ArticlePDF Available

Amino acid δ13C and δ15N analyses reveal distinct species‐specific patterns of trophic plasticity in a marine symbiosis


Abstract and Figures

Compound‐specific isotope analyses (CSIA) and multivariate “isotope fingerprinting” track biosynthetic sources and reveal trophic interactions in food webs. However, CSIA have not been widely applied in the study of marine symbioses. Here, we exposed a reef coral (Montipora capitata) in symbiosis with Symbiodiniaceae algae to experimental treatments (autotrophy, mixotrophy, heterotrophy) to test for trophic shifts and amino acid (AA) sources using paired bulk (δ13C, δ15N) and AA‐CSIA (δ13CAA, δ15NAA). Treatments did not influence carbon or nitrogen trophic proxies, thereby not supporting nutritional plasticity. Instead, hosts and symbionts consistently overlapped in essential‐ and nonessential‐δ13CAA (11 of 13 amino acids) and trophic‐ and source‐δ15NAA values (9 of 13 amino acids). Host and symbiont trophic‐δ15NAA values positively correlated with a plankton end‐member, indicative of trophic connections and dietary sources for trophic‐AA nitrogen. However, mass balance of AA‐trophic positions (TPGlx–Phe) revealed heterotrophic influences to be highly variable (1–41% heterotrophy). Linear discriminant analysis using M. capitata mean‐normalized essential‐δ13CAA with previously published values (Pocillopora meandrina) showed similar nutrition isotope fingerprints (Symbiodiniaceae vs. plankton) but revealed species‐specific trophic strategies. Montipora capitata and Symbiodiniaceae shared identical AA‐fingerprints, whereas P. meandrina was assigned to either symbiont or plankton nutrition. Thus, M. capitata was 100% reliant on symbionts for essential‐δ13CAA and demonstrated autotrophic fidelity and contrasts with trophic plasticity reported in P. meandrina. While M. capitata AA may originate from host and/or symbiont biosynthesis, AA carbon is Symbiodiniaceae‐derived. Together, AA‐CSIA/isotope fingerprinting advances the study of coral trophic plasticity and are powerful tools in the study of marine symbioses.
Content may be subject to copyright.
Limnol. Oceanogr. 9999, 2021, 118
© 2021 The Authors. Limnology and Oceanography published by Wiley Periodicals LLC on
behalf of Association for the Sciences of Limnology and Oceanography.
doi: 10.1002/lno.11742
Amino acid δ
C and δ
N analyses reveal distinct species-specic
patterns of trophic plasticity in a marine symbiosis
Christopher B. Wall ,
*Natalie J. Wallsgrove,
Ruth D. Gates,
Brian N. Popp
Hawaii Institute of Marine Biology, University of HawaiiatM
anoa, Honolulu, Hawaii
Pacic Biosciences Research Center, University of HawaiiatM
anoa, Honolulu, Hawaii
Department of Earth Sciences, University of HawaiiatM
anoa, Honolulu, Hawaii
Compound-specic isotope analyses (CSIA) and multivariate isotope ngerprintingtrack biosynthetic
sources and reveal trophic interactions in food webs. However, CSIA have not been widely applied in the study
of marine symbioses. Here, we exposed a reef coral (Montipora capitata) in symbiosis with Symbiodiniaceae algae
to experimental treatments (autotrophy, mixotrophy, heterotrophy) to test for trophic shifts and amino acid
(AA) sources using paired bulk (δ
C, δ
N) and AA-CSIA (δ
). Treatments did not inuence carbon
or nitrogen trophic proxies, thereby not supporting nutritional plasticity. Instead, hosts and symbionts consis-
tently overlapped in essential- and nonessential-δ
(11 of 13 amino acids) and trophic- and source-δ
values (9 of 13 amino acids). Host and symbiont trophic-δ
values positively correlated with a plankton
end-member, indicative of trophic connections and dietary sources for trophic-AA nitrogen. However, mass bal-
ance of AA-trophic positions (TP
) revealed heterotrophic inuences to be highly variable (141% hetero-
trophy). Linear discriminant analysis using M. capitata mean-normalized essential-δ
with previously
published values (Pocillopora meandrina) showed similar nutrition isotope ngerprints (Symbiodiniaceae
vs. plankton) but revealed species-specic trophic strategies. Montipora capitata and Symbiodiniaceae shared
identical AA-ngerprints, whereas P. meandrina was assigned to either symbiont or plankton nutrition. Thus,
M. capitata was 100% reliant on symbionts for essential-δ
and demonstrated autotrophic delity and con-
trasts with trophic plasticity reported in P. meandrina. While M. capitata AA may originate from host and/or
symbiont biosynthesis, AA carbon is Symbiodiniaceae-derived. Together, AA-CSIA/isotope ngerprinting
advances the study of coral trophic plasticity and are powerful tools in the study of marine symbioses.
Scleractinian reef corals are mixotrophic cnidarians that
form a mutualistic symbiosis with Symbiodiniaceae micro-
algae (formerly Symbiodinium) (LaJeunesse et al. 2018). The
success of the coral-Symbiodiniaceae symbiosis in nutrient-
poor tropical and subtropical seas is underpinned by nutri-
tional exchanges. Symbiont photosynthesis supports coral
metabolism and calcication (Muscatine et al. 1981) by pro-
viding the host with low-molecular-weight compounds
(i.e., glucose, organic and amino acids), glycoconjugates, and
free fatty acids (Trench 1971b; Markell and Trench 1993;
Papina et al. 2003; Whitehead and Douglas 2003), as well as
additional compounds from photosynthetic-dependent dis-
solved inorganic nitrogen (DIN) assimilation and the uptake
of dissolved free amino acids (DFAAs) (Wang and Doug-
las 1998; Kopp et al. 2013). In return, the coral supplies its
endosymbionts with chemical building blocks derived from
host metabolism (i.e., nitrogenous waste, metabolic inorganic
carbon) (Wang and Douglas 1998). Supplemental to
Symbiodiniaceae nutrition (collectively, autotrophy), reef
corals exploit a variety of water column resources including
dissolved organic compounds, bacteria, detritus, and a range
of phytoplankton and zooplankton to meet energetic needs
(collectively, heterotrophy) (Houlbrèque and Ferrier-
Pagès 2009).
Heterotrophic sources of nutrition are vital to reef
corals, as symbiont-derived autotrophic products are low
in nitrogen and phosphorus (Falkowski et al. 1984). In
healthy individuals, prey capture may account for 1550%
of energy demand (Houlbrèque and Ferrier-Pagès 2009);
This is an open access article under the terms of the Creative Commons
Attribution License, which permits use, distribution and reproduction in
any medium, provided the original work is properly cited.
Additional Supporting Information may be found in the online version of
this article.
Author Contribution Statement: C.B.W., R.D.G., and B.N.P. designed the
project. C.B.W. and N.J.W. performed laboratory analyses. C.B.W. analyzed
the data. C.B.W., B.N.P., and N.J.W. wrote the manuscript.
although, host and endosymbiont assimilation of dissolved
organic (i.e., DFAA, urea) and inorganic nitrogen (i.e., NO
4) may fully sustain coral nitrogen budgets (Grover
et al. 2008). The relative contribution of heterotrophy to the
diet of corals can vary according to coral species (Palardy
et al. 2005), environmental conditions (Fox et al. 2018), and
physiological states (Grottoli et al. 2006). Corals are under
mounting threats from local and global stressors, and trophic
plasticity contributes to coral physiological resilience under
environmental change (Grottoli et al. 2006). Therefore, testing
the causes and consequences of nutritional plasticity in corals
is of substantial interest.
Stable isotope analyses have been widely applied to study
organismal physiology, trophic ecology, and biogeochemical
cycles in terrestrial and marine ecosystems. In particular, varia-
tions in carbon and nitrogen isotopic composition have been
used to determine the trophic ecology of reef corals across spa-
tiotemporal scales (Alamaru et al. 2009) and in response to
physiological stress (Baumann et al. 2014) or discrete
Symbiodiniaceae associations (Baker et al. 2018; Wall
et al. 2020). In some corals, greater reliance on heterotrophic
feeding can be driven by light attenuation across depths
(Alamaru et al. 2009), seawater turbidity (Anthony and
Fabricius 2000), or in regions with high or variable oceanic
productivity (Fox et al. 2018). In addition, heterotrophy may
increase in some corals in response to disruption of the coral-
Symbiodiniaceae symbiosis during bleaching (Grottoli
et al. 2006). Carbon isotope values are most often applied to
corals as a means of identifying greater contributions of
C-depleted heterotrophic prey relative to
autotrophic photosynthates (Muscatine et al. 1989; Laws
et al. 1997). Based on these principles, more negative δ
values in the host, or host values relative to those measured in
the symbionts (i.e., hostsymbiont δ
C values [i.e., δ
]), have been used to indicate a greater reliance on heterotro-
phic nutrition (Rodrigues and Grottoli 2006).
Early pioneering studies showed species-specic isotope
values in corals in low-light, deep habitats, with lower δ
and higher δ
N values interpreted as greater reliance on
heterotrophy (Muscatine et al. 1989; Muscatine and
Kaplan 1994). However, there are important considerations
and limitations to isotope-based inferences on animal
diets, especially in the study of symbioses. For instance,
changes in tissue composition (i.e.,
C-depleted tissue lipids,
protein : lipid : saccharide ratio) (Cooper et al. 2011; Wall
et al. 2019), photosynthesis : respiration ratio (Swart
et al. 2005), and shifts in symbiont communities (Wall
et al. 2020) can all coincide with changes in depth, with each
also inuencing tissue δ
C values. In the case of nitrogen,
spatial and temporal variability in nitrogen sources at the base
of the food web can lead to different isotope values in con-
sumers despite similar trophic levels (Heikoop et al. 2000).
Trophic enrichment factors observed in conventional
predatorprey interactions are also attenuated in corals due to
internal nitrogen cycling between host and symbiont
(Reynaud et al. 2009). Therefore, metabolic effects, tissue com-
position and turnover rates, and different isotopic composi-
tions of source compounds at the base of the food web have
each contributed to the uncertainty in deciphering exibility
from delity in coral trophic strategies.
Fundamental questions remain in our understanding of the
mechanisms governing patterns of stable isotope values in the
coral-Symbiodiniaceae symbiosis. Compound-specic isotope
analyses (CSIA) of individual macromolecules, such as amino
acids and fatty acids (FAs) (see review by Ferrier-Pagès and
Leal 2019), may be useful in disentangling biological changes
in an organisms nutrition (or metabolism) from
unconstrained variance in the form of fractionation-mediated
effects. Accordingly, CSIA may clarify central processes within
the coral holobiont (i.e., nutrient cycling) and illuminate tro-
phic interactions previously obscured in the study of
mixotrophic symbioses. While bulk tissue isotope analyses
represent a collective average of all macromolecules in a sam-
ple, CSIA use molecular approaches based on established met-
abolic pathways and their isotopic discrimination through
trophic transfer and consumer biochemistry (McMahon and
Newsome 2019). For instance, most animals are incapable of
synthesizing all amino acids de novo; therefore, certain amino
acids propagate through food webs without isotope fraction-
ation and are effective tracers of consumer diets (Ohkouchi
et al. 2017; McMahon and Newsome 2019). CSIA also have
greater resolution and can simultaneously provide informa-
tion on dietary food sources, nutrition, and physiology of an
organism (Ohkouchi et al. 2017; Whiteman et al. 2019).
