Species distribution models and mitochondrial DNA phylogeography suggest an extensive biogeographical shift in the high-intertidal seaweed Pelvetia canaliculata

Journal of Biogeography (J. Biogeogr.) (2014)

ORIGINAL ARTICLE

Species distribution models and mitochondrial DNA phylogeography suggest an extensive biogeographical shift in the high-intertidal seaweed Pelvetia canaliculata Jo~ao Neiva*†, Jorge Assis†, Francisco Fernandes, Gareth A. Pearson and Ester A. Serr~ao

Centro de Ci^encias do Mar, Centro de Investigacß~ao Marinha e Ambiental – Laboratorio Associado, Universidade do Algarve, Gambelas, Faro 8005-139, Portugal

ABSTRACT

Aim Species distributions have been continuously adjusting to changing climatic conditions throughout the glacial–interglacial cycles. In the marine realm, evidence suggests that latitudinal range shifts, involving both spatial expansions and trailingedge contractions, may represent a common response to climatic oscillations. The biogeographical histories of coastal organisms, however, have been inferred primarily using molecular markers, potentially overlooking past range dynamics beyond contemporary rear edges. In this study we combined species distribution models (SDMs) and mitochondrial DNA (mtDNA) data to investigate the biogeographical history of the high-intertidal seaweed Pelvetia canaliculata. We investigated the hypotheses that its distribution is set by both marine and terrestrial climates and that its range has shifted northwards since the Last Glacial Maximum. Location North-eastern Atlantic intertidal from Portugal to Norway. Methods In total, 432 individuals at 27 sites covering the extant range of Pelvetia canaliculata were sampled and sequenced for a c. 500 bp mtDNA intergenic spacer. A niche model was developed using marine and terrestrial variables. Range dynamics were reconstructed based on the geographical patterns of genetic variation and on the SDM projections for the Last Glacial Maximum (LGM) and the present. Results The best distribution models incorporated both marine and terrestrial variables. LGM projections revealed suitable habitat between southern Morocco and the periglacial shorelines of the Celtic Sea. Pelvetia canaliculata exhibited a highly structured phylogeography, being subdivided into three largely disjunct lineages, two of them endemic to Iberia. The central/northern European lineage exhibited the highest haplotypic diversity and showed a consistent decline in nucleotide diversity and haplotypic richness at higher latitudes.

*Correspondence: Jo~ao Neiva, Centro de Ci^encias do Mar, Universidade do Algarve, Gambelas, 8005-139 Faro, Portugal. E-mail: [email protected] †These authors contributed equally to this work.

ª 2014 John Wiley & Sons Ltd

Main conclusions Assuming species/climate equilibrium, SDMs supported the hypothesis of a post-glacial latitudinal range shift. Molecular variation revealed contrasting demographic behaviours in Iberian and periglacial regions. In Iberia the low haplotypic diversity suggested complex range dynamics that are not fully captured by SDM projections. Periglacial regions, identified as the source of poleward colonization, were inferred to have been comparatively more stable. Greater attention should be paid to marine range dynamics at low-latitude range margins, particularly in genetically structured low-dispersal species exhibiting southern endemic variation. Keywords Climatic refugia, Last Glacial Maximum, north-eastern Atlantic, Pelvetia canaliculata, phylogeography, range shift, species distribution model, trailing-edge.

http://wileyonlinelibrary.com/journal/jbi doi:10.1111/jbi.12278

1

J. Neiva et al.

INTRODUCTION The Pleistocene glacial–interglacial cycles have periodically imposed significant changes on the spatial distribution of biomes, ecosystems and species (Hofreiter & Stewart, 2009). During glacial peaks, such as the Last Glacial Maximum (LGM, c. 20 ka), temperate species commonly became restricted to southern (and cryptic periglacial) glacial refugia (Hewitt, 1999; Provan & Bennett, 2008). Climate amelioration eventually allowed species to expand their ranges poleward, often associated with the erosion (Hampe & Petit, 2005) or migration (e.g. Davis & Shaw, 2001) of southern range margins. Although the impact of glacial advances and the role of glacial refugia in poleward (re)colonizations are well documented, range dynamics at the southern distributional edges have been largely ignored and remain insufficiently understood (Stewart & Dalen, 2008; Stewart et al., 2010; Kettle et al., 2011). The glacial–interglacial cycles also exerted a strong and direct influence on the distribution of coastal species. In the north-western Atlantic, the compression of isotherms and the loss of ice-free rocky intertidal substrate led to the regional extirpation of many taxa during the LGM (Ilves et al., 2010; Waltari & Hickerson, 2012). In the north-eastern Atlantic, the ice sheets reached as far south as the British Isles and the dramatic drop in sea level resulted in the complete emersion of several shallow seas and the English Channel (Menot et al., 2006). Consequently, the littoral communities currently found in these areas could only colonize them after the LGM, originating from periglacial or more southern refugia (Provan et al., 2005; Hoarau et al., 2007; Remerie et al., 2009; Neiva et al., 2012a; Provan & Maggs, 2012). As most marine organisms leave a poor fossil record, their range dynamics have been inferred primarily using molecular data. However, because genes can only be sampled from extant populations, even geographically comprehensive, multi-locus data-sets will overlook past range dynamics beyond contemporary rear-edges. This inherent limitation of molecular markers is particularly relevant for coastal species because mounting evidence suggests similarly responsive leading and trailing edges (Sunday et al., 2012). Indeed, important southern range contractions attributable to post-glacial (Kettle et al., 2011) and ongoing climate warming (Perry et al., 2005; Wernberg et al., 2011; Nicastro et al., 2013) have been documented for both vagile and low-dispersal coastal species. Species distribution models (SDM) are a useful approach to generate spatially explicit information that can be used to validate and/or complement biogeographical inferences based on molecular markers. SDMs build a mathematical representation of the realized niche of species based on their known distribution in environmental space, allowing predictions of the potential species distributional range across time periods for which empirical or modelled environmental data are available (Elith & Leathwick, 2009). A variety of terrestrial studies have successfully combined phylogeographical and hindcasting SDM data to investigate climate-driven range dynamics

