Molecular Phylogenetics and Evolution 52 (2009) 329–339

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

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

Phylogenetic relationships and species circumscription in Trentepohliaand Printzina (Trentepohliales, Chlorophyta)

Fabio Rindi *, Daryl W. Lam, Juan M. López-BautistaDepartment of Biological Sciences, The University of Alabama, P.O. Box 870345, 425 Scientific Collections Bldg., Tuscaloosa, AL 35487-0345, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 28 October 2008Revised 12 January 2009Accepted 15 January 2009Available online 7 February 2009

Keywords:18S rRNA geneChlorophytaMolecular systematicsMorphological convergencePrintzinarbcL geneSubaerial algaeSubstitutional saturationTrentepohliaTrentepohlialesUlvophyceae

1055-7903/$ - see front matter � 2009 Elsevier Inc. Adoi:10.1016/j.ympev.2009.01.009

* Corresponding author. Present address: Martin Rversity of Ireland, Galway, Ireland. Fax: +353 91 5250

E-mail address: fabio.rindi@nuigalway.ie (F. Rindi)

Subaerial green microalgae represent a polyphyletic complex of organisms, whose genetic diversity ismuch higher than their simple morphologies suggest. The order Trentepohliales is the only species-richgroup of subaerial algae belonging to the class Ulvophyceae and represents an ideal model taxon to inves-tigate evolutionary patterns of these organisms. We studied phylogenetic relationships in two commongenera of Trentepohliales (Trentepohlia and Printzina) by separate and combined analyses of the rbcL and18S rRNA genes. Trentepohlia and Printzina were not resolved as monophyletic groups. Three main cladeswere recovered in all analyses, but none corresponded to any trentepohlialean genus as defined based onmorphological grounds. The rbcL and 18S rRNA datasets provided congruent phylogenetic signals andsimilar topologies were recovered in single-gene analyses. Analyses performed on the combined 2-genedataset inferred generally higher nodal support. The results clarified several taxonomic problems andshowed that the evolution of these algae has been characterized by considerable morphological conver-gence. Trentepohlia abietina and T. flava were shown to be separate species from T. aurea; Printzina lage-nifera, T. arborum and T. umbrina were resolved as polyphyletic taxa, whose vegetative morphologyappears to have evolved independently in separate lineages. Incongruence between phylogenetic rela-tionships and traditional morphological classification was demonstrated, showing that the morphologicalcharacters commonly used in the taxonomy of the Trentepohliales are phylogenetically irrelevant.

� 2009 Elsevier Inc. All rights reserved.

1. Introduction

Subaerial green microalgae are among the most widespread andevolutionarily diverse organisms inhabiting terrestrial environ-ments. Their systematics have been traditionally problematic andsurrounded by great controversy. In the past, their classificationwas entirely based on morphological characters (Smith, 1950; Ettland Gärtner, 1995). However, since the morphology of most spe-cies is unicellular and affected by great plasticity, the set of reliablecharacters useful for systematic purposes is limited. In the last15 years, DNA sequence data have become gradually availableand have led to important insights into the evolution of theseorganisms. It has become clear that terrestrial green algae are ahighly polyphyletic group originating from colonization of terres-trial environments by many separate lineages of aquatic (bothfreshwater and marine) algae (Huss et al., 1999; Lewis andMcCourt, 2004; Lewis and Lewis, 2005; Watanabe et al., 2006; Le-wis, 2007; Cardon et al., 2008). The adaptation to a terrestrial life-style has resulted into a convergent morphology in algae belongingto widely separated evolutionary lineages (López-Bautista et al.,

ll rights reserved.

yan Institute, National Uni-05..

2007). A direct consequence of this phenomenon is that the classi-fication at the genus and species level requires a reassessmentbased on a polyphasic combination of different approaches (Prösc-hold and Leliaert, 2007). At present, the limited amount of DNA se-quence data available for most groups is the main hurdle to thecompletion of such a reassessment.

The order Trentepohliales is an ideal model taxon to illustratethe complexity of the problems that affect the systematics and tax-onomy of subaerial green algae. This order includes microchloro-phytes occurring in a wide range of habitats in areas with humidand rainy climates. Although most diverse and abundant in tropicalregions, the Trentepohliales are also found in temperate zones andoccur on a large variety of substrata, such as bark, leaves and fruitsof vascular plants, wood, metal, plastic, stones and concrete(Thompson and Wujek, 1997; López-Bautista et al., 2002; Rindiand López-Bautista, 2007). They are also among the most commonphycobionts in lichens (Chapman and Good, 1983; Nakano andIhda, 1996; Chapman and Waters, 2001). When present in largeamounts, their occurrence is revealed by the red, orange or yellowcolor produced by the accumulation of carotenoids in their cells.The order contains the single family Trentepohliaceae and is distin-guished from all other green algae by the combination of the fol-lowing features: presence of b-carotene and haematochrome;absence of pyrenoids in the chloroplast; flagellar apparatus with

330 F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339

unique ultrastructure (described in detail by Chapman and Henk,1983 and Roberts, 1984); transverse cell walls with plasmodes-mata; and presence of a characteristic reproductive structure, thesporangiate lateral (a highly modified branch consisting of an api-cal cell that is swollen basally and tapers apically into a short neck,on which a spherical or oval zoosporangium is borne).

This unusual combination of characters has been a source ofgreat confusion in the classification of the Trentepohliales at alltaxonomic levels, in particular the class level. The possible inclu-sion of these algae in at least three different classes was suggestedin the past (López-Bautista and Chapman, 2003, and referencestherein), as well as the elevation to the rank of a class, the Tren-tepohliophyceae (Van den Hoek et al., 1995). López-Bautista andChapman (2003) provided robust evidence for placement in theclass Ulvophyceae based on 18S rRNA sequence data. The molecu-lar evidence currently available suggests that the Trentepohlialesform a well-supported monophyletic group sister to a clade com-posed of the order Dasycladales and the clade Cladophorales–Siphonocladales (López-Bautista and Chapman, 2003).

At the genus level, the classification of the Trentepohliales hasbeen traditionally based on the scheme proposed by Printz(1939). In this monograph, five genera (Cephaleuros, Phycopeltis,Physolinum, Stomatochroon, Trentepohlia) were recognized on mor-phological basis. Cephaleuros Kunze includes 17 species and con-sists of filaments irregularly branched, free or coalescent toproduce irregular discs that grow below the cuticle or the epider-mis of leaves of vascular plants. Species of Phycopeltis Millardetconsist of creeping filaments, usually coalescing to produce moreor less regular discs epiphytic on leaves, stems and fruits of vascu-lar plants; the genus currently includes 26 species. StomatochroonPalm includes four species with diminutive habit, growing inleaves of vascular plants as filamentous endophytes, the presenceof which is revealed by sporangiate laterals protruding out of thehost plant’s stomata. Members of Trentepohlia Martius form smalltufts or compact crusts consisting of numerous branched filamentsgrowing on a large variety of natural and artificial substrata; this isthe largest and earliest-described genus, including about 40 spe-cies. Physolinum was erected by Printz (1921) for a single species,Trentepohlia monilia De Wildeman, due to the production of apl-anospores; however, in the more recent literature the validity ofthis character has been disputed and Physolinum has not been sep-arated from Trentepohlia (Flint, 1959; Cribb, 1970; Sarma, 1986).Printzina was erected by Thompson and Wujek (1992) for 9 taxawhich were formerly included in Trentepohlia, and a newly de-scribed species, P. ampla. Shape of zoosporangia, arrangement ofthe sporangiate laterals and development of the erect part of thethallus were the main characters on which Thompson and Wujek(1992) based the separation of the two genera.

