Dredge efficiency on natural oyster grounds in Delaware Bay and its application in monitoring the Eastern oyster (Crassostrea virginica) stock in Delaware, USA

Fisheries Research 186 (2017) 292–300

Contents lists available at ScienceDirect

Fisheries Research journal homepage: www.elsevier.com/locate/fishres

Full length article

Dredge efficiency on natural oyster grounds in Delaware Bay and its application in monitoring the Eastern oyster (Crassostrea virginica) stock in Delaware, USA Frank Marenghi a,1 , Kathryn Ashton-Alcox b , Richard Wong c , Bellamy Reynolds a , Gulnihal Ozbay a,∗ a

Delaware State University, 1200 North DuPont Highway, Dover, DE 19901, USA Haskin Shellfish Research Laboratory, Rutgers University, 6959 Miller Avenue, Port Norris, NJ 08349, USA c Delaware Division of Fish and Wildlife, 3002 Bayside Dr., Dover, DE 19901, USA b

a r t i c l e

i n f o

Article history: Received 6 August 2015 Received in revised form 3 October 2016 Accepted 13 October 2016 Handled by Dr. P. He Available online 21 October 2016 Keywords: Eastern oyster Crassostrea virginica Dredge efficiency Delaware Bay

a b s t r a c t The annual Natural Oyster Ground Survey is the primary tool used in the management of the oyster fishery in the Delaware portion of Delaware Bay, USA. This survey monitors relative abundance, annual mortality, and recruitment of oysters from commercially important beds and tracks relative changes in the overall stock. Studies have shown that the commercial dredge gear used for this and similar surveys does not collect live oysters, boxes, and cultch with equal efficiency. We compared catch rates of live oysters, boxes, and cultch using this gear to adjacent diver-collected samples. Dredge efficiency varied by bed with the lowest efficiency on the bed with the least fishing pressure, possibly due to a higher degree of reef consolidation. Dredge efficiencies between size classes (20–65 mm, 65–75 mm, >75 mm) were highly variable, although the dredge may be biased towards the capture of live oysters and larger (>65 mm) individuals. Oyster abundance (not including spat) estimated from annual surveys and corrected for dredge efficiency on the commercially important Delaware oyster beds increased from approximately 240 million to 491 million individuals between 2007 and 2008. A comparison of three estimation techniques (bed-specific efficiencies, mean efficiencies, and bootstrapping) is discussed in the broader context of estimating population abundance, size structure, recruitment, and mortality, when corrected and uncorrected for dredge efficiency. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Eastern oyster, Crassostrea virginica Gmelin, 1791, beds in Delaware Bay historically supported large oyster fisheries. Industry and managers adopted practices to maintain the population for the continuation of the fishery from as early as 1873 (Fegley et al., 2003). Annual landings for Delaware Bay were consistently between 1 and 2 million bushels until the middle of the 20th century when oyster mortalities caused by Haplosporidium nelsoni (MSX) and Perkinsus marinus (Dermo) were first documented (Ford and Haskin 1987; Ford, 1996). Declines in landings are also attributed to fluctuations in imported seed and market demand,

∗ Corresponding author. E-mail address: [email protected] (G. Ozbay). 1 Current address: Maryland Department of Natural Resources, 580 Taylor, Ave B-2, Annapolis, MD 21401, USA. http://dx.doi.org/10.1016/j.fishres.2016.10.007 0165-7836/© 2016 Elsevier B.V. All rights reserved.

particularly in the years of the Great Depression (Ford, 1997; Fegley et al., 2003). The effect of fishing cannot be quantified because the stock was not regularly monitored after a 1910 survey (Moore, 1911) and removals from the natural beds could not be distinguished from planted stock (Ford, 1997). Nevertheless, these landings may have been unsustainable due to declining oyster abundance even in the absence of disease epizootics (Moore, 1911; Fegley et al., 2003). Like New Jersey, Delaware traditionally had a transplant fishery in which small oysters experiencing slower growth and lower mortality on up-bay public seed beds were transplanted onto private leases down-bay in higher salinity waters, increasing their growth rate and market quality prior to harvest later in the year or in future years (Powell et al., 1997; Ford et al., 1999; Greco, 2009). Legislation enacted in 2001 established a quota-based fishery in which oysters were sold directly from the natural seed beds (Cole et al., 2009). The annual Natural Oyster Ground Survey is the primary tool used in the management of the oyster fishery in Delaware. Since

F. Marenghi et al. / Fisheries Research 186 (2017) 292–300

1974, the Delaware Department of Natural Resources and Environmental Control (DNREC) has collected relative abundance, mortality, and recruitment data from the major oyster beds in the Delaware portion of Delaware Bay to track relative changes in the stock (Greco, 2009). The annual oyster harvest quota in Delaware is based on the relationship between harvest and relative abundance from the annual survey in prior years (Cole et al., 2009). Commercial oyster dredges or smaller dredges have been used in surveys in both Delaware and Chesapeake bays for many decades (Cole, 1988; Fegley et al., 2003; Mann et al., 2004; Tarnowski, 2005; Volstad et al., 2008; Greco, 2009). Inefficiency and bias in dredge sampling have been noted for at least two decades (Chai et al., 1992; Powell et al., 2002; Mann et al., 2004). The dredge with its iron tooth bar is designed to ideally pull the topmost oysters off a reef, leaving behind the associated reef material. In areas of highly consolidated reef or with rock, a dredge may ‘bounce’ somewhat, while in softer, less consolidated areas, the teeth are likely to dig in and pick up more objects in its path. In Virginia, a dredge survey has been supplemented with quantitative sampling with patent tongs since 1993 (Mann et al., 2004). The New Jersey survey was improved in 2000 by using dredge calibration in conjunction with measured swept areas and tow volumes (Powell et al., 2002, 2007). Powell et al. (2002, 2007) determined that oysters were captured with greater efficiency than boxes (dead oysters with articulated valves) while boxes were picked up with greater efficiency than cultch (oyster shell only). Larger oysters and boxes are often collected with greater efficiency than smaller oysters and boxes, although the tendency of oysters to attach to each other makes this evaluation difficult (Powell et al., 2002, 2007). The dredge may simply not pick up some oysters while others may pass through the mesh. If not accounted for by dredge calibration, these biases may have large ramifications in estimating numbers of oysters and boxes based on dredge samples. For example, if dredge efficiency for live oysters varies significantly from that for boxes, annual mortality will be under- or overestimated with potentially large management implications. If the dredge is biased with regard to oyster shell length, this will alter the perceived size class distribution, e.g., the estimation of the number of market-sized animals. Our study determines survey-employed commercial dredge efficiencies on four commercially important oyster beds for various material types (live oyster, box, cultch), and size classes (20–65 mm, 65–75 mm, >75 mm) in 2007 in Delaware (Dredge Calibration). We then convert efficiencies into catchability coefficients to expand survey dredge catches to absolute oyster abundances. These abundances and harvest monitoring results can be used by DNREC to determine annual exploitation rates, providing information used when setting quotas. Differences between efficiency-corrected and uncorrected estimates of abundance, annual mortality, and recruitment as well as the implications of three estimation techniques (bed-specific efficiencies, mean efficiencies, and bootstrapping) are discussed.

