CONTENTS Executive Summary Introduction Methods Results & Discussion Conclusions Acknowledgments References Appendix 1
Northeast Fisheries Science Center Reference Document 09-20
River Herring Discard Estimation, Precision, and Sample Size AnalysisSE Wigley, J Blaylock, and PJ Rago
NOAA's National Marine Fisheries Service, 166 Water St, Woods Hole MA 02543
Web version posted January 15, 2010Citation: Wigley SE, Blaylock J, Rago PJ. 2009. River Herring Discard Estimation, Precision, and Sample Size Analysis. NEFSC Ref Doc 09-20; 15 p. Available from: National Marine Fisheries Service, 166 Water Street, Woods Hole, MA 02543-1026, or online at http://nefsc.noaa.gov/nefsc/publications/
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.
A total of 106,455 pounds (149% CV) of river herring was discarded during the 12-month period from July 2007 through June 2008. The river herring discards occurred primarily in the New England small mesh otter trawl fleet, and to a lesser extent in the New England shrimp trawl, New England large mesh trawl and Mid-Atlantic small mesh otter trawl fleets.
The total number of sea days (9,002 days) needed to achieve a 30% CV for river herring discard estimates is 40% less than those needed for the 15 Standardized Bycatch Reporting Methodology (SBRM) species groups (15,125 days). On a fleet by fleet basis, the number of sea days needed to achieve a 30% CV for river herring discard estimates was either the same or less than for the 15 SBRM species groups with the exception of one fleet: New England large mesh otter trawl where the river herring species group needed 2,827 more sea days than the 15 SBRM species groups.
The low number of observed trips in the Mid-Atlantic gillnet fleets and the limitations of observing all discards associated with the fishing practices of the high-volume fisheries are two sources of uncertainty in the discard estimation and sample size analysis.
Concerns have been expressed by governmental and non-governmental groups regarding the discarding of river herring by commercial fishing fleets operating off the northeastern coast of the United States. River herring is a species group comprising two individual species: alewife, Alosa pseudoharengus, and blueback herring, Alosa aestivalis. This document provides estimates of river herring discards, with associated precision, and the number of sea days needed to achieve discard estimates over a range of precision levels. This document then compares the number of sea days needed for river herring with sea days needed for the 15 Standardized Bycatch Reporting Methodology (SBRM) species groups for the same time period derived in the 2009 SBRM analysis. The SBRM is an omnibus amendment to the 13 fishery management plans in the Northeast region that establish review and reporting requirements for bycatch monitoring and observer coverage as well as a process for prioritizing observer coverage when there is a shortfall in funding (NMFS 2007).
River herring is managed by the Atlantic States Marine Fisheries Commission (ASMFC). Background information on river herring is available on the ASMFC web site: http://www.asmfc.org/
The 2009 SBRM analysis data set was expanded to include the river herring species group. This data set includes data from the Northeast Fisheries Observer Program (NEFOP), the Vessel Trip Report (VTR, including logbooks from the surfclam and ocean quahog fishery) database, the Northeast Fisheries Science Center (NEFSC) commercial landings database, and the NOAA Fisheries Marine Recreational Information Program (MRIP) recreational landings from July 2007 through June 2008.
This analysis used the discard estimation methods described in the SBRM analysis (Wigley et al. 2007). Total discards of river herring from July 2007 through June 2008 were estimated using a combined d/k ratio estimator (Cochran 1963), where d = discarded pounds of river herring and k = the kept pounds of all species. Total discards (in weight) of river herring, were derived by multiplying the estimated discard rate of each fleet by the corresponding fleet landings in the VTR database, and then summing over fleets. In this report, the term coefficient of variation (CV) is defined as the ratio of the standard error of the total discards divided by the total discards. Appendix 1 presents the equations used in the analysis. We assumed 100% mortality for all discarded fish.
