China’s bottom trawl sheries and their global impact

China dominates the world’s highly destructive bottom trawl sheries (BTF), active in 30 countries and landing 28% of BTF catch. We created the rst time series for China’s BTF, from 1950 to 2018, and examined their national and global impacts. Between 1978 and 1997, China’s BTF eet increased 47 fold (in numbers) and 26 fold (in engine power), driving enormous acceleration in Asia’s shing capacity. China embarked on BTF globally from 1985, particularly in East Asia and Africa. Such distant water sheries (DWF) absorb about 20% of China’s BTF capacity. China’s rampant BTF raises signicant concerns: sh availability plummeted wherever China shed, domestically and in DWF; there are strong indications of shing through the food web and shing indiscriminately in China’s EEZ; and the mean trophic level of catch eaten by humans has declined. Urgent management interventions are needed to stem such ‘slash-and-burn’ shing practices in China and worldwide.


Introduction
The expansion of bottom trawl sheries (BTF)-sheries that 'bulldoze' demersal marine life and habitat non-selectively-has long been a great concern around the world1,2. Studies have indicated that the number of global shing vessels was twice the level that would produce maximum sustainable pro ts, losing potential bene ts of at least $50 billion in 2004 alone3. As the world's dominant commercial shing practice, BTF has consistently contributed c. 25% biomass of global annual catches since late 1980s, including 60% of the discards from that biomass4. Substantial evidence has accumulated to show that BTF extracts non-targeted and rare marine species indiscriminately, degrades benthic marine ecosystems, unleashes contaminants from sediments5-7. BTF also come into con ict with small-scale and selective sheries, engage in illegal activities and may be crewed by slavery or forced labor8.
Moreover, BTF are generally unpro table and rely on government subsidies to survive9. Therefore, curtailing BTF should be a high priority in developing sustainable marine sheries and conserving biodiversity1, especially as we try to meet UN Sustainable Development Goal 14 (Life Below Water) by 203010.
Mitigating impacts of BTF requires countries to reduce both their shing effort and constrain their footprints at seas. Historically, shing efforts and footprints of BTF have generally moved offshore from domestic to distant waters, and from developed (e.g. European Union, North America) to developing regions (e.g. Asia, Africa)11-13, largely driven by resource availability, technology development, policies, and pro tability6, 14,15. Over the past few decades, some developed nations have stabilized their overall shing effort and footprints both in domestic and distant waters16-19. But developing countries, especially in Asia, have been expanding their share for decades along with increasing their economic power, foreign investment, and global consumption20-24. It is thus vital for the world to understand the trajectory of bottom trawl sheries (BTF) in these rising shing powers… and their impacts on the ocean sustainability.
A great challenge in probing BTF is that many developing countries have not gathered or published data on their shing capacity and shing effort16, 17,25,26. In the past ten years, global shing capacity and effort have been frequently reconstructed16-18. A common trend is that such capacity and effort expanded rapidly from the late 1970s throughout 2010; notably this trend is dominated by Asian eets which continue to grow17,18. Such a prominent rise is likely due to the rapid development of trawling since 1970s in Asia, where BTF is massive (e.g. ~ 83,000 BTF vessels targeting shrimp) but poorly monitored and managed19, 27. In literature, however, few studies have focused on reconstruction of shing capacity of BTs in developing countries in Asia. This knowledge gap makes it di cult for the world to understand their impacts on the ocean and for policy making to mitigate these impacts.

Results
Development phases. We found that the history of China's BTF could be split into four eras: (i) First Era (1950-1963, E1), characterized by shing through the food web while moving offshore; (ii) Second Era (1964-1978, E2), characterized by diversi cation and more powerful vessels due to technology development; (iii) Third Era (1979-1996, E3), characterized by explosive growth after the economic reform and moving into distant waters beyond C4S; and (iv) Fourth Era (1997-2018, E4), characterized by fewer but more powerful vessels and growing distant-water sheries (hereafter, DWF) See text for the explanation of the development phases in Supplementary Information SI 2.
Total shing capacity. We showed that the total capacity of Chinese BTF was very low (in relative terms, < 1500 trawlers) in E1 and E2, but increased dramatically from E3 after China embarked on a policy of economic reform (1978) (Fig. 1a). The tally peaked at very large numbers (~ 70,000 trawlers, 8 GW) around 1997 (Fig. 1a)-the highest known for any nation-and then declined to ~ 30,000 vessels in 2018 (6 GW, Fig. 1a). China's BTF capacity in DWF beyond C4S grew quickly in just two decades, once it had started in 1985 (45 trawlers per year, r2 = 0.99; Fig. 1b). By 2018, their capacity accumulated to 1500 vessels and 1.2 GW (Fig. 1b), equal to 20% of the gross engine power of all Chinese trawlers.
