Confluence Region Research Articles

Time-averaged flow structure in the central region of a stream confluence

Previous process-oriented field studies of stream confluences have focused mainly on fluvial dynamics at or immediately downstream of the location where the confluent flows enter the downstream channel. This study examines in detail the spatial evolution of the time-averaged downstream velocity, cross-stream velocity, and temperature fields between the junction apex, where the flows initially meet, and the entrance to the downstream channel. A well-defined, vertically oriented mixing interface exists within this portion of the confluence, suggesting that lateral mixing of the incoming flows is limited. The downstream velocity field near the junction apex is characterized by two high-velocity cores separated by an intervening region of low-velocity or recirculating fluid. In the downstream direction, the high-velocity cores move inwards towards the mixing interface and high-velocity fluid progressively extends downwards into a zone of scour, resulting in an increase in flow velocity in the centre of the confluence. The cross-stream velocity field is dominated by flow convergence, but also includes a component associated with a consistent pattern of secondary circulation. This pattern is characterized by two surface-convergent helical cells, one on each side of the mixing interface. The helical cells appear to be the mechanism by which high-momentum fluid near the surface is advected downwards into the zone of scour. For transport-ineffective flows, the dimensions and intensities of the cells are controlled by the momentum ratio of the confluent streams and by the extant bed morphology within the confluence. Although the flow structure of formative events was not measured directly in this study, documented patterns of erosion and deposition within the central region of the confluence suggest that these events are dynamically similar to the measured flows, except for the fact that formative flows are not constrained by, but can reshape, the bed morphology. The results of this investigation are consistent with and augment previous findings on time-averaged flow structure in the downstream portion of the confluence. © 1998 John Wiley & Sons, Ltd.

Hepatic heterogeneity in the response to ATP studied in the bivascularly perfused rat liver.

The zonation of the purinergic action of ATP in the hepatic parenchyma was investigated in the bivascularly perfused rat liver by means of anterograde and retrograde perfusion. Livers from fed rats were used, and ATP was infused according to four different experimental protocols: (A) anterograde perfusion and ATP infusion via the portal vein; (B) anterograde perfusion and ATP via the hepatic artery; (C) retrograde perfusion and ATP via the hepatic vein; (D) retrograde perfusion and ATP via the hepatic artery. The following metabolic parameters were measured: glucose release, lactate production and oxygen consumption. The hemodynamic effects were evaluated by measuring the sinusoidal mean transit times by means of the indicator-dilution technique. ATP was infused during 20 min at four different rates (between 0.06-0.77 micromol min[-1] g liver[-1]; 20-200 microM) in each of the four experimental protocols. The results that were obtained allow several conclusions with respect to the localization of the effects of ATP along the hepatic acini: (1) In retrograde perfusion the sinusoidal mean transit times were approximately twice those observed in anterograde perfusion. ATP increased the sinusoidal mean transit times only in retrograde perfusion (protocols C and D). The effect was more pronounced with protocol D. These results allow the conclusion that the responsive vasoconstrictive elements are localized in a pre-sinusoidal region; (2) All hepatic cells, periportal as well as perivenous, were able to metabolize ATP, so that concentration gradients were generated with all experimental protocols. Extraction of ATP was more pronounced in retrograde perfusion, an observation that can be attributed, partly at least, to the longer sinusoidal transit times. In anterograde perfusion, the extraction of ATP was time-dependent, a phenomenon that cannot be satisfactorily explained with the available data; (3) ATP produced a transient initial inhibition of oxygen uptake when protocols A and B were employed. These protocols are the only ones in which the cells situated shortly after the intrasinusoidal confluence of the portal vein and the hepatic artery were effectively supplied with ATP. The decrease in oxygen consumption was more pronounced at low ATP infusions when protocol B was employed. These observations allow the conclusion that the former phenomenon is localized mainly in cells situated shortly after the intrasinusoidal confluence of the portal vein and hepatic artery. Oxygen consumption in all other cells, especially the proximal periportal ones, is increased by ATP; (4) In agreement with previous data found in the literature, glycogenolysis stimulation by ATP was more pronounced in the periportal region. The cells that respond more intensively are not the proximal periportal ones, but those situated in the region of the intrasinusoidal confluence of the portal vein and the hepatic artery.

