The research plan will be implemented by a combination of Time-Series Surveys and Process Studies carried out in three regions of the Southern Ocean: the Antarctic Peninsula region, the eastern Weddell Sea, and the Indian Ocean sector. Field studies should be preceded by modelling. This section describes the criteria for field studies and suggests how key measurements be made.
Two specific field programmes are recommended:
The winter study would focus on overwintering strategies of target species and would last for a minimum of 6 months.
The summer study would focus on processes of foraging, reproduction and recruitment in relation to physical dynamics, and would last at minimum 6 months.
Three principal areas of operation are recommended (Figure 1). These are the:
The choice of these areas presents important opportunities for Southern Ocean GLOBEC. First, it is possible to make geographical comparisons with respect to key processes such as overwintering strategies of zooplankton, or foraging behavior of top predators. Second, it is possible to compare ecosystems with different top predators (e.g. a penguin-dominated system in the Peninsula region vs. a crabeater-dominated system in the eastern Weddell Sea). Third, it will be possible to compare the influences of sea-ice, which may vary greatly among regions.
The Antarctic Peninsula region is broadly defined to range from the Bellingshausen Sea in the west, to the Scotia Sea and South Georgia in the east, and to the Peninsula itself in the south. This region provides a rich historical background and an abundance of shore-based predator populations. It is an area where krill, as well as many other target species, are abundant. A number of countries routinely operate research vessels in this area, including Argentina, Brazil, Chile, China, Japan, Korea, Spain, UK and USA. Airfields are maintained at Rothera (UK) on Adelaide Island and Eduardo Frei Base (Chile) on King George Island. Shore stations operated by many nations provide the possibility for complementary process studies.
The eastern Weddell Sea region is centered on the Lazarev Sea, which encompasses the eastern arm of the Weddell Gyre. Target species of importance include krill and crabeater seals; Adelie penguin colonies are notably absent. Research vessels are regularly operated in this region by Germany, Russia, South Africa, and the Nordic countries (Finland, Norway and Sweden).
The Indian Ocean sector, broadly defined, ranges at its southernmost extremity from Enderby Land (35 degrees E) to the Shackleton Ice Shelf (100 degrees E). A number of studies have been carried out in coastal areas of this region, notably in Lutzow-Holm Bay (37 degrees E) and Prydz Bay (70 degrees E). Important target species in this region include krill, crabeater seals and Adelie penguins. Countries that routinely operate research vessels in the vicinity are Australia, China, France and Japan. Continental stations at Syowa (Japan), Davis (Australia) and Tunshan (China) are complemented by a year-round sub-Antarctic base on Kerguelen Island (France).
The selection of specific research sites in each of the three Southern Ocean GLOBEC study areas will be determined at the next stage of the implementation process. A regional planning meeting for each research area will be convened to choose a specific site and to coordinate activities in accordance with scientific and logistic criteria described in the following section.
Time-series surveys are designed to adequately resolve temporal and spatial scales relevant to the key questions posed. The guiding principle is that surveys be nested in both space and time.
4.2.1 Criteria for Time-Series Surveys
a) Time scales that resolve population dynamics and seasonal variability
The relevant time scale of processes implied by the key questions is of order 1 to 10 weeks. For example, the factors affecting reproduction of zooplankton, which includes maturation, feeding history, and lipid accumulation, occur on this time scale. Similarly, the role of sea-ice dynamics in affecting the foraging and reproductive success of many top predators is expected to become apparent on a time scale of 1 to 10 weeks.
Continuous observations are required for a minimum period of 6 months to adequately resolve the population dynamics of either zooplankton or top predators. This period of time is necessary to observe mechanisms that lead up to reproduction, that affect the survival of young, and that determine recruitment - which for most species occurs on an annual time scale.
b) Eddy-resolving spatial scales
Mesoscale circulation dynamics contribute significantly to the retention and re-distribution of animal populations in the pelagic ecosystem. Time-series surveys must therefore be capable of resolving the Rossby radius at high latitudes, which may be of order 10 km. To determine the total size of the survey region involves two sets of tradeoffs. An area that is too large will lack synopticity. On the other hand, an area that is too small would have to be sampled more frequently than is logistically feasible.
