Leatherback Sea Turtles and the California/Oregon Drift Gillnet Fishery

Leatherback Sea Turtles and the California/Oregon Drift Gillnet Fishery Prepared for the California Seafood Council by LGL Ecological Research Associates, Inc Bryan, Texas November 2001 i a S e a f o o d T a s t e T h e T r a d r n f o l i i t i o n C a California Seafood Council PO Box 1951, Buellton, CA 93427 (805 693-5430

LEATHERBACK SEA TURTLES AND THE CALIFORNIA/OREGON DRIFT GILLNET FISHERY By Benny J. Gallaway, Ph.D. LGL Ecological Research Associates, Inc. 1410 Cavitt Street Bryan, TX 77801 For California Seafood Council Contract No. CSC-R97-01 20 November 2001

TABLE OF CONTENTS Page INTRODUCTION POPULATION STATUS.1 7 DISTRIBUTION, MOVEMENTS, AND MIGRATION 14 Nesting and Homing.14 Juvenile Distributions and Movements 15 Adult Females.16 Local (California/Oregon Movements.. 20 THE CALIFORNIA/OREGON DRIFT GILLNET FISHERY...25 Annual Patterns of Effort Quarterly Patterns of Effort Monthly Patterns of Effort 25 26.30 OBSERVER DATA.32 MANAGEMENT RECOMMENDATIONS 35 LITERATURE CITED 38 APPENDICES i

LIST OF FIGURES Figure Page 1 Area of the NMFS preferred Alternative 4 closure showing year and location of leatherback sea turtles taken during the month of September...4 2 Area of the NMFS preferred Alternative 4 closure showing year and location of leatherback sea turtles taken during the month of October 3 Area of the NMFS preferred Alternative 4 closure showing year and location of leatherback sea turtles taken in November and January....5.6 4 The distribution of leatherback nesting colonies around the world 8 5 Post-nesting movements of 3 leatherback sea turtles tracked by satellite from the Caribbean Island of Trinidad 17 6 Post-nesting movements of 9 leatherback sea turtles from Mexiquillo Beach, Mexico.18 7 Summer-fall distributions of the 12 to 15 C isotherms on the west coast of North America based on long-term averages 22 8 Leatherback CPUE by latitude..24 9 Annual levels of fishing and observer effort, 1990-2000.. 28 10 Annual fishing effort for the California/Oregon Drift Gillnet Fishery for quarters 3, 4, and the 1 st quarter of the following year, 1990-2000..29 11 Monthly patterns of fishing effort in the California/Oregon Drift Gillnet Fishery, 1998-2000 31 12 Annual observer effort for the California/Oregon Drift Gillnet Fishery for quarters 3, 4, and the 1 st quarter of the following year..33 13 Catch-per-unit-effort of leatherback sea turtles for fishing seasons 1990-2000 expressed as number of turtles taken per observed set..34 14 Industry proposal for an alternative time/area closure, shaded area would be closed to fishing from 1 September to 31 January. 37 LIST OF TABLES Table 1 Quarterly and annual observer effort, fishing effort, observed leatherback catch, CPUE data, and estimated fishery take of leatherbacks, 1990-2000 Page.27 ii

LEATHERBACK SEA TURTLES AND THE CALIFORNIA/OREGON DRIFT GILLNET FISHERY By Benny J. Gallaway, Ph.D. LGL Ecological Research Associates, Inc. 1410 Cavitt Street Bryan, TX 77801 INTRODUCTION The National Marine Fisheries Service s (NMFS Biological Opinion of October 23, 2000 determined that the issuance of a Marine Mammal Protection Act (MMPA permit to the California/Oregon drift gillnet fishery for swordfish and sharks (CA/OR DGF would likely jeopardize Pacific populations of leatherback and loggerhead sea turtles (NMFS 2000. This determination was based upon estimates that as many as 17 leatherbacks and 11 loggerheads might be killed each year (the estimate for loggerheads is for El Niño years only in a fishery characterized by about 3,000 sets per year. After review of the NMFS (2000 BO, Gallaway (2001, Appendix 1 suggested that the estimates of take and mortality were unrealistic and that the future operation of the CA/OR DGF would result in an impact below the jeopardy threshold. NMFS, in a letter dated 10 July 2001 from Dr. William T. Hogarth, Acting Assistant Administrator for Fisheries (also in Appendix 1, did not agree with the conclusions that were reached in Gallaway (2001. NMFS agreed that overall effort in this fishery has declined, but defended their approach of using effort averaged over the past three years (1997, 1998, and 1999 as a reasonable estimate of how the fishery might operate in the next three years. Data are now available for all of 2000 and preliminary data are available for the first quarter of 2001 (Appendix 2. This new information shows that the effort continued to decline in 2000 (1,936 sets in 2000, as compared to 2,634 sets in 1999 and 3,353 sets in 1998 and that the average number of total sets over the most recent three-year period was 2,641 sets. We continue to believe that these new data support the contention that the fishery effort will most likely continue to decline or, at most, remain stable below 2,000 sets on an annual basis. With regards to leatherback takes, NMFS argued that the observed reduction in the leatherback entanglement rate since the implementation of the Pacific Offshore Cetacean Take Reduction Plan (TRP could not be demonstrated to be statistically significant based upon only three years of data. They further noted that it was not possible to conclude that the apparent reduction was associated with the regulatory measures of the TRP based upon the limited availability of the data. NMFS suggested some alternative explanations that might account for the reduction including: 1

