Thermal Stratification in Saldanha Bay (South Africa)

Estuarine, Coastal and Shelf Science (1999) 49, 877–890 Article No. ecss.1999.0550, available online at http://www.idealibrary.com on

Thermal Stratiﬁcation in Saldanha Bay (South Africa) and Subtidal, Density-driven Exchange with the Coastal Waters of the Benguela Upwelling System

P. M. S. Monteiroa and J. L. Largierb aCouncil for Scientiﬁc and Industrial Research, P.O. Box 320, Stellenbosch, 7599, South Africa bDepartment of Oceanography, University of Cape Town, Rondebosch 7701, South Africa; Currently at: Scripps Institution of Oceanography, San Diego, CA 92093-0209, U.S.A.

Received 1 February 1999 and accepted in revised form 3 August 1999

A study was conducted to understand the mechanisms driving observed subtidal variability in the stratification of Saldanha Bay, located in the southern Benguela system. It was found that the 6–8 day period variability in bay stratification was caused by the inflow and outflow of cold upwelled water driven by changing baroclinic pressure gradients between the coastal and bay domains. The direction and magnitude of the pressure gradients were governed by coastal upwelling activity and a lag in the response of the bay to changes in density structure in the coastal ocean. When the pressure gradients were bayward and cold water was being driven into the bay the cycle was termed to be in an ‘ active phase ’ and the reverse was termed the ‘ relaxation phase ’. The upwelling-favourable equatorward wind stress impacted the bay stratification in two ways: on the regional scale, wind drives upwelling and governs the inflow–outflow of cold upwelled bottom water, which strengthens stratification; conversely, on the local bay scale, wind drives vertical mixing, which weakens stratification. A four-phase model is used to describe the observed variability in stratification in the bay. The associated density-driven exchange flows are capable of flushing the bay in 6–8 days, about one-third of the time for tidal exchange alone (c. 25 days). These inflows of cold bottom water are ecologically critical as they supply nutrients to the bay and thus impose a control on new production within the bay environment. Further ecological implications of this bay–ocean exchange include export of phytoplankton new production to the coast, limitation of the risk of harmful algal blooms (HABs) and the division of the system into two distinct ecosystems (bay and lagoon). 1999 Academic Press

Keywords: baroclinic; shelf-bay; exchange; stratiﬁcation; upwelling; phytoplankton-dynamics; Benguela; Saldanha Bay

Introduction that baroclinic pressure gradients associated with the bay–ocean density difference drive the subtidal For much of the 10-month upwelling season exchange of water between the bay and the shelf. (August–May), Saldanha Bay (South Africa) is These bay–ocean density (temperature) differences thermally stratified with a surface warmed upper layer result from wind-driven upwelling of cold water over and a lower layer of cold upwelled water that has been the shelf at the mouth of Saldanha Bay. Typical time advected into the bay. Fluctuations in water tempera- scales of the wind forcing are 6–8 days (Shannon, tures and stratification correspond to fluctuations in 1985; Shannon & Nelson, 1996; Monteiro et al., the strength and persistence of regional-scale synoptic 1998). This is discussed as a local dynamic prob- winds and local surface heat fluxes (Shannon & lem, with the depth of isotherms at the mouth of the Nelson, 1996; Monteiro et al., 1998). The link bay controlling the bay–ocean density differences. between stratification in Saldanha Bay and exchange The problem of whether the isotherm depth is pri- between bay and ocean has been a focus of research marily forced by remote shelf-wave propagation since a 2-year thermistor string record provided a clear or by changes in the regional wind field will not illustration of how stratification within a bay varied on be considered here. While the focus of this study is synoptic time scales (Monteiro et al., 1998). In this on baroclinic processes, owing to their impact on study data is analysed from an intense 2-week exper- stratification and nutrient dynamics, we recog- iment which included CTD surveys and a second nize that shelf-bay exchange is also due to baro- thermistor string in the outer bay. Our hypothesis is tropic processes. For example, higher-frequency aE-mail: [email protected] boundary forcing mechanisms, such as tidal and

0272–7714/99/120877+14 $30.00/0 1999 Academic Press 878 P. M. S. Monteiro and J. L. Largier intertidal motions are also important at shorter time open bays (e.g. Drakes Bay, California and St Helena scales. Bay, Southern Benguela) appear to be primarily due In previous work, two recognizable extreme phases to upwelling that results from wind stress on the of synoptic scale stratification variability in Saldanha surface of the bay itself. Bay were identified. The ‘ active phase ’ was used to As might be expected, these intrusions of bottom describe the condition when Saldanha Bay becomes water have significant implications in respect of strongly stratified due to the bottom inflow of cold understanding the biogeochemical and ecological upwelled water. The ‘ relaxation phase ’ was used to functioning of the Bay. Shelf-bay coupling governs describe the condition when the bay returns to an phytoplankton new production, the upper and lower isothermal condition as cold bottom water retreats limits of the carrying capacity of Saldanha Bay for (Monteiro et al., 1998). These phases appear to be filter-feeder mariculture production, the low inci- closely correlated to fluctuations in the equatorward dence of harmful algal blooms (HABs) in the Bay, the wind stress that drives coastal upwelling. While this ecological de-coupling of the bay and Langebaan previous work recognizes the importance of ocean Lagoon systems and the biogeochemistry of pollut- forcing, the description of phases was based more on ants. The vertical flux of nutrients across the thermo- the symptom, water column temperatures in Small cline within the bay was addressed by Monteiro et al. Bay, than on the cause, density driven inflows of (1998). The objective of this study is to describe the cold bottom water. In this paper, the view to focus mechanism that governs sub-tidal horizontal ocean– attention on the underlying reason for the intrusion of fluxes, which is a pre-requisite for the vertical turbu- a cold bottom layer is revised. While clear evidence lent flux of nitrate across the thermocline to the is provided for the importance of gravity-driven euphotic zone in the bay. Both temporal and spatial intrusions of dense bottom water, which would occur water column data are used to show how this cold even in the absence of wind stress on the surface of the water alternately fills the bay up to a depth of about bay itself (Largier, 1996; Chadwick et al., 1996; 5 m and then drains and mixes away in response to Monteiro et al., 1998), this dynamic is modified fluctuations in wind forcing. somewhat by Ekman dynamics which may extend into Saldanha Bay (Lentz, 1992; Spolander, 1996). We consider this direct wind forcing of bay currents to be Materials and methods secondary . Some comparable observations on thermal density- The study area driven exchange have been obtained from other bays—particularly interesting are other bays con- Saldanha Bay is a coastal embayment located in the nected to eastern boundary upwelling systems (e.g. southern Benguela upwelling system (33S, 18E) California, Chile). Typical of these bays is the juxta- approximately 100 km north of Cape Town. Before position of cold ocean water with surface-warmed bay 1975 it comprised two main sections; an Outer Bay of water, producing a bay–ocean density gradient that predominantly coastal character and an Inner Bay can result in vertically sheared exchange flow. Basal (Figure 1). Its physiography was altered in 1975 by inflow of cold bottom water has been observed in San the construction of a 4 km long jetty and the Marcus Francisco Bay (Largier, 1996), northern Monterey Island causeway (Weeks et al., 1990)(Figure 1). Bay (Graham & Largier, 1997), Tomales Bay (Largier These modifications divided the Inner Bay into two et al., 1998), San Diego Bay (Chadwick et al., 1996; parts which have been commonly referred to as Small Largier et al., 1997) and Mission Bay (Largier et al., Bay (14·3106 m2) and Big Bay (42·2106 m2) 1996, 1997), in addition to Saldanha Bay (Monteiro (Figure 1). Saldanha Bay is linked to the Benguela et al., 1998). However, there are important differences System to the west and a large shallow tidal lagoon in the dynamics owing to differences in width and (Langebaan Lagoon) to the south. The nominal depth of the mouth versus size and depth of bay, and boundaries between Saldanha Bay and adjacent differences in the relative strengths of tidal forcing, systems are marked in Figure 1. The boundary wind forcing, freshwater inflow and heating and between the relatively shallow (5–20 m) and strongly evaporation rates (Largier et al., 1997). In the case of stratified Saldanha Bay and the more oceanic outer narrow-mouth bays, such as Tomales, San Diego and bay is depicted as line A. Outer bay on the seaward San Francisco, it appears that the intrusions are side of this boundary is a semi-protected body of primarily density-driven, although the high-frequency deeper (25–50 m) water of largely oceanic character. structure is strongly influenced by tide. In contrast, The boundary with the tidal and very shallow cold water intrusions to Monterey Bay and other more Langebaan Lagoon is depicted as B. Langebaan Density driven shelf-bay exchange 879

