Study Area

The Trident Turning Basin is part of the Naval Ordnance Testing Unit facilities located within the Cape Canaveral Air Force Station on the north side of the Port Canaveral, Brevard County, Florida (Fig. 1). Constructed in the mid-1970s, it is a man-made embayment approximately 600 m wide × 1200 m long, encompassing approximately 0.8 km². The entire perimeter of the basin, with the exception of a concrete seawall behind a large wharf on the east side, is lined with granite boulder riprap. There are 4 areas of open water behind the wharf. A narrow shelf extends along the perimeter of the basin, with water depths at high tide ranging from 0.5 m along the north wall to 5 m along the southwest wall. The water depth in the main area of the basin is approximately 13 m. The substrate of the basin is a mixture of sand, shell, and mud. The basin is close to the ocean entrance to Port Canaveral, allowing the unrestricted circulation of seawater with the ebb and flow of tides. A wedge-shaped mat of macroalgae covers the intertidal and subtidal areas of the riprap boulders along the southwest, west, north, and northeast walls of the basin. Approximately 5 m in vertical width at the southwest corner of the basin and almost completely within the subtidal zone, it becomes progressively narrower with a greater proportion of it in the intertidal zone as one progresses along the southwest and west walls to the north and northeast walls. The bulkhead behind the wharf is largely devoid of macroalgae. The subtidal and intertidal portions of the boulder riprap along the east wall of the basin south of the wharf have been covered by the accretion of sand over the course of the study.

Capture

Turtles were captured using tangle nets and dip nets. The tangle nets consist of multifilament nylon twine mesh hung from a braided polypropylene top line and a braided polypropylene continuous lead-core bottom line. Two tangle nets were deployed at various locations over the shallow shelf along the perimeter of the basin during each netting session: one 238 m in length, 3.7 m in depth, with a 40-cm stretch mesh size (knot to knot) and the other 229 m in length, 3.7 m in depth, with a 30.5-cm stretch mesh size. The nets were suspended from floats attached at regular intervals to the top line during deployment and frequently checked by pulling hand over hand along the top line from the bow of a 4.6-m boat. The dip nets were long handled (2.5 m) with large hoops (0.8 m in diameter). Two to 4 individuals worked from the bow and gunnels of a 5.2-m boat using the dip nets to opportunistically capture turtles feeding or swimming along the margins of the basin. Net soak time and capture effort with dip nets varied per day from < 2 to > 6 hrs.

Following capture, turtles were transported to shore, where the SCL (i.e., nuchal notch to tip of pygal scute), straight carapace width, head width, and body depth were measured with forestry calipers. The curved carapace length, curved carapace width, plastron length, plastron to vent length, and plastron to tail-tip length measurements were made with a fiberglass cloth tape. All measurements were to the nearest 0.1 cm using protocol described by Bolten (1999). The body mass of each captured animal was measured to the nearest 0.1 kg using a digital scale. All turtles were examined for scute aberrance, injury, and evidence of disease. When captured for the first time, each turtle was tagged externally on the trailing edge of the front flippers using inconel alloy flipper tags supplied by the Archie Carr Center for Sea Turtle Research, University of Florida, Gainesville. A passive internal transponder tag (AVID brand manufactured by Destron-Fearing Co) was inserted subcutaneously in the right front flipper proximal to the wrist joint. If needed, flipper tags were replaced on recaptured turtles. Dorsal and ventral views of turtles were photographed for all captures. The turtles were released back into the basin the day they were captured, except those whose injuries or poor physical condition appeared to negatively affect their reasonable chance of survival in the wild. These turtles were transported to a state sanctioned care facility for examination and rehabilitation via coordinating through the Florida Fish and Wildlife Conservation Commission.

A total of 146 d of capture effort were expended over the 18 yrs of data collection in the Trident Basin (Appendix 1). Effort necessarily varied over the years because of Navy operations, weather conditions, and funding.

Statistical Analysis

To explore the possibility of change in the distribution of green turtle body sizes, the SCL data from individuals were grouped by each calendar year of the study. Linear regression was used to look for a trend in mean SCLs obtained by year.

The data set for each year included the SCL measurements from turtles captured for the first time during that year and not recaptured, from the last capture of intrayear recaptures, and from turtles initially captured in a previous year and then recaptured during the year of that data set. Although this might violate the assumption of independence between the years by including interyear recaptures, including those recaptures provided a more representative sample of turtles in the basin during each year as opposed to using first-time capture data alone. For comparison, data sets using the same criteria were constructed for juvenile green turtle SCL data obtained at our cIRL study site (Fig. 1).

