Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
Contents lists available at SciVerse ScienceDirect
Journal of Experimental Marine Biology and Ecology
journal homepage: www.elsevier.com/locate/jembe
Testing an autonomous acoustic telemetry positioning system for
fine-scale space use in marine animals
Zy Biesinger a,⁎, Benjamin M. Bolker b, Douglas Marcinek a, Thomas M. Grothues c,
Joseph A. Dobarro c, William J. Lindberg a
a
b
c
Program in Fisheries and Aquatic Sciences, University of Florida, 7922 NW 71st Street, Gainesville, FL 32653 USA
Department of Mathematics and Statistics, McMaster University, 1280 Main Street West, Hamilton, Ontario L8S 4K1, Canada
Institute of Marine and Coastal Sciences, Rutgers, The State University of New Jersey, 800 c/o 132 Great Bay Boulevard, Tuckerton, NJ 08087-2004, USA
a r t i c l e
i n f o
Article history:
Received 31 December 2012
Received in revised form 6 June 2013
Accepted 9 June 2013
Available online xxxx
Keywords:
Acoustic telemetry
Autonomous positioning
Autonomous underwater vehicle
Long-term tracking
Telemetry capabilities
Telemetry limitations
a b s t r a c t
We tested the capabilities and limitations of a novel autonomous acoustic positioning telemetry system with
data from fifteen field deployments off the Florida coast. Telemetry array coverage areas ranged between 100
and 300 m across. For fixed transmitters within the array, the fraction of transmissions leading to high-quality
calculated position estimates averaged 44%, with wide variation. Positional accuracy was about 2 m. The choice
of filtering strictness represented a trade-off between the accuracy and frequency of positions. There was
substantial temporal variation, but no clear pattern (e.g., daily or tidal correlations) in frequency of positions.
There was no spatial bias within the array. Array performance for stationary transmitters was robust to user
errors in sound speed and hydrophone position estimates. Performance was less robust for a transmitter
attached to an autonomous underwater vehicle moving through the array, with 22% of transmissions leading
to position estimates. Overall the system produced reliable results, but as the use of acoustic telemetry in
complex ecological studies increases it is important to recognize technological requirements and limitations.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
The need to understand complex space-use, behavioral, and ecological
processes of animals in challenging aquatic systems is driving the
development of novel acoustic telemetry technologies (Espinoza et al.,
2011b; Lucas and Baras, 2000; Niezgoda et al., 2002; O'Dor et al., 1998).
New positioning telemetry systems allow data collection at finer scales,
over larger areas and longer times, and in less accessible locations than
ever before (Andrews et al., 2011; Hanson et al., 2007; Parsons et al.,
2003; Voegeli et al., 2001). Such detailed data enhance our ability to
address complex ecological questions and are particularly important as
spatial management tools (e.g., design of marine protected areas) are
increasingly used to manage exploited species (Cooke et al., 2005; Feary
et al., 2011). Though acoustic telemetry technologies are powerful tools,
their proper use requires an understanding of their requirements and
limitations. This is especially true for positioning telemetry arrays,
which provide fine-scale animal position estimates.
Some active acoustic technologies require animals to be captured
and fitted with acoustic tags, which allow the measurement of specific
individuals' behavior (Heupel et al., 2006). In addition to uniquely
identifying individuals, tags can report a wide range of environmental,
⁎ Corresponding author at: Department of Fisheries and Wildlife, Oregon State University,
Nash Hall, Room 104, Corvallis, OR 97331, USA. Tel.: +1 541 737 1859; fax: +1 541 737
3590.
E-mail address: [email protected] (Z. Biesinger).
0022-0981/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jembe.2013.06.007
physiological, and behavioral data (Carey and Lawson, 1973; Wolcott,
1995). Tagged individuals can be manually located with a directional
hydrophone (Collazo and Epperly, 1995; Johnson et al., 2009; Zeller,
1999) giving position estimates. Tag detections by an omnidirectional
hydrophone can be interpreted as presence/absence data or as position
estimates with error around the position determined by the detection
capability of the receiver (Clements et al., 2005; Heupel et al., 2004).
Some work has been done using presence/absence information from
multiple hydrophones and calculating two- and three-dimensional
short-term center of activity locations, but not precise location
estimates at a single time (Heupel et al., 2012; Simpfendorfer et al.,
2012; Simpfendorfer et al., 2002).
In contrast, with acoustic positioning telemetry, when a particular
tag transmission is received by at least three hydrophones, the twoor three-dimensional position solution can be calculated to within a
few meters or less (Bergé et al., 2012; O'Dor et al., 1998; Voegeli et al.,
2001), using differences in the times a single transmission arrives at
multiple hydrophones (e.g., hyperbolic trilateration, Niezgoda et al.,
2002). This technology requires the important distinction between a
transmission's detection by individual hydrophones and its calculated
position solution when detected by at least 3 hydrophones. In most instances, positioning telemetry can provide precise position estimates
more often and for longer periods than manual tracking, allowing for
finer scale movement and habitat use studies (Espinoza et al., 2011a).
The advantages of positioning arrays come with the extra requirements of very precise clock synchronization (typically achieved by
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
continuous cable or radio communication), enough hydrophones to
cover the geographic scale of interest, accurate position estimates of
each hydrophone, and an estimate of sound speed. Though all
acoustic technologies are affected by the aquatic acoustic environment and deployment details, positioning telemetry technologies
are particularly susceptible to noise, acoustic conditions, and user
errors (Bergé et al., 2012; Cote et al., 1998; Heupel et al., 2006;
O'Dor et al., 1998; Welsh et al., 2012).
New positioning telemetry systems, for example the Lotek
Wireless© WHS 3050 MAP (Cooke et al., 2005; Cote et al., 1998;
Niezgoda et al., 2002) or Vemco© VPS (Andrews et al., 2011;
Espinoza et al., 2011b) use autonomous hydrophones, stationary
beacon tags, and post-deployment positioning software, instead of
continuous communication, to compensate for clock differences. In
these systems, autonomous hydrophones potentially allow arrays to
be deployed over larger areas and in deeper waters than traditional
communicating positioning systems, without surface or shore exposure. Positioning telemetry has been used to study animal space and
habitat use in many settings (Bégout and Lagardère, 1995; Hanson
et al., 2007; Klimley et al., 2001; Semmens, 2008).
The added sophistication of positioning telemetry, especially
autonomous systems, necessitates a clear understanding of system requirements and limitations in order to correctly interpret telemetry
output. Post-processing computation produces metrics of the quality,
or accuracy, of each calculated position solution, allowing users
(e.g., of Lotek© systems) or the manufacturer (e.g., of Vemco© or HTI©
systems) to filter telemetry output (Cooke et al., 2005; Hanson et al.,
2007; Niezgoda et al., 2002). Studies using positioning systems have
reported accuracies of 2 m or better (Cooke et al., 2005; Cote et al.,
1998; Espinoza et al., 2011b; Niezgoda et al., 2002; O'Dor et al., 1998;
Semmens, 2008). The reported fraction of tag transmissions resulting
in reliable position solutions varied between 4 and 75%, depending
heavily on the acoustic environment, tag strength, array geometry,
and array spacing (Brown et al., 2010; Cooke et al., 2005; Geraldi and
Powers, 2011; Niezgoda et al., 2002).
