Time series of in-situ particle properties and sediment trap fluxes in the coastal upwelling filament off Cape Blanc, Mauritania

Progress in Oceanography 137 (2015) 1–11

Contents lists available at ScienceDirect

Progress in Oceanography journal homepage: www.elsevier.com/locate/pocean

Time series of in-situ particle properties and sediment trap fluxes in the coastal upwelling filament off Cape Blanc, Mauritania N. Nowald a,⇑,1, M.H. Iversen a,b,c,1, G. Fischer a,b, V. Ratmeyer a, G. Wefer a a

MARUM – Center for Marine Environmental Sciences, University of Bremen, Leobener Str., 28359 Bremen, Germany Geosciences Department, University of Bremen, Klagenfurter Str., 28359 Bremen, Germany c Helmholtz Young Investigator Group SEAPUMP, Alfred Wegener Institute for Polar and Marine Research, Am Handelshafen 12, 27570 Bremerhaven, Germany b

a r t i c l e

i n f o

Article history: Received 9 May 2014 Received in revised form 10 December 2014 Accepted 22 December 2014 Available online 31 December 2014

a b s t r a c t We compared particle data from a moored video camera system with sediment trap derived fluxes at 1100 m depth in the highly dynamic coastal upwelling system off Cape Blanc, Mauritania. Between spring 2008 and winter 2010 the trap collected settling particles in 9-day intervals, while the camera recorded in-situ particle abundance and size-distribution every third day. Particle fluxes were highly variable (40–1200 mg m2 d1) and followed distinct seasonal patterns with peaks during spring, summer and fall. The particle flux patterns from the sediment traps correlated to the total particle volume captured by the video camera, which ranged from1 to 22 mm3 l1. The measured increase in total particle volume during periods of high mass flux appeared to be better related to increases in the particle concentrations, rather than to increased average particle size. We observed events that had similar particle fluxes, but showed clear differences in particle abundance and size-distribution, and vice versa. Such observations can only be explained by shifts in the composition of the settling material, with changes both in particle density and chemical composition. For example, the input of wind-blown dust from the Sahara during September 2009 led to the formation of high numbers of comparably small particles in the water column. This suggests that, besides seasonal changes, the composition of marine particles in one region underlies episodical changes. The time between the appearance of high dust concentrations in the atmosphere and the increase lithogenic flux in the 1100 m deep trap suggested an average settling rate of 200 m d1, indicating a close and fast coupling between dust input and sedimentation of the material. Ó 2015 Elsevier Ltd. All rights reserved.

Introduction The export of organic matter from the surface to the deep ocean plays a key role in the global carbon cycle and is a crucial process for the sequestration of atmospheric carbon dioxide. This occurs by the conversion of carbon dioxide into organic matter via phytoplankton growth in the sunlit surface waters. The organic matter may be removed from the surface ocean if the phytoplankton form settling aggregates. This process is known as the ‘‘Biological Pump’’ (Broecker, 1982) and its efficiency is mainly a function of the remineralisation rate and sinking velocity of the organic matter. The export of particulate organic carbon typically occurs via settling of large marine particles, like marine snow (aggregates >500 lm) or fecal pellets, which are formed from a large variety of organic and inorganic particles in the upper ocean. Particle mass ⇑ Corresponding author. Tel.: +49 421 218 65612; fax: +49 421 218 65605. E-mail addresses: [email protected] (N. Nowald), [email protected] (M.H. Iversen), gfi[email protected] (G. Fischer), [email protected] (V. Ratmeyer), [email protected] (G. Wefer). 1 Co-first authors. http://dx.doi.org/10.1016/j.pocean.2014.12.015 0079-6611/Ó 2015 Elsevier Ltd. All rights reserved.

fluxes in the deep ocean are mostly quantified by the application of sediment traps, which collect sinking material over a known time interval and area. However, the sample resolution of deep ocean sediment trap collections is rather low (days to weeks to months) and does not determine which particles determine the mass flux (Lampitt et al., 1993). First, because this method integrates particle sizes and numbers and second, because individual aggregate structures are not preserved in the sample cups due to their fragile nature. The latter makes the collection and study of individual aggregates extremely difficult and was only achieved at the ocean surface using Scuba techniques (e.g. Alldredge and Gotschalk, 1988). Samples of single aggregates below Scuba depths are rarely available despite the improvements of marine technology during the past 30 years. The development of non-destructive methods, like vertically profiling camera systems, provided valuable information about the in-situ abundance and size distribution of marine particulate matter through the entire water column (Honjo et al., 1984; Asper, 1987; Ratmeyer and Wefer, 1996; Gorsky et al., 1992; Lampitt et al., 1993). Most camera profiles show a dramatic decrease in the particle abundance within the upper few hundred

2

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11

meters of the water column, indicating that the processes determining the amount of material that is being exported, already take place in the upper ocean (Iversen et al., 2010; Stemmann et al., 2004; Jackson and Checkley, 2011). In contrast to sediment traps, in-situ cameras can directly detect intrusions of laterally advected material into distinct depth layers (Fischer et al., 2009) and the resuspension of particles above the seafloor (Walsh and Gardner, 1992; Karakas et al., 2006). However, vertical profiles must be regarded as snapshots, because they only measure the actual particle distribution over depth at a certain time point at a certain place. Furthermore, data from profiling systems provide no information about composition, mass and settling velocities of individual particles. Although Guidi et al. (2008) and Iversen et al. (2010) were able to accurately estimate particle fluxes from high resolution camera profiles, these calculations are still based on certain assumptions regarding size-specific mass and settling velocities of the particles due to a general lack of knowledge about individual particle properties. Time series of paired measurements of optical systems and sediment traps combine the advantages of both methods. However, such studies are rare due to the fact that commercial cameras for long term deployments do not exist and most systems described in the literature are prototypes developed by certain institutes and are no standard study tools. Lampitt et al. (1993), for instance, deployed a camera system at a mooring depth of 270 m in the Porcupine Abyssal Plain for a period of 5 months. The system took pictures every 8.5 min in order to study the seasonal and the daily variations of marine particles. Ratmeyer and Wefer (1996) were able to compare particle abundance and size, acquired by a still image camera, with particle fluxes over three months at a location north of the Canary Islands. Their system was programmed in 4.3 day intervals and allowed to correlate one flux peak with an increase in total particle volume. Because the studies by Lampitt et al. (1993) and Ratmeyer and Wefer (1996) proved that optical systems are able to record flux events, we developed and deployed an optical system together with sediment traps in a mooring array for two years. The aim of the study was to record the seasonal variation in particle abundance and size-distribution with a camera-system in combination with vertical flux patterns collected with sediment traps in order to study transport processes and in-situ particle properties over longer periods in the upwelling system off Cape Blanc, Mauritania. Video recordings and trap samples were collected during two subsequent deployments of the eutrophic CBeu mooring located 120 nm off the coast, between spring 2008 and early winter 2010. The video camera was deployed within the mooring array at a depth of 1100 m, one hundred meters above the sediment trap, and was programmed to record 30 s of video in three day intervals. Materials and methods The CBeu mooring was first deployed in 2003 and has been serviced on an annual basis ever since. The data presented here are from the CBeu6 and CBeu7 moorings that were recovered and redeployed during the RV Poseidon cruise 365 and the RV Maria S. Merian cruise 11 in 2008 and 2009, respectively. The CBeu6 deployment period was from the 26th of April 2008 to the 28th of March 2009, followed by the CBeu7 mooring deployed between the 1st April 2009 and the 28th of Febuary 2010. The moorings were equipped with a sediment trap and a Multi Sensor Platform (MSP) both described below (Table 1). Multisensor platform (MSP) The MSP is a hexagonal glass fibre reinforced plastics frame with a height of 2.20 m, a diameter of 1 m and a total weight of