Taxonomy and systematics of Trentepohlia and Printzina havebeen traditionally among the most complicated problems in theclassification of the green algae. The evolutionary origins of theseorganisms are obscure; it is presently unknown which morpholog-ical characters bear phylogenetic significance and which characterstates should be considered primitive and derived. In general, thetaxonomy of these algae is still very unclear and in need of a de-tailed reassessment. Morphological characters which have beenused for taxonomic purposes include the following: shape and sizeof vegetative cells; relative development of the prostrate parts ver-sus the erect parts of the thallus; branching pattern; presence ofhair-like protrusions; arrangement of sporangiate laterals; shapeand size of zoosporangium and enlarged cell bearing it (‘‘suffultorycell”); arrangement of gametangia; shape of apical cells; features ofcell wall (such as color and presence of corrugations); and type ofsubstratum colonized (Hariot, 1890; De Wildeman, 1900; Printz,1921, 1939; Sarma, 1986; Ettl and Gärtner, 1995). In practice, itis not infrequent to observe a complete and uninterrupted range

of variation between different character states in the same spe-cies, sometimes in the same specimen. Some species of Tren-tepohlia and Printzina are characterized by great plasticityrelated to environmental factors, and their morphology mayvary considerably depending on the environmental conditions(Howland, 1929; Nakano and Handa, 1984; Thompson and Wu-jek, 1997; Rindi and Guiry, 2002). As in other groups of ulvo-phycean green algae (i.e., Verbruggen et al., 2007a; De Clercket al., 2008), there are extensive ‘‘grey” zones between tradi-tional species defined on morphology. Collection of specimensfor which the morphology does not correspond satisfactorilyto any of the species currently included in these genera is acommon situation in field investigations, and for this reasonspecies of Trentepohlia are often identified at the genus level,sometimes only at family or order level. Another factor thatgreatly complicates taxonomic matters is the fact that severalspecies were described by early authors, who provided only avery vague description (Linnaeus, 1753, 1759; Agardh, 1824;Kützing, 1843; Zeller, 1873); for some of them, no type speci-mens or other authentic specimens are available (i.e., Trentepoh-lia aurea and T. iolithus). For this reason, the identity of somespecies has become very confused in the course of the time,and the circumscription of several morphologically similar taxaremains to this day an unsolved matter.

DNA sequence data for the Trentepohliales have become avail-able relatively recently (López-Bautista and Chapman, 2003;López-Bautista et al., 2003, 2006). López-Bautista et al. (2006) isthus far the only study that has considered the phylogeny of theTrentepohliales at the genus and species level using molecular data(18S rRNA gene sequences). This investigation suggested a lack ofcorrespondence between molecular phylogeny and traditionalclassification based on morphological characters, Cephaleurosseemingly being the only monophyletic genus. These results sug-gested that subcuticular habit, heteromorphic life history andoccurrence of zoosporangia in clusters were the only morphologi-cal characters with potential phylogenetic significance. However,the use of a single-gene dataset along with a relatively limited tax-on sampling restricted the extent of the conclusions drawn in thatstudy.

Because of their morphological and ecological diversity andwidespread distribution, Trentepohlia and Printzina represent anideal subject to investigate the evolutionary mechanisms responsi-ble of morphological diversification in terrestrial eukaryotic micro-algae. The objective of this study is to infer phylogeneticrelationships within Trentepohlia and Printzina through analysesof the nuclear-encoded 18S rRNA and the chloroplast-encodedlarge subunit of the ribulose-1,5-bisphosphate carboxylase/oxy-genase (rbcL) genes. This investigation extends the dataset of 18SrRNA sequences available and provides the first published datasetof rbcL sequences for these organisms. By comparing our molecularresults with the morphology-based classification of the order Tren-tepohliales, we provide new insights into the phylogeny of thesealgae.

2. Materials and methods

2.1. Collections, taxon sampling and vouchering

Samples of Trentepohlia and Printzina were collected by theauthors in the course of previous studies or obtained from exter-nal collaborators (Supplementary Table 1). Collections weremade mainly in French Guiana (Rindi and López-Bautista,2008), Panama (Rindi et al., 2008a), western Ireland (Rindi andGuiry, 2002) and the Southeastern USA. The specimens werestored in sealable plastic bags and identified at the best possiblelevel of taxonomic resolution in the laboratory. Whenever possi-

F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339 331

ble, the strains collected were isolated into unialgal culturesusing liquid and agarized Jaworski’s Medium (Tompkins et al.,1995) and Bold’s Basal Medium with vitamins (Starr and Zeikus,1987). Voucher specimens were deposited in the University ofAlabama herbarium (UNA) and in the phycological herbariumof the National University of Ireland, Galway (GALW).

2.2. DNA extraction, PCR and sequencing

DNA was extracted from 43 strains of Trentepohlia, Printzinaand Cephaleuros (Supplementary Table 1) using the QiagenDNeasy Plant Mini Kit (Qiagen, Valencia, California). With rela-tively few exceptions (Trentepohlia cfr. abietina, T. dusenii, T. flav-a, T. umbrina from Tirrenia, Italy, and T. cfr. umbrina from FrenchGuiana) the extraction was performed on specimens isolated intounialgal cultures. For all PCR amplifications, a GeneAmp PCR2700 thermal cycler (Applied Biosystems, Foster City, California)was used. PCR of the 18S rRNA and rbcL genes was performed astwo overlapping fragments using several combinations of prim-ers (Supplementary Table 2). Whereas most primers were ob-tained from the literature, some were designed by J.M.L.-B.using Oligo6.89 (Molecular Biology Insights, Cascade, Colorado).Each 13 lL reaction contained: 4.4 lL of water; 1.25 lL of 10�reaction buffer; 1.25 lL of MgCl2 (25 mM); 1.25 lL of dATP,dCTP, dGTP and dTTP cocktail (8 mM); 0.625 lL of each primer(10 mM); 0.1 lL of Taq (New England BioLabs, Ipswich, Massa-chusetts); 2.5 lL of betaine (5 M); and 1.0 lL of total genomicDNA. For both 18S rRNA and rbcL, the normal PCR protocol con-sisted of an initial denaturing phase of 10 s at 96 �C, followed by40 cycles of 94 �C for 1 min, 50 �C for 1 min and 72 �C for1.5 min, with a final extension of 8 min at 72 �C. The PCR prod-ucts were examined for correct length, yield and purity underUV light on 1% agarose gels stained with ethidium bromide,and subsequently purified using the Qiagen MinElute Gel Extrac-tion Kit (Qiagen). The amount of DNA in PCR products was quan-tified with a NanoDrop ND-1000 spectrophotometer (NanoDropTechnologies, Wilmington, Delaware). PCR products were se-quenced in both directions using Big Dye version 3.1 (AppliedBiosystems) and sequence readings were obtained with an ABI3100 automated sequencer (Applied Biosystems). Sequence chro-matograms were aligned and contigs were created using Sequen-cher 4.5 (Gene Codes Corporation, Ann Arbor, Michigan).

2.3. Exploration of the characteristics of the molecular datasets

Three different methods were used in order to estimate therelative amounts of phylogenetic signal and noise in the 18SrRNA and rbcL datasets. First, skewness in the distribution ofthe data was examined by calculating the g1-statistic (Hillisand Huelsenbeck, 1992) using 10,000 random trees in PAUP*4.0b10 ( Swofford, 1998). The values obtained were comparedwith the empirical threshold values reported by Hillis and Huel-senbeck (1992) to verify the non-random distribution of thedata. Second, potential substitutional saturation in the datasetsanalyzed was evaluated graphically by plotting the uncorrectedp-distances against the Maximum Likelihood-corrected distancesas determined with the evolutionary model which yielded thebest fit to the data (estimated using the corrected Akaike Infor-mation Criterion (AICc) in ModelTest 3.7; Posada and Crandall,1998). Separate plots were produced for datasets with and with-out outgroup taxa. Finally, a test of substitutional saturationbased on the ISS statistic, as proposed by Xia et al. (2003) andimplemented in DAMBE (Xia and Xie, 2001), was performed sep-arately for the 18S rRNA and rbcL datasets, as well as for thethree codon positions of rbcL separately.

2.4. Sequence alignment and phylogenetic analyses

Phylogenetic analyses were performed separately on the twogenes and on a combined 18S rRNA-rbcL dataset. For the 18S rRNAgene, an alignment of 1618 basepairs was used for phylogeneticanalyses. The alignment was constructed by adding sequences ofTrentepohliales to the alignment of Siphonocladales of Leliaertet al. (2007), which was prepared on the basis of the secondarystructure information, and adjusted by eye. The dataset approxi-mately corresponds to positions 78–1697 of the complete se-quence of Volvox carteri f. nagariensis X53904 (Rausch et al.,1989). Beside the newly sequenced strains, most of the 18S rRNAsequences of Trentepohliales available in GenBank were used (Sup-plementary Table 1). The choice of the outgroup taxa was based onrecommendations by Verbruggen and Theriot (2008), and madeafter preliminary analyses in which different sets of taxa, moreand less closely related to the Trentepohliales, were used as out-groups. At present, the ulvophycean orders Cladophorales/Siphonocladales and Dasycladales are the groups which are knownas most closely related to the Trentepohliales (López-Bautista andChapman, 2003). The preliminary analyses indicated that membersof these orders were a suitable choice as outgroup taxa, and the fol-lowing sequences were selected: Cladophora rupestris Z35319,Cladophora vagabunda Z35316, Valonia utricularis Z35323, Bornetel-la nitida Z33464 and Polyphysa peniculus Z33472.