Recently, carbon AA-CSIA (Fox et al. 2019), nitrogen AA-CSIA
(Fujii et al. 2020; Martinez et al. 2020), and FA-CSIA (Teece
et al. 2011) have been applied in corals to examine nutritional
modes at biological (i.e., within and among species) and envi-
ronmental levels (i.e., shallow and mesophotic habitats). How-
ever, to date, few CSIA of reef corals exist and AA-CSIA have
been limited to ecological studies without experimental
manipulation of feeding in controlled laboratory conditions.
Here we use bulk isotope and AA-CSIA of carbon and nitro-
gen to test for trophic plasticity in the reef-building coral
Montipora capitata (Dana, 1846) exposed to manipulative
nutrition treatments (light-by-feeding treatments). While the
inuence of light and food availability on bulk tissue isotope
values in corals has been extensively studied, this work repre-
sents the rst tandem analysis of δ
and δ
values for
corals and Symbiodiniaceae (and a heterotrophic food source)
under manipulative nutrition regimes. We assessed changes in
individual δ
and δ
values and the dynamic patterns
of multivariate amino acid data. Multivariate analyses of
, termed essential amino acid ngerprints,
have previously been used to identify origins of amino acids
across food webs and primary producers (Larsen et al. 2009).
Multivariate amino acid ngerprinting reects an organisms
capacity for amino acid biosynthesis and not the isotope ratio
Wall et al. Coral nutrition and amino acid isotopes
of the sample, which can vary due to local environmental
conditions. Therefore, AA-CSIA ngerprinting data can incor-
porate a wide variety of samples across a range of geographic
locations or environmental settings to determine nutritional
sources. Considering few AA-CSIA data exist for coral and
Symbiodiniaceae, we examined isotope values at several scales
to determine sources of variation while limiting a priori
assumptions. Using individual δ
and δ
values and
isotope ngerprints consisting of 13 amino acids carbon
(essential and nonessential) and nitrogen (trophic and source),
we hypothesized that coral autotrophy under ecologically rele-
vant light treatments would result in convergence in coral-
Symbiodiniaceae amino acid values and isotope ngerprints.
Second, we hypothesized that heterotrophy treatments would
cause divergence in coral-Symbiodiniaceae amino acid values
in favor of greater overlap with isotope values of the plankton
community, and this effect would be most pronounced in the
coral host. Finally, we combined our data with other publi-
shed examples of essential-δ
values in corals to examine
patterns in coral-Symbiodiniaceae and heterotrophic food
sources to test for species-specic patterns of trophic plasticity
and amino acid sources within a single analytical framework.
Materials and methods
Coral collection
On 05 December 2017, two colonies (20 cm diameter) of
M. capitata were collected at <1 m depth from The Point Reef
at Moku o Loe (Coconut Island) at the Hawaii Institute of
Marine Biology (HIMB) (2125059.600N, 15747011.700 W) on the
windward side of Oahu, Hawaii. Colonies were spatially sepa-
rated (5 m from each other) and at different perpendicular dis-
tances to the edge of the reef crest (ca. 5 vs. 10 m). While
symbiont communities were not assessed, we chose colonies
visually identied as brownand orangecolor morphs,
which are common in K
aneohe Bay and are known to harbor
different symbiont communities (Wall et al. 2020). Colonies
were fragmented into nine ramets of approximately equal size
(ca. 10 cm height) and placed into three ow-through aquaria
(50 liter) at a density of three ramets from each coral genet per
treatment (n= 2 genets treatment
). Aquaria received ltered
seawater (1.0 μm) from K
aneohe Bay maintained at ambient
conditions (ca. 25C) using in-tank heaters, chilled seawater,
and a real-time temperature control system (Apex Controllers,
Neptune Systems); ow was provided by a submersible
aquarium pump. Photosynthetically active radiation (PAR)
conditions were provided by light emitting diode lamps
programmed on a 12L : 12D cycle and a diel ramp. Irradiances
increased from 07:00 to 11:30 hrs to peak irradiances of
475 μmol photons m
, and decreasing from 14:30 to
19:00 hrs; daily integrated light intensities were 12 mol
photons m
. All corals were allowed to acclimate to tank
conditions for 7 d before the start of light and feeding
treatments (detailed below). During the acclimation period,
corals were not fed supplemental heterotrophic prey.
Experimental treatments
Corals were exposed to three experimental nutrition
treatments in the aquaria used in post-collection acclimation:
light and no heterotrophic feeding (Light-Not Fed [L-NF]),
light and heterotrophic feeding (Light-Fed [L-F]), and darkness
with heterotrophic feeding (Dark-Fed [D-F]). Treatments
were dened as interactions of combined light-by-feeding
conditions to produce treatments based on ratios of
autotrophy to heterotrophy: 1 : 0 (L-NF, autotrophic), 1 : 1
(L-F, mixotrophic), and 0 : 1 (D-F, heterotrophic). Although
M. capitata is not found in complete darkness in nature, it has
the capacity to survive and persist under extremely low-light
conditions. Montipora capitata is commonly found at water
depths of 40100 m in the Auau Channel separating the
Hawaiian Islands of Maui and Lanai where PAR is about 1%
of surface levels (Pyle et al. 2016). This species also survives
under extremely low-light conditions in shallow habitats of
aneohe Bay where surface light is attenuated by 86% at 8 m
depth (Wall et al. 2020). While the D-F treatment has limited
ecological relevance, this treatment was designed to bring a
cessation to Symbiodiniaceae photosynthesis during the
experimental period.
Corals in feeding treatments were provided with plankton
freshly collected using a plankton tow (63-μm mesh) off the
leeward side of HIMB and size-fractioned to remove debris
greater than 250 μm. Microscopy showed this plankton size
fraction (63250 μm) to be dominated by copepods, zoea, and
gastropod larvae. A concentrated plankton sample collected
on 19 January 2018 (63250 μm) was saved as a heterotrophic
food source isotopic end-member. Plankton was provided to
corals in feeding treatments thrice weekly for 3 h during mid-
day (13 : 0016 : 00 h). Previous work has demonstrated
plankton and POM samples from seawater adjacent to patch
reefs in K
aneohe Bay show limited spatial and seasonally vari-
ability in δ
C and δ
N values and low δ
N variability
(2) due to size fractions (Wall et al. 2020) (Supporting
Information Table S1). Therefore, it is reasonable to expect iso-
tope values in the plankton samples of a consistent size frac-
tion were relatively stable during our short experimental time
period. Concentrated tow materials were added directly into
treatment tanks while turning off the supply of seawater
entering tanks to avoid prey escape; nal plankton concentra-
tion (mean ± SE) in each tank was 1630 ± 322 plankton L
Unfed corals experienced the same environmental conditions
as fed corals during feeding (i.e., periodic cessation of ow-
through seawater). Prey ingestion rates for M. capitata were
not quantied, as the goal of the feeding treatments was to
give corals ad libitum access to plankton (Edmunds 2011) to
test the inuence of heterotrophy on tissue isotope values.
Montipora capitata feeding has been previously shown to range
from 5 to 32 plankton per gram of coral ash-free dry weight
Wall et al. Coral nutrition and amino acid isotopes
biomass per hour in bleached and pigmented colonies
(Grottoli et al. 2006). Corals were exposed to treatments for
29 d (14 December 201711 January 2018) and immediately
frozen at 80C until further processing.
Tissue stable isotope analysis
Coral tissues were removed from the skeleton using an air-
brush lled with double-distilled water (ddH
O) and attached
to a compressed air cylinder. To obtain enough tissue for AA-
CSIA, the three replicate coral ramets within each treatment
tank were pooled to produce a single coral blastate for each
genet within a treatment (n= 2 genets per treatment). The iso-
lated tissue blastate was kept on ice, briey homogenized, and
ltered through 20-μm mesh to remove skeletal debris (Wall
et al. 2020). Host and symbiont tissues were separated by
repeated centrifugation and ddH
O rinses (Muscatine
et al. 1989), and isolated tissues were lyophilized and stored at
room temperature until analyzed. Isotopic values are reported
in delta values (δ) using per mill () notation relative to stan-
dard materials (Vienna Pee-Dee Belemnite [V-PDB] and atmo-
spheric N
standards [Air] for carbon and nitrogen,
respectively). Samples for bulk tissue carbon (δ
C) and nitro-
gen (δ
N) isotope analyses and C : N ratios for coral host,
symbiont algae and plankton tissues (ca. 1 mg) were packed in
tin capsules and measured on a Costech elemental
combustion system coupled to a Thermo-Finnigan Delta Plus
XP isotope ratio mass spectrometer. Sample analytical
accuracy and precision (δ
C and δ
N) was <0.2(see
Supporting Information Appendix S1).
Individual amino acid isotope analyses
Isotopic analysis of amino acids in all samples (coral host,
symbiont algae, plankton) was performed by subjecting tissue
samples to acid hydrolysis, carboxyl terminus esterication,
and amine group triuoroacetylation (Hannides et al. 2013;
Shih et al. 2020) (see Supporting Information Appendix S1).
Acid hydrolysis was performed by heating (150C) approxi-
mately 15 mg of tissue in 6 N HCl, evaporating to dryness,
redissolving hydrolysate in 0.01 N HCl, ltering (0.2-μm poly-
ethersulfone lter), purifying by cation exchange (Dowex
50WX8-400), and amino acid elution with ammonium
hydroxide. Hydrolyzed tissues were esteried and the amine
group was triuoroacetylated. Finally, solvent extraction in P-
buffer (KH
in milli-Q water, pH 7) was used
to further purify samples (Shih et al. 2020). Chloroform was
used to partition acylated amino acids, and following solvent
evaporation sample triuoracetylation was repeated to maxi-
mize derivitization. Samples were stored frozen in 3 : 1 methy-
lene chloride : TFAA at 20C until analyzed.
