Improving the quality of information on Scottish anglerfish fisheries: making use of fishers' data
1 FRS Marine Laboratory, PO Box 101, Victoria Road, Aberdeen AB11 9DB, UK
2 NAFC Marine Centre, Port Arthur, Scalloway, Shetland, UK
Correspondence to H. Dobby: tel: +44 1224 876544; fax: +44 1224 295511; e-mail: h.dobby{at}marlab.ac.uk
Dobby, H., Allan, L., Harding, M., Laurenson, C. H., and McLay, H. A. 2008. Improving the quality of information on Scottish anglerfish fisheries: making use of fishers’ data. – ICES Journal of Marine Science, 65: 1334–1345.In recent years, the International Council for the Exploration of the Sea (ICES) Working Group on the Assessment of Northern Shelf Demersal Stocks has been unable to provide an analytical assessment for anglerfish. One of the reasons for this has been the poor quality of the commercial catch-and-effort data, with ICES and the European Commission’s Scientific, Technical, and Economic Committee for Fisheries (STECF) stressing the need for reliable information on which to base estimates of stock status. In response, and following consultation with the fishing industry, an anglerfish tallybook project was implemented in Scotland as part of a long-term approach to providing better data. Tallybooks are completed on a haul-by-haul basis. Skippers record catches of anglerfish (by size category) and other species where possible, together with information on haul location, duration, and depth. Individual vessel catch rates are calculated and used to provide insights into temporal trends in the stock and the spatial distribution of the fishery. The history of the fishery and management advice are summarized, and an overview of the tallybook project is provided. Catch rates are analysed using a generalized additive modelling approach which incorporates seasonal, annual, spatial, and vessel-dependent effects. The results show increased catch rates between 2006 and 2007.
Keywords: catch rate, fishers' data, GAMs, landings per unit effort, tallybook
Received 11 January 2008; accepted 17 August 2008.
 |
Introduction |
|---|
 Top Introduction MethodsResultsDiscussionReferences |
|---|
Landings of anglerfish from the waters around Scotland are primarily white-bellied anglerfish (Lophius piscatorius) and generally only a small percentage (
6%) of black-bellied anglerfish (Lophius budegassa; Laurenson et al., 2008a), with the two species being landed and processed together. Historically, anglerfish was an unimportant component of the catch of mixed species demersal fisheries around Scotland, often being discarded. However, as traditional whitefish stocks have declined in the region, anglerfish have become increasingly important. Before 1985, the total reported landings of anglerfish from ICES Subareas IV (the North Sea) and VI (West of Scotland and Rockall) had been <10 000 t. However, during the late 1980s and early 1990s, the fishery, prosecuted mainly by vessels from France, the UK (Scotland), Ireland, and Spain in Subarea VI, and the UK (Scotland), Denmark, and Norway in Division IVa, expanded rapidly, with landings peaking at almost 35 000 t in 1996 (Figure 1).
|
The rapid increase in reported landings in Divisions VIa and IVa (the most important areas for the fishery) was followed by a notable decline in subsequent years [before the introduction of restrictive total allowable catches (TACs)], indicating that the earlier increase was almost certainly based on the expansion of the fishery and increases in its efficiency (ICES, 2000, 2003). Subsequent declines in reported landings are likely to be the result of restrictive TACs, as explained below.
As the fishery in Division VIa developed, the precautionary TAC set for ICES management areas VI, XII, XIV, and Vb became restrictive and there was a tendency for vessels to misreport landings as coming from the North Sea, where no restrictions on landings were laid down. When the TAC for the North Sea was introduced in 1999 (management areas IIa and IV), it was set at a level consistent with historically reported landings, so the practice of area misreporting was institutionalized. The TACs for both areas were precautionary, and there was increased pressure on ICES to provide assessments for anglerfish.
In subsequent years, ICES attempted to provide advice in support of an analytical TAC (based on an accepted assessment and projection) by exploring a number of assessment methods at the ICES Working Group on the Assessment of Northern Shelf Demersal Stocks (WGNSDS). Based on particle-tracking studies (Hislop et al., 2001) and microsatellite DNA analysis carried out as part of an EU-funded research project (Anon., 2001), the northern shelf stock of anglerfish was at that time considered for assessment purposes to consist of Division IIIa, Subarea IV, and Division VIa (although management was still based on the areas defined above, and projected TACs were split according to recent landings proportions). Therefore, area misreporting was not a crucial issue when compiling the catch data required for an assessment. Because of problems with age determination of anglerfish, approaches focused on making use of length-structured catch data. The development of an appropriate method of assessment was further hampered by a lack of reliable biological information, such as on growth, maturity, and stock distribution, and also by limited survey data (ICES, 2005).
