Decline of the Sea Turtles: Causes and Prevention (1990)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

6 Sea Turtle Mortality Associated wig Human Activities ea turtles on nesting beaches are most susceptible to mortality asso- ciated with human activities at the egg, hatchling, and nesting female stages and in coastal waters at the subadult (including juve- nile) and adult stages. They are vulnerable to diverse potentially lethal interactions with human activities, situations including direct predation and habitat modification, incidental capture or entanglement in fishing gear, and physical damage caused by dredging of shipping chan- nels, collisions with ships and boats, and oil-rig removal or other under- water explosions. Each species in the pelagic environment is vulnerable to ingestion of plastics, debris, and petroleum residues. The species differ in behavior and habitat requirements, so they can be affected differently by various human activities. The recognized sources of mortality related to human activities are list- ed in order of estimated importance in Table 6-1 for all life stages, and order-of-magnitude mortality estimates are presented in Table 6-2 for juvenile plus adult loggerheads and Kemp's ridleys. The latter table includes the committee's judgment of the certainty of the information on which the estimates were based and lists the preventive and mitigative measures that are in place or being developed. The preventive and mit- igative measures are described and evaluated in detail in Chapter 7. The present chapter discusses the information on each mortality factor associ 74

75 Sea Turtle Mortality Associated with Human Activities TABLE 6-1 A qualitative ranking of the relative importance of various mortality factors on juveniles or adults, eggs, and hatchlings with an indication of mortality caused primarily by human activities. Sources are listed in order of importance to juveniles or adults, because this group includes the life stages with greatest reproductive values. Life Stage Primarily Human Juveniles Source of Mortality Caused to Adults Eggs Hatchlings Shrimp trawling yes high none unimportant Other fisheries yes medium to low none unimportant Non-human predators no low high high Weather no low medium low Beach development yes low medium low Disease no low unimportant low Dredging yes low unimportant unimportant Entanglement yes low unimportant low Oil-platform removal yes low none unimportant Collisions with boats yes low none unimportant Directed take yes low medium unimportant Power plant entrainment yes low none unimportant Recreational fishing yes low none unimportant Beach vehicles yes low to unimportant medium unimportant Beach lighting yes low to unimportant unimportant medium Beach replenishment yes unimportant low low Toxins yes unknown unknown unknown Ingestion of plastics, yes unknown none unknown debris ated with human activities first for eggs and hatchlings and then for juve- niles through adults. The analyses in Chapter 5 on the reproductive value of various life stages called attention to the mortality factors that are most important for juveniles and adults in the ocean and inshore marine habitats. The most important identifiable source of mortality for loggerhead and Kemp's rid- leys is incidental capture in shrimp trawls (Table 6-21; other fisheries and fishery-related activities are also important, but collectively only one-tenth as important as shrimp trawling. Dredging, collisions with boats, and oil- rig removal are also important, but only one-hundredth as important as  shrimp trawling. Mortality from entrainment in power plants and directed capture of juveniles and adults is believed to be generally low. Parasites,

a' o . - v) a' cd ~ o ~ o an ~ ~v) A Cal o a' Ct C) ~ A a: oc . o o V) ad (d .Z ~ O ~ Cal - oc Ct A O ~ ~ O U) ._ Cal Cal ~ Q -a o V) O _ am ~1 4 -  Cal V) U) au a' ~ ~ ._ .~ V) ~ An, Cal ~ _ so so O ~ O .= ¢ a; Cal - ._ ad Ct au ._ Cal .~ ._ Cal a' .~ PA ._ Cal *a Ct O .e at: O aim a' . - o A A U) o ~ o ~  a' 76 ~o .E ~ =, ~ ~a., S~ S ~ ~.~ 0, ° ~E E ~ C) Cd o ~ V) ~ ~ ~ o ~ ~2 ~ C ~o oo o ~o U ~o o o o o o , mY ~U ~U~ O ~', U~N ~U~ o o o o o U-` o ~o o o o o U ~U ~U~ ._ 2 ~ ,', ~ ._ U) Ct - 11 ._ C~ a; 4= o 11 *

77 Sea Turtle Mortality Associated with Human Activities toxins, and ingestion of plastics and other debris also constitute prob- lems, but present information does not allow quantitative estimates of annual mortality related to them. MORTALITY OF SEA TURTLE EGGS AND HATCHLINGS Beach Erosion and Accrelion Erosion of nesting beaches can result in loss of suitable nesting habi- tat. Erosion rates are influenced by dynamic coastal processes, including sea-level rise. Human interference with natural processes through coastal development and associated activities has resulted in accelerated erosion rates in some localities and interruption of natural shoreline migration. Accretion (deposition of beach sediments) also kills eggs in a nest. Beach Armoring Where beach-front development occurs, a site is often fortified to pro- tect the property from erosion. Shoreline engineering is expensive and is virtually always carried out to save structures, not sandy beaches; it usu- ally accelerates beach erosion (NRC, 19871. Several types of shoreline engineering, collectively referred to as beach armoring, include sea walls, rock revetments, riprap, sandbag installations, groins, and jetties. Those structures can cause severe adverse effects on nesting turtles and their eggs. Beach armoring can result in permanent loss of a dry nesting beach through accelerated erosion and prevention of natural beach and dune accretion, and it can prevent or deter nesting females from reaching suitable nesting sites. Clutches deposited seaward of the structures can be inundated at high tide or washed out by increased wave action near the base of them. As the structures fail and break apart, they spread debris on the beach, which can further impede access to suitable nesting sites and result in a higher incidence of false crawls (non-nesting emergences of females) and trapping of hatchlings and nesting turtles. Sandbags are particularly susceptible to rapid failure, which results in extensive debris on nesting beaches. Rock revetments, riprap, and sandbags can cause nesting turtles to abandon nesting attempts or to construct egg cavities of improper size and shape. Groins are designed to trap sand during transport in longshore cur- rents, and jetties might keep sand from flowing into channels. Those structures prevent normal sand transport and accrete beaches on one side of the structure while starving opposite beaches, thereby causing severe

78 Decline of the Sea Turtles erosion (NRC, 1987) and corresponding degradation of nesting habitat. Even widely spaced groins can deter nesting. Drift fences, also commonly called sand fences, are erected to build and stabilize dunes by trapping sand that moves along the beach and pre- venting excessive sand loss. They also protect dune systems by deterring public access. Because of their construction, improperly placed drift fences can impede nesting and trap emergent hatchlings. Beach Nourishment Beach nourishment consists of pumping, trucking, or otherwise depositing sand on the beach to replace what has been lost to erosion. Beach nourishment can disturb nesting turtles and even bury turtle nests during the nesting season. The sand brought in might differ from native beach sediments and can affect nest-site selection, digging behavior, incu- bation temperature (and hence sex ratios), gas-exchange characteristics in incubating nests, moisture content of a nest, hatching success, and hatch- ling emergence success (Mann, 1977; Ackerman, 1980; Mortimer, 1982b; Raymond, 1984; Nelson, 19861. Beach nourishment can result in severe compaction or concretion of the beach. The trucking of sand to protect beaches can itself increase compaction. Significant reductions in nesting success on severely compacted beach- es have been documented (Raymond, 19841. Nelson and Dickerson (1989a) evaluated compaction on 10 nourished east coast Florida beaches and concluded that five were so compacted that nest digging was inhibit- ed and another three might have been too compacted for optimal dig- ging. They further concluded that, in general, beaches nourished from off- shore borrow sites are harder than natural beaches and that, although some might soften over time through erosion and accretion of sand, oth- ers can remain hard for 10 years or more. Nourished beaches develop steep escarpments in the midbeach zone that can hamper or prevent access to nesting sites. Nourishment projects involve use of heavy machinery, pipelines, increased human activity, and artificial lighting. They are normally conducted 24 hours a day and can adversely affect nesting and hatching activities. Pipelines and heavy machinery can create barriers to nesting females emerging from the surf and crawling up the beach, and so increase the incidence of false crawls. Increased human activity on a project beach at night might cause further disturbance to nesting females. Artificial lights along a project beach and in the nearshore area of the borrow site might deter nesting females and disori- ent emergent hatchlings on adjacent nonproject beaches.

79 Sea Turtle Mortality Associated with Human Activities Artificial Lighting Extensive research has demonstrated that emergent hatchlings' princi- pal cues for finding the sea are visual responses to light (Daniel and Smith, 1947; Hendrickson, 1958; Carr and Ogren, 1960; Ehrenfeld and Carr, 1967; Dickerson and Nelson, 19891. Artificial beachfront light from buildings, streetlights, dune crossovers, vehicles, and other sources has been documented in the disorientation of hatchling turtles (McFarlane, 1963; Philibosian, 1976; Mann, 1977; Fletemeyer, 1980; Ehrhart, 19839. The results of disorientation are often fatal. As hatchlings head toward lights or meander along the beach, their exposure to predators and likeli- hood of desiccation are greatly increased. Disoriented hatchlings can become entrapped in vegetation or debris, and many hatchlings have been found dead on nearby roadways and in parking lots after being struck by vehicles. Hatchlings that find the water might be disoriented after entering the surf zone or while in nearshore water. Intense artificial light can even draw hatchlings back out of the surf (Carr and Ogren, 1960; pers. comm., L. Ehrhart, University of Central Florida, 1989~. In 1988, 10,155 disoriented hatchlings were reported to the Florida Depart- ment of Natural Resources. The problem of artificial beachfront lighting is not restricted to hatch- lings. Carrel al. (1978), Ehrhart(1979), Mortimer(1982b), andWithering- ton (1986) found that adult green turtles avoided bright areas on nesting beaches. Raymond (1984) indicated that adult loggerhead emergence pat- terns were correlated with variations in beachfront light in southern Bre- vard County, Florida, and that nesting females avoided areas where beachfront light was most intense. Witherington (1986) noted that logger- heads aborted nesting attempts at a greater frequency in lighted areas. Problem lights might not be restricted to those placed directly on or near nesting beaches. The background glow associated with intensive inland light, such as that emanating from nearby large metropolitan areas, can deter nesting females and disorient hatchlings that are navigating the nearshore waters. Cumulatively, along the heavily developed beaches of the southeastern United States, the negative effects of artificial light are profound. Beach Cleaning Several methods are used to remove human-caused and natural debris from beaches, including mechanical raking, hand raking, and hand pick- ing of debris. In mechanical raking, heavy machinery can repeatedly tra

80 Decline of the Sea Turtles verse nests and potentially compact the sand above them; it also results in tire ruts along the beach that might hinder or trap emergent hatchlings (Hosier et al., 19811. Mann (1978) suggested that mortality within nests can increase when beach-cleaning machinery exerts pressure on soft beaches with large-grain sand. Mechanically pulled rakes and hand rakes can penetrate the surface and disturb a sealed nest or might even uncover pre-emergent hatchlings near the surface of the nest. In some areas, col- lected debris is buried on the beach; this can lead to excavation and destruction of incubating egg clutches. Disposal of debris near the dune line or on the high beach can cover incubating egg clutches, hinder and entrap emergent hatchlings, and alter nest temperatures. Mechanical beach cleaning is sometimes the sole reason for extensive nest relocation. Increased Human Presence Resident and tourist use of developed (and developing) nesting beach- es can adversely affect nesting turtles, incubating egg clutches, and hatch- lings. The most serious threat caused by increased human presence on the beach is the disturbance of nesting females. Nighttime human activity can cause nesting females to abort nesting attempts at all stages of the process. Murphy (1985) reported that beach disturbance can cause turtles to shift their nesting beaches, delay egg-laying, and select poor nesting sites. Davis and Whiting (1977) reported significantly higher rates of false crawls on nights when tagging patrols were active on an otherwise remote, undeveloped nesting beach. Nesting beaches heavily used by pedestrians might have low rates of hatchling emergence, because of compaction of the sand above nests (Mann, 1977), and pedestrian tracks can interfere with the ability of hatchling loggerheads to reach the ocean (Hosier et al., 19811. Campfires and the use of flashlights on nesting beaches disorient hatchlings and can deter nesting females (Mortimer, 19893. Recreational Beach Equipment Recreational material on nesting beaches (e.g., lounge chairs, cabanas, umbrellas, boats, and beach cycles) can deter nesting attempts and inter- fere with incubating egg clutches and the seaward journey of hatchlings. The documentation of false crawls near such obstacles is increasingly common as more recreational equipment is left in place all night on nest- ing beaches. There are also reports of nesting females that become entrapped under heavy wooden lounge chairs and cabanas on southern

