Cryptic diversity, geographical endemism and allopolyploidy in NE Pacific seaweeds

Abstract

Background

Molecular markers are revealing a much more diverse and evolutionarily complex picture of marine biodiversity than previously anticipated. Cryptic and/or endemic marine species are continually being found throughout the world oceans, predominantly in inconspicuous tropical groups but also in larger, canopy-forming taxa from well studied temperate regions. Interspecific hybridization has also been found to be prevalent in many marine groups, for instance within dense congeneric assemblages, with introgressive gene-flow being the most common outcome. Here, using a congeneric phylogeographic approach, we investigated two monotypic and geographically complementary sister genera of north-east Pacific intertidal seaweeds (Hesperophycus and Pelvetiopsis), for which preliminary molecular tests revealed unexpected conflicts consistent with unrecognized cryptic diversity and hybridization.

Results

The three recovered mtDNA clades did not match a priori species delimitations. H. californicus was congruent, whereas widespread P. limitata encompassed two additional narrow-endemic species from California - P. arborescens (here genetically confirmed) and P. hybrida sp. nov. The congruence between the genotypic clusters and the mtDNA clades was absolute. Fixed heterozygosity was apparent in a high proportion of loci in P. limitata and P. hybrida, with genetic analyses showing that the latter was composed of both H. californicus and P. arborescens genomes. All four inferred species could be distinguished based on their general morphology.

Conclusions

This study confirmed additional diversity and reticulation within NE Pacific Hesperophycus/Pelvetiopsis, including the validity of the much endangered, modern climatic relict P. arborescens, and the identification of a new, stable allopolyploid species (P. hybrida) with clearly discernable ancestry (♀ H. californicus x ♂ P. arborescens), morphology, and geographical distribution. Allopolyploid speciation is otherwise completely unknown in brown seaweeds, and its unique occurrence within this genus (P. limitata possibly representing a second example) remains enigmatic. The taxonomic separation of Hesperophycus and Pelvetiopsis is not supported and the genera should be synonymized; we retain only the latter. The transitional coastline between Point Conception and Monterey Bay represented a diversity hotspot for the genus and the likely sites of extraordinary evolutionary events of allopolyploid speciation at sympatric range contact zones. This study pinpoints how much diversity (and evolutionary processes) potentially remains undiscovered even on a conspicuous seaweed genus from the well-studied Californian intertidal shores let alone in other, less studied marine groups and regions/depths.

Background

Species diversity, endemism and structure are predicted to be lower in marine vs terrestrial assemblages because populations are less prone to diverge when dispersal potential is high and prominent barriers are absent [1–3]. Molecular markers, with their unprecedented power to delineate and describe biological relationships, show nevertheless a rather intricate and multi-layered picture of marine biodiversity. Many poor dispersers and/or habitat specialists often display extreme levels of cryptic phylogeographic structure across their wide distributions (e.g. [4, 5]), including divergent regional lineages of uncertain biological status (e.g. [6, 7]). True cryptic (and pseudo-cryptic, when morphologically distinguishable a posteriori) species are also commonly identified within morphologically conserved, hyper-diverse and/or poorly-studied taxonomic groups [8–10], and even within large, ecologically dominant organisms such as canopy-forming kelp [11, 12]. In the NE Atlantic, for instance, two new fucoids have been genetically identified (within existing taxa) in the past 10 years alone [13, 14].

Remarkably, the number of endemic (geographically restricted) cryptic species, often recovered within allegedly wide-ranging congeners, is also accumulating, particularly throughout tropical/subtropical archipelagos (e.g. [15–17]). Endemism may be a stable feature in the evolutionary history of some species, while for others just a transient state. For instance, nascent species are often spatially restricted, at least for some time [18]. Species may also endure climatically adverse periods in rather restricted “refugia”, before expanding their ranges again or becoming extinct [19–21]. In the terrestrial realm, many modern (interglacial) “climatic relicts” are concentrated in atypically cooler (or otherwise very heterogeneous) areas, such as southern mountain ranges [22]. Similar patterns may emerge, at least at the population level, in marine systems, for instance associated with cooler upwelling areas [23].