150 kg. It was equipped with a particle video-recording system (PVS) and a Falmouth Scientific 3DACM Conductivity, Temperature and Depth sensor (CTD). The MSP was placed 100 m above the sediment trap and, according to the pressure sensor, was located at 1110 m depth during the CBeu6 deployment and at 1145 m depth during CBeu7. The CTD measured temperature, pressure and conductivity and was further equipped with a compass and an acoustic current meter, providing absolute current direction and velocity. Additionally, a tiltmeter provided the tilt angle and tilt direction of the MSP. The CTD was programmed to collect 30 s of data every 6 h and averaging these measurements into a single value which was saved in its memory buffer. The Particle Video-recording System PVS consisted of a Sony HD camcorder with a standard 1080i HD resolution (1440  1080 pixels). The optical setup of the camera was based on a still camera systems used and described by Honjo et al. (1984), Asper (1987) and Ratmeyer and Wefer (1996). The PVS was programmed via a microcontroller connected to the camera’s LANC interface, which was programmed to record a video sequence of 30 s every three days at midnight. Each video sequence was illuminated with a 50 Hz Deep Sea Power & Light Seastrobe 2000, which was mounted at a distance of 45 cm perpendicular to the optical axis of the camcorder. The strobe light created a 12 cm wide slab of light whereby a defined sample volume of 6.47 l was illuminated. For size calibration, a scale was mounted at the opposite side of the strobe and was visible in the recordings. The entire unit was powered by a 12V/38Ah DSPL battery. The total recording time was limited by the 60 min recording length of the Mini-DV tapes, used by the camera. Unfortunately, a fault in the firmware prevented recordings during the CBeu6 deployment after the 31st of December 2008, resulting in a gap in the recordings between January and April 2009. Upon recovery of the PVS, the DV tape was digitalized into one AVI-file with full 1080i HD resolution. Subsequently, the Adobe Premiere Software package was used to separate the AVI file into individual video sequences of 30 s length, according to their recording date. Due to memory limitations of the image analysis software, we only used the first 15 s of the each sequence. Individual frames from each sequence were grabbed and saved as a BMP image stack, providing a total of 375 images for a 15 s video recording at a frame rate of 25 FPS. The BMP image stack was converted into an 8 bit grey scale image stack with 256 grey values, to ease the image analysis. Many video sequences could not be used, because the autofocus of the camera activated accidently for unknown reasons. As a result, the camera focus was outside the illuminated sample volume, thus, 79 of 196 video sequences were discarded. This amounted to a total number of 57 usable video sequences from the CBeu6 deployment and 60 usable video sequences from the CBeu7 mooring. The 8 bit grey scale image stacks were analysed using the IMAGEJ image analysis software. An 8 bit greyscale image covers 256 grey scale values, where a pixel with a grey value of 0 appears black and a pixel with a value of 255 appears bright white. By setting a grey value threshold, the background can be subtracted and the software recognises areas with a grey scale value above a certain threshold as a particle. Pixels below the threshold were treated as background and were excluded from the data set. The image analysis software extracted the size of individual particles recognised on each of the 375 frames that belonged to one sequence as a circular area, given in mm2. These data were saved to a worksheet, which was used for further calculations. The particle abundance was calculated by counting all particles found in every single frame and by dividing this number by 375. Hence, the particle abundance is the total number of particles averaged over all frames. From the area, the Equivalent Spherical Diameter (ESD) of each individual particle was calculated and likewise averaged over all frames to one value. The total particle

3

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11 Table 1 Sampling periods and sampling intervals of the CBeu sediment trap and particle video-recording system. CBeu mooring

PVS

Trap

Mooring #

Position (LAT/LON)

Waterdepth (m)

Depth (m)

Sampling period

Rec. interval (d)/Rec. time (s)

# of recordings

Depth (m)

Sampling period

Sampling interval (d)

# of samples

6

20°45.1 N/18°41.9 W

2699

1110

3/30

57

1210

40

20°44.6 N/18°42.7 W

2761

1145

3/30

60

1245

04/26/2008– 03/22/2009 04/05/2009– 02/28/2010

8.5

7

04/26/2008– 12/31/2008 04/07/2009– 02/19/2010

9

37

volume was the sum of all particle volumes, calculated from the ESDs, averaged over 375 frames. The smallest particle diameter that could be accurately determined with the optical setup of the PVS was 200 lm. Frames containing free swimming zooplankton were excluded from the measurements. This was the case in two recordings (8/18/2008 and 2/19/2010). It was not possible to identify different particle types by the video recordings and to assign them to specific particle populations. Particulate matter recorded by the video camera could be phytoplankton aggregates and/or fecal pellets produced by zooplankton. When we use the term particle, we refer to the entire particle spectra recorded by the camera unless otherwise stated. It was not possible to measure the sinking velocities of individual particles as initially intended. This was due to unexpected influence of the particle tracks by the movement of the platform. This influence caused the particles to move along rather turbulent tracks instead of a discrete direction as expected. Furthermore, the compass data from the CTD indicated that the platform was in constant motion and rotation. The platform rotation indicated a tidal signal, with a more or less 180° rotation every 6 h. Neither the video recordings, nor the CTD data enabled us to determine whether the trajectories of the particles were natural or were biased by the rotation of the platform. Thus, the moored video camera systems appear incapable of measuring in-situ sinking velocities of particles.

acquired by the camera. This was done by sorting the particles abundance and size distribution obtained for each video sequence into 21 logarithmically spaced size bins. The size bins from each video sequence were used to calculate the particle size spectra (n) from the particle number concentration (DC) within a given size range (Dd):

n¼

DC ½# m3 cm1  Dd

ð1Þ

The particle size spectra can be used to calculate mass fluxes, as described by Guidi et al. (2008) and Iversen et al. (2010). To do this, the authors assumed that the total mass flux (F) was equal to the mass flux spectrum integrated over the entire particle size spectra. Using diameter (d) as particle size measure, this can be described as:

F¼

Z

1

1

nðdÞmðdÞwðdÞdðdÞ ½g m2 d 

ð2Þ

0

where n(d) (# m3 cm1) is the particle size spectra per small size range as a function of diameter d (cm), m(d) is the particle mass (this can be as gram dry weight, carbon, nitrogen, etc.), and w(d) (m d1) is the average sinking velocity of the particles in a given size range. The method is based on the assumption, that particle mass (m) and sinking velocity (w) can be expressed by a power function as a function of particle diameter (d):

Sediment trap B

The sediment trap was moored at 1210 m depth during the CBeu6, and at 1245 m during the CBeu7 deployment period. The trap was a large aperture sediment trap of the Kiel type with 40 cups and an opening area of 0.5 m2 (Kremling et al., 1996). The sampling duration of each collection cup was 8.5 days during CBeu6 and 9 days during CBeu7. The cups were filled with GF/F filtered seawater with NaCl added to increase the salinity to 40 per mil. Additionally, the traps were poisoned with HgCl2 prior to the deployment. Upon trap recovery, swimmers were removed manually and the trap material was additionally wet-sieved through a 1 mm nylon mesh in the laboratory. All fluxes therefore refer to the size of <1 mm. The trap samples were homogenized and split into several subsamples which were subsequently used for analysis of total mass, particulate organic carbon, carbonate, lithogenic material, and opal. The calcium carbonate fraction consisted of pteropods, foraminifera and nanofossils (mainly coccoliths and coccolithophorids) whereby the latter group made up the majority of the carbonate fluxes with contributions between 64% and 94% at the mesotrophic sediment trap site further offshore (Fischer et al., 2007). Estimating fluxes from particle size-distribution and abundance We used the method described in Guidi et al. (2008) and Iversen et al. (2010) to estimate particle fluxes from the particle sizes

wm ¼ Ad

ð3Þ

In order to calculate the mass fluxes from the particle size spectra using Eqs. (2) and (3), the A and B values must be known. Following Guidi et al. (2008) and Iversen et al. (2010), a minimization procedure can be used to find the A and B vales, which provide the best fit between fluxes collected with the deep ocean sediment traps and fluxes calculated from the particle size spectra for each video sequence. The MatLab function fminsearch was used to find the A and B values, which returned the minimum log-transformed differences (DFC) between the sediment trap flux (F) and the fluxes estimated from the size spectra (FE):