81 Sea Turtle Mortality Associated with Human Activities Florida nesting beaches (pers. comm., S. Bass, Gumbo Limbo Nature Cen- ter, 1989; pers. comm., J. Hoover, Dade County Beach Department, 1989~. Recreational beach equipment placed directly above incubating egg clutches can hamper emergent hatchlings and can destroy eggs by pene- tration directly into a nest (pers. comm., C. LeBuff, Caretta Research, Inc., 19891. Beach Vehicles The operation of motor vehicles on turtle nesting beaches is still per- mitted in many areas of Gulf of Mexico and Atlantic states (e.g., Florida, North Carolina, and Texas). Some areas restrict night driving, and others permit it. Driving on beaches at night during the nesting season can dis- rupt the nesting process and result in aborted nesting attempts. The adverse effect on nesting females in the surf zone can be particularly severe. Headlights can disorient emergent hatchlings and vehicles can strike and kill hatchlings attempting to reach the ocean. The tracks and ruts left by vehicles traversing the beach interfere with the ability of hatchlings to reach the ocean. The time spent in traversing tire tracks and ruts can increase the susceptibility of hatchlings to stress and predation during transit to the ocean (Hosier et al., 19811. Driving directly above incubating egg clutches compacts the sand and can decrease hatching success or kill pre-emergent hatchlings (Mann, 19771. In many areas, beach-vehicle driving is the only reason nests have to be relocated. Vehicular traffic on nesting beaches also contributes to erosion, especially during high tides or on narrow beaches, where driving is concentrated on the high beach and foredune. Exotic Dune and Beach Vegetafion Non-native vegetation has been intentionally planted in or has invaded many coastal areas and often displaces native species, such as sea oats, beach morning glory, railroad vine, sea grape, dune panic grass, and pen- nywort. The invasion of such destabilizing vegetation can lead to increased erosion and degradation of suitable nesting habitat. Exotic veg- etation can also form impenetrable root mats, which can prevent proper nest-cavity excavation, and roots can penetrate eggs, cause eggs to desic- cate, or trap hatchlings. The Australian pine (Casuarina equisetifolia) is particularly detrimen- tal. Dense stands of that species have taken over many coastal strand areas throughout central and southern Florida, causing excessive shading

82 Decline of the Sea Turtles of the beach. Studies in southwestern Florida suggest that nests laid in the shaded areas are subjected to lower incubation temperatures, which can alter the natural hatchling sex ratio (Marcus and Maley, 1987; Schmelz and Mezich, 19881. Fallen Australian pines limit access to suitable nest sites and can entrap nesting females. Davis and Whiting (1977) reported that nesting activity declined in Everglades National Park where dense stands of Australian pine took over native beach vegetation. Schmelz and Mezich (1988) indicated that dense stands of Australian pines in south- western Florida affect nest-site selection and cause increased nesting in the middle beach area and higher ratios of false crawls to nests compared with areas of native vegetation. MORTALllY OF SEA TURTLE JUVENILES AND ADULTS Shrimp Fishing Description of the Fishery The shrimp fishery has the highest product value of any fishery in the United States. It also is the most important human-associated source of deaths of adult and subadult sea turtles. Sea turtles are captured in shrimp trawls towed along the bottom behind shrimping vessels. The vessels might tow one to four otter trawls. An otter trawl consists of a heavy mesh bag with tapered wings on each side that funnel shrimp into the cod end, or bag, of the net. To keep the trawl near the bottom and achieve horizonal opening of the mouth of the trawl, a weighted otter board is positioned at the front of each wing to serve as a hydrofoil. Tur- tles swimming, resting, or feeding on or near the bottom in the path of a trawl are overtaken and enter the trawl with the shrimp. What is often perceived as the U.S. shrimp fishery is actually a number of fisheries. Seven species of shrimp are harvested in the fishery: brown shrimp (Penaeus aztecus), white shrimp (P. setiferus), pink shrimp (P. duorarum), seabobs (Xiphopenaeus kroyeri), royal red shrimp (Hymen openaeus rousts, rock shrimp (Sicyonia brevirostr~s), and trachs (Tra- chypenaeus spy. Each shrimp species is taken by a distinct fishery, and the several fisheries are differentiated according to fishing depths, season- al landings, vessel and gear, fishing localities, fishing techniques, and other characteristics. The most valuable shrimp species in the United States are brown, white, and pink. For example, in 1985, U.S. commercial shrimp catches were 122,000 metric tons in the gulf and 13,000 metric tons in the south Atlantic. The white shrimp fishery is the most important in the U.S. south Atlantic; the brown shrimp fishery is more important in the gulf.

83 Sea Turtle Mortality Associated with Human Activities Brown shrimp range along the north Atlantic and Gulf of Mexico coasts from Martha's Vineyard, Massachusetts, to the northwestern coast of Yucatan. The range is not continuous, but is marked by an apparent absence of brown shrimp along Florida's west coast between the Sanibel and the Apalachicola shrimping grounds (Farfante, 19691. In the U.S. Gulf of Mexico, catches are highest along the coasts of Texas, Louisiana, and Mississippi. Brown shrimp can be caught at depths of 100 m or more, but most come from depths less than 50 m. The season begins in May, peaks in June and July, and declines to an April low (Gulf of Mexico Fishery Management Council, 19811. White shrimp range along the Atlantic coast from Fire Island, New York, to Saint Lucie Inlet, Florida, and along the gulf coast from the mouth of the Ochlockonee River, in the Florida panhandle, to Campeche, Mexico. In the gulf, there are two centers of abundance: one along the Louisiana coast and one in the Campeche area. White shrimp are com- paratively shallow-water shrimp; most of the catch comes from depths less than 25 m. The catch has a major peak in late summer and early fall, with an October high and a minor peak of over-winter shrimp with a peak in May. The largest catches occur west of the Mississippi River to the Freeport, Texas, area, although the catch is considerable along the entire north central and western gulf and south Atlantic. Pink shrimp range along the Atlantic from the lower Chesapeake Bay to the Florida Keys and around the gulf coast to the Yucatan peninsula. Major concen- trations exist off southwestern Florida and in the southeastern part of the Gulf of Campeche. The two major pink shrimp grounds in the United States are the Tortugas and Sanibel grounds in southwestern Florida. The pink shrimp catch comes mainly from depths less than 50 m, with a maxi- mal catch from 20-25 m. Most of the catch is taken off Florida and is greatest in the southwestern waters of the state. The catch is high from October through May. In the south Atlantic, white shrimp account for the majority of landings in Georgia and the Atlantic coast of Florida. In South Carolina, small landings of white shrimp in the spring are augmented by a much larger catch in the fall. The spring white shrimp fishery is based on adults that have over-wintered, whereas the fall catch is based almost entirely on young of the year. White shrimp are caught in North Carolina principally during the fall, but the catch is much smaller than that of brown and pink shrimp (Carder et al., 19741. Brown shrimp predominate in the North Carolina fishery. During some years, catches of brown shrimp exceed those of white shrimp in South Carolina as well. The peak of the brown shrimp harvest occurs during the summer in all four south Atlantic states. Brown shrimp enter and leave the Florida east coast fishery earlier than in the other three states. In the south Atlantic, pink shrimp are of major

84 Decline of the Sea Turtles commercial significance only in North Carolina, where they account for about one fourth of the total shrimp landings. Fishing for pink shrimp usually begins in the spring and ends by midsummer. Other minor shrimp species are often fished incidentally or during the offseasons of the major shrimp fisheries. A targeted rock shrimp fishery exists in the south Atlantic off northern Florida from August to January. In recent years, vessels in the western gulf have focused their effort on "tracks" during the late winter and spring months. The trach catch is pri- marily from depths of 20-50 m. The royal red shrimp fishery is relatively insignificant, occurring at depths of 250-550 m; harvesting and marketing obstructions have limited this fishery. Seabobs are caught most often in shallow waters at 13 m or less and in the open ocean; along the Louisiana coast, catch rates are highest in October-December. The various fisheries share some similarities, in socioeconomic makeup and biology, but there are important contrasts, principally in depth of operation. The similarities and differences might have an important bear- ing on turtle bycatch. Many of the fisheries for shrimp, especially of the three major species, are timed and located in relation to the life histories of the shrimp. For example, several discrete fisheries constitute the "gulf brown shrimp fishery." Juvenile and subadult brown shrimp live in bays and estuaries and are harvested by the inshore fishery. The shrimping vessels used are usually small, from 6 to 30 m long; most are about 15 m long. As the shrimp mature, they migrate offshore. Vessels fish near shore out to a depth of 25 m, especially for subadult and adult white and pink shrimp. The larger vessels of the gulf type begin almost exclusive harvest of the species (adult brown and pink shrimp) in water deeper than 25 m; these vessels are generally 20-30 m long. As the maturing brown shrimp continue to migrate into the deeper gulf waters, the smaller inshore ves- sels are limited, and only the larger vessels can gain access to the fishery. The offshore fishery provides the basis for the adult brown shrimp fish- ery. The pink shrimp fleet off Florida uses a variety of vessels of different sizes but is associated primarily with larger offshore boats. White and brown shrimp are caught in bays and estuaries in some states by the smaller inshore vessels. Unlike the adult brown shrimp fleet, which uses larger vessels, the adult white shrimp fleet uses vessels of all sizes. Distribution and Intensity of the Fishery The distribution and intensity of fishing in inside waters were calculat- ed from raw data and summaries provided to the committee by NMFS. For waters outside the coast, the information was taken from Appendix F.

85 Sea Turtle Mortality Associated with Human Activities Fishing effort is measured in effort-days (24-hr days of towing time) per boat, regardless of variations in vessel size, the number and size of nets it tows, and water depth. That probably underestimates effort outside the coastal beaches, compared with bays, rivers, and estuaries, because off- shore boats tend to be larger and tow more nets for longer periods than the inside boats. The unit of effort is used by NMFS in the Gulf of Mexi- co and the Atlantic Ocean and is estimated annually with the cooperation of individual state agencies. Quarterly effort in offshore waters is plotted by fishing zone from Texas to Maine for 1987 and 1988 (Figure 6-11. Data used are the best available currently from NMFS. The shrimp fishery is intense, totalling about 373,000 24-hr days per year from the Mexican border in the gulf to Cape Hatteras in the Atlantic during 1987 and 1988. The intensity is much greater in the gulf 345,000 days-compared with the Atlantic's 28,000 days; 92% of the total effort is expended in the gulf. Most effort is offshore of the coastal beaches, about 249,000 days, or 67% (67% for the gulf and 68% for the Atlantic). The rest is expended in the bays, estuaries, and rivers, 33% of the total effort. The most intense fishery outside the coastal beaches is off Texas and Louisiana and includes 83% of the effort off the coastal beaches and 55% of the total effort from Maine to the Mexican border. Shrimp fishing off Mexico near the U.S. border has been low or absent in recent years. Although fishing efforts in the bays, estuaries, and rivers are similar in the Atlantic and Gulf of Mexico, distinct differences occur in shrimping efforts directed toward offshore waters. Brown and pink shrimp are important fisheries in the Gulf; therefore, more shrimping effort is expended in deeper waters of the gulf than in the Atlantic, where white shrimp dominate the fishery. Statistical reporting procedures vary between the Atlantic and gulf data bases (pers. comm., I. Nance, NMFS, 19891. Areas of effort are reported by distance from shore in the Atlantic and by depth in the gulf. Because of differences associated with the slope of the gulf's continental shelf, a comparison of effort by distance from shore would be impractical; however, because white shrimp is the principal Atlantic fishery, effort focuses on a relative shallower and nearshore fishery. For 1987 and 1988, NMFS data indicate that 92% of the Atlantic effort outside of coastal beaches was from 0 to 5 km, 3% was from 5 to 20 km, and 4% was farther than 20 km offshore. In contrast, in the gulf outside of coastal beaches, 65% of the effort was in water shallower than 27 m (a depth contour that ranges from approximately 14-50 km offshore), while 24% was between 27 m and 48 m; another 11% was deeper than 48 m. Seasonally, effort for the fisheries outside of coastal beaches is greatest in summer and fall, lower in spring, and least in winter. In 1987 and

86 Sea Turtle Mortality Associated with Human Activities FIGURE 6-1 Shrimp-fishing effort, 1987-1988, by season. Fishing zones are shown on horizontal axis (see Figure 4-13. Data from Appendix F. 45000 30000 D A y S D A y S 1 5000 o 45000 30000 1 5000  SHRIMPING EFFORT 19B7 · Fall ~ Summer [:1 Spring O Winter Van ........ . . . ~. . . 21 18 15 12 9 6 3 24 27 30 33 36 39 42 SHRIMPING EFFORT 1988 O- ~- ~ 1 1 At 21 18 15 12 9 6 3 24 27 30 33 36 39 42 1988, 33% of the effort was in summer, 31% in fall, 24% in spring, and 12% in winter. This pattern largely represents that of the western gulf; local variations from this pattern occur. Off Georgia and the Carolinas, lit- tle fishing takes place in winter, whereas off the Atlantic coast of northern Florida, effort is more uniform through the year and includes significant winter fishing. Along the gulf coast of Florida, fishing is most intense in winter and spring.

