Introduction

Populations of species that are involved in range expansion often have small effective population sizes and can fix beneficial mutations through a combination of selection and genetic drift that often includes asexual reproduction (Hallatschek et al. 2007; Hallatschek and Nelson 2010; Song et al. 2011). Asexual reproduction can be an effective mechanism to establish a new population; however, it is not considered to confer adaptive advantage over longer evolutionary time scales because of asexuals presumed lack of genetic diversity (Otto 2009). Contrary to this idea, recent work with an asexual, clonal yeast that was dividing mitotically showed rapid local adaptation to salinity stress (Bell and Gonzalez 2011; Dhar et al. 2011). This adds to our understanding of persistent asexual reproduction in populations at the extremes of their ranges (e.g., in algae: Dixon 1965; Pearson and Murray 1997; Sosa and Lindstrom 1999; and references therein).

The intertidal red alga Porphyra umbilicalis has a typical sexual life history in the northeastern (NE) Atlantic (e.g., Brodie et al. 2008), but individuals in at least part of the northwestern (NW) Atlantic metapopulation appear to reproduce exclusively by asexual neutral spores released from blades (Blouin et al. 2007). The current described range for P. umbilicalis is from Long Island Sound to Greenland in the NW Atlantic and from Norway south to Portugal in the NE Atlantic (Sears 2002; Brodie et al. 2008; Teasdale et al. 2009). Influences on the NW Atlantic’s intertidal marine biota include extirpation by glaciation and recolonization by both natural processes and anthropogenic vectors (Maggs et al. 2008; Brawley et al. 2009; Muhlin and Brawley 2009; Adey and Hayek 2011). Given the lack of observed sexual reproduction in the NW Atlantic P. umbilicalis and recolonization of the NW Atlantic shores (from Europe or small NW Atlantic refugia) by P. umbilicalis after the last glacial maximum (LGM), the putative asexual population in Maine may be comprised of a group of clones. This is tested here with AFLPs (Vos et al. 1995).

In addition to investigating whether or not P. umbilicalis is clonal in Maine, we evaluate AFLP-relevant contamination of our field collections, provide an in silico approach to identifying additional potential contaminants, and discuss the importance of their removal when conducting an investigation with AFLP DNA fingerprinting. AFLPs are considered to be highly reproducible and robust against methodological artifacts, but can suffer from different types of scoring errors. Many of the nearly 7000 AFLP studies cited in the BIOSIS Citation Index make inferences concerning individuals or populations from wild-collected material. Some of these AFLP studies evaluated methodological or methylation errors (Donini et al. 1997; Bonin et al. 2004) or the proper analysis of AFLP data to avoid confounding results (Meudt and Clarke 2007; e.g., from heterozygotes, multiple alleles, and deviations from Hardy–Weinberg equilibrium; Bonin et al. 2004; Foll et al. 2010). Largely missing from the literature, aside from general warnings, is how to ensure removal of non-target DNA from AFLP DNA fingerprinting studies.

Materials and methods

Individual P. umbilicalis thalli were collected from three intertidal locations in the North Atlantic: Schoodic Point, ME (44° 20′ 20″ W 680 2′ 42″ (n = 25)); Cobscook Bay, ME (44o 52′ 54″ W 67° 7′ 53″ (n = 25)); and Sidmouth, England (50° 40′ 38″ W 3° 14′ 23″ (n = 17); Fig. 1). Porphyra umbilicalis haplotype P.um1 (UTEX #LB 2951, isolated from Schoodic Point in 2008 and being sequenced by The Joint Genome Institute, JGI) was included as a positive control against methodological artifacts in all of the AFLP reactions and was included in our analyses (adjusted sample size for Schoodic Point, n = 26). The Maine samples were examined for epiphytes and endophytes by brightfield microscopy immediately after collection. As a precaution against contaminating DNA, each individual was washed vigorously in soapy, deionized water for 1 min, and duplicate samples were frozen in liquid nitrogen (LN2). To further determine whether our general sample preparation successfully removed contaminants, we performed AFLP analysis on another group of wild-collected plants (n = 3 individuals) from Schoodic as follows: The AFLP profiles for these individuals were generated in duplicate from equivalent longitudinal sections of each blade that were either (1) rubbed with paper towels (treatment: minimal cleaning), (2) cleaned in the manner described above (treatment: vigorous cleaning), or (3) cleaned and then subcultured in the antibiotic conditions described below for the English site (treatment: vigorous cleaning and antibiotics).

