Deep-Sea Research I 47 (2000) 505}531

Diel variation in the vertical distribution of particulate matter ('0.15 mm) in the NW Mediterranean Sea investigated with the Underwater Video Pro"ler Lars Stemmann*, Marc Picheral, Gabriel Gorsky Laboratoire d'Oce& nanographie Biologique et d'Ecologie du Plancton Marin, URA2077, Station Zoologique, B.P. 28, F-06230 Villefranche-sur-Mer, France Received 6 October 1997; received in revised form 5 March 1999; accepted 10 June 1999

Abstract Day/night variations in the size distribution of the particulate matter '0.15 mm (PM) were studied in May 1995 during the DYNAPROC time-series cruise in the northwestern Mediterranean Sea. Data on vertical distributions of PM ('0.15 mm) and zooplankton were collected with the Underwater Video Pro"ler (UVP). The comparisons of the UVP data with plankton net data and POC data from water bottles indicated that more than 97% of the particles detected by the UVP were non-living particles (0.15 mm) and that the PM contributed 4}34% of the total dry weight measured on GF/F "lters. Comparison of seven pairs of day and night vertical pro"les performed during the cruise showed that in the upper 800 m, the mean size and the volume of particles was higher at night than during the day. During the night, the integrated volume of the PM increased on average by 32$20%. This increase corresponded to a shift of smaller size classes ((0.5 mm) towards the larger ones ('0.5 mm). During the day, the pattern was reversed, and the quantity of PM '0.5 mm decreased. During the study period, the standing stock of PM (60}800 m) decreased from 7.5 to less than 2 g m~2 but the diel variations persisted, except for two short periods in the super"cial layer following a wind event. The cyclic feeding activity induced by the diel vertical migration of zooplankton could be the best candidate to explain the observed diel #uctuations in the size classes of PM in the water column. However, our results also suggest that in the upper layer additional driving forces such as the increase of the level of turbulence after a wind event or the modi"cation of the zoo- and phytoplankton community can in#uence the PM temporal evolution. ( 2000 Elsevier Science Ltd. All rights reserved. Keywords: Particulate matter; Diel variation; Large aggregates; Vertical distribution; Underwater Video Pro"ler; Mediterranean Sea

* Corresponding author. Tel.: 00-33-4-93-76-38-40; fax: 00-33-4-93-76-38-34. E-mail address: [email protected] (L. Stemmann) 0967-0637/00/$ - see front matter ( 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 7 - 0 6 3 7 ( 9 9 ) 0 0 1 0 0 - 4

506

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

1. Introduction Data on temporal distribution and #ux of particulate matter provide indication of the balance between production and removal processes in the water column (Bishop et al., 1986). Among the di!erent size classes of marine particles, particulate matter larger than 100 lm is considered to play an important role in the export of organic matter to the ocean interior (Fowler and Knauer, 1986). Moreover, the particle volume distribution indicates that a large part of the particle mass is found in the 0.1}3 mm size range (Jackson et al., 1997). Advances in our understanding of characteristics, abundance or temporal dynamics of large particulate matter was, until recently, limited by the low vertical resolution of the conventional methods such as water bottles, in situ pumps and sediment traps. These methods are suitable for chemical and biological analyses, or for mass measurements, but are not appropriate for abundance and size estimations. The characteristics, quantity and dynamics of marine snow have been reviewed by Alldredge and Silver (1988). They concluded that: (1) knowledge concerning the production and the breakdown of large aggregates is lacking, and (2) it is necessary to develop appropriate technologies for in situ measurement of their characteristics. Non-destructive optical methods may provide much better particle characterisation and resolution than the conventional sampling methods. Photographic equipment can produce several hundred images per deployment and video methods several thousand. Honjo et al. (1984) developed an in situ photographic technique for the study of large, millimetre-size particle distributions at meso- and bathy-pelagic depths. Asper et al. (1992), Gardner and Walsh (1990), Lampitt et al. (1993a) and Walsh and Gardner (1992) modi"ed the system and used it for vertical pro"ling of the water column or as a moored time lapse camera system. Eisma (1990) and Ratmeyer and Wefer (1996) improved the resolution of the image acquisition and studied particles from a few microns in size to hundreds of microns. Asper (1987) and Diercks and Asper (1997) used a video camera to "lm particles as they settled in sediment trap collectors. Gorsky et al. (1992,1999) and Davis et al. (1992) video recorded particles and zooplankton in the water column. Signi"cant diel variations of marine snow in the northeast Atlantic were observed by Lampitt et al. (1993b) with a time lapse camera moored during several months at 270 m depth. Late in the morning, marine snow concentrations were threefold higher than during the rest of the day. The authors suggested that the mid-water biota and its migratory behaviour was responsible for the observed diel variability. Gardner et al. (1995), using a beam transmissometer, observed diel variation in beam attenuation pro"les in the upper equatorial Paci"c waters. The beam attenuation data indicated that particles (20 lm in size increased by 70% during the day and decreased proportionally during the night. They observed regular day/night oscillations in the thickness of the mixed layer and concluded that the small sized particles produced within the surface mixed layer during the day may be diluted when the mixed layer expands during the night. Alternatively, the small particles may aggregate into larger ones that exceed the detection limit of the transmissometer.

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

507

Ruiz (1997) tested by model the three main processes considered to be possible causes for the daily cycles of marine aggregates observed by Lampitt et al. (1993b): (1) phytoplankton cell diel growth within the aggregate, (2) grazing by the migrating zooplankton and (3) diel turbulence in the mixed layer. In his model, neither the "rst nor the second process alone was able to generate diel variations. But, the third process, the daily cycle of turbulence, caused a clear cyclic behaviour of particulate matter. The author suggested that the diel variation in marine turbulence could lead to diel aggregation in the mixed layer and that the susbsequent downward export of the fast sinking aggregates would account for the diel variations observed at 270 m depth by Lampitt et al. (1993b). However, such a mechanism would create a periodic pulse of sedimentation of large particles and would result in the appearance of a structured signal in the vertical distribution of PM. This was not observed in the vertical pro"les of Honjo et al. (1984), Lampitt et al. (1993a) or Ratmeyer and Wefer (1993). Therefore, additional mechanisms have to be considered. In order to better de"ne the factors controlling the diel variation of the particulate matter, high-resolution data obtained on di!erent time and space scales are needed. In this paper we present data on diel variations in concentration and size of particulate matter ('0.15 mm) within the western Mediterranean waters, observed by the Underwater Video Pro"ler (Gorsky et al., 1992) during the May 1995 French-JGOFS DYNAPROC cruise. Our main objectives were to quantify these variations and to evaluate their potential origins.

