Download Data integration for conservation: Leveraging

Ecological Informatics 20 (2014) 13–26
Contents lists available at ScienceDirect
Ecological Informatics
journal homepage: www.elsevier.com/locate/ecolinf
Data integration for conservation: Leveraging multiple data types to
advance ecological assessments and habitat modeling for marine
megavertebrates using OBIS–SEAMAP
Ei Fujioka a,⁎, Connie Y. Kot a, Bryan P. Wallace a,b, Benjamin D. Best a, Jerry Moxley a, Jesse Cleary a,
Ben Donnelly a, Patrick N. Halpin a
a
b
Marine Geospatial Ecology Lab, Nicholas School of the Environment, Duke University, P.O. Box 90328, Durham, NC 27707, United States
Oceanic Society, Washington, DC, United States
a r t i c l e
i n f o
Article history:
Received 12 April 2013
Received in revised form 18 January 2014
Accepted 19 January 2014
Available online 23 January 2014
Keywords:
Biogeographic database
Spatio-temporal ecological assessments
Satellite telemetry
Sea turtle nesting
Line-transect sighting
Photo-identification
a b s t r a c t
Spatially explicit conservation efforts to identify, designate, and prioritize protected areas or biologically significant areas require analyses beyond basic species distribution and abundance studies, including assessments of
migration patterns, habitat use, and ecological drivers of behavior. With the advent of alternate survey methods
and platforms within the marine environment (e.g. satellite telemetry, passive acoustics, photo identification,
nesting site monitoring and genetic sampling) in addition to traditional shipboard or aerial visual surveys, researchers have been developing novel analytical and modeling methodologies to fulfill such in-depth ecological
assessments. This trend has raised interests and needs not only in filling spatial, temporal and ‘ecological’ gaps
but also in the synthesis of these disparate data from multiple methods/platforms. OBIS–SEAMAP, a thematic
node of the Ocean Biogeographic Information System (OBIS) specializing on marine megavertebrates, takes a
unique approach to data integration into the OBIS–SEAMAP database to enable novel applications of a global
biogeographic database. In this paper, we summarize our efforts to accomplish this integration and to develop
novel mapping and visualization tools available on the OBIS–SEAMAP web site. We also discuss advantages
and implications of an integrated database in advancing ecological assessments and modeling efforts based on
preliminary assessments of the OBIS–SEAMAP data and derived products. Finally, we make critical suggestions
for the design and function of biogeographic databases to make contributed data more useful for conservation
efforts.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
Spatial conservation efforts require ecological understanding
beyond the distribution or abundance of the target species for more
effective conservation strategies and management. This need is particularly clear for the designation of marine protected areas or the delineation
and prioritization of biological hotspots. For example, the Convention on
Biological Diversity (CBD) has defined seven criteria for the identification
of Ecologically or Biologically Significant Areas (EBSAs) in the open ocean
and deep seas: 1) Uniqueness or rarity; 2) Special importance for life history of species; 3) Importance for threatened, endangered or declining
species and/or habitats; 4) Vulnerability, fragility, sensitivity, slow recovery; 5) Biological productivity; 6) Biological diversity; and 7) Naturalness
(CBD, 2009). Identification of EBSAs that meet these criteria relies
on spatio-temporal assessments of migration patterns, habitat uses
⁎ Corresponding author.
E-mail address: [email protected] (E. Fujioka).
1574-9541/$ – see front matter © 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.ecoinf.2014.01.003
(breeding and foraging), site fidelity, behavioral differences by sex or
age class, and other ecological aspects (CBD, 2009; Olsen et al., 2009;
Redfern et al., 2006). Such global scale ecological assessments require extensive collections of high-quality biological data, which was the driving
purpose for establishing global biogeographic databases such as the
Ocean Biogeographic Information System (OBIS; http://www.iobis.org;
Ausubel et al., 2010; Costello and Vanden Berghe, 2006; Grassle, 2000).
In the case of marine megavertebrates (e.g. marine mammals, seabirds
and sea turtles), most data in such databases come from museum specimens or sightings recorded during visual surveys (both line-transect
and opportunistic surveys by ship or plane). Visual sightings provide confirmation of species' presence in surveyed locations and are often used to
examine marine biodiversity across taxa, geographic distribution and
abundance of species (Mora et al., 2008; Tittensor et al., 2010; Webb
et al., 2010). Given the costs and restricted spatial and temporal coverage
of visual surveys (Ackleh et al., 2012; Marques et al., 2009), however,
these platforms are limited in filling spatio-temporal gaps for marine
megavertebrates, many of which are long-distance migratory or elusive,
deep-diving species (Gedamke et al., 2001; Olsen et al., 2009). Moreover,
visual surveys are not optimal platforms to collect animal characteristics
14
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
OBIS-SEAMAP Web site
Mapping engine
Non-spatial data access
Standardized point / line tables
Level 1
Layer
definition
Level 2
Individual
dataset
Individual
dataset
Level 3
Relevant
resource
Relevant
resources
Individual
dataset
…
Relevant Geographic
data
resources
Hard disk
OBIS-SEAMAP Database
Fig. 1. The hierarchical structure embodying the three data levels in the OBIS–SEAMAP information system.
such as age, sex, body size, or genetic information, which results in ‘ecological gaps’ in biogeographic databases that hinders in-depth assessments of animal ecology.
To fill spatio-temporal or ecological gaps, other technologies and
methodologies have been developing (i.e. Birdlife International, 2010;
Lascelles et al., 2012; Teloni et al., 2008), including animal tracking,
photo-identification, genetic stock delineations and passive acoustic
monitoring (PAM). Data collected through these methods or platforms
(hereafter ‘data types’) are complementary to those from visual surveys
and provide a wealth of additional information on animal behavior and
ecology. Combined with novel analytical methodologies, the data
collected with various data types facilitate a growing number of studies
investigating animal behavior, migration patterns, habitat uses and
population structures for seabirds (Camphuysen et al., 2012; Louzao
et al., 2009; Skov et al., 2007), marine mammals (DeAngelis et al.,
2011; Mate et al., 2011; Pirotta et al., 2011; Weller et al., 2012), and
sea turtles (Godley et al., 2010; Seminoff et al., 2012).
These demands and requirements highlight the need for biogeographic databases to take a comprehensive approach to filling in
data gaps by incorporating a wide variety of data types in one place
and to make them available for researchers to select the most appropriate suite of datasets that meet their analytical or modeling goals
Table 1
Data sources and citations used in the case studies. The #records column for Case Study 2 show the number of records of sperm whales with the total number of records in a dataset in
parentheses. (*) Contributing providers can be seen by going to the species page and opening up the Datasets on Map window. For more details, see http://seamap.env.duke.edu/content/
help/DatasetsOnMap.