c) Comparability among survey sites
Survey design in the Antarctic Peninsula, Indian Ocean, and eastern Weddell Sea areas should be sufficiently similar to permit unambiguous comparison of results between time-series surveys. The shape of the survey might be altered to accommodate local oceanographic dynamics, but the overall size of the survey area and the timing of cruises should provide for comparable resolution in space and time.
d) Adjacency to shore-based predator populations
GLOBEC field studies that include shore-based predator populations (e.g. penguins) should cover a region that encompasses the ocean adjacent to the shore-based colony. The potential foraging range of the species under study must be considered when determining the dimensions of the study region.
e) Augmented effort along survey lines
Many nations ply standard supply routes to and from Antarctica (Fig. 1). An effort should be made to locate time-series survey sites along or near such routes so that supplemental underway measurements can be made (e.g. seal counts, bioacoustics, meteorology). Measurements along survey lines will augment temporal coverage and provide latitudinal comparisons.
f) Standard measurements
Measurements of key variables should be standardised both within and among time-series survey sites. For example, zooplankton should be sampled with the same type of nets and at the same depth horizons; predator populations should be censused using standard techniques. An outline of key measurements and techniques is given in Section 4.2.2.
4.2.2 Measurements and methods for Time-Series
Surveys
The following suite of measurements may be regarded as essential elements of a SO GLOBEC time-series survey; they are common to addressing many of the questions posed. For example, the stage-frequency distribution, depth distribution and abundance of zooplankton must be measured in time-series surveys in order to understand overwintering strategies (Zooplankton Question 5), factors affecting larval survival (Zooplankton Question 3), and the allocation of prey items among top predators (Predator Question 4). This section summarises recommended measurements, and suggests methods that might be used to make such measurements.
a) Stage-frequency distribution of zooplankton
Methods: nets, multi-nets, ROV, video plankton recorder, continuous plankton recorder.
b) Zooplankton abundance and depth distribution
Methods: nets, multi-nets, ROV, video plankton recorder, continuous plankton recorder, acoustics, optical plankton recorder
c) Physical dynamics and water mass properties
Methods: CTD, ADCP, Lagrangian drifters, moored instruments (e.g. current meters), satellite data (e.g. AVHRR), underway surface measurements
d) Distribution and production of particulate food
Methods: POC, PON, fluorometry, HPLC pigments, particulate ATP, species composition, particle counters, flow cytometry. Primary production measurements, microbial dynamics.
e) Sea-ice distribution and dynamics
Methods: Satellite, visual observations, meteorology (wind forcing), CTD-geostrophy and ADCP-local currents (fluid forcing).
f) Sea-ice characteristics
Methods: Satellite (e.g. SSMI), visual, coring, under-ice observations
g) UV-B radiation
Methods: Spectral radiometer
h) Predator population dynamics
Methods: Age structure methods (e.g. otolith analysis in fishes, tooth analysis in seals, etc.)
i) Predator distribution and abundance
Methods: Fishes & squid: nets, acoustics; Birds & seals: visual observations, aerial surveys, telemetry
j) Predator foraging behavior
Methods: Gut content analysis, behavioral observations, TDR
k) Predator reproduction
Methods: Fishes & squid: larval distribution, abundance; Birds & seals: behavioral observations, egg & pup production
The above measurements and methods are representative of those recommended for application during Southern Ocean GLOBEC time-series surveys. Detailed recommendations of appropriate methods are given in GLOBEC Report No. 5. It is recognised that additional measurements may be required in certain studies, and that not all measurements will be necessarily required in others. Furthermore, certain observations will be more powerful when carried out in conjunction with Process Studies (please see Section 4.3).
4.2.3 Generic Time-Series Survey design
This section describes the generic features of the time-series surveys, based on criteria given in Section 4.2.1. The mesoscale cruise plan, and its integration with Process Studies (Section 4.3), are outlined in Figure 2. The surveys should be planned carefully under consideration of local features pertaining to hydrography (current speeds and directions) and topography and possess the following characteristics:
a) Spatial scales
It is recommended that survey activities be carried out at three spatial scales: macroscale, mesoscale and fine scale. The macroscale (dimensions of order 1000 km) encompasses the mesoscale survey area, and is appropriate for observations of phenomena that impinge on smaller scales. Primary observational tools include satellite imagery, Lagrangian drifters and aerial surveys.
