Figure 1.
Distributions of vocalizing humpback whales and spawning herring populations in Fall 2006.
Spatial distribution of vocalizing humpback whales coincides with the time and location of spawning Atlantic herring distributions in Fall 2006. Humpback whale vocalizations are found to be distributed along the northern flank of Georges Bank, coinciding with dense herring shoals (>0.20 fish/m2, red shaded areas) imaged using active OAWRS system [18] and diffuse herring populations (≈0.053 fish/m2, bounded by magenta line) obtained from conventional fish finding sonar (CFFS) line-transect data from NEFSC Annual Fall Herring Surveys [18], [63]. The green shaded areas indicate the overall humpback whale call rate densities (number of calls/[(min) (50 nmi)2]) measured with our large aperture array. All data represent means between September 22 and October 6, 2006. The dashed magenta line represents the southern bound of the NEFSC survey tracks [18], [63]. The black trapezoid indicates Stellwagen Bank [158].
Figure 2.
Daytime distributions of vocalizing humpback whales and diffuse herring populations.
Spatial distribution of vocalizing humpback whales coincides with the locations of diffuse herring populations during daytime hours. In daylight, the vast majority of the humpback whale vocalizations originate within areas containing diffuse herring populations (≈0.053 fish/m2, bounded by magenta line) [63]. The green shaded areas indicate the daytime humpback whale call rate densities (number of calls/[(min) (50 nmi)2]) measured with our large aperture array. All data represent daytime means between September 22 and October 6, 2006. The dashed magenta line represents the southern bound of the NEFSC survey tracks [18], [63]. The daytime hours are between sunrise and sunset (06:00:01 to 18:00:00 EDT). The black trapezoid indicates Stellwagen Bank [158].
Figure 3.
Nighttime distributions of vocalizing humpback whales and dense herring shoals.
Spatial distribution of vocalizing humpback whales coincides with the locations of dense evening herring shoals during nighttime hours. At night, vocalizing humpback whales become concentrated at and near dense evening herring shoals (>0.20 fish/m2, red shaded areas) that form along the northern flank of Georges Bank and call rates increase dramatically [18]. The green shaded areas indicate the nighttime humpback whale call rate densities (number of calls/[(min) (50 nmi)2]) measured with our large aperture array. All data represent nighttime means between September 22 and October 6, 2006. The magenta line bounds the areas with diffused herring populations (≈0.053 fish/m2). The dashed magenta line represents the southern bound of the NEFSC survey tracks [18], [63]. The data shown are for nighttime hours between sunset and sunrise the next day (18:00:01 to 06:00:00 EDT). The black trapezoid indicates Stellwagen Bank [158].
Figure 4.
Humpback whale call-rate is synchronized with Atlantic herring shoal population density over a diurnal cycle.
(A) Mean humpback whale call rate (black line within gray standard deviation over 15 minute bins) over a diurnal cycle and mean herring shoal areal population density (blue line with standard deviation indicated by the blue error bars) from September 28 to October 3. When the areal population density of the diffuse daytime herring populations reaches a critical threshold of approximately 0.2 fish/m2 (red dashed line) near sunset, the herring population density drastically increases at a rate of roughly 5 fish/m2 per hour [18] to form evening shoals. (B) Diurnal humpback whale call rate follows a synchronous pattern with 0.82 correlation coefficient and 0–15 minute time lag between the two time series in (A). The period from roughly 2–6 EDT contains a data gap.
Figure 5.
Percentage of semi-diurnal period containing different classes of humpback whale vocalizations for day and night.
A roughly three-fold percentage increase is found at night for repetitive non-song calls, which are primarily responsible for the overall diurnal dependence of observed humpback whale vocalizations. Humpback whale songs showed negligible mean variation compared to standard deviations for day (15.7%±18%) versus night (19.1%±15%). Percentages were calculated using the approaches discussed in Section 3.2. The total percentage, the sum of all four categories, exceeds 100% because different call types could occur within overlapping time windows. The “No calls detected”, however, is mutually exclusive with the other categories. Here the daytime hours are between sunrise and sunset (06:00:01 to 18:00:00 EDT) and nighttime hours are between sunset and sunrise the next day (18:00:01 to 06:00:00 EDT).
