Biology of the Shortfinned Eel Anguilla Obscura in Lake Te Rotonui, Mitiaro, Cook Islands 1

Biology of the Shortfinned Eel Anguilla obscura in Lake Te Rotonui, Mitiaro, Cook Islands 1

D. J. JELLYMAN 2

ABSTRACT: Lake Te Rotonui, a shallow depression lake in the center of Mitiaro Island, southern Cook Islands, contains freshwater eels despite having no surface connection to the sea. During a survey of the eel population in July 1988, all ofthe 287 eels captured using fyke nets and gaffs were Anguilla obscura, although it is possible that A. megastoma and perhaps A. marmorata also occur in small numbers. Ages of eels were found from burnt otoliths; it was assumed that otolith zones were formed annually, although this could not be validated. Growth rates were slower than those of other tropical eel species, being similar to those of temperate species. Eels fed exclusively on Oreochromis mossambica, which was abundant in the lake. The relatively slow growth in the presence of abundant food may be due to high and stressful summer water temperatures. From length and age frequency distributions, it is suggested that recruitment of glass-eels into the lake is intermittent and via submarine outfalls. A review of the limited larval information suggested that A . obscura spawns to the east of Tahiti, with larvae transported west and south by the South Equatorial Current.

ALTHOUGH THE TEMPERATE SPECIES of fresh MATERIALS AND METHODS water eels (genus Anguilla) have been widely studied, there have been few studies of the Study Area tropical species, especially those of the Pacific region. Anguilla obscura is a tropical short The island of Mitiaro (190 l' S, 1570 3' W) finned species that ranges from western New is in the southern Cook Islands, ca. 230 km Guinea to Tahiti (Ege 1939). Biological data northeast of Rarotonga. Mitiaro is 2500 ha available for this species are limited larval in area and saucer-shaped in cross section , information (Jespersen 1942, Castle 1963, with a raised coral limestone reef (makatea) Matsui et al. 1970), meristic features (Ege surrounding a central depression (420 ha) 1939, Beumer et al. 1981, Marquet and occupied by swamps, lakes, and four small Lamarque 1986), distribution (Ege 1939, Cas basalt "islands" (Figure 1). Cliffs around the tle 1968), and brief observations on some outer margin of the island rise to a maximum aspects of freshwater biology (Marquet and height of 15 m above mean sea level (M.S.L.) Lamarque 1986). The present study is the first and the lakes are ca. 2 m above M.S.L. The that examines growth rates. It was carried island has a human population of 250 living out as part of a New Zealand Government in villages adjacent to Omutu landing, an Foreign Aid project to assess the eel popula artificial gap in the fringing coral platform. tion of Lake Te Rotonui and to advise on the Lake Te Rotonui has a surface area of 70 commercial viability of the stock. ha at a water level of 2 m above M.S.L. , although this may increase to 114 ha at maximum levels. The western margin is shal low and convoluted and the lake increases in 1 Manu script accepted 7 February 1991. depth to the east , with a maximum depth 2 Freshwater Fisheries Centre, Ministry of Agriculture and Fisheries, P.O. Box 8324, Riccart on, Christchurch, of 2-2.5 m to the south of the Parava Track. New Zealand. Because of the shallow and exposed nature 362 Biology of Anguilla obscura-JELLYMAN 363

