Genome-wide assessment elucidates connectivity and the evolutionary history of the highly dispersive marine invertebrate Littoraria flava ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Biological Journal of the Linnean Society, 2021, 133, 999–1015. With 5 figures. Genome-wide assessment elucidates connectivity and the evolutionary history of the highly dispersive marine invertebrate Littoraria flava (Littorinidae: Gastropoda) THAINÁ CORTEZ1,*, , RAFAEL V. AMARAL1, THADEU SOBRAL-SOUZA2 and Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 SÓNIA C. S. ANDRADE1, Departamento de Genética e Biologia Evolutiva, Universidade de São Paulo, SP, Brazil 1 Departamento de Botânica e Ecologia, Universidade Federal do Mato Grosso, Cuiabá, MT, Brazil 2 Received 2 February 2021; revised 13 March 2021; accepted for publication 14 March 2021 An important goal of marine population genetics is to understand how spatial connectivity patterns are influenced by historical and evolutionary factors. In this study, we evaluate the demographic history and population structure of Littoraria flava, a highly dispersive marine gastropod in the Brazilian intertidal zone. To test the hypotheses that the species has (1) historically high levels of gene flow on a macrogeographical spatial scale and (2) a distribution in rocky shores that consists of subpopulations, we collected specimens along the Brazilian coastline and combined different sets of genetic markers (mitochondrial DNA, ITS-2 and single nucleotide polymorphisms) with niche-based modelling to predict its palaeodistribution. Low genetic structure was observed, as well as high gene flow over long distances. The demographic analyses suggest that L. flava has had periods of population bottlenecks followed by expansion. According to both palaeodistribution and coalescent simulations, these expansion events occurred during the Pleistocene interglacial cycles (21 kya) and the associated climatic changes were the probable drivers of the distribution of the species. This is the first phylogeographical study of a marine gastropod on the South American coast based on genomic markers associated with niche modelling. ADDITIONAL KEYWORDS: ecological niche modelling – Littorinidae – marine connectivity – phylogeography – population genomics. INTRODUCTION that this presumption is not entirely true (Bucklin, 2000; Cowen, 2000; Launey, 2002; Taylor, 2003; Molluscs are the second largest phylum in terms of the Rynearson & Armbrust, 2004; Sotka et al., 2004; number of species and play a fundamental role in rocky Carini & Hughes, 2006; Kramarenko & Snegin, 2015). shore communities. Similar to most coastal marine The genetic differentiation of the populations of rocky invertebrates, these species often have a sedentary shore molluscs depends not only on dispersal capacity adult phase following a pelagic larval period, during but also on larval duration and behaviour, local which gene flow occurs among populations. Therefore, adaptation, hydrographical barriers and life-history larval dispersal success potentially plays a crucial role traits (Palumbi, 2003; Ayre et al., 2009). in connectivity in populations, recolonization rate, Highly heterogeneous environments, such as rocky species distribution and local population persistence/ shore intertidal zones, present physical and biological extinction over time in marine ecosystems (Levin, gradients over short geographical distances. Such 2006; Cowen & Sponaugle, 2009). It has been widely heterogeneity can be measured by several predictors assumed that the pelagic larval phase facilitates and could favour different genotypes in different dispersal over broad scales, in turn promoting high microhabitats (Carini & Hughes, 2006; Funk et al., 2012; levels of gene flow and geographically uniform genetic Kramarenko & Snegin, 2015). Littorinids are a taxonomic structure. Nonetheless, several studies have shown group of herbivorous gastropods that inhabit the global intertidal zones. For example, Littoraria flava (King & *Corresponding author. E-mail: thainacortez@usp.br and Broderip, 1832) has a continuous and broad distribution soniacsandrade@ib.usp.br in the supratidal areas of rocky shores and estuarine © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015 999
1000 T. CORTEZ ET AL. environments of the Western Atlantic Ocean (Beasley south-eastern (SE) and southern (S) regions of the et al., 2005; Reid et al., 2010; Rodrigues et al., 2016; Lima Brazilian coast (Fig. 1) under the Instituto Chico et al., 2017). Although the adults are relatively sedentary, Mendes de Conservação da Biodiversidade licence its planktotrophic larvae show great dispersal capacity, no. 56726-1. In six of the 11 localities, samples were with durations ranging between 3 and 10 weeks (Reid, collected along horizontal transects in the supralittoral 1986, 1999; Rios, 1994). zone toward the sea, according to the experimental The gene flow dynamics in L. flava have been design of Andrade & Solferini (2007). The distances assessed from allozyme markers (Andrade et al., 2005; among the sites were established using the following Andrade & Solferini, 2007), which revealed a moderate formula: Dn = 2n−1, where D is the distance in metres genetic structure at the macrogeographical scale and and n is the sequence number of the site. The samples strong variation across only a few metres, based on were collected within 1 m2 of the marked point at sites Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 subpopulations within single rocky shores. These 0, 4, 8, 16, 32 and 64 m along the transect. Three sites findings suggest asynchronous spawning associated with the highest specimen abundance in each transect with recurrent colonizations, as mostly reported for were sampled. The approach allowed the investigation direct developing species (Janson, 1987; Johannesson of genetic structure at both the macro and micro spatial & Tatarenkov, 1997; Tatarenkov & Johannesson, 1999). scales. For mtDNA and ITS-2 sequencing, at least two The ecological importance of L. flava (Reid, 1986, individuals from each locality were selected, without 1999) and its extensive geographical distribution, high discerning the sites within transects. The collected abundance and highly dispersive larvae with short individuals were stored in liquid nitrogen until DNA generation time are conducive for investigations on extraction. genetic connectivity in marine ecosystems at both the broad and narrow scales. Here, we aimed to assess the population dynamics DNA extraction, amplification and sequencing and phylogeographical patterns underlying the Genomic DNA was extracted according to the CTAB connectivity of a non-model organism, L. flava. To protocol of Doyle & Doyle (1987). DNA integrity gain insights into the demographic history of the was checked by 1% agarose gel electrophoresis species, we combined genetic data from genomic single and quantified using a dsDNA BR Assay kit nucleotide polymorphisms (SNPs), mitochondrial DNA (Invitrogen, Carlsbad, CA, USA) on a Qubit v3 (mtDNA) and nuclear internal transcribed spacer 2 fluorometer (Invitrogen). Partial sequences of the (ITS-2) with palaeoclimatic simulations to predict the mitochondrial genes cytochrome oxidase subunit palaeodistribution of the species. Based on the dispersal 1 (COI), 16S ribosomal RNA (16SrRNA) and the capacity of L. flava in the highly heterogeneous ITS-2 were obtained for subsequent analyses. The intertidal environment, we (i) reconstructed a likely COI fragment was amplified using the primer pair demographic scenario using coalescent methods in LCO1490/HCO2198 (LCO1490: 5′-GGTCAACAA Brazilian populations to determine their connectivity ATCATAAAGATATTGG-3′, HCO2198: 5′-TAAAC levels and (ii) expected to reveal the genetic structure TTCAGGGTGACCAAAAAATCA-3′, Folmer et al., at the microspatial scale (Andrade & Solferini, 2007). 