The lack of leatherback takes in 2000 could be that the majority of the observed sets were south of Point Pinos in an area where there has historically been a lower entanglement rate; The apparent reduction in takes could also be the result of a change in environmental conditions rather than the effects of the TRP; and The reduction could be the result of an overall reduction in the number of leatherback turtles in ocean waters off California. Based upon these explanations, NMFS concluded that the implementation of the reasonable and prudent alternative (RPA identified in the BO to protect leatherback and loggerhead sea turtles 1 was necessary (NMFS 2000. We address these explanations below. On 20 August 2001, NMFS (2001 released an environmental assessment of the implementation of the RPA identified in the NMFS (2000 BO. This assessment included some new alternatives. The original NMFS proposal to protect leatherbacks was to close the area from Point Conception (134º27 N north to 45ºN and west to 129ºW from August 15 to October 31 for a period of three years (2001-2003. Historically, five leatherbacks had been observed taken outside this time area, and 18 had been taken within the closed area. This was taken to indicate that there would be a 78% reduction in leatherback takes if the closure was enacted. Assuming that, on average, 39 leatherbacks would be entangled over the next three years (13 per year times 3 years, the proposed closure would reduce the total take to nine leatherbacks of which six might die. These levels of take and mortality were determined as being not likely to result in jeopardy to the leatherback sea turtle population. Of interest, the data for the last three years show that the leatherback take in the fishery was eight or nine individuals based on two observed captures over the period. Both turtles were released alive. The Take Reduction Team (TRT at its May 2001 meeting suggested a different measure that could provide the same level of leatherback protection as the RPA proposed by NMFS in the BO, yet still allow fishing to occur north of Point Conception during September and October. This alternative (Alternative 3 in NMFS 2001 was to close the area from 36º15 N to 45ºN and west to 129ºW from August 15 to November 15, for a period of three years (2001-2003. Again, only 5 of the 23 historical entanglements since 1990 have occurred within the time area closure which suggests an equivalent level of protection as provided by the NMFS proposal while allowing some fishing to occur north of Point Conception. No leatherbacks have been taken in the CA/OR DGF in August over the entire period of record. This closure could reasonably be changed to 1 September to 15 November. NMFS (2001 modified the area aspect of the TRT alternative based upon satellite telemetry tracking data obtained from two leatherback sea turtles that were tagged in 1 The focus of this response is leatherback sea turtles. We will address loggerheads in a future response. 2

Monterey Bay in September 2000. This proposal, the Preferred Alternative 4, was to close the area bounded by a line extending from shore at Point Sur (36º18.5 N to the point 34º27 N, 123º35 W, west to 125ºW, north to 45ºN, then east to where the 45ºN parallel reaches land, from August 15 to November 15, for a period of three years (2001-2003. This Alternative is in effect as of this year under the interim final rule. Again, there is no basis for initiating this closure prior to 1 September. In Figures 1-3, we show the area of the proposed closure and the distribution of the historical takes in September (Figure 1, October (Figure 2, and November-January (Figure 3. Documented takes have begun in September (4 of the 23 taken since 1990, peaked in October (14 of 23 and 5 of 23 have been taken over the three-month period November (2-December (2-January (1. Since our original review of the NMFS (2000 BO we have conducted a review of the literature and other information regarding leatherback sea turtle populations, especially as this information relates to the populations affected by the CA/OR DGF. We have requested and obtained from NMFS the updated fishing and observer effort data compiled by the California Department of Fish and Game (CDF&G for the period 1990 through the first quarter of 2001. These data were available on a quarterly level of temporal resolution. In order to evaluate monthly trends in effort by area, we obtained monthly fishing effort data by fishing block directly from the CDF&G. Month-by-block fishing effort data having comparable input data were available for the period 1997 to 2000. It is our understanding that observer data were used to allocate effort to fishing blocks if available; if not, logbook data were used. Landings data were used to allocate effort only in instances where there were no observer or logbook data. We have also requested and received from NMFS, a summary of the protected species take and disposition data by date and haul location, and have requested a tabulation of the location and date of each observed haul from the observer program since 1990. These data are being compiled but have yet to be received. Based upon analyses of these data and the review of additional literature information, we have updated and reevaluated our original leatherback assessment. Below, we first characterize the status of leatherback populations worldwide, and then discuss selected aspects of their biology (migration and movements that are pertinent to this assessment. We next discuss the CA/OR DGF in terms of the temporal and spatial patterns of fishing effort and the corresponding observations of leatherback takes since 1990. Lastly, we propose additional alternatives for time/area closures that would afford equal protection as that proposed by NMFS, and greater opportunity for fishing. 3

