Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 Evaluation of the Marine r Mammal Research Program DESCRIPTION OF MMRP RESULTS The sound source used by the Acoustic Thermometry of Ocean Climate (ATOC) experiment is acoustically different from the source used in the previous Heard Island Feasibility Test (HIFT), in part due to concerns about potential effects on marine mammals and in part to the shorter distance between the ATOC sources and receivers. In fact, information obtained from HIFT indicated that a less intense sound source (HIFT used a level of 221 dB) could be used for ocean basin-scale studies such as the ATOC experiment (Baggeroer and Munk, 1992~. The ATOC source level was thus reduced to 195 dB. This difference corresponds to a 400-fold decrease in sound intensity. The 75-Hz ATOC signal was transmit- ted from sources located at 850- and 980-m depths off the coasts of Hawaii and California, respectively, for 20 minutes every 4 hours every fourth day, under the standard protocol of a 2 percent duty cycle (see footnote on page 15~. This standard protocol was varied somewhat in both the California and the Hawaii transmissions, depending on the needs of MMRP investigators, although the transmissions were not optimized for studies of marine mammals. The California source transmitted for experimental periods of 2 to 4 days, separated by at least 4 days with no transmissions. The Kauai source used a similar protocol during the first season, followed by the standard protocol of 1 day of transmissions every 4 days. One exception was that the duty cycle of the Kauai source was increased to 8 percent in the summer of 1999, after the humpback whale season, in accor- dance with the environmental impact statement (EIS) for the Kauai source (ARPA and NOAA, 1995~. 21
22 MARINE MAMMALS AND LOW-FREQUENCY SOUND For all ATOC transmissions, the source level in dB increased linearly from 165 to 195 dB and the power increased logarithmically over a period of 5 minutes preceding the 20-minute, full-intensity transmission. This ramp-up period was designed on the assumption that it would allow marine mammals the opportunity to move away and avoid exposure to the sound if it annoyed them. ~ As a result of the decrease in source level and the use of a sound ramp-up, the potential for acoustic impact on marine mammals presumably has been reduced in ATOC compared with HIFT. The sound level at the 850- to 980-m deep sources should diminish to approximately 130 dB at the sea surface, plus or minus a small fraction of this value due to the Lloyd mirror effect.2 Thus, marine mammals that spend most of their time in surface waters potentially are exposed to much lower received sound levels3 than the source level, although deep-diving species could be exposed to higher levels when diving near the source. Geometric spreading also diminishes sound levels in all directions from a source, so that the received level is expected to be about 135 dB at a radial distance of 1 km from the source and 129 dB at 2 km (see Figure 1.1~. The Committee assessed all available information and concluded that the Marine Mammal Research Program (MMRP) was not able to demonstrate a lack of significant effects of ATOC transmissions on marine mammals. The MMRP did not provide unequivocal evidence about the effects of ATOC transmissions because (1) the MMRP data were not fully analyzed as of April 1999 and (2) several of the observational programs were not designed in accordance with the suggestions made in another NRC report (NRC, 1996) that may have helped reduce the ambiguity of the results. It would have been impossible for the MMRP program to conclusively demonstrate a lack of subtle or long-term effects within the short period of the program and the program did produce some useful results that advance our understanding of the effects of sound on marine mam- mals. However, it is important for those designing future studies to recognize that simply not detecting reactions is not by itself sufficient evidence that there is no significant impact. {Costa and Williams (1999) estimated that sustained swimming speeds of many marine mammals are about 2 m/sec. Thus, the time needed for a marine mammal to swim from near the source to a distance at which the received level would be 120 db (5.6 km) would be 47 min. The time needed to reach the 130-dB received level distance (1.8 km) would be 15 min. Thus, the characteristics of the ATOC signal ramp-up period could expose marine mammals to levels of sound of 130 to 165 dB for periods of as much as 15 min. 2The Lloyd mirror effect creates a diminished or augmented pressure of sound from an under- water source either located near the water-air boundary or when received near that boundary. It is caused by the interference between direct and surface-reflected waves and thus creates alternating sound nulls and peaks around this level. 3The "received level" is the sound pressure level measured at the animal, that is, the level to which it is actually exposed. The received level is lower than the source level, depending on the distance between the source and the animal, the sound frequency, and environmental factors.
