Incidental capture, direct mortality and delayed mortality of sea turtles in Australia's Northern Prawn Fishery

Transcription

1 Marine Biology (1996) 125: Springer-Verlag 1996 I. R. Poiner 9 A. N. M. Harris Incidental capture, direct mortality and delayed mortality of sea turtles in Australia's Northern Prawn Fishery Received: 13 September 1995 / Accepted: 15 December 1995 Abstract The species composition, catch and mortality rates of sea turtles captured incidentally by the tiger prawn fishery on Australia's northern coast in 1989 and 1990 were estimated by monitoring the fishery's catch. In 1990, the delayed rate of mortality from damage was estimated and the size composition was measured. Five species of turtles were captured: the flatback (Natator depressa, 59% of the total), loggerhead (Caretta caretta, 10%), olive ridley (Lepidochelys olivacea, 12%), green turtle (Chelonia mydas, 8%) and hawksbill (Eretmochelys imbricata, 5%). The turtle catches varied with water depth: the highest catch rates (0.068 _ turtles per trawl) were from trawls in water between 20 and 30 m deep, relatively few turtles (10%) were captured in water deeper than 40 m (25% of trawls). Catch rates varied with time of year: the highest catch rates were ( 0.013) turtles per trawl in winter. There was no significant difference in the overall catch rate (X 2 = 0.047; p = ; df= 1) but a significant difference in mortality rate 0~ 2 = 3.99; p < 0.05; df-- 1) between the two years. The incidence of capture in the commercial fishery was (_ ) turtles per trawl towed for about 180 min, with ( ) turtles per trawl drowning in the nets. There were no significant differences in the catch and mortality rates between the two years for any of the turtle species except the loggerhead, which had a significantly (X 2 = ; p ; df= 1) lower catch rate in 1990 ( turtles per trawl) than in 1989 ( turtles per trawl), and a significantly higher mortality in 1990 (33%) than in 1989 (19%). Catch rates and mortality varied between the species: the flatback had the highest catch rate (0.030 _ turtles per trawl) but the lowest mortality (10.9%); the loggerhead had a catch rate of turtles per Communicated by G. F. Humphrey, Sydney I. R. Poiner (12~) ' A. N. M. Harris CSIRO Division of Fisheries, Marine Laboratories, PO Box 120, Cleveland, Queensland 4163, Australia trawl, and high mortality (21.9%); the olive ridley had a catch rate of turtles per trawl and a low mortality (12.5%); the green turtle's catch rate was per trawl and mortality 12.0%; the hawksbill had the lowest catch rate ( turtles per trawl) but highest mortality (26.4%). Based on the fishing effort ( d for 1989 and d for 1990), we estimate that ( _+ 424) turtles were caught and returned to the sea in 1989 and ( 404) in 1990, of which drowned in 1989 and 943 +_ 187 in In 1990, an estimated 25% of all captured turtles suffered some non-lethal damage; an estimated 21% of turtles were captured comatose and 4% were injured. We conclude that, considering other threats, trawl-induced drowning is not the major impact on turtle populations in northern Australia, but that measures to reduce drowning and delayed mortality would be desirable. Introduction Otter trawls are used by offshore vessels fishing for penaeid prawns, the most important fisheries resource in northern Australia. Dragged along the seabed at speeds close to four knots, these trawls do more than capture prawns: they disturb the substrate and capture animals as a by-catch. Trawling may adversely affect the ecology of the trawl grounds by changing the abundance and composition of by-catch species (Craik et al. 1990; Harris and Poiner 1991). In particular, the effect of prawn-trawling on sea turtles is alleged to be a major problem in the conservation of these species (Henwood and Stuntz 1987; Chan et al. 1988; Murphy and Hopkins-Murphy 1989). However, there are few reliable data on the capture and mortality of sea turtles by commercial prawn trawlers, except in waters of the USA (Carr 1977; Hillestad et al. 1977, 1982; Schoop and Ruckdeschel 1982; Henwood and Stuntz 1987; Magnuson 1990).