4

5

6

POPULATION STATUS Ross (1982 assembled data on the population status of leatherbacks and estimated that there were between 29,000 and 45,000 adult females worldwide. Pritchard (1982, based upon a brief aerial survey of Pacific Mexican beaches, added an estimated 70,000 turtles for that area alone. When he included estimates for additional beaches not included in the Ross (1982 estimate, he with a single stroke nearly tripled the world estimate by raising the total estimate to on the order of 115,000 adult females. Spotila et al. (1996a, working with a Costa Rican nesting assemblage of leatherbacks, became interested in determining how important this colony was to the overall population. They found that, in 1994, the world population had dramatically declined since the early 1980 s and that, with only 20,000 to 30,000 adult females in the world population, this species was in imminent danger of extinction. Spotila et al. (1996b revised the initial estimates to 34,529 adult females (26,177 to 42,878. These new estimates were based on data from 28 documented nesting colonies (Figure 4. Given that these estimates were still less than one-third of Pritchard s (1982 estimates, the authors again contended that leatherbacks were on the road to extinction. A generic, regional comparison of the estimated changes between the Pritchard (1982 and the Spotila et al. (1996b estimates is shown: 1982 1996 % Reduction (Increase Atlantic 15,000 27,608 (184% East Pacific 87,000 4,638 95% West Pacific & Indian 13,000 2,283 82% 115,000 34,529 Atlantic populations of leatherbacks had almost doubled whereas Pacific populations had declined. The Western Pacific and Indian Ocean Region population was estimated to have declined by 82%, and a 95% decline was estimated for the Eastern Pacific Region. Spotila et al. (1996b believed that leatherbacks were in most danger in the Indian Ocean and Western Pacific, and healthiest in the Atlantic. It is worthwhile to note that the population size for adult females has often been estimated from a survey or other estimate of the number of nests at a particular locality. The number of nests is divided by an average of the number of times a female nests during a given year. Spotila et al. (1996b used five nest/female for this aspect of the equation. This value is then multiplied by the re-migration or re-nesting interval (years between nesting to obtain the total population of adult females (i.e., not all are on the beach in a given year. Spotila et al. s (1996b estimates used an average re-migration or re-nesting interval of 2.5 years. Spotila et al. (2000 report that they have now determined that the average re-nesting is 3.7 years for leatherbacks at Playa Grande, Costa Rica. If this value is representative for Pacific populations in general, then the absolute estimates 7

8

for recent years being calculated based on a multiplier of 2.5 may be being underestimated by as much as 148% (3.7/2.5. While we do not argue that major declines have not occurred in the Eastern Pacific and in some Western Pacific nesting colonies, review of the literature (Ross 1982; Pritchard 1982; Spotila et al. 1996a,b; Pritchard 1996; Spotila et al. 2000 led us to question the magnitude of the declines that are implied to have occurred in the Western Pacific population between the early 1980 s and the present. The results of our review are summarized below. The Western Pacific population is of particular interest since the leatherbacks taken by the CA/OR DGF are most likely from this population (see NMFS 2000, 2001; Gallaway 2001. Spotila et al. (1996b estimated that, in 1996, the Western Pacific population consisted of about 1,838 nesting female leatherbacks and that the Indian population consisted of only 445 nesting females. The Irian Jaya, Bird s Head Peninsula, Papua New Guinea population was the largest of the three colonies included in the Western Pacific estimate with an estimated size of 1,625 nesting females (650 nesters each year assuming a re-nesting interval of 2.5 years. Assuming leatherbacks can withstand 1% annual mortality inflicted by humans, they declared that anthropogenic mortality from all causes should not exceed 18 adult females from this population per year. However, updated data (NMFS 2000 show that some 1,163 to 1,533 leatherbacks nested at Irian Jaya in 1996 suggesting a population size of about 2,966 females (2,908 to 3,833. Pritchard (1982, relying upon Salm (1981, estimated that, in the early 1980 s, the size of this population was about 4,000 animals. There were 7,311 nests at this locality in 1981 suggesting some 1,236 to 1,630 nesting animals (NMFS 2000, see also Gallaway 2001 in Attachment 1. These levels equate to a population size of about 3,583 (3,090 to 4,075 adult female leatherbacks. Given these estimates (3,583 in the early 1980 s and 2,966 in 1996, the Irian Jaya population reflected an estimated 17% decline between the early 1980 s and the mid-1990 s. The second colony reported by Spotila et al. (1996b from the Western Pacific was the Papua New Guinea, North Coast colony which was estimated to presently consist of 188 (125 to 250 adult females. Pritchard (1982 did not estimate the population size at this locality but had added a factor of perhaps 3,000 for this Melanesian region. Nesting in this region was characterized as dilute but extremely widespread along most of the northern coast of New Guinea, as well as at certain localities in the Bismarck Islands, Solomons, New Hebrides, and as far east as Fiji. I do not know to what extent these estimates are comparable to the Spotila (1996b estimates, but I suspect they are not based upon the differences in geographic regions that were addressed. Ross (1982 provides data that suggest some 250 adult females comprised the nesting colony at Papua New Guinea at the time of his paper. If this is true, this population reflects as much as a 25% reduction between the early 1980 s and the mid 1990 s. However, since the upper end of Spotila s 95% CI is also 250 adult females, this population may in fact be stable. 9

