You are on page 1of 15

Environ Monit Assess (2019) 191:633

https://doi.org/10.1007/s10661-019-7808-2

Development and application of a low-cost rapid


assessment system for coastal benthic habitats
Caryl S. Benjamin & Patrick Lawrence P. Cadeliña &
Aletta T. Yñiguez & Cesar L. Villanoy

Received: 12 March 2019 / Accepted: 8 September 2019


# Springer Nature Switzerland AG 2019

Abstract Information on community structure and spa- no significant differences in estimates of percent coral
tial distribution of benthic habitats are important in cover and diversity of benthic functional groups for
marine conservation and management. Coral areas, of- majority of sites while estimates of community structure
ten monitored by trained scuba divers, are of importance were very similar. ARAICoBeH is cheaper and requires
for the ecosystem goods and services they provide. The less time to gather the same amount of data as that of the
data gathered may, however, be limited due to cost and photo transect method. The spatial distribution of coast-
time constraints. These restrictions and the continuing al benthic habitats and community structure of coral-
decline in coral reef health contribute to the need to dominated areas in El Nido, Palawan, Philippines, are
develop rapid methods to efficiently document the dis- included to illustrate the application of the method.
tribution and status of coral reefs. A Rapid Assessment
Instrument for Coastal Benthic Habitats (ARAICoBeH) Keywords Mapping . Coral assessment . Coral
System was developed to enable large-scale but low- monitoring . Rapid assessment . Community structure
cost coastal benthic habitat characterization and map-
ping without compromising accuracy. Inquiries on com-
munity structure and status in coral-dominated areas are
Introduction
also possible with the data collected using the method.
The instrument, as well as the data processing and
Corals are important coastal resources because of the
mapping algorithm are described in detail in this paper.
ecosystem goods and services that they provide
A comparative study was conducted between
(Connell 1978; Moberg and Folke 1999). According to
ARAICoBeH and the frequently used underwater photo
Burke et al. (2011), approximately 850 million people or
transect method. Variables compared were percent coral
one-eighth of the world’s population live within 100 km
cover, functional group diversity, community structure
of a coral reef and most probably derive some benefit
as well as time and monetary requirements. There were
from these reefs. Majority of these people depend on
reefs for their sustenance and livelihood and come from
developing countries and island nations. Unfortunately,
C. S. Benjamin : P. L. P. Cadeliña (*) : A. T. Yñiguez : there is a global trend of decline in coral reef health
C. L. Villanoy
Marine Science Institute, University of the Philippines, Diliman,
(Knowlton 2001; McClanahan 2002; Pandolfi et al.
1101 Quezon City, Philippines 2003; Wilkinson 2008). Coral reefs are vulnerable to
e-mail: plcadelina@msi.upd.edu.ph local sources of stress such as terrestrial runoff
C. S. Benjamin
(Fabricius 2005) and eutrophication (Tomascik and
Chair of Ecoclimatology, Technische Universität München, Sander 1985; Szmant 2002) as well as global stressors
Hans-Carl-von-Carlowitz-Platz 2, 85354 Freising, Germany such as global warming and ocean acidification (Hoegh-
633 Page 2 of 15 Environ Monit Assess (2019) 191:633

Guldberg et al. 2007; Pandolfi et al. 2011). Local activ- Transect methods are often preferred in studies
ities alone like overfishing and destructive fishing, assessing environmental impacts in medium scales
coastal development, and watershed-based or marine- while quadrats are suited for fine-scale surveys
based pollution threaten more than 60% of the world’s (Beenaerts and Berghe 2007). These methods give good
coral reefs (Burke et al. 2011). Documentation of the estimates of percent coral cover and diversity but require
current state of coral reefs is therefore a research priority extensive time underwater as well as expertise in both
to provide baseline information, monitor changes due to scuba diving and identification of benthic components
threats and disturbances as well as recoveries (Gardner (Hill and Wilkinson 2004).
et al. 2005; Bruno and Selig 2007; Baker et al. 2008; Advancements in technology have led to the exten-
Wilkinson 2008; Wilkinson and Souter 2008). sive use of digital video surveys and underwater pho-
Resource maps are excellent tools for coastal re- tography to collect photo quadrats (PQs) as an alterna-
source management and conservation because large tive to in situ methods (Lam et al. 2006; Leujak and
amounts of information (e.g., spatial distribution and Ormond 2007). PQs are collected along transects and
status of resources) can be reduced into a simple visual percent cover of the different benthic components are
representation (Tortell 1992). Coral reef habitat maps estimated from points superimposed over the PQs using
have been produced using remote sensing for resource an image analysis software (English et al. 1997; Houk
assessment (Andréfouët and Riegl 2004; Gilbert et al. and Van Woesik 2006; Kohler and Gill 2006). Advan-
2006) and conservation (Aswani and Lauer 2006). tages over the in situ methods include reduction in time
However, the accuracy and level of detail derived from spent underwater and creation of permanent records
satellite and airborne images are limited to geomorpho- (Leujak and Ormond 2007). Previous studies have also
logical classes and coarse habitat discrimination (e.g., established the superiority of these methods in terms of
coral versus non-coral) because there is high variability efficiency and precision in detecting differences in coral
in accurately discerning finer taxonomic habitat classes communities between sites and over time (Weinberg
and density estimates using supervised classification 1981; Murdoch and Aronson 1999; Ninio et al. 2003;
(Green et al. 1996; Mumby et al. 1997; Andréfouët Brown et al. 2004; Lam et al. 2006; Leujak and Ormond
et al. 2006). There is a need for a method to produce 2007).
maps with more detailed information to provide insights These scuba-based methods, however, have limita-
on indicators of reef status (Scopélitis et al. 2010). tions in terms of underwater time, sensitivity to weather
There is a variety of existing coral survey methods conditions as well as access to scuba facilities. These are
which require various levels of expertise as well as also often used at a narrow range of depths and are
resources (Dethier et al. 1993; Beenaerts and Berghe therefore limited to only a fraction of a benthic zone.
2005; Lam et al. 2006; Leujak and Ormond 2007; Jokiel To address these limitations, methodologies using re-
et al. 2015). It is suggested that choice of method should motely operated vehicles (ROVs), towed cameras, or
be based on scale, level of detail required as well as drop cameras to remotely collect photos or videos have
availability of resources. Methods to gather fine-scale been developed (Mitchell and Coggan 2007; Grizzle
and detailed data are ideal for scientific research while et al. 2008). All three have both advantages and disad-
large-scale but less detailed data collection are more vantages. ROVs have been widely used for prolonged
practical for management decisions (Hill and deep reef surveys (Williams and Mahon 2004; Lam
Wilkinson 2004). et al. 2006; Lirman et al. 2007), but these can be bulky,
Earlier methods for coral census required in situ data difficult to deploy and control, and are expensive.
acquisition. The manta tow method involved a snorkeler Towed cameras are usually housed in sledges dragged
being towed by a boat running at constant speeds. across the seafloor and are therefore not ideal for hard
Estimates of benthic cover were recorded during regular substrates and have the tendency to damage fragile
stops. The method can be used to survey large areas in a benthic species (Service and Golding 2001). Converse-
short period of time but has a tendency to underestimate ly, drop cameras are convenient, affordable, and can be
abundance and is ideally combined with benthic moni- deployed in areas where towed and ROV systems can-
toring methods involving scuba diving (Hill and not be deployed, such as rocky or high-energy environ-
Wilkinson 2004). Other commonly used underwater ments (e.g., Cole et al. 2001). All three can be used to
survey methods are the transect and quadrat methods. survey large spatial scales. However, most studies
Environ Monit Assess (2019) 191:633 Page 3 of 15 633

