All sea turtle species are listed as Threatened or Endangered under the US Endangered Species Act. Data from commercial fisheries have been used to quantify anthropogenic effects on sea turtle health and mortality and identify potential areas for bycatch mitigation (Lewison et al. 2004; Casale et al. 2008; Finkbeiner et al. 2011; Stokes et al. 2011). Conversely, recreational fisheries are an understudied area of sea turtle biology and current literature suggests that the most endangered sea turtle, the Kemp's ridley (Lepidochelys kempii) (International Union for Conservation of Nature 2015), is the predominant species represented in recreational bycatch in the United States (Rudloe and Rudloe 2005; National Marine Fisheries Service and US Fish and Wildlife Service [NMFS and US FWS] 2015; Coleman et al. 2016b; Seney 2016). The ingestion of hooks by sea turtles can decrease survival by increasing the risk of coelomitis, hemorrhaging, septicemia, and chronic fibrosis of the esophageal wall (Orós et al. 2004; Casale et al. 2008). Fishing line attached to hooks may also increase the potential for health risks as the normal peristaltic movements of the gastrointestinal tract (GIT) can tighten the line, causing intestinal plication, obstruction, ulceration, perforation, or intussusception (Bjorndal et al. 1994; Watson et al. 2005; Parga 2012). Because these health issues have the potential to decrease survival rates, understanding the effects of incidental capture on sea turtle populations is crucial in identifying areas of conservation and management concern.

Stokes et al. (2011) found that hook size played a significant role in ingestion probability in loggerheads (Caretta caretta), as smaller turtles were unable to ingest larger hooks. Steen et al. (2014) used radiographs to measure the prevalence of ingested hooks within populations of freshwater turtles and found that turtle size affected the hook ingestion rates of red-eared sliders (Trachemys scripta elegans) and snapping turtles (Chelydra serpentina), though this proved inconsistent as size effects were not noted in a separate population of C. serpentina. Ingesta transit rates can be used to detect obstructions or abnormal gastrointestinal motility. Ingesta transit times in C. caretta have been previously defined using barium (Di Bello et al. 2006; Valente et al. 2008), and Birse and Davenport (1987) found that C. caretta gut clearance time decreased as temperature increased. Casale et al. (2008) noted that only a small percentage of C. caretta captured by longlines were able to successfully excrete ingested fishing hooks. Though hook ingestion occurs in freshwater turtles and sea turtles, hook transit times (when passage is successful) have never been studied.

Though recreational fisheries interactions have been identified as a conservation concern for L. kempii in the northern Gulf of Mexico (nGOM) (Coleman et al. 2016a, 2016b), no work has been done to analyze the proportion of turtles affected. The Institute for Marine Mammal Studies (IMMS) in Gulfport, Mississippi, evaluates and rehabilitates sea turtles that have been incidentally captured by recreational fisherman along the Mississippi coast. The objectives of this study were to measure the prevalence and transience of fishing hooks in L. kempii in the Mississippi Sound. We addressed these objectives by 1) evaluating radiographs of turtles brought to IMMS for presence/location of fishing hooks, 2) monitoring turtles for passage of ingested fishing hooks (noting transit time when successful), and 3) analyzing how these data relate to turtle size, hook size, and position within the GIT.

METHODS

Sea turtles included in this study were admitted to IMMS from Mississippi waters from 2012 to 2015 for medical evaluation and rehabilitation. Most turtles were admitted following incidental hook-and-line capture. Some were also reported as incidental trawl, dredge, or entanglement captures or live strandings. NMFS defines a stranding as any turtle that washes ashore or is found floating, generally in a weakened condition (National Oceanic and Atmospheric Administration [NOAA] 2013). Radiographs (dorsoventral, cranial–caudal, and lateral views) and standard morphological measurements were taken for all turtles upon arrival at the facility. Analysis of turtle size was done using the distance from the midline of the nuchal notch to the posterior notch of the supracaudal scutes (mSCL). Turtles that presented with fishing hooks had their hooks removed if they were located externally, in the oral cavity, or at a position in the esophagus where the hook could be located and removed during initial examination (Kaletsch et al. 2014; Coleman et al. 2016b).