In 2002, a modified catch-at-size analysis (Sullivan et al., 1990), which made use of commercial catch-at-length data, effort data, and a recruitment index derived from survey data, was accepted by ICES as the basis for advice. The method addressed known problems associated with age readings and the rapid expansion of the fishery and provided a satisfactory fit to the data. Concomitantly, a precautionary reference point for fishing mortality F (Fpa) was proposed for the stock based on the value implied by 35% pristine spawning-stock biomass per recruit [proposed by the ICES Advisory Committee on Fishery Management (ACFM), adapted from Mace and Sissenwine, 1993; ICES, 1997]. The proposed precautionary reference point was very low (much lower than the estimated F) because of particular life-history characteristics (a high age-at-maturity), which suggested that anglerfish may be vulnerable to high rates of exploitation. The ICES advice following the precautionary approach was therefore for a reduction in TAC, such that F would be below Fpa, implying a substantial reduction in fishing mortality.
Based on the advice from 2002 and successive years, TACs were reduced. However, the Scottish fishing industry felt that these successive reductions in TAC were not consistent with an apparent upturn in stock abundance, which they had widely observed on fishing grounds over a number of years. In subsequent years (2003–2005), TACs proved highly restrictive, providing a strong incentive to discard and to underreport landings. As a result, reported landings became increasingly unrepresentative of total catch, and assessments based on them were progressively unreliable. Although it is acknowledged that underreporting of landings has been a problem in the fishery, it is thought that the situation has improved a lot since the implementation of the registration of buyers and sellers legislation in the UK and Ireland in late 2006, under Council Regulation (EEC) No. 2847/93 (ICES, 2007a).
In addition to the poor quality of commercial catch data, concerns about the uses of recruitment abundance indices as auxiliary data, derived from data collected on traditional groundfish surveys, were raised by ACFM. Catch rates of anglerfish in the surveys, which are designed to provide indices of abundance for other demersal species, are generally low because of the use of unsuitable groundgear and typically poor coverage of the area and depth distribution of anglerfish. Extensive exploration of the available survey data was carried out at the WGNSDS meeting in 2004. The surveys investigated showed few common trends and little internal consistency, leading the working group (WG) to conclude that they were unlikely to provide a reliable indication of year-class strength.
As a result of the deteriorating quality of commercial data and inadequate survey data, the WG in 2004 felt unable to present an analytical assessment for anglerfish. ICES concluded that the stock status be considered "unknown" in relation to precautionary limits and advised (ICES, 2004) that effort in the fishery should not be allowed to increase and that the fishery be accompanied by mandatory programmes to collect catch-and-effort data on both target fish and bycatch.
In response, FRS (Fisheries Research Services, Aberdeen) embarked on a new scientific programme for improving data availability for assessing and managing anglerfish, consisting of a number of elements: a new industry/science dedicated anglerfish survey (Fernandes et al., 2007), increased observer coverage (in 2005 and 2006), and a tallybook scheme. During late 2004 and 2005, a number of fishers had provided data from their private diaries in an effort to improve understanding of the fishery and the state of the stock. Analysis of the diary data provided valuable information to ICES (ICES, 2005) on temporal and spatial trends in catch rate. Following the success of this data-collation exercise, a formal scheme for collecting catch-rate data was proposed.
Commercial catch per unit effort (cpue) data are often used to obtain indices of abundance for fish stocks, and in particular for stocks with a large spatial distribution for which fishery- independent scientific surveys would be difficult or costly to carry out (Campbell, 2004; Nishida and Chen, 2004). Although FRS has recently implemented an annual fishery-independent, multi-vessel anglerfish survey (Fernandes et al., 2007), the commercial catch rate data collected as part of the tallybook scheme are a much richer dataset, covering a broad area with continuous, high-resolution sampling. It is therefore considered that with appropriate standardization to account for factors influencing catchability, these data could provide useful information on relative stock abundance.
Here, we describe the tallybook data-collection scheme, provide an overview of the data collected, and present an analysis of catch rates from the first two years of the scheme.
|
Methods |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Data
Extensive discussions with the fishing industry during 2005 resulted in FRS implementing the anglerfish tallybook project at the start of 2006. The project is part of a long-term approach to providing better information on the anglerfish fishery and the state of the stock and is being operated in conjunction with fisher organizations [Scottish Fishermen’s Federation (SFF), Fishermen’s Association Limited, and Pêcheurs de Manche et Atlantique] and the North Atlantic Fisheries College (NAFC) Marine Centre, Shetland. These organizations have been responsible for distributing the tallybooks, coordinating the returns, and allocating vessel codes before the anonymous tallybook sheets are forwarded to FRS. Effort has been made to involve the main fishers for anglerfish, but also vessels fishing on the continental shelf for which anglerfish is more of a bycatch.
The tallybooks are completed on a haul-by-haul basis, skippers being asked to record catches (landings and discards) of anglerfish (by size category) and other species where possible, together with information on haul location, duration, and depth, and to provide a description of the gear used. Additionally, information on catches of mature females, in terms of number of mature fish per haul, has also been requested. Each tallybook sheet also contains a field for vessel name and registration number, but this is overwritten and replaced with a code before being sent to FRS. The SFF and NAFC retain the record of vessel names and allocated codes, to which FRS has no access. Data storage is currently in a relational database at FRS. Tallybook returns from vessels based in Shetland are processed by staff at NAFC, then forwarded electronically to FRS.
