Is habitat restoration effective for brown trout and Atlantic salmon populations?

A part of Jonny Grey’s role as a Professor in Practice is to forge links with the academic community and facilitate research projects. It was serendipitous timing last year when Lizzie Thomas contacted WTT, asking if we had any data relating to our numerous restorations — Jonny’s TROUT project, Tackling Resilience On Underperforming Tributaries, was just beginning to bear fruit. Lizzie is a Zoology graduate from the University of Southampton and has just completed her MSc research in Ecology, Evolution and Conservation at Imperial College, London. She was particularly interested in understanding the effects of habitat restoration on salmonids given their ecological and economic value and, here, summarises the findings from her thesis. 

On these pages, I don’t really have to state that effort to restore salmonid (trout and salmon) is relatively common and increasingly widespread throughout the UK, Europe and North America. The aim is to increase the amount of and/​or improve salmonid habitat to bolster salmonid numbers and population resilience. Projects vary tremendously in size, from small-scale spawning gravel addition and in-stream structure addition, to large-scale dam removal and liming (to reduce acidification in rivers). So, the goal of habitat restoration measures is to support a sustainable wild population in contrast to population restoration efforts that simply aim to increase salmonid numbers by stocking.

DSC 4951
DSC 0219
DSC 1458

To date, there must have been thousands of salmonid habitat restoration projects undertaken across the globe, yet little work has been carried out aiming to synthesise the outcomes. In particular, there are very few studies that have actually calculated the impact of habitat restoration on Atlantic salmon and brown trout. Identifying this research gap in the scientific literature marked the beginning of my MSc research project.

My time at Imperial coincided with Covid-19 and restrictions on fieldwork in 202021, so a desk-based research project synthesising the results of salmonid habitat restoration projects seemed a good avenue to explore. Without the possibility of fieldwork, I thought I could use the time, working from my desk, as an opportunity to develop a new skill — meta-analysis. Don’t panic! A meta-analysis is a powerful statistical tool that synthesises the results of many individual studies to give an overall review of trends. So, instead of investigating first-hand the effects of salmonid habitat restoration projects out in the field, my project instead would encompass analysing the results of many salmonid restoration projects that had already been completed.

The process of a meta-analysis starts by defining a research question; in my research, What is the effectiveness of habitat restoration on Atlantic salmon and brown trout?”. Following on, a thorough literature search was conducted to identify restoration projects and academic studies that relate to the research question. Data, in the form of salmon and/​or trout density (fish/​100m2) before and after restoration (or at control and impact sites), were extracted from each restoration project/​study. Once all the data had been extracted and tidied, the real analysis began. An effect size”, that is the difference between the salmonid density before and after restoration (or salmonid density between control and restored sites), was calculated for each project/​study. The meta-analysis then summarised the results of these individual projects/​studies to provide insight into trends associated with habitat restoration. 

I found 51 projects providing comparable data, ranging from Spain to North America and including five restoration projects undertaken by WTT: barrier removal on Haw Beck and the River Aire, weir notching on Eastburn Beck, and large woody material addition on Flasby Beck and White Beck. I then conducted multiple analyses to: (i) understand the effect of all restoration types on each species; (ii) understand the effect of each restoration type on each species; and (iii) understand the influence of different factors, such as river characteristics, land use and stocking, on the response of the two salmonid species. 

DSC 0841
DSC 0834

The first, and perhaps most important finding of my study was that both brown trout and Atlantic salmon responded positively to habitat restoration. This is very encouraging given the unknowns (ie lack of funding for monitoring) and questions surrounding whether habitat restoration is effective. 

Next, I then investigated how salmon and trout respond to different restoration efforts. These restoration efforts included: spawning gravel addition, in-stream structure addition (eg large woody material), channel restoration, barrier removal, nutrient restoration and liming. The analysis suggested no difference in effectiveness of different restoration types on these two salmonid species. For example, effects of small-scale restoration such as gravel bed addition and woody debris addition, were not significantly different from the effects of larger-scale restoration efforts, such as liming and dam removal. This result highlights that both small-scale and large-scale salmonid habitat restoration is effective. It is particularly encouraging given that most habitat restoration projects are limited by budget and feasibility and are thus focused on small-scale restoration efforts.

L Thomas Fig jpg
A statistical representation of 'effect sizes' for different restoration types on salmonid density

Finally, almost as an added-extra’ analysis, I investigated the impact of different factors (formally known as moderators) that may influence the effect of restoration on salmonids. In the meta-analysis world, this is called moderator analysis. I thought this important given that many of the restoration projects/​studies that I used in my thesis also included data on project age (time in years since restoration efforts were implemented), geographical location, river characteristics (e.g., river width, river depth), riparian land-use, and whether artificial stocking occurred. Other important moderators that are often reported in the literature, including annual discharge, flow velocity, riverbed substrate and pH, were insufficiently detailed in the projects/​studies I found, and so I was unable to include them in the moderator analysis. 

The only moderator I included that had a significant influence on the impact of restoration on the two salmonids was project age. Salmonid density increased with project age. In other words, salmon density increased with time since restoration efforts were implemented and is intuitive to a point: salmonid populations often take years to decades to respond to restoration but will reach a carrying capacity eventually. As often is the case, there is a real need for long-term monitoring (i.e., 10+ years) which is difficult to fund. Most studies in the scientific literature call for longer-term monitoring of restoration projects. 

DSCPDC 0000 BURST20210607114609274
DSC 5935
DSC 1843 2

Whilst none of the other moderators significantly influenced the impact of habitat restoration on salmonids, further analysis on the presence of the stocking moderator threw up some interesting results. In rivers without stocking, salmonid populations significantly increased in response to the habitat restoration. However, where stocking was carried out, salmonid populations did not respond to habitat restoration. This specific result mirrors that found in another meta-analysis of Finnish river restorations by Marttila et al., 2019 and indicates that all the good work to improve rivers for wild fish will amount to nothing if artificial stocking continues. Therefore, it is important to focus resources to directly address the factors limiting salmonid production, rather than attempting to increase salmonid populations through regular stocking events. Sounds like the WTT ethos!

Some of you may be thinking, why was there was no sub-analysis investigating the effects of habitat restoration on each age-group of salmonids? This was due to insufficient reporting of each age-cohort in the data. Habitat requirements of salmonids are very much life stage-specific, and change quite dramatically throughout the life-cycle. If the data allowed, it would have been very interesting to investigate both how restoration as a whole influences different life stages of salmonids, and also if the effect of restoration on each life stage differed between each restoration effort. This is certainly a future lead to explore.

I hope to continue my research project by increasing the number of salmonid habitat restoration projects in my database, giving me a little more statistical power in the meta-analysis that would be required for a scientific journal publication. In the meantime, I would welcome any questions or comments regarding my thesis and also any further suggestions of salmonid restoration projects (either brown trout or Atlantic salmon) that could be added to my database. I would also like to thank Prof Jonathan Grey and the WTT for providing data on the some of the Trust’s brown trout habitat restoration projects.

Lizzie (elizabethlethomas@​gmail.​com)