Data source and link to the OBIS–SEAMAP web site
Case study 1: Integration of multiple sea turtle data types
The State of the World's Sea Turtles
http://seamap.env.duke.edu/swot
130 Datasets for Caretta caretta
http://seamap.env.duke.edu/species/173830
58 Datasets for Chelonia mydas
http://seamap.env.duke.edu/species/173833
82 Datasets for Dermochelys coriacea
http://seamap.env.duke.edu/species/173843
Case study 2: Applications of integrated data for modeling work
BOEM Sperm Whale Seismic Study: S-Tag Argos Telemetry
http://seamap.env.duke.edu/dataset/810
SEFSC GoMex Oceanic 1992
http://seamap.env.duke.edu/dataset/13
SEFSC GoMex Oceanic 1993
http://seamap.env.duke.edu/dataset/17
SEFSC GoMex Oceanic 1994
http://seamap.env.duke.edu/dataset/19
SEFSC GoMex Oceanic 1996
http://seamap.env.duke.edu/dataset/25
SEFSC GoMex Oceanic 1997
http://seamap.env.duke.edu/dataset/27
SEFSC GoMex Oceanic 1999
http://seamap.env.duke.edu/dataset/29
SEFSC Gomex Shelf 1994
http://seamap.env.duke.edu/dataset/35
Gomex Sperm Whale Survey 2000
http://seamap.env.duke.edu/dataset/96
Data type
#Records
Citations or providers
Nesting
6008
Kot et al., 2014 and references therein.
Visual/telemetry
157,301
See the species page (*)
Visual/telemetry
18,190
See the species page (*)
Visual/telemetry
10,369
See the species page (*)
Telemetry
3171
(3171)
3
(2342)
4
(3270)
6
(2822)
2
(3602)
1
(2128)
2
(1307)
1
(2092)
30
(297)
Jochens et al., 2006, 2008
Visual
Visual
Visual
Visual
Visual
Visual
Visual
Visual
NOAA Southeast Fisheries Science Center
NOAA Southeast Fisheries Science Center
NOAA Southeast Fisheries Science Center
NOAA Southeast Fisheries Science Center
NOAA Southeast Fisheries Science Center
NOAA Southeast Fisheries Science Center
NOAA Southeast Fisheries Science Center
NOAA Southeast Fisheries Science Center
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
(Pennisi, 2010; Sala and Knowlton, 2006; Scholes et al., 2012). The
Ocean Biogeographic Information System–Spatial Ecological Analysis of Megavertebrate Populations (OBIS–SEAMAP; http://seamap.
env.duke.edu; Halpin et al., 2009), a thematic node of OBIS that specializes on marine megavertebrates, is a rare initiative in that it has
the ability to store multiple data types including opportunistic and linetransect visual sightings, animal tracking data, photo-identification data
(sighting history and fin images), PAM data, seabird colony observations,
marine mammal stranding data and sea turtle nesting data. Improved
integration of these multiple data types would help to fill data gaps and
provide the means for significant improvements in ecological assessments and habitat modeling. More specifically, OBIS–SEAMAP has been
addressing the following technical challenges: 1) how to integrate different data types while maintaining the differences among them without
losing the vital information associated with each data type; 2) how to
map and visualize different types of data in a consistent, informative
way; and 3) how to extend the database structure and tools for a
particular data type without losing the integrity of other types.
Herein, we first summarize our integration efforts and the improvement of the mapping and visualization tools on the OBIS–SEAMAP web
site. Next, we present preliminary assessments of the integrated database to discuss how the integration would contribute to 1) filling data
gaps; 2) advancing modeling work; and 3) enhancing conservation
efforts. For these assessments, we developed two case studies using
sea turtle data collected worldwide and sperm whale (Physeter
macrocephalus) data in the Gulf of Mexico. Finally, we critically evaluate
the current capability and issues surrounding biogeographic databases
and propose future directions.
2. Materials and methods
2.1. Classification of multiple data types into three structuring levels
As of December 2013, the OBIS–SEAMAP database holds more than
3.1 million visual sighting records from ship/airplane-based surveys
and over 710,000 telemetry location records from tagged animals, a
majority of which are tagged sea turtle data shared through the Satellite
15
Tracking and Analysis Tool (STAT) program (Coyne and Godley, 2005).
The database also holds a large volume of sighting history and fin
images from several photo-identification catalogs and an increasing
number of PAM data recorded by stationary or towed hydrophones. In
addition, over 6500 sea turtle nesting records from over 3000 nesting
sites worldwide have been contributed to the State of the World's Sea
Turtles initiative (SWOT; http://www.seaturtlestatus.org), a distinct
node of OBIS–SEAMAP (Kot et al., 2014).
To assess the similarities and differences among data types, we
reviewed all of the data types in the OBIS–SEAMAP database and sorted
out what fundamental information all types possess and how remaining
parts of the data could be grouped and linked to the fundamental information. A record in a visual survey dataset represents a species occurrence or encounter at a given location, and at a specific time and date
during the survey. Most of the shipboard or aerial line-transect survey
datasets submitted to OBIS–SEAMAP hold attributes that are required
for the calculation of the distance to the sighted animals from the survey
platform and the estimation of species density, such as sighting angles,
survey conditions (e.g. Beaufort sea state and glare) and the observer's
positions (Best et al., 2012; Buckland et al., 1993; Thomas et al., 2010).
Line-transect datasets are often accompanied with effort data in the
form of a time series of ship or airplane locations. Telemetry studies
and photo-identification studies both trace individual animals' movements over time and provide unique identifiers of animals or tagging
devices and animal characteristics (e.g. age class, sex). However, telemetry data record the consecutive movement of an animal in a relatively
short period of time, whereas photo-identification data reveal animal
movement at sporadic times over a longer period. Additional information accompanying telemetry and photo-identification data are also
different. Telemetry data often include quality flags (e.g. Argos quality
indicator) and parameters to the filtering algorithm applied whereas
photo-identification data contain classification of images (e.g. image
quality and distinctiveness) and fin features (e.g. top notch or
trailing edge). Sea turtle nest monitoring counts nesting females,
clutches or crawls at fixed locations (i.e. nesting beaches) at various intervals (e.g. daily, and weekly) for one nesting season. Nesting data also
include information describing monitoring efforts (e.g. start and end
Fig. 2. Enhanced mapping and visualization tools for the integrated database. The map represents all data types of green sea turtles (visual sightings: circle, telemetry: triangle, nesting
sites: square) along with survey tracks and animals' movement pathways within a study area (pink boundary). The temporal distribution graph displays the number of animals per
data type stacked (visual sightings: orange, telemetry: red, nesting: blue).
16
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
Fig. 3. All data types for Caretta caretta, including visual sightings (circles), telemetry locations (triangles), nesting beaches (squares) and the RMUs (areas hatched in blue), are mapped
together on the OBIS–SEAMAP web site. The symbols are color-coded by dataset. Data are filtered for 1996–2013.
regarded as relevant resources. Therefore, these resources need to
be integrated to help researchers gain as much insight from their collected data as possible.
In summary, we identified three data structuring levels: level 1)
fundamental information required to map and visualize the data as
points (i.e. locations, date/time, species identification, group size and
data type) and lines (i.e. effort tracklines and travel paths of animals
tagged). These attributes are regarded as common or minimum required ones that are needed for all datasets submitted to OBIS–SEAMAP.
Animal characteristics such as sex and age class are also principal.
Telemetry locations of tagged animals or sighting history of photoidentified animals are often accompanied with these attributes but visual surveys are usually not able to determine such characteristics. Thus,
animal characteristics are grouped into level 1 but not regarded as
required attributes; level 2) additional attributes unique to a data type
(e.g. sighting angles and survey conditions in a line-transect survey)
or particular dataset (e.g. in-situ oceanographic measurements) that
dates of monitoring efforts and percent of beaches covered by
monitoring).
