ArticlePDF Available

Abstract and Figures

Mobulid rays, a group of closely related filter-feeders, are threatened globally by bycatch and targeted fisheries. Their habitat use and feeding ecology are not well studied, and most efforts have focused on temporally limited stomach content analysis or inferences from tagging data. Previous studies demonstrate a variety of different diving behaviors across species, which researchers have interpreted as evidence of disparate foraging strategies. However, few studies have examined feeding habitats and diets of multiple mobulid species from a single location, and it is unclear if the proposed differences in diving and inferred foraging behavior are examples of variability between species or regional adaptations to food availability. Here, we use stable isotope data from mobulids landed in fisheries to examine the feeding ecology of 5 species at 3 sites in the Indo-Pacific. We use Bayesian mixing models and analyses of isotopic niche areas to demonstrate dietary overlap between sympatric mobulid species at all of our study sites. We show the degree of overlap may be inversely related to productivity, which is contrary to prevailing theories of niche overlap. We use isotope data from 2 tissues to examine diet stability of Manta birostris and Mobula tarapacana in the Philippines. Finally, we observe a significant but weak relationship between body size and isotope values across species. Our findings highlight challenges to bycatch mitigation measures for mobulid species and may explain the multi-species mobulid bycatch that occurs in a variety of fisheries around the world.
Content may be subject to copyright.
MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
Vol. 580: 131–151, 2017
https://doi.org/10.3354/meps12304 Published September 29
INTRODUCTION
Mobulid rays are a group of closely related, highly
derived filter-feeders that, in many cases, have over-
lapping habitats and geographic distributions (Cou-
turier et al. 2012). Periods of cladogenesis within the
Mobulidae coincide with periods of global warming,
and speciation within the family has occurred as re -
cently as within the last million years (Kashiwagi et
al. 2012, Poortvliet et al. 2015). Hypotheses for the
drivers of these speciation events include fragmenta-
tion of productive upwelling environments and
reduced food availability during extended periods of
global warming, and physical barriers to dispersal
and connectivity during ice ages (Poortvliet et al.
2015). The most re cently diverged mobulid species
© Inter-Research 2017 · www.int-res.com*Corresponding author: j8stewart@ucsd.edu
Trophic overlap in mobulid rays:
insights from stable isotope analysis
Joshua D. Stewart1, 2,*, Christoph A. Rohner3, Gonzalo Araujo4,
Jose Avila5, Daniel Fernando2,6,7, Kerstin Forsberg5, Alessandro Ponzo4,
Joshua M. Rambahiniarison4, Carolyn M. Kurle8, Brice X. Semmens1
1Scripps Institution of Oceanography, University of California San Diego, La Jolla, CA 92093, USA
2The Manta Trust, Dorset DT2 0NT, UK
3Marine Megafauna Foundation, Praia do Tofo, Inhambane, Mozambique
4Large Marine Vertebrates Research Institute Philippines, Jagna, Bohol 6308, Philippines
5Planeta Oceano, Lima 15074, Peru
6Department of Biology and Environmental Science, Linnaeus University, Kalmar 39182, Sweden
7Blue Resources Trust, Colombo 00700, Sri Lanka
8Division of Biological Sciences, University of California San Diego, La Jolla, CA 92093, USA
ABSTRACT: Mobulid rays, a group of closely related filter-feeders, are threatened globally by
bycatch and targeted fisheries. Their habitat use and feeding ecology are not well studied, and
most efforts have focused on temporally limited stomach content analysis or inferences from tag-
ging data. Previous studies demonstrate a variety of different diving behaviors across species,
which researchers have interpreted as evidence of disparate foraging strategies. However, few
studies have examined feeding habitats and diets of multiple mobulid species from a single loca-
tion, and it is unclear if the proposed differences in diving and inferred foraging behavior are
examples of variability between species or regional adaptations to food availability. Here, we use
stable isotope data from mobulids landed in fisheries to examine the feeding ecology of 5 species
at 3 sites in the Indo-Pacific. We use Bayesian mixing models and analyses of isotopic niche areas
to demonstrate dietary overlap between sympatric mobulid species at all of our study sites. We
show the degree of overlap may be inversely related to productivity, which is contrary to prevail-
ing theories of niche overlap. We use isotope data from 2 tissues to examine diet stability of Manta
birostris and Mobula tarapacana in the Philippines. Finally, we observe a significant but weak
relationship between body size and isotope values across species. Our findings highlight chal-
lenges to bycatch mitigation measures for mobulid species and may explain the multi-species
mobulid bycatch that occurs in a variety of fisheries around the world.
KEY WORDS: Niche overlap · Mixing model · Feeding ecology · Bycatch risk
Resale or republication not permitted without written consent of the publisher
Mar Ecol Prog Ser 580: 131–151, 2017
132
shows evidence of hybridization (Walter et al. 2014),
suggesting that differences in behavior and habitat
use, as opposed to geographic isolation, may be the
primary factors driving and maintaining speciation
(Kashiwagi et al. 2011, 2012).
Over the past decade, demand for mobulid gill
plates in Asian medicine has led to targeted fisheries
and increased bycatch retention of mobulid rays
(Couturier et al. 2012). While these growing targeted
fisheries have catalyzed focused conservation and
scientific attention for mobulids (Ward-Paige et al.
2013, White et al. 2015, Lewis et al. doi:10.7287/
peerj. preprints.1334v1), bycatch of these species
likely impacted populations long before large-scale
targeted fisheries began. Mobulid rays are vulnera-
ble to incidental capture in gill nets, purse seines,
trawls, and even long lines (Croll et al. 2016). More-
over, their low annual reproductive output and con-
servative demographic characteristics make them
highly susceptible to fisheries-induced population
declines (Dulvy et al. 2014, Pardo et al. 2016), even
when catch rates are low. Understanding horizontal
and vertical habitat use, which are likely driven by
foraging in many cases, can help determine when
mobulid rays are most vulnerable to incidental cap-
ture in fisheries and can aid in the development of
bycatch mitigation measures (Stewart et al. 2016b).
Several studies have examined the horizontal and
vertical movements of mobulid rays using satellite
telemetry (Canese et al. 2011, Croll et al. 2012, Braun
et al. 2014, Jaine et al. 2014, Thorrold et al. 2014,
Stewart et al. 2016b), and others have used direct
observations to identify feeding patterns and prey
sources (Notarbartolo di Sciara 1988). The results of
these studies demonstrate differences in habitat use
(e.g. Croll et al. 2012, Thorrold et al. 2014), vertical
movements (e.g. Canese et al. 2011, Croll et al. 2012,
Braun et al. 2014, Jaine et al. 2014, Thorrold et al.
2014, Stewart et al. 2016b), and in some cases prey
sources between mobulid species (Notarbartolo di
Sciara 1988). However, few studies have examined
the feeding ecology of multiple mobulid species in a
given region or within a single species at multiple
locations (see Notarbartolo di Sciara 1988, Sampson
et al. 2010). Consequently, it remains unclear whether
the observed differences in foraging behavior and
habitat use are consistent between species or simply
a result of regional variations in resource availability.
Armstrong et al. (2016) demonstrated that mantas
require zooplankton to reach threshold densities
before feeding becomes energetically profitable, and
therefore regional patterns in prey availability could
conceivably have a greater influence on mobulid
feeding behavior than morphological and behavioral
differences between species.
Researchers are increasingly using stable isotope
analysis as a tool for inferring the trophic ecology of
animals, providing insight into trophic niche separa-
tion (Cherel et al. 2007, Plass-Johnson et al. 2013),
trophic overlap (Foley et al. 2014, Jackson et al.
2016), diet composition (Semmens et al. 2009), and
diet shifts (Ben-David et al. 1997, MacNeil et al.
2005). Stable isotopes of carbon (δ13C) and nitrogen
(δ15N) are 2 of the most commonly used isotopes in
ecological studies, as they can provide information
on feeding locations and prey types due to pre-
dictable changes in δ13C across habitats (France
1995) and increases in δ15N at higher trophic levels
(Owens 1987). Importantly, the use of stable isotopes
to infer trophic dynamics requires a number of ex -
perimentally validated parameters, including frac-
tionation and tissue incorporation rates (Gannes et
al. 1997), which can be challenging to obtain for
species that are not easily studied in a laboratory
setting. Further, factors such as diet composition,
body condition, organism size, and metabolic rate
can influence isotopic incorporation, and few studies
have explicitly examined these effects (Gannes et
al. 1997, Newsome et al. 2010). While acknowledg-
ing these limitations, isotope ana lysis is particularly
useful in marine species, as it can provide a large
temporal window into the diet of animals that are
otherwise challenging to observe in the wild for
long periods (e.g. MacNeil et al. 2005, Cherel et al.
2007). In most regions, mobulid rays are present spo-
radically and unreliably at sites accessible by re -
searchers, making direct observations challenging
(Couturier et al. 2012). With the exception of coastally
associated reef manta rays (e.g. Anderson et al. 2011,
Jaine et al. 2012) and occasional opportunistic obser-
vations (e.g. Stewart et al. 2016b), direct observations
of feeding behavior in mobulids are rare and largely
restricted to surface feeding and daylight hours,
complicating efforts to characterize the trophic ecol-
ogy of this family. These characteristics make stable
isotope analysis a powerful tool for studying the
trophic ecology of mobulid rays.
In this study, we examine the trophic ecology and
isotopic niche separation of 5 species of mobulid rays
using stable isotope analysis. We use a Bayesian mix-
ing model to assess the diet contribution of prey
sources identified and sampled from stomach con-
tents at 1 study site, and we compare the isotopic
niche areas and isotopic niche separation between
species and across regions. We use samples collected
from fisheries landings at 3 sites throughout the Indo-
Stewart et al.: Trophic overlap in mobulid rays 133
Pacific to examine how resource partitioning may
change across regions with varying productivity and
resource availability.
MATERIALS AND METHODS
Study sites and sample collection
Our study sites included (1) the coast of northern
Peru, (2) Sri Lanka, and (3) the Bohol Sea in the
Philippines (Fig. 1). We analyzed muscle samples
from 5 species of mobulid rays (4 species per site)
from Peru (n = 46), the Philippines (n = 192), and Sri
Lanka (n = 102) and liver samples from 2 species in
the Philippines (n = 30) (Table 1).
Peru
In Peru, we collected skeletal muscle samples of
Manta birostris (n = 3), Mobula japanica (n = 20),
Mobula munkiana (n = 18), and Mobula thurstoni
(n = 5) at landing sites throughout the Tumbes
region, in the north of the country, from August 2012
through May 2013. Fishers typically caught mobulid
rays within approximately 10 to 30 km of shore (over
the continental shelf) as non-discarded bycatch in
purse seines and gill nets targeting tuna. Rays were
landed either whole, gutted, or in many cases only as
pectoral fins. We only included samples in this study
from individuals that could be confidently identified
from images recorded at the time of tissue collection
(Stevens 2014). Where possible, we recorded disc
width of landed individuals (Table 1).
Sri Lanka
In Sri Lanka, we collected skeletal muscle samples
from M. birostris (n = 37), M. japanica (n = 27), Mob-
ula tarapacana (n=27), and M. thurstoni (n = 11)
landed at fish markets in Negombo, in the west of the
country, and Mirissa, in the south, from November
2010 through August 2013. In most cases, fishers
reported catching the mobulids within 20 km of shore
on the edge of the continental shelf, while other sam-
ples were collected from mobulids caught on long-
range expeditions within and outside the exclusive
economic zone and off the continental shelf. Rays
were landed gutted, and we recorded disc width of
individuals where possible (Table 1).
Fig. 1. Study sites. Bounding boxes for chl asatellite data are displayed in the lower map, while bathymetric maps of the study
sites are displayed within those bounding boxes for (a) Peru, (b) Sri Lanka, and (c) Philippines. Orange circles indicate the
locations of primary tissue sample collection sites for each region
Mar Ecol Prog Ser 580: 131–151, 2017
Philippines
In the Philippines, we analyzed samples collected
from M. birostris (n = 42), M. japanica (n = 42), M.
tarapacana (n = 35), and M. thurstoni (n = 73) at a
landing site in Jagna, on the island of Bohol, from
December 2012 through May 2014. Mobulids were
landed primarily during the dry season from mid-
November to mid-June. Fishers captured mobulids in
the central and eastern Bohol Sea, typically between
5 and 50 km from shore at night and in the top 30 m
of the water column over depths greater than 1000 m.
We collected skeletal muscle tissue and recorded
disc width where possible (Table 1). We collected
stomach contents from a subset of individuals across
all species, and we collected liver samples from a
subset of M. tarapacana (n = 15) and M. birostris (n =
15). In some cases, prey sources were present in a
number of different stomach content samples but in
quantities that were too small to prepare for stable
isotope analysis (e.g. copepods, chaetognaths, and
pteropods; see Rohner et al. 2017 for a detailed ana -
lysis of stomach contents). While these prey sources
did not make up a substantial portion of the diet dur-
ing the period when stomach contents were col-
lected, their relative importance may change through-
out the year. Because isotope analysis allows for
dietary in sights over a longer tissue integration
period, we chose to include these sources in our ana -
lyses. To obtain adequate material for isotope ana -
lysis, we performed 1 plankton tow in Pintuyan, South-
ern Leyte, during February 2016. We only included
copepods, chaetognaths, and pteropods from the
plankton tow in isotope analyses; no other species or
groups that were not present in stomach contents
were included.
