ICES Journal of Marine Science: Journal du Conseil Advance Access originally published online on September 20, 2007
ICES Journal of Marine Science: Journal du Conseil 2007 64(9):1800-1819; doi:10.1093/icesjms/fsm145
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Taking account of catchability in groundfish survey trawls: implications for estimating demersal fish biomass
1 Fisheries Research Services, Marine Laboratory, PO Box 101, 375 Victoria Road, Aberdeen AB11 9DB, UK
2 Wageningen–IMARES, PO Box 68, 1970 AB IJmuiden, The Netherlands
Correspondence to H. M. Fraser: tel: +44 1224 295439; fax: +44 1224 295511; e-mail: h.fraser{at}marlab.ac.uk
Fraser, H. M., Greenstreet, S. P. R., and Piet, G. J. 2007. Taking account of catchability in groundfish survey trawls: implications for estimating demersal fish biomass. – ICES Journal of Marine Science, 64.Groundfish surveys are a key component of current scientific data monitoring and data-collection activities undertaken in support of fisheries management. Recent requirements to develop and implement an ecosystem approach to management are placing increasing and varied demands on such datasets. Successfully incorporating ecosystem and environmental objectives within fisheries management will, for example, require greater understanding of foodweb trophodynamics, which in turn requires detailed information on the abundance and distribution of fish predators and prey species on spatial scales hitherto rarely considered. However, no trawl gear catches all the fish in its path, so density estimates provided by such trawl samples do not reflect true densities of fish. Catchability in a trawl gear is affected by many factors and varies both between species and between different sized conspecifics, and therefore has the capacity to confound our understanding of predator–prey interactions and of the relative abundance of different species and size classes of fish at any point in time or space. To overcome such problems, estimates of the catchability of each size class of each species sampled in a given survey are required for trawl sample densities to be raised to the actual densities of fish present at each location sampled. Here, we present a method for estimating catchability coefficients for 1-cm size classes of fish species sampled by the Grande Overture Vertical trawl during the third quarter ICES International Bottom Trawl Survey. The catchability coefficients obtained are applied to survey data collected between 1998 and 2004 to examine annual variation in actual abundance of the demersal fish assemblage in the North Sea.
Keywords: catchability, demersal fish, GOV, IBTS
Received 14 March 2007; accepted 27 July 2007; advance access publication 20 September 2007.
 |
Introduction |
|---|
 Top Introduction Methodology and analysisDiscussionAppendixReferences |
|---|
Assessing the abundance of organisms is perhaps the most critical aspect of any study of ecological processes. In studies of predator–prey interactions, knowledge of predator abundance is essential if total predation loadings are to be assessed (Hislop et al., 1991; Wanless et al., 1998; Furness, 2002). Likewise, without knowing the abundance of prey organisms, these predation loadings cannot be converted to prey mortality rates (Greenstreet et al., 1997). As we move from traditional fisheries management towards an ecosystem approach to management (Hall and Mainprize, 2004; Frid et al., 2005), such considerations need to be extended to include interactions between fish predators and a larger number of prey species belonging to a wider variety of taxa.
Groundfish surveys have been carried out in the North Sea in some form or other since the early 1900s (Rijnsdorp et al., 1996; Greenstreet et al., 1999). Such research surveys provide estimates of the abundance of each fish species sampled at a particular location. However, no trawl gear samples all the individuals present in its path, and catch rates of fish of different species and size in a given fishing gear vary considerably. Vertical distributions of many species vary with time of day, affecting the availability of fish to demersal trawl gears (Michalsen et al., 1996; Casey and Myers, 1998; Benoît and Swain, 2003). Different species of fish behave differently ahead of the trawl gear. Some are herded into the path of the net by the action of the otter doors and trawl sweeps on the seabed stirring up a sediment cloud (Ramm and Xiao, 1995; Bublitz, 1996; Somerton, 2004), others show net-avoidance behaviour (Main and Sangster, 1981), and variation in swimming endurance influences which fish fall back into the net (Winger et al., 1999, 2000). Factors that influence the catch efficiency and selective properties of trawl gears include sweep length (Engås and Godø, 1989), mesh size (Suuronen and Millar, 1992), net spread (Rose and Nunnallee, 1998; von Szalay and Somerton, 2005), trawl speed and duration (Ehrich and Stransky, 2001; Somerton and Weinberg, 2001; Weinberg et al., 2002), and the size and type of trawl groundgear (Engås and Godø, 1989; Walsh, 1992). Consequently, the catchability of particular species and sizes of fish varies between different fishing gears, dependent on the characteristics of the gear (ICES, 2004a), so trawl surveys provide gear-biased perceptions of the actual abundance of different species and size classes at a particular time and location. To estimate actual species densities, survey trawl estimates of catch density need to be converted to estimates of actual density by taking into account the catchability of the fish involved in the particular gear employed (Harley and Myers, 2001).
Several studies have attempted to estimate the total biomass of the North Sea fish community by scaling survey catch rate data to biomass estimates of the main commercial species determined from virtual population analysis (VPA) (Yang, 1982; Daan et al., 1990; Sparholt, 1990). Estimates of total fish abundance provided by these studies have been used in investigations of North Sea foodweb dynamics (Bryant et al., 1995; Greenstreet et al., 1997), but such studies only considered total biomass and total catch rates, making no attempt to consider the biomass and catch rates of different age and size classes of each species. Given the importance of size structure in aquatic foodwebs (Boudreau and Dickie, 1992; Duplisea and Kerr, 1995; Rice and Gislason, 1996; Duplisea et al., 1997; Jennings et al., 2002), and the importance placed on fish size in assessing the impact of fishing on fish communities in proposals for Ecological Quality Objectives for fish communities (ICES, 2001, 2006b), we attempted to address this shortcoming in the approach adopted here.
Here, ICES International Bottom Trawl Survey (IBTS) quarter 3 (Q3) data, Dutch Beam Trawl Survey (BTS) Q3 data, and ICES stock assessment data are analysed to model the catchability of each 1-cm size class of demersal fish caught in the Q3 IBTS. The catchability coefficients derived are then used to estimate the biomass of each species present in each ICES rectangle. Summing across rectangles, we present annual variation in the total biomass of the demersal fish community of the North Sea.
|
Methodology and analysis |
|---|
|
TopIntroductionMethodology and analysisDiscussionAppendixReferences |
|---|
Data sources
For the purposes of this study, only data covering the time period 1998–2004 were analysed. Before this, not all participating countries used the same Grande Overture Vertical (GOV) trawl gear, and tow durations varied. After 1998, trawl gear and duration was standardized across the entire survey, producing a spatially consistent dataset with which to estimate spatial variation in fish abundance. Two primary survey datasets were used in this analysis, the ICES IBTS and the Dutch BTS. The IBTS is an internationally coordinated survey covering the whole of the North Sea (Figure 1a). The BTS covers the southeastern, western, central eastern and part of the northern North Sea (Figure 1a). Both surveys take place at approximately the same time of year, in August/September.
|
Raw survey data were cleaned following procedures described by Daan (2001). Numbers-at-length of all species considered to be part of the demersal fish community caught in each trawl sample were quantified (Appendix), together with information on location, distance towed, and area swept by the gear. Otoliths collected from the assessed species (haddock, Melanogrammus aeglefinus; whiting, Merlangius merlangus; cod, Gadus morhua; Norway pout, Trisopterus esmarkii; and plaice, Pleuronectes platessa) provide age/length keys used to convert numbers-at-length to numbers-at-age in each tow. Hauls with outlier tow distances, swept areas, or duration were excluded from the analysis.
