On Turbulence Production by Swimming Marine Organisms in the Open Ocean and Coastal Waters

SEPTEMBER 2010 R O U S S E A U E T A L . 2107

SHANI ROUSSEAU,* ERIC KUNZE,RICHARD DEWEY,KEVIN BARTLETT, AND JOHN DOWER SEOS, University of Victoria, Victoria, British Columbia, Canada

(Manuscript received 15 December 2009, in ﬁnal form 24 April 2010)

ABSTRACT

Microstructure and acoustic profile time series were collected near Ocean Station P in the eastern subarctic North Pacific and in Saanich Inlet at the south end of Vancouver Island, British Columbia, Canada, to examine production of turbulent dissipation by swimming marine organisms. At Ocean Station P, although a number of zooplankton species are large enough to generate turbulence with Reynolds numbers Re . 1000, biomass densities are typically less than 103 individuals per cubic meter (,0.01% by volume), and turbulent kinetic energy dissipation rates « were better correlated with 16-m vertical shear than acoustic backscatter layers. In Saanich Inlet, where krill densities are up to 104 individuals per cubic meter (0.1% by volume), no dramatic elevation of dissipation rates « was associated with dusk and dawn vertical migrations of the acoustic backscatter layer. Dissipation rates are a factor of 2 higher [h«i 5 1.4 3 1028 Wkg21, corresponding to buoyancy Re 5 h«i/(nN2) ; 140] in acoustic backscatter layers than in acoustically quiet waters, regardless of whether they are vertically migrating. The O(1 m) thick turbulence patches have vertical wavenumber spectra for microscale shear commensurate with the Nasmyth model turbulence spectrum. However, the turbulence bursts of O(1025 Wkg21) proposed to occur in such dense swarms appear to be rare. Thus far, intense turbulent bursts have been found infrequently, even in very dense aggregations O(104 individuals per cubic meter) characteristic of coastal and high-latitude environs. Based on sampling to date, this corresponds to a frequency of occurrence of less than 4%, suggesting that turbulence production by the marine biosphere is not efficient.

1. Introduction quantify the possible impact on ocean turbulence and mixing. Turbulence production in the stratified ocean interior Energetic arguments suggest that significant turbulent is largely attributed to breaking internal gravity waves dissipation rates might be generated by aggregations of generated by (i) atmospheric forcing at the surface and swimming marine organisms. Based on the Reynolds (ii) tidal flow over topography (Munk and Wunsch 1998). numbers of individual organisms, Huntley and Zhou While it has long been known that swimming marine (2004) predicted turbulent kinetic energy dissipation organisms produce a turbulent wake (Wiese and Ebina « ; 25 21 1995; Yen 2000; Yen et al. 2003; Catton et al. 2008), the rates O(10 Wkg ) for organisms ranging from research focus has been on the energetic and detection 0.5-cm-long zooplankton to cetaceans, a value compara- consequences for these organisms (e.g., Enders et al. ble to that driven by storms in surface waters. Globally, 2003; Pitchford et al. 2003). Little has been done to Dewar et al. (2006) estimated that up to 1 TW might be available from swimming marine organisms, assuming 30 Gt of biomass [equivalent to ;O(0.1%) by volume if distributed over the upper 100 m], comparable to the * Current affiliation: Que´bec-Oceán, De´partement de Biologie, Universite´ Laval, Quebec City, Quebec, Canada. 2 TW needed for abyssal mixing, 0.8 TW of deep-ocean tidal dissipation (Egbert and Ray 2001), and 1 TW of deep-ocean wind forcing (Wunsch and Ferrari 2004); Corresponding author address: Eric Kunze, SEOS, University for comparison, global human power consumption of Victoria, P.O. Box 3055, STN CSC, Victoria BC V8W 3P6, Canada. was 15 TW in 2008. Munk (1966) came up with simi- E-mail: [email protected] lar numbers for the energy available for biologically

DOI: 10.1175/2010JPO4415.1

Ó 2010 American Meteorological Society Unauthenticated | Downloaded 09/28/21 11:28 AM UTC 2108 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 40 generated turbulence by assuming that 20% of bi- to 105 are high enough that a high turbulent mixing effi- ological consumption was used for swimming. Because ciency (g ; 0.2) would be expected. Moreover, they were the bulk of the ocean’s biomass is found in the upper few accompanied by strong temperature-gradient micro- hundred meters and his primary interest was in abyssal structure (though the complicated thermohaline vari- mixing, he did not pursue this thread further. Bi- ability in Saanich Inlet precludes quantifying a reliable ologically generated turbulence is unlikely to play a sig- mixing efficiency) and density overturns on the length nificant role in the meridional thermohaline circulation scales expected for the observed dissipation rates [LO 5 because of biomass’s intensification toward the surface («/N3)1/2]. and coastal zones. Even more anecdotal evidence comes from a single If 1 TW is distributed over the upper 100 m, it im- profile collected in a dense aggregation of krill being fed plies average turbulent kinetic energy dissipation rates upon by a pod of humpback whales in Soquel Canyon, h«i ; 5 3 1028 Wkg21 [corresponding to a buoyancy California, during August 2008, which revealed dissipa- Reynolds number h«i/(nN2) 5 500]. This is well within tion rates of 1026 Wkg21, the highest during the cruise, the detection limit of conventional microstructure pro- exceeding dissipation rates found in the near-bottom filers (Dewey et al. 1987). stratified turbulent layer (Kunze et al. 2010) by an order Much of any turbulence generated by swimming ma- of magnitude. In contrast, Rippeth et al. (2007) found no rine organisms is likely to occur in the surface boundary elevated turbulence associated with diel migration of the layer, which is already well mixed by atmospheric forc- acoustic backscatter layer in 13 dawn and dusk time series ing. However, many schooling species exhibit diel ver- from the western continental shelf of the British Isles. tical migration through stratified waters of the upper Another issue is how often biologically generated tur- pycnocline. During daylight, they evade predation by bulence does any mixing. Measurements by Gregg and resting in the dark below 100-m depth, surfacing at Horne (2009) in Monterey Bay fish aggregations ex- nightfall to feed (Forward 1988). Biologically generated hibited high turbulent kinetic energy dissipation rates, turbulence from such diel vertical migratory species could as previously reported by Farmer et al. (1987), but with potentially mediate transport of nutrients and dissolved (i) microscale shear spectra depleted at low wavenumber gases between the surface mixed layer and underlying and (ii) very weak temperature-gradient microstructure 2 2 stratified waters where mixing processes are not well xT and hence low mixing efficiency g 5 N xT /(Tz«) # understood (Johnston and Rudnick 2009). This could 0.0022, where xT is the turbulent temperature variance contribute to providing nutrients to maintain biological dissipation rate. On the other hand, Lorke and Probst productivity in the often nutrient-depleted euphotic zone. (2010) inferred mixing efficiencies g 5 0.2 associated with Assuming an upper-ocean buoyancy frequency N 5 turbulence generated by shoals of perch in Lake Con- 1022 rad s21 and an upper-bound mixing efficiency g ; 0.2 stance but reported dissipation rates « two orders of (Osborn 1980; Oakey 1982), h«i ; 5 3 1028 Wkg21 cor- magnitude smaller than predicted by either the bio- responds to diapycnal eddy diffusivities K 5 gh«i/N2 ; mechanical arguments of Huntley and Zhou (2004) or 1024 m2 s21. Purely physical mesoscale (McGillicuddy metabolic arguments. Gregg and Horne (2009) suggested and Robinson 1997; McGillicuddy et al. 1998, 1999) and that the fish aggregations responsible for their high tur- submesoscale (Mahatavan and Tandon 2006) mecha- bulent dissipation rates created shear too close to the nisms have also been proposed for transporting nutri- Kolmogorov scale (n3/«)1/4 ; O(1 cm) (Tennekes and ents into the euphotic zone (Klein and Lapeyre 2009). Lumley 1972), which was then viscously squelched rather Kunze et al. (2006) reported observational evidence than generating broadband turbulence and efficient mix- for biologically generated turbulence in Saanich Inlet, ing, an argument consistent with grid-generated turbu- British Columbia (BC), Canada, a sheltered fjord at the lence experiments (Itsweire et al. 1986). In contrast, south end of Vancouver Island. During the 28 April 2005 shear-driven turbulence [reduced shear Vz 2 2N . 0(Sun dusk upward migration of an acoustic backscatter layer et al. 1998) or, equivalently, gradient Richardson number composed of the krill species Euphausia pacifica with Ri , 0.25] is generated at the largest (Ozmidov) scales up to 104 individuals per cubic meter (0.1% by volume; («/N3)1/2 ; O(1 m), allowing turbulence to overcome De Robertis et al. 2003), they observed a coincident 10– stratification and cascade energy down to Kolmogorov