Mean shing capacity: Within C4S, the horsepower per vessel (HpV) of Chinese BTF has grown consistently (from 65 to 198 kW) except for a shock from 1978 to 1980 (Fig. 1c), when a large number of small private trawlers emerged after economic reform34, 35. In contrast, the HpV of China's BTF beyond C4S uctuated dramatically and peaked in mid-1990s (1500 kW); notably a new rise has occurred since 2013, as China's sheries moved towards fewer, more powerful boats32 (Fig. 1c). We found Chinese trawlers were generally more powerful than its other marine shing vessels. With C4S, the ratio between HpV of BTF and of other Chinese marine sheries increased dramatically from 1980 to early 1990s, but then plummeted before stabilizing at around 5.0 (Fig. 1d). Such a ratio in shing eets beyond C4S showed a similar shape of trajectory with a quick rise in early 1990s but then declined to around 1.0 by 2010 and stabilized (Fig. 1d).
Total catch: Total landings by Chinese BTF grew fastest in E3, and then stabilized after 1996 before a large new growth after 2013 (Fig. 2a). The total landed value showed a similar trajectory, but generally increased more continuously (Supplementary Fig. S2.1). The percentage of Chinese marine catch derived from BTF increased consistently in the rst three eras (Fig. 2b). The proportion peaked around 60%, and then leveled off until a new rise after 2013 to around 70% (from 2015 -2018, Fig. 2b). The proportion of BTF landings from China's claimed EEZ gradually declined from ~ 98 to 70% by 1975, and then bounced back to 90% by 1984, before a new drop to ~ 40% (i.e., 60% were from distant waters; Fig. 2b). This indicates Chinese BTFs initially moved offshore, partly driven by the no-trawl zone policy (1955), and then moved inshore after the shery agreement with Japan (effective in 1975)32. Chinese trawlers have, however, increasingly operated in distant waters since China started to develop DWF beyond C4S in 198532. The catch share from C4S then gradually declined from 100% in 1985 to 60% in 1997, then bounced back again slightly before dropping to ~ 55% after 2013 (Fig. 2b).
Fishing e ciency (CPUE & VPUE): Within C4S, catch per unit effort (CPUE) of Chinese BTF initially doubled from 1950 to 1954, peaking at 6.8 t / (kW · year), and then halved by 1962 (Fig. 2c). After 1962 CPUE was boosted again to 5.3, largely by technology development35 (Fig. 2c). But this rise did not last long and the CPUE plunged in 1970s to less than 1 t / (kW · year) by 1984 (Fig. 2c). After China implemented summer moratorium within C4S in 1981, the CPUE declined slower and then gradually rose to nearly 1.5 t / (kW · year) by 2018 (Fig. 2c). Beyond C4S, CPUE plummeted even more strikingly from 55 to 5 t / (kW · year) in just one decade from 1985 (when it was 118 times the value within C4S) to 1995 (when it became only 3 times the value within C4S) (Fig. 2d). Landed value per unit effort (VPUE) demonstrated similar trajectories in both analyses (Figs. 2c&d).
Mean trophic level (MTL) in China's claimed EEZ. Our results suggest that China's BTF in its claimed EEZ showed two signs of problematic shing behavior (Fig. 3a). The rst is shing down/through the food web, which means increased representation of lower trophic levels in the catch36,37. This behavior was evident in China's BTF and in its all marine sheries within China's claimed EEZ, during E1, E2, and part of E3 (Fig. 3a). For instance, we found the MTL of China's BTF landings that were directly consumed by humans (hereafter, MTLℎ) declined at a rate of -0. Log-relative-price index (LRPI) in China's claimed EEZ. The LRPI represents the log-transformed slope of a linear relationship between price and trophic level of the sheries stocks38. In a healthy shery, species at higher trophic levels generally hold higher prices and thus generate a positive LRPI value, while a bellshape trajectory is common when high-valued sheries stocks are gradually being depleted38. In China's claimed EEZ, we found that the LRPI of all species (or sh species only) of China's BTF generally followed a bell-shape trajectory (Fig. 3b), suggesting a gradual decline on high trophic-level sh stocks, in line with the MTL (Fig. 3a). However, it should be noted that the negative values in the early eras likely arose from overestimates of prices for shrimps and shes at lower trophic levels in previous studies 39,40 ( Supplementary Fig. S2.4).