Nonlinear regression was used to look for trends in the proportion of first-time captures vs. recaptures, and in the proportion of turtles at last recapture that were < 35 cm SCL vs. those ≥ 35 cm SCL. The data were compiled by year from 1993 to 2011.

The Fisher's exact test was used to determine whether the decrease in capture effort over the last 8.5 yrs of the study (when the mean number of capture days dropped to 4.7/yr from 10.4/yr for the previous 9.5 yrs; Appendix 1) affected the proportion of first-time captures vs. recaptures and the time interval between first-time and most recent recapture.

To look for a trend in abundance, the mean population estimates per year from June 1998 to August 2011 were plotted using linear regression. An extraordinarily high estimate in February 2011 pulled the regression line upward, so the data were analyzed both with and without that estimate. We used Kruskal-Wallis to determine whether there was seasonal variation in population estimates. The estimates were compiled into four data sets, July–September (Summer), October–December (Fall), January–March (Winter), and April–June (Spring). The data were analyzed both with and without the February 2011 estimate. Dunn's multiple-comparison test was used to determine which of the seasonal estimates were significantly different. Statistical analyses were calculated using Graphpad Prism for Mac OS X 5.0d (Graphpad Software, San Diego, CA).

Population Estimates

Starting in June 1998 (but excluding the capture sessions in June, August, and September 2002 due to a lack of funding), the size of the Trident Basin juvenile green turtle population was estimated after each capture session using the Lincoln-Peterson mark and recapture method for closed populations. Each of the green turtles caught and returned to the basin during a capture session was marked with a unique number painted on the carapace with nontoxic white paint. A visual survey was conducted within 2 d of a capture session by slowly cruising around the perimeter of the basin in a 4.6-m boat with two observers in the bow. The marked and unmarked turtles observed were treated as “captures”. With such a short time interval between the conclusion of the capture session and the visual survey, we considered the effect of any immigration or emigration of individuals on the estimate to be negligible, thus meeting the assumption of a closed population. The population estimate was calculated using the formula:

where

  =  the estimate of the population size at the time of marking,

  =  the number of individuals marked during the netting session,

  =  the total number of individuals observed during the follow-up survey, and

  =  the number of marked individuals observed during the follow-up survey.

A 95% confidence interval is constructed for each estimate using the formula

Captures

The Trident Basin was almost exclusively used by juvenile green turtles. From 17 June 1993 through 12 August 2011, there were 1560 captures of juvenile green turtles in the Trident Basin. Of those, 770 (49.4%) were first-time captures and 790 were recaptures. The only other marine turtle species captured were 10 juvenile loggerheads, all of which were severely emaciated, and a juvenile hawksbill (Eretmochelys imbricata) hand-captured and released in the basin during an underwater fish survey in March 2009 (K. Holloway-Adkins, pers. comm., March 2009).

Population Size Classes

The SCL measurements of the first-time captures ranged from of 20.0 to 48.4 cm, with a mean of 29.6 cm, 3.7 standard deviation (SD). The SCL measurements of all green turtle captures ranged between 20.0 and 52.0 cm, with a mean of 31.7 cm, 5.4 SD. Figure 2a reveals a significant decrease in Trident Basin green turtle mean SCL over the course of the study (R²  =  0.5162, p  =  0.0005), whereas the cIRL green turtle mean SCL (Fig. 2b) did not show a meaningful trend (R²  =  0.1507, p  =  0.1005).

View Figure

• Download as Powerpoint
• Download Figure

Residency

As noted above, 790 captures were of 252 turtles that had been previously captured in the basin 1 or more times (mean  =  3.2 recapture events, range  =  1–22 recapture events). The shortest interval between initial and most recent recapture was 0.08 yr (29 d), and the longest interval was 12.4 yrs (4539 d). The mean interval between initial capture and most recent recapture was 2.3 yrs. There was a significant decrease in proportion of recaptures (R²  =  0.6448; Fig. 3a) and a significant decrease in the proportion of recaptures of turtles with ≥ 35 cm SCL (R²  =  0.5389; Fig. 3b) over the span of the study. After recording a relatively uniform increase in long-term recaptures over the first 13 yrs of study, a gap between short-term (< 3 yrs) and long-term (> 7 yrs) recaptures occurred in 2006 (Fig. 4). Between 2008 and 2011, we had only 1 recapture of > 3 yrs (3.7 yrs). Comparing the days of capture effort and number of captures between 1993 and 2002 (104 d and 966 captures) with those between 2003 and 2011 (42 d and 594 captures), a significantly greater number of captures per day of effort occurred during the latter time period (9.3 vs. 14.1, respectively; Fischer's exact test 2-tailed p  =  0.0314; i.e., despite the reduced effort, we encountered more turtles, so should have recaptured individuals in short-term, intermediate, or long-term residence if they had been present).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Population Estimates