15-min burst of turbulence with dissipation rates of scales through nonlinear interactions, provided LO LK, 1025 to 1024 Wkg21 between 50- and 100-m depths, before being viscously damped. consistent with the predictions of Huntley and Zhou Although capable of generating turbulence, individ- (2004). These measurements had microstructure shear ual zooplankton and small fish may not be able to mix spectra that fit the Nasmyth (1970) model turbulence efficiently because their lengths are comparable to the spectra. Buoyancy Reynolds numbers Re 5 «/(nN2) 5 104 Kolmogorov length scales. However, the coordinated

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC SEPTEMBER 2010 R O U S S E A U E T A L . 2109 movement of a dense aggregation might produce turbu- anywhere from a 1 in 10 to 1 in 1000 chance of sampling lence at the thickness of such aggregations with outer (i) in an aggregation. Our Saanich Inlet data were col- (Ozmidov) scales of O(1–10 m) (Kunze et al. 2007; Catton lected in a regime of widespread krill aggregations, so the et al. 2008). If so, the density and swimming behavior of chance of sampling is O(1). We will focus on the efficacy such aggregations will be important factors in the gener- of turbulence production (ii) in this paper. ation of turbulent mixing. This parameter space remains To accumulate statistics of biologically generated tur- unexplored. bulence and better constrain the occurrence of turbu- Katija and Dabiri (2009) examined the problem of lence production in conjunction with aggregations of water being dragged with an object moving in a fluid and swimming marine organisms, an observational program concluded that this transport may be important regard- was embarked on to collect more data over a 3-yr period. less of whether turbulence is produced. More thorough analysis of the problem, albeit in the inviscid limit 2. Data and background (Thiffeault and Childress 2010), suggests very low induced diffusivities K 5 0.27Un‘4 ; 1026 m2 s21 asso- a. Field methods ciated with krill swarm densities even as high as 104 To examine turbulence generation by swimming zoo- individuals per cubic meter, where U is the swimming plankton, microstructure and acoustic backscatter pro- speed, n is the density in individuals per cubic meter, file time series were collected during the dusk and dawn and ‘ is the organism’s length scale. The strong de- vertical migrations of acoustic backscatter layers. Six pendence on length scale ‘ in this expression is at odds time series were gathered near Ocean Station P (OSP) with Katija and Dabiri’s claims. It suggests that ag- in the eastern subarctic North Pacific (488–508N, 1458W) gregations of larger organisms may be important for during 7–11 June 2007 and 11 were gathered in Saanich mixing. Both of these works ignore the role of molec- Inlet, British Columbia (488409N, 1238309W), during ular diffusion, which will eradicate microscale property May and June of 2006–08. A total of 394 microstructure differences and reduce net mixing efficiency. profiles were collected using a Rockland Scientific teth- Saanich Inlet has weak background dissipation rates ered freefall vertical profiler (available online at http:// O(1029 Wkg21) (Gargett et al. 2003; Kunze et al. 2006), www.rocklandscientific.com): 62 near OSP and 332 in corresponding to diffusivities K # 0.02 3 1024 m2 s21.If Saanich Inlet. The profiler is instrumented with fine- the elevated dissipation rates reported by Kunze et al. scale Seabird temperature and conductivity as well as (2006) occurred every dusk and dawn, they would dom- microscale shear, temperature and conductivity sen- inate mixing in the inlet, raising daily average diffu- sors. With a fall speed of about 0.6 m s21, each cast sivities to (4–40) 3 1024 m2 s21, a factor of 200–2000 took roughly 20 min near OSP and 6 min in Saanich higher than the background level. If a 1025 Wkg21 Inlet. burst (Huntley and Zhou 2004) were to occur for Simrad EK60 multiple-frequency (only 38-, 120-, and 20–30 min day21, daily averaged dissipation rates « ; 200-kHz, equivalent to 4-, 1-, and 0.8-cm wavelengths, O(1026 Wkg21) would be produced. In the open ocean, are used here) acoustic backscatter data are utilized for such a burst needs only to occur every 2–30 days to the OSP datasets while a 200-kHz (0.8-cm wavelength, produce an average diffusivity of 1024 m2 s21 and only 12-cm resolution) single-frequency ASL bioacoustics once every 1–12 months to produce average dissipation sensor (MacLennan and Simmonds 1992) was used in rates comparable to those of a Garrett–Munk internal Saanich Inlet. At OSP, the 120-kHz proved to be most wave field (« 5 5K N2, where K 5 0.1 3 1024 m2 s21; 0 0 useful for identifying and tracking acoustic backscatter Garrett 1979; Gregg 1987). layers. ADCP velocity profiles were also collected at The impact of biologically generated turbulence on OSP (38-kHz, 16-m resolution) and during the 2008 ocean mixing will depend on (i) the distribution and Saanich Inlet time series (300-kHz, 4-m resolution). density of dense swimming aggregations, (ii) how often and effectively such aggregations generate turbulence, b. Data processing and (iii) the mixing efficiency g of such turbulence. Their detection will also depend on the strength of turbulence The turbulent kinetic energy dissipation rate « 5 2 generated by physical processes: for example, wind forc- (15/2)nuz under the assumption of isotropy, where n 5 ing or nocturnal cooling of the surface mixed layer might 1026 m2 s21 is the kinematic molecular viscosity and mask any biological signal. Dense krill swarms in the uz is the microscale vertical shear. Where quoted, tur- open ocean are typically 10–1000 m long and separated bulent eddy diffusivities K 5 g«/N2 were inferred as- by several kilometers (Weber et al. 1986; Nicol 1986; suming a mixing efficiency g 5 0.2 following standard Barrange et al. 1993; Zhou and Dorland 2004), suggesting microstructure practice (Osborn 1980; Oakey 1982).