Mesoscale surveys would be accomplished by research vessels. It is recommended that, at the latitude of the Southern Ocean, spacing between grid lines and stations should not exceed 20 km. The area of interest would ideally be about 40,000 km2, or 200 km on a side. However, assuming a station time of 3 h, such a survey would require about 20 days under favorable weather conditions. As this is likely to be too long, a smaller area, e.g. 25,000 km2, or 160 km on a side might well have to be chosen, as this would take about 15 days to survey. Surveys of this extent should be adequate to characterise temporal variability on monthly time scales, which is within the 1-10 week time scale indicated for many key processes.
Fine scale surveys would be carried out before and after the mesoscale survey, using the same research vessel. These surveys would be nested within the mesoscale survey region. The area of interest would be approximately 1000 km2, or 30 km on a side. With spacing of grid lines and stations at 6 km intervals, and assuming a station time of 2 hrs, fine scale surveys could be completed within several days. Surveys of this extent might well be sufficient for characterising small scale variability in the horizontal. However, depending on the species and region under study, it might be more appropriate to choose smaller spacing between grid lines and stations.
b) Temporal scales
Relevant time scales range from min-hr (e.g. foraging behavior) to approximately 6 months (e.g. studies of overwintering strategy), with most falling in the range of 1-10 weeks. The purpose of the time-series surveys is to resolve processes occurring at these time scales. Surveys should be repeated in the area of interest at least once per month. A series of 2-week surveys will, over a 6-month period, provide unprecedented resolution of key processes.
c) Shape of survey areas
It is recommended that the shape of the survey be determined by the area in which it is conducted. Thus, a survey that crosses a broad frontal region might be made rectangular in order to accommodate features on either side of the front (e.g. 400 x 100 km rather than 200 x 200 km), or irregular to accomodate a shoreline. The total survey area should remain constant (i.e. 40,000 km2), in order to (a) retain a total survey period of approximately 2 weeks, and (b) not exceed available logistic resources.
d) Inclusion of survey lines
Traditional transit routes to and from the Antarctic continent provide the opportunity for adjunct studies that would supplement, but not replace, time-series surveys. Common transit routes (Fig. 1) include:
e) Multiple ship participation
Time-series surveys will need to be approximately 6 months long to adequately address key scientific questions. Three different vessels may be required to complete such a survey. Each vessel would commit to a minimum of a 6-week cruise, with the first and last 2-week periods being for time series surveys and the middle 2-week period reserved for process studies.
It will not always be feasible to carry out the relevant process studies for zooplankton and predators simultaneously, which is in contrast to Time Series Surveys where both categories of organisms can and should be monitored during the same cruise. In the following, zooplankton and predators are hence dealt with separately.
The following key questions which assess the impact of physical processes operating over different temporal and spatial scales on zooplankton population dynamics will be addressed by process studies:
Zooplankton question 3: Factors affecting successful reproduction
Zooplankton question 4: Physical processes influencing larval survival and recruitment to the adult population
Zooplankton question 5: Zooplankton overwintering strategies
Each of these processes will have to be studied in the context of the life history cycle of the various species identified as key organisms. This enables one to break down the processes into elements that can be studied individually.
a) Factors affecting successful reproduction (Question 3)
Assessing the success of reproduction will require study of each of the following aspects which are likely to be constrained by specific physical factors:
b) Larval survival and recruitment to the adult population (Question 4)
This is a key process in the life cycle of all organisms that is strongly dependent on environmental conditions but poses considerable difficulties for adequate assessment. There is a pressing need to develop methodology and model simulations to study the following component processes:
c) Overwintering strategies (Question 5)
Available information on the different overwintering strategies of the various species suggests that there is a gradient from total diapause of the entire population to active foraging throughout the winter. There also seem to be regional differences in overwintering strategy within the same species for which variation in factors triggering a particular type of winter behaviour may be responsible. In order to assess the gamut of overwintering behaviours the following studies need to be carried out:
Biochemical composition: These studies could utilise e.g. dry : wet mass ratios, protein and lipid ratios. In many cases it will be necessary to split the latter into wax esters, triglycerides and polar lipids.