Catches (or at least landings) are variously recorded in kilogrammes or boxes. For boxes, each vessel has a nominal "box weight" associated with it, and at the data-entry stage, box numbers by size category are converted into weight. A number of data-checking procedures and "fill-ins" are also carried out at this stage: for example, missing depth units or gear descriptions are inserted by referring to previous hauls of that vessel.
To date, tallybook returns have been received from 37 fishing vessels with a wide spatial coverage (almost 13 500 hauls), as shown in Figure 2. Most of the fishing activity reported was in the northern North Sea (around Shetland), and off the north and west coasts of Scotland, along the shelf edge. A smaller, but still significant, number of records come from Rockall, Stanton Bank, and the Fladen Ground. The location of these places in relation to Scotland and the ICES Divisions is shown in Figure 3a.
|
|
In all, 12 vessels have been continuous participants throughout the duration of the project, and the data presented and analysed below are restricted to those vessels. The vessels are typical of the larger sample in that they consist of vessels apparently targeting Nephrops (low average catch rates of anglerfish) and those for which anglerfish is a more important component of their catch. The spatial distribution of fishing activity, which is also shown in Figure 2, is similar to that of the larger sample, although the subset contains no hauls from the more inshore waters around the south of the Outer Hebrides and Stanton Bank, or the more southerly part of the shelf edge. Hauls carried out there were by vessels that were only short-term participants in the scheme.
Figure 3a depicts the spatial distribution of officially reported Scottish anglerfish landings (EU logbook data) averaged over 2006 and 2007 (adjusted for area misreporting; ICES, 2008). The tallybook data show good spatial coverage of the areas of high reported Scottish landings, the only exception being the lack of data from the statistical rectangle directly south of the Outer Hebrides, which in recent years has had average landings of some 200 t.
Table 1 shows the total landings of the 12 tallybook vessels compared with the total reported landings of all Scottish vessels landing into Scotland for 2006 and 2007 by month and ICES Division. Coverage varies from area to area and month to month, but the most notable feature is that, in several cases, the landings recorded in the tallybooks significantly exceed the officially reported landings (e.g. January in Division VIb). This is likely to be a result of area misreporting (both years) and underreporting of official landings, particularly in 2006 before the introduction of buyers and sellers legislation, and highlights the unreliability of the official data. Total official landings in 2007, i.e. summed over all areas, may be more representative of actual landings. The tallybook landings in 2007 are 20% of the total annual reported landings.
|
According to official data (Table 2), the fishery at Rockall (VIb) is prosecuted by relatively few vessels (14 in 2006, 26 in 2007), some 20% of which are covered by tallybook data. In Divisions IVa and VIa, far more vessels are involved in the fishery, although a large number of them, particularly in Division VIa, report annual landings of >1 t. Vessel coverage by the tallybook scheme is therefore poorer in these areas than at Rockall. Most hauls are made in depths of 100–400 m, although there are a significant number of hauls from depths between 600 and 800 m, which are the result of some vessels making occasional trips into deeper water (Figure 4).
|
|
As most of the fishing activity recorded by these 12 vessels was in Divisions IVa (North Sea), VIa (West of Scotland), and VIb (Rockall) in water <1000 m deep, the statistical analysis presented here excludes the small number of records from outside these areas.
Analysis
Although vessels were asked to record discards on the tallybook sheets, only 5 of the 12 provided such information. For the other vessels, it is not clear whether a blank record implies zero discards or whether it is the result of incomplete data recording. Therefore, as just 15% of the trips in the subset of data contain any information on discards, landings data were used in the analysis. In the tallybooks, landings are recorded in size categories such as "small", "medium", and "large". However, it has not yet been possible to reconcile these categories with sizes of anglerfish, so total catch rate (in kg h–1) was calculated. As a first step, exploratory analysis, including calculation of basic summary statistics and scatterplots of catch rate (or, strictly speaking, landings per unit effort, lpue) against the other variables recorded in the tallybooks, was carried out to help identify possible relationships among the variables.
A statistical modelling approach using generalized additive models (GAMs) was used to determine temporal and spatial trends in lpue. The most common method for standardizing cpue data, first demonstrated by Gavaris (1980), is the use of generalized linear models (GLMs; Nelder and Wedderburn, 1972). The basic assumption of a GLM is that there is a linear relationship between some function (g) of the expected value of the response variable (yi) and the explanatory variables (xj), which can be a numerical or categorical (discontinuous) variable:
|
| (1) |
In recent years, the use of this approach as a means for standardizing commercial catch rate data has become increasingly common (Goñi et al., 1999; Battaile and Quinn, 2004; Bishop et al., 2004), and an overview of current approaches is given by Maunder and Punt (2004) and Venables and Dichmont (2004). GAMs are an extension to GLMs (Hastie and Tibshirani, 1990), in which the linear predictor is replaced by
|
| (2) |
The distribution and link function are chosen by examining plots of deviance residuals against fitted values for systematic patterns and trends. Within the R framework, a wide range of distributional assumptions and link functions can be explored, including standard distributions (such as Gaussian, gamma, and Poisson) and more-flexible distributions known as "quasi" distributions. This family of distributions is defined by a link function and variance, but the dispersion parameter is estimated, so allowing the model to produce a better fit to overdispersed data. Here, a quasi-distribution with logarithmic link and variance proportional to the mean appeared to be a fair assumption.