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM 23 The MMRP was awarded $3 million to conduct its work over 5 years. The Committee did not examine how this funding was allocated to different activities, or whether the funding was adequate to meet the goals, staffing levels, or any other management matters, so it is impossible to determine whether the program was hampered by inadequate funding for the necessary tasks, poor planning and execution of observations, constraints placed on the program by the ATOC experi- mental design or regulatory requirements, the difficulty of working with large whales, or other factors. Although the MMRP observations did show some indications that the ATOC signal did not have a short-term effect on nearby populations of marine mammals and there were no obvious mass mortalities of marine mammals or abandonment of the ATOC source areas by marine mammal species under observation, there was little detailed observational evidence of the effect of the ATOC signal on individual whales. The MMRP results and the committee's evaluation of the significance of the results are given in Table 2.1. The Committee makes a number of recommendations in Chapter 5, based on the MMRP experiences, about the need for peer-reviewed research, multidisci- plinary research teams, proactive research programs not linked to specific acous- tic experiments, and the need to devote sufficient financial and human resources to ensure timely data analysis and publishing of results. Because the MMRP did not provide unambiguous results about the effects (or lack thereof) of the ATOC transmissions, the Committee cannot state unequivocally whether or not ATOC transmissions should continue. (The Committee was not asked to make such an assessment, but the question arose in the natural course of the Committee's discussions.) Instead, in the event of ATOC continuation or other large-scale acoustic tomography experiments, the Committee offers some criteria that should be considered and some mitigation measures that may reduce concerns about such experiments. California ATOC Source The goals of the California portion of the MMRP included (1) sampling the distribution and abundance of marine mammals in the vicinity of the source, (2) testing for differences in those distributions when the source was on and off, and (3) measuring diving responses of a marine mammal (the northern elephant seal, Mirounga angustirostris) as it passed the source while returning to its rook- ery. Because of the distance of the source from land, shore-based observations were precluded, so the distributions of marine mammals were sampled using aerial surveys. Observations were also conducted from a small boat for photo- graphic identification of humpback whales and blue whales and for enumeration of other marine mammals. Aerial surveys provided a more or less synoptic picture of whale distribution throughout the survey area in a single day. These observations were designed to test whether the distribution of whales around the sound source would differ when the source had been transmitting for 1 to 3 days,
24 MARINE MAMMALS AND LOW-FREQUENCY SOUND TABLE 2.1 MMRP Experiments/Observations, Results, and Significance Experiment/Observations Results ATOC MI California Results Naturally migrating elephant seals (Costa et al., 1999) Translocated elephant seals (Costa et al., 1999) Distributions of marine mammals by aerial observations (Calambokidis, 1999) Although the power of the test was low (or = 0.05; power = 0.178), analysis of the five male seals returning when the source was on (2-9 km) versus the 11 control male seals (2-66 km) shows no significant difference in closest distance to the source (t = -1.524, df = 14, p = 0.15). Data analysis is incomplete, but male seals exposed to received levels ranging from 118 to 137 dB in the 60- to 90-Hz band did not seem to change their dive patterns. · A high diversity of species (including six endangered marine mammal species) and numbers of individuals was observed both when the ATOC source was on and off. Humpback whales within a 40 * 40 km inner box centered around the ATOC source were sighted on average 2.4 km farther from the sound source when it was on versus off. No significant difference was apparent in sightings of humpbacks in an 80 * 80 km outer box excluding the inner box. Sperm whales showed a similar response, but data are complicated by dependence of clustered sightings of subgroups. Risso's dolphins were also found farther from the source in the experimental condition within 24 hours of a transmission. Behavior of marine mammals was generally similar comparing exposure to control conditions. No significant differences were observed in the orientation of humpback and sperm whales, but limited sightings of blue whales suggest they oriented more toward the sound source during the transmissions, a trend nearing the p = 0.05 significance level. . . . . Althou which l appear sound, deviatil resulted ATOC The fir highest seals in was 13 The set lack of of the ~ even th certain] · No app of sigh exposu statistic in the ( While ] the are; some near th the sou away (
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM mce 25 ATOC MMRP Conclusion NRC Committee Conclusion = 0.05; ,eals versus no the , exposed dB in the their dive ndangered ndividuals was on per box Bled on roe when ance was ;0 * 80 km whales mplicated bgroups. om the n 24 hours . .. y slmllar No ;, but ey ing the · Although the one migrating seal for which there was a dive record did not appear to react to the initiation of the sound, there does appear to be a deviation in the dive pattern that resulted from the cessation of the ATOC transmission. . The first significant finding was that the highest level of exposure to ATOC for seals intentionally placed near the source was 137 dB in the 60- to 90-Hz band. · The second significant finding was the lack of a dramatic response to operation of the ATOC sound source in any seal, even though the seals could almost certainly hear the source. · No apparent differences in number of sightings comparing control and exposure periods, but there were statistically significant differences in the distribution of some species. · While humpback whales did not vacate the area during ATOC transmissions, some whales shifted from using areas near the sound source (<14 km from the source) to areas slightly farther away (14-28 km from the source). . No evidence was obtained to indicate that the ATOC source disrupts the geographic locations of migrating elephant seals, but the power of the test is limited by small sample size and lack of data on female seals. More data are needed on influence of ATOC on dive patterns of migrating seals. · These are the only data on possible effects of ATOC on diving behavior. Statistical procedures did not correct for multiple tests; test results need to be reanalyzed before final conclusions can be drawn. Given the high diversity and number of animals, and the possibility that the Pioneer Seamount is a critical habitat area for marine mammals, it is probably not a good area for the ATOC source to be located for decades of operation. ATOC has discontinued transmissions from this source. Not enough data are available to determine whether most species showed avoidance or attraction. Significant differences in distance of humpback whales from the source indicate an avoidance response, but the scale of this response is small enough that this is unlikely to impact availability of habitat for the population. Aerial survey data suggest possible vertical avoidance response for some species. For example, there is some evidence for increased sightings of sperm whales during exposure, indicating that exposed whales might spend more time at the surface, a potential vertical avoidance response, but this was not studied in sperm whales in greater detail. continued