1994). The primer sets 16SrRNAH/16SrRNAR To the best of our knowledge, this is the first study (16SrRNA-H: 5′-CGCCTGTTTATCAAAAACAT-3′, to elucidate the evolutionary history and connectivity 16SrRNA-R: 5′-CCGGTCTGAACTCAGATCACGT-3′, patterns of a widespread marine invertebrate on the Palumbi, 1991) and ITS-1F/jfITS-1-3r (ITS-1F: Brazilian coast by combining genome-wide genetic 5′-GTTTCCGTAGGTGAACCT-3′, Rokicka et al., markers, mtDNA and palaeoclimate simulations. 2007; jfITS-1-3r: 5′-GAGCCGAGTGATCCACCGC TAAGAGT-3′, Dawson & Jacobs, 2001) were used for amplification of 16SrRNA and ITS-2 partial sequences. MATERIALS AND METHODS Polymerase chain reactions (PCRs) were carried To assess the population dynamics and demographic out in a total volume of 20 µL containing 1 × PCR history of L. flava, we applied the methodological steps buffer, 2.5 U Taq DNA Polymerase, 1.5 mM MgCl 2, illustrated in Supporting Information Fig. S1 and 200 μM of each dNTP, 0.35 µM of each oligonucleotide detailed in the following sections. and ultrapure water to achieve the reaction volume (Taq PCR Master Mix, Qiagen). The cycling profile on the thermal cycler involved a denaturation step at Sampling and experimental design 94 °C for 1 min, 35 cycles of denaturation at 94 °C for Ninety-three L. flava individuals were collected 1 min; annealing temperature for 1 min, extension at between January 2018 and January 2019 from 11 72 °C for 1 min; and extension at 72 °C for 10 min. locations distributed across the north-eastern (NE), Annealing temperatures were 52.5 °C for COI, 51 °C © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
L. FLAVA CONNECTIVITY AND DEMOGRAPHIC HISTORY 1001 Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 Figure 1. Distribution of sampling locations along the Brazilian coastline. The blue dots indicate sampling collection points, and the orange dots indicate the occurrence points of Littoraria flava, according to GBIF, used in the niche modelling analyses. for 16SrRNA and 49 °C for ITS-2. PCR products were Library construction and SNP calling purified with polyethylene glycol solution 15% (PEG) Individual libraries were generated based on the and then amplified for sequencing using a BigDye genotyping-by-sequencing method described by Terminator Cycle Sequencing Kit v.3.1 (Applied Elshire et al. (2011). Briefly, the genomic DNA of each Biosystems) at the Myleus Sequencing Facility (Belo sample was digested with the PstI restriction enzyme Horizonte, Brazil). (5′-CTGCAG-3′) and ligated to barcode and common Chromatograms were analysed using Geneious adaptors with appropriate sticky ends. The products v.9.1.8 (Biomatters Ltd), which was used to perform were grouped into sets of 40–53 samples and then BLAST searches (http://blast.ncbi.nlm.nih.gov) to check amplified by PCR using generic primers matching the for contamination or sequencing errors. Individual common adaptors under the following conditions: 5 min consensus sequences were aligned using MEGA v.7.0 at 72 °C, 30 s at 98 °C, 18 cycles of 10 s at 98 °C, 30 s at (Kumar et al., 2016). The 16SrRNA and COI data 65 °C and 30 s at 72 °C, and an extension step of 5 min were concatenated with DnaSP v.5.10.01 (Librado & at 72 °C. The presence of spare adapters and the sizes Rozas, 2009) into a mitochondrial haplotype (mtDNA). of the DNA fragments were assessed by quantification Based on the heterozygous sites, the phased haplotypes on an Agilent 2100 Bioanalyzer (Agilent Technologies) of ITS-2 were obtained with PHASE (Stephens & with the Agilent DNA 1000 kit and by quantitative Donnelly, 2003) from DnaSP, according to IUPAC PCR on a Light Cycler 480II (Roche) with a Kapa ambiguity codes and using a minimum posterior Biosystems kit for library quantification. Libraries probability of 0.9 (Garrick et al., 2010). Finally, the were constructed by EcoMol Consultoria (Piracicaba, mtDNA and ITS-2 haplotypes were concatenated with SP, Brazil), and sequencing was performed at the DnaSP into a final single haplotype. In summary, three Center for Functional Genomics Applied to Agriculture different datasets were obtained: the mitochondrial and Agroenergy (Animal Biotechnology Laboratory, DNA (mtDNA with COI + 16SrRNA), nuclear DNA LZT/ESALQ/USP, Piracicaba, SP, Brazil) on a HiSeq (ITS-2) and concatenated (mtDNA + ITS-2) datasets. 2500 platform (Illumina Inc., San Diego, CA, USA). © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
1002 T. CORTEZ ET AL. The SeqyClean pipeline (Zhbannikov et al., and 200 sampling increments with a burn-in of 20 000 2017) was used to filter out sequences smaller than were applied, in addition to an adaptive heating 50 bp and remove adapter sequencess and other scheme. contaminants from the UniVec database (NCBI, ftp:// ftp.ncbi.nlm.nih.gov/pub/UniVec/). The iPyrad v.0.7.28 program (Eaton, 2014) was used to assign reads to Single nucleotide polymorphisms individual samples and to edit and cluster reads into Clustering analyses was performed on SNPs from (1) consensus sequences through paralogue identification. all sampled individuals from the 11 localities and (2) The density of missing data per locus was analysed individuals within each transect, separately, using the through matrix occupancy (de Medeiros & Farrell, Bayesian method in STRUCTURE v.2.3.4 (Pritchard 2018). The obtained VCF file (Variant Call Format) et al., 2000). After the most likely lambda (λ) was Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 was converted to other program-specific input formats determined according to the author’s recommendation, using PGDSpider v.2.1.15 (Lischer & Excoffier, 2012). the population number (K) was allowed to vary from PLINK (Purcell et al., 2007) was used to remove SNPs 1 to 11. The program ran each K-value 20 times with with minimum allele frequency lower than 1%, missing a burn-in of 50 000, followed by 1 000 000 MCMC genotypes higher than 20% and linkage disequilibrium iterations. The most likely number of genetic groups (r2 = 0.50). was selected based on ΔK using STRUCTURE HARVESTER (Earl & vonHoldt, 2012) and graphically illustrated using the pophelper package (Francis, 2017) Genetic diversity implemented in R. Discriminant analysis of principal The number of haplotypes (H), number of polymorphic components (DAPC, Jombart, 2008) was implemented sites (S), number of segregating sites (θ S ) and using the adegenet package to describe the genetic nucleotide diversity (θ π) were measured for mtDNA variance among the data. The analysis was performed and ITS-2 data per location in Arlequin v.3.5 (Excoffier with the clusters identified by STRUCTURE as & Lischer, 2010). Multi-loci estimates of expected prior information, and with a number of principal heterozygosity (H E ) and observed heterozygosity components (PCs) selected according to the α-score (HO) were calculated for SNPs, also in Arlequin. The function. Genetic differentiation was assessed with fixation index FIS (Weir & Cockerham, 1984) across loci AMOVA, following the methodology and hierarchical and its significance were computed with the adegenet levels applied for mtDNA and ITS-2 markers. Within package v.1.4 (Jombart, 2008) in R (R Core