http://www.mesopp.eu/

Title

Name (affiliation)Name (affiliation)

Linking the mesopelagic components of ecosystem models

to acoustic vertical echosounder observations

Roland Proud1*, Martin J. Cox2, Rudy Kloser3, Nils Olav Handegard4, & Andrew S. Brierley1

1Pelagic Ecology Research Group, Scottish Oceans Institute, Gatty Marine Laboratory, University of St Andrews, UK, 2Australian Antarctic Division, Kingston, Tasmania, Australia, 3CSIRO Oceans and Atmosphere Flagship, GPO Box 1538, Hobart, TAS

7001, Australia, 4Institute of Marine Research, PO Box 1870, Nordnes, 5817 Bergen, Norway.

©MESOPP consortium

Modelling Ecosystems

• Food-web models e.g. Size-based models (figure; Blanchard et al. 2009, 2011, 2012).• End-to-end models e.g. Atlantis.• Habitat models.

Framework for open ocean ecosystem monitoring with acoustics – integrated with ecosystem models

Fish and Fisheries 2012

©MESOPP consortium

Mesopelagic component

• Formed of nekton and zooplankton.• Resides between 200 and 1000m during the daytime.• Migrates daily to feed at the surface at night (Diel Vertical Migration; DVM).• DVM facilitates the biological carbon pump.

©MESOPP consortium

Uncertainty

Estimates of global (open-ocean) mesopelagic fish biomass1 Gt Gjoseter and Kawaguchi (1980) Ocean trawl data14.3 to 19.5 Gt Irigoien et al. (2014) Acoustics< 1.4 Gt Jennings and Collingridge (2015) Macroecological model2.4 Gt Anderson et al. (2018) Food-web model1.8 to 16 Gt Proud et al. (2018) Acoustics

• Tropic efficiency may vary between 5 and 20% per trophic level.• Most ecosystem models do not account for this variability.• We need to use observations to validate mesopelagic component of

ecosystem models…Picture from Acoustic Laboratory for Ecological Studies (http://aleslab.blogspot.co.uk/2012/08/)

©MESOPP consortium

Acoustic Observations (IMOS)

Public data since 2010

16 vessels monitored

8 produce >80% of data

6 of these fishing vessels~ 300,000 km – focus multi-frequency vessels at 18 and 38 kHz

www.imos.org.au

©MESOPP consortium

Acoustic-Ecological Model reference data set most at 38 kHz

http://www.mesopp.eu/data/catalogue/Following ICES standardised metadata format for acoustic data

©MESOPP consortium

Acoustic selectivity at 38 kHz

Gas-bladdered fish and siphonophores are the dominant sources of backscatter in the mesopelagic zone.

Factor of 30 Factor of 1000

©MESOPP consortium

Acoustic observation model

Predict acoustic signal from output of ecosystem models and then compare to acoustic observations.

Group 1

Group 1

Group 2

Ecosystem

©MESOPP consortium

Predicting acoustic signal

To predict acoustic backscatter (sa, area-backscattering coefficient) from ecosystem model output we need to know the size and depthdistributions of the group and also the proportion of the group that have gas bladders:

sa ≅Abundance × proportion × 10 TS(size, depth) 10

Area

Where TS is the mean target strength of gas bladders and the area is specified by the ecosystem model. This can then be compared to acoustic observations:

sa = 200

1000

svdz ,

Where sv is volume backscattering coefficient and z is depth.

©MESOPP consortium

Predicting acoustic signal

Where that information is not output from the ecosystem model, we use values taken from the literature:

Size

range

(mm)

Depth

range

(m)

Orientation Gas-

bladder

volume (%

of body

volume)

Length-

width

ratio

Proportion

with gas-

bladders

Fish 20-

300R1

200-

1000

N(0,30) 0.01-2.63R3 4-12R4 0-1

Siphonophore

pneumatophore

0.01-

40R2

200-

1000

N(0,30) NA NA 0-1

R1: www.fishbase.orgR2: (Barham 1963, Robison et al. 1998, Lavery et al. 2007)R3: (Yasuma & Yamamura 2010)R4: (Flynn et al. 2012)

©MESOPP consortium

Predicting acoustic signal

We also need to consider spatial and temporal resolutions of ecosystem models:

Scale Atlantis Ecopath

(high

latitude)

Ecopath

(Kerguelen)

SEAPODYM MIZER WOMBAT

Space Multiple

irregular

polygons

Single

region

Single region Various

(0.08°- 2°)

2 boxes Various

(0.1-1°)

Depth 10 depth

layers

NA NA 6 depth

layers based

on euphotic

depth

NA 2 depth

layers

Time User defined

(e.g. 12

hours,

monthly,

annually)

Monthly Monthly 7 days Various

(weekly,

monthly,

annual)

Various

(weekly,

monthly,

annual)

Large spatial and temporal extent require more acoustic data

©MESOPP consortium

Acoustic observation model

Acoustic observations can be partitioned into groups (e.g. anatomical). Multi-frequency methods

Group 1

Group 2

Frequency 1

Frequency 2

Faculty of Edit this on the Slide Master The University of Adelaide

Vessel acoustics pelagic habitat, what biologically

does this represent?

38 kHz

18 kHz

18 kHz >>

38 kHz

Highly

influenced by

resonance

scattering by

depth

18kHz <<

38 kHz

©MESOPP consortium

Summary

• We can validate the mesopelagic components of ecosystem models by converting predicted abundance/biomass to acoustic signal and comparing to echosounder observations i.e. Acoustic Observation Model.

• To convert model output to acoustic signal, we need information concerning the size and depth distribution of the group and also the proportion that have gas bladders.

• Information that is not provided by the ecosystem model can be obtained elsewhere (e.g. literature and catch data).

• We can use archived echosounder observations e.g. IMOS (www.imos.org.au) and MESOPP (www.mesopp.eu) to make the comparison – particularly at multi-frequencies.

• Partitioning echosounder observations into different anatomical groupings may help reduce uncertainty in the Acoustic Observation Model.