The desired end point of an assessment survey is to provide defensible quantitative estimates, on a river and/or region specific basis, of
(1) demographics of the live population by size class (as traditionally used in fishery management),
(2) age class (as desired by both fishery management and restoration activity), and
(3) status of the underlying shell substrate (reef) structure. All three are then addressed in the context of the long-term data sets extending backwards in time.
The patent tong oyster stock assessment survey began in fall 1993 for a subset of the current locations, with serial improvements being made in field methods and expansion of the survey footprint through 1998. The footprint continues to expand to this day as additional restoration sites are added. Methodological changes to include the discrimination of shell substrate types and greater fidelity in demographic descriptions have also been made. Methods have been described in peer publications by Mann et al. (2009), Southworth et al. (2010) and Harding et al. (2010). A detailed description with historical notes, based on Mann et al. (2009) follows. The surveyed areas are based on prior designations of public oyster grounds (Baylor Ground after Baylor 1896), and later updated by Haven et al. (1981). These delineations include a variety of bottom types deemed suitable for oysters including oyster reef (also termed oyster rock) through shell-mud and shell sand mixtures that were either originally marginal to more compact reef habitat or are the product of decades of disturbance by fishing and other anthropogenic activity. The survey design adopted in 1993 was based on the approach employed by NEFSC-NMFS in its offshore clam surveys; essentially treat each area as a stratum within which sampling will be effected on a random basis. In 1993 the approach was to mark the corners of the defined region, overlay a grid with the grid points being given progressive numbers, then sample grid points in the order determined by a random number table. The procedure of Bros and Cowell (1987) is then employed to assure adequacy of sampling within each strata – basically a plot of standard error of the running mean versus the number of samples collected within a strata, the desired minimum number of samples being determined by the point at which the plot levels and indicates decreasing information for each additional added sample.
The described approach has both strengths and weaknesses. All surveys where distribution of target species within a boundary is unknown are a compromise of time invested in mapping to determine sampling on the one hand, versus time and resources on the other. The current Virginia survey attempts to cover 180 distinct sampling areas covering approximately 8,900 acres in one calendar month with a vessel and crew that can sample 80 to 100 individual sites per day with a hydraulic patent tong under optimal weather conditions. Twenty calendar days in the field provides for approximately 1650 sites, or approximately one site (one sq. meter) every 5.2 acres (21,000 m2). It is clearly impractical to attempt a detailed mapping at scales finer than that provided by Haven and colleagues for the vast area at hand (notably Haven and field crew spent 4 years in the field producing their most recent generation maps, and the distance between their transect swaths was still of an order greater than certain reef footprints). Thus the employment of the above described approach. Retrospectively, the option exists to cumulatively add data by annual increments to the database describing the resource within each area (=strata) and re-stratify within the boundaries based on distribution of both oysters and shell substrate. This requires care in that re-stratification redirects the limited resources (1650 samples upper limit) with gains in fidelity in some regions being traded against losses in others. In addition, re-stratification cannot jeopardize continuity of long-term data. A related project has been initiated to compare, as a retrospective exercise on a virtual population developed by overlaying the entire geo-referenced Virginia historical data set on a single GIS layer, options for random, fixed station and adaptive sampling regimes to develop oyster population estimates.