o 1km I t I

FIGURE I. Mitiaro Island showing Lake Te Rotonui, the Parava work site (asterisk), the area fished (diagonal lines), and the location of the freshwater outfall (star). 364 PACIFIC SCIENCE, Volume 45, Octobe r 1991 of the lake and the presence of frequent Capture and Handling Techniques sea breezes, it is unlikely that any thermal stratification would occur. Eels were sampled using 12 unbaited, The lake does not have a surface outlet, and single- wing fyke nets (mesh size, 20 mm ; drainage is presumed to be via subterranean leader, 2.9 m). Each net was numbered, set at conduits through the porous makatea. The right angles to the shore, and its location only observed outfall is a seepage area north recorded. Nets were lifted and reset each of the airstrip (Figure I) where flow was morning and eels were taken to a landing estimated at 4-5 liters/sec, although the point adjacent to the Parava Track (Figure I) islanders are aware ofsome seepage ofsimilar for processing . Additional eels were gaffed by rate at the Omutu landing. During the survey, one of the local men, who would quietly 20-27 Ju ly 1988, the lake level fell at a paddle his canoe until sighting an eel head rate of 10.8 mm/day. Evaporation from the emerged from the algal mat. The eel could lake calculated from air temperature data is then normally be gaffed with a swift upward estimated at 3.5 mm/day (I. Jowett, MAF, thrust and pulled into the canoe. pers. comm.), meaning that the remaining fall Eels from fyke nets were anesthetized in a (7.3 rum /day) must have been due to sub solution of benzocaine. The length of all marine outflow, equivalent to 59 liters/sec , eels caught was recorded (± I mm), and the less the seepage observed. weight of most was measured to the nearest Between 20 and 27 July the midday surface 109. Where eel numbers allowed, a minimum water temperature fell from 30°C to 19.5°C. of 10 pairs of otoliths were taken from each Conductivity of the lake water ranged from 50-mm length group represented. The stom 267 to 285 mS/m. No aquatic macrophytes ach contents of these eels were also recorded were seen in the lake, bu t the bottom was using a points system, where 0 point s denoted covered with a 0.5-m mat of blue-green an empty stomach, 40 points a full stomach, algae and the decomposition products of this and 45 points a distended stomach. Where mat. The alga has been tentati vely identified digestion permitted, food items were identi as Coelospharerium (Chro ococcaceae). No fied and their numbers recorded. aquatic invertebrates were seen in the lake, Three experiments were carried out as steps although adult damselflies (Odonata) were to estimate the number of catchable eels common. in the lake (i.e., eels > 350 mm and hence Six species of fish were caught. Native large enough to be fyke-netted). The first was species were the eel Anguilla obscura ("tuna") carried out to provide information on the and the e1eotrid Eleotris fusca ("kokopu"). number of eels accessible to a net set at one Of the introduced species, guppies (Poecilia location. The technique itself was a progres reticulata), mosquitofish (Gambusia affinis), sive removal experiment using nets set for and the cichlid Oreochromis mossambica were three or more successive nights at the same abundant around the shallow margins of the site. The second experiment was designed to lake. Mosquitofish were introduced to the give a measure of the effective area of a net ; it Cook Islands before 1948 (Krumholz 1948), was presumed that this area would equate to probably from Hawaii. Together with guppies the average home range of an individual eel. they provide a means of controlling mos That mo st eels within a population exhibit quitoes. Oreochromis mossambica was intro localized mo vements within a home range has duced to several South Pacific islands during been established for A. rostrata by LaBar and the mid- to late 1950s (Maciolek 1984). The Facey (1983) and Bozeman et al. (1985). For population in Lake Te Rotonui was dense and this experiment, eels captured at one sho reline stunted (author' s unpublished data). Occa