87 Sea Turtle Mortality Associated with Human Activities Fishing effort in the gulf has grown steadily since 1960. The increase has been by a factor of about 2.5 in 30 years. The proportion of the effort in rivers, estuaries, and bays has remained about the same during this growth period. In the Atlantic, comparable data were not available, but from 1984 to 1988, total effort ranged from 24,000 to 34,000 24-hr days per year, reaching a maximum in 1986. Seasonal Changes in Stranding, Shrimp-Fishing Effort, and Turtle Abundance The recent abundance of stranded sea turtles and the intensity of shrimp fishing vary from the western Gulf of Mexico along the coast to the Gulf of Maine and from season to season. The distributions of strand- ings are complex interactions between trawling intensity and the abun- dance of sea turtles and other factors. The relationship between strand- ing and fishing intensity takes on a different perspective when viewed on short and long time and space scales. For example, the highest stranding rate does not occur off Texas and Louisiana (Figure 4-3), where shrimp fishing is now most intense (Figure 6-1~; turtle abundance is lower there ~. ~ in, , now than along the south Atlantic coast. Such broad-scale comparisons do not provide evidence of the present effects of trawling, because they do not account for historical changes in the abundance of turtles in rela- tion to past shrimping and other mortality factors. The relation between turtle stranding and fishing effort on an interme- diate scale i.e., seasonal changes in areas that differ in the ratio of turtle abundance to shrimping effort permits an interesting, but speculative interpretation. Sites chosen for our analysis were those with NMFS con- tractual stranding surveys: Texas (zones 17-21), the gulf coast of Florida (zones 4 and 5), the northern Florida's Atlantic coast (zones 29-31), and Georgia-South Carolina (zones 31 and 321. Those four areas span a range of ratios of turtle abundance in aerial surveys (number of turtles sight- ed/10,000 km2) to shrimp fishing effort (10,000 24-hr days of fishing) from about two for Texas to about 2,500 off Florida, or a factor of about 1,250 in the ratio of abundance of turtles to fishing effort in the two states. In only two of the eight examples (Figure 6-2) was turtle stranding positively correlated with fishing effort (p = 0.051. One of those exam- ples, from the Atlantic coast of Florida in 1988, was used by Schroeder and Maly (1989) as evidence for a direct relation between stranding and fishing effort. The relation between stranding and effort is more complex than the simple argument that more shrimping effort equals more turtle stranding. Models of fishing-induced mortality have produced insights that can be  applied to the present situation. Results of an examination of the season

Decline of the Sea Turtles FIGURE 6-2 Seasonal changes in sea-turtle strandings on ocean beaches and shrimp-fishing effort offshore of ocean beaches at four locations along the Gulf of Mexico and the Atlantic coast for 1987 and 1988. The four areas differ greatly in the abundance of turtles sighted from aerial ocean surveys and shrimp-fishing effort. Texas (zones 18-21) had the fewest turtles per unit of shrimp fishing, followed by western Florida (zones 4-5), and Georgia and South Carolina (zones 31-32?; the largest number of turtles per unit of shrimp-fishing effort was for Atlantic north Florida (zones 28-301. The correlations and p values are those for a sim- ple linear regression. Data from Appendixes E and F. r-0.10 p>0.1 TEXAS- 19117 _ r.-0.27 p,0. 1 TEXAS- 198B 90 1 500VA EFFORT_~3, ~ ''' ":.\1 Q° 100~0 ,~. STRANDINGS "' .'', ~.30 SOOO JAN APR JlJI Ot . JAN AM JUL OCT 1200T ~ B.OC In ~40c - o IL IL CJ GEORGIAN CAROUNA-1~7 z00 r-~0.23 peon 240 `~1800 STR~N~  I 900 ~ SO O O JAN APR JUL OCT GUM COAST OF S. FLORIDA.19B7 ..+0.82 p~o.o1 . 30 5 lo 7,: EFFORT _ ~ STRANDINGS JAN APE JUL OCT ATLANTIC FLORIDA-1#~7 90^ 600 300 t SO ~20 __ 4~ GULF COAST OF S. FLORIDA-19BB 600 i. .0.53 p-0.06 400 EFFORT 200 L O i-. At' ~_ _ JAN APR JUL OCT 20 ~ ~ ,~n a '° m 0 2100 ..~-V.~ A,-.. 60 . STRANDIN(3S P- - ~- Coo ~4 23 0r ~ lo - JAN APR JUL OCT GEORGIANS CAROLINA-1968 r.+0.28 v>0.1 ATLANTIC FLORIDA-1988 r.+0.58 p0.05 90 r.~0.06 p-0.1 120 900 p  \: 600~_~ o O O JAN "R JUL DOT JO HER JUL EFFORT | OCT

89 Sea Turtle Mortality Associated with Human Activities al changes in fishing effort and stranding in each area and year (Figure 6- 2) suggest an analogy with those models. In Texas, for example, strand- ing reached a maximum in April and then declined as effort increased; later in the summer, effort was high, but few turtles were stranded. One possible interpretation is that trawling has eliminated most of the turtles in that area by early summer. Alternative explanations could be that the turtles migrate through the area in the spring (Pritchard and Marquez M., 1973; Timko and Kolz, 1982) and that oceanic conditions in the spring differ from those in the other seasons and tend to bring more dead float- ing turtles to the beach than in other seasons (Amos, 19891. A pattern with some similar features was observed on the gulf coast of southern Florida. The decline in strandings occurred while effort was high, as would be expected from fishing-induced mortality. Effort declined, but the decline in turtle stranding began before effort dropped to low levels. In the Georgia-South Carolina region, stranding was greatest as the fishing effort was increasing early in the season, but later declined as effort continued to increase. That pattern is consistent with the interpre- tation that fishing effort locally depleted the turtles by middle to late sum mer. Finally, on the Atlantic coast of northern Florida, stranding reached a maximum as effort increased and then began to decline-sharply in 1987, marginally in 1988. That pattern is also consistent with the effects of fish- ing-induced mortality on a fixed or limited population size. The only case of no major decline when effort was high was the northern Florida exam- ple of Schroeder and Maley (19891. That area has the most turtles per unit of shrimping effort among the four locations examined and would be the most likely to support a direct relation between stranding and effort over an extended period with modest levels of fishing effort relative to the standing stock of sea turtles. We cannot eliminate the alternative hypotheses from the existing data on turtle migration and ocean currents. However, these observations are consistent with models of fishing-induced mortality, and that suggests that this is a likely hypothesis. It might explain the lack of a significant posi- tive correlation between seasonal fishing effort and turtle stranding in all but two of the eight examples: the relationship between fishing effort and abundance of the fished species often are out of phase. We note that neither significant positive nor nonsignificant negative correlations between seasonal changes in stranding and shrimping effort are by themselves enough to reveal the influence of shrimping on strand- ing. The relationships are more complex on these broad temporal and spatial scales in response both to shrimping effort and to changes in turtle abundance. The influence of shrimping on turtles cannot be excised

90 Decline of the Sea Turtles from the seasonal patterns only by a simple linear regression analysis. More incisive analyses, as presented below, are needed to tease apart the relationship. Strong Evidence of Shrimp Trawling as an Agent of Sea Turtle Mortality One central charge of this committee is to evaluate available evidence to assess whether incidental catch of sea turtles during shrimp trawling is indeed a cause of sea turtle mortality and, if so, to estimate the magnitude and importance of this mortality. Sea turtles are undoubtedly caught in large numbers during shrimp trawling. For example, the primary source of tag returns from female Kemp's ridleys tagged at the nesting beach at Rancho Nuevo (84% of 129 returns) has come from incidental capture of the turtles and reporting of tag numbers by cooperative shrimpers (Pritchard and Marquez M., 1973; Marquez M. et al., 19891. Furthermore, observers on vessels conducting commercial shrimp trawling have report- ed large numbers of sea turtle captures (Hillestad et al., 1978; Roithmayr and Henwood, 19821. Even if individual fishermen catch few turtles, the size of the shrimp fleet and the effort exerted result in a collective catch that is "large," although not all sea turtles that are caught in shrimp trawls necessarily die as a result. In a recent review, 83% of 78 papers on the incidental cap- ture of all Atlantic sea turtle species in fishing operations inferred that shrimp trawling is a major source of mortality (Murphy and Hopkins-Mur- phy, 19891. We consider below five observations that, when taken together, consti- tute a compelling demonstration that incidental capture during shrimp trawling is the proximate cause of mortality of substantial numbers of sea turtles. Relation Between Sea Turtle Mortality in Trawls and Tow Time The most convincing data available to assess whether shrimp trawling is responsible for sea turtle deaths come from NMFS studies relating the time that a trawl was allowed to fish (tow time) to the percentage of dead sea turtles among those captured. Henwood and Stuntz (1987) published a linear equation showing a strong positive relation between tow time and inci- dence of sea turtle death. They concluded that "the dependence of mor- tality on tow time is strongly statistically significant (r = 0.98, p < 0.0011." The committee analyzed the data set used by Henwood and Stuntz to clarify in detail the relationship between tow times and mortality. Death rates are near zero until tow times exceed 60 minutes; then they rise rapidly with increasing tow times to around 50% for tow times in excess

91 Sea Turtle Mortality Associated with Human Activities of 200 minutes. That pattern is exactly what would be expected if trawl- ing were causing the drowning of an air-breathing animal. Death rates never reach 100%, because some turtles might be caught within 40-60 minutes of lifting the net from the water. The data provide the functional relation between other correlative relations, namely, between fishing activity and dead turtles or population trends. Under conditions of involuntary or forced submergence, as in a shrimp trawl, sea turtles maintain a high level of energy consumption, which rapidly depletes their oxygen store and can result in large, potentially harmful internal changes. Those changes include a substantial increase in blood carbon dioxide, increases in epinephrine and other hormones asso- ciated with stress, and severe metabolic acidosis caused by high lactic acid concentrations. In forced submergence, a turtle becomes exhausted and then comatose; it will die if submergence continues. Physical and biological factors that increase energy consumption, such as high water temperature and increased metabolic rates characteristic of small turtles, would be expected to exacerbate the harmful effects of forced submer- gence because of trawl capture. Drowning can be defined as death by asphyxiation because of submer- gence in water. There are two general types of drowning: "dry" and "wet." In dry drowning, the larynx is closed by a reflex spasm, water is prevented from entering the lungs, and death is due to simple asphyxia- tion. In wet drowning, water enters the lungs. For nearly drowned tur- tles, the wet type would be more serious, because recovery could be greatly compromised by lung damage due to inspired seawater. The exact mechanism of sea turtle drowning is not known, but a diagnostic condition of the wet-drowning syndrome the exudation of copious amounts of white or pink froth from the mouth or nostrils has been observed in trawl-captured turtles. Turtles captured in shrimp trawls might be classified as alive and live- ly, comatose or unconscious, or dead. A comatose turtle looks dead, having lost or suppressed reflexes and showing no sign of breathing for up to an hour. The heart rate of such a turtle might be as low as one beat per 3 minutes. Lactic acid can be as high as 40 mM, with return to normal values taking as long as 24 hours. It takes 3-5 hours for lactic acid to return to 16-53% of peak values induced by trawl capture. Although the fate of comatose turtles directly returned to the sea is unknown, it is reasonable to assume that they will die (Kemmerer, 19891. In 1989, NMFS conducted a tow-time workshop to analyze data on tow times and turtle conditions from seven research projects. The projects spanned 12 years, during which 4,397 turtles were encountered. The numbers of dead and comatose turtles increased with tow time (Figure 6