2. Methods 2.1. Field sampling Data were collected during the DYNAPROC cruise (DYNAmics of rapid PROCesses in the water column) carried out from May 1st to June 1st, 1995, aboard the French R/V Suroit. The studied site was located at 43325@N, 07352@E in the Ligurian Sea. This station is 28 nautical miles o! Nice, France, and the water depth is 2300 m (Fig. 1). In the NW Mediterranean the continental shelf is narrow and the slope is steep (o! Nice, the 1000 m isobath is located only 5 nautical miles o!shore). The studied station has been sampled monthly since 1988 and is the site of a longterm survey program (Dynamic and Atmospheric Fluxes in the Mediterranean, DYFAMED), part of the France-JGOFS activities. The DYNAPROC cruise was composed of 4 successive legs, each one week long. During legs 1 and 3 the stability of the hydrologic structure was monitored (1) by continuous data acquisition of physical and chemical parameters along a coast open sea transect using the THES system (Prieur et al., 1993) and an Acoustic Doppler Current Pro"ler (ADCP RDI at 150 Khz) and (2) by CTD casts (SEABIRD 911) performed on a 16-station grid around the studied site. Water column processes were investigated during legs 2 and 4 and included deployments of drifting sediment traps, multiple plankton net tows, CTD-Rosette water sampler and the Underwater Video

508

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

Fig. 1. UVP sampling strategy during DYNAPROC cruise: *Table indicating the time of the UVP casts for particulate matter (PM). *Sampled station location in the Ligurian Sea. *Sampling grid in the water column (0}1000 m) for all the UVP casts during the cruise. The 16 continuous lines represent the pro"les made for zooplankton observations and the 24 dashed lines the pro"les for particle abundance and size measurements. Two casts (one during leg 2 and one during leg 4) did not achieve the programmed depth because of technical problems.

Pro"ler (UVP). More details about the sampling strategy are given in Andersen and Prieur (2000). 40 vertical pro"les were carried out with the UVP from May 10 to 16 and from May 26 to June 1 (legs 2 and 4). The UVP was deployed from the surface to 1000 m. 24 pro"les were made to assess the vertical distribution of particulate matter (PM) and 16 to determine the vertical distribution of zooplankton (Fig. 1).

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

509

2.2. The Underwater Video Proxler The Underwater Video Pro"ler (UVP, Gorsky et al., 1992) records and provides information on abundance, size and shape of objects larger than 0.15 mm situated in its "eld of view at a rate of 25 Hz. The image acquisition unit consists of: (1) a Hi-8 camcorder, (2) an electronic control unit with a data logger, (3) lead-acid, 12 v batteries, (4) two lighting systems, one with a delimited collimated light "eld of 0.28 l in front of the camera, the second composed of four waterproof spots (Birns Inc.) illuminating a volume of approx. 70 l of water for zooplankton recording and (5) a Seabird SBE-19 CTD. One 0}1000 m pro"le contains 25 000 images, which are processed by a system composed of: (1) a 512]512 pixel real time digitizer with 20 min storage capacity (YBL Inc.), (2) a Hi8 video tape recorder, (3) custom made image analysis software (FX Informatique, Inc.), (4) a PC with MATLAB (MATH WORKS Inc.) data analysis software. For the quantitative study of particles, the camera records objects illuminated in 1.5 cm thick volume of water (0.28 l). Since the lowering speed is 1 m s~1, one frame is recorded every 4 cm, and the successive images do not overlap. The recorded pro"le is digitised and can be automatically treated by the image analysis software. The zooplankton are recorded in a larger, non-structured, water volume (&70 l) illuminated by four watertight spots. The lowering speed is 0.5 m s~1, and each organism can be observed on several successive frames. Calibration experiments in a sea water tank using measured marine aggregates, organisms and calibrated particles showed that the UVP's detection limit was 100 lm in Equivalent Spherical Diameter (Stemmann, 1998). In this paper we present data on the diel variation of particles larger than 150 lm. 2.3. PM quantixcation Number of particles per frame, their surface and maximum length expressed in pixels were measured with our custom program and saved in an ASCII "le. These data were further processed by data analysis software developed in our laboratory. This software allows: (1) the merging of the CTD and UVP data, (2) conversion of particle surface and length to metric units using a calibration function obtained in the laboratory on a variety of natural particles, (3) estimation of particle Equivalent Spherical Diameter (ESD) and volume, (4) data management for further numerical or statistical analyses. Dry weight of particles is calculated using the equations of Alldredge and Gotschalk, (1988) obtained in Monterey Bay on aggregates sampled in the 0}40 m depth layer. These algorithms were recently con"rmed by Alldredge (1998) for particles of di!erent origin. 2.4. Zooplankton identixcation The large illuminated volume and the low vertical speed allowed the detection of organisms such as sarcodina, medusa, siphonophora, ctenophora, chaetognatha,

510

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

mollusca, crustacea, appendicularia, "sh and others. The determination of the taxonomic groups depended on the size of the detected organism. The con"guration of the UVP used during the DYNAPROC cruise allowed the determination of the higher taxonomic levels (order or family) only. The detailed UVP data on the macroplankton distribution will be presented elsewhere. In this paper we deal with the detection and enumeration of smaller organisms ((3 mm, mainly copepods). The estimations were based on the escape behaviour of organisms recorded on several successive frames. 2.5. Data analysis In a typical vertical pro"le more than 100,000 di!erent particles can be detected. In this work we analyzed every second frame (one image every 8 cm).The observed size range (from 0.15 mm to several mm) was divided into 28 ESD classes. We conducted one statistical and one numerical analysis, a two way ANOVA (Scherrer, 1984) and a hierarchical #exible clustering (Legendre, 1984), to quantify and test the signi"cance of the observed variability in the vertical distribution of particles during the DYNAPROC cruise. Sunset occurred at 18h30 at the beginning of the cruise and at 18h00 at the end of the cruise. The day UVP pro"les were done before sunset (between 15h30 and 17h00), and the night pro"les, which could not be taken consistently at the same hour, were performed after sunset (between 21h35 and 3h07 local time). 2.5.1. Vertical distribution Data from every second frame of the vertical pro"les were grouped to obtain a 5 m vertical resolution. Thus the sampling volume of each data point corresponded to 17.5 l. To test the signi"cance of the variations at di!erent time scales such as: (i) the time variability during the cruise (termed cruise variability hereafter) and (ii) the diel variability, we executed a two way ANOVA on the mean size and total volume of PM in the super"cial (60}100 m) and deep (800}840 m) layers (Table 1). To simplify the reading of this paper, we will use the term &size' instead of ESD. Normality and homoscedasticity were tested on raw data using respectively the Kolmogorov}Smirnov test for goodness of "t and the log-ANOVA test (Scherrer, 1984). 2.5.2. Size distribution To detect water layers with similar PM size distribution we determined the size distributions of particles and grouped them by hierarchical clustering. All the pro"les obtained during the cruise were divided vertically into 22 layers. They were 20 m thick in the super"cial 150 m layer (70 l sample volume) and 50 m thick below this depth (175 l sampling volume). Thus we could consider 506 layers, each with the corresponding size distribution determined with a minimum count of 1000 particles. As particles of a few millimetres were rare ((0.01%), all the particles '1.5 mm were summed in the last size class. The hierarchical #exible clustering was performed on a matrix of Kolmogorovdistances among the size distributions (Legendre, 1984). This distance, based on the absolute di!erences between two observed cumulative frequency distributions, has the advantage of a statistical test and is powerful in the detection of variation between two

0.10$0.04

0.11$0.02

Night

0.78$0.15

Night

Day

0.44$0.09

0.20$0.01

Night

Day

0.20$0.01

0.32$0.02

Night

Day

0.23$0.01

Day

0.11$0.02

0.06$0.01

0.36$0.05

0.33$0.08

0.21$/0.01

0.19$0.01

0.23$0.01

0.22$0.01

13/14 May

0.08$0.01

0.06$0.02

0.14$0.06

0.13$0.06

0.21$0.01

0.20$0.01

0.27$0.02

0.20$0.01

15/16 May

!No data, F "observed value for F, F "critical value for F. 4 # "n and n "degree of freedom. 1 2

Volume (800}840 m)

Volume (60}100 m)

ESD (800}840 m)

ESD (60}100 m)

12/13 May

0.11$0.02

0.11$0.04

0.25$0.10

0.15$0.05

0.20$0.01

0.19$0.01

0.24$0.02

0.20$0.01

27/28 May

0.11$0.02

0.07$0.01

0.17$0.05

0.16$0.05

0.20$0.01

0.19$0.01

0.20$0.01

0.21$0.01

28/29 May

0.54$0.11

0.07$0.04 0.10$0.04

! !

