The biology of red algae in the genus Bostrychia (Rhodomelaceae) has been well studied over many years (reviewed in Zuccarello and West 2011). While the phylogeny and taxonomy have been clarified many questions still remain. One aspect is the highly diverse clade consisting mostly of algae identified as Bostrychia moritziana (Sonder ex Kutzing) J. Agardh or B. radicans (Montagne) Montagne. This clade, designated the B. moritziana-B. radicans species complex, contains one morphologically distinct species, B. pilulifera Montagne, plus a parasite of Bostrychia (Bostrychiocolax australis Zuccarello & West) (Zuccarello and West 2006). The other members of the clade consist of seven subclades of specimens resembling either B. moritziana or B. radicans, or subclades containing samples that can be identified as either (Zuccarello and West 1997, 2003, Zuccarello et al. 1999). It is clear that these subclades could be recognized as distinct species, based on genetic differentiation and reproductive incompatibility (spermatia attach to trichogynes but no carposporophytes are produced) (Zuccarello and West 2003, 2006). Data from crossing experiments suggest that even within these subclades reproductive incompatibility may be present (Zuccarello and West 1995, 2003, Zuccarello et al. 1999). Therefore based on the biological species concept (Mayr 1942), species exist even within these subclades. As yet no consistent morphological character has been found to define any of the lineages within the B. moritziana-B. radicans species complex, except for the 2 above mentioned species. In a continuing attempt to clarify the taxonomy within this species complex we made detailed observations of some unusual isolates in culture.

The Rhodomelaceae, to which Bostrychia belongs, are polysiphonous, with a consistent and evolutionarily conserved mode of pericental cell division (Hommersand 1963). Pericental cells on Bostrychia are further transversely divided one or more times into tier cells (Hommersand 1963, King and Puttock 1989), in some species cortication is also evident (King and Puttock 1989). Sexual reproduction is by an alternation of isomorphic generations (Polysiphonia-type) with reproductive structures (tetrasporangia, spermatangia, carpogonia).

Programmed cell death (PCD), also known as apoptosis, while a well studied phenomenon in many organisms including plants (Van Doorn 2011) is poorly studied in red algae but it has been alluded to in the death of hair cells and trichoblasts, as well as formation of regular thallus perforations (Garbary et al. 2012). Cell repair, after artificial cell wounding, has also been studied in red algae (Kim et al. 1988, 1995, Kim and Fritz 1993). Kim et al. (1988) described three types of cell repair patterns (elongation type, fusion type, and non-fusion-type). A combination of both PCD and repair response has not previously been documented in red algae.

During our investigations of algae in the Bostrychia moritziana-B. radicans species complex we noticed unusual development in certain culture isolates maintained over years in culture and these isolates are in a unique molecular lineage. We describe these isolates as a new species within the B. moritziana-B. radicans species complex and document possible PCD in these strains.

Methods for collection, isolation and maintenance of cultures are presented in West and Zuccarello (1999) and West (2005). To promote slower growth low nutrient levels were used initially (2 mL modified-Provasoli-media [MPM] enrichment per liter of sterile seawater) and 18-22℃, low light levels, <5 μmol photons m^-2 s^-1 cool white fluorescent or LED lighting at 12 : 12 LD daily cycle. The seawater was adjusted to 30 practical salinity units (psu) with Milli-Q water (Millipore Corp., Billerica, MA, USA). Careful observations with a dissecting stereomicroscope were made and short (1-4 mm) apices checked for contaminating epiphytes were excised with micro-forceps and subcultured in 50 × 70 mm crystallizing dishes containing 30 psu seawater medium with 10 mL MPM enrichment per liter. Most of these unialgal cultures were maintained in 18-22℃, 5-20 μmol photons m^-2 s^-1 cool white fluorescent or LED lighting at 12 : 12 LD daily cycle.