26 TABLE 2.1 Continued MARINE MAMMALS AND LOW-FREQUENCY SOUND Experiment/Observations Results ATOC MI Hawaii Results Observations from shore stations during ATOC transmissions (Franker and Clark, l999a,b) Observations from boats during playbacks (Franker and Clark, 1998a) Results of bottom-mounted recorders (Franker and Clark, l999b) · Analysis of whales observed near shore, approximately 14 km from the ATOC source, . . showed a statistically significant increase in the time and distance between successive surfacings as a function of estimated received level of ATOC transmissions. These whales were exposed to levels up to 130 dB in the 60- to 90-Hz band. Analysis of sighting rates of inshore whales showed slightly higher rates during control versus transmission periods, but the difference was not statistically significant. No difference was found in the distance of whales from shore, but distance from the source was not analyzed. Only 11 of 50 playbacks exposed whales to received levels >120 dB in the M-sequence band.a These whales did not show a pronounced avoidance response. Two behavioral variables showed a significant increase with increasing exposure: the distance traveled and time taken between successive surfacings. No difference in the amount of energy in the band of humpback song was detected comparing 20 minutes before transmission, during, and after transmission, from one recorder placed offshore near the ATOC source and four recorders placed inshore near the main concentration of whales. The dis inshore signific · · ret slgnlllc that ins receive Whale differ s and pla Overal] M-seq~ be dete biologi respond There song pi This do stopper to franc average There ~ . . singing from tt
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM 27 ATOC MMRP Conclusion NRC Committee Conclusion . ce, .n the time s as a OC 1 to levels s showed is not , found in ance from received These ice cant ance he band d after shore near d inshore . . The distribution and abundance of inshore whales did not change significantly, but there were significant changes in diving behavior that increased with increasing received level, up to 130 dB. Whale tracks and bearings did not differ significantly between control and playback conditions. · Overall, subtle responses to M-sequence playbacks could only be detected statistically, but the biological significance of these responses is uncertain. There was no change in received song power in response to ATOC. This does not mean that no whales stopped or started singing in response to transmissions but rather that the average song level did not change. There was no widespread change in singing behavior or movement away from the area. . . · There is a data gap in testing for changes in the distribution and abundance of whales near the ATOC source. Given evidence for changes in the diving behavior of whales exposed to low levels far from the source, there is a clear need to study changes in the behavior of whales near the ATOC source in order to evaluate the potential impact of behavioral disruption. It is uncertain whether a change in the time and distance between surfacings is a biologically meaningful measure of the effects of the ATOC source. The limited number of whales exposed to received levels >120 dB limits the power of overall tests. The same behavioral responses were observed in scaled playbacks as were detected in the operation of ATOC. This replication increases confidence in the robustness of this response. A comparison of total sound energy in the band of humpback song power shows no change just before, during, or after exposure. This is a very crude response measure, which would miss potentially important responses (e.g., half of singers stop, half double their source level). In addition, much of the power in the "song band" could stem from sources other than humpback songs and this unmeasured background would not be expected to change in response to an ATOC transmission. This dilutes the power of the test. Details of the movement patterns and songs of individual whales singing near the source need to be studied before it will be possible to evaluate the effect of ATOC on singing whales fully. The appropriateness and statistical power of this method were not tested. continued
28 TABLE 2.1 Continued MARINE MAMMALS AND LOW-FREQUENCY SOUND Experiment/Observations Results ATOC MI Results of aerial surveys (Mobley et al., 1999) . Of five major areas in the Hawaiian Islands, Kauai/Niihau had the second-highest number of humpbacks sighted, after the four-island region west of Maui, but Penguin Banks west of Molokai had the highest density after corrections for observing effort. An average of 5.3 humpbacks and 0.6 sperm whales were sighted per survey during 1994 within 40 km of the source location before ATOC transmission. An average of 7.0 humpbacks and 0.75 sperm whales were sighted per survey within 40 km of the source during 1998 after the source was activated. This higher rate in 1998 may reflect better sighting conditions. Humpbacks prefer water <200 m deep, but 30 percent were sighted in 200-2,000 m depths. Mean distances from shore and from the source were higher for humpback whales when the source was on, but difference was not statistically significant. . For hu] in dista from sl results, not ope operati Sperm · r as inert were fc island NOTE: The first two columns of the table were provided to Christopher Clark and Dan Costa for review before the report was published. aAn M-sequence signal is a phase-modulated tonal signal. compared with after at least 4 days without ATOC transmissions. Acoustic surveys also were planned by MMRP investigators, and acoustic measurements were attempted using ATOC receivers. The planned acoustic surveys did not yield any data because of the failure of deployed equipment. The following null hypotheses were tested using aerial, visual, and acoustic surveys (ARPA, 1995, p. C-12: Ho There is no detectable difference in sighting rate, distribution, orientation, general activity, or group size of different species (or groups of species) between