Team, 2013) transects, each site was considered a population unit, with 10 000 permutations, 10 000 dememorization and which were also tested for the isolation-by-distance 100 000 Markov chains Monte Carlo (MCMC) steps (IBD) model using the Mantel test implemented in (P < 0.05). adegenet (10 000 permutations). For that, based on the localities’ coordinates, the geographical distances were transformed into Euclidean distances using the dist Population structure function of R. Mitochondrial DNA and ITS-2 Minimum spanning networks (MSNs) were constructed Demographic history and niche-based in PopART v.1.7 (Leigh & Bryant, 2015) for mtDNA modelling and ITS-2. Arlequin was used to calculate genetic To explore historical demography from mtDNA and differentiation based on the unbiased FST estimator ITS-2 data, the demographic processes over time θ (Weir & Cockerham, 1984), where the significance were assessed using Tajima’s D (Tajima, 1989) and was determined by running 10 000 permutations Fu’s F S (Fu, 1997) neutrality tests, and mismatch and 100 000 MCMC steps in analysis of molecular distribution analyses (Harpending et al., 1993), variance (AMOVA, P < 0.05). To avoid mixing different performed by Arlequin with 10 000 permutations. population units, the AMOVA followed hierarchical Considering the absence of a clear fossil record levels, considering (1) individuals within a location and molecular clocks of L. flava for these markers, and (2) individuals within a region (South, Southeast estimation of the divergence times between clades was and Northeast) as population units. For each marker, conducted using BEAST v.2.6.2 (Drummond et al., Migrate v.4.4.3 (Beerli, 2009) was used to estimate 2012) for the concatenated dataset (mtDNA + ITS-2) long-term gene flow by calculating the population size with the GTR+G nucleotide substitution and a time- parameter (θ) and the average number of migrants calibrated topology based on the estimated L. flava per generation (Nm) using the formula: Nm = θi × Mj→i. origin, ±36 Mya (Reid et al., 2010). Parameters Two parallel runs with 2 × 105 recorded genealogies, applied were the uncorrelated relaxed-clock, and © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
L. FLAVA CONNECTIVITY AND DEMOGRAPHIC HISTORY 1003 400 million steps were sampled every 5000 steps, The niche-based models were developed using five with a burn-in of 400 000. The tree was rooted with mathematical algorithms to improve the reliance of Littoraria intermedia as the outgroup (accession models, which were (1) envelope score from Bioclim (Booth numbers KT149308 and KT149304). The convergence et al., 2014); (2) Domain–Gower distance (Carpenter et al., of effective sample size was assessed using the Tracer 1993); (3) support vector machines (Tax & Duin, 2004); v.1.7.1 program (Rambaut & Drummond, 2003), and (4) maximum entropy (MaxEnt) (Phillips & Dudík, 2008) the resulting tree was edited with FigTree v.1.4.4 and (5) Random Forest (Breiman, 2001). All models were (Rambaut, 2007). developed to predict mH and LGM climate conditions Alternative evolutionary scenarios and relevant using the dismo (Hijmans et al., 2017) and kernlab population parameters were estimated using (Karatzoglou et al., 2004) R packages. To evaluate the Fastsimcoal v.2.6 (Excoffier & Foll, 2011) based on models, we used a bootstrap method to randomize the Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 the 6094 SNPs. This program uses the minor allele occurrence points into two subsets of 2 k-folding (training site frequency spectrum (SFS) parameter, computed and test), repeated 20 times to decrease data collinearity. with Arlequin (1000 bootstraps), to infer demographic We conducted an ensemble forecasting approach (Araujo and population parameters. We tested three distinct & New, 2007) to predict a consensual map with the demographic scenarios: the first assumed an expansion frequency of species in each site. Finally, we produced one of the Northeastern populations in the past (divergence map with current species distribution patterns, one for time backward in time), where all individuals migrate the mH and one for the LGM. to the other two regions. The second and third models assumed expansion events with post-migration in the Southeastern and Southern populations, respectively. Data availability The time of divergence (TDIV) was allowed to range The raw de-multiplexed sequences generated in from one to 500 000 generations back in time. For this study are available in NCBI SRA as BioProject each model, we ran 50 independent replicates, each Accession PRJNA656564. mtDNA and ITS-2 sequences including 40 estimation loops (-L 40) with 300 000 (-n were deposited in GenBank and accession numbers 300 000) coalescence simulations. The probability of are listed in the Supporting Information (Table S1). each model given the observed data was determined based on both the maximum likelihood value and Akaike’s information criterion (AIC). The palaeodistribution of L. flava was fitted RESULTS through ecological niche modelling (ENM) based on information on known occurrence locations and Datasets climate variables to predict the niches of species A total of 93, 63 and 46 samples were analysed (Ferrier & Guisan, 2006; Stigall, 2012; Alvarado- for the SNPs mtDNA, and ITS-2, respectively Serrano & Knowles, 2014). We collected species (Table 1; Supporting Information, Table S1). The occurrence records from the GBIF dataset (https:// differences in the sizes of dataset samples among the www.gbif.org/) and our sampled locations, totalling 59 markers resulted from difficulties in obtaining the occurrence points (Fig. 1). Eighteen marine variables amplification products, especially for ITS-2. At least available on the MARSPEC database (http://marspec. six individuals per location were sequenced, except at weebly.com/) (Sbrocco & Barber, 2013) for current, PIF, which allowed us to conduct comparative analysis mid-Holocene (mH, 6 kya) and Last Glacial Maximum among the distinct datasets. The COI and 16SrRNA (LGM, 21 kya) temporal scenarios were used as sequences were 626 and 487 bp, respectively, with 31 environmental predictors for model fitting. Marine and 19 polymorphic sites. The single mitochondrial variables were downloaded at a spatial resolution haplotype measured 1113 bp, with 46 variables sites, of 5 arc-minutes (~10 × 10 km, in Equator region) 13 parsimony-informative sites and 33 singletons. (Sbrocco, 2014), clipped to the South Atlantic Ocean, ITS-2 sequences were 405 bp with 11 polymorphic spanning 50°S to 15°N latitude, and 80°W to 45°W sites, ten parsimony-informative sites and one longitude. A factorial analysis with Varimax rotation singleton. The concatenated dataset (mtDNA + ITS-2) was conducted to select non-correlated predictors was 1518 bp, in which 44 sites were polymorphic and that explain environmental variation in the study parsimony-informative. area (see details in Sobral-Souza et al., 2015). The The library construction resulted in 322 479 123 reads predictors included bathymetry, plan curvature, of 93 L. flava individuals (BioProject PRJNA656564). profile curvature, salinity of the saltiest month, Seqyclean removed ~29% of the reads (Supporting annual salinity range and annual sea temperature Information, Table S2). Of the remaining 227 380 970 range (°C). sequences, the first quality filter, using iPyrad, © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
