Effects of increased mesh size on catch and fishing power of coral reef fish traps

Fisheries Research 39 (1999) 275±294

Effects of increased mesh size on catch and ®shing power of coral reef ®sh traps David Robichauda,b,1,*, Wayne Huntea,b,c, Hazel A. Oxenfordb,c a

Department of Biology, McGill University, 1205 Ave. Docteur Pen®eld, MontreÂal, Que., Canada H3A 1B1 b Bellairs Research Institute of McGill University, Holetown, St. James, Barbados c Marine Resource and Environmental Management Program, University of the West Indies, St. Michael, Barbados Received 11 June 1997; accepted 5 August 1998

Abstract Effects of increased mesh size on catch and ®shing power of coral reef ®sh traps (Antillean design) were investigated in the Barbados west coast trap ®shery by experimental ®shing with commercial traps (maximum aperture 4.1 cm) and large mesh traps (maximum aperture 5.5 cm). Large mesh traps caught 53±63% less ®sh by number and 51% less by weight than the commercial traps. The ®sh in the commercial traps were signi®cantly smaller by length, body depth and weight than those in large mesh traps, and a signi®cantly higher percentage were immature. The effect of mesh size on ®shing power of traps was investigated by comparing the catch rates of ®sh large enough to be retained by both trap types (i.e., body depths >5.5 cm; termed adjusted catch). The adjusted catch of large mesh traps was 24±35% lower by number and about 30% lower by weight than that of commercial traps, indicating that the ®shing power of large mesh traps is substantially lower than that of commercial traps. The squeezability hypothesis and the visual image hypothesis were tested as explanations for the reduced ®shing power of large mesh traps by comparing catches of commercial traps, large mesh traps and experimental traps (where experimental traps were designed to have a similar visual image as commercial traps but similar ®sh retention capacity as large mesh traps). The higher ®shing power of commercial traps is generated primarily by a difference in catch rates in the 5.5±6.0 cm body depth size class; i.e., the size class which might feasibly squeeze through the 5.5 cm maximum aperture of large mesh traps. This strongly supports the squeezability hypothesis as an explanation for the higher ®shing power of small mesh traps. We could ®nd no de®nitive evidence indicating that reduced visual image of traps, whether created by structural differences (trap construction) or biotic differences (number of ®sh already in a trap), decreases ingress rates to traps and hence explains the lower ®shing power of large mesh traps. These results will facilitate the incorporation of the reduced ®shing power effect into yield per recruit models which can be used to predict catch rate changes in ®sheries in which the minimum trap mesh size is increased. # 1999 Elsevier Science B.V. All rights reserved. Keywords: Coral reef ®sh; Barbados; Antillean ®sh traps; Fishing power; Mesh size selectivity; Squeezability; Visual image

*Corresponding author. Tel.: +1-709-778-0349; fax: +1-709778-0655; e-mail: [email protected] 1 Present address: c/o Fisheries Conservation Chair, Fisheries and Marine Institute, Memorial University of Newfoundland, PO Box 4920, St. John's, NF, Canada A1C 5R3.

1. Introduction Coral reefs support many ®sh species that are commercially important to island countries through-

0165-7836/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved. PII: S0165-7836(98)00198-2

276

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

out the world (Roberts and Polunin, 1993; McClanahan and Kaunda-Arara, 1996). Trawl net and hookand-line ®shing are dif®cult in coral reef areas and demersal traps are therefore the predominant gear used to harvest reef ®sh (Dammann, 1980). Since reef ®sh resources are accessible and inexpensive to harvest, and since the ®sheries are largely unregulated, reef ®sh have been over®shed in most countries (Appeldoorn et al., 1987). Over®shing has been attributed to excess ®shing effort (i.e., too many traps per reef area) and to small trap mesh sizes capturing small and immature ®sh (Stevenson, 1978; Lee and Al-Baz, 1989; Sary et al., 1991; Beliaeff et al., 1992), leading to a loss of potential yield through both growth and recruitment over®shing (Bohnsack et al., 1989; Rosario and Sadovy, 1991; Hunte and Mahon, 1994). Fishery managers are therefore considering an increase in the legal minimum mesh size (Aiken and Haughton, 1987; Burnett-Herkes et al., 1988; McConney, 1996), a management tool that has also been proposed for trawl (Smolowitz, 1983), gillnet (Ehrhardt and Die, 1988) and trapnet (Meyer and Merriner, 1976) ®sheries. This is an attractive management option for coral reef ®sh because alternate management tools, such as area closures and reduced ®shing effort, may be more dif®cult to enforce (Mahon and Drayton, 1990). A major constraint to imposing larger minimum mesh sizes in coral reef trap ®sheries is the expectation that catch rates will decline sharply in the short term, since some portion of ®sh retained by traps of smaller mesh will be too small to be retained by traps of larger mesh. Munro (1983) discusses assumptions that the size of ®sh caught in a trap, speci®cally body-depth, is a direct function of trap mesh size. This effect on catch has been reported in several studies in which traps of different mesh sizes have been simultaneously ®shed (e.g. Olsen et al., 1978; Stevenson and Stuart-Sharkey, 1980; Ward, 1988; Bohnsack et al., 1989), but note the indication in the last two studies that traps of very small mesh size may have lower catch rates than traps of intermediate mesh. The magnitude of the short-term catch rate reduction and hence the probability of obtaining the ®shers' cooperation with the transition to larger mesh, may vary with the degree of overexploitation of the reef ®sh resource. Since one effect of over-exploitation is a decrease in the mean ®sh size in the population, the short-term reduction in catch

rates accompanying increased mesh size may be greater for more over-exploited populations, such as the reef ®sh stocks in Barbados (Mahon and Drayton, 1990). One objective of this study is to document the short-term changes in catch that may be expected with an increase in trap mesh size in a heavily overexploited ®shery such as the Barbados west coast trap ®shery. A second potential constraint to imposing larger minimum mesh sizes is that there have been few studies of the proportion of immature ®sh retained by traps (but see Wolf and Chislett, 1974; Stevenson, 1978; Rosario and Sadovy, 1991), and the proportion of immature ®sh which can be harvested without reducing future yield through recruitment over®shing is unknown (Hunte and Mahon, 1994). The contention that transition to a larger mesh will result in long-term bene®ts by reducing the proportion of immature ®sh retained is therefore questionable. A second objective of this study is to document the changes in percentage of immature ®sh retained that might be expected with an increase in trap mesh size in the Barbados west coast trap ®shery. Catch rates of large mesh traps are expected to increase in the long term following a transition to a larger mesh in the ®shery, not only through reduced recruitment over®shing, but also because the mean ®sh size in the population should increase to an equilibrium with the new mesh size and growth over®shing may be reduced (see Hunte and Mahon (1994) for a review; Sary et al. (1997) for indications of higher catch rates of large mesh traps three years after their introduction to a Jamaican coral reef ®shery). These last two effects on catch rates following transition to a larger mesh in the ®shery can be modeled using yield per recruit analyses (see Hunte and Mahon (1994) for review). This would be useful to resource managers contemplating the imposition of larger mesh sizes, but the ability to predict the rate of increase will be constrained if the ®shing power of large mesh traps is lower than that of small mesh traps, where ``®shing power'' refers to the relative catch rate of traps on ®sh large enough to be retained by either mesh (see Section 2). Preliminary results from previous studies suggest that the ®shing power of large mesh traps may be lower than that of small mesh traps (e.g. Hartsuijker and Nicholson, 1981; Munro, 1983; Burnett-Herkes et al., 1988; Moran and Jenke, 1990). A third objective

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

of this study is to investigate whether the ®shing power of traps can be expected to change with changes in mesh size in coral reef trap ®sheries. Two hypotheses have been proposed to explain the apparent reduction in ®shing power of large mesh traps in coral reef ®sheries, but neither has been quantitatively tested. Separating these hypotheses is important in resource management since it will explicitly determine how reduced ®shing power with increased mesh size should be incorporated into yield per recruit models. The ®rst hypothesis, called the squeezability hypothesis (Hartsuijker, 1982), proposes that ®sh whose body depths are slightly larger than the maximum aperture of the mesh can squeeze through it, and that this could cause the ®shing power of large mesh traps to be lower than that of small mesh traps. For example, in the present study, ®sh with body depths slightly larger than 5.5 cm (the maximum aperture of the large mesh traps) would be retained by the commercial traps, but might squeeze through the mesh of large mesh traps, thereby ensuring that the catch of large mesh traps on ®sh with body depths equal to or slightly greater than 5.5 cm is lower than that of commercial traps on these same size classes. One way to test the squeezability hypothesis is to compare, for commercial (small mesh) and large mesh traps, the catch rate of different size classes of ®sh with body depths larger than 5.5 cm. The prediction is that, if squeezability is an important factor, the greatest difference in catch rates should occur in size classes immediately larger than 5.5 cm, with the catch rate difference becoming negligible at larger ®sh sizes. A fourth objective of this study is to compare the catch rates of commercial and large mesh traps on ®sh of different size classes, thereby assessing whether squeezability can explain reductions in ®shing power for large mesh traps. A second hypothesis for reductions in ®shing power of large mesh traps, proposed by several authors (e.g. Ward, 1986, 1987; Bohnsack et al., 1989; Moran and Jenke, 1990; Newman and McB Williams, 1995), can be called the visual image hypothesis. It assumes that ®sh are attracted to solid structures on the reef (Samples and Sproul, 1985; Bohnsack, 1989) out of curiosity, for food, for sleeping sites or for refuge from predators (Randall, 1963; Stone et al., 1979; Matthews, 1985; Hixon and Beets, 1989). Small mesh

