phylogenetic analysis of freshwater sponges provide evidence for

Available online at www.sciencedirect.com

www.elsevier.com/locate/ympev

Molecular Phylogenetics and Evolution 45 (2007) 875–886

Phylogenetic analysis of freshwater sponges provide evidencefor endemism and radiation in ancient lakes

Martin J. Meixner a, Carsten Luter b, Carsten Eckert b, Valeria Itskovich c, Dorte Janussen d,Thomas von Rintelen b, Alexandra V. Bohne a, Johannes M. Meixner a, Wolfgang R. Hess e,*

a SMB, R.- Breitscheidstr. 70, D-15562 Rudersdorf, Germanyb Museum fur Naturkunde der Humboldt-Universitat zu Berlin, Invalidenstr. 43, D-10115 Berlin, Germany

c Limnological Institute of the Siberian Branch of Russian Academy of Sciences, Laboratory of Analytical/Bioorganic Chemistry, Ulan-Batorskaya 3,

RUS-664033 Irkutsk, Russiad Forschungsinstitut und Naturmuseum Senckenberg, Senckenberganlage 25, D-60325 Frankfurt am Main, Germany

e University of Freiburg, Institute of Biology II, Faculty of Biology, Experimental Bioinformatics, Schanzlestr. 1, D-79104 Freiburg, Germany

Received 7 December 2006; revised 15 August 2007; accepted 6 September 2007Available online 18 September 2007

Abstract

Morphologic and phylogenetic analysis of freshwater sponges endemic to lakes in Central Sulawesi, Siberia and South-East Europe ispresented. We also analyzed several cosmopolitan sponge species from Eurasia and North America and included sponge sequences frompublic databases. In agreement with previous reports [Addis, J.S., Peterson, K.J., 2005. Phylogenetic relationships of freshwater sponges(Porifera, Spongillina) inferred from analyses of 18S rDNA, COI mtDNA, and ITS2 rDNA sequences. Zool. Scr. 34, 549–557], the meta-niid sponge Corvomeyenia sp. was the most deeply branching species within a monophyletic lineage of the suborder Spongillina. Pachy-

dictyum globosum (Malawispongiidae) and Nudospongilla vasta (Spongillidae), two morphologically quite distinct species from Sulawesiwere found in a joint clade with Trochospongilla (Spongillidae) rendering Trochospongilla paraphyletic. Furthermore, Ochridaspongia sp.,another Malawispongiidae, clustered far away from that clade, together with Ephydatia fluviatilis, making the latter family polyphyletic.The Lubomirskiidae endemic to Lake Baikal, Lubomirskia abietina, Baikalospongia bacillifera, B. intermedia, and Swartschewskia papyr-

acea formed a well-supported clade that was most closely linked to the genus Ephydatia (99.9% identity over a total length of 2169 con-catenated nucleotide positions). Our study indicates the frequent and independent origin of sponge species endemic to differentfreshwater ecosystems from a few cosmopolitan founder species. The highly specific primer sets newly developed here facilitate workon the molecular phylogeny and DNA barcoding of sponges.� 2007 Elsevier Inc. All rights reserved.

Keywords: Demospongiae; Lake Baikal; Mitochondrial cytochrome oxidase subunit I; Molecular taxonomy; Porifera; Sulawesi lakes; Translationalexceptions

1. Introduction

The majority of extant Porifera are restricted to the mar-ine environment, but a few taxa within the Demospongiaelive in freshwater habitats. All freshwater sponges werecombined into a new haplosclerid suborder Spongillina,comprising six extant families, Spongillidae, Lubomirskii-dae, Malawispongiidae, Metaniidae, Metschnikowiidae

1055-7903/$ - see front matter � 2007 Elsevier Inc. All rights reserved.

doi:10.1016/j.ympev.2007.09.007

* Corresponding author. Fax: +49 761 203 6996.E-mail address: [email protected] (W.R. Hess).

and Potamolepidae (Manconi and Pronzato, 2002). TheSpongillidae show a worldwide distribution, whereas theother five families are endemic or are geographicallyrestricted. Sequence analyses of 18S and 28S rDNA, themitochondrial cytochrome oxidase I gene (COI), andITS2 rDNA have been proven useful to infer their phyloge-netic relationships (Addis and Peterson, 2005; Borchielliniet al., 2004; Nichols, 2005). Based on such data, the mono-phyletic origin of freshwater sponges has been suggested(Addis and Peterson, 2005), which is not in contrast to veryrecent work suggesting reassessment of the order

mailto:[email protected]

876 M.J. Meixner et al. / Molecular Phylogenetics and Evolution 45 (2007) 875–886

Haplosclerida (Redmond et al., 2007). In fact, several anal-yses have pointed out that freshwater sponges are not trueHaplosclerida: According to Borchiellini et al. (2004), theSpongillina form a monophylum outside the marine Haplo-sclerida (represented by Halichondria mediterranea),whereas in Nichols (2005) the Haplosclerina are a well-suported clade with exclusion of the Spongillina. Moremolecular analyses such as those introduced by Borchielliniet al. (2004) and Nichols (2005) would be helpful to illumi-nate the phylogenetic relationships between freshwater andmarine haplosclerids and to other demosponges in moredetail.

The origin of Spongillidae has been dated to a timebetween 183 and 141 MYa by using both minimum evolu-tion (ME) and maximum likelihood (ML) molecular clocks(Peterson and Butterfield, 2005). The phylogenetic relation-ships between and within the freshwater sponge familiesare not fully resolved yet; however, it has been suggestedthat the genus Ephydatia is paraphyletic and that speciesbelonging to the families Lubomirskiidae, Metschnikowii-dae and Malawispongiidae may have arisen from speciesof that genus (Addis and Peterson, 2005).

To further elucidate the relationships among freshwaterporifera, a DNA-based phylogenetic analysis of speciesendemic to geographically and geologically distinct fresh-water lakes could be informative. For this purpose,sponges were sampled in old tectonic lakes on Sulawesi(Indonesia), in lakes Baikal and Dzhegetai-Kul (SE Sibe-ria) as well as in Lake Ohrid (Macedonia). This collectionwas complemented by common freshwater sponges fromlakes and rivers in Brandenburg and Mecklenburg (NEGermany) and from river Rhine (W Germany) and byfreshwater sponge-derived sequences from publicdatabases.

Lake Poso and Lake Matano in Central Sulawesi areoligotrophic and in many geological aspects very similar.Based on their endemic faunal elements, Sarasin and Sara-sin (1905) assumed a Late Miocene to Pliocene age; how-ever, according to new fission track data, Late Tertiaryto Early Pleistocene (�2 MYa) appears more likely (Bellieret al., 2006). The epilimnion temperature ranges from 27 to31 �C in both lakes and varies only slightly from 27 �C inthe hypolimnion (Haffner et al., 2001). The largely endemicsponge fauna of these lakes was neglected for almost a cen-tury. The most recent report from 1901 described three spe-cies (Weltner, 1901), all without gemmules, the typicalasexual reproducing stages for spongillids.

Lake Baikal (Siberia) is the largest and oldest lake in theworld since it can be dated back to the early Miocene(Martinson, 1936, 1938). Due to its age and special geolog-ical situation without large inlets and only one outlet, aunique fauna and flora developed in Lake Baikal over sev-eral geological epochs. It possesses an endemic spongefa-una, together with several ubiquitous species of thefamily Spongillidae. A typical underwater photo fromLake Baikal is shown in Fig. 1a with ‘‘sponge forests’’ con-sisting mainly of branching Lubomirskia baicalensis and

other incrusting Lubomirskiidae as the dominant megafa-unal element. With a biomass of more than 1 kg/m2 insome littoral areas (Kozhow, 1963), sponges play a prom-inent ecological role in the shallow water habitats (6–20 mdepths) of Lake Baikal. But sponges are found also in thedeeper zones of Baikal and have been collected from largedepths, as far as down to 889 m (Rezvoi, 1936). Tempera-tures of Lake Baikal are low throughout the year: The sur-face water varies between �1 and about +10 �C for Winterand Summer, respectively. Below ca. 30 m water depth, thetemperature is constant at 4 �C. Lake Dzhegetai-Kul islocated about 750 km SW from Lake Baikal on the footof Tannu-Ola Mountains at an elevation of about1000 m. The oligotrophic lake has a maximum depth of17 m and is probably of Holocene age, about 10,000 yearsago. The lake is habitat of the non-gemmulating spongeBaikalospongia dzhegatajensis Rezvoi (1936). The first men-tion of sponges from Lake Baikal was by the traveler P.S.Pallas (1771), who described the most prominent littoralspecies under the name Spongia baicalensis (now Lubomirs-

