J. Phycol. 39, 775–788 (2003) NEW INSIGHTS IN THE TAXONOMY OF THE CERAMIUM SINICOLA COMPLEX: RESURRECTION OF CERAMIUM INTERRUPTUM (CERAMIACEAE, RHODOPHYTA) 1 Tae Oh Cho,2 Suzanne Fredericq Department of Biology, University of Louisiana at Lafayette, Lafayette, Louisiana 70504-2451, USA Steven N. Murray Department of Biological Science, California State University, Fullerton, California 92834-6850, USA and Sung Min Boo Department of Biology, Chungnam National University, Daejon 305-764, Korea in the degree of cortication, with C. interruptum having interrupted cortication at branching points and in lower thallus parts and with C. sinicola having incomplete cortication only in the lower thallus part. Dawson (1944) recognized these species as varieties within the C. sinicola complex after comparing the degree of cortication in these taxa. Dawson’s taxonomic decision has subsequently been widely adopted in the literature (Abbott and Hollenberg 1976). Ceramium codicola has been reported from the northeast Pacific and is a well-defined species. Although C. codicola is distinguished by bulbous and pigmented rhizoids and by being a strict epiphyte on Codium, this species strongly resembles C. sinicola in gross morphological features, habitat, and distribution (Dawson 1950, Abbott and Hollenberg 1976, Cho et al. 2001). Both species have small and densely corticated thalli and are epiphytic on Codium along southern California and northern Baja California where their ranges overlap (Dawson 1950). We recently collected C. interruptum, C. sinicola, and C. codicola in all reproductive stages from a littoral tide pool and subtidal zone in the northeast Pacific. We reassess their taxonomic ranks and determine evolutionary trends in cortication patterns and attachment modes of these three species as inferred from a comparative morphological study and three molecular data sets (plastid-encoded rbcL, the RUBISCO spacer, and nuclear encoded small subunit [SSU] rDNA sequences). Phylogenetic relationships of the Ceramium sinicola complex (C. interruptum and C. sinicola) including C. codicola were studied using nucleotide sequences of rbcL and small subunit rDNA, and the RUBISCO spacer was used for sequence comparison of each species. A reassessment of the taxonomic rank and the evolutionary trend within the complex was inferred from a comparative morphological study and molecular data sets based on 11 samples from eight populations from the Pacific coast of the United States and Mexico. Intraspecific relationships were poorly resolved, but the resurrection of C. interruptum as a distinct species was strongly supported by both morphological and molecular data. Ceramium interruptum is distinguished by the combination of the following features: thalli uncorticated at the first internode above the dichotomy, presence of four corticating filaments, 7–11 segments between branching points, rhizoids digitate, and epiphytic on a variety of hosts. Our molecular analyses show that C. sinicola is the sister group to C. codicola, and C. interruptum is basal to them. These phylogenetic relationships allowed for an assessment of the trend in the evolution of cortication pattern and attachment mode to the host. Key index words: Ceramium; C. interruptum; C. sinicola; Ceramiales; morphology; phylogeny; Rhodophyta; rbcL; RUBISCO spacer; SSU rDNA; taxonomy Setchell and Gardner (1924) provided a taxonomic treatment for the genus Ceramium from Baja California and the Gulf of California, Mexico, in which they described several new species, including C. interruptum from the vicinity of La Paz, Baja California Sur, and C. sinicola from Ensenada, Baja California Norte. These species are closely related by having incomplete cortication in the lower thallus region but are separated by only one or two distinguishing features materials and methods Morphology. Samples of C. interruptum, C. sinicola, and C. codicola were collected on the Pacific coast of the United States and Mexico and sorted according to morphology and host under a stereomicroscope. Each identified sample was preserved in 4% formaldehyde/seawater for morphological observations. Microscopic observations were made from materials stained with 1% aqueous aniline blue acidified with 0.1% diluted HCl. A total of 25 individuals from 10 tufts collected from each site were selected for measurement of quantitative characters. Herbarium abbreviations follow Holmgren et al. (1990). DNA extraction, amplification, and sequencing. Replicate samples for DNA extraction were desiccated in silica gel after identification under a stereomicroscope in the laboratory. Genomic DNA was extracted using a hexadecyltrimethylammonium bro- 1 Received 2 Author 10 October 2002. Accepted 2 April 2003. for correspondence: e-mail txc7221@louisiana.edu 775 776 TAE OH CHO ET AL. mide (CTAB) method (Doyle and Doyle 1987) as presented in Cho et al. (2003). Amplifying (F7-R753 and F645-RrbcS start) and sequencing (F7, F645, R753, RrbcS start) primers of the rbcL are listed in Freshwater and Rueness (1994), Lin et al. (2001), and Gavio and Fredericq (2002). PCR and sequencing protocols are as described in Lin et al. (2001). Primers (G01-G10, G02-G14, G04-G13, G06G07) of the SSU rDNA, developed by Saunders and Kraft (1994), were used in the present study, and the PCR and sequencing protocols are as described in Cho et al. (2003). The primers for the RUBISCO spacer are listed in Brodie et al. (1998). PCR conditions consisted of 4 min at 94 C for denaturation, followed by 25 cycles of 1 min at 94 C, 1 min at 40 C, and 2 min at 72 C with a final 10 min extension cycle at 72 C. Sequences were determined for both forward and reverse strands using an ABI Prism 3100 Genetic Analyzer (PE Applied Biosystems, Foster City, CA, USA) with