ArticlePDF Available

Using photographic identification to monitor sea turtle populations at Perhentian Islands Marine Park in Malaysia


Abstract and Figures

Perhentian Islands Marine Park is home to foraging and nesting Green Turtles (Chelonia mydas) and Hawksbill Turtles (Eretmochelys imbricata) but little information is available other than nesting trends and hatching success. We used photographic identification (photo-ID) methods to identify individuals and to determine their sex ratios, habitat use, and site fidelity. We collected 1,826 sightings between 2009 and 2015 from conservation projects (998 in-water sightings, 184 nesting sightings) and members of the public (639 in-water sightings, five nesting sightings), and used NaturePatternMatch (NPM) software and manual visual matching to identify individuals. We identified 120 (minimum) to 131 (maximum) individual Green Turtles, including a maximum of 17 males, 58 females and 56 turtles of unknown sex from both in-water and nesting beach sightings. We identified 20 (minimum) to 23 (maximum) individual Hawksbills of unknown sex from in-water sightings. Green Turtles were sighted most frequently at seagrass beds and Hawksbills only among coral reefs. We resighted 47 Greens and eight Hawksbills between one and 144 times (mean = 23.1 times). Nesting Greens also showed strong site fidelity, although the nesting home range for some individuals included different beaches on adjacent islands within approximately 30 km. We identified boat-related injuries in eight turtles and mortalities of two turtles. Our study suggests that photographs from conservation projects and members of the public were appropriate for photo-ID to provide information on the turtle populations in the Perhentian Islands Marine Park.
Content may be subject to copyright.
Herpetological Conservation and Biology 12(2):350–366.
Submitted: 10 September 2016; Accepted: 20 March 2017; Published: 31 August 2017.
Copyright © 2017. Seh-Ling Long
All Rights Reserved.
Basic biological, ecological, and population
demographic information is essential to species
conservation and management. The identication of
individuals within a population allows the study of
growth rates, age structure, sex ratios, survivorship,
residency, distribution, movement patterns, and
population size, which are important for ecological and
behavioral studies (Wells and Scott 1990; Wilson et al.
2006; Holmberg et al. 2009; Bjorndal et al. 2013). In
sea turtles, tagging using ipper tags and/or passive
integrated transponder (PIT) tags are common methods
used to recognize individuals and track their movements
(Luschi et al. 1996; James et al. 2007). Tags of all types
are more often applied to sexually mature female turtles
due to ease of attachment during the nesting process.
However, information on nesting females only may
underestimate the population size due to the paucity of
knowledge on adult males, sub-adults, and juveniles that
spend most of their time at foraging grounds and out at
sea (Schoeld et al. 2008). More recently, photographic
identication (photo-ID) has been increasingly used for
in-water population and behavioral studies of sea turtles
(Jean et al. 2010; Su et al. 2015; Araujo et al. 2016).
Photographic identication has been used in long-
term studies of large and long-lived species (Würsig
and Jeerson 1990; Baird et al. 2008). It is gaining
popularity as a non-invasive alternative to tagging
where animal capture and tag attachment is not feasible
due to nancial, logistical, or ethical reasons (Thompson
and Wheeler 2008; Su et al. 2015). Many species
bear unique markings or natural patterns that allow
individuals to be identied from photographs, e.g.,
Tigers (Panthera tigris; Hiby et al. 2009) and Pink River
Dolphins (Inia goerensis; Gomez-Salazar et al. 2011).
In Cheloniidae, the facial scale patterns are unique to
individuals and stable over a period of at least 3 y for
Hawksbill Turtles (Eretmochelys imbricata; Dunbar et
al. 2014) and up to 11 y for Green Turtles (Chelonia
mydas; Carpentier et al. 2016). The scale patterns on
the top of the head (Lloyd et al. 2012) and scute patterns
on the carapace (Hall and McNeill 2013) are also
useful for recognizing individual turtles. Other distinct
features, such as barnacle patterns, could also help
with identication (Hall and McNeill 2013). Several
1Institute of Oceanography and Environment, Universiti Malaysia Terengganu, 21030 Kuala Nerus,
Terengganu, Malaysia
2Reef Check Malaysia, Wisma Central, Jalan Ampang, 50450 Kuala Lumpur, Malaysia
3Perhentian Turtle Project, Ecoteer, Kampung Pulau Perhentian, 22300 Kuala Besut, Terengganu, Malaysia
4Corresponding author, email:
    
            
         
  
           
  
 
        
Key Words.—Chelonia mydas; citizen science; computer-assisted pattern recognition; Eretmochelys imbricata;
Green Turtle; Hawksbill Turtle; individual identification
Long and Azmi.—Photo-ID to monitor sea turtles at Perhentian Islands of Malaysia.
methods of photo-ID for sea turtles have been developed
(see Reisser et al. 2008; Schoeld et al. 2008; Jean et
al. 2010; Lloyd et al. 2012; Su et al. 2015). Although
not error free, computer-assisted identication systems
can improve identication capability and accelerate the
process, which is especially needed for large databases
(Carter et al. 2014; Dunbar et al. 2014).
In many places, viewing of sea turtles on nesting
beaches and in the water as a major tourist attraction
has created an opportunity to use citizen scientists
(volunteers and members of the public) in data collection
for population studies (Campbell and Smith 2006; Bell
et al. 2009; Williams et al. 2015). Citizen science can
be dened as public participation in scientic research,
often in collaboration with or under the supervision of
researchers (Dickinson et al. 2010). It has educational
value as it promotes knowledge and increases public
awareness, engagement, and appreciation towards
nature (Cohn 2008). Some photo-ID studies on land
and in the water have used citizen scientists to collect
photographs (Wee and Subaraj 2009; Carpentier et al.
2016). However, data gathered by citizen scientists may
be inaccurate, inconsistent, or unreliable due to their
limited or lack of knowledge, training, and motivation
in scientic research, leading to diculties during
data analysis (Cohn 2008). Error and bias could occur
from non-standardized sampling methods, non-uniform
sampling eort across time and space, or erroneous
reporting (Dickinson et al. 2010). Appropriate training
and supervision can enhance the reliability of data
collected by citizen scientists, providing invaluable data
sets for researchers to study populations (Danielsen
et al. 2014). If well trained and supervised, citizen
science provides a cost-eective way of collecting
data through increased manpower and sampling eort,
allowing biodiversity monitoring over greater spatial
and temporal extents (Goredo et al. 2004).
Four species of sea turtles, Leatherback
(Dermochelys coriacea), Olive Ridley (Lepidochelys
olivacea), and Green and Hawksbill turtles, occur in the
State of Terengganu in Malaysia, and only two species,
Green and Hawksbill turtles, nest and reside in the
waters of Perhentian Islands Marine Park (Chan 2006).
Approximately 20–30% of all Green Turtle nesting
reported in Terengganu occurs on these islands (Sarala
Aikanathan and Jeanne Mortimer, unpubl. report).
To protect nesting habitats and increase hatchling
production, the Terengganu State Department of
Fisheries (DoF) set up a hatchery and declared several
high density nesting beaches as turtle sanctuaries
(Siow and Moll 1982; Chan 2006). During the nesting
season from April to October, DoF rangers patrol the
beaches and relocate nests to the hatchery. Nesting data,
such as counts of landings, nests, eggs incubated and
hatchlings, serve as a baseline for monitoring nesting
trends and hatching success. Little is known, however,
about the in-water turtle populations because research
and conservation work have focused mainly on nesting
sites. Moreover, there is a lack of mark-recapture data.
We used photo-ID as an alternative to tagging, and
we integrated citizen science to gather more sighting
data on Green and Hawksbill turtles that is otherwise
not possible with limited resources. The objective
of this study was to assess nesting and in-water
turtle populations of these species in the Perhentian
Islands Marine Park using sightings data collected by
conservation projects and members of the public. We
used photo-ID methods to identify individual turtles
for developing a sea turtle photo-ID database and to
determine their sex ratios, habitat use, and site delity.
 .—The Perhentian Islands Marine Park
(5°53'49''N 102°43'45''E) is a tourist destination located
in the South China Sea, 21 km o the mainland of
Terengganu on the north-eastern coast of west Malaysia
(Fig. 1A). The marine park consists of several islands,
and only Perhentian Besar (867.3 ha) and Perhentian
Kecil (524.8 ha) are inhabited. The islands lie on a
shallow continental shelf extending from the mainland
with an average maximum depth of 30 m (Simon
Harding et al., unpubl. report), and experience a
seasonal weather pattern with higher rainfall during the
Northeast monsoon from October until March (Suhaila
et al. 2010). The water temperature ranges from 26–
31° C (Wetzelhuetter et al. 2014), and a thermocline is
present at dierent depths in dispersed geographic areas
(Simon Harding et al., unpubl. report). The horizontal
underwater visibility varies spatially from 7.1–15.5
m (Simon Harding et al., unpubl. report); however,
during the monsoon, the visibility can decrease. There
are currents owing northward (April to August) and
southward (November to March) along the Terengganu
coastline (Mohd Akhir 2012), but the islands are more
exposed to local tidal currents that show no prevailing
current direction (Simon Harding et al., unpubl. report).
Surrounding the islands are coral reefs, with the
mainly fringing reefs sloping to a sandy bottom at 20
m and seamount type reefs (Toda et al. 2007; Simon
Harding et al., unpubl. report). There are monospecic
seagrass beds of Enhalus acoroides or Halophila minor,
as well as mixed-species of the latter with Halophila
decipiens, Halophila ovalis, and Halodule pinifolia
(Muta Harah and Japar Sidek 2013). Habitat diversity
is high and patchily distributed, ranging from areas
unsuitable for benthic growth to those supporting high
cover of benthic organisms (Simon Harding et al.,
unpubl. report). The islands were protected as a marine
park in 1994 under the Malaysian Fisheries Act of 1985.
Herpetological Conservation and Biology
The law prohibits touching or taking of any marine
resources whether alive or dead without permission, and
it bans water skiing, speed boat racing, jet skiing, and
shing activities within a radius of two nautical miles
from the lowest tide level of the islands. Less harmful
activities such as snorkeling, scuba diving, swimming,
underwater photography, and kayaking are allowed
and occur before and after the monsoon from March
to October. Peak tourist season is June to August with
the annual number of tourists arriving to the marine
park exceeding 100,000 since 2012, and increasing to
180,481 in 2015 (Department of Marine Park Malaysia,
unpubl. data). There are at least 30 dive sites at natural
and articial reefs and no fewer than 13 snorkel sites,
mostly around fringing reefs or seagrass areas adjacent
to the beach (Tourism Planning Research Group, unpubl.
report). Some of the sites are for diving and snorkeling,
but snorkelers often stay in shallower areas.
. (A) The Perhentian Islands Marine Park within the Terengganu waters in peninsular Malaysia (SEATURTLE.ORG Maptool.
2002. SEATURTLE.ORG, Inc. Available from [Accessed 06 December 2016]), and (B) the location of
turtle sightings in the water and on beaches. Note: The snorkel site in front of Tiga Ruang Beach is also known as Tiga Ruang.
Long and Azmi.—Photo-ID to monitor sea turtles at Perhentian Islands of Malaysia.
   .—The Perhentian
Turtle Project (PTP) carried out photo-ID surveys at
foraging and nesting grounds between April and mid-
October 2015. In the rst month, we trained four interns
(who stayed for the whole study period) on snorkel
surveys and beach monitoring, where they learned about
species and track identication, biometric sampling,
photo-ID methods, free diving, and record keeping.
