Introduction

Large-bodied consumers often exert disproportionately strong impacts on the ecology and chemistry of the ecosystems they inhabit because they frequently consume large volumes of food and redistribute much of this intake in new forms and in new locations (Owen-Smith 1988; Bonaldo et al. 2014; Roman et al. 2014; Estes et al. 2016). The influence of these large consumers on material transfer is amplified in cases where they meet their energy demands via the intake of large volumes of low-quality foods. African elephants, for example, consume > 5 tons yr−1 of well-defended foods (e.g., grasses and trees) and subsequently excrete the majority of this low-quality biomass intake. This high volume consumption and excretion have a defining influence on the morphology and biodiversity of African savannas (Jones et al. 1994; McCauley et al. 2006; Pringle et al. 2007; Asner 2009). Understanding how high-throughput mega-consumers select among available food sources and what the biochemical implications are of the large quantities of waste they generate can improve our view of system chemistry and ecosystem ecology.

Here we report upon the biogeochemical characteristics of the intake and fecal production of an understudied, but potentially important marine mega-consumer: Bolbometopon muricatum, the bumphead parrotfish (Fig. 1). Reaching 1.4 m in total length, B. muricatum is the largest parrotfish in the world and among the largest coral reef fishes (Choat et al. 2006). B. muricatum was once considered abundant or common throughout tropical and subtropical reefs in the Pacific and Indian Oceans, but this widespread species is now considered uncommon or rare in much of that range (Donaldson and Dulvy 2004). B. muricatum feed principally on living corals and recolonized coral substrate—food items that, in respect to mass, are mostly composed of indigestible coral-derived calcium carbonates (McCauley et al. 2014). Because of this especially low-quality diet, B. muricatum must consume very high volumes of these foods. It is estimated that a B. muricatum individual takes in between 4.48 and 5.69 tons of reef material annually (Bellwood et al. 2003; McCauley et al. 2014; Hoey and Bellwood 2008). While foraging on coral rock material (i.e., reef bench lacking living coral but colonized by a diverse assemblage of other biota) is a common feeding mode in many parrotfish and acanthurids (Bellwood and Choat 1990; Choat et al. 2002), the heavy reliance of B. muricatum on live corals is relatively unique (Rotjan and Lewis 2006; Bonaldo et al. 2014). It is not known whether the inclusion of this live coral represents a shift toward increased optimality in resource acquisition. Understanding why this diet partitioning occurs is especially important in the case of B. muricatum, as intense predation of live corals by B. muricatum may alter patterns of coral abundance and diversity on reefs (McCauley et al. 2014). Likewise, the consumption of coral substrate without live coral polyps present (e.g., coral pavement, rubble) may serve as a major agent of bioerosion and can influence the physical structure of reef environments (Bellwood et al. 2003; Perry et al. 2015).

Fig. 1
figure 1

Bolbometopon muricatum pictured here a delivering a bite upon a living scleractinian coral colony and b having just released feces, both at Palmyra Atoll. Photograph credit D. McCauley

Full size image

The coral forage consumed by B. muricatum is pulverized in their pharyngeal mill, reducing this often structurally complex intake into fine particulate matter. As so little of this consumed material can be assimilated, B. muricatum excretes an estimated 4.8 tons of feces, largely as inorganic calcium carbonate particulate, back upon reefs annually (McCauley et al. 2014; Perry et al. 2015). Feces from other fish and vertebrates have been shown to have important effects on the biochemistry and ecology of ecosystems in a broad variety of contexts (Bray et al. 1981; Meyer and Schultz 1985; Young et al. 2010). The high fraction of inorganic calcium carbonate generally found in parrotfish feces differentiates it from other reef fish feces; however, parrotfish feces has been found to contain measurable volumes of lipids and proteins and to host diverse bacterial assemblages that can be quite biologically active (Bailey and Robertson 1982; Smriga et al. 2010). It remains unknown whether and how the especially large volume of feces generated by B. muricatum may influence reef biogeochemical processes. Nutrient enrichment is not the only avenue through which fecal deposition by B. muricatum could influence reefs. As crushed calcium carbonate is known to have a strong buffering capacity in other settings (Pytkowicz 1967), the constant, large volume of calcium carbonate-rich feces generated by B. muricatum could potentially influence the alkalinity and pH of seawater at localized scales, or in regions with restricted water flow. Given great concern about the effects of rapidly changing pH levels in coral reefs, it is imperative to understand whether changes in abundance of B. muricatum and ecologically similar species may further impact pH in these systems.