Imaging was done using a Quantum Quest HF QG-500-2 set at 68 kV and 200 mA. Hook presence and location were noted based on dorsoventral radiographs. In our analysis of the prevalence of fishing hooks we were only concerned with hooks that were present prior to the event for which the turtle was admitted. Radiographs were evaluated to ensure that single interaction events with multiple hooks were not included as multiple interactions, as several fishing rigs exist that utilize multiple hooks on one line (Fig. 1A). The hooks included in our analysis of passage were any hook, from the current capture event or prior, that was not removed during the initial examination of the turtle. Some turtles had ingested more than 1 hook, resulting in a larger number of hooks than turtles. Turtles were housed in saltwater tanks at similar conditions throughout their observation period, and were offered a regular diet of shrimp (Penaeidae spp.) and/or capelin (Mallotus vilosus) daily.

View Figure

• Download as Powerpoint
• Download Figure

Ingested hooks (located too deep to be seen or removed upon initial exam) were monitored for passage through repeated radiographs to gauge mobility, and by checking turtle enclosures for defecated hooks daily. Turtles with immobile hooks were monitored to ensure that they could eat and pass feces without hindrance, and in a small number of cases veterinarians chose to remove the hook surgically. If hook passage was successful, transit time was recorded and the hook was collected, measured, and identified as a J, circle, or treble hook. In addition, hooks were noted to be partial hooks if they were missing any part of the eye, shank, bend, or point. Only complete hooks were included in our measurement analysis. Hook length was measured as the distance from the top of the eye to the outside of the bite/throat, and hook width was measured at the widest point between the outer edges of the shank and bend. Turtles, once deemed to be in good health by a veterinarian, had flipper and passive integrated transponder tags applied as appropriate for their size and were released back into Mississippi waters (Coleman et al. 2016b).

For our analysis of transit time, only mobile hooks were evaluated. The locations of ingested hooks were placed into 3 groups based on intake radiographs. Upper GIT (T_U) was defined as appearing at or cranial to the second rib head, mid GIT (T_M) was defined as between the second and fourth rib heads, and lower GIT (T_L) was defined as at or caudal to the fourth rib head. If a hook appeared to be in between sections (e.g., the eye of the hook appeared below the second rib head, but the bend and barb appeared above it), then it was placed into the region that contained more than half the total length of the hook based on radiograph.

All statistical analyses were done using RStudio (version 3.2.2, RStudio Team 2015), and confidence intervals for the prevalence data were found using the binom package, method asymptotic (Dorai-Raj 2014). The a priori significance level was set as α = 0.05. We used the Kruskal-Wallace 1-way analysis of variance to examine the significance of mSCL on the prevalence of fishing hooks, and a linear regression model was used to examine the relationship between mSCL and ingested hook size. Multiple linear regression models were used to examine the relationships between transit time and position, transit time and mSCL, and transit time and hook size.

RESULTS

During our 4-yr study period a total of 631 unique L. kempii were admitted and radiographed a total of 882 times (including recapture events). Of these 882, 864 were brought to the facility as the result of recreational hook and line interactions, while 18 were admitted for other reasons. Of the 18, 13 were strandings, 3 were trawl captures, 1 was a dredge capture, and 1 was found entangled. Additionally, 133 catch-and-release events were reported.

