 | ||
 |     |  |
| |||||||||
(The hard-copy version of this report (see ordering information at the bottom of this document) contains 21 figures comprised of drawings, distribution maps, bar graphs, and pie diagrams. To reduce download times, only those figures that portray data analyses are being made available here.) Bycatch and its Reduction in the Gulf of Mexico Gulf and South Atlantic Fisheries Development Foundation, Inc. SUMMARY Concerns over the magnitude and species composition of bycatch, discards, and incidental finfish mortality associated with shrimp trawling prompted a 1990 amendment to the Magnuson Fishery Conservation and Management Act that mandated the development of a Bycatch Reduction Research Program. As part of a multi-organizational response to this mandate, the Gulf and South Atlantic Fisheries Development Foundation, Inc. (Foundation) coordinated the development of a strategic planning document - A Research Plan Addressing Finfish Bycatch in the Gulf of Mexico and South Atlantic Shrimp Fisheries. The long-term goal of this comprehensive four-year plan was to provide reliable information about bycatch in the southeast U.S. shrimp trawl fishery. Subsequently, the Foundation focused its programmatic contribution on four of the eight high-priority objectives outlined in that Plan:
In 1993, the Foundation began placing observers aboard voluntarily participating commercial shrimp trawlers to collect fishery-dependent data concerning shrimp trawl bycatch. From 1993-1996, Foundation-contracted observers logged 2,320 days aboard trawlers, collecting information on 3,162 shrimp trawl tows. According to NMFS Galveston, the programmatic partners' pooled data set consists of ~5700 tows, thus Foundation generated data comprises about 55% of the total. Characterization sampling by Foundation observers included 403 days in the Gulf of Mexico collecting data on 479 shrimp trawl tows, 34 days in the South Atlantic producing data for 34 tows, and an additional 14 days were logged monitoring the catch of 20 tows in the rock shrimp fishery operating off the Atlantic Florida coast. Greater effort was expended evaluating the exclusion efficiency of various bycatch reduction devices (BRD's) and turtle-excluder-devices (Ted's). Observers logged 1426 days in the Gulf of Mexico collecting data on 1696 tows examining BRD's, and 244 tows examining TED's. In the South Atlantic, observers spent 443 days evaluating 689 BRD tows. According to two separate NMFS analyses, over 450 and 150 taxa have been identified in trawls from the Gulf of Mexico and South Atlantic respectively; the average catch was 27 kg (~60 lb.) of biomass per hour of trawling. The bycatch to shrimp catch ratios generated by these data were in stark contrast to an often-quoted ratio of 10:1. For the Gulf of Mexico, the bycatch to shrimp ratio was 5¼:1, and for the South Atlantic it was 4½:1. More importantly, the generalization of a 10:1 bycatch ratio has been often misquoted to represent the finfish to shrimp ratio when in fact, in the Gulf of Mexico the finfish to shrimp ratio was 4.2:1, and in the South Atlantic the ratio was 2.8:1. General ecological concerns aside, bycatch reduction has a more pragmatic consequence for fishery management: alleviation of incidental mortality on heavily fished finfish stocks. For example, juveniles of a common Gulf of Mexico species (red snapper, Lutjanus campechanus) and a common South Atlantic species (weakfish, Cynoscion regalis) have been particularly identified as fishes impacted by shrimp trawling. Similar impacts are hypothesized for two other species (king mackerel, Scomboromorus cavalla, and Spanish mackerel, S. maculatus) which occur in both the Gulf of Mexico and the South Atlantic. This juvenile mortality is thought to effect recruitment to the fishable stocks, thus restricting allocations for directed recreational and commercial fishing efforts. The initial goal of the Bycatch Program was to reduce incidental mortality on these species, and others, by 50%. To address this goal, the Foundation evaluated the exclusion efficiency of various bycatch reduction devices (BRD's) and turtle-excluder-devices (TED's). Two general types of BRD's were tested: fisheyes (cone-shaped metal frames inserted into the codend webbing to provide an escape hole), and expanded mesh (large square-hung mesh located in the front of the codend usually encircling a funnel). Three different hard-grid TED's were tested either against each other or against a "naked" net (net without a TED). Two soft TED's were compared against naked nets and against each other, and one was compared against a hard TED. Fish exclusion and shrimp retention were very dependent upon the placement and configuration of the BRD. A total of 12 different fisheye configurations were tested; some configurations were represented by minimal sample sizes and no analyses were conducted on the data. For the Gulf of Mexico, we found that only one configuration - a 5" X 12" placed 30 meshes back from the start of the codend - successfully met mortality reduction goals for red snapper by eliminating 38% of the individuals per net-hour; however there was a 6% shrimp loss associated with this gear. Other fisheye configurations reduced the number of red snapper by 20-25%, but none met the minimum 50% mortality reduction criterion. Four configurations of expanded mesh were extensively tested; three of these provided good fish exclusion, including red snapper (10-25%), without an associated shrimp loss, but only one configuration met the red snapper mortality reduction criterion. Other limited testing occurred for four other BRD concepts; none of these gears met the combined goals of shrimp retention and fish exclusion, but two of them warrant further evaluations. For the South Atlantic, none of the fisheye or expanded mesh configurations met the weakfish reduction goals during Foundation tests; however when combined with other researcher's data, the 5" X 12" fisheye at 30 meshes and the 3-bar expanded mesh BRD did exclude 40% or more of the weakfish taken per net-hour with less than a 5% difference in shrimp catch. Most TED testing was against naked (no TED) nets in the offshore (> 15 fm) northwest Gulf of Mexico; other tests, comparing one TED to another, were conducted over a wider area. Results of these tests indicated hard-grid TED's (Anthony Weedless, Georgia-grid, Super-Shooter, Seymore) did not contribute substantially to the reduction of finfish catch; this does not mean that they do not contribute to the mechanical exclusion of large fishes, but that they did not tend to reduce the catch of smaller fishes. Two soft TED's (Morrison, Andrews 5") did have substantial (40-60%) total finfish exclusion capabilities; the Andrews TED excluded >70% of the red snapper, and the Morrison excluded ~25% of this species. TED's have been mandated for several years now, and finfish mortality reductions attributable to soft TED use should be incorporated in the stock assessments for "key" species such as red snapper and weakfish, dependent upon the percent of the trawlers that have used soft TED's over that time period. Summary - The southeastern shrimp fishery may, in certain areas and at certain times, have an unwanted bycatch far exceeding the targeted shrimp catch. In recent years there has been a contribution to bycatch reduction by the industry through their use of TED's, and bycatch can be further reduced through the use of BRD's. It is imperative, however, that this contribution does not also induce an economic hardship through a concurrent loss of shrimp from the catch. Otherwise, any benefits stemming from enhancement of finfish stocks would be negated by the additional burden on the shrimp industry. If this can be accomplished, in the long-run, reduction of finfish bycatch in the shrimp fishery is ecologically and economically beneficial to the southeast US fishing industry, and thus to the general public and the nation. BYCATCH AND ITS REDUCTION IN THE GULF OF MEXICO AND SOUTH ATLANTIC SHRIMP FISHERIES INTRODUCTION Among U.S. fisheries, shrimp traditionally ranks first or second in value and in the top ten in volume, with the majority of production coming from the southeast (South Atlantic and Gulf of Mexico) coastal states. Southeast fishers land approximately 250 million pounds of shrimp annually valued at $450 to $500 million, accounting for >80% of landings and >90% of value of the entire U.S. shrimp industry. In the southeastern U.S., the shrimp fishery provides substantial direct (harvesting, processing, distribution, retail) and indirect (boat building and maintenance, fishing gear/fuel suppliers) economic opportunities, and the influence of this fishery is felt throughout the U.S. because of widespread demand for this delicacy. The gear with the widest applicability in this economically dominant fishery, the otter trawl, is non-selective with a large incidental harvest (bycatch); bycatch varies seasonally and regionally between one- and seven-fold the quantity of shrimp caught. Based on shrimp landings, bycatch may exceed 1.5 billion pounds annually. Unfortunately, there is a limited market or use (and limited incentive to develop a market) for this catch, thus most is discarded. Incidental mortality attributable to shrimping activities is thought to impact recruitment capabilities of several overfished finfish stocks. Reduction of this bycatch and the associated incidental mortality is desirable both from an ecological and economic perspective. Concerns over the magnitude and species composition of bycatch, discards, and finfish incidental mortality associated with shrimp trawling stem from various factors. Ecologically, we know that ocean resources are finite and must be utilized efficiently. Secondly, increased urbanization of the southeastern region (leading to environmental degradation and a yet unquantified effect on fish stocks) has produced an increased awareness, and in some cases mis-perceptions, of fishing activities and the effects of fishing in coastal waters. Finally, on an economic basis, incidental mortality may contribute to problems of overfishing. For example, juveniles of a common Gulf of Mexico species (red snapper, Lutjanus campechanus) and a common South Atlantic species (weakfish, Cynoscion regalis) have been particularly identified as fishes impacted by shrimp trawling. Similar impacts are hypothesized for two other species (king mackerel, Scomboromorus cavalla, and Spanish mackerel, S. maculatus) which occur in both the Gulf of Mexico and the South Atlantic. Shrimp trawl mortality may impact recruitment for these species, thus limiting the resource available for directed recreational and commercial fishing efforts. These general concerns about finfish bycatch in the shrimp fishery prompted a 1990 amendment to the Magnuson Fishery Conservation and Management Act. This