277

traps offer a more solid visual image since they have more wire per unit area of trap surface. The hypothesis therefore suggests that small mesh traps may have higher ®shing power than large mesh traps because ingress rates are higher. If this hypothesis is correct, any large mesh trap designed to have a similar visual image to a commercial (small mesh) trap should have a catch rate similar to the commercial trap on ®sh large enough to be retained by both mesh sizes. Visual image can be affected by factors other than structural trap design. For example, early aggregation of small ®sh in commercial traps may increase the visual image of the trap, thereby resulting in higher subsequent ingress rates for ®sh of all sizes than that which occurs in large mesh traps. We refer to such factors, potentially affecting the visual image of traps, as biotic factors; and to aspects of trap design that could affect the visual image of traps as structural factors. The ®nal objective of this study is to test the visual image hypothesis by investigating whether increased visibility, through either structural (large mesh vs. small mesh) or biotic (the number of ®sh already in a trap) changes, results in increased ingress rates and hence increased ®shing power of coral reef ®sh traps. 2. Materials and methods 2.1. Field Study 1 The Antillean arrowhead traps with a single, horseneck funnel used in this study are described by Munro et al. (1971). Each trap measured 2 ft (0.61 m) in height, 5 ft (1.52 m) in width, and 4 ft (1.22 m) in length, giving a volume of 40 ft3 (1.13 m3). The traps consisted of hexagon shaped wire mesh made of galvanized steel, framed by wooden sticks nailed and strapped together with wire. This is the most common trap design used in the Barbados commercial ®shery. The wire mesh of the traps was of the same gauge but different mesh size. Half of the traps were made with the mesh size typically used in the Barbados commercial ®shery (maximum aperture 4.1 cm), and are referred to as ``commercial traps''. The rest were made of mesh of maximum aperture 5.5 cm, and are referred to as ``large mesh traps''. All experimental ®shing was conducted on three fringing reefs on the west coast of Barbados, namely

278

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

Heron Bay, South Bellairs and North Bellairs reefs. All traps were placed on the seaward edge of the spurand-groove zone of the reef, on sandy patches, no more than 2 m away from a coral spur (see Lewis (1960) for a description of the zonation on Barbados fringing reefs). The ®shing was conducted in May of 1986, 1990, 1991 and 1996. On each ®shing occasion, commercial traps and large mesh traps were ®shed in pairs, i.e., a commercial trap was set relatively close to a large mesh trap. The rationale was to control for effects of variation in the availability of ®sh at different trap sites on catch rate differences between trap types. Soak time (the time over which the trap was allowed to ®sh) was identical for all trap pairs. It was typically two days, but varied from 1 to 4. A total of 92 soaks was conducted during the study, 46 for each of the trap types. On the last day of each soak, the trap pairs were hauled and the taxon, fork length (cm), maximum body depth (cm) and weight (g) of each ®sh caught were recorded for each trap. Weights were not measured directly in 1996, but were estimated using log weight versus log fork length relationships for each species (Munro, 1983; Bohnsack and Harper, 1988; Robichaud, unpublished data). Fork lengths were also used to estimate the percentage of immature ®sh in a trap, based on length at sexual maturity data in Cummings (1968); Robertson and Warner (1978, 1978); Munro (1983); Hunte and Mahon (1994), for the more common species in the catch. 2.2. Field Study 2 Twelve ®sh traps were used. Four were commercial traps and four were large mesh traps similar to those described for Field Study 1. The additional four traps are subsequently called ``experimental traps''. They were identical in size and similar in design to the commercial traps, except that a single 3 ft2 ft (91.5 cm61 cm) panel was replaced by the mesh of maximum aperture 5.5 cm used in the large mesh traps. The panel replaced was perpendicular to the sea bed when ®shing, thus maximizing the probability that a ®sh in the trap would swim through it. The rationale for the trap design was to produce a trap with a similar visual image to the commercial traps, but with a retention capacity similar to the large mesh traps. Twelve permanent trap sites were established, four in the spur-and-groove zone of the South Bellairs

fringing reef and eight in the spur-and-groove zone of the North Bellairs fringing reef. All trap sites were in sandy grooves, no more than 1 m away from a coral spur and at least 27 m away from the next nearest trap site (i.e., there were non-overlapping circles of at least 573 m2 around each trap). The ®shing was conducted between February and June 1996. Traps of the three designs were randomly assigned to the twelve sites. Soaks were ®ve days long, and were started on the day of each quarter moon. During each of the ®rst four days of the soak, at approximately the same time each day, all traps were visited, the number of ®sh of each species in the trap was recorded, and the length of each ®sh estimated. On the last day of each soak, the traps were hauled and the taxon, total length (cm) and maximum body depth (cm) of each ®sh caught were recorded for each trap. The ®sh were then returned live to the sea. Repeatability (Becker, 1992) of total length and body depth measurements were 0.990.0012 and 0.940.012, respectively. Total lengths were used to estimate the percentage of immature ®sh in a trap for the more common species (see data sources listed in Section 2.1). After one complete lunar cycle (four soaks), the traps were randomly moved between trap sites such that at the end of the study, every site was host to each trap type in a random order for the duration of a lunar cycle. A total of 144 soaks was conducted, 48 for each of the trap types. To improve the accuracy of underwater ®sh length estimation, the observers were trained for two weeks prior to the start of the study. The observers estimated the lengths of ®sh in traps. The traps were then hauled, the lengths of all ®sh recorded, and the observers' estimates compared to the measured lengths. This process was repeated until the mean estimation error for all ®sh on a given day was less than 10%. Body depths of ®sh in the traps prior to hauling were not estimated underwater, but were calculated using species-speci®c allometric relationships between body depth and total length (Robichaud, 1996). Using data on the size of each ®sh of each species in each trap on subsequent days, daily ingress was calculated as the number of ®sh known to have entered the trap during the previous 24 h period, irrespective of whether the ®sh subsequently escaped, died in the trap, or remained captive for the remainder of the soak (Munro, 1974). Ingress calculated in this way does not

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

account for ®sh which enter and escape during the interval between observations, or for ®sh which have been previously observed but which escape or are consumed and then replaced by a conspeci®c of equal size. The daily ingress estimated in this way is therefore an underestimate of the number of ®sh which have actually entered the trap (Munro, 1974). The problems associated with estimating daily ingress accurately were somewhat alleviated in this study by noting ®sh body markings to better identify individuals. Furthermore, precise individual identi®cation was possible for 14.4% of ®sh caught because they were tagged with numerically coded anchor tags. 2.3. Data handling and analysis Fishing power is de®ned as the relative vulnerability of a group of ®sh to the gear types in question (Ricker, 1975):

arcsine(root p) transformations; Sokal and Rohlf, 1995). The Wilcoxon two-group test, the non-parametric analog of the t-test, was therefore used throughout. The Wilcoxon two-group test was also used to test the squeezability hypothesis by comparing the number of ®sh in each 0.5 cm body depth size class for each trap type. The same test was used to compare catch on the fourth day of soak and daily ingress for each trap type in Field Study 2. In testing the visual image hypothesis, effects of the three trap designs on catch (as the number of ®sh per trap haul) were compared using two-way ANOVAs with site and trap design as the independent variables and catch as the dependent variable. When signi®cant differences were detected, pairwise comparisons were conducted using Tukey's HSD tests. Although the catch data deviated somewhat from the assumptions of normality and homogeneity of variance, ANOVAs

…catch rate in gear X†…stock fished by gear X†ÿ1 …catch rate in reference gear†…stock fished by reference gear†ÿ1 Since in this study there is only one stock of ®sh in the area of all trap types, the relative vulnerability reduces to ``relative catch rate''. The group of interest is that containing ®sh large enough to be retained by both mesh types (i.e., ®sh with a body depth greater than 5.5 cm). To compare the ®shing power of commercial and large mesh traps, all ®sh with body depths less than or equal to 5.5 cm were removed from the catch, since these ®sh would not be retained by the large mesh traps (maximum aperture 5.5 cm). The catch of traps excluding ®sh less than or equal to 5.5 cm body depth is subsequently referred to as ``adjusted catch'' or ®shing power. The parameters compared between commercial traps and large mesh traps were: length, body depth and weight of individual ®sh, catch rate and adjusted catch rate (as number of ®sh per trap and weight of ®sh per trap) and percentage of immature ®sh caught per trap. None of these parameters were normally distributed in either Field Study 1 or Field Study 2 (Shapiro±Wilk tests; p<0.001 for all parameters), and none could be normalized using standard transformation techniques (e.g. loge(x‡1), root(x‡0.5) and

279

:

were used because there is no non-parametric analog of the two-way ANOVA, and because ANOVAs are considered to be robust against deviations from these assumptions when sample sizes are large and equal (Maxwell and Delaney, 1990), as was the case in this study. Effects of trap design on daily ingress were investigated using a single factor ANOVA with daily ingress of ®sh large enough to be retained by either mesh size as the dependent variable and trap design as the independent. When signi®cant differences were detected, pairwise comparisons were conducted using Tukey's HSD tests. To assess whether the presence of ®sh in traps in¯uenced subsequent ingress, daily ingress of ®sh large enough to be retained by either mesh size was regressed against the number of ®sh (of all sizes) in the traps on the day previous to the observation period. In this case, the method described in Elliot (1983) was used to determine that x0.71 was the appropriate transformation to homogenize the variance. Since there was little difference between the results based on the transformed and non-transformed