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC 2110 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 40

This might lead to overestimates of K because lower is orders of magnitude below that in coastal waters mixing efficiencies are expected for low-Reynolds-number (Goldblatt et al. 1999; Miller et al. 1984; Mackas and (Re , 200) turbulence (Gargett et al. 1984) and were re- Tsuda 1999), with densities up to 103 individuals per cubic ported in aggregations of small fish in Monterey Bay meter. None of these four species exhibits strong or (Gregg and Horne 2009). Onboard accelerometers were consistent diel vertical migration (Mackas et al. 1993). checked to ensure no contamination of the microscale Chaetognaths are the most abundant vertically migra- shear by instrument vibration. After despiking the mi- tory species (Goldblatt et al. 1999) but are acoustically croscale shear uz to remove particle impacts, vertical transparent. Migrating acoustic backscatter layers are wavenumber spectra resembled the Nasmyth (1970) dominated by 1.5-cm euphausiids, 0.15-cm pteropods, turbulence model spectrum. Despiked shear profiles and 2.8-cm myctophid fish of O(1 individual per cubic were broken into half-overlapping 4-m segments, Fourier meter) (Trevorrow 2005). transformed, then fit iteratively to the Nasmyth (1970) d. Saanich Inlet site model spectrum (Oakey 1982) over the resolved wavenumber range, which typically lies between but A 75–80-m sill at the mouth of Saanich Inlet limits includes neither the (i) Ozmidov (0.1–1 m) nor (ii) circulation of its deep-basin waters. The inlet is a reverse Kolmogorov (0.001–0.01 m) length scales. Dissipation estuary with its primary sources of freshwater outside rates « are then based on the shear variance integrated the inlet: Cowichan River to the north during winter and over the Nasmyth model spectrum. This methodology the Fraser River freshet during summer (Takahashi et al. resolves dissipation rates in the range 10210 , « , 1977; Gargett et al. 2003). Both wind and tidal forcing are 1026 Wkg21 with a factor of 2 uncertainty. The 4-m weak, so background turbulence dissipation rates are segments do not allow detection of turbulence at the low O(1029 Wkg21). length scale of an individual euphausiid but were used In addition to major spring and autumn phytoplankton to relate dissipation rate to the presence and movement blooms, smaller blooms occur fortnightly during neap of O(1 m) thick groups of krill. Temperature micro- tides throughout the summer (Parsons et al. 1983). Several structure confirmed that strong temperature gradients mechanisms have been proposed for the high primary accompanied strong microscale shear signals. However, productivity in Saanich Inlet (;490 gC m22 yr21, Timothy the complicated small-scale thermohaline structure of and Soon 2001; Grundle et al. 2009). Parsons et al. sug- Saanich Inlet precludes accurately quantifying turbulent gested intrusions of a strong tidal mixing front near the xT and mixing efficiency g, so this was not pursued here. mouth of the inlet, but these excursions proved not to Backscattering intensity is expressed in terms of vol- penetrate far enough into the fjord and Gargett et al. ume backscattering strength Sv 5 TS 1 10 log10(n), (2003) reported that the surface flow was outward during where n is the number of individuals per cubic meter and spring tides. Gargett et al. reported weak surface flow TS is the individual target strength (Clay and Medwin during neaps. They suggested that stronger spring tidal 1977). Diel biases in acoustic sampling of krill that ap- mixing just outside the inlet led to pressure gradients that pear to be due to differences in their orientation while drove surface waters outward and intermediate waters feeding and resting have been reported by Simard and inward, upwelling the nutrient-rich intermediate waters Sourisseau (2009). into the euphotic zone. Given expected lags in the system, this is consistent with primary production peaks every c. Ocean Station P site neap; nutrients are stirred throughout the inlet in about OSP has been studied for more than 50 yr (Tabata and a week by a counterclockwise circulation and a tidally Weichselbaumer 1992; Mackas et al. 1993; Freeland generated eddy field (E. Kunze 2010, unpublished man- et al. 1997), starting with data collected by Canadian uscript). Because the inlet’s circulation is weak, primary weatherships from 1956 to 1982 and continuing with the production remains local (Gargett et al. 2003). The ongoing Institute of Ocean Sciences Line P program spring–neap cycle is unresolved by our sampling. The (DFO 2009). OSP lies near the confluence of the sub- high primary productivity and low deep-water renewal polar and subtropical gyres in one of three high-nitrate, rate induce seasonally anoxic conditions below 100-m low-chlorophyll regions in the World Ocean (Boyd et al. depth (Jaffe et al. 1999). 1999). Annual primary productivity is about 140 com- High primary productivity also supports dense ag- pared to 250–500 gC m22 yr21 in BC coastal waters, the gregations of 1–2-cm-long Euphausia pacifica with up highest range corresponding to Saanich Inlet’s typical to 104 individuals per cubic meter (De Robertis 2002; productivity (Timothy and Soon 2001). As a conse- De Robertis et al. 2003; Parsons et al. 1983; Mackie quence, the zooplankton biomass, composed primarily of and Mills 1983; Beveridge 2007), one to two orders of four species of 0.2–0.6-cm copepods (Mackas et al. 1993), magnitude higher than typical open-ocean densities

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC SEPTEMBER 2010 R O U S S E A U E T A L . 2111

FIG. 1. Average proﬁles of (left)–(right) temperature T, salinity S, density r, buoyancy frequency N, 16-m shear, turbulent kinetic energy dissipation rate «, and diapycnal eddy diffusivity K (assuming mixing efﬁciency g 5 0.2) at OSP. The horizontal axes for (middle)–(right) panels are logarithmic.

(Greenlaw 1979) and corresponding to 0.1% by volume. are reduced during daylight, when the acoustic back- We rely on these extensive past measurements to assume scatter layer rests just above the ;100-m-deep oxycline. that the migrating acoustic backscatter layer is dominated by Euphausia pacifica. This species dominates the diel migrating acoustic backscatter signal (278 to 3. Results 271 dB re 1 m), though the layer may also contain less a. Ocean Station P abundant amphipods, ctenophores, copepods, juvenile fish and acoustically transparent chaetognaths. Using During our 6–11 June 2007 sampling, typical summer the same bioacoustic profiler, Beveridge (2007) inferred stratification was observed (Fig. 1; Freeland et al. 1997), higher zooplankton densities near the inlet mouth during with a well-mixed surface layer overlying a seasonal spring and summer, in agreement with Parsons et al. thermocline between 40 and 60 m of buoyancy frequency (1983) finding a front at the inlet’s mouth. Individual 2 3 1022 rad s21 (period ;5 min). A remnant winter krill swim at speeds up to 19 cm s21. However, even mixed layer of weaker stratification was observed be- during vertical migration, their swimming activity tends tween 60 and 90 m (Ueno et al. 2007) above the per- to be more random and episodic than synchronized manent halocline spanning 100–150-m depth (buoyancy (De Robertis et al. 2003). Swimming speeds in Saanich frequency 2 3 1022 rad s21). Greater depths are

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC 2112 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 40

FIG. 2. Sample profile time series of 120-kHz acoustic backscatter (grayscale) and turbulent kinetic energy dissipation rate « (red) from the (top) 6 Jun 2007 dusk and (bottom) 7 Jun 2007 dawn near OSP in the eastern subarctic North Pacific (488–508N, 1458W). The red index bar at the bottom of the rightmost profile ranges from log(«) 529to27(inWkg21). This open-ocean site is characterized by (i) thick nonmigrating acoustic backscatter layers at 40–50 m and, to a lesser extent, 110–120 and 170–180 m, and (ii) two migrating backscatter layers that rest below 250-m depth during the day and migrate into the upper 50 m at night with the more intense migrating layer lagging about half an hour at dusk and leading about half an hour at dawn. The acoustic backscatter data also reveals considerable finestructure. Persistent layers of high dissipation rate « are found in the 40–50-m-depth nonmigrating backscatter layer, which coincides with the seasonal thermocline, and near 100–120 m, coincident with the permanent halocline (Fig. 1). Higher 16-m ADCP shears are found at both these depths. characterized by a weaker pycnocline stratified by both Its acoustic signature as a function of frequency was temperature and salinity (buoyancy frequency 0.4 3 consistent with copepods and euphausiids (Lavery et al. 1022 rad s21). ADCP 16-m shear was elevated in both 2007). Based on the volume backscattering strength Sv, the seasonal thermocline (1022 s21) and permanent hal- copepod abundances are 102 to 103 individuals per cubic ocline (0.7 3 1022 s21). meter in the near-surface zooplankton layer, consistent Asingleverticalopen236-mm mesh net tow at the OSP with both our net tows and sampling during May 1996 open-ocean site found that four 0.2–0.6-cm-long copepod by Goldblatt et al. (1999). species and three 0.8–2-cm-long chaetognath species The 25–90-m-thick migrating layer observed above contributed 60% and 20% of the biomass, respectively, 300-m depth at OSP also had an acoustic signature in the upper 250 m, consistent with Mackas et al. consistent with euphausiids and copepods, although less (1993). Euphausiids were relatively low in abundance, pronounced than in the nonmigrating layer and often though the 0.56-m diameter of the net may have been consistent with Trevorrow’s (2005) finding of a mixed too small to sample accurately this species due to net signal from euphausiids and myctophid fish. Trevorrow avoidance (Lawson et al. 2008; Zhou et al. 1994). Pre- (2005) and Marlowe and Miller (1975) inferred 0.03 vious studies at OSP have found abundant euphausiids individuals per cubic meter for myctophids at 15230-m in summer (Goldblatt et al. 1999). depth and 1–10 individuals per cubic meter shallower A nonmigrating 10–25-m-thick acoustic backscatter than 100-m depth during night for euphausiids, although layer at 50-m depth (Fig. 2) coincided with the seasonal the euphausiids may have been undersampled due to net halocline and shallow high-shear (0.8 3 1022 s21) layer. avoidance. Trevorrow (2005) reported pteropods and