2. Methods 2.1. Dredge calibration We conducted a gear efficiency study in October 2007 to compare the oyster catch rates from DNREC survey dredge gear to adjacent diver-collected samples. The survey vessel is 19 m in length and uses a stern-towed, commercial dredge (1.33 m width with 24 teeth measuring 9.5 cm long, spaced 6.4 cm apart). The dredge bag is constructed with 6.0 cm rings with a 5.4 cm inside diameter spaced 6.4 cm apart. There was no diving plane. This is the same design as is used throughout the Delaware Bay for both commercial oystering and survey sampling in New Jersey. Sam-

293

ples were collected and processed using methods developed for the New Jersey oyster beds (Powell et al., 2002, 2007). Twentyseven dredge tows (paired with diver transects described below) were taken across the four most important natural oyster beds in Delaware: Ridge, Lower Middle, Over-the-Bar, and Silver (Fig. 1, Table 1). Three study sites were located on Ridge and two study sites were situated on each of the other beds. Study sites were intentionally located in the middle portions of each bed, away from the bed peripheries, and in close proximity to the DNREC annual survey stations. Three replicate tows were taken at each location. Each tow lasted approximately 45 s. This relatively short tow time is necessary so the dredge does not overfill and begin to push material along the bottom (Powell et al., 2002; Mann, 2010). For each tow, total bottom time and positions were recorded using the Global Positioning System (GPS) onboard the survey vessel. Tow distance was calculated from GPS positions recorded every 5 s over a 45 s tow. The total volume of bottom material collected by the dredge was recorded and a one-bushel (35 L) subsample was taken from each tow. At each site, a leaded line was deployed immediately adjacent (within ∼5 m) and parallel to the dredge path (Powell et al., 2002). Twelve replicate 0.25 m2 quadrats were sampled by divers along the undisturbed transect. Divers took all loosely consolidated reef material on the bottom within the quadrat and placed it in the dive bag. In the event that large quantities of shell were present, the diver took enough of the upper portion of the shell base to fill the dive bag (Powell et al., 2002). There are two assumptions inherent in this method: (1) the dredge is less than 100% efficient but the diver can collect everything within the quadrats and (2) the dredge tow and the diver transect are close enough to independently sample two very similar portions of the oyster bed. All dredge and dive bag samples were stored in a refrigerated walk-in cooler at approximately 10 ◦ C prior to sorting. Each sample (bushels and dive bags) was sorted into categories: live oysters, boxes, spat (oyster <20 mm), and cultch. The volume of each material was recorded and the total shell length (longest dimension) of all oysters and boxes was measured. Counts of live oysters and boxes were divided into three size classes for analysis: smalls, 20–65 mm in length; sub-markets, 65–75 mm (Kraeuter et al., 2007); and markets, >75 mm. All sizes of cultch were sorted together. The area of the bottom swept by the gear was determined by multiplying distance towed (m) × dredge width (m). Counts of oysters and boxes per sample bushel and cultch volume per sample bushel were standardized per square meter swept. The volumes of material collected from the diver transects were also standardized per square meter sampled and oyster, box and cultch density (m−2 ) were calculated in a similar fashion. Dredge efficiency (e) was calculated as the ratio of oyster density in the dredge sample (m−2 ) to oyster density in the dive sample (m−2 ) for oysters and boxes; the dredge efficiency for cultch was Table 1 Oyster bed names, their area (m2 ), and sampling intensity for the 2007 and 2008 Delaware Natural Oyster Grounds surveys. Bed name

Area (m2 )

Number of Survey Tows

Ridge Silver Over-the-Bar Lower Middle Red Buoy Woodland Beach Pleasanton’s Rock Drum Persimmon Tree

1,527,540 1,297,023 1,163,836 1,050,197 715,267 391,709 246,381 166,655 50,588

15 10 6 6 3 6 3 3 2

Total

6,609,196

54

294

F. Marenghi et al. / Fisheries Research 186 (2017) 292–300

Fig. 1. Map of Delaware Natural Oyster Grounds in the annual survey (DNREC 2008).

calculated as the ratio of cultch volume in the dredge sample (m−2 ) to cultch volume in the dive sample (m−2 ). The catchability coefficient (q) is the multiplier by which dredge-corrected densities

are obtained and is equivalent to 1/e. The oysters, boxes, or cultch density from a dredge tow must be multiplied by the q-value to account for what the dredge did not catch to match the assumed

F. Marenghi et al. / Fisheries Research 186 (2017) 292–300

295

Table 2 Mean dredge efficiencies (%) for three size classes of oysters and boxes and for cultch on the four study beds. All-Size, calculated from the entire size-frequency; Bed Means, calculated from All-Size oyster, All-Size box, and cultch by bed. Bed

Ridge Silver Over-the-Bar Lower Middle All-Bed Mean

Oysters

Boxes

20–65 mm

65–75 mm

>75 mm

All-Size

20–65 mm

65–75 mm

>75 mm

All-Size

8.9 15.2 12.9 12.3 12.3

11.4 20.4 16.1 20.4 16.8

15.1 21.2 13.1 10.3 15.6

10.9 18.9 13.3 10.4 13.8

8.2 7.3 33.0 16.9 15.9

12.4 16.8 10.3 5.3 12.0

10.9 14.7 15.4 5.5 12.4

7.1 14.6 14.3 6.2 11.2

Fig. 2. Dredge efficiency (mean ± SE) for live oysters and boxes in three size classes (20–65 mm, 65–75 mm, >75 mm) in the dredge calibration study.