Figure 6.
Spectrograms of a typical “meow”, “bow-shaped” call and “feeding cry” observed during OAWRS 2006 experiment.
(A) “Meow” is a roughly 1.4 second duration, frequency modulated downsweep signal (570 to 380 Hz) with a center frequency of roughly 475 Hz. (B) “Bow-shaped” call has a roughly 2.4 second duration, downsweep frequency modulated section (510 to 395 Hz) followed by a short upsweep coda with a center frequency of roughly 440 Hz. (C) “Feeding cry” consists of (1) a main section that lasts approximately 3.5 seconds with frequency oscillations between 500 Hz and 540 Hz and (2) a 2 second long frequency-modulated ending section.
Figure 7.
Spectrograms of typical repetitive “meows” observed during OAWRS 2006 experiment in the Gulf of Maine.
Four 70-s time series containing repetitive meows are shown in (A) – (D) recorded 5-s apart, on October 1, 2006 between 19:10:00 EDT and 19:14:55 EDT.
Figure 8.
Spectrograms of a typical repeated humpback whale song theme observed during OAWRS 2006 experiment.
A repeated humpback whale song theme, starting at (A) 23:17:44 EDT and (B) 23:49:01 EDT and each lasting roughly 1 minute, was recorded on October 2, 2006 from a singing humpback whale in the northern flank of Georges Bank.
Figure 9.
Humpback song occurrence rate is constant in the periods “before” and “during” OAWRS survey transmissions.
The mean percentage of a diurnal cycle containing humpback whale song in the periods “before” and “during” OAWRS survey transmissions, as defined in Section 2.2, remains constant, indicating the transmissions had no effect on humpback whale song over the entire passive 400-km diameter survey area of the Gulf of Maine including Stellwagen Bank.
Table 1.
Percentage of time the Risch et al. statistical test [34] incorrectly finds whales respond to sonar when no sonar is present using annual humpback whale song occurrence data reported from single sensor detections at Stellwagen Bank [35] in time-dependent ambient noise.
Table 2.
The Risch et al. statistical test is applied to the same humpback whale song occurrence data reported in Risch et al. [34] over the 33-day period from September 15 to October 17 for 2008 and 2009, with either of these two years as the test year and the other as the control year.
Figure 10.
Reported humpback whale Stellwagen Bank song occurrence [35] shows large natural variations within and across years.
Large natural variations in humpback whale song occurrence reported from single sensor detections at Stellwagen Bank [35] in time-dependent ambient noise within and across years are common in the absence of sonar. Line plots of reported single sensor daily humpback whale song occurrence at Stellwagen Bank in hours/day (A) for the entire year and (B) from September 15 to October 17, in 2006 and 2008 [35]. Many periods lasting roughly weeks where high song occurrence episodes are found in one year but not in another when no sonars are present are indicated by black arrows in (A). The reported reducing change in humpback whale song occurrence, to zero [34], [35], occurred in the “before” period while the OAWRS vessels were inactive and docked on the other side of Cape Cod from Stellwagen Bank, at the Woods Hole Oceanographic Institution, due to severe winds for days before OAWRS transmissions for active surveying began on September 26, 2006, as marked by the black arrow in (B). This shows that Risch et al. [34] analysis violates temporal causality.
Figure 11.
Quantifying large differences in the reported humpback whale song occurrence at Stellwagen Bank [35] across years.