Modern phylogenetic analyses have also clarified marine species relationships and boundaries, often challenging traditional taxonomic classifications (e.g. [24–26]). Phylogenetic discordance and/or genomic admixture reveal that hybridization and gene introgression occur in a diverse range of marine taxa (reviewed in [27]). Perhaps unsurprisingly, hybridization has been found to be particularly frequent among sessile broadcast spawners (displaying external fertilization after release of the gametes in the water column) that form dense aggregations of closely related species, such as some corals [28] and canopy-forming seaweeds [29–31]. A range of sexual and clonal corals is suspected to be of actual hybrid origin [32, 33], but marine hybrid speciation, including via allopolyploidy (i.e., associated with chromosome multiplication), is otherwise very poorly documented. Evidence for ancient polyploidization events and for modern (allo)polyploid taxa has been accumulating across much of the tree of life [34–38], and its incidence and evolutionary significance may be more general than previously appreciated.

Hesperophycus californicus and Pelvetiopsis limitata are the extant members of two monotypic genera of canopy-forming fucoid seaweeds (Fucaceae, Heterokontophyta) endemic to the temperate rocky-shores of the north-east Pacific. Both occur in mid/high intertidal assemblages, although with complementary biogeographical distributions. Warm-temperate H. californicus is present from Punta Eugenia (Mexico) to Monterey Bay (California, USA), whereas cold-temperate P. limitata occurs from Point Conception (California) to Vancouver Island (British Columbia, CAN). A third putative species, P. arborescens, was described from just south of Monterey [39], but has never enjoyed much practical recognition and use. Collectively, these species/genera form a monophyletic lineage that is sister to Fucus [40], a much more speciose genus (8+ species) with a centre of diversity in the north Atlantic. The genus Fucus has been the focus of much physiological, ecological, and phylogeographical research [41–43], but reconstructing its evolutionary history has long been complicated by poor morphological discrimination, low marker resolution, incomplete lineage sorting, and hybridization/introgression [14, 24, 31, 44, 45]. Hesperophycus-Pelvetiopsis have received considerably less attention, but preliminary tests for newly developed molecular markers immediately revealed multiple taxonomic and phylogenetic conflicts consistent with cryptic diversity and reticulate evolution.

This study aimed to identify all genetic entities present within this complex, and to describe their phylogenetic relationships, morphological differences and biogeographical ranges. We reveal a single genus (we retain Pelvetiopsis) with four species, adding to this complex two narrow-endemic, pseudo-cryptic species from California (P. arborescens and P. hybrida sp.nov.), the latter of recent allopolyploid origin, a process so far unknown in brown seaweeds.

Methods

Sampling, DNA isolation and amplification, genotyping and sequencing

A comprehensive panel of populations of Pelvetiopsis limitata sensu lato (i.e., including putative P. arborescens) and Hesperophycus californicus was sampled along the entire ranges of these species in the NE Pacific. Preliminary results showed that P. arborescens and a fourth entity were genetically distinct, so additional populations (mostly from the Californian counties of Monterey and San Luis Obispo) were sampled to include all the morphological and genetic diversity present in the complex. Whole individuals were also sampled for descriptions of the general morphology of each recovered genetic entity. These were collected at the sites where they had been genetically identified and morphometric measurements, photographs and herbarium pressings were made. In total, 36 population samples containing 16 individuals (ca. 576 individuals) from 30 sites spanning Punta Baja (Baja California, Mexico) to Cape Meares (N Oregon, USA) were analysed (Additional file 1). Samples of apical tissue were collected every meter along 30 m, or haphazardly along 25–100 m, depending on the local spatial patchiness at each site, and were preserved dehydrated in silica-gel crystals until DNA extraction. Genomic DNA was extracted from ca. 8 mg dehydrated tissue using the Nucleospin® 96 Plant II kit (Macherey-Nagel Duren, Germany), and diluted 1:100.

