Hidden Biodiversity of the Extremophilic Cyanidiales Red Algae

Molecular Ecology (2004) 13, 1827–1838 doi: 10.1111/j.1365-294X.2004.02180.x

BlackwellHidden Publishing, Ltd. biodiversity of the extremophilic Cyanidiales red algae

CLAUDIA CINIGLIA,*‡ HWAN SU YOON,†‡ ANTONINO POLLIO,* GABRIELE PINTO* and DEBASHISH BHATTACHARYA† *Dipartimento di Biologia vegetale, Università ‘Federico II’, via Foria 223, 80139 Napoli, Italy, †Department of Biological Sciences and Center for Comparative Genomics, University of Iowa, 210 Biology Building, Iowa City, Iowa 52242, USA

Abstract The Cyanidiales is a group of asexual, unicellular red algae, which thrive in acidic and high temperature conditions around hot springs. These unicellular taxa have a relatively simple morphology and are currently classified into three genera, Cyanidium, Cyanidioschyzon and Galdieria. Little is known, however, about the biodiversity of Cyanidiales, their population structure and their phylogenetic relationships. Here we used a taxonomically broadly sampled three-gene data set of plastid sequences to infer a robust phylogenetic framework for the Cyanidiales. The phylogenetic analyses support the existence of at least four distinct Cyanidiales lineages: the Galdieria spp. lineage (excluding Galdieria maxima), the Cyanidium caldarium lineage, a novel monophyletic lineage of mesophilic Cyanidium spp. and the Cyanidioschyzon merolae plus Galdieria maxima lineage. Our analyses do not support the notion of a mesophilic ancestry of the Cyanidiales and suggest that these algae were ancestrally thermo-acidotolerant. We also used environmental polymerase chain reaction (PCR) for the rbcL gene to sample Cyanidiales biodiversity at five ecologically distinct sites at Pisciarelli in the Phlegrean Fields in Italy. This analysis showed a high level of sequence divergence among Cyanidiales species and the partitioning of taxa based on environmental conditions. Our research revealed an unexpected level of genetic diversity among Cyanidiales that revises current thinking about the phylogeny and biodiversity of this group. We predict that future environmental PCR studies will significantly augment known biodiversity that we have discovered and demonstrate the Cyanidiales to be a species-rich branch of red algal evolution. Keywords: biodiversity, Cyanidiales, extremophile, phylogeny, plastid genes, red algae Received 25 November 2003; revision received 13 February 2004; accepted 13 February 2004

recently at hot springs, in an acidic river, and at a deep Introduction ocean site (DeLong & Pace 2001; Lopez-Garcia et al. 2001; Extremophiles are organisms that thrive at an extremely Moon-van der Staay et al. 2001; Amaral Zettler et al. 2002). high or low pH (e.g. < 3), temperature (e.g. > 50 °C), salinity, Most thermo-acidophiles are prokaryotes (i.e. Archea or desiccation and pressure (Rothschild & Mancinelli 2001), Bacteria) with one notable exception, the Cyanidiales. relying on specialized enzymes for survival (Hough & Cyanidiales is a group of asexual, unicellular red algae, Danson 1999). These enzymes have great potential for which thrive in acidic (pH 0.5–3.0) and high temperature biotechnological and pharmaceutical applications [e.g. Taq (50–55 °C) conditions around hot springs and/or acidic polymerase (Brock 1997)]. Despite their proclivity for hostile sulphur fumes (Pinto et al. 2003). Cyanidiales have a environments, extremophiles are a highly diverse group relatively simple morphology (see Fig. 1A) consisting with an abundance of novel taxa having been discovered of spherical thick-walled cells containing one plastid (i.e. chloroplast), 1–3 mitochondria, a nucleus, a vacuole Correspondence: Debashish Bhattacharya. Fax: (319) 335 1069; and storage products (Merola et al. 1981; Sentsova 1991; Ott E-mail: [email protected] & Seckbach 1994; Albertano et al. 2000; Pinto et al. 2003). ‡These authors contributed equally to this work. Three genera (Cyanidium, Cyanidioschyzon and Galdieria)

1828 C. CINIGLIA ET AL.

Fig. 1 The Cyanidiales red algae. (A) TEM micrograph of a crypto-endolithic strain of Galdieria sulphuraria. The abbreviations denote the following: m = mitochondrion, n = nucleus, p = plastid, v = vacuole. Scale bars = 1 µm. (B) The environmental Sites A–E used to collect Cyanidiales at Pisciarelli in the Phlegrean Fields, Italy (see text for details).

and six species are presently recognized in this order based Environmental polymerase chain reaction (PCR) (i.e. on morphological characters such as cell shape, number and cultivation-independent surveys) has been used effectively shape of plastids, characters of the cell wall, presence–absence to assess species biodiversity in the field (DeLong & Pace of vacuoles, the pattern of cell, division and the number 2001; Lopez-Garcia et al. 2001; Moon-van der Staay et al. of autospores in sporangia. However, species delimitation 2001; Amaral Zettler et al. 2002). In this study, we recon- remains unclear due to a paucity of diagnostic character structed the phylogeny of the Cyanidiales in trees that sets for each taxon (Albertano et al. 2000; Merola et al. 1981; included all known genera and species using a concate- Sentsova 1991; Ott & Seckbach 1994; Pinto et al. 2003). nated set of three plastid genes (psaA, psbA, and rbcL). In Molecular phylogenetic studies suggest that the Cyanid- addition, we collected environmental samples and analysed iales represent one of the most ancient groups of algae, strains isolated from different locations worldwide, as well having diverged about 1.3 billion years ago at the base as employing more extensive sampling from different of the Rhodophyta (Müller et al. 2001; Yoon et al. 2002b). sites and habitats (i.e. both extremophilic and mesophilic) Given their long evolutionary history, it is intriguing that in Italy to assess the diversity of the Cyanidiales. Phyloge- only a handful of recognized morphological species have netic analyses of rbcL sequences from the environmental survived in this lineage. Three possible explanations for samples revealed an unanticipated level of genetic diversity this observation are (1) the Cyanidiales have always been in the Cyanidiales and helped us to define the major species poor, perhaps similar to other ancient algal groups evolutionary lineages in this group. There was also a such as the Glaucophyta (Helmchen et al. 1995); (2) many striking pattern of intrapopulation structure that suggests Cyanidiales lineages have diverged over evolutionary environmental conditions play a major role in partitioning time, but only a few have survived recurrent extinctions or Cyanidiales genotypes. a recent extinction event (there is, however, no fossil record for this group to test this idea); or (3) we grossly underes- Materials and methods timate the genetic diversity of Cyanidiales due to a reliance on a limited character set that may reflect strong selection Sampling of cultured isolates and DNA-sequencing against morphological variation rather than genetic homo- geneity [e.g. as in Bangia spp. (Butterfield 2000)]. Recent The sources and GenBank Accession nos of the sequences analyses (Gross et al. 2001; Pinto et al. 2003) show limited that were determined in this study are summarized in support for the third hypothesis, although no detailed Table 1. Twenty-seven existing cultures that have been study has yet been published on Cyanidiales biodiversity. gathered from around the world and deposited in the