At the time of analysis, samples were evaporated under N
at room temperature and redissolved in 50100 μL ethyl ace-
tate. Carbon stable isotope composition of amino acids was
determined using a Thermo Scientic MAT 253 mass spec-
trometer coupled to a Thermo Scientic Trace GC Ultra gas
chromatograph via a Thermo Scientic ConoIV(see Arthur
et al. 2014). For δ
C determination, internal reference com-
pounds of known isotopic compositionaminoadipic acid
(AAA) and norleucine (Nor) and underivitized n-C
were co-injected with samples and used to determine accuracy
and precision. Between triplicate runs of each sample a suite
of amino acids of known isotopic composition that were pre-
pared alongside the samples were analyzed; the AAA, Nor, and
were also co-injected with these amino acids. For isoto-
pic correction of unknown amino acids, an average correction
factor was derived from the amino acid suites run before and
after the triplicate sample analysis and applied to measured
isotope ratios (see details in Arthur et al. 2014). All amino acids
were analyzed in triplicate and isotopic values are reported in
δ-notation relative to V-PDB.
Nitrogen stable isotope composition of amino acids was
determined using a Delta V Plus mass spectrometer interfaced
with a Trace GC gas chromatograph through a GC-C III com-
bustion furnace (980C), reduction furnace (650C), and liq-
uid nitrogen cold trap (see details in Hannides et al. 2013). All
samples were analyzed in at least triplicate. Internal reference
compounds (AAA and Nor) were co-injected with samples and
served as a normalizer for sample amino acids δ
N values and
an internal tool for monitoring combustion reactor degrada-
tion and sample injection accuracy. Measurement accuracy
was determined using the known δ
N value of aminoadipic
acid to determine the measured δ
N value of norleucine, and
vice versa. In addition, a full amino acid reference suite (15
amino acids) of known isotopic composition was also ana-
lyzed before and after each three sample measurements. A pro-
cess blank (subject to the same hydrolysis and derivatization
steps) was analyzed in the same manner as samples and did
not contain detectable amino acids.
Amino acid classication
The classication of amino acids using carbon isotope
values is often grouped as acquired through diet (i.e., essen-
tial) and synthesized de novo from cellular carbon pools (i.e.,
nonessential). Essential amino acids include isoleucine (Ile),
leucine (Leu), lysine (Lys), methionine (Met), phenylalanine
(Phe), threonine (Thr), and valine (Val). The intact carbon
skeletons of essential amino acids are synthesized in a variety
of pathways by bacteria and primary producers. Consumers
acquire essential amino acids through dietary proteins and
incorporated them into tissues expressing little isotope frac-
tionation. Conversely, the nonessential amino acids glycine
(Gly), serine (Ser), alanine (Ala), glutamic acid (Glx), aspartate
(Asp), proline (Pro), arginine (Arg), and tyrosine (Tyr) are syn-
thesized from carbon pools or direct routing of nonessential
amino acids sourced from dietary protein (see McMahon and
Newsome 2019).
Amino acid nitrogen isotope values are controlled by the
enzymatic processes of transamination and deamination,
which lead to isotope fractionation (e.g., OConnell 2017).
Wall et al. Coral nutrition and amino acid isotopes
Protein δ
N values for amino acids are classied as source
and trophicamino acids (sensu Popp et al. 2007) and are
based on empirical studies of nitrogen incorporation into
amino acids. Source amino acids, including Phe, Met, Lys,
show minimal C N bond breakage in their metabolism and
discriminate little between diet and consumer and are thought
to represent baseline nitrogen values in the absence of trophic
discrimination. In contrast, the trophic amino acids, including
Glx, Asp, Ala, Ile, Leu, Pro, and Val, undergo extensive trans-
amination and deamination; thus, their δ
N values increase
with each trophic transfer (Ohkouchi et al. 2017).
The essential (EAA) and nonessential (NEAA) and trophic
and source amino acids have been previously applied in
birds, mammals, and invertebrates (see Whiteman
et al. 2019). However, corals can synthesize some amino
acids de novo and translocation of amino acids between sym-
biont partners has obscured origins of amino acid synthesis
(Trench 1971b; Fitzgerald and Szmant 1997; Wang and
Douglas 1999); therefore, applying these bins to corals
should be considered with caution. Due to this uncertainty,
we follow the previously published examples of essential
and nonessential amino acid grouping for carbon isotope
analysis of corals (Fox et al. 2019) and other animals
(Whiteman et al. 2019). For trophic and source amino acids,
we used amino acid designation outlined in Whiteman
et al. (2019) and include those amino acids where informa-
tion is limited, unknown, or where behavior is source-like
(i.e., glycine, serine) as source-amino acids.
Trophic position and trophic proxies
Trophic position was calculated as the difference between
N values of glutamic acid (glutamate, Glx) and phenylala-
nine, following (Chikaraishi et al. 2009):
Trophic position TPGlx Phe
ðÞ=δ15NGlx δ15 NPhe
where an assumed βvalue of 3.4for differences in δ
values of glutamic acid and phenylalanine in primary pro-
ducers and a 7.6enrichment (Δvalues) in
N between glu-
tamic acid and phenylalanine at each trophic transfer.
Fractionation factors for δ
in prokaryotic and eukaryotic
phytoplankton cultures and isolated Symbiodiniaceae are sim-
ilar and support both the grouping of amino acids in these
taxa as fractionatingand nonfractionatinggroups and the
application of TP
calculations (McCarthy et al. 2013;
Fujii et al. 2020) (see Supporting Information Appendix S1 for
details on testing canonical βand Δvalues). Uncertainty in
trophic position was determined using propagation of errors
(Jarman et al. 2017). To estimate fractional contributions of
heterotrophy to coral nutrition, we used trophic positions cal-
culated from δ
values of glutamic acid and phenylala-
nine and a simple linear two-component mixing model that
yields a propagated uncertainty (Phillips and Gregg 2001).
This approach uses a rst-order Taylor series approximation to
estimate the propagated variance in the contribution of each
An additional proxy for trophic position that does not rely
on assumed values of βor Δ, the difference in the δ
weighted mean values for trophic and source amino acids
(Bradley et al. 2015), was also calculated using trophic (ala-
nine, leucine, glutamic acid) and source amino acids (lysine,
phenylalanine), following (Hayes et al. 1990):
where δ
is the nitrogen isotopic value of a specied
amino acid within the trophic or source category, and σ
is the
standard deviation of the amino acid based on triplicate analysis
(Shih et al. 2020).
Summed variance in trophic amino acid δ
The summed variance (ΣV) index was applied to trophic-
values as a metric for heterotrophic resynthesis of
organic matter by bacteria (McCarthy et al. 2007), which is a
potential supplemental source of Symbiodiniaceae nutrition.
The sum of the δ
variance among individual trophic
amino acids (alanine, leucine, proline, aspartic acid, and glu-
tamic acid) was used to calculate ΣVfor host and symbiont tis-
sues according to experimental treatment.
Data analyses
Amino acid carbon and nitrogen isotope values in the coral
and Symbiodiniaceae were examined in a permutational mul-
tivariate ANOVA (PERMANOVA) using Euclidean distance
matrices to test for differences in tissue fraction (host or sym-
biont) and nutrition treatments in the package vegan
(Oksanen et al. 2019). In order to meet requirements for PER-
MANOVA, amino acid isotope values were analyzed as abso-
lute values (for carbon) or with a constant applied (in case of
nitrogen isotope values). Principal components analyses (PCA)
of a scaled and centered correlation matrix were then applied to
amino acid carbon and nitrogen isotope values. Correlation vec-
tors were plotted on PC-biplots for amino acids showing signi-
cant correlations with PCs. Analyses were grouped at the level of
(1) tissue fraction and (2) treatment to visualize the relationship
between coral-Symbiodiniaceae amino acid isotope values and
those observed in the pooled plankton sample (i.e., the hetero-
trophic food source). Carbon and nitrogen isotope values for
corals and Symbiodiniaceae bulk tissue, individual amino acids,
and indices (e.g., trophic position, δ
N-weighted mean, ΣV)
were analyzed using a linear model with tissue fraction (host or
symbiont) and nutrition treatments as xed effects.
To compare the multivariate essential-δ
from M. capitata from K
aneohe Bay (this study) with those of
Wall et al. Coral nutrition and amino acid isotopes
the only other example of δ
values in reef corals and
Symbiodiniaceae (Palmyra Pocillopora meandrina) in relation to
allochthonous food sources (plankton and particulate organic
carbon [POM]) (Fox et al. 2019), we used a linear discriminant
analysis (LDA). The LDA included δ
C values of six essential
amino acids (isoleucine, leucine, lysine, phenylalanine, threo-
nine, valine) in hosts, symbionts, plankton, and POM. Isotope
values were normalized to their sample mean to allow for
comparison of essential-δ
patterns among groups (Larsen
et al. 2009). Using mean-normalized essential-δ
printing has been applied in other systems to trace sources of
carbon through terrestrial and aquatic food webs (Larsen
et al. 2013) and removes baseline spatiotemporal variability in
C values (Larsen et al. 2020). Using nutrition sources (sym-
bionts, plankton-POM), we obtained predictions of group
membership from leave-one-out cross-validation using the
function lda in the package MASS (Venables and Rip-
ley 2002). Amino acid coefcients contributing most to
class selection were obtained and the predict function was
used to assign coral host membership to nutrition groups
for both M. capitata and P. meandrina. Linear discriminant
axes (LD1, LD2) were plotted with ellipses (95% standard
deviation of group mean) for nutrition sources. Raw
values in Palmyra plankton and POM are
not distinct (Fox et al. 2019), therefore ellipses were drawn
according to autotrophic (symbiont-derived) or heterotro-
phic (plankton + POM) nutrition sources.
To supplement the LDA from Hawaii and Palmyra, we sub-
sequently applied the same LDA approach above using values
of the coral holobiont (Acropora haraonis + Symbiodiniaceae)
and plankton (> 333 μm) from the Red Sea (McMahon
et al. 2015) using mean-normalized essential-δ
values for
ve essential amino acids (isoleucine, leucine, phenylalanine,
threonine, valine [lysine not available]). This analysis offers a
comparison of hosts, symbionts, holobionts, and planktonic
food sources across three locations and provides insight into
the sources of coral nutrition and comparability of planktonic
food sources among coral reefs.
Finally, we sought to test (1) whether Symbiodiniaceae
show different essential-δ
ngerprints compared to
diverse groups of cultured and free-living plankton and (2)
how similar essential-δ
ngerprints are in diverse plank-
ton assemblages. To accomplish this, we surveyed the litera-
ture and compiled values of the ve essential-δ
plankton, POM, and microalgae using examples from
aneohe Bay (Oahu, Hawaii, this study), Station Aloha
(open ocean, north of Oahu, Hannides et al. 2013), Palmyra
(Fox et al. 2019), the Red Sea (McMahon et al. 2015), and in
cultured microalgae (grouped chlorophytes, chrysophytes,
cyanobacteria, diatoms, and haptophytes; Larsen et al. 2013).
Pooling the cultured microalgae was previously justied using
mean-normalized essential-δ
values due to limited differ-
ences among groups (Larsen et al. 2013). Mean-normalized
values in these free-living/cultured plankton,
which represent a range in heterotrophic food sources, were
compared to those of Symbiodiniaceae isolated in corals from
Hawaii and Palmyra using PCA of a scaled and centered correla-
tion matrix.