To estimate the catchability of the GOV, information on the abundance of fish was required. For commercial species, estimates of total numbers of fish-at-age in the North Sea were obtained from the stock assessments carried out by the Working Group on the Assessment of Demersal Stocks in the North Sea and Skagerrak (WGNSSK) (ICES, 2005). In the case of cod, the VPA assessment from the 2004 WGNSSK was used (ICES, 2004b), because in 2005, cod were classified as an observation stock with the consequence that an update of the VPA was not considered appropriate (ICES, 2005).
Determining fish density from trawl samples
To calculate fish density at each trawl location, estimates of the area sampled by the fishing gear were required. For the BTS, no information on gear geometry was necessary because the gear used in the survey had a fixed width of 8 m. For the IBTS, such information, i.e. door- and wing-spread, were recorded in the IBTS database, but these fields were not always completed. To estimate values for the missing data, a regression analysis was performed on 5 years of Scottish net geometry data so that mean door- (SG) and wing-spread (SW) could be estimated from the depth (D) recorded at the station (SG = 33.25 log D + 15.74, n = 290, p < 0.001; and SW = 6.85 log D + 5.89, n = 290, p < 0.001, respectively). The average distance towed (across all recorded tow distances) was used where this value was not recorded (there were no significant differences in tow distance between vessels). Once these gaps in the database had been filled, two measures of the area swept by the trawl gear were determined; the area swept between the otter doors, and the area swept between the wings of the net.
Knowing the area swept, two density estimates could be determined for each trawl sample. The first, based on the area swept by the entire gear, was considered most appropriate for two species, haddock and whiting, which are herded into the path of the net by the sediment cloud stirred up by the otter doors and sweeps (Main and Sangster, 1981; Wardle, 1983; Sangster and Breen, 1998; Breen et al., 2004). The second, based on the area swept by the net alone, was considered to be the most appropriate for the other demersal fish species. Use of area swept by the net alone for all species would have resulted in catchability coefficients >1 for haddock and whiting. The geometric mean density of fish calculated over all trawls carried out in each rectangle was used to estimate the density of fish in each ICES rectangle.
Development of catchability models
The catchability coefficient, q, of a fishing gear determines what fraction of the actual number of a given category of fish (e.g. of species s and size class l, or species s and age class a) in the path of the trawl (Ns,l or Ns,a) is caught (Cs,l or Cs,a), e.g. King (1995). Therefore,
|
| (1) |
Catchability coefficients for the species assessed
With no knowledge of the actual numbers of fish present at each trawl location, it is not possible to estimate q for each trawl sample individually. However, average q can be estimated across the whole North Sea if the numbers of fish in the North Sea estimated from the IBTS trawl survey data can be compared with estimates of total abundance in the area. For example, the abundance estimates provided by the ICES VPA stock assessment process (e.g. ICES, 2005) for the species assessed. Within the demersal fish community, and considering only those species frequently found in IBTS samples, these included haddock, whiting, cod, Norway pout, and plaice.
Determining the numbers of fish in the North Sea based on the IBTS
The number of fish of each assessed species (s) and age class (a) in each rectangle covered by the Q3 IBTS was estimated by multiplying the rectangle IBTS density estimates by the area of each rectangle. Density estimates in rectangles not surveyed in 1 year within the normal IBTS area (Figure 1a) were estimated by interpolation. Survey-based estimates of the numbers of each category of fish (Nsurv,s,a) were then derived by summing the individual estimates across all rectangles in the IBTS area.
Correcting for differences in area covered by the IBTS and stock assessments
ICES Area IV covers the whole of the North Sea, but not all of Area IV was covered by the IBTS (Figure 1a). Raising factors were used to multiply estimates of the numbers of fish determined for the area covered by the IBTS to make these estimates equivalent to the numbers expected had all of Area IV been surveyed. To take account of the fact that fish were not evenly distributed across the North Sea, raising factors were determined and applied independently for each of five subareas (Figure 1b).
The stock assessments made for the five demersal species regularly caught in the IBTS survey cover different geographic areas. To compare estimates of species abundance-at-age derived from the IBTS and stock assessments, the proportion that ICES Area IV made up of the total area included in each stock assessment was required (Table 1). The VPA estimates of numbers-at-age of each species were multiplied by these correction factors to provide equivalent abundance estimates for ICES Area IV alone (Table 2). This assumed that the density of fish was, on average, the same in ICES Area IV as in the other ICES Areas that made up each stock assessment area. This assumption was tested by examining landings data to check whether estimates of the numbers of fish in ICES Area IV, based on the assessments, would have altered radically had the stocks been allocated between ICES Area IV and outside ICES Area IV pro rata on the basis of landings rather than area. The two methods produced essentially similar allocations.
|
|
Correcting for differences in the timing of the IBTS and stock assessments
The stock assessments provide estimates of the numbers-at-age of the assessed species on 1 January each year (ICES, 2005). The Q3 IBTS data were collected from July to October. Over this intervening period, mortality would have reduced the numbers of fish present, so to make appropriate comparisons, this reduction in the number of fish present in the sea needs to be taken into account. Estimates of both fishing (F) and natural mortality (M) for each age class of each of the assessed species are provided by the assessment reports (ICES, 2004b, 2005). Therefore, the number of fish (by species s and age a) present in Q3 (NQ3,s,a) can be calculated from the numbers present on 1 January (NJan1,s,a) and the combined mortality rate Zs,a (where Zs,a = Fs,a + Ms,a). These mortality rates are calculated for the year, so Z must be multiplied by Y, the length of the intervening period as a proportion of a year (assuming an average of 9 months between 1 January and the time of the Q3 surveys, Y = 0.75), i.e.
|
| (2) |
Numbers of 0-group haddock are available in the stock assessment as of 1 July. The above equation was applied to the numbers of 0-group haddock using Y = 0.25 to give the number of 0-group haddock at the end of Q3. The stock assessment for Norway pout was available for Q3, so no mortality was applied to the estimates of numbers-at-age for this species.