CYANIDIALES BIODIVERSITY 1829

Table 1 Sample information and GenBank Accession nos for taxa included in the phylogenetic analyses. The Accession nos of sequences determined in this study are shown in bold text

Taxa Source psaA psbA rbcL

Bangiales Bangia atropurpurea SAG 33.94 AY119698 AY119734 AY119770 Bangia fuscopurpurea SAG 59.81 AY119699 AY119735 AY119771 Porphyra purpurea GenBank NC_000925 NC_000925 NC_000925 Compsopogonales Compsopogon coeruleus SAG 36.94 AY119701 AY119737 AF087116 Cyanidiales Cyanidioschyzon merolae DBV 201 JAVA AY119693 AY119729 AY119765 DBV 001 NAPS AY119694 AY119730 AY119766 DBV 202 NAMN — — AY541296 Cyanidium caldarium RK1 NC_001840 NC_001840 NC_001840 DBV 019 SIPE AY541281 AY541289 AY541297 DBV 182 JAVA AY541282 AY541290 AY541298 DBV 020 APAS — — AY541299 C. sp. — Monte Rotaro Monte Rotaro AY391362 AY391365 AY391368 Monte Rotaro 19 — — AY541300 Monte Rotaro 20 — — AY541301 C. sp. — Sybil Sybil AY391363 AY391366 AY391369 Galdieria daedala IPPAS P508 AY541283 AY541291 AY541302 Galdieria maxima IPPAS P507 AY391364 AY391367 AY391370 Galdieria partita IPPAS P500 AY541284 AY541292 AB18008 Galdieria sulphuraria-A SAG 108.79 AY119695 AY119731 AY119767 UTEX 2393 AY541285 X52758 AF233069 DBV 011 CEMD AY541286 AY541293 AY541303 DBV 018 CNASC AY541287 AY541294 AY541304 DBV 015 NAFG — — AY541305 DBV 017 NASF — — AY541306 DBV 021 MEVU — — AY541307 DBV 074 JAVA — — AY541308 DBV 135 AZUF — — AY541309 Galdieria sulphuraria-B DBV 009 VTNE AY119696 AY119732 AY119768 DBV 012 BNTE AY541288 AY541295 AY541310 DBV 063 AGCS AY119697 AY119733 AY119769 DBV 002 NAPS — — AY541311 Environmental sample Pisciarelli-A1 — — AY541312 Pisciarelli-A12 — — AY541313 Pisciarelli-B15 — — AY541314 Pisciarelli-B19 — — AY541315 Pisciarelli-B20 — — AY541316 Pisciarelli-C1 — — AY541317 Pisciarelli-C2 — — AY541318 Pisciarelli-C16 — — AY541319 Pisciarelli-D1 — — AY541320 Pisciarelli-D5 — — AY541321 Pisciarelli-D15 — — AY541322 Pisciarelli-E10 — — AY541323 Pisciarelli-E11 — — AY541324 Pisciarelli-E12 — — AY541325 Porphyridiales Bangiopsis subsimplex PR21 AY119700 AY119736 AY119772 Dixoniella grisea SAG 39.94 AY119702 AY119738 AY119773 Flintiella sanguinaria SAG 40.94 AY119704 AY119740 AY119774 Porphyridium aerugineum SAG 1380–2 AY119705 AY119741 AY119775 Rhodella violacea SAG 115.79 AY119706 AY119742 AY119776 Rhodosorus marinus SAG 116.79 AY119708 AY119744 AY119778 Stylonema alsidii SAG 2.94 AY119709 AY119745 AY119779

1830 C. CINIGLIA ET AL.

Table 1 Continued

Taxa Source psaA psbA rbcL

Rhodochaetales Rhodochaete parvula UTEX LB 2715 AY119707 AY119743 AY119777 Florideophycidae Chondrus crispus Nova Scotia, Canada AY119710 AY119746 U02984 Palmaria palmata Maine, USA AY119711 U28165 U28421 Thorea violacea SAG 51.94 AY119712 AY119747 U28421 Chlorophyta Mesostigma viride GenBank NC_002186 NC_002186 NA Nephroselmis olivacea GenBank NC_000927 NC_000927 NA Glaucophyta Cyanophora paradoxa GenBank NC_001675 NC_001675 NA

NA: the rbcL gene of the green and glaucophyte algae are of a cyanobacterial origin, whereas those in the red algae and red algal-derived plastids are of proteobacterial origin [e.g. Valentin, Zetsche (1990)].