The 42 Lincoln-Peterson green turtle population estimates from June 1998 through August 2011 ranged from 27 to 224 turtles, with a mean and 95% confidence interval of 61 ± 10 (Fig. 5a). Linear regression of the annual estimates with the February 2011 estimate (224) included (R²  =  0.1957, p  =  0.1132) and excluded (R²  =  0.0166, p  =  0.6610) showed no significant trend in population (Fig. 5b). Analysis of the seasonal estimates with the February 2011 estimate included revealed a significantly higher population in the basin during the spring compared with the fall, but not with that in the winter or summer at p ≤ 0.05 (Fig. 5c; K-W  =  11.24, p  =  0.0105; Dunn's multiple-comparison spring vs. fall p  =  0.0381, spring vs. summer p  =  0.0567, spring vs. winter, and winter vs. fall and summer p > 0.1). With the February 2011 estimate excluded, the spring population was significantly higher than both the summer and fall estimates, but not that of the winter at p ≤ 0.05 (K-W  =  10.7, p  =  0.0135; Dunn's multiple-comparison spring vs. fall p  =  0.0311, spring vs. summer p  =  0.0472, spring vs. winter, and winter vs. fall and summer p > 0.1).

View Figure

• Download as Powerpoint
• Download Figure

Population Size Class Distribution and Residency

This long-term study has documented a significant decline in the mean SCL of the Trident Basin green turtles. It has also documented a decline in the proportion of recaptures vs. first-time captures, the interval between first-time and most recent recapture, and in the SCL of recaptures over the past 18 yrs. Although the decline in overall recaptures and long-term recaptures might have been a result of the decline in monitoring effort during the last 8.5 yrs (Appendix 1), this is an unlikely explanation.

Rather, our findings are likely the result of 2 interrelated factors. First, the number of juvenile green turtles seeking developmental habitat on Florida's east coast has increased significantly as demonstrated by results of 2 other long-term, in-water abundance studies: one at our cIRL study site (Ehrhart et al. 2007), and the other in the nearshore waters of Hutchinson Island (Witherington et al. 2006). These increases correspond to the increase in green turtle nesting in Florida and Costa Rica over the past 25 yrs (Ehrhart et al. 2007; Chaloupka et al. 2008), and is supported by the findings of 2 genetic studies. Bagley (2003) found that the juvenile green turtles captured in the Trident Basin, at our cIRL study site, and over the nearshore Sabellariid worm rock reefs 4 km south of Sebastian Inlet came primarily from the Florida and/or Mexico (53%–94%) and Costa Rica (3%–40%) breeding genetic stocks. Likewise, Bass and Witzell (2000) determined that 42% and 53% of the sampled juvenile green turtles from the nearshore waters off Hutchinson Island came from the Florida and/or Mexico and Costa Rica breeding stocks, respectively.

Second, the green turtles in the Trident Basin are primarily foraging on the macroalgae growing on the rock riprap lining the basin (Redfoot 1997). The declines in mean SCL, proportion of recaptures vs. first-time captures, the interval between first-time and most recent recapture, and the SCL of recaptures are probably related to the limited biomass, and perhaps the nutritive value of the macroalgae in the basin. We have observed that the algal mat growing on the accessible areas of the rocks in the basin is closely cropped, while that growing in the cracks between rocks (inaccessible to even the smallest turtles) is uncropped. Kubis et al. (2009) analyzed the juvenile green turtle growth rates in 3 disparate developmental habitats on Florida's east coast, the Trident Basin, cIRL, and Sabellariid worm rock reef study sites noted above. They found that in the Trident Basin growth rates increased to a maximum at approximately 35 cm SCL and then declined. However, at both other sites, growth rates increased until turtles attained a maximum of 50–55 cm SCL, and then declined beyond that size. The authors attributed the limited growth of the Trident Basin turtles to the paucity of available algae. This suggests that available food resources in Trident Basin cannot sustain larger individuals, which might cause them to avoid the basin or to depart shortly after arriving.

Records of Trident Basin green turtles recaptured in other developmental habitats on Florida's east coast support this hypothesis. Five turtles tagged at Trident Basin have been recaptured at our cIRL study site, one in the cooling water intake canal at the St. Lucie Power Plant on Hutchinson Island, and one west of Haulover Canal in the northern Indian River Lagoon. In each of these instances, the turtle was either close to the upper limit of the Trident Basin size distribution when it was known to have left the basin or was relatively small when last captured in the basin and had grown appreciably when recaptured elsewhere (Table 1). Although these 6 turtles are only 0.1% of the 770 individual green turtles captured in the basin, considered together, they present a compelling anecdote suggesting that low nutrient availability in Trident Basin might hinder turtle growth and preclude residency of larger animals.