Cultch

Bed Mean

4.9 5.4 9.1 22.5 9.4

7.2 13.0 12.2 14.7 11.3

Fig. 3. Dredge efficiency (mean ± SE) for live oysters, boxes, and cultch in the dredge calibration study. Oyster and box dredge efficiencies calculated using all sizes of oysters or boxes in each tow.

analyses were performed using Minitab Version 15.1.30.0. Results are expressed as efficiency (e) percentage unless otherwise noted. 100% efficiency of a diver at the same site (Powell et al., 2002, 2007). Catchability coefficients were calculated for the three previously mentioned size classes for both oysters and boxes and for cultch volume within individual dredge tows and then averaged by bed. Additional q-values were calculated for all sizes of oysters and of boxes (All-Size) for each tow. The All-Size efficiencies represent the degree to which the dredge caught live oysters or boxes of any size compared to diver-collected samples. Thus, there were 9 q-values per each of the 27 dredge tows. In some cases (primarily Ridge site 1 and Lower Middle site 1), results were excluded for size classes where there were no oysters or boxes caught and q = 0. In a few cases on Lower Middle site 1, the dredge caught more than the diver for a particular size class and the resulting q-value was discarded. Catchability coefficients were averaged for each category by bed. To determine the extent to which size class, bed, and material type (live oyster, box, or cultch) affected dredge efficiency, we employed two-way nested analysis of variance (ANOVA) with material type and size class nested within bed. All factors were considered fixed effects. A three-way ANOVA would have led to an unbalanced design, therefore, we ran one ANOVA by bed and material type (live oyster, box, cultch) and a second ANOVA by bed and size class (20–65 mm, 65–75 mm, >75 mm). A post-hoc two-way nested ANOVA using dredge efficiency for oysters (AllSize) was performed by material type and bed. Nested ANOVAs were employed because the intrinsic properties of oyster reefs may influence the proportion of oysters to boxes to cultch and their sizes due to potential differences in recruitment and harvest history by bed (Boudreaux et al., 2006; Southworth et al., 2010). The transformed (natural log + 1) q-value was the dependent variable in all the hypothesis tests. Assumptions of normality (KolmogorovSmirnov test, p > 0.15) and heterogeneity of variances (Levene’s test, p > 0.2) were satisfied by the transformed data. All statistical

2.2. Stock abundance estimates In 2007 and 2008, the annual DNREC Natural Oyster Ground survey was conducted using the same vessel and dredge used for the dredge calibration study (Cole et al., 2009; Greco, 2009). The survey consists of a fixed station design, with stations located uniformly across beds. Samples were taken at 2–15 fixed stations on each commercially important bed in Delaware waters, for a total of 54 samples in each year (Table 1, Fig. 1). The number of samples is proportional to the size of the bed, covering roughly one sample per 30 acres. Dredge tow durations were approximately 45 s. Coordinates of each tow were logged from the on-board GPS, the total volume of material collected was measured, and a onebushel subsample was taken. Samples were stored and processed as described for the dredge calibration study. Quality control was provided by examining a random subsample of cultch from each bushel carefully in a well-lit space to determine if any spat, oysters, or boxes were missed during sorting. If detected, bushel results were adjusted proportionately. Oyster bed areas (Table 1) were taken from a hydroacoustic core sampling study conducted by the DNREC Division of Soil and Water Conservation (Wilson et al., 2006). Uncorrected densities of oysters and boxes from the stock abundance survey were calculated as described for the dredge calibration analysis and expressed as numbers per m2 of dredge-swept bottom. The All-Size catchability coefficient (q-value) for oysters from the dredge calibration analysis was then applied to the oyster densities and the All-Size box q-value was applied to the box densities to correct for dredge efficiency. To calculate corrected spat densities, we used the q-value for each material type to which the spat was attached. The qvalues were bed-specific when used for the four beds that were in the dredge calibration study and means from the study beds

296

F. Marenghi et al. / Fisheries Research 186 (2017) 292–300

Table 3 Average densities (m−2 ) from the Natural Oyster Ground Surveys for three size classes of oysters and boxes as well as spat (<20 mm). All densities were corrected for dredge efficiency. Bed-specific q-values were used for beds from the dredge calibration study and mean q-values were used for non-study beds. Although oysters are divided into size classes, the All-Size q-value was used to correct for dredge efficiency. Spat densities used the q-value for each material type on which the spat set. 2007

2008

Bed name

>75 mm

65–75 mm

20–65 mm

<20 mm

>20 mm Boxes

>75 mm

65–75 mm

20–65 mm

<20 mm

>20 mm Boxes

Ridge Silver Over-the-Bar Lower Middle Red Buoy Woodland Beach Pleasanton’s Rock Drum Persimmon Tree

8.6 16.5 17.8 34.5 20.8 38.0 26.4 20.6 5.1

4.6 5.7 5.7 8.6 4.1 9.8 8.3 6.9 2.0

10.6 10.3 8.8 13.0 11.5 19.9 15.8 15.9 3.8

81.1 52.0 33.0 168.5 244.9 14.8 192.5 208.8 3.9

7.7 15.7 11.5 15.0 10.9 17.6 32.8 23.7 4.2

6.6 13.5 19.8 18.8 10.5 23.6 13.4 14.7 31.7

3.7 3.1 6.3 8.2 4.1 6.5 4.9 6.6 7.9

23.1 49.1 33.8 56.0 19.8 11.6 108.2 87.9 11.4

15.9 99.6 15.9 44.6 12.3 6.5 58.7 68.9 6.2

19.0 18.1 14.3 13.0 11.3 14.6 24.4 24.7 7.0

Table 4 Annual mortality rate per bed (% from box counts) uncorrected and corrected for dredge efficiency; corrected using All-Size q-values for oysters and boxes; bedspecific q-values used for dredge calibration study beds, mean q-value used for other beds. 2007

skewness in the distribution of the observed data (Efron, 1987; Haddon, 2001). 3. Results

