Migration Behavior of Anguilla celebesensis Silver Eels within their Tomini Bay Spawning Area

© 2023 Academia Sinica, Taiwan Open Access Migration Behavior of Anguilla celebesensis Silver Eels within their Tomini Bay Spawning Area Ryotaro Manabe1, Takatoshi Higuchi2,3 , Shun Watanabe4,5,*,§ , Fadly Y. Tantu6,*,§ , Hagi Y. Sugeha7, Hiroki Kaneko8, Michael J. Miller9, Seishi Hagihara10 , Tatsuki Yoshinaga11 , Augy Syahailatua7,12, Sam Wouthuyzen7,12, Triyanto13, Kawilarang W. A. Masengi14, Katsufumi Sato15 , Jun Aoyama2, and Katsumi Tsukamoto9 1Ainan Fisheries Division, Agriculture, Forestry, and Fisheries Promotion Department of the Nanyo Regional Bureau, Ehime Prefectural Government, 2420 Johenko, Ainan-cho Minamiuwa-gun, Ehime 798-4194, Japan. E-mail: [email protected] (Manabe) 2Center for International and Local Research Cooperation, Atmosphere and Ocean Research Institute, The University of Tokyo, 1-19-8 Akahama, Otsuchi, Iwate 028-1102, Japan. E-mail: [email protected] (Aoyama) 3UMR BOREA [MNHN, CNRS, SU, IRD, UCN, UA], Station Marine de Dinard, 38 rue du Port Blanc 35800 Dinard, France. E-mail: [email protected] (Higuchi) 4Department of Fisheries, Faculty of Agriculture, Kindai University, 3327-204 Nakamachi, Nara 631-8505, Japan. *Correspondence: E-mail: [email protected] (Watanabe) 5Agricultural Technology and Innovation Research Institute, Kindai University, 3327-204 Nakamachi, Nara 631-8505, Japan 6Department of Aquaculture, Faculty of Animal Husbandry and Fishery, Tadulako University, Jl. Soekarno Hatta Km 9 Palu, 94118 Central Sulawesi, Indonesia. *Correspondence: E-mail: [email protected] (Tantu) 7Research Centre for Oceanography - National Research and Innovation Agency, Jl. Pasir Putih 1, Ancol Timur, Jakarta 14430, Indonesia. E-mail: [email protected] (Sugeha); [email protected] (Syahailatua); [email protected] (Wouthuyzen) 8Department of Marine Science and Resources, Nihon University, 1866 Kameino, Fujisawa, Kanagawa 252-0880, Japan 9Department of Aquatic Bioscience, The University of Tokyo, 1-1-1 Yayoi, Bunkyo, Tokyo 113-8657, Japan. E-mail: [email protected] (Miller) 10Nanae Freshwater Station, Field Science Center for Northern Biosphere, Hokkaido University, 2-9-1, Sakura-cho, Nanae, Kameda-gun, Hokkaido 041-1105, Japan. E-mail: [email protected] (Hagihara) 11School of Marine Biosciences, Kitasato University, 1-15-1 Kitasato, Minami-ku, Sagamihara, Kanagawa 252-0373, Japan. E-mail: [email protected] (Yoshinaga) 12Research Centre for Collaboration on Aquatic Ecosystem in Eastern Indonesia. Jl. Ir. Putuhena, Ambon 97233, Indonesia 13Research Center for Limnology and Water Resources-National Research and Innovation Agency, JL. Raya Bogor Km 46. Bogor 16911, Indonesia. E-mail: [email protected] (Triyanto) 14Faculty of Fisheries and Marine Science, Sam Ratulangi University, Jl. Kampus UNSRAT Bahu, Manado, Sulawesi Utara 95115, Indonesia. E-mail: [email protected] (Masengi) 15Atmosphere and Ocean Research Institute, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8564, Japan. E-mail: [email protected] (Sato) §SW and FYT contribute equally to this work. Received 20 April 2023 / Accepted 10 July 2023 / Published 19 September 2023 Communicated by Hin-Kiu Mok The tropical Celebes eel, Anguilla celebesensis, has a short migration between its spawning and growth habitats. Its spawning areas were hypothesized to be in Tomini Bay and the Celebes Sea after collecting their small leptocephali. However, there is no information about the silver eel oceanic spawning migration behavior of A. celebesensis. To better understand their short-distance spawning migration behavior, four large female silver eels (Eel 1–4) were equipped with pop-up satellite archival tags (PSATs) and released near the mouth of the Poso River in Tomini Bay of Sulawesi Island on 22 February (Eel 1–3) and 11 March 2010 (Eel 4). All PSATs ascended in Tomini Bay and transmitted their data. Eel 3 and 4 provided clear records of consistent diel vertical migration (DVM: eight days-Eel 3, 13 days-Eel 4) with daytime dives to mean depths of 444.7 m (Eel 3) and 539.0 m (Eel 4), where mean temperatures were 9.1°C (Eel 3) and 7.7°C (Eel 4), and nighttime ascents to mean depths of 132.8 m (Eel 