The third estimate included in the recent Western Pacific population estimate was the Malaysia, Terengganu colony estimated to consist of 25 nesting females (Spotila et al. 1996b. This population has a well-documented decline. Adult females in this population were estimated at over 3,000 individuals in 1968 declining to 100 by 1984. Since 1984, the adult female population has decreased from ~100 turtles to about 25 adult females. From the above, the three Western Pacific populations included in Spotila et al. s (1996b consisted of 1,838 adult females; 1,625 at Irian Jaya; 188 (125 to 250 at Papua New Guinea, and 25 at Terengganu. However, the Spotila et al. (1996b estimate for the 1996 population size at Irian Jaya appears to be in error. The NMFS (2000 updated data suggest a population of ~2,966 adult female leatherbacks at Irian Jaya in 1996. Thus, for the Western Pacific region, the adult female leatherback population in 1996, for the three colonies included in the estimate, was on the order of 3,179 turtles, not 1,838 turtles. In the early 1980 s the estimated population size for these three colonies in the Western Pacific region was 4,033; 3,583 for Irian Jaya (NMFS 2000, 250 for Papua New Guinea (Ross 1982, and 200 at Terengganua (Spotila et al. 1996b. Thus, the reduction in the Western Pacific population of adult leatherback females in 1996 as compared to the early 1980 s, based upon data restricted to the same nesting assemblages, was on the order of 21%, not 82%. Further, the Western Pacific population in 1996 was as strong as the Eastern Pacific population. Assuming leatherbacks can withstand 1% annual mortality inflicted by humans, total anthropogenic mortality for the Western Pacific population should not exceed 32 adult female leatherbacks. Not everyone agreed with Spotila et al. (1996b that leatherbacks were faced with extinction. Pritchard (1996 noted that he had taken the same global information on leatherback numbers and trends as Spotila et al. (1996b and had reached different conclusions. He examined populations for which adequate population trend data were available, looking for common threads or stresses in those cases for which serious population decline had occurred. He determined that there was no evidence of overall decline in leatherbacks in the Atlantic, and noted that on many nesting areas there had been significant increases in recent years. In the Indian Ocean, he observed that Hughes (1996 had documented steady increases of nesting animals in Tongaland for several decades. Moreover, he observed that these animals represented only a component of an overall population having a much larger geographic distribution, but that there were no recent data for these populations. Pritchard (1996 stated that elsewhere in the Indian Ocean, leatherback colonies were few and small, but there was no evidence that really large colonies had ever occurred in this region. He suggested that the ubiquity of coral reefs along both mainland and insular shores was likely the limiting factor for leatherback nesting in this region. Pritchard (1996 agreed that there was a problem in the Pacific, especially the Eastern Pacific. He considered these nesting declines to be real, even though he believed his 1982 data were possibly being used as a baseline for subsequent estimates to a greater degree than the quality of the data would justify. He noted that the same surveys used to generate the population estimates showed that adult females were being killed on many of the beaches and 10

that subsequent ground-truthing indicated that egg collection was also rampant. He stated that these two factors alone would have decimated the population, and that the numerous atsea factors documented by Spotila et al. (1996b only served to make the situation worse, i.e., the decline steeper. As he noted, the well-documented examples of serious leatherback nesting declines (e.g., Terengganu, Malaysia; Playa Grande, Costa Rica were cases where almost all of the eggs laid by the entire nesting colony had been harvested for many years. He concluded there is no turtle indeed no organism that can tolerate such interception of its reproductive effort. Pritchard (1996 observed that many of the leatherback populations that had declined are now protected. This is the case for the large Western Pacific population at Irian Jaya and other locales in this region, as well as for many Eastern Pacific populations, especially in Mexico and Central America. Pritchard (1996 also raised the issue of not considering long-term, natural cycles of considerable amplitude which are characteristic of leatherback populations. He noted that it was easy to conceive of natural factors that could serve to reduce the recruitment success of very large or dense nesting populations. The described factors included destruction of nests by other females nesting on the same spot, buildup of resident hatchling predator concentrations on both the beaches and in nearshore habitats, food competition, and so on. The apparent worldwide rarity of nesting leatherbacks only a few decades ago, he concluded, may in part have reflected simply that most colonies remained undiscovered, but it might also in part have reflected reality. There may have been fewer leatherbacks then. For example, the oldest residents at the major Playa Grande, Costa Rica leatherback nesting beach reported that this beach used to be a ridley arribada beach until the ridleys were displaced by the insurgent leatherback colony. Pritchard (1996 chided Spotila et al. (1996b for their use of the term extinction. He stated that this term is very absolute and should not be used casually. He did not consider extirpation of leatherbacks throughout extensive parts of their global range to constitute extinction. He noted that projections of any global species to extinction based upon a few years of trend data needed to take into account the concept of the demostat (Hardin 1993. This concept refers to the tendency of animal populations to expand until they reach an asymptotic level, to which they will tend to return after reduction following stress or temporary increase. Without a demostat, virtually all animal populations would either rise to infinity or drop to zero. In reality, most populations of most species most of the time will oscillate around a modal value, with deviation from this value ultimately corrected as limiting factors come into play. When entirely new conditions permanently change the rules of the game, the demostat will be re-set in a new position perhaps a much lower one, if new sources of mortality are introduced. But it will not normally hit zero, except when reproduction has, for some reason, been totally thwarted, or in the case of species whose habitats have been destroyed, or oceanic island species whose demographics are so sensitive and absolute population levels so low as to tolerate no disturbance at all. 11