2

(Waltari et al., 2007; Carnaval et al., 2009; Schorr et al., 2012), but only very recently have the advantages of such integrative approaches been applied in the marine realm (Bigg et al., 2008; Provan & Maggs, 2012; Waltari & Hickerson, 2012). Intertidal seaweeds are good models to investigate marine range shifts. Their distributions are climatically defined (Breeman, 1990; L€ uning, 1990), they have particularly good distributional records, and they commonly display strong phylogeographical structure that can be linked to their past range dynamics (Hoarau et al., 2007; Fraser et al., 2009; Coyer et al., 2010; Neiva et al., 2012a). Seaweed distributions have traditionally been linked to ocean sea surface temperature (SST) isotherms, with range boundaries set by lethal or sublethal conditions limiting growth, reproduction or survival (van den Hoek, 1982; Breeman, 1988; L€ uning, 1990). For intertidal species subjected to daily immersion–emersion cycles, the terrestrial climate may play an additional role (Helmuth et al., 2002; Martınez et al., 2012), but this possibility has seldom been tested. In this paper we use an intertidal fucoid seaweed to investigate climate-driven biogeographical shifts along the north-eastern Atlantic coasts by integrating SDM and phylogeographical approaches. Pelvetia canaliculata (L.) Decaisne & Thuret (channelled wrack) is a perennial, hermaphroditic fucoid that forms, together with several other kelp/fucoid seaweeds, an endemic structural assemblage that typifies the cold-temperate shores of the north-eastern Atlantic (L€ uning, 1990). It lives in the high intertidal, on sheltered to semi-exposed rocky shores. It forms a complex symbiotic relationship with the fungus Mycophycias ascophylli, which is also present in Ascophyllum nodosum (Toxopeus et al., 2011). This species is hypothesized to have experienced extensive range shifts during the glacial–interglacial cycles because its modern distribution [from northern Portugal (40º N) to European Russia (71º N)] encompasses both ice-free and previously glaciated regions. In this study we develop a SDM for P. canaliculata based on marine and terrestrial climatic variables and transfer it to the LGM. We compare range projections with the historical biogeography of P. canaliculata inferred from mitochondrial (mtDNA) data. MtDNA has been widely used in fucoid phylogeography (Hoarau et al., 2007; Coyer et al., 2010), and a few studies show its general concordance with nuclear (microsatellite) data (Neiva et al., 2012a,b). We hypothesize that the range of P. canaliculata (1) is set by both marine and terrestrial climate variables, and (2) has globally migrated northwards since the LGM. We discuss the demographic and genetic implications of range shifts for low-dispersal marine species.

MATERIALS AND METHODS Occurrence and environmental data Predictive distribution maps for Pelvetia canaliculata were developed using a transferable model trained with occurrence

Journal of Biogeography ª 2014 John Wiley & Sons Ltd

Biogeographical shift in Pelvetia canaliculata data in relation to current environmental conditions, and projected onto distinct climate scenarios. Presence data were collated from several sources, including field observations, research articles, reports and online databases (see Appendix S1 in Supporting Information). Records were only considered when geographical locations were unequivocal and explicit down to shore/beach level. A variable number of geographically close occurrences were removed to ensure a similar record density throughout the species’ range, thus reducing the effects of spatial autocorrelation (Beatty & Provan, 2011). Records were gridded (0.25º resolution cells) and for modelling purposes those found within the same cell were only considered once. Current climatic data were obtained from the National Oceanic and Atmospheric Administration (NOAA/OAR/ ESRL PSD, Boulder, CO, USA, http://www.esrl.noaa.gov/psd/), and summarized in several long-term (1980–2010) monthly mean metrics describing the seasonal range variation of salinity, sea surface and air temperatures, and relative air humidity (see Appendix S2). Models included also tidal amplitude and intertidal availability. We define intertidal availability as the coastal area between hydrographic zero and the local maximal tidal level. Climatic data for the LGM were derived from two atmospheric circulation models (ACM): CCSM4 and MIROC5 (Earth System Grid Federation, http://www.esgf.org). These cover the range of climatic uncertainty of available models (Ramstein et al., 2007) and have been widely implemented in SDM (Schorr et al., 2012; Gassert et al., 2013). Intertidal availability was recalculated for 120 m depth. All predictors were gridded by bilinear interpolation to match the resolution of the distribution data. Transferable distribution models and predictive maps SDMs were developed using three machine-learning algorithms known for high predictive performance: boosted regression trees (BRT; De’ath, 2007), maximum entropy (Maxent; Phillips et al., 2006), and multivariate adaptive regression splines (MARS; Leathwick et al., 2005). While Maxent uses presence-only data, BRT and MARS use both species presence and absence. For these, pseudo-absences (PAs) were computed by randomly selecting cells at least two degrees from any presence (as many PAs as presences for BRT and 100 PAs for MARS; the ‘2º far’ method of BarbetMassin et al., 2012). Multiple models were performed with 70% of the distribution records (i.e. training data) in relation to all possible combinations of non-correlated climate predictors (Pearson’s correlation between pairs R2 < 0.75). BRT models were built using a tree complexity of 5, a bag fraction of 0.5, the optimal number of trees was determined by 10-fold cross validation on the training data, and the learning rate was set as the fastest reaching > 1000 trees (Elith et al., 2008; Millar & Blouin-Demers, 2012). Maxent 3 was implemented in R Journal of Biogeography ª 2014 John Wiley & Sons Ltd

using the package dismo (R Development Core Team, 2012). For Maxent and MARS the default parameter settings were used as implemented in R, an approach known to retrieve consistently good results (Phillips & Dudık, 2008; Bedia et al., 2011). For each model a predictive map was generated and reclassified into a binary presence–absence surface based on a threshold that maximized the sum of sensitivity and specificity (Phillips et al., 2006; Thuiller et al., 2009). Models were evaluated in terms of discriminatory power (i.e. the capacity to differentiate presences from absences) against testing data by the area under the receiver operating characteristic curve (AUC; Allouche et al., 2006). This iterative approach was performed 50 times per method, regenerating pseudoabsences at each step and randomizing training and testing data. The subsets of climate predictors with the highest predictive accuracy for each modelling method were selected as those with the highest mean AUC, using paired Wilcoxon tests between the mean AUC values of all combinations of predictors (Guisan et al., 2007) under H0: no differences in discriminatory power between subset of predictors. Modern and LGM distributions were produced as weighted averages of single-models trained with the full set of distribution records in relation to the most discriminatory subset of predictors. These were projected onto remote sensing and ACM climate data. The weights of each modelling method were based on the discriminatory power as measured by the mean AUC (i.e. weighted average ensemble as defined by Marmion et al., 2009). The potential for model transferability to the ACMs was inferred by true skill statistic (TSS) comparing the modern distribution predicted with data from MIROC5 and CCSM4 (averaged 1980–2010) against the records of occurrence and pseudo-absences. The ensemble for the LGM was produced by merging the outcomes from both CCSM4 and MIROC5. Past predictions lack the support of distribution records so no threshold was implemented on the predictive maps. These were reclassified as probability of occurrence. Ice sheets were reconstructed following Peltier (1994). All analyses were performed using the packages gbm, dismo, mda, SDMTools, raster and biomod for R (R Development Core Team, 2012). Sampling and molecular analyses Populations of Pelvetia canaliculata (n = 27) were sampled from Portugal (41º N) to Norway (67º N; Table 1), covering most of its distributional range. At each site, 3–5 cm apices were excised from 16 individuals sampled along a 50–200 m transect and stored in silica gel. Genomic DNA was extracted from approximately 8 mg of dried tissue using the Nucleospin Multi-96 plant kit (Macherey-Nagel, D€ uren, Germany). A ‘universal’ fucoid/kelp primer pair targeting the 23S/ trnK intergenic spacer (mtIGS) was developed based on a partial alignment of mtDNA genome of Fucus vesiculosus (Secq et al., 2006; GenBank accession AY494079), Laminaria 3

