Genetic variation within and among asexual populations of Porphyra umbilicalis Kützing (Bangiales, Rhodophyta) in the Gulf of Maine, USA

Population structure

Our study considered two populations collected from estuarine tidal rapid sites (Dover Point and Wiscasset), and five populations from open-coastal sites (Fort Stark, Quoddy Head, Schoodic Point, Reid State Park, and Nubble Light). The estuarine tidal rapid sites were very similar to each other; individuals from the estuarine tidal rapid sites were collected from bridge pilings in the high intertidal zone where few other organisms were found, presumably because larger organisms would be damaged by the currents produced by tidal flow. The open-coastal sites were more diverse, and individuals were collected from a number of different rock types, exposures, and in areas that experienced considerable variation in wave action and force.

Despite the habitat similarity of the Dover Point and Wiscasset locations, the population from Dover Point shared its lowest genetic distance (i.e. its highest genetic similarity) with the neighboring open-coastal populations at Fort Stark (d=0.066) and Nubble Light (d=0.041). The population at Wiscasset shared the lowest genetic distances with its closest sampled open-coastal population at Reid State Park (d=0.206). The population from Wiscasset was very divergent from other populations, with genetic distances ranging from 0.206 with its closest open-coastal population at Reid State Park to 0.557 with the population at Schoodic Point (Table 4). The genetic differences are largely driven by a high frequency of genotype 5 in the Wiscasset population, which is relatively rare in other populations (Figure 1). Such a high frequency of an otherwise rare genotype could suggest the possibility of a bottleneck during colonization of the bridge pilings at Wiscasset, and the presence of this genotype at the closest open-coastal population at Reid State Park could suggest the two populations share a source population.

Genotype 5 was also present in populations from Fort Stark and Nubble Light, but was not present at Dover Point, and therefore probably does not represent a genotype that is particularly well adapted to estuarine habitats. More importantly, however, the high genetic distances between Dover Point and Wiscasset and the divergent genotype frequencies suggest that isolation-by-distance is more important to population structure than selection within habitat. The estuarine tidal-rapid populations at Dover Point and Wiscasset are more closely related to their closest open-coastal population than they are to each other.

Despite this, isolation-by-distance is not the dominant force shaping population structure among all populations in the Gulf of Maine. A Mantel test among all populations found no correlation between Nei’s unbiased genetic distances and geographic distances (Mantel test: R²=-0.10, p=0.60). Analysis of Molecular Variance (AMOVA; Excoffier et al. 1992) partitions the variation in allele frequencies within and among populations, and more variation among populations than within populations might suggest genetic isolation and differentiation due to genetic drift or selection. For P. umbilicalis populations in the Gulf of Maine, however, AMOVA revealed that variation was partitioned almost equally within and among populations; slightly more of the variation was within populations (58%), though a considerable proportion was partitioned among populations (42%, Φ_PT=0.42, p<0.001). There is thus no evidence for habitat selection or genetic isolation by distance using these genetic markers.

Lacking direct evidence of habitat selection and isolation-by-distance among all populations, the population structure of P. umbilicalis in the Gulf of Maine may be dictated largely by stochastic forces, such as bottlenecks (as for Wiscasset) and genetic drift. Isolation-by-distance appears to be an important factor dictating population structure of estuarine populations, but not an important factor dictating population structure within the Gulf of Maine as a whole. This does not mean that selection for habitat does not occur. The lack of recombination among genotypes means that adaptation in one population may occur along a different trajectory from that in a population in a similar habitat, rather than in parallel with that population. Putatively neutral genetic markers such as these are probably poor markers for identifying population differentiation in asexual populations, because there is no guarantee that the genomes of individuals containing the same EST-SSR are identical. Based on these data presented here, it appears the initial distribution of these genotypes was stochastic, with the exception of estuarine populations, which were probably founded from nearby open-coastal populations after the initial distribution of genotypes. Subsequent diversification from the initial stochastically-dispersed lineages could have occurred in each population, and without sampling on a more genomic scale, there is no way to know if habitat selection shapes population structure in the Gulf of Maine.

New Hampshire lineages

The most genetic diversity at EST-SSR loci within populations was found at Fort Stark and Dover Point, New Hampshire. The population at Dover Point, however, was sampled more often than the other populations, which could suggest that there is a temporal effect on diversity. In addition to being the most genotypically diverse, populations from New Hampshire were more closely related to each other and to populations in southern Maine than to populations from northern Maine based on Nei’s unbiased genetic distances calculated using genotype frequencies (Table 4, Figure 3). The shared genotypes (particularly genotype 3) among them suggest that Fort Stark or other nearby southern Maine populations such as Nubble Light, rather than distant populations from Maine, provide some of the genetic material that the population at Dover Point is derived from.