(a) 18°E

St Helena Bay Cape Columbine

33°S Saldanha Bay Langebaan Lagoon

Benguela Upwelling

System

Cape Town

(b)

S Cape 33° Y5 Saldanha Agulhas Small Bay

PFF MS1 SSF Z5 IOJ PC 2' Y1 Marcus Is MS2 Big Bay Z1 Malgas Is. X3 A 4' X1 Outer Bay Peninsula B Jutten Is. Southern Benguela Upwelling System Langebaan Lagoon 56' 58' 18° 2' E F 1. Map of the layout of Saldanha Bay. Saldanha Bay is located poleward of an important upwelling centre (Cape Columbine) in the southern Benguela upwelling system. The bay system comprises an Outer Bay of largely coastal upwelling domain character and an Inner Bay which was physically divided into two hydrodynamically distinct sections: Big Bay and Small Bay by a 4 km long iron ore jetty (IOJ). The boundary into Langebaan Lagoon to the south is shown (b). Also shown are the positions of the CTD temperature section stations (X, Y and Z lines) as well as the time series moorings in Small Bay (MS1) and the boundary (X3). Wind data were measured at the Port Control site (PC). The 5 m-depth contour, which marks an important hydrodynamical and ecological boundary, is also shown. 880 P. M. S. Monteiro and J. L. Largier waters tend also to be warmer and saltier than and re-stratification linked to the first event (25/2–1/3) Saldanha waters (Shannon & Stander, 1977; Largier were very rapid, suggesting the sudden passage of a et al., 1997). mixing front past the thermistor moorings. The first mixing event appears to be true wind driven mixing as Sampling the isothermal water column at both thermistor-string sites has a temperature similar to the average tempera- Sampling was carried out over a 13-day period from ture of the stratified water column prior to mixing. 24 February to 8 March 1997 (24/2 to 8/3 in the date The larger amount of cold water at the bay mouth site convention used subsequently in this paper). The results in a horizontal temperature gradient, even after location of the CTD stations and the thermistor mixing. The second mixing event, observed only in strings are depicted in Figure 1. The X-line stations Small Bay, appears to have occurred in response to the span the Outer Bay (X1 to X3). The Y-line stations disappearance of the cold bottom water from this span Small Bay and the Z-line spans Big Bay. Lines location and resultant decrease in water column are oriented across-isobath and are located away from stability. Re-stratification is rapid and marked by a the boundaries. CTD (conductivity–temperature– drop in bottom temperature, indicating the onset of a depth) profiles were carried out with a Seabird new intrusion of bottom water. Following both events, SBE-19 CTD/O/pH pumped instrument which al- the surface temperature starts increasing slowly once lowed slower deployment rates best suited for the the water column is stratified, consistent with surface slower response O2 (Oxygen: SBE 23) and pH (SBE heating. However, at the bay mouth site [Figure 2(b)], 18) sensors as well as higher vertical resolution. The the surface temperature does not increase immediately Small Bay thermistor string was an 11 m Aanderaa following re-stratification (27/2), but only starts in- chain with 1 m intervals between sensors logging creasing once the inner bay has re-stratified (1/3). hourly over the period of the field trip. It is located at After the second mixing event in Small Bay, the positions MS1 (Monitoring Station 1) in Figure 1. new intrusion of bottom water has warmer than The thermistor string at the mouth of the Inner Bay after the first—consistent with the idea that colder (station X3) was a string of 17 independent Onsett water was raised to a greater height over the shelf by Hobo-T thermistors, deployed at 1 m vertical spacing stronger winds and stronger coastal upwelling during and logging data every 12 min. It is located at position 25–28/2. MS2 in Figure 1. The two thermistor chain records overlap for a period of 10 days (24/2 to 6/3). Sea-level Wind and sea level data. Northward and eastward data were obtained from two sites in the Bay, both components of the hourly wind records from using stilling well type units; the Navy site (SAN) and Saldanha Bay and Cape Columbine are shown in the CSIR sensor at the end of the iron ore jetty (IOJ) Figure 2(c,d). Both orthogonal components are de- (Figure 1). Local wind data were obtained from the picted for completeness, but the equatorward (north- Port Captain’s Control tower and coastal wind data ward) component alone is a good index of the strength were obtained from Cape Columbine, located on the of upwelling along the open coast. It should be noted open coast approximately 50 km north of Saldanha that the inertial oscillation period at the latitude of Bay (G. Nelson: SFRI, Cape Town). Saldanha Bay is approximately 22 h so wind events with duration of a day or less have little or no impact on coastal upwelling dynamics. Results Fluctuations in the northward wind component at Saldanha Bay and Cape Columbine are in phase, Time series data although wind speed at the Saldanha Bay site is Thermistor data. Both thermistor string times series greater during upwelling wind events (e.g. 25/2). The exhibit periods of stratification and periods when eastward component of wind exhibits different the water column is mixed [Figure 2(a,b)]. The iso- patterns at the two sites [Figure 2(d)]. At Cape thermal or weakly stratified periods occurred during Columbine, this component is very weak, illustrating the very strong wind event of 25/2 to 1/3 and to a the coast-polarised nature of winds at that site. At lesser extent during the 5/3 wind event. Only during Saldanha Bay, a strong diurnal signal is evident the earlier event was the water column mixed at the (amplitude about 5 ms1). bay mouth, remaining stratified during the later event. Subtidal sea-level data are plotted in Figure 2(e). At the Small Bay thermistor site, stratified periods The tidal signal was removed with a 40-h low were observed during weak winds (<5 ms1), specifi- pass filter and corrections were made for changes in cally 24–25/2, 1–4/3 and 5–8/3. Both de-stratification atmospheric pressure, thus providing a dynamically Density driven shelf-bay exchange 881