4

Iberia Viana do Castelo, PT Samieira (Ria de Pontevedra), ES Lires, ES A Coru~ na (Ria de A Coru~ na), ES Porcia, ES Lastres, ES Noja, ES Txatxarramendi/Ea, ES Brittany, English Channel and southern Ireland Piriac-sur-Mer, FR Le Cabellou, FR Roscoff, FR Saint-Briac-sur-Mer, FR Plymouth, UK St. Ives, UK Dunworley, IE western/northern Ireland, Wales and Scotland Gleninagh Quay (Galway Bay), IE Ballyhoorisky, IE Rhosneigr, UK Easdale (Firth of Lorn), UK Orkneys, UK Aberdeen, UK Tholen, NL Northern countries Faroe Islands, DK Mildevegen, NO Hylla (Trondheimsfjord), NO Fauske (Saltfjord), NO Herdisarvik, IS

Geographical region Population, country

53º08′ 55º15′ 53º09′ 56º17′ 58º57′ 57º08′ 51º36′ 61º55′ 60º14′ 63º50′ 67º15′ 63º51′

23 24 25 26 27

N, N, N, N, N,

N, N, N, N, N, N, N,

W W W W W W E

6º55′ W 4º59′ W 11º25′ E 15º20′ E 21º46′ W

9º13′ 7º45′ 4º17′ 5º39′ 3º15′ 2º05′ 4º06′

W W W W W W W

16 17 18 19 20 21 22

2º33′ 3º55′ 3º59′ 2º08′ 4º10′ 5º29′ 8º46′

47º23′ 47º51′ 48º44′ 48º38′ 50º21′ 50º13′ 51º35′

9 10 11 12 13 14 15 N, N, N, N, N, N, N,

41º42′ N, 8º51′ W 42º25′ N, 8º44′ W 43º00′ N, 9º15′ W 43º22′ N, 8º23′ W 43º34′ N, 6º53′ W 43º30′ N, 5º16′ W 43º29′N, 3º32′ W 43º23′ N, 2º35′ W

Lat, Long

1 2 3 4 5 6 7 8

Map Code 128 16 16 16 16 16 16 16 16 110 15 16 16 15 16 16 16 96 16 16 16 16 16 16 16 80 16 16 15 16 16

n A,B A A A A A A A,B A,B C C C C C C C C C C C C C C C C C C C C C C

Clades

C13(6), C17(10) C1(14), C14, C15 C1(15) C1(12), C16(3), C17 C1(16)

C1(16) C1(3), C24(8), C25(2), C26, C27, C28 C1(16) C1(11), C29(3), C31(2) C17(16) C12, C17(12), C18, C20(2) C10(16)

C2(2), C18(12), C19 C1(15), C3 C4(9), C5(2), C6, C7, C8, C29, C30 C1(2), C21(11), C22, C23 C1(16) C9(16) C1(14), C11(2)

A2(16) A2(16) A1(13), A3(2), A4 A1(16) A1(16) A1(16) A1(3), A6(4), B1(9) A1, B1(3), B2(8), B3(4)

Haplotypes 8 1 1 3 1 1 1 3 4 17 3 2 7 4 1 1 2 12 1 6 1 3 1 4 1 6 2 3 1 3 1

Nhap

670 – – 342 – – – 625 692 773 362 125 692 467 – – 233 683 – 733 – 508 – 442 – 458 500 242 – 425 –

H (10-3)

238 – – 72 – – – 293 194 233 147 19 271 127 – – 35 154 – 184 – 103 – 90 – 76 149 37 – 67 –

p (10-5)

Table 1 Pelvetia canaliculata 23S/trnK intergenic spacer (mtIGS) haplotype frequencies, number of haplotypes, haplotype diversity (H) and nucleotide diversity (p) for each geographical region and sampling site. PT, Portugal; ES, Spain; FR, France; IE, Ireland; UK, United Kingdom; NL, The Netherlands; DK, Denmark; NO, Norway; IS, Iceland.

J. Neiva et al.

Journal of Biogeography ª 2014 John Wiley & Sons Ltd

Biogeographical shift in Pelvetia canaliculata digitata (Secq et al., 2002; AJ344328) and Saccharina spp. (Yotsukura et al., 2010; AP011493–AP011499). Primers (F: 5′-TGGGTAGTTTGACTGGGGCGGT-3′, R: 5′-ACGGTT CCAATACCCACACCTGC-3′) were designed in conserved flanking regions to amplify a large, 1500 bp fragment spanning the mtIGS. After sequencing, new primers specific to the mtIGS of P. canaliculata were designed (F 5′-GGAGGTG CAAGAGCTGCAAGGT-3′; R 5′-TCGAACTCCCGTCTTCGT GCTT-3′). Polymerase chain reactions (PCRs) were performed in a 20 lL volume containing 19 GoTaq Flexi buffer (Promega, Madison, WI, USA), 2.0 mm MgCl2, 125 lm each dNTP, 0.5 lm each primer, 1 U GoTaq Flexi DNA Polymerase (Promega) and 2 lL of 1:100 diluted DNA. An initial denaturation step (94 °C, 5 min) was followed by 35 cycles of 94 °C for 30 s, 60 °C for 30 s and 72 °C for 1 min, and a final extension step (72 °C, 10 min). MtIGS amplicons were cleaned with ExoSap (Fermentas, Waltham, MA, USA) and sequenced in an automated capillary sequencer (Applied Biosystems, Carlsbad, CA, USA) at the Centro de Ci^encias do Mar (CCMAR Portugal). The geographical distribution of haplotypes was mapped and their relationships were inferred using the medianjoining algorithm implemented in Network 4.6 (Bandelt et al., 1999). Haplotype (Hhap) and nucleotide diversities (phap) within populations and selected geographical regions were calculated with DnaSP 5 (Librado & Rozas, 2009). RESULTS Species distribution model for Pelvetia canaliculata The modelling methods used to predict the distribution of P. canaliculata had mixed performances, with mean AUC values ranging from acceptable to outstanding (mean AUC range from 0.7 to 0.981). BRT had the highest discriminatory power [AUC (mean  SD) = 0.900  0.071; AUCmax = 0.981)], followed by Maxent (AUC = 0.848  0.058; AUCmax = 0.973) and MARS (mean AUC = 0.815  0.069; AUCmax = 0.953). In all methods, the predictors with more explanatory value were maximum summer air temperature, minimum winter sea surface temperature, maximum summer sea surface temperature and tidal coefficient (AUC = 0.948  0.019, 0.897  0.009 and 0.883  0.011 for BRT, Maxent and MARS, respectively). These four predictors were considered in all models for Ensemble purposes. The comparison of actual occurrences with the distribution predicted with data from both CCM4 and MIROC5 circulation models revealed excellent transferability (CCSM4TSS = 0.874; MIROC5TSS = 0.891).