kia baicalensis). Dybowski (1880) wrote the first mono-graph on all baikalian sponges known in the 19thcentury, which he united into the genus Lubomirskia. Inthe early 20th century two further genera were described,Baikalospongia Annandale, 1914 and Swartschewskia Mak-ushok, 1927. A revised system of the sponges from LakeBaikal was published by Rezvoi (1936), who also describedBaikalospongia dzhegatajensis, which he attributed to theLubomirskiidae. Some authors assigned this species tothe Spongillidae (Greze and Greze, 1958; Kozhov, 1972),however, in Muller et al. (2006) it was positioned againwithin the Lubomirskiidae. B. dzhegatajensis shows mor-phological affinities to Baikalospongia in spicule shapeand skeletal architecture; however, until this work nomolecular data existed for this species and its phylogeneticaffiliation remained unresolved. In a thorough revision ofthe baikalian sponges, Efremova (2001, 2004) described anew genus, Rezinkovia, and several new species of theLubomirskiidae. According to Efremova (2001), the fol-lowing valid genera (and species) are currently known fromLake Baikal: Lubomirskiidae: Lubomirskia (4 sp.), Baikalo-

spongia (4 sp., 1 ssp.), Swartschewskia (2 sp.) and Rezinko-

via (2 sp.); Spongillidae: Ephydatia (2 sp.), Spongilla (1 sp.),Eunapius (1 sp.) and Trochospongilla (1 species).

The oligotrophic Lake Ohrid being 286 m is the deepestlake in SE Europe. It is situated on the Macedonian–Alba-nian border and covers an area of 340 km. The hypolim-nion temperature varies only slightly from 6 �C and thesurface layer never cools below 3.94 �C (Outcalt and Allen,1981). Because of the absence of geophysical and sedimen-tological studies the absolute geological age is stillunknown. Spirkovski et al. (2001) speculated that the lakewas formed between 4 and 10 MYa ago, but earlier dataindicated an age between 2 and 3 MYa (Stankovic, 1960).Lake Ohrid represents a refuge for numerous endemicfreshwater organisms, whose close relatives can be foundas fossil remains from the Tertiary period (Stankovic,

Fig. 1. (a) Underwater photo from Lake Baikal, about 6 m depth. (b) Gemmules of the Spongillidae, Spongilla lacustris with gemmules, scale 100 lm. (c)Pachydictyum globosum from Lake Poso, Sulawesi, upper part of spicule bundle, scale 40 lm. (d) Pachydictyum globosum, n. ssp., Skeletal architecture,scale 200 lm. (e) Pachydictyum globosum, n. ssp., detail of skeleton, scale 40 lm. (f) Baikalospongia bacillifera, Lubomirskiidae, from Lake Baikal,spicules, scale 50 lm. (g) Baikalospongia intermedia, Lubomirskiidae, spicules, scale 50 lm. (h) Baikalospongia fungiformis, Lubomirskiidae, scale 50 lm.(i) Lubomirskia baicalensis, Lubomirskiidae, spicules, scale 50 lm. (j) Rezinkovia echinata, Lubomirskiidae, scale 30 lm, (k) Swartschewskia papyracea,Lubomirskiidae, scale 30 lm. (l) Heterotula multidentata, Spongillidae, spicules, scale 30 lm.

M.J. Meixner et al. / Molecular Phylogenetics and Evolution 45 (2007) 875–886 877

1960). Five different sponge species were described fromthis lake (Arndt, 1938; Gilbert and Hadzisce, 1984; Had-zisze, 1953) and one, Ochridaspongia sp., is included here.Based on molecular sequence data from nuclear and mito-chondrial gene loci, we tested if sponges endemic to thesedifferent freshwater ecosystems would share a common orpartially common evolutionary history or originated inde-pendently from each other from a few cosmopolitan foun-der species.

2. Materials and methods

2.1. Investigation areas, sampling, and taxonomic

characterization

An overview on all species analyzed in this study is givenin Table 1. Sponges from Lake Poso and Lake Matanowere collected by SCUBA diving during binational Ger-man–Indonesian expeditions in 2003 and 2004. The hereinvestigated species from Lake Baikal were collected bySCUBA diving during expeditions in 1997–1999, within a

Russian-Japan joint research program to Lake Baikal. Aspecimen of Baikalospongia dzhegatajensis from LakeDzhegetai-Kul was kindly provided by Dr. Igor Ivanov(Academy of Sciences, Kysyl, Russia). A set of commonSpongillidae was collected in the lakes Schmaler Luzin(Germany, Mecklenburg, 2003), Flakensee/Kalksee (Ger-many, Berlin/Brandenburg area, 2004) and in the LocknitzRiver (Germany, Berlin/Brandenburg area, 2003). Trocho-

spongilla horrida (Weltner, 1901) was collected on the leftembankment of Rhine River near Neuburg in November2005 and kindly provided by Dr. Jochen Gugel (Universityof Stuttgart, Germany). Ochridaspongia sp. was collected inLake Ohrid, and kindly provided by Dr. Christian Albr-echt (University of Giessen, Germany). As outgroups tofreshwater sponges, sequences were obtained from the mar-ine sponges Pseudaxinella reticulata, Microciona prolifera

and Scopalina ruetzleri (collected and kindly provided byDrs. Klaus Rutzler and Wolfgang Sterrer).

All sponges were inspected for possible contaminationby invertebrate commensals. Part of each sample was fixedimmediately in 70% ethanol for taxonomic identification

Table 1Species used in this work following the classification of Systems Porifera (Hooper and van Soest, 2002)

Species/sample Lineage Origin/reference COI GenBankAccession No.

18 rRNA GenBankAccession No.

VoucherNo.

Baikalospongia bacillifera(Dybowski, 1880)

Haplosclerida;Lubomirskiidae

Russia, Lake Baikal, W shore, Bolshie Koty, 14–15 m, July 2000 DQ167167 DQ176775 ZMB Por12657

Baikalospongia dzhegatajensis(Rezvoi, 1936)


Russia, Tuvinian Autonom Republic, Lake Chagytai, N shore; leg.:Igor Ivanov, 1999

EF025856 n.d. ZMB Por12690

Baikalospongia intermedia(Dybowski, 1880)


Russia, Lake Baikal, W shore, Bolshie Koty, 14–15 m, July 2000 DQ167168 DQ167154 ZMB Por12653

Corvomeyenia sp. Addis03–32 Haplosclerida;Metaniidae;

(Addis and Peterson, 2005) DQ176781 DQ176774 —

Ephydatia cooperensis (Petersonand Addis, 2000)

Haplosclerida;Spongillidae

(Peterson and Addis, 2000) DQ087505 AF140354 —

Ephydatia sp. Haplosclerida;Spongillidae

Germany, Rudersdorf (nr. Berlin), Lake Kalksee, May 2004 DQ167173 n.d. ZMB Por12660

Ephydatia fluviatilis 1 (Linnaeus,1759)


Germany, Erkner (nr. Berlin), Lake Falkensee, December 2003 AJ843884 AJ705048 —



Germany, Rudersdorf (nr. Berlin), Lake Kalksee, July 2004 DQ167174 DQ167158 ZMB Por12658



COI: (Addis and Peterson, 2005), 18S : (Richelle-Maurer et al.,2006)

DQ176777 AY578146 —

Ephydatia muelleri 1(Lieberkuhn, 1855)


(Peterson and Addis, 2000) (Addis and Peterson, 2005) DQ176778 AF121110T —

Ephydatia muelleri 2(Lieberkuhn, 1855)


Germany, Rudersdorf (nr. Berlin), Lake Unterer Kalksee, May 2004 DQ167172 DQ167159 ZMB Por12659

Eunapius fragilis 1(Leidy, 1851) Haplosclerida;Spongillidae

(Peterson and Addis, 2000) (Addis and Peterson, 2005) DQ176779 AF121111 —

Eunapius fragilis 2 (Leidy, 1851) Haplosclerida;Spongillidae

USA, North Carolina, October 2003 AJ843882 n.d. —

Eunapius carteri (Bowerbank,1863)