2 814 From several data sets collected between 1979 and 1988 in Australia's Northern Prawn Fishery, Poiner et al. (1990) estimated that an average of 5730 ( ) turtles were caught each year, of which an average of 344 ( 125) drowned; but that after the introduction in 1987 of management measures to reduce effort in the fishery, the number captured declined to about 4114 ( 1369) turtles in 1988, of which an estimated 247 ( _+ 90) turtles drowned. Turtle catches varied with the duration of the trawl and water depth and mortality varied with trawl duration. There is little information on the catch, mortality rates and the size composition of captured sea turtles at a species level in the Northern Prawn Fishery. Furthermore, there are few assessments of the injury to the turtles that may be caused by trawling. Poiner and Harris (1994) reported the first year's data of a two-year program monitoring the catch, mortality rates and size composition of turtles in the Northern Prawn Fishery. This paper presents the results of the two-year monitoring program. It aims to investigate the catch of sea turtles in 1989 and 1990, presenting information on the size composition and injury rates, Estimates are made of the sea turtles captured and killed in the fishery as well as the likely population sizes of turtles on the trawl ground. The fishery Australia's Northern Prawn Fishery extends from Cape York (142 ~ in the east to Cape Londonderry (127 ~ in the west, and includes the Gulf of Carpentaria, the Arnhem Land coast and Joseph Bonaparte Gulf(Fig. 1). Six species of sea turtles live in the prawntrawling areas of northern Australia: the flatback (Natator depressa), olive ridley (Lepidochelys olivacea), green turtle (Chelonia mydas), hawksbill (Eretmochelys imbricata), loggerhead (Caretta caretta), and leather- back (Dermochelys coriacea) (Poiner et al. 1990). The first four nest on beaches adjacent to trawling areas in the Northern Prawn Fishery (Cogger and Lindner 1969; Bustard 1972; Limpus et al. 1983; Limpus and Reed 1985). The Northern Prawn Fishery is divided into a banana prawn fishery and a tiger prawn fishery based on species caught, area fished, season, and mode of fishing. The trawls in the banana prawn fishery for a few weeks in April and May are mainly daytime, short (between 1 and 15 min), and targeted on dense schools of prawns detected by aerial surveillance or characteristic echosounder traces, so turtles are rarely caught. In contrast, the tiger prawn fishery operates at night, from April to November, with a peak in August-September. The main fishing grounds are further offshore. Unlike banana prawns, tiger prawns do not form dense schools, so more fishing time is spent trawling; each trawler typically makes about four trawls of about 180 min each per night. A logbook program has operated in the Northern Prawn Fishery since the early 1970s. The data include the catch in weight of species groups (banana prawns and tiger prawns) as boat-days specified on the basis of a six-nautical-mile grid. The total catch recorded in the logbooks can be checked against landings, since most of the product is exported and therefore subject to inspection. Thus the total effort (boat-days) expended in the fishery for any given year can be calculated from the logbook and landings data (Somers and Taylor 1981; Somers 1993). The tiger prawn fishery increased from an average of 9417 ( 1 688) boat days per year in to (_+ 1735) boat days per year in (Somers and Taylor 1981; Somers 1990b). In 1987, seasonal closures were extended to a total of 6 mo over two periods (December to mid-april and mid- June to August). Consequently the effort in the tiger prawn fishery actually declined to an average of _ 1761 boat days per year for Fig. 1 Map of Australia's northern coastline showing eastern and western boundaries of Northern Prawn Fishery 10 ~ 130 ~ 140 ~ pe York 10 ~ L ondonperry [ ~. 20' 20 ~ 130 ~ 140~

3 815 (Somers 1990b). Furthermore, trawling was banned from some nearshore areas to protect juvenile prawn nurseries (Australian Fisheries Service 1989). Materials and methods Seven volunteer fishermen in 1989 and 11 in 1990 monitored the operations of commercial trawlers fishing for tiger prawns in the Northern Prawn Fishery (Somers 1993). All the fishermen were trained in turtle identification and data collection. For a minimum of seven consecutive fishing days per month during the 1989 and 1990 tiger prawn seasons, the fishermen recorded the turtle catches and identified the species of turtle. In the 1989 season, the turtles were recorded as alive or dead. In the 1990 season, the curved carapace length of captured turtles was measured and the condition of the turtles was scored into one of four categories - healthy, injured, comatose or dead (Table 1). In both years the number of trawls, the hours fished and the location of trawls were recorded each day. The average headrope length of the trawls was 47.9 m, and the average duration of trawls was 156 min in 1989 and 186 min in 1990, with an average of 4.1 trawls per day fished in both years. Data analysis The incidence of a turtle in a trawl catch was a rare event of a total of trawls recorded, had no turtles, 285 caught one turtle, 19 caught two turtles and 1 caught three turtles. The catches of turtles from each trawl were assumed to be random variables and independent of each other. The distribution of these variables takes values of O, 1, 2 and 3, with probabilities ofpo, P~, P2 and P3 respectively, where: Po + P1 + P2 + P3 = 1. (1) From elementary probability theory, the mean catch rate (R) is: 3 R= ~ ip~ = PI + 2P2 + 