The rbcL alignment consisted of 879 basepairs (correspondingto the positions 232–1110 of the complete sequence of Chlorellavulgaris AB001684; Wakasugi et al., 1997) and was constructedby eye using MacClade 4.05 (Maddison and Maddison, 2002). Noindels were created in the rbcL alignment. The choice of the out-group taxa was based on the same considerations as for 18S rRNA.In this case, since no published rbcL sequences of Cladophorales/Siphonocladales are currently available, members of the Dasycla-dales and Bryopsidales were used as outgroup taxa: Batophora oer-stedii AY177748, Bornetella nitida AY177746, Polyphysa peniculusAY177743, Bryopsis hypnoides AY942169 and Halimeda discoideaAB038488. The combined rbcL-18S rRNA dataset consisted of2496 basepairs and included 23 taxa for which both rbcL and 18SrRNA could be sequenced successfully. Batophora oerstedii, Borne-tella nitida and Polyphysa peniculus were used as outgroups. Thecombined sequences used for these taxa were originally obtainedfrom the same samples, sequenced for 18S rRNA by Olsen et al.(1994) and for rbcL by Zechman (2003).

Phylogenetic inference was based on Maximum Parsimony(MP), Maximum Likelihood (ML) and Bayesian inference (BI). MPwas performed using PAUP* 4.0b10; the analysis consisted of aheuristic search with random taxon addition (100 replicates),tree-bisection–reconnection branch swapping, steepest descentand Multrees option enabled. For rbcL and the combined 18SrRNA-rbcL dataset, the model-based analyses (ML and BI) wereperformed on partitioned datasets, applying separate models toeach partition. Three partitions were used for rbcL (first, secondand third codon position); four partitions were used for the com-bined dataset (the three codon positions of rbcL and the whole18S rRNA gene). Rate differences between partitions were accom-modated with ‘‘rate amplifier” parameters in the models of mod-el-based analyses, using the default settings of the programsused. ML was performed using Treefinder (version April 2008; Jobbet al., 2004). The models and parameters selected by Treefinderwere applied. For 18S rRNA, the model selected by Treefinder un-der the AICc criterion was GTR + G; for rbcL, the models selectedwere GTR + G for the first codon position, TVM + G for the secondposition and J1 + G for the third position. For MP and ML, nodalsupport was assessed by non-parametric bootstrap analysis (Fel-senstein, 1985) with 1000 resamplings.

332 F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339

BI was performed using MrBayes 3.04 (Huelsenbeck and Ron-quist, 2001). In each BI analysis, two parallel runs of four MonteCarlo Markov Chains (one cold and three incrementally heated)were conducted for 2,000,000 generations, with tree samplingevery 100 generations. Priors and probability proposals set as de-fault in MrBayes were used. The stationary distribution of the runswas verified before to stop the analysis; it was assumed that thestationary distribution was reached when the average standarddeviations of split frequencies between the two runs was lowerthan 0.01. If the stationary distribution was not reached at theend of the runs, the analysis was continued for additional1,000,000 generations. The burn-in phase was assessed by plottingthe number of generations against the likelihood scores and deter-mining where the analysis reached stationarity; 100,000 to300,000 generations were discarded as burn-in, and the remainingsamples were used to construct the majority-rule consensus trees.For each partition, the model was set to GTR + I + G by setting thenumber of substitution types to 6 with gamma shape parameterand proportion of invariable sites; the parameters were unlinkedand allowed to vary across partitions.

In relevant cases, the significance of differences between thetopologies inferred in the analyses and other possible topologieswas tested using Shimodaira–Hasegawa (SH) tests (Shimodairaand Hasegawa, 1999) or Approximately Unbiased (AU) tests (Shi-modaira, 2002) as implemented in Treefinder.

3. Results

3.1. Characteristics of the molecular datasets

In the Trentepohliales, the rbcL gene showed a considerablyhigher substitution rate than the 18S rRNA gene (Table 1). In theingroup taxa, 387 variable sites (44%) occurred in the rbcL align-ment; 327 of these (37%) were parsimony-informative. Only 175variable sites (=10.8%) were found in the 18S rRNA alignment,117 of which (7.2%) were parsimony-informative. The highest p-uncorrected sequence distances were 21.1% for rbcL (betweenPrintzina lagenifera from French Guiana and Trentepohlia umbrinafrom Pavia, Italy) and 5.5% for 18S rRNA (between Trentepohliaumbrina from Ireland and Physolinum monile from Louisiana StateUniversity culture collection). For all datasets, with and withoutoutgroup taxa, the g1-statistic was considerably more negative(left-skewed) than the empirical threshold values of Hillis andHuelsenbeck (1992), indicating that the alignments were signifi-cantly more structured than random data (suggesting limited sat-uration). In 18S rRNA, the saturation plot for the ingroup taxashowed a near-linear correlation between ML-corrected and

Table 1Characteristics of the datasets analyzed. Figures without brackets refer to datasets withstatistic, tests marked with an asterisk indicate significant results (=indicative of little ssymmetrical topologies.

rbcL (total) rbcL (1st position

Characters analyzed 879 293Parsimony-informative sites 327 (374) 70 (87)Variable parsimony-uninformative sites 60 (56) 17 (21)Invariable sites 492 (449) 206 (185)Highest uncorrected p-divergence within

ingroup taxa21.1%

Number of most parsimonious trees/length insteps

3/1584(5/2061)

Measure of skewness �0.559123 �0.461276(g1-value) (�0.480546) (�0.442506)ISS statistic 0.327/0.726* 0.204/0.659*(ISS/ISS.C, value for 32 taxon data subsets) (0.325/0.695)* (0.222/0.676)*

uncorrected p-distances, suggesting little saturation (Fig. 1A andB). In the rbcL gene, similarly limited saturation was observed forthe first and second codon positions, as well as for the whole gene(Fig. 1C–H). As expected, evidence of partial saturation was ob-served for the third codon position, for which the plot tended to le-vel off with increasing genetic distance (Fig. 1I and J). However, themajority of the parsimony-informative positions in rbcL were lo-cated in the third position (226 out of 327, =69%) and analyses inwhich the third position was excluded did not improve phyloge-netic resolution. MP and ML analyses performed only on first andsecond positions resulted in mostly unresolved trees; althoughthe number of most parsimonious trees recovered was relativelylow (seven with outgroups, three without outgroups), the boostrapanalyses recovered very few well-supported nodes. Similar resultswere obtained in analyses performed on the translated amino acidsequence, in which large polytomies and few resolved nodes oc-curred (results not shown). Analyses performed only on the thirdposition recovered three most parsimonious trees with outgroupsand five without outgroups. Although they were not able to resolvethe internal nodes, they provided a higher number of supportednodes and their results were generally more similar to analysesperformed on the whole gene. The results of the saturation testsbased on the ISS statistic did not indicate substantial saturation.The ISS.C values were significantly lower than the ISS values, whencalculated by DAMBE for both symmetrical and asymmetrical trees(indicating little saturation; Table 1). Only the test performed onthe third position of rbcL in the dataset inclusive of outgroups gavenot significant results for a highly asymmetrical tree (suggestingsubstantial saturation in the case of a highly asymmetrical tree);however, the results of the same test were significant in the caseof a symmetrical tree.

3.2. Phylogenetic analyses

The topologies recovered in the rbcL and 18S rRNA trees weregenerally congruent. No major discrepancies occurred between dif-ferent methods of phylogenetic reconstruction; differences wereonly observed in the placement of taxa or clades that received littleor no statistical support.