Dynamics of the Brazil‐Malvinas Confluence based on inverted echo sounders and altimetry

We use data from Geosat altimeter and from 10 inverted echo sounder (IES) moorings deployed in the SW Atlantic Ocean off the Argentine continental shelf to investigate several aspects of the dynamics of the upper layer in the Brazil‐Malvinas Confluence region. We use the altimeter data to estimate the sea height anomalies at each IES location and use the IES data to compute the upper layer thickness, taken in this work to go to the depth of the 8°C isotherm. We first discuss the sea height and upper layer thickness variations caused by the passage of the Brazil Current, Malvinas Current, and warm anticyclonic and cold cyclonic eddies. We introduce a two‐layer model in which we decompose the sea height into its baroclinic and barotropic contributions. We then propose a method to monitor the thickness of the upper layer and the barotropic and baroclinic transports as a function of the sea height anomalies and the statistics of the upper layer thickness and reduced gravity for the region. We compute the reduced gravity values from the slope of a linear fit between the sea height anomalies and the upper layer thicknesses. We estimate the reduced gravity values for this region to range from 0.005 to 0.011 m s−2. We also estimate the mean barotropic sea height difference using two methods: conservation of mass and conservation of potential vorticity. Finally, we compute the time series for the baroclinic and barotropic transports during the Geosat Exact Repeal Mission time period. Our results suggest that the mean baroclinic transport in the upper layer decreases from 12 Sv at around 35°S to 7 Sv at 37°S. Our results also indicate that there is a significant barotropic contribution to the upper layer transport in the confluence region.

Experiment studies circulation in the western South Atlantic

Observations in the Brazil Current region between 20° and 30°S are revealing many interesting oceanographic features. In the austral winter of 1993, what appears to be an anomalous northward penetration of waters from the Brazil/Malvinas confluence region into the South Brazil Bight was observed (Figure 1).

Assimilation of Sea Surface Height Data into an Isopycnic Ocean Model

Abstract Two different methods for assimilating sea surface height data into an isopycnic primitive equation model were developed and tested for the idealized case of an analytical Gaussian warm core (anticyclonic) eddy, then implemented using remotely sensed data from the Brazil–Malvinas confluence region. The first method makes a geostrophic assumption about the flow to relate sea surface height field gradients to the model velocities. The second method nudges the model sea surface height itself toward the observed values using a linear vertical influence function in the upper layers. The relationship between the surface height observations and the layer interface displacements is derived from observations of eddies in different regions of the Atlantic Ocean. Both assimilation methods were successful in transferring the dynamical influence of the sea surface height measurements deep into the water column, but a combination of both gave the best results. The application of both methods reproduced the det...

Hydrographic conditions in the surface layers over the slope-open ocean transition area near the Brazil-Malvinas confluence during austral summer 1990

The Confluence 3 cruise during February 1990 provides the first hydrographic survey of the Brazil-Malvinas Confluence region with such a refined spatial resolution over the continental slope. It is also the first survey in the region ever obtained in an austral summer. The austral summer 1990 atmospheric conditions were anomalously warm over the region. The surface waters were separated from the underlying waters by a thin and sharp seasonal thermocline which hid the real location of the Brazil-Malvinas thermocline front on infrared satellite imagery. The surface salinity was influenced by the large Rio de La Plata runoff which was anomalously high by a factor of 3. Salinity values less than 32 psu were encountered. These low salinity waters were entrained by the surface east-west front away from the coast. Four types of fronts were observed in the surface layer: (i) an east-west front associated with the structure of the seasonal thermocline separating cold, low salinity, oxygen supersaturated and productive waters to the south from warm, salty, weakly undersaturated and poor chlorophyll a waters to the north, (ii) a north-south front in salinity due to the outflow of the Rio de La Plata, (iii) a shelf break front between the shelf water and the Malvinas Water characterized by higher Chl a values, and (iv) a front between the Malvinas Water and the Brazil-Malvinas return flow.

RIVER CONFLUENCES ON ALLUVIAL STREAMS WITH EQUAL DISCHARGE AND EQUAL BED WIDTHS

The experiment, for river confluences on alluvial streams at 450 are undertaken. Two models with different conditions are carried out. Suitable mathematical models to simulate river confluence problems are designed. The experiment data are used to calibrate the mathematical models. The flow pattern, flow velocity vectors and bed configuration especially at the confluence region are studied comparatively with the experiment results. Each term of momentum equation for both fixed flat and movable bed are investigated along the main channel. The effect of eddy viscosity on flow pattern on the right side of main channel at the junction is considered. Finally satisfied results and good agreement between physical and numerical models are gained.