Comparing IBTS and assessment ICES Area IV numbers-at-age to estimate q
Having corrected both the IBTS- and assessment-derived estimates of the numbers of fish (by species s and age a) in the same area and at the same time (ICES Area IV and Q3) these two values can be substituted into Equation (1), and rearranged to solve for q, i.e.
|
| (3) |
Catchability (q)-at-length
As the survey data are length-based, estimates of q-at-age of each assessed species were converted to q-at-length through application of the appropriate age/length key. Relationships fitted to these data then allowed q to be estimated for each 1-cm length class of each species (Figure 2). For haddock, a polynomial relationship was used to calculate q-at-length (Appendix). For Norway pout and plaice, q for young/small fish differed markedly from that for older/larger fish, so no trend in q with increase in length could be determined for the smaller fish of these two species. All Norway pout 
11 cm were therefore assigned the same value of q, equal to the mean for all 0-group Norway pout, and all plaice 17 cm were assigned the same value of q, equal to the mean value for 1-group plaice. For older plaice, q decreased linearly with length, providing a linear regression equation with which to estimate q for any plaice >17 cm. No such trend was apparent for Norway pout, so all pout 
12 cm were assigned a value of q equal to the mean for all Norway pout aged 1+. For cod, q increased with length-at-age and a polynomial relationship was used to estimate q for any length of cod. For all age classes of whiting, q decreased with increasing length, and a linear relationship provided the best fit to the data. The stock assessments provide no information on the abundance of 0-group whiting, so it was not possible to estimate q directly for whiting <21 cm. All whiting 21 cm were assigned a q of 0.62. All values of q-at-length are summarized in the Appendix.
|
Catchability coefficients for non-assessed species
Catch ratio
The IBTS Q3 survey caught 95 demersal species over the period 1998–2004. To calculate the total biomass of all demersal fish in the North Sea, or in any particular rectangle, catchabilities for the non-assessed species are also required. To establish q-at-length for non-assessed species, catch ratios between the GOV and the 8-m beam trawl (8BT) were examined. To find GOV and 8BT trawls that were carried out at approximately the same time and place, each statistical rectangle was divided into nine mini-rectangles of equal size. GOV and 8BT hauls located within the same mini-statistical rectangle in the same year were considered paired hauls. All fish were assigned to 5-cm length classes and the density of each species (s) at each length class (l) in both the GOV (FGOV,s,l) and 8BT (F8BT,s,l) samples were determined. Catch ratios (RC,s,l) were then calculated:
|
| (4) |
In some cases, a species/length-class combination was represented in either the GOV or the 8BT, but not in the other gear, i.e. a positive/zero or a zero/positive paired haul. Catch ratios could only be used to estimate q for any given species and length-class combination where positive/positive "paired haul" data were available, i.e. where both the GOV and the 8BT hauls contained fish of the same species and length class. Mean catch ratios for each species and length class were determined by summing the estimates for the individual paired-haul samples, and dividing by H, the total number of positive/positive paired hauls where the focal species and length-class combination was sampled by both gears:
|
| (5) |
Table 3 gives the mean catch ratio (RC) for the five assessed species and 24 other species for which RC data were available.
|
For 21 of the non-assessed species and length-class combinations where valid mean catch ratios could be determined, catch ratios were generally negative, i.e. fish densities in the beam trawl were greater than those in the GOV, and where positive, never exceeded 0.45. Two of the assessed species, plaice and cod, also had negative catch ratios across their entire length ranges (Table 3). Catchability (qs,l) of each length class (l) of each non-assessed species (s) with valid mean catch ratio data (RC,s,l) could then be determined knowing the mean cod and plaice catch ratios (RC,PLA,l and RC,COD,l) and catchabilities (qPLA,l and qCOD,l) at the same length class using the equation
|
| (6) |
The resulting catchability estimates are given in Table 4.
|
Best-fit curves were plotted through these estimates of q to explore the relationships between q and length for each species and to allow q to be estimated for every 1-cm length class of each species. For many species, sufficient data were available to allow curve fitting. In all instances, a polynomial function provided the best fit. To estimate q for each 1-cm length class for these species, the polynomial expression was used for the length range over which RC data were available. Where it was necessary to extrapolate beyond the data range, a constant q equal to the q obtained at the extreme of the data range was assumed. Figure 3 shows examples of the application of this approach to estimating q at each 1-cm length for common dab (Limanda limanda) and grey gurnard (Eutrigla gurnardus). This approach was adopted for all species in Table 3 where three or more RC-derived q values were available. For species where only two RC values were available to provide estimates of q in a 5-cm length class, an alternative approach was adopted. If the RC trends between the two length classes for which data were available for the non-assessed species in question were in the same direction as the trends between the same length classes for both plaice and cod, then a linear relationship between q and length between the two length ranges with RC data was assumed. If extrapolation beyond these length classes was necessary, q was again assumed to remain constant and equal to the value obtained at the extreme ranges of the data (e.g. bullrout, Myoxocephalus scopius, in Figure 3). If, on the other hand, the RC trends between the two length ranges for the non-assessed species in question differed from the trends observed for either cod or plaice, then a constant q, equal to the average of the two values of q derived from the application of Equation (6) to the RC data, across all 1-cm length classes of the non-assessed species was assumed (e.g. solenette, Buglossidium luteum, in Figure 3).
|
There were three non-assessed species with positive catch ratios over the majority of size classes for which there were data available; poor cod (Trisopterus minutus), bib (Trisopterus luscus), and saithe (Pollachius virens). These three species were treated differently and are discussed later.
Odds ratio
For a number of species and size-class combinations that featured in the mini-rectangle paired haul dataset, only zero/positive or positive/zero paired haul data were available. Under these circumstances, it was not possible to determine valid catch ratios, so their catchability in the GOV trawl could not be estimated using the approach above. However, even in that situation, odds ratios could be calculated (Fleiss, 1981; Tollit et al., 1997; Holland et al., 2005). For all species (s) and 5-cm length classes (l) for which any mini-statistical-rectangle data were available, odds ratios (RO) were calculated:
|
| (7) |
Logarithms to the base 10 of the odds ratio (log RO) were taken so that the positive values indicate species/length-class combinations of fish taken preferentially by the GOV and negative values indicate species/length-class combinations taken preferentially by the 8BT. The numerical value of the log-transformed index indicates orders of magnitude differences in the relative catchability of the two gears. Log odds ratio (log RO) could only be calculated when both I and U were positive. If I or U = 0 for a particular species/length-class combination, then the data were excluded, because log RO could not be calculated on either zero or infinity values.
For each of the 21 non-assessed species where q for 5-cm length classes had been determined from catch ratio data (Table 4), q was significantly related to log RO (Figure 4). This relationship was then used to estimate q for an additional 15 species and 5-cm length classes for which values of log odds ratio could be calculated, but for which no catch ratio data were available. Figure 5, using examples for four of these species, demonstrates how catchability-at-length coefficients were estimated for these 15 additional non-assessed species, following a procedure similar to that applied to the catch ratio data in Figure 3. Relationships were fitted in the cases of species with three or more data points, with constant values applied outside their data ranges (e.g. spotted dragonet, Callionymus maculatus, and turbot, Psetta maxima; Figure 5). Where only two data points were available, the average q was used (e.g. ling, Molva molva, and thickback sole, Microchirus variegatus; Figure 5).
|
|
Remaining non-assessed species
Examination of the catch-ratio-at-length signatures of poor cod suggests that, unlike Norway pout, they became increasingly better caught by the 8BT, compared with the GOV, as they increased in length (Table 3). As poor cod have high positive catch ratios at smaller size, similar to those of Norway pout (Table 3), the first half of Equation (6) was used to calculate the catchability of poor cod using the catch ratios and q-at-length of Norway pout instead of plaice.