Eckert and Sarti (1997 found that, in addition to the nesting population decline observed in Costa Rica, the formerly large Mexican nesting assemblage in the Eastern Pacific was in serious decline. The documented decline at one of the index beaches (Playa Mexiquillo was on the order of about 23% per year between 1984 and 1995. A comprehensive survey in 1996 showed that the decline was not attributable to emigration from this particular locality within the overall range, but rather that a true population decline was in progress. Fewer than 1,000 females nested in all of Pacific Mexico in 1996. In 1982, Pritchard (1982 had estimated on the order of 70,000 leatherbacks nested on Pacific beaches of Mexico in the early 1980 s. Satellite tracking conducted by Eckert and his colleagues showed that Mexican-nesting leatherbacks navigate to South American waters after egg laying is complete. Other tracking and tagging data support this contention, and demonstrate that Costa Rican leatherbacks also move south after nesting (e.g., Chandler 1991, Marquez and Vellanueva 1993 and Morreale et al. 1996. The region into which the leatherbacks move coincides spatially with an area utilized by the Chilean gillnet and longline swordfish fishery. This fishery is the largest in South America. The Chilean fishery was characterized by a large expansion after 1980. The gillnet fleet alone expanded from 4,777 days-at-sea in 1987 to 40,692 days-at-sea in 1993 (Weidner and Serrano 1997. Eckert and Sarti (1997 estimate that this fishery took on the order of 800 leatherbacks in 1990, and likely more than twice that number (1,600+ between 1990 and 1996. Using data from Frazier and Montero (1990, they estimated that 80% of the gill-netted turtles die. Considering the Peruvian fisheries as well, Eckert and Sarti (1997 deduced that a minimum of 2,000 leatherbacks were killed annually by the combined swordfish operations of Peru and Chile. This, they believed, was a major source of mortality accounting for the decline of the Eastern Pacific leatherback population. The increasing frequency and severity of El Niño events may constitute a change in habitat affecting greatly the Eastern Pacific leatherback population. These animals utilize the highly productive upwelling areas in the Southern Pacific off South America as their primary foraging grounds. The productivity of these areas is greatly reduced when El Niño events occur. Trenberth and Hoar (1995 report that there has been a tendency for more frequent El Niño events and fewer La Niña events since the late 1970 s. They note that aspects of the most recent warming in the tropical Pacific from 1990 to 1995 (which are connected to but not synonymous with El Niño, were unprecedented in the climate record. In 1988-89, 1,367 leatherbacks nested on Playa Grande, the fourth largest leatherback nesting population in the world (Spotila et al. 2000. Four years later (the approximate re-nesting interval, 506 turtles nested there in 1994-95 and in 1998-99, there were only 117 (Spotila et al. 2000. Unprecedented habitat change since 1990 was coincident with the Eastern Pacific leatherback population decline. These changes may have also played a significant role in the decline of the Eastern Pacific leatherback population over the last decade. Spotila et al. (2000 report that the Eastern Pacific population has continued to decline since the 1996 paper. They note that the Eastern Pacific population in 1999 contained only 1,690 adult females down from 4,638 in 1995. The total adult and subadult female population 12

was estimated at 2,955 individuals. They estimated that the Eastern Pacific population could tolerate an annual loss of 17 adult and 13 subadult females. The South American gillnet and longline fisheries appear to be one of the major factors accounting for the decline of the Eastern Pacific leatherback population (Eckert and Sarti 1997. In addition, quality of the foraging grounds has likely declined significantly in association with increased frequency and severity of El Niño events over the past decade. The level to which this population can rebound under the present climatic regime is therefore uncertain. However, the same climatic events affecting the turtles on the foraging grounds also detrimentally affect the fisheries, causing them to decrease. Further, the Eastern Pacific populations are protected strongly at the nesting beaches. We think it unlikely, therefore, that this population will be extirpated. However, it is also unlikely that the population will grow to former levels in the near to intermediate feature. We also think this example provides a rationale for why the Western Pacific population has not dramatically declined and may be increasing. The largest source of fishing mortality on this population was the North Pacific high seas driftnet fishery targeting squid and tuna (Eckert and Sarti 1997. This fishery operated in the area primarily utilized by leatherbacks from the Western Pacific population. This fishery peaked during the 1980 s and early 1990 s. In 1990-1991, this fishery took on the order of 1,000 leatherbacks of which 10 to 100% were believed killed (Eckert and Sarti 1997. This fishery declined after 1991, and was terminated in 1993 by a United Nations decree. Nesting trends at the Irian Jaya beaches show the lowest level in 1985 (3,000 nests during the peak of the fishery. When the record resumed in 1993, the number of nests had already increased to about 4,000. The trend then reflected a small but steady increase for the period 1993-1995, and a step increase of 2,235 more nests in 1996 than in 1995 (see Figure 3 in Gallaway 2001, Attachment 1. This population appears to be increasing following closure of the North Pacific high seas drift net fishery. In summary, we believe the data support the contention that, relative to the 1980 s, Atlantic populations of leatherbacks are at least stable and are presently at a high level of abundance; Indian Ocean populations are small and scattered and with some exceptions have probably always been so over most of this area; Western Pacific populations may have declined but if so only moderately; and the Eastern Pacific population has dramatically declined. A major shift in environmental quality on the foraging grounds utilized by the Eastern Pacific population has coincided with this most recent decline, but mortality from the swordfish fisheries off Chile and Peru has been a major factor accounting for the decline. The Western Pacific leatherback population has appeared to increase following closure of the North Pacific high seas drift net fishery. We conclude that Pacific leatherbacks are not in imminent danger of extinction. However, the combined effects of fishing mortality and habitat change may hold the Eastern Pacific population at a low population level for the foreseeable future. 13