3P> (2) i-0 and the variance (V) is: V = P1 + 4P2 + 9P3 - R 2. (3) Table 1 Definition of categories of turtle condition used during 1990 fishing season Category Healthy Injured Comatose Dead Definition Moving, flapping, aggressive (especially loggerheads); care needs to be taken to avoid personal injury from flippers or biting (especially loggerheads) Healthy but with wounds from trawling (e.g. head damage, broken flippers, cuts, retroverted anus) or sharks (e.g. flippers bitten off during net retrieval) Dazed, slow or few movements; can be lifted or shoved with minimum of caution; signs of recovery if left on board till next trawl shot No movement; head limp, extefided, and flops to the ground; no sign of breathing; eyes do not respond to touch; no recovery if left on deck, sometimes smelly The total numbers of turtles (C) captured by the fishery was estimated by: C = R-T (4) where R is the catch rate and T is the total fishing effort. The standard error of C (ac) is associated with both standard errors of R and T. Assuming that R and T are independent of each other, the standard error of C = RT can be written as: ~c = ~ + 4 r ~ + 4 ~. (5) The 95 % confidence interval of turtle effort was estimated to be 90 to 110% (Somers personal communication) so we have: 1.968r = 0.1. (6) T The 95% confidence limit of the catch rates was calculated as 1.96 ac. Although the catch of turtles in each trawl was not assumed to have a normal distribution, from the central limit theorem in probability theory, the estimated means can be assumed to have a normal distribution given such a large sample. Differences in the catch and mortality rates between the two years were tested by the )~2 test for differences in probabilities (Conover 1980). To compare the catch and mortality rates with the Gulf of Mexico and southern North Atlantic data of Henwood and Stuntz (1987), the effort (E) values were standardised to reflect hours towed with a single 30.5 m head-rope trawl net, using the following formula E = N'(H/30.5 m). (T/60), where N = number of trawls, H = headrope length (m), and T is the duration (min) of a trawl. The population of turtles on the trawl grounds was estimated for two depth zones (10 to 40 m, 41 to 90 m) from the density of turtles and the area of the zones. The density of turtles was calculated from turtle catch-rates and the area swept by a trawl of average headrope length (47.9 m) with a 60% opening and towed at 3.2 knots for an average time of rain (shallow zone = 10 to 40 m) or min (deeper zone = 41 to 90 m). The area of the depth zones was extracted from a data base of the depth at the centre of each 0.1 ~ x0.1 ~ rectangle in a logbook grid-referencing system of the Northern Prawn Fishery. Results Species composition A total of 165 turtles was captured in 1989 in a sample of trawls, and 161 in 1990 in a sample of trawls. Of these, 17 were reported drowned in 1989 and 29 in 1990 (Table 2). The flatback turtle (Natator depressa) was the main species caught (59%), but numbers of loggerheads (Caretta caretta, 10%) and olive ridleys (LepidocheIys olivacea, 12%) were also caught, and green turtles (Chelonia mydas, 8%) and hawksbills (Eretmochelys imbricata, 5%) were occasionally captured. No leatherback turtles (Dermochelys coriacea) were caught. Of the turtles caught, 7% were unidentified at the species level (Table 2). Distribution The western and southern parts of the Gulf of Carpentaria were the most heavily fished tiger prawn areas of

4 816 Table 2 Species composition, number of turtles caught, numbers and percent of total drowned Species Caught Drowned Caught Drowned No. (%) No. (%) No. (%) No. (%) Flatback (Natator depressa) 89 (54) Loggerhead (Caretta caretta) 26 (16) Olive ridley (Lepidochelys olivacea) 24 (15) Green turtle (Chelonia mydas) 14 (8) Hawksbill (Eretmochelys imbricata) 6 (4) Unidentified 6 (4) Total 165 (100) 7 (7.9) 103 (64) 14 (13.6) 5 (19.2) 6 (4) 2 (33.3) 2 (8.3) 16 (10) 3 (18.8) 2 (14.3) 11 (7) 1 (9.1) 1 (16.7) 9 (6) 3 (33.3) 0 (0.0) 16 (10) 6 (37.5) 17 (10.3) 161 (100) 29 (18.0) the Northern Prawn Fishery in 1989 and 1990 (Fig. 2a). The fishermen taking part in the monitoring study fished widely over the Northern Prawn Fishery, covering the whole fishery in extent, and their catches reflecting the intensity of effort (Fig. 2b). Their records showed that turtles were caught throughout northern Australia, but the numbers reflected mainly the intensity of fishing effort, so that most turtles were caught around Groote Eylandt (Fig. 3a). Spatial differences between catches of turtle species were less clear after allowing for fishing effort and chance occurrence. All five species of turtles were caught in the heavily fished central area (Figs. 3 and 4). Loggerhead turtles were not caught in Joseph Bonaparte Gulf (western Northern Prawn Fishery) (Fig. 4a) and green turtles were not caught in either Joseph Bonaparte Gulf or Albatross Bay (eastern Northern Prawn Fishery) (Fig. 4b). Catch rates No significant differences were found in the overall catch rates 0~2= 0.047; p = ; df= 1), but the proportion of dead turtles caught was significantly higher in 1990 (18%) than in 1989 (10.3%) (X 2 = 3.99; p < 0.05; df = 1) (Table 3). The incidence of capture in the commercial fishery was ( ) turtles per trawl. The loggerhead catch rate in 1990 ( _ turtles per trawl) was significantly (Z = ; p = ; df= 1) lower than in 1989 ( _ turtles per trawl; Table 3). No significant differences were found in the catch rates of the other species in 1989 and t990 (Table 3). Catch rates and the proportion caught dead differed with species: the flatback had the