Saithe were not caught in sufficient numbers in the Q3 GOV survey to be treated in the same way as the other assessed species. Moreover, only large saithe were caught and catch ratio data were therefore only available for the larger 5-cm size classes (Table 3). At that size, there was no obvious correspondence between the variation in catch ratios with length for saithe and the other gadoid species. Consequently, saithe q-at-length was given the average q-at-length of the three assessed gadoids; haddock, whiting, and cod. Similarly, there was no obvious correspondence between the variation in catch ratios with length for bib and Norway pout and poor cod, so the average q-at-length of Norway pout and poor cod was attributed to bib.
Application of the processes described earlier provided estimates of q for each 1-cm length class of 44 species, including the five assessed species. However, it was not possible to determine either valid catch ratios or odds ratios for many of the rarest species sampled in the IBTS survey. The 39 non-assessed species for which q was calculated were divided into four groups based on their body morphology; roundfish, flatfish, elasmobranchs (skates and rays), and elasmobranchs (dogfish). The average q-at-length of fish in each group was then determined. The 51 demersal species for which catch ratio or odds ratio data were not available were then assigned to the same four groups and assumed to have the group average q-at-length for each 1-cm size class. This final step provided an estimate of q for each 1-cm length class of every demersal fish species sampled by the GOV. For all species and length classes, the steps by which q was estimated are summarized in the Appendix.
To estimate the biomass of each species, numbers-at-length were converted to biomass-at-length using published (Coull et al., 1989) and unpublished length–weight relationships (FRS unpublished data; IMARES unpublished data).
Estimate of total North Sea demersal fish biomass
Accounting for species- and size-related catchability in the GOV trawl density estimates resulted, on average, in a 4.55–6.5 increase in the ensuing estimates of total North Sea demersal fish biomass (Figure 6). When catchability was taken into account, biomass varied between 3.8 million tonnes in 2004 and 7.5 million tonnes in 1999, compared with estimates of 837 000 t in 2004 and 1.4 million tonnes in 2000 determined using the raw GOV density data. The difference between the two trend-lines also varied between years, giving rise to slightly different patterns of annual variation in demersal fish biomass. The 1999 biomass estimate based on density data raised to account for catchability was a factor of 6.8 higher than the biomass estimate based on the raw GOV density data, whereas in 2002, the two estimates differed by only a factor of 4.1.
|
Variation in the extent to which accounting for catchability affected biomass estimates was primarily attributable to changes in the size structure of the demersal fish assemblage and the different catchability coefficients assigned to fish of different body length. In this case, the greater difference between the two biomass estimates in 1999 was caused by the influx of the strong 0-group haddock cohort, clearly illustrated in Figure 7. Figure 7 shows biomass estimate trends for five assessed species and the five most abundant non-assessed species. The effect of species-related differences in catchability is reflected in the difference between the pairs of trend-lines. For most species, the trend-lines co-vary closely, but this is not the case for haddock. In 1999 and 2000, haddock biomass estimates based on catchability-corrected density were factors of 22 and 6, respectively, greater than the biomass estimate derived from the raw density data. In all other years, the difference between the two estimates varied by a factor of 3.6–4.9. In 1999, haddock biomass based on the catchability-corrected density data peaked at
3 million tonnes, accounting for the increase in the estimated biomass of the whole demersal fish assemblage. The large numbers of 0-group haddock present in 1999, combined with the low catchability of these fish in the GOV (Figure 2), explain the exceptionally large biomass of haddock that year, and the effects of taking account of catchability on the perception of the size structure of the haddock population were particularly marked (Figure 8).
|
|
|
Discussion |
|---|
|
TopIntroductionMethodology and analysisDiscussionAppendixReferences |
|---|
The need to consider catchability
Estimates of catchability were generally low, suggesting that raw trawl survey density data seriously underestimated actual densities, so estimates of population biomass derived from such data will be too low. Some global North Sea foodweb models, such as multispecies virtual population assessment (MSVPA) (ICES, 2006a), avoid such issues by using estimates of predator and prey abundance derived directly from the VPA. However, they exclude most species recorded in the North Sea, so cannot be considered to represent comprehensive North Sea foodwebs. Other North Sea scale foodweb studies (e.g. Greenstreet et al., 1997; Heath, 2005; Jennings et al., 2007) have attempted to overcome the problem by using survey data raised to account for species-specific variation in catchability (e.g. Yang, 1982; Sparholt, 1990). Applying such catchability corrections suggested total North Sea demersal fish biomass estimates of
6 million tonnes, comparable with our estimates. However, our analyses also show that catchability varies with fish length, a factor ignored in the earlier studies. In marine foodwebs, predators eat smaller prey (Pope et al., 1994; Scharf et al., 2000; Floeter and Temming, 2003), so taking account of size-related variation in catchability is critical if the relative abundances of large and small individuals are to be assessed adequately and the trophodynamics of marine foodwebs appropriately modelled (e.g. Hall et al., 2006; Pope et al., 2006). Other foodweb studies, particularly those that address subscales of the North Sea, are based directly on survey data (e.g. Greenstreet et al., 1998; Greenstreet, 2006), and these will certainly be subject to error introduced by the unknown catchabilities of each species in the survey gear employed. Marine protected areas (MPAs) may be used to safeguard areas of high species diversity requiring studies of spatial variation in diversity to provide the appropriate advice. In areas dominated by species of small body size such as the northwestern North Sea where Norway pout are dominant (Greenstreet and Hall, 1996; Greenstreet et al., 1999), or in areas dominated by species with low catchability, raw survey data will tend to downplay the true level of dominance, resulting in inflated estimates of species diversity. Failure to take account of catchability in the survey gear in such studies may provide inappropriate advice leading to the incorrect siting of MPAs.
Limitations and assumptions of the methodology
To estimate catchability in the survey trawl, we need an estimate of the actual numbers of fish in the path of the gear. We presume that, for the five assessed species, dividing stock assessment estimates of the numbers of fish (at age or length) in the North Sea by the area of the North Sea provides such estimates. We therefore determine q-at-length for these species by simply applying Equation (3) to the IBTS- and assessment-derived North Sea density estimates. Application of these q values to the IBTS survey densities and raising to the area of the North Sea should therefore replicate annual variation in the stock assessments of the five assessed species; in general, this was the case. Occasional higher-than-expected IBTS-derived biomass estimates were invariably associated with one or two exceptionally large single trawl samples of these species. These generated density estimates which, when raised by the area of the ICES rectangle concerned, produced exceptionally high biomass estimates in those rectangles. In such circumstances, not even the use of the geometric mean was sufficient to control the resulting bias.