location were fin-clipped for later recognition siona l milkfish, Chanos chanos, were seen; and then released at the capture site. The these fish resulted from a small trial stocking follo wing night five net s were set equidistan tly in 1984. at a radius 75 m from the capture site, and any Biology of Anguilla obscura-JELLYMAN 365 marked eels were recorded and removed. On for determining the age of freshwater eels subsequent nights, the five nets were set at (e.g., Aprahamian 1987). For this, whole oto radii of 50 m and 25 m, respectively. liths were held directly over a hot Bunsen The third experiment was a mark-recapture burner flame. After burning, the otolith was study using individually numbered metal broken transversely using gentle pressure from strap tags that were clamped at the base ofthe a scalpel blade and the two halves embedded pectoral fin. Eels not killed for otolith removal by the base in a small quantity of silicon or fin-clipped for the home range estimate rubber adhesive placed on a microscope slide. were tagged and released at the capture site. A drop ofmicroscope immersion oil placed on Population size was estimated by both the the broken surface highlighted the ring forma Schumacher and Schnabel multiple census tion, and otoliths were viewed using a bin techniques (Ricker 1975). ocular microscope (30 x power) with strong side illumination. An overall readability index (scale 1-5) was Meristics recorded for each pair of otoliths, and the Proportional body measurements and ver otolith diameter (across the shortest axis) was tebral counts are widely used to distinguish measured with a graduated eyepiece. Age in between the various species of Anguilla. As it fresh water was calculated by counting the was thought possible that the Australasian number of dark (hyaline) "winter" zones. shortfinned eel, Anguilla australis, might Where available, all four halves of each pair occur in small numbers, several measure ofotoliths were read; if age varied, the modal ments were taken to identify eels to species age was recorded, or if one half was notice level. Anguilla australis and A. obscura can ably more readable, age from that portion was be differentiated by origin of the dorsal fin taken. relative to the anus and the position ofthe eye Statistics used were simple linear regression. relative to the lower jaw (Ege 1939). Measure ments recorded were total length, predorsal length, preanal length, length of lower jaw, RESULTS and the distance from the gape to a perpendic Eel Species ular line passing through the midpoint of the eye. Indices of the origin of dorsal and anal All eels examined (287) were non variegated fins and the position of the eye relative to and shortfinned. Proportional measurement jaw length were calculated according to the data confirmed that they were A . obscura. method of Ege (1939) and Beumer et al. Measurements of dorsal fin origins and eye (1981). Vertebral counts were made from position from a sample of 30 eels (422-845 X rays of a sample of 19 eels collected from mm) gave mean indices of 5.2 (range 3.1-7.2, Lake Te Rotonui in 1978 by P. R. Todd S.D . ± 1.1) and 44.5 (range 33.3-53.6, S.D. ± (Freshwater Fisheries Centre, Christchurch). 5.7), respectively. Although the mean fin ori gin index was within the range recorded by Ege (1939), the mean eye index was outside Condition and Growth Ege's range . However, Ege's data were for The condition factor (K) of individual eels generally smaller eels than in the present was calculated using the formula: study . When Ege's data (given as mean eye index per 100-mmlength group) are combined K = W X 106 jL 3.3 8 with similar data from the present study, there where W = weight in grams, L = length in is a strong linear relationship between eye mm, and 3.38 = the exponent derived from position and total fish length (r = 0.95, P < the length-weight relationship. 0.01). Vertebral counts ranged from 101 to Age was determined by viewing burned 104, with a mean of 103.79 (S.D . ± 0.69, otoliths; this technique has been widely used n = 19). 366 PACIFIC SCIENCE, Volume 45, October 1991