Culture Collection of the Dipartimento di Biologia vegetale, Field sampling and environmental PCR Naples (DBV, Italy), the Culture Collection of Microalgae of the Institute of Plant Physiology, Saint Petersburg [IPPAS, We sampled biomats of Cyanidiales from five different Russia (http://wdcm.nig.ac.jp/CCINFO/CCINFO.xml? habitats at Pisciarelli (Naples, Italy), which is a hydrothermal 596)] and the Sammlung von Algenkulturen, Pflanzenphy- area located in the central part of the Quaternary volcanic siologisches Institut der Universität Göttingen [SAG, complex of the Phlegrean Fields. These sites had distinct Germany (http://wwwuser.gwdg.de/∼epsag/phykologia/ environmental regimes and were named Sites A–E (see epsag.html)] were used in the study. We have included Fig. 1B). Site A had a population of Cyanidiales growing three isolates of Cyanidium caldarium and two represent- on the rocks surrounding a hot sulphur spring where the atives of Cyanidioschyzon merolae, from both Italian and Asian sulphur steams were intense and the temperature ranged sites, two mesophilic strains of Cyanidium sp. collected from 45 to 55 °C. Site B was a dry, crypto-endolithic site in in two Italian caves (Sybil cave at Cumar, Naples and which the biomat grew inside the layers of the alunite opal Monte Rotaro, Ischia), and 10 strains of Galdieria collected rocks and the temperature ranged from 18 to 30 °C. Site from different sites around the world. These latter isolates C was a relatively dry, interlithic site in which the Cyani- comprise three Russian species (Galdieria maxima, Galdieria diales grew within a fissure that was enveloped in sulphur daedala, and Galdieria partita) with the remainder being emissions and was covered by a thin and crumbly sulphate strains of Galdieria sulphuraria. crystal layer. Site D was a humid but low temperature Algal tissue was ground with glass beads by using a region where the Cyanidiales lived on muddy soil flanking Mini-BeadBeater (Biospec Products, Inc., Bartlesville, OK, the stream that flows from the hot spring. The temperature USA). Total genomic DNA was extracted using the DNeasy gradually decreased at Site D, reaching around 25–40 °C. Plant Mini Kit (Qiagen, Santa Clarita, CA, USA). PCR were Finally, at Site E, the biomat grew on hot rocks lying on the conducted using specific primers for the plastid genes psaA soil behind the hot springs. and psbA (Yoon et al. 2002a; 2002b). Three novel degenerate In each case, environmental material was collected with primers were used to amplify the rbcL gene: rC475F; 5′- a spatula near the middle of the particular site. The algal AAAACTTTCCAAGGRCCWGC-3′, rC910r; 5′-TTWCCT- material was transported to the laboratory in 50 mL Falcon GCTCTRTGTAARTG-3′, rCR; 5′-GCWGTTGGTGTYTC tubes. The Cyanidiales cells are extremely stable and did HACWAAATC-3′. These primers were designed on the not require any further handling prior to DNA extraction. basis of sequence comparisons of the plastid gene in Between 10 and 20 mg of material was used for each DNA Cyanidium and in other red algae. The PCR products were preparation as described above. For Site B, the surface rock purified with the QIAquick PCR purification kit (Qiagen) was first removed and the crypto-endolithic cells were and used for direct sequencing using the BigDyeTM collected. After genomic DNA was extracted from the cells Terminator Cycle Sequencing Kit (PE-Applied Biosystems, in these natural populations, the rbcL gene was amplified Norwalk, CT, USA), and an ABI-3100 at the Center for using PCR. The amplification products were cloned into Comparative Genomics at the University of Iowa. the pGEM-T vector (Promega). A total of 90 clones were

© 2004 Blackwell Publishing Ltd, Molecular Ecology, 13, 1827–1838 CYANIDIALES BIODIVERSITY 1831 picked (up to 20 per site) and characterized by the patterns bootstrap replicates (Felsenstein 1985) were analysed using produced by restriction enzyme digests (i.e. using, DraII the ME-LogDet and ME-GTR approaches. In the Bayesian and RsaI). Three isolates from the DBV collection that inference of the DNA data (mrbayes version 3.0b4, represented the known genera (Cyanidium, Cyanidioschyzon Huelsenbeck & Ronquist 2001), we used the site-specific and Galdieria) of Cyanidiales were used as positive con- GTR + Γ (ssGTR) model with separate model parameter trols. One to two clones representing each distinct banding estimates for the three data partitions (i.e. 1st, 2nd and pattern were sequenced from these environmental 3rd codon positions). Metropolis-coupled Markov chain samples and added to the existing rbcL data set. All cultured Monte Carlo (MCMCMC) from a random starting tree was Cyanidiales from the DBV collection in Naples are available initiated in the Bayesian inference and run for 2000 000 from G. P. Although we do not have vouchers of each generations. Trees were sampled each 1000 cycles. Four chains environmental sample used in this work, additional collec- were run simultaneously of which three were heated and tions from these sites are also available upon request one was cold, with the initial 200 000 cycles (200 trees) from G. P. being discarded as the ‘burn-in’. Stationarity of the log likelihoods was monitored to verify convergence by 200 000 cycles (results not shown). A consensus tree was made Phylogenetic analyses with the remaining 1800 phylogenies to determine the Sequences were aligned manually using SeqPup (Gilbert posterior probabilities at the different nodes. For the analysis 1995). The alignment used in the phylogenetic analyses is of rbcL, the same settings were implemented in the available upon request from D. B. We analysed two different Bayesian inference as described above, except for the use data sets. In the first analysis, we concatenated partial of a two-partition evolutionary model (i.e. 1st and 2nd sequences of the three plastid genes psaA (1395 nt), psbA codon positions). (957 nt) and rbcL (1215 nt) found in 17 Cyanidiales, 15 non- In addition to the DNA analyses, phylogenetic inference Cyanidiales red algae, two green algae and a glaucophyte as using protein data was also done for the three-gene and the outgroup. In the second analysis of the rbcL data set, we rbcL data sets. An optimal ML tree (ML-protein) was inferred included 27 representatives of the Cyanidiales from the in each case using ‘proml’ (phylip version 3.6, Felsenstein cultured strains and GenBank of a worldwide distribution 2002) and the JTT evolutionary model with 10 random- (three American, four Asian and three Russian) and 17 sequence version additions and global rearrangements. local populations from Italy. We added 14 environmental Bootstrap analysis was performed with these data using samples from Pisciarelli. We used non-Cyanidiales red the JTT model (2000 replicates for the three-protein data algae as the outgroup in this analysis because the rbcL gene set, 100 replicates for the rbcL data set) as described above. of the green and glaucophyte algae are of cyanobacterial To assess the stability of the monophyletic groups iden- origin, whereas those in the red algae and red algal-derived tified in the three-gene and rbcL plastid trees, we generated plastids are of proteobacterial origin (Valentin & Zetsche alternate trees in which critical branches in the ME-LogDet 1990). We excluded third codon positions from the rbcL tree were rearranged and compared to the likelihood (under data set to reduce the possible misleading effects of muta- the Tamura–Nei + Γ model) of the ‘best’ tree (Tamura & tional saturation in the DNA sequences (for details, see Nei 1993). Phylogenetic support for these trees was assessed Pinto et al. 2003). using the one-sided Kishino–Hasegawa test (Kishino & Phylogenetic trees were inferred with minimum evolu- Hasegawa 1989; Goldman et al. 2000) implemented in tree- tion (ME), Bayesian inference and maximum likelihood puzzle version 5.1 (Schmidt et al. 2002). (ML) methods. In the ME analyses of the three-gene data set, we generated distance matrices using LogDet transfor- Results mation and the general time reversible model (Rodriguez et al. 1990) with estimations of nucleotide frequencies, the Phylogeny of the concatenated three-gene data set shape parameter of the gamma distribution to accommo- date rate variations across sites and the proportion of A total of 3567 nt and 1189 aa from the three plastid genes invariant sites (GTR + I + Γ model) with the paup* compu- psaA, psbA, and rbcL in 17 representatives of the Cyanidiales ter program (Swofford 2002). The parameter estimates (for a total of 45 novel sequences) and representatives of for the GTR + Γ + I model were estimated using paup* and the major lineages of Bangiophycidae red algae (Müller a starting ME tree was built with HKY-85 distances. Ten et al. 2001; Yoon et al. 2002b) were used to infer the phylo- heuristic searches with random-addition-sequence start- genetic relationships of the Cyanidiales. The ME-LogDet ing trees and tree-bisection–reconnection (TBR) branch tree of the concatenated sequences shows strong support rearrangements were performed to find the optimal ME for the monophyly of the Cyanidiales, with these taxa trees. Best scoring trees were held at each step. To test the forming a sister group to the rest of the red algae (ME-LogDet stability of monophyletic groups in the ME analyses, 2000 bootstrap = 100%, ML-protein bootstrap = 91% and