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC SEPTEMBER 2010 R O U S S E A U E T A L . 2113 chaetognaths in this layer as well, although the latter are acoustically transparent. Species composition at the station 180 km south of OSP differed from that at Station P. Two layers were observed to migrate downward at dawn with similar speeds, one preceding the other by 40 min. The acoustics suggest a greater abundance of the larger myctophids in the earlier denser migrating layer that descends below 250-m depth, whereas smaller euphausiids, copepods, or pteropods likely dominate in the later 25–30-m-thick layer, which rested at 180-m depth during daylight. The migrating acoustic backscatter layers traveled at 4.4 cm s21, at the low end of the average 5210 cm s21 swimming speeds of individual euphausiids during migration reported by De Robertis et al. (2003). Using Huntley and Zhou’s (2004) equation 1.5 3 105M0.63 , Re , 6.7 3 105M0.52 with euphausiid mass M 5 1025 kg c FIG. 3. Probability distribution functions of log(«) at OSP; the yields 1600 , Rec , 1700, whereas Re 5 uL/n 5 800– station 180 km south is similar. Probability distribution functions 1700 for 1.7-cm-long euphausiids and 900 for 2–6-cm-long in the migrating acoustic backscatter layer (green) and where the myctophid fish (Trevorrow 2005) swimming at 3 cm s21 acoustics detected no zooplankton (black) are similar, whereas (Baird and Jumper 1995), transitional for generation those in the 40–50-m-depth near-surface layer (red) are skewed toward values over an order of magnitude higher. of turbulence with mixing (Thorpe 2005; Gargett et al. 1984). 28 21 Dissipation rates rarely exceed 10 Wkg (Fig. 2) 27 21 in our measurements, with background levels of and (4 6 1) 3 10 Wkg [Re ; O(1000)] 180 km 10210 Wkg21. Microstructure shear spectra are consis- south. However, because this is at the base of the sur- tent with the universal Nasmyth (1970) turbulence model face mixed layer, these values could also be due to atmospheric forcing (Brainerd and Gregg 1993; Hosegood spectrum at both short and long wavelengths (lz 5 0.02–1 m), in contrast to Gregg and Horne’s (2009) et al. 2008). In both cases, the buoyancy Reynolds finding turbulent shear spectra depleted at longer tur- numbers are low. In two of the surface-layer acoustic bulent wavelengths within Monterey Bay fish aggrega- backscatter time series, dissipation rates were distrib- tions. With buoyancy Reynolds numbers Re 5 «/(nN2) ; uted identically to those in the deeper waters. These O(100), lower than those from Huntley and Zhou’s probability distribution functions were not lognormal, formula, turbulence may not be isotropic and mixing pointing to the presence of multiple processes (Gurvich efficiencies may fall below 0.2. and Yaglom 1967; Yamazaki and Lueck 1990). Mean At stratified depths of 60–250 m, dissipation rate dissipation rates and 95% confidence limits were there- distributions were not significantly different inside and fore calculated using the bootstrap method (Efron and outside layers of high acoustic backscatter (Fig. 3), with Gong 1983). averages of (1.1 6 0.9) 3 1029 Wkg21 [Re 5 10–100]. The weak turbulence measured at OSP is better cor- In contrast to many measurements in the pycnocline related with 16-m shear (or gradient Richardson number (e.g., Gregg and Sanford 1988), our dissipation rates Ri) than with acoustic backscatter. Dissipation rates « « are not proportional to N2, so inferred turbulent diffu- are better correlated (Spearman p , 0.05, r 5 0.3–0.8) sivities K were not independent of stratification, perhaps with 16-m Ri below 60-m depth than with 120-kHz vol- because of the sharpness of the two pycnoclines. In both ume backscattering strength Sv (20.36 , r , 0.37). The the migrating acoustic backscatter layer and acoustically Spearman rank correlation coefficient, quiet waters, the average dissipation rate was at the low n end of typical pycnocline values (Moum and Osborn 6 å (x y )2 1986; Gregg 1987; Ledwell et al. 1993). i i r 5 1 i51 , In the nonmigrating acoustic backscatter layer, which n(n2 1) spans 20–60-m depth, including waters inside the surface mixed layer and the high-shear seasonal thermo- is the nonparametric equivalent of the Pearson corre- cline, dissipation rates were significantly higher with an lation coefficient but uses rank order xi and yi rather 28 21 average (4 6 1) 3 10 Wkg [Re ; O(100)] at OSP than numeric values Xi and Yi (Sokal and Rohlf 1981),

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC 2114 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 40

FIG. 4. Average profiles of temperature T, salinity S, density r, buoyancy frequency N, 16-m shear, turbulent kinetic energy dissipation rate «, and diapycnal eddy diffusivity K (assuming g 5 0.2) in Saanich Inlet. so it is suitable for variables with non-Gaussian distri- the exception of two time series in 2007 during flood tide, butions such as dissipation rate. Correlation was higher all time series were collected during slack tide. The with both 16-m Ri and volume backscattering (r 5 buoyancy frequency is 2 3 1022 rad s21 in the upper 0.5–0.8) if 15–60-m depths were included. However, in 20–80 m, falling to 1022 rad s21 at 80–120-m depth, the shallow nonmigrating backscatter layer, high shear mostly because of salt stratification (Fig. 4). ADCP 1-m and high backscatter are collocated, so their impacts shears were only collected during the May 2008 mea- cannot be separated. surements and were about 0.5 3 1022 s21 throughout Given that turbulence dissipation may lag the gener- the water column, although higher shears were some- ation mechanism by as much as one buoyancy period times observed near the surface. (Tennekes and Lumley 1972), we also performed lagged Eleven dawn and dusk time series were collected in correlations ranging up to one buoyancy period. Lagged Saanich Inlet where intense biologically generated tur- correlations did not have higher values than the unlagged bulence was previously reported in 1 of 2 dusk time se- results. ries (Kunze et al. 2006). A strong 20–50-m-thick migrating acoustic backscatter layer was present in all Saanich Inlet b. Saanich Inlet dawn and dusk time series (Fig. 5), traversing from Saanich Inlet measurements were collected over 9– a 100-m daytime resting depth to the surface at dusk in 11 June 2006, 8–10 May 2007, and 7–9 May 2008. With 60–90 min with migration speeds of 0.7–3.1 cm s21. The

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC SEPTEMBER 2010 R O U S S E A U E T A L . 2115

FIG. 5. Sample profile time series of 200-kHz acoustic backscatter (grayscale) and turbulent kinetic energy dissipation rate « (red) from (top) 9 and 10 Jun 2006 dusks and (bottom) 10 Jun 2006 and 10 May 2007 dawns in Saanich Inlet (488409N, 1238309W). The red index bar at the bottom of the rightmost profile ranges from log(«) 529to27 (in W kg21). This coastal inlet is characterized by extremely dense swarms of Euphausia pacifica of up to 104 individuals per cubic meter that migrate from a dense surface layer at night to depths of ;80 m during the day, although the acoustic backscatter layer often appears more dispersed during the day.