Genotypes 3 at Dover Point and Fort Stark, and genotype 6 at Fort Stark may represent more recent mutations than the other genotypes sampled in this study. They differ from genotype 1 by one allele at different loci, and neither genotype is found in populations from Maine. The degree of similarity among genotypes 1, 3 and 6 suggests the genotypes are related (Figure 2), and the widespread geographic distribution of genotype 1 and its presence among the group of putative recombinant genotypes (discussed below, shown in Table 3) suggest that genotype 1 is probably the ancestral genotype from which genotypes 3 and 6 are derived. Genotype 3 and 6 thus represent lineages created by more recent mutations than those found in other genotypes. If we assume a step-wise migration from the NE Atlantic Ocean via Iceland and maritime Canada in the post-glacial era, as Ingolfsson (1992) does for rocky shore invertebrates, then Fort Stark as the southernmost population may represent the most derived population of Porphyra umbilicalis in this study. The population may represent an accumulation of various genotypes that evolved in more northerly locations and moved south with the prevailing current in the Gulf of Maine. Further studies using controlled seasonal collections on more southern populations would be necessary to test this theory, however, as other modes of transport across the Atlantic were possible (Brawley et al. 2009).

Genotype 6 is the only genotype unique to a single population and is found only at Fort Stark (Figure 1). The genotype was highly reproducible in replicate collections (data not shown), individuals possessing the genotype had 100% identity to voucher P. umbilicalis samples within the sequenced region of rbcL-rbcS intergenic spacer, and the genotype could be re-sampled from a precise location in the field. Individuals of that genotype appear to be limited to a single rock on which individuals experience considerable turbulence of low velocity due to the position of that rock in the surf. Individuals tend to have shorter thalli and are of a lighter color. By contrast, genotype 3 was very common at Fort Stark in other locations along the shoreline, and fairly common at Dover Point. The number of samples at Dover Point is small due to the small size of the population, and subsequent collecting trips to the Dover Point site to augment the collection have been difficult due to construction at the U.S.-4/NH-16 bridge (www.newington-dover.com). Our most recent trips have not found any P. umbilicalis individuals. Indeed, the transect sites in West et al. (2005) have apparently been covered by the pilings for the new bridge.

Putative recombinant genotypes?

Four common genotypes (genotypes 1, 2, 4, 5; Table 3) were the product of either convergent evolution of alleles (homoplasy) or a recombination event. The mode of asexual reproduction by mitotic differentiation of blade cells means that there is no recombination of genetic material in populations of Porphyra umbilicalis in the NW Atlantic, and therefore linkage among loci should be complete, regardless of their location on the chromosomes. The linkage among loci 2179 and 2219 is not complete, because alleles A and B from locus 2179 were found with either allele A or B from locus 2219. The lack of complete linkage among these loci suggests that some alleles arose independently in different lineages through convergent evolution, or that a recombination event disrupted the linkage between these two loci. Mitotic recombination could not explain these data because only the haploid life history stage of this species is present in the NW Atlantic, and thus a mitotic recombination event would be undetectable unless it involved unequal crossing over.

The most parsimonious explanation for these data is a recombination event, which casts doubt on the assumption that these populations reproduce exclusively asexually in the NW Atlantic. However, if a local recombination event created these genotypes, we might expect to find all four possible recombinants in a single population. All four putative recombinant genotypes were sampled, but all four were not sampled in a single population. A limited number of populations and individuals from each population were sampled, however, and the missing recombinant genotype from each population could easily have been overlooked by chance.

While these putative recombinant genotypes appear to be the strongest evidence thus far that sexual reproduction in the NW Atlantic could exist, researchers have looked for evidence of sexual reproduction in the NW Atlantic in the past and come up lacking (Blouin and Brawley 2012). In light of that research, a number of possible explanations exist to explain the data presented here without invoking sexual recombination in the NW Atlantic. Our data could be explained by convergent evolution of microsatellite lengths (Figure 4, Scenario 1), which has been found to exist in other organisms (Garza and Freimer 1996, Ortí et al. 1997). Alternatively, the putative recombinant genotypes could represent genotypes resulting from a historic recombination event that took place before the ability to reproduce sexually was apparently lost (i.e. before colonization of the NW Atlantic, Figure 4, Scenario 3) and tissue carrying those genotypes could have subsequently migrated to the NW Atlantic. Given the ambiguity of the origin of these genotypes and previous evidence to the contrary (Blouin and Brawley 2012), it is not possible at this time to claim definitively that sexual reproduction occurs in NW Atlantic populations of P. umbilicalis, nor is it possible to differentiate between these alternative theories that do not invoke recombination in the NW Atlantic.

Figure 4:

Four possible scenarios that explain colonization of the NW Atlantic from the NE Atlantic that attempt to reconcile data from this study and others that find no sexual reproduction in the NW Atlantic.

How many introductions?

