation in terms of restriction endonucleases. Proceedings of Tamura K, Nei M (1993) Estimation of the number of the National Academy of Sciences, USA, 76, 5269–5273. nucleotide substitutions in the control region of Norðdahl H, Ingo´lfsson O´ ,Pe´tursson HG, Hallsdo´ttir M (2008) mitochondrial DNA in humans and chimpanzees. Molecular Late Weichselian and Holocene environmental history of Biology and Evolution, 10, 512–526. Iceland. Jo¨kull, 58, 343–364. Tamura K, Dudley J, Nei M, Kumar S (2007) MEGA4: Nylander J (2004) MrAIC.pl: Program Distributed by the Author. molecular evolutionary genetics analysis (MEGA) software Evolutionary Biology Centre, Uppsala. version 4.0. Molecular Biology and Evolution, 24, 1596–1599. Page TJ, Humphreys WF, Hughes JM (2008) Shrimps Down Thornton IWB (1983) Vicariance and dispersal: confrontation or Under: Evolutionary Relationships of Subterranean compatibility? GeoJournal, 7, 557–564. Crustaceans from Western Australia (Decapoda: Atyidae: Torsvik TH, Mosar J, Eide EA (2001) Cretaceous–Tertiary Stygiocaris). PLoS ONE, 3, 1–12. geodynamics: a North Atlantic exercise. Geophysical Journal Palumbi S, Martin A, Romano S, McMillan WO (1991) The International, 146, 850–866. Simple Fool’s Guide to PCR, version 2.0. University of Hawaii, Vandel A. (1964) Biospe´leólogie : La biologie des animaux Honolulu. cavernicoles. Gauthier-Villars, Paris. Paradis E, Claude J, Strimmer K. (2004) APE: analyses of Verovnik R, Sket B, Trontelj P (2004) Phylogeography of phylogenetics and evolution in R language. Bioinformatics, 20, subterranean and surface populations of water lice Asellus 289–290. aquaticus (Crustacea: Isopoda). Molecular Ecology, 13, 1519– Penny D (2005) Relativity of molecular clocks. Nature, 436, 183– 1532. 184. Vogel S (2008) Cryosphere: fire and ice. Nature Geoscience, 1, Philippe H, Forterre P (1999) The rooting of the universal tree 91–92. of life is not reliable. Journal of Molecular Evolution, 49, 509– Witt JDS, Threloff DL, Hebert PDN (2006) DNA barcoding 523. reveals extraordinary cryptic diversity in an amphipod Porter ML (2007) Subterranean biogeography: what have we genus: implications for desert spring conservation. Molecular learned from molecular techniques? Journal of Cave and Karst Ecology, 15, 3073–3082. Studies, 69, 179–186. Xia X, Xie Z (2003) DAMBE: software package for data Proudlove GS, Wood PJ, Harding PT et al. (2003) A review of the analysis in molecular biology and evolution. Journal of status and distribution of the subterranean aquatic Crustacea Heredity, 92, 371–373. of Britain and Ireland. Cave and Karst Science, 30, 53–74. Xia X, Xie Z, Salemi M, Chen L, Wang Y (2003) An index of Provan J, Bennett KD (2008) Phylogeographic insights into substitution saturation and its application. Molecular cryptic glacial refugia. Trends in Ecology & Evolution, 23, 564– Phylogenetics and Evolution, 26, 1–7. 571. Yang Z (2007) PAML 4: phylogenetic analysis by maximum R core Development Team (2005) R: A Language and likelihood. Molecular Biology and Evolution, 24, 1586–1591. Environment for Statistical Computing. Foundation for Zaksˇek V, Sket B, Gottstein S, Franjevic´ D, Trontelj P (2009) The Statistical Computing, Vienna. limits of cryptic diversity in groundwater: phylogeography of Rambaut A, Drummond AJ (2003) Tracer version 1.2. University the cave shrimp Troglocaris anophthalmus (Crustacea: of Oxford, Oxford. Decapoda: Atyidae). Molecular Ecology, 18, 931–946. Rogers AR, Harpending H (1992) Population growth makes Zuckerkandl E, Pauling L (1965) Evolutionary divergence and waves in the distribution of pairwise genetic differences. convergence in proteins. In: Evolving Genes and Proteins (eds Molecular Biology and Evolution, 9, 552–569. Bryson V, Vogel HJ), pp. 97–166. Academic Press, New York. Rundgren M, Ingo´lfsson O (1999) Plant survival in Iceland during periods of glaciation? Journal of Biogeography, 26, 387– This study presents a part of Etienne Kornobis’ PhD project. 396. His project focuses on population structure and phylogenetics Sadler JP (1999) Biodiversity on oceanic islands: a of two endemic groundwater amphipod species in Iceland. palaeoecological assessment. Journal of Biogeography, 26, 75–87. Snaebjorn Pa´lsson studies phylogeography and population Schubart CD, Diesel R, Hedges SB (1998) Rapid evolution to ¨ genetics of various organisms. Bjarni K. Kristjańsson has terrestrial life in Jamaican crabs. Nature, 393, 363–365. focused his research on the study of distribution, ecology and Sigmundsson F, Sæmundsson K (2008) Iceland: a window on biodiversity of freshwater organisms including groundwater North-Atlantic divergent plate tectonics and geologic amphipods in Iceland. Jorundur Svavarsson studies mainly processes. Nordic Geoscience and 33rd IGC Episodes, 31, 92– ¨ morphology and distribution of marine invertebrates. 97. Sinton J, Gro¨nvold K, Sæmundsson K (2005) Postglacial eruptive history of the Western Volcanic Zone, Iceland. Geochemistry Geophysics Geosystems, 6, 1–34. Supporting Information Stillman JH, Reeb CA (2001) Molecular phylogeny of eastern Pacific porcelain crabs, genera Petrolisthes and Pachycheles, Additional supporting information may be found in the online based on the mtDNA 16S rDNA sequence: phylogeographic version of this article. and systematic implications. Molecular Phylogenetics and Fig. S1 Bayesian phylogenetic tree of C. islandicus popula- Evolution, 19, 236–245. tions, based on 128 sequence variation of the COI gene. Only

2010 Blackwell Publishing Ltd 2530 E. KORNOBIS ET AL. posterior probabilities superior to 0.5 are shown. The scale is years. ESS = Effective sampling size, i.e. the number of inde- given in number of substitutions per sites. pendent samples that the trace is equivalent to Fig. S2 Bayesian phylogenetic tree of C. islandicus populations Table S2 Haplotype frequencies at each sampling location of based on 103 sequence variation of the 16S gene. Only poster- C. islandicus populations in Iceland. Haplotype names corre- ior probabilities superior to 0.5 are shown. The scale is given spond to names in Fig. 2 and Table S3 in number of substitutions per sites. Table S3 Variable sites of the COI and 16S haplotypes of C. is- Fig. S3 Mismatch distribution plots. Groups are the monophy- landicus populations in Iceland. The ﬁrst two characters of the letic groups of C. islandicus populations in Iceland: A, A¢,C,D, haplotype names refer to COI haplotypes and correspond to E, F for both genes COI (a) and 16S (b). The A group is not the names in Fig. 2 and Table S2. The last number corresponds included for 16S since it lacked variation. to the 16S haplotype Table S1 Divergence time estimates for each monophyletic Please note: Wiley-Blackwell are not responsible for the con- group of C. islandicus populations in Iceland, calculated with tent or functionality of any supporting information supplied COI and 16S genes combined. Conﬁdence intervals (Hpd = by the authors. Any queries (other than missing material) 95% highest posterior density interval) are presented in million should be directed to the corresponding author for the article.

2010 Blackwell Publishing Ltd

Paper 2

Discordance in variation of the ITS region and the mitochondrial COI gene in the subterranean amphipod Crangonyx islandicus

J Mol Evol DOI 10.1007/s00239-011-9455-2

Discordance in Variation of the ITS Region and the Mitochondrial COI Gene in the Subterranean Amphipod Crangonyx islandicus

Etienne Kornobis • Snæbjo¨rn Pa´lsson

Received: 4 April 2011 / Accepted: 19 July 2011 Ó Springer Science+Business Media, LLC 2011

Abstract The amphipod Crangonyx islandicus is a Keywords Crustacea Ribosomal DNA Internal recently discovered species endemic to Iceland. Popula- transcribed spacer COI Duplication Population tions of C. islandicus are highly structured geographically structure Concerted evolution and genetically. The COI and 16S mitochondrial genes confine six monophyletic groups which have diverged for up to 5 million years within Iceland, and may present two cryptic species. To investigate the potential cryptic species Introduction status we analyse here the internal transcribed spacers (ITS1 and ITS2) and compare its variation with the patterns Crangonyx islandicus is an endemic amphipod species obtained with the mtDNA. The ITS regions present much dwelling in groundwater underneath lava fields in Iceland. less divergence among the geographic regions in compar- Analysis of variation in the mitochondrial COI and ison with the mtDNA, distances based on ITS1 are corre- 16S genes has revealed cryptic species diversity within lated with the COI distances as well as with geographic C. islandicus (Kornobis et al. 2010). The populations are distances, but most of the variation is observed within highly structured geographically and confine at least six individuals. The variation in the ITS region appears to have monophyletic groups which have diverged within Iceland been shaped both by homogenization effect of concerted for up to 5 million years. The geographical patterns of the evolution and divergent evolution. A duplication of 269 genetic variation indicate that C. islandicus survived base pairs is found in the ITS1 of all individuals from the repeated glaciation periods in Iceland in sub-glacial refugia southern populations, its divergence from its paralog at the tectonic plate boundary, probably facilitated by appears to predate the split of the different groups within geothermal activity. As convergent evolution in morpho- Iceland but some evidence point to rapid diversification logical traits is common among subterranean species, after the split. This duplication does not affect the sec- molecular markers have been widely used for detection of ondary structures found in the 30 and 50 ends of the cryptic diversity and species delimitation (e.g., Lefe´bure sequence, suggested to have a role in the excision of the et al. 2006; Fisˇer et al. 2008). An application of the species ITS1. Compensatory base changes within the ITS2 screening thresholds on C. islandicus, developed by Hebert sequences which have been suggested to be a species et al. (2004) and Witt et al. (2006), based on the COI indicator were not detected. variation gave evidence for two provisional species, one with a wide distribution in Iceland, and a second one found in northeastern Iceland (Kornobis et al. 2010). Electronic supplementary material The online version of this article (doi:10.1007/s00239-011-9455-2) contains supplementary The mtDNA COI region has been the main marker used material, which is available to authorized users. in the Barcoding Life initiative (see for review: Savolainen et al. 2005). The use of a sole mtDNA marker may lead to & E. Kornobis ( ) S. Pa´lsson an over- or under-estimation of species diversity (e.g., Roe Department of Life- and Environmental Sciences, University of Iceland, Askja, Sturlugata 7, 101 Reykjavik, Iceland and Sperling 2007), for example the COI gene displays e-mail: [email protected] high levels of variation even among conspecifics in

123 J Mol Evol copepod species (Goetze 2003). As single gene trees and Materials and Methods species trees have shown incongruencies (see Roe and Sperling 2007), additional DNA barcodes have been pro- Molecular Work posed such as the ITS2 to identify species and have proved to be highly effective (Yao et al. 2010). The internal The nuclear ITS region was amplified using primers ITS1 transcribed spacer regions (ITS1 and ITS2) of the ribo- F and 28S R (Table A1 in Supplementary Material) from somal RNA have been commonly used as suitable markers 18 individuals already extracted by Kornobis et al. (2010). for molecular taxonomy (Tang et al. 2003) including Sampling locations of the specimens are presented in crustacean species (Chu et al. 2001; Ota et al. 2010). Fig. 1. PCR amplifications were performed in 20 ll, con- Criteria based on the variation within the ITS2 region, such taining 0.15 mM dNTPs, 0.1% tween, 0.5 lg/ll of BSA, as the occurrence of compensatory base changes (CBC) are 19 Taq buffer, 0.35 lM primers, 0.5 unit of Taq highly informative for species identification for plants and (New England Biolabs), and 10–100 ng of template. fungi (Mu¨ller et al. 2007). Variation within the ITS1 has Amplifications were obtained with the PCR conditions: been suggested to be suitable for intraspecies analysis and 94°C 4 min, followed by 39 cycles of 94°C 30 s, anneal- has been used to assess divergence between populations ing temperature ranged from 50 to 60°C depending on the e.g., in the crustacean Penaeus japonicus (Chu et al. 2001). samples, for 45 s, 72°C 1.5 min, with a final elongation The ITS2 region has been shown to be less informative for step at 72°C for 10 min. PCR products were cut out of the intraspecies variation than ITS1 (Areekit et al. 2009), thus gel and purified using the Nucleospin Extract II Kit the variation in both regions might be particularly infor- (Macherey–Nagel). They were then ligated to TOPO mative for genetic analyses of C. islandicus populations. vector (TOPO TA Cloning Kit, Invitrogen) and cloned in TM Most rRNA genes in prokaryotes and eukaryotes are chemically competent cells DH5a -T1R. Plasmids puri- subject to concerted evolution (Dover et al. 1982; Ganley fied using the Nucleospin Plasmid Kit (Macherey–Nagel) and Kobayashi 2007, see for review Nei and Rooney 2005), were sequenced in both directions and with internal where the member genes are assumed to evolve as an unit primers (Table A1 in Supplementary Material), one to four in concert within a gene family. Homogenization of the clones per individual, using ABI BigDye Terminator v3.1 TM members of the gene family results from unequal cross- (Applied Biosystems) and ran on ABI PRISM 3100 overs and gene conversion (see Nei and Rooney 2005) and Genetic Analyser. Raw sequences were checked and edited affects the intraspecific variability (e.g., Bower et al. 2008), using BioEdit 7.0.9.0. (Hall 1999). Sequences have been opposite to the effect of divergent evolution, following a submitted to GenBank under accession number from death–birth process, where duplications or intraspecific JN258055 to JN258095. paralogs will differ more from each other than interspecific orthologs (Ota and Nei 1994; Ambrose and Crease 2010). The different mechanisms of evolution for the mitochondrial gene COI and the ITS regions are known to have led to different patterns in intraspecific variations for various F 22 species (e.g., Hansen et al. 2006; Navajas et al. 1998; E Carlini et al. 2009). A difference in variation of nuclear and 18 mitochondrial markers is expected due to the haploidy and 17 uniparental inheritance of the latter, resulting in four times 19 smaller effective population size than for a diploid nuclear A’ C 8 marker. In addition, lack of recombination within the 1 Latitude (°N) B mtDNA is expected to reduce the variation due to back- 5 D 2

11 ground selection (Charlesworth et al. 1993) and hitch- 4 9 13 A hiking (Maynard-Smith and Haig 1974). This difference of 12

mtDNA and nuclear markers may be further augmented 63.5 64.5 65.5 66.5 when considering a nuclear gene family, with multiple 24 22 20 18 16 14 gene copies (Mano and Innan 2008). Longitude (°W) The aims of this study are twofold: ﬁrst, to compare the genetic structures observed in an earlier study for the COI Fig. 1 Sampling locations of the groundwater amphipod Crangonyx and 16S mitochondrial genes among C. islandicus popu- islandicus in Iceland analysed in this study. Numbers and capital lations to the patterns of variation for the nuclear ITS1 and letters refer to sites and different monophyletic groups deﬁned by the COI and 16S variation in Kornobis et al. (2010). The volcanic active ITS2 genes, and second to evaluate the cryptic species zone is displayed in gray. Glaciers are in light gray. Longitudes and diversity within C. islandicus. latitudes are shown in degrees on the x- and y-axes, respectively

123 J Mol Evol

Alignment, Annotation, and Secondary Structure The correlation of the ITS1 and ITS2 pairwise genetic distances, and the genetic distances based on mtDNA COI Complete sequences were trimmed into separate datasets (Kornobis et al. 2010) were tested with a Mantel test, using 18S, ITS1, 5.8S and ITS2. 18S, ITS1, and 5.8S were the ade4 package (Dray and Dufour 2007) in R. In order to annotated by using annotations available for the amphipod avoid bias in the Mantel test due to different sampling species Diporeia hoyi, as well as by detecting highly con- effort between locations, we tested also the correlation on a served regions (18S and 5.8S) in an alignment encompass- subset of distances, including solely two clones (when ing other Crangonyctoidean species (in prep.). The ITS2 available) per location. region was defined using the MCMC analysis available at To characterize the structure among populations, a stan- the ITS2 Database III (Koetschan et al. 2010) choosing the dard AMOVA based on pairwise distances and haplotype eukaryote model, with maximum E values \0.1 (since the frequencies were computed using Arlequin 3.5 (Excoffier suggested value of E \ 0.001 did not lead to successful and Lischer 2010). The observed variation was partitioned, annotation for all the sequences), and a minimum size of among mtDNA monophyletic groups, as defined by Korn- 150 nt (as suggested in the ITS2 annotation software). obis et al. (2010), among individuals within monophyletic The ITS1 and ITS2 sequences were aligned using group and within individuals. Due to the limited number of RNAsalsa 0.8.1 (Stocsits et al. 2009) to produce simulta- samples obtained for ITS1 and ITS2, the only mtDNA neously the alignment and secondary structure for each monophyletic groups considered were A, A0, D, and E. sequence. The consensus structure constrain, necessary as Results were compared with a corresponding result from an input for a RNAsalsa run, was obtained with RNAalifold AMOVA based on COI pairwise distances computed in (Bernhart et al. 2008). After checking nakedly, no ambig- Arlequin. To avoid a bias due to larger samples of the COI uous parts in the alignments were found. The highly con- than for the ITS1 and ITS2, the AMOVA presents the served 18S, 5.8S, and 28S regions were aligned by eye in average of an analysis of ten random samples of the COI BioEdit (Hall 1999). The program 4SALE (Seibel et al. sequences where the number of sequences is equal to the 2008) was used to plot the consensus secondary structures, number of sequenced clones obtained for the ITS1. and to identify potential cryptic species by searching for CBC in the ITS2 region. Mountain plots (see Hofacker 2003) were drawn to compare the different structures. Results

Sequence Variation Genetic Diversity, Comparison with COI mtDNA Diversity and Population Structure In total, 41 complete sequences were obtained for the ITS1 region and 30 for the ITS2 and 5.8S. Forty partial Variability indices (i.e., segregating sites, phylogenetically sequences were obtained from the 30 end of the 18S gene. informative sites, haplotype, and nucleotide diversity) were For the mitochondrial genes, a total of 128 sequences of the summarized using DnaSP v5 (Librado and Rozas 2009), COI (658 bp) and 103 of the 16S (414 bp) were previously the APE (Paradis et al. 2004), and PEGAS (Paradis 2010) obtained by Kornobis et al. (2010). Diversity indices are packages implemented in R (R Core Development Team summarized in Table 1 for each DNA region. The ITS1 2005). The relationships between ITS1 haplotypes were region is highly variable in length, ranging from 363 to characterized by a reduced median network (Bandelt et al. 672 bp. The length variation is mainly due to a 269 bp 1995) computed in Network v4.6 (available at http://www. duplication (called ‘‘RII’’ in this article) solely present in fluxus-engineering.com/sharenet.htm) and redrawn in R. southern populations (Tables 1, 2; Fig. 2), and was iden- The optimal model of evolution for each dataset was tified in a dot matrix view using http://blast.ncbi.nlm.nih. chosen according to the maximum likelihood tree with gov. Interestingly, the long version of 672 bp of the ITS1 respect to substitutions, transition/transversion ratio, pro- contains an open reading frame (ORF) of 107 amino acids portion of invariant sites, and the gamma distribution located in the middle of the sequence (Fig. 2). Minor parameters selected with PhyML (Guindon and Gascuel length variation is also due to variation in number of TGA 2003) and implemented in the R-package APE. The Akaike tandem repeats surrounding RII and CT repeats (Table 2; Information Criterion (AIC) (Akaike 1974) was used to Fig. 2). The nucleotide diversity of ITS1 and ITS2 nuclear detect the model which best fitted the data. Two types of regions were much less variable than observed for the COI genetic distances were calculated, one based on the best and 16S mitochondrial genes but showed more variation fitted model of nucleotide mutations and a second one than the 18S and 5.8S regions (Table 1). The number of based on differences in number of tandem repeats and indel haplotypes for the ITS1 and ITS2 is however large com- events, each considered as a point mutation. pared to the sample size and the number of haplotypes