utilizing these methods focus either on visualization of and a makeshift fin that acts as a rudder is attached to a
sublittoral communities (Jones et al. 2000; Bates et al. half meter metal pole (Fig. 1a). A 3-kg lead weight is
2004) or on ground-truthing data in acoustic studies attached to the pole with a sturdy rope at 0.75 m from
(Beaman and Harris 2005; Roberts et al. 2005). Only a the camera. The instrument is lowered and towed from
few studies utilize remote photo/videography for habitat the boat using a calibrated rope (Fig. 1b). Depth and
assessment in shallow areas (Schneider et al. 1987; location are continuously measured per second and are
Dunbabin et al. 2005; Pacunski et al. 2008; Walcott recorded by a single beam echo sounder transducer
et al. 2014). (Garmin Airmar Tilted™ Element, and P19–20°) and a
A Rapid Assessment Instrument for Coastal Benthic GPS/transducer viewer (GPSMap 421s), respectively.
Habitats (ARAICoBeH) is a drop camera system that This GPS/transducer viewer is powered by a portable
was developed as an alternative to scuba-based coral lithium 33,000-mAh power bank. A handheld GPS
reef and shallow-water benthic habitat survey methods. (optional) is used to easily mark the starting point of
Cost of the system is lower than ROVs because it takes each transect. The GPS attached to the transducer may
advantage of the low cost of action cameras, which are also be used for this purpose.
compact, rugged, and waterproof. Unlike scuba-based Sampling using the ARAICoBeH System can be
methods, ARAICoBeH System includes positional data done using a small boat manned by three individuals.
(latitude and longitude) for each PQ collected in the One person takes note of the start and end time for each
field. The spatial distribution and status of benthic hab- transect. The same person determines the starting points
itats (e.g., percent live coral cover) can therefore be of transects based on distances between transects mea-
generated as an output. The method significantly re- sured using the GPS. The second person deploys the
duced the time necessary to survey larger areas due to instrument according to the current depth less 1 m to
the elimination of scuba diving and the development of prevent disturbance and damage of benthic communi-
an algorithm to automate time-intensive and repetitive ties. The third person is responsible for directing the
post-survey processing tasks. course of the boat and for keeping the speed at a mini-
The objectives of this study are to develop, test, and mum (0.5 knots) to optimize quality of pictures. Each
apply a low-cost rapid monitoring system to assess transect extends from a depth of 3 to 15 m and is
coastal benthic habitats. This is an end-to-end system perpendicular to the shore.
from the instrument to the data processing and mapping The camera time should be synchronized with the
algorithm. A comparative study between the GPS time. The GoPro Hero 3 Silver camera displays the
ARAICoBeH System and the frequently used underwa- current time only as hours and minutes, but these values
ter photo transect method in terms of estimates of coral are corrected to achieve accuracy to the second on the
reef health as well as time and cost requirements are time a picture was taken. This is done by taking a few
presented. The spatial distribution of coastal benthic pictures of the GPS time at the start of each sampling
habitats and community structure of coral-dominated day. The correction of the difference in seconds between
areas in El Nido, Palawan, Philippines, are included to the GPS time and the time the picture was taken is
illustrate the application of the method. included in the data processing script. Data collected
during a field survey include pictures or PQs, start and
end times for each transect, GPS tracks, and echo sound-
Materials and methods er depths.

ARAICoBeH system Data processing algorithm

Instrument design and operation Thousands of pictures of benthic habitats or PQs can be
collected within a few days of field sampling using
ARAICoBeH System is composed of three compo- ARAICoBeH. Selection of photos, geotagging, picture
nents: underwater photography, acquisition of location file organization, data management, and finally map-
information, and depth monitoring. Pictures are taken ping are simplified with the development of an algo-
underwater at a 2-s interval using an underwater action rithm written in the R language (R Core Team 2015) to
camera (GoPro Hero 3 Silver with waterproof housing), automate these tasks.
633 Page 4 of 15 Environ Monit Assess (2019) 191:633

Fig. 1 a ARAICoBeH
instrument, b during deployment
(a) (b)
showing its components: (1) un-
derwater camera attached to a (2) 6
20-cm metal pole with (3) acrylic
1
fin to serve as a rudder. (4) A 25-
cm rope attaches the pole to (5) a
3-kg weight which acts as stabi-
lizer. All these are tethered to the
boat via (6) a calibrated 20-m
mountaineering rope

4
3

Management and processing of a large quantity the Coral Point Count with Excel Extension (CPCe)
of picture files often require long periods of time software (Kohler and Gill 2006). Stratified random
and/or computing resources. To address this, the method is employed wherein each PQ is divided into
algorithm works with the picture information and five rows and two columns, with a single point random-
not the actual picture files during the initial stages ly overlaid in each cell. This method is used in order to
of data processing. The Exchangeable Image File reduce the chance of points clumping in each image
Format (EXIF) information from all the pictures (Kohler and Gill 2006). The benthic component under
collected from the field, specifically the filename each point is then identified manually by an observer
as well as date and time taken, is first extracted using a key (Table 1) modified from English et al.
and saved into a single data frame. The total space (1997). The key focused on coral life forms (e.g.,
and computer power, required to process this data Acropora branching, coral massive), other benthic func-
frame, is lower by several magnitudes as compared tional groups (i.e., other invertebrates, macroalgae,
to that of thousands of picture files. Geographic seagrasses), and abiotic factors (i.e., sand, rubble).
coordinates are added to each picture information CPCe generates the results in several spreadsheets
by matching the time the picture was taken and but the merging into a single file is automated in the
time recorded from the GPS. A subset of pictures data-processing algorithm. Geographic coordinates are
to be used for analysis and mapping is then se- then reattached to each quantified PQ. Some PQs may
lected according to a user-specified spatial interval. be rejected and excluded during image analysis due to
An interval of 2 m measured from the geographic image quality. A complete datasheet with geographic
coordinates associated with each picture was used coordinates and cover of biotic and abiotic components
in this study. The selected pictures are then copied is generated for further statistical analysis. The same
to a new folder. A map of the locations of these data is used to generate maps to visualize the spatial
pictures may be generated to visualize extent of variability of live hard coral cover. The relative compo-
survey. sition in terms of major benthic categories in different
Image analysis to quantify the different components areas of the surveyed site may also be compared through
of the benthic ecosystem to percent cover is done using pie charts. The data processing algorithm is
Environ Monit Assess (2019) 191:633 Page 5 of 15 633