Fig. 2 Phylogeny of the Cyanidiales inferred from minimum evolution (ME) analysis using the LogDet transformation of the combined plastid DNA sequences of psaA, psbA, and rbcL. Results of a ME-LogDet bootstrap analysis are shown above the branches, whereas the bootstrap values from a protein maximum likelihood analysis using the JTT evolutionary model are shown below the branches. Only bootstrap values > 60% are shown. The thick nodes represent > 95% Bayesian posterior probability for clades using the site-specific GTR model.

posterior probability = 1.0, Fig. 2). Within the Cyanidiales, consistent with the ME-LogDet tree shown in Fig. 2, except four lineages are resolved with strong support. The first is for the branching order of the non-Cyanidiales red algae the Galdieria spp. lineage (excluding Galdieria maxima), which (trees not shown). consists of two subclades. The second lineage is composed of the Cyanidium caldarium strains, the third is a new clade rbcL phylogeny formed by mesophilic Cyanidium sp., and the fourth includes Cyanidioschyzon merolae, which groups with Galdieria maxima The ME-LogDet tree of rbcL sequences was inferred with strong bootstrap support. The interrelationship from a data set of 810 nt (excluding 3rd codon positions) of these four distinct lineages is poorly resolved in Fig. 2, and 405 aa from 41 Cyanidiales, including 14 representatives with only the early branching of the Galdieria lineage having of environmental samples. This tree (Fig. 3) shows a very moderate bootstrap support (ME-LogDet = 78%, ML-protein similar topology to the three-gene tree. The four groups are = 64%). The ME-GTR and ML-protein trees are generally again resolved. The Galdieria lineage is demarcated clearly

Fig. 3 Phylogeny of the Cyanidiales inferred from a minimum evolution (ME) analysis using the LogDet transformation of the rbcL sequences. Results of a ME-LogDet bootstrap analysis are shown above the branches, whereas the bootstrap values from a protein maximum likelihood analysis using the JTT evolutionary model are shown below the branches. Only bootstrap values > 60% are shown. The thick nodes represent > 95% Bayesian posterior probability for clades using the site-specific GTR model. Sequences from the environmental survey are represented by ‘Pisciarelli-[followed by] collection site’ and are marked with asterisks.

into two subgroups, Galdieria-A and Galdieria-B. The repre- to Italy. The Cyanidium caldarium sequences and the sentatives of Galdieria-A have a worldwide distribution mesophilic-Cyanidium sp. lineages maintain a sister-group (i.e. Italy, United States, Mexico, Indonesia, and Russia), relationship [now with more support (LogDet = 85%, whereas members of the Galdieria-B lineage are restricted posterior probability = 0.95)] with the addition of more

(1) Galdieria-A and Galdieria-B (excluding Galdieria maxima); (2) Cyanidium caldarium; (3) mesophilic Cyanidium spp.; and (4) Cyanidioschyzon merolae and Galdieria maxima (see Figs 2 and 3). Morphology is normally a valuable tool in systematics; however, in the case of the unicellular Cyani- diales, it appears to not be sufficiently informative to establish a robust taxonomy (e.g. Pinto et al. 2003). Below we discuss in more detail each of the four lineages we have identified in our study, their constituent taxa and the environmental conditions in which they were found.

