Scholarly article on topic 'In situ observations of a doliolid bloom in a warm water filament using a video plankton recorder: Bloom development, fate, and effect on biogeochemical cycles and planktonic food webs'

In situ observations of a doliolid bloom in a warm water filament using a video plankton recorder: Bloom development, fate, and effect on biogeochemical cycles and planktonic food webs Academic research paper on "Biological sciences"

0
0
Share paper
Academic journal
Limnology and Oceanography
OECD Field of science
Keywords
{""}

Academic research paper on topic "In situ observations of a doliolid bloom in a warm water filament using a video plankton recorder: Bloom development, fate, and effect on biogeochemical cycles and planktonic food webs"

LIMNOLOGY

OCEANOGRAPHY

Limnol. Oceanogr. 60, 2015, 1763-1780 © 2015 The Authors Limnology and Oceanography published by Wiley Periodicals, Inc.

on behalf of Association for the Sciences of Limnology and Oceanography

doi: 10.1002/lno.10133

In situ observations of a doliolid bloom in a warm water filament using a video plankton recorder: Bloom development, fate, and effect on biogeochemical cycles and planktonic food webs

Kazutaka Takahashi,*1 Tadafumi Ichikawa,2 Chika Fukugama,1 Misaki Yamane,1 Shigeho Kakehi,3 Yuji Okazaki,3 Hiroshi Kubota,2 Ken Furuya1

department of Aquatic Bioscience, Graduate School of Agricultural and Life Sciences, The University of Tokyo, Tokyo, Japan

2Research Center for Fisheries Oceanography and Marine Ecosystem, National Fisheries Research Institute, Fukuura, Kanazawa-ku, Yokohama, Japan

3Fisheries Management and Oceanography Division, Tohoku National Fisheries Research Institute, Shiogama, Japan

I Abstract

We investigated distribution patterns of a doliolid (Dolioletta gegenbauri) bloom in relation to the physical environment using a video plankton recorder in the Oyashio-Kuroshio mixed water region. Using 12 km transects, doliolid blooms were encountered at a horizontal scale of about 2-3 km, which corresponds to submeso-scale physical events. Doliolids were also consistently encountered in the subsurface layer above the pycnocline in warmer (> 14°C) and higher-salinity (> 34) water masses, and seawater density was the most critical factor affecting distribution depth. Compared to previous studies, the density and biomass of the blooms observed in this study (77 mgC m"3 and 4600 inds m"3) were highest in the open ocean. Bloom formation consisted of two phases;first, the seeding population of a nurse stage increased rapidly to 2000 inds m"3 by asexual reproduction, followed by asexual production of phorozooids. Estimated population clearance rates revealed that these dense patches could potentially sweep the surrounding water within 2-3 d. The incidence of exhausted and shrunken zooids was significantly correlated with patch density, suggesting that mortality was due to overgrazing. Shrunken doliolids appeared to sink below the pycnocline, corresponding to 8-17% of the particulate organic carbon flux at 150 m. Hydromedusae, pelagic polycheates, and sapphirinid copepods preyed on the doliolids. These results indicate that doliolids, which were seeded by populations originating from the Kur-oshio, formed dense blooms in response to submesoscale physical events and would alter the sinking particle properties (i.e., biological pump) and the epipelagic food web structure through their grazing and mortality.

An important goal of marine ecology studies is to understand the distribution patterns of marine organisms and the structure and functioning of pelagic ecosystems, as doing so can be used to predict how environmental changes and human activities will impact the distribution patterns of marine fauna and the biology of the ocean (Kiorboe 2011). Within this context, the ephemeral, patchy, and sporadic nature of doliolid blooms makes these events very difficult to investigate in any detail (Deibel and Paffenhofer 2009).

*Correspondence: akazutak@mail.ecc.u-tokyo.ac.jp

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

Additional Supporting Information may be found in the online version of this article.

Doliolids are pelagic tunicates with a complex life cycle that includes polymorphic asexual and sexual reproductive stages (Paffenhofer and Koster 2011);having both stages means that doliolids are capable of responding to favorable conditions by producing offspring quickly (Deibel 1998;Deibel and Paffenhofer 2009). As doliolids are nonselective filter feeders that feed on particulate organic matter ranging in size from bacteria to copepod eggs (Crocker et al. 1991;Paffenhofer et al. 1995), they are considered to have a marked effect on plankton community structure and biogeochemical cycles in the area where their blooms occur (Deibel 1985;Paffenhofer et al. 1995).

Despite their potential significance in pelagic ecosystems, relatively little is known about the factors that cause doliolid blooms and the fate of the blooms. Given the sudden appearance and disappearance of blooms by asexual reproduction, doliolids are considered to be adapted to event-

Fig. 1. Satellite images of (A, C) surface-water temperature and (B, D) Chl a concentrations around sampling area for late May 2009 (8-d composite) derived from MODIS Aqua data (Ocean Watch, NOAA). Transect lines for VPR observations are also shown.

scale rather than seasonal-scale changes in environmental conditions (Deibel and Lowen 2012). For example, in the South Atlantic Bight (SAB) off the coast of Florida, U.S.A., high rates of phytoplankton production in response to upwelling are a primary condition necessary for the occurrence of doliolid blooms along the continental shelf (Deibel 1985;Paffenhofer et al. 1995). Deibel and Paffenhofer (2009) concluded that thaliacean blooms (doliolids and salps) require a broad, shallow continental shelf, a strong boundary current with eddies and meanders, and along-shelf winds that promote upwelling. However, as patches of thaliaceans also occur in the open ocean (Tsuda and Nemoto 1992; Takahashi unpubl), further investigations on the patch dynamics of thaliacean blooms in relation to the physical oceanographic conditions are necessary. In particular, knowledge of the factors affecting the termination process and fate of the bloom is crucial to our understanding of the role of doliolid blooms in marine biogeochemical cycles.

In addition to their sporadic occurrence, the fragile gelatinous bodies of doliolids make them difficult to collect using nets, further complicating investigations on the ecology of these organisms. In particular, information on the nurse stage, which is critical for asexual reproduction and consequently for causing blooms, is very difficult to obtain as the dorsal cadophore is typically lost when sampling is performed using nets. In addition, the resolution of net sam-

pling is often not sufficient to reveal the vertical and horizontal scales of doliolid patchiness. To overcome these problems, nonintrusive and consecutive optical sampling techniques are considered preferable. Paffenhofer et al. (1991) observed the vertical profiles of doliolids by video filming from a manned submersible. More recently, the three-dimensional mesoscale structural characteristics of a salp swarm (Thalia democratica) was captured using an optical plankton recorder (Everett et al. 2011). Although these studies successfully clarified some of the factors affecting the distribution of thaliacean patches, the resolution of their analyzes was not sufficient for relating the patch dynamics of the patches to environmental factors. We, therefore, attempted to accurately clarify the distribution patterns of doliolid blooms in relation to the hydrographical conditions in the Oyashio-Kuroshio mixed water region in the Northeastern Pacific using a video plankton recorder (VPR). In doing so, we sought to infer the processes underlying the development, fate, and effect of the doliolid bloom on the biogeochemical cycle and planktonic food web.

