National Electrofishing Programme for Scotland (NEPS) 2021: analysis

Discussion

A third NEPS survey was undertaken in 2021. When combined with previous surveys these data provide a strong quantitative evidence base from which to assess spatial and temporal variability in the densities of salmon and trout, the status of salmon, and the potential influence of pressures including water quality and farm to wild genetic introgression. Such robust evidence and understanding provide an important underpinning to target management actions to ensure good quality habitat and healthy fish populations as envisaged by the Blue Economy Vision for Scotland and Wild Salmon Strategy. With additional work and / or resources, the assessment methods developed for salmon under NEPS could be extended to other species and reporting contexts. This could include Habitats Directive, Water Framework Directive and Conservation Regulations, although this is not without challenges (see below for further discussion).

NEPS 2021 employed a new single year survey design that evolved in response to feedback from local fisheries managers and logistical challenges identified in previous years. The new survey design added value by incorporating local monitoring and reporting requirements where possible. While the new design offered many advantages and increased flexibility over previous surveys it also posed substantial challenges in terms of data analysis and interpretation, particularly through the inclusion of larger 5^th Strahler order rivers in some regions. Nevertheless, it was possible to overcome these challenges to present new and improved data, whilst also providing data comparable to previous years and surveys.

Post-stratification of the NEPS survey data allowed assessments of density and status for individual rivers including Special Areas of Conservation where it was also possible to suggest approaches for classification that could meet the needs of Site Condition Monitoring. A comparison between juvenile salmon density estimates and rod catch at national and regional scales provided assurance that the abundance metrics underlying the adult and juvenile assessment methods in Scotland are broadly coherent. Although water quality data were collected during surveys in previous years, this was the first year that these data have been presented as part of NEPS reporting. When considered alongside assessments of status these data provide a powerful basis for identifying environmental pressures acting on salmon, thereby informing targeted conservation and restoration activities. These issues are discussed in detail below.

Capture Probability

Capture probability models are essential for harmonising the data collected under NEPS (and from other electrofishing data sources) and for producing reliable, unbiased density estimates. Without appropriate modelling, spatio-temporal variability in apparent abundance (i.e. fish counts) can be confounded with spatio-temporal variability in capture probability (Millar et al., 2016, Dauphin et al., 2018, Glover et al., 2018, Malcolm et al., 2019a, Glover et al., 2019) leading to erroneous understanding of the status of stocks and potentially poor management decision making. The capture probability model presented in this report was similar to that from previous years (Malcolm et al., 2019b, 2020). Trout were generally more catchable than salmon, parr were more catchable than fry, and fish were more catchable on the first pass than subsequent passes, with the between pass differences in capture probability being greater for parr. It is possible that some of this variability reflects differences in habitat use among species and lifestage or the effects of different average size, with larger fish typically being more catchable.

Some new Organisation – Teams were established in 2021 due to new individuals undertaking fieldwork, or movement of individuals between organisations. It was necessary to make some pragmatic decisions to group a few small Teams within larger Organisations where there were insufficient data to derive the capture probability for individual Teams. The effects of Organisation-Team and Hydrometric Area (which could also include other effects including water quality and equipment) continued to be difficult to disentangle given the spatially constrained range of sites typically sampled by individual Organisation-Teams. Nevertheless, their combined effect was large emphasising the need for further investigation to identify the key underlying causes. Three important areas that could be examined would be the effects of electrical conductivity, equipment and fish size. These controls have been characterised under the NEPS programme and their effects would be expected to vary regionally. Unfortunately, these data were not routinely recorded in the past so modelling of historical data might not be possible.

Consistent with previous analyses there was a positive linear effect of year in the final capture probability model. This indicates that capture probability increased over time at the national scale. This has major consequences for the use of single pass and timed electrofishing for assessing population trends. Specifically, the use of uncalibrated single pass or timed data could result in biased trend assessments, potentially leading to inappropriate management decisions (Glover et al., 2019). NEPS 2021 continued to include multi-pass data in the survey that allow capture probability models to be fitted that address spatial and temporal biases to provide a reliable assessment of spatio-temporal variability that is often absent from other more ad-hoc surveys. These complex models have been made available (through an online R Shiny application) that allow NEPS collaborators to harmonise and correct data collected for other local purposes^[1].