amendment mandated that the Secretary of Commerce establish a program under the jurisdiction of the South Atlantic and Gulf of Mexico Fishery Management Councils to assess the impact of incidental harvest by the shrimp trawl fishery on fishery resources. Since 1990, the Gulf and South Atlantic Fisheries Development Foundation, Inc. (Foundation) has been involved in a series of projects, as part of a cooperative multi-organizational multi-year program, to address this mandate. This report summarizes the results of Foundation contributions to this program. METHODS The design for the overall Bycatch Program was cooperatively conceived and approved by multi-organizational oversight and monitoring panels (Table 1). The primary oversight group, the Bycatch Steering Committee, was composed of 34 representatives from the commercial fishing industry, environmental organizations, recreational groups, NMFS, Sea Grant, the Gulf of Mexico and South Atlantic Fishery Management Councils, the Atlantic and Gulf States Marine Fisheries Commissions, academic institutions, and state management agencies. The 16-member Technical Review Panel had the responsibility of designing appropriate sampling and analytical protocols. Additionally, four members of a Statistical Panel had the responsibility of evaluating sampling protocols in relation to statistical requirements associated with data stratification and analysis. The Gear Review Panel (GRP), an 8-member group, evaluated the success of the various bycatch reduction devices being tested, and played a crucial role by providing recommendations for increasingly more detailed and stringent evaluations of specific designs. Extensive descriptions of the established protocols are detailed in:
The reader is referred to these documents for detailed information on sampling protocols; only pertinent portions of these protocols are highlighted in this document for the purpose of describing the work performed. All bycatch-reduction-devices (BRD's) were evaluated according to the GRP's "Bycatch Reduction Device Test Criteria". All BRD's were subjected to a 4-step testing process; advancement from one step to the next required review of data and approval by the GRP:
For BRD evaluations, Foundation emphasis was directed to those designs that achieved "operational testing" status according to the GRP. This included two general gear types: fisheyes (cone-shaped metal frames inserted into the codend webbing to provide an escape hole), and expanded mesh (large square-hung mesh located in the front of the codend usually encircling a funnel). Although some general BRD configurations are defined in this text, for presentation purposes, many of the BRD's are identified by the computer codes that were assigned to each configuration. See Appendix I for a description of each configuration. Additionally, this report describes the location of several fisheyes according to the number of meshes that the BRD was placed from the start of the codend or bag; for example: "a 5" X 12" oval fisheye placed 30 meshes back from the start of the bag". This reference point is associated with bags that were 120 meshes in length; obviously this position would represent a different proportional distance to a long (140) or short (80) mesh bag. On the other hand, based on our results, there does not appear to be a direct relationship between the proportional distance and the size of the fisheye. It has been recommended that the placement be defined as meshes, or another measure, from the tie-off rings. Unless otherwise noted in this report, the tests were conducted using 120 mesh bags. To conduct these tests, the Foundation arranged for observers to collect data aboard voluntarily participating commercial shrimp trawlers. All sampling was fishery-dependent with data collected on trawls fished under normal working conditions. For each tow sampled, whether it was for characterization or BRD evaluations, detailed information concerning gear configurations, locations, times, and catch was recorded by the observer. Total catch of sample nets was weighed. For characterization, the observer randomly selected a net for sampling, and took a well-mixed sample of the catch from that net. Sample size equaled 12 kg (26.5 pounds) per hour of tow (large catches or extended tows were sub-sampled). For BRD evaluations, a '1-basket' {ca. 30 kg or 70 pounds} sample from both a control (without BRD) and experimental (with BRD) net was collected. Such paired-net sampling for BRD evaluations allowed direct comparisons of the catch between two nets on a single trawl effort; to do so required that the two nets be fishing with the same general efficiency. To offset the possibility that the two nets were not fishing with similar efficiency, the BRD was moved to the "control" net for half the samples (the "old" BRD net then became the control). By doing so, any biases associated with a specific net were introduced into both the control and BRD data, thus providing a more accurate reflection of the BRD's effectiveness. Whether sampling was for characterization or BRD evaluation, species (finfish and invertebrates) within these samples were identified, enumerated, weighed as a species lot, and individuals of selected "key" species were measured. The total number of tows sampled per trip was dependent upon the fishing activity of the vessel and the logistic restraints imposed from the time required to complete one sampling effort (i.e. one sampling effort may not have been completed before the next tow was brought aboard, thus this next tow was not sampled). In addition to examining the efficiency of new gears, the Gear Review Panel recommended that additional investigations be made concerning the bycatch exclusion capabilities of several turtle-excluder-device (TED) designs. Certain finfish stock assessments are based on data generated without TED's either through TED-exempt fishery-independent surveys, or from fishery-dependent data generated prior to the introduction of TED's to the fishery. Bycatch target goals should include the contribution derived from TED's. Such a credit was applied by the Atlantic States Marine Fisheries Commission for initial BRD certifications by the South Atlantic states. Concurrent federal regulations for the South Atlantic have not incorporated a "credit for TED's", nor have the proposed federal Gulf of Mexico regulations included such a credit. To address this need, the Foundation conducted TED exclusion tests in the Gulf of Mexico comparing the exclusion capabilities of four hard TED's against each other or against "naked" nets (nets without TED's), two soft TED's (Andrews 5" and Morrison) against "naked" nets and against each other, and an Andrews 5" against hard TED's. Naked net versus a TED testing was only permitted in waters deeper than 15 fathoms west of 89° west longitude, and with a four-hour tow time restriction. Tests conducted outside this area were required to have an operational TED in all nets that were fished. The results of these tests were compared to similar testing for TED exclusion efficiency in the South Atlantic (supported in part by the Foundation) that was conducted jointly by the University of Georgia Marine Extension Service and the South Carolina Department of Natural Resources. Data Processing: Data collection and management procedures were designed to ensure minimal data entry errors. The data sheets were proofed for accuracy and completeness at the end of each trip by the observers, and the data were then given to data entry personnel for computerization. Any irregularities or questionable/confusing entries on the data sheets were noted by data entry personnel who referred to the observer for confirmation of corrections, as necessary. Once the data were computerized, a printout of the computer file was cross-checked against the original data sheets for keypunch errors, corrections were made to the computer file, and the edited version checked again against the previous version. Once the data files were verified as accurate, the data were archived in a pooled, multi-organizational data set at the NMFS Galveston Lab, and an additional set of the data was supplied to Foundation personnel. Data Analysis: Characterization - For characterization analyses, two types of samples could be used - true characterization sampling of the catch in a standard trawl equipped with a TED, and the sampling of the "control" net during BRD evaluations. This "control" net was a standard trawl net equipped with a TED. The "control" net samples differed only in that a limited number of species were identified, whereas true characterization samples identified the vast majority of species in the trawl. The Foundation did not focus its efforts on characterization studies, thus this report relies on a NMFS analysis of the full programmatic database. In general, those analyses were conducted with the following methodologies: The species number and weight for the entire tow of the designated sample net were extrapolated using the ratio of the sample weight (or number) vs.. the total net weight:
These extrapolated values were then converted to a catch-per-unit-effort (CPUE) based on the hours towed:
Mean (+/- SE) catch rates for the trip (or other unit of time) were then calculated based on the summation of these CPUE values for all successful tows sampled during a particular trip or time period. BRD Evaluation - One of the more pragmatic consequences of this research program has been the development of management regulations requiring the use of BRD's in the fishery to reduce finfish incidental mortality. Based on stock assessments for the two key finfish species, red snapper in the Gulf of Mexico and weakfish in the South Atlantic, the goal of the Bycatch Research Program was to reduce shrimp trawl incidental mortality by 50%. The South Atlantic additionally has an interest in achieving similar reductions for Spanish and king mackerels. The goal is to reduce mortality on the Age 0 and Age I age classes; these are the dominant ages taken by trawlers. In the South Atlantic, two target options have been defined for the weakfish and mackerels: a 50% fishing mortality reduction or a 40% reduction in the numbers of fish taken per net-hour, but the 50% mortality reduction is still the only criterion for evaluating red snapper reductions in the Gulf of Mexico. Calculating such a reduction in fishing mortality requires examining the reductions for specific length classes within each year class. Then the length-specific reductions must be incorporated into the survivorship calculations for each length group within these year classes. Because of the need for very specific stock assessment details concerning the various 'cohort' strengths of red snapper within the Age 0 and Age I classes, calculating actual mortality reduction was relatively impossible for most BRD