Crab and lobster fisheries - stock assessments: results 2016 to 2019

2. Data Collection and Methods

2.1. Assessment Areas

For assessment purposes, the Scottish creel fishing grounds are divided into 12 assessment areas as shown in Figure 1. These areas are based on the historical system for reporting Scottish landings data (Thomas, 1958), subsequently revised to include two offshore areas – Papa, which lies to the west of Shetland, and Sule, which is to the north and west of Orkney and includes the Rona, Sulisker and Sule-Skerry banks. Some Scottish assessment areas extend outside Scottish Territorial Waters. On the east of Scotland, the South East assessment area extends beyond the Scottish EEZ border, while on the west coast, the Clyde assessment area stops short of the Irish EEZ border. There is some fishing on grounds outside the assessment areas. Currently these areas support only small fisheries and landings data are monitored for any change in importance.

2.2. Landings Data

The assessments use official landings data, which provide location of capture (by ICES rectangle), the species and the weight landed into ports in Scotland. These data are collated by Marine Scotland Compliance from sales notes, logbooks and FISH1 forms. The UK iFISH database was used to extract landings data for all crab and lobster species mentioned in this report. Marine Scotland’s FIN database was used for inspecting and performing corrections in landings from rectangle 40E4 (see below). Data for brown crab landings from the Republic of Ireland (collected by the Irish Sea-Fisheries Protection Authority – SFPA) were compiled and provided by the Irish Marine Institute. These data were not used in the assessments but have been included to illustrate brown crab landings by nation on a statistical rectangle basis to the west coast of Scotland.

ICES rectangle 40E4 straddles two assessment areas: the Clyde and the South Minch. Officially, reported landings data are available only for the whole rectangle. Exploratory data analysis was performed with the aim of developing a working method to split landings in rectangle 40E4 between the two areas.

FISH1 form data (2017-2020), which contain positional information on where fishing activities take place, were used to make a comparison between fishing location and the subsequent landing port. Results of the analysis indicated that the location of the landing port, either East (Clyde) or West (South Minch) of the Kintyre peninsula, was a good proxy for fishing location. Therefore, the E/W location of the landing port was used to allocate landings from rectangle 40E4 to either the Clyde or the South Minch areas for all trips.

2.3. Catch-at-length Data

2.3.1. Landings-at-length

Landings length-frequency data were collected by MSS as part of its market sampling programme with data extractions carried out from MSS’ Fisheries Management Database (FMD). Historical Shetland sampling data were provided by Shetland UHI (formerly the North Atlantic Fisheries College (NAFC) Marine Centre) with the permission of the Shetland Shellfish Management Organisation. Since 2010, data from fisheries in Shetland have been collected and provided by staff at Shetland UHI under the Memorandum of Understanding (MoU) between Shetland UHI and MSS. From 2012, landings length-frequency data collected by Orkney Sustainable Fisheries (OSF), mostly from the Orkney assessment area (but also from Papa and Sule), have been shared with MSS. These additional data have been used in combination with data collected by MSS from the same areas. All the sampling data are held in the MSS Fisheries Management Database (FMD).

Sampling measurements are taken as carapace length (CL) for lobsters, measured from the eye-socket to the centre of the base of the thorax carapace, and carapace width (CW) for brown and velvet crabs, measured across the widest part of the body, not including any spines.

In general, sampling effort is focused in those areas where fisheries are most important. However, the timing of landings is rather unpredictable and sampling is to some extent opportunistic, which explains the variability in numbers sampled and the occurrence of zeros for certain species in some areas. Brown crabs and lobsters may be retained for a period of time in holding tanks after being landed, which makes them easier to access for sampling. In contrast, velvet crabs are often landed in remote harbours and promptly dispatched to fishing processors or shipped abroad. This makes it more difficult to get samples, particularly in the Inner Hebrides (South Minch) in the west of Scotland.

MSS aims to conduct crab and lobster stock assessments every three years using length-frequency data from the most recent three years. However, due to delays in staffing to conduct this work, this report uses four years of sampling data (2016-2019). Before the report was concluded, landings information for 2020 became available and therefore the most recent landings data are also presented for the three species although the assessments were conducted for 2016-2019. The numbers of animals measured, number of trips and percentage of sampled fishing trips (quotient between the number of trips sampled and the total number of trips extracted from iFISH) by assessment area are shown for each species in Table 3, Table 4 and Table 5.

