Delineation of Two Sibling Red Algal Species, Gracilaria Gracilis and Gracilaria Dura (Gracilariales, Rhodophyta), Using Multiple DNA Markers: Resurrection of the Species G. Dura Previously Described in the Northern Atlantic 200 Years Ago¹

Species delineation is an old methodological problem that has recently received an increased level of attention in the context of the barcoding approach (Chase et al. 2005, DeSalle et al. 2005, Petit and Excoffier 2009). We address this question for the cosmopolitan Gracilaria Grev. genus of red seaweeds, in which many species are poorly delimited (Gurgel and Fredericq 2004). The taxonomy of this genus is particularly difficult due to the relatively small number of diagnostic characters available compared to the high number of species described and their wide distribution (Gurgel et al. 2004). In this genus, the precise taxonomic identification of cylindrical specimens is often problematic, or nearly impossible, when only small and nonreproductive individuals are available in the field. Short sequences of nuclear or organelle DNA were thus developed in the 1990s, such as the plastid-encoded intergenic RUBISCO spacer (Destombe and Douglas 1991) and the nuclear ITS spacer regions (Goff et al. 1994), which allowed cohesive taxonomic entities to be defined and to delineate the species in this genus. More recently, different genetic markers have been used for red algal taxonomy (Provan et al. 2004, Saunders 2005, Robba et al. 2006). For a more accurate identification of sibling species, and because detection of hybridization/introgression cannot be reliably accomplished by examination of a single DNA region, Chase et al. (2005) suggested the use of a multilocus approach including mitochondrial, plastid, and nuclear ribosomal DNA markers.

In this paper, we investigate one of these taxonomic problems using two species: G. gracilis and G. dura. Both species are very similar morphologically and occur in the lower intertidal and upper subtidal. G. gracilis was originally described as Fucus verrucosus by Hudson (1762) based on material collected in the United Kingdom. This species was for a long time confused with Gracilariopsis longissima under the name Gracilaria verrucosa (Huds.) Papenf. (Steentoft et al. 1995). The wide amphi-Atlantic distribution of G. gracilis (Bird and Kain 1995) was confirmed by the use of cytoplasmic and nuclear DNA markers (Destombe and Douglas 1991, Wattier et al. 1997). This species is commonly found on northern Atlantic coasts from Morocco to Norway (Guillemin et al. 2008) and on southern Atlantic coasts of Namibia (Wattier et al. 1997, Iyer et al. 2005) and Argentina (Destombe and Douglas 1991). G. dura was described for the first time in 1822 by C. Agardh from samples collected on the Atlantic coast of southern Spain (Cadiz) by Cabrera (C. Agardh 1822); later this species was also reported on the northern coast of Brittany by Crouan and Crouan (1830) (cited in Chalon 1905 and Piquenard 1912) and also on the Atlantic coast of Morocco (Raphaélis 1929, Dangeard 1949). The plants collected in the Mediterranean Sea by J. Agardh (1842) in the Gulf of Naples were referred to as G. dura var. lyra.

In Morocco, G. dura and G. gracilis can easily be distinguished based on their gross morphology. G. dura has a bright red color and small lateral branches often located on the same side of the main axis (see Gayral 1958, p. 359), whereas G. gracilis usually has dark reddish-brown, irregularly branching thalli (Steentoft et al. 1995). In northern Europe, distinguishing between these two terete species is much more difficult. Thus, they are probably regularly misidentified and generally recorded under the name of G. gracilis. Recently, a barcoding approach based on a sequence of the RUBISCO spacer indicated that these two morphological species, occurring together in the same site in Morocco, corresponded to two sister species (Guillemin et al. 2008). In this context, the observed weak morphological difference between G. dura and G. gracilis in the northern Atlantic compared to the coast of Morocco raises the question as to whether this difference is due to environmental factors or to hybridization events.

By using a combination of markers from the three different genome compartments—mitochondria, plastid, and nuclear—the aim of this study was to delineate more precisely these two morphological species and to detect the occurrence of putative hybridization.

Materials and methods

Sampling and DNA extraction. Forty-two samples from 22 locations were selected to be representative of the G. gracilis and G. dura species distribution (Fig. 1 and Table S1 in the supplementary material). Samples were stored in silica gel for later molecular analysis. DNA was extracted from the algae using a modified Chelex method (Cohen et al. 2004). Individuals from Morocco previously identified as G. gracilis (MGR3) and G. dura (MGR177a) and deposited as dried samples preserved in silica gel in the collection at Roscoff-UMR7144 were used as references (Guillemin et al. 2008).

Results

Alignment of the rbcL revealed 17 polymorphic sites over a total length of 1,309 bp (1.29%) and gave rise to eight chlorotypes (C1–C8; GenBank accession numbers GQ229494–GQ229500, Table 1, Fig. 2). The cox2-cox3 spacer sequences yielded an alignment of 308 bp of which 11 sites were polymorphic (3.6%) (Table 1). Ten mutations and one insertion–deletion mutation were observed, resulting in a total of six mitotypes designated M1 to M6 (Table 1, Fig. 2, GenBank accession numbers GQ229501–GQ229506). The ITS2 was 336 bp long and showed 27 polymorphic sites (18 mutations and nine insertion–deletion mutations, Table 1), resulting in a total of 11 ribotypes (designated R1 to R11; GenBank accession numbers GQ229507–GQ229517, Fig. 2).