In the analysis, the sampling unit was a fishing trip. The same broad stratification scheme used in the SBRM was employed, where trips were partitioned into fleets using six classification variables: calendar quarter, geographic region, gear type, mesh, access area, and trip category. Calendar quarter was based on landed date and used to capture seasonal variations in fishing activity and discard rates. Two broad geographical regions were defined: New England (NE) and Mid-Atlantic (MA) based on port of departure; ports located from Maine to Rhode Island constituted the NE region and ports located in states from Connecticut southward constituted the MA region. Gear type was based on Northeast gear codes (negear). Some gear codes were combined: sink, anchored and drift gillnets, and single and paired mid-water trawls, (Table 1), and trips for which gear was unknown were excluded. Mesh size groups were formed for otter trawl and gillnet gear types. For otter trawls, two mesh groups were formed: small (mesh less than 5.5 inches) and large (5.5 inch mesh and greater). For gillnets: three mesh groups were formed: small (mesh less than 5.5 inches); large (mesh between 5.5 and 7.99 inches) and extra large (mesh 8 inches and greater). Two access area categories were formed: access area (AA) and open (OPEN). Trips participating in the US/Canada access area, ‘B-day’ category and other quota monitoring programs could not be identified in the VTR database, and hence these trips were grouped by the other stratification variables and therefore not partitioned separately. The sea scallop fishery was divided into General (GEN) and Limited (LIM) category trips. All other fisheries were combined into a category called ‘all’.
Of the 44 fleets examined in the analysis, 22 fleets had little or no observer data (all quarterly cells were missing for a fleet, or sparse observer coverage existed across all quarters for a fleet). No discard estimation was performed for these fleets, and all 22 fleets were designated as in need of pilot coverage (Table 1). Pilot coverage is defined as a minimum level of observer coverage to acquire bycatch information with which to calculate variance estimates that in turn can be used to further define the level of sampling needed (NMFS 2004). For each of the remaining 22 fleets, estimates of river herring discards were derived. Eight fleets had limited observer data (i.e., one or no observed trips in a calendar quarter), and an imputation approach was used to ‘fill in’ the missing (or incomplete) information using data from an adjoining stratum. In this imputation procedure, only the temporal stratification (i.e., calendar quarter) was relaxed to half year or entire year, recognizing that seasonal variations may occur that will thus not be accounted for. For 14 fleets, sufficient observer data existed to estimate discards without imputation (Table 1).
Sample Size Analysis
A sample size analysis was conducted to estimate the number of baseline trips and sea days needed to monitor river herring discards in each fleet. As described in the SBRM analysis (and given in Appendix 1), the number of trips and sea days needed to achieve a given precision level was based on the variance of the total river herring discard estimates. Sample size (trips and sea days) associated with the SBRM precision standard for discard estimates (30% CV) were derived along with 5 other precision levels: 10%, 20%, 40%, 50% and 60% CV. The sample size analysis was performed using trips as the sampling unit of the analysis, and then converting the number of trips to sea days by multiplying by the weighted mean trip length (Table 2) - where the weighting factor was the quarterly number of VTR trips.
When total discards could not be estimated due to little or no observer coverage (no data) or when total discards were zero (no variance), the sample size (number of trips) was determined using a pilot coverage level set to 2% of the quarterly VTR trips for a fleet, with a minimum of 3 trips per quarter (12 trips per year) and a maximum of 100 trips per quarter (400 trips per year). The 2% pilot coverage level is the same as used in the 2009 SBRM analysis. The quarterly trips were then multiplied by the quarterly mean VTR trip length to derive quarterly sea days. The quarterly trips and quarterly sea days were then summed for annual number of trips and sea days. The pilot coverage may result in too much coverage in cases where little or no observer coverage may actually be needed.
Importance filters were used in SBRM to provide a standardized protocol to further refine the number of baseline sea days based on (a) the importance of the discarded species relative to the total amount of discards by a fleet, and (b) the total fishing mortality due to the discards. Three filters (i.e., unlikely cell filter; fraction of discard filter; and fraction of mortality due to discards filter) are applied simultaneously. The unlikely cell filter eliminates sea days associated with fleets where species and gear combinations are unlikely or infeasible. The unlikely cell filter can act as an ‘override’ mechanism in situations where pilot coverage is evoked due to no variance (observer coverage indicates zero discards). A detailed description of the SBRM importance filters is given in Wigley et al. (2007). As in the 2009 SBRM analysis, the baseline sea days for river herring were filtered using a 95% cut-point in the discard mortality filter, and a 98% cut-point for the total mortality filter due to discards. The unlikely cell filter was not updated to include the river herring species groups as all fleets were considered ‘likely’ to discard river herring.