20.0 MS1 18.0 1 m 16.0 mean 14.0 11 m

Temperature 12.0 10.0 5–02–1997 16 17 18 19 20 21 22 23 24 25 26 27 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15–03–1997 1 mrt 19.0 MS2 17.0 15.0 1 m 13.0 11 m

Temperature 11.0 9.0 5–02–1997 16 17 18 19 20 21 22 23 24 25 26 27 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15–03–1997 1 mrt 25.0 Northward 15.0 PortCon

0.0

–5.0 Wind(N) m/s Wind(N) CapeCol –15.0 5–02–1997 16 17 18 19 20 21 22 23 24 25 26 27 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15–03–1997 1 mrt 25.0 Eastward 15.0 PortCon

5.0

–5.0 Wind(E) m/s Wind(E) CapeCol –15.0 5–02–1997 16 17 18 19 20 21 22 23 24 25 26 27 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15–03–1997 1 mrt 1.25 Sea Level 1.15

1.05 10J – SL Sea Level (m) 0.95 5–02–1997 16 17 18 19 20 21 22 23 24 25 26 27 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15–03–1997 1 mrt F 2. The time series data sets. The thermistor chain temperature time series from the moorings in Small Bay (a) and the bay–coastal domain boundary (b) depict the variability at 1 m, 6 m and 11 m depth. The equatorward (c) and zonal (d) components of the wind at Saldanha Bay and Cape Columbine. They show the enhanced equatorward speeds and zonal variability in Saldanha Bay. The sea level variability in Saldanha Bay (e) processed through a 40 h low pass filter to reveal the variability linked to upwelling intensity. Upwelling conditions are characterized by low sea levels. meaningful synthetic sub-surface pressure (which will and high sea-levels during calm or downwelling be referred to as ‘ sea-level ’ for simplicity). Sea-level winds. Although fluctuations in sea-level appear to variability in this time scale is generally well matched lead wind fluctuations by several hours, the record with the strength of upwelling-favourable winds length is not long enough relative to event time scales [Figure 2(c)]: low sea-levels during upwelling winds to calculate meaningful correlation statistics. 882 P. M. S. Monteiro and J. L. Largier

Saldanha Bay Cross section showing Temperature contours

x1 x2 x3 y1 y2 y3 y4 y5 0 m 0 m 15.00 17.00 19.00 15.00 17.00 19.00 15.00 19.00 19.00 13.00 17.00 15.00 17.00 13.00 15.00 17.00 17.00 15.00 13.00 15.00 –5 m 15.00 –5 m 13.00 13.00 13.00 13.00 13.00 –10 m –10 m 11.00 11.00 11.00 11.00 11.00 11.00 –15 m –15 m

–20 m –20 m 11.00

–25 m –25 m

–30 m –30 m

–35 m (a) 24/2/1997 –35 m (b) 25/2/1997

0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m 0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m

0 m 17.00 0 m

–5 m 13.00 15.00 –5 m 13.00 15.00 17.00 17.00 14.00 11.00 13.00 16.00 11.00 –10 m 13.00 15.00 –10 m 13.00 15.00 14.00 11.00 14.00 13.00 11.00 –15 m –15 m

–20 m –20 m

–25 m –25 m

–30 m –30 m

–35 m (c) 27/2/1997 –35 m (d) 28/2/1997

0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m 0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m x1 x2 x3 y1 y2 y3 y4 y5 0 m 0 m 15.00 17.00 15.00 15.00 17.00 13.00 13.00 15.00 15.00 13.00 15.00 –5 m 13.00 13.00 –5 m 15.00 13.00 13.00

13.00 13.00 13.00 –10 m –10 m 13.00

–15 m –15 m 11.00 11.00

–20 m 11.00 –20 m

11.00 –25 m –25 m

–30 m –30 m

–35 m (e) 3/3/1997 –35 m (f) 6/3/1997

0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m 0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m F 3. (a–f)

Wind data show that the sampling period CTD temperature sections (24/2–8/3) started in a brief calm period, soon after a CTD sections [Figure 3(a–j)] depict the thermal protracted upwelling event (16/2–23/2). This calm structure along sections extending from the coastal period was followed by a short but very strong ocean waters (station X1) through the bay-shelf upwelling event (25/2–1/3), 2 days of non-upwelling boundary (station X3) into either Small Bay (stations wind and a second upwelling event that was much Y1–Y5) or Big Bay (stations Z1–Z5). These sections weaker (3/3–5/3). During the ﬁnal days winds were can be related to the thermistor time series obtained weak (6/3–8/3). close to stations X3 and Y3. The plots are arranged Density driven shelf-bay exchange 883

Saldanha Bay Cross section showing Temperature contours

x1 x2 x3 z1 z2 z3 z4 z5 x1 x2 x3 z1 z2 z3 z4 z5 0 m 0 m 15.00 15.00 15.00 15.00 15.00 15.00 13.00 13.00 –5 m 15.00 –5 m 11.00 13.00 13.00 13.00 11.00 13.00 13.00 11.00 11.00 15.00 –10 m 11.00 13.00 –10 m 13.00 11.00 11.00 11.00 11.00 –15 m –15 m 11.00

–20 m –20 m

–25 m –25 m

–30 m –30 m

–35 m (g) 28/2/1997 –35 m (h) 2/3/1997

0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m 0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m

