Assessing the Functional Diversity of Herbivorous Reef Fishes Using a Compound-Specific Stable Isotope Approach

Assessing the functional diversity of herbivorous reef fishes using a compound-specific stable isotope approach

Thesis by

Matthew D Tietbohl

In Partial Fulfillment of the Requirements

For the Degree of

Master of Science

King Abdullah University of Science and Technology, Thuwal, Kingdom of Saudi Arabia

EXAMINATION COMMITTEE PAGE

The thesis of Matthew D Tietbohl is approved by the examination committee.

Committee Chairperson: Dr. Michael Berumen Committee Members: Dr. Xose Anxelu Moran, Dr. Mark Tester, Dr. Simon Thorrold

ABSTRACT

Assessing functional diversity of herbivorous reef fishes using a compound-

specific stable isotope approach

Matthew D Tietbohl

Herbivorous coral reef fishes play an important role in helping to structure their environment directly by consuming algae and indirectly by promoting coral health and growth. These fishes are generally separated into three broad groups: browsers, grazers, and excavators/scrapers, with these groupings often thought to have a fixed general function and all fishes within a group thought to have similar ecological roles. This categorization assumes a high level of functional redundancy within herbivorous fishes.

However, recent evidence questions the use of this broad classification scheme, and posits that there may actually be more resource partitioning within these functional groupings. Here, I use a compound-specific stable isotope approach (CSIA) to show there appears to be a greater diversity of functional roles than previously assumed within broad functional groups. The δ13C signatures from essential amino acids of reef end- members (coral, macroalgae, detritus, and phytoplankton) and fish muscle were analyzed to investigate differences in resource use between fishes. Most end-members displayed clear isotopic differences, and most fishes within functional groups were dissimilar in their isotopic signature, implying differences in the resources they target.

No grazers closely resembled each other isotopically, implying a much lower level of functional redundancy within this group; scraping parrotfish were also distinct from excavating parrotfish and to a lesser degree distinct between scrapers. This study highlights the potential of CSIA to help distinguish fine-scale ecological differences

4 within other groups of reef organisms as well. These results question the utility of lumping nominally herbivorous fishes into broad groups with assumed similar roles.

Given the apparent functional differences between nominally herbivorous reef fishes, it is important for managers to incorporate the diversity of functional roles these fish play.

ACKNOWLEDGEMENTS

I could not be more grateful to the many people who contributed to my thesis. I am so thankful of the support of my family, who offered love and support always despite being on another side of the world. My KAUST family, those in the Reef Ecology Lab, Red

Sea Research Center, and elsewhere have also been very wonderful and many people offered helpful pointers and words throughout this process. A huge thanks also goes out to all those who helped me find and collect algae and corals with an especially huge thanks to Dr. Darren Coker and Tane Sinclair-Taylor for all their help around the process of collecting the fish. I am highly grateful for the work Leah Houghton and Dr.

Simon Thorrold put in with helping to run the samples analyzed in my thesis, and to

Camrin Braun for helping me troubleshoot the IsotopeR model. Tane and Melissa

Pappas also deserve a big thanks for helping to edit the final figures in this thesis. I would also like to thank Shobhit Agrawal for his help and words of advice for coding in

RStudio; this help is very much appreciated! This project would not have been feasible if I had not had the joy of meeting Dr. Howard Choat, whose brief lecture on parrotfish ecology helped to formulate the idea that led to this thesis. I also want to thank Dr.

Kelton McMahon, whose work at KAUST inspired me to apply for the Master’s program, and who was able to offer useful comments on isotope analysis during the writing of this manuscript. Dr. Thomas Larsen also offered many key comments on isotope ecology useful in this writing. None of this would have been possible if it were not for Dr. Michael Berumen, whom I am ever thankful for deciding to bring me in as a

Master’s student and being an overall great advisor and person. Big thanks to my thesis committee as well for their thoughtful comments and patience. John Gittings, Luis Silva, and Gordon Byron deserve special thanks for many useful chats throughout the data analysis and writing process and who helped to keep me focused and happy!

TABLE OF CONTENTS Page EXAMINATION COMMITTEE APROVALS FORM 2 ABSTRACT 3 ACKNOWLEDGEMENTS 5 TABLE OF CONTENTS 6 LIST OF ABBREVIATIONS 7 LIST OF FIGURES 8 LIST OF TABLES 9 MAIN TEXT 10 1. INTRODUCTION 10 2. MATERIALS AND METHODS 19 2.1 Samples Collection 19 2.2 Sample Processing and Analysis 21 2.3 Data Analysis 24 3. RESULTS 26 3.1 End-Members 26 3.2 Fishes 30 4. DISCUSSION 34 4.1 Separation of End-Members 34 4.2 Fish Functional Diversity 41 4.3 Gut Microbial Community Effects 45 4.4 Ecosystem Functions & Management Implications 47 4.5 Future Directions 50 5. CONCLUSION 53 6. REFERENCES 54 7. APPENDIX I 67 7.1 IsotopeR Results 67 7.2 Modeling 68

LIST OF ABBREVIATIONS

CSIA – compound specific stable isotope analysis PCA – principal component analysis PC – principal component GBR – Great Barrier Reef RPM – revolutions per minute GC-C-irm-MS – gas chromatography-combustion-isotope ratio monitoring mass spectrometry MCF - methyl chloroformate GC – gas chromatograph RUBISCO - ribulose-1,5-bisphosphate carboxylase/oxygenase PEP-CK - phosphoenolpyruvate carboxykinase SCFA – short chain fatty acids EBM – ecosystem based management δ13C and δ15N –

Where X is either carbon (C) or nitrogen (N) and R = 13C/12C or 15N/14N (the corresponding ratio of heavy to light isotopes)

LIST OF FIGURES

Figure 1. Principal component analysis of end-members Figure 2. Principal component analysis of herbivorous reef fishes Figure 3. Principal component analysis of end-members and herbivorous reef fishes Figure 4. Principal component analysis of macroalgae and detrital end-members Figure 5. Mixing model results from the IsotopeR Bayesian mixing model

LIST OF TABLES

Table 1. Results from Kruskal-Wallis test comparing δ13C of end-members Table 2. Results from Dunn’s Test on each species after comparing δ13C values Table 3. Eigenvalues, variance, and cumulative variance for the five principle components (PC) of the principle component analysis run on the δ13C values of the five essential amino acids for reef end-members, herbivorous fishes, and both run together Table 4. Results from the principle component analysis (PCA) run on the δ13C values of essential amino acids for reef end-members, herbivorous fishes, and both run together

1. INTRODUCTION

Herbivores are an important component of ecosystems throughout the world where they have multiple ecosystem functions (Huntly 1991). These include, but are not limited to, aiding nutrient flow though the environment (Bazely & Jefferies 1986;

Detling 1988; Metcalfe et al. 2013), increasing biodiversity (Olff & Richie 1998; Falk et al. 2015; but see Hobbs & Huenneck 1992), and changing the physical structure of their environment (Jones et al. 1997; Belliveau et al. 2002; Nolte et al. 2013). The functional importance of herbivores is not only limited to terrestrial habitats, but extends into the marine realm (Smetacek et al. 1990; Jones & Andrews 1990; Estes et al. 1998).

On coral reefs, herbivores help structure benthic communities (Smith et al. 2010) and regulate primary production (Burkepile & Hay 2006). Herbivorous reef organisms range from invertebrates, such as urchins or snails, to vertebrates, such as sea turtles or fishes (Ogden & Lobel 1978; Goatly et al. 2012). Herbivorous reef fishes have diets that vary greatly among species; understanding what and how they eat is useful for classifying these fishes into different functional groups (Green & Bellwood 2009). While specific functional groups are somewhat debated, they are typically defined to include browsers, grazers, scrapers, and excavators (Green & Bellwood 2009; Bonaldo et al.

2014). Browsers include species from the family Kyphosidae, some surgeonfish

(Acanthuridae) of the genus Naso, some rabbitfishes (Siganidae) and parrotfishes

(Calotomus spp. and Leptoscarus spp.), and spadefish (Platax spp) (Bellwood et al.

2006). Browsers feed almost entirely on macroalgae and associated epiphytes (Hoey &

Bellwood 2009). By cropping down large stands of macroalgae, browsers prevent over- shading of corals and reduce algae-coral contact, which can help increase the survival of

11 young corals (Box & Mumby 2007). Grazers include many species of surgeonfishes

(Acanthuridae), some species of rabbitfishes, and pygmy angelfishes of the genus

Centropyge (Green & Bellwood 2009); grazers are responsible for grazing down the turf algae on coral reefs, preventing algae from overgrowing or smothering corals, and removing macroalgae before it reaches a substantial size (McCook et al. 2001, Bellwood et al. 2004). Scrapers and excavators include all non-browsing parrotfish (Labridae).

These species scrape the carbonate substrate of the reef, removing algae and creating newly cleared settlement space for juvenile corals while also contributing to the creation of sand and its movement around reefs (Bruggermann et al. 1996, Bonaldo et al. 2014).

These functional groups play complementary roles to each other, with the overall effect of an intact community aiding the settlement and growth of corals while helping prevent the establishment of macroalgae (Hughes et al. 2005; Adam et al. 2015).

Grazers, scrapers, and excavators may help maintain an ecosystem in a more coral- dominated state while preventing the excessive growth of turf and macroalgae (Heenan

& Williams 2013); browsers are key species that may help shift an ecosystem from one dominated by algae to one dominated by corals by removing large macroalgae (Bellwood et al. 2006, Nyström et al. 2008). Species within a functional group that share the same niche can be said to have redundant roles; this redundancy within functional groups is known as functional redundancy (Nyström 2006). Ecosystems tend to be more resilient to forces of change with higher levels of functional redundancy (McCann 2000). Recovery from a high prevalence of macroalgae to lower levels after disturbances, such as large storms or coral bleaching, may be more likely or faster on reefs with intact herbivorous fish communities containing higher levels of functional redundancy (Hughes 1994;

Hughes et al. 2007; Cheal et al. 2010).

Within functional groups, herbivores are often thought to have high levels of functional redundancy, though this may not always be the case (Burkepile & Hay 2007).

For example, on the Great Barrier Reef (GBR) the browser functional group has been shown to have limited redundancy among species and locations, with a single species often dominating the consumption of macroalgae despite relatively high abundances of other known browsers (Bellwood et al. 2006; Hoey & Bellwood 2009). More subtle differences may exist too. It has been found on the GBR that different species do not browse equally, with some species of browsing rabbitfish limiting their consumption to the leaves of macroalgae while Naso unicornis and Kyphosus vaigiensis mostly consume the thallus (Streit et al. 2015). Even though rabbitfishes have a high number of bites and may consume lots of macroalgae, they do not remove a significant amount of algal biomass compared to other species feeding on different parts of the same algae (Fox &

Bellwood 2008). These studies demonstrate that certain species may have disproportionate effects on macroalgae cover and that exactly what these fishes are eating can affect their functional role. Recently several grazing species, including some rabbitfishes and surgeonfish of the genus Zebrasoma, have been found to represent a likely sub-division of coral reef grazers; these fishes may be better known as “croppers” due to their ability to feed on the apical portion of algae within tiny nooks and crannies that other grazers cannot reach due to their jaw morphology (Brandl & Bellwood 2016).

Different feeding rates and times have also been reported for herbivorous fishes thought to be consuming the same material; while this could also be a mechanism to avoid competition, it may also be an indication of fishes targeting different algae whose nutrient content can vary throughout the day (Zemke-White et al. 2002). Many more nuances and subtle differences among species may exist which could hint at more

13 functional diversity and lower redundancy within these groups (Fox et al. 2009). Higher levels of functional diversity or limited functional redundancy could indicate reef communities are actually less resilient to change than previously thought. As specialized herbivorous fishes are removed or lost from an ecosystem, the function they used to fulfill is potentially left empty if no other species share the same niche or can adapt to fill it. This can potentially lead to the loss of ecosystem functions (Nyström et al. 2008).

Interactions between fishes and their environment, as with any system, are important to understand because a system with an intact food web and network of interactions is more likely to be resilient to disturbances (Gao et al. 2016). The distinctions of the roles among species within a given functional group are crucial to a better understanding how coral reefs are likely to respond to stressors they face, including climate change and fishing pressure (Mantyka & Bellwood 2007; Green & Bellwood 2009; Bellwood et al.

2011).

Because it is critical to understand species’ functional roles and it is not always easy to define these (Petchey & Gaston 2006), it is important to use a diverse suite of methods to clarify how species within each functional group actually function within coral reef ecosystems (J Choat, personal communication, Choat et al. 2004). Dietary studies and algal transplant/observational experiments have yielded important information about which food items herbivorous reef fishes target and consume, greatly improving our understanding these species’ functional roles (Choat et al. 2002; Mantyka

& Bellwood 2007; Hoey & Bellwood 2009; Streit et al. 2015). Although these studies may be informative about what these fishes choose to eat, they may not paint a complete picture of these fish’s diet in their more complex natural environment. Conventional dietary studies may only represent a snapshot in time, which may fail to take into account

14 reliance on certain food sources over others; this may be further complicated by the varying nutritional content of different food items from which nutrients may be differentially assimilated into fish tissue (Raubenheimer & Simpson 1997). Depending on the nutritional condition of an organism and what food is available to it, organisms may change their diet or nutrient assimilation to ensure a balanced nutrient ratio is reached in their tissues (Halford et al. 2000; Mayntz et al. 2005). This can also complicate the implications of feeding trials which are unable to interpret feeding preferences with regards to the fishes’ nutrient condition. Additionally, dietary analysis can be highly labor intensive, and it is time consuming to properly sample an organisms diet over a sufficiently long time period (Bearhop et al. 2004). Identifying particular food items within the gut may also be difficult due the mechanical processing or trituration that occurs after consumption by many herbivorous fishes (Bellwood & Choat 1990).

Determining exactly which item a fish is consuming and integrating nutrients from when feeding over a highly heterogeneous substrate may not always be easy, as the ability to identify fine-scale differences in targeted resources is limited (de la Morinière et al.