123 J Mol Evol

Table 1 Comparison between nuclear rDNA regions and mitochondrial COI and 16S genes 18S (41) ITS1 (41) 5.8S (30) ITS2 (30) COI (128) 16S (103) bp 144 363–672 309–311 613–618 658 414 S 444105389 24 IS 0 8 1 2 69 21 p 0.398% 0.746% 0.364% 0.646% 2.146% 1.033% k 5 26 4 24 35 17 H 18.80% 90.10% 14.30% 95.20% 93.30% 90.50% Numbers in parentheses present the total number of clones (nucDNA) or haplotypes (mtDNA) bp length of the fragment in base pairs, S number of segregating sites, IS number of phylogenetically informative sites, p nucleotide diversity, k number of haplotypes, H haplotype diversity observed for the COI and 16S sequences. The nucleotide and 93% for the ITS2, depending on how the variation is diversity within monophyletic groups, defined by COI and analyzed. The difference is larger when the variation 16S genealogy (Kornobis et al. 2010), is higher for the ITS1 among individuals within groups is also considered. This is and ITS2 than for the COI (Table 2). Nucleotide diversity clearly different from the results obtained from the mtDNA among clones within individuals ranges from 0 to 1.7%. COI sequences where only 12.55% of the variation is Based on the AIC, the best fit model was the HKY within sample locations. However, the geographic struc- (Hasegawa et al. 1985) for the ITS1 and the F84 (Felsen- ture, observed with mtDNA genes, holds to a certain stein and Churchill 1996) for the ITS2. The TN93 model degree when a sample size comparable to the one obtained (Tamura and Nei 1993), the most similar substitution for the ITS1 and ITS2 was used for the COI (Table 3). model to HKY and implemented in R, was used for pair- When the indel variation (RII and microsatellites) were wise distance calculations for the ITS1. considered as point mutations, 23% of the ITS1 variation is among groups defined by the mtDNA genealogy (Fig. 5). Secondary Structure of the ITS Regions This value is reduced to 8% when the indels were not considered. Including indels in the partition of the ITS2 Two main secondary structures are observed for the ITS1 variation did not result in larger differentiation among within C. islandicus (Fig. 3) due to the occurrence of the groups (Table 3). 269 bp duplication, otherwise the structures are similar The pairwise ITS1 genetic distances increase with (Fig. 4). The slight differences in length of the helices geographic distances (Mantel test, P \ 0.001, see Fig. 6a) observed in Fig. 4 can be either explained by single point but not for the ITS2 (P = 0.84). This correlation for the mutations or variation in the number of repeats in the ITS1 stands even when the Mantel test was conducted microsatellites TGA1 and TGA2. The occurrence of the solely on two clones per location (P \ 0.001). Although, insert in the southern population is not affecting the structure the ITS1 pairwise genetic distances are much smaller than of the common region RI (Figs. 3, 4). The ITS1 secondary the COI distances they were correlated (Mantel test, structure presents three conserved helices for all individuals P \ 0.05, see Fig. 6b), variation in ITS2 was not related studied (Fig. 3). Interestingly, two of these helices are to the COI genetic distances (P = 0.92). Considerable occurring at the 50 and 30 ends of the ITS1. One CBC was variation is observed in the genetic distances of the ITS1 observed in the ITS1 sequence comparing one clone from for a given COI genetic distance (Fig. 6b) which is partly one individual from northern Iceland to the other sequences. explained by the large variation witnessed within indi- The few variable sites observed in the ITS2 sequences viduals for the ITS1. COI nucleotide variation for the do not lead to noticeable changes in the secondary struc- complete dataset is more than three times higher than the tures predicted by RNAsalsa. The secondary structure of ITS2 nucleotide variation and the proportion of informa- the ITS2 is characterized by two long helices at the 50 and tive sites for the COI is also much higher (Table 1). This 30 end of the sequence with terminal ramifications. No variation in the COI led to the identification of clear CBC were observed among the ITS2 sequences. geographic groups, whereas in this study most of the clusters are inferred from the ITS1 variations and are Geographic Patterns much less geographically structured (Fig. 7). Although, most of the variation in ITS1 is found within individuals, Most of the variation of the ITS1 and ITS2 is found within the genetic variation among ITS1 sequences support individuals (Table 3), or 59 and 75% for the ITS1 and 82 the pattern of geographic structure observed with the

123 J Mol Evol

R ΙΙ R Ι 100 9 18S ITS1 5.8S ITS2 28S TGA1 TGA2 CT ORF 100 SD

9 Fig. 2 The internal transcribed region of the nuclear rRNA complex p in C. islandicus. Location of the ITS1 and ITS2 region is shown with respect to the rRNA genes 18S, 5.8S, and 28S. Regions RI and RII represent a duplicated region. RII is only present in southern populations. The ORF region corresponds to a 107-amino acid open reading frame. Microsatellites are displayed in light gray and are KS designated as TGA1, TGA2, and CT i N COI sample 100 9 standard deviation of the nucleotide diversity SD 100 SD 9 p KS nucleotide diversity, i N p c N ITS2 100 9

). TGA 1, TGA 2 and CT shows the numbers of tandem repeats of the corresponding microsatellites. The Fig. 3 The secondary structure of the ITS1 in C. islandicus. The -

100 SD consensus secondary structure obtained with RNAsalsa is displayed

9 for the northern (a) and southern (b) populations. The location of the number of segregating sites; p two microsatellites is indicated by TGA1 and TGA2. The structures S framed by broken lines (numbered 1–3) represent the only three ) or absent (

? helices which are conserved among all individuals studied. The asterisk and the arrow indicate, respectively, the 50 end of the ITS1 0 0 KS and its orientation (5 –3 ). The duplicated region (RII) in b is

i displayed in light gray N c

N mitochondrial genes (Figs. 5, 7). The E and F mtDNA 0 number of haplotypes; groups from the North are distinct from the A, A , B, and K C mtDNA groups from the western volcanic zone in . RII is coded as present (

2 central and southwestern Iceland by a single mutation (Figs. 1, 5, and 7). The occurrence of the 269-bp indel (RII in the ITS1) and the length variation of the tandem repeats TGA1 and TGA2 supports further the distinction between the northern and southern regions (Table 2;

number of individuals; Fig. 1). Except for one individual, the D group from the i N 4 0 7 3 1 3 8 1.29 0.85 3 1 2 2 0.22 0.35 3 2.26 1.44 0.16 0.30 4 0 7, 8 6 3 2 2 0.18 0.25 6 3 5 5 0.27 0.30 6 2.78 1.96 0.11 0.19 5, 6 9, 10 7 9 4 9 22 0.79 0.48 3 2 3 6 0.65 0.60 9 3.09 4.07 0.23 0.25 5 9 7 1 1 1 NA NA NA 1 1 1 NA NA NA 1 1 NA NA NA 5, 6 9, 10, 11, 12 7, 8 11 5 10 21 0.64 0.42 6 3 6 11 0.60 0.45 11 4.24 6.52 0.29 0.27 6 9, 10 7 1 1 1 NA NA NA 1 1 1 NA NA NA 1 1 NA NA NA 5, 6 9, 10, 11 6, 7 10 3 10 18 0.57 0.40eastern 10 3 volcanic 10 25 0.86 zone 0.49 in southern 10 5.24 6.69 Iceland 0.34 0.31 is also different from the central and southwestern populations (A, A0,B, - - ? ? ? ? R II TGA 1 TGA 2 CT ? and C) by a single mutation (Fig. 1, 7). Most of the segregating sites in the other regions, 18S, 5.8S, and Variation within mtDNA monophyletic groups for ITS1 and ITS2 compared to the COI ITS2, are singletons and their only effect on the network

number of clones; (Fig. 7) are to increase the length of the branches already 0 c F E D A A C B COI values present the averageN values calculated for 100 samples of the original dataset with number of individuals corresponding to the number of clones obtained for the ITS1 Indel regions correspond to regions in Fig. Table 2 mtDNA gr. Indel in ITS1 and ITS2 ITS1 drawn. Two informative sites were found in the ITS2

123 J Mol Evol

Fig. 4 Mountain plot representing the secondary structure of the ITS1 for all Crangonyx islandicus individuals. a Structures of the southern populations. b Structures of the northern populations. The position of the segregating sites is shown by ticks on the x axis in the lower graph. Triangles represent the positions of the phylogenetically informative sites. RI and RII (a) are delimiting the two duplicates. Microsatellites are highlighted by gray regions and annotated TGA1 and TGA2. The dotted line in b corresponds to the duplicated RII

region, one mutation which is in all clones of a single distances between stem and loop regions in RI did not individual and a second mutation shared by only two differ (Fig. 8). clones from the central and the southwestern regions. A single informative site found in the 5.8S region distinguishes the two different clones from northeastern Iceland Discussion from the other sequences. The ITS region in C. islandicus defines well-conserved Indel Region in the ITS1 secondary structures which are maintained despite some variation, mainly in length. Three helices are found, two of The regions RI and RII of the ITS1 were found in all 26 them occurring at the 50 and 30 ends of the ITS1 may have a clones from the southern populations, whereas RII was role in the maturation of the ribosomal RNAs. The sec- absent from all 9 clones from the northern populations ondary structure is known to have an important role in the (Table 2; see Figs. 2, 4). The variation observed in the excision of the ITS regions during the processing of the sequence of the 269-bp indel (RII) is clearly lower than the rRNA genes (van Nues et al. 1995;Coˆte´ and Peculis 2001). one observed within RI, but the variation between the two According to Gottschling and Plo¨tner (2004), a helix paralogs RI and RII within individuals is considerably structure near the 30 end of the ITS1 appears to be highly higher than within the two regions (Fig. 8). conserved among the eukaryotes secondary structures, Further analyses of the variation within stems and loops though the taxon sampling for these regions is still limited. and between the RI and RII regions shows distinct patterns. The duplication RII, occurring in the middle of the ITS1 The variation in number of segregating sites in RI is around sequence, is not affecting the secondary structure of the two times higher in the loops while similar numbers are regions near both 50 and 30 ends. No variation was observed observed within stem and loop in RII (Table 4, Fisher’s test in the secondary structure of the ITS2 within C. islandicus. P = 0.019). Comparison of variation between RI and RII, Variation along the ITS regions is determined, as other with respect to stem and loop, showed significantly dif- member of gene families by the relative rates of homoge- ferent values (Table 4, Fisher’s test P = 0.044). The nization due to gene conversion and unequal crossing over, highest ratio of variable sites was found when considering and divergence due to mutation, natural selection, and loops in RI and stems in RII. Similarly, genetic distances genetic drift (see Nei and Rooney 2005). The ITS1 is more within clones between RI and RII were higher considering variable than the ITS2 in this study as observed also in stem regions in RII than loops, whereas the average other studies at the intra-specific level (e.g., Areekit et al.

123 o Evol Mol J 123

Table 3 Analyses of molecular variances of C. islandicus populations in Iceland based on COI, ITS1, and ITS2 pairwise distances and relative frequencies COI ITS1 ITS2 Indel No indel Indel No indel df % Var % SD U Stat SD df % Var U Stat % Var U Stat df % Var U Stat % Var U Stat

Among groups 3 87.72 1.97 0.88*** 0.02 3 23.44 0.23*** 7.83 0.08* 3 9.29 0.093 16.37 0.16** Among locations within groups 9–12 -0.28 2.76 -0.04 0.22 Among individuals within groups 11 17.58 0.23*** 17.24 0.19* 7 -2.48 -0.027 1.6 0.02 Within locations 20–23 12.55 2.55 Within individuals 21 58.98 74.93 14 93.19 82.02 The groups present the monophyletic groups (A, A0, D, and E) defined by the mtDNA genealogy from Kornobis et al. (2010). The results for the COI are averages of ten AMOVAs realized with ten different samples of the sequences with sample sizes equal to the ITS1 and ITS2. ITS1 and ITS2 results are both presented with and without the indel events considered as point mutations df degree of freedom, % var percentage of variance for each hierarchical levels, % SD standard deviation of the percentage of variance, U Stat U statistics, SD standard deviation observed among the ten samples for COI gene *P \ 0.05, **P \ 0.01, ***P \ 0.001 i.6 Fig. for 16S and COI genes mt 2010 on based islandicus C. tree phylogenetic bayesian 5 Fig. ihteT9 usiuinmdl n eghdfeecsin differences length and omitted model, is corrected (duplication) sites, substitution RII per in TN93 Variation substitutions microsatellites. of the number on with based are distances O eei itne ( distances genetic COI lineages E ftemnpyei rusi iloso er.Pseirprobability the Posterior ancestor in years. common displayed of recent are millions most values in the groups to monophyletic time the the of represents node each at a niaig05ncetd usiuinprhnrdsites hundred per substitution nucleotide 0.5 indicating bar and .TergosR n I fteIS r on nmtDNA in found are ITS1 the of RII and RI regions The ). F T1gntcdsacsvru egahcdsacs( distances geographic versus distances genetic ITS1 the on plotted ITS1 the of (RII) indel large the of Existence fo otenIead oss nyrgo I h number The RI. region only possess Iceland) northern (from A – D ouain nIead\mdfidfo onbse al. et Kornobis from \(modified Iceland in populations fo otenIead,weesmDAlineages mtDNA whereas Iceland), southern (from b within ) box rnhsaedant cl,wt the with scale, to drawn are Branches . rnoy islandicus Crangonyx Genetic . a and ) J Mol Evol

2009). The variable sites are more commonly found in the loop structures, although to a lesser degree in ITS2, possibly indicating different selective pressures along the secondary structures. The ITS1 was found to be highly variable in length, with the occurrence of the two microsatellites at both ends of a large duplicated region. Microsatellites have been suggested to trigger crossing- over (Brandstro¨m et al. 2008) and could have played a role in the apparition of the duplicated region RII present in southern population of C. islandicus. Variation in length of the ITS1 has been commonly reported among species (e.g., Ko and Jung 2002; Gamerschlag et al. 2008; Chu et al. 2001) and within species (e.g., Bower et al. 2008; Ko and Jung 2002; van Herwerden et al. 1999). Such length variation has been observed in the amphipod species Gamm- arus minus (Carlini et al. 2009) and Gammarus tigrinus (Kelly et al. 2006) but to a lesser extent, their ITS1 length ranging, respectively, from 298 to 324 bp and from 550 to 563 bp. It is noticeable that the length of the ITS1 of the southern population of C. islandicus is among the largest Fig. 7 Unrooted reduced median network of the ITS1 haplotypes of C. islandicus populations. The size of the pies refers to observed found among crustacean species (cf. Chu et al. 2001). frequencies of haplotypes. Pie shadings refer to groups defined by Similarly, the ITS2 in C. islandicus is found to be among geographic areas and mtDNA genealogy from Kornobis et al. (2010) the largest one by comparison with sequences available in and presented in Figs. 1 and 5. The length of the branches are drawn the ITS2 database (Koetschan et al. 2010). in proportion to the number of mutations between haplotypes The population genetic structure based on the ITS regions is only partly comparable to the patterns revealed by the mtDNA variation of the barcoding marker COI (see Korn- obis et al. 2010). The variation within the ITS regions are only giving a weak support for the main geographic groups obtained from the mtDNA genealogy (Kornobis et al. 2010). The northern populations (mtDNA groups E and F) clustered together, whereas the mtDNA genealogy shows an early divergence of the northeastern group (F) from all the other groups, including E from northern Iceland. Nonetheless, the ITS1 variation is correlated with geographic distances among populations. The small divergence found among the populations of C. islandicus for the ITS1 and ITS2, reflects lower substitution rates than in the mtDNA COI and 16S genes in this species. ITS1 has already been shown to be less Genetic distances (subs/sites) informative than the COI for intra- and inter-specific comparisons for amphipods (Kelly et al. 2006; Carlini et al. 2009) and flatworms (Hansen et al. 2003; Hansen et al. 2006). 0.00 0.02 0.04 0.06 0.08 12345678 Whether this is due to the homogenization effect of con- Comparisons certed evolution, some other selective constrains, or solely due to increased effective population size of a nuclear marker Fig. 8 Pairwise genetic distances within C. islandicus based on ITS1. present in a gene family (Mano and Innan 2008) is not Boxplot 1 Distances between all individuals for region RI. Boxplot 2– known. The high variation in this study within individuals 8 present only information obtained from southern populations where RII is found. 2 Distances between individuals for RI. 3 Distances indicates that the homogenization is not strong, although it between individuals for RII. 4 Distances between RI and RII within may retard the substitutions of the segregating variants individuals. 5 and 6 Distances between RI and RII within individuals within populations. A larger effective population size may considering the secondary structure of RII, pairwise distances in stem result in increased variation within groups and consequently regions (5) and loop regions (6). 7 and 8 Distances between RI and RII within individuals considering the secondary structure of RI, to lower estimates of population subdivision such as the Fst pairwise distances in stem regions (7) and loop regions (8) and in longer time to fixation of segregating variants.

123 J Mol Evol

Table 4 Comparison of Comparison Regions Structure SSprop. number of segregating sites in stems and loops of ITS1 and Within regions R0 Stem 4 0.053 between regions RI and RII Loop 4 0.095 R I Stem 15 0.091 Loop 15 0.183 R II Stem 17 0.088 The number of ﬁxed mutations Loop 9 0.077 between RI and RII are in RI RII SSprop. parenthesis S number of segregating sites, Between regions Stem Stem 26 (2) 0.195 S prop proportion of segregating Loop Loop 12 (0) 0.240 sites out of the total number of Stem Loop 2 (0) 0.074 sites for each types of secondary structure Loop Stem 10 (2) 0.385

In contrast with the study based on the COI and 16S, structure supports reduced effect of homogenization. More differentiation among populations observed for the ITS frequent changes are observed in sites which were in loops regions does not point to any cryptic species. CBC sup- of RI but are present in stems of RII (see Fig. 8), such porting the occurrence of cryptic speciation (Mu¨ller et al. mutations may be favored as they contribute to the struc- 2007) were not observed in the ITS2 sequences. Two COI tural stability of RII. species delimitations criteria (Lefe´bure et al. 2006; Hebert Evolution of the ITS regions within C. islandicus et al. 2004; Witt et al. 2006) applied to the C. islandicus appears to be shaped both by concerted and divergent populations led to different conclusions about the new evolution. This makes any interpretations of the geographic species status of the F group (Kornobis et al. 2010). The patterns and species delimitation within this species more fact that no CBC were present among the ITS2 sequences difficult than with the variation obtained by the COI is though not sufficient to reject the hypothesis that mtDNA sequences. A larger genome wide study allowing C. islandicus is a species complex. In a study by Mu¨ller to separate the loci specific patterns from the rest of the et al. (2007), only 77% of the taxa which did not display genome is though needed to understand the evolution of the any CBC were actually belonging to the same species. ITS regions in C. islandicus. The high divergence between RI and RII in the ITS1, in comparison with the variation within both regions, may Acknowledgments We are grateful to the University of Iceland and indicate that the duplication event might have occurred the Icelandic Research Council (Rannis) for financial support. We want also to thank the editor and the anonymous reviewer who pro- before the split of the populations in northern and southern vided helpful suggestions on the draft manuscript. Iceland for 1–5 million years ago according to the time estimates based on COI and 16S variation (Kornobis et al. 2010). The duplicated region RII could have been lost in References both mtDNA lineages present in northern Iceland. An alternative and more parsimonious explanation is that the Akaike H (1974) A new look at the statistical model identification. duplication event RII occurred 1–1.3 millions years ago, IEEE Trans Autom Control AC-19:716–723 diversified from RI and spread throughout southern Ice- Ambrose C, Crease T (2010) Evolution of repeated sequences in the land. Accordingly, RII has been shaped by a divergent ribosomal DNA intergenic spacer of 32 arthropod species. J Mol evolution whereas RI has been conserved by purifying Evol 70:247–259 Areekit S, Singhaphan P, Khuchareontaworn S, Kanjanavas P, selection. Such a divergent evolution, either due to drift or Sriyaphai T, Pakpitchareon A, Khawsak P, Chansiri K (2009) selection, where the duplication may have acquired a new Intraspecies variation of Brugia spp. in cat reservoirs using function and interspecific orthologs are more similar to complete ITS sequences. Parasitol Res 104:1465–1469 one another than they are to intra-specific paralogs (Ota Arnheim N (1983) Concerted evolution of multigene families. In: Nei M, KoehnN RK (eds) Evolution of genes and proteins. Sinauer, and Nei 1994; Nei and Rooney 2005), has an opposite Sunderland, pp 38–61 effect than the homogenization caused by concerted evo- Bandelt HJ, Forster P, Sykes BC, Richards MB (1995) Mitochondrial lution (Arnheim 1983). There is some support for the portraits of human populations using median networks. Genetics hypothesis of divergent evolution of RII compared to RI. 141:743–753 Bernhart SH, Hofacker IL, Will S, Gruber AR, Stadler PF (2008) The difference observed between RII and the rest of the RNAalifold: improved consensus structure prediction for RNA ITS1 in the amount of variation relative to the secondary alignments. BMC Bioinform 9:474

