Photo-ID and telemetry highlight a global whale shark hotspot in Palawan, Philippines

The Philippines is home to the second largest known population of whale sharks in the world. The species is listed as endangered due to continued population declines in the Indo-Pacific. Knowledge about the connectivity within Southeast Asia remains poor, and thus international management is difficult. Here, we employed pop-up archival tags, data mining and dedicated effort to understand an aggregation of whale sharks at Honda Bay, Palawan, Philippines, and its role in the species' conservation. Between Apr and Oct 2018, we conducted 159 surveys identifying 117 individual whale sharks through their unique spot patterns (96.5% male, mean 4.5 m). A further 66 individual whale sharks were identified from local operators, and data mined on social media platforms. The satellite telemetry data showed that the whale sharks moved broadly, with one individual moving to Sabah, Malaysia, before returning to the site <1 year later. Similarly, another tagged whale shark returned to the site at a similar periodicity after reaching the Malay-Filipino border. One individual whale shark first identified in East Kalimantan, Indonesia by a citizen scientist was resighted in Honda Bay ~3.5 years later. Honda Bay is a globally important site for the endangered whale shark with connectivity to two neighbouring countries, highlighting the need for international cooperation to manage the species.

Movements from photo-iD. One  More interestingly, individual ID-051, a 3 m male, was first identified in East Kalimantan, Indonesia on Dec 29 2013 by a citizen scientist who submitted the encounter to Wildbook for Whale Sharks, and was resighted in www.nature.com/scientificreports www.nature.com/scientificreports/ Honda Bay on Oct 19 2018 whilst on survey. On this occasion the shark was visually estimated to measure 4 m T L . This represents the first international whale shark match between the Philippines and Indonesia using photo-ID.
Lagged identification rate and residency. Model H (Table 1) best fitted the empirical data through a combination of mortality, emigration, reimmigration and residency parameters. The LIR declined quickly following initial identification (Fig. 2) and continued to decline before rising after mean 92.9 days, then again after mean 362.0 days, and 738.3 days, and never quite reaching zero after 1,539.2 days. The model estimated a mean 41.1 ± 13.5 whale sharks at the study site at any one time, residing a mean 6.4 ± 2.9 days within the study area, whilst spending 58.2 ± 25.5 days outside. Mortality or permanent emigration rate was estimated at 0.00097 ± 0.00056.

Whale shark tracking.
Only five of the ten miniPAT tags popped up and transmitted any data. The five sharks were tracked for a mean 136.6 ± 51.3 days (range 61-200 days). Of these 5, data transmission was very limited (summary in Table 2), and none were physically retrieved. Four of the 5 tracked individuals (P-1346, P-1128, P-1126, P-1123) moved south between deployment and pop-up date (Fig. 3). Given the limited data transmitted, and the spatial accuracy of these tags (~50 km), the movement presented here is an estimate and not absolute (an animation with 50%, 95% and 99% confidence intervals is presented as Supplementary Video 1). However, it is clear that these tracked animals moved south during the overall tracking period (Fig. 3). Individual P-1125 was tracked for a total of 200 days and transmitted more complete data packages of its locations. The animal appears to have first moved northeast towards Cuyo Islands in the northern Sulu Sea, before returning south via the Cagayancillo archipelago and TRNP, where the tagged popped-up southeast of Honda Bay (Fig. 3).
