Supplementary Materials for

Transcription

1 advances.sciencemag.org/cgi/content/full/4/6/eaat2504/dc1 Supplementary Materials for The economics of fishing the high seas Enric Sala, Juan Mayorga, Christopher Costello, David Kroodsma, Maria L. D. Palomares, Daniel Pauly, U. Rashid Sumaila, Dirk Zeller This PDF file includes: Published 6 June 2018, Sci. Adv. 4, eaat2504 (2018) DOI: /sciadv.aat2504 Supplementary Materials and methods fig. S1. Distributions of vessel characteristics, high-seas fishing effort relative to total fishing effort, and nonlinear relationships between length, tonnage, and engine power. fig. S2. Vessel encounters on the high seas (2016). fig. S3. Fishing effort by gear type. fig. S4. Effort versus catch. table S1. High-seas fishing fleets and effort by country (2016). table S2. High-seas fishing fleets and fishing effort by gear type. table S3. AIS coverage of high-seas fleet and scale factors by country, Regional Fisheries Management Organization, gear type, and vessel size in table S4. Average labor costs per day by country, gear type, and vessel size class. table S5. High-seas costs, catch, revenue, subsidies, and profits by country. table S6. High-seas fishing costs by reefers and bunker vessels per flag state. table S7. Summary of high-seas economics by country, gear type, and FAO region (scaled results). References (51 55)

2 Supplementary Materials and methods The characteristics of the high seas fleet We characterized the high seas fleet by identifying each vessel's flag state, overall length, gross tonnage, engine power, auxiliary engine power, crew size, design speed, and specific fuel consumption. When available, these characteristics were obtained from official vessel registries, and data gaps were filled using machine learning and regression models. For AIS fishing vessels the fraction of vessels with registry information are: flag state (98%), gear type (78%), length (82%), tonnage (73%), engine power (80%), auxiliary engine power (7%), and crew size (60%). All of Indonesia s VMS vessels had information on gear type, length, and tonnage and 68% had engine power. For reefers and bunkers, we had complete information on flag, type, length, and tonnage; 75% had engine power and 56% had crew size. Flag States Flag states for 97% of the fishing vessels, and 100% of the bunkers and reefers, were obtained by combining official registries and the AIS MID-code (7). Remaining data gaps - representing ~2% of all AIS positions - were manually inspected and filled using online databases to assign correct flag states. Only 28 vessels remain with unknown flag states. For vessels with non-sovereign flag countries we added and used their sovereign flag state (e.g., France for Reunion). With the exception of 15 vessels flagged in Belize, Panama, and El Salvador whose ownership was clearly traced to Spanish companies, we used flag states for the analysis. In total, 72 flag states are represented on the high seas fishing fleet, with only 5 countries - China, Taiwan, Japan, Indonesia, and Spain - accounting for 72% and 78% of the fishing vessels and the time spent on the high seas in 2016, respectively. Vessels with unknown flag states accounted for less than 1% of vessels and time at sea. For bunkers and reefers, 25 flags are represented with five nations - Panama, Russia, Indonesia, Kiribati, and Liberia - accounting for 69% and 64% of the vessels and encounters on the high seas in 2016, respectively. Gear type To fill in gaps in vessel gear type, we used the inferred labels from GFW's convolutional neural net and for each vessel we selected the inferred label with the highest confidence level across five years ( ). The gear type for a subset of 123 fishing vessels, representing less than 1% of the AIS position, remains unidentified. For details on the methods and the accuracy of GFW s neural net see (9). Length, tonnage, engine power Data gaps for vessel length, tonnage, and engine power were filled using the inferred characteristics from GFW's neural net algorithms. For a small subset of fishing vessels whose length (n = 3), tonnage (n = 62) and engine power (n = 127) was not inferred by

3 GFW models due to relatively low activity (these fishing vessels represent less than 0.1% of all AIS positions), we used nonlinear regressions to fill data gaps. Auxiliary engine power Using available data on the auxiliary engine power of 1,156 distinct fishing vessels from the EU, covering all gear types except squid jiggers, we used conditional inference random forests to fill in data gaps for all GFW fishing vessels. We randomly split our sample into training (70%, 811 vessels) and testing sets, and trained a first random forest using vessel length, engine power, and gear type as predictor variables. Since the available data did not contain all possible gear types, a second random forest was ran without gear type as a predictor variable. Both random forests were run with 500 trees, 10-fold cross validation resampling with 5 repeats, and the RSME was used to tune the mtry parameter and select the optimal model. The final models had similar cross validation performance (RMSE 1 : 269.1, R 1 2 : 0.77, RMSE 2 : 280.8, R 2 2 : 0.74) and the variable importance metric suggests that vessel tonnage is the most important predictor of auxiliary engine power. Nevertheless, when compared with the testing set, the model with gear type had better out of sample performance (RMSE 1 = 242, RMSE 2 = 274) suggesting that, while gear type is relatively less important than tonnage for predicting engine power, is not an irrelevant predictor. Our results show that, on average, the auxiliary engine power of a fishing vessel is 31% of its main engine power. This estimate is lower but similar to what s been reported by the European Environmental Agency (39%). For fishing vessels of the high seas fleet, we find that the average auxiliary engine power represents 47% of the power of the main engine. For bunker and reefer vessels that operate in the high seas fleet, we had no available information on auxiliary engine power; thus, we use the average fraction for cargo vessels reported by the European Environmental Agency (24%) Crew Using a dataset of 3,840 vessels for which we had crew size information, we used conditional inference random forests to fill in data gaps for all GFW fishing vessels. We randomly split our sample into training (70%, 2,690 vessels) and testing sets, and trained a first random forest using vessel length, engine power, gear type, and flag state as predictor variables. Since the available data did not contain all possible combinations of gear type and flag state, we trained two more random forests excluding flag state, and excluding both flag state and gear type from the predictors. Similarly to the auxiliary engine power, all random forests were trained with 500 trees, 10-fold cross validation resampling with 5 repeats, and the RSME was used to tune the mtry parameter and select the optimal model. The results show a good fit in all cases, with the model that incorporates gear type and flag performing slightly better than the rest (RSME: 7.46, R 2 = 0.83). For bunkers and reefers we used a similar approach with a dataset of 179 vessels for which we had information on their crew size and ran two conditional inference random forests: 1) with vessel type (bunker or reefer), length, tonnage, engine power, and flag state and 2) excluding flag state. The first model performs better both in the cross