Germany, Erkner (nr. Berlin), Channel to Locknitz River, July 2004 DQ167175 DQ167160 ZMB Por12661

Lubomirskia abietina(Swartschewski, 1901)


Russia, Lake Baikal Listvianka, 12 m, July 2000 DQ167170 DQ167156 ZMB Por12655

Lubomirskia baicalensis (Pallas1771)


Russia, Lake Baikal, Listvianka, 8 m, July 2000 DQ167169 DQ167155 ZMB Por12654

Microciona prolifera (Ellis &Solander, 1786)

Poecilosclerida;Microcionidae

U.S., Massachusetts coast near Woods Hole, 2004 AJ843888 AJ705047 —

Nudospongilla vasta 1(Weltner, 1901)


Indonesia, Sulawesi, Lake Matano, S shore, nr. cave entrance, 2003 DQ167179 DQ167165 ZMB Por12651

Nudospongilla vasta 2(Weltner, 1901)


Sulawesi, Lake Matano, S shore, nr. cave entrance, 2003 DQ167180 DQ167166 ZMB Por12652

Ochridaspongia sp. Haplosclerida;Malawispongiidae

Macedonia, Lake Ohrid, Ohrid Bay, 35–15 m, leg.: ChristianAlbrecht, September 2004

EF025855 n.d. ZMB Por12689

Pachydictyum globosum(Weltner, 1901)

Haplosclerida;Malawispongiidae

Indonesia, Sulawesi, Lake Poso, Cape Bancea, 2003 DQ167177 DQ167163 ZMB Por12649

878M

.J.

Meix

ner

eta

l./

Mo

lecula

rP

hy

log

enetics

an

dE

volu

tion

45

(2

00

7)

87

5–

88

6

Pac

hyd

icty

um

glo

bo

sum

n.

ssp

.H

aplo

scle

rid

a;M

alaw

isp

on

giid

aeIn

do

nes

ia,

Su

law

esi,

Lak

eP

oso

,C

ape

Ban

cea,

2003

DQ

16

717

8D

Q1

67

16

4Z

MB

Po

r12

650

Pse

ud

axin

ella

reti

cula

ta(R

idle

y,18

81)

Hal

ich

on

dri

da;

Axi

nel

lid

aeB

erm

ud

a,F

latt

sIn

let

31M

ay20

03,

leg.

:W

olf

gan

gS

terr

eran

dK

lau

sR

utz

ler

AJ

84

38

94

AJ

705

04

6—

Sco

pal

ina

ruet

zler

i(W

ied

enm

eyer

,19

77)

Hal

ich

on

dri

da;

Dic

tyo

nel

lid

ae18

S:

Ber

mu

da,

Fla

tts

Inle

t31

May

2003

,le

g.:

Wo

lfga

ng

Ste

rrer

and

Kla

us

Ru

tzle

r;C

OI:

(Nic

ho

ls,

2005

)A

J6

21

546

AY

5619

76—

Sp

on

gill

ala

cust

ris

1(L

inn

aeu

s,17

59)

Hap

losc

leri

da;

Sp

on

gill

idae

Ger

man

y,E

rkn

er(n

r.B

erli

n),

Lo

ckn

itz

Riv

er,

No

vem

ber

2003

AJ

84

38

83

AJ

703

89

0Z

MB

Po

r12

666

Sp

on

gill

ala

cust

ris

2(L

inn

aeu

s,17

59)

Hap

losc

leri

da;

Sp

on

gill

idae

Ger

man

y,R

ud

ersd

orf

(nr.

Ber

lin

),K

alk

grab

enC

han

nel

,Ju

ly20

04D

Q1

67

17

6D

Q1

67

16

1Z

MB

Po

r12

662

Sp

on

gill

ala

cust

ris

3(L

inn

aeu

s,17

59)

Hap

losc

leri

da;

Sp

on

gill

idae

(Pet

erso

nan

dA

dd

is,

2000

)n

.d.

AF

1211

12—

Sw

arts

chew

skia

pap

yrac

ea(D

ybo

wsk

i,18

80)

Hap

losc

leri

da;

Lu

bo

mir

skii

dae

Ru

ssia

,L

ake

Bai

kal

,C

entr

alb

asin

,B

ols

hie

Ush

kan

iIs

lan

d,

12m

;le

g.:

Dir

kB

on

se,

Au

gust

2003

DQ

16

717

1D

Q1

67

15

7Z

MB

Po

r12

656

Tro

cho

spo

ngi

lla

ho

rrid

a(W

eltn

er,

1901

)H

aplo

scle

rid

a;S

po

ngi

llid

aeG

erm

any,

Neu

bu

rgam

Rh

ein

,R

hin

eR

iver

;le

g.:

Joch

enG

uge

l,20

05E

F0

25

854

AY

60

932

0Z

MB

Po

r12

665

Tro

cho

spo

ngi

lla

pen

nsy

lvan

ica

Hap

losc

leri

da;

Sp

on

gill

idae

(Pet

erso

nan

dB

utt

erfi

eld

,20

05)

DQ

0875

03D

Q08

7507

New

seq

uen

ces

are

inb

old

.O

rigi

ns

are

on

lygi

ven

for

spo

nge

ssa

mp

led

du

rin

gth

isw

ork

;n

.d.,

no

td

eter

min

ed.


and another part was frozen in liquid nitrogen for molecu-lar analysis. Species used for molecular genetic approacheswere identified according to spicule morphology and skele-tal architecture (Fig. 1b–i,k). Samples of all specimens arekept in the Museum of Natural History at Humboldt-Uni-versity Berlin, Germany.

2.2. Isolation of total genomic DNA

About 0.5 g fresh or ethanol-preserved material washomogenized in a mortar under liquid nitrogen and trans-ferred into 15 ml Falcon tubes using a spatula chilled inliquid nitrogen. 1 ml (v/w) 2· CTAB-buffer (2% CTAB,100 mM Tris–HCl pH 8.0, 20 mM EDTA, 1.4 M NaCl)was added, vortexed and incubated for 15 min at 65 �C ina water bath. One total volume chloroform/isoamylalcohol(24:1) was added and vortexed for 1 min. For phase separa-tion, samples were centrifuged for 10 min at 2500g at roomtemperature (RT). The upper phase was transferred intonew tubes, the exact amount checked and 1/5 volume 5·CTAB buffer (5% CTAB, 0.35 M NaCl) added, mixed andincubated at 65 �C for 10 min. Following chloroformextraction as before, 1–1.5 vol 1· CTAB (1% CTAB,50 mM Tris–HCl pH 8.0, 10 mM EDTA) was given to theupper phase for precipitation of DNA for 1 h at RT. Aftercentrifugation at 2500g for 20 min, the pellet was resus-pended in 1 ml of HS-TE (10 mM Tris–HCl pH 8.0,1 mM EDTA, 1 M NaCl). In some cases the pellet neededto be incubated at 65 �C for 10 min to achieve completeresuspension. DNA was precipitated from this suspensionby 3 vol of 96% ethanol/3 M NaOAc 30:1. Followingcentrifugation and a 70% ethanol wash step, the pellets wereair-dried and finally resuspended in 50 ll of sterile water.

2.3. Oligonucleotide primers, PCR conditions, molecular

cloning and sequencing

PCR was performed in a total volume of 50 ll, contain-ing the following: 5 ll 10· reaction buffer (Qiagen), finalconcentration of 3 mM MgCl2, 2.5 ll dNTP mix (2.5 mMeach), 20 pmole of each primer and 1 U Taq polymerase(Qiagen). PCRs were performed on a PTC-200 thermalcycler (MJ research). The PCR regime for amplificationwas: 1.5 min at 96 �C, followed by 35 cycles with each30 s 96 �C; 40 s 53–58 �C; 3 min 72 �C, and a final exten-sion step at 72 �C for 30 min. PCR fragments were eitherpurified directly using Qiaquick columns, or by elutionfrom 1% agarose gels and subsequent column purification.The purified DNA fragments were either cloned into vectorpDrive (Qiagen) or sequenced directly with the respectivePCR primers driving the sequencing reaction. The primerpair HCO2198/ LCO1490 (Table 2, Pop et al., 2003) wasused for amplification of a fragment of the mitochondrialcytochrome oxidase subunit I gene (COI) from Spongilla

lacustris. Exploiting sequence information from marinesponges, we subsequently modified these primers to beeither 30 truncated by 5 nucleotides (HCO2198B), or con-