2008

Bed name

Uncorrected Corrected Uncorrected Corrected

Ridge Silver Over-the-Bar Lower Middle Red Buoy Woodland Beach Pleasanton’s Rock Drum Persimmon Tree Overall Mortality (All Beds)

10.8 25.4 25.8 16.2 14.3 12.8 26.7 23.4 17.6 19.7

24.5 32.6 26.2 21.1 22.9 20.7 39.3 35.4 27.5 26.2

17.6 16.3 11.2 9.8 15.5 10.2 18.9 16.4 7.2 15.0

36.3 21.6 19.2 13.5 24.7 25.9 16.2 18.4 12.1 21.7

were used for the remaining five surveyed beds. Absolute abundances of oysters, boxes, and spat were calculated by multiplying the efficiency-corrected densities (m−2 ) by the total bed areas. To facilitate a comparison with the historical survey of Moore (1911), oyster densities were converted to bushels per acre using 380 oysters per bushel (Grave, 1912). Boxes (dead oysters with articulated shells representing recent deaths) were used to calculate recent mortality rates as commonly practiced in similar oyster studies (Ford et al., 2006; Powell et al., 2006). The fraction of boxes (boxes/boxes + live oysters) found in samples represents the percentage of oysters that had died within the past year, a de facto annual mortality rate. We calculate annual mortality rates for each bed and across all beds. Market (oysters >75 mm) to spat and live oysters to boxes ratios were calculated for further descriptions of stock status. Absolute abundances of live oysters, boxes, and spat corrected for dredge efficiency were calculated three ways for the 2007 and 2008 surveys using: (1) mean All-Size catchability coefficients from the four dredge calibration study beds applied to all nine beds, (2) bed-specific All-Size catchability coefficients for the four dredge calibration study beds applied to each of those beds and the aforementioned All-Size mean used for the remaining five beds, and (3) a bootstrapping routine including bias-corrected 90% confidence intervals in which catchability coefficients for each material type were randomly selected from all those measured in the dredge calibration experiment and applied to the uncorrected density of each tow for 2000 iterations (Efron, 1987; Haddon, 2001). Bootstrapping allows determination of the sampling distribution of each bed by repeated and random resampling with replacement from the observed data. Confidence intervals were created from the resultant distribution without having to transform data or satisfy parametric statistical assumptions (Efron, 1987; Rochowicz, 2011). Bias-corrected confidence intervals were calculated to account for

3.1. Dredge calibration Average dredge efficiencies ranged from a low of 4.9% for capture of cultch on Ridge to a high of 33.0% for catching small (20–65 mm) boxes on Over-the-Bar (Table 2). The dredge always left behind far more material than it caught, thus the catchability coefficients (q-value multipliers) necessary to correct dredge catches for what was missed are correspondingly high. There was no statistically significant difference in dredge efficiency between the three size classes of live oysters or boxes (df = 8, F = 0.71, p = 0.682; Fig. 2). Because of this, we used the q-values calculated from the entire size frequency (All-Size) of live oysters or boxes for the remainder of the analyses. Dredge efficiencies between material types (live oyster, boxes, cultch) were not significantly different (df = 8, F = 1.36, p = 0.235), and varied from live oysters (13.8%), to boxes (11.2%), to cultch (9.4%) (Table 2, Fig. 3). Bed-specific means calculated by averaging All-Size oyster, AllSize box, and cultch efficiencies varied noticeably among beds although not significantly (df = 3, F = 2.45, p = 0.073). The lowest mean was on Ridge (7.2%) while efficiencies were similar on Silver, Over-the-bar, and Lower Middle averaging 13.0%, 12.2%, 14.7%, respectively (Table 2). 3.2. Stock abundance estimates and population indices Live oyster and box densities for each size class including spat, from the 2007 and 2008 stock surveys were corrected for dredge efficiency using All-Size bed-specific q-values for Ridge, Silver, Over-the-bar, and Lower Middle, and the All-Size mean q-value for the remaining five beds (Table 3). Oyster densities decreased from 2007 to 2008 for the two largest size classes on nearly all beds with a 19% average decrease for oysters >75 mm and an 8% average decrease for oysters in the 65–75 mm size class. The relatively large spat set of 2007 survived well enough to increase the average oyster density for the 20–65 mm size class by a factor of 3.7 (Table 3). Although the largest size classes of oysters decreased in density between 2007 and 2008, all estimates of total abundance (oysters >20 mm) increased due to the abundance of 20–65 mm individuals in 2008 (Table 3). For the same reason, although box densities increased on six of the nine beds between 2007 and 2008, the annual mortality rate decreased between years due to the increased abundance of small oysters (Tables 3 and 4). The risk of not correcting densities for dredge efficiency is illustrated in Tables 4–6 that show uncorrected versus corrected values

F. Marenghi et al. / Fisheries Research 186 (2017) 292–300 Table 5 Ratio of market-sized (>75 mm) oysters to spat, uncorrected and corrected for dredge efficiency; corrected used All-Size q-values for oysters and boxes. Spat corrections used the q-value for each material type on which the spat set. Bed-specific q-values were used for dredge calibration study beds and the mean q-value used for other beds. 2007

2008

Bed name

Uncorrected

Corrected

Uncorrected

Corrected

Ridge Silver Over-the-Bar Lower Middle Red Buoy Woodland Beach Pleasanton’s Rock Drum Persimmon Tree

0.16 0.80 1.12 0.36 0.14 3.86 0.29 0.21 2.61

0.11 0.32 0.54 0.21 0.09 2.56 0.14 0.10 1.32

0.53 0.30 2.40 0.71 1.32 5.95 0.34 0.31 8.59

0.42 0.14 1.25 0.42 0.85 3.61 0.23 0.21 5.13

Table 6 Ratio of all live oysters to boxes uncorrected and corrected for dredge efficiency; corrected used All-Size q-values for oysters and boxes; bed-specific q-values used for dredge calibration study beds, mean q-value used for other beds. 2007