4 validation and the testing set validation and the vessel tonnage and type are the most important predictors (RSME: 4.72, R 2 = 0.54). Design speed The design speed of a vessel's main engine is used to determine its load factor at any given time and thus it's an important characteristic to quantify fuel consumption. However, this formation is very scarce. In some cases, maximum observed speed is used as a proxy, in others it is estimated from other vessel characteristics. We take the latter approach and use an estimated relationship between main engine power and design speed from a sample of 23,296 fishing vessels from the IHS fairplay database (Naya Olmer, ICCT, personal correspondence, April, 2017) S design = Engine Power Engine Power 2 (eq. S1) where engine power is in KW and design speed in knots (R 2 = 0.42; p value < 0.01). Our results show that the average design speed of all fishing vessels and high seas vessels is 11 and 11.8 knots, respectively. These estimates are very close to the average reported by the International Maritime Organization: 11.5 knots. Specific fuel consumption Country-level Specific Fuel Consumption parameters (SFC; gr/kw-hr) were available for China (280), EU member states (270), Iceland (250), Norway (250), South Korea (260), and Russia (250). For the remaining countries we use vessel length-specific factors: 240 for vessels <12 meters, 220 for vessels between meters, and 180 for vessels over 24 meters. For the auxiliary engines, we use the SFC reported by the EEA and for bunkers and reefers we use country-specific data when possible and fill in gaps with the lowest estimate provided by the EEA (203 gr/kw-hr) Fishing effort In 2016, the high seas fleet spent cumulatively 509,710 days in international waters and fished for 7.07 million hours. In terms of energy, this fishing effort corresponds to 6,743 million kw hours. Detailed methodology of AIS-inferred fishing effort can be found in (9). The high seas under the jurisdiction of the South Pacific Regional Fisheries Management Organization (SPRFMO) has the most fishing effort and one of the largest fleet, with more than 140,000 fishing days, over 1,100 vessels from 28 different flag states, and a fishing intensity of 45 kw hours per square kilometer. The WCPFC and IATTC follow with 136,000 (1636) and 104,000 (864) fishing days (vessels), respectively. In terms of

5 FAO regions, the Eastern Central Pacific (FAO region 77) is the region with the largest high seas fleet: 836 vessels from 19 different flag states. In turn it is also the region with highest fishing effort (~19 % of total). This is likely correlated with this region being the largest of all. Accounting for surface area, the FAO region with most fishing intensity becomes the Southwest Atlantic, followed by the Southeast Pacific and the Northwest Pacific. The time spent on the high seas ranges drastically, with some vessels spending a full 365 days at sea on On average, squid jiggers, pots and trap vessels, and drifting long lines spend significantly more time in the high seas than other vessels such as purse seiners or pole and line vessels. Encounters with reefer and bunker vessels An encounter is defined by an event in which two vessels remain within 500 meters of each other for longer than 3 hours while traveling at speed less than 2 knots at least 10 km away from any known anchorage. We filter these events for those when one the vessels was identified as a refrigerated cargo or bunker vessel and the other as a fishing vessel (51). For more details on the methodology see the report The Global View of Transshipment by Global Fishing Watch, available at In total, we observed of 1,559 encounters between these vessels and the high seas fleet, of which 1,386 take place in international waters. The flags with most reefer and bunker vessels are Panama, Kiribati, Liberia, and Russia and the high seas fishing fleets that most rely on them are Japanese, Taiwanese, South Korean, and Chinese long liners, as well as Chinese and Taiwanese squid jiggers, and Chinese and Russian trawlers. Most high seas encounters take place in the Eastern Central Pacific (FAO 77, 17%) and the Southwest Atlantic (FAO 41, 17%), followed by the Southeast Pacific (15%). In terms of RFMOs, most encounters take place in the WCPFC (34%), followed by the IATTC (28%), and the ICCAT (28%). Gaps in AIS We define a gap event as any time when a fishing vessel stops transmitting for more than 24 hours and where both the start and end of the gaps are further than 100 nautical miles from shore. This distance from shore cutoff serves to exclude gaps where the vessel turns off their AIS upon coming to port. Overall, total time associated with gaps represents ~16% of total time at sea. There is variability in the magnitude of these gaps across countries with Taiwanese and Chinese vessels having the largest median time spent in gaps: 30% and 18% respectively. Gear types also differ in the amount of gap time with squid jiggers and purse seiners being the worst offenders: 20% and 21%, respectively. More details on the magnitude of global ASI gaps can be found in (9). We did not account for the fact that some vessels may have acquired AIS part way through the year, and thus were not broadcasting for some of their activity in About 17% of the high seas vessels broadcast AIS for the first time in If these vessels are similar to the vessels that had AIS before 2016, that would suggest that are missing an addition 5-10%