We also used three separate methods to estimate catchability-at-length for each of the 90 non-assessed demersal species sampled by the Q3 IBTS. Application of these catchability coefficients to the IBTS density data provided estimates of the biomass of each non-assessed species in the North Sea that were equivalent to the biomass estimates of the five assessed species provided by the stock assessments. In fact, as the catch ratio and odds ratio analyses used to derive catchability coefficients for the non-assessed species rely on the cod and plaice biomass estimates, any changes in the assessment methodology would affect not only the assessed species, but also the catchability coefficients and subsequent biomass estimates of all the non-assessed species. For example, the values of both natural (M) and fishing mortality (F) used critically affect the estimates of biomass produced by the VPA. Should mortality rates used in the VPA be underestimated, the immediate consequence would be that the VPA estimate of biomass will also be an underestimate. Under such circumstances, our catchability coefficients would be revised down, so all our estimates of biomass for the non-assessed species would be higher. Conversely, should mortality rates be overestimated in the assessments, the stock biomass estimates would be too high, requiring our catchability coefficients to be increased, and our biomass estimates for the non-assessed species would be lower.
Currently there is considerable uncertainty with respect to the real values of both M and F. Values of F used in the stock assessments could be too low if levels of discarding and misreported landings are seriously underestimated. Discard levels in cod, haddock, and whiting assessments have been monitored and used in the VPA for many years, but monitoring of plaice discarding is relatively recent, and estimates of discard mortality have only been incorporated into the stock assessment in the past few years. At present, sensitivity analysis of plaice discard mortality estimates has still to be carried out (ICES, 2007). M estimates derived from recent single-species stock assessments use the last key run of the MSVPA model for cod, haddock, whiting, and Norway pout (ICES, 2006a). Parameter values in the MSVPA are themselves derived from diet studies carried out in 1991 (Hislop et al., 1997). These data are now seriously outdated. Populations of both predators and prey have changed substantially in the intervening period, so the potential for current estimates of natural mortality to be erroneous is certainly present.
At present, the VPA for plaice assumes a constant M of 0.1 across all ages. Such low levels of natural mortality are very unlikely among smaller, younger fish. The consequences of this are that our estimates of q for the smaller size classes of plaice are likely to be too high. Use of lower values of qPLA,l for the smaller size classes (l) in Equation (6) would affect all results of qs,l for species where catchability-at-length was derived using the catch ratio analysis. As these values of q were used to parameterize the odds ratio q function (Figure 4), this would also be affected. Finally, the averaging method also relies on the catch ratio and odds ratio estimates of q, so revision of the smaller plaice q-at-length would affect the estimates of all non-assessed species over the length range involved. To investigate what effect a higher M for plaice would have on the catchability coefficients of all non-assessed species, the numbers of 1-group plaice from the VPA were multiplied by 1.5 and the number of 2-group plaice by 1.25, and all values of q-at-length were recalculated. Because q-at-length for cod remained unchanged, when the new values of q for plaice were used in Equation (6), the resulting q-at-length for the non-assessed species was reduced, but only by a small amount. This reduction in q would mean that the resulting total North Sea biomass estimate would be 5% higher. Therefore, although our actual estimates of q-at-length for each of the non-assessed species are clearly dependent on the parameter values used in the current stock assessments, any revision to these values resulting in new estimates of cod and plaice biomass-at-length, could easily be incorporated using our methods to produce revised catchability coefficients for all non-assessed species.
To estimate q for non-assessed species in the GOV, we used catch ratio and odds ratio analysis to compare catches in the 8BT and the GOV. In interpreting the results of these analyses, we make an important assumption. We assume that variation in the catch ratios and odds ratio is driven entirely by variation in catchability in the GOV rather than in the 8BT. For now, we see no alternative to this approach and recognize this as a potential weakness of the method employed. However, for many of the species considered, small-bodied roundfish and flatfish with strong benthic behavioural characteristics, it seems reasonable to assume that such fish would be much better sampled by the 8BT, an assumption strongly supported by the number of negative catch ratios, frequently of high negative numerical value (Table 3).
No 0-group abundance estimates for whiting, cod, or plaice were provided by the stock assessments. Estimates of catchability of the smaller size classes of these species were therefore based on the application of specific rules, so were prone to greater uncertainty. For example, given that the catchability of both 0-group haddock and Norway pout is lower than their 1-group conspecifics, it is entirely possible that 0-group whiting have a lower catchability than 1-group whiting, but no data were available that directly supported this. In the absence of such data, it was deemed more prudent to assume that 0-group whiting had similar catchability to the smallest size class of whiting for which data were directly available. There is some suggestion that 0-group whiting are indeed more effectively sampled by the GOV trawl than 0-group haddock (D. G. Reid, FRS Marine Laboratory, unpublished data).
Several species of demersal fish were recorded in the BTS, but do not appear in the IBTS dataset (Appendix). These fish, clearly part of the North Sea demersal assemblage, are small in size and benthic in habit, and were simply not sampled by the GOV. Application of catchability raising factors will therefore affect studies of species diversity (i.e. evenness, where it is the distribution of individuals between species that is of concern). However, they cannot influence studies of species richness. Some species will always be unsampled in the GOV-based IBTS, and application of any raising factor to a zero catch will still give a zero result.
During the IBTS, some countries change groundgear depending on the part of the North Sea in which they are fishing. Likewise, the BTS has equipped the standard 8BT with a flip-up rope when covering the central and northern North Sea. Currently, the effect of changing groundgear on estimates of catchability has not been examined. However, some studies have suggested that catchabilities of both roundfish and flatfish may be affected by groundgear choice (Main and Sangster, 1981; Ehrich, 1987; Engås and Godø, 1989; Walsh, 1992). This may be an issue deserving further consideration if changes in groundgear are frequent in the survey, or if the groundgear used changes in future.
Comparisons with previous estimates of catchability
Both Yang (1982) and Sparholt (1990) considered the total catchability of standard species, relating total survey catch rates to total biomass estimates derived from VPA or MSVPA. This approach has obvious problems given that catchability in the GOV was always likely to be size-dependent (Harley and Myers, 2001; Benoît and Swain, 2003), and because there is clear evidence of changes in the size structure of the North Sea demersal fish assemblage, both within and between species (Jennings et al., 1999; Olsen et al., 2005; Greenstreet and Rogers, 2006). The results of this study show that the catchability in the GOV is indeed strongly dependent on fish size, smaller size classes of many species tending to be sampled less well than their larger conspecifics. However, many of the catchability-at-length relationships tended to be unimodal in shape whereby, with further increase in length beyond a certain critical length, the largest fish of many species, including haddock, whiting, and plaice, were increasingly poorly sampled. Such unimodal relationships may well be characteristic of groundfish surveys with their relatively short tow duration. The largest fish of many species may well be able to swim fast enough, and for long enough, to avoid falling back into the net and being caught. The capacity to avoid capture in this way would be expected to increase with increasing fish length (Wardle, 1989; Winger et al., 1999, 2000).