We consider two main questions in this report. First we ask: What properties of B. muricatum forage may influence their feeding? To answer this question we identified differences in the chemical, mineralogical, and energetic content of its two principal food classes: living corals and other coral material lacking visible living coral. We use this information to interpret why B. muricatum partitions its feeding between these two major food sources. Second we ask: What are the biogeochemical implications of the high volume of fecal output of this species? To measure the potential effects of this excretion, we examined the nutrient content of field-collected B. muricatum feces and examined the influence of B. muricatum feces on seawater alkalinity and pH. Leveraging diverse toolkits from chemistry and material science, we gain understanding of the biogeochemical characteristics of these reef processes and identify feedbacks of this consumption on their surrounding ecosystems, rather than simply enumerating bulk rates of intake and output. Resolving these issues in the case of B. muricatum is particularly important because (1) B. muricatum is part of a globally imperiled Tribe whose numbers appear to be rapidly declining (Dulvy and Polunin 2004; Comeros-Raynal et al. 2012) and (2) B. muricatum is one of the most ecologically important benthic consumers in the coral reef communities where it exists, and it is predicted to shape the diversity, abundance, and size structure of corals in these communities as well as general dynamics of bioerosion and reef persistence (Bellwood et al. 2003; Bonaldo et al. 2014).

Materials and methods

Study site

Field data were collected at the Palmyra National Wildlife Refuge (5°52′N, 162°04′W). B. muricatum has been much depleted by fishing across its range (Dulvy and Polunin 2004; Comeros-Raynal et al. 2012), but it, and numerous other large reef fish, remains relatively abundant at Palmyra (Stevenson et al. 2007; McCauley et al. 2010, 2012). All sampling was conducted in backreef sites (landward side of reef crest; depth 1–5 m) of the atoll.

Elemental, energetic, and mineralogical content of B. muricatum forage

B. muricatum feed primarily on living and coral rock material (McCauley et al. 2014). “Living coral” will be used throughout to refer to macroscopic living scleractinian corals. Here we define “coral rock” to mean all materials consumed from reef substrates that contained no visible (> 25 mm) adult living corals. These coral rock reef substrates are highly biodiverse and highly biologically active. They can host a diverse assemblage of surface-colonizing and endolithic biota including turf and encrusting algae, crustose coralline algae, endolithic algae, boring reef invertebrates, newly settled microscopic coral recruits, and biofilms.

To understand more about how the properties of these two major forage classes guide feeding selectivity in B. muricatum, we measured the elemental composition, mineral structure, and energetic content of live corals and coral rock material. Samples were removed from the feeding areas of B. muricatum at Palmyra Atoll, placed in sterile containers, and frozen. Live coral fragments (living tissue plus skeleton) were removed from the living scleractinian corals in the following three genera: n = 13 Pocillopora, n = 3 Acropora, and n = 2 Montipora. Each of these genera is known to be consumed by B. muricatum. Pocillopora, in particular, has been previously observed at Palmyra to be preferentially consumed by B. muricatum (McCauley et al. 2014), and has been found elsewhere to be fed upon in proportion to its availability (Hoey and Bellwood 2008).