Methods

Field research

This study was conducted onboard the R. V. Soyo-Maru of the Fisheries Research Agency from 29 May 2009 to 2

June 2009 in the Oyashio-Kuroshio mixed water region in the Northwestern Pacific Ocean (Fig. 1). We deployed a VPR (Color Auto-VPR, SeaScan), which captures images using a charge-coupled device camera at a resolution of 1024 X 1024 pixels and 15 frames per second along two transect lines across a warm water filament during daytime (07: 36-14: 53 h). The field of view of the image was calibrated to 37.9 X 37.9 X 89.0 mm (height X width X depth, 37.04 im pixel"1), giving an image volume of 127.8 mL and objects within this field were confirmed to be captured as in-focus images (Ichikawa 2008). The VPR was towed from the starboard side of the vessel at a speed of 2-3 knots and continuously lowered and raised 16-18 times between the surface and a depth of 50 m along the two 12 km- to 14 km-long transects. Cable release and heaving speed were approximately 0.2 m s"1. The VPR system also included a CTD sensor (MCTD, Falmouth Scientific), which logged temperature and salinity data. Chlorophyll a (Chl a) concentration was measured with calibrated fluorescence sensor (COMPACT-CLW, JFE Advantech, Ltd.) attached to the frame of the VPR. At both ends of the transects, mesozooplankton sampling was conducted using a Vertical Multiple Plankton Sampler net (50 cm X 50 cm opening, 100 im mesh, Terazaki and Tomatsu 1997) from a depth of 50 m to the sea surface to determine the abundance and species composition of the doliolid community.

VPR data analysis

All in-focus images were extracted as regions of interest (ROI) using the Auto VPRdeck software (Seascan) and saved to disk as TIFF files. All ROIs were sorted manually and the different stages of doliolids were identified and counted. Selected images were merged with CTD data using a time stamp to analyze the effect of environmental factors on doliolid distribution, and hydrographic and distribution data of doliolids were plotted for analysis using Ocean Data View software (ver. 4.5.1, http://odv.awi.de). The length of the doliolids was measured directly in the extracted images using the image processing program, ImageJ (National Institutes of Health, rsb.info.nih.gov/ij/). The body lengths (BL) of doliol-ids in images taken from various angles (excluding the lateral side) were estimated based on the allometric relationship between BL (in mm) and the diameter of the oral aperture ring (DOR in mm) of doliolids, both of which were determined using images taken from a lateral view of animals before general analysis. The following equations were used to estimate BL in the different doliolid lifecycle stages:

Gonozooids : BL = 3.0455Dor + 0.9857,R2 = 0.72892 (n = 242)

Phorozooids : BL = 2.7406Dor + 1.1363,R2 = 0.70892 (n = 113)

The equation used for gonozooids (Eq. 1) was also applied to oozooids and unidentified solitary zooids when their images were not taken from lateral side. The number of trophozooids on the dorsal cadophore of nurses (NT) was determined based on the length of the cadophore (LC in mm) using the following equation (Eq. 3) applied to in-focus images of nurses that were captured along the two observation transects:

NT = 2.2525LC8966, R2 = 0.75976 (n = 118) (3)

During the manual sorting of the images, potential doliolid predators were also identified.

Biomass and potential clearance capacity

Using the length frequency data of doliolids, we estimated the carbon biomass and potential clearance capacity that could attributed to the patches of doliolid populations along the transects (see Supporting Information Appendix 1). Biomass estimates of the doliolid populations were calculated based on length (BL in mm)-weight (Cw in igC) relationship using the following equation (Eq. 4) described by Gibson and Paffenhofer (2000):

Cw = 0.4643 BL2 3119 (4)

Although the equation was developed to analyze gono-zooids, we applied the equation to all of the stages observed in this study, including oozooids, trophozooids, and phoro-zooids due to their morphological similarity. The mean BL of trophozooids was taken as 1.5 mm based on the direct measurements of VPR images.

The length (BL in mm)-clearance rate (CR in mL h"1 ind"1) regressions first described by Crocker et al. (1991) were used to estimate the clearance rates of individual adult zooids in the doliolid patches. The equations used for calculations are as follows:

Gonozooids : Log CR = 1:34 log BL"0:73 (5)

Phorozooids : Log CR = 1.77 log BL-1.53 (6)

The equation used for gonozooids (Eq. 5) was also applied to oozooids and unidentified solitary zooids. The clearance rates of polymorphic colonial stages were only estimated for nurse colonies by multiplying the number of trophozooids derived from the above-mentioned allometry equation (Eq. 3) by the mean clearance rate of the trophozooids (2.5 mL zooid-1 h-1, Tebeau and Madin 1994).

Incidence of exhausted individuals

It has been suggested that overgrazing by doliolid blooms is like responsible for the termination of the bloom itself (Deibel and Paffenhofer 2009). Based on this assumption, we

hypothesized that the occurrence of exhausted individuals is a density-dependent characteristic of doliolid populations. To confirm this hypothesis, we compared the proportion of exhausted individuals in a given patch (see Supporting Information Appendix 1). The ratio of the width to length (shrink ratio [SR]) of doliolid zooid in the VPR images was then used as an index of individual vitality, as we noticed that doliol-ids shrink in size and appear thinner when they are exhausted. We conducted onboard observation using recently collected doliolids during the cruise to observe the morphological changes associated with starvation and exhaustion in doliolids. Doliolids were collected by vertical tow from a depth of 50 m to surface with a conical plankton net (diameter: 80 cm, mesh: 100 im) fitted with a 3L nonfiltering cod end. Ten intact gonozooids harvested in this way were then gently transferred to acid-washed, six-well tissue culture plates using a wide bore pipette, with each well (diameter: 3.5 cm, depth: 2 cm) containing about 10 mL of filtered seawater. The wells were then incubated at sea surface temperature (ca., 16°C) in the dark. All individuals were inspected for vitality under dissecting microscope at 4-h intervals and were photographed with a digital camera attached to the microscope when they died to determine the mean SR of dead individuals.

