CONTENTS Introduction Methods Results Discussion Acknowledgments Literature Cited Appendix Tables & Figures
Northeast Fisheries Science Center Reference Document 06-24
Christopher D. Orphanides1 and Kathryn D. Bisack2
Analysis of Virginia Fisheries Effort as a Component in the Development of a Fisheries Sampling Plan to Investigate the Causes of Sea Turtle Strandings
1National Marine Fisheries Serv., Narragansett Lab., 28 Tarzwell Dr., Narragansett RI 02882
2National Marine Fisheries Serv., Woods Hole Lab., 166 Water St., Woods Hole MA 02543-1026
Web version posted November 27, 2006Citation: Orphanides CD, Bisack KD. 2006. Analysis of Virginia fisheries effort as a component in the development of a fisheries sampling plan to investigate the causes of sea turtle strandings. US Dep Commer, Northeast Fish Sci Cent Ref Doc 06-24; 49 p.
Information Quality Act Compliance: In accordance with section 515 of Public Law 106-554, the Northeast Fisheries Science Center completed both technical and policy reviews for this report. These predissemination reviews are on file at the NEFSC Editorial Office.
Annual reported sea turtle strandings in Virginia peaked at 506 animals in 2003 (Swingle et al. 2004), with May and June having the highest number of strandings in all years (NMFS 2002). In 2002, NMFS enacted measures aimed at reducing sea turtle mortality in Virginia pound net gear (NMFS 2002).
This report analyzes commercial effort during May-July in 2001, 2002, and 2003 in other Virginia fisheries, as a step towards allocating fishery observer coverage as part of a long-term bycatch monitoring plan. Such a plan typically comprises several stages as enumerated in the NMFS National Approach to Standardized Bycatch Monitoring Programs (NOAA, 2004). Before observer coverage begins, anecdotal or other evidence (such as strandings) often indicate that bycatches are occurring. Using this anecdotal information, a baseline observer program is then established to assess whether a systematic program is needed to estimate bycatch. At the conclusion of the baseline stage, sufficient data are generally available to determine if there is indeed a bycatch problem. If so, the sampling program then moves to a pilot stage to obtain refined bycatch estimates, and estimates of the bycatch variance. During the pilot stage, the fishery is commonly sampled proportional to effort. After several years of sampling, the sampling program typically advances to an optimal sampling design, which requires an assessment of how many sampling trips are needed to determine the total bycatch with an acceptable coefficient of variation (CV). A more developed observer program may be implemented which allows alternative stratification designs to be deployed in an optimal approach. If regulations are developed and enacted to reduce bycatches, sampling intensity may need to increase to attain acceptable CVs if observed bycatches decrease as a consequence of the management regimen, because the associated variances of the bycatch rates usually increase when this occurs. By following these steps, bycatches can be systematically assessed, then monitored in the future.
The present study is at the baseline observer program stage. The Sea Turtle Stranding and Salvage Network (STSSN) has documented sea turtles strandings on Virginia beaches for 25 years, particularly during spring when sea turtles migrate into the Chesapeake Bay. In 2001 stranding levels jumped following increased strandings in the previous five years. Also, in spring 2001 a NMFS observer found three loggerhead turtles entwined in the large mesh of two pound net leaders. Sea turtle interactions in the region have been documented in the sea scallop dredge (Murray 2005), bottom otter trawl (Murray 2006), and gillnet fisheries (NMFS unpublished data). It is hypothesized that some strandings may have been caused by interactions with these and other fisheries; however, further observation is necessary to confirm or deny the link between fishery interactions and strandings.
The objectives of this study were to (a) document commercial fishing effort for fisheries that may interact with sea turtles in the Virginia state waters and adjacent federal ocean waters, and (b) use this information to begin to develop an observer sampling plan for these fisheries. The study examined sea turtle strandings data obtained during May through July in 2001-2003. Fishing effort and sea turtle strandings data were examined by region and gear type.