At least three of the five tracked whale sharks returned to Honda Bay. Two from the animals that transmitted data (P-1125 and P-1128; Fig. 3), and P-1122 which was resighted by citizen scientists on the Oct 30 2017 still carrying a heavily fouled tag (Table 3). Individual P-1396 was resighted in Honda Bay during whale shark surveys on the Jul 7 2018 not carrying the tag and no obvious scarring to the naked eye.  (Fig. 3). Individual P-1128 appeared to have been following a similar path, but the tag popped up northeast of the Turtle Islands, near the Malay-Filipino border. Both P-1128 and P-1346 were resighted in Honda Bay on May 23 and 28 2018 respectively during whale shark surveys (Table 3)  n/a n/a n/a n/a n/a n/a n/a n/a 39720 P-1122 M 3.5 19-Jul-17 n/a n/a n/a n/a n/a n/a n/a n/a 39741 P-1124 M 6.0 20-Jul-17 n/a n/a n/a n/a n/a n/a n/a n/a 39742 n/a n/a 5.0 22-Jul-17 n/a n/a n/a n/a n/a n/a n/a n/a 39748 P-1127 M 2.5 22-Jul-18 n/a n/a n/a n/a n/a n/a n/a n/a Table 2. Summary of whale sharks tagged in Honda Bay, Palawan, with pop-archival tags in July 2017. *Distance (km) reflects the minimum straight line between deployment and pop-up location. Tags #39701, #39720, #39741, #39742 and #39748 failed to transmit any data. Diving. To estimate the vertical velocity during descent, we selected dives that were characterised by a clear V-shape, and where sharks went from shallower water (<50 m) to >500 m deep. Given the gaps in the time-series data transmitted (intervals of 7.5-10 min), we present estimates that are indicative of the minimum vertical velocity of descent for sharks P-1125, P-1126 and P-1128. A total of 11 dives >500 m were recorded amongst these three sharks (range 516-1,009.5 m). The mean descent vertical velocity was estimated at 0.32 ± 0.29 ms −1 (range 0.004-0.988 ms −1 ). For individual P-1128, all dives >500 m (n = 7) took place between 04:40 pm and 07:40 am, whereas individual P-1126 only dive >500 m took place at 12:52 pm, and P-1125 dives were at 07:00 am, 03:30 pm and 10:00 pm.

Model Name Parameters ΔQAIC
A total of 189 deep dives (>200 m) were performed by sharks P-1125, P-1126 and P-1128 between Jul 22 2017 and Jan 18 2018. Two thirds (67.7%) of these dives were performed between 06:00 pm and 05:59 am, with the remainder of deep dives (32.3%) taking place between 06:00 am and 05:59 pm, daylight hours in the region. Most deep dives were transmitted from individual P-1125 (98), consistently performing a third (34.7%) of deep dives during between 06:00 am and 05:59 pm. Some periodicity was observed in the temporal deep diving of all sharks, with marked absences of deep dives ( Supplementary Fig. 3).

Discussion
Honda Bay, Palawan, is a globally important whale shark hotspot with 117 individuals identified in a single season through dedicated photo-ID, and a further 66 identified through data mined from social media platforms and other sources. This small juvenile, male-dominated aggregation appears to occur seasonally between Apr and Oct, with some individuals returning yearly to feed on small fishes and krill. Through photo-ID, individuals were matched to other sites in the Philippines. Combined with satellite telemetry, we report the first international return movement of whale sharks in Asia, with one individual moving to Lahad Datu in Malaysian Borneo through the Tawi-Tawi Kinabatangan strait, and returning to Honda Bay less than a year later. A second shark was also tracked to the edge of the Malay-Filipino border in the south Sulu Sea, and the animal returned to Honda Bay ca. a year later. Two other tags also moved to the southern Sulu Sea, and the fifth tag that transmitted   www.nature.com/scientificreports www.nature.com/scientificreports/ travelled north before returning to the central Sulu Sea. We also report the first photo-ID match between the Philippines and Indonesia by that of a small male identified in East Kalimantan, Indonesian Borneo, in late 2013, and resighted in Honda Bay during our 2018 seasonal work.