2.3.2. Discards

Discard data are not regularly collected for the crab and lobster fisheries in Scotland, and any mortality due to discarding practices is not taken into account in these assessments. Anecdotal information and recent ad hoc unpublished studies suggest that crab and lobster discard rates in the target creel fisheries are variable and occasionally high (>50% by number). Discards typically comprise those animals that do not meet landing regulations such as undersized individuals, post-moult animals with soft shells, v-notched female lobsters or berried female crabs. Discard survival rates are likely to be high (Rodrigues et al., 2021), and for assessment purposes landings are assumed equal to catch.

2.3.3. Raising and Data Quality

Length frequency data obtained from market sampling and official landings data were combined to provide a raised annual landings-at-length distribution. Data were averaged over four years (2016-19) and aggregated into 5 mm size classes for brown crab and lobster, and 3 mm size classes for velvet crab for use in the Length Cohort Analysis (LCA).

Landings-at-length data were not available for the three species in all assessment areas. The decision-making process to select which areas had sufficient data to run stock assessments is presented in Annex A. Four parameters were used to categorise the quality of the sampling data: number of trips/animals sampled; number of years for which data are available (maximum four years – 2016-19); sampling seasonality; and the shape of the length frequency distribution (LFD). For each species/area combination, these parameters were classified in one of three categories (“poor”/”ok”/”good”). Stock assessments were not run for areas where one or more of the parameters was classified as “poor” (see Annex A).

2.4. Biological Data

Information about the growth of British crabs and lobsters used in the LCA comes mainly from tagging studies carried out in the 1960s and 70s (Hancock and Edwards, 1966; Hancock and Edwards, 1967) (Table 6). Estimates of the von Bertalanffy growth parameters: asymptotic length (L_∞) and instantaneous growth rate (K) have been estimated using Ford-Walford plots (Chapman, 1994; Tallack, 2002; Mouat et al., 2006) (Table 6). Length-weight relationships (parameters a and b shown in Table 6) are from MSS (unpublished) market sampling measurements of length and weight. Different, area specific, biological parameters were applied in Shetland assessments as these are available from research on these species conducted in Shetland (see Leslie et al., 2010 for data sources).

2.5. Size-based Indicators and Sex Ratio

1

In theory, size-based indicators or length-based indicators (LBI) can provide information about the effects of fishing pressure on a stock. Fishing mortality affects the abundance of populations but also their age and size distribution. As cohorts age and grow, the effects of mortality accumulate over time. With increasing mortality, fewer older and larger individuals can be expected in a stock and associated catches. Life history processes such as maturation and fecundity depend on size, such that the size structure of the stock affects the reproductive potential and capacity for recovery (Trippel, 1998; Moland et al., 2010).

In this report, several size-based indicators were explored for the period over which sampling data are available (brown crab and lobster – from 1981; velvet crab – from 1987) for each species, by sex and assessment area. Any years in which the number of animals sampled was less than 50 and species from assessment areas sampled only once (Sule for lobster, Mallaig for brown crab, and North Coast, Papa, Sule for velvet crab) were excluded.

Size at first capture (L_c), the mean size in the landings of individuals above L_c, and the mean size of the largest 5% of individuals in the landings (L_max5%) were examined and plotted. The size at first capture, L_c, was calculated as the size at half the maximum frequency in the ascending part of a size frequency distribution (ICES, 2014b). Ideally, the size at first capture, L_c, should be above the size at first maturity, L_mat, to allow for reproduction before being caught by the fishery (Myers and Mertz, 1998).

As a reference point for the mean size, we plotted the F_MSY proxy L_F=M (the expected length in the landings when fishing mortality is equal to natural mortality). The proxy depends on the ratio of natural mortality M and the growth rate K from the von Bertalanffy equation. A ratio of 1.5 is assumed in standard data-limited cases where K and M are uncertain (Prince et al., 2015). Other crab and lobster species have been characterized as having an M/K ratio around 1.5, with unexploited stocks dominated in numbers by juveniles (Prince et al., 2015). Species with lower M/K ratios, such as Nephrops, exhibit an adult modal size (Prince et al., 2015). Given that the parameter values for M and K as used in the LCA (Table 6) are not recent (originating from Chapman (1994)) and additionally, that natural mortality is difficult to estimate, we use the default ratio of 1.5 for the calculation of reference point L_F=M , such that L_F=M =0.75L_c+0.25L_∞ (ICES, 2014b). L_∞ refers to the asymptotic size parameter from the von Bertalanffy growth equation (ICES, 2014b). As for M and K, L_∞ estimates were also obtained from historical data (Chapman, 1994) with the exception of those applied to the Shetland area (Table 6).