2008

Bed name

Uncorrected

Corrected

Uncorrected

Corrected

Ridge Silver Over-the-Bar Lower Middle Red Buoy Woodland Beach Pleasanton’s Rock Drum Persimmon Tree

8.26 2.94 2.87 5.15 5.98 6.84 2.75 3.27 4.69

3.09 2.07 2.82 3.75 3.36 3.84 1.54 1.83 2.63

4.69 5.13 4.28 8.79 5.44 5.10 9.25 7.89 12.90

1.75 3.62 4.20 6.40 3.05 2.87 5.19 4.43 7.24

for three parameters. We focus here not on the specific values of these ratios but the implications of using dredge efficiencycorrected and uncorrected ratios. Correcting for dredge efficiency in Table 4 increases the annual mortality rate estimates in all but one case (Pleasanton’s Rock in 2008). Not accounting for dredge efficiency underestimates overall mortality by 6.5% and 6.7% in 2007 and 2008, respectively. On Ridge in 2008, mortality without correction would have been underestimated by a factor of 2 (Table 4). In Table 5, the efficiency-corrected ratios of >75 mm oysters to spat were always less than uncorrected ratios. The difference ranged from 22 to 60% depending on bed and year. Similarly in Table 6, the corrected ratios of >20 mm oysters to boxes were always less than the uncorrected. Efficiency-corrected ratios of market to spat and live oysters to boxes (Tables 5 and 6) are the result of using the All-Size catchability coefficients for live oysters, boxes, and spat for each bed. Spat are not included in the live oyster to box ratios. The efficiencycorrected ratios of oysters to spat were 2–60% less than uncorrected ratios, depending on bed and year. The corrected ratios of oysters to boxes were 2–63% less than uncorrected ratios depending on bed and year. The degree to which the efficiency-correction affected the live oysters to boxes ratios was quite variable between beds. The uncorrected ratio for Over-the-bar would have values overestimated by 2%, while ratios at Ridge would have been overestimated by 60% in both years. Ratios on many of the beds would have been overestimated by 27–44%, indicating there were many more live oysters to dead than actually present (Table 6). Oyster abundance, excluding spat, on the Delaware Natural Oyster Grounds increased from 2007 to 2008 by a factor of 1.6 to 2.0 depending on the estimate used and annual mortality decreased between years by between 5 and 11% (Table 7). The bootstrap method incorporates the range of calculated efficiencies according to the distribution of the observed data. Per-tow efficiencies ranged

297

Table 7 Delaware Natural Oyster Grounds population estimates for 2007 and 2008, excluding spat. A. Abundance calculated for all nine survey beds and for the four beds from the dredge calibration study (Ridge, Silver, Over-the-Bar, Lower Middle) using three methods: bootstrap, mean efficiency, and bed-specific efficiency for the study beds but mean efficiency for non-study beds. B. The bootstrap abundance estimate with 90% confidence intervals. A Method

Abundance

Annual Mortality

2007 Bootstrap Mean efficiency Bed-specific efficiency

239,897,622 253,627,672 247,807,662

27% 30% 26%

2008 Bootstrap Mean efficiency Bed-specific efficiency

490,725,864 402,834,641 386,013,950

16% 24% 22%

B 90% Confidence Interval Year

Oyster Population Estimate

Lower Bound

Upper Bound

2007 2008

239,897,622 490,725,864

84,254,525 156,301,610

479,651,813 972,059,541

between 1 and 85% in the dredge calibration study although they were generally very low with a quarter of them less than 5% and three-quarters less than 16%. Incorporating this many low dredge efficiencies (higher q-values) also explained the large confidence intervals that were produced (Table 7B). The method using the mean efficiency across all beds produced the highest estimate of mortality (Table 7A). Although there is no ‘gold standard’, the comparison of these three methods informs the relationship between dredge efficiency and the precision of the abundance estimate on the beds. 4. Discussion 4.1. Dredge calibration Dredges are designed to allow smaller particles to pass through while retaining larger ones. Although efficiency and oyster size are positively related in other dredge efficiency experiments (Powell et al., 2002), this relationship was not supported statistically in our results. No pattern in dredge efficiency was detected for boxes of different sizes. Size-based efficiency is partially confounded due to the nature of oysters of assorted size classes to grow together in clusters. We chose to use dredge efficiencies based on the full size-frequency of oysters greater than 20 mm for abundance estimates. Estimates of recruitment (spat counts) were determined by using efficiency estimates for the material on which the spat set, the amount and size of which likely varies within and between beds. The proportion of single oysters or boxes, and those in clusters and in combination could be calculated and their lengths measured in future studies. Although we failed to detect significant differences between material types (live oyster, box, cultch), we chose to use the observed material-specific catchability coefficients to estimate abundances in order to reflect the nominal differences seen in our study and other similar studies (Fig. 3). It is possible, for example, that boxes break more easily when collected by dredge than by diver. Decreased capture efficiency for boxes is consistent with findings from multiple dredge efficiency studies on at least ten oyster beds in the New Jersey portion of the Delaware Bay (Powell et al., 2002, 2007; J. Morson 2015, personal communication). Annual mortality estimates from box counts differ dramatically when cor-