6 of activity of the fishing fleet. This missing activity, if accounted for, would have the effect of increasing costs and thus further lowering profits. AIS coverage of high seas fleets An important consideration is estimating the fraction of the world's high seas fleet that we can observe from AIS data. This is a difficult task because clear points of reference of the number of fishing vessels for each country and, by gear type, that operate in distinct geographic regions of the high seas are, to our knowledge, not available. Nevertheless, to address this issue, we focused on the most important fishing nations and did exhaustive searches on RFMO registries, government websites, and scientific literature, for the latest estimates of fleet capacities. Often, information is aggregated by RFMO or geographic region, and to separate high seas vs. within EEZ fishing vessels we had to make assumptions about the behavior of the missing fleet. In general, we assume that the fraction of a fleet we do not see, behaves with the same intensity and spatial patterns as the ones we observe. Here we summarize our review of reference points our findings and report how we use them to estimate the fraction of the high seas fleet we cannot see with AIS. (table S4) China The 2016 China Fisheries Yearbook reports that the nation's distant water fleet was comprised of 2,460 and 2,512 vessels in 2014 and 2015, respectively. This is a significant increase from the 1,989 vessels reported to operate in 2011, and reflects China's plan for expansion of its distant water fleets. China's Ministry of Agriculture s five-year plan to develop the nation's DWF (due in 2010) aimed to increase the DWF to 2,200 boats including 840 boats on the high seas (52). This target was not reached, as the 2011 China Fisheries Yearbook report a DWF of 1,899 vessels. Lastly, a report published in 2008 states that the DWF in 2007 was composed of 500 squid jiggers, about 400 tuna boats, approximately 800 trawling boats, and over 100 were purse seiners and other fishing vessels. Another source of information are the RFMOs. China's 2016 report to the WCPFC Commission reports 429 long liners and 20 purse seines actively fishing in the region in The ICATT Biennial report , reports that the number of vessels from China operating in the Atlantic Ocean increased from 13 in 2014 to 24 in 2015 to 41 in All of these vessels are longliners, targeting bigeye tuna, bluefin tuna, and sharks. Lastly, the 2016 Report of the 21st Session of the Indian Ocean Tuna Commission, IOTC reports that China's active capacity for tropical tunas is 54 vessels and 13 vessels for albacore and swordfish and China's National report to the RFMO reports 53 active vessels in Given the available information, we can make the following statements about China's AIS coverage: We observe 2,170 Chinese vessels fishing on foreign EEZ and international waters in This suggests we see 86 % of the DWF reported in 2015.

7 We observe 838 fishing vessels in the high seas fleet. This is very close to the 840 target stated by the Ministry of Agriculture. We see 411 longliners (drifting longlines only) in the WCPFC region that fish on foreign EEZ or the high seas. This is 96% what the WCPFC reports in Similarly, we see 20 purse seiners fishing on the WCPFC high seas area in 2016, which corresponds to what the RFMO reports. The ICCAT report 41 active Chinese longliners operating in 2016 and we observe 47. We observe 27 longliners in the IOTC in 2016, which is 40% of the 67 vessels reported. Overall, we see 350 squid jiggers, which suggests we are missing 150 vessels. We observe a total of 1,108 Trawlers fishing on foreign EEZ or the high seas. This is more than what the 800 boats reported in Combining purse seines with gillnets, driftnets, pole and line, and vessel of unknown gear, we see a total of 172 vessels fishing on foreign EEZ or the high seas. This is more than what the 100 boats reported in 2007 for the "purse seines and other fishing gears" category reported in Taiwan Taiwan s distant water fleet is reported to be more than 2,000 vessels fishing in more than 26 countries by Taiwan s Fisheries Agency. Another source estimates that, in 2014, the distant water fishing fleet numbered 1,300 vessels and included 34 tuna purse seine vessels, 332 large scale tuna longline vessels, 746 small scale tuna longline boats, 99 squid jiggers and 91 torch light net vessels for Pacific Saury. The OPRT reports that in 2016, there were 296 Taiwanese large registered vessels (> 100 GT) targeting tuna. Additionally, the WCPFC reports that in 2015, there were three Taiwanese fleets operating in the region: 1) large-scale long liners (>100 GRT): 76 active vessels, 2), small-scale long liners:1,306 vessels operating both inside Taiwan s EEZ and beyond, and 3) distant-water purse seiners: 34 active vessels. In turn, the ICCAT reports that in 2015 the number of longliners was 117; 75 fished for big-eye and 43 targeted albacore. The CCBST reports 70 Taiwanese vessels active in 2015 targeting southern bluefin tuna and the IOTC reports that Taiwan s active capacity in 2016 was 233 vessels for tropical tunas and 111 for swordfish and albacore. Given the available information, we can make the following statements about Taiwan s AIS coverage: We observe 698 Taiwanese vessels fishing on foreign EEZ and international waters in This suggests we see 35% of the DWF using 2000 boats as the reference point. Of these, 593 (85%) belong to the high seas fleet and comprise the secondlargest after China. We see 36 Taiwanese-flagged purse seiners, which is slightly more than reported in 2014.