DF C ¼

X 2 ½logðF ;i Þ  logðF E;i Þ

ð4Þ

i

Using the logarithmic transformation ensures equal weight to the differences for both large and small fluxes. We created different sets of A and B values by using the minimization procedure for (i) all trap fluxes and size spectra throughout a whole year = annually determined A and B values and (ii) by using the trap fluxes and particle size spectra for the different seasons to determine the A and B values = seasonally determined A and B values (see section ‘Comparison between estimated and measured trap-derived mass fluxes’). However, the minimization procedure needs several trap and estimated flux values as input parameters. Therefore, the few particle size spectra recorded during spring 2009 were not sufficient to determine seasonally A and B values for that period.

4

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11

Aqua MODIS satellite imagery Satellite data presented here were downloaded from the GES DISC – Goddard Earth Science Data and Information Services Center (http://disc.sci.gsfc.nasa.gov/). Monthly chlorophyll a data from February 2008 to February 2010 were extracted from a 1°  1° box around the CBeu trap location (20°N – 21°N, 18°W – 19°W). A 1° box was chosen, because it largely represents the catchment area of deep ocean sediment traps (Siegel and Deuser, 1997). A monthly Aerosol Optical Depth @ 550 nm (AOD) dataset was extracted for the same period, but from a larger box around CBeu. Due to the higher mobility and transportation pathways of windblown dust, the box was extended to the north and further to the east towards the coast (20°N – 22°N, 19°W – 17°). Regional settings The CBeu mooring site is located approximately 120 nm miles offshore Cape Blanc, Mauritania (20°45 N/18°42 W; Fig. 1), with a bottom depth of around 2800 m. It is part of the Canary Current System (CC), flowing southwards along the NW African coast and is one of the four major Eastern Boundary Upwelling Ecosystems (EBUEs) (Fréon et al., 2009) (Fig. 1). EBUEs are characterised by high primary production in the surface water and high export and transfer of organic carbon from the ocean surface to the deep-sea. Although they cover only 1% of the global ocean area, they contribute to up to 15% of the global primary production (Behrenfeld and Falkowski, 1997; Carr, 2002). According to Carr (2002) and Carr et al. (2006), the CC is the most productive EBUE with an annual primary production of 535 g C m2 year1 (Bory and Newton, 2000). The study site in the CC, unlike to other EBUEs, is characterised by predominantly carbonate-secreting primary producers, mainly nanofossils (coccolithophorids). Biogenic opal (diatoms) occurs in lower percentages of around 5% on average (of total mass flux) at the mesotrophic Cape Blanc site further offshore, which may be related to a rather low supply of dissolved silicate from the subsurface waters (Fischer et al., 2007; Ragueneau et al., 2000). Fischer and Karakas (2009) argued that a large portion of the settling material in

the Cape Blanc area consists of densely packed and fast sinking fecal pellets containing high amounts of nanofossils. At the eutrophic CB mooring, the contribution of biogenic opal to total mass is about 2– 3-fold higher compared to the mesotrophic site. However, diatoms are not the dominating primary producers at CBeu. The study area belongs to the Cape Blanc inter-gyre region (19– 24°N), which is characterized by a weaker seasonality (regularly with peaks in winter-spring and fall) and a persistent large offshore extension of chlorophyll (Lathuilière et al., 2008). According to Cropper et al. (2014), our study area is situated on the southern rim of the strong and permanent coastal upwelling zone (21– 26°N). Persistently blowing trade winds lead to coastal upwelling and an advection of the cold- and nutrient-rich surface water towards the open ocean. The Mauritanian upwelling system shows an immediate biological response, in terms of high chlorophyll concentration, to increased wind forcing (Mittelstaedt, 1991; Pradhan et al., 2006). The trade winds increase in intensity in late winter (Barton, 1998; Nykjaer and Van Camp, 1994), being highest during spring and early summer, and then decline (Fischer et al., 1996). During summer and fall, the Mauritania Current (MC) moves northwards to about 20°N (Mittelstaedt, 1991) and meets the CC. There, a salinity front is formed in the subsurface waters, known as the Cape Verde Frontal Zone (CVFZ, Zenk et al., 1991). Here, salty and nutrient-poor North Atlantic Central Water (NACW) is separated from the nutrient-rich and cooler South Atlantic Central Water (SACW), which provides a significant part of the source water for the upwelling in this region (Hughes and Barton, 1974). At the CVFZ, both water masses may be upwelled and mixed laterally, resulting in the formation of frontal eddies (Meunier et al., 2012). The region off Cape Blanc is characterised by a prominent structure, known as the ‘giant Cape Blanc filament’ (Van Camp et al., 1991). Although the upwelling belt is generally only 50– 70 km wide, cold water containing a high biomass of phytoplankton, is laterally advected up to 450 km offshore (Hernandez-Guerra and Nykjaer, 1997) mainly during the winter-spring season (Fischer et al., 2009). The filament is largest in winter and spring (Fischer et al., 2009) showing some interannual variability (Pradhan et al., 2006).

Fig. 1. (a) Study site showing surface currents and prevailing wind systems with position of the CBeu mooring, located within the highly dynamic upwelling system off Cape Blanc, Mauritania. (b) Aqua MODIS True Color image from 21th September 2009 (http://oceancolor.gsfc.nasa.gov), showing a large dust plume over the study area. The entire event lasted from 14th September to the 26th September 2009.

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11

The study area is strongly influenced by wind-blown dust from the Sahara (Fig. 1b) and is characterised by the largest dust deposition worldwide (Mahowald et al., 1999; Jickells et al., 2005). The shallow trade wind layer transports lithogenic material (dust) year-round from the Sahara/Sahel into the marine realm between the Canary Islands and the Cape Verde Islands (Pye, 1987 and references therein). The trade wind system, here known as the Harmattan, shows highest intensities in winter. The westward blowing Saharan Air Layer (SAL) above the trade wind layer originates from the African Easterly Jet, a mid-tropospheric wind system (Prospero, 1990) and is mostly active in summer season transporting fine grained dust down to the Caribbean (Prospero, 1990).

5

Results Particle abundance, size and volume During both years, the PVS returned particle concentrations ranging between 1 and 20 particles l1 (Fig. 2a), with an average concentration of 5.5 ± 3 particles l1. The average particle abundance during the CBeu6 deployment was 4.5 ± 2 particles l1 and 25% lower compared to the CBeu7 mooring period (6 ± 3.5 particles l1). In 2008, increased particle abundances were mainly observed during spring and fall/winter, with values around 10 particles l1. The summer and fall seasons in 2008 were characterized by rather small variations in concentration. The most remarkable

Fig. 2. Comparison of in-situ particle parameters abundance (a), ESD (b) and volume (c) from the video camera with sediment trap derived total mass fluxes (d) and bulk constituents (e–i). We selected four events from the flux record (E1-E4, grey shaded bars) that illustrated changing particle properties during different periods of high particle fluxes.