0.53$0.15

0.20$0.01

0.20$0.01

0.34$0.02

0.32$0.02

30/31 May

0.47$0.13

0.13$0.06

! !

0.32$0.02

0.22$0.02

29/30 May

Anova

n "11 1 F "0.69 4 n "24 2 F "3.26 #

n "13 1 F "1.12 4 n "28 2 F "2.99 #

n "11 1 F "0.97 4 n "24 2 F "3.26 #

n "13 1 F "1.49 4 n "28 2 F "2.99 #

Log test

Table 1 Mean and standard deviation of particle size (in mm) and volume (in ppm) in 2 layers (60}100 and 800}840 m depth) during the 7 day/night comparisons. Each value is calculated from 8 observations. The Log-Anova test critical and observed values for F are given in the last column" L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531 511

512

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

distributions (Scherrer, 1984). After the grouping, we performed the Chi square test to check if the distribution of identi"ed groups in space and time was similar during the day and night.

3. Results 3.1. Vertical abundance proxles Of the 24 abundance pro"les of particles obtained during legs 2 and 4, the seven day/night pairs are presented in Fig. 2. Three pairs of depth pro"les were obtained during the second leg (Fig. 2a). The "rst two daytime pro"les (UVP casts 8 and 11) show similar vertical distribution with a high concentration layer between 100 and 280 m reaching a maximum of 30 particles l~1. The third daytime pro"le (UVP cast 17) shows a similar trend but with lower abundance, especially in the mid-water layers (maximum of 15 particles l~1). The strongest di!erences between the day and night vertical pro"les during leg 2 were found for the "rst pair (UVP casts 8 and 9). The night pro"le (UVP cast 9) shows that the maximum abundance of particles is located 100 m deeper than in the corresponding day pro"le (UVP cast 8). Below 800 m, the concentration of particles was similar during the night and day in all the comparisons. Pro"les obtained during leg 4 (Fig. 2b) were less strati"ed than during leg 2 and they showed no peaks in the 60}300 m layer. In contrast to the pro"les of leg 2, maxima of concentration are observed deeper than 200 m depth with concentrations ranging between 6 and 18 particles l~1. In the more stable upper layers, the abundance ranged only from 6 to 12 particles l~1. The lowest concentrations were found after May 28 (UVP cast 28) until the end of the cruise. A day/night di!erence in the abundance is particularly marked in the second and third comparisons of leg 4 (UVP casts 27}28 and 32}33). Although the day and night PM abundance vertical distributions showed some di!erences during legs 2 and 4, no typical diel pattern could be inferred from the shape of these pro"les. 3.2. Vertical size proxles Vertical pro"les of day/night mean size are shown in Fig. 3. The mean size ranged from 0.17 to 0.40 mm. During the two legs, the general patterns were similar, with larger particles in the upper layers (&60}200 m) and smaller particles in the deeper layers. An exception was the cast 12 pro"les, obtained the night of 05/14, when the maximum size of particles was found in the 350}400 m deep layer. Day/night di!erences in particulate size were observed throughout the water column but were more pronounced in the upper layers. During the night, the mean size was up to 1.5 times larger than during the day in the 7 comparisons. During leg 4, after May 28 (UVP cast 27), particles remained large in the day pro"le, whereas during

Fig. 2. Vertical distribution of particulate matter abundance in numbers l~1. The plots represent the day (thin line) and night (bold line) pro"les during leg 2 (2a) and leg 4 (2b). Each data point represents the abundance measured in 17.5 l of water.

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531 513

Fig. 3. Vertical pro"les of mean size in millimetre calculated from the particle size distribution (ESD" Equivalent Spherical Diameter). The plots represent the day (thin line) and night (bold line) pro"les during leg 2 (3a) and leg 4 (3b).

514 L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

515

leg 2 we observed a reduction in size during the day. The smallest di!erences between day and night mean sizes were observed for pro"les 27 and 28. 3.3. ANOVA The ANOVA was performed on the data from the 7 pairs of day/night pro"les. In all cases the ANOVA tests indicated that the cruise and diel variations were signi"cant in terms of both mean size and total volume of PM (Table 2, p(0.05). The analysis of the probabilities of the null hypothesis of non-signi"cant e!ects of cruise and diel variations on the PM distribution (Table 2) indicates that (1) the levels of signi"cance were higher (lower probability) in the sub-surface layer (60}100 m) than in the deep layer (800}840 m) except in the case of particle volume cruise variation, (2) the diel variability of particle size was higher than the cruise variability in both Table 2 Results of the two way Anova from Table 1. Mean squares, degrees of freedom (df), F values and probabilities to accept the null hypothesis (P) are given for both factors (cruise and diel variations) Source of variation Size (60}100 m) Size (800}840 m) Volume (60}100 m) Volume (800}840 m)

Daily Diel Daily Diel Daily Diel Daily Diel

Mean square 0.032 0.0687 0.00026 0.00141 0.61 0.45 0.0052 0.00083

df

F

P

6 1 5 1 6 1 5 1

108.2 230.4 2.97 15.89 59.8 44.23 6.9 11.08

(0.0001 (0.0001 0.015 0.000 (0.0001 (0.0001 (0.0001 0.0012

Fig. 4. Particles size distribution (4a) and cumulative distribution (4b) from di!erent depths (every 20 m between 0 and 1000 m) at two stations (station 9 and 12). These distributions are representative of all the calculated distributions during the cruise (see text for details).

516

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

upper and deep strata, and (3) diel variability of particle volume was lower than the cruise variability at 60}100 m depth but not at 800}840 m depth. 3.4. Size distribution The size distributions and cumulative distributions calculated from 0 to 1000 m depth for UVP cast 09 and 12 are presented in Fig. 4. Their patterns are characteristic of all the 506 distributions analysed during the cruise (but not shown in Fig. 4 because of the large number of size distributions). The common feature in all cases was the general decrease in particle abundance with increasing size (Fig. 4a). The median size ranged from 0.15 to 0.34 mm (Fig. 4b). Classi"cation of the particle size distributions is shown in Fig. 5. At the Kolmogorov distance of 80, four groups can be isolated. The mean size cumulated distribution of the 4 groups are presented in Fig. 6a and b. The di!erence in the size distribution is mainly due to the increase of the frequency of particles '0.5 mm that contributed by 25% of the total abundance in group 1 (large particle distributions) and by 5% in group 4 (small particle distributions). The di!erence is even greater when the frequencies are expressed in terms of volume. Particles '0.5 mm correspond to more than 90% in group 1 (large particle distributions) and to less than 40% of the total volume in group 4 (small particle distributions). The spatial and temporal position of the four separated groups is given in Fig. 7. Large particles (group 1) are found in the upper layers (0}500 m) whereas small particles (group 4) occupy both the upper and the deeper layers. We also noted that during leg 2, the vertical extent of the large particles distribution was deeper than

Fig. 5. Dendrogram of the hierarchical #exible clustering (b"0.5) on a matrix of Kolmogorov distances between 506 cumulative size distributions. At a Kolmogorov distance of 80, four types of size distribution can be separated.