Few studies have purposefully evaluated positioning telemetry system performance, e.g., the fraction or accuracy of position solutions
(Bergé et al., 2012; Ehrenberg and Steig, 2002; Espinoza et al., 2011b),
temporal and spatial variations in system performance, or the effects
of user input errors on performance. The degree of temporal variability
in array performance is expected to reflect the array's susceptibility to
temporal changes in acoustically important environmental variables,
for example, temperature, salinity, weather and sea state, or water
stratification. Spatial variation is expected to reflect performance differences due to array deployment geometry.
In 7 steps, we evaluated Lotek's© autonomous positioning system for
several stationary and one moving transmitter. (1) We calculated the
overall fraction and temporal variability of transmissions detected by
single hydrophones at different distances. This can be done without
deploying an entire array and aids in initial array design (Cooke et al.,
2005). (2) We calculated the overall fraction and temporal variability
of the fraction of transmissions resulting in position solutions with the
array deployed at different spacings, and explored the interaction
between data filtering and position solution accuracy. Detection and
position solution fractions from our large datasets may be suggestive
of position solution probabilities in future animal studies. (3) We estimated position solution accuracy of filtered data. (4) We described
the spatial variation in the fraction and accuracy of position solutions.
(5) We calculated the impact of sound speed changes (e.g., due to
changing water temperature) or estimation errors on position solution
fraction and accuracy. (6) We calculated the effect of errors in hydrophone position estimates on position solution fraction and accuracy.
Using data from one transmitter attached to an Autonomous underwater vehicle, (7) we calculated the overall fraction of transmissions
resulting in position solutions and position solution accuracy and calculated the impact of sound speed changes and errors in hydrophone
47
position estimates on position solution fraction and accuracy. Results
presented here using tags of known position, complement array performance using tags implanted in fish, and thus of unknown position,
described in Biesinger et al. (2013).
2. Materials and methods
2.1. Study system
This study was conducted 30 km off the Florida coast in the Gulf of
Mexico in 13 m of water, where the seafloor was characterized by a
mix of low-relief hard-bottom and sand-bottom habitats (Parker et
al., 1983). Hard-bottom was characterized by emergent limestone
often covered with a veneer of sand and shell rubble, and typically
sustained low algal, sponge, and soft-coral growth less than 0.5 m
tall. Sand-bottom was characterized by deeper, bare sand. Part of
the Steinhatchee Fisheries Management Area, our experimental,
artificial reef system (Fig. 1) consisted of clusters of four, immediately
adjacent, hollow cement hemispheres about 1 m tall, with holes
allowing fish access to the interior. Each reef cluster had a 4 m2 footprint. Each telemetry array deployment centered on a single reef,
with no others within the array. Smaller experimental reefs in the
system consisted of single cement hemispheres. All locations had
relatively little turbulence from wave action or boat traffic, and no
other structures, e.g., docks or hardened shorelines. We continuously
measured water temperature with an Acoustic Doppler Current
Profiler (ADCP, Teledyne RDI© Workhorse Sentinel, 600 kHz) approximately 1 km away.
2.2. Telemetry system: transmitters, hydrophones, and software
2.2.1. Transmitters
We used uniquely coded Lotek© 76 kHz transmitters of two basic
types: tags, normally attached to animals, with (MA-TP16-25, 2 s
interval, transmission length: 250 ms, 156 dB re 1 μPa at 1 m) and
without temperature and pressure sensors (MA-16-25, 2 s interval,
transmission length: 250 ms, 156 dB re 1 μPa at 1 m), and stationary
beacons, used for clock synchronization, with (MA-TP16-50, 20 s
interval, transmission length: 250 ms, 156 dB re 1 μPa at 1 m) and
without temperature and pressure sensors (MA-16-50, 20 s interval).
The MAP system uses a code-division-multiple-access (CDMA)
scheme encoding information in waves that are extracted from a
noise carrier signal based on correlation; it does not require time
sharing of the acoustic channel and is therefore very robust against
code collision and noise interference (Niezgoda et al., 2002). Because
of this, the timing among signals can be and is invariant to 0.00001 s,
in contrast to other common coding schemes such as pulse interval
coding (PIC) that randomize transmission intervals (Grothues,
2009). Each complete transmission, termed a symbol, lasts on the
order of milliseconds and is comprised of three codes, short acoustic
bursts conveying ID and, optionally, sensor data. The duration of a
complete symbol, or one of its constituent codes, is always short
relative to the time that even a fast fish can move a single body
length. Although these tags supported sensors, those data are not
used to evaluate system performance and we do not report it here.
2.2.2. Hydrophone array
We used an array of five autonomous submersible dataloggers
(Lotek© WHS 3050 MAP, 76 kHz), which we call hydrophones, each
consisting of an actual omnidirectional hydrophone, a receiver, a
datalogger, and a battery pack. Each hydrophone unit was mounted
about 2 m above the seafloor using either posts driven into the
underlying rock (for deployments lasting longer than a day) or temporary, weighted posts with a surface buoy adding vertical stability
and independent GPS position estimates. All array deployments
used the same basic geometry: a central hydrophone 10 m northeast
48
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
Fig. 1. Map of locations of experimental, artificial reef system in the northeastern Gulf of Mexico. Each reef consisted of 1 or 4 hollow cement hemispheres about 1 m tall, with holes
allowing fish access to the interior. Each telemetry array deployment centered on a single 4-unit reef, with no others within the array.
of the reef and four hydrophones set a given distance in each of the cardinal directions. Depending on the deployment, the outer hydrophones
were 50 to 150 m from the reef. Beacons were suspended 0.5 m above
the hydrophone and attached via monofilament line and a float.
Prior to deployment, hydrophones must be set to record receptions
in symbol or code mode. Symbol mode saves only completely detected
symbols (i.e., all three codes of a single transmission). Code mode saves
individual codes, making it possible to sometimes reconstruct incomplete symbols (e.g., only one or two codes) in a process called Partial
Symbol Reconstruction (PSR). PSR is facilitated within the native
software (Section 2.2.3) by checking the timing of isolated codes from
incomplete symbols against their earlier or subsequent occurrence in
complete symbols and works because of the precise clock and invariant
transmission timing. This potentially has important consequences to
performance in marginal acoustic conditions because it requires that
only one third of a signal be logged by at least 3 hydrophones.