Results

Oceanographic conditions

Two transects were set across a warm water filament originating from the Kuroshio Extension (Fig. 1). After identifying an area with high Chl a concentration (Chl a) along the northern edge of the warm water filament, a VPR tow (Transect 1) was undertaken across the front. A vertical TS profile revealed that the existence of a cold (ca., 14°C), low salinity (<34.2) water mass from the Oyashio region extending over the western half of Transect 1, forming pronounced front with warm (> 15°C) saline (> 34.2) water in the middle of the transect (Fig. 2). The pycnocline around frontal area was located at 35-40 m, extending to 40-50 m toward both ends of Transect 1 (Fig. 2). Generally, Chl a concentrations along Transect 1 were high above the pyc-nocline, and a particularly high Chl a concentration (> 1 ig L"1) was observed in the subsurface layer around the frontal area as well as in the area of the cold water mass (Fig. 2G). Transect 2 was set approximately 40 km to the east of Transect 1 (Fig. 1); judging by the relatively higher salinity (> 34.4) and lower temperature (<14.5°C) observed along this transect, the transect appeared to be located in the warm water mass that formerly associated with the Kur-oshio Extension. A vertical TS profile of the Transect 2 revealed that the physical environment of this area was relatively homogenous and that the pycnocline was consistently located at a depth of 35 m to 40 m (Fig. 2B,D,F). However, as in the Transect 1, the Chl a concentration

along Transect 2 was generally high immediately above the pycnocline, while the pronounced increase of high Chl a concentration at the surface was present in the middle of the Transect 2 (Fig. 1H).

Doliolids in A-VPR images and in net samples

In total, 11,127 and 14,223 doliolid individuals were enumerated in the VPR images captured along Transects 1 and 2, respectively. Nurse stages (old oozooids) were easily distinguished from other stages due to the presence of elongated dorsal cadophores and broad muscle bands (Fig. 3A). With the exception of the nurses, the lifecycle stages of some individuals could be assigned with confidence when they were suitably oriented and in focus (Fig. 3B-D);how-ever, some of the solitary zooids could not be identified due to the insufficient resolution of images. No eggs or larvae were observed in the images, probably because they were too small to be captured by the VPR. Examination of the net samples revealed that Dolioletta gegenbauri accounted for 97100% of the doliolid populations (solitary zooids) and remains of Doliolum denticulatum and Doliolum nationalis were also identified.

Distribution of doliolids along transect lines

Numerous high-density doliolid patches were observed along both transects (Fig. 4). On Transect 1, a doliolid patch with a scale of about 2 km was observed semi-continuously for about 4 km across the front between the cold water mass and warm water filament (Fig. 4A). This patch was strongly associated with the water mass, which was characterized by having a high Chl a concentration and a vertically homogenous seawater density (sigma-t) extending from above the pycnocline to the subsurface (Fig. 4E,G). On Transect 2, a doliolid patch with a scale of about 3 km was observed at the middle of the transect, with smaller patches observed before and after this main patch at intervals of about 4 km (Fig. 4B). No structural characteristics in terms of temperature and salinity characteristics were apparent on Transect 2 (Fig. 2B,D,E); however, close examination of the seawater density profiles revealed isopycnal outcrops from a depth of about 20 m, slightly above the pycnocline, into the surface water around the patches of doliolids (Fig. 4F).

Generally, doliolids occurred in the subsurface layer, and the lower limit of the distribution appeared to correspond with pycnocline depth (Fig. 4). Peaks in doliolid distribution were observed at approximately 25 m and 17 m on Transect 1 and 2, respectively (Fig. 5A,B). In terms of physical factors, doliolids were abundant in areas with relatively high temperatures (> 14°C) and salinities (> 34.2), with peak doliolid densities observed in water masses with a seawater density (sigma-t) of approximately 25.5 (Fig. 5). Although the distribution patterns of nurse stages were generally similar to those of the other stages, the centers of their occurrence did not always coincide with those of the other stages (Fig. 4). For instance, the highest densities of nurses encountered at the

Transect 1 Transect 2

146,95°E 147°E 147.05°E 147.TE 147.45°E 147,S°E 147.55°E I47.6°E

]46.95°E 147°E I47.05°E I47.PE I47.45°E 147.5°E I47.5S°E 147.6"E

147°E 147.05°E !47.]°E I47.45°E 147.S"E I47.55°E 147.6SE

Longitude

Fig. 2. Environmental and physicochemical parameters examined along Transect 1 (left column) and 2 (right column).Vertical profiles of (A, B) temperature, (C, D) salinity, (E, F) seawater density (sigma-t), and (G, H) Chl a concentration. Dashed lines in E, F, G, H indicate the position of the pyc-nocline. Aligned small dots indicate actual sampling points data obtained.

westernmost end of Transect 2 where total doliolid densities were moderate (Fig. 4 C,D). The highest numbers of nurses were observed at 21-30 m, which was 4-5 m deeper than the depths at which the other solitary zooids were observed along the transects (Fig. 5A,B). In addition, compared to the other zooids, these peaks in nurse densities tended to occur slightly warmer and more saline water masses (Fig. 5C-F).

To clarify the processes underling the development of doliolid blooms, the relationships between the population structure and abundance were analyzed for different patches (Table 1, see also Supporting Information Appendix 1). The

density of doliolids in the patches considered for analysis ranged from 120 inds m"3 to 4654 inds m"3, which was significantly related to the proportion of nurse stages within each population (R2 = 0.44, n = 14, p < 0.05, Fig. 6A). However, nurse density increased together with total density up to about 2000 inds m"3, before decreasing thereafter (Fig. 6B). This tendency was more apparent in the relationship between total doliolid density and the mean length of the nurse zooid and its chain (dorsal cadophore length);both of these nurse-size parameters increased with total doliolid population density up to about 2000 inds m"3, before decreasing

» gf > C V

> A •

F ^F I D

— - -

Fig. 3. In situ photographs of doliolids taken by the VPR at sampling stations in the Oyashio-Kuroshio mixed water region. Scale bar = 1 mm, (A) nurse with dorsal cadophore, 24.2 m, 12: 21 h, 29 May. (B) Phorozooid, 29.5 m, 08: 04 h, 29 May. (C) Gonozooid, 27.5 m, 08: 30 h, 29 May. (D) Oozooid, 15.6 m, 08: 12 h, 29 May.

markedly to approximately half of their maximum length (Fig. 6C,D). Conversely, the mean BL of the other solitary zooids (mostly gonozooids) was positively related to the total density of all doliolids (Fig. 6C).

Nurse density increased together with proportion of nurse stages within each population up to about30%, while density of nurse decreased when they accounted for more than 31% of the population (Fig. 7A). The mean length of the nurse zooid showed significant relationships with its chain length (R2 = 0.85, n = 14, p < 0.05, Fig. 7B);both of these size parameters appeared to be related to proportion of nurse stages in the patches. For example, nurse (zooid and chain)

length increased until the proportion of nurses in the patch reached about 30% but nurses became considerably smaller when they accounted for more than 31% of the population (Fig. 7C,D).