In general, rbcL showed better resolution than 18S rRNA, recov-ering a higher number of well-supported nodes (BP > 60%,PP > 0.9). In the rbcL phylogeny, three main clades were recoveredwith moderate to high support (Fig. 2). A clade sister to all otherTrentepohliales (hereafter indicated as clade 1; Fig. 2) includedspecies of Trentepohlia and Printzina with very different morphol-ogy and widespread geographical distribution (such as Printzinabosseae, Trentepohlia flava, T. iolithus and T. umbrina), as well as

out outgroups; figures in brackets refer to datasets including outgroups. For the ISS

aturation); the ISS.C values indicated are the values calculated by DAMBE assuming

) rbcL (2nd position) rbcL (3rdposition)

18S rRNA rbcL + 18S rRNA

293 293 1618 249631 (40) 226 (247) 117 (424) 380 (644)11 (15) 32 (20) 58 (142) 95 (188)251 (238) 35 (26) 1443 (1052) 2021 (1664)

5.5% 10.0%

4/398 2/1257(31/1049) (1/2009)

�0.397632 �0.571262 �0.751748 �0.973715(�0.382366) (�0.478957) (�1.985496) (�1.394987)0.152/0.659* 0.431/0.659* 0.309/0.779*(0.147/0.659)* (0.480/0.675)* (0.194/0.778)*

0

0.05

0.1

0.15

0.2

0.25

0.3

0 0.2 0.4 0.6 0.8 1

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

0

0.01

0.02

0.03

0.04

0.05

0.06

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

0

0.05

0.1

0.15

0.2

0.25

0.3

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

0

0.05

0.1

0.15

0.2

0.25

0 0.1 0.2 0.3 0.4 0.5

ML-corrected distancesUn

corr

ecte

d p-

dist

ance

s

0

0.05

0.1

0.15

0.2

0.25

0 0.1 0.2 0.3 0.4 0.5

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0 0.05 0.1 0.15 0.2 0.25 0.3

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

0

0.02

0.04

0.06

0.08

0.1

0.12

0 0.05 0.1 0.15 0.2 0.25 0.3

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

0

0.1

0.2

0.3

0.4

0.5

0.6

0 0.5 1 1.5 2 2.5 3

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

00.050.10.150.20.250.30.350.40.450.5

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

ML-corrected distances

Unco

rrec

ted

p-di

stan

ces

81 )spuorgtuo htiw( S81 S (without outgroups)

rbcL (with outgroups) rbcL (without outgroups)

rbcL 1st (without outgroups) rbcL 1st (with outgroups)

rbcL 2nd (with outgroups) rbcL 2nd (without outgroups)

rbcL 3rd (with outgroups) rbcL 3rd (without outgroups)

A B

C D

E F

G H

I J

Fig. 1. Analysis of saturation of the 18S rRNA and rbcL genes by plotting ML-corrected distances against uncorrected p-distances. Graphs in the left column are for databaseswith outgroups; graphs in the right column are for databases without outgroups. The databases concerned are specified for each graph.

F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339 333

Fig. 2. Phylogram inferred from Maximum Likelihood analysis of the rbcL gene inTrentepohlia, Printzina and related taxa, with bootstrap support (BP) and Bayesianposterior probabilities (PP) indicated at the nodes. From left to right (or from top tobottom, above and below branches), support values at nodes correspond toMaximum Parsimony BP, Maximum Likelihood BP and Bayesian PP. BP values lowerthan 60% and PP lower than 0.85 are not reported. The tree was rooted usingoutgroup sequences of Dasycladales and Bryopsidales (not shown) specified inSection 2.

Fig. 3. Phylogram inferred from Maximum Likelihood analysis of the 18S rRNAgene in Trentepohlia, Printzina and related taxa, with bootstrap support (BP) andBayesian posterior probabilities (PP) indicated at the nodes. From left to rightsupport values at nodes correspond to Maximum Parsimony BP, MaximumLikelihood BP and Bayesian PP. BP values lower than 60% and PP lower than 0.85are not reported. The tree was rooted using outgroup sequences of Cladophoralesand Dasycladales (not shown) specified in Section 2. New sequences produced inthe present study are marked with an asterisk.

334 F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339

some tropical strains of Trentepohlia that could not be identifiedwith certainty at the species level. The relative positions of taxain this clade were consistent between different analyses, but notalways supported by high BP values and Bayesian PP; a well-sup-ported separation, however, was found between some strains ofTrentepohlia (T. flava, T. iolithus, T. umbrina and T. cfr. umbrina)and the other taxa within the clade (Fig. 2). The other Trentepohli-ales sequenced belonged mainly to two relatively well-supportedclades. One of these (hereafter indicated as clade 2; Fig. 2) includednumerous strains of Trentepohlia and Printzina, collected mostly inthe tropics. The vegetative morphology of these algae corre-sponded primarily to two widely distributed tropical species,Printzina lagenifera (type species of Printzina) and Trentepohliaarborum. In the case of Printzina lagenifera, some strains couldnot be identified with certainty at the species level because theydid not become reproductive; it was therefore impossible to ob-serve the flask-shaped gametangia, which represent the diagnosticcharacter of this species. Neither Printzina lagenifera nor Trentepoh-lia arborum produced well-supported monophyletic groups and ingeneral reconstruction of relationships in this clade was impossi-ble, since almost none of the basal nodes received any support.The other clade (hereafter indicated as clade 3; Fig. 2) was roughlycorrespondent to the genus Cephaleuros. It included several strainsof this genus obtained from culture collections, but it was rendered

paraphyletic by a strain of Trentepohlia dialepta from CCAP culturecollection (the identity of which, however, is unclear; see Section4). Some additional taxa, such as Trentepohlia aurea (type speciesof Trentepohlia), Trentepohlia cfr. abietina from Spain and Printzinasp. (F223) did not belong to any of the three main clades; due tothe lack of support in the internal branches, their position relativeto the other clades could not be clarified.

Despite lower resolution, the 18S rRNA dataset provided phylo-genetic patterns in agreement with the rbcL results (Fig. 3). The rel-ative positions of individual taxa were nearly identical and nomajor differences between the two phylogenies were found. Thesame three main clades were present; only the clade 3 (the Ceph-aleuros-clade), however, received strong nodal support. The clades2 and 3 and Trentepohlia aurea were clustered in a common cladewith high statistical support; the position of Trentepohlia aureaand Printzina sp. (F233) was again unresolved. Some additionalspecies for which rbcL could not be sequenced were included inthe 18S rRNA phylogeny. Trentepohlia abietina was placed in theclade 1, sister to a clade formed by three strains of Trentepohliaumbrina; Physolinum monile occurred in the clade 2, where it wasrecovered with high support as sister taxon to a strain of Printzinacfr. lagenifera from Hawaii.

Analyses carried out on the combined dataset 18S rRNA-rbcL in-cluded 23 taxa and provided the results with the highest BP and PP

F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339 335

support (Fig. 4). In general, all clades which received support insingle-gene analyses were recovered with higher support in thecombined analyses. The three main clades were again recovered.As inferred in the 18S rRNA phylogeny, the clades 2 and 3 and Tren-tepohlia aurea and Printzina sp. (F223) occurred in a common,highly-supported clade. However, relationships within the clade2 were again problematic, as most internal nodes were unresolved;once again, Printzina lagenifera was not monophyletic. The positionof Trentepohlia aurea was well resolved in the Bayesian analysis, inwhich this species was sister to Cephaleuros with high PP; suchrelationship, however, received no support in the MP and ML boot-strap analyses. The position of Printzina sp. (F223) was againunresolved.

4. Discussion

4.1. Characteristics of the molecular datasets

Previous trentepohlialean molecular datasets were limited to arelatively low number of 18S rRNA and phragmoplastin gene se-quences (López-Bautista and Chapman, 2003; López-Bautistaet al., 2003, 2006). The results presented here are in agreementwith previous studies based only on 18S rRNA sequences and thetopology of our trees corresponds well with that of López-Bautistaet al. (2006), which was based on a more limited taxon sampling.Compared to these data, rbcL sequences were expected to providebetter resolution at the genus and species level, and the results ofthis study have basically confirmed this prediction. Among algae,the rbcL gene has proved very useful for testing phylogenetichypotheses at a range of different taxonomic levels, even whenonly partial sequences are used. As a protein-coding gene, it ispotentially affected by problems of substitutional saturation atthe third codon (wobble) position. In some studies this has beenregarded as a major impediment to phylogenetic inference, andthe third positions have been eliminated from the analyzed data-sets (Nozaki et al., 2000; Nakada et al., 2008). However, the major-ity of the rbcL-based investigations has concluded that, even whenthe third position shows evidence of partial saturation, it still con-

Fig. 4. Phylogeny of Trentepohlia, Printzina and related taxa inferred from MaximumLikelihood analysis of the combined dataset rbcL-18S rRNA. From left to rightsupport values at nodes correspond to Maximum Parsimony BP, MaximumLikelihood BP and Bayesian PP. BP values lower than 60% and PP lower than 0.85are not reported. The tree was rooted using outgroup sequences of Dasycladales(not shown) specified in Section 2.