Live coral pieces were removed using a chisel or pliers. All coral rock samples were removed using a chisel. Live coral and coral rock samples were collected in such a way as to mimic the size, shape, and depth of material that B. muricatum take from each of these two forage classes, with field removal of samples performed adjacent to a known B. muricatum bite wherever possible (McCauley et al. 2014). The average size of these divot-type coral rock bites was measured in the field (n = 64) during focal follows of B. muricatum to be 3.7 length × 2.6 width × 0.4 depth cm (SE ± 0.2 × 0.2 × 0.05, respectively) (McCauley et al. 2014), and calipers were used to help match the size of coral rock sample collections to these natural mean dimensions. In the laboratory, living coral and coral rock materials were freeze-dried, milled to a fine powder in a SPEX mill using polymethacrylate balls, and thoroughly homogenized.

The percent N of living and deal coral samples was measured via elemental analyzer (Carlo-Erba NA 1500; n = 18 live coral; n = 17 coral rock). A subset of these samples was then analyzed to determine the abundance of a suite of twenty-three additional physiologically and ecologically important elements using energy-dispersive X-ray fluorescence (XRF; n = 4 live Pocillopora, n = 2 live Acropora, and n = 2 live Montipora; n = 8 coral rock) with an XEPOS spectrometer. Finely powdered samples for XRF measurement were placed in a Teflon cup over a Prolene membrane. X-ray fluorescence spectra were acquired using an XEPOS spectrometer from four spots on each cup, and results were averaged, following normalization to the Compton scattering-background. We also analyzed a subset of samples (n = 6 live Pocillopora, n = 1 live Acropora, and n = 1 live Montipora; n = 7 coral rock) following wet-digestion (HNO3/H2O2), using an inductively coupled plasma emission spectrometer (ICP; Thermo 6300, both axial- and radial-viewing geometry), which provides greater sensitivity when analyzing the abundance of elements with lower atomic numbers, allowing us to examine B and P more accurately. Samples were interspersed with processed blanks and QA/QC standards created from NIST-traceable SPEX ICP standards.

To compare the calorimetric content of live coral versus coral rock samples, we analyzed both food types using differential scanning calorimetry (DSC) on a TA Instruments DSC Q100 (V9.9, Build 303). Samples of live coral and coral rock (n = 6 live Pocillopora, n = 2 live Acropora, and n = 2 live Montipora, n = 10 coral rock) of between 8.00 and 12.00 mg were weighed in crimped-Al combustion pans, and the reference pan was filled with a comparable mass of laboratory-synthesized amorphous calcium carbonate (ACC). We selected ACC as a reference material to remove complicating endothermal transitions observed in biogenic carbonates during pilot runs with empty reference pans. Researchers have identified ACC as a common precursor to carbonate minerals in coral and other marine species (Addadi et al. 2003; Michel et al. 2008; Radha et al. 2010). The DSC temperature program heated samples (in hydrocarbon-free air at 40 ml min−1) from room temperature to 105 °C at a rate of 25 °C min−1. Samples were then heated to 465 °C at a rate of 10 °C min−1. In the interval 110–465 °C, a slower rate of heating allowed greater resolution of potential exothermic/endothermic transitions, since this is the interval at which most organic substances digested by B. muricatum demonstrate exothermic transitions. Finally, samples were heated to 550 °C at a rate of 50 °C min−1. Thermograms produced by the DSC were analyzed by integrating under the largely exothermic curve from 200 to 465 °C, obtaining a standardized enthalpy of combustion for each sample on TA’s Universal Analysis 2000 software.

To better understand how potential differences in the mineralogy of live coral versus coral rock materials shape B. muricatum feeding, we analyzed both forage types using X-ray powder diffraction (XRD, SSRL beamline 11-2, λ = 0.9744 Å) in Debye–Scherrer geometry. Diffraction rings were integrated using the software Fit2D (ESRF) by comparing to a laboratory standard of the salt LaB6. Live coral (composited from 2 Pocillopora samples) and coral rock (likewise composited from two samples) samples (collected and prepared as above) were analyzed. The XRD patterns obtained in these analyses were queried against the PDF/4 + database of the International Centre for Diffraction Data (ICDD 2015), including major lines to identify calcite, aragonite, and dolomite.