Hook Prevalence

Of our total sample, 12.5% (95% CI, 10.4%–14.8%) of turtles displayed evidence of prior hook-and-line interaction. The annual proportions of L. kempii with gear from prior angling interactions were 9.1% (95% CI, 4.7%–13.5%), 11.6% (95% CI, 7.5%–15.8%), 16.3% (95% CI, 11.8%–20.7%), and 11.8% (95% CI, 7.5%–16.0%) for 2012, 2013, 2014, and 2015, respectively. While these percentages represent the proportion of the sampled turtles that had evidence of prior interaction, some turtles contained multiple hooks that were determined to be from multiple prior interactions (Fig. 1B). While 110 turtles contained hooks from prior events, there were 128 total prior events noted due to some turtles containing 2–3 hooks. Because of this the proportions of retained gear that were determined to be from previous events were 10.9% (95% CI, 6.1%–15.7%), 12.5% (95% CI, 8.2%–16.8%), 18.2% (95% CI, 13.5%–22.8%), and 13.1% (95% CI, 8.7%–17.6%) for 2012, 2013, 2014, and 2015, respectively. Two of 18 turtles admitted for reasons other than hook-and-line capture showed evidence of prior interactions with fishing gear. One was a stranding with hook and fishing line present in the oral cavity. The other turtle was found entangled in fishing line that extended from a hook embedded in the caudal esophagus to a nearby crab trap. Additionally, 145 of the 864 (16.8%) hook-and-line captured L. kempii were admitted to rehab with no fishing gear present (Table 1). This includes instances where the hook fell from the turtle without human intervention and those where the angler presumably removed the hook themselves.

Table 1. Prevalence of fishing hooks from incidentally hook-and-line captured Lepidochelys kempii presented for rehabilitation from the Mississippi Sound, including hooks from current and previous interactions.

View Fullscreen

The mSCL for the sampled L. kempii ranged from 19.2 to 63.0 cm (30.3 ± 3.86 cm SD), with the majority (87.5%) being juveniles between 25.0 and 35.0 cm. Carapace length was not significantly related to the probability of gear presence (H = 0.867, df = 1, p = 0.352). Thirty-eight ingested hooks were measured upon passage; length ranged from 1.20 to 6.67 cm and width ranged from 1.00 to 2.95 cm. A linear regression model showed that the effect of turtle size on the size of ingested hooks was related to hook width (p = 0.002) but not hook length (p = 0.063). Overall, the adjusted r² value of this model shows that carapace length accounts for 21.9% of the variation in hook sizes in our sample.

Passage and Transit Time

Seventy-five L. kempii with 82 hooks were monitored for hook passage and transit time. Transit times and embedded hook locations had linear relationships with position (T_U, T_M, and T_L). Of the 82 ingested hooks, 22 (26.8%) were noted as being immobile and showed no movement upon repeated radiographs. Of the immobile hooks, 17 were in the upper GIT, 4 were in the mid GIT, and 1 was in the lower GIT. We were unable to consistently identify fishing hook types by radiograph alone due to the various angles of hooks given in imagery.

Though 60 (73.2%) hooks passed successfully, 10 were not collected upon defecation. Successful passage was noted for these based on repeated radiographs, and transit time was recorded for 50 hooks. In our study of transit times, T_U (n = 12) was 13.50 ±  3.59 d SD with a range of 8–19 d, T_M (n = 15) was 8.40 ± 4.67 d SD with a range of 1–18 d, and T_L (n = 23) was 4.48 ± 2.76 d SD with a range of 1–11 d. The group of successfully passed hooks included J hooks (n = 41), circle hooks (n = 4), and partial hooks (n = 15) that were missing some part of the hook prior to time spent in rehabilitation. Generalized linear models showed that carapace length and hook size did not significantly affect transit time (Table 2).

Table 2. Generalized linear models of hook transit time (days) in relation to carapace length (mSCL) and hook size in Lepidochelys kempii. T_U is upper gastrointestinal tract (GIT), T_M is mid GIT, and T_L is lower GIT.

