Project Progress

Data Modelling

With the help of other MorFish project staff and collaborators, project postdoc Stephen Gregory has been developing statistical models to interrogate GWCT & INRA’s long-term salmon datasets. These statistical models correspond to two projects:

A statistical model to estimate spawning salmon stock using validated automatic fish counters

Salmon stocks in England, France, Scotland, Ireland, Wales and North America, are routinely quantified using automatic fish counters. In principle, automatic fish counters can quantify the spawning stock of migrating salmonids with little human intervention. In reality, the data must be validated to ensure they relate to the target species and the counters themselves require continuous maintenance; water is a very destructive and corrosive environment and counters frequently suffer periods of malfunction.

Stephen is developing a statistical model that estimates consistent daily counts of salmonids from noisy and incomplete automatic fish counter data. The model will account for incomplete validation, periods of malfunction and covariates of salmonid movement rates. It will be tested using synthetic data and then applied to 15 years of river Frome salmonid counter data (Figure 1).


It is being developed as a general model that can be applied to automatic fish counter data on other rivers in the UK and France, and elsewhere in the world.

A statistical model to investigate climate effects on salmon parr sizes

Salmon parr have been monitored on the Frome, Oir and Scorff for between 15 and 25 years, and for each fish captured (by electrofishing), their size is recorded. There are suggestions that changes in parr condition could be responsible for recent decreases in abundances of Atlantic salmon populations, either by reducing the number of salmon migrating to sea or by affecting their survival at sea or the timing of their migrations. Yet, there are few studies that have shown that changing environmental conditions are linked to changes in salmon parr lengths.

Stephen is developing a statistical model to investigate long-term changes in salmon parr lengths and whether they have been affected by river temperature and river flow. He has aligned the parr data for all three rivers, assembled datasets of possible environmental drivers for all three rivers and has developed models to explain observed variation and changes in salmon parr lengths.

By understanding the influence of environmental drivers on parr lengths, we can begin to speculate about the possible consequences of changes in river temperature and river flows on parr lengths and, ultimately, salmon population abundances.