Both Yang (1982) and Sparholt (1990) defined several fish groups on the basis of body form and habit. Each of these groups was headed up by one or more of the assessed species, for which survey catch rates could be related to VPA biomass estimates, to act as the group standard species. Catchability in the GOV of all species in each group was then assumed to be equal to the catchability of the group standard species. All flatfish (except sole, which were placed in a group of their own) were placed in the same group, headed up by plaice as the standard species. In making these assignments, both Yang and Sparholt assumed that all flatfish had similar catchability in the GOV to that of plaice. However, catchabilities of flatfish in the GOV vary markedly, with several species, e.g. common dab, being sampled well (ICES, 2004a). Our results confirm this; GOV catchability coefficients of 20 cm common dab were around 0.9, a value close to one and an order of magnitude greater than the coefficient of 0.09 for plaice of similar size. In fact, plaice tended to have a lower catchability-at-length than most other flatfish of similar size. Our results suggest, therefore, that the biomass estimates provided for many flatfish species by Yang (1982) and Sparholt (1990) were in fact overestimates, perhaps by as much as an order of magnitude. The earlier studies suggested that the common dab was one of the most abundant species in the North Sea. Our results would now refute this. This has a considerable implication for some of the published foodweb studies that used the earlier estimates of biomass. For example, Greenstreet et al. (1997) had considerable difficulty reconciling consumption by their demersal benthivore fish group with production by their benthic invertebrate prey groups. Our data would now tend to suggest that their demersal benthivore consumption rates had been overestimated, so greatly compensating for the discrepancy between prey consumption and production.
Ehrich et al. (in ICES, 2004a) compared the catchability of fish in several different gears in the North Sea. Their analysis showed that species such as small gobies (Pomatoschistus minutus), small flatfish species (Buglossidium luteum, Arnoglossus laterna), dragonets (Callionymus lyra), and some larger flatfish species such as plaice had very low catchabilities in the GOV, agreeing with our results here. Gadoids and gurnards also had catchabilities close to one, similar to our estimates. Meta-analysis carried out by Harley and Myers (2001) used trawl survey datasets from North America, Europe, and New Zealand to look at length-specific catchabilities for a number of different species groups. They found that catchability was greater for haddock than for cod, implying behavioural differences between the two species even though they share many morphological similarities. They also found that the catchability of plaice was <0.1 in the North Sea. Their results again agree closely with ours.
This study specifically calculated q-at-length for demersal fish in Q3. Therefore differences in catchability attributable to seasonal variation in the availability of a species to the GOV have not been examined. Previous researchers have looked at diurnal (Michalsen et al., 1996; Casey and Myers, 1998; Korsbrekke and Nakken, 1999; Benoît and Swain, 2003) changes in catchability, as well as seasonal differences (Harley and Myers, 2001). The last authors found that catchability in research surveys was greater in summer and autumn than in winter and spring. However, even such a seasonal affect could be linked to diurnal variation if, during winter, a greater proportion of survey trawl samples was taken during the hours of darkness. These differences should be considered when using trawl data from other quarters of the year, suggesting the need for further work in this area.
The science required to underpin the implementation of an ecosystem approach to management clearly needs to expand into areas beyond the traditional expertise of fisheries scientists of the past (Heath, 2005). For this to happen, scientists need to access an increasing breadth of different types of information. The importance of groundfish surveys in this new regime is bound to increase far beyond the scope of their original remit. It is therefore critical that as marine scientists we interpret correctly the data that these surveys provide: understanding catchability is therefore a critical step.
|
Appendix |
|---|
|
TopIntroductionMethodology and analysisDiscussionAppendixReferences |
|---|
Table showing all fish species caught in both the Q3 IBTS 1998–2004 and the 8-m BTS from the same period. The table shows whether a species was considered demersal or pelagic. Only the species labelled demersal were included in the analysis. The table indicates how the catchability of each of the 95 demersal species caught in the GOV was derived, whether it was derived using the VPA assessment (assessed), using the catch ratio information from the mini-statistical rectangles (catch ratio), using the odds ratio (odds ratio), as the average q-at-length of either the roundfish, flatfish, skates and rays, or dogfish groups (average), or whether it was derived some other way (other, see text). The relationships used to calculate the catchability-at-length for each of the 95 demersal fish species caught in the Q3 IBTS, 1998–2004, are also given.
|
|
Acknowledgements |
|---|
This work was supported by the Scottish Executive Environment and Rural Affairs Department and funded by the European Commission "Managing Fisheries to Conserve Groundfish and Benthic Invertebrate Species Diversity (MAFCONS EC project Q5RS-2002-00856). We thank the staff at ICES for providing the IBTS data, Paul Fernandes, Coby Needle, and other members of the Sea Fisheries Group at FRS, Rob Fryer, Colin Miller, and two anonymous reviewers for providing helpful comments that greatly improved the manuscript.
|
References |
|---|
|
TopIntroductionMethodology and analysisDiscussionAppendixReferences |
|---|
-
Benoît H. P., Swain D. P. Accounting for length- and depth-dependent diel variation in catchability of fish and invertebrates in an annual bottom-trawl survey. ICES Journal of Marine Science (2003) 60:1298–1317.
Boudreau P. R., Dickie L. M. Biomass spectra of aquatic ecosystems in relation to fisheries yield. Canadian Journal of Aquatic Sciences (1992) 49:1528–1538.
Breen M., Dyson J., O’Neil F. G., Jones E. G., Haigh M. Swimming endurance of haddock (Melanogrammus aeglefinus L.) at prolonged and sustained swimming speeds, and its role in their capture by towed fishing gears. ICES Journal of Marine Science (2004) 61:1071–1079.
Bryant A. D., Heath M. R., Broekhuizen N., Ollason J. G., Gurney W. S. C., Greenstreet S. P. R. Modelling the predation, growth and population dynamics of fish within a spatially-resolved shelf-sea ecosystem model. Netherlands Journal of Sea Research (1995) 33:407–421.
Bublitz C. G. Quantitative evaluation of flatfish behaviour during capture by trawl gear. Fisheries Research (1996) 25:293–304.[CrossRef][Web of Science]
Casey J. M., Myers R. A. Diel variation in trawl catchability: is it as clear as day and night? Canadian Journal of Fisheries and Aquatic Sciences (1998) 55:2329–2340.
Coull K. A., Jermyn A. S., Newton A. W., Henderson G. I., Hall W. B. Length/weight relationships for 88 species of fish encountered in the North East Atlantic. Scottish Fisheries Research Report (1989) 43:81.
Daan N. The IBTS database: a plea for quality control. ICES Document CM 2001/T: 03 (2001).