123 J Mol Evol

Bower JE, Dowton M, Cooper RD, Beebe NW (2008) Intraspecific Hebert PDN, Stoeckle MY, Zemlak TS, Francis CM (2004) concerted evolution of the rDNA ITS1 in Anopheles farauti Identification of Birds through DNA Barcodes. PLoS Biol 2: sensu stricto (Diptera: Culicidae) reveals recent patterns of 1657–1663 population structure. J Mol Evol 67:397–411 Hofacker IL (2003) Vienna RNA secondary structure server. Nucleic Brandstro¨m M, Bagshaw AT, Gemmell NJ, Ellegren H (2008) The Acids Res 31:3429–3431 relationship between microsatellite polymorphism and recombina- Kelly DW, MacIsaac HJ, Heath DD (2006) Vicariance and dispersal tion hot spots in the human genome. Mol Biol Evol 25:2579–2587 effects on phylogeographic structure and speciation in a Carlini DB, Manning J, Sullivan PG, Fong DW (2009) Molecular widespread estuarine invertebrate. Evolution 60:257–267 genetic variation and population structure in morphologically Ko KS, Jung HS (2002) Three nonorthologous ITS1 types are present differentiated cave and surface populations of the freshwater in a polypore fungus Trichaptum abietinum. Mol Phylogenet amphipod Gammarus minus. Mol Ecol 18:1932–1945 Evol 23:112–122 Charlesworth B, Morgan MT, Charlesworth D (1993) The effect of Koetschan C, Fo¨rster F, Keller A, Schleicher T, Ruderisch B, deleterious mutations on neutral molecular variation. Genetics Schwarz R, Mu¨ller T, Wolf M, Schultz J (2010) The ITS2 134:1289–1303 Database III-sequences and structures for phylogeny. Nucleic Chu KH, Li CP, Ho HY (2001) The first internal transcribed spacer Acids Res 38:275–279 (ITS-1) of ribosomal DNA as a molecular marker for phylogenetic Kornobis E, Pa´lsson S, Kristjańsson BK, Svavarsson J (2010) and population analyses in crustacea. Mar Biotechnol 3:355–361 Molecular evidence of the survival of subterranean amphipods Coˆte´ C, Peculis B (2001) Role of the ITS2-proximal stem and (Arthropoda) during Ice Age underneath glaciers in Iceland. Mol evidence for indirect recognition of processing sites in pre-rRNA Ecol 19:2516–2530 processing in yeast. Nucleic Acids Res 29:2106–2116 Lefe´bure T, Douady CJ, Gouy M, Gibert J (2006) Relationship Dover G, Coen E, Strachan T (1982) The Dynamics of sequence between morphological taxonomy and molecular divergence turnover and concerted evolution of families of rDNA, histone within Crustacea: proposal of a molecular threshold to help genes and non-coding DNA of Drosophila. Heredity 49:141 species delimitation. Mol Phylogenet Evol 40:435–447 Dray S, Dufour AB (2007) The ade4 package: implementing the Librado P, Rozas J (2009) DnaSP v5: a software for comprehensive duality diagram for ecologists. J Stat Softw 22:1–20 analysis of DNA polymorphism data. Bioinformatics 25: Excoffier L, Lischer HEL (2010) Arlequin suite ver 3.5: a new series 1451–1452 of programs to perform population genetics analyses under Mano S, Innan H (2008) The evolutionary rate of duplicated genes Linux and Windows. Mol Ecol Resour 10:564–567 under concerted evolution. Genetics 180:493–505 Felsenstein J, Churchill GA (1996) A hidden Markov model approach Maynard-Smith J, Haig J (1974) The hitch-hiking effect of a to variation among sites in rate of evolution. Mol Biol Evol 13: favourable gene. Genet Res 23:23–35 93–104 Mu¨ller T, Philippi N, Dandekar T, Schultz J, Wolf M (2007) Fisˇer C, Sket B, Trontelj P (2008) A phylogenetic perspective on Distinguishing species. RNA 13:1469–1472 160 years of troubled taxonomy of Niphargus (Crustacea: Navajas M, Lagnel J, Gutierrez J, Boursot P (1998) Species-wide Amphipoda). Zool Scr 37:665–680 homogeneity of nuclear ribosomal ITS2 sequences in the spider Gamerschlag S, Mehlhorn H, Heukelbach J, Feldmeier H, D’Haese J mite Tetranychus urticae contrasts with extensive mitochondrial (2008) Repetitive sequences in the ITS1 region of the ribosomal COI polymorphism. Heredity 80:742–752 DNA of Tunga penetrans and other flea species (Insecta, Nei M, Rooney A (2005) Concerted and birth-and-death evolution of Siphonaptera). Parasitol Res 102:193–199 multigene families. Annu Rev Genet 39:121–152 Ganley ARD, Kobayashi T (2007) Highly efficient concerted Ota T, Nei M (1994) Divergent evolution and evolution by the birth- evolution in the ribosomal DNA repeats: total rDNA repeat and-death process in the immunoglobulin VH gene family. Mol variation revealed by whole-genome shotgun sequence data. Biol Evol 11:469–482 Genome Res 17:184–191 Ota Y, Tanaka K, Hirose M, Hirose E (2010) Description of a new Goetze E (2003) Cryptic speciation on the high seas; global species of Elaphognathia (Isopoda: Gnatheidae) from Japan phylogenetics of the copepod family Eucalanidae. Proc R Soc based on morphological and molecular traits. J Crustac Biol Lond B 270:2321–2331 30:710–720 Gottschling M, Plo¨tner J (2004) Secondary structure models of the Paradis E (2010) Pegas: an R package for population genetics with an nuclear internal transcribed spacer regions and 5.8S rRNA in integrated-modular approach. Bioinformatics 26:419–420 Calciodinelloideae (Peridiniaceae) and other dinoflagellates. Paradis E, Claude J, Strimmer K (2004) APE: analyses of phylog- Nucleic Acids Res 32:307–315 enetics and evolution in R language. Bioinformatics 20:289–290 Guindon S, Gascuel O (2003) A simple, fast, and accurate algorithm R Core Development Team (2005) R: a language and environment for to estimate large phylogenies by maximum likelihood. Syst Biol statistical computing. Foundation for Statistical Computing, Vienna 52:696–704 Roe A, Sperling F (2007) Population structure and species boundary Hall T (1999) BioEdit: a user-friendly biological sequence alignment delimitation of cryptic Dioryctria moths: an integrative editor and analysis program for Windows 95/98/NT. Nucleic approach. Mol Ecol 16:3617–3633 Acids Symp Ser 41:95–98 Savolainen V, Cowan RS, Vogler AP, Roderick GK, Lane R (2005) Hansen H, Bachmann L, Bakke TA (2003) Mitochondrial DNA Towards writing the encyclopedia of life: an introduction to variation of Gyrodactylus spp. (Monogenea, Gyrodactylidae) DNA barcoding. Phil Trans R Soc Lond B 1462:1805–1811 populations infecting Atlantic salmon, grayling and rainbow Seibel PN, Mu¨ller T, Dandekar T, Wolf M (2008) Synchronous visual trout in Norway and Sweden. Int J Parasitol 33:1471–1478 analysis and editing of RNA sequence and secondary structure Hansen H, Martinsen L, Bakke TA, Bachmann L (2006) The alignments using 4SALE. BMC Res Notes 1:91 incongruence of nuclear and mitochondrial DNA variation Stocsits RR, Letsch H, Hertel J, Misof B, Stadler PF (2009) Accurate supports conspecificity of the monogenean parasites Gyrodacty- and efficient reconstruction of deep phylogenies from structured lus salaris and G. thymalli. Parasitology 133:639–650 RNAs. Nucleic Acids Res 37:6184–6193 Hasegawa M, Kishino H, Yano T (1985) Dating of the human-ape Tamura K, Nei M (1993) Estimation of the number of nucleotide splitting by a molecular clock of mitochondrial DNA. J Mol Evol substitutions in the control region of mitochondrial DNA in 22:160–174 humans and chimpanzees. Mol Biol Evol 10:512–526

123 J Mol Evol

Tang B, Zhou K, Song D, Yang G, Dai A (2003) Molecular are critical for processing of internal transcribed spacer 2 from systematics of the Asian mitten crabs, genus Eriocheir (Crusta- Saccharomyces cerevisiae precursor ribosomal RNA. J Mol Biol cea: Brachyura). Mol Phylogenet Evol 29:309–316 250:24–36 van Herwerden L, Blair D, Agatsuma T (1999) Intra- and interindi- Witt JDS, Threloff DL, Hebert PDN (2006) DNA barcoding reveals vidual variation in ITS1 of paragonimus westermani (Trema- extraordinary cryptic diversity in an amphipod genus: implica- toda: digenea) and related species: implications for phylogenetic tions for desert spring conservation. Mol Ecol 15:3073–3082 studies. Mol Phylogenet Evol 12:67–73 Yao H, Song J, Liu C, Luo K, Han J, Li Y, Pang X, Xu H, Zhu Y, van Nues RW, Rientjes JMJ, Morre SA, Mollee E, Planta RJ, Venema Xiao P, Chen S (2010) Use of ITS2 region as the universal DNA J, Raue HA (1995) Evolutionarily conserved structural elements barcode for plants and animals. PLoS One 5:e13102

123

Paper 3

Molecular taxonomy and phylogenetic aﬃnities of two groundwater amphipods, Crangonyx islandicus and Crymostygius thingvallensis, endemic to Iceland

Molecular Phylogenetics and Evolution 58 (2011) 527–539

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier.com/locate/ympev

Molecular taxonomy and phylogenetic afnities of two groundwater amphipods, Crangonyx islandicus and Crymostygius thingvallensis, endemic to Iceland ⇑ Etienne Kornobis a, , Snbjrn Plsson a, Dmitry A. Sidorov b, John R. Holsinger c, Bjarni K. Kristjnsson d a Department of Biology, University of Iceland, Askja, Sturlugata 7, 101 Reykjavik, Iceland b Institute of Biology and Soil Science, Far Eastern Branch of the Russian Academy of Sciences, Vladivostok 690022, Russia c Department of Biological Sciences, Old Dominion University, Norfolk, VA, USA d Department of Aquaculture and Fish Biology, Hlar University College, Heyri 1, 550 Saurkrkur, Iceland article info abstract

Article history: The amphipod superfamily Crangonyctoidea is distributed exclusively in freshwater habitats worldwide Received 20 October 2010 and is characteristic of subterranean habitats. Two members of the family, Crangonyx islandicus and Revised 17 December 2010 Crymostygius thingvallensis, are endemic to Iceland and were recently discovered in groundwater under- Accepted 20 December 2010 neath lava elds. Crangonyx islandicus belongs to a well-known genus with representatives both in North Available online 30 December 2010 America and in Eurasia. Crymostygius thingvallensis denes a new family, Crymostygidae. Considering the incongruences observed recently between molecular and morphological taxonomy within subterranean Keywords: species, we aim to assess the taxonomical status of the two species using molecular data. Additionally, rDNA the study contributes to the phylogenetic relationships among several crangonyctoidean species and spe- Crustacea Amphipoda cically among species from four genera of the family Crangonyctidae. Given the available data we con- Crangonyctoidea sider how the two Icelandic species could have colonized Iceland, by comparing geographical origin of the Molecular phylogeny species with the phylogeny. Alignment methods Regions of two nuclear (18S and 28S rRNA) and two mitochondrial genes (16S rRNA and COI) for 20 different species of three families of the Crangonyctoidea were sequenced. Four different methods were used to align the RNA gene sequences and phylogenetic trees were constructed using bayesian and maximum likelihood analysis. The Crangonyctidae monophyly is supported. Crangonyx islandicus appeared more closely related to species from the Nearctic region. Crymostygius thingvallensis is clearly divergent from the other species of Crangonyctoidea. Crangonyx and Synurella genera are clearly polyphyletic and showed a geographical association, being split into a Nearctic and a Palearctic group. This research conrms that the studied species of Crangonyctidae share a common ancestor, which was probably widespread in the Northern hemisphere well before the break up of Laurasia. The Icelandic species are of particular interest since Iceland emerged after the separation of Eurasia and North America, is geographically isolated and has repeatedly been covered by glaciers during the Ice Age. The close relation between Crangonyx islandicus and North American species supports the hypothesis of the Trans-Atlantic land bridge between Greenland and Iceland which might have persisted until 6 million years ago. The status of the family Crymostygidae is supported, whereas Crangonyx islandicus might represent a new genus. As commonly observed in subterranean animals, molecular and morphological taxonomy led to different conclusions, probably due to convergent evolution of morphological traits. Our molecular analysis suggests that the family Crangonyctidae needs taxonomic revisions. Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction and meristic criteria. The two species are Crangonyx islandicus,a new species within a known genus in the family Crangonyctidae Two endemic species of the exclusively freshwater superfamily (Svavarsson and Kristjnsson, 2006), and Crymostygius thingvallen- Crangonyctoidea, characteristic of subterranean habitats, were re- sis, a new monotypic family (Crymostygidae) (Kristjnsson and cently discovered in Iceland and described using morphological Svavarsson, 2004). According to morphological taxonomy, Crang- onyx islandicus belongs to the genus Crangonyx, distributed in ⇑ North America and Eurasia. The discovery of these two new species Corresponding author. Fax: +354 525 4632. in Iceland, which was repeatedly covered by glaciers during the E-mail addresses: [email protected] (E. Kornobis), [email protected] (S. Plsson), sidorov @biosoil.ru (D.A. Sidorov), [email protected] (J.R. Holsinger), [email protected] cold periods of the Ice Age (Pleistocene, from 2.59 Myr to (B.K. Kristjnsson). 12,000 years ago), has raised questions about when and how they

1055-7903/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2010.12.010 528 E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539 colonized Iceland. A recent study has found strong evidence for methods can have profound effects on phylogenies and are of spe- survival of Crangonyx islandicus in sub-glacial refugia in Iceland. cial concern for the ribosomal genes where insertions and/or dele- Using mtDNA divergence among monophyletic and geographically tions are commonly observed. Numerous sequence alignment isolated populations, Kornobis et al. (2010) showed that the ob- methods have been developed recently (Edgar and Batzoglou, served divergence occurred within Iceland, during the Ice Age 2006; Novk et al., 2008; Bradley et al., 2009). We apply different and even before its onset. alignment techniques and weighting schemes to evaluate the im- The existence of Iceland has been traced to a geological hotspot pact of the alignment and variation along the sequence. which migrated southeastward about 40 Mya from the East Green- land coast to its current location at the boundary of the Atlantic 2. Material and methods ridge and the Greenland–Scotland Transverse Ridge (Lawver and Mller, 1994; Lundin and Dor, 2002). Geological and biological 2.1. Samples evidence support that the Greenland–Scotland Ridge, submerged at great depths at present, was above sea level from early Cenozoic Twelve amphipod species were collected from 15 locations in to late Miocene, rst as a continuous land bridge and later as a North America, Europe and Asia (Table 1) and were preserved in chain of islands (McKenna, 1983; Eldholm et al., 1994; Grmsson 96% ethanol. For the construction of phylogenetic trees, sequences et al., 2007; Poore, 2008; Denk et al., 2010). Fossil records show of 18S RNA (14 species) and 28S RNA (7 sp.) genes together with that plants migrated along this land bridge between Scotland and the mtDNA 16S (4 sp.) and COI (5 sp.) genes (Table A1) from 17 proto-Iceland until 24 Mya and until 15 Mya or even 6 Mya be- species in total belonging to the suborder Gammaridea were ob- tween Greenland and proto-Iceland (Grmsson et al., 2007; Denk tained from Genbank (http://www.ncbi.nlm.nih.gov/genbank/), et al., 2010). Because Crangonyx islandicus and Crymostygius thingv- 13 of them belonging to the superfamily Crangonyctoidea. The allensis belong to a superfamily of exclusively freshwater amphi- total sample resulted in 20 different species of the crangonyctoids pods, it is possible that their ancestors colonized Iceland through in ve genera. The number of species sequenced for each gene this land connection rather than by marine ancestors. varied (Table A1). To evaluate the divergence among species and The taxonomy of subterranean organisms based on morpholog- genera of the Crangonyctidae, comparisons were made with corre- ical information faces several problems. The morphological traits sponding divergence between 18S sequences from the amphipod are generally characterized both by progressive (e.g. elongation superfamilies Gammaroidea (33 species), Lysianassoidea (3), of the trunk and/or sensory appendages) and regressive evolution Talitroidea (10) and Eusiroidea (7), and 28S sequences from (loss of eyes and pigmentation) (Porter, 2007; Vinl et al., Gammaroidea (54), Lysianassoidea (18) and Talitroidea (4), 2008). These troglomorphic traits are apparently shaped by similar obtained from Genbank. selective pressures (absence of light, low nutrients, low oxygen) encountered in subterranean systems worldwide, which have led to convergent evolution of morphological traits among different 2.2. Molecular work species. The abundance of such traits may hinder the identication of morphologically informative characters for phylogenetic recon- DNA was extracted using 6% Chelex 100 (Biorad) from whole struction (Englisch et al., 2003; Wiens et al., 2003). Taxonomic specimens or only a pereopod from bigger specimens. Two regions revision using molecular markers is thus especially important for of the nuclear genome, the 18S RNA and 28S RNA genes, and two of subterranean species (Fišer et al., 2008). Molecular taxonomy of the mitochondrial genome, COI and 16S RNA genes, were amplied groundwater amphipods has been widely used in the past decade using primers summarized in Table 2. PCR amplications were per- (Englisch et al., 2003; Fišer et al., 2008) and has indicated that formed in 10 ll, containing 0.15 mM dNTPs, 0.1% Tween, 0.5 lg/ll morphological markers are to a great extent homoplasic. In addi- of BSA, 1X Taq buffer, 0.35 lM primers, 0.5 unit of Taq (New Eng- tion, molecular analysis has recently revealed cryptic diversity land Biolabs) and 10–100 ng of template. Amplications were ob- ° (Lefbure et al., 2006a; Kornobis et al., 2010) and taxonomic incon- tained with the PCR conditions: 94 C 4 min, followed by 39 ° sistencies among various groundwater amphipod species (Lefbure cycles of 94 C 30 s, annealing temperature ranging from 45 to ° ° et al., 2006a). 55 C (depending on species and genes) for 45 s, 72 C for 1.5 min ° The aim of this study is to evaluate the taxonomic status of the (1 min for both mt genes), with a nal elongation step at 72 C two endemic species Crymostygius thingvallensis and Crangonyx for 6 min. The mtDNA fragments were puried and sequenced di- islandicus and to contribute to the phylogenic classication of the rectly as in Kornobis et al. (2010). The 18S and 28S PCR products crangonyctoidean species, with a special focus on the family Cran- were cut out of the gel and puried using the Nucleospin Extract gonyctidae. In addition we consider the phylogenetic relationships II Kit (Macherey–Nagel), ligated to TOPO vector (TOPO TA Cloning with respect to the species distribution (Palearctic or Nearctic) to Kit, Invitrogen) and cloned in chemically competent cells DH5a™- R infer putative colonization routes, focusing on the colonization of T1 . Plasmids puried using the Nucleospin Plasmid Kit (Mache- Iceland. The study is based on DNA sequence variation among rey–Nagel) and Exosap puried PCR products were sequenced in three families (Crymostygidae, Crangonyctidae, Pseud- both directions and with internal primers (Table 2) using ABI Big- ocrangonyctidae) of the superfamily Crangonyctoidea and among Dye Terminator v3.1 (Applied Biosystems). Three to ve clones per ve genera (Crangonyx, Amurocrangonyx, Bactrurus, Synurella and individual of the nuclear genes were sequenced and run on an ABI Stygobromus) in the family Crangonyctidae for two nuclear genes PRISM™ 3100 Genetic Analyzer. Raw sequences were checked and (18S and 28S ribosomal RNA genes), and two mitochondrial genes edited using BioEdit 7.0.9.0. (Hall, 1999). Sequences have been sub- (Cytochrome oxydase subunit I COI and 16S ribosomal RNA). mitted to GenBank under accession numbers from HQ286000 to The subunits, 18S and 28S, of ribosomal genes have been exten- HQ286020 and from HQ286022 to HQ286037. sively used as markers for phylogenies at the family level, e.g. within amphipods (Englisch et al., 2003; MacDonald et al., 2005; 2.3. Alignment Lefbure et al., 2006a) and even for deeper phylogenies within crustaceans (Mallatt and Giribet, 2004; Page et al., 2008). Although Four different methods were used to align the RNA sequences: the use of RNA genes for phylogenetic purposes has been success- (1) ClustalW2 with default parameters (Larkin et al., 2007). Clu- ful, some methodological problems exist (Edgar and Batzoglou, stalW has been extensively used, but newly developed algorithms 2006). Like the tree building methods (Morrison, 2008), alignment are considered to be more accurate (Morrison, 2009). We chose E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539 529

Table 1 Species included in the phylogenetic analyses with geographic origin and sampling date. References are given for specimens obtained from genbank. Species from the Gammaridae and Megaluropidae families were used as outgroups. NA stands for not available.

Family Species Locality Latitude Longitude Reference or year of sampling Crangonyctidae Amurocrangonyx arsenjevi Khabarovsk, Russia 47.902 135.340 This study (2005) Crangonyctidae Bactrurus brachycaudus St. Louis Co., Missouri, USA NA NA Englisch and Koenemann (2001) Crangonyctidae Bactrurus mucronatus Saline Co., Illinois, USA 37.680 88.420 Englisch and Koenemann (2001) Crangonyctidae Bactruruspseudo mucronatus Oregon Co., Missouri, USA 36.815 91.181 Englisch and Koenemann (2001) Crangonyctidae Crangonyx chlebnikovi Perm, Russia 57.446 57.017 This study (2005) Crangonyctidae Crangonyx chlebnikovi Perm, Russia 57.061 57.528 This study (2003) Crangonyctidae Crangonyx oridanus Gainsville, Florida, USA 81.657 30.275 Slothouber Galbreath et al. (2009) Crangonyctidae Crangonyx forbesi St. Louis Co., Missouri, USA 38.616 90.701 Englisch AND Koenemann (2001) Crangonyctidae Crangonyx islandicus Thingvallavatn, Iceland 64.241 21.053 This study (2007) Crangonyctidae Crangonyx islandicus Svartarvatn, Iceland 65.469 17.233 This study (2007) Crangonyctidae Crangonyx islandicus Klapparos Kopasker, Iceland 66.361 16.400 This study (2008) Crangonyctidae Crangonyx pseudogracilis Lake Charles, Louisiana, USA 30.262 93.221 Slothouber Galbreath et al. (2009) Crangonyctidae Crangonyx serratus Virginia, USA NA NA MacDonald et al. (2005) Crangonyctidae Crangonyx sp. Barnishee Slough, Tenessee, USA 35.579 89.963 Slothouber Galbreath et al. (2009) Crangonyctidae Crangonyx subterraneus Biberach, Schwarzwald, Germany 48.326 8.126 Fišer et al. (2008) Crangonyctidae Stygobromus gracilipes (H3970) West Virginia, USA 39.302 77.850 This study (2002) Crangonyctidae Stygobromus gracilipes (H4070) Virginia, USA 39.014 78.276 This study (2000) Crangonyctidae Stygobromus mackini USA NA NA Englisch and Koenemann (2001) Crangonyctidae Stygobromus stegerorum Virginia, USA 38.301 79.255 This study (2000) Crangonyctidae Synurella ambulans Ljubljana, Slovenia 46.050 14.500 This study (2010) Crangonyctidae Synurella dentata Preble Co., Ohio, USA 39.772 84.722 Englisch et al. (2003) Crangonyctidae Synurella sp. Lake Charles, Louisiana, USA 30.262 93.221 Slothouber Galbreath et al. (2010) Crymostygiidae Crymostygius thingvallensis Thingvallavatn, Iceland 64.241 21.053 This study (2007) Gammaridae Gammarus abstrusus Lushan, Sichuan, China 30.280 102.970 Hou et al. (2007) Megaluropidae Megaluropus longimerus Curacao, Karibische See NA NA Englisch et al. (2003) Niphargidae Niphargus fontanus Grundwasser am Ruhrufer, Germany NA NA Englisch and Koenemann (2001) Niphargidae Niphargus kochianus Grundwasser am Ruhrufer, Germany NA NA Englisch et al. (2003) Pseudocrangonyctidae Procrangonyx primoryensis Primory, Russia 47.185 138.743 This study (2003) Pseudocrangonyctidae Procrangonyx primoryensis Primory, Russia 47.256 138.800 This study (2002) Pseudocrangonyctidae Pseudocrangonyx korkishkoorum Primory, Russia 43.100 131.547 This study (2006)

Table 2 Oligonucleotides used for PCR and sequencing of the four different genes.