All individuals were screened for the 23S/trnK mitochondrial intergenic spacer (mtIGS), as this marker was previously shown to provide good inter- and intra-specific resolution in the sister genus Fucus, and some propensity to cross species barriers [29, 31, 44]. The ‘universal’ fucoid/kelp primers of [46] were used to amplify a large 1.5 kb fragment spanning the mtIGS in a few individuals. New primers were designed in conserved genic flanking regions from these sequences to specifically target the mtIGS of Hesperophycus-Pelvetiopsis (see Additional file 2 for primer sequences and amplification details).

A geographically diverse panel of individuals representing all genetic entities inferred with the mtIGS was used to assess cross-amplification and polymorphism of 40 microsatellite loci developed de novo for P. limitata. These were identified in silico by screening with MsatCommander [47] a genomic DNA library (Biocant, Cantanhede, Portugal; http://www.biocant.pt/) using shotgun 454 pyrosequencing. From these, 13 loci were selected to produce multi-locus genotypes for all Hesperophycus-Pelvetiopsis individuals (see Additional file 2 for primer sequences and amplification conditions).

Amplified fragments were analyzed in an ABI PRISM 3130xl automated capillary sequencer (Applied Biosystems) at CCMAR, Portugal. Amplicons were pre-cleaned with ExoSap (Fermentas, Waltham, MA, USA) and sequences were aligned, proofread, and edited in GENEIOUS 4.8 (Biomatters; http://www.geneious.com). Microsatellite alleles were manually scored in STRand [48] using the GeneScan™ 500 LIZ™ size standard (Applied Biosystems).

Data analysis

Haplotype/sequence networks were constructed in Network 4 (www.fluxus-engineering.com) using the Median-Joining (MJ) algorithm [49]. Phylogenetic relationships were reconstructed using Bayesian and Maximum Likelihood (ML) inference methods. Nucleotide substitution models (using three substitution schemes) were compared with jModelTest 2 [50] and the best-fit model selected based on the Akaike information criterion. Bayesian analyses were performed using MrBayes 3 [51]. Two parallel Metropolis-coupled Markov chain Monte Carlo searches, each with four chains (3 ‘heated’), were run for 10⁶ generations, sampling trees and parameters every 10² generations. The number of substitution rates, among-site rate variation, and base frequency priors were set according to the substitution model selected, leaving the remaining options as default. Average standard deviation of split frequencies between runs and cold chains Log-likelihood stationarity were inspected to assess inter-run convergence and run length sufficiency, respectively. Based on the latter, 10⁵ generations (1000 trees) were discarded as burn-in. The remaining 18 000 trees sampled were used to produce 50% majority-rule consensus trees and to calculate branch posterior probabilities. Maximum likelihood analyses were performed with PhyML 3 [52] using the ATGC bioinformatics platform (http://www.atgc-montpellier.fr/phyml/). Nodal support was calculated using 1000 bootstraps. Trees were rooted with Fucus spp. Inter-specific divergence within Pelvetiopsis and within the two major lineages of Fucus was compared by estimating Kimura’s two parameter (K2P) distances between ancestral haplotypes defining each recovered species.

Summary statistics of the microsatellite genetic diversity, including microsatellite allele frequencies, mean allelic richness (A), Nei’s gene diversity (H _E), observed heterozygosity (H _O), and inbreeding coefficients (F _IS) were calculated with Genetix 4 [53] for all populations. The number and delimitation of multilocus genotypic clusters and its congruence with mtIGS phylogroups was visually inspected with a factorial correspondence analysis (FCA) implemented in Genetix, and with a Bayesian, model-based genetic admixture analysis implemented in STRUCTURE 2 [54, 55]. In the latter, individuals were pooled into one dataset for analyses, without a priori population assignments. Each number of assumed genetic clusters (K, set sequentially from 2 to 4) was run five times using a burn-in of 200,000 iterations and a run-length of 1,000,000 iterations.

