Fig 1.
Reef sites sampled along the East Red Sea coast (Saudi Arabia).
Top left figure showing the sites and their respective region (North, Center, South). Top right figure shows the average of sea surface temperature in Celsius (SST) from May 2015 to May 2017, i.e., during most of the deployment periods. The bottom left, and bottom right figure shows the average for the same time interval for Chlorophyll-a in mg m-3 (Chla) and photosynthetic active radiation (PAR) in Einstein m-2 d-1, respectively. Maps were designed using ArcMap (Version 10.7.1.), Environmental Systems Research Institute, Inc., Redlands (esri.com) by Ute Langner. SST data was obtained from NASA Goddard Space Flight Center, Ocean Ecology Laboratory, Ocean Biology Processing Group.
Fig 2.
Alpha diversity matrix in the Red Sea basin.
A) Boxplot with confidence interval of the mean number of operational taxonomic units (OTUs) per region. B) Boxplot with confidence interval of the mean number of organisms per region. C) Means and confidence intervals of the number of individuals of Arthropoda per region. D) OTUs accumulation curves per ARMS for each region (North, Central, and South) and all ARMS pooled together (All). E) Rarefaction curves per region (north, central, and south) and with all samples pooled together. F) Means and confidence intervals of the number of OTUs per region visualized as a box plot for Arthropoda, G) for Annelida, and H) for Mollusca. I) The expected number of species in 20 individuals [ES20]. The letters a and b denote groups significantly different tested by Tukey HSD test. Data in figures A, H, and I were normally distributed and with equal variances, therefore a One-Way ANOVA was performed. Figures B, C, F, and G were analyzed using Kruskal-Wallis as a not parametric test. ARMS used in each analysis are noted in the S1 File. Images from: Sander Scheffers (hermit crab), Dieter Tracey (Polychaeta), and Tracey Saxby (goby and nudibranch), IAN Image Library (ian.umces.edu/imagelibrary).
Fig 3.
Venn diagram showing the number of operational taxonomic units (OTUs) unique to each region and shared between the regions (North, Central, and South) using a random subsample of nine ARMS per region.
The associated circle graphs show the relative composition of cryptobiome at the Phylum level calculated in terms of the relative number of OTUs per region (numbers shown in the center of each circle). The phyla Platyhelminthes, Nemertea, and Sipuncula were grouped under the category “others”.
Fig 4.
Multivariate visualizations of the distance-based redundancy analysis (dbRDA) showing differences between the community structure of cryptic organisms of the ARMS sampled in the north (N), central (C), and south (S) regions.
The axis represents the percentage of constrained variation. A) The arrows mark the contribution of the non-correlated explanatory variables to the variation observed. Sea surface temperature (SST), Chlorophyll-a concentration (Chla), and photosynthetic active radiation (PAR) were obtained from remote sensing and calculated as averages of the monthly means over five years prior to ARMS retrieval dates. Reef community variables from reef surveys at 10 m depths as in percentage cover of hard corals (HC), percentage cover of macroalgae (MA), percentage cover of soft coral (SC), percentage cover of turf algae (Turf), percentage cover of bare substrates (Abiotic; i.e., rock, sand, and dead coral) were also correlated with observed cryptic biodiversity patterns. B) The arrows mark the contribution of the OTUs that significantly aided to the differences obtained in community structure between two regions.
Table 1.
Variables driving the parent and two first nodes (split points) in the regression trees summed overall response variables.
Figures in the supporting information shows the complete tree for each variable. Turf, HC and Abiotic stands for means percentage cover of turf algae, hard coral, and not live substrate respectively. Average SST (sea surface temperature) and PAR (photosynthetic active radiation) are satellite-derived variables calculated over a period of 5 years.
Fig 5.
CART model using regression tree to observe critical breakpoints in the OTU numbers as a proxy for species richness of the cryptobiome in response to SST, Chla (and its correlated variable POC), PAR, HC, SC, Turf, and MA.
Table 2.
Summary of the key environmental variables and split values identified for which the abundance of each OTU were highest and lowest.
Percentage cover of turf algae (Turf), not living substrates (Abiotic), and hard coral (HC), and the five-year average of the remote sensing variables sea surface temperature (SST), and photosynthetic active radiation (PAR) were the key environmental variables determining split values of the dominant OTU abundance.