Science Evidence Data and Digital Portfolio Annual Report 2024 - 2025

Salmon and Freshwater Fisheries (SFF)

Headlines

• Our Adult Salmon Assessment Team produced an annual stock status update to accompany the Conservation Regulations (in June 2024) and oversaw the preparation of the Scottish contribution to the ICES Working Group on North Atlantic Salmon (WGNAS), In addition, they led on the development of national adult salmon stock assessment methods and contributed to the North Atlantic Salmon Organisation (NASCO) work through support and input to the UK delegation.
• Our Epidemiology Team have contributed to Marine Climate Change Impact Partnership through various meetings and writing the Impacts on Aquaculture Report.
• An international collaboration led by the Epidemiology Team developed methods towards communication of knowledge strength in sea lice dispersal modelling.
• In July and August 2024, the Marine Directorate and the Helmsdale District Salmon Fishery Board had the fish counter on the River Helmsdale refurbished to help ensure the continued provision of this valuable data.
• Our Scientists have improved accessibility in the Official Statistics Reports produced in SFF.

I. The Scottish Salmon and Sea Trout Fishery Statistics published annually are a summary of rod and net catch and effort, for each fishing season. It is based on returns of salmon and sea trout fisheries throughout Scotland.

II. The 2024 Scottish Shellfish Farm Production Survey published since 1996, which achieves a 100% rate annually, is now fully accessible including the development of a Shiny Application for data visualisation.

III. Efforts are ongoing to bring the Scottish Government (annual) Scottish Fish Farm Production Survey published since 1979, which also achieves a 100% return rate, to improved accessibility.

Key Work in 2024 - 25

1. Collect, collate and publish salmon and sea trout catch data as Official Statistics. Collect, collate and publish data on salmon counts. Develop a national network of fish counters. Develop and apply automated tools to examine changes in salmon growth; undertake modelling of in-river abundance, age, length, sex ratio and egg requirements; update methods. Assess and report on the status of salmon stocks nationally, and internationally (ICES WGNAS); provide science support to policy for NASCO.

Contributes to:

Annual legislation to conserve salmon stocks and manage fisheries in Scotland. Scottish input into international stock assessment and management (ICES, NASCO).

2. Annual licensing.

Contributes to:

Consenting for freshwater stocking purposes and control illegal methods and times of fishing.

3. Collect and analyse data on juvenile salmonids including the National Electrofishing Programme for Scotland (NEPS). Run Girnock and Baddoch Index Monitoring Sites. Collect, store, quality control and analyse data collected under the Scotland River Temperature Monitoring Network (SRTMN).

Contributes to:

Feeding into the management of Scottish Salmon and Trout; improve assessment of sea trout in a national and international context; understanding population processes and the effects of environmental change.

4. Developing and validating sea lice modelling and international collaboration work with Norway Institute of Marine Research; refining environmental sampling methods for planktonic sea lice and sea trout; eel monitoring; understanding smolt migratory

behaviour through rivers and nearshore environments using acoustic tracking tagging and sampling.

Contributes to:

Advice on interactions between farmed and wild salmon populations; develop/provide methods to validate sea lice dispersal models; and application of genetics to inform origin of stocks; fulfil UKs international obligations regarding eel management, advice on planning applications for salmon and sea trout and feed into SEPA’s Sea Lice Risk Assessment Framework.

5. Collect and quality control data and publish as Official Statistics for Shellfish and Finfish Aquaculture Production Surveys; provide analysis and advice on epidemiology.

Contributes to:

Independent audit of Finfish/Shellfish aquaculture, international obligations to the Food and Agriculture Office, food security and regional planning and control of diseases to maintain fish health status.

6. Business and site management, workforce management, quality management, H&S and training at Freshwater Lab sites at Faskally, Montrose, Traps at Deeside and West Coast Field Station.

Contributes to:

Fleet Maintenance; Site Security; Asset/ Equipment Transportation and Disposal; H&S Management and Training and ensure accreditation with UKAS standards for Lab facilities.

7. Provide advice on pressures on salmon, trout and eels.

Contributes to:

Delivery of Wild Salmon Strategy and Vision for Sustainable Aquaculture.

Case Study: Automated extraction of salmon growth history using deep learning

Salmon scales grow as the fish grows in length and deposit rings, much like tree rings, that can be measured to determine how fast or slow a fish grew throughout its life, and how old it was when the scale was removed. The growth of salmon – especially while they are at sea – is related to how well they survive and how long they stay at sea. However, manually counting and measuring growth rings from scales is labour intensive.

Together with collaborators at the University of St Andrews, scientists from the Marine Directorate developed a machine learning algorithm based on over 1,000 scale images and over 50,000 annotated features to predict the position of growth rings from high resolution images of wild Atlantic salmon scales. The problem was approached in an innovative two step way: first, the centre of the scale is detected using a convolutional neural network (CNN) and transects are extracted from the image which contain the growth rings (step 1 in figure below). A second (trained) CNN predicts where the rings are on the image (step 2 in figure below). From these predictions, the spacing between the rings can be counted and detailed growth histories for individual fish calculated (step 3 in figure below). The work was published in an Open Access journal in 2024.

The new tool performs well and rapidly, cutting the time taken to process a single scale transect from 10-15 minutes to seconds. With an average precision of 99% when detecting the centre of a scale and of 95% when detecting an individual growth ring, the tool performs similarly to another human being. The code to run the tool has been made publicly available. It is currently undergoing evaluation to test its use on salmon scale archives from other salmon stocks through government and academic collaborators in the USA, Ireland and Norway.

The information on salmon growth derived from the machine learning approach can support the management of wild Atlantic salmon stocks in Scotland, and internationally, by providing insight into how environmental factors impact on fish size, maturation, growth and survival and ultimately, how populations respond to climate change.