To estimate hydrophone positions on the seafloor, we attached a
Garmin GPS 76 (2 m accuracy, recording every 5 s) to a rod rising 1 m
above a surface buoy tied tautly to the seafloor, so that the buoy submerged below, and the GPS remained above the wave action. Held in
place for 5 min, the mean of 100 recorded GPS positions became the
hydrophone position estimate. With accurate hydrophone position estimates the array should be capable of calculating positions with about
1 m accuracy within the array (Cooke et al., 2005; Niezgoda et al.,
2002). Because of deployment geometry, the details of trilateration,
and the discretization of time by hydrophones, if a transmitter moves
outside the perimeter of a circular array, positional accuracy drops
drastically. Moving outside the array essentially reduces the available
information to presence/absence or general direction and distance
(Simpfendorfer et al., 2002), and temperature, and depth, but not an
accurate position solution (Cooke et al., 2005; Cote et al., 1998).
2.2.3. Positioning software: ALPS
At the end of a deployment, Lotek's© positioning software, the
Asynchronous Logger Positioning System (ALPS, version 2.18), combined hydrophone GPS position estimates, a sound speed estimate,
and detection records from all five hydrophones to calculate each
transmitter's two-dimensional position solutions. It is useful to distinguish a detection by a single hydrophone from a position solution
calculated by ALPS using detections from three or more hydrophones
(Cote et al., 1998; Lagardère et al., 1990). The use of multiple beacons
during array deployments was primarily a redundancy to ensure
clock drift compensation during post-processing; only one beacon
was required to process data through ALPS. However, during an animal movement study, temporal and spatial variations in the number
of position solutions calculated for beacons could be used to account
for some of the variation in animal tag position solutions.
ALPS produced a time-series of Easting–Northing position solutions,
including sensor data if available, and several quality metrics (Cooke et
al., 2005; Hanson et al., 2007; Niezgoda et al., 2002). One metric from
the matrix algebra underlying ALPS calculations, the condition number
(CN) also called the error amplification factor, describes the stability or
sensitivity of the position solution matrix. High CN values suggest less reliability in the matrix calculation. With data collected in code mode, ALPS
could run without performing PSR, i.e., only complete symbols produced
position solutions. Alternatively, ALPS could use PSR to reconstruct the
transmitter ID of partially detected symbols, though sensor data were
lost. To compare data collected in different modes, note that results
from code mode without PSR are equivalent to results from symbol
mode. Except where noted, results include PSR. We used R (R Core
Development Team, 2012) for filtering and further calculations of ALPS
output, as described below. Telemetry data and R scripts to reproduce
all figures are in Supplementary Files 1 and 2.
2.3. Field deployments
To test array performance we conducted one detection trial
between one transmitter and one hydrophone, three trials testing
the array at different spacings, two trials testing spatial variation in
performance within the array, and nine fish tagging studies, all
summarized in Table 1 and described in Supplementary File 3. Here
we report results from stationary transmitters.
2.4. System performance
2.4.1. Detection fraction
The probability of a transmission producing a position solution is
directly related to the (non-independent) probabilities of detection
by at least three hydrophones, which, in turn, depend on the distance
between the transmitter and each hydrophone. To characterize the
detection probabilities, for each transmitter–hydrophone pair we calculated how the overall detection fraction (# detections per # transmissions) varied with distance. Acoustic transmission through water
is affected by temperature, salinity, sea state, or water stratification.
To explore how detection fractions varied through time, reflecting
changes in the acoustic environment, we examined the detection
fraction for each deployment hour. We did not measure salinity or
sea state, or identify any stratification events.
2.4.2. Position solution fraction
Data from Deployments A and C–O were used to examine how the
fraction of position solutions (# position solutions per # transmissions)
was affected by array spacing (Table 1). Using independent GPS transmitter position estimates, we explored the utility of CN. For each transmitter we examined the number of position solutions below various
CN values. We examined the relationship between the Northing locations of position solutions and their associated CN values, relative to
the GPS Northing estimate. From these analyses, we examined a range
of filtering strategies and report two extremes: no filtering and stringent
filtering (only keep solutions with CN b 1.5). We did not apply other
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
49
Table 1
Array deployment details.
Deployment
Dates
A
B
Cd
D
E
F
G
H
I
J
K
L
M
N
O
2007
2008
2008
2009
2009
2009
2009
2009
2009
2009
2009
2009
2009
2009
2009
a
b
c
d
Dec 7–2008 Jan 15
July 22
Oct 9–Dec 7
April 23
May 7
May 7
June 1
June 3
June 1–June 17
July 10–July 27
Aug 3–Aug 20
Aug 24–Sep 8
Sep 14–Oct 1
Oct 12–Oct 27
Nov 16–Nov 30
Activity
Location IDa
Collection modeb
Spacing (m)c
Total duration
Number of transmitters
Fish study
Detection trial
Fish study
Spacing trial
Spacing trial
Spacing trial
Internal trial
Internal trial
Fish study
Fish study
Fish study
Fish study
Fish study
Fish study
Fish study
1
1
1
1
1
1
2
2
2
3
4
1
5
6
3
Symbol
Symbol
Code
Code
Code
Code
Code
Code
Code
Code
Code
Code
Code
Code
Code
50
200, 300
125
125
100
150
100
100
100
100
100
100
100
100
100
38 days
12 min
59 days
77 min
136 min
105 min
377 min
252 min
16 days
17 days
17 days
15 days
17 days
15 days
14 days
5
1
5
5
7
7
14
14
5
5
5
5
5
4
4
(hb)
(hb)
(hb)
(hb)
(hb)
(hb)
(hb)
(hb)
(hb)
(sb)
(sb)
(hb)
(sb)
(hb)
(sb)
Each location was classified as predominantly hard-bottom (hb) or sand-bottom (sb).
Hydrophone setting for data collection mode.
Spacing between central and marginal hydrophones, except Deployment B which was spacing between a single hydrophone and transmitter pair.
Deployment using the AUV and a moving transmitter on 2008 Oct 10.
more sophisticated techniques, e.g., filtering on two or more ALPS quality
metrics (Hanson et al., 2007) or Kalman filtering (Grewal and Andrews,
2008). In this study, except where noted, results represent stringently filtered data.
Position solutions from Deployment A, collected in symbol mode,
were not augmented by PSR. To compare these results with PSR
results, using Deployments I-O, we calculated the percent increase
in position solution fractions from non-PSR to PSR for each transmitter, then calculated the mean percent increase. For Deployment A, we
increased each position solution fraction by the overall mean percent
increase. Finally, we examined how the hourly position solution fraction varied through time.