Photo-ID was effectively employed to identify 183 individual whale sharks at Honda Bay, with a significant contribution (36%) from the general public as citizen scientists. The slow-swimming, and relatively benign nature of the whale shark makes it easy to photograph these animals underwater 28 , however, the quality of the photographs needs improving. The number of identification images usable from the number of encounters obtained through other data sources was relatively low (66 out of 230). An educational process to the tourists, perhaps as part of the interaction briefing, about the importance of collecting photo-ID data and how they can independently submit their encounters to Wildbook for Whale Sharks, would maximise data collection 15 . Not only can the data be used for mark-recapture models, but it can also provide insight into the size, sex and threats (e.g. propeller cuts, rope entanglement) individual whale sharks are exposed to.
Whale sharks visiting Honda Bay were mostly juveniles (mean 4.5 m total length), similar to that observed in Djibuti (3.7 m) 47 , Saudi Arabia (4.3 m) 21 , Bahia de la Paz, MX (4 m) 17 , Taiwan (4.6 m) 48 and Christmas Island (4.6 m) 49 . They are however considerably smaller than whale sharks observed elsewhere in the Philippines as estimated visually (Cebu, 5.5 m 10 , Southern Leyte, 5.7 m 27 , Donsol, 6.5 m 19 ), and at other Indo-Pacific aggregations like Mozambique (6.7 m) 50 , Qatar (6.9 m) 7 or the Galapagos Islands (10.4 m) 41 . It appears Honda Bay might play a role in the developmental stage of these small juvenile whale sharks, particularly for males (96.5% of identified individuals). Only one adult male was observed, suggesting this is unlikely a mating ground for the species, but rather a targeted or opportunistic foraging ground.
Honda Bay is a globally important site for whale sharks, with maximum likelihood methods estimating a mean ~41 whale sharks at any one time within the survey area. Using similar methods, Araujo et al. 27  Whale sharks were encountered active feeding in small tuna boils, targeting the same prey the tunas were. We identified the main species as the Philippine anchovy Encrasicholina oligobranchus, although other species were probably present throughout the season. On occasion, the sharks were encountered feeding on krill species together with other filter feeders including Mobula birostris, M. kuhlii and M. japanica, and Balaenoptera edeni (Authors, unpub. data). These co-occurrences with other megafauna highlight a high degree of prey availability in the area. It is yet unclear if the sharks visit Honda Bay to specifically forage on small fishes or krill, or perhaps both. Their low resighting rate within season, and the low estimated residency as calculated through modified maximum likelihood methods, suggests these whale sharks might visit Honda Bay to opportunistically forage on available prey, before moving elsewhere. Interestingly peaks in productivity as inferred from chl-α are highest between Nov and Feb, like in Lahad Datu, Malaysia, to the south, and like in the Bohol Sea to the east (Authors, unpub. data 54 ), and thus it remains unclear the main drivers for these long-distance movements. Unlike other coastal sites where whale sharks are known to spend considerable amounts of time as explained through their lagged identification rate (e.g. ~31 d, Mafia Island, Tanzania 52 ~27 d, Pintuyan, Southern Leyte 27 ; ~50 d, Donsol, Luzon 19 ), whale sharks visiting Honda Bay appear to be short-term visitors as that observed in Honduras (~12 d) 53 , the Red Sea (~12 d) 21 or Mozambique (~5 d) 52 . Further work into their detailed habitat use whilst in Honda Bay will elucidate this, and perhaps molecular approaches to better understand their foraging ecology 52 . www.nature.com/scientificreports www.nature.com/scientificreports/ Only five of the ten tags deployed transmitted data. This failure has been reported in pop-up archival tags before with studies reporting ~50% transmission success 55 . This is a considerable rate of failure given the cost of these tags, satellite time, field costs involved, and the invasiveness on the target animals. Individual P-1122 was resighted 103 days after tagging still carrying the tag. It looked heavily fouled and that tag failed to report altogether. It could perhaps be a fouling issue by which the tags become too heavy to pop-up, or even to detach altogether from the animal. Similarly, individual P-1396 was resighted 350 days after tagging (tag never reported) but with no tether and no obvious scar to the naked eye, suggesting the tag, tether and anchor might have been pulled out. Given the cost of these tags and the animal welfare implications, careful considerations should be made given their poor success at least in this region. Nonetheless, results yielding conservation and management implications can arguably outweigh invasive methods employed such as drag from fouled tags or tag deployment. Araujo et al. 45 deployed towed SPOT5 tags on 17 juveniles and obtained basin-wide connectivity data as well as more detailed habitat use data for coastal areas, albeit a shorter tracking time (~64 days). Perhaps an adaptation of these tags (e.g. dorsal fin clamp 56 , or dorsal fin tether 57 ) might yield better results.