Results

A total of 19 mtIGS haplotypes (plus outgroups) were recovered. The mtIGS recovered three main clades that partially conflicted with a priori species delimitations. H. californicus was congruent, whereas three well resolved phylogroups were apparent within P. limitata sensu lato (Fig. 1). These included (see results below and Discussion): 1) P. limitata sensu stricto (the most widespread, hereafter simply referred to as P. limitata), 2) P. arborescens, and 3) a new, undescribed cryptic species, P. hybrida sp. nov., genealogically closer to H. californicus. MtDNA clades were well resolved (i.e. had high statistical support) but resolution was insufficient to establish the sequence of cladogenesis (lineage splitting, Fig. 1b). Mean sequence distances between Hesperophycus-Pelvetiopsis clades ranged between 0.057 (P. arborescens vs H. californicus) and 0.122 (H. californicus vs P. limitata), the same magnitude of that found between the two major lineages of Fucus (0.086, Table 1). Haplotype diversity within inferred entities was rather low (2–6 haplotypes, Fig. 1a), with 23 populations (~64%) fixed for a single haplotype (data not shown).

Fig. 1

Genealogical relationships of Pelvetiopsis spp. based on mtIGS data. a MtIGS haplotype network; the dots represent inferred, unsampled haplotypes and the small rectangles represent inferred indels. b Reconstructed 50% majority-rule consensus phylogenetic tree. Numbers above and below the branches are Bayesian posterior probabilities (>0.90) and maximum likelihood bootstrap support values (>70), respectively. For clarity, outgroup (Fucus) branches were collapsed (tip triangles), with the horizontal length of the triangle representing the distance from the branches’ common node to the tip of the longest branch

Table 1 MtIGS divergence in Fucus and Pelvetiopsis

Patterns of microsatellite cross-amplification and polymorphism varied markedly between loci, with four (Pl310, Pl51, Pl54, Pl311) failing to produce clear products in at least one species, and four (Pl32, Pl36, Pl51, Pl53) revealing inter-specific differences only (Additional file 3). Populations of H. californicus and P. arborescens typically exhibited homozygote excess (Additional file 1). In contrast, fixed heterozygosity was apparent in a high proportion of loci that successfully amplified in P. limitata (6/13) and P. hybrida (8/10) (Additional file 3), resulting in extremely high (and significant) heterozygote excess (-1.0 < F _IS < -0.6) in all their populations (Additional file 1). The occurrence of 3 alleles at a single locus was very rarely observed. In P. hybrida, each allele in heterozygote loci was identical or similar in size to either P. arborescens or H. californicus, although some peak imbalance (allele drop-out) was noticeable at loci Pl41 and Pl27 (Additional file 3). Homozygosity was only observed when P. arborescens or H. californicus produced similar-sized alleles (Pl53), or when amplification failed in one of them (Pl51). In P. limitata, alleles were shared with H. californicus or P. arborescens, or were species-specific. The congruence between the recovered genotypic clusters and the mtDNA clades was absolute (Fig. 2). As expected given its fixed heterozygosity and allele sizes, both Structure (K = 3; Fig. 2a) and FCA (Fig. 2b) analyses indicated that P. hybrida was genetically intermediate (hybrid) between H. californicus and P. arborescens. Contrastingly, allele states didn’t reveal any obvious genealogic relationship between P. limitata and its extant congeners.

Fig. 2

Genotypic entities within Pelvetiopsis as assessed with microsatellite loci. a Structure plot for K = 3, showing inferred taxonomic entities at the bottom and mtDNA clades at the top. b FCA plot, with individuals labelled according to inferred genetic entities. Note the congruence between sequence and typing data and between different analyses