Table 2List of oligonucleotide primers used in this study

Target gene/purpose Designation Sequence Reference

mt COI amplification HCO2198 TAAACTTCAGGGTGACCAAAAAATCA(finishes off at a W ‘‘stop’’ codon)

(Pop et al., 2003)

mt COI amplification LCO1490 GGTCAACAAATCATAAAGATATTGG (Pop et al., 2003)mt COI amplification, alternative

primer setHCO2198B TAAACTTCAGGGTGACCAAAAAATCAAAA This paper

mt COI amplification, alternativeprimer set

LCO1490B GTCAACAAATCATAAAGATATTGGGACT This paper

COI sequencing COIsqf TAGGATCTGCTTTTGTAGAGCAAGG This paperCOI sequencing COIsqr ATTGCCATATCAACCGATCCCCC This paperCOI amplification of

Baikalospongia, Ephydatia andSpongilla

COI-fresh-fw CAGGCATGATAGGTACAGCATTTAG This paper

COI amplification ofBaikalospongia, Ephydatia andSpongilla

COI-fresh-rv CTCCYCCAGCAGGATCAAAG This paper

COI amplification alternative forSulawesisponges

COI-fw-M GGCATGATAGGTACAGCATTTAG This paper

COI amplification alternative forSulawesisponges

COI-sula-rv CTCCYCCAGCAGGATCAAARAA This paper

18S rRNA amplification 18a_mdl AACCTGGTTGATCCTGCCAGT (Medlin et al., 1988)18S rRNA amplification 18b_mdl TGATCCTTCTGCAGGTTCACCTAC (Medlin et al., 1988)18S rRNA sequencing 18SFdwn CGCGGTAATTCCAGCTCCAATAGCG This paper18S rRNA sequencing 18SFup CTACGAGCTTTTTAACTGCAACAACTTTAATATACGC This paper18S rRNA sequencing 18SRdwn CAAATTAAGCCGCAGGCTCCACTCC This paper18S rRNA sequencing 18SRup GTG GAG CCT GCG GCT TAA TTT GAC TCA A This paper


tain one additional nucleotide at the 50 end and four at the30 end (LCO1490B). However, the frequent and tight asso-ciation between sponges and other eukaryotic organismsturned out as a particular technical challenge due to thefrequent amplification of contaminants. To avoid the con-taminants, we developed a more specific primer set, COI-fresh-fw/COI-fresh-rv, taking several unique positions intoaccount. These primers worked excellent for the amplifica-tion of COI fragments from Baikalospongia, Ephydatia andSpongilla. However, for the sponges from Sulawesi, thereverse primer needed to be modified again, by slightlyshifting it along the known COI sequences, reducing spec-ificity through introduction of a fourfold degeneracy, yield-ing primer COI-sula-rv. In order to have matching physicalproperties, COI-fw-M was developed which is based onCOI-fresh-fw, but two nucleotides shorter at the 50 end.As a result we obtained 20 new high quality COI sequencesfrom freshwater sponges and 23 new sequences altogetherfrom all sponges analyzed. The complete list of oligonucle-otides is given in Table 2.

For amplification of 18S rDNA, we used the primer pair18a_mdl/18b_mdl (Medlin et al., 1988). However, onlypartial sequences were obtained for Lubomirskia baicalensis

and Pachydictyum globosum n. ssp. For Ochridaspongia sp.it was not possible to obtain rDNA amplicons. Altogether,15 novel nearly complete and two partial 18S rDNAsequences were obtained. In all cases, both DNA strandswere sequenced using standard protocols and the Big DyeTerminator chemistry version 1.1 (Applied BiosystemsInc.) on an Applied Biosystems 373 or 377 automatedDNA sequencer. Sequencing data were manually cor-

rected, edited, blasted against Genbank and aligned usingABI software SeqEd, SeqAlign, or Lasergene (DNAstar).All sequences have been deposited in GenBank, accessionnumbers are given in Table 1.

2.4. Inference of phylogenetic relationships

A multiple alignment of 18S and COI sequences was per-formed using Clustal X 1.81 (Thompson et al., 1994). Multi-ple alignment parameters were 15 for gap opening and 6.66for gap extension. In case of the mitochondrial COI gene,lengths of the initial PCR fragments differed due to theuse of different primer sets for the different species. There-fore, the 28 sequences selected for analysis were between496 and 681 nt long with an overlap of 486 nt between allof them. The main criterion for selecting outgroups wasthe availability of a COI as well as of an 18S rDNAsequence. In preliminary analyses we tested several addi-tional COI sequences of different marine sponges belongingto distinct phylogenetic groups (Suberites domuncula, Pros-

uberites laughlini, Geodia media and Haliclona amphioxa)which we downloaded from GenBank but found that theyhad no effect on the relationships between the here com-pared freshwater sponges. In addition, selected freshwatersponge sequences were obtained from Genbank. Theseincluded Ephydatia cooperensis, E. muelleri, Eunapius fragi-

lis, Corvomeyenia sp. and Trochospongilla pennsylvanica ofNorth American origin (Addis and Peterson, 2005; Petersonand Addis, 2000; Peterson and Butterfield, 2005).

The total alignment length of 18S rDNA sequences was1694 nt. For Ochridaspongia sp. it was not possible to get

Baikalospongia bacilliferaLubomirskia abietina

Baikalospongia intermediaEphydatia muelleri 2Lubomirskia baicalensisEphydatia muelleri 1Ephydatia sp.Ephydatia fluviatilis 1Baikalospongia dzhegatajensisEphydatia cooperensisSwartschewskia papyraceaOchridaspongia sp.

Ephydatia fluviatilis 2Spongilla lacustris 2Spongilla lacustris 1

Eunapius carteriEunapius fragilis 2Eunapius fragilis 1

Trochospongilla horridaNudospongilla vasta 1Nudospongilla vasta 2Trochospongilla pennsylvanica

Pachydictyum globosumPachydictyum globosum n. ssp.

Corvomeyenia sp.Pseudaxinella reticulataScopalina ruetzleri

Microciona prolifera

0.02

E

A

M

P

F

GH

I

JLO

N

99919999P

71928184M

83949090N

--7175O

79-9188L

-82--J

--7884I

74927476H

87988888G

9599--F

-959494E

-84--A

MPMLNJMEnode

ig. 2. Minimum evolution phylogenetic tree of freshwater sponges basedn COI. A 486 bp COI fragment was compared and analyzed for 28pecies. The marine species Pseudaxinella reticulata, Scopalina ruetzleri

nd Microciona prolifera served as outgroups. The species names aredicated in the tree. Bootstrap support values from 10,000 repetitions are

iven for the nodes labeled by capital letters: ME (distance tree), NJdistance tree), and MP as implemented in MEGA 3.1. The ML analysisas performed in the program Tree-Puzzle. Different models of substi-

ution were compared, the search options for the tree shown (length.4129452) were LogDet as model of substitution, heterogeneous patternmong lineages and uniform rates among sites. Gaps in the alignmentere completely deleted and the initial tree was found through NJ.


an 18S sequence since the sample was contaminated bymites (Arachnidae) and except Lubomirskia baicalensis

and Pachydictyum globosum n. ssp. for which we obtainedonly truncated sequences (1143 bp for P. globosum n. ssp.and 1140 for L. baicalensis). Phylogenetic trees were con-structed under different optimality criteria (neighbour join-ing (NJ), minimal evolution (ME) and maximumparsimony (MP)) as integrated in MEGA 3.1 (Kumaret al., 2004). TREE-PUZZLE 5.2 (Schmidt et al., 2002)and MrBayes 3.1 (Huelsenbeck and Ronquist, 2001) wereemployed for maximum likelihood (ML) and Bayesiananalyses, respectively. Sequences of the marine speciesPseudaxinella reticulata, Microciona prolifera and Scopali-

na ruetzleri were included as outgroups. Bootstrap analysiswith 1000 repetitions was performed under ML, MP andME approaches in MEGA 3.1; 10,000 quartet puzzlingsteps were performed in TREE-PUZZLE 5.2 and at least1,000,000 generations were generated in MrBayes 3.1.