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

517

Fig. 6. Mean cumulative size distributions in terms of abundance (6a) and volume (6b) for the four groups isolated by hierarchical clustering (see Fig. 5). Group 4 (small particles) is represented by squares and group 1 (large particles) by circles.

Table 3 Occurrence of the four isolated groups by the clustering between 60 and 200 m. The hypothesis of identical occurrence of the four groups during day and night can be rejected as shown by the higher observed Chi-square value (X2obs) than the critical value at 5% (X2c)

Group Group Group Group Sum

1 2 3 4

Day

Night

Sum

3 9 12 14 38

14 5 8 7 34

17 14 20 21 72

df"3 X2obs"11.84 X2c at 5%"7.81

during leg 4. An important di!erence in the vertical distribution of the groups was observed between day and night in the upper layer. Particles were larger at night (dark grey, Fig. 7) in the upper layers (60}200 m) than during the day (light grey, Fig. 7). The Chi2 test of the frequency of occurrence of each group in the 60}200 m depth layer showed that the di!erences between day and night distributions of the groups were signi"cant (p(0.05, Table 3). 3.5. Particle abundance In order to estimate the evolution in time of PM, we calculated the mean abundance for two di!erent size classes (from 0.15 to 0.5 mm and '0.5 mm) in three layers (60}100 m, 200}400 m and 600}800 m). The results, shown in Fig. 8, exhibited lower particle abundance for both size-classes during leg 4 than during leg 2. The decrease of PM abundance was regular during leg 2 (except on May 13) in the three layers but with a maximum e!ect in the two upper layers. During leg 4, the two size classes

Fig. 7. Vertical and temporal location of the 4 groups isolated by clustering during the two legs. Note the diel variability in the upper layers (60}450 m) by the diel changing of grey levels (lighter at day and darker at night). The stars mark the pro"les used in Figs. 2 and 3 and used in the ANOVA test. The dashed rectangle at the top of each column indicates the night pro"les.

518 L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

519

Table 4 Ratio between night and day abundance for two size classes (ESD(0.5 and ESD'0.5 mm) during the cruise at three depth (60}100, 200}400 and 600}800 m) Water layer

12/13

13/14

15/16

27/28

28/29

29/30

30/31

min

max

mean

std

Particles (0.5 mm 60}100 m 0.47 200}400 m 1.32 600}800 m 1.22

0.67 0.85 1.21

0.23 0.81 0.70

0.56 1.05 1.19

1.10 1.48 0.45

0.85 1.72 no data

0.80 0.95 1.11

0.23 0.81 0.70

1.10 1.72 1.45

0.67 1.17 1.15

0.28 0.34 0.25

Particles '0.5 mm 60}100 m 1.57 200}400 m 1.36 600}800 m 1.31

0.93 1.49 1.63

1.05 1.20 1.08

1.26 2.42 1.81

1.15 1.97 1.58

1.64 5.94 no data

1.17 2.32 2.07

0.93 1.20 1.08

1.64 5.94 2.07

1.25 2.39 1.58

0.26 1.64 0.35

showed inverse trends at 60}100 m and 200}400 m depth: the large particle abundance increased while the small particle abundance decreased. The day/night variability was more pronounced for the larger particles ('0.5 mm) than for the small size class ((0.5 mm). 3.5.1. Upper layers (60}100 m) The size classes of small and large particles showed di!erent evolution with time. An opposite diel variability, with the large particle abundance increasing and small particle abundance decreasing at night, was distinct during the day/night comparison of the pro"les of May 12}13, 27}28, 28}29, 29}30, 30}31. The comparisons done May 13}14 and 16}17 show that the small particle abundance decreased while the concentration of large particles remained constant or increased. Abundance of large particles was on average 1.25 times higher at night than during the day, whereas small particle abundance was on average 1.5 times lower at night than during the day (Table 4). 3.5.2. Deeper layers (200}400 and 600}800 m) The day/night variation was observed during the two legs but was particularly marked during leg 4 for particles '0.5 mm at 200}400 m depth. In this layer particles '0.5 mm showed a strong diel variability while small particles did not show diel variation. Large particle ('0.5 mm) abundance was, on average, 2.4 (maximum 5.9) times higher at night than during the day between 200 and 400 m depth. Small particle concentrations increased on average by a factor of 1.17 during the night. Between 600 and 800 m depth the average night increase was a factor of 1.14 for particles (0.5 mm and 1.58 for particles '0.5 mm. 3.6. PM integrated volume in the 60}800 m layer Total particle volume, integrated from 60 to 800 m depth, decreased from 0.14}0.22 l m~2 at the beginning of leg 2 to 0.03}0.12 l m~2 during leg 4 (Fig. 9). The

520

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

Fig. 8. Mean abundance of particles and standard deviation for two size classes, ESD(0.5 mm (circles) and ESD'0.5 mm (squares) in three layers (60}100, 200}400, 600}800 m depth). Midnight is indicated by the dashed vertical lines.

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

521

Fig. 9. PM integrated (60}800 m) volume during the DYNAPROC cruise. Midnight is indicated by the dashed vertical lines. The two wind events are indicated by the arrows.

integrated volume of particles was on average 32$20% higher at night than during the day.

4. Discussion 4.1. General conditions at the studied site Previous studies have shown that the studied station is located in the central water of the Ligurian Sea and that it is separated from the coastal water by a permanent geostrophic front (Boucher et al., 1987; BeH thoux et al., 1988; Sournia et al., 1990). This frontal structure limits the extent of lateral input of small particles ((20 lm) into the central zone (Copin, 1988; Durrieu de Madron et al., 1990). It also acts as a barrier restraining the dispersal of coastal planktonic communities o!shore (Boucher et al., 1987; Pedrotti and Fenaux, 1992). ADCP data obtained during the DYNAPROC cruise by Andersen and Prieur (2000) have shown that in the studied area advective processes were weak. The time-average horizontal current was lower than 2 cm s~1 throughout the observational period, resulting in an eastward integrated displacement of 30 km in one month. The coast-open sea CTD transects performed during legs 1 and 3 showed that the station was located outside the frontal in#uence in the central zone of the Ligurian Sea. Four hydrological grid surveys, undertaken on May 1, 8, 18 and 24 (16 CTD casts around the sampled station), demonstrated that the water mass did not undergo signi"cant change during the period of study (Andersen and Prieur, 2000). Thus the variations in the distribution of PM observed during the cruise may be of local, open sea origin. The integrated phytoplankton and bacterial biomass, the primary production and the concentration of dissolved and particulate lipids decreased from the beginning to the end of the cruise (Andersen and Prieur, 2000; Vidussi and Marty, 2000; Goutx