To reinforce XRD measurements, the two major classes of B. muricatum forage materials (n = 4 live Pocillopora, n = 2 live Acropora, and n = 1 live Montipora; n = 7 coral rock) were analyzed using Fourier transform infrared spectroscopy (FTIR) on a Thermo-Nicolet Nexus 470 (Madison WI) spectrometer. Ball-milled live coral and coral rock powders were mixed at 1% (w/w) with spectral-grade potassium bromide, pressed into mulls, and spectra collected in transmission-mode using FTIR over the frequency range of 5000–400 cm−1 (2–25 µm wavelength) at a resolution of 2 cm−1 and co-addition of 512 interferograms. Raw spectra were auto-corrected for the presence of ambient CO2 and H2O vapor in the sample compartment using OMNIC software. Spectra were converted into absorbance by rationing single-beam spectra with a blank KBr mull, as described in Raab and Vogel (2004). To confirm the FTIR spectral assignments, one composited sample from two each live coral and coral rock were compared over the range 50–2000 cm−1 for their FT-Raman spectra using a Thermo-Nicolet DXR Raman microscope and 532 nm laser-excitation at the Lawrence Berkeley National Lab, BL 5.4.1. FT-Raman spectral assignments followed Urmos et al. (1991).

Nutrient content of B. muricatum feces

To evaluate the possible effects of B. muricatum defecations on reef nutrient cycling, we assayed the N and P content of B. muricatum feces. For these tests, subsamples of fresh feces were collected using 100-ml sterile syringes immediately after feces has been defecated by B. muricatum and had settled onto a relatively clean portion of the substrate (Smriga et al. 2010). Effort was taken to minimize the collection of benthic substrate or detritus that might contaminate the sample of feces. The feces and seawater collected by this method were stored frozen. To measure P concentrations (n = 15), samples were filtered in the laboratory onto pre-leached ash-less filter papers (Whatman #40), freeze-dried, ground to a fine powder (as above), homogenized, Kjeldahl-digested, and analyzed on a continuous flow autoanalyzer (Alpkem Flow Solution IV). Percent N was measured from a separate set of samples of B. muricatum feces (n = 22) collected using the same methods as above, prepared using methods described above for coral, and analyzed on an elemental analyzer.

Effects of B. muricatum excretion on seawater alkalinity and pH

To determine whether the calcium carbonate-rich feces of B. muricatum have any influence on core parameters of local water chemistry, we measured the total alkalinity and pH of seawater in B. muricatum feces produced in the field (feces n = 14, reference n = 14). Water samples containing an ambient mix of feces and seawater were collected in acid-washed glass bottles in the densest portion of the cloud of feces defecated by B. muricatum. Reference samples containing only seawater were collected at the same water depth in like fashion in an area 5 m outside of the region where B. muricatum defecated. Mercuric chloride was added as a preservation agent to all feces and reference samples, and samples were sealed in airtight glass bottles, and then stored in a cool, dark location until analyzed.

The TA of the samples was evaluated from the proton balance at the alkalinity equivalence point, 4.5, at 25 °C using an open-cell titration. This method utilized a multi-point hydrochloric acid titration of seawater (Dickson 1981). The instrument program used a Levenberg–Marquardt nonlinear least-squares algorithm to calculate the TA from the potentiometric titration data. The program was patterned after those developed by Dickson (1981), Dickson et al. (2007) and Johansson and Wedborg (1982). The least-squares algorithm of the potentiometric titrations also gave initial pH as calculated from the initial electromotive force (EMF), the standard potential of the electrode system (E°). The semi-automated system consisted of a Metrohm 765 Dosimat titrator, an Orion 720A pH meter with a ROSS glass pH electrode (Orion, model 810100), and a double junction Ag/AgCl reference electrode (Orion, model 900200), and a water-jacketed pen titration cell (Millero et al. 1993). The system used ~ 60 g of sample, which was determined gravimetrically to ± 0.003 g. The seawater samples were equilibrated to a constant temperature of 25 ± 0.1 °C with a water bath (NESLAB, RTE-10). A typical titration records the stable solution EMF (deviation less than 0.09 mV) and adds enough acid to change the voltage a pre-assigned increment (~ 13 mV). The electrode was calibrated for each titration as part of the fitting procedure, using certified reference material Batch 112 provided by A. Dickson (Scripps Institution of Oceanography). We determined instrument precision at ~ 2 µmol/kg, with average value measured at 2222.8 (n = 2) and the certified value at 2223.26.