the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit (PE Applied Biosystems). Alignment and phylogenetic analyses. All generated sequence data of rbcL, SSU rDNA, and the RUBISCO spacer were compiled (Table 1), and the data sets for each marker were manually aligned with Sequencher (Gene Codes Corp., Ann Arbor, MI, USA), then exported for analysis. The rbcL and SSU rDNA sequences were compiled together with published sequences (Cho et al. 2003). Phylogenetic analyses of the rbcL, SSU rDNA, and the concatenated rbcL  SSU rDNA data were performed using the maximum likelihood (ML) and maximum parsimony (MP) algorithms available in PAUP* (v. 4.0b10, Swofford 2002). A partition homogeneity test of concatenated rbcL  SSU rDNA sequences was performed (Farris et al. 1994) as implemented in PAUP using 1000 replications with the heuristic search option. For the ML analyses, the aligned sequences were first analyzed with the software Modeltest (v. 3.0, Posada and Crandall 1998), which compared different models of DNA substitutions in a hierarchical hypothesis-testing framework to select a base substitution model that best fits our sequence data. The optimal model for the rbcL sequence was a GTR (general time reversible model, Rodriguez et al. 1990)  G (gamma distribution). The parameters were as follows: assumed nucleotide frequencies A  0.3200, C  0.1522, G  0.2108, T  0.3169; Table 1. Taxa and collecting information of the Ceramium species used in the analyses of rbcL, SSU, rDNA, and the RUBISCO spacer with their Gen Bank Accession numbers. Species Regiona C. codicola J. Agardh BB-V C. codicola BB-T C. interruptum Setchell et Gardner CCSP-V C. interruptum CCSP-T C. interruptum CCSP C. interruptum DP C. interruptum BCS C. sinicola Setchell et Gardner CCSP-V C. sinicola CCSP-T C. sinicola DP C. sinicola LB Collecting information (location; date; collector; stages) Boiler Bay, Oregon, USA; 8. vii. 1998; T. O. Cho & G. I. Hansen Boiler Bay, Oregon, USA; 8. vii. 1998; T. O. Cho & G. I. Hansen; Tetrasporic Crystal Cove State Park, California, USA; 5. xii. 1999; T. O. Cho & S. Murray; Vegetative Crystal Cove State Park, California, USA; 5. xiii. 1999; T. O. Cho & S. Murray; Tetrasporic Crystal Cove State Park, California, USA; 5. xiii. 1999; T. O. Cho & S. Murray Dana Point, California, USA; 4. xii. 1999; T. O. Cho & S. Murray EL Tecolote, Baja California Sur, Mexico; 20. v. 2000, T. O. Cho & R. R-. Rodriguez Crystal Cove State Park, California, USA; 5. xii. 1999; T. O. Cho & S. Murray; Vegetative Crystal Cove State Park, California, USA; 5. xii. 1999; T. O. Cho & S. Murray; Tetrasporic Dana Point, California, USA; 4. xii. 1999; T. O. Cho & S. Murray Laguna Beach, California, USA; 3. xii. 1999; T. O. Cho & S. Murray Gamp, Kyungbuk, Korea; 17. ix. 1998; S. M. Boo & T. O. Cho Host rbcL SSU rDNA RUBISCO spacer Codium fragile AY155522 AY155510 AY155236 Codium fragile AY155523 AY155511 AY155237 Codium fragile AY155524 AY155512 AY155238 Codium fragile AY155525 AY155513 AY155239 Corallina sp. AY155526 AY155514 AY155240 Gelidium sp. AY155527 AY155515 AY155241 Corallina sp. AY155528 AY155516 AY155242 Codium fragile AY155529 AY155517 AY155243 Codium fragile AY155530 AY155518 AY155244 Codium fragile AY155531 AY155519 AY155245 Codium fragile AY155532 AY155520 AY155246 Centroceras clavulatum — — AY155533 AY155521 — (C. Agardh) Montagne Ceramium affine Setchell — — AF521797a AF460859a — et Gardner C. horridum Setchell — — AF521796a AF460858a — et Gardner C. inkyuii Cho, Fredericq — — AF521798a AF460860a — et Boo C. paniculatum Okamura — — AF521802a AF460864a — C. tenerrimum (Martens) — — AF521804a AF460866a Okamura a From Cho et al. (2003). BB-V, Boiler Bay–vegetative; BB-T, Boiler–tetrasporangial; CCSP-V, Crystal Cove State Park–vegetative; CCSP-T, Crystal Cove State Park–tetrasporangial; CCSP, Crystal Cove State Park; DP, Dania Point; BCS, Baja California Sur; LB, Laguna Beach. CERAMIUM SINICOLA COMPLEX 777 Fig. 1. Ceramium interruptum (a–j  CNUK C02138). (a) Thallus showing pseudodichotomous branching and interrupted cortication (arrow). Scale bar, 200 m. (b) Apical region with spines. Scale bar, 50 m. (c) Cross-section through node showing axial and periaxial cells. Scale bar, 100 m. (d) Alternate sequence formation of four primary cortical cells from periaxial cell. Scale bar, 20 m. (e) Complete cortication in upper thallus part. Scale bar, 100 m. (f) Incomplete cortication in lower thallus part. Scale bar, 100 m. (g) Interrupted cortication just above branching point. Scale bar, 100 m. (h) Longitudinal section through upper thallus part. Scale bar, 100 m. (i) Incompletely corticated axis (arrow) in lower thallus part. Scale bar, 200 m. (j) Cross-section through prostrate part showing rhizoids with digitate tips. Scale bar, 100 m. Ax, axial cell; C1–4, sequence of cortical cell formation; P, periaxial cell. 778 TAE OH CHO ET AL. substitution rate matrix with A-C substitutions  1.3341, A-G  4.1845, A-T  3.7774, C-G  0.7365, C-T  25.5051, G-T  1.0000; rates for variable sites assumed to follow a gamma distribution with shape parameter  0.1513. The optimal model of SSU rDNA was a TrN  I (Invariable sites)  G. The parameters were as follows: assumed nucleotide frequencies A  0.2477, C  0.2096, G  0.2807, T  0.2620; substitution rate matrix with A-C substitutions  1.0000, A-G  2.6188, A-T  1.0000, C-G  1.0000, C-T  5.7923, G-T  1.0000; proportion of sites assumed to be invariable  0.7680 and rates for variable sites assumed to follow a gamma distribution with shape parameter  0.6635. The optimal model of rbcL  SSU rDNA was TrN  I  G. The parameters were as follows: assumed nucleotide frequencies A  0.2808, C  0.1875, G  0.2475, T  0.2843; substitution rate matrix with A-C substitutions  1.0000, A-G  3.3343, A-T  1.0000, C-G  