highest catch rate ( _ turtles per trawl) but the lowest proportion dead (10.9%); the olive ridley had the next highest catch rate ( _ ) but a higher proportion dead (12.5%); loggerheads had a lower catch rate ( _ ) but a high proportion dead (21.9%); the green turtle's catch rate was turtles per trawl and mortality 12.0%, and the hawksbill had the lowest catch rate ( turtles per trawl) but high mortality (26.4%) (Table 3). Turtle catch rates varied with water depth, with the highest rates recorded in shallow water (< 40 m; Fig. 5). The highest catch rates were ( ) turtles per trawl in the 10 to 20 m depth range and ( _ ) turtles per trawl in the 20 to 30 m depth range. There is very little trawling in water < 10 m deep in the Northern Prawn Fishery. Only 33 turtles (10%) were captured in depths > 40 m, despite 25% of trawls occurring at these depths. The catch rates of all species were greatest in water < 30 m deep. Turtle catch rates also varied with time of year, with the highest rates for all species recorded in mid-june and August (the fishery was closed from 15 June to 1 August and December to mid-april; Fig. 6). The highest catch rates were ( ) turtles per trawl in June and ( +_ ) turtles per trawl in August. The olive ridley also had a relatively high catch rate in April. Sizes The size of turtles captured in 1990 ranged from 25 to 105 cm curved carapace length (Table 4). All the captured olive ridleys were adult-sized. Juveniles, immature subadult and adult-sized turtles of the other species were caught, but most were of the larger size classes (Table 4). Total catch and total numbers dead in catch The prawn trawlers averaged 4.1 trawls per day in 1989 and Their average duration was 156 rain in 1989 and 186min in 1990, but in both years there were four trams per day of,,~180 min duration on 85% of fishing days. Therefore, the total number of trawls expended on tiger prawn fishing in the Northern Prawn Fishery

5 817 _ la I 1 I 1 I I uu I I I I 1 1 I t I 1 ~ 10 ~ _. 12 ~ - o ~ 14 ~ t ~ I I I I I I I 1 1 I 1 I m I I I I I I I I b 10 ~ ~ - ~ o ~Nl3f' g/ 12 ~ ~ o~ ol 14 ~ <33 ~,m,6,, 24 ~ 126 ~ 128 ~ 130 ~ 132 ~ 134 ~ 136 ~ 138 ~ 140 ~ 142~ Fig and 1990 (pooled)distribution over Australian Northern Prawn Fishery of number of totat boat-days fished for tiger prawns in prawn fishery (a), and trawl shots fished in the monitoring program (b) in 1989 and 1990 was estimated by multiplying the total effort (27049 boat-days for 1989 and for 1990; Australian Fisheries Service, Canberra) by four. The total number of turtles and the number of each species (except for the loggerhead) captured in the fishery in each year were estimated by multiplying the number of trawls by the pooled estimated turtle-catch rates. For loggerheads, since the yearly catch rates were significantly different, they were estimated for each year. The number of turtles drowned was estimated by multiplying the number of trawls by the estimated mortality rates for each year. The average number of turtles of each species found dead in the catch was estimated from the percentage of dead turtles recorded in the monitoring program. The estimated total number of turtles captured in the Northern Prawn Fishery in 1989 was (_+ 424) and in 1990, 5238 (+ 404). Of these, 567 (+ 140) turtles in 1989 and 943 ( _+ 187) in 1990 had drowned (Table 5). Flatbacks accounted for 43% of drownings in 1989 and 49% in 1990; loggerheads accounted for 20%

6 818 t }' I I I" I I 1 ~t l I ' t I 1 I I I I "-J~,,l~.~ J _ - eo lip- --~ at 10 ~ 12<> I1 a ~. 14 ~ 1 I I t I I I I I I I I I I I I [- I F I I I I I I -I I I I I I I 9 I I I " I "-L..,-..~.L b 10 ~ 12 ~ 14 ~ O 0~ / I I I I I I I [ I I I I '1 I 1 '1 I (3 10" s 12 ~ o ).~/l 14 ~ o 124 ~ I I I I I I I 1 I I I 1 I I 126 ~ 128 ~ 130 ~ 132 ~ 134 ~ 136 ~ 138 ~ 140 ~ 1 42~ Fig, 3 a Turtles of all species; b flatback, Nataror depressa; c olive ridley, Lepidochelys olivacea. Number of turtles caught in monitoring program and 1990 (years pooled) distribution over Australian Northern Prawn Fishery

7 819 t' a l l l l f----q---r-'---i--t I 1 I t I t.i,l-,j i10 o 12 o 0 14 o I L i I I t,,i i I I I I I I I 1 I I f ~'''''''''''''''' "~1o0 '1 I t I I I I 1 I I 1 I " I I I 1 I I " 1 I I I I I I I -I I I I I I I---P--I-- I "--~1_ t 13 9,. j 10 ~ 12 ~ 14 o I I.I.[ I 1 I I I I I I I I _1 124 ~ 126 ~ 128 ~ 130 ~ 132 ~ 134 ~ 136 ~ "138 ~ 140 ~ 142~ Fig. 4 a Loggerhead, Caretta caretta; b green turtle, Chelonia mydas; c hawksbill, Eretmochelys imbricata. Number of turtles caught in monitoring program and 1990 (years pooled) distribution over Australian Northern Prawn Fishery

8 820 Table 3 Estimated catch rates, standard errors and mortality of the five species of turtles captured in 1989 (3 313 trawls) and 1990 (3 092 trawls) monitoring program, and 1989 and 1990 data pooled. Catch rates are presented both as number per trawl (Turtles per trawl) and number per unit effort standardised to reflect hours towed with a 30.5 m-headrope prawn trawl-net (Turtles/std net-h). Catch rates Species Turtles per trawl (SE) of Gulf of Mexico (n = 50 turtles) and southern North Atlantic (n = 478 turtles) prawn fisheries are also listed for three of turtle species captured (loggerhead Caretta caretta, Kemp's ridley Lepidochelys kempi, and the green turtle Chelonia mydas) (from Henwood and Stnntz 1987) (CI confidence interval) Full specific names are given in Table 2 Turtles/std net-h (CI) Mortality (%) Northern Prawn Fishery 1989 Flatback Loggerhead Olive ridley Green turtle Hawksbill Unidentified All species 1990 Flatback Loggerhead Olive ridley Green turtle Hawksbill Unidentified All species 1989 