2003; Clements et al. 2016). One way around these potential issues in determining diet is isotope analysis.

In ecological studies, isotopic information has many uses including investigations of trophic status and tracking an organisms’ assimilation of nutrients by assessing carbon isotopes (Layman et al. 2015). Bulk analysis of stable 15N isotopes ratio values (δ15N) is often used to investigate trophic relationships between organisms and is useful in this sense because of the relatively consistent fractionation of δ15N between trophic levels that results in higher trophic levels having higher δ15N signatures (Carassou et al. 2008;

Logan & Lutcavage 2013). Fractionation refers to the change in isotopic abundance in

15 molecules and is related to differences in the kinetics and enzymatic reaction rates (Fry,

2006). While this fractionation is important for trophic analysis of food webs, it becomes problematic when using δ13C isotopic analysis. The analysis of bulk δ13C isotope signatures is often used to analyze the diets of organisms and trace the flow of carbon through a food web from an initial source (Peterson & Fry 1987; Karasov & del Rio

2007). Different producers at the bottom of the food chain, including plants, algae, and bacteria, may have unique isotopic signatures due to differences in carbon fixation and enzyme kinetics (Fry & Sherr 1989). These signatures are then incorporated into the tissue of their consumers; bulk isotopic analysis of δ13C signatures is therefore a useful way to investigate dietary source utilization as it integrates these isotopic signatures from multiple food items over time allowing a better analysis of an organism’s true diet.

Depending on the turnover time of the tissue sampled, this may allow an analysis of diet on the timescale of weeks to months or even years (Plass-Johnsen et al. 2015). However, fractionation can affect δ13C signatures as well; the true isotopic signature of a consumer may become obscured by fractionation making it difficult to infer its actual diet (Post

2002; Popp et al. 2007; McMahon et al. 2013b). Fractionation correction factors have been previously used to compensate for this, but specific investigation of isotopic fractionation has shown wildly variable values between tissue type (Tieszen et al. 1983;

Cherel et al. 2005) and species (Mill et al. 2007) making generalizations of fractionation effects between species problematic (Gannes et al. 1997; McCutchan et al. 2003). Still in some cases, bulk analysis of carbon isotopes has been useful in identifying dietary sources within multiple systems (Vander Zanden & Vabeboncoeur 2002; Layman 2007;

O’Farrell et al. 2014). But in systems with high numbers of potential carbon sources like coral reefs, bulk analysis becomes more challenging (Layman 2007; McMahon et al.

2016). Without proper identification of δ13C values at the base of the food chain, which determine consumer δ13C values, it is difficult to determine if changes in consumer δ13C are due to changes in food web structure or simply changes in the δ13C values of basal producers (Post 2002; Graham et al. 2009). However, technological advances in stable isotope analysis can help to address these concerns (Meier-Augenstein 1999; Sessions

2006).

Compound specific stable isotope analysis (CSIA) is similar to bulk analysis except that the analysis is limited to certain biomolecules rather than all sources of carbon within a tissue. One group of molecules often subject to CSIA is amino acids, specifically essential amino acids. Essential amino acids for animals, as defined by

Borman et al. (1946), are amino acids that are not internally synthesized at adequate rates for normal growth. These essential amino acids must be acquired through an organism’s diet, and are relatively conserved throughout a wide range of organisms (Fitzgerald &

Szmant 1997). The isotopic signature of these essential amino acids are determined by producers, also known as end-members, such as plants, bacteria, fungi, and algae which can synthesize amino acids de novo (Larsen et al. 2009). Because these different lineages use different substrates or have different catabolic pathways, they have been found to have unique isotopic δ13C fingerprints (Scott et al. 2006; Larsen et al. 2012; Larsen et al.

2013). Since these essential amino acids are assimilated from an organism’s diet and undergo minimal processing after initial consumption, the δ13C values of these may remain nearly unchanged from end-members to consumers (Jim et al. 2006; McMahon et al. 2010). If the basal δ13C values are known from a given ecosystem, CSIA can be used to determine which carbon sources a consumer may rely on as long as basal sources differ (Hopkins & Ferguson 2012; Arthur et al. 2014; McMahon et al. 2016).

In this study, I investigated the sources of carbon supporting herbivorous fishes on a Red Sea coral reef. While herbivory has been well studied on coral reefs of the Indo-

Pacific and Western Atlantic, the amount of scientific literature investigating the importance of and processes of herbivory on Red Sea coral reefs is comparatively limited

(Berumen et al. 2013). Hoey et al. (2016) has documented a higher diversity of parrotfish and a difference in functional assemblages between the Red Sea and Arabian Gulf, with more excavators present on Red Sea reefs. However, Khalil et al. (2013) documented overall densities of herbivorous fishes in the central Red Sea as being one to two orders of magnitude lower than reefs in the Gulf of Aqaba and Caribbean. There clearly appears to be variation in the composition of nominally herbivorous fishes throughout the Red

Sea. Given the multiple, bleaching events that have occurred over the past seven years

(Furby et al. 2013; Monroe et al. in review) a diverse suite of herbivores is likely needed to prevent reefs from becoming permanently dominated by macroalgae. It is important to further investigate the diversity of herbivorous fishes’ role as this may affect how they are able to help these reefs recover.

By further examining coral reef end-members for group-specific signatures, this study first expands upon the investigation of variation in δ13C values in end-members previously completed by Larsen et al. (2013) who found a clear separation between multiple end-members within temperate marine ecosystems. I hypothesized that the δ13C values of essential amino acids of different end-members, largely macroalgae but also including coral, microalgae/detritus, and phytoplankton, would be substantially different, allowing the possibility to distinguish between different groups, e.g. Chlorophyta from

Phaeophyta. The next goal of my study was to use the fishes’ δ13C values to determine if differences in these signatures related to differences in resource use. McMahon et al.

(2016) were able to clearly distinguish a broad range of reef fishes based on their δ13C signatures, including herbivores, planktivores, and picivores, and were further able to determine their reliance on different sources of carbon on a cross-shelf gradient of reefs using a Bayesian mixing model. Focusing solely on herbivorous fishes from a single reef,

I attempted to determine how much detail CSIA of essential amino acids could discern.

The δ13C values of different functional groups of fishes were hypothesized to differ based on assumed dissimilarities in diet. Whether or not there is apparent ecological diversity within any of the groups of herbivorous fishes, determined in this case by different isotopic signatures, is important to know because this information may directly relate to how well certain ecosystem processes of coral reefs in the Red Sea may adapt to stressors such as fishing or coral bleaching (Khalil et al. 2013). Coral reefs in the Red Sea are already heavily fished (e.g., Jin et al. 2012; Spaet & Berumen 2015), and it is possible that driving species to low abundance or local extinction could potentially eliminate important ecosystem services herbivorous fish normally supply to the region’s reefs.

2. MATERIALS AND METHODS

2.1 Sample Collection

All samples were collected from Qita’ Al-Gursh (22°25'50.89"N, 38°59'41.99"E), also known as Shark Reef, located ~8.5 km offshore from King Abdullah Economic City.

Shark Reef is a small atoll fringed with living coral, with a perimeter ~2.42 km and area of ~0.31 km2, about a third of which is a sandy bottomed and coral rubble-strewn lagoon.

End-members (n = 46) and fishes (n= 50) were collected from Shark Reef between

March 28th and May 2nd, 2016. Fourteen species characterized as source end-members from four different groups were haphazardly collected from the windward, lagoon, and leeward side of Shark Reef: macroalgae, Halimeda discoidea (n = 4) and Caulerpa serrulata (n = 3) (Chlorophyta); Dictyota aff. bartayrsii (n = 3, hereafter referred to as

Dictyota bartayrsii) and Turbinaria ornata (n = 3) (Phaeophyta); Porolithon sp (n = 3) and Galaxaura aff. rugosa (n = 3, hereafter referred to as Galaxaura rugosa)

(Rhodophyta); Symploca hydnoides (n = 3) and a filamentous Cyanobacteria sp (n = 3)

(Cyanophyta), corals, Acropora aff. lamarcki (n = 3, hereafter referred to as Acropora lamarcki), Porites lutea (n = 3), and their Symbiodinium spp (n = 6); phytoplankton, using adult Aurelia aurita (n = 9) as a proxy; and detritus/microalgae, using Holothuria atra (n = 6) as a proxy (following McMahon et al. (2016)). All macroalgae, coral, and sea cucumber samples were collected on snorkel or SCUBA and stored on ice once onboard. Jellyfish were collected with a net from via boat and three collected at one time were pooled into one sample to ensure enough protein was present for isotopic analysis.

Previous studies have found little to no difference in the δ13C of essential amino acids in algae and other members at the base of the food web, even over large differences, so end- member δ13C values over the span of Shark Reef weren’t expected to vary significantly

20 between individuals collected on the windward or leeward side (Larsen et al. 2013;

McMahon et al. 2016). Holothuria atra was used as a proxy for the detrital pool as previously described by McMahon et al. (2016). It is known that sea cucumbers consume diatoms and other microalgae in addition to detritus (Uthicke, 1999), so the δ13C signature of H. atra may actually represent an integration of epilithic microalgae and detritus, much of which is processed by microbes (Wilson et al., 2003). However, only 8

% and 7 % at most of the organic content of its diet is reported to originate from bacterial biomass and blue-green algae respectively (Moriarty 1982). This means H. atra’s δ13C signature should more closely resemble detritus without approaching the distinct δ13C signatures seen in microbes or microalgae (Larsen et al. 2009; Larsen et al. 2013).

Aurelia aurita was used as a proxy for phytoplankton production with three adults combined into one sample size due to the small concentration of protein in a single adult jellyfish (Hirst & Lucas 1998). Adult A. aurita feed mostly on copepods, many of which feed either directly on phytoplankton and ciliates or other herbivorous copepods (Ishii &

Tanaka 2001). Because the δ13C signature of essential amino acids shows almost no fractionation between diet and consumer due to the inability of the consumer to synthesize these (McMahon et al. 2010, 2013) it was believed A. aurita and H. atra would be viable proxies for phytoplankton production and detritus. Additionally, turf algae, detritus and microalgae < 125 μm, endolithic algae, calanoid copepods, cultured

Prochlorococcus, Lobophora variegata, Acanthophora spicifera, Tydemania expeditionis, and more samples of cyanobacteria and microbial mats were sampled as end-member sources though these remain yet to be analyzed as of this writing.

Herbivorous fishes from Shark Reef were collected by barrier netting (~30 m length, 1 m height, nylon mesh, hole size 3 cm X 3 cm) and stored on ice after capture

21 according to KAUST IBEC protocol 15IBEC10. Individuals were collected from different herbivorous functional groups as outlined by Green and Bellwood (2009).

Scrapers were represented by Chlorurus sordidus (n = 6), Hipposcarus harid (n = 5),

Scarus ferrugineus (n = 6), Sc. frenatus (n = 6), and Sc. niger (n = 7). Grazers included

Acanthurus nigrofuscus (n = 3), A. sohal (n = 3), Ctenochaetus striatus (n = 6), and

Zebrasoma desjardinii (n = 2) while browsers were represented by Kyphosus cinerascens

(n = 1), Naso elegans (n = 3), and N. unicornis (n = 2). Six individuals were collected in total for each species, additionally including Siganus stellatus argenteus, Siganus luridus,

Stegastes nigricans, and Plectroglyphidodon leucozonus, though only the fishes presented in this study have been analyzed to date.

2.2 Sample Preparation and Analysis

Coral tissue and Symbiodinium were processed as in Ziegler et al. (2014) with some minor modifications. Coral tissue was blown off the skeleton with an air brush containing 4 °C chilled 4% NaCl solution, and the tissue slurry was homogenized for one minute with an Ultra-Turrax mixer at speed setting four (IKA, Germany). The slurry was then centrifuged (eppendorf 5810 RHamburg, Germany) to separate the denser

Symbiodinium cells from coral tissue. The Acropora lamarcki slurry was spun at 1000

RPM at 4 °C for 5 minutes; the brown pellet of Symbiodinium was kept, and the supernatant decanted and spun three times more, once again at 1000 RPM at 4 °C for 5 minutes, then twice at 4000 RPM at 4 °C for 6 minutes. In all cases the pellet was discarded as it was mixed in color with dark brown (Symbiodinium) and pale (coral tissue) components. The supernatant was carefully decanted each time and the final supernatant was spun down at 4000 RPM at 4 °C for 12 minutes to collect the coral tissue in the pellet. Porites lutea samples were processed similarly with a slightly

22 different centrifuge protocol. The slurry was spun twice at 1000 RPM at 4 °C for 5 minutes as before, and the pellet was kept both times for Symbiodinium. The supernatant was spun at 2000 RPM at 4 °C for 4 minutes; the mixed pellet was discarded and supernatant spun again at 4000 RPM at 4 °C for 6 minutes. The pellet was mostly coral tissue, so after removing the supernatant the surface was gently washed with 4% NaCl solution and added to the supernatant solution while the rest of the bottom-half of the pellet was discarded. The supernatant was spun down one final time at 4000 RPM at 4 °C for 12 minutes to collect a pure, pale pellet of coral tissue for isotopic analysis.

Symbiodinium and coral tissue pellets were then frozen at – 40 °C.

All macroalgae samples were kept on ice until processing. Epiphytes and epizoic fauna were removed with a razor and tweezers, and samples were then rinsed with Milli-

Q water. Adult jellyfish were rinsed briefly with Milli-Q water and then frozen in batches of three, with all three being collected from the location on same day. Muscle tissue was dissected from sea cucumbers and rinsed with Milli-Q water after the skin had been scrapped off. Dorsal white muscle was removed from fishes with care taken to make sure no sample was contaminated with blood or scales. After processing, all samples were frozen at – 40 °C before being lyophilized (freeze-dried) for at least 48 hours before homogenization with mortar and pestle.