1.0000, C-T  10.6558, G-T  1.0000; proportion of sites assumed to be invariable  0.7733 and rates for variable sites assumed to follow a gamma distribution with shape parameter  0.8339. The ML tree was generated by a heuristic search of 100 random additions holding one tree at each step under the invoked settings for the respective base substitution model. Support for nodes was determined by calculating bootstrapping proportion values (Felsenstein 1985) using 100 bootstrap replicates for the ML analysis. MP trees of the rbcL, SSU rDNA, and rbcL  SSU rDNA were inferred from a heuristic search option. For the MP analyses, the uninformative characters were also excluded and support for nodes was determined by calculating bootstrapping proportion values (Felsenstein 1985) using 1000 replicates for MP analyses. results Ceramium interruptum Setchell et Gardner 1924, p. 775 (Figs. 1 and 2) Holotype: Marchant, no. 78m, UC. Type locality: Eureka, near La Paz, BCS, Mexico. Representative specimens: Crystal Cove State Park, Orange Co. California, USA (T. O. Cho and S. N. Murray, 05.xii.1999, CNUK 002138, 002440, vegetative, CNUK 002131-002132, 002439, tetrasporic, CNUK 002293, 002400, male and female), Dana Point, Orange Co., California, USA (T. O. Cho and S. N. Murray, 04.xii.1999, CNUK TC310, TC207, vegetative); El Tecolote, Baja California Sur, Mexico (T. O. Cho and R. R. Rodriguez, 20.v.2000, CNUK TC265, tetrasporic). Thalli are delicate, rose-pink in color, 0.8–1.5 cm high (Fig. 1a), consisting of interwoven prostrate axes giving rise to erect pseudodichotomous axes. Erect axes bear elongate, forcipulate and incurved apical regions with multicellular spines (Fig. 1b). Axial cells are spherical to cylindrical, reaching 158  20  141  15 m at the level of the fourth dichotomy away from the apex. Seven to nine periaxial cells are cut off obliquely from the upper part of each parent axial cell and remain at the nodes after axial cell elongation (Fig. 1c). All periaxial cells produce corticating filaments that form the cortex. Four primary cortical cells are produced from each periaxial cell in an alternate sequence and develop into corticating filaments (Fig. 1d). The first two cortical cells are cut off obliquely from the anterior end of the periaxial cells and grow acropetally; the remaining two are produced obliquely from the posterior end and grow basipetally. Both the acropetal and basipetal corticating filaments are 10–13 cells long. The cortication of erect axes is complete (Fig. 1, a, e, and h) except at the first internode above a fork (Fig. 1, a and g) and in the lower thallus part (Fig. 1, f and i). The cortication of the interwoven prostrate axes is incomplete, covering the axial cells only at the nodes. Branches are regularly pseudodichotomous. Branching takes place at intervals of 7–10 (average, 8.4  1.2) axial cells in the main axes and at the intervals of 8–11 (average, 8.9  1.4) cells in the lateral axes. In addition, adventitious branches rarely develop from periaxial cells in the lower thallus parts. A spine sometimes occurs on the abaxial face near the apex. Each spine is composed of three to four cells with a multicellular base (Fig. 1b). Gland cells develop from cortical cells of acropetally and basipetally corticating filaments and are ovoid to angular, averaging 13  3  12  2 m. Rhizoids attached to Corallina and other algae are multicellular, uniseriate, and basally digitate (Fig. 1j). Rhizoids on Codium fragile (Suringar) Hariot are slender, branched and nonbulbous, of varying length depending on the texture of the host, and are produced from periaxial and cortical cells of interwoven prostrate axes. Spermatangia are first produced adaxially and later make a whorl around the cortical axis (Fig. 2, a and b). Spermatangial parent cells develop from cortical cells and produce one to two spermatangia terminally (Fig. 2c). Spermatangia are colorless and elliptical to spherical, measuring 4  1  3  1 m in size. Cystocarps are borne on the abaxial face of the upper thallus and are surrounded by five to seven involucral branchlets with incomplete cortication (Fig. 2, d and e). Mature cystocarps are spherical, 225  62 m long and 252  50 m in diameter, and composed of carposporangia transformed from gonimolobe cells. Tetrasporangia are completely immersed in the cortex (Fig. 2, f and g), whorled at the nodes in the upper part, and scattered irregularly in the middle thallus region (Fig. 2g). They are produced initially from periaxial cells in the upper part (Fig. 2h) and then from inner cortical cells in the mid region (Fig. 2i). One tetrasporangium develops per tetrasporangial initial (Fig. 2, h and i). Tetrasporangia are tetrahedrally divided, spherical to ellipsoidal, and average 51  5  40  5 m excluding the sheath and 57  4  47  7 m with the sheath. Ceramium interruptum is epiphytic on a diverse range of hosts such as articulated coralline algae (e.g. Corallina pinnatifolia (Manza) Dawson, C. vancouveriensis Yendo, and Lithothrix aspergillum Gray), Gelidium spp., Laurencia spp., and Codium fragile and occurs in the intertidal to subtidal zones of wave-exposed sites. It also was found to occur mixed in with C. sinicola on Codium fragile at Crystal Cove State Park, Orange Co., California. Ceramium sinicola Setchell et Gardner 1924, p. 773 (Figs. 3 and 4) Holotype: No. 1366, Herb. Calif. Acad. Sci., collected by Ivan M. Johnston (No. 67b). CERAMIUM SINICOLA COMPLEX 779 Fig. 2. Ceramium interruptum (a–c  CNUK C002440, d–e  CNUK C002400, f–i  CNUK C002439). (a) Male thallus with spermatangia. Scale bar, 200 m. (b) Surface of axis bearing spermatangia. Scale bar, 50 m. (c) Cross-section through thallus showing formation of spermatangia. Scale bar, 50 m. (d) Female thallus bearing cystocarps. Scale bar, 200 m. (e) Cystocarp surrounded by involucral branches. Scale bar, 100 m. (f) Tetrasporangial thallus. Scale bar, 200 m. (g) Corticated axis showing tetrasporangia completely immersed in cortex. Scale bar, 100 m. (h) Cross-section through node of tetrasporic axis. Scale bar, 50 m. (i) Longitudinal section through thallus with tetrasporangia. Scale bar, 100 m. Ax, axial cell; Cy, cystocarp; Iv, involucral branches; P, periaxial cell; S, spermatangium; T, tetrasporangium. 