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM 29 ATOC MMRP Conclusion NRC Committee Conclusion .s, per of gion west cat had the ing effort. m whales n 40 km . . msslon. rm whales e source . This tiny pths. lurch were roe was on, ant. . . For humpbacks no significant changes in distance from source or distance from shore were noted from the 1994 results, when the ATOC source was not operating, to 1998, when it was operating. Sperm whales, previously described as infrequent in Hawaiian waters, were found offshore of all five island regions. · Humpback whales in Hawaii showed a pattern of increased distance from the source when it was operating (average = 19 km) compared to off (average = 17 km). This difference, based on 28 exposure sightings, was not statistically significant. A similar pattern in California (16.9 km for exposure versus 14.5 km for control), based on 105 exposure sightings in the 40 * 40 km inner box, was highly statistically significant. It is likely that humpbacks in both places have similar responses, and that difference in significance stems from limited sample sizes in Hawaii. The possible shift in distribution at great ranges from the source suggests the need for a behavioral study on responses to ATOC signals targeting animals near the source. · Kauai is an important habitat for the expanding population of humpback whales wintering in Hawaii. Coarse analysis does not suggest a mass evacuation of the area within 40 km of the ATOC source. However, enough humpbacks were sighted in water >200 m depth to justify a targeted study of humpbacks near the Kauai source location. The presence of offshore sperm whales also suggests that a targeted study on the impacts of long-term ATOC signals on this endangered species would be appropriate. n Costa for surveys conducted when the source is on and when it is off, and as a function of distance from the source. Ho There is no detectable difference in cetacean acoustic behavior (i.e., call types, rates, structure, or sequence patterns) between measurements from record- ings made when the source is on and when it is off, and as a function of distance from the source. Appropriate tests of these hypotheses assume that the precision of the mea- surements and the statistical power of the tests are great enough to demonstrate
30 MARINE MAMMALS AND LOW-FREQUENCY SOUND actual differences; that is, the probability of a false negative result is small. Studies of marine mammals in the wild are so difficult (due to problems of observing animals that spend much of their time underwater) and the results so imprecise (because of natural variability and low sample sizes) that it is easy to imagine that such studies might not detect differences that could reflect biologi- cally significant impacts. Only for tagged elephant seals exposed to California transmissions were analyses of precision or power presented, and thresholds for biological significance were not suggested. These factors make it difficult to evaluate the validity of the MMRP's negative results, especially for species other than elephant seals. Aerial surveys were conducted from November 1995 to October 1998. Dur- ing control surveys there were 1,524 marine mammal sightings4 involving 29,826 animals. During experimental surveys (source on), there were 1,617 marine mammal sightings, involving 27,874 animals. Not only were there more sightings in both the experimental and the control conditions than expected, there was a larger diversity of marine mammals sighted than expected. At least eight species of small- and medium-sized toothed whales were observed, four species of seals, five baleen whale species, and two species of large toothed whales. The most frequently sighted large whales were humpback whales (482 sightings) and sperm whales (349 sightings), numbers large enough to permit statistical tests for differ- ences between control and experimental surveys. Statistical analyses of these data were not completed by the time of the Committee's April 1999 meeting. Aerial surveys revealed that humpback whales were significantly further from the source when it was on than when it was off. A similar pattern was found for sperm whales, but the statistical significance was complicated by seasonal differ- ences in distribution (Calambokidis, l999~. Calambokidis also found an increas- ing trend in the number of humpback whales off the U.S. west coast from 1988 to 1996 (6.7 percent) and from 1996 to 1998 (9.3 percent), using photoidentified whales and mark-recapture calculations, indicating that the ATOC source did not negatively affect the population level of this species. Elephant seals are important research subjects in relation to the effects of the ATOC source because they have sensitive low-frequency hearing (Kastak and Schusterman, 1998), swim in the pelagic zone, and routinely dive near the depth of the deep sound channel.5 This species has breeding areas near the California ATOC source site, and these animals are excellent subjects for tag attachment (tags are subsequently removed or shed during molting; D. Costa, University of 4A sighting is one group of marine mammals, regardless of number. 5The deep sound channel or SOFAR (Sound Fixing And Ranging) channel occurs at a depth in the ocean at which ``sound rays propagating close to horizontally are trapped by refraction, reducing spreading loss and avoiding surface and bottom losses,, (Richardson et al., 1995). The SOFAR channel is found at approximately 1,000 m in the open ocean and approaches the sea surface in polar regions.
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM 31 California, Santa Cruz, personal communication, 1999~. Satellite tags were used to track the locations of naturally migrating individuals. A total of 26 adult males were followed during their natural migration; five tracks were observed when the source was on, and 11 tracks were monitored when the source was off. Only a few tracks of these naturally migrating seals passed near the ATOC source, and there was no obvious avoidance, based on nearest approach to the ATOC source. Based on aerial surveys, elephant seals were found at the same distance from the source whether the source was on (50 sightings) or off (35 sightings). Translocation experiments were used to obtain a larger sample size of seals exposed near the source. In these experiments, archival tags designed to record received levels of sound and dive patterns were attached to juvenile elephant seals removed from a rookery. Thirteen elephant seals were translocated near the ATOC source when the source was operating, and five seals were translocated to the same site when the source was off. The maximum measured received levels of the ATOC sound for each of the 13 seals in this experiment ranged from 118 to 137 dB. MMRP investigators conducted an extensive statistical analysis of dive patterns of the translocated seals (including a variety of measures, such as time of dive and depth) comparing (1) the dive before the source was turned on, (2) the first dive that started after the source was turned on, (3) the second dive after the source was turned off, and (4) an average of dives measured over 18 hours following the second dive after the source was turned off. The comparisons conducted thus far suggest that there was not a statistically significant change in the dive behavior of translocated seals in response to ATOC transmissions at received levels of 118 to 137 dB. However, the preliminary statistical analysis comprised hundreds of individual tests. These must be merged into one overall analysis, with proper correction for significant results that can occur by chance when a large number of statistical tests are run. Hawaii ATOC Source The Hawaiian observations focused on humpback whales and were planned to include aerial visual surveys, passive acoustic monitoring, and shore-based surveys of reactions to ATOC transmissions (off Kauai) and playbacks of hump- back whale vocalizations (off Hawaii). The only peer-reviewed paper analyzing the responses of marine mammals to the ATOC signal to date is that of Frankel and Clark (1998a). This paper did not report on research involving the ATOC source in its site off the north coast of Kauai but described a series of playback experiments using a smaller vessel-deployed source off the coast of Hawaii in a much better site for observing whales. This source was operated at 172 dB, with a frequency bandwidth of 60 to 90 Hz (the same as ATOC). The source was a vessel moored each day in an area of high whale density off the leeward coast of Hawaii, in a position allowing excellent monitoring of humpback whales from a shore station. Unlike most MMRP observations, timing of operation of the