Hydrographic conditions in the Brazil‐Malvinas Confluence during austral summer 1990

The Confluence 3 cruise during February 1990 provides the first hydrographic survey of the Brazil‐Malvinas Confluence region ever obtained with such a refined spatial resolution over the continental slope. The closely spaced stations on the slope permit the observation of the branching of the Malvinas Current near 40°S. The eastern branch returns south, while the eastern one is squeezed along the continental slope, managing to continue northward as far north as 36.3°S. The Brazil‐Malvinas (B‐M) front has a north‐south orientation on the slope and an east‐west direction offshore. The B‐M front is marked by very strong gradients in temperature and salinity. Fine scale vertical structures and intrusions are numerous in the vicinity of the front and cover a wide range of scales. The largest intrusion observed is a 50‐km‐wide, 350‐m‐thick subsurface lense of Thermocline Water present on the Malvinas side of the front. The fine scale vertical structure results in very high temperature and salinity fluxes across the front of the order of 2×10−2°C m−1 s−1 and 10−3 psu m−1 s−1. Two types of mode waters are identified as follows: Subtropical Mode Waters north of the front and Subantarctic Mode Waters on both sides of the front. The Subantarctic mode waters (SAMW) appear to recirculate near the western boundary. A young variety of SAMW (lighter, very oxygenated) propagates northward along the continental slope with the Malvinas Current. It detaches from the coast near 37°S and is strongly altered in the frontal region. It is then entrained southward with the Malvinas return flow and Brazil Current extension. The Confluence 3 hydrographic survey provides a good description of the North Atlantic Deep Water (NADW) and Circumpolar Deep Water (CDW) convergence near the western boundary. Strong, fine scale vertical structures are observed in the vicinity of the NADW/CDW front, and the associated temperature and salinity fluxes induced are strong, 5×10−3°cm/s and 2×10−3 psu.m/s, respectively.

Physical Oceanography of the Southwest Atlantic Ocean

FROM ITS LITTORAL margin to the open ocean, the westem South Atlantic (Fig. 1) is marked at all depths by circulation patterns and exchange processes that are centrally important to the region- al marine resources and local economies, and to the global flux of heat and dissolved substances. Among other important characteristics, the Southwest Atlantic (SWA) is characterized by the presence of the Brazil Current (BC), a warm west- ern boundary current that, while weaker than the Gulf Stream in terms of mass transport, is energet- ically comparable to its North Atlantic counterpart, particularly in the region of confluence with the northward-flowing Malvinas Current (MC) at approximately 38°S. Because of the wide range of issues needed to be understood in terms of the physical oceanography, this oceanic region has been addressed by several important scientific pro- grams, a few of which are listed in Table 1 and indicated on Fig. 2. Results from these and other programs are summarized here and recommenda- tions for future efforts are offered. Basic Patterns of Circulation Inshore from the northern Brazil Current, over a continental shelf that widens gradually toward the south, are regions where seasonal wind-driven upwelling of cold, nutrient-rich waters occurs. These are important elements of the shelf circula- tion and they have major impacts on biological productivity. In particular, the Cabo Frio region (Fig. 1) has long been known for its active wind- induced upwelling, and new measurements from surface drifters now show that a counter-circula- tion exists on the shelf in the Santos Bight inshore of the Brazil Current (Stevenson and de Souza, 1994). Occasionally, this may be associated with

Scale interaction in the Western Pacific Monsoon

The lower-tropospheric scale interactions occurring in the summer monsoon of the western North Pacific are reviewed and summarised in a conceptual model. Diabatic heating produces a circulation with similar characteristics to those that are observed. In the lower troposphere the advection of vorticity by the divergent wind produces a compact, and more intense response than in the upper levels. Subsequent phase dispersion westward, and group propagation eastwards, lead to a monsoon depression in convectively suppressed conditions, a westerly jet with cross-equatorial flow, and a strong confluence region to the east of the monsoon depression.

Weddell Sea shelf water in the Bransfield Strait and Weddell-Scotia Confluence

The unusual stratification of the waters in the Weddell-Scotia Confluence between the Scotia and Weddell Seas and in the Bransfield Strait is traced to the influence of shelf waters from the northwest Weddell Sea. The shelf waters span the density range encompassed by the warm, salty Circumpolar Deep Water (CDW) of the Antarctic Circumpolar Current, and the colder and slightly fresher CDW in the Weddell Sea. An isopycnal mixture of these three source waters flows eastward from the tip of the Antaractic Peninsula in to the Weddell-Scotia Confluence region, and westward north of the Peninsula, where it flows downslope to renew the deep waters of the Bransfield Strait. This mixing scheme can occur year-round, in contrast to some previous explanations of the stratification in the region, which relied on the (unobserved) winter convective overturn of the water column.