780 TAE OH CHO ET AL. Fig. 3. Ceramium sinicola (a–j  CNUK C02510). (a) Vegetative thallus. Scale bar, 500 m. (b–c) Apical regions. Scale bars, 40 m. (d) Cross-section through thallus showing axial, periaxial, and cortical cells. Scale bar, 40 m. (e) Alternate sequence formation of five primary cortical cells from periaxial cell. Scale bar, 40 m. (f) Upper thallus part with complete cortication. Scale bar, 100 m. (g) Lower thallus part with incomplete cortication. Scale bar, 100 m. (h) Branching point with complete cortication. Scale bar, 100 m. (i) Lower thallus with irregular branches. Scale bar, 200 m. (j) Rhizoids showing slender, rod-shaped, and non-bulbous rhizoids. Scale bar, 100 m. Ax, axial cell; C, cortical cell; C1–5, sequence of cortical cell formation; P, periaxial cell. CERAMIUM SINICOLA COMPLEX 781 Fig. 4. Ceramium sinicola (a–j  CNUK C002359). (a) Male thallus. Scale bar, 200 m. (b) Surface of cortex bearing spermatangia. Scale bar, 40 m. (c) Cross-section through male thallus showing layer of spermatangia completely surrounding thallus. Scale bar, 100 m. (d) Female thallus bearing cystocarps. Scale bar, 200 m. (e) Gonimolobe, gonimoblast, and fusion cell formed from fusion product of supporting cell, foot cell, and axial cell. Scale bar, 40 m. (f) Mature cystocarp surrounded by involucral branches. Scale bar, 100 m. (g) Tetrasporangial thallus. Scale bar, 200 m. (h) Detail of cortex showing tetrasporangia completely immersed in cortex. Scale bar, 40 m. (i) Cross-section through node of tetrasporic axis. (j) Cross-section through internode of tetrasporic axis. Scale bars, 100 m. Ax, axial cell; Cy, cystocarp; Fu, fusion cell; G, gonimoblast; Gl, gonimolobe; Iv, involucral branches; S, spermatangium, T, tetrasporangium. 782 TAE OH CHO ET AL. Fig. 5. Ceramium codicola (a–g  CNUK C00084). (a) Vegetative thallus. Scale bar, 200 m. (b) Apical region. Scale bar, 40 m. (c) Cross-section through node showing axial, periaxial, and cortical cells. Scale bar, 100 m. (d) Alternate sequence formation of five primary cortical cells from periaxial cell. Scale bar, 40 m. (e) Branching point with complete cortication. Scale bar, 100 m. (f) Lower thallus part. Scale bar, 200 m. (g) Detail of rhizoidal cells showing bulbous rhizoid (arrows). Scale bar, 100 m. Ax, axial cell; C1–5, sequence of cortical cell formation; P, periaxial cell. Type locality: Ensenada Bay, Baja California Norte Mexico. Representative specimens: Crystal Cove State Park, Orange Co., California, USA (T. O. Cho and S. N. Murray; 05.xii.1999, CNUK TC202, male and female, CNUK TC297, tetrasporic); Laguna Beach, Orange Co., California, USA (T. O. Cho and S. N. Murray; 03. xii; 1999, CNUK 002510, vegetative, CNUK TC226, tetrasporic); Dana Point, Orange Co., California, USA (T. O. Cho and S. N. Murray; 04.xii.1999, CNUK 002359, male and female, CNUK TC299, vegetative, male, female and tetrasporic). Thalli are delicate, rose-pink in color, 0.3–0.4 cm high (Fig. 3a). They mostly consist of erect pseudodichotomous axes and short interwoven prostrate axes bearing numerous adventitious branches. Erect axes bear blunt, forcipulate, and incurved apical regions (Fig. 3, b and c). The axial cells are spherical to cylindrical and reach 163  14  126  20 m at the level of the third dichotomy away from the apex. Seven to eight periaxial cells are cut off obliquely from the upper part of each parent axial cell and remain at the nodes after axial cell elongation (Fig. 3d). All periaxial cells produce corticating filaments that form the cortex. Five primary cortical cells are produced from each periaxial cell in an alternate sequence (Fig. 3e). The first two primary cortical cells are cut off obliquely from the anterior end of periaxial cells and grow acropetally; the second two are produced obliquely from CERAMIUM SINICOLA COMPLEX the posterior end and grow basipetally; the fifth one is produced on the outside face of periaxial cells. The acropetal and basipetal corticating filaments are both five to seven cells long. The mature cortex at the upper part of erect axes is complete (Fig. 3, f and h), whereas the cortex in the interwoven prostrate axes and the base of the erect axes is incomplete (Fig. 3g). Branches are pseudodichotomous. Branching takes place at intervals of six to eight (average, 6.7  1.0) axial cells in the main axes and at the intervals of six to eight (average, 6.6  0.9) cells in the lateral axes. In addition, many adventitious branches develop from periaxial cells of erect axes. Rhizoids are multicellular, uniseriate, pigmented, rod-shaped, and branched near the tip. They are slender, branched and non-bulbous, produce extensions in bundles on Codium fragile, and are of various lengths depending upon the texture of the host (Fig. 3, i and j). Rhizoids are produced from periaxial and cortical cells of short interwoven prostrate axes and are located on the basal region of erect axes. Spermatangia are produced adaxially first and later make a whorl around the cortical axis later (Fig. 4, a and b). Spermatangial parent cells develop from cortical cells and produce one to two spermatangia terminally (Fig. 4c). Spermatangia are colorless and elliptical to spherical, measuring 3  1  3  1 m in size. Cystocarps are borne on the abaxial face of the upper thallus and are