134
Region/species No. of Disc width Converted mass Tissue turnover δ15N (‰) δ13C (‰) LE δ13C (‰) Bulk C:N LE C:N
samples (cm) (mean, kg) (mean, d)
Peru
Manta birostris 3 365.3 ± 265.9 601.5 550.8 10.4 ± 1.3 −16.7 ± 1.0 – 3.1 ± 0.1
Mobula japanica 20 172.0 ± 61.2 42.9 324.9 10.9 ± 1.0 −17.4 ± 0.4 – 3.1 ± 0.1
Mobula munkiana 18 143.4 ± 40.5 25.3 292.4 12.5 ± 0.2 −17.1 ± 0.2 – 3.2 ± 0.1
Mobula thurstoni 5 90.4 ± 8.5 8.0 231.9 9.9 ± 0.9 −17.6 ± 0.3 – 3.2 ± 0.1
Philippines
M. birostris 42 438.5 ± 55.4 1024.8 612.8 9.6 ± 0.4 −16.5 ± 0.7 – 3.2 ± 0.2
M. birostris liver 15 – 245.1 9.1 ± 0.4 −22.9 ± 0.7 −19.8 ± 1.3 17.9 ± 4.8 7.9 ± 1.5
M. japanica 42 192.2 ± 24.3 62.7 350.5 9.9 ± 0.4 −16.2 ± 0.4 – 3.1 ± 0.1
Mobula tarapacana 35 237.9 ± 49.6 171.8 428.7 9.8 ± 0.6 −16.1 ± 0.6 – 3.1 ± 0.1
M. tarapacana liver 15 – 171.5 10.1 ± 0.6 −21.9 ± 0.7 −19.5 ± 0.9 14.5 ± 4.0 7.3 ± 1.2
M. thurstoni 73 147.4 ± 24.1 31.0 304.4 9.5 ± 0.4 −16.2 ± 0.5 – 3.1 ± 0.1
Sri Lanka
M. birostris 37 253.9 ± 54.6 208.0 445.4 10.6 ± 0.8 −17.6 ± 0.4 – 3.2 ± 0.1
M. japanica 27 207.0 ± 21.9 80.6 368.5 10.7 ± 0.5 −17.8 ± 0.5 – 3.2 ± 0.1
M. tarapacana 27 205.0 ± 44.2 111.4 393.1 10.9 ± 0.8 −17.4 ± 0.6 – 3.2 ± 0.1
M. thurstoni 11 126.3 ± 19.9 20.2 279.3 10.4 ± 0.3 −18.0 ± 0.4 – 3.2 ± 0.1
Philippines stomach contents
Euphausiids 40 – – – 7.6 ± 0.5 −18.3 ± 0.4 – 3.7 ± 0.2
Copepodsa 3 – – – 6.9 ± 0.4 −19.8 ± 0.8 – 3.9 ± 0.5
Chaetognathsa 3 – – – 8.4 ± 0.4 −18.8 ± 0.5 – 3.3 ± 0.2
Pteropods 1 – – – 5.24 −12.2 8.2 –
Pteropodsa 6 – – – 6.2 ± 0.1 −14.6 ± 2.5 – 4.3 ± 0.9
Myctophids 11 – – – 9.0 ± 0.8 −17.9 ± 0.9 – 3.3 ± 0.4
Sardinella spp. 6 – – 9.7 ± 0.4 −17.8 ± 0.8 – 4.1 ± 0.6
Cubiceps spp. 3 – – – 8.4 ± 0.3 −17.3 ± 0.3 – 3.5 ± 0.1
aSamples collected in a zooplankton tow
Table 1. Summary information for tissues collected from mobulids, including mass conversions and tissue turnover rates, and their prey.
All values are means ± SD. All mobulid tissues were skeletal muscle unless otherwise noted. Liver tissues were high in lipids, which leads
to lower δ13C values; therefore, we lipid extracted (LE) the liver tissues and report δ13C values for intact (bulk) and LE liver tissues (see
‘Materials and methods’)
Stewart et al.: Trophic overlap in mobulid rays
We visually sorted prey collected in mobulid stom-
ach contents and plankton tows to the species level
except for copepods, which we pooled across species
to obtain enough material for isotope analysis. We
further sorted prey species using microscopy (details
reported in Rohner et al. 2017). In 2 cases, fish sam-
ples collected from M. tarapacana stomachs could
not be visually identified. We extracted DNA from
these samples at Scripps Institution of Oceanography
using a Qiagen DNeasy kit and sequenced 16S genes
for genetic identification. We matched these se -
quences to existing barcodes using the basic local
alignment search tool database (Madden 2002),
which identified them as Sardinella sp. and Cubiceps
sp. Myctophids were typically found whole and intact
in M. birostris stomach contents, and we subsampled
these to include a portion of skin, connective tissue,
and muscle, which we homogenized and included as
a single sample per fish. In all cases, only 1 fish was
present per stomach content sample (n = 11). The
remains of the partially digested Sardinella sp. and
Cubiceps sp. from the M. tarapacana stomach con-
tents were homogenized for analysis after removing
degraded tissue. We found multiple specimens of
Sardinella sp. (n = 6) and Cubiceps sp. (n = 3) per
stomach content sample, and we analyzed each indi-
vidual fish as a separate prey sample. Euphausiids,
copepods, and chaetognaths were each pooled
(within species) to obtain sufficient tissue for isotope
analysis. We took 1 pooled subsample of euphausiids
per stomach content sample (n = 40 samples). We
separated partially digested from intact euphausiids
within each stomach content sample and only in -
cluded intact euphausiids in our subsamples for iso-
tope analysis. We took multiple pooled subsamples of
copepods (n = 3 samples) and chaetognaths (n = 3
samples) from 1 plankton tow. In the case of ptero-
pods, 1 individual constituted 1 sample, and we ob -
tained specimens from 1 stomach content sample
(n = 1 individual) and 1 plankton tow (n = 6 individu-
als) (Table 1). While additional samples of prey spe-
cies collected by plankton tows across a longer
period of time would have been ideal, logistical con-
straints restricted our available samples to a single
plankton tow to supplement stomach contents in the
present study.
Sample storage
As remote tropical sampling locations generally
did not allow for preferred preservation methods
such as freezing, we stored tissue, stomach content,
and plankton tow samples in 95% ethanol. While
there is no consensus on the effect of ethanol on δ13C
and δ15N stable isotope values (Sarakinos et al. 2002,
Barrow et al. 2008, Burgess & Bennett 2017), most
preservation studies on fish muscle and fin tissue
indicate there are negligible effects of ethanol on
δ15N values, and the shift in δ13C values after long-
term storage in ethanol is low (typically a mean
increase of 0.5 to 1.5‰) (Kaehler & Pakhomov 2001,
Sarakinos et al. 2002, Kelly et al. 2006, Vizza et al.
2013, Stallings et al. 2015). Additionally, Burgess &
Bennett (2017) suggest that storage of elasmobranch
tissues in ethanol mimics the effects of urea extrac-
tion, perhaps by acting as a solvent. In Peru, samples
were stored in 96% ethanol + 0.1 mM EDTA, as they
were initially intended for genetic analysis. To our
knowledge, there are no studies that examine the
effects of ethanol and EDTA preservation on isotope
values. Sample preservation studies have examined
the impacts of DMSO storage on isotope values
(Lesage et al. 2010), and in at least 1 case, a DMSO +
EDTA solution was used (Hobson et al. 1997). The
mean effect of DMSO and DMSO + EDTA storage on
isotope values was similar for both δ13C (−4.74 vs.
−5.1) and δ15N (−0.7 vs −0.9), suggesting that EDTA
does not have a substantial additional impact on iso-
tope values. Importantly, where EDTA was added to
a DMSO buffer, the variance of isotope values did
not change as compared to frozen samples (Hobson
et al. 1997), which would impact estimates of iso-
topic niche area in the present study. The storage
and preservation of samples in ethanol (or ethanol +
EDTA) makes them challenging or impossible to
compare with other isotopic values in the literature
for the same species or potential source items that
were preserved differently. However, these storage
methods should not affect the estimation of isotopic
niche area or within-region comparison of sources
and consumers given the consistent preservation
methodology utilized across samples. We did not
compare isotope values among samples that were
preserved only in ethanol with those preserved in
ethanol + EDTA.
Isotope analysis
For stable isotope sample preparation, we soaked
samples in deionized water for 5 min and then rinsed
them to remove debris and residual ethanol from
storage. We then freeze-dried approximately 10 mg
(wet) of tissue from each individual for 24 h using a
FreeZone 2.5 freeze drier (Labconco). In the case of
135
Mar Ecol Prog Ser 580: 131–151, 2017
136
liver samples, we freeze-dried 30 mg (wet) of tissue
from each individual for 72 h.
The high lipid content of elasmobranch livers can
result in lower δ13C values (Logan & Lutcavage 2010,
Kim & Koch 2012). We therefore lipid-extracted liver
samples using petroleum ether (Dobush et al. 1985),
following the protocol outlined in Kim & Koch (2012).
As lipid extraction with petroleum ether appears to
affect δ15N values (Parng et al. 2014), we subsampled
liver prior to lipid extraction and used non-lipid-ex-
tracted samples for δ15N values and lipid-extracted
samples for δ13C values. We did not lipid-extract mus-
cle tissue, as their carbon to nitrogen (C:N) ratios fell
below the C:N threshold of 3.5 suggested by Post et
al. (2007), indicating these tissues had sufficiently low
lipid content that would not affect δ13C values (see
Table 1). We did not lipid-extract prey samples be-
cause their C:N values (3.8 ± 0.7, mean ± SD) indi-
cated they were not lipid-rich as defined by Newsome
et al. (2010), and the tissue quantity of many prey
samples (especially copepods and chaetognaths) was
so low that lipid extraction was not practical. However,
to assess the possible impacts of the lipid content of
prey samples on our mixing models, we did apply
mathematical lipid normalization equations (see ‘Ma-
terials and methods: Statistics and mixing models’).
We homogenized pteropods whole without acid-
washing or removing carbonate shell components,
which may have artificially increased δ13C values for
pteropods in our results (Mateo et al. 2008).
We homogenized dried mobulid muscle tissue and
prey fish samples using a Wig-L-Bug dental amal-
gamator and mobulid liver and zooplankton prey
samples manually using a mortar and pestle. We then
packaged between 0.5 and 1.0 mg of powdered tis-
sue in 5 × 9 mm tin capsules (Costech). Samples were
analyzed at the University of California Santa Cruz
Stable Isotope Laboratory using an NC2500 elemen-
tal analyzer (CE Instruments) interfaced to a Delta
Plus XP isotope ratio mass spectrometer (Thermo -
Finnigan) with an acetanilide standard.
We estimated tissue-specific stable isotope turn-
over times for skeletal muscle by calculating the
average mass of each species based on species-
specific disc width to mass conversions from Notar-
bartolo di Sciara (1988) and using the body mass tis-
sue incorporation rates for carbon and nitrogen in
teleosts and elasmobranchs described in Kim et al.
(2012b). No disc width to mass conversion is avail-
able for M. birostris, so we used the conversion for-
mula for M. tarapacana, which is the second-largest
mobulid species in the present study. MacNeil et al.
(2006) estimated the incorporation rates of nitrogen
in liver tissue to be approximately 40% of the incor-
poration rates of nitrogen for muscle tissue in a con-
trolled feeding experiment using freshwater sting -
rays, and we therefore applied this scaling factor to
our muscle tissue nitrogen incorporation rates for M.
birostris and M. tarapacana to arrive at approxi mate
nitrogen incorporation rates for liver.
Statistics and mixing models
We performed all statistical analyses using the R
software program (R Core Team 2016). We used the
Stable Isotope Bayesian Ellipses in R (SIBER) pack-
age (Jackson et al. 2011) to create standard ellipses
representing relative isotopic niche widths in bivari-
ate δ13C and δ15N space, where the standard ellipse
represents bivariate standard deviation. We gener-
ated Bayesian credible intervals of the standard
ellipse area of each species as well as for the overall
mobulid assemblage (all species combined) within
each region. We computed the pairwise overlaps be -
tween mobulid species within each region and used
the mean proportional overlap among species as a
metric for community overlap. Bayesian estimation of
standard ellipses allows for an unbiased estimate of
relative isotopic niche area even at small sample
sizes, in contrast with metrics such as convex hulls,
which are positively correlated with sample size
(Jackson et al. 2011). Therefore, our median esti-
mates of isotopic niche metrics should not be im -
pacted by small sample sizes, although uncertainty
and therefore credible interval width should be
greater. We excluded M. birostris in Peru from iso-
topic niche area comparisons, as the extremely large
credible intervals resulting from small sample size
(n = 3) made comparisons uninformative. However,
we did include M. birostris in the mean proportional
isotopic niche overlap calculations for Peru to keep
the total number of species constant at each site to
allow for comparisons across regions. For all SIBER
analyses, we used 2 chains of 10 000 iterations with a
burn-in of 1000 and thinning of 10.
To determine the contribution of prey sources to
mobulids’ diets in the Philippines, we used MixSIAR,
a Bayesian stable isotope mixing model (Stock &
Semmens 2016). We included 7 prey sources in the
model: euphausiids, copepods, chaetognaths, mycto -
phids, Sardinella, Cubiceps, and pteropods. We sub-
sequently made 2 combinations of 3 groups a post -
eriori by summing their model output posterior
distributions based on their functional role, because 7
sources would likely be confounded with only 2 trac-
Stewart et al.: Trophic overlap in mobulid rays
ers (δ15N and δ13C) (B. X. Semmens et al. unpubl.).
The grouping scenarios were (1) zooplankton (chae -
to gnaths, copepods, euphausiids), fish (myctophids,
Sardinella, Cubiceps), and pteropods; and (2) epi -
pelagic (chaetognaths, copepods, Sardinella), meso-
pelagic (euphausiids, myctophids, Cubiceps), and
ptero pods. We kept pteropods separate from the
aggregated groupings due to the dissimilarity of their
isotopic signature and possible influences of carbon-
ate shells on the measured isotope values. We did not
combine these groups a priori (before inclusion in the
model) be cause the combined source groups were
not bi variate normally distributed, which is an expec-
tation of the mixing model (B. X. Semmens et al.
unpubl.).
Dietary lipids in prey sources may be routed directly
to consumer tissues (Newsome et al. 2010, Parng et
al. 2014), making it more appropriate to leave prey
samples with lipids intact. However, to ac count for
uncertainty in the importance of dietary lipid con-
tent, as well as possible influences of carbonate shells
in our pteropod samples, we ran 2 versions of our
mixing models. The first included bulk prey samples
and bulk consumer samples, while the second used
correction factors to account for lipid content and car-
bonate shells. For the second mixing model, we
applied muscle tissue-specific fish C:N correction
factors for δ13C (Logan et al. 2008) to our myctophid,
Sardinella, and Cubiceps sources and euphausiid
species-specific C:N correction factors for δ13C to our
chaetognath, copepod, euphausiid, and pteropod
sources, because euphausiids were the only marine
invertebrates for which correction factors were avail-
able (Logan et al. 2008). After accounting for lipid
content, we applied a correction factor of −6 δ13C
to pteropod samples, representing the mean differ-
ence in whole versus acid-washed marine gastropods
reported in (Mateo et al. 2008). We ran each version
of the mixing model (bulk and lipid/ carbonate cor-
rected) with both informative and uninformative pri-
ors as specified below.