(a)

20°W

0°

20°E

40°E 70°N

1

0.75

0.5

0.25

0

50°N

PalaeoCeltic Sea

30°N

30°N

(b) 70°N

50°N

Cantabrian Sea

30°N

30°N 20°W

0°

20°E

40°E

Figure 1 Range projections for Pelvetia canaliculata during (a) the Last Glacial Maximum and (b) the present. Coloured areas along the shorelines indicate the predicted probability of occurrence. In panel (a) the maximum extension of ice sheets is depicted in white and the shoreline was modelled to a lower sea level ( 120 m).

regions such as the palaeo-shorelines of the Celtic Sea and western Ireland (Fig. 1a). At the southern range margin, during full glacial conditions it could potentially have extended far beyond its current limit. Suitable habitat would have been present throughout Portugal, the Moroccan Atlantic coast and the Canary and Madeira archipelagos. Conditions might have been similarly favourable in the western Mediterranean, although high probabilities of occurrence were only supported for the regions with higher tidal coefficients. Persistence regions, with high probabilities of occurrence both for the present day and the LGM, ranged from north-western Iberia and the moving shorelines of the Celtic Sea, with the exception of the Cantabrian Sea, where the probability of occurrence was spatially variable across the two periods.

Hindcast distributions

Mitochondrial DNA phylogeography

The Ensemble hindcast supported a significant change in the potential distribution of P. canaliculata between the LGM and today (Fig. 1). As expected, LGM projections revealed that it could only have been present in marginal periglacial

In total, 39 mtIGS haplotypes (GenBank accession numbers KC143772–KC143810) were identified in the 429 individuals of Pelvetia canaliculata. The median-joining network revealed three major mtIGS lineages displaying well-defined,

Journal of Biogeography ª 2014 John Wiley & Sons Ltd

5

J. Neiva et al. (b)

(a)

Figure 3 Nucleotide diversity (p, columns) and number of haplotypes (stars) of Pelvetia canaliculata within four selected geographical regions.

Figure 2 Genealogy and distribution of 23S/trnK intergenic spacer (mtIGS) haplotypes throughout the range of Pelvetia canaliculata. (a) Contemporary range in the north-eastern Atlantic (black shoreline) and location of sampling sites (white circles). Pie charts depict haplotype frequencies at each site (see Table 1 for location and haplotype IDs). Haplotypes are coloured as in (b). The black arrow on the low left corner marks the historical southern limit by 1990. (b) Parsimony network of mtIGS haplotypes. Sampled haplotypes are represented by circles sized to their global frequency. Links represent a single nucleotide change, black dots represent inferred, unsampled haplotypes, and black squares represent small indels. Shared haplotypes (A1, A2, B1, C1, C17, C18 and C29) are depicted in strong colours, whereas private haplotypes are represented by paler versions of the colours of their closest shared haplotype. Two haplotype radiations within lineage C are also depicted in distinct shades of green.

largely disjunct geographical distributions (Table 1, Fig. 2). Phylogroup A, composed of haplotypes A1 and four related ones, was present throughout Iberia. A second Iberian phylogroup, B, composed of B1 and two related haplotypes, was restricted to the easternmost part of the Cantabrian Sea. Phylogroup C, composed of C1 and 30 related haplotypes, was distributed along central and northern European shores, from north-western France to Norway and Iceland. Phylogroups were defined by one interior and widespread (within its phylogroup range) haplotype (A1, B1 and C1), although a few derived haplotypes dominated in restricted geographical regions (e.g. A2 in W Iberia and C17 in Scotland/Faroes) or single populations (e.g. C9 in St. Ives and C10 in the Netherlands). Distinct lineages were only found mixed in eastern Cantabria (Noja and Txatxarramendi, lineages A and B). Lineage A was interior in the network, with similar divergences from B and C. Only the three interior haplotypes (A1, B1 and C1) plus four derived ones (A2, C17, C18 and 6

C29) were shared among at least two populations. The remaining 32 haplotypes were population-specific. Haplotypic diversity was commonly very low within populations (Table 1); 12 (44%) were fixed for a single haplotype, and in only two (Roscoff and Ballyhoorisky) were more than five haplotypes detected. These had the highest haplotypic diversities, but Noja exhibited the highest nucleotide diversity, as it was composed of two distinct lineages (A and B) with similar frequencies. Haplotypic but not nucleotide diversity was lower in Iberia (lineages A and B) than in Central/Northern Europe (lineage C). Within the former, nucleotide diversity was highest in the Brittany/Channel region and lowest in the Nordic Countries (Table 1, Fig. 3). DISCUSSION Species distribution model analyses Our niche model revealed that the large-scale distribution of Pelvetia canaliculata is set by both marine and terrestrial climatic conditions, particularly those related to temperature. Its southern distributional limit in northern Portugal is shared with many intertidal/shallow subtidal cold-temperate seaweeds (Lima et al., 2007). In north-western Iberia, where these seaweeds are ubiquitous, summer SSTs are atypically low relative to central Portugal or the inner Cantabrian Sea, as a result of persistent (and often intense) upwelling. Pelvetia canaliculata, however, extends its distribution into the inner Cantabrian Sea, where summer SSTs can reach 20 °C (Dıez et al., 2012). In central Portugal, where P. canaliculata is currently absent, mean summer SSTs are slightly lower, but a Mediterranean terrestrial climate prevails. Pelvetia canaliculata was present at Berlenga Island (off central Portugal) at least until 1992 (see Appendix S3), and its c. 240 km northwards retreat may be related to the SST warming trend observed in this already marginal (and mostly sandy) shoreline over the past three decades (Lima et al., 2007; Nicastro et al., 2013). Journal of Biogeography ª 2014 John Wiley & Sons Ltd

Biogeographical shift in Pelvetia canaliculata Assuming species/climate equilibrium, the LGM projections imply a post-glacial expansion from the palaeo-shorelines of the Celtic Sea into Iceland, northern Norway and the White Sea (c. 18º N). They also imply a marked range compression at its southern trailing edge, from Morocco and possibly the Mediterranean to its present rear edge in northern Portugal (c. 14º N). Its actual occurrence (and later extirpation) south of Portugal cannot be demonstrated without direct fossil evidence, but is in line with archaeological evidence of coldwater fish in the western Mediterranean around the LGM (Cortes-Sanchez et al., 2008; Kettle et al., 2011). SDMs similarly hindcast suitable habitat far beyond modern rear-edges in a diverse range of cold-temperate organisms (Provan & Maggs, 2012; Waltari & Hickerson, 2012). Smaller range shifts attributable to (so far much milder) global warming have already been documented (Wernberg et al., 2011; Nicastro et al., 2013) or forecast (Jueterbock et al., 2013) for other seaweeds. Although fucoid propagules disperse only locally (Pearson & Serrao, 2006), rafting of detached fertile fronds occurs (Norton, 1992; Kinlan & Gaines, 2003; Neiva et al., 2012b). The kelp/fucoids present in Norway and Iceland confirm that many cold-temperate seaweeds (including P. canaliculata) have successfully tracked their habitats as the ice sheets retreated. Glacials typically last much longer than interglacials (the last glacial lasted between c. 120 and 11.5 ka), so it would be expected that P. canaliculata would have had time to equilibrate with its glacial environment (i.e. colonize north-western Africa). If so, its range suffered a latitudinal shift instead of a global contraction(s)/expansion(s) into/ from geographically restricted region(s). Conceptually, rangeshifts differ from classical contraction/expansion models in several aspects (Fig. 4; see also Bennett & Provan, 2008; Stewart et al., 2010). Range shifts require southern and northern range margins acting as mismatched leading and (a)