View Fullscreen

All 22 turtles with immobile ingested hooks were released as the ingested gear did not appear to impact their ability to process food during observation. Six turtles that were released with immobile ingested hooks were recaptured at later dates. These included 5 that were released with hooks in the upper GIT, and one that was released with a hook in the mid GIT. The mid GIT hook had passed successfully prior to the recapture, and the time between release and recapture was 12 d. One of the upper GIT hooks, observed 92 d after the turtle's initial release, was a small partial hook that included the barb and part of the bend. As it was in the same state upon initial capture, it is unclear if the hook was broken in its capture event or if it had been embedded and already degraded before the initial report and response. The other 4 showed clear signs of degradation. One turtle was initially in rehabilitation/observation for 280 d, during which time the hook appeared completely immobile and intact. Thirty-eight days after release this turtle was captured again, and radiographs revealed that the hook had degraded as the majority of the shank was no longer present (Fig. 1C–D). The other 3 turtles were recaptured after 27, 264, and 8 d. The eye and part of the shank had degraded in 2 hooks, and the third was missing a metal clasp that had been attached to the hook during its time in observation. All recaptured L. kempii previously released with embedded hooks were in good body condition and did not appear emaciated or otherwise debilitated.

DISCUSSION

While 12.5% of our sampled turtles had fishing gear present from previous interactions, 16.8% of turtles admitted after incidental hook-and-line capture had no gear present from their reported interactions. Because of this, it can be assumed that the proportion of L. kempii in the Mississippi Sound affected by fisheries interactions is likely higher than our calculated confidence interval range of 10.4%–14.8% and that many do not retain gear. Hooks that the turtles were caught with and reported from were not included as prevalence data because the turtles were not at large with the attached gear. Though wounds sustained from removing hooks and releasing turtles at the time of capture could still influence a turtle's survival by increasing the risk of infection, rehabilitation mitigates potential negative health complications. In addition to the 882 L. kempii sampled in our study, 133 catch-and-release events were noted. These include events where the turtles retain fishing gear (when the anglers cut the line to release the turtle, the turtle breaks the line, etc.) and where the turtles do not retain fishing gear (when the angler removes the hook and releases the turtle, the hook becomes dislodged before being landed, etc.). It is likely that some anglers would be more inclined to report interactions in which they cannot remove the hook themselves, and because of this our sample may not be a true representation of incidental capture hook locations.

Our analysis of ingested hooks returned results similar to the Stokes et al. (2011) finding that larger turtles were able to ingest larger hooks, and hook width was the limiting factor as it likely controls which hooks can be ingested. The small number of circle hooks in our sample complements the finding that hook width is the primary factor influencing ingestion probability as circle hooks have a wider bend than J hooks, proportional to hook size, but we do not know the ratio of circle to J-hooks used by anglers in this region. While it may be intuitive that prevalence of fishing gear would increase with size, as larger turtles can engulf a larger proportion of available hooks, our data show that hook prevalence rates do not change significantly in relation to L. kempii age/size. Mortality is inversely related to turtle size (Lutz et al. 2002), and the fact that most of the sea turtles involved in fisheries interactions in the Mississippi Sound are juveniles means that these interactions could exacerbate the high mortality rate of this age/size class (NMFS and US FWS 2015).