92 Decline of the Sea Turtles 39. Small increases in tow time between 45 and 125 minutes resulted in large, steep increases in the numbers of dead and comatose turtles. For most tow times, there were more comatose than dead turtles. Few turtle deaths were related to tow times of less than 60 minutes. Tow times are thus a critical element in determining turtle mortality associated with shrimp trawls. Coinciclence of Opening and Closing of Shrimp Sec son with Changes in Turtle Stranding on Adjacent Beaches in Texas anal South Carolina Murphy and Hopkins-Murphy (1989) used the data on sea turtle stranding in South Carolina in 1980-1986 to seek a temporal relation between the opening of the ocean shrimp fishery and the rate of stranding. In South Carolina, the Sea Turtle Stranding and Salvage Network (STSSN) has pro- vided complete and reliable coverage of the ocean beaches for several years. The opening of the ocean shrimp fishery took place between May 16 and June 26 and varied from year to year. The 7-year total number of strandings (190 carcasses) in the 2-week periods just after the opening of the fishery was 5 times as large as the number of strandings in the 2-week periods immediately before the opening (38 carcasses). Although that does not conclusively demonstrate a causal relationship, repetition of the FIGURE 6-3 Relation between the percentage of dead or dead and comatose loggerheads as a function of tow time of trawls. Total number of turtles captured was 4,397. Compiled by the committee from raw data provided by NMFS that were the basis for Henwood and Stuntz's (1987) calculations. 100 75 PERCENT 50 25 O {I DEAD ' ~ D"D'O - TOSE job 0 60 120 180 240 TOW TIME (MIN) · . . .

93 o Ct Ct .~ V) o Cal v) V) . - U) o of ., o Ct of .= o C~ C~ a' o p U) a: 4 - C~ C~ s~ ~n o Q  ._ V) Ct 1 ~9 ~ ,-1 Z ¢ E~ E~ V) au -1- 0 a' ~ .~ C~ - p Q O ._. ~ ~ P~ *= .= C~ o Ps '=: ¢ a~ ~ a, o ¢ V) ~ O o ~ ~ O 4o~ C) o 4 - p Q 4- Ct p C~ U)  Ct Q oo ~9o G ~O . .. o oo 0 ~r o oo o oo oo oo U~U~ o oo r~ ~oo o oo ~ ~oooo oo oooo oo ooo ~G~G~  . - a~ o - ~ o ~v) c's Q_ O cnv: C~ 0 a~au C~ a~ a' oc O _ ~ U) ~ V) _ Q - 0 =' C~ O U: ~ s~ O o o C, O ;= V) =o - ~ Ct O *

94 Decline of the Sea Turtles FIGURE 6-4 Sea turtle strandings on beaches before and after opening or closing of shrimping seasons in South Carolina and Texas. Statistical analysis of differences is in Table 6-3. (Compiled from NMFS data9 MARINE TURTLE CARCASSES FROM SOUTH CAROLINA BEACHES 160 In . . 120 100 TURTLE STRANDINGS PER 80 2 WEEKS 60 40 l ~ 204: 2 - 0-2 ~2 2~ WEEKS WEEKS OPENING WEEKS WEEKS BEFORE BEFORE 2 WEEKS AFTER AFTER LOGGERHEAD CARCASSES FROM TEXAS BEACHES 20 18 16 TURTLE 1 2 TRANDINGS PER 10 2 WEEKS 8 2 w -tat A A` o 2-4 WEEKS 0-2 WEEKS ~2 WEEKS 2" WEEKS BEFORE BEFORE AFTER AFTER CLOSURE LOGGERHEAD CARCASSES FROM TEXAS BEACHES  18 16 ~ 10~un-80 O 15-May-81 17-May-82 0 1 6-May 83 15Jun-84 2Wun-85 -X 18~un-86 ILK 4~un-87 -28~)un-88 -1~un-89 1~un-80 O 22-May-81 ~ 25-May~2 O 27-May-83 16-May-84 20-May-85 X 10-May-86 OK 1~un-87 -1~un-88 ~ 1Wu1-80 14 - /: 0 16~u1-81 12- ~ // \ ~ 1Wu1-82 TURTLE 10 - - ~ \ o 15~u1-83 2 WEEKS ~ ~ ~s~i 87 o 2 - WEEKS 0-2 WEEKS 0-2 WEEKS 2 - WEEKS -1 6~1u1-88 BEFORE BEFORE AFTER AL I tR OPENING

95 Sea Turtle Mortality Associated with Human Activities large increase in stranding after the beginning of shrimping, despite varia- tion in the date of the beginning of shrimping, strongly suggests that shrimp trawling is the proximate cause of the large increase in dead sea turtles found on South Carolina beaches after the opening of shrimp sea- son. To evaluate further the potential effect of shrimp trawling on the num- bers of sea turtles found dead on South Carolina beaches, we followed the lead of Murphy and Hopkins-Murphy (1989) and segregated stranding data into two-week intervals (the first and second halves of each month, because of how the data were compiled) for the IO-year STSSN data base (1980-19891. The 2-week interval in which the-fishery opened was desig- nated as the "2 weeks after opening," unless the opening occurred at the end of the 2 weeks, in which case the next 2-week interval was called the "2 weeks after opening." We compiled the strandings not only for the 2- week intervals before and after opening of the shrimp season, but also for the 2-week periods before and after that 4-week period, for a total of four 2-week periods (Table 6-3, Figure 6-49. We then used the Wilcoxon signed-ranks test (a paired-sample nonparametric test) to compare strand- ings in each pair of successive 2-week periods. The 3.7-fold increase in turtle strandings that occurred in the 2 weeks after opening has a two- tailed probability of 0.006 of occurring by chance. No other contrast between successive 2-week intervals had a probability of less than 0.10. This analysis thus implies that shrimp trawling was indeed responsible for the increase in turtle strandings. It is an especially strong analysis, in that the increase observed with the opening of the fishery was independent of seasonal changes (the date of opening varied widely from May 16 to June 261. We also used the STSSN data base for Texas beaches for the 9 years of 1980-1988 to evaluate the effects of fishery closing and opening on stranding of loggerheads (Table 6-3, Figure 6-41. The changes in four consecutive 2-week periods were compared and analyzed as for the South Carolina data. The application of the nonparametric tests demon- strated that the sixfold increase in loggerhead stranding between the 2 weeks before and the 2 weeks after opening of the Texas brown shrimp fishery had a two-tailed probability of 0.04 (Table 6-31. Differences between 2-4 weeks before and 0-2 weeks before intervals were not statis- tically significant. As in the South Carolina case, the statistical tests sug- gest that loggerhead stranding increased significantly when shrimp trawl- ing opened in Texas. Finally, we analyzed in the same manner how stranding rates changed at the time of closing of the Texas brown shrimp fishery (Table 6-3, Fig- ure 6-41. Loggerhead stranding decreased between the 2 weeks before and the 2 weeks after closing by a factor of 2.7. The probability that that

96 Decline of the Sea Turtles decrease occurred by chance with a 2-tailed test was 0.01 (Table 6-33. The contrast between the first two periods (2-4 weeks before versus 0-2 weeks before closing) had a probability of 0.56. A decline in stranding did occur between 0-2 weeks after and 2-4 weeks after closing, p=0.008. Consequently, although a large and statistically significant decline in log- gerhead stranding had occurred after closing of the Texas brown shrimp fishery, the decline continued to occur between the last two periods. Given the uncertainty as to how long it takes for dead turtles to reach the beach, those results are consistent with either an effect of brown shrimp- ing on sea turtle stranding or a general decline in sea turtle stranding dur- ing the period for other reasons. Because the dates of closure varied from May 10 to June 1, we interpret the decline to be fishery related. Stranding in the three cases South Carolina opening, Texas opening, and Texas closing-changed by factors of 3.9, 5.0, and 4.5, based on the 4 weeks before and 4 weeks after opening. We conclude that, in those locations and at those times, approximately 70-80% of the stranded turtles were caught in shrimp trawls. Taken along with the results on tow time given above, these results provide strong evidence of the crucial role of shrimp fishing on turtle mortality. Relation Between Loggerhecd Populations and Shrimping Effort Along the southeastern Atlantic coast, loggerhead populations are declining where shrimp fishing is intense off the nesting beaches. They are not declining, however, where shrimping effort is low or absent. Nesting populations in South Carolina and Georgia (Figure 3-lc,4) are declining, whereas those in central and southern Florida are not and might even be increasing (Figure 3-le,J5. Shrimping effort declines markedly to the south at Cape Canaveral (Figure 6-1) so, for example, the population at Hutchinson Island is subject to essentially no shrimp fishing off the nest- ing beach (fewer than 17 effort-days per year) whereas the populations at Little Cumberland Island, Georgia, and Cape Island, South Carolina, have intense fisheries (about 400-7,000 days per year per fishing zone). Fur- ther evidence of the relation between shrimping effort and turtle popula- tion declines is found in the lower stranding rates of loggerheads in fish- ing zones 26-28 at Canaveral and south, where effort is low, even though these zones have the highest density of nesting loggerheads (Figure 4-21. Shrimping effort declines from about 1,000 effort-days per year in zone 28 to almost none in zones 27 and 26. In contrast, effort increases irregularly north of zone 28 to a maximum of 5,000-7,000 in zone 32, off South Car- olina. Because the turtles aggregate off the nesting beaches during the nesting seasons between their multiple nestings, the absence of the fish- ery would be expected to reduce mortality and contribute to the mainte

97 Sea Turtle Mortality Associated with Human Activities nance or growth of local nesting populations, as was observed south but not north of Canaveral. Quantification of Sec-TurtIe Mortality in Shrimp Trawls Incidental catch and mortality of sea turtles in shrimp trawls have been estimated on the basis of interviews with vessel captains (Anon., 1976; Anon., 1977; Cox and Mauerman, 1976; Rabalais and Rabalais, 1980; Rayburn, 1986) and direct observation by fishery observers on commercial shrimping vessels (Hillestad et al., 1978; Ulrich, 1978; Roithmayr and Henwood, 1982; Hen- wood and Stunts, 19871. Henwood and Stuntz (1987) provide the most complete assessment of sea turtle capture and mortality for the south Atlantic and Gulf of Mexico shrimp fisheries. Their study, based on more than 27,000 hours of observed trawling, estimated an annual incidental capture of approximately 47,000 sea turtles, with an estimated mortality of about 11,000. The study suffers from the a posterior) approach of esti- mating capture and mortality from Programs not specifically designed for 1 . ~r _ ·_ ~ v ~ , ~ that purpose and, therefore, Is limited in its ability to account for possible differences in capture and mortality related to such variables as species, season, depth, and geographic location. Although the statistics have been debated (Clement Assoc., 1989; Murphy and Hopkins-Murphy, 1989), the estimates are conservative because of the approach taken. Points of con- tention with the estimates of mortality include the use of data from a Arch .~tudv of the use of turtle excluders on trawlers, the representa _ , ~ a. . . ~ - . a . - 1 ~ . ~ ~ Ott ~d ~ 1 ~ ~ ~ 1 ~ ~;^ 1 tiveness of tishlng o1strlout1on between resear`;~1 bLUQ1~ Ally ~llllll~l~lal shrimping, the precision of mortality estimates based on the method used to calculate mortality rate, the magnitude of mortality estimates based on the assumption that all comatose sea turtles survive, and the magnitude of mortality estimates based on the complete omission of inside waters (waters landward of the barrier islands, including bays, sounds, etc.) (Table 6-41. The objective of the trawler excluder study was to design and use an apparatus that would effectively prevent the incidental capture of sea tur- tles in existing shrimping gear. Shrimp fishermen fished with commercial fleets in both the Gulf of Mexico and the south Atlantic. Sixty-two percent of the trips were in the south Atlantic, where 95% of the loggerheads and 78% of the Kemp's ridleys were caught (Table 6-51. Georgia, fishing zone 31, accounted for 74% of the total south Atlantic trips and 58% of the catch of loggerheads and 71% of the catch of Kemp's ridleys in the south Atlantic. For the south Atlantic, the estimated catch rate for the trawler !~1~ C'til~l~T~T~C~ c~tr~mgl~r infl~`~nr~1 by the catch rate: off Georgia; the  Georgia catch rate was lower than the other zones sampled. Similarly, in the gulf, the catch rates reflected activity off Texas and Louisiana, which comprised 75% of the effort. Eliminated from consideration in this study v _ CCAL~lil~l~1 OL~-V vv ~1 ~ll^1 y A_ ~ ~