13 D fyke-netted 12

11 I gaffed 10

o Z 6

40 0 500 600 70 0 800 Length (mm)

FIGURE 2. Length freque ncy histograms of all eels captured (n = 287).

tion for linear regression was Catches and Siz es ofEels L = 0.0031 (D) + 0.7949 (n = 118) A total of 264 eels were caught over seven nights' fishing. Catches per net ranged from where L = fish length in mm and D = otolith o to 26, with overall catch per unit effort diameter in mm. (CPUE) being 3.14 eels per net per night. An Only one pair of otoliths was rejected as additional 23 eels were gaffed. unreadable although otoliths were generally The smallest eel caught by fyke net was 337 considered difficult to read. Hyaline zones mm, although sizes < 380 mm were not well tended to be broad and diffuse rather than represented (Figure 2). The mean length of narrow and distinct, meaning that the fyke-netted eels (538 mm, S.D . ± 105mm) was differenti ation between adjacent hyaline and smaller than that for gaffed eels (647 mm, opaque zones was often difficult to determine. S.D. ± 74 mm). Ages ranged from 6 to 28 yr in fresh water. The length-weight relationship was The age-length relationship for eels is shown in Figure 3. The average growth curve is loge W = 3.3832 (loge L) - 15.3713 shown by the line and was derived from the (n = 145, r = 0.99) equation: where L = length in mm and W = weight in loge L = 0.5817 (loge A) + 4.7238 grams. (n = 117,r=0.71) The mean condition factor ofeels (n = 145) was 2.12 (range 1.62-2.91, S.D. ± 0.23), and where L = length in mm and A = age in there was no relationship between total length years. and condition. The first 69 eels netted were all aged and so consti tute an unbiased (unstratified) sample. Con sequently, otoliths taken from these were Age and Growth used to generate the age frequency distribu Otolith diameter was strongly correl ated tion (Figure 4). Mean age from this sample with eel length (r = 0.88, P < 0.01). The equa- was 16.4 yr (S.D . ± 3.6). Biology of Anguilla obscura-JELLYMAN 367