Genes Primer name Primer sequence (50–30) References 18S 18SF CCTAYCTGGTTGATCCTGCCAGT Englisch et al. (2003) 18S 700R CGCGGCTGCTGGCACCAGAC Englisch et al. (2003) 18S 1500R CATCTAGGGCATCACAGACC Englisch et al. (2003) 18S R TAATGATCCTTCCGCAGGTT Englisch et al. (2003) 18S F700+ AATTCCAGCTTCAGCAGCAT This study 28S 28SF TTAGTAGGGGCGACCGAACAGGGAT Hou et al. (2007) 28S700F AAGACGCGATAACCAGCCCACCA Hou et al. (2007) 28S1000R GACCGATGGGCTTGGACTTTACACC Hou et al. (2007) 28SR GTCTTTCGCCCCTATGCCCAACTG Hou et al. (2007) 28Sa TTGGCGACCCGCAATTTAAGCAT Cristescu and Hebert (2005) 28Sb CCTGAGGGAAACTTCGGAGGGAAC Cristescu and Hebert (2005) COI LCO1490 GGTCAACAAATCATAAAGATATTGG Folmer et al. (1994) HCO2198 TAAACTTCAGGGTGACCAAAAAATCA Folmer et al. (1994) 16S 16Stf GGTAWHYTRACYGTGCTAAG Macdonald et al. (2005) 16Sbr CCGGTTTGAACTCAGATCATGT Palumbi et al. (1991)

this method for comparison with other alignment techniques. (2) The mean sequence identity was estimated for each alignment MAFFT 6 (Katoh et al., 2009), with Q-INS-i strategy in order to take method and for each RNA gene, using the APE package (Paradis into account the secondary structure of the RNA. This method was et al., 2004)inR(R Development Core Team, 2010). According to chosen for the recent implementation of secondary structure Hall (2008) and following the results of Kumar and Filipski search (Katoh and Toh, 2008) and its good ratio accuracy/computa- (2007), an alignment of non-coding DNA that presents a mean se- tion time (Edgar and Batzoglou, 2006). (3) FSA 1.15.2, a statistical quence identity of 66% ensures about 50% of alignment accuracy. alignment software with a limited runtime and which avoids ef- Variable alignment accuracies higher than 50% are considered to ciently false/positive alignments (Bradley et al., 2009), was used have little effect on the phylogenetic reconstruction, both for with a gap factor of 1 and the Tamura Nei model. (4) RNAsalsa Bayesian and Maximum likelihood methods (Ogden and Rosen- 0.8.1 (Stocsits et al., 2009), a recent software which uses secondary berg, 2006). structure information for adjusting and rening the sequence The optimal model of evolution for each dataset was chosen alignment, was used with default parameters and the consensus according to the maximum likelihood tree with respect to substitu- secondary structure of preliminary alignments of the RNA se- tions, transition/transversion ratio, proportion of invariant sites quences as a structural constraint. The COI sequences were aligned and the gamma distribution parameters selected with PhyML by eye using BioEdit. (Guindon and Gascuel, 2003) and implemented in the R-package 530 E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539

APE. The Akaike Information Criterion (AIC) (Akaike, 1974) was gene, we reconstructed trees following the same procedure as de- used to detect the model which best tted the data, as recom- scribed above, separately with two combined datasets: one with mended by Posada and Buckley (2004). the nuclear genes (18S and 28S), and another with the mitochon- Nuclear datasets were partitioned in stem and loop regions drial genes (COI and 16S). according to the consensus structure obtained in RNAalifold The results from the different alignment methods were com- (Bernhart et al., 2008) for each alignment method. Variation, pared by considering the variation in number of trees obtained including the number of phylogenetically informative sites and with MrBayes (Ronquist and Huelsenbeck, 2003) and their likeli- gaps at different regions, outside hairpins, in stems and at internal hoods using Tracer (Rambaut and Drummond, 2003), then summa- and terminal loops, was scored for the four different methods. rized with the mean values and high posterior density interval Mountain plots were used to compare the secondary structures (HPD). As the overall likelihood value depends on the number of obtained by RNAalifold. bases compared, the best alignment method was selected after adjusting the log likelihoods of the trees using ANCOVA, with the 2.4. Phylogenetic analysis number of bases in the alignment as a covariate.

Phylogenetic trees were constructed separately for each gene 2.5. Variation within Crangonyctoidea compared with other amphipod fragment as the available number of taxa for the different genes superfamilies varied, and also based on combined datasets. The analysis based on different genes may provide a stronger support for the observed Pairwise evolutionary distances between 18S and 28S se- clusters of species, or indicate some gene-specic evolutionary quences, aligned with ClustalW2, were calculated using the APE divergence. package with the evolutionary model selected from the PhyML Search for optimal trees was carried out with MrBayes v3.1.2 analysis. The distances were grouped at different taxonomic levels (Ronquist and Huelsenbeck, 2003) and PhyML (Guindon and (within and between genera, within and between families) in order Gascuel, 2003) using the pre-specied evolutionary model (with to assess both the taxonomic status of the Icelandic species and to the lowest AIC value). MrBayes is particularly suited for our analy- evaluate the classication within the family Crangonyctidae. For sis since it allows partitioning of the data according to secondary comparing divergence within this family with divergence within structure and implementation of indels as morphological charac- other amphipod families, we computed distances at the same tax- ters. In MrBayes, the parameters of the selected model were opti- onomic levels for four other amphipod superfamilies available for mized during searches as recommended by Ronquist and the 18S in Genbank (Gammaroidea, Lysianassoidea, Talitroidea, Huelsenbeck (2003), running two independent MCMC with one Eusiroidea) and three superfamilies for the 28S (Gammaroidea, 6 cold and three hot chain searches during 2 10 generations for Lysianassoidea and Talitroidea). the 16S and 4 106 generations for the other genes, sampled every 100 generations. In most cases a 10% burn-in was estimated suf- cient after checking graphically for convergence to stable ln L 3. Results scores with Tracer (Rambaut and Drummond, 2003). We compared posterior probabilities of the splits between runs, as well as during 3.1. Sequences runs using default parameters in the program AWTY (Nylander et al., 2008). High correlation was observed between posterior Sequences of the 18S gene ranged in length from 2310 bp to probabilities of the splits between runs (Pearson correlation test: 2440 bp (Genebank accession numbers: HQ286012–HQ286018) r > 0.98, p < 2.2 1016) and no particular trend diagnosing lack while the 28S sequences (HQ286019–HQ286020 and HQ286022– of convergence was observed (see Supplementary material HQ286024) ranged from 850 to 1440 bp. The length of the se- Fig. A1). The secondary structure and the indels were taken into quences retrieved from Genbank varied from 1150 to 2500 bp for account for the nuclear genes with commands implemented in the 18S and from 820 to 1290 bp for the 28S sequences. Crymosty- MrBayes. The doublet model of evolution was applied to the stems gius thingvallensis had the longest fragments among the Crang- in order to take into account compensatory mutations. Two differ- onyctoidea sequences for both 18S and 28S genes, reecting its ent partitions were used in order to test the sensitivity of the tree unique phylogenetic status within the superfamily. The variation topology to the partitioning: runs were performed considering four in length is due to several indels, ranging from 1 to 53 bases. The different partitions (outside hairpins, stems, internal and terminal genetic distances observed between clones from the same individ- loops) and for just two partitions (stems and loops). Indels were ual never exceeded 0.9%, irrespective of the alignment method and coded as binary characters and included as a morphological data- the nuclear gene. We therefore chose a single clone from each indi- set. PhyML does not provide these options. Branch supports were vidual as representative of each species. The 16S sequences ranged assessed by using the approximate likelihood ratio test (aLRT) in from 362 to 411 bp (Genebank accession numbers: HQ286000– PhyML, applying the non-parametric method based on a Shimoda- HQ286011). No indels and no stop codons were observed in the ira–Hasegawa-like procedure (Anisimova and Gascuel, 2006). Both 582 bp long COI alignment (HQ286025–HQ286037). MrBayes and PhyML trees were rooted by the most closely related taxa available in Genbank outside the superfamily: the 18S with 3.2. Alignment and secondary structure Megaluropus longimerus and the 28S with Gammarus abstrusus (see Table 1)(Englisch et al., 2003). Different alignment methods resulted in different alignments of The secondary structure of the 16S fragment was not taken into sequences as reected by the overall lengths and the number of account in the phylogenetic reconstruction, due to the short se- phylogenetic informative sites for each RNA gene dataset (Table 3). quences. A codon model of evolution was applied to the complete The number of phylogenetically informative sites and sites where COI dataset in MrBayes, which allows for different rates for synon- gaps were introduced varied among the alignment methods. The ymous and non-synonymous substitutions as well as on 0-fold number of gaps introduced was smallest with ClustalW2 and larg- degenerated codon positions in order to avoid phylogenetic signal est with FSA, for 18S: 532 (ClustalW2), 724 (MAFFT), 1277 (FSA) saturation. The COI and 16S phylogenies were rooted by G. abstru- and 664 (RNAsalsa), and for 28S: 446 (ClustalW2), 474 (MAFFT), sus (see Table 1), as used in previous amphipod phylogenies (Eng- 1094 (FSA) and 554 (RNAsalsa). Variation among species depends lisch et al., 2003). In addition to the phylogenies computed for each on the regions of secondary structure. Stems were generally least E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539 531

Table 3 The COI trees reconstructed with the different datasets were Comparison of the likelihoods between genes and alignment methods. The table similar in topology. To avoid a misleading phylogenetic signal presents the results from the Bayesian analysis for the four alignment methods. ln L: due to saturation, we choose the conservative strategy to present the log likelihoods of the Bayesian trees, HPD: high posterior density interval of the log likelihoods, bp: number of base pair in the alignment, IS: number of the tree based on 0-fold degenerated sites. phylogenetically informative sites. 3.3. Phylogenetic relationships among species of Crangonyctoidea, Methods Genes -lnL 95%HPD bp IS based on nuclear genes 18S 12020 12010-12040 2629 439 ClustalW 28S 9859 9849-9870 1561 518 16S 4592 4582-4602 424 262 Species belonging to the Crangonyctidae form a well-supported 18S 12140 12130-12150 3224 352 monophyletic group for both 18S and 28S genes (Fig. 3), indepen- FSA 28S 9984 9974-9995 2115 427 dently of the alignment method (posterior probability: pp > 0.95). 16S 4459 4440-4471 689 249 Crymostygius thingvallensis is clearly differentiated from both Nip- 18S 11350 11340-11360 2840 426 MAFFT 28S 9588 9578-9598 1660 481 hargidae and Crangonyctidae based on both the 18S and 28S phy- 16S 4553 4543-4564 431 255 logenies (Fig. 3). The 28S dataset aligned with ClustalW2 and the 18S 11910 11900-11920 2663 345 18S dataset aligned with RNAsalsa are the only datasets supporting RNAsalsa 28S 10040 10030-10050 1595 466 an early divergence of the Crymostygidae from Niphargidae and 16S 4299 4290-4309 455 260 Crangonyctidae, which cluster together (Table 4). All other phylogenies clustered Crymostygius thingvallensis together with the Cran- gonyctidae. The family Crangonyctidae is composed of two variable: in 18S, the proportion of variable sites ranged from 0.055 monophyletic groups (Fig. 3 and Table 4). One group is formed to 0.072, and the largest differences were found among the loops by the genera Bactrurus and Stygobromus clustering together with (0.116–0.175). The difference in number of informative sites be- the European Synurella ambulans and species of Crangonyx from tween four regions (outside hairpins, stems, internal and terminal Eurasia (i.e. Crangonyx chlebnikovi for the 18S and both C. chlebnik- loops) was signicant, independent of the alignment (Fisher exact ovi and Crangonyx subterraneus for the 28S). The other group is 6 tests: P < 3.44 10 ), for both 18S and 28S datasets. composed of species of Crangonyx from North America and Iceland The mean sequence identity for all nuclear gene alignments (in both phylogenies) and the species of Synurella from North was high, ranging from 71% to 95%, and is substantially higher America (18S phylogeny only). Independent from the alignment than the percentage needed to ensure the alignment accuracy method and the gene studied, all phylogenies support that Crang- required for phylogenetic reconstruction (66%, see Hall, 2008; onyx is clearly polyphyletic (Fig. 3 and Table 4). For example, spe- Kumar and Filipski, 2007). The 16S mean sequence identity val- cies of Crangonyx from Eurasia (C. chlebnikovi and C. subterraneus) ues are more marginal and ranged from 60% to 70%; the max- are more closely related to Stygobromus and Bactrurus (pp > 0.95) imum value was obtained with the FSA alignment. Based on the than to species of Crangonyx from North America and Iceland AIC, the best-t model, independently chosen for all genes, was (Fig. 3a). Similarly, S. ambulans from Europe is more closely related the GTR + I + G model (Lanave et al., 1984). The TN93 model to species of Stygobromus and Bacturus (pp > 0.95) than to species (Tamura and Nei, 1993), the most similar substitution model of Synurella species from North America (Synurella sp. and Synurella to GTR and implemented in R, was used for pairwise distance dentata), based on the 18S dataset. This dataset also supports a calculations. group formed by Amurocrangonyx arsenjevi, Bactrurus, Stygobromus, Similar consensus secondary structures were obtained with the and the species of Crangonyx and Synurella from Eurasia (pp > 0.95, different alignment methods of the 18S and 28S genes as shown by see Fig. 3a). the mountain plots (Fig. 1), except for the rst part of the RNAsalsa The 18S phylogenies also show an early divergence of Crangonyx alignment of the 18S gene which exhibits much shorter stem struc- islandicus from other Crangonyx from North America, which are tures than those observed with other alignment methods. The FSA more closely related to Synurella sp. and S. dentata (Fig. 3a). This pat- alignments induced a consensus secondary structure with rela- tern is observed, though less frequently, for 28S phylogenies and tively longer loops for both genes, along with shorter stem struc- combined datasets, depending on the alignment and the phyloge- tures for the rst part of the 18S. No difference was observed netic methods used (Table 4). Moreover, the divergence observed between phylogenetic trees constructed considering the four parti- between Crangonyx islandicus and other Crangonyx greatly exceed tions based on the secondary structure (outside hairpins, stems, the one observed between species of the genera Stygobromus and internal and terminal loops) or two partitions (stems and loops) Bactrurus or even between those species and A. arsenjevi (Fig. 3a). in MrBayes. Conversely, phylogenies based on the 28S dataset, not encompassing The sample of most probable trees after burn-in in Bayesian species of Synurella from North America, support the monophyly of analysis varied considerably in size between the datasets aligned the group formed by species of Crangonyx from North America and with different methods. Samples of trees based on FSA and Clu- Crangonyx islandicus (pp > 0.95, see Fig. 3b). The species of Crangonyx stalW2 were dominated by few, highly probable, different topol- from North America (Crangonyx forbesi, Crangonyx sp., Crangonyx ogies whereas MAFFT-aligned datasets produced a sample of pseudogracilis and Crangonyx oridanus) only clustered together in numerous different topologies with low posterior probabilities. the phylogeny based on MAFFT alignment for the 18S gene. None- RNAsalsa showed intermediate numbers of most probable trees. theless, the early divergence of C. forbesi from the other species is The log likelihood of the trees based on the different alignments only supported by informative indels. All 18S phylogenies con- showed clear patterns which depended on the length of the se- structed without indels as morphological characters strongly sup- 2 quence alignments (ANCOVA, adjusted R = 0.998) (Fig. 2). The ported the monophyly of the North American species of Crangonyx FSA alignments produced signicantly the most likely trees in (pp > 0.95, data not shown). all cases (Table 3 and Fig. 2). FSA is a conservative method, avoid- The different alignment methods result in slight differences in ing stringently false homologies, and consequently shows align- topology, which are mostly in the external nodes of the trees, ments with less informative sites (Table 3). For these reasons grouping together species of Stygobromus, S. ambulans, C. chlebnik- we chose to present the trees based on FSA alignment techniques ovi and C. subterraneus. These changes appeared even for nodes and discuss the differences obtained with the other alignment which were highly supported in the phylogeny based on FSA methods. alignment. The monophyly of the genus Bactrurus, based on 18S, 532 E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539

(a)

ClustalW2 FSA MAFFT RNAsalsa height 0204060

0 500 1000 1500 2000 2500 position (bp)

(b)

ClustalW2 FSA

0 MAFFT RNAsalsa 04 height 02

0 500 1000 1500 position (bp)

Fig. 1. Mountain plots of the secondary structure obtained with the four alignment methods, for the 18S genes (a) and 28S genes (b). The height of the lines reects the secondary structure: the higher the peaks, the longer the stem. A horizontal line presents loop regions.

observed above with different alignment techniques (Fig. A2). Trees based on the combined dataset and aligned with MAFFT ClustalW2 FSA and FSA showed the same topology, different from the one based Mafft on ClustalW2 alignment. They highly support an early divergence RNAsalsa of Crangonyx islandicus from C. forbesi and C. pseudogracilis (pp = 0.81 for FSA and pp = 1 for MAFFT). The species of Stygobro- mus clustered together and C. chlebnikovi clustered with S. ambulans, but both with less support (0.51 < pp < 0.92).

3.4. Phylogenetic relationships among species of Crangonyctoidea, based on mitochondrial genes

The phylogenies based on the mtDNA fragments of COI and Mean log likelihood 16S genes are less well supported, probably due to the short sequence lengths and signal saturation (Fig. 4). Nonetheless, COI and 16S phylogenies support the polyphyly of the genus Crang- onyx. The phylogenetic relationships at a deeper level are less clear, due to the faster rate of evolution of these two genes.

−12000 −10000 −8000 −6000 −4000 Due to the particularly weak resolution of the COI gene for these 6.0 6.5 7.0 7.5 8.0 phylogenetic relationships, we choose not to present it in Table 4. Log of number of bases In the 16S phylogeny, the cluster with Crymostygius thingvallensis and the two species of Pseudocrangonyctidae (Pseudocrangonyx Fig. 2. Evaluation of the likelihoods for trees based on the different genes and korkishkoorum and Procrangonyx primoryensis) from Eastern Asia alignment methods. Mean log likelihoods of the trees obtained with MrBayes are is relatively well supported (supported by all alignment tech- presented with respect to the logarithm of the length of the aligned sequences in base pairs. Lines present the prediction from analysis of covariance (ANCOVA) niques with 0.61 < pp < 1). This cluster could not be veried by (adjusted R2 = 0.998). Both linear coefcients (intercept and slope) for the FSA the nuclear gene phylogenies due to difculties in amplifying se- alignment were signicantly different from the estimates based on the other quences for P. korkishkoorum and P. primoryensis with the prim- alignment methods (p < 0.008). ers used. The 16S dataset supports, though weakly (pp = 0.77), the clustering of Crangonyx islandicus with the North American species Crangonyx serratus. As obtained with the 18S, 28S and is supported by all alignments whereas Stygobromus is only mono- 16S datasets, C. chlebnikovi and S. ambulans appeared more clo- phyletic in the tree based on ClustalW2 alignment (Table 4). The sely related to species of Stygobromus than to other species of monophyly of Stygobromus was never supported by the 28S phy- Crangonyx from North America (Figs. 3 and 4a). logenies, independent of the alignment (Table 4). The COI phylogeny (Fig. 4b) shows poor resolution for the The tree based on the 18S–28S combined datasets shows a sim- internal nodes of the tree. The species of Crangonyx from North ilar topology. The changes observed are due to the positioning of America (C. pseudogracilis, Crangonyx sp. and C. oridanus) clus- Crangonyx islandicus, C. forbesi, C. chlebnikovi and S. ambulans,as ter together (pp = 1) but no conclusion can be drawn from this E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539 533

(a) (b)

Fig. 3. Bayesian phylogenetic tree of the crangonyctid species, together with Crymostygius thingvallensis and two Niphargus species, based on 18S (a) and 28S (b) sequences aligned with FSA. Posterior probability values are displayed in the box. The tree is rooted by Megaluropus longimerus for the 18S and by Gammarus abstrusus for the 28S; outgroups are not shown. Stars and squares correspond to Palearctic and Nearctic species respectively. Branches are drawn to scale, with the bar indicating 0.05 expected changes per site. dataset about their phylogenetic relationship with Crangonyx 4. Discussion islandicus. Synurella sp., an inhabitant from North America, is closely related to A. arsenjevi and the North American species The different methods of alignment and phylogenetic recon- of Crangonyx (pp = 0.86). Crymostygius thingvallensis did not clus- struction applied in this study allow several main conclusions ter with any of the species integrated in the COI phylogeny. C. to be made: (1) The taxonomic status of Crymostygius thingvall- chlebnikovi from the Ural Mountains clustered with P. primoryensis as a new family is conrmed. (2) Crangonyx islandicus ensis (pp = 1) from Eurasia, which belongs to the family may represent a new genus. (3) As commonly observed among Pseudocrangonyctidae. subterranean species, discrepancies appear between molecular The tree based on 16S-COI combined dataset shows an early and morphological taxonomy, which may have resulted from divergence of Crangonyx islandicus and C. pseudogracilis from the convergent evolution in morphological traits. Several species of cluster formed by the rest of the species (Fig. A3). Stygobromus gra- Crangonyctidae need taxonomical revision. (4) Phylogeographic cilipes and S. ambulans are clustered together (pp = 0.91). Crymosty- patterns indicate colonization of Iceland from the Nearctic via gius thingvallensis is more closely related (pp = 0.98) to the cluster Greenland. formed by C. chlebnikovi and P. primoryensis (pp = 1). 4.1. Molecular and morphological taxonomy 3.5. Molecular divergence The tree topologies varied less among phylogenetic reconstruc- The genetic divergence observed for both 18S and 28S genes tion methods than among alignment methods, which showed only within the crangonyctoid genera is exceptionally high compared minor differences, particularly in external branches, even for well- to available data from other amphipod genera (Fig. 5). This diver- supported nodes. This emphasizes the need in future studies on gence is considerably reduced when Crangonyx islandicus, C. chleb- RNA phylogenies to consider different alignment methods, as com- nikovi, C. subterraneus and S. ambulans are omitted from the monly done when results from different tree reconstruction meth- comparison, in which case the divergence observed between ods are compared. Crangonyx islandicus and other species of Crangonyx is more similar The apparent monophyly of species in the family Cran- to the comparison between genera than within. The variation be- gonyctidae is strongly supported by the 18S and 28S nuclear tween Crymostygius thingvallensis and species from other families genes. The classication of Crymostygidae as a monotypic family of Crangonyctoidea is comparable to the one observed between within the superfamily Crangonyctoidea (Kristjnsson and families (Fig. 5). Svavarsson, 2004) is further supported by the phylogenies Divergence among Crangonyx islandicus populations for both presented in this study and it appears as a sister family to the 18S and 28S sequences was much less than observed for the Crangonyctidae. Its status is also supported by the genetic dis- mtDNA genes, COI and 16S, as reported by Kornobis et al. (2010) tances observed between Crymostygius thingvallensis and the (data not shown). Interestingly, the only three phylogenetically two other families of the Crangonyctoidea, which are similar to informative sites found along an 800 bp fragment of the 28S sup- genetic distances observed between families among other port an earlier divergence of the populations in northern Iceland amphipod superfamilies. A further investigation is needed to from populations in southern Iceland. assess their phylogenetic relationships with species of 534

Table 4 Support of different hypothesis according to different phylogenetic reconstruction methods and alignment methods for the RNA genes. NA: insufcient data for testing a hypothesis for a particular gene. %: percentage of trees supporting the hypothesis among all computed trees. # trees: number of relevant trees to test a hypothesis.

