Study Area.

In 2008 and 2009, we accompanied a waterman while he trapped a 22-km² area along the Nanticoke River (Dorcester and Wicomico counties) on the eastern shore of the Chesapeake Bay in Maryland (∼ 38.47991, −75.81641, WGS84). Trapping locations occurred in small estuaries (salinity 0–6 parts per thousand; Vee Gee STX-3 refractometer) and covered several kilometers from the main river branch.

Trappable Area.

We collected coordinates (18N UTM) for each trapping location to measure the trappable area available to the waterman. A “general” trappable area was created by placing a minimum bounding polygon around the trap locations that conformed to the shoreline of the river, streams, and tributaries. This general polygon was used as a boundary to clip features from the National Hydrography Dataset waterbody polygons (http://datagateway.nrcs.usda.gov, accessed 13 April 2016), resulting in a trappable area along bodies of water. The trappable area polygon was overlaid on an aerial image of the surrounding area and the perimeter of the polygon was manually edited to match the water boundaries of the aerial image, to show the potential trappable area along the bodies of water in the bay. Data layers were edited and manipulated in ArcGIS v.10.2 (ESRI Inc, Redlands, CA). Distances between trapping locations were determined using the “pointdistances” tool in Geospatial Modelling Environment (Beyer 2012).

Harvesting Methods.

The waterman we observed relied entirely on standard hoop-net–style traps (see Plummer 1979) baited with American gizzard shad (Dorosoma cepedianum), which was both fresh and frozen on different occasions, and cut behind the head to expose blood. Our discussions with other watermen in the Working Group indicated that this appears to be the standard method used by most turtle trappers in Maryland. Typically, 15–16 hoop traps were deployed at any one time and the locations of the traps were moved several kilometers after each trapping period (ca. every 48 hrs), so that no location was trapped more than once per season. The waterman claimed that moving traps put the least amount of harvesting pressure on any particular area for that season (J. Edwards, pers. comm., May 2008), although these were anecdotal claims. Traps were checked every 48 hrs and contained plastic flotation objects (e.g., milk or laundry soap bottle) to prevent turtle drowning. The traps were allowed to free-float with the changing tide by tying the trap to a polyvinyl chloride pipe inserted into the substrate.

Size structure data on harvested snapping turtles were collected by measuring the curved carapace length (CCL) of captured turtles by measuring the sagittal distance along the surface of the top shell to the nearest 1 mm with a measuring tape. Body mass was measured to the nearest 0.1 kg using a Pesola® scale. Males were identified as having a cloacal opening posterior to the carapacial rim (Mosimann and Bider 1960 for diagrams; Ernst et al. 1994), and all turtles < 20 cm CCL were considered to be juveniles (Mosimann and Bider 1960; approximation from Christiansen and Burken 1979). Turtles under the minimum length (or size) limit were not harvested and released. Using a Dremel® tool, we gave all released turtles a permanent mark on the carapace following Ernst et al. (1974). We calculated catch per unit effort (CPUE) for 2008 and 2009 catches by dividing the total number of captured turtles that were above the minimum size limit (9.5 inches [24.1 cm] in 2008 and 11 inches [27.9 cm] in 2009) by total effort (trap-hours).

Wholesale Turtle Dealer.

To assess the harvest on a broader spatial scale, we visited a wholesale turtle dealer and distributor of turtles in Millington, Maryland (Turtles DeLuxe, Kent County), in 2008 and 2009 to record information from snapping turtles brought from several other sites in Maryland and other states to this central processing and distribution business. Population size data were collected as described above only on Maryland-harvested snapping turtles. Turtles were not marked during this part of the study because they were either killed or shipped to markets after we had handled them.

Statistical Methods.

We conducted statistical tests using Program R 2.12.1 (R Core Development Team 2010, www.cran.r-project.org) with α = 0.05. We used 1-way analysis of variance (ANOVA) to test for differences between male and female size (CCL) and to test for differences in CPUE between years. Residuals of ANOVAs were tested for normality using a Shapiro-Wilk goodness-of-fit test and for homogeneity of variances using a Bartlett's test because of differences in sample sizes. We used a contingency table analysis (VassarStats) corrected for continuity to test the association between year and sex. We used a Wilcoxon signed-rank test to test for significant differences in population structure (size classes) between tidal and nontidal habitats because residuals were not normally distributed. Association between site and sex ratio was tested using a contingency table. All means reported are presented as ± standard error (SE).