280

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

data, the simpler, non-transformed results are presented here. 3. Results 3.1. Catch of traps During Field Study 1, a total of 592 ®sh from 40 species was caught in the 92 soaks conducted. Of these, 432 ®sh from 35 species were caught in the commercial traps, and 160 ®sh from 23 species in the large mesh traps. During Field Study 2, a total of 2148 ®sh from 41 species were caught in the 96 soaks conducted. Of these, 1461 ®sh from 38 species were caught in the commercial traps, and 687 ®sh from 28 species in the large mesh traps (Table 1). The number of ®sh caught per trap was 63.0% lower for large mesh traps than for commercial traps in Field Study 1 (Table 1; Zˆÿ4.26, p<0.0001) and 53.0% lower in Field Study 2 (Table 1; Zˆÿ4.69, p<0.0001). The weight of ®sh caught per trap was 51.2% lower for large mesh traps than in commercial traps (Table 1; Zˆÿ3.48, pˆ0.0005). In both ®eld studies, large mesh traps caught signi®cantly bigger ®sh than commercial traps. Length was 11.7% greater for individuals in large mesh traps than in commercial traps in Field Study 1 (Table 2; Zˆ3.73, pˆ0.0002) and 8.75% greater in Field Study 2 (Table 3; Zˆ8.30, p<0.0001). Body depth was 18.0% greater for individuals in large mesh traps than for those in commercial traps in Field Study 1 (Table 2; Zˆ10.56, p<0.0001) and 11.9% greater in Field Study 2 (Table 3; Zˆ15.40, p<0.0001). Weight was 47.2% greater for individuals in large mesh traps (134.8 g) than for those in commercial traps (91.6 g; Zˆ4.91, p<0.0001). Consistent with these size differences, the percentage of immature ®sh caught was 105% higher for commercial traps than for large mesh traps in Field Study 1 (Table 2; Zˆÿ2.25, pˆ0.025) and 28.2% higher in Field Study 2 (Table 3; Zˆÿ2.92, pˆ0.0036). The tendency of commercial traps to have higher catch rates, to catch smaller ®sh and to catch a higher percentage of immature ®sh was also evident in most cases when species were considered separately (see Tables 1, 2, 3). In both ®eld studies, most of the species caught by the commercial traps that were not caught by the large mesh traps were those that were relatively rare in the commercial traps (Table 1). This suggests that the

lower catch rates of the large mesh traps can largely explain the absence of these species from large mesh traps. The exceptions in Field Study 1 were french grunts (Haemulon ¯avolineatum), longspine squirrel®sh (Holocentrus rufus) and redband parrot®sh (Sparisoma aurofrenatum) which were common in commercial traps but were either not caught or caught in very small numbers in large mesh traps (Table 1). In Field Study 2, french grunts and redband parrot®sh were again notable exceptions (Table 1). The catch rate difference between trap types for these species may simply re¯ect the small body depth characteristic of these species. In both ®eld studies, only a few species (®ve in Field Study 1; three in Field Study 2) were caught by the large mesh traps and not by the commercial traps (Table 1), but the catch rates were negligible. In Field Study 1, only one individual of three of the ®ve species was captured, and in Field Study 2, no more than two individuals of each of the three species were captured. None of the species captured only by large mesh traps in Field Study 1 were the same as those in Field Study 2, con®rming that the data do not indicate any speciesspeci®c selectivity for these species by large mesh traps. In both ®eld studies, the relative abundance of the different species in the catch did not differ between trap types for those species taken by both trap types (relative abundance of species in commercial traps is signi®cantly correlated with relative abundance in large mesh traps; for Field Study 1: rsˆ0.65, pˆ 0.0038; for Field Study 2: rsˆ0.76, p<0.0001). 3.2. Fishing power of traps The 92 soaks conducted during Field Study 1 produced an adjusted catch (see Section 2) of 391 ®sh from 35 species. Of these, 237 ®sh from 29 species were in the adjusted catch of commercial traps, and 154 ®sh of 23 species were in the adjusted catch of the large mesh traps (Table 1). The 96 soaks conducted during Field Study 2 produced an adjusted catch of 1417 ®sh from 35 species. Of these, 808 ®sh from 30 species were in the adjusted catch of commercial traps, and 609 ®sh from 27 species were in the adjusted catch of the large mesh traps (Table 1). In both ®eld studies, the adjusted catch of commercial traps was signi®cantly lower than the total catch (Table 1; for Field Study 1: number of ®sh per trap:

Common name

Ocean surgeon Blue tang Banded butterflyfish Orangespotted filefish Yellowtail damselfish French grunt Stoplight parrotfish Blackbar soldierfish Rock beauty Redband parrotfish Princess parrotfish Queen parrotfish Coney Longspine squirrelfish Sergean major Smooth trunkfish Spanish hogfish Doctorfish Graysby Foureye butterflyfish Damselfish Web burrfish Striped parrotfish Ballonfish French angelfish Caesar grunt Mahogany snapper Redtail parrotfish Squirrelfish Redfin parrotfish Spotted moray Rock hind Spotted scorpionfish Smallmouth grunt Whitespotted filefish Bermuda chub Honeycomb cowfish

Species

Acanthurus bahianus Acanthurus coeruleus Chaetodon striatus Cantherhines pullus Microspathodon chrysurus Haemulon flavolineatum Sparisoma viride Myripristis jacobus Holacanthus tricolor Sparisoma aurofrentum Scarus taeniopterus Scarus vetulaa Epinephelus fulvus Holocentrus rufus Abudefduf saxatilis Lactophrys triqueter Bodianus rufusa Acanthurus chirurgus Epinephelus cruentatus Chaetodon capistratus Stegastes spp. Chilomycterus antillarum Scarus croicensis Diodon holocanthus Pomacanthus paru Haemulon carbonarium Lutjanus mahogoni Sparisoma chrysopterum Holocentrus adscensionis Sparisoma rubripinne Gymnothorax moringa Epinephelus adscensionis Scorpaena plumieri Haemulon chrysargyreum Cantherhines macrocerus Kyphosus sectatrix Lactophrys polygonia

93 (36) 39 (20) 30 (30) 17 (17) 10 (6) 55 (21) 4 (3) 32 (9) 15 (14) 31 (19) 30 (24) 3 (2) 10 (10) 11 (1) 1 (1) 3 (0) 1 (1) 1 (1) 1 (1) 6 (5) 6 (0) 2 (1) 5 (1) 0 (0) 2 (2) 0 (0) 1 (1) 0 (0) 6 (0) 0 (0) 5 (4) 1 (1) 1 (1) 2 (0) 3 (3) 0 (0) 1 (1)

5285.9 (2647.7) 2023.6 (1424.6) 1556.6 (1556.6) 1717.3 (1717.3) 621.7 (389.2) 4160.9 (1727.1) 1149.0 (578.9) 2382.2 (853) 1434.9 (1388.7) 3026.0 (2004.1) 5307.1 (4695) A (A) 2926.9 (2926.9) 795.8 (106.5) 78.0 (78) 73.2 (0) A (A) 58.8 (58.8) 127.7 (127.7) 189.0 (166.3) 216.9 (0) 238.0 (117.6) 515.4 (124.5) 0 (0) 447.3 (447.3) 0 (0) 72.3 (72.3) 0.0 (0) 914.7 (0) 0.0 (0) 4642.3 (3683.6) 1050.0 (1050) 394.3 (394.3) 108.3 (0) 1094.3 (1094.3) 0 (0) 278.6 (278.6)

34 (31) 19 (19) 5 (5) 27 (26) 2 (2) 0 (0) 4 (4) 9 (8) 10 (10) 2 (2) 5 (5) 6 (6) 6 (6) 0 (0) 0 (0) 2 (1) 0 (0) 0 (0) 3 (3) 12 (12) 0 (0) 1 (1) 0 (0) 3 (3) 0 (0) 4 (4) 2 (2) 0 (0) 0 (0) 1 (1) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 1 (1)

2468.3 (2315.8) 1291.9 (1291.9) 278.1 (278.1) 2918.0 (2841) 157.5 (157.5) 0 (0) 1185.9 (1185.9) 859.2 (778.9) 1010.9 (1010.9) 191.5 (191.5) 1234.2 (1234.2) 1610.2 (1610.2) 1818.5 (1818.5) 0 (0) 0 (0) 153.9 (111.7) 0 (0) 0 (0) 1337.0 (1337) 378.8 (378.8) 0 (0) 618.8 (618.8) 0 (0) 1101.4 (1101.4) 0 (0) 814.0 (814) 332.4 (332.4) 0 (0) 0 (0) 244.0 (244) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 290.0 (290)

727 (320) 245 (188) 76 (64) 41 (40) 63 (51) 44 (0) 35 (30) 25 (3) 22 (19) 16 (5) 8 (4) 13 (10) 12 (9) 14 (1) 15 (6) 12 (0) 14 (13) 13 (5) 11 (11) 0 (0) 8 (0) 6 (4) 7 (2) 1 (1) 4 (4) 3 (2) 2 (0) 5 (4) 0 (0) 3 (3) 0 (0) 3 (1) 2 (2) 1 (0) 0 (0) 2 (2) 0 (0)

Number

Number

Number

Weight (g)

Commercial traps

Large mesh traps

Commercial traps Weight (g)

Field Study 2

Field Study 1

293 (232) 124 (121) 58 (55) 48 (47) 30 (30) 0 (0) 30 (30) 1 (1) 20 (20) 0 (0) 1 (1) 13 (13) 6 (6) 4 (0) 8 (8) 6 (3) 7 (7) 5 (4) 4 (4) 0 (0) 0 (0) 4 (4) 0 (0) 6 (4) 4 (4) 2 (2) 4 (4) 3 (3) 0 (0) 1 (1) 0 (0) 0 (0) 1 (1) 0 (0) 0 (0) 0 (0) 0 (0)

Number

Large mesh traps

Table 1 Number (and weight in Field Study 1) of fish caught in commerical and large mesh traps during 46 soaks (average duration two days) of each trap type in the Field Study 1 and 48 five-day soaks of each trap type in Field Study 2

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294 281

Spotfin butterflyfish Spotted drum Spotted goatfish Yellow goatfish Bar jack Bluestriped grunt Bridled burrfish Cardinal soldierfish Glasseye snapper Greater soapfish Lane snapper Porcupinefish Sailors choice Schoolmaster

Chaetodon ocellatus Equetus punctatus Pseudupeneus maculatus Mulloidchthys martinicus Caranx ruber Haemulon sciurus Chilomycterus antennatus Plectrypops retrospinis Priacanthus cruentatus Rypticus saponaceus Lutjanus synagris Diodon hystrix Haemulon parra Lutjanus apodus

Adjusted catch (see text for explanation) is in parentheses. a This species has missing weight data in 1996.