Samples were then shipped to Woods Hole Oceanographic Institution where they were analyzed by gas chromatography-combustion-isotope ratio monitoring mass spectrometry (GC-C-irm-MS). Samples were acid hydrolyzed with 6 M hydrochloric acid and derivatized by adding 35 μl methanol, 30 μl pyridine, and 15 μl methyl chloroformate (MCF) mixed by votexing for 30 seconds after each addition. Amino acids analysis by GC-C-irm-MS can be confounded by the highly polar groups found on amino

23 acids, such as carboxylic acid; therefore, prior to analysis, amino acids should be derivatized by adding less polar groups (Klee & Grob 1985; Silfer et al. 1991; McMahon et al. 2011). The MCF method of derivatization was used because it is a rapid, one-step process and decreases the derivative carbon contribution, which aids the accurate determination of sample δ13C values (Chen et al. 2010; Walsh et al. 2014). Amino acid-

MCF derivates were separated from the reaction mixtures by liquid:liquid extraction using chloroform, varying the amount of chloroform depending on the concentration of the sample: reaction mixture volume. The resulting solution was mixed and then stratified, and the organic layer of amino acid-MCF derivates were separated and used for GC-C-irm-MS analysis. Derivatized samples were injected into GC columns in splitless mode at 260 °C then separated on an Agilent VF-23ms column (30 m length,

0.25 mm inner diameter, and 0.25 mm film thickness; Agilent Technologies,

Wilmington, Delaware, USA) in an Agilent 6890N gas chromatograph (GC, Littleton,

Colorado, USA). Sample concentrations for all amino acids were adjusted so that all peaks gave a minimum 0.5 volt output. Separated amino acid peaks were combusted online in a Finnigan GC-C continuous flow interface at 950 °C and measured as CO2 with a Thermo Finnigan MAT253 mass spectrometer (Bremen, Germany). Runs were standardized by using intermittent pulses of a CO2 reference gas of known isotopic composition. Samples were run in duplicate, and amino acid standards of known isotopic concentration were run every three samples. Standards consisted of amino acid of known isotopic composition from Sigma Aldrich. Cod, Gadus morhua, muscle samples were also run as a well-characterized biological sample, with a mean reproducibility of ± 0.25

%, to help determine if sample measurements were within realistic bounds (Houghton, pers. comms.). Carbon isotope compositions were recovered for the essential amino acids

24 valine, isoleucine, leucine, threonine, and phenylalanine and the non-essential amino acids alanine, glycine, proline, aspartic acid, and glutamic acid.

2.3 Data Analysis

The δ13C values for five essential amino acids, valine, leucine, isoleucine, threonine, and phenylalanine, were tested for normality with a Shapiro-Wilks test. Data were not normally distributed for any amino acid (valine: w = 0.87787, p-value =

0.00018; isoleucine: w = 0.86163, p-value = 6.26E-05; leucine: w = 0.91149, p-value =

0.00194; threonine: w = 0.93220, p-value = 0.01011; phenylalanine: w = 0.89191, p- value = 0.00046), and the variance was not homogeneous (Fligner-Killeen test: X2 =

5.0044, df = 4, p = 0.2868). Therefore, the Kruskal-Wallis test was run to determine if there were significant differences between any end-members and Dunn’s test was run post-hoc to determine which species differed. Differences between end-member and fish

δ13C values were visualized with a principle component analysis (PCA). Individual factors maps plotted with factor variables, the essential amino acids, were used to determine which amino acids were driving differences between end-members and fishes.

In order to quantify the relative contribution of various end-members to fishes, a

Bayesian stable isotope mixing model was used. The IsotopeR (with GUI) package, version 0.5.4 (Hopkins & Ferguson 2012), was run in RStudio, version 0.99.903, separately for each functional group. Groups were run separately to increase the model’s predictive capability. Models run with large numbers of sources may quickly lose their predictive capabilities, so to reduce any model bias, the number of end-member sources input into the model was limited to the number of isotope measurements (n) +1 as recommended by Hopkins & Ferguson (2012). The IsotopeR model also includes a parameter for the concentration of each isotope sampled within end-members. Due to the

25 limited amount of data on amino acid concentrations of tropical algae and corals, these were assumed to be equal between essential amino acids. A small trophic discrimination factor of 0.1 ± 0.1‰ was chosen to represent the minimal fractionation between end- members and consumers sensu McMahon et al. (2016). The model was set to run with three Markov chains, 10,000 Markov chain Monte Carlo burn-ins and runs, and a thinning rate of 100. Model outputs compared the relative contribution of end-member sources for all individual fishes as a percentage.

3. RESULTS

3.1 End-Members

The δ13C values of coral reef end-member essential amino acids were found to differ by species (Table 1). Table 2 shows the z-scores and p-values from a post-hoc

Dunn’s test that showed which species did differ from each other. Aurelia aurita and

Turbinaria ornata differed the most compared to other species, differing from seven and nine other species respectively. For both coral species, coral tissue and symbiont showed similar patterns of significance but Acropora was significantly different than Porites.

Chlorophyts were not significantly different from each other; Rhodophyts were not significantly different as well, though they did approach the significance threshold

(Dunn’s Test: z-score = -1.551162, p-value = 0.0604). Cyanophyts and Phaeophyts were significantly different from each other. Holothuria atra was only significantly different from some macroalgae including T. ornata, filamentous Cyanobacteria, Galaxaura rugosa, and Halimeda discoidea.

Table 1. Results of a Kruskal-Wallis test comparing the essential amino acid δ13C values between each end-member species. Amino Acid Chi-square df p-value Species Valine 41.87974 13 6.85E-05 Isoleucine 43.81075 13 3.29E-05 Leucine 43.53932 13 3.66E-05 Threonine 43.47632 13 3.75E-05 Phenylalanine 32.99803 13 4.49E-05

Table 2. Results of a Dunn’s test on each species after comparing essential amino acids between species with a Kruskal-Wallis test. The z-score is shown on top and below is the p-value. Significantly different p- values are shown in bold. Acropora Acropora Aurelia Caulerpa Filamentous Dictyota Galaxaura Halimeda Holothuria Porites Porites Porolithon Symploca Species lamarcki Symbiodinium aurita serrulata Cyanobacteria bartayrsii rugosa discoidea atra lutea Symbiodinium species hydnoides Acropora -0.030414 Symbiodinium 0.4879

Aurelia 0.790788 0.821203 aurita 0.2145 0.2058 Caulerpa -0.760373 -0.729958 -1.551162 serrulata 0.2235 0.2327 0.0604 Filamentous 0.304149 0.224564 -0.486639 1.064523 Cyanobacteria 0.3805 0.369 0.3133 0.1435 Dictyota -0.608299 -0.577884 -1.399087 0.152074 -0.912448 bartayrsii 0.2715 0.2817 0.0809 0.4396 0.1808

Galaxaura -1.3672 -1.338257 -2.159461 -0.608299 -1.672822 -0.760373 rugosa 0.0856 0.0904 0.0154 0.2715 0.0472 0.2235 Halimeda -0.325149 -0.292634 -1.170538 0.487724 -0.650299 0.325149 1.13802 discoidea 0.3725 0.3849 0.1209 0.3129 0.2577 0.3725 0.1276 Holothuria -1.808688 -1.773568 -2.721812 -0.930685 -2.15989 -1.106285 -0.228281 -1.596596 atra 0.0352 0.0381 0.0032 0.176 0.0154 0.1343 0.4097 0.0552 Porites -2.463611 -2.433196 -3.254399 -1.703237 -2.76776 -1.855312 -1.094938 -2.308561 -1.036044 lutea 0.0069 0.0075 0.0006 0.0443 0.0028 0.0318 0.1368 0.0105 0.1501

Porites -2.422196 -2.402781 -3.223985 -1.672822 -2.737245 -1.824897 -1.064523 -2.276046 -1.000924 0.030414 Symbiodinium 0.0075 0.0081 0.0006 0.0472 0.0031 0.034 0.1435 0.0114 0.1584 0.4879 Porolithon -2.919835 -2.88942 -3.710624 -2.159461 -3.223985 -2.311536 -1.551162 -2.7925 -1.562847 -0.0456224 -0.486639 species 0.0018 0.0019 0.0001 0.0154 0.0006 0.0104 0.0604 0.0026 0.059 0.3241 0.3133 Symploca -1.855312 -1.824897 -2.6461 -1.094938 -2.159461 -1.247013 -0.486639 -1.658262 -0.333641 0.608299 0.577884 1.064523 hydnoides 0.0318 0.034 0.0041 0.1368 0.0154 0.1062 0.3133 0.0486 0.3693 0.2715 0.2817 0.1435 Turbinaria -3.07191 -3.041495 -3.862699 -2.311536 -3.376059 -2.463611 -1.703237 -2.95886 -1.738448 -0.608299 0.63817 -0.152074 -1.216598 ornata 0.0011 0.0012 0.0001 0.0104 0.0004 0.0069 0.0443 0.0015 0.0411 0.2715 0.2615 0.4396 0.1119

The principle component analysis (PCA) visualized overall patterns in isotopic signatures between end-members. Over 83% of the variance between end-members was explained by the first two principle components (PC, Table 3), with phenylalanine as the most important loading for PC1 and isoleucine and threonine as the most important loadings for PC2 (Table 4). The PCA displayed the large differences between the phytoplankton end-member and both corals from the rest of the macro and microalgae

(Figure 1). It demonstrated the close association between coral tissue and coral symbionts and showed that Acropora and Porites corals were more isotopically distinct from each other. There was not tight clustering of macroalgae within broader groups. The two

Chlorophyts were clustered near each other with filamentous Cyanobacteria. Rhodophyts and Phaeophyts did not cluster closely together: Dictyota bartayrsii was more similar to the sea cucumber microalgae proxy and Halimeda discoidea, Turbinaria ornata clustered with the calcareous Porolithon sp, and Galaxaura rugosa clustered more tightly with the cyanobacteria Symploca hydnoides rather than Porolithon sp.

Table 3. Eigenvalues, variance, and cumulative variance for the five principle components (PC) of the principle component analysis run on the δ13C values of the five essential amino acids for reef end- members, herbivorous fishes, and both run together. Over 75% of the variance was explained by the first two PCs in all cases, with end-members being the best described by the first two PCs.

Eigenvalues Variance (%) Cumulative Variance (%) End-Members Dimension 1 3.3377 66.7540 66.7540 Dimension 2 0.8268 16.5356 83.2896 Dimension 3 0.5876 11.7517 95.0414 Dimension 4 0.1362 2.7232 97.7645 Dimension 5 0.1118 2.2355 100 Fishes Dimension 1 2.6302 52.6034 52.6034 Dimension 2 1.2109 24.2185 76.8219 Dimension 3 0.7893 15.7857 92.6076 Dimension 4 0.2459 4.9174 97.5250 Dimension 5 0.1238 2.4750 100 Combined Dimension 1 2.7780 55.5593 55.5593 Dimension 2 1.2387 24.7733 80.3326 Dimension 3 0.5071 10.1412 90.4738 Dimension 4 0.3223 6.4454 96.9192 Dimension 5 0.1540 3.0808 100

Table 4. Results from the principle component analysis (PCA) run on the δ13C values of essential amino acids for reef end-members, herbivorous fishes, and both run together. Phenylalanine was the most important loading vector for PC1 for end-members and the combined PCA, while threonine was most important, in a negative direction, for PC1 for herbivorous fishes.

PC1 PC2 PC3 PC4 PC5 End-Members Valine 0.4847 -0.3978 -0.0727 -0.7755 0.0106 Isoleucine 0.3998 0.5663 0.5423 -0.0982 -0.4574 Leucine 0.4718 -0.4488 -0.1995 0.5370 -0.4977 Threonine 0.3535 0.5638 -0.7360 0.0025 0.1244 Phenylalanine 0.5075 -0.0417 0.3357 0.3170 0.7263 Fishes Valine 0.5378 -0.2318 0.2729 -0.5947 -0.4783 Isoleucine -0.3902 0.3062 0.7543 -0.3391 0.2647 Leucine 0.4958 0.1475 0.5210 0.2595 -0.0579 Threonine -0.5578 -0.2460 0.1829 0.2595 -0.7263 Phenylalanine -0.0385 -0.8776 0.2274 0.0798 0.4126 Combined Valine 0.5013 -0.4092 0.0957 0.0991 0.7498 Isoleucine 0.3703 0.5372 -0.6644 -0.3208 0.1729 Leucine 0.4633 -0.4554 0.0442 -0.5820 -0.4871 Threonine 0.3334 0.5783 0.7296 -0.1476 0.0191 Phenylalanine 0.5345 0.0457 -0.1229 0.7258 -0.4127

Figure 1. Principal component analysis on coral reef end-members using the δ13C signature of five essential amino acids valine, isoleucine, leucine, threonine, and phenylalanine. The numbers in parentheses are the percent variation explained by each axis. Vectors depict the direction in which amino acids drive differences in the location of end-members on the first two dimensions; their relative equality in length depicts their equal importance in determining variation for these first two dimensions. Images next to the points are shaded outlines of each organism.

3.2 Fishes

The PCA for fishes revealed not all functional groups clustered together (Figure

2). Most parrotfish clustered in the same area with the notable exception of Chlorurus

31 sordidus. Grazers were spread throughout the PCA with Ctenochaetus striatus well separated from the other species. Zebrasoma desjardinii clustered with the browser

Kyphosus cinerascens; Acanthurus nigrofuscus was near but distinct from A. sohal which clustered more closely with both browsing Naso species. Threonine was the most important loading in the negative direction for PC1 followed by valine and leucine in the positive direction. PC1 and PC2 together explained only 76.8% of the variance (Table 3).

Phenylalanine was the most important loading for PC2 in the negative direction as well

(Table 4). When end-members were run together with fishes in the PCA, the first two

PCs described over 80% of the variance (Figure 3). All fishes were well constrained within the end-members with the exception of Ct. striatus. End-members and fishes largely maintained their clustering pattern and distribution, except for A. nigrofuscus, which clustered more closely to Z. desjardinii and K. cinerascens. Phenylalanine was the most important loading for PC1, while threonine and isoleucine were the most important loadings for PC2 (Table 3). Results from the IsotopeR mixing model output are further discussed in Appendix I.