Galdieria-A and -B lineages The systematics of the genus Galdieria is yet to be defined clearly. A recent study that utilized a combined morphological and physiological approach revealed substantial Fig. 4 Results of the environmental survey showing the distribution of Cyanidiales species/lineages clones with respect to complexity in this genus, with critical characters such as the eco-physiological conditions at Pisciarelli, Italy. Up to 20 cell size and number and shape of plastids providing no clones were sampled from each site. In this figure, G-A is Galdieria- objective basis for discriminating among the four species A, G-B is Galdieria-B, Cy is Cyanidium, Cz is Cyanidioschyzon and Galdieria sulphuraria, Galdieria daedala, Galdieria partita and unid. is unidentified taxa. Galdieria maxima. In addition, high levels of rbcL sequence divergence within the genus made it difficult to refute or confirm the existing taxonomy of Galdieria (Pinto et al. 2003). sequences. Cyanidioschyzon merolae, Galdieria maxima and In this study, we have gained a much broader perspective one of the environmental samples (C16) are part of a weakly on this genus and can now separate it clearly into two supported lineage. subclades, Galdieria-A and Galdieria-B, with Galdieria maxima distantly related to these species (see Fig. 2 and below). There are, however, no distinctive morphological or Distribution of Cyanidiales in the environmental samples ultrastructural characters that distinguish these taxa The results of the analysis of environmental samples are (for details, see Pinto et al. 2003). summarized in Fig. 4. We surveyed a total of 90 clones It is intriguing that all of the samples from Site B in our from five sites. Members of the Galdieria-A lineage were environmental survey were positioned in the Galdieria-B found at Sites A (2/20 clones), D (5/19 clones) and E (10/ lineage (see Figs 2 and 3). The Galdieira-B population at Site 13 clones), all of which were relatively humid conditions, B grows beneath the surface of rocks (see Fig. 1B), where whereas members of Galdieria-B were found only at Sites B there is presumably a moderate humidity. Eight sequences (19/19) and C (8/20), which are relatively dry. It is interesting from Site C were also positioned in the Galdieria-B lineage. that all the representatives of Site B, which is a crypto- The Site C biomat grows in a relatively dry area between endolithic habitat, was only of the Galdieria-B type. Cyanidio- the fissures of rocks that are exposed to sulphur emissions. schyzon merolae coexisted with Galdieria-A (i.e. Sites A, D These results suggest that Galdieria-B may be adapted and E) and with Cyanidium caldarium at Site C (11/20). to dry conditions. Although we have no direct evidence for this hypothesis, it is interesting that Galdieria sulphuraria DBV 002 differs markedly from studied members of the Discussion Galdieria-A clade (e.g. DBV 074, SAG 107–79 and SAG 107– 89), in that it can exist at extreme salt concentrations (8– Biodiversity of the extremophilic Cyanidiales 10% NaCl) and is unable to utilize nitrate (Pinto et al. 2003). Our study was aimed primarily at assessing the extent Perhaps these traits are adaptations to its crypto-endolithic of biodiversity of Cyanidiales at different environmental habitat. In general, ecophysiological studies show that habitats in the Phlegrean Fields and to incorporate this desiccation tolerance is a restrictive factor for the distribution assessment of biodiversity into a broadly sampled and of the Cyanidiales (Gross 1999). Taken together, our results robust phylogenetic framework. As a result of our approach, suggest that the Galdieria-B lineage may be adapted speci- we have identified four well-supported lineages of Cyani- fically to dry environments and dominates this ecophysio- diales in addition to establishing the phylogenetic positions logical niche in the crypto-endolithic site. Unlike the of the three traditionally recognized genera Cyanidium, Galdieria-B lineages, the Galdieria-A strains were collected Cyanidioschyzon, and Galdieria. The four lineages are at the hydrothermal areas (e.g. hot springs) and established