A decrease in mean size may be due to a reduction in the number of large individuals in the population due to fishing pressure but may also be as a result of good recruitment, which causes an increase in the number of small individuals. To mitigate the effects of variable recruitment, the mean size of the largest 5% of landed individuals L_max5% was examined (ICES, 2014b; Probst et al., 2013). The mean size of the largest 5% was compared to the reference point 0.9 L_∞ which is considered to be a better F_MSY proxy than the precautionary reference point of 0.95 L_∞ (Miethe and Dobby, 2016). A declining trend in the mean size of largest animals can be associated with a reduction in the abundance of large individuals due to the effects of fishing mortality. Due to data-limitation, we used available values of life history parameters derived for particular regions such as Shetland, Hebrides and Firth of Forth for other areas (Table 6, Table 7).

Trends in the average annual sex ratio in the sampled landings were explored for each of the three species, again excluding years in which fewer than 50 animals were sampled.

2.6. Survey Data

The habitat preferences and spatial distribution of Scottish crustacean stocks are poorly understood and the stocks are often described as data-limited for stock assessment purposes. By using fishery-independent surveys in the form of scallop trawl and dredge surveys, estimates of abundance and recruitment of more commonly caught by-catch species (such as brown crab) can be assessed. In this report, the spatial distribution of brown crab was described for three study areas East Coast (Figure 18), West Coast (Figure 19) and Shetland (Figure 20), and abundance and recruitment indices for each area developed using methodologies outlined by Mesquita et al., 2021. While these study areas do not directly match the assessment areas outlined in the other assessment methods in this report, they do cover larger regional areas of Scotland and give an indication of abundance and recruitment trends given the data available. Data was provided by fishery-independent dredge and trawl surveys carried out by Marine Scotland Science research vessels between 2008 and 2020, bathymetry data from the UK Hydrographic Office by OceanWise Ltd and sediment layers from the British Geological Survey (BGS) (Cooper et al., 2013; OceanWise, 2015). Survey data for Shetland was limited to dredge only data due to the low number of trawl survey stations sampled in the inshore waters of the islands. Additionally, the dredge survey was not completed in Shetland in 2008, 2012, 2014 or 2015 due to poor weather.

2.6.1. Spatial Distribution

Geostatistics were used to explore the spatial distribution of brown crabs using catch rates from dredge and trawl surveys. For each study area, analysis was performed separately for males, females and juveniles under 100mm carapace width (CW). Due to fluctuations in crab catch rates, including the presence of zeros (no crabs caught), the data was logarithmically transformed as follows:

where Cij is the log-transformed catch rate (response variable), 𝑟̅_ij is the catch rate in station i and year j, and _j is the mean catch rate in year j.

Variogram models were then fitted to the transformed dredge and trawl surveys in each year using the estimator:

where γ is the calculated semi-variance for each dataset, Z(x_i) is the catch rate of brown crabs at sampled station x_i, Z(x_i + h) is the catch rate value separated from x_i by a lag distance h (measured as a straight line, in km) and N(h) is the number of observation pairs separated by h. A distance h of 10km was used for the dredge survey and 25km for the trawl survey (number of lags = 15) (Mesquita et al., 2021). Following the approach proposed by Rivoirard et al., 2000, the variograms for each survey per year were combined to obtain a mean variogram standardized by the yearly sample variance. The mean variograms were then visually inspected and used to fit either spherical or nested linear models to the data (Mesquita et al., 2021). Kriging was then applied using the predictor variables distance to coast, depth and BGS sediment type.

2.6.2. Abundance and Recruitment Indices

An abundance estimate of brown crab in each study area was estimated using generalized additive models (GAMs). Two sets of GAMs (one for dredge surveys and one for trawl surveys) were created using the catch rate per haul (numbers of crabs caught per square meter) as the response variable, using the Tweedie distribution to account for zero observations and skewed data distribution.