Statistical analyses

A MANOVA was used to assess differences in ICP and XRF-derived elemental concentrations. Other differences in the elemental and nutrient concentrations and caloric content of coral foods of B. muricatum, as well as differences in the total alkalinity and pH between feces and reference samples were compared using Welch’s two-sample t tests. Statistics were computed in Program R (R Core Team 2012) and MINITAB version 14.

Results

Elemental, energetic, and mineralogical composition of forage

Comparisons of the nutrient content of live coral and coral rock foods consumed by B. muricatum revealed unique patterns of enrichment and depletion. Eleven of the twenty-four elements measured were significantly different between live coral and coral rock samples (Table 1). Eight of these were higher in coral rock material than in live corals and three elements were higher in living coral. IR spectroscopy confirmed that the lack of difference in N content detected through EA was paralleled by low detectability of protein. Though the IR spectra point to differences in the average chain length of lipids between live coral and coral rock materials, no qualitative differences in abundance could be seen, and we found low detectability of carbohydrates.

Table 1 Nutrient content of live coral and coral rock, principal food sources of Bolbometopon muricatum
Full size table

Measurements of caloric content of B. muricatum potential food sources demonstrated that coral rock materials had a caloric content (as referenced to ACC) that was approximately three times greater than values observed for live corals (Fig. 2; t = 6.4; df = 14.6, P < 0.001).

Fig. 2
figure 2

Comparisons of the energy content of living coral and coral rock—major diet items of reef consumer Bolbometopon muricatum. This box and whisker plot shows the median energetic content at the bolded line, with the range of the box representing 25th and 75th quartiles, and whiskers extending to minimum and maximum values observed

Full size image

X-ray diffraction of live coral versus coral rock showed considerable overlap in calcite- and aragonite-derived bands, but in the case of coral rock, additional lines associated with magnesian calcites appeared as well, principally at d-spacings of 2.9791, 2.3648 and 1.811 Å. Dolomite was not observed. In Supplemental Figure 1, whole diffractograms are shown, as well as best library matches for each.

FTIR spectra for replicate live coral versus coral rock showed some overlap: a combination of aragonite- and calcite-associated IR absorptions at 713 and 700 cm−1, the ν1 line of aragonite (1083 cm−1), and additional organic absorptions features near 2620 cm−1 (Fig. 3). IR spectra from coral rock showed three principal differences from live coral: (1) calcite ν2 absorptions showed clear splitting of the main 856 cm−1 peak/843 cm−1 shoulder to include a band at 873 cm−1; (2) the ν4 absorptions (known to be sensitive to Mg++ substitution; Luzinova et al. 2011) possessed a weak shoulder at 718 cm−1; and (3) altered lipid composition as judged by ratios of several bands between 2980 and 2820 cm−1 (Supplemental Figure 2). Together these bands arise from the acyl chains of lipids; the ratio of methylene (> CH2) spacers to terminal methyl (–CH3) provides a bulk estimate of lipid chain lengths. While overall intensity of the lipid absorbances is not very different between live coral and coral rock material, average chain lengths differ, consistent with contrasting biological colonization. In the context of the differing mineralogy of the two substrates, we include a spectrum for the laboratory-synthesized ACC used as a reference in the DSC measurements for this manuscript (Supplemental Figure 3); comparison of this standard material suggests ACC as a minor component of these coralline materials.