NEPS 2021: Changes to strata

Changes were made to the NEPS survey design to address challenges related to the allocation of over samples (replacement sites for those that could not be sampled), allow sampling of larger 5^th order rivers in selected regions as requested by local fisheries managers, and allow locally driven sampling and reporting requirement to be incorporated into the larger NEPS programme with associated benefits for both local and national management.

In the NEPS 2018-19 survey design the spatial distribution of strata matched NEPS reporting regions, which in return received a standard allocation of 30 samples (Malcolm et al., 2019b). This was appropriate from a survey design perspective. However, it caused logistical problems in regions supported by more than one local delivery organisation because over samples could fall anywhere in the region and affect the relative number of sites to be sampled by the different organisations. To address this problem strata were sub-divided where multiple organisations fell within a reporting region. For reporting purposes, it remained possible to post-stratify the data and recombine strata to match reporting regions from previous years.

Additional strata were also added to assess fish populations above barriers on the Shin and in the Forth regions. These changes to the design came with additional local resource and increased the overall value of the NEPS programme both locally and nationally. Where possible this concept should be extended in future years to maximise the robustness of local fisheries monitoring while also improving efficiency given the extensive method development and overhead costs incurred in running NEPS.

NEPS 2021: Changes to the sample frame

The target population for NEPS is rivers that are accessible to salmon (below physical barriers), support salmon fisheries (are assessable under Conservation Regulations) and can be sampled by wading and electrofishing. In 2018 and 2019 the sample frame (i.e. the approximate spatial extent of the target population) included Strahler river orders 2-4. First order rivers often ran dry, were too small to sample reliably or were impossible for salmon to access. Fifth order (and larger) rivers were considered on average to be too large to reliably sample across the country. However, some local fisheries managers considered that 5^th order rivers could be sampled and requested that this be considered in revisions to the NEPS survey. In response, local fisheries managers were asked to identify whether they wanted fifth order rivers to be included in the sample frame on a region-by-region basis. At the same time, some local fisheries managers took the opportunity to remove smaller rivers from the sample frame where physical access was known to be impossible. The result was substantial changes to affected regional benchmarks, increased spatial variability in benchmarks and a requirement to harmonise data between survey years to allow for appropriate comparisons of abundance between regions and years. These challenges were addressed through development of appropriate spatial data, an extended suite of analyses and careful survey design. However, the increased flexibility has come at a cost in terms of the complexity of analyses now required.

The NEPS benchmark model (Malcolm et al., 2019a) predicts that salmon densities should increase with upstream catchment area and thus river order. Given the regionally variable inclusion of fifth order rivers, it was important to explore whether patterns of spatial variability in benchmark density were at least broadly reflected in the observed juvenile density data. It was also important to establish whether inter-annual variability in densities within river orders were also broadly consistent between river orders. Salmon fry and parr densities increased with river order and inter-annual variability in densities within river order were indeed broadly comparable across river orders. This suggests that spatial variability in the benchmark was appropriate and that temporal variability in observed densities should be similar regardless of the particular sample frame used.

The observation that salmon densities increase and trout densities decrease with river order could reflect one of two potential processes. Firstly, hydraulic and sedimentary characteristics and food availability (broadly habitat quality) could vary systematically across the sampled rivers (Wyrick and Pasternack, 2014) in a way that favours salmon and constrains trout. For example, the proportion of run, riffle and glide habitat characterised by higher velocities favoured by drift feeding salmon could increase in a downstream direction, while the percentage of pools and other less favoured slow water habitats could decline. Alternatively, local delivery organisations could introduce a systematic bias into the survey by micro-siting survey locations to avoid deeper water areas in larger rivers that could not be fished by wading based electrofishing. This could have the effect of sampling areas that are on average associated with higher salmon densities in larger rivers. The available site-wise habitat data collected during electrofishing shows that the percentage of pool habitat surveyed decreased with river order, while the percentage of run and riffle increased. What is not clear is whether this reflects genuine changes in habitat availability or local survey bias. Importantly, these observations do not affect the ability of NEPS to assess the status and abundance of freshwater fish as the data supporting the benchmark model will be biased in a similar way to the NEPS survey. However, interpretation of the observed spatial variability would become important if it were used to try and scale juvenile production across whole river systems, potentially underestimating the value of smaller rivers to whole system production. The extent of any biases could be determined by asking local managers to characterise habitat at the precise location of the allocated NEPS sample and then at the location that they were able to access the river for survey (i.e. following micro-siting). However, a comprehensive assessment of the true value of rivers of different size would require development of alternative survey methods capable of sampling larger, and particularly deeper rivers.

