JPR Advance Access originally published online on October 4, 2009
Journal of Plankton Research 2009 31(12):1441-1452; doi:10.1093/plankt/fbp092
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
An advanced laser-based fluorescence microstructure profiler (TurboMAP-L) for measuring bio-physical coupling in aquatic systems
1 Faculty of Marine Science, Tokyo University of Marine Science and Technology, 5-7, Konan 4, Minato-ku, Tokyo 108-8477, Japan 2 Oceanographic Lab, JFE Alec Co., 7-2-3 Ibukidai-higashi, Nishi, Kobe 651-2242, Japan
* CORRESPONDING AUTHOR: hide{at}kaiyodai.ac.jp
Received on June 3, 2009; accepted on September 7, 2009
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
An advanced model of the Turbulence Microstructure Acquisition Profiler, TurboMAP-L, to investigate the coupling between phytoplankton and turbulence in aquatic systems is described. The profiler provides simultaneous measurement of turbulence, hydrographic and biological parameters at sampling rates between 64 and 512 Hz. Specifically, the addition of a new laser fluorescence-based sensor extends the measurement of in situ chlorophyll fluorescence distributions to millimeter scales. Complementary information on phytoplankton patch and particulate matter size and spatial structure are obtained through the attachment of a separate CMOS mini-camera system. Images of the TurboMAP-L LED fluorescence/turbidity sensor sample volume are obtained with 330 x 330 µm resolution. Results with respect to the performance of the laser sensor and camera system are presented from laboratory tests and field experiments conducted in coastal waters off Tokyo, Japan.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
In aquatic ecosystems, the measurement of phytoplankton distributions provides the basis for the elucidation of the physical and biological processes which influence plankton ecology (Steele, 1989
; Smetacek and Cloern, 2008). For phytoplankton, small-scale turbulence may directly influence fundamental biological processes such as nutrient diffusion (Karp-Boss et al., 1996), cell division (Berdalet, 1992), herbivorous grazing (Kiørboe and Saiz, 1995) and aggregate formation (Kiørboe, 2001). Coincident in situ measures of phytoplankton, environmental and turbulent fields at submetre scales are therefore fundamental to obtaining a greater understanding of the interplay between phytoplankton and the physical environment as well as the consequences of these interactions for planktonic ecosystems and biogeochemical cycling.
Technical advances have been central to submetre scale measures of phytoplankton distributions. Owing to the ubiquitous presence of fluorescent chlorophyll pigments in phytoplankton, fluorometers are the principal devices used for the in situ measurement of phytoplankton biomass (Lorenzen, 1966; Denman and Gargett, 1995). For the measurement of undisturbed phytoplankton distributions, free-fall fluorescence microstructure profilers with high-resolution sensors located at the front end of instruments are necessary (Desiderio et al., 1993; Doubell et al., 2006). The application of fluorescence microstructure profilers has been important in revealing the occurrence and ecological significance of phytoplankton thin layers in stratified systems (Alldredge et al., 2002; McManus et al., 2003).
Recently, simultaneous measurements of fluorescence and turbulence microstructure using the Turbulence Ocean Microstructure Acquisition Profiler, TurboMAP (Wolk et al., 2002), have extended the investigation of microscale phytoplankton distributions into the surface mixed layer. Although results from these studies indicated a degree of coupling between phytoplankton and turbulent mixing processes (Yamazaki et al., 2006), the complex patterns and high levels of phytoplankton variability measured over scales of centimeters to tens of centimeters differed significantly from the observed scalar fields (Mitchell et al., 2008). Further, fluorescence microstructure profiling with millimeter-scale resolution in well-mixed waters has revealed increased levels of microscale phytoplankton patchiness characterized by irregularly spaced millimeter to centimeter-scale patches of highly elevated fluorescence concentrations (Doubell et al., 2006). These results suggest a dynamic and complex seascape composed of small phytoplankton patches and steep gradients that may be an intrinsic feature of highly turbulent systems. To date, coincident measurement of turbulence and fluorescence microstructure below centimeter scales by high-resolution profilers has not been reported.
For the exploration of plankton and particulate matter structure below centimeter scales, in situ imaging techniques have been used. Imaging systems have included the use of photographic devices (MacIntyre et al., 1995; Tiselius and Kuylenstierna, 1996), two-dimensional (2D) laser fluorometers (Franks and Jaffe, 2001, 2008), video plankton recorders (Davis et al., 2005) and 3D digital holography (Katz et al., 1999;. Watson, 2004). These techniques have provided new insights into the in situ structure and composition of plankton communities. However, they are not integrated with microstructure profilers to provide simultaneous biological and physical parameters at small scales, a challenge that TurboMAP-Laser (TurboMAP-L) can address.
