From dirty to delicacy? Changing exploitation in China threatens the world's largest amphibians

1. Determining the dynamics and sustainability of human interactions with threatened species is essential to inform evidence- based conservation, but data can be challenging to collect across large areas and multiple user groups. 2. Chinese giant salamanders Andrias spp. are critically depleted across China. Wild populations were exploited during the 20th century, and more recently to support a large- scale farming


| INTRODUC TI ON
Effective conservation of species that are threatened by exploitation requires a detailed understanding of human-wildlife interactions, including current patterns, levels and drivers of harmful behaviours by different user groups, the scales across which such interactions occur, and whether behaviour change over time has affected the sustainability of interactions (Nilsson et al., 2020;St John et al., 2013).
Determining the dynamics and sustainability of human-wildlife interactions is of particular importance for conservation-priority species occurring within social-ecological systems in human-dominated landscapes, for example across eastern and southeast Asia, where threatened biodiversity coexists with high human population densities and is heavily exploited for local consumption and wider trade (Carter et al., 2012;Coggins, 2002;Liu et al., 2016;Zhang & Yin, 2014). Establishing robust baselines to inform evidencebased conservation for such systems is an interdisciplinary activity that typically requires the use of social science methods (Bennett et al., 2017;Lischka et al., 2018;Newing, 2010). However, such data can be challenging to collect systematically across large areas, or in an integrated manner for illegal activities or multiple key user groups, such as across supply chains from hunters to consumers (Challender & MacMillan, 2014).
Chinese giant salamanders (Cryptobranchidae: Andrias spp.) are the world's largest amphibians, reaching almost 2 m in length and >50 kg in weight (Sowerby, 1925a(Sowerby, , 1925bWang et al., 2004). They were historically distributed in fast-flowing montane rivers and streams across a large area of central, eastern and southern China, with records from 18 Chinese provinces or equivalent administrative regions Fei et al., 2006;Huang, 1982;Wang et al., 2004). However, recent range-wide surveys have revealed that wild populations are now critically depleted or extirpated across China . Allopatric populations have traditionally been interpreted as the single widely distributed species Andrias davidianus, which is assessed as Critically Endangered by IUCN (Liang et al., 2004), but recent genetic analysis has shown that they constitute a complex of at least three species, including the South China giant salamander A.
Chinese giant salamanders are recognised as global conservation priorities on the basis of evolutionary history (Gumbs et al., 2018).
Anthropogenic modification of freshwater systems (e.g. pollutant emissions; alteration of flow regimes and water turbidity through damming) has reduced the availability of suitable giant salamander habitat (Dai et al., 2009;Wang et al., 2004). However, giant salamander populations have also been lost from unmodified habitats across China that still support diverse amphibian communities and abundant prey Tapley et al., 2015, in press). Giant salamanders were designated as a State 2 Protected Animal in China in 1988, with this national legislation making hunting or collection illegal without an official permit (Dai et al., 2009).
Despite their protected status, range-wide giant salamander declines have been attributed primarily to overexploitation (Tapley et al., in press).
The first western description of Chinese giant salamanders during the 19th century observed that 'such animals are a valuable alimentary resource for the inhabitants of the country' (Blanchard, 1871), and exploitation both for local consumption and for the domestic luxury food market occurred throughout the 20th century (Cunningham et al., 2016;Dai et al., 2009;Huang, 1982;Liu, 1950;Simoons, 1991;Sowerby, 1925aSowerby, , 1925bWang et al., 2004). The use of giant salamanders in traditional Chinese medicine (TCM) has been documented for over 2000 years (Strassberg, 2002), with a wide range of perceived medicinal benefits (He et al., 2018). However, the extent to which these historical interactions impacted wild populations is unclear. It is also suggested that cultural taboos, associated with the concept that giant salamanders were 'dirty' or bad luck, may have limited the effects of exploitation until recently (Cunningham et al., 2016). A vast-scale giant salamander farming industry has developed rapidly in China since the 2000s (Figure 1). Whereas some level of captive breeding has been achieved by salamander farms, this industry has led to further pressure on wild populations to supply breeding stock, as well as posing risks of pathogen transfer, competition and hybridisation from farm releases or escapes (Cunningham et al., 2016;Turvey et al., 2018;Yan et al., 2018).
Farms include large self-contained and privately owned facilities (company model); smaller-scale rearing of salamanders in individual stocking of farms with wild-caught animals, and our findings highlight an important gap in the effectiveness of China's conservation protection for some of its highest-priority threatened species. Tackling this problem will likely require multiple coordinated approaches, including enforcement of legislation, increased penalties for removing giant salamanders from the wild, permanent identification of captive-bred giant salamanders, and consumer-focused interventions to reduce urban demand.

K E Y W O R D S
Andrias, Chinese giant salamander, cultural taboos, illegal wildlife trade, interview survey, overexploitation, traditional Chinese medicine, wildlife farming households, often in association with company-model farms (smallholder model); and breeding co-operatives, often with a centralised facility (Cunningham et al., 2016).
Understanding the causes and impacts of these various pressures on wild populations, identifying which specific demands and user-groups are the greatest drivers of exploitation, and assessing the effectiveness of existing legislation and conservation management, are all necessary before appropriate mitigations can be developed to combat further giant salamander declines. However, numerous crucial aspects of giant salamander exploitation remain unclear, including whether population declines are recent or predate the farming industry; levels of stocking of wild-caught versus farm-bred individuals in farms, and the legality of exploitation of wild populations to stock farms; and what factors drive and regulate consumer demand, including the relative importance of consumption for food versus medicinal use, whether any cultural restrictions limit consumption, and why giant salamanders are considered desirable to eat in China. There is an urgent need for improved baselines to understand the dynamics and drivers of giant salamander decline, and the sustainability or otherwise of past and present interactions with wild populations.
In order to strengthen the scientific evidence-base on past and present activities associated with giant salamander declines, we