(b)

Figure 4 Schematic representation of hypothetical range changes and refugia across a glacial–interglacial cycle in (a) the classical contraction/expansion model, and (b) latitudinal range shift. Journal of Biogeography ª 2014 John Wiley & Sons Ltd

trailing edges during warming and cooling periods. They do not entail dramatic changes in range size, although these can occur if climatic envelopes become too compressed. Finally, genetic diversity (as a proxy for long-term persistence) is expected to be higher where glacial and interglacial distributions overlap (here between Iberia and Ireland), instead of corresponding to species maximum range contractions as in common definitions of terrestrial glacial refugia (Hewitt, 2000). Mitochondrial DNA phylogeography The existence of distinct Iberian and Central/Northern mtDNA phylogroups, physically separated by the large distributional gap in the south-western French coast, represents the most striking feature of P. canaliculata phylogeography. The two endemic mtDNA lineages indicate that Iberia has provided a long-term climatic refugium, as unique genetic variation was able to evolve and persist there. This region, however, had a very limited (if any) contribution to the post-glacial colonization of Northern Europe. The refugial status of Iberia is puzzling, however, when considering the region’s low haplotypic diversity. Phylogroup A is far less diverse than phylogroup C, and than other endemic Iberian lineages of fucoid species with similar cold-temperate distributions (Neiva et al., 2012a). A hypothetical population bottleneck in Iberia, that would explain its current low regional genetic diversity, cannot be attributed to glacial periods. Indeed, lineage C clearly persisted at much colder periglacial latitudes. SDMs projected an extended distribution of P. canaliculata throughout northern Africa suggesting that Iberia was a central climatic optimum. A more plausible hypothesis is that Iberian P. canaliculata regressed during warm periods. Some controversy remains as to whether or not the summer air temperatures during the Holocene Climatic Optimum (c. 9–5 ka) were warmer than today in south-western Europe (Davis et al., 2003; Brewer et al., 2007; Wanner et al., 2008). Archaeological evidence in northern Spain is consistent with warmer SST during the mid-Holocene (Clark, 1971; Bailey & Craighead, 2003; Marın et al., 2011). Unstable range dynamics attributable to contemporary decadal-scale meteorological events may offer an alternative explanation. Several structural seaweeds have been experiencing regional range shifts along the Cantabrian Sea in the last century, including range expansions and contractions from/into colder north-western Iberia (Arrontes, 2002; Fernandez & Anad on, 2008; Fernandez, 2011). Similar contraction/expansion cycles have not been documented in P. canaliculata in modern times, but may have occurred during earlier periods. If so, they would certainly contribute to the observed genetic depletion in this region, as recurrent extinctions and leading-edge colonizations would eliminate most of its regional genetic diversity. Projecting the range of P. canaliculata for additional points in time could potentially improve the resolution of its spatial dynamics in Iberia and beyond (see Espındola et al., 2012). 7

J. Neiva et al. In sharp contrast with Iberian phylogroups, phylogroup C was both widespread and diverse. Its vast distribution indicates this lineage was the source of the post-glacial poleward colonization. A clear decrease of haplotypic richness and nucleotide diversity (but not haplotypic diversity) was observed between the Brittany/Channel region and the Nordic countries. The former is a diversity hotspot in several other seaweeds, reflecting its proximity to periglacial shorelines (Provan et al., 2005; Hoarau et al., 2007; Olsen et al., 2010). The large number of private haplotypes and even small haplotype radiations found around Brittany suggest that in the long term this region had larger and/or more stable populations than Iberia. The distribution of the most common haplotypes (C1 and C17) suggests an expansion proceeding along two distinct routes, as in other seaweeds (Provan et al., 2005; Hoarau et al., 2007; Neiva et al., 2012b). The first wave migrated westwards and northwards following the submersion of the English Channel and the Celtic and Irish seas, allowing the central haplotype C1 to spread along southern English shores, Ireland and western Scotland. After the re-establishment of the passage between the Channel and the North Sea (c. 7.5 ka), a second wave expanded along the east coast of England, where the derived haplotype C17 eventually spread further north along eastern Scotland. The first wave apparently arrived earlier in Norway, as this region is dominated by haplotype C1. The haplotypes found in Iceland are consistent with colonization via Norway or Scotland, and not via the shortest (and thus expected) route provided by the Faeroes archipelago. Range shifts and phylogeographical structure Pelvetia canaliculata is subdivided into well-defined phylogroups whose contribution to the global range/niche occupancy differs markedly. Phylogroup B is restricted to a short stretch of coastline whereas phylogroup C occupies the entire range above 45º N. These differences, under strict niche conservatism, are assumed to reflect primarily the effects of demographic history acting on pre-existing patterns of lineage distribution and divergence. However, the distinct evolutionary histories and selective regimes experienced by lineages occupying different geographical/environmental spaces are likely to promote regional accumulation of unique and potentially adaptive genetic variability (Hampe & Petit, 2005; Pearman et al., 2010). Acknowledging the existence of multiple phylogeographical lineages is especially important in the context of climateinduced range shifts. The demographic behaviours and fate of regional populations of P. canaliculata at the glacial–interglacial transition were quite diverse. Phylogroup C survived along peri-glacial shorelines and experienced an extensive post-glacial expansion whereas southern lineages were inferred to have suffered a global range reduction. Even if extinctions at the trailing edge are counteracted by the establishment of populations at the leading edge, the spatial dynamics of lineages may result in significant losses of 8