Time-at-temperature reported herein is similar (66-77%) to that reported elsewhere for juvenile whale sharks in the tropics 38,39,45 , yet warmer than that reported for whale sharks at cooler aggregations 40,55,58 . Interestingly, at localities with higher average water temperatures, size at which 50% of males reach maturity was reportedly smaller (7.0 m, Gulf of Mexico 8 ; 7.3 m, Qatar 7 ; 6.8 m, Philippines 19 ) than that at more temperate localities (8.1 m, Ningaloo Reef 59 ; 9.2 m, western Indian Ocean 50 ). Perhaps juvenile sharks spending the majority of their time in warmer, steadier conditions, allow for faster growth and development 60 . Further investigation into the growth, temperature and the effects of global ocean warming on this endangered species are necessary.
The mean vertical velocity estimated herein for dives >500 m deep was 0.32 ms −1 (max. 0.99 ms −1 ) for 3 sharks of mean 4 m T L . Tyminski et al. 61 reported a mean of 0.68 ms −1 in the Gulf of Mexico for 5 sharks ranging from 6-8.5 m T L , with a maximum of 1.83 ms −1 descent vertical velocity on a 7.5 m individual. Arguably, these sharks were ca. double in size from the ones reported herein which could explain the slower descent velocity. The difference in descent vertical velocity between the two studies could also be a consequence of the sampling rate employed where here we used 7.5-10 min intervals, and Tyminski et al. 61 employed 3 s intervals. This discrepancy in the sensitivity of the data could also explain the slower descent speeds reported here. However, using 5 s intervals at Oslob, Cebu, vertical velocities were slower to those reported herein on similar sized sharks (Authors, unpub. data). It is possible that the speed of descents is dependent on the reason for these deep dives and, for example, a predatory avoidance dive (e.g. leatherback turtle 62 ) will likely be at a higher speed than an exploratory dive 63 .
Time  58 and with Tyminski et al. 61 when the sharks were inshore of the Yucatan Peninsula. Interestingly there were prolonged periods of no deep-diving behaviour at all (Supplementary Fig. 3) that is likely associated with coastal, shallow habitats where there are no adjacent waters >200 m deep, such as Honda Bay. The diel vertical movement is believed to be linked with foraging opportunities 2,58 , in this case, when the sharks are in Honda Bay without access to deep-waters, they stay on the shelf, yet when they leave they likely perform more regular deep dives, and this could explain the periodicity in deep-diving behaviour reported here.

conclusions
Our results highlight Honda Bay as a global hotspot for the endangered whale shark where they visit seasonally to feed on small fishes and krill. We used archival tags coupled with photo identification to understand the movements of whale sharks from this area, and report the first international return migration in Asia using these techniques. We demonstrate the usefulness of monitoring social media platforms to generate data on endangered species, and encourage the education of tourists at whale shark hotspots through citizen science programmes to aid monitoring efforts 15 . Coupled with the occurrence of other threatened, endangered and protected (ETP) species, Honda Bay has been declared a Marine Key Biodiversity Area. Understanding critical habitats for whale sharks is one conservation priority for the species 12 , and here we provided evidence that supports Honda Bay as an important habitat, and the strait between Tawi-tawi (PH) and Lahad Datu (MY) as an important migratory corridor for the species.
