Understanding how mesopredators partition their diet and the identity of consumed prey can assist in understanding the ecological role predators and prey play in ecosystem trophodynamics. Here, we assessed the diet of three common coral reef mesopredators; Pseudochromis flavivertex, Pseudochromis fridmani, and Pseudochromis olivaceus from the family Pseudochromidae, commonly known as dottybacks, using a combination of i) visual stomach content analysis, ii) stomach content DNA metabarcoding (18S, COI), and iii) stable isotope analysis (δ¹⁵N, δ¹³C). In addition, P. flavivertex is found in two distinct color morphs in the Red Sea, providing an opportunity to analyze intra-morph differences. These techniques revealed partitioning in the dietary composition and resource use among species. Arthropods comprised the main dietary component of P. flavivertex (18S > 60%; COI > 10%), and P. olivaceus (18S = 57.2%) while P. fridmani ingested predominantly mollusks (18S = 51.3%, COI = 24.6%). Despite being small predators, microplastics were found in the gut content of some of these fishes. Stable isotope analysis showed differences in species' isotopic niche breadth and trophic position. Pseudochromis olivaceus presented the largest isotopic niche (SEA_C = 1.61‰²), while P. fridmani showed the smallest isotopic niche (SEA_C = 0.45‰²) among species. Although the two techniques used for stomach content analysis did not show differences in the diet within color morphs of P. flavivertex, they differed in the isotopic niche and resource use. Despite our limited sampling, our findings provide evidence of species-specific differences in the trophic ecology of dottybacks and demonstrate their important role as predators of cryptic invertebrates and small fishes. This study highlights the importance of combining several approaches (short-term: visual analysis and DNA metabarcoding; and long-term: isotope analysis) when assessing the feeding habits of coral reef fish, as they provide complementary information necessary to delimit their niches and understand the role that small mesopredators play in coral reef ecosystems.

Fish and tissue collection 

A total of 125 adult individuals from three species and two color morphs of dottybacks (Pseudochromis flavivertex M1 = 48; P. flavivertex M2 = 28; P. olivaceus = 24, and P. fridmani = 25) were collected (Figure 1) in the central Red Sea off the coast of Thuwal, Saudi Arabia (22.2833° N, 39.1000° E) between 09:00 and 12:00 h of August and September 2021. Fish collections took place in different reefs along a cross-shelf gradient (coastal, mid shore, offshore) to broadly cover all available shallow reef habitats for these species in the central Red Sea, and to maximize the diversity of dietary items detected for each dottybacks species rather than compare habitats. Not all dottybacks species were collected across the different shelves because not all species were found in the same location during the sampling period (see ESM Table 1 for individual sample details). Fishes were collected under KAUST IACUC Protocol 19IACUC03.

Fish were caught using a hand net and clove oil solution (1:10 clove oil: ethanol) as an anesthetic and euthanized by immersion in buffered Tricaine Methanesulfonate (MS-222: 250 mg/l seawater). Fish were photographed on the boat immediately after capture to document the color pattern and preserved on ice in situ. Once in the laboratory, total (TL mm) and standard length (SL mm), wet weight (mg), and sex (when possible) were recorded for each individual (ESM Table 1). The gut cavity was opened and the entire intact stomach and intestine were removed using sterile dissection tools from the fish within 3 hours after collection and preserved separately in sterile Eppendorf tubes containing 96% ethanol at -20° C for analysis. A section (~1 cm²) of dorsal muscle tissue was removed from each individual and preserved in a sterile vial at -20° C for isotope analysis. All dissecting tools were sterilized between individuals and between tissues within individuals in a three-step process by rinsing and incubating them in a series of falcon tubes containing 1:10 bleach and sterile water, sterile water, and ethanol.