6

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11

increase in the particle abundance was recorded in early fall 2009, with maximum concentration of 20 particles l1. After mid-fall 2009, a drop in particle abundance was observed with lower concentrations compared to spring and summer. Although the majority of the particles had sizes which were below the marine snow range (0.5 mm), the average particle ESD was 0.78 ± 0.1 mm for the entire sampling periods (Fig. 2b). This indicated the existence of few, but considerably large aggregates in the water column. The variations in average particle sizes throughout the years were rather low. The average ESD was 0.8 ± 0.1 mm and 0.76 ± 0.07 mm for the CBeu6 and CBeu7 deployments, respectively. Increased particle diameters >1 mm were observed during spring and early summer 2008 as well as between fall and winter in 2008. The variability of particle diameters was rather low between spring 2009 and winter 2010. The total particle volume was the most variable particle property throughout the two years with values ranging from 1 mm3 l1 in mid-September 2008 to 22 mm3 l1 in December 2008 (Fig. 2c). The average value for the total particle volume during the CBeu6 deployment in 2008 was 6 ± 5 mm3 l1, while it was almost twofold lower during 2009/2010 (3.5 ± 2 mm3 l1). The total particle volume increased when both particle abundance and average diameter, or either one of the two parameters increased. High total particle volumes were observed during the first half of spring 2008 and during the onset of summer 2008. The highest measured total particle volume was observed during the fall to winter transition in 2008, reaching concentrations of up to 22 mm3 l1. Observations of high total particle volume (20 mm3 l1) were made in early fall 2008 due to one single large particle captured in the recordings. The same was true for the high total particle volume observed in a single video sequence during early winter 2010. Particle fluxes Particle mass fluxes were highly variable throughout the two years and ranged between 1200 mg m2 d1 in spring 2008 and 40 mg m2 d1 in winter 2010 (Fig. 2d). Generally, the flux followed distinct seasonal patterns with highest fluxes from spring until early summer and during early fall. During the fall to winter transition in 2008/2009, an event with high mass fluxes was observed but did not re-occur during the transition in 2009/2010, indicating some interannual variability. In part, similar flux patterns with highest fluxes during winter-spring were reported by Mollenhauer et al. (in press) for the same location between July 2003 and April 2007. However, interannual variability is evident, i.e. a secondary flux peak was found in autumn 2005. In terms of contribution to the mass flux, calcium carbonate was by far the most important bulk constituent and contributed up to 50% to the total mass flux (Fig. 2d and e). Thus, the carbonate flux pattern matches total mass. The majority of the calcium carbonate fluxes were nanofossils while foraminifera and pteropods were only present in minor amounts (data not shown). The annual average carbonate flux was 130 ± 90 mg m2 d1, with a maximum flux of 366 mg m2 d1during early spring 2008 (Fig. 2e). Lithogenic material was the second most important constituent and average flux for both years was 71 ± 76 m2 d1 largely following total mass (Fig. 2f). The lithogenic flux was highest during early spring 2008 and early fall 2009, reaching 400 mg m2 d1. However, there was also a shorter period of two weeks during late spring 2009 where the lithogenic flux dropped to almost zero. The flux peak during early fall 2009 is noteworthy, as the lithogenic flux contributed with about 50% to the total mass flux. Biogenic opal played a minor role for the total mass fluxes off Cape Blanc (Fig. 2g) which appears to be typical for the area (Fischer et al., 2007). Generally, elevated opal fluxes were observed during spring and summer with fluxes above

150 mg m2 d1. High opal fluxes were also recorded during fall/ winter 2009, but generally seemed low during fall with rates below 40 mg m2 d1. The trends in organic carbon flux followed those observed for total mass flux (R2 = 0.91) (Fig. 2d and h). There were positive correlations between organic carbon flux and the flux of calcium carbonate (R2 = 0.53), lithogenic material (R2 = 0.79) and biogenic opal (R2 = 0.85). The organic carbon fluxes were highest during spring and summer with maximum values of 80 mg C m2 d1, which was considerably higher than the average organic carbon flux of 16 ± 13.5 g C m2 d1 for both deployment years. Overall, organic carbon flux contributed between 3.2% and 8.1% to the total mass fluxes,with an average value of 5.6 ± 1.1%. Comparison between sediment trap mass fluxes and optical measurements Settling material was collected throughout the whole study period. Individual export events were characterized by noticeable changes in particle abundance and sizes (Fig. 2). Accordingly, we selected four events E1, E2, E3 and E4 (Fig. 2, grey shaded areas) which document significant changes in flux and particle properties. Some flux events, for instance in spring-summer 2009, were not reflected by the particle properties. Event E1 occurred during the first half of spring 2008 and had the highest mass fluxes observed over the entire two year period. Both particle abundance and diameter increased to levels above average values, resulting in high particle volumes (8.6–11.11 mm3 l1) throughout a period of three weeks. Event E2 occurred in early summer 2008 and had high total mass fluxes above 400 mg m2 d1, which were a result of low particle concentrations concurrent with a moderate increase in the particle diameters. This resulted in relatively high particle volumes ranging between 4 mm3 l1 and 14 mm3 l1. Event E3 occurred during the fall to winter transition in 2008. The situation was comparable to E1, where both particle abundance and diameter increased to levels above average. However, during E3, the total particle volume was much higher than those observed for the E1 event. The total particle volume measured during E3 had the highest value of the entire dataset and coincided with a massive pulse in mass flux (see Fig. 2). The CBeu7 deployment only showed one noticeable event, E4, which occurred during the first half of fall 2009 and reached very high fluxes of almost 700 mg m2 d1. During this event, particle abundance was the highest observed during the entire study (20 particles l1). However, the average particle diameters were lower than the annual average in ESD. In comparison to the other three events, event E4 had low opal fluxes but very high flux of lithogenic material, contributing to 50% to the total mass fluxes. Comparison between estimated and measured trap-derived mass fluxes Estimating mass fluxes from annually and seasonally determined A and B values, show different results. The estimated mass fluxes that were calculated from annually determined A and B values (Fig. 3a, black lines) overestimated the measured mass fluxes in fall and winter, but underestimated them during spring and summer (Fig 3a, grey bars). When the A and B values were determined for each season (Fig. 3b, black lines), the flux estimation became more accurate. The coefficient of determination of the linear regression between the measured and the estimated fluxes is far better when the A and B values were determined for each season individually (R2 = 0.56) in comparison to the annually determined A and B values (R2 = 0.36) (see Fig. 3, insert).

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11

Discussion In this study we compared seasonal patterns of particle fluxes derived from sediment traps with observations of particle abundance, total volume, concentration, and size-distribution at 1100 m depths from 2008 to 2010. The particle abundances from the PVS were in the range of those measured with a profiling particle camera system (ParCa) at 1100 m depth at the CBeu station during previous studies (Fischer et al., 2009; Nowald et al., 2006; Iversen et al., 2010). Stemmann et al. (2002) compiled a global data set of particle abundances in which all the deep ocean measurements were within the range of those observed in this study. Ratmeyer and Wefer (1996) deployed the ParCa at 995 m depth in a oligotrophic system north of the Canary Islands and found particle abundances which were 10 times higher than our abundances, however, their particle volume concentrations were in the lower range of those observed in this study. The similarities in particle abundance between different in-situ camera systems suggest a good optical capture efficiency and a good comparability between the PVS and previous systems, despite strong seasonal and regional differences in particle abundance, total volumes, and particle fluxes. Supporting this, the CBeu station generally had two times higher particle fluxes than the mesotrophic station (CBmeso) which was located 80 nm west of CBeu (Fischer et al., 1996, 2007; Fischer and Karakas, 2009). We observed a tight coupling between major flux peaks and total particle volume. This suggests that the camera was capable of recording the sinking particles making up the total flux and even capture individual flux patterns and flux pulses. Thus, the data from the camera seemed to provide a realistic representation of particle abundance and size-distribution, despite a rather low temporal resolution of the recordings of 30 s every third day. However, the correlation coefficient for linear regression between mass flux and total particle volume was rather low (R2 = 0.18). Because the sampling technique and sample intervals of the sediment traps and the optical system were so different, a linear regression might not be suitable to compare both data sets. Although the particle abundances showed considerable agreement with the trends in total mass flux, especially in 2009/2010 (CBeu7), the total particle volume was the better choice when comparing the sediment trap data with the optical measurements.