The whale shark has been protected in the Philippines since 1998 (FAO 193), and in Malaysia since 1999 (Fisheries Regulation of 1999), with a general understanding that poaching is low. However, concerns remain about the illegal take of these animals in the region, in light of fisheries operating in the south of China 13 that probably have extended fishing grounds into Malay and Filipino waters. Although juvenile whale sharks might not move as much as originally thought 15,52 , they still undergo long-distance movements, or move regionally crossing international boundaries 27,39,40,45,63 as also reported herein. The results presented here that confirm the movements between the Philippines and Malaysia, and Indonesia, therefore add to this connectivity evidence, and further emphasizes the need for international cooperation to manage this Endangered species. Our results support the objectives of the Coral Triangle Initiative, the Sulu-Sulawesi Seascape Project 66 , and the Concerted Actions for Whale Sharks under CMS (UNEP/CMS/Concerted Action 12.7, 2017) amongst others, to enhance the management and conservation of the whale shark through trilateral collaboration between Indonesia, Malaysia and the Philippines. The whale shark connectivity corridor identified here in the south Sulu Sea including Tawi-tawi and Jolo (PH), Sandakan, Kinabatangan, Kunak and Lahad Datu (MY), appears to be a key area of concern for the species, as has also been highlighted for other ETP species like marine turtles. A trilateral Boat-based surveys and photo-iD. Dedicated boat-based surveys in Honda Bay were conducted between April 23 rd and October 21 st 2018. We employed two survey platforms to find whale sharks in the Bay: small outriggered pumpboats (7.9 m, 10 hp), and large tourist bangkas (15 m, 90 hp), similar to those described by Araujo et al. 27 . Pumpboat surveys started from Barangay San Jose (9.7982N, 118.7724E), whereas bangka surveys operated from the Puerto Princesa City baywalk (9.7441N, 118.7301E; Fig. 1). Whale sharks were haphazardly searched for within the central and south western part of the Bay (see Fig. 1) when sea state conditions were <Beaufort 3, and swell was <1 m. Birds and fish boils were used as sighting cues to aid in finding the whale sharks. Upon encountering a whale shark, researchers recorded the location on a handheld GPS, entered the water and collected photographic identification of any whale sharks in the water. photo-iD. Photographs of the left and right flank were taken for identification purposes. Left identification images were prioritised over the right, as it's the current international standard. Only images of sufficient standard that allowed visual identification of the left and right flank were utilised. Whale shark sex was determined by the presence (male) or absence (female) of claspers as confirmed from photographs. No sex was assigned if there was no photographic evidence to confirm it. Maturity in males was externally assessed following Norman and Stevens 59 and Rohner et al. 50 . Maturity in females could not be visually determined. The size of the whale sharks was visually estimated based on items of known length within proximity (i.e. boats and/or other swimmers). Although this method carries an inherent level of error [ref. 70 , it can be used to determine the size class of the study animals 10 . Whale shark behaviour was noted during each encounter and when a whale shark was actively feeding and conditions permitted, a sample of the main prey was collected.
Data mining. Photo-identification data collected by members of the general public were used to understand whale shark movements. Data was extracted between April and August 2018 from popular social media platforms (i.e. ©Facebook, ©Instagram, ©Twitter and ©YouTube). Systematic searches for specific keywords was conducted using Boolean operators to increase positive matches. Each video or photograph extracted was further processed for identification purposes and inputted into a database including date posted, user, quality and whether it could be used for identification. A second researcher validated extracted and processed images, and visually matched them against a localised database. Where date of encounter could not be verified from the user, the date of upload was used as the encounter date; an assumption necessary to maximise data collection. photo-iD validation. Identification images from dedicated photo-ID effort and different data sources were processed and sorted into separate folders, each corresponding to a different individual whale shark. These were visually matched against a localised database. A second researcher working independently confirmed, sorted and identified images visually. A third researcher, independent from the first two, ran an identification photograph of each individual through the program I 3 S (http://www.reijns.com/i3s) 71 containing identification images from all sites in the Philippines (Authors, unpub. data). New, unidentified individuals were uploaded onto the online whale shark database 'Wildbook for Whale Sharks' (www.whaleshark.org). The presence of each individual whale shark was recorded on a spreadsheet.