All fish Mean per trap (all fish)

Common name

Species

Table 1 (Continued )

(0) (0) (0) (0) (0) (0) (0) (0) (0) (0) (1) (0) (0) (0)

432 (237) 9.4 (5.2)

0 0 2 0 0 0 0 1 0 0 1 0 0 0 43371.0 (29879) 942.8 (649.5)

0.0 (0) 0.0 (0) 242.8 (0) 0.0 (0) 0 (0) 0 (0) 0 (0) 71.2 (0) 0.0 (0) 0.0 (0) 170.1 (170.1) 0.0 (0) 0.0 (0) 0.0 (0)

(0) (0) (0) (0) (0) (1) (0) (0) (0) (0) (0) (1) (0) (0)

160 (154) 3.5 (3.3)

0 0 0 0 0 1 0 0 0 0 0 1 0 0 21164.5 (20812.5) 460.1 (452.4)

0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 584.3 (584.3) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 285.6 (285.6) 0 (0) 0 (0)

(2) (0) (0) (0) (0) (0) (1) (0) (1) (0) (0) (0) (0) (0) 1461 (808) 30.4 (16.8)

2 0 0 2 1 0 1 0 1 0 0 0 1 0

Number

Number

Number

Weight (g)

Commercial traps

Large mesh traps

Commercial traps Weight (g)

Field Study 2

Field Study 1

(0) (2) (0) (0) (0) (0) (0) (0) (0) (1) (0) (0) (0) (1) 687 (609) 14.3 (12.7)

0 2 0 0 0 0 0 0 0 1 0 0 0 1

Number

Large mesh traps

282 D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

283

Table 2 Mean fork length, mean body depth and mean percent of fish immature for the more common species caught in commercial and large mesh traps during 46 soaks (average duration two days) in Field Study 1 Species

Acanthurus bahianus Acanthurus coeruleus Chaetodon striatus Cantherhines pullus Microspathodon chrysurus Haemulon flavolineatum Sparisoma viride Myripristis jacobus Holacanthus tricolor Sparisoma aurofrenatum Scarus taeniopterus Scarus vetula Epinephelus fulvus Holocentrus rufus Abudefduf saxatilis Lactophrys triqueter Bodianus rufus Acanthurus chirurgus Epinephelus cruentatus Chaetodon capistratus Stegastes spp. Chilomycterus antillarum Scarus croicensis Diodon holocanthus Pomacanthus paru Haemulon carbonarium Lutjanus mahogoni Holocentrus adscensionis Sparisoma rubripinne Epinephelus adscensionis Scorpaena plumieri Haemulon chrysargyreum Cantherhines macrocerus Lactophrys polygonia Pseudupeneus maculatus Haemulon sciurus Plectrypops retrospinis Lutjanus synagris Diodon hystrix All fish

Common name

Ocean surgeon Blue tang Banded butterflyfish Orangespotted filefish Yellowtail damselfish French grunt Stoplight parrotfish Blackbar soldierfish Rock beauty Redband parrotfish Princess parrotfish Queen parrotfish Coney Longspine squirrelfish Sergeant major Smooth trunkfish Spanish hogfish Doctorfish Graysby Foureye butterflyfish Damselfish Web burrfish Striped parrotfish Balloonfish French angelfish Caesar grunt Mahogany snapper Squirrelfish Redfin parrotfish Rock hind Spotted scorpionfish Smallmouth grunt Whitespotted filefish Honeycomb cowfish Spotted goatfish Bluestriped grunt Cardinal soldierfish Lane snapper Porcupinefish

Commercial traps

Large mesh traps

Immature (%)

Mean FL (cm)

Mean BD (cm)

Immature (%)

Mean FL (cm)

Mean BD (cm)

1.1 71.8 0 0 10.0 0

13.5 (14.8) 11.9 (13.5) 11.9 (11.9) 17.2 (17.2) 12.2 (12.5) 15.5 (16) 23.4 (20.8) 13.8 (15.3) 14.6 (14.7) 17.9 (18.5) 21.1 (21.8) 21.7 (22.3) 26.9 (26.9) 16.1 (18.1) 13.1 (13.1) 8.6 (b) 26.0 (26) 13.8 (13.8) 20.0 (20) 8.8 (9.1) 10.7 (b) 12.8 (13.3) 17.9 (19.5) b b () 19.4 (19.4) b b () 16.8 (16.8) 16.9 (b) b b () 36.5 (36.5) 25.8 (25.8) 15.0 (b) 28.3 (28.3) 20.6 (20.6) 19.4 (b) b b () 11.9 (b) 23.0 (23) b b ()

5.5 (6.1) 6.1 (7) 7.2 (7.2) 7.8 (7.8) 5.8 (6.2) 5.4 (5.8) 6.5 (7.3) 5.2 (6) 6.8 (6.9) 5.8 (6) 6.4 (6.7) 6.4 (6.9) 8.1 (8.1) 4.8 (6) 7.5 (7.5) 3.7 (b) 8.4 (8.4) 6.3 (6.3) 5.8 (5.8) 6.1 (6.3) 4.8 (b) 4.8 (6.1) 5.3 (5.6) b b () 13.0 (13) b b () 7.3 (7.3) 5.0 (b) b b () 10.0 (10) 17.3 (17.3) 4.4 (b) 14.8 (14.8) 8.7 (8.7) 5.1 (b) b b () 4.5 (b) 6.8 (6.8) b b ()

0 47 0 7 0

14.9 13.6 12.2 17.4 13.5 b b () 24.4 15.1 15.3 20.2 24.2 28.3 26.6 b b () b b () 11.2 b b () b b () 29.3 10.1 b b () 26.9 b b () 21.1 b b () 22.1 22.6 b b () 23.9 b b () b b () b b () b b () 24.9 b b () 30.6 b b () b b () 19.5

(19.5)

6.2 (6.3) 6.9 (6.9) 7.3 (7.3) 7.9 (8) 6.2 (6.2) b b () 8.5 (8.5) 6.1 (6.1) 7.4 (7.4) 6.9 (6.9) 7.6 (7.6) 9.2 (9.2) 8.3 (8.3) b b () b b () 5.4 (6.3) b b () b b () 8.8 (8.8) 6.4 (6.4) b b () 7.1 (7.1) b b () 8.6 (8.6) b b () 7.0 (7) 7.1 (7.1) b b () 7.5 (7.5) b b () b b () b b () b b () 8.8 (8.8) b b () 11.0 (11) b b () b b () 7.0 (7)

15.4 (16.6)

6.1 (6.9)

17.2 (17.4)

7.2 (7.3)

a

31.3 0 0 6.7 a

0 0 a a a a a

0 a a a b a b a a b a a a a a a b a a b

8.4

b a

0 0 0 0 a

0 b b a b b a

0 b a b a b a a b a b b b b a b a b b a

4.1

(15) (13.6) (12.2) (17.5) (13.5) (24.4) (15.2) (15.3) (20.2) (24.2) (28.3) (26.6) (15.1) (29.3) (10.1) (26.9) (21.1) (22.1) (22.6) (23.9)

(24.9) (30.6)

The parameters for adjusted catch (see text for explanation) are in parentheses. Percent immature only calculated for the most common species in this catch. b None of this species were caught in this trap type during this Field Study. a

Zˆ2.75, pˆ0.0059; weight of ®sh per trap: Zˆ2.11, pˆ0.035; for Field Study 2: number of ®sh per trap: Zˆ4.12, p<0.0001), but total catch and adjusted catch for large mesh traps did not differ signi®cantly

(Table 1; for Field Study 1: number of ®sh per trap: Zˆ0.091, pˆ0.93; weight of ®sh per trap: Zˆ0.020, pˆ0.98; for Field Study 2: number of ®sh per trap: Zˆ0.41, pˆ0.68).