Difference in humpback whale song occurrence reported from single sensor detections at Stellwagen Bank [35] in time-dependent ambient noise across years exceeds that of the “during” period most of the time when no sonars are present. (A) Difference in mean humpback whale song occurrence at Stellwagen Bank over respective 11-day periods with 1-day increment in 2006 and 2008, (B) histogram of difference in mean humpback song occurrence over 11-day periods between 2006 and 2008 when no sonar is present, i.e. excluding the “during” period from September 26 to October 6. Periods when the difference in means of respective 11-day periods is greater than (red dots) and less than (blue dots) that of the “during” period are indicated in (A). The difference in means fluctuates randomly throughout the year, exceeding the “during” period 57.8% of the time (most of the time) when no sonars are present, indicating that there is nothing unusual about such differences, which are actually the norm.
Figure 12.
Reported annual humpback song occurrence at Stellwagen Bank [35] are uncorrelated between years over 11-day periods.
Annual humpback whale song occurrence reported from single sensor detections at Stellwagen Bank [35] in time-dependent ambient noise are uncorrelated over 11-day periods across years. (A) Correlation coefficient between 2006 and 2008 humpback whale song occurrence time series over 11-day period with 1-day increment (B) histogram of the correlation coefficient in (A). The correlation coefficient of the annual humpback whale song occurrence time series over 11-day periods across years obeys a random distribution peaking at zero correlation about which it is symmetric, showing that correlation in trend between years is random and quantitatively expected to be zero with roughly as many negative correlations as positive ones. The correlation coefficient between the humpback whale song occurrence across years smoothly transitions from negative values in the “before” period, showing no similarity or relation in trend between years just before the 2006 OAWRS survey transmission period, to some of the highest positive correlations obtained between years in the “during” period. This demonstrates high similarity and relation in trend between years during the 2006 OAWRS active survey transmission period, which contradicts the results of the Risch et al. [34] study.
Figure 13.
Wind-dependence of mean detection range for single sensor at Stellwagen Bank[34], and OAWRS receiver array.
The green shaded areas indicate the overall vocalizing humpback whale call rate densities (number of calls/[(min) (50 nmi)2]) determined between September 22 and October 6, 2006 by our large aperture receiver array towed along several tracks (black lines). The mean detection ranges for the single sensor at Stellwagen Bank are in blue and for the OAWRS receiver array are in red, where Stellwagen Bank is marked by yellow shaded regions. These detection ranges are determined by the methods described in Section 3.5 given a humpback whale song unit source level of approximately 180 dB re 1 µPa and 1 m which is the median of all published humpback whale song source levels [93], [101], [102], [152]–[154]. The error bars represent the spread in detection range due to typical humpback whale song source level variations (Section 3.5). Under (A) low wind speed conditions vocalizing whales are within the mean detection area for a single Stellwagen Bank sensor but for (B) higher wind speeds most vocalizing whales are outside the mean detection area of the same sensor, which results in reduction of detectable whale song occurrence by the single sensor [34] at Stellwagen Bank.
Figure 14.
Wind-speed increase causes reduction in humpback song occurrence at Stellwagen Bank.
Average wind speed increase from the “before” to the “during” period at Stellwagen Bank causes reduction in the percentage of time humpback whale songs are within mean detection range of a single Stellwagen Bank sensor. (A) Averaged wind speed measured at the NDBC buoy [72] closest to Stellwagen Bank over the “before,” “during,” and “after” 11-day periods; and (B) percentage of the time vocalizing humpback whales localized by our large aperture array are within the mean detection range of the single sensor [34] at Stellwagen Bank in the “before” and “during” periods, using waveguide propagation methods and whale song parameters described in Section 3.5. Since the OAWRS experiment was conducted only up to October 6, 2006, the humpback whale source distribution in the “after” period was not measured and we do not investigate the percentage of time that humpback whales are within the mean detection range of the single sensor at Stellwagen Bank [34] for the “after” period. The triangles represent the mean wind speed and the solid ticks represent the standard deviation of the wind speed over the respective 11-day periods.
Figure 15.
Humpback song occurrence detectable by single sensor matches reported humpback song occurrence at Stellwagen Bank [34].