32 MARINE MAMMALS AND LOW-FREQUENCY SOUND source was determined by whale monitoring rather than the predetermined ATOC transmission schedule. In this study, if the shore observers could follow a whale or group of whales for 25 minutes, they would radio the playback vessel and instruct the boat to start an experiment. On a randomized schedule, during 50 of the 85 trials the vessel transmitted the ATOC signal; the other 35 trials were no- sound controls. The shore observers were unaware of which condition was being employed during any given trial. The shore team attempted to continue to ob- serve the whales during the 25-minute experimental period and for a 25-minute post-trial phase. The estimated received level (based on empirical measurements at different ranges and bearings from the playback vessel) at the whales' location during playback varied from ambient (near 90 dB) to nearly 130 dB. Statistical analyses of whale tracks and swimming directions revealed no difference in these factors between experimental and control conditions. How- ever, this is difficult to interpret because the analysis combined data from whales located so far from the playback signal that the signal was buried in ambient noise, with only a few whales exposed to received levels high enough to expect the possibility of a response. Simple nonparametric comparisons of speed, dura- tion, and distance between surfacings of the humpback whales also showed no difference between control and playback conditions. It appears that the swim- ming direction of whales with respect to the playback source was not analyzed, even though this is the critical measure for determining an avoidance response. There was a slight significant increase in the time and distance between succes- sive surfacings at increasing received levels of playback; this strengthens the Committee's concerns about conclusions of no effect in these pooled data. Of the 85 trials, an observed whale passed within the 120-dB isopleth at a range of about 400 m in 11 playback trials and five control trials. A potential avoidance reaction was observed in one of these 11 playback trials; a similar "avoidance reaction" also was observed during control observations with no sound. Two potential approaches were observed during playback. The limited sample size of animals exposed to received levels greater than 120 dB limits the power of conclusions regarding lack of effects. Although the sample size of whales exposed to playbacks louder than a 120-dB received level was small, the results imply that most whales would be unlikely to show an avoidance response when exposed to sound levels of 90 to 130 dB. The responses observed were no stronger than those elicited when vessels approached whales in the study area. The Committee received several unpublished manuscripts from MMRP investigators about the responses (or lack thereof) of humpback whales to the ATOC source 14 km north of Kauai's coast. The Committee was told that aerial survey results suggested there may be resident populations of sperm whales and short-finned pilot whales (Globicephala macrorhynchus) in the offshore waters, but the Committee was not presented with any data on the distribution or potential responses of these two species when exposed to the ATOC sound. Rather,
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM 33 MMRP studies concentrated on humpback whales, as indicated in the EIS for the Kauai source (ARPA and NOAA, 1995~. The observation effort focused on inshore waters roughly 10 km from the source. Transmission loss measurements suggest that the received level of the ATOC signal in the inshore waters preferred by humpback whales (less than 200 m deep) is less than 120 dB in the 60- to 90-Hz frequency range.6 Although many analyses found no difference in responses between control and transmission conditions, some statistically significant differ- ences apparently were observed, even though most whales appeared to be exposed to levels less than 120 dB. As in the playback experiments off the island of Hawaii, the distance between successive surfacings increased with increasing received level of ATOC transmissions off Kauai during 1998 (p = 0.0017~. This could have resulted if the whales either were swimming faster or stayed under water longer between surfacings. The Quick-Look Report of the Hawaii ATOC-MMRP (Franker and Clark, 1998b) (an unpublished and unreviewed account)7 concluded that "there were no acute or short-term effects of the ATOC transmissions on marine mammals."8 The Committee questions whether a conclusion this broad can be reached at this time using the data provided. The report does, in fact, present evidence for some short-term behavioral changes in response to the ATOC sound source by hump- back whales. Even more important, the Committee questions the ability of the MMRP to show the absence of any response. The failure to observe an effect could result from a number of factors, including the specific conditions of the experiment and lack of sufficient statistical power (resulting from an insufficient number of observations or the statistical test chosen). This concern is particularly heightened for the Hawaii MMRP study in which most observations were made far from the source and no results were presented on responses of the marine mammal species most commonly sighted offshore near the source (sperm whales and pilot whales). Contrary to plans listed in the EIS for the Kauai source (ARPA and NOAA, 6The original predictions in the initial EIS were based on a spherical transmission loss from the 20*1Og(range) relationship. That is how the 40-km radius "zone of influence" was determined. This contour is approximately 7.5 km south of the ATOC source. Actual measurements of the ATOC transmissions found that the 120-dB received level occurred approximately 4.8 km south of the source. At the 200-m contour, the received level was approximately 111 dB in the 60- to 90-Hz ATOC band (C. Clark, Cornell University, personal communication, 1999). This was more consis- tent with the predictions of a cylindrical equation model, which terminated the 120-dB isopleth at the 200-m depth contour. 7The Quick-Look reports were undoubtedly designed as a means to disseminate research results rapidly and widely to try to achieve open access to research results and keep the public informed, both worthy goals. However, the Committee determined that the Quick-Look format was generally counter-productive because it widely disseminated non-peer-reviewed results and did not encourage timely peer review and publication of research results. 8http://atoc.ucsd.edu/HIquicklookrpt.html, accessed 10/13/99.