The similarities of these different data types were condensed to
where (latitude and longitude of the location), when (date and time)
and what (species identification). This fundamental information is accompanied with a variety of attributes depending on data types, survey
objectives and survey methods, which is usually denoted in a tabular
format corresponding to individual records. In addition to the fundamental information and tabular attributes, there are relevant resources
collected during surveys or derived from the data that would help
understand and investigate the data further. For example, it is
more informative and better meets project's objectives to visualize
the locations of individual animals captured in a photo-identification
along with their fin images. Likewise, model outputs derived from
biological data should be visualized along with the input data
(Barlow et al., 2009). Supplemental georeferenced data such as inland locations where animals were tagged and released are also
68
5
58
116
Telemetry
Nesting
Visual
5
Telemetry
296
5
115 Nesting
Nesting
Telemetry
6
16
240
Visual
Visual
155
2 056
Loggerhead
308
16
9
399
45
5
587
Green
Leatherback
Fig. 4. Venn diagrams representing spatial complementarity of three data types for loggerhead, green and leatherback sea turtles. Numbers without an arrow represent the number of 1degree cells of each data type. Numbers with an arrow represent the number of 1-degree cells in the intersection of two or three data types.
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
Table 2
Seasonal gaps represented by the number of records of visual sightings and telemetry
locations for loggerhead, green and leatherback sea turtles. Seasons are defined as:
winter — December, January and February; spring — March, April and May; summer —
June, July and August; fall — September, October and November.
Loggerhead
Green
Leatherback
season
Visual
Telemetry
Visual
Telemetry
Visual
Telemetry
Winter
Spring
Summer
Fall
2111
4172
4672
1592
21,129
22,862
52,374
38,757
81
77
76
32
3971
2201
4097
7021
398
344
1289
769
1124
1783
641
87
are suitable to store in database tables. These attributes are optional and
attributes are provided in formats that vary between datasets; and level
3) relevant resources. These resources are also optional and usually provided as files (e.g. fin images from a photo-identification survey or rasters of model outputs) that may or may not be suitable to be put in a
relational database. Resources can be associated with an entire dataset
or an individual observation record in a dataset.
17
method or value range can vary (e.g. numeric code or string vocabulary
for glare). Instead of standardizing these attributes to fit them in a single
database schema, we took a straightforward approach where the attributes were stored in the OBIS–SEAMAP database in their original format. Since attributes at level 2 were not standardized, we did not
design the database and online tools to use these attributes for any statistical calculations or comparisons among datasets. The linkage between the standardized tables (level 1) and individual dataset tables
(level 2) was established using dataset identification numbers at a
dataset level and row identification numbers at a record level. For
level 3, we devised a means of linking datasets or records in the standardized point table with relevant resources such as fin images and
overlaying relevant geographic data (e.g. model outputs) over the biological data. For individual records, the link information is stored in a
column in the standardized table. Relevant resources are usually stored
in a hard disk on the OBIS–SEAMAP server. Relevant geographic data are
defined as map layers in the OBIS–SEAMAP map engine. Geographic
data can be stored in a hard disk in the original format (e.g. ESRI
shapefile) or imported into the OBIS–SEAMAP database depending on
how they are visualized and assessed.
2.2. Implementation of a hierarchical structure for data integration
2.3. Enhancements of mapping and visualization tools
We developed a hierarchical structure representing the three data
levels that maximizes the flexibility of the database structure and data
processing, while retaining valuable information from each data type
(Fig. 1). At the top level (level 1), the fundamental information was
put together in standardized, spatially-enabled tables for point and
line data. Though being common attributes, values can be provided in
different formats (e.g. single tabular column having ‘2000/10/20’ or
three columns having ‘2000’, ‘Oct’ and ‘20’ for an observation date).
The OBIS–SEAMAP data manager cleaned and standardized the attribute values (e.g. ‘2000-10-20 10:15:00’ for an observation date and
time). Each record in the standardized tables was given a flag indicating
the data type. Tabular attributes unique to individual datasets were
stored in separate tables at level 2. Some attributes (e.g. glare) are common in datasets of the same data type but coding scheme, recording
One of the imperative challenges in mapping multiple data types simultaneously was to avoid misinterpretation of data while achieving a
consistent data processing and visualization of multiple data types. For
example, although the OBIS–SEAMAP web site had been providing
gridded summaries of the data representing the number of records
and animals per unit cell, it had counted individual locations from
telemetry data as unique animals, which would result in an overestimate of the species abundance. For improved interpretation of different
data types, we infused the data type as a parameter in the rendering
processes and most of the statistical calculations to produce more
appropriate results per data type before combining them in visual
outputs (e.g. gridded summary maps or time series graphs). With the
improved calculations, for example, an individual animal tagged or
Fig. 5. Green turtle (Chelonia mydas) visual sightings (circle), telemetry locations (triangle) and Northwest Atlantic RMU (hatched area). The green and yellow triangle points off the RMU
are telemetry locations of two juveniles.
18
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
Fig. 6. Loggerhead (Caretta caretta) visual sightings (circle), telemetry locations (triangle) and RMUs in the mid-Atlantic.
photo-identified is counted as one animal no matter how may telemetry
locations or sightings of that animal fall within a unit cell. We also implemented different representations for each data type on a map. Furthermore, we added query options that allow users to filter point data
based on data types.
Since the hierarchical structure unrolls from the standardized table
to individual dataset tables at the second level, we had to develop a
blanket mechanism to abstract the differences among the dataset tables
in order to display dataset-specific attributes in a consistent fashion. The
blanket mechanism has two phases. First, it determines what attributes
are appended to a record or dataset of interest. Second, it extracts values
of these attributes from the dataset table using the record or dataset
identifier to display online.
To visualize relevant resources associated with a record or dataset,
we developed a similar mechanism that locates the resources using
the link information in the standardized table and makes tailored
views on the map interface depending on the type of the resources.
For example, images are displayed with a HTML IMG tag while georeferenced data are rendered through the OBIS–SEAMAP map engine
according to the layer definition that specifies symbols, colors, etc.
Fig. 7. Leatherback (Dermochelys coriacea) visual sightings (circle), telemetry locations (triangle) and RMUs in the North Atlantic.
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
19
Fig. 8. The SWSS telemetry, photo-identification, and genetic sample data were merged into a single dataset “BOEM Sperm Whale Seismic Study (SWSS) S-Tag Argos Telemetry” (http://
seamap.env.duke.edu/dataset/810). Each location can be identified with all attributes (inset) associated with it including sex, age class and a photo-identified fin image.
To demonstrate the application of a biogeographic database integrating multiple data types, we describe two case studies with visual,
qualitative assessments conducted using tools available on the OBIS–
SEAMAP web site, summary statistics of records within the OBIS–
SEAMAP database, and spatial analyses on the data downloaded from
OBIS–SEAMAP using ArcGIS Desktop version 10.1 (ESRI, 2012).
the number of cells that had only one data type present with cells that
had multiple data types present. We also examined seasonal differences
for the number of records OBIS–SEAMAP currently had from visual
sightings vs. telemetry data. In addition, we examined how many visual
sighting and telemetry points lay outside of the RMUs. While the OBIS–
SEAMAP database holds sea turtle visual sighting data as early as 1758,
we filtered data from 1996 through 2013 for this case study because
both telemetry and visual sighting data were available for those years.