Table 1 Identification key of functional groups modified from Comparison of ARAICoBeH to underwater photo
English et al. (1997) used to quantify benthic components using
transect method
the software Coral Point Count with Excel extension (CPCe,
Kohler and Gill 2006)
Estimates of reef health
Key Functional/abiotic group

ACB Acropora branching


The comparative study was undertaken in three different
locations in the Philippines (Fig. 2). These were among
ACD Acropora digitate
several coral reef areas assessed under research pro-
ACE Acropora encrusting
grams funded by the Philippine government: The Na-
ACS Acropora branching
tional Assessment of Coral Reef Environments
ACT Acropora tabulate
(NACRE) program of the Department of Science and
CB Coral branching
Technology and the Coral Reef Visualization (CoRVA)
CE Coral encrusting
program of the Department of Environment and Natural
CF Coral foliose
Resources. Field surveys were conducted in January,
CM Coral massive
March, and June 2015 for VIP, Sablayan, and El Nido,
CMR Coral mushroom
respectively.
CS Coral submassive Following the sampling strategy for ecological field
CHL Blue coral methods by van Woesik et al. (2009) for the underwater
CME Fire coral photo transect, three sites were chosen per location with
CTU Organ-pipe coral two stations per site. Figure 2 shows the three sites per
SC Soft coral location: VIP (13.773° N, 120.875° E)—site 1: Ligpo,
AN Anemones site 2: Dive and Trek, site 3: Twin Rocks; Sablayan
ASC Ascidians (12.827 N, 120.767° E)—site 1: North Pandan Island,
GOR Gorgonians site 2: South Pandan Island, site 3: mainland Mindoro;
OIV Other invertebrates and El Nido (11.127° N, 119.320° E)—site 1: Miniloc,
SP Sponges site 2: Shimizu-Inatula, and site 3: Pangulasian. Four
ZO Zoanthids 50-m transects were laid following different depth con-
AA Algal assemblage tours: 15, 10, 7, and 5 m. The data from site 2: South
HA Halimeda Pandan Island in Sablayan, Mindoro was limited to the
MA Macroalgae 5-m depth due to the steepness of the slope. Pictures
TA Turf algae were taken at 1-m intervals along the transect line. The
SG Seagrasses distance between the camera and the seafloor is kept
CA Coraline algae constant at 1.3 m using a PVC frame which allows
DC Dead coral coverage of at least 1 m2. The frame includes a 1-m
DCA Dead coral with algae PVC base used for scaling.
R Rubble ARAICoBeH surveys were done at the same
RCK Rock study sites immediately after completing the un-
S Sand derwater photo transect method. Three transect
SI Silt tows were conducted per station, with at least
W Water 50-m intervals between transects. These were per-
TWS Tape, wand, shadow
pendicular to the shore and targeted depths from 3
U Unknown
to 15 m. The length of each transect varied de-
pending on the bathymetry of the site.
Estimates of live hard coral cover, functional group
diversity, and community structure were compared be-
tween the two methods in the different field locations.
conveniently saved as two R scripts: pre-CPCe and Functional group diversity was used as an alternative to
post-CPCe and is maintained and accessible upon re- species diversity because finer-resolution taxonomic
quest at https://github.com/biome-upmsi. classification is very difficult based on photographs.
633 Page 6 of 15 Environ Monit Assess (2019) 191:633

Fig. 2 Map for the comparative a) VIP 1


study showing three sites in red
2
box for each location. a Verde 117 E 120 123 126
Island Passage—site 1; Ligpo, 21
site 2: Dive and Trek, site 3: N
Twin Rocks b Sablayan, Occi-
dental Mindoro—site 1: North
Pandan,
18
site 2: South Pandan, 0 2.5 km
site 3: mainland Mindoro c El 3
Nido, Palawan—site 1: Miniloc,
site 2: Shimizu-Inatula, b) Sablayan
1
site 3: Pangulasian 15

A 2

B
12

C
0 2.5 km
3

9
c) El Nido

1
6 2

0 2.5 km
3

Data were analyzed at the site level to compare the Time and monetary costs
methods at a finer spatial scale with the objective of
minimizing the sources of variation. A location level The time and monetary costs of data collection and
analysis would introduce more variation due to differ- analysis are also compared in this study. Comparison
ences between sites. Data for live hard coral cover and of required time was estimated for 50 PQs. Field survey
diversity were tested for normality (Shapiro-Wilk test) time included the collection time of PQs using the
and homogeneity of variances (Levene’s test). Two-way ARAICoBeH System and the underwater photo transect
ANOVA with method (ARAICoBeH System or Under- method. Post-survey data processing components in-
water photo transect) and site (1,2, or 3) as factors were cluded time needed to download data from the camera,
used for data that satisfied assumptions of normality and selection of photos for analysis, geotagging, and analy-
homogeneity of variances and the non-parametric sis in CPCe as well as mapping.
Kruskal-Wallis test for data that did not satisfy the Monetary costs were divided into the one-time ex-
assumptions. Non-metric multidimensional scaling pense for necessary equipment and total amount to
(nMDS) on untransformed data based on Bray-Curtis cover all expenses (e.g., boat and scuba tank rental, daily
similarities and analysis of similarities (ANOSIM; wages, etc.) for collecting 400 PQs. This is equivalent to
Clarke and Warwick 1994) were used to compare esti- data from one site with two stations each and four
mates of community structure. The R statistic in transects for each station obtained by one diver for the
ANOSIM indicates that communities are well separated underwater photo transect which is a standard sampling
when R > 0.75, overlapping but clearly different when strategy as suggested by van Woesik et al. (2009). The
R > 0.50 and barely separable when R < 0.25. All statis- monetary cost for the ARAICoBeH System was esti-
tical analyses were done in R (R Core Team 2015). mated also for 400 PQs.
Environ Monit Assess (2019) 191:633 Page 7 of 15 633