522

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

et al., 2000). In the sediment traps, the C/N ratio increased and the lipid/POC ratio and the protein content decreased (Goutx et al., 2000). From May 10 to 31, we observed a decrease of the particle abundance and the fading away of the mid-water PM peak (Fig. 2a and b). The PM abundance found at the end of the cruise was similar to the summer abundance usually found in this area by the DYFAMED program between 1992 and 1996 (Stemmann, 1998). The entire data set demonstrates that the DYNAPROC cruise was carried out during the transition period between mesotrophic and oligotrophic situation. 4.2. Nature of detected particles We could not directly sample the recorded particles, and the indications concerning the origin of PM had to be deduced from indirect observations. In order to estimate the origin of PM and the contribution of living organisms to the concentration of the PM detected by the UVP, we compared (i) the PM calculated dry weight with particle data obtained by water "ltration and (ii) the number of particles with the concentration of zooplankton recorded by the UVP and with the concentration of zooplankton collected by nets. (i) PM estimated by the UVP and by xltration. Particulate carbon at 150 m depth was determined in samples from SBE 911 CTD/Rosette water bottles "ltered through Millipore GF/F "lters (J.C. Marty, personal communication). The UVP particulate carbon weight was estimated to be 20% of the dry weight calculated from the Alldredge and Gotschalk (1988) equation for marine snow. The mean PM calculated from the UVP data obtained between 100 and 200 m represented from 4 to 34% (mean 12%) of the "ltered carbon (Table 5). The highest ratio, found early (03 : 07 h) on May 14, was 2.5 times higher than the ratios found for the other days. Except for this value, the ratios observed are similar to those calculated for other oligotrophic areas (Stemmann, 1998). Taking into account that the porosity of a GF/F "lter is 0.7 lm, that the estimations of the UVP PM were calculated for the size class '0.15 mm and that the coe$cients used are characteristic of marine snow, lowdensity aggregates, particles larger than 0.15 mm constituted up to one-third of the particulate matter pool. (ii) Comparison between PM and zooplankton abundance. On the UVP records, small crustacean zooplankton ('0.5 mm) were detected because of their escape behaviour. The density of this zooplankton fraction was always lower than 0.1 ind. l~1, (Fig. 10). The zooplankton density was on average lower than 0.035 ind. l~1 in the 60}400 m layer and lower than 0.014 ind. l~1 in the deeper strata (Fig. 10). In contrast, the large particle ('0.5 mm) abundance maximum was of 5.4 particles l~1, with mean abundance of 1.3 particles l~1 between 60 and 400 m depth and 0.6 particles l~1 deeper (Fig. 8). These observations indicate that, in abundance, zooplankton of the same size range account for a maximum of only 2}3% of the objects detected by the UVP in the upper and deeper layers. Results obtained with plankton nets of various mesh were compared to the distribution of particles detected by the UVP. We have focused our comparisons on the copepods, because they form the major part of the catches in the size range

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

523

Table 5 Mean dry weight of PM ('150 lm) calculated for the 100}200 m layer (UVP data) and the dry weight of PM ('0.7 lm) "ltered at the depth of 150 m (J-C. Marty, personal communication) Date

UVP 100}200 m lg l~1

Filter 150 m lg l~1

UVP/Filter (%)

11/05/95 12/05/95 14/05/95 15/05/95 16/05/95 28/05/95 31/05/95

6.4 7.4 6.8 2.2 1.2 2.1 3.5

48 94 20 27 30 36 32

13.3 7.9 34.1 8.2 4.1 5.9 10.9

Fig. 10. Vertical distribution of zooplankton (nbr m~3) detected by the UVP in three day pro"les (a) and three night pro"les (b). Station number, date and hour of the pro"les are given in the legend.

detected by the UVP. The results obtained during the DYNAPROC cruise by a modi"ed Longhurst-Hardy Plankton Recorder equipped with a 0.2 mm mesh (Nival, personal communication) showed that the concentrations of copepods in the upper 120 m ranged from 0.08 to 0.6 ind. l~1. For particles '0.2 mm, concentrations recorded by the UVP ranged from 5 to 10 particles l~1 in the same depth layer. At the studied site in June 1994, Gasser (1996), using a Bioness multiple closing net with 0.5 mm mesh size, found concentrations of copepods ranging from 0.01 in deep layers ('200 m) to 0.05 ind. l~1 in the upper 100 m layer. During the DYNAPROC cruise, we found PM ('0.5 mm) concentrations ranging from 0.8$0.6 in the deep layers to 1.6$1.3 particles l~1 in the upper layer, more than 100 times the copepod abundance estimated by nets with a 0.5 mm mesh size. These comparisons indicate that the values estimated by the UVP and by the net tows are similar and that more than 97% of the objects detected by the UVP are non-living particles.

524

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

4.3. Diel PM variability The observed diel variability of vertical pro"les was not in#uenced by modi"cation of the particles' optical properties in changing light regimes during the daytime, since our data analysis started at depth below 60 m for all the pro"les. Below this depth, the measured daytime background light value of the frames remained identical to that of the nighttime pro"les or to that of deep layers. The ANOVA test demonstrated that the day/night variations a!ected signi"cantly the PM size and volume not only in the sub-super"cial layers but also as deep as 800 m (Table 2). The mean size showed more variability during one diel cycle than during the whole cruise in both upper deep layers. In the 60}100 m layer particle volume varied more during the cruise than during one diel cycle. Inversely in the 800}840 m layer, the volume varied more during one diel cycle than during the cruise. This suggests that during the studied period the water column diel PM variability contributed to a major part of the PM total variability. The diel variation observed by the UVP resulted in a nighttime increase of the integrated PM volume of 32$20% (Fig. 9). This increase can be explained by aggregation of size classes below the limit of detection of the UVP ((0.15 mm) into detectable size classes. The contribution of particles larger than 0.5 mm to the total volume was smaller than 40% during the day but exceeded 90% at night. On average, the large particles were 1.25 times more abundant and the small particles 1.5 times less abundant at night than during the day (Table 4). In the upper layers (60}100 m), this shift toward the large particles was associated with a diminution of the abundance of particles (0.5 mm. This inverse relationship between the two size classes was particularly marked during leg 4 at 60}100 m depth (Fig. 8). In the deep layers, the increase in mean size was linked mostly to the diel variation of particles larger than 0.5 mm. The night increase of PM '0.5 mm was more constant in the two deeper layers than in the upper layer, where this increase was not observed during the nights of May 13}14 and 15}16. These observations suggest that the increase in size of particles in upper and deeper layers is based on di!erent processes. These diel variations involved a larger stock of particles during leg 2 and a smaller one during leg 4, when the system became more oligotrophic. The intensity of diel #uctuations (expressed as the ratio between night and day abundance values), although variable, showed no consistent trend during the cruise (Table 4). In a situation where lateral advection is not signi"cant, as shown for the studied period, all these observations suggest that (1) a substantial part of the particulate volume in the sea undergoes signi"cant variation in size within a period of one day, (2) the processes leading to this diel variation are successive, complementary and highly active throughout the water column. 4.4. Potential origin of diel variations Considering that the sampled station was not in#uenced by the frontal activity and that the horizontal advection was weak, we can suppose that the PM diel variation is