researchers. Once the catch reductions for specific lengths within the age classes were known, the data were forwarded to NMFS for final calculations of mortality reductions. All final mortality reduction estimates for any of the BRD designs and configurations were provided to the appropriate Council or other management agency by NMFS personnel. The only values that we could readily derive from our portion of the database were overall reductions in either number per net-hour, or weight per net-hour, or specific reductions for individual size groups within the age classes. In some cases, a 25% reduction in the number taken per net-hour translated to a >50% reduction in mortality on the overall catch because the percent reductions for certain sizes were substantial; no direct correlation existed however between number excluded and mortality reduction because of the variations in the sizes/ages taken during the different seasons. In the case of red snapper, it would be beneficial if, in the future, a more readily applicable measure for the effectiveness of a BRD was developed, such as the percent reduction in number per hour that was substituted for weakfish in the South Atlantic. Of course, for all tests and evaluations, it was equally important to consider the changes in shrimp catch attributable to the various BRD designs. The initial Bycatch Research Plan specifically identified that a researcher's goal should be to produce a BRD that addresses finfish reduction while also retaining at least 95% of the shrimp. This has not been an identified criterion for acceptance or rejection of a particular BRD, but a benchmark for achievement. In any event, for any BRD recommended (or certified) for use, it was imperative that shrimp retention data be available to the shrimpers. Obviously, it would be to a shrimper's benefit to select a BRD that had the least shrimp loss, but if a shrimper preferred to use a BRD that met finfish requirements but retained less shrimp, that would be the shrimper's prerogative. All analyses included comparisons of the shrimp retention, total biomass reduction, and specific finfish reductions for the various BRD designs. Exclusion capabilities of several TED's were also evaluated using the same procedures. To calculate the reduction attributable to a specific BRD or TED, total biomass reduction for a particular tow was calculated as:
For the various species, reductions were calculated by:
Only tows which have representatives of a species in one or both nets are used for these calculations. Total absence of a species from both nets does not provide information on reduction-gear efficiency. Additionally, if a species was considered "select", then the extrapolations were unnecessary. These means were tested for significant difference (p < 0.05) through the use of paired t-tests according to the following hypotheses: Ho : µ control - µ BRD = 0 Ha : µ control - µ BRD ¹ 0 FINDINGS Sampling Overview: From 1993 through 1996, observers logged 2,320 days in southeast U.S. waters aboard voluntarily participating shrimp trawlers. Performing both characterization and BRD-evaluation sampling, these observers collected information on 3,162 shrimp trawl tows. This included 2,385 BRD tows, 244 TED tows, and 533 characterization tows. According to NMFS Galveston, the 1992-1996 pooled programmatic data set consists of ~5,700 tows, thus the 1993-1996 Foundation generated data comprises about 55% of the total. Characterization sampling by Foundation observers included 403 days in the Gulf of Mexico collecting data on 479 shrimp trawl tows, 34 days in the South Atlantic producing data for 34 tows, and an additional 14 days and 20 tows of effort collected preliminary characterization data on the rock shrimp fishery operating off the Atlantic Florida coast. Greater effort was expended on evaluations of the exclusion efficiency of various BRD's and TED's. Observers logged 1,426 days in the Gulf of Mexico collecting data on 1,696 tows examining BRD's (Figure 1), and 244 tows examining TED's. In the South Atlantic, observers spent 443 days documenting 689 BRD-evaluation tows (Figure 2). Characterization Sampling: The following information is based on data available in two different NMFS documents:
According to the NMFS, true characterization sampling consisted of approximately 1,700 tows (of which the Foundation contributed 533), but when the samples taken from the "control" net during BRD evaluations are included in the analysis, the programmatic database consists of ~5,700 tows, and ~4,600 of those were used to characterize the catch in the fishery (some tows without complete information could not be appropriately extrapolated and used). More than 450 taxa have been identified in trawls from the Gulf of Mexico, and the average catch was approximately 27 kg of biomass per hour of trawling. Shrimp constituted 16% of the total catch by weight, other invertebrates 17%, and finfish 67% (Figure 3). The 10 most abundant species by weight were (in descending rank): longspined porgy (15%), brown shrimp (9%), Atlantic croaker (9%), inshore lizardfish (6%), pink shrimp (3%), and Gulf butterfish (3%), with lesser blue crab, white shrimp, longspined swimming crab and brown rock shrimp each comprising 2 % of the catch. A special concern in the Gulf catch has been the occurrence of juvenile red snapper. According to two additional analyses -one by NMFS the other by Texas Shrimp Association - red snapper constituted about 0.4 - 0.5% of the total catch by weight, ranking 74th by number (2.5 individuals/hr) and 48th by weight (0.14 kg/hr). Based on the estimated shrimp effort in the Gulf of Mexico, this may translate to 35 million individuals annually; NMFS recommended substantial reductions in shrimp trawl incidental mortality to rebuild this overfished stock. For the South Atlantic, about 150 taxa have been found in shrimp trawls, and the average catch rate was about 26 kg of biomass per hour towed. Shrimp were 18% of the catch by weight, other invertebrates comprised 31%, and 51% of the catch was finfish (Figure 3). The ten most abundant species were: cannonball jellyfish (14%), white shrimp, spot, and Atlantic menhaden each at 9%, brown shrimp and other jellyfish each at 8%, Atlantic croaker (6%), southern kingfish and blue crab each at 4%, and star drum at 3%. Special finfish species of concern included weakfish, king mackerel, and Spanish mackerel; although these species represent a minimal contribution to the total catch, based on stock assessments for these species, management agencies have recommended their exclusion from trawls as a method to enhance recruitment and increase the size of their stocks. The bycatch to catch ratios produced from this new study were in stark contrast to an often quoted bycatch to shrimp ratio of 10:1. For the Gulf of Mexico, the bycatch to shrimp ratio was 5¼:1, and for the South Atlantic it was 4½:1. More importantly, the generalization of a 10:1 bycatch ratio has often been misquoted to represent the finfish to shrimp ratio when in fact, in the Gulf of Mexico the finfish to shrimp ratio was 4.2:1, and in the South Atlantic the ratio was 2.8:1. Seasonal finfish to shrimp ratios (by weight) were:
Additionally, the regional Gulf and South Atlantic finfish to shrimp ratios by weight were (Gulf areas are < or > 60 ft depth contour; no such stratified analysis was conducted for the South Atlantic; numbers in parentheses indicate the sample size in numbers of tows for each category):
The Gulf Florida region identified above includes the entire western Florida coastline, and the environment and species composition in the northern half of this region is distinct from that found in the southern part. In a specific analysis conducted by the Foundation on its own data, the three general categories - finfish, shrimp, and other invertebrates - were each equally represented in the south Florida (Tortugas) pink shrimp fishery. In other words, the finfish to shrimp ratio was 1:1. Bycatch-Reduction-Device (BRD) Evaluations: Two general types of bycatch reduction devices (BRD's) were examined: fisheyes (cone-shaped metal frames inserted into the codend webbing to provide an escape hole) (Figure 4), and expanded mesh (large square-hung mesh located in the front of the codend usually encircling a funnel) (Figure 5). Eleven different configurations of four sizes of fisheyes were tested during 1,031 tows in the Gulf of Mexico (Figure 6). Six configurations of two sizes of fisheyes were tested during 407 tows in the South Atlantic (Figure 7). Four configurations of expanded mesh were extensively tested; all four were tested during 536 tows in the Gulf of Mexico (Figure 8), and two configurations were tested during 270 tows in the South Atlantic (Figure 9). Limited testing also occurred for four other BRD concepts during 133 tows in the Gulf and 12 tows in the South Atlantic. Two fisheye and three expanded mesh configurations excluded substantial quantities of finfish while retaining most of the shrimp; one additional industry-developed BRD may prove acceptable, but needs further evaluation concerning reductions of "key" finfish species. Based on the data we collected, only two BRD's (1 fisheye and 1 expanded mesh) successfully met the red snapper reduction criterion for the Gulf of Mexico (Table 2), and none of the gears met the minimum weakfish reduction criterion in the South Atlantic (Table 3). Fish exclusion and shrimp retention are very dependent upon the placement and configuration of the BRD. For example, although it might seem logical that a small hole in the front of the codend would exclude less fish and retain more shrimp than a large hole in the back of the codend, this is not necessarily the case. Apparently, several interrelated factors effect the behavior and subsequent exclusion of both shrimp and fish. For each configuration we tested, the resulting exclusions changed in varying amounts. A 5" X 12" oval (football-shaped) fisheye was tested in three positions. When placed 30 meshes back from the start of the bag along the top-center line (EE code), it reduced 38% of the red snapper individuals per net-hour, but lost a statistically significant 6% of the shrimp in Gulf of Mexico tests. In the South Atlantic, this configuration resulted in an insignificant 1% shrimp loss, but for our tests, it only reduced weakfish by 12%, king mackerel by 21%, and Spanish mackerel by 31%. When these data were combined with other programmatic data sets, the overall results indicated that this configuration met the 40% reduction in number per net-hour criterion set for the South Atlantic. When placed 45 meshes back along the top centerline (EK code) in 140 mesh bags in Gulf of Mexico tests, it lost similar amounts of shrimp, and only reduced the number of red snapper caught per net hour by 25%. In 1993 South Atlantic tests using 120 mesh bags, this configuration