surrounded by five to six involucral branchlets with incomplete cortication (Fig. 4, d and f). After presumed fertilization, the supporting cell enlarges, cutting off an auxiliary cell, with the auxiliary cell then dividing into a foot cell and a gonimoblast initial and the supporting cell, foot cell, and axial cell 783 fusing into a fusion cell product (Fig. 4e). The first gonimolobe is cut off terminally (Fig. 4e) from the gonimoblast initial, followed by the production of one or two additional gonimolobe initial laterally. Mature cystocarps are spherical, 298  64 m long and 253  60 m in diameter, and composed of carposporangia transformed from gonimolobe cells. Tetrasporangia are completely immersed in the cortex (Fig. 4, g and h). They are whorled at the nodes of the upper part and scattered irregularly in the middle thallus region. They are produced initially from periaxial cells in the upper part (Fig. 4i) and then from inner cortical cells in the middle part (Fig. 4j). One tetrasporangium develops from each tetrasporangial parental cell (Fig. 4i). Tetrasporangia are tetrahedrally divided, spherical to ellipsoidal, and average 42  13  37  8 m, excluding the sheath. Ceramium sinicola is epiphytic and restricted to its host Codium fragile (Suringar) Hariot in the subtidal zone of wave-exposed sites. We found it to occur mixed in with C. interruptum on Codium fragile at Crystal Cove State Park, Orange Co., California. Ceramium codicola J. Agardh 1894, p. 23 (Figs. 5 and 6) Thalli of C. codicola are 0.8–1.8 cm high and pseudodichotomously branched (Fig. 5a). Erect axes have short, blunt, forcipulate, and incurved apical regions (Fig. 5b). Periaxial cells are oblique and five to six in number (Fig. 5c). All periaxial cells produce corticating filaments forming the cortex. Five primary cortical cells are produced from each periaxial cell in an alternate sequence (Fig. 5d). The first two are cut off Fig. 6. Maximum likelihood trees inferred from 16 sequences of eight Ceramium species and one specimen of Centroceras clavulatum, used as the outgroup. (a) Phylogenetic inference from rbcL sequence data matrix. Ln likelihood of the tree was 4038.0029. (b) Phylogenetic inference from SSU rDNA sequence data matrix. Ln likelihood of the tree was 3521.0171. Bootstrap proportion values ( 50%) are shown for all ML (top) and MP (bottom) analyses. 784 TAE OH CHO ET AL. obliquely from the anterior end of periaxial cells and grow acropetally. The second two are produced obliquely from the posterior end and grow basipetally. The last one is produced on the outside face of the periaxial cell. The cortex is complete across the entire thallus (Fig. 5, e and f). Rhizoids are produced near the base and are multicellular, uniseriate, pigmented, slender, rarely branched, forming bundles that are bulbous at the tip. The rhizoids are of various length depending upon the texture of the host, Codium fragile (Fig. 5g). Rhizoids are cut off from periaxial and cortical cells at the base of erect axes. Phylogenetic analyses. The 1431-bp portion of the 1467-bp rbcL gene (97.5% sequenced) analyzed included 181 parsimony informative sites. Ceramium interruptum differed from C. sinicola by 3.5%–3.8% sequence divergence and from C. codicola by 4.2%–4.3% sequence divergence (Table 2). There was a difference of 1.2%–1.5% sequence divergence between C. sinicola and C. codicola. Three (CCSP on Codium, CCSP on Corallina, and DP on Gelidium) of the four C. interruptum populations were identical, and two (LB and DP) of three C. sinicola populations had identical sequences; the Baja California Sur population of C. interruptum differed from other populations by 0.5% sequence divergence, whereas the Crystal Cove population of C. sinicola differed from other populations by 0.8% sequence divergence. The sequences from different stages (vegetative and tetrasporic) within a population did not differ. In the SSU rDNA data set, a 1727-bp portion was sequenced that included 103 informative sites. Ceramium interruptum differed from C. sinicola by 0.8%–0.9% sequence divergence and from C. codicola by 0.8% sequence divergence. There was a difference of one to two nucleotides or 0.1%–0.06% sequence divergence between C. sinicola and C. codicola (Table 2). Sequences from the four C. interruptum populations did not differ from each other in any position; likewise, the Crystal Cove population of C. sinicola differed from the other two populations by one nucleotide or 0.06% sequence divergence. The sequences from different reproductive stages within a population did not differ from each other. The 105-bp RUBISCO spacer was completely sequenced for the C. interruptum, C. sinicola, and C. codicola populations. The RUBISCO spacer sequences for C. interruptum differed from C. sinicola by 2.9% sequence divergence and from C. codicola by 24% sequence divergence. The RUBISCO spacer sequences of C. sinicola were also different from that of C. codicola Table 2. Comparison of rbcL, SSU rDNA, and RUBISCO spacer gene sequence divergence values (%) between species. rbcL SSU rDNA RUBISCO spacer C. interruptum : C. sinicola C. interruptum : C. Codicola C. interruptum : C. codicola 3.5–3.8 0.8–0.9 2.9 4.2–4.3 0.8 24.0 1.2–1.5 0.06–0.1 24.0 by 24% sequence divergence (Table 2). Each population was identical to one another within each species. The RUBISCO spacer sequences of C. codicola were anomalous in that it contained two large indels (deletions) that were not present in the C. interruptum and C. sinicola vouchers. Because the indel in C. codicola remains unexplained, all RUBISCO spacer sequences were removed from the phylogenetic analysis. ML phylogenetic trees (Fig. 6) were obtained from the alignment of the separate rbcL and SSU rDNA sequences. Centroceras clavulatum was