Trophic niches reflect compositional differences in microbiota among Caribbean sea urchins

Study site

The Northern coast of Puerto Rico is characterized by a very narrow shelf and high-energy sandy beaches, due to the action of northeast trade winds and North Atlantic winter storms. Substrate composition of sites are made up of carbonate rocks; and due to the high annual precipitation levels and the discharge of rivers, high sediment discharges are common in the north (Veve & Taggart, 1996). Surveys were conducted during February of 2019 at three shallow-water sites (1-2 m depth) of the Northeastern coast of Puerto Rico (Fig. 1), being the same depth for each species. The sites chosen for the study were Cerro Gordo located in Vega Baja (CG, 18°29′05.81″N, −66°20′20.23″W), Isla de Cabra in Cataño (IC, 18°28′26.32″N, −66°08′18.82″W), and Mar Azul in Luquillo (MA, 18°23′15.01″N, −65°43′11.26″W). Temperature (°C), salinity (‰), and pH measurements were collected in situ at each site using the quality meter instrument Pro2030 (https://www.ysi.com/pro2030). We used the average of 5 environmental samples of each parameter per site. Oceanic water conditions were recorded for the three sites, and no differences were found (Table S1).

Sample collection and processing

Six adults of the species Diadema antillarum, Echinometra lucunter, and Tripneustes ventricosus were collected at all the three sites. Additionally, three adults of Lytechinus variegatus were collected in CG and IC, but no individuals of L. variegatus were found in MA. The species E. lucunter (red urchin) and D. antillarum (black urchin) were collected associated to hardground biotopes (fringing reefs), whereas L. variegatus (green urchin) and T. ventricosus (white urchin) were collected in a back-reef lagoon biotope, covered by seagrass beds of Thalassia testudinum leaves, upon which they graze and ingest the leaves. Seawater was collected (1L) in a sterile container from the area where the sea urchins were caught in the reef biotope. Additionally, from the seagrass beds, T. testudinum samples were also collected and stored in 50 mL falcon tubes at each site. Sea urchins and seagrass bed samples were put in separate bags with seawater. All samples were put in a foam cooler at the collection site and immediately transported to the lab where they were processed. Sampling of these four species was approved by the Department of Natural and Environmental Resources of Puerto Rico permit number DRNA-2019-IC-003.

In total, 60 animals were anesthetized using 20 mM MgCl₂, a solution that is used in aquaculture as a suitable nontoxic anesthetic (Arafa, Sadok & Abed, 2007). Animals were first acclimatized in seawater in 100 mL glass beakers with 25 mL seawater for at least 10 min or until they attached as per the approved IACUC protocol A#-5301118. Once attached, an additional 25 mL sterile MgCl₂ was added. Experimentally induced anesthesia was monitored after a standardized exposure time of 15 min until all sea urchins detached from the walls of the beaker. Once detached, animals were carefully moved by hand to a tray that was placed in a −80 °C ultra-low freezer for 10 mins prior to dissection.

Using flame-sterilized scissors and a metal tray, sea urchins were dissected and opened with an equatorial cut, at the level of the maximum diameter, circumnavigating the mouth. The side along the peristomial membrane was lifted from the sea urchin, while still maintaining the integrity of the digestive tract (Whalen, 2008). The digestive tract (gut tissue, including esophagus, stomach, and intestine, were removed from the sea urchin with scissors, and transferred with a pair of tweezers to a sterilized petri dish. Gut content samples (mostly pellets with few pieces of intestinal tissue) were transferred to 2 ml microtubes and stored in the freezer at −80 °C until DNA extraction procedures. The reef water was filtered using a 0.45 µm membrane device. Membranes then were transferred to a falcon tube and stored at −80 °C for further analyses.