Backward stepwise selection, a procedure that involves initially including all likely covariates in the model, then sequentially removing those that do not help explain the variance in the data, was used to select an appropriate set of explanatory variables. At each stage, the least significant explanatory covariate is removed, the new model refitted, then tested against the original model using an F-test (Hastie and Tibshirani, 1990). The procedure is repeated until the reduced model provides a significantly different fit (p < 0.05) from the original model. Smooth terms are represented using penalized regression splines, with smoothing parameters selected by cross-validation.
For this study, the response variable lpue (kg h–1), was modelled, with month, year, depth, and spatial effects. Clearly, the catch rates are likely to differ significantly between vessels operating in the same spatio-temporal stratum as a result of both gear type and vessel size. From the descriptions given in the tallybooks, two factors were considered in the model: gear was categorized as either single or twin-rig, and mesh size as <100 or 
100 mm. Following the approach of Battaile and Quinn (2004), an additional factor, vessel ID, was included to account for other vessel effects, such as boat size/power and experience/skill of the skipper and crew. Month and depth were included as smooth terms, and area (either statistical rectangle or ICES Division) was modelled as a categorical variable.
Preliminary analysis conducted as part of this study focused on modelling the lpue on a haul-by-haul or daily basis. However, the residuals of these models showed significant temporal autocorrelation (lagged by either haul number or day), which conflicts with the assumption that the values of lpue are independent observations. In the analysis presented here, serial autocorrelation in the residuals was eliminated by aggregating the catch-rate data at the trip level, then using this as the response variable in the GAM. The resulting dataset has >450 records. Typically, fishing activity occurs over a number of statistical rectangles on a single trip, so the measure of trip location was taken to be the ICES statistical rectangle where the greatest number of hauls was carried out. The depth associated with each trip was taken to be the average over all hauls in the trip, and the date was taken as the start of the trip. The residuals from the final model were then examined for serial correlation. A comparison of models for the residuals from the final model, including and excluding an autoregressive correlation structure (using the nlme library; Pinheiro and Bates, 2000), was carried out to determine whether there was still significant serial correlation in the data.
|
Results |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Average catch rates
Monthly average anglerfish lpue ranged from <10 kg h–1 for boats likely to be targeting Nephrops (in shelf areas) to >100 kg h–1 for some whitefish boats during the early part of the year. Typical temporal trends in average lpue by vessel are shown in Figure 5. Some of the vessels show a marked seasonal pattern of reduced catch rates during summer and higher average catch rates from November to April, like that shown for Boat 3 in the Figure. Considering average monthly catch rates alone, it is unclear whether there has been any overall interannual variation in average catch rate. A number of vessels appear to have higher monthly average catch rates over some or all of 2007 than in the equivalent period of 2006 (e.g. Boats 1 and 3), but others show no change (e.g. Boat 2).
|
Figures 6a and b are scatterplots of lpue against depth and month for all vessels and years combined. Values of lpue were high across almost the full range of depths (except hauls shallower than 100 m). However, a greater proportion of the hauls which were carried out in depths <300 m have a relatively low lpue (<50 kg h–1), suggesting that depth may be a significant explanatory variable in any lpue standardization. Lpue plotted against month shows a wide range of values at all times of a year, although mean values (across all vessels, years, and areas) are lower during summer. Figure 6c is a boxplot of lpue by ICES Division. The point estimate of median lpue in Division IVa (28 kg h–1) is less than half that for Division VIa (70 kg h–1). Although there is a lot of variability in lpue within each area (as a result of the inclusion of all vessels, all years, and all areas), the differences in the point estimates do suggest that it is worth considering ICES Division as an explanatory factor when modelling lpue. Further, the distribution of officially recorded landings (Figure 2a) suggests that some regions of each ICES Division are more important, in terms of landings, than others. This may be related to catch rate (or relative abundance) in these areas, so spatial effects are also considered at the ICES statistical rectangle level.
|
Statistical analysis
The full model initially considered in the analysis included month, year, depth, area (ICES Division), mesh, gear type, and vessel ID as covariates. Interaction terms were included to investigate the possibility of there being different temporal effects in different ICES Divisions. Month, year, vessel ID, ICES Division, and depth were significant terms. When vessel ID was included, the variables describing gear and mesh were not significant explanatory variables. Of the interaction terms, a seasonal effect by area (ICES Division) was retained in the model, but incorporating a single annual effect for all areas was adequate.
