BANGIA FUSCOPURPUREA ( DILLWYN ) LYNGBYE ( BANGIALES , RHODOPHYTA ) IN TAIWAN

Samples of the freshwater red algae, Bangiadulcis atropurpurea, collected from the mountain waterfalls and its close species, Bangia fuscopurpurea, collected from coasts, were phylogenetically analyzed. The sequences of the rbcL gene and RuBisCO spacer region of the freshwater Bangiadulcis atropurpurea species were identical or similar to those of B. atropurpurea from Japan, North America and Europe. This result indicated that the freshwater Bangiadulcis species from Taiwan shared a common ancestor with the three above-mentioned populations and formed a distinct clade from the marine Bangia species in the phylogenetic trees. It is suggested that all the previous records on marine Bangia species should be revised and the name B. fuscopurpurea be used for the marine species in Taiwan. In this study, the freshwater alga B. atropurpurea presents a new record in the algal flora of Taiwan. This finding is important for the protection of the biodiversity of red algal flora, and provides useful information on the ecological conservation of the species in Taiwan.


INTRODUCTION
The red-algae genus Bangia Lyngbye is widespread along the coast of Taiwan.In the marine environment, Bangia generally occupies a habitat in the high intertidal zone (Sheath, 1984).Bangiadulcis atropurpurea (A.Roth) W. A. Nelson is reported from freshwater, and it was originally described by Roth (1806) as Conferva atropur-purea, which was later included in the genus Bangia by C. Agardh (1824).Geesink (1973) concluded that the freshwater B. atropurpurea and the marine B. fuscopurpurea (Dillwyn) Lyngbye are conspecific.On the basis of karyology and DNA sequence analysis, freshwater Bangia sp. are placed on a separate and well-supported branch, and the name B. atropurpurea has been proposed to represent this lineage (Müller et al., 1998(Müller et al., , 2003)).Later, Nelson (2007) indicated that the freshwater species could be distinguished from other members of the Bangiales based on cytological, morphological, ecological, and phylogenetic characters.Thus, Nelson used Bangiadulcis to describe the freshwater species.
In Taiwan, B. atropurpurea has been recorded in many studies (Okamura, 1936;Shen and Fan, 1950;Tanaka, 1950Tanaka, , 1952;;Fan, 1953;Tokida, 1954;Rho, 1958;Chiang, 1962Chiang, , 1973;;Taniguti, 1976;Sheath and Cole, 1984).All the specimens were collected from marine environments, and we believe that the name used in these studies was incorrect.Molecular tools have recently proved useful in various studies of algae, facilitating the elucidation of relationships among taxa, especially among red algae (Coleman and Goff, 1991;Müller et al., 1998Müller et al., , 2003;;Wang et al., 2005).However, there has not been any molecular method-based taxonomic or phylogenetic study of Bangia species from Taiwan.The present study was undertaken to re-examine the Bangia species found in Taiwan.We collected marine Bangia species from the coasts of northern Taiwan and Kinmen County (the county lies in the sea roughly 190 km west of the Taiwan) and two samples of freshwater Bangia species from the mountain waterfalls of northern and central Taiwan.In addition to confirming their phylogenetic relationship based on the rbcL gene sequence, the RuBisCO spacer region and morphological characters are also described in detail.

Morphological observations
Samples of marine Bangia species were collected from three sites in northern Taiwan and Kinmen County.Two samples of freshwater Bangiadulcis sp. were collected, one from the Youling Waterfall,  (Swofford, 2003).Each character in the DNA sequence analysis was analyzed with equal weight.The distance-matrix values were estimat-ed according to the Kimura two-parameter model (Kimura, 1980).The robustness of tree topology was estimated by 100 replicates of bootstrap resampling, using the maximum parsimony (Mp) method.The A+T content (%) was calculated using the GeneDoc software (Nicholas et al., 1997).

