A fecal sample was isolated from the hindgut of each of 76 fish (n = 6–14 individuals per species; 9 species; see Table S1 for replication) so that fecal material examined had passed through the entirety of the fish digestive tract. We measured the volume and mass of each fecal sample and then examined it for the presence of live Symbiodiniaceae cells (n = 12 each of sediment and water samples were also analyzed). For this, we used a Neubauer hemocytometer and a light microscope, and samples were stained with Trypan Blue to identify Symbiodiniaceae cells that were live at the time of preservation (Figure S1; Strychar and Coates, 2004; Zhang et al., 2008). We detected live Symbiodiniaceae cells in all obligate corallivore feces (n = 40, across four species) and in ~ 82% of facultative corallivore feces (n = 18 of 22, across three species). Live Symbiodiniaceae cells were detected in ~ 36% of grazer/detritivore feces (n = 5 of 14, across two species), ~ 42% of sediment samples (n = 5 of 12) and ~ 8% of seawater samples (n = 1 of 12).
Live Symbiodiniaceae cell densities ml− 1 sample differed among species and environmental samples (Fig. 2; Overall Kruskal-Wallis test results: chi-squared = 85.2132, df = 10, p-value = 0), with higher mean live cell densities observed in the feces of each obligate corallivore species than in feces of grazer/detritivore species and sediment and water samples (Fig. 2; Table S2). Mean live Symbiodiniaceae cell concentrations in individual obligate corallivore species (Fig. 2) ranged from 7.96·105 ± 5.1·105 (Amanses scopas) to 7.6·106 ± 2.5·106 (Chaetodon reticulatus) ml− 1 of feces and were 4–5 orders of magnitude higher than sediment samples (22.75 ± 22.37) and 6–7 orders of magnitude higher than seawater samples (0.23 ± 0.51). Mean live Symbiodiniaceae cell densities in feces of facultative corallivore species were more variable than obligate corallivores and ranged from 2.1·104 ± 2.8·104 (Chlorurus spilurus) to 2.1·106 ± 1.9·106 (Chaetodon pelewensis) ml− 1 feces. Overall, live Symbiodiniaceae cell densities in corallivore feces in this study were 1–3 orders of magnitude higher than values previously reported from the corallivorous Carribean parrotfish Sparisoma viride (live and dead cells combined: 3.2·103-8.9·103 cells ml− 1; Castro-Sanguino and Sánchez, 2012), including the most comparable species in our study, Chlorurus spilurus, which is also a facultatively corallivorous parrotfish. The higher cell densities observed in this study are likely due to dietary differences among fishes. The corallivores examined here generally rely on corals for a larger part of their diets (Harmelin-Vivien and Bouchon-Navaro, 1983, but note that no data are available for C. spilurus) than S. viride (Castro-Sanguino and Sánchez, 2012), and are therefore expected to ingest higher abundances of Symbiodiniaceae cells.
The long-term viability of Symbiodiniaceae cells in feces was further confirmed through culturing (attempted from 61 fish fecal pellets using 24 replicate wells per sample; 1464 wells total). Six to ten weeks after cultures were instigated, live Symbiodiniaceae cells in coccoid, mitotic and motile life stages were documented in at least one replicate culture well from 54% of obligate corallivore feces (n = 19 of 35 samples), 11% of facultative corallivore feces (n = 2 of 18 samples), and 38% of grazer/detritivore feces (n = 3 of 8 samples) indicating that the feces of various reef fish consumers contain Symbiodiniaceae cells that are competent for division and long-term survival.