8 We observe 239 long liners over 100 GT. This represents between 72% and 81% of that reported. For the small-scale long liners (< 100GT), we see 352, representing 47% of the entire DWF. We see 52 Taiwanese squid jiggers, representing 53% of that reported in On the WCPFC, we see 109 large-scale long liners fishing on foreign EEZ or the high seas. This is more than the 75 reported by the RFMO On the small-scale class (<100 GT), we see 267, which represent 36 % of the DWF in the RFMO. However, the extra vessels we see in the >100 GT category are all less than 250GT. Given the available information from WCPFC, and that the inferred tonnages have a RMSE of 373, we think is reasonable to downgrade those vessels to the smaller size class. The ICCAT reports 117 active Taiwanese long liners operating in 2016 and we observe 50 (43%). The CCSBT reports 70 long liners in 2015 and we observe 159, which is more than twice as much. We observe 187 long liners and purse seiners on the IOTC which, compared to what they report as Taiwan s active capacity (233 vessels for tropical tunas and 111 for swordfish and albacore), is only 54%. Japan The Japanese government, in the 2013 Census of Fisheries available at Portal Site of Official Statistics of Japan, 2015, and the THE 90TH STATISTICAL YEARBOOK OF MINISTRY OF AGRICULTURE, FORESTRY AND FISHERIES : MAFF, presents a detailed profile of the country s fishing fleet. In this document, Japan s distant water fleet in 2013 was composed of: seven trawlers, 29 purse seiners, 197 longliners and 26 pole and line. More recently, the OPRT reports that there were 208 Japanese registered vessels >300 tons targeting tuna in In terms of the RFMOS, Japan national report to the IOTC states that in 2015, only 52 long liners and 3 purse seiners operated in the region. In the WCPFC s Tuna Fishery Yearbook (2015), it is stated that 111 distant water and offshore long liners, 75 pole and line, and 70 purse seiners actively fish in the RFMO area. However, the number of purse seine vessels licensed by the Japan Fisheries Agency (JFA) to fish in tropical waters has been capped at 35 since 1997 (> 350 GT). The national report by Japan to the WCPFC gives more detail and states that of the WDF and offshore long liners, 18 are between tons, 24 are between tons and 69 are greater than 200 tons. In the Atlantic, the ICCAT reports 72 long liners targeting bluefin are the only fleet present in the Atlantic Ocean and the CCBST website reports that 74 Japanese vessels were actively targeting bluefin in Lastly, Japan s report to the IATTC states that there 77 long liners in the Eastern Pacific Ocean in Given the available information, we can make the following statements about Japan s AIS coverage:

9 We observe 519 vessels in the distant water fleet and 478 in the high seas fleet. These numbers are higher than what the census reported in 2013 as 260 boats. We see four trawlers in the DWF, which is 57% of the seven reported in the 2013 census. We see 213 long liners in 2016, which is slightly more that reported by the OPRT. We observe 43 long liners and three purse seiners on the IOTC in 2016, representing 83% and 100%, respectively. On the WCPFC, we see 274 longliners >50GT on the DWF which is more than reported in the yearbook. By size class, we see: 93 between , 66 between and 109. This suggests that we see more than what the RFMO report for long lines in all categories. For pole and line vessels, we see 75, which is what has been reported. For purse seiners, we see 35 which is the cap given by Japan s fisheries agency. We see 69 long liners in the ICCAT area, 85 on the CCSBT area, and 62 on the IATTC area. These numbers suggest we see 96% of longliners targeting bluefin in the ICCAT, 115% of those reported on the CCSBT, and 81% of those reported on the IATTC. South Korea Korea s fisheries sector assessment by WWF, 2016 reports that the country s distant water fleet is comprised by 342 registered vessels (315 of which are active in 2013), owned by 71 companies, and add up to 202,172 GT. The composition of the DWF is: 150 tuna long liners, 32 tuna purse seiners, 93 trawlers (13 of which operate on the high seas), 32 squid jiggers, 14 Saury stick-held net dippers, 21 vessels using other fishing gear. More recently, the Korean Maritime Institute reports 348 active distant water vessels in Additionally, OPRT reports that there are 110 Korean registered vessels (> 100 GT) targeting tuna in In terms of the RFMOs, Korea s National Report to the IOTC (2016) states that in 2015, only 14 long liners (three were between GT and 11 between GT) and 5 purse seines (1 between 1,000-2,000 GT and 4 between 2,000-3,500) operated in the IOTC region. In the Atlantic, the ICCAT Biennial Report states that in 2015, four Korean longliners operated in the region and there was no activity by purse seiners. The CCBST website reports 10 Korean vessels active in 2015 and fishing for southern bluefin tuna. Lastly, the Korea s National Report to the WCPFC reports 96 long liners and 25 purse seiners actively fishing in the RFMO area in Given the available information, we can make the following statements about Korea s AIS coverage: We observe 325 vessels in the distant water fleet in 2016, which is greater than the active vessels in 2013 but 93 % of the reported vessels in We see 150 long liners in 2016 (100%). We observe 75 (>100%) purse seiners. However, if we exclude a cluster that fishes on Japan s EEZ we see 30, which is 94% of the reported estimate.