Abundance of trout

It is not currently possible to assess the status of brown trout due the absence of a suitable density model from which to estimate benchmark densities. This remains a priority for method development in the future. In the absence of a benchmark it is still possible to assess changes in density among years. At a national scale trout fry and parr densities have declined between 2018 and 2021, although densities in 2019 and 2021 remained broadly comparable. Sea trout populations (as indicated by rod catch) have been in long-term decline and it is possible that changes in juvenile trout abundance at least partially reflect these changes, although it is not possible to readily separate individuals of resident and migratory origin.

Abundance and status of salmon

The numbers of juvenile salmon fry and parr in the NEPS 2021 survey provide a strongly contrasting picture of population health. Salmon parr densities were lower than observed in the 2018 and 2019 surveys reflecting multiple years of relatively low spawner numbers. Grade 1 regions were restricted to the north of Scotland and the national-scale density was only ca. 60% of the benchmark. In contrast, there was an increase in salmon fry densities between 2019 and 2021 to levels which exceeded the national benchmark. In some regions (e.g. Annan, Ayrshire, Nith, Clyde) this led to improvements in status. It is likely that improvements in salmon fry densities largely reflected local improvements in spawner returns. However, it is also possible that substantial reductions in older age classes of salmon (parr) reduced competition and predation (Bacon et al., 2015) allowing for compensatory improvements in the survival of fry. Although higher salmon fry densities were not observed across all regions, the overall picture was better than 2019 and was more positive than that observed for parr.

NEPS assessments in the context of statutory reporting requirements

Where possible, multiple reporting needs should be addressed through a common set of monitoring activities and methods. This maximises efficiencies, reduces costs associated with data collection and reporting, and ensures greater consistency of procedure between assessments in different statutory contexts. At face value juvenile assessment methods could meet many of the reporting requirements of the Habitats Directive, Water Framework Directive and Conservation Regulations as they relate to Atlantic salmon and other freshwater fish populations. However, the specific requirements and drivers vary between legislation so it is useful to assess where NEPS data and assessments can be used at present, or where further work and developments would be required. Specifically, there is a need to consider issues relating to the reference level (e.g. Benchmark), spatial coverage of samples (sample frame), and the frequency and density of sampling.

The benchmark for NEPS was derived from a national analysis of juvenile salmon densities (Malcolm et al., 2019a) that modelled the abundance of salmon fry and parr across Scotland from 3848 multi-pass electrofishing site visits between 1997 and 2015. The resulting density predictions were then scaled to better reflect the densities expected in a near-natural catchment, given adequate spawner returns to stock available habitat. The aim of the benchmark was thus to define a reference level that was close to saturated habitat (or Smax in stock-recruitment terms). However, direct comparisons with stock-recruitment derived references are challenging because well-defined stock-recruitment relationships only exist for very few locations in Scotland (Gurney et al., 2010), and of those, only two sites (Girnock and Baddoch Burn) also have adequate data on juvenile salmon abundance to make suitable inferences (Glover et al., 2018).

The Habitats Directive aims to maintain and restore favourable conservation status for species of interest that includes Atlantic salmon. Both the benchmark and methods deployed for NEPS would appear consistent with definitions specified under the Common Standards Monitoring Guidance for Freshwater Fauna (JNCC, 2015), with Grade 1 appearing to be consistent with Favourable Conservation Status.