3) and 112.4 m (Eel 4), where Citation: Manabe R, Higuchi T, Watanabe S, Tantu FY, Sugeha HY, Kaneko H, Miller MJ, Hagihara S, Yoshinaga T, Syahailatua A, Wouthuyzen S, Triyanto, Masengi KWA, Sato K, Aoyama J, Tsukamoto K. 2023. Migration Behavior of Anguilla celebesensis silver eels within their Tomini Bay Spawning area. Zool Stud 62:46. doi:10.6620/ZS.2023.62-46. Zoological Studies 62:46 (2023) doi:10.6620/ZS.2023.62-46 1 © 2023 Academia Sinica, Taiwan mean temperatures were 20.6°C (Eel 3) and 23.4°C (Eel 4). Eel 3 and 4 started to dive to deeper water around nautical dawn and swam up to shallower water around sunset. During nighttime, both eels swam in deeper and colder water during nights with moonlight than during nights without moonlight, and there was a negative linear relationship between experienced water temperatures with the moon in the sky and the lunar age for the eels. The A. celebesensis daily rhythm of DVM behaviors was similar to spawningmigration DVM behaviors of other anguillid species. Essential life history characteristics of A. celebesensis appear to be a short migration between freshwater growth habitat and ocean spawning habitat, and high GSI values with advanced gonadal development in downstream-migrating silver eels. Key words: Celebes eel, Diel vertical migration, Lunar cycle, Spawning migration, Migration distance BACKGROUND Over the past decade, technological advances in telemetry have transformed the ability to observe aquatic animal behavior and movements (Hussey et al. 2015). The well-known long distance spawning migrations of freshwater eels fascinated biologists and stimulated research (Aoyama 2009), and now they have been studied using new technologies (Béguer-Pon et al. 2017). The early stages of the migration behavior patterns of anguillid silver eels were studied using acoustic tracking methods (see Béguer-Pon et al. 2017), and then more recently the oceanic migration behaviors were studied using pop-up satellite archival transmitting tags (PSATs). The PSATs collect depth, temperature, and light intensity data over long periods up to several months while attached to the silver eels. Many PSATs studies have been conducted on the oceanic migration behaviors of silver eels of the temperate species of Anguilla dieffenbachii (Jellyman and Tsukamoto 2010; Watanabe et al. 2020), A. anguilla (e.g., Aarestrup et al. 2009; Righton et al. 2016; Wright et al. 2022), A. japonica (Manabe et al. 2011; Chen et al. 2018; Higuchi et al. 2018 2021), A. rostrata (e.g., Béguer-Pon et al. 2017), and A. australis (Koster et al. 2021) and also of tropical silver eels of A. marmorata, A. megastoma, A. obscura, and A. bicolor pacifica (e.g., Schabetsberger et al. 2015 2021; Chen et al. 2018). These studies showed that anguillid silver eels, regardless of the species, consistently perform diel vertical migration (DVM) behaviors during their spawning migrations in offshore waters. These DVM behaviors consist of swimming at much deeper depths during daytime (mostly > 800 m) and then coming up to shallower depths at night (mostly < 200 m). These DVM behaviors of anguillid eels are thought to be for predator avoidance, although thermoregulation to control metabolic rate and gonad maturation may also play a role (Aarestrup et al. 2009; Jellyman and Tsukamoto 2010; Manabe et al. 2011). Higuchi et al. (2021) found three Japanese eels were ingested by predators near dawn at depths when the eels start the decent phase of their DVM, and potential predators were estimated to be tuna and swordfish that have heat conservation ability. European eels and American eels were also documented to be preyed on by several types of fish predators or marine mammals (Righton et al. 2016) and by porbeagle sharks (Lamna nasus) and likely Atlantic bluefin tuna (Thunnus thynnus) in the Gulf of St. Lawrence (Béguer-Pon et al. 2012). Because these DVM behaviors to swim deeper during the day to reduce predation from various predators are performed by both temperate and tropical anguillid silver eels, they may be part of the original characteristics of the genus Anguilla that appears to have originated in tropical regions (see Aoyama 2009). One place where many tropical anguillid