Daan N., Bromley P. J., Hislop J. R. G., Nielsen N. A. Ecology of North Sea fish. Netherlands Journal of Sea Research (1990) 26:343–386.
Duplisea D. E., Kerr S. R. Application of a biomass size spectrum model to demersal fish data from the Scotian shelf. Journal of Theoretical Biology (1995) 177:263–269.[CrossRef][Web of Science]
Duplisea D. E., Kerr S. R., Dickie L. M. Demersal fish biomass size spectra on the Scotian Shelf, Canada: species replacement at the shelfwide scale. Canadian Journal of Fisheries and Aquatic Sciences (1997) 54:1725–1735.
Ehrich S. The proportion of young cod escaping under the G.O.V.-trawl rigged with heavy bobbin footrope. ICES Document CM 1987/B: 28. (1987) 8.
Ehrich S., Stransky C. The influence of towing time, catch size and catch treatment on species diversity estimates from groundfish surveys. Archive of Fishery and Marine Research (2001) 49:37–44.[Web of Science]
Engås A., Godø O. R. The effect of different sweep lengths on the length composition of bottom-sampling trawl catches. Journal du Conseil International pour l’Exploration de la Mer (1989) 45:263–268.
Fleiss J. L. Statistical Methods for Rates and Proportions. (1981) London: John Wiley and Sons.
Floeter J., Temming A. Explaining diet composition of North Sea cod (Gadus morhua): prey size preference vs. prey availability. Canadian Journal of Fisheries and Aquatic Sciences (2003) 60:140–150.
Frid C. L. J., Paramor O. A. L., Scott C. L. Ecosystem based fisheries management: progress in the NE Atlantic. Marine Policy (2005) 29:461–469.[CrossRef][Web of Science]
Furness R. W. Management implications of interactions between fisheries and sandeel-dependent seabirds and seals in the North Sea. ICES Journal of Marine Science (2002) 59:261–269.
Greenstreet S. P. R. Does the prohibition of industrial fishing for sandeels have any impact on local gadoid populations? In: Management of Marine Ecosystems, Zoological Society of London Conservation Biology Series 12—Boyd I. L., Wanless S., Camphuysen C. J, eds. (2006) Cambridge, UK: Cambridge University Press. 223–235.
Greenstreet S. P. R., Bryant A. D., Broekhuizen N., Hall S. J., Heath M. H. Seasonal variation in the consumption of food by fish in the North Sea and implications for food web dynamics. ICES Journal of Marine Science (1997) 54:243–266.
Greenstreet S. P. R., Hall S. J. Fishing and the ground-fish assemblage structure in the north-western North Sea: an analysis of long-term and spatial trends. Journal of Animal Ecology (1996) 65:577–598.[CrossRef][Web of Science]
Greenstreet S. P. R., McMillan J. A., Armstrong F. Seasonal variation in the importance of pelagic fish in the diet of piscivorous fish in the Moray Firth, NE Scotland: a response to variation in prey abundance? ICES Journal of Marine Science (1998) 55:121–133.
Greenstreet S. P. R., Rogers S. I. Indicators of the health of the North Sea fish community: identifying reference levels for an ecosystem approach to management. ICES Journal of Marine Science (2006) 63:573–593.
Greenstreet S. P. R., Spence F. E., McMillan J. A. Fishing effects in northeast Atlantic shelf seas: patterns in fishing effort, diversity and community structure. 5. Changes in structure of the North Sea groundfish species assemblage between 1925 and 1996. Fisheries Research (1999) 40:153–183.[CrossRef][Web of Science]
Hall S. J., Collie J. S., Duplisea D. E., Jennings S., Bravington M., Link J. A length-based multispecies model for evaluation community responses to fishing. Canadian Journal of Fisheries and Aquatic Sciences (2006) 63:1344–1359.
Hall S. J., Mainprize B. Towards ecosystem-based fisheries management. Fish and Fisheries (2004) 5:1–20.[CrossRef][Web of Science]
Harley S. J., Myers R. Hierarchical Bayesian models of length-specific catchability of research trawl surveys. Canadian Journal of Fisheries and Aquatic Sciences (2001) 58:1569–1584.
Heath M. R. Regional variability in the trophic requirements of shelf sea fisheries in the Northeast Atlantic, 1973–2000. ICES Journal of Marine Science (2005) 62:1233–1244.
Hislop J., Bromley P. J., Daan N., Gislason H., Heesen H. J. L., Robb A. P., Skagen D., et al. Database report of the stomach sampling project, 1991. ICES Cooperative Research Report (1997) 219:421.
Hislop J. R. G., Robb A. P., Bell M. A., Armstrong D. W. The diet and food consumption of whiting (Merlangius merlangus) in the North Sea. ICES Journal of Marine Science (1991) 48:139–156.
Holland G. J., Greenstreet S. P. R., Gibb I. M., Fraser H. M., Robertson M. R. Identifying sandeel Ammodytes marinus sediment habitat preferences in the marine environment. Marine Ecology Progress Series (2005) 303:269–282.[CrossRef][Web of Science]
ICES. Report of the ICES Advisory Committee on Ecosystems. ICES Cooperative Research Report (2001) 249:75.
ICES. Report of the Working Group on Fish Ecology, 2–7 April 2004, ICES Headquarters, Copenhagen. ICES Document CM 2004/G: 09. (2004) a:249.
ICES. Working Group on the Assessment of Demersal Stocks in the North Sea and Skagerrak (WGNSSK), 7–16 September 2004, Bergen, Norway. ICES Document CM 2005/ACFM: 07. (2004) b:783.
ICES. Working Group on the Assessment of Demersal Stocks in the North Sea and Skagerrak (WGNSSK), 6–15 September 2005, ICES Headquarters, Copenhagen. ICES Document CM 2006/ACFM: 09. (2005) 958.
ICES. Report of the Study Group on Multispecies Assessments in the North Sea (SGMSNS). 20–25 February 2006, ICES Headquarters, Copenhagen. ICES Document CM 2006/RMC: 02. (2006) a:75.
ICES. Report of the ICES Advisory Committee on Fishery Management, Advisory Committee on the Marine Environment and Advisory Committee on Ecosystems, 2006. ICES Advice, Books 1–10 (2006) b6:310.
ICES. Working Group on the Assessment of Demersal Stocks in the North Sea and Skagerrak (WGNSSK), 5–14 September 2006, ICES Headquarters. ICES Document CM 2007/ACFM: 35. (2007) 1160.
Jennings S., De Oliveira J. A. A., Warr K. J. Measurement of body size and abundance tests of macroecological and food web theory. Journal of Animal Ecology (2007) 76:72–82.[CrossRef][Web of Science][Medline]
Jennings S., Greenstreet S. P. R., Reynolds J. D. Structural change in an exploited fish community: a consequence of differential fishing efforts on species with contrasting life histories. Journal of Animal Ecology (1999) 68:617–627.[CrossRef][Web of Science]
Jennings S., Pinnegar J. K., Polunin N. V. C., Warr K. J. Linking size-based and trophic analyses of benthic community structure. Marine Ecology Progress Series (2002) 226:77–85.[CrossRef][Web of Science]
King S. Fisheries Biology, Assessment and Management. (1995) Oxford: Fishing News Books. 338.