2.4.3. Position solution accuracy
Assessing position solution accuracy was complicated by the difficulty of making accurate, independent estimates of hydrophone and
transmitter positions on the seafloor. We used two approaches to assess
accuracy. First, using data from Deployments A and C–O (beacons only
from Deployments G and H), we compared mean ALPS position solutions of each transmitter with independent estimates made using GPS
buoys. This approach cannot separate the errors of each positioning
method. Second, we used successive position solutions of a transmitter
to form a cloud of Easting–Northing locations where the centroid is the
mean position solution. Tighter clouds suggest higher accuracy of individual position solutions. We examined the consistency of position solutions through time by calculating the range of the central 90% of
solutions along the Northing axis. This approach can reveal temporal
variation in accuracy but not systematic biases.
2.4.4. Spatial variation
The geometry defined by a transmitter and the hydrophones may
affect the accuracy and probability of producing a position solution
(Cote et al., 1998; Niezgoda et al., 2002). To map spatial variation in position solution fractions within the array, using data from Deployments
G and H, we calculated the position solution fractions of 14 transmitters
periodically moved to 63 different locations within the array. Because
transmitters were not at all 63 locations simultaneously, spatial
variation is confounded with temporal variation in aquatic acoustic
conditions, for example changes in water flow or sea state. The impacts
of such temporal changes are not well described. Nevertheless, strong
systematic spatial bias should be discernible, if present.
2.4.5. Sound speed
The speed of a single transmission through water affects the
differences in arrival times at each hydrophone (Niezgoda et al.,
2002). When processing data through ALPS, only one sound speed
can be specified, even though temporal changes or spatial differences
in environmental conditions, such as water temperature, will change
the actual sound speed. In our system water conditions are generally
spatially homogeneous. To evaluate the effect of temporal sound
speed changes or estimation errors on the probability and accuracy
of position solutions, we examined data from the four beacons of
Deployment K, which experienced a relatively constant temperature
(29.9 to 31.6 °C). We calculated the nominal speed of sound in
water (1545 m s−1 at 30 °C; Wilson, 1960). We also calculated the
sound speeds corresponding to 20, 25, 35, and 40 °C and ran ALPS
once for each of the five sound speeds (1521, 1533, 1545, 1554,
1562 m s−1). We calculated the change in position solution fractions
and the displacement of the mean position solution location for each
beacon.
2.4.6. Hydrophone positions
To evaluate the impact of errors in hydrophone GPS position
estimates (i.e., researchers recording incorrect or inaccurate GPS locations for hydrophones) on position solution fractions and mean position
solutions, using the four beacons from Deployment K, we ran ALPS once
using the best GPS position estimate of the central hydrophone, then
five more times, artificially displacing the central hydrophone 2, 4, 6,
8, and 10 m to the east. That is, we did not physically move the hydrophone, instead we changed the position entered to the ALPS software.
For each artificial displacement we calculated the change in the overall
position solution fraction and the displacement of the mean position
solution location for each beacon.
2.5. Autonomous underwater vehicle: REMUS
To assess array performance of a moving transmitter, at the beginning of telemetry array Deployment C on 10 Oct 2008, we attached a
sensor tag to the underside of a propeller-driven autonomous underwater vehicle (AUV, REMUS-100, Hydroid, Inc., Pocasset, MA, 36 kg,
1.6 m length, 0.2 m diameter). The AUV navigated itself by computing its range to two acoustic transponders, thus vehicle navigation
accuracy depended upon the accuracy of transponder placement relative to positions specified during mission planning. Transponders
were deployed on weighted/buoyed lines from the boat using the
Garmin 76 with 2 m accuracy. Prior to deployment, the AUV was set
to follow a course at a constant depth of 9 m that passed through
the array along a linear path, then again along a path making one
125° turn. Repeating this course at 1.8 m s−1, the AUV passed
through the array multiple times spending a total of almost 17 min
50
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
within the array. While the AUV self noise and that of its sensors does
not appreciably interfere with detection and decoding of even distant
tags (Grothues, 2009; Grothues et al., 2010), the hull and propeller
wash produce a baffling effect that could challenge detection of a
tag attached directly to it by distant receivers behind it, creating a
geometrical bias in the probability of simultaneous detections by
opposite hydrophones and therefore trilateration.
For the tag attached to the AUV, using the same data processing and
filtering steps as described, we calculated the fraction of transmissions
resulting in position solutions and the degree of agreement with the
AUV's self-reported position estimates. As with stationary transmitters,
to describe the impact of changes or errors in sound speed we calculated the speed of sound at the measured water temperature and temperatures above and below (10, 15, 20, 25, 30 °C). We ran ALPS once for
each of the five corresponding sound speeds (1490, 1507, 152, 1535,
and 1546 m s−1). We calculated the change in position solution fractions and examined the distributions of the differences between position solutions from the nominal sound speed and from shifted sound
speeds. To describe the impact of errors in hydrophone GPS position estimates, we repeated the steps described for stationary transmitter
analyses, running ALPS separately for each artificial displacement of
the central hydrophone 2, 4, 6, 8, and 10 m to the east. We calculated
the change in position solution fractions and examined the distributions
of differences.
3. Results
3.1. Detection fraction
For each transmitter/hydrophone pair we calculated the overall detection fraction (Fig. 2) and the hourly detection fractions (e.g., Fig. 3a).
Given the amount of hourly variability, we identified each detection
fraction as based on more or less than 6 h of data. In general, there
was no consistent change in detection fraction across the distances tested. At distances between 50 and 200 m there was no clear decreasing
pattern. Comparing the hourly detection fraction of one transmitter by
all hydrophones (or similarly the hourly detection fraction of all transmitters by one hydrophone) during individual deployments showed
that during some periods, hourly detections were uniformly high or
low, while at other times there was little consistency. These temporal
relationships were complicated by the spatial relationship of the hydrophones; for example, in the first 30 h of Deployment L (Fig. 3a, shaded
region) there was a drop in detections of the central beacon by the central, north, and east hydrophones. In contrast, during hours 210 to 280
only the east and south hydrophones showed a drop in detections.
Such patterns presumably reflect temporal changes in the spatial structure of the acoustic environment.
3.2. Position solution fraction
With stringent filtering (CN b 1.5), on average 44% of transmissions resulted in position solutions for central transmitters and 13%
for marginal transmitters (overall range: 0 to 94%; 4a and b). Examining the Northing location of position solutions and corresponding CN
values, relative to the GPS Northing estimate (e.g., Fig. 4c and d), we
found that as CN filtering values decreased, a higher percentage of
remaining position solutions were accurate to 2 m or less. This
analysis highlights the interplay between the choice of the desired accuracy, strictness of filtering, number of remaining position solutions,
and probability that a given position solution meets that accuracy.