Figure 2. Principal component analysis on coral reef herbivorous fishes using the δ13C signature of five essential amino acids valine, isoleucine, leucine, threonine, and phenylalanine. The numbers in parentheses are the percent variation explained by each axis. Vectors depict the direction in which amino acids drive differences in the location of end-members on the first two dimensions; the longer vector length for phenylalanine demonstrates its slightly greater importance for detecting differences in the first two dimensions. Images next to the points are shaded outlines of each fish species.

Figure 3. Principal component analysis on coral reef end-members and herbivorous fishes using the δ13C signature of five essential amino acids valine, isoleucine, leucine, threonine, and phenylalanine. The numbers in parentheses are the percent variation explained by each axis. Vectors depict the direction in which amino acids drive differences in the location of end-members on the first two dimensions. Note the position changes mostly for fishes and not end-members in this analysis. Symbol shape and color is the same from the previous two figures.

4. DISCUSSION

The compound-specific isotope analysis (CSIA) of the δ13C signatures of essential amino acids proved to be quite useful in distinguishing fine scale differences in the isotopic signatures of nominally herbivorous coral reef fishes related to differences in resource use. Despite an assumed similarity in diet between all parrotfish species, this technique found interspecific differences in the δ13C signatures between scrapers and excavating parrotfish and even to some degree within scraping parrotfishes. It also distinguished differences in the group of grazing surgeonfishes, with no two species closely resembling each other (Figure 2). There may be great potential using the CSIA technique with other reef functional groups, like invertivores, to expose other instances of hidden diversity within reef organisms. The broad functional group labels normally used to define herbivorous fishes on coral reefs appear to be less useful than previously thought. Because this technique demonstrates apparent functional differences within these groups, it implies that management plans used to manage herbivorous fishes may need to be reconsidered in order to protect this diversity of roles.

4.1 Separation of End-Members

A wide range of potential food sources or end-members for coral reef herbivores were sampled in this study to use the supposed diversity of δ13C signatures to reconstruct their dietary contributions to herbivorous fishes. There were some broad differences between end-members, including both corals and jellyfish, which separated out well from the rest of the macroalgae based off their δ13C values of essential amino acids (Figure 1).

In previous studies, microalgae cultures (Larsen et al. 2013) and herbivorous copepods

(McMahon et al. 2016) have been used to represent phytoplankton production; however, sampling these requires access to cultures and sorting through a diverse suite of plankton

35 for certain copepods is highly time consuming. In this study Aurelia jellyfish were used as a proxy for phytoplankton production as it was easily collected and likely well represented phytoplankton production. Pelagic δ13C values comes from a multitude of photosynthetic organisms in the Red Sea (Shaikh et al. 1986) that retain a range of metabolic pathways that may fractionate δ13C values differently (Falkowski 1991).

However, this range of possible δ13C values is integrated in the essential amino acids of

Aurelia aurita which differed from benthic end-members in being more depleted in δ13C, especially for valine (average δ13C = -23.81± 0.17), leucine (average δ13C = -25.11±

0.16), and phenylalanine (average δ13C = -27.09 ± 0.27). This distinction between generally more negative pelagic and higher benthic δ13C values has been well documented for a range of aquatic and marine habitats (Vander Zanden & Vadeboncoeur

2002; Hobson et al 2002; Wyatt et al. 2012) including with analysis of essential amino acids (Larsen et al. 2013). If Aurelia was a poor proxy, it would have likely been much closer to the benthic sources of production and enriched in δ13C, but its generally lower

δ13C values support the notion that Aurelia is an easy to collect, useful proxy for phytoplankton production.

Both coral species’ δ13C values were nearly identical to their symbionts, which likely reflects the corals’ high reliance on the symbionts as an autotrophic source of essential amino acids (McCloskey & Muscatine 1984; Muscatine 1990, Wang & Douglas

1999). At greater depths with less light, coral symbionts will increase the efficiency at which they photosynthesize, but may be more limited in the amount of nutrients they share with their coral host (McCloskey & Muscatine 1984; Titlyanov et al. 2001). To compensate for this loss of input from symbionts, corals may increase their reliance on heterotrophic carbon, evidenced by increasing similarity to pelagic δ13C with increasing

36 depth (Muscatine et al. 1989), though this might not always be the case (Ziegler et al.

2014). In this study however, all corals were sampled from a relatively shallow depth of less than ten meters where they should rely heavily on symbionts as a source of carbon, which would help explain the high similarity in δ13C values of essential amino acids between each coral and its symbiont.

Notably though, while both Acropora and Porites were isotopically similar to their respective Symbiodinium, they were clearly separated from each other in PCA space. Genetically distinct cultures of Symbiodinium have been shown to have unique structures of isoenzymes (Schoenberg & Trench 1980) and different metabolite profiles

(Klueter et al. 2015) which does imply the Symbiodinium in all Acropora and Porites corals sampled were genetically distinct with a slightly different synthesis pathway for essential amino acids. Coral species is known not to be the main determinant of symbiont distribution on coral reefs; environmental factors, such as light or temperature, are play stronger roles in structuring symbiont communities among corals (Rowan & Knowlton

1995; LaJeunesse 2005). Additionally, a diverse range of corals in the Red Sea have been shown to have a limited diversity in their symbiont host, limited to Symbiodinium Clades

A and C according to Karako-Lampert et al. (2004) and Sawall et al. (2014). So given the similarity in location from which corals were collected and the low diversity of

Symbiodinium clades in the Red Sea, it cannot be ruled out that both colonies harbored similar symbiont communities, and the difference in isotopic signatures resulted from small scale variation in environmental factors, like temperature, that affected the anabolism of essential amino acids (Gorospe & Karl 2011). However, there is potentially a high amount of diversity even within Symbiodinium clades (LaJeunesse et al. 2004) so even if the Acropora and Porites corals harbored Symbiodinium from the same clade,

37 they may have different variants with different amino acid synthesis pathways, though further testing would be needed to confirm this.

While both corals and the phytoplankton end-members showed good separation in PCA space, detritus and macroalgae showed a much less clear pattern. Though the detritus end-member was generally distinct from the other macroalgae, it was not significantly different from Dictyota bartayrsii, the nearest macroalgae as well as a few others close to it in the PCA analysis (Table 2; Figure 1). Though McMahon et al. (2016) found a distinct separation of detritus and macroalgae using the same sea cucumber species as a proxy, a different macroalgae was used in their analysis. Much of the detrital biomass found on coral reefs is known to originate from algal biomass, though turf algae appears to be the main supplier (Wilson et al. 2003). If the detritus Holothuria atra was feeding on was mostly derived from macroalgae, it could explain the greater similarity found in this study between detritus and macroalgae end-members.

Though it was expected that there would be greater differences between the different types of macroalgae, this was not found to be the case in this analysis. In terrestrial ecosystems, different species of plants may be distinguished isotopically by their specific mechanism of photosynthesis; C3 plants such as trees, fix carbon by combining phosphate ribulose-1,5-bisphosphate with CO2 via the enzyme ribulose-1,5- bisphosphate carboxylase/oxygenase (RUBISCO) to form two three-carbon sugars while

C4 plants, such as grass species, fix carbon in a two-step process by first fixing CO2 first into a four carbon sugar before being transferred to a separate layer of cells where it is re- fixed by RUBISCO (Furbank & Taylor 1995). These different chemical reactions result in different fractionation patterns of δ13C which in turn results in the ability to distinguish isotopically between plants and plant contributions to soil content (Bender 1968). The

38 majority of fractionation in marine algae is also purported to come from the initial step of photosynthesis by the action of RUBISCO (Parks & Epstein 1960). Slight variations in the affinity of RUBISCO for CO2, whether directly taken up from the water or created

- from HCO3 through phosphoenolpyruvate carboxylase (Moroney & Somanchi 1999), are known to exist between species of marine algae (Raven 1997). Different reaction rates of RUBISCO would result in small isotopic differences in the precursor molecules in amino acid synthesis (Smith et al. 1961). The initial isotopic differences in these molecules could be further exaggerated by any variations in enzymes involved later on in the synthesis of essential amino acids. And while RUBISCO appears to be the prominent fixation enzyme in Chlorophyts and Rhodophyts, some Phaeophyts seem to rely more heavily on phosphoenolpyruvate carboxykinase (PEP-CK) to incorporate CO2 in addition to RUBISCO (Kremer & Kuppers 1977). The use of an additional enzyme in carbon fixation means that these Phaeophyts might be even more isotopically distinct from other marine algae. Indeed this distinction between δ13C values of Phaeophyts, Rhodophyts, and Cyanobacteria and other microalgae essential amino acids has been found in a temperate system using CSIA (Larsen et al. 2013). In the current study, this was found to be the case with only Turbinaria ornata, which was notably separated from other algae except for Porolithon sp (Figure 1). The other Phaeophyt, Dictyota bartayrsii, was found to group in more of a shotgun cluster with Chlorophyts and filamentous Cyanobacteria.

To see if there might be further separation of macroalgae based on their essential amino acid δ13C isotope signatures obscured by the distinctly isotopically separate phytoplankton and coral end-members, a separate PCA was run with just macroalgae and detritus (Figure 4). This analysis showed that T. ornata as isotopically more distinct than

Porolithon sp which now clustered tightly with the filamentous Cyanobacteria. Fucoid

39 algae like T. ornata often exhibit large differences with other macroalgae, including other

Phaeophyts, likely due to yet-undescribed differences in the biosynthesis pathway of essential amino acids (Larsen, pers. comm.) or related to the heavier use of PEP-CK in carbon fixation (Kremer & Kuppers 1977). Galaxaura rugosa was also more distinct in this second PCA plot; the reasons for this are unclear, but since Rhodophyts are reported to use the same enzymes in carbon fixation and use similar precursor molecules in the formation of amino acids this distinction could result from variation in the kinematics of enzymes involved further along in the synthesis of essential amino acids resulting in a different fractionation pattern (Abelson & Hoering 1961; Kremer & Kuppers 1977).

Figure 4. Principal component analysis of macroalgae and detritus end-member samples using the δ13C signature of five essential amino acids valine, isoleucine, leucine, threonine, and phenylalanine. The numbers in parentheses are the percent variation explained by each axis. Vectors depict the direction in which amino acids drive differences in the location of end-members on the first two dimensions. Detritus shows a more clear separation from macroalgae in this analysis, while the rest of macroalgae samples are generally distinct from one another, likely due to differences in the reaction rates of enzymes involved in carbon fixation.

Both Chlorophyts were relatively close to one another and distinct from most other algae, likely a result of similar RUBISCO reaction rates and subsequent amino acid synthesis steps (Raven 1997). Single celled cyanobacteria have been reported to have distinct δ13C signatures of essential amino acids (Larsen et al. 2013) though this distinction appears to be lost when comparing larger colonies of cyanobacteria like were used in this study.

Larger colonies of cyanobacteria may heterotrophically utilize detrital carbon from other macroalgae, which could explain the lack of distinction even in this further analysis

(Bardgett et al. 2007).

In summary, broad similarities in the enzymes involved in carbon fixation in marine algae and subsequent synthesis reactions in the formation of essential amino acids could result in the lack of broadly distinct groups related to higher algae taxonomy. The small differences though, are very likely due to unique fractionation patterns related to the first steps of carbon fixation, which differ at least slightly between different algal species (Raven 1977; Moroney & Somanchi 1999). To further clarify patterns of isotopic values of marine algae, a broader variety of samples will need to be analyzed, ideally with further investigation into the biochemical reactions involved in carbon fixation and amino acid synthesis.

4.2 Fish Functional Diversity

Previous studies have shown the ability to use CSIA to separate fish and other marine consumers based on their assimilation of carbon from various sources (Larsen et al. 2013; Arthur et al. 2014; McMahon et al. 2016). In this study, this technique demonstrated the ability to distinguish differences within herbivorous fishes, even those reported to have nearly identical diets (Paddack et al. 2006; Green & Bellwood 2009).

This shows the power of this technique to resolve fine-scale differences in carbon uptake and assimilation in marine fishes. The grazer functional group demonstrated this well, as no two fish closely resembled the isotopic signature of another (Figure 2). Acanthurus nigrofuscus and A. nigrofuscus are reported to have highly similar diets, consisting mostly of filamentous red (Polysiphonia) and green filamentous or turfing algae

(Montgomery et al. 1989; Alwany et al. 2005; Marshell & Mumby 2012; Miyake et al.

2016). Zebrasoma desjardinii is generally reported to feed on the same kinds of algae

(Montgomery et al. 1989; Miyake et al. 2016) though its diet may also include substantial amounts of fish feces and gelatinous zooplankton (Alwany 2008; Bos et al.

2016). Ctenochaetus striatus grazes over the same turfing substrate as the Acanthurid grazers and Z. desjardinii, but has previously been reported to consume mostly detritus and microalgae (Montgomery et al. 1989, Choat et al. 2002; Marshell & Mumby 2012).

Recent analysis of fatty acid profiles have shown that Ct. striatus appears to actually be targeting and assimilating energy from benthic diatoms within the epilithic algae matrix

(Clements et al. 2016). Based on these reports of the diets of these fishes, it would be expected for A. nigrofuscus and A. sohal to be highly similar in their isotopic signatures,

Z. desjardinii to be mostly similar, though perhaps to a slightly lesser degree, and it would be expected for Ct. striatus to differ more substantially compared to the other grazers. However, the isotopic data seem to tell a different story where these species shows a different pattern of δ13C values implying resource partitioning within grazers.

This could be explained either through fishes specifically targeting different varieties of algae or certain areas on the reef. Brandl and Bellwood (2016) have demonstrated centimeter-scale differences in the location of grazing in herbivorous fishes.