Morphological observations
Bangiadulcis atropurpurea (Roth) W. A.  , B). Mature filaments become polysiphonous by longitudinal cell divisions (Figs.1A,  C).The plants are anchored to the substratum by a rhizoidal part, which grows downward from the lower cells through the common outer wall (Fig. 1D).Asexual reproduction occurs by monospores (Fig. 1E).An unknown bud-like structure was observed on a filament (Fig. 1F).Holotype − Roth's collection was destroyed in World War II (Koster, 1957).Chialing River near pehpei, Szechwan, China; Amahata-gawa, Yamanashi, Japan; Great Lakes except Lake Superior, Canada; North America; and Taiwan.

Specimens
Habitat − plants are epilithic, growing on stones submerged in a well-aerated, moderately agitated, and clear pool under a waterfall, at a mean depth of 2-5 cm.In Taiwan, they are found Descriptions − plants are filamentous, with unbranched dark-purple filaments.Young filaments initially comprise a single row of cylindrical cells (Fig. 2A).Mature filaments become polysiphonous by longitudinal cell divisions (Fig. 2B).Asexual reproduction is not observed.World distribution − Ireland, Europe; western Africa; eastern North America; eastern South America; Caribbean Basin; Mediterranean Basin; eastern Asia; western North and central America; Southeast Asia; western South America; and Atlantic Islands.
Habitat − plants are epilithic, growing on rocks in the high intertidal zone.
Remarks − Bangia fuscopurpurea is dioecious.Spermatangia develop in packets by division of an undifferentiated cell, and carpogonia produce a small protrusion for the attachment of spermatia.Asexual reproduction occurs by vegetatively produced monospores.This species occurs only in marine habitats.

The rbcL gene
A set of 23 rbcL sequences of selected species (Table 1), including 1 020 aligned sites with 289 potentially parsimony-informative characters, was analyzed.Five most parsimonious trees were obtained in the Mp analysis based on a heuristic search performed using the stepwise addition of 100 random replications.One of these trees is shown in Fig. 1, wherein we have shown the branches supported by ≥50% bootstrap values in the Mp analysis.The tree length was 524 steps, with a consistency index (CI) of 0.7023 and retention index (RI) of 0.8497.We encountered difficulty in amplifying the rbcL gene for the collection from pinglang Bridge, Keelung City, Taiwan, and therefore the related information was not included in the analysis.The freshwater Bangiadulcis species clade was well supported by bootstrap analysis (100%), and as shown in Fig. 3, it represents a distinct clade from the marine Bangia species.

RuBisCO spacer region
The RuBisCO spacer region in the specimens of both the marine and freshwater Bangiadulcis species was 77-bp long and A/T rich, with a nucleotide composition ranging from 70.13% to 77.92% A+T.Both the freshwater specimens from Taiwan (TY* and NT*) were found to be identical.The freshwater specimens from Ireland (IR*) differed from those collected from Taiwan by 1 bp; this difference was due to an A↔C transversion.Furthermore, the freshwater specimens from America and Europe differed from those collected from Taiwan by 1 bp; this difference was due to an A↔G transition.Both the marine specimens from Kinmen County (KM1 and KM2) were identical to that from Canada (NF) and differed by 3 bases from the marine specimen from Taiwan (KL; 8.29% sequence divergence), which was identical to the specimens from North Carolina (NC) and Massachusetts (MA) (Table 2).
We have not shown the trees inferred from the parsimony analysis of the RuBisCO spacer region because of the lack of parsimony-informative sites.parsimony analysis of the RuBisCO spacer region resulted in 12 most parsimonious trees (data not shown).These trees were based on 7 potentially parsimony-informative characters within the 77-bp region of the spacer; the tree length was 27 steps, with CI = 0.8889 and RI = 0.8000.There was considerable homoplasy in 6 characters, because of which we could not clarify the relationship among these specimens.