2.4.1. Case study 1: integration of multiple sea turtle data types
For the first case study, we integrated the SWOT nesting data, which
had been technically separated from the main OBIS–SEAMAP database,
with other sea turtle data types (i.e. visual sightings and telemetry data)
in OBIS–SEAMAP. SWOT is a partnership among the Oceanic Society, the
IUCN Marine Turtle Specialist Group (MTSG) and the OBIS–SEAMAP initiative along with the network of more than 500 sea turtle experts.
Every year, SWOT collects sea turtle nesting data from around the
world and disseminates the collected data through the OBIS–SEAMAP
web site (http://seamap.env.duke.edu/swot; Kot et al., 2014). Each
record in the SWOT database holds the location of the beach, species
observed and count, reported as an exact or binned number of nesting
females, clutches, or crawls. Additional attributes include beach
names, beach length and details of monitoring effort according to the
SWOT minimum data standards (MDS; SWOT Scientific Advisory Board,
2011). In addition to nesting data records, the SWOT database holds relevant sea turtle biogeographic data, including species distribution ranges,
DNA sampling sites and regional management units (RMUs), which are
geographically explicit areas delineated by integrating sea turtle nesting
beaches, genetic stocks, mark-recapture studies, and telemetry data for
each species (Wallace et al., 2010). By fitting the SWOT nesting data
into the three data levels, the location of the beach, species monitored
and count fell in level 1. Other attributes, such as beach name, beach
length, and monitoring effort details, were grouped in level 2. Biogeographic data were classified as relevant resources at level 3.
Using the integrated sea turtle data (Table 1), we addressed spatiotemporal complementarity of data types for loggerhead (Caretta caretta),
green (Chelonia mydas) and leatherback (Dermochelys coriacea) sea turtles by aggregating data per 1-degree grid per data type and comparing
2.4.2. Case study 2: applications of integrated data for modeling work
The second case study is aimed at illustrating applications of a
merged dataset to modeling work using an intensive data set from the
Sperm Whale Seismic Study (SWSS) supported by the US Minerals
Management Service (now Bureau of Ocean Energy Management —
BOEM; Jochens et al., 2006, 2008; Table 1). The SWSS conducted linetransect surveys, recorded visual sightings of marine mammals and
deployed Argos telemetry tags when possible. In addition, skin samples
and fin images were taken for certain individuals. Animal characteristics
such as sex, age class, body size and haplotypes were identified by skin
samples or visual observations, or estimated with images. In this case
study, we associated Argos transmitted locations of tagged sperm
whales with animal characteristics and a link to a fin image if the animal
was photo-identified. Argos data (e.g. locations and date/time) with animal characteristics were appended to the standardized point table
(level 1). Fin images were classified as relevant resources and put in
the server hard disk (level 3).
Then, we overlaid the merged telemetry data over a summer predictive habitat layer for sperm whales in the Gulf of Mexico derived from
an environmental niche model developed by the Strategic Environmental Research and Development Program (SERDP) project (Best et al.,
2007, 2012; Department of the Navy, 2007a, 2007b). The model
ingested 49 sperm whale sighting data from 8 shipboard and aerial
line-transect surveys registered in OBIS–SEAMAP (Table 1). Using the
interactive tool on the OBIS–SEAMAP site that allows for selection of individual animals based on sex, we compared the movements of male
and female sperm whales over the predictive habitat layer. Habitat
suitability from the predictive habitat layer was sampled and summarized for each animal for all Argos locations from the SWSS dataset.
2.4. Data sources and assessment methods for case studies
20
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
Fig. 9. The summer predictive habitat suitability layer derived from the SERDP project (http://seamap.env.duke.edu/search/?app=serdp) with summer telemetry data of a) all 34 females
and b) all 6 males from the SWSS dataset. The color scheme of the habitat suitability layer captured from the web site is shown to the right.
3. Results
3.1. Mapping and visualization of integrated data
The integration of multiple data types using the hierarchical data
structure allows users to apply most of the advanced mapping and visualization tools available on the OBIS–SEAMAP web site (e.g. interactive
time-series charts or histograms for oceanographic variables sampled)
while mapping multiple data types simultaneously. Moreover, the hierarchical structure and the mechanism to abstract the differences among
data types contributes to a compatible look and interface that presents
disparate attributes from different datasets and to custom-tailored visualization tools specific to a data type (e.g. displaying fin images for
photo-identification data). The mechanism also grants us the capability
to incorporate any sorts of geographic data relevant to a particular
dataset or species (e.g. species distribution maps, study areas or an
extension of the beach covered by nest monitoring efforts) that help
comprehend the details of the data.
Since most of the statistical calculations are done per data type, time
series graphs are able to display summary statistics of the data mapped
(e.g. the number of animals in a designated study area aggregated per
year) with bars per data type stacked (Fig. 2). Similarly, points on the
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
21
Fig. 10. Three females identified as 3700846, 3705648, and 3705726 stayed in low habitat suitability areas in summer.
map are given distinct symbols according to the data type (Fig. 2). As the
fundamental information of line data is stored in the standardized table,
it becomes possible to map telemetry tracklines and survey tracklines
from multiple datasets or multiple species on a single map, which provides useful information to clarify which direction an animal migrated
to, how often the species of interest was encountered during a transect
survey or how well transects were located compared with actual animal
movements (Fig. 2). Through these improvements, researchers are able
to explore spatio-temporal distributions of diverse sets of data while
clearly distinguishing the scientific implications that different data
types deliver.
3.2. Case study 1
The enhanced flexibility of OBIS–SEAMAP enables an extension and
consolidation of separate applications into a unified framework. Originally launched as a separate database and application, SWOT is now
fully integrated into OBIS–SEAMAP, both data-wise and tool-wise. Researchers are able to assess a variety of sea turtle data including visual
sighting, telemetry and nesting data along with the SWOT RMUs using
exploratory tools tailored for SWOT (e.g. filtering nesting beaches by
country or color-coding symbols by colony size) combined with those
commonly available on the OBIS–SEAMAP web site (Fig. 3).
For all of the three sea turtle species assessed, 1-degree cells containing either visual sighing or telemetry data only outnumbered those with
both visual sighting and telemetry data (Fig. 4). Most of the cells for
nesting data do not overlap with other sea turtle data types, since
nesting sea turtle data are collected on beaches (land) where shipboard
or aerial surveys do not cover and tagged animals do not usually visit
Table 3
Summary statistics of habitat suitability (HS) sampled on transmitted locations of three
female sperm whales, all females and all males from the SWSS telemetry dataset and all
visual sightings used in the model.
Group
Min HS
Mean HS
Max HS
STD
#Records
Three females
All females combined
All males combined
Visual sightings
0.0004
0.0004
0.003
0.0009
0.0480
0.0969
0.1275
0.1594
0.0851
0.3864
0.4353
0.3088
0.0196
0.0653
0.0787
0.054
98
767
142
49
(except during the nesting seasons). The seasonal comparison of the
number of visual sighting and telemetry data revealed that the telemetry data for loggerheads and greens compensate fewer visual sightings
in fall than the other three seasons (Table 2). Across the seasons, whereas there are more leatherback visual sightings in summer and fall than
in winter and spring, telemetry data demonstrates the opposite trend
with more data in winter and spring than summer and fall.