tains a good deal of phylogenetic signal (Gontcharov et al., 2004;De Clerck et al., 2006). Usually, analyses performed on the firstand second positions provide a high number of most parsimonioustrees and poor resolution, whereas analyses limited to third posi-tions yield results similar to analyses performed on the whole gene(Lewis et al., 1997; McCourt et al., 2000; Zechman, 2003; Gontcha-rov et al., 2004; De Clerck et al., 2006). The rbcL dataset analyzed inthis study shows a higher substitution rate than in other groups ofgreen algae (a likely reflection of this high variability is the factthat it has proved impossible to design rbcL primers working satis-factorily for all trentepohlialean taxa on the whole length of thegene). Parsimony-informative positions account for 37% of the to-tal sites, whereas in most green algal orders they usually range be-tween 25% and 30% (Zechman, 2003; Lam and Zechman, 2006;Rindi et al., 2007, 2008b). Third positions include 69% of the parsi-mony-informative sites in our dataset and the distance plots showevidence of saturation, especially when the outgroups are in-cluded. At the same time, however, analyses performed only onthird positions yielded a higher number of resolved nodes thananalyses performed on the first and second positions. The g1 scoresand the ISS values calculated for the whole gene suggest that the ef-fect of the third position does not cause substantial harm to thephylogenetic informativeness of the dataset. In consideration ofthis, and of the fact that the phylogenetic signals provided by rbcLand 18S rRNA are largely congruent, we regard the retention ofrbcL third positions as an appropriate solution for the aims of thepresent investigation. The combination of these two genes hasproved to be a powerful tool of inference at several taxonomic lev-els, extracting successfully the phylogenetic signal of the fast-evolving rbcL and resolving better shallow evolutionary relation-ships, where 18S rRNA lacks variability (Gontcharov et al., 2004).In our case, although not resolutive for several internal nodes(especially in the clade 2), the use of the combined dataset repre-sents an improvement and provides to most nodes higher supportthan single-gene analyses. This situation was observed in almostevery study in which rbcL and 18S rRNA were combined (e.g., Hep-perle et al., 1998; Hayden and Waaland, 2002; Gontcharov et al.,2004; Nakada et al., 2008).

4.2. Implications for the systematics of Trentepohlia and Printzina

Our molecular analyses revealed a phylogenetic scenario in sub-stantial contrast with the traditional taxonomic scheme of Tren-tepohlia and Printzina based on morphology (Printz, 1939).Members of Trentepohlia and Printzina were present in all threemajor clades recovered in our phylogenies. None of these cladescorresponded with the current morphological circumscription ofany trentepohliacean genus, with the only partial exception ofCephaleuros. All strains of Cephaleuros sequenced were clusteredin a clade that received high support in all three datasets analyzedand under all methods of inference used. Cephaleuros was mono-phyletic in the 18S rRNA and combined analyses, but in the rbcLphylogeny a strain of Trentepohlia dialepta from CCAP made it para-phyletic. This suggests that T. dialepta should be transferred toCephaleuros. Trentepohlia dialepta was originally described byNylander (1862) as Coenogonium dialeptum using material fromBrazil. As Cephaleuros, this species is widespread on leaves of treesin tropical regions (Hariot, 1889a; Printz, 1939) and shares withthis genus some important morphological similarities, in particularthe habit of the erect filaments and the arrangement of the zoospo-rangia in clusters at the top of the erect axes (Salleh and Milow,1999). It is noteworthy, however, that the strain from CCAP wasisolated from material from Kampala, Uganda, ‘‘thought to be a cof-fee parasite” (http://www.ccap.ac.uk/strain_info.php?Strain_-No=483/2). Species of Cephaleuros grow in tropical andsubtropical regions as subcuticular epiphytes on leaves of many

336 F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339

plants, including coffee (Chapman, 1984; Thompson and Wujek,1997; Chapman and Waters, 2001). Therefore, the possibility thatthe strain in question is a Cephaleuros which was misidentified infirst instance cannot be excluded. It is also worth mentioning thatspecies of both Trentepohlia and Cephaleuros, when growing in cul-ture, assume a free filamentous morphology and become almostindistinguishable.

The large majority of the strains of Trentepohlia and Printzina se-quenced belonged to either clade 1 or clade 2. Both clades includestrains with very different morphology, and it is presently impos-sible to determine any morphological synapomorphies definingthese two clades. It is clear that none of the morphological charac-ters traditionally used for taxonomic purposes (Hariot, 1890; DeWildeman, 1891, 1900; Printz, 1939; Sarma, 1986; Ettl and Gärt-ner, 1995) are phylogenetically significant. Ecological traits suchas type of habitat and substratum have also been used to distin-guish species and groups of species in the Trentepohliales (Printz,1939). In particular, in the original description of Printzina, Thomp-son and Wujek (1992) reported the type of habitat as one of themain characters distinguishing this genus from Trentepohlia(Printzina associated with shaded forest habitats, Trentepohliaoccurring at more exposed sites). Our results show that these char-acters are also phylogenetically irrelevant, since they are not asso-ciated with any well-supported monophyletic group; algaegrowing on both natural and artificial substrata and in both ex-posed and sheltered habitats are mixed together in both clades.Geographical distribution may be in part a better indicator of phy-logenetic relationships. Whereas most of the strains of the clade 2were identified as Printzina lagenifera or Trentepohlia arborum, twowidespread tropical and subtropical algae (De Wildeman, 1891,1900; Printz, 1939; Rindi et al., 2005, 2008a), species associatedmainly with temperate regions (Trentepohlia abietina, T. flava, T. io-lithus and T. umbrina) belong to the clade 1. However, exceptions tothis situation are present in both clades (e.g., Printzina bosseae inthe clade 1, Printzina cfr. lagenifera from Italy and P. lagenifera fromIreland in the clade 2) and we feel that further taxon sampling isnecessary before drawing generalizations in this regard. Someadditional taxa (Trentepohlia aurea, T. cfr. abietina from Spain andPrintzina sp. F223) are not associated with any of the three mainclades and their positions were not fully resolved in any of thedatasets analyzed. In the case of Trentepohlia aurea, this is particu-larly problematic, since this is the type species of Trentepohlia.Therefore, its position affects directly the genus-level classificationin the whole order.

At present, the genus-level taxonomy in the Trentepohliales isbased entirely on morphology. The subdivision of the six generais straightforward and founded on easily observable characters,such as habit, morphology of the erect parts relative to the pros-trate parts, and arrangement of the sporangiate laterals (Printz,1939). Thompson and Wujek (1992, 1997) introduced some addi-tional characters related to the morphology of the zoosporangia(shape of the zoosporangium and position of the escape pore).Our molecular analyses present strong evidence that such classifi-cation will need major modifications based on phylogenetic princi-ples. Designations at the genus level might be an appropriatesolution for our clades 1, 2 and 3. This, however, would introduceradical rearrangements in the nomenclature and in the genus cir-cumscription of the Trentepohliales; we feel that in the present sit-uation this would be premature. Decisions on these matters shouldbe made only after additional molecular datasets able to resolvethe relationships of lineages eligible as genera (in particular the po-sition of Trentepohlia aurea relative to the other main clades) be-come available. Even more important will be the availability ofhigh-quality sequences of two genera that we could not includein our databases, Phycopeltis and Stomatochroon. Unfortunately,species of Phycopeltis are virtually impossible to grow in culture

and so far we have been unable to obtain reliable sequences fromthis genus. Additionally, the type species Phycopeltis epiphyton(Millardet, 1870) is one of the smallest-sized in the genus andcan be detected with the unaided eye only when it produces largepopulations. The same considerations apply to Stomatochroon,which, for its small size and elusive nature (this alga lives insideof the substomatal chambers of plant leaves), is the least recordedtrentepohlialean genus (Thompson and Wujek, 1997; López-Bau-tista et al., 2002). In the course of our field surveys, we have neverfound specimens of this genus.

4.3. Morphological versus phylogenetic species delineation inTrentepohlia and Printzina

Our molecular results indicate that the circumscription ofspecies in Trentepohlia and Printzina is a difficult matter. The corre-spondence between morphological and phylogenetic circumscrip-tion varies greatly from species to species and needs to beevaluated case by case; it is clear that a species concept based onlyon morphological features is inadequate for the Trentepohliales (asfor most other groups of terrestrial green algae) and the use ofmolecular data will play an essential role in the taxonomy of thesealgae in the future. Given the morphological homoplasy revealedby our results, the availability of a molecular marker suitable forDNA barcoding would be an ideal solution for species definitionin the Trentepohliales. We feel that the substitution rate of rbcLwould be adequate for DNA barcoding in this order, and in factother studies have suggested the suitability of rbcL for this purposein other groups of Ulvophyceae (Verbruggen et al., 2007b). How-ever, easy amplification through primers designed from conservedsites is a fundamental requirement for barcoding purposes, and inthe case of the Trentepohliales rbcL does not satisfy such require-ment. The preparation of our database required the use of multiplecouples of primers, and several samples could not be amplified atall. In general, the development of molecular markers for DNA bar-coding is still a highly debated topic for green algae and plants; todate there is not yet a marker which has been accepted as a univer-sal DNA barcode (Kress and Erickson, 2008) and intensive work inthis area is still going on.