To examine whether Symbiodiniaceae communities in corallivore feces constitute a potential source of symbiont cells for uptake by local corals, we characterized community compositions of Symbiodiniaceae in all our samples (fish feces, sediment and water) as well as in three common coral species (Acropora hyacinthus, Pocillopora species complex, and Porites lobata species complex; n = 12 per coral species; Table S1) by sequencing the internal transcribed spacer-2 (ITS-2) region of Symbiodiniaceae rDNA. The coral species examined here harbor unique Symbiodiniaceae communities (Putnam et al., 2012; Rouzé et al., 2019) and are members of coral genera frequently targeted by the corallivores in this study (Pratchett, 2014). At the genus level, symbiont communities differed among coral species, obligate corallivore feces, grazer/detritivore feces, and facultative corallivore feces, reef sediments and seawater (PERMANOVA: df = 6, F = 17.3, R2 = 0.58, p = 0.001; Fig. 3, Figure S2). Symbiodiniaceae communities in obligate corallivore feces overall most closely resembled symbiont compositions in the Pocillopora species complex and Porites lobata species complex corals (Table S3 and S4). On average, sequenced reads from obligate corallivore feces were dominated by similarities to the genus Cladocopium (72–98% of reads), with low relative abundances of reads identified as the genus Durusdinium (2–27% of reads) and Symbiodinium (0–1% of reads). These findings are consistent with observations (this study and Pratchett, 2014) that corallivores at our study sites mainly feed on pocilloporid and poritid corals. Facultative corallivore feces were similar to sediment and seawater samples (Table S3 and S4); these feces contained, on average, less Cladocopium (26–81%), but more Durusdinium (11–73%) and Symbiodinium (0–8%) than obligate corallivores (Fig. 3). Grazer/detritivore feces were distinct from all other sample categories based on pairwise PERMANOVA test results (Table S4). The Symbiodiniaceae genera Breviolum and Fugacium, which are rare in Mo’orean corals (Putnam et al., 2012), were detected only in grazer/detritivore feces (Fugacium only), reef-associated seawater (Breviolum only) and sediment samples (Fig. 3, S2; Table S3).
Relative abundance of hits to a given Symbiodiniaceae genus in corals, obligate corallivore feces, facultative corallivore feces, grazer/detritivore feces, and reef-associated sediment and water (all included samples had > 1,000 reads). Different letters indicate significant differences in Symbiodiniaceae community compositions based on pairwise PERMANOVA tests (p < 0.05, using the Benjamini-Hochberg correction for multiple comparisons) using Bray-Curtis distances based on randomly subsampled (n = 12) untransformed data (Table S1, S4). Overall PERMANOVA test results: df = 6, F = 17.3, R2 = 0.58, p = 0.001. Included samples (and sample sizes) for each sample category in the figure are as follows: Corals: Acropora hyacinthus, ACR (11); Pocillopora species complex, POC (12); Porites lobata species complex, POR (12). Obligate corallivores: Amanses scopas, AMSC (7); Chaetodon lunulatus, CHLU (8); Chaetodon ornatissimus, CHOR (14); Chaetodon reticulatus, CHRE (11). Facultative corallivores: Chaetodon pelewensis, CHPE (8); Chaetodon citrinellus, CHCI (6); and Chlorurus spilurus CHSP (8). Grazer/Detritivores: Ctenochaetus flavicauda, CTFL (6); and Ctenochaetus striatus, CTST (6). Sediment and water: Sediment, SED (12); Water, WAT (7). For data used, see Additional file 2.
To generate the first estimates of the daily dispersal of live Symbiodiniaceae cells by corallivorous fish at the reef scale, we applied a bootstrap approach to the product of observed egestion rates, fecal pellet sizes (from in situ field observations), fish densities (Brooks, 2018), and live Symbiodiniaceae cell densities and fecal pellet densities (from ex situ measurements) for two obligate corallivores (C. ornatissimus and C. reticulatus) and one facultative corallivore (C. citrinellus). This resulted in a probability distribution describing the likely number of live cells dispersed by each species per 100 m2 per day (Fig. 4). The mean (± 95% CI) estimated dispersal rates for the obligate corallivores were 1.01·108 ± 6.5·106 and 1.27·108 ± 4.61·106 cells per 100 m2 d− 1, respectively; these were three orders of magnitude higher than the estimated mean for the facultative corallivore (3.32·105 ± 6.21·104). Differences between obligate versus facultative corallivore estimates were mainly driven by higher densities of C. ornatissimus and C. reticulatus individuals at our study site (4.7–8.8 times higher than C. citrinellus, Table S5), and higher Symbiodiniaceae densities in their feces (3.7–7.5 times higher than C. citrinellus, Table S5, Fig. 2). Further, mean fecal pellet size, fecal density and egestion rates were also 1.1–4.5 times higher for the two obligate corallivores, relative to the facultative corallivore (Table S5).