species are present and short-distance spawning migrations occur is the Indonesian Seas region. At least five species (A. celebesensis, A. interioris, A. marmorata, A. borneensis, A. bicolor pacifica) of tropical anguillid eels are present in the region around Sulawesi Island of north-central Indonesia (Ege 1939; Sugeha et al. 2001; Aoyama et al. 2003 2018; Kuroki et al. 2006; Sugeha 2012). One of the common species on Sulawesi Island is the Celebes eel (Tsukamoto et al. 2020), A. celebesensis, which has spawning areas in the Celebes Sea and Tomini Bay based on collections of small eel larvae (Aoyama et al. 2003 2018). Because of the small size of Tomini Bay (Fig. 1), the spawning migration of A. celebesensis silver eels is shortest known spawning migration distance among anguillid eels. The spawning migration distance of this species in Tomini Bay was presumed to be only 80–300 km (Aoyama et al. 2003), which contrasts with the much longer migration distances of temperate eels of about 900–8000 km (Aoyama 2009). Moreover, A. celebesensis silver eels migrating out from Lake Poso of central Sulawesi Island have high gonadosomatic index (GSI) values and remarkably welldeveloped oocytes just before downstream migration page 2 of 16Zoological Studies 62:46 (2023) © 2023 Academia Sinica, Taiwan Fig. 1. (A) Map of the Indonesian Archipelago region showing Tomini Bay (red rectangle). (B) The release (yellow star) and pop-up locations of the four tagged Celebes eels (red circles), and the location of the CTD hydrographic profile (black circle). -500 -500 -500 -500 -1000 -1500 -1000 -1000 -1500 -2000 -2500-3000 -3500 -500 1000 2000 -1000 -1500 -1500 Poso Lak e Release location Eel 3 Eel 4 Eel 2 Eel 1 120˚E 121˚E 122˚E 123˚E 124˚E 2˚S 1˚S 0˚ 1˚N CTD Sulawesi Island 500 100 km 90˚E 100˚E 110˚E 120˚E 130˚E 140˚E 150˚E 10˚S 0˚ 10˚N 20˚N −5000 0 Elevation (m) 5000 (A) (B) Pacific Ocean Indian Ocean Celebes Sea South China Sea Tomini Bay page 3 of 16Zoological Studies 62:46 (2023) © 2023 Academia Sinica, Taiwan compared to the A. marmorata that also migrate out of the lake (Hagihara et al. 2012 2020; Arai 2014), but spawn offshore far to the north in an overlapping area with A. japonica (Kuroki et al. 2009). Because of the possibly unique short-distance spawning migration of A. celebesensis in Tomini Bay, studying their migratory behavior could be useful to enhance our understanding of the ecology and evolution of anguillid eels. Therefore, the objective of this study was to examine the spawning migration behavior of A. celebesensis in Tomini Bay using PSATs. Furthermore, we also report a possible predation event on a tagged eel detected by analysis of the range of temperature differences between ambient water temperature and tagrecorded temperatures. MATERIALS AND METHODS Eel specimens, tag attachment, and release site Eight migrating Anguilla celebesensis were obtained from commercial fisherman using large weirs in the outlet of Lake Poso (1°45'30"S, 120°38'30"E) on Sulawesi Island, Indonesia during late February and early March 2010. Detailed information about the weirs was described in Hagihara et al. (2018c). Eels captured by the weirs can be regarded as migrating eels that had just started their downstream migration from Lake Poso to Tomini Bay, which is consistent with their silvering-related characteristics, gonadal development, and reproductive hormone levels (Hagihara et al. 2012 2020). These eels were transported in plastic bags (110 cm × 85 cm) in water containing one-third seawater while being cooled with ice until being held in a wooden box (100 cm × 200 cm × 100 cm deep) in the seawater along the coast of Tomini Bay for several days before release. The MiniPAT (Wildlife Computers, USA) used in the study weighed 53 g, were 115 mm in length (excluding antenna) and were pressure tested to a depth of 2000 m. The three tags used on 22 February 2010 (Eel 1–3) were preprogrammed to record water temperature (resolution ± 0.05°C) and depth (resolution ± 0.5 m) every 75 seconds. The tag of Eel 4 used on 10 March 2010 was preprogrammed to record water temperature and depth every five minutes. Moreover, ascents were set at 14 days for Eel 1–3 and at 60 days for Eel 4 after