Minimum criteria for sampling young-of-year (YOY) American eel were established by the Atlantic States Marine Fisheries Commission (ASMFC) in their Fishing Management Plan (FMP) for the species (download pdf), with the Technical Committee approving sampling gear and methods. The timing and placement of gear must coincide with periods of peak YOY shoreward migration. At a minimum, the gear must fish during flood tides during nighttime hours. The sampling season is designated as a minimum of 4 days per week for at least 6 weeks, or for the duration of the run. At least 1 site must be sampled in each jurisdiction. The entire catch of YOY eels must be counted from each sampling event and a minimum of 60 glass eels (if present per system) must be examined for length, weight, and pigmentation stage weekly.
Due to the importance of the eel fishery in Virginia, the methods used must ensure proper sampling coverage in space and time, and provide reliable recruitment estimates. To provide the necessary spatial coverage and to assess suitable locations, numerous sites were evaluated previously. Final site selection was based on known areas of glass eel concentrations, accessibility, and specific physical criteria (e.g., proper habitat) suitable for glass eel recruitment to the sampling gear.
Four sites were selected: 2 on the York River and 1 each on the Rappahannock and James rivers. The 2 sites on the York River are Bracken's Pond and Wormley Pond. Bracken's Pond is located along the Colonial Parkway at the base of the Yorktown Naval Weapons Station Pier and is less than 100 meters from the York River; the tide often reaches the spillway. This site was chosen as a primary site in 2000 with gear comparisons performed throughout the sampling season. Wormley Pond is located on the Yorktown Battlefield and drains into Wormley Creek, which has a tidal range that routinely reaches 50 centimeters in depth at the spillway. This site was not sampled in spring 2000. Kamp's Millpond drains into the eastern branch of the Corrotoman River, a tributary to the Rappahannock River. Kamp's Millpond covers approximately 80 acres and is located upstream of Route 790, just north of Kilmarnock. The final collection site on the James River is Wareham's Pond, which is located in the Kingsmill area of James City County. Wareham's Pond drains directly into the James River, which is about 100 meters away, though high tides may reach the end of the spillway.
Irish Eel Ramps
Irish eel ramps were used to collect eels at all sites. The ramp configuration successfully attracts and captures small eels in tidal waters of Chesapeake Bay. Ramp operation requires a continuous flow of water over the climbing substrate and the collection device; continuous flow was accomplished through a gravity feed. Hoses were attached to the ramp and collection buckets to allow for quick removal of eels for sampling. EnkamatTM erosion control material on the ramp floor provided a textured climbing surface. The ramps were placed on an incline (15-45°) with the ramp entrance and textured mat extending into the water. The ramp entrance was placed in shallow water (< 25 cm) to prevent submersion of the entire ramp. The inclined ramp and an additional 4o incline of the substrate inside the ramp provided sufficient slope to create attractant flow. A hinged lid provided access for cleaning and flow adjustments.
Only eels in the ramp's collection bucket (not on the climbing surface) were recorded. Trap performance was rated on a scale of 0 to 3 (0 = new set; 1 = gear fishing; 2 = gear fishing, but not efficiently; 3 = gear not fishing). Water temperature, air temperature, and precipitation were recorded during most site visits. All eels were enumerated and placed above the impediment, with any subsample information recorded, if applicable. Specimens less than or equal to ~ 85 mm total length (TL) were classified as YOY, while those > 85 were considered elvers. These lengths correspond to the two distinct length-frequency modes observed in the 2000 survey, which likely reflects differing year classes (Geer 2001; note: eels longer than 254 mm TL are considered yellow phase eels, although this is not explicitly stated in Geer 2001). Length, weight, and pigmentation stage (see Haro and Krueger 1988) were recorded from 60 eels weekly.