Although this initial model gave a reasonable fit to the data (explaining some 60% of the deviance), using ICES statistical rectangle as the area explanatory variable rather than ICES Division (but retaining ICES Division in the interaction term) resulted in a much improved fit to the data (explaining more than 80% of the deviance). However, when including statistical rectangle rather than ICES Division as the area factor, depth is no longer a significant term (p >> 0.01) in the model, because trips associated with a particular statistical rectangle have similar average fishing depths. A summary of this, the final model, is shown in Table 3.
|
There is a relatively good relationship between observed lpue and fitted values (Figure 7a). Plots of the deviance residuals show a relatively symmetrical distribution, with no obvious departures from normality (Figure 7b and c), and the residual variance shows no significant changes through the range of fitted values (Figure 7d). Additionally, there are no systematic patterns in the residuals when plotted against the explanatory variables (Figure 8). A model of the residuals which allowed for an autoregressive term (correlation in the residuals of successive trips grouped at a boat level) was not significantly different from those of a model with no autocorrelation (p
0.15).
|
|
The estimated effects of interest in this final model are shown in Figure 9. Although the annual effects from the model are quite uncertain, the catch rates reported in 2007 are estimated to be significantly higher (by
15%) than those in 2006. The estimated seasonal effects are consistent with previously observed seasonal patterns in the fishery, whereby catch rates decrease during summer (ICES, 2005). Although this pattern is evident across all ICES Divisions, the timing and magnitude of the changes is somewhat different among areas. The lowest values in the North Sea are around June, in Division VIa in September, and at Rockall in July/August. The magnitude of the intra-annual fluctuations is estimated to be much greater in Division VIa and VIb than in Division IVa, lpues estimated to increase by around 80% over the last 4 months of the year. Estimated spatial effects are shown in Figure 3b. The statistical rectangles with greatest effect are at the eastern and southeastern side of Rockall, and west of Shetland and the Outer Hebrides. The areas of lowest standardized lpue are in the North Sea, at Fladen.
|
|
Discussion |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Temporal patterns
The analysis presented here shows a significant increase in lpue between 2006 and 2007. Other sources of information also indicate a buoyant anglerfish stock recently. Analysis of commercial catch rates obtained from data collected by observers on board Scottish fishing vessels showed increases recently (ICES, 2007a), although this analysis has not been updated to include 2007 data. In addition, preliminary estimates from the Scottish anglerfish survey show biomass increases of
30% between 2006 and 2007, although the broad confidence intervals on the estimates mean that the increase is not statistically significant (Fernandes, 2008). The estimates of seasonal effects show large increases in catch rate at the end of the year for both Divisions VIa and VIb. This is consistent with previously observed seasonality of catch rates in these areas (ICES, 2005), and with fishers’ experience. Lowest catch rates are estimated for June in IVa and for September in VIa. This differs somewhat from the diary data analysis of ICES (2005), which also used GAMs and found lowest catch rates on the shelf edge in July and at Fladen around September/October. These differences could reflect the use of a more aggregated dataset in this analysis (at a trip rather than a day level) or maybe by slightly different definition of the areas in the two studies. ICES (2005) estimated seasonal patterns for the shelf edge (across IVa and VIa) and Fladen separately, whereas here we estimated separate seasonal effects for each ICES Division. Appropriate spatial stratification for the estimation of seasonal patterns in tallybook catch rates need to be investigated further. The estimated seasonal trends in catch rate at Rockall (lowest in June) are consistent with those documented by ICES (2005).
The seasonal patterns in the northern shelf anglerfish fishery are well documented, and although both large- and small-scale anglerfish movements have been observed (Laurenson et al., 2005), the migration patterns are not well understood. Anglerfish are thought to spawn in deep water (Fulton, 1903; Bowman, 1920; Hislop et al., 2001) along the shelf edge north and west of Scotland, with a protracted spawning period over the first five months of the year (Afonso-Dias and Hislop, 1996). Observed changes in catch rate may therefore be linked to movements to spawning areas or to movements of the whole stock. However, given that few fish of spawning size are collected from market samples at any time of the year (ICES, 2006), it seems unlikely that the variations in catch rate are caused only by spawning behaviour. Making use of additional tallybook information on size composition and catches of spawning females may help explain the patterns.
Spatial effects
Although only half the vessels in our sample made fishing trips to Rockall, there is sufficient spatial and temporal overlap of their fishing activity with that of other vessels (in both Divisions VIa and IVa) to be able to estimate spatial and vessel effects. The apparent high lpue at Rockall is not just a consequence of larger vessels fishing at Rockall. The 2006 and 2007 anglerfish surveys carried out by FRS found the greatest densities (kg km–2) at the eastern and southeastern edges of Rockall (Fernandes, 2008), consistent with our findings. The spatial effects estimated by the model across the rest of the area of study are also relatively consistent with those shown by Fernandes (2008), with moderately high catch rates west of Shetland and much lower ones across the northern North Sea.
Data
The validity of the results is obviously strongly dependent on the accuracy of the data being supplied. Participation in the scheme is voluntary, with no immediate pay-off, and it is thought that the vessels taking part have made a real commitment to the project and to providing accurate returns. Tallybook data are made anonymous before being sent to FRS, so there is little incentive for vessels to misreport landings. Data validation has been carried out by FRS observers when they happen to be on board vessels involved in the tallybook scheme, but such validation covers <2% of the trips analysed. In addition, the data were scrutinized thoroughly before analysis, to check for errors that may have crept in either during data entry or while the tallybook sheets were being completed. Records that included hauls of atypical duration or too many hauls per day, for example, were excluded at the outset. As the vessels are anonymous, and the official data (both effort and landings) have in the recent past been of such poor quality (through misreporting), it is not possible to carry out data comparisons to ascertain the accuracy of our data. In other parts of the world (Starr and Vignaux, 1997), experience suggests that fisher information from voluntary logbook schemes can provide useful and accurate data on catch rate (and size composition), which may be useful for stock assessment.