0 m 0 m 17.00 13.00 15.00 15.00 13.00 15.00 15.00 13.00 15.00 15.00 –5 m –5 m 13.00 13.00 13.00 13.00 13.00 13.00 12.00 13.00 –10 m –10 m 13.00 11.00 12.00 12.00 12.00 –15 m 11.00 –15 m 11.00

–20 m –20 m 11.00 11.00

11.00 –25 m –25 m

–30 m –30 m

–35 m (i) 3/3/1997 –35 m (j) 6/3/1997

0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m 0 m100 m 200 m 300 m 400 m 500 m 600 m 700 m 800 m 900 m 1000 m F 3. (g–j) F 3. The CTD temperature sections along the XY (a–f) and XZ (g–j) lines (see Figure 1 layout). These sections show the spatial variability of stratiﬁcation between the bay and coastal domains on selected days (dates are given) during the time series [Figure 2(a,b)].

into two groups, which separately depict the sections a negligible impact on the density variability. Thus, into Small Bay [Figure 3(a–f)] and those into Big temperature measured in the CTD sections and Bay [Figure 3(g–j)]. These Small Bay and Big Bay thermistor chains can be interpreted in terms of sections share the same data in the X1–X3 part of temperature (heat balances) or density (dynamical the section, as can be seen when sections into the balances). two sub-embayments were obtained on the same CTD sections for 24/2 and 25/2 [Figure 3(a,b)] day. In some instances (e.g. 24/2 and 26/2) the characterize the water column during the first coastal domain was not sampled due to extreme stratified period prior to the onset of the strong sea conditions and the sections were truncated. As wind event late on the 25 February [Figure 2(c)]. the sections into Small Bay (Stations Y1–Y5) was This stratification is particularly strong within Small sampled more often, discussion will focus primarily Bay, owing to the presence of a surface heated upper on these data. The less intensely sampled Big Bay layer with temperatures over 20 C. The thermocline sections are included to illustrate the extent to which is at 3–8 m and has a gradient of about 1 Cm1, the Small Bay thermistor string data reflect the water the strongest observed during this experiment. The column structure over the whole bay. In this study, water is warmest along the shallow northern shore of temperature was used as a suitable proxy for density the bay. Between 24/2 and 25/2, one can see the because the variability of salinity is minimal and has disappearance of the coldest water from the bay: for 884 P. M. S. Monteiro and J. L. Largier example, at the bay-mouth station X3, the 11 C Water less than 10C has disappeared from the bay isotherm dropped from 12 m to 20 m and water and sub-11 C water has almost disappeared. The colder than 10 C (at 20 m on 24/2) had receded 12 C isotherm has dropped by 10 m from the pre- completely out of Outer Bay (not visible at depths vious day. Below the thermocline, isotherms still slope up to 35 m). seaward and one can expect the draining of 11–12 C CTD sections for 27/2 and 28/2 [Figure 3(c,d,g)] water to continue. Above the thermocline, isotherms characterize the water column on two occasions dur- slope shoreward in response to greater surface heating ing the first isothermal period in Small Bay. On the of shallow water resident in the bay. Near-surface first day of the strong southerly wind event (26/2), it temperatures exceed 17 C in the middle of Small was not possible to obtain CTD data. Nevertheless, Bay. an isothermal water column was observed seaward to The CTD sections for 6/3 [Figure 3(f,j)] follow the thermistor string at the mouth of the bay, where further draining of cold water on 4/3 and mixing of the the water was about 2 C cooler than at the thermistor water column in Small Bay on 5/3. On 6/3 the surface string location in Small Bay. This horizontal gradient layer has been re-established by surface warming and in temperature (warmer water farther into the bay) is the lower layer has been re-established by a new also evident in the CTD section for 27/2. In this intrusion of cold water. Although a sharp thermocline section one can already see the new intrusion of has not developed, the total temperature difference bottom water at the mouth of Small Bay (station Y1), over the water column is 5 C at the thermistor string consistent with the re-stratification recorded by the (station Y3) in Small Bay. mouth thermistor string [Figure 2(b)] at station X3. The CTD section for 28/2 includes the Outer Bay and Discussion shows the elevated isotherms over the shelf and the more developed intrusion, with sub-11 C water The combination of time-series and section data extending into both Small and Big Bay. The contin- provides temporal and spatial views of the wind- ued wind over the bay, however, maintains a relatively correlated, cold, bottom water intrusions that drive deep surface mixed layer which, appears as an iso- fluctuations in bay stratification. Wind plays two roles thermal water column inshore of the 10 m isobath. in bay stratification. Surface wind stress on the bay Barotropic and baroclinic tidal effects cause the lead- tends to mix the water column whereas surface wind ing edge of the bottom layer to advance and retreat on stress on the ocean drives upwelling over the shelf, tidal time scales—resulting in 13 C water and near- which tends to stratify the bay. Stratification is also bottom stratification being observed brieflyatthe enhanced by surface heating of the bay. It is the Small Bay thermistor string (station X3) on 28/2 balance between vertical mixing and buoyancy fluxes [Figure 2(a)]. Also of interest on 28/2 is the presence that leads to the changes in thermal stratification in of the warmest water in the middle of Small Bay Saldanha Bay. There is a positive feedback as this [Figure 3(d)]. stratification enables intrusion of cold water, which in The CTD section for 2/3 [Figure 3(h)] character- turn strengthen the stratification. There are two parts izes the structure of the water column on the second to the discussion. Firstly, an explanation is provided day of the stratified period following the strong wind for the observed stratification in the bay and its mixing event—a period of weak or non-upwelling relation to density-driven exchange flows between the winds. Very cold water (<10C) has intruded far into bay and the coastal ocean. Secondly, the implications Big Bay (inshore of the 10 m isobath) and a typical of this relationship for bay biogeochemistry and strong, shallow stratification has been re-established ecology are addressed. within the Bay. A thin layer of warmer near-surface water (>15 C) extends from the shore out into the Shelf–bay density differences and bay stratification ocean, well beyond the mouth. While near-surface isotherms in the Outer Bay are level following the Horizontal bay–ocean differences in temperature lead intrusion of cold bottom water, at depth isotherms are to a pressure gradient that changes with depth. Thus, already sloping seaward. This is in response to the at depth, pressure gradients are either more landward lowering of isotherms over the shelf (e.g. 10C water is or more seaward than those near-surface. This baro- found at 25 m at the outermost Station X1 and at clinic pressure gradient then controls the preferential 10 m at the innermost Station X3). low-frequency intrusion or draining of cold bottom The CTD sections for 3/3 [Figure 3(e,i)] were water to or from the bay. Given this dynamical sampled in a period characterized by weak winds and view, we re-define the idea of active and relaxation further draining of cold bottom water from the bay. phases: Density driven shelf-bay exchange 885