River herring discards were estimated for 22 of the 44 fleets (Table 2). Of the 22 fleets with discard estimates, 6 fleets had discard estimates greater than zero and 16 fleets had discard estimates equal to zero. A total of 106,455 pounds of river herring was discarded during the 12-month period from July 2007 through June 2008. By fleet, discards of river herring ranged between 0 and 95,744 pounds, with largest amount occurring in the New England small mesh otter trawl fleet. The coefficient of variation of the total river herring discard estimate was 149%. For the six fleets where estimated river herring discards were greater than zero, the coefficient of variation ranged between 31% and 166% (Table 2).
Sample Size Analysis
Sample size coverage levels (baseline trips and sea days) to assess river herring discards are presented in Table 2. To achieve a 30% CV on river herring discards, a total of 6,923 baseline trips and 15,395 baseline sea days would be needed (Table 2). In six fleets, sample sizes (5,317 trips and 12,107 days) were based on the variance of total discards, while sample sizes for 38 fleets (1,606 trips and 3,288 days) were based on pilot coverage. The sample size based on pilot coverage (due to no data or no variance) contributed 23% and 21% to the total baseline trips and sea days, respectively. By fleet, the total number of baseline trips ranged between 6 and 1,662 trips while the total number of baseline sea days ranged between 10 and 4,060 days (Table 2).
For the six fleets where estimated river herring discards were greater than zero, the number of baseline trips and sea days needed to estimate discards over a range of precision levels are presented in Tables 3a and 3b and Figure 1. The sample sizes reported in Tables 3a and 3b are subtotals associated with the six fleets. To obtain a grand total, the subtotal must be added to the pilot coverage subtotal given in the text above. For example, to achieve a 20% CV based on river herring discards, a total of 11,423 baseline trips [9,817 trips (Table 3a) + 1,606 trips based on pilot coverage] and 25,454 days [22,166 days (Table 3b) + 3,288 days based on pilot coverage] would be needed.
The importance filter eliminated the sea days associated with the 16 fleets with zero discards of river herring. As well, three other fleets (MA large mesh otter trawl, MA small mesh otter trawl, and New England mid-water trawl) were eliminated from coverage because these fleets contributed to less than 5% of total discards and less than 2% of the total mortality due to discards. The total number of filtered sea days needed to achieve a 30% CV is 9,002 days (Table 2). By fleet, the number of filtered sea days needed to achieve a 30% CV for the total discards of river herring ranged between 0 and 4,060 days (Table 2).
The total number of filtered sea days needed for river herring may be an overestimate due to the inclusion of sea days associated with fleets that would have been eliminated had the unlikely cell filter been updated for river herring. The sea days associated with the clam dredge fleets, handline fleets, and the fish, conch, lobster, and crab pot fleets would most likely have been eliminated as these gear types rarely capture river herring. The total number of filtered sea days needed for river herring will be improved when observer data are available for all the fleets (no sea day estimates based on pilot coverage) and the unlikely cell filter is updated for this species group. In 2011, a three-year evaluation of the SBRM will occur and, at that time, bycatch information and estimation methods will be evaluated including an evaluation of the unlikely cell filter for all species.
This analysis used the SBRM Omnibus Amendment cut-points of 95% and 98% to eliminate sea days associated with fleets that contribute the smallest amount to total discards and total mortality due discards for river herring, respectively. Table 4 reveals that had another cut-point been used for the total discard filter, for example 93% instead of 95%, eliminating fleets contributing to the lower 7% of total discards instead of the lower 5% of total discards, then the 4,060 days associated with the NE large mesh otter fleet would have been filtered out and the total number of filtered sea days for river herring would have been 4,942 days instead of 9,002 days.