Trophic discrimination factors, or the differences in
prey and consumer isotope values that result from
metabolic processes, are an important aspect of mix-
ing models, but the determination of appropriate
trophic discrimination factors has remained a hotly
contested debate in the literature (Gannes et al. 1997,
Caut et al. 2008, Hussey et al. 2010b). Determining
discrimination factors generally requires laboratory
experiments or controlled feeding studies, which are
not practical or readily available for many species,
including mobulids. Couturier et al. (2013) proposed
discrimination factors of 2.4‰ for nitrogen and 1.3
for carbon based on stable isotope values of wild reef
manta rays and their putative prey sources. These
values fall within the range of other published elas-
mobranch discrimination factors, although laboratory
experiments demonstrate considerable variation,
even within species (Hussey et al. 2010a, Kim et al.
2012a, Malpica-Cruz et al. 2012). One advantage of
Bayesian mixing models is their ability to incorporate
uncertainty in parameter estimates and propagate
this uncertainty throughout the model (Moore &
Semmens 2008). We used mean values of 2.4‰ for
nitrogen and 1.3‰ for carbon (Couturier et al. 2013)
but used a standard deviation of 1‰ for both isotopes
in our mixing models to account for the variability in
laboratory-derived trophic discrimination factors, the
uncertainty of the discrimination factors proposed for
wild reef manta rays, and possible differences be -
tween mobulid species. Specifically, we selected a
standard deviation of 1‰, as it adequately covers the
range of δ15N and δ13C trophic discrimination factors
reported in laboratory experiments with elasmo-
branchs (Hussey et al. 2010a, Kim et al. 2012a,
Malpica-Cruz et al. 2012).
In addition to trophic discrimination factors, Baye -
sian mixing models require prior distributions for
proportional diet contribution of all prey sources. We
ran 2 different iterations of our mixing model. The
first model used priors that are uninformative on the
simplex, where all possible sets of diet proportions
are given equal weight in the prior distributions (B.
X. Semmens et al. unpubl.). Priors are specified using
a Dirichlet distribution, where an uninformative prior
has a probability αi= 1/(n sources) for each source i.
The second model used semi-informative priors,
which we based on the presence and abundance of
the 7 prey sources in stomach content samples from
each species. An informative prior could be specified
as αi,j = n records of prey species iin stomach con-
tents of consumer species j. However, in some cases,
1 or 2 prey sources dominated stomach content sam-
ples or may have been the only prey sources present
within a species’ stomach contents, which would
have led to extremely informative priors. Using very
strong priors essentially eliminates the utility of a
mixing model, as the posterior distributions will sim-
ply reflect the priors (B. X. Semmens et al. unpubl.).
Furthermore, the stomach contents were collected
from January to April and therefore are representa-
tive of the mobulids’ diets during only a third of the
year. Muscle tissue represents the integrated diet
over a much longer period (Table 1) and is therefore
useful in determining longer-term integration of diet
contributions. For these reasons, we qualitatively
137
Mar Ecol Prog Ser 580: 131–151, 2017
relaxed the stomach content-informed priors so that
the model had a reasonable chance of including any
of the possible prey sources, while at the same time
providing some weight to the diet preferences
implied by stomach content data. Prior distributions
for each source are presented in Fig. S1 & Table S1 in
the Supplement at www. int-res. com/ articles/ suppl/
m580 p131_ supp. pdf, and a posteriori aggregated
priors are presented in ‘Results: Mixing models’ (see
Fig. 5). We ran all mixing models with 3 chains of
300 000 iterations, a burn-in of 200 000, and thinning
of 100. We assessed convergence of all Bayesian ana -
lyses using visual inspection of chain convergence
and autocorrelation plots as well as Gelman-Rubin
diagnostics (Gelman & Rubin 1992).
To assess relationships between trophic character-
istics and body size, we used Bayesian linear mixed
effects models. To do this, we set up models with iso-
tope values (δ15N and δ13C) as the response (normally
distributed), disc width and regions as fixed effects,
and species as a random effect. We used a multivari-
ate ANOVA to examine isotopic differences between
lipid-extracted liver samples and muscle samples in
M. birostris and M. tarapacana. Because MANOVAs
do not provide significance levels for each variable,
we used paired t-tests to examine within-individual
differences in δ15N and δ13C values between liver
and muscle of individuals from which we collected
both tissue types. We applied Bonferroni corrections
for repeated measures and report corrected p-values.
Environmental data
We used the xtractomatic package (Mendelssohn
2015) in R to obtain monthly chl avalues from MODIS
satellites for the regions surrounding each of our
study sites over the past 10 yr to characterize the long-
term patterns and variability in regional productivity.
Our bounding boxes were 7.5° S, 85° W to 0°N, 80° W
(Peru); 2.5° N, 75° E to 10° N, 85° E (Sri Lanka); and
5° N, 120°E to 12.5° N, 130°E (Philippines) (see Fig. 1).
We selected these bounding boxes based on the ex-
pected distribution of mobulids in each region, which
was supported by bycatch data from the eastern
equatorial Pacific (Croll et al. 2016), interviews with
fishermen regarding capture locations in Sri Lanka,
and records of long-distance movements in several
Mobula species (Thorrold et al. 2014, Francis & Jones
2017). We averaged chl avalues over the entire
bounding box for a given region and smoothed
monthly averages into seasonal (3 mo) averages for
plotting purposes.
RESULTS
Isotopic niche areas
The stable isotope values from the 4 mobulid spe-
cies, and their resulting isotopic niche spaces, largely
overlapped among species for those sampled in Sri
Lanka and the Philippines, whereas there was a
greater separation in isotope values and isotopic
niche space among species collected in Peru (Figs. 2
& 3). The mean proportional isotopic niche overlap
among species increased from Peru to Sri Lanka to
the Philippines (0.10, 0.33, 0.36, respectively), al -
though there was a high degree of overlap between
Bayesian credible intervals of these estimates (Fig. 3,
Table S2 in the Supplement). Satellite-derived chl a
values indicate Peru had the greatest mean primary
productivity of our 3 study sites, followed by Sri
Lanka and the Philippines (1.04, 0.48, 0.19 mg m−3,
respectively), while Sri Lanka had the greatest vari-
ability (SD) in chl avalues followed by Peru and the
Philippines (SD: 0.38, 0.27, 0.06 mg m−3, respectively)
(Fig. 3).
Manta birostris and Mobula tarapacana, the largest
of the mobulids, had larger isotopic niche areas than
Mobula japanica and Mobula thurstoni in Sri Lanka
and the Philippines, while in Peru, M. japanica and
M. thurstoni had much larger isotopic niche areas
than Mobula munkiana (Fig. 4). We found no pattern
between ellipse area and region that was consistent
across species. For example, while the median ellipse
area of M. japanica was largest in Peru and smallest
in the Philippines, the median ellipse area of M.
thurstoni was largest in Peru and smallest in Sri
Lanka, and the ellipse area of M. birostris was larger
than M. tarapacana in the Philippines but smaller in
Sri Lanka (Fig. 4). We report pairwise comparisons
of isotopic niche area credible intervals and median
values of isotopic niche areas in Table S3.
Size-based differences in isotopic signatures
Our mixed effects models demonstrated significant
but weak relationships between body size (disc
width) and isotope values. The median slope across
all mobulids demonstrated an increase in the δ15N
value of 0.21‰ per meter increase in disc width and
an 89.2% probability of a slope greater than 0. Con-
versely, mobulids displayed a negative slope in δ13C
values, with a median decrease of 0.23‰ per meter
in crease in disc width across all species and a 97.47%
posterior probability of a slope less than 0. Species-
138
Stewart et al.: Trophic overlap in mobulid rays
level relationships and significance values are re-
ported in Figs. S2 & S3 in the Supplement.
Mixing models
The mixing model results from the Philippines sup-
port the data from the isotopic niche ellipses indicating
that the diets of mobulids in the region are largely
overlapping (Fig. 5). Using lipid- and carbonate-cor-
rected prey sources had a minor impact on aggregated
posterior distributions, typically reducing the diet
contribution from pteropods and slightly increasing
the contribution from fish or zooplankton (Table 2,
Table S4). All 4 species of Mobula appeared to con-
sume similar proportions of the 3 aggregated prey
groups (Fig. 5, Table 2): model outputs suggest that
diets were dominated by zooplankton, had a substan-
tial input from the pteropod source, and included a
smaller but still notable proportion of fish. In many
cases, the discrete (non-aggregated) sources were
confounded, evidenced by the influence of prior spec-
ifications on posterior distributions (e.g. euphausiids
in M. birostris and M. tarapacana, Cubiceps in M.
japanica) (Fig. S1 & Table S5). Aggregated posterior
distributions for fish and zooplankton groups, how-
ever, were robust to prior specifications (Fig. 5,
Table 2). Using a semi-informative prior did not sub-
stantially impact median estimates of diet proportions
but did substantially reduce credible intervals and the
139
Fig. 2. Isotope data with SIBER ellipses and sources (Philippines). (a) Philippines, (b) Philippines with sources, (c) Peru, (d) Sri
Lanka. Ellipses in (a), (c), and (d) represent maximum likelihood standard ellipses for each species, while analyses presented
in ‘Materials and methods’ and ‘Results’ were performed on Bayesian estimates of standard ellipses (not shown). Bars in (b)
represent mean ± SD for each dietary source in the Philippines included in the mixing models. Sources in (b) are corrected for
trophic discrimination, including the additional uncertainty in trophic discrimination factors that was incorporated into the
mixing models. Legend for species colors and shapes in all regions is listed in (a). See Table 1 for full species names
Mar Ecol Prog Ser 580: 131–151, 2017
140
skewedness and non-normality of posterior distribu-
tions. The 2 exceptions to this were the proportion of
fish in the diet of M. japanica, which shifted from 0.33
to 0.15 (median) from the uninformative to semi-infor-
mative prior, and the normality of the posterior distri-
bution of the proportion of pteropods in the diet of M.
thurstoni. At the discrete source level, both pteropods
and copepods tended to be robust to changes in prior
specification across species (Fig. S1). There was more
variability in diet contributions from the epipelagic
and mesopelagic aggregated sources across species
than from the fish and zooplankton groups (Table 2).
However, epi pelagic and mesopelagic groupings
were less robust to prior specifications. There was no
seasonal trend in isotope values for euphausiid prey
samples, which were collected across multiple months
and years. Variability in euphausiid isotope values
was similar to the variability in prey samples collected
in a single plankton tow (Table 1), suggesting that
prey samples from the single plankton tow adequately
capture the variability in prey sources for the purpose
of diet reconstruction.
Liver samples
The mean effects of lipid extraction on the isotope
values from liver samples were an increase of 2.79
on δ13C and a decrease of 0.11‰ on δ15N (Table 1). In
all analyses and discussion of liver samples, we used
the δ13C values from lipid-extracted samples and the
δ15N values from corresponding bulk (non-lipid-
extracted) samples unless otherwise specified. De -
spite repeatedly sonicating liver samples in petro-
leum ether until the solution was clear instead of a
dark orange color (indicating that lipids had been
successfully removed), the C:N ratios of liver samples
remained high (5.6 to 10.3; Table 1). There was a sig-
nificant relationship between the bulk C:N ratio and
the change in δ13C between lipid-extracted and
bulk liver samples (p = 0.015; Fig. S6). The linear
regression equation for both species combined was:
δ13Clipid-extracted δ13Cbulk = 0.089 × C:Nbulk + 1.328
However, the fit of this relationship was poor (R2=
0.137). Muscle tissue isotope values from individuals
paired with liver samples were no different from the
overall muscle tissue isotope values for either M.
birostris (δ15N: 9.6 ± 0.5; δ13C: −16.6 ± 0.8) or M. tara-
pacana (δ15N: 9.8 ± 0.6; δ13C: −16.2 ± 0.6), indicating
that the subsample of individuals with both muscle
and liver tissue was representative of the full set of
samples.
The multivariate ANOVA indicated that liver sam-
ples were significantly different from muscle samples
in both M. birostris (p < 0.001) and M. tarapacana
(p < 0.001). Paired t-tests indicated that the δ15N val-
ues were significantly different between muscle and
liver in M. birostris (p = 0.049) but not M. tarapacana
Fig. 3. Niche overlap and environmental data from Peru, Sri
Lanka, and the Philippines. (a) Mean proportional overlap be-
tween species’ isotopic niche areas. Rectangles represent the
50, 75, and 95% credible intervals (dark to light shading, re-
spectively), and black dots represent the mode values. (b) Box-
plots of monthly mean chl avalues for each of the 3 study re-
gions (boundaries defined in ‘Materials and methods’) across a
10 yr period from 2006 to 2016. (c) Smoothed 3 mo averages of
chl avalues across the same 10 yr period as in (b) for Peru
(solid line), Sri Lanka (dashed), and the Philippines (dotted)
Stewart et al.: Trophic overlap in mobulid rays
(p = 0.53). The δ13C values were significantly differ-
ent between muscle and liver in both species (p <
0.001).
DISCUSSION
Niche overlap
Trophic niche partitioning according to body size is
commonly observed in nature and is a fundamental
theory explaining coexistence of similar species in
habitats with limited resources (Hutchinson 1957).
For example, in marine fishes, body size frequently
correlates with mouth gape, which in turn deter-
mines the maximum prey size a consumer can target
(Scharf et al. 2000). In filter-feeders, the mechanism
that would facilitate trophic niche partitioning is less
clearly linked to body size, as prey items are typically
orders of magnitude smaller than a filter-feeder’s
mouth gape. Nevertheless, sympatric rorqual whales
demonstrate resource partitioning across a range of
body sizes despite morphological similarities, most
likely a function of behavioral differences (Santora et
al. 2010, Gavrilchuk et al. 2014). Like rorqual whales,
mobulid rays are morphologically similar but span a
range of body sizes across species. However, despite
the wide range of body sizes sampled in our study,
we observed a high degree of isotopic overlap for
most species within each region. Previous studies
found trophic niche overlap between several mobu-
lids in the Gulf of California, Mexico (Notarbartolo di
Sciara 1988, Sampson et al. 2010), and our results
demonstrate a similar pattern for 5 mobulid species
at 3 separate locations across the Indo-Pacific. Addi-
tionally, we observed only weak and most likely eco-
logically irrelevant relationships between body size
and δ13C and δ15N values in mobulids, further
demonstrating inter- and intraspecific trophic simi-
larities regardless of body size.