A posteriori, all four genetic entities could be recognized based on their general morphology (Fig. 3). A combination of several characteristics, including maximum length, frond width, presence and distribution of cryptostomata in vegetative and reproductive parts, and shape of receptacles, was particularly helpful to confirm species identities in the field (Additional file 4). The large sampling effort of this study allowed us to establish the distribution range of each species (Fig. 4). The southernmost species, H. californicus, was genetically confirmed from Baja California to Monterey Bay whereas cold-temperate P. limitata was confirmed from Point Conception to N Oregon. P. arborescens and P. hybrida exhibited much more restricted (and apparently non-overlapping) distributions centered in the Monterey (Big Sur) and San Luis Obispo counties, respectively. The first was mostly found in the Carmel area, with the southernmost population (Pacific Valley) representing an isolated collection. P. hybrida was also found in remote San Miguel Island, and its distribution in other Channel Islands remains a possibility. No species were collected on the mainland between Mexico and Point Conception.

Fig. 3

General morphology and shape of receptacles of Pelvetiopsis spp. a Hesperophycus californicus (=Pelvetiopsis californica), b P. limitata, c P. hybrida, and d P. arborescens

Fig. 4

Inferred latitudinal distributions of Pelvetiopsis spp. Distributions based on the geographical origin of genetically-determined species samples (colour codes as in other figures), and also based in literature records. Note the overlapping ranges of species between Point Conception and Monterey Bay

Discussion

Crypticism, allopolyploidy and biogeography of Pelvetiopsis spp.

Our analyses revealed multiple taxonomic, phylogenetic and genomic inconsistencies within Hesperophycus-Pelvetiopsis, notably the occurrence of 1) more genetic entities than the two initially recognized species, 2) species/gene conflicts, and 3) stable, genome-wide heterozygosity in two entities. Such conflicts often reflect incorrect taxonomy (e.g. cryptic species) or natural evolutionary processes such as incomplete lineage sorting and introgressive hybridization, among others [56]. Here, both types of factors were readily identified.

In contrast with H. californicus, genetic data clearly supported multiple entities within P. limitata sensu lato. One of these undoubtedly corresponds to the previously described P. arborescens, as confirmed by its characteristic morphology and occurrence at the type locality (Point Lobos, [39]). Another one, the newly recognized pseudo-cryptic Pelvetiopsis hybrida, was found to be genetically intermediate between H. californicus and P. arborescens, with all available evidence indicating an allopolyploid origin (see below). All four entities corresponded to genetically discontinuous clusters that were found a posteriori to further exhibit distinctive morphological characteristics. All combinations of species pairs (with the exception of P. arborescens with P. hybrida) were found locally co-occurring in at least a few sites between Point Conception and Monterey Bay (personal observations). Examples include P. limitata, P. arborescens and H. californicus all along the Monterey peninsula, P. hybrida and H. californicus throughout San Luis Obispo County, and P. limitata and P. hybrida in the same county around Hazards (Montaña de Oro State Park). Despite close proximity, local ongoing hybridization between species was never detected. Given their genetic and morphological integrity in sympatry, these four entities (P. limitata, P. arborescens, P. hybrida, and H. californicus) clearly correspond to good biological species.

Molecular analyses nevertheless revealed a key hybridization event in the evolutionary history of Hesperophycus-Pelvetiopsis. Specifically, P. hybrida (and to a lesser extent P. limitata) exhibited fixed heterozygosity in a high proportion of microsatellite loci, with allele states clearly indicating the contribution of H. californicus and P. arborescens genomes. Genomic heterozygosity in sexually reproducing organisms is not expected beyond F1 hybrids, because independent (Mendelian) segregation of chromosomes and successive generations of mating necessarily generate a proportion of homozygotes while pulling genotypic frequencies of populations towards Hardy-Weinberg equilibrium. In selfing species and/or with very restricted gametic dispersal, inbreeding would actually promote an excess of homozygotes, as observed here in H. californicus and P. arborescens, and in other fucoids sharing similar life-history and reproductive traits [57, 58]. P. hybrida cannot result from the recurrent formation of F1 hybrids, since currently it doesn’t overlap its distribution with P. arborescens, one of its inferred ancestors. A clonal hybrid swarm is also unlikely because P. hybrida produces fully functional sexual gametes (pers. observ.) and also because vegetative propagation by Fucaceae has never been reported in intertidal wave-swept habitats. Clonal lineages are present only in sheltered environments such as salt-marshes, estuarine mudflats and land-locked bays, and exhibit distinctive morphologies characterized by dwarf and curly habit, vegetative proliferation and lack (or nearly so) of reproductive structures and anchoring holdfasts ([59–62]). Attached, open-coast clones have only been reported from the atidal, brackish Baltic-Sea, where they result from the reattachment of breaking vegetative fragments [63]. This process requires motionless time (weeks) on a substrate to develop rhizoids, something that can happen in the Baltic subtidal where sea surface can be still for prolonged periods (for example, when it freezes, leaving the algae in still water underneath), but not in the wave-swept intertidal rocky-shores sites where P. hybrida occurs.