98 Decline of the Sea Turtles TABLE 6-4 Points of contention and potential sources of bias to estimated mortality as cal- culated by Henwood and Stuntz (1987). Contention/Potential Bias Impacts on Estimate Use of trawl excluder study provided a biased sample because fishermen fished where turtles were. Fishermen fished with fleet and were not controlled by contracting agency. A sea turtle "hot spot" (Cape Canaveral channel) (see Figure 4-1) was eliminated from study. Geor- gia (fishing zone 31 (see Figure 4-1)) account- ed for the majority of the study effort and catch in south Atlantic; Texas and Louisiana (fishing zones 15-19 (see Figure 4-1)) account- ed for the majority of the study effort in the Gulf of Mexico. The study was used to calcu- late catch rates for the south Atlantic and the gulf. Overall catch (and hence, mortality) was estimated by multiplying catch rates by com- mercial fishing effort as determined separately by NMFS for the gulf and south Atlantic. No significant bias was detected. Fishing effort in study did not reflect In the Gulf of Mexico, 65% of the commercial true commercial fishery. Data were biased. effort was exerted in waters <27 m (1988), whereas 84% of the effort reported in the study was in waters <27 m. If catch rates are partitioned by depth (<27 m and >27 m), based on sea turtle distribution (see Chapter 4), the 19% oversampling of water <27 m results in an overestimate of catch (and mor- tality) of about 24% for the gulf. Precision of mortality estimates is erroneous. Methods used did not incorporate variability of mortality rate into variability of estimated mortality. (Product estimated captures times mortality rate.)  All comatose sea turtles were assumed to survive. This produces an underestimate, because not all comatose sea turtles do survive. Captures in inside waters were not included, thus reported estimates are low. Reported limits of confidence intervals would be widened, thus increasing the uncertainty about the estimated mortality. If all comatose turtles died, the estimate of mortality would increase from about 11,000 to about 32,000. Reported mortality estimates are underesti- mates due to omission of inside waters data. Approximately 37% of total shrimping effort occurs in inside waters. Depending on species, total estimated mortality might be higher by a factor of 1.6 or from 11,000 to 18,000.

99 Sea Turtle Mortality Associated with Human Activities TABLE 6-5 Distribution of effort (number of tows) and capture of loggerhead and Kemp's ridley sea turtles from trawler excluder study. Statistical ZoneTows (no.) Loggerheads (no.) Kemp's Ridleys (no.) 193 3 2476 3 360 2 15160 0 16111 0 17110 1 181,340 5 2 19169 1 ~ Total Gulf of Mexico2,519 14 4 30308 50 313,024 161 10 32527 41 4 33209 23 Total Atlantic4,068 275 14 Source: Partial data set from Henwood and Stuntz (1987) and W. Stuntz (pers. comm.) were the catch and effort data from the Cape Canaveral ship channel and surrounding area (approximately 24 km). This local area harbors large concentrations of sea turtles throughout the year, and high turtle catch rates there do not reflect those occurring outside the Canaveral area (Hen- wood and Stuntz, 19871. Elimination of those data provided conservative estimates of catch rates for the south Atlantic. Distribution of effort by depth in the Gulf of Mexico in the Henwood and Stuntz (1987) study is biased toward shallower waters than are usual or typical for the commercial shrimp fleet. The commercial fleet exerted 65% of total offshore shrimping effort in 27 m or less in 1988 (pers. comm., E. Klima, NMFS, 1989), whereas 84% of the total effort reported by Henwood and Stuntz (1987) was in 27 m or less. Statistically signifi- cant differences in capture rates among depths were not found in these data, and the data were pooled to provide the best estimates of capture rates in the Gulf of Mexico. However, information discussed in Chapter 4 strongly suggests that turtle abundance is negatively correlated with depth. The confidence intervals associated with estimates of mortality in Hen- wood and Stuntz (1987) do not incorporate the uncertainties associated with the estimated mortality rate, so they portray a lower level of uncer- tainty than is reflected by the data. Incorporating that uncertainty would broaden the confidence intervals about the estimates of mortality.

100 Decline of the Sea Turtles Henwood and Stuntz (1987) restricted their analysis of mortality rate (number of dead turtles per unit of tow time) to turtles classified as dead; they excluded turtles classified as comatose. Recent work (pers. comm., P. Lutz, University of Miami, 1989; Stuntz and Kemmerer, 1989) indicates that some comatose sea turtles die even after proper resuscitation tech- niques have been applied and the turtle becomes active. Internal injuries that are not visible in turtles landing on deck and are not initially totally debilitating are considered a factor in delayed mortality of trawl-caught sea turtles (pers. comm., D. Owens, Texas A&M University, 19891. If some or all comatose sea turtles die as a result of trawling, the Henwood and Stuntz study underestimates sea turtle mortality by a factor of as much as 3 (Figure 4-31. A final underestimate results from the Henwood and Stuntz (1987) study having considered only shrimping effort in waters outside the coastal beaches. Because 33% of the total shrimping effort in 1987 and 1988 occurred in rivers, estuaries, and bays and because sea turtles (espe- cially young Kemp's ridleys) are found in these waters, total mortality from the shrimp fishery could be higher than the Henwood and Stuntz estimates by a factor of as much as 1.6. That possibility is based on the assumption that the abundance of turtles is the same inside and outside and the assumption that a unit of effort is equal inside and outside; nei- ther of those assumptions is precisely true (nor known, for that matter). The limitations of the data and the criticisms of methods used do not detract from the basic findings of the Henwood and Stuntz study. With its assumption that all comatose turtles survive and its omission of all turtle capture and mortality estimates for inside waters, the approach taken by Henwood and Stuntz results in a marked underestimate of total sea turtle mortality associated with the shrimp fishery. Relation Between Sea Turtle Stranding and Spatiotemporal Paltern of Shrimp Trawling in North Carolina The northern limit of the geographic zone of ocean shrimp trawling occurs at Ocracoke Inlet, North Carolina. Data compiled by Street (1987) on sea turtle stranding on ocean beaches in North Carolina exhibit a spatiotemporal pattern that closely matches that of trawl fishing in the ocean offshore of that state. South of Ocra- coke Inlet, where offshore shrimp trawling continues from about May through September, 86% of the 545 sea turtle strandings observed in 1980- 1986 occurred in those months. In contrast, north of Ocracoke Inlet, where no shrimp trawling occurs, but where a winter trawl fishery for flounder exists, 85% of the 456 sea turtle strandings recorded on ocean beaches in 1980-1986 occurred during the October-April period (Street, 19871. The spatiotemporal switch in the season and location of apparent

101 Sea Turtle Mortality Associated with Human Activities sea turtle mortality suggests that shrimp trawling causes substantial mor- tality of sea turtles south of Ocracoke Inlet in North Carolina. The winter mortality of sea turtles to the north might be caused by groundfish trawl- ing or by temperature shocks in the colder-water biogeographic province north of Cape Hatteras. Other Fisheries, Discorded or Lost Gear, and Marine Debris Turtles are caught and killed in finfish trawls, seines, pompano gill nets in Florida (pers. comm., L. Ehrhart, University of Central Florida, March 1990), various kinds of passive fishing gear (such as gill nets, weirs, traps, and long lines), lost fishing gear, and other debris. We con- clude that the mortality associated with these and related factors is about one-tenth that associated with shrimp trawling (Table 6-21. Collectively, the nonshrimp fisheries constitute the second largest source of mortality of juvenile to adult sea turtles. That statement is based on the observa- tions documented below by region. The assessment of sea turtle mortality attributed to entanglement in stationary or fixed fishing gear is difficult, because of the disparity and discontinuity of reliable data. It is fair to assume that in some localities and with some types of fishing gear, entrapment and entanglement occur fairly often, but the resulting turtle deaths might not be as consistent. Most of the entangled or entrapped turtles are subadults and adults. Fishermen appear to be reasonably cooperative in efforts to set live sea turtles free. However, dead turtles are set adrift and might later be accounted for as strandings. The ratio of dead turtles set adrift to those counted as stranded is not adequately documented. If the approximate 4:1 ratio documented by Murphy and Hopkins-Murphy (1989) is consid- ered valid, some total estimate of mortality can be made. On the basis of yearly stranding data with mortalities directly associated with encounters with fixed fishing gear, a yearly estimate of a maximum of 45-400 sea tur- tle deaths is reasonable. That is only a crude estimate; more research and monitoring are necessary to document and understand the interac- tion of sea turtles with fixed fishing gear. Estimates of worldwide losses and discards of commercial fishing gear including plastic nets, lines, and buoys range from 1,350 to 135,000 metric tons of gear per year (Merrell, 1980; Welch, 19881. Recre- ational fishing in the United States is undoubtedly another important source of marine debris, including bait bags and lost and damaged gear (Pruter, 19871. NMFS recreational fishing statistics indicate that more than 81 million recreational fishing trips are made annually to marine waters (NMFS, 1986a,b).

102 Decline of the Sea Turtles It is especially difficult to document the deaths caused by this source or to separate them from deaths caused by fixed but unattended fishing gear. Yet, sea turtles (and other marine life) are particularly vulnerable to commercial fishing gear that has been lost or abandoned at sea. Such gear continues to catch and entangle marine life indiscriminately, causing injury, strangulation, starvation, and drowning (Carr, 1987; Laist, 1987; McGavern, 1989; Gregg, 19881. Deaths of green turtles, hawksbills, logger- heads, Kemp's ridleys, and leatherbacks have been caused by entrapment and entanglement in fishing gear (Mayer, 19851. Monofilament line is the most common type of debris to entangle turtles. Other debris includes rope, trawl netting, gill netting, plastic sheets, and plastic bags. Fishing- related debris is involved in about 68% of all cases of sea turtle entangle- ment (O'Hare and Iudicello, 19871. Other stationary or passive fishing gear that has caused deaths of turtles includes pound nets, long lines, sturgeon nets, and nylon and monofilament gill nets (Van Meter, 19831. Leatherbacks and green turtles are prone to entangling their front flippers and heads in buoy ropes or discarded twine (O'Hare et al., 1986; O'Hara and Iudicello, 19871. The largest authenticated leatherback ever recorded became entangled in whelk-fishing lines and drowned; fishermen cut the dead turtle loose, and the carcass washed up the next day on a beach in Wales (Morgan, 19891. Sea turtle entanglement in monofilament fishing line is a common problem. It is not usually related to active fishing; rarely is a fishhook reported attached to the line. In several cases, a turtle was entangled on line snagged on underwater structures or reefs, which caused constriction and necrosis of the limbs or drowning (O'Hare et al., 19861. Balazs (1985) acquired reports of 60 cases worldwide of turtle entan- glement involving monofilament line, rope, netting, and cloth debris. Of the 60 cases, 55 (92%) involved single animals, and 38% of all the turtles were dead or died later. Five species from 10 locations worldwide were reported; Kemp's ridleys were not included. Green turtles accounted for 19 (32%) of the 60 cases, and immature turtles were affected more often than adults in all the species represented except the leatherbacks. Only adult leatherbacks were reported entangled; immature leatherbacks are rarely reported anywhere. Monofilament line, with no fishhooks attached, accounted for 20 (33%) of the cases; segments or snarls of rope, 14 (23%~; pieces of trawl or webbing, 12 (20%~; and monofilament net, 8 (13%~. Fishing-related debris was involved in 41 (68%) of all the cases. New England Leatherbacks and Kemp's ridleys become entangled in lobster gear (O'Hare et al., 19861. Balazs (1985) reported a dead leatherback from Rhode Island that had a longline hook embedded in its flipper, with rope