Trappable Area.

Of 82 trapping locations, the estimated trappable area for the waterman was 2047 ha for 2008 and 2009. The distance between successive trapping locations averaged 1372 m and ranged from 120 to 16,551 m. The distance between all trapping locations averaged 10,286 m and ranged from 120 to 26,931 m.

Population Sex Ratio and Size Structure.

The waterman we observed captured 199 turtles (147 males, 52 females) over 4 sampling trips (8 d total) in May 2008 and 414 snapping turtles (235 males, 179 females) in 8 sampling trips (16 d total) in May 2009 (Table 1). Sex ratios differed between years (χ²₁ = 16.03, p < 0.0001). Mean CCL in 2008 was 27.1 ± 0.36 cm and 29.5 ± 0.38 cm for females and males, respectively. In 2009, female and male mean CCL was 27.0 ± 0.23 cm and 30.6 ± 0.28 cm, respectively. Male CCL was larger than females in both years (2008: F_1,198 = 13.11, p = 0.0004; 2009: F_1,413 = 87.12, p < 0.0001). There was no difference in female CCL between years (F_1,229 = 0.02, p = 0.88), but males in 2009 were approximately 3.5% larger in CCL than in 2008 (F_1,380 = 5.17, p = 0.02). For 2008 and 2009, mean male mass was 5.5 ± 0.17 and 6.0 ± 0.17 kg, respectively, while mean female mass was 4.33 ± 0.16 and 3.94 ± 0.11 kg, respectively. When correcting for CCL with analysis of covariance (ANCOVA), we found an interaction between CCL and log-transformed body mass (t = 6.04, p < 0.001). An interaction plot showed that males were heavier than females of the same length until approximately 28 cm CCL, similar in mass between 28 and 30 cm CCL, and weighing less than females when > 30 cm CCL.

Table 1. Hypothetical snapping turtle size limits (inches) with effective size class protected based on combined 2008 and 2009 harvester data from Maryland. Columns show cumulative totals and percent protected (reproductive individuals only). For example, in 2008, 10.0 inches would protect 9 females and 26 males, representing 17% of all females and 18% of all males harvested, respectively.

View Fullscreen

Effects of Size Limits.

Through 2008, Maryland DNR required a minimum size for harvested snapping turtles of 9.5 inches CCL (24.1 cm). This resulted in the protection of 13% of the turtles captured (Table 1; i.e., 25 of 199 turtles captured were returned to the water as being legally undersized for harvesting). The percent protected from harvesting between the sexes did not differ significantly in 2008; of 52 females captured, 14% were too small to be harvested, compared with 12% of 147 males (χ²₁ = 0.05, p = 0.82).

Based on our data from 2008, the Working Group concluded that the size limits in effect were not protecting a sufficient proportion of females and voted to increase the minimum size legal for harvesting in 2009 to 11 inches CCL (27.9 cm). This regulatory change had a significant impact on the turtle harvest. Using data from the same waterman from 2008, this change in size limit resulted in protecting 60% of 179 females and 26% of 235 males captured (Table 1). Unlike 2008, size limits were significantly more likely to protect females than males in 2009 (χ²₁ = 48.7, p < 0.001). In addition, the waterman elected to release an additional 45 turtles (7% of total catch over 2 yrs) because they were marginally above (< 1 inch [2.54 cm]) the legal size limit and he wanted to avoid erroneously harvesting any turtles that were below the size limitations.

Catch Rates.

Mean CPUE for 2008 and 2009 seasons were 5.76 ± 0.28 and 3.37 ± 0.36 legal turtles/100 trap-hrs, respectively (Fig. 1). The CPUE was significantly lower in 2009 (ANOVA; F_1,11 = 18.61, p = 0.0015). Data from July 2009 were not used in this analysis to prevent any potential influence of seasonality.

View Figure

• Download as Powerpoint
• Download Figure

Wholesale Market Data.