7

Fig. 4 indicates that events with high mass fluxes occur through different combinations of average particle diameter and total particle abundance. During periods with low particle fluxes (<200 mg m2 d1, blue dots in Fig. 4), particle diameters were generally small and the particle abundances low, whilst high mass fluxes occurred in situations with an increase in both particle abundance and diameter (E1 and E3) or in situations with a high number of small particles (E4). Nonetheless, high total fluxes were also observed during E2, though particle abundance was very low without a pronounced increase in particle size. The high flux events could thus be formed as a result of different particle sizedistributions and abundance combinations. Still, there seemed to be an upper limit of mass fluxes, created either due to many small or few large particles but not as a result of many large particles (see encircled data points in Fig. 4). The particle abundances and size-distributions, which we observed at 1100 m depth was a result of physical (e.g. aggregation and disaggregation) and biological processes (e.g. consumption and remineralisation) occurring in the overlaying waters. Since settling marine snow aggregates are typically formed from surface ocean primary producers (e.g. Trent et al., 1978; Alldredge and Gotschalk, 1988; Smetacek, 1985) and may be ballasted by lithogenic material (e.g. Hamm, 2002; Iversen et al., 2010), we compared the particle fluxes, particle abundance, volume concentrations, and average sizes with satellite derived ocean color and aerosol optical thickness data (Fig. 5a and b). Increased chlorophyll concentration in the ocean surface did not generally coincide with high mass fluxes determined by the sediment trap. For example, chlorophyll concentration between spring and early summer 2008 was high (1.8–3.5 mg m3), which corresponded with the flux events E1 and E2. By contrast, chlorophyll concentrations were in a similar range during fall and winter 2008, but particle fluxes were much lower. Upwelling in the study area occurs year-round (Cropper et al., 2014) with higher intensities during winter-spring, which generally results in increased chlorophyll concentrations. However, the relationship between the chlorophyll standing stock in a 1°  1° grid above the trap and deep ocean mass flux (e.g. of organic carbon) is not straightforward. This seemed due to various processes forming and destroying larger sinking particles in the surface and subsurface layer. Further, there seemed to be a closer link between deep ocean

Fig. 3. Comparison between measured particle fluxes from the sediment traps (grey bars) and estimated fluxes (black lines) from PVS derived particle size spectra (Eqs. (2) and (3)). Estimated fluxes were derived from annually determined A and B values (a, upper panel) and from seasonally determined A and B values (b, lower panel). Note that the gap in the estimated fluxes during winter 2008/2009 was related to a malfunctioning of the PVS. No fluxes for the season of spring 2009 could be calculated, since too few particle size spectra were determined from the video recordings during this period to determine A and B values with the minimization procedure.

8

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11

Fig. 4. Relationships between average ESD (Equivalent Spherical Diameter), total particle abundance and particle mass flux. Particle abundance is plotted as a function of ESD and grouped into six color coded ranges of total mass flux, as determined by the sediment traps.

fluxes and the entire size of the Cape Blanc filament (Fischer et al., 2009), than between fluxes and chlorophyll concentrations acquired from a static area of surface ocean above the trap location. Guidi et al. (2009) could explain 68% of the global variation in mass fluxes with changes in the size structure of the

phytoplankton community and integrated chlorophyll a through the euphotic zone. On the other hand, Ternon et al. (2010) did not find any correlation between POC flux and increased phytoplankton biomass in the Mediterranean Sea. In fact, Ternon et al. (2010) observed highest fluxes during periods before an increase in the surface chlorophyll concentrations, which contradicts our common understanding of production, particle formation and export. Generally, one would expect temporal delay between the production and arrival of the material in the deep-sea, which can be somewhat inferred from the data presented in Fig. 5. However, the flux of particles is generally influenced by several different parameters such as the input of ballast (e.g. Iversen and Ploug, 2010; Ploug et al., 2008; Fischer et al., 2009), particle formation processes (e.g. Karakas et al., 2009) and sinking velocities (Iversen and Ploug, 2010; Fischer and Karakas, 2009). Until we fully understand these influences on particles transport, it remains difficult to link chlorophyll standing stocks and primary productivity to the export and transfer of particles to the deep-ocean. However, the events E1, E2 and E3 were most likely results of high chlorophyll standing stock and increased primary production in the ocean surface. All three events occurred during periods of a high chlorophyll concentration, which lasted over several weeks to months and biogenic opal mainly produced by diatoms shows distinct flux maxima (Fig. 2g). Further, high mass and particulate organic carbon fluxes coincided with the occurrence of organic rich, comparably large and/or numerous aggregates (Fig. 4), which were potentially formed during periods of high primary productivity (e.g. Smetacek, 1985). However, the particulate matter of the individual events was characterized by large differences regarding chemical composition and physical properties. The mass and carbon fluxes of E1 were 1.5-fold higher than during E3, although particle abundances and sizes were similar during both events, though the total volumes were lower at E1.This implies that the composition of the individual particles during the two events was different with more dense, organic rich and faster settling particles during E1. On the other hand, events E2 and E3 had similar fluxes (average mass flux was 584 and 511 mg m2 d1 for E2 and E3, respectively), but E2 had lower particle abundances than E3. This indicates that composition of the particles collected during E2 were denser and more organic rich than those collected during E3. It is

Fig. 5. Temporal variability of satellite derived chlorophyll standing stock (a) and Aeorsol Optical Depth (b) between 2008 and early 2010. Grey shaded bars indicate the four defined events E1 to E4. The red bar in late summer 2009 indicates the occurrence of a large dust plume.