284

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

Table 3 Mean total length, mean body depth and mean percent of fish immature for the more common species caught in commercial and large mesh traps during 48 five-day soaks of each trap type in Field Study 2 Species

Common name

Commercial traps Immature (%)

Acanthurus bahianus Acanthurus coeruleus Chaetodon striatus Cantherhines pullus Microspathodon chrysurus Haemulon flavolineatum Sparisoma viride Myripristis jacobus Holacanthus tricolor Sparisoma aurofrenatum Scarus taeniopterus Scarus vetula Epinephelus fulvus Holocentrus rufus Abudefduf saxatilis Lactophrys triqueter Bodianus rufus Acanthurus chirurgus Epinephelus cruentatus Stegastes spp. Chilomycterus antillarum Scarus croicensis Diodon holocanthus Pomacanthus paru Haemulon carbonarium Lutjanus mahogoni Sparisoma chrysopterum Sparisoma rubripinne Epinephelus adscensionis Scorpaena plumieri Haemulon chrysargyreum Kyphosus sectatrix Chaetodon ocellatus Equetus punctatus Mulloidchthys martinicus Caranx ruber Chilomycterus antennatus Priacanthus cruentatus Rypticus saponaceus Haemulon parra Lutjanus apodus All fish

Ocean surgeon Blue tang Banded butterflyfish Orangespotted filefish Yellowtail damselfish French grunt Stoplight parrotfish Blackbar soldierfish Rock beauty Redband parrotfish Princess parrotfish Queen parrotfish Coney Longspine squirrelfish Sergeant major Smooth trunkfish Spanish hogfish Doctorfish Graysby Damselfish Web burrfish Striped parrotfish Ballonfish French angelfish Caesar grunt Mahogany snapper Redtail parrotfish Redfin parrotfish Rock hind Spotted scorpionfish Smallmouth grunt Bermuda chub Spot fin butterflyfish Spotted drum Yellow goatfish Bar jack Bridled burrfish Glasseye snapper Greater soapfish Sailors choice Schoolmaster

2.1 64.1 5.3 19.5 0 0 a

100 0 0 0 a

0 0

a a a

92.3 0 a a a a

100 33.3 a a a a a

100 a a b

0

a a a

b a b

20.0

Large mesh traps

Mean TL (cm)

Mean BD (cm)

15.7 14.2 11.6 16.8 15.8 17.3 22.5 15.9 16.3 18.5 20.8 26.6 26.6 22.0 13.4 10.8 26.4 14.1 27.0 11.7 17.0 18.8 21.0 13.8 20.6 19.1 22.5 26.8 29.3 23.5 18.3 29.4 15.0 b b () 26.5 23.0 22.0 19.0 b b () 17.8 b b ()

5.5 (5.9) 6.6 (7.1) 6.3 (6.5) 6.6 (6.6) 6.1 (6.3) 4.8 (b) 7.1 (7.4) 5.0 (6) 6.9 (7.1) 5.5 (6.3) 5.5 (6.2) 7.3 (7.8) 7.7 (8.5) 4.9 (5.8) 5.5 (6) 4.3 (b) 7.1 (7.3) 5.6 (6.3) 8.1 (8.1) 4.7 (b) 6.5 (7.2) 5.5 (6) 6.3 (6.3) 8.4 (8.4) 5.8 (6.5) 4.9 (b) 6.3 (6.5) 7.7 (7.7) 8.3 (14.5) 6.9 (6.9) 4.3 (b) 9.1 (9.1) 8.8 (8.8) b b () 5.4 (b) 5.5 (b) 6.5 (6.5) 5.8 (5.8) b b () 5.3 (b) b b ()

(16.7) (15.3) (11.8) (16.9) (16.2) (b) (23.6) (18.3) (17) (20.1) (22.6) (28.5) (29.1) (23.8) (14.4) (b) (26.8) (15.9) (27) (b) (18.8) (20) (21) (13.8) (22.9) (b) (23.3) (26.8) (47.5) (23.5) (b) (29.4) (15) b

() (b) (22) (19) (b)

16.0 (17.1)

The parameters for adjusted catch (see text for explanation) are in parentheses. Percent immature only calculated for the most common species in the catch. b None of this species were caught in this trap type during this Field Study. a

5.9 (6.5)

Immature (%) 0 52.4 0 6.3 0 b a

100 0 b

0 a

0 0 a a a

80.0 0 b a b a

100 0 a a a b a b b b a b b b b a b a

15.6

Mean TL (cm)

Mean BD (cm)

16.7 15.4 12.4 17.0 16.1 b b () 28.2 18.0 17.0 b b () 27.5 28.8 28.9 24.9 14.7 14.5 29.9 16.3 25.6 b b () 16.6 b b () 20.0 16.5 26.0 30.4 25.7 24.8 b b () 24.5 b b () b b () b b () 24.0 b b () b b () b b () b b () 26.5 b b () 40.0

5.9 (6) 7.1 (7.1) 6.6 (6.7) 6.5 (6.6) 6.3 (6.3) b b () 8.7 (8.7) 5.8 (5.8) 7.2 (7.2) b b () 8.3 (8.3) 8.1 (8.1) 8.5 (8.5) 5.4 (b) 6.2 (6.2) 5.8 (6.8) 7.9 (7.9) 6.6 (6.8) 7.3 (7.3) b b () 6.7 (6.7) b b () 6.8 (7.6) 10.3 (10.3) 7.4 (7.4) 8.4 (8.4) 7.5 (7.5) 7.3 (7.3) b b () 6.0 (6) b b () b b () b b () 7.4 (7.4) b b () b b () b b () b b () 8.0 (8) b b () 11.5 (11.5)

(17) (15.5) (12.5) (17) (16.1) (28.2) (18) (17) (27.5) (28.8) (28.9) (b) (14.7) (16.9) (29.9) (16.9) (25.6) (16.6) (21.3) (16.5) (26) (30.4) (25.7) (24.8) (24.5)

(24)

(26.5) (40)

17.4 (17.6)

6.6 (6.7)

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

The adjusted catch of large mesh traps, as indicated by the number of ®sh caught per trap, was 35.0% lower than that of commercial traps in Field Study 1 (Table 1; Zˆÿ2.00, pˆ0.046) and 24.6% lower in Field Study 2 (Table 1; Zˆÿ2.50, pˆ0.012). The adjusted catch of large mesh traps, as indicated by the weight of ®sh caught per trap, was lower than that of commercial traps in Field Study 1, but the difference was not statistically signi®cant (Table 1; Zˆ ÿ1.56, pˆ0.12). The results con®rm that the ®shing power of commercial traps is greater than that of large mesh traps. The tendency for adjusted catch rates to be higher for commercial traps than for large mesh traps was also evident in most cases when species were considered separately (Table 1). The body depth of ®sh in the adjusted large mesh catch was 5.8% greater than that in the adjusted commercial catch in Field Study 1 (Table 2; Zˆ4.62, p<0.0001) and 3.1% greater that in Field Study 2 (Table 3; Zˆ3.52, pˆ0.0004). The length of ®sh in the adjusted large mesh catch was signi®cantly longer (2.9%) than in the adjusted commercial catch in Field Study 2, but not in Field Study 1 (for Field Study 1: Table 2; Zˆ0.89, pˆ0.37; for Field Study 2: Table 3; Zˆ2.12, pˆ0.034). The weight of individual ®sh in the adjusted commercial catch (113.9 g) did not differ signi®cantly from that in the adjusted large mesh catch in Field Study 1 (137.8 g; Zˆ1.38, pˆ0.17). 3.3. The squeezability hypothesis In Field Study 1, the catch of commercial traps began to rise sharply in the 4.0±4.5 cm body depth size class, reaching a peak in the 5.5±6.0 cm size class and declining at larger size classes (Fig. 1(a)). In Field Study 2, the catch of commercial traps again began to rise in the 4.0±4.5 cm size class, reaching a peak in the 5.0±5.5 cm size class and declining at larger size classes (Fig. 1(b)). The decline in catch of ®sh in size classes larger than 6.0 cm presumably re¯ects the reduced abundance of ®sh in these larger size classes. In Field Study 1, the catch of large mesh traps rose sharply in the 5.5±6.0 cm body depth size class, reaching a peak in the 6.0±6.5 cm size class and declining at larger size classes (Fig. 1(a)). In Field Study 2, the catch of large mesh traps began to rise in the 5.0±5.5 cm size class, rose sharply to a peak in the

285

5.5±6.0 cm size class, and declined at larger size classes (Fig. 1(b)). In both ®eld studies, the difference in catch rates of the two trap types on ®sh large enough to be retained by both trap types (i.e., greater than 5.5 cm) is greatest in the 5.5±6.0 cm size class and declines at larger size classes (Fig. 1(a) and (b)). In both cases, the catch rate difference was not statistically signi®cant by the 6.0± 6.5 cm size class (for Field Study 1: Zˆÿ1.76, pˆ 0.079; for Field Study 2: Zˆÿ1.47, pˆ0.14) and was negligible in all larger size classes (Fig. 1(a) and (b)). The data from both studies indicate that the higher catch rates of commercial traps than large mesh traps on ®sh large enough to be retained by both trap types (i.e., the higher ®shing power of commercial traps) are driven primarily by a difference in catch rates of the 5.5±6.0 cm size class (for Field Study 1: Zˆÿ3.69, pˆ0.0002; for Field Study 2: Zˆÿ2.45, pˆ0.014). These results strongly support the squeezability hypothesis as an explanation for the reduced ®shing power of large mesh traps. They indicate that the ability of ®sh of body depth slightly greater than 5.5 cm to escape from large mesh traps can fully explain the reduced ®shing power of large mesh traps. It is possible that the squeezing escape of ®sh in the 5.5±6.0 cm body depth size class from large mesh traps only occurs when ®sh are stressed at the time of trap hauling. To test this, we compared the catch rates of ®sh in different size classes in the two trap types on Day 4 in Field Study 2 (i.e., on the day before trap hauling). The pattern of catch of the different size classes in the two traps is very similar to that obtained when the trap is hauled (compare Fig. 1(c) with (b)). The difference in catch rates of the two trap types on ®sh larger than 5.5 cm is driven completely by differences in catch rates of ®sh in the 5.5±6.0 cm, 6.0± 6.5 cm and 6.5±7.0 cm size classes; differences in catch rates on ®sh larger than 7.0 cm body depth are negligible (Fig. 1(c)). This suggests that the squeezing escape of ®sh occurs during normal soak time, and not only at the time of haul. 3.4. The visual image hypothesis 3.4.1. Structural visual image The data acquired in Field Study 2 indicated a signi®cant effect of both trap design (F2,108ˆ12.12, p<0.0001) and site (F11,108ˆ5.62, p<0.0001) on num-