For some species with relatively stable and well-definedmorphology, all strains included in our analyses produced well-supported clades. Conversely other taxa, as delineated on morpho-logical basis, appear to be paraphyletic or polyphyletic entities thatare better regarded as complexes of cryptic species. Trentepohliaaurea represents a good example of the first case. This species,which is the generitype of Trentepohlia, has always been relativelywell defined from a morphological point of view (e.g., Martius,1817; Kützing, 1854; Hansgirg, 1886; Hariot, 1889b; Printz,1939) and the morphology of the numerous specimens depositedin European herbaria under this name is generally constant (FabioRindi, personal observation). Our molecular data confirm its sepa-ration from similar species, such as Trentepohlia abietina and T. flav-a, which has been very controversial in the past (e.g., Hariot,1889b; De Wildeman, 1900; Printz, 1921, 1939; Cribb, 1970).

Conversely, Trentepohlia umbrina, T. arborum and P. lageniferarepresent polyphyletic morphotaxa forming complexes of crypticspecies evolved through morphological convergence. For Tren-tepohlia umbrina, the European samples that we sequenced forma well-supported monophyletic group (Fig. 3) and we considerthem genuinely representative of this species (described by Küt-zing (1843) for corticolous material from southern Germany).However, an identical or very similar morphology was also foundin a sample from Louisiana belonging to the clade 2 and a samplefrom French Guiana which, although occurring in the clade 1, wasseparated from European T. umbrina with high support (Fig. 2). ForTrentepohlia arborum and Printzina lagenifera the situation is more

F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339 337

complicated. Specimens morphologically corresponding to thesespecies occur in the clade 2, which in the overall scenario of tren-tepohlialean evolution represents a very problematic lineage. Inthis clade most internal branches are very short when comparedto the terminal branches, and the support for most internal nodesis non-existent in all datasets analyzed. Reducing the length of ter-minal branches by adding more taxa may be expected to benefitfuture phylogenetic reconstructions. But the fact that the samepatterns are observed in two genes with very different characteris-tics (such as rbcL and 18S rRNA) suggests that the short internalbranches are the result of a fast evolutionary radiation, and ulti-mately only the use of additional molecular datasets (such as ITSand the D1–D2 28S rRNA) may provide better resolution. For thesereasons, we cannot exclude that the inclusion of further loci in ourmolecular phylogeny would eventually resolve Printzina lageniferaand Trentepohlia arborum as monophyletic taxa. However, for rbcLan average uncorrected p-sequence distance of about 10% was ob-served between strains morphologically corresponding to Printzinalagenifera (with a maximum of 11.38% between P. lagenifera fromPanama and P. lagenifera from Ireland). A similar situation was ob-served between strains morphologically corresponding to Tren-tepohlia arborum, for which an uncorrected p-sequence distanceas high as 10.43% (between T. arborum from French Guiana andT. arborum from Panama) was found. Furthermore, we performedSH and AU tests in order to test differences between the treesrecovered by our ML analyses and trees in which the strains mor-phologically referable to Printzina lagenifera and Trentepohlia arbo-rum were constrained into monophyletic groups. For all threedatabases (rbcL, 18S rRNA, rbcL + 18S rRNA) the tests showed sig-nificant differences, rejecting the monophyly of P. lagenifera andT. arborum. Such evidence suggests that P. lagenifera and T. arborumare unlikely to be individual species, and more probably representcomplexes of cryptic species with convergent morphology. Thiswill make necessary a precise characterization of their taxonomicidentities, which in the case of Printzina lagenifera will be particu-larly difficult. This species was described by Hildebrand (1861)from bark of vines in a greenhouse hosting palms in Germany. Al-most certainly, the material used by Hildebrand for the descriptionwas introduced as epiphyte on tropical plants; without knowledgeof the vector plant’s origin, it is impossible to know the actual ori-gin of the alga.

4.4. Conclusions

Terrestrial algae of the genera Trentepohlia and Printzina repre-sent a group of organisms in which morphological traits are notindicative of evolutionary relationships. These genera do not formmonophyletic groups and their separation is not justified bymolecular data. If their separation is to be maintained, their mor-phological circumscription will have to be reassessed radically.The morphology of these algae is affected by great homoplasy,which has caused convergence of separate taxa into an almostidentical morphology. Because of this, the concept of species inTrentepohlia and Printzina (and in the Trentepohliales in general)cannot be based exclusively on morphological grounds. The taxo-nomic identity of each species will have to be circumscribed caseby case, using a combination of molecular data and morphologicalinvestigations in the field and in culture. We also feel that othertypes of genotypic and phenotypic characters (chromosome num-ber, secondary metabolites, ultrastructural characteristics) deserveto be investigated in depth, as they could reveal the synapomor-phies which cannot be found in the morphology.

Finally, it is becoming increasingly obvious that the morpholog-ical convergence observed in Trentepohlia and Printzina is a generaltrait of the evolution of subaerial green algae. This phenomenonhas been a great impediment in the systematics of these organisms

and has delayed a full appreciation of their genetic diversity. Whilecases of morphological convergence have already been docu-mented for species of the classes Trebouxiophyceae and Chloro-phyceae (Huss et al., 1999; Rindi et al., 2007), this study and ourrecent discovery of the genus Spongiochrysis (Rindi et al., 2006)show that this phenomenon has also taken place for terrestrialmembers of the Ulvophyceae. The overall message emerging fromthese studies is that life in terrestrial environments involves someconstraints for which a diminutive size and the reduction to a lim-ited range of morphological forms are favored. At present, it isimpossible to establish which environmental factors are the mainconstraints forcing such reduction and convergence (and thealmost complete absence of fossil record represent a substantiallimitation to propose hypotheses in this regard). However, we feelthat a combination of systematic studies with physiological, bio-chemical and ecological investigations (focusing in particular ondispersal mechanisms) might lead to major insights into this inter-esting aspect of terrestrial algal evolution.

Acknowledgments

The study was funded by the US National Science Foundation(Systematics Program DEB-0542924 to J.M.L.-B.), with partialsupport from the College of Arts and Sciences and the Depart-ment of Biological Sciences, The University of Alabama. F.R. isgrateful to the Marine Institute of Ireland for financial supportunder the Beaufort Biodiscovery Programme. D.W.L. acknowl-edges the University of Alabama for financial support in the formof Aquatic Biology and Graduate Council Fellowships. Assistancefrom Britton O’Shields with DNA extractions and PCR, as wellas from Brook Fluker and Wally Holznagel with DNA sequencing,were greatly appreciated. We are very grateful to Frederik Lelia-ert for stimulating discussion, suggestions which greatly helpedto improve the manuscript, and the 18S rRNA alignment ofSiphonocladales that was used for the preparation of the 18SrRNA alignment. Recommendations by Heroen Verbruggen andan anonymous reviewer were of great help to improve the origi-nal version of the manuscript. Assistance by Frédéric Lecorre(Floramazone) during fieldwork in French Guiana and by the staffof the S.T.R.I. Station of Barro Colorado Island (in particular OrisAcevedo) during fieldwork in Panama is gratefully acknowledged.Collections of Trentepohliales received from Ricardo Tsukamoto,Alison Sherwood, Michael Guiry, Benito Gómez-Silva, PatrickMartone, Willem Prud’homme Van Reine and Paul Broady weregreatly appreciated. Dr. Bruno De Reviers is gratefully acknowl-edged for the loan of an authentic specimen of Trentepohlia duse-nii (PC0110846) from PC.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ympev.2009.01.009.

References

Agardh, C.A., 1824. Systema Algarum. Berling, Lund.Cardon, Z.G., Gray, D.W., Lewis, L.A., 2008. The green algal underground:

evolutionary secrets of desert cells. Bioscience 58, 114–122.Chapman, R.L., 1984. An assessment of the current state of our knowledge of the

Trentepohliaceae. In: Irvine, D.E.G., John, D.M. (Eds.), Systematics of the GreenAlgae. Academic Press, London, pp. 233–250.

Chapman, R.L., Good, B.H., 1983. Subaerial symbiotic green algae: interactions withvascular plant hosts. In: Goff, L.J. (Ed.), Algal Symbiosis: A Continuum ofInteraction Strategies. Cambridge University Press, London and New York, pp.173–204.

Chapman, R.L., Henk, M.C., 1983. Ultrastructure of Cephaleuros virescens(Chroolepidaceae, Chlorophyta). IV. Absolute configuration analysis of the

338 F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339

cruciate flagellar apparatus and multilayered structure in pre- and post-releasegametes. Am. J. Bot. 70, 1340–1355.