Under sexual reproduction (including selfing), only allopolyploidy can account for the generationally stable, genome-wide hybridity of P. hybrida. At least in selfing species, allopolyploidy is more likely to result from the self-fusion of unreduced gametes of F1 hybrids than via triploid bridges [64], although the actual pathways can only be guessed at. In any case, genome duplication may have been the key event restoring hybrid fertility, since P. hybrida originates from relatively divergent species. Despite the fixed heterozygosity of individuals, the very low number of individuals exhibiting three alleles (four were never observed) in at least one locus indicates that most homeolog loci actually occur, as expected in low-dispersal selfing species, in the homozygous state.

Organellar genomes are maternally inherited in brown seaweeds [65, 66]. Mitochondrial sequence data thus allow us to identify a northern female H. californicus and a male P. arborescens as the diploid ancestors of P. hybrida. The small mtDNA divergence and the (mainly) identical microsatellite allele states further argue for a relatively recent evolutionary origin. However, we show that mtDNA haplotypes are no longer shared between P. hybrida and H. californicus, and that the modern ranges of P. hybrida and P. arborescens do not overlap. Complete mtDNA lineage sorting and range shifts can occur on relatively short time-scales (single or few glacial cycles, [7, 29]), but here the results clearly indicate that hybridization was not extremely recent and is not ongoing presently. In P. limitata, microsatellite data was also compatible with multiple homeologue chromosomes, but allele states didn’t reveal any obvious relationship with any of its extant conspecifics. Ongoing karyotyping and sequencing of multiple nuclear loci will allow confirmation of the genealogical relationships and ploidies inferred here for H. californicus, P. arborescens and P. hybrida, and explore the hypothesis that P. limitata corresponds to a second, cryptic paleopolyploid species.

The hypothesis that P. limitata had also an allopolyploid origin is striking because currently this cold-temperate species shows the widest latitudinal span of all recognized species (ca. 14°), and is the only species occurring north of Monterey Bay. By comparison, P. hybrida showed a much more restricted range in mainland California and in San Miguel Island (Northern Channel Islands). Low-dispersal organisms might take some time to migrate and colonize new habitats. It can be hypothesized that this rather recent species is still in the process of equilibrating with its climatic niche (i.e., still expanding its range). Remarkably, its inferred paternal ancestor, P. arborescens, was only found in an even narrower, non-overlapping stretch of coastline spanning about 0.7° in latitude (<130 km). This degree of endemism is rare in temperate marine species, and in related fucoids is normally associated with newly emerged, extreme habitats (e.g. Fucus radicans in the brackish Baltic Sea, [45]) and/or past climates (e.g. F. virsoides in the Adriatic Sea, [67]). Within its small range, P. arborescens was always found in small scattered patches and only along a high intertidal fringe typically above all other intertidal fucoids (JN, EAS, GAP, pers. obs.). Notwithstanding its dynamic evolutionary history, currently this species may be better described as an extremely vulnerable climatic relict [22]. Finally, the most southern species, H. californicus, also exhibited a fairly wide latitudinal range, although south of Point Conception this was very discontinuous and mostly restricted to offshore islands (e.g. Northern and Southern Channel Islands) and prominent capes (e.g Punta Baja) [68, 69]. This species exhibited a marked phylogeographical break at Point Conception that should be further investigated.