(a) The ‘ active phase ’ refers to a bayward baroclinic cold water). Decreases in stratification are due to pressure gradient, which occurs when the coastal cooling of the surface (through surface heat fluxes), domain is colder than the bay. Isotherms slope warming of deeper water (through outflow of colder down into the bay and cold water below the water) or greater mixing of the water column by tides sloping isotherms is preferentially advected into (bottom stress) and winds (surface stress). In the the bay. Thus, ‘ active ’ implies that cold water is present study of Saldanha Bay, there is a net heating being actively fed into bay from the coastal of the surface (typically 130–170 W m2 in summer; domain. The active phase usually occurs during Monteiro et al., 1998) which is less variable than the upwelling winds and immediately following these other synoptically varying factors (e.g. bottom inflow/ winds, before the ocean isotherms lower. outflow, wind mixing). In essence, the changes in (b) The ‘ relaxation ’ phase refers to a seaward baro- stratification during this 1997 experiment reflect an clinic pressure gradient, which occurs when the interaction between the advective buoyancy flux bay is colder than the ocean. Isotherms slope up associated with the intrusion and draining of bottom into the bay and cold water below the sloping water and the vertical mixing due to surface wind isotherms is preferentially advected out of the bay. stress. Tidal currents are weak and although there is Thus, ‘ relaxing ’ implies that cold water in the some evidence for tidal fluctuations in stratification, bay is no longer being held up or pushed in by the subtidal fluctuations in stratification are dominated cold water in the ocean and that the bay stratifi- by wind-related mixing. Depending on the strength of cation is relaxing—i.e. the cold bottom water is the wind, stratified and mixed water columns can be draining out of the bay. The relaxation phase observed during both ‘ active ’ and ‘ relaxation ’ usually occurs during weak or poleward winds, phases. In Small Bay there were two periods during following an upwelling wind event. which the water column was mixed (25–27/2 and 4–5/3) and three periods during which the water col- Active phases were observed from 25/2 to 1/3 umn was stratified (24–25/2, 1–3/3 and 6–9/3). The associated with the strong upwelling wind event and first and second stratified periods were during a relax- from 4/3 to 6/3 associated with the second upwelling ation phase and weak winds. The third stratified period wind event. The only significant relaxation phase was was during a weakly active phase. The first mixed observed during weak poleward winds from 1/3 to 3/3. period was during an active phase, but with very strong The strength and direction of the ocean–bay baro- wind. The second mixed period was at the end of a clinic pressure gradient are reflected in Figure 4, relaxation phase, during moderate to weak winds. which compares the 11-m temperatures at the two In summary, stratification in Saldanha Bay is con- thermistor sites. MS1 indexes the Bay and MS2 trolled by winds which act through a combination of indexes offshore coastal waters, although this outer two mechanisms: control of the exchange of sub- site is only an approximation as it is located at the thermocline water to and from Saldanha Bay mouth of the bay. Observed ocean–bay temperature through wind-driven coastal upwelling, and control differences are greater than 2 C and may be sustained of vertical shear through local wind-driven vertical for a day, or even longer if vertical mixing is strong mixing. The time-dependent relationship between enough to preclude intrusion. Thus active indicates these mechanisms is presented below as a 4-phase the potential for intrusion, not necessarily an active model. lower layer inflow. For example, the active phase conditions on 26/2 did not lead to bottom water Time-dependence in stratification and ocean–bay intrusion because mixing dominated the stratification exchange balance and a vertically sheared exchange flow could Given the typical time scale of upwelling winds not develop. Clearly, the relationship between sub- (6–8 days), one can describe phases of a conceptual thermocline bay–ocean exchange of bottom water and stratification cycle in Saldanha Bay and show how bay stratification is complex. While the efficacy of the closely it correlates with observed temperature vertical exchange flow during active and passive sections [Figure 3(a–j)] and time series (Figure 2). In phases is controlled by vertical stratification and mix- the model cycle, this phase starts the end of a relaxing, this stratification in turn depends on the intrusion ation period, when the bay is characterized by some activity. form of weakened thermal stratification. What are the key terms that contribute to the bay stratification balance? Increases in stratification are Phase 1—onset of equatorward wind, local mixing and due to heating of the surface (through surface heat coastal upwelling. Within hours of the start of fluxes) and cooling of deeper water (through inflow of upwelling winds, mixing effects within the bay act 886 P. M. S. Monteiro and J. L. Largier

19.0

18.0

Active 17.0

Active

16.0

MS1 - Temperature (11 m)