Fig. 1
figure 1

Collecting sites in the North Atlantic with a detailed map showing where in Maine and England collections were made

Full size image

The Schoodic Point and Cobscook Bay collections of P. umbilicalis from ~300 × 30 m sections of high-mid-intertidal zone at each site were collected using a transect line and a random numbers table. A 10-knotted string dropped onto the clumps so located allowed individuals to be collected randomly because knot position was determined with a random numbers’ table. Sidmouth (UK) individuals were collected haphazardly at low tide from a single boulder (~1 m3) on a predominantly sandy beach, blotted dry, and express-transported for our preparation. Sidmouth samples were examined for endophytes and epiphytes with light microscopy, vigorously washed in soapy water for 1 min, and separated into individual cultures for 1 week. All Sidmouth individuals were producing either male or female gametes. For culture, individual thalli were placed in 1 L Erlenmeyer flasks with 800 mL 1 μm-filtered, autoclaved seawater with ½ Provasoli’s enrichment solution (WES; West and McBride 1999), 4 mg L-1 GeO2, 10 mg/L penicillin G, and 2.5 mg L−1 dihydrostreptomycin sulfate (1× PenStrep; Guillard 2005). The cultures were gently bubbled (~1 L min−1) at 60 μmol photons m−2 s−1, 10 °C, 14:10 (L:D in Percival I36LLVL incubators (Percival Scientific, Inc., Perry, IA)). All individuals were determined to be free of endophytes or epiphytes by light microscopy after 1 week, and duplicate samples were frozen in LN2.

DNA was extracted from cleaned, frozen tissue using a DNeasy Plant Mini Kit (Qiagen Inc. Gaithersburg, MD) following the manufacturer’s instructions. The Small Plant Genome Mapping Kit (Applied Biosystems Carlsbad, CA) was used to generate the AFLP profiles for this investigation according to the manufacturer’s instructions (after Vos et al. 1995). Briefly, 50 ng P. umbilicalis DNA was digested with EcoRI and MseI endonucleases (New England Biolabs, Ipswich, MA). Ligation of adaptors for preselective amplification was carried out overnight at room temperature (T4 ligase, New England Biolabs). Preselective PCR amplification was carried out with no selective bases (EcoRI+ 0 and MseI+ 0; Iitsuka et al. 2002; Yang et al. 2003; Niwa et al. 2004). Subsequently, selective PCR amplification of targets was accomplished with four fluorescently labeled EcoRI primer sets (EcoRI-AC/MseI-CAA, EcoRI-AG/MseI-CAA, EcoRI-AG/MseI-CG, EcoRI-AC/MseI-CT). Selective PCR products were analyzed on an ABI 3730 analyzer to create a multilocus genotype (MLG) from a presence–absence binary matrix using GeneMarker 1.8 software (SoftGenetics LLC State College, PA). Duplicate samples were extracted and amplified from every individual used throughout this investigation to ensure repeatability of AFLP scoring. When the presence or absence of any marker was inconclusive, additional preselective and selective PCRs were performed and compared to the original duplicates. Only those bands that occurred in duplicate runs were scored as present. Additionally, our laboratory has generated genome scaffolds assembled by JGI for 3 planctomycete contaminants known to survive antibiotic treatment in P.um1 cultures. We subjected these scaffolds to in silico AFLP analysis. Any bands from the P. umbilicalis individuals also found in the in silico AFLP analysis of the planctomycete scaffolds were removed from further analysis.

To investigate the genotypic diversity and the presence of clones within the Maine populations, we used the tools included within the GenoDive 2.0b22 software package (Meirmans and Van Tienderen 2004). Genetic distances were calculated using an infinite allele model (IAM), which assumes that a single mutation changes the state of any detected marker within an individual’s MLG. The bimodal distribution of the pairwise distances and a comparison to the maximum pairwise distance was used to set an appropriate threshold for the assignment of a MLG to a single multilocus lineage (MLL, or “clone”) for the Maine populations. We tested for the presence of clonal structure within the dataset by randomizing AFLP markers over individuals and comparing the observed diversity with the randomized set (Gomez and Carvalho 2000). If the threshold were set too high, we would over-estimate the amount of variation in the MLG attributable to scoring error or somatic mutation and fail to reject the possibility of sexual reproduction within those individuals assigned to a single clone (i.e., sexual reproduction is suggested by p sex > 0.05). Diversity associated with putative clonal asexuals compared to individuals known to be sexually reproductive was calculated with the following indices: number of genotypes, number of effective genotypes, and Simpson’s diversity index. We also performed an analysis of molecular variance (AMOVA) to examine genetic diversity within and between sites. Finally, the genetic similarity between individuals was calculated using the Primer v6 software (Primer-E Ltd., Ivybridge, UK).