800 ·...... 70 0 · .. • • • • • .. A · 600 ··• · · ·.. • E 500 .. • . . Ie · E • • • •· · · or. • .... 400 • OJ ··· l: · . ·• · .....CIl • · ··· 300

200

100

0 5 10 15 20 25 30 Freshwater age (yea rs)

FIGURE 3. Age-length relationship of individual eels. The curved line is the log-transformed least squares regression referred to in the text.

12 ~ J 10 .: \ j' V %8 6 /

4 /

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 A ges ( years)

FIGURE 4. Frequency distribution of ages from a random sample of fyke-netted eels (n = 69). The continuous line shows the three-point moving average. 368 PACIFIC SCIENCE, Volume 45, October 1991

TABLE I range experiment, at a distance of 25 m from

STOMACH FULLNESS OF EELS (n = 117) the release point at the lake margin . The seven other recaptures were all tagged eels.Of these, STOMACHS three had moved 30 m from their release point, one moved 100 m, one moved 150 m, FULLNESS NO. % and two moved 250 m. The mean distance moved was 108 m. Because of the variability E m pty 43 37 <± 20 17 in these few data, it was not possible to 41-Z 1 6 5 establish an area for home range of eels with 21- 4 3 13 11 any confidence. i -full 20 17 Population estimates for nets set for three Full-distended 15 13 or more nights were also quite variable (Table 3). Because substantial numbers ofeels could Diet still be caught after three nights' fishing, the estimates for the six fishing nights were con Because stomachs examined (n = 117) came sidered to be the more appropriate data. from those eels killed for oto lith removal, the These gave a mean population of 37 eels sample was stratified by length group but available to each net. covered the full size range of eels caught. It had been anticipated that this population Stomach fullness is given in Table 1. The estimate would be incorporated with the average score for stomachs that contained home range estimate to provide an estimate of food (n = 74) was 26.1 points. eel density. The density estimate could then be Only three food categories were recorded extrapolated for the whole lake. In practice, from stomachs. These were Oreochromis, because the home range data proved inconclu unidentifiable fish (due to degree of digestion), sive such total population estimates were not and algae. A summary of the information carried out. (Table 2) shows the dominance of Oreo Population estimates from the modified chromis in the diet, and it is very likely that all Schnabel method and the Schumacher meth the "unidentified fish" were also Oreochromis. od were 1005 eels (95% CL, 460-3652) and The maximum number of Oreochromis per 1055 eels (95% CL, 598-3127), respectively. stomach was four and the largest Oreochromis These estimates were relative to a shoreline found was 140 mm. Oreochromis also domi length of 900 m. nated the diet of gaffed eels (n = 16), con Again, a lack of a sound estimate of home stituting 76% of all points. range caused difficulty in converting this shoreline length to the effectivearea fished by Population Estimates the nets. If the modal recapture distance (30 m) is used as a conservative estimate, then Eight marked eels were recaptured , only the Schnabel and Schumacher estimates are one of which was caught during the home relative to an area of 2.7 ha. Assuming a

TABLE 2

STOMACH CONTENTS OF EELS (n = 117)

NO. OF NO. OF STOMACHS % OF AVERAGE POINTS INDIVIDUAL MAXIMUM NO. FOOD ITEM CONTAINING ITEM POINTS PER STOMACH ORGANISMS PER STOMACH

Empty 43 Oreochromis 45 85.7 36.8 68 4 Unidentified fish 18 10.3 11.1 5+ 3 Algae 16 4.0 4.8 Biology of Anguilla obscura-JELLYMAN 369

TABLE 3

CATCHES OF EELS FROM FYKE NETS ON SUCCESSIVE N IGHTS AT THE SAME SITE

NIGHT TOTAL POPULATION CL NET 2 3 4 5 6 CATCH ESTIMATE (95%)