The biomass of these doliolid patches ranged between 1.1 IgC L"1 and 77.0 igC L"1 (Table 1), and the estimated population clearance rates for a given patch varied from 0.9 L h"1 m"3 to 20.3 L h"1 m"3 (Table 1 and Supporting Information Appendix 1). The water volume swept by each doliolid population corresponded to 2-49% of their surrounding water per day, assuming that filtering activity was continuous. The trophozooids on the dorsal cadophore of nurses are

Fig. 4. Vertical profiles of doliolid distribution and related environmental variables along the Transect 1 (left column) and 2 (right column). Distribution of (A, B) total doliolids and (C, D) nurse stages (old oozooids). Dashed lines in A, B, C, D indicate the position of the pycnocline. Vertical profiles of (E, F) seawater density (sigma-t) with fine scale in the vicinity of patches and (G, H) Chl a concentration. Location of doliolid patches (> 500 ind m"3) indicated by white lines in E, F, G, and H.

considered to contribute considerably to population clearance rates, and can account for 14-96% of total population clearance rates (Table 1)

Incidence of exhausted individuals and flux by dead doliolids

Shipboard observations revealed that doliolids in filtered seawater shrunk with time, and all of them died within 12 h after capture (Fig. 8A,B). The mean SR (body width/length) of dead doliolids was 0.315 ± 0.05 (n = 10). In the field, individuals with SR of <0.4 were also commonly encountered (Fig. 8D,E), accounting for 9-30% of all solitary zooids in the various patches. Proportion of shrunken individuals in a patch was positively related to the density of total zooids

and negatively related to the percentage of nurses in a patch (Fig. 9). The muscle bands of the shrunken zooids in the VPR images were characteristically opaque (Fig. 8) when compared with healthy individuals with high SR (Fig. 3). Some of the individuals with lower SR were deformed or bent and appeared exhausted (Fig. 8F)

On the whole, shrunken doliolids (SR < 0.4) in the population below pycnocline (> 40 m) were relatively more abundant (20-49%) than those in patches above the pycnocline (Table 2). Assuming that the exhausted doliolids below the pycnocline eventually died and sank as detritus, the potential flux associated with the exhausted doliolids was estimated to be 16-29 mgC m"2 d"1, corresponding to 8-17% of the POC flux at a depth of 150 m (Table 2).

Fig. 5. Frequency histograms showing the occurrence of doliolids relative to various environmental factors along profile in surface layer (0-50 m); (A, B) depth, (C, D) temperature, (E, F) salinity, and (G, H) seawater density (sigma-t) along Transect 1 (left column) and 2 (right column). Proportion of the nurse stage in each patch is indicated by solid bars.

Table 1. Environmental variables and characteristics of population structure and estimated clearance rate of separate doliolids' patches. Patch ID and position along the transcends are given in Supporting Information Appendix 1.

Population Total

Mean clearance Population clearance

Mean Biomass body Mean Mean rates of clearance rate of Potential

Chi a of total Density Density %of length of body length solitary rates of doliolid cleacance

Depth Mean conc. doliolids of total of nurse nurse solitary length of nurses' zooids trophozooids population of resident

Patch range temperature Mean (ra (/jgC doliolids (ind in the zooids of nurse cadophore (mL h"1 (mL h"1 (mL h"1 water

Transect ID (m) (°C) salinity L-1) L-1) (ind m~3) m"3) population (mm) (mm) (mm) m"3) m"3) m"3) (% d"1)

1 1A 12-23 14.58 34.07 0.43 10.7 1111 23 2.1 3.7 3.4 2.96 1426 30 1456 3

1 1B 2-30 15.02 34.20 0.75 33.2 2261 244 10.8 4.4 3.5 5.44 2793 7680 10473 25

1 1C 2-34 15.47 34.34 0.48 45.8 2807 349 12.4 4.7 3.0 4.84 3688 8863 12551 30

1 1 D 2-20 15.63 34.37 0.06 1.1 131 48 37.0 2.8 2.6 4.26 50 340 390 1

1 1 E 21-31 15.53 34.37 0.72 6.3 555 192 34.5 3.6 3.2 4.45 357 573 930 2

1 1 F 2-30 15.45 34.36 0.73 2.1 179 76 42.3 3.3 3.0 5.55 76 800 876 2

2 2A 15-31 14.42 34.43 0.30 48.0 1864 575 30.9 4.3 5.9 11.56 1583 13747 15331 37

2 2B 10-31 14.34 34.43 0.53 38.1 1956 390 20.0 4.3 5.3 10.74 1808 13550 15358 37

2 2C 15-26 14.44 34.42 0.48 27.0 1359 340 25.1 4.4 4.6 10.24 1203 5003 6205 15

2 2D 2-27 14.46 34.40 0.58 77.0 4654 156 3.4 4.7 3.7 6.22 5466 3983 9448 23

2 2E 1-27 14.46 34.29 0.83 58.1 3926 284 7.2 4.3 3.9 7.53 4180 10428 14607 35

2 2F 5-21 14.47 34.41 0.37 27.5 1692 235 13.9 4.2 4.8 9.01 1546 2693 4239 10

2 2G 5-21 14.45 34.41 0.56 6.0 404 78 19.3 3.9 4.5 7.27 273 526 798 2

2 2H 5-21 14.41 34.41 0.63 21.0 1270 114 8.9 4.5 4.4 7.28 1435 1746 3180 8

2000 3000 4000 5000

1000 2000 3000 4000 5000

Density of doliolids (ind m~3)

Fig. 6. Relationships between total doliolid density and nurse population parameters in different patches; (A) proportion of nurses in a patch (%), (B) nurse density, (C) mean length of zooids, and (D) mean length of nurse chain (dorsal cadophore length). Linear regressions (solid lines) in (A) indicate that the relationship is significant (n = 14, R2 = 0.44, p < 0.05). Relationship between total doliolid density and mean length of solitary zooids is also shown in (D). Each data point refers to a defined patch indicated in Supporting Information Appendix 1.

Potential doliolid predators

Some of the captured images showed various planktonic organisms associating with doliolids, indicating that they were potential predators of the doliolids. For example, the jellyfishes Phialella fragilis and Solmundella bitentaculata were observed with doliolids in their tentacles (Fig. 10A,B). Density of S. bitentaculata was positively related with that of nurse stage along Transect 1 (data not shown). Some images showed a pelagic polycheate, Tomopteris sp., apparently attacking and biting opaque objects (Fig. 10C,D). The presence of pigmented guts and muscle bands in the opaque objects suggests that potential prey items were shrunken doliolids. As reported previously by Takahashi et al. (2013), sapphirinid copepods were the most frequently encountered

organisms associated with the doliolids. In particular, Sap-phirina nigromaculata was often found on the chains of nurses as well as on or within the solitary zooids (Fig.10E,F). Mean densities of hydromedusae, polychaetes and sapphirinid copeods upper 50 m in net-samples made at both ends of transects were 135 inds m"3, 53 inds m"3, and 7 inds m"3, respectively.