Chapman, R.L., Waters, D.A., 2001. Lichenization of the Trentepohliales—complexalgae and odd relationships. In: Seckbach, J. (Ed.), Symbiosis. Kluwer AcademicPublishers, Dordrecht, pp. 359–371.

Cribb, A.B., 1970. A revision of some species of Trentepohlia especially fromQueensland. Proc. R. Soc. Qld 82, 17–34.

De Clerck, O., Leliaert, F., Verbruggen, H., Lane, C.E., De Paula, J.C., Payo, D.A.,Coppejans, E., 2006. A revised classification of the Dictyoteae (Dictyotales,Phaeophyceae) based on rbcL and 26S ribosomal DNA sequence analysis. J.Phycol. 42, 1271–1288.

De Clerck, O., Verbruggen, H., Huisman, J.M., Faye, E.J., Leliaert, F., Schils, T.,Coppejans, E., 2008. Systematics and biogeography of the genus Pseudocodium(Bryopsidales, Chlorophyta), including the description of P. Natalense sp. nov.from South Africa. Phycologia 47, 225–235.

De Wildeman, E., 1891. Les Trentepohlia des Indes Neerlandaises. Ann. Jard. Bot.Buitenzorg 9, 127–142.

De Wildeman, E., 1900. Les algues de la flore de Buitenzorg. E.G. Brill, Leiden.Ettl, H., Gärtner, G., 1995. Syllabus der Boden-, Luft- und Flechtenalgen. Gustav

Fischer Verlag, Stuttgart, Jena and New York.Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using the

bootstrap. Evolution 39, 783–791.Flint, E.A., 1959. The occurrence of zoospores in Physolinum Printz. New Phytol. 58,

267–270.Gontcharov, A.A., Marin, B., Melkonian, M., 2004. Are combined analyses better than

single gene phylogenies? A case study using SSU rDNA and rbcL sequencescomparisons in the Zygnematophyceae (Streptophyta). Mol. Biol. Evol. 21, 612–624.

Hansgirg, A., 1886. Prodromus der algenflora von Böhmen. Erster theil enthaltenddie Rhodophyceen, Phaeophyceen und einen theil der Chlorophyceen. Arch.Naturwiss. Landesdurchf. Böhmen 5, 1–96.

Hariot, P., 1889a. Notes sur le genre Trentepohlia Martius (Suite). J. Bot. Paris 3, 378–388.

Hariot, P., 1889b. Notes sur le genre Trentepohlia Martius (Suite). J. Bot. Paris 3, 366–375.

Hariot, P., 1890. Notes sur le genre Trentepohlia Martius (Suite). J. Bot. Paris 4, 178–180.

Hayden, H.S., Waaland, J.R., 2002. Phylogenetic systematics of the Ulvaceae (Ulvales,Ulvophyceae) using chloroplast and nuclear DNA sequences. J. Phycol. 38,1200–1212.

Hepperle, D., Nozaki, H., Hoenberger, S., Huss, V.A.R., Morita, E., Krienitz, L., 1998.Phylogenetic position of the Phacotaceae within the Chlamydophyceae asrevealed by analysis of 18S rDNA and rbcL sequences. J. Mol. Evol. 47, 420–430.

Hildebrand, F., 1861. Über einen Chroolepus mit Zoosporenbildung. Bot. Zeit. 13, 81–85.

Hillis, D.M., Huelsenbeck, J.P., 1992. Signal, noise and reliability in molecularphylogenetic analyses. J. Hered. 83, 189–195.

Howland, L.J., 1929. The moisture relations of terrestrial algae. IV. Periodicobservations of Trentepohlia aurea Martius. Ann. Bot. 43, 173–202.

Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogeny.Bioinformatics 17, 754–755.

Huss, V.A.R., Frank, C., Hartmann, E.C., Hirmer, M., Kloboucek, A., Seidel, B.M.,Wenzeler, P., Kessler, E., 1999. Biochemical taxonomy and molecular phylogenyof the genus Chlorella sensu lato (Chlorophyta). J. Phycol. 35, 587–598.

Jobb, G., von Haeseler, A., Strimmer, K., 2004. TREEFINDER: a powerful graphicalanalysis environment for molecular phylogenetics. 2004. BMC Evol. Biol.Available from: <http://www.biomedcentral.com/1471–2148/4/18>.

Kress, W.J., Erickson, D.L., 2008. DNA barcodes: genes, genomics and bioinformatics.Proc. Natl. Acad. Sci. USA 105, 2761–2762.

Kützing, F.T., 1843. Phycologia generalis. F.A. Brockhaus, Leipzig.Kützing, F.T., 1854. Tabulae phycologicae oder Abbildungen der Tange. IV. Privately

published, Nordhausen.Lam, D.W., Zechman, F.W., 2006. Phylogenetic analyses of the Bryopsidales

(Ulvophyceae, Chlorophyta) based on Rubisco large subunit gene sequences. J.Phycol. 42, 669–678.

Leliaert, F., De Clerck, O., Verbruggen, H., Boedeker, C., Coppejans, E., 2007.Molecular phylogeny of the Siphonocladales (Chlorophyta:Cladophorophyceae). Mol. Phylogenet. Evol. 44, 1237–1256.

Lewis, L.A., 2007. Chlorophyta on land: independent lineages of green eukaryotesfrom arid lands. In: Seckbach, J. (Ed.), Algae and Cyanobacteria in ExtremeEnvironments. Springer, Dordrecht, pp. 571–582.

Lewis, L.A., Lewis, P.O., 2005. Unearthing the molecular phylodiversity of desert soilgreen algae (Chlorophyta). Syst. Biol. 54, 936–947.

Lewis, L.A., McCourt, R.M., 2004. Green algae and the origin of land plants. Am. J.Bot. 91, 1535–1556.

Lewis, L.A., Mishler, B.D., Vilgalys, R., 1997. Phylogenetic relationships of theliverworts (Hepaticae), a basal embryophyte lineage, inferred from nucleotidesequence data of the chloroplast gene rbcL. Mol. Phylogenet. Evol. 7, 377–393.

Linnaeus, C., 1753. Species Plantarum, vol. 2. Impensis Laurentii Salvii, Holmiae(Stockholm), pp. 561–1200.

Linnaeus, C., 1759. Systema naturae per regna tria naturae, vol. 2. Stockholm, pp.825–1384.

López-Bautista, J.M., Chapman, R.L., 2003. Phylogenetic affinities of theTrentepohliales inferred from small-subunit rDNA. Int. J. Syst. Evol. Microbiol.53, 2099–2106.

López-Bautista, J.M., Rindi, F., Casamatta, D., 2007. The systematics of subaerialalgae. In: Seckbach, J. (Ed.), Algae and Cyanobacteria in Extreme Environments.Springer, Dordrecht, pp. 601–617.

López-Bautista, J.M., Rindi, F., Guiry, M.D., 2006. Molecular systematics of thesubaerial green algal order Trentepohliales: an assessment based onmorphological and molecular data. Int. J. Syst. Evol. Microbiol. 56, 1709–1715.

López-Bautista, J.M., Waters, D.A., Chapman, R.L., 2002. The Trentepohlialesrevisited. Constancea 83: http://ucjeps.berkeley.edu/constancea/83/lopezetal/trentepohliales.html.

López-Bautista, J.M., Waters, D.A., Chapman, R.L., 2003. Phragmoplastin, green algaeand the evolution of cytokinesis. Int. J. Syst. Evol. Microbiol. 53, 1715–1718.

Maddison, D.R., Maddison, W.P., 2002. MacClade 4.05: Analysis of Phylogeny andCharacter Evolution. Sinauer Associates, Sunderland, MA.

von Martius, C.F.P., 1817. Flora Cryptogamica Erlangensis. J.L. Schrag, Nürnberg.McCourt, R.M., Karol, K.G., Bell, J., Helm-Bychowski, M., Grajewska, A.,

Wojciechowski, M.F., Hoshaw, R.W., 2000. Phylogeny of the conjugating greenalgae (Zygnemophyceae) based on rbcL sequences. J. Phycol. 36, 747–758.

Millardet, M.A., 1870. De la germination des zygospores dans les genres Closteriumet Staurastrum et sur un genre nouveau d’algues chlorosporées. Mém. Soc. Sci.Nat. Strasbourg 6, 37–51.

Nakada, T., Nozaki, H., Pröschold, T., 2008. Molecular phylogeny, ultrastructure andtaxonomic revision of Chlorogonium (Chlorophyta): emendation ofChlorogonium and description of Gungnir gen. nov. and Rusalka gen. nov.. J.Phycol. 44, 751–760.