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11

possible that this was due to a high carbonate ballasting which resulted in higher sinking velocities (Iversen and Ploug, 2010; Fischer and Karakas, 2009) whereby the transfer time from surface to 1100 m depth was shorter and thus, more organic carbon was preserved within the individual aggregates (e.g. Francois et al., 2002). The event E4 differed from the previous events by having low surface chlorophyll, low biogenic opal flux but a high total mass flux. The satellite images indicated that the chlorophyll concentration during the month before and during E4 was much lower (<1 mg m3) than for the events E1-E3. In contrast to E1 and E3, event E4 was preceded by high values in aerosol optical thickness. E4 had particle abundances three times higher than the annual average while the ESDs were slightly smaller than the annual average. The chemical composition of event E4 showed a very high flux of lithogenic material, which contributed with roughly 50% to the total mass flux. On the other hand, biogenic opal flux was lower in comparison to the other events. Previous studies have shown that incorporation of lithogenic material can result in the formation of small and dense aggregates (e.g. De La Rocha et al., 2008; Hamm, 2002) and/or the fragmentation of larger aggregates into smaller particles (Passow and De La Rocha, 2006). This could explain the high abundance of small particles during E4. The high content of lithogenic material is mainly due to the deposition of fine-grained Saharan/Sahelian dust (Iversen et al., 2010; Fischer and Karakas, 2009; Bory and Newton, 2000). This is supported by Aqua MODIS satellite True Color images (Fig. 1b) and AOD data (Fig. 5b). The AOD record showed a seasonal pattern, with high dust loads in the atmosphere during the summer seasons of 2008 and 2009. The abrupt decrease in the AOD values in late September 2009, overlapped with event E4 recorded in the deep ocean. More importantly, the True Color images indicated a giant atmospheric dust plume which occurred between the 14th September and the 26th September. Although the AOD values were high during the entire summer period, we hypothesize that E4 was triggered by this short-termed, high magnitude dust plume observed from the True Color images. We estimated a delay time of approximately 5 days between the first occurrence of the dust plume (14thof September) and the increase in lithogenic flux at 1100 m (19th of September). This might suggest a close coupling between dust deposition and export of particles with settling rates of 200 m d1, which is typical for marine snow and fecal pellets (Ploug et al., 2008; Iversen and Ploug, 2010) and within the range observed for particles off Cape Blanc (Fischer and Karakas, 2009; Iversen et al., 2010). For the northern Canary Current system, Neuer et al. (2004) described such a close coupling between dust deposition and subsequent export and transfer of organic carbon to the deep ocean. A fast transmission of the atmospheric dust signal via lithogenic fluxes to the 2000 m deep North Atlantic was described by Brust et al. (2011). However, a close coupling between dust and deep ocean carbon flux in the NE tropical Atlantic Ocean was questioned by Bory et al. (2002). The lithogenic flux was only slightly higher during E1 compared to E4 but the True Color images indicated no dust plume and AOD values were only moderately increased before and during the occurrence of E1 (Fig. 5b). We can only speculate why lithogenic fluxes were high during E1. Ternon et al. (2010) proposed a mechanism in the Mediterranean Sea, whereby the fine-grained lithogenic material was stored in the upper water column for longer periods of time, until it was scavenged by settling organic particles. Due to low biological activity, small sized mineral particles could remain in the water column, which prevented the transfer of the lithogenic material via organic aggregates to the deep. Export of the lithogenic material to the deep-sea could have been triggered by a bloom, which was also previously suggested by Bory and Newton (2000) and Bory et al. (2002). They discussed that a rapid

9

transfer of lithogenic material was coupled to biological activity in the surface waters. The high chlorophyll in the ocean surface in the months before the occurrence of E1 indicates the regular spring bloom of phytoplankton which is also documented with high biogenic opal fluxes. The lithogenic material stored in the water column could have been scavenged by settling particles and subsequently transferred to the deep-sea, as suggested by Ternon et al. (2010). The lithogenic material settling during event E1 obviously did not induce a fragmentation of the particles as observed for E4. E4 only differed from the other events by having low biogenic opal (diatom) fluxes, which suggests no strong phytoplankton bloom in spring 2008. Iversen and Ploug (2010) observed that diatom dominated marine snow particles were larger than marine snow formed by nanofossils (coccolithophorid) or a mixture of nanofossils and diatoms. This might suggest that the combination of high lithogenic fluxes with a low contribution of diatoms could have resulted in the formation of many small particles during E4. Obviously, there is a general trend of larger average particle sizes during periods with high biogenic opal flux. Both the present study and that by Fischer and Karakas (2009) suggest that high mass and organic carbon fluxes are closely related to high lithogenic and/or carbonate fluxes off Cape Blanc. The combination between long-term deployments of in-situ camera systems and sediment traps showed that export and transfer of different particle types was strongly influenced by seasonal changes and the input of wind-blown dust, which affected the composition and size-distribution of the settling particles. Previous studies have shown that particles from different regions have different biogeochemical characteristics (Guidi et al., 2008; Iversen et al., 2010). This study showed that such differences may also occur within one region due to seasonal changes in the environmental conditions and dust input. For instance, major carbonate producers such as nanofossils (coccolithophorids) were present year round with a maximum in summer/fall whilst diatoms predominantly bloomed during winter-spring. Thus, particle composition and properties change seasonally. Furthermore, the influence from zooplankton flux feeding has been observed at high rate in the upper 100 m of the water column off Cape Blanc (Iversen et al., 2010). These influences can both be seasonal and diel and may cause particle transformations into (i) several small particles if large particle are fragmented by swimming and feeding activity (e.g. Goldthwait et al., 2004) or (ii) particle formation in the form of fecal pellet production (e.g. Ebersbach and Trull, 2008). To support our hypothesis, that particles properties are changing seasonally, we calculated particle fluxes similar to the trap collected fluxes by using the method of Guidi et al. (2008). Seasonally determined A and B parameters provided better flux estimates than annually determined A and B parameters (Fig. 3a and b), suggesting that the composition of the settling particles in 1100 m water depth changed seasonally. Considering the high rates of degradation, consumption and transformation of the settling material in the water column, it is quite surprising that deep ocean particles still show clear seasonal differences in particle characteristics. It suggests that little particle transformation occurs once the aggregates escape the zone of high biological activity in the surface and subsurface waters. This supports previous studies which have suggested limited degradation in the meso-pelagic and deep ocean (Iversen et al., 2010; Iversen and Ploug, 2013). Lampitt et al. (1993) observed similar seasonal differences in the physical nature of the particle pool over short periods, however, their study was from shallower water depth (270 m) and would thus have a more direct link to the euphotic zone where most aggregation processes occur. The results from the present study show that a combined deployment of sediment traps and optical systems offers a novel

10

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11

approach to link changes in bulk flux over time to the composition of the individual particles settling downwards and constituting the fluxes. We are convinced that this type of study, which focuses on particle-specific dynamics, will improve our understanding of the mechanisms controlling particle transport and flux in the ocean at different seasonal and regional scales. These results show a great potential and can be improved further through higher sampling rates and a higher optical resolution of the camera. A higher optical resolution will shift the lower particle size range from the 200 lm, which the PVS had, to e.g. 73 lm as used by McDonnell and Buesseler (2012) or 100 lm as used by Checkley et al. (2008), but it would also lower the volume of each image and, thereby, likely capture fewer of the large scarce particles. However, it was observed that peak fluxes could occur due to many small or few large particles, but not as a result of many large particles (Fig. 4). This suggests that both large and small particles can make up the majority of the exported material and that one cannot just focus on one end of particle size-spectra. A more important improvement would be to have cameras and traps at several water depths simultaneously. This would allow a full water column perspective of the influence from different phytoplankton blooms, atmospheric dust deposition, and the activity and vertical migration of zooplankton on the export and transfer of organic carbon. Additionally, a better temporal resolution (e.g. day-night cycles) would improve our knowledge on the impact of zooplankton on particle formation and settling.

Acknowledgments We would like to thank the masters and crews of the RV POSEIDON and RV MARIA S. MERIAN for their support and assistance during deployment and recovery of the moorings. The authors are grateful for the valuable comments of two anonymous reviewers. We also would like to thank G. Ruhland, M. Klann and B. Donner for their help in the laboratory as well as J. Renken for the development of the camera system. The study was funded through DFG-Research Center/Cluster of Excellence ‘‘The Ocean in the Earth System’’ at the University of Bremen.