Twenty volunteers (who stayed 7–14 d) also received
training but may not have acquired all the skills needed
during their stay. Therefore, we conducted every survey
in a team of two to three people where we and trained
interns led and supervised the volunteers (if any) in data
We conducted one to two turtle photo-ID surveys a
day for 5–7 d a week. Each survey lasted 2–3 h between
0900–1300 or 1400–1800. We surveyed predominantly
around the seagrass area at Teluk Pauh (Fig. 1B)
with one observer on a kayak to look out for turtles
breathing at the water surface while the rest snorkeled
to look for turtles underwater. The survey area where
we haphazardly looked for turtles was approximately
4 ha and less than 10 m deep. As sightings of turtles
were sporadic elsewhere, we surveyed other sites
opportunistically. When a turtle was sighted, we free
dived to around eye level of the turtle and photographed
the right and left sides of the face from a minimum
distance of 1.5 m. We also photographed the dorsal
view of the turtle. We used four camera models to
photograph in-water turtles (Canon PowerShot G15 and
PowerShot G16 in their respective underwater housings,
Canon, Inc., Ota, Tokyo, Japan; GoPro Hero 3, GoPro,
Inc., San Mateo, California, USA; and Olympus TG-3,
Olympus Corp., Shinjuku, Tokyo, Japan).
Every night during 1900–0800, we surveyed the
beach with DoF rangers at Tiga Ruang Turtle Sanctuary
(Fig. 1B). After encountered turtles had completed the
oviposition process, we checked the turtles for tags
and measured the curved carapace length. We then
used a digital camera (Canon PowerShot G15, Canon
PowerShot G16, or Olympus TG-3) to take facial
photographs at approximately the eye level of the turtle
from a distance of approximately 0.4 m without camera
ash, but by shining a red LED headlight toward the
head. An articial light source was not necessary when
photographing turtles after 0630. For all encountered
in-water and nesting individuals, we recorded the tag
number (if any) and sighting information, such as date,
time, and location. In addition to the photographs
collected in 2015, we also had photographs of in-
water and nesting turtles taken opportunistically since
2012. During the study period, we also gave a brieng
with instruction on photo-ID data collection to project
leaders of three conservation projects who already
had a background in marine biology and conservation,
including beach monitoring experience and knowledge
of species identication. They contributed sightings
of nesting turtles, as well as photographs of turtles that
they encountered during leisure snorkel and dive trips
since 2011.
.—Members of the public could
submit photographs of turtles via the PTP email (turtle@ and social media sites (e.g., Facebook
and Instagram). There were posters and banners at
various locations around the islands and at the mainland
jetty to provide information on photo-ID, as well as
illustrated guidelines to watch turtles. Upon receiving
the photographs, we contacted the contributor to verify
sightings information because photographs from social
media sites did not have metadata. We recorded the
sighting location as unknown for sightings with no
location information or vague location descriptions, the
sighting date as the month and year the photographs
were taken if the date was unknown, the sighting time
as day or night if the exact time was unknown, and the
camera model if such information was available.
  .—We
distinguished the sex of in-water turtles using tail length
dimorphism whereby individuals with evidently longer
tails that extended signicantly beyond the carapace
were identied as adult male (Wibbels 1999), otherwise
they were assumed to be of undetermined sex or adult
females if they had a tag from a nesting beach. We used
the software NaturePatternMatch (NPM), which detects,
recognizes, and compares natural patterns for individual
identication (Stoddard et al. 2014). The NPM is a less-
species specic adaptation of the Manta Matcher (MM)
identication software for manta rays (Manta spp.; see
Town et al. 2013). Other computer-aided identication
systems tested on hard-shelled sea turtles require the user
to manually select reference points (Dunbar et al. 2014),
code (Jean et al. 2010), or outline the facial scales (Carter
et al. 2014). Instead, the NPM uses the Scale-Invariant
Feature Transform (SIFT) algorithm to automatically
select distinct features for image matching (Lowe 2004).
The selected features appeared to be robust to changes
in location, scale, rotation, illumination, 3D viewpoint,
noise, and occlusion (Lowe 2004, Stoddard et al. 2014).
For example, the SIFT algorithm could detect distinct
points from the pineal spots of Leatherback Turtles and
recognize the same individual from photographs that
varied in illumination, resolution, and viewing angle
(Pauwels et al. 2008). It could also identify manta rays
correctly despite the presence of occlusions (e.g., sh)
and extreme image noise (Town et al. 2013). Thus,
automated software using the SIFT algorithm could
recognize individuals from photographs taken under
challenging conditions underwater or at night.
Herpetological Conservation and Biology
The automated matching process consisted of
(1) image preprocessing; (2) image enhancement;
(3) feature detection and extraction (per left or right
prole); and (4) feature matching. First, we cropped
the left facial scales and rotated the cropped images
to a horizontal angle using GNU Image Manipulation
Program (GIMP) 2.8.14 (Free Software Foundation,
Inc., Boston, Massachusetts, USA; Fig. 2A). Then,
we used NPM, which converted the facial scale images
to grayscale and enhanced the images after noise lter
and contrast adjustment (Fig. 2B). The NPM software
detected and extracted distinct features from the left
facial scales using the SIFT algorithm (Lowe 2004; Fig.
2C), and matched the extracted features against the left
face of all individuals in the database. It then displayed
a ranked list of the left face matches from most to least
similar based on a similarity score (0–1), along with
a condence score (0–1) to indicate how reliable the
rankings are (Fig. 3A). We repeated the process for
the right facial scales to get a ranked list of the right
face matches (Fig. 3B). We visually checked the list
for the correct match (if any) and looked through all the
matches if the condence score was low.
For photographs that did not show the facial scales
clearly enough, we did visual comparison manually
using an identication tree based on the (1) species; (2)
sex; (3) number of post-ocular scales on the right and
left faces; and/or (4) number of temporal and parietal
scales on top of the head (modied from Schoeld et
al. 2008; Lloyd et al. 2012; Su et al. 2015). We used
the facial scale patterns, as well as the natural markings
and scars on the top of the head, shell, and ippers to
identify individuals (Fig. 4). We identied an individual
as a new turtle when we found no match, whereas a
match from either side of the face indicated resighted
turtles. We added every new turtle to a database and
grouped them as identied individuals with photographs
of both sides of the face, left side only, or right side only.
We assigned new turtles an ID to specify the species,
individual, and sex (if known), only if they were sighted
alive with both sides of the face photographed. We gave
new turtles with only one facial prole an ID after seeing
. An example of the photo-ID process for a Green Turtle (Chelonia mydas) where a cropped image of the left facial scale (A)
is converted into grayscale and enhanced (B) before NPM extracts distinct features using SIFT (C). (Photographed by Seh-Ling Long).
Long and Azmi.—Photo-ID to monitor sea turtles at Perhentian Islands of Malaysia.
them again with both sides of the face photographed.
After the rst observer identied the individual from
a new sighting, a second observer checked to conrm
the identity of the individual. To avoid errors and
misidentication, we checked all of the sightings again
to verify identications.
  .—We gathered 1,826 turtle
sightings from conservation projects (n = 1,182) and
members of the public (n = 644), which consisted of
underwater sightings from 2009 to 2015 (n = 1,637)
and nesting sightings from 2011 to 2015 (n = 189). All
sightings occurred between the late January and early
November. The locations of six underwater sightings
and one nesting sighting were unknown whereas the
remaining sightings were from 13 dive and/or snorkel
sites (n = 1,631) and seven beaches (n = 188; Fig. 5).
Due to inadequate information on sighting per unit eort
and frequency per site, it was not possible to quantify
eorts and abundance based on the number of sightings
per site alone.
   .—In-water
sightings from both conservation projects (n = 998) and
members of the public (n = 639) showed diculties in
capturing photographs from a standardized angle for
observations where animals were moving constantly.
Other factors, such as water visibility, light conditions,
and whether or not the turtle was blocked by sh, also
aected the quality of the photographs. By positioning
the camera around the eye level of turtles, all nesting
photographs (n = 184 sightings) from conservation
projects showed the face of the turtle in close-up.
Meanwhile, members of the public photographed
nesting turtles (ve sightings) from a standing position
using ash. In nesting photographs, sometimes the
retraction of the head of the turtle and sand covered a
small part of the facial scales (90 sightings). Successful
identication of both in-water and nesting sightings
was possible using the NPM, as was manual matching,
. Examples of NPM output showing a ranked list of the left (A) or right face (B) matches based on a similarity score of 1 (perfect
match) to 0 (no match). Every matched image shows the facial scales of a dierent individual. The lename is the turtle ID and the
identity indicates the left (L) or right face prole (R). A high condence score indicates that the query image is signicantly more similar
to the rst-ranked matched image than any other images whereas a low condence requires the user to check a larger number of matches
for the correct match (Town et al. 2013). Here, the fourth-ranked and the rst-ranked matches are the correct match for the left side (A)
and right side (B), respectively, showing that the query images match individual P27.
Herpetological Conservation and Biology
so long as photographs were not overly out of focus
and taken at a horizontal and vertical angle of < 45°
where the majority of the facial scales were visible.
For in-water sightings, the scale patterns on top of the
head, scute patterns of the carapace, barnacle patterns,
injuries, and scars, were also useful for identication.
Of the total sightings, 79 (4.3%) were not recognizable
because the photographs were too blurred, including the
only two nesting sightings of Hawksbill Turtles.
The left and right facial patterns were non-identical
for all identied turtles of both species. Identication
of resighted turtles was possible using only one side of
the face, as long as both sides of the facial prole were
already available in the database. Unlike Green Turtles,
the facial scales of Hawksbills showed higher similarity
among individuals and had a smaller number of facial
scales and features for matching. They also showed less
variation in the number and shape of the post-ocular,
. Examples of natural markings and scars used for identication: the number, shape and arrangement of the facial scale patterns
(A, B, C showing the number of post-ocular scales) and at the top of the head (D, E, F showing the number of temporal and parietal scales
surrounding the frontoparietal [FP] scale), the natural markings (G, H) and injuries (I, J) on the shell, or missing a portion of the front (K)
or rear ippers (L). (Fig. 1A photographed by Thomas Brown, Fig. 1B photographed by Charlotte. E. Babbs, Fig. 1C and 1I photographed
by Nicholas J. Tolen, Fig. 1D photographed by Andrea Szalai, Fig. 1E photographed by Kevin Heitzman, Fig. 1F photographed by
Nazirul A. Amin, Fig. 1G photographed by Seh-Ling Long, Fig. 1H and 1L photographed by Petros Persad, Fig. 1J photographed by
William Forster, and Fig. 1K photographed by Rahmat A. Wahab).
Long and Azmi.—Photo-ID to monitor sea turtles at Perhentian Islands of Malaysia.
temporal and sub-temporal scales, and thus, required
the combination of all scales including the tympanic and
central scales for identication. Time intervals between
the rst and last sightings of resighted turtles varied from
5–2,212 d for Green Turtles (78 turtles) and 8–490 d for
Hawksbill Turtles (eight turtles). During this period,
there were no changes in the shape and arrangement of
the scales on the face.
Overall, we identied 96.2% of the sightings from
conservation projects and 94.7% from the public, adding
up to 1,747 sightings. This total represented 96.1% of
the Green Turtle sightings (n = 1,766) and 81.7% of the
Hawksbill Turtle sightings (n = 60), which belonged
to 115 turtles with left and right proles (104 Greens
and 11 Hawksbills), 22 turtles with left prole only (13
Greens and nine Hawksbills) and 17 turtles with right
prole only (14 Greens and three Hawksbills; Table
1). There were 1,567 resightings of Green Turtles, of
which 127 were repeated sightings of the same turtles on
the same day. In contrast, all 26 Hawksbill resightings
occurred on dierent days. All the single-prole turtles
were seen only once. The identied turtles in the water
. (A) The distribution of Green Turtles (Chelonia mydas) and (B) Hawksbill Turtles (Eretmochelys imbricata) in the Perhentian
Islands Marine Park based on sightings from conservation projects and public individuals between 2009 and 2015.
were not the same individuals as the turtles identied on
nesting beaches.
