Diel migration and swimbladder resonance of small fish: some implications for analyses of multifrequency echo data

Diel migration and swimbladder resonance of small fish: some implications for analyses of multifrequency echo data Olav Rune Godø, Ruben Patel, and Geir Pedersen 1143 Godø, O. R., Patel, R., and Pedersen, G. 2009. Diel migration and swimbladder resonance of small fish: some implications for analyses of multifrequency echo data. ICES Journal of Marine Science, 66: 1143 1148. Many fish with swimbladders exhibit diel vertical migrations (DVM). Ascents and descents of hundreds of metres occur, and altered swimbladder volume and buoyancy can result from incomplete secretion and resorption of gas. When acoustic observations are made near the resonance frequency of the swimbladder, the estimated fish biomass can be positively biased. When multiple-frequency echosounders are used, the frequency response of the backscatter might vary temporally and spatially and compromise the effectiveness of conventional target-identification methods. In this paper, variations in backscatter from mesopelagic fish are studied using data collected west of the British Isles with a five-frequency echosounder (Simrad EK60). Two acoustic layers, one dominated by pearlsides (Maurolicus muelleri) and the other by myctophids (Myctophidae), were monitored during their DVM. The frequency responses of the layers changed systematically, mainly characterized by increases in the nautical-area-backscattering coefficient (s A ) values at 18 khz relative to those at 38 khz. This could have been caused by changes in the resonance frequencies of fish swimbladders, as they expanded and contracted during ascent and descent. Two s A maxima in the myctophids layer suggest the presence of two types of target with different scattering characteristics. Models of sound scatter from myctophid swimbladders suggest that these peaks have resulted from resonance scattering. The s A at 18 khz attributed to M. muelleri also peaked, but at the maximum depth of their distribution. Spatial and temporal changes in the frequency responses of fish should be taken into account when pelagic fish communities are surveyed with multiple-frequency echosounders. Keywords: diel migration, mesopelagic fish, multiple-frequency analysis, species identification, swimbladder resonance. Received 8 August 2008; accepted 15 February 2009; advance access publication 16 April 2009. O. R. Godø, R. Patel and G. Pedersen: Institute of Marine Research, PO Box 1870, Nordnes, 5817 Bergen, Norway. Correspondence to O. R. Godø: tel: þ47 55 23 86 75; fax: þ47 55 23 85 31; e-mail: olavrune@imr.no. Introduction A wide acoustic bandwidth was once thought to play an important role in remote species identification (MacLennan and Holliday, 1996; Simmonds et al., 1996). Since then, frequency-dependent backscattering has been studied for various species and species groups, and these frequency responses are now commonly used in studies of marine ecosystems for target identification (Horne, 2000; Korneliussen and Ona, 2000; Korneliussen et al., 2008). For this method to be effective, the frequency responses of targets and nearby scatterers must be stable temporally and spatially, or have predictable dynamics. Aspects of diel vertical migration (DVM; Neilson and Perry, 1990) can modulate the frequency responses of fish. Physoclists and physostomes might not be neutrally buoyant during parts of their DVM; and variations in their swimbladder volumes will cause their backscattering cross section (s bs ; Arnold and Greer, 1992; Huse and Ona, 1996) and resonance frequency ( f r ) to change. Changes in buoyancy can alter the distributions of fish pitch angle relative to the horizontal plane (u; MacLennan and Holliday, 1996; Godø et al., 2006). This can also cause frequencydependent variations in their nautical-area-scattering coefficient (s A ; MacLennan et al., 2002). Changes in swimbladder volume will influence all frequencies similarly and hence might have only minor effects on analyses of multiple-frequency data. In contrast, changes in s bs and s A from variations in u are larger at higher frequencies; and changes in f r resulting from changes in swimbladder volume generally affect s bs and s A