and 1990 pooled Flatback Loggerhead Olive ridley Green turtle Hawksbill Unidentified All species Gulf of Mexico Loggerhead Kemps ridley Green turtle Total Southern North Atlantic Loggerhead Kemps ridley Green turtle Total (0.0029) (0.0014) (0.0016) (0.0008) (0.0015) (0.0007) (0.0011) (0.0005) (0.0007) (0.0004) (0.0008) (0.0004) (0.0040) (0.0020) (0.0033) (0.0013) (0.0008) (0.0003) 33, (0.0014) (0.0005) (0.0011) (0.0004) (0.0010) (0.0003) (0.0064) 0,0010 (0.0005) (0.0043) (0.0016) (0.0022) (0,0009) (0.0009) (0.0004) (0.0010) (0.0004) 12, (0.0008) (0.0003) (0.0006) (0.0003) (0.0008) (0.0003) (0.0029) (0.0012) (0.0008) (0.0004) (O.0002) (0.0008) (0.0039) (0.0008) (0.0003) (0.0041) r o.oe ~0.04 co,r- ~ 0.02 i 1= I-- t { lt0 20 ~ 30 ' 4'0 50 ' 6 0 Depth (m) Fig. 5 Turtle catch rate (all species combined) ( 4-1 SE) as a function of depth (m) { 0.12 ~ 0.10.~ 0.08 E 0.08 (9 I I I i 6 I I I I April May June July Aug Sept Oct Nov Fig. 6 Turtle catch rate (all species combined) ( +_ 1 SE) related to month. Fishery is closed from December to mid-april and mid-june to August each year {

9 821 Table 4 Length-frequency and numbers of juvenile, sub-adult and adult-sized turtles caught in 1990 for the five species of turtles captured in the fishery. Turtles were grouped into three ageclasses from the midpoints of their curved carapace lengths following Marquez (1990). Full specific names are given in Table 2 (Unid unidentified) Flatback Loggerhead Olive ridley Green turtle Hawksbill Unid Length-frequency (curved carapace-length midpoints) 25 cm cm cm cm cm cm cm cm cm Age classes juveniles sub-adults adults Table 5 Estimated number (SE) of turtles caught and drowned, in Northern Prawn Fishery in 1989 and Numbers estimated by multiplying annual number of trawls fished by estimated catch rates (separate for loggerheads, pooled for 1989 and 1990 for other species and all turtles). Numbers of all turtles drowned were estimated separately for each year. Note best estimate for all turtles is not a summation of the estimates for each species (see "Results - Total catch and total numbers dead in catch"). Full specific names are given in Table 2 Species Caught (SE) Drowned (SE) Caught (SE) Drowned (SE) Flatback (291) 355 (79) (277) 337 (76) Loggerhead 849 (184) 163 (74) 200 (82) 67 (47) Olive ridley 675 (115) 84 (45) 643 (109) 80 (43) Green turtle 422 (87) 51 (29) 402 (83) 48 (28) Hawksbill 253 (67) 68 (34) 241 (63) 64 (32) Unidentified 371 (91) 101 (42) 354 (87) 96 (40) All turtles (424) 567 (140) (404) 943 (187) Table 6 Number of turtles (n) and percentage scored as healthy, injured, comatose or drowned in 1990 when extracted from trawl net. Data are shown for each species and for all species pooled; full specific names are given in Table 2 Species (n) Healthy Injured Comatose Drowned (%) (%) (%) (%) Flatback (103) Loggerhead (6) Olive ridley (16) Green turtle (11) Hawksbill (9) Unidentified (16) Total (161) of drownings in 1989 and 9% in 1990; the other species each accounted for < 10% of drownings in both years (Table 5). Delayed mortality Overall, 21% of captured turtles were comatose when removed from the net and 4% were injured (Table 6). Most of these (comatose 22%, injured 5%) were flat- back turtles, as they dominated the catch in 1990 (Table 6). The percentage of comatose individuals in the other species ranged from 0 to 44%, and the number injured from 0 to 18%, although low sample sizes (n = 6 to 16) make these estimates unreliable (Table 6). Notes made by the Volunteer fishermen indicated that ~ 50% of the comatose turtles recovered after ~ 30 rain on the vessel deck, and many of the injuries appeared superficial. However, we have no data on their rate of survival after their return to the sea.

10 822 Discussion Species composition Of the five species of turtles captured, three comprised 81% of the catch: the flatback (Natator depressa, 59%); the loggerhead (Caretta caretta, 10%) and the olive ridley (Lepidochelys olivacea, 12%). The green turtle (Chelonia mydas, 8%) and the hawksbill (Eretmochelys imbricata, 5%) were occasionally captured. Of the turtles captured, 7% were not identified. The only noticeable change in composition between the two years was that the proportion of loggerheads in the catch declined from 16% in 1989 to 4% in Catch and mortality rates The estimated catch rate of turtles in the Northern Prawn Fishery was ( ) turtles per trawl. This rate was not significantly different between 1989 and 1990 or between Poiner et al.'s (1990) estimates of _ turtles per trawl, X2= 0.756, p = The catch rate was dependent on depth: the highest rates were in water between 10 and 40 m deep, while few turtles were captured below 40 m (Fig. 5). The rate also varied within the year: the highest catch rates were in the winter months of June and August (Fig. 6), shortly before nesting in the early summer months. The proportion of dead turtles differed significantly between the two years: 10% in 1989, 18% in The mortality in 1989 did not differ significantly from the estimates of Poiner et al. (1990). However, the mortality in 1990 was higher when injured and comatose categories were also recorded. Either the dead turtles were underscored in 1989, or some comatose individuals were considered dead in The shorter average duration of trawls in 1989 (156 rain) compared to 1990 (186 rain) may also possibly explain some of the difference in mortality. However, neither of the estimates includes any subsequent mortality due to damage during trawling or to capture stress. Survivorship of comatose turtles could alter mortality rates considerably. If 50% of the comatose