Ecological Structure among Migrant and Resident Seabirds of the Scotia--Weddell Confluence Region

We quantitatively assess seasonal changes in community structure and habitat selection among seabirds in the Scotia-Weddell Confluence region, Antarctica. Additionally, we discuss the biological and physical factors underlying the patterns. Data were derived from strip-transects on closely-coordinated multidisciplinary cruises that characterized the physics and biology during spring 1983, autumn 1986 and winter 1988. We describe for the first time ever for the Southern Ocean seasonal changes in seabird communities in terms of composition, using cluster analysis, as well as relative density and diversity among species. Sea-surface temperature, distance to the pack ice edge and ice type, all physical characteristics of habitat, were the most important environmental variables that affected assemblage composition

Wind‐forced variations in the Brazil‐Malvinas confluence region as simulated in a coarse resolution numerical model of the South Atlantic

The observed seasonal and interannual fluctuations in the Brazil‐Malvinas confluence region are investigated using a wind‐driven isopycnic coordinate model of the South Atlantic Ocean south of 10°S. The model is configured on a rotated Mercator grid with 2° horizontal resolution and five constant‐density layers in the vertical. In order to model the passage of the Antarctic Circumpolar Current (ACC) across the basin, the grid is augmented by a channel extension to the west of Drake Passage and east of 50°E, having the width of Drake Passage. A series of benchmark experiments with annual mean climatological forcing shows that (1) when bottom topography is included, one observes a reduction in Drake Passage transport in agreement with previous studies, as well as a northward shift in the Brazil Current separation latitude, (2) an increase in Drake Passage transport to realistic values does not cause any further northward shift in the separation point, and (3) the model is relatively insensitive to the choice of lateral boundary conditions. A second set of experiments, in which the forcing is by seasonal climatological wind data and the Drake Passage transport is relaxed to a constant annual mean value, indicates the presence of a semiannual signal in the annual transport cycles for the Malvinas Current and for the ACC through Drake Passage. That signal is significantly damped in the Brazil Current region, and the amplitude of each cycle is reduced in comparison to observations. When the value to which the Drake Passage transport is relaxed is allowed to vary in time, the semiannual wind‐forced oscillation in the Malvinas region remains evident, with additional superimposed variations related to the variations in the transport forcing. The final experiments are forced by a 10‐year data set of realistic wind stress values, providing 10‐year time series of model output for analysis of the interrelationships of the principal Southwestern Atlantic currents. A high correlation at the semiannual period is found to exist among the cycles of Drake Passage transport, Malvinas Current transport, and seasonal movements of the Brazil‐Malvinas confluence latitude, while the Brazil Current transport cycle exhibits a significant energy peak only at the annual period. We conclude that the locally wind‐forced semiannual signal south of the confluence is significantly damped before reaching the Brazil Current region by several factors: friction, the opposing flow of the current itself, and the inability of the Malvinas to penetrate the subtropical circulation that is confined to the upper model layers.

Spatial and temporal scales in altimetric variability in the Brazil‐Malvinas current confluence region: Dominance of the semiannual period and large spatial scales

Two years of Geosat data are used to investigate the space and time scales of the mesoscale variability in the Brazil‐Malvinas Current confluence region. The mesoscale activity is highly inhomogeneous and anisotropic. The inhomogeneity is characterized by low values of sea level variability in the Malvinas Current (less than 8 cm, i.e., eddy kinetic energy less than 150 cm2 s−2), intermediate values in the Brazil Current (typically 16 cm/800 cm2 s−2) and high values in the Brazil‐Malvinas frontal region (30 cm/1700 cm2 s−2). The anisotropy is marked with meridional variances of velocity that are typically three times larger than zonal ones in the region of maximum variability. Mesoscale variations are dominated by relatively large spatial scales and low‐frequency fluctuations. Contrary to similar spectra derived from Geosat data over the North Atlantic (Le Traon, 1991; Le Traon et al., 1990) or over the whole southern ocean (Chelton et al., 1990) there is very little energy at the annual period. In the mean spectrum, the energy is at least 3 times smaller at the annual period than at the semiannual period or at shorter periods. The semiannual signal, with length scales of about 400–500 km, has a clear northward propagation and is maximum in the frontal region. It is probably associated with the semiannual wave that dominates the atmospheric circulation in the southern hemisphere. However, the mechanism responsible for this semiannual signal in the frontal region remains unclear. Beyond this dominant semiannual frequency, the altimetry‐derived mesoscale fluctuations exhibit other energetic signals, especially signals at periods between 75 and 150 days and spatial scales of 500–600 km with a westward propagation which have characteristics consistent with the dynamics of barotropic Rossby waves.