Niche overlap theory posits that overlap de creases
as interspecific competition increases, for ex ample
when resources are scarce and species are forced
to specialize to outcompete sympatric competitors
141
Fig. 4. Niche area by species and region. (a) Between-region comparisons for each species and (b) between-species compar-
isons for each region. Rectangles represent the 50, 75 and 95 % credible intervals (dark to light shading, respectively), and black
dots represent the mode values. We excluded Manta birostris collected in Peru due to large credible intervals that resulted from
small sample size (n = 3). See Table 1 for full species names
Mar Ecol Prog Ser 580: 131–151, 2017
142
Fig. 5. Mixing model estimates of diet contributions for a posteriori aggregated sources. Colored density distributions represent
a posteriori aggregated prior specifications and grey histograms represent posterior distributions in either the uninformative
(left) or informative (right) model runs. The far right panel compares posterior distributions between the uninformative and in-
formative model runs. Rectangles represent the 50, 75 and 95% credible intervals (dark to light shading, respectively), and
black dots represent the mode values (Fig. S1 in the Supplement at www. int-res. com/ articles/ suppl/ m580 p131_ supp. pdf shows
the prior distributions that were specified for each source in the model as well as source-specific posterior distributions
before a posteriori aggregation)
Stewart et al.: Trophic overlap in mobulid rays
(Pianka 1974, 1981). However, this niche overlap the-
ory (Pianka 1974) has not been definitively supported
in observational studies (Porter & Dueser 1982). An al-
ternate explanation for our re sults is that trophic niche
overlap in mobulid rays follows an inverse pattern,
where overlap increases as resources become more
scarce (e.g. Porter & Dueser 1982). A mechanism for
this increasing overlap in nutrient-poor regions may
143
Fig. 5 (continued)
Mar Ecol Prog Ser 580: 131–151, 2017
be an increased reliance on high-biomass prey
patches that are sparsely distributed and often domi-
nated by 1 or a few prey species (e.g. Rohner et al.
2015, Armstrong et al. 2016). For example, in Peru,
where strong equatorial up welling leads to overall
high surface primary productivity, zooplankton prey
is likely to be abundant. The density of vertically mi-
grating zooplankton and fish is influenced by surface
productivity (Croll et al. 2005, Hazen & Johnston
2010), and therefore an overall increase in zooplank-
ton abundance (surface-associated and deep scatter-
ing layers) is expected in a highly productive region
such as Peru. Under these conditions, different mobu-
lid species may be able to take advantage of their evo-
lutionarily distinct traits (e.g. size, maximum depth
tolerance, thermal inertia) to maximize foraging suc-
cess by feeding on their preferred zooplankton prey,
re sulting in greater trophic separation. In contrast, in
oligotrophic tropical waters such as our study site in
the Philippines, zooplankton should be lower in abun-
dance and more patchily distributed. Numerous stud-
ies demonstrate trophic niche separation and resource
partitioning in both pelagic and benthic predators in
similar oligotrophic regions (e.g. Young et al. 2010,
Heithaus et al. 2013, Pardo et al. 2015). However, fil-
ter-feeding elasmobranchs such as mobulids appear
to require prey densities to exceed a threshold level to
make feeding energetically profitable (Armstrong et
al. 2016). Consequently, as re gio nal productivity de-
clines, there may be fewer prey patches of adequate
density, resulting in multiple sympatric species con-
verging on the same high-density prey sources and
therefore greater trophic overlap such as that ob-
served in the present study. This is also supported by
observations of mobulid captures in the Philippines.
Multiple mobulid species are frequently captured in
the same gill nets, which are set at night over deep
water in the Bohol Sea when euphausiids are abun-
dant near the surface. Between 2013 and 2014, 25%
of 790 re corded fishing trips captured more than 1
species of mobulid in a single net (J. M. Rambahiniari-
son unpubl.). Whale sharks, another filter-feeding
elasmobranch, also rely on dense and often monospe-
cific prey patches to survive in oligotrophic regions
(Rohner et al. 2013, 2015), while sympatric rorqual
whales exhibit trophic niche separation in highly pro-
ductive polar foraging grounds (Santora et al. 2010,
Gavrilchuk et al. 2014). It is possible that prevalent
theories of niche overlap (May & Mac Arthur 1972,
Pianka 1974, 1981) do not adequately describe the
trophic dynamics of sympatric marine filter-feeders
due to the prey density thresholds they require to
meet energetic demands. It may also be the case that
large, long-lived, mobile animals are able to escape
bottlenecks in resource availability that would other-
wise lead to persistent trophic niche differentiation in
less productive environments. Importantly, isotopic
overlap does not translate directly to trophic overlap,
as consumers may feed on mixtures of taxonomically
distinct but isotopically similar prey items that would
lead to similar consumer isotope signatures. However,
analysis of stomach contents in the Philippines
verified that the mobulids’ diets converge for at least 6
mo of the year (Rohner et al. 2017), effectively ground
truthing our inferences from isotope data. This
demonstrates the benefits of combining these 2 ap-
proaches in dietary studies, and future analysis of
mobulid stomach contents in Sri Lanka and Peru
would help validate the inferences made here. Our
observations of these relationships between trophic
dynamics and regional productivity are limited be-
cause we have observations from only 3 locations.
144
Species/source Uninformative Semi-informative
median (95% CI) median (95% CI)
Manta birostris
Zooplankton 0.64 (0.26−0.84) 0.64 (0.47−0.79)
Fish 0.19 (0.06−0.48) 0.14 (0.04−0.32)
Pteropods 0.14 (0.03−0.47) 0.21 (0.08−0.33)
Epipelagic 0.55 (0.21−0.78) 0.33 (0.16−0.51)
Mesopelagic 0.28 (0.11−0.56) 0.45 (0.27−0.67)
Mobula japanica
Zooplankton 0.52 (0.27−0.70) 0.65 (0.48−0.80)
Fish 0.33 (0.12−0.61) 0.15 (0.04−0.32)
Pteropods 0.15 (0.03−0.31) 0.19 (0.10−0.30)
Epipelagic 0.37 (0.15−0.61) 0.21 (0.06−0.41)
Mesopelagic 0.48 (0.21−0.72) 0.60 (0.37−0.78)
Mobula tarapacana
Zooplankton 0.59 (0.08−0.82) 0.60 (0.43−0.74)
Fish 0.15 (0.02−0.83) 0.19 (0.08−0.36)
Pteropods 0.21 (0.01−0.43) 0.20 (0.12−0.30)
Epipelagic 0.28 (0.04−0.77) 0.19 (0.06−0.40)
Mesopelagic 0.52 (0.04−0.85) 0.60 (0.37−0.78)
Mobula thurstoni
Zooplankton 0.59 (0.31−0.79) 0.67 (0.48−0.88)
Fish 0.20 (0.04−0.52) 0.12 (0.03−0.37)
Pteropods 0.18 (0.02−0.43) 0.17 (0.01−0.40)
Epipelagic 0.43 (0.19−0.67) 0.32 (0.14− 0.54)
Mesopelagic 0.37 (0.13−0.64) 0.48 (0.28−0.75)
Table 2. Median diet contributions of a posteriori aggre-
gated prey groups from the MixSIAR Bayesian mixing
model. Note that epipelagic/mesopelagic groups are differ-
ent combinations of the same sources in zooplankton/fish
groups. Pteropods were kept separate from both grouping
scenarios. The groupings and specification of uninforma-
tive and semi-informative priors are explained in ‘Materials
and methods’
Stewart et al.: Trophic overlap in mobulid rays
Nevertheless, this topic warrants further study in
mobulids and perhaps filter-feeders more broadly.
Mobula munkiana appears to be an exception to
the pattern of trophic overlap as the only mobulid in
our study with an isotopic niche that is almost
entirely non-overlapping with other mobulids in the
same region. M. munkiana had no overlap with
Manta birostris or Mobula thurstoni and only a 6%
median isotopic niche overlap with Mobula japanica
(Table S2 in the Supplement). Notarbartolo di Sciara
(1988) found a similar pattern in the Gulf of Califor-
nia, with M. munkiana stomach contents dominated
by mysids, as compared with the euphausiid prey
found in M. japanica and M. thurstoni. Our results
suggest that this trophic niche separation may be
consistent for the species throughout its range and
that M. munkiana may be feeding on an entirely dif-
ferent prey source and/or in an entirely different
region from other mobulids in Peru. This is supported
in part by differences in landing seasons between M.
munkiana and other mobulids in Peru (K. Forsberg
unpubl.), suggesting that M. mun ki ana may be pres-
ent along the coast during a different time of year
than other mobulids, perhaps due to differences in
foraging patterns.
δ13C and δ15N values can provide insights into
feeding locations and prey types due to predictable
changes in δ13C across marine habitats (France 1995)
and increases in δ15N at higher trophic levels (Owens
1987). In predatory elasmobranchs, changes in iso-
topic values are common as individuals grow and
either change habitats, in the case of an ontogenetic
shift, or access larger prey items at higher trophic
levels (Estrada et al. 2006, Hussey et al. 2011). Borrell
et al. (2011) found a positive relationship between
both δ13C and δ15N and body size in whale sharks
from the northwestern Indian Ocean, suggesting that
ontogenetic shifts may also occur in elasmobranch
filter-feeders. We found weak relationships between
disc width and isotope tracers across all mobulid spe-
cies. The observed decrease in δ13C and increase in
δ15N with increasing disc width could indicate a shift
to higher trophic level, offshore prey sources in older
and larger individuals within species. However, the
magnitude of the relationship we found (~0.2‰ per
meter disc width) is minimal in comparison to the
observed variability in isotope values within a given
size class, which often exceeded 2‰ for individuals
less than 10 cm apart in size. This suggests that mob-
ulids likely do not experience an ontogenetic shift in
feeding behavior and trophic level (δ15N) nor in habi-
tat (δ13C), although there may be some weak overall
effect of disc width on trophic dynamics or isotopic
fractionation. Notarbartolo di Sciara (1988) found dif-
ferences in stomach contents between juvenile and
adult M. thurstoni in the Gulf of California but sug-
gested this was more likely due to the season when
either size class was sampled as opposed to a true
ontogenetic dietary shift. The individuals sampled in
our study spanned a range of disc widths from juve-
niles to mature adults in all regions and species with
the exception of M. thurstoni in Peru. This suggests
that both juvenile and adult mobulids may occupy
the same habitats and target the same prey, as pro-
posed for M. birostris by Stewart et al. (2016a). This is
further supported by captures of both mature and
juvenile mobulids in the same nets in the Philippines
(J. M. Rambahiniarison pers. obs.).
The overlap we observed between species’ isotopic
niches is surprising given the diversity of vertical
habitat use and foraging behaviors recorded in mob-
ulids through observational studies and archival tag
deployments. M. japanica and closely related Mob-
ula mobular appear to spend the majority of their
time in near-surface habitats, shallower than 50 m
depth (Canese et al. 2011, Croll et al. 2012, Francis &
Jones 2017). M. birostris makes deeper foraging
dives and appears to spend substantial amounts of
time below 100 m (Stewart et al. 2016b). Mobula
tarapacana routinely undertakes deep dives below
800 m — in some cases over 1800 m for periods of 1
to 4 h, presumably to access bathypelagic scattering
layers of fishes (Thorrold et al. 2014). These vertical
movements are generally linked to observed or inf -
erred foraging behavior and represent a high degree
of vertical segregation that could in turn lead to dif-
fering trophic niches. However, all of these observa-
tions were made in different regions, and our results
suggest that the variability in observed behaviors
may be a result of regional differences in the location
of high-density zooplankton prey as opposed to con-
sistent differences in feeding behavior between spe-
cies. The few studies that examine multiple mobulid
species within a single region support this conclu-
sion. Notarbartolo di Sciara (1988) ob served similari-
ties in the euphausiid-dominated stomach contents of
adult M. japanica and M. thurs toni in the Gulf of Cal-
ifornia, Mexico, which Sampson et al. (2010) later
confirmed using isotope analysis. While the majority
of stomachs collected from M. tarapacana in Notar-
bartolo di Sciara (1988) were empty, there were
traces of euphausiids, among other crustaceans, and
1 stomach contained numerous remains of fishes.
Paired with our mixing model results, this suggests
that M. tarapacana is more piscivorous than the other
mobulids. However, the increased proportion of fish
145
Mar Ecol Prog Ser 580: 131–151, 2017
146
in the diet of M. tarapacana is minor in our results,
and it is unlikely that the extreme energy expendi-
ture of deep dives to the bathypelagic zone (Thorrold
et al. 2014) would justify such a modest dietary con-
tribution. We again posit that such behaviors are
likely region-specific, and that M. tarapacana from
the Philippines may not undertake these types of for-
aging excursions. Archival tag deployments on M.
tarapacana in the Philippines could provide insights
into vertical habitat use and differences between this
region and previous studies in the eastern Atlantic.
Mixing models
Stomach content collections from mobulids landed
in the Philippines allowed us to examine dietary over-
lap in greater detail. Diet contributions from the fish
and zooplankton aggregated source groups were sim-
ilar across species, with the majority of the diet coming
from the zooplankton group and lower diet contribu-
tions coming from pteropods and the fish group. Our
aggregated posterior distributions were robust to our
choice of priors, and median values of diet contribu-
tions from fish, zooplankton, and ptero pods were sim-
ilar regardless of our use of uninformative or semi-in-
formative priors. The most notable ex ception to this
was M. japanica. M. japanica muscle tissue had the
highest mean δ15N value of all mobulids in the Philip-
pines, placing them closest to the fish sources we
identified from M. birostris and M. tarapacana stom-
ach contents. In our model using an uninformative
prior, M. japanica appeared to exceed all other
species in their consumption of fish. Even when using
semi-informative priors with a lower expected contri-
bution of fish in the diet, our model results still indi-
cated that M. japanica consumes a greater proportion
of fish than M. thurstoni and M. birostris and a similar
proportion to M. tarapacana, despite M. tarapacana
and M. birostris being the only mobulids with
stomach content samples that contained fish. There
are several possible explanations for the discrepancy
between stomach contents and mixing model results
for M. japanica. The first possibility is that M. japanica
is more piscivorous during the rainy season, and the
landings and stomach content sampling during the
dry season did not reflect the overall diet. While
telemetry data suggest that M. japanica is restricted
mainly to near-surface waters (Croll et al. 2012), at
least some individuals make regular dives to depths of
200 to 300 m (Francis & Jones 2017). Myctophids and
Cubiceps sp. such as those found in the stomach con-
tents of M. birostris and M. tarapacana, migrate verti-
cally to depths that are easily within the recorded
diving depths of M. japanica, while Sardinella sp. typ-
ically inhabit near-surface waters. Smaller mobulids
have been recorded actively feeding on schools of fish
in the western Pacific (Heinrichs 2009), and it is possi-
ble that M. japanica engages in the same behavior in
the Philippines. An alternative ex planation is that our
trophic discrimination factors are incorrect, given
the variability in published discrimination factors for
elasmobranchs and the lack of experimentally derived
discrimination factors for mobulids. There is some evi-
dence to suggest that trophic discrimination factors
may vary with body size and compelling evidence to
suggest a relationship between trophic enrichment
and dietary protein content (Newsome et al. 2010).