1004 T. CORTEZ ET AL. Table 1. Geographic coordinates of sampling locations and sample sizes of each genetic marker. GPS coordinates Sample size Location Sites Abbr. Lat. (°S) Long. (°W) SNPs mtDNA ITS-2 Northeast Region (NE) Sabiaguaba NA SBF 3°47′24″ 38°25′23″ 11 7 6 Pier das Algás NA ALF 9°36′57″ 35°44′13″ 6 5 2 Southeast Region (SE) Anchieta 4, 8, 64 m ACF 20°48′37″ 40°39′39″ 8 2 5 Gamboa 0, 2, 32 m GAF 20°53′19″ 40°45′55″ 8 3 4 Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 Barra de São João 0, 8, 32 m SJF 22°35′55″ 41°59′25″ 10 7 5 Praia da Gorda 0, 8, 32 m PGF 22°43′48″ 41°58′20″ 10 5 3 Praia Dura 4, 16, 64 m DF 23°29′32″ 45°09′55″ 9 10 7 Araçá 4, 16, 64 m ARF 23°48′47″ 45°24′31″ 8 10 4 South Region (S) Santo Antônio NA STF 27°30′46″ 48°30′57″ 7 7 4 Ribeirão da Ilha NA RBF 27°42′45″ 48°33′40″ 14 5 4 Ponta de Ribeirão da Ilha NA PIF 27°49′54″ 48°34′14″ 2 2 2 Total 93 63 46 The table presents the abbreviation (Abbr.) used of each location and whether the sampling was performed randomly (NA) or through transects. The sample size is given for each genetic marker (SNPs, mtDNA, and ITS-2). Table 2. Diversity indices based on mtDNA (COI + 16SrRNA), ITS-2, and SNPs of Littoraria flava. mtDNA ITS-2 SNPs Location N H θS θπ N H θS θπ N HO HE FIS Northeast Region SBF 7 6 6.5 4.5 6 12 2.6 3.6 11 0.16 0.18 0.09 ALF 5 4 3.8 3.2 2 4 1.6 1.8 6 0.22 0.26 0.10 Southeast Region ACF 2 2 3.0 3.0 5 4 3.2 3.8 8 0.20 0.21 0.05 GAF 3 2 1.3 1.3 4 4 3.8 4.5 8 0.18 0.21 0.11 SJF 7 6 4.4 4.3 5 8 3.0 3.6 10 0.17 0.19 0.10 PGF 5 4 5.7 5.4 3 6 3.1 3.3 10 0.16 0.19 0.11 DF 10 7 3.8 2.8 7 14 2.2 3.3 9 0.18 0.20 0.08 ARF 10 7 4.2 3.4 4 8 2.3 2.3 8 0.19 0.22 0.14 South Region STF 7 7 3.6 3.1 4 6 3.1 3.8 6 0.20 0.26 0.16 RBF 5 3 3.8 3.6 4 6 3.9 4.6 7 0.19 0.24 0.17 PIF 2 2 3.0 3.0 2 4 3.2 3.6 2 0.50 0.54 0.05 Sample size (N), haplotype number (H), nucleotide differences (θS), nucleotide diversity (θπ), observed heterozygosity (HO), expected heterozygosity (HE), and fixation index (FIS) are shown for each location according to the genetic marker. Significant values of FIS and differences between HE and HO (P < 0.05) are shown in bold. Abbreviations as in Table 1. retained 19 133 SNPs within 2249 loci, with 12.67% Genetic diversity of data missing per locus on average. Eight samples Nucleotide diversity (θπ) between individuals was often were removed owing to the relatively large amount of low, ranging from 1.33 in Gamboa (SE) to 5.40 in Praia missing data (>35% per individual). The number of raw da Gorda (SE) for mtDNA and from 1.83 in Alagoas (NE) reads per sample ranged from 203 705 to 10 356 692 to 4.64 in Gamboa (SE) for ITS-2 (Table 2). In contrast, (Table S3). After PLINK filtering, we obtained 6094 the segregating sites were often high, with the highest SNPs within 1572 loci from 85 individuals. values observed in Sabiaguaba (NE) for mtDNA and in © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
L. FLAVA CONNECTIVITY AND DEMOGRAPHIC HISTORY 1005 Ribeirão da Ilha (S) for ITS-2. SNP analyses revealed represent a population unit, the mtDNA did not reveal that the observed heterozygosity across all loci was significant structuring (F ST = 0.024, P = 0.1283) in significantly lower than the expected heterozygosity in contrast to the observation based on ITS-2 (FST = 0.057, five out of 11 localities (P < 0.05) (Table 2). Eight of the P = 0.0016) and SNPs (FST = 0.006, P = 0.008). 11 average FIS values across the loci were significant, The pairwise FST calculated based on mtDNA and with the highest being 0.167 in Ribeirão da Ilha (S). ITS-2 revealed that of the 55 pairwise tests, four The mtDNA MSN had a star-shaped configuration, and 25 comparisons, respectively, were significant with the most abundant haplotypes as sources (Fig. (Fig. 3A). For SNPs, 13 out of 55 were significant 2A), and multiple unique haplotypes diverging based comparisons, mostly in the Southeastern locality Praia on only a few substitutions. The most diverse haplotype da Gorda (SE). Overall, such pairwise comparisons included individuals from all coastal regions. There was were not consistent with patterns of IBD for any local Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 no clear geographical pattern of haplotype distribution population. across localities for either ITS-2 or mtDNA. The Bayesian STRUCTURE assignment analysis revealed clear distinction among three clusters (K = 3): the first is mainly composed of individuals Population structure from the Northeast and Southeast regions; the Among localities second, individuals from the Southeast and South The results of AMOVA with all genetic markers regions; and the third, individuals from the three revealed most of the total variance within localities regions, with no predominance of any of the groups and low yet significant FST and FSC indices (mtDNA (Fig. 2B, C). The scatter plot from the results of FST = 0.077, ITS-2 FST = 0.162 and SNPs FST = 0.014) DAPC revealed slight differentiation among the (Table 3). Assuming all individuals from a region three clusters (Fig. 2D). Figure 2. Haplotype networks and clustering analysis of Littoraria flava. A, minimum spanning network of mtDNA, to the left, and ITS-2 data, to the right. Each node represents a haplotype, its size corresponds to the haplotype frequency and its colours indicate the locality. Black nodes represent inferred ancestral nodes. B, assignment plot based on STRUCTURE analyses with 6094 SNPs and K = 3. Each vertical bar corresponds to one individual, and the colour indicates the proportion of membership in each genetic cluster. C, K = 3 inferred by STRUCTURE HARVESTER for the genetic assignment based on 6094 SNPs. D, DAPC density plot for 85 L. flava individuals. Dots represent individuals, with colours denoting the genetic groups identified by STRUCTURE. Abbreviations are as in Table 1. © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
1006 T. CORTEZ ET AL. Table 3. Analysis of molecular variance (AMOVA) results from all genetic markers. Each group corresponds to a region from the Brazilian coast (Northeast, Southeast, and South). Marker Source of variation d.f. Variance components % Variation mtDNA Among groups 2 0.002 0.10 Among localities 8 0.15 7.57 FST = 0.077* Among individuals 53 179.84 92.32 FSC = 0.076* Total 63 179.99 ITS-2 Among groups 2 0.01 0.46 Among localities 8 0.33 15.78 FST = 0.162** Among individuals 65 174.09 83.77 Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 FSC = 0.158** Total 75 207.83 SNPs Among groups 2 0.09 0.22 Among localities 8 0.17 0.40 FST = 0.014** Among individuals 159 4198.64 99.38 FSC = 0.011* Total 169 4198.91 The results indicate the source of variation with its degrees of freedom (df), variance of component, and percentage variation (% Variation). * P < 0.05; ** P < 0.01. Figure 3. Population differentiation and migration results for mtDNA, ITS-2 and SNP markers from 11 sampled locations. A, heatmaps of the pairwise FST for each genetic marker. Colour coding illustrates the observed FST value. Only significant indices are shown (P < 0.05). On the left, the localities are grouped according to the coastline regions. B, migration rates per generation according to Migrate-n analysis for both mtDNA and ITS-2. On the right, migration probabilities according Fastsimcoal for SNPs. The circle sizes and colours represent the rates of migrants from the source locality (rows) toward the receiver locality (columns). Abbreviations are as in Table 1. The migration rates per generation, estimated localities, Praia Dura (SE), Araçá (SE) and Praia using Migrate-n for mtDNA and ITS-2, were da Gorda (SE) (Fig. 3B; Supporting Information, inconsistent among markers for the Southeastern Table S4). While the mtDNA indicated that the © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