10 We see 30 (94%) squid jiggers. We see 13 (100%) trawlers on the high seas. In the IOTC region, we see 13 long liners in 2015 (93%) and three purse seiners (60%). In the ICCAT, we see 20 long liners (>100%) and 0 and purse seiners in This is more long liners than reported but consistent regarding purse seiners. In the CCSBT region, we see 18 long liners (>100%). In the WCPFC region, we see 113 longliners (>100%) and 27 purse seiners (>100%). Spain The 2017, EU Annual Economic Report (AER) states that Spain has 221 distant water vessels (>40 meters) active in By gear type, the fleet is composed of 30 purse seiners (> 40 meters), 33 demersal trawlers (>40 meters), 39 demersal trawlers (24-40 meters; not active on the high seas), and 83 long liners (24-40 meters). Another source, Cepesca, report that Spain has 205 vessels fishing in foreign non-eu water and the high seas. Of these, 89 are trawlers, 29 purse seiners, 3 bottom long liners and 84 drifting long liners. In terms of RFMOs, the WCPFC Tuna Fishery Yearbook (2015) and Spain s Annual report to the WCPFC report only five Spanish long liners and four purse seiners actively fishing in the RFMO area. In the IOTC, 14 purse seiners operated in 2016, more specifically, all operations have taken place on FAO region 51 and 19 long liners operated in 2016, 18 in 2015, and 22 in 2014 IOTC. The IATTC reports that in 2016, only two Spanish purse seiners operated in the Eastern Tropical Pacific and Spain s report on its activity in the region suggest there were 25 long liners fishing for swordfish in Another source of information is FFA s Global Tuna Market & Industry Dynamics report. This document states that only four Spanish-flagged vessels operate in the region plus two vessels with flags from El Salvador. According to the report, in 2007 EU firms controlled 84 purse seiners (20% of global vessels) but only 57 of these were EU-flagged. In the 2000s Spanish-owned and foreign-flagged purse seiners used at least eight flags from the Seychelles, five from Ecuador, five from Ghana, two from El Salvador and some from other Latin American and West African countries. Given the available information, we can make the following statements about Spain s AIS coverage: We observe 354 vessels in Spain DWF and 240 belong to the high seas fleet. Both fleets are larger than the 205 vessels reported by CEPESCA. We see 51 purse seiners in the DWF and 32 belong to the high seas fleet. Again, both fleets are larger than the 29 purse seiners reported by CEPESCA in We see 17 DWF purse seiners > 40 meters. The EU reports 30 in If we include Spanish-owned vessels flagged in Bonaire, El Salvador, Panama and Belize, our number increases to 31.

11 In the IOTC region, we see four purse seiners, which represent 29% of the 14 reported to be active in In the IATTC region, we see three purse seiners, which is one more than reported by the RFMO. We see 22 longliners, which is 88 % of the reported number from the RFMO. In the WCPFC we see three purse seiners and three long liners, which is 75% and 60% of the reported numbers. We see 87 trawlers in the DWF (98% of what CEPESCA reports). We see 39 trawlers > 40 meters in length, which is more than what the EU says. In the size class, we see 43 trawlers, again slightly more than the 39 that the EU AER reports in We see 135 longliners in the DWF, which is more than the 87 that CEPESCA reports. We see 96 longliners between 24 and 40 meters, which is slightly more than the 83 reported by the EU AER. United States The 2017 U.S Report to the WCPFC states there were 133 long liners, 23 long liners based in American Samoa and 37 purse seiners fishing in the RFMO region in Additionally, they report that only six vessels participated in the South Pacific albacore fishery in The last IATTC quarterly report of 2016, states that the U.S had 25 purse seiners fishing on the region. Lastly, the ICCAT Biennial Report states that 103 and 110 longliners operated in the RFMO region in 2015 and 2014, respectively. Given the available information, we can make the following statements about USA s AIS coverage: We see 38 purse seiners fishing on Hawaii s EEZ, foreign EEZ and the high seas of the WCPFC region. This suggests we have ~100% of the fleet. We see 20 purse seiners on foreign and international waters on the IATTC, representing 80% of the reported number. If we include all purse seiners that fish in US waters we see 66 vessels, which is more than the reported number. We see seven trawlers, which is slightly more than reported. We see 94 drifting long liners fishing in Hawaii s EEZ, foreign EEZs, and the high seas of the WCPFC region. This represents a total of 71% of the 133 long liners base in Hawaii and California, and 62% if we include vessels based on American Samoa. We see 57 drifting long lines in the ICCAT region in 2016, which is 55% of that reported by the RFMO. Of these, we see only 13 fishing on the high seas. Vanuatu Vanuatu s fleet operates predominantly in the Western and Central Pacific and the nation s annual report to WCPFC states that there were 3 purse seiners and 74 long line DWF and offshore vessels active in Additionally, OPRT reports that there are 34 Vanuatu registered vessels (> 100 GT) targeting tuna in 2016.

12 Given the available information, we can make the following statements about Vanuatu s AIS coverage: We see three purse seiners and 50 long lines in the WCPFC Region. This represents 100% and 68%, respectively, of the reported numbers. We see 39 long liners (> 100 GT), which is slightly more than reported. In summary, we have 100% coverage of purse seiners and long liners > 100GT. For smaller longliners (<100GT), we see 32.5% of the reported numbers. Portugal The 2017 EU Annual Economic Report (AER) states that in 2015, 20 vessels (>24 meters) comprised the distant water fleet, and 42 in Additionally, they state that only four long liners over 40 meters fish in foreign waters or the high seas. Between meters there are 16, and between there are 27. In terms of RFMOs, the IOTC reports that seven Portuguese long liners operated in 2016 and six in The EU WCPFC report 2017 states that Portugal fleet in the Pacific is limited to one long liner in the WCPFC and two in the IATTC. However, as many as 10 longliners are authorized to fish within the WCPFC. Given the available information, we can make the following statements about Portugal s AIS coverage: We see one long liner in the WCPFC and two in the IATTC, which correspond to the reported numbers. We see seven long liners fishing within the IOTC region, which again coincides with what has been reported. Overall, we see 45 longliners fishing on foreign EEZ or international waters. Five are larger than 40 meters (100% of the AER number), 29 are between meters and 11 between 12 and 24 meters. In summary we see 100% of the longliners in the IOTC, WCPFC and IATTC RFMOS. Overall, we see 100% of the Portuguese Distant Water fleet defined in the AER by vessels over 24 meters fishing on foreign EEZ and the high seas. France The 2017, EU Annual Economic Report (AER) states that the French industrial fleet of Purse Seiners consisted of 21 vessels in 2015, including the five vessels registered on the island of Mayotte, all over 40 meters in length. In terms of RFMOs, the IOTC reports that 12 French purse seiners (>40m) and 19 long liners (flagged in Reunion, >15 m) are authorized to fish in 2016 and in The list of active vessels from the ICCAT reports that there 46 purse seiners active in the region; 24 of them are more than 40 meters in length. Given the available information, we can make the following statements about France s AIS coverage:

13 Overall, we see 23 purse seiners fishing on foreign EEZ or high seas. Of these, 22 are greater than 40 meters, which is slightly more than the reported numbers from the AER. We see eight purse seiners in the IOTC region, which represents 67% of the reported number. We see 11 long liners in the IOTC, which represents 58% of the reported number. We see 19 purse seiners (>40 m) in the ICCAT region, which represents 79 of the reported number. Other countries The Seychelles National Report to the IOTC states that in 2014, the number of purse seiners and long liners was 11, and 36 respectively. We observe six and 32, respectively. Mexico s latest IATTC quarterly report of 2016 states there were 49 purse seiners operating in the Eastern Tropical Pacific; 21 are >70 meters, 22 between 50 and 70 meters, and three between meters. We observe 14, 16 and two, respectively. Colombia s latest IATTC quarterly report of 2016 states there were 14 purse seiners operating in the Eastern Tropical Pacific. The CLAV lists reports that 11 of those are >40 meters and three are between 24 and 40 meters. We observe 12 Colombian purse seiners in GFW. Ecuador s active industrial fleet is reported to be 40 vessels: Of these, IATTC reports there are 15 long liners. We observe three of the long liners and 14 of the purse seiners. Labor costs database Upper bound The EU 2017 Annual Economic report provides estimates of total cost of labor (crew wages and unpaid labor value) and total days at sea for the distant water fleets of Spain, France, Portugal, Italy, and Lithuania in Using this information, we estimated the average labor cost per day by nation, gear type, and size class. Additionally, the 90th statistical yearbook of Japan's Ministry of Agriculture, Forestry, and Fisheries (MAFF) provides information of the total labor cost and total fishing days by type of fishery and vessel size class. This information is aggregated for 2014 and reported per fishery enterprise. Moreover, the Korean Maritime Institute provides estimates of the average wages of fishermen per year, as well as the average number of days at sea per year for different gear types and size classes. Using this information, we estimate the average daily wage per crew for the main gear types. Lastly, for Chile, average labor costs per day are reported by Subpesca for segments of the shrimp trawlers, other trawlers, and purse seine fleets.

14 For China, we use the upper bound estimate of the monthly wages per crew member of the distant water squid jigging fleet ($500). For Taiwan, we use the country s current minimum monthly wage ($639). Combining all the above data sources we estimate the average labor cost per day by gear type and size class, which we use to fill in gaps and estimate the high bound of fishing costs. Lower Bound For China, we use the lower bound estimate of monthly wages per crew member of the squid jigging fleet ($350 USD). For Taiwan, we use the typical rate offered to foreign crew members in Taiwanese Distant Water vessels reported by Greenpeace ($300 per month) (53). For the United States, we use estimates of labor costs for the Hawaii-based long line fleet (54). This report estimates that on average, small tuna long liners (< 17m) spend 132,000 US dollars on labor cost per year, medium sized longliners spend (17-22 m) spend 171,000, and the large (>22 meters) spend 181,000. Using the average number of days at sea of these vessels we obtained estimates of labor cost per day. For Vanuatu, we use the average annual payment to crew members ($7,294) reported in (55) combined with the observed average days at sea to estimate labor cost per day. Lastly, we obtained mean nominal hourly earnings and labor cost per employee in the category of "Skilled agricultural, forestry and fishery workers" reported by ILO in US dollars for Argentina, Australia, Belize, Colombia, Cyprus, Ecuador, Estonia, Latvia, Malaysia, Mauritius, New Zealand, Norway, Poland, Russia, (ILO, accessed 10/28/17). Assuming an eight-hour workday and adjusting for inflation, we estimate daily earnings and labor cost per employee in 2014 dollars. For the remaining countries, representing ~8% of total high seas effort, we used regional, gear type, and size-specific average labor costs from the lower bound estimates described above. Fractions of total costs The EU 2015 Annual Economic Report and the 2013 Japanese Fisheries Yearbook provide detailed information of the cost structure of the distant water fleets. Using this data, we estimate the average fraction that fuel and labor cost represent from the total costs (these include: depreciation, repair, maintenance, rights, other variable costs, other non-variable costs). The resulting fractions by vessels size class are: 0.43 (12-18 meters), 0.44 (12-24 meters), 0.45 (24-40 meters) and 0.47 (> 40 meters).

15 Supplementary Figures A. B. trawlers squid jigger pots and traps purse seines other fishing unknown pole and line drifting longlines set longlines trollers set gillnets trawlers squid jigger pots and traps other fishing pole and line purse seines unknown drifting longlines set longlines set gillnets trollers length (m) C. D Crew size E. F. engine power (KW) y = 0.98 x 1.8 trawlers purse seines squid jigger pots and traps unknown other fishing pole and line drifting longlines set longlines trollers set gillnets Engine power (KW) drifnets squid jigger pots and traps drifting longlines pole and line purse seines trawlers trollers set longlines set gillnets Fraction of fishing effort on high seas tonnage (GT) y = x length (m) Known Predicted length (m) Known Predicted fig. S1. Distributions of vessel characteristics, high-seas fishing effort relative to total fishing effort, and nonlinear relationships between length, tonnage, and engine power. Distribution of vessel length (A), engine power (B), and crew size (C) by gear type. Fractions of fishing effort in the high seas vs. EEZ by gear type (D). Nonlinear relationships between length and engine power (E), and between length and tonnage (F).