However, given the similarity in both the stomach
content samples and body sizes of M. japanica and M.
thurstoni, we would ex pect a similar shift for both spe-
cies in our model. Given the differences between the
2 species’ isotopic signatures, we find it more likely
that there is a true dietary difference and a greater
contribution from an enriched δ15N source fish or
possibly an unsampled source —to M. japanica in the
rainy season. Diet contributions from our epipelagic
and mesopelagic source aggregations tended to be
more variable across species than fish and zooplank-
ton groups. However, we found epipelagic and meso-
pelagic groups to be less informative; first, because
they tended to be more sensitive to prior specifications
than the fish and zooplankton groups (Table 2) and,
second, because they provide fewer insights into
habitat use, as both the epipelagic prey and the verti-
cally migrating mesopelagic prey can be accessed by
mobulids at the surface at night.
The non-aggregated source contributions from our
model outputs provide additional information on
dietary sources and possible differences between
species. However, our mixing model results on spe-
cific prey sources should be interpreted with caution,
both because the number of sources (7) is far greater
than the number of tracers we used (2) and because
of the apparent sensitivity of the sources to prior
specifications (Fig. S1). Several sources stand out for
their consistency across priors and species. In all
cases, our mixing models estimated a higher propor-
tion of pteropods in the diets of mobulids than
expected based on their presence in stomach con-
tents. Their prevalence in the estimated diet propor-
tions is undoubtedly due to the geometry of the
source and consumer isotope values: without ptero-
pods, the consumers would not be contained within
the sources in bivariate isotope space, and therefore
pteropods must be included in a sufficient proportion
Stewart et al.: Trophic overlap in mobulid rays
for the mixing model to come to a mathematical solu-
tion. Our mixing model runs with carbonate correc-
tions for pteropod shells led to reductions in pteropod
diet contributions, illustrating this concept. It is possi-
ble that mobulids are feeding heavily on pteropods in
the rainy season, when they are not landed by fishers
and sampled, explaining the paucity of pteropods in
stomach content samples. Alternatively, there may
be an additional, unknown prey source that is iso-
topically similar to pteropods (either bulk or carbon-
ate corrected) but not included in our mixing models.
Burgess et al. (2016) hypo thesized that M. birostris in
Ecuador may be relying heavily on mesopelagic prey
sources, which are higher in δ13C and lower in δ15N
as compared with surface zooplankton, similar to the
pteropod source in the present study. It is possible
that our pteropod source is isotopically similar to
unsampled mesopelagic prey sources in the Philip-
pines and that all mobulid species rely heavily on
these prey items. However, the pteropod source
(with or without carbonate corrections) was isotopi-
cally distinct from both mesopelagic fishes and verti-
cally migrating euphausiids that were sampled from
stomach contents, suggesting that pteropods may in
fact be an important diet item for these species in the
Philippines.
Importantly, our aggregated posterior distributions
may artificially inflate the proportion of fish included
in the diet due to our inclusion of 3 fish sources. In the
discrete source posteriors, the proportion of each fish
in the diet of M. japanica and M. thurstoni abuts zero
under the semi-informative prior, suggesting that the
proportion of the diet coming from each fish species
is negligible very different from the posterior dis-
tributions for fish in M. tarapacana and M. birostris
(Fig. S1). However, when these posterior distribu-
tions are summed for the a posteriori aggregated fish
group, the resulting diet proportion is inflated.
Diet-switching
We analyzed multiple tissue types in mobulids
from the Philippines to directly examine the possible
diet-switching that our mixing model results sug-
gested. Liver is a metabolically more active tissue
than muscle and therefore has a faster turnover rate
(Tieszen et al. 1983, Hobson & Clark 1992). Previous
studies have used liver and muscle tissue to examine
dietary stability in elasmobranchs, with isotopic dif-
ferences between the 2 tissue types interpreted as
evidence of seasonal diet switching (MacNeil et al.
2005). The isotopic signatures of the liver samples we
collected in the Philippines were significantly differ-
ent from the paired muscle samples from the same
individuals. The greatest difference between tissues
was in δ13C for both M. birostris and M. tarapacana.
However, liver and muscle tissues are expected to
fractionate δ13C differently, with lower δ13C values in
liver samples even after lipid extraction (Pinnegar &
Polunin 1999, MacNeil et al. 2005). Controlled exper-
iments and studies of wild populations of teleost and
elasmobranch fishes suggest that this difference in
δ13C fractionation between liver and muscle tissue
may range from 0.5 to 2‰ (Pinnegar & Polunin 1999,
MacNeil et al. 2005, Hussey et al. 2010a), while the
shift in δ13C between our muscle and liver samples
was, on average, 2.8‰. It is therefore likely that
some portion of the difference in δ13C is due to vari-
able fractionation rates between the 2 tissues, while
some portion is due to true dietary shifts. This may
explain the discrepancies between the frequency of
some prey sources in stomach contents and the mix-
ing model outputs of diet proportions for M. birostris
and M. tarapacana and possibly the other mobulid
species for which liver samples were not collected
(Fig. S1). If liver and muscle tissue isotope values
were equal, it would suggest a stable diet throughout
the tissue integration period of the metabolically
slower tissue, while differences suggest diet-switching
(MacNeil et al. 2005). However, the C:N ratios of
lipid-extracted liver tissues re mained higher than
corresponding muscle tissues, which may indicate
that lipid extraction was not successful, confounding
the interpretation of these results.
We found small but significant differences in δ15N
between tissues in M. birostris and no significant dif-
ference in δ15N between tissues in M. tarapacana.
However, δ15N may also fractionate differently be -
tween tissue types, although with a smaller effect
than δ13C. Hussey et al. (2010a) found δ15N values to
be 0.37 to 0.89‰ higher in bulk muscle tissue than in
bulk livers of 3 captive sharks and 0.11 to 1.18‰ higher
in lipid-extracted muscle than in lipid- ex tracted liv-
ers. Similarly, MacNeil et al. (2005) found δ15N values
to be, on average, 0.03 and 0.57‰ higher in lipid-
extracted muscle tissue than in lipid-extracted livers
of 2 species of non-captive sharks that they con-
cluded had relatively stable diets. While our compar-
ison of δ15N in bulk muscle and bulk liver tissue
seems to suggest that M. birostris may switch its diet
seasonally in the Philippines, adding a correction fac-
tor of 0.5‰ to liver δ15N values to account for possible
fractionation differences between tissues inverts
these results, with δ15N becoming significantly differ-
ent between tissues in M. tarapacana and not signif-
147
Mar Ecol Prog Ser 580: 131–151, 2017
icant in M. birostris. These findings illustrate how
even small uncertainties in stable isotope ecology,
especially in fractionation factors, can have substan-
tial impacts on the interpretation of results, and they
highlight the frequently repeated need for species-
specific laboratory validation of parameters that are
used in isotope analysis.
CONCLUSIONS
Our findings contribute to a limited but growing
body of knowledge on the habitat use and ecology of
mobulid rays, which are highly vulnerable to ex -
ploitation due to their demographic characteristics
(Dulvy et al. 2014, Pardo et al. 2016). Both bycatch
and targeted fisheries appear to contribute to global
declines in mobulid abundance (Ward-Paige et al.
2013, White et al. 2015, Croll et al. 2016). Targeted
fisheries can be managed with legislation banning
the capture of mobulids, but bycatch remains a more
challenging and persistent threat due to the ubiquity
of mobulid bycatch in artisanal and commercial fish-
eries of all types (Croll et al. 2016). The apparent sim-
ilarity in diets and overlapping isotopic niches be -
tween mobulids in this study are consistent with the
spatial overlap in bycatch of the various mobulid spe-
cies (Croll et al. 2016). Identifying the spatial and
temporal patterns of mobulids’ primary zooplankton
prey (for example euphausiids in the Philippines)
could aid in predicting their occurrence and relative
vulnerability to bycatch-prone fisheries. Our results
further indicate that both adults and juveniles are
targeting similar prey and are thus overlapping with
and susceptible to the same fisheries pressures,
which has implications for the catchability of differ-
ent life stages and consequently the development of
age- or size-based fisheries management strategies.
Further research is necessary to corroborate many of
the patterns observed here over larger spatial and
temporal scales and will aid in our understanding of
habitat use by mobulids and the development of fish-
eries bycatch mitigation strategies.
Acknowledgements. Funding for fieldwork in the Philip-
pines was provided by the Ocean Park Conservation Foun-
dation Hong Kong and the PADI Foundation. Fieldwork in
the Philippines was approved by the local government of
Jagna and the Sangguniáng Bayan of Jagna (Bohol), and
samples were collected under a memorandum of agreement
with the Department of Agriculture, Bureau of Fisheries and
Aquatic Resources (DA-BFAR) Region 7 for the project
‘Visayas Marine Megafauna Research, Conservation and
Education project’. C.A.R. was supported by 2 private trusts.
J.A. and K.F. were supported by the Marine Conservation
Action Fund at the New England Aquarium, the Disney
Conservation Fund, the abc* Foundation, WildAid, and
Shark Savers. Funding for fieldwork in Sri Lanka was pro-
vided by the Save Our Seas Foundation and the Marine
Conservation and Action Fund at the New England Aquar-
ium. Funding for isotope analysis was provided by the Cen-
ter for the Advancement of Population Assessment Method-
ology. J.D.S. was supported by NOAA ONMS Nancy Foster
Scholarship NA15NOS4290068 and a fellowship from the
Robert & Patricia Switzer Foundation. We thank all the
volunteers of Large Marine Vertebrates Research Institute
Philippines for data collection and Maita Verdote and Dr. Jo
Marie Acebes (Balyena.org) for the support, advice, and
information shared during the mobulid fishery monitoring
project in Bohol. We are grateful for the support of the fish-
ers from Jagna; DA-BFAR Region 7; the Department of Envi-
ronment and Natural Resources Region 7; and the local
government unit of Jagna (Bohol). We thank Michelle Rob-
bins, Courtney Pinns, Ana Mendoza, Taylor Smith, and Kara
Reynolds for their assistance with sample preparation for
isotope analysis and Elena Duke and the Burton Lab at
Scripps Institution of Oceanography for their assistance with
genetic barcoding of unidentifiable stomach contents. Robert
Rubin provided insightful discussion on niche partitioning
and body size. We thank Wilmer Purizaca Ayala, Planeta
Océano volunteers, local fishermen, and community mem-
bers for their support in data collection in Peru. We thank
Anthony Richardson and the zooplankton lab at CSIRO,
Brisbane, for their help with sorting stomach contents prior
to isotope analysis.
LITERATURE CITED
Anderson RC, Adam MS, Goes JI (2011) From monsoons to
mantas: seasonal distribution of Manta alfredi in the
Maldives. Fish Oceanogr 20: 104−113
Armstrong AO, Armstrong AJ, Jaine FRA, Couturier LIE
and others (2016) Prey density threshold and tidal influ-
ence on reef manta ray foraging at an aggregation site on
the Great Barrier Reef. PLOS ONE 11: e0153393
Barrow LM, Bjorndal KA, Reich KJ (2008) Effects of preser-
vation method on stable carbon and nitrogen isotope val-
ues. Physiol Biochem Zool 81: 688−693
Ben-David M, Flynn W, Schell M (1997) Annual and sea-
sonal changes in diets of martens: evidence from stable
isotope analysis. Oecologia 111: 280−291
Borrell A, Aguilar A, Gazo M, Kumarran RP, Cardona L
(2011) Stable isotope profiles in whale shark (Rhincodon
typus) suggest segregation and dissimilarities in the diet
depending on sex and size. Environ Biol Fishes 92:
559−567
Braun CD, Skomal GB, Thorrold SR, Berumen ML (2014)
Diving behavior of the reef manta ray links coral reefs
with adjacent deep pelagic habitats. PLOS ONE 9:
e88170
Burgess KB, Bennett MB (2017) Effects of ethanol storage
and lipid and urea extraction on δ15N and δ13C isotope
ratios in a benthic elasmobranch, the bluespotted
maskray Neotrygon kuhlii. J Fish Biol 90: 417−423
Burgess KB, Couturier LIE, Marshall AD, Richardson AJ,
Weeks S, Bennett MB (2016) Manta birostris, predator of
the deep? Insight into the diet of the giant manta ray
through stable isotope analysis. R Soc Open Sci 3: 160717
148
Stewart et al.: Trophic overlap in mobulid rays
Canese S, Cardinali A, Romeo T, Giusti M, Salvati E, Angio-
lillo M, Greco S (2011) Diving behavior of the giant devil
ray in the Mediterranean Sea. Endang Species Res 14:
171−176
Caut S, Angulo E, Courchamp F (2008) Caution on isotopic
model use for analyses of consumer diet. Can J Zool 86:
438−445
Cherel Y, Hobson KA, Guinet C, Vanpe C (2007) Stable iso-
topes document seasonal changes in trophic niches and
winter foraging individual specialization in diving pred-
ators from the Southern Ocean. J Anim Ecol 76: 826−836
Couturier LIE, Marshall AD, Jaine FRA, Kashiwagi T and
others (2012) Biology, ecology and conservation of the
Mobulidae. J Fish Biol 80: 1075−1119
Couturier LIE, Rohner CA, Richardson AJ, Marshall AD and
others (2013) Stable isotope and signature fatty acid
analyses suggest reef manta rays feed on demersal zoo-
plankton. PLOS ONE 8: e77152
Croll DA, Marinovic B, Benson S, Chavez FP, Black N, Ter-
nullo R, Tershy BR (2005) From wind to whales: trophic
links in a coastal upwelling system. Mar Ecol Prog Ser
289: 117−130
Croll DA, Newton KM, Weng K, Galván-Magaña F, O’Sulli-
van J, Dewar H (2012) Movement and habitat use by the
spine-tail devil ray in the Eastern Pacific Ocean. Mar
Ecol Prog Ser 465: 193−200
Croll DA, Dewer H, Dulvy NK, Fernando D and others
(2016) Vulnerabilities and fisheries impacts: the uncer-
tain future of manta and devil rays. Aquat Conserv Mar
Freshw Ecosyst 26: 562−57
Dobush GR, Ankney CD, Krementz DG (1985) The effect of
apparatus, extraction time, and solvent type on lipid
extractions of snow geese. Can J Zool 63: 1917−1920
Dulvy NK, Pardo SA, Simpfendorfer CA, Carlson JK (2014)
Diagnosing the dangerous demography of manta rays
using life history theory. PeerJ 2: e400
Estrada JA, Rice AN, Natanson LJ, Skomal GB (2006) Use of
isotopic analysis of vertebrae in reconstructing ontoge-
netic feeding ecology in white sharks. Ecology 87: 829−834
Foley CJ, Bowen GJ, Nalepa TF, Sepúlveda MS, Höök TO,
Ramcharan C (2014) Stable isotope patterns of benthic
organisms from the Great Lakes region indicate variable
dietary overlap of Diporeia spp. and dreissenid mussels.