Page 6/24
Fishing-in-balance index (FIBI) in China's claimed EEZ. The FIBI is commonly used to re ect sheries' geographic expansion (FIBI increase) or contraction (FIBI decrease). It goes up if catches increase faster than would be predicted by trophic level declines, and it goes down if catches fail to compensate for a decrease in trophic level25. We found China's BTF in its claimed EEZ gradually expanded in the E1 and E2 at a consistent rate (Fig. 3c), that expansion then sped up in E3 (after economic reform) and slowed down in E4 (after China rati ed UNCLOS in 1996 and started to reduce total shing capacity)32. This trajectory is generally in line with China's all marine sheries estimated by the Sea Around Us Project (SAUP)41, although the latter varied more prominently especially in E3 (Fig. 3c).
Catch composition in China's claimed EEZ. We found prominent shifts in the dominance of different stock assemblages over the four eras in China's claimed EEZ, again providing strong indications of (i) shing through food webs, (ii) biomass trawling, and (iii) shing based on availability (Fig. 4, see more details in Supplementary Figs S2.5-S2.10). For instance, in E1, large sh are the dominant stock assemblage, followed by shrimps and medium-sized sh in terms of percentages of landings. However, dominance of large and medium sh gradually declined (though not depleted, Supplementary Figs S2.5&2.6) as other species at lower trophic levels (e.g. jelly sh) increased their share (Fig. 4a), suggesting shing through the food web. In E4, the dominance of shrimps and jelly sh vanished after 2002. Since then, the catch shares of most assemblages have tended to converge between 10 and 20% (Fig. 4a), meaning that Chinese BTF has become less selective. The catch share of crabs & lobsters (mainly crabs) increased (Fig. 4a), likely because they became relatively more available (Supplementary Fig. S2.9). In contrast, the dominance measured by landed value showed a totally different picture (Fig. 4b); this was dominated by the high prices of shrimps with surprisingly high relative values of jelly sh during parts of E3 and E4.
Global contribution. We found China has greatly increased its share in global BTF, especially between 1985 and 1996, when its contribution to global landings rose from 4% to 21% (Fig. 5a). Three other developing nations in Asia (i.e. Indonesia, Vietnam, and Thailand) also rose to sit among the top ve exploiters between 1994 and 2014 (in terms of mean contribution to the landings). In contrast, the top ve shing powers between 1950 and 1970 (i.e. USA, Spain, Russia, Portugal, and Japan) have generally shrunk their shares, with only the USA still found within the top 5 contributors between 1994 and 2014 (Fig. 5a). Notably, the global landings of BTF peaked at 37.6 Mt in 1989 and then generally declined, although the catch remained relatively high (> 25 Mt; Fig. 5b). Such a transition roughly matches with the decline on CPUE (and VPUE) of China's BTF beyond C4S (Fig. 2d).  (Fig. 7a,e,h,I,j,k) and was also dominant in EEZs that BTF from other foreign nations hardly touched (e.g. Côte d'Ivoire; Fig.   7b,c,d,f,g,l).

Discussion
We reveal a seven-decade modi cation of China's BTF (in capacity, catch, and footprint) and identify its impacts on global sheries and marine conservation. First, ours is the only reconstruction of a long-term timeseries of BTF in any developing countries, where sheries data are poorly reported to FAO, and only the third such reconstruction for any BTF globally6,42. Our study suggests that the dramatic rise in shing capacity of Asian eets in the late 1970s was partly driven by Chinese bottom trawlers17,18. The reconstructed capacity data also enables us to reveal a dramatic decline on CPUE (and VPUE) for China's BTF both in and beyond C4S; notably, the CPUE for China's BTF beyond C4S decreased much more strikingly than the CPUE for all marine sheries in the same regions, especially Asia and Africa18. The new ndings highlight the importance of reconstructing shing capacity6,18, especially for data-poor developing countries in Asia (with its considerable shing power). Second, from a sustainability and conservation perspective, our study suggests China's BTF in its claimed EEZ included shing through food webs, biomass trawling, and shing based on availability43,44, rather than simply shing down the food web37,45. Such destructive practices by China appear also to have penetrated to the distant waters, given the plunging CPUE in waters beyond C4S. Management actions to stem such 'slash-and-burn' shing practices at sea are needed, though not only for China, if we are about to achieve the UN Sustainable Development Goal in marine sheries by 203010.