Fig. 3
figure 3

FT-Raman and FTIR spectra identifying mineral matrices of live coral and coral rock materials ingested by Bolbometopon muricatum. FTIR spectra are mean of seven individual collections for each of the two dietary inputs, whereas each FT-Raman spectra are composited materials from three individual collections

Full size image

Nutrient content of feces and impacts on seawater properties

The nitrogen and phosphorus contents of field-collected B. muricatum feces were overall very low (mean % N = 0.34, SE ± 0.02; mean % P = 0.0002, SE ± 0.002).

Total alkalinity levels measured in B. muricatum feces (mean = 2216 μmol kg−1, SE ± 5) and reference (mean = 2221 μmol kg−1, SE ± 6) water samples were not significantly different (t = − 1.16, df = 3, P = 0.27). A minor, but statistically significant, difference was observed in pH between the feces (mean = 7.99, SE ± 0.009) and reference (mean = 8.00, SE ± 0.008) water samples (t = − 2.77, df = 13, P = 0.02).

Discussion

Nutritional and energetic quality of dominant foods

The large size and powerful feeding morphology of B. muricatum endow them with the unique mechanical capacity to consume and process large volumes of living scleractinian coral. Yet they divide their consumption relatively evenly between living corals and coral rock substrates colonized by other reef biota (McCauley et al. 2014).

Our assays of the energy content of live coral and coral rock materials suggested that coral rock contained approximately three times more caloric value than live corals (Fig. 2). Previous methods for measuring energy content of coral forage have focused on the caloric density of living tissues only and thus predictably yielded higher values than our measurements, which included both tissue and skeleton material (Glynn and Krupp 1986; Hourigan 1987; Tricas 1989; Haramaty 1991; Choat et al. 2002). However, since dietary value includes processing time and effort associated with non-living components, samples resembling whole bites are likely a more meaningful assay for determining the food preference of species that ingest skeleton along with living tissue.

Coupling previous estimates of intake rates of B. muricatum on live coral and coral rock (2140 and 2330 kg yr−1 indv, respectively, McCauley et al. 2014) with our calorimetry measurements, we estimate that an individual B. muricatum processes approximately 2752 kJ d−1 derived solely from coral-associated (i.e., live coral and coral rock) materials. These intake rates do not account for the energetic contributions obtained from the consumption of foods other than live coral and coral rock, although by mass such sources are estimated to comprise less than 0.5% of the diet of these fish. Such estimations must be viewed as upper bound estimates of caloric intake by B. muricatum for it is unknown what proportion of these calories estimated using these methods can in fact be digested by B. muricatum. Epilithic algal cell walls, for example, represent one class of materials found in coral rock that may not be digestible by B. muricatum and, consequently, would represent inaccessible calories.

The ostensible advantages of coral rock as a food pool is only moderately well reflected in the diet distributions of B. muricatum. Focal follows of B. muricatum at Palmyra, where these nutrient and energy measurements were taken, indicate that B. muricatum consumes coral rock in proportion to availability, while strongly preferring Pocilloporid live corals and avoiding macroalgae and other live corals (McCauley et al. 2014).

Although we found similarly low levels of macronutrients N and P in both forage types, we found elevated concentration of eight biologically important elements in the coral rock materials sampled relative to living corals (Table 1). The higher abundance of these specific nutrients in coral rock samples likely derives from the fact that the non-living skeletal matrix of the corals has been recolonized extensively, both superficially and interstitially, by a diverse array of algae, invertebrates, and microbes (e.g., turf and crustose coralline algae, endolithic algae, boring reef invertebrates, newly settled microscopic coral recruits, biofilms), that serve as rich sources of nutrients and energy, likely in forms of proteins and lipids (Campion-Alsumard et al. 1995; Clements et al. 2016). Parrotfish in other contexts have exhibited tendencies to feed preferentially on coral substrates with denser macroborer colonization (Rotjan and Lewis 2005). It remains, however, unknown and untested in the context of this research exactly which macronutrients and biologically important elements are directly detectable by B. muricatum during feeding or if and how the presence of these constituents can affect the longer-term evolution of feeding preferences for this species.