at lower frequencies. Therefore, variations in u and f r can directly alter the s A measured at different frequencies. In addition, if the f r is very close to the measurement bandwidth, the s A will vary temporally throughout the DVM period and spatially throughout the species depth range. This paper examines some potential problems in acoustic surveys associated with resonance scattering. For the traditional survey frequencies from 18 to 400 khz, such problems might only arise for small fish with swimbladders, e.g. young-of-the-year and mesopelagic fish, and at the lowest frequencies. Myctophids have swimbladders and they represent a dominant group of mesopelagic species in the North Atlantic. As they age, the gas in their swimbladders is gradually replaced by lipids (Neighbors and Nafpaktitis, 1982). Gas volume in myctophids is therefore not necessarily dependent on fish size. Their swimbladder gas volume is often less than that of other fish of a similar size. Myctophids exhibit extensive DVM (Pearcy et al., 1977), probably # 2009 International Council for the Exploration of the Sea. Published by Oxford Journals. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

1144 O. R. Godø et al. without the ability to compensate continuously for pressure changes. When marine populations are studied with multifrequency echosounders, it must be considered that small-swimbladdered fish like myctophids could become resonant during part of their vertical excursions. Resonance in swimbladdered fish is a wellknown phenomenon (Holliday, 1972; Nero et al., 2004), which has also been taken into account in multiple-frequency studies of deep-water fish (Kloser et al., 2002). Chapman et al. (1974) studied the frequency response of fish at medium and high frequencies (2 20 khz) using underwater explosions. Clear day vs. night differences were observed in f r, which were attributed to changes in swimbladder volumes during the DVM period. To date, most such acoustic studies have described variations in s A vs. coarse bins of depth and time-of-day. An exception is Love et al. (2004), who combined broad-bandwidth acoustics and intensive trawling, and demonstrated time-depth resonance phenomena in mesopelagic fish surviving in or close to an anoxic environment during the day. In the current study, changes in the frequency responses of three layers of mesopelagic fish are examined over several DVM periods, focusing on times of ascent and descent. Such changes could compromise the efficiencies of multiple-frequency analyses. Methods Data were collected with a multifrequency echosounder (Simrad EK60) on MV Eros from 24 to 27 March 2007 in an area to the west of the British Isles (Figure 1). Despite forming part of the annual blue-whiting survey, the selected area had few blue whiting. The ship is a combined purse-seiner and trawler, and the echosounder transducers (centre frequencies f ¼ 18, 38, 70, 120, and 200 khz) were mounted on a keel that can be lowered. Because of the depth range of mesopelagic fish, only data from the lower three frequencies were analysed. The signal-to-noise ratio at 70 khz limited the analysis to depths of,600 m. The echosounder was calibrated before the survey in accordance with standard procedures (Foote et al., 1987). The multifrequency measurements of volume-backscattering strengths (S v ) were processed with the large-scale survey system (Korneliussen et al., 2006). The S v were averaged in cells (1852 m horizontal 10 m vertical) and apportioned to the following categories (Figure 2): (i) migrating deep layer: a layer of mesopelagic fish migrating from 400 to 500 m by day to the surface at night; (ii) non-migrating deep layer: a layer of non-migrating mesopelagic fish below the migrating deep layer and partly mixing with them by day; and (iii) migrating shallow layer: a layer of fish migrating between 200 m by day and the surface at night. The layers were identified during both day and night with a standard pelagic-sampling trawl (Åkratrål) with a 20 35 m vertical opening and a codend of 22 mm mesh (Wenneck et al., 2008). The frequency dependence of the s A was examined for changes in f r. The ratio at each f between s A and the s A at 38 khz was denoted the relative frequency response (RFR; Korneliussen