Use of moored acoustic instruments to measure short-term variability in abundance of Antarctic krill

Upward-looking acoustic Doppler current profilers (ADCPs) (300 kHz) and echosounders (125 kHz) were deployed on moorings at South Georgia to measure abundance of Antarctic krill continuously over several months. Echoes from krill were identified using the theoretical difference in echo intensity at 300 and 125 kHz and scaled to krill density using target strengths appropriate for krill in the region: krill size was determined from diet samples from furseals and penguins foraging near the moorings. A method using water flow past the moorings was developed to convert time-based acoustic observations of krill to area-based abundance estimates. Flow past the stationary moorings was treated analogously to motion along-track of a research vessel through a nominally stationary body of water during a conventional survey. The moorings thus provide a Eulerian view of variation in krill abundance. This is ecologically instructive for South Georgia, where krill are generally passive drifters on currents and where temporal fluctuations in abundance have significant consequences for krill-dependent predators. Moorings were positioned on routine research vessel survey transects, and validity of the mooring method was assessed by comparison of mooring and vessel observations. Krill density estimates from the moorings were not statistically different from vessel estimates in adjacent time periods. A time series of krill density from a mooring revealed step-changes that were not seen during short-term vessel surveys. Moorings deliver data over time scales that cannot be achieved from research vessels and provide insight on environmental factors associated with variation in krill abundance at South Georgia. Mooring data may aid ecosystem-based management. Acknowledgments The moorings described here were funded by a grant from the UK Natural Environment Research Council awarded under the Antarctic Funding Initiative to E.J.M., A.S.B., D.G.B., and M.A. Brandon (Open University). We are grateful to the masters, officers, and crew of RRS James Clark Ross for deployment and recovery of the moorings. We thank colleagues at Bird Island, South Georgia, for krill length-frequency data used to determine target strength, and we thank colleagues on RRS James Clark Ross for providing acoustic survey data used in the shipmooring comparison—in particular, Jon Watkins for his contribution to provision of these data and discussion of sampling issues, Mark Brandon and Sarah Jenkins for discussion of CTD data, Geraint Tarling for advice on ADCP processing, and John Simmonds for advice on implementation of the SONAR equation. We also thank logistics personnel at British Antarctic Survey for their efforts in recovering a mooring that was dragged prematurely from the seabed by a fishing vessel. Limnol. Oceanogr.: Methods 4, 2006, 18–29 © 2006, by the American Society of Limnology and Oceanography, Inc. LIMNOLOGY and OCEANOGRAPHY: METHODS ing summer data (almost no winter data exist) show apparent year-to-year variation in krill abundance at South Georgia, few data are available on variation within years, and the possibility that apparent interannual variability is a temporal alias of shorter-term, intra-annual variation cannot be discounted. In fact, on the few occasions when more than one research cruise per year has made measurements at the island, marked withinyear variations in krill abundance have been detected (for example 2 g m–2 wet mass in October 1997 versus 21 g m–2 in January 1998; Brierley et al., 1999b). Data from predators breeding ashore at South Georgia suggest that krill availability can vary significantly throughout the course of the breeding season (for example, significant changes in the foraging behavior of Antarctic furseals have been detected; Mori and Boyd, 2004), but a lack of corroborating independent at-sea observations has prevented direct verification of this. Evidence from elsewhere around Antarctica suggests that local krill abundance can change substantially within seasons (McClatchie et al., 1994; Siegel et al., 1998). Understanding the scale, timing, and causes of such variation is key to understanding, managing, and conserving the South Georgia pelagic marine ecosystem. Low krill abundance at the peak of the predator breeding season, for example, could have far graver ecosystem consequences than would krill scarcity during midwinter, and these two scenarios would have different implications for krill fishery management in an ecosystem context (Constable et al., 2000). Present at-sea sampling regimens cannot resolve such short-term differences in krill abundance, however, and it has been impossible so far to link predator breeding performance indices to krill abundance (i.e., to describe functional responses of predators to varying prey availability) because of the mismatch between the time scales of the vessel surveys (~2-week duration) and the breeding season (4 to 5 months) (Reid et al., 2005). Furthermore, it has not yet been possible to determine the causes of short-term changes in krill abundance at South Georgia. The krill population at the island is not self-sustaining, and it has been suggested that abundance changes partly due to fluctuations in the position of the Sub-Antarctic Circumpolar Current Front (SACCF) that acts as a conveyor belt transporting krill to South Georgia from the Antarctic Peninsula (Hofmann et al., 1998; Murphy et al., 1998; Thorpe et al., 2004). The SACCF is a dynamic feature that retroflects around the eastern tip of South Georgia. When it meanders toward the northern shelf, it may deliver krill to the island, in which case increases in krill abundance should be associated with the arrival of water with the properties (temperature and salinity) of the SACCF. To date, lack of contemporaneous oceanographic and krill abundance data have prevented this hypothesis from being tested empirically. We deployed moorings at South Georgia to measure krill abundance and oceanographic parameters continuously and so gain insight on possible variations in abundance, and causes thereof, that could not be achieved with conventional shipbased sampling. Use of moorings to investigate temporal variability in pelagic ecosystems is not new (e.g., Cochrane et al., 1994; Fischer and Visbeck, 1993; Tarling, 2003), and a forward look at priorities for zooplankton research (Marine Zooplankton Colloquium 2, 2001) identified the use of remote sensing tools such as acoustic moorings as key to future studies of zooplankton hotspots. To our knowledge, however, previous mooring-based studies have not attempted to scale the point observations they gather to account for variability in water flow past them. In regions where current velocity is not uniform, scaling for flow will be important because, to take an extreme example, a period of apparent sustained high acoustic backscatter (planktic biomass) could be due to a single discrete aggregation in an otherwise empty background remaining stationary over the mooring: in that case the view from the mooring would be of continually high abundance, whereas the more widespread regional view would be of generally low abundance. In the case of krill, which are characterized by extremely patchy distributions (most biomass is in compact, high-density swarms), setting mooring-based observations in a wider context is particularly important (for example, predators cannot be expected to forage for our convenience directly over moored instruments), and we need to understand both temporal and spatial variation in abundance (Trathan et al., 1993). In this article, we report a method we have developed that uses the rate of water flow past the mooring to scale timebased observations of krill abundance. Analysis of data from what we have called “virtual survey transects” enables quantitative biomass estimates to be calculated. We describe analysis of data from a mooring deployed off shelf at South Georgia between November 2004 and January 2005 to illustrate our method. The method has potential wide application for studies of planktic ecosystems. Materials and procedures Acoustic surveys are conducted routinely to estimate abundance of pelagic species such as herring, pollock, and krill (Simmonds and MacLennan, 2005). During these surveys, calibrated echo intensity data are recorded from pings transmitted downwards into the water column at regular intervals (typically 1 s) at one or more frequencies (typically 38, 120, and 200 kHz) from a research vessel progressing along predefined survey transects (typically at 10 knots). Each ping is timeand position-stamped (using GPS) and provides the volume backscattering coefficient (sv, m –1) for regular depth bins (the size of which depend on the rate at which the echosounder samples the echo wave) down the water column. Further averaging down the water column or integrating along track leads to volumetric (e.g., mean volume backscattering strength [MVBS], Sv, dB re 1 m –1) or areal (e.g., nautical area scattering coefficient [NASC], sA, m 2 n.mi–2) measures of sound scattering that can be scaled using target strength (TS, dB re 1 m2) to provide various measures of animal abundance. Target strength at a given frequency usually varies as a function of animal size, and physical samples of the target species have to be obtained to determine size (usuBrierley et al. Short-term variation in krill abundance