34 MARINE MAMMALS AND LOW-FREQUENCY SOUND 1995), there were no analyses of the vocal behavior of individual humpback whales exposed to the ATOC signal. Instead, there was only a general assess- ment of the total energy at humpback whale vocalization frequencies in the area, which could be misleading given that all vocal activity in the area was summed for pro- and postexposure. Considering the analyses conducted to date, the possibility that the ATOC signal might affect the song production of humpback whales cannot be eliminated. We do not know the functions) of humpback songs, but they may be a reproductive "advertisement" display, as for the songs of some birds (Tyack, 1981~. The limited data presented by the MMRP made it impossible to draw any but the most tentative conclusions about the effects of ATOC sounds. Based on the material presented, baleen whales, sperm whales, and elephant seals in California, and humpback whales in Hawaii did not show profound avoidance responses to the ATOC signal. However, complete analyses and peer review are required before any more definitive conclusions can be reached. COMPARISON OF THE RECOMMENDATIONS OF NRC (1996) AND MMRP RESPONSES Although it was difficult to evaluate the MMRP in midcourse, the 1996 NRC interim report contained the following conclusions: The California ATOC source transmissions did not appear to cause a major change in the distribution of marine mammals. 2. The constrained sound characteristics and conditions used during the MMRP-controlled ATOC transmissions impeded the project's ability to answer fundamental questions concerning the impact of ATOC-like noises on marine mammals. 3. Several changes in the plans for the Hawaiian MMRP studies (eventually concluded in 1998) could provide more definitive information about the potential of ATOC sound to affect marine mammals and other organisms. Specifically, shore-based observations should be conducted for the entire 6 months of ATOC transmissions, and the effectiveness of observational methods should be vali- dated using playbacks of relevant natural sounds conducted within visual range of the shore station. The Committee reviewed plans for MMRP research in its 1996 interim report and believed that the value of the work could have been enhanced considerably with some modifications in the proposed study. Not only would these changes have strengthened the ability of the MMRP to test the effects of the ATOC sound on marine organisms, the additional data obtained would provide useful insight into broader questions about the effects of low-frequency sounds on marine mam- mal behavior. Much of the MMRP research focused on statistical tests of whether
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM 35 behavioral indicators varied significantly between transmission and control con- ditions. The Committee favored tests in which the biological significance of any such changes could be evaluated, which would require a broader investigation into the effects of noise on the normal behavioral ecology of each species. The goal of the MMRP was to determine whether the ATOC transmissions might adversely affect reproduction or survival of marine mammal populations. The methods available to measure population trends are crude, and the cause of conserving populations will not be met by waiting until a threat actually causes a measurable decline in populations. Therefore, it is useful to use more short-term measures as indicators of potential adverse impact. For example, if marine animals avoid or leave critical habitats because of a human disturbance, the animals may enter suboptimal habitats, with eventual negative effects on feeding and/or reproduction. Proxies selected to measure adverse impact should be easily measured animal behaviors that, if disrupted, would have significant negative impacts on marine mammal reproduction and longevity. The apparent avoidance reactions observed in the California MMRP studies are good examples of rel- evant measures; the impact can be related to the percentage of habitat lost or can be estimated by comparing the quality of the habitat the whales left compared to that to which they moved. Other elements of the MMRP studied behavioral changes that are less rel- evant. For example, the Hawaii MMRP analyzed the distance traveled and time spent between surfacings for humpback whales and found a statistically signifi- cant trend for these measures to increase with increasing exposure to ATOC transmissions. Even though these results are statistically significant, it is difficult to interpret their possible biological significance. We suggest that future studies of this sort carefully select behavioral and physiological measures that can more easily be related to potential adverse impact. Basic research in the behavioral ecology of many species is required to direct these choices. For example, the more we know about the foraging ecology of a species, the better we can interpret the biological significance of disruption of feeding behavior, or movement to different feeding areas. Since humpback song is known to be involved in the breeding behavior of humpback whales and the ATOC sound could have disrupted singing, the Hawaii MMRP observations would have benefited from selecting behavioral studies that could more easily be related to potential impact on song and thus breeding behavior. NRC (1996) offered two suggestions and one point for further consideration. The first suggestion concerned maintenance of the onshore observation station on Kauai for the full duration of source testing to allow time for additional playback experiments. The second suggestion concerned the need for prompt analysis and dissemination of MMRP results. An additional point concerned other marine species that are potentially ensonified by ATOC sounds. For each of these issues the following sections will present recommendations of the NRC from its 1996 interim report, followed by the MMRP responses.
