Underwater observation of Acropora tenuis
We observed spawning of the scleractinian coral A. tenuis (n=11) on a section of the fringing reef of Sesoko Island, Okinawa, Japan (26°37’43.9”N 127°51’43.3”E), by snorkeling or SCUBA diving for six nights from May 26 to May 31, 2018. Floats were released from 4 of the 11 tagged colonies (ten15, ten21, ten27, and ten31). The area we where we monitored the floats measured approximately 1 × 1 km2.
Sperm concentration in situ
To trace gamete bundles released from the tagged colonies, floats with fluorescent light bars (Fig. 1 A) were released directly above the spawning colonies approximately 10 min after the colonies started spawning. We followed the floats by kayak, and collected 1 L seawater near each float at 9 min intervals for approximately 1 h after the release of the floats (Fig. 1 B). The collected seawater was brought back to Sesoko Station within 30 min, and the sperm concentrations were measured with a Thoma hemocytometer according to a previous study.12 The sperm in the collected seawater was measured five times in 200 × 200 µm. When no sperm was found, the concentration was described as being below 2.5×104 sperm/mL. It took about 1 h from collecting the seawater to the start of sperm concentration measurement, and by this time most of the eggs in the seawater had already been fertilized.
Genetic diversity of spermatozoa in seawater and fertilized eggs
The genotypes of sperm and fertilized eggs were determined using microsatellite markers developed for Acropora.24 Eggs in the collected seawater were transferred to fresh 0.22 µm filtered seawater, and fertilized eggs that had completed embryogenesis were fixed with 99.5% ethanol 3 days after collection. The remaining seawater was filtered and sperm were trapped on a membrane (mixed cellulose ester gridded at 0.45 μm, Merck Millipore, MA, USA). The membranes were soaked with 1 mL CHAOS solution (4 M guanidine thiocyanate, 0.1% [v/v] N-lauryl sarcosine sodium salt, 0.1 M β-mercaptoethanol, 10 mM Tris-HCl pH 8.0) and the DNA was extracted using a Wizard SV genomic DNA purification system (Promega, WI, USA). DNA was extracted from the fertilized eggs following a previous study with some modifications.12 Fertilized eggs were kept in filtered seawater for 2 days and fixed in 99.5% EtOH. The fixed larvae were treated with 20 μL f lysis buffer (100 mM NaCl, 10 mM Tris-HCl [pH 8.0], 0.3% [w/v] Triton X-100, 0.3% [w/v] Tween 20) containing 1 g/mL proteinase K for 2.5 h at 55°C and heated at 95°C for 5 min. The supernatant was used for PCR reactions for genotyping. As a negative control for sperm detection in seawater, seawater was collected in the daytime following the same protocol. We sampled eight times at 9 min intervals after a float was released.
The sperm in seawater and fertilized eggs were genotyped with seven microsatellite markers with FAM or BIC (12406m3, 11543m5, 11401m4, 441m6, 11292m4, 10366m5, and 12130m5; Supplementary Information 124). The allele diversity of these seven markers was marked and the fragment amplification was stable. Subsequently, we used the seven markers to verify the presence of the alleles. Fragments were analyzed with a DNA sequencer (Applied Biosystems 3730xl or 3130xl) with GeneScan 500 LIZ dye size standard (Thermo Fisher, MA, USA). Peaks were measured with Microsatellite Analysis v1.0 software (Applied Biosystems by Thermo Fisher, MA, USA). For analyses of seawater containing sperm, peaks below 100 were excluded as alleles (Supplementary Fig. 1). The ratio of each peak in the samples (Rx) was calculated as Rx1 = (Hx1/H1 + H2 + … + Hx). Here, Hx1 indicates the height of one peak among the others, and H1 +… Hx is the sum of the heights of all peaks. In the negative seawater control, marker 12406m3 was used and three alleles (176, 179, and 185) were detected in one of the seven samples. These alleles were not detected in sperm in this study (153–174). Thus, the alleles detected in sperm from seawater were treated as those from the released sperm.
The Welch t-test was used to examine the differences in allele ratios. A GLMM was performed with the glmmML package in R ver. 3.1.25 Each allele was treated as a random effect and the binominal distribution was used as a family.