| MATERIAL AND ME THODS
Between May 2013 and June 2016, we conducted interviews with four different respondent groups in China: rural households living in close proximity to potential giant salamander habitat; company members in company-model salamander farms; county-level government fishery bureau officials; and urban respondents representing potential end-consumers (Table 1). Different standard questionnaires containing multiple open-ended questions were used for each respondent group (Text S1). All interviews were conducted in Chinese (either standard Mandarin or a local dialect) by the authors and Chinese field teams comprising staff and students from local academic institutions, who received training in standardised interview techniques before fieldwork commenced.
We explained that we were conducting interviews to research knowledge of local wildlife and aquatic resources, obtained informed verbal consent before all interviews, and informed respondents they could stop at any time. Verbal rather than written consent was obtained because many respondents had low levels of literacy, and/or were cautious about signing documents. Rural and urban questionnaires were anonymous; names and positions of farm company members and government officials were initially recorded to allow future follow-up if necessary, but were not re- Interviews with farm company members and government officials were mainly conducted in person, with some county-level fisheries staff contacted via email by province-level or municipality-level fisheries staff (i.e. staff at higher administrative levels) on our behalf.
Farms were identified through discussion with fisheries staff. Farm interviews gathered data on the history of the farm, stock composition and origin, and stocking preferences. Government interviews gathered data on local legislation and enforcement including reports of illegal poaching or trading, permits to collect wild salamanders, and numbers of licensed farms under their jurisdiction. Interviews with members of the general public living in urban centres were conducted in three provincial capital cities (Guiyang, Kunming, Xi'an), with respondents selected through random encounters for faceto-face interviews in parks, museums and residential districts. Urban interviews gathered data on demographic characteristics of respondents; whether they knew of anyone who had eaten giant salamander (to establish levels of urban salamander consumption across respondents' wider social networks and beyond their direct personal experience, e.g. if salamanders are preferentially consumed only at high-status banquets); and reasons why they might want to eat giant salamander. Additional data also collected during interviews have been analysed and reported elsewhere Turvey et al., 2018;Tapley et al., in press).
Salamander last-consumption dates were reported in a variety of formats (e.g. years ago; decadal ranges), and were converted to direct calendar years for analysis. Responses of 'when I was small' or 'when I was a child' were randomly assigned a date within the year range when the respondent was between 8 and 15 years old, and we converted all other records following Turvey, Bryant, et al. (2017) unless reported dates were too vague to include in analysis.
We constructed multivariate additive generalised linear models (GLMs) to investigate factors predicting the likelihood that: (a) rural respondents had eaten giant salamander; (b) urban respondents knew someone who had eaten giant salamander; and (c) urban respondents wanted to eat giant salamander because of specific reasons reported in >5% of interviews. All GLMs had binary responses ('yes'/'no'), binomial error structure and logit link function. GLMs investigating factors associated with rural respondents having eaten salamanders included respondent age, gender, ethnicity and province as fixed effects, and were conducted using two model structures: (a) including ethnicity as a binary variable (Han vs. non-Han ethnic groups); (b) including all ethnic groups represented in our dataset by ≥10 respondents as separate categories (n = 11). GLMs investigating urban respondent responses included respondent age, gender, education level and city as fixed effects. We also used a chisquare test to investigate whether rural respondents who reported that giant salamanders were 'unlucky' or 'bad luck' showed a different likelihood of reporting having eaten them. We analysed all data using r version 3.2.3 (R Development Core Team, 2015).
In total, 234 respondents (8.0%) reported that salamanders could be used for TCM, and 203 respondents reported specific uses.
These were most frequently associated with healing or modifying skin (often specifically using salamander skin or mucus), including treating burns (n = 118), skin diseases such as leprosy (n = 9) or wounds/cuts (n = 4), and improving one's looks (n = 14 To control for the fact that fewer respondents will have been alive (and thus able to eat salamanders) longer ago, data were recalculated relative to the total number of respondents who were alive in each decade, and retain a highest last consumption peak in the 1980s (Figure 3b).
Of these 12 counties, officials in only one (Xingwen County, Sichuan) had issued permits to collect wild salamanders, but officials in four counties provided reports of illegal salamander poaching or trading, with fines given for illegal poaching in one county.    or 'baby-fish' (supposedly so-called because giant salamanders sound like crying babies) is said to be common across China (Dai et al., 2009;Huang, 1982;Wang et al., 2004), this name was reported by a relatively small subset of respondents, and was often linked to a perceived association with dead babies rather than baby-like vocalisations. Older historical accounts also refer to differing local names (Liu, 1950;Strassberg, 2002 (Jiao et al., 2012;Zhang et al., 2020). Nuanced region-specific management policies may therefore be necessary to regulate local rural interactions with giant salamander populations effectively. However, extensive consumption is documented across the likely distributions of all recently recognised Chinese giant salamander species (Turvey et al., 2019;Yan et al., 2018).

| D ISCUSS I ON
Overexploitation for TCM poses a major threat to many threatened species, both within China and through international trade Cheng et al., 2017;O'Malley et al., 2017;Wong, 2019).
Rural respondents were aware of multiple medicinal uses for giant salamanders, which are similar to advertised uses for giant salamanders in modern-day commercial TCM (He