Application of ARAICoBeH system differences in estimates of live hard coral cover in two
out of three sites in all three study locations (Fig. 3).
In order to test and illustrate the efficacy of Live hard coral cover estimates for VIP yielded no
ARAICoBeH System as a method for large-scale as- significant differences in sites 2 (p value = 0.618) and
sessment, a more extensive field survey was conducted 3 (p value = 0.096). For Sablayan, there were no signif-
in the town of El Nido, Palawan, on November 2015. icant differences between the methods in sites 1 (p
This survey was independent of the survey done in three value = 0.326) and 3 (p value = 0.753), while for El
locations in El Nido for method comparison (June 2015). Nido, no significant differences were found in sites 1
El Nido is a protected area in the Philippines as part of (p value = 0.460) and 2 (p value = 0.140). The differ-
the National Integrated Protected Areas System. The ences in 1 out of three sites in each study location were
town center lies next to Bacuit Bay, which is a 150- artifacts of the site and survey conditions. The direction
km2 bay opening to the West Philippine Sea and consists of the wind, when the survey was done in site 1 in VIP,
of 45 islands. Marine resources in Bacuit Bay are said to was towards the shore which led to difficulty in maneu-
be diverse; with estimated 910 fish species, more than vering the boat. This created a problem specifically for a
400 coral species, 8 seagrass species, 3 species of en- method comparison study wherein a complete overlap
dangered sea turtles and presence of large marine verte- in the area sampled was necessary. The differences in
brates (e.g., sharks, whales, and rays; CCRES 2014). site 2 in Sablayan could be due to the bathymetry of the
Two of the primary livelihoods of the El Nido commu- area. Underwater transect data was reduced to only one
nity are derived from coral reefs: fishing and tourism depth (5 m) because the area beyond that depth was a
(Cesar 2000; Cola et al. 2005). wall. In contrast, ARAICoBeH still covered a depth
The boat-based ARAICoBeH transects were made at gradient starting at 3 m. The differences in El Nido site
least 1 km apart targeting areas that have coral reefs 3 can also be attributed to abrupt changes in bathymetry.
present based on the 2010 Global Distributions of Coral These changes led to problems in keeping the distance
Reefs from United Nations Environment Program– of the camera from the reef constant. Quality of some
World Conservation Monitoring Centre (UNEP- pictures is affected and is therefore discarded from
WCMC 2010). The data analysis and visualization al- analysis.
gorithms were then applied to the data gathered in the In terms of functional group diversity between the
field survey. two methods, there were also no significant differences
in majority of the study sites (Fig. 4). There was no
significant difference in site 3 (p value = 0.085) in VIP.
Results and discussion The same was observed in all sites in Sablayan: site 1 (p
value = 0.894), site 2 (p value = 0.328), and site 3 (p
Comparison of coral reef indicators derived value = 0.354). Differences were also not significant in
from ARAICoBeH and underwater photo transect sites 2 (p value = 0.068) and 3 (p value = 0.967) in El
method Nido, Palawan. The comparison of functional diversity
highlighted the higher standard error in the
The feasibility of using ARAICoBeH System as an ARAICoBeH estimates relative to that of the underwa-
alternative to the commonly used underwater photo ter photo transects. This can be expected because it
transect method was assessed. The criteria used for covers a continuous depth gradient, while underwater
comparison were three common indicators of coral reef photo transects were laid at discrete depths only.
health: live hard coral cover, functional group diversity, ARAICoBeH therefore captures the variability expected
and community structure. On the average, the three in a coral reef at a depth gradient of 3 m to 15 m.
study locations differed in terms of these indicators For community structure, the two methods also
relative to each other. Coral cover is lower in VIP yielded highly similar results in most sites. Non-metric
relative to Sablayan and El Nido while diversity is multidimensional scaling (nMDS) plots of community
higher in El Nido and Sablayan relative to VIP. Live structures for all the sites quantified from PQs collected
hard coral cover showed that estimates using the using the two methods are compared in Fig. 5. The
ARAICoBeH system and the underwater photo transect yellow and blue polygons represent the two datasets
method were mostly similar. There were no significant derived from ARAICoBeH System and Underwater
633 Page 8 of 15 Environ Monit Assess (2019) 191:633

Fig. 3 Comparison of live hard a) Verde Island b) Sablayan, c) El Nido,


coral cover. Error bars are Passage Mindoro Palawan
standard errors of the mean 100

Hard coral cover (%)


75

50

25

1 2 3 1 2 3 1 2 3

Site

ARAICoBeH System Underwater photo transects

photo transects, respectively. These large overlaps in the R = 0.011, p = 0.066). Exceptions are sites 1 in El Nido
colored polygons suggest that community structure data (R = 0.28, p = 0.003) and 2 (R = 0.33, p = 0.001) in
were highly similar for three sites in VIP and two for Sablayan with slightly higher R values but still with
Sablayan (1 and 3) and El Nido (2 and 3). These visual clear overlaps. This is consistent with the results for
results from nMDS for all sites in the three study loca- coral cover (Fig. 3) wherein there was a significant
tions were consistent with the results of ANOSIM, difference in site 2 in Sablayan and a higher standard
showing that the community structures estimated by error for ARAICoBeH in site 1 El Nido. Both can be due
the two methods were barely separable (R < 0.25) in to the rapid changes in the bathymetries of the sites.
majority of the sites. The R values of these sites are as The three metrics analyzed here, namely live hard
follows: VIP (site 1: R = 0.074, p = 0.001; site 2: R = coral cover, functional diversity, and community struc-
0.028, p = 0.2; site 3: R = 0.13, p = 0.001), Sablayan ture, are the typical parameters used to evaluate the
(site 1: R = 0.13, p = 0.001; site 3: R = 0.011, p = condition of coral reefs (van Woesik et al. 2009;
0.001), and El Nido (site 2: R = 0.19, p = 0.001; site 3: Licuanan and Aliño 2014). Based on our results, the

Fig. 4 Comparison of Shannon a) Verde Island b) Sablayan, c) El Nido,


diversity. Error bars are standard Passage Mindoro Palawan
errors of the mean 1.50
Shannon Diversity index

1.25

1.00

0.75

0.50

1 2 3 1 2 3 1 2 3

Site

ARAICoBeH System Underwater photo transects


Environ Monit Assess (2019) 191:633 Page 9 of 15 633

Fig. 5 Non-metric multidimensional scaling (nMDS) ordination Each point is a composition (functional group percent cover) of
based on Bray-Curtis similarities characterizing community struc- one photoquadrat. Stress values for all sites are < 0.3 which
ture gathered using ARAICoBeH System (yellow) and underwater indicates a relatively good representation but needs careful inter-
photo transect (blue) for each site within the three study locations. pretation (Clarke 1993)