Can J Fish Aquat Sci 71: 1784−1795
France RL (1995) Carbon-13 enrichment in benthic com-
pared to planktonic algae: foodweb implications. Mar
Ecol Prog Ser 124: 307−312
Francis MP, Jones EG (2017) Movement, depth distribution
and survival of spinetail devilrays (Mobula japanica)
tagged and released from purse-seine catches in New
Zealand. Aquat Conserv Mar Freshw Ecosyst 27: 219−236
Gannes LZ, O’Brien DM, Martinez del Rio C (1997) Stable
isotopes in animal ecology: assumptions, caveats, and a call
for more laboratory experiments. Ecology 78: 1271−1276
Gavrilchuk K, Lesage V, Ramp C, Sears R, Bérubé M,
Bearhop S, Beauplet G (2014) Trophic niche partitioning
among sympatric baleen whale species following the col-
lapse of groundfish stocks in the Northwest Atlantic. Mar
Ecol Prog Ser 497: 285−301
Gelman A, Rubin DB (1992) Inference from iterative simula-
tion using multiple sequences. Stat Sci 7: 457−472
Hazen EL, Johnston DW (2010) Meridional patterns in the
deep scattering layers and top predator distribution in
the central equatorial Pacific. Fish Oceanogr 19: 427−433
Heinrichs S (2009) Hunting mobulas of Misool. www.blue
sphere media. com/2009/10/hunting-mobulas-of-misool/
(accessed on 23 January 2017)
Heithaus MR, Vaudo JJ, Kreicker S, Layman CA and others
(2013) Apparent resource partitioning and trophic struc-
ture of large-bodied marine predators in a relatively pris-
tine seagrass ecosystem. Mar Ecol Prog Ser 481: 225−237
Hobson KA, Clark RG (1992) Assessing avian diets using
stable isotopes I: turnover of 13C in tissues. Condor 94:
181−188
Hobson KA, Gloutney ML, Gibbs HL (1997) Preservation of
blood and tissue samples for stable-carbon and stable-
nitrogen isotope analysis. Can J Zool 75: 1720−1723
Hussey NE, Brush J, McCarthy ID, Fisk AT (2010a) δ15N and
δ13C diet−tissue discrimination factors for large sharks
under semi-controlled conditions. Comp Biochem Physiol
A Mol Integr Physiol 155: 445−453
Hussey NE, MacNeil MA, Fisk AT (2010b) The requirement
for accurate diet−tissue discrimination factors for inter-
preting stable isotopes in sharks. Hydrobiologia 654: 1−5
Hussey NE, Dudley SFJ, McCarthy ID, Cliff G, Fisk AT
(2011) Stable isotope profiles of large marine predators:
viable indicators of trophic position, diet, and movement
in sharks? Can J Fish Aquat Sci 68: 2029−2045
Hutchinson GE (1957) Concluding remarks. Cold Spring
Harb Symp Quant Biol 22:417–427
Jackson AL, Inger R, Parnell AC, Bearhop S (2011) Compar-
ing isotopic niche widths among and within communities:
SIBER Stable Isotope Bayesian Ellipses in R. J Anim
Ecol 80: 595−602
Jackson MC, Woodford DJ, Bellingan TA, Weyl OLF and
others (2016) Trophic overlap between fish and riparian
spiders: potential impacts of an invasive fish on terres-
trial consumers. Ecol Evol 6: 1745−1752
Jaine FRA, Rohner CA, Weeks SJ, Couturier LIE, Bennett MB,
Townsend KA, Richardson AJ (2014) Movements and
habitat use of reef manta rays off eastern Australia: off-
shore excursions, deep diving and eddy affinity revealed
by satellite telemetry. Mar Ecol Prog Ser 510: 73−86
Jaine FRA, Couturier LIE, Weeks SJ, Townsend KA, Bennett
MB, Fiora K, Richardson AJ (2012) When giants turn up:
sighting trends, environmental influences and habitat
use of the manta ray Manta alfredi at a coral reef. PLOS
ONE 7: e46170
Kaehler S, Pakhomov EA (2001) Effects of storage and
preservation on the d13C and d15N signatures of selected
marine organisms. Mar Ecol Prog Ser 219: 299−304
Kashiwagi T, Marshall AD, Bennett MB, Ovenden JR (2011)
Habitat segregation and mosaic sympatry of the two spe-
cies of manta ray in the Indian and Pacific Oceans:
Manta alfredi and M. birostris. Mar Biodivers Rec 4: e53
Kashiwagi T, Marshall AD, Bennett MB, Ovenden JR (2012)
The genetic signature of recent speciation in manta rays
(Manta alfredi and M. birostris). Mol Phylogenet Evol 64:
212−218
Kelly B, Dempson JB, Power M (2006) The effects of preser-
vation on fish tissue stable isotope signatures. J Fish Biol
69: 1595−1611
Kim SL, Koch PL (2012) Methods to collect, preserve, and
prepare elasmobranch tissues for stable isotope analysis.
Environ Biol Fishes 95: 53−63
Kim SL, Casper DR, Galván-Magaña F, Ochoa-Díaz R,
Hernández-Aguilar SB, Koch PL (2012a) Carbon and
nitrogen discrimination factors for elasmobranch soft tis-
sues based on a long-term controlled feeding study. Env-
iron Biol Fishes 95: 37−52
149
Mar Ecol Prog Ser 580: 131–151, 2017
Kim SL, del Rio CM, Casper D, Koch PL (2012b) Isotopic
incorporation rates for shark tissues from a long-term
captive feeding study. J Exp Biol 215: 2495−2500
Lesage V, Morin Y, Rioux E, Pomerleau C, Ferguson SH,
Pelletier E (2010) Stable isotopes and trace elements as
indicators of diet and habitat use in cetaceans: predicting
errors related to preservation, lipid extraction, and lipid
normalization. Mar Ecol Prog Ser 419: 249−265
Logan JM, Lutcavage ME (2010) Stable isotope dynamics in
elasmobranch fishes. Hydrobiologia 644: 231−244
Logan JM, Jardine TD, Miller TJ, Bunn SE, Cunjak RA, Lut-
cavage ME (2008) Lipid corrections in carbon and nitro-
gen stable isotope analyses: comparison of chemical
extraction and modelling methods. J Anim Ecol 77:
838−846
MacNeil MA, Skomal GB, Fisk AT (2005) Stable isotopes
from multiple tissues reveal diet switching in sharks. Mar
Ecol Prog Ser 302: 199−206
MacNeil MA, Drouillard KG, Fisk AT (2006) Variable uptake
and elimination of stable nitrogen isotopes between tis-
sues in fish. Can J Fish Aquat Sci 63: 345−353
Madden T (2002) The BLAST sequence analysis tool. In:
McEntyre J, Ostell J (eds) The NCBI handbook. Nat -
ional Center for Biotechnology Information, Bethesda,
MD
Malpica-Cruz L, Herzka SZ, Sosa-Nishizaki O, Lazo JP,
Trudel M (2012) Tissue-specific isotope trophic discrimi-
nation factors and turnover rates in a marine elasmo-
branch: empirical and modeling results. Can J Fish
Aquat Sci 69: 551−564
Mateo MA, Serrano O, Serrano L, Michener RH (2008)
Effects of sample preparation on stable isotope ratios of
carbon and nitrogen in marine invertebrates: implica-
tions for food web studies using stable isotopes. Oecolo-
gia 157: 105−115
May RM, MacArthur RH (1972) Niche overlap as a function
of environmental variability. Proc Natl Acad Sci USA 69:
1109−1113
Mendelssohn R (2015) xtractomatic: extracts environmental
data from ERD’s ERDDAP web service. http://coastwatch.
pfel.noaa.gov/xtracto/
Moore JW, Semmens BX (2008) Incorporating uncertainty
and prior information into stable isotope mixing models.
Ecol Lett 11: 470−480
Newsome SD, Clementz MT, Koch PL (2010) Using stable
isotope biogeochemistry to study marine mammal ecol-
ogy. Mar Mamm Sci 26: 509−572
Notarbartolo di Sciara G (1988) Natural history of the rays of
the genus Mobula in the Gulf of California. Fish Bull 86:
45−66
Owens NJP (1987) Natural variation in 15N in the marine
environment. Adv Mar Biol 24: 389−451
Pardo SA, Burgess KB, Teixeira D, Bennett MB (2015) Local-
scale resource partitioning by stingrays on an intertidal
flat. Mar Ecol Prog Ser 533: 205−218
Pardo SA, Kindsvater HK, Cuevas-Zimbrón E, Sosa-Nishizaki
O, Pérez-Jiménez JC, Dulvy NK (2016) Growth, pro -
ductivity, and extinction risk of a data-sparse devil ray.
Sci Rep 6: 33745
Parng E, Crumpacker A, Kurle CM (2014) Variation in the
stable carbon and nitrogen isotope discrimination factors
from diet to fur in four felid species held on different
diets. J Mammal 95: 151−159
Pianka ER (1974) Niche overlap and diffuse competition.
Proc Natl Acad Sci USA 71: 2141−2145
Pianka ER (1981). Competition and niche theory. In: May
RM (ed) Theoretical ecology: principles and applica-
tions, 2nd edn. Sinauer Associates, Sunderland, MA,
p 167−196
Pinnegar JK, Polunin NVC (1999) Differential fractionation
of δ13C and δ15N among fish tissues: implications for the
study of trophic interactions. Funct Ecol 13: 225−231
Plass-Johnson JG, McQuaid CD, Hill JM (2013) Stable iso-
tope analysis indicates a lack of inter- and intra-specific
dietary redundancy among ecologically important coral
reef fishes. Coral Reefs 32: 429−440
Poortvliet M, Olsen JL, Croll DA, Bernardi G and others
(2015) A dated molecular phylogeny of manta and devil
rays (Mobulidae) based on mitogenome and nuclear
sequences. Mol Phylogenet Evol 83: 72−85
Porter JH, Dueser RD (1982) Niche overlap and competition
in an insular small mammal fauna: a test of the niche
overlap hypothesis. Oikos 39: 228−236
Post DM, Layman CA, Arrington DA, Takimoto G, Quat-
trochi J, Montaña CG (2007) Getting to the fat of the mat-
ter: models, methods and assumptions for dealing with
lipids in stable isotope analyses. Oecologia 152: 179−189
R Core Team (2016) R: a language and environment for sta-
tistical computing. R Foundation for Statistical Comput-
ing, Vienna
Rohner CA, Couturier LIE, Richardson AJ, Pierce SJ, Preb-
ble CEM, Gibbons MJ, Nichols PD (2013) Diet of whale
sharks Rhincodon typus inferred from stomach content
and signature fatty acid analyses. Mar Ecol Prog Ser 493:
219−235
Rohner CA, Armstrong AJ, Pierce SJ, Prebble CEM and
others (2015) Whale sharks target dense prey patches
of sergestid shrimp off Tanzania. J Plankton Res 37:
352−362
Rohner CA, Burgess KB, Rambahiniarison JM, Stewart JD,
Ponzo A, Richardson AJ (2017) Mobulid rays feed on
euphausiids in the Bohol Sea. R Soc Open Sci 4: 161060
Sampson L, Galván−Magaña F, De Silva-Dávila R, Aguiniga-
Garcia S, O’Sullivan JB (2010) Diet and trophic position
of the devil rays Mobula thurstoni and Mobula japanica
as inferred from stable isotope analysis. J Mar Biol Assoc
UK 90: 969−976
Santora JA, Reiss CS, Loeb VJ, Veit RR (2010) Spatial asso-
ciation between hotspots of baleen whales and demo-
graphic patterns of Antarctic krill Euphausia superba
suggests size-dependent predation. Mar Ecol Prog Ser
405: 255−269
Sarakinos HC, Johnson ML, Vander Zanden MJ (2002) A
synthesis of tissue-preservation effects on carbon and
nitrogen stable isotope signatures. Can J Zool 80:
381−387
Scharf FS, Juanes F, Rountree RA (2000) Predator size-
prey size relationships of marine fish predators: inter-
specific variation and effects of ontogeny and body size
on trophic-niche breadth. Mar Ecol Prog Ser 208:
229−248
Semmens BX, Ward EJ, Moore JW, Darimont CT (2009)
Quantifying inter-and intra-population niche variability
using hierarchical Bayesian stable isotope mixing mod-
els. PLOS ONE 4: e61871−9
Stallings CD, Nelson JA, Rozar KL, Adams CS, Wall KR,
Switzer TS, Winner BL, Hollander DJ (2015) Effects of
preservation methods of muscle tissue from upper-
trophic level reef fishes on stable isotope values (δ13C
and δ15N). PeerJ 3: e874
150
Stewart et al.: Trophic overlap in mobulid rays 151
Stevens GMW (2014) Field guide to the identification of
mobulid rays (Mobulidae): Indo-West Pacific. Manta
Trust, Dorchester
Stewart JD, Beale CS, Fernando D, Sianipar AB, Burton RS,
Semmens BX, Aburto-Oropeza O (2016a) Spatial ecol-
ogy and conservation of Manta birostris in the Indo-
Pacific. Biol Conserv 200: 178−183
Stewart JD, Hoyos-Padilla EM, Kumli KR, Rubin RD (2016b)
Deep-water feeding and behavioral plasticity in Manta
birostris revealed by archival tags and submersible
observations. Zoology 119: 406−413
Stock BC, Semmens BX (2016) Unifying error structures in
commonly used biotracer mixing models. Ecology 97:
2562−2569
Thorrold SR, Afonso P, Fontes J, Braun CD, Santos RS, Sko-
mal GB, Berumen ML (2014) Extreme diving behaviour
in devil rays links surface waters and the deep ocean.