The turtle dealer in Millington was the major wholesale buyer of snapping turtles harvested from the Mid-Atlantic States and some turtle harvesters came from outside of the region (e.g., from Ohio, Connecticut) to sell their catch (M. Johnson, Turtles DeLuxe, pers. comm., June 2008). Even turtles purchased directly from watermen by other local Maryland seafood dealers (e.g., Cool Ice Company, Cambridge, Maryland) sell to this particular dealer (J. Edwards, waterman, pers. comm., May 2009). Harvested turtles purchased during and after June were primarily from states other than Maryland, supporting the unofficial season observed by Maryland turtle harvesters (M. Johnson, pers. comm., June 2008).

In 2008, large males were butchered immediately upon arrival at the dealer, whereas females and small males were kept alive in a cold storage unit in plastic containers (2 or 3 per container) or large plastic drums (10–15/drum). Purchases ranged from 4500 to 9000 kg/d in May and June (M. Johnson, pers. comm., June 2008). Females were sold alive to aquaculture businesses both domestic and abroad. In 2008, the only snapping turtles available to us for data collection at the dealer were females (mean CCL, 28.4 ± 0.23 cm; mean mass, 4.5 ± 0.11 kg; n = 159), but these females were stored communally and origin could not be determined for each individual. However, in 2009, handling of turtles was different in that turtles were separated by size instead of sex, with larger individuals of both sexes butchered immediately for meat and all others being shipped live for sale in China. Sales to aquacultures in Maryland were not reported in 2009.

Prices (presented in US dollars) were determined by live weight and typically ranged from $1.35 to $2.00/lb ($2.98–$4.41/kg) per individual, regardless of size or sex. In July 2008, however, prices at Turtles DeLuxe were reduced below $1.00/lb ($2.20/kg; below typical market value). This reduction was an attempt to dissuade harvesters from collecting more turtles (M. Johnson, pers. comm., July 2008) because harvesters were providing more turtles than the dealer could sell.

Per capita price (based on live mass) paid to watermen was directly related to turtle mass, with most sales being between $1 and $2/lb ($2.20–$4.41/kg; J. Edwards, pers. comm., May 2008). Total gross income per trapper was estimated to be between $14,000 and $28,000 for a 31-d season, with an estimated daily harvest of 450 lbs (204 kg) over 31 d at $1–$2/lb. A regression analysis showed a positive exponential relationship between CCL and per capita price (R² = 0.8622), with the equation per capita price = 0.4407e^0.1087(CCL).

Data from Waterman.

Although we worked with only one waterman for field data collection in 2008–2009, conversations with him and other watermen in Maryland indicated that the practices he followed were similar to those used by most (but not all) of the harvesting community. Working with more watermen would have been beneficial, but their participation in the study was strictly voluntary and no other individuals came forward.

Overall, harvesting practices were different from what we expected in several aspects. First, there is an unofficial, but predictable, self-imposed season on snapping turtle harvesting in Maryland that typically begins in late April (although as early as late March) and ends between late May and early June. Most commercial turtle harvesters we spoke with focus on collecting snapping turtles as a “bridge” between other, larger fisheries, such as oysters (Crassostrea virginica) during the winter months and blue crab (Callinectes sapidus) from June through September (J. Edwards, pers. comm., May 2008). We were unable to find any waterman who harvested snapping turtles for the turtles' entire active season, and few did so after blue crab harvesting began. Evidence of this unofficial season was seen at the wholesale turtle dealer we worked with, where, after the beginning of June, a majority of snapping turtles purchased were from Delaware and Virginia, not Maryland. Why trappers in Delaware and Virginia would have different trapping seasons from those in Maryland is not known. The turtle dealer also saw an overall decline in purchases starting at the end of July (M. Johnson, pers. comm., June 2008).

Based on this information, the Working Group decided that any regulations based on a limited access in the spring could undermine profitability of the harvest. Although reliance on such a self-imposed season could be problematic if economic conditions made harvesting snapping turtles more profitable year-round, the Working Group felt that imposing such regulations at this time was not required.

Besides the apparent economic and opportunistic appeal of trapping snapping turtles when other fisheries are closed or are not profitable, harvesting is also driven by turtle body size and activity patterns. Brown and Brooks (1993) attributed differences in movement activity between male and female snapping turtles to sexual strategies, with males having more and longer movements during the mating season in April and May. Males engage in combat and will compete with other males for access to females, so male activity is greater as they actively seek out females (Kiviat 1980; Brown and Brooks 1993). Snapping turtles also show sexual size dimorphism, with males being larger than females (Kiviat 1980; Galbraith et al. 1988; present study). Thus, targeting the larger sex during the height of activity patterns makes seasonal income from snapping turtle harvests higher during the reproductive season. Furthermore, market pricing is determined by the live mass of each turtle. Therefore, larger individuals will give a higher per capita income than relatively smaller individuals, meaning harvesters have an incentive to target large males during periods of more activity (Close and Seigel 1997).