ARAICoBeH system can provide realistic estimates for method requires much longer time for the field survey.
these measurements without the need for scuba divers Selection of photos collected using ARAICoBeH for
with a few exemptions in terms of bathymetry of the analysis and geotagging which involves running a pre-
area. written script in R takes about 5 min. On the other hand,
the diver in the underwater photo transect method takes
Time and monetary cost analysis a picture of the benthic habitat at specified intervals and
these pictures are ready for the next step which is the
A comparison of time and monetary cost for sampling analysis using the CPCe software. The manual classifi-
and analysis are summarized in Tables 2 and 3, respec- cation benthic groups from photos using CPCe are
tively. Overall, the time needed for a coastal benthic common for both methods and thus require the same
habitat analysis with 50 PQs from field survey to gen- amount of time. Visualization of results using the
eration of results is shorter with ARAICoBeH by ARAICoBeH method involves running a pre-written
13 min. Field survey time for ARAICoBeH for 50 script. Results include maps of live hard coral cover as
PQs takes an average of 7 min while the survey time well as general benthic composition. All results are in a
using the underwater photo transect method takes about single datasheet which is a convenient start to statistical
25 min. The need to set up scuba diving gear and lay analysis. Visualization of results from an underwater
transects before the PQs are collected in the scuba-based photo transect survey would require more time when
633 Page 10 of 15 Environ Monit Assess (2019) 191:633

Table 2 Comparison of average time required to collect and process 50 photoquadrats using ARAICoBeH and the underwater photo
transect method

Activity Time needed (min)

ARAICoBeH System Underwater photo transect

Field survey 7 25
Selection of photos for analysis and geotagging 5 0
Photo analysis using CPCe 50 50
Visualization of results 5 5
Total time 67 80

starting a monitoring or assessment program because the however, assumed that a plotting script is already avail-
researcher would need to write script for this or run a able, and a similar amount of time was identified for
series of procedures in plotting software. It was, visualization of results for both methods.

Table 3 Comparison of monetary cost to collect and process 400 photoquadrats using ARAICoBeH and the underwater photo transect
method. Amounts are in US dollars and based on prices and salaries in the Philippines in 2015

ARAICoBeH System Underwater photo transect

Unit cost Quantity Total cost Unit cost Quantity Total cost

I. Equipment
GoPro Hero 3 Silver with underwater case 300.00 1 300.00 300.00 1 300.00
Metal pole 2.20 1 2.20 – – –
Marine rope 160.80 1 160.80 – – –
1-m kernmantle rope 10.80 1 10.80 – – –
3-kg lead weight 65.00 1 65.00 – – –
GPS/transducer viewer 440.00 1 440.00 – – –
Single beam transducer 150.00 1 150.00 – – –
Power bank 107.00 1 107.00
Makeshift fin 3.00 1 3.00 – – –
Handheld GPS – – – 283.50 1 283.50
Dive computer – – – 278.80 2 557.60
50-m transect tape – – – 21.45 4 85.80
PVC camera frame – – – 16.00 1 16.00
Full SCUBA set, excluding tank – – – 750.70 2 1501.40
Total 1238.80 2744.30
II. Expenses per site
Scuba tank rental – – – 6.50 4 26.00
Boat rental 65.00 1 small boat × 0.5 day 32.50 110.00 1 big boat × 1 day 110.00
Gasoline 10.00 0.5 gal × 0.5 day 5.00 20.00 1 gal × 1 day 20.00
Salary: boat operator/s 10.00 1 person × 0.5 day 5.00 10.00 2 persons × 1 day 20.00
Salary: researchers 20.00 2 persons × 0.5 day 20.00 20.00 2 persons × 1 day 40.00
Total 62.50 216.00
Grand total 1301.30 2960.30
Environ Monit Assess (2019) 191:633 Page 11 of 15 633

Fig. 6 Live hard coral cover (%)


map of El Nido, Palawan, in the
Philippines during the 2014
survey. Inset of map in red box
shows coral cover variability at
a finer scale

Monetary costs were also evaluated and compared the underwater photo transect survey method requires
for both methods (Table 3). Equipment costs for the an initial equipment expense of USD 2744.30. The
ARAICoBeH System amount to USD 1238.80 while difference is mainly due to the expensive cost of scuba

Fig. 7 Community structure map


of benthic habitats of El Nido,
Palawan, in the Philippines
during the 2014 survey. Red
boxes indicate boundaries of the
clustering of transects
633 Page 12 of 15 Environ Monit Assess (2019) 191:633

diving equipment. Sampling a site with two stations Guntao while the transects in the central islands of
each takes less time when using the ARAICoBeH meth- Miniloc, Shimizu, Entalula, and Pangulasian are pooled
od. This translates to savings in terms of boat rental together. There are two clusters in the north, and these
expenses, gasoline as well as salary for both boat oper- are transects from the north and south of the island of
ators and researchers. Total boat rental costs inclusive of Cadlao. Two more clusters were created from the tran-
gasoline and boat operators for sampling using sects surveyed inside the bay: Bay north and Bay south.
ARAICoBeH are USD 62.5 per day while it costs about The sites with the highest live coral cover are found in
USD 216.00 for the underwater photo transect surveys. Guntao and the central islands, while the lowest are
The difference in cost is also due to the need for a bigger found in the clusters of the inner Bacuit Bay: Bay north
boat for transport of dive equipment. The big diving and Bay south. Live hard corals were the most dominant
boats are usually manned by two people and consume in these two island clusters while the rest of the clusters
more gasoline. Tank rentals are cheap, but services may were dominated by sand and other abiotic components.
not be readily available in remote areas. Additional costs The highest algal cover was found in Bay north and Bay
may be incurred for transport of scuba diving tanks and south.
in some cases an air compressor to the survey location. Results of the ARAICoBeH survey agree with the
previous rapid ecological assessment conducted by
Distribution and composition of benthic habitats in El Turak and Devantier (2010), wherein similar sites in
Nido, Palawan Miniloc island and Guntao have coral covers that are
higher than the average for all sites sampled in El Nido,
A more extensive survey independent of the method while other sites monitored (e.g., inner Bacuit Bay,
comparison was conducted in the town of El Nido, Cadlao) were found to have significantly lower coral
Palawan, to illustrate the application of the method. cover. Average score of live hard coral cover for El Nido
The survey covered an approximate total length of was 34.48%, considering all the sites surveyed.
72 km of coastal waters in more than 10 islands and Disregarding all ‘sandy’ areas within El Nido, which
was completed within a period of 4 days. These results comprised about 30% of all analyzed PQs, the average
were based on a subsample of 1676 PQs from a total of score increases the average to 47.48%. This was similar
27,299 pictures, using the pre-processing ARAICoBeH to the initial findings of the recent national reef assess-
algorithm which ran for 30 min. Manual scoring of PQs ment in the Philippines wherein El Nido was catego-
was conducted over a period of 3 days, while the data rized as ‘excellent’, with a hard coral cover score of over
merging and mapping step were carried out for about 44% which was twice the national average (Licuanan
20 min using the post-processing algorithm. Overall, it et al. 2017).
took about 8 days to complete, from data collection to Licuanan and Gomez (1988) previously established
visualization. The application of ARAICoBeH System that the fringing reefs in El Nido are consistent with
to assess the benthic habitats of El Nido highlighted the Miyadi’s Embayment Principle (1944, cited by
capacity of the system to detect trends and differences at Horikoshi 1988), which states that the variability in
different scales. The output maps can be used to focus coral community structure is a function of the relative
on a benthic category of interest such as live hard coral position within a bay. This is due to the different envi-
cover and to visually show differences at finer scales ronmental gradients prevalent in the area or the degree
(e.g., transect level, Fig. 6, red box). The distance be- of exposure of benthic communities in the bay. One
tween PQs for this study was set to 2 m, but the resolu- important environmental factor that influences the El
tion of the survey can still be improved by decreasing Nido coral communities is sedimentation rate
the distance since pictures are available at two-second (Quibilan and Aliño 2006). Higher sedimentation rates
intervals. This level of detail can be used as a decision- may be observed inside the bay relative to the other
support tool for managers working at smaller scales areas as a result of river input. Sedimentation is known
such as village leaders and people’s organizations. to be detrimental to coral health (Fabricius 2005). The
Figure 7 illustrates presentation of results at a differ- increased sediment output from Manlag River in the
ent scale and can be used to highlight differences be- 1980s has been shown to affect coral health with ob-
tween sites. The transects were clustered into six sites served decreases in coral cover, diversity, and colony
according to location. The western-most cluster is sizes particularly in sites within Bacuit Bay (Hodgson
Environ Monit Assess (2019) 191:633 Page 13 of 15 633