The rich coastal biota of California, including its luxuriant seaweed flora, has been extensively surveyed in the past decades; the intertidal fucoid assemblages in particular are conspicuous and easily accessible for amateurs and trained phycologists, and are routinely included in coastal monitoring programs (e.g MARINe consortium). The discovery of a new Pelvetiopsis illustrates how much cryptic diversity may be hiding within more problematic groups and in more remote areas and depths. Remarkably, all Pelvetiopsis were present along the transitional coastline between Point Conception and Monterey Bay, either as relict (P. arborescens) or evolutionarily recent (P. hybrida) endemics, or, for more widespread species, overlapping the southernmost (cold-temperate P. limitata) and northernmost (warm-temperate P. californica) parts of their ranges. This modern hotspot of biodiversity, refugial persistence and evolutionary opportunity probably results from its relatively stable climatic history and high spatial habitat heterogeneity. The narrow endemic species nevertheless raise conservation concerns and deserve finer detail monitoring, particularly P. arborescens. Its restricted range, rarity and ecologically marginal vertical distribution make it particularly vulnerable to demographic stochasticity and to extreme El Niño events and ongoing climatic changes.

The identification of three phylogenetic clades comprising four species (one of hybrid origin) calls for a taxonomic revision of the complex (Table 2). Specifically, available data do not support the historical separation of Hesperophycus and Pelvetiopsis, and a new combination for Hesperophycus californicus (Pelvetiopsis californica) is therefore proposed. In addition to the formal description of the allotetraploid P. hybrida, the practical recognition of rare P. arborescens is also warranted.

Table 2 Taxonomic update of Hesperophycus-Pelvetiopsis

Originality of allopolyploid speciation

The recognition of two narrow-endemic, pseudo-cryptic species within Pelvetiopsis is certainly remarkable, but the allopolyploid nature of P. hybrida is truly unexpected and completely original both from a taxonomic (brown seaweeds) and ecological (intertidal) perspective. Variance in genome size estimates in seaweeds have been attributed to palaeo-polyploidization events [35, 70], including in Fucales [71], but direct genetic evidence for modern polyploids remains scarce [38, 72] and nearly non-existent in brown seaweeds (but see [60]. Autopolyploids are harder to detect genetically and may be underreported [73], but not allopolyploids exhibiting stable genomic hybridity (e.g. [74]). For instance, unlike P. hybrida, suspected Macrocystis x Pelagophycus allopolyploids could not be confirmed genetically based on this benchmark (revised in [75]). Surprisingly, positive correlates of plant polyploidy, including perenniality and ability to self, are common in many speciose genera of canopy-forming brown seaweeds (Fucus, Sargassum, Cystoseira, Laminaria, Saccharina), some of which also display some propensity to hybridize. For instance, in the closely related and well-studied genus Fucus, hybridization between related species has been frequent and ongoing but the primary evolutionary consequence has only been introgressive gene-flow [29–31].

Why is allopolyploidy featuring in recent Pelvetiopsis evolution while virtually unknown across its entire class? Globally, whole genome duplication appears to require relatively stringent conditions; hybridization for instance is taxonomically widespread and frequent [76], whereas polyploidy is common in plants [77] and a few vertebrate lineages [78] but infrequent in most other taxonomic groups [37, 79, 80]. Even in plants, high variation in the frequency of polyploidy implies different propensities for polyploidization, or at least polyploid establishment, in different lineages [81]. In related Fucus, chemotactic attraction of sperm by eggs is cross-specific [82], and the potential for gamete contact in mixed stands can be high (e.g., during synchronous joint spawning events, but see [83]). Yet, hybridization (leading to introgression) is only known between related species and is often asymmetric [29, 30]. For instance, natural hybrids normally involve male (sperm) F. serratus x female (eggs) F. evanescens, presumably because sperm allocation is much higher in dioecious vs hermaphrodite species [84]. Speciation of P. hybrida on the other hand involved the interaction of two divergent self-compatible hermaphroditic species. This is unlikely to be a coincidence. Allopolyploids seem to originate more frequently when crosses involve genomically divergent species ([85]; but see [86]) because more severe meiotic conflicts presumably prevent diploid hybrids from backcrossing/introgressing with parental species (or to self), and result in higher numbers of unreduced (diploid) gametes. This is consistent with inter-lineage allopolyploid speciation in Pelvetiopsis vs intra-lineage (most likely homoploid) introgression in sister Fucus. It does not really offer, however, a general explanation for why allopolyploidy is otherwise so rare among brown seaweeds.