and all injured survive, the mortality rate would be ( ) turtles per trawl or 25% of captured turtles. However, if we assume the 1990 comatose and injury rates representative for the fishery, and that all the comatose and injured turtles die after release, the estimated mortality rate for the fishery would increase to ( _ ) turtles per trawl, or 39% of captured turtles. The mortality rates of the flatback, olive ridley and green turtles were relatively low ( -~ 11%), but the rates for the loggerhead and the hawksbill turtles were high (> 20%). These differences may reflect the relative abilities of the five species to resist drowning. Comparisons with Gulf of Mexico and southern North Atlantic Adjusted to one hour's fishing with a 30.5 m net, the catch rate of turtles in the Australian Northern Prawn Fishery [mean=0.0113, 95% confidence interval (CI) = turtles] is higher than the rate reported for the Gulf of Mexico (mean=0.0031, 95% CI = turtles), by Henwood and Stuntz (1987) but lower than the rate they reported for the southern North Atlantic (mean =0.0487, 95% CI = turtles) (Table 3). Most prawn trawling in the southern North Atlantic fishery occurs in water depths <18 m and, as in the Northern Prawn Fishery, catch rates vary with water depth, with the highest catch rates in water,~14 m deep. In the Gulf of Mexico, prawn trawling occurs in depths up to 80 m (Rothschild and Brunenmeister 1984), but unlike the other two fisheries, the turtle catch rate appears to be fairly constant over all depths down to 30 m. The turtle mortality rates for the Gulf of Mexico and southern North Atlantic prawn fisheries were estimated as 29 and 21% of captures, respectively (Henwood and Stuntz 1987; present Table 3) which is higher than either of the estimates (10 and 18%, respectively) from the Northern Prawn Fishery. The difference may be due to different species having different mortality rates. The loggerhead (Caretta caretta) dominates the American catches: 94% of the southern North Atlantic and 86% of the Gulf of Mexico catches. The same species constitutes a small component of the northern Australian catch (10%), but its estimated mortality rate is similar to the American rates (22%). The loggerhead therefore appears to be particularly susceptible to drowning. In contrast, the dominant turtle in the Northern Prawn Fishery catch, the flatback (N. depressa = 59%), has a low mortality rate: 11%. This species is endemic to northern Australia and tends to be found inshore in relatively shallow ( < 40 m), muddy waters; it possibly has a higher resistance to drowning in trawls (11% mortality) than the other species. The difference in the overall mortality rates of turtles in the American and Australian fisheries may, therefore, be due to the susceptibility to drowning of the dominant species. Trawl capture, mortality and impact on turtle populations Our analyses suggest that between and sea turtles are captured each year in the Northern Prawn Fishery. About 14% drown in the trawl nets, and another 25% are injured or comatose when brought on board. The number of turtles captured and drowned is mainly dependent on the total effort expended in the fishery (Poiner et al. 1990). In 1989 the fishing effort was

11 boat-days, and an estimated (_+ 424) turtles were captured, of which 567 ( _ 140) drowned on capture and at most (all injured and comatose), a further could have died later. In 1990, the fishing effort was boat-days and an estimated ( + 404) turtles were caught, of which 943 ( 187) drowned on capture and at most (all injured and comatose) a further could have died later. The Northern Prawn Fishery is currently being restructured to remove boats from the fishery: the number of boats has declined from 200 in 1990 to ~ 128 boats in 1994 (Sachse and Robins 1995). Although it is difficult to predict the impact of the restructuring, fishing effort is unlikely to increase above values. Sea turtles are long-lived, reaching sexual maturity after many years (Limpus and Walters 1980; Carr 1986) for example possibly > 40 yr for Caretta caretta (Limpus 1985). Mature individuals have a long breeding life, but with gaps between breeding years (Bustard 1972). When breeding, they produce large numbers of eggs in several clutches a year. The sea turtles are strongly faithful to the few foraging areas and breeding sites, which are usually far apart (Carr 1986). Sexual determination is fixed by the temperature at which the egg is incubated (Mrosovsky et al. 1984), but little is known about the sex ratios of the populations, or the size and genetic relationships of breeding populations (Gyuris and Limpus 1988). Extensive tagging of hatchlings and other age classes has yielded little data on the natural mortality rate of turtles (Mrosovsky 1983; Carr 1986; Hirth 1978). Thus, there is a paucity of information about the population size, age structure and age-specific survivorship of sea turtle populations in general, let alone those of northern Australia. Threats to northern Australian sea turtle populations include the commercial harvesting of eggs and turtles for meat in Indonesia and elsewhere, degrada- tion of foraging areas, destruction of habitat and changes to breeding sites, ingestion of floating plastic debris, harvesting of turtles and eggs by traditional hunters, and incidental capture in the trawl, drift and set nets of commercial fisheries. Without information on the total population size, age structure, age-specific survivorship of the turtle populations and turtle distributions (patch dynamics), the size and age structure of the segment of the population caught by