Gut content DNA metabarcoding 

DNA extraction

A total of 56 gut tracts (P. flavivertex M1 = 21; P. flavivertex M2 = 13; P. olivaceus = 10, and P. fridmani = 12) were dissected, and the entire gut content (excluding tissue from the host fish to minimize dietary masking by the host DNA) extracted using the DNeasy PowerSoil DNA extraction Kit (QIAGEN) following the manufacturer’s protocols with the following modifications: samples were incubated with 12.2 µl of Proteinase K and 60 µl of Solution C1 at 56° C at 200 rpm overnight, to ensure that all the tissues were completely lysed (Leray et al., 2013) and final DNA elution was done at 50 µl. Extractions were performed under sterile conditions in a dedicated space in the Bioscience Core Laboratory at KAUST to minimize the risk of contamination. Filter tips were used, and negative controls were included in all the extraction rounds. The concentration of DNA extracts was determined using a Qubit dsDNA HS (high sensitivity, 0.2 to 100 ng) Assay Kit with a Qubit 4.0 fluorometer (Thermo Fisher Scientific) according to the manufacturer’s protocols. DNA samples were normalized to 5 ng/µl.

Library preparation and sequencing

The Illumina 16S Metagenomic Sequencing Library Preparation guide was followed (Illumina, 2019) with modifications. Two sets of universal primers were used to perform the polymerase chain reactions (PCRs), one amplifying a 313 bp segment of the mitochondrial cytochrome c oxidase subunit I (COI) gene  (mlCOIintF: 5' – GGWACWGGWTGAACWGTWTAYCCYCC – 3'; jgHCO2198: 5’–TAIACYTCIGGRTGICCRAARAAYCA –3; (Geller et al., 2013; Leray et al., 2013) and the other amplifying a 450 bp segment of the eukaryotic V4 region of the nuclear small subunit ribosomal RNA (18S rRNA) gene (Uni18SF: 5' – AGGGCAAKYCTGGTGCCAGC – 3'; Uni18SR: 5’– GRCGGTATCTRATCGYCTT – 3'; (Zhan et al., 2013). These primers were selected due to their versatility in detecting a wide range of metazoan prey items, and they work well for the identification of marine invertebrates (Casey et al., 2019; Coker et al., 2023; Geller et al., 2013; Leray et al., 2013; Rey et al., 2020; Zhan et al., 2013).

Three PCR replicates were conducted per sample. Each PCR reaction was run in a total volume of 25 µl: 1 μl of 10 μM forward primer, 1 μl of 10 μM reverse primer, 12.5 μl of KAPA HiFi HotStart DNA Polymerase (HotStart and Ready Mix formulation, KAPA Biosystems), 8 μl of RNA free water and 2.5 μl of genomic DNA. PCR was performed under the following conditions: Initial denaturation at 95° C for 3 min, followed by 35 cycles: 98° C for 20 s, 46° C (COI) or 50° C (18S) for 60 s, and 72° C for 90 s, and a final extension at 72° C for 5 min. Each PCR contained 61 samples, including 56 gut-content DNA samples, three extraction negative controls, one PCR negative control, and one PCR positive control containing genomic DNA from one fish species (Acanthopagrus berda). PCR reactions were verified on 1.0% agarose gels and QIAxcel (QIAGEN, Hilden, Germany) with the QIAxcel DNA Screening Kit, using a QX alignment marker of 15 bp – 3 kb and method AM320. Electropherograms were analyzed with BioCalculator 3.0 (QIAGEN). The PCR reaction performed with negative control extractions confirmed the absence of contaminants (no bands). All successfully replicated PCR reactions were pooled into a single product for a final volume of 75 μl.

PCR products were bead cleaned using Agencourt AMPure XP beads (Beckman Coulter, Brea, CA, USA) at a concentration of 0.8x vol/vol. For library preparation, dual indices were attached to the amplicon of each sample using Nextera DNA Library Prep Kit and the Nextera XT Index Kit (Illumina). Each indexing PCR reaction was run in a volume of 50 μl: 5 μl of Index Primer 1 (N7xx), 5 μl of Index Primer 2 (S5xx), 25 μl of 2x KAPA HiFi HotStart Ready Mix (KAPA Biosystems), 10 μl of PCR Grade water and 5 μl of PCR product. The PCR amplification included an initial denaturation at 95° C for 3 min, followed by 8 cycles of 95° C for 30 s, 55° C for 30 s, 72° C for 30 s, and a final extension at 72° C for 5 min. PCR products were cleaned and normalized using the SequalPrep Normalization Plate (96) Kit (Invitrogen).