References Alldredge, A.L., Gotschalk, C., 1988. In situ settling behaviour of marine snow. Limnology and Oceanography 33 (3), 339–351. Asper, V.L., 1987. Measuring the flux and sinking speed of marine snow aggregates. Deep Sea Research Part I 34 (1), 1–17. Barton, E.D., 1998. Eastern Boundary of the North Atlantic: Northwest Africa and Iberia. In: Robinson, A.R., Brink, K. (Eds.), The Global Coastal Ocean. John Wiley & Sons, New York, Chichester, Weinheim, Brisbane, Singapore, Toronto, pp. 29– 67 (vol. 11). Behrenfeld, M.J., Falkowski, P.G., 1997. Photosynthetic rates derived from satellitebased chlorophyll concentration. Limnology and Oceanography 42 (1), 1–20. Bory, A.J.M., Newton, P.P., 2000. Transport of airborne lithogenic material down through the water column in two contrasting regions of the eastern subtropical North Atlantic Ocean. Global Biogeochemical Cycles 14 (1), 297–315. Bory, A., Dulac, F., Moulin, C., Chiapello, I., Newton, P.P., Guelle, W., Lambert, C.E., Bergametti, G., 2002. Atmospheric and oceanic dust fluxes in the northeastern tropical Atlantic ocean: how close a coupling? Annales Geophysicae 20, 2067– 2076. Broecker, W.S., 1982. Ocean Chemistry During Glacial Times. GeochimCosmochimActa 46, 1698–1705. Brust, J., Schulz-Bull, D.E., Leipe, T., Chavagnac, V., Waniek, J.J., 2011. Descending particles: from the atmosphere to the deep ocean: A time series study in the subtropical NE Atlantic. Geophysical Research Letters 38, L06603. http:// dx.doi.org/10.1029/2010GL045399. Carr, M.-E., 2002. Estimation of potential productivity in Eastern Boundary Currents using remote sensing. Deep-Sea Research Part I 49, 59–80. Carr, M.-E., Friedrichs, M.A.M., Schmeltz, M., 2006. A comparison of global estimates of marine primary production from ocean colour. Deep Sea Research Part II 53, 73–98. Checkley Jr., D.M., Davis, R.E., Herman, A.W., Jackson, G.A., Beanlands, G.A., Regier, L.A., 2008. Assessing plankton and other particles in situ with the SOLOPC. Limnology Oceanographic 53 (5, part 2), 2123–2136.

Cropper, T.E., Hanna, E., Bigg, G.R., 2014. Spatial and temporal seasonal trends in coastal upwelling off Northwest Africa, 1981–2012. Deep-Sea Research II 86, 94–111. De La Rocha, C.L., Nowald, N., Passow, U., 2008. Interactions between diatom aggregates, minerals, particulate organic carbon, and dissolved organic matter: Further implications for the ballast hypothesis. Global Biogeochemical Cycles 22 (4). Ebersbach, F., Trull, T.W., 2008. Sinking particle properties from polyacrylamide gels during the Kerguelen Ocean and Plateau compared Study (KEOPS): Zooplankton control of carbon export in an area of persistent natural iron inputs in the Southern Ocean. Limnology and Oceanography 53 (1), 212–224. Fischer, G., Karakas, G., 2009. Sinking rates and ballast composition of particles in the Atlantic Ocean: implications for the organic carbon fluxes to the deep ocean. Biogeosciences 6, 85–102. Fischer, G., Donner, B., Ratmeyer, V., Davenport, R., Wefer, G., 1996. Distinct year-toyear particle flux variations off Cape Blanc during 1988–1991: Relation to delta super(18)O-deduced sea-surface temperatures and trade winds. Journal of Marine Research 54 (1), 73–98. Fischer, G., Karakas, G., Blaas, M., Ratmeyer, V., Nowald, N., Schlitzer, R., Helmke, P., Davenport, R., Donner, B., Neuer, S., Wefer, G., 2007. Mineral ballast and particle settling rates in the coastal upwelling system off NW Africa and the South Atlantic. International Journal of Earth Sciences. http://dx.doi.org/10.1007/ s00531-007-0234-7. Fischer, G., Reuter, C., Karakas, G., Nowald, N., Wefer, G., 2009. Offshore advection of particles within the Cape Blanc filament, Mauritania: Results from observational and modelling studies. Progress in Oceanography 83, 322–330. Francois, R., Honjo, S., Krishfield, R., Manganini, S., 2002. Factors controlling the flux of organic carbon to the bathypelagic zone of the ocean. Global Biogeochemical Cycles 16 (4). http://dx.doi.org/10.1029/2001GB001722. Fréon, P., Barange, M., Aristegui, J., 2009. Eastern Boundary Upwelling Ecosystems: integratuve and comparative approaches. Progress in Oceanography 83, 1–14. Goldthwait, S., Yen, J., Brown, J., Alldredge, A., 2004. Quantification of marine snow fragmentation by swimming euphausiids. Limnology and Oceanography 49 (4), 940–952. Gorsky, G., Aldorf, C., Kage, M., Picheral, M., Garcia, Y., Favole, J., 1992. Vertical distribution of suspended aggregates determined by a new underwater video profiler. Annales de l’Institutocéanographique 68, 275–280. Guidi, L., Jackson, G.A., Stemmann, L., Miquel, J.C., Picheral, M., Gorsky, G., 2008. Relationship between particle size distribution and flux in the mesopelagic zone. Deep Sea Research Part I 55 (10), 1364–1374. Guidi, L., Stemmann, L., Jackson, G.A., Ibanez, F., Claustre, H., Legendre, L., Picheral, M., Gorsky, G., 2009. Effects of phytoplankton community on production, size and export of large aggregates: a world-ocean analysis. Limnology and Oceanography 54 (6), 1951–1963. Hamm, C.E., 2002. Interactive aggregation and sedimentation of diatoms and claysized lithogenic material. Limnology and Oceanography 47 (6), 1790–1795. Hernandez-Guerra, A., Nykjaer, L., 1997. Sea surface temperature variability off north-west Africa: 1981–1989. International Journal of Remote Sensing 18 (12), 2539–2558. Honjo, S., Doherty, K.W., Agrawal, Y.C., Asper, V.L., 1984. Direct optical assessment of large amorphous aggregates (marine snow) in the deep ocean. Deep Sea Research Part A 31 (1), 67–76. Hughes, P., Barton, E.D., 1974. Stratification and water mass structure in the upwelling area northwest Africa in April/May 1969. Deep Sea Research I 21, 611–628. Iversen, M.H., Ploug, H., 2010. Ballast minerals and the sinking carbon flux in the ocean: carbon-specific respiration rates and sinking velocities of marine snow aggregates. BiogeoSciences 7, 2613–2624. http://dx.doi.org/10.5194/bg-72613-2010. Iversen, M.H., Ploug, H., 2013. Temperature effects on carbon-specific respiration rate and sinking velocity of diatom aggregates – potential implications for deep ocean export processes. Biogeosciences 10. http://dx.doi.org/10.5194/bg-104073-2013. Iversen, M.H., Nowald, N., Ploug, H., Jackson, G.A., Fischer, G., 2010. High resolution profiles of vertical particulate organic matter export off Cape Blanc, Mauritania: degradation processes and ballasting effects. Deep Sea Research Part I 57, 771– 784. Jackson, G.A., Checkley Jr., D.M., 2011. Particle size distribution in the upper 100 m water column and their implications for animal feeding in the plankton. DeepSea Research I 58, 283–297. Jickells, T.D., An, Z.S., Andersen, K.K., Baker, A.R., Bergametti, G., Brooks, N., Cao, J.J., Boyd, P.W., Duce, R.A., Hunter, K.A., Kawahata, H., Kubilay, N., laRoche, J., Liss, P.S., Mahowald, N., Prospero, J.M., Ridgwell, A.J., Tegen, I., Torres, R., 2005. Global iron connections between desert dust, ocean biogeochemistry, and climate. Science 308 (5718), 67–71. Karakas, G., Nowald, N., Blaas, M., Marchesiello, P., Frickenhaus, S., Schlitzer, R., 2006. High-resolution modeling of sediment erosion and particle transport across the northwest African shelf. Journal of Geophysical Research 111. http:// dx.doi.org/10.1029/2005JC003296. Karakas, G., Nowald, N., Schäfer-Neth, C., Iversen, M.H., Barkmann, W., Fischer, G., Marchesiello, P., Schlitzer, R., 2009. Impact of particle aggregation on vertical fluxes of organic matter. Progress in Oceanography 83, 331–341. Kremling, K., Lentz, U., Zeitzschell, B., Schulz-Bull, D.E., Duinker, J.C., 1996. New type of time-series sediment trap for the reliable collection of inorganic and organic trace chemical substances. Review of Scientific Instruments 67, 4360–4363.