deployment. In case of premature death of the tagged eel or detachment of the tag from its body, the MiniPAT tags were preprogrammed to commence a premature release setup and then transmit data after reaching the surface. The premature release setup initiates compulsorily release of the tag at > 1800 m or after 72 h at a constant depth (± 2.5 m). However, this procedure did not work in two of the PSAT tags, because the tag did not surface after 72 h at 1800 m or deeper, but they did eventually detach from the weight that was part of the tag attachment harness. The four largest A. celebesensis (Eel 1–4) were chosen for attachment with PSAT (Table 1). Before release, they were individually removed from the wooden box, anaesthetized with eugenol (2‰) in a plastic bucket (80 cm × 80 cm), their lengths and body weights were measured, and the silvering index was determined based on Okamura et al. (2007) and Hagihara et al. (2012). Moreover, dorsal fin tissue was sampled for DNA species identification of each individual. Surgical tagging was carried out according to Manabe et al. (2011). Tagged eels were retained in a plastic bag (110 cm × 85 cm) with sea water until they recovered from anesthetization. Releases were conducted separately on 22 February and 11 March 2010. Eel 1–3 were released near the mouth of the Poso River in Tomini Bay from a boat at a point where the water depth was more than 200 m deep during daytime from 11:18 to 13:35 on 22 February, and Eel 4 was released at the same point (120°45'E, 1°22'S) at 12:45 on 11 March (Fig. 1B, Table 1). The eels were later genetically confirmed as being A. celebesensis by comparing their mitochondrial DNA 16S ribosome RNA sequences to existing sequences of the species (Aoyama et al. 1999). The eel release points and PSATs surfacing locations in the southern part of the inner/western region Table 1. Biological data of the four Celebes eels studied with PSAT tagging, and eel release and tracking information. TL: total length; BW: body weight; SI: silvering index Eel no. TL (mm) BW (g) SI Release timing Release location Pop-up timing Pop-up location Tracked duration (day) Longitude Latitude Longitude Latitude Eel 1 887 1400 S1 22 February 2010 11:18 120˚45'E 1˚22'S 14 March 2010 05:29 121˚10'E 0˚46'S 20 Eel 2 937 2050 S1 22 February 2010 12:32 120˚45'E 1˚22'S 09 March 2010 04:10 121˚11'E 1˚17'S 15 Eel 3 904 1640 S1 22 February 2010 13:35 120˚45'E 1˚22'S 09 March 2010 06:43 120˚47'E 0˚33'S 15 Eel 4 899 1350 S1 22 March 2010 12:45 120˚45'E 1˚22'S 17 June 2010 18:24 120˚30'E 0˚49'S 121 page 4 of 16Zoological Studies 62:46 (2023) © 2023 Academia Sinica, Taiwan of Tomini Bay were plotted over a bathymetric chart (SRTM_PLUS, Olson et al. 2014; Generic Mapping Tools, Wessel et al. 2013, Figs. 1B, S1). Definition of behavioral patterns To analyze the swimming depths and experienced water temperature of A. celebesensis in detail, the tracking records of the whole tracking periods were visually assessed and separated into periods of adaptation (I), DVM (II), predation (III), abnormal vertical behavior (IV), tag drifting or stopped position (V), and tag sinking (VI; Fig. 2). Then, the exact transition points between each status were determined Fig. 2. Depth and temperature records of the four Celebes eels tagged with PSATs and released into Tomini Bay. The swimming depths (blue lines) and water temperatures (red lines) are shown in each panel. Tracking data were separated into periods of adaptation (I), diel vertical migration (II), predation (III), abnormal vertical behavior (IV), tag drifting or stopped position (V), and tag sinking (VI). The yellow, half yellow, and black circles show days of full moon, first quarter moon, and new moon, respectively. The black inverted triangles show surfaced PSATs. Eel 4 recorded data until 0:00 on 10 April 2010 (D), but the tag remained at the same depth until 20 May, which is not shown. IV V IV V VIIIIIII III IV V 2000 1600 1200 800 400 0 0 10 20 30 2000 1600 1200 800 400 0 2000 1600 1200 800 400 0 2000 1600 1200 800 400 0 0 10 20 30 0 10 20 30 0 10 20 30 22 23 24 25 26 27 28 1 32 4 65 7 98 10 11 12 13 14 15 February 2010 March 2010 11 14 17 20 23 26 29 1 74 10 March 2010 April 2010 Swimming