did not lose much shrimp (1%), but it also did not exclude much of the finfish, and did not apparently exclude sea trout (no distinction was made between the two species during these tests). Based on successes achieved by North Carolina researchers, the Foundation also tested another modification of the 5" X 12" fisheye. The fisheye was still 30 meshes from the start of the bag, but offset 15 meshes to the outside of top-center (MJ code). Results differed dramatically between the Gulf of Mexico and the South Atlantic. In the Gulf, it had 20-80% finfish reductions, including a 25% red snapper reduction, but it lost a significant 8% of the shrimp. In the South Atlantic, this configuration didn't appear to function well, having minimal (<20%) fish exclusions for those species that were excluded; some were not excluded at all. Early in the program, we also used two 5" X 12" fisheyes placed 45 meshes back in a 140-mesh bag and at nearly horizontal ("3 o'clock" and "9 o'clock") positions (FE code). Compared to a single fisheye, this configuration lost twice as much shrimp (16%), and surprisingly it did not reduce the catch of red snapper. A 4" X 7" oval shaped fisheye produced mixed results. When placed 30 meshes back from the start of the bag along the top center line (ED code), it lost only 4% of the shrimp in both Gulf of Mexico and South Atlantic tests. In the Gulf, it only reduced red snapper by 21%, but in the South Atlantic, this configuration had the best weakfish reduction - 34% - of any BRD tested by our group in this region. We moved it back to the 45-mesh position (EJ code) for Gulf tests, and it excluded large numbers of snapper (35%), but it lost 17% of the shrimp. These were exceptionally high shrimp losses, and the losses may be an artifact of the sample size (see page 17). When moved even farther back to the posterior third of the bag (about six feet from the tie-off) along the top center line (EP code), it did not lose much shrimp (1%), but it also didn't exclude substantial numbers of fish. Surprisingly, these tests were conducted aboard a trawler that has used this configuration for years. However, our tests were conducted primarily in inshore and near-shore waters of Louisiana where the younger and smaller fishes reside. We never tested this configuration in offshore (> 10 fm) waters; further study is warranted. Given results from other researchers, we also tested an intermediate sized (4.5" X 9") fisheye placed 30 meshes back from the start of the bag (ET code). Initially tested by researchers in North Carolina with good results, subsequent South Atlantic testing by Georgia researchers were similarly positive. We installed this BRD on a trawler in the western Gulf to see if it would work equally well for red snapper; the results were confusing. The configuration did not lose any shrimp, and it had good finfish reductions ranging from 20% to 70% for the various species. On the other hand, it did not reduce red snapper catches at all, and substantial quantities of red snapper were taken during the trials. This is contradictory to most other tests we have conducted. Although it is more difficult to stimulate red snapper to leave a trawl (see discussions later in this report), usually when other finfishes are commonly excluded, red snapper reductions also occur. We only had 12 successful tows with this gear configuration; additional testing is warranted to further examine red snapper reduction capabilities of this BRD. A diamond shaped fisheye (LA code), commonly used in North Carolina, was monitored aboard a North Carolina trawler working off the Florida Keys. Only nine tows were sampled, but the BRD net had substantially less shrimp (14%) than the control net. Otherwise, it did well on finfish reductions. Two expanded mesh designs, originally developed by NMFS, had very similar results. The original design had 3 bars of large 4-5" mesh (8-10" stretch) hung on the square and encircling a funnel (AA code). The funnel itself was modified to include a laterally extended flap on the terminal end. This design was later modified to include two additional bars of large mesh located beside the funnel flap in anticipation of additional exclusion capability (AE code). This modification did not enhance or impair the BRD's capabilities; both designs had no shrimp loss and excluded about 25% of the red snapper individuals per net-hour in the Gulf of Mexico tests, and about 30% of the weakfish and >40% of the mackerels in the South Atlantic tests. NMFS subsequently modified the funnel so that it was smaller in diameter and made from a stiffer webbing in order to increase its functional longevity (AT code), but initial tests with this gear on both coasts suggested that the modification reduced the BRD's effectiveness for both shrimp retention and finfish exclusion. Another modification of this design (AS code) evolved through testing by two Texas fishers (L. Jones, H. Davis) in the Gulf of Mexico; exceptional results were achieved with this configuration. Shrimp loss was a non-significant 4%, and red snapper reductions equaled 40%. The design was simplified from the original by forming the funnel from panels cut from the side of the codend. Thus, large openings were created where the panels are folded in, precluding the need for the large-mesh webbing. Most importantly, a cone shaped deflector located behind the funnel appears to be the major improvement with this gear. Given the apparent success of the cone, this device will be used with other BRD's to determine if it can enhance their exclusion efficiency. Three other gears that partially combined the TED and the BRD were tested during single trips but none produced substantially positive results. A gear developed by the NMFS (IA code) incorporated funnel-shaped webbing behind the TED with framed openings in the exterior of the bag behind the TED; this design also used a hummerwire deflector to stimulate the fish to move to the openings. There was a minimal 3% shrimp loss, and relatively good finfish exclusion, but little red snapper exclusion (6%). Further testing might be beneficial, but other problems with this configuration prompted NMFS to suggest that additional modifications were necessary. No more testing was conducted on this concept. An industry designed BRD had a framed bottom opening located behind the TED (NA code). The designer (C.J. Kiffe) modified this gear several times, and we tested his fourth version during a trip in the eastern Gulf of Mexico in 1996. There was no shrimp loss associated with this gear, but finfish make up such a limited contribution to the catch in this region that the efficiency of the device is still uncertain; additional testing in the western Gulf of Mexico, especially in areas inhabited by red snapper, would be beneficial. A modification to the Busken TED was tested for "proof of concept" in the northeast Gulf of Mexico, but shrimp loss was unacceptable at 9%, and finfish losses were minimal; additional modifications of this gear were needed. The analyses discussed so far in this report have included all successful tows completed in each region, without consideration of where the tows were conducted or what type of shrimping gear was being used, thus more specific analyses were needed to address some management issues. For the Gulf of Mexico, BRD's will only be required west of Cape San Blas, Florida, or in NMFS Statistical Zones 8-21. Results for many analyses would not change with this area restriction; for example only sporadic records of red snapper occurred in Statistical Zones 1-7, thus most, or all, of the data for this key finfish species would be the same. As noted earlier, shrimp retention is dependent upon the target species, and pink shrimp which are more common in the eastern Gulf of Mexico have a higher exclusion rate (Figure 10); thus shrimp retention should be expected to change when the eastern Gulf data are excluded from the analysis. Additionally, for both the Gulf and South Atlantic region, there is a possibility that soft TED's (Morrison, Andrews, Taylor, etc.) may be decertified for use; thus an analysis of the effectiveness of these gears with only hard TED's was also warranted. At this writing, for the Gulf of Mexico, only the EE BRD has been identified for certification because it is the only BRD to date that has been shown to exclude sufficient quantities of red snapper (the AS BRD reported herein also meets that criterion); this configuration is also widely accepted by the South Atlantic states. For the Gulf of Mexico, an analysis of the EE BRD efficiency in Statistical Zones 8-21 (with all TED types) reduced the sample size from 197 (see Table 3) to 142 and showed that there was a non-significant shrimp loss of 4% by weight (compared to a significant 6% for the total Gulf) with a total finfish weight reduction of 22% (p < 0.05); red snapper reduction by number of individuals per net-hour was the same (38%) as no red snapper were collected in EE fisheye tests in the eastern Gulf of Mexico. Further restriction of this analysis to include only those tows using a hard TED further reduced the sample size to 80 tows. Results showed a non-significant 5% shrimp loss by weight, a 24% total finfish weight reduction, and a 51% reduction of red snapper by number. A similarly restricted analysis for EE BRD tests with a hard TED in the South Atlantic reduced the number of tows in the sample from 81 to 47, and yielded a 3% shrimp loss (compared to an overall 1% loss - see Table 3), and a 26% reduction in weakfish (compared to 12% with all data). Turtle-Excluder-Device (TED) Evaluations: The Foundation conducted several TED tests during this program (Table 4). All naked net tests were conducted in offshore (>15 fm) waters in the northwest the Gulf of Mexico. Two hard TED's (a Georgia grid-style and an Anthony Weedless) were tested against each other, and a Super-Shooter was tested against a naked net. None of these tests indicated that the hard TED's contributed to bycatch reduction; they did not indicate any shrimp loss either. Specific attention was paid to testing the Andrews 5" soft TED against a variety of other TED's, or against a naked net. As far as the soft TED's were concerned the Andrews 5" was extremely effective at reducing the bycatch of finfish, including red snapper (77% reduction in number per net-hour). On the negative side, this TED had a rather high shrimp loss rate (16%). Tests of the Andrews against two different hard TED's provided similar results of exclusion efficiency as tests against a naked net; again indicating a minimal contribution of hard TED's to bycatch reduction. The Morrison TED also reduced finfish bycatch substantially, with many common fishes being excluded by 20-40%; there was a 23% reduction in the number of red snapper taken per net hour as well. This TED also lost substantial quantities of shrimp (13%). In tests comparing the exclusion rates of a Morrison to an Andrews, as one might expect, the values were basically the difference between the exclusion capabilities of each against a naked net. These data were compared to a previously conducted South Atlantic TED evaluation study (supported in part by funds to the Foundation) by the University of Georgia Marine Extension Service (UGMAREX) and the South Carolina Department of Natural Resources. That study compared the exclusion of three different TED's to a naked net (Table 5). Results for each TED varied according to the specific area and time of the test, but in general, the hard TED's had relatively good exclusion of common finfishes such as spot and Atlantic croaker. This contrasted sharply with our Gulf-based results, which showed little exclusion from hard TED's. During at least one series of South Atlantic tests, all three TED nets caught 10-20% less brown shrimp than the naked net; these July tests in South Carolina were substantially different from the results of a June test in Georgia waters suggesting either a gear problem or a very specific local phenomenon. The Morrison TED did consistently reduce the catch of weakfish by 15-25%. These studies provide an indication of the exclusion capabilities of TED's, and a benchmark for management agencies that wish to incorporate a "credit for TED's" to meet the reduction requirements for certain species. Such a credit is only applicable if stock assessments do not include material from the post-TED implementation period. From these data it is apparent that TED's, used in combination with the most efficient BRD's, will meet many of the bycatch reduction requirements, and some TED's may singly meet those criteria as well. Shrimp and Fish Exclusion Considerations: Shrimp: Shrimp retention has long been a controversial issue concerning TED's. As noted above, shrimp loss was greater with the soft TED's compared to most hard TED's. In the South Atlantic study, a Morrison soft TED lost 10-15% more shrimp than a "naked" net, with the specific value dependent on the area and time of testing; this was a very similar value to our comparisons between a Morrison TED and a naked net in offshore Gulf waters (13%). The Andrews 5" TED lost an even greater amount. The Andrews TED net caught 16% less shrimp than a naked net, and in tests against a Morrison TED, the Andrews TED produced 6% less shrimp (i.e. the Morrison lost ~10% compared to what a naked net would have produced). It should be noted however that shrimp loss is a known characteristic of soft TED's, but many fishers prefer these TED's because of the finfish reduction which allows them to fish more efficiently. In other instances, certain TED's may be a necessity; for example, the Andrews TED is one of the few TED's that can be used successfully, without excessive clogging, in the west Florida shelf area. Shrimp retention values for any specific BRD design are very dependent upon the species of shrimp being targeted. White shrimp seem to be excluded the least, followed by brown shrimp, with pink shrimp being the most difficult to retain; these differences have been examined for some BRD configurations. During 1993, a 5X12" fisheye placed 45 meshes back from the start of the bag (EK code) was tested extensively in both the Gulf and South Atlantic. In the South Atlantic, during brown shrimp season, the shrimp loss was about 6-7%. This was similar to results in the Gulf of Mexico where almost all testing occurred during brown shrimp fishing. However, later in the year in the South Atlantic, as the white shrimp contribution increased in the catch, there was actually a nominal shrimp gain. Overall, there was a non-significant 1% shrimp loss over the year of testing in the South Atlantic. Similarly, gears which have worked quite well in the northwestern Gulf or in the South Atlantic Bight have worked poorly in the Florida Keys where pink shrimp are dominant. We analyzed the changes in catch for both red snapper and shrimp across the Gulf of Mexico using a 5" X 12" fisheye at 30 meshes (EE code) (Figure 10) according to two depth strata (< > 15 fathoms). Shrimp loss for this gear was exceptionally high (10%) in the eastern Gulf of Mexico where pink shrimp dominated the catch. For the western Gulf of Mexico, where brown shrimp dominated the catch, near-shore (< 15 fm) shrimp loss was 5% off both Texas and Louisiana, but there was no shrimp loss in the offshore strata. For both near-shore and offshore, red snapper reductions were substantial. Overall, for most fisheyes tested in the Gulf of Mexico, there has been at least a 5% difference in the shrimp catch of a net equipped with a BRD (see Table 2), and a consistent minimal (1-4%) difference in shrimp catch with fisheyes in the South Atlantic (see Table 3). Shortly after BRD's regulations were implemented in the South Atlantic, fishers complained of very high (as much as 35%) shrimp losses, and these losses were documented by state DNR observers. This report is not intended to be an economic analysis, but we did ask one trawler owner - shrimp processor to calculate what a 5% reduction in his landings would mean to his operation. A 5% reduction in the gross annual production, and gross income, (assuming no market response to the 5% drop in availability) would translate directly to a 5% loss in crew share which was approximately a $1400 reduction based on 1996 landings of the operation. More importantly, this would translate to a 58% decrease in profits to the owner, and a 20-30% loss in profit to the processor (depending on availability or substitute foreign product). It should be remembered however that a shrimp "loss" per tow or per trip may not equate to a similar reduction in production over the course of a year. For example, although TED's do lose some shrimp, annual production of shrimp in the southeast U.S. has been relatively stable; individuals trawlers may be producing less, but overall the fishery is continuing to produce an equivalent amount of shrimp. On the other hand, if a loss per trip or per trawler does equate to an actual reduction in landings (annual production) for that owner, this may substantially impact the economics of that particular operation. At least part of the shrimp loss problem that is often attributed to the BRD may actually stem from the between-net variation in catches between and within specific tows, and may in part explain the exceptionally high losses noted by shrimpers in the South Atlantic shortly after BRD regulations were implemented. As an example of the problems that can be encountered with reaching a conclusion on BRD efficiencies, the Foundation's data set accumulated on a 5" X 12" fisheye @ 30 meshes (EE code) was divided into two distinct geographic zones: NMFS Statistical Zones 1-7 (Florida west coast south of Cape San Blas) and Zones 12-21 (Louisiana through Texas). There was a substantial variation in shrimp catch rates (kg/net-hour) between tows for both zone groupings (Figure 11a). A shrimp "loss" of 60% or more occurred on some tows; conversely, on other tows just prior/subsequent to such a "losing" tow, a shrimp "gain" of the same magnitude would occur. For the purposes of this discussion, and to simplify the graphic presentation, all difference values that were larger than +/-40% were discarded from the data set. As illustrated in Figure 11a, shrimp catch fluctuations of +/- 20% between nets within a tow were common; this can result in a change of as much as +/- 40% between tows. One might argue that the BRD caused any and all shrimp losses; however if the BRD is the only variable involved, then one would need to apply the same credit, instead of blame, to the BRD for those instances where a substantial shrimp gain occurred in the BRD net. Obviously neither argument can be logically pursued by considering the BRD the only variable; a substantial inter-net variability in catch rate exists within a single tow and between tows. Looking at these data in more detail, in a second analysis (Figure 11b1), the difference in the running mean (moving average) catch rate (kg/net-hr) in control nets was compared to the running mean catch in the BRD nets {a running mean is the mean catch of control net tows 1+2 compared to the mean catch in the BRD net for tows 1+2, then the mean catch of control net tows 1+2+3 compared to the mean catch in the BRD net for tow 1+2+3, etc.), and again the differences in catch between these two values was often substantial on a tow-by-tow additive basis. For Statistical Zones 1-7, the running mean difference was relatively stable with increasing numbers of tows in the samples; the percent difference in shrimp catch rate for the BRD net was 5-10% less than the shrimp catch rate in the control net. However, for the western Gulf sample (Statistical Zones 12-21) the running mean was substantially impacted by the variable shrimp catches between nets, especially during the first few summations when the sample size was small. The first tow in this series (Tow #94) had a 6% greater shrimp catch in the BRD net, but the very next tow had 16% less shrimp in the BRD net, and there was a substantially larger catch (in kg) for this second tow, thus the running mean kg/hr was strongly influenced by this second tow, and the differences in catch rate were also impacted. With the addition of this second tow, the running mean difference was -15% for the BRD net. On Tow #99, a similar 16% shrimp "loss" occurred, and these two "losses" strongly influenced the running mean differences for several subsequent tows. Conversely, on Tow #127, a 40% larger shrimp catch was noted for the BRD net, and this value strongly influenced the running mean; at this point, the running mean difference changed from -5% to zero (no overall difference for the average catch between the BRD and Control nets for Tows 94-127). However, a "loss" of 14% on the very next tow (#128), followed by a 38% "loss" on Tow #129 drove the running mean difference back down to about an 8% overall "loss" rate. The running mean difference then gradually rose to a final value of -3.5% difference between the BRD net and Control net for the entire sample (Tows 94-234). To better illustrate how a single tow, or sub-set of tows, within the entire data set could influence the resulting differences in overall catch rate, a second random sort of the data sets was conducted (Figure 11b2) {no graphic such as is illustrated in Figure 11a is provided for this second run}. In this run, the Statistical Zone 1-7 data set started with a 4% increase in the BRD net, followed by some substantial losses on the ensuing tows, which drove the running mean difference down to nearly a 10% shrimp loss for the first 10 tows. The running means then continued to fluctuate between -6% and -9%. More drastic changes occurred in the western Gulf data set (Statistical Zones 