selected as the outgroup in the analyses because Centroceras is a closely related genus based on morphological traits. The ML trees of both the rbcL (Fig. 6a) and SSU rDNA (Fig. 6b) data for 16 taxa of Ceramium were composed of two clades with robust bootstrap support, one containing the assemblage C. horridum, C. interruptum, C. sinicola, and C. codicola, the other including C. affine, C. paniculatum, C. tenerrimum, and C. inkyuii. All samples of C. interruptum were also placed in a strongly supported (99% for ML, 100% for MP [rbcL]; 96% for ML, 99% for MP [SSU rDNA]) monophyletic clade clearly separated from both C. sinicola and C. codicola. The BCS population of C. interruptum was basal to the other populations. Whereas all the nodes of rbcL tree (Fig. 6a) were supported by bootstrap values 50%, most nodes of the SSU rDNA tree (Fig. 6b) were not. Parsimony analyses of both rbcL and SSU rDNA produced trees of identical topology to the ML results. A ML phylogenetic tree (Fig. 7) was obtained from the alignment of the combined rbcL  SSU rDNA after performing the partition homogeneity test (P  0.53). Ceramium horridum was used as outgroup because it was the sister taxon in the analyses of each sequence data set. This combined tree revealed three clades, all of which are well supported by high bootstrap. Especially, the well-supported clustering of C. interruptum populations indicates this species is distinguished by four cortical filaments, incomplete cortication on branching point and base, and digitate rhizoids (Fig. 7). Parsimony analyses produced a tree of similar topology to the maximum likelihood result. discussion Ceramium interruptum is distinguished from C. sinicola by the combination of the following features: thalli interrupted at the first internode above the dichotomy, presence of four corticating filaments, 7–11 segments between branching points, rhizoids digitate, and epiphytic on a variety of hosts (Table 3). The reinstatement of C. interruptum is supported by the phylogenetic analyses inferred from rbcL and SSU rDNA sequences and by sequence comparison with the RUBISCO spacer. The important diagnostic feature in Ceramium is the developmental pattern of the cortical cells from periaxial cells (Womersley 1978). The cortical filaments in Ceramium are typically composed of three to five cells (Womersley 1978, Cho et al. 2001). In most Ceramium species, including C. interruptum, each of CERAMIUM SINICOLA COMPLEX 785 Fig. 7. Maximum likelihood tree inferred from the combined rbcL  SSU rDNA sequence data matrix of the Ceramium sinicola complex, with distribution of character states among taxa used in this study. Ceramium horridum was used as an outgroup, and ln likelihood of the tree was –5237.1226. Bootstrap proportion values ( 50%) are shown for ML (top) and MP (bottom) analyses. the periaxial cells typically cuts off two primary cortical cells acropetally and also two basipetally (Hommersand 1963, Womersley 1978, Cho et al. 2001), but in some species (e.g. C. recticorticum Dawson), there are three primary cortical cells, two cut off acropetally and one basipetally (Cho et al. 2002). However, C. sinicola and C. codicola each possess five primary cortical cells, two cut off acropetally, two basipetally, and the fifth cut off on the outside face of the periaxial cells. Therefore, the number and the development of corticating filaments is a valid taxonomic character not only for grouping some species (Cho et al. 2002), but also for distinguishing two similar-looking species (Cho et al. 2003) such as C. interruptum and C. sinicola. Our observations indicate that the interrupted cortication at the branching point in C. interruptum is stable throughout all developmental stages and samples and clearly represents one of the most important distinguishing features characterizing this species. The degree of cortication has been used as a significant diagnostic feature for species level taxonomy in the genus Ceramium (e.g. Dixon 1960, Womersley 1978, Cho et al. 2000, 2001), although it may depend on the grow- ing season and microhabitat in some species (Garbary et al. 1978, Cormaci and Motta 1987). According to the original descriptions of Setchell and Gardner (1924), C. interruptum, C. johnstonii Setchell et Gardner, and C. sinicola may be distinguished from one another only by the degree of cortication and the developmental formation of tetrasporangia. Dawson (1944, 1950) compared the degree of cortication in these three species and was of the opinion that they formed a series within a single species: C. sinicola var. typicum, C. sinicola var. interruptum, C. sinicola var. johnstonii. However, C. interruptum is clearly distinguished from C. sinicola by the interrupted cortication just above the branch forkings, although its tetrasporangial deposition is similar to that of C. sinicola. Some samples of C. sinicola have slightly separating cortication at the upper branches, numerous lateral branches, and tetrasporangia formed in two to three whorls, which were characteristic for C. johnstonii according to Setchell and Gardner (1924). Therefore, C. johnstonii may be synonymous with C. sinicola. Some Ceramium species have distinct rhizoids that are closely related to the texture of the host. Ceramium 786 Table 3. TAE OH CHO ET AL. Comparison of morphological features among Ceramium interruptum, C. sinicola, and C. codicola. C. interruptum (this study) Thallus length (cm) Habit Cortication of Erect thallus Basal thallus part No. of cortical filaments Direction of dominant corticating filaments No. of periaxial cells No. of axial cells between branches along main axis No. of axial cells between branches along main axis Spines Tip of rhizoids Size of tetrasporangia (m) Size of cystocarps (m) Host Habitat