  .—There were 57 Green
Turtles with photographs of both facial proles, and of
these, 14 were males, one was female, and the remaining
42 were of unknown sex. Of nine turtles with the left
prole only, one was a male, and of eight turtles with the
right prole only, there were two males (Table 1). The
maximum number of individual turtles was 74, while
the minimum number of individual turtles was 67 if
the right-prole male and unknown sex turtles were the
same individuals as the left-prole male and unknown
sex turtles. About 80% of the Green Turtles occurred
in mixed seagrass beds at depths of about 3–11 m,
mainly feeding at Teluk Pauh and Atas Busong. The rest
occurred on coral reefs at Batu Layar and Shark Point
where the depths varied from 5–15 m. Only two turtles
found at a seagrass area at Teluk Pauh were also sighted
on coral reefs at Batu Layar, which is about 1.1 km
away. Forty seven turtles were resighted up to 144 times
(mean = 23.1, SD = 32.7) at the same site on dierent
days over about 6.1 y. Two of the resighted turtles had
a tag on both front ippers but we could trace the tags
of only one turtle to its original tagging source. One of
them was tagged by the Sea Turtle Research Unit of the
Universiti Malaysia Terengganu (SEATRU UMT) when
it nested approximately 30 km away at Redang Island
(Fig. 1A) between May and July 2013. This turtle was
rst seen in the water at Teluk Pauh in September 2009
and was resighted foraging at the same site every year
during 2012–2015 between April and October, except in
2013 where there was only one sighting in September.
It had an injury on its carapace when photographed in
September 2015 (Fig. 4I). Eight other resighted turtles
also had healed or new injuries with six of them showing
evidence of boat strikes or propeller cuts (Fig. 4J–L).
Furthermore, there were four sightings of dead Green
Turtles, and photographs revealed boat-related injuries
on two of them, while cause of death was unknown
for the other two. Of four mortalities, one turtle had
decomposed beyond recognition. One turtle matched an
individual from the database, which was spotted three
times at Teluk Pauh. We presumed the remaining two
turtles to be new individuals because there was no match
in the photo-ID database.
  .—There were 11
Hawksbill Turtles with both facial proles, nine with
the left prole only and three with the right prole only
(Table 1). There could have been a maximum of 23
individual turtles or a minimum of 20 turtles, if all three
right-prole turtles were the same individuals as the left-
prole turtles. The sex of all the Hawksbill Turtles was
unknown. The observed Hawksbills occurred on coral
reefs that were 5–15 m deep. Eight turtles were seen at
the same site two to six times on dierent days over a
period of 490 d between 2014 and 2015. Only one of the
resighted turtles used two dierent sites that are about
3.7 km apart, namely Teluk Kerma and Teluk Pauh. One
Hawksbill Turtle was missing a portion of the shell at its
left rear ipper.
.—We identied 47 individual
female turtles with photographs of both facial proles,
plus four turtles with the left prole only and six turtles
with the right prole only (Table 1). If all of these turtles
were dierent individuals, there was a maximum of 57
nesting turtles. If all turtles with the left prole only
were the same individuals as the right-prole turtles,
there was a minimum of 53 nesting turtles. There were
no remigrants over the 5-y period. All the turtles were
sighted once, except for 31 turtles that laid between two
to nine nests within a season in 2015. The time interval
between the rst and subsequent sighting of the same
turtle ranged from 1–65 d. Of these renesters, 25 nested
on one specic beach, while the other six turtles nested
on multiple beaches in the Perhentians. Three renesters
had tags from the DoF, and records were found for two
of them. Both were tagged while nesting at Redang
. Numbers of identied Green Turtles (Chelonia mydas) and Hawksbill Turtles (Eretmochelys imbricata) in the water and on
nesting beaches of the Perhentians, Malaysia, based on their facial proles and the minimum (min.) and maximum (max.) numbers of
unique individuals.
Green Turtles Hawksbill Turtles
Facial prole Facial prole
Population Sex
side Min. Max.
side Min. Max.
In-water Male 14 1 2 16 17 - - - - -
Female 1 - - 1 1 - - - - -
Unknown 42 86 50 56 11 9 3 20 23
Nesting Female 47 4 6 53 57 - - - - -
Total 104 13 14 120 131 11 9 3 20 23
Herpetological Conservation and Biology
Island in 2009 and 2010, respectively, and were among
the six females that used multiple nesting sites in the
Perhentians in 2015.
Photographic identication provided a way to assess
two distinct in-water and nesting sea turtle populations
in the Perhentian Islands Marine Park. Between 2009
and 2015, there were 67–74 Green Turtles including
16–17 adult males and one adult female sighted in the
water, and 20–23 Hawksbill Turtles of undetermined
sex sighted between 2014 and 2015. There were another
53–57 female Green Turtles that nested from 2011 to
2015. For in-water turtles, we could distinguish adult
males using tail length dimorphism and adult females
by tracing tags to a nesting site. The rest could either
be sexually immature males or females of dierent life
stages (i.e., juvenile, sub-adult, or adult). Measuring
turtle carapace length using paired-laser photogrammetry
could provide more information on population structure
and size distributions to determine which size classes
occupy an area (Araujo et al. 2016).
 .—Green Turtles were the
predominant species sighted at seagrass beds, with a
few observed in coral reef areas, whereas sightings
of Hawksbills occurred only in coral reef areas. All
resighted individuals displayed high levels of delity to
a particular site, and only a few were seen at more than
one site. Studies have shown that sea turtle habitat use
and movement patterns within a foraging home range
are related to their diet (León and Bjorndal 2002; Read
and Limpus 2002), the availability of food (Makowski et
al. 2006; Berube et al. 2012), and shelter (Makowski et
al. 2006). Sea turtles may have more than one preferred
site within their foraging range (Semino and Jones
2006) and they can develop delity to specic foraging
areas when there are sucient resources (Makowski et
al. 2006). Green Turtles in the western South Atlantic
spent a longer time in shallow areas where food was
more abundant than turtles in areas where food was less
abundant (Reisser et al. 2013). Although Green Turtles
feed primarily on seagrass, they are known to also feed
on algae found in coral reef areas (Read and Limpus
2002). Hawksbill Turtles were found more frequently
at rocky reef areas compared to sandy bottom areas
o the North Pacic coast of Costa Rica, presumably
because the rocky reefs sustained a higher diversity of
Hawksbill prey species, including sponges, algae, and
invertebrates (Carrión-Cortez et al. 2013). The sighting
of Green Turtles mostly at seagrass beds and Hawksbill
Turtles only in coral reef areas in the Perhentians could
also be related to their diet and the availability of food
within an area to which they repeatedly returned.
One Green Turtle photographed at an in-water site
had ipper tags from a nesting beach 30 km away from
the Perhentians. It nested at Redang Island between
May and July 2013 and was found at the same foraging
site in the Perhentians before (2009 and 2012) and
after (September 2013 onwards) nesting, suggesting
short distance migrations between foraging and nesting
grounds. Previous satellite tracking of ve nesting
Green Turtles from Redang Island showed only long
distance post-nesting migrations over 670 to 1,700
km to foraging grounds in neighboring countries, and
no turtles moved northward (Papi et al. 1995; Luschi
et al. 1996). Post-nesting Green Turtles tracked from
mainland Terengganu had also migrated long distances
to foraging grounds (van de Merwe et al. 2009). Short-
distance migrations are less well known among Green
Turtle populations in our study region, but have been
previously reported elsewhere. One example is the
Green Turtles that Whiting et al. (2008) tracked from
Cocos (Keeling) Islands where all six migrated fewer
than 40 km to foraging areas after nesting. Green
Turtles nesting in the Galapagos Islands have also been
observed to reside within 75 km of the nesting beach
(Semino et al. 2008). In Costa Rica, some post-nesting
Green Turtles moved to foraging grounds as close as 5
km to the nesting beach, while some turtles migrated
as far as 1,086 km (Blanco et al. 2012). Migration
has a high energy cost, and is aected by various
factors, including body size, energy, and mortality
cost (Alerstam et al. 2003). Short-distance migrations
require less energy and can occur if foraging and nesting
grounds are in close proximity (Whiting et al. 2008).
Because the turtle tagged on Redang Island showed
residency at its foraging ground, the Perhentian Islands
probably met its resource needs, such as adequate food
and shelter. It may be a tness strategy for the turtle that
saves energy by performing short-distance migrations
between foraging and nesting grounds.
Photographs showed boat strikes and propeller cut
injuries in eight turtles and mortalities in two turtles
in the Perhentians. With rapid tourism development,
injuries to and mortalities of resident turtles due to
increased boat trac around the islands may aect
breeding populations elsewhere. We did not examine
the nesting turtles for injuries, but they could also
be vulnerable to boat strikes because nesting turtles
usually stay close to the nesting beach during breeding
season (van de Merwe et al. 2009). So far there are no
regulations for speed limits in the Perhentian Islands
Marine Park and beyond two nautical miles from the
islands. As Green Turtles cannot avoid boats moving
faster than 4 km/h (Hazel et al. 2007), designation of
conservation zones to regulate boat activities, such as
limiting the boat speed and boat density, are possible
actions that can be taken, as in National Marine Park of
Long and Azmi.—Photo-ID to monitor sea turtles at Perhentian Islands of Malaysia.
Zakynthos, Greece, to mitigate boat impacts to turtles
in nearshore foraging grounds (Schoeld et al. 2013).
However, sucient evidence-based information that is
site-specic, such as seasonal dierences in the use of
an area by turtles and boats, is required to appropriately
make such recommendations (Schoeld et al. 2013).
 .—Green Turtles were the most
frequently sighted species in nesting areas. The majority
of renesters showed strong site delity returning to
the same beach, while only some individuals showed
weaker nesting site delity by nesting on multiple
beaches. Evidence from tagged females indicated that
some turtles nesting at Perhentian Islands had nested
at Redang Island in previous years. This suggests that
the nesting season home ranges of some adult females
span across islands within a range of 30 km. It is well-
established that Green Turtles possess a high degree of
delity in returning to previous nesting grounds (Plotkin
2003) as well as their natal rookeries for nesting (Allard
et al. 1994). They may return to a specic nesting site
(Luschi et al. 1996; Ekanayake et al. 2010), or use
multiple nest sites within a considerable area spanning a
few hundred kilometers (Bjorndal et al. 1983). As there
are also other islands in the vicinity of the Perhentians,
it is possible that the rookery consists of a number of
dierent nesting beaches on any of those islands. There
were female Green Turtles returning to the Perhentians
to nest again within one nesting season. However, due
to inadequate and incomplete data prior to 2015, there
was no record showing that nesting females from the
photo-ID database were remigrants returning to the
Perhentians for nesting in subsequent years. Thus,
long-term monitoring initiatives and mark-recapture
data will greatly contribute to understanding the nesting
population in the Perhentians.
   .—Most photo-
ID studies on hard-shelled sea turtles have used facial
scale patterns for recognizing individuals photographed
underwater (Reisser et al. 2008; Jean et al. 2010;
Chassagneux et al. 2013; Su et al. 2015; Carpentier et al.