All statistical analyses were performed in R version 3.6.1
(R Core Team 2019). Data and code to reproduce analyses and
gures are available at GitHub (
CSIA-corals) and archived at Zenodo (
Bulk tissue isotope analyses
Carbon isotope values did not differ between host and sym-
biont tissue fractions or in response to treatments (p0.190)
(Supporting Information Table S2, Fig. 1a). Nitrogen isotope
values were higher in the host than the symbiont (p= 0.005)
(Fig. 1b) but were not affected by treatments (p=0.831). Mean
Fig. 1. (ae) Bulk tissue δ
C and δ
N values and C : N for coral hosts and Symbiodiniaceae symbionts and their relative differences (δ
exposed to three Light-by-Feeding nutrition treatments and their relation to a pooled plankton sample. Treatments are LNF (LightNot Fed, autotro-
phic), LF (LightFed, mixotrophic), DF (DarkFed, heterotrophic). Boxplots are n=2, except for plankton (n= 1).
Wall et al. Coral nutrition and amino acid isotopes
C : N ranged from 5.3 to 7.0 and did not differ between host
and symbiont tissues (p= 0.420). Tissue C : N was lower in the
D-F treatment compared to others (p= 0.025), with symbiont
C : N declining and approaching those in plankton
(C : N = 4.4) (Fig. 1c). The difference in isotope values in the
host relative to the symbiont for carbon (p= 0.064) and nitro-
gen was not affected by treatments (p= 0.769) (Fig. 1d,e),
although δ
tended to be 0.51lower in the D-F treat-
ment relative to others. Overall, the carbon and nitrogen iso-
tope values of the pooled plankton sample were lower (δ
and higher (δ
N) than those values observed in both the coral
host and Symbiodiniaceae symbionts (Fig. 1).
Amino acid isotope analyses
Thirteen amino acids were extracted from coral tissue,
Symbiodiniaceae, and a pooled plankton end-member
(Supporting Information Table S3). Mean coral host δ
values ranged from 24.3 to 8.1and were generally higher
than those observed in the symbionts (9 of 13 amino acids)
but differences in host and symbiont δ
values ranged
from 0.8 to +3.8. Carbon isotope values in plankton
amino acids were on average 5.5 and 4.7lower than those
in coral host and Symbiodiniaceae, respectively (Supporting
Information Table S3). Coral δ
values ranged from 1.1
to +7.0but were variable in relation to δ
values in the
symbiont. Individual coral δ
values ranging from 2.3 to
+2.7relative to Symbiodiniaceae, with glycine, threonine,
and tyrosine in particular having higher δ
N values in the
symbionts. δ
values in the plankton sample were on
average 2.5 higher relative to corals and Symbiodiniaceae.
PERMANOVA results showed the multivariate carbon iso-
tope ngerprint (i.e., δ
) of 13 amino acids did not differ
between corals and their symbionts (p= 0.056), nutrition
treatments (p= 0.470), or their interaction (p= 0.771)
(Table 1). The multivariate nitrogen isotope ngerprint
(i.e., δ
), however, did differ between host and symbiont
tissue fractions (p= 0.017), but was not affected by treatments
(p= 0.911) or the fraction ×treatment interaction (p= 0.599)
(Table 1).
PCA showed two PCs explained 81% (carbon) and 74%
(nitrogen) of variance in amino acid isotope values in the
coral, Symbiodiniaceae, and pooled plankton sample (Fig. 2).
The PCA ellipses followed PERMANOVA results, with host and
symbiont amino acids overlapping for δ
values (Fig. 2a)
but showing clear separation for δ
values (Fig. 2b). Treat-
ment ellipses showed δ
values were most distinct
between the L-F and D-F treatments (Fig. 2c). δ
overlapped but showed a more constrained distribution in the
L-F treatment (Fig. 2d). In relation to the plankton sample,
vectors for individual amino acid δ
C values were negatively
correlated with PC1, which had a positive correlation with the
plankton sample. Therefore, carbon PCAs for tissue fraction
and treatment effects on δ
values showed poor conver-
gence or relation to those of the plankton (Fig. 2a,c). In
contrast, amino acid vectors for δ
values in the host,
symbionts, and the plankton were all negatively correlated
with PC1. Six amino acid vectors (glutamic acid, alanine,
aspartic acid, and valine) were particularly correlated with the
plankton sample and also aligned more with host-ellipses over
those in the symbiont. In contrast, symbiont δ
and vectors positively correlated with PC2 were phenylala-
nine, tyrosine, and threonine (Fig. 2b). Treatments showed
poor correlation with δ
values in the plankton,
where ellipses for greater heterotrophy food availability
(i.e., L-F, D-F) overlapped with those where heterotrophy was
withheld (Fig. 2d).
Carbon isotope values of individual amino acids
For the 13 individual amino acids, carbon isotope values in
host and symbiont fractions only differed for glycine
(p= 0.004) and glutamic acid (p= 0.004) (Supporting Informa-
tion Table S4), which were lower in symbiont tissues relative
to the host (Fig. 3a). However, trends in δ
values were
observed for alanine (p=0.059) and proline (p=0.055), which
tended to also have lower isotope values in symbiont tissues
(Supporting Information Fig. S1a). Nutrition treatments
(i.e., light ×feeding) showed limited effects on δ
, with
the exception of glycine (p= 0.016), which was signicantly
lower in L-F relative to D-F treatment (post hoc: p= 0.013)
and lower in both the autotrophic and mixotrophic treat-
ments (i.e., LNF, LF) (post hoc: p0.172) (Fig. 3a,
Supporting Information Fig. S1c). The remaining amino acids
were not affected by treatments (p0.076) (Supporting Infor-
mation Table S4). Carbon isotope values in amino acids of the
Table 1. Results of PERMANOVA testing effects of tissue fraction
and nutrition treatment on amino acid carbon and nitrogen iso-
tope values.*
Factor df SS R
Amino acid carbon isotope values
Fraction 1 6.931 0.153 1.724 0.056
Treatment 2 7.895 0.174 0.982 0.470
Fraction ×treatment 2 6.369 0.141 0.792 0.771
Residual 6 24.128 0.532
Total 11 45.323 1.000
Amino acid nitrogen isotope values
Fraction 1 7.065 0.182 2.042 0.017
Treatment 2 4.753 0.122 0.687 0.911
Fraction ×treatment 2 6.315 0.162 0.912 0.599
Residual 6 20.763 0.534
Total 11 38.896 1.000
df, degrees of freedom; SS, sum of squares.
Fractionis host coral or symbiont Symbiodiniaceae tissue. Treatment
represents combination of Light-by-Feeding nutrition treatments: Light
Not Fed, LightFed, DarkFed. Bolded p-value represents signicant
effects (p< 0.05).
Wall et al. Coral nutrition and amino acid isotopes
pooled plankton sample were generally lower for all amino
acids measured in the host and symbiont; however, δ
values for valine, glycine, and serine overlapped with those
observed in the plankton (Fig. 3a, Supporting
Information S1a).
Nitrogen isotope values of individual amino acids
Four amino acid nitrogen isotope values differed in host
and symbiont fractions. δ
values were lower in symbiont
fractions for leucine (p= 0.002), proline (p= 0.001), and
aspartic acid (p= 0.036), but the isotope values of tyrosine
were higher in the symbiont relative to the host (p= 0.017)
(Supporting Information Table S5; Fig. 3b, Supporting Infor-
mation Fig. S1b). δ
values did not differ between host
and symbiont fractions for the other nine amino acids
(p0.103) (Supporting Information Table S5; Fig. 3,
Supporting Information Fig. S1b). Nutrition treatments did
not inuence δ
values in any of the 13 amino acids,
although a trend for lower δ
values in heterotrophy
treatments (i.e., L-F, D-F) was observed for leucine
(p= 0.069) (Fig. 3b, Supporting Information Fig. S1d). There
was no signicant difference in source amino acid δ
values in the plankton, coral hosts, or symbionts, and tro-
phic position calculations using glutamic acid and phenylal-
anine (e.g., Eq. 1, TP
symbiont fractions (p= 0.223) or in response to treatments
(p= 0.998). Thus, source-δ
values alone were ineffec-
tive in identifying sources of nutrition in M. capitata.
Symbiont Plankton
Fig. 2. (ad) Principal component analyses of δ
C(left) and δ
N(right) values of individual amino acids in coral hosts, Symbiodiniaceae symbionts,
and a pooled plankton sample in relation to tissue fractions (a,b) and treatments (c,d). Ellipses represent 90% standard deviation with arrows for individ-
ual amino acids being signicant (p< 0.05) correlation vectors.
Wall et al. Coral nutrition and amino acid isotopes
Overall, the mean SD) host and symbiont TP
ranged from 0.90 ± 0.08 to 1.35 ± 0.24, with the highest tro-
phic positions measured in corals from treatments that were
not fed; the plankton end-member TP
was 2.00 ± 0.22
(Supporting Information Table S6; Fig. S2a).
Percent heterotrophy and δ
Differences between plankton TP
and those in coral
host and symbionts were used to estimate fractional contribu-
tions of heterotrophy using TP
in a mass balance two-
member mixing model that accounted for three end-member
trophic positions. First, the TP
of Symbiodiniaceae pri-
mary producers is 1.0 (Supporting Information Table S6; but
see also Martinez et al. 2020; Fujii et al. 2020); therefore, 100%
autotrophic nutrition in the coral host would occur if the syn-
thesis of glutamic acid by the coral host is entirely derived
from translocation of symbiont-derived nitrogen (amino or
organic acids). Second, assuming the Δ
value for hetero-
trophic feeding by coral hosts is similar to other heterotrophic
organisms and zooplankton consumers (TP
of 2.00; this
study, Fujii et al. 2020), we anticipate corals feeding on 100%
heterotrophy in the form of primary producers and algal detri-
tus would have a TP
of 2.0. These primary producers
could be free-living algae or in hospite Symbiodiniaceae, as
corals can ingest phytoplankton (Leal et al. 2014) as well as
digest symbionts (Titlyanov et al. 1996; Tanaka et al. 2018).
Lastly, as a third end-member, we consider a diet of
zooplanktivory. The pooled plankton sample was a mix of
microzooplankton and macrozooplankton at TP
2.00 ± 0.20. Therefore, we consider the host coral feeding het-
erotrophically on zooplankton to have a TP
of 3.0,
which is similar to a mean TP
of 2.86 for a temperate
heterotrophic octocoral (Grossowicz et al. 2020). Using the
difference in TP
between plankton, symbiont and host,
the fractional contribution of heterotrophy to host nutrition
ranged from (mean ± SD) 1% ± 9% to 41% ± 15%, being
higher in mixing models assuming heterotrophy of TP
2.0 relative to 3.0 (Supporting Information Table S6). Surpris-
ingly, these calculations resulted in higher % heterotrophy in
the Light-Not Fed treatment (2141% heterotrophy) compared
to treatments where corals were fed (16% heterotrophy)
(Supporting Information Table S6).