To elucidate the structure and dynamics of microscale phytoplankton variability in relation to turbulent mixing, an advanced version of the TurboMAP profiler (Wolk et al., 2002, 2006), TurboMAP-L, has been developed. In addition to the environmental sensors found on preceding models, TurboMAP-L is fitted with a second shear sensor and a new laser-based fluorescence sensor capable of millimeter-scale resolution. Complementary information on the 2D size and spatial structure of phytoplankton and particle aggregation is obtained through the attachment of a separate compact digital camera system to the profiler. Similar imaging devices have successfully been used to study the foraging behavior of marine mammals (Otani et al., 1998; Watanabe et al., 2003), but, until now, the technology has not been applied to the study of microscale particle aggregation. The major objective of this article is to provide a description and assessment of the TurboMAP-L fluorescence sensors and image logging system. We demonstrate the capability of the combined systems for the complementary measurement of in situ microscale phytoplankton and particle structure.
|
METHOD |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
TurboMAP-L system
TurboMAP-L is a free-fall instrument (Fig. 1) which carries sensors for turbulent shear and temperature gradient (
u/z, v/z and T/z), biological (in vivo chlorophyll fluorescence) and turbidity microstructure as well as standard hydrographic parameters (conductivity, temperature, depth) to a depth of ca. 500 m. Table I provides the characteristics for each sensor. The instrument weight is 30 kg in air, measures 2 m in length, 0.12 m in diameter and the sensor electronics are housed within the titanium pressure casing. For ease of handling, the device is assembled in two interconnecting parts. Sensors are mounted forward of the rounded nose to ensure the measurement of undisturbed fields. Four (detachable) 0.25 m long titanium prongs extend from the nose to protect the sensors during deployment. Sensors are sampled at a rate between 64 and 512 Hz for different channels and data are transmitted in real time to a shipboard computer via a 5 mm Kevlar tether which is anchored to the end of the rear pressure casing. Deployment from the ship's trawl deck is made using an electronic winch. The winch operator adjusts the rate at which the tether is released to maintain a loose connection between the profiler and the ship during descent. Typical profiling speeds are between 0.50 and 0.80 m s–1. Reversal of the winch direction returns the instrument rapidly to the ship following each cast.
|
|
Laser fluorescence probe
A new laser fluorescence probe has been designed to obtain improved resolution over the previous light emitting diode (LED) fluorescence/turbidity probe. To minimize flow distortion of the sampled fluorescence field, the fluorescence probe has an aerodynamic design (Fig. 1C). The positioning of both the excitation and the receiver diode on a single flat surface further reduces the possibility of recirculation within the sample volume due to mixing caused by irregularities in the probe shape. A blue diode laser (NDHB510APA; Nichia Chemical Industries Ltd) is used for chlorophyll excitation (peak 410 nm) and is projected at 45º outward and into the oncoming flow. Detection of the backscatter fluorescence occurs within a cylindrically shaped (10 x 2 mm, length x diameter) sample volume centered 8 mm in front of the optical receiver diode (640–1000 nm). The reduced sample volume of the laser probe (32 µL), in comparison with the LED probe (
4 mL), allows for measurement of the fluorescence field with increased spatial resolution. Calibration and testing of the ranges and sensitivity to fluorescent sources (e.g. sodium fluorescein and pure chlorophyll a solutions) were carried out in the laboratory using standard methods (Wolk et al., 2001, 2002). The normally arbitrary units used to measured fluorescence are calibrated, so that the output units are approximately equivalent to µg L–1 of chlorophyll a.