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

525

controlled by local processes. Although 7 day/night comparisons are not su$cient to demonstrate a regular cycle, the repetition of the same pattern observed during the cruise supports the hypothesis that the diel variation of the PM vertical distribution in the water column is a cyclic process. Di!erent physical or biological mechanisms may be responsible for cyclic variations in aggregation or removal of the particulate organic matter from a given layer. 4.4.1. Physical mechanisms Gardner et al. (1995) observed regular day/night oscillations in the load of submicrometer particles in the mixed layer and concluded that the small sized particles produced within the surface mixed layer may be diluted when the mixed layer expands during the night or that the small particles could be aggregated into larger ones not detected by the transmissometer. A more recent study suggested that diel #uctuations of the turbulence, caused by day/night di!erences in the heating of the upper water by solar radiation, could cause a cyclic behaviour of particulate matter by modifying their encounter rate (Ruiz, 1997). However, the observed deep diel variations cannot be exclusively the signature of the downward export of PM previously aggregated in the upper layers, since such a mechanism would imply the detection of intermediate layers of particles with di!erent characteristics. We did not observe such a vertical distribution during the DYNAPROC cruise. Moreover, the sinking speed of such particles aggregated in the upper layers would exceed the fastest known sedimentation rate. The sinking speed required for a large particle produced at the beginning of the dark period at 8}9 p.m. to reach 800 m depth at midnight would be approximately 4800}6000 m day~1. Such values have never been observed or predicted. Maximum speeds found in the literature are 2470$990 m day~1 for the fecal pellets of salps (Caron et al., 1989). Since during the DYNAPROC cruise the production of large particles was not restricted to the upper layers, other processes able to modify the particulate pool in the water column have to be considered. 4.4.2. Biological mechanisms Lampitt et al. (1993b) observed signi"cant diel variations of marine snow in the northeast Atlantic at 270 m depth with three-fold higher volume concentrations late in the morning relative to other times of the day. They suggested that marine snow at this depth was supplied from a small subset of the upper ocean marine snow diel production with a high and uniform sinking rate. The authors suggested that the mid-water biota and their migratory behavior were responsible for the observed diel variability in the PM distribution. In contrast to Lampitt et al. (1993b), we found higher mean size and higher total volume of particles at night. It is possible that this di!erence arises from di!erent sampling strategies. The system used by Lampitt et al. (1993b) was mounted on a long-term mooring. Floating or moored objects are known to attract plankton and nekton. For example, the problem of swimmers and their fecal production in sediment traps or "sh catches under #oating objects are well known (Seiler and Brandt, 1997).

526

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

Moreover, the concentrations of small "sh under the #oating objects are higher during the day than during the night (Massuti and Renones, 1994; Tanaka and Oozeki, 1996). According to our results, at night the abundance of the zooplankton was higher in the super"cial layer between 40 and 120 m with a mean of 39 organisms m~3 than between 200 and 400 m depth (20 organisms m~3). During the day the trend was inverse with a mean of 28 organisms m~3 in the super"cial 40}120 m layer and with a mean of 35 organisms m~3 between 200 and 400 m (Fig. 10). This pattern could be caused by the higher concentration of migrating organisms in the upper layers at night and their location in the deeper layers during daytime. If this migrating behaviour is similar in the North Atlantic, then the daytime biological activity at the depth of the moored camera of Lampitt et al. (1993b) could be high and the results could re#ect a local activity rather than a general trend in the water column. During the DYNAPROC cruise, the UVP was lowered at a speed of 1 m s~1, a speed that the recorded zooplanktonic organisms could attain only for short distances. As the plankton and nekton diel vertical migration innvolves a variety of organisms with di!erent feeding behaviours, with a wide range of vertical distributions between 100 and 1000 m depth (Mackas and Bohrer, 1976; Pearre, 1973; Roe, H.S.J., 1984; Simard et al., 1985; Simard et al., 1986) and with di!erent timing of migration (Plueddemann and Pinkel, 1989; Baussant, 1993), it may a!ect PM distribution in all the water column and not only in the upper layers. Moreover, the upward and downward migration enhances the feeding activity of mid-water non-migratory populations (Roe and Badcock, 1984; Yoon, 1995) and can also a!ect the pool of detritic particles at di!erent depths. These observations indicate that the diel vertical migration is the best candidate to explain the diel transformation of PM observed during the cruise well below the mixed layer. 4.5. Role of the diel vertical migration In the Ligurian sea, previous studies on zooplankton vertical distribution showed extended migration of di!erent herbivorous, omnivorous and carnivorous species (Andersen and Sardou, 1992; Andersen et al., 1992; Sardou and Andersen, 1993; Gasser, 1996; Baussant, 1993; Yoon, 1995). During the DYNAPROC cruise, diel vertical migration of zooplankton was suggested by higher zooplankton biomass measured at night than during the day (Andersen and Prieur, 2000). Sediment traps deployed at 200 m depth with 4-h collecting period showed a diel signal of biotracers (POC, N-acetylated glucosamine, protein and chloroplast lipids), suggesting that night feeding activity was the process responsible for sedimentation of fresh material (Goutx et al., 2000). It is thus possible that in the upper layers, the diel vertical migration of zooplankton and the associated increase of feeding rates in#uenced the rate of production of large particles (feces and other aggregates) and was responsible for the observed day/night variations. In order to evaluate this impact in the 60}100 m layer, we calculated the fecal pellet production of copepods (larger than 1 mm) able to produce fecal pellets '0.15 mm. For this, we used the density of copepods reported by Gasser (1996) for the same

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

527

location during a previous cruise. He found a density of 10}50 copepods m~3 in the upper 100 m. Assuming (1) that the migrators occupy the 60}100 m layer, (2) that the fecal pellet production corresponds to the values published in the literature (40}80 pellets ind.~1 day~1, Butler and Dam, 1994) and (3) that one fourth of the pellets were produced in the time lag between 6h00 pm and midnight, the migration of 1 to 12.5 (10 to 25% of day copepod abundance) copepods per m~3 would increase particle abundance by 0.01 to 0.25 pellets l~1. During the cruise the nighttime increase of particles '0.5 mm ranged from 0.3 to 2.1 particles per litre. Thus the copepod fecal production alone could account for a considerable part (up to 80% in some cases) of the increase of particles '0.5 mm in the upper layers (60}100 m). 4.6. Perturbations in the PM diel cycle The regular diel variation was altered during two periods. The "rst followed a strong wind burst during the night of May 12}13 (maximum wind '16 m s~1). This wind event induced a strong mixing of the upper 10 m and a vertical advective event by Ekman pumping of deeper and cooler water (Andersen and Prieur, 2000). After the wind event most of the parameters indicating a downward #ux of organic matter increased, and two nights later (May 14}15) the #uxes measured in the drifting sediment traps were the highest (Goutx et al., 2000). On May 15, the POC measured in sediment traps decreased by 50% and remained low until the end of leg 2. No diel variations were observed in the traps on May 15 and 16 (Goutx et al., 2000). The 0}200 m integrated phytoplankton biomass decreased from approx. 35 mg m~2 on May 13 to 14 mg m~2 on May 16 (Andersen and Prieur, 2000). The PM 60}800 m integrated volume estimated by the UVP decreased from 0.19 ml m~2 the night of May 13}14 to 0.07 ml m~2 the night of May 15}16 (Fig. 9). In the 60}100 m layer, PM concentrations also showed a decreasing trend after the wind event, but the following nights (13}14 and 15}16; we have no data from the night of May 14}15) the abundance of particles '0.5 mm did not increase as observed during the other nights (Fig. 8). In contrast, in the 200}400 m layer the concentration of particles ('0.5 mm) increased during both nights with a maximum e!ect the night of May 13}14 (Fig. 8). It is possible that the modi"cation of the PM diel variation after the wind event could result from the depth dependent modi"cation of the aggregation-disaggregation processes. For example, after the wind event in the 60}100 m layer the enhanced turbulence due to the wind event could enhance the aggregation of zooplankton and the grazing on phytoplankton (Andersen and Prieur, 2000). The nighttime aggregation of PM and its #ux to deeper layers (200}400 m) may be controlled by the zooplankton diel migration. This may explain the highest ratio calculated between the PM'0.15 mm observed by the UVP and the PM'0.7 lm obtained by "ltration on the night of May 13}14 (Table 5). The same night, PM #ux measured at 200 m depth in sediment traps was also high (Goutx et al., 2000). The lack of PM'0.5 mm diel variations in both layers and in the sediment trap data the night of May 15}16 could be due to the consequent impoverishment of the water column and thus the reduction of nighttime aggregation.