Our methods were similar to those used by Casale et al. (2008) in treating C. caretta caught on drifting longlines (i.e., removing hooks when able, and monitoring turtles' health when hooks are ingested). In this setting, mortality from ingesting fishing gear was high, and deaths were attributed mostly to hooks, and occasionally to attached branchlines. Casale et al. (2008) suggested a high (94%) mortality rate for C. caretta with ingested hooks from drifting longlines, with only 6 successfully expelled hooks seen during their study. Our observations of L. kempii with ingested hooks suggest far fewer complications, but it must be considered that many hooks are removed from individuals as part of the rehabilitation process prior to release and our sample is mostly angled turtles that are actively foraging (turtles that are not feeding are unlikely to be caught). These differences in ingestion effects could also be attributed to the differences in the fisheries' methods and gears. Drifting longlines have hours of soak time which may allow a sea turtle time to deeply ingest hooks, which could potentially result in substantially more damage during gear retrieval, due to the hooks penetrating the lining of the stomach or intestine instead of the esophagus (Orós et al. 2004; Parga 2012). In addition, these turtles are usually larger, and are lifted by the line onto the boat before being released. In a recreational fishery setting, a sea turtle would likely be reeled in soon after the turtle ingests the bait/hook, thus resulting in a higher proportion of shallow hookings compared to more passive methods of fishing. While hook wounds in the oral cavity and upper esophagus can still cause health issues, their survival rates have been found to be higher and they are more easily treated in rehabilitation (Casale et al. 2008). While our sample from the Mississippi Sound appears to show less impact from fishing gear than studies of commercial fisheries elsewhere, it is important to note that our sample was largely derived from hook-caught turtles that were actively foraging (e.g., anorexic turtles would not be sampled).

Sea turtles in commercial fisheries also face increased mortality as a result of fishing line attached to hooks. Fishing line is often considered the most dangerous part of the gear (Bjorndal et al. 1994; Casale et al. 2008; Parga 2012). Ingested fishing line can cause intussusception, or when anchored to an immobile hook, can travel into the intestine where peristalsis causes it to tighten and sever the intestinal wall (Casale et al. 2008; Parga 2012). Though some turtles in our study were observed with ingested fishing line, none were diagnosed with complications. This could also be attributed to differences between commercial and recreational fisheries. Fishermen that release the turtles on site are able to reel in a juvenile L. kempii, cut the line, and release it with a shorter length of line present, which is not always possible with the larger turtles seen in commercial fisheries. An example of this was described by Casale et al. (2008), where it was noted that lengths of retained branchline that were found to have caused death in C. caretta ranging from 51.5 to 540 cm. All ingested lengths of line collected with passed hooks in our sampled L. kempii were relatively short, if present, and none reached the length of 51.5 cm. Turtles may still face line-related injuries in recreational fisheries if the line breaks or is cut to release the turtle, leaving a long length of line that can be ingested; if remaining external, line could be subject to tangling and biofouling. At least one L. kempii with an immobile hook had line detach and pass successfully, though we cannot be sure how many immobile hooks had line associated with them as the lines used in recreational fisheries are often not radio-opaque.

Nemoz et al. (2004) described European pond turtles (Emys orbicularis) with fish hooks embedded in their esophagus. These turtles were found weakened and with evident weight loss, and one was later recaptured with a second hook embedded close to the first hook. The majority of the immobile hooks seen in L. kempii in the present study did not appear to impact the turtles' ability to survive or eat during their time in observation. The 6 L. kempii that were released with gear and recaptured in the present study showed no emaciation or debilitation, and one had a separate hook pass through its GIT unhindered by the previously embedded hook (which had partially degraded). In all cases where the hook was determined to be embedded/immobile in the upper GIT based on repeated radiographs, the hook had turned so that the eye was directed distally. The same was observed in E. orbicularis (Nemoz et al. 2004) and Apalone spinifera (Steen et al. 2014) radiographs. This indicates that when the hook shank is directed distally on radiograph the chances of it being or becoming immobilized are increased. A hook with the shank oriented posteriorly is more likely to embed the point of the hook in tissue than if the shank of the hook was directed cranially (Fig. 1C–D). Thus, the orientation of an ingested hook as viewed in radiographs can be used to gauge likely mobility in the upper GIT. While the same does not hold true for the mid and lower GIT due to the various angles of overlapping sections of intestine observed in 2-dimensional imaging, most hooks that will become stuck do so in the upper GIT where orientation can be defined more clearly. The lack of complications may be attributed to the larger size of sea turtles, which when compared with freshwater turtles like E. orbicularis, would allow more room in the GIT for ingesta to travel past an immobile hook. Our resampling of turtles with immobile hooks also suggests that a sea turtle's natural environment and diet may play a significant role in the degradation of hooks over time. A good example is seen in the L. kempii that had a completely intact immobile hook for 280 d of captive observation but was reradiographed 38 d after release, when most of the hook's shank had degraded (Fig. 1C–D). A study by Broadhurst et al. (2007) found that some yellowfin bream (Acanthopagrus australis) were able to eject ingested hooks between 6 and 56 d of observation. All of these hooks showed some degree of oxidation and they were often broken in 2 pieces. Though the composition of hooks is likely to affect the time that it takes them to degrade (McGrath et al. 2011), we did not identify hook compositions in our study. Additionally, no hooks in our study were able to decay into multiple pieces and then be completely expelled within a turtle's observation period, though 15 turtles who were admitted containing ingested partial hooks (having contained them for an indeterminable amount of time) did successfully expel them.