Rabbitfishes, Zebrasoma surgeonfishes, and A. nigrofuscus preferentially feed within nooks and small crevices while larger Acanthurid surgeonfish, Ct. striatus, and parrotfishes fed mostly on relatively flat surfaces around these microtopographic refuges; the species with the ability to graze within these refuges were named as “croppers” within the grazing functional group. Given this distinction, one would expect Z. desjardinii and A. nigrofuscus to be highly similar in their isotopic signatures, though this was not found to be the case (Figure 2). This further implies an even higher degree of partitioning within these croppers.

While parrotfishes are broadly reported to feeding on turfing, filamentous algae, and detritus, there are some slight variations between species in reported diets. Though

43 the vast majority of bits were taken over short turfing algae, Lokrantz et al. (2008) reported that Chlorurus sordidus also foraged over calcareous algae; Scarus niger utilized this surface less, but also consumed massive coral species about ten percent of the time. Hipposcarus harid has been reported to feed on decaying seagrass blades and sponges as well (de la Torre-Castro et al. 2008), though these reports are from individuals within seagrass beds, which cannot be found on Shark Reef (pers. obs.).

Other reports for the species analyzed in this study show a high consumption of turfing algae and a small reliance on massive corals (Afeworki et al. 2011; Afeworki et al. 2013;

Bonaldo et al. 2014). Given these reports, one would expect that all parrotfish sampled in this study might cluster together, though this was not quite the case. All the scraping species did cluster quite closely (Figure 2) though there was still some noticeable separation between Sc. niger and H. harid at a small scale. The lone excavating species sampled, C. sordidus, was noticeably different from the other parrotfish, implying the use and assimilation of a different carbon source. Again, the recent report of Clements et al.

(2016) sheds some light on why this difference might exist. In their report, Clements et al. (2016) posit that scraping and excavating parrotfish are specifically targeting epilithic and endolithic cyanobacteria. After a thorough analysis of feeding morphology, fatty acid analysis, and bulk isotope analysis, their review concludes the combined data are most consistent with a parrotfish diet heavily reliant on these cyanobacteria and other micro- autotrophs on or within the reef substratum. Euendolithic, or rock boring, algae are known to have successional patterns just as turf algae on reefs and terrestrial plants do

(Tribollet 2008), where bare reef are first colonized by euendolithic cyanobacteria and some Chlorophyts and then transition to a community dominated by the green alga

Ostreobium and Phaeophila dendroides. This shift in species is also related to a shift in

44 the depth of the reef substrate, as Ostreobium is able to penetrate the substratum by a few millimeters (Grange et al. 2015). Excavating parrotfishes seem to preferentially target euendolithic communities in the later successional stage, while scraping parrotfish prefer early stage endolithic communities. If these different groups of parrotfishes are targeting different endolithic communities as posited by Clements et al. (2016), then this could explain the difference between C. sordidus and the other parrotfish. Just as there seemed to be some degree of niche partitioning between the two cropping surgeonfishes, there may very well be some small degree of this in scraping parrotfish, evidenced by the slight differences between Sc. niger and H. harid from the other two scarids.

Of all the browsers collected in this study, only a small subset was analyzed. So any analysis must be interpreted cautiously. However, given the minimal individual variation between the other samples, it is likely that they are a decent representation of the unanalyzed sample’s isotopic values. Browsers were partitioned into two groups, one with the single Kyphosid, Kyphosus cinerascens, and the other with Naso elegans and N. unicornis. All three species are reported to have broadly similar diets, with a high reliance on fucoid algal species such as Turbinaria and Sargassum (Clements & Choat

1997, Choat et al. 2002, Rasher et al. 2013). However, Choat et al. (2002) reported the diet of K. cinerascens to consist mostly of red thallate algae, such as Laurencia spp, and green and red filamentous algae. Naso elegans has also been reported to consume red thallate algae as well by Miyake et al. (2016). In all the browsers collected in this study, most had stomachs full of solely brown fucoid algae, though some individuals of N. unicornis and N. elegans had consumed other algae as well (pers. obsv.). This difference in fucoid algae consumption could be the factor driving the difference between these species, though again, it’s prudent to be cautious in assuming differences based on a

45 single species. What is interesting to note, is that in Figure 3, none of the browsers clustered closely with Turbinaria ornata, indicating that something other than dietary content could be structuring the isotopic signatures. If the actual diet of these fishes is not fully responsible for their isotopic signature, it is very likely that the isotopic signatures are altered by the fishes’ gut microbial community. This effect has been previously identified in a marine vertebrate herbivore (Arthur et al. 2014) and will be discussed further in the following section.

4.3 Gut Microbial Community Effects

The gut microbiota is known to contribute to digestion and affect the nutrition and growth among other factors in fishes as well as other animals (Ghanbari et al. 2015).

Since it has been reported that microbes and bacteria may have their own unique isotopic signature (Larsen et al. 2009) and many herbivorous fishes may receive nutrition from their gut microbes (Clements & Choat 1995), it is reasonable to assume that fishes receiving nutrients from their gut microbiota might be perceived to have a microbial based diet even if this is not the case. This has been shown to be the case in an analysis of sea turtles by Arthur et al. (2014). They found that sea turtles feeding on seagrass diets had isotopic signatures more similar to bacteria, while those feeding on a more carnivorous diet in the pelagic zone more closely resembled microalgae that represent the base of the pelagic food web. This study implies that herbivorous vertebrates that rely on their gut microbiota to break down algae they consume might not resemble the food item they are consuming isotopically. Indeed, this seems to be the case with the herbivorous fishes in this study. Since K. cinerascens is known to rely mostly on fucoid algae, like T. ornata, and had stomachs full of the same algae, it would be expected to closely resemble this algae in its isotopic signature. However, this was not the case when both were plotted

46 together (Figure 3) which implies an impact of the gut microbial community on its isotopic signature. The same is also likely to be true of N. unicornis and N. elegans which are also known to rely on fucoid algae as a source of nutrients and have extensive gut microbiota that actively ferment consumed algae (Choat et al. 2002, Miyake et al. 2015,

Ngugi pers. comm.). Other grazers have been reported to rely on fermentative gut microbes as part of their nutrition, but to a lesser extent than browsing species; this is based on the concentrations of short chain fatty acids (SCFA) in their gut, which are indicative of microbial carbohydrate fermentation (Choat et al. 2002). Zebrasoma and

Acanthurus species closely related to A. sohal are species reported to have intermediate concentrations of SCFA in their gut. Analysis by Miyake et al. (2015) shows a similar microbial community of A. sohal and A. nigrofuscus collected from the Red Sea. With an intermediate reliance on microbial fermentation for nutrition and similar gut microbial communities, it might be expected to see similar isotopic signatures between these species as was seen for both Naso species which were also similar in their gut microbiota. One might suspect a convergence in isotopic signature between these two grazers but this was not the case. In this case, it appears to be a difference in diet that drives the difference in isotopic signature, which may be explained by the fine-scale differences in feeding microhabitat. Zebrasoma desjardinii had a highly similar gut microbial community to both Naso species (Miyake et al. 2015); if the gut microbes of fishes relying on hindgut fermenting bacteria fully structured their isotopic signatures, then Z. desjardinii, N. elegans, and N. unicornis should all have highly similar isotopic signatures. This was not observed (Figure 2), which implies that the gut microbial community is not likely to be the main factor structuring essential amino acid isotope signatures, except in fishes heavily reliant on hindgut fermentative microbes. Most

47 species of parrotfish and some surgeonfish, such as Ct. striatus, have hindguts characterized by isovalerate, a SCFA byproduct of protein fermentation (Davila et al.

2013). This, paired with a relatively transient gut microbiota (Miyake et al. 2015), implies a diet high in protein and a low reliance of the core gut microbiota for nutrient assimilation, meaning there is likely a minimal isotopic effect of the gut microbial community in parrotfish and some surgeonfishes.

4.4 Ecosystem Functions and Management Implications

The diversity in isotopic signatures of these herbivorous fishes implies the utilization of a broader diversity of resources than normally thought. Here, it may be important to note that the term ‘herbivore’ may not be an accurate descriptor for all of these fishes. Herbivore well describes browsing fishes that seem to rely on hind gut fermenting microbes to help digest complex algae cell walls, similar to the digestion process of terrestrial herbivores (Crossman et al. 2005). Parrotfishes and some surgeonfishes, like Ctenochaetus striatus, however, may be better described as microphagous because of their apparent targeting of micro- and euendolithic algae

(Clements et al. 2016). However, removing the herbivore label from these species does not downplay their importance in coral reef functioning. Fishes labeled grazers, scrapers/excavators, and browsers all still play key roles in helping to structure coral reef ecosystems. Given this, it is becoming apparent that the roles of these broad functional groups are more complex and diverse than previously thought. Grazers do not simply function as reef lawnmowers cropping down all turf algae and preventing the establishment of macroalgae, but each have a unique place in the scheme of fish-reef interactions. The combined selective grazing of cropping species within small crevices helps to promote the survival of recently settled coral recruits while grazing around these

48 tiny refuges may help prevent the shading of them; both kinds of grazers are needed to reduce the growth of turf algae and aid coral settlement (Brandl & Bellwood 2016). The functional role of bioerosion and clearing of space on reefs may also be divided among parrotfishes. Excavators, especially large ones like the bumphead parrotfish, are often thought to have a much greater impact on bioerosion processes on reefs (Bellwood et al.

2003). But the difference may not be just in the amount of reef material that is eroded, but where this occurs. Scrapers targeting early successional stages of endolithic algae focus their feeding over a smaller area, which helps maintain the substrate in an early stage of succession (Clements et al. 2016). This positive feedback loop appears to also facilitate coral recruitment as crustose coralline algae may also be abundant in the preferred feeding area of scraping parrotfishes (Adam et al. 2015). The targeting of reef substrate in early successional stages is likely to result in the focusing of bioerosion over certain areas or reefs. Excavating parrotfish range over a larger swath of reef (Adam et al. 2015) and may have a more diffuse effect on bioerosion across the entire reef-scape.

Both forms of bioerosion likely play important roles in the clearing of new space for organisms to settle and in coral reef carbonate accretion (Clements et al. 2016), though more detailed investigation is needed to understand how scrapers versus excavators may affect this process.

The CSIA of essential amino acid δ13C signatures helped reveal functional diversity within the nominally herbivorous fishes on coral reefs. This higher level of diversity related to their isotopic signatures supports previously reported differences within some of these groups (Brandl & Bellwood 2016, Clements et al. 2016). What this also means is that there may be much less functional redundancy within these functional groups. Though ecological diversity is known to be important to maintaining healthy

49 ecosystems resilient to disturbances (Bellwood et al. 2004, Hughes et al. 2010), functional redundancy is also an essential part of healthy and resilient ecosystems

(Nyström 2006). Previous studies have demonstrated limited functional diversity within browsers in other reef systems (Bellwood et al. 2006; Hoey & Bellwood 2009; Streit et al. 2015) and the data presented here seem to be in agreement with this. Though not all browsing fish samples have been analyzed to date, browsers appear to be divided into two groups: Naso surgeonfish and Kyphosus cinerascens. This limited functional diversity of the fishes that are thought to be responsible for preventing macroalgae overgrowth on reefs means this functional role may be vulnerable to disturbances to these fish populations (Green & Bellwood 2009). Though the other nominally herbivorous fish functional groups are more specious, they may also have lower functional redundancy due to high degrees of specialization within grazers and scrapers/excavators. Lower levels of functional redundancy within these groups could lower the resilience of these processes to disturbances, like overfishing (Hoey & Bellwood 2009). This could greatly increase the chance of important ecological roles, like cropping algae in crevices aiding coral recruit survival, being lost if care is not taken to preserve them.

Ecosystem based management (EBM) is a management strategy that tries to maximize ecosystem resilience and health, and has seen an increase in use and application in the marine realm, including coral reef ecosystems over the past few decades (Levin & Lubchenco 2008; Fenner 2012; Dominguez-Tejo et al. 2016). The

EBM of coral reefs should allow important ecological functions to remain intact by focusing management efforts on preserving important species interactions (Mumby &

Steneck 2008). The broad functional group classification scheme for nominally herbivorous fishes has been useful for managers trying to implement EBM, and simple

50 guidelines for monitoring these species have been created to help managers better understand the vulnerability of their management areas to change (Green & Bellwood

2009). While these convenient guidelines can allow managers to protect these processes, these processes are more complex as previously stated and demonstrated in this study.

Without expanding the actual functional roles of these species and including this in

EBM, key services could disappear if certain fish species are lost because of exclusion from management plans or assumed functional redundancy within the broader grouping.

Previous guidelines could be expanded to include coppers, scrapers, and excavators as separate functional groups. However, this again simplifies the ecological role of these species by assuming similar roles within only slightly smaller groupings. The isotopic differences found in this study between Acanthurus nigrofuscus, and Zebrasoma desjardinii, two croppers, implies that even though they appear to have similar functional roles, there may be yet undescribed but important differences between these species; this may also apply to scraping parrotfishes. Managers should try and take care when deciding management plans to preserve the total diversity of herbivorous reef fish functions when possible so they ensure all ecosystem functions are included in their management plan, with special care taken to preserve functions with low levels of redundancy (Navia et al. 2016). It may not be possible to protect every fish species on a coral reef, but managers should be aware of the functions that might be lost if certain species decline in abundance (Sale 2008).

4.5 Future Directions

Though the compound-specific stable isotope analysis of δ13C from essential amino acids proved quite useful for examining ecological differences between nominally

51 herbivorous fishes, there is still room to improve the power of this analysis as further discussed below:

• The patterns of δ13C values of end-members, especially algae require further

examination. Though most algae were generally separated from each other in

multivariate space, no distinctions were found between taxonomic groupings.

However, algae in this study came from a broad number of families, and it could be

possible that stronger differences would emerge by sampling more species as well as

some closely related individuals.