The analysis conducted here is based on lpue and does not include discard information, because this has so far not been widely recorded by the vessels participating in the scheme. However, this is unlikely to significantly undermine the conclusions on spatial and temporal distribution drawn from the statistical analysis because, by and large, discarding in the Scottish anglerfish fishery appears to be relatively limited. Of the hauls included in the tallybook dataset, some 700 contain information on discards, with 25% of these having zero discards and the median discard rate being <5% by weight. Although it is not clear whether this is a true reflection of current practice, it does tie in with the results of other studies (Kunzlik et al., 1995; Laurenson, 1999) which found low levels of discarding in the Scottish anglerfish fishery. Additionally in 2005/2006, overall discards obtained by the NAFC from observer sampling were calculated to be 0.3% by weight (CHL, unpublished data).
Discard data would still be useful, however, for enhancing knowledge of the dynamics and distribution of the stock, and participating vessels are still encouraged to complete this record in the tallybook. Such information could, for example, indicate the presence of a particularly strong year class if high levels of discards (of small fish) were recorded, which could result in increased levels of biomass in subsequent years. It has been suggested that shallow water, particularly around Shetland, may act as nursery areas for juvenile anglerfish (Laurenson et al., 2008b), and large numbers of small anglerfish have been caught in this area on the Scottish anglerfish survey (Fernandes, 2008). Further information on the spatial distribution of discards of small fish may confirm this hypothesis and also help to identify other nursery areas across the northern shelf. It is not clear why the discard field is left blank so frequently, although it has been suggested by FRS observers that skippers of fishing vessels (who fill in the tallybook) often will not be aware of how much or how little has been discarded because they are not on deck while the catch is being processed. A fuller explanation to participants of why these data are important may improve completion rates.
Participants were also asked to supply information on the quantity of other species caught (or landed), to assist with fishery definition. Most participating vessels do not complete this section of the tallybook, precluding further analysis. Previous attempts at defining anglerfish fisheries (STECF, 2007) have shown that the vast majority of Scottish anglerfish landings come from mixed trawl fisheries catching a range of demersal species and Nephrops, with the landings of anglerfish as a relatively low percentage of the total. However, those analyses were based on data reported officially at a trip level, so are likely to suffer from the misreporting problems discussed above. The tallybook scheme represents an opportunity to improve our understanding of the fisheries in which anglerfish are caught, so fishers are still encouraged to supply the mixed fishery data.
Concluding remarks
The tallybook scheme described here has delivered relatively extensive information on the spatial and depth distribution of catch rates in the Scottish anglerfish fishery, which is unavailable (or unreliable) from other sources such as EU logbook data. The project was initiated as part of a long-term approach to obtaining better commercial fishery information, so continued participation by a large number of vessels would result in a more valuable dataset that could be used to provide information on temporal changes in fishery catch rates and potentially the state of the stock, as illustrated in the analysis presented above. In all, 37 boats have been involved with the scheme, but the number of boats returning tallybooks dwindled to 12 by mid-2007. Reduced participation has also been a problem in other schemes of this type, and Johnson and van Densen (2007) described two such examples. The F-project, implemented in the Netherlands to obtain cpue and discard data for use in the assessment of flatfish species, has had limited success, but currently has just a few skippers participating, resulting in poor coverage of a fishery’s distribution. Similarly, limited participation in a programme set up in the northeastern US to provide real-time data on the mid-Atlantic squid fishery has meant that the aim of real-time management of the fishery has yet to be accomplished. In 2001, the Institute of Marine Research (IMR), Bergen, Norway, started a fisher self-sampling scheme (known as the "reference fleet") to collect catch data and biological information (Helle and Pennington, 2004). Participation in that scheme has remained relatively high (ICES, 2007b), but in contrast to the tallybook scheme described here, fishers receive payment for their involvement.
ICES (2007b) lists a number of incentives which are deemed important for motivating fishers to participate in data-collection schemes, including a form of compensation, such as increased TAC or direct payment. However, ICES (2007b) also stress that knowing that participation in such schemes is essential and/or useful in stock management is an effective motivating factor in itself. Clearly this has been sufficient encouragement for some Scottish anglerfish fishers to remain members of the tallybook scheme, but not enough for others. The initial enthusiasm for the tallybook scheme may have been the result of fishers’ belief that there would be some immediate reward (perhaps in terms of an increased TAC) for providing data. It is clear that the objectives of such data-collection schemes need to be fully discussed with and understood by potential participants at the outset. Johnson and van Densen (2007) stress the need for good communication in successful cooperative projects. Feedback from our project has been given to the industry in the form of personalized summary reports and occasional presentations, with the aim of retaining interest and support for the scheme. The anonymous nature of the tallybook scheme makes the communication process rather difficult. All discussion of the data with individual participants has to be carried out through an intermediary such as the SFF or the NAFC. This is not ideal, and possible ways to improve the situation, such as holding data workshops or discussion groups, are currently being considered.