12-21); there were shrimp catch "losses" during the first 5 tows (Tows #94-99), followed by some substantially positive catch differences in Tows #105-115. These positive values resulted in a running mean difference that indicated the BRD net was actually out-producing the Control net for the next 50 tows, then there were a series of tows where the BRD net had a lower shrimp catch, and as the sample size increased, the shrimp "loss" returned to its slightly negative terminal value of -3.5%. Obviously, these drastic differences (either positive or negative) are not entirely attributable to the BRD, and are more likely reflective (in part) of catch variability and efficiency between nets on a single tow. The importance of this analysis is directly related to current BRD certification protocols that have been established by the South Atlantic Fishery Management Council, which will also be used by the various South Atlantic states. Those protocols prescribe a 30-tow series be conducted where appropriate numbers of selected finfish species are taken. The data set used for Figure 11 readily illustrated that after only 30 tows, or for any 30 tow sub-set within the data set, there will still be substantial variability in the end result. The fluctuations seen in these shrimp catches are also representative of the fluctuations that will be seen in the BRD's ability to exclude finfish. Because of this variability it is probable that the end result values may not indicate that the BRD was certifiable even though it actually was contributing substantial finfish reduction without any shrimp loss. Conversely, and more importantly, it is also highly probable that after the 30-tow test, a BRD that actually does little to reduce finfish and/or has a substantial shrimp loss may wind up being certified. There is no easy solution to this dilemma. The 30-tow requirement follows minimal statistical requirements, but in actuality may not be a sufficient sample size. To successfully complete such a 30-tow series will probably require at least one to two weeks of the shrimp fisher's time. Conversely, to require an even larger sample size that will reduce statistical uncertainty would be an excessive and undo hardship on the fisher who is testing the BRD. Additionally, the data sets included samples taken aboard several trawlers over an extended, and some of the changing trends may reflect changing efficiencies over time by individual trawlers, as well as differences in efficiency (success in reducing bycatch) between trawlers in the fleet. However, the major differences are more likely attributable to inter-net variability on a single tow. Optimally, a shrimp fisher wants the nets to fish equally and as efficiently as possible. On the other hand, certain tows may be subjected to different concentrations of shrimp in the area swept by each net (shrimp are patchily distributed, and although over time the catch will be similar, on a short-term basis, the catch can be very different), different environmental conditions (dragging the edge of a channel or trough), different fishing techniques (e.g. all turns are made to port during a tow, thus the starboard net sweeps a wider area), or other conditions that effect the quantity of the catch in a particular net. These kinds of differences need to be recognized by shrimp fishers, so that, if possible, they can avoid situations that cause the problem. To help fishers recognize and realize these variables, over the course of the Bycatch Program, Foundation personnel or their representatives held numerous workshops throughout the southeast to interact directly with fishers. Because fishers are not usually able to attend meetings of the various fishery management agencies and must often depend on second-hand information concerning such issues, early on, these workshops were designed to convey background information about the Bycatch Program, along with the status of current tests, to the fishers. More importantly, the workshops provided considerable feedback about BRD's and BRD regulations from the industry. Realizing the value of such feedback from the industry, the last series of workshops was designed to specifically address fishers' concerns, and to discuss potential solutions to problems being encountered during day-to-day fishing activities. Working cooperatively with Texas Sea Grant and University of Georgia Marine Extension Service personnel, with the help of many local Sea Grant and industry liaisons, over 84 individual meetings with single or small groups of fishers were conducted in more than 40 ports from South Carolina to Texas. One conclusion noted by shrimpers was the same as we had documented in our efforts: brown shrimp were excluded at a higher rate than white shrimp. Causes of excessive shrimp loss identified by shrimpers included:
Relatedly, some differences in shrimp catch rates between the Gulf of Mexico and the South Atlantic may stem from differences in the fishing tactics, and our sampling. From available video monitoring, fisheyes often let out the majority of fish during haulback, thus the Gulf vs.. South Atlantic differences in shrimp catch may represent differences in the general depth of fishing. Most Gulf sampling effort has been in offshore waters whereas the sampling of South Atlantic fishery was conducted in very shallow near-shore waters. These offshore Gulf trawlers spend long periods at sea, working through rough weather that might not be encountered by South Atlantic boats simply because returning to port is easier for the South Atlantic fleet. It has been noted that shrimp loss is worse in rough weather, especially during a downsea haulback where belching may be a problem. Similarly, "wash-down" at the end of the tow can lead to a shrimp loss. The former phenomenon occurs when a wave surges the boat and nets forward and up, and as the boat returns to the wave trough, the nets sink back more rapidly than the catch. Thus, the catch remains nearer the mouth, the TED, and the BRD; any of which could then be used for escape by the catch, including shrimp. "Wash-down" is practiced by many captains; when the nets are at the surface, the vessel is run forward at high speed. This pushes the catch to the tail of the net, and eliminates a substantial portion of mud collected in the nets. However, the turbulence created in the bag during "wash-down" may actually eject part of the catch (including shrimp) from the bag through the BRD opening(s). Using an underwater video camera mounted near a fisheye, researchers from Texas A&M Marine Advisory Service and the University of Georgia Marine Extension Service documented that both these actions can and do contribute to additional shrimp loss. Many of these factors are unavoidable characteristics of the fishery; different vessels have different horsepower, fishing capabilities and available gear. However, fishers must be made aware that these factors will affect the shrimp catch when using a BRD; only then can the fishers attempt to address the problems, if possible, to minimize the impacts. Red Snapper: The reductions of red snapper are of critical importance for management options on this species in the Gulf of Mexico. The goal is to reduce shrimp trawl incidental mortality by 50%; unfortunately, none of the BRD's tested directly achieved a 50% reduction in the catch rate. Many of the BRD's did exclude > 20% of the red snapper. Soft TED's also help reduce red snapper catches; the Morrison TED excluded 23% of the fish per hour, and nets equipped with the Andrews 5" TED had 77% less red snapper than a naked net. Red snappers were ubiquitously distributed in the western Gulf of Mexico continental shelf waters, and sporadically recorded on the west Florida shelf (Figure 12). Looking specifically at the western Gulf of Mexico, red snapper occurred in 393 of 719 total fisheye tows (Figure 13), and 201 of 403 expanded mesh tows (Figure 14). Their distribution tended to be offshore of the 10-fathom contour except off south Texas where near-shore captures were not uncommon. This distribution is reflected in the proposed management options for federal waters as well; those regulations would only mandate BRD use west of Cape San Blas, Florida. In reviewing the data by season, reduction levels for red snapper were strongly correlated to seasons and the associated size of the fish occurring on the shrimp grounds; this was true for both fisheyes (Figure 15) and expanded mesh (Figure 16). At least according to the smallest fish available in our samples, Age-0 red snapper recruit to the shrimp grounds in late summer-early fall at a size of about 50 mm (2") fork length (FL). If smaller fish (< 50 mm FL) do recruit to the grounds earlier, they apparently do not occur in the final catch; it may be that these smaller fish can pass through the meshes in the codend. The juveniles continue to reside in these regions through Age-I (mean size about 175 mm {7"} FL). {For stock assessment purposes all fish become Age I at the start of the calendar year (i.e. at 6 months of age), but for purposes of this discussion, ages represent biological ages where spawning birthdates are mid-summer.} Quarterly stratification of the red snapper length-frequencies taken in both the "control" and "BRD" net indicated that during the first calendar quarter the majority individuals were small Age-0 fish less than 120 mm (4.7") FL, and at this size, there was very little reduction. As the year progressed, and these fish increased in size to larger than 110 mm FL (4.33"), reductions increased, but only for those fish larger than 100 mm (3.9"). By mid-summer, when most red snapper on the shrimp grounds were larger than 100 mm, reductions were substantial. During the fall, when the next year-class began recruiting, overall reductions were reduced again because of limited reduction on these smaller fish, but there was still good reductions of the few larger fish taken. This size-limited reduction has a kinematic basis. According to NMFS Pascagoula researchers (the reader is referred to NOAA Tech. Memorandum NMFS SEFC 327 "Status report on the potential of gear modifications to reduce finfish bycatch in shrimp trawls in the southeastern United States 1990-1992" for more detailed information), fish smaller than 100 mm (~4") cannot sustain a swimming speed to overcome the speed of the trawl, thus they tire and are pushed to the back of the cod end during an extended tow-time. Red snapper 60-100 mm can sustain a swimming speed of only about 0.2-0.3 meters (7.9-11.8") per second, and have a burst speed of about 0.6-1.0 meters (23.6-39.4") per second, whereas the trawl is being pulled forward at about 1.3 meters (57") per second (= 2.5 knots {nautical miles per hour}). Once the juveniles attain a size larger than 110 mm (4.33"), they can produce the needed sustainable swimming speed to keep a constant position within the net, and seek escape routes. A second problem with red snapper centers on an innate optomotor behavior pattern. It is a reef-oriented species, and the structure provided by the shrimp net itself is a stimulant which attracts the fish, and they will "take up station" within the net. Additionally, the juvenile fish will respond to water flow patterns within the net, and will take up station in turbulent or low-flow areas. Most of these BRD gears are designed to alter water flow and thus act as an attractant to these fish; however the optomotor response to remain with the solid structure of the net is difficult to alter, thus the fish tend to remain inside the net even when located in the immediate vicinity of the BRD. Some underwater video observations have demonstrated that red snapper that exit the net through a BRD may actually turn around and re-enter the net through the BRD. Although an overall 50% reduction was not achieved, BRD's tested have achieved substantial reductions for larger Age-0 red snapper, and most Age I fish (i.e. fish larger than 100 mm {4.33"} FL). After more detailed examination by NMFS personnel concerning the catch rate reductions for discreet size classes, it was determined that, for some BRD's, the 25-30% reduction in individuals actually achieved a >50% fishing mortality reduction for the combined Age 0 - Age I group. According to a NMFS-authored report (Some considerations in determining bycatch reduction requirements - Nichols, et al. 1995)