Genomic DNA extractions

Genomic DNA was extracted from 0.22 µM membranes of filtered seawater, sea grassss and gut fecal pellets from each of the four sea urchin species from reef water and seagrass biotopes (∼200 mg). We used the QIAGEN PowerSoil™ kit (QIAGEN LLC, Germantown Road, Maryland, USA), following manufacturer’s instructions with the following modifications: (1) gut content sample homogenization (3000 r.p.m. for 2 min at room temperature) in a PowerLyzer homogenizer (QIAGEN LLC, Germantown Road, Maryland, USA), and (2) Elution was performed using 100 µl of sterile PCR water previously warmed at 65 °C, to increase DNA yield, allowed to remain on the filter for 5 min incubation at room temperature before the final centrifugation step. A Qubit^® dsDNA HS (High Sensitivity) Assay Kit was used to assess DNA concentration (ranging from 5–100 ng/ul) of purified extracts using the Qubit^® Fluorometer at room temperature (Waltham, Massachusetts, USA).

16S-rRNA gene amplifications and Illumina sequencing

DNA from the gut content samples were normalized to 4 nM during l6S library prep. We amplified the V4 hypervariable region of the 16S ribosomal RNA marker gene (∼291 bp) using the universal bacterial primers: 515F (5′GTGCCAGCMGCCGCGGTAA3′) and 806R (5′GGACTACHVGGGTWTCTAAT3′) in the Earth Microbiome Project (http://www.earthmicrobiome.org/emp-standard-protocols/16s/) (Caporaso et al., 2012) using previously reported conditions (Abarca et al., 2018). We used the Illumina MiSeq Reagent kit 2 ×250 bp to sequence the 16S amplicons. The 16S-rRNA reads were deposited in QIITA (Gonzalez et al., 2018) Bioproject ID 12668, and the raw sequences are available in the European Nucleotide Archive ENA Project: PRJEB40117; ERP123720.

Bioinformatic analyses and statistical tests

Alpha diversity and taxonomic plots

Alpha diversity measures of Chao 1 (richness), were plotted as rarefaction curves and boxplots. For alpha diversity statistical tests, we used the script compare_alpha_diversity.py in QIIME, to compare the diversity between groups of samples in each metadata category. These statistical tests were nonparametric t-tests with Monte Carlo permutations to determine the p-value. We considered a rarefaction level of 6,700 reads for all the 60 samples together including water and seagrass, while for analyses of gut samples (n = 54) the rarefaction level was 7,000 reads. Barplots revealing phyla were computed using QIIME (Caporaso et al., 2010) and those at the genus-level with MicrobiomeAnalyst (https://www.microbiomeanalyst.ca/) using the same parameters.

Additionally, we used the group_significance.py script in QIIME, which compares OTU frequencies across animal species, to ascertain whether or not there are statistically significant differences between the OTU abundance in the different groups (using a Kruskal-Wallis test) S elected boxplots of significantly different taxa , were generated using ggplot2 (Wickham, 2009) (Team, 2020) and taxa with p-values <0.05 were marked with an asterisk. The core microbiome was calculated in QIIME and accounted for those OTUs shared by 50% of the samples across the four species. The list of OTUs was added to a web-based tool for Venn diagrams and plotted using http://www.interactivenn.net/index.html (Heberle et al., 2015).

All statistical test results are summarized in (Table S2).

Echinoid-associated microbiota is distinct from environmental samples

Environmental (water and seagrass samples) separate clearly from host-associated sea urchin digesta (permanova p-value = 0.001; ANOSIM, p-value = 0.01; Fig. 2A, Table S2). Alpha diversity was significantly different between seagrass and echinoid samples (t-test, p-value = 0.003), similarly to reef water and gut samples (t-test, p-value = 0.003, Fig. 2B), but no remarkable differences were found between reef water and seagrass samples (t-test, p-value = 0.336). Composition was similar between reef water and seagrass in terms of the relative abundance of Proteobacteria and Bacteroidetes; while Euryarchaeota was only found in water samples and Cyanobacteria was more abundant in seagrass samples. Seagrass samples were dominated by the cyanobacteria Rivularia, while in contrast water samples were dominated by various groups such as Rhodopirellula, Paramoritella, NS5 marine group or Prochlorococcus (Fig. 2C). Gut samples had dominant Prolixibacter, Propionigenium, Photobacterium and Desulfotalea; while water samples were dominated by Paramoritella, Blastopirellula, Prochlorococcus, NS5 marine group, and other uncultured bacteria, while seagrass samples were dominated by Rivularia (Fig. 2D).