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC 2116 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 40 migrating layer sometimes appeared to spread over the precludes reliable estimates of turbulent temperature entire sampled water column (Fig. 5) and sometimes variance dissipation xT and mixing efficiency g. Dissi- appeared to abruptly change depth, although this was pation rates « of 1027 Wkg21 (buoyancy Reynolds possibly aliased horizontal variability because of ship drift. numbers Re ; 103) were frequently observed at depth, During the June 2006 dusks, a second more diffuse layer whereas values of 1025 Wkg21 were only found shal- migrated 20 m below the first. Volume backscattering lower than 20-m depth and may have been generated strength Sv was higher (240 dB) at the surface during by surface cooling. night than during the migration period (269 to 252 dB), Average turbulent diapycnal diffusivities were about consistent with the organisms orienting vertically while 0.1 3 1024 m2 s21 between 20- and 80-m depth and 0.01 3 migrating (Simard and Sourisseau 2009), then aggregat- 1024 m2 s21 over 80–160-m depth assuming g 5 0.2. ing densely at the surface. Nonmigrating zooplankton For the inferred buoyancy Reynolds numbers below near the surface will also contribute to the acoustic signal 200, mixing efficiencies g may be smaller (Gargett et al. and fish may join the aggregation to feed (Greenlaw 1984). No significant correlation (Spearman coefficient 1979). Reasons for the slower migration rate in Saanich p 5 0.71) could be found between dissipation rates Inlet compared to OSP are not known. The depth range « and 4-m ADCP shear. traversed by the backscatter layer is smaller. As well, Excluding the upper 30 m to remove the influence most zooplankton species are thought to use light in- of nocturnal cooling, average dissipation rates h«i are a tensity as a proxy for vulnerability to visual predators factor of 2 higher when volume backscattering strength (Boden and Kampa 1965). Thus, variations in water Sv is high in 8 of the 11 time series using threshold Sv of opacity, cloud cover, sunlight, and lunar cycle may all 267 (2006–07) and 270 (2008) dB (Fig. 6). Average dis- influence migration rates. This dependence on light in- sipation rates are 1.4 3 1028 Wkg21 [buoyancy Reynolds tensity is also likely size dependent (De Robertis 2002). number Re 5 «/(nN2) 5 140] in high acoustic backscat- Responses to light intensity can be modified by chem- ter and 0.7 3 1028 Wkg21 (Re 5 70) in acoustically ical cues from visual predators and food availability quiet waters. Dissipation rates « were highest for the (Ringelberg 1995; Forward and Rittschoft 2000). June 2006 sampling (time series 1–4) during full moon Individual Euphausia pacifica swim at speeds of 3– and slack tide, lowest for May 2007 (time series 5–7) 19 cm s21 at 608 to the horizontal to reduce visibility to during a third quarter moon, also during slack. The May predators (De Robertis et al. 2003), with the bulk swim- 2008 sampling (time series 8–11) also had a quarter moon ming at the low end of the speed range. With lengths of and slack tide. Average volume backscattering strengths 0.012–0.022 m, this implies Reynolds numbers UL/n of Sv were significantly (by 2–12 dB) higher in high dissi- up to 2400 for individuals (Torres 1984). Using 20–50-m pation rates in 5 of 11 time series using a threshold « 5 aggregation thicknesses implies Reynolds numbers of 1028 Wkg21. Spearman coefficients indicated signifi- 3 3 105, suggesting that krill swarms would generate cant positive correlations (p , 0.05) between volume turbulence much more effectively than individuals. With backscattering strength Sv and kinetic energy dissi- typical target strength of 279 dB (Trevorrow et al. 2005), pation rates « over 30–120-m depth in all but the May we deduce densities of 30–800 individuals per cubic meter 2008 time series. in the migrating layer and up to 8000 near the surface at Mean dissipation rates did not differ significantly be- night. However, these results are uncertain as they as- tween migrating and nonmigrating acoustic backscatter sume a monospecific euphausiid population, which was layers, suggesting that nonmigrating krill produce simi- not confirmed. lar turbulence levels as migrating swarms. Lagging up to Dissipation rates exceeding 1026 Wkg21 (Re 5 104) one buoyancy period in time increased the correlations were observed in 73 out of 376 profiles (Fig. 5). How- in five time series but did not change overall results. ever, high dissipation rates were observed in acoustically Migrating and daylight resting layers exhibit the same quiet waters as well as in the acoustic backscatter layers, nonlognormal probability density functions for dissipa- suggesting other sources of turbulence production in tion rate seen at OSP. The dusk time series collected on the inlet. For dissipation rates exceeding 1029 Wkg21, 28 April 2005 (Kunze et al. 2006) was the only one ex- (i) the microscale shear spectra match the Nasmyth hibiting dramatically higher dissipation rates within the (1970) model spectra over wavelengths of 0.01–1 m, migrating acoustic backscatter layer. No patterns could (ii) microscale temperature-gradient spectra are well be discerned in the standard deviations or skewnesses of resolved over 0.003–3 m, and (iii) resolved microstructure the probability distribution functions. temperature-gradient variance is well correlated with During a dusk time series collected during July 2009, kinetic energy dissipation rates «, although the com- there was again no intensified turbulence associated plicated finescale water-mass variability in Saanich Inlet with upward migration of the acoustic backscatter layer.

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC SEPTEMBER 2010 R O U S S E A U E T A L . 2117

FIG. 6. Average dissipation rates h«i with 95% conﬁdence limits below 30-m depth for 11 dawn and dusk time series collected in Saanich Inlet during 2006–08 binned by high and low acoustic backscattering signal (267 dB in 2006 and 2007 and 270 dB in 2008, based on the backscattering probability distribution function). Dissipation rates are on average a factor of 2 larger in high acoustic backscattering.

However, earlier in the day, three bursts of 1026 Wkg21 backscatter layer, « was comparably correlated with back- (Re 5 104) appeared in single profiles (separated by less scatter and 16-m Ri. Based on the acoustic backscatter than 6 min) as the ship drifted over the western slope of strength from three frequencies, we infer zooplankton the inlet. These bursts appeared to be correlated with densities of less than 103 individuals per cubic meter in elevated horizontal and vertical velocity rather than the nonmigrating backscatter layer, comparable to what high acoustic backscatter. Similar phenomena may have is found elsewhere in the open ocean. The migrating been responsible for the 28 April 2005 burst (Kunze et al. backscatter layer was likely composed of euphausiids 2006), although the latter’s duration was 5–7 profiles. and myctophid fish. Overall, we conclude that no un- Of five dawn time series collected during August 2006 ambiguous evidence for biologically generated turbu- in California continental margin waters (water depth lence could be identified near open-ocean site OSP 1000 m) off Point Sur, only one exhibited evidence for where abundances were orders of magnitude lower than elevated dissipation rates in the upper 50 m. This co- in the coastal ocean. The intermittent occurrence of higher incided with the highest acoustic backscatter signal of dissipation rates « seemed best explained by shear-driven the cruise. turbulence below 60 m and atmospheric forcing in the nonmigrating layer. In Saanich Inlet, the krill species Euphausia pacifica 4. Summary dominates the acoustic backscatter signal with densities To obtain statistics of turbulence production by swim- of up to 104 individuals per cubic meter. These are among ming marine organisms, profile time series were collected the densest aggregations in the world, comparable to of acoustic backscatter, finescale vertical shear, and mi- those found in swarms off Antarctica (Hamner et al. 1983; croscale kinetic energy dissipation rates spanning the diel Zhou and Dorland 2004) and in the Arctic (Siegel 2000). dawn and dusk vertical migration of the acoustic back- No recurrence of a burst of 1025 Wkg21 dissipation rate scatter layer. Six time series were collected near OSP « (Kunze et al. 2006) was observed in our 11 dawn and during June 2007, representing the first attempt to look dusk time series. Below the surface layer, dissipation rates for biologically generated turbulence in the deep open were elevated by a factor of 2 in acoustic backscattering ocean. Eleven time series were collected in Saanich Inlet, layers (« 5 1.4 3 1028 Wkg21,implyingK # 0.3 3 a sheltered coastal fjord, during May or June of 2006, 1024 m2 s21 for N 5 1022 rad s21 and a mixing efficiency 2007, and 2008. g # 0.2) compared to acoustically quiet waters. At OSP, turbulent dissipation rates « were better cor- At both sites, turbulent dissipation rates associated related with 16-m vertical shear or gradient Richardson with vertically migrating acoustic backscatter layers number (r 5 0.4–0.8) than with acoustic volume back- were an order of magnitude below predictions (Huntley scattering strength Sv. Below a nonmigrating layer at and Zhou 2004). Thus, turbulence production by the 40–50-m depth, average « was similar inside and out- marine biosphere appears to be inefficient, particularly side acoustic backscattering layers. In a nonmigrating in the deep ocean.