© 2004 Blackwell Publishing Ltd, Molecular Ecology, 13, 1827–1838 CYANIDIALES BIODIVERSITY 1835 in culture collections [i.e. DBV, IPPAS, SAG, UTEX (http:// mesophilic isolates were reported with morphological www.bio.utexas.edu/research/utex/)]. The finding of and ecophysiological descriptions (Hoffmann 1994), their Galdieria-B at only the Italian sites should not necessarily systematic position remained unknown. Our study is the be taken to indicate an endemic species, because we may first to demonstrate the phylogenetic relationship of meso- find members of this group if we sample dry thermophilic philic Cyanidium sp. within the Cyanidiales. The concaten- environments in other regions of the world. ated three-gene and rbcL trees both support a sister An interesting feature of the Galdieria-A lineage is the relationship of the mesophilic and extremophilic (Cyanidium clear sorting of the lineages based on geographical origin. caldarium) Cyanidium lineages and their distinct phylo- The taxa collected from Italy, the United States, Mexico, genetic position with regard to the acido-thermophilic Russia (except Galdieria maxima, but see below) and Indo- Galdieria-A and -B lineages (see Figs 2 and 3). This result is, nesia each form a distinct monophyletic clade in Fig. 3. however, without ML bootstrap support in the three-gene This suggests that long-distance dispersal may be limited tree (Fig. 2). in Galdieria-A, or alternatively that our sampling was not In another study from our laboratory (Yoon et al. 2004), complete enough at the different sites to discover more the 16S rRNA and the first and second codon positions rare migrants. Support for the first idea comes from the of five plastid genes (psaA, psaB, psbA, rbcL and tufA) were intriguing distribution pattern of the Cyanidiales. These analysed in a combined data set of 5177 nt. This tree included algae are scattered throughout the world; however, their seven Cyanidiales representing the four lineages identified habitat is locally restricted to thermal and acidic sites. here plus 14 non-Cyanidiales red algae, chromists, green Geographic isolation may therefore be expected among these and glaucophyte algae for a taxonomically broad sample taxa due to their inability to undergo long-distance migra- of 46 plastid genomes. The trees inferred from this data tion caused by reduced tolerance to desiccation and set robustly support (bootstrap with ME-GTR + I + Γ = 100% mesophilic conditions, such as low temperature and neutral and unweighted maximum parsimony = 97%, posterior pH (Brock 1978; Gross 1999). Combined with the apparent probability = 1.0) the monophyly of the mesophilic Cyanid- absence of resting spores in Cyanidiales, members of ium spp. and the existence of a super-clade comprised of Galdieria-A may undergo rapid interpopulation divergence Cyanidium caldarium RK1 and Cyanidioschyzon merolae DBV culminating in some cases in speciation events. This would, 201 + Galdieria maxima. The Galdieria lineage is therefore the of course, not hold for the mesophilic Cyanidium sp. (see earliest divergence in the Cyanidiales in this analysis. This below), about which we currently know very little, both latter result is also found in Figs 2 and 3, albeit with poorer with regard to phylogenetic breadth and distribution at support. We describe in some detail the results of the other mesophilic sites. For the remaining acido-thermophilic six-gene analysis because the phylogenetic position of the taxa (i.e. Cyanidium caldarium, Cyanidioschyzon merolae, mesophilic Cyanidium spp. is a critical issue in Cyanidiales Galdieria-B and other species in this genus), the prevailing evolution. This group potentially holds the key for identi- view is that these cells are dispersed infrequently via wind, fying the ancestral condition of both the Cyanidiales and of streams and oceanic currents (see Discussion in Brock 1978; all red algae. The results presented here, and in particular Gross et al. 2001). to those of Yoon et al. (2004) resolve this issue by providing support for the idea that mesophily is a derived character in the ancestrally thermo-acidophilic Cyanidiales. Cyanidium caldarium and the mesophilic lineages To assess alternative hypotheses about Cyanidiales The Cyanidium caldarium lineage forms a strongly supported evolution, we rearranged the position of the Cyanidiales group that is surprisingly broadly distributed. In stark lineages relative to each other in the three-gene and rbcL contrast to Galdieria-A, Cyanidium caldarium includes closely ME-LogDet trees (Figs 2 and 3) and in the six-gene tree (tree related isolates from Indonesia (DBV 182, Java), Italy (DBV of highest likelihood in the Bayesian posterior distribution) 019, Siena; DBV 020, Acqua Santa; C2, Pisciarelli) and in Yoon et al. (2004) and compared the log-likelihoods Japan (RK1), suggesting that these taxa have an effective using the one-sided KH-test. In these rearrangements, the dispersal mechanism(s). Another intriguing result of our mesophilic taxa were moved to the branch uniting all analysis of Cyanidium spp. is the identification of a novel the red algae, the Cyanidiales or the Galdieria lineage (see monophyletic lineage of mesophilic taxa. We collected these Figs 2 and 3). Results of the KH-test for these rearrange- mesophilic strains from the Sybil cave at Cuma and from ments are summarized in Table 2 and show that positioning Monte Rotaro, where the habitats are nonacidic (pH 7.0– the mesophilic Cyanidium spp. at the base of the red algae is 7.2) and nonthermal. The mesophilic Cyanidium sp. grows rejected with the three-gene and six-gene data sets, whereas on the bottom of the walls on the shaded side of the Sybil a basal position in the Cyanidiales is strongly rejected by cave. Mesophilic Cyanidiales were first described by the latter data set and has a low probability (albeit nonsig- Schwabe as a new algal species, C. chilense, from a cave on nificant) in the three-gene data set (P = 0.074). Movement the Chilean coast (Schwabe 1936). Even though some of the of the mesophilic clade to the base of the Galdieria lineage

Table 2 Results of the one-sided KH-test of ∆ Rearrangement Tree log-like. KH-prob. rearranged trees inferred from the plastid three-gene, rbcL and six-gene data sets. Meso., ME-LogDet Tree (Fig. 2) Three-gene Best 1.000 Galdi., Cyanid., G. max. and Russian are meso- Meso. >> base of all reds Three-gene 79.25 < 0.001** philic Cyanidium spp., the Galdieria-A and -B Meso. >> base of Cyanid. Three-gene 21.57 0.074 clades, Cyanidiales, Galdieria maxima, and the Meso. >> base of Galdi. Three-gene 21.23 0.065 Russian isolates of Galdieria daedala and G. max. >> Russian Three-gene 715.69 < 0.001** Galdieria partita, respectively. The asterisks G. max. >> base of Galdi. Three-gene 261.37 < 0.001** indicate significance at *P < 0.050 and **P < ME-LogDet Tree (Fig. 3) rbcL Best 1.000 0.010, ∆ log-like. indicates the difference Meso. >> base of all reds rbcL 3.74 0.265 in log-likelihood between the best and Meso. >> base of Cyanid. rbcL 3.74 0.272 rearranged trees and >> denotes the move- Meso. >> base of Galdi. rbcL 3.90 0.260 ment of taxa to different positions in the G. max. >> Russian rbcL 130.59 < 0.001** trees G. max. >> base of Galdi. rbcL 22.79 0.024* Best Bayesian Tree (from Yoon et al. unpubl.) Six-gene Best 1.000 Meso. >> base of all reds Six-gene 109.72 < 0.001** Meso. >> base of Cyanid. Six-gene 68.83 < 0.001** Meso. >> base of Galdi. Six-gene 70.26 < 0.001** G. max. >> Russian Six-gene 677.89 < 0.001** G. max. >> base of Galdi. Six-gene 212.37 < 0.001**