103 Sea Turtle Mortality Associated with Human Activities attached. Although there have been no reports of sea turtle entanglement in gill nets in New England, reports of leatherbacks with cuts, severed limbs, or chafing marks suggest the possibility. Fretey (1982) published an extensive inventory of flipper injuries among leatherbacks in the large French Guiana nesting colony; some of these animals are known to come from feeding grounds in the northeastern United States. Balazs's (1985) compilation of worldwide incidence of sea turtle entanglement indicated that 11% of the 55 cases investigated involved monofilament net (O'Hare et al., 19869. New York Bight Turtle mortalities have resulted from lobster-pot lines and pound nets. Between 1979 and 1988, 58 stranded sea turtles reported in the New York Bight exhibited signs of entanglement with debris or inactive or fixed fishing gear. The Okeanos Ocean Research Foundation in New York reported two dead leatherbacks entangled in lobster gear in 1986 (O'Hare et al., 19861. Lobster-pot floatlines are a major source of entanglement, because they can be more than 180 m long in offshore waters and virtual- ly undetectable below the surface. Six of 10 leatherbacks were caught in lobster-pot lines, and one entangled and drowned (Balazs, 1985; Sadove and Morreale, 19891. In Long Island Sound, fixed pound-net gear traditionally captures the most sea turtles, predominantly Kemp's ridleys, but also green turtles, leatherbacks, and loggerheads (Morreale and Standora, 19891. The Okeanos Ocean Research Foundation has accumulated numerous reports of sea turtles, especially Kemp's ridleys, entrapped in pound nets in east- ern Long Island. Surveyed fishermen indicate catching 10-20 turtles per year. That might be important, because more than 100 licensed fisher- men were using pound nets in the region in 1986. It might not constitute a mortality problem, if the turtles are simply enclosed in the heart or head of the net until released, but deaths can occur if the turtles get tangled in the hedging or stringers (Balazs, 1985; Sadove and Morreale, 19891. Doc- umented cases from 1986 include seven Kemp's ridleys, four loggerheads, and two green turtles captured; all but four were released alive (O'Hare et al., 19861. Balazs (1985) reported a leatherback that was found dead, tangled in rope. Debris in the water column or at the surface, such as floating line, can entangle turtles during normal activities, such as surfac- ing to breathe (Balazs, 1985; Sadove and Morreale, 19891. Mid-Atiantic and Chesapeake Bay The principal fishery-caused mortality in the mid-Atlantic and the Chesapeake Bay is in the pound-net fishery in the bay during the summer and the finfish trawl fishery for flounder off the coast in the winter. Doc

104 Decline of the Sea Turtles umentation is best for the effects of the pound-net fishery. Other fishery- related mortality results from gill nets, crab-pot lines, and occasionally even rod-and-reel fishing. Some deaths in gill nets occur off Delaware (O'Hare et al., 19869. Almost all turtle stranding during October and November in Virginia and adjacent waters of North Carolina occurred on the ocean front where heavy flounder trawling takes place off the coast. Some of the stranded turtles showed net marks and might have drowned in fish trawls. The sea- sonality of stranding in North Carolina north and south of Cape Hatteras implicates the flounder fishery as the source of mortality. Low-tempera- ture deaths also might have contributed to the stranding. Further evalua- tion of this fall or winter mortality is warranted (Barnard et al., 19891. An estimated 50-200 sea turtles strand from all causes in and around the Chesapeake Bay each year (Keinath et al., 1987; personal communica- tion., D. Barnard and I. Keinath, Virginia Institute of Marine Science, October 19891. Stranding data for 1979-1988 were analyzed by Barnard et al. (1989) and D. Barnard and J. Keinath (pers comm., Virginia Institute of Marine Science, 19891. Of the turtles examined, 20% had definite net marks indicating death by pound net, gill net, or other fishing gear; 47% had no outward sign of injury or were very decomposed. The 20% figure is lower than previously estimated by Bellmund et al. (1987) and Keinath et al. (19871. Crab-pot lines and pound-net leads probably contributed to many of these deaths. Stranding of dead turtles in and around Chesapeake Bay typically begins in mid-May. That pattern coincides with the deployment of pound nets in May. However, pound nets are in use through October, whereas strandings tend to cease by early July. The higher number of strandings early in the season might be related to the emaciated or weakened state of turtles entering the bay after a long migration (Bellmund et al., 1987; pers. comm., D. Barnard and I. Keinath, Virginia Institute of Marine Sci ence, October 19891. Many pound-net deaths might be related to the inability of sick or injured turtles to avoid fixed nets during periods of strong tidal flow (Musick, 1988; Barnard et al., 19891. Pound-net hedging or leaders with stringers produced the highest mortality rates for turtles, 0.7 per net, espe- cially in strong currents. Pound-net leads composed of small mesh from top to bottom were associated with insignificant mortality rates. The tur- tle entanglement was 0.4 per net for open-water nets, compared with 0.1 per net for embayments and protected areas; the difference might be the result of the stronger currents in open water (Bellmund et al., 19871. In areas with weak currents, live turtles caught in pound nets apparently can move in and around the nets without becoming entangled (Bellmund et al., 1987), as evidenced by live turtles marked and released from one net

105 Sea Turtle Mortality Associated with Human Activities that have later been recaptured in the same or a nearby net and by the observation of a few loggerheads crawling out over the head netting as the net was being worked (Lutcavage, 19811. It is unlikely that stranded turtles without visible constrictions were killed in pound nets. Entangled turtles in pound nets die and begin to decompose in situ; they do not drift free to strand on shore (Bellmund et al., 19871. None of five dead turtles entangled in pound-net hedging dur- ing 1984 came loose over 5 weeks. However, the rotting turtle eventually bloats; as it decomposes, it tears free, floats away, and strands (Lutcavage, 19811. One dead and marked loggerhead from pound-net hedging stranded 5 days later 10 km from the net. Various reports have assessed the sources of mortality of dead sea tur- tles stranded on inshore beaches and shores in and around the Chesa- peake Bay. A total of 645 dead turtles, including 527 loggerheads and 28 Kemp's ridleys, stranded between May 1979 and November 1981. Necropsies of some loggerheads implicated enteritis and drowning (Lut- cavage and Musick, 19851. A sample of 71 turtles from 1979 included 25 with a determinable cause of death; seven of the deaths were caused by pound nets. Of the 57 turtles sampled in 1980, 21 had a determinable cause of death, and 19 deaths were caused by pound nets. In addition to pound-net deaths, one turtle died in a haul seine, one after being caught on a long line, and two in crab-pot lines (Lutcavage, 19811. Of the 124 turtles sampled in 1981, 11 had determinable deaths, and four deaths were caused by pound nets (Lutcavage and Musick, 19851. Confirmed netting deaths from 1979 to 1983 numbered 53 (19% of the determinable deaths); only four turtles (1.4%) died as a result of non-net fishing gear. Of the 83 dead stranded turtles examined in 1984,10 (12%) had evidence of constriction, and 20 (24%) were in pound or gill nets (Bellmund et al., 19871. Definite net-related deaths in the Chesapeake Bay during some summers from 1979 to 1984 ranged from 3% to 33% of the total number of stranded turtles (Lutcavage and Musick, 19851. Early reports indicated that the cause of death could be determined for about half the 980 stranded sea turtles recorded between 1979 and 1987 and that almost 40% could be attributed to entanglement in gill or pound nets (Keinath et al., 19871. However, reanalysis of the data available for 1979-1988 determined that fewer turtle deaths (approximately 20%) could be definitely attributed to entanglement in pound or gill nets, or other fishing gear (Barnard et al., 19891. South Atlantic Sea turtle deaths other than those caused by shrimp fishing have occurred in the south Atlantic in association with oceanic gill nets, large ocean set nets, and tuna and billfish long lines.

106 Decline of the Sea Turtles Turtle mortality associated with gill-net fisheries in the Carolinas starts In early spring and is maximal in April. The South Carolina Wildlife and Marine Resources Department reported that oceanic gill net fisheries for Atlantic sturgeon, shad, and shark have caused the deaths of loggerheads, Kemp's ridleys, and green turtles (pers. comm., S. Murphy, S.C. Wildlife and Marine Resources, 19891. In 1980-1982, about 217 turtles stranded in the early spring in connection with large ocean nets used to catch Atlantic In 10~-1C)RC, the .~tllr~on season was closed inmid-Aoril. and sturgeon. ~ ,_ ~ ,,, ~ DO the carcass count Decreased to about cup furrier. In 1986, there was no sturgeon season, and only about 18 turtles died in the spring. Illegal drift nets for the shad fishery and shark fishery were probably responsible tor most of the 36 carcasses reported in May 1989, including eight leatherbacks. Sea turtles are caught infrequently on long lines in the gulf and Atlantic (Swordfish Management Plan, 19851. On the basis of data from the 1979 Japanese long-line observer program, 12 turtles (including two leatherbacks) were caught in the gulf and 17 (including nine loggerheads) were caught in the Atlantic. During 1980, the same observer program r~n~rter1 an r,~rtl~.~ captured. The greatest number were captured in Jan- uarv-March in the gulf and in September-lanuary in the Atlantic. Seven ~ ~ ~ [ . , . c, ~ percent of the turtles captured died in gulf long-line fishery and 30% in incidental captures in the Atlantic (O'Hare et al., 19861. One unidentified turtle in 1987 and one leatherback in 1988 were hooked, as reported by observers on Japanese long-line Vessels fishing in the northwest Atlantic fishery conservation zone (FFOP, 1988, 19891. Leatherbacks tend to get hooked (either in the mouth or the flipper area), whereas loggerheads are prone to entanglement in the ganglion lines attached to the main line. In Florida, there were five recent confirmed sightings by divers of sea turtles entangled in monofilament fishing line on reefs and wrecks (pers. comm., J. Halusky, N.E. Florida Sea Grant Extension, May 19899. Two of them were rescued and released, and three were dead. Balazs (1985) reported 10 cases of turtle entanglement in Florida in 1978-1984: one green turtle, alive; five loggerheads, including three dead; and five hawksbills, alive. Georgia. Balazs also reported a dead loggerhead in Of the 11 cases, six involved monofilament fishing line, two involved rope, two involved line and netting. ~. gill or other netting, and one involved both  Gulf Coast A study along the Texas coast during 1986 and 1987 encountered entanglement of 25 turtles in discarded net and monofilament line. Entanglement was identified as the probable cause of death of seven; the remainder were stranded alive. Nine of the 25 turtles were Kemp's rid

107 Sea Turtle Mortality Associated with Human Activities leys, and the others were loggerheads, hawksbills, green turtles, and leatherbacks. The turtles were entangled in fishing line and hooks, shrimp trawls, onion sack, net and rope, tar, crab trap, and trot line. The study concluded that the pronan~ty that a sea turtle in Texas coastal waters would come into contact with marine debris is high, and that com- mercial and recreational fishermen and their discarded gear were respon- sible for most of entanglements (Plotkin and Amos, 1988; Ross et al., 19891. Balazs (1985) reported five entangled turtles in Texas in 1977-1983, including one live green turtle, three live hawksbills, and one dead hawksbill. Four of the entanglements involved monofilament fishing line, the other a piece of plastic onion bag. Amos (1989) reported that, in 77 recorded strandings of hawksbills in Texas since 1972, the incidence of entanglement in plastic was high 22% of those in which such informa- tion was recorded. The most common form of entanglement occurred when turtles' necks or limbs were caught in woven plastic produce sacks. Monofilament fishing line wrapped around limbs has also been recorded. No entanglements of recent posthatchlings have been noted, only entan- glements of yearlings. An anecdote from Paul Raymond of the NMFS Law Enforcement Divi- sion provides a dramatic example of the problem. An abandoned pom- pano trammel net (three panels) of monofilament was seized on October 16, 1989, off the beach (near shore) near Wabasso, Florida (Indian River County). It had been set 6 days before and left unchecked. In it were 10 juvenile green turtles and one juvenile loggerhead, all entangled and drowned. Pompano trammel nets are tethered in very shallow water near shore by fishermen in small boats. The industry is not well organized or documented as to size, season, or distribution. Nets set inshore (behind the coastal regulation lines) must be attended, but that is not required by Florida for nets outside the line. The net in question here had been aban- doned. An unattended but not abandoned net can also kill turtles. Dredging Dredging of harbors and entrance channels can kill sea turtles (Hop- kins and Richardson, 19843. A comprehensive survey of records and pro- ject reports recognized 149 confirmed incidents of sea turtles entrained by hopper dredges working in two shipping channels from 180 to 1990 (Table 6-6) (pers. comm., J.I. Richardson, University of Georgia, April 19901. Only verifiable records of fresh kills or live turtles were included in this table, explaining the slight difference in total counts between this survey and other reports (Rudloe, 1981; Joyce, 19821. Three species of sea

