Species-specific abundance and biomass

Species-specific abundance and biomass estimates are the end result of fisheries acoustics surveys.  Converting acoustical energy to abundance is a multi-step procedure.  Volume or area backscattering coefficients are scaled to numerical density [# m^-3] by σ_bs.  σ_bs are obtained from in situ measurements and/or TS-L regressions.  Densities are vertically integrated to give areal densities [m²•m^-2] along the cruise track, and areal densities are scaled to the survey area.  In freshwater assessment, areal density is typically given per ha or sometimes per m².  The marine convention is to scale areal density to square nautical miles [1 nmi² = (1852)²•m²].  All these steps have been outlined in this website and the SOP. 

The final step is to convert fish abundance to species-specific abundance and biomass.  This requires separation of densities by species or age groups through samples, location, or TS distribution.  Biomass is then calculated from this densities and observed weights or if necessary from a length to weight regression.  We have presented some approaches to the problem of separating acoustic density estimates by species.  This problem requires more attention as approaches to species estimation differ among the different Great Lakes and a consistent approach is needed to provide managers with useful fish abundance estimates.  This issue also requires a better understanding of trawl selectivity.  We envision this to be the main issue in need of further development.

Biomass can also be obtained directly from acoustic estimates using biomass specific TS equations (e.g., Warner et al. 2002) or by converting TS bins to fish lengths followed by conversion of fish lengths to biomass (e.g., Brandt et al. 1991).  Using biomass specific TS regressions to derive average σ_bs per g fish will probably give reasonable numbers as long as fish length distribution is relatively constant.  Biomass calculations from these equations will be biased when comparing biomass from small and large fish because acoustics backscatter is related to the area of individual fish, not the mass of individual fish.  The second method introduces unknown biases associated with the wide distribution of fish TS obtained from a single fish (see also Separating Groups of Interest) and the non-linearity in both L-W regressions and TS-L regressions.  Therefore, we cannot recommend the second method, although it is possible that positive and negative biases will offset each other.