C. sinicola (this study) 0.9  0.2 Creeping and erect Erect 0.3  0.1 Interrupted Incomplete 4 Acro- and basipetal Mostly complete Incomplete 5 Acro- and basipetal C. codicola (Cho et al. 2001) Erect 1.0  0.4 Complete Complete Acropetal 5 7–9 8.4  1.2 7–8 6.7  1.0 5–6 8.3  1.4 8.9  1.4 6.6  0.9 7.4  1.4 Present Mostly digitate 51  5  40  5 225  62  252  50 Articulated coralline algae, Gelidium spp., Codium fragile Intertidal and subtidal zone Present Mostly rod-shaped 42  13  37  8 298  64  253  60 Codium fragile Absent Bulbous 47  4  37  5 380  72  360  76 Codium fragile Subtidal zone Subtidal zone interruptum is epiphytic on a variety of hosts such as Corallina, Gelidium, and Codium and has two types of rhizoidal shape that follow the texture of the host’s surface: on hosts with a compact cortex like Gelidium and Corallina, the rhizoids have mostly digitate tips, whereas such tips are absent when epiphytic on Codium fragile. Our C. sinicola samples were mostly collected on Codium fragile, although Setchell and Gardner (1924) found this species to be entangled with Laurencia. Ceramium sinicola produces slender and rod-shaped rhizoids that penetrate the texture of the host to which they attach. Ceramium codicola is strictly epiphytic on Codium fragile; it produces bulbous rhizoids for strong attachment among the utricles of the latter, most likely a result of successful adaptation to the morphology of its specific host. Sites of tetrasporangial initiation and the degree of cortication may be closely related in Ceramium species (Womersley 1978, Cho et al. 2001). In the completely corticated species such as C. codicola, C. kondoi Yendo, and C. pacificum (Collins) Kylin, tetrasporangia develop from both periaxial and inner cortical cells, whereas in the incompletely corticated species such as C. cimbricum H. Petersen in Rosenvinge, C. californicum J. Agardh, and C. gardneri Kylin, they develop only from periaxial cells (Cho et al. 2001). Although cortication is interrupted on the basal thallus, tetrasporangial development in C. interruptum and C. sinicola reveals the same features as species having complete cortication. Tetrasporangia of C. interruptum and C. sinicola are produced from both periaxial and inner cortical cells, and one tetrasporangium originates from a single parental cell. First, tetrasporangia are produced in a single whorl at the nodes of the upper axis and then in two to three whorls on the middle axis. Setchell and Gardner (1924) characterized C. sinicola as having a single whorl of tetrasporangia and used this as one of the key characters discriminating this taxon. However, our observa- tions indicate that this tetrasporangial arrangement cannot be used to distinguish this species. Our sequences of rbcL, SSU rDNA, and the RUBISCO spacer revealed sufficient sequence divergence between C. interruptum and C. sinicola to warrant species recognition for C. interruptum. The interspecific rbcL sequence divergence (3.5%–3.8%) between C. interruptum and C. sinicola is similar to other Ceramium species (2.5%–3.8%) (Cho et al. 2003) and within the range of other red algae, for example, between species of Porphyra (1.3%) (Brodie et al. 1996) or Grateloupia (2%–10%) (Gavio and Fredericq 2002). The interspecific SSU rDNA sequence divergence between these species (0.8%–0.9%) is also similar to other Ceramium species (0.5%–1.0%) (Cho et al. 2003) and more than the interspecific differences (0%–0.4%) reported for the Gelidiales (Bailey and Freshwater 1997). RUBISCO spacer sequence divergence between C. interruptum and C. sinicola was 2.9%. These molecular results support the resurrection of C. interruptum as a distinct taxon from C. sinicola. The sequence matrices of rbcL and SSU rDNA revealed intraspecific variation, with rbcL differing by 0.5%–0.8% within populations of both C. interruptum and C. sinicola and with SSU rDNA differing by 0.06% within C. sinicola. The sequence divergence between and within species is higher in rbcL than in SSU rDNA. Higher interspecific sequence variation values for the rbcL than for SSU rDNA have been reported for the Gelidiales (Bailey and Freshwater 1997). Contrary to the general view that the RUBISCO spacer evolves faster than rbcL and SSU rDNA (Maggs et al. 1992), the RUBISCO spacer sequences of Ceramium obtained in this study did not exhibit any variation within a species. Hence, the RUBISCO spacer may not be a useful tool appropriate for distinguishing intraspecific relationships of Ceramium species. Based on morphological grounds, C. interruptum and C. sinicola have been considered closely related on the CERAMIUM SINICOLA COMPLEX 787 Fig. 8. Hypothetical evolution in number of corticating filaments (a), degree of cortication (b), and holdfast (c) among Ceramium interruptum, C. sinicola, and C. codicola. basis of cortication, and C. sinicola and C. codicola have been confused with one another because of their morphological similarity and occurrence on the same host (Setchell and Gardner 1924, Dawson 1950, Abbott and Hollenberg 1976). Our morphological data reveals that C. sinicola has character states that are intermediate between those of C. interruptum and C. codicola (incomplete cortication only on the base, mostly rodshaped rhizoids, and necessity of Codium fragile as a specific host); we view the paraphyletic nature of C. sinicola based on rbcL and SSU rDNA sequences in our molecular analyses as the result that some populations have recently speciated or are speciating in a taxon corresponding morphologically to C. codicola. We prefer to recognize the two groups of C. sinicola as belonging to one species because their RUBISCO spacer sequences are identical, and we recognize C. codicola as a species separated from C. sinicola based on significant rbcL and RUBISCO