36 MARINE MAMMALS AND LOW-FREQUENCY SOUND Maintenance of the Shore Station and Playback Experiments At its October 1996 meeting the Committee was presented with plans for maintenance of the shore observation station at the Hawaii field site during the 1997 ATOC transmissions to observe behavioral changes in humpback whales during exposure to ATOC sounds. Observation of marine mammals from similar shore stations provided useful baseline data during a previous playback study of humpback whales off the coast of the island of Hawaii (ARPA and NOAA, 1995, Appendix G) during the 1993 to 1994 season and earlier off the coast of California (Malme et al., 1983, 1984~. According to plans for the ATOC sound transmis- sions in 1997, the shore station observers were to be in place for only 4 to 6 weeks. This period was designed to provide the minimum amount of information needed for comparison with the 1993 to 1994 baseline data, with no margin for unforeseen circumstances. The Committee disagreed with this minimal effort and recommended that the shore station be maintained and used throughout the humpback whale season off Kauai (e.g., during the entire 6 months of MMRP-controlled ATOC transmis- sions). This suggestion was based on the conclusion that additional very useful data (see below) could be obtained from continuation of the shore-based observa- tions, especially with the addition of natural sound playbacks near the shore station. Extending the field season also would have increased the sample size of observations, making it more likely that significant effects of the ATOC trans- missions would have been detected, if such effects actually occurred. Shore- based observations are important because they provide a means of observing marine mammals without introducing the confounding effects of nearby vessels. Although the shore site was probably outside the area within which an effect would be expected, such observations should have been able to determine whether inshore humpback whales, rather than offshore near the ATOC source, would be affected. The MMRP did conduct shore observations for 6 weeks from Febru- ary 9 to March 20, 1998. Observations were not extended beyond this time. According to the MMRP, as of 1996, ATOC transmissions in California and an ATOC-like test signal played off Hawaii had little observed effect on marine mammal behavior, at least in terms of surface tracks and the number, frequency, or depth of dives. Interpretation of these findings is complicated, however, because there had been no observed response to ATOC signals. Thus, it was impossible to establish the validity of the method that had been used to study ATOC's effects. Simply stated, the Committee could not choose between the conclusion that the ATOC signal had little or no effect and the alternative view that the observational methods used were not sensitive enough, or not designed appropriately, to detect the effects of these sounds. In the interest of facilitating future investigations into the effects of ATOC or ATOC-like sound sources on marine mammals, it is essential that an effort be made to define protocols that are useful scientifically and relevant to the decisions that must be made. Such proto-
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM 37 cots could specify how observers must confirm before their observation program begins that their observation techniques can measure the variables of interest- behavior associated with critical activity as well as the minimum statistical power that can be tolerated, so that significance of disruption can be ascertained. Analyses should be framed not only to test for any detectable change, but also to estimate the percentage of time a behavior is disrupted, the amount of energy wasted, and/or the probability that the disruption will prevent animals from achieving the goal of the activity. The Committee's dilemma could have been overcome if the MMRP had been able to demonstrate that its observation methods were valid. The most direct means to test the methods would have been to increase the intensity of the sound source until a response was observed, to create a direct estimation of the stimulus-to-response relationship. However, the ATOC source cannot be oper- ated at levels higher than 195 dB for technical reasons, and it is unknown whether higher levels would produce a measurable response without being unduly harm- ful to marine mammals. Therefore, the Committee suggested an alternative approach that of broadcasting noises other than ATOC signals that would affect marine mammal behavior in a way that is detectable by the same (or similar) methods used in the ATOC study. In earlier studies of marine mammal responses to playback of auditory stimuli, Clark and others (Clark and Clark, 1980; Malme et al., 1983; Tyack, 1983; Mobley et al., 1988; Frankel et al., 1995) have shown that animals respond strongly to certain natural vocalizations, such as the calls given by other members of the same species or the vocalizations of a predator, such as the killer whale (Orcinus orca). Although use of a non-ATOC stimulus would not allow validation of the specific response to the ATOC stimulus, it would at least have validated that MMRP investigators could observe (with sufficient precision and accuracy) such things as startle, flight, and vocal responses, which could be produced by many different stimuli, whether ATOC sounds or killer whale calls. The Committee recommended that the MMRP incorporate natural sounds into its research during its 1997 to 1998 studies, taking into account the results of the playback studies cited above. The MMRP did not conduct extensive vessel-based playbacks of natural sounds, although a vessel was used during this time to observe whales. Frankel and Clark (1998a) reported on one playback of an Alaskan humpback whale feeding call, although the results were ambiguous. If playback of these natural sounds had elicited a strong observable behavioral response from the subjects, that response would have provided an important validation for the ob- servational method used by the MMRP to test ATOC's effects on the behavior of marine mammals. Measurement techniques must be calibrated with a stimulus that produces a measurable response. Such a calibration allows a scientist to distinguish between a true lack of response and a response that was unmeasurable by the chosen technique. For example, in the case of singing humpback whales, had the researchers tested enough singing animals with a biological sound (e.g.,