As transit times were shown to not be influenced by either turtle size or hook size, we can assume that ingesta transit rates, even for foreign bodies (where passage is successful), are correlated to factors that regularly influence metabolism in reptile physiology: health and temperature (Skoczylas 1978; Birse and Davenport 1987; Spencer et al. 1998). While Valente et al. (2008) found no statistically significant difference in transit times for C. caretta in water ranging from 16.27°C to 23.86°C, it was noted that turtles in 20.0°C–23.60°C water tended to have faster transit times than in water under 20.0°C. Similarly, Birse and Davenport (1987) found that transit time in C. caretta decreased as temperature increased between 20.0°C and 30.0°C, with the effect being greatest between 20.0°C and 25.0°C. Lepidochelys kempii included in this study were generally kept between 25.0°C and 30.0°C. Despite the differences in temperature, the average and range of T_U (13.50 d, 8–19 d) in the study reported here is similar to that seen in the Valente et al. (2008) study of passage in C. caretta at lower temperatures (13.19 d, 5–21 d to empty 85% of ingested markers). The results of this and previous studies together suggest that sea turtles may have a broad range of temperatures for optimal GIT function, and while transit times may vary inversely with temperature, it may not be significant.

For rehabilitation and radiographic interpretation we identified multiple factors that can be used in judging hook mobility. The upper GIT was where most hooks became embedded. If a hook moves past this point, the chances of it passing are increased. The orientation of a hook in the upper GIT can be used as an additional identifier of a hook's mobility/immobility, as a barb that is not pointed toward the cranial end of the animal indicates that it may be embedded in tissue. Lastly, our reported transit times based on location within the GIT provided means for monitoring mobility and the potential for medical complications. For example, a turtle presented with an ingested hook in the upper GIT that does not show any movement after 20 d will probably not pass the hook. However, we noted evidence of hooks degrading over time in sea turtles. Turtles presented hooked in the oral cavity and upper esophagus had their hooks removed during their initial examination. Butcher et al. (2007) showed that when a hook is embedded in a fish's buccal cavity the hook often dislodges and is subsequently ingested. We think this is also likely in turtles and may account for some of the transient and embedded hooks in our sample.

While studies on commercial fisheries' interactions with sea turtles are well represented in research literature, there has been little effort to describe the relationship between sea turtles and recreational fisheries. Our study highlights some of the differences in the effects of incidental captures in these fisheries. To further understand this issue, future studies should include necropsy information from turtles that wash ashore in areas with recreational fisheries, as our sample represents L. kempii that already survived a period after initial ingestion and did not stop attempting to feed. Additionally, surveys of fishing effort and gear used in areas where sea turtles regularly interact with recreational fisheries would be beneficial in measuring relationships between sea turtle interactions, gear, and effort. Information collected from our large sample of L. kempii adds to the body of knowledge concerning the understudied areas of recreational captures and sea turtle biology in the Mississippi Sound while providing means to gauge potential human impacts on the nGOM population.