298

F. Marenghi et al. / Fisheries Research 186 (2017) 292–300

rected for dredge efficiency (Table 4). Using uncorrected estimates almost always underestimated mortality. Mean dredge efficiency for oysters on the four study beds was always <22%, meaning that the dredge is a relatively inefficient sampling tool on a per-area basis because it caught <22% of the oysters the diver was able to collect from the bottom. The dredge efficiency at Ridge was lower than the other three beds in the study; half that of Lower Middle, where the dredge was most efficient (7.2% and 14.7%, respectively). These overall and between-bed differences in efficiency are similar to those reported for oyster beds in the New Jersey portion of Delaware Bay (Powell et al., 2007; see Table 1). For example, the average dredge efficiencies for AllSizes of oysters and boxes and for cultch over the four beds in this study were 13.8%, 11.2%, and 9.4% respectively (Table 2). The same averages over nine sites for the New Jersey study (using an identical dredge towed from the port side) were 16.9%, 11.4%, and 8.4% (Powell et al., 2007; see Table 1). Dredge gear has the ability to reduce the three dimensional structure and heterogeneity of oyster reefs; the action of the dredge while fishing and the culling that occurs on deck breaks up many of the large oyster clusters (Rothschild et al., 1994; Lenihan and Peterson, 2004). Dredges function with increased efficiency on loose unconsolidated reefs compared with unfished reefs (Powell et al., 2002). Ridge is the largest of Delaware’s Natural Oyster Grounds and historically supported the majority of the fishing effort until 2005 (Whitmore and Cole, 2003; Whitmore and Greco, 2005). Although it was closed to fishing due to low survey indices for fourteen non-consecutive years since 1977, the closures included 2005 through 2008. The closure to fishing leading up to and including the dredge calibration study may have allowed Ridge oyster reefs to consolidate into more numerous or more complex clumps, leading to the observed low dredge efficiency for live oysters (10.9%, Table 2). The low harvest pressure (less than 5%) from 2005 to 2007 on Lower Middle (Greco, 2009) may be reflected in its similarly low dredge efficiency for live oysters (10.4%, Table 2). Over-the-Bar contributed approximately 10% of the harvest from 2005 to 2007 with dredge efficiencies for oyster in an intermediate range (13.3%, Table 2). Since 2005, Silver has accounted for the majority (up to 45%) of the fishing effort in Delaware waters (Whitmore and Cole, 2003; Greco, 2009) and it had the highest dredge efficiency for live oysters in the study (18.9%, Table 2). Due to the varied dynamics of an oyster reef, dredge efficiency should be expected to change over time (Powell et al., 2007). For example, although Lower Middle experienced less than 5% of the Delaware oyster harvest between 2005 and 2007, in 2008, approximately 40% of the harvest came from this bed (Greco, 2009). In addition to fishing pressure, setting patterns may also affect reef consolidation over time and thus, dredge efficiency. More frequent and intense recruitment events on a bed will likely result in more and larger oyster clumps and decreasing dredge efficiency, while beds with lower and less frequent recruitment might be less clumpy. From 2001 to 2006, leading up to the 2007 dredge calibration study there was little or no spat set on any of the Delaware oyster beds (Greco, 2009) so this effect could not be determined here. In earlier years, however, recruitment events were more variable between the beds, with some having consecutive years of high recruitment, e.g., Ridge and Lower Middle from 1997 to 2000, at the same time that others, e.g., Over-the-Bar and Woodland Beach, had much lower recruitment (Greco, 2009). Another factor that might affect dredge efficiency on oyster beds is the occurrence of a large or ongoing mortality event such as a freshet or a disease outbreak. Periodic assessments of the same beds may be necessary for dredge calibration to remain accurate over time. Efficiency estimates should therefore be made on the greatest number of beds practical i.e., all surveyed beds, due to the variability present between oyster beds in a region, even those in close proximity. This

difference was observed most dramatically on Ridge, the bed with the lowest efficiency (7.2%). The large range (1–85%) of per-tow efficiencies in this study simultaneously suggests that within-bed variability may be as great as the variability between beds, so whether or not more dredge efficiency estimates are made, increasing the number of samples within beds during surveys may be a prudent way to refine estimates of abundance and mortality when resources are limited (Cochran, 1977; Smith, 1996; Southworth et al., 2010). 4.2. Stock abundance estimates and population indices By using our dredge efficiency calculations, it is possible to provide an estimate of total oyster abundance for the commercially important beds of the Delaware Natural Oyster Grounds. We calculated abundance using mean and bed-specific efficiencies and got estimates ranging from 240 to 250 million oysters in 2007 to 390–490 million oysters in 2008 (Table 7A and B). That is at least a 56% increase in the number of oysters between 2007 and 2008 and is attributed to the spat set that occurred in the fall of 2007 which was the largest in Delaware since 2000 (Greco, 2009; Delaware Division of Fish and Wildlife, 2010) and translated into an increase from the <20 mm size class in 2007 to the 20–65 mm size class in 2008. Oyster density for this size-class increased by an average of 32.4 oysters m−2 (Table 3). There was a coincident decrease in natural mortality from 2007 to 2008 of at least 5% overall (Table 4). The observed decrease in density for the >65 mm oysters may be explained in part to fishery removals during 2008, although on seven of the nine beds density decreased by a factor of 2 for the larger oysters while about 85% of the harvest came from Lower Middle and Silver (Greco, 2009) and Ridge remained closed during the 2008 season (Greco, 2009). For an historical perspective, Moore (1911) classified oyster densities in Delaware Bay into four categories: Dense, Scattering, Very scattering, and Depleted. Observed oyster densities for eight of the nine survey bars in 2007 and 2008 (Table 3) would have met the criteria for “Dense growth” >150 bushels per acre; >14 oysters m−2 ). Only Ridge in 2008 and Persimmon Tree in 2007 may have been considered by Moore to be “Very scattering” (<75 bushels per acre; <7 oysters m−2 ) or “the least productive bottom capable of furnishing a livelihood to the dredgers” (Moore, 1911). Summing the product of the efficiency-corrected density of live individuals on each bed and bed area for all beds provides an estimate of the total number of oysters present on the Delaware Natural Oyster Grounds but does not provide an unbiased confidence interval with which to compare annual change. Confidence intervals calculated by bootstrapping are often more accurate than those obtained with standard methods (Efron, 1987; Efron and Tibshirani, 1993). Despite this, the 90% confidence intervals generated around our population estimate were large. This was due to several factors, including the variability in the dredge efficiency estimates per-tow. Oyster densities before correction for dredge efficiency were also variable between tows, among and within beds. The ratio of market-sized (>75 mm) oysters to spat (<20 mm) is an indicator of the long-term survival of a bed because more spat (ratios <1) are required to compensate losses due to mortality (Haskin Shellfish Research Laboratory, 2013). A high ratio (>1) suggests a lack of recruitment to the area and a low ratio (<1) indicates there are enough spat in the population to replace the large oysters that will eventually die (Haskin Shellfish Research Laboratory, 2008). A low ratio alone does not indicate persistence of the population because the chance of the number of spat being greater than the number of markets is greater in years when the market population is low. These ratios are most useful when evaluated along with other indicators at appropriate time scales (years to decades). Furthermore, a ratio <1 is likely not required every