2016), out of water (Dunbar et al. 2014), and on nesting
beaches at night (Valdés et al. 2014; Chew et al. 2015).
Lloyd et al. (2012), Hall and McNeill (2013), and Dunbar
et al. (2014) showed that individual identication is also
possible using photographs of the dorsal view of the
head and carapace. For nesting turtles, we only used the
facial scale patterns to identify individuals because only
the facial scales of turtles illuminated by the headlight
were visible in nesting photographs. We were also better
able to photograph the face of every nesting turtle close-
up (as close as 0.4 m) compared to constantly moving
in-water turtles. In-water photographs showed facial
scales at varying angles, but other visible features, such
as the scales on the top of the head and scute patterns
of the carapace, could aid with identication. These
features were less available for nesting turtles because
of the diculties in photographing the dorsal view of
the head and shell using headlights as a light source and
when the turtles were camouaging their nests. The
sand that landed on the head and shell further limited
the observation of the scale and scute patterns.
Natural markings need to be stable throughout
the lifespan of an animal to use these features to
consistently identify individuals in a population. In our
study, the facial scales were the most reliable features to
identify in-water and nesting individuals of both species
because the patterns were distinct for every individual
and remained the same over at least 2,212 d for Green
Turtles and 490 d for Hawksbill Turtles. Carpentier et
al. (2016) also showed that individual Green Turtles
were distinguishable from their facial scale patterns
over 3,954 d (almost 11 y), while the longest period
recorded for Hawksbills was 1,155 d (Dunbar et al.
2014). Obtaining both left and right facial proles was
important to avoid double-counting and to successfully
identify a new turtle. Once we had photographically
documented the left and right proles of an individual,
we were able to identify resighted turtles even if only
one side of the face was photographed. Others have also
stressed the importance of photographing both sides of
the face to increase successful identication (Chew et al.
2015; Su et al. 2015).
The head and shell scale patterns, injuries, scars,
coloration, and epibionts (e.g., barnacles) served as
complementary and secondary features to aid with
identication. Instantaneous recognition of certain
individuals with obvious markings was possible.
However, this could lead to a bias towards recognizing
easily identiable individuals in poor-quality
photographs (Davies et al. 2012). These secondary
features helped with the identication of turtles within a
season, but to use them to identify individuals over time
could be challenging because some of these features may
change (Hall and McNeill 2013). For example, some
scars could be permanent, but injuries could change
throughout the healing process. Furthermore, using
solely the scales on the top of the head to recognize
individuals resulted in a lower successful identication
rate, and Lloyd et al. (2012) suggested using the shell
markings for distinguishing individuals sharing similar
head patterns. The secondary features alone might not
be sucient to successfully identify most individuals,
but using them along with the facial scales could
increase successful identication.
For both species, manual matching was faster when
there were fewer than 20 individuals in the database.
Manual matching becomes more time consuming and
labor-intensive with large amounts of photographs
Herpetological Conservation and Biology
(Arzoumanian et al. 2005; Chew et al. 2015), in which
the time used to manually match new photographs
is proportional to the size of the database (Dunbar et
al. 2014). When the database increased, automated
matching reduced the amount of time used to manually
search through every individual in the database. The
NPM was less suitable for Hawksbill Turtles due to the
considerably smaller database, and manual matching
was faster. The smaller number of scales and a high
similarity in facial patterns shared by all Hawksbill
Turtles could also possibly result in the NPM producing
fewer distinguishable features between individuals and
a lower successful identication rate.
Automated identication software can deal with
photographs taken under challenging conditions
(Pauwels et al. 2008; Town et al. 2013) while reducing
the time required to match a new photograph against all
individuals in the database (Carter et al. 2014; Dunbar et
al. 2014). However, the quality of photographs is critical
when using software because poor-quality photographs
can reduce the match success (Kelly 2001; Speed et al.
2007). Moreover, software also produced false positive
matches when matching individuals not previously
in the database, whereas visual matching resulted in a
high number of correct matches (Dunbar et al. 2014).
Successful identications were usually possible using
visual matching even if photographs were not clear
enough for software identication (Davies et al. 2012).
Considering the drawbacks and restrictions of both
methods, we combined the use of automated pattern
matching software and manual visual comparison for
matching in-water and nesting photographs of highly
variable quality.
 .—Our results indicated that
photographs gathered by the public can be as useful as
those collected by trained personnel from conservation
projects. Providing information and training on photo-
ID data collection helps ensure that photographs are
suitable for photo-ID. Even so, there are limitations
and biases for using photo-ID and opportunistic citizen
science data. For example, some individuals in the wild
can become accustomed to human presence (Whittaker
and Knight 1998), and therefore are more likely to be
photographed than those that avoid humans (Kostas
2015). Such bias associated with photo-ID could lead
to heterogeneity in sighting probabilities, which will
underestimate the actual population (Sosa-Nishizaki et
al. 2012). The use of citizen science could also result
in observation and geographical biases from uneven
sighting eorts over time (Bell et al. 2009) and at
dierent sites (Dennis and Thomas 2000), depending on
the weather conditions (Koss et al. 2009), availability
and behavior (e.g., motivations and preferences) of
citizen scientists (Koss et al. 2009; Isaac and Pocock
2015; Boakes et al. 2016). In Reunion Island, the
turtle distribution increased with sighting eorts that
were dependent on the weather and tourism activity
(Chassagneux et al. 2013). Similarly, the sighting
eorts by citizen scientists in the Perhentians were
restricted to areas for diving and snorkeling during the
tourist season. Therefore, the turtle distribution patterns
in our study probably did not reect the true distribution
of turtles in the Perhentians.
Furthermore, reporting bias could happen when
citizen scientists provide inadequate or incomplete
information (Isaac and Pocock 2015). In our study,
the sighting information from citizen scientists was
not always available or accurate. Tourists reported the
location using English names that were not standardized
and could be misleading. For example, Turtle Beach
could mean any of the nesting beaches, while Turtle
Point might be the seagrass areas of Teluk Pauh or Atas
Busong. Additionally, tourists from dierent countries
had dierent time zones set on their cameras, which
resulted in incorrect sighting date and time. We could
resolve this, but not always, through conrmation with
the tourists. Although citizen scientists used dierent
types of cameras, identication was possible as long as
the photographs were not blurry, and the facial scales
were visible. Receiving photographs in their original
size could increase successful identication because
there were dierences in the image size and quality
between photographs submitted through email and
social media. The image size of photographs received
through social media had been reduced whereas
photographs received through email remained bit-for-
bit identical to the originals. We also lacked data on
the sighting eorts by citizen scientists (i.e., time per
dive/snorkel and frequency per site), which is often an
issue associated with using citizen science (Richardson
et al. 2012; Williams et al. 2015). Hence, increased
survey time and frequency and/or density of turtles per
site could contribute to a higher number of sightings at
a particular site in the Perhentians, although additional
data are required to determine this.
The photo-ID method is less invasive than capture
and tagging methods, but there might be impacts of
human-sea turtle interactions from photographing sea
turtles. Bell et al. (2009) observed that Hawksbill
Turtles were less disturbed by divers than Green Turtles
in Cayman Islands. At Mayotte Island, Green Turtles
generally avoided snorkelers except for one location
where they seemed to be unaected by human presence
(Roos et al. 2005). The dierent responses to human
presence imply that interactions with humans may alter
(Meadows 2004; Taquet et al. 2006) or even negatively
impact turtle behaviors (e.g., Hawksbills spent less time
searching for food, feeding and breathing when divers
approached them; Hayes et al. 2016). Apart from that,
Long and Azmi.—Photo-ID to monitor sea turtles at Perhentian Islands of Malaysia.
the presence and behavior of tourists, including the use
of ash cameras, could disturb nesting turtles (Jacobson
and Lopez 1994; Waayers et al. 2008). Thus, it requires
training to know when and how to photograph nesting
turtles at night. The use of citizen science in photo-ID
can be useful in studying sea turtles in water and on
nesting beaches but practical guidelines, training, and
supervision are needed to minimize human impacts and
ensure that the data collection is reliable.
.—A photo-
ID database provided information on the in-water and
nesting sea turtle populations in the Perhentian Islands
Marine Park, including the number of individuals, sex
ratios, habitat use, and threats from human-sea turtle
interactions. Photographic identication also provides
mark-recapture data useful for future population models
to estimate population parameters, such as abundance,
survival, recruitment, residency, and population size,
and growth rate, as demonstrated in the studies of
other species, e.g., whales (Gowans et al. 2000) and
sharks (Holmberg et al. 2009; Gore et al. 2016). All
of this information can contribute towards informing
local governments in developing sustainable tourism,
conservation, and management practices in marine
parks. Thus, photo-ID can be a useful tool for long-term
population monitoring, assessing emerging or dominant
threats, and evaluating conservation status. Increased
sighting eorts at other beaches and in other nearshore
and deeper waters, as well as during the o-tourist
season, can enhance understanding of the temporal and
spatial turtle distributions, site delity, and movement
patterns in the marine park. However, due to the weather
conditions during the Northeast monsoon, poor water
visibility could impede the use of photo-ID. Regional
collaboration to share photo-ID databases would enable
the tracking of turtles to determine connectivity between
foraging and nesting sites at a regional scale (Schoeld
et al. 2008; Dunbar et al. 2014; Su et al. 2015). Such
understanding could then be used to assist in the
development of regional management and conservation
plans for sea turtles.
Acknowledgments.—We sincerely thank Daniel
Quilter for his support and invaluable advice, the sta
members, interns and volunteers of the PerhentianTurtle
Project, Blue Temple Conservation, Bubbles Turtle and
Reef Conservation Project and Perhentian Community
and Teaching Project, especially Nicholas John Tolen,
Petros Persad, Charlotte Emily Babbs, Muhammad Azri
Fakrullah Mohd Azhar, Neil Hinds, Sabina Gramaglia-
Hinds, Thomas Horsell, Shauna Tay Siew Li, Ng Chew
Tsann, Csaba Szilvási, and Kevin Heitzman, as well as
the DoF rangers, including Noor Hisam Bin Mohammad,
Ismail Bin Yaakob, Amran Bin Mohammad, Muhammad
Ghazali, Saiful Nizam Bin Ismail, Mohamad Syukiiran
Deraman, and Che Husin Bin Che Harum, for their
support and assistance in the eld, the members of
the public, especially Rahmat Bin Abd Wahab, for the
photographs, Christopher Town for kindly providing
access to the NaturePatternMatch, William Forster and
Thomas Brown for testing the software, and SEATRU
UMT for conrming the tag numbers. We are grateful
for the support from the local communities, government
agencies, resorts, chalets, dive centers, restaurants, and
cafes. This research was conducted in collaboration
with the Terengganu State Department of Fisheries
(Prk. Tr. 2214) and under the research permit from the
Department of Marine Park Malaysia (JTLM 630-7
Jld. 6 (12)). This research was nancially supported
by National Marine Aquarium and Durham University
Charities Kommittee.
Alerstam, T., A. Hedenström, and S. Åkesson. 2003.
Long-distance migration: evolution and determinants.
Oikos 103:247–260.
Allard, M.W., M.M. Miyamoto, K.A. Bjorndal, A.B.
Bolten, and B.W. Bowen. 1994. Support for natal
homing in Green Turtles from mitochondrial DNA
sequences. Copeia 1994:34–41.
Araujo, G., J. Montgomery, K. Pahang, J. Labaja,
R. Murray, and A. Ponzo. 2016. Using minimally
invasive techniques to determine Green Sea Turtle
Chelonia mydas life-history parameters. Journal of
Experimental Marine Biology and Ecology 483:25–
Arzoumanian, Z., J. Holmberg, and B. Norman. 2005.