DISTRIBUTION, MOVEMENTS, AND MIGRATION The leatherback sea turtle has the largest geographic range of any sea turtle, indeed even any reptile (Pritchard 1976, Pritchard and Trebbau 1984, Eckert 1999. They have been reported from 60ºN to 42ºS latitude in the Pacific Ocean (Stinson 1984 and in all major oceans (Groombridge 1982. Unlike other sea turtles, leatherbacks can tolerate extreme temperature variations. They routinely migrate to tropical waters for nesting, after which they travel vast distances to highly-productive, cold-water regions for foraging. They have been observed to be capable of swimming vigorously in waters of 0ºC (Goff and Lien 1988. Frair et al. (1972 reported leatherbacks to maintain core temperatures 15ºC above ambient water temperature of 5ºC. The factors enabling this ability have variously been ascribed to cylindrical body form, large body mass, a thick layer of epidermal fat, the presence of counter current heat exchanges in the flippers, the relatively low freezing for lipids, dark body color and, potentially, heat-generating adipose tissue (Starbird et al. 1993 and references contained therein. The ability of leatherbacks to maintain elevated internal temperature relative to the environment has been defined as gigantothermy (Spotila et al. 1991, Spotila and Standora 1985, Palidino et al. 1990. Nesting and Homing The distribution of nesting is restricted to between 30ºN and 20ºS latitudes (Starbird et al. 1993, see also Figure 4. Dutton et al. (1999 noted that while mixing of leatherback stocks occurred in foraging areas (e.g., near Hawaii, NMFS 2000, NMFS 2001, the segregation of maternal lineages between Eastern and Western Pacific subpopulations of leatherbacks suggests that natal homing occurs. However, they also noted that the evidence is not clearcut, since several subpopulations (including ones in different ocean basins; e.g., the Atlantic and Indian Oceans were genetically indistinguishable. The results of the Dutton et al. (1999 study estimated that natal homing was much stronger for island than for mainland nesting assemblages. For example, the Mexico and Costa Rican populations were genetically indistinguishable despite being 1,500 km apart. In contrast, there has never been a single observation of exchange between the Guianas and the island of Trinidad, only 400 to 500 km apart. The authors suggest that imprinting cues necessary for natal homing are not sufficiently strong to resolve mainland beaches (particularly since the mainland coasts used are so dynamic, whereas the islands used for nesting represent more distinctive (and more discrete nesting habitat to a leatherback. Dutton et al. (1999 conclude that natal homing instincts may not be as rigid in leatherbacks as in the other sea turtles. They suggest that this bodes well for conservation of this species as it allows more flexibility to exploit new reproductive habitat and to recolonize beaches where populations have been extirpated. The authors suggest that the evolutionary history of leatherbacks has been one of extinction and recolonization and 14

perhaps their ability to colonize rapidly relative to other sea turtles is one reason why this species has persisted through previous population bottlenecks. Juvenile Distributions and Movements Eckert (1999a found that the distribution of juveniles was generally restricted to tropical waters warmer than 26ºC until they exceeded 100 cm in carapace length as measured over the curve. The data reflected seasonal migration to higher latitudes by juveniles when warmer waters are available. Eckert (1999a suggested that it was conceivable that young leatherbacks were not capable of gigantothermy due to their small mass and relatively large surface, and were thus restricted to warm water habitats. Under these conditions, one would expect a gradual increase in thermal range as mass increased. He noted, however, that this was not the case. The data showed a very sharp break with turtles larger than 100 cm suddenly found in waters as cold as 12ºC. His interpretation was that either the turtles reached some physiological threshold (implying that the ability to stay warm has a developmentally-induced component as well as a mass component, or there is some behavioral change which causes turtles to move to colder waters as they enter this size class. Juvenile leatherbacks are restricted to waters >26ºC (Eckert 1999a. Thus, they characteristically occur outside the temperature range (18 to 20ºC where most swordfish fisheries operate. How long it takes a juvenile leatherback to attain 100 cm in length becomes an important recovery issue. Large numbers of juveniles were produced in the Eastern Pacific in the mid-1980 s. The currently accepted growth models (Zug and Parham 1996 suggested these juveniles would have reached 100 cm within about four years, moved into the Chilean swordfish grounds and many would already have been killed. The growth model suggests that these turtles would reach observed size at maturity in as little as nine years, putting them back on the nesting grounds in the mid-1990 s. There was no population increase in the mid- 1990 s or even to date. Under this scenario, the recovery potential of Eastern Pacific leatherbacks has already been substantially diminished. However, the growth model assumed that lines of arrested growth on the bones used to age leatherbacks occur on an annual basis. Periods of arrested growth may not be annual, especially for patch feeders like leatherbacks. Arrested growth could actually occur when the turtles make long migrations between foraging areas, or during periods of low food abundance. Eckert (1999a also pointed out that the model relied heavily on a single data point to set the shape of the growth curve. This data point was a 42.7-cm long turtle that had one growth ring and was therefore assumed to be one-year old. There was no independent reason to believe this turtle was really one year old. The assumptions used in the growth model are potentially flawed, and it is probably not reliable. There may still be large numbers of juveniles in the developmental pipeline and, if so, these could still be available to help restore the depleted Eastern Pacific population. However, even if alive, few may survive if the South American fisheries have remained large (Eckert 1999a. Adult Females. 15