Development of a Suggested Sampling Plan to Allocate Observer Coverage
The coastal Virginia bycatch monitoring program is in the early stages of development. As a first step in developing a suggested sampling plan, we examined the deployment of commercial fishing effort that occurred at the approximate time and areas coincident with turtle strandings. After accounting for logistic constraints, this information was then used to allocate observer coverage proportional to fishing effort.
Times of the year and fishing areas adjacent to shores having high levels of strandings were recommended as priorities for observer coverage. To detect fine scale patterns of sea turtle strandings and commercial fishing activity, our temporal stratification was weekly. Concentrating on areas with high levels of turtle strandings, we restricted our spatial strata to the southern portion of the Chesapeake Bay and to federal waters (i.e., Statistical Areas 625 and 631, Figure 1) adjacent to Virginia.
We next decided which fishing gears should be suggested for observer coverage. Fishing gear with documented or suspected sea turtle interactions in Virginia waters (state and federal) include gillnet, haul seine, scallop trawl, bottom fish trawls, purse seine, pound net, pot gear, and handline. Pound net and pot gear were excluded from analysis here because separate alternative platform sampling programs have been enacted for fisheries using these gears.
Data Sources and Analysis
Data from four sources were used in our analyses:
- Virginia Marine Resources Commission (VMRC)
- National Marine Fisheries Service (NMFS) Vessel Trip Report (VTR)
- Commercial Fisheries Database System (CFDBS)
- Virginia sea turtle strandings data.
VMRC, VTR, and CFDBS report commercial fishery landings by fish species by trip. VMRC records provide data on the catch landed in state waters, while the VTR and CFDBS records provide catch data from both state and federal waters. Most, but not all, of Virginia area fisheries data are represented in the state VMRC and federal VTR data. CFDBS data were used to assess purse seine effort, because this fishery is not included in either the VMRC or VTR data bases. The purse seine was the only fishery for which the CFDBS data were used. Sea turtle strandings data provided information on the turtle species, location of each reported stranding, the date of the stranding, and the condition of the turtle.
To investigate the relationship between sea turtle strandings and fishing effort, the VMRC, VTR, and strandings data were stratified spatially and temporally by the same strata. The temporal stratification comprised one week periods from April 29 to July 30. The spatial stratification used four regions within Chesapeake Bay and six regions outside of the Bay. Fishing locations were identified using latitude and longitude bearings recorded on the VTRs and the Virginia state-defined regions provided in the VMRC records. The four quadrants inside the bay were created by combining a number of the smaller Virginia state-defined regions and include the northeast (NE), northwest (NW), southeast (SE), and southwest (SW) (Figure 1). Quadrants outside the bay encompassed both state and federal waters. Waters outside of the bay and within 3 miles of shore were defined as state waters, and fishing effort in these waters was recorded in both the VMRC and VTR records. We separated state waters at 37° north latitude: north Atlantic (NA) and south Atlantic (SA). Fishing effort in state waters was removed from the VTR summaries to avoid double counting of effort. Fishing effort in federal waters was reported in the VTRs and included Statistical Areas 625, 626, 631, and 632. These statistical areas were also separated at 37° north latitude. Areas 625 and 631 were adjacent to the Virginia shore, had an eastern extent of 75° west longitude, and extended in a north-south direction one degree of latitude (Figure 1). Statistical areas 632 and 626 were the one-degree square eastern counterparts of Areas 631 and 625.
The VMRC and VTR data were further stratified by gear type. Within each stratum (year, week, region, gear type) the number of trips and total landings were summed. Both fishing trips and landings were investigated because trips were not always comparable across gear types. For example, the purse seine fishery had very few trips but a large amount of landings, whereas the handline fishery had numerous trips but comparatively few landings.
It was not possible to stratify purse seine fishing effort by the same strata as the rest of the data because:
- limited fishing location data were available in the CFDBS, and
- trips tended to be longer than one week, making it inappropriate to bin the trips into weeks as done with the other fisheries data.
Instead, purse seine effort was spatially divided into bay and ocean strata, and all recorded trips during May through July in a single year were grouped into one time period representing the study period for that year.