Discussion

The VPR observations successfully detected doliolid blooms in the Oyashio-Kuroshio mixed water region of the Northwestern Pacific Ocean. Dense patches with densities

of >2000 inds

consistently encountered at

(D !»

% of nurses in a patch

<D izj

10 20 30 40

% of nurses in a patch

cd ¿S

"cO <D

Cl tw O

¡50 fl u

Mean length of nurse zooid (mm)

0 10 20 30 40

% of nurses in a patch

Fig. 7. Relationship between various patch properties of nurses and their proportion or size in various patches. (A) nurse stage density vs. proportion of nurses in a patch (%). (B) Mean length of nurse chain vs. mean length of nurse zooid, (C) mean length of nurse zooid vs. proportion of nurses in a patch, (D) mean length of nurses, chain vs. proportion of nurses in a patch. Liner regressions (solid lines) in (B) indicate that the relationship is significant (n = 14, R2 = 0.85, p < 0.05). Each data point refers to a defined patch indicated in Supporting Information Appendix 1.

horizontal scales of about 2-3 km along two 12 km-long transects. These dense patches corresponded to areas with high Chl a concentrations, which were in turn associated with structures consisting of vertically homogenous seawater densities that resembled submesoscale upwelling or water mixing events. The relationships between these physical factors and the incidence of doliolid (D. gegenbauri) blooms have been extensively examined in the SAB off the coast of Florida, Georgia and the Carolinas in the U.S.A. (Deibel and Paffenhofer 2009). Doliolid patches in the SAB are associated with intrusions of cool, high-nutrient bearing water from aphotic depths, which are in turn associated with increased phytoplankton productivity (Deibel and Paffenhofer 2009). As these intrusions are shelf-break upwellings that have spun

off from cyclonic meanders and eddies of the western wall of the Gulf Stream, the dimensions of the doliolid patches can approach those of stranded intrusions themselves, that is, > 100 km in the along-shelf direction and up to 60-80 km in the cross-shelf direction (Deibel and Paffenhofer 2009). The full scale of the doliolid patches in the Oyashio-Kur-oshio mixed water region described in this study are considered to be smaller than those of the mesoscale patches in the SAB. This was partly due to the difference of sampling design;for example, VPR observations are useful for examining distribution patterns at subkilometer scales, whereas net sampling is effective for assessing distribution at a scale of several tens of kilometers. However, the physical mechanisms that enhance primary production at the study site,

Fig. 8. Images of shrunken doliolids at various stages of exhaustion. Scale bar = 1 mm. Photomicrographs of (A) healthy normal gonozooid and (B) dead shrunken gonozooid. (C, D, and E) in situ images of doliolid zooids at various stages of exhaustion taken by the VPR. SR = body width/BL. (F) VPR images of doliolids that have shrunken and become deformed in situ, possibly due to exhaustion. Time and depth measurements taken are also shown.

which is deeper than 5400 m, are very different from those at the SAB, which is shallower than 40 m and which experiences shelf-break upwelling frequently. In this respect, the

occurrence of dense doliolid blooms in the offshore area surveyed in this study is somewhat unusual as most doliolid blooms have been observed in coastal and neritic areas

(Deibel and Paffenhofer 2009). Furthermore, the findings reported here imply the existence of unique mechanisms in the Oyashio-Kuroshio mixed water region that regulate biological production.

The physical events associated with the doliolid blooms in the Oyashio-Kuroshio mixed water region in this study can be divided into two types: the first is the existence of a vertically homogenous structure above the pycnocline at the edge of the warm water filament on Transect 1 (Fig. 4E). Although the detailed mechanism has not yet been resolved, this structure might correspond to the submesoscale increase in mixing at frontal area, which would increase the upward

-'-□--y = 0.16339 + 2.6663 x 10"5x R2= 0.42245 ""■"y = 0.03883 + 1.8469 x 10"5x R2= 0.34437

/ • • • •

• <0.5 --□--<0.4 -■-<0.3

D B-nr"

№ 0 1000 2000 3000 4000 5000 Density of doliolids (ind m~3)

Fig. 9. Relationship between the proportion of doliolids at various stages of exhaustion and total doliolid density. Liner regressions (solid lines) in the figures indicate that the relationship is significant (p < 0.05). Each data point refers to a defined patch indicated in Supporting Information Appendix 1.

flux of nitrates and resulting in an increase of Chl a (Fig. 4G) (Johnston et al. 2011;Kaneko et al. 2013). Conversely, no structural features, such as frontal areas, were observed along Transect 2, which was instead characterized by relatively homogenous water masses (Fig. 2B,D,F). However, examination of the fine scale sigma-t profile of Transect 2 revealed the possible existence of isopycnal outcrops from a depth of about 20 m, slightly above the pycnocline, into the surface waters areas where doliolid patches were observed (Fig. 4F). These structures are considered to resemble local upwelling caused by the Langmuir circulation (Barstow 1983), although the horizontal scale of the circulation cells (several km) associated with the outcrops were unusually larger than generally known (Thorpe 2004). Thus, the upwelling-like structures observed along Transect 2 are considered to be regulated by as yet unknown submesoscale mechanisms that warrant investigation in the future (H. Kaneko, personal communication).

The doliolid patches observed in this study were generally observed in the subsurface layers above the pycnocline in the warmer (> 14°C) and higher salinity (> 34) water masses, suggesting that their seed populations originated in the Kuroshio and Kuroshio Extension. Among the physical variables examined, seawater density is likely to be the most critical factor affecting the distribution depth of the doliolids (Fig. 5). This would agree with the distribution patterns observed in the SAB where peaks in doliolid abundance are frequently observed in the pycnocline or thermocline (Deibel 1985;Paffenhofer et al. 1991). As in the SAB, nurses were distributed in slightly deeper layers than the other solitary zooid stages (Paffenhofer et al. 1991, 1995). Unlike the solitary zooids that do not swim actively, nurses show active swimming behavior under conditions of food scarcity and high predation pressure (Paffenhofer and Koster 2011). Consequently, the distribution patterns of nurses are considered to reflect food availability and risk of predation. In addition, their close proximity to the pycnocline means that they can respond quickly to any phytoplankton blooms associated

Table 2. Density, biomass possible contribution of shrunken doliolids of which shrink ratio (SR) < 0.4 to vertical flux below the pyc-nocline. Location of sampling sites D1-D4 are given in Supporting Information Appendix 1.