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC 2118 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 40

5. Discussion of efficient turbulent mixing by marine organisms may depend on the synchronized behavior of individuals At the OSP open-ocean site, where zooplankton den- within an aggregation, but the nature of this behavior is sities are up to 103 individuals per cubic meter in the poorly known. At high densities, they may engage in nonmigrating surface layer and 1–10 individuals per cubic synchronized swimming to improve their swimming ef- meter in the migrating layer, no elevated turbulence was ficiency and reduce risks of predation (Abrahams and identified unambiguously associated with the acoustic Colgan 1985; Jensen et al. 1998), but only a few studies backscatter layer. This suggests that the frequency of exist on krill aggregation behavior (Hamner et al. 1983; intense turbulence production by swimming marine or- Jensen et al. 1998). They are thought to respond to the ganisms in the open ocean is low, although we caution presence of predators, potential mates, and even tur- that, based on the statistics outlined in the introduction, bulence. In Saanich Inlet, individual euphausiids swim in sampling with six time series may be inadequate. bursts punctuated by rests and swim somewhat erratically, In Saanich Inlet, where krill densities are up to 104 in- even during diel migration (De Robertis et al. 2003). dividuals per cubic meter (0.1% by volume), events of Given the heterogeneity of dense (104 individuals per O(1025 Wkg21) (Kunze et al. 2006), as predicted by cubic meter) aggregations, even in most highly productive Huntley and Zhou (2004), also occur infrequently (not coastal and high-latitude waters (Weber et al. 1986; Nicol more than 8% of the time). Because of the spatial het- 1986; Barrange et al. 1993; Zhou and Dorland 2004), and erogeneity of aggregations of swimming marine organ- the less than 4% frequency with which they apparently isms, biologically generated turbulence is expected to be generate intense turbulence, exploring these questions an intermittent phenomenon. However, the inlet mea- presents a considerable sampling challenge. The ship- surements were in laterally broad layers of high acoustic board campaign approach taken here is neither cost ef- backscatter where, if Huntley and Zhou were correct, fective nor efficient. Continuous time series are needed to turbulence should be expected consistently. Taken to- place a lower bound on the frequency of intense turbulent gether with the findings of (i) Rippeth et al. (2007) of bursts associated with swimming marine organisms and no elevated turbulence in 13 dawn and dusk time series on relate such bursts to swimming behavior. the western continental shelf of the British Isles, (ii) Lorke and Probst (2010) of dissipation rates associated with Acknowledgments. We thank the captains and crews shoals of perch two orders of magnitude below the pre- of the MSV John Strickland and CCGS John P. Tully for dictions of Huntley and Zhou (2004), and (iii) Gregg and their assistance. Chris MacKay and Ian Beveridge played Horne (2009) that dense aggregations of small fish can critical roles in the data collection, Moira Galbraith an- generate elevated turbulent kinetic dissipation rates « not alyzed the OSP zooplankton samples, Doug Yelland in- accompanied by elevated mixing, our results suggest that structed us in operation of the Tully acoustics, and Marie intense biologically generated turbulence occurs infre- Robert served as chief scientist. MATLAB routines quently and biologically generated turbulent mixing were used to process data from the EK60 were de- occurs even more rarely. Thus, production of turbulence veloped by Richard Towler (NOAA Alaska Fisheries by the ocean biosphere does not appear to be an effective Science Center 2009, personal communication), ADCP process in contrast to the energetic arguments (Munk by Richard Dewey (University of Victoria 2009, personal 1966; Huntley and Zhou 2004; Dewar et al. 2006). With communication), and VMP by Kevin Bartlett (University measured dissipation rates « low,theissueofwhetherthe of Victoria 2009, personal communication). Ann Gargett mixing efficiency g is high (0.2) or low (0.0022; Gregg and and an anonymous reviewer are thanked for their use- Horne 2009) in our data is moot, so it was not examined. ful comments on the manuscript. The VENUS obser- Based on the ;4% occurrence of turbulence bursts in vatory program kindly loaned us the ZAP bioacoustic the limited sampling to date, we suggest that, rather than profiler. The Rockland Scientific microstructure in- 1 TW being available for turbulence production by strumentation was purchased with an NSERC CFI swimming marine organisms (Dewar et al. 2006), turbu- grant. Additional support for this work came from lence generation by the global marine biosphere may be NSERC Discovery Grants, an NSERC Shiptime Grant, no more than 0.04 TW (40 GW). Biologically generated and ONR Grant N0000140810700. turbulence may contribute most of the mixing in locations of weak physically generated turbulence and high biomass concentration such as the Arctic, but these are REFERENCES unlikely to be important for the global ocean. Abrahams, M. V., and P. W. Colgan, 1985: Risk of predation, hy- In light of the very different results of Gregg and drodynamic efficiency and their influence on school structure. Horne (2009) and Lorke and Probst (2010), generation Environ. Biol. Fishes, 13, 195–202.

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC SEPTEMBER 2010 R O U S S E A U E T A L . 2119