Fishing effort by gear type was examined for patterns consistent with patterns in sea turtle strandings from early May through the last week of June. Correlation analysis was explored but was not included in this report. This and other similar types of statistical analysis were not appropriate because:
- there is an unknown probable temporal lag between a turtle death and its stranding,
- the location of a turtle death may be distant from its stranding location, and
- binned fishing effort may not represent the week of an interaction if a trip spanned weekly periods.
Instead, examination of fishing effort relative to turtle strandings was limited to comparative figures and tables. This analysis focused on the areas where most strandings occurred (SE, SW, NA, and SA areas).
Virginia Sea Turtle Strandings
Loggerhead sea turtles accounted for 81% of all Virginia sea turtle strandings from 2001 to 2003 (Table 1). Kemp’s ridleys accounted for 11% of total strandings, and green, leatherback, and unidentified turtles each constituted less than 4% of total strandings. More than 70% of the yearly strandings in Virginia occurred between May and July, with 49% in the month of June alone.
During the May to July period, strandings consistently peaked across years from early to mid June (Table 2). The vast majority of stranded turtles had body conditions classified as moderately or severely decomposed (category 3 or greater; Swingle et al. 2004). Total strandings for the peak months were greatest in the SE, SW, and NA in decreasing order (Table 2). In two of the three years, the SE had the most strandings followed by the SW, and in 2002 the NA region had the most strandings (Table 3). Strandings along Virginia’s ocean coastline (NA and SA) were also relatively high in some years. Northern regions of the Chesapeake Bay had considerably fewer reported strandings than other regions.
VMRC Fisheries Effort - State WatersDuring the May to July period in 2001-2003, commercial fishery harvests from the Chesapeake Bay and Virginia ocean state waters totaled 14.1, 14.8 and 12.7 million pounds, respectively (Table 4). Six types of gear accounted for 95% of the catch from state waters: pot gear (34%), pound nets (27%), anchored gillnets (13%), haul seines (10%), drift gillnets (6%), and pealer pots (5%). Those same gear types accounted for 85% of the trips: pot gear (45%), pound nets (5%), anchored gillnets (6%), haul seines (1%), drift gillnets (2%), and pealer pots (25%).
State water landings and number of trips were higher inside the Chesapeake Bay (92% to 95% of VMRC totals) than outside (5%-8% of VMRC totals; Table 4 and Table 5). Pot and pound net gear accounted for the majority of landings and trips in the NE and NW areas of the Bay. In the SE and SW areas of the Bay, anchored and drift gillnet, pots, haul seines, and pound nets caught the majority of catch; however, pot gear and pat and tong fishing accounted for the majority of the trips. The major distinction between these two southern regions is that haul seine gear was fished almost exclusively in the SW and accounted for approximately 10% of the total catch in this region. In the NA and SA regions, gillnets and pot gear contributed significant amounts of landings and trips, particularly along the northern shore.
The largest differences between the amount of landings and the amount of effort (e.g. number of trips) occurred in the haul seine, crab pound net, scrap rake, and pat and tong fisheries (Table 4 and Table 5). The haul seine fishery had large landings with few trips, while the other fisheries had low landings and many trips.
VTR Fisheries Effort - Federal Waters
During May through July 2001-2003, the majority (80-85%) of the catch in Federal waters (Figure 2, Figure 3, and Figure 4) was landed in the following gear: scallop dredge, scallop trawl, bottom fish otter trawl, mid-water trawl, and gillnet gear (in Statistical Areas 625, 626, and 632). Area 631 contributed little to the overall catch. However, these same gear types only comprised 40% of all trips. Handline accounted for an additional 35% of all trips, but only 2% of landings (Table 6 and Table 7).
The bottom otter trawl fishery had the highest landings in 2001 and 2003 when it accounted for 40% and 57%, respectively, of the total annual landings. In 2001 and 2002, the majority of landings were in Area 626 (83% and 87%), while in 2003 the majority of landings were split between Areas 626 (48%) and 632 (43%; Table 6).