A 8 6 6 - * 20 36 20-86 B 8 1 0 9 9 9- 9 Means 8.0 3.5 3.0 15 16 15-20 C 10 20 0 6 10 0 46 54 46-67 D 14 5 2 3 8 2 34 40 34-51 E 6 0 4 2 3 0 15 16 15- 21 Means 10.0 8.3 2.0 3.7 7.0 0.7 32 37 32-47

*- = not fished. uniform distribution of eels throughout the (1939) for A . obscura. It is probable that such lake, this equates to total population esti relatively high values are due to the larger eels mates of 26,050 and 27,350 eels, respectively. in the present study, as Ege's data indicate Equivalent estimates using the mean recap that these indices are positively related to ture distance of 108mare 7240 and 7600 eels. increasing size. Vertebral counts (mean 103.8) were similar to those given by Ege (1939) for A . obscura from the Cook Islands (mean DISCUSSION 103.9) and identical to those of Marquet and Lamarque (1986) for A. obscura from French All eels caught in the present study were Polynesia. nonvariegated and shortfinned, which con Although there have been many growth firmed their identify as A. obscura. Although studies of temperate eels, there have been single specimens of the shortfinned eel A. few studies on tropical eels, and none on australis have been recorded from both Fiji A . obscura. From data in Marquet and and Tahi ti (Ege 1939), the validity of these Lamarque (1986), it can be calculated that identifications is doubtful, and hence it is they considered that A . obscura reached thought unlikely that A. australis occurs in the lengths of420 mm and 530 mm by the end of Cook Islands. Anguilla megastoma, a varie the first and second years, respectively, in gated longfinned species, has been recorded fresh water. To date, growth rates this rapid from the Cook Islands (Ege 1939), but A. for Anguilla spp. have only been recorded in marmorata, another variegated longfinned culture conditions (e.g., Usui 1974), and it species, has not been recorded even though seems unlikely that such rapid growth would the Cook Islands are within its geographic occur in the wild. range. No longfinned eels were found during Use of otoliths for determination of age of the present survey, but local people indicated tropical fish is successful for some species but that a larger species of eel, corresponding in not others (Samuel et al. 1987). The technique description to A . megastoma or A . marmorata, has already been used for the tropical fresh was sometimes caught in Lake Te Rotonui but water eel A . nebulosa (Frost 1955,Pantulu and was uncommon. Thus the presence of species Singh 1962, Balon 1975). Although she used other than A . obscura in Lake Te Rotonui is oto liths to determine age in A. nebulosa, suspected but not confirmed. Frost (1955) expressed some doubt about the Measurements for fin origin and the posi annual formation of oto lith zones. Pantulu tion of the eye relative to jaw length were, and Singh (1962) studied zone formation on respectively, outside or at the upper end of otolith margins over a year and concluded the range of measurements recorded by Ege that formation was indeed annual. Reviewing 370 PACIFIC SCIENCE, Volume 45, October 1991 the formation of annual zones on the bony recorded some fish in the diet of small eels tissue of tropical fish, Balon (1975) suggested (< 200 mm). Because of the cryptic behavior that this formation was not due to climatic ofjuvenile eels, it is unlikely that they would factors alone but also to an endogenous commence feeding on fish immediately upon annual rhythm in fish growth that is regulated their arrival as glass-eels (mean length ca. but not governed by environmental factors. 50 mm [Ege 1939, Marquet and Lamarque In the present study, it was not possible to 1986]) in Lake Te Rotonui. Although no validate the annual formation ofotolith zones benthic invertebrates were observed in the in A . obscura. Alternating hyaline and opaque lake, it is probable that juvenile eels are zones were present, but , as mentioned previ principally benthic scavengers for their first ously, the boundaries between zones were fewyears in Lake Te Rotonui before changing less distinct than in the temperate eels that to an exclusive fish diet. I am familiar with (A . australis and A. Given the abundant supply offish prey, it is dieffenbachii). The hyaline zone was relatively surprising that the growth rate of eels in Lake wide and sometimes split into several super Te Rotonui was not faster, approaching that numerary rings. Despite these differences, it of A. nebulosa. A possible explanation lies in was assumed that, like A . nebulosa, the zoning the temperature extremes of the lake itself. in otoliths of A. obscura is annual in formation During the sampling period, the maximum and can be used to establish age. water temperature recorded was 30.0°C, al Growth rates for A. nebulosa labiata (Frost though higher water temperatures would be 1955, Balon 1975) and A. nebulosa nebulosa expected during summer. (Pantulu and Singh 1962) were considerably No data are available on temperature toler faster than those for A . obscura in the present ances of A. obscura or other tropical eels, but study . Although A . obscurais a tropical spe temperatures greater than 30°C are stressful cies, growth rates from the present study were for temperate species, with the upper lethal comparable to those of temperate species limit for A . anguilla being 38°C (Sadler 1979). (e.g., A. mossambica [McEwan and Hecht For temperatures above the optimum of 1984], A. anguilla [Sinha and Jones 1967], 26SC, activity and appetite decline in A . A . rostrata [Ogden 1970],A. australis [Burnet anguilla, although metabolic rate continues 1969], and A. reinhardtii [Sloane 1984]). to increase (Seymour 1989). Perhaps high Eels examined from Lake Te Rotonui had daily water temperatures in combination with fed exclusively on Oreochromis, a shallow low levels of dissolved oxygen at night from water demersal species, which was available algal respiration produce some physiological over a size range of 8-200 mm. In common stress during summer with resultant slow with Oreochromis elsewhere, it is likely that growth. This is consistent with islanders' ob some breeding occurs throughout the year servations that the now-abandoned baited (e.g., Huet 1972). During breeding, nest trap ("hinaki") fishery was only successful guarding by male Oreochromis probably from May to September, as higher water increases their vulnerability to nocturnal temperatures outside these months caused predation by eels. In addition to small eels to become torpid and "hibernate in the Oreochromis, guppies and mosquitofish were mud ." also present in the shallows but were not Cook Island specimens of A . obscura found in eel stomachs. No eels < 300 mm (present study) were heavier than specimens were captured, thus diet of juvenile eels is of equivalent length from French Polynesia unknown, but small fish would be readily (Marquet and Lamarque 1986). Similar geo available. graphic variation in length-weight relation The incidence of fish in the diet of eels ships is seen in A. nebulosa, where two studies invariably increases with increasing size ofthe (Frost 1955, Wickstrom and Enderlein 1988) eel (e.g., Jellyman 1989). Fish do not become indicate that the species is lighter than A. an important part of the diet of A . australis obscura for a given length, while a third until eels are > 300 mm, but Sloane (1984) (Balon 1975) indicates the reverse. Biology of Anguilla obscura -JELLYMAN 371