Density of total Density of shrunk Biomass of Potential flux by % of the doliolids

Surveyed Depth doliolids* doliolids* shrunk doliolids exhausted doliolids* flux to POM

area (m) (ind m"3) (ind m23) (igC m"3) (mgC m"2 d"1) flux§ (d"1)

D1 40-60 12.0 2.6 72.0 29.3 17

D2 40-60 9.2 2.5 49.0 19.9 12

D3 40-60 9.8 2.0 32.5 13.2 8

D4 40-60 3.9 1.9 38.2 15.5 9

*Only solitary zooids were counted. individuals with SR < 0.4.

^Assuming sinking rate of 407m d_1 (Takahashi et al. 2013). §Calculated using value of POC flux at 150 m, 173.4 mgC mT2 d2

(Takahashi etal. 2013).

B ^ * § * JmL W nJL T F

f I jj /

u » 2 A -

1 • _ v \

Fig. 10. In situ photographs of potential doliolid predators taken using the VPR at sampling stations in the Oyashio-Kuroshio mixed Water Region. Scale bar = 1 mm. (A) hydromedusa of Phialella fragilis, 38.0 m, 08: 17 h, 29 May, (B) narcomedusa of S. bitentaculata, 24.2m, 12: 21, May 29, (C, D) pelagic polychaete, Tomopteris sp., 13.8-21.1 m, 14: 53 h, 29 May, (E) cyclopoid copepod, S. nigromaculata female in the tunic of a zooid, 19.1 m, 14: 57 h, 29 May, (F) S. nigromaculata female associated with nurse's chain, 29.2 m, 08: 39 h, 29 May.

with any upwelling events, resulting in enhanced asexual reproduction.

The maximum density and biomass of some of the doliolid patches observed in this study are among the highest levels reported to date, particularly in the open ocean

(Deibel 1985;Deibel and Paffenhofer 2009). The mean length of solitary zooids (primarily gonozooids, which were more abundant) was positively related to total density, suggesting that blooms were associated with an increase in the size of individuals (Fig. 6C). As phorozooid start to produce

gonozooid asexually at a length of 5 mm and gonozooid begin to release egg/sperm at a length of 6 mm (Paffenhofer and Gibson 1999), individual growth of solitary zooids would be an inevitable factor to form dense bloom of doliolid. The density of doliolids in all of patches was negatively related to the proportion of nurses (Fig. 6A). As abundance and mean nurse length were small in patches with a high proportion of nurses (> 31% of total proportion;Fig. 7A,C,D), we consider that these nurse-dominated patches correspond to the seeding populations of the doliolid bloom, with the high proportion of nurses reflecting the situation just before and after the bloom. Nurse density and size (zooid and dorsal cadophore length) both increased with total doliolid density up to about 2000 inds m"3, implying that the growth of nurse stages was associated with an increase in doliolid densities in the first phase of the bloom. Furthermore, the decrease in nurse density observed in those populations that had total densities > 2000 inds m"3 corresponded with a decrease in nurse sizes (Fig. 6B-D), indicating that the large nurses began to disappear from very dense patches and that further increases in total doliolid densities were largely attributed to asexual reproduction by phoro-zooids, which produce the gonozooids that are the dominant stage of intensive bloom (Deibel 1998). These data lend support to the importance attributed by Deibel and Paffenhofer (2009) to the critical role the nurse stage plays in patch formation. In addition, our findings also support the hypothesis that doliolid blooms are dependent on the asexual production of phorozooids that are produced by the cadophores of nurses (Deibel and Lowen 2012). The life history parameters of doliolids, that is, feeding, growth of gono-zooids, and fecundity of phorozooids vary under various prey concentrations and temperatures (Gibson and Paffenhofer 2000, 2002) and thus the extent to which the patches examined in this study would develop under a variety of possible environmental conditions is unknown. However, the findings presented here suggest that the process underlying doliolid bloom formation can be attributed to two different asexual stages; nurses (old oozooids) establish new populations when they encounter favorable environments (e.g., submesoscale upwelling), following the disappearance of the large nurses, the phorozooids drive rapid population growth by producing gonozooids, which results in intensive blooms. The polymorphic asexual phase of doliolids, which is unique among pelagic tunicates (Deibel and Lowen 2012), is very well suited for exploiting instances of highly localized and enhanced production, such as event-scale occurrences like submesoscale upwelling. Furthermore, this would be the beneficial for maximizing opportunity for sexual fertilization and the production of gonozooids, which fertilize the eggs internally as gonozooids cannot swim actively (Deibel and Lowen 2012).

Our estimates revealed that the dense patches of doliolids potentially sweep all of their surrounding water within a sev-

eral days (Table 2). As further support for our calculations, the location of doliolid patches with higher clearance rates (e.g., 2A, 2B, 2E, Table 2, also see Supporting Information Appendix 1) coincided closely areas that had low Chl a concentrations (Fig. 4H), although no significant relationship was found between Chl a concentration and patch density with entire data set (see Table 1) probably due to the difference in initial conditions in terms of environment variables and doliolid population structure. These estimates are comparable with those of doliolid clearance rates in the SAB, where doliolid populations have been estimated to clear 25120% of their resident water volume per day (Deibel 1985; Paffenhofer et al. 1995). However, in those studies, the contribution of the trophozooids on the nurses' chain was not considered in the estimates of clearance rates. Even though we used relatively lower estimates for individual trophozooid clearance rates (i.e., 2.5 mL zooid"1 d"1; Tebeau and Madin 1994), our estimates showed that the contribution of tropho-zooids to population clearance rates was considerable. Recent studies have shown that the clearance rates of trophozooids have been estimated to be > 100 mL zooid"1 d"1 in a large colony, which is comparable to the clearance rates of gono-zooids of the same size (Paffenhofer and Koster 2011). Consequently, the actual contribution of grazing by trophozooids may even exceed our estimates. However, we realized that the size of the trophozooids on the nurse chains at our study sites were relatively similar to each other (see Fig. 3A);in the SAB, trophozooids at the distal end of the cadophore were generally larger than those at the proximal end (Paffenhofer and Koster 2011). While it is not clear whether this variation could be attributed to genotypic or phenotypic differences, detailed information on the physiological characteristics of the nurse stages is required to better understand the development of doliolid blooms and the effect they have on the pelagic ecosystems in the study site. As doliolid blooms generally persist for more than a week (Deibel 1985;Paffenhofer 2013), it is likely that the doliolids in the study site eventually cleared almost all food particles in their immediate environment, as has been suggested in the SAB, while most of fecal pellet would be mineralized before export to depth as their fecal pellet sink very slowly (Patonai et al. 2011).