An astronomical pattern-matching algorithm for
computer-aided identication of Whale Sharks
Rhincodon typus. Journal of Applied Ecology
Baird, R.W., A.M. Gorgone, D.J. McSweeney, D.L.
Webster, D.R. Salden, M.H. Deakos, A.D. Ligon,
G.S. Schorr, J. Barlow, and S.D. Mahay. 2008.
False Killer Whales (Pseudorca crassidens) around
the main Hawaiian Islands: long-term site delity,
inter-island movements, and association patterns.
Marine Mammal Science 24:591–612.
Bell, C.D., J.M. Blumenthal, T.J. Austin, G. Ebanks-
Petrie, A.C. Broderick, and B.J. Godley. 2009.
Harnessing recreational divers for the collection of
sea turtle data around the Cayman Islands. Tourism
in Marine Environments 5:245–257.
Berube, M.D., S.G. Dunbar, K. Rützler, and W.K. Hayes.
2012. Home range and foraging ecology of juvenile
Hawksbill Sea Turtles (Eretmochelys imbricata) on
inshore reefs of Honduras. Chelonian Conservation
and Biology 11:33–43.
Herpetological Conservation and Biology
Bjorndal, K.A., A.B. Meylan, and B.J. Turner. 1983.
Sea turtles nesting at Melbourne Beach, Florida, I.
Size, growth and reproductive biology. Biological
Conservation 26:65–77.
Bjorndal, K.A., J. Parsons, W. Mustin, and A.B. Bolten.
2013. Threshold to maturity in a long-lived reptile:
interactions of age, size, and growth. Marine Biology
Blanco, G.S., S.J. Morreale, H. Bailey, J.A. Semino,
F.V. Paladino, and J.R. Spotila. 2012. Post-nesting
movements and feeding grounds of a resident East
Pacic Green Turtle Chelonia mydas population
from Costa Rica. Endangered Species Research
Boakes, E.H., G. Gliozzo, V. Seymour, M. Harvey, C.
Smith, D.B. Roy, and M. Haklay. 2016. Patterns of
contribution to citizen science biodiversity projects
increase understanding of volunteers’ recording
behaviour. Scientic Reports 6:33051.
Campbell, L.M., and C. Smith. 2006. What makes them
pay? Values of volunteer tourists working for sea
turtle conservation. Environmental Management
Carpentier, A.S., C. Jean, M. Barret, A. Chassagneux,
and S. Ciccione. 2016. Stability of facial scale
patterns on Green Sea Turtles Chelonia mydas over
time: a validation for the use of a photo-identication
method. Journal of Experimental Marine Biology
and Ecology 476:15–21.
Carrión-Cortez, J., C. Canales-Cerro, R. Arauz, and
R. Riosmena-Rodríguez. 2013. Habitat use and
diet of juvenile Eastern Pacic Hawksbill Turtles
(Eretmochelys imbricata) in the North Pacic Coast
of Costa Rica. Chelonian Conservation and Biology
Carter, S.J.B., I.P. Bell, J.J. Miller, and P.P. Gash. 2014.
Automated marine turtle photograph identication
using articial neural networks, with application
to Green Turtles. Journal of Experimental Marine
Biology and Ecology 452:105–110.
Chan, E.H. 2006. Marine turtles in Malaysia: on the
verge of extinction? Aquatic Ecosystem Health and
Management 9:175–184.
Chassagneux, A., C. Jean, J. Bourjea, and S. Ciccione.
2013. Unravelling behavioral patterns of foraging
Hawksbill and Green Turtles using photo-
identication. Marine Turtle Newsletter 137:1–5.
Chew, V.Y.C., H.C. Liew, and J. Joseph. 2015.
Photographic identication of Green Turtles
(Chelonia mydas) at Redang Island, Malaysia.
Marine Turtle Newsletter 146:1–6.
Cohn, J.P. 2008. Citizen science: can volunteers do real
research? BioScience 58:192–197.
Danielsen, F., P.M. Jensen, N.D. Burgess, R. Altamirano,
P.A. Alviola, H. Andrianandrasana, J.S. Brashares,
A.C. Burton, I. Coronado, N. Corpuz, et al. 2014.
A multicountry assessment of tropical resource
monitoring by local communities. Bioscience
Davies, T.K., G. Stevens, M.G. Meekan, J. Struve, and
J.M. Rowclie. 2012. Can citizen science monitor
Whale-Shark aggregations? Investigating bias
in mark-recapture modelling using identication
photographs sourced from the public. Wildlife
Research 39:696–704.
Dennis R.L.H., and C.D. Thomas. 2000. Bias in
buttery distribution maps: the inuence of hot
spots and recorder’s home range. Journal of Insect
Conservation 4:73–77.
Dickinson, J.L., B. Zuckerberg, and D.N. Bonter.
2010. Citizen science as an ecological research tool:
challenges and benets. Annual Review of Ecology,
Evolution and Systematics 41:149–172.
Dunbar, S.G., H.E. Ito, K. Bahjri, S. Dehom, and L.
Salinas. 2014. Recognition of juvenile Hawksbills
Eretmochelys imbricata through face scale
digitization and automated searching. Endangered
Species Research 26:137–146.
Ekanayake, E.M.L, R.S. Rajakaruna, T. Kapurusinghe,
M.M. Saman, D.S. Rathnakumara, P. Samaraweera,
and K.B. Ranawana. 2010. Nesting behaviour of the
Green Turtle at Kosgoda Rookery, Sri Lanka. Ceylon
Journal of Science (Biological Sciences) 39:109–
Goredo, S., C. Piccinetti, and F. Zaccanti. 2004.
Volunteers in marine conservation monitoring: a
study of the distribution of seahorses carried out
in collaboration with recreational scuba divers.
Conservation Biology 18:1492–1503.
Gomez-Salazar, C., F. Trujillo, and H. Whitehead. 2011.
Photo-identication: a reliable and noninvasive tool
for studying Pink River Dolphins (Inia georensis).
Aquatic Mammals 37:472–485.
Gore, M.A., P.H. Frey, R.F. Ormond, H. Allan, and
G. Gilkes. 2016. Use of photo-identication and
mark-recapture methodology to assess Basking
Shark (Cetorhinus maximus) populations. PLoS
ONE 11, 1–22.
Gowans, S., H. Whitehead, J.K. Arch, and S.K. Hooker.
2000. Population size and residency patterns
of Northern Bottlenose Whales (Hyperoodon
ampullatus) using the Gully, Nova Scotia. Journal of
Cetacean Research and Management 2:201–210.
Hall, A.G., and J.B. McNeill. 2013. Inferring sea turtle
recapture rates using photographic identication.
Herpetological Review 44:561–569.
Hayes, C.T., D.S. Baumbach, D. Juma, and S.G. Dunbar.
2016. Impacts of recreational diving on Hawksbill
Sea Turtle (Eretmochelys imbricata) behaviour
Long and Azmi.—Photo-ID to monitor sea turtles at Perhentian Islands of Malaysia.
in a marine protected area. Journal of Sustainable
Tourism 25:79–95.
Hazel, J., I.R. Lawler, H. Marsh, and S. Robson.
2007. Vessel speed increases collision risk for the
Green Turtle Chelonia mydas. Endangered Species
Research 3:105–113.
Hiby, L., P. Lovell, N. Patil, N.S. Kumar, A.M.
Gopalaswamy, and K.U. Karanth. 2009. A Tiger
cannot change its stripes: using a three-dimensional
model to match images of living Tigers and Tiger
skins. Biology Letters 5:383–386.
Holmberg, J., B. Norman, and Z. Arzoumanian. 2009.
Estimating population size, structure, and residency
time for Whale Sharks Rhincodon typus through
collaborative photo-identication. Endangered
Species Research 7:39–53.
Isaac, N.J.B., and M.J.O. Pocock. 2015. Bias and
information in biological records. Biological Journal
of the Linnean Society 115:522–531.
Jacobson, S.K., and A.F. Lopez. 1994. Biological
impacts of ecotourism: tourists and nesting turtles
in Tortuguero National Park, Costa Rica. Wildlife
Society Bulletin 22:414–419.
James, M.C., S.A. Sherrill-Mix, and R.A. Myers. 2007.
Population characteristics and seasonal migrations
of Leatherback Sea Turtles at high latitudes. Marine
Ecology Progress Series 337:245–254.
Jean, C., S. Ciccione, E. Talma, K. Ballorain, and J.
Bourjea. 2010. Photo-identication method for
Green and Hawksbill Turtles - rst results from
Reunion. Indian Ocean Turtle Newsletter 11:8–13.
Kelly, M.J. 2001. Computer-aided photograph matching
in studies using individual identication: an example
from Serengeti Cheetahs. Journal of Mammalogy
Koss, R.S., K. Miller, G. Wescott, A. Bellgrove, A.
Boxshall, J. McBurnie, A. Bunce, P. Gilmour, and D.
Ierodiaconou. 2009. An evaluation of Sea Search as a
citizen science programme in marine protected areas.
Pacic Conservation Biology 15:116–127.
Kostas, P. 2015. In-water behaviour of the Loggerhead
Sea Turtle (Caretta caretta) under the presence of
humans (Homo sapiens) in a major Mediterranean
nesting site. P. 47 In Book of Abstracts of 35th Annual
Symposium on Sea Turtle Biology and Conservation.
Kaska, Y., B. Sönmez, O. Türkecan, and C. Sezgin
(Eds.). MACART Press, Denizli, Turkey.
León, Y.M., and K.A. Bjorndal. 2002. Selective feeding
in the Hawksbill Turtle, an important predator in
coral reef ecosystems. Marine Ecology Progress
Series 254:249–258.
Lloyd, J.R., M.Á. Maldonado, and R. Staord. 2012.
Methods of developing user-friendly keys to
identify Green Sea Turtles (Chelonia mydas L.)
from photographs. International Journal of Zoology
Lowe, D.G. 2004. Distinctive image features from
scale-invariant keypoints. International Journal of
Computer Vision 60:91–110.
Luschi, P., F. Papi, H.C. Liew, E.H. Chan, and F.
Bonadonna. 1996. Long-distance migration and
homing after displacement in the Green Turtle
(Chelonia mydas): a satellite tracking study. Journal
of Comparative Physiology A 178:447–452.
Makowski, C., J.A. Semino, and M. Salmon. 2006.
Home range and habitat use of juvenile Atlantic
Green Turtles (Chelonia mydas L.) on shallow
reef habitats in Palm Beach, Florida, USA. Marine
Biology 148:1167–1179.
Meadows, D. 2004. Behavior of Green Sea Turtles in
the presence and absence of recreational snorkelers.
Marine Turtle Newsletter 103:1–4.
Mohd Akhir, M.F. 2012. Surface circulation and
temperature distribution of Southern South China
Sea from global ocean model (OCCAM). Sains
Malaysiana 41:701–714.
Muta Harah, Z., and B. Japar Sidek. 2013. Occurrence
and distribution of seagrasses in waters of Perhentian
Island Archipelago, Malaysia. Journal of Fisheries
and Aquatic Science. 8:441–451.
Papi, F., H.C. Liew, P. Luschi, and E.H. Chan. 1995.
Long-range migratory travel of a Green Turtle
tracked by satellite: evidence for navigational ability
in the open sea. Marine Biology 122:171–175.
Pauwels, E.J., P.M. de Zeeuw, and D.M. Bounantony.