Although there are acknowledged problems with the Scottish tallybook scheme, such as incomplete data and falling participation, the importance of the data is recognized by industry representatives and scientists. WGNSDS has recommended a workshop on anglerfish assessment and advice (ICES, 2008), part of which will focus on the use of self-sampled (tallybook) data in the assessment and advisory process. Further discussions will therefore take place with the Scottish fishing industry on possible improvements to the current feedback/communication format, with the aim of fuller completion of the tallybook records and retaining current participation levels, as well as the continued usefulness of the data.
|
Acknowledgements |
|---|
We thank the skippers of the vessels taking part in the Scottish tallybook scheme and Rory Campbell of the SFF for assisting with the coordination of the scheme. In addition, comments and suggestions by Steve Cadrin and an anonymous reviewer led to a much improved paper.
|
References |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
-
Afonso-Dias I. P., Hislop J. R. G. The reproduction of anglerfish Lophius piscatorius Linnaeus from the north-west coast of Scotland. Journal of Fish Biology (1996) 49:18–39.[CrossRef][Web of Science]
Anon. The distribution and biology of anglerfish and megrim in waters to the west of Scotland. (2001) Oban, Scotland: Scottish Association for Marine Science. 88. Final Report European Commission—DGXIV Study Contract 98/096, Coordinator J. D. M. Gordon.
Battaile B. C., Quinn T. J. Catch per unit effort standardization of the eastern Bering Sea walleye pollock (Theragra chalcogramma) fleet. Fisheries Research (2004) 70:161–177.[CrossRef][Web of Science]
Bigelow K. A., Boggs C. H., He X. Environmental effects on swordfish and blue shark catch rates in the US North Pacific longline fishery. Fisheries Oceanography (1999) 8:178–188.[CrossRef][Web of Science]
Bishop J., Venables W. N., Wang G. Analysing commercial catch and effort data from a penaeid trawl fishery: a comparison of linear models, mixed models, and generalised estimating equations approaches. Fisheries Research (2004) 70:179–193.[CrossRef][Web of Science]
Bowman A. The eggs and larvae of the angler (Lophius piscatorius L.) in Scottish waters. A review of our present knowledge of the life history of the angler. (1920) 1. Fishery Board for Scotland, Scientific Investigations 1919. 42.
Campbell R. A. CPUE standardisation and the construction of indices of stock abundance in a spatially varying fishery using general linear models. Fisheries Research (2004) 70:209–227.[CrossRef][Web of Science]
Fernandes P. G. The 2007 northern shelf anglerfish survey. (2008) 8. Working Document to the Working Group on Northern Shelf Demersal Stocks, 2008 http://www.ices.dk/reports/ACOM/2008/WGNSDS/WGNSDS_Working_Documents_2008.pdf.
Fernandes P. G., Armstrong F., Burns F., Copland P., Davis C., Graham N., Harlay X., et al. Progress in estimating the absolute abundance of anglerfish on the European northern shelf from a trawl survey. (2007) 14. ICES Document CM 2007/K: 12.
Fulton T. W. The distribution, growth and food of the angler (Lophius piscatorius). Twenty-first Annual Report of the Fishery Board for Scotland (1902) (1903) 187–217. Part III.
Gavaris S. Use of a multiplicative model to estimate catch rate and effort from commercial data. Canadian Journal of Fisheries and Aquatic Sciences (1980) 37:2272–2275.
Goñi R., Alvarez F., Adlerstein S. Application of generalized linear modeling to catch rate analysis of western Mediterranean fisheries: the Castellón trawl fleet as a case study. Fisheries Research (1999) 42:291–302.[CrossRef][Web of Science]
Hastie T., Tibshirani R. Generalized Additive Models. (1990) London: Chapman and Hall. 335.
Helle K., Pennington M. Survey design considerations for estimating the length composition of the commercial catch of some deep-water species in the northeast Atlantic. Fisheries Research (2004) 70:55–60.[CrossRef][Web of Science]
Hislop J. R. G., Gallego A., Heath M. R., Kennedy F. M., Reeves S. A., Wright P. J. A synthesis of the early life history of the anglerfish, Lophius piscatorius (Linnaeus 1758) in northern British waters. ICES Journal of Marine Science (2001) 58:70–86.
ICES. Report of the Study Group on the Precautionary Approach to Fisheries Management, Copenhagen, Denmark, 5–11 February 1997. (1997) 41. ICES Document CM 1997/Assess: 07.
ICES. Report of the Working Group on the Assessment of Northern Shelf Demersal Stocks, June 1999. (2000) 468. ICES Document CM 2000/ACFM: 01.
ICES. Report of the Working Group on the Assessment of Northern Shelf Demersal Stocks, August/September 2002. (2003) 625. ICES Document CM 2003/ACFM: 04.