• End-members were all collected at the same time in the morning, but it is possible

there could be some small changes isotopic signatures throughout the day. It is

known that the nutritional quality of algae can change during the day depending on if

and when photoinhibition occurs (Zemke-White et al. 2002) though it remains to be

tested if this could significantly alter isotopic values.

• Ontogenetic and seasonal variation in essential amino acid δ13C signatures were

not investigated in study, as only adult fish were collected and all samples were

collected over about a month long period in late spring. Ontogenetic variation in bulk

isotopic signatures has been previously documented in some parrotfish species

(Plass-Johnsen et al. 2013), and the fine resolution of CSIA may prove useful at

discerning even more subtle changes in resource use. In fishes, the isotopic turnover

time for muscle can vary by species (Gaston & Suthers 2004; Perga & Gerdeaux

2005) though is reported to be in the range of a few weeks for some nominally

herbivorous fishes (Plass-Johnsen et al. 2015). If the turnover time is longer for some

species, it could obscure seasonal changes in the isotopic values; in this instance it

would be preferable to sample blood or liver tissue which has a faster turnover rate

(Buchheiser & Latour 2010). Further investigation of muscle isotope turnover times

may be warranted for essential amino acids as well to see if there is any difference

when compared to bulk muscle tissue turnover rates.

• Analysis of δ13C was used in this study to trace the flow of nutrients from end-

members through to fishes. However, δ15N may also be measured as a way to further

investigate nutrient flow and trophic structure (Chikaraishi et al. 2009). Including this

analysis with δ13C analysis would allow for a much clearer understanding of coral

reef fish food web ecology in the future. The analysis of fatty acids in addition to

essential amino acid isotopes may even further allow for a more detailed picture of

interactions between fishes and their targeted resources (Piché et al. 2010).

5. CONCLUSION

Using the δ13C values of essential amino acids proved a useful technique for investigating the functional diversity within nominally herbivorous coral reef fishes.

Fishes often thought to be consuming the same material, notably grazing surgeonfishes and parrotfishes, appear to be targeting different algal or microbial material over the reef.

Even within groups, such as cropping surgeonfishes, which are thought to consume the same apical portions of algae within crevices on the reef, were found to be isotopically distinct. Scraping parrotfishes, which also seem to feed over similar reef substrate, appear to have differences in their preferred diet. This highlights the true diversity of fish-reef interactions that we are only beginning to understand (Clements et al. 2016). The compound-specific isotope analysis approach used proved highly useful at perceiving these differences, and has great potential to be expanded to investigate fine-scale ecological dissimilarities in other groups of fishes and reef organisms. These results also highlight the importance coral reef managers should place on preserving ecosystem functioning to promote reef resilience; the interactions between nominally herbivorous fishes and algae are highly complex and should not be considered equal between all fishes within a broadly labeled grouping. Until more is understood about the fine-scale details of these interactions, managers should remain cautious when outlining management plans for these ecologically important reef fishes to try and preserve the diversity of the key functions these fishes carry out.

REFERENCES

Abelson, P. H., & Hoering, T. C. (1961). Carbon isotope fractionation in formation of amino acids by photosynthetic organisms. Proceedings of the National Academy of Sciences, 47(5), 623-632.

Adam, T. C., Kelley, M., Ruttenberg, B. I., & Burkepile, D. E. (2015). Resource partitioning along multiple niche axes drives functional diversity in parrotfishes on Caribbean coral reefs. Oecologia, 179(4), 1173-1185.

Afeworki, Y., Bruggemann, J. H., & Videler, J. J. (2011). Limited flexibility in resource use in a coral reef grazer foraging on seasonally changing algal communities. Coral Reefs, 30(1), 109-122.

Afeworki, Y., Zekeria, Z. A., Videler, J. J., & Bruggemann, J. H. (2013). Food intake by the parrotfish Scarus ferrugineus varies seasonally and is determined by temperature, size and territoriality. Marine Ecology Progress Series, 489, 213-224.

Alwany, M. A. (2008). Behavioural responses towards different light intensities of two Red Sea surgeonfishes. Egyptian J. Exp. Biol.(Zool.), 4, 199-204.

Alwany, M., Thaler, E., & Stachowitsch, M. (2005). Territorial behaviour of Acanthurus sohal and Plectroglyphidodon leucozona on the fringing Egyptian Red Sea reefs. Environmental Biology of Fishes, 72(3), 321-334.

Arthur, K. E., Kelez, S., Larsen, T., Choy, C. A., & Popp, B. N. (2014). Tracing the biosynthetic source of essential amino acids in marine turtles using δ13C fingerprints. Ecology, 95(5), 1285-1293.

Bardgett, R. D., Richter, A., Bol, R., Garnett, M. H., Bäumler, R., Xu, X., Lopez-Capel E, Manning DA, Hobbs PJ, Hartley IR, Wanek W. (2007). Heterotrophic microbial communities use ancient carbon following glacial retreat. Biology Letters, 3(5), 487-490.

Bazely, D. R., & Jefferies, R. L. (1986). Changes in the composition and standing crop of salt-marsh communities in response to the removal of a grazer. The Journal of Ecology, 693-706.

Bearhop, S., Adams, C. E., Waldron, S., Fuller, R. A., & MacLeod, H. (2004). Determining trophic niche width: a novel approach using stable isotope analysis. Journal of Animal Ecology, 73(5), 1007-1012.

Belliveau, S. A., & Paul, V. J. (2002). Effects of herbivory and nutrients on the early colonization of crustose coralline and fleshy algae. Marine Ecology Progress Series, 232, 105-114.

Bellwood, D. R., & Choat, J. H. (1990). A functional analysis of grazing in parrotfishes (family Scaridae): the ecological implications. Environmental Biology of Fishes, 28(1-4), 189-214.

Bellwood, D. R., Hoey, A. S., & Choat, J. H. (2003). Limited functional redundancy in high diversity systems: resilience and ecosystem function on coral reefs. Ecology Letters, 6(4), 281-285.

Bellwood, D. R., Hoey, A. S., & Hughes, T. P. (2011). Human activity selectively impacts the ecosystem roles of parrotfishes on coral reefs. Proceedings of the Royal Society of London B: Biological Sciences, rspb20111906.

Bellwood, D. R., Hughes, T. P., Folke, C., & Nyström, M. (2004). Confronting the coral reef crisis. Nature, 429(6994), 827-833.

Bellwood, D. R., Hughes, T. P., & Hoey, A. S. (2006). Sleeping functional group drives coral-reef recovery. Current Biology, 16(24), 2434-2439.

Bender, M. M. (1968). Mass spectrometric studies of carbon 13 variations in corn and other grasses. Radiocarbon, 10(2), 468-472.

Berumen ML, Hoey AS, Bass WH, Bouwmeester J, Catania D, Cochran JE, Khalil MT, Miyake S, Mughal MR, Spaet JL, Saenz-Agudelo P. (2013). The status of coral reef ecology research in the Red Sea. Coral Reefs, 32(3), 737-748.

Bonaldo, R. M., Hoey, A. S., & Bellwood, D. R. (2014). The ecosystem roles of parrotfishes on tropical reefs. Oceanography and Marine Biology: An Annual Review, 52, 81-132.

Borman, A., Wood, T. R., Black, H. C., Anderson, E. G., Oestekling, M., Womack, M., & Rose, W. C. (1946). The role of arginine in growth, with some observations on the effects of argininic acid. Journal of Biological Chemistry, 166, 585-594.

Bos, A. R., Cruz-Rivera, E., & Sanad, A. M. (2016). Herbivorous fishes Siganus rivulatus (Siganidae) and Zebrasoma desjardinii (Acanthuridae) feed on Ctenophora and Scyphozoa in the Red Sea. Marine Biodiversity, 1-4.

Box, S. J., & Mumby, P. J. (2007). Effect of macroalgal competition on growth and survival of juvenile Caribbean corals. Marine Ecology Progress Series, 342, 139-149.

Brandl, S. J., & Bellwood, D. R. (2016). Microtopographic refuges shape consumer- producer dynamics by mediating consumer functional diversity. Oecologia, 1-15.

Bruggemann, J. H., Van Kessel, A. M., Van Rooij, J. M., & Breeman, A. M. (1996). Bioerosion and sediment ingestion by the Caribbean parrotfish Scarus vetula and Sparisoma viride: implications of fish size, feeding mode and habitat use. Marine Ecology Progress Series, 134, 59-71.

Buchheister, A., & Latour, R. J. (2010). Turnover and fractionation of carbon and nitrogen stable isotopes in tissues of a migratory coastal predator, summer flounder (Paralichthys dentatus). Canadian Journal of Fisheries and Aquatic Sciences, 67(3), 445-461.

Burkepile, D. E., & Hay, M. E. (2006). Herbivore vs. nutrient control of marine primary producers: context-dependent effects. Ecology, 87(12), 3128-3139.

Burkepile, D. E., & Hay, M. E. (2011). Feeding complementarity versus redundancy among herbivorous fishes on a Caribbean reef. Coral Reefs, 30(2), 351-362.

Carassou, L., Kulbicki, M., Nicola, T. J., & Polunin, N. V. (2008). Assessment of fish trophic status and relationships by stable isotope data in the coral reef lagoon of New Caledonia, southwest Pacific. Aquatic Living Resources, 21(1), 1-12.

Cheal, A. J., MacNeil, M. A., Cripps, E., Emslie, M. J., Jonker, M., Schaffelke, B., & Sweatman, H. (2010). Coral–macroalgal phase shifts or reef resilience: links with diversity and functional roles of herbivorous fishes on the Great Barrier Reef. Coral Reefs, 29(4), 1005-1015.

Chen, W. P., Yang, X. Y., Hegeman, A. D., Gray, W. M., & Cohen, J. D. (2010). Microscale analysis of amino acids using gas chromatography–mass spectrometry after methyl chloroformate derivatization. Journal of Chromatography B, 878(24), 2199-2208.

Cherel, Y., Hobson, K. A., & Hassani, S. (2005). Isotopic discrimination between food and blood and feathers of captive penguins: implications for dietary studies in the wild. Physiological and Biochemical Zoology, 78(1), 106-115.

Chikaraishi, Y., Ogawa, N.O., Kashiyama, Y., Takano, Y., Suga, H., Tomitani, A., Miyashita, H., Kitazato, H. and Ohkouchi, N. (2009). Determination of aquatic food‐web structure based on compound‐specific nitrogen isotopic composition of amino acids. Limnology and Oceanography: Methods, 7(11), 740-750.

Choat, J., Clements, K., & Robbins, W. (2002). The trophic status of herbivorous fishes on coral reefs I: dietary analysis. Marine Biology, 140(3), 613-623.

Choat, J. H., Robbins, W. D., & Clements, K. D. (2004). The trophic status of herbivorous fishes on coral reefs II: food processing modes and trophodynamics. Marine Biology, 145(3), 445-454.

Clements, K. D., & Choat, J. H. (1995). Fermentation in tropical marine herbivorous fishes. Physiological Zoology, 355-378.

Clements, K. D., & Choat, J. H. (1997). Comparison of herbivory in the closely-related marine fish genera Girella and Kyphosus. Marine Biology, 127(4), 579-586.

Clements, K. D., German, D. P., Piché, J., Tribollet, A., & Choat, J. H. (2016). Integrating ecological roles and trophic diversification on coral reefs: multiple lines of evidence identify parrotfishes as microphages. Biological Journal of the Linnean Society.

Crossman, D. J., Choat, J. H., & Clements, K. D. (2005). Nutritional ecology of nominally herbivorous fishes on coral reefs. Marine Ecology Progress Series, 296, 129- 142.

Davila, A. M., Blachier, F., Gotteland, M., Andriamihaja, M., Benetti, P. H., Sanz, Y., & Tomé, D. (2013). Intestinal luminal nitrogen metabolism: role of the gut microbiota and consequences for the host. Pharmacological Research, 68(1), 95-107. de la Morinière, E. C., Pollux, B. J. A., Nagelkerken, I., Hemminga, M. A., Huiskes, A. H. L., & Van der Velde, G. (2003). Ontogenetic dietary changes of coral reef fishes in the mangrove-seagrass-reef continuum: stable isotopes and gut-content analysis. Marine Ecology Progress Series, 246, 279-289. de la Torre-Castro, M., Björk, M., Eklöf, J., & Rönnbäck, P. (2008). Seagrass importance in food provisioning services: fish stomach content as a link between seagrass meadows and local fisheries. Western Indian Ocean Journal of Marine Science, 7(1).

Detling, J. K. (1988). Grasslands and savannas: regulation of energy flow and nutrient cycling by herbivores. In Concepts of Ecosystem Ecology (pp. 131-148). Springer New York.

Domínguez-Tejo, E., Metternicht, G., Johnston, E., & Hedge, L. (2016). Marine Spatial Planning advancing the Ecosystem-Based Approach to coastal zone management: A review. Marine Policy, 72, 115-130.

Estes, J. A., Tinker, M. T., Williams, T. M., & Doak, D. F. (1998). Killer whale predation on sea otters linking oceanic and nearshore ecosystems. Science, 282(5388), 473-476.

Falk, J. M., Schmidt, N. M., Christensen, T. R., & Ström, L. (2015). Large herbivore grazing affects the vegetation structure and greenhouse gas balance in a high arctic mire. Environmental Research Letters, 10(4), 045001.

Falkowski, P. G. (1991). Species variability in the fractionation of δ13C and δ12C by marine phytoplankton. Journal of Plankton Research, 13(supp1), 21-28.

Fenner, D. (2012). Challenges for managing fisheries on diverse coral reefs. Diversity, 4(1), 105-160.

Fitzgerald, L. M., & Szmant, A. M. (1997). Biosynthesis of ‘essential’ amino acids by scleractinian corals. Biochemical Journal, 322(1), 213-221.

Fox, R. J., & Bellwood, D. R. (2008). Remote video bioassays reveal the potential feeding impact of the rabbitfish Siganus canaliculatus (f: Siganidae) on an inner-shelf reef of the Great Barrier Reef. Coral Reefs, 27(3), 605-615.

