350 
Fishery Bulletin 11 7(4) 
1 acoustic transmitter. The ADLs (G6a+\ Cefas Technology 
Ltd., Lowestoft, UK) were programmed to record triaxial 
acceleration at 25 Hz, depth at 1 Hz, and temperature at 
0.033 Hz. If possible, ADLs were paired with externally 
attached continuous acoustic transmitters (V9, Vemco, 
Bedford, Canada) with 5-s delays, to facilitate recovery 
of the ADLs through recapture after 5-7 d. Most of these 
tags were attached to the first dorsal fin directly by using 
monofilament (Fig. 2A). Alternatively, 2 ADLs were indi¬ 
rectly attached to the dorsal fin of individuals by tethering 
them to custom float packages (Fig. 2B). Float packages 
incorporated both a very high frequency (VHF) transmit¬ 
ter (MM110, Advanced Telemetry Systems Inc., Isanti, 
MN) and a galvanic timed release (GTR; International 
Fishing Devices Inc., Jupiter, FL). The GTRs were set to 
corrode after approximately 7 d, allowing the float package 
to detach and rise to the surface, where the VHF transmit¬ 
ter facilitated recovery following the technique of Lear and 
Whitney (2016), and eliminating the need for recapture. 
All animals were tagged with a coded 69-kHz acoustic 
transmitter (V9), which was epoxied to a rototag and 
externally attached to the second dorsal fin following the 
methods of Poulakis et al. (2013). These transmitters 
emitted unique acoustic signals on a random delay 
between 80 and 160 s to minimize interference of signals 
from multiple tags and to maximize battery life. These 
transmitters communicated to an array of 35 omnidirec¬ 
tional VR2W acoustic receivers (Vemco) placed through¬ 
out the Peace River estuary (Fig. 1). They were also used 
to facilitate recovery of the ADLs through recapture if 
necessary. After tagging, smalltooth sawfish were released 
at their site of capture. 
Work with animals was conducted under endangered 
species permit no. 15802, issued by the National Marine 
Fisheries Service. 
Data processing and analyses: data loggers 
Once ADLs were recovered, raw data were downloaded 
from them and prepared for analysis by using Igor Pro 
(vers. 6.22; WaveMetrics Inc., Portland, OR) and the appli¬ 
cation Ethographer (Sakamoto et al., 2009). Data from the 
first 12 h of ADL deployments were excluded to eliminate 
any behavioral effects of capture stress following tagging 
and release (Whitney et al., 2012). Static and dynamic 
acceleration were separated by using a 3-s box smoother, 
which was sufficient to remove the tailbeat signal from the 
static acceleration traces (Shepard et al., 2008), and both 
components were used to analyze body movement and 
position throughout the monitoring periods. K-means clus¬ 
tering analyses were run on the sway (tailbeat) accelera¬ 
tion axis to separate resting and active periods (Sakamoto 
et al., 2009; Whitney et al., 2010). Overall dynamic body 
acceleration (ODBA) (Wilson et al., 2006; Gleiss et al., 
2011) was calculated as the sum of the absolute value of 
1 Mention of trade names or commercial companies is for identi¬ 
fication purposes only and does not imply endorsement by the 
National Marine Fisheries Service, NOAA. 
Figure 2 
Photographs of acceleration data loggers (ADLs) attached to 
the first dorsal fins of 2 of the 10 smalltooth sawfish ( Pris- 
tis pectinata ) caught and tagged in the Peace River, Florida, 
from May 2014 through November 2015. (A) An ADL on an 
individual that was 1.7-m stretch total length (STL), paired 
with a continuous acoustic transmitter (not shown) to enable 
recapture and recovery of the logger. (B) A 1.8-m-STL indi¬ 
vidual, with the logger tethered to a float package that incor¬ 
porated a very high frequency transmitter. A galvanic timed 
release was used to attach these float packages, allowing 
recovery of the ADL without recapture. 
the 3 dynamic acceleration axes. We used ODBA to deter¬ 
mine burst events, calculating the 98th percentile of ODBA 
for each individual by using means for each 1-s segment of 
data and subsequently classifying burst events as periods 
during which ODBA was continuously above this threshold 
for more than 3 s (Gleiss et al., 2017). 
Because of permitting restrictions, no validation (ground- 
truthing) trials were run for this species to specifically link 
behaviors to their corresponding acceleration traces; how¬ 
ever, ground-truthing has been conducted for the large- 
tooth sawfish (Pristis pristis ) (senior author and A. Gleiss, 
unpubl. data), as well as for the lemon shark (Negaprion 
brevirostris) (Brewster et al., 2018). These studies estab¬ 
lished that chafe behavior (dorsal rubbing) is defined by a 
distinctive W-shaped trace in the heave acceleration axis. 
Therefore, burst events with similar distinctive W-shaped 