Photography was done using bright field optics on a Zeiss GFL microscope (Carl Zeiss, Jena, Germany) with a Canon G3 camera (Canon, Tokyo, Japan) and Photoshop CS to capture images. Aniline Blue stain (0.02%) in 50% Karo syrup was used to stain tetrasporangial stichidia of thalli fixed in 5% formaldehyde-seawater. All other images were of living specimens.

The methods for DNA extraction and amplification of rbcL are identical to those described in Zuccarello and West (2006). Sequences were edited (new Genbank Accession numbers KC768865-KC768876), assembled and aligned using the Geneious 6.0 software package (Biomatters, http://www.geneious.com/). Alignment was straight forward as no gaps were found in the data set. Maximum-parsimony trees (MP) were constructed in PAUP* 4.0b10 (Swofford 2002), using the heuristic search option, 500 random sequence additions, tree-bisection-reconnection (TBR) branch swapping. Support for individual internal branches was determined by bootstrap analysis as implemented in PAUP*. For bootstrap analysis, 1,000 bootstrap data sets were generated from resampled data (10 random sequence additions, 1million rearrangements per replicate). Maximum-likelihood trees (ML) were constructed in RAxML v.7.2.6 (Stamatakis 2006). Tree likelihoods were estimated with site specific general-time-reversible (GTR) and site specific rate heterogeneity (G) model for each codon position. Bootstrap support values were calculated with 1,000 replicates. Bayesian inference analysis was performed with MrBayes v3.1.2 (Ronquist and Huelsenbeck 2003). Analyses consisted of two independent runs of one cold and three incrementally heated chains, and 3 × 10⁶ generations with sampling every 100 generations. Posterior probabilities were calculated using a Metropolis-coupled Markov chain Monte Carlo approach. The log files of the runs were checked with Tracer v1.4.1 (Rambaut and Drummond 2007) and a burning sample of 1,000 trees was removed before calculating the majority rule consensus trees in MrBayes. All bioinformatic analyses with MrBayes were carried out on

the freely available Bioportal (http://www.bioportal.uio.no).

The rbcL data set of 54 taxa consisted of 1,163 characters of which 430 were parsimony-informative. The MP / ML and Bayesian trees were congruent in all supported branches. The topology of the ML tree is shown in Fig. 1. The data clearly showed the seven lineages B. moritziana-B. radicans species complex, plus B. pilulifera, as presented previously (West et al. 2006, Zuccarello and West 2006), while support between lineages is poor. Lineage 2 is strongly supported and consists of three supported subclades. One of these lineages contains two isolates from the western Pacific (4613 and 4588), which displayed unusual morphology from their day of collection. This unusual morphology and the molecular data lead to us to propose a new species for this clade:

Uniseriate filaments entirely monosiphonous with apical cell divisions, branching not apically derived but formed by lateral bud cells from intercalary cells or by single intercalary cell deaths followed by lateral growth of an adjacent cell. Tetrasporangial stichidia terminal with 2-4 segments, each segment bearing 2-3 tetrasporangia with 2 cover cells. Females with subterminal procarp and 3 sets of pericental cells around the axial cell. Terminal cystocarps with a distinct ostiole. Males with terminal spermatangial stichidia. Cell death occurs at variable intervals along the filaments and results in: 1) ‘non-fusion division’ by adjacent cells on both sides rejoining the filament; 2) cell division of only one adjacent cell resulting in branching at that site; 3) filament fragmentation forming new thalli.

Type localities. Polysiphonous Bostrychia species and monosiphonous specimens with cells of similar appearance to those of Bostrychia were collected on mud and pneumatophores of the mangrove Sonneratia sp. associated with the palm Nypa fruticans (Thunberg) Wurmberg, on Feb 8, 2006, near Okat Harbor (05°21.406′ N, 162°57.966′ E), Kosrae, Micronesia. No reproduction was observed on the monosiphonous specimens in the field. This solate (4613) is designated as the holotype. Isolate 4588 is designated as the paratype and was obtained on Avicennia pneumatophores at Achang Bay Resort, Guam (13°15.371′ N, 144°41.063′ E), Feb 12, 2006. Type specimens are deposited in the National Herbarium of New South Wales (NSW), Royal Botanic Gardens, Mrs Macquaries Road, Sydney NSW 2000, Australia.