and Dixon 2000). Higher algal cover and a higher Acknowledgements Special thanks to the Philippine Associa-
tion of Marine Sciences (PAMS) for the opportunity to present this
percentage of the abiotic components in the clusters in
paper.
the inner bay drive the difference from the Guntao and
central island clusters which are dominated by hard Funding information This study was funded by the National
corals. Assessment of Coral Reef Environments (NACRE) research pro-
This survey demonstrates the effectiveness of gram of the Department of Science and Technology (DOST) and
using ARAICoBeH System as a survey tool to the Coral Reef Visualization and Assessment (CoRVA) research
program of the Department of Natural and Environmental Re-
rapidly collect data over large spatial scales, as sources (DENR). Patrick Lawrence P. Cadeliña was also supported
well as to hasten the analysis work using its com- by an Outright Research Grant through the University of the
panion pre- and post-processing algorithms. This Philippines Diliman Office of the Vice-Chancellor for Research
system would be helpful in developing countries and Development (UP-OVCRD) during the revision and fine-
tuning of this paper.
like the Philippines where monitoring to inform
coral reef resource management is limited by man- Compliance with ethical standards
power and certified scuba divers.
Conflict of interest The authors declare that they have no con-
flict of interest.
Conclusions

The assessment of ARAICoBeH as an alternative to the


underwater photo transect method showed that it
References
yielded estimates of common indicators of reef health,
such as hard coral cover and functional diversity, that are
very similar to the results from the scuba-based method Andréfouët, S., & Riegl, B. (2004). Remote sensing: A key tool for
interdisciplinary assessment of coral reef processes. Coral
across different levels of reef health and diversity. This Reefs, 23(1), 1–4.
shows that the ARAICoBeH System can be used to Andréfouët, S., Muller-Karger, F. E., Robinson, J. A., Kranenburg,
assess and monitor the state of shallow marine benthic C. J., Torres-Pulliza, D., Spraggins, S. A., & Murch, B.
habitats without compromising accuracy. It should, (2006, June). Global assessment of modern coral reef extent
and diversity for regional science and management applica-
however, be noted that there are some limitations. The
tions: A view from space. In Proceedings of the 10th inter-
method is not applicable to sites with a steep slope (e.g., national coral reef symposium (Vol. 2, pp. 1732–1745).
reef wall) and with sudden changes in bathymetry. En- Okinawa: Japan.
vironmental conditions, such as water turbulence, sedi- Aswani, S., & Lauer, M. (2006). Incorporating fishermen’s local
mentation, current, and wind velocities during survey knowledge and behavior into geographical information sys-
tems (GIS) for designing marine protected areas in Oceania.
times, should also be considered as these would defi- Human Organization, 65(1), 81–102.
nitely affect quality of images taken. It is also not Baker, A. C., Glynn, P. W., & Riegl, B. (2008). Climate change
possible to identify corals and coral associated organ- and coral reef bleaching: An ecological assessment of long-
isms to higher taxonomic detail. Sizes of areas covered term impacts, recovery trends and future outlook. Estuarine,
Coastal and Shelf Science, 80(4), 435–471.
in each PQ differ due to the variability in distance of the
Bates, C. R., Moore, C. G., Malthus, T, Harries, D. B., Austin, W.,
camera from the substrate. Units are therefore presently Mair, J. M., & Karpouzli, E. (2004). Broad scale mapping of
limited to percent cover but means to scale the photos sublittoral habitats in the sound of Barra, Scotland. Scottish
are currently being undertaken. Natural Heritage Report: 005, 147 pp.
Most coral reefs are found in less-developed coun- Beaman, R. J., & Harris, P. T. (2005). Bioregionalization of the
George V shelf, East Antarctica. Continental Shelf Research,
tries where monitoring and research programs are often
25(14), 1657–1691.
limited by availability of resources. The lower time and Beenaerts, N., & Berghe, E. V. (2005). Comparative study of three
monetary cost of using ARAICoBeH would allow re- transect methods to assess coral cover, richness and diversity.
source managers and researchers to increase the spatial Western Indian Ocean Journal of Marine Science, 4(1), 29–
and temporal scale of their assessments. The method 38.
Beenaerts, N., Berghe, E. V. (2007). Comparative Study of Three
presented here can be applied to other areas to gather Transect Methods to Assess Coral Cover, Richness and
spatially explicit ecological data necessary for the con- Diversity. Western Indian Ocean Journal of Marine Science
servation and management of marine benthic habitats. 4(1)
633 Page 14 of 15 Environ Monit Assess (2019) 191:633