Digestive system: Activity here could be assessed by, for example, histology and enzyme content or activity studies.
Process studies will be necessary to address the key questions 2-6 listed above (Section 3.2). Whereas questions 2 and 6 are focussed on performance of predators in relation to sea ice and on overwintering strategies respectively, the questions 3 - 5 encompass various aspects of predator/prey relationships. With regard to the latter questions, it should be pointed out that defining methods and approaches for studying the dynamic interactions between top predators and their prey has received much less attention than, for instance, the population dynamics of either zooplankton prey or top predators taken separately. The development of research programme elements to address each of the key questions will require cooperation by scientists with expertise on sea ice, prey ecology and top predators. Only a small proportion of the necessary research can be carried out from shore-based stations. The variety and complexity of the ship-based research necessitates (a) very careful planning of interactive studies; (b) dedicated time on Time Series Surveys; (c) and Process Study cruises dedicated to predator-prey research.
The following consists of an overview of the approaches and methods in use for the study of predator population dynamics and the relevant processes and mechanisms.
a) Population dynamics
b) Foraging ecology
This comprises studies of:
c) Energetics and physiology
d) Growth
Various aspects of growth reflect different aspects of interactions between predators and their environment. Growth (in mass and/or morphometrics) of dependent offspring reflects parental performance over the breeding season. Growth rates of juveniles and adults (chiefly accessible via annual or daily growth layers revealed by analysis of sections of teeth, otoliths and statoliths) integrate a range of interactions over daily to annual time scales.
e) Sampling and observation systems
A full definition of the sampling and observation systems required to address the predator-prey interaction element of SO-GLOBEC will need to await the definition of the research programmes required to address the key questions. However, it may be helpful at this stage to summarise some aspects of likely sampling requirements and to note those methods and sampling systems that are being, or will need to be, developed in order to have effective field programmes. GLOBEC Report No. 5 (Table 2) summarises some relevant details and highlights several key requirements in the field of development of new sampling systems.
At the first Southern Ocean GLOBEC science planning workshop, held in 1991, modelling was identified as a necessary and integral component of the programme. Results from that workshop (U.S. GLOBEC Report No. 5), recommended three essential criteria:
In June 1993 a second workshop was convened to further the development of a science plan; the recommendations from that workshop (GLOBEC Report No. 5) further highlighted the need for modelling as part of a SO GLOBEC programme. Two specific modelling activities were recommended:
These two activities should be emphasised in planning for the SO GLOBEC programme. Models should be developed in advance of field studies and then tested and improved with the incoming data.
a) Purpose
A conceptual model would provide a framework to (1) formulate, synthesise and address scientific questions, and (2) organise and direct field programmes. Key scientific questions posed in SO GLOBEC have the common goal of understanding the role of physical and biological interactions in determining population variability. The output of field programmes could then be integrated with the conceptual model to answer the central questions.
b) Examples of key questions
c) Approach
One approach for developing a SO GLOBEC conceptual model is to develop individual modules that describe interactions between the physical environment, food resources, intermediate predators, such as zooplankton and krill, and top predators. These modules represent choices for processes and their mathematical representation. Within each module are submodules that describe specific processes and potentially each module could stand alone. For example, the physical environment module could include large-scale circulation, mixed layer dynamics and sea-ice processes. The food resources module could allow for bio-optical approaches for estimating pelagic primary production and sea-ice algal production as well as pelagic and sea-ice heterotrophic production. The components of the other modules would be expanded similarly to include the needed processes. Within the biological processes modules sufficient detail needs to be allowed so that the important parts of biological life histories are included, i.e. size and stage structured populations. The development of the modules of the conceptual model should help in highlighting processes that are critical and thus need measurement.
d) Implementation
The development of a conceptual model for SO GLOBEC studies would consist of two major phases. First, the basic components of the model must be identified.