The movement of females leaving the nesting beach have been monitored using satellite telemetry (Eckert 1999b. In 1995, three turtles from Trinidad (Atlantic were equipped with transmitters (Figure 5. Two of these were tracked for one year; the other transmitter failed prematurely when it was damaged by a fisherman who caught the turtle. Both the remaining turtles traveled north to about 45ºN latitude after nesting. One traveled up the western side of the Atlantic whereas the other first traveled east across the Atlantic between the 15º and 20º latitudes before traveling northward. Both remained in these northern areas until November, then they moved south to an area off the African coast where seasonal upwelling creates favorable foraging opportunities (see Figure 5. These turtles demonstrated a clear preference for waters between 16-18ºC. In 1997, Eckert (1999b deployed 10 transmitters on leatherback females nesting at Mexiquillo Beach, Mexico (9 in January, 1 in November. One of the turtles was killed by a poacher; the transmitter failed on a second turtle. All the remaining turtles renested at least once after deployment. Despite leaving the nesting grounds on different days, all traveled almost identical pathways from the nesting beach to the south (Figure 6. They were diverted west when they encountered the North Equatorial Current at about 8ºN, probably due to the rapid westward movement of this current. At about 2ºN, all but one of the turtles routed themselves back to the east, until reaching about 5ºS, One of these continued southeast all the way to the coastal waters off Peru and then onto Chile. The others turned towards the northwest and then appeared to begin a meander pattern. The turtle which did not turn back east after passing the North Equatorial Current moved westward across the South Pacific, ultimately reaching about 30ºS. This turtle had the longest record, about 18 months. The destination for these Eastern Pacific leatherbacks appeared to be the highly productive and cool waters off Peru and Chile (see Figure 6, much as the Trinidad turtles moved to upwelling areas off Western Africa. However, in 1997, an El Niño event was in progress there was no upwelling and subsequently primary production was poor (Eckert 1999b. Upon encountering this anomaly, the turtles appeared to become confused and wander aimlessly. Ultimately, the transmitters on turtles in this region all failed, whereas 16

17

18

the transmitter on the turtle which had traversed to the other side of the Pacific continued to operate. The turtles off South America may have died or been killed and the transmitter simply sank to the bottom. Eckert (199b provides the following discussion of these turtles: That the turtles died is worthy of further consideration. The longest tracking record in this study was the only turtle who did not move toward S. America. Instead this turtle moved toward the western Pacific, and its transmitter functioned for 18 months. The following scenario must be considered. Leatherbacks leaving the nesting colonies of Mexico and Costa Rica may be resource-stressed by a long reproductive season with limited food and the high energetic requirements brought about by the demands of reproduction or elevated water temperatures. When they leave, their greatest need is to replenish energy stores (e.g., fat, and they must move to areas where food is concentrated (e.g., upwelling regions of the S. Pacific. However, if food is not present in these areas, as is the case during an El Niño, the turtles may have few options and will starve. Such a strategy seems quite risky for the turtles, but it may be that the rarity of such strong oceanic perturbations mean this strategy is usually successful. Given the reported increase in El Niño events over the last 25 years, their increased strength, and the turtle s apparent inability to modify their migratory behavior, these climatic anomalies may be contributing to the population decline exhibited by this species in the Pacific Ocean. Testing this hypothesis will require substantial follow-on tracking work to determine if the patterns defined here also occur in non-el Niño years. The ability to monitor individuals in a population is valuable when trying to determine threats that a population might encounter when it can no longer be observed. This study has demonstrated the usefulness of such an approach to understanding the population decline in Mexico. As we noted in Eckert and Sarti (1997 it is clear that the Pacific Ocean waters off S. America are critical to the survival of the leatherback nesting populations of Mexico and Central America. It is also clear that high seas fisheries in the Pacific have not only jeopardized the leatherback sea turtle, but have also squandered many years of leatherback conservation at the nesting beaches. While there has been mortality of nesting turtles and their nests due to poaching on the nesting beaches the proportion of turtles killed pales in comparison to the incidental take by fisheries. Further the rate of population decline is too fast for overharvest of eggs (as in Malaysia, where almost 100% egg take took over 50 years to cause a population crash to be a primary reason for the population crash in Mexico. As we proposed in Eckert and Sarti (1997, blaming this population decline on the killing of nesting females or egg harvest is naïve now that we have demonstrated that leatherbacks are subject to such high mortality on the open ocean at such a critical period of 19

their lives. We are left to conclude that the combined impacts of the driftnet fisheries; the increasing frequency of El Niños, the continuing egg harvest; the kills of nesting females and the killing of leatherbacks off S. America have contributed to the destruction of the largest population of leatherbacks in the world. I found no satellite tracking data for post-nesting movements of Western Pacific leatherback turtle populations. However, transmitters have been deployed in late summer on leatherbacks in Monterey Bay in 2000 (NMFS 2001 and again in 2001. The two turtles tracked in 2000 moved south-southwest across the Pacific Ocean and are presently in the Mariana Trench area (Scott Eckert, pers. comm.. The turtles tracked in summer and fall of 2001 moved northward along the California coast to the Farallon Island area where they are apparently foraging. These data suggest that leatherbacks offshore California are largely representatives of the Western Pacific population. These data support the genetics data which show that all leatherbacks that have stranded or have been taken in the fisheries off California and Oregon have been representatives of the Western Pacific nesting assemblages (NMFS 2000, NMFS 2001, each citing P. Dutton et al., in press and P. Dutton, NMFS, pers. comm.. Local (California/Oregon Movements NMFS (2000, 2001 imply that leatherbacks from the Eastern Pacific population move northward along the Mexican coast into California with their appearance corresponding with the summer arrival of the 18º to 20ºC isotherms. These turtles are suggested to move farther northward with a warming environment into and beyond northern California, being documented to occur as far north as southern Alaska. They interpret the August peak in sightings of leatherbacks in the coastal waters of California to possibly imply a southward return movement of adults to the Mexican nesting beaches where breeding occurs in winter. This movement scenario is not supported by much if any of the available data. All the genetics data to date (as acknowledged by NMFS 2000, 2001 suggest California leatherbacks taken in the fishery or that have stranded on California beaches are of Western not Eastern Pacific origin. Further, Stinson (1984 has provided evidence that most of the leatherbacks in northern California and Oregon enter the coastal zone in July from offshore in association with the 13º to 15ºC isotherms. There is a separate but disjunct group of leatherbacks (probably of Mexican origin that is abundant south of Point Conception in July and they continue to be seen in this coastal area as late as September (Stinson 1984. The arrival of these turtles is associated with the arrival and occurrence of the 18º to 20ºC isotherms, but these turtles and the associated warm temperatures are not characteristic of the coastal regions in question for the entire period between October of one year and July of the next year. These turtles presumably flux into the southern California bight area from Mexico in May and June, are abundant from July to September, then retreat southward back into Mexico by October. There have been no observed takes of leatherbacks in CA/OR DGF south of Point Conception in the interim between July and September, or even in October (see Figures 1-3. The fishery does not operate in these coastal waters until 15 August, and 20