and Ona, 2000): RFR f ¼ s Af =s A38 : The s A values were thresholded before the RFR f values were calculated using Equation (1) and averaged in temporal bins: daytime (09:00 17:00 UTC); night-time (21:00 05:00 UTC); and dawn and dusk (the hours between). Low-frequency, isotropic scattering functions describing resonant scattering from gas-filled prolate spheroids and bubbles (Andreeva, 1964; Weston, 1967; Ye, 1997) were used to estimate swimbladder sizes with resonances at 18 khz. Table 1 gives these sizes for depths of 200 and 400 m, representing the depths of RFR 18 maxima. Comparisons with the observed RFR 18 might indicate the presence of resonance. The least-squares method was used to find the diameters of the major axes of a bubble and prolate spheroid. In this context, low-frequency means that the wavelength was much larger than the major (a) and minor (b) axes of the swimbladder. ð1þ Figure 1. Cruise track of the MS Eros. The data in this paper were collected on the two northern tracks. Results The biomass in the study area exhibited significant DVM (Figure 2). It was difficult to separate the migrating and nonmigrating deep layers of fish in the deep water by day, and the merging of layers probably included some mixing. Therefore, a line was drawn on the echogram from the depth where the layers met in the morning to the depth where they separated in the evening. Furthermore, partial separation of the layers permitted further delineation by day. The migrating shallow layer was easily separated from the deep scattering layers by day; however, it could not be observed at all at night, when it had ascended into the surface blind zone. Catch information from the pelagic trawls indicated the presence of a wide variety of species, but only a few species were caught in significant numbers and of these only a few had a swimbladder. Only fish with swimbladders were regarded as being relevant to the current analysis. Judging from the relative

Diel migration and swimbladder resonance of small fish 1145 Figure 2. An 18-kHz echogram exemplifying the DVM during one cycle (midnight to midnight). For the upper migrating shallow layer, the RFRs at 18, 38, 70, 120, and 200 khz were plotted vs. time intervals throughout a vertical migration. For the lower non-migrating deep layer and the intermediate migrating deep layer, the RFRs were only calculated at 18, 38, and 70 khz, because of low signal-to-noise ratios at the higher frequencies. The times of sunrise (left arrow) and sunset (right arrow) are indicated. Depth (m) and colour scale (db) are indicated on the left and right axes, respectively. Table 1. Modelled resonance at 18 khz, including estimates of requisite swimbladder dimensions and the corresponding RFR 18 at depths of 200 and 400 m. Depth (m) Bubble (mm) Estimated Spheroid major-axis (ka 5 0.2) (mm) RFR 18 Bubble (mm) Observed Spheroid major-axis (ka 5 0.2) (mm) RFR 18 200 0.8 2.7 12.4 1.0 3.2 3.3 400 1.2 3.8 9.0 1.4 4.6 2.6 The observed swimbladder dimensions are estimated from the model constrained with the observed RFR 18. The acoustic wave number is denoted as k. Table 2. Details of fish with swimbladders caught regularly within the study area. Common name Family/species Swimbladder Myctophid Myctophidae Gas- and lipid-filled Fish length (cm) Trawl depth range (m) 8 10 0 500 Hatchetfish Sternoptychidae Gas-filled 4 10 0 500 Pearlside Sternoptychidae/ Maurolicus muelleri Gas-filled 4 7 0 170 species abundance in the catches, the main acoustic backscatterers in the three layers were myctophids (Myctophidae), hatchetfish (Sternoptychidae), and pearlsides (Maurolicus muelleri; Table 2). The catch depths indicated that the upper migrating layer was dominated by pearlsides, and the migrating and non-migrating deep layers were mostly formed by myctophids. Hatchetfish, although less abundant than myctophids, were most frequently associated with the deep migrating layer. The s A values of the three layers varied substantially over time (Figure 3), although s A at all three f values were similar for the nonmigrating deep layer (Figure 3c), and s A at 18 khz was higher than that at 38 and 70 khz for the other two layers (Figure 3a and b). The maxima in the time-series of RFR f were identified for the two