Baird, R. C., and G. Y. Jumper, 1995: Encounter models and deep- Freeland, H., K. Denman, C. S. Wong, F. Whitney, and R. Jacques, sea fishes: Numerical simulations and the mate location problem 1997: Evidence of change in the winter mixed-layer in the in Sternoptyx diaphana (Pisces, Sternoptychidae). Deep-Sea northeast Pacific Ocean. Deep-Sea Res. I, 44, 2117–2129. Res. I, 42, 675–696. Gargett, A. E., T. R. Osborn, and P. W. Nasmyth, 1984: Local Barrange, M., D. G. M. Miller, I. Hampton, and T. T. Dunne, 1993: isotropy and the decay of turbulence in a stratified fluid. Internal structure of Antarctic krill Euphausia superba J. Fluid Mech., 144, 231–280. swarms based on acoustic observations. Mar. Ecol. Prog. Ser., ——, D. Stucchi, and F. Whitney, 2003: Physical processes associ- 99, 205–213. ated with high primary production in Saanich Inlet, British Beveridge, I. A., 2007: Acoustic observations of zooplankton Columbia. Estuarine Coastal Shelf Sci., 56, 1141–1156. distribution in Saanich Inlet, an intermittently anoxic fjord. Garrett, C., 1979: Mixing in the ocean interior. Dyn. Atmos. M.S. thesis, Dept. of Biology, University of Victoria, Oceans, 3, 239–265. 147 pp. Goldblatt, R. H., D. L. Mackas, and A. G. Lewis, 1999: Meso- Boden, B. P., and E. M. Kampa, 1965: An aspect of euphausiid zooplankton community characteristics in the NE subarctic ecology revealed by echo-sounding in a fjord. Crustaceana, 9, Pacific. Deep-Sea Res. II, 46, 2619–2644. 155–173. Greenlaw, C. F., 1979: Acoustical estimation of zooplankton pop- Boyd, P. W., P. J. Harrison, and B. D. Johnson, 1999: The Joint ulations. Limnol. Oceanogr., 24, 226–242. Global Ocean Flux Study (Canada) in the NE subarctic Pa- Gregg, M. C., 1987: Diapycnal mixing in the thermocline. J. Geo- cific. Deep-Sea Res. II, 46, 2345–2350. phys. Res., 92, 5249–5286. Brainerd, K., and M. C. Gregg, 1993: Diurnal restratification and ——, and T. B. Sanford, 1988: The dependence of turbulent dissi- turbulence in the oceanic surface mixed layer. I: Observations. pation on stratification in a diffusively stable thermocline. J. Geophys. Res., 98, 22 645–22 656. J. Geophys. Res., 93, 12 381–12 392. Catton, K. B., D. R. Webster, and J. Yen, 2008: Can krill mix ——, and J. K. Horne, 2009: Turbulence, acoustic backscatter and the ocean? Preprints, Ocean Sciences Meeting, Orlando, FL, pelagic nekton in Monterey Bay. J. Phys. Oceanogr., 39, 1097– ASLO, 209. 1114. Clay, C. S., and H. Medwin, 1977: Acoustic Oceanography: Prin- Grundle, D. S., D. A. Timothy, and D. E. Varela, 2009: Variations ciples and Applications. John Wiley and Sons, 544 pp. of phytoplankton productivity and biomass over an annual cy- De Robertis, A., 2002: Small-scale spatial distribution of the eu- cle in Saanich Inlet, a British Columbia fjord. Cont. Shelf Res., phausiid Euphausia pacifica and overlap with planktivorous 29, 2257–2269. fishes. J. Plankton Res., 24, 1207–1220. Gurvich, A. S., and A. M. Yaglom, 1967: Breakdown of eddies and ——, C. Schell, and J. S. Jaffe, 2003: Acoustic observations of probability distributions for small-scale turbulence. Phys. Fluids, the swimming behavior of the euphausiid Euphausia pacifica. 10, S59–S65. J. Mar. Sci., 60, 885–898. Hamner, W. M., P. P. Hamner, S. W. Strand, and R. W. Gilmer, Dewar, W. K., R. J. Bingham, R. L. Iverson, D. P. Nowacek, 1983: Behavior of Antarctic krill Euphausia superba: Che- L. C. St. Laurent, and P. H. Wiebe, 2006: Does the marine moreception, feeding, schooling and molting. Science, 220, biosphere mix the ocean? J. Mar. Res., 64, 541–561. 433–435. Dewey, R. K., W. R. Crawford, A. E. Gargett, and N. S. Oakey, Hosegood, P. J., M. C. Gregg, and M. H. Alford, 2008: Re- 1987: A microstructure instrument for profiling oceanic tur- stratification of the surface mixed layer with submesoscale bulence in coastal bottom boundary layers. J. Atmos. Oceanic lateral density gradients: Diagnosing the importance of the Technol., 4, 288–297. horizontal dimension. J. Phys. Oceanogr., 38, 2438–2460. DFO, cited 2009: Fisheries and oceans Canada—Line P program. Huntley, M. E., and M. Zhou, 2004: Influence of animals on tur- History of oceanographic sampling at Ocean Weather Sta- bulence in the sea. Mar. Ecol., 273, 65–79. tion PAPA and Line P. [Available online at http://www. Itsweire, E. C., K. N. Helland, and C. W. van Atta, 1986: The pac.dfo-mpo.gc.ca/science/oceans/data-donnees/line-p/index- evolution of grid-generated turbulence in a stably-stratified eng.htm.] fluid. J. Fluid Mech., 162, 299–338. Efron, B., and G. Gong, 1983: A leisurely look at the bootstrap, the Jaffe, J. S., M. D. Ohman, and A. De Robertis, 1999: Sonar esti- jackknife and cross-validation. Amer. Stat., 37, 36–48. mates of daytime activity levels of Euphausia pacifica in Egbert, G. D., and R. Ray, 2001: Estimates of M2 tidal energy Saanich Inlet. Can. J. Fish. Aquat. Sci., 56, 2000–2010. dissipation from TOPEX/Poseidon altimeter data. J. Geo- Jensen, K. H., P. J. Jakobsen, and O. T. Kleiven, 1998: Fish phys. Res., 106, 22 475–22 502. kairomone regulation of internal swarm structure in Daphnia Enders,E.C.,D.Boisclair,andA.G.Roy,2003:Theeffectof pulex (Caldocera: Crustacea). Hydrobiologia, 368, 123–127. turbulence on the cost of swimming for juvenile Atlantic Johnston, T. M. S., and D. L. Rudnick, 2009: Observations of the salmon (Salmo salar). Can. J. Fish. Aquat. Sci., 60, 1149– transition layer. J. Phys. Oceanogr., 39, 780–797. 1160. Katija, K., and J. O. Dabiri, 2009: A viscosity-enhanced mechanism Farmer, D. D., G. B. Crawford, and T. R. Osborn, 1987: Temper- for biogenic ocean mixing. Nature, 460, 624–626. ature and velocity microstructure caused by swimming fish. Klein, P., and G. Lapeyre, 2009: The oceanic vertical pump in- Limnol. Oceanogr., 32, 978–983. duced by mesoscale and submesoscale turbulence. Annu. Rev. Forward, R. B., 1988: Diel vertical migration: Zooplankton pho- Mar. Sci., 1, 351–375, doi:10.1146/annurev.marine.010908. tobiology and behavior. Oceanogr. Mar. Biol. Ann. Rev., 26, 163704. 361–393. Kunze, E., J. F. Dower, I. Beveridge, R. Dewey, and K. P. Bartlett, ——, and D. Rittschoft, 2000: Alteration of photo-responses in- 2006: Observations of biologically-generated turbulence in volved in diel vertical migration of a crab larva by fish mucus a coastal inlet. Science, 313, 1767–1770. and degradation products of mucopolysaccharides. J. Exp. ——, ——, R. Dewey, and E. A. D’Asaro, 2007: Mixing it up with Mar. Biol. Ecol., 245, 277–292. krill. Science, 318, 1239.

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC 2120 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 40