15.0 C) °

14.0 Temperature ( Temperature

13.0

12.0 MS2 - Temperature (11 m)

11.0

10.0

Relaxed

9.0 5–02–1997 161718192021222324252627281234567891011121314 15–03–1997 1 mrt F 4. The relative variability of the temperature record at 11 m between the thermistor chain mooring in Small Bay (MS1) and bay–coastal domain boundary (MS2 at X3) (see Figure 1). It shows that the ‘ active ’ phase seen in the periods 25/2 to 1/3 and 5/3 to 7/3 corresponds to a condition when the coastal domain is colder than the bay and the ‘ relaxation ’ phase during 1/3 to 4/3 corresponds to a condition when the opposite holds. alone and the residual thermocline is eroded, perhaps was solely due to the strength of the wind to the bottom in the shallower parts of the bay. (>20ms1)—sufficient to overcome the strongly Examples of the onset of equatorward winds are found stratified conditions. The mixing effect is seen both in on 25/2 and 4/3 [Figure 2(d)]. The latter example on the time series temperature records [Figure 2(a,b)]as 4/3 serves as the typical case where mixing of the well as in the temperature sections [Figure 3(a–j)]. water column in Small Bay was preceded by a At the same time, the upwelling winds over the shelf gradual weakening of stratification [Figure 2(a)]. drive offshore Ekman transport and cold water The first example on 25/2 is less typical as mixing upwells near the coast, resulting in the presence of Density driven shelf-bay exchange 887 cold water at the shallow mouth of Saldanha Bay temperature section from 24/2 [Figure 3(a)]. Tem- about a day after the wind starts. peratures of the shallow surface layer (<5 m) in Small Bay were as high as 20 C and underlain by a Phase 2—intrusion of cold bottom water. Coastal strong thermocline (>1 Cm1) and a bottom layer upwelling occurs over the time scale of an inertial with temperatures of about 11 C. period (approximately 22·0 h at 33 latitude). The subsequent intrusion of cold water into the bay occurs Phase 4—draining of cold bottom water. In contrast to at the speed of an internal gravity wave (order the immediate end to vertical mixing as winds weaken, 0·1ms1), thus taking another 20 to 30 h to intrude upwelling structures over the shelf are in geostrophic the 5–10 km separation between the coastal and bay equilibrium and they will only decay slowly as energy domains. Thus, although one may observe some pre- is lost to friction (Smith, 1995). Typically, upwelled liminary cooling earlier, the full intrusion of cold isotherms lower over a few days. This can be seen bottom water will only be complete about 1·5–2·5 by comparing the temperature record in Small Bay days after the onset of an upwelling wind. This can be [Figure 2(a)] with the wind record [Figure 2(c)] from seen by comparing the timing of the onset of equator- 1/3 to 5/3. An immediate increase in stratification was ward winds with the timing of the response of the observed with the weakening of upwelling winds on temperature field at the thermistor strings [Figure 1/3. Only a day later (corresponding to the 22-h 2(a–e)]. These records show that the onset of inertial oscillation period) did bottom water start upwelling winds in the late evening of 3/3 was only warming and bay stratification start weakening in reflected as an intrusion of cold water in Small Bay on response to the decay of upwelling structures in the 5/3 some 30 h later [Figure 2(a,c)]. coastal domain [Figure 2(a,c)]. The slower decay of Resulting from this dense bottom intrusion, coastal upwelling structures and its impact on bay stratification is strengthened and re-established from stratification is emphasized in the initial part of the below. So, while the stratifying bottom water intrusion thermistor record in small Bay (MS1). It can be seen and the de-stratifying vertical mixing are coupled in that the initial stratification induced by the preceding that they are both forced by the same wind, they are period of persistent equatorward winds was not somewhat de-coupled by the temporal lag in the two destroyed by the short (c. 1 day) wind reversal period effects. Wind mixing happens a day to a day and [Figure 2(a,c)]. Strong upwelling conditions were a half prior to the complete intrusion of new re-established on 25/2 in a time scale that matched sub-thermocline water. the inertial oscillation period so that the baroclinic This second phase can be delayed or even pre- pressure gradient, which maintained the cold water cluded when the wind is strong enough to maintain intrusion, was only weakened rather than reversed vertical mixing as this prevents the development of [Figure 2(a,c)]. vertical shear and the intrusion of a lower layer. This In a hypothetical 6-day cycle, the first 3 days can be seen in Figure 3(d,g) which shows that (Phases 1 and 2) are the active phase and the second although a strong baroclinic pressure gradient was 3 days (Phases 3 and 4) are the relaxation phase. The established between the coastal and bay domains, the strongest stratification is observed on days 2–5. The persistently strong winds prevented its intrusion into most ‘ active ’ day is day 2 and the most ‘ relaxing ’ the bay. Once wind mixing weakened on 1/3, bottom days are days 5 and 6. In reality, the wind is highly water flowed into the bay and strong stratification was variable, rarely starting suddenly and rarely stopping established [Figures 2(a) and 3(h)]. suddenly. Further, the cycle may be short circuited (skipping phase 4) if there is only a brief break in Phase 3—enhanced stratification. As a result of this upwelling winds. In this case, the bay will remain strengthening and shoaling of the thermocline, sur- stratified (e.g. 24–25/2) and it is unusual that the new face warming becomes trapped in a thinner surface upwelling wind event will be strong enough to mix the layer and surface temperatures increase. Thus, water column as happened late on 25/2. Nevertheless, stratification is strengthened further. While a steady this conceptual cycle presents a pattern that one can state may occur if this wind continues, it is about look for in the observations of reality. this time that the upwelling winds weaken. As the local wind forcing disappears, the surface heating is Shelf-bay exchange fluxes limited to an even shallower surface layer and stratification may increase further. The intense stratifi- One of the most important aspects of the density- cation due to the combined action of cold water driven ocean–exchange flow is its impact on the intrusion and solar heating is best shown in the renewal of bay water. Earlier studies that viewed 888 P. M. S. Monteiro and J. L. Largier

T 1. Volumes (106 m3) of the surface and bottom layer in three key parts of Saldanha Bay during the period of maximum intrusion of cold bottom layer up to a depth of 5 m

Bay area Depth range volumes (106 m3) (see Figure 1) 0–5m 5m–Bottom Total

Outer Bay 137·8 501·8 639·6 Inner Bay (Small and Big Bays) 251·8 344·4 596·2 Langebaan Lagoon 64·23·868·0

