Investigation of the relationship between fish stocks, ecological quality ratings (Q-Values), Environmental factors and degree of Eutrophication

Summary: Final Report of the ERTDI-funded project 2000-MS4-M1. Authors: Fiona Kelly, Trevor Champ, Neasa McDonnell, Mary Kelly-Quinn, Simon Harrison, Alison Arbuthnott, Paul Giller, Mike Joy, Kieran McCarthy, Paula Cullen, Chris Harrod, Phil Jordan, David Griffiths, and Robert Rosell.

Published: 2007

Pages: 6

Filesize: 36 KB

Format: pdf


MS4 Final Report :: Environmental Protection Agency, Ireland

In Ireland the water quality of streams and rivers has been assessed using macroinvertebrates, chemistry and macroflora since the 1960s. The Water Framework Directive (EC Directive 2000/60/EC) (WFD) lists fish amongst the biological elements which should be used for the classification of ecological status of surface waters (rivers, lakes and transitional waters [estuaries]). This project was supported under the EPA ERTDI (2000–2006) programme to: (i) assess the impact of water quality, as evidenced by the Environmental Protection Agency (EPA) Quality Rating System (Q-values) on riverine fish stocks; (ii) assess the feasibility of using fish assemblages as indicators of ecological quality and (iii) develop a predictive model which would have application in the context of the WFD. Investigation of specific questions regarding eutrophication pressures was also required. The project was awarded to and executed by an alliance of state agencies and academic institutions north and south of the border

Using Comité Européen de Normalisation (CEN)-compliant electric fishing methods, a comprehensive dataset of fish and habitat variables was generated at 374 river locations, mostly in wadable 1st to 4th order streams, across the full range of EPA water-quality ratings. Archival material for fish and habitat variables was sourced for another 145 sites. Established EPA protocols were used to assess water quality at each location surveyed.

The study established that there is a relationship between fish-community composition and Q-values. Non-salmonids dominate the fish community at ‘poor’-quality (Q2–3) sites but decrease to <10% of the fish population at ‘high’-quality (Q4–5 and Q5) sites, whereas salmonids dominate the community at high-quality sites and decrease to <20% at poor-quality sites. It was statistically possible to separate a number of fish groups in relation to Q-values. Salmonids (1+ and older) were the best indicators of water quality in terms of species composition (%) and abundance (no. fish m-2), as indicated by Q-values. Salmonids (1+ and older) divided into five distinct groups: Group 1 (Q2–3), Group 2 (Q3), Group 3 (Q3–4 and Q4), Group 4 (Q4–5) and Group 5 (Q5) in terms of species composition. In terms of abundance, the 1+ and older salmonids divided into three significantly different groups: Group 1 (Q2–3 and Q3), Group 2 (Q3–4 and Q4) and Group 3 (Q4–5 and Q5). Moreover, the abundance of 1+ and older salmon was significantly different between moderate (Q3–4) and good-quality (Q4) sites. These simple metrics can be used to separate the ‘high/good’ and ‘good/moderate’ boundaries for the WFD for fish (particularly for wadable river sites). Separation of the good/moderate boundary (i.e. Q4/Q3–4) is particularly important but it is a relatively subtle change indicated by 1+ and older salmon, which is only applicable to locations downstream of impassable barriers.

Using the fish-community data generated by the project, a predictive model was developed for fish in rivers. All sites achieving a Q-value of Q4–5 and Q5 were considered ‘high’-quality or possible reference sites. The observed/expected (O/E) scores grouped by Q-values were significantly different for the six Q-value groups; the differences were between the lower Q-values but some overlap occurred at the higher Q-values. The discriminant model assessment showed that sites were correctly assigned to bio-groups and the distribution of reference site O/E scores was similar to many published RIVPACS and AUSRIVAS models using fish and invertebrates. This suggests that the model produced here is robust and to the standard of other similar models in use worldwide. While the correlation with Q-scores was strong and positive, there was no significant difference between the reference sites and the Q3–4 sites. <...>