The spatial wavenumber response of the laser probe was determined by towing the sensor and a fast response thermistor (FP07) at 0.1 m s–1 through a warm, plume of sodium fluorescein in a laboratory channel following the methods of Wolk et al. (2006). Temperature spectra were corrected using the sensor's empirical response function (Gregg, 1999). Fluorescence and temperature spectra matched over wavenumbers up to the resolution of the FP07 sensor (400 cycles m–1 or 2.5 mm; Fig. 2). Since the molecular diffusivity for the sodium fluorescein is two orders of magnitude smaller than the molecular thermal diffusivity, the fluorescence structure can persist on a longer timescale than temperature and has more power at high wavenumbers. For the fluorescence signal reduced variance due to the anti-aliasing filter of the recording system occurred at wavenumbers above 1000 cycles m–1. Hence, the spatial resolution of the sensor was determined to be at least 1 mm, and no correction of the fluorescence spectra is required.
|
Image logging system
The Digital Still Logger camera system (DSL 190-VDTII; Little Leonardo Co., Japan), hereafter referred to as DSL, measures 2.2 cm in diameter, 13.3 cm in length and weighs 0.082 kg in air. A sealed aluminum casing houses a 1024 x 1280 pixel, 24 bit CMOS camera with a 4.4 mm lens and is suitable for operations to a depth of 190 m. DSL is powered by an internal lithium manganese-dioxide battery and contains additional sensors for the measurement of pressure and temperature. Images are sampled at a selectable rate (maximum 5 Hz). The camera shutter speed is 66 ms. Data are compressed before being stored in a 2 GB internal flash memory and communications with the logger are carried out using a USB2.0 interface.
The LED light source of the fluorescence/turbidity probe provides illumination for the DSL camera. The probe uses an array of six blue (peak 460 nm) LEDs which are arranged on a circumference of 20 mm diameter and are tilted 30° toward the centre of this circle (Wolk et al., 2001, 2006). Intersection of the light beams creates a sample volume with a focal point 15 mm above the center of the probe. The DSL camera is fitted onto TurboMAP-L using an aluminum frame which attaches to profiler's protective prongs (Fig. 1B). The focal depth of the camera is 10 cm and is aligned with the center of the LED sample volume. The camera lens aperture of f/2.8 provides sufficient depth of field to ensure that the camera is in focus over the entire sample volume depth (2 cm). Images are orientated normal to the direction of travel, and with a field of view of 6.0 x 7.5 cm, the camera system has pixel resolution of 59 x 59 µm.
Inspection of field images and laboratory tests with phytoplankton showed that the DSL resolution was too coarse to identify individual cells. This resulted in a loss of information across adjacent pixels. To establish the camera's spatial resolution, it was necessary to determine the filter characteristics of the DSL. In a laboratory tank, we measured the camera's response to a point source smaller than the resolution of the camera. Aligning the peak light intensity measured along two adjacent pixels in both the x and the y directions, an average point spread function (PSF) was formed (Fig. 3A). The Fourier transform of the PSF gives the system's optical transfer function (OTF). Following the method of Franks and Jaffe (2001), a Monte Carlo simulation was performed and the simulated data were used to calculate the filter characteristics of the camera. The OTF was convolved with random time-series in the frequency domain and the average spectrum was calculated (Fig. 3B). The OTF showed that the system behaves as a low-pass filter with a sharp reduction in variance beginning at 10 cycles cm–1. We assume the half power point at 30 cycles cm–1 to be a conservative estimate of the cameras spatial resolution (330 x 330 µm).
|
With a shutter speed of 66 ms and a pixel resolution 59 µm, the movement of objects while the camera shutter is open inevitably causes some streaking in images. To quantify the influence of particle motions on particle size estimates, we towed the camera longitudinally and laterally past a series of 0.5 cm diameter circles in a laboratory channel at speeds ranging from 6.0 to 19.0 cm s–1. With the camera moving in the longitudinal plane (i.e. front-on toward circles), images were not subject to streaking; however, the relative diameter of circles depended on their distance from the camera lens. Hence, objects imaged in the field are sized, assuming that they are located in the middle of the illuminated sample volume with accuracy for the diameter of ±10%. For lateral movement, streaking was a linear function of cameras tow velocity (r2 = 0.96, P < 0.01):
|
|
Images taken in situ are a 2D projection of particle movements through the illuminated sample volume. At the descent speed of TurboMAP-L, the dominant trajectory of particles is into the imaging plane. Relative to the time/space scale of images, the lateral movement of small particles or phytoplankton may be caused by cell motility or small-scale velocity fluctuations generated by turbulent shear. In regions of strong turbulent mixing, such as those just below the sea surface, typical values of turbulent kinetic energy dissipation are in the order of 10–6 W kg–1 and will generate turbulent velocities (0.5 cm s–1) much greater than the swimming speed of phytoplankton (Yamazaki and Squires, 1996). In these conditions, according to the above relation, lateral motions due to turbulence do not cause the elongation of particles in images by more than 8 pixels.