Holotype. Dried specimen from culture 4613. NSW 904958.

Paratype. Dried specimen from culture 4588. NSW 904960.

Holotype and paratype cultures. Isolates 4588 and 4613 are available at the Korean Marine Plants Collection, Chungnam National University, 220 Gung-dong, Youseong-gu, Daejeon 305-764, Korea.

RcbL sequence of holotype: Genbank accession numbers KC768866

The monosiphonous filaments are quite uniform with a cell diameter of 28-30 μm and cell length varying from 48-80 μm. Filament growth is by apical cell division (Fig. 2C). Each cell has a central nucleus, 8-10 μm in diameter, suspended in an axial cytoplasmic strand. Chloroplasts are purple to pink, discoid to oblong or polygonal (3-5 μm) and positioned in the peripheral cytoplasm and in the central axial strand. Lateral branching does not occur at the apex as seen in polysiphonous Bostrychia species but occurs from intercalary cells at various intervals along the filaments (Fig. 2A-C). Rhizoidal filaments are composed of thin elongate cells, these originate intercalarily and do not resemble other rhizoid types seen in this genus (e.g., cladohaptera) and occasionally show somewhat contorted growth (Fig. 2D & E).

Cell death and repair. Cell death is evident throughout the monosiphonous filaments. This phenomenon has been noticed over several years in these isolates and is a regular phenomenon. The regularity and numbers of cell death indicates that this is a due to a developmental process and is not due to injury. When a cell protoplast collapses a homogeneous green to brown colour appears within the dead cell (Figs 2G, H & 3A-G). A dark brown collar is seen at either end of the dead cell (Figs 2G & 3A-G). Within 1-2 days one or both adjacent cells elongate slightly (Fig. 3A), change from a vegetative cell to a meristematic cell with a darker red and denser protoplast that divides transversely and grows into the dead cell space (Fig. 3A-C). These new cells are similar to apical cells in the monosiphonous filaments. Cell death may also take place in a cell at filament branch nodes (Fig. 3B). The cell adjacent to the dead basal cell of the lateral branch elongates forming a new cell that grows through the dead cell (Figs 2H & 3B). The other adjacent cell in the main filament often does not change (Figs 2H & 3B).

In other cases, possibly when the position and distance between the adjacent cells change, the apical cell on one side will curve away from the axis of the filament (Fig. 3D) and form a new branch that can break away forming a new filament. When a filament breaks at a dead cell both cells adjacent to the dead cell become shoot tips of new filaments (Fig. 3E). Also if two or more cells in a sequence die the adjacent live cell divides, elongates laterally and forms a new shoot apex (Fig. 2G).

Tetrasporophytes. Terminal tetrasporangial stichidia are numerous (Fig. 4A). Development is initiated with the formation of a series of pericentral cells by 2-4 axial cells, adjacent to the apical cell (Fig. 4B & C). Often the apical cell resumes cell divisions and vegetative growth so that mature stichidia become intercalary (Fig. 4A). The pericentral cells divide transversely forming tier cells (Fig. 4B) like those of normal polysiphonous Bostrychia species. The lower cell serves as a stalk cell and the upper cell becomes a tetrasporangial initial. The stalk cell also divides

to form 1-2 cover cells that project out from the lower part of each mature tetrahedrally divided sporangium (68-77 μm diameter) (Fig. 4D & E). Discharged spores are about 30-36 μm diameter and the spores divide transversely once, the lower cell first forming a slightly narrow rhizoid and the upper cell secondarily forming a monosiphonous shoot (Fig. 4F). No pericentral cells were observed during vegetative growth of any sporelings.