Average humpback whale song occurrence detectable by a single hydrophone at Stellwagen Bank in time-dependent ambient noise in the “before” and the “during” periods matches the reported humpback whale song occurrence by Risch et al. [34]. Using the measured wind speeds at Stellwagen Bank [72] (Figure 14), the measured spatial distribution of vocalizing humpback whales (Figure 1), and constant song production rates (Figure 9) measured by our large-aperture array, the detectable song occurrence over the “before” and “during” period are found to be within ±18% of the reported means [34], much less than the standard deviations of reported song occurrence[34], using waveguide propagation methods and whale song parameters described in Section 3.5. Before and during OAWRS survey transmissions, this figure shows that reported variations in song occurrence at Stellwagen Bank by Risch et al. [34] are actually due to detection range changes caused by wind-dependent ambient noise, through well established physical processes [20], [73].
Table 3.
Typical anthropogenic noise sources at Stellwagen Bank.
Table 4.
Received mean intensity of typical anthropogenic noise sources at Stellwagen Bank.
Figure 16.
Vocalizing humpback whale bearings measured by our large-aperture receiver array.
Examples of vocalizing humpback whale bearings measured on (A) October 2 and (B) October 3, 2006. Almost all humpback whale vocalizations are found to originate from North-Northeast Georges Bank directions (purple shaded areas) and the Great South Channel directions (green shaded areas), but none originates from Stellwagen Bank directions (red shaded areas). All vocalizing humpback whale bearings are measured from the true North in clockwise direction with respect to the instantaneous spatial locations of towed horizontal receiver array center. The techniques used here for resolving source bearing ambiguity about the horizontal line-array's axis are described in Section 3.3. The shaded bars on the x-axis indicate the operation time periods of the towed array.
Table 5.
Temporal and spectral characteristics of humpback whale non-song calls.
Table 6.
Risch et al. statistical test statements [34].
Table 7.
Possible outcomes of each pairwise comparison between the mean humpback whale song occurrence in the 11-day period of the
33-day period in the
year and that in the
11-day period of the
33-day period in the
year in the Risch et al. statistical test [34].
Figure 17.
Histogram of the measured humpback whale song unit source levels.
The humpback whale song unit source levels measured from more than 4,000 recorded song units during the same 2006 Gulf of Maine experiment discussed here at the same time and at the same location approximately follow a Gaussian distribution and are in the range 155 to 205 dB re 1 µPa and 1 m with a mean of 179.8 dB re 1 µPa and 1 m and a median of 179.4 dB re 1 µPa and 1 m, which are within 0.6 dB of the median of all published humpback whale song unit source levels of 180 dB re 1 µPa and 1 m [93], [101], [102], [152]–[154]. The solid and dashed gray lines represent the mean and the median of the measured humpback song unit source levels, respectively.
Figure 18.
Autocorrelation of Vu et al. [35] humpback whale song occurrence time series in 2006 and 2008.
The e-folding time scale of the Vu et al. [35] annual humpback whale song occurrence time series is (A) 18 days for 2006 and (B) 21 days for 2008. The roughly 20-day coherence time scale shows that the humpback song occurrence gradually changes over periods longer than the 11-day periods analyzed in Risch et al. [34]. It is noteworthy that (1) the humpback song occurrence dropped to zero in the “before” period, and (2) only after a time period consistent with the measured coherence time scale of song occurrence, within which temporal processes are correlated, did song occurrence begin to increase in the “during” period (Figure 10). The Risch et al. [34] analysis then violates temporal causality because the correlated processes that caused the reduction in humpback song occurrence started days before the OAWRS survey transmissions began, yet the analysis and conclusions of Risch et al. [34] offer no other explanation than these survey transmissions for the reduction. Both time series show high correlation at a time lag of roughly seven months due to increases in song occurrence during the spring and fall seasons (Figure 10), separated by roughly seven months.