DOI: 10.14466/CefasDataHub.157

Description

Mesozooplankton abundance data gathered as part of an annual autumn pelagic survey in the English Channel, Celtic Sea and Bristol Channel, running from 2012-2016. Samples were obtained using twin ringnet deployments and subsequently processed back at the lab using ZooScan image analysis. Count abundances from semi-automatic image classifications on Plankton Identifier (and later, EcoTaxa) were then converted into abundances per m3 using estimates of filtered volume from flowmeter readings. The current dataset comprises mesozooplankton results from the POSEIDON survey series which ran until 2016, although results have continued to be obtained as part of the expanded PELTIC survey series, which has been ongoing since 2017. ZooScan image analysis of mesozooplankton samples was initially trialed with samples collected in the first annual survey in 2012. Samples were collected with twin ringnet deployments and KC flowmeter attachments onboard the RV Cefas Endeavour, with image analysis conducted on those obtained using an 80µm ringnet of 0.5m diameter. Deployments were made at set station locations along transect lines which concentrated on UK territorial waters around the Cornish Peninsula, from the Bristol Channel south of Wales and eastern Celtic Sea to the western side of the English Channel. The transect lines stations are located within were designed with their spacing and orientation across the prevailing bathymetry of the area.

Preserved samples were sent back to the Plankton Lab at the Cefas Lowestoft Laboratory and thereafter split into fractions and prepped for image analysis using ZooScan. To obtain a representative subsample, a minimum of 200 zooplankton specimens needed to be captured in the images where possible from each sample, with rescans conducted as necessary. Images of objects processed from the sample were then taxonomically classified within Plankton Identifier software.

Once uploaded into either Plankton Identifier, scanned objects from each sample were predicted into taxa categories with the provision of a learning set to aid automatic image sorting using Random Forest algorithms. These predicted classifications are validated afterward by an experienced zooplankton taxonomist and the metadata downloaded with the individual taxa counts. Filtered volume estimates were obtained based on calculations using flowmeter rotor revolution counts recorded for each station. Objects captured in images were separated into zooplankton taxa and detritus groups and abundances estimated in terms of individuals per m3, using counts and filtered volume and raising factor calculations based on the sample fractions used. An agreed subset of the data may be selected for QC checks by a second experienced taxonomist, ensuring quality control in the image classifications.

Contributors

Almeida, Nevena / Bouch, Paul / Pettigrew, James / Close, Hayden / Cefas Plankton Team / POSEIDON crew

Subject

Plankton / Zooplankton / Monitoring / Pelagic / Taxonomy / Ecosystem / Plankton abundance per unit volume of the water column

Start Date

12/10/2013

End Date

01/11/2016

Year Published

2024

Version

1

Citation

Almeida et al. (2024). Mesozooplankton abundance in the Bristol Channel, Celtic Sea and English Channel 2012-2016. Cefas, UK. V1. doi: https://doi.org/10.14466/CefasDataHub.157

Rights List

Open Government Licence

DOI

10.14466/CefasDataHub.157