Because of difficulties in establishing the A. marmorata, were present almost year effective fishing area of nets and a reliable round in an estuary in the Philippine Islands. estimate of eels present within the area fished, Spawning grounds of A. obscura are it was not possible to establish a total popula unknown. Leptocephali have been recorded tion estimate with any certainty. For example, from four areas: near Tahiti and Fiji (Jesper using modal and mean recapture distances sen 1942), near the New Hebrides (Castle produced a 3.6-fold difference in such esti 1963), and off northeastern New Guinea mates . Assuming the smaller estimate (7240 (Matsui et al. 1970). The larva occurring 7600 eels) to be a conservative one, then farthest east (Tahiti) was also the smallest (27 correspondingly conservative biomass esti mm), indicating dispersal from east to west. mates (relative to the mean size offyke-netted Jespersen (1942) presumed that this small size eels of 538 mm) are 38-40 kgjha of eels indicated proximity of the spawning ground, accessible to fyke nets. but as larvae have been recorded at 10° Both the length and age frequency distribu latitude north of Tahiti, it is likely that the tions for A . obscura show a nearly normal spawning ground is northeast of Tahiti, possi distribution, whereas positively skewed distri bly east of the Marquesas Islands. butions would be anticipated in populations Larval transport would be similar to that where recruitment is unrestricted and some proposed for A. australis by Jellyman (1987), harvesting of adult eels takes place. Certainly with larvae transported to New Guinea and the present limited gaff fishery appears selec the east coast ofAustralia by the westerly flow tive for larger eels (> 500 mm) and should of the South Equatorial Current, while a accentuate such skewness. A likely explana southwest branch of the same current would tion for this lack of skewness is that annual convey larvae to New Caledonia, Tonga, recruitment is intermittent, possibly due to Samoa, Fiji, and the Cook Islands. the difficulty of glass-eels in locating the sub marine outfall(s). Islanders have no knowl edge of recruitment ofjuvenile eels from the ACKNOWLEDGMENTS sea and hence believe that eels breed within I wish to thank the New Zealand Ministry the lake. Castle (1968) recorded a similar ofForeign Affairs for funding this study, with situation from Rennell Island (Solomon particular thanks to Megan Adams for her Islands) where Lake Tegano contains A. assistance in negotiations with the Cook Is obscura but has no surface connection with land Government. I am grateful to the able the sea. Here also the islanders believe that and cheerful field assistance given by Ruru eels breed in the lake although a connection Maoate of the Cook Islands Department of through the coral is now known. Marine Resources and also to the administra Recruitment times for A. obscura glass-eels tive assistance ofNed Howard ofthat depart are not well known. Ege (1939) obtained ment. I also thank Barry Biggs, Department samples of glass-eels from New Caledonia of Scientific and Industrial Research, Christ from March to July and October, while church, New Zealand, for identification ofthe Marquet and Lamarque (1986) captured blue-green algae. Finally, I acknowledge all glass-eels in Tahiti from October to April with the practical help and friendship ofthe people the main periods being January and April. ofMitiaro Island, especially Ariki Tiki Tetava These limited data indicate some seasonality and the Island Council. ofrecruitment, with time differences probably reflecting distance from spawning grounds. If correct, this is at variance with the suggestion LITERATURE CITED of Jespersen (1942) that tropical eels spawn throughout the year, or at least for a large part APRAHAMIAN, M. W. 1987. Use ofthe burning of it. It also differs from information con technique for age determination in eels tained in Tabeta et al. (1976) where glass-eels (Anguilla anguilla (L.» derived from the of two tropical species, A. celebensis and stocking of elvers. Fish. Res. 6: 93-96. 372 PACIFIC SCIENCE, Volume 45, October 1991

