In temperate regions, one of the most critical determinants of present range-wide genetic diversity was the Pleistocene climate oscillations, the most recent one created by the last glacial maximum (LGM). This study aimed to describe Nereocystis luetkeana genetic structure across its entire range (Alaska to California) and test different models of population connectivity within the Salish Sea. This region was colonized after the LGM and has been under increased disturbance in recent decades. We utilized microsatellite markers to study N. luetkeana genetic diversity at 53 sites across its range. Using higher sampling density in the Salish Sea, we employed a seascape genetics approach and tested isolation by hydrodynamic transport and environment models. At the species distribution scale, we found four main groups of genetic co-ancestry, Alaska; Washington with Vancouver Island's outer coast and Juan de Fuca Strait; Washington's inner Salish Sea; and Oregon with California. The highest allelic richness (AR) levels were found in California, near the trailing range edge, although AR was also high in Alaska. The inner Salish Sea region had the poorest diversity across the species distribution. Nevertheless, a pattern of isolation by hydrodynamic transport and environment was supported in this region. The levels of allelic richness and genetic differentiation suggest that during the LGM, bull kelp had both northern and southern glacial refugia in Haida Gwaii and Central California, respectively. Genetic diversity in Northern California sites seems resilient to recent disturbances, whereas the low levels of genetic diversity in the inner Salish Sea are concerning.

Sampling natural populations

We sampled fifty-three sites across the geographic range of N. luetkeana, ranging from Herring Island, Alaska (59.65°N, 151.59°W) to Cambria Bay, California (35.53°N, 121.09°W), from May 2016 to August 2017. We collected a higher sampling site density inside the Salish Sea, the inner water body composed of the Strait of Georgia in British Columbia, Canada, and Puget Sound in Washington, USA (Figure 1, Table 1). Within each site, the average number of specimens collected was 40, ranging from 7 to 51. We sampled specimens haphazardly, separated by at least 2 meters, by cutting 2-4 cm pieces of blade tissue in a non-destructive manner. We wiped the sampled blade tissue to remove epiphytes before storing the tissue in silica gel desiccant for preservation until DNA extraction. Next, we used a Tissue Lyser II (Qiagen, Valencia, CA) to homogenize the silica-dried tissue to a fine powder before extracting DNA using the DNeasy Nucleospin 96 Plant Kit II (Machery-Nagel, Duren, Germany) following the kit protocol.

Microsatellite loci genotyping

We characterized microsatellite regions for N. luetkeana and used seven of the resulting microsatellite loci (Ner-2, Ner-4, Ner-6, Ner-9, Ner-11, Ner-13, and Ner-14, see article online supplementary material). We prepared PCRs in a total reaction volume of 15 µL comprised of 10 µM primer, 10 mM dNTP's per base (Promega, Madison, WI), 25 mM MgCl2, 3.0 µl 5X PCR buffer, and 0.5 U GoTaq Polymerase. Thermocycler conditions consisted of a 5-minute denaturation step at 95ºC, followed by 33 cycles of 20 seconds each at 95ºC, 20 seconds at an annealing temperature of 57ºC–61ºC, 30 seconds at 72ºC followed by a final elongation step of 20 minutes at 72ºC using an Eppendorf thermocycler (Eppendorf, USA). We sized microsatellite PCR fragments using fragment analysis on a 96-capillary DNA sequencer ABI 3730xl at the Madison Biotechnologies Center. We scored the resulting microsatellite fragments with STRand (https://www.vgl.ucdavis.edu/informatics/strand.php) and binned them into integer allele codes with the R (R Core Team, 2016) package “MsatAllele” (Alberto, 2009). The presence of null alleles was evaluated with MICRO-CHECKER v.2.2.3 (Van Oosterhout et al., 2004).

The format of the data is Genepop version 4.2 (Rousset, 2008) with 3 digits per allele.