exchange as being primarily tidally driven estimated a growth limiting nutrient NO3 is naturally advected flushing period of 25 days (Shannon & Stander, 1977; into the relatively shallow (<20 m) Saldanha Bay. Weeks et al., 1990). These subtidal density-driven Coupled with vertical turbulent flux across the exchange flows flush the bay more rapidly. The thermocline, this supply of NO3 drives new produc- intrusion of cold upwelled water into Saldanha Bay tion in Saldanha Bay (King & Devol, 1979; Monteiro typically reaches a depth of approximately 5 m. This et al., 1998; Smaal et al., 1997/1998). Nitrate is a allows some useful volume flux calculations to be limiting nutrient in the Benguela upwelling system carried out which can be used to estimate the flushing (Andrews & Hutchings, 1980; Brown & Hutchings, period of the bay. The volume and area statistics 1987; Hutchings et al., 1995) and its physical supply obtained from a GIS for the bay are summarized in rate into Saldanha Bay also regulates bay-scale phyto- Table 1 (G. Bosch, CSIR, pers. comm.). plantkon new production. The dynamics of phyto- These volume data show that each intrusion of cold plankton new and regenerated production are then bottom water into the Inner Bay (Small and Big Bays) dependent on the correct understanding of the inter- replaces 58% of the total volume of Inner Bay. Two action of both outcomes of density driven flows: upwelling cycles would flush 84% of the volume of advective intrusion and the variability of thermocline Inner Bay—in a time of 12–16 days, about half the strength. The evolution of bay stratification, described flushing time estimated on the basis of tidal flushing above as a 4-phase cycle, linked to the intrusion of alone. NO3 rich cold water provides the basis for an Furthermore, wind-driven vertical entrainment of improved understanding of the evolution of phyto- cold bottom water into the surface layer will further plankton bloom dynamics within the bay. Nutrients reduce this period. It has been estimated on the basis are available for phytoplankton new production dur- of the Pollard, Rhines and Thompson (1973) model ing the Phase 2 and Phase 3 regimes. During Phase 2, that the average daily entrainment rates into the characterized by strong wind-driven entrainment, 6 3 1 surface layer in Saldanha Bay are 20 10 m day NO3 supply rates to the upper euphotic zone are (Spolander, 1996; Monteiro et al., 1998). This large and allow phytoplankton biomass to bloom entrainment results in a rectified flow of water through without nutrient limitation. During this phase the the system, similar to estuarine circulation: in at the main limitation on production is the light flux for bottom and out at the surface. Therefore, during a photosynthesis. The rapid decline of turbulence in windy stratified period of 3 days, an additional Phase 3 limits the supply of NO3 across the thermo- 6 3 60 10 m of surface water leaves the bay. Some cline which leads to the rapid depletion of NO3 in fraction of this is old surface layer water. If all of the the surface layer. During this phase the physical exported water is ‘ old ’, then >90% of the volume of supply of NO3 becomes the main factor limiting the Saldanha Bay would be exchanged in 12–16 days. If rates of new production. the flushing period is taken as the e-folding time, then this would give a flushing period of less than one cycle (6–8 days). Carrying capacity for filter-feeder production Phytoplankton new production in Saldanha Bay Biogeochemical and ecological implications of governs the food supply and hence the carrying density driven exchange capacity of the bay for industrial scale farming of filter feeders (Grant et al., 1998; Pitcher & Calder, 1998). Phytoplankton new production The magnitude of the carrying capacity is dependent Density-driven intrusion of upwelled water is the only on both the NO3 flux from intruded upwelled water, natural mechanism by which the phytoplankton which governs the upper limit, and the proportion of Density driven shelf-bay exchange 889 the resulting new production exported from the bay distinct ecosystems as a result of the interaction be- domain, which governs the lower limit. It is the tween basin geomorphology and the dynamics of the magnitude of this latter fraction that limits the food- density-driven exchange between the bay and the supply-based carrying capacity for mussel farming in coastal ocean. The boundary between the two eco- Saldanha Bay (Grant et al., 1998; Smaal et al., 1997/ system types is the 5-m isobath and it applies to both 1998). The improved understanding of the dynamics the Langebaan Lagoon as well as the shallow parts of of the exchange, particularly net surface outflow over the eastern shore of Saldanha Bay. upwelling cycles has allowed the uncertainties of carrying capacity estimates to be constrained to real- Water quality considerations istic levels. As discussed earlier, the magnitude of phytoplankton new production exported out of the This large-scale subtidal exchange with the ocean is bay in the surface layer is a combination of advective also important in terms of the potential impacts of losses due to the rectified wind-driven circulation and eutrophication over the inner shelf and the fate of diffusive losses due to in-and-out movement of sur- pollutants discharged into the bay (Monteiro et al., face water in the presence of biomass gradients across 1997, 1999). On the one hand, variable basal intru- the bay-ocean boundary. The net outflows of surface sions minimize the chances of long-term ecological layer water from Saldanha Bay also limits the oppor- stress, which could arise from a persistent seasonal tunities for coastal blooms to be imported. This is of stratification, but on the other hand, these intrusions importance to the filter feeder farming industry can important hypoxic or anoxic inner shelf waters because it lowers the risk of exposure to harmful algal into Saldanha Bay (Monteiro, unpubl. data). These blooms (HABs) which are common in the coastal can therefore play a key role in the variability of redox waters offshore of Saldanha Bay in the late summer and pH, which are state variable for the partitioning of period (Pitcher & Boyd, 1996). pollutants between solid and dissolved phases in the bay. The relatively low turbidity and oligotrophic character of these shallow (<5 m) parts of the system De-coupling of Saldanha Bay and Langebaan Lagoon make them highly sensitive to eutrophication resulting ecosystems from the land based discharge of nutrients. Evidence The intrusion of cold NO3 rich water into Saldanha for this sensitivity has been observed through the Bay attains equilibrium between density-driven forc- occurrence of large Ulva sp. blooms in both Saldanha ing, surface heating and mixing induced by wind when Bay and Langebaan Lagoon linked to fish processing the top of the thermocline is at approximately 5 m waste and storm water/sewage effluents (Monteiro depth. This means that those parts of the system et al., 1997). which are less than 5 m deep, which includes the ecologically unique Langebaan Lagoon, will not be Conclusion exposed to a direct NO3 supply and will not support significant phytoplankton new production. The impli- This study was able to support the hypothesis that the cation that those shallow areas are largely oligotrophic observed subtidal variability in stratification in optically clear waters is borne out by observation. A Saldanha Bay was primarily driven by the baroclinic limited magnitude of phytoplankton biomass derived pressure gradients between the coastal and bay from production in Big Bay reaches Langebaan domains. The time scales of this variability (6–8 days) Lagoon through the active tidal pumping exchange are linked to those of upwelling-favourable wind mechanism through