Hypothesis ClustalW2 FSA MAFFT RNAsalsa Total % # trees 18S 28S 16S 28S–18S 18S 28S 16S 28S–18S 18S 28S 16S 28S–18S 18S 28S 16S 28S–18S .Kroi ta./MlclrPyoeeisadEouin5 21)527–539 (2011) 58 Evolution and Phylogenetics Molecular / al. et Kornobis E. MrBayes Crangonyx monophyly 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 63 16 Crymostygiidae sister family of Crangonyctidae 1 0 0 1 1 1 0 1 1 1 0 1 0 1 0 1 10 63 16 Bactrurus monophyly 1 NA NA NA 1 NA NA NA 1 NA NA NA 1 NA NA NA 4 100 4 Stygobromus monophyly 1 0 0 0 0 0 0 1 0 0 0 1 0 0 0 1 4 25 16 Synurella monophyly 0 NA NA NA 0 NA NA NA 0 NA NA NA 0 NA NA NA 0 0 4 Crangonyctidae monophyly 1 1 0 1 1 1 0 1 1 1 0 1 1 1 0 1 12 75 16 Crangonyx islandicus diverged early from North American Crangonyx 10NA0 10NA1 1011 1011 9 5616 Amurocrangonyx cluster with Stygobromus and Bactrurus 1NA0NA1NA0NA1NA0NA1NA0NA4 508 Crangonyctidae divided m two main groups (described m the text) 1 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 14 88 16 Crangonyx chlebnikovi clustered with Stygobromus and Bactrurus 1111 1111 1111 1111 1610016 Synurella ambulans clustered with Stygobromus and Bactrurus 1111 1111 1111 1111 1610016 Pseudocrangonyx, Procrangonyx and Crymostygius cluster together NA NA 1 NA NA NA 1 NA NA NA 1 NA NA NA 1 NA 4 100 4 PhyML Crangonyx monophyly 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 16 Crymostygiidae sister family of Crangonyctidae 1 0 0 1 1 1 0 1 1 1 0 1 1 1 0 1 11 69 16 Bactrurus monophyly 1 NA NA NA 1 NA NA NA 1 NA NA NA 1 NA NA NA 4 100 4 Stygobromus monophyly 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 6 16 Synurella monophyly 0 NA NA NA 0 NA NA NA 0 NA NA NA 0 NA NA NA 0 0 4 Crangonyctidae monophyly 1 1 0 1 1 1 0 1 1 1 0 1 1 1 0 1 12 75 16 Crangonyx islandicus diverged early from North American Crangonyx 1011 1111 1011 1010 127516 Amurocrangonyx cluster with Stygobromus and Bactrurus 1 NA 0 NA 0 NA 0 NA 1 NA 0 NA 1 NA 0 NA 3 38 8 Crangonyctidae divided in two main groups (described in the text) 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 15 94 16 Crangonyx chlebnikovi clustered with Stygobromus and Bactrurus 1111 1111 1111 1111 1610016 Synurella ambulans clustered with Stygobromus and Bactrurus 1111 1111 1111 1111 1610016 Pseudocrangonyx, Procrangonyx and Crymostygius cluster together NA NA 1 NA NA NA 1 NA NA NA 1 NA NA NA 1 NA 4 100 4 E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539 535

(a) (b)

Fig. 4. Bayesian phylogenetic tree of the crangonyctid species, together with Crymostygius thingvallensis, based on 16S sequences aligned with FSA including two Niphargus species (a) and based on COI sequences (b). Posterior probability values are displayed in the box. The trees are rooted by Gammarus abstrusus (not shown on the gure). Stars and squares correspond to Palearctic and Nearctic species respectively. Branches are drawn to scale, with the bar indicating 0.05 expected changes per site.

(a) (b) 0.1 0.2 0.3 0.4 0.00 0.05 0.10 0.15 12345 6 7 123 4567

Fig. 5. Comparison of genetic distances within Crangonyctoidea with distances within and among other amphipod families. Genetic distances from ClustalW2 aligned dataset compared among various taxonomic groups for 18S (a) and 28S (b). #1: within genera, #2: within genera without Crangonyx islandicus, S. ambulans, Crangonyx chlebnikovi, #3: between genera, #4: between genera without Crangonyx islandicus, S. ambulans, Crangonyx chlebnikovi, #5: between Crangonyx islandicus other Crangonyx, #6: between families without Crymostygius thingvallensis, #7: between Crymostygius thingvallensis and other families. Horizontal lines represent the maximum genetic distance observed for other amphipod families: broken line: between species within genera, dotted line: between genera, mixed: between families. 536 E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539

Pseudocrangonyctidae, which is described as a closely allied rangonyx and Pseudocrangonyx are included within the sister family to the Crangonyctidae (Holsinger, 1994). A further Crangonyctidae (Holsinger, 1994). As described by Sidorov and study is needed to assess the monophyly and the phylogenetic Holsinger (2007), Amurocrangonyx is a apparent sister genus of relationships among families of the superfamily Crangonyctoidea, Crangonyx; this phylogenetic relation is partially supported by including for instance other species of Crangonyctidae, (e.g. the 18S tree in our analysis. Crangonyx africanus) and members of the families Sternophysin- Holsinger (1977) presented two opposing theories to explain gidae, Paramelitidae and Neoniphargidae from the Southern the evolution of the crangonyctid genera: (1) One lineage evolved Hemisphere. into Crangonyx and Synurella and another into Stygobromus and The observed phylogenies in this study suggest that species Bactrurus. (2) One lineage evolved into Bactrurus and Crangonyx of Crangonyctidae need taxonomic revisions in order to restore and another into Synurella and Stygobromus. The 18S phylogenies the monophyly of their respective genera. Phylogenies based on support, though weakly, an early divergence of Synurella and all genes, except for some 28S phylogenies, cluster Crangonyx Crangonyx from the Bactrurus and Stygobromus groups. The same islandicus separately from the genus Crangonyx. Svavarsson and phylogenetic relationships between these four genera have already Kristjnsson (2006) hesitated to describe Crangonyx islandicus been observed by Holsinger (1994), based on morphological syna- as a new genus based on morphological differences. The ab- pomorphies. Nonetheless, the polyphyly observed for both Crang- sence of sternal gills, lateral lobe of head truncated, pereopod onyx and Synurella encourage a revision of the classication of 7 longest, uropod 3 short with inner ramus naked, and the Crangonyx, Synurella and Stygobromus before reaching a nal con- presence of only a single strong spine on the inner plate of clusion on the evolution of the crangonyctid genera. More speci- the maxilliped are a unique combination of morphological traits cally, Crangonyx islandicus, C. chlebnikovi, S. ambulans and S. dentata which distinguishes Crangonyx islandicus from the other species need taxonomic revision in order to restore the monophyly of the of the genus Crangonyx (Svavarsson and Kristjnsson, 2006). Our genera. molecular data supports that these characters might be given The group formed by Stygobromus, Bactrurus, Amurocrangonyx more weight in the phylogenetic classication and leads us to and the species of Crangonyx and Synurella from Eurasia included envisage Crangonyx islandicus as a new genus. The similar ge- in this study are almost exclusively eyeless (Holsinger, 1977; netic distances observed between genera within Crangonyctidae Sidorov and Holsinger, 2007). The exception is the presence of and between Crangonyx islandicus and other Crangonyx also sup- vestigial eyes composed of a cluster of ommatidia in S. ambulans port the new generic status of Crangonyx islandicus. Phylogenies (see Karaman, 1974; Arbacˇiauskas, 2008). Conversely, most spe- that include more species of Crangonyx and Synurella from cies of Crangonyx from North America (Holsinger, 1977), S. den- North America may assist in lending more support to this tax- tata (Hubricht, 1943) and Crangonyx islandicus (Svavarsson and onomic revision. Kristjnsson, 2006) possess vestigial eyes. The two groups are Mitochondrial DNA variation in Crangonyx islandicus popula- reciprocally monophyletic and as vestigial eyes are found in tions revealed extensive diversity, suggesting a cryptic species the more distantly related Crymostygius thingvallensis (Kristjnsson complex. Maximum divergence was observed for a population and Svavarsson, 2004), it suggests that vestigial eyes is an ances- in northeastern Iceland which had diverged from the other pop- tral trait. The phylogeny reects that the evolutionary trend to- ulations in Iceland about 5 millions years ago (Kornobis et al., ward the loss of the eyes might have appeared more rapidly in 2010). Much less variation is observed with the conservative the group formed by Crangonyx and Synurella from Eurasia and RNA genes. The few segregating sites in the 28S gene group in Bacturus and Stygobromus from North America than in Crang- northeastern populations with populations from northern onyx and Synurella from North America. A further study of the Iceland. phylogeographical patterns among species of Crangonyctidae is Our molecular study supports the monophyly of the genus needed to get a better understanding of regression of the eye Bactrurus. In accordance with previous classication based on structure. morphological data, C. oridanus and C. pseudogracilis appeared closely related to each other, based on the 18S and COI phylog- 4.2. Geographical patterns and the colonization of Iceland enies (Zhang and Holsinger, 2003). The 18S marker supports an early divergence of C. forbesi from those two species, as de- Species both from Crangonyx and Synurella showed genetic clus- scribed by the morphological analysis of Zhang and Holsinger tering that did not support the previous morphological taxonomy (2003). but which can instead be explained by their geographical distribu- Both Crangonyx and Synurella are polyphyletic. The monophyly tion. C. chlebnikovi and C. subterraneus, both found in the Palearctic of the genus Stygobromus is poorly supported by the molecular region, showed clear divergence from species of Crangonyx from data. Consequently, revision of the classication within the Cran- the Nearctic region. Synurella exhibits the same geographical pat- gonyctidae is needed as well as the morphological and meristic tern, with S. ambulans in Europe being more closely related to spe- taxonomic criteria previously used, which appear to be homopla- cies of Crangonyx in the Palearctic than to species of Synurella from sic. Genetic distances between genera within Crangonyctidae are the Nearctic. This phylogeographic pattern and the monophyly of larger than distances between genera in several other families of the crangonyctid species tend to support a hypothesis that sug- amphipods, suggesting that morphological differences among taxa gests the occurrence of an ancestor to the family that was wide- of the Crangonyctidae are subtle. spread in Laurasia before its breakup in Late Paleocene (about The classication of C. chlebnikovi based on morphology is un- 60 Mya). der revision (Sidorov et al., unpublished). The presence of a lateral Considering that the closest relatives of Crangonyx islandicus lobe on the head, which is rather prominent and narrowly were found to be Nearctic species, as well as the high diversity rounded anteriorly with inferior antennal sinus, is a morphologi- within the Nearctic species of Crangonyctidae and the geological cal characteristic common to the genus Bactrurus but not to the age of Iceland, we conclude that Crangonyx islandicus colonized genus Crangonyx (Sidorov et al., 2010). P. primoryensis and P. kor- Iceland from Greenland as hypothesized by Kristjnsson and kishkoorum, in the family Pseudocrangonyctidae, are supported as Svavarsson (2007). Their ancestor either followed proto-Iceland closely related species by the 16S gene phylogenies. Nonetheless, as it departed from Greenland or colonized Iceland via the more data is necessary to reach a conclusion about their phyloge- Greenland–Iceland land bridge. An alternative colonization route netic relationships with the Crangonyctidae, especially if Proc- along the Scotland-Faeroe-Iceland ridge is not supported. Since E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539 537

Crangonyx islandicus and Crymostygius thingvallensis belong to the submerged by the sea. Similar examples of organisms predating freshwater superfamily Crangonyctoidea, their respective ances- the age of their habitat are known from the Galapagos archipel- tors are more likely to have inhabited freshwater as well. We ago, e.g. among iguanas (Rassmann, 1997) and from the Hawai- can therefore argue that freshwater contacts occurred between ian islands, where organisms such as Drosophila and Hawaiian Iceland and Greenland. A tolerance to limited salinity (about 1/ honey creepers colonized new islands as the older islands dis- 100 of sea salinity) of a related species, C. pseudogracilis (see appeared by erosion and subduction due to plate tectonics Slothouber Galbreath et al., 2010) indicates that there might (see Fleischer et al., 1998). though have been a possible route of colonization when the land The records of freshwater amphipods from other formerly glaci- bridge was reduced to a chain of islands. Two Synurella species ated areas are rare and no such species or fossil has been described (S. jakutana and S. levanidovorum) from the coast of the Okhotsk from Greenland. In cases where freshwater amphipods have been Sea are known to inhabit brackish waters (Sidorov, unpublished). found in formerly glaciated areas, post-glacial colonization cannot Resistance of the Icelandic species to salinity has not been be ruled out, e.g. in the Alps (Lefbure et al., 2006b). Fissure areas tested. and geothermal energy have been hypothesized as providing suit- Past colonization events from Greenland to Iceland through able refugia during the Ice Age for subterranean species in Iceland the Greenland–Iceland land bridge have been inferred from fos- (Kornobis et al., 2010). The absence of such geological regions else- sil records for mammals (McKenna, 1983) and plant species where in formerly glaciated areas of the Northern Hemisphere (Grmsson et al., 2007; Denk et al., 2010). As the ice sheet cov- might explain the limited distribution of groundwater amphipods ered Iceland repeatedly during the glacial period of the Ice Age, in these regions. most species which colonized Iceland before the Ice Age became extinct. The only species which might have survived in Acknowledgments refugia below the glacier were the groundwater species, such as amphipods, but Crangonyx islandicus is the only known spe- We acknowledge Cene Fišer for providing valuable samples cies which has survived glaciation in Iceland (Kornobis et al., of Synurella ambulans. We want to thank Virginia Escudero 2010). Crangonyx islandicus and Crymostygius thingvallensis are and Jaume Cuart Castell for laboratory assistance. We are thus probably the oldest known inhabitants of Iceland and their grateful to the University of Iceland Research Fund and the early colonization may predate the oldest rock formation in Ice- Icelandic Research Council (Ranns) for nancial support. Two land, 16 Mya (Moorbath et al., 1968). The oldest rock forma- anonymous reviewers made comments that improved the tions of the Iceland hotspot date back to 40 Mya and are manuscript. found in the Faeroe Islands and East Greenland. As Iceland drifted apart at the tectonic boundaries, only the youngest part of Iceland remained above sea level around the hotspot beneath Appendix A Iceland. Terrestrial organisms present on Iceland would need to colonize the younger parts from the older ones before being See Table A1.

Table A1 Genbank accession numbers of the sequences for each gene used for tree construction.

This Study 18S 28S 16S COI Amurocrangonyx arsenjevi HQ286015 HQ286007 HQ286025 Crangonyx chlebnikovi HQ286017 HQ286023 HQ286001 HQ286026 Crangonyx islandicus F HQ286006 HQ286030 Crangonyx islandicus F HQ286005 HQ286029 Crangonyx islandicus S HQ286013 HQ286020 HQ286004 HQ286027 Crymostygius thingvallensis HQ286012 HQ286019 HQ286009 HQ286032 Procrangonyx primoryensis HQ286011 HQ286033 Pseudocrangonyx korkishkoorum HQ286010 Stygobromus gracilipes 1 HQ286016 HQ286022 HQ286002 HQ286034 Stygobromus gracilipes 2 HQ286003 HQ286035 Stygobromus stegerorum HQ286014 HQ286024 HQ286008 HQ286036 Synurella ambulans HQ286018 HQ286000 HQ286037 From Genbank Bactrurus brachycaudus AF202979 Bactrurus mucronatus AF202978 Bactrurus pseudomucronatus AF202985 Crangonyx oridanus AJ966709 AJ968911 Crangonyx forbesi AF202980 EU693287 Crangonyx pseudogracilis AJ966705 EF582940 EF582845 AJ968900 Crangonyx serratus AY926703 Crangonyx sp. AJ966706 AY529053 Crangonyx subterraneus EU693288 Gammarus abstrusus EF582903 EF582950 EF582855 EF570304 Megaluropus longimerus DQ378035 Niphargus fontanus AF202981 EF617304 EF028464 Niphargus kochianus AF419221 EU693308 Stygobromus mackini DQ377995 Synurella ambulans EF617236 Synurella dentata AF419233 Synurella sp. AJ966706 AJ968913 538 E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539