F. Marenghi et al. / Fisheries Research 186 (2017) 292–300

year for the bed to maintain itself due to the naturally irregular recruitment patterns of oysters. In this study, correcting for dredge efficiency caused the market-to-spat ratio to decrease, suggesting recruitment would have been underestimated without the correction. Two beds (Over-the-Bar in 2007 and Red Buoy in 2008) had an uncorrected market to spat ratio >1, suggesting that recruitment limitation may be occurring. The efficiency-corrected ratio was <1, suggesting that recruitment may be adequate here and that the population may increase over the near-term (Table 5). The ratio of live oysters to boxes is an indication of relative mortality and may inform the relationship between live oysters and shell availability and provide information on long-term bed viability. Corrected live oyster to box ratios in this study were lower (by as much as 63%) than uncorrected ratios in all cases (Table 6). Differences in dredge efficiency between oysters and boxes could skew the ratios of live oysters to boxes and, while not significantly different in this study, patterns in the efficiency among live oysters, boxes, and cultch indicate these categories should be retained for use in estimating abundance pending further study. Failing to account for different efficiencies with which the dredge collects boxes and live oysters if they exist, can result in an under- or overestimate of mortality rate from box counts, especially if boxes only persist for one year or less (Ford et al., 2006; Powell et al., 2006). It was difficult to separate distinct size classes of oysters in this study due to their tendency to grow together in clumps and the lack of significant differences in dredge efficiency between size classes. Therefore, using the entire size frequency (All-Size) efficiencies is the most prudent method for dredge-efficiency correction for the annual Delaware survey at this time. The degree to which bed-specific efficiencies are reflected in the Delaware population assessment is uncertain, however intriguing patterns require further study to examine differences in dredge efficiency between beds. Calculating abundance by incorporating the per-tow efficiencies in the bootstrap method affected the population estimates more than choosing between the bed-specific and mean efficiency corrections (Table 7A). The catchability coefficients for the All-Size category for live oysters and boxes across beds would be a reasonable approach to calculate the annual stock abundance estimate for the Delaware Natural Oyster Grounds but may not be as accurate as the bootstrap method if within and between-bed dredge efficiencies are as variable for all the beds as for the four used in the dredge calibration study. Annual mortality estimates appear to be less sensitive to the methods examined (Table 7A). Incorporating dredge efficiency into the Delaware Natural Oyster Ground survey in this way has led to a quantitative estimate of the Eastern oyster population in Delaware for the first time.

Acknowledgements We thank Captain Mike Garvilla, Mike Greco, and crew of the R/V First State. Dr. Eric Powell and Rebecca Marzec from Haskin Shellfish Research Laboratory were instrumental in training, logistics, and programming for this project. We also thank Keleigh Provost, Johnna Fay, Oluchi Ukaegbu, and Kate Rossi-Snook in Dr. Ozbay’s Lab who helped process samples. Karen Capossela provided helpful comments on the manuscript. We also would like to thank to Mrs. Laurieann Phalen in Dr. Ozbay’s Lab for her assistance with the manuscript formatting. This research was conducted in conjunction with the Delaware Bay Oyster Restoration Task Force which includes the Cumberland County Empowerment Zone, the Delaware and New Jersey Shellfish Industry, the Delaware Department of Natural Resources and Environmental Control, the Delaware River and Bay Authority, the Delaware River Basin Commission, Delaware State University College of Agriculture and Related Sciences, New Jersey Department of Environmental Protec-

299

tion, the Partnership for the Delaware Estuary, Rutgers University’s Haskin Shellfish Research Laboratory and the U.S. Army Corps of Engineers, Philadelphia District. This program was funded by the U.S. Army Corps of Engineers, Philadelphia District (W912BU-07P-0218).