Shipboard observations revealed that the ratio of doliolid width to length (SR) decreased to 0.32 when they were exhausted and died, although the exact cause of death was unknown. The proportion of shrunken doliolids (SR < 0.4) in the patches along the transects were negatively related with the total solitary zooid density, suggesting that the cause of the shrunken doliolids in the dense patches were dying due to a shortage of food. While the fate of the shrunken doliol-ids with SR > 0.4 is uncertain, we believe that SR is appropriate for measuring the degree of the decline in solitary zooids, as individuals with lower SR often appeared to be dying (Fig. 8F). It has previously been emphasized that

mortality due to predation (i.e., grazing food web) is highly prevalent among planktonic marine organisms (e.g., Hirst and Kiorboe 2002). However, our findings suggest that mortality due to starvation may frequently be responsible for the termination of doliolid blooms.

Although total zooid densities were extremely low beneath the pycnocline, shrunken individuals consistently accounted for a large portion of the doliolid population, suggesting that these shrunken doliolids sank from upper layers. Our estimates indicate the flux attributed to dead doliolids corresponded to 8-17% of the particulate sinking flux at 150 m and that this contribution of dead doliolids would increase at the termination phase of the bloom due to the depletion of prey attributed to overgrazing. As the rate of sinking of dead doliolids is typically fast (407 m d"1; Takahashi et al. 2013), the flux due to the dead doliol-ids would be an efficient way to transport the materials to deeper waters.

Although jellyfish and teleosts, such as mackerel Scomber spp. and the mesopelagic fish species, Ceratoscopelus warmingii and Myctophum asperum, have been reported to prey on doliolids (Itoh 2009;Takagi et al. 2009) relatively few studies have been published on the potential predators of doliol-ids to date (Harbison 1998). This study confirmed that invertebrates, such as a hydromedusae and pelagic poly-cheates preyed on doliolids. In addition, a recent study revealed that the copepod, S. nigromaculata, is also a predator of doliolids (Takahashi et al. 2013). As sapphirinid cope-pods are frequently associated with the dorsal cadophore of the nurse stage, it is considered possible that these copepods may affect the growth and grazing activities of the nurse population (Paffenhofer 2013). As abundance of S. nigroma-culata, common species in this region is strongly correlated with that of doliolid (Takahashi unpubl), the potential role of these copepods in the development of doliolid blooms is considered important. The relatively small size of the invertebrate predators observed in this study relative to the size of the doliolids being preyed on was considered to indicate the importance of the trophic pathway associated with the doliolid bloom, which does not follow the typical size-dependent predator-prey rule (Hansen et al. 1994). Although tunicate blooms are generally considered to be a trophic "dead-end" in the pelagic food web (Verity and Sme-tacek 1996), this study showed that doliolids are utilized as prey by various organisms and thus, that the role of doliol-ids in the grazing food chain may be a more important than previously thought.

The results of this study also clarify aspects of the sub-mesoscale distribution and population structure of the doliolid bloom, as well as how these factors affect the development and fate of the bloom. Although the in situ observations obtained using a VPR in this study were advantageous for revealing the patch dynamics of these fragile gelatinous organisms, further investigations of the

physical characteristics and horizontal dimension of the mesoscale structures are still required to clarify their effects at a regional scale in ecosystems. In addition to doliolids, salps are also known to form dense blooms in the study area from spring to summer (Tsuda and Nemoto 1989; Takahashi unpubl). It is thus likely that bloom formation by thaliaceans is a component of the succession process of the planktonic community during the summer in this region, even though their occurrence is sporadic and ephemeral. As shown in this study, outburst by thaliaceans mirrors increases in local primary production. Thus, clarifying the mechanisms responsible for bloom development and the ecological consequences of these blooms is considered to be central to our understanding the pelagic ecosystem of the Oyashio-Kuroshio mixed water region, which is highly productive and functions as nursery grounds for numerous small pelagic fishes (Yatsu et al. 2005;Watanabe 2009). For instance rapid development of doliolid bloom in response to local increase of primary production would result in the deleterious effects with submesoscale to suspension feeding copepods by removal of food particles useful for copepods and ingestion of copepod eggs and nauplii by doliolids (Deibel 1985;Paffenhofer et al. 1995). Therefore frequent occurrence of doliolid bloom would be adverse to some planktivorous fish like Japanese sardine which primarily prefer calanoid copepods (Takagi et al. 2009). Conversely, occurrence of doliolid bloom may be beneficial to the recruitment of chum mackerel as juvenile of mackerel is known to prey on doliolid as a main prey (Itoh 2009). Physiological characteristic of chum mackerel which show faster growth at higher water temperature (Kamimura et al. 2015) also matches with occurrence pattern of doliolids in this region of which density increase with temperature (Takahashi unpubl).

References

Barstow, S. F. 1983. The ecology of Langmuir circulation: A review. Mar. Environ. Res. 9: 211-236. doi:10.1016/0141-1136(83)90040-5 Crocker, K. M., A. L. Alldredge, and D. K. Steinberg. 1991. Feeding rates of the doliolid, Dolioletta gegenbauri, on diatoms and bacteria. J. Plankton Res. 13: 77-82. doi: 10.1093/plankt/13.1.77 Deibel, D. 1985. Blooms of the pelagic tunicate, Dolioletta gegenbauri: Are they associated with Gulf Stream frontal eddies? J. Mar. Res. 43: 211-236. doi:10.1357/ 002224085788437307 Deibel, D. 1998. The abundance, distribution and ecological impact of doliolids, p. 171-186. In Q. Bone [ed.], The biology of pelagic tunicates. Oxford Univ. Press. Deibel, D., and G.-A. Paffenhofer. 2009. Predictability of patches of neritic salps and doliolids (Tunicata, Thaliacea). J. Plankton Res. 31: 1571-1579. doi:10.1093/plankt/fbp091