Results

Sensitivity analysis for non-target DNA

A total of 566 polymorphic markers were scored for all individuals in this study. The in silico analysis of planctomycete scaffolds revealed that the presence of 10 AFLP bands in the P. umbilicalis preparations (1.8 % of P. umbilicalis bands) represent fragment sizes that may be contributed by known planctomycete contaminants of P. umbilicalis; these planctomycete bacteria remain associated with blades after vigorous washing. The putative contaminant markers were removed from further analysis, leaving a total of 556 markers. Verification of our cleaning protocol using a subset of 55 informative bands showed that vigorous cleaning was the most significant factor in reducing variation due to contaminants in AFLP profiles between preparation methods. Contamination from epiphytic bacteria and microalgae was still responsible for up to 15 % of the bands detected after minimal cleaning; however, another 1–2 bands/individual were removed in the vigorous cleaning and antibiotic treatment compared to vigorous cleaning alone (Fig. 2).

Fig. 2
figure 2

Single linkage dendrogram generated from Bray–Curtis similarity coefficient showing that vigorous washing of field-collected samples greatly reduced contamination from bacteria and microalgae living on the surface of wild-collected P. umbilicalis

Full size image

Similarity analyses

Markers in this study revealed 58 distinct MLLs (21, 20, and 17 for Schoodic, Cobscook, and Sidmouth, respectively), two MLLs, one at each Maine site, were comprised of 6 individuals each, whereas no clonal signature was found in the Sidmouth samples (Table 1). The frequency distribution of pairwise distances was weakly bimodal, suggesting the presence of non-sexual genomic rearrangements (Meirmans and Van Tienderen 2004; also see Lasso 2008 for discussion). The threshold for defining clonal lineages for the Maine collections was set to 2.7 % (Fig. 3). This threshold corresponds to the 2–4 % threshold cited in previous studies (e.g., Arens et al. 1998; Winfield et al. 1998; Van der Hulst et al. 2003; Douhovnikoff and Dodd 2003; Lasso 2008). The probability that the 10 individuals comprising these two clones came from sexual recombination is not supported (p sex = 0.001 for those clones at both Maine sites, Table 1). AMOVA results showed that the greatest amount of genetic diversity was found within sites rather than among sites (Table 2), and the Bray–Curtis genetic similarity analysis showed that the Sidmouth site exhibits the greatest within-site diversity as compared to the individuals from either of the two Maine sites (Fig. 4).

Table 1 Descriptive statistics of genetic diversity parameters; n—sample size; T—threshold for multilocus genotype; MGD—maximum genetic distance in IAM; Clones MLL—number of Multilocus lineages detected at threshold; Clonal fraction—(n-Clones)/n; Divpop—Simpson’s diversity index; MLL = 1×—multilocus lineage sampled once; MLL >1×—multilocus lineage sampled more than once; p sex—probability that sexual recombination is responsible for the repeated genotypes assigned to a MLL
Full size table
Fig. 3
figure 3

Frequency distribution of genetic distances calculated by IAM for the Maine collection multilocus genotypes, the number of clones represented at each distance, and the threshold cutoff (vertical line) for assignment to the same multilocus lineage (MLL, clone). Genetic distances to the left of the threshold are assumed to be the result of non-sexual genomic rearrangements

Full size image
Table 2 Analysis of molecular variance (AMOVA) for P. umbilicalis individuals from Cobscook, Schoodic, and Sidmouth
Full size table
Fig. 4
figure 4

Multidimensional scaling ordination plot derived using the Bray–Curtis coefficient showing similarity of P. umbilicalis using AFLPs for individuals in this study (UK = Sidmouth, B = Schoodic Point, M = Cobscook Bay)

Full size image

Discussion

AFLP analysis of P. umbilicalis. Although this study identified one clone at Schoodic that was sampled 6× and another clone at Cobscook that was sampled 6×, P. umbilicalis in Maine is neither a single clone nor a small subset of clones. We are not able to determine robustly whether the remaining 39 MLLs from the Maine collections are asexual clones or were produced by sexual reproduction. However, the genetic similarity of all Maine individuals was high, and we have never observed sexual reproduction in the field in the northwest Atlantic, and that the closest source of sexual individuals is in Europe (Brodie and Irvine 2003; Blouin et al. 2007; Brodie et al. 2008). As expected, the (sexual) Sidmouth population displayed greater within-site variation when examined with AFLP DNA fingerprinting than the (putative asexual) Maine populations. The larger variation found within the Sidmouth site in comparison with the two Maine sites (Fig. 4) is further demonstrated by the scales over which the individuals in Maine and Sidmouth were collected. The Maine individuals were collected over several hundred meters of shoreline at each site whereas Sidmouth individuals were all from a single small boulder.