16 fig. S2. Vessel encounters on the high seas (2016). Global distribution of encounters between fishing vessels and reefer/bunker vessels (A). Top 14 fishing fleets with most encounters with reefers and bunkers in the high seas (B).

17 fig. S3. Fishing effort by gear type. Total high seas fishing effort, and fishing effort for the top five gear types (2016).

18 fig. S4. Effort versus catch. Relationship between high seas fishing effort (kw days) and catch (tons) by country and FAO region.

19 Supplementary Tables table S1. High-seas fishing fleets and effort by country (2016). Country No. Vessels Avg. length (m) Total tonnage (GT) Installed Capacity (KW) Avg. crew size Fishing days Fishing hours (x1000) Fishing kwh (x10^6) China 838 (23%) ,5061 (29%) 2142 (30%) 2455 (36%) Taiwan 593 (16%) ,870 (20%) 1526 (22%) 1284 (19%) Japan 478 (13%) ,466 (14%) 1040 (15%) 737 (11%) Indonesia 450 (12%) ,641 (10%) 426 (6%) 128 (2%) Spain 252 (7%) ,514 (6%) 438 (6%) 225 (3%) South Korea 200 (6%) ,102 (8%) 567 (8%) 575 (9%) United States 141 (4%) (2%) 141 (2%) 79 (1%) Russia 73 (2%) (2%) 110 (2%) 287 (4%) Portugal 60 (2%) (2%) 120 (2%) 91 (1%) Vanuatu 52 (1%) (2%) 151 (2%) 167 (2%) Norway 41 (1%) (0%) 20 (0%) 73 (1%) Canada 36 (1%) (0%) 14 (0%) 55 (1%) Seychelles 35 (1%) (1%) 64 (1%) 63 (1%) Mexico 28 (1%) (1%) 23 (0%) 81 (1%) NA 28 (1%) (0%) 28 (0%) 32 (0%) France 23 (1%) (0%) 12 (0%) 31 (0%) Fiji 21 (1%) (0%) 30 (0%) 17 (0%) Ecuador 17 (0%) (0%) 12 (0%) 22 (0%) Sri Lanka 17 (0%) (0%) 4 (0%) 1 (0%)

20 Faroe Islands 16 (0%) (0%) 8 (0%) 22 (0%) Panama 15 (0%) (0%) 5 (0%) 12 (0%) New Zealand 13 (0%) (0%) 9 (0%) 11 (0%) Colombia 12 (0%) (0%) 20 (0%) 58 (1%) Micronesia 12 (0%) (0%) 4 (0%) 7 (0%) Kiribati 11 (0%) (0%) 4 (0%) 11 (0%) United Kingdom 11 (0%) (0%) 11 (0%) 13 (0%) Ireland 9 (0%) (0%) 1 (0%) 1 (0%) Latvia 9 (0%) (0%) 24 (0%) 27 (0%) Malaysia 9 (0%) (0%) 18 (0%) 13 (0%) Marshall Islands 9 (0%) (0%) 3 (0%) 7 (0%) South Africa 9 (0%) (0%) 3 (0%) 2 (0%) Venezuela 9 (0%) (0%) 4 (0%) 10 (0%) Chile 8 (0%) (0%) 9 (0%) 15 (0%) Ghana 8 (0%) (0%) 6 (0%) 15 (0%) Belize 6 (0%) (0%) 8 (0%) 7 (0%) Estonia 5 (0%) (0%) 10 (0%) 25 (0%) Germany 5 (0%) (0%) 5 (0%) 13 (0%) Lithuania 5 (0%) (0%) 4 (0%) 12 (0%) Maldives 5 (0%) (0%) 4 (0%) 4 (0%) Ukraine 5 (0%) (0%) 12 (0%) 19 (0%) Mauritius 4 (0%) (0%) 3 (0%) 2 (0%) Australia 3 (0%) (0%) 0 (0%) 0 (0%) Bonaire 3 (0%) (0%) 1 (0%) 1 (0%) Brazil 3 (0%) (0%) 3 (0%) 1 (0%)

21 Iceland 3 (0%) (0%) 0 (0%) 2 (0%) Iran 3 (0%) (0%) 2 (0%) 1 (0%) Thailand 3 (0%) (0%) 1 (0%) 0 (0%) Denmark 2 (0%) (0%) 1 (0%) 6 (0%) Guatemala 2 (0%) (0%) 0 (0%) 0 (0%) India 2 (0%) (0%) 1 (0%) 1 (0%) Papua New Guinea 2 (0%) (0%) 0 (0%) 1 (0%) Senegal 2 (0%) (0%) 0 (0%) 0 (0%) Tanzania 2 (0%) (0%) 4 (0%) 4 (0%)

22 table S2. High-seas fishing fleets and fishing effort by gear type. Gear type No. Vessels Avg. length (m) Total tonnage (GT) Installed Capacity (KW) Avg. crew size Fishing days Fishing hours (x1000) Fishing kwh (x10^6) Drifting Longlines 2165 (60%) (69%) 5047 (71%) 3719 (55%) Purse seines 457 (13%) (6%) 263 (4%) 394 (6%) Squid jigger 453 (13%) (17%) 1231 (17%) 1490 (22%) Trawlers 255 (7%) (5%) 351 (5%) 979 (15%) Pole and line 93 (3%) (1%) 59 (1%) 55 (1%) Unknown 84 (2%) (1%) 42 (1%) 45 (1%) Set longlines 44 (1%) (0%) 28 (0%) 21 (0%) Set gillnets 39 (1%) (0%) 12 (0%) 3 (0%) Pots and traps 15 (0%) (0%) 31 (0%) 34 (0%) Trollers 8 (0%) (0%) 4 (0%) 2 (0%) Driftnets 5 (0%) (0%) 2 (0%) 2 (0%)