Field images acquired by the DSL camera are initially cropped to 340 x 340 pixels (2.0 x 2.0 cm) over the central region closest to the LED light source where the incident irradiance is strongest. This provides an image which corresponds closely to the measured LED fluorescence/turbidity sensor sample volume. Owing to the exponential decrease of the LED irradiance from the light source (Wolk et al., 2001, 2006), images are corrected for non-uniform illumination across the image area following a previously described method (Franks and Jaffe, 2001, 2008). To do this, a beam correction pattern is formed from the average of all the images in a profile. This averaged image is then smoothed using a 15 x 15 pixel averaging filter. Each image is then divided by the beam correction pattern to normalize for variations in incident irradiance.
Owing to the large number of images taken during a profile, all images are screened for streaking using the technique described in Franks and Jaffe (2001, 2008). The extent of streaking is quantitatively determined by the relative isotropy or anisotropy of the 2D spectrum of each image (Fig. 4). Briefly, each image is first multiplied by a Butterworth window and the 2D image spectrum is calculated. Contours of the 2D spectrum in an unstreaked image approach a radially symmetrical pattern, whereas in a streaked image, the contours form elongate ellipses with the major axis oriented perpendicular to the direction of streaking. The ratio of the 1D variance at each wavenumber along two pairs of perpendicular lines is estimated. Images are rejected if the average ratio for one pair was <1.5 or >1.5. On the basis of the above criteria, no images were rejected. This result indicates that TurboMAP-L provided a stable platform for deployment of the camera system and the observed images are not significantly altered by the motion of the profiler or the movement of particles.
|
E = 1.5 or 1/1.5.)Finally, binary images of individual particles above a threshold light intensity value are formed (see Franks and Jaffe, 2008
). Quantification of individual particle properties, such as minor and major axis length and particle area, are then determined using the MATLAB regionprops function (The Mathworks Inc., Natick, MA, USA). Threshold value selection, set as the mean profile light intensity in this study, is based on the minimum value that would reliably discriminate individual particles. Trials showed that the choice of threshold value did not significantly alter particle size estimates or the slope of the particle size-abundance spectrum.
TurboMAP-L has been deployed in different coastal areas on numerous occasions. We present the initial results of TurboMAP-L profiling in combination with the DSL camera. The experiment was conducted from R/T V Seiyo Maru at night on 8 February 2008 at a station situated near the Tokyo Bay entrance (35°16'26''N, 139°43'14''E) and during the day on 29 May 2008 in adjacent Sagami Bay (35°02'56''N, 139°43'20''E), Japan. Ten repeated TurboMAP-L profiles were made at each site. For each profile, turbulent kinetic energy dissipation rates (
) were calculated by integrating the shear spectrum over 2 s data intervals. Standard CTD and fluorescence profiling was conducted using a Seapoint system. The DSL camera system was configured to take images continuously at a rate of 5 Hz giving images vertically separated by 10–16 cm. Once deployed, the camera recorded images continuously to the limit of its internal memory.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Field assessment: TurboMAP-L laser probe
To demonstrate the performance of the fluorescence laser probe in relation to the LED probe and the Seapoint fluorescence probe mounted on a separate CTD, we use data obtained in Sagami Bay. Figure 5 shows the distribution of shear, temperature, density and fluorescence measured by TurboMAP-L in Sagami Bay. The considerable microscale structure in shear and relatively uniform distribution of temperature observed down to a depth of 30 m indicated an actively turbulent mixing layer. Below this layer, the water column was thermally stratified. The average turbulent kinetic energy dissipation rates for the mixing layer (10–30 m) and stratified waters below (30–100 m) were 1.87 x 10–7 and 6.96 x 10–9 W kg–1, respectively.
|
Fluorescence profiles measured by TurboMAP-L (Fig. 5C) and Seapoint system (Fig. 6A) showed a general decrease in fluorescence with depth, particularly below the surface mixed layer. Comparison of 1 m averaged fluorescence levels measured by each device showed each of the TurboMAP-L fluorescence sensors provided fluorescence values consistent with conventional fluorescence sensors attached to CTD assemblies (Fig 6A and B). The measured distribution of phytoplankton biomass becomes increasingly patchy when measured with centimeter (LED probe) and millimeter (laser probe) scale resolution (Fig. 5C). The reduced sample volume of the laser probe measured increased levels of fluorescence variability. Fluorescence distributions measured by the laser probe detailed multiple point fluorescence peaks ranging in size from
0.4 to 4.08 cm width. Fluorescence peaks were characterized by steep boundary gradients where >15-fold increases in the signal intensity were common. Peaks were separated by regions of low and constant fluorescence values. For TurboMAP-L, mean fluorescence intensity estimates from each fluorescence probe showed r2 values >0.6 when concentrations are averaged over spatial scales >10 cm (Fig. 6C). The rapid decrease in correlation observed at smaller length scales is not surprising, since the observed patch scale was in the order of millimeters to centimeters and the separation distance between two probes is 12 cm.