108 Decline of the Sea Turtles TABLE 6-6 Reported sea turtle incidents by species during dredging activities from 1980 to 1990. Site Year Loggerhead Green Turtle Unidentified Total Cape Canaveral 1980 50 3 1871 Entrance Channel 1981 1 1 13 1984-85 3 0 69 1986 5 0 05 1988 8 2 1828 1989-90 0 6 17 Totals 67 12 44123 King's Bay 1987-88t 7 1 19 Entrance Channel, 1988 3 0 27f Georgia and Florida 1989 9 0 110 Totals 19 1 26 Fragments of sea turtle carcasses not identified to species. It is assumed that most are log- gerheads. tInitia1 construction dredging for Trident submarine base. fTwo Kemp's ridleys caught in 1988 at Kings's Bay, Georgia. Source: Richardson, 1990. turtles were taken, including two individuals of the endangered Kemp's ridley. Although some entrained specimens were identified, it is estimat- ed that 90% or more of the incidents involved the loggerhead. Nearly all sea turtles entrained by hopper dredges are dead or dying when found, but an occasional small green turtle has been known to survive. Entrapment and death of turtles by hopper dredges first became an issue of concern at the Port Canaveral Entrance Channel, Florida, in 1980 after unusually high concentrations of loggerheads were noted in the area (Carr et al., 19819. Seventy-seven loggerheads were reported killed in 1980 dur- ing the removal of 2.5 million cubic yards (1.9 x 106 m3) of sediment from the channel (Rudloe, 1981;.Toyce, 19821. The rate of turtle take varied among dredges, ranging from 0.038 turtle entrained per hour (dredge McFarland) to 0.121 turtle entrained per hour (dredge Long Island) Joyce, 1982~. The very high number of turtles taken was not repeated in subse- quent years for several reasons. First, the Long Island, because it seemed to pose the greatest threat, was transferred immediately to other areas. Sec- ond, a program of gear modification to the drag heads was initiated at that  time. Finally, the loggerheads did not seem to use the Canaveral Channel in the same numbers in later years. By 1989, the rate of sea turtle capture in surveys in the channel were about one-tenth the rates recorded in 1978- 1983 (pers. comm., A. Bolten, University of Florida, 19891.

109 Sea Turtle Mortality Associated with Human Activities Although the most serious loss of turtles in hopper dredges occurs within the Port Canaveral Entrance Channel, smaller numbers have been taken at King's Bay Entrance Channel. Twelve turtles (10 juvenile logger- heads, one adult loggerhead, one juvenile green turtle) were taken dur- ing some 20,000 hours of construction dredging (Slay and Richardson, 19881. The rate of capture was less than 0.001 turtle per dredge hour. The loss of turtles to hopper dredges in other entrance channels is not yet known, but other entrance channels from North Carolina to Florida will be surveyed by NMFS to assess any potential effects on sea turtles (pers. comm., J. Richardson, University of Georgia, 19901. Data are being gathered through additional observer programs to answer the question, and the numbers of sea turtles taken are expected to be considerably smaller than observed at Port Canaveral. The data are not available, but it would not be unusual for 1,000 hours of maintenance dredging to be needed per channel per year. Collisions with Bools Another source of mortality to sea turtles associated with human activi- ty is collision with vessels. The regions of greatest concern are those with high concentrations of recreational-boat traffic, such as the south- eastern Florida coast, the Florida Keys, and the many shallow coastal bays in the Gulf of Mexico. Of the turtles stranded on the gulf and Atlantic coasts of the United States, 6% of 1,847 strandings in 1986, 7% of 2,373 in 1987, and 9% of 1,991 in 1988 had boat-related injuries for an average of about 150 turtles per year (Schroeder, 1987; Schroeder and Warner, 1988; Teas and Martinez, 19899. In most cases, it was not possi- ble to determine whether the injuries resulted in death or were post . . . mortem injuries. In the Chesapeake Bay region, boat-propeller wounds accounted for approximately 7% of the deaths of sea turtles stranded in 1979-1988 whose causes were determinable (Barnard et al., 1989), or about five to seven turtles per year. If we assume that half the boat-collision injuries documented by the STSSN were the primary causes of death of the stranded sea turtles in 1986-1988, and only about 20% of the dead turtles wash ashore, about 400 turtles are killed by boat collisions each year along the gulf and Atlantic coasts of the United States outside of coastal beaches. That esti- mate might be low, because the strandings include only the ocean beach- es (boat collisions with turtles also occur in inside waters), and an animal with an open wound has an increased probability of predation and thus a further reduction in probability of stranding. The estimate might be

110 Decline of the Sea Turtles high, because more than half of the turtles might have been hit when they were already dead from other causes and were floating. Petroleum-Platform Removal The use of explosives in removal of petroleum structures became con- troversial with respect to turtle mortality in 1986. From March 19 to April 19, 1986, 51 turtles, primarily Kemp's ridleys, were found dead on beach- es of the upper Texas coast. Ten petroleum structures in the nearshore area of the strandings had been removed with explosives during the peri- od. Shrimping was at a seasonal low, and circumstantial evidence suggest- ed that at least some of the strandings were due to underwater explosions used in removal of the structures (Klima et al., 19881. Further evidence of the serious effects of the explosions included the stranding of 41 bot- tlenose dolphins (Tursiops truncatus) and large numbers of dead fish (Klima et al., 19881. After those incidents, attention focused on the possible effects of petroleum-platform removal. In July 1986, 11 sightings of at least three turtles (two loggerheads and one green turtle) occurred during the removal of a platform 30 miles south of Sabine Pass, Texas. What appeared to be a dead or injured turtle drifting with the current 10 feet below the surface was reported 1.5 hours after detonation of explosives (Gitschlag, 19891. Six sightings of loggerheads were reported at five other removal sites, and a green turtle was observed at another location. Those sightings and strandings resulted in a consultation under Section 7 of the Endangered Species Act of 1973 between NMFS and the Minerals Manage- ment Service (MMS). Oil and gas companies wishing to use underwater explosives were thereafter required to submit permit requests to MMS. Obtaining a permit requires use of qualified observers to monitor sea tur- tles near platforms and in some cases to remove turtles to a safe location away from the potential impact of explosive charges. Data collected by NMFS since 1986 support an association between tur- tles and some offshore platforms. Divers have reported that turtles com- monly associate with offshore structures (Rosman et al., 19879. Gitschlag (1989) reported that 36 turtle sightings near platforms scheduled to be removed were made during 1987-1988 by the NMFS observer program. Another 30 turtles were observed during that period at structures not scheduled for removal (personal communication, G. Gitschlag, NMFS, 19893. A recent NMFS observer effort indicated that turtle concentrations near a petroleum platform could be large. Twelve turtles were collected and removed from one structure off Texas in September 1989 (pers. comm., G. Gitschlag, NMFS, 19891.

111 Sea Turtle Mortality Associated with Human Activities Additional reports confirm the association of turtles with offshore plat- forms. Lohoefener (1988) used aerial surveys and found hard-shelled sea turtles (cheloniids) to be associated with platforms offshore of the Chan- deleur Islands (Louisiana), although their study did not indicate an associ- ation of sea turtles with platforms in the western Gulf of Mexico. They determined the daytime probability of one or more cheloniids near a platform off the Chandeleur Islands to be about 0.27 within 500 m of the structure, 0.50 within 1,000 m, and 0.65 within 1,500 m. West of the Mis- sissippi River, the probability of one or more cheloniids within 500 m of a randomly selected platform would be about 0.04, within 1,000 m about 0.08, and within 1,500 m about 0.13. Only larger turtles and only turtles on or near the surface are usually seen by aerial surveys, so the figures given should be considered low. Although information on association of sea turtles with energy plat- forms is sparse, the potential for mortality must be considered genuine. It is difficult to document a cause-effect relation between turtle deaths and offshore explosions, because no dead animals have been recovered at removal sites and freshly killed turtles sink and might drift a long way by the time putrefaction causes them to float. Association of turtles with the structures is not random; platforms apparently provide a resting place or a location where food is readily available (Klima et al., 19881. From March 1987 through 1988, 69 platforms and 39 caissons or other single- pile structures were removed in gulf waters of Louisiana and Texas. MMS estimated that there were 3,434 platforms in the federal outer continental shelf as of December 1986 and predicted that 60-120 structures would be removed each year for the next 5 years (MMS, 19881. Continuing research should identify more specifically the negative effects of explo- sive removal of offshore structures. Safeguards for protection of turtles near structures scheduled for removal are essential. To estimate the numbers of sea turtles that might be killed or injured by explosions in the future, we assumed that the injury and mortality zone will extend no farther than 1,000 m from the structure being removed (Klima et al., 19884. For the Chandeleur Islands area, where the highest densities were seen, Lohoefener et al. (1988) used aerial surveys and estimated a 0.5 probability that a turtle would be visible within 1,000 m of a given structure during the day. If about 100 platforms in gulf waters of Louisiana and Texas will be removed each year over the next 10 years, a total of 8-50 turtles each year could be killed or injured with- out protective intervention. That estimate is biased downward for two reasons: first, an aerial survey samples only during the day, when turtles are known to forge away from resting sites. Second, turbidity in the Gulf of Mexico may reduce visibility from the air, especially west of the Missis- sippi. Yet, Klima et al. (1988) estimated higher densities of turtles in this

112 Decline of the Sea Turtles region during the observer programs. If only half of the turtles are seen during aerial surveys, the estimate could reach as high as 100 turtles pos- sibly affected each year over the 10-year period. Other uses of explosives also might have an effect. Petroleum seismo- graphic exploration ana m1lllary maneuvers can urn =~plO~lV=~. Their impact on turtle mortality has not been measured, but it might exist. In contrast, turtles nesting in areas adjacent to military bombing activities (e.g., on eastern Vieques Island, Puerto Rico) might actually benefit, because the control of human access and the danger of unexploded rounds greatly reduce the presence of egg poachers (pers. comm., P. Pritchard, Florida Audubon Society, October 19891. ~ . . . . . · . - . ~ 1 ~ in: .. ~ ~ Entrainment of Sea Turtles in Power-Plan' Intake Pipes Sea turtles can become entrained in intake pipes for cooling Upstater at coastal power plants. The best-documented case is that of St. Lucie unit 2 in southeastern Florida. At that facility, nets are constantly set and moni- tored in the intake canal to remove sea turtles. In 1976-1988, 122 (7.5%) of the 1,631 loggerheads and 18 (6.7%) of the 269 green turtles entrapped in the canal were found dead, for an average of about 11 turtles per year (Applied Biology Inc., 1989a). Four Kemp's ridleys were found dead dur- ing the same period. No dead leatherback or hawksbill has been found there (Applied Biology Inc., 1989a). Deaths resulted from injuries sus- tained in transit through the intake pipe, from drowning in the capture ~ cat nets, and perhaps from causes before entrainment. At four other power plants in Florida (Port Everglades, Turkey Point, Cape Canaveral, and Riviera Beach), 21 turtles (loggerheads, green turtles, and one hawksbill) were entrained in the systems from May 1980 through December 1988. ^ ~ ~ ~ Of the 21, seven were found dead (four ot which were loggerheads), for an average of about one per year. At the Port Ever- glades plant, 25-30 hatchlings were also entrained in the system, and a few of them died (Applied Biology Inc., 1989b). Other turtle deaths at coastal power plants have been reported in New Jersey (Eggers, 1989), North Carolina, and Texas (pers. comm., T. Hen- wood, NMFS, 1989; pers. comm., B. Schroeder, Florida Department of Natural Resources, 19891. They were sporadic and apparently involved few turtles. For example, the Delaware Bay Power Plant in New Jersey entrapped 38 turtles in 9 years 26 loggerheads (18 dead) and 12 Kemp's ridleys (six dead), for an average of about three per year. Two factors cause an unusually high entrainment rate at the St. Lucie  unit 2 power plant in Florida. First, the continental shelf is narrow in that area, and that seems to cause the normally high density of turtles passing