Recruitment indices were also calculated using GAMs in a two-part hurdle model as described by Mesquita et al. (2021), with the response variable of the catch rate of crabs (N m⁻²) under 100 mm CW. As very few crabs below 100 mm CW are caught in the trawl surveys, only dredge survey data is used in the calculation of recruitment indices. Explanatory variables for the abundance models and the recruitment models were sediment type (BGS data, categorical variable with four levels), depth, distance to coast, year (2008–2020), and geographical position (latitude and longitude plus an interaction term between these) at each sampling station. Models were selected using a backward stepwise selection based on the analysis of deviance of fitted models.

2.7. Length Cohort Analysis (LCA)

Age determination is generally not possible for animals that moult, and therefore the application of age-structured assessment methods to crustacean stocks is problematic (Smith and Addison, 2003). Length Cohort Analysis (LCA) (Jones, 1984) is a commonly used method of assessing stocks for which commercial catch length frequency distribution data are available. LCA was used by MSS in previous assessments of the Scottish crab and lobster stocks (Mill et al., 2009; Mesquita et al., 2011; Mesquita et al., 2016), by Shetland UHI to assess crustacean stocks around Shetland (Leslie et al., 2007; Leslie et al., 2010) and also by Cefas (Centre for Environment, Fisheries and Aquaculture Science) to assess crab and lobster stocks in England (Cefas, 2020a; Cefas, 2020b).

The LCA method uses the commercial catch size composition data (length frequency data) and estimates of growth parameters and natural mortality to estimate fishing mortality at length. The key assumption of the approach is that the length distribution is representative of a typical cohort over its lifespan. However, this is only true of length frequency data from a single year if the population is in equilibrium, therefore, LCA is usually applied to data averaged over a number of years during which it is assumed that there were no major systematic changes in the stock (e.g. in recruitment and exploitation rates). LCA also assumes uniform growth among animals. The results of LCA can be used to predict changes in the long-term (equilibrium) stock biomass and yield-per-recruit (YPR) based on changes in mortality, potentially resulting from fishing effort reductions or changes to minimum landing size regulations. The approach gives an indication of the exploitation of the stock in terms of growth overfishing, but not recruitment overfishing.

LCAs were applied to all stocks where sampling data were collected and considered to be sufficient (Annex A). To account for the differences in growth and length-weight relationships (Table 6), males and females were assessed separately. For the Shetland stocks of brown crab and velvet crab, two assessments were conducted using the two alternative sets of biological parameters (Table 6). The LCA provides estimates of fishing mortality (F) for each length class which are averaged over a fixed length range for each species and sex to give an estimate of average fishing mortality (F_bar) for each stock. Using a fixed length range in the calculation of F_bar enables comparisons of the recent F value (2016-19) with those calculated in previous assessments (2006-08 and 2009-12, 2013-15) and offers the potential to detect trends in F.

2.8. Reference Points

The results of LCA were used to calculate yield-per-recruit and biomass-per-recruit (BPR) relative to changes in fishing mortality, which provide an indication of stock status in terms of growth overfishing. The relationship between YPR and F is typically dome-shaped – low levels of F result in low landings as few individuals are caught, whilst high levels of F may also result in a reduction in yield (and biomass) from a particular cohort as animals are caught before they have had time to grow to a size that would contribute much weight to the yield (growth overfishing). In between these lies F_MAX, the fishing mortality rate that maximizes YPR for a particular selection pattern. For data-limited stocks such as crabs and lobsters it is not possible to directly estimate the maximum sustainable yield (MSY) and hence, in line with previous assessments, F_MAX was used as a proxy for F_MSY. This approach has been widely used by ICES (ICES, 2010) for other crustaceans such as Nephrops (e.g. ICES, 2014a). F_MSY proxy values for the majority of the stocks were estimated from the most recent per-recruit analysis, based on the results of the LCA of 2016-2019 landings-at-length data. For lobster stocks in two areas (North Coast and Ullapool), 2016-19 sampling data were not available and so for these stocks, reference points were retained from previous round of assessments. All F_MSY proxy values remain preliminary and may be modified following further data exploration and analysis. A summary of stock status in terms of fishing mortality in relation to F_MSY was provided for each stock. A stock was classified as being fished “at F_MSY” when the estimated F was within 10% of F_MSY.