To compare array performance during Deployment A (collected in
symbol mode without PSR) with other deployments (collected in code
mode with PSR), using only data from Deployments I–O we calculated
the overall mean percent increase in position solution fraction due to
PSR to be 527% (range: 0 to 5871%). We increased position solution fractions for Deployment A by this amount (Fig. 5). For both unfiltered and
Fig. 2. Overall detection fractions (# detections per # transmissions) by individual hydrophones at various distances. Closed circles and ‘X’s represent detection fractions
from deployments lasting more or less than 6 h, respectively.
filtered data, wider array spacing generally showed lower position solution fractions, with shorter duration deployment showing greater variability (Fig. 5). Again, this pattern was not as strong as expected. There
was substantial variation among deployments, with some wider spaced
deployments performing as well as narrowly spaced deployments. This
is particularly notable for central transmitters on long deployments. As
expected, transmissions from within the array generally produced more
high quality position solutions than from the margins (Fig. 5b; Cooke et
al., 2005; Cote et al., 1998). Hourly position solution fractions ranged
from 0 to 1 (e.g., Fig. 3b) and reflected the temporal variability of hourly
detection fractions (e.g., Fig. 3a), so that low detection fractions by two
or three hydrophones often led to low position solution fractions.
3.3. Position solution accuracy
The first method for assessing position solution accuracy compared
mean position solutions with GPS position estimates. For transmitters
within the hydrophone array, the distance between the mean ALPS
position solution and GPS position estimate was less than 2 or 3 m, 86
and 95% of the time, respectively, with a maximum distance of 11.9 m
(see also Biesinger, 2011). Even for transmitters at array margins,
where performance expectations were lower, 30 and 73% of mean
ALPS position solutions were less than 3 and 10 m from the GPS position estimates, respectively, with a maximum difference of 81 m. The
second assessment method examined position solution consistency
through time. Over an entire deployment, consistency among position
solutions for a single transmitter was high for central transmitters and
for many marginal transmitters, showing central 90% ranges less than
3 m (Biesinger, 2011).
3.4. Spatial variation
Position solution fractions from 63 locations during Deployments
G and H ranged from 0 to 0.98 (Fig. 6) without showing any clear
spatial patterns except decreased fractions near array margins. For
transmitters at array margins, the cloud of position solutions was
generally larger than for transmitters fully within the array. For example, the mean position solution for the beacon at the south hydrophone (Fig. 6 point A) falls between widely spaced lines of individual
position solution locations. In contrast, all individual position solutions are nearly equal to the mean position solution for a central tag
(Fig. 6 point B). Performance patterns from this test were confounded
by the combination of spatial and temporal variations, as well as the
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
51
Fig. 3. Temporal variation in hourly detection (# detections per # transmissions) and hourly position solution fractions (# position solutions per # transmissions). a) Hourly detection
fractions of the beacon at the central hydrophone by each hydrophone during Deployment L. Positioning of the five graphs reflects actual deployment geometry. b) Hourly position
solution fractions of the same transmitter. As examples of the complex relationship between detections and position solutions, two time periods are highlighted as shaded regions.
During hours 0–30, the center, north, and east hydrophones show detection decreases; then during hours 210 –280, the east and south hydrophones show detection decreases. Both
periods show resultant position solution decreases.
Fig. 4. Relationship between condition number, CN, and position solution fraction and accuracy. a) Each curve shows the fraction of position solutions removed by filtering out
points with progressively smaller condition numbers. b) The same curves as in a), below CN = 3, show the range of the fraction of position solutions remaining (0 to 0.94)
when filtered at CN = 1.5. c) The shade of each hexagonal box represents the number of position solutions for the central beacon of Deployment A sharing Northing and CN combinations, up to CN = 20. d) The same data as in c) plotted up to only CN = 3. Of 64,611 total position solutions, after filtering at CN = 1.5, 71% remain and of those, 97% are within
1 m of the GPS Northing estimate (solid dashed lines).
52
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
Fig. 5. Mean position solution fractions from array deployments at various spacings. To represent the range of quality of position solutions, a) and b) show unfiltered and filtered
(CN b 1.5) solutions, respectively. Array spacing indicates the distance from center to outer hydrophones. Closed and open circles represent central and marginal transmitters,
respectively, in deployments longer than 6 h. Position solution fractions from deployments shorter than 6 h are indicated by ‘X’s, these are primarily from the spacing and internal
performance trials. Closed and open triangles indicate position solution fractions of central and marginal transmitters, respectively, in Deployment A, adjusted to estimate the potential increase if PSR had been possible. At each array spacing, data points have been slightly offset (jittered) horizontally to distinguish points.
short duration of the trial. Longer duration trials would reveal longer
term spatial variation in performance but obscure shorter term temporal variation.
maximum displacement of the mean from the nominal location was
18.9 m though most were within 5 m.
3.6. Hydrophone positions
3.5. Sound speed
For the central beacon, the largest position solution fraction decrease,
from its maximum, 0.69, at the nominal sound speed (1545 m s−1
at 30 °C) to 0.56 (at 1533 m s−1 at 25 °C) was 17%. The largest displacement in the mean position solution location was 1.4 m (Biesinger,
2011). For marginal beacons, there was little change in the already low
position solution fractions, ranging between 0.001 and 0.06. The
The position solution fraction for the central, north, and south
beacons increased slightly (e.g., 3% from 0.69 to 0.71 for the central
beacon) with a 2 m artificial eastward displacement, then dropped
(e.g., 17% to 0.57 for the central beacon) with further displacements
(Biesinger, 2011), suggesting that the 2 m artificial displacement
corrected an error in the GPS position estimate of the central or
other hydrophone, effectively bringing the estimated array geometry
into closer agreement with the true deployment geometry. With an
artificial displacement of 10 m the mean position solution location
of the central beacon moved 1.9 m. The marginal beacons' position
solution fractions at all artificial displacements ranged between
0.001 and 0.05; the maximum displacement of the mean location
was 26 m.
3.7. Autonomous underwater vehicle: REMUS
Fig. 6. Spatial variation in position solution fractions during the internal performance
trial, Deployments G and H. Open circles show locations of mean position solutions
for individual transmitters. The size of the circle indicates the position solution fraction. Closed circles show target transmitter deployment locations where all position
solutions were removed during filtering. All position solutions for two transmitters
(Points A and B) are indicated by ‘X’s to compare differences in consistency of central
and marginal position solution locations over time. For point B, the ‘X’s all occur at the
same location, creating a solid dot inside the mean position solution circle.
The tag attached to the AUV, spent 1003 s within the array over the
course of several passes (Fig. 7). Of the 501 transmissions sent during
this time, 105, or 22%, of these resulted in position solutions. For each
position solution we calculated the distance between it and the AUV's
self-calculated position. These distances ranged from 0.38 to 7.15 m,
with 88 and 65% being less than 3 and 2 m, respectively. The position
solution fractions obtained when running ALPS using different sound
speeds did not change substantially, ranging from 20 to 22%. Also, running ALPS using different sound speeds had little effect on differences
between ALPS position solutions and the AUV's self-reported position
estimates: between 79 and 86% of position solutions were within 3 m
of the AUV's position estimates, and between 54 and 64% were within
2 m. As artificial displacements of the central hydrophone increased
from 2 to 10 m, position solution fractions decreased from 20 to 19%.