|
Fluorescence spectra calculated from the TurboMAP-L probes provide information on the structure of the microscale phytoplankton field and potentially the causal mechanisms for its formation (Yamazaki et al., 2006
). To compare the capacity of each probe to measure fluorescence variability with spatial scale, we examined fluorescence spectra calculated over 4 s (1024 point) sections of data (Fig. 7). The limited wavenumber response of the LED fluorescence probe was corrected using the probes spatial transfer function giving a maximum spatial resolution of 2 cm (Wolk et al., 2006). In general, the shape of the phytoplankton spectrum derived from each probe was matching at low wavenumbers (1–10 cycles m–1) and varied from white (Fig. 7A) to red (Fig. 7B). At higher wavenumbers (10–100 cycles m–1), the laser probe's ability to resolve smaller scales (i.e. separate peak structures) consistently resulted in the measurement of increased variance. This caused the phytoplankton spectrum to flatten as predicted by Franks (Franks, 2005) and is consistent with scaling arguments that suggest volume-averaged measures of phytoplankton concentrations transition from being distributed as a continuum to discrete at length scales of 10 cm (Siegel, 1998).
|
Field assessment: DSL camera
The successful imaging of objects required sufficient contrast between the LED-illuminated sample volume and background light field. The camera system does not work well in surface waters during daylight hours. Useable images from the camera system during day time profiling in Sagami Bay (Fig. 5) were obtained at depths below which the ambient light conditions were adequately dark (60 m). In contrast, the DSL camera during night time operation in Tokyo Bay was free from the ambient light problem.
Figure 8 shows the concurrent vertical distribution of temperature, density, shear and fluorescence measured by TurboMAP-L in Tokyo Bay. Tidal currents and frequent strong wind events during winter mix the relatively shallow water column at the bay entrance (ca. 50 m depth). Temperature variation was <0.7°C over the profiled depth and <0.07°C down to 30 m depth. The average turbulent kinetic energy dissipation rate for the upper water column (6–30 m) was 6.31 x 10–7 W kg–1. Below 30 m depth in the bottom boundary layer, strong mixing, evidenced by the measurement of multiple density overturns, was associated with an increase in the average turbulent kinetic energy dissipation rate to 1.01 x 10–6 W kg–1. The buoyancy frequency was in the order of 10–3 and 10–2 s–1 in the top and bottom layers, respectively. The observed level and pattern of turbulence in the upper water column were consistent with the relatively uniform distribution of hydrographic parameters and confirm that the variability observed in fluorescence profiles or images is not associated with finescale features, such as pycnocline or nutricline development, typical of more stable water columns.
|
Closer inspection of the fluorescence microstructure measured TurboMAP-L in Tokyo Bay clearly demonstrates the increased levels of microscale patchiness measured by the laser probe in comparison with the LED probe (Fig. 9). Laser-derived fluorescence distributions were dominated by high-intensity fluorescence peaks separated by a background of low and constant values. Peaks were defined over multiple points and again were characterized by steep boundary gradients. For each profile, maximum peak concentrations >15-fold relative to background values were observed. Millimeter- to centimeter-sized peaks dominated profiles and the largest peak width recorded was 2.80 cm. Smaller peaks generally displayed a monotonic rise and decay in fluorescence, whereas larger peaks showed varying morphologies characterized by multiple intra peak fluctuations in fluorescence.
|
Visual inspection of the corresponding in situ images obtained by the DSL camera revealed a strong consistency with the spatial structure of fluorescence measured by the laser probe (Fig. 9). Particles ranged from 0.16 to 15.02 mm in size (major axis length) and were separated by sub-millimeter to millimeter length scales. High particle abundances were observed throughout the water column in Tokyo Bay (Fig. 10), and there were no consistent changes in the abundance or size distribution with depth or time.