38 MARINE MAMMALS AND LOW-FREQUENCY SOUND killer whale vocalizations), they could have determined which behavioral param- eters showed statistically significant changes (e.g., song elements, diving behav- ior, evacuation of the area) and provided a baseline of comparison for other stimuli. Because the MMRP did not add this component, the difficulty of inter- preting the MMRP' s results remains. Changes that might have indicated significant effects of ATOC transmis- sions include · significant changes in singing patterns (would need to correlate with calving rates); · significant flight of animals from the source area (significant either in distance, speed, duration, or movement into harm's way); · significant reduction in calves produced; and/or · significant abandonment of area by identified individuals in later years. Need for Prompt Data Analysis One of the problems faced in preparation of the 1996 NRC report was the lack of analyzed data from a number of MMRP field studies, particularly those conducted in Hawaii. Thus, it was difficult to assess the quality and significance of this work and to make suggestions for future ATOC-related marine mammal studies. In its 1996 report the NRC noted that such a situation, if it persisted, would compromise the Committee's future work, and it would not be able to conclude whether there are deleterious effects of the ATOC sound source on marine mammals (or other organisms). The Committee expressed its hope that it would receive a full analysis of MMRP observations and conclusions at least 2 months prior to its final meeting. The NRC strongly recommended that data analysis and presentation be the highest priority for investigators in both Hawaii and California and that sufficient funds be set aside to enable a complete and expeditious evaluation of the data. The Committee's 1999 meeting was sched- uled for approximately 6 months after completion of MMRP observations to allow time for analysis to be completed. The Committee received the June 1998 Quick-Look Report on the Hawaii ATOC-MMRP Results several weeks before its April 1999 meeting. This Quick-Look Report included some preliminary analyses and indicated that more extensive analyses would be forthcoming. The Committee did not receive any completed analyses or conclusions before the meeting. The MMRP investigators at the meeting asserted (as did the Quick- Look Report) that considerable further analysis was needed to interpret the data properly. In discussions with MMRP investigators during the April 1999 meet- ing, it was clear that only limited funds and personnel were available during the final year of the MMRP and that this shortage continues to jeopardize the quality and timeliness of the scientific products of the MMRP.
EVALUATION OF THE MARINE MAMMAL RESEARCH PROGRAM Ensonified Species Other than Marine Mammals 39 In addition to the two specific suggestions for MMRP research, the Com- mittee noted that the EIS for both the Hawaii and the California sources included analyses of the effects of the ATOC sound on other biota, including marine turtles, fish, and other organisms (ARPA, 1995; ARPA and NOAA, 1995~. The only study published for other vertebrates from ATOC-funded research was for an experimental study of ATOC-like sounds on fish (Klimley and Beavers, 1998~. The NRC (1994, pp.53-53) specifically pointed out that a major concern for all low-frequency ensonification is not only effects on marine mammals but also the potential effects of such sounds on other components of marine mammal food chains, such as fish or zooplankton, and on other endangered species (e.g., turtles). The Committee strongly supports this assertion and continues to be concerned that low-frequency sound may have implications for a far wider range of the marine biota than is being studied at the moment. This is of particular importance for sound sources such as ATOC that will be operated in one place for years at a time. In addition, a number of studies suggest that ATOC-like sounds may be very attractive to many species of sharks (Myrberg, 1972, 1978; Myrberg et al., 1976~. Sharks attracted to ATOC sounds could be affected adversely by these sounds, and ATOC equipment could be jeopardized by sharks. The lack of study on marine organisms other than mammals makes it impossible to infer the poten- tial impact of long-term deployment of ATOC-like sources in areas used by . . . sensitive species. SIGNIFICANCE OF THE MMRP TO RESEARCH USES OF SOUND Data presented by the MMRP were inconclusive regarding the effects of the ATOC sound on marine mammals. The Committee considers that existing data from the MMRP and other sources such as recent work motivated by the 1994 NRC report and funded by ONR (e.g., Au et al., 1997) suggest, however, that there is no cause for alarm about the short-term effects of ATOC sources on dolphins and most seals because they do not dive to depths that would allow them to encounter the source at levels they could hear well. However, there is cause for concern because we cannot totally rule out short- and long-term effects of ATOC, particularly on baleen whales and sperm whales. Optimally designed studies are needed on the long-term effects of high-intensity sound sources (e.g., interference with communication and reproductive activities, exclusion from critical habitat). ATOC investigators plan to apply for funding and permission to continue ATOC transmissions in Hawaii for another 5 years (the California source has been terminated). ATOC investigators plan to conduct aerial surveys near the Kauai ATOC source to monitor the distribution and abundance of marine mam- mals to advance the understanding of possible long-term acoustic impacts (P.
40 MARINE MAMMALS AND LOW-FREQUENCY SOUND Worcester, Scripps Institution of Oceanography, personal communication, 1999~. It is outside the Committee's charge to comment on whether ATOC should be allowed to proceed. However, if it does proceed, monitoring of marine mammal behavior and responses to the ATOC transmissions should continue as an integral part of the experimental design in order to improve the ability to evaluate the impact of ATOC during the next 5 years of Hawaii ATOC transmissions. As part of this continued evaluation, there should be annual reports of all yearly data to an oversight body not associated with ATOC (e.g., the Marine Mammal Com- mission or National Oceanic and Atmospheric Administration), with the authority to terminate transmissions if there is evidence of significant deleterious effects from long-term exposure. ATOC scientists should be required to notify the oversight group immediately if they detect significant adverse effects on marine mammals. Continuation of ATOC transmissions should be conditional on timely publication of marine mammal results. Chapter 5 includes a discussion of appro- priate monitoring that should be considered if ATOC is approved to continue.