Female gametophytes. Carpogonial branches develop on a subterminal cell (Fig. 5A) of the monosiphonous branches and the shoot elongates if no fertilization occurs and the procarps become intercalary (Fig. 5B & C). The female thalli did not persist in culture so consequently it was not possible to investigate the further details of procarp-carposporophyte development but it appears that the carpogonial branch base is partially enclosed by two sets of prominent sterile cells (Fig. 5C). Mature cystocarps of about 350 μm diameter were observed only twice with a well developed pericarp and terminal ostiole, producing about 40 carpospores (Fig. 5E & F).

Male gametophytes. Terminal spermatangial stichidia are about 50 μm wide and up to 400 μm long (Fig. 5D). These stichidia are quite similar to the spermatangial stichidia borne on monosiphonous laterals of some polysiphonous B. moritziana (King and Puttock 1989).

The monosiphonous thalli of isolate 4588 showed no spore reproduction in field or culture specimens. Propagation in culture resulted entirely from cell death and breakage along the filaments producing many short filaments. This isolate has slightly smaller cells than isolate 4613 (22-26 μm diameter and 46-53 μm long).

The consistent morphology of these unusual isolates and their phylogenetic position as a well-supported lineage within the Bostrychia moritziana-B. radicans species complex warrants that they be described as a unique species of Bostrychia. Finding morphological characters for the growing number of ‘molecular species’ (i.e., species recognition based solely on molecular distinctness) continually being identified in red algae is problematic for taxonomy and field identification. Part of the problem lies with the shortage of characters available in many algal genera. In Bostrychia, this problem is acutely seen. There are an extremely limited number of characters to identify species (King and Puttock 1989, Zuccarello and West 2006) leading to many polyphyletic species that cannot be described (B. simpliciuscula Harvey ex J. Agardh) or monophyletic but highly diverse species complexes (B. moritziana-B. radicans) (Zuccarello and West 2006, in this paper). Our ability to collect and culture many isolates combined with careful observations, has at least resolved a morphologically consistent evolutionary lineage in this case. Monosiphony has occasionally been observed in 27 other culture isolates (Table 1). In all cases monosiphony is short-lived and isolates revert to polysiphony, or switch back and forth. These occasionally monosiphonous isolates are found in several lineages of the B. moritziana-B. radicans species complex but mostly in lineage 2 with 42% andlineage 7 with 30%.

Isolates 4588 and 4613 were the only field samples we observed to be monosiphonous and have remained so in culture for 7 years. The molecular evidence and reproductive characters were essential in linking B. anomala to the B. moritziana-B. radicans species complex.

A most unusual feature of B. anomola is the regular pattern of cell death in this species. PCD has not been studied in red algae although it has been hypothesized in various red algal developmental events (e.g., trichoblasts abscission, thallus perforations) but no definitive molecular information was obtained (Garbary et al. 2012). As cell death in B. anomola often leads to the fragmentation of the filament it could be an effective means of vegetative propagation. Mangrove-associated Bostrychia species are often vegetative in the field, although in culture they can produce sexual or asexual structures (West and Zuccarello 1999). Asexual reproduction of lineages with the B. moritziana-B. radicans species complex is common in some regions and does not seem to effect the ability of these populations to maintain large population sizes (West et al. 1992, West and Zuccarello 1999). Cell death and subsequent filament propagation could give B. anomala a competitive advantage in certain environments.

When filaments do repair they mostly follow a pattern, recognized as the “non-fusion-type” as proposed by Kim et al. (1988). Cell division is initiated in the cells adjacent to the dead cell and the elongating cells abut within the cytoplasmic space of the dead cell. The cells newly in contact must be able to adhere as the filament is not weaker at this location. Both PCD and wound repair are phenomena in red algae that warrant increased scientific research.

Our data clearly indicates that an evolutionary lineage within the B. moritziana-B. radicans species complex has consistent morphological characters that warrant its description as an identifiable species, B. anomala. These isolates also have unusual cellular development that should be further explored.