At 10 m artificial displacement, the range of differences between ALPS
and the AUV was 3.4 to 17.5 m different. With 2 m artificial displacement, only 42 and 7% position solutions were within 3 and 2 m, respectively, of the AUV's position estimates. With artificial displacements 4 m
or greater, the number position solutions within 3 m of the AUV estimates quickly decreased to zero.
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
53
Fig. 7. Autonomous underwater vehicle self-reported positions (―) and telemetry array calculated position solutions (●) during 8 passes through the acoustic array (▲). Travel
direction was always from east to west. Hash marks at the axes indicate 50 m.
4. Discussion
These results suggest that Lotek's© acoustic positioning system can
provide accurate data on animal fine-scale space use and movement.
It was capable of deployments covering areas larger and farther from
shore than cabled or radio systems (Andrews et al., 2011; Klimley et
al., 2001), being essentially limited only by the power of the transmitter.
Without the need to be connected to and exposed at the surface, the
array can be deployed in deeper waters with less risk of tampering. Autonomous positioning systems are thus well suited for home range or
pathway studies of animal movement ecology in larger, deeper,
farther-offshore environments for longer durations. The results also
54
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
highlight some of the challenges and limitations of acoustic telemetry.
Unrecognized and unmeasured spatial and temporal variations in the
acoustic environment can significantly impact array performance. Researchers must be aware of these limitations and interpret telemetry
data accordingly.
With our configuration and acoustic environment, Lotek's© autonomous positioning telemetry system was able to cover an area up to
300 m across (Fig. 5), calculating transmitter positions with 2 m accuracy or better (e.g., Fig. 4b). The overall fraction of transmissions
giving good position solutions varied substantially, covering the full
range from 0 to 1 when very stringently filtered (Fig. 5). Position solution fractions varied widely from hour to hour (Fig. 3b), but there
was no clear spatial bias (Fig. 6). For stationary transmitters, array
performance was robust to changes or errors in sound speed and
hydrophone position estimates. An acoustically challenging, moving
tag showed poorer array performance and more sensitivity to deployment errors.
Temporal and spatial variations in detections and position solutions,
observed even in our relatively controlled deployments, have important
implications for animal studies where tag locations are not independently known. Temporal variability leads to non-uniform sampling,
with successive time periods experiencing different sampling intensities. Such unrecognized variation can lead to incorrect interpretations
of home range size, habitat use, or movement paths, especially when
temporal or spatial patterns exist in animal space use. One potential
aid to describing and compensating for performance variability during animal studies is the use of stationary sentinel transmitters (e.g., beacons),
providing an index of the acoustic environment and temporal sampling
bias. However, even multiple sentinels are unlikely to adequately
describe spatial variation and thus are unlikely to be appropriate for
simple temporal or spatial weighting.
The potential for periods of poor array performance means that
researchers must take care to design telemetry deployments that
meet minimum performance requirements. For example, in a study examining habitat use and home range size of a slow moving fish species,
knowing individual positions every 10 min might provide sufficient
temporal resolution. Tagging individuals with transmitters with a
10 s burst interval and subsampling position solutions to obtain only
one position every 10 min are likely to provide sufficient temporal
resolution even during times of poor array performance. In spite of
such logistical challenges, acoustic positioning technology has been
successfully used in animal space-use studies (Biesinger, 2011;
Biesinger et al., 2013; Espinoza et al., 2011a; Grothues et al., 2012).
Though the power of the transmitter and the distance between it
and a hydrophone clearly affect detections (Kilfoyle and Baggeroer,
2000), we did not observe a clear decrease over the range of 50 to
250 m (Fig. 2). Similarly, there was no clear decrease in array performance with increasing array spacing (Fig. 5), though most deployments used a 100 m spacing. There is an expected and observed
decay in signal sound pressure level (SPL) with distance following
the inverse power law (Grothues, 2009; Grothues et al., 2012), but
the signal interpretation (and therefore logging) of CDMA does not
require high SPL. Therefore, logging effectiveness should have been
(and here appears to be) poorly related to distance, but only until a
distance threshold is reached, and this threshold was beyond the
tested hydrophone distances. Future arrays might consider wider
spacing, but the detection fraction drop off is likely to be precipitous
at the threshold and to be reached more often in dynamic acoustic
positions. Therefore, it is imperative that the choice of array coverage,
transmitter power, and array performance be made with care and
tested prior to application.
Array performance was relatively robust to sound speed and
hydrophone position estimate errors, with 20 °C temperature differences and 10 m hydrophone position errors decreasing position solution fractions by, at worst, 17% and moving the average position
solution location, at worst, 1.4 m for central transmitters. However,
this issue is more complicated than our tests reveal. Because acoustic
positioning telemetry uses sound speed and distances between hydrophones, an error in one may compound or compensate for another
in spatially variable ways, with resulting variation in spatial impacts
on array performance. For example, the increase in the position
solution fraction with a 2 m artificial displacement of the central hydrophone suggests a situation where the GPS estimate of either the
central or other hydrophone position was wrong and the artificial
movement brought the relative geometry closer to the true hydrophone arrangement. Errors in central vs. marginal hydrophone
estimates could affect performance differently, and errors in multiple
position estimates will combine in complicated ways depending on
which hydrophones have errors and which are used in the calculation
of any given position solution. In practice these observations suggest
that taking extra time and care to ensure that hydrophone position
estimates are as accurate as possible will increase the frequency and
accuracy of position solutions and increase the quality of fish
space-use descriptions.
Because the distances (and directions) between hydrophones, and
not their absolute positions, are used to calculate position solutions, errors in one position estimate, or sound speed estimate, may compensate
or compound errors in another. The complexity of these interactions
and the challenge of compensating for errors highlight the importance
of obtaining the best possible sound speed and hydrophone position estimates. This is particularly highlighted by impacts to the AUV-borne
tag.
The hollow hull and propeller wake of the AUV challenged the array
function for the tag it carried relative to expectations for a tagged fish.
The fraction of position solutions for the AUV-borne tag was only
about half that of central stationary tags, 22 versus 44%, respectively
and about equal to that of marginal stationary tags. Position solution
accuracy decreased significantly as artificial hydrophone displacement
increased, emphasizing the importance of accurate equipment placement. In testing a different positioning telemetry technology, Espinoza
et al. (2011b) reported a small but statistically insignificant decrease
in position accuracy for transmitters towed by a boat compared to
stationary transmitters. They reported large differences in position
estimates between positioning telemetry and active acoustic tracking
for tags in moving shark, though it is unclear how much of the differences were due to each positioning method. Our results and those of
Espinoza et al. (2011b) suggest that positioning telemetry arrays
perform differently for stationary and moving transmitters, suggesting
caution in biological studies. The challenge of confirming telemetry
array performance in biological studies depends on the ability to obtain
independent and accurate position estimates of free moving animals,
though again we point to animal space-use studies successfully using
acoustic positioning technology (Biesinger, 2011; Biesinger et al.,
2013; Espinoza et al., 2011a; Grothues et al., 2012).