Nakano, T., Handa, S., 1984. Observations on Trentepohlia lagenifera (Hild.) Wille(Chlorophyceae, Trentepohliaceae). Jpn. J. Phycol. 32, 354–363.

Nakano, T., Ihda, T., 1996. The identity of photobionts from the lichen Pyrenulajaponica. Lichenologist 28, 437–442.

Nozaki, H., Misawa, K., Kajita, T., Kato, M., Nohara, S., Watanabe, M.M., 2000. Originand evolution of the colonial Volvocales (Chlorophyceae) as inferred frommultiple chloroplast gene sequences. Mol. Phylogenet. Evol. 17, 256–268.

Nylander, W., 1862. Quelques observations sur le genre Coenogonium. Ann. Sci. Nat.Bot. ser. 4 16, 83–96.

Olsen, J.L., Stam, W.T., Berger, S., Menzel, D., 1994. 18S rDNA and evolution in theDasycladales (Chlorophyta): modern living fossils. J. Phycol. 30, 729–744.

Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution.Bioinformatics 14, 817–818.

Printz, H., 1921. Subaerial algae from South Africa. Kong. Norsk. Vidensk. Selsk.Skrift. 1, 3–41.

Printz, H., 1939. Vorarbeiten zu einer monographie der Trentepohliaceen. Nytt Mag.Naturvbidensk. 80, 137–210.

Pröschold, T., Leliaert, F., 2007. Systematics of the green algae: conflict of classic andmodern approaches. In: Brodie, J., Lewis, J. (Eds.), Unravelling the Algae: ThePast, Present and Future of Algal Systematics. The Systematics AssociationSpecial Volume Series 75. CRC Press, Boca Raton, London and New York, pp.123–153.

Rausch, H., Larsen, N., Schmitt, R., 1989. Phylogenetic relationships of the greenalgae Volvox carteri deduced from small-subunit ribosomal RNA comparisons. J.Mol. Evol. 29, 255–265.

Rindi, F., Guiry, M.D., 2002. Diversity, life history and ecology of Trentepohlia andPrintzina (Trentepohliales, Chlorophyta) in urban habitats in western Ireland. J.Phycol. 38, 39–54.

Rindi, F., Lam, D.W., López-Bautista, J.M., 2008a. Trentepohliales (Ulvophyceae,Chlorophyta) from Panama. Nova Hedwigia 87, 421–444.

Rindi, F., Guiry, M.D., López-Bautista, J.M., 2008b. Distribution, morphology andphylogeny of Klebsormidium (Klebsormidiales, Charophyceae) in urbanenvironments in Europe. J. Phycol. 44, 1529–1540.

Rindi, F., López-Bautista, J.M., 2007. New and interesting records of Trentepohlia(Trentepohliales, Chlorophyta) from French Guiana, including the description oftwo new species. Phycologia 46, 698–708.

Rindi, F., López-Bautista, J.M., 2008. Diversity and ecology of Trentepohliales(Ulvophyceae, Chlorophyta) in French Guiana. Cryptogamie Algol. 29, 13–43.

Rindi, F., López-Bautista, J.M., Sherwood, A.R., Guiry, M.D., 2006. Morphology andphylogenetic position of Spongiochrysis hawaiiensis gen. et sp. nov., the firstknown terrestrial member of the order Cladophorales (Ulvophyceae,Chlorophyta). Int. J. Syst. Evol. Microbiol. 56, 913–922.

Rindi, F., McIvor, L., Sherwood, A.R., Friedl, T., Guiry, M.D., Sheath, R.G., 2007.Molecular phylogeny of the green algal order Prasiolales (Trebouxiophyceae,Chlorophyta). J. Phycol. 43, 811–822.

Rindi, F., Sherwood, A.R., Guiry, M.D., 2005. Taxonomy and distribution ofTrentepohlia and Printzina (Trentepohliales, Chlorophyta) in the HawaiianIslands. Phycologia 44, 270–284.

Roberts, K.R., 1984. The flagellar apparatus of Batophora and Trentepohlia and itsphylogenetic significance. In: Irvine, D.E.G., John, D.M. (Eds.), Systematics of theGreen Algae. Academic Press, London, pp. 331–341.

Salleh, A., Milow, P., 1999. Notes on Trentepohlia dialepta (Nylander) Hariot(Trentepohliaceae, Chlorophyta) and sporangia of some other species ofTrentepohlia Mart. From Malaysia. Micronesica 31, 373–378.

Sarma, P., 1986. The freshwater Chaetophorales of New Zealand. Beih. NovaHedwigia 58, 1–169.

Shimodaira, H., 2002. An approximately unbiased test of phylogenetic treeselection. Syst. Biol. 51, 492–508.

Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods withapplications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116.

Smith, G., 1950. The Freshwater Algae of the United States. McGraw-Hill, New York.

F. Rindi et al. / Molecular Phylogenetics and Evolution 52 (2009) 329–339 339

Starr, R.C., Zeikus, J.A., 1987. UTEX—the culture collection of algae at the Universityof Texas at Austin. J. Phycol. 23 (Suppl.), 1–47.

Swofford, D.L., 1998. PAUP*. Phylogenetic Analysis Using Parsimony (*and OtherMethods). Version 4.0b10. Sinauer Associates, Sunderland, MA.

Thompson, R.H., Wujek, D.E., 1992. Printzina gen. nov. (Trentepohliaceae), includinga description of a new species. J. Phycol. 28, 232–237.

Thompson, R.H., Wujek, D.E., 1997. Trentepohliales: Cephaleuros, Phycopeltis andStomatochroon. Morphology, Taxonomy and Ecology. Science Publishers,Enfield, New Hampshire.

Tompkins, J., Deville, M.M., Day, J.G., Turner, M.F., 1995. Culture collection of algaeand protozoa. Catalogue of strains. The Culture Collection of Algae and Protozoa,Institute of Freshwater Ecology, Ambleside.

Van den Hoek, C., Mann, D.G., Jahns, H.M., 1995. Algae: An Introduction toPhycology. Cambridge University Press, Cambridge.

Verbruggen, H., Littler, D.S., Littler, M.M., 2007a. Halimeda pygmaea and Halimedapumila (Bryopsidales, Chlorophyta): two new dwarf species from fore reefslopes in Fiji and the Bahamas. Phycologia 46, 513–520.

Verbruggen, H., Leliaert, F., Maggs, C.A., Shimada, S., Schils, T., Provan, J., Booth, D.,Murphy, S., De Clerck, O., Littler, D.S., Littler, M.M., Coppejans, E., 2007b. Speciesboundaries and phylogenetic relationships within the green algal genus Codium(Bryopsidales) based on plastid DNA sequences. Mol. Phylogenet. Evol. 44, 240–254.

Verbruggen, H., Theriot, E.C., 2008. Building trees of algae: some advances inphylogenetic and evolutionary analysis. Eur. J. Phycol. 43, 229–252.

Wakasugi, T., Nagai, T., Kapoor, M., Sugita, M., Ito, M., Ito, S., Tsudzuki, J., Nakashima,K., Tsudzuki, T., Suzuki, Y., Hamada, A., Ohta, T., Inamura, A., Yoshinaga, K.,Sugiura, M., 1997. Complete nucleotide sequence of the chloroplast genomefrom the green alga Chlorella vulgaris: the existence of genes possibly involvedin chloroplast division. Proc. Natl. Acad. Sci. USA 94, 5967–5972.

Watanabe, S., Mitsui, K., Nakayama, T., Inouye, I., 2006. Phylogenetic relationshipsand taxonomy of sarcinoid green algae: Chlorosarcinopsis, Desmotetra,Sarcinochlamys, gen. nov., Neochlorosarcina and Chlorosphaeropsis(Chlorophyceae, Chlorophyta). J. Phycol. 42, 679–695.

Xia, X., Xie, Z., 2001. DAMBE: data analysis in molecular biology and evolution. J.Hered. 92, 371–373.

Xia, X., Xie, Z., Salemi, M., Chen, L., Wang, Y., 2003. An index of substitutionsaturation and its application. Mol. Phylogenet. Evol. 26, 1–7.

Zechman, F.W., 2003. Phylogeny of the Dasycladales (Chlorophyta, Ulvophyceae)based on analyses of Rubisco large subunit (rbcL) gene sequences. J. Phycol. 39,819–827.

Zeller, G., 1873. Algae collected by Mr. S. Kurz in Arracan and British Burma,determined and systematically arranged. J. Asiat. Soc. Bengal 42,175–193.