Brown, E. K., Cox, E., Jokiel, P. L., Rodgers, S. K., Smith, W. R., Hill, J., & Wilkinson, C. (2004). Methods for ecological monitor-
Tissot, B. N., et al. (2004). Development of benthic sampling ing of coral reefs (Vol. 117). Townsville: Australian Institute
methods for the coral reef assessment and monitoring pro- of Marine Science.
gram (CRAMP) in Hawai’i. Pacific Science, 58(2), 145–158. Hodgson, G., & Dixon, J. A. (2000). In H. S. J. Cesar (Ed.),
Bruno, J. F., & Selig, E. R. (2007). Regional decline of coral cover Collected Essays on the Economics of Coral Reefs El Nido
in the indo-Pacific: Timing, extent, and subregional compar- revisited: Ecotourism, logging and fisheries (pp. 55–68).
isons. PLoS One, 2(8), e711. Kalmar: CORDIO.
Burke, L., Selig, E., & Spalding, M. (2011). Reefs at risk revisited. Hoegh-Guldberg, O., Mumby, P. J., Hooten, A. J., Steneck, R. S.,
Washington, DC: World Resource Institute. Greenfield, P., Gomez, E., Harvell, C. D., Sale, P. F.,
Capturing Coral Reef and Related Ecosystem Services. (2014). Edwards, A. J., Caldeira, K., Knowlton, N., Eakin, C. M.,
CCRES annual report 2014. Retrieved from http://ccres. Iglesias-Prieto, R., Muthiga, N., Bradbury, R. H., Dubi, A., &
net/images/uploads/publications/78/ccres_annual_ Hatziolos, M. E. (2007). Coral reefs under rapid climate
report_2015_for_web.pdf. change and ocean acidification. Science, 318(5857), 1737–
Cesar, H. (2000). Impacts of the 1998 coral bleaching event on 1742.
tourism in El Nido, Philippines. RI, USA: Coastal Resources Horikoshi, M. (1988). So-called “embayment degree” recognized
Center, University of Rhode Island. in the coastal regional ecosystem in Ryukyu and Palau.
Clarke K. R. (1993) Non-parametric multivariate analyses of Galaxea, 7, 197–210.
changes in community structure. Austral Ecology 18(1): Houk, P., & Van Woesik, R. (2006). Coral reef benthic video
117–143 su r v e ys fa c i l i t a t e lo n g - t e r m m o n i t o r i n g i n t h e
Clarke, K. R., & Warwick, R. M. (1994). An approach to statistical Commonwealth of the Northern Mariana Islands: Toward
analysis and interpretation. Change in marine communities, 2. an optimal sampling strategy. Pacific Science, 60(2), 177–
Cola, R., de la Calzada, R. J., Palma, J., & Tongson, E. (Eds.). 189.
(2005). El Nido: Working together for environmental en- Jokiel, P. L., Rodgers, K. S., Brown, E. K., Kenyon, J. C., Aeby,
forcement. Quezon City: WWF-Philippines. G., Smith, W. R., & Farrell, F. (2015). Comparison of
Cole, R., McComb, P., & Sait, J. (2001). Use of drop video to map methods used to estimate coral cover in the Hawaiian islands.
habitats in a high energy shallow reef environment. In Video PeerJ, 3, e954.
sensing of the size and abundance of target and non-target Jones, L. A., Hiscock, K., & Connor, D. W. (2000). Marine habitat
fauna in Australian fisheries-a national workshop (p. 74). reviews. A summary of ecological requirements and sensi-
Connell, J. H. (1978). Diversity in tropical rain forests and coral tivity characteristics for the conservation and management of
reefs. Science, 199(4335), 1302–1310. marine SACs. Joint Nature Conservation Committee,
Dethier, M. N., Graham, E. S., Cohen, S., & Tear, L. M. (1993). Peterborough, UK: JNCC, 178.
Visual versus random-point percent cover estimations: Knowlton, N. (2001). The future of coral reefs. Proceedings of the
‘Objective’ is not always better. Marine Ecology Progress National Academy of Sciences, 98(10), 5419–5425.
Series, 96, 9. https://doi.org/10.1073/pnas.091092998.
Dunbabin, M., Roberts, J., Usher, K., Winstanley, G., & Corke, P. Kohler, K. E., & Gill, S. M. (2006). Coral point count with excel
(2005). A hybrid AUV design for shallow water reef naviga- extensions (CPCe): A visual basic program for the determi-
tion. In Proceedings of the international conference on robot- nation of coral and substrate coverage using random point
ics and Autiomation. (pp. 2105–2110). Barcelona. count methodology. Computers & Geosciences, 32(9),
English, S., Wilkinson, C., & Baker, V. (1997). Survey manual for 1259–1269.
tropical marine resources (2nd ed.). Townsville: Australian Lam, K., Shin, P. K., Bradbeer, R., Randall, D., Ku, K. K.,
Institute of Marine Science. Hodgson, P., & Cheung, S. G. (2006). A comparison of video
Fabricius, K. E. (2005). Effects of terrestrial runoff on the ecology and point intercept transect methods for monitoring subtrop-
of corals and coral reefs: Review and synthesis. Marine ical coral communities. Journal of Experimental Marine
Pollution Bulletin, 50(2), 125–146. Biology and Ecology, 333(1), 115–128.
Gardner, T. A., Cote, I. M., Gill, J. A., Grant, A., & Watkinson, A. Leujak, W., & Ormond, R. F. G. (2007). Comparative accuracy
R. (2005). Hurricanes and Caribbean coral reefs: Impacts, and efficiency of six coral community survey methods.
recovery patterns, and role in long-term decline. Ecology, Journal of Experimental Marine Biology and Ecology,
86(1), 174–184. 351(1), 168–187.
Gilbert, A., Andréfouët, S., Yan, L., & Remoissenet, G. (2006). Licuanan, W. Y., & Aliño, P. M. (2014). A proposed framework
The giant clam Tridacna maxima communities of three for a national coral reef assessment program. Philippine
French Polynesia islands: Comparison of their population Science Letters, 7(1), 201–206.
sizes and structures at early stages of their exploitation. Licuanan, W. Y., & Gomez, E. D. (1988). Coral reefs of the
ICES Journal of Marine Science, 63(9), 1573–1589. northwestern Philippines: A physiognomic-structural ap-
Green, E. P., Mumby, P. J., Edwards, A. J., & Clark, C. D. (1996). proach. In Proceedings of the 6th international coral reef
A review of remote sensing for the assessment and manage- symposium (Vol. 3, pp. 275–280). Townsville.
ment of tropical coastal resources. Coastal Management, Licuanan, A. M., Reyes, M. Z., Luzon, K. S., Chan, M. A. A., &
24(1), 1–40. Licuanan, W. Y. (2017). Initial findings of the nationwide
Grizzle, R. E., Brodeur, M. A., Abeels, H. A., & Greene, J. K. assessment of Philippine Coral Reefs. Philippine Journal of
(2008). Bottom habitat mapping using towed underwater Science, 146(2), 177–185.
videography: Subtidal oyster reefs as an example of applica- Lirman, D., Gracias, N. R., Gintert, B. E., Gleason, A. C. R., Reid,
tion. Journal of Coastal Research, 241, 103–109. R. P., Negahdaripour, S., & Kramer, P. (2007). Development
Environ Monit Assess (2019) 191:633 Page 15 of 15 633