results in a significantly worse tree using the six-gene data. et al. 2001; Yoon et al. 2002b; Pinto et al. 2003), although the The rbcL data set does not significantly reject these enigmatic relationship is hard to explain using traditional rearrangements, but this presumably reflects the lower taxonomic criteria that rely on morphological and ecophy- resolving power in a single-gene analysis. Given the results siological characters. Galdieria maxima, for example, has a of the KH-test, in particular using the most informative six- larger cell dimension (10–16 mm), is spherical in shape gene data set, and the bootstrap and Bayesian results for the with one multilobed plastid, contains several vacuoles and three-gene tree (Fig. 2), we suggest that the mesophilic cell division occurs through the formation of sporangia Cyanidium spp. are not a basal divergence in either the entire with 4–16 spores, rather than by binary fission as in red algal clade or in the Cyanidiales. This is consistent with Cyanidioschyzon merolae. In addition, Galdieria maxima is the idea that mesophily is a derived condition in the ances- able to live mixotrophically (i.e. the simultaneous capacity trally thermo-acidophilic Cyanidiales. Our data do not for photosynthesis and for the uptake of particulate and/ resolve, however, the issue of whether the first red algae or dissolved organic compounds) and heterotrophically, were mesophiles or extremophiles, although we currently unlike Cyanidioschyzon merolae which is autotrophic and favour mesophily because the early divergences in the restricted to acido-thermophilic sites (Pinto et al. 2003). sister groups of the reds, the green and glaucophyte algae Results of the KH-test, however, robustly reject [P < 0.001 (e.g. Friedl et al. 2000; Moreira et al. 2000) are all mesophilic. (except for P = 0.024 with rbcL), see Table 2] the forced This hypothesis could potentially be overturned if future monophyly of the Russian Galdieria species (i.e. Galdieria biodiversity sampling shows the Cyanidiales to be para- daedala, Galdieria maxima and Galdieria partita) or the posit- phyletic with extremophilic members that branch at the ioning of Galdieria maxima at the base of the Galdieria base of the mesophilic red algae. lineage. Although our data argue convincingly for a distant phylogenetic relationship of Galdieria maxima to the other Russian Galdieria species, an alternative explanation for Cyanidioschyzon and Galdieria maxima lineages our results (until now unsubstantiated) is that the Galdieria Cyanidioschyzon merolae is distinct from Cyanidium and maxima clone may represent a contaminant in the ‘typical’ Galdieria with regard to the following morphological Galdieria maxima culture. Light microscopic analyses do characters: the cells of Cyanidioschyzon merolae have a more not, however, show obvious cell size heterogeneity in this elliptical shape, they lack cell walls or vacuoles and they IPPAS culture. Finally, one of the environmental samples divide through binary fission rather than through endo- we have identified (C16) is also a member of the Cyani- spores, as is the case for Cyanidium caldarium and Galdieria dioschyzon + Galdieria maxima lineage (Fig. 3), although we sulphuraria (Albertano et al. 2000). Our analyses showed, have no morphological data for this taxon. In summary, however, a sister group relationship between the morpho- our results potentially reveal the traditional systematic logically distinct Cyanidioschyzon merolae and Galdieria scheme to have grossly underestimated the biodiversity maxima. This result is consistent with previous reports (Gross of Cyanidiales. More extensive sampling from different

© 2004 Blackwell Publishing Ltd, Molecular Ecology, 13, 1827–1838 CYANIDIALES BIODIVERSITY 1837 natural habitats will be required to elucidate in greater mophilic sites will further broaden our understanding of detail the biodiversity and phylogeny of this group. Cyanidiales taxon diversity and phylogeny.

Environmental survey Acknowledgements

The environmental PCR and the restriction enzyme This work was supported by grants awarded to D. B. by the approach allowed us to survey efficiently the population National Science Foundation (DEB 01–07754, MCB 02–36631). structure of Cyanidiales, thereby leading to the identification This paper is dedicated respectfully to Wolfgang Gross, who was of novel lineages such as Galdieria-B and the establishment an active member of the phycology and Cyanidiales community. of previously unrecognized species relationships. In the latter case we found, for example, that Galdieria-A and References Cyanidioschyzon merolae coexist at all the humid sites (i.e. Sites A, D and E). Within the populations, Cyanidioschyzon Albertano P, Ciniglia C, Pinto G, Pollio A (2000) The taxonomic merolae was more abundant than Galdieria-A at the hydro- position of Cyanidium, Cyanidioschyzon and Galdieria: an update. Hydrobiologia 433 thermal habitats (e.g. 90% at Site A, 74% at Site D), whereas , , 137–143. Amaral Zettler LA, Gomez F, Zettler E et al. (2002) Microbiology: Galdieria-A dominated Site E (77%), where the sulphur eukaryotic diversity in Spain’s River of Fire. Nature, 417, 137. fumes arose from the rocks. In stark contrast, Galdieria-B Brock TD (1978) The genus Cyanidium. In: Thermophilic Microorgan- preferred relatively dry habitats (i.e. Sites B and C). Taken isms and Life at High Temperatures (ed. Starr PM), pp. 255–301. together, the environmental survey appears to be a useful Springer-Verlag, New York. tool for ecophysiological and phylogenetic studies. Our Brock TD (1997) The value of basic research: discovery of Thermus demonstration of significant intraspecific variation in Cyani- aquaticus and other extreme thermophiles. Genetics, 146, diales shows the potential of this technique for studying 1207–1210. Butterfield NJ (2000) Bangiomorpha pubescens n. gen., n. sp.: impli- genetic differentiation on a geographical scale probably cations for the evolution of sex, multicellularity, and the Meso- down to the level of within-population spatial structure. proterozoic/Neoproterozoic radiation of eukaryotes. Paleobiology, 26, 386–404. DeLong EF, Pace NR (2001) Environmental diversity of bacteria Conclusions and archaea. Systematic Biology, 50, 470–478. In conclusion, we provide a robust phylogeny of the Felsenstein J (1985) Confidence limits on phylogenies: an approach Evolution 39 Cyanidiales using a concatenated data set of three plastid using the bootstrap. , , 783–791. Felsenstein J (2002) PHYLIP 3.6. Distributed by the author. Depart- genes and an environmental PCR survey. We establish the ment of Genetics, University of Washington, Seattle. existence of at least four distinct Cyanidiales lineages Friedl T, Besendahl A, Pfeiffer P, Bhattacharya D (2000) The distri- that incorporate the three traditional genera. These are bution of group I introns in lichen algae suggests that licheniza- the Galdieria spp. lineage, the Cyanidium caldarium lineage, the tion facilitates intron lateral transfer. Molecular Phylogenetics and mesophilic Cyanidium spp. lineage, and the Cyanidioschyzon Evolution, 14, 342–352. merolae plus Galdieria maxima lineage. Although we have Gilbert DG (1995) SEQPUP, a Biological Sequence Editor and Analysis Program for the Macintosh Computer maintained the traditional species concepts in our analyses, . Indiana University, Bloom- ington, IN. inspection of the branch lengths that define within-lineage Goldman N, Anderson JP, Rodrigo AG (2000) Likelihood-based diversity suggest that the Cyanidiales could plausibly be tests of topologies in phylogenetics. Systematic Biology, 49, split into many more species than is currently the case. 652–670. Within-lineage (often species) sequence divergence in the Gross W (1999) Revision of comparative traits for the acido- and Cyanidiales is comparable, for example, to between-order thermophilic red algae Cyanidium and Galdieria. Kluwer Academic divergences in the non-Cyanidiales red algae [e.g. compare Publishers, London. Compsopogon coeruleus (Compsopogonales) to Rhodochaete Gross W, Heilmann I, Lenze D, Schnarrenberger C (2001) Biogeography of the Cyanidiaceae (Rhodophyta) based on 18S parvula (Rhodochaetales)]. It is possible, however, that the ribosomal RNA sequence data. European Journal of Phycology, high divergence rates in the Cyanidiales are explained by 36, 275–280. an elevated mutation rate in these taxa, resulting poten- Helmchen TA, Bhattacharya D, Melkonian M (1995) Analyses tially from DNA damage in their extreme environments. of ribosomal RNA sequences from glaucocystophyte cyanelles Our analyses also reject the notion of a mesophilic ancestry provide new insights into the evolutionary relationships of of the Cyanidiales and suggest that this lineage was plastids. Journal of Molecular Evolution, 41, 203–210. Evolu- ancestrally thermo-acidotolerant. The environmental PCR Hoffmann L (1994) Cyanidium-like algae from caves. In: tionary Pathways and Enigmatic Algae: Cyanidium Caldarium survey was a powerful approach for discovering Cyani- (Rhodophyta) and Related Cells (ed. Seckbach J), pp. 175–182. diales biodiversity and was easy to implement with these Kluwer Academic Publishers, London. terrestrial algae. We predict that future analyses using this Hough DW, Danson MJ (1999) Extremozymes. Current Opinion in approach at many other, until now poorly sampled, extre- Chemical Biology, 3, 39–46.