spacer gene sequence divergence values between it, C. interruptum and C. sinicola. Analysis of the combined rbcL  SSU rDNA data set allows us to assess the trend in the evolution of cortication pattern and attachment mode to the host. The ML tree from the combined data set (Fig. 7) and diagrams of morphological features (Fig. 8) reveal that the phylogenetic relationships of these species follow an evolutionary trend in the number of corticating filaments, degree of cortication, and shape of the rhizoid tips according to host as follows: cortication evolved toward an increase in the number of cortical filaments and complete cortication; host selection evolved toward one host; and the bulbous rhizoid tip of C. codicola may be the result of successful adaptation to its specific host, Codium fragile. T. O. C. thanks Dr. R. R. Rodriguez and Dr. G. I. Hansen for their logistical support in Baja California, Mexico and Oregon, USA. We thank W. T. Stam and two anonymous reviewers for improving the manuscript. This study was supported by a KOSEF postdoctoral fellowship to T. O. C., NSF grant DEB-9903900 to S. F., and KOSEF grant R01-2000-00074 (2002) to S. M. B. Abbott, I. A. & Hollenberg, G. J. 1976. Marine Algae of California. Stanford University Press, California. xii  827 pp. Agardh, J. G. 1894. Analecta algologica. Continuatio II. Lunds Univ. Årsskr. 30:1–99. Bailey, J. C. & Freshwater, D. W. 1997. Molecular systematics of the Ge- 788 TAE OH CHO ET AL. lidiales: inferences from separate and combined analyses of plastid rbcL and nuclear SSU gene sequences. Eur. J. Phycol. 32:343–52. Brodie, J., Hayes, P. K., Barker, G. L. & Irvine, L. M. 1996. Molecular and morphological characters distinguishing two Porphyra species (Rhodophyta: Bangiophycidae). Eur. J. Phycol. 31:303–8. Brodie, J., Hayes, P. K., Barker, G. L., Irvine, L. M. & Bartsch, I. 1998. A reappraisal of Porphyra and Bangia (Bangiophycidae, Rhodophyta) in the northeast Atlantic based on the rbcL-rbcS intergenic spacer. J. Phycol. 34:1069–74. Cho, T. O., Boo, S. M. & Hansen, G. I. 2001. Structure and reproduction of the genus Ceramium (Ceramiales, Rhodophyta) from Oregon, USA. Phycologia 40:547–71. Cho, T. O., Choi, H.-G., Hansen, G. & Boo S. M. 2000. Corallophila eatoniana comb. nov. (Ceramiaceae, Rhodophyta) from the Pacific coast of North America. Phycologia 39:323–31. Cho, T. O., Fredericq, S. & Boo, S. M. 2003. Ceramium inkyuii sp. nov. (Ceramiaceae, Rhodophyta) from Korea; a new species based on morphological and molecular evidence. J. Phycol. 39:237–47. Cho, T. O., Rodriguez, R. R. & Boo, S. M. 2002. Developmental morphology of a poorly documented alga, Ceramium recticorticum (Ceramiaceae, Rhodophyta), from the Gulf of California, Mexico. Cryptog. Algol. 23:277–89. Cormaci, M. & Motta, G. 1987. Osservazioni sulla morfologia di Ceramium rubrum (Hudson) C. Agardh (Rhodophyta, Ceramiales) in coltura e considerazioni sulla sua tassonomia. Bol. Acad. Gioenia Sci. Nat. 20:239–51. Dawson, E. Y. 1944. The marine algae of the Gulf of California. Allan Hancock Pac. Exped. 3:189–454. Dawson, E. Y. 1950. A review of Ceramium along the Pacific coast of North America with special reference to its Mexican representatives. Farlowia 4:113–38. Dixon, P. S. 1960. Studies on marine algae of the British Isles: the genus Ceramium. J. Mar. Biol. Assoc. UK 39:331–74. Doyle, J. J. & Doyle, J. L. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. 19:11–5. Farris, J. S., Källersjö, A. G., Kluge, A. G. & Bult, C. 1994. Testing significance of incongruence. Cladistics 10:315–9. Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783–91. Freshwater, D. W. & Rueness, J. 1994. Phylogenetic relationships of some European Gelidium (Gelidiales, Rhodophyta) species, based on rbcL nucleotide sequence analysis. Phycologia 33:187–94. Garbary, D. J., Grund, D. & McLachlan, J. 1978. The taxonomic status of Ceramium rubrum (Huds.) C. Ag. (Ceramiales, Rhodophyceae) based on culture experiments. Phycologia 17:85–94. Gavio, B. & Fredericq S. 2002. Grateloupia turuturu (Halymeniaceae, Rhodophyta) is the correct name of the non-native species in the Atlantic known as Grateloupia doryphora. Eur. J. Phycol. 37: 349–60. Holmgren, P. K., Holmgren, N. H. & Barnett, L.C. 1990. Index Herbariorum. I. The herbaria of the world. Regn. Vegetab. 120: 1–693. Hommersand, M. H. 1963. The morphology and classification of some Ceramiaceae and Rhodomelaceae. Univ. Calif. Publ. Bot. 35:165–366. Lin, S.-M., Fredericq, S. & Hommersand, M. H. 2001. Systematics of the Delesseriaceae (Ceramiales, Rhodophyta) based on large subunit rDNA and rbcL sequences, including the Phycodryoideae, subfam. nov. J. Phycol. 37:881–99. Maggs, C. A., Douglas, S. E., Fenety, J. & Bird, C. J. 1992. A molecular and morphological analysis of the Gymnogongrus devoniensis (Rhodophyta) complex in the north Atlantic. J. Phycol. 28:214–32. Posada, D. & Crandall, K. A. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14:817–8. Rodriguez, F., Oliver, J. F., Marin, A. & Medina, J. R. 1990. The general stochastic model of nucleotide substitution. J. Theor. Biol. 142:485–501. Saunders, G. W. & Kraft, G. T. 1994. Small-subunit rRNA gene sequences from representatives of selected families of the Gigartinales and Rhodymeniales (Rhodophyta). 1. Evidence for the Plocamiales ord. nov. Can. J. Bot. 72:1250–63. Setchell, W. A. & Gardner, N. L. 1924. The marine algae. Expedition of the California Academy of Sciences to the Gulf of California in 1921. Proc. Calif. Acad. Sci. Ser. 4 12:695–949. Swofford, D. L. 2002. PAUP*: Phylogenetic Analysis Using Parsimony (* and many Other Methods). Version 4.0b10. Sinauer, Sunderland, MA. Womersley, H. B. S. 1978. Southern Australian species of Ceramium Roth (Rhodophyta). Aust. J. Mar. Freshw. Res. 29:205–57.