528

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

The second wind event (wind speed up to 13 m s~1), which took place on May 26, induced a mixing process that could increase the standing stock of the phytoplankton at the beginning of leg 4, as described in Andersen and Prieur (2000). The cloudy weather and the subsequent decrease in the super"cial primary production together with the strong grazing pressure (Goutx et al., 2000) could drastically decrease the small particle standing stock and lead to the low PM concentrations observed during Leg 4 (Fig. 8). On May 29, the concentration of particles '0.5 mm remained high in the upper 60}100 m layer, although the diel variability persisted (Fig. 8). This pattern seemed to characterise the 60}100 m layer but not the deeper strata, where the two size classes co-varied. However, from May 29 to 30, we observed a loss of 75% of the small particles in the 200}400 layer and 65% in the 600}800 m layer. The origin of this feature can be related to the reinforcing of the oligotrophic regime, to the shift toward the pico- and nanoplanktonic assemblages (Vidussi et al., 2000) and subsequently to the change in the zooplankton community (Andersen, personal communication). However, additional data are needed to corroborate this hypothesis.

5. Conclusion The DYNAPROC cruise was carried out in the central Ligurian Sea during the transition period from mesotrophic to oligotrophic regime. The entire data set demonstrated that the water mass did not undergo signi"cant change during the period of study, suggesting that the variations in the distribution of PM observed during the cruise are of local, open sea origin. Our results indicate that, in the 60}800 m water column, the size distribution of particulate matter undergoes signi"cant diel changes. During the night, production of PM larger than 0.5 mm prevailed, resulting in an upward shift in size, but during the day, the aggregated particles were removed or disaggregated, resulting in a downward shift in size of particles. Having considered di!erent physical and biological processes, we suggest that the nocturnal aggregation in the water column is induced mostly by the diel variation in zooplankton feeding activity related to their diel vertical migration. The short-term modi"cations in PM diel variations after a wind event might be linked to the more intense aggregation processes induced by greater turbulence in the mixed layer. Our data support the hypothesis that PM dynamics result from a tight coupling between the living and the non-living particulate pool on short time scales (hours-days) that should not be overlooked in models of the particle dynamics in the sea.

Acknowledgements This research was a part of L.S.'s Ph.D. thesis undertaken in the framework of the CNRS-INSU French JGOFS program and of the Mediterranean Targeted Project

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

529

II-MATER (MTP II-MATER). We acknowledge the support of the European Commission's Marine Science and Technology (MAST) Programme under contract MAS3-CT96-0051. This is contribution number MTP II-MATER/037. We thank the chief scientist, V. Andersen, for organising the cruise and the captain and the crew of R/V Suroit for their e$cient assistance at sea.

References Alldredge, A.L., Gotschalk, C., 1988. In situ settling behavior of marine snow. Limnology and Oceanography 33 (3), 339}351. Alldredge, A.L., Silver, M.W., 1988. Characteristics, Dynamics and Signi"cance of Marine Snow. Progress in Oceanography 20, 41}82. Alldredge, A., 1998. The carbon, nitrogen and mass content of marine snow as a function of aggregate size. Deep-Sea Research I 45, 529}541. Andersen, V., Prieur, L., 2000. One-month study in the open NW Mediterranean Sea (DYNAPROC experiment, May 1995): Overview of the hydrogeochemical structures and e!ects of wind events. Deep-Sea Research I 47, 397}422. Andersen, V., Sardou, J., 1992. The diel migrations and vertical distributions of zooplankton and micronekton in the North-western Mediterranean Sea. 1. Euphausiids, mysids, decapods and "shes. Journal of Plankton Research 14 (8), 1129}1154. Andersen, V., Sardou, J., Nival, P., 1992. The diel migrations and vertical distributions of zooplankton and micronekton in the North-western Mediterranean Sea. 2. Siphonophores, hydromedusae and pyrosomids. Journal of Plankton Research 14 (8), 1155}1169. Asper, V.L., 1987. Measuring the #ux and sinking speed of marine snow aggregates. Deep-Sea Research 39, 939}952. Asper, V.L., Honjo, S., Orsi, T.H., 1992. Distribution and transport of marine snow in the Panama Basin. Deep-Sea Research 39 (6), 939}952. Baussant, T., 1993. Analyse spatio-temporelle du macroplancton et du micronecton observeH s par eH chosondage en mer Ligure (MeH diterraneH e Nord-Occidentale). The`se de Doctorat UniversiteH Pierre et Marie Curie Paris VI Paris 165 pp. BeH thoux, J.P., Prieur, L., Bong, J.H., 1988. Le courant Ligure au large de Nice. In: Minas, H.J., Nival, P. (Eds.), OceH anographie peH lagique MeH diterraneH ene. Oceanologica Acta 9, 59}67. Bishop, J.K.B., Conte, M.H., Wiebe, P.H., Roman, M.R., Langdon, C., 1986. Particulate matter production and consumption in deep mixed layers: observations in a warm-core ring. Deep-Sea Research 33 (11/12), 1813}1841. Boucher, J., Ibanez, F., Prieur, L., 1987. Daily and seasonal variations in the spatial distribution of zooplankton populations in relation to the physical structure in the Ligurian front. Journal of Marine Research 45 (1), 133}173. Butler, M., Dam, H.G., 1994. Production rates and characteristics of fecal pellets of the copepod Acartia tonsa under simulated phytoplankton bloom conditions: Implications for vertical #uxes. Marine Ecological Progress Series 114 (1}2), 81}91. Caron, D.A., Madin, L.P., Cole, J.J., 1989. Composition and degradation of salp fecal pellets: implication for vertical #ux in oceanic environments. Journal of Marine Research 47, 829}850. Copin-Montegut, C., 1988. Major elements of suspended particulate matter from the western Mediterranean Sea. OceH anographie PeH lagique MeH diterraneH ene. Minas, H.J., Nival, P., Paris, Oceanologica Acta. Special issue 9, 95}102. Davis, C.S., Gallager, S.M., Solow, A.R., 1992. Microaggregation of Oceanic Plankton Observed by Towed Video Microscopy. Science 257, 230}232. Diercks, A.R., Asper, V.L., 1997. In situ settling speeds of marine snow aggregates below the mixed layer: Black Sea and Gulf of Mexico. Deep-Sea Research I 44 (3), 385}398.