Although more Age 0's are in the trawl, the percent of Age I's caught in relation to the stock size of Age I is greater (according to NMFS estimates). Thus, the substantial reductions achieved by the time fish reach Age I (ca. 120 mm or 4"), provide sufficient reduction in incidental mortality on the combined group. These same size-based concerns, combined with the time of year that sampling occurred, may have been one of the factors that influenced the results for several other BRD's (Figure 17) and TED's (Figure 18). The majority of our tests for a small fisheye (ED code) occurred in December through February, and the majority of the fish taken were 70-110 mm FL; only a 21% exclusion was obtained overall, and almost none for these smaller size categories. Tests on the MJ BRD occurred from July through October, and the most common length category was 90-100 mm; for the 90-100, 100-110, and 110-120 mm classes, reductions were good: 45%, 52%, and 69% respectively, but no reductions were noted for fish < 90 mm, and these small fish comprised 40% of the fish taken. Similarly, the fall (November-January) testing of the very effective AS BRD did not result in the exclusion of any red snapper smaller than 100 mm; these fish comprised about 30% of the total number of red snapper taken. The 100-110 mm fish were excluded at 34%, and then all larger size categories had >50% exclusions. Exclusions attributable to the two soft TED's were not so size related. For the Morrison TED, an relatively consistent exclusion rate of 10-25% was seen in the commonly occurring size classes which ranged from 60-110 mm FL. The reductions attributable to the Andrews TED were even more dramatic with all size classes ranging from 40-220 mm having at least a 60% reduction. Weakfish: It would seem logical that this same size issue applies to fishes of the South Atlantic region as well. Most weakfish taken are less than 175 mm {~7"} FL; these are primarily Age 0, but fish >150 mm FL could be small Age I fish. In any event, the conclusions concerning the ability to generate sustainable swimming speeds inside a trawl would be applicable to a wide variety of small (< 100 mm) finfishes, including weakfish and mackerels. This would be especially true for small weakfish; this is a rather delicate juvenile that would likely tire quickly and be injured easily in a trawl. Goals for reduction are also set at 50% mortality reductions or 40% of the individuals taken per net-hour, but according to the BRD efficiencies (Table 3), it was not achieved for any of the BRD's. All four BRD's tested in this region excluded weakfish at different rates We initially thought that this size factor was an explanation for our limited success in reducing this species in the South Atlantic; on the other hand, length-frequencies for several of the BRD's tested did not support the size related exclusion capability. The small fisheye (ED code) had the best exclusion rate of any fisheye tested (34% -see Table 3) (Figure 19 - top), although the sample size was rather small, and testing was limited mainly to July. Interestingly, the smaller size classes (60-110 mm FL) were excluded by about 40-70%, but very little exclusion was achieved for fishes >110 mm FL. The length frequency of weakfish caught during tests of the medium-sized fisheye placed 15 meshes off top-center (MJ code) (Figure 19 - bottom) was bimodal in distribution with a good representation of weakfish at 80-110 mm FL and another group at 130-150 mm FL. This bimodality stemmed from the sampling effort which included the month of July (8 tows) and then December through February (55 tows); the smaller fish were taken in the winter sampling and the larger fish during the summer. The July sampling, although only a few tows, produced large numbers of weakfish; by contrast the winter sampling produced very few fish per tow, but enough tows were completed to collect a substantial number of fish. As with the ED BRD, the better reductions were seen in the smaller sizes (20-60% for 60-110 mm FL fish) taken in the winter; again no reductions were noted for fish the larger fish taken in the summer. For the primary BRD's tested in South Atlantic waters, it was possible to look at the length-frequencies on a quarterly basis, similar to the analyses conducted for red snapper in the Gulf of Mexico. In general, these analyses provided similar length-frequencies to the less-sampled BRD's discussed above, except that surprisingly, the quarterly size distribution did not increase over time. Unlike the results for the MJ BRD where weakfish were decidedly smaller in the winter than in the summer, there were no distinct periodic modal size shifts for either the EE or AA BRD. From a biological standpoint, the catch is primarily composed of Age 0 fishes, and once this age class recruits to the shrimp grounds, the length-frequencies should reflect the growth of these individuals over time much like the quarterly length frequencies generated for red snapper in the Gulf of Mexico (Figures 15 and 16). Unfortunately, there are three variables that tend to mask such distinctions: (1) different spawning periods according to a north-south gradient, (2) a protracted spawning period which produces numerous cohorts within an age class, and (3) a north-and-south movement of trawlers along the coast during the year stemming from (a) local open and closed seasons and (b) a shift in the target effort on different shrimp species over the year. For the 5" X 12" fisheye @ 30 meshes (EE code), almost all testing was in the second half of 1994, and interestingly, the size frequency for the third quarter (July-September) had a stronger contribution of larger fish (120-150 mm FL) than the size frequency for the fourth quarter (October-December) (Figure 20). Almost no exclusion was seen in the third quarter tests, but in the fourth quarter, there was a 20-40% exclusion for all size classes. A similar size shift was noted for the MJ BRD, but in that case, the magnitude was much greater. In this instance, all third quarter sampling occurred in South Carolina waters, and all fourth quarter sampling was in Georgia waters; thus the shifted size distribution between quarters may have been related to very local spawning and recruitment patterns. Comparisons of the size distribution by quarter for the expanded mesh-extended funnel BRD (AA code) further compounded what appears to be a complex situation. Testing was conducted during all four quarters of 1995, plus the fourth quarter of 1994 (Figure 21). The first quarter length-frequency had a mode of 100-110 mm; the second quarter had a mode of 100-110 mm with a strong contribution by 90-100 mm fish; the third quarter had a mode of 90-100 mm fish, and the fourth quarter had a mode of 90-100 mm as well. Nearly all fourth quarter fish larger than 110 mm were taken in 1994. Reduction rates in all four quarters were similar, ranging from 20-40% for the more common (smaller) size groups, but little exclusion was achieved for the larger fish collected during the fourth quarter of 1994. The only other BRD tested in 1995 was the MJ BRD (Figure 19), and in that sampling, the fourth quarter fish were similar in size (80-110 mm) to fish taken in the fourth quarter 1995 AA BRD tests, but third quarter sampling was dominated by fish 130-150 mm; much larger than the 1995 third quarter fish taken during AA BRD tests. However, the MJ BRD tests were conducted in southern Georgia - northeast Florida waters, whereas the AA BRD tests were in northern Georgia and South Carolina waters. This further supported the suggestion that there was some geographic difference in the population(s) of weakfish between South Carolina and northeast Florida during a common timeframe. In addition to a geographic differences in sizes of weakfish within a timeframe, there appeared to be a distinct difference in the size distribution of weakfish occurring on the shrimp grounds between 1994 and 1995 within the South Carolina - Georgia area. When looking at the results from 1994 EE BRD tests, all third quarter sampling was in South Carolina and these tests produced a size distribution dominated by fish 120-150 mm FL. By contrast, the 1995 third quarter AA BRD data, which were also collected primarily in South Carolina, produced a length frequency dominated by fish of only 80-110 mm FL. The 1994 fourth-quarter length frequency from the AA BRD tests in Georgia had a distribution of 80-200 mm with a mode of 120 mm as did the 1994 fourth quarter EE BRD tests. The 1995 fourth quarter AA BRD sampling produced a length frequency with a mode of only 90-mm (range 60-110 mm). Only the 1995 South Georgia - northeast Florida data set included large weakfish. There were no readily discernable differences in the types of fishing gears (net types or TED's) or towing speeds being employed during 1994 versus 1995, in many cases the sampling during both years occurred on the same trawlers, and in some cases on the same fishing grounds. If there were no fishing gear variations, then the inter-year size differences must have had a biological basis. This situation may have been influenced by environmental factors such as temperature or salinity which may have affected either the spawning period or age class growth, or both. The end result was that only a portion of the size range of weakfish known to exist on South Carolina - Georgia shrimp grounds