References Boudreaux, M.L., Stiner, J.L., Walters, L.J., 2006. Biodiversity of sessile and motile macrofauna on intertidal oyster reefs in Mosquito Lagoon, Florida. J. Shellfish Res. 25, 1079–1089. Chai, A., Homer, M., Tsai, C., Goulletquer, P., 1992. Evaluation of oyster sampling efficiency of patent tongs and an oyster dredge. N. Am J. Fish. Manage. 12, 825–832. Cochran, W.G., 1977. Sampling Techniques, third edition. John Wiley, Sons. Cole, R., Coakley, J., Greco, M., Wong, R., 2009. Oyster Management White Paper – 2004, Updates Through 2009. Delaware Division of Fish & Wildlife, pp. 11. Cole, R.W., 1988. Natural Oyster Ground Survey – October 1988. Delaware Division of Fish & Wildlife, Dover, DE http://www.dnrec.delaware.gov/fw/Fisheries/ Pages/Fisheries.aspx. Delaware Division of Fish & Wildlife, 2010. Survey of Natural Oyster Beds. Delaware Department of Natural Resources & Environmental Control, Division of Fish & Wildlife, pp. 29 http://www.dnrec.delaware.gov/fw/Fisheries/Pages/ Fisheries.aspx. Efron, B., Tibshirani, R., 1993. An Introduction to the Bootstrap. Chapman & Hall/CRC, Boca Raton, pp. 436. Efron, B., 1987. Better bootstrap confidence intervals. J. Am. Stat. Assoc. 82, 171–185. Fegley, R.S., Ford, S.E., Kraeuter, J.N., Haskin, H.H., 2003. The persistence of New Jersey’s oyster seedbeds in the presence of oyster disease and harvest: the role of management. J. Shellfish Res. 22, 451–464. Ford, S.E., Haskin, H.H., 1987. Infection and mortality patterns in strains of oysters Crassostrea virginica selected for resistance to the parasite Haplosporidium nelsoni (MSX). Am. Soc. Parasit. 73, 368–376. Ford, S.E., Powell, E., Klink, J., Hofmann, E., 1999. Modeling the MSX parasite in Eastern oyster (Crassostrea virginica) populations. I. Model development, implementation, and verification. J. Shellfish Res. 18, 475–500. Ford, S.E., Cummings, M., Powell, E.N., 2006. Estimating mortality in natural assemblages of oysters. Estuaries Coasts 29, 361–374. Ford, S.E., 1996. Range extension by the oyster parasite Perkinsus marinus into the northeastern US: response to climate change? J. Shellfish Res. 15, 45–56. Ford, S.E., 1997. History and present status of molluscan shellfisheries from Barnegat Bay to Delaware Bay. In: MacKenzie, C.L., Burrell, V.G., Rosenfield, A., Hobart, W.L. (Eds.), The History, Present Condition, and Future of the Molluscan Fisheries of North and Central America and Europe. U. S. Department of Commerce (NOAA Technical Report NMFS 127). Grave, C., 1912. A Manual of Oyster Culture in Maryland. Fourth Rep. Board Shell Fish Commissioners, Baltimore, Maryland, pp. 75. Greco, M.J., 2009. Natural Oyster Ground Survey, 2008. Delaware Department of Natural Resources & Environmental Control, Division of Fish & Wildlife, Dover, Delaware (34) http://www.dnrec.delaware.gov/fw/Fisheries/Pages/Fisheries. aspx. Haddon, M., 2001. Modelling and Quantitative Methods in Fisheries. Chapman & Hall/CRC, Boca Raton, pp. 406. Haskin Shellfish Research Laboratory, 2008. Report of the 2008 Stock Assessment Workshop (10th SAW) for the New Jersey Delaware Bay Oyster Beds. Haskin Shellfish Research Laboratory, Rutgers University, Bivalve, NJ111 http://hsrl. rutgers.edu/SAWreports/SAW2008. pdf. Haskin Shellfish Research Laboratory, 2013. Report of the 2013 Stock Assessment Workshop (15th SAW) for the New Jersey Delaware Bay Oyster Beds. Haskin Shellfish Research Laboratory, Rutgers University, Bivalve NJ133 http://hsrl. rutgers.edu/SAWreports/SAW2013.pdf. Kraeuter, J.N., Ford, S.E., Cummings, M., 2007. Oyster growth analysis: a comparison of methods. J. Shellfish Res. 26, 479–491. Lenihan, H.S., Peterson, C.H., 2004. Conserving oyster reef habitat by switching from dredging and tonging to diver-harvesting. Fish. B-NOAA 102, 298–305. Mann, R., Southworth, M., Harding, J.M., Wesson, J., 2004. A comparison of dredge and patent tongs for estimation of oyster population. J. Shellfish Res. 23, 387–390. Mann, R., 2010. Quantitative sampling of oyster populations and habitat for fishery stock assessment and ecological restoration. In: Sellner, K.G. (Ed.), Virginia Oyster Restoration Review Workshop Summary. Chesapeake Research Consortium Ref#10-171, Edgewater, MD, p. 67. Moore, H.F., 1911. Condition and extent of the natural oyster beds of Delaware. U.S. Bur. Fish. Doc. 745, 29. Powell, E.N., Klinck, J.M., Hofmann, E.E., Ford, S.E., 1997. Varying the timing of oyster transplant: implications for management from simulation studies. Fish. Oceanogr. 6, 213–237. Powell, E.N., Ashton-Alcox, K.A., Dobarro, J.A., Cummings, M., Banta, S.E., 2002. The inherent efficiency of oyster dredges in survey mode. J. Shellfish Res. 21, 691–695. Powell, E.N., Kraeuter, J.N., Ashton-Alcox, K.A., 2006. How long does oyster shell last on an oyster reef? Estuar. Coast. Shelf Sci. 69, 531–542.

300

F. Marenghi et al. / Fisheries Research 186 (2017) 292–300

Powell, E.N., Ashton-Alcox, K.A., Kraeuter, J.A., 2007. Reevaluation of Eastern oyster dredge efficiency in survey mode: application in stock assessment. N. Am. J. Fish. Manage. 27, 492–511. Rochowicz Jr., J.A., 2011. Bootstrapping analysis, inferential statistics and EXCEL. Spreadsheets Educ. (eJSiE) 4 (3) http://epublications.bond.edu.au/ejsie/vol4/ iss3/4. Rothschild, B.J., Ault, J.S., Goulletquer, P., Herald, M., 1994. Decline of the Chesapeake Bay oyster population: a century of habitat destruction and overfishing. Mar. Ecol. Prog. Ser. 111, 29–39. Smith, S.J., 1996. Analysis of data from bottom trawl surveys. In: Lassen, H. (Ed.), Assessment of Groundfish Stocks Based on Bottom Trawl Surveys. NAFO Scientific Council Studies, 28, pp. 25–53. Southworth, M., Harding, J.M., Wesson, J.A., Mann, R., 2010. Oyster (Crassostrea virginica Gmelin 1791) population dynamics on public reefs in the Great Wicomico River, Virginia, USA. J. Shellfish Res. 29, 271–290.

Tarnowski, M., 2005. Maryland Oyster Population Status Report, 2003 and Fall Surveys. Maryland Department of Natural Resources Shellfish Program and Cooperative Oxford Laboratory, Maryland Department of Natural Resources, Annapolis, MD33, Publ. No. 17–1072005-62. Volstad, J.H., Dew, J., Tarnowski, M., 2008. Estimation of annual mortality rates for eastern oysters (Crassostrea virginica) in Chesapeake Bay based on box counts and application of those rates to project population growth of C. virginica and C. ariakensis. J. Shellfish Res. 27, 525–534. Whitmore, W.H., Cole, R.W., 2003. Delaware Commercial Shellfish Harvest and Status of the Fisheries 2001–2002. Division of Fish & Wildlife, Dover, Delaware. Whitmore, W.H., Greco, M.J., 2005. Delaware Commercial Shellfish Harvest and Status of the Fisheries 2003–2004. Division of Fish & Wildlife, Dover, Delaware. Wilson, B.D., Bruce, D.G., Madsen, J.A., 2006. Mapping the distribution and habitat of oysters in Delaware Bay. 26th Annual ESRI International User’s Conference Technical Paper, 39.