Relationship between visual counts and call detection rates of gray whales (Eschrichtius robustus) in Laguna San Ignacio, Mexico
(Color online) Map of Laguna San Ignacio (LSI), including 2008 and 2010 deployment location of acoustic hydrophone (solid triangle), which is similar to the cabled hydrophone location used by (Dahlheim, 1987). Visual surveys of gray whale population studies are subdivided into three zones indicated by dashed lines. An additional site in 2010 (filled circle) was deployed 1.5 km away to evaluate propagation conditions.
(Color online) Example spectrograms of gray whale signals in Laguna San Ignacio recorded in 2008. (a) S1 call, showing a train of pulses with energy between 100 Hz and 1 kHz with peak frequencies between 300 and 800 Hz; (b) S4 call, with pulse bandwidth between 100 and 1500 Hz, with peak frequencies between 150 and 300 Hz, and total durations between 0.5 and 1.5 s; (c) S3 call, ranging between 90 and 300 Hz and with call durations between 1 and 2 s. The S1 and S4 call spectrograms are imaged using a 256 pt FFT with 75% overlap on a Hanning-windowed time series sampled at 6.25 kHz; the S3 spectrogram uses a 1024 pt FFT. Image intensity is in units of power spectral density (dB re 1 μPa2/Hz).
(Color online) Deployment configuration of bottom-mounted acoustic recording station. Recorders are anchored at 8–15 m below the surface, separated by 33 m of polypropylene rope. An acoustic transponder is attached between a recorder and anchor to facilitate recovery. The second recorder was intended as a backup, so data from only one recorder is used in this study.
(Color online) Cumulative histograms of gray whale call type distributions over a 24-h period using six non-consecutive days of data between February 10 and March 10, 2008. The solid line indicates the mean call rate per hour, averaged over the 6 days. The dashed lines indicate the 5 and 95% confidence limits of the mean hourly call rate synthesized from 1000 simulations of N random times drawn from a uniform 24-h distribution, where N is the total number of calls observed in each subplot. (a) S1 calls, with a mean hourly rate of 198 calls per hour (solid line) and confidence limits of 174 and 220 calls per hour; (b) S4 calls, with a mean hourly rate of 29 calls per hour and confidence limits of 19 and 37 calls per hour; (c) S3 calls, with a mean hourly rate of 21 calls per hour and confidence limits of 13 and 27 calls per hour.
(Color online) 2D matrix quantifying boat transits detected by the recorder during daylight hours over the 6 days analyzed. The intensity scale indicates the transits detected during that hour and day. The maximum number of transits per hour (18) occurred on March 4 at 14:00 and then slightly less on February 28 at 11:00.
(Color online) Visual and raw acoustic call counts of Eastern Pacific gray whales in LSI in 2008, measured on 6 days over 30 days. (a) Visual counts of mother-calf pairs (dotted), single animals (dashed), and total animals (solid) in the lower and middle zones of the lagoon. Solid dots indicate census counts from the lower zone only, while open squares show the combined counts from the lower and middle zones. (b) Raw call detection counts over 24 h periods that encompass times of visual surveys. Three types of calls are plotted: S1 (circle), S4 (square), and S3 (triangle). The S1 counts are further subdivided into daytime only (dotted line), nighttime only (dashed line), and total daily count (solid line).
(Color online) Modeling and measurements of propagation loss factor α in Eq. (1): (a) Modeled transmission loss (dashed lines) as a function of propagation range, averaged over 25 frequencies evenly spaced between 350 and 750 Hz. Two source depths of 0.25 and 5 m are shown along with the transmission loss modeled by the best-fit α value (solid lines). The modeled environment was a 10 m deep Pekeris waveguide with 1650 m/s bottom speed and 1.5 g/cc density, typical values for sand. The receiver depth was 10.25 m, simulating a buried instrument. (b) Empirical measurements of integrated received levels between 100 and 500 Hz during a direct approach by a motorboat toward the recorder during a 2010 deployment. A single point in this plot was estimated by averaging four 4096 point FFT snapshots, overlapped 50%, which translates into 1.15 s of data using the 12.5 kHz sampling rate. The power spectral density was integrated between 100 and 500 Hz (the dominant frequency band of the boat) and then converted to logarithmic units, yielding units of dB re 1 μPa. The best-fit logarithmic propagation law is shown as a solid line. Beyond 300 m range the boat noise was buried in the background.
(Color online) Background noise levels in the lagoon, with resulting call multiplier factors. (a) Background noise levels in dB re 1 μPa rms, integrated between 350 and 750 Hz, using a 1024-pt FFT, overlapped 50%, and retaining the minimum level detected every two minutes, to suppress effects of impulsive noise. The diel effects of land and sea breezes are readily apparent. (b) Hourly averages of the data shown in (a), during days of visual surveys. Dashed lines indicate daylight hours, solid lines nighttime hours. The dots indicate the geometric and arithmetic mean of the ambient noise over a 24-h period. (c) Resulting hourly call multiplication factors ([N/Nref]2/α) used in Eq. (1), with α = 2.55 and Nref = 104 dB re 1 μPa (rms), using the same pattern scheme as (b) for indicating daytime and nighttime.
(Color online) Relative changes in visual survey counts, raw S1 acoustic call counts, and noise-adjusted S1 call counts over the observation period. The normalized lower-zone single animal visual count (dashed line) and total lower-zone animal count (solid line) are identical on all plots, and show the number of animals sighted in the lower zone on a given date, divided by the count on Feb. 10, the first observation date. The top row (a, b, c) shows the relative changes in the S1 acoustic call count during daytime (dotted line with circles), nighttime (dashed line with circles) and over the entire 24-h period (solid line with circles). The bottom row (d, e, f) plots the relative changes in the square root of the S1 acoustic call count. The first column (a, d) plots the raw acoustic count data, the second (b, e) and third columns (c, f) show the noise-adjusted call counts using α = 1.65 and 2.55, respectively.
Article metrics loading...
Full text loading...