Calculations of δ
ΣV(i.e., Sum-V) did not differ
among host or symbiont fractions (p= 0.088) or in response
to treatments (p= 0.656) (Supporting Information Fig. S2b),
and δ
-weighted means did not differ between the host
and symbionts or between source and trophic amino acid
(p0.626) (Supporting Information Fig. S3).
C essential amino acids patterns among coral studies
The δ
C values for amino acids from the pooled K
Bay, plankton sample (63250 μm) displayed similar patterns
to those reported for plankton (> 163 μm) and seawater POM
(> 0.7 μm) in Palmyra (Fox et al. 2019), although an offset was
observed, with lower δ
C values for the Hawaii plankton
sample, particularly for nonessential amino acids (Supporting
Information Fig. S4). Using mean-normalized δ
C values for
the six essential amino acids (i.e., δ
) of autotrophic and
heterotrophic food sources from Hawaii and Palmyra, linear
discriminant analysis cross-validation correctly identied class
membership (71% overall), with 100% success rates for symbi-
onts, but less so for plankton and POM (60% and 25%,
Plankton +
Fig. 3. (a)δ
C and (b)δ
N values of individual amino acids in coral host, Symbiodiniaceae symbionts, in response to nutrition treatments relative to a
pooled plankton sample. Treatments are LightNot Fed (LNF), LightFed (LF), DarkFed (DF); values are mean ± SD (n= 2), except for the plankton
sample (n= 1). X-axis symbols indicate signicant differences (p< 0.05) between fractions (host and symbiont, *) and treatments ().
Wall et al. Coral nutrition and amino acid isotopes
respectively), which largely overlapped in their essential-
values and are poorly differentiated nutritional sources
(Fig. 4a). The rst linear discriminant (LD1) explained 98% of
group variation, and the essential amino acids contributing
most to group separation along LD1 were valine (LD1 scaling:
0.64), leucine (0.55), isoleucine (0.50), and phenylalanine
(0.34). The LDA predictions of coral host group membership
showed the essential-δ
ngerprint of all Hawaii
M. capitata corals to be autotrophic (i.e., grouped with symbi-
onts), and M. capitata corals showed a high degree of overlap
with symbionts in LD plots (Fig. 4). In contrast to M. capitata,
Palmyra P. meandrina were more variable and plotted across
autotrophic and heterotrophic food source groups (Fig. 4). The
larger training data set of essential-δ
values from Hawaii
and Palmyra assigned six of 19 P. meandrina corals to hetero-
trophic food sources, in agreement with group membership
reported by Fox et al. (2019). Adding Red Sea coral and plank-
ton samples to the mean-normalized essential-δ
resulted in 83% effective group membership among food
sources (symbionts, plankton, POM), with 94% and 100% of
plankton and symbionts group membership being correctly
assigned. Here, LD1 accounted for 99% of group variance. Six
corals (all from Palmyra) were again assigned to heterotrophic
food sources, whereas Hawaii and Red Sea corals were all
assigned to symbiont food sources (Fig. 4b).
PCA of mean-normalized isotope values for ve essential-
of cultured microalgae and plankton/POM samples
across ve studies showed 62% of variance was explained by
two PCs. Greater data dispersion were observed for microalgae
relative to the plankton and POM; however, both groups over-
lapped (Supporting Information Fig. S5a). Adding
Symbiodiniaceae symbionts into the data matrix increased
variance explained (71% at two PCs). Symbionts were sepa-
rated based on their location (Hawaii, Palmyra) and/or host
coral (M. capitata, P. meadrina), yet symbionts were distinct
and did not overlap with microalgae, plankton, or POM
sources (Supporting Information Fig. S5b).
Climate change and environmental stress threaten reef-
building corals; however, the inherent or exible exploitation
of heterotrophic nutrition can benet corals experiencing abi-
otic stress and dysbiosis. Previous work using feeding assays
Fig. 4. Linear discriminant analysis of mean-normalized amino acid carbon isotope ngerprints using (a) six essential amino acids (isoleucine, leucine,
lysine, phenylalanine, threonine, tyrosine, valine) and (b)ve essential amino acids (without lysine). Ellipses represent 95% condence ellipses for each
nutrition group (autotrophy [symbiont] or heterotrophy [plankton-POM]). In (a) and (b), Hawaii data are host and symbiont (n= 6 [fragments from two
reef genets in three experimental treatments]) and plankton (n= 1 [63250 μm]); Palmyra data are host (n= 19), symbionts (n= 11) (from 19 colonies
[10 m]) and plankton (n= 9 [>163 μm]) and POM (n= 8 [>0.7 μm]) at four sites (Fox et al. 2019). Red Sea data in (b) are host+symbiont (holobionts,
n= 23 colonies) and plankton (n= 23 [>333 μm]) from eight sites (McMahon et al. 2015).
Wall et al. Coral nutrition and amino acid isotopes
(Grottoli et al. 2006) and isotope analyses of bulk tissues
(Rodrigues and Grottoli 2006) and lipids (Baumann
et al. 2014) has shown M. capitata grown in the laboratory uti-
lizes heterotrophic nutrition to survive stressful events includ-
ing bleaching. However, eld data collected across
environmental gradients (Wall et al. 2020) and during and
after in situ bleaching events (Wall et al. 2019, but see ex situ
bleaching Rodrigues and Grottoli 2006), do not support a ten-
dency towards trophic plasticity in this species. Using both
bulk and AA-CSIA, we show minimal responses of M. capitata
to nutrition treatments and do not observe this species to
exploit opportunities for greater prey capture when resources
are available, despite changes in autotrophic inputs. Results of
a linear discriminant analysis of mean-normalized essential-
values showed autotrophic (symbiont) nutrition
supported all M. capitata corals regardless of treatment. Treat-
ments did drive greater variability in δ
and declining
for D-F corals in particular, while also reducing C : N
and glycine-δ
N values. Across all treatments, host and sym-
biont trophic positions (i.e., TP
, difference in the δ
weighted mean values for trophic and source amino acids)
were lower than the plankton TP (1 vs. plankton TP
of 2), suggesting limited trophic connections to plankton-
derived nutrition compared to other cnidarian predators
(Grossowicz et al. 2020). While treatments were not meant to
produce changes in coral physiology at the scale of cellular
bleaching, and results are not meant to be exhaustive in test-
ing for heterotrophic plasticity at ecological scales, we never-
theless conclude healthy M. capitata exhibit low nutritional
plasticity based on food availability or environmental
was lower in the D-F treatment; however, it appears
incorrect to interpret this trend as a sign of greater heterotro-
phy due to the fact that host and symbionts δ
C values were
highest in this treatment (16.6 to 15.1) and did not
approach δ
C values of the plankton (21.8). Therefore,
higher δ
C values in the D-F treatment corresponded with
lower C : N, which together suggest corals (and
Symbiodiniaceae in particular) may have experienced a
decline in lipid biomass. Perturbed light : dark cycles can
affect coral circadian clock genes (Hoadley et al. 2011) and
prolonged darkness can cause behavior changes or energetic
decits that inuence polyp extension, feeding rates, and tis-
sue biomass (Clayton and Lasker 1982). In the temperate
coral, Astrangia poculata polyp extensions were higher in fed
vs. unfed corals in both symbiotic and aposymbiotc corals
(Burmester et al. 2018), and in darkness Cladocora caespitosa
corals had lower feeding rates and reduced lipid biomass com-
pared to illuminated treatments (Hoogenboom et al. 2010). In
tropical corals, darkness can also cause lower prey ingestion
rates and reduce biomass (Clayton and Lasker 1982), which
may be due to the hosts requirement for autotrophic energy
to support the capture and ingestion of prey (Leal et al. 2015).
In the context of the present study, changes in bulk δ
values likely relate to changes in tissue composition and the
substrates used in metabolism, rather than changes in feeding
modes (Swart et al. 2005; Wall et al. 2019). The absence of
detectable changes in heterotrophy in treatments where
M. capitata were fed (L-F, D-F) emphasizes that laboratory
M. capitata exhibits a delity to autotrophic nutrition despite
changes in food availability, a potential for prey capture to be
dependent on symbiont photosynthesis, and the importance
of autotrophy and photoacclimation as factors shaping the
depth distribution (Wall et al. 2020) and physiological
responses of M. capitata to environmental stress.
The lack of treatment effects across analyses of bulk tissue
and amino acids provide consistent evidence for poor capacity
for trophic plasticity in M. capitata; yet, heterotrophy is impor-
tant to coral energy budgets (Houlbrèque and Ferrier-
Pagès 2009) and M. capitata does consume plankton (Grottoli
et al. 2006). Indeed, M. capitata percent heterotrophy calcula-
tions based on TP
indicate 141% of nitrogen from
sources supplemental to symbiont translocation (Supporting
Information Table S6); however, higher percent heterotrophy
did not correlate with prey availability or the absence of light.
This apparent paradox may be explained by a combination of
limited contributions of heterotrophy (relative to autotrophy)
to holobiont amino acid biosynthesis, the translocation (and
recycling) of amino acid building blocks between symbiont
partners, and the combined capacity of the coral holobiont to
obtain amino acids through diet/assimilation and de novo syn-
thesis (see below). Bulk tissue δ
N and δ
C values can vary
substantially among coral colonies of the same species in
response to symbiont communities (Wall et al. 2020) and spa-
tial scales (meters to kilometers) due to environmental hetero-
geneity (i.e., water ow, light, upwelling) (Teece et al. 2011;
Fox et al. 2019; Radice et al. 2019; Wall et al. 2019), which
may affect feeding efforts (Palardy et al. 2005) and the ability
to detect trophic changes (Fox et al. 2019). In the present
study, colonies differed in their distance to the patch reef
slope (and deeper lagoon water). We expect the close proxim-
ity (ca. 5 m) and similar depth (< 1 m) of corals collected in
the present study likely minimized effects of environmental
inuences on bulk isotope values prior to coral collections. To
what degree coral amino acid isotope values differ in response
to host and/or symbiont biological traits vs. environmental
conditions is uncertain. Therefore, testing of colony-level vari-
ability (linked to host genotypes, symbiont communities) and
environmental inuences on amino acid stable isotope values
should be further examined under eld and laboratory
Essential amino acid isotope ngerprinting
Our analysis of carbon and nitrogen isotope values in
amino acids show signicant overlap in M. capitata host and
symbiont amino acid isotope ngerprints in both PC and lin-
ear discriminant plots (Figs. 2, 4). In particular, δ
are similar for both essential and nonessential amino acids,
Wall et al. Coral nutrition and amino acid isotopes
indicating a shared source of carbon for host and symbiont
amino acids biosynthesis. Surprisingly, the PCA showed it was
the presence or absence of light (and not feeding) that pro-
duced the largest differences in δ
Cngerprints (Fig. 2c),
indicating a signicant inuence of symbiont photosynthesis
on coral essential amino acids. However, when examining
C values and differences in tissue fractions and treatment
effects (Supporting Information Table S4), we show only glu-
tamic acid and glycine δ
C (both nonessential-AA) differed
between host and symbiont tissue fractions, with only glycine
being inuenced by nutrition treatments (Supporting Infor-
mation Fig. S1c). The most reasonable explanation of these
trends appears to be that M. capitata uses the same proportion
of dietary carbon sources (i.e., autotrophy : heterotrophy) and
most, if not all, amino acids may be synthesized and shared
within the coral holobiont with minimal direct contributions
from allochthonous nutrition. Tissue isotope values integrate
nutritional signals over long periods; therefore, the absence of
nutrition treatment effects (δ
C and δ
N) in the present
study (1 month) may also be a result of tissue turnover rates
in the host and symbiont (Tanaka et al. 2018; Rangel
et al. 2019) and different assimilation (and metabolism) of car-
bon and nitrogen derived from autotrophic vs. heterotrophic
nutrition (Krueger et al. 2018). Nitrogen turnover time in par-
ticular may be prolonged (212 months) and differ between
host and symbiont tissues and the inputs from dissolved or
heterotrophic nitrogen sources (Tanaka et al. 2018; Rangel
et al. 2019). Therefore, more and potentially longer manipula-
tive feeding studies are required to determine how AA-CSIA
differ among colonies with different environmental histories
(i.e., depth and light environment) and to test for the inu-
ence of tissue turnover times and colony-level traits (physiol-
ogy, genetic) on AA-CSIA data.