23 table S3. AIS coverage of high-seas fleet and scale factors by country, Regional Fisheries Management Organization, gear type, and vessel size in Scale Flag iso3 RFMO Gear type Vessel size AIS vessels Reference point factor TWN WCPFC Drifting longlines (0,250] Tons TWN WCPFC Drifting longlines (250,2000] Tons VUT WCPFC Drifting longlines (0,100] Tons VUT WCPFC Drifting longlines (100,2000] Tons CHN ICCAT Drifting longlines ALL CHN IOTC Drifting longlines ALL CHN WCPFC Drifting longlines ALL CHN WCPFC Purse seines ALL TWN CCSBT Drifting longlines ALL TWN ICCAT Drifting longlines ALL TWN IOTC Drifting longlines ALL TWN WCPFC Purse seines ALL JPN CCSBT Drifting longlines ALL JPN IATTC Drifting longlines ALL JPN ICCAT Drifting longlines ALL JPN IOTC Drifting longlines ALL JPN WCPFC Drifting longlines ALL JPN IOTC Purse seines ALL JPN WCPFC Purse seines ALL KOR CCSBT Drifting longlines ALL KOR ICCAT Drifting longlines ALL KOR IOTC Drifting longlines ALL KOR WCPFC Drifting longlines ALL

24 KOR IOTC Purse seines ALL KOR WCPFC Purse seines ALL ESP WCPFC Drifting longlines ALL ESP IOTC Drifting longlines ALL ESP IATTC Drifting longlines ALL ESP IATTC Purse seines ALL ESP IOTC Purse seines ALL ESP WCPFC Purse seines ALL USA ICCAT Drifting longlines ALL USA WCPFC Drifting longlines ALL USA WCPFC Purse seines ALL USA IATTC Purse seines ALL USA WCPFC Trollers ALL VUT WCPFC Purse seines ALL SYC IOTC Drifting longlines ALL SYC IOTC Purse seines ALL COL IATTC Purse seines ALL ECU IATTC Drifting longlines ALL ECU IATTC Purse seines ALL TWN ALL DWF Squid jigger ALL JPN ALL DWF Pole and line ALL JPN ALL DWF Trawlers ALL KOR ALL DWF Squid jigger ALL KOR ALL DWF Trawlers ALL FRA ALL DWF Purse seines (40,100] Meters MEX ALL DWF Purse seines (0,40] Meters MEX ALL DWF Purse seines (50,70] Meters

25 MEX ALL DWF Purse seines (70,100] Meters CHN ICCAT Squid jigger ALL CHN IOTC Squid jigger ALL CHN WCPFC Squid jigger ALL CHN SPRFMO Squid jigger ALL table S4. Average labor costs per day by country, gear type, and vessel size class. Avg. labor cost per day (low bound) Avg. labor cost per day (high bound) Country Gear type Length class China Squid jigger (40,200] China Drifting longlines (40,200] Taiwan Drifting longlines (24,40] Japan Drifting longlines (40,200] Taiwan Drifting longlines (40,200] South Korea Drifting longlines (40,200] China Drifting longlines (24,40] Spain Drifting longlines (24,40] Japan Drifting longlines (24,40] Indonesia Drifting longlines (18,24] Taiwan Drifting longlines (18,24] Indonesia Purse seines (24,40] Indonesia Drifting longlines (24,40] Russia Trawlers (40,200] Taiwan Squid jigger (40,200] Vanuatu Drifting longlines (40,200]

26 United States Drifting longlines (18,24] Norway Trawlers (40,200] Japan Pole and line (24,40] Indonesia Purse seines (18,24] China Trawlers (40,200] Mexico Purse seines (40,200] Japan Drifting longlines (18,24] Japan Pole and line (40,200] United States Purse seines (40,200] Spain Drifting longlines (40,200] Portugal Drifting longlines (24,40] Spain Trawlers (40,200] Spain Purse seines (40,200] Seychelles Drifting longlines (40,200] Spain Fixed gear (18,24] South Korea Squid jigger (40,200] China Purse seines (40,200] South Korea Drifting longlines (24,40] France Purse seines (40,200] United States Drifting longlines (24,40] South Korea Trawlers (40,200] Fiji Drifting longlines (24,40] Ecuador Purse seines (40,200] Taiwan Purse seines (40,200] South Korea Purse seines (40,200] Colombia Purse seines (40,200] Spain Fixed gear (24,40]

27 Seychelles Drifting longlines (24,40] Portugal Drifting longlines (18,24] Faroe Islands Trawlers (40,200] Panama Purse seines (40,200] Kiribati Purse seines (40,200] Marshall Islands Purse seines (40,200] Venezuela Purse seines (40,200] table S5. High-seas costs, catch, revenue, subsidies, and profits by country. Summary of scaled results by country. π = profits before subsidies, π* = profits after subsidies. All monetary values are US$ millions. FSM = Federates States of Micronesia, PNG = Papua New Guinea, RMI = Republic of the Marshall Islands, U.K. = United Kingdom. Country Catch (tonnes) Revenue Cost low bound Cost high bound π low bound π high bound Subsidies π* low bound π* high bound China Taiwan South Korea Spain Japan Ecuador Indonesia Russia Mexico United States Norway France Seychelles Panama

28 Ghana Colombia PNG Vanuatu Venezuela Sri Lanka Belize Portugal New Zealand Kiribati Faroe Islands Thailand Iceland RMI Chile Fiji Estonia Guatemala FSM Iran India Ukraine Denmark Lithuania Ivory Coast Germany U.K Namibia