Index Calculation Procedure
A review of the index calculation procedure was undertaken in 2009 to investigate the use of the geometric mean catch for days during which 95% of the glass eels were captured. The rationale for the review was based on an observation concerning data from a glass eel monitoring site on the Potomac River (Gardy's Millpond; Tuckey and Fabrizio 2009). In 2000, at Gardy's Millpond, 291 glass eels were collected, of which 262 were used to calculate the 95% geometric mean index (18.3); in 2009, 231 glass eels were collected, of which 223 were used to calculate the 95% geometric mean index (1.6). The actual difference in numbers of glass eels used in the calculation is 39 (counting only those eels captured during the 95% recruitment window) and a difference of 54 days of effort, but the index in 2000 is 11 times greater than that in 2009. Is the index obtained by the 95% geometric mean method affected by daily fluctuations in recruitment when effort is "adjusted" by the 95% cut-off value? To answer that question, a theoretical analysis was conducted for three possible recruitment patterns and resultant indices were compared for: (1) a single peak recruitment event, (2) constant recruitment throughout the sampling period, and (3) episodic recruitment exhibiting multiple peaks during the sampling period. For this analysis, effort was constant and equal to 30 trap days and the total number of eels arriving during the recruitment period was 1,000 glass eels for each recruitment scenario. Three recruitment indices were calculated:
- the simple, arithmetic average over the time period sampled,
- the geometric mean using the 95% cut-off, and
- the area-under-the-curve (AUC; Olney and Hoenig 2001).
If the arithmetic average is used to calculate the index, all three recruitment patterns yield the same index value -- 33.3 because the total number of glass eels captured and total effort are the same. One problem with using the average as an index of abundance for glass eels is that catches do not follow a normal distribution (a necessary assumption), and thus, this measure of central tendency may not accurately reflect ‘average' conditions during the recruitment period. Furthermore, this approach requires adoption of constant effort year after year; if effort changes, then the index value may change as well. For example, adding a single week of sampling during which no eels are captured will reduce the average (index) to 27.0 in this example. Targeting the timing of sampling to coincide with recruitment for a species that migrates from the continental shelf and exhibits yearly fluctuations in timing is difficult, if not impossible. Timing of recruitment may vary due to water temperature, wind patterns or other factors that are not predictable and a fixed period of sampling may miss recruitment of glass eels if ingress occurs earlier or later than expected.
Indices based on the 95% geometric mean differ markedly for the three recruitment scenarios and range from 29.4 (episodic pattern) to 300.0 (peak pattern). The reason for this variation is that the number of zero catches included in the calculation depends on the recruitment pattern even though eliminating 5% of the low catches attempts to reduce that influence. If daily recruitment patterns do not change appreciably among years, then the 95% method for index calculation will work as expected. However, if recruitment patterns change each year such that in one year, glass eels arrive in a single week but the following year, eels trickle in over a period of two months, then the 95% geometric mean will produce incomparable results. The 95% geometric mean method is highly dependent on the underlying daily recruitment pattern, and appears to work best when ingress during the sampling period is fairly consistent.
The last index calculation method examined was the AUC; values resulting from this method were equal (1000.0). The AUC method is not sensitive to differences in annual sampling effort that may result in additional days with zero catches. More importantly, the index can easily accommodate variations in daily recruitment patterns that may be environmentally driven and vary from year to year.
One goal of recruitment monitoring is to allow comparison of relative recruitment between years with the underlying assumption that a constant relationship exists between the observed (calculated) index and the actual abundance of recruits. The index should be free from the influence of sampling variations that occur from year to year and should be invariant to within-year fluctuations in recruitment. The periodicity in recruitment that occurs within a single year is certainly of interest and may lead to insights into factors affecting recruitment variability, but the calculation of the index should not be affected by that pattern. A census that counts 500 eels recruiting to a pond in two days and no eels for the remaining 48 days of sampling compared with a census that counts 20 eels per day for 50 days should both result in a tally of 1,000 eels or an equivalent index. The current approach for calculating a recruitment index (based on the 95% geometric mean) appears to fall short of this goal. Results from this analysis were presented to the ASMFC American Eel Technical Committee and Stock Assessment Subcommittee meeting in Annapolis, MD September 14 - 16, 2009.