River herring is not a species group included the SBRM analysis, thus the number of sea days needed to monitor the river herring species group can be compared with the 15 SBRM species groups. A comparison of the filtered sea days for river herring and the 15 SBRM species groups (Table 2) reveals fewer total sea days would be needed to monitor only river herring (Table 2). However, on a fleet basis, only the NE large mesh otter trawl fleet needed more sea days for river herring than for the 15 SBRM species groups. As noted above, if the SBRM cut-point for total discards was relaxed to 93% and eliminated the sea days associated with the lower 7% of the total river herring discards, then the sea days from the NE large mesh otter trawl would be filtered out and the total number of sea days needed to monitor the 15 SBRM species groups would be greater than the sea days needed to monitor river herring for all fleets.
The total number of sea days needed for the 15 SBRM species groups (15,125 days) and for river herring (9,002 days) exceed the total number of sea days that the Northeast Fisheries Science Center (NEFSC) has funding to support. The NEFSC has funds to support a total of 6,283 days [4,283 days (Table 2) plus 1,940 US/Canada Quota-Monitoring sea days and 60 Discovery days; NEFSC 2009b]. When a funding shortfall exists, the SBRM Sea Day Prioritization Process (NEFMC 2007; NMFS 2007; NEFSC 2009b) is invoked; the resulting 2009 sea days, by fleet, are given in Table 2.
Misreporting of gear code in the VTR database is evident for the NE shrimp trawl fleet. The seasonal nature (December to April) of the northern shrimp fishery contributes to the inadvertent misreporting of gear code (OTS = otter trawl, shrimp and OTF = otter trawl fish) by fisherman who switch between these two fisheries. Thus caution is warranted when interpreting results regarding the NE shrimp trawl fleet. There is a need for improved data auditing and the identification of Special Access Program trips within the VTR database. Additionally, the low number of observed trips in Mid-Atlantic gillnet fleets contribute to uncertainty in the river herring discard estimates. Increased observer coverage will confirm the extent to which these fleets encounter river herring.
From 2005 onwards, the NEFOP has informally collected information to characterize unique fishing practices (fish pumping, use of sorting grates at the pump entrance and on deck, and net release) of high-volume fisheries like the paired and single mid-water trawl fleets. In 2010, the NEFOP manual describing data collection will be formally expanded to include sampling protocols designed to systematically characterize the fishing practices of high-volume fisheries. It is important to emphasize that only information systematically collected has been used in the current discard estimation analysis. Thus, discards associated with underwater net release or those associated with sorting grate at the pump entrance have not been included. Therefore, the discard estimates presented here may underestimate river herring discards associated with the high-volume fisheries. In the future, discard estimation methods may be able to incorporate the expanded data collection information and summarize the frequency of underwater net releases.
River herring is a non-targeted species group captured in various commercial trawl fleets. The 2009 SBRM Annual Discard Report (NEFSC 2009a) summarizes where and when (statistical area and calendar quarter) observers have recorded the disposition of catch (kept or discarded) and the weight (pounds) of alewife and blueback herring among the commercial fleets (NEFSC 2009a, Section II, Tables 4c and 5c). Some trawl fleets that encounter river herring are retaining these species while other fleets are discarding them. Generally, the mid-water trawl fleets are retaining river herring, while the otter trawl fleets are not.
Nominal catch (landings only) estimates of river herring during 2005 to 2007 have been derived by Cieri et al. (2008). In their study, a ratio estimator of observed river herring bycatch (kept) to Atlantic herring landings was developed based upon NEFOP observer data, Maine Department of Marine Resource portside sampling data, and VTR data. A stratification scheme was also used that included: year, calendar quarter, area, and gear type. Estimates of the annual nominal catch of river herring during 2005-2007 were 285,833, 171,973 and 1,686,617 pounds, respectively. It is important to note that the Cieri et al. (2008) estimates are for nominal catch of river herring as a non-targeted (hence their use of the term ‘bycatch’) species in four commercial fishing fleets. In this context, the ‘bycatch’ estimates are not equivalent to the discard estimates derived in the present analysis. In the present analysis, discards represent the component of total catch that are not retained; total catch = landings (target and non-target) + discards.
This analysis indicates that during the July 2007 through June 2008 time period, river herring were primarily discarded in the NE small mesh otter trawl fleet, and to a lesser extent in the NE shrimp trawl, NE large mesh trawl and MA small mesh otter trawl fleets.