Teasdale and Klein (2010) recently reported that the genetic divergence found in Porphyra from both sides of the Atlantic is consistent with a recolonization of North America by a European population sometime after the LGM through drift or commerce (see Provan and Bennett 2008 for review). It should be noted that transatlantic introductions through shipping vectors (e.g., Brawley et al. 2009) would almost certainly have included sexual individuals given their prevalence in Europe (Brodie et al. 2008). Thus, the apparent lack of sexual differentiation in Maine P. umbilicalis remains notable. There is mounting evidence that many littoral animals and algae survived the last LGM in NW Atlantic refugia (Young et al. 2002; Govindarajan et al. 2005; Maggs et al. 2008; Muhlin and Brawley 2009; Coyer et al. 2011b) and that repeated cycles of glaciation have affected NW Atlantic intertidal algal genetic diversity more than in the NE Atlantic (Cunningham 2008; Olsen et al. 2010; Coyer et al. 2011a). This suggests that P. umbilicalis could have experienced a strong bottleneck effect from a previous glacial period and that this population might have survived the LGM in offshore refugia as a “trailing” edge population (Hampe and Petit 2005; Provan and Maggs 2012). Although asexual reproduction could favor persistence of P. umbilicalis in the NW Atlantic, there are numerous related Bangiales (e.g., B. birdiae, P. purpurea) with transatlantic distributions that reproduce primarily by sexual reproduction throughout their distribution, including in the NW Atlantic (Sutherland et al. 2011). Thus, the apparent absence of sexual reproduction in P. umbilicalis in Maine remains puzzling, and additional phylogeographic studies of P. umbilicalis and sister taxa may help solve the conundrum.

In addition to the potential for cryptic non-target DNA discussed below, genome rearrangements mediated by transposable elements (TEs) or viruses are known to be a major source of genomic variation for photosynthetic organisms (Slotkin and Martienssen 2007) and could account for the differences in AFLP profiles for putative asexuals (Maine collections) in our study. Further, theoretical and empirical assessments of the ability of asexual species to adapt to new environments find that rates of mitotic recombination may be adequate to allow asexual populations to adapt as quickly as sexual ones to new environments (Feschotte and Pritham 2007; Birky 2010). Finally, equivalent nucleotide substitution rates have been found in both sexual and asexual bdelloid lineages (Welch and Meselson 2001; Mandegar and Otto 2007). If similar rates of substitution or recombination exist in P. umbilicalis, this could explain the absence of sexual differentiation observed in extensive field and culture studies yet the high levels of AFLP-determined diversity demonstrated in this study.

Identification and removal of non-target DNA

Non-target DNA from bacteria and microalgae are potentially confounding agents in AFLP DNA fingerprinting; however, the potential for contamination is rarely addressed formally in AFLP studies (Bonin et al. 2004). Simply wiping our field samples after collection removed some non-target DNA; however, vigorous washing in deionized water with detergent successfully removed nearly all of the non-target DNA as compared to vigorous washing followed by treatment with two antibiotics (penicillin and streptomycin). Caution should be used when applying our protocol to other organisms because our cleaning procedure was specifically designed to remove non-target DNA while not damaging the P. umbilicalis. Porphyra umbilicalis has a simple flat morphology, making it relatively easy to clean blades as described. Additionally, the epiphytic organisms that were removed in our vigorous cleaning treatment are adapted to living in full salinity seawater (32 PSU). The efficacy of our cleaning procedure was improved by the ability of deionized water to cause many organisms growing on the surface of the P. umbilicalis thalli to rupture, whereas P. umbilicalis is tolerant to these conditions.

Additional, non-target DNA sources may remain cryptic because they are not removed by the most rigorous cleaning protocols that still allow for successful extraction of target DNA. We were able to identify and remove potential planctomycete contamination because of our knowledge that planctomycete bacteria survive in unialgal Porphyra cultures that have been treated with penicillin and streptomycin, and because of the availability of their complete genomes. Planctomycete bacteria are resistant to β-lactams (penicillin G) and aminoglycosides (dihydrostreptomycin sulfate), and P. umbilicalis is killed by many antibiotics that eliminate epiphytic planctomycetes (Fuerst et al. 1997; Cayrou et al. 2010). In our approach, we have removed these bands from our analysis because they offer a potential source of contamination; further analysis is required (excision and sequencing of bands removed from analysis) to prove that these bands are planctomycete in origin. Although some investigators predict that whole genome analysis will replace molecular methods like the AFLP technique (e.g., Schlotterer 2004), it is just as likely that whole genome approaches will still be too expensive for many targets for the foreseeable future (Meudt and Clarke 2007; Gaggiotti 2010). Although most investigations do not have readily available reference genomes for known contaminants, recent advances in sequencing technology have resulted in a dramatic increase in the genetic data now available in public databases including whole genome sequences of many bacteria. Our in silico data successfully identified previously cryptic non-target DNA; however, we cannot rule out the presence of additional non-target DNA that was not removed by our cleaning procedure (e.g, bacteria living within the cell wall of P. umbilicalis). Even so, the application of an appropriate cleaning protocol and application of in silico AFLP analysis could identify potential contaminants from species known to be associated as parasites or commensals with the studied organism, and greatly improve the ability to interpret AFLP data from field-collected organisms.