genetic diversity (see Dalen et al., 2007; Nicastro et al., 2013). Perhaps the biggest limitation of species-based SDMs such as the one developed here is that they do not take into account these historical components of diversity and their potential importance for the overall functional diversity (and thus evolutionary potential) of species (Jump & Pe~ nuelas, 2005). The fate of contracting lineages may be further complicated by density-barrier effects. Pelvetia canaliculata is able to disperse non-locally and to keep track of its habitat. Yet, the numerous phylogeographical breaks throughout its range show that, despite dispersal, gene-flow is negligible even at relatively small spatial scales. Contrasting outcomes of dispersal into newly available (colonization) and into occupied (immigration) habitats suggest important density-barrier effects (De Meester et al., 2002). These have been invoked to explain the remarkable persistence of genetic breaks in other structural seaweed species (Fraser et al., 2009; Neiva et al., 2012a). If so, the unique Iberian lineages may be trapped between increasingly unsuitable conditions at the warming trailing-edge (see Jueterbock et al., 2013) and more northern regions where lineage C has long been established. CONCLUSIONS Our results suggest that Pelvetia canaliculata has experienced a substantial post-glacial latitudinal range shift, and highlight how behaviours of different subpopulations facing climatic change can be diverse and independent. Contracting trailing edges and latitudinal shifts have been documented or predicted for many coastal species and may prove to be widespread in the marine realm. Pelvetia canaliculata harbours unique genetic variation at its southern limit that may be at risk. Marginal populations may be extirpated if conditions become increasingly unsuitable, and the fate of contracting lineages may be further complicated by density-barrier effects. These concerns may apply to other highly-structured lowdispersal species. Given the paucity of marine fossil records, palaeodistributional modelling should be integrated more often with molecular information, particularly to reconstruct glacial distributions at and beyond modern trailing edges. ACKNOWLEDGEMENTS We thank the editor and anonymous referees for useful suggestions; Agnes Mols Mortensen, Alice Neiva, Ana Lara, Daniel Brazier, Duarte Neiva, Herre Stegenga, Karl Gunnarson, Mahaut de Vareilles, Martin Atrill, Michael Burrows, Paul Brazier, Richard Ticehurst, Rui Catarino and T^ania Pereira for kindly providing samples; and Marta Valente (CCMAR) for the sequencing work. This work was funded by FCT (Fundacß~ao para a Ci^encia e a Tecnologia) through projects EDGES (PTDC/AAC-CLI/109108/2008) and EXTANT (EXCL/AAG-GLO/0661/2012) and fellowships to J.A. (doctoral, SFRH/BD/65702/2009) and J.N. (postdoctoral, SFRH/BPD/88935/2012). Journal of Biogeography ª 2014 John Wiley & Sons Ltd

Biogeographical shift in Pelvetia canaliculata REFERENCES Allouche, O., Tsoar, A. & Kadmon, R. (2006) Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). Journal of Applied Ecology, 43, 1223–1232. Arrontes, J. (2002) Mechanisms of range expansion in the intertidal brown alga Fucus serratus in northern Spain. Marine Biology, 141, 1059–1067. Bailey, G.N. & Craighead, A.S. (2003) Late Pleistocene and Holocene coastal palaeoeconomies: a reconsideration of the molluscan evidence from northern Spain. Geoarchaeology – An International Journal, 18, 175–204. Bandelt, H.J., Forster, P. & Rohl, A. (1999) Median-joining networks for inferring intraspecific phylogenies. Molecular Biology and Evolution, 16, 37–48. Barbet-Massin, M., Jiguet, F., Albert, C.H. & Thuiller, W. (2012) Selecting pseudo-absences for species distribution models: how, where and how many? Methods in Ecology and Evolution, 3, 327–338. Beatty, G.E. & Provan, J. (2011) Comparative phylogeography of two related plant species with overlapping ranges in Europe, and the potential effects of climate change on their intraspecific genetic diversity. BMC Evolutionary Biology, 11, 29. Bedia, J., Busque, J. & Gutierrez, J.M. (2011) Predicting plant species distribution across an alpine rangeland in northern Spain. A comparison of probabilistic methods. Applied Vegetation Science, 14, 415–432. Bennett, K.D. & Provan, J. (2008) What do we mean by ‘refugia’? Quaternary Science Reviews, 27, 2449–2455. Bigg, G.R., Cunningham, C.W., Ottersen, G., Pogson, G.H., Wadley, M.R. & Williamson, P. (2008) Ice-age survival of Atlantic cod: agreement between palaeoecology models and genetics. Proceedings of the Royal Society B: Biological Sciences, 275, 163–U113. Breeman, A.M. (1988) Relative importance of temperature and other factors in determining geographic boundaries of seaweeds: experimental and phenological evidence. Helgolander Meeresuntersuchungen, 42, 199–241. Breeman, A.M. (1990) Expected effects of changing seawater temperatures on the geographic distribution of seaweed species. Expected effects of climatic change on marine coastal ecosystems (ed. by J.J. Beukeuma), pp. 69–76. Kluwer Academic Publishers, Dordrecht, The Netherlands. Brewer, S., Guiot, J. & Torre, F. (2007) Mid-Holocene climate change in Europe: a data-model comparison. Climate of the Past, 3, 499–512. Carnaval, A.C., Hickerson, M.J., Haddad, C.F.B., Rodrigues, M.T. & Moritz, C. (2009) Stability predicts genetic diversity in the Brazilian Atlantic forest hotspot. Science, 323, 785–789. Clark, G.A. (1971) The Asturian of Cantabria: subsistence base and evidence for post-Pleistocene climatic shifts. American Anthropologist, 73, 1244–000.

Journal of Biogeography ª 2014 John Wiley & Sons Ltd

Cortes-Sanchez, M., Morales-Mu~ niz, A., Sim on-Vallejo, M.D., Bergada-Zapata, M.M., Delgado-Huertas, A., L opez-Garcıa, P., L opez-Saez, J.A., Lozano-Francisco, M.C., RiquelmeCantal, J.A., Rosell o-Izquierdo, E., S onchez-Marco, A. & Vera-Pelaez, J.L. (2008) Palaeoenvironmental and cultural dynamics of the coast of Malaga (Andalusia, Spain) during the Upper Pleistocene and early Holocene. Quaternary Science Reviews, 27, 2176–2193. Coyer, J.A., Hoarau, G., Van Schaik, J., Luijckx, P. & Olsen, J.L. (2010) Trans-Pacific and trans-Arctic pathways of the intertidal macroalga Fucus distichus L. reveal multiple glacial refugia and colonizations from the North Pacific to the North Atlantic. Journal of Biogeography, 38, 756–771. Dalen, L., Nystr€ om, V., Valdiosera, C., Germonpre, M., Sablin, M., Turner, E., Angerbj€ orn, A., Arsuaga, J.L. & G€ otherstr€ om, A. (2007) Ancient DNA reveals lack of postglacial habitat tracking in the arctic fox. Proceedings of the National Academy of Sciences USA, 104, 6726–6729. Davis, B.A.S., Brewer, S., Stevenson, A.C. & Guiot, J. (2003) The temperature of Europe during the Holocene reconstructed from pollen data. Quaternary Science Reviews, 22, 1701–1716. Davis, M.B. & Shaw, R.G. (2001) Range shifts and adaptive responses to Quaternary climate change. Science, 292, 673–679. De Meester, L., G omez, A., Okamura, B. & Schwenk, K. (2002) The Monopolization Hypothesis and the dispersalgene flow paradox in aquatic organisms. Acta Oecologica, 23, 121–135. De’ath, G. (2007) Boosted trees for ecological modeling and prediction. Ecology, 88, 243–251. Dıez, I., Muguerza, N., Santolaria, A., Ganzedo, U. & Gorostiaga, J.M. (2012) Seaweed assemblage changes in the eastern Cantabrian Sea and their potential relationship to climate change. Estuarine Coastal and Shelf Science, 99, 108–120. Elith, J. & Leathwick, J.R. (2009) Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics, 40, 677–697. Elith, J., Leathwick, J.R. & Hastie, T. (2008) A working guide to boosted regression trees. Journal of Animal Ecology, 77, 802–813. Espındola, A., Pellissier, L., Maiorano, L., Hordijk, W., Guisan, A. & Alvarez, N. (2012) Predicting present and future intra-specific genetic structure through niche hindcasting across 24 millennia. Ecology Letters, 15, 649–657. Fernandez, C. (2011) The retreat of large brown seaweeds on the north coast of Spain: the case of Saccorhiza polyschides. European Journal of Phycology, 46, 352–360. Fernandez, C. & Anad on, R. (2008) La cornisa Cantabrica: un escenario de cambios de distribuci on de comunidades intermareales. Algas, 39, 30–32. Fraser, C.I., Nikula, R., Spencer, H.G. & Waters, J.M. (2009) Kelp genes reveal effects of subantarctic sea ice during the