The model should then be reviewed, revised and published in the open literature. Through the peer review system a consensus can be reached as to the relevant components for the conceptual model. The second phase is to develop the actual models for the modules of the larger model. It is envisioned that the development of the modules for the conceptual model would be done concurrently by several groups with expertise in a particular area. However, this distributed approach to development of the conceptual model will require that it be done in such a way that the individual components are compatible. Thuis, it is important that the groups involved in the development of the conceptual model recognise that it is part of a larger effort.
a) Availability
At present a three-dimensional time-dependent circulation model with embedded biology that describes the Antarctic marine pelagic ecosystem does not exist. Most of the few models available for the Antarctic marine system have been designed for circulation studies. The Fine Resolution Antarctic Model (FRAM) provides estimates of the large-scale flow associated with the Antarctic Circumpolar Current (The FRAM Group, 1991). The model was formulated for open ocean thermohaline and geostrophic flows and for the most part the simulated distributions accurately reproduce flow at this scale. However, the accuracy of the simulated circulation in coastal and shelf regions is questionable since shelf circulation dynamics were not explicitly included in FRAM. More relevant, however, is that the spatial resolution of FRAM is not adequate for most of the key biological questions posed in SO GLOBEC.
Finer scale circulation models have been developed for some regions of the Antarctic (e.g. Capella et al., 1992). These models are more amenable to inclusion of shelf circulation processes and can provide simulated circulation distributions at scales that can be matched to many biological processes. An important aspect of these regional models is the inclusion of a sea-ice component.
Much of the biological activity in marine systems occurs in the upper portion of the water column, i.e. the mixed layer. In Antarctic regions, the stability (or lack of stability) of the mixed layer has been suggested as one of the important controls for biological production. Vertical and time-dependent sea ice-mixed layer models exist (e.g., Lemke et al. 1990). However, these have not yet been applied to a wide range of Antarctic environments.
b) Needed models and resources
The paucity of existing circulation models for the Antarctic may be regarded as a handicap to developing a strong modelling effort for SO GLOBEC. However, this is not necessarily the case. Several existing circulation models are generally available that can be adapted for the Antarctic. These include the Semi-spectral Primitive Equation Model (Haidvogel et al. (1991), the S-Coordinate Rutgers University Model (Song and Haidvogel, 1994), the GFDL Ocean Circulation Model (Cox, 1984), the Isopycnal Model (Bleck and Boudra, 1986; Oberhuber, 1986) and the Princeton Ocean Model (Blumberg and Mellor 1987). These models should be adapted to the needs of SO GLOBEC. Two general areas of emphasis are identified:
For the time frame prior to SO GLOBEC field studies, existing computer resources should be sufficient for development of needed circulation models. Limiting factors are likely to be the availability of data sets and human resources to develop and implement the models. The latter limitation could be overcome by making support available for students and researchers to do modelling studies.
a) Availability
The availability of models for any part of the Antarctic marine pelagic system is poor. To date, models that treat primary production processes, pelagic and sea-ice, are unavailable, although some are now in development. Similarly, models for higher trophic levels are few; examples include a modelling study for the early life stages of krill (Hofmann et al., 1992), and another for the population dynamics of Calanoides acutus (Huntley et al., 1994). Models need to be developed for older krill life stages, other zooplankton and top predators.
SO GLOBEC could make a major contribution to modelling and subsequent understanding of the Antarctic system. Many of the key scientific questions can be addressed by modelling studies, perhaps more easily than with field studies. Processes involved in krill overwintering, interactions of salps and krill, and the allocation of food (e.g. krill) among multiple predators, are just some examples of key scientific questions which models could resolve. Process studies and laboratory experiments could provide the inputs for some model parameters and the model could be used to integrate and synthesise observations.
b) Needed models and resources
a) Available techniques
Data assimilation is an area that is rapidly developing and is now being used with ocean circulation models (Brasseur and Nihoul, 1994). First attempts are now being made to use data assimilation with biological models (Lawson et al., 1994). Many of the issues associated with assimilation of data into biological models are discussed in GLOBEC Report No. 6; these issues are also relevant to SO GLOBEC. The development of assimilation methods for biological data should be encouraged and incorporated into SO GLOBEC programmes.
b) Needed models and resources
Assuming that data assimilation techniques for biological data are available, the limiting factor in application of these techniques is adequate data sets. The development of data assimilative models for the Southern Ocean system will require close cooperation between modellers and experimentalists. The importance of this cooperation cannot be overemphasised.