The baseline sea days needed to attain a 30% CV for discard estimates of river herring are slightly higher than the total filtered sea days needed to monitor 15 SBRM species groups. When the importance filter is applied to the river herring species group, the sea days associated with 19 of the 22 fleets with estimated discard were eliminated, indicating that the discarding of river herring is a minor component of total discards and total catch of river herring in the majority of fleets. The total number of filtered sea days needed to monitor river herring is 40% less than those needed to monitor the 15 SBRM species groups. On a fleet by fleet basis, the number of filtered sea days needed to monitor river herring was either the same (due to pilot coverage) or less than for the 15 SBRM species groups with the exception of one fleet: NE large mesh otter trawl where the river herring species group needed 2,827 more sea days than the 15 SBRM species groups.
The low number of observed trips in the Mid-Atlantic gillnet fleets and the limitations of observing all discards associated with the fishing practices of the high-volume fisheries are two sources of uncertainty in this analysis. More observer coverage, particularly in the Mid-Atlantic region, is needed to derive discard estimates and sample sizes. Direct discard estimation will reduce the dependency on pilot coverage for sample sizes estimates.
The discard estimation used a broad stratification approach designed to encompass all species groups considered in the SBRM analysis. The discard estimates reported here may not necessarily correspond directly with the discard estimates derived specifically for river herring due to differences in stratification. It is expected, however, that estimates would be in the same order of magnitude.
We wish to thank all the NEFOP observers for their diligent efforts to collect the discard information used in this analysis. We thank Jeff Hansen for his technical assistance and our reviewers for their helpful comments on this report.
Cieri, M, G. Nelson, and M. Armstrong. 2008. Estimates of river herring bycatch in the directed Atlantic herring fishery. Report prepared for the Atlantic States Marine Fisheries Commission, Washington, DC. September 23, 2008. 17 p.
Cochran, W.L. 1963. Sampling Techniques. J. Wiley and Sons. New York.
National Marine Fisheries Service (NMFS). 2004. Evaluating bycatch: a national approach to standardized bycatch monitoring programs. U. S. Dep. Comm., NOAA Tech. Memo. NMFS-F/SPO-66, 108 p. On-line version, http://www.nmfs.noaa.gov/by_catch/SPO_final_rev_12204.pdf
National Marine Fisheries Service (NMFS). 2007. Magnuson-Stevens Fishery Conservation and Management Act Provisions; Fisheries of the Northeastern United States; Northeast Region Standardized Bycatch Reporting Methodology Omnibus Amendment. Federal Register, Vol. 73, No. 18, Monday, January 28, 2008. pages 4736-4758. Available on-line at:http://www.gpoaccess.gov/fr/retrieve.html
New England Fishery Management Council (NEFMC), Mid-Atlantic Fishery Management Council and National Marine Fisheries Service. 2007. Northeast Region Standardized Bycatch Reporting Methodology: An Omnibus Amendment to the Fishery Management Plans of the New England and Mid-Atlantic Fishery Management Councils. June 2007. 642p. Available on-line at: http://www.nefmc.org/issues/sbrm/
Northeast Fisheries Science Center (NEFSC). 2009a. 2009 Standardized Bycatch Report Methodology Annual Discard Report (Section I and II). Presented in the NEFMC and MAFMC. February 2009. 1560 p. Available on-line at: http://nefsc.noaa.gov/femad/fishsamp/fsb/
Northeast Fisheries Science Center (NEFSC). 2009b. Standardized Bycatch Reporting Methodology Proposed 2009 Observer Sea Day Allocation Consultation and Prioritization Process Response to Comments from New England and Mid Atlantic Fishery Management Councils and the Northeast Regional Office. Presented to Northeast Regional Coordinating Committee 1 April 2009. 15 p. Available on-line at: http://nefsc.noaa.gov/femad/fishsamp/fsb/
Wigley S.E, P.J. Rago, K.A. Sosebee, and D.L. Palka. 2007. The analytic component to the Standardized Bycatch Reporting Methodology Omnibus Amendment: sampling design and estimation of precision and accuracy (2nd edition). U.S. Dep. Commer., Northeast Fish. Sci. Cent. Ref. Doc. 07-09; 156 p. Available on-line: http://nefsc.noaa.gov/publications/crd/crd0709/