trawlers, and indirect mortality rates, it is difficult to estimate the impact of trawling on the population. In 1990 we measured the carapace length of the captured turtles. All the captured olive ridleys were adults. Most individuals of the other species were sub-adult and adult- sized, but some were juveniles, immature sub-adults and mature adults (Table 4). However, except for the flatback turtle (n = 103), the sample sizes (n = 6 to 16) for the other species were too low to be useful. When these data are available, stage-based population models similar to those developed for loggerhead turtles in the southeastern USA (Crouse et al. 1987) should generate more robust estimates of the impact of trawling on northern Australian turtle populations and advice on the appropriate management response. We can use the catch rates to estimate the turtle densities in the Northern Prawn Fishery. For all species, most trawl catches are made in the km 2 portion of the Northern Prawn Fishery between 10 and 40m water depth, with few catches in the km 2 deeper than 40 m. If we assume the trawl catch rates of turtles to be representative of the density of turtles in the Northern Prawn Fishery, there would be at least ftatbacks, 4700 loggerheads, olive ridleys green turtles and hawksbills, in the fishery (Table 7)~ Because turtles spend only a proportion of their time on the bottom and can swim fast enough to avoid the path of the trawl (Musick et al. Table 7 Estimated number of turtles ( _+ SE) in Northern Prawn Fishery calculated from 1989 and 1990 recorder programs. Total of 4874 trawls was recorded from 10 to 40 m water depth range and 1302 trawls from 41 to 90 m. Full specific names are given in Table 2 Depth (m) No. caught No./trawl No./100 km z Total No. Flatbaek (0.0027) (1331) (0.0031) (3458) Loggerhead (0.0012) (580) (0.0011) (1193) Olive ridley (0.0012) (597) (0.0019) (2063) Green turtle (0.0009) (434) (0.0019) (2063) Hawksbill (0.0007) (359) (0.0008) (844) Unidentified (0.0010) (508) (0.0008) (844)

12 824 Table 8 Estimated total number of turtles in Northern Prawn Fishery, annual trawl catch, and estimated annual mortality directly due to trawling. Full specific names are given in Table 2 Species Total no. No. No. Mortality caught drowned (%) Flatback Loggerhead Olive ridley Green turtle Hawksbill Unidentified I991), estimates made without correcting for turtle behaviour must underestimate the actual densities, and consequently the size, of the turtle populations on the trawl grounds. Mortality estimates are < 2% of the trawl- ground population size for all species except the loggerhead (2.39%) and hawksbill (2.99%) (Table 8). However, these values overestimate mortality in the northern Australian population. Shallow waters <10m (~-, km 2) are rarely trawled in the Northern Prawn Fishery (Somers 1993), and trawl catches cannot be used to estimate turtle density at these depths. Some species may have higher densities in shallow water (< 10 m) than in deeper water (>10 m). Green turtles for example, would be strongly associated with their seagrass feeding pastures, which in the Gulf of Carpentaria are usualiy intertidal and subtidal to ~ 5-8 m water depth (Poiner et al. 1987). The whole Northern Prawn Fishery is responsible for the direct or indirect mortality of a maximum of ~2 100 turtles of four species each year. This is much lower than other human-induced mortalities. For example, the indigenous harvest of the green turtle (Chelonia mydas) in the Torres Strait by Torres Strait Islanders and the Kiwai people of Papua New Guinea is of the order of 5200 to 6300 turtles per year (Johannes and MacFarlane 1991; Marsh and Saalfeld 1989); the annual catch of the Indonesia commercial fishery is estimated to be > adult green (C. mydas) and hawksbill (Eretmochelys imbricata) turtles per year (H.A. Reichart unpublished data). The data are inadequate, but we consider that trawler-induced mortality in the Northern Prawn Fishery has less impact upon turtle populations than other hazards in northern Australia. Somers (i994) simulated the impact upon a loggerhead turtle population of an annual mortality of 3%, similar to that of the Northern Prawn Fishery. The population declined at 0.4% per year and would take yr to extinction if no density-dependent feedback mechanism inherent in the theory of fishing were invoked. However, any measures that reduce drowning and delayed mortalities of turtles would be desirable. Options the fishery may have to reduce turtle mortality include restrictions on fishing around seasonal nesting times (already done in the Gulf of Carpentaria), public- ity on the importance of retaining comatose turtles until recovery, shorter trawl tows ( < 90 rain), and the development of alternative trawl gear (Somers 1990a). Acknowledgements J. Brosnan, S. Cook, K. Emery, L. Graham, N. Laird, A. Pawlowski, D. Smith, D. Boulton, D. Holland, G. Akerstrom, G. Gallop, I. Stapleton, J. Faulkner, K. Munro, L. McKenzie and S. Robson collected the data for the study. Mr I. Somers (CSIRO Marine Laboratories) helped interpret the fishery logbook data. Dr Y. Wang and Mr J. Kerr (CSIRO Institute of Plant Production and Processing Biometrics Unit) provided some statistical advice. Mr I. Somers and Dr S. Blaber made valuable comments on the manuscript. Mrs L. Maxwell assisted with editing. References Australian Fisheries Service (1989) "Northern Prawn Fishery 1989 Information Booklet". Australian Fisheries Service, Department of Primary Industries and Energy, Canberra, Australia Bustard R (1972) Sea turtles. Collins, London Craik W, Glaister J, Poiner I (eds) (1990) The