Trophic structure in open waters of the marginal ice zone in the Scotia-Weddell confluence region during spring (1983)

The structure of the food web was investigated in open waters adjacent to the marginal ice zone in the southern Scotia Sea in spring 1983. Diets were defined for dominant zooplankton, micronekton, and flying seabird species and then aggregated by cluster analysis into feeding groups. Most zooplankton were omnivorous, feeding on phytoplankton, protozoans, and in some cases, small metazoans (copepods). Only two species were found to be exclusively herbivorous:Calanoides acutus andRhincalanus gigas. Micronekton were carnivores with copepods being the dominant prey in all their diets. The midwater fishElectrona antarctica was the dominant food item in seven of the nine seabird species examined. Cephalopods, midwater decapod shrimps and carrion were also important in the diets of a few seabird species. Comparison (cluster analysis) of diets in spring with other seasons (winter, fall) indicated that over half the species examined (18 of 31) had similar diets in all seasons tested. The significant intraspecific shifts in diet that did occur were attributable to regional, seasonal, and interannual effects. A scheme is presented that describes the major energetic pathways through the open water ecosystem from phytoplankton to apex predators. At the base are phytoplankton and protozoans which are the principal food resource for the biomass copepods and krill. Krill and the biomass copepods are the principal forage of the midwater fishElectrona antarctica which, in turn, is the central diet component of flying seabirds as well as important food for the Antarctic fur seal and cephalopods. Krill are a major diet element for the fur seal and cephalopods, and the principal food of the minke whale.

On the formation of central water and thermocline ventilation in the southern hemisphere

A statistical method is used to measure the similarity between the surface T-S characteristics of the region 30°–45°S with that of the vertical T-S relationship between 10° and 20°S. This follows Iselin's notion, first proposed in 1939 ( Transactions of the American Geophysical Union, 3, 565–599) that the Central Water of the subtropical oceans originates in the Subtropical Convergence (STC), where it is formed by subduction of the fluid through Ekman pumping out of the mixed layer, and propagates towards the tropics y isopycnal spreading. His classical schematic diagram shows how the surface T-S characteristics along a meridional track across the STC coincide with the T-S characteristics of a vertical profile through the thermocline at lower latitudes. Iselin's original concept depicts, in the geostrophic regime, a strictly zonal movement of water and neglects any reference to gyre-scale oceanic circulation within a basin. The technique described here accounts for subduction along the isopycnals, and includes both meridional and zonal circulation within and between the basins. Results are presented for each of the four seasons in the three major oceanic basins using the 1° square Levitus ( Climatological Atlas of the World Ocean, National Oceanic and Atmospheric Administration, Rockville, 1982) data set. An RMS statistics is calculated based on minimising the difference in spiciness between the T-S points that lie along the same isopycnal at the surface and at depth. (Spiciness is a quantity that compares the relative contributions of temperature and salinity to density change, and is defined by a certain orthogonality condition with density.) Results show that the water most likely to be subducted to form the thermocline waters of the entire tropical South Pacific (10°–20°S) from 155°E to 100°W is the surface water of the STC in the western South Pacific between 155°E and 130°W. Formation occurs mainly during winter, but evidence for subduction is found in other seasons. The results indicate the presence of a “shadow zone” in the southeastern section of the South Pacific gyre. For the Indian Ocean, no shadow zone is present, and highest correlations again occured during winter in the STC from 72°E to 110°E. In the South Atlantic Ocean, Central Water formation occurs both in the Brazil/Malvinas Current confluence region, and also in the subtropical western Indian Ocean, indicating leakage from the Agulhas retroflection.