L. FLAVA CONNECTIVITY AND DEMOGRAPHIC HISTORY 1007 rates of migrants toward Praia da Gorda (SE) were Demographic history and ecological niche close to zero, the ITS-2 data showed rates almost modelling 300 times higher for the same locality. The opposite The population events investigation highlighted the trends were observed in Praia Dura (SE) and Araçá contrasting patterns between the mtDNA and ITS-2 (SE), where the mtDNA indicated rates of migrants data. While the mtDNA data only revealed significant approximately 50–100 times higher than the rates negative D for the Northeastern localities (P < 0.05; associated with the ITS-2 data. All other locations Supporting Information, Table S5), the ITS-2 data had high numbers of migrants – usually greater revealed significant negative values of Fu’s FS test for than 200, either sender or receiver – indicating high eight out of 11 localities, and only Praia da Gorda (SE) levels of gene flow, even across large distances. The showed a significant Harpending’s raggedness index. Fastsimcoal results for 6094 SNPs revealed high The topology provided by BEAST did not recover Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 probabilities of past migration among populations a geographical pattern at the origin of the current from different regions of the coastline, such as between distribution of L. flava populations (Figure S3). Santo Antônio (S) and Alagoas (NE) and between Nonetheless, the tree topology with several diverging Ribeirão da Ilha (S) and Barra de São João (SE) branches suggests demographic expansion over (Fig. 3B). The highest probability of migration was time. The first ancestral divergence seemed to have from Santo Antônio (S) and Praia Dura (SE). generated two major clades: one (~16.7 Mya) mostly represented by the Southeastern localities and the second (~27.03 Mya) comprising individuals from all Within transects regions of the Brazilian coastline. STRUCTURE clustering analyses based on SNPs The optimal-fitting model tested by Fastsimcoal, from individuals sampled along the transects according to the highest likelihood (-2903.53) showed no signs of genetic structure, according to and the lowest AIC (13,327), was the one in the sites established during sampling. Instead, two which the population divergence was initiated clusters (K = 2) were detected for Gamboa (SE) and in the Southeastern region toward the other Praia da Gorda (SE), and five clusters (K = 5) were two regions (Figure 4). The best model assumes identified for all remaining locations (Supporting i n i t i a l u n i d i r e c t i o n a l g e n e f l o w, f o l l o w e d b y Information, Fig. S2). Nonetheless, there was no bidirectional migration across all regions. The low clear microgeographical pattern for any location. variation in likelihoods (Δ likelihood = 417.90) The results of both pairwise F ST and Mantel’s obtained from each replicate of the best model tests among the sites within transects revealed no indicated robust estimates. The divergence time statistically significant correlation. from the Southeastern populations was ~150 000 Figure 4. Comparisons among all demographic models, where the best-fitting scenario assumes the divergence of Littoraria flava populations originating in the Southeastern region around 150 000 generations backward in time. A, illustrations of demographic models tested using Fastsimcoal with topologies showing the origins of divergence from the Northeast, Southeast and South regions. Numbers below the models are divergence times, Akaike information criterion values and the delta likelihood (difference between maximum possible and maximum obtained model likelihood in log10 units) of each model. B, boxplots showing the log10 likelihood distributions based on the observed SFS. Individuals were grouped according to their region of the Brazilian coastline, represented by the different colours. © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
1008 T. CORTEZ ET AL. Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 Figure 5. Modelled distributions of Littoraria flava for (A) Last Glacial Maximum (21 kya), (B) mid-Holocene (6 kya) and (C) present (0 kya) scenarios. The suitability value predicts how adequate the environment is for the species occurrence. generations ago. The second-best model (highest obtained robust evidence from coalescent analysis and likelihood = −2894.81, AIC = 13 355) assumed a palaeodistribution reconstruction that the distribution divergence (~417 generations ago) from the Southern of L. flava has retracted since the LGM, probably due region, and presented greater variation of estimates to increasing sea surface temperature (SST), which has (Δ likelihood = 423.83). already been suggested for other littorinids. ENM indicated a geographical retraction from the LGM to the present time (Fig. 5). Geographical stability was proposed under the present and mH Intra-populational diversity and climate scenarios. The current and mH distribution microgeographical variation showed that L. flava was potentially distributed across almost the entire Brazilian coast, in addition Genetic diversity analysis based on mtDNA and to French Guiana, Suriname, Guyana, Venezuela and ITS-2 revealed greater diversity within populations part of Colombia. However, a considerable reduction and numerous shared sequences among locations. was noted in the distribution during the LGM, with Similar patterns have been reported for littorinids, the most suitable areas for L. flava being restricted to with both direct and planktotrophic development the Southeastern locations of the Brazilian coast. (Je Lee & Boulding, 2009; Evangelisti et al., 2017; Nehemia et al., 2019), and seem to be caused by high gene flow among populations and large population sizes. There was no clear structuring within transects based on the delimited sites, which DISCUSSION is inconsistent with the result of a previous study To the best of our knowledge, this is the first investigation (Andrade & Solferini, 2007). This difference could be of the evolutionary history and connectivity patterns caused by the different mutation rates of SNPs and of a widely distributed marine invertebrate along the allozyme markers (Johannesson & Tatarenkov, 1997; Brazilian coastline based on genomics in combination Sunnucks, 2000; Schlötterer, 2004) and, because with ENM. Our combined genetic data support the allozymes are involved in metabolic functions, hypothesis that L. flava has been dispersing over natural selection might be involved (Tatarenkov & large distances for a long time, with the populations Johannesson, 1999; Johannesson et al., 2004; Carini distributed in distant locations currently interconnected & Hughes, 2006; Kramarenko & Snegin, 2015). The by gene flow. Besides a high rate of gene flow, the observed microgeographical structure within the genetic structure exhibited three clusters distributed transect did not exhibit an IBD model (Supporting across the study area, which could be due to oceanic Information, Fig. S2). The observed structure among circulation and local environmental conditions. We sites, with cluster numbers varying from two to © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