Appendix B. Supplementary material Lanave, C., Preparat, G., Saccone, C., Serio, G., 1984. A new method for calculating evolutionary substitution rates. J. Mol. Evol. 20, 86–93. Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan, P.A., McWilliam, Supplementary data associated with this article can be found, in H., Valentin, F., Wallace, I.M., Wilm, A., Lopez, R., Thompson, J.D., Gibson, T.J., the online version, at doi:10.1016/j.ympev.2010.12.010. Higgins, D.G., 2007. ClustalW and ClustalX version 2. Bioinformatics 23, 2947– 2948. Lawver, L.A., Mller, R.D., 1994. Iceland hotspot track. Geology 22, 311–314. Lefbure, T., Douady, C.J., Gouy, M., Trontelj, P., Briolay, J., Gibert, J., 2006a. References Phylogeography of a subterranean amphipod reveals cryptic diversity and dynamic evolution in extreme environments. Mol. Ecol. 15, 1797–1806. Akaike, H., 1974. A new look at the statistical model identication. IEEE Trans. Lefbure, T., Douady, C.J., Malard, F., Gibert, J., 2006b. Testing dispersal and cryptic Automat. Contr. AC-19, 716–723. diversity in a widely distributed groundwater amphipod (Niphargus Anisimova, M., Gascuel, O., 2006. Approximate likelihood ratio test for branches: a rhenorhodanensis). Mol. Phylogenet. Evol. 42, 676–686. fast, accurate and powerful alternative. Syst. Biol. 55, 539–552. Lundin, E., Dor, A., 2002. Mid-Cenozoic post-breakup deformation in the Arbacˇiauskas, K., 2008. Synurella ambulans (F. Mller, 1846), a new native passive margins bordering the Norwegian–Greenland Sea. Mar. Petrol. amphipod species of Lithuanian waters. Acta Zool. Lithuan. 18, 66–68. Geol. 19, 79–93. Bernhart, S., Hofacker, I., Will, S., Gruber, A., Stadler, P., 2008. RNAalifold: improved MacDonald, K.S., Yampolsky, L., Duffy, J.E., 2005. Molecular and morphological consensus structure prediction for RNA alignments. BMC Bioinform. 9, 474– evolution of the amphipod radiation of Lake Baikal. Mol. Phylogenet. Evol. 35, 486. 323–343. Bradley, R., Roberts, A., Smoot, M., Juvekar, S., Do, J., Dewey, C., Holmes, I., Pachter, L., Mallatt, J., Giribet, G., 2004. Further use of nearly complete 28S and 18S rRNA genes 2009. Fast statistical alignment. PLoS Comput. Biol. 5, 1–15. to classify Ecdysozoa: 37 more arthropods and a kinorhynch. Mol. Phylogenet. Cristescu, M., Hebert, P., 2005. The Crustacean Seas – an evolutionary perspective Evol. 40, 772–794. on the Ponto-Caspian peracarids. Can. J. Fish. Aquat. Sci. 62, 505–517. McKenna, M., 1983. Holarctic landmass rearrangement, cosmic events, and Denk, T., Grmsson, F., Zetter, R., 2010. Episodic migration of oaks to Iceland: Cenozoic terrestrial organisms. Ann. Mo. Bot. Gard. 70, 459–489. evidence for a North Atlantic land bridge in the latest Miocene. Am. J. Bot. 97, Morrison, D., 2008. A framework for phylogenetic sequence alignment. Plant Syst. 276–287. Evol. 282, 127–149. Edgar, R., Batzoglou, S., 2006. Multiple sequence alignment. Curr. Opin. Struct. Biol. Morrison, D., 2009. Why would phylogeneticists ignore computerized sequence 16, 368–373. alignment? Syst. Biol. 58, 150–158. Eldholm, O., Myhre, A.M., Thiede, J., 1994. Cenozoic tectono-magmatic events in the Moorbath, S., Sigurdsson, H., Goodwin, R., 1968. K–Ar ages of the oldest exposed North Atlantic: potential palaeoenvironmental implications. In: Boulter, M.C., rocks in Iceland. Earth Planet Sci. Lett. 4, 197–205. Fisher, H.C. (Eds.), Cenozoic Plants and Climates of the Arctic. NATO ASI Series, Novk, A., Mikls, I., Lyngs, R., Hein, J., 2008. StatAlign: an extendable software vol. 127. Springer, Berlin, pp. 35–55. package for joint Bayesian estimation of alignments and evolutionary trees. Englisch, U., Coleman, C.O., Wgele, J.W., 2003. First observations on the phylogeny Bioinformatics 24, 2403–2404. of the families Gammaridae, Crangonyctidae, Melitidae, Niphargidae, Nylander, J.A.A., Wilgenbusch, J.C., Warren, D.L., Swofford, D.L., 2008. AWTY (are we Megaluropidae and Oedicerotidae (Amphipoda, Crustacea), using small there yet?): a system for graphical exploration of MCMC convergence in subunit rDNA gene sequences. J. Nat. Hist. 37, 2461–2486. Bayesian phylogenetics. Bioinformatics 24, 581–583. Englisch, U., Koenemann, S., 2001. Preliminary phylogenetic analysis of selected Ogden, T., Rosenberg, M., 2006. Multiple sequence alignment accuracy and subterranean amphipod crustaceans, using small subunit rDNA gene sequences. phylogenetic inference. Syst. Biol. 55, 314–328. Org. Divers. Evol. 1, 139–145. Page, T., Short, J.W., Humphrey, C.L., Hillyer, M.J., Hughes, J.M., 2008. Molecular Fišer, C., Sket, B., Trontelj, P., 2008. A phylogenetic perspective on 160 years of troubled systematics of the Kakaducarididae (Crustacea: Decapoda: Caridea). Mol. taxonomy of Niphargus (Crustacea: Amphipoda). Zool. Scr. 37, 665–680. Phylogenet. Evol. 46, 1003–1014. Fleischer, R., McIntosh, C., Tarr, C., 1998. Evolution on a volcanic conveyor belt: Palumbi, S., Martin, A., Romano, S., McMillan, W.O., 1991. The Simple Fools Guide to using phylogeographic reconstructions and K–Ar-based ages of the PCR, Version 2.0. University of Hawaii, Honolulu. pp. 43. Hawaiian Islands to estimate molecular evolutionary rates. Mol. Ecol. 7, Paradis, E., Claude, J., Strimmer, K., 2004. APE: analyses of phylogenetics and 533–545. evolution in R language. Bioinformatics 20, 289–290. Folmer, O., Black, M., Hoeh, W., Lutz, R., Vrijenhoek, R., 1994. DNA primers for Poore, R.H., 2008. Neogene Epeirogeny and the Iceland Plume. Ph.D. Dissertation, amplication of mitochondrial cytochrome c oxidase subunit I from diverse University of Cambridge, Cambridge. metazoan invertebrates. Mol. Mar. Biol. Biotechnol. 3, 294–299. Porter, M.L., 2007. Subterranean biogeography: what have we learned from Grmsson, F., Denk, T., Smonarson, L.A., 2007. Middle Miocene oras of molecular techniques? J. Cave Karst Stud. 69, 179–186. Iceland—the early colonization of an island? Rev. Palaeobot. Palyno. 144, Posada, D., Buckley, T., 2004. Model selection and model averaging in 181–219. phylogenetics: advantages of Akaike information criterion and Bayesian Guindon, S., Gascuel, O., 2003. A simple, fast, and accurate algorithm to estimate approaches over likelihood ratio tests. Syst. Biol. 53, 793–808. large phylogenies by maximum likelihood. Syst. Biol. 52, 696–704. R Development Core Team, 2010. R: A Language and Environment for Statistical Hall, G.H., 2008. Phylogenetic Trees made Easy, third ed. Sunderland, Computing. Foundation for Statistical Computing, Vienna. Massachusetts. Rambaut, A., Drummond, A.J., 2003. Tracer Version 1.2. University of Oxford, Hall, T., 1999. BioEdit: a user-friendly biological sequence alignment editor and Oxford. analysis program for Windows 95/98/NT. Nucl. Acids. Symp. Ser. 41, 95–98. Rassmann, K., 1997. Evolutionary age of the Galpagos iguanas predates the age of Holsinger, J.R., 1977. A review of the systematics of the Holarctic amphipod family the present Galpagos islands. Mol. Phylogenet. Evol. 7, 158–172. Crangonyctidae. Crustaceana 4, 244–281. Ronquist, F., Huelsenbeck, J., 2003. MrBayes 3: Bayesian phylogenetic inference Holsinger, J.R., 1994. Pattern and process in the biogeography of subterranean under mixed models. Bioinformatics 19, 1572–1574. amphipods. Hydrobiologia 287, 131–145. Sidorov, D.A., Holsinger, J.R., 2007. Amurocrangonyx, a new genus of subterranean Hou, Z., Fu, J., Li, S., 2007. A molecular phylogeny of the genus Gammarus amphipod (Crangonyctidae) from the Russian far east, with a redescription of (Crustacea: Amphipoda) based on mitochondrial and nuclear gene sequences. the poorly known Crangonyx arsenjevi and comments on biogeographic Mol. Phylogenet. Evol. 45, 596–611. relationships. J. Crustacean Biol. 27, 660–669. Hubricht, L., 1943. Studies in the Nearctic freshwater Amphipoda, III. Am. Midl. Nat. Sidorov, D.A., Pankov, N.N., Krasheninnikov, A.B., 2010. A Bactrurus-like 29, 683–712. subterranean amphipod (Crangonyctidae) from the Ural Mountain karst Karaman, G.S., 1974. Genus Synurella Wrzes. In Yugoslavia with remarks on its all region. In: Abstracts of 20th International Conference of Subterranean world known species, their synonymy, bibliography and distribution (fam. Biology, Postojna, p. 145. Gammaridae). Poljopriv. Sumar 20, 83–133. Slothouber Galbreath, J.G.M., Smith, J.E., Becnel, J.J., Butlin, R.K., Dunn, A.M., 2010. Katoh, H., Toh, H., 2008. Recent developments in the MAFFT multiple sequence Reduction in post-invasion genetic diversity in Crangonyx pseudogracilis alignment program. Brief. Bioinform. 9, 286–298. (Amphipoda: Crustacea): a genetic bottleneck or the work of hitchhiking Katoh, K., Asimenos, G., Toh, H., 2009. Multiple alignment of DNA sequences with vertically transmitted microparasites? Biol. Invas. 12, 191–209. MAFFT. In: Posada, D. (Ed.), Bioinformatics for DNA Sequence Analysis. Humana Stocsits, R.R., Letsch, H., Hertel, J., Misof, B., Stadler, P.F., 2009. Accurate and efcient Press, New York. reconstruction of deep phylogenies from structured RNAs. Nucl. Acids. Res. 37, Kornobis, E., Plsson, S., Kristjnsson, B.K., Svavarsson, J., 2010. Molecular evidence 6184–6193. of the survival of subterranean amphipods (Arthropoda) during Ice Age Svavarsson, J., Kristjnsson, B.K., 2006. Crangonyx islandicus sp. nov., a subterranean underneath glaciers in Iceland. Mol. Ecol. 19, 2516–2530. freshwater amphipod (Crustacea, Amphipoda, Crangonyctidae) from springs in Kristjnsson, B.K., Svavarsson, J., 2004. Crymostygidae, a new family of lava elds in Iceland. Zootaxa 1365, 1–17. subterranean freshwater gammaridean amphipods (Crustacea) recorded from Tamura, K., Nei, M., 1993. Estimation of the number of nucleotide substitutions in subarctic Europe. J. Nat. Hist. 38, 1881–1894. the control region of mitochondrial DNA in humans and chimpanzees. Mol. Biol. Kristjnsson, B.K., Svavarsson, J., 2007. Subglacial refugia in Iceland enabled Evol. 10, 512–526. groundwater amphipods to survive glaciations. Am. Nat. 170, 292–296. Vinl, R., Witt, J.D.S., Bradbury, J.H., Jazdzewski, K., Grabowski, M., Sket, B., 2008. Kumar, S., Filipski, A., 2007. Multiple sequence alignment: in pursuit of homologous Global diversity of amphipods (Amphipoda; Crustacea) in freshwater. DNA positions. Genome Res. 17, 127–135. Hydrobiologia 595, 241–255. E. Kornobis et al. / Molecular Phylogenetics and Evolution 58 (2011) 527–539 539

Wiens, J., Chippindale, P., Hillis, D., 2003. When are phylogenetic analyses misled Zhang, J., Holsinger, J.R., 2003. Systematics of the Freshwater Amphipod Genus by convergence? A case study in Texas cave salamanders. Syst. Biol. 52, Crangonyx (Crangonyctidae) in North America. Virg. Mus. Nat. Hist., Memoir 501–514. No. 6, 274 pp. Errata

The two COI sequences, reported to be from Crangonyx chlebnikovi and Pro- crangonyx primoryensis in Paper 3 were found after the publication to result from contaminations and are not amphipod sequences. Thus their records have been deleted from genbank (accession numbers HQ286026 and HQ286033) and a new tree, omitting these two sequences and with the parameters described in the article, was reconstructed for the COI gene (Fig 7.1). Compared to the tree presented in the article, the corrected tree (Fig 7.1) shows more support for the group formed by Amurocrangonyx arsenjevi and the North American Crangonyx and Synurella species. C. islandicus still appears to have diverged early from this group. Stygobromus species are clustering in a monophyletic group. Synurella ambulans is more closely related to the Stygobromus group than to Synurella sp. as observed in the 18S tree in Paper 3. Nonetheless, this cluster is poorly supported in the corrected COI tree (Fig 7.1). This Errata is not inﬂuencing the main conclusions drawn in Paper 3. Synurella ambulans

Stygobromus stegerorum

Stygobromus gracilipes 2

Stygobromus gracilipes 1

Crangonyx sp.

Crangonyx pseudogracilis

Crangonyx floridanus

Synurella sp.

Amurocrangonyx arsenjevi

Crangonyx islandicus T1 p<0.8 0.8>p>0.95 Crangonyx islandicus N p>0.95 Crangonyx islandicus F

Crymostygius thingvallensis

0.05

Figure 7.1: Bayesian phylogenetic tree of the crangonyctid species, together with Crymostygius thingvallensis based on COI sequences. Posterior probability values are displayed in the box. The tree is rooted by Gammarus abstrusus (not shown on the ﬁgure). Stars and squares correspond to Palearctic and Nearctic species respectively. Branches are drawn to scale, with the bar indicating 0.05 expected changes per site. Crangonyx lineages are highlighted in red.

Paper 4

Phylogenies of Crangonyctoidea species based on the ITS regions (ITS1 and ITS2)

ABCBDEFFBAABFABBBB

EABBFB

ABCDEFABB

DDEDFBBDE

ABCDEFCFAA

BBDBBC D CDBDEBC DCFCACACCDACADAFABCBEDFC

EDFCDDCACACDCACFDACFDFECDCCBADCDDFCAABCDCCDACDBC

FCECDACDBFADCDCDCCACCFCBECDCDCCACCDCFCEDC

BCCDBFADCACDCCCDCCEDCBCABCFCFCDADFACCDBCFC

DCACFACBECACFACFCACAFAEFCACEDADCABCABCABFCBFADCCFC

ACFCACFCADCDCDCCEDCCBACDCCFCCDCDCDABCBCC

DFCDCDADCDCECFCEACBACDFCABCFDACCCEABCFCBFACADFACFCAC

AFCDCFEFECCFDACFDFEC CBBC DBBC DC CDCFCCCFCBEC

DCDCEFCACCBACDCCFCCDADFACCDCFEFECDCFCBFCCFC

CECCFCDBFADCACFEACDDCABCABFCCFCCAECACC

FDDCEBCDFEDFACACCACFCAFDCDCCCFCEFACDADFACCFCCFCDAFAC

CFCDCFEFECADFACCFCFEFACCFCBCACFCDCFCDCDBCACFCFBAC

EFCDCEFCFCFCFFDCDECDFCDAFCCACFCBCCCBACCFCCCDADC

FDCCCABCCDFEDFACDCACCEECCFCBFACFEFAC

ABCDA

BBDBBDCDBDEBDCFCACDACACABCACBEDFC

CDDCACACDCAFCDCDDCCDDCDCAFCCCCDCC

ACDCCBCDCDCCDACAFCDCDDCCBACDCCFCEDC

BCCDCCDABCACCFCECDACDBFADCEFCFCFDACFDFECCFCFC

DACACACFCDCCCBACCDCCFDFCFCBECDCACFACC

D ACCEDADCCDCCABCCFCADCDCDA CD ACDC

DDCFCCCCDFCFCACCFCBDCFDCDBCDCFAB

CAFDCFDACDCBACCDCCDCCFCADCEEAFCCCDCC

EDCBCDCCCECFCCDFAACDBCDCAFACACBCEDCFCDC C

DFEBCCCFEACCCDADFACACDACDCCEFCBCFDCFCDCCABACFCDCC

CDCCDCDCCACCDABCDADFACAFACACECFCABCCCDADFACDCCDCC

CDCFDCFCFCCCFCDABCCACBCDCFCDCCDCDCDCCDFAEDC

AFFCCACACDCAAEFCFCAFABEACBADCECDCFCCFCEFDDCDBFADC

ACCBADCBCACFCAFAADFACCABCADCBADCFDAFCFCAFABEAC

ACDCFCACFACBFACDFAACCFCFECCACDCCACCCDADFACCFC

CBACACEDFAC CBDEBC CFCABCAFCFCFACACACFCDCC

DCABCDADFACACCEFCDFCCDADFACACCCDADFACACCCCEDFACFEFECFDC

CAFCFCFCDCFCCFCDADFACDCAFACEDFACCCBACDCADCAFC

DCFEFECDCCFCDCDCEADCCACFCABCCFCCCEFCFCDCCDADC

CDCECFCDCCFCDCEACCDCDCCCFEFECAFABCCECFCDCC

AFAACACFCCDBCDCEDFCBECDCCCCDCFEFECCFCCBAC

DACFCCDCCECFCFEFCBACBCDBCFCDCCACCCCEACFCDC

CDCAFCDCCCFCACFCFCDEDCDCFCEFCCFCFCCFCDC

CDACCFACFECACFCDACFCBFACDFAACCFCFCDACACDBDEB

DCBBDBBAFCACCFCDACDBFADCDCFCDEDFCFCFACFDFECCFC

BE CD CAFCFCCEEAF AFADCCACFCDADFACACFCFEFECCFCC

BACAFACFCDACDCFCAFCDFCFCBFACFCFEFAC

EFDBFAB

DFACCFCCBACDCFDACCCACAEABCECFACCDCCDCC

FABCFCDAACAFACDBFADCFBADCCBECDBFADCFCBDCDCACC

FCEDACADBADCADBADCCBECAC CDBDEB CDCCFC

AFAFCDFACFABCFCADBCACDCABCCDEFCCCCACFCDC

FBCACAECACFCBFACDDCAFCBBDACABACDCDABCDFCCDCBACCACFDACC DCCDCFCDDAD DA A DAF DFAFEBAFE DCCDABCC DBFAD DA CEAD DBFAD DA CEAD DBFAD DBDEB ABDDDFCD DBFAD DBDEB DDCDCD DBFAD DBADB CFAC DBFAD FABDB FCABAADC DBFAD FABDB ABAADC DBFAD FABB ABAADC FBAAD BBDBB ABDDDFCD ADBAD DBBA DCD FC ADBAD DBDBDBBDB DDACFD FCFCDC ADBAD DBDD ACFDFAECD FC

ABCDEFAD

CDCFDFCCCACDCCEABCCCCADCCCACCCFCC C

DBCACCEDCAFDCFDACDCCBACDCDAACEABCACCCDCCCDC

CCDAADFACCCACCCFDAABCCCCCFCBCCCCDCEC

CCACCEAFCCDCCBDCADCDCBCCFDFCAADFACCFDACAFCFC

CAFACCCCACCCCCCCCCCDDABCFDFECDBCCCFCCC

ABCCFCDCCCCCCCACAFCDCADCBDFACFCDFCCCCCACCEFCCEFC

EFCCFCBCDCEAACEABCFCEACFDFCCAFCDDBCABDFCFCCFCCC

ABCAFCAFBCDCCACADCFFCCCDACEAACEABCFCEAC

DACAFCDDBCCECACFCAFACDCAFCAFDCACDCCEABCCABC

ADF CCACAFCCFCACCCAAAEDCCDCCCC CFAC

DCDCECCCDCAFCEABCAAFCCDC

FBCAFCCFCABEFACACECCCDAADFACDCEAB

ACD ACECCC AACCECFCDC C ECFCDC C ECFCDC CAF ACFE CC AACCECFCDC

EAAEEDCBCD

CEDACCFCCBACCACEABCFCDFDFCECCDDCDCD

DCCDCCAFAABCCBACDFCFCCCCFCCDCFCCCCFCCAFCDCACACDC C

DABFCECAFCECBDCCFDFACDCDDACCFCCDBFCDCABCFCBEAAC

ACCEFCDCCCCCBACDCACEABCFDFACCFCDCCDDADCAC

CDFDDCCFDCFCDCCCCFCDCCABCFCEDFCCDAECDECCC

DCDCAAECACCCFCFFADCEBCCAFABDFCCABCFCDCFEFECCFC

AACCDCCCDFCDCACCCFCDCCCEFCABCFACACFCCDFDDC

DCECFCFDCFCFEFECAFCFCFCCDFCECDCCFCAADCFEFECCEC

AFCACFCCABCEDAFCCCDCFEFECCFCFCACECCAC C

FCFDFCCDABFCCFCCDCCDCAFCDCFEFECCDCECCFECFDACEABC

DDCCFAFCFCDCCAFCDEFCDDFCCECFEFECFDACEACDCDCAEFCCDC

DDCECDCFEFCAFCDACDFCFCDCCCCDABFCCFCCDCC

AFCFCBDCDCACFCDCCDCECDCCAACABCDFACDCDFCDCDABFCDCC

ECDCFACAAAEDCDCFEFECAEFDE CEFCDCCCDADFACACDC

FEFECDCEDACAFCEFDACFCCDC

DADAAFCAACACECCDADCAFCBFADCADFACAFCDCEFACAAFCDCCFFC

C EDA CEAB CDC CAD CD C D C CD C FC CDDB CDDAC F C D C CA C C C

FCCDCCCDCCFFADCEADFACAFACFCCDCCBACDCEFCC

ABCDCDACCECACDCFCDFACEABCFFDFAABCADFAFCCFFCEABC

F C ADFAF C DF C A C DDAD C DF CFFBADEADDDFAF C AF C DEFC

FFABCCFADCCCEFACCDCDFDFCDCCDABCFCFCDAECAACFCAFC

CEACDCDECCDCFCDDBCCCDACDFACAFACCDACCDC

ECFCFFCFCCACFCAFFCFCDFDCDCCCDDCDCECCCBFCDCFCC

FFCCFCCECFDACCFCDBFADCDCADBADCDACCDCFCFCC

FDACCECACFCCDFDDCFDCFCDCCCDCFDDCDDADCCDADFACCFC

BFCCFCCBACDCFCCFFCDCFADFCAFCFCDCDFACAAFCDCCDCF C

DADFACCFCBFCCCDCCDCFCBFCCFACFACADFAFCBA

CDCCFCFCDFDFCCCDCCDCFCCCDABCAFCDDCC

FEFCAFCFCABCAABCFCEABCFCAFDCDFACFCAFCFCCCAFCFCC

DDBCAFCACCCDDCDCFCDCCCECDCDCFEFECDABCAFCCD C

EFCEABCAFCCFCDCCAFCDCCAACEFEFECEFAFEFACCCDEDFCFC

FCCDABCFCDCFEFECADFACFCFCBFACFEFACDCFCDCCCFCEFAC

DADFACCFCCDCFEFCAFCAFCEABCDCEFAFEFACCAACFCFCCEC

ACAFCFCFDCDFAFACAFDCCFCFCCDEDFCACCDCFACDADFACDCFFCAFCDC

DFCFFCDFCCAFCCEFDFACEABCFCDCDDBCACCDCFCDCCEFCCFC C

CDCFCDCDEDFCCEABCCFFDCCDCDCCFCFBACCFCFCCACFCD C

CCCDCCEDCBCFCFCFCFDACCCCDCCECDDCACFACFEC

DFAFACAFDCFCACFCACFCFEACCEFCACCDCFACDFACDCFADFCEABCDC

DFCFFCAFCCEFDFA

CCBACAEABCCCDCCDCDFDFCCCAAAEDCFABCCACACCBDCAFC

FDFACACFCCDFDDCCDCCDECCCCCCDCAAAEDCCFABCDBCCFC

ACFCDFDFCECFCDCABCEFACDFCFCCCCFCCACDACFCAACAFCFCCFFABCFCC

DECDDFCFCCAACFCCCFCDCEECDFDFACCFCCCFCECCADCFCD C

ACCCFCDCCAEABCFCAADFACFCFCCCCFCCDCFCCCCFCC

CFCFADCAADFACDFFCAFCCCEFACCDCACCDCE

CBFCCFCCDCCBACACAFCACACABCDADCDCCDCFCCACFC

DBCDCFCFCFCCFCFDDCDCDFAEDCFCFCEFDDCACABCCCCECC

DBCDDCFDACCBDCACCFDCECFACDBCFDCFCDDBCBFCECDBCFDDC

CBFCDADFACCFCCDCCACDCCAFACACDCCDBCEADFACFCDCC

FFCACFCCBACCDBADBCDCBDEBDCACFCCBACCDBDDABC

CCECCFDCDFCDBCCFCFCACEFACEAFCDCCCCCCFCDF C

CDCBFCCFCDFCDCFCDFCAFCFCFFDCBFCCAFCFCCCCCCC

DDBCECCDFCDCABFCDBCACCCFDCACCCDCDCFACBFCCCCC

CFCFCCFFCDCABFCCFCCBACDCDCABFCABCACCFDCACCDCC

FFABCFCCFFCCAFACDFACDFEBCEFCADCFCFCBFCCFCCDCC

CCCCFFCCFCCCCFCDBFADCDCADBADCACAECDC

DDCFCFCDCCCCFDDCDCCCCEFDDCACECACFCECFDAC

CFCCDFDDCEFACAAFCACABFCACCDCFCACFCCBACDCCCCBAC

FDACFCCFCBFACADFACAF

FBCDADFACACBFCDCEFACECCFCCDCCBACDBCDACACCBFCACBACACDCDACFDFAFACC FCBFCDADFACDCCFFCDCCDEDFCADFABCCFCDADCCCDCBACCEFACDADFACACDCCFC FCDFDFCAEABCDCCACCFCDCFCFCBA

