Beyond Chl-a: Are large-scale observatories of plankton

abundance and diversity possible with currently available

technologies?

Why are we here?

11/05/17 Soutenance HDR Fabien Lombard

(on the 40 already published papers)

- up to 8 papers from « Ocean obs » mentionning imaging and or optical methods

- up to 11 papers from « Ocean obs » mentionning plankton observations

+ Breakout session on Ecosystem Health and Biodiversity under discussions

Why are we here?

Plankton is central in most biogeochemical fluxes and have then a strong retroaction on climate

Plankton is the foundation of most marine trophic chains and is central for fisheries, aquaculture…

Plankton is very sensible to any environmental changes

Phyto- and Zooplankton abundance and diversity have been tagged as

Essential Ocean Variables (EOVs) by GOOS (Global Ocean Observing System)

Essential Climate Variables (ECVs) under GCOS (Global Climate Observing System)

11/05/17 Soutenance HDR Fabien Lombard

Why are we here?

Plankton, among other “biological variables”, is the more easy to

collect and survey at global scale : High scalability potential

meeting with high societal needs (RI)

Miloslavich et al 2018

Plankton characteristics and constrains

Wide range of size/taxa/functions

Number decreasing with size

Needed sampling volume increase with size < 1mL for nano (0.2-2µm) < 1 L for pico (2-20µm) < 1m3 for micro (20-200µm) <1000m3 for mesoplankton (200-2000)

Consequences: - we need water sampling + nets to sample the full range of plankton

11/05/17 Soutenance HDR Fabien Lombard Blanchard et al 2017

Desirables properties for global observations?

Always a compromise between several desirable properties: - Taxonomic resolution - Quantitative power - Observing the full plankton range - Potential scalability and interoperability

No distinctions

Functional groups

Full morphological taxonomy

Population/cryptic species

Only on selected types (e.g. phytoplankton)

Applicable on the full

plankton types

High scalability potential

Too costly (HR/expertise/ $$)

Not comparable between experts

Available techniques and their potential?

Bulk measurements and mass

Diagnostic pigments

Taxonomic counts

Genomic tools

Bio-acoustic

Bio-optics

Quantitative imaging

« not the best

but quite

equilibrated »

Taxonomic

Quantitative

Generalist

Scalable

An example of meta-imaging approach

11/05/17 Soutenance HDR Fabien Lombard IFCB

An example of meta-imaging approach

11/05/17 Soutenance HDR Fabien Lombard

Tara Ocean

station 210

Labrador Sea

Lessons learned from end-to-end approach

Imaging can bring a common and unified measurment

across plankton types

Never trust the “specifications” of an instrument

What is seen is not

what is seen efficiently

Strong need for

cross-calibration

between instruments

11/05/17 Soutenance HDR Fabien Lombard

Where do we go?

Requirements for future plankton observation systems

Sampling every time we can!!

Standardized, intercalibrated and cross-compared

measurements

An holistic view of plankton

Data that fulfils modelling needs (beyond « total mass »):

types of planktons, sizes and biomasses

Distributed and open data sharing (FAIR)

Going toward automated measurements ?

11/05/17 Soutenance HDR Fabien Lombard

Ecotaxa : a first step toward distributed data

Free exploration of validated

images

Launched in 2016 … 85 million

images now (41% validated)

Powerful filters

Powerful manual annotation (>20 000 day-1)

Explicit taxonomy

Ecotaxa : a first step toward distributed data

A free collaborative tool for hosting, sorting, annotating taxonomically and sharing images

Built-in automatic classification algorithms (random forest, deep learning) to help sorting images

632 users/ 197 organisations (about 40 simultaneous users)

About 11 types of instruments (AMNIS, IFCB, Cytosense in test, flowcam, zooscan, zoocam, UVP, LOKI, eHFCM, bioscope, planktoscope… still growing)

Full control of dataset and permissions by data-owner

Ecotaxa : a first step toward distributed data

Particles and marine snow data

Fully calibrated, fully exportable

Future: World Wide Web of Plankton Image Curation

-3 more servers (US, Brazil, Japan) under a joint

infrastructure

-Scalling up from million to billion images

-linking with OBIS and other global repository

-better recognition algorithms

-better export and visualisation of results

UVP-6

N° 1 1989 N° 2 1991 N° 3 1996 N° 4 2001 N° 5 2008

N° 6 2019

(planned date or release)

UVP-6

First goal

Need to recover the

sensor (for the moment)

UVP-6

11/05/17 Soutenance HDR Fabien Lombard

UVP-6 Currently under tests for integration on several vectors (gliders, Argo floats, moorings, ROV)

and cross calibrations

Already planned in several large scale projects

(H2020, ERC, ANR, ANR jcjc)