L. FLAVA CONNECTIVITY AND DEMOGRAPHIC HISTORY 1009 five, could be due to variation in environmental monomorphic sites were only found in Anchieta (SE), and temporal conditions, in addition to variance in the null alleles are unlikely to generate heterozygosity reproductive success, which together could result deficits in so many populations (Carlson et al., 2006; in differences in the genetic composition of larvae Crooks et al., 2013). The Wahlund effect is a plausible that colonize an area. Such interactions might explanation for our results if there was chaotic result in apparently chaotic genetic patchiness recruitment of cohorts from different origins or if (Hedgecock et al., 2007; Liu & Ely, 2009; Villacorta- many breeding groups constituted each population, Rath et al., 2018), primarily because only a small which might explain the structuring within the fraction of individuals contribute to the subsequent transects. Although such mechanisms may not be generations (Hedgecock & Pudovkin, 2011). Natural entirely responsible for our results, we cannot exclude selection driven by heterogeneous environmental the possibility that they have shaped the population Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 conditions, either during the larval stage or after dynamics of L. flava (Andrade et al., 2003, 2005). settlement, might also generate similar patterns Heterozygote deficiency could also be explained by (Hedgecock et al., 1994). Few marine studies have gene flow dynamics according to a metapopulation attempted to link environmental factors with model, where the history of extinctions would yield genetic differentiation patterns at the fine spatial low heterozygosity rates and the frequent population scale (Berkström et al., 2013; DiBattista et al., 2017; turnover would decrease genetic variation among Coscia et al., 2020). local populations (Gilpin, 1991; Smedbol et al., 2002). Andrade & Solferini (2007) found significant temporal variation among individuals living in the same rocky shore. Furthermore, based on the sizes Population divergence and gene flow of individuals, previous work (our unpubl. data) has The findings of the our study corroborate the also demonstrated the asymmetric abundance of assumption of interconnected populations across broad L. flava adults and juveniles on distinct rocky shores spatial scales with high levels of gene flow (Figs 2, 3; during the same period, indicating asynchronous Supporting Information, Table S4), given the lack of a and asymmetrical rates of arrival of new recruits. clear geographical pattern (Fig. 2A). The findings also If asynchronous colonization is true for the species, suggest high numbers of migrants per generation and we could also assume a possibility of local extinction high effective population size (Fig. 3B). Nevertheless, at any time. Our findings seem to be consistent with there were some discordant patterns between mtDNA the hypothesis that interconnected populations and ITS-2 marker results for Praia Dura, Araçá are composed of heterogeneous larval and recruit and Praia da Gorda, probably due to the different cohorts, resulting in unrecognizable local variation evolutionary dynamics and inheritance mechanisms, patterns. as already reported from phylogeographical studies The three clusters identified by STRUCTURE do not (King et al., 1999; Chu et al., 2001; Presa et al., 2002; highlight a geographical pattern through the species Ni et al., 2012; Santos-Neto et al., 2016). Overall, the distribution (Fig. 2B). There is a clear distinction migration probabilities estimated from Fastsimcoal between the clusters mostly represented by individuals are quite similar across all regions, supporting the from Northeastern + Southeastern localities and hypothesis of gene flow among distant locations. The Southeastern + Southern localities (clusters in highest migration probabilities obtained from SNPs yellow and blue in Fig. 2B, respectively). The clusters, are consistent with the non-significant pairwise FST however, are not composed of the same individuals that comparisons (Fig. 3). share the two most common mtDNA haplotypes (Fig. The low values of FST observed for SNPs (Fig. 3A) 2A). Within our study area, the passage of cold fronts suggests large effective population sizes or high along the Cabo Frio coast (Rio de Janeiro) is capable levels of genetic connectivity among sites (Bohonak, of completely modifying the wind patterns (Carbonel, 1999; Marko & Hart, 2011). Nonetheless, we observed 2003; Coelho-Souza et al., 2012), directly affecting significant heterozygote deficiency and high positive FIS coastal circulation and SST variation, explaining our estimates in several localities (Table 2). Such patterns palaeodistribution simulation patterns. Considering have already been reported for L. flava (Andrade et al., the geographical proximity of Cabo Frio and Praia 2005) and many other marine invertebrates (Addison da Gorda (~22 km), which seems to be the transition & Hart, 2005; Costantini et al., 2007; Knutsen et al., area between the two major genetic clusters, these 2003), usually attributed to natural selection, null coastal events may be the sources of such disturbances alleles, inbreeding and the Wahlund effect (Gajardo in larvae dispersal of L. flava, resulting in genetic et al., 2002; Whitaker, 2004). Because we removed differentiation, and possibly the retraction during the individuals with a high amount of missing data and LGM. Such an explanation has already been proposed © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
1010 T. CORTEZ ET AL. in population genetics and community studies in MSN topology, palaeodistribution and coalescent lobsters and fishes in the same region (Ekau, 1999; simulations (Slatkin & Hudson, 1991; Rogers, 1995; Maggioni et al., 2003; Freitas & Muelbert, 2004). The Avise, 2009). However, the fossil record for littorinids third genetic cluster (in red in Fig. 2B) may represent from Brazil is scarce and unsuitable for testing this those individuals that were able to cross the expansive hypothesis. The tree topology was also consistent with distance across the three coastal regions, despite the a continuum of population expansion, as there are oceanographic and climatic disturbances, due to the several clades with individuals from all regions of the plastic larval behaviour influenced by biophysical coastline originating from the initial clade, without conditions (Leis & Clark, 2005; Berumen et al., 2012; any apparent geographical isolation (Supporting Faillettaz et al., 2018). Because no study has adopted Information, Fig. S3). a similar genomic approach in studying the genetic In the present study, we have provided a first glimpse Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 structure of a marine invertebrate organism in the of the demographic processes that have shaped the Brazilian coastline, we could not compare these results population structure of a widespread intertidal with other studies of high-dispersal species along the Brazilian littorinid. Although our demographic same area. inferences based on genetic data do not reveal the actual demographic processes that occurred in the past, by integrating palaeodistribution simulations Demographic history and niche modelling with our genetic data, we obtained robust evidence While the mtDNA data revealed significant Tajima’s that L. flava has experienced some type of departure D only for Sabiaguaba and Alagoas, the ITS-2 data from population equilibrium since LGM, probably revealed that more than half of the localities had due to increasing SST. Despite comprehensive significant negative values of Fu’s