Though we have assessed performance with overall and hourly
measures, most telemetry studies rely on the accuracy of individual
position solutions or the probability of obtaining position estimates
with a desired frequency, similar to our AUV-borne tag analysis. Our
position solution fractions may be seen as estimates of position
solution probabilities for similar array deployments, i.e., what is the
probability that a given position solution is accurate to 2 m or that a
deployment will produce accurate position solutions at least every
minute. Furthermore, depending on the goals of a particular study,
for example, describing overall space or habitat use, or analyzing individual animal movement paths, higher frequency or greater accuracy
may be more important. To assess array performance in this light, one
must consider the trade-off between the frequency and accuracy of
position solutions, and the probability that each solution meets the
desired accuracy; each new array configuration or acoustic environment may perform differently.
The capability of new acoustic positioning telemetry technologies to
collect animal movement data at finer resolutions, over longer times,
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
and in less accessible locations makes it possible to address a wider
range of movement and ecological questions than ever before. With
the ability to simultaneously track many individuals, this technology
makes it possible to address density-dependent and species–species interactions. In conjunction with detailed aquatic and environmental
data, it is possible to explore fine-scale, habitat-dependent processes.
As emphasized by others (Andrews et al., 2011; Espinoza et al.,
2011b), the choice of telemetry system and design should be driven
by the study objectives, animal behavioral characteristics, and user
requirements. For example, with the Vemco© VPS array, the manufacturer retains responsibility for filtering and ensuring the quality of telemetry results, leaving the researcher to interpret the behavioral and
ecological significance of data. This compares with the Lotek© system,
where the researcher has access to raw data and responsibility to correctly interpret both quality and biological significance of telemetry
output, potentially allowing more insight into the aquatic and acoustic environment. Regardless of the chosen system, the research question must drive the study design and the telemetry deployment
details. For example, in a study of habitat use or home range size of
wide ranging animals, one might choose larger spatial coverage and
longer duration (i.e. less frequent tag transmissions extending battery
life), sacrificing smaller scale animal movements. In a study of finer
resolution behaviors, like movement pathways or response to
individual predator encounters, to obtain frequent, highly accurate
position estimates, one could cover a smaller area (or increase the
number of hydrophones) and use short-lived tags with more frequent
transmissions. The increasing availability of spatial data will give researchers an increasingly spatial perspective of animal interactions
and improve our understanding of ecological pattern and process.
Acknowledgments
We thank A. Valle-Levinson for the use of the ADCP, T. Bacheler,
E. Pierce, and M. Meadows for help with fieldwork, and M. Sisak and
S. Moussavi at Lotek Wireless©. Z.B. was supported as a D.M. Smith
Marine Fellow and an E.T. York Presidential Fellow at the University
of Florida. The National Marine Fisheries Service, MARFIN Program
(Grant nos. NA04NMF4330075 and NA07NMF4330123 to W.J.L.)
partially supported this work. Further support came from the Disney
Worldwide Conservation Fund, the American Academy of Underwater
Sciences, and the Aylesworth Foundation. We thank two anonymous
reviewers for several suggested improvements. [ST]
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://
dx.doi.org/10.1016/j.jembe.2013.06.007.
References
Andrews, K.S., Tolimieri, N., Williams, G.D., Samhouri, J.F., Harvey, C.J., Levin, P.S., 2011.
Comparison of fine-scale acoustic monitoring systems using home range size of a
demersal fish. Mar. Biol. 158, 2377–2387.
Bégout, M.-L., Lagardère, J.-P., 1995. An acoustic telemetry study of seabream (Sparus
aurata L.): first results on activity rhythm, effects of environmental variables and
space utilization. Hydrobiologia 300 (301), 417–423.
Bergé, J., Capra, H., Pella, H., Steig, T., Ovidio, M., Bultel, E., Lamouroux, N., 2012. Probability
of detecting and positioning error of a hydro acoustic telemetry system in a fastflowing river: intrinsic and environmental determinants. Fish. Res. 125–126, 1–13.
Biesinger, Z., 2011. Quantifying the Relationships among Habitat, Behavior and Performance of Gag, Mycteroperca microlepis. Dissertation University of Florida.
Biesinger, Z., Bolker, B.M., Marcinek, D.M., Lindberg, W.J., 2013. Gag (Mycteroperca
microlepis) space-use correlations with landscape structure and environmental
conditions. J. Exp. Mar. Biol. Ecol. 443, 1–11.
Brown, H., Benfield, M.C., Keenan, S.F., Powers, S.P., 2010. Movement patterns and
home ranges of a pelagic carangid fish, Caranx crysos, around a petroleum platform
complex. Mar. Ecol. Prog. Ser. 403, 205–218.
Carey, F.G., Lawson, K.D., 1973. Temperature regulation in free-swimming bluefin tuna.
Comp. Biochem. Physiol. 44A, 375–392.
55
Clements, S., Jepsen, D., Karnowski, M., Schreck, C.B., 2005. Optimization of an acoustic
telemetry array for detecting transmitter-implanted fish. N. Am. J. Fish. Manag. 25,
429–436.
Collazo, J.A., Epperly, S.P., 1995. Accuracy tests for sonic telemetry studies in an estuarine environment. J. Wildl. Manage. 59, 181–188.
Cooke, S.J., Niezgoda, G.H., Hanson, K., Suski, C.D., Phelan, F.J.S., Tinline, R., Philipp, D.P.,
2005. Use of CDMA acoustic telemetry to document 3-D positions of fish: relevance
to the design and monitoring of aquatic protected areas. Mar. Tech. Soc. J. 39, 17–27.
Cote, D., Scruton, D.A., Niezgoda, G.H., McKinley, R.S., Rowsell, D.F., Lindstrom, R.T.,
Ollerhead, L.M.N., Whitt, C.J., 1998. A coded acoustic telemetry system for high
precision monitoring of fish location and movement: application to the study of
nearshore nursery habitat of juvenile Atlantic cod (Gadus morhua). Mar. Tech.
Soc. 32, 54–62.
Ehrenberg, J.E., Steig, T.W., 2002. A method for estimating the “position accuracy” of
acoustic fish tags. ICES J. Mar. Sci. 59, 140–149.
Espinoza, M., Farrugia, T.J., Lowe, C.G., 2011a. Habitat use, movements and site fidelity
of the gray smooth-hound shark (Mustelus californicus Gill 1863) in a newly
restored southern California estuary. J. Exp. Mar. Biol. Ecol. 401, 63–74.