N. Nowald et al. / Progress in Oceanography 137 (2015) 1–11 Lampitt, R.S., Hillier, W.R., Challenor, P.G., 1993. Seasonal and diel variation in the open ocean concentration of marine snow aggregates. Letters to Nature 362 (22), 737–739. Lathuilière, C., Echevin, V., Levy, M., 2008. Seasonal and intraseasonal surface chlorophyll-a variability along the northwest African coast. Journal of Geophysical Research: Oceans 13 (C5), C05007. Mahowald, N., Kohfeld, K., Hansson, M., Balkanski, Y., Harrison, S.P., Prentice, I.C., Schulz, M., Rodhe, H., 1999. Dust sources and deposition during the last glacial maximum and current climate: a comparison of model results with paleodata from ice cores and marine sediments. Journal of Geophysical ResearchAtmospheres 104 (D13), 15895–15916. McDonnell, A.M.P., Buesseler, K.O., 2012. A new method for the estimation of sinking particle fluxes from measurements of the particle size distribution, average sinking velocity, and carbon content. Limnology Oceanography Methods 10, 329–346. Meunier, T., Barton, E.D., Barreiro, B., Torres, R., 2012. Upwelling filaments off Cape Blanc: interaction of the NW African upwelling current and the Cape Verde frontal zone eddy field? Journal of Geophysical Research: Oceans 117 (C8), C08031. Mittelstaedt, E., 1991. The ocean boundary along the northwest African coast: Circulation and oceanographic properties at the sea surface: Progess in. Oceanography 26, 307–355. Mollenhauer, G., Basse, A., Kim, J.-H., Damsté, J.,S.,S., Fischer, G. (in press). A fouryear record of UK’37 and TEX86-derived sea surface temperature estimates from sinking particles in the filamentous upwelling region off Cape Blanc, Mauritania. Deep Sea Research Part I. Neuer, S., Torres-Padron, M.E., Gelado-Caballeo, M.D., Rueda, M.J., Hernandez-Brito, J., Davenport, R., Wefer, G., 2004. Dust deposition to the eastern subtropical North Atlantic gyre: Does ocean’s biogeochemistry respond? Global Biogeochemical Cycles 18. http://dx.doi.org/10.1029/2004GB002228. Nowald, N., Karakas, G., Ratmeyer, V., Fischer, G., Schlitzer, R., Davenport, R., Wefer, G., 2006. Distribution and transport processes of marine particulate matter off Cape Blanc (NW-Africa): results from vertical camera profiles. Ocean Science Discussions 3, 903–938. Nykjaer, L., Van Camp, L., 1994. Seasonal and interannual variability of coastal upwelling along northwest Africa and Portugal from, 1981to 1991. Journal of Geophysical Research 99, 197–207. Passow, U., De La Rocha, C., 2006. Accumulation of mineral ballast on organic aggregates. Global Biogeochemical Cycles 20, 1–7. Ploug, H., Iversen, M.H., Fischer, G., 2008. Ballast, sinking velocity, and apparent diffusivity within marine snow and zooplankton fecal pellets: implications for substrate turnover by attached bacteria. Limnology and Oceanography 53 (5), 1878–1886. Pradhan, Y., Lavender, S.J., Hardman-Mountford, N.J., Aiken, J., 2006. Seasonal and inter-annual variability of chlorophyll-a concentration in the Mauritanian

11

upwelling: observation of an anomalous event during 1998–1999. Deep Sea Research (Part II, Topical Studies in Oceanography) 53, 1548–1559. Prospero, J.M., 1990. Mineral-aerosol transport to the North Atlantic and North Pacific: The impact of African and Asian sources. In: Knap, A.H., (Ed.), The longrange atmospheric transport of natural and contaminant substances: Dodrecht, Mathematical and Physical Sciences. Kluwer Academic Publishers, p. 19–52. Pye, K., 1987. Aeolian Dust and Dust Deposits. Elsevier, New York, p. 334. Ragueneau, O., Treguer, P., Leynaert, A., Anderson, R.F., Brzezinski, M.A., DeMaster, D.J., Dugdale, R.C., Dymond, J., Fischer, G., Francois, R., Heinze, C., Maier-Reimer, E., Martin-Jezequel, V., Nelson, D.M., Queguiner, B., 2000. A review of the Si cycle in the modern ocean: recent progress and missing gaps in the application of biogenic opal as a paleoproductivity proxy. Global Planet Change 26 (4), 317– 365. Ratmeyer, V., Wefer, G., 1996. A high resolution camera system (ParCa) for imaging particles in the ocean: System design and results from profiles and a threemonth deployment. Journal of Marine Research 54 (3), 589–603. Siegel, D.A., Deuser, W.G., 1997. Trajectories of sinking particles in the Sargasso Sea: modeling of statistical funnels above deep-ocean sediment traps. Deep-Sea Research I 44 (9–10), 1519–1541. Smetacek, V., 1985. Role of sinking in diatom life-history cycles: ecological, evolutionary and geological significance. Marine Biology 84, 239–251. Stemmann, L., Gorsky, G., Marty, J.C., Picheral, M., Miquel, J.C., 2002. Four-year study of large-particle vertical distribution (0–1000 m) in the NW Mediterranean in relation to hydrology, phytoplankton, and vertical flux. Deep-Sea Research II 49 (11), 2143–2162. Stemmann, L., Jackson, G.A., Gorksy, G., 2004. A vertical model of particle size distributions and fluxes in the midwater column that includes biological and physical processes; Part II, Application to a three year survey in the NW Mediterranean Sea. Deep-Sea Research I 51 (7), 885–908. Ternon, E., Guieu, C., Loye-Pilot, M.-D., Leblond, N., Bosc, E., Gasser, B., Miquel, J.C., Martin, J., 2010. The impact of Saharan dust on the particulate export in the water column of the North Western Mediterranean Sea. Biogeosciences 7, 809– 826. Trent, J.D., Shanks, D., Silver, M.W., 1978. In situ and laboratory measurements on macroscopic aggregates in Monterey Bay, California. Limnology and Oceanography 23, 626–635. Van Camp, L., Nykjær, L., Mittelstaedt, E., Schlittenhardt, P., 1991. Upwelling and boundary circulation off Northwest Africa as depicted by infrared and visible satellite observations. Progress in Oceanography 26, 357–402. Walsh, I.D., Gardner, W.D., 1992. A comparison of aggregate profiles with sediment trap fluxes. Deep Sea Research I 39 (11/12), 1817–1834. Zenk, W., Klein, B., Schroder, M., 1991. Cape verde frontal zone. Deep Sea Research Part A. Oceanographic Research Paper 38 (Supplement 1(0)), 505–530.