While there are limited δ
data for essential and non-
essential amino acids in corals and heterotrophic sources (Fox
et al. 2019), our results align well with earlier ndings.
Notably, essential and nonessential-δ
values show signif-
icant overlap between hosts and symbionts, with most amino
acids in hosts and symbionts being enriched in
C relative to
those amino acids in the plankton-POM. In addition, corals
and Symbiodiniaceae showed greater separation from plank-
ton for essential-δ
values, in particular for isoleucine,
lycine, and threonine (this study, Fox et al. 2019). Capturing
the variability in amino acid isotope values of heterotrophic
sources (i.e., POM, range of plankton size classes) was beyond
the scope of the present study. However, bulk δ
C and δ
values across a range of plankton size classes (including those
used here, 63250 μm) in K
aneohe Bay show little spatiotem-
poral variance (Supporting Information Table S1, see also Wall
et al. 2020). In Palmyra, where δ
C and δ
N values in zoo-
plankton communities also show little difference among
pelagic and lagoonal habitats, δ
C values of essential and
nonessential amino acids did not differ between plankton and
POM (Fox et al. 2019). Therefore, in the absence of signicant
spatiotemporal variation in plankton community composition
and constitutive biochemical pathways, amino acid isotope
values may be stable at the base of the food web. Therefore,
appear to be particularly useful in testing for
trophic interactions linked to colony-level changes in feeding
behavior and/or environmental factors (Teece et al. 2011; Fox
et al. 2018; Radice et al. 2019), but more tests are needed to
evaluate sources of variability in amino acid isotope values
(and food quality) in coastal marine food webs and coral reef
ecosystems (McMahon et al. 2015).
Identifying the sources of essential amino acids in M. cap-
itata was enabled by (i) using linear discriminant analysis of
mean-normalized essential-δ
values (isoleucine, leucine,
lysine, phenylalanine, threonine, tyrosine, valine); (ii) leverag-
ing previously published δ
values of coral-
Symbiodiniaceae and heterotrophic food sources (McMahon
et al. 2015; Fox et al. 2019); (iii) and comparing
Symbiodiniaceae isolated from corals in Hawaii and Palmyra
to a larger collection of natural zooplankton assemblages and
cultured microalgae (Hannides et al. 2013; Larsen et al. 2013).
Evaluating these data sets revealed several compelling nd-
ings. First, PCA revealed considerable overlap between micro-
algae and zooplankton mean-normalized essential-δ
ngerprints (Supporting Information Fig. S5a), and when
Symbiodiniaceae were added to the data matrix their δ
ngerprint was unique compared to microalgae and zooplank-
ton (Supporting Information Fig. S5b). While there is variabil-
ity in plankton end-members that requires further analysis,
nevertheless, the consistent difference in Symbiodiniaceae
related to plankton suggests such essential-δ
are promising for detecting differences between autotrophic
and heterotrophic food sources (i.e., Symbiodiniaceae-derived
vs. prey capture) in corals across space and time. Using plank-
ton, coral hosts, Symbiodiniaceae, and a coral holobiont (host
+symbiont), we nd similar overlap in plankton samples
(Hawaii, Palmyra, Red Sea) (Fig. 4b), which were distinct from
those of Symbiodiniaceae (Hawaii, Palmyra). Coral holobionts
from the Red Sea, like those from coral hosts in Hawaii, over-
lapped with essential-δ
ngerprints of the symbionts
and showed limited overlap with plankton, whereas Palmyra
corals were equally spread between symbiont and plankton
groups (Supporting Information Fig. S4a,b). Overall, these data
show that plankton samples have conserved essential-δ
ngerprints that are different from those in the
Symbiodiniaceae symbionts and the essential-δ
prints of corals reect the proportions of these food sources
coral hosts consume, which varies according to coral species,
locations, or both.
Using the analytical approach employed by Fox
et al. (2019) of essential-δ
ngerprints of six essential
amino acids in isolated coral hosts and Symbiodiniaceae sym-
bionts, we found LDA to be effective in assigning corals to
food sources based on the proportions of these sources to ani-
mal diets. The Hawaii plankton sample was the approximate
Wall et al. Coral nutrition and amino acid isotopes
centroid for the probability ellipse of the larger group of het-
erotrophic food sources from Palmyra (Fox et al. 2019) and
again provides support for using mean-normalized essential-
to effectively compare food sources across spatial scales
(Fig. 4) (Larsen et al. 2020). Coral hosts generally grouped with
their respective Symbiodiniaceae symbionts, but holobiont-
specic patterns in essential-δ
values were apparent,
suggesting differences in biosynthetic capacities and/or feed-
ing modes. We found all M. capitata corals grouped with auto-
trophic (symbiont) nutrition, whereas P. meandrina showed
trophic plasticity (41% of essential amino acid carbon from
heterotrophy; Fox et al. 2019) and grouped with both symbi-
ont and plankton nutritional sources. Therefore, our results
using individual amino acids, essential-δ
and feeding manipulation consistently show M. capitata to be
largely reliant on autotrophic nutrition for essential amino
acid biosynthesis. Considering the limitations in available
data for AA-CSIA, it is important to continue verifying isotope
ngerprint approaches among coral species and reef locations.
In this task, linear discrimant analyses of mean-normalization
of essential-δ
ngerprints appear promising for under-
standing coral trophic ecology and testing nutritional
Amino acid nitrogen isotope values and trophic position
Trophic amino acids undergo transamination and reect
trophic connections between dietary sources and consumers,
whereas source amino acids show minimal isotopic discrimi-
nation between consumers and their diet and propagate
through food webs with minimal bond-breakage. Measured
differences between phenylalanine in both M. capitata host
and Symbiodiniaceae fractions relative to plankton were small
and glutamic acid values overlapped for the host and symbi-
ont, indicating a shared position as the base of the food web
(Supporting Information Fig. S1). Our values are similar to
those reported for the coral Acropora digitifera from Japan and
identify corals and their symbionts as having a trophic posi-
tion of 1a value indicative of primary producers. In a
larger sampling of corals from Japan, TP
in host and
symbionts ranged from 0.8 to 1.4 and was indicative of a
range of heterotrophic dependence for nitrogen (<27% hetero-
trophy). Host TP
was also positively correlated with
both bulk symbiont δ
N values and TP
, which was
interpreted as a sign of enhanced feeding in polluted near-
shore habitats (Fujii et al. 2020). However, corals may not shift
trophic positions even across extreme environmental change.
For instance, Stylophora pistillata mean TP
was similar in
mesophotic and shallow hosts and symbionts (1.31.7); thus,
heterotrophy represented the same proportion of energy to S.
pistillata metabolism (35%) across a range of habitats and
environmental conditions (Martinez et al. 2020).
We show M. capitata are autotrophic (TP
= 1) and the
most abundant heterotrophic food source was likely
microalgae-derived detritus and/or digested symbiont cells
= 1), with contributions from zooplanktivory being
small (TP
= 2). Interestingly, M. capitata mean percent
heterotrophy was lower in fed-treatments (16%) compared to
the Light-Not Fed treatment (heterotrophy 2141%). Consid-
ering this effect was observed in both genotypes used in the
study, treatment effects on TP
and percent heterotrophy
may be greater than colony-specic attributes related to feed-
ing effort. While intriguing, due to the low statistical power
(small effect size and low replication) and no difference in
either TP
or the difference in the δ
N-weighted mean
values for trophic and source amino acids of hosts and symbi-
onts, caution is required in interpreting these percent hetero-
trophy calculations from TP
. However, these results
suggest corals feed less often on prey of TP
of 2.0 and
microalgae-derived detritus, and not zooplanktivory, is the
dominant source of heterotrophy identied in previous stud-
ies using these calculations (Fujii et al. 2020; Martinez
et al. 2020). As more AA-CSIA data sets become available for
coral and invertebrates (Shih et al. 2020), it will be important
for fundamental assumptions in these methods to be evalu-
ated (i.e., canonical constants, enrichment factors, prey
sources) as these may vary in response to diet quality or due
to underlying differences in the potential for corals and their
Symbiodiniaceae or microbiomes to synthesize amino acids
and demonstrate nutritional exibility or delity.
Using the trophic-δ
ngerprint in PC plots, we
observed trophic-δ
values in the host tended to be
higher than those in the symbiont, and source-δ
in the plankton were substantially higher (up to 7) than
both host and symbiont. However, the single sample of the
plankton end-member in the present study limits the infer-
ences of patterns in corals relative to the plankton. Neverthe-
less, similarities in trophic- and source-δ
values for both
host and symbiont tissue fractions indicate a shared source of
amino acid nitrogen and the inuence of holobiont nitrogen
conservation and/or recycling (Falkowski et al. 1984). Dis-
solved nitrogen is rapidly assimilated by Symbiodiniaceae
(< 1 h) and transferred to the host (6 h), presumably in the
form of amino acids and glycoconjugates (Kopp et al. 2013).
Similarly, dietary end products (i.e., NH +
4and CO
) of host
planktivory are assimilated and recycled by Symbiodiniaceae,
with <23% of heterotrophic carbon and nitrogen pools being
assimilated by the symbionts in <6 h (Krueger et al. 2018).