Fox, R. J., Sunderland, T. L., Hoey, A. S., & Bellwood, D. R. (2009). Estimating ecosystem function: contrasting roles of closely related herbivorous rabbitfishes (Siganidae) on coral reefs. Marine Ecology Progress Series, 385, 261-269.

Fry, B. (2006). Stable Isotope Ecology. Springer Science & Business Media.

Fry, B. (2013). Alternative approaches for solving underdetermined isotope mixing problems. Marine Ecology Progress Series, 472, 1-13.

Fry, B., & Sherr, E. B. (1989). δ13C measurements as indicators of carbon flow in marine and freshwater ecosystems. In Stable Isotopes in Ecological Research (pp. 196-229). Springer New York.

Furbank, R. T., & Taylor, W. C. (1995). Regulation of photosynthesis in C3 and C4 plants: a molecular approach. The Plant Cell, 7(7), 797.

Gannes, L. Z., O’Brien, D. M., & del Rio, C. M. (1997). Stable isotopes in animal ecology: assumptions, caveats, and a call for more laboratory experiments. Ecology, 78(4), 1271-1276.

Gao, J., Barzel, B., & Barabási, A. L. (2016). Universal resilience patterns in complex networks. Nature, 530(7590), 307-312.

Gaston, T. F., & Suthers, I. M. (2004). Spatial variation in δ13C and δ15N of liver, muscle and bone in a rocky reef planktivorous fish: the relative contribution of sewage. Journal of Experimental Marine Biology and Ecology, 304(1), 17-33.

Ghanbari, M., Kneifel, W., & Domig, K. J. (2015). A new view of the fish gut microbiome: advances from next-generation sequencing. Aquaculture, 448, 464-475.

Goatley, C. H., Hoey, A. S., & Bellwood, D. R. (2012). The role of turtles as coral reef macroherbivores. PloS One, 7(6), e39979.

Gorospe, K. D., & Karl, S. A. (2010). Small-scale spatial analysis of in situ sea temperature throughout a single coral patch reef. Journal of Marine Biology, 2011.

Graham, B. S., Koch, P. L., Newsome, S. D., McMahon, K. W., & Aurioles, D. (2010). Using isoscapes to trace the movements and foraging behavior of top predators in oceanic ecosystems. In Isoscapes (pp. 299-318). Springer Netherlands.

Grange, J. S., Rybarczyk, H., & Tribollet, A. (2015). The three steps of the carbonate biogenic dissolution process by microborers in coral reefs (New Caledonia). Environmental Science and Pollution Research, 22(18), 13625-13637.

Green, A. L., & Bellwood, D. R. (Eds.). (2009). Monitoring functional groups of herbivorous reef fishes as indicators of coral reef resilience: a practical guide for coral reef managers in the Asia Pacific Region (No. 7). IUCN.

Halford, J. C., Smith, B. K., & Blundell, J. E. (2000). Neural and Metabolic Control of Macronutrient Intake. CRC Press

Heenan, A., & Williams, I. D. (2013). Monitoring herbivorous fishes as indicators of coral reef resilience in American Samoa. PloS one, 8(11), e79604.

Hirst, A. G., & Lucas, C. H. (1998). Salinity influences body weight quantification in the scyphomedusa Aurelia aurita: important implications for body weight determination in gelatinous zooplankton. Marine Ecology Progress Series, 165, 259-269.

Hobbs, R. J., & Huenneke, L. F. (1992). Disturbance, diversity, and invasion: implications for conservation. Conservation Biology, 6(3), 324-337.

Hobson, K. A., Fisk, A., Karnovsky, N., Holst, M., Gagnon, J. M., & Fortier, M. (2002). A stable isotope (δ13C, δ15N) model for the North Water food web: implications for evaluating trophodynamics and the flow of energy and contaminants. Deep Sea Research Part II: Topical Studies in Oceanography, 49(22), 5131-5150.

Hoey, A. S., & Bellwood, D. R. (2009). Limited functional redundancy in a high diversity system: single species dominates key ecological process on coral reefs. Ecosystems, 12(8), 1316-1328.

Hoey, A. S., Feary, D. A., Burt, J. A., Vaughan, G., Pratchett, M. S., & Berumen, M. L. (2016). Regional variation in the structure and function of parrotfishes on Arabian reefs. Marine Pollution Bulletin, 105(2), 524-531.

Hopkins III, J. B., & Ferguson, J. M. (2012). Estimating the diets of animals using stable isotopes and a comprehensive Bayesian mixing model. PLoS One, 7(1), e28478.

Hughes, T. P. (1994). Catastrophes, phase shifts, and large-scale degradation of a Caribbean coral reef. Science-AAAS-Weekly Paper Edition, 265(5178), 1547-1551.

Hughes, T. P., Bellwood, D. R., Folke, C., Steneck, R. S., & Wilson, J. (2005). New paradigms for supporting the resilience of marine ecosystems. Trends in Ecology & Evolution, 20(7), 380-386.

Hughes, T. P., Rodrigues, M. J., Bellwood, D. R., Ceccarelli, D., Hoegh-Guldberg, O., McCook, L., Moltschaniwskyj, N., Pratchett, M.S., Steneck, R.S. and Willis, B. (2007). Phase shifts, herbivory, and the resilience of coral reefs to climate change. Current Biology, 17(4), 360-365.

Hughes, T. P., Graham, N. A., Jackson, J. B., Mumby, P. J., & Steneck, R. S. (2010). Rising to the challenge of sustaining coral reef resilience. Trends in Ecology & Evolution, 25(11), 633-642.

Ishii, H., & Tanaka, F. (2001). Food and feeding of Aurelia aurita in Tokyo Bay with an analysis of stomach contents and a measurement of digestion times. In Jellyfish Blooms: Ecological and Societal Importance (pp. 311-320). Springer Netherlands.

Jim, S., Jones, V., Ambrose, S. H., & Evershed, R. P. (2006). Quantifying dietary macronutrient sources of carbon for bone collagen biosynthesis using natural abundance stable carbon isotope analysis. British Journal of Nutrition, 95(06), 1055-1062.

Jin, D., Kite-Powell, H., Hoagland, P., & Solow, A. (2012). A bioeconomic analysis of traditional fisheries in the Red Sea. Marine Resource Economics, 27(2), 137-148.

Jones, G. P., & Andrew, N. L. (1990). Herbivory and patch dynamics on rocky reefs in temperate Australasia: the roles of fish and sea urchins. Australian Journal of Ecology, 15(4), 505-520.

Jones, C. G., Lawton, J. H., & Shachak, M. (1997). Positive and negative effects of organisms as physical ecosystem engineers. Ecology, 78(7), 1946-1957.

Khalil, M. T., Cochran, J. E., & Berumen, M. L. (2013). The abundance of herbivorous fish on an inshore Red Sea reef following a mass coral bleaching event. Environmental Biology of Fishes, 96(9), 1065-1072.

Karako-Lampert, S., Katcoff, D. J., Achituv, Y., Dubinsky, Z., & Stambler, N. (2004). Do clades of symbiotic dinoflagellates in scleractinian corals of the Gulf of Eilat (Red Sea) differ from those of other coral reefs? Journal of Experimental Marine Biology and Ecology, 311(2), 301-314.

Karasov, W. H., & del Rio, C. M. (2007). Physiological Ecology: How Animals Process Energy, Nutrients, and Toxins (p. 744). Princeton: Princeton University Press.

Klee, M. S., & Grob, R. L. (1985). Modern Practice of Gas Chromatography. John Wiley & Sons.

Klueter, A., Crandall, J. B., Archer, F. I., Teece, M. A., & Coffroth, M. A. (2015). Taxonomic and environmental variation of metabolite profiles in marine dinoflagellates of the genus Symbiodinium. Metabolites, 5(1), 74-99.

Kremer, B. P., & Küppers, U. (1977). Carboxylating enzymes and pathway of photosynthetic carbon assimilation in different marine algae—Evidence for the C4- pathway? Planta, 133(2), 191-196.

LaJeunesse, T. C. (2005). “Species” radiations of symbiotic dinoflagellates in the Atlantic and Indo-Pacific since the Miocene-Pliocene transition. Molecular Biology and Evolution, 22(3), 570-581.

LaJeunesse, T. C., Thornhill, D. J., Cox, E. F., Stanton, F. G., Fitt, W. K., & Schmidt, G. W. (2004). High diversity and host specificity observed among symbiotic dinoflagellates in reef coral communities from Hawaii. Coral Reefs, 23(4), 596-603.

Larsen, T., Taylor, D. L., Leigh, M. B., & O'Brien, D. M. (2009). Stable isotope fingerprinting: a novel method for identifying plant, fungal, or bacterial origins of amino acids. Ecology, 90(12), 3526-3535.

Larsen, T., Wooller, M. J., Fogel, M. L., & O'Brien, D. M. (2012). Can amino acid carbon isotope ratios distinguish primary producers in a mangrove ecosystem? Rapid Communications in Mass Spectrometry, 26(13), 1541-1548.

Larsen, T., Ventura, M., Andersen, N., O’Brien, D. M., Piatkowski, U., & McCarthy, M. D. (2013). Tracing carbon sources through aquatic and terrestrial food webs using amino acid stable isotope fingerprinting. PLoS One, 8(9), e73441.

Layman, C. A. (2007). What can stable isotope ratios reveal about mangroves as fish habitat? Bulletin of Marine Science, 80(3), 513-527.

Layman, C. A., Giery, S. T., Buhler, S., Rossi, R., Penland, T., Henson, M. N., Bogdanoff AK, Cove MV, Irizarry AD, Schalk CM., & Archer, S. K. (2015). A primer on the history of food web ecology: fundamental contributions of fourteen researchers. Food Webs, 4, 14-24.

Levin, S. A., & Lubchenco, J. (2008). Resilience, robustness, and marine ecosystem- based management. Bioscience, 58(1), 27-32.

Logan, J. M., & Lutcavage, M. E. (2013). Assessment of trophic dynamics of cephalopods and large pelagic fishes in the central North Atlantic Ocean using stable isotope analysis. Deep Sea Research Part II: Topical Studies in Oceanography, 95, 63- 73.

Lokrantz, J., Nyström, M., Thyresson, M., & Johansson, C. (2008). The non-linear relationship between body size and function in parrotfishes. Coral Reefs, 27(4), 967-974.

Mantyka, C. S., & Bellwood, D. R. (2007). Direct evaluation of macroalgal removal by herbivorous coral reef fishes. Coral Reefs, 26(2), 435-442.

Marshell, A., & Mumby, P. J. (2012). Revisiting the functional roles of the surgeonfish Acanthurus nigrofuscus and Ctenochaetus striatus. Coral Reefs, 31(4), 1093-1101.

Mayntz, D., Raubenheimer, D., Salomon, M., Toft, S., & Simpson, S. J. (2005). Nutrient- specific foraging in invertebrate predators. Science, 307(5706), 111-113.

McCann, K. S. (2000). The diversity–stability debate. Nature, 405(6783), 228-233.

McCloskey, L. R., & Muscatine, L. (1984). Production and respiration in the Red Sea coral Stylophora pistillata as a function of depth. Proceedings of the Royal Society of London B: Biological Sciences, 222(1227), 215-230.

McCook, L., Jompa, J., & Diaz-Pulido, G. (2001). Competition between corals and algae on coral reefs: a review of evidence and mechanisms. Coral Reefs, 19(4), 400-417.

McCutchan, J. H., Lewis, W. M., Kendall, C., & McGrath, C. C. (2003). Variation in trophic shift for stable isotope ratios of carbon, nitrogen, and sulfur. Oikos, 102(2), 378- 390.

McMahon, K. W., Fogel, M. L., Elsdon, T. S., & Thorrold, S. R. (2010). Carbon isotope fractionation of amino acids in fish muscle reflects biosynthesis and isotopic routing from dietary protein. Journal of Animal Ecology, 79(5), 1132-1141.

McMahon, K. W., Fogel, M. L., Johnson, B. J., Houghton, L. A., & Thorrold, S. R. (2011). A new method to reconstruct fish diet and movement patterns from δ13C values in otolith amino acids. Canadian Journal of Fisheries and Aquatic Sciences, 68(8), 1330- 1340.

McMahon, K. W., Hamady, L. L., & Thorrold, S. R. (2013b). Ocean ecogeochemistry: a review. Oceanography and Marine Biology: An Annual Review, 51, 327-374.

McMahon, K. W., Thorrold, S. R., Houghton, L. A., & Berumen, M. L. (2016). Tracing carbon flow through coral reef food webs using a compound-specific stable isotope approach. Oecologia, 180(3), 809-821.

Meier-Augenstein, W. (1999). Applied gas chromatography coupled to isotope ratio mass spectrometry. Journal of Chromatography A, 842(1), 351-371.

Metcalfe, D. B., Asner, G. P., Martin, R. E., Silva Espejo, J. E., Huasco, W. H., Farfán Amézquita, F. F., Carranza‐Jimenez L, Galiano Cabrera DF, Baca LD, Sinca F, & Quispe, H. (2014). Herbivory makes major contributions to ecosystem carbon and nutrient cycling in tropical forests. Ecology Letters, 17(3), 324-332.

Miyake, S., Ngugi, D. K., & Stingl, U. (2015). Diet strongly influences the gut microbiota of surgeonfishes. Molecular Ecology, 24(3), 656-672.

Miyake, S., Ngugi, D. K., & Stingl, U. (2016). Phylogenetic Diversity, Distribution, and Cophylogeny of Giant Bacteria (Epulopiscium) with their Surgeonfish Hosts in the Red Sea. Frontiers in Microbiology, 7.

Moroney, J. V., & Somanchi, A. (1999). How do algae concentrate CO2 to increase the efficiency of photosynthetic carbon fixation? Plant Physiology, 119(1), 9-16.

Mumby, P. J., & Steneck, R. S. (2008). Coral reef management and conservation in light of rapidly evolving ecological paradigms. Trends in Ecology & Evolution, 23(10), 555- 563.