depth (m) Date (A) Eel 1 (B) Eel 2 (C) Eel 3 (D) Eel 4 Experienced water temperature (°C) page 5 of 16Zoological Studies 62:46 (2023) © 2023 Academia Sinica, Taiwan using R (R Core Team 2015) and the ‘changepoint’ package (Killick and Eckley 2014), which is explained by Higuchi et al. (2021). The adaptation phase is the period when some eels showed irregular vertical movements before they began performing DVM behaviors. The abnormal vertical behavior phase is the period when the eel ceased showing DVM, possibly because of reduced swimming stamina, or effects of predation. The tag drifting period is when the PSAT drifts on the sea surface after detaching from the eel. Data analysis of diel vertical migration To analyze the eels’ DVMs (behavior pattern II, Fig. 2C, D) in detail, the daily DVMs were divided into four basic DVM periods of daytime, ascent, nighttime, and descent based on the methods reported in Westerberg et al. (2014), Chow et al. (2015), and Higuchi et al. (2021). In brief, the start and end points were defined as periods that had ≥ 10 consecutive minutes of either ascending or descending depths during each transition period in both shallower nighttime and deeper daytime depths. The swimming depths and water temperatures were compared among the four DVM periods of each of the two eels that showed regular DVM using Kruskal-Wallis tests followed by Steel-Dwass tests. The swimming depths and water temperatures of each of the four DVM periods were compared between the two eels using Brunner-Munzel tests. For nighttime analysis periods, the two tagged eels (Eel 3 and 4) with tracks that covered more than half of a lunar cycle (lunar age: 0–15/15–30) were used to check analysis correlation accuracy between lunar age and the swimming depth and experienced water temperature. Astronomical factors such as lunar age (http://koyomi.vis.ne.jp/) and time of sunrise and sunset (https://keisan.casio.jp/exec/system/1184726771) were estimated at an intermediary point between the release and surfacing locations of each tag (Higuchi et al. 2021). The nighttime swimming depths were separated into when the moon was present in the sky (moon altitude > 0) and absent (moon altitude < 0), and they were compared by Brunner-Munzel tests during each night. The weather conditions such as cloud cover or rainfall were not taken into account in this analysis. The correlations between each astronomical factor and swimming depth and experienced water temperature in each of the four DVM periods were tested by normal/ partial Spearman’s rank correlation coefficient. Estimation of predation To detect possible predation events for the tagged eels, the recorded depths and water temperature data of each PSAT were compared to the environmental water temperature structure at the hydrographic observation station using Conductivity Temperature Depth profiler (CTD, Sea-Bird Electronics, USA) in Tomini Bay (120°45'E, 1°2'S, Figs. 1B, S2), which was in the center of the area where the eels were likely swimming. The CTD was made during a R/V Hakuho Maru research cruise in the central Indonesian Seas (JAMSTEC, Japan, 9 March 2010, KH-09-5, Leg 7, St. 6) to collect eel larvae (Aoyama et al. 2018) at the time of the eel tagging experiment. If differences between the tag-recorded temperature and the same-depth CTD profile temperature (Fig. S2) exceeded a standard criterion during tracking, tagged eels were judged to be preyed on by an endothermic predator. To establish a criterion of the temperature difference for detecting predation events, the difference between the experienced temperature of Eel 3 and environmental CTD water temperature was used (Higuchi et al. 2021). Large predators such as tunas and swordfish typically have large-scale vertical movement behaviors. The peritoneal cavity temperature of some endothermic fish is higher than the ambient water temperature just after diving from shallower/ warm to deeper/cold layers and lower than ambient water temperature just after rising from deeper/cold to shallower/warm layer (Musyl et al. 2003; Hino et al. 2019). The change point of mean and dispersion data of the temperature difference was used to determine if predation occurred using R (R