migrating layers (Figure 4a and b). For the deep migrating layer, the RFR 18 exhibited maxima at 03:00, 08:00, 15:00, and 19:00 (local time), with maxima close to 3 and minima below 2. The RFR 70 varied near 1. The RFR 18 for the migrating shallow layer fell from a maximum of 2 in the early morning to a minimum of 1 at night-time. The RFR 70 displayed no trend and varied near 1. To explore possible changes in f r vs. depth, the RFR 18 were plotted in 10-m bins (Figure 5). To obtain a full depth range of the myctophids, data from the migrating and non-migrating deep layers were combined. These RFR 18 exhibited maxima near 200 m and 380 m, and minima between these depths, at the surface and at 600 m. Again, RFR 18 varied much more than RFR 70. The models indicate that resonance at 200 and 400 m produced much larger RFR values (Table 1) than the maxima recorded in this study. The migrating shallow layer yielded similar results. Specifically, there were systematic changes in RFR 18 and reasonably constant RFR 70 vs. depth (Figure 6). The RFR 18 had largest values with poorly defined peaks at the surface and near 600 m, and minima of 1 from 80 to 120 m. The day and night RFR f values did not differ. Discussion According to Butler and Pearcy (1972), there are three general types of myctophid swimbladder, namely gas-filled, fat-invested, and atrophied. Small fish tend to have ellipsoid, thin-walled, gasfilled bladders. As they grow, there is an increasing probability of bladder modification through lipid filling, reduced size, or shrinking and thickening of walls. Hence, smaller individuals can have larger s bs than larger fish. Details of the species and individuals in this study are not available, but Butler and Pearcy (1972) suggest that swimbladder volumes of:

1146 O. R. Godø et al. Figure 4. The RFR 18 (red) and RFR 70 (blue) for the migrating shallow layer (a) and the deep layers (b) by time-of-day. The s A,10 m 2 nautical mile 22 were thresholded before calculating RFR values. Near midnight, the RFR values are missing for the migrating shallow layer, because it ascends into the echosounder s blind zone. Figure 3. The s A by time-of-day for (a) the migrating shallow layer, (b) migrating deep layer, and (c) the non-migrating deep layer as measured by 18 (continuous line), 38 (dotted line), and 70 khz (hatched line). (i) Stenobrachius leucopsarus and Stenobrachius nannochir are 4 5% of their body volume, and this reduces to 0.4% when their length is.70 mm; (ii) Lampanyctus ritteri and Lampanyctus regalis are 2% of their body volume, and this reduces to 0.01% in fish.80 mm (not fat-invested); and (iii) Diaphus theta are 7% of their body volume, and this reduces to 0.3% in fish.25 mm. Yasuma et al. (2003) studied D. theta with lengths ranging from 26.9 to 77.4 mm and found a ¼ 2.4 8.7 mm. The expected volume of swimbladders resonant near 18 khz for the current study was estimated from the literature (Table 2). Because of the limited biological sampling and a lack of studies of the lipid content of the swimbladders of deep-migrating fish, it is unclear why RFR 18 peaked at two different depths. Swimbladder and body volumes are inversely related to body size and lipid content, making estimates of bubble dimensions very uncertain. Adding to this complexity is the multipurpose use of the swimbladder. The lipid content has a buoyancy function, but its major purpose is thought to be as an energy reserve (Phleger, 1998). Hence, both size and energy conditions determine the acoustic properties of these fish. A spectrum of potential population structures could therefore have produced the two peaks for the myctophids. If the radii of the major and minor axes of the swimbladders of hatchetfish and M. muelleri at the surface comprise 30 40% of the total fish length, the swimbladder axes of these species will be in the ranges 1.2 4 cm in hatchetfish and 1.2 2.8 cm in M. muelleri. Assuming no gas exchange and compression of a spherical swimbladder following Boyle Mariotte s law, the swimbladder radii of hatchetfish at a depth of 200 m should be roughly 3.0 0.5 mm, and at 400 m roughly 1.3 1.9 mm. Hatchetfish could have played some role in the deep migrating layer, although their relative abundance in the catches was small.