Lagged identification rate. Photo-ID data extracted from data mining sources and collected through dedicated effort was used to calculate the lagged identification rate (LIR), defined as the probability that an individual will be resighted at the study site after a certain time lag 26 . The LIR was modelled using the 'Movement' module in the program SOCPROG 2.7 72 , and was used to estimate residency, population size and mortality or permanent emigration. Models were tested for a combination of open, closed, emigration, reimmigration and mortality population parameters (Table 1). This approach operates under the assumption that identified individuals have equal probability of recapture over time, no marks change or can be lost, and individuals can leave and re-enter the site. The quasi-Akaike information criterion (QAIC), was used to account for over-dispersion of the data, and select the best-fitting model 73 .
The best-fit model was bootstrapped for 100 repetitions to generate standard errors and 95% confidence intervals 74 . (2019) 9:17209 | https://doi.org/10.1038/s41598-019-53718-w www.nature.com/scientificreports www.nature.com/scientificreports/ Satellite tagging. Ten whale sharks were opportunistically tagged with pop-up archival tags (miniPATs, Wildlife Computers, Washington, USA) between July 19 th and 23 rd 2017 in Honda Bay, Palawan (Table 2). Upon encountering a whale shark, we used a Hawaiian-sling spear pole to attach the tag to the whale sharks' left flank, between the base of the dorsal fin and the first lateral ridge. The tag was attached with a titanium dart that is inserted 10 cm into the subdermal tissue of the shark, tethered by a stainless steel line. It is unclear how long darts remain embedded for, but 5 tags recovered by Araujo et al. 44 using the same darts were attached to the tethers upon retrieval. Darts were sharpened and cleaned with 95% ethanol prior to deployment. Tags were deployed on first attempt to minimise disturbance to the animals. Tags were not coated with antifouling. Tags were programmed to pop off after pre-determined intervals of 150, 180 and 300 days ( Table 2). Tags recorded on-off-on histograms for depth and temperature for 12 hrs for eight days on, four days off, and four days on, with depth bins of <5 m, 5-15 m, 15-30 m, 30-60 m, 60-100 m, 100-200 m, 200-300 m, 400-500 m, 500-700 m, 700-1,000 m, 1,000-2,000 m, and temperature bins of <0 °C, 0-5 °C, 5-10 °C, 10-15 °C, and then in 2.5 °C increments until 32.5 °C, and then 32.5-45 °C. MiniPATs collected light-level data which was used to determine location. No tags were physically recovered in order to retrieve the full time-series dataset in this study. For those tags that transmitted sufficient depth time-series data, we calculate the number of deep-dives (>200 m) during day and night, and estimate the vertical velocity of extreme dives (>500 m) based on pre-set intervals of 7.5-10 min.
All data retrieved from the tags was transmitted through the ARGOS satellite system and downloaded from Collecte Localisation Satellites (www.argos-system.cls.fr). Wildlife Computers' DAP3 processor was used to estimate light-level data into geographic locations of the tags deployed. This processor uses a Hidden Markov Model (HMM) with a multi-step algorithm at a 0.25° grid size with light levels, bathymetry data (ETOPO-1) 75 , tagging location (GPS), pop-up location (first ARGOS Location Quality 3), any other GPS location assigned by the user, sea surface temperature (SST; NOAA OI SST V2, http://www.esrl.noaa.gov/psd), and a mean animal speed of 3.5 km h −1 47,76 . Light-level locations had a mean radius error of ~50 km, and 50%, 95% and 99% confidence intervals were generated (Wildlife Computers, 2018), presented herein as Supplementary Video 1.

Data availability
All identification data is hosted on the online database 'Wildbook for Whale Sharks' (www.whaleshark.org). Tag data will be made freely available upon manuscript publication.