effects of fishing. Aust J mar Freshwat Res 41:1-197 Carr A (1977) Crisis for the Atlantic ridley. Marine Turtle News 4:~3 Carr A (1986) New perspectives on the pelagic stage of sea turtle development. NOAA natn mar Fish Serv tech Memo US Dep Commerce SEPFC-190:1-36 Chan EH, Liew HC, Mazlan AG (1988) The incidental catch of sea turtles in fishing gear in Teringganu, Malaysia. Biol Conserv 43 : 1-7 Cogger HG, Lindner DA (1969) Marine turtles in northern Australia. Aust Zool 15: Conover WJ (1980) Practical nonparametric statistics. John Wiley & Sons, New York Crouse DT, Crowder LB, Caswelt H (1987) A stage based population model for loggerhead sea turtles and implications for conservation. Ecology 68:1412-I423 Gyuris E, Limpus CJ (1988) The loggerhead turtle, Caretta caretta, in Queensland: population breeding structure. Aust Wildl Res 15: Harris AN, Poiner IR (1991) Changes in species composition of demersal fish fauna of Southeast Gulf of Carpentaria, Australia, after 20 years of fishing. Mar Biol 111 : HenwoOd TA, Stuntz WE (1987) Analysis of sea turtles captures and mortalities during commercial shrimp trawling. Fish Bull US 85 : Hillestad HO, Richardson JI, McVea C, Watson JM (1982) Worldwide incidental capture of sea turtles. In: Bjorndal K (ed) Biology and conservation of sea turtles. Proceedings ofa World Conference on Sea Turtle Conservation. Smithsonian Institution Press, Washington, DC, pp Hillestad HO, Richardson JI, Williamson GK (1977) Incidental capture of sea turtles by shrimp trawlermen in Georgia. Report to the National Marine Fisheries Service by Southeastern Wildlife Services Inc, Athens, Georgia (No further publication details available) Hirth HF (I978) A model for the evolution of green turtle Chelonia mydas remigrations. Herpetologica 34: Johannes RE, MacFarlane JW (1991) Traditional fishing in the Tortes Strait Islands. Australian Goverment Publishing Service, Canberra Limpus C (1985) A study of the loggerhead sea turtle, Caretta earetta in eastern Australia. Doctoral dissertation. University of Queensland Limpus C, Parmenter CJ, Baker V, Fleay A (1983) The Crab Island sea turtle rookery in north eastern Gulf of Carpentaria. Aust Wildl Res 10:

13 825 Limpus C, Reed PC (1985) Green sea turtles stranded by cyclone Kathy on the south-western coast of the Gulf of Carpentaria. Aust Wildl Res 12: Limpus C J, Walters DG (1980) The growth of immature green turtles Chelonia rnydas under natural conditions. Herpetologica 36: Magnuson JJ (1990) Decline of sea turtles. Causes and prevention. National Academy Press, Washington, DC Marsh H, Saalfeld WK (1989) The distribution and abundance of sea turtles in the Torres Strait region. Report to Australian Fisheries Service, Canberra The Great Barrier Reef Marine Park Authority, Townsville and The Fisheries Management Branch of the Queensland Department of Primary Industries, Brisbane (March 1989, Internal publication) Marquez RM (1990) FAO species catalogue. Vol 11. Sea turtles of the world. An annotated and illustrated catalogue of sea turtle species known to date. FAO Fish Synopsis 125 (11): 1-81 Mrosovsky N (1983) Conserving sea turtles. British Herpetological Society, London Mrosovsky N, Hopkins-Murphy SR, Richardson JI (1984) Sex ratios of sea turtles: seasonal changes. Science, NY 225: Murphy TM, Hopkins-Murphy SR (1989) Sea turtle and shrimp fishing interactions: a summary and critique of relevant information. Centre for Marine Conservation, Washington, DC Musick JA, Barnard D, Keinarth JA (1991) Prediction of trawl fishery impacts on seaturtles: model. In: Salmon M, Wyneken J (eds) Proceedings of llth Annual Workshop on Seaturtle Biology and Conservation. National Oceanic and Atmospheric Administration, US Department of Commerce, Washington, pp (NOAA Tech Minut NMFS-SEFSC 302) Poiner IR, Buckworth RC, Harris A (1990) The incidental capture and mortality of sea turtles in Australia's Northern Prawn Fishery. Aust J mar Freshwat Res 41 : Poiner IR, Harris A (1994) The incidental capture and mortality of sea turtles in Aiastralia's Northern Prawn Fishery. Proceedings of Australian Marine Turtle Conservation Workshop, Seaworld Nara Resort, November Queensland Department of Environment and Heritage and Australian Nature Conservation Agency, Canberra, ACT, pp Poiner IR, Staples DJ, Kenyon K (1987) The seagrass commmunities of the Gulf of Carpentaria, Australia. Aust J mar Freshwat Res 38: Rothschild B J, Brunenmeister SL (1984) The dynamics and management of shrimp in the northern Guff of Mexico. In: Gulland JA, Rothschild BJ (eds) Penaeid shrimps - their biology and management. Fishing News Books Ltd, Surrey, pp Sachse M, Robins C (1995) Northern Prawn Fishery: data summary Australian Fisheries Management Authority, Canberra Schoop CR, Ruckdeschel CR (1982) Increasing turtle strandings in the south-east United States: a complication factor. Biol Conserv 23: Somers IF (1990a) Manipulation of fishing effort in Australia's penaeid prawn fisheries. Aust J mar Freshwat Res 41:1-12 Somers IF (1990b) Expected long term average catch in the NPF. Aust Fish 49(6): Somers IF (1993) Commercial penaeid prawn species distributions in the Gulf of Carpentaria. Aust J mar Freshwat Res 45: Somers IF (1994) Modelling loggerhead turtle populations. Proceedings of Australian Marine Turtle Conservation Workshop, Seaworld Nara Resort, November Queensland Department of Environment and Heritage and Australian Nature Conservation Agency, Canberra, ACT, pp 14~145 Somers IF, Taylor BR (1981) Fisheries statistics relating to the declared management zone of the Australian Northern Prawn Fishery, Rep mar Lab CSIRO 138:1-13