Out of 18,094 green turtle records (266 visual sightings and 17,828
telemetry locations), 30 visual sightings and 243 telemetry locations
lay outside of the green turtle RMUs mainly in the North Atlantic
(Fig. 5). A few loggerheads traveled outside the loggerhead RMUs
with 42 visual and 650 telemetry records (Fig. 6). There were 7736 records of leatherbacks, among which 61 visual sightings and 88 telemetry locations fell outside of the leatherback RMUs concentrating in the
North Atlantic and the eastern side of Scotland (Fig. 7).
3.3. Case study 2
The integration of multiple data types from the SWSS surveys resulted in a single dataset containing telemetry locations of 51 tagged sperm
whales (34 females, 6 males and 11 unknown) among which 21 whales
were linked to fin images and all 40 sexed animals were associated with
haplotypes. The compiled dataset provides researchers with the ability
to not only trace individual sperm whale movements over time but
also see detailed animal information including sex, age class and a fin
image of the whale (Fig. 8).
When overlaid on the SERDP summer habitat suitability map for
sperm whales in the Gulf of Mexico, the compiled SWSS telemetry
dataset showed that most of the tagged females remained around offshelf areas in the Gulf where high habitat suitability was predicted
(Fig. 9a). Some females, however, frequented low habitat suitability
areas in the northwestern part of the Gulf. While summer visual
sightings of sperm whales within the boundary of the model presented
an average suitability of 0.1594 (n = 49, SD = 0.054), Argos transmitted locations of three female sperm whales (3700846, 3705648, and
3705726; a full list of individuals is seen in Jochens et al., 2008) loomed
in areas with an average suitability around 0.0480 (n = 98, SD =
0.0196) in the summer. This average for the three females is lower
than a mean suitability of 0.0969 for all 34 females (n = 767, SD =
22
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
Fig. 11. Movements of the male sperm whale 2505725 from the SWSS telemetry dataset in summer (left); the fall movement was also overlaid on the fall density layer in the Atlantic (right).
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
23
Fig. 12. Seasonal sighting rates (the number of records [yellow bars] divided by effort hours [gray line]) for visual sightings of loggerheads from 46 line-transect surveys accompanied with
effort data. Sighting rates are 0.23, 0.37, 0.15 and 0.10 for winter, spring, summer and fall, respectively. The chart was captured from the OBIS–SEAMAP web site.
0.0653) though the difference is not statistically significant (Fig. 10 &
Table 3).
Male sperm whales tended to have wider ranges than females
where an average suitability range (difference between the maximum
and minimum suitability values) for males is 0.2566 (SD = 0.0734;
Fig. 9b) whereas that for females is 0.1168 (SD = 0.0668). One of the
male sperm whales identified as 2505725 (body size: 11 m, haplotype:
C) showed distinct migratory patterns crossing between the Gulf of
Mexico and US east coast in summer and fall (Fig. 11).
4. Discussion
4.1. Advantages of an integrated biogeographic database
One of the most direct benefits of an integrated biogeographic
database with advanced mapping and visualization tools is that it provides complementary, multi-perspective views of biological data for a
species of interest. The ease of spatio-temporal exploration of a variety
of data will promote qualitative assessments of animal distribution
and migration patterns, which expedites data mining and accession of
data appropriate to planned analytical work. While OBIS–SEAMAP primarily remains as an exploratory framework or discovery portal, novel
approaches that we developed bridge qualitative and quantitative assessments, the latter of which is supported by the data download functionality with multi-faceted criteria (i.e., selection of multi-species, time
period, data type, and region). The flexibility of the OBIS–SEAMAP database and application framework also enables reintegration of quantitative analyses or modeling work back into the database and to visualize
results with source data for further investigation.
4.2. Case study 1: enhancing conservation efforts with the integrated data
Since sea turtles cover large geographic ranges and present complex
life history patterns (Godley et al., 2010; James et al., 2005; Polovina
et al., 2004), biological assessments using the combination of telemetry,
visual sighting and nesting data are required to fill spatio-temporal and
ecological gaps and to generate a more resolute understanding of total
range and distribution, habitat use, and migratory patterns of the sea
turtles. Thus, the integration of the SWOT and OBIS–SEAMAP databases
was a perfect coupling that can greatly facilitate global assessments. The
case study presented here verified that the integrated database covered
a wider geographic area and supplemented seasonal gaps of each data
type.
Further assessments of spatio-temporal gaps in data and strategic
planning of visual sighting and tagging surveys will need to take into
account survey effort. For example, when constrained to sighting records from visual line-transect surveys with effort data, the sighting
rate (the number of records per effort hour) of loggerheads in summer
is the second lowest (Fig. 12). This low rate is in contrast with visual
sightings (line-transect plus opportunistic) and telemetry data of
loggerheads that are most numerous in summer (Table 2). This may indicate that summer line-transect surveys have been conducted in areas
where loggerheads visit less frequently in summer or that a low sighting
rate in summer reflects loggerhead's different behavior from other
seasons causing them to be more difficult to visually observe during a
line-transect survey.
The SWOT initiative is expanding its database to include available
satellite telemetry data in an effort to identify “Important Turtle
Areas” throughout the world's oceans. The OBIS–SEAMAP/SWOT database will facilitate the assessments of finer-scale spatial and temporal
variation in densities of turtle movements or presence, which in turn allows for improvements of the existing RMUs. The existence of records
falling outside of an RMU illustrated in our case study will give researchers an opportunity to review the adequacy of and reexamine
the boundaries of current RMUs. First, subadult green turtles visit waters off Cape Cod (Lazell, 1980), which was confirmed by the telemetry
data from two juveniles identified as 37189 (McClellan and Read, 2007)
and 78066a (data provided by Connie Merigo, New England Aquarium
through the STAT program) that traveled outside of the Northwest
Atlantic RMU (Fig. 5). Visual sightings of green turtles were also found
outside of the RMU though the age class was not known. Second, a
recent study documented trans-Atlantic migrations of juvenile loggerheads (Eckert et al., 2008). Although a similar migration was illustrated
in our case study, a middle section of the pathway is not contained in the
loggerhead RMUs (Fig. 6). More data collection and investigations will
be needed to assess if this pathway is regularly followed by many juveniles. Third, although the congregation of leatherbacks at the sea off
Nova Scotia, Canada has been studied (James et al., 2005) and is covered
by the leatherback northwest Atlantic RMU, the visual and telemetry
data indicate that leatherbacks travel further north (Fig. 7). Although
RMUs were delineated to be core areas based on data from various
data types and not intended to encapsulate all locations of sea turtle
data, these points are worth considering when RMU boundaries are
revisited in the future. The redefinition of RMUs or “RMUs 2.0” will
directly advance international and regional conservation processes
such as defining EBSAs or promoting sustainable fishing practices with
reduced bycatch.
4.3. Case study 2: implications of integrated data for modeling work
The compiled SWSS dataset was a first attempt in the OBIS–SEAMAP
database to merge Argos telemetry data and photo-identification images along with animal characteristics obtained from multiple methodologies, which in turn can be explored all together for a better account of
sperm whale distribution. Sperm whales are known to favor waters
with depths of 400–1200 m (Watwood et al., 2006) and display sexual
segregation with females remaining in small-ranging social groups at
tropical and subtropical latitudes while males migrate solitarily or in
24
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
Fig. 13. Sequences of telemetry locations of individual animals in the OBIS–SEAMAP database, such as those of (Sterna paradisaea; http://seamap.env.duke.edu/dataset/705; Egevang et al., 2010; left) are aggregated by 1-degree cell per species and per
day before being uploaded into OBIS (right).