Espinoza, M., Farrugia, T.J., Webber, D.M., Smith, F., Lowe, C.G., 2011b. Testing a new
acoustic telemetry technique to quantify long-term, fine-scale movements of
aquatic animals. Fish. Res. 108, 364–371.
Feary, D.A., Burt, J.A., Bartholomew, A., 2011. Artificial marine habitats in the Arabian
Gulf: review of current use, benefits and management implications. Ocean Coast.
Manag. 54, 742–749.
Geraldi, N.R., Powers, S.P., 2011. Subtle changes in prey foraging behavior have cascading
effects in a shallow estuary. Mar. Ecol. Prog. Ser. 427, 51–58.
Grewal, M.S., Andrews, A.P., 2008. Kalman Filtering: Theory and Practice Using
MATLAB, 3rd ed. Wiley Press, USA.
Grothues, T.M., 2009. A review of acoustic telemetry technology and a perspective on
diversification in coastal tracking arrays. In: Nielsen, J.L., Arrizabalaga, H.,
Fragoso, N., Hobday, A., Lutcavage, M., Sibert, J. (Eds.), Tagging and Tracking of
Marine Animals with Electronic Devices. Springer, pp. 77–90.
Grothues, T.M., Dobarro, J., Eiler, J., 2010. Collecting, interpreting, and merging fish
telemetry data from an AUV: remote sensing from an already remote platform.
Proceedings of AUV 2010. IEEExplore, Publisher, Monterey Bay, CA (9 pp.).
Grothues, T.M., Able, K.W., Pravatiner, J.A., 2012. Winter flounder (Pseudopleuronectes
americanus Walbaum) burial in estuaries: acoustic telemetry triumphs and tribulations.
J. Exp. Mar. Biol. Ecol. 438, 125–136.
Hanson, K.C., Cooke, S.J., Suski, C.D., Niezgoda, G., Phelan, F.J.S., Tinline, R., Philipp, D.P.,
2007. Assessment of largemouth bass (Mictopterus salmoides) behaviour and activity
at multiple spatial and temporal scales utilizing a whole-lake telemetry array.
Hydrobiologia 582, 243–256.
Heupel, M.R., Simpfendorfer, C.A., Hueter, R.E., 2004. Estimation of shark home ranges
using passive monitoring techniques. Environ. Biol. Fish. 71, 135–142.
Heupel, M.R., Semmens, J.M., Hobday, A.J., 2006. Automated acoustic tracking of aquatic
animals: scales, design and deployment of listening station arrays. Mar. Freshw.
Res. 57, 1–13.
Heupel, M.R., Simpfendorfer, C.A., Olsen, E.M., Moland, E., 2012. Consistent movement
traits indicative of innate behavior in neonate sharks. J. Exp. Mar. Biol. Ecol.
432–433, 131–137.
Johnson, R., Bester, M.N., Dudley, S.F.J., Herman Ooshuizen, W.H., Meÿer, M., Hancke, L.,
Gennari, E., 2009. Coastal swimming patterns of white sharks (Carcharodon
carcharias) at Mossel Bey, South Africa. Environ. Biol. Fish. 85, 189–200.
Kilfoyle, D.B., Baggeroer, A.B., 2000. The state of the art in underwater acoustic
telemetry. IEEE J. Ocean. Eng. 25, 4–27.
Klimley, A.P., Le Boeuf, B.J., Cantara, K.M., Richert, J.E., Davis, S.F., Van Sommeran, S.,
2001. Radio-acoustic positioning as a tool for studying site-specific behavior of
the white shark and other large marine species. Mar. Biol. 138, 429–446.
Lagardère, J.-P., Ducamp, J.-J., Favre, L., Mosneron Dupin, J., Spérandio, M., 1990. A
method for the quantitative evaluation of fish movements in salt ponds by acoustic
telemetry. J. Exp. Mar. Biol. Ecol. 141, 221–236.
Lucas, M.C., Baras, E., 2000. Methods for studying spatial behaviour in freshwater fishes
in the natural environment. Fish Fish. 1, 283–316.
Niezgoda, G., Benfield, M., Sisak, M., Anson, P., 2002. Tracking acoustic transmitters by
code division multiple access (CDMA)-based telemetry. Hydrobiologia 483, 275–286.
O'Dor, R.K., Andrade, Y., Webber, D.M., Sauer, W.H.H., Roberts, M.J., Smale, M.J., Voegeli,
F.M., 1998. Applications and performance of radio-acoustic positioning and
telemetry (RAPT) systems. Hydrobiologia 371–372, 1–8.
Parker, R.O., Colby, D.R., Willis, T.D., 1983. Estimated amount of reef habitat on a
portion of the U.S. south Atlantic and Gulf of Mexico continental shelf. Bull. Mar.
Sci. 33, 935–940.
Parsons, D.M., Babcock, R.C., Hankin, R.K.S., Willis, T.J., Aitken, J.P., O'Dor, R.K., Jackson,
G.D., 2003. Snapper Pagrus auratus (Sparidae) home range dynamics: acoustic
tagging studies in a marine reserve. Mar. Ecol. Prog. Ser. 262, 253–265.
R Core Development Team, 2012. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria3-900051-07-0
(http://www.R-project.org. Accessed 15 April 2012).
Semmens, B.X., 2008. Acoustically derived fine-scale behaviors of juvenile Chinook
salmon (Oncorhynchus tshawytscha) associated with intertidal benthic habitats in
an estuary. Can. J. Fish. Aquat. Sci. 65, 2053–2062.
Simpfendorfer, C.A., Olsen, E.M., Heupel, M.R., Moland, E., 2012. Three-dimensional
kernel utilization distributions improve estimates of space use in aquatic animals.
Can. J. Fish. Aquat. Sci. 69, 565–572.
Simpfendorfer, C.A., Heupel, M.R., Hueter, R.E., 2002. Estimation of short-term centers
of activity from an array of omnidirectional hydrophones and its use in studying
animal movements. Can. J. Fish. Aquat. Sci. 59, 23–32.
56
Z. Biesinger et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 46–56
Voegeli, F.A., Smale, M.J., Webber, D.M., Andrade, Y., O'Dor, R.K., 2001. Ultrasonic
telemetry, tracking and automated monitoring technology for sharks. Environ.
Biol. Fish. 60, 267–281.
Welsh, J.Q., Fox, R.J., Webber, D.M., Bellwood, D.R., 2012. Performance of remote acoustic receivers within a coral reef habitat: implications for array design. Coral Reefs
31, 693–702.
Wilson, W.D., 1960. Equation for the speed of sound in sea water. J. Acoust. Soc. Am. 32,
1357.
Wolcott, T.G., 1995. New options in physiological and behavioural ecology through
multichannel telemetry. J. Exp. Mar. Biol. Ecol. 193, 257–275.
Zeller, D.C., 1999. Ultrasonic telemetry: its application to coral reef fisheries research.
Fish. Bull. 97 (4), 1058–1065.