530

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

Durrieu de Madron, X., Ny!eler, F., Godet, C.H., 1990. Hydrographic structure and nepheloid spatial distribution in the Gulf of Lions continental margin. Continental Shelf Research 10 (9}11), 915}929. Eisma, D., Schuhmacher, T., Boekel, H., VanHeerwaarden, J., Franken, H., Laan, M., Vaars, A., Eijgenraam, F., Kalf, J., 1990. A camera and image-analysis system for in situ observation of #ocs in natural waters. Netherlands Journal of Sea Research 27 (1), 43}56. Fowler, S.W., Knauer, G.A., 1986. Role of large particles in the transport of elements and organic compounds through the entire water column. Progress in Oceanography 16, 147}194. Gardner, W.D., Chung, S.P., Richardson, M.J., Walsh, I.D., 1995. The oceanic mixed-layer-pump. Deep-Sea Research II 42 (1}2), 757}775. Gardner, W.D., Walsh, I.D., 1990. Distribution of macroaggregates and "ne-grained particles across a continental margin and their potential role in #uxes. Deep-Sea Research 37 (3), 401}411. Gasser, B., 1996. La structuration verticale des communauteH s meH sozooplanctoniques a` travers des fronts hydrodynamiques. The`se de Doctorat en OceH anologie. Pierre et Marie Curie, Paris VI Paris. Gorsky, G., Aldorf, C., Kage, M., Picheral, M., Garcia, Y., Favole, J., 1992. Vertical distribution of aggregates determined by a new underwater video pro"ler. Annales de l'Institut OceH anographique 68 (1}2), 275}280. Gorsky, G., Picheral, M., Stemmann, L., 1999. Use of the Underwater Video Pro"ler for the study of Aggregate Dynamics in the North Mediterranean. Estuarine, Coastal and Shelf Science, in press. Goutx, M., Momziko!, A., Striby, L., Andersen, V., Marty, J.C., Vescovali, I., 2000. High frequency #uxes of labile compounds in the central Ligurian Sea, northwestern Mediterranean. Deep-Sea Research I 47, 535}558. Honjo, S., Doherty, K.W., Yogesh, C., Agrawal, Y.C., Asper, V.L., 1984. Direct optical assessment of large anomorphous aggregates (marine snow. in the deep ocean). Deep-Sea Research 31, 67}76. Jackson, G.A., Ma$one, R., Costello, D.K., Alldredge, A., Logan, B.E., Dam, H.G., 1997. Particle size spectra between 1 lm and 1 cm at Monterey Bay determined using multiple instruments. Deep-Sea Research 44 (11), 1739}1767. Lampitt, R.S., Wishner, K.F., Turley, C.M., Angel, M.V., 1993a. Marine snow studies in the Northeast Atlantic Ocean: Distribution, composition and role as a food source for migrating plankton. Marine Biology 116 (4), 689}702. Lampitt, R.S., Hillier, W.R., Challenor, P.G., 1993b. Seasonal and diel variation in the open ocean concentration of marine snow aggregates. Nature 362, 737}739. Legendre, L., Legendre, P., 1984. Ecologie numeH rique 1. Le traitement multiple des donneH es eH cologiques. P. d. l. U. d. QueH bec, QueH bec, Masson 1 (2), 260. Massuti, E., Renones, 1994. Observaciones sobre la comunidad de peces pelagicos asociados a objetos #otantes en aguas oceanicas de Mallorca. Boletin del Instituto Espanol de Oceanogra"a. Madrid 10 (1), 81}93. Mackas, D., Bohrer, R., 1976. The biology of Thysanoessa raschii (M Sars) with a comparison of its diet with that of Meganyctifaness norvegica (M Sars), in: Barnes, H. (Ed.) Some Contempary Studies in Marine Sciences. Goerges Allen and Unwin, London, 1966. Pearre, S., 1973. Vertical migration and feeding in Sagitta elegans Verril. Ecology 54, 300}314. Pedrotti, M.L., Fenaux, L., 1992. Dispersal of echinoderm larvae in a geographical area marked by upwelling (Ligurian Sea, NW Medietrranean). Marine Ecology Progress Series 86, 217}227. Plueddemann, A., Pinkel, J.R., 1989. Characterization of the patterns of diel migration using a Doppler sonar. Deep-Sea Research 36 (4), 509}530. Prieur, L., Montegut, Copin C., Claustre, H., 1993. Biophysical aspects of `ALMOFRONT-1a, an intensive study of a geostrophic frontal jet. Annales de l'Institut OceH anographique 69 (1), 71}86. Ratmeyer, V., Wefer, G., 1996. A high resolution camera system (ParCa) for imaging particles in the ocean: System design and results from pro"les and a three-month deployment. Journal of Marine Research 54, 589}603. Roe, H.S.J., 1984. The diel migrations and distributions within a mesopelagic community in the North East Atlantic. 4. The copepods. Progress in Oceanography 13, 353}388. Roe, H.S.J., Badcock, J., 1984. The diel migrations and distributions within a mesopelagic community in the North East Atlantic. 5. Vertical migrations and feeding of "sh. Progress in Oceanography 13, 389}424.

L. Stemmann et al. / Deep-Sea Research I 47 (2000) 505}531

531

Ruiz, J., 1997. What generates daily cycles of marine snow? Deep-Sea Research I 44 (7). Sardou, J., Andersen, V., 1993. Micronecton and macroplancton in the Ligurian Sea (Mediterranean Sea): Diel migrations and vertical distributions. Oceanologica Acta 16 (4), 381}392. Scherrer, B., 1984. Biostatistique. G. M. eH diteur, Canada Boucherville 1 (1), p. 850. Seiler, S., Brandt, A., 1997. Seasonal occurence of planktonik Crustacea in sediment trap samples at three depth horizons in the Greenland Sea. Polar Biology 17 (4), 337}349. Simard, Y., Lacroix, G., Legendre, L., 1985. In situ twilight grazing rythm during diel vertical migrations of a scattering layer of Calanus xnmarchicus. Limnology and Oceanography 30, 598}606. Simard, Y., Lacroix, G., Legendre, L., 1986. Diel vertical migrations and nocturnal feeding of a dense coastal krill scattering layer (Thysnoessa raschi and Meganyctiphanes norvegica) in strati"ed surface waters. Marine Biology 91, 93}105. Sournia, A., Brylinski, J.M., Dallot, S., LeCorre, P., Leveau, M., Prieur, L., Froget, C., 1990. Fronts hydrologiques au large des co( tes franc7 aises: Les sites-ateliers du programme Frontal. Oceanologica Acta 13 (4), 413}438. Stemmann, L., 1998. Particulate matter spatio-temporal analysis using a new video system, in the northwestern Mediterranean Sea. In#uence of biological production, terrigeneous inputs and hydro-dynamical forcings. Doctorat de l'UniversiteH OceH anologie Biologie. PARIS VI Paris 180 Tanaka, Y., Oozeki, 1996. Where are the eggs of the Paci"c saury, Cololabis saira?. Ichthyological Research 43 (3), 329}333. Vidussi, F., Marty, J.C., Chiaverini, J., 2000. Phytoplankton pigment variations during the transition from spring bloom to oligotrophy in the northwestern Mediterranean Sea. Deep-Sea Research, this issue. Walsh, I.D., Gardner, W.D., 1992. A comparison of aggregates pro"les with sediment trap #uxes. Deep-Sea Research 39 (11/12), 1817}1834. Yoon, W.D., 1995. Contribution des Organismes Macrozooplanctoniques au #ux de la matie`re organique. The`se de Doctorat OceH anologie. UniversiteH Pierre et Marie Curie, Paris VI Paris 137 pp.