the narrow channels connecting stress. Wind drives the observed stratification this lagoon to the bay proper (Shannon & Stander, dynamics in Saldanha Bay in two ways. Through 1977). This small export flux of phytoplankton bio- coastal upwelling, regional winds drive the variability mass from Saldanha Bay supports a limited amount of of coastal water temperatures that govern bay–ocean filter feeder production at the northern end of the exchange. At the same time, local winds drive vertical lagoon. However, the energy supply to the system as a mixing and entrainment of intruded upwelled waters. whole is largely supported by regenerated production A 4-phase conceptual model was formulated to of benthic diatoms in the subtidal areas and salt marsh account for the time dependence observed. The vegetation in the inter-tidal areas. This regenerated coupling between horizontal exchange and vertical production system supports the highest diversity of entrainment has important implications for the benthic invertebrates in South Africa (Day, 1959; Saldanha Bay ecosystem. These include governing Christie & Moldan, 1977). The Saldanha–Langebaan phytoplankton new production, export of new pro- system is therefore divided into low ecologically duction, limiting the risk of the incidence of harmful 890 P. M. S. Monteiro and J. L. Largier algal blooms in the bay, and dividing the system into Coastal and Estuarine Dynamics (Aubrey, D. G. & Friederichs, eutrophic and oligotrophic sub-systems. C. T., ed.). Coastal and Estuarine Studies (A.G.U), vol. 53, pp. 227–241. Largier, J. L., Hollibaugh, J. T. & Smith, S. V. 1997 Seasonally hypersaline estuaries in Mediterranean-climate regions. Estuarine, Acknowledgements Coastal and Shelf Science 45, 789–797. Largier, J. L., Smith, S. V., Hollibaugh, J. T. & Fischer, A. 1998 This work was conducted as a collaborative research Interactions between ocean and river forcing of estuarine thermo- effort with financial support from the CSIR, Sea haline structure: Tomales Bay, California. To be submitted to Fisheries Research Institute (SFRI), the Foundation Estuaries (in prep.) Lenz, S. J. 1992 The surface boundary layer in coastal upwelling for Research Development (FRD) and the Depart- regions. Journal of Physical Oceanography 24, 1517–1539. ment of Oceanography at the University of Cape Monteiro, P. M. S., Anderson, R. J. & Woodborne, S. 1997 15Nas Town. We thank Prof. Geoff Brundrit and Roy van a tool to demonstrate the contribution of fish-waste derived Ballegooyen for constructive discussions and for nitrogen to an Ulva bloom in Saldanha Bay. South African Journal of Marine Science 18, 1–9. handling the time series data, Bruce Spolander for Monteiro, P. M. S., Spolander, B., Brundrit, G. B. & Nelson, G. assistance in fieldwork and data reduction. Lucille 1998 Shellfish culture in the Benguela System: estimates of Schonegevel for preparing the temperature contour nitrogen driven new production in Saldanha bay using two physical models. Journal of Shellfish Research 17, 3–13. plots and Gordon Bosch for extracting water volume Monteiro, P. M. S., Luger, S., Pretorius, J. P. & van Ballegooyen, data from the GIS. R. 1999 Numerical modelling of eutrophication and particle dynamics in Saldanha Bay in order to predict the transport and fate of trace metals using the Delft3D-FLOW and -WAQ models. References Submitted to the South African Journal of Science Pitcher, G. C. & Boyd, A. J. 1996 Across-shelf and alongshore Andrews, W. R. H. & Hutchings, L. 1980 Upwelling in the dinoflagellate distributions and the mechanisms of red tide Southern Benguela Current. Progress in Oceanography 9, 1–81. formation within the southern Benguela upwelling system. In Brown, P. C. & Hutchings, L. 1987 The development and decline Harmful and Toxic Algal Blooms (Yasumoto, T., Oshima, Y. of phytoplankton blooms in the southern Benguela upwelling & Fukuyo, Y., eds). Intergovernmental Oceanographic system: 2. Nutrient relationships. In The Benguela and Comparable Commission of UNESCO, Paris, pp. 243–246. Ecosystems (Payne, A. I. L., Gulland, J. A. & Brink, K. H., eds). Pitcher, G. C. & Calder, D. 1998 Shellfish culture in the Benguela South African Journal of Marine Science 5, 393–410. System: phytoplankton and the availability of food for com- Chadwick, D. B., Largier, J. L. & Cheng, R. T. 1996 The role of mercial mussel farms in Saldanha Bay, South Africa. Journal of thermal stratification in tidal exchange at the mouth of San Diego Shellfish Research 17, 3–13. Bay. In Buoyancy Effects on Coastal and Estuarine Dynamics Pollard, R. T., Rhines, P. B. & Thompson, R. O. R. Y. 1973 The (Aubrey, D. G. & Friederichs, C. T., eds). Coastal and Estuarine deepening of the wind mixed layer. Geophysical Fluid Dynamics 4, Studies (A.G.U), vol. 53, pp. 153–174. 381–404. Christie, N. D. & Moldan, A. 1977 Distribution of benthic macro- Shannon, L. V. & Stander, G. H. 1977 Physical and chemical fauna in Langebaan Lagoon. Transactions of the Royal Society of characteristics of water in Saldanha Bay and Langebaan Lagoon, South Africa 42, 273–284. South Africa. Transactions of the Royal Society of South Africa 42, Day, J. H. 1959 The biology of Langebaan Lagoon: a study of the 441–460. effect of shelter from wave action. Transactions of the Royal Society Shannon, L. V. 1985 The Benguela Ecosystem Part I: Evolution of of South Africa 35, 475–547. the Benguela, physical features and processes. In Oceanography Graham, W. M. & Largier, J. L. 1997 Upwelling shadows as and Marine Biology. An Annual Review. 23 (Barnes, M., ed.). nearshore retention sites: the example of northern Monterey Bay. Aberdeen University Press, Aberdeen, pp. 105–182. Continental Shelf Research 17, 509–532. Grant, J., Stenton-Dozey, J., Monteiro, P., Pitcher, G. & Heasman, Shannon, L. V. & Nelson, G. 1996 The Benguela: Large scale K. 1998 Shellfish culture in the Benguela System: a carbon features and processes and system variability. In The South budget of Saldanha Bay for raft culture of Mytilus galloprovincialis. Atlantic: Present and Past Circulation (Wefer, G., Berger, W. H., Journal of Shellfish Research 17, 41–49. Siedler, G. & Webb, D., eds). Springer-Verlag, Berlin. Hutchings, L., Pitcher, G. C., Probyn, T. A. & Bailey, G. W. 1995 Smaal, A. C., Prins, T. C., Dankers, N. & Ball, B. 1997/1998 The chemical and biological consequences of coastal upwelling. Minimum requirements for modelling bivalve carrying capacity. In Upwelling in the Ocean: Modern Processes and Ancient Records Aquatic Ecology 31, 423–428. (Summerhayes, C. P., Emeis, K.-C., Angel, M. V., Smith, R. L. Smith, R. L. 1995 The physical processes of coastal upwelling & Zeitzschel, B., eds). John Wiley & Sons, Chichester, p. 65–82. systems. In Upwelling in the Ocean: Modern Processes and Ancient King, F. D. & Devol, A. H. 1979 Estimates of vertical eddy Records (Summerhayes, C. P., Emeis, K.-C., Angel, M. V., diffusion through the thermocline from phytoplankton nitrate Smith, R. L. & Zeitzschel, B., eds). John Wiley & Sons, uptake rates in the mixed layer of the eastern tropical Pacific. Chichester, pp. 39–64. Limnology and Oceanography 24, 645–651. Spolander, B. 1996 Entrainment in Saldanha Bay. M.Sc. Thesis. Largier, J. L. 1996 Hydrodynamic exchange between San Francisco Department of Oceanography, University of Cape Town, South Bay and the ocean: the role of ocean circulation and stratification. Africa. In San Francisco Bay (Hollibaugh, J. T., ed.). AAAS Pacific Weeks, S. J., Boyd, A. J., Monteiro, P. M. S. & Brundrit, G. B. Division, pp. 69–104. 1991 The currents and circulation in Saldanha Bay after 1975 Largier, J. L., Hearn, C. J. & Chadwick, D. B. 1996 Density deduced from historical measurements of drogues. South African structures in ‘ low flow estuaries ’.InBuoyancy Effects on Journal of Marine Science 11, 525–535.