The interim size limit in 2008 in Maryland protected male and female snapping turtles equally, but protected only a minority of both genders. The increased 11 inches CCL (27.9 cm) minimum size limit was put in place in 2009 with the goal of protecting 50% of reproductive females. This was, in fact, successful (Table 1), but negatively affected the harvest rates achieved by the watermen. During the 2009 season, CPUE was reduced significantly and at least some harvesters increased effort by fishing more traps over a larger area each day (J. Edwards, pers. comm., July 2008). Effort, however, was limited considerably because of the fixed amount of time available for reaching traps during high tide. For instance, the number of hours one trapper can realistically spend emptying traps is 5–6 hrs during high tide because water levels are too low for boat access at other times (P.W.C., pers. obs.). Thus, access to traps may act as an indirect way of limiting harvest effort. In addition, because tidal areas represent a small portion of the range of snapping turtles in Maryland, a source–sink dynamic (Pulliam 1988) may exist between the tidal harvested turtles and the nonharvested nontidal turtles, potentially allowing a sustainable harvest situation. However, we stress that direct evidence neither supports nor refutes this possibility, so the effect this limitation has on the recovery of population size and the potential existence of source–sink dynamics should be investigated further.

In addition to the physical limitations of harvesting in tidal areas, the waterman we observed claimed he moved the locations of each individual trap up to several kilometers after it was recovered and before it was reset, allowing for just a single trapping period each year at any particular area (J. Edwards, pers. comm., July 2009). The spatial analysis shows that traps were placed over a relatively large area (2047 ha) and distances among trapping locations (mean = 1372 m; range = 120–16,551 m) exceed many long-distance turtle movements recorded in the literature. For example, Hammer (1969) saw females in South Dakota move distances > 3.4 km, but most individuals moved significantly less. Similarly, a nesting female in Canada was observed to move 2.02 km, a much greater distance than any male movement (Pettit et al. 1995). We are hesitant to compare our results with the literature because a majority of studies investigating the spatial ecology of snapping turtles have focused on freshwater lakes and streams in Michigan and Canada (Obbard and Brooks 1980, 1981; Congdon et al. 1987; Pettit et al. 1995), where the dynamics may differ greatly from the tidal estuaries in Maryland. A management plan limiting pressure of harvests spatially would be very difficult to enforce.

Harvest Market Data.

The rather centralized market for snapping turtles in the Mid-Atlantic was unexpected, especially given the fact that trappers came to this site not only from Maryland, but from states as far away as Ohio and Connecticut. Prices paid per unit mass varied widely, even during the 2 yrs of this study, suggesting that the profitability of the snapping turtle commercial harvest was highly responsive to market forces. However, gross income for some trappers could be extremely high for a short season fishery, suggesting that snapping turtle harvesting could be profitable for experienced trappers where turtles remain abundant. In addition, the disparate destination for harvested turtles in 2008 (females to China, males butchered for meat) versus 2009 (large animals of both genders butchered) shows the impact of global market forces on demand for different components of the population. Finally, the owner of the turtle dealership we visited was convicted in 2012 by the state of New York for attempting to traffic in prohibited wildlife. The turtle dealership was sold to a new owner, who now continues its operation as of 2014.

We believe the Working Group represents a successful cooperation of stakeholders. The Working Group made decisions on recommendations only after considerable discussion that included all parties. The need for field data on local snapping turtle populations, to allow for accurate management decisions, was considered a priority among the Working Group members. Although successful as a stakeholder meeting, the success for the sustainability, in terms of survival of snapping turtles and continuation of natural processes that may be affected by snapping turtle harvests, must be further investigated. In addition, the specific definition of sustainability goals (species-, population-, or ecosystem-specific) of the Working Group was not discussed (see Mace and Reynolds 2001). However, the diversity of the members of the Working Group likely ensured a wide variety of interests were represented.