Traditional dietary trace metals were unlikely to be the selective agent for feeding, with no significant difference in levels of Cu, Ni, Mo, Co, Zn or Cd observed—with one exception, the micronutrient Mn, in which coral rock samples were depleted relative to live coral. Halogens are not usually considered in the context of nutrition, but Cl, Br, and I can play an important role in the physiology of marine organisms (e.g., in the context of defense compound/hormone production). XRF demonstrated coral rock to be highly enriched in both Br and I, when expressed as a molar ratio to [Cl]. This comparison of coral material [Br]/[Cl], [I]/[Cl] to mean ocean water values (1.57 × 10−3; Davis et al. 1998; 3 × 10−6; Winchester and Duce 1967) strongly suggests that organisms colonizing the coral rock substrates are bio-concentrating these elements from seawater. Marine algae are commonly associated with this bioaccumulation and production of complex halogenated secondary metabolites, with red algae being the most prolific (Cabrita et al. 2010; Bedke and Vanderwaal 2011). In addition, symbiotic cyanobacteria associated with sponges produce large quantities of halogenated secondary metabolites containing both chlorine and bromine (Unson et al. 1994). We found that iodine concentrations were particularly enriched, showing a nearly 20-fold increase in concentration in coral rock substrate as compared to live coral forage. The source of this iodine in the coral rock could be derived from sources such as turf or encrusting red or brown algae, sponges, or boring molluscs and other invertebrates (Das et al. 1981; Heyland and Moroz 2005). Though halogenated compounds are usually examined in the context of defense and toxicity (Hay et al. 1987), B. muricatum does not appear to present avoidance. If B. muricatum is capable of assimilating some portion of the halogenated compounds in this coral rock, then this food source may play an important role in its development and growth.

The major diet types of B. muricatum also differed in respect to their structural chemistry. The coral rock, for example, exhibited greater substitution of Mg, K, Sr, and Na for Ca (significant at P < 0.001), an elemental signature akin to a ‘high Mg’ calcite. In addition, X-ray diffraction, FTIR spectroscopy, and X-ray-fluorescence collectively indicated that the mineral mass of coral rock B. muricatum forage reflected a Mg-calcite form different from the mixture found in the live coral samples of calcite and aragonite. Coral skeletons are known to be dominated by aragonite, with calcite pointing to the presence of other calcifying organisms in the sample (Goffredo et al. 2012). The difference in the structural composition of these food types may be attributed to the benthic colonizers themselves. Crustose coralline algae, which principally utilize Mg-calcite (de Vrind-de Jong and de Vrind 1997; Addadi et al. 2003) was likely present in the coral rock sampled, along with a diverse group of other colonizers. Mineral properties of forage have consequences for their consumers, with higher substitution of Mg in the calcite lattice resulting in a mechanically harder skeletal structure (Wang et al. 1997), while also having higher solubility (Woosley et al. 2012) once ground and digested.

Rationalizing food preference and diet composition is challenging, as the consumer is not only seeking to maximize nutrient and energy intake, but to also balance these benefits with other direct and indirect costs associated with foraging for and processing of each food item (Schoener 1971; Hughes 1980; Targett et al. 1986; Targett and Targett 1990; Gerking 1994; Hay et al. 1994). Surprisingly, we found that coral rock material was by most measures a richer food source than living corals, despite some living corals being actively targeted. Additional experimental work is necessary to determine which of the compounds we have examined or other characteristics of the forage are most critical to determining the diet choices of B. muricatum, including work to understand what is actually assimilated and thus contributes to nutrition. While some species of marine fish appear to strategically target diet items based primarily on nutrient or energy content (Lobel 1981; Tricas 1989), others appear to make diet selections based on other criteria (Hay et al. 1994; Rotjan and Lewis 2005). Moreover, the relative importance of energy versus specific nutrients may vary by species, ecological context, and even ontological stage of the consumer.