BALON, E. K. 1975. The eels of Lake Kariba: movements and inshore population sizes distribution, taxonomic status, age, growth of American eels in Lake Champlain, and density. J. Fish BioI. 7:797-815. Vermont. Trans. Am. Fish. Soc. 112: 111 BEUMER, J. P., R. G. PEARSON, and L. K . 116. PENRIDGE. 1981. Pacific shortfinned eel, McEWAN, A., and T. HECHT. 1984. Age Anguilla obscura Gunther, 1871 in Austra and growth of the longfin eel Anguilla lia: Recent records of its distribution and mossambica Peters, 1852 (Pisces: Anguil maximum size. Proc. R. Soc. Queens!. 92: lidae) in Transkei rivers. S. Afr. J. Zoo!' 85-90. 19: 280-285. BOZEMAN, E. L., G. S. HELFMAN, and T. MACIOLEK, J. A. 1984. Exotic fishes in Hawaii RICHARDSON. 1985. Population size and and other islands of Oceania. Pages 131 home range of American eels in a Georgia 161 in W. R. Courtenay, Jr. , and J. R. tidal creek . Trans. Am. Fish. Soc. 114: 821 Stauffer, Jr., eds. Distribution, biology, 825. and management of exotic fishes. Johns BURNET, A. M. R. 1969. The growth of New Hopkins University Press, Baltimore. Zealand freshwater eels in three Canterbury MARQUET, G., and P. LAMARQUE. 1986. streams. N. Z. J. Mar. Freshwater. Res. Acquisitions recentes sur la biologie des 3 :376 -384. anguillies de Tahiti et de Moorea (Polynesia CASTLE, P. H. J. 1963. Anguillid leptocephali Francaise): A. marmorata, A. megastoma, in the southwest Pacific. Zoo!' Pub!. Victo A. obscura. Vie Milieu 36 :311 -315. ria Univ. Wellington 33: 1-14. MATSUI, I., T. TAKAI, and A. KATAOKA. 1970. ---. 1968. Anguilla obscura on Rennell Leptocephalae of the eel Anguilla obscura Island. Nat. Hist. Rennell Is. Br. Solomon found in the stomachs of skipjack tuna Is. 5 :61-66. Katsuwonus pelamis caught near New EGE, V. 1939. A revision ofthe genus Anguilla Guinea. J. Shimonoseki Univ. Fish. 19: 25 Shaw. A systematic, phylogenetic and 28. geographical study. Dana-Rep. Carlsberg OGDEN, J. C. 1970. Relative abundance, food Found. 16. habits, and age of the American eel, FROST, W. E. 1955. Observations on the biolo Anguilla rostrata (Le Sueur) in certain New gy ofeels (Anguilla spp.) of Kenya Colony, Jersey streams. Trans. Am. Fish . Soc. 99: East Africa. Colon. Off. Fish. Pub!. 6. 54-59. HUET, M. 1972. Textbook of fish culture. PANTULU, V. R., and V. D. SINGH. 1962. On Breeding and cultivation of fish. Fishing the use of otoliths for the determination News (Books), London. of age and growth of Anguilla nebulosa JELLYMAN, D. J. 1987. Review of the marine McClelland. Proc. Indian Acad. Sci. Sect. life history of Australasian temperate spe B:263-275. cies of Anguilla. Am. Fish. Soc. Symp . RICKER, W. E. 1975. Computation and inter 1: 276-285. pretation ofbiological statistics offish pop - - -. 1989. Diet of two species of fresh ulations. Fish. Res. Board Can. Bul!. 191. water eel (Anguilla spp.) in Lake Pounui, SADLER, K. 1979. Effects of temperature on New Zealand. N. Z. J. Mar. Freshwater the growth and survival ofthe European eel Res. 23: 1-10. Anguilla anguilla L. J. Fish Bio!. 15:499 JESPERSEN, P. 1942. Indo-Pacific leptoceph 507. alids of the genus Anguilla. Systematic and SAMUEL, M. , C. P. MATHEWS, and A. S. biological studies. Dana-Rep. Carlsberg BAWAZEER. 1987. Age and validation ofage Found. 22. from otoliths for warm water fishes from KRUMHOLZ, L. A. 1948. Reproduction in the Arabian Gulf. Pages 253-265 in R. C. the western mosquito fish Gambusia affinis Summerfelt and G. E. Hall, eds. Age and affinis (Baird and Girard) and its use in growth offish. Iowa State University Press, mosquito contro!. Eco!. Monogr. 18: 1-43. Ames. LABAR, G. W., and D. E. FACEY. 1983. Local SEYMOUR, E. A. 1989. Devising optimum Biology of Anguilla obscura- JELLYMAN 373

feeding regimes and temperatures for the occurrence of anguiIIid elvers in Cagayan warmwater culture of eel, Anguilla anguilla River, Luzon Island, the Philippines. Bull. L. Aquae. Fish. Manage. 20 : 129- 142. Jpn. Soc. Sci. Fish. 42:421-426. SINHA, V. R. P., and J. W. JONES. 1967. On USUI, A. 1974. Eel culture. Fishing News the age and growth of the freshwater eel (Books), London. (Anguilla anguilla). J. Zoo!. 153: 99- I 17. WICKSTROM, H., and O. ENDERLEIN. 1988. SLOANE, R. D. 1984. Distribution, abundance, Notes on the occurrence of two tropical growth and food offreshwater eels (Anguilla species of Anguilla in reservoirs in south spp .) in the Douglas River, Tasmania. eastern Sri Lanka and preliminary data on Aust. J. Mar. Freshwater Res. 35 :325-339. the populations. Aquae. Fish. Manage. TABETA, 0., T. TANIMOTO, T. TAKAI, I. 19: 377-385. MATSUI, and T. IMAMURA . 1976. Seasonal