To enhance our knowledge on the mesopelagic community and the importance of the organisms inhabiting this domain, key players in various regions of the world need to be identified, their abundance assessed, and their ecosystem function determined.

The mesopelagic domain is defined as the depth region between 200 and 1000 m depth. This deep twilight zone is challenging to study, in particular to obtain quantitative data on the abundance of organism that reign here.

To be able to explore this depth region a multitude of instrumentation and gear, new and traditional, are normally applied. In this work echosounders are a necessity to sample the entire mesopelagic portion of the water column simultaneously. Acoustic techniques have, however, some limitations. Sound waves can rapidly penetrate to ocean depths and backscattering from organisms and maybe physical boundaries can be visualized within seconds. However, higher frequencies (>120 kHz) do not reach as deep as the lower frequencies (≤ 38 kHz). Current acoustic systems are normally carried on the ship hull but can also be lowered many hundredth of meters using other platforms. This will obviously improve operation range for the higher frequencies. Traditionally, the echosounders were designed to transmit and receive at a single narrow frequency, and there was a need for several transducers to cover a greater band of frequencies. Today, these systems are adjoined with broadband systems that allows transmit and reception of chirp pulses, that is, over a frequency range. Along with new developments in backscattering models, our understanding of organisms backscattering has improved (Figure 1), allowing for advancement of taxonomic identification of mesopelagic and other scatterers.

Figure 1: Acoustic scattering classes

Various post-processing systems allows analysis of these acoustic signals (e.g. the Large Scale Survey System – LSSS), and an associated Acoustic Feature Library is developed to store exemplary acoustic characteristics of particular species or group of organisms. Figure 1 shows a principle for separating main acoustic classes: (1) gas-filled targets are resonant; (2) hard targets show an asymptotic increase with frequency; and (3) fluid-like targets shows fluctuation with increasing frequency.

Figure 2 shows acoustic variables for some acoustic library categories (ALC). The naming of the ALC indicates which species the training data are based. ALC_herring are acoustic features based on data verified to be herring, although the acoustic features of ALC_herring is likely to fit organisms acoustically similar to herring, such as sardine. Figure 2, right panel, shows the relative frequency response, r(f) = sA(f)/sA(38 kHz), which is the Nautical Area Scattering Coefficient at frequency f relative to the most common acoustic frequency used in fisheries acoustics, 38 kHz. The right panel shows the variable R(18)=10log10[r(18kHz)] plotted against R(200). The 2-dimensional plots show ellipses, although the data itself are multi-dimensional, i.e. hyper-ellipses. Little overlap between hyper-ellipses makes categorization (“species identification”) good.

Figure 2. ALC- Acoustic Library Categories. Relative frequency responses R(18 kHz) vs. R(200 kHz) for the ALC of importance for the case studies. The ellipses are statistical boundaries on each category that contain 90% of the observations.

Some mesopelagic species have small swimbladders and are therefore resonant in the commonly used frequency range. The resonance frequency depends on swimbladder-size and depth, so that the resonance frequency may not be one of the available frequencies. Figure 3 shows two acoustic categories shown above (ALC_resonant_18 and ALC_herring_nvg), and two additional ones where the features are extracted from Mueller’s pearlside (Maurolicus muelleri) and Glacier lantern fish (Benthosema glaciale). Figure 3 shows that the targets may be near, but not always at the resonance frequency.

Figure 3. Acoustic features of some mesopelagic species.

The use of various trawl types allows classification of targets caught in the trawl. Advanced instrumentation and photographic equipment can also be attached to trawl cod ends for electronic imaging of trawl catch along the trawl path (e.g. Deep Vision). In this way, we can visually identify and measure the size of the organisms detected by the acoustics and with some certainty say at which depth a specific organism has been caught — which is otherwise quite tricky since the physical catch measured on board is a mixture of everything captured along the path trawled. The Deep Vision system allows counting and taxonomic identification of a considerable portion of a catch without the trawl catch being brought on deck of the research vessel. See also https://deepvision.no/deep-vision-acoustic-integration.

Figure 4. Example echogram of acoustic data with trawl path (left). Images taken inside the trawl with Deep Vision, verify the presence of mesopelagic crustaceans and fishes at ~200 m depth (right).

As part of our contributions to SUMMER, we along with partners, will use some of the above techniques and supplementary methods to aid our understanding of the mesopelagic community, its composition, key organism abundance and biomass, as well as its connectivity with the epipelagic domain.

Text/Figures/Photo: Institute of Marine Research, Bergen (Tor Knutsen, Rolf Korneliussen, Shale Rosen).