ICES. Report of the ICES Advisory Committee on Fishery Management and Advisory Committee on Ecosystems, 2004. (2004) 1544. ICES Advice, 1 and 2.
ICES. Report of the Working Group on the Assessment of Northern Shelf Demersal Stocks (WGNSDS), 10–19 May 2005, Murmansk, Russia. (2005) 757. ICES Document CM 2006/ACFM: 13.
ICES. Report of the ICES Advisory Committee on Fishery Management, Advisory Committee on the Marine Environment and Advisory Committee on Ecosystems, 2006. (2006) 271. ICES Advice, Book 5.
ICES. Report of the Working Group on the Assessment of Northern Shelf Demersal Stocks (WGNSDS), 8–17 May 2007, Galway, Ireland. (2007) a:852. ICES Document CM 2007/ACFM: 22.
ICES. Report of the Workshop on Using Fishers to Sample Catches (WKUFS), 5–6 June 2007, Bergen, Norway. (2007) b:45. ICES Document CM 2007/ACFM: 24.
ICES. Report of the Working Group on the Assessment of Northern Shelf Demersal Stocks (WGNSDS), 15–21 May 2008, Copenhagen, Denmark. (2008) 735. ICES Document CM 2008/ACOM: 08.
Johnson T. R., van Densen W. L. T. Benefits and organization of cooperative research for fisheries management. ICES Journal of Marine Science (2007) 64:834–840.
Kunzlik P. A., Newton A., Jermyn S. Exploitation of monks (Lophius spp.) and megrims (Lepidorhombus spp.) by Scottish fishermen in ICES Division VIa (west of Scotland). (1995) 125. EEC Contract No MA-2-520.
Laurenson C. H. Investigation of the monk fishery in northern European Union waters. (1999) 31. DG XIV Study Project No 96/086 http://ec.europa.eu/fisheries/publications/studies/bio/1309R03B96086.pdf.
Laurenson C. H., Dobby H., McLay H. A. The Lophius budegassa component of monkfish catches in Scottish waters. ICES Journal of Marine Science (2008) a 65:1346–1349.
Laurenson C. H., Dobby H., McLay H. A., Leslie B. Biological features of the Lophius piscatorius catch in Scottish waters. ICES Journal of Marine Science (2008) b 65:1281–1290.
Laurenson C. H., Johnson A., Priede I. G. Movements and growth of monkfish Lophius piscatorius tagged at the Shetland Islands, Northeastern Atlantic. Fisheries Research (2005) 71:185–195.[CrossRef][Web of Science]
Mace P. M., Sissenwine M. P. How much spawning per recruit is enough? Canadian Special Publication of Fisheries and Aquatic Sciences (1993) 120:101–118.
Maunder M. M., Punt A. E. Standardizing catch and effort data: a review of recent approaches. Fisheries Research (2004) 70:141–149.[CrossRef][Web of Science]
Nelder J. A., Wedderburn R. W. M. Generalized linear models. Journal of the Royal Statistical Society, Series A (1972) 135:370–384.[CrossRef]
Nishida T., Chen G. Incorporating spatial autocorrelation into the general linear model with an application to the yellowfin tuna (Thunnus albacares) longline CPUE data. Fisheries Research (2004) 70:265–274.[CrossRef][Web of Science]
Pinheiro J. C., Bates D. M. Mixed-Effects Models in S and S-PLUS. (2000) New York: Springer. 528.
R Development Core Team. R: a language and environment for statistical computing. (2007) Vienna, Austria: R Foundation for Statistical Computing. http://www.r-project.org/.
Rodríguez-Marín E., Arrizabalaga H., Ortiz M., Rodríguez-Cabello C., Moreno G., Kell L. T. Standardization of bluefin tuna, Thunnus thynnus, catch per unit effort in the baitboat fishery of the Bay of Biscay (Eastern Atlantic). ICES Journal of Marine Science (2003) 60:1216–1231.
Starr P. J., Vignaux M. Comparison of data from voluntary logbook and research catch-sampling programmes in the New Zealand lobster fishery. Marine and Freshwater Research (1997) 48:1075–1080.[CrossRef][Web of Science]
STECF. Report of the Scientific, Technical and Economic Committee for Fisheries: Evaluation of the Report of the STECF- SGRST 06-03 Working Group on Anglerfish. (2007) 77. SEC(2007) 468 http://ec.europa.eu/fisheries/publications/factsheets/legal_texts/sec_2007_468_en.pdf.
Sullivan P. J., Lai H-L., Gallucci V. F. A catch-at-length analysis that incorporates a stochastic model of growth. Canadian Journal of Fisheries and Aquatic Sciences (1990) 47:184–198.
Venables W. N., Dichmont C. M. GLMs, GAMs, GLMMs: an overview of theory for applications in fisheries research. Fisheries Research (2004) 70:315–333.
Wood S. N. Generalized Additive Models. An Introduction with R. (2006) Chapman and Hall/CRC. 392.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. H. Laurenson, H. Dobby, H. A. McLay, and B. Leslie Biological features of the Lophius piscatorius catch in Scottish waters ICES J. Mar. Sci., October 1, 2008; 65(7): 1281 - 1290. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||