|
Particle size-abundance spectra calculated for both the Tokyo and the Sagami Bay systems followed a strong power law distribution (Fig. 11). The size-abundance spectrum for all particles measured in Tokyo (n = 171950) and Sagami Bay (n = 10 487) peaked at
0.35 mm. The reduction in the abundance of particles measuring smaller than 0.35 mm in size is likely to be an artifact caused by the limited resolution of the camera (Fig. 3). For Tokyo Bay, two distinct scaling regions were observed. Particles ranging from 0.34 to 3.65 mm had a slope of –1.66 (r2 = 0.98, P < 0.01) and larger particles a slope of –5.07 (r2 = 0.97, P < 0.01). For Sagami Bay, the slope of the size-abundance spectrum, –2.98 (r2 = 0.94, P < 0.01), was consistent across the range of particle sizes. Particle area increased with particle size with a slope of 1.57 (r2 = 0.92, P < 0.01) and 1.56 (r2 = 0.79, P < 0.01) in Tokyo and Sagami Bay, respectively. If particles were circular, the size-area spectrum would scale with a slope of 2, indicating that the particles in the DSL images were slightly elongated.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
For the first time, we show the simultaneous measurement of turbulent shear and phytoplankton distributions throughout the euphotic zone with millimeter scale resolution (Figs 5 and 8). A new laser fluorescence sensor has revealed the distribution of phytoplankton in separate millimeter to centimeter scale patches. The extent of phytoplankton patchiness measured is consistent with previous measures of microscale fluorescence distributions made with millimeter-scale resolution (Franks and Jaffe, 2001
, 2008; Doubell et al., 2006); however, our observations are supported by corresponding measures of small-scale turbulence. The relatively uniform distribution of environmental parameters (i.e. temperature, density) observed in the upper water column of both systems (Figs 5 and 8) are the result of active mixing shown by the measured turbulence intensity (10–7–10–6 W kg–1). Under these conditions, we observed high levels of phytoplankton spatial heterogeneity at the millimeter–centimeter scales (Fig. 9) where current phytoplankton distribution models predict a maximum of homogeneity (Siegel, 1998). The patterns of fluorescence variability measured by the laser sensor are supported by the vertical distribution, high abundance and size of particles identified by the DSL camera (Figs 10 and 11). This consistency suggests that imaged particles are predominately phytoplankton aggregates. Imaged particles were typically separated by sub-millimeter- to millimeter-scale distances (Fig. 9). Larger fluorescence peaks displayed internal fluctuations in fluorescence over one or two points (Fig. 9). These fluctuations are consistent with the packaging of smaller phytoplankton aggregates (i.e. particles resolved by the camera) separated by distances smaller than the resolution of the laser sensor. Hence, it is likely that the peak structures identified by the laser probe constitute patches of increased biomass which may include individual phytoplankton cells as well as chains and aggregates. Testing this hypothesis requires high-resolution water sampling that has not been developed yet.
Coagulation theory states that the stickiness and concentration of phytoplankton and other suspended particles, as well as the turbulent fluid shear responsible for driving collisions among particles, are the three dominant factors relevant to the formation of marine aggregates (Kiørboe, 2001). Winter and spring phytoplankton assemblages in Tokyo and connected Sagami Bay are dominated by diatoms (Kanda et al., 2003; Nakane et al., 2008), in particular Chaetoceros spp., typically measuring <25.0 µm in size (Nakane et al., 2008). The slopes of the size spectra for particles <3.65 mm in size in Tokyo Bay (Fig. 11A1) and in Sagami Bay (Fig. 11B1) are consistent with previous laboratory (Jackson et al., 1995) and field (Kiørboe et al., 1998) measures of phytoplankton particle size spectra dominated by Chaetoceros species. The demonstrated ability of diatoms, including Chaetoceros spp., to form aggregates due to physical coagulation (Kiørboe and Hansen, 1993; Jackson et al., 1995) coupled with the high levels of shear observed in Tokyo Bay (Fig. 8A) may potentially explain the high abundance of particles observed here. Steeper slopes, as was observed for larger particles (>3.65 mm) in Tokyo Bay, are expected for shear-driven coagulation (McCave, 1984). Since the Kolmogorov scale is 1–3 mm for the observed turbulence intensity in Tokyo Bay (Fig. 8B), the observed shift in the scaling of the particle size-abundance spectrum may be due to a transition between inertia dominated to viscous dominated regime. In contrast, the Kolmogorov scale for the observed turbulence intensity at depths >60 m in Sagami Bay (Fig. 5B) is 3–10 mm. This is larger than the majority of particles measured in Sagami Bay and no change in the slope of the particle size-abundance spectra was observed (Fig. 11B).