Scallop dredge gear was second in terms of landings during 2001-2003. Scallop dredge and trawl effort were both primarily located in offshore Area 626 (Table 6 and Table 7; Figure 2, Figure 3, and Figure 4).
Most gillnet gear was fished in Area 625, although the number of gillnet trips in Area 626 increased in 2003 relative to previous years (Table 6 and Table 7; Figure 2, Figure 3, and Figure 4). Gillnets accounted for between 4 and 10% of the study area’s annual landings during 2001-2003 (Table 6). The majority of landings from Area 625 in all three years was from gillnets, and Area 625 had the most gillnet trips in all years except 2002 (Table 6 and Table 7).
Large differences were detected between the amount of trips and landings, by gear type. This scenario applies to handlines, sink gillnet, and fish pots. In these fisheries, total number of fishing trips may be a more appropriate measure of activity than total amount of catch.
Purse Seine Effort - Commercial Fisheries Database System
The purse seine fishery was the only fishery assessed using the CFDBS data because landings from this fishery are not recorded in the VTR or VMRC databases. Purse seine fishing in the study region and time period consisted of a limited number of multi-day trips, but a large amount of landings (Table 8). During 2001-2003, 88% of purse seine landings occurred inside Chesapeake Bay.
Sea Turtle Strandings Relative to Fisheries Effort
There were no consistent patterns in fishing effort that corresponded to peaks in turtle strandings. Although gillnet landings often peaked at the same time as peaks in strandings (Figure A1, panel 3) or just before peaks in strandings (Figure A2, panels 2 and 3), this was not a consistent phenomenon. Similar patterns are also seen among gillnet trips (Figure A11, panel 2, Figure A12, panel 3, and Figure A15, panel 2), though again, this was not consistent. As discussed previously, analyses beyond illustrating the basic patterns in the data were not possible due to probable spatial and temporal inaccuracies in the data.
Development of Sampling Plan to Allocate Observer Coverage
Three different plans for allocating observer coverage were developed based on varying levels (and cost) of observer effort (Table 9). These plans differed in the sampling period and number of sampling days, but all allocated observer effort in proportion to fishing effort at the 10% level. Each plan sampled gillnets, haul seines, scallop trawls, and fish trawls. Gillnets and haul seines are included for sampling in Virginia state waters, and gillnets, scallop trawls and fish trawls are proposed for sampling in federal waters.
Plan 1 deploys observers during May through July and requires 205 sea days. Plan 2 has observer coverage from mid-May to the end of June and requires 121 sea days. Finally, Plan 3 deploys observers only in June (the peak month for strandings) and requires 69 sea days.
Observer coverage in the gillnet fishery accounts for most of the proposed sampling days (87% in plans 1 and 2 and 90% in plan 3). More than 89% of the proposed sampling days in the Chesapeake Bay and state waters are in the gillnet fishery, and about 60% of the observer coverage in federal waters is in the gillnet fishery. Sampling inside the Bay accounts for about 60% of all sampling; sampling in state waters account for about 25%; and sampling in federal waters accounts for between 10% and 14% of the total proposed observer coverage. These percentages vary slightly by plan because of different time periods are covered, but sampling days are largely distributed in similar proportions across all plans.
Plan 1 would observe both the peak and tails of the turtle stranding distribution, while providing some cushion if the peak shifts in time. Plans 2 and 3 cover 82% and 75%, respectively, of the historical turtle bycatches during the May-July time period. However, for the fisheries being observed, only 58% and 32% of the effort occurs during the time periods of plans 2 and 3, respectively. Plans 2 and 3 reduce observer cost but may miss the peak turtle strandings period due to inter-annual variability of turtle residence in the area.