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
25
To overcome these challenges and make public biogeographic databases better suited to various ecological assessments and conservation
efforts, it is crucial to extend and improve their data schemas. There
are on-going efforts to extend database schemas. For example, in partnership with the US Integrated Ocean Observing System (IOOS) and
other US federal agencies, USGS/OBIS-USA is establishing the Marine
BioGeography data standard (MBG), which is an enhanced implementation of Darwin Core designed to hold richer data such as absence,
abundance, tracking and sampling details (Fornwall et al., 2012). This
effort, which OBIS–SEAMAP is participating in, will extend the utility
of the data held in OBIS-USA.
small groups to high latitudes along continental shelf-edges (Griffin,
1999; Rice, 1989; Teloni et al., 2008). These ecological preferences
were qualitatively confirmed by the SWSS telemetry data in which the
female sperm whales remained around the continental shelf in the
Gulf of Mexico and the tagged male sperm whale moved further
offshore (Figs. 8 & 9).
When compared with the SWSS telemetry locations, the SERDP
model results were consistent with off-shelf areas in the Gulf of Mexico,
although a lower habitat suitability prediction in the northwestern part
seemed to contradict the numerous telemetry observations there
(Figs. 9 & 10). This may be the result of other environmental conditions
or inherent variability with their distribution. Moreover, it is generally
unreasonable to extrapolate tracks from only a few individuals to
capture the distribution of a full population (Aarts et al., 2008). Given
that only 49 sperm whale sightings were ingested in the habitat
model, a similar issue of adequacy should be also addressed in a modeling work based on visual sightings.
The track of the male sperm whale 2505725 reasonably suggested a
migratory corridor between shelf habitats of the Gulf of Mexico and US
east coast, separated by the tip of Florida where the lowest suitability
was predicted (Fig. 11). Predicted density surfaces based on linetransect surveys favor capturing behaviors with the greatest residence
time (e.g. foraging and breeding) and not ephemeral ones (i.e. migration). Hence this migratory route is not captured well in the seasonal
average habit model.
These assessments highlight the need for examining multiple data
types in order to fill spatio-temporal, ecological gaps when evaluating
areas of importance for different animal behaviors that may also vary
by characteristics (e.g. age and sex). This is especially the case for
sperm whales that display strong sexual segregation (Pirotta et al.,
2011), which are also hard to detect visually (Ackleh et al., 2012).
Thus, the simultaneous visualization of these various types of data and
ability to interactively explore them spatially and temporally yields an
extremely useful synoptic view for planning or modeling purposes.
We strive to improve the integration of multiple data types and to
develop novel mapping and visualization tools that allow researchers
to discover, explore and assess integrated data with complementary,
multi-perspective views. This effort will help in determining reliable
conservation strategies when used within a justified, systematic analytical framework (Lascelles et al., 2012). The case studies demonstrate
advantages of our approach to leveraging multiple data types to
advance ecological assessments and habitat modeling for marine
megavertebrates using OBIS–SEAMAP. Qualitative assessments similar
to these case studies, exploring data through time, space, taxa, and
data type, can be conducted directly using the OBIS–SEAMAP web site
and its array of tools. As OBIS–SEAMAP project members are deeply involved in the development of the OBIS and OBIS-USA databases and
their portals, lessons learned from our efforts in developing OBIS–
SEAMAP can be applied to these databases (Fujioka et al., 2012). It is
equally important to facilitate collaboration in mutual data exchanges
among taxon/data type/region specific biogeographic databases, as
seen in the SWOT and OBIS–SEAMAP integration, to fill data gaps, and
advance analytical methodologies for a better understanding of the biology and ecology of marine megavertebrates.
4.4. Necessity of extended data schema for biogeographic databases
Acknowledgments
There are several challenges to incorporating data from multiple
data types into a single biogeographic database. While occurrence data
from biogeographic databases indicate broad-scale species distribution
or biodiversity (Evans and Hammond, 2004; Sala and Knowlton, 2006),
occurrence data alone are generally not suitable for other modeling
approaches such as estimates of absolute abundance, density or identification of important areas based on ecological aspects (e.g. migration,
feeding and breeding). For example, a density estimate using visual
sighting data from line-transect surveys requires additional attributes
such as sighting angles, weather conditions and observer's positions
as well as effort data (Barlow et al., 2009; Buckland et al., 1993).
Although the OBIS–SEAMAP database holds such attributes and effort
data, they are dropped when the OBIS–SEAMAP data are transferred
to biogeographic databases that rely on data schema derived from
Darwin Core (Darwin Core Task Group, 2009), originally intended to
document records of museum specimen (OBIS, 2005; Wieczorek et al.,
2012).
Furthermore, reviewing a time series of transmitted positions from
tagged animals within a biogeographic database could be misleading
when these data are used without unique identifiers of individual animals for abundance estimates or density models. For example, multiple
positions from one tagged animal may come across as multiple animals
counted once for each position. To alleviate such misrepresentation,
sequences of telemetry locations of individual animals in the OBIS–
SEAMAP database are aggregated per species, per day and per 1degree grid before being uploaded to OBIS, so the resulting grids at
least illustrate the presence of species at a given location on a given
day. This results in a loss of finer scale accuracy and the capability to
assess individual animals' movements (Fig. 13).
The OBIS–SEAMAP project, launched in 2002 through the support
from the Alfred P. Sloan Foundation, is supported by a funding from
NOPP/NSF (NSF-OCE-07-39199 & NSF-OCE-11-38046). We thank
NOAA NMFS, the US Navy, BOEM and other US federal agencies for
their data contribution and collaborative efforts to improve OBIS–
SEAMAP. The accumulation of sea turtle telemetry data would not be
possible without the support by Michael Coyne, seaturtle.org. For the
SWOT initiative and its database, our acknowledgement goes to Rod
Mast and Brian Huthchinson at the Oceanic Society and all the nesting
data contributors worldwide as well as the organizers of WIDECAST.
We also would like to give thanks to the past OBIS–SEAMAP project
members and the current ones who were not directly involved in this
paper work, including Andy Read and Kim Urian at Duke University,
Melissa Soldevilla at NOAA Southeast Fisheries Science Center, and
Andrew DiMatteo at the US Navy. Above all, we express the deepest
appreciation to OBIS–SEAMAP data providers worldwide.
4.5. Conclusion
References
Aarts, G., MacKenzie, M., McConnell, B., Fedak, M., Matthiopoulos, J., 2008. Estimating
space-use and habitat preference from wildlife telemetry data. Ecography 31,
140–160.
Ackleh, A.S., Ioup, G.E., Ioup, J.W., Ma, B., Newcomb, J.J., Pal, N., Sidorovskaia, N.A.,
Tiemann, C., 2012. Assessing the Deepwater Horizon oil spill impact on marine mammal population through acoustics: endangered sperm whales. J. Acoust. Soc. Am. 131
(3), 2306–2314.
Ausubel, J., Baker, D.J., Clark, M.R., Garcia, S.M., Halpin, P.N., Mannix, H., 2010. Making marine life count: a new baseline for policy. PLoS Biol. 8 (10).