3. Results

3.1. Selection of sponges

In the current study, four sponges from Lake Poso andLake Matano included two morphologically differentsponges belonging to the genus Pachydictyum. One wasthe previously described Pachydictyum globosum Weltner,1901 and the other was a similar but as yet undescribedPachydictyum globosum n. ssp. (later to be described as aproper species). These two Pachydictyum were morpholog-ically very close but not identical (Fig. 1c–e). The other twosponges, originating from Lake Matano, were identified asNudospongilla vasta. These two sponges differed in shapeand global morphology from each other but were uniformin spicule shape and megascleres architecture. Their mega-scleres were similar to those from Nudospongilla vasta

(Weltner, 1901), but no gemmules could be found, thusthe species definition meets the same problem as with theBaikal sponges, were gemmules are also absent. Further-more, five sponge specimens from Lake Baikal and onefrom Lake Dzhegatai Kul (Baikalospongia dzhegatajensis)were included. The sponges from Lake Baikal were identi-fied as Baikalospongia bacillifera Dybowski, 1880, Baikalo-

spongia intermedia Dybowski, 1880, Lubomirskia abietina

Swartschewski, 1901, Lubomirskia baicalensis Pallas 1771and Swartschewskia papyracea Dybowski, 1880. Identifica-tion of all species was first based on a thorough morpho-logical characterization (Fig. 1f–l). The Spongillidae havea worldwide distribution, whereas the families Lubomir-skiidae and Malawispongiidae are endemic or at least aregeographically restricted. Therefore, we included severaladditional samples, Ochridaspongia sp., and a number ofdifferent individuals of common Spongillidae. The latterwere identified as Ephydatia muelleri, E. fluviatilis, Eunapiusfragilis and Eunapius carteri, as Spongilla lacustris andTrochospongilla horrida. For details and references seeTable 1.

3.2. Molecular phylogenies based on the mitochondrial COI

gene

Sequence identity among all compared COI sequenceswas very high, ranging between 96% and 100%. The out-groups COI sequences shared 82–84% identity with eachof the ingroup sequences. Two of the outgroups (Pseudax-

inella reticulata, Halichondrida, and Microciona prolifera,Poecilosclerida) were chosen on the basis of a previousanalysis in which, based on 18S and 28S rDNA, Spongilli-na (freshwater Haplosclerida) had been joined in a clade‘‘G4’’ together with these (and other) orders of marinesponges (Borchiellini et al., 2004) whereas marine Haplo-sclerida had been in a different clade ‘‘G3’’. Finally, ourthird outgroup (Scopalina ruetzleri) had previously beenfound in a joint clade A with freshwater sponges basedon large subunit rDNA sequences (Nichols, 2005). Phylo-genetic trees were calculated under different models ofmolecular sequence evolution (Fig. 2). The two individualSpongilla lacustris 1 and 2 from Germany shared an iden-tical COI sequence with each other, as did the two Nudo-

pongilla vasta 1 and 2, but also the two morphologically

Fosaing(wt0aw


different Pachydictyum globosum. Surprisingly, COIsequences recovered were not consistent with the morpho-logical assignment of different species and genera. Forexample, Ephydatia muelleri, Baikalospongia intermedia

and Lubomirskia baicalensis shared identical sequenceswith each other as did E. cooperensis (from N America)with E. fluviatilis 1 and 2 from Germany and with B. dzheg-

atajensis and Swartschewskia papyracea, moreover theMalawispongiidae Ochridaspongia sp. was part of thisgroup. The metaniid sponge Corvomeyenia sp. was at abasal position within the spongillids. The Sulawesiansponges, Pachydictyum globosum, Pachydictyum globosum

n. ssp. and Nudospongilla vasta 1 and 2, were joined in amonophyletic group together with Trochospongilla species.This clade was the second-most deeply branching clusteramong freshwater sponges. The position of an Ephydatia

sp. which we collected from a German site remained unre-solved in clade E since it differed at two distinct positionsboth from the E. fluviatilis and the E. muelleri sequences.While the basic topology of all reconstructed trees wasidentical, we found that the separation power of the COIdata was not very high and some branches did not receivehigh statistical support due to the similarity among thecompared sequences.

Deducing the encoded amino acids from COI sequences,the presence of translational exceptions was apparent, sinceUGA frequently codes for tryptophan. There are four tryp-tophan codons, at positions 114–116, 180–182, 249–251and 429–431 within the partial 486 bp COI fragment. How-ever, we noticed differences in the frequency of how oftenUGA is actually used as tryptophan codon. In the marinesponges Pseudaxinella reticulata and Scopalina ruetzleri aswell as Corvomeyenia sp. UGA was present in all fourcases. For the freshwater sponges (except Corvomeyenia

sp.) the triplet UGA was present three times, at the first,second and forth position, and UGG present once (249–251). In Microciona prolifera, UGG was present for thefirst occurrence (114–116), different from all other speciesstudied here.

3.3. Molecular phylogenies based on 18S rDNA

Phylogenies employing the 18S rDNA were similar tobut not identical to those obtained using COI data: Cor-

vomeyenia sp. was again the most deeply branching speciesamong freshwater sponges and the clade consisting ofPachydictyum globosum, Nudospongilla vasta, Trochospong-illa horrida and Trochospongilla pennsylvanica was locatedat the second-most basal position within the Spongillina(not shown). The two individual Spongilla lacustris 2 and3 from Germany, the two Ephydatia muelleri, one fromNorth America (E. muelleri 1) and one from Germany(E. muelleri 2), as well as the two individual Nudopongilla

vasta 1 and 2, each shared an identical 18S rDNA witheach other, as we had found for their COI sequences.Another set of species with identical sequences consistedof Baikalospongia intermedia, B. bacillifera and Lubomirs-

kia abietina. These Lubomirskiidae (including Swarts-

chewskia) formed a clade with Ephydatia muelleri as itssister-taxon, which is different from the COI analysis, how-ever support at this level in the tree was poor. The Swarts-

chewskia papyracea 18S rDNA varied at three positionsfrom the E. fluviatilis or E. cooperensis sequence.

3.4. Inference of molecular phylogenies based on

concatenated sequences

Inference of phylogenetic relationships between fresh-water sponges based on either the 18S rDNA or COI dataalone suffered from the high similarity among all comparedsequences. For instance, in the 18S dataset, the two leastrelated freshwater species, Lubomirskia abietina and Cor-vomeyenia sp. had still 1654 of 1683 bp in common (98%identity, 29 divergent sites), whereas the differentiationbetween Pachydictyum globosum and Nudospongilla vasta

was based on only two divergent positions. Since the resultsobtained on basis of COI and 18S were very similar, thetwo datasets were concatenated for those 21 specimensfor which both sequences were available, resulting in a2177 bp alignment. In this comparison, Lubomirskia abieti-na and Corvomeyenia sp. still had an identity of 97.8%,however, this is statistically more robust since it is basedon 2122 shared and 47 divergent, i.e. phylogeneticallyinformative sites.

Based on this alignment, very robust results wereobtained in the ML analysis with Tree-Puzzle as indicatedby the support values of >96 for all nodes except C and D(Fig. 3). Identical tree topologies were obtained under allmodels employed except MP and nodes C and D whichwere reversed in the Bayesian compared to the non-Bayes-ian analysis (Fig. 3). The statistical support was somewhatlower for MP, caused by the smaller number of parsimony-informative sites. All freshwater sponges except Corvom-

eyenia belonged to one of three larger clades labeled (i)–(iii), each beginning with the node E, H or J in Fig. 3and each containing different Spongillidae. These groupsor clades are (i) Malawispongiidae (except Ochridaspongia

sp. which was not present in the concatenated analyses),two different Trochospongilla and Nudospongilla; (ii) a jointclade consisting of the two genera Eunapius and Spongilla,each represented by two sequences; (iii) all Lubomirskiidaeand all species belonging to the genus Ephydatia.