We have demonstrated that the TurboMAP-L profiler provides an integrated platform for the investigation of biological and physical processes at the oceanic microscale. To our knowledge, we have shown, for the first time, coincident measures of fluorescence and turbulence microstructure made with millimeter-scale resolution and corresponding images of particle size and abundance. The complementary information obtained from this integrated system will considerably expand our understanding of the processes influencing aggregation processes at the microscale in the ocean. Although the mechanisms responsible for the formation and coupling between aggregates and phytoplankton patches require further investigation, our results show that extensive aggregate and microscale phytoplankton patchiness may be a common feature under strong turbulent conditions. Considering the rich composition of marine aggregates, which includes organic and inorganic materials such as transparent exopolymeric particles, fecal pellets, phytoplankton, including chain forming diatoms and microbial communities, the observed microscale patches and particles are expected to be of key ecological significance. Such observations are critical to the testing of current theoretical models of coagulation and phytoplankton patch generation, with wider implications for understanding the mechanisms responsible for the structuring planktonic food webs and the fate primary production in aquatic ecosystems.
|
FUNDING |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
This study was funded by a Grant-in-Aid for Science Research (B2) 16310005 from Japan Society for the Promotion of Science. M.J.D. was supported by a Japan Society for the Promotion of Science post-doctoral fellowship (JSPS-07770).
|
ACKNOWLEDGEMENTS |
|---|
Thank you to the Captain and crew of the R/TV Seiyo Maru for their assistance with the deployment of the CTD and TurboMAP-L systems.
|
Notes |
|---|
Corresponding editor: Roger Harris
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Alldredge A. L., Cowles T. J., MacIntyre S., et al. Occurrence and mechanisms of formation of a dramatic thin layer of marine snow in a shallow Pacific fjord. Mar. Ecol. Prog. Ser. (2002) 233:1–12.[CrossRef]
Berdalet E. Effects of turbulence on the marine dinoflagellate Gymnodinium nelsonii. J. Phycol. (1992) 28:267–272.[CrossRef][Web of Science]
Davis C. S., Thwaites F. T., Gallager S. M., et al. A three-axis fast-tow digital video plankton recorder for rapid surveys of plankton taxa and hydrography. Limnol. Oceanogr.: Methods (2005) 3:59–74.
Denman K. L., Gargett A. E. Biological–physical interactions in the upper ocean: the role of vertical and small scale transport processes. Annu. Rev. Fluid Mech. (1995) 27:225–255.[Web of Science]
Desiderio R. A., Moore C., Lantz C., et al. Multiple excitation fluorometer for in situ oceanographic applications. Appl. Optics. (1993) 36:1289–1296.
Doubell M. J., Seuront L., Seymour J. R., et al. High-resolution fluorometer for mapping microscale phytoplankton distributions. Appl. Environ. Microbiol. (2006) 72:4475–4478.
Franks P. J. S. Plankton patchiness, turbulent transport and spatial spectra. Mar. Ecol. Prog. Ser. (2005) 294:295–309.[CrossRef]
Franks P. J. S., Jaffe J. S. Microscale distributions of phytoplankton: initial results from a two-dimensional imaging fluorometer. Mar. Ecol. Prog. Ser (2001) 220:59–72.[CrossRef]
Franks P. J. S., Jaffe J. S. Microscale variability in the distributions of large fluorescent particles observed in situ with a planar laser imaging fluorometer. J. Mar. Syst. (2008) 69:254–270.[CrossRef]
Gregg M. C. Uncertainties and limitations in measuring 
and T 
. J. Atmos. Ocean Tech. (1999) 16:1483–1490.[CrossRef]
Jackson G. A., Logen B. E., Alldredge A. L., et al. Combining particle size spectra from a mecocosm experiment measured using photographic and aperture impedance (Coulter and Elzone) techniques. Deep-Sea Res. II (1995) 42:139–157.[CrossRef]
Kanda J., Fujiwara S., Kitazato H., et al. Seasonal and annual variation in the primary production regime in the central part of Sagami Bay. Prog. Oceanogr. (2003) 57:17–29.[CrossRef]
Katz J., Donaghay P., Zhang J., King S., Russell K. Submersible holocamera for detection of particle characteristics and motions in the ocean. Deep-Sea Res. I-Oceanogr. Res. Pap. (1999) 46:1455–1481.