Barlow, J., Ferguson, M.C., Becker, E.A., Redfern, J.V., Forney, K.A., Vilchis, I.L., Fiedler, P.C.,
Gerrodette, T., Ballance, L.T., 2009. Predictive modeling of cetacean densities in the
eastern Pacific Ocean. Final report of SERDP Project SI-1391 (Available at http://
www.serdp.org/content/download/6700/87757/file/SI-1391-FR.pdf).
26
E. Fujioka et al. / Ecological Informatics 20 (2014) 13–26
Best, B.D., Halpin, P.N., Fujioka, E., Read, A.J., Qian, S.S., Hazen, L.J., Schick, R.S., 2007.
Geospatial web services within a scientific workflow: predicting marine mammal
habitats in a dynamic environment. Ecol. Inform. 2, 210–223.
Best, B.D., Halpin, P.N., Read, A.J., Fujioka, E., Good, C.P., LaBrecque, E.A., Schick, R.S., et al.,
2012. Online cetacean habitat modeling system for the U.S. East Coast and Gulf of
Mexico. Endanger. Species Res. 18, 1–15.
Birdlife International, 2010. Marine Important Bird Areas Toolkit: Standardized Techniques for Identifying Priority Sites for the Conservation of Seabirds at Sea. BirdLife
International, Cambridge UK (Version 1.2: February 2011).
Buckland, S.T., Anderson, D.R., Burnham, K.P., Laake, J.L., 1993. Distance Sampling: Estimating Abundance of Biological Populations. Chapman & Hall, London.
Camphuysen, K., Shamoun-Baranes, J., Bouten, W., Garthe, S., 2012. Identifying ecologically important marine areas for seabirds using behavioural information in combination
with distribution patterns. Biol. Conserv. 156, 22–29.
CBD, 2009. Expert Workshop on Scientific and Technical Guidance on the Use of Biogeographic Classification Systems and Identification of Marine Areas Beyond National
Jurisdiction in Need of Protection, Ottawa, Canada.
Costello, M.J., Vanden Berghe, E., 2006. ‘Ocean biodiversity informatics’: a new era in marine biology research and management. Mar. Ecol. Prog. Ser. 316, 203–214.
Coyne, M.S., Godley, B.J., 2005. Satellite Tracking and Analysis Toos (STAT): an integrated
system for archiving, analyzing and mapping animal tracking data. Mar. Ecol. Prog.
Ser. 301, 1–7.
Darwin Core Task Group, 2009. Darwin Core. WWW document. http://www.tdwg.org/
standards/450.
DeAngelis, M., Saez, L., MacNeil, J., Mate, B., Moore, T., Weller, D., Perryman, W., 2011.
Spatio-temporal modeling of the Eastern Pacific gray whale's (Eschrichtius robustus)
migration through California, Oregon, and Washington. 19th Biennial Conference
on the Biology of Marine Mammals, Tampa, Florida, USA. November 27–December
2, 2011.
Department of the Navy, 2007a. Navy OPAREA Density Estimates (NODE) for the Northeast OPAREAS: Boston, Narragansett Bay, and Atlantic City. Naval Facilities Engineering
Command, Atlantic; Norfolk, Virginia. Contract N62470-02-D-9997, Task Order 0045.
Prepared by Geo-Marine, Inc., Plano, Texas.
Department of the Navy, 2007b. Navy OPAREA Density Estimates (NODE) for the Southeast OPAREAS: VACAPES, CHPT, JAX/CHASN, and Southeastern Florida & AUTECAndros. Naval Facilities Engineering Command, Atlantic; Norfolk, Virginia. Contract
N62470-02-D-9997, Task Order 0060. Prepared by Geo-Marine, Inc., Hampton,
Virginia.
Eckert, S.A., Moore, J.E., Dunn, D.C., Van Buiten, R.S., Eckert, K.L., Halpin, P.N., 2008. Modeling loggerhead turtle movement in the Mediterranean: importance of body size and
oceanography. Ecol. Appl. 18, 290–308.
Egevang, C., Stenhouse, I.J., Phillips, R.A., Petersen, A., Fox, J.W., Silk, J.R.D., 2010. Tracking
of Arctic terns Sterna paradisaea reveals longest animal migration. Proc. Natl. Acad.
Sci. U. S. A. 107, 2078–2081.
ESRI, 2012. ArcGIS Desktop: Release 10.1. Environmental Systems Research Institute, Redlands, CA.
Evans, P.G.H., Hammond, P.S., 2004. Monitoring cetaceans in European waters. Mammal
Rev. 34 (1), 131–156.
Fornwall, M., Gisiner, R., Simmons, S., Moustahfid, H., Canonico, G., Halpin, P., Goldstein,
P., Fitch, R., Weise, M., Cyr, N., Palka, D., Price, J., Collins, D., 2012. Expanding biological
data standards development processes for US IOOS: visual line-transect observing
community for mammal, bird, and turtle data. US IOOS Summit Community White
Paper, July 2012.
Fujioka, E., Vanden Berghe, E., Donnelly, B., Castillo, J., Cleary, J., Holmes, C., McKnight, S.,
Halpin, P., 2012. Advancing global marine biogeography research with open source
GIS software and cloud computing. Trans. GIS 16 (2), 143–160.
Gedamke, J., Costa, D.P., Dunstan, A., 2001. Localization and visual verification of a complex minke whale vocalization. J. Acoust. Soc. Am. 109 (6).
Godley, B.J., Barbosa, C., Bruford, M., Broderick, A.C., Catry, P., Coyne, M.S., Formia, A., Hays,
G.C., Witt, M.J., 2010. Unravelling migratory connectivity in marine turtles using multiple methods. J. Appl. Ecol. 47 (4), 769–778.
Grassle, J.F., 2000. The Ocean Biogeographic Information System (OBIS): an on-line,
worldwide atlas for accessing, modeling and mapping marine biological data in a
multidimensional geographic context. Oceanography 13, 5–7.
Griffin, R.B., 1999. Sperm whale distributions and community ecology associated with a
warm-core ring off Georges Bank. Mar. Mamm. Sci. 15 (1), 33–51.
Halpin, P.N., Read, A.J., Fujioka, E., Best, B.D., Donnelly, B., Hazen, L.J., Kot, C., Urian, K.,
LaBrecque, E., Dimatteo, A., Cleary, J., Good, C., Crowder, L.B., Hyrenbach, K.D., 2009.
OBIS–SEAMAP: the world data center for marine mammal, sea bird, and sea turtle
distributions. Oceanography 22 (2), 104–115.
James, M.C., Ottensmeyer, C.A., Myers, R.A., 2005. Identification of high-use habitat and
threats to leatherback sea turtles in northern waters: new directions for conservation.
Ecol. Lett. 8, 195–201.
Jochens, A., Biggs, D., Engelhaupt, D., Gordon, J., Jaquet, N., Johnson, M., Leben, R., Mate, B.,
Miller, P., Ortega-Ortiz, J., Thode, A., Tyack, P., Wormuth, J., Würsig, B., 2006. Sperm
whale seismic study in the Gulf of Mexico; Summary Report, 2002–2004. U.S. Dept.
of the Interior, Minerals Management Service, Gulf of Mexico OCS Region, New
Orleans, LA. OCS Study MMS 2006-034 (352 pp.).