Korsbrekke K., Nakken O. Length and species dependent diurnal variation of catch rates in the Norwegian Barents Sea bottom trawl surveys. ICES Journal of Marine Science (1999) 56:284–291.
Main J., Sangster G. I. A study of the fish capture process in a bottom trawl by direct observations from a towed underwater vehicle. Scottish Fisheries Report (1981) 23.
Michalsen K., Godø O. R., Fernö A. Diel variation in the catchability of gadoids and its influence on the reliability of abundance indices. ICES Journal of Marine Science (1996) 53:389–395.
Olsen E. M., Lilly G. R., Heino M., Morgan M. J., Bratten J., Dieckmann U. Assessing changes in age and size at maturation in collapsing populations of Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences (2005) 62:811–823.
Pope J. G., Rice J. C., Daan N., Jennings S., Gislason H. Modelling an exploited marine fish community with 15 parameters – results from a simple sized based model. ICES Journal of Marine Science (2006) 63:1029–1044.
Pope J. G., Shepherd J. G., Webb J. Successful surf-riding on size-spectra: the secret of survival in the sea. Philosophical Transactions of the Royal Society of London, Series B (1994) 343:41–49.
Ramm D. C., Xiao Y. Herding in groundfish and effective pathwidth of trawls. Fisheries Research (1995) 24:243–259.[CrossRef][Web of Science]
Rice J., Gislason H. Patterns of change in the size spectra of numbers and diversity of the North Sea fish assemblage, as reflected in surveys and models. ICES Journal of Marine Science (1996) 53:1214–1225.
Rijnsdorp A. D., van Leeuwen P. I., Daan N., Heessen H. J. L. Changes in abundance of demersal fish species in the North Sea between 1906–1909 and 1990–1995. ICES Journal of Marine Science (1996) 53:1054–1062.
Rose C. S., Nunnallee E. P. A study of changes in groundfish trawl catching efficiency due to differences in operation width, and measures to reduce width variation. Fisheries Research (1998) 36:139–147.[CrossRef][Web of Science]
Sangster G. I., Breen M. Gear performance and catch comparison trials between a single trawl and a twin rigged gear. Fisheries Research (1998) 36:15–26.[CrossRef][Web of Science]
Scharf F. S., Juanes F., Rountree R. A. Predator size-prey size relationships of marine fish relationships of marine fish predators: interspecific variation and effects on ontogeny and body size on trophic niche breadth. Marine Ecology Progress Series (2000) 208:229–248.[CrossRef][Web of Science]
Somerton D. A. Do Pacific cod (Gadus macrocephalus) and walleye pollock (Theregra chalcogramma) lack a herding response to the doors, bridles, and mudclouds of survey trawls? ICES Journal of Marine Science (2004) 61:1186–1189.
Somerton D. A., Weinberg K. L. The affect of speed through the water on footrope contact of a survey trawl. Fisheries Research (2001) 53:17–24.[CrossRef][Web of Science]
Sparholt H. An estimate of the total biomass of fish in the North Sea. Journal du Conseil International pour l’Exploration de la Mer (1990) 46:200–210.
Suuronen P., Millar R. B. Size selectivity of diamond and square mesh codends in pelagic herring trawls: only small herrings will notice the difference. Canadian Journal of Fisheries and Aquatic Sciences (1992) 49:2104–2117.
von Szalay P. G., Somerton D. A. The effect of net spread on the capture efficiency of a demersal survey trawl used in the eastern Bering Sea. Fisheries Research (2005) 74:86–95.[CrossRef][Web of Science]
Tollit D. J., Greenstreet S. P. R., Thompson P. M. Prey selection by harbour seals, Phoca vitulina, in relation to variations in prey abundance. Canadian Journal of Zoology (1997) 75:1508–1518.
Walsh S. J. Size-dependent selection at the footgear of a groundfish survey trawl. North American Journal of Fisheries Management (1992) 12:625–633.[CrossRef]
Wanless S., Harris M. P., Greenstreet S. P. R. Summer sandeel consumption by seabirds breeding in the Firth of Forth, south-east Scotland. ICES Journal of Marine Science (1998) 55:1141–1151.
Wardle C. S. Fish reactions to towed fishing gears. In: Experimental Biology at Sea—MacDonald A., Priede I. G., eds. (1983) New York: Academic Press. 167–195.
Wardle C. S. Understanding fish behaviour can lead to more selective fishing gears. In: Proceedings of the World Symposium on Fishing Gear and Fishing Vessel Design.—Fox S. G., Huntington J., eds. (1989) Newfoundland, Canada: Marine Institute, St Johns.
Weinberg K. L., Somerton D. A., Munro P. T. The effect of trawl speed on the footrope capture efficiency of a survey trawl. Fisheries Research (2002) 58:303–313.[CrossRef][Web of Science]
Winger P. D., He P., Walsh S. J. Swimming endurance of American plaice (Hippoglossoides platessoides) and its role in fish capture. ICES Journal of Marine Science (1999) 56:252–265.
Winger P. D., He P., Walsh S. J. Factors affecting the swimming endurance and catchability of Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences (2000) 57:1200–1207.
Yang J. An estimate of the fish biomass in the North Sea. Journal du Conseil International pour l’Exploration de la Mer (1982) 40:161–172.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  G. J. Piet, R. van Hal, and S. P. R. Greenstreet Modelling the direct impact of bottom trawling on the North Sea fish community to derive estimates of fishing mortality for non-target fish species ICES J. Mar. Sci., October 1, 2009; 66(9): 1985 - 1998. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. M. Fraser, S. P. R. Greenstreet, and G. J. Piet Selecting MPAs to conserve groundfish biodiversity: the consequences of failing to account for catchability in survey trawls ICES J. Mar. Sci., January 1, 2009; 66(1): 82 - 89. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. P. R. Greenstreet, H. M. Fraser, and G. J. Piet Using MPAs to address regional-scale ecological objectives in the North Sea: modelling the effects of fishing effort displacement ICES J. Mar. Sci., January 1, 2009; 66(1): 90 - 100. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. P. R. Greenstreet Biodiversity of North Sea fish: why do the politicians care but marine scientists appear oblivious to this issue? ICES J. Mar. Sci., November 1, 2008; 65(8): 1515 - 1519. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. M. Fraser, S. P. R. Greenstreet, R. J. Fryer, and G. J. Piet Mapping spatial variation in demersal fish species diversity and composition in the North Sea: accounting for species- and size-related catchability in survey trawls ICES J. Mar. Sci., May 1, 2008; 65(4): 531 - 538. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||