FS, and only Praia understanding of the biotic and abiotic factors da Gorda had a significant Harpending’s raggedness influencing local species abundance and distribution index. Nevertheless, both sets of data presented clear on rocky shores, the interactive effects of such evidence that L. flava populations had expanded features over large temporal scales remain unclear. demographically, as supported by both coalescent Our combined genetic data support the hypothesis and palaeodistribution simulations. Both approaches that L. flava has been dispersing over large distances revealed geographical and population expansion from for a long time, with the populations distributed in the Southeastern area (Figs 4, 5), and the ENM results distant locations currently interconnected by gene suggested that the events occurred during the LGM flow. Here, we propose that during glacial periods, (21 kya). These findings indicate that past climate L. flava underwent geographical and population change could have had effects on broad-scale species retraction followed by subsequent expansion during distribution or diversity, and highlight their effect on multiple global glacial–interglacial cycles. Such population expansion after cooler and drier periods. patterns can be reliably used to explain the current Previous palaeoclimate investigations suggest diversity patterns of other marine species in rocky that the tropics were generally significantly cooler shores in the South Atlantic. during the LGM, even with regard to SST (Stute et al., 1995; Otto-Bliesner et al., 2006). Simulations for the mH (6 kya) revealed a small but significant ACKNOWLEDGEMENTS amount of annual cooling over the tropical oceans, often associated with reduced levels of methane and We are grateful to the Instituto de Biociências – USP, annual solar anomalies. Overall, precipitation in and the Centro de Biologia Marinha da Universidade oceans seems to have increased between the LGM and de São Paulo (CEBIMar) staff; to all colleagues at the mH (Otto-Bliesner et al., 2006). Organisms from the Laboratório de Diversidade Genômica, especially the intertidal ecosystem tend to respond more rapidly Cecili Mendes, for assisting in the fieldwork; to EcoMol to climate change by, for instance, an alternation in staff for library preparation; and to Darwin server their geographical distribution ranges (Williams administrators for their assistance. We are also grateful & Morritt, 1995; Menge, 2000; Menge et al., 2007). to Cristina Yumi Miyaki, Vera Nisaka Solferini and Thus, the range of a species sensitive to climatic Gustavo Muniz Dias for helpful suggestions during fluctuation, such as L. flava, could have extended the discussion of the results, and the two anonymous into the Brazilian coast in response to progressive reviewers for their comments. We would like to thank warming following the LGM, a process that would have Editage (www.editage.com) for English language involved serial bottleneck events. Such scenarios have editing. This work was supported by CAPES and the already been suggested for other littorinids (Je Lee & Fundação de Amparo à Pesquisa do Estado de São Boulding, 2009; Silva et al., 2013; Evangelisti et al., Paulo (FAPESP, processes 2015/20139-9, 2018/05118-3 2017), and would explain our neutrality test results, and 2019/11478-5). The authors declare that they have © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
L. FLAVA CONNECTIVITY AND DEMOGRAPHIC HISTORY 1011 no known competing financial interests or personal Bucklin A. 2000. Population genetic variation of Calanus relationships that could have appeared to influence finmarchicus in Icelandic waters: preliminary evidence of the work reported in this paper. genetic differences between Atlantic and Arctic populations. ICES Journal of Marine Science 57: 1592–1604. Carbonel C. 2003. Modelling of upwelling–downwelling cycles caused by variable wind in a very sensitive coastal system. REFERENCES Continental Shelf Research 23: 1559–1578. Addison JA, Hart MW. 2005. Spawning, copulation and Carini G, Hughes J. 2006. Subdivided population structure inbreeding coefficients in marine invertebrates. Biology and phylogeography of an endangered freshwater Letters 1: 450–453. snail, Notopala sublineata (Conrad, 1850) (Gastropoda: Alvarado-Serrano D, Knowles L. 2014. Ecological Viviparidae), in Western Queensland, Australia: population niche models in phylogeographic studies: applications, genetic structure and phylogeography of N. sublineata. Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 advances and precautions. Molecular Ecology Resources Biological Journal of the Linnean Society 88: 1–16. 14: 233–248. Carpenter G, Gillison A, Winter J. 1993. DOMAIN: a flexible Andrade S, Magalhaes C, Solferini V. 2003. Patterns of modelling procedure for mapping potential distributions of genetic variability in Brazilian Littorinids (Mollusca): a plants and animals. Biodiversity and Conservation 2: 667–680. macrogeographic approach. Journal of Zoological Systematics Carlson CS, Smith JD, Stanaway IB, Rieder MJ, and Evolutionary Research 41: 249–255. Nickerson DA. 2006. Direct detection of null alleles in SNP Andrade S, Medeiros H, Solferini V. 2005. Homogeneity genotyping data. Human Molecular Genetics 15: 1931–1937. test of Hardy–Weinberg deviations in Brazilian littorinids: Chu K, Li C, Ho H. 2001. The first Internal Transcribed evidence for selection? Journal of Molluscan Studies 71: Spacer (ITS-1) of ribosomal DNA as a molecular marker for 167–174. phylogenetic and population analyses in Crustacea. Marine Andrade S, Solferini V. 2007. Fine-scale genetic structure Biotechnology 3: 355–361. overrides macro-scale structure in a marine snail: nonrandom Coelho-Souza S, López M, Guimarães J, Coutinho R, recruitment, demographic events or selection?. Biological Candella R. 2012. Biophysical interactions in the Cabo Frio Journal of the Linnean Society 91: 23–36. upwelling system, southeastern Brazil. Brazilian Journal of Araujo M, New M. 2007. Ensemble forecasting of species Oceanography 60: 353–365. distributions. Trends in Ecology & Evolution 22: 42–47. Costantini F, Fauvelot C, Abbiati M. 2007. Fine-scale Avise J. 2009. Phylogeography: retrospect and prospect. genetic structuring in Corallium rubrum: evidence of Journal of Biogeography 36: 3–15. inbreeding and limited effective larval dispersal. Marine Ayre D, Minchinton T, Perrin C. 2009. Does life history Ecology Progress Series 340: 109–119. predict past and current connectivity for rocky intertidal Coscia I, Wilmes S, Ironside J, Goward-Brown A, invertebrates across a marine biogeographic barrier? O’Dea E, Malham S, McDevitt A, Robins P. 2020. Fine- Molecular Ecology 18: 1887–1903. scale seascape genomics of an exploited marine species, the Beasley C, Fernandes C, Gomes C, Brito B. 2005. Molluscan common cockle Cerastoderma edule, using a multimodelling diversity and abundance among coastal habitats of Northern approach. Evolutionary Applications 13: 1854–1867. Brazil. Ecotropica 11: 9–20. Cowen R. 2000. Connectivity of marine populations: open or Beerli P. 2009. How to use MIGRATE or why are Markov closed? Science 287: 857–859. chain Monte Carlo programs difficult to use. Population Cowen RK, Sponaugle S. 2009. Larval dispersal and marine Genetics for Animal Conservation 17: 42–79. population connectivity. Annual Review of Marine Science 1: Berkström C, Lindborg R, Thyresson M, Gullström M. 443–466. 