A BF BF BF DA DBDEB DBADB BBDBB D DBDBDBBDB DBDD DBBA FABDB FABB D FDDCADFA AAFCCDADFA

DADFACDCA ECCE DADCAF DFACAF EFACAAF

DCCFCDACADFACAFACCFCFCAFCCDCFCCDDCFCDCCC

CDCCDCFCCAEDCCCFCCCDCFCFCFCAFCCCFCCDCC

BACDCAFCACAFCACFCCDDBCDCAFCBDCAFADCFCDCCAFCCFCDAA C

AFDCFCCCDCBACCDEDFCAFCFCC

EBCDACCCECBFCDBCFDDCCAFCDCCFCAFBDCCFCFDDCACDCFCDCC FCEFDDCACCCECBFCCFCACCDACAECACFACFECDCAADFCCDCDCDBDEBCC BBDBBCDBDD

BADEDCBCD

CDCFEFECCFCCBACACABCCDBCFCACAECACFACFECDFDCFC C

EFABCAFCEFADAFCACDCFCEBCDCFDFCFCCBACECDCFCACFDF C

DADFACDCECDBCACACFCCFEFECDDCABCCAFCECAFAFCFCCDCFEFEC

EDACAFCFCEFDACFCABCCCFDAAB CDBBADBDBDBBDB D C

BDEBFCFDAABCFCFCDACDCFCFCFABCACDCADCFCDC

FEFECCF CBBDBBC D CDD ECACDCDCAFCCFCF C

DDCACADCAFCCFCFCECFCDCBACACDCEBCFCEADFACDCABCBC

FCAFCAFDCCABCCCDFDFACCFCDCFEFECDABCFCFCCFDDCFEFEDCDAC

EBBCDCFEFECCACDBFADCACDCFCADBADCACAFCCDDCDCDBCBA CDBC DBDBBDBC CDC BDEBC CFABC DBC CDBC DDC CBBC DBBC B CD BADBCCDCACACFABCBCDCDFCCECCCDFCDCFFCCFCFEFE

DABEECAABCFCDBCECCACFEFECFDACCFCACCFCACBADBC

DDCFCDCECCACACAFCFCFACCFCBFCDFEBCDAADFACCFCACDCAC

ABCC

EBBEFDACFCCFCDCFEFECCFCCECCABFCCFCACFCFCDCFEFECFCABCFCDC FCBCFCFCCAFDCACFCCBACADFACCFCACDCDCDCC CDC CFABC C DB

CDCCFCFCDFDFCABCCDCCCCABCCFCAADCFBACCCDBCACF C

AFAC CFABBC ACDDCFCDCABCDCCFCBECC CFDB C

BADBCDCDACACFCFCDCCFCFCDFDFCABCCFCFCCACCFC

DACDBFADCBDCDDCFABCEFABCFBFCACBECACFCFFCEFC

A C F C B C D C C F C F C DFDF C FFD C C FD C D C C F C C C FFD C C

DBB DDCDCAFADFCCADBADCDCDBFADCA CBDEB ABCDC

CFCFCACCFCDACDBFADCACDCBADBCDCEFABCFBFCDCDC

CCDFCFCFCACCFCBECFABCFDCFCBDEB

EBCBFACFCCDBFADCDCADBADCACDCCFCFCDFDFCCCDCCFEFCAFCFC ABCAABCFCCFCACFCCDBBACDECCDCCDCFABCFFDCDECDCDCAADFCC FCEFAFEFACCAF

CFBCCFCFCCFCDCCFCEFEFECDADFACABCCDCCCFCE

DADFACDCAFADCDFAFACAFDCCFCFCCABCDFCDFCFFCCCABCFC

DCFEFECADFACFCFCBFACFEFACDCDCFCCFCFFDCDECCCAFCCFC

BECFDAABCDBBCDCFCACCFCDACDBFADCCFCFCACFFCEFCC

FCDCFEFECACACACFCBFACFEFACACDCCDFCFCFCDFAACC

FCFCAFDCDCCECDCAFDCDCCFCECDCFEFECABCDCCAFDC

DCCECCCDCDFEDFACCFCFCAFDFCDFCACDCDCAFDCDC

EBCBFACFCCDBFADCDCADBADCACDCCCECDCFEFECDADFACFEFCAFCAFC ECDCDCFEFECCDABCAFCDCCFCACFCCDBBACDECCDCCDCFABCFC FFDCDECDCCFCECDCFEFECCFCFFDCDECDCCFCECCDCDCACDCDABCFCFCAFDC FDACCAFCDCCEFACDCDCFEFECDADFA

EFEFECDADFACEACFCCDCFDACCDABCACCAFCDAA CAC DB

ACD CDBB AFCFCFCACCCCFCCDCCECDCFCC

AABCFCBECCFBCBADBCDCACDCFCCCCBADBCDC

ADCDCFFCEFCAFCABCFFDCDECACFCBCDCCFCEFACDADFAC

CABC

EBCDFEDFACCFCBFACABDCACCDCCECDCAFDCDCCCEFEFECDADFACECAFDC DCCCFCCECDADFACCAFDCDCCFCCECDADFACECAFDCDCCFCCECDADFA

CFCDCCCABCCCDCDCAFCFBCDCFCFCCDCCFCFC

DFDFCDCCCDBFADCACAECACFCCFCDCAACACFCBECCC C

BADBCBDEBCDCFDBCDCFCFCCADCFBCDBBDD

CFCDCABFCCFCDBFADCACDCCACFCFCFCCABEACFC

FCFCECCDACCFCCECCADFACAFCACFCCDABFCDCCDCFCDFEDFAC C

FCBFACABDCCACCECABC

CFBCCFCFCFDACAFCFCFCDFDFCDCCCACAADCFCFCFBCCFCFC

FDACCFCCDCCBCCACFCDCCCDFAFACAFDCFCFCACCFCFC

FEACDBBCBDEBCBADBCACFDBCDCFBCDCDFCC

CDCCFCDFCFFCCCDCCCDCCFCCCCFBCCFCFC

DCCCFCCDADFACDCABEFCAFCFCBACFDACCFCCDCCDFDFCDFAFAC

AFDCCCCFCDADFCAFCFCCDCCAFCCCCCBACDFCFFC C

CDCCFA

DCDA

C BA C F C A C FA C FE C D C C F C F C DFDF C D C C E C C EF C DAC

BACCFCDBFADCDCCFCCDCCADCBCACFCDC CBB

DBBAABCDCCDACACDCABFCCFCFCDFCDAACDBFADCDCADBADC

CBECCCACCFCDBFADCDACAECACFACFECACFACCBFAC

DFAA CFC F CD ACDFEBCCACCFCADCCDDADCDCABCFC

D BE C D C FA C D C A C A C F C D C C BD C C F C DFDF C E C C F C BFAC

FEFACFCCBECFABCDDCDDFACACFCBACDCCFCCDCFCFC

DFDFCCCCFCBECFABCDCCEFCACFCCDCCFCACFCDCC

CCCDCFDACAACCFCDBFADCDACDCDFAEDCFCBECDCACFECACC

FCFCFE

CDFAFACAFDCDBCFCDCFACCFCCCDCCDFDFCCDCDFC

CFCCDFAFACAFDCCFCDFCAFCFCCDCCFCFCDCACECC

DACCFCABCEFAFEFACDFCCACFCCBACACDCFCDCDFEDFACCFCBFACABDC

DCDCFCADCBACDFCFCDACCCFCCFAECCFCCADFABCFC

BFACADFACFDACACFCDCFEFECECDCFCCECACCFCCABFCDCFCE C

CFCCDADFACCABCFDADCCBA

CBFCCFCCDCCDCECFCCABCDADCDBCFCDACACFEACCBFC

DCCFFCCFCCACDCACFCAFDCDDCACCEFDDCACCFCDCC C

CCDCBCECFDCFCECDDADCACFCCDFDDCCFDDCDCEFDDC

FEBCFCECCEFDDCACDDADCACAAFCDCBCDBFCDCDDCCFCCFDDC

BCADCFCDCCCDCCDACCFCECCFCCBACCBFACACDFAEDC

ECFCFCECCEBCACABFCDCFCECBACDCDCCCAFC

FCDCCCFCFFCEFDFACEBCECDCCFFCDEDFACAFAADFACCC

FACACFCCFCDAAFCCFCCDCCFCCAFCDCEFADCDCFEFECACFCDCCDCFC

FEFEDCDFFCCFCAACCDCCDFCDCDCFCCDCCCFCDCCCACDC

FCDCFEFECCFCCECAABCCFCCECACCFDDCCDCDCFC

ECCDBCEADFACAADFCFDFCFCECECACFACFECDCFCEBCCFC C

BACCFCAAABCCDCCDFCACCFCAACCFCCDCCDFCFCEFAC

DCDCFEFECCABCCCEFACCDCACCFCADCFCDCACCCFCDC

CCDAFACFCABCCFEFECCCFCCDCFCECCEFACAAFC C

CFCCDCDFCFCCCFCCDCCBCDFDCFCCACDACAFCFCCBACDCAADFC

FDFCFCCBACECDCEFAD

CBCCFCCCFCDBFADCDCCFCCBACACDCAADCFCDCDC

CFCDBCADCEEAFCACDDCAFCDCFCFEACCBACDFCFCDACDCECDAC

CDCCEECFEFCCEABCFCCBACBCCFCDCCCDACFCDCCBCFCDC

CCCBACACCEECCBFACFEFACCDCFCFEC CBDEB C DC

ADFACDFCFCAFDCACCACDCCACCDFEBCAFCCECCBFACFEFECDBC

BDEBCEDFACACFCDCCFDCFCCDCCAFADCBCCFCCDCCDC

F C BFA C D C E C C F C DAB C C A C D C A C AFAADFA C C DADFA C C F C B C A C FC

DBFAD C AF C F C AF C EFAD C DF C D C FDF C F C D C D C C E C F C C BFAC

DFAACDFCAFCFDADC

FFFF

DACCCCCCDFCFCFDFAFADCCAFAADFACCDCEFCFC

DCCCCCADCCECDCDFCBFACACCBFCC

DFCCDCCACCECCFDCCCDACACECFEFECAFACC

CDABFCCAAC

CCCCDDCCDDCCDDDCCCAADCDDACCBFCDCCFFCDADFACAC

FCADCAFCAFDCFDACDCCCDACDADCDCAFC

DCCCDEDFCCBACDCCCDCFEFECEACACC

FCCEACCCCCFCADCFCDCACCAAFCBAFACCABCAFCAC

CABCACDFCEACACC

DCCEECCCCDCDDBCAFABCFCEDAFCADBDCCBAFCCFDFCFC

BDCCCCEFACECDABFCAFCABCDEDCDCABCFEBEFCEACACC

EACCDECCCCACDFCDCDEDFCDBAFCFCFADFCDBCBACCDAECAAC

FCAC

DADCCDCCDDACCECCDDCCDACCCDACCCABCC

FEDFCECCEDCCFDACCADCAABCFDFCAFCECCCCAC

DCCCAAFCDCEACABADCECDABFCAFCDCDDACBDCCACC

EACACCCC

DCCFCCCFACCCACACDDDCDFDDCAFCADCCFC

FCAFAADFACCEBACAFDCFDACDCBACCDFCC

ADCCECCDCCCEFACCFCDCFEFECCFCCAFDCFDACDCCC

ACDCFACADCFDCAFC

DCCCADCCDCFEFECCEACACC

ACCCCAFCCDEACCCDBCCCCCCCECBACACFC

EDDCFEFECAAFCDCECACEDCBCCC

EDCCACCACCCECDDACCFCADCCAFDCFDACDCBACAC

CDCACEDCADADCFAADCC

CCFCCCCDDCCEFCCCCCEABCCDCFEFECAC

DEDCDCEFCACFEFACCBFACFCACAFC

CCACCEFCCCCDDCCCCCCCAFDFACDCDCC

DFDFACC

AEDCCCCACFCCFADFABCEFADCDFCCDCEFAFEFACFEBCDDFACFEACC

EFACECCCC

FDCCFCCCCACCEACCDCCECCCCEFCCCCC

DFDDCECDCFEFECCBCEACACCC

ACCCCACCABCCAFCCCEDCFDCDCBFACDAAFAC

CFCBEDFCDACDBDEBCDCBBDBBCACFCDCC

BFCC

ACCCCACCADCACDADFACCFCCBACDCFCAFADCCBCACFC

EFDDCDACDBCADAECCCCCC

ACCCCAFCCDDCCCEDCACCFCEADCCEFDDC

DACFDCEABCCBCEDFCBDACACDCCC

AFDCCDDCCCFBADCDCCDACCEFDDCDFCBDDADCDAC

EFDDCCCEDFACECCDFCAFC

DDCCDDFDCCDCCACCCCCFCCDEDFABCEFADCEFAFEFACDFCCCC

ADCCDCCCDCCDCFDCCACDDACCCACDFDCAADFAC

DFCCCCFFACCADCEFABCDCDCBDACBACDDCDCC

EFCCCCDCCDCCDCAEACCDFDCCCEFADCDDACCFDACADC

CDCCDFCCC

DCECCAFCCCCCCDFDCCDCCDECCCEFADACCFEFEDC

FCDCAFADCCABCCFDCDACDCCCDDCAADCEC

ADCCCCAC

DDACCDECCFACCCCDDCCBFACDCEFACACCDBEDBCAADFAC

ABACCDCCDABCCCCCFACFEFECCFCFACDACDCADC

FAADCDCAFCDFAACFCDEDCACDCEACDCCACEDFCAC

DDCCECCCCFACDCCDDBABCACBFACDDFDBCCDACADFAC

AFACDCDADCDDCCAACDFACFFCFCAC

CCFCDCCCDCDBEDBCDCAFCCFDFAFADCEFABCEDFACCFDFAFADC

EFABCADC

EFCCCCCCEFEFECDACACBFACCFCDEDCCEDC

FDFACCBFCC

ACCCCDDCCCCCECAEDCDDACDCAFABCCCECDCDC

FEFECDABFCEABCCCCFC

FDCCDCCDCCCBDACDCBFACDADFACACABCCCADCC

DCEFAC

FAFCCFCCFCCACCFDCCCEDFCDCAAFCFEFACCCBACC

FEFECCEACAC

DDCCAFCCCDCBDEBCCCDCEFDDCDFCDACEFDDC

ADCDBFADCCABCACDDCACACDCFDDC

DBCCDCCACCDBCCECCCDACCFCDCFEFECCCACFAADCAADFACC

BFACFEFACDCFEFEDCEFACDCAFC

CCEACCDDCCEFCCCCCAFCACAFDCDCBCCE

FEFECDABFCAADFA

DCCBCCAECCECCDCCACCDBCCECCECCADCCCCCCCCBACDCFCEADC

CDCCDFCDCDADCCCC

AFBF

CDFCFCFDCDCAFFACAFCACDCCABCCAABCDEDCDCCDCBDFECFCFC

AAFCCDCDCECDCFCDACDCEACDCCADADCEF

CDED

EBCEADFACFCFFCACFCDBFADCACDCCADBADCACFCFCDCDCCFCFCEC CCDCCCFCACDCBDEBCCBADBCCDD

EBCBFACFCCDBFADCDCADBADCACDCCFCCECFEFCAFCFCABCAABCFC CFCACFCCDBBACDCDCAADFCCFCEFAFEFACCAF

Paper 5

Classiﬁcation of Crangonyx islandicus based on morphological characters and comparison with molecular phylogenies