Purified libraries were pooled with unique indices and quantified using a Bioanalyzer (Agilent) and a Qubit Fluorometer (Invitrogen) using a dsDNA HS Assay Kit. A Final concentration of the library was calculated using quantitative PCR (qPCR) and normalized to 4nM, and 20% phiX was used for internal control. Pair-end sequencing (2 x 300 bp) was performed on an Illumina MiSeq platform with a MiSeq v3 Reagent Kit (Illumina). Sequence data were automatically demultiplexed using MiSeq Reporter (V2), and forward and reverse reads were assigned to samples.

Bioinformatics pipeline

The USEARCH-UNOISE3 pipeline (Edgar, 2010, 2016) was used to process 18S rRNA and COI reads. For 18S rRNA gene-based amplicon libraries, raw reads were decontaminated of phiX, quality-filtered (QV = 25), and adapter-trimmed using the BBDuk tool from the BBMap suite (Bushnell, 2022). Primers were removed using cutadapt (Martin, 2011). USEARCH-UNOISE3 was used to conduct all further reads processing: Paired-end reads merging, deduplication, sequence alignment, denoising, chimera removal, clustering, and the generation of a zero-radius OTU (zOTU) table. Input sequences with low abundances (i.e., representing < 8 reads) were discarded by the default version of the USEARCH-UNOISE3 pipeline. Filtering of the zOTUs was performed based on BLASTn (Altschul et al., 1990) query coverage analysis (>70% query coverage), the post-clustering curation algorithm LULU (Frøslev et al., 2017), and SortMeRNA v4.3 (Kopylova et al., 2012). Finally, BLASTn and BLAST-QC (Torkian et al., 2020), of the final available zOTU sequences were performed against the latest available NCBI nucleotide database to assign taxonomy. To generate the full NCBI taxonomic lineage information for all of the identified hits, the Entrez Direct (Kans, 2010) and TaxonKit (Shen & Ren, 2021) tools were used. The taxonomy of all zOTU sequences that had >98% identity hits in the nt database was further verified against the curated EukRibo database v2 (Berney et al., 2022). All zOTUs representing bacterial rRNA genes or non-target genes were excluded from the analysis. The relative abundance of zOTU sequences in each library was estimated as a percentage of the whole library.

To process COI amplicon libraries, a similar pipeline was followed as with the 18S rRNA gene amplicons (except for the SortMeRNA filtering step). Apart from the curation of the COI zOTU sequences on LULU, the sequences were additionally filtered based on a visual screening of stop-codons and frameshift errors on SeaView (Gouy et al., 2010) after aligning the corresponding amino-acid sequences on MACSE (Ranwez et al., 2011). The taxonomy of all fish-affiliating zOTU COI sequences that had >98% identity hits in the nt database was further verified against the curated MitoFish database (Iwasaki et al., 2013) and that of the remainder sequences against the BOLD database (Ratnasingham & Hebert, 2007).

All zOTUs representing fish (Phylum- Chordata; Class- Actinopterygii) were removed from the analysis of the 18S sequences as the 18S rRNA genes representing the host fish species (Pseudochromis flavivertex, P. olivaceus, and P. fridmani) were not available in the NCBI database, and therefore could not be recognized among the other fish zOTUs present in the stomach content. COI sequences belonging to the host fish species were available in the NCBI database. Thus, all zOTUs matching those sequences were removed from the analysis whilst retaining all other zOTUs assigned to Actinopterygii. These “self-hits” (e.g., all zOTUs identified as P. flavivertex, P. olivaceus, and P. fridmani) were removed to exclude host tissue and to avoid wrong inferences of cannibalism or trophic linkages among the studied species. Though cannibalism of smaller individuals could happen, this likely represents an insignificant fraction of pseduochromid diets. Likewise, zOTUs corresponding to sequences identified as endoparasitic protozoa (e.g. Apicomplexa) were removed from the analysis as these organisms were most likely living in the host gut rather than being eaten by them. Finally, zOTUs that were identified only as Eukarya were labeled as Unidentified along with the zOTUs without matches (<85% identity match) on NCBI.