Our reconstructed capacity timeseries reveals new insights on the effects of China's policies on BTF. The astonishing rise in BTF capacity since 1979 was largely due to the nation's economic reform, which privatized vessels and freed the seafood market32,46. The decline in BTF capacity since 1997 might have been driven by the Double Control (on vessels and horsepower) and the Asian Financial Crisis in 199746,47, as well as the low CPUE and VPUE. Interestingly, we suggest that China's vessel buyback programs (started in 2002), which aimed to reduce shing capacity32, might actually have halted a decline in capacity. One explanation is that shers remained in BTF in the hope of receiving yet higher buyback allowances in later years, and were facilitated in this intent by the fuel subsidies that started in 200632. Vessel buyback programs have been widely employed in many sheries around the world (e.g. New Zealand, Australia, US, Canada) for capacity reduction or vessel modernization1,48. Studies have suggested that, when such a program is used to reduce shing capacity, a catch-share system is often needed as an additional measure to prevent shers remained in the sheries from intensifying the race for sh48. The dearth of such a system in China's BTF may explain the intensi ed biomass trawling observed in recent decades32,46.
Our study raises signi cant concerns over sustainability and conservation in China's BTF, including a striking collapse in CPUE in both C4S and distant waters beyond C4S. The plunging CPUE (and VPUE) in C4S is most likely due to sheries depletion32,49, although it might also be affected by other factors (e.g. Sino-Japanese Fishery Agreement 1975)32,50. Such a plunge may not arise from China's BTF alone. For instance, in the 1960s and 1970s, Japanese bottom trawlers were much more powerful and may have driven the decline of some benthic sh stocks in the East China Sea50. As well, many other shing entities (e.g., Russia, EU, South Korea) have long been involved in the same distant EEZs where China operates. Fisheries in some of these waters (e.g. the Atlantic coast of Africa) may have already been depleted by these nations long before China started in 198551, as hinted by the decline in global BTF landings after 1989. However, we nd that China has gradually overtaken other nations in shing many of these waters in recent decades, while also expanding its footprints to foreign EEZs that other shing powers rarely exploit. Many of these EEZs are in Africa which is most affected by IUU shing52, while some of them (e.g. the Persian Gulf) are homes to fragile coral reefs which have also been threatened by climate change53. Worries about sustainability of BTF in a global sense are heightened by (i) the revealed history of destructive trawling by China within its claimed EEZ and (ii) China's current pattern of expanding its footprints by sending more powerful vessels to distant waters.
We nd that China's BTF had disproportionally greater impacts upon species of higher trophic levels. Given that trawling is non-selective, with most taxa removed in proportion to their availability, we might imagine that MTL would not change much over time54. The fact that MTL has declined in China's claimed EEZ from 1950 to 1988 might result from two reasons. First, initially, Chinese trawlers did selectively target the schooling sh stocks of higher trophic levels (e.g., large yellow croakers), causing higher shing mortality of these species32,49. Second, trawl pressure in this region was so high that it in uenced resilience of species to varying extents, with higher tropic levels being most affected45,55. It is likely that both mechanisms have played a role in driving the decline on MTL.
We propose a new metric to measure sheries sustainability, i.e. MTLℎ. MTL or marine trophic index has been widely adopted as an indicator for sustainable sheries or ecosystem health by scientists44.
However, in line with other criticisms36,44, we suggest that using MTL to measure sustainability could be misleading in the case of bottom trawling. Bottom trawling can show increases in MTL at certain times even under egregious shing pressure. One example would be China's BTF after a sheries moratorium directed at protecting high trophic level species and lasting several months32. The consequent increase in juveniles of these species would help to disguise over shing of their adults, by in ating the MTL, as shown by our results. In contrast, using MTLℎ largely prevented such a misleading effect. We thus propose MTLℎ as an important improvement that is suitable for monitoring the sustainability of sheries, especially non-selective ones like BTF.