4. Discussion

From a molecular point of view, sponges are stronglyundersampled, resulting in a lack of molecular phyloge-netic markers for sponges and of sequence information ingeneral. The exact position of several sponge taxa andthe monophyly of the phylum Porifera is a matter of debate(Adams et al., 1999; Borchiellini et al., 2004,2001; Cavalier-Smith et al., 1996; Collins, 1998; Medina et al., 2001). Todate, about 7500 sponge species have been described andthe same number is expected still to await discovery (Hoo-

Fig. 3. Phylogenetic relationships among ubiquitous and endemic species of freshwater sponges based on a multiple alignment of concatenated nucleotidesequences for COI and 18S rDNA (21 sequences with 2177 positions alignment length). The number of constant sites was 1810 (=83.1% of all sites). Theshown tree resulted from a Bayesian analysis. Different individuals of the same species are labeled by numbers. All freshwater sponges studied here belongto one of three clades (i)–(iii). Clade (i) contains Trochospongilla, Nudospongilla and Pachydictyum; (ii) Spongilla lacustris and Eunapius; (iii) consists of allLubomirskiidae and Ephydatia. The nodes have been labeled by capital letters (A)–(M) and the posterior probabilities for each split or statistical supportvalues resulting from bootstrapping analysis (10,000 times; for ME, NJ and MP) or 10,000 steps of Quartett puzzling (ML) are given in the table if 70 orgreater; *nodes C and D were reversed in Bayesian compared to non-Bayesian analysis but were only weakly supported. For Bayesian analysis, theevolutionary model was set to GTR with gamma-distributed rate variation across sites and a proportion of invariable sites. The standard deviation of splitfrequencies was 0.007454 after 2,000,000 generations and a sampling frequency of 100 resulting in 20001 samples. The potential scale reduction factor wasreasonable close to 1.0 (1.000–1.005). The total tree length is 3.213703, the six reversible substitution rates as estimated by the program are:r(A � C) = 0.161, r(A � G) = 0.203, r(A � T) = 0.0118, r(C � G) = 0.050, r(C � T) = 0.493, r(G-T)=0.07998. The four stationary state frequencies are:pi(A) = 0.256, pi(C) = 0.201, pi(G) = 0.265, pi(T) = 0.277; the shape of the gamma distribution of rate variation across sites alpha = 0.07475; theproportion of invariable sites pinvar = 0.913278. As substitution model, HKY was used in ML, Kimura 2-parameter in ME, NJ and MP. The transition/transversion ratio as estimated from data set was 2.62; the pyrimidine transition/purine transition ratio 0.80. The number of parsimony-informative siteswas 168 from a total of 2161. MP analysis resulted in 8 trees of length 525 (CI = 0.849524, RI = 0.784741, RCI = 0.666656). Differences between the MPtrees were exclusively in clade (iii) and affected the placement of E. muelleri and the other three Ephydatia species. The marine species P. reticulata,S. ruetzleri and M. prolifera served as outgroups in the non-Bayesian analysis (not shown).


per and van Soest, 2002). To speed up the identificationprocess of both known and new species, a combinationof morphological and molecular methods would be desir-able. Mitochondrial COI gene 50 end sequences were sug-gested as molecular marker for barcoding eukaryoticorganisms (Blaxter, 2004; Hebert et al., 2003). The COIgene is ubiquitous; in several groups of organisms such asbirds, polychaetes or insects, it was shown to be well con-served but also to be variable enough to serve as phyloge-netic and taxonomic marker. Mitochondrial markers havebeen introduced recently for sponge systematics (Addis andPeterson, 2005; Nichols, 2005; Schroder et al., 2003). Basedon the here presented data set, we extend these studies bythe observation of translational exceptions within thesponge mitochondrial COI gene, in which the opal codonUGA frequently encodes tryptophan instead of terminat-ing translation. While this finding alone is in consent withthe mitochondrial codon usage of most invertebrates,

including the closely related cnidarians, it is interesting tonote that the extent and frequency of these exceptions cor-relates with the results of the phylogenetic analysis. Forinstance, Corvomeyenia sp. is at the most basal positionamong freshwater sponges (Fig. 2) and it has all four trans-lational exceptions in common with the marine spongesP. reticulata and S. ruetzleri, whereas all other freshwatersponges use the triplet UGA three times but once UGGto code for tryptophan. However, more work on longermitochondrial sequences is required to see to what extentthese translational exceptions correlate with phylogeny.

The critical test of DNA barcoding is whether it enablesdiscrimination between closely related species. Comparisonof 50 COI sequences from 13,000 pairs of congeneric speciesshowed a mean divergence of 11.3% (Hebert et al., 2003).With the striking exception of some cnidarians (sea anem-ones, corals, and some jellyfish), 98% of the species pairsexhibited sequence divergences greater than 2%. Since


sponges are a sister phylum to cnidarians it may not sur-prise that we found also for the here investigated differentspecies of freshwater sponges a COI sequence divergenceof only 2% as the maximum. While additional poriferanCOI sequences could be expected helpful for barcoding,sponge COI here did not allow discrimination between clo-sely related species. COI sequences appeared to be too con-served also in an analysis of populations of the marinedemosponges Crambe crambe in the Mediterranean andeastern Atlantic (Duran et al., 2003) and of Astrosclera

willeyana in the Red Sea and Indo-Pacific (Worheide,2006). Thus, COI alone appears not as a particularly suit-able marker for barcoding closely related sponges from thesea or freshwater. As an alternative, it should be used incombination with other molecular markers. In fact, thedegree of sequence divergence observed here was aboutthe same in case of COI and of 18S rDNA, encouragingus to concatenate the two datasets. With a similarapproach, robust phylogenetic trees were previouslyobtained for 5 species of Demospongiae (Addis and Peter-son, 2005).

Our data extend the existing information on the phy-logeny and taxonomy of freshwater sponges. For instance,based on the analyses presented in Figs. 2 and 3, there isno clade supporting Spongillidae as a monophyleticgroup. The metaniid sponge Corvomeyenia sp. was previ-ously suggested basal to the spongillids (Addis and Peter-son, 2005). That hypothesis depended upon the method ofanalysis and sequence and was found in two of three anal-yses of COI and a combined 18S rDNA + COI analysis,whereas two of three analyses of 18S rDNA weakly sup-ported Trochospongilla pennsylvanica as the basal taxon.Our study confirmed the hypothesis by Addis and Peter-son (2005), since all analyses provided strong supportfor Corvomeyenia sp. as the most basal spongillid species,on basis of the concatenated data set as well as of 18SrDNA and COI alone.

Another set of interesting observations was made for thegenus Ephydatia. COI sequences were identical for E. flu-

viatilis 1 and 2 from two different sites in Northern Ger-many with the published sequence of E. cooperensis. Incontrast, the 18S rDNA sequence of E. fluviatilis 3(Richelle-Maurer et al., 2006) differed in one nucleotidevery close to the 30 end from that of E. cooperensis (Peter-son and Addis, 2000). This single nucleotide difference haspreviously been used as evidence to differentiate these twospecies. When we compared E. fluviatilis 3 with E. fluviatilis

1 and 2 and E. cooperensis, an identical sequence was foundalso for the 18S rDNA in the here compared region. There-fore, we analysed 12 additional samples of E. fluviatilis forthe critical 18S rDNA 30 end (not shown), which were allfound to be identical with E. cooperensis, E. fluviatilis 1and 2 and to the exclusion of the E. fluviatilis 3 sequence.Thus, in this case, their very high genetic similarity as indi-cated by the shorter COI fragments was confirmed.Although a single nucleotide difference might result fromerrors, especially at the very beginning or very end of a

sequence, this result might indicate the presence of micro-heterogeneity within the E. fluviatilis 18S rDNA 30 end.Whether species with identical or one base maximum differ-ence in the 18S rDNA and COI sequences are separate spe-cies cannot be decided. Previous analysis of the ITS2rDNA region showed a difference of 10 nt between E. flu-viatilis and E. cooperensis as well as between B. bacillifera

and L. baicalensis (Addis and Peterson, 2005). Thus, themore variable ITS rDNA region appears more suitable asmarker at this level. We collected one Ephydatia sp. froma German site which differed at two distinct positions fromthe E. fluviatilis, E. cooperensis and the E. muelleri COIsequences, and hence was placed at an intermediate posi-tion (see Fig. 2). Based on its spicule morphology, we werenot able to assign it to either of the known species. Hencethis specimen might represent a distinct species or resultfrom hybridization and recombination between E. fluviatil-

is and E. muelleri. Further work including additional locisuch as ITS rDNA would be required to resolve its positionbut was beyond the scope of the current study.