Huelsenbeck JP, Ronquist F (2001) mrbayes: Bayesian inference of Rothschild LJ, Mancinelli RL (2001) Life in extreme environments. phylogenetic trees. Bioinformatics, 17, 754–755. Nature, 409, 1092–1101. Kishino H, Hasegawa M (1989) Evaluation of the maximum Schmidt HA, Strimmer K, Vingron M, von Haeseler A (2002) likelihood estimate of the evolutionary tree topologies from DNA tree-puzzle: maximum likelihood phylogenetic analysis sequence data, and the branching order in Hominoidea. Journal using quartets and parallel computing. Bioinformatics, 18, of Molecular Evolution, 29, 170–179. 502–504. Lopez-Garcia P, Rodriguez-Valera F, Pedros-Alio C, Moreira D Schwabe GH (1936) Über einige Blaualgen aus dem mittleren und (2001) Unexpected diversity of small eukaryotes in deep-sea südlichen Chile. Verhandlungen des Deutschen Wissenschaftlichen Antarctic plankton. Nature, 409, 603–607. Vereins zu Santiago de Chile, 3, 113–174. Merola A, Castaldo R, Gambardella P, Musachio R, Taddei R Sentsova OY (1991) Diversity of acido-thermophilic unicellular (1981) Revision of Cyanidium caldarium: three species of acido- algae of the genus Galdieria (Rhodophyta, Cyanidiophyceae). philic alage. Giornale Botanico Italiano, 115, 189–195. Botanicheskii Zhurnal, 76, 69–79. Moon-van der Staay SY, De Wachter R, Vaulot D (2001) Oceanic Swofford DL (2002) PAUP*: Phylogenetic Analysis Using Parsimony 18S rDNA sequences from picoplankton reveal unsuspected (*and Other Methods) 4.0b8. Sinauer, Sunderland, MA. eukaryotic diversity. Nature, 409, 607–610. Tamura K, Nei M (1993) Estimation of the number of nucleotide Moreira D, Le Guyader H, Phillippe H (2000) The origin of red substitutions in the control region of mitochondrial DNA in algae and the evolution of chloroplasts. Nature, 405, 69–72. humans and chimpanzees. Molecular Biology and Evolution, 10, Müller KM, Oliveira MC, Sheath RG, Bhattacharya D (2001) Ribos- 512–526. omal DNA phylogeny of the Bangiophycidae (Rhodophyta) Valentin K, Zetsche K (1990) Rubisco genes indicate a close and the origin of secondary plastids. American Journal of Botany, phylogenetic relation between the plastids of Chromophyta and 88, 1390–1400. Rhodophyta. Plant Molecular Biology, 15, 575–584. Ott FD, Seckbach J (1994) New classification for the genus Cyanid- Yoon HS, Hackett JD, Bhattacharya D (2002a) A single origin of the ium Geitler 1933. In: Evolutionary Pathways and Enigmatic Algae: peridinin- and fucoxanthin-containing plastids in dinoflagellates Cyanidium Caldarium (Rhodophyta) and Related Cells (ed. Seckbach J), through tertiary endosymbiosis. Proceedings of the National pp. 145–152. Kluwer Academic Publishers, London. Academy of SciencesUSA, 99, 11724–11729. Pinto G, Albertano P, Ciniglia C et al. (2003) Comparative Yoon HS, Hackett JD, Ciniglia C, Pinto G, Bhattacharya D, (2004) approaches to the taxonomy of the genus Galdieria merola A molecular timeline for the origin of photosynthetic eukaryotes. (Cyanidiales, Rhodophyta). Cryptogamie Algologie, 24, 13–32. Molecular Biology and Evolution, 21, 809–818. Rodriguez F, Oliver JL, Marin A, Medina JR (1990) The general Yoon HS, Hackett JD, Pinto G, Bhattacharya D (2002b) The single, stochastic model of nucleotide substitution. Journal of Theoretical ancient origin of chromist plastids. Proceedings of the National Biology, 142, 485–501. Academy of Sciences USA, 99, 15507–15512.