It should be noted that P. hybrida was found co-occurring in sympatry with its ancestor P. californica (= H. californicus) throughout much of its distribution. Niche differentiation is acknowledged as a pre-requisite, or at least facilitator, allowing initial establishment and long-term survival of viable (but numerically rare) neopolyploids as stable species, by alleviating potential competition with longer-established diploid ancestors [87–90]. Both polyploidy and hybridization have saltation effects, and any adaptive phenotypic trait may be more readily conserved (and potentially reinforced) due to cytotype isolation. In any case, ecological differences need not be great. Floristic meta-analyses show that budding speciation (new species arising within the ranges of older, widespread persisting taxa) is common in plants, with sister species often differing in subtle habitat preferences conducive to fine-scale parapatry/allopatry [18]. The importance of fine-scale environmental heterogeneity and of reproductive traits (such as selfing or timing of reproduction) is also known to promote species divergence in the shallow marine realm. In sister Fucus, for instance, the integrity and vertical distribution of closely related species is maintained along strong vertical gradients by small differences in stress tolerance, even in the face of moderate gene-flow [14]. Similar patterns are also apparent in mixed Pelvetiopsis stands occurring between Point Conception and Monterey. The average vertical ranges of P. hybrida and P. arborescens were always above those of H. californicus and P. limitata where they co-occurred (unpublished data). Likewise, very local (e.g. boulder level) patterns of distribution suggest an important role for wave action in limiting their distributions. Whether these ecological differences are among the immediate effects of allopolyploidization is unclear, since they cannot be separated from other evolutionary processes (e.g. drift, selection) driving subsequent niche divergence [91].

Because of the very limited genomic differentiation of P. hybrida from its extant ancestors, this species is particularly suited to investigate how divergent parental genomes are transferred and functionally integrated in a new species and might contribute to adaptive diversification. The rocky intertidal zone provides one of the best natural laboratories for examining the relationships between abiotic constraints, biotic interactions, and ecological patterns in nature [92]. A few sites where different combinations of species coexist have already been identified, and field observations and experimental work may identify niche differences allowing coexistence at these local scales, including among species with partially shared genetic backgrounds.

Conclusions

This study revealed a much more diverse and evolutionarily complex picture of NE Pacific Pelvetiopsis than previously anticipated. In addition to the confirmation of the Big Sur endemic P. arborescens, molecular data allowed for the first time the identification of a stable allopolyploid marine species (P. hybrida) with clearly discernable phylogenetic relationships (♀ H. californicus x ♂ P. arborescens), morphology, and geographical distribution. Ongoing analyses of nuclear sequence data support our inferences regarding P. hybrida, and are consistent with the hypothesis of P. limitata corresponding to a second, cryptic paleopolyploid. Allopolyploid speciation remains otherwise unknown in brown seaweeds, and until some ecological, biological or reproductive originality is identified, its unique occurrence within Pelvetiopsis will remain enigmatic. The diversity hotspot between Point Conception and Monterey Bay likely reflects its relatively stable climatic history and its regionally high habitat heterogeneity. This conspicuous seaweed genus from the well-studied Californian intertidal illustrates how much diversity (and evolutionary processes) potentially remains undiscovered in other (less studied) marine groups and regions.