Our analyses included four of the six currently recog-nized families of living freshwater sponges (Manconi andPronzato, 2002) and provide some clues for the possibleevolution and phylogenetic affiliation of endemic species.The Malawispongiidae family (Manconi and Pronzato,2002) is based on weak morphological characters, such asthe absence of gemmules and contains five monotypic gen-era, Malawispongia, Cortispongilla, Ochridaspongia, Pachy-

dictyum and Spinospongilla, all inhabiting ancient tectoniclakes in Africa, Europe and Asia (Malawi, Tanganyika,Kinneret, Ohrid and Poso). The wide geographical distanceof these isolated regions makes a close phylogenetic rela-tionship between these genera doubtful. Indeed, our analy-sis of COI from Ochridaspongia sp. surprisingly suggestedthis sponge as associated to the Ephydatia-Lubomirskiidaeradiation rather than being closely related to Pachydict-

yum, making the Malawispongiidae polyphyletic. More-over, the DNA data do not support a close relatedness ofthe Sulawesi sponges Pachydictyum and Nudospongilla toSpongilla but rather to Trochospongilla since we found inall analyses high support for a joint clade of Pachydictyum,Nudospongilla and Trochospongilla. Within this clade, thegenus Trochospongilla is paraphyletic with the generaPachydictyum and Nudospongilla more closely related toT. pennsylvanica than to T. horrida. Our data suggests acommon ancestor to group (i) in Fig. 3.

The origin of sponges from Lake Baikal has been enig-matic until recently. Spicules of the endemic Lubomirskii-dae have been found in Late Miocene sediments (at least24 MY old). Detailed records of sponge spicules in sedi-ments from drilling cores of Lake Baikal (Martinson,1936, 1938, 1940, 1948) demonstrate that it was possibleto distinguish the Lubomirskiidae taxa at a species level(although Martinsons determination of four Spongillidaespecies on account of isolated spicules should be consideredwith caution). Thus, according to the fossil record, theLubomirskiidae ought to be an old still extant taxon of


endemic freshwater sponges. Previously, the degree ofnucleotide substitutions within the 18S rDNA of differentBaikalian sponges was considered too low to drawunequivocal conclusions (Itskovich et al., 1999). However,with �630 nt the 18S rDNA sequences had been very shortin that analysis. In our COI-18S rDNA concatenated data-set, Lubomirskiidae had between 2161 (Baikalospongia

bacilifera) and 2167 (Swartschewskia papyracea) of thecompared 2169 nt in common with Ephydatia fluviatilis,corresponding to a 99.6–99.9% sequence identity. In con-trast, Spongilla lacustris had a 1.2% sequence divergencefrom the Lubomirskiidae (2145–2147 shared positions),and was separated into a different clade (ii) in Fig. 3. There-fore, our data show that the sponges from Lake Baikalbranch off from the other Spongillina together withEphydatia fluviatilis, E. cooperensis and E. muelleri. Thisis essentially what Addis and Peterson (2005) have found,thus confirming and extending those results. Thus, allavailable data strongly indicate a common origin of theendemic Lubomirskiidae together with the ubiquitousgenus Ephydatia whereas the genera Ephydatia and Baik-

alospongia are not monophyletic. Moreover, the close relat-edness between Lubomirskiidae and the genus Ephydatia

constitutes an interesting parallel to the Malawispongiidaein which Pachydictyum is more closely related to the ubiq-uitous genus Trochospongilla. Within the genus Ephydatia,coherence exists in the E. cooperensis–E. fluviatilis part asindicated by the identity between E. cooperensis and thetwo E. fluviatilis of Central European origin, even in theconcatenated sequences. Thus E. cooperensis, although itlacks the ability of gemmule formation, might be a morerecent split off from the cosmopolitan E. fluviatilis.

Indirectly, our data has shed some light on the originand loss of gemmules. Our data reinforce a view that theloss of gemmules may well have happened several timesduring the evolution of endemic taxa from the Spongilli-dae. The cosmopolitan Spongillidae have been found asisolated spicules in freshwater sediments of the Mesozoicand possibly also from Late Palaeozoic, whereas the oldestfossile gemmules date at least 112 MY, to the Early Creta-ceous (Volkmer-Ribeiro and Reitner, 1991). However, theability to form gemmulae might be traced back to the ori-gin of the Spongillina, going along with, or even as a pre-requisite for, the sponge colonization of freshwater envi-ronments. For instance, there might be no evolutionaryadvantage for gemmules in a stable lake environment, suchas Lake Baikal. Gemmules are absent in the Lubomirskii-dae, and even the Baikalian Spongillidae show a tendencyto reduced gemmule formation, probably due to the stableenvironmental conditions of Lake Baikal. Finally, ouranalyses clearly support the monophyletic origin of fresh-water-adapted sponges with the outgroups chosen andsmall number of species included. Endemic species of dif-ferent lake systems seem to have originated from a cosmo-politan founder species such as Ephydatia andTrochospongilla more frequently and independently as pre-viously thought.

Acknowledgments

Molecular analyses and SEM studies were supportedby the Systematics Research Fund of The SystematicsAssociation and The Linnean Society of London withinthe project ‘‘Taxonomy, phylogeny and biogeographyof freshwater sponges from old lakes in Sulawesi’’ toC.L. Collecting trips to Sulawesi were supported by theDeutsche Forschungsgemeinschaft (focus program ‘‘Radi-ations’’). DFG (DFG 436 RUS/17/2001) also supportedthe research visit of VI in Berlin for the study of baika-lian sponges. Additional financial support from theOcean Genome Legacy Foundation to WRH is gratefullyacknowledged. We would like to thank Dr. Igor Ivanov(Academy of Science, Kysyl, Russia), Dr. Christian Albr-echt (University of Giessen, Germany), Dr. Jochen Gugel(University of Stuttgart, Germany), Klaus Rutzler(Smithsonian Institution, Washington DC), WolfgangSterrer (Marine Aquarium Bermuda) and Dirk Bonsefor kindly providing sponge material for this study. Wewould like also to thank Sofia Efremova and YoshikiMasuda for help in the collection and identification ofsamples from Lake Baikal and two reviewers for criticalbut very helpful comments.

References

Adams, C.L., Mcinerney, J.O., Kelly, M., 1999. Indicators of relationshipsbetween poriferan classes using full-length 18S rRNA gene sequences.Mem. Queensl. Mus. 44, 33–43.

Addis, J.S., Peterson, K.J., 2005. Phylogenetic relationships of fresh-water sponges (Porifera, Spongillina) inferred from analyses of 18SrDNA, COI mtDNA, and ITS2 rDNA sequences. Zool. Scr. 34,549–557.

Arndt, W., 1938. Spongiologische Untersuchungen am Ochridasee. Arch.Hydrobiol. 34, 48–80.

Bellier, O., Sebrier, M., Seward, D., Beaudouin, T., Villeneuve, M.,Putranto, E., 2006. Fission track and fault kinematics analyses for newinsight into the Late Cenozoic tectonic regime changes in West-CentralSulawesi (Indonesia). Tectonophysics 413, 201–220.

Blaxter, M.L., 2004. The promise of a DNA taxonomy. Phil. Trans. R.Soc. Lond. B 359, 669–679.

Borchiellini, C., Chombard, C., Manuel, M., Alivon, E., Vacelet, J.,Boury-Esnault, N., 2004. Molecular phylogeny of Demospongiae:implications for classification and scenarios of character evolution.Mol. Phylogenet. Evol. 32, 823–837.

Borchiellini, C., Manuel, M., Alvion, N., Boury-Esnault, N., Vacelet, J.,Le Parco, Y., 2001. Sponge paraphyly and the origin of Metazoa. J.Evol. Biol. 14, 171–179.

Cavalier-Smith, T., Allsopp, M.T.E.P., Chao, E.E., Boury-Esnault, T.N.,Vacelet, J., 1996. Sponge phylogeny, animal monophyly and the originof the nervous system: 18S rRNA evidence. Can. J. Zool. 74, 2031–2045.

Collins, A.G., 1998. Evaluating multiple alternative hypotheses for theorigin of Bilateria: an analysis of 18S rRNA molecular evidence. Proc.Natl. Acad. Sci. USA 95, 15458–15463.

Duran, S., Pascual, M., Turon, X., 2003. Low levels of genetic variation inmtDNA sequences over the western Mediterranean and Atlantic rangeof the sponge Crambe crambe (Poecilosclerida). Mar. Biol. 144, 31–35.

Dybowski, W., 1880. Studien uber die Spongien des Russischen Reichesmit besonderer Berucksichtigung der Spongien-Fauna des Baikal-Sees.Memoires de l‘academie imperiale des sciences de St.-Petersbourg, VII.Ser. 27, 1–71.


Efremova, S.M., 2001. Sponges (Porifera). In: Timoshkin, O.A. (Ed.),Index of Animal Species Inhabiting Lake Baikal and its CatchmentArea. Novosibirsk Nauka, Novosibirsk, pp. 179–192.