2008. Leatherbacks matching by automated image
recognition. Pp. 417–425 In Advances in Data
Mining. Medical Applications, E-Commerce,
Marketing, and Theoretical Aspects: 8th Industrial
Conference, ICDM 2008 Leipzig, Germany, July
16–18, 2008 Proceedings. Perner, P. (Ed.). Springer,
Berlin, Germany.
Plotkin, P. 2003. Adult migrations and habitat use. Pp.
225–241 In The Biology of Sea Turtles Volume II.
Lutz, P.L., J.A. Musick, and J. Wyneken (Eds.). CRC
Press, Boca Raton, Florida, USA.
Read, M.A., and C.J. Limpus. 2002. The Green Turtle,
Chelonia mydas, in Queensland: feeding ecology
of immature turtles in Moreton Bay, Southeastern
Queensland. Memoirs of the Queensland Museum
Reisser, J., M. Proietti, P. Kinas, and I. Sazima. 2008.
Photographic identication of sea turtles: method
description and validation, with an estimation of tag
loss. Endangered Species Research 5:73–82.
Reisser, J., M. Proietti, I. Sazima, P. Kinas, P. Horta, and
E. Secchi. 2013. Feeding ecology of the Green Turtle
(Chelonia mydas) at rocky reefs in Western South
Atlantic. Marine Biology 160:3169–3179.
Herpetological Conservation and Biology
Richardson, J., A.G. Wood, A. Neil, D. Nowacek, and
M. Moore. 2012. Changes in distribution, relative
abundance, and species composition of large whales
around South Georgia from opportunistic sightings:
1992 to 2011. Endangered Species Research 19:149–
Roos, D., D. Pelletier, S. Ciccione, M. Taquet, and
G. Hughes. 2005. Aerial and snorkelling census
techniques for estimating Green Turtle abundance on
foraging areas: a pilot study in Mayotte Island (Indian
Ocean). Aquatic Living Resources 18:193–198.
Schoeld, G., K.A. Katselidis, P. Dimopoulos, and J.D.
Pantis. 2008. Investigating the viability of photo-
identication as an objective tool to study endangered
sea turtle populations. Journal of Experimental
Marine Biology and Ecology 360:103–108.
Schoeld, G., R. Scott, A. Dimadi, S. Fossette, K.A.
Katselidis, D. Koutsoubas, M.K.S. Lilley, J.D. Pantis,
A.D. Karagouni, and G.C. Hays. 2013. Evidence-
based marine protected area planning for a highly
mobile endangered marine vertebrate. Biological
Conservation 161:101–109.
Semino, J.A., and T.T. Jones. 2006. Diel movements
and activity ranges of Green Turtles (Chelonia
mydas) at a temperate foraging area in the Gulf of
California, Mexico. Herpetological Conservation
and Biology 1:81–86.
Semino, J.A., P. Zárate, M. Coyne, D.G. Foley, D.
Parker, B.N. Lyon, and P.H. Dutton. 2008. Post-
nesting migrations of Galápagos Green Turtles
Chelonia mydas in relation to oceanographic
conditions: integrating satellite telemetry with
remotely sensed ocean data. Endangered Species
Research 4:57–72.
Siow, K.T., and E.O. Moll. 1982. Status and conservation
of estuarine and sea turtles in West Malaysian waters.
Pp. 339–347 In Biology and Conservation of Sea
Turtles, Proceedings of the World Conference on
Sea Turtle Conservation. Bjorndal, K.A. (Ed.).
Smithsonian Institution Press, Washington D.C.,
Sosa-Nishizaki, O., E. Morales-Bojórquez, N. Nasby-
Lucas, E.C. Oñate-González, and M.L. Domeier.
2012. Problems with photo identication as a
method of estimating abundance of White Sharks,
Carcharodon carcharias: an example from
Guadalupe Island, Mexico. Pp. 393–404 In Global
Perspectives on the Biology and Life History of the
White Shark. Domeier, M.L. (Ed.). CRC Press, Boca
Raton, Florida, USA.
Speed, C.W., M.G. Meekan, and C.J.A. Bradshaw. 2007.
Spot the match wildlife photo-identication using
information theory. Frontiers in Zoology 4:2. DOI:
Stoddard, M.C., R.M. Kilner, and C. Town. 2014.
Pattern recognition algorithm reveals how birds
evolve individual egg pattern signatures. Nature
Communications 5:4117.
Su, C.M., C.T. Huang, and I.J. Cheng. 2015. Applying a
fast, eective and reliable photographic identication
system for Green Turtles in the waters near Luichiu
Island, Taiwan. Journal of Experimental Marine
Biology and Ecology 467:115–120.
Suhaila, J., S.M. Deni, W.Z.W. Zin, and A.A. Jemain.
2010. Trends in Peninsular Malaysia rainfall data
during the Southwest monsoon and Northeast
monsoon seasons: 1975–2004. Sains Malaysiana
Taquet, C., M. Taquet, T. Dempster, M. Soria, S.
Ciccione, D. Roos, and L. Dagorn. 2006. Foraging
of the Green Sea Turtle Chelonia mydas on seagrass
beds at Mayotte Island (Indian Ocean), determined
by acoustic transmitters. Marine Ecology Progress
Series 306:295–302.
Thompson, P.M., and H. Wheeler. 2008. Photo-ID-
based estimates of reproductive patterns in female
Harbor Seals. Marine Mammal Science 24:138–146.
Toda, T., T. Okashita, T. Maekawa, B.A.A.K. Alan,
M.K.M. Rajuddin, R. Nakajima, W. Chen, K.T.
Takahashi, B.H.R. Othman, and M. Terazaki.
2007. Community structures of coral reefs around
Peninsular Malaysia. Journal of Oceanography
Town, C., A. Marshall, and N. Sethasathien. 2013. Manta
Matcher: automated photographic identication of
manta rays using keypoing features. Ecology and
Evolution 3:1902–1914.
Valdés, Y.A., J.A. Ricardo, F.B. Trelles, and O.E.
Abad. 2014. First assay of photo-identication
in marine turtles’ nesting population. Revista de
Investigaciones Marinas 34:43–51.
van de Merwe, J.P., K. Ibrahim, S.Y. Lee, and J.M.
Whittier. 2009. Habitat use by Green Turtles
(Chelonia mydas) nesting in Peninsular Malaysia:
local and regional conservation implications.
Wildlife Research 36:637–645.
Wee, Y.C., and R. Subaraj. 2009. Citizen science and the
gathering of ornithological data in Singapore. Nature
in Singapore 2:27–30.
Waayers, D., D. Newsome, and D. Lee. 2008. Research
note observations of non-compliance behaviour by
tourists to a voluntary code of conduct: a pilot study
of turtle tourism in the Exmouth region, Western
Australia. Journal of Ecotourism 5:211–222.
Wells, R.S., and M.D. Scott. 1990. Estimating Bottlenose
Dolphin population parameters from individual
identication and capture-release techniques. Report
of the International Whaling Commission Special
Issue 12:407–415.
Long and Azmi.—Photo-ID to monitor sea turtles at Perhentian Islands of Malaysia.
Wetzelhuetter, C., A. Chelliah, and S.Y. Chen.
2014. Coastal ecosystem: coral reef environment
monitoring of coral reefs in the marine parks of
Terengganu and Tioman, Malaysia, using Reef
Check methods and the implementation into the
marine park management. Pp. 75–108 In Coastal
Ecosystems: Types, Sustainable Management and
Conservation Strategies. Masura, T. (Ed.). Nova
Science Publishers, Hauppauge, New York, USA.
Whiting, S.D., W. Murray, I. Macrae, R. Thorn, M.
Chongkin, and A.U. Koch. 2008. Non-migratory
breeding by isolated Green Sea Turtles (Chelonia
mydas) in the Indian Ocean: biological and
conservation implications. Naturwissenschaften
Whittaker, D., and R.L. Knight. 1998. Understanding
wildlife responses to humans. Wildlife Society
Bulletin 26:312–317.
Wibbels, T. 1999. Diagnosing the sex of sea turtles
in foraging habitats. Pp. 139–143 In Research and
Management Techniques for the Conservation of Sea
Turtles. Eckert, K.L., K.A. Bjorndal, and F.A. Abreu-
Grobois (Eds.). International Union for Conservation
of Nature Species Survival Commission Marine
Turtle Specialist Group Publication No. 4,
Washington, D.C., USA.
Williams, J.L., S.J. Pierce, M.M.P.B. Fuentes, and M.
Hamann. 2015. Eectiveness of recreational divers
for monitoring sea turtle populations. Endangered
Species Research 26:209–219.
Wilson, S.G., J.J. Polovina, B.S. Stewart, and M.G.
Meekan. 2006. Movements of Whale Sharks
(Rhincodon typus) tagged at Ningaloo Reef, Western
Australia. Marine Biology 148:1157–1166.
Würsig, B., and T.A. Jeerson. 1990. Methods of photo-
identication for small cetaceans. Report of the
International Whaling Commission Special Issue
Herpetological Conservation and Biology
  is currently pursuing her Ph.D. research on human-sea turtle
interactions in relation of local livelihoods at Universiti Malaysia Terengganu.
She received her B.A.Sc. in Conservation and Biodiversity Management from
Universiti Malaysia Terengganu and M.Sc. in Conservation and Biodiversity
from University of Exeter. Upon graduation, she was a Research Assistant for the
Green Turtle Program at Tortuguero, Costa Rica. After returning to Malaysia, she
was the Project Leader for the Perhentian Community and Conservation Project,
working with the local communities on environmental and educational outreach
activities to raise public awareness. Her passion for sea turtle conservation has led
her to set up the Perhentian Turtle Project. (Photographed by Björn Johansson).
   received his B.A.Sc. in Biodiversity Conservation and
Management from Universiti Malaysia Terengganu, Malaysia. During his
undergraduate study, he developed a deep passion towards environmental
conservation, especially for the marine environment. In 2014, he obtained his rst
eld experience with turtle conservation while undertaking a summer internship
with Ecoteer in Perhentian Islands, Malaysia. The following year, he helped
to set up the Perhentian Turtle Project with goals of learning and gaining more
experience in the eld. While he is still in his early twenties, he is keen to learn
about turtle conservation work and pursue his Masters degree in the near future.
(Photographed by Charlotte E. Babbs).
... For example, despite the widespread usage of toe clipping and subcutaneous implants (elastomer or tags) in anurans (Brow 1997), these methods are invasive and controversial because their use may be harm the frog, especially in arboreal species in which adhesive discs are essential to perch (e.g., Clarke 1972, May 2004, Funk et al. 2005. e oto rap ic dentification et od (PIM) is a useful, non-invasive marking alternative that can be used to distinguish individuals of species that possess characteristic features or natural markings Würsig 1977, Long andAzmi 2017). Many computerassisted systems (pattern recognition or photographic matching software) have been developed that enable researchers to process large numbers of photographic images in relatively short timeframes. ...
... Many computerassisted systems (pattern recognition or photographic matching software) have been developed that enable researchers to process large numbers of photographic images in relatively short timeframes. Thus, due to its bio o ica o istic financia and et ica advantages, PIM has been increasingly used in research studies of an array of taxa, such as insects and sea stars (Chim andTan 2012, Caci et al. 2013), sharks and rays (Tienhoven et al. 2007, Marshall andPierce 2012), marine and fres ater te eost fis es orreia et al. 2014, Dala-Corte et al. 2016, aquatic and terrestrial mammals (Kniest et al. 2010, Bolger et al. 2012, and turtles and lizards (Knox et al. 2013, Long andAzmi 2017), as well as toads and salamanders (Gamble et al. 2008, Caorsi et al. 2012. ...