Echinoid collection sites and species explain differences in the microbiota L. variegatus, an herbivorous echinoid, has the most distinct microbiota

Differences between community structure were detected among sea urchin species (PERMANOVA p-value = 0.004; ANOSIM, p-value = 0.048). Alpha diversity was significantly higher among L. variegatus compared to E. lucunter (t-test, p-value = 0.006) and between L. variegatus and D. antillarum and (t-test, p-value = 0.012; Fig. 3B, Table S2). The relative abundance of the microbiota at the phyla-level showed higher dominance of Planctomycetes, Actinobacteria and Cyanobacteria in L. variegatus compared to the other species. There were similar amounts of Bacteroidetes, Proteobacteria and Fusobacteria among all species (Fig. 3C). The most abundant bacterial genera were Prolixibacter, Propionigenium and Photobacterium, but in some samples of D. antillarum and T. ventricosus, Kistimonas were particularly abundant (Fig. 3D).

Community analyses by animal collection site (location), showed significant differences among the three sites, PERMANOVA p-value =0.001; ANOSIM, p-value = 0.01). Microbiota from animals collected in Luquillo clustered together in axis 1, while axis 2 separated mostly Cerro Gordo from Cataño (Fig. 4A, Table S2). Alpha diversity analyses showed less richness in Luquillo samples, nevertheless, there were no significant differences among sites (Fig. 4B, Table S2). At the Phyla level within all sites, dominant groups included Bacteroidetes, Fusobacteria, and Proteobacteria. Fusobacteria were reduced in samples collected in Cerro Gordo. Bacteroidetes increased in Cerro Gordo, which separated in the NMDS from the other sites (Figs. 4A, 4C). The Phyla Cyanobacteria and Bacteroidetes were more abundant in Cataño (where L. variegatus was collected). All species collected from Cataño or Cerro Gordo presented higher richness on their samples indifferently of the sea urchin species Fig. 4D). At the genus level, we found that Prolixibacter (Bacteroidetes), Propionigenium (Fusobacterium), Photobacterium and Vibrio (Proteobacteria) were present across all samples (Fig. 2E). The genus Prolixibacter was more abundant in Echinometra lucunter samples from Cataño and Luquillo (Fig. 4F), and Photobacterium were more abundant in IC and MA (Fig. 4F). Only 161 OTUs were considered core among all four echinoids, when computing the core microbiota of OTUs present in 50% of the samples Fig. S1).

Reef species D. antillarum (black echinoid) and E. lucunter (red echinoid) have a distinct microbiome from L. variegatus (green echinoid) and T. ventricosus (white echinoid) occupying the seagrass niche

We then determined the taxonomic biomarkers significantly associated (p < 0.05) with each of the four sea urchin species. D. antillarum had Sedimitomix, Ferrimonas, and Desulfotalea. Another species collected in the reef niche was E. lucunter, dominated by Thalassospira and Vibrio and shared dominant Prolixibacter and Photobacterium with D. antillarum (Fig. 5). The species T. ventricosus exhibited higher relative abundance of Propionigenium with respect to the other three sea urchins, but not significantly. L. variegatus, also collected in the seagrass bed, had a significantly higher dominance of Pleurocapsa, Planctomyces, Rhodopirellula, Pelagibius, and Blastopirellula (Fig. 5) and the highest number of unique OTUs, when considering a core microbiome of 50% (Fig. S1). Interestingly, its core microbiota shared a similar number of OTUs with D. antillarum and E. lucunter (Fig. S1). Certain similarities in species composition among the species collected in the different niches (seagrass and reef) determined significant differences in beta diversity (PERMANOVA p-value = 0.007; ANOSIM, p-value = 0.01; Fig. 6A, Table S2) but no differences in alpha diversity (Fig. 6B). It comes with low R values associated with ANOSIM, confirming the low dissimilarity (some overlap) between the groups of samples. Samples from seagrass biotopes displayed higher abundances of the phyla Fusobacteria, Lentisphaerae, and Planctomycetes, while having lower concentrations of Verrucomicrobia. In contrast, Bacteroidetes was more abundant in sea urchin species from reef niche (Fig. 6C). The most abundant genera were Prolixibacter in the reef niche, Propionigenium and Photobacterium at both trophic niches, but mostly in the seagrass bed (Fig. 6D).