Deibel, D., and B. Lowen. 2012. A review of the life cycles and life-history adaptations of pelagic tunicates to environmental conditions. ICES J. Mar. Sci. 69: 358-369. doi: 10.1093/icesjms/fsr159 Everett, J. D., M. E. Baird, and I. M. Suthers. 2011. Three-dimensional structure of a swarm of the salp Thalia democratica within a cold-core eddy off southeast Australia. J. Geophys. Res. 116: C12046. doi:10.1029/2011JC007310 Gibson, D., and G.-A. Paffenhofer. 2000. Feeding and growth rates of the doliolid, Dolioletta gegenbauri Uljanin (Tuni-cata, Thaliacea). J. Plankton. Res. 22: 1485-1500. doi: 10.1093/plankt/22.8.1485 Gibson, D., and G.-A. Paffenhofer. 2002. Asexual reproduction of the doliolid, Dolioletta gegenbauri Uljanin (Tuni-cata, Thaliacea). J. Plankton. Res. 24: 703-712. doi: 10.1093/plankt/24.7.703 Hansen, B., P. K. Bjornsen, and R. J. Hansen. 1994. The size rqtion between planktonic predators and their prey. Lim-nol. Oceanogr. 39: 395-403. doi:10.4319/lo.1994.39.2.0395 Harbison, G. R.. 1998. The parasites and predators of Thaliacea, p. 187-214. In Q. Bone [ed.], The Biology of pelagic Tinicates. Oxford University Press. Hirst, A. G., and T. Kiorboe. 2002. Mortality of marine planktonic copepods: Global rates and patterns. Mar. Ecol. Prog. Ser. 230: 195-209. doi:10.3354/meps230195 Ichikawa, T. 2008. Analysis of the community structure of meso-zooplankton in the Oyashio and Transition zone using Video Plankton Recorder VPRII). Bull. Fish Res. Agen. 24: 23-104. Itoh, H. 2009. Feeding habits of larval and juvenile fish collected by a shore seine for shirasu (whitebait) in the inner part of Suruga Bay, central Japan. Bull. Jpn. Soc. Fish. Oceanogr. 73: 181-189. (In Japanese) Johnston, T. M. S., D. L. Rudnick, and E. Pellas-Sanz. 2011. Elevated mixing at a front. J. Geophys. Res. 116: C11033. doi:10.1029/2011JC007192 Kaneko, H., I. Yasuda, K. Komatsu, and S. Itoh. 2013. Observations of vertical turbulent nitrate flux across the Kur-oshio. Geophys. Res. Lett. 40: 3123-3127. doi:10.1002/ grl.50613

Kamimura, Y., M. Takahashi, N. Yamashita, C. Watanabe and A. Kawabata. 2015. Larval and juvenile growth of chub mackerel Scomber japonicus in relation to recruitment in the western North Pacific. Fish. Sci. 81: 505-513. doi:10.1007/s12562-015-0869-4 Kiorboe, T. 2011. How zooplankton feed: Mechanisms, traits and tradeoffs. Biol. Rev. 86: 311-340. doi:10.1111/j.1469-185X.2010.00148.x Paffenhofer, G.-A. 2013. A hypothesis on the fate of blooms of doliolids (Tunicata, Thaliacea) J. Plankton Res. 35: 919-924. doi:10.1093/plankt/fbt048 Paffenhofer, G.-A., L. P. Atkinson, T. N. Lee, P. G. Verity, and L. R. Bulluck. 1995. Distribution and abundance of thaliaceans and copepods off the southeastern U.S.A. dur-

ing winter. Cont. Shelf Res. 15: 255-280. doi:10.1016/ 0278-4343(94)E0004-6 Paffenhofer, G.-A., and D. M. Gibson. 1999. Determination of generation time and asexual fecundity of doliolids (Tunicata, Thaliacea). J. Plankton Res. 21: 1183-1189. doi: 10.1093/plankt/21.6.1183 Paffenhofer, G.-A., and M. Koster. 2011. From one to many: On the life cycle of Dolioletta gegenbauri Uljanin (Tunicata, Thaliacea). J. Plankton Res. 33: 1139-1145. doi:10.1093/ plankt/fbr001

Paffenhofer, G.-A., T. B. Stewart, M. J. Youngbluth, and T. G. Bailey. 1991. High-resolution vertical profiles of pelagic tunicates. J. Plankton Res. 13: 971-981. doi:10.1093/ plankt/13.5.971 Patonai, K., H. El-Shaffey, and G.-A. Paffenhofer. 2011. Sinking velocities of fecal pellets of doliolids and calanoid copepods. J. Plankton Res. 33: 1146-1150. doi:10.1093/ plankt/fbr011

Takagi, K., A. Yatsu, H. Itoh, M. Moku, and H. Nishida. 2009. Comparison of feeding habits of myctophid fishes and juvenile small epipelagic fishes in the western North Pacific. Mar. Biol. 156: 641-659. doi:10.1007/s00227-008-1115-8 Takahashi, K., T. Ichikawa, H. Saito, S. Kakehi, Y. Sugimoto, K. Hidaka, and K. Hamasaki. 2013. Sapphirinid copepods as predators of doliolods: Their role in doliolid mortality and sinking flux. Limnol. Oceanogr. 58: 1972-1984. doi: 10.4319/lo.2013.58.6.1972 Tebeau, C. M., and L. P. Madin. 1994. Grazing rates for three life history stages of the doliolid Dolioletta gegenbauri Uljanin (Tunicata, Thaliacea). J. Plankton Res. 16: 1075-1081. doi:10.1093/plankt/16.8.1075 Terazaki, M., and C. Tomatsu. 1997. A vertical multiple opening and closing plankton sampler. J. Adv. Mar. Sci. Tech. Sci. 3: 127-132. Thorpe, S. A. 2004. Langmuir circulation. Annu. Rev. Fluid Mech.

36: 55-79. doi:10.1146/annurev.fluid.36.052203.071431 Tsuda, A., and T. Nemoto. 1992. Distribution and growth of salps ina Kuroshio warm-core ring during summer 1987. Deep-Sea Res. 1A 39: S219-S229. doi:10.1016/S0198-0149(11)80013-0 Verity, P. G., and V. Smetacek. 1996. Organism life cycles, predation, and structure of marine pelagic ecosystems. Mar. Ecol. Prog. Ser. 130: 277-293. doi:10.3354/ meps130277

Watanabe, Y. 2009. Recruitment variability of small pelagic fish populations in the Kuroshio-Oyashio transition region of the western North Pacific. J. Northw. Atl. Fish. Sci. 41: 197-204. doi:10.2960/J.v41.m635 Yatsu, A., T. Watanabe, M. Ishida, H. Sugisaki, and L. D. Jacobson. 2005. Environmental effects on recruitment and productivity of Japanese sardine Sardinops melanostic-tus and chub mackerel Scomber japonicus with recommendations for management. Fish. Oceanogr. 14: 263-278. doi:10.1111/j.1365-2419.2005.00335.x

Acknowledgments

We thank the captains and crew of the research vessel, Soyo-Maru, and all of the participants of the cruises for their cooperation at sea. We also thank R. Fukuhara for data analysis of the video plankton recorder images and Y. Nishibe for helpful comments on data processing. H. Kaneko and T. Okunishi provided helpful comments for understanding the physical environment of the sampling sites. A portion of this research was supported by a grant for "Studies on Prediction and Application of Fish Species Alternation" by the Research and Development Department of the Agriculture Forestry and Fisheries Research Council,

and by Grants-in-Aid for Scientific Research on Innovative Areas and Scientific Research on Innovative Areas (B) from the Ministry of Education, Science, Sports and Culture (No. 24310007 and 24121005 to K.T.).

Submitted 15 December 2014 Revised 1 June 2015 Accepted 15 June 2015

Associate editor: Thomas Anderson