Abstract
We compare estimates of krill density derived from gliders to those from contemporaneous and previous ship-based surveys. Our comparisons cover several temporal and spatial scales within two strata around the northern Antarctic Peninsula (off Cape Shirreff on the north side of Livingston Island and in the Bransfield Strait). Our objective is to explore the feasibility of using gliders to supplement or replace vessel-based surveys of fishery resources. We deployed two long-duration Slocum G3 gliders manufactured by Teledyne Webb Research (TWR), each equipped with a suite of oceanographic sensors and a three-frequency (38, 67.5, and 125 kHz, each single-beam) Acoustic Zooplankton Fish Profiler. We used the acoustic data collected by these gliders to estimate biomass densities (g⋅m–2) of Antarctic krill (Euphausia superba). The two gliders were, respectively, deployed for 82 and 88 days from mid-December 2018 through mid-March 2019. Off Cape Shirreff, glider-based densities estimated from two repeat small-scale surveys during mid-December and January were 110.6 and 55.7 g⋅m–2, respectively. In Bransfield Strait, the glider-based estimate of biomass density was 106.7 g⋅m–2 during December–January. Contemporaneous ship-based estimates of biomass density, from a multi-ship broad-scale krill survey (Macaulay et al., 2019) restricted to the areas sampled by the gliders, were 84.6 g⋅m–2 off Cape Shirreff and 79.7 g⋅m–2 in Bransfield Strait during January. We compared two alternative krill-delineation algorithms (dB differencing and SHAPES); differences between biomass densities estimated by applying these algorithms were small and ranged between 4 and 7%. Alternative methods of sampling krill length-frequency distributions (LFDs) (nets or predator diets), which are required to convert acoustic energy to biomass density, also influenced the glider-based results. In Bransfield Strait, net-based estimates of biomass density were 6% less than those based on predator diets. Off Cape Shirreff the biomass density of krill estimated from a net-based LFD was 20% greater than that based on predator diets. Development of a variance estimator for glider-based biomass surveys is ongoing, but our results demonstrate that fisheries surveys using acoustically-equipped gliders are feasible, can provide density estimates to inform management, and may be conducted at lower cost than ship surveys in some cases.
Highlights
Long-term trends in the biomasses of fishes and mesozooplankton have traditionally been derived from ship-based surveys that integrate both acoustic and net sampling (e.g., Greenlaw, 1979; Johannesson and Mitson, 1983; Demer and Conti, 2003; Reiss et al, 2008; Fielding et al, 2014; Eriksen et al, 2016; Simonsen et al, 2017; Jech et al, 2018)
Excessive variability in vehicle roll and pitch can greatly impact acoustically-derived biomass estimates, but our gliders were relatively stable throughout their deployments
The length-frequency distribution (LFD) for krill collected from net tows conducted during the multi-ship survey were similar to the LFD from penguins that foraged in the Bransfield
Summary
Long-term trends in the biomasses of fishes and mesozooplankton have traditionally been derived from ship-based surveys that integrate both acoustic and net sampling (e.g., Greenlaw, 1979; Johannesson and Mitson, 1983; Demer and Conti, 2003; Reiss et al, 2008; Fielding et al, 2014; Eriksen et al, 2016; Simonsen et al, 2017; Jech et al, 2018). We believe that autonomous vehicles will become a mainstay of fisheries surveys because such vehicles provide opportunities to measure and monitor marine ecosystems in a manner that minimizes competition for ship time; provides a continuous, three-dimensional, realtime representation of the ecosystem state; and reduces operational expenses. A variety of autonomous vehicles have been used to collect data on marine ecosystems. These platforms include slowmoving (0.2–1.0 m s−1) buoyancy-driven gliders (Eriksen et al, 2001; Sherman et al, 2001; Webb et al, 2001; Davis et al, 2002; Rudnick et al, 2004), faster-moving (1–2 m s−1) surface airfoils (Ghani et al, 2014), and propeller-driven vehicles (Brierley et al, 2002; Moline et al, 2015). The support networks (e.g., ship resources for deployment and recovery plus laboratory and office infrastructure for maintenance and piloting) required to operate autonomous vehicles scale with the size of the vehicle
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