286

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

Fig. 1. The numbers of fish caught in each 0.5 cm body depth size class (a) during Field Study 1; (b) during Field Study 2; and (c) observed in traps on the fourth day of soak (i.e., the day before trap haul) during Field Study 2. Field Study 1 consisted of 46 soaks of commercial traps and 46 soaks of large mesh traps (average soak time was two days). Field Study 2 consisted of 48 five-day soaks of commercial traps and 48 five-day soaks of large mesh traps. The shaded areas correspond to the ``squeezing zone'', i.e., the size classes which appear to be able to squeeze through the apertures of the large mesh. Mesh size aperture of commercial traps is 4.1 cm. Mesh size aperture of large mesh traps is 5.5 cm.

ber of ®sh caught per trap, but the interaction between trap design and site was not signi®cant (F22,108ˆ1.53, p>0.05). The mean catch per trap of the commercial traps (30.4 ®sh) was signi®cantly higher than the mean

catch of both the experimental traps (18.5 ®sh; Q3,108ˆ4.97, p<0.01) and the large mesh traps (14.3 ®sh; Q3,108ˆ6.71, p<0.01; see Fig. 2). Mean catch per trap of the experimental traps did not differ signi®-

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

287

Fig. 2. The mean catch per trap (dark bars) and mean adjusted catch per trap, i.e., catch less fish whose body depth is 5.5 cm (light bars) taken in 48 five-day soaks of large mesh traps, experimental traps and commercial traps in Field Study 2. Error bars are 1 SE.

cantly from that of the large mesh traps (Q3,108ˆ1.74, p>0.05; Fig. 2). This does not support the visual image hypothesis which predicts that experimental traps should have a higher catch rate than large mesh traps. When ®sh of body depths <5.5 cm were removed from the catch of all traps, there was no longer an effect of trap design (F2,108ˆ1.65, p>0.05), although the site effect and interaction were signi®cant (F11,108ˆ6.00, p<0.0001; F22,108ˆ1.68, pˆ0.043). This result again does not support the visual image hypothesis which predicts that commercial traps and experimental traps should have higher catch rates than large mesh traps on ®sh large enough to be retained by all mesh sizes. Since daily ingress is the number of ®sh which entered (and remained) in a trap between successive 24 h observation periods, and since ®sh with body depth <5.5 cm could have escaped from experimental and large mesh traps between successive observation periods, ®sh with body depths <5.5 cm were not considered when ingress rates were compared between trap types, i.e., ``adjusted'' ingress rates were compared. Mean daily adjusted ingress differed signi®cantly between trap types (F2,712ˆ5.23, pˆ0.005).

The mean adjusted ingress into commercial traps (5.99 ®sh per day) was signi®cantly higher than the mean adjusted ingress into large mesh traps (4.26 ®sh per day; Q3,712ˆ4.46, p<0.01; Fig. 3). The mean adjusted ingress into experimental traps (4.79 ®sh per day) was intermediate between, and not signi®cantly different from, mean adjusted ingress into commercial or large mesh traps (Fig. 3; for commercial vs. experimental: Q3,712ˆ3.12, p>0.05; for large mesh vs. experimental: Q3,712ˆ1.35, p>0.05). The higher ingress rates into commercial traps compared to large mesh traps can be interpreted as supporting the visual image hypothesis for increased ®shing power of small mesh traps; i.e., ®shing power is higher because ingress rates are higher, since small mesh traps are more visible. However, it is surprising in this context that the ingress rates of large mesh traps and experimental traps do not differ, since the experimental traps were designed to have a similar visual image as the commercial traps. For ®sh with body depths >5.5 cm, ingress rates of the 5.5±6.0 cm size class into commercial traps are signi®cantly higher than that into experimental traps

288

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

Fig. 3. The mean daily ingress (number of fish per day) of fish with body depths >5.5 cm into large mesh traps, experimental traps and commercial traps during 48 five-day soaks of commercial traps, 48 five-day soaks of large mesh traps and 48 five-day soaks of experimental traps (Field Study 2). Error bars are 1 SE.

(Zˆÿ2.56, pˆ0.01), but ingress rates of this size class into experimental and large mesh traps do not differ signi®cantly (Zˆÿ1.56, pˆ0.12). The apparent differences in ingress between trap types for the 6.0± 6.5 cm size class (Fig. 4) are not statistically signi®cant (for commercial vs. experimental: Zˆ1.075, pˆ0.45; for experimental vs. large: Zˆÿ0.33, pˆ 0.74; for commercial vs. large mesh: Zˆÿ0.99, pˆ0.32). For ®sh with body depths greater than 7.0 cm, no differences in ingress rates to the different trap types are apparent (Fig. 4). This pattern of ingress rates of the different size classes into the different trap types is very similar to the pattern observed for catch rates of the different size classes for the different trap types (Fig. 1). The data in Fig. 4 could be interpreted as supportive of the visual image hypothesis in that ingress to commercial traps appears greater than to experimental traps which appears greater than to large mesh traps. However, the data suggest that

this is only true for ®sh whose body depths are slightly larger than 5.5 cm, i.e., ®sh in the ``squeezability zone'' (see Fig. 1). This issue is discussed further below. 3.4.2. Biotic visual image There was a signi®cant positive relationship between adjusted daily ingress and the number of ®sh in the trap on the previous day (Fig. 5) for commercial traps (Yˆ4.66‡0.081X, pˆ0.0048), for experimental traps (Yˆ3.07‡0.18X, p<0.0001), for large mesh traps (Yˆ3.01‡0.15X, p<0.0001) and for the three trap types combined (Yˆ3.46‡0.13X, p<0.0001), but the model explained little of the variance in all cases 2 2 2 ˆ 0:033; rexperimental ˆ 0:15; rlarge …rcommercial mesh ˆ 2 0:10; rpooled ˆ 0:093†. These results can be interpreted as supporting the visual image hypothesis, i.e., the greater the number of ®sh in a trap, the greater the visual image and the higher the ingress rate, but

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

289

Fig. 4. The total number of fish in each 0.5 cm body depth size class known to have entered traps between successive 24 h observation periods during 48 five-day soaks of commercial, experimental and large mesh traps (Field Study 2). The shaded area corresponds to the ``squeezing zone'', i.e., the size classes which appear to be able to squeeze through the apertures of the large mesh. Mesh size aperture of commercial traps is 4.1 cm. Mesh size aperture of large mesh traps is 5.5 cm.

alternative explanations for the results are considered below. 4. Discussion Under the current model for the catch rate of Antillean ®sh traps which uses ®sh body depth and trap mesh size to predict retention of ®sh in traps and hence catch rate, large mesh traps are expected to have lower catch rates than small mesh traps when harvesting the same ®sh population (Munro, 1983). Consistent with this, the large mesh traps had signi®cantly lower catch rates than the commercial traps in this study, as indicated both by the number of ®sh per trap and weight of ®sh per trap. The results suggest that transition of the Barbados trap ®shery from the current commercial trap mesh size (inch and a quarter mesh, i.e., maximum aperture of about 4.1 cm) to a larger mesh size (inch and a half, i.e., maximum aperture of about 5.5 cm) would result in substantial short-term reductions in catch rate (in the range 53±63% by number of ®sh caught per trap; about 51% by weight

of ®sh caught per trap). Stevenson and Stuart-Sharkey (1980) reported a catch rate reduction of about 45% by number of ®sh per trap and 14% by weight of ®sh per trap in traps of 1.5 in. mesh size compared to 1.25 in. mesh size (the same mesh size differences as in this study) during experimental ®shing on the west coast of Puerto Rico; and Nicholson and Hartsuijker (1983) reported a reduction of 15% by weight of ®sh per trap and 7% by number of ®sh per trap in moving from 1.25 in. mesh to 1.5 in. mesh on the south Jamaica shelf and Pedro Bank (Jamaica). The reductions recorded in the present study are more severe, and may re¯ect the state of extreme over-exploitation believed to characterize the west coast trap ®shery in Barbados. The size of ®sh caught by the large mesh traps was signi®cantly bigger than that caught by commercial traps as indicated by length, weight and body depth. Consistent with this, about 8±20% of the ®sh caught by the commercial traps were immature, compared to about 4±16% of those caught by the large mesh traps. It is important to note that these are pooled percentages for all species and that there are substantial

290

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

Fig. 5. The adjusted daily ingress (i.e., ingress for fish with body depth >5.5 cm) as a function of number of fish caught on the previous day for each of the five soak days of the 48 soaks of each trap type. The relationships are significant for commercial traps (Yˆ4.66‡0.081X, pˆ0.0048), for experimental traps (Yˆ3.07‡0.18X, p<0.0001) and for large mesh traps (Yˆ3.01‡0.15X, p<0.0001), but the model explained 2 2 2 little of the variance in all cases …rcommercial ˆ 0:033; rexperimental ˆ 0:15; rlarge mesh ˆ 0:10†.