Populations of Porphyra umbilicalis found on NE Atlantic shores reproduce both sexually and asexually (Brodie and Irvine 2003), but NW Atlantic populations are not known to reproduce sexually (Blouin et al. 2007, Blouin 2010, Blouin and Brawley 2012). If NW Atlantic populations are derived from NE Atlantic populations (Teasdale and Klein 2010), it is not clear why NW Atlantic populations are apparently incapable of reproducing sexually (though to our knowledge, no one has studied ecological differences in sexual and asexual populations in the NE Atlantic that may prevent sexual populations from becoming established in the NW Atlantic). Efforts have been made to detect the signatures of sexual reproduction and there is no current evidence that sexual reproduction occurs (Blouin and Brawley 2012).

However, an asexual mode of reproduction would have been highly advantageous for P. umbilicalis as it colonized the NW Atlantic from European refugia (Figure 4, Scenario 1). The advantage of asexual reproduction or other forms of self-compatibility in colonizing species has long been known to invasive species biology (Baker and Stebbins 1965). Organisms that are capable of reproducing asexually are considered more likely to be successful in colonizing new habitats than species that require an alternative mating type. For species with alternative mating types to be successful in colonizing new habitats, the colonization event by necessity must include sufficient numbers of both mating types to ensure reproduction. The statement has been codified as “Baker’s Law” (Baker 1967, Cheptou 2011), and support has been found in a number of noxious, terrestrial invasive species (Hollingsworth and Bailey 2000, Hao et al. 2011). Most P. umbilicalis individuals are dioecious when sexually reproductive (Brodie and Irvine 2003, Brodie et al. 2008), and thus would have required the introduction of multiple mating types to sustain a sexual population. Some individuals contain both male and female reproductive structures (Holmes and Brodie 2004), but it is not clear precisely how common this trait is in P. umbilicalis, or how heritable that trait is. Holmes and Brodie (2004) found that only 5% of Porphyra dioica individuals in Devon, U.K., which is usually dioecious like P. umbilicalis, contained both male and female reproductive parts, and P. umbilicalis individuals apparently with both male and female reproductive tissue in the Faroe Islands are probably equally rare (personal communication, Agnes Mols-Mortensen). In addition, if a rare individual containing both male and female reproductive structures were to colonize NW Atlantic shores, it is not clear what percentage, if any, of subsequent generations would be similar. Male reproductive structures are apparently absent from NW Atlantic populations of P. umbilicalis at this time (Bird and McLachlan 1992). The ability to reproduce asexually via neutral spores would have provided the initial colonist (whether that colonist was an asexual individual as in Figure 4 Scenario 1, or an individual with both male and female reproductive parts as in Figure 4 Scenario 2) the ability to establish a population and persist following a single introduction.

Support for Baker’s Law is somewhat limited (Rambuda and Johnson 2004, Grimsby et al. 2007, Barrett et al. 2008), however, because multiple introductions appear to be quite common in terrestrial plants (reviewed in Dlugosch and Parker 2008) and marine colonizers/invaders (Dlugosch and Parker 2008, Brawley et al. 2009), thereby eliminating the advantage of clonal invasives. Indeed, there is evidence that the colonization event of P. umbilicalis did in fact involve multiple individuals. If we assume that NW Atlantic populations are incapable of sexual reproduction, then it is possible the putative recombinant genotypes identified in the present study represent a recombination event that occurred prior to the colonization of NW Atlantic shores. If these putative recombinants are relics of ancestral, sexual populations in Europe, then multiple introductions from refugia in Europe would have occurred for all four putative recombinant genotypes to occur in the NW Atlantic (Figure 4 Scenario 3).

Therefore the answer at this time to the question regarding how many introductions of P. umbilicalis from the NE Atlantic to the NW Atlantic occurred remains unclear. Teasdale and Klein (2010) suggested there was just one introduction of material from European refugia, but their markers were relatively conserved ribosomal intron and internal transcribed spacers, which could have missed cryptic genetic diversity. Given that P. umbilicalis is capable of asexual reproduction, a single introduction could have established the populations in the NW Atlantic, and asexual reproduction would have been an advantage to colonizers (Figure 4 Scenario 1 and 2). However, our data suggest that multiple introductions from the NE Atlantic could have occurred (i.e. all four putative recombinant genotypes can be found in the Gulf of Maine; Figure 4 Scenario 3). If we allow for the possibility of multiple introductions, however, it begs the question of why individuals capable of sexual reproduction were not included, and why we do not find sexual populations in the Gulf of Maine that were themselves capable of producing the recombinant genotypes (Figure 4 Scenario 4). The introduction of a single individual containing both male and female reproductive parts seems to explain all aspects of the data (Figure 4 Scenario 2), but the apparent rarity of these individuals in the NE Atlantic makes this possibility seem unlikely.