peak effort occurs in October and November after these Eastern Pacific leatherbacks have presumably moved southward. The Eastern Pacific leatherback population is at minimal, if any risk, from the CA/OR DGF. North of Point Conception, leatherback sightings in coastal waters are restricted to July- September (Stinson 1984. An explosion of sightings in the coastal zone occurs in July along Oregon in conjunction with the arrival of the 13º to 15ºC isotherms. From that point, Stinson (1984 observed that some leatherbacks moved south into the coastal waters of northern and central California during late July, August, and September. Others moved north from Oregon as part of the Alaska Gyral current appearing later in the season in waters further north. Stinson (1984 demonstrated that these inferred seasonal movements corresponded with the seasonal distribution of the 13º to 15ºC isotherms (Figure 7. In June, 13º to 15ºC waters begin to spread laterally along the coast (top left panel of Figure 7 as opposed to being configured as a narrow band intersecting the coast (e.g., bottom right panel of Figure 7, November. By July, these waters have spread into Canadian and Alaska regions, as well as southward to at least San Francisco Bay. In July, these thermal contours are compressed tightly against the coast. Wherever, in July, the 13-15ºC waters meet the northern coast there is a corresponding increase in leatherback sightings (Stinson 1984. She further observed that these isotherms reached their most northerly position in August, and that leatherbacks appeared from central California to the Gulf of Alaska wherever temperatures were 13-15ºC. This pattern continued through September. In October, the 13-15ºC isotherm withdraws from the north and few turtles are seen there. Stinson (1984 believed that the leatherbacks entering the coastal zone of central and north California in July were turtles that had moved from offshore habitats to the coastal zones of Oregon and northern California. Following arrival, they then spread laterally to the north and south within the coastal zone. She discounted that these turtles were migrants from southern California. She observed 1 even if all the leatherbacks that began to arrive in southern California region in spring continued their coastal trek north into Oregon, their numbers could not account for the July explosion of sightings in Oregon; and 2 they were not found in the intervening area between southern California and Oregon during May or June as the 13-15ºC isotherms were passing northward along central and northern California. Instead, leatherbacks did not occur until the warmer offshore isotherms have expanded shoreward, pushing the 13-15ºC isotherm tight against the coast of North America and this first happens in July along Oregon. 21

22

The movement patterns inferred by Stinson (1984 are supported by the temporal and latitudinal distribution of leatherback catch rates as documented by the CA/OR DGF observer program. First, leatherback takes have not been documented in July or August. During this period the fishery operates more than 75 nm offshore through 15 August, and leatherbacks would be inferred to be foraging in nearshore coastal waters and widely dispersed from Point Conception to southern Alaska. After 15 August, the fishery is able to fish within 75 nm but no takes have been observed in August. Historically, takes begin in September, and the four that have been taken in the fishery have occurred offshore in the region between San Francisco Bay and the CA/OR border (see Figure 1. September is near the end of the period (July-September when leatherbacks are abundant in the coastal zone (Stinson 1984. Some 61% of the historical takes in the CA/OR DGF have occurred in October. They have largely occurred in offshore waters and the range of the takes extends from Oregon to Point Conception (see Figure 2. This sharp increase in takes may reflect completion of the exodus of leatherbacks from inshore to offshore habitats (as well as from northern to more southerly regions that occurs in association with changes in the thermal environment (see Figure 7. Barlow (pers. comm. 2001, Appendix 3 examined latitudinal differences in leatherback entanglement rates in the CA/OR DGF as determined from observer data. With some slight differences, the results of our analyses of the same data (Figure 8 are in agreement. The area between 36.5 and 38.0 N exhibits the highest entanglement rate (0.016 leatherbacks per set; Figure 8. The entanglement rates for the two geographic regions north of this mid- Californian region are about half this magnitude (rate = 0.007 to 0.008/set, whereas the two cells to the immediate south exhibit entanglement rates that are about half again lower (0.003 to 0.004 leatherbacks/set; see Figure 8. Entanglement rates south of 33.5 N are an order of magnitude lower than observed in adjacent cells to the north; i.e., about 0.0004 leatherbacks/set versus 0.003 to 0.004 per set. The higher CPUE s in the northerly regions as compared to southern regions may reflect that a larger fraction of leatherbacks initially move to the highly-productive northerly regions offshore Canada and southern Alaska as compared to the fraction that moves to the south off southern California. The peak CPUE in the central California region off San Francisco Bay may be the result of a return migration for leatherbacks that summered in northern regions, supplemented by turtles that summered to the south moving north. The Farallon Islands occur in this central California region and this area is known to be a high-value forage area for leatherbacks in late fall (Scott Eckert, pers. comm., 2001. 23