Animal monitoring research involving mark-recapture techniques increasingly requires non-invasive methods of individual identification. The photographic identification method (PIM) is an excellent tool for this purpose and has been applied successfully to many taxa. However, the utility of PIM is a function of species-specific features that are judged suitable for a given target-species. Herein, the suitability of inguinal color patterns for photo identification of individuals of Pithecopus gonzagai are evaluated by comparing two widely used computer-assisted photographic matching programs (I³S and Wild.ID). Both programs accurately identified more than 70% of individuals in the top 20 potential matching photographs. Wild.ID was slightly better than I³S in matching efficiency and has a faster processing time. Thus, PIM is useful to identify individual P. gonzagai; however, before implementing the technique in animal-monitoring studies of other taxa, one must evaluate the suitability PIM for the target species and calibrate the relative efficiency of the software programs in identifying individuals.
... Current population genetics approaches require invasive tissue sampling or blood draws -for sea turtles this frequently is conducted during nesting, physically interacting with nesting females and probably inducing additional stress -alternatively one egg needs to be sacrificed per clutch to obtain maternal DNA without direct tissue/blood sampling (Adams et al., 2019;Calmanovici et al., 2018;Gadagkar et al., 2005;Gatto et al., 2018;Jensen et al., 2019;Komoroske et al., 2018;Long & Azmi, 2017;Shamblin et al., 2011). ...
Elusive aquatic wildlife, such as endangered sea turtles, are difficult to monitor and conserve. As novel molecular and genetic technologies develop, it is possible to adapt and optimize them for wildlife conservation. One such technology is environmental (e)DNA – the detection of DNA shed from organisms into their surrounding environments. We developed species‐specific green (Chelonia mydas) and loggerhead (Caretta caretta) sea turtle probe‐based qPCR assays, which can detect and quantify sea turtle eDNA in controlled (captive tank water and sand samples) and free ranging (oceanic water samples and nesting beach sand) settings. eDNA detection complemented traditional in‐water sea turtle monitoring by enabling detection even when turtles were not visually observed. Furthermore, we report that high throughput shotgun sequencing of eDNA sand samples enabled sea turtle population genetic studies and pathogen monitoring, demonstrating that non‐invasive eDNA techniques are viable and efficient alternatives to biological sampling (e.g. biopsies and blood draws). Genetic information was obtained from sand many hours after nesting events, without having to observe or interact with the target individual. This greatly reduces the sampling stress experienced by nesting mothers and emerging hatchlings, and avoids sacrificing viable eggs for genetic analysis. The detection of pathogens from sand indicates significant potential for increased wildlife disease monitoring capacity and viral variant surveillance. Together, these results demonstrate the potential of eDNA approaches to ultimately help understand and conserve threatened species such as sea turtles.
... A variety of photo-based pattern detection and matching programs developed using information theory can identify and discriminate biological patterns in various contexts. For example, conservationists have used matching algorithms to identify individual newts [24], manta rays [25], whale sharks [26] and sea turtles [27] based on their distinctive skin and shell patterns for re-sighting purposes. Information theory-based methods lack the subjectivity of earlier methods of pattern detection and recognition and take less time to complete [25,26]. ...
Full-text available
Maternal signatures are present in the eggs of some birds, but quantifying interclutch variability within populations remains challenging. Maternal assignment of eggs with distinctive appearances could be used to non-invasively identify renesting females, including hens returning among years, as well as to identify cases of conspecific brood parasitism. We explored whether King Rail ( Rallus elegans ) eggs with shared maternity could be matched based on eggshell pattern. We used NaturePatternMatch (NPM) software to match egg images taken in the field in conjunction with spatial and temporal data on nests. Since we had only a small number of marked breeders, we analyzed similar clutch images from a study of Eurasian Common Moorhens ( Gallinula chloropus chloropus ) with color-banded breeders for which parentage at many nests had been verified genetically to validate the method. We ran 66 King Rail clutches ( n = 338 eggs) and 58 Common Moorhen clutches ( n = 364 eggs) through NPM. We performed non-metric multidimensional scaling and permutational analysis of variance using the best egg match output from NPM. We also explored whether eggs could be grouped by clutch using a combination of egg dimensions and pattern data derived from NPM using linear discriminant analyses. We then scrutinized specific matches returned by NPM for King Rail eggs to determine whether multiple matches between the same clutches might reveal maternity among nests and inform our understanding of female laying behavior. To do this, we ran separate NPM analyses for clutches photographed over several years from two spatially distant parts of the site. With these narrower datasets, we were able to identify four instances where hens likely returned to breed among years, four likely cases of conspecific brood parasitism, and a within-season re-nesting attempt. Thus, the matching output was helpful in identifying congruent egg patterns among clutches when used in conjunction with spatial and temporal data, revealing previously unrecognized site fidelity, within-season movements, and reproductive interference by breeding females. Egg pattern data in combination with nest mapping can be used to inform our understanding of female reproductive effort, success, and longevity in King Rails. These methods may also be applied to other secretive birds and species of conservation concern.
... Photography and videography can provide important information about animal behavior, natural history, and populations (Hartley 1948, Davies et al. 2012, Pimm et al. 2015, Long and Azmi 2017, Weise et al. 2017, Caven et al. 2019a, Hawkes et al. 2020. Use of photography and videograpy to observe foraging Whooping Cranes enabled us to document species-specific diet items being consumed without interrupting the cranes' natural foraging activities. ...
The Aransas-Wood Buffalo population of Whooping Cranes (Grus americana) migrates approximately 4000 km through the central Great Plains biannually, between their breeding and wintering grounds. Whooping Cranes depend on stopover sites to provide secure resting locations and the caloric resources necessary to complete their migration, such as the USFWS-designated critical habitat area in the Central Platte River Valley (CPRV) of Nebraska. This area includes braided river habitat characterized by low-elevation and submerged sandbars, which provide important roosting and foraging opportunities for migrating Whooping Cranes. We used long-range photography, videography, and behavioral scan sampling to document forage items consumed by Whooping Cranes during an 11-day stopover in this area during the fall of 2019. We identified 3 adult-plumage Whooping Cranes and 1 colt consuming 16 individual vertebrates of at least 6 different species during the stopover. In total, we documented Whooping Cranes consuming 7 Channel Catfish (Ictalurus punctatus), 5 ray-finned fish (Actinopterygii), 1 sunfish (Centrarchidae), 1 carp/minnow relative (Cypriniformes), 1 perch relative (Percidae), and 1 Leopard Frog relative (Lithobates sp.). We estimated prey item lengths using the average exposed culmen measurements for adult Whooping Cranes and approximated their nutritional value using log-transformed length–weight regression equations with taxon-specific intercepts and slopes from secondary data sources. We estimated that aquatic vertebrate forage made up a significant portion of Whooping Crane daily energy requirements and provided substantial amounts of calcium, phosphorus, and protein not present at high levels in waste grains also consumed during migration. Additionally, we documented territorial behavior by adult Whooping Cranes during migration and evidence of adults teaching their colt to forage. Our study demonstrates the utility of photography and videography to natural history research and indicates that aquatic vertebrates may be a relatively regular part of Whooping Crane diet in the CPRV. <>
... Photo-ID with photographs provided by citizen scientists has recently been used for studies of sea turtle ecology (e.g. Long and Azmi, 2017;Dunbar et al., 2021b). However, data collection by citizen scientists may be inaccurate or unreliable owing to lack of knowledge, experience, and motivation in scientific research, making analysis difficult (Cohn, 2008). ...
The accurate identification of wildlife is necessary for biological research and population monitoring. In sea turtles, artificial tags have commonly been used for individual identification, but tagging has low long-term reliability because of inherent problems such as tag loss and the stress caused by invasive tagging. Photographic identification (Photo-ID) has been increasingly used as an alternative to tagging. Mostly, facial scute patterns have been used as Photo-ID means for sea turtles; however, for Photo-ID of turtles on nesting beaches, there is a risk of disturbing individuals when attempting to access the head region and capture the entire face. Thus, here, we focused on the posterior part of the carapace (fifth vertebral scute), which can be photographed more easily and less invasively, as a useful natural marker for the identification of nesting green turtles (Chelonia mydas). To establish the utility of the carapace for individual identification, we used photographs (167 images from 77 individuals) collected in a nesting survey conducted over 28 years at Ishigaki Island, Japan, and verified the long-term identifiability of carapace. We initially matched individual turtle images with the HotSpotter program to assess the automated recognition of images, and then conducted a blind test to visually validate the results of automated recognition. High matching accuracy was achieved, especially within the same nesting season (98.1%; 52 out of 53 queries), and logistic regression indicated that the matching accuracy was over 70% when the photographing interval was ≤4 years, which is longer than the mean remigration interval for nesting green turtles on Ishigaki Island. In the blind test of visual identification, 95.4% of the image pairs were correctly judged (as those of the same individual or different individuals). These results show that a query image can be successfully identified in the database even if automated matching to the same individual fails or if the query is taken from a newly recruited individual. This study shows the long-term identifiability of nesting green turtles using carapace photographs and proposes a practical Photo-ID method that is conducive to citizen science.
... For example, the online tools of citizen science, such as social media, gaming, monitoring apps and centralised websites for volunteer recruitment (Newman et al., 2012), do not require participants to be physically engaged with experts. Participants are merely required to follow instructions, collect the required data and load them onto websites, social media or other online platforms (Long and Azmi, 2017;Wilson et al., 2015;Winterton et al., 2012). Although such one-way interaction using online tools results in limited physical engagement between experts and participants, in some circumstances, experts regard internet communication to be as important for the participants as face-to-face interaction (Bela et al., 2016). ...
Community-based monitoring is increasingly recognised as one solution to sustainable environmental management. However, the development of community-based monitoring has led to confusion or misconceptions regarding other similar initiatives. Through a review of the characteristics and synthesising criteria of effective community-based monitoring, this article addresses how to distinguish community-based monitoring from other forms of community engagement research. A review of relevant community-based monitoring literature identifies the characteristics of and knowledge gaps in procedures and governance structures. Additionally, evidence of common benefits, challenges and lessons learned for successful community-based monitoring are deliberated. As an outcome of the review, the article synthesises a set of community-based monitoring criteria as follows: (1) efficacy of initiatives, (2) technicality aspects, (3) feedback mechanisms and (4) sustainability. These synthesised criteria will be instrumental in designing customised community-based monitoring initiatives for environmental sustainability.
... Similarly, Calmanovici et al. (2018) found that photos taken of turtles underwater at different angles, distances, and in different light conditions reduced the accuracy of matches using I 3 S Pattern. Likewise, Long and Azmi (2017) were able to more successfully identify individual turtles through NPM if photos were taken at horizontal and vertical angles <45 • from where face scales were visible. In a more recent study, Steinmetz et al. (2018) also found photographs taken at high vertical or horizontal angles usually resulted in poorer potential matches among photographs of nesting hawksbill (Eretmochelys imbricata) turtles on Mahé Island in the Seychelles. ...