9

J. Neiva et al. Last Glacial Maximum. Proceedings of the National Academy of Sciences USA, 106, 3249–3253. Gassert, F., Schulte, U., Husemann, M., Ulrich, W., Rodder, D., Hochkirch, A., Engel, E., Meyer, J. & Habel, J.C. (2013) From southern refugia to the northern range margin: genetic population structure of the common wall lizard, Podarcis muralis. Journal of Biogeography, 40, 1475–1489. Guisan, A., Graham, C.H., Elith, J. & Huettmann, F. (2007) Sensitivity of predictive species distribution models to change in grain size. Diversity and Distributions, 13, 332–340. Hampe, A. & Petit, R.J. (2005) Conserving biodiversity under climate change: the rear edge matters. Ecology Letters, 8, 461–467. Helmuth, B., Harley, C.D.G., Halpin, P.M., O’Donnell, M., Hofmann, G.E. & Blanchette, C.A. (2002) Climate change and latitudinal patterns of intertidal thermal stress. Science, 298, 1015–1017. Hewitt, G.M. (1999) Post-glacial re-colonization of European biota. Biological Journal of the Linnean Society, 68, 87–112. Hewitt, G. (2000) The genetic legacy of the Quaternary ice ages. Nature, 405, 907–913. Hoarau, G., Coyer, J.A., Veldsink, J.H., Stam, W.T. & Olsen, J.L. (2007) Glacial refugia and recolonization pathways in the brown seaweed Fucus serratus. Molecular Ecology, 16, 3606–3616. van den Hoek, C. (1982) The distribution of benthic marine algae in relation to the temperature regulation of their life histories. Biological Journal of the Linnean Society, 18, 81–144. Hofreiter, M. & Stewart, J. (2009) Ecological change, range fluctuations and population dynamics during the Pleistocene. Current Biology, 19, R584–R594. Ilves, K.L., Huang, W., Wares, J.P. & Hickerson, M.J. (2010) Colonization and/or mitochondrial selective sweeps across the North Atlantic intertidal assemblage revealed by multitaxa approximate Bayesian computation. Molecular Ecology, 19, 4505–4519. Jueterbock, A., Tyberghein, L., Verbruggen, H., Coyer, J.A., Olsen, J.L. & Hoarau, G. (2013) Climate change impact on seaweed meadow distribution in the North Atlantic rocky intertidal. Ecology and Evolution, 3, 1356–1373. Jump, A.S. & Pe~ nuelas, J. (2005) Running to stand still: adaptation and the response of plants to rapid climate change. Ecology Letters, 8, 1010–1020. Kettle, A.J., Morales-Mu~ niz, A., Rosell o-Izquierdo, E., Heinrich, D. & Vøllestad, L.A. (2011) Refugia of marine fish in the northeast Atlantic during the Last Glacial Maximum: concordant assessment from archaeozoology and palaeotemperature reconstructions. Climate of the Past, 7, 181–201. Kinlan, B.P. & Gaines, S.D. (2003) Propagule dispersal in marine and terrestrial environments: a community perspective. Ecology, 84, 2007–2020. Leathwick, J.R., Rowe, D., Richardson, J., Elith, J. & Hastie, T. (2005) Using multivariate adaptive regression splines to

10

predict the distributions of New Zealand’s freshwater diadromous fish. Freshwater Biology, 50, 2034–2052. Librado, P. & Rozas, J. (2009) DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics, 25, 1451–1452. Lima, F.P., Ribeiro, P.A., Queiroz, N., Hawkins, S.J. & Santos, A.M. (2007) Do distributional shifts of northern and southern species of algae match the warming pattern? Global Change Biology, 13, 2592–2604. L€ uning, K. (1990) Seaweeds: their environment, biogeography and ecophysiology. Wiley-Interscience, New York. Marın, A.B., Gonzalez-Morales, M.R. & Estevez, J. (2011) Paleoclimatic inference of the mid-Holocene record of monk seal (Monachus monachus) in the Cantabrian Coast. Proceedings of the Geologists Association, 122, 113–124. Marmion, M., Parviainen, M., Luoto, M., Heikkinen, R.K. & Thuiller, W. (2009) Evaluation of consensus methods in predictive species distribution modelling. Diversity and Distributions, 15, 59–69. Martınez, B., Viejo, R.M., Carre~ no, F. & Aranda, S.C. (2012) Habitat distribution models for intertidal seaweeds: responses to climatic and non-climatic drivers. Journal of Biogeography, 39, 1877–1890. Menot, G., Bard, E., Rostek, F., Weijers, J.W.H., Hopmans, E.C., Schouten, S. & Damste, J.S.S. (2006) Early reactivation of European rivers during the last deglaciation. Science, 313, 1623–1625. Millar, C.S. & Blouin-Demers, G. (2012) Habitat suitability modelling for species at risk is sensitive to algorithm and scale: a case study of Blanding’s turtle, Emydoidea blandingii, in Ontario, Canada. Journal for Nature Conservation, 20, 18–29. Neiva, J., Pearson, G.A., Valero, M. & Serr~ao, E.A. (2012a) Fine-scale genetic breaks driven by historical range dynamics and ongoing density-barrier effects in the estuarine seaweed Fucus ceranoides L. BMC Evolutionary Biology, 12, 78. Neiva, J., Pearson, G.A., Valero, M. & Serr~ao, E.A. (2012b) Drifting fronds and drifting alleles: range dynamics, local dispersal and habitat isolation shape the population structure of the estuarine seaweed Fucus ceranoides. Journal of Biogeography, 39, 1167–1178. Nicastro, K.R., Zardi, G.I., Teixeira, S., Neiva, J., Serr~ao, E.A. & Pearson, G.A. (2013) Shift happens: trailing edge contraction associated with recent warming trends threatens a distinct genetic lineage in the marine macroalga Fucus vesiculosus. BMC Biology, 11, 6. Norton, T.A. (1992) Dispersal by macroalgae. British Phycological Journal, 27, 293–301. Olsen, J.L., Zechman, F.W., Hoarau, G., Coyer, J.A., Stam, W.T., Valero, M. & Aberg, P. (2010) The phylogeographic architecture of the fucoid seaweed Ascophyllum nodosum: an intertidal ‘marine tree’ and survivor of more than one glacial–interglacial cycle. Journal of Biogeography, 37, 842– 856.