Efremova, S.M., 2004. New genus and new sponge species of the familyLubomirskiidae Rezvoy, 1936. In: Timoshkin, O.A. (Ed.), Index ofAnimal species inhabiting Lake Baikal and its catchment area.Novosibirsk Nauka, Novosibirsk, pp. 1261–1278.

Gilbert, J.J., Hadzisce, S., 1984. Taxonomic notes on the shallow-waterendemic sponges of Lake Ohrid, Yugoslavia, with a description of twonew species and a redescription of Spongilla stankovici. Arch.Hydrobiol. 99, 331–339.

Greze, V.N., Greze, I.I., 1958. Ozero Chagytai. Izvestija VsesojuznogoGeograficeskogo Obshestva 90, 279–284.

Hadzisze, S., 1953. Beitrag zur Kenntnis der Spongillidenfauna der großenMazedonischen Seen (Dorja-, Prespa- und Ohridsee). Rec. Trav. Stat.Hydrobiol. Ohrid 3, 73–103.

Haffner, G.D., Hehanussa, P.E., Hartoto, D. (Eds.), 2001. The biologyand physical processes of large lakes in Indonesia. Lakes Matano andTowuti, pp. 183–192.

Hebert, P.D.N., Cywinska, A., Ball, S.L., deWaard, J.R., 2003. Biologicalidentifications through DNA barcodes. Proc. R. Soc. Lond., Ser. B.270, 313–322.

Hooper, J.N.A., van Soest, R.W.M., 2002. Systema PoriferaA Guide tothe Classification of Sponges. Kluwer Academic/Plenum Publishers,Dordrecht, New York.

Huelsenbeck, J.P., Ronquist, F., 2001. MrBAYES: Bayesian inference ofphylogenetic trees. Bioinformatics 17, 754–755.

Itskovich, V.B., Belikov, S.I., Efremova, S.M., Masuda, Y., 1999.Phylogenetic relationships between Lubomirskiidae, Spongillidae,and some marine sponges according partial sequences of 18S rDNA.Mem. Queensl. Mus. 44, 275–280.

Kozhov, M.M., 1972. Otcherki po Baikalovedeniju (Essays on BaikalScience). Vostochno-Sibirskoe knishnoe Izdatelstvo, Irkutsk.

Kozhow, M., 1963. Lake Baikal and its Life. W. Junk-Verlag, Den Haag.Kumar, S., Tamura, K., Nei, M., 2004. MEGA3: Integrated software for

molecular evolutionary genetics analysis and sequence alignment.Brief. Bioinform. 5, 150–163.

Manconi, R., Pronzato, R., 2002. Suborder Spongillina subord. nov:Freshwater sponges. In: Hooper, J.N.A., van Soest, R.W.M. (Eds.),Systema Porifera: A Guide to the Classification of Sponges. Kluwer,New York, pp. 921–1019.

Martinson, G.G., 1936. Razpredelenie spikul gubok v skvashine glu-bokogo burenija u c. Posolska na Baikale. Doklady Akademii NaukSSSR 4, 261–264.

Martinson, G.G., 1938. Iskopaemaya spongiofauna tretichnych otlosheniiPribaikalya. Doklady Akademii Nauk SSSR 21, 212–214.

Martinson, G.G., 1940. Materialy k issledovaniyu iskopaemoi mikro- ispongiofauny Pribaikalya. Trudy Baikalskoi Limnologicheskoi stanzii10, 425–450.

Martinson, G.G., 1948. Iskopaemye gubki iz Tunkinskoi kotloviny vPribaikalye. Doklady Akademii Nauk SSSR 61, 897–900.

Medina, M., Collins, G.A., Silberman, J.D., Sogin, M.L., 2001. Evalu-ating hypothesis of basal animal phylogeny using complete sequencesof large and small subunit rRNA. Proc. Natl. Acad. Sci. USA 98,9707–9712.

Medlin, L.K., Elwood, H.J., Stickel, S., Sogin, M.L., 1988. The charac-terization of enzymatically amplified eukaryotic 16S-like rRNA-codingregions. Gene 71, 491–499.

Muller, W.E.G., Schroder, H.C., Wrede, P., Kaluzhnaya, O.V., Belikov,S.I., 2006. Speciation of sponges in Baikal-Tuva region: an outline. J.Zool. Syst. Evol. Res. 44, 105–117.

Nichols, S.A., 2005. An evaluation of support for order-level monophylyand interrelationships within the class Demospongiae using partialdata from the large subunit rDNA and cytochrome oxidase subunit I.Mol. Phylogenet. Evol. 34, 81–96.

Outcalt, S.I., Allen, H.L., 1981. The thermal response of Lake Ohrid,Yugoslavia, to meteorological conditions. Arch. Hydrobiol. 91, 181–191.

Peterson, K.J., Addis, J.S., 2000. Clypeatula cooperensis gen. n., sp n., anew freshwater sponge (Porifera, Spongillidae) from the RockyMountains of Montana, USA. Zool. Scr. 29, 265–274.

Peterson, K.J., Butterfield, N.J., 2005. Origin of the Eumetazoa: testingecological predictions of molecular clocks against the Proterozoicfossil record. Proc. Natl. Acad. Sci. USA 102, 9547–9552.

Pop, A.A., Wink, M., Pop, V.V., 2003. Use of 18S, 16SW rDNA andcytochrome c sequences in earthworm taxonomy (Oligochaeta, Lum-bricidae). Pedobiologia 47, 428–433.

Redmond, N.E., van Soest, R.W., Kelly, M., Raleigh, J., Travers,S.A., McCormack, G.P., 2007. Reassessment of the classificationof the Order Haplosclerida (Class Demospongiae, Phylum Porif-era) using 18S rRNA gene sequence data. Mol. Phylogenet. Evol.43, 344–352.

Rezvoi, P.D., 1936. Freshwater sponges of the USSR. In: Rezvoi, P.D.(Ed.), The Fauna of the USSR. Acad. Sci., Moscow, Leningrad, pp.1–42.

Richelle-Maurer, E., Boury-Esnault, N., Itskovich, V.B., Manuel, M.,Pomponi, S.A., Van de Vyver, G., Borchiellini, C., 2006. Conservationand phylogeny of a novel family of non-Hox genes of the Antp class inDemospongiae (Porifera). J. Mol. Evol. 63, 222–230.

Sarasin, P., Sarasin, F., 1905. Reisen in Celebes ausgefuhrt in den Jahren1893–1896 und 1902–1903, Kreidel, Wiesbaden.

Schmidt, H.A., Strimmer, K., Vingron, M., von Haeseler, A., 2002.TREE-PUZZLE: maximum likelihood phylogenetic analysis usingquartets and parallel computing. Bioinformatics 18, 502–504.

Schroder, H.C., Efremova, S.M., Itskovich, V.B., Belikov, S., Masuda, Y.,Krasko, A., Muller, I.M., Muller, W.E.G., 2003. Molecular phylogenyof the freshwater sponges in Lake Baikal. J. Zool. Syst. Evol. Res. 41,80–86.

Spirkovski, Z., Avramovski, O., Kodzoman, A., 2001. Watershed man-agement in the Lake Ohrid region of Albania and Macedonia. LakesReserv.: Res. Manage. 6, 237–242.

Stankovic, S., 1960. The Balkan Lake Ohrid and its living world. DenHaag W. Junk, Den Haag.

Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W:improving the sensitivity of progressive multiple sequence alignmentthrough sequence weighting, positions-specific gap penalties andweight matrix choice. Nucl. Acids Res. 22, 4673–4680.

Volkmer-Ribeiro, C., Reitner, J., 1991. Renewed study of the typematerial of Palaeospongilla chubutensis Ott and Volksheimer (1972).In: Reitner, J., Keupp, H. (Eds.), Fossil and Recent Sponges. Springer,Berlin, pp. 121–133.

Weltner, W., 1901. Susswasserspongien von Celebes (SpongillidenstudienIV). Arch. f. Naturgesch. Beiheft 1901, 187–204.

Worheide, G., 2006. Low variation in partial cytochrome oxidase subunitI (COI) mitochondrial sequences in the coralline demosponge Astro-

sclera willeyana across the Indo-Pacific. Marine Biol. 148, 907–912.

phylogenetic analysis of freshwater sponges provide evidence for

Documents