differences in this parameter between species. For many species, the percentage of immature ®sh in both trap types was zero; for blue tangs (Acanthurus coeruleus) it was as high as 71.8% in the commercial traps and 52.4% in the large mesh traps. Hunte and Mahon (1994) emphasized that there will be substantial variation between species in this parameter since size at sexual maturity varies considerably between reef ®sh species; and hence that there will be considerable variation between species in susceptibility to recruitment over®shing. The percentage of immature ®sh in the traps is low compared to that reported for other studies. Wolf and Chislett (1974) reported that about 96% of silk snapper (Lutjanus vivanus) caught in 1.25 in. traps on Saba Bank and about 50% of the silk snapper caught in 1.5 in. traps were immature. Beliaeff et al. (1992) reported that almost all of the yellowtail snapper (Ocyurus chysurus) and white grunt (Haemulon plumieri) caught in traps of aperture 3.8 cm off Guadeloupe were juveniles; and Stevenson (1978) reported that a signi®cant proportion of red hind (Epinephelus guttatus), bar jack (Caranx ruber),

white grunt, spotted goat®sh (Pseudopeneus maculatus) and redband parrot®sh (Sparisoma aurofrenatum) caught in traps with maximum apertures of 3.4 and 4.2 cm off Puerto Rico were sexually immature. The low percent of immature ®sh caught in the traps in this study may again re¯ect severe overexploitation in the Barbados trap ®shery. Reef ®sh species which reach sexual maturity at a large size, and are therefore vulnerable to recruitment over®shing, may already exist in negligible numbers in the reef ®sh population. Large mesh traps caught signi®cantly bigger ®sh and a smaller percentage of immature ®sh than commercial traps in this study. This supports the possibility that the catch rates of large mesh traps could increase in the longer term following transition of the ®shery to large mesh traps. Sary et al. (1997) documented an increase in catch rates of large mesh traps three years after the introduction of large mesh traps to a reef ®sh ®shery in Jamaica. Catch rate increase would occur if the decrease in harvest of immature ®sh ultimately results in increased ®sh abundance through increased recruitment, if the size of the ®sh

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

population shifts to larger sizes in equilibrium with the larger mesh traps in the ®shery, and if the transition to a larger mesh reduces growth over®shing. However, comprehension of stock±recruitment relationships in harvested coral reef ®sh is too rudimentary to allow even semi-quantitative a priori predictions on how changes in the percentage of immature ®sh harvested might change recruitment strength and hence population abundance, and modeling catch rate increases that might result from shifts in ®sh population size structure and from reductions in growth over®shing, is complicated if the ®shing power of traps decreases with increases in mesh size. What is evident is that, given the substantial variation between reef ®sh in size at sexual maturity, increase in ®sh abundance occurring through reduced recruitment over®shing, in response to increase in mesh size of traps, will differ considerably between species. An important result of this study is that the ®shing power of the large mesh traps was signi®cantly lower than that of the commercial traps. When ®sh small enough to escape from large mesh traps (body depth 5.5 cm) were removed from the catch of commercial traps, the catch rates of large mesh traps remained lower than that of the commercial traps, in the range 25±35% by number of ®sh caught per trap and about 30% by weight of ®sh caught per trap. Results of previous studies can also be interpreted as indicating that large mesh traps have lower ®shing power than small mesh traps (e.g. Hartsuijker and Nicholson, 1981; Munro, 1983), and this has been attributed to either ``squeezing'' (Hartsuijker, 1982; Ward, 1988) or to visual image effects of the ®sh trap (e.g., Luckhurst and Ward, 1987; Burnett-Herkes et al., 1988; Bohnsack et al., 1989; Newman and McB Williams, 1995). The squeezability hypothesis was tested in this study by comparing the catch rates of commercial and large mesh traps on ®sh of different size classes. The data indicate that the higher ®shing power of commercial traps than large mesh traps is generated primarily by a difference in catch rates on ®sh in the 5.5±6.0 cm body depth size class, and secondarily by a catch rate difference in the 6.0±6.5 cm size class; i.e., in the size classes which might feasibly be expected to squeeze through the 5.5 cm maximum aperture of the large mesh traps. Catch rate differences between traps were negligible for ®sh with body depths larger than 6.5 cm. These results are strongly supportive of the

291

squeezability hypothesis for the higher ®shing power of small mesh traps. The visual image hypothesis would predict higher catch rates, through higher ingress rates, for small mesh traps across all ®sh size classes. The fact that squeezability may be a common and important aspect of the mode of operation of ®sh traps is supported by results presented by Hartsuijker and Nicholson (1981). These authors point out that the mean length of ®sh retained in traps is consistently larger than the mean length predicted by ®sh body depth. In the present study, the fact that the catch rate in commercial traps (maximum aperture 4.1 cm) peaks in the 5.0±6.0 cm body depth size class and not in the 4.0±5.0 cm size class (see Fig. 1(a) and (b)) support the occurrence of squeezing over the latter size range, since younger/smaller ®sh are likely to be more abundant in the population than older/larger ®sh. Squeezability may be an important factor in¯uencing catch in other gear types. For example, Meyer and Merriner (1976) reported that escapement of ®sh from trapnets is affected by body form, body ®rmness and swimming ability of the ®sh. The species which are most likely to squeeze through mesh smaller than their maximum girth are those which are circular in crosssection, which are powerful swimmers, and which have compressible bodies. The visual image of a trap may be affected by the structural design of the trap, referred to here as the structural visual image, or by the number of ®sh in a trap, referred to here as the biotic visual image. The visual image hypothesis for the increased ®shing power of small mesh traps, in the context of the structural visual image, was tested by comparing the ingress rates of ®sh to commercial, large mesh and experimental traps. The experimental traps were designed to have a visual image similar to the commercial traps by having similar mesh size over most of the trap area, but size-speci®c retention capacity similar to large mesh traps by having a panel of similar mesh size as the large mesh traps. If visual image effects were the principal reason for the higher ®shing power of small mesh traps, through increased ingress rates, ingress rates and catch rates of ®sh large enough to be retained by all mesh sizes (i.e., adjusted ingress rates and adjusted catch rates) should be similar for commercial and experimental traps, but higher for these trap types than for large mesh traps. No sig-

292

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

ni®cant differences between the adjusted catch rates of the three trap designs were detected, but daily ingress rates were directly measured and did differ signi®cantly between trap types. As expected under the visual image hypothesis, ingress rates to commercial traps were signi®cantly higher than that to large mesh traps. Ingress to the experimental traps was intermediate between that to the commercial and large mesh traps, but not statistically different from either. The ingress rates to the experimental traps (mean 4.79 ®sh per day) were closer to those of large mesh traps (mean 4.26 ®sh per day) than to those of commercial traps (mean 5.99 ®sh per day). This is not expected under the visual image hypothesis, since the experimental traps were designed to have a similar visual image, and hence similar ingress rates to the commercial traps. It is important to note that ingress is measured here as the number of ®sh known to have entered a trap between successive observation periods that were 24 h apart. It is therefore likely that some ®sh which have entered a trap will escape between successive observation periods, and hence that ingress rates will be underestimated (Munro, 1974); but the escape rate, and hence the degree of underestimation, need not be the same between trap types. Squeezing escape (for the ®sh of body depths >5.5 cm considered here) is likely to be higher for large mesh traps than experimental traps and negligible for commercial traps (see rationale previously discussed), and hence ingress rates are likely to be more underestimated for large mesh traps than experimental traps, and more underestimated for these trap types than for commercial traps. It is therefore possible that squeezability effects can explain the differences in ingress rates estimated in this study for the different trap types. Ingress rates of ®sh in different body depth size classes are of particular interest given the possibility that squeezability effects may bias the ingress rates to different trap types measured in this study. It was only for ®sh in the body depth size range 5.5±6.5 cm that estimated ingress rates were higher for commercial traps than experimental traps, and higher for experimental traps than large mesh traps. Ingress rates to the different trap types do not differ for ®sh in larger size classes. If the visual image of traps in¯uences ®sh ingress rates, the effect would be expected across all ®sh size classes. It seems most unlikely that it should be restricted only to ®sh whose body depths are

slightly larger than the maximum aperture of the large mesh used in the experimental traps and large mesh traps in this study (5.5 cm). On the other hand, this is precisely the size range of ®sh over which squeezing escape is likely to occur from the experimental and large mesh traps. These results strongly support the suggestion that the observed differences in ingress rates to the different trap types arise through different degrees of underestimation of ingress rates for the different trap types as a consequence of different squeezing escape rates from the different trap types. The visual image hypothesis, in the context of the biotic visual image, was investigated by examining the effect of the number of ®sh in a trap on ingress in the subsequent 24 h period. Subsequent ingress increased signi®cantly with the number of ®sh in a trap for all trap types, but the model explained a negligible proportion of the variance in all cases, and the slopes were shallow, such that the effect of number of ®sh in a trap on subsequent ingress was very small. In any event, the effect need not be the consequence of an enhanced visual image of the trap. A similar effect would arise through any social aggregation or schooling tendencies of reef ®sh species. Perhaps more importantly, there was a signi®cant effect of trap site on catch rates in this study, and this could explain the weak correlation observed between the number of ®sh in a trap and subsequent ingress. At a good trap site, the number of ®sh in a trap would be high and subsequent ingress would be high; at a poor trap site, both the number of ®sh in a trap and the subsequent ingress would be low. In conclusion, our results con®rm that larger mesh traps have lower ®shing power than smaller mesh (commercial) traps in coral reef trap ®sheries. We provide strong support for the hypothesis that the higher ®shing power of small mesh traps arises through higher squeezing escape rates from large mesh traps of ®sh expected to be retained by both trap types. We could ®nd no de®nitive evidence indicating that the higher ®shing power of small mesh traps arises through an enhanced visual image effect which results in higher ingress rates to the traps. The fact that the reduced ®shing power of larger mesh traps results primarily from squeezability rather than visual image effects will make it easier to incorporate the reduced ®shing power effect into yield per recruit models for predicting catch rate changes in ®sheries which intend to use larger mesh traps as a management