was available during 1995 testing. The varying successes and failures concerning BRD efficiency and the inherent biological shifts in the weakfish stock structure on an annual and sub-regional basis are good examples of the problems that will be encountered when attempting to certify a BRD for use over a broad area. At least according to the data available, exclusion was greater for weakfish that were less than 110 mm FL, and in fact, almost no exclusion occurred for fishes larger than this size. If exclusion is size-dependent, this may affect future BRD certifications; testing conducted during periods when large fish are abundant will more likely produce negative results, whereas tests conducted during periods when small fish are abundant may be more likely to produce positive results. For example, none of the BRD tests conducted in the South Atlantic by the Foundation met the minimum criterion (50% reduction in mortality or a 40% reduction in the number of fish taken per net-hour) for weakfish or mackerels. Our database on the EE BRD (a design which has been accepted for use by most South Atlantic states), totaled 97 tows of which 82 were considered "successful" and usable for analysis; this is far and above the required 30 tows outlined in the South Atlantic protocols. Regardless, if size selection is a factor, based on our 12% weakfish reduction, which may have been related to the larger sized fish being taken, this BRD would have failed miserably. Similarly, the quarterly analyses of the expanded mesh - extended funnel BRD suggested that the results for some individual quarters might have met the criteria, but over the 15 month test period, which included larger fish taken during the fourth quarter of 1994, the reductions were insufficient. With the distinct differences in size classes noted between 1994 and 1995, tests conducted in 1994 had no exclusion, and if these had been certification trials, the BRD in question would have failed the test. On the other hand, with the smaller fish that occurred in 1995, that same testing would have certified the BRD. Given the variability's associated with the seasonal and spatial differences in the occurrence and recruitment of the "key" finfishes and their variable reduction rates under different testing conditions, it is quite likely under the new certification procedure that functional BRD's may be rejected and perhaps non-functional BRD's may be accepted. Either way, the results may be biased as to the overall contribution that the BRD makes at reducing incidental mortality on the South Atlantic juvenile weakfish stock. In summary, there are numerous variables associated with testing and certifying BRD's for use in the fishery. Based on the time of year and the size of the finfish occurring on the shrimp grounds, bycatch reduction may vary. It would seem logical that larger fish would be excluded at a better rate than small fish, and this was corroborated for red snapper in the Gulf of Mexico, but at least based on our data collected in the South Atlantic, just the opposite was true for weakfish. Annual fluctuations in time of spawning and recruitment strength to a particular area may impact test results (and subsequent actual performance in the fishery) for any BRD. Additionally, one must be very careful in assessing the differences in shrimp catch between a BRD-equipped net and a non-BRD net; between net variability within a tow and between-tow variability within a series of tows may mask the BRD's actual influence on the shrimp catch. These problems leave a BRD investigator between a "rock and hard place"; testing should be conducted under conditions reflecting the commercial fishery, but commercial fishers will be hard pressed to commit the time needed to collect data for a minimum number of tows where the location of the sampling may be driven by presence of the bycatch instead of the presence of commercially viable quantities of shrimp. On the other hand, at least given the variability associated with our testing during a four-year period, this "minimum" number of tows may actually not be sufficient to provide accurate information concerning the BRD's actual efficiency. CONCLUSIONS Reduction of finfish bycatch in the shrimp fishery has many positive ecological and economic ramifications, if it can be accomplished without a concurrent reduction in the shrimp catch; otherwise the economic impact on the shrimp fishery might far outweigh any benefits provided to finfish stocks and their fisheries. With a reduction in unwanted bycatch, the industry should realize a reduced cost to harvest and process the catch, as well as a higher quality product. These reductions will also provide indirect benefit by ameliorating a negative perception about "waste" in this fishery, and make a positive ecological impact on the faunal community inhabiting areas where shrimp are abundant. Just as importantly, reduction in juvenile finfish mortality is anticipated to increase available stocks of commercially and recreationally important fishes, thus alleviating user-group conflicts stemming from current catch restrictions. This Program would have been impossible to begin, let alone complete, without the assistance that was offered by the southeastern commercial shrimp industry. The many owners and operators of shrimp trawlers from North Carolina to Texas are to be commended for their voluntary participation and their dedicated interest in this program. Their direct participation in the field operations, and their interaction with scientists, managers, and gear technologists was invaluable in completing the objectives of the Program. Table 1. Membership of the various oversight committees and panels associated with the Cooperative Research Program Addressing Finfish Bycatch in the Gulf of Mexico and South Atlantic Shrimp Fisheries
Table 2. Gulf of Mexico shrimp trawl fishery bycatch reduction evaluations conducted by the Gulf and South Atlantic Fisheries Development Foundation, Inc. between January 1993 through December 1996. Refer to Appendix I for bycatch reduction device codes. Numbers in parentheses indicate the number of paired tows (simultaneous tow of a BRD equipped net against a "control" net {without a BRD} where the species or category occurred in either one or both nets {zero catch in both nets does not indicate BRD effectiveness}. A minus (-) in front of a number indicates a reduction in catch rate in the BRD net compared to the control net, a plus (+) indicates an increased catch rate in the BRD net. Asterisk (*) indicates the difference in catch rate between the control and BRD net was significant (p < 0.05) with a paired t-test. Results for major categories for two other limitedly tested BRD's (HC and LA) are listed in footnote 1.
† approx. half the testing in FL Keys where majority of listed finfish species do not commonly occur; other tests in western Gulf (29 tows) focused on snapper and shrimp reductions without concurrent finfish reduction analysis. This sized fish-eye was also tested in 2 other positions, primarily in the northwestern Gulf, looking specifically at shrimp and red snapper catches: @ 5 meshes from start of bag (36 tows) biomass in the BRD net was -16%, total finfish +14%, shrimp -7%, red snapper +9%; @ 45 meshes (9 tows) biomass reduction was -28%, shrimp -17%, and red snapper -35%. tested in eastern Gulf only (Statistical Zones 2-7) where majority of listed finfish do not occur commonly. footnote 1: LA tested in FL Keys only (n = 9): biomass -32%*, total finfish -30%*, shrimp -14%; HC tested in FL Keys only (n = 8): biomass -0-, total finfish +11%, shrimp -2%. Table 3. South Atlantic shrimp trawl fishery bycatch reduction evaluations conducted by the Gulf and South Atlantic Fisheries Development Foundation, Inc. between January 1993 through December 1996. Refer to Appendix I for bycatch reduction device codes. Numbers in parentheses indicate the number of paired tows (simultaneous tow of a BRD equipped net against a "control" net {without a BRD} where the species or category occurred in either one or both nets {zero catch in both nets does not indicate BRD effectiveness}. A minus (-) in front of a number indicates a reduction in catch rate in the BRD net compared to the control net, a plus (+) indicates an increased catch rate in the BRD net. A star (*) indicates the difference in catch rate between the control and BRD net was significant (p < 0.05) with a paired t-test. The last three species were added to the analyses in 1995 because of their common occurrence in the trawls in this region.
† no distinction between weakfish (Cynoscion regalis) and silver seatrout (C. nothus) was made during these tests Table 4. Catch rate differences of various TED's tested. First TED listed for each column is considered the experimental gear; second TED (or naked net) is considered the control. All numbers indicate a reduction in the catch rate unless preceded by a plus (+) sign which indicates an increase in the catch rate of the experimental gear. Maximum number of successful tows is indicated in parentheses in headings; for specific species or species groups, only tows that contained the species or species group in one or both nets were included in analysis. A dashed line entry (--) indicates that the sample size was two or less (including zero) tows, thus any values generated were most likely not indicative of any true reduction effectiveness. An asterisk (*) indicates a statistically significant (p< 0.5) value from zero (no reduction) based on a paired t-test.
Appendix 1. Descriptions of Bycatch-Reduction-Devices (BRD's) according to their designated computer codes used in this report. Some fisheye computer code descriptions were generic as to the BRD's placement (i.e. front third of codend); for these descriptions, more specific locations are noted as tested. Some codes listed in this appendix are not discussed in the text; this list includes all BRD's examined by the Foundation - some were minimally tested and results are footnoted or no analyses were conducted.
"This document served as part of a completion report to the National Marine Fisheries Service MARFIN program in conjunction with National Oceanographic and Atmospheric Administration Award No. NA57FF0285. The views expressed herein are those of the author and do not necessarily reflect the views of NOAA or any of its sub-agencies." A hard copy of this report is available by mailing five dollars ($5.00 US [to cover copying and mailing]) to - Gulf & South Atlantic Fisheries Development Foundation, Suite 997, Lincoln Center., 5401 W. Kennedy Blvd., Tampa, FL 33609; [telephone: (813) 286-8390; email: steve.branstetter@worldnet.att.net]. Payment must be in US funds (in advance is preferred); personal checks are fine, but credit card orders cannot be handled. |
|
Email comments,
questions about the website or report broken links to
SFA |