Photo identification (PID) in animal studies has been a widely used method for identifying individuals of many species based on unique natural markings and patterns. The use of PID has facilitated investigations in which residency, home ranges, and growth rates have been assessed. However, many PID studies in the past have relied heavily on manual photo matching. More recently, computer-assisted PID programs have been used to identify individuals of different sea turtle species, and reduced time investment in identifying individuals within specific populations. Still, some computer-based PID programs require significant time investment in ensuring photos are captured at consistent angles and lighting conditions, pre-processing image manipulations, and post-processing manual matching confirmation of potential matches provided by the program. For PID to be an effective, time and money saving mechanism for wildlife research and conservation, these common drawbacks need to be addressed with a computer-assisted PID program that reduces manipulation and time investment burden, and consistently provides accurate and reliable results. In this study, we evaluated the accuracy of matching individual face images using the HotSpotter (HS) PID program by building a database of 2136 images of hawksbill (Eretmochelys imbricata) turtles, then querying the database with 158 new images to find matches for individual turtles. Overall, we found that with almost no pre-processing manipulation, and with images from highly variable underwater conditions, qualities, and angles, HS correctly matched individuals in the first choice 80% of the time, increasing to 91% in the first six choices. When assessing in-water images only, accuracy for matching increased from 84% in the first choice, to 94% by the sixth choice. We suggest that the integration of HS technology into a global, web-based PID system will increase the ability to remotely identify individual marine organisms on a global scale, and improve usability for community scientists who may have little to no technical training.
... It is due to the colour of the carapace may change. The Scale Invariant Feature Transform (SIFT) and Nature Pattern Match (NPM) were used by [10] that allows the identification of sea turtles only one side of the face. ...
Up to now, identification of sea turtle species mainly for tracking the population usually relied on flipper tags or through other physical markers. However, this approach is not practical due to the missing tags over some period. Due to this matter, we propose a photo identification system of the individual sea turtle based on the convolutional neural network (CNN) using a pre-trained AlexNet CNN and error-correcting output codes (ECOC) SVM. Experiments were performed on 300 images obtained from Biodiversity Research Center, Academia Sinica, Taiwan. Using Alexnet and ECOC SVM, the overall accuracy achieved is 62.9%. The results indicate that features obtained from the CNN are capable of identifying photo of sea turtles.
Full-text available
Sea turtles spend most of their life cycle in foraging grounds. Nevertheless, given the logistical ease, bulk of studies are carried out in nesting grounds. One strategy to gather information in foraging grounds is the use of cost-effective and non-invasive techniques that allow public participation. The present study aimed to use photographic-identification to investigate the spatio-temporal distribution of Chelonia mydas and Eretmochelys imbricata . Furthermore, we describe the occurrence of fibropapillomatosis. This work was carried out at subtropical rocky reefs of the Brazilian coast in Arraial do Cabo (22°57’S, 42°01’W), within a sustainable conservation unit. The images were obtained through social media screening, citizen science and intentional capture. A total of 641 photos (between 2006 and 2021) and 19 diving forms (between November 2019 and March 2020) were obtained. All diving forms presented at least one turtle. The photo-id identified 174 individuals of C. mydas , with 45 resighted individuals. E. imbricata had 32 individuals identified, and 7 individuals resighted. The median interval between the first and last individual sighting was 1.7 years for C. mydas and 2.4 years for E. imbricata . Fibropapillomatosis was only observed in C. mydas , with a prevalence of 13.99% and regression in 2 individuals (10.00%). The results indicated Arraial do Cabo as an important development area for sea turtles with resident individuals showing fidelity for at least 6 years. As the region accumulates anthropogenic impacts, it is critical to propose significant management measures to protect sea turtles from continuous and synergistic impacts, mainly on resident individuals.
Coastal monitoring and management can be enhanced by citizen science (CS), especially because Information and Communications Technology (ICT) enables remote public engagement in CS. To date, this type of engagement would benefit from a thorough investigation to highlight its supportive role to coastal CS, identify its unique challenges, and propose solutions and research avenues to sustain its development. A systematic scoping review and qualitative analysis of scientific papers (N = 53) described the state of the art in remote public engagement in coastal CS. The analysis revealed unique advantages of remote engagement in coastal CS, including flexibility, social inclusivity, and organised communication management. Challenges, mainly associated with technology, could be solved by simplifying interfaces to streamline communication and participation in CS. The review identified research gaps and confirmed the potentially positive contribution of remote engagement in coastal CS, which could enhance coastal monitoring, management, public participation, and stewardship.
Full-text available
The often opportunistic nature of biological recording via citizen science leads to taxonomic, spatial and temporal biases which add uncertainty to biodiversity estimates. However, such biases may also give valuable insight into volunteers’ recording behaviour. Using Greater London as a case-study we examined the composition of three citizen science datasets – from Greenspace Information for Greater London CIC, iSpot and iRecord - with respect to recorder contribution and spatial and taxonomic biases, i.e. when, where and what volunteers record. We found most volunteers contributed few records and were active for just one day. Each dataset had its own taxonomic and spatial signature suggesting that volunteers’ personal recording preferences may attract them towards particular schemes. There were also patterns across datasets: species’ abundance and ease of identification were positively associated with number of records, as was plant height. We found clear hotspots of recording activity, the 10 most popular sites containing open water. We note that biases are accrued as part of the recording process (e.g. species’ detectability) as well as from volunteer preferences. An increased understanding of volunteer behaviour gained from analysing the composition of records could thus enhance the fit between volunteers’ interests and the needs of scientific projects.
Full-text available
Following centuries of exploitation, basking sharks (Cetorhinus maximus) are considered by IUCN as Endangered in the Northeast Atlantic, where they have now been substantially protected for over two decades. However, the present size of this population remains unknown. We investigated the use of photo-identification of individuals' dorsal fins, combined with mark-recapture methodology, to investigate the size of populations of basking shark within the west coast of Scotland. From a total of 921 encounters photographed between 2004 and 2011, 710 sharks were found to be individually identifiable based on dor-sal fin damage and natural features. Of these, only 41 individuals were re-sighted, most commonly both within days of, and close to the site of, the initial encounter. A smaller number were re-sighted after longer periods of up to two years. A comparison of the distinguishing features of individuals on first recording and subsequent re-sighting showed that in almost all cases these features remained little changed, suggesting the low re-sighting rate was not due to a loss of distinguishing features. Because of the low number of re-sighting we were not able to produce reliable estimates for the long-term regional population. However , for one 50 km diameter study area between the islands of Mull, Coll and Tiree, we were able to generate closed-population estimates for 6–9 day periods in 2010 of 985 (95% CI = 494–1683), and in 2011 of 201 (95% CI = 143–340). For the same 2011 period an open-population model generated a similar estimate of 213 (95% CI = 111–317). Otherwise the low rate and temporal patterning of re-sightings support the view that such local basking shark populations are temporary, dynamic groupings of individuals drawn from a much larger regional population than previously supposed. The study demonstrated the feasibility and limitations of photo-identification as a non-invasive technique for identifying individual basking sharks.
Full-text available
The Perhentian Island archipelago, Terengganu in east coast of Peninsular Malaysia is relatively in isolation from the mainland and poor accessibility has led to the paucity of observation on marine animals and plants e.g., seagrasses for the archipelago. Field surveys on seagrasses around Perhentian Besar Island and Perhentian Kecil Island was conducted for occurrence and distribution of seagrasses. Plants were sampled by direct collection during low tide or by snorkelling and SCUBA diving over four years (2007-2010) as part of the inventory of marine plants. Five species of seagrasses comprising the rare larger-bodied, Enhalus acoroides (L.f.) Royle and the small-bodied Halophila decipiens Ostenfeld, H. minor (Zoll.) den Hartog, H. ovalis (R. Brown) Hooker and Halodule pinifolia (Miki) den Hartog were recorded around the two islands. Enhalus acoroides grew in isolated patches at shallow depths of 0.5-1.5 m of the littoral zone usually exposed during the low spring tides at Tanjung Batu Lepir. Halophila decipiens, H. minor, H. ovalis and Halodule pinifolia grew as monospecific or mixed populations with no distinct zonation distributed at deeper depths of 4.6-12.0 m of the sublittoral zones at Channel in front of Department of Fisheries, Channel in front of Coco Hut and Kg. Tanjung Aur. Seagrass characterized by small-bodied with flaccid leaves e.g., Halophila decipiens, H. minor, H. ovalis and Halodule pinifolia are uniquely equipped for survival under extreme physical oceanographic due to seasonal north-east monsoons prevailing at Perhentian islands, making them less vulnerable to wave action and become detached. The morphological descriptions of the five species are also described.
This paper presents a method for extracting distinctive invariant features from images that can be used to perform reliable matching between different views of an object or scene. The features are invariant to image scale and rotation, and are shown to provide robust matching across a substantial range of affine distortion, change in 3D viewpoint, addition of noise, and change in illumination. The features are highly distinctive, in the sense that a single feature can be correctly matched with high probability against a large database of features from many images. This paper also describes an approach to using these features for object recognition. The recognition proceeds by matching individual features to a database of features from known objects using a fast nearest-neighbor algorithm, followed by a Hough transform to identify clusters belonging to a single object, and finally performing verification through least-squares solution for consistent pose parameters. This approach to recognition can robustly identify objects among clutter and occlusion while achieving near real-time performance.
The hawksbill sea turtle (Eretmochelys imbricata) is a critically endangered species encountered by recreational divers in marine protected areas (MPAs) circumtropically. Few studies, however, have examined the impacts of recreational diving on hawksbill behaviours. In 2014, we collected turtle sightings surveys and dive logs from 14 dive operations, and conducted in-water observations of 61 juvenile hawksbill turtles in Roatán, Honduras, to determine if differences in dive site use and diver behaviours affected sea turtle behaviours in the Roatán Marine Park. Sightings distributions did not vary with diving pressure during an 82-day study period. We found the amount of time turtles spent eating, investigating and breathing decreased when approached by divers. Our results suggest diver interactions may negatively impact sea turtle behaviours, however it is unknown if recreational diving has a cumulative effect on turtles over time. We recommend that MPA managers should implement monitoring programmes that assess the impacts of tourism on natural resources. We have established monitoring of hawksbills as representatives of the marine habitat in an MPA, which has the potential to be heavily impacted by dive tourism, and provide recommendations for continued monitoring of the resource. This is an Original Manuscript of an article published by Taylor & Francis in the Journal of Sustainable Tourism on May 9, 2016, available at
Between 2008 and 2013 the status of coral reefs was examined using Reef Check survey methods on 19 selected reefs along the eastern coast of the Malaysian Peninsula, around Perhentian, Redang and Tioman Islands. The goal of the study was to determine the status of these reefs and to monitor continuous change on the reef caused by human and non-human factors. Data on indicator fish, indicator invertebrate and indicator substrate as well as environment, socio-economic and human impacts was collected. The assessment of this data demonstrates that the two most obvious impacts were warm water bleaching and sewage pollution. Overall the study shows that live coral cover did not change significantly over the past six years. The results of these surveys were implemented into the Marine Parks management. First the survey technique was introduced to Marine Park personnel for long term monitoring. Priority and resilience areas where mapped out and established. Furthermore, a bleaching response plan was designed for Malaysia after the 2010 bleaching event and was implemented jointly with park management. In addition, pollution caused by growing ecotourism was recognized as a major impact to the coral reef status and measures like waste control were introduced and successfully established. Reef rehabilitation efforts in the form of coral transplanting were also undertaken at sites where the natural reef had suffered damage due to human and natural impacts. This reef rehabilitation project was used as a platform to involve tourism operators, local villagers and park managers in hands-on conservation in order to promote a better understanding of the importance of coral reefs. It was demonstrated that coral transplanting can be successfully used in reef rehabilitation.
The diet of immature Green Turtles, Chelonia mydas, from the Moreton Banks and Flathead Gutter sites of Moreton Bay, included the available seagrass species and some of the available species of algae. Some animal material was ingested, as were fruits of the Grey Mangrove, Avicennia marina. Volumetrically, the seagrass Halophila ovalis and the red algae Gracilaria cylindrica and Hypnea spinella were the most important components of the diet.