Journal of Biogeography ª 2014 John Wiley & Sons Ltd

Biogeographical shift in Pelvetia canaliculata Pearman, P.B., D’Amen, M., Graham, C.H., Thuiller, W. & Zimmermann, N.E. (2010) Within-taxon niche structure: niche conservatism, divergence and predicted effects of climate change. Ecography, 33, 990–1003. Pearson, G.A. & Serrao, E.A. (2006) Revisiting synchronous gamete release by fucoid algae in the intertidal zone: fertilization success and beyond? Integrative and Comparative Biology, 46, 587–597. Peltier, W.R. (1994) Ice Age paleotopography. Science, 265, 195–201. Perry, A.L., Low, P.J., Ellis, J.R. & Reynolds, J.D. (2005) Climate change and distribution shifts in marine fishes. Science, 308, 1912–1915. Phillips, S.J. & Dudık, M. (2008) Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography, 31, 161–175. Phillips, S.J., Anderson, R.P. & Schapire, R.E. (2006) Maximum entropy modeling of species geographic distributions. Ecological Modelling, 190, 231–259. Provan, J. & Bennett, K.D. (2008) Phylogeographic insights into cryptic glacial refugia. Trends in Ecology and Evolution, 23, 564–571. Provan, J. & Maggs, C.A. (2012) Unique genetic variation at a species’ rear edge is under threat from global climate change. Proceedings of the Royal Society B: Biological Sciences, 279, 39–47. Provan, J., Wattier, R.A. & Maggs, C.A. (2005) Phylogeographic analysis of the red seaweed Palmaria palmata reveals a Pleistocene marine glacial refugium in the English Channel. Molecular Ecology, 14, 793–803. R Development Core Team (2012) R: a language and environment for statistical computing. Austria, Vienna. Ramstein, G., Kageyama, M., Guiot, J., Wu, H., Hely, C., Krinner, G. & Brewer, S. (2007) How cold was Europe at the Last Glacial Maximum? A synthesis of the progress achieved since the first PMIP model-data comparison. Climate of the Past, 3, 331–339. Remerie, T., Vierstraete, A., Weekers, P.H.H., Vanfleteren, J.R. & Vanreusel, A. (2009) Phylogeography of an estuarine mysid, Neomysis integer (Crustacea, Mysida), along the north-east Atlantic coasts. Journal of Biogeography, 36, 39–54. Schorr, G., Holstein, N., Pearman, P.B., Guisan, A. & Kadereit, J.W. (2012) Integrating species distribution models (SDMs) and phylogeography for two species of Alpine Primula. Ecology and Evolution, 2, 1260–1277. Secq, M.P.O., Kloareg, B. & Loiseaux-de Goer, S. (2002) The mitochondrial genome of the brown alga Laminaria digitata: a comparative analysis. European Journal of Phycology, 37, 163–172. Secq, M.P.O., Goer, S.L., Stam, W.T. & Olsen, J.L. (2006) Complete mitochondrial genomes of the three brown algae (Heterokonta: Phaeophyceae) Dictyota dichotoma, Fucus vesiculosus and Desmarestia viridis. Current Genetics, 49, 47–58.

Journal of Biogeography ª 2014 John Wiley & Sons Ltd

Stewart, J.R. & Dalen, L. (2008) Is the glacial refugium concept relevant for northern species? A comment on Pruett and Winker 2005. Climatic Change, 86, 19–22. Stewart, J.R., Lister, A.M., Barnes, I. & Dalen, L. (2010) Refugia revisited: individualistic responses of species in space and time. Proceedings of the Royal Society B: Biological Sciences, 277, 661–671. Sunday, J.M., Bates, A.E. & Dulvy, N.K. (2012) Thermal tolerance and the global redistribution of animals. Nature Climate Change, 2, 686–690. Thuiller, W., Lafourcade, B., Engler, R. & Ara ujo, M.B. (2009) BIOMOD – a platform for ensemble forecasting of species distributions. Ecography, 32, 369–373. Toxopeus, J., Kozera, C.J., O’Leary, S.J.B. & Garbary, D.J. (2011) A reclassification of Mycophycias ascophylli (Ascomycota) based on nuclear large ribosomal subunit DNA sequences. Botanica Marina, 54, 325–334. Waltari, E. & Hickerson, M.J. (2012) Late Pleistocene species distribution modelling of North Atlantic intertidal invertebrates. Journal of Biogeography, 40, 249–260. Waltari, E., Hijmans, R.J., Peterson, A.T., Nyari, A.S., Perkins, S.L. & Guralnick, R.P. (2007) Locating Pleistocene refugia: comparing phylogeographic and ecological niche model predictions. PLoS ONE, 2, 563. Wanner, H., Beer, J., Buetikofer, J., Crowley, T.J., Cubasch, U., Flueckiger, J., Goosse, H., Grosjean, M., Joos, F., Kaplan, J.O., Kuettel, M., Mueller, S.A., Prentice, I.C., Solomina, O., Stocker, T.F., Tarasov, P., Wagner, M. & Widmann, M. (2008) Mid- to Late Holocene climate change: an overview. Quaternary Science Reviews, 27, 1791–1828. Wernberg, T., Russell, B.D., Thomsen, M.S., Gurgel, C.F.D., Bradshaw, C.J.A., Poloczanska, E.S. & Connell, S.D. (2011) Seaweed communities in retreat from ocean warming. Current Biology, 21, 1828–1832. Yotsukura, N., Shimizu, T., Katayama, T. & Druehl, L.D. (2010) Mitochondrial DNA sequence variation of four Saccharina species (Laminariales, Phaeophyceae) growing in Japan. Journal of Applied Phycology, 22, 243–251. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1 Mapped records of Pelvetia canaliculata used to build the niche models, and geographical locations and sources of the records used to model the niche for Pelvetia canaliculata. Appendix S2 Environmental predictors used for modelling the distribution of Pelvetia canaliculata. Appendix S3 Photographic record (1992) of now extinct Pelvetia canaliculata from Berlenga Island, central Portugal, and photographic record of a herbarium specimen of Pelvetia canaliculata collected as drift near Lisbon (19th century, Herbarium of Lisbon University).

11

J. Neiva et al. BIOSKETCHES ~o Neiva is a post-doctoral researcher at the CCMAR, University of Algarve. His research is focused on the historical biogeJoa ography and population genetic structure of temperate fucoid seaweeds. Jorge Assis is a graduate student at CCMAR. His research is focused on species distribution modelling, landscape ecology and patterns of population connectivity. Author contributions: J.N., J.A., G.A.P. and E.A.S. conceived the study. J.N. and F.F. collected samples and contacted collaborators abroad, performed the laboratory work and analysed the genetic data. J.A. performed the SDM analyses. All authors interpreted results. J.N. and J.A. led the writing, which was critically revised by G.A.P. and E.A.S. All authors read and approved the final version of this manuscript.

Editor: Christine Maggs

12

Journal of Biogeography ª 2014 John Wiley & Sons Ltd