Over longer periods, host metabolism can supply up to 80%
of Symbiodiniaceae nitrogen (Tanaka et al. 2015). In the con-
text of our study, the large difference in δ
between M. capitata (host and symbionts) and the plankton,
particularly for those amino acids predicted to show connec-
tions between diet and consumers (i.e, trophic-δ
), does
not support a signicant role of zooplanktivory in the diet of
M. capitata or as a signicant source of nitrogen in amino acid
biosynthesis. Instead, we suggest a dominant inuence of de
novo biosynthesis of trophic and source amino acids by the
host and/or Symbiodiniaceae in the M. capitata holobiont,
Wall et al. Coral nutrition and amino acid isotopes
with the possibility for a signicant yet variable inuence of
low trophic position food sources (e.g., digested symbiont
cells, microalgae) on trophic amino acids.
The role of microbial symbionts in coral amino acids is
uncertain. To test for microbial inuences on δ
we applied an isotope proxy indicative of reworking of organic
matter by heterotrophic bacteria and amino acid resynthesis
based on trophic-δ
values (i.e., ΣV) (McCarthy
et al. 2007). ΣVvalues for M. capitata host and symbiont tis-
sues were below those reported for the sponge Mycale grandis
near southern K
aneohe Bay (isolated sponge-cells and associ-
ated microbes ΣVrange 2.33.0). In this case, higher ΣVin
M. grandis, indicated microbial symbionts as a source for trans-
located amino acids (Shih et al. 2020). We found ΣVvalues
were low (1) and resembled those reported for oceanic zoo-
plankton (McCarthy et al. 2007) and did not differ signi-
cantly in host and symbiont tissues or among treatments
(Supporting Information Fig. S2b). ΣVcalculations may differ
depending on the number of trophic amino acids included in
calculations; nevertheless, we show microbial resynthesis of
amino acids does not appear to drive patterns in coral trophic-
Sources of amino acids: A combination of host and
symbiont biosynthesis
CSIA may provide new insights into the trophic ecology of reef
corals and clarify trends observed in bulk tissue isotope analyses;
however, questions regarding the source of amino acids in reef
corals remain unanswered. Symbiotic anthozoans are able to
acquire many amino acids from their endosymbiont
Symbiodiniaceae (Markell and Trench 1993). For instance, symbi-
ont xed
C was tracked into anemone and zoanthid host pro-
tein and constituent amino acids (alanine, glutamic acid)
(Trench 1971a,b) and was also observed in seven amino acids in
the proteins of Aiptasia pulchella (Wang and Douglas 1999). The
release of glycoconjugates rich in essential amino acids by freshly
isolated Symbiodiniaceae also suggests Symbiodiniaceae as a
source for cnidarian host amino acids (Markell and Trench 1993).
ThegenomeofSymbiodinium kawagutti (formerly of Clade F)
(LaJeunesse et al. 2018) further reveals that this endosymbiont
contains complete biosynthesis pathways for nine amino acids
(except lysine and histidine) and may supply these to its coral
host (Acropora digitifera), which lacks these pathways (Lin
et al. 2015).
The role of the host is not insignicant, however, and cni-
darians are capable of de novo synthesis of essential amino
acids. Incubating ve symbiotic and nonsymbiotic
scleractinian corals with
C-radiolabeled amino acid precur-
sors (glucose, glutamic acid) and tricarboxylic acid cycle inter-
mediates (lysine, valine) showed scleractinians can synthesize
16 of 20 protein amino acids, eight of which are considered
essential (Fitzgerald and Szmant 1997). In a separate study,
Aiptasia pulchella not only acquired the majority of amino
acids from its endosymbionts, but also synthesized
methionine and threonine (Wang and Douglas 1999). Recent
genomic evidence, however, is revealing diversity in the
capacity of cnidarians to synthesize amino acids. For instance,
reef corals in the Robusta clade have a complete histidine bio-
synthesis pathway, but this pathway may be absent within
the Complexa clade and sea anemones (Ying et al. 2018).
Therefore, genomic studies appear to be a promising avenue
to distinguish origins of amino acids and the diverse capacities
for amino acid biosynthesis within the Cnidaria and
Our ndings using AA-CSIA cannot distinguish between
the de novo synthesis of amino acids by the host or the trans-
location of amino acids from the symbiont to the host. How-
ever, our data provide new insights into the amino acid
biosynthesis pathways and their relationship to coral trophic
ecology. First, coral and Symbiodiniaceae δ
C values in essen-
tial and nonessential amino acids were highly similar to each
other, yet distinct from those in the plankton. These ndings
indicate heterotrophy offers a limited contribution to
M. capitata holobiont amino acids and modifying this contri-
bution (i.e., through experimental nutrition treatments) may
be difcult to inexorableanding further supported by lack
of clear nutrition treatment effects on TP
The apparent trophic delity in M. capitata under experimen-
tal nutrition regimes contrasts with the exibility observed in
P. meandrina colonies in situ (Fox et al. 2019). Therefore, tro-
phic plasticity may be determined by species-specic traits
(behavior, morphologies) or genomic capabilities for amino
acid biosynthesis. Second, in a similar pattern as observed
with carbon, δ
values rarely differed between host and
symbiont fractions, but δ
ngerprints suggest an inu-
ence of planktivory on trophic-δ
values to a greater
extent than source-δ
values, although this effect is small.
N values of phenylalanine in all members (host, symbiont,
plankton) are quite similar. Thus, TP
calculations may
be problematic in the absence of large shifts in heterotrophi-
cally derived glutamic acid, which we did not observe despite
considerable differences in treatment conditions. Finally, low
and ΣVvalues suggest corals exist at a lower trophic
position than zooplankton (indicating greater feeding on
microalgae detritus compared to zooplanktivory), and there is
no clear inuence of microbial processing of dissolved organic
matter and microbial amino acid resynthesis in corals or
Symbiodiniaceae δ
values. Corals feed on a variety of
plankton size classes, and heterotrophic nutrition can vary
due to biological and behavioral attributes (i.e., polyp size,
physiological state, feeding effort) and across environments
(Palardy et al. 2005). Therefore, as the application of AA-CSIA
expands in coming years, it will be important for ecological
studies to be matched with manipulative feeding studies of
Wall et al. Coral nutrition and amino acid isotopes
varying diet type and quality using corals with different emer-
gent properties.
Data Availability Statement
All data and code to generate gures and perform analyses
are archived and openly available at GitHub (https://github.
com/cbwall/CSIA-corals) and archived at Zenodo (http://doi.
Alamaru, A., Y. Loya, E. Brokovich, R. Yam, and A. Shemesh.
2009. Carbon and nitrogen utilization in two species of
Red Sea corals along a depth gradient: Insights from stable
isotope analysis of total organic material and lipids.
Geochim. Cosmochim. Acta 73: 53335342. doi:10.1016/j.
Anthony, K. R., and K. E. Fabricius. 2000. Shifting roles of het-
erotrophy and autotrophy in coral energetics under varying
turbidity. J. Exp. Mar. Bio. Ecol. 252: 221253. doi:10.
Arthur, K. E., S. Kelez, T. Larsen, C. A. Choy, and B. N. Popp.
2014. Tracing the biosynthetic source of essential amino
acids in marine turtles using δ
Cngerprints. Ecology 95:
12851293. doi:10.1890/13-0263.1
Knowlton. 2018. Climate change promotes parasitism in a coral
symbiosis. ISME J. 12:921930. doi:10.1038/s41396-018-0046-8
Baumann, J., A. G. Grottoli, A. D. Hughes, and Y. Matsui.
2014. Photoautotrophic and heterotrophic carbon in
bleached and non-bleached coral lipid acquisition and stor-
age. J. Exp. Mar. Bio. Ecol. 461: 469478. doi:10.1016/j.
Bradley, C. J., N. J. Wallsgrove, C. A. Choy, J. C. Drazen, E. D.
Hetherington, D. K. Hoen, and B. N. Popp. 2015. Trophic
position estimates of marine teleosts using amino acid com-
pound specic isotopic analysis. Limnol. Oceanogr.
Methods 13: 476493. doi:10.1002/lom3.10041
Burmester, E. M., A. Breef-Pilz, N. F. Lawrence, L. Kaufman,
J. R. Finnerty, and R. D. Rotjan. 2018. The impact of auto-
trophic versus heterotrophic nutritional pathways on col-
ony health and wound recovery in corals. Ecol. Evol. 47:
14171412. doi:10.1002/ece3.4531
Chikaraishi, Y., and others. 2009. Determination of aquatic
food-web structure based on compound-specic nitrogen
isotopic composition of amino acids: Trophic level estima-
tion by amino acid δ
N. Limnol. Oceanogr. Methods 7:
740750. doi:10.4319/lom.2009.7.740
Clayton, W. S., and H. R. Lasker. 1982. Effects of light and
dark treatments on feeding by the reef coral Pocillopora
damicornis (Linnaeus). J. Exp. Mar. Bio. Ecol. 63: 269279.
Cooper, T. F., M. Lai, K. E. Ulstrup, S. M. Saunders, G. R.
Flematti, B. Radford, and M. J. H. van Oppen. 2011.
Symbiodinium genotypic and environmental controls on
lipids in reef building corals. PLoS One 6: e20434. doi:10.
Edmunds, P. J. 2011. Zooplanktivory ameliorates the effects of
ocean acidication on the reef coral Porites spp. Limnol.
Oceanogr. 56: 24022410. doi:10.4319/lo.2011.56.6.2402
Falkowski, P. G., Z. Dubinsky, L. Muscatine, and J. W. Porter.
1984. Light and the bioenergetics of a symbiotic coral. Bio-
science 34: 705709. doi:10.2307/1309663
Ferrier-Pagès, C., and M. C. Leal. 2019. Stable isotopes as
tracers of trophic interactions in marine mutualistic symbi-
oses. Ecol. Evol. 9: 723740. doi:10.1002/ece3.4712
Fitzgerald, L. M., and A. M. Szmant. 1997. Biosynthesis of
essentialamino acids by scleractinian corals. Biochem. J.
322: 213221. doi:10.1042/bj3220213
Fox, M. D., and others. 2018. Gradients in primary production
predict trophic strategies of mixotrophic corals across spa-
tial scales. Curr. Biol. 28: 33553363.e4. doi:10.1016/j.cub.
Fox, M. D., E. A. Elliott Smith, J. E. Smith, and S. D. Newsome.
2019. Trophic plasticity in a common reef-building coral:
Insights from δ
C analysis of essential amino acids. Funct.
Ecol. 33: 22032214. doi:10.1111/1365-2435.13441
Fujii, T., Y. Tanaka, K. Maki, N. Saotome, N. Morimoto, A.
Watanabe, and T. Miyajima. 2020. Organic carbon and nitro-
gen isoscapes of reef corals and algal symbionts: Relative inu-
ences of environmental gradients and heterotrophy.