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

tool, since only those ®sh in size classes immediately larger than the maximum aperture of the large mesh will be affected. Acknowledgements We would like to thank Ian Popple, Matt Chapman, Dana Haggarty, Kelly Skinner, Marianne Corless, Julie Bauman, Christopher Hamacher, Paul Robichaud, Suzanne Dorsey, Emma Cronin and several McGill ®eld course students for help in the ®eld. Don Kramer, Jim Grant, Matt Chapman, Gareth Lawson, Derek Roff, Gray Stirling, Richard Preziosi and Marianne Corless provided advice, comments and criticisms. This work was supported by personal research grant AO264 from the Natural Sciences and Engineering Research Council of Canada to WH, and by a MacArthur Foundation grant to Bellairs Research Institute through the Caribbean Conservation Association. References Aiken, K.A., Haughton, M., 1987. Status of the Jamaica reef fishery and proposals for its management. Proc. Gulf Carib. Fish. Inst. 38, 469±484. Appeldoorn, R., Dennis, G.D., Monterossa Lopez, O., 1987. Review of shared demersal resources of Puerto Rico and the Lesser Antilles region. In: Mahon, R. (Ed.), Report and Proceedings of the Expert Consultation on Shared Fishery Resources of the Lesser Antilles Region. FAO Fish. Rep. 383, pp. 36±106. Becker, W.A., 1992. Manual of Quantitative Genetics, 5th ed. Academic Enterprises, Pullman, WA. Beliaeff, B., Louis, M., Souprayen, J., 1992. Preliminary results on the efficiency and selectivity of the chevron-shaped trap, a traditional fishing device in the Guadeloupe. Proc. Gulf Carib. Fish. Inst. 41, 150±165. Bohnsack, J.A., 1989. Are high densities of fishes at artificial reefs the result of habitat limitation or behavioral preference? Bull. Mar. Sci. 44, 631±645. Bohnsack, J.A., Harper, D.E., 1988. Length±weight relationships of selected marine reef fishes from the southeastern United States and the Caribbean. NOAA Tech. Mem. NMFS-SEFC-215. Bohnsack, J.A., Sutherland, D.L., Harper, D.E., McClellan, D.B., Hulsbeck, M.W., Holt, C.M., 1989. The effects of fish trap mesh size on reef fish catch off southeastern Florida. Mar. Fish. Rev. 51(2), 36±46. Burnett-Herkes, J., Luckhurst, B., Ward, J., 1988. Management of Antillean trap fisheries ± Bermuda's experience. Proc. Gulf Carib. Fish. Inst. 39, 5±11.

293

Cummings, W.C., 1968. Reproductive habits of the sergeant major, Abudefduf saxatilis (Pisces, Pomacentridae) with comparative notes on four other damselfishes in the Bahamas Islands. Ph.D. Thesis, University of Miami, Miami, FL. Dammann, A.E., 1980. Caribbean reef fish: Fish traps and management. Proc. Gulf Carib. Fish. Inst. 32, 100±105. Ehrhardt, N.M., Die, D.J., 1988. Selectivity of gill nets used in the commercial spanish mackerel fishery of Florida. Trans. Am. Fish. Soc. 117, 574±580. Elliot, J.M., 1983. Some methods for the statistical analysis of samples of benthic invertebrates. Freshwater Biological Association Scientific Publication no. 25. Hartsuijker, L., 1982. A re-assessment of the stocks of reef fish on Pedro Bank. Tech. Rep. Proj. FAO TCP/JAM/8902: Potfish. Surv. Pedro Bank, Fish. Div., Jamaica, no. 4. Hartsuijker, L., Nicholson, W.E., 1981. Results of a potfishing survey on Pedro Bank (Jamaica): the relations between catch rates, catch composition, the size of fish and their recruitment. Tech. Rep. Proj. FAO TCP/JAM/8902: Potfish. Surv. Pedro Bank, Fish. Div., Jamaica, no. 2. Hixon, M.A., Beets, J., 1989. Shelter characteristics and Caribbean fish assemblages: experiments with artificial reefs. Bull. Mar. Sci. 44, 666±680. Hunte, W., Mahon, R., 1994. Size selection of fish by the mesh of fish traps. CARICOM Fishery Research Document, CARICOM, St. Vincent. Lee, J.U., Al-Baz, A.F., 1989. Assessment of fish stocks exploited by fish traps in the Arabian Gulf area. Asian Fish. Sci. 2, 213± 231. Lewis, J.B., 1960. The coral reefs and coral communities of Barbados. W. I. Can. J. Zool. 38, 1133±1152. Luckhurst, B., Ward, J., 1987. Behavioral dynamics of coral reef fishes in Antillean fish traps at Bermuda. Proc. Gulf Carib. Fish. Inst. 38, 528±546. Mahon, R., Drayton, N., 1990. Trap fishery management in Barbados: the fisherman's perspective. Caribb. Mar. Stud. 1, 94±100. Matthews, K.R., 1985. Species similarity and movement of fishes on natural and artificial reefs in Monterey Bay. California. Bull. Mar. Sci. 37, 252±270. Maxwell, S.E., Delaney, H.D., 1990. Designing experiments and analyzing data. A Model Comparison Approach, Brooks/Cole Publishing Company, CA. McClanahan, T.R., Kaunda-Arara, B., 1996. Fishery recovery in a coral-reef marine park and its effect on the adjacent fishery. Cons. Biol. 10, 1187±1199. McConney, 1996. Draft management plan for the fisheries of Barbados. Barbados government Document. Meyer, H.L., Merriner, J.V., 1976. Retention and escapement characteristics of pound nets as a function of pound-head mesh size. Trans. Am. Fish. Soc. 3, 370±379. Moran, M., Jenke, J., 1990. Effects of fish trap mesh size on species and size selectivity in the Australian north west shelf trap fishery. Fishbyte 8(2), 8±13. Munro, J.L., 1974. The mode of operation of Antillean fish traps and the relationships between ingress, escapement, catch and soak. J. Cons. Int. Explor. Mer. 35, 337±350.

294

D. Robichaud et al. / Fisheries Research 39 (1999) 275±294

Munro, J.L., 1983. Caribbean coral reef fisheries resources. ICLARM studies and reviews 7, International Center for Living Aquatic Resources Management, Manila, Philippines. Munro, J.L., Reeson, P.H., Gaut, V.C., 1971. Dynamic factors affecting the performance of the Antillean fish trap. Proc. Gulf Carib. Fish. Inst. 23, 184±194. Newman, S.J., McB Williams, D., 1995. Mesh size selection and diel variability in catch of fish traps on the central Great Barrier Reef Australia: a preliminary investigation. Fish. Res. 23, 237± 253. Nicholson, W., Hartsuijker, L., 1983. The state of the fisheries resources of the Pedro Bank and South Jamaica shelf. FAO Fish. Rep. 278(Suppl.), 215±254. Olsen, D.A., Dammann, A.E., LaPlace, J.A., 1978. Mesh selectivity of West Indian fish traps. Mar. Fish. Rev. 40(7), 15±16. Randall, J.E., 1963. An analysis of the fish populations of artificial and natural reefs in the Virgin Islands. Carib. J. Sci. 3, 31± 47. Ricker, W.E., 1975. Computation and interpretation of biological statistics of fish populations. Bull. Fish. Res. Board Can. 191. Roberts, C.M., Polunin, N.V.C., 1993. Marine reserves: simple solutions to managing complex fisheries? Ambio 22, 363± 368. Robertson, D.R., Warner, R.R., 1978. Sexual patterns in the Labroid fishes of the western Caribbean, II: The parrotfishes. Smithsonian Contrib. Zool. 255, 1±26. Robichaud, D., 1996. Factors affecting catch, fishing power and trappability in the Barbados trap fishery. M.Sc. Thesis, McGill University, MontreÂal, Canada. Rosario, A., Sadovy, Y., 1991. Trap mesh selectivity off the west coast of Puerto Rico (January±December 1990), Fisheries Research Laboratory, Puerto Rico Department of Natural Resources.

Samples, K.C., Sproul, J.T., 1985. Fish aggregation devices and open-access commercial fisheries: a theoretical inquiry. Bull. Mar. Sci. 37, 305±317. Sary, Z., Miller, M., van Barnveld, W., Picou-Gill, M., Woodley, J.D., 1991. Facilitating change in artisanal fishery practice: the two-for-one trap exchange programme at Discovery Bay. Jamaica. Proc. Gulf Carib. Fish. Inst. 44, 283±296. Sary, Z., Oxenford, H.A., Woodley, J.D., 1997. Effects of an increase in trap mesh size on an overexploited coral reef fishery at Discovery Bay. Jamaica. Mar. Ecol. Prog. Ser. 154, 107±120. Smolowitz, R.J., 1983. Mesh size and the New England groundfishery ± applications and implications. NOAA Tech Rep. NMFS SSRF 771. Sokal, R.R., Rohlf, F.J., 1995. Biometry, 3rd ed. Freeman, New York. Stevenson, D.K., 1978. Management of a tropical fish pot fishery for maximum sustainable yield. Proc. Gulf Carib. Fish. Inst. 30, 95±115. Stevenson, D.K., Stuart-Sharkey, P., 1980. Performance of wire fish traps on the western coast of Puerto Rico. Proc. Gulf Carib. Fish. Inst. 32, 173±193. Stone, R.B., Pratt, H.L., Parker, R.O., Davis, G.E., 1979. A comparison of fish populations on an artificial and natural reef in the Florida Keys. Mar. Fish. Rev. 41, 1±11. Ward, J., 1986. Fish trap workshop report ± GCFI 1986. Proc. Gulf Carib. Fish. Inst. 37, 251±252. Ward, J., 1988. Mesh size selection in Antillean arrowhead fish traps. In: Venema, S.C., MoÈller-Christensen, J., Pauly, D. (Eds.), Contributions to Tropical Fisheries Biology. FAO Fish Rep. 389, pp. 455±467. Warner, R.R., Robertson, D.R., 1978. Sexual patterns in the Labroid fishes of the western Caribbean, I: The wrasses. Smithsonian Contrib. Zool. 254, 1±27. Wolf, R.S., Chislett, G.R., 1974. Trap fishing explorations for snapper and related species in the Caribbean and adjacent waters. Mar. Fish. Rev. 36(9), 49±61.