The Water Framework Directive (WFD) aims to restore rivers to Good Ecological Status. This is defined as representing only slight disturbance from natural conditions. However, WFD also requires status to be reported at five different levels (High, Good, Moderate, Poor, Bad) and that the assessments consider species composition, abundance and age structure. In this context current NEPS assessments would need to be extended to include additional classification categories consistent with WFD definitions, and also additional native fish species (e.g. trout and eels). Based on WFD definitions it is likely that High and Good categories would be consistent with NEPS Grade 1 classifications. The addition of new categories to the NEPS classification scheme would not necessarily involve substantial development work in the case of salmon because classifications are typically based on ecological quality ratios between observed and expected (benchmark) abundances. However, the addition of new species is more challenging and would involve the development of new species and life-stage specific benchmarks.

Conservation Regulations assessments are based on estimates of maximum sustainable yield (MSY) derived from adult-adult stock recruitment relationships. These references are likely to be lower than the NEPS benchmark, although in reality this would be extremely challenging to assess given available stock- recruitment data. One pragmatic solution could be to compare densities at the benchmark, with those observed under MSY for the only two suitable datasets (Girnock and Baddoch) and then scale the benchmark based on the proportional difference in abundance. However, this would require substantial new analyses.

The NEPS sample frame covers wadeable rivers, below impassable barriers in catchments supporting salmon fisheries. This means that all of the catchments that need to be assessed for Conservation Regulations are included in the sample frame. However, the sample frame does not include all of the waterbodies that need to be assessed for WFD, or all of the catchments that need to be assessed for salmon under the Habitats Directive. Specifically the NEPS sample frame does not include smaller catchments that do not support salmon fisheries, areas above physical barriers, large and deep rivers or lochs that cannot be sampled by wading and electrofishing. It would be possible to extend the sample frame to include additional rivers or to supplement the current design with complementary but consistent designs for other rivers depending on resource availability and sampling limitations. Nevertheless, this does not constrain use of the data that are already available.

The density of samples in the NEPS survey reflects the size of the regional strata, allocation of samples to strata (currently 30 samples per NEPS 2018-19 strata), availability of funding, the ability of the fisheries management sector to complete surveys over the summer and the desire to obtain adequate data to assess status and trends. Sample density is thus a pragmatic balance between what is desirable from a scientific and management perspective and what can be achieved given available financial and practical resources. WFD assessments are conducted at a waterbody scale and thus a key driver is to obtain as many site-wise assessments as possible in different waterbodies. Habitats Directive requires assessments of status and trends at the scale of each SAC, which are catchment, sub-catchment and indeed multi-catchment scales (North Harris). Conservation Regulations require assessments for each river catchment supporting a salmon fishery. In all cases the confidence of reliable assessment will improve as the number of samples within each reporting area increases (and as inter-site variability decreases). In this report we did not assess the status of any spatial extents where there were less than five samples in each year of NEPS surveys. This decision was practical based on the minimum data requirements of the spsurvey R package, but also clearly represents a lower limit on the sample numbers that would be desired. This means that there were no assessments for smaller rivers or SACs. If assessments were required for every individual SAC and catchment then an increase in sample density would be required in those areas that currently receive few samples. This would be technically feasible by varying sample numbers and / or strata numbers, but would come with resource implications.

WFD has a requirement to report on the status of waterbodies once in a five year period. Habitats Directive requires reporting every six years and Conservation Regulations assessments need to be completed annually. NEPS surveys were carried out in 2018, 2019 and 2021 with funding support variably coming from Marine Scotland (2018-2021), Crown Estate Scotland (2019, 2021), Nature Scot (2018) and SEPA (2018). Additionally, the fisheries management sector provide support in-kind since the funding available for NEPS cannot cover all incurred costs. Funding for future iterations of a nation-wide NEPS programme are not guaranteed and this has potential consequences for statutory and other reporting requirements where it is envisaged that NEPS data could contribute. It is possible that NEPS assessments could be completed on a bi-annual basis and still provide useful and near-continuous assessment data as each survey includes two cohorts (fry and parr), but less frequent sampling also limits the value of the data in terms of assessing recruitment between life stages.