Nat Commun 5: 4274
Tieszen LL, Boutton TW, Tesdahl KG, Slade NA (1983) Frac-
tionation and turnover of stable carbon isotopes in ani-
mal tissues: implications for δ13C analysis of diet. Oecolo-
gia 57: 32−37
Vizza C, Sanderson BL, Burrows DG, Coe HJ (2013) The
effects of ethanol preservation on fish fin stable isotopes:
Does variation in C: N ratio and body size matter? Trans
Am Fish Soc 142: 1469−1476
Walter RP, Kessel ST, Alhasan N, Fisk AT and others (2014)
First record of living Manta alfredi × Manta birostris
hybrid. Mar Biodivers 44: 1−2
Ward-Paige CA, Davis B, Worm B (2013) Global population
trends and human use patterns of Manta and Mobula
rays. PLOS ONE 8: e74835
White ER, Myers MC, Flemming JM, Baum JK (2015) Shift-
ing elasmobranch community assemblage at Cocos
Island — an isolated marine protected area. Conserv Biol
29: 1186−1197
Young JW, Lansdell MJ, Campbell RA, Cooper SP, Juanes F,
Guest MA (2010) Feeding ecology and niche segregation
in oceanic top predators off eastern Australia. Mar Biol
157: 2347−2368
Editorial responsibility: Stephen Wing,
Dunedin, New Zealand
Submitted: March 1, 2017; Accepted: August 11, 2017
Proofs received from author(s): September 25, 2017
... Les raies manta sont des filtreurs qui occupent principalement les habitats côtiers des eaux intertropicales du monde entier se nourrissant aussi bien dans les upwellings productifs des zones équatoriales que dans les milieux oligotrophes des zones tropicales Stewart et al. 2017). Les raies manta de récif sont observées en agrégation de plusieurs individus pouvant aller jusqu'à des centaines dans certaines régions comme les Maldives, le Mexique, ou au Mozambique, par exemple (Law 2010 ;Kitchen-Wheeler et al. 2010). ...
Thesis
Espèce emblématique et néanmoins menacée qui peuple les mers des régions tropicales et subtropicales du monde entier, la raie manta de récif (Mobula alfredi) est bien présente en Nouvelle-Calédonie. La population de l’archipel n’a cependant encore jamais été étudiée et la compréhension de sa biologie, son écologie, sa dynamique des populations et ses mouvements est encore limitée à l’échelle globale. L’acquisition de connaissances de références pourrait jouer un rôle essentiel pour la conservation de l’espèce. Cette thèse tente de décrire la population, sa structure et l’écologie spatiale de raies manta de Nouvelle-Calédonie en utilisant des approches diverses combinant la science participative à la télémétrie satellite et la génomique. Un suivi de cinq ans permettant la collecte de 1741 photo-identifications de raies manta a permis de connaitre les caractéristiques et la distribution de la population, d’estimer son abondance et d’obtenir un premier aperçu de son utilisation de l’espace et des potentielles sources de blessures. L’utilisation de 21 balises satellites a permis d’obtenir des données de déplacements et de comportements de plongées plus. Finalement, le séquençage du génome de 92 échantillons dont 73 provenant de quatre sites en Nouvelle-Calédonie et 19 de deux sites sur la côte Est de l’Australie a permis de révéler l’existence d’une structure génétique à l’échelle régionale et locale. Les résultats présentés dans cette thèse apportent les premières données sur la population de raies manta de récifs en Nouvelle-Calédonie et proposent les mesures et précautions qui devraient être prises pour évaluer et mitiger les possibles perturbations présentes et futures.
... Community-level isotope metrics [36] were generated using Stable Isotope Bayesian Ellipses in R (SIBER) package [44], following Pereira et al. [13], to evaluate trophic diversity, trophic redundancy, and niche breadth. Similarities in trophic structure among substrates and seepage activity were determined by computing pairwise ellipse overlaps based on Bayesian posterior estimates using 2 chains of 20,000 iterations with a burn-in of 1,000 and thinning of 10 [45]. ...
Article
Full-text available
Continental margins host methane seeps, animal falls and wood falls, with chemosynthetic communities that may share or exchange species. The goal of this study was to examine the existence and nature of linkages among chemosynthesis-based ecosystems by deploying organic fall mimics (bone and wood) alongside defaunated carbonate rocks within high and lesser levels of seepage activity for 7.4 years. We compared community composition, density, and trophic structure of invertebrates on these hard substrates at active methane seepage and transition (less seepage) sites at Mound 12 at~1,000 m depth, a methane seep off the Pacific coast of Costa Rica. At transition sites, the community composition on wood and bone was characteristic of natural wood-and whale-fall community composition, which rely on decay of the organic substrates. However, at active sites, seepage activity modified the relationship between fauna and substrate, seepage activity had a stronger effect in defining and homogenizing these communities and they depend less on organic decay. In contrast to community structure, macrofaunal trophic niche overlap between sub-strates, based on standard ellipse areas, was greater at transition sites than at active sites, except between rock and wood. Our observations suggest that whale-and wood-fall sub-strates can function as stepping stones for seep fauna even at later successional stages, providing hard substrate for attachment and chemosynthetic food.
... Mobulids can feed in different strata of the water column, such as near-surface, epipelagic, mesopelagic, pelagic, and demersal sources (Couturier et al. 2013;Stewart et al. 2017;Burgess et al. 2018;Peel et al. 2019), and they can perform vertical migrations (Thorrold et al. 2014) to feed on high-density zooplankton associated with the thermocline (Stewart et al. 2016). There are only a few studies to date analyzing dietary data of mobulids, in which it was possible to observe that mobulids generally feed on copepods, euphausiids, small fishes, fish larvae and eggs, decapods, mysids, and chaetognaths (Couturier et al. 2012;Rohner et al. 2017b;Stewart et al. 2018;Coasaca-Céspedes et al. 2018). ...
Article
Although mobulids (Mobulidae) are threatened with extinction, basic information about their biology and ecology is still lacking. Therefore, we analyzed impacts of fishing, morphology, coloration, and stomach content of mobulid carcasses stranded on the coast of Ilha Comprida, southeastern Brazil, found in protected areas of sustainable use. The carcasses were of a juvenile female manta ray (believed to be Mobula cf. birostris), two adult females M. birostris (one with a partially aborted fetus), and two adult M. hypostoma (one female with an aborted fetus, and one male). The stomach content of a M. birostris consisted almost exclusively of a newly reported prey item, the sergestid Acetes americanus, highlighting the variation of prey items in the species’ diet. This is the first record of a pregnant M. birostris in the South Atlantic Ocean and the first account of A. americanus as a mobulid prey item. The reported mobulids have been impacted by fishing activities, even within sustainable use protected areas, highlighting the importance of improving surveillance in the area and the communication between the fishing communities, scientists, and managers for designing better conservation plans for the species. Future research and conservation efforts are urgently needed to better understand the spatial and temporal distribution of mobulids, as well of its prey A. americanus, to further confirm that mobulids use this site as a feeding and nursery area and to reduce the anthropogenic impacts on these endangered populations.
... The species occurs in neritic and oceanic waters, including around oceanic islands, pinnacles and seamounts (Gadig et al. 2003;Mendonça et al. 2012;Marshall et al. 2019), usually related to highly productive upwelling areas . As a filter-feeder species, M. thurstoni forage on zooplankton patches, including small shrimp, crabs and fish (Notarbartolo-di-Sciara 1987, 1988Stewart et al. 2017). Although able to adapt and shift diet according to the food availability (Couturier et al. 2012), the bentfin devil rays have specialized feeding habits, which comprise mostly euphausiids species (Gadig et al. 2003;Rohner et al. 2017;Mukharror et al. 2018). ...
Article
The bentfin devil ray (Mobula thurstoni) is a migratory elasmobranch species with a wide distribution range. Despite the recent increase in mobulid research, critical habitats and home ranges are still being identified for these threatened species. In the present study, photo and video records opportunistically gathered by SCUBA diving effort were used to identify individuals and habitat usage by M. thurstoni in a marine protected area in the western equatorial Atlantic Ocean, Fernando de Noronha Archipelago (FNA). The bentfin devil ray was identified in five distinct records, in different years and sites around the archipelago. All the males showed developed claspers, suggesting mature individuals using the area. The UNESCO heritage site of FNA is considered an area of high biological importance, containing essential habitats for several species of fish, turtles and marine mammals. However, habitat usage by devil ray species is poorly reported in the region; therefore, the present study presents the first report of M. thurstoni at FNA, which adds the fifth mobulid ray species recorded in the region. Additionally, these results correspond to the second record of living specimens of M. thurstoni in Brazilian jurisdictional waters, highlighting new information on the species’ distribution and the home range of mobulids in Brazil.
... In the Eastern Pacific, the occurrence of devil rays is connected to higher oceanic productivity, such as the upwelling systems along the coast of Baja California Peninsula, Galapagos Islands, and Peru, and the oceanic upwellings of the Costa Rican Dome (Lezama-Ochoa et al., 2019). Different dietary studies from stomach contents and stable isotope analysis suggest that most mobulid species are predominantly planktivorous and, except for M. munkiana, overlap in their trophic preference of euphausiids as principal prey (Notarbartolo di Sciara, 1988;Sampson et al., 2010;Stewart et al., 2017). ...
Article
Full-text available
Introduction: Identifying critical habitats for vulnerable elasmobranch species is crucial for effective conservation measures. The Munk’s devil ray (Mobula munkiana) is endemic to the Eastern Pacific, but yet little is known about its biology, ecology, and habitat use. As filter feeders, it is assumed that this species concentrates at high-productive upwelling regions, such as the Costa Rican Dome. Like many elasmobranchs, its populations are highly depleted and require urgent information to inform better conservation measures. Objective: The study was conducted to gain information on a unique behavior observed in juvenile M. munkiana, so further information can be provided on early life stages of this vulnerable species. Methods: From June to September 2017 and in August 2018, the feeding behavior of juvenile Mobula munkiana was observed in two shallow bays located at Punta Descartes, North Pacific Costa Rica. Individuals were captured using a non-lethal method to obtain data on size, weight, and sex distribution. Plankton samples (n = 100) were taken at both bays throughout the months to infer diet composition. Results: Munk’s devil rays showed a repetitive swimming movement parallel to the beach, feeding exclusively in the shallow breaking zone of the low tide waves at depth <50cm. A total of 12 M. munkiana (11 live and one found dead) indicated a juvenile feeding aggregation ranging from 490 – 610mm in disk width and 1400 – 2300gr in weight. The sex ratio (males to females) was 3:1. Zooplankton of the order Mysidacae was found in the highest abundance in the breaking zone. Conclusions: The specific behavior and seasonal occurrence of juvenile Munk’s devil rays in this area seem to be driven by prey abundance. More research is needed to conclude the presence of reproductive adults at deeper depths and the year-round habitat use of Punta Descartes. The area is threatened by unsustainable development and requires realistic management strategies to guarantee the survival of vulnerable species and their critical habitats.
... In the Eastern Pacific, the occurrence of devil rays is connected to higher oceanic productivity, such as the upwelling systems along the coast of Baja California Peninsula, Galapagos Islands, and Peru, and the oceanic upwellings of the Costa Rican Dome (Lezama-Ochoa et al., 2019). Different dietary studies from stomach contents and stable isotope analysis suggest that most mobulid species are predominantly planktivorous and, except for M. munkiana, overlap in their trophic preference of euphausiids as principal prey (Notarbartolo di Sciara, 1988;Sampson et al., 2010;Stewart et al., 2017). ...
Book
Full-text available
The North Pacific and the South Pacific of Nicaragua is a region of great biological, geological, economic and social wealth. There the dry forest mixes with the rain forest, the sea with the islands and nature with the people. It is a region influenced by the Papagayo upwelling, the emergence of cold marine waters during the dry season, which generates an abundance of life in the sea. As a sample of this marine wealth, in this Special Issue, more than half of the contributions are dedicated to advances in the knowledge of the marine biodiversity of the region. Contributions from the social sciences, geology and physics of the region are also included. Within these areas, the publications provide information on maritime border management, archaeology and sustainable tourism, coastal geology, projected climate changes, as well as various oceanographic aspects of the area. We hope that this Special Issue on the North Pacific of Costa Rica and the South Pacific of Nicaragua will promote more research in the region and help inform decision-making processes and educational activities. We thank the authors for their manuscripts, as well as the more than sixty reviewers who with their comments and suggestions helped to improve the quality of the manuscripts.
... In the Eastern Pacific, the occurrence of devil rays is connected to higher oceanic productivity, such as the upwelling systems along the coast of Baja California Peninsula, Galapagos Islands, and Peru, and the oceanic upwellings of the Costa Rican Dome (Lezama-Ochoa et al., 2019). Different dietary studies from stomach contents and stable isotope analysis suggest that most mobulid species are predominantly planktivorous and, except for M. munkiana, overlap in their trophic preference of euphausiids as principal prey (Notarbartolo di Sciara, 1988;Sampson et al., 2010;Stewart et al., 2017). ...