Microorganisms 8: 1221. doi:10.3390/microorganisms8081221
Grossowicz, M., E. Shemesh, S. Martinez, Y. Benayahu, and D.
Tchernov. 2020. New evidence of Melithaea erythraea colo-
nization in the Mediterranean. Estuar. Coast. Shelf Sci.
236: 106652. doi:10.1016/j.ecss.2020.106652
Grottoli, A. G., L. J. Rodrigues, and J. E. Palardy. 2006. Hetero-
trophic plasticity and resilience in bleached corals. Nature
440: 11861189. doi:10.1038/nature04565
Grover, R., J.-F. Maguer, D. Allemand, and C. Ferrier-Pagès.
2008. Uptake of dissolved free amino acids by the
scleractinian coral Stylophora pistillata. J. Exp. Biol. 211:
860865. doi:10.1242/jeb.012807
Hannides, C. C. S., B. N. Popp, C. A. Choy, and J. C. Drazen.
2013. Midwater zooplankton and suspended particle
dynamics in the North Pacic Subtropical Gyre: A stable
isotope perspective. Limnol. Oceanogr. 58: 19311946. doi:
1990. Compound-specicisotopicanalyses:Anoveltoolfor
reconstruction of ancient biogeochemical processes. Org.
Geochem. 16: 11151128. doi:10.1016/0146-6380(90)
Heikoop, J. M., J. J. Dunn, M. J. Risk, T. Tomascik, H. P.
Schwarcz, I. M. Sandeman, and P. W. Sammarco. 2000.
N and δ
C of coral tissue show signicant inter-reef var-
iation. Coral Reefs 19: 189193. doi:10.1007/
Wall et al. Coral nutrition and amino acid isotopes
Hoadley, K. D., A. M. Szmant, and S. J. Pyott. 2011. Circadian
clock gene expression in the coral Favia fragum over diel
and lunar reproductive cycles. PLoS One 6: e19755. doi:10.
Hoogenboom, M., R. Rodolfo-Metalpa, and C. Ferrier-Pages.
2010. Co-variation between autotrophy and heterotrophy
in the Mediterranean coral Cladocora caespitosa. J. Exp. Biol.
213: 23992409. doi:10.1242/jeb.040147
Houlbrèque, F., and C. Ferrier-Pagès. 2009. Heterotrophy in
tropical scleractinian corals. Biol. Rev. Camb. Philos. Soc.
84:117. doi:10.1111/j.1469-185X.2008.00058.x
Jarman, C. L., and others. 2017. Diet of the prehistoric popula-
tion of Rapa Nui (Easter Island, Chile) shows environmen-
tal adaptation and resilience. Am. J. Phys. Anthropol. 164:
343361. doi:10.1002/ajpa.23273
Kopp, C., and others. 2013. Highly dynamic cellular-level
response of symbiotic coral to a sudden increase in envi-
ronmental nitrogen. MBio 4: e00052e00013. doi:10.1128/
Krueger, T., J. Bodin, N. Horwitz, C. Loussert-Fonta, A. Sakr, S.
Escrig, M. Fine, and A. Meibom. 2018. Temperature and
feeding induce tissue level changes in autotrophic and het-
erotrophic nutrient allocation in the coral symbi osisA
NanoSIMS study. Sci. Rep. 8: 12710. doi:10.1038/s41598-
LaJeunesse, T. C., J. E. Parkinson, P. W. Gabrielson, H. J.
Jeong, J. D. Reimer, C. R. Voolstra, and S. R. Santos. 2018.
Systematic revision of Symbiodiniaceae highlights the
antiquity and diversity of coral endosymbionts. Curr. Biol.
28: 25702580.e6. doi:10.1016/j.cub.2018.07.008
Larsen, T., D. L. Taylor, M. B. Leigh, and D. M. OBrien. 2009.
Stable isotope ngerprinting: A novel method for identify-
ing plant, fungal, or bacterial origins of amino acids. Ecol-
ogy 90: 35263535. doi:10.1890/08-1695.1
Larsen, T., T. Hansen, and J. Dierking. 2020. Characterizing
niche differentiation among marine consumers with amino
acid δ
Cngerprinting. Ecol. Evol. 10: 77687782. doi:10.
Larsen, T., M. Ventura, N. Andersen, D. M. OBrien, U.
Piatkowski, and M. D. McCarthy. 2013. Tracing carbon
sources through aquatic and terrestrial food webs using
amino acid stable isotope ngerprinting. PLoS One 8:
e73441. doi:10.1371/journal.pone.0073441
Laws, E. A., R. R. Bidigare, and B. N. Popp. 1997. Effect of
growth rate and CO
concentration on carbon isotopic frac-
tionation by the marine diatom Phaeodactylum tricornutum.
Limnol. Oceanogr. 42: 15521560. doi:10.4319/lo.1997.42.
Leal, M. C., C. Ferrier-Pagès, R. Calado, M. E. Thompson, M. E.
Frischer, and J. C. Nejstgaard. 2014. Coral feeding on
microalgae assessed with molecular trophic markers. Mol.
Ecol. 23: 38703876. doi:10.1111/mec.12486
Leal, M. C., K. Hoadley, D. T. Pettay, A. Grajales, R. Calado,
and M. E. Warner. 2015. Symbiont type inuences trophic
plasticity of a model cnidarian-dinoagellate symbiosis.
J. Exp. Biol. 218: 858863. doi:10.1242/jeb.115519
Lin, S., and others. 2015. The Symbiodinium kawagutii
genome illuminates dinoagellate gene expression and
coral symbiosis. Science 350: 691694. doi:10.1126/
Markell, D. A., and R. K. Trench. 1993. Macromolecules
exuded by symbiotic dinoagellates in culture: Amino acid
and sugar composition. J. Phycol. 29:6468. doi:10.1111/j.
Martinez, S., and others. 2020. Energy sources of the depth-
generalist mixotrophic coral Stylophora pistillata. Front.
Mar. Sci. 7: 988. doi:10.3389/fmars.2020.566663
McCarthy, M. D., R. Benner, C. Lee, and M. L. Fogel. 2007.
Amino acid nitrogen isotopic fractionation patterns as indi-
cators of heterotrophy in plankton, particulate, and dis-
solved organic matter. Geochim. Cosmochim. Acta 71:
47274744. doi:10.1016/j.gca.2007.06.061
McCarthy, M. D., J. Lehman, and R. Kudela. 2013. Com-
pound-specic amino acid δ
N patterns in marine algae:
Tracer potential for cyanobacterial vs. eukaryotic organic
nitrogen sources in the ocean. Geochim. Cosmochim. Acta
103: 104120. doi:10.1016/j.gca.2012.10.037
McMahon, K. W., S. R. Thorrold, L. A. Houghton, and M. L.
Berumen. 2015. Tracing carbon ow through coral reef
food webs using a compound-specic stable isotope
approach. Oecologia 180: 809821. doi:10.1007/s00442-
McMahon, K. W., and S. D. Newsome. 2019. Chapter 7
Amino acid isotope analysis: A new frontier in studies
of animal migration and foraging ecology, p. 173190.
In K.A.HobsonandL.I.Wassenaar[eds.],Trackingani-
mal migration with stable isotopes, 2nd ed. Academic
Muscatine, L., L. R. McCloskey, and R. E. Marian. 1981.
Estimating the daily contribution of carbon from zooxan-
thellae to coral animal respiration. Limnol. Oceanogr. 26:
601611. doi:10.4319/lo.1981.26.4.0601
Muscatine, L., J. W. Porter, and I. R. Kaplan. 1989. Resource
partitioning by reef corals as determined from stable
isotope composition: I. δ
C of zooxanthellae and animal
tissue vs depth. Mar. Biol. 100: 185193. doi:10.1007/
Muscatine, L., and I. R. Kaplan. 1994. Resource partitioning by
reef corals as determined from stable isotope composition
N of zooxanthellae and animal tissue versus depth.
Pac. Sci. 48: 304312.
OConnell, T. C. 2017. Trophicand sourceamino acids in
trophic estimation: A likely metabolic explanation.
Oecologia 184: 317326. doi:10.1007/s00442-017-3881-9
Ohkouchi, N., and others. 2017. Advances in the application
of amino acid nitrogen isotopic analysis in ecological and
biogeochemical studies. Org. Geochem. 113: 150174. doi:
Wall et al. Coral nutrition and amino acid isotopes
Oksanen, J., and others. 2019. vegan: Community ecology
package. 631637.
Palardy, J. E., A. G. Grottoli, and K. A. Matthews. 2005. Effects
of upwelling, depth, morphology and polyp size on feeding
in three species of Panamanian corals. Mar. Ecol. Prog. Ser.
300:7989. doi:10.3354/meps300079
Papina, M., T. Meziane, and R. van Woesik. 2003. Symbiotic
zooxanthellae provide the host-coral Montipora digitata with
polyunsaturated fatty acids. Comp. Biochem. Physiol. B
Biochem. Mol. Biol. 135: 533537. doi:10.1016/s1096-4959
Phillips, D. L., and J. W. Gregg. 2001. Uncertainty in source
partitioning using stable isotopes. Oecologia 127: 171179.
Popp, B. N., B. S. Graham, R. J. Olson, C. C. S. Hannides, M. J.
Lott, G. A. López-Ibarra, F. Galván-Magaña, and B. Fry.
2007. Insight into the trophic ecology of yellown tuna,
Thunnus albacares, from compound-specic nitrogen iso-
tope analysis of proteinaceous amino acids. Isotopes as
Indicators of Ecological Change 173190. doi:10.1016/
Pyle, R. L., and others. 2016. A comprehensive investigation
of mesophotic coral ecosystems in the Hawaiian Archipel-
ago. PeerJ 4: e2475. doi:10.7717/peerj.2475
R Core Team. 2019. R: A language and environment for statis-
tical computing. R Foundation for Statistical Computing.
Radice, V. Z., O. Hoegh-Guldberg, B. Fry, M. D. Fox, and S. G.
Dove. 2019. Upwelling as the major source of nitrogen for
shallow and deep reef-building corals across an oceanic
atoll system. Funct. Ecol. 33: 11201134. doi:10.1111/
Rangel, M. S., D. Erler, A. Tagliaco, K. Cowden, S. Scheffers,
and L. Christidis. 2019. Quantifying the transfer of prey
N signatures into coral holobiont nitrogen pools. Mar.
Ecol. Prog. Ser. 610:3349. doi:10.3354/meps12847
Reynaud, S., P. Martinez, F. Houlbrèque, I. Billy, D. Allemand,
and C. Ferrier-Pagès. 2009. Effect of light and feeding on
the nitrogen isotopic composition of a zooxanthellate
coral: Role of nitrogen recycling. Mar. Ecol. Prog. Ser. 392:
103110. doi:10.3354/meps08195
Rodrigues, L. J., and A. G. Grottoli. 2006. Calcication rate
and the stable carbon, oxygen, and nitrogen isotopes in