Diel migration and swimbladder resonance of small fish 1147 Figure 5. The RFR 18 (left) and RFR 70 (right) by depth for combined migrating and non-migrating deep layers. The s A,2m 2 nautical mile 22 were thresholded before calculating RFRs. Figure 6. The RFR 18 (left) and RFR 70 (right) by depth for the migrating shallow layer. The s A,2m 2 nautical mile 22 were thresholded before calculating RFRs. Maurolicus muelleri was found in shallower water and was presumably the main contributor to the echo abundance in this layer. Their estimated swimbladder at 200 m is 6.0 8.0 mm, and this falls outside the expected resonance size. The increase in RFR 18 towards the extremes of the distribution range could be a first indication of resonance. Resonance is, however, unlikely to occur at both depths, and contamination from the deep migrating layer in the extremes of its distribution range could have contributed to this effect. Moreover, modification of behaviour as an effect of altered buoyancy (Arnold and Greer, 1992; Huse and Ona, 1996; Godø et al., 2006) could have had an effect. This study has demonstrated that resonance in fish with swimbladders could greatly affect the relative scattering response and degrade classifications of sound scatterers. The 18 khz recordings were particularly susceptible to resonance, and this frequency should be used with caution. These findings are supported by the calculated theoretical values of Kloser et al. (2002), who estimated a resonance peak for myctophids at 20 khz. On the positive side, the dynamics of the resonance phenomenon includes information about the fish and their behaviour. The results of this study also support the use of broad-bandwidth acoustics (Nero et al., 2004) for a better resolution of the temporal and

1148 O. R. Godø et al. depth dynamics of the phenomenon. Multiple-frequency, broadbandwidth echosounders should therefore preferably be used in studies of the behavioural characteristics and species compositions of mesopelagic fish. Conclusion Both deep (mainly myctophids) and shallow (mainly M. muelleri) migrating layers exhibited significant temporal variation in the RFR 18, but relatively little in the RFR 70. The deep layers also revealed two well-defined peaks in RFR 18 vs. depth, at the time of vertical migration. In contrast, the migrating shallow layer exhibited its highest RFR 18 values at the extremes of this layer s depth range, and a minimum near 100 m. Perhaps the RFR 18 values from both layers were affected by changes in f r. This could be important for properly apportioning S v to species during acoustic-survey analyses. These results demonstrate that swimbladder resonance complicates echo classifications, specifically when small fish with extensive vertical migrations are involved. Future studies of the resonance phenomenon with multifrequency or broad-bandwidth echosounders should enhance our understanding of the behavioural dynamics of such fish and consequently improve our ability to classify them remotely. Moreover, the development of such methods should facilitate an enhanced understanding of open-ocean ecosystems. Acknowledgements We are grateful to Kenneth Foote of the Woods Hole Oceanographic Institution for valuable discussion during the analysis, and three anonymous referees for useful guidance during the final preparation of the manuscript. The study was partly funded by the Research Council of Norway, through the Eco-Fish project (Project number 173536). References Andreeva, I. B. 1964. Scattering of sound by air bladders of fish in deep sound-scattering ocean layers. Soviet Physics Acoustics, 10: 17 20. Arnold, G. P., and Greer, W. M. 1992. Vertical movements of cod (Gadus morhua L.) in the open sea and the hydrostatic function of the swimbladder. ICES Journal of Marine Science, 49: 357 372. Butler, J. L., and Pearcy, W. G. 1972. Swimbladder morphology and specific gravity of myctophids off Oregon. Journal of the Fisheries Research Board of Canada, 29: 1145 