Future work: Water quality as a predictor of abundance

Water quality influences capture probability and is a critical control on juvenile salmon survival and abundance through effects on fish physiology (Malcolm et al., 2014), in-stream productivity and food availability (Williams et al., 2009). Water quality is not included in the current benchmark model (Malcolm et al., 2019a), nor will it be included in the forthcoming trout benchmark model, due to the absence of historic water quality data at electrofishing sites underpinning benchmark models. However, it is a potentially important predictor that could further explain within and between catchment differences in salmonid abundance. Water quality data collected during NEPS and provisionally reported here would allow further investigation of these effects and could be incorporated into future assessments at site-wise scales. With further large-scale spatial modelling of water quality (e.g. Smart et al., 2001; Monteith et al., 2015) these data could be included in future benchmark models. This requirement was highlighted in the NEPS 2018 report and remains a priority (Malcolm et al. 2019b).

Future work: Identifying pressures acting on freshwater fish populations

In recent years there have been attempts to characterise the pressures acting on Atlantic salmon to inform management, habitat restoration and regulation of potentially detrimental activities (Forseth et al., 2017). NEPS provides a strong quantitative framework for characterising pressures at multiple spatial scales. Recently tissue samples from a sub-set of NEPS sites were used to undertake the first national scale assessment of the effects of farm to wild genetic introgression on wild salmon populations (Gilbey et al., 2021). In 2021 tissue samples were obtained from all NEPS sampling sites across the country and laboratory analysis is currently underway. This will provide an unprecedented quantitative assessment of the scale of introgression effects on wild Atlantic salmon in Scotland and allow reporting at regional scales comparable with NEPS status assessments.

For the first time this report has included a large scale characterisation of water quality across Scotland using the data collected under NEPS. While one-off spot samples do not replace the need for more frequent and long-term monitoring, such surveys are able to fulfil two of the major goals of water quality monitoring, namely identifying pollutant sources and informing locations for conservation and restoration (Dupas et al., 2019). Mapping of NEPS data revealed strong spatial patterns in water quality reflecting both natural (e.g. geology, soils, climate) and anthropogenic (e.g. nutrient pollution, sulphur deposition) influences that can directly or indirectly affect survival and productivity of salmon and other freshwater fish species (Malcolm et al., 2014; Forseth et al., 2017). There have been strong concerns in recent years over the impacts of excess nutrients on ecological systems (Jarvie et al., 2018). Anthropogenic sources of nutrients can include atmospheric deposition, agriculture, leaking septic tanks and sewage (Edwards and Withers, 2008). High nutrient concentrations can cause eutrophication (Rankinen et al., 2019), damaging algal blooms and reductions in dissolved oxygen. Some forms of nitrogen e.g. ammonia and nitrite can also be directly toxic. In the case of ammonia the relative contributions of the dominant non-toxic form (ammonium, NH₄⁺) and toxic free ammonia (NH₃) is controlled by temperature and pH, with greater risks associated with high pH and temperature. There are thus clear risks associated with water quality that could be exacerbated by climate change that threatens to both increase temperature and reduce flows with consequent effects on chemical concentrations (Charlton et al., 2018).

Future Work: Development of benchmarks for brown trout and European eel

There is a desire to improve assessment of both brown trout and European eel. This has been highlighted by WGTRUTTA and WGEEL ICES working groups. The addition of new benchmarks would add value to the NEPS programme and increase the potential utility of the work in the context of WFD. The development of an assessment method for trout is also a commitment in the Scottish Government's response to the Salmon Interactions Working Group (SIWG) to support the sustainable development of aquaculture in Scotland.

Future Work: NEPS survey designs

During 2022 there have been substantial efforts to develop a new multi-year design that could be used to support future NEPS surveys. This has involved the development of new spatial datasets and consultation with local fisheries managers to further refine the spatial extent of the sample frame to identify rivers that can be sampled by wading and electrofishing. Where local resources have allowed (and where there is a local desire), new strata have been created to improve the spatial scale of NEPS assessments within larger regions. For the remainder of the 2022/23 financial year MS will focus on the delivery of a new survey design, the development of necessary R code, spatial data and methods required to support future NEPS programmes should these be prioritised for future funding.