and application of a video-mosaic survey technology to Service, M., & Golding, N. (2001). Procedural guideline no 3-14:
document the status of coral reef communities. In situ survey of sublittoral epibiota using towed sledge video
Environmental Monitoring and Assessment, 125(1–3), 59– and still photography. In J. Davies, J. Baxter, M. Bradley, D.
73. https://doi.org/10.1007/s10661-006-9239-0. Connor, J. Khan, E. Murray, W. Sanderson, C. Turnbull, &
McClanahan, T. R. (2002). The near future of coral reefs. M. Vincent (Eds.), Natura 2000 marine monitoring hand-
Environmental Conservation, 29(4), 460–483. book. UK marine SACs project (pp. 331–337). Peterborough:
Mitchell, A., & Coggan, R. (2007). 19 remote video techniques. In UK.
R. Coggan, J. Populus, J. White, K. Sheeran, F. Fitzpatrick, & Szmant, A. M. (2002). Nutrient enrichment on coral reefs: Is it a
S. Piel (Eds.), Review of standards and protocols for seabed major cause of coral reef decline? Estuaries, 25(4), 743–766.
habitat mapping (pp. 179–203). MESH Mapping European https://doi.org/10.1007/BF02804903.
Seabed Habitats, INTERRG European Program. Tomascik, T., & Sander, F. (1985). Effects of eutrophication on
Moberg, F., & Folke, C. (1999). Ecological goods and services of reef-building corals. Marine Biology, 87(2), 143–155.
coral reef ecosystems. Ecological Economics, 29(2), 215– https://doi.org/10.1007/BF00539422.
233. Tortell, P. (1992). Coastal zone sensitivity mapping and its role in
Mumby, P. J., Green, E. P., Edwards, A. J., & Clark, C. D. (1997). marine environmental management. Marine Pollution
Coral reef habitat mapping: How much detail can remote Bulletin, 25(1–4), 88–93.
sensing provide? Marine Biology, 130(2), 193–202. Turak, E., & DeVantier, L. (2010). Coral biodiversity, marine
Murdoch, T. J., & Aronson, R. B. (1999). Scale-dependent spatial tourism and conservation priorities in. El Nido, Palawan.
variability of coral assemblages along the Florida reef tract. https://doi.org/10.13140/RG.2.1.4390.9282.
Coral Reefs, 18(4), 341–351.
UNEP-WCMC, WorldFish Centre, WRI, TNC. (2010). Global
Ninio, R., Delean, S., Osborne, K., & Sweatman, H. (2003).
distribution of warmwater coral reefs, compiled from multi-
Estimating cover of benthic organisms from underwater vid-
ple sources (listed in "Coral_Source.mdb"), and including
eo images: Variability associated with multiple observers.
IMaRS-USF and IRD (2005), IMaRS-USF (2005) and
Marine Ecology Progress Series, 265, 107–116. https://doi.
Spalding et al. (2001). Cambridge (UK): UNEP World
org/10.3354/meps265107.
Conservation Monitoring Centre. URL: data.unep-wcmc.
Pacunski, R. E., Palsson, W. A., Greene, H. G., & Gunderson, D.
org/datasets/13
(2008). Conducting visual surveys with a small ROV in
shallow water. In Marine habitat mapping technology for Van Woesik, R., Gilner, J., & Hooten, A. (2009). In Centre for
Alaska (pp. 109–128). Marine Studies, Gerhmann Building, The University of
Pandolfi, J. M., Bradbury, R. H., Sala, E., Hughes, T. P., Bjorndal, K. Queensland (Ed.), Standard operating procedures for repeat-
A., Cooke, R. G., et al. (2003). Global trajectories of the Long- ed measures of process and state variables of coral reef
term decline of coral reef ecosystems. Science, 301(5635), environments. Coral reef targeted research and capacity
955–958. https://doi.org/10.1126/science.1085706. building for management program (p. 4072). St Lucia.
Pandolfi, J. M., Connolly, S. R., Marshall, D. J., & Cohen, A. L. Walcott, J., Eckert, S., Oxenford, H. A., & Horrocks, J. A. (2014).
(2011). Projecting coral reef futures under global warming Use of a towed camera system to investigate benthic habitat
and ocean acidification. Science, 333(6041), 418–422. use by inter-nesting female hawksbill sea turtles. Endangered
https://doi.org/10.1126/science.1204794. Species Research, 24(2), 159–170.
Quibilan, M., & Aliño, P. (2006). Coral community structure of Weinberg, S. (1981). A comparison of coral reef survey methods.
western Philippine reefs I: Spatial patterns. In Proceedings of Bijdragen tot de Dierkunde, 51(2), 199–218.
the 10th international coral reef symposium (pp. 341–350). Wilkinson, C. (2008). Status of coral reefs of the world: 2008.
Okinawa, Japan. Townsville: Global Coral Reef Monitoring Network and
R Core Team. (2015). R: A language and environment for statis- Reef and Rainforest Research Centre.
tical computing. Vienna: R Foundation for Statistical Wilkinson, C. R., & Souter, D. N. (Eds.). (2008). Status of
Computing Retrieved from http://www.R-project.org/. Caribbean coral reefs after bleaching and hurricanes in
Roberts, J. M., Brown, C. J., Long, D., & Bates, C. R. (2005). 2005. Global Coral Reef Monitoring Network.
Acoustic mapping using a multibeam echosounder reveals Williams, S., & Mahon, I. (2004). Simultaneous localisation and
cold-water coral reefs and surrounding habitats. Coral Reefs, mapping on the great barrier reef. In robotics and automation,
24(4), 654–669. 2004. Proceedings. ICRA’04. 2004 IEEE international confer-
Schneider, D. C., Gagnon, J. M., & Gilkinson, K. D. (1987). ence on (Vol. 2, pp. 1771–1776). IEEE. Retrieved from:
Patchiness of epibenthic megafauna on the outer Grand banks http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1308080.
of Newfoundland. Marine Ecology Progress Series, 39, 1–13.
Scopélitis, J., Andréfouët, S., Phinn, S., Arroyo, L., Dalleau, M.,
Publisher’s note Springer Nature remains neutral with regard to
Cros, A., & Chabanet, P. (2010). The next step in shallow
jurisdictional claims in published maps and institutional
coral reef monitoring: Combining remote sensing and in situ
affiliations.
approaches. Marine Pollution Bulletin, 60(11), 1956–1968.

You might also like