We estimated the number of live Symbiodiniaceae cells dispersed by the obligate corallivores Chaetodon ornatissimus (CHOR) and Chaetodon reticulatus (CHRE) and the facultative corallivore Ctenochaetus citrinellus (CHCI) per 100 m2 per day by applying a bootstrap approach (1000 iterations) to the equation T=gSWCF. The estimated number of dispersed Symbiodiniaceae cells (T) is the product of five variables: A fish species-specific constant representing the estimated number of egestions in an eight-hour day (g); observed fecal pellet sizes in cm (S); measured densities of fecal samples in g cm−1 (W); measured densities of Symbiodiniaceae cells g−1 feces (C); and observed fish densities per 100 m2 (F). Due to variation in fish species distributions, most data for CHOR and CHRE were collected on the fore reef, whereas data for CHCI were collected on the back reef. See Table S5 for replication. For data used see Additional file 2.
This work demonstrates that obligate corallivore feces contain live Symbiodiniaceae cells at densities two to seven orders of magnitude higher than other environmental reservoirs, such as the water column, sediments and macroalgae (this work; Fujise et al., in review; Littman et al., 2008), and up to three orders of magnitude higher than feces of the one other corallivorous fish that has been examined: the Caribbean parrotfish S. viride (Castro-Sanguino and Sánchez, 2012). Corallivore feces that come in direct contact with live coral colonies are likely to facilitate the transmission of Symbiodiniaceae cells between colonies. We therefore conducted surveys to estimate the relative frequency at which corallivore feces fell onto such colonies. We observed that egested feces often fell apart into several segments as they fell through the water column, and segments of 91% of egested feces landed on live corals at our fore reef site (n = 20 of 22 feces). Segments of only 10% of feces fell onto live corals at our back reef site (n = 3 of 30 feces), which is consistent with lower mean (± SD) live coral cover in this reef zone (44.8% ± 10.8 coral cover on the fore reef; 15.3% ± 7.10 back reef, ANOVA, df = 1, F = 44.91, p < 0.001). These results indicate that fish predators commonly mediate the dispersal of Symbiodiniaceae cells to prospective coral host colonies on Pacific reefs, especially in areas with relatively high coral cover.
Our findings suggest that the feces of obligate corallivores and at least some facultative corallivores constitute significant but underexplored environmental ‘hotspots’ of Symbiodiniaceae on coral reefs; such feces may supply Symbiodiniaceae cells to potential hosts directly, or to other environmental reservoirs as they disintegrate (Castro-Sanguino and Sánchez, 2012; Nitschke et al., 2016). Corals have been shown to take up Symbiodiniaceae from sediments and seawater (Coffroth et al., 2006; Lewis and Coffroth, 2004; Nitschke et al., 2016). Thus, corals are likely capable of incorporating Symbiodiniaceae from fecal material as well. Although experimental validation for stony corals is needed, this has already been demonstrated for the sea anemone Aiptasia pulchella (Muller Parker, 1984). Taken together, our findings broadly suggest that some predators have indirect positive effects on prey health; such predators supply prey with microorganisms that support their function.
The dispersal of beneficial microorganisms may become increasingly important (Foo et al., 2017; Rebollar et al., 2016) as anthropogenic stressors disrupt animal and plant microbiomes, leading to disease and mortality (Hughes et al., 2017; Jiménez and Sommer, 2017; Petton et al., 2015; Wilkinson, 2008). Our results suggest that shifts and declines in fish communities due to overfishing (Bellwood and Choat, 2011; Hawkins et al., 2007; McClanahan et al., 1999) and habitat degradation (Pratchett et al., 2006; Viviani et al., 2019) may contribute to an unexplored issue on reefs: altered (or reduced) dispersal of Symbiodiniaceae, a key member of the coral microbiome. Coral-Symbiodiniaceae partnerships have been increasingly disrupted over the past four decades, resulting in coral reef decline (Hughes et al., 2017; Wilkinson, 2008). To help corals tolerate stress and mitigate reef degradation (Peixoto et al., 2017), probiotic solutions of beneficial Symbiodiniaceae (Morgans et al., 2020) and bacteria (Rosado et al., 2019) are currently being developed. We show that corallivores egest feces containing high densities of live Symbiodiniaceae cells directly onto corals. This behavior may constitute a routine, global-scale ‘restoration effort’ that inoculates corals with natural probiotics derived from nearby colonies. It is thus important that healthy fish assemblages are maintained on reefs as the potentially stabilizing effect of corallivores on coral microbiomes is investigated.