1150. Chapman, R. P., Bluy, O. Z., Adlingto, R. H., and Robison, A. E. 1974. Volume reverberation in northern waters. Journal of the Acoustical Society of America, 55: 418. Foote, K. G., Knudsen, H. P., Vestnes, G., MacLennan, D. N., and Simmonds, E. J. 1987. Calibration of acoustic instruments for fish density estimation: a practical guide. ICES Cooperative Research Report, 144. Godø, O. R., Hjellvik, V., and Tjøstheim, D. 2006. Diurnal variation in frequency response of gadoids in the Barents Sea. ICES Document CM 2006/I: 23. Holliday, D. V. 1972. Resonance structure in echoes from schooled pelagic fish. Journal of the Acoustical Society of America, 51: 1322 1332. Horne, J. K. 2000. Acoustic approaches to remote species identification: a review. Fisheries Oceanography, 9: 356 371. Huse, I., and Ona, E. 1996. Tilt angle distribution and swimming speed of overwintering Norwegian spring spawning herring. ICES Journal of Marine Science, 53: 863 873. Kloser, R. J., Ryan, T., Sakov, P., Williams, A., and Koslow, J. A. 2002. Species identification in deep water using multiple acoustic frequencies. Canadian Journal of Fisheries and Aquatic Sciences, 59: 1065 1077. Korneliussen, R. J., Diner, N., Ona, E., Berger, L., and Fernandes, P. G. 2008. Proposals for the collection of multifrequency acoustic data. ICES Journal of Marine Science, 65: 982 994. Korneliussen, R. J., and Ona, E. 2000. Some applications of multiple frequency echo sounder data. ICES Journal of Marine Science, 59: 291 313. Korneliussen, R. J., Ona, E., Eliassen, I., Heggelund, Y., Patel, R., Godø, O. R., Giertsen, C., et al. 2006. The large-scale survey system LSSS. Proceedings of the 29th Scandinavian Symposium on Physical Acoustics, Ustaoset, 29 January 1 February 2006. Love, R. H., Fisher, R. A., Wilson, M. A., and Nero, R. W. 2004. Unusual swimbladder behavior of fish in the Cariaco Trench. Deep Sea Research I: Oceanographic Research Papers, 51: 1 16. MacLennan, D. N., Fernandes, P. G., and Dalen, J. 2002. A consistent approach to definitions and symbols in fisheries acoustics. ICES Journal of Marine Science, 59: 365 369. MacLennan, D. N., and Holliday, D. V. 1996. Fisheries and plankton acoustics: past, present, and future. ICES Journal of Marine Science, 53: 513 516. Neighbors, M. A., and Nafpaktitis, B. G. 1982. Lipid compositions, water contents, swimbladder morphologies and buoyancies of 19 species of midwater fishes (18 myctophids and 1 neoscopelid). Marine Biology, 66: 207 215. Neilson, J. D., and Perry, R. I. 1990. Diel vertical migrations of marine fishes: an obligate or facultative process? Advances in Marine Biology, 26: 115 168. Nero, R. W., Thompson, C. H., and Jech, J. M. 2004. In situ acoustic estimates of the swimbladder volume of Atlantic herring (Clupea harengus). ICES Journal of Marine Science, 61: 323 337. Pearcy, W. G., Krygier, E. E., and Cutshall, N. H. 1977. Biological transport of Zn-65 into deep-sea. Limnology and Oceanography, 22: 846 855. Phleger, C. F. 1998. Buoyancy in marine fishes: direct and indirect role of lipids. American Zoologist, 38: 321 330. Simmonds, E. J., Armstrong, F., and Copland, P. J. 1996. Species identification using wideband backscatter with neural network and discriminant analysis. ICES Journal of Marine Science, 53: 189 195. Wenneck, T. D., Falkenhaug, T., and Bergstad, O. A. 2008. Strategies, methods, and technologies adopted on the RV G. O. Sars MAR-ECO expedition to the Mid-Atlantic Ridge in 2004. Deep Sea Research II: Topical Studies in Oceanography, 55: 6 28. Weston, D. E. 1967. Sound propagation in the presence of bladder fish. Underwater Acoustics, 2: 55 88. Yasuma, H., Sawada, K., Ohshima, T., Miyashita, K., and Aoki, I. 2003. Target strength of mesopelagic lanternfishes (family Myctophidae) based on swimbladder morphology. ICES Journal of Marine Science, 60: 584 591. Ye, Z. 1997. Low-frequency acoustic scattering by gas-filled prolate spheroids in liquids. Journal of the Acoustical Society of America, 101: 1945 1952. doi:10.1093/icesjms/fsp098