DISCUSSION
The genus Bangia has a long evolutionary history (1.2 × 10 9 y; Butterfield et al., 1990;Butterfield, 2000) and a divergence time of 174 × 10 6 to 265 × 10 6 y between freshwater and marine species (Müller et al., 1998).Based on the rbcL gene sequence analysis in this study, we found considerable sequence divergence between the marine and freshwater species.
In this study, the values of A-T richness and length of the RuBisCO spacer region of the examined species were within the range reported for Rhodophyta (Destombe and Douglas, 1991;Kamiya et al., 1998).Due to homoplasy and the small number of phylogenetically informative characters, we could not clarify the relationship among the various Bangia species.However, we observed some basic trends that are consistent with those indicated by the rbcL gene analysis.For instance, the 2 freshwater specimens collected from northern and central Taiwan (TY* and NT*, respectively; Fig. 3) were identical with regard to the spacer region, as were the marine specimens collected from Alaska, Northwest Territories, Greenland, New Hampshire and Rhode Island (AK, NWT, GLD, NH and RI, respectively).Furthermore, all the freshwater specimens of Bangia species were nearly identical, as revealed by the analyses of two genes in a previous study (Müller et al., 2003).
The finding of monophyly of the freshwater Bangiadulcis species based on the rbcL gene analysis in this study was consistent with that based on independent analyses of the rbcL and 18S RNA genes and a combined analysis of the two genes in a previous study (Müller et al., 2003, Nelson, 2007).From this study and those of Müller et al. (1998Müller et al. ( , 2003)), it is clear that all the freshwater species formed a separate lineage.With regard to the marine Bangia species, several distinct entities were revealed by the 18S RNA gene analysis performed in a previous study (Müller et al., 2003), but not by the rbcL gene analysis conducted in this study.However, all marine Bangia species should be recognized as B. fuscopurpurea until the complex relationship among these species can be resolved (Sheath and Müller, personal communication).Besides confirming the taxonomic relationship among the first-recorded freshwater Bangiadulcis species, the present study indicates that the marine Bangia species are distinct from the freshwater Bangiadulcis species found in Taiwan.It is suggested that all the previous records of marine Bangia species should be revised and the name B. fuscopurpurea be used for the marine species in Taiwan.
The role of taxonomy should never be underestimated.We need more taxonomic data and convenient methods to tackle algal identification.
Here we use the rbcL gene and RuBisCO spacer region to distinguish Bangiadulcis atropurpurea from its close species.
Bangiadulcis atropurpurea (as B. atropurpurea) has been indicated as an endangered species in Serbia (Simić et al., 2007), Slovakia (Marhold and Hindak, 1998) and Germany (Kusber et al., 2005).It has even been declared Table 2. pairwise comparison of distance matrices and variation in the aligned sequences of the RuBisCO spacer region (77 bp) among Bangiadulcis and Bangia populations.The lower matrix shows the number of base-pair changes between entities.The upper matrix shows sequence divergence between entity pairs, calculated using the Kimura two-parameter model.(Siemińska, 1992) and only distributed in one locality in Finland (Eloranta and Kwandrans, 2004).We concluded that taxonomic knowledge is vital to the conservation of the algae, and DNA barcoding provides at least a potential approach.

Fig. 1 .
Fig. 1.Bangiadulcis atropurpurea (A.Roth) W. A. Nelson.A. Unbranched filaments.Young filaments initially comprising a single row of cylindrical cells (arrowheads).Mature filaments become polysiphonous by longitudinal cell divisions (arrows).B. Close-up of young filaments, showing a single row of cylindrical cells.C. Close-up of mature filaments, showing longitudinal cell divisions.D. Rhizoidal part of a plant.E. Asexual reproduction by monospore(s) (arrow).F. Unknown bud-like structure on a filament.

Fig. 3 .
Fig. 3.A set of 23 rbcL sequences of selected species, including 1020 aligned sites with 289 potentially parsimony-informative characters, was analyzed.This figure shows 1 of the 5 most parsimonious trees generated by performing a heuristic search with the stepwise addition of 100 random replications.Bootstrap values are shown above the branches that were supported by ≥50% bootstrap values in the maximum parsimony (Mp) analysis.The tree length was 524 steps, with the consistency index (CI) of 0.7023 and the retention index (RI) of 0.8497.

Table 1 .
Location (abbreviation), collection information, and GenBank accession number of Bangia species used in this study.