Jochens, A., Biggs, D., Benoit-Bird, K., Engelhaupt, D., Gordon, J., Hu, C., Jaquet, N., Johnson,
M., Leben, R., Mate, B., Miller, P., Ortega-Ortiz, J., Thode, A., Tyack, P., Würsig, B., 2008.
Sperm whale seismic study in the Gulf of Mexico: synthesis report. Dept. of the Interior, Minerals Management Service, Gulf of Mexico OCS Region, New Orleans, LA. OCS
Study MMS 2008-006 (341 pp.).
Kot, C.Y., DiMatteo, A., Fujioka, E., Wallace, B., Hutchinson, B., Cleary, J., Halpin, P., Mast, R.,
2014. The State of the World's Sea Turtles Online Database: Data Provided by the
SWOT Team and Hosted on OBIS–SEAMAP. Oceanic Society, Conservation International, IUCN Marine Turtle Specialist Group (MTSG), and Marine Geospatial Ecology
Lab, Duke University (http://seamap.env.duke.edu/swot).
Lascelles, B.G., Langham, G.M., Ronconi, R.A., Reid, J.B., 2012. From hotspots to site protection: identifying marine protected areas for seabirds around the globe. Biol. Conserv.
156, 5–14.
Lazell Jr., J.D., 1980. New England Waters: critical habitat for marine turtles. Copeia 2,
290–295.
Louzao, M., Bécares, J., Rodríguez, B., Hyrenbach, K., Ruiz, A., Arcos, J., 2009. Combining
vessel-based surveys and tracking data to identify key marine areas for seabirds. Marine Ecology Progress Series 391, 183–197.
Marques, T.A., Thomas, L., Ward, J., DiMarzio, N., Tyack, P.L., 2009. Estimating cetacean
population density using fixed passive acoustic sensors: an example with Blainville's
beaked whales. Acoust. Soc. Am. 125 (4), 1982–1994.
Mate, B.R., Bradford, A., Tsudulko, G., Vertyankin, V., Ilyashenko, V., 2011. Late-feeding
season movements of a Western North Pacific gray whale off Sakhalin Island,
Russia and subsequent migration into the eastern North Pacific. 19th Biennial Conference on the Biology of Marine Mammals, Tampa, Florida, USA. November 27–December
2, 2011.
McClellan, C.M., Read, A.J., 2007. Complexity and variation in loggerhead sea turtle life
history. Biol. Lett. 3 (6), 592–594.
Mora, C., Tittensor, D., Myers, R., 2008. The completeness of taxonomic inventories for
describing the global diversity and distribution of marine fishes. Proc. R. Soc. B Biol.
Sci. 275 (1631), 149–155.
OBIS, 2005. The OBIS Schema Version 1.1. Available at http://iobis.org/node/304.
Olsen, E., Budgell, W.P., Head, E., Kleivane, L., Nøttestad, L., Prieto, R., Silva, M.A., Skov, H.,
2009. First satellite-tracked long-distance movement of a sei whale (Balaenoptera
borealis) in the North Atlantic. Aquat. Mamm. 35 (3), 313–318.
Pennisi, E., 2010. Filling gaps in global biodiversity estimates. Science 330, 24.
Pirotta, E., Matthiopoulos, J., MacKenzie, M., Scott-Hayward, L., Rendell, L., 2011. Modelling sperm whale habitat preference: a novel approach combining transect and
follow data. Mar. Ecol. Prog. Ser. 436, 257–272.
Polovina, J.J., Balazs, G.H., Howell, E.A., Parker, D.M., Seki, M.P., Dutton, P.H., 2004.
Forage and migration habitat of loggerhead (Caretta caretta) and olive ridley
(Lepidochelys olivacea) sea turtles in the central North Pacific Ocean. Fish.
Oceanogr. 13 (1), 36–51.
Redfern, J.V., Ferguson, M.C., Becker, E.A., Hyrenbach, K.D., et al., 2006. Techniques for
cetacean-habitat modeling. Mar. Ecol. Prog. Ser. 310, 271–295.
Rice, D.W., 1989. Sperm Whale, Physeter macrocephalus Linnaeus, 1758. Academic Press,
London.
Sala, E., Knowlton, N., 2006. Global marine biodiversity trends. Annu. Rev. Environ.
Resour. 31, 93–122.
Scholes, R.J., Walters, M., Turak, E., Saarenmaa, H., Heip, C.H.R., Tuama, É.Ó., Faith, D.P.,
Mooney, H.A., Ferrier, S., Jongman, R.H.G., Harrison, I.J., Yahara, T., Pereira, H.M.,
Larigauderie, A., Geller, G., 2012. Building a global observing system for biodiversity.
Curr. Opin. Environ. Sustain. 4, 139–146.
Seminoff, J.A., Benson, S.R., Arthur, K.E., Eguchi, T., Dutton, P.H., et al., 2012. Stable isotope
tracking of endangered sea turtles: validation with satellite telemetry and d15N analysis of amino acids. PLoS ONE 7 (5), e37403. http://dx.doi.org/10.1371/journal.
pone.0037403.
Skov, H., Durinck, J., Leopold, M.F., Tasker, M.L., 2007. A quantitative method for evaluating the importance of marine areas for conservation of birds. Biol. Conserv. 136,
362–371.
SWOT Scientific Advisory Board, 2011. The State of the World's Sea Turtles (SWOT)
minimum data standards for nesting beach monitoring. Technical Report
(24 pp.).
Teloni, V., Mark, J.P., Patrick, M.J.O., Peter, M.T., 2008. Shallow food for deep divers:
dynamic foraging behavior of male sperm whales in a high latitude habitat. J. Exp.
Mar. Biol. Ecol. 354, 119–131.
Thomas, L., Buckland, S.T., Rexstad, E.A., Laake, J.L., et al., 2010. Distance software: design
and analysis of distance sampling surveys for estimating population size. J. Appl. Ecol.
47, 5–14.
Tittensor, D.P., Mora, C., Jetz, W., Lotze, H.K., Ricard, D., Vanden Berghe, E., Worm, B., 2010.
Global patterns and predictors of marine biodiversity across taxa. Nature 466 (7310),
1098–1101.
Wallace, B.P., DiMatteo, A.D., Hurley, B.J., Finkbeiner, E.M., Bolten, A.B., et al., 2010.
Regional management units for marine turtles: a novel framework for prioritizing
conservation and research across multiple scales. PLoS ONE 5 (12).
Watwood, S., Miller, P.O.J., Johnson, M.P., Madsen, P.T., Tyack, P.L., 2006. Deep-diving
foraging behaviour of sperm whales (Physeter macrocephalus). J. Anim. Ecol. 75,
826–835.
Webb, T.J., Vanden Berghe, E., O'Dor, R., 2010. Biodiversity's big wet secret: the global
distribution of marine biological records reveals chronic under-exploration of the
deep Pelagic Ocean. PLoS ONE 5 (8).
Weller, D.W., Klimek, A., Bradford, A.L., Calambokidis, J., et al., 2012. Movements of gray
whales between the western and eastern North Pacific. Endanger. Species Res. 18,
193–199.
Wieczorek, J., Bloom, D., Guralnick, R., Blum, S., Döring, M., et al., 2012. Darwin core: an
evolving community-developed biodiversity data standard. PLoS ONE 7 (1), e29715.
http://dx.doi.org/10.1371/journal.pone.0029715.