Impacts of feces on reef ecosystems

Feces from reef fishes other than B. muricatum, as well as reef invertebrate communities, have been reported previously to be particularly high in nutrient content (Pinnegar and Polunin 2006) and as such, to constitute a potentially important source of nutrients and energy for recipient organisms and ecosystems (Bray et al. 1981; Bailey and Robertson 1982; Geesey et al. 1984; Meyer and Schultz 1985; Tribollet 2008). Overall rates of fish excretion, for example, have been found in some systems to contribute 25 times more N to certain reefs than all other biotic and abiotic sources pooled (Burkepile et al. 2013). Our analyses of the nutrient content of B. muricatum feces shows, however, that it contains relatively little N and P (see Pinnegar et al. 2007: %N = 1.46, %P = 1.73). As such, even though individual B. muricatum output a large volume of feces onto reefs (approximately 4 tons or greater annually; Bellwood et al. 2003; McCauley et al. 2014), it does not seem likely that this input has the capacity to induce the same fertilizing impact on reef ecosystems that is hypothesized for other piscine feces (Meyer and Schultz 1985; Pinnegar and Polunin 2006). It is important to note, however, that a sizable fraction of B. muricatum feces consists of fine particulates; ~ 25% < 125 um in B. muricatum feces sampled via gut contents analysis in the Great Barrier Reef (Hoey and Bellwood 2008). While smaller particulates were detected in the fecal samples we collected from reef substrates for nutrient analysis (e.g. ~ 18% < 75 μm and ~ 30% 76–212 μm; McCauley et al. 2014), it is highly likely that some fraction of the smallest particles do not settle to the reef substrate and would not be collected using our fecal sampling methods. Consequently, our analysis may underrepresent total nutritional output contained in B. muricatum feces. Furthermore, the analysis of feces alone does not consider the potential contributions of other excretory products of fishes (i.e., urine) which have been shown to shape nutrient budgets (Allgeier et al. 2016).

Investigations of these B. muricatum feces deposits showed little influence on seawater total alkalinity and pH at localized scales. There was no difference in the total alkalinity of B. muricatum feces from ambient seawater, providing no evidence that excreted calcium carbonate has a significant buffering effect on localized areas. Additionally, differences in feces and ambient pH were minor (i.e., Δ 0.01 pH) and do not suggest that B. muricatum alters localized seawater pH conditions to an ecologically significant degree, especially when compared to natural diurnal fluctuations (Ohde and van Woesik 1999). Despite this lack of direct effect on seawater chemistry, feces production by B. muricatum may yet exert an important influence on overall carbonate dynamics in reef ecosystems. Bioeroding Holothuroidea, for example, has been shown to increase alkalinity and exhibit distinct effects on the coral reef; however, these studies were conducted with a specimen held in an aquarium for 24 h, rather than in a field setting under natural flow conditions (Schneider et al. 2011). The transformation of tons of reef-bound, structurally complex carbonates by B. muricatum into more mobile sediment forms may yet have an important influence on reef ecosystems, by increasing transport of carbonates across reef boundaries or otherwise altering reef carbonate budgets (Bellwood 1995; Perry et al. 2015). Calcium carbonate derived from fish feces inputs (this material produced as a result of gastrointestinal precipitation) has been documented to play a major role in marine carbonate dynamics in other contexts (Wilson et al. 2009). Evaluation of the importance of these potential impacts of B. muricatum excretion will require further investigation. Experiments tracking change in carbonate chemistry in areas where B. muricatum have been extirpated or recovered over time could, for example, provide more meaningful insight into the cumulative effects of B. muricatum.

In conclusion, this work allows us to develop a holistic characterization of how B. muricatum both shapes its environment, and how it’s foraging behavior is shaped by the environment. Our analyses of the chemistry and structure of B. muricatum’s forage materials in the coral reef landscape where it feeds will help make more sense of the foraging decisions made by this species. Our examination of the in situ measured chemistry of its feces also shape our expectations of the influence it may have on nutrient budgets and patterns of ecosystem ecology. Similar combinations of integrated insight derived from material science, marine chemistry, and foraging ecology can likely be usefully applied to furthering our understanding of other marine mega-consumers or species with similarly influential roles in marine ecosystems.