On deep-current and hydrographic observations from a mudwave region and elsewhere in the Argentine Basin

Near-bottom current measurements from both sides of a mudwave, in an extensive mudwave region of the mid-Argentine Basin, are examined for evidence of lee waves. Such waves appear to exist when the across-crest flow exceeds a critical value [predicted by Flood (1988), Deep-Sea Research, 35, 973–983] of about 9 cm s−1. The observations agree with the predictions of Blumsack and Weatherly (1989, Deep-Sea Research, 36, 155–172) in three of four cases. However, if it is assumed that the flow perturbations induced by lee waves are shifted slightly downcurrent from that predicted by the Blumsack and Weatherly model, then there is agreement in all four cases. The mudwave region is characterized as having a strong (∼ 10 cm s−1), steady (eddy kinetic energy < mean kinetic energy) current. The westward flow is consistent with the anticyclonic abyssal gyre centered about the Zapiola Drift inferred by Flood and Shor (1987, Deep-Sea Research, 35, 973–983). Near-bottom flow measurements made further to the west, in the region of confluence of the Brazil and Malvinas Current Extensions, reveal relatively energetic fluctuations, and equatorward flowing deep western boundary currents along the continental slope. Along the continental rise a poleward flow is observed, consistent with that inferred by Reid (1989, Progress in Oceanography, 23, 149–244). Too-cold bottom layers were found to be common only on the lower continental slope (4000 m depth) and at the mid-basin, mudwave site.

Analysis of satellite sea surface temperature time series in the Brazil‐Malvinas Current Confluence region: Dominance of the annual and semiannual periods

We study the dominant periodic variations of sea surface temperature (SST) in the Brazil‐Malvinas Confluence region from a satellite‐derived data set compiled by Olson et al. (1988). This data set is composed of 202 sea surface temperature images with a 4 × 4 km resolution and extends over 3 years (from July 1984 to July 1987). Each image is a 5‐day composite. The dominant signal, as already observed by Podesta et al. (1991), has a 1‐year period. We first fit a single‐frequency sinusoidal model of the annual cycle in order to estimate mean temperature, amplitude, and phase at 159 points uniformly distributed over the region. The residuals are generally small (less than 2°C). The largest departures from this cycle are located either in the Brazil‐Malvinas frontal region or in the southeastern part of the region. Other periods in SST variations are identified by means of periodograms of the 159 residual time series in which the annual cycle has been substracted. The periodograms show that a semiannual frequency signal is present at almost every location. The ratio of the semiannual amplitude to the annual amplitude increases southward from 0% at 30°S to reach up to 45% at 50°S. In the south the semiannual signal creates an asymmetry, and the resulting (total) annual cycle has a cold period (winter) longer than the warm one (summer). In the frontal region the annual and semiannual signals have an important interannual variation. This semiannual frequency is associated with the semiannual wave present in the atmospheric forcing of the southern hemisphere. Differential heating over the mid‐latitude oceans and the high‐latitude ice‐covered Antarctic Continent has been suggested as the cause of this semiannual wave (Van Loon, 1967).

A mixed ray tracing/full wave ICRH mode conversion model

A mixed ray tracing/full wave code is presented. The adopted 1D full wave model gives a realistic description of the mode conversion in the ion-ion hybrid region as it accounts for temperature oblique incidence poloidal magnetic field and finite parallel electric effects. Each ray entering the confluence region is stopped and a full wave computation is subsequently made. After that, a set of subrays carrying the transmitted, reflected and converted power is relaunched. The code is suited to study scenarios where a non-negligible fraction of the power carried by the fast wave is lost to the ion Bernstein or the ion cyclotron wave.

Does prey preference affect habitat choice in Antarctic seabirds?

Diet composition of the members of 2 seabird species assemblages in the Scotia-Weddell Confluence region, Antarctica, was investigated during 3 seasons/years: spring 1983, autumn 1986, and winter 1988. One assemblage frequented the pack ice and the other was present in adjacent open waters; most members of the latter species assemblage vacated the Antarctic during winter. We sought answers to 2 questions: (1) Did the 2 species assemblages depend on food webs that differed substantially; and (2) Was there a trophic basis to explain why the pack-ice assemblage did not expand into the open water left vacant by the other during winter? To test an a priori hypothesis that diet was affected by habitat, diet samples were obtained from birds encountered in 3 habitats: open water, sparse concentrations of ice, and heavy ice cover. Cluster analysis showed broad overlap in seabird diet regardless of species, habitat (ice/water mass) or year. Seabirds exploited prey largely according to ranked availability, although they appeared to choose the larger fish and crustaceans over smaller crustaceans. Myctophids in particular, but also krill and squid, were the main prey groups. Diets did not separate on the basis of predator/prey size to any appreciable degree even though a lOOO-fold difference in predator size existed. Feeding success, as indicated by fullness of stomach, of the members of the 2 species assemblages was highest when in their respective usual habitats. We conclude that open-water seabirds lack the specialized foraging behaviors required to exploit the pack-ice environ­ ment and that open waters during the winter offer much poorer feeding conditions than those of the pack ice.