Materials and Methods
All sea stars were collected between April and October 2016 across six
sites in central Oregon. These included two northern sites
~0.7 km apart near Depoe Bay, at Fogarty Creek on 20 Oct
2016 (44.8386, -124.0588) and Boiler Bay on 23 May 2016 (44.8303,
-124.0608). The two central sites sampled were Smelt Sands (44.3212,
-124.1081, on 16 Aug 2016) and Yachats Beach (44.3114, -124.1086, on 8
May 2016); theses were 55 km southward from Boiler Bay, and 0.2 km apart
from one another. The two southern sites were 7.4 km southward on Cape
Perpetua and were 4.7 km from one another; these were Strawberry Hill on
1 Jul 2016 (44.2492, -124.1154) and Tokatee Klootchman on 24 Apr 2016
(44.2037, -124.1170).
Animals were collected by hand at low tide. Arm length were recorded
(center to longest arm) for each animal, and only adults (with
>3cm arm length) were scored for disease status and had
tissue sampled. We recorded disease symptoms as per Menge et al. (2016);
these included, in order of severity: lesions, arm loss, twisted arms,
deflated (loss of turgor pressure), and disintegrating/dying. Animals
were considered healthy if none of these symptoms nor any injuries
existed (e.g. regrowing arms). Tube feet (~5-10) were
collected from each animal using scrubbed and sterilized forceps, then
stored in 1.5mL microcentrifuge tubes containing 1 mL of 95% ethanol.
All samples were stored on ice and then at –20°C until ready for DNA
isolation.
For genotyping, we included only individuals with the highest
symptomatic scores in order to avoid misdiagnosis, as well to include
only individuals that were truly not resistant to SSWS (i.e. not in the
process of recovery after mild disease). In total, 82 individuals were
included in the symptomatic group and 112 in the asymptomatic group, but
these proportions ranged across sites (Table 1). DNA was extracted using
the E.Z.N.A Tissue DNA kit (Omega Biotek, Norcross, GA) and quantified
using a fluorescence method (Quant-iT dsDNA Assay Kit, ThermoFisher,
Waltham, MA). We used the 2bRAD protocol for genotyping SNPs (Wang et
al. 2012), following the original published protocol, but using the
enzyme AlfI. We also used adaptors with ”NN” overhangs to target 100%
of restriction sites. Multiplexed individuals were pooled at
approximately equimolar amounts (after quantification via quantitative
PCR) and sequenced across five lanes of an Illumina HiSeq 3000 as 50-bp
single reads.
Adaptors were trimmed and low-quality reads were filtered (<30
phred) using publicly available scripts
(https://github.com/Eli-Meyer/2brad_utilities/). Cleaned reads
were mapped to the reference P. ochraceus genome (NCBI;
GCA_010994315.1)
using SHRiMP (Rumble et al., 2009), reporting the top three maximum hits
per read. We used Stacks v.1.35 (Catchen, Hohenlohe, Bassham, Amores, &
Cresko, 2013; Catchen, Amores, Hohenlohe, Cresko, & Postlethwait, 2011)
to call genotypes using default parameters, with samples coded by
disease status group (symptomatic/asymptomatic). Additional filtering
parameters used included: selection of a single (first) SNP per stack,
removal of loci that were genotyped in only one group, the minimum minor
allele frequency (MAF) set to 0.025, and a minimum minor allele count
set to four. Finally, the percentage of individuals needed to process a
locus were adjusted for each population so that a minimum of eight
samples were genotyped in a group. In VCFtools (Danecek et al., 2011),
retained only biallelic SNPs and loci with minimum depth of 6.