Harmful algal blooms (HABs), which can be lethal in marine species and cause illness in humans, are increasing worldwide. In the Gulf of Mexico, HABs of Karenia brevis produce neurotoxic brevetoxins that cause large-scale marine mortality events. The long history of such blooms, combined with the potentially severe effects of exposure, may have produced a strong selective pressure for evolved resistance. Advances in next-generation sequencing, in particular genotyping-by-sequencing, greatly enable the genomic study of such adaptation in natural populations. We used restriction site-associated DNA (RAD) sequencing to investigate brevetoxicosis resistance in common bottlenose dolphins (Tursiops truncatus). To improve our understanding of the epidemiology and aetiology of brevetoxicosis and the potential for evolved resistance in an upper trophic level predator, we sequenced pools of genomic DNA from dolphins sampled from both coastal and estuarine populations in Florida and during multiple HAB-associated mortality events. We sequenced 129 594 RAD loci and analysed 7431 single nucleotide polymorphisms (SNPs). The allele frequencies of many of these polymorphic loci differed significantly between live and dead dolphins. Some loci associated with survival showed patterns suggesting a common genetic-based mechanism of resistance to brevetoxins in bottlenose dolphins along the Gulf coast of Florida, but others suggested regionally specific mechanisms of resistance or reflected differences among HABs. We identified candidate genes that may be the evolutionary target for brevetoxin resistance by searching the dolphin genome for genes adjacent to survival-associated SNPs.
Stacks analysis of common bottlenose dolphin RAD sequences
Results of Stacks analysis of RAD sequencing data generated from pooled genomic DNA from common bottlenose dolphins sampled along the Florida Gulf of Mexico coastline. Sample pools include live coastal dolphins from the Florida Panhandle (PLC, N=17) and central-west Florida (CWLC, N=26) and live estuarine dolphins from St. Joseph Bay in the Panhandle (PLE, N=12) and Sarasota Bay in central-west Florida (CWLE, N=25). Additional sample pools include samples from bottlenose dolphin strandings (dead samples) during unusual mortality events (UMEs) associated with harmful algal blooms (HABs) in the Panhandle in 1999 (PU99, N=16) and 2004 (PU04, N=35) and during HABs between 1992 and 2006, including a UME in 2005-2006, in central-west Florida (CWU, N=25). Two groups of samples (PLC and CWU) were divided for sequencing into pools of excellent and good quality genomic DNA. PL represents samples from the Florida Panhandle that cannot be definitively attributed to either coastal or estuarine dolphin populations.
Stacks parameters are described in the Methods section of Cammen et al. (2015). Chromosome accession numbers refer to scaffolds in the bottlenose dolphin genome: Ttru_1.4/turTru2 (GCA_000151865.2). The spreadsheet includes both the alleles detected at each RAD locus as well as the depth of read coverage within the sample pool for the given alleles. Analyses described in the manuscript were conducted on a subset of these RAD loci that passed additional filters: 1 SNP with a sample-wide minor allele frequency greater than 0.10; present in at least 2 sample pools.
Stacks_results.csv
RAD locus allele frequencies
Allele frequencies for 7,431 polymorphic (single SNP) RAD loci that were assessed for an association with bottlenose dolphin survival following exposure to harmful algal blooms. Allele frequencies for each sample pool were calculated based on the proportion of reads sequenced for each SNP variant (see “Stacks analysis of RAD data” Dryad file). Depth indicates the total number of reads per locus and sample pool; sample size indicates the number of individuals sequenced in the sample pool (note this varies for PLC and CWU because individuals of good and excellent quality DNA were separated during library preparation).
RADallelefreq.csv
Fisher’s Exact Test for differences in allele frequency between live and dead bottlenose dolphins
Results (p values) of Fisher’s Exact Test for differences in allele frequency between pools of live and dead dolphins. Sample pool abbreviations as in description for “Stacks analysis of common bottlenose dolphin RAD sequences.” Fisher’s exact tests, which were implemented in R, use a contingency table approach to assess deviations from a null hypothesis of no association between genotype and phenotype for each locus independently.
FishersExactTest.csv
BayeScan analyses of comparisons of live and dead bottlenose dolphins
Results of BayeScan analyses of comparisons of RADseq allele frequencies between live and dead dolphins. Sample pool abbreviations as in description for “Stacks analysis of common bottlenose dolphin RAD sequences.” BayeScan (v2.1) implements a Bayesian approach to assess evidence for selection at a given locus by comparing population- and locus-specific components of FST across multiple loci (Foll and Gaggioti 2008).
BayeScan_results.csv
Smoothed pairwise genetic differentiation (FST) and overall nucleotide diversity (pi)
Smoothed pairwise genetic differentiation (FST) and overall nucleotide diversity (pi) across bottlenose dolphin RAD loci. Sample pool abbreviations as in description for “Stacks analysis of common bottlenose dolphin RAD sequences.” FST was calculated from nucleotide diversity at each population and across the pooled populations, where nucleotide diversity was calculated using allele counts (see equation 2 in Hohenlohe et al. 2010). Accession numbers refer to scaffolds in the killer whale genome (Oorc_1.1, GCA_000331955.2). Smoothed average FST was calculated using a Gaussian function with theta of 150kb and windows truncated at 3σ from centre in both directions
smoothFst&Pi.csv