Karp-Boss L., Boss E., Jumars P. A. Nutrient fluxes to planktonic osmotrophs in the presence of fluid motion. Oceanogr. Mar. Biol. Annu. Rev. (1996) 34:71–107.
Kiørboe T. Formation and fate of marine snow: small-scale processes with large-scale implications. Sci. Mar. (2001) 65:57–71.
Kiørboe T., Hansen J. L. S. Phytoplankton aggregate formation: observations of patterns and mechanisms of cell sticking and the significance of exopolymeric material. J. Plankton Res. (1993) 15:993–1018.
Kiørboe T., Saiz E. Planktivorous feeding in calm and turbulent environments, with emphasis on copepods. Mar. Ecol. Prog. Ser. (1995) 122:135–145.[CrossRef]
Kiørboe T., Tiselius P., Mitchell-Innes B., et al. Intensive aggregate formation with low vertical flux during an upwelling-induced diatom bloom. Limnol. Oceanogr. (1998) 43:104–116.
Lorenzen C. J. A method for the continuous measurement of in-vivo chlorophyll concentration. Deep-Sea Res. (1966) 13:223–227.
MacIntyre S., Alldredge A. L., Gotschalk C. C. Accumulation of marine snow at density discontinuities in the water column. Limnol. Oceanogr. (1995) 40:449–468.
McCave I. N. Size spectra and aggregation of suspended particles in the deep ocean. Deep-Sea Res. (1984) 31:329–352.[Web of Science]
McManus M. A., Alldredge A. L., Barnard A. H., et al. Characteristic distribution and persistence of thin layers over a 48 hour period. Mar. Ecol. Prog. Ser. (2003) 261:1–19.[CrossRef]
Mitchell J. G., Yamazaki H., Seuront L., et al. Phytoplankton patch patterns: seascape anatomy in a turbulent ocean. J. Mar. Syst. (2008) 69:247–253.[CrossRef]
Nakane T., Nakaka K., Bouman H., et al. Environmental control of short-term variation in the plankton community of inner Tokyo Bay, Japan. Estuar. Coast. Shelf Sci. (2008) 78:796–810.[CrossRef]
Otani S., Naito Y., Kawamura A., et al. Diving behavior and performance of harbour porpoises, Phocoena phocoena, in Funka Bay, Hokkaido, Japan. Mar. Mammal Sci. (1998) 14:209–220.[CrossRef]
Siegel D. A. Resource competition in a discrete environment: why are plankton distributions paradoxical? Limnol. Oceanogr. (1998) 43:1133–1146.
Smetacek V., Cloern J. E. On phytoplankton trends. Science (2008) 319:1346–1348.
Steele J. H. The ocean ‘landscape. Landsc. Ecol. (1989) 3:185–192.[CrossRef]
Tiselius P., Kuylenstierna M. Growth and decline of a diatom spring bloom: phytoplankton species composition, formation of marine snow and the role of heterotrophic dinoflagellates. J. Plankton Res. (1996) 18:133–155.
Watanabe Y., Mitani Y., Sato K., et al. Dive depths of Weddell seals in relation to vertical prey distribution as estimated by image data. Mar. Ecol. Prog. Ser. (2003) 252:283–288.[CrossRef]
Watson J. HoloMar: a holographic camera for subsea imaging of plankton. Sea Tech. (2004) 45:53–55.
Wolk F., Seuront L., Yamazaki H. Spatial resolution of a new micro-optical probe for chlorophyll and turbidity. J. Tokyo Uni. Fish. (2001) 87:13–21.
Wolk F., Yamazaki H., Seuront L., et al. A new free fall profiler for measuring biophysical microstructure. J. Atmos. Ocean. Tech. (2002) 19:780–793.[CrossRef]
Wolk F., Yamazaki H., Li H., et al. Calibrating the spatial response of bio-optical sensors. J. Atmos. Ocean. Technol. (2006) 23:511–516.[CrossRef]
Yamazaki H., Squires K. D. Comparison of oceanic turbulence and copepod swimming. Mar. Ecol. Prog. Ser. (1996) 144:299–301.[CrossRef]
Yamazaki H., Mitchell J. G., Seuront L., et al. Phytoplankton microstructure in fully developed oceanic turbulence. Geophys. Res. Lett. (2006) 33:L01603. doi: 10.1029/2005GL024103.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||