Discussion

The standard DNA-barcoding approach based on a single gene (such as cytochrome oxydase subunit 1 gene) sequence divergence as a tool for species delimitation has been strongly criticized because it will not necessarily provide sufficient resolution needed to recognize the large range of species targeted by the barcoding initiative (DeSalle et al. 2005). As proposed by Chase et al. (2005), in our study, we used multilocus barcodes including three independent markers (plastid, mitochondria, and nuclear) to explore species limits in the G. gracilis species complex. This method not only improves the detection of species boundaries, but it also allows the incidence of hybridization and introgression to be tested, a phenomenon that occurs at a high frequency in plants and probably also in algae (Chase et al. 2005). Whatever the genetic marker, the intergroup divergence between G. gracilis and G. dura is five to 10 times higher than the intragroup divergence. These results are congruent with the boundaries used as guidelines for decisions concerning species validity (i.e., intergroup should be 10 times greater than intragroup differences: Hebert et al. 2004) and with previous taxonomic studies of the Gracilariaceae (Bellorin et al. 2002, Cohen et al. 2004, Yang et al. 2008). Indeed, the intergroup sequence pair-wise divergences observed in the present study for the ITS2, the rbcL gene, and the cox2-cox3 spacer fit with the values obtained previously between closest species of Gracilariaceae. The congruent divergences highly suggest that these two taxa correspond to two different genetic entities that have evolved separately and, therefore, correspond two distinct species according to the unified species concept developed by de Queiroz (2007), for which species are separately evolving metapopulation lineages. The fact that genetic diversity is higher between two individuals of G. gracilis and G. dura occurring together at the same geographic site than between two individuals of G. gracilis from Norway and South Africa or of G. dura from Morocco and Brittany indicates true discontinuities between groups (i.e., lineage separation) rather than genetic differentiation due to isolation by distance.

G. dura was first described, based on its morphological characteristics, from the northern Atlantic, in Morocco (Raphaélis 1929, Dangeard 1949), in southern Spain (Agardh 1822), and in Brittany (Chalon 1905, Piquenard 1912). However, because of the lack of clear-cut morphological character differences between G. dura and G. gracilis, the species G. dura disappeared totally from the inventory of the northern Atlantic flora after 1950 (Feldmann 1954, Steentoft and Farnham 1997). This species has also been reported from the Mediterranean Sea, in the Gulf of Naples by Agardh (1842) as G. dura var. lyra. However, the comparison of the rbcL sequences for G. dura from the Atlantic (GenBank numbers GQ229498 and GQ229499, this study) and from the Mediterranean (GenBank number AY651058, Gargiulo et al. 2006) clearly demonstrates that these two species are highly divergent and correspond to two different cryptic species (K2P pair-wise divergence >10%). Recently, Gargiulo et al. (2006) suggested a reexamination of these Mediterranean samples of G. “dura” from southern Italy that grouped, in all molecular analyses, with the clade containing Hydropuntia species.

The multilocus approach used in this study suggests that the two samples collected in Brittany and Wales and previously assigned to G. gracilis according to their rbcL gene plastid sequences by Gurgel and Fredericq (2004), while showing a genetic divergence of 0.95% (i.e., AY049399 and AY049400), correspond respectively to G. gracilis and G. dura. This finding suggests that the distribution range of G. dura (from Morocco to Wales) is much more restricted than that of G. gracilis (Wattier et al. 1997, Iyer et al. 2005, Gargiulo et al. 2006).

In the present study, the observed linkage disequilibrium between chloroplast and mitochondrion confirms uniparental inheritance of organelles. Organelle inheritances were studied only in a few Rhodophyta species (Zuccarello et al. 1999a,b, 2002, Choi et al. 2008). Organelle genomes were shown to be preferentially transmitted by the mother. However, in Porphyra yezoensis, low frequencies of paternal leakage and biparental inheritance were also reported (Choi et al. 2008).

Our data report high but incomplete linkage disequilibrium since “incongruent” cytotypes (i.e., a combination of G. dura mitotypes with G. gracilis chlorotypes) were observed in three individuals only and in sites where both species are co-occuring. The sample size does not allow us to definitively conclude between the two explanations for such incongruent associations: either shared ancestral polymorphism or interspecific hybridization. However, since incongruence was exclusively observed in sympatric populations, we speculate that hybridization is the most likely explanation. While the ITS regions have proved particularly helpful for inferring relationship at lower taxonomic levels, cases of recombination and gene conversion in these ribosomal repeat regions were reported following hybridization in higher plants and fungi (Hugall et al. 1999, Hughes and Petersen 2001). In our study, the clear relationship between cytoplasmic and ribosomal markers in G. gracilis and in G. dura suggests either that hybridization between the two species is uncommon or that ribosomal gene conversion occurred differently according to the parental species. The three putative hybrids (i.e., individuals with incongruent cytotypes) observed in our study advocate that hybrids between G. dura and G. gracilis may have been formed at some time, the result of either a mitochondrion or a chloroplast introgression.

Contrary to Phaeophyceae (such as Fucales: Coyer et al. 2007; and Laminariales: Druehl et al. 2005), only a few reports of hybridization have been published for red seaweeds. Using both nuclear and plastid DNA, Niwa et al. (2009) demonstrated the occurrence of plastid introgression from P. yezoensis to P. tenera. A possible case of differential organelle inheritance was also reported in Mastocarpus stellatus lineages by Zuccarello et al. (2005) in which some plants in natural populations exhibited a mixed cytoplasm composed of a mitochondrial haplotype of one breeding group with the plastid haplotype of the other breeding group.

Our genetic analysis of divergence between two sibling species based on a multilocus approach confirms the existence of two sibling species under the name of G. gracilis in the northern Atlantic. It illustrates clearly that sequences from different DNA compartments are needed to correctly diagnose species and retrace their phylogenetic history.