Core Team 2015) with the ‘changepoint’ package (Killick and Eckley 2014). After predation occurred (behavior pattern III), the tag-recorded depth and temperature would reflect the swimming depth and peritoneal cavity temperature of the predator. Based on the pattern of the depth variation and temperature characteristics, the general type of the predator was estimated. RESULTS Pop-up location and tracking period All four PSATs that were attached to each eel were eventually released from the eels or egested from a predator and then reached the surface and transmitted their archived data to an ARGOS satellite (Fig. 1B, Table 1). The tags came off of the eels after 15–121 days (Table 1). The pop-up locations for three tags (Eel 1, 3, and 4) were in the southern part of the western region of Tomini Bay and the other tag (Eel 2) ascended near the Tomini Bay coast (Fig. 1B). The tags that surfaced offshore were located over different slope areas of the shallow ridge (< 500 m) that separates two deeper page 6 of 16Zoological Studies 62:46 (2023) © 2023 Academia Sinica, Taiwan basins that are > 1500 m deep. The distance from release location to pop-up locations of the four tagged A. celebesensis ranged from 48.6–91.1 km (Fig. 1B). Behavioral statuses The changepoint analyses defined the different status-periods for tracks in four individual eels. There were either three (Eel 1 and 2) or five (Eel 3 and 4) changepoints from the start to the end of the recorded experienced water temperature data of each eel, resulting in the behavioral patterns being divided into two (Eel 1 and 2) or four (Eel 3 and 4) periods (Fig. 2). Pattern I shows irregular vertical movement that occurred until the changepoint when DVM behaviors started, which were defined as behavior pattern II (normal DVM) (Fig. 2C, D). Abnormal vertical behaviors and remaining at almost the same positions were designated as patterns IV and V, respectively (Fig. 2). However, the temperature data during the pattern IV period of Eel 1 showed a DVM pattern was only within a narrow depth range, so it was classified as abnormal. Behavior pattern III was predation (Fig. 2C), which will be described later. Finally, behavior pattern VI consisted of the tag suddenly sinking to the release trigger-depth (recorded maximum depth: 1653 m, Fig. 2C). Comparisons of diel vertical migration statuses for 2 Celebes eels The DVM behaviors of Eel 3 and 4 transitioned between deeper/colder layers in daytime (Eel 3: mean depth 444.7 m, Eel 4: 539.0 m, both ranges: 39–611.6 m; Eel 3: mean water temperature 9.1°C, Eel 4: 7.7°C, both ranges: 7–13.1°C; Table S1, including a spike-like ascent by Eel 3 after sunrise: Fig 3A) and shallower/ warmer layers in nighttime (Eel 3: 132.8 m, Eel 4: 112.4 m, both ranges: 10–212 m; Eel 3: 20.6°C, Eel 4: 23.4°C, both ranges: 14.3–29.3°C; Table S1, including several spike-like ascents by Eel 3: Fig 3A), with clear daily rhythms (Fig. 3). The analysis periods of the 2 eels that showed DVMs were performed for 21 days (Eel 3: 8 days around full moon, Eel 4: slightly > 13 days from new moon to first quarter moon, Fig. 2C, D). The detailed analysis periods of the DVM of Eel 3 and 4 for correlations between nighttime swimming depth and lunar age had > 15 days. Swimming depths and experienced water temperatures of Eel 3 and 4 each had significant differences among their three phases of daytime, ascent or descent, and nighttime (all p < 0.001; Fig. 4) but were not different between ascent and descent (all p > 0.16; Fig. 4). Furthermore, the swimming depths and experienced water temperatures of Eel 3 and 4 had significant differences between each of their 4 phases of daytime, ascent, nighttime and descent (all p < 0.001, swimming depths: figure 4A vs B, experienced water temperatures: Fig. 4C vs D). The swimming depths and experienced water temperatures of Eel 4 in daytime were deeper and colder than Eel 3 and the swimming depths and experienced water temperatures of Eel 3 in nighttime were shallower and warmer than Eel 4 (Fig. 4). Nighttime behavior and lunar cycle Eel 3 and 4 showed several patterns of nighttime swimming depths that were likely related to their locations, the lunar cycle, and whether the moon