Muscatine, L. (1990). The role of symbiotic algae in carbon and energy flux in reef corals. Ecosystems of the World, 25, 75-87.

Muscatine, L., Porter, J. W., & Kaplan, I. R. (1989). Resource partitioning by reef corals as determined from stable isotope composition. Marine Biology, 100(2), 185-193.

Montgomery, W. L., Myrberg, A. A., & Fishelson, L. (1989). Feeding ecology of surgeonfishes (Acanthuridae) in the northern Red Sea, with particular reference to Acanthurus nigrofuscus (Forsskål). Journal of Experimental Marine Biology and Ecology, 132(3), 179-207.

Navia, A. F., Cruz-Escalona, V. H., Giraldo, A., & Barausse, A. (2016). The structure of a marine tropical food web, and its implications for ecosystem-based fisheries management. Ecological Modelling, 328, 23-33.

Nolte, S., Esselink, P., Smit, C., & Bakker, J. P. (2014). Herbivore species and density affect vegetation-structure patchiness in salt marshes. Agriculture, Ecosystems & Environment, 185, 41-47.

Nyström, M. (2006). Redundancy and response diversity of functional groups: implications for the resilience of coral reefs. AMBIO: A Journal of the Human Environment, 35(1), 30-35.

Nyström, M., Graham, N. A. J., Lokrantz, J., & Norström, A. V. (2008). Capturing the cornerstones of coral reef resilience: linking theory to practice. Coral Reefs, 27(4), 795- 809.

O'Farrell, S., Bearhop, S., McGill, R. A., Dahlgren, C. P., Brumbaugh, D. R., & Mumby, P. J. (2014). Habitat and body size effects on the isotopic niche space of invasive lionfish and endangered Nassau grouper. Ecosphere, 5(10), 1-11.

Ogden, J. C., & Lobel, P. S. (1978). The role of herbivorous fishes and urchins in coral reef communities. Environmental Biology of Fishes, 3(1), 49-63.

Olff, H., & Ritchie, M. E. (1998). Effects of herbivores on grassland plant diversity. Trends in Ecology & Evolution, 13(7), 261-265.

Paddack, M. J., Cowen, R. K., & Sponaugle, S. (2006). Grazing pressure of herbivorous coral reef fishes on low coral-cover reefs. Coral Reefs, 25(3), 461-472.

Park, R., & Epstein, S. (1960). Carbon isotope fractionation during photosynthesis. Geochimica et Cosmochimica Acta, 21(1-2), 110-126.

Petchey, O. L., & Gaston, K. J. (2006). Functional diversity: back to basics and looking forward. Ecology Letters, 9(6), 741-758.

Peterson, B. J., & Fry, B. (1987). Stable isotopes in ecosystem studies. Annual Review of Ecology and Systematics, 293-320.

Perga, M. E., & Gerdeaux, D. (2005). ‘Are fish what they eat’all year round?. Oecologia, 144(4), 598-606.

Piché, J., Iverson, S. J., Parrish, F. A., & Dollar, R. (2010). Characterization of forage fish and invertebrates in the Northwestern Hawaiian Islands using fatty acid signatures: species and ecological groups. Marine Ecology Progress Series, 418, 1-15.

Plass-Johnson, J. G., McQuaid, C. D., & Hill, J. M. (2013). Stable isotope analysis indicates a lack of inter-and intra-specific dietary redundancy among ecologically important coral reef fishes. Coral Reefs, 32(2), 429-440.

Plass‐Johnson, J. G., McQuaid, C. D., & Hill, J. M. (2015). The effects of tissue type and body size on δ13C and δ15N values in parrotfish (Labridae) from Zanzibar, Tanzania. Journal of Applied Ichthyology, 31(4), 633-637.

Post, D. M. (2002). Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology, 83(3), 703-718.

Rasher, D. B., Hoey, A. S., & Hay, M. E. (2013). Consumer diversity interacts with prey defenses to drive ecosystem function. Ecology, 94(6), 1347-1358.

Raubenheimer, D., & Simpson, S. J. (1997). Integrative models of nutrient balancing: application to insects and vertebrates. Nutrition Research Reviews, 10(01), 151-179.

Raven, J. A. (1997). Inorganic carbon acquisition by marine autotrophs. Advances in Botanical Research, 27, 85-209.

Rowan, R., & Knowlton, N. (1995). Intraspecific diversity and ecological zonation in coral-algal symbiosis. Proceedings of the National Academy of Sciences, 92(7), 2850- 2853.

Sale, P. F. (2008). Management of coral reefs: where we have gone wrong and what we can do about it. Marine Pollution Bulletin, 56(5), 805-809.

Sawall, Y., Al-Sofyani, A., Banguera-Hinestroza, E., & Voolstra, C. R. (2014). Spatio- temporal analyses of Symbiodinium physiology of the coral Pocillopora verrucosa along large-scale nutrient and temperature gradients in the Red Sea. PloS one, 9(8), e103179.

Schoenberg, D. A., & Trench, R. K. (1980). Genetic variation in Symbiodinium (= Gymnodinium) microadriaticum (Freudenthal), and specificity in its symbiosis with marine invertebrates. I. Isoenzyme and soluble protein patterns of axenic cultures of Symbiodinium microadriaticum. Proceedings of the Royal Society of London B: Biological Sciences, 207(1169), 405-427.

Scott, J. H., O'Brien, D. M., Emerson, D., Sun, H., McDonald, G. D., Salgado, A., & Fogel, M. L. (2006). An examination of the carbon isotope effects associated with amino acid biosynthesis. Astrobiology, 6(6), 867-880.

Sessions, A. L. (2006). Isotope‐ratio detection for gas chromatography. Journal of Separation Science, 29(12), 1946-1961.

Shaikh, E. A., Roff, J. C., & Dowidar, N. M. (1986). Phytoplankton ecology and production in the Red Sea off Jiddah, Saudi Arabia. Marine Biology, 92(3), 405-416.

Silfer, J. A., Engel, M. H., Macko, S. A., & Jumeau, E. J. (1991). Stable carbon isotope analysis of amino acid enantiomers by conventional isotope ratio mass spectrometry and combined gas chromatography/isotope ratio mass spectrometry. Analytical Chemistry, 63(4), 370-374.

Smetacek, V., Scharek, R., & Nöthig, E. M. (1990). Seasonal and regional variation in the pelagial and its relationship to the life history cycle of krill. In Antarctic Ecosystems (pp. 103-114). Springer Berlin Heidelberg.

Smith, D. C., Bassham, J. A., & Kirk, M. (1961). Dynamics of the photosynthesis of carbon compounds II. Amino acid synthesis. Biochimica et Biophysica Acta, 48(2), 299- 313.

Smith, J. E., Hunter, C. L., & Smith, C. M. (2010). The effects of top–down versus bottom–up control on benthic coral reef community structure. Oecologia, 163(2), 497- 507.

Spaet, J. L., & Berumen, M. L. (2015). Fish market surveys indicate unsustainable elasmobranch fisheries in the Saudi Arabian Red Sea. Fisheries Research, 161, 356-364.

Streit, R. P., Hoey, A. S., & Bellwood, D. R. (2015). Feeding characteristics reveal functional distinctions among browsing herbivorous fishes on coral reefs. Coral Reefs, 34(4), 1037-1047.

Tieszen, L. L., Boutton, T. W., Tesdahl, K. G., & Slade, N. A. (1983). Fractionation and turnover of stable carbon isotopes in animal tissues: implications for δ13C analysis of diet. Oecologia, 57(1-2), 32-37.

Titlyanov, E. A., Titlyanova, T. V., Yamazato, K., & Van Woesik, R. (2001). Photo- acclimation dynamics of the coral Stylophora pistillata to low and extremely low light. Journal of Experimental Marine Biology and Ecology, 263(2), 211-225.

Tribollet, A. (2008). The boring microflora in modern coral reef ecosystems: a review of its roles. In Current developments in bioerosion (pp. 67-94). Springer Berlin Heidelberg.

Uthicke, S. (1999). Sediment bioturbation and impact of feeding activity of Holothuria (Halodeima) atra and Stichopus chloronotus, two sediment feeding holothurians, at Lizard Island, Great Barrier Reef. Bulletin of Marine Science, 64(1), 129-141.

Vander Zanden, M. J., & Vadeboncoeur, Y. (2002). Fishes as integrators of benthic and pelagic food webs in lakes. Ecology, 83(8), 2152-2161.

Walsh, R. G., He, S., & Yarnes, C. T. (2014). Compound‐specific δ13C and δ15N analysis of amino acids: a rapid, chloroformate‐based method for ecological studies. Rapid Communications in Mass Spectrometry, 28(1), 96-108.

Wang, J. T., & Douglas, A. E. (1999). Essential amino acid synthesis and nitrogen recycling in an alga–invertebrate symbiosis. Marine Biology, 135(2), 219-222.

Ward, E.J., Semmens, B.X., & Schindler, D.E. (2010). Including source uncertainty and information in the analysis of stable isotope mixing models. Environmental Science and Technology, 44, 4645-4650.

Wegley, L., Edwards, R., Rodriguez‐Brito, B., Liu, H., & Rohwer, F. (2007). Metagenomic analysis of the microbial community associated with the coral Porites astreoides. Environmental Microbiology, 9(11), 2707-2719.

Wilson, S. K., Bellwood, D. R., Choat, J. H., & Furnas, M. J. (2003). Detritus in the epilithic algal matrix and its use by coral reef fishes. Oceanography and Marine Biology, 41, 279-310.

Wyatt, A. S. J., Waite, A. M., & Humphries, S. (2012). Stable isotope analysis reveals community-level variation in fish trophodynamics across a fringing coral reef. Coral Reefs, 31(4), 1029-1044.

Zemke-White, L. W., Choat, J., & Clements, K. (2002). A re-evaluation of the diel feeding hypothesis for marine herbivorous fishes. Marine Biology, 141(3), 571-579.

Ziegler, M., Roder, C. M., Büchel, C., & Voolstra, C. R. (2014). Limits to physiological plasticity of the coral Pocillopora verrucosa from the central Red Sea. Coral Reefs, 33(4), 1115-1129.

APPENDIX I

Figure 5. Results from a mixing model representing the relative percent contribution of different end- members to the isotopic concentration of all fishes. Species are labeled left to right alphabetically by functional group with browsers, followed by grazers, and excavators and scrapers. The relative contribution was determined from the δ13C values of five essential amino acids (valine, isoleucine, leucine, threonine, and phenylalanine) using the IsotopreR Bayesian stable isotope mixing model. Due to the similarity between some of the macroalgae samples, both Chlorophyts, filamentous cyanobacteria, and Dictyota bartayrsii were combined for analysis.

7.1 IsotopeR Results

Results from the IsotopeR mixing model showed a high reliance on coral as a source of essential amino acids in all parrotfishes (Appendix I, Figure 5). Three species, Scarus ferrugineus, Sc. frenatus, and Hipposcarus harid all showed nearly identical output, with about two-thirds of their essential amino acids from coral and slightly less than one-third from detrital/microalgae. Scarus niger was similar to these parrotfish in its reliance on coral, but relied much more on Porolithon sp. while receiving only 1.35 ± 0.12 % of its

68 essential amino acids from detritus. Chlorurus sordidus relied most heavily on coral with

91.63 ± 0.02 % of its essential amino acids ascribed to this source. All grazers appeared to assimilate essential amino acids nearly exclusively from Turbinaria ornata;

Acanthurus nigrofuscus, A. sohal, and Zebrasoma desjardinii all received more than 99% of their essential amino acids from this source. Ctenochaetus striatus was reported to rely on T. ornata by only 90.15 ± 0.13 %, and relied on Symploca hydnoides next most heavily by 5.13 ± 0.03 %. All browsers had highly similar outputs and relied most heavily on T. ornata (84.70 ± 0.23 % for all) followed by S. hydnoides (11.16 ± 0.18 % for all) as their main sources of carbon for essential amino acids.

7.2 Modeling

Even given the isotopic differences between browsing fishes and the algae they are known to consume, the model used in this study was still able to resolve them well and showed a high reliance on Turbinaria ornata in all species (Appendix I, Figure 5).

The model predictions for others species were not well constrained given what we already know about these grazers, scraping, and excavating species. The model output for grazers placed a very high reliance on T. ornata in all species, even more so than browsers. This can likely be attributed to the fact that they model input was missing an important component of their diet, namely turfing algae. While there does appear to be differences in exactly what kind of turfing algae or other microalgae they may be targeting based off their isotopic signature, no single sample of turfing algae has been sampled to date, so any algae they might normally consume was missing from the model.

Further analysis of turfing algae should be analyzed after collection from different reef microhabitats to help confirm the distinction between some of these grazers. The model outputs also reported all species of parrotfish to have a high reliance on coral. This was

69 not anticipated since most previous studies have identified them as detrital feeders, and given the recent revelations of their diet (Clements et al. 2016), it was expected all species might cluster closer to the detrital end-member, which may more likely resemble the isotopic signature of endolithic algae. Given the universally high reliance on coral though, it may be the case that the isotopic signature of coral samples more closely resembles that of endolithic algae. This remains to be tested yet, as these organisms are difficult to sample (Choat pers. comm.; McMahon pers. comm.), but given the complexity of microbes, including cyanobacteria, that live on corals this must be considered when running future models that include endolithic algae (Wegley et al.

2007). But when running the model in future scenarios after more samples have been measured, it will be important to take care of limiting the number of sources input to the model. Isotope mixing models become undetermined and lose the number of unique possible solutions when the sources outnumber the measurements, or isotope values, by more than n + 1 where n is the number of isotope measurements (Hopkins & Ferguson

2012; Fry 2013). To achieve this, isotopically indistinct end-members may be required to be lumped together for analysis or excluded if it is well known that end-member will not contribute to the consumer’s diet (Ward et al 2010). But given how well the model seemed to work with the browsing herbivores, it demonstrates the potential to reveal resource utilization patterns in the other fishes once the correct sources have been input.