Table 1. Juvenile green turtles tagged in the Trident Basin, Florida, June 1993 to August 2011, and then recaptured at a remote location. SCL  =  straight carapace length.

View Fullscreen

The decline in green turtle SCL distributions and recapture rates at Trident Basin are in contrast to patterns observed among green turtles at our cIRL study site. We attribute the lack of a SCL trend and the consistently low recapture rate of 12.8% (Ehrhart et al. 2007) in the cIRL green turtle population to the greater biomass of macroalgae available to them and to their movement within the lagoon and between the lagoon and nearshore Sabellariid worm rock reefs (W. Redfoot, unpubl. data, 2009).

Though juvenile green turtles in the 20–52-cm SCL size classes have been recorded in other developmental habitats along the Atlantic and Gulf coasts of the United States (Table 2), the Trident Basin population is noteworthy in that the SCL range for all green turtle captures in the basin did not exceed 52.0 cm SCL and the mean was only 31.7 cm. The 2 places where the reported SCL means and the ranges were homologous to those of the Trident Basin green turtles also had similar habitats—the rock riprap jetties at Brazos Santiago Pass (Coyne 1994) and the Mansfield Channel in Texas (Shaver 1994). Coyne (1994) found that the green turtles in the Brazos Santiago Pass were foraging on the macroalgae growing on the jetty rocks. Shaver (1994) suggested that the green turtles around the Mansfield Channel jetties were also feeding on the macroalgae growing there. Moreover, the green turtles captured in the lagoon foraging habitat adjacent to Brazos Santiago Pass, South Bay–Mexiquita Flats (Coyne 1994), and cold-stunned green turtles in Laguna Madre, Texas (Shaver 1990) were similar in SCL means and ranges to the green turtles in Florida developmental habitats other than the Trident Basin.

Table 2. Carapace length data obtained from green turtles in developmental habitats along the US Atlantic and Gulf of Mexico coasts. Measurements are in centimeters. SCL  =  straight carapace length, MSCL  =  minimum straight carapace length, CCL  =  curved carapace length, TCL  =  total carapace length, *  =  converted from inches.

View Fullscreen

Population Estimates

The broad range of population estimates obtained over 13 yrs (Fig. 5a) were probably an artifact of environmental factors such as water clarity, water temperature, wind conditions, and other variables during capture sessions and visual surveys. Despite the obfuscating effect of the factors noted above on the data, two patterns emerged. As seen in Fig. 5b, the lack of a trend in annual population estimates indicates the Trident Basin is generally at or near its carrying capacity for green turtles. Second, the population is significantly higher during the winter and spring as compared with the summer and fall (Fig 5c). That contrasts to our cIRL study site, where capture rates (denoting a higher abundance of green turtles) are higher during the fall and winter as compared with those in the spring and summer (Ehrhart et al. 2007). Had we been able to pick only days with good environmental conditions to conduct capture sessions and visual surveys, the population estimates would probably have been more precise and greater abundance during the fall and winter might have been detected. Despite these complicating factors, the population estimates indicate that the basin generally serves as habitat for 61 ± 10 juvenile green turtles, and abundance sometimes exceeds 100 individuals.

Conservation Implications

The National Research Council's report on the assessment of sea turtle status and trends (2010) identified gaps in sea turtle demography data, including the lack of information on the juvenile age classes in neritic and estuarine habitats. It emphasized the need for those gaps to be filled in order to adequately predict population trends in these long-lived species. For example, the effect of higher than normal rates of mortality at an immature stage of a species life cycle may not be known until nest numbers start to decline 10–20 yrs later. If the SCL distribution of green turtles found in the Trident Basin and at Brazos Santiago Pass and the Mansfield Channel in Texas are the norm rather than the exception, then similar habitats along the coast of the United States from North Carolina to Texas may be providing developmental habitat, either seasonally or year-round, for a substantial number of small juvenile green turtles. Demographic data for juvenile green turtles are not only lacking for most ports and jetty-stabilized inlets, but are lacking for many nearshore and estuarine areas that may also serve as developmental habitat. As competition for available forage (as documented by this study) causes an increase in size-class partitioning and a decrease in residency, those habitats will become increasingly important to the continued recovery of this species. The insights into juvenile green turtle population characteristics and trends gained through long-term, in-water monitoring efforts such as ours underscore the value of such studies. They supply data needed to monitor and interpret sea turtle abundance trends, which are required to refine targets of management actions to more effectively promote population recovery.