was present in the sky or absent. Their swimming depths and experienced water temperatures each had significant differences between with-moon and withoutmoon in the sky (all p < 0.001, Fig. 5, Table S2). Eel 3 showed the most distinct changes in swimming depths and experienced different water temperatures at night compared with Eel 4. Eel 3 swam deeper than Eel 4 at night with-moon and swam shallower than Eel 4 at night without-moon (all p < 0.001, Fig. 5A vs B, Table S2). As a result, Eel 3 experienced cooler water temperatures than Eel 4 at night with-moon and experienced warmer water temperatures than Eel 4 at night without-moon (all p < 0.001, Fig. 5C vs D, Table S2). There were significant linear relationships between swimming depths or water temperatures with the moon in the sky and the lunar age for the two eels (rho = 0.28: swimming depths and rho = -0.19: water temperatures, all p < 0.05, Fig. 6A, C). However, there were no significant linear relationships between swimming depths or water temperatures without the moon in the sky and the lunar age for the two eels (rho = 0.21: swimming depths and rho = -0.26: water temperatures, all p > 0.05, Fig. 6B, D). Regardless of moon presence, the eels did not stay at the same depth both during night and day times and were constantly moving up and down in small depth increments (Fig 3A, B). Predation The range of temperature differences between ambient water temperature and tag-recorded temperatures of Eel 1, 2, and 4, when there was no evidence of experiencing a predation event, was from -3.6 to +3.5°C (Fig. S3A, B, D). Therefore, if a tagged eel showed a magnitude of temperature difference that exceeded ± 4°C, it was suggested to be ingested by a predator with heat conservation ability. According to this, Eel 3 was estimated to be ingested by a predator with heat conservation ability, because this tag recorded temperature differences of -9.3 to +15.3°C, Figs. 7, page 7 of 16Zoological Studies 62:46 (2023) © 2023 Academia Sinica, Taiwan S3C). The change point determination analyses of the sudden changes in the vertical distribution behaviors of Eel 3 suggested that it was ingested by a predator when swimming at 197.5 m at 22:59 on the night of 4 March 2010 (Fig. 7). There were clear differences between the ambient and experienced temperatures of Eel 3 before, during and after the predation event (Fig. S3C). Furthermore, the tag-recorded temperatures of Eel 3 before and during predation were distributed within and then out side of the range that was used as the determination criteria for predation events (Fig. 7). Therefore, the PSAT of Eel 3 was apparently recording the peritoneal cavity temperatures of the endothermic predator from ingestion to egestion (Fig. 7). The predator of Eel 3 repeated vertical migrations from near the surface to as deep as 405 m each day for four days (Fig. 7). Since the depth records of this tagged eel showed that the PSAT fell directly to 1800 m just before tag surfacing, it is presumed that the tag was ejected by the predator and the tag float and weighted harness were separated after the tag sank to that depth (Fig. 7). DISCUSSION Diel vertical migration of Celebes eels in their spawning area The present study released PSAT-tagged A. celebesensis silver eels at the edge of their Tomini Bay spawning area of eastern Sulawesi Island, Indonesia. These silver eels were caught at the outlet area of Lake Poso, where they have been found to have more Fig. 3. The swimming depths (blue lines: A, B) and experienced water temperatures (red lines: C, D) during the diel vertical migrations (stage II) of Eel 3 (left panels: A, C) and Eel 4 (right panels: B, D) with each daily data set plotted together. The swimming depths of each eel are separated into four phases of daytime, ascent, nighttime, and descent. 