121

ABBABCDAEFBCDEFFB AABBABEFBBAB EFABBEBFA

ABCADEFAAEFAAA ABCDEFABECFAF

ABCDEFCFAA

BCFACDCBEDFCDACACACFCDCDCDCCCFCDBFADC

DACACACADCFEBEFCFCDFACBACCAECFECDCCDADFACACEDCBC

CFDFCF FCDFACDCBCCACCFCADCCFCCFCAC

AAFCCFCBECECCFACFECCDACDFCAABCFCBADCDAAFACCFCAFC

FCAD CBCC ACDCFCEDADCA CFCAC D CEBBCCECFC

BADCDDFCACCDBCCABCCDCCFCCAFCEDFAC C

FCDCCDCCFCACDCEBBCCEDCBCCFCDBFADCDAC

DCFCBADCDDFCFDFCDCDCACCFCFFCFFADCBCACFCBADC

FDAFCCACEFDDCACFDDCFCCCBADCFCEBBCC

ECDCFCBCCFCADCCBADCAAFCCFDCEBBAC

ABEFCAFCFCDFFCCCEDCDFD

EFDDCBEDFCBCCBCABBDCADFACD AA

BCFCDDCCAFCACDCEFDDCBEDFCDACACFCDC

DD C C AF C C A C A C B C F C F C DF C DA C DBFAD C EFC

ADCACFCAECEBDADFCDFACBACCECCFCACDCADCACFCAFC

CFCFDFACDCACDCDFCCCCBDACEABCFCCBCDDACEFACDEFCAFC

ABACDCADFACCDCBECBCCFDACCACCCACECACFCADC

DBCCABCCEACCDDCCAFDCCACCABCCDCC

CDCCEDCCBCACFCDCCCFDFCFCFCDFACDCBC

ACCFCADCDCECFDFCAFCDCDCACDCADCDCDAFCDCABCFC

DCDCD

DCCFCCFDCDCBBDACDFFCCDADFACACDCACFC

DCCEDFACDBCFCDACBACACDCDCDDFACCAFCFACDCCFC

FCBDBCACDCCABABCCCFEDCDCECFCCAACDCCFCDABC

AECFCABCCDCABCEDFACACFDFCDCCFCFCEDFACCEDFACC

DDFCBACDCBDCCDFCFDCEDFACDCDFCDBCAFDCACFCDCC

ECFDFCFCAFADFACDBCFCEDFACEFCDCECAFACDCDCFECFDFCFCAC

DCEACDFCBDADFACACEDFCBDA

CDACDBFADCDABCFCBE BCC ACDCEACDFCEFDDC

DACDCCCFCACDCBDCACFCDCCBADCDDFCDFCCBC

CDDACBCECCDCCCABFDFACDCCCCDBCEFDDCAC

DCCABADCFCDCFBCCAFADCCCFCDCCCCDFECEC

ECFCBFACDFAACDCCBADCDFDCDFAEDCDBCFCBCCACC

BACDCCCDCCDCFDFCFCBCCACDCFACACFCDCCC

BCCACCEDADCACFCACDCEBBDCCCDFCFCEBBCDC

ECBCDCFDCFCFCFCACCFCBECBCCDBCCABCCFAACFDFCFCFC

ACCFCCCDFCFCFCBDCCDCCBCBECCFCADCBCCFEBCFC A C F C ADF C F C F C A C A C FA C BAD C B C FBDFAB C EBBC D C

FCAACFACBCACAFCACDAFACFCDEDFCFACDFAACFCFCFCDCAC

BADCFDAFCDCEFACFCBFCCDDCEFACAFACFBECBC

CDFAACCFCAFCFCBCCACFACDACCCFCACACDCC

BADCDDFAFACECDCACDABCFDCBACDCACFCADFACCFCDAACCDC

FCBCCACBFACDDACDCCBADCFDAFCDCCDFAEDCAFFCFCACDEFC

AFCFCDFACACFCADCDCACECCFACBADCDDACCFCFCADC

BCC ACCDBCCABCCACDCDCCEFCDDCCBFAC

DFAACECEFCACADFACDEFCFCADFACEFCFCDCAFCFEBCFDFC

AFCDCCDCCDFCFEBCDDFDCABCCDC

CECCFACDDACACDACDCCEBBFCFCCDCDCFCCAFAC

EABCFDAFADCDDFCCACFCDCCCFECCBDACFDFCDCAC

CEDCDFDCDCCADBCEABCFACDDACFCFCDCCCDCACCACAC

CFCCABCDADCAFACDBFADCBCDBCCABCCDDCCAFDCC

ECFCDCCDCDCFCDADFACACDAFDFCCDADFACACCACDCFCFCBFAC

DFAACCAFCADBCDCEACBEDFCDAFDFCAFCAFCCACCCACDC

DCDCCDCAFDFCFACDFCCCACECFCDCCDCEFCACBCACCAC

DCDCCBFAC

CFACFECFCDACDACACFCDACFCBFACDAAFACCFDFAEDAFCFC

ADCACCBCDCBCDCACDACAFCDCBCDCCEDCDFDC

CDAFACCCCFFADCBADCACFCEFDFACFACACCFCADC

CCFCACDCFCBADCFDAFCCCFCBFACDFAA

BAB

FBE

CECFCCBADCDDFCACCDBCCABCCFCDEDFCFCBADC DAAFACCFCCDCFCADCEBBCEDADCAFCFCFCADC

ACCBCCDCCACEFDCDFADCAAEDCCAFCDCDFCDDBC

CEFCACBACCDFACCFFDCACAAAEDCCFFCCEFCDCDC

FCCFCDCFCCFCADCCCEBBCCDDCCEAECAAAEDC

CEBBC DCDCEACCF CDCAFADCBADCFCDEFCAAAEDCABCC

DACCCAFACCFDDCCAFCCDCEBBABCC

DACCCECFCACECABCCECACAFACCFCDDFCCAFFCDCECC

FCA

CCFCDCDACCBFACFCDCCBADCDCEDCDFDCCDDCFC

BADCDFDCCAFCBDCDBCCFCBFFCDFCAAAEDCCAFCDC

DDCDCFCBADCAFACCFCEBBCDCCABCCFCECBCD

ACABCCDBCBCBACACCABCCDCBDCEEAFF

AFCCDDCCCECFCCFCDDFCFDFCACCDBCCABC

CDCCBCCFEFCEABCDAFACDCFACFCCDAFACDDACDC

EF C AF CF CBD C D C CF CA C C DDB CFA C CEAB C AD C DAC

ECFCFDACCFCFCDAAECFCDCEFCAFCFCBDCCACFCAC

DDBCFACCCFACDDACCDACEABCEADCAFDCFCACDCFC

FCDCFEFCEABCFCABCAABCFCAFCACFCCDDBCDDACFCDCCACC

CCFCDCCACCDCAAAEDCDCCACCDCCFCFCC

FEFCEABCFCFCDFDFCDCDCEFCFDAABCCDDFCFDFCDCCDCBC

CEFBECECCDBCCABCCCFACBADCBCDCACCFCBEC

BFFCABCFCFCBCDCCCDCCCDBFADCACFCBE CBFFC AC

FACDCABFCFECAFFCFCBFACDFAACCACDCDCEFBECCFC

DCBADCBACFCAFCBFFCACDCDBCCABCCDCBDCE

EAFFACACFCFCDFACCFCDBFADCDACCCDCACFCDC

BFACFCDCCFCCCBCCDCDCEFCCFCDBFADCACC FDACCECAECFECACFCDCCCCDADCFCCFCCBCDCFEFCEABC

FCDCDFDCDCACACFCDCCDCCDAFADCCEC CEBBCDABCDC

FCDCACACFCDCC

ABFAABBBABEFB

DADFACACACDCDCFCFCBFACFCCFFABCFCDFACFCFCACACDAEC

DCACCDCACDCDCCAFCFCDFAFACAFDCCFCFCFDACCBADCDC

EDCDFDCAFCFCDFCFFCDFCCCFFCDCFCBFACCCACACAFC

E C FF C D C C F C FD C FD C F C F C C F C DAD C A C A C C DAAB C FC

BFACDEFDFACACFCDCCEABCFCDCDDBCDCFCDCCACCACCFDAC

C F C EA C AFA C AB C C DDD C C AB C C AB C C DB C C

ABCCAFCCDDCCDDCCAFCCACCABCC

DADEDCDDDFACDFAFACAFDCCDEDFCEABCCFCFBCCFCFC

ACCDCBFCCDDADCCFCECFCCDFDCCDDADCCDCAAAEDCC

FCACFCDCEAFFCDCCFCFCAECACFCDDA

BADCABCDFACDCCFCECDFDCDCDFDCCFCAFDFECDCEACCFC

BADCBCDCFCCDCCEDCBACEABCFCCDDBCDDACFCDCC

DCCAFCCDCBADCDFDCFCFCEDCFCCDABCFCDADFACCFCFDFCCDC

CFCCDDFCACCDBCCABCCAFCFCFCDACDCACCFCFCCDCAC

C CBC D CBFF C F CAD CD FC FCFCDC CECBCDC

EBBC A C FCA C EBB CBFF DF D CDBCBC BA C C

DADFACDCEDFAACCEABCFCCDFACDEDFCCDCDDFCABAADCCFCCDFACDCFFC

AFCDCACDFCFFCAFCAC

ABE

BAD C FDF C C F C DDF C D C C D CBCCEFCCCDDFCACFFDCC EECCCDCACDDEFCCDCAAAEDCCFDCCAFAACDC

FCFCAAAEDCDDCCFCACCDCCCCACFCFCDDFC

FDFCCAAAEDCCFCFDFCDCEFCEDFAC CF CCAAAEDC C

EBBCDDADCDCDCEACDCCFDFCCFCDDFCCACFCECCACCDCDC

CCFCFC C EBB D F C AFFAB C DC CF C AB CAAAFA C C AEFDFA C C

EBBDCCABCCDACCACABCCDCCACFACCDCCDCFC

FDFCCFACDDFCACCABCCDBCCDCAFCCCCFDCCACCC

FCCDBCCCACFCBACCDCCFCACFDFCACCCCCCE C

CFCABCCFACDDFCCEBB

ABBABEF

AFACCFCFCEDADCBCCACFFCACDCEBBCFCFCDFDFC

ECFCFEFCFCFCCDBCCABCCDCCFCFCAFCEFBECACFCDCDBCADFC

CFCFCABCCDCCCFCDACBECA CBDCC BFC D CCDBC CFACFC

CFCDCBCCFCBFCBECEFCAFCFCEBBCDC

FCBC ACFCDAECDACBCFCAF CBFFC ABC CDCCDCEFCC

DDFCAFFABCDCDDFCCAFFCDADFACAFCFCFCDCCFCFCFCCDDFC

DACBAABCCEFACACCFDCCCFCFCEFCFCDFABCFCBFACDFAACC

FEBBDCFCAAFCFCFCADCBCCCFCDCCFCABC

AABCFCCCDFDFACDCFCFCCDCCDACEFCFCACFCBFAC

DFAACCFCDACBECACCDBCDCABCCCDBCDCFCBABABC

DACCFCACCFCBFCBECFCACEFABCAFCCBCAFACFCBDC

BECDCFCAFACACFCFCCFCEBECACABCCDCCACEFDCDFADCCF C

FCFACDCFCFCBFACDFAACDCCCDCEFEDFCFCFCDC

FCCEFBECCDCDBCCABCCFCDCFCCFDFCFCBDCBEC

ABCDCCFCFCFCDACBE CBFC D CCDB C F C F C AF CBDCE

EAFF CF CBFFC EFBECECCDBCCABCCAFACFCBE CBC DC

FC F C BE CBCC D C FA C AB C CBDC BE C FCA CB CFEEE D CBBE DDCACFACDCFCDCABCDCCFCFC

BCCA

BCDEFAC BFACDFAACF CBCC ACDCCBCCFCACDCCFCDDFCACCDBCDC ABCCDCFEFCCDAECDACCFCACFCAFCFCBFFCEFBECDCACCDBCDCABCCC BADCDDFCFDFCDCADCDCABCAFCACCCDCACCBCACCCCACCDFDCFCDDADCCA C BECACCDBCCABCCDCADCCFCABFCACACCADFCACFACFEACDCDCAFC

BACECAFCFCDAECDACFCBDCCACEFBECDCECDCDC

FCEFCFAFCEBBCDCFCBCDCDCAFCCAFDFCDFCDCB

FABCCDCCCBECCCFCEBBCDCFCBCFCDC

DCFDFCACDDFCCACAFABEACFCCDCFCACCFCBFCBECCDCCAC

EFDCDFADCCFCBACDCCABCAAB CFC CCBADC

DAAFACAFCEBBDCCBFDCAFCDCFCAC BCDEA BFACDFAACF CBCC ACDCCBCCFCACDCCFCDDFCACCDBCDC ABCCFEFCCDAECDACDCFCAFCBDCEEAFFCDCEFBECCBADCDDFCFDFCDC ADCDCABCAFCACCCDCACCBCACCCCACCDFDCFCDDADCCACBECACCDBC C ABCCDCADCCFCABFCACACCADFCACFACFEACDCDCAFC

CDCBA CFCA DDCCDFCAFCF CCB CFE DCC

AFDFCFCF CBDCCBECABCCDCCACACACDDCFCFCDFACDCCDBCC

ABCCCACFCADCAADCFCFCCBCDCBDCCBECCACC

FCBDCBECCFECCBCDCFCACFCDCDCACFDFCACFCDDFCC

ACFABCDCECCFADEDCCFCACCCFCDCEACFCCDBCCABCC

FACFDFCEBBCCDCCDACBADCFCCBCDCFCBDCCBEC

ACACFCEFCCECBCACBECEBBCAFCF BCDEACDADCBFACFCCDBFACACFBFCAFCBDCEEAFFCDCCCECFACDAAFC DECDCADCACFCCCFCACFCCFBCFCDBCFCEFBECACFCCDCDCDCFCDCAFCFCDC AADFABCCFCDBCCAF

EFABBBEBFA

CBCDCCBADCDDFCDCFCAF CBDCEEAFFCACCBEFC

AFCFCBACDCCFCEDCECFDCFCFCFCAF CBFFC ABCCCDCC

ABCFCFCBA CBC AC FACDCF CBFFC ACDCCCDFCFC

BCCACFDCCBCADCFCDCFCBCFCBDCCDCBFCBE FCACCCDFCFCFCFCADCACFDCFCFCFCEDADCCACFC

EDCBACABCCDCABCCCACEFDCDFADCDCCB CF

DDCCDFCFCEBBCACACFCEFCCFCEDCDFDCABCEBBCFC

FCBCDCCCFCAECACACFCDCCACFCDCDCDBCADFCCFCFCDC

FACACEFCAFCFCADCCCFCBACDCCBCADCFACBEABC

ABCCCDCCFCADEBBCDCDCCCDFCFCDCFCFDCFCF

CFCFCBCCACCFCADCACFCCBDCCEDCDFDCFCFCAC

DCEFABCAFCACCFCBECEBBCDCECBCDABCCDCABCCDCCACEFDC

DFAD

ABFAABBBABEFB

CDCBFCCFCBCACCDCDCDACABCDFCAFCFCDFAFACAFDC

CFCFCDACFCDCCBCDABCFCFCDFCFFCDCCDCBFA C

CAFCDCECFFCDCCFCFBDCCCACFCDCCDCCDCABC FC

DCDCDCFFCCABCABAADFCABCFDFCABAADFCDBCACACCF CBCC DCC

FDCDACFCCCDFCFCDCCFCFCCF CCBC DCF CBDCC BECDCFACC

DBCCABCCDCFCBECAFCFCDBFCACDCCACCFCCDBCBECC

ACAFADFCFCFCBDCCDCBFCBECDCCDCABC

A A A A

FA ADFACFCBFACDFAACDCDCBFCDB CBCC DCDBFADCACBACDCDCC BADCCFCDBCACDCEDCDFDCCDCCCDBFADCACCDCDFACAAFCCCDEC CDCCFCFDCFFCCFCFCCACCACFCDC BCDEAADFACFCBFACDFAACDCCBCDCDCACDBCBCCACCFCACFCDCDCAC ABECCCACFCFCDCACDCCFCFCDCACACDCACAAF

CDCDADFACDCCFCBFACDFAACDBCDBFADCACDC

CEDCDFDCDCFCDAECDCACCFCACCCFFCCFCCABAADFCCFCDFC

FFCDABCFCCBACAFCDCDCDCACDCCACECDCCECFCFCAAFC

FDCDABCDDADCCFCBACCDCCBFACADFACFCACFCDCACDFCFC

DAC A A AB CBDE BCAFE ACC

ACACDADFACDBCBCCACBECDFCDBCDCABCCACFCDCDDBCDAECBFCACAAFC CECCACCFDDCCCFCDC CCFCBADCFDAFCAFCFCFCDACDCCFCBACDCCCDC

CCDCCDCAFFCDFFCCACEFDCABCCDCCCCFCCDFACC

ABAADFCDADFACFCBADCDDFCFDFCDCFCEFCDCCEDCDFDCACDCC

DFCECFCFCDCECCFDDCDDADCADCCFCCBCEFABCACCFDFAFADCC

CBDAFCCFCCDFACCDCDDFCDCBDCCCFDCCCFCDCDCCFCC

FCDCCFDCCDCCFCCFCCFCCDFACCFCCBCACCDDFCCDCC

DCCCFCCDCCDECCFDCCFCACFFCFDFCFCCFCFCABFCCDFACCF C

DDFCDCCFCBFCDFACFCDECDCECCFCDFCCDCCEFABCFCFA C

FDFCACACDCBFADCCFDAF

CDFACABCBFACAFDCCFCCDCCBCFCFCADCCBCC

ACDCFCDCEBBCCFCFCDCFCFACFACCBADCAC

DCCCDCDCDCCBFCCDDCEFACACBADCFDAFCCFDFCCDDFCCFC

ECCACCFCDDCDBACCBDFCCDCCFCECCFADEDCCCCDCAFADC

CFCACDCCBEFAFCCFCCCCECBCDDCEBBACDC

EBB

CAFBDFACCFCFCEDADCACACFCBCCDBCCABCCACFCDFCFCDAC

BFACDFAACDCFCACBECACCFCDEFCCCDBCCFDFCB

DFCFBFBEEFCAFCFCACDCCDDFCBECCDBCCABC

C FA C DA C F C DDF C BA C C EFA C C F C F C BFA C DFAA C C

BADCFDAFCDDCDCCFCDCCBFACEFACDFCFCDACCCDCAFCFC ED C BA C C A C C DB C C ABCCCFCBCCFCFCADC

BCCFCFDAFCDCABCDADCDCDCCCADFACCACAFACBDCFDCDBC

EFCFCDDFCDCFCCEAFDCFCDEDFCFCFFADCBECFDFECCFCDCCFC

AFACCFCBCCBECAABCFCFACBFACAFADFACFCFCAC

EBBC DCFCF CBCC A CFCAC D CEBBC DCEFCABCAC

ACEEAC

CABCCFCDDFCFDFCECACFACFECCEDCFCAFABEACFCAAAEDCC

FCEFCEDFACCFCFDFCEDFACFCBECCACACFCDCCACC

ECFCDCABCABFCFCFCCDCACACAFCFCFCECCEFDFACD C

CCFCCDCCBACACDCCEAFFCDCFCDFACCDCCED C

BCACFCDCCFCFCFCEDFACCACCBCCCFCADCFC

FCDCFDFCCFCDDFCACCDBCDCABCCBCCEFCDCCFCEFCDC

AFAFCACDABCFCFCBADCAFADCEFCDCCFEBCBADCDDACC

FCBFADCFCABFCEDFACCFACCACCFCECDEFCAFCFFADCAC

CFDFEC

CBCAFCFCDCDCCFCFCBCEBBFCDC

BCFDCFCBDCFCDCAAFCAFAEFACACABAADCDCFCDADCDC

BCDCDCACDBCACFDCDCDFCFCADCDBCCABCCEBBCAC

CCAFDACFCDCFDCECCDFCCECDCDCDCDCADAFCFCBBDADC

BACCAFDACDCDCDDFCDFCCFCADCCAFCDCDCECABCFCDEC

CFCFCADCDCEDADCFFACDFCAFCFCDCDCFDCDCDFCFCD

DCDCFCFCCFCADCACACFEBCABEFCAFCFCEDCBAC C

FC DCFCFCAD CBCC ACFCDCDCABCCFCFCFCEDADC

ACEBBCDCFCAACDCCCDFCFCFCBECEBBCDCECBCDC

ACABECFCFCFCDCCEDCDCBADCADFACABFCFEBCCECFCFC

AAFCEFCCBCCACAECACFCEDCBCFCFCDEDCCBC

ACBADC FDAFCEFCCACEFDDCACCCEFCFCFCFCFDF C

F ACCDFCFCFCFCADCFCFDCFCFCCDADCECECADACDC CDCCAABCFCBADCDCEDCAFDCFCFCACDCCBC

FCAAFAEFCACEDADCACCCDFCBADCFCCBDCFCBDCBEC

ACDCFCFACAFAEFACACFCADCCBDCBECACFCFCACAFAEFC

BEC CBCC ACFCADCDCFCCBECFABCDFACACCFCDAACFFC

DBCCABCCCECCFCACACEDADCECCDACCFCACAFAEFACC

DCFFADCDFCCFCACBCDCFCBDCCBECACFCADCABADCDC

EDADCFCFCEDCDDACAFCDFCDCDCABCCFCACCFCFCADC

BCCACDCAFCACCCDFCFCFCEBBCDCECBCDCBDCACFCDCC

FFABCDBCCABCCFACFDF CBBE ACACCDFCBADCFCFC

BDCC BE C D C F C FCA C F C C BAD C DFE C C F CECBCDC DC

EBB

CDCCDADFACFCFCFDFCCFCDDFCACCDBCDCABCC

CCAFCACCFCDBFADCDCFCDCCACFCEDCBACAADFCDC

ABCBCCDCCBFCEFACDBCFCAFDFCBDCCBADCDDFCABC

ACCFCDAADFAC CBCC ACCFCADCDCCFFCFCCEFADC

DFAACAFACFDCDBCBDCCDFACFCDCACDCBFACDFAACDBC

BCCACFCFFADCBFACADFACACBFCDFACCBDFCBFCDCCFCBFC

DADFACFCECBCDCDCEBBCBECCECCDCCCDCCCCDCDC

ABCCAADFCFDFCFACDDACECCECFCACFCBFACDFAACCFC

DBFA

BADCDCEDCDFDCDCFCEFABCDCAB C CF C FC FC

BCC ACEFABCFCDAAFCCFCA CDCCEDCECCFCCBCAFC

ABCCFCACACAADCDCECFCAFCBDCFEBCFCACDDFACC

EAECBADCDFECAFCEFCAFCECFCACACFCDBCBCDCCBC

ABCFCFCDFAACAFCFCBFBECCFCDFDCFCFCAFCBFACDFAAC

FCEDADCDCFCADCACCFCBE CBC CBCACFCBADC

DDF C C C DB C EFDD C A C C F C C F C F C A C DDFA C C FAC

ADCCCACEDCDDACCFCCCFCDBFADCACCFCCAFC BFACDFAACEFCCDCFACCFCBCCACDCCFEACCDCCDFAEDCAFFC

CDA CAF CFC D CFCBADCAADCF CAD CA CF

FCB D B F C F C BAD C D C F C ED C E C C AF C B C C

DADAAFCACCFCBFACDFAACAFCDFCFCAFCFDACCEFCDC

AFCDCDCFCAFCEFADCAFACDABCDADFACACFCBFACDCDCC

CFCBACEFCFCFCCBFACDFAA

DAEA

CDFCFCFDCCACDCAFFACDCAFCACCAABCDEDCDCDCDCDDDC

CACCEABCFCABCCDFAEDCDDFCCDCBDFECFCFCAAFCCDCDCECDC

FCDACDCEACDCCADADCEF

DADEDCCCBFFDFCCCCDCCDFACDACACCAFEDADCDFC

ECCFCECC

ECCABCCAFDCCADCCCBADCACDBCCDAC

ACFCBECECBCDCABCACAFCEFDDCDAFDFCCBFCBECFCCCC

DCCCEECCCCDCDDBCAFABCFCEDAFCADBDCCBAF CCBFC

EEEFCEBCC

FACCCCBCCDDBCACCAFAEFCCFCDEFCDFFCC

CACAAFCCDABFCDFF

FCCACCDFFDBEFDCCCEFCAAFCCBDACACFCDEFFAC

DCFCCDDACFDCFDCACFECBCFCC

ABCC CEDEECEBECCEEBBDCECE

DDBCCDFADCEECEFACCC ABCCCBEBDCBEDDBCCBEDBCAFDC

FFACBCDABFCC

ABCC CEDECEEBBDCECBCDBCEBE

CEEBEEEAFADCFAEFACFCBCCC

ABCCCAFACCFCFBAFCACEFDDCECBCDCFFCDBFADC

CFCAFCACAFCFCCDACDCBBDACDFAA CCBFCBE

CFCCC

ABC C C DA CC C C FD C D C FDFA C FDFE C CBCC C BC

DBFADCBECECC

CCECCDCCCCDDCBACCDCBDACFCEFCCAADCFAC

ECACFECCCC

DDDCCCFAEFACFCFCBCCFCADCECBFFCCACEBDADCAFC

D C C AF C D C C C A C FA C C AABD C D C AFAEFA C D C DDADC

CFCBABDBEACCC

DCCCABCCCFDFACCFCFCADCEFDDCDACBECEBB

DBFADCCBECC

A C C C C AF C C C DD C C C ED C A C C F C EAD C C

EFDDCDACFDCEABCCBCEDFCBDACACDCCFFBCFCCC

A C C C C A C C AB C C C AF C C C ED C FD C DC

BFA C DAAFA C C F C BEDF C DA CBCC FC D CBDCEC EAFF

ACFCDCCFFBFCEACFECCC

AFDCCCDDCCCFBADCDCCDACCEFDDCDFCBDDADC

DACEFDDCCCEDFACECCBFCEBFECBCC DFCCCCFFACCADCEFABCDCDCBDACBACDDCBBCC

DCCCDFACDDCDDBFCACFCDCACFADCBDAC

FCECBCEFBCC

EACCCCAFCADCDBFADCBCCBCCCEFDDCADCACFC

BEDFCCFCCFDFACDCABBDADCAADFACEBBCFCCC

ACCEFCCCCCFDCFDCFCCFCDADCCDCAAFCFDAFC

DCDCBFACFCCDEBCC

DDA C C DE C C C FA C C C C DD C C BFA C D C EFA C A C C DBEDBC

CCBDECC

CFCCDCCCCDBEDBCDCAFCCFDFAFADCEFABCEDFACC

FDFAFADCEFABCAD

ACCFCCEACCCFACFACDDACAAADFCDFACDC

DFACACCDADCACFCFCADCCBFCEBCFCFCECCEE

CDCC

ACCCABCCCDBCBCCDCCBECCEFDDCDACDBFADC

C F C EAD C D C DF CAF C DC AFA C C F C C CBCC BAC DCFCC

ABBDACDFAACCBFCBECFCCC

DDCCCAFCC CBCC FC CCDCEFDDCDFCDAC

EFDDCADCDBFADCCABCACDDCACACDCCCECC

ACCAFFCCDACCDECCDACCCFCCCDCAAFCC

DACADCEFDDCACDFCBCCFCCC

DBCCCABCC CEDECEBEBDCBCBCE CBEDBCABAADCEECCDFEDCAFCACCC EA

DC DDFCDCDDFCFDFCECACFACFECDCACCDBCCABC F CAD CAFCCFDCCCDCCCFD CAFCCCDCCCFDC CFAECEC CC C CAEECC CC C CCACCDAD C C FACDC CFC CFADCF CA FACCDDCDC CF CDF FACDCFC CAFCCFCFF CAFCDCCCFFCCCFCDCFF FACDE C C FAEECC CCFDCCFD CFADCCFDC F CECCDF CF CDF FF ACDC CF CFADCF CA F ACDE C C F ACCDDCCCC C CDF CF F ACDCFC CAFCCFCFF CAFCDCCCFFCCCFCDCFF F CECCD CF CDF F AEECC CCFDCCFD CEEDCCFDC F ADCCCCC CC CDF CF F AEEED C C F FADCDCE CF CDF FAECE CC CC F AEC CC CC FACCDDCC CF CDF

ACCDDDCCCCC CCE C C ACDC CF CFCA CCACDCEF EEECC CDF CF FACCDDCFCCC CEC CDF CF EBCDEA

BCDEFAC BFACDFAACF CBCC ACDCCBCCFCACDCCFCDDFCACCDBCDC ABCCFEFCCABCAABCDCFCAFCBFFCDCEFBE BCDEAC BFACDFAACF CBCC ACDCCBCCFCACDCCFCDDFCACCDBCDC ABCCFEFCCABCAABCDCFCAFCBDCEEAFFCDCEFBE BCDEAC DACCEDCBCDCCFCCDCBADCDDFCCDBFADCACCBADC DDFCFDFCDCACCDBCDCABCCDCADCDCABCAFCACCCDCACCBCACCCCACC DFDCFCDDAD BCDEAC DACFCEDCBCDCCFCCDCBADCDDFCCDBFADCACAACC ACFCDCCCBADCDDFCFDFCDCACCDBCCABCCDCADCDCABCAFCACCCD C ACCBCACCCCACCDFDCFCDDAD