Article
Full-text available
Introduction: Identifying critical habitats for vulnerable elasmobranch species is crucial for effective conservation measures. The Munk's devil ray (Mobula munkiana) is endemic to the Eastern Pacific, but yet little is known about its biology, ecology, and habitat use. As filter feeders, it is assumed that this species concentrates at high-productive upwelling regions, such as the Costa Rican Dome. Like many elasmobranchs, its populations are highly depleted and require urgent information to inform better conservation measures. Objective: The study was conducted to gain information on a unique behavior observed in juvenile M. munki-ana, so further information can be provided on early life stages of this vulnerable species. Methods: From June to September 2017 and in August 2018, the feeding behavior of juvenile Mobula munki-ana was observed in two shallow bays located at Punta Descartes, North Pacific Costa Rica. Individuals were captured using a non-lethal method to obtain data on size, weight, and sex distribution. Plankton samples (n = 100) were taken at both bays throughout the months to infer diet composition. Results: Munk's devil rays showed a repetitive swimming movement parallel to the beach, feeding exclusively in the shallow breaking zone of the low tide waves at depth <50cm. A total of 12 M. munkiana (11 live and one found dead) indicated a juvenile feeding aggregation ranging from 490-610mm in disk width and 1400-2300gr in weight. The sex ratio (males to females) was 3:1. Zooplankton of the order Mysidacae was found in the highest abundance in the breaking zone. Conclusions: The specific behavior and seasonal occurrence of juvenile Munk's devil rays in this area seem to be driven by prey abundance. More research is needed to conclude the presence of reproductive adults at deeper depths and the year-round habitat use of Punta Descartes. The area is threatened by unsustainable development and requires realistic management strategies to guarantee the survival of vulnerable species and their critical habitats.
Article
Oceanic manta rays Mobula birostris are distributed in tropical and subtropical oceans with aggregation sites typically at oceanic islands and seamounts. However, in some regions the species is found in coastal habitats close to highly productive areas. Recently, an important aggregation site for oceanic manta rays was identified along the southern coast of Bahía de Banderas, Mexico. We conducted weekly monitoring trips to the aggregation hotspots over a period of 4½ yr, recording manta sightings and collecting zooplankton samples, to investigate relationships between sightings and prey availability. We evaluated relationships between manta sightings and physical, biological, and environmental variables, finding a seasonal signal of manta occurrence with a peak in sightings around April, although there were substantial deviations from this trend in some years. Our results show effects of the El Niño Southern Oscillation on sighting rates, with more mantas during La Niña phases. We found significant relationships between manta sightings and sea surface temperature, moon phase, and tidal range. Manta sightings were negatively related to fish egg densities, positively related to copepod and cladoceran densities, and showed no relationship to euphausiid density. Mantas may be feeding on mesopelagic prey in the submarine canyon adjacent to the southern coast and basking in shallow waters through the day during thermal recovery periods. Identifying the seasonal patterns of occurrence and the environmental drivers of sightings will support the development of strategies to mitigate anthropogenic threats, which are common in this coastal population. KEY WORDS: Co-occurrence · Filter-feeder · Elasmobranch · Seasonality · GAM · Zooplankton
Article
Full-text available
In 2018, the giant manta ray was listed as threatened under the U.S. Endangered Species Act. We integrated decades of sightings and survey effort data from multiple sources in a comprehensive species distribution modeling (SDM) framework to evaluate the distribution of giant manta rays off the eastern United States, including the Gulf of Mexico. Manta rays were most commonly detected at productive nearshore and shelf-edge upwelling zones at surface thermal frontal boundaries within a temperature range of approximately 20–30 °C. SDMs predicted highest nearshore occurrence off northeastern Florida during April, with the distribution extending northward along the shelf-edge as temperatures warm, leading to higher occurrences north of Cape Hatteras, North Carolina from June to October, and then south of Savannah, Georgia from November to March as temperatures cool. In the Gulf of Mexico, the highest nearshore occurrence was predicted around the Mississippi River delta from April to June and again from October to November. SDM predictions will allow resource managers to more effectively protect manta rays from fisheries bycatch, boat strikes, oil and gas activities, contaminants and pollutants, and other threats.
Article
Full-text available
Australian cownose rays (Rhinoptera neglecta) and whitespotted eagle rays (Aetobatus ocellatus) are large myliobatiform rays that co‐occur off temperate eastern Australia. Here, we performed stable‐isotope analyses (δ13C, δ15N and δ34S) on fin clips of both species to gain insights into their trophic interactions and isotopic niches and assess the effect of preservation (ethanol‐stored versus frozen) on isotopic values of fin clip tissue of R. neglecta. Linear mixed models identified species as the main factor contributing to variation among δ15N and δ34S values, and disc width for δ13C. Bayesian ecological niche modelling indicated a 57.4 to 74.5% overlap of trophic niches, with the niche of R. neglecta being smaller and more constrained. Because values of δ13C were similar between species, variation in isotopic niches were due to differences in δ15N and δ34S values. Linear mixed models failed to detect differences in isotopic values of ethanol‐stored and frozen fin tissue of R. neglecta. This study provides the first examination of the trophic ecology of R. neglecta and the comparison of isotopic niche with A. ocellatus, which will facilitate future research into the trophic interactions of these species and aid better resource management. This article is protected by copyright. All rights reserved.
Article
Full-text available
Mobulid rays have a conservative life history and are caught in direct fisheries and as by-catch. Their subsequent vulnerability to overexploitation has recently been recognized, but fisheries management can be ineffective if it ignores habitat and prey preferences and other trophic interactions of the target species. Here, we assessed the feeding ecology of four mobulids (Manta birostris, Mobula tarapacana, M. japanica, M. thurstoni) in the Bohol Sea, Philippines, using stomach contents analysis of fisheries specimens landed between November and May in 2013–2015. We show that the mobulids feed heavily on euphausiid krill while they are in the area for approximately six months of the year. We found almost no trophic separation among the mobulid species, with Euphausia diomedeae as the major prey item for all species, recorded in 81 of 89 total stomachs (91%). Mobula japanica and M. thurstoni almost exclusively had this krill in their stomach, while M. tarapacana had a squid and fish, and Ma. birostris had myctophid fishes and copepods in their stomachs in addition to E. diomedeae. This krill was larger than prey for other planktivorous elasmobranchs elsewhere and contributed a mean of 61 364 kcal per stomach (±105 032 kcal s.e., range = 0–631 167 kcal). Our results show that vertically migrating mesopelagic species can be an important food resource for large filter feeders living in tropical seas with oligotrophic surface waters. Given the conservative life history of mobulid rays, the identification of common foraging grounds that overlap with fishing activity could be used to inform future fishing effort.
Article
Full-text available
Devil rays (Mobula spp.) face intensifying fishing pressure to meet the ongoing international demand for gill plates. The paucity of information on growth, mortality, and fishing effort for devil rays make quantifying population growth rates and extinction risk challenging. Furthermore, unlike manta rays (Manta spp.), devil rays have not been listed on CITES. Here, we use a published size-at-age dataset for the Spinetail Devil Ray (Mobula japanica), to estimate somatic growth rates, age at maturity, maximum age, and natural and fishing mortality. We then estimate a plausible distribution of the maximum intrinsic population growth rate (rmax) and compare it to 95 other chondrichthyans. We find evidence that larger devil ray species have low somatic growth rate, low annual reproductive output, and low maximum population growth rates, suggesting they have low productivity. Fishing rates of a small-scale artisanal Mexican fishery were comparable to our estimate of rmax, and therefore probably unsustainable. Devil ray rmax is very similar to that of manta rays, indicating devil rays can potentially be driven to local extinction at low levels of fishing mortality and that a similar degree of protection for both groups is warranted.
Article
Full-text available
Mixing models are statistical tools that use biotracers to probabilistically estimate the contribution of multiple sources to a mixture. These biotracers may include contaminants, fatty acids, or stable isotopes, the latter of which are widely used in trophic ecology to estimate the mixed diet of consumers. Bayesian implementations of mixing models using stable isotopes (e.g. MixSIR, SIAR) are regularly used by ecologists for this purpose, but basic questions remain about when each is most appropriate. In this study, we describe the structural differences between common mixing model error formulations in terms of their assumptions about the predation process. We then introduce a new parameterization that unifies these mixing model error structures, as well as implicitly estimates the rate at which consumers sample from source populations (i.e. consumption rate). Using simulations and previously published mixing model datasets, we demonstrate that the new error parameterization outperforms existing models and provides an estimate of consumption. Our results suggest that the error structure introduced here will improve future mixing model estimates of animal diet. This article is protected by copyright. All rights reserved.
Article
Full-text available
Mobulid rays are protected in New Zealand, but the spinetail devilray Mobula japanica is caught as bycatch in skipjack tuna purse seine fisheries. Between 2005 and 2014, rays were recorded in 8.2% of observed purse seine sets. Rays were caught during summer, with a ‘hotspot’ (24.3% of sets) near the shelf edge off North Island over seabed depths of 150–350 m. Rays were usually brailed aboard with the tuna catch from successful sets, but were often entangled in the bunt of the net during unsuccessful sets. Observers tagged nine rays with popup archival tags to obtain preliminary information on their post-release survival, and spatial and vertical movements. Seven of the nine tags reported data, and four of those rays died within 2–4 days of release. All four rays that died had been brought aboard entangled in the bunt. The three surviving rays were all brailed aboard with the tuna catch. One surviving ray remained near New Zealand for 2.7 months during summer, and the other two migrated 1400–1800 km northward to tropical waters near Vanuatu and Fiji at minimum speeds of 47 and 63 km day−1 at the end of summer. Archive data from one ray showed that it made regular vertical movements of 25–100 m amplitude, but spent most of its time shallower than 50 m, more so during the night (89.6%) than the day (76.6%), and mainly experienced temperatures of 18−22 °C. Dives deeper than 200 m were usually made during the day or twilight. All three surviving rays typically moved between the surface and 200–300 m daily, and reached greatest depths of 649 m, 1000 m and 1112 m, respectively, substantially exceeding the previous depth record for this species of 445 m. Recommendations are made for reducing purse seine mortality of mobulid rays by avoiding areas of high ray abundance, avoiding setting on ray-associated tuna schools, and adopting best-practice methods of returning rays to the sea from the net or vessel. Copyright
Article
Full-text available
Studies on resource sharing and partitioning generally consider species that occur in the same habitat. However, subsidies between linked habitats, such as streams and riparian zones, create potential for competition between populations which never directly interact. Evidence suggests that the abundance of riparian consumers declines after fish invasion and a subsequent increase in resource sharing of emerging insects. However, diet overlap has not been investigated. Here, we examine the trophic niche of native fish, invasive fish, and native spiders in South Africa using stable isotope analysis. We compared spider abundance and diet at upstream fishless and downstream fish sites and quantified niche overlap with invasive and native fish. Spider abundance was consistently higher at upstream fishless sites compared with paired downstream fish sites, suggesting that the fish reduced aquatic resource availability to riparian consumers. Spiders incorporated more aquatic than terrestrial insects in their diet, with aquatic insects accounting for 45–90% of spider mass. In three of four invaded trout rivers, we found that the average proportion of aquatic resources in web-building spider diet was higher at fishless sites compared to fish sites. The probability of web-building and ground spiders overlapping into the trophic niche of invasive brown and rainbow trout was as high as 26 and 51%, respectively. In contrast, the probability of spiders overlapping into the trophic niche of native fish was always less than 5%. Our results suggest that spiders share resources with invasive fish. In contrast, spiders had a low probability of trophic overlap with native fish indicating that the traits of invaders may be important in determining their influence on ecosystem subsidies. We have added to the growing body of evidence that invaders can have cross-ecosystem impacts and demonstrated that this can be due to niche overlap.
Article
Information on the movements and population connectivity of the oceanic manta ray (Manta birostris) is scarce. The species has been anecdotally classified as a highly migratory species based on the pelagic habitats it often occupies , and migratory behavior exhibited by similar species. As a result, in the absence of ecological data, population declines in oceanic manta have been addressed primarily with international-scale management and conservation efforts. Using a combination of satellite telemetry, stable isotope and genetic analyses we demonstrate that, contrary to previous assumptions, the species appears to exhibit restricted movements and fine-scale population structure. M. birostris tagged at four sites in the Indo-Pacific exhibited no long-range migratory movements and had non-overlapping geographic ranges. Using genetic and isotopic analysis, we demonstrate that the observed movements and population structure persist on multi-year and generational time scales. These data provide the first insights into the long-term movements and population structure of oceanic manta rays, and suggest that bottom-up, local or regional approaches to managing oceanic mantas could prove more effective than existing, international-scale management strategies. This case study highlights the importance of matching the scales at which management and relevant ecological processes occur to facilitate the effective conservation of threatened species.
Article
Foraging drives many fundamental aspects of ecology, and an understanding of foraging behavior aids in the conservation of threatened species by identifying critical habitats and spatial patterns relevant to management. The world’s largest ray, the oceanic manta (Manta birostris) is poorly studied and threatened globally by targeted fisheries and incidental capture. Very little information is available on the natural history, ecology and behavior of the species, complicating management efforts. This study provides the first data on the diving behavior of the species based on data returned from six tagged individuals, and an opportunistic observation from a submersible of a manta foraging at depth. Pop-off archival satellite tags deployed on mantas at the Revillagigedo Archipelago, Mexico recorded seasonal shifts in diving behavior, likely related to changes in the location and availability of zooplankton prey. Across seasons, mantas spent a large proportion of their time centered around the upper limit of the thermocline, where zooplankton often aggregate. Tag data reveal a gradual activity shift from surface waters to 100–150 m across the tagging period, possibly indicating a change in foraging behavior from targeting surface-associated zooplankton to vertical migrators. The depth ranges accessed by mantas in this study carry variable bycatch risks from different fishing gear types. Consequently, region-specific data on diving behavior can help inform local management strategies that reduce or mitigate bycatch of this vulnerable species.
Article
Mobula thurstoni was the most abundant (58% of the catch), followed by M. japanica (30%), M. munkiana (9%), and M. tarapacana (3%). The study area served as a nursery ground for M. thurstoni, a summer feeding and mating ground for M. thurstoni and M. japanica, and a wintering ground for M. munkiana and young M. thurstoni. Data on size, weight, sex ratio, life history, seasonality, feeding habits, behavior, habitat, and symbionts are presented. Size segregation was a common feature of M. thurstoni, M. japanica and M. munkiana. Summer prey were almost exclusively the euphausiid Nyctiphanes simplex; the mysid Mysidium sp. dominated in winter. -from Author