——, K. J. Morrice, C. MacKay, K. Bartlett, E. E. McPhee-Shaw, Oakey, N. S., 1982: Determination of the rate of dissipation of J. B. Girton, and S. R. Brody, 2010: Stratified turbulent turbulent energy from simultaneous temperature and velocity bottom boundary layers in canyons. Extended Abstracts, shear microstructure measurements. J. Phys. Oceanogr., 12, Ocean Sciences Meeting, Portland, OR, Amer. Geophys. 256–271. Union, 37. Osborn, T. R., 1980: Estimates of the local rate of vertical diffu- Lavery, A. C., P. H. Wiebe, T. K. Stanton, G. L. Lawson, sion from dissipation measurements. J. Phys. Oceanogr., 10, M. C. Benfield, and N. Copley, 2007: Determining dominant 83–89. scatterers of sound in mixed zooplankton populations. J. Acoust. Parsons,T.R.,R.I.Perry,E.D.Nutborwn,W.Hsieh,and Soc. Amer., 122, 3304–3326. C. M. Lalli, 1983: Frontal zone analysis at the mouth of Saanich Lawson, G. L., P. H. Wiebe, T. K. Stanton, and C. J. Ashijan, 2008: Inlet, British Columbia, Canada. Mar. Biol., 73, 1–5. Euphausiid distribution along the western Antarctic Peninsula— Pitchford, J. W., A. James, and J. Brindley, 2003: Optimal foraging Part A: Development of robust multi-frequency acoustic in patchy turbulent environments. Mar. Ecol., 256, 99–110. techniques to identify euphausiid aggregations and quantify Ringelberg, J., 1995: Changes in light intensity and diel vertical euphausiid size, abundance and biomass. Deep-Sea Res. II, migration—A comparison of marine and fresh-water envi- 55, 412–431. ronments. J. Mar. Biol. Assoc. U. K., 75, 15–25. Ledwell, J. R., A. J. Watson, and C. S. Law, 1993: Evidence for slow Rippeth, T. P., J. C. Gascoigne, J. A. M. Green, M. E. Inall, mixing across the pycnocline from an open-ocean tracer- M. R. Palmer, J. H. Simpson, and P. J. Wiles, 2007: Turbulent release experiment. Nature, 364, 701–703. dissipation of coastal seas. Science. [Available online at http:// Lorke, A., and W. N. Probst, 2010: In situ measurements of tur- www.scienceonline.org/cgi/eletters/313/5794/1768.] bulence in fish shoals. Limnol. Oceanogr., 55, 354–364. Siegel, V., 2000: Krill (Euphausiacea) demography and variability Mackas, D. L., and A. Tsuda, 1999: Mesozooplankton in the eastern in abundance and distribution. Can. J. Fish. Aquat. Sci., 57 and western subarctic Pacific: Community structure, seasonal (S3), 151–167. life histories and interannual variability. Prog. Oceanogr., 43, Simard, Y., and M. Sourisseau, 2009: Diel changes in acoustic 335–363. and catch estimates of krill biomass. J. Mar. Sci., 66, 1318– ——, H. Sefton, C. B. Miller, and A. Raich, 1993: Vertical hab- 1325. itat partitioning by large calenoid copepods in the oceanic Sokal, R. R., and F. J. Rohlf, 1981: Biometry. 2nd ed. W.H. Freeman, subarctic Pacific during spring. Prog. Oceanogr., 32, 620 pp. 256–294. Sun, C.-J., W. D. Smyth, and J. N. Moum, 1998: Dynamic in- Mackie, G. O., and C. E. Mills, 1983: Use of the Pisces IV sub- stability of stratified shear flow in the upper equatorial Pacific. mersible for zooplankton studies in coastal waters of British J. Geophys. Res., 103, 10 323–10 337. Columbia. Can. J. Fish. Aquat. Sci., 40, 763–776. Tabata, S., and W. E. Weichselbaumer, 1992: An update of the MacLennan, D. N., and E. J. Simmonds, 1992: Fisheries Acoustics. statistics of hydrographic/CTD data taken at Ocean Station P Chapman and Hall, 325 pp. (May 1956-September 1990). Institute of Ocean Sciences Mahatavan, A., and A. Tandon, 2006: An analysis of mechanisms Canadian Data Rep. of Hydrography and Ocean Sciences for submesoscale vertical motion at oceanic fronts. Ocean 107, 75 pp. Modell., 14, 241–256. Takahashi, M., D. L. Seibert, and W. H. Thomas, 1977: Occasional Marlowe, C. J., and C. B. Miller, 1975: Patterns of vertical distri- blooms of phytoplankton during summer in Saanich Inlet, BC, bution and migration of zooplankton at Ocean Station ‘‘P.’’ Canada. Deep-Sea Res., 24, 775–780. Limnol. Oceanogr., 20, 824–844. Tennekes, H., and J. L. Lumley, 1972: A First Course in Turbulence. McGillicuddy, D. J., Jr., and A. R. Robinson, 1997: Eddy-induced MIT Press, 300 pp. nutrient supply and new production in the Sargasso Sea. Deep- Thiffeault, J.-L., and S. Childress, 2010: Stirring by swimming Sea Res. I, 44, 1427–1450. bodies. Phys. Lett., 374A, 3487–3490, doi:10.1016/j.physleta. ——, and Coauthors, 1998: Influence of mesoscale eddies on new 2010.06.043. production in the Sargasso Sea. Nature, 394, 263–266. Thorpe, S. A., 2005: The Turbulent Ocean. Cambridge University ——, R. Johnson, D. A. Siegel, A. F. Michaels, N. R. Bates, and Press, 439 pp. A. H. Knap, 1999: Mesoscale variations of biogeochemical Timothy, D. A., and M. Y. S. Soon, 2001: Primary production and properties in the Sargasso Sea. J. Geophys. Res., 104, 13 381– deep-water oxygen content of two British Columbian fjords. 13 394. Mar. Chem., 73, 36–51. Miller, C. B., B. W. Frost, H. P. Batchelder, M. J. Clemons, and Torres, J. J., 1984: Relationship of oxygen consumption to swim- R. E. Conway, 1984: Life histories of large, grazing copepods ming speed in Euphausia pacifica II: Drag, efficiency and in a subarctic ocean gyre: Neocalanus plumchrus, Neocalanus a comparison with other swimming organisms. Mar. Biol., 78, cristatus, and Eucalanus bungii in the Northeast Pacific. Prog. 231–237. Oceanogr., 13, 201–243. Trevorrow, M. V., 2005: The use of moored inverted echo sounders Moum, J. N., and T. R. Osborn, 1986: Mixing in the main ther- for monitoring mesozooplankton and fish near the ocean mocline. J. Phys. Oceanogr., 16, 1250–1259. surface. Can. J. Fish. Aquat. Sci., 62, 1004–1018. Munk, W., 1966: Abyssal recipes. Deep-Sea Res., 13, 707–730. ——, D. L. Mackas, and M. C. Benfield, 2005: Comparison of multi- ——, and C. Wunsch, 1998: Abyssal recipes II: Energetics of tidal frequency acoustic and in situ measurements of zooplankton and wind mixing. Deep-Sea Res. I, 45, 1977–2010. abundances in Knight Inlet, British Columbia. J. Acoust. Soc. Nasmyth, P., 1970: Oceanic turbulence. Ph.D. Thesis, University of Amer., 117, 3547–3588. British Columbia, 69 pp. Ueno, H., E. Oka, T. Suga, H. Onishi, and D. Roemmich, 2007: Nicol, S., 1986: Shape, size and density of daytime surface swarms Formation and variation of temperature inversions in the of the euphausiid Meganyctiphanes norvegica in the Bay of eastern subarctic North Pacific. Geophys. Res. Lett., 34, Fundy. J. Plankton Res., 8, 29–39. L05603, doi:10.1029/2006GL028715.

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC SEPTEMBER 2010 R O U S S E A U E T A L . 2121

Weber, L. H., S. Z. El-Sayed, and I. Hampton, 1986: The variance Yen, J., 2000: Life in transition: Balancing inertial and viscous spectra of phytoplankton, krill and water temperature in the forces by planktonic copepods. Biol. Bull., 198, 213–224. Antarctic Ocean south of Africa. Deep-Sea Res., 33, 1327– ——, J. Brown, and D. R. Webster, 2003: Analysis of the flow field 1343. of the krill Euphausia pacifica. Mar. Freshwater Behav. Wiese, K., and Y. Ebina, 1995: Propulsion jet of Euphausia superba. Physiol., 36, 307–319. J. Mar. Biol. Assoc. U. K., 75, 43–54. Zhou, M., and R. D. Dorland, 2004: Aggregation and vertical mi- Wunsch, C., and R. Ferrari, 2004: Vertical missing, energy and the gration behavior of Euphausia superba. Deep-Sea Res. II, 51, general circulation of the oceans. Annu. Rev. Fluid Mech., 36, 2119–2137. 281–314. ——, W. Nordhausen, and M. Huntley, 1994: ADCP measurements Yamazaki, H., and R. Lueck, 1990: Why oceanic dissipation rates of the distribution and abundance of euphausiids near the are not lognormal. J. Phys. Oceanogr., 20, 1907–1918. Antarctic Peninsula in winter. Deep-Sea Res. I, 41, 1325–1445.

Unauthenticated | Downloaded 09/28/21 11:28 AM UTC