113 Sea Turtle Mortality Associated with Human Activities along the coast to be concentrated near the shore, where the coolant- water intake tube is. Second, that part of the coast appears to be on the main coastal migratory route for turtles in the region. Therefore, mortality rates for this power plant should be considered separately. A total mor- tality estimate of about 11 turtles per year might be expected in the future at current population densities: about 9.4 loggerheads, one green turtle, and 0.3 Kemp's ridley. For other power plants, far less is known. According to the Edison Electric Institute (1987), 98 power-generating facilities use ocean or estu- arine water for their cooling systems along the gulf and Atlantic coasts of the United States. If we assume that rates of turtle capture from the five power plants discussed above (excluding the St. Lucie facility) are typical for the remaining coastal facilities between New York and Texas, we can estimate an annual mortality of 48 loggerheads and 13 Kemp's ridleys (loggerheads, 98 power plants x 0.48 per year; Kemp's ridleys, 98 power plants x 0.13 per year). Adding in the estimates from the St. Lucie plant raises the loggerhead to 57 per year and Kemp's ridley to 13 per year. An important consideration for the future is that, as turtle populations increase, we would expect an increase in the number of animals entrained in the facilities, and as human populations increase, more power plants might be built. Directed Take Directed take of sea turtles and their eggs is illegal in the United States and along the Caribbean and gulf coasts of Mexico. Some illegal take does occur in the United States and Mexico, but the numbers are proba- bly negligible. Loss of eggs and adult Kemp's ridleys at Rancho Nuevo is minimal, because protection has been provided. Although directed take of sea turtles is widely considered to affect populations, at least locally (Pritchard, 1980), the committee was unable to quantify the extent of the problem. Toxicology Tissues and eggs from several species of sea turtles in the southeastern United States, Ascension Island in the South Atlantic, the coast of France, and other geographic regions have been analyzed for organochlorine compounds, heavy metals, hydrocarbons, and radionuclides (Hillestad et al., 1974; Thompson et al., 1974; Stoneburner et al., 1980; Clark and Krynitsky, 1980, 1985; Witkowski and Frazier, 1982; Bellmund et al.,

114 Decline of the Sea Turtles 19851. Turtles were found to be contaminated to various degrees in all the studies cited. However, because of the lack of data on physiological effects of the pollutants in sea turtles, their effect on survival cannot be estimated. Additional studies are needed to determine extents of contami- nation and the physiological effects of the contaminants. Ingestion of Plastics and Other Debris About 24,000 metric tons of plastic packaging is dumped into the ocean each year (Welch, 19881. Nationwide 10-20% of beach debris is expanded polystyrene foam and 40-60% is other plastic (McGavern, 19891. An estimated 1-2 million birds and more than 100,000 marine mammals and sea turtles die from eating or becoming entangled in plastic debris each year, including netting, plastic fishing line, packing bands, and Styrofoam (Welch, 1988; McGavern, 1989; Sanders, 19899. Sea turtles ingest a wide variety of synthetic drift items, including plas- tic bags, plastic sheeting, plastic particles, balloons, Styrofoam beads, and monofilament fishing line. Specific reports have been related to green turtles in Hawaii, Florida, and Texas; loggerheads in Georgia, Florida, Texas, and Virginia; hawksbills in Florida and Hawaii; and leatherbacks in New York, New Jersey, Massachusetts, and Texas (Wallace, 1985; O'Hara et. al., 19869. Turtles can mistake plastic bags and sheets for jellyfish or other prey. Ingestion of those items can cause intestinal blockage; release toxic chemicals; reduce nutrient absorption; reduce hunger sensation; inhibit feeding and mating activity; diminish reproductive performance by leaving the turtle unable to maintain its energy requirements and cause suffocation, ulceration, intestinal injury, physical deterioration, malnutri- tion, and starvation (Wehle and Coleman, 1983; Wallace, 1985; O'Hara et al., 1986; Bryant, 1987; Farrell, 1988; Gramentz, 1988; Welch, 1988; McGavern, 19891. Absorption of toxic plasticizers (such as polychlorinated biphenyls) is also possible as a result of ingestion. Some plasticizers can concentrate in tissues, and the toxic ingredients can cause eggshell thinning, tissue dam- age, and aberrant behavior (Wehle and Coleman, 1983; O'Hara et al., 19861. Plastic bags blocked the stomach openings of 11 of 15 leatherbacks that washed ashore on Long Island during a 2-week period. Ten had four to eight quart-sized bags, and one had 15 quart-sized bags (San Francisco Chronicle, 1983; Balazs, 1985; O'Hara et al., 19861. In South Africa, Hughes extracted a ball of plastic from the intestine of an emaciated leatherback; when unraveled, it measured 9 x 12 ft. or 2.7 x 3.7 m (Bal

115 Sea Turtle Mortality Associated with Human Activities ads, 1985; Coleman, 19871. In September 1988, the largest leatherback ever recorded (914 kg) was found dead on a beach in Wales. The cause of death was listed as drowning due to entanglement, but a tightly com- pacted piece of plastic (15 x 25 cm) blocked the entrance to the small intestine and might have contributed to death (Eckert and Eckert, 19881. Accumulation of pollutants and plastic debris found in sargassum drift- lines might be a source of mortality of turtles through ingestion (Mayer, 19851. Floating debris is concentrated by natural processes along lines of convergence between discrete water masses, in the core of major current gyres, or on beaches and submerged rocky outcrops. Driftlines along margins of small temporary eddies or areas of downwelling can accumu- late floating debris and provide feeding areas for turtles. Young turtles are passive migrants in offshore driftlines and can contact buoyant debris. In 1979-1988 in the New York Bight area, necropsies were performed on 116 sea turtles. Various amounts of synthetic materials were found in 10 of 33 leatherbacks, three of 35 loggerheads, one of four green turtles, and none of 44 Kemp's ridleys. Most prevalent were plastic bags, small pieces of plastic sheeting, monofilament line, small pieces of variously colored plastic, and numerous small polystyrene balls. There was strong evidence in some animals that ingestion of synthetic materials caused their deaths. There is little information on the residence times and cumu- lative effects of synthetic materials in marine animals. These observations are not well suited to quantify the frequencies of ingestion (Sadove and Morreale, 19891. Studies conducted along the Texas coast in 1986-1988 documented the effects of marine debris on sea turtles (Plotkin and Amos, 1988; Stanley et al., 1988; Plotkin, 19891. They were significantly affected by ingestion of, and to a smaller extent entanglement in, marine debris. Necropsies of Kemp's ridleys, loggerheads, and green sea turtles revealed that the intestines of at least 65 of 237 turtles examined contained marine debris, such as plastic bags, Styrofoam, monofilament fishing line, polyethylene beads, aluminum foil, tar, glass, and rubber. In a 22-month study, plastic was found in nearly 80% of the turtle stomachs that contained debris and in turtles from about 97% of the beaches surveyed (Stanley et al., 19881. All five species found in the Gulf of Mexico had eaten or were ensnared by debris. Reports of debris ingestion by species indicated that green turtles had the highest incidence (32%) followed by loggerheads (26%), leatherbacks (24%), hawksbills (14%), and Kemp's ridleys (4%~. For all species except the leatherback, immature turtles were involved more frequently than adults. The distribution of debris types was as follows: plastic bags and sheets (32.1%), tar balls (20.8%), and plastic particles (18.9%~. NMFS scientists, on the basis of the results of autopsies conducted

116 Decline of the Sea Turtles since 1978, estimated that one-third to one-half of all turtles have ingested plastic products or byproducts (Cottingham, 19881. Mortality associated with ingestion of plastics and debris cannot be accurately quantified from available data. Of the turtles examined, green turtles ingested plastic debris most frequently, followed by loggerheads and leatherbacks. Research is needed to develop accurate postmortem techniques to determine the role of plastic ingestion on turtle deaths. However, many reported stranded turtles are in an advanced state of decomposition, so it is difficult to determine exact causes of death, although indigestible stomach contents might still be identifiable. It is possible that the enactment of Annex V of the International Convention for the Prevention of Pollution from Ships (called bLARPOL for "marine pollution") might affect the amount of plastic that sea turtles are likely to encounter in the future; but, considering the life span of plastic and the amount already present in the oceans, the possible deleterious effects of plastic on sea turtles and other wildlife will be present for generations to come. SUMMARY Sea turtles are susceptible to human-caused deaths through their entire life, from nesting females, eggs, and hatchlings on beaches to juveniles and adults of both sexes in offshore and inshore waters. The committee found that the most important source of mortality on eggs and hatchlings at present on U.S. beaches is from non-human preda- tors, whose abundance is often associated with human disturbance, but other factors are beach development, directed take, beach vehicles, and beach lighting. The most important source of mortality for juveniles to adults in the coastal zone is shrimp trawling. Other factors judged to be of significance for juveniles and adults are other fisheries and entangle- ment in lost fishing gear and marine debris. Order-of-magnitude estimates of human-caused mortality on juvenile to adult loggerheads and Kemp's ridleys were made by the committee. Shrimp trawling accounts for 5,000-50,000 loggerhead and 500-5,000 Kemp's ridley mortalities per year. Other fisheries and discarded fishing gear and debris account for 500-5,000 loggerhead and 50-500 Kemp's rid- ley mortalities. Dredging, collisions with boats, and oil-rig removal each account for 50-500 loggerhead and 5-50 Kemp's ridley deaths. Entrain- ment in electric power plants and directed take each account for fewer than 50 turtle deaths per year. Based on the committee's evaluation, about 86% of the human caused mortalities on juveniles and adults result

117 Sea Turtle Mortality Associated with Human Activities from shrimp trawling. The committee recognized the possible effects of plastic ingestion and marine debris but was unable to quantify them. The strong evidence that shrimp trawling is the primary agent for sea turtle mortality caused by humans comes from five lines of analysis and information First, the proportion of sea turtles caught in shrimp trawls that are dead or comatose increases with an increase in tow time from 0% during the first 50 minutes to about 70% after 90 minutes. Second, the numbers of turtles stranding on the coastal beaches consistently increased in a steplike fashion when the shrimp fishery opened in South Carolina and Texas and decreased when the fishery closed in Texas. Because the openings and closings were on different dates in different years, the change in strandings can be ascribed to the fishery rather than to date per se. The change in stranding rate indicates that 70 to 90% of the turtles stranded at those times and places were killed in shrimp trawls. Based on analysis of data from loggerheads, these stranded tur- tles were also in the life stages with the highest reproductive values. Third, loggerhead nesting populations are declining in Georgia and South Carolina where shrimp fishing is intense, but appear to be increasing far- ther south in central to southern Florida where shrimp fishing is low or absent. Fourth, the estimate in the literature of 11,000 loggerheads and Kemp's ridleys killed annually by shrimp trawling was judged by the committee to be an underestimate, possibly by a factor of three to four, because that estimate accounted for neither mortality in bays, rivers, and estuaries nor the likely deaths of most comatose turtles brought onto the deck of shrimp trawlers. Many of the comatose turtles will die even when released back into the water. Fifth, in North Carolina, turtle strand- ing rates increase in summer south of Cape Hatteras when the shrimp fleet is active and north of Cape Hatteras in winter when the flounder trawling is active.