Although our study focused on China, the same worrisome trajectory of BTF has been found in many other countries. For instance, the trajectories of the shing capacity and CPUE of China's BTF resemble those of UK's BTF over 118 years from 1889 to 20076. China's BTF have gradually moved offshore and increased shing pressures in distant waters. Such changes are found in the footprints of the global commercial shing eets12,13, as well as the development history of BTF in other nations in Europe, North America, and more recently in Asia6, 15,27,56. This type of shing practice resembles the 'slashand-burn' agricultural practice on land and impairs our progress towards the UN Sustainable Development Goal 14-Life below Water by 203010. However, the development of BTF is often entangled with sh & animal farming22,23, increasing consumption demand32, and other short-term socioeconomic concerns (e.g., employment)57, especially in developing countries58. Such entanglements may explain the di culty of curtailing BTF in China, although the Chinese Ministry of Agriculture has long indicated a plan to do so through national policies32. In the future, the focus must be implementing policy effectively through management interventions that decouple short-sighted socioeconomic interests from BTF, in favour of long-term stability of ocean ecosystems and human food security.

Methods
Study area. While examining China's BTF as a whole, we also compared shery indices (as described below) for four ocean areas that overlapped each other (see detailed information in Fishery indices. We sought to create a comprehensive pro le of China's BTF and compare it to China's other (or all) marine capture sheries. To this end, we collected and analysed data in four categories of shery assessment: (i) total & mean shing capacity, (ii) total yield & shing e ciency, (iii) shery health, and (iv) catch composition (see details in Table 1 and more on SI 1). These categories contained a total of 14 indices. Eleven of these are commonly used in shery studies6,25,37-40,45,59-61. In addition, we devised an adjusted mean trophic level for catches directly consumed by humans (MTLℎ) and deployed two dominance indices of catch composition that are not commonly used (Table 1). We examined whether the MTLℎ could indicate the extent of biomass trawling when compared with the original MTL for all BTF landings. The two dominance indices represented the (i) contribution of each stock assemblage to the total catch, and (iii) contribution of each assemblage to the total landed value of the catch. We de ned seven stock assemblages for China's BTF (e.g., large sh, medium sh; Table 2).
Data sources. We derived our 14 shery indices based on data collected from a total of ve sources (Table 1) Supplementary Table S1.1). The fth source, SAUP data, contained reconstructed catch and landed value speci cally for different shery stocks (some even to the species/genus level) by different types of gears including bottom trawl and other sub-categories in the dataset (e.g. 'shrimp trawl', 'beam trawl', and 'dredge'). Here we considered all these demersal destructive gears as shing gears used in BTF.
Fishing capacity reconstruction. The timeseries datasets for shing capacity (i.e., vessels and engine horsepower) of China's BTF were reconstructed from 1950 to 201833. The reconstruction process was focused on three approaches: (i) interpolating missing data points within the time frame of collected data (using GAM & LOESS, 'gam', 'loess.as', and 'loess' functions from the 'mgcv' and 'fANCOVA' r packages)65,66, (ii) extrapolating data to earlier or later years beyond the time frame (using ARIMA, 'auto.arima' function in the 'forecast' r package)67, and (iii) calculating total capacity from regional records (e.g. BTF horsepower in the East China Sea) based on estimated ratios. We used a bootstrapping approach to estimate 95% con dence intervals wherever it was applicable17. Although similar approaches have been used by other studies17,18, our work differed in two ways. First, we validated our estimates with independent data whenever available (e.g. piecemeal regional records of bottom trawlers).
Our primary rule was that the estimated metric for China's BTF (e.g. number of vessels) must fall between (i) the sum of the same metric across provinces for which data were available (lower bound) and (ii) the same metric for any category of vessel that included BTF (e.g. all motorized catchers; higher bound).
Second, for extrapolation, we used other models (GAM or LOESS) instead of ARIMA when the latter derived less acceptable estimates (e.g. negative values or violated our primary rule)33.
Fishery yield reconstruction. The catch timeseries were originally derived from SAUP database and were updated for the period from 1950 to 198433. SAUP allocated many records (n = 9826) of 'reported' and 'unreported' catches from distant waters beyond C4S to China's BTF before 1985. Yet it is widely recognized that China's shing eets did not enter waters beyond C4S until 1985, as SAUP acknowledges68. Presumably mistakes occurred during the spatial disaggregation process in the original SAUP study62. These errors for the early period  were corrected by (i) removing the 'unreported' catches beyond C4S (mostly discards which are rare in China's sheries) and (ii) reallocating the 'reported' estimates beyond C4S to the closest (and likely most appropriate) regions within C4S33.
Further calculations and mapping. We used our collected and reconstructed data to calculate shery indices for mean capacity, shing e ciency, shery health, and catch composition (see details in SI 1