2013. Assessing connectivity in a tropical embayment: fish Crooks L, Carlborg Ö, Marklund S, Johansson AM. migrations and seascape ecology. Biological Conservation 2013. Identification of null alleles and deletions from SNP 166: 43–53. genotypes for an intercross between domestic and wild Berumen M, Almany G, Planes S, Jones G, Saenz- chickens. G3: Genes, Genomes, Genetics 3: 1253–1260. Agudelo P, Thorrold S. 2012. Persistence of self- Dawson M, Jacobs D. 2001. Molecular evidence for cryptic recruitment and patterns of larval connectivity in a marine species of Aurelia aurita (Cnidaria, Scyphozoa). The protected area network: Connectivity Patterns in An MPA Biological Bulletin 200: 92–96. Network. Ecology and Evolution 2: 444–452. DiBattista J, Travers M, Moore G, Evans R, Newman S, Bohonak A. 1999. Dispersal, gene flow, and population Feng M, Moyle S, Gorton R, Saunders T, Berry O. structure. The Quarterly Review of Biology 74: 21–45. 2017. Seascape genomics reveals fine-scale patterns of Booth T, Nix H, Busby J, Hutchinson M. 2014. BIOCLIM: dispersal for a reef fish along the ecologically divergent the first species distribution modelling package, its early coast of Northwestern Australia. Molecular Ecology 26: applications and relevance to most current MAXENT 6206–6223. studies. Diversity and Distributions 20: 1–9. Doyle J, Doyle J. 1987. A rapid DNA isolation procedure for Breiman L. 2001. Random forests. Machine Learning 45: small quantities of fresh leaf tissue. Phytochemical Bulletin 5–32. 19: 11–15. © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
1012 T. CORTEZ ET AL. Drummond A, Suchard M, Xie D, Rambaut A. 2012. genotypes, lost alleles, and systematic bias in parameter Bayesian phylogenetics with BEAUti and the BEAST 1.7. estimation. BMC Evolutionary Biology 10: 118. Molecular Biology and Evolution 29: 1969–1973. Gilpin M. 1991. The genetic effective size of a metapopulation. Earl D, vonHoldt B. 2012. STRUCTURE HARVESTER: a Biological Journal of the Linnean Society 42: 165–175. website and program for visualizing STRUCTURE output Harpending H, Sherry S, Rogers A, Stoneking M. 1993. and implementing the Evanno method. Conservation The genetic structure of ancient human populations. Current Genetics Resources 4: 359–361. Anthropology 34: 483–496. Eaton D. 2014. PyRAD: assembly of de novo RADseq loci for Hedgecock D, Launey S, Pudovkin A, Naciri Y, phylogenetic analyses. Bioinformatics 30: 1844–1849. Lapègue S, Bonhomme F. 2007. Small effective number Ekau W. 1999. Topographical and hydrographical impacts of parents (Nb) inferred for a naturally spawned cohort on zooplankton community structure in the Abrolhos Bank of juvenile European flat oysters Ostrea edulis. Marine region, East Brazil. Archive of Fishery and Marine Research Biology 150: 1173–1182. Downloaded from https://academic.oup.com/biolinnean/article/133/4/999/6272942 by guest on 10 August 2021 47: 307–320. Hedgecock D, Li G, Nelson K. 1994. The central stock of Elshire R, Glaubitz J, Sun Q, Poland J, Kawamoto K, Northern anchovy (Engraulis mordax) is not randomly Buckler E, Mitchell S. 2011. A robust, simple genotyping- mating population. California Cooperative Oceanic Fisheries by-sequencing (GBS) approach for high diversity species. Investigations Report. 35: 121–136. PLoS ONE 6: e19379. Hedgecock D, Pudovkin A. 2011. Sweepstakes reproductive Evangelisti F, Bellucci A, Sabelli B, Albano P. 2017. The success in highly fecund marine fish and shellfish: a review periwinkle Echinolittorina punctata (Mollusca: Gastropoda) and commentary. Bulletin of Marine Science 87: 971–1002. tracked the warming of the Mediterranean Sea following the Hijmans R, Phillips S, Leathwick J, Elith J, Hijmans MRJ. Last Glacial Maximum. Marine Biology 164: 34. 2017. Package ‘dismo’. Circles 9: 1–68. Excoffier L, Foll M. 2011. fastsimcoal: a continuous-time Janson K. 1987. Allozyme and shell variation in two marine coalescent simulator of genomic diversity under arbitrarily snails (Littorina, Prosobranchia) with different dispersal complex evolutionary scenarios. Bioinformatics 27: abilities. Biological Journal of the Linnean Society 30: 1332–1334. 245–256. Excoffier L, Lischer H. 2010. Arlequin suite ver 3.5: a new Je Lee H, Boulding E. 2009. Spatial and temporal population series of programs to perform population genetics analyses genetic structure of four northeastern Pacific littorinid under Linux and Windows. Molecular Ecology Resources 10: gastropods: the effect of mode of larval development on 564–567. variation at one mitochondrial and two nuclear DNA Faillettaz R, Paris C, Irisson J. 2018. Larval fish swimming markers. Molecular Ecology 18: 2165–2184. behavior alters dispersal patterns from marine protected Johannesson K, Lundberg J, André C, Nilsson P. 2004. areas in the North-Western Mediterranean Sea. Frontiers in Island isolation and habitat heterogeneity correlate with Marine Science 5: 97. DNA variation in a marine snail (Littorina saxatilis): island Ferrier S, Guisan A. 2006. Spatial modelling of biodiversity isolation and habitat heterogeneity. Biological Journal of the at the community level. Journal of Applied Ecology 43: Linnean Society 82: 377–384. 393–404. Johannesson K, Tatarenkov A. 1997. Allozyme variation in Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R. 1994. a snail (Littorina saxatilis) – Deconfounding the effects of DNA primers for amplification of mitochondrial cytochrome microhabitat and gene flow. Evolution 51: 402–409. c oxidase subunit I from diverse metazoan invertebrates. Jombart T. 2008. adegenet: a R package for the multivariate Molecular Marine Biology and Biotechnology 3: 294–299. analysis of genetic markers. Bioinformatics 24: 1403–1405. Francis R. 2017. pophelper: an R package and web app to Karatzoglou A, Smola A, Hornik K, Zeileis A. 2004. analyse and visualize population structure. Molecular Kernlab—an S4 package for kernel methods in R. Journal of Ecology Resources 17: 27–32. Statistical Software 11: 1–20. Freitas D, Muelbert J. 2004. Ichthyoplankton distribution King P, Broderip W. 1832. Description of the Cirrhipeda, and abundance off Southeastern and Southern Brazil. Conchifera and Mollusca: in a collection formed by the Brazilian Archives of Biology and Technology 47: 601–612. Officers of HMS Adventure and Beagle employed between Fu Y. 1997. Statistical tests of neutrality of mutations against the years 1826 and 1830 in surveying the southern coasts population growth, hitchhiking and background selection. of South America: including the Straits of Magalhaens Genetics 147: 915–925. and the Coast of Tierra Del Fuego. Zoological Journal 5: Funk W, McKay J, Hohenlohe P, Allendorf F. 2012. 332–349. Harnessing genomics for delineating conservation units. King TL, Eackles MS, Gjetvaj B, Hoeh WR. 1999. Trends in Ecology & Evolution 27: 489–496. Intraspecific phylogeography of Lasmigona subviridis Gajardo G, Cancino J, Navarro J. 2002. Genetic variation (Bivalvia: Unionidae): conservation implications of range and population structure in the marine snail Chorus discontinuity. Molecular Ecology 8: S65–S78. giganteus (Gastropod: Muricidae), an overexploited endemic Knutsen H, Jorde PE, André C, Stenseth NC. 2003. Fine‐ resource from Chile. Fisheries Research 55: 329–333. scaled geographical population structuring in a highly Garrick R, Sunnucks P, Dyer R. 2010. Nuclear gene mobile marine species: the Atlantic cod. Molecular Ecology phylogeography using PHASE: dealing with unresolved 12: 385–394. © 2021 The Linnean Society of London, Biological Journal of the Linnean Society, 2021, 133, 999–1015
You can also read