700 600 500 400 300 200 100 0 0 5 10 15 20 25 30 0 3 12 15 21 912 18 6 0 3 12 15 21 912 18 6 Swimming depth (m) Time of day (A) (C) (B) (D) Eel 3 Eel 4 Experienced water temperature (°C) Nighttime DaytimeDaytime SunriseSunset Ascent Descent page 8 of 16Zoological Studies 62:46 (2023) © 2023 Academia Sinica, Taiwan advanced gonadal development levels than the A. marmorata silver eels that also migrate out of the lake (Hagihara et al. 2012; Arai 2014), but spawn in the western North Pacific (Kuroki et al. 2009). Catches of small larvae of A. celebesensis indicate that the Celebes eels spawn within Tomini Bay (Aoyama et al. 2003 2018), so the present study is the first study to record the DVM of A. celebesensis within their spawning area while having more advanced levels of reproductive development. Two of the Celebes eels released into Tomini Bay showed continuous patterns of very clear and regular DVM behaviors for more than a week, which were the same type of DVM behaviors seen in previous tagging studies on anguillid eels. The two eels showed some minor differences in their depths occupied during their DVMs, partly due to different release times in relation to the lunar cycle, but they both mostly stayed between about 50–200 m at night. Eel 3 mostly moved down to maximum depths where the temperature was 8–9°C (~440–520 m) and Eel 4 moved down to a very consistent 7–8°C (~520–610 m) during the day. Both eels gradually swam deeper with sun altitude during the day, indicating that they appeared to be responding to light intensity even at depths of about 400–600 m. These features of DVM behaviors have been observed in some other tropical and temperate anguillid species studied with PSAT tags (Schabetsberger et al. 2015 2021; Higuchi et al. 2021; Watanabe et al. 2020). There were significant relationships between Fig. 4. The swimming depths and experienced water temperatures of each eel during the four phases of daytime (Dy), ascent (As), nighttime (Nt), and descent (Ds). Blue box plots (A, B) show comparisons of the swimming depths and red box plots (C, D) show experienced water temperatures during each of the four phases for Eel 3 (left panels: A, C) and Eel 4 (right panels: B, D). Dy As Nt Ds 0 100 200 300 Eel 3 400 500 Swimming depth (m) Phase 600 700 30 25 20 15 10 5 Dy As Nt Ds Eel 4 (A) (B) (C) (D) Experienced water temperature (°C) page 9 of 16Zoological Studies 62:46 (2023) © 2023 Academia Sinica, Taiwan Wessel P, Smith WHF, Scharroo R, Luis J, Wobbe F. 2013. Generic mapping tools: improved version released. Eos Trans Am Geophys Union 94:409–410. doi:10.1002/2013EO450001. Westerberg H, Sjöberg N, Lagenfelt I, Aarestrup K, Righton D. 2014. Behaviour of stocked and naturally recruited European eels during migration. Mar Ecol Prog Ser 496:145–157. doi:10.3354/ MEPS10646. Wouthuyzen S, Aoyama J, Sugeha HY, Miller MJ, Kuroki M, Minegishi Y, Suharti SR, Tsukamoto K. 2009. Seasonality of spawning by tropical anguillid eels around Sulawesi Island, Indonesia. Naturwissenschaften 96:153–158. doi:10.1007/ s00114-008-0457-x. Wright RM, Piper AT, Aarestrup K, Azevedo JMN, Cowan G, Don A, Gollock M, Ramallo SR, Velterop R, Walker A, Westerberg H, Righton D. 2022. First direct evidence of adult European eels migrating to their breeding place in the Sargasso Sea. Sci Rep 12:1–6. doi:10.1038/s41598-022-19248-8. Supplementary materials Fig. S1. A higher resolution bathymetry map showing release (yellow star) and pop-up locations of the four tagged Celebes eels (red circles) in Tomini Bay. (download) Fig. S2. The depth changes of water temperature (red dots), salinity (blue dots), sigma-t (orange dots), and chlorophyll-a (green dots) in the upper 1000 m at the CTD hydrographic station in Tomini Bay (black circle in Fig. 1) during KH-09-5. These data of water temperature were used to compare to experienced swimming depth and water temperature from the PSATs data. (download) Fig. S3. Differences between tag-recorded and observed environmental temperatures in the CTD hydrographic cast for Eel 1 (A), Eel 2 (B), Eel 3 (C), and Eel 4 (D). The gray shaded areas show the same temperatures ranges (-4 to 4°C). (download) Table S1. The means and standard deviations of swimming depths and experienced water temperatures separated into four phases of daytime, ascent, nighttime, and descent during DVM behavior (II) of Eel 3 and 4. (download) Table S2. The means and standard deviations of swimming depths and experienced water temperatures with moon and without moon of Eel 3 and 4 that showed DVM behavior (II) in nighttime. (download) page 16 of 16Zoological Studies 62:46 (2023)