Sea turtles typically arrive in the Chesapeake Bay in late May and early June (Mansfield et al. 2002). Patterns in fishing effort and strandings data suggest that specific possible fisheries should be systematically sampled for evidence of interactions. Documentation of the fishing effort during the general time and area of high levels of stranded turtles provides a starting point to develop a systematic sampling plan; however, it is only a starting point for several reasons. First, stranding patterns and fishing activity were not consistent across the three years examined. Second, the commercial fishery landings data were binned by week according to when a vessel returned to the port and landed its catch. This binning may have obscured the relationship between fishing effort and strandings as fish caught at sea in one week could have been landed the following week. Third, a turtle interaction may occur in one week, and the turtle could then float a great distance and land on a distant beach a week or two later far away from the time and location of the actual fishery interaction. For example, a turtle interaction could occur along the southern Atlantic coast of Virginia and be carried by tides into the Bay. Finally, several fisheries/gears may contribute to the pattern of sea turtle strandings, with no single fishery responsible for the majority of the strandings.
These difficulties illustrate the need for observer data. Using observer data to examine sea turtle-fishery interactions would be far more straightforward. Commercial fishing data would not have to be binned since the exact location and time of the interaction would be known. Also, observing an interaction would take away the guesswork and inaccuracies involved in trying to estimate when and where a stranded turtle may have interacted with a particular fishery.
This study would not have been possible without the diligent data collection by the Virginia Sea Turtle Strandings Program administered by the Virginia Institute of Marine Science. We thank all the volunteers who collected these data and the organizations that contributed to the data collection such as the Virginia Aquarium and Marine Science Center. Our thanks also go to the fishermen who reported their fishing effort, to the Virginia Marine Resources Commission for collecting and sharing their fishing data, and to the NOAA staff who made the VTR and CFDBS data available for analysis. Without these efforts in data collection and preparation, the analyses in this report would not have been possible.
Mansfield KL, Seney EE, Fagan MA, Musick JA, Frisch KL, Knowles AE. 2002. An evaluation of interactions between sea turtles and pound net leaders in the Chesapeake Bay, Virginia. Report to NMFS, Contract #: EA1220-02-SE-0075. Virginia Institute of Marine Science; 118 p.
Mooreside PD. 2000. Integration of physical oceanography with spatio-temporal patterns of stranded sea turtles in North Carolina. Durham [NC]: Duke University, master’s thesis. 56 p.
Murray KT. 2006. Estimated average annual bycatch of loggerhead sea turtles (Caretta caretta) in US Mid-Atlantic bottom otter trawl gear, 1996-2004. Northeast Fisheries Science Center Reference Document 06-19; 34 p.
Murray, KT. 2005. Total bycatch estimate of loggerhead turtles (Caretta caretta) in the 2004 Atlantic sea scallop (Placopecten magellanicus) dredge fishery. Northeast Fisheries Science Center Reference Document 05-12; 22 p.
National Marine Fisheries Service. 2002. Sea turtle conservation measures for the pound net fishery in Virginia waters of the Chesapeake Bay (F/ER/2002/00438). Endangered Species Act Section 7 Consultation Biological Opinion; 83 p.
National Oceanic and Atmospheric Administration. 2004. Evaluating bycatch: a national approach to standardized bycatch monitoring programs. US Dep Commer, NOAA Tech Memo NMFS-F/SPO-66; 108 p.
Swingle MW, Barco SG, Harry CT, Trapani C. 2004. Sea turtle stranding response on Virginia’s Eastern shore. Virginia Aquarium and Marine Science Center Scientific Report 2004-03; 24 p.
 Examples of alternative designs are observing all components of the fishery every few years and not at all in between, or observing only some components of the fishery every year with the goal that all components of the fishery are observed every few years.
 Although this sampling plan does not include gillnet and scallop dredge fisheries in Areas 626 and 632, other NMFS programs (not described in this paper) have allocated observer coverage to this portion of the fishery.
 The pound net alternative platform sampling program requires an experienced observer to operate a vessel independent of the fishing vessel to check for turtles caught in the pound leader. The second platform provides observer safety and improved observation abilities. In the case of pot gear, it is possible that if an interaction has occurred, the turtle will unlikely be seen when the pot is hauled out of the water. Estimating bycatch in this fishery may be best achieved through an alternative platform approach using sonar and a diver to check for bycatch before the pot is pulled from the water.
 Given numerous unknowns after a turtle death (e.g.. buoyancy of a dead turtle over time, tidal currents, large scale oceanographic currents, wind-driven movement, etc.; Mooreside 2000), it is extremely difficult to ascertain where a stranded turtle died.
 VTR effort analyzed in this paper is likely to be a slight underestimate of the true fishing effort. Some VTR records had missing latitude and longitude information, and therefore these records were not included in the analysis since it could not be determined where the fishing effort occurred. The VTR data for the Northeast region (US Atlantic waters north of North Carolina), had missing location information on 4.8%, 4.7%, and 5.25% of all recorded trips for 2001, 2002, and 2003, respectively. VTR data that was recorded as landed in Virginia had missing location information for 7.9%, 8.4%, and 11% of recorded trips for 2001, 2002, and 2003, respectively. Since records were selected by location instead of state, fishing effort within the study region could have been landed in other ports outside of Virginia. Therefore, the percentage of missing VTR location information is likely somewhere between the percentages for the whole Northeast VTR dataset and the percentages for Virginia.
 If interactions are found to be gear-specific in state waters, the same interactions are likely to occur in sections of the Bay not analyzed and sampling methods could be transferred to these regions.
 Strandings data are considered opportunistic data. The Virginia sea turtle strandings network systematically surveys various sections of the coastline for strandings, but other areas are surveyed irregularly and strandings are often reported to the stranding network by the public. Thus, effort across shorelines is not standardized and differences in the timing of reported strandings could be due to the amount of public visitation to a particular shoreline; that is, the NW may have a lower level of strandings than other regions in part because less effort (or visitation) takes place in this region. Also, the date when a turtle is found may not represent the date on which it washed up on shore, and the turtle’s condition may depend on how long it has been on land before it was detected.
 One observer sea day cost approximately $1,150 in FY2006. Therefore, 121 sea days requires a budget of $139,150.
Appendix Table A1. 2001 May-July VMRC trips and turtle strandings by week and water body Table A2. 2002 May-July VMRC trips and turtle strandings by week and water body Table A3. 2003 May-July VMRC trips and turtle strandings by week and water body Table A4. 2001 May-July VTR trips and turtle strandings by week and water body Table A5. 2002 May-July VTR trips and turtle strandings by week and water body Table A6. 2003 May-July VTR trips and turtle strandings by week and water body Figure A1. Southeast region 2001-2003 VMRC landings by gear type and turtle strandings Figure A2. Southwest region 2001-2003 VMRC landings by gear type and turtle strandings Figure A3. North Atlantic region 2001-2003 VMRC landings by gear type and turtle strandings Figure A4. South Atlantic region 2001-2003 VMRC landings by gear type and turtle strandings Figure A5. Statistical area 625 2001-2003 VTR landings by gear type and turtle strandings Figure A6. Statistical area 631 2001-2003 VTR landings by gear type and turtle strandings Figure A7. Statistical area 626 2001-2003 VTR landings by gear type and turtle strandings Figure A8. Statistical area 632 2001-2003 VTR landings by gear type and turtle strandings Figure A9. Southeast region 2001-2003 VMRC trips by gear type and turtle strandings Figure A10. Southwest region 2001-2003 VMRC trips by gear type and turtle strandings Figure A11. North Atlantic region 2001-2003 VMRC trips by gear type and turtle strandings Figure A12. South Atlantic region 2001-2003 VMRC trips by gear type and turtle strandings Figure A13. Statistical area 625 2001-2003 VTR trips by gear type and turtle strandings Figure A14. Statistical area 631 2001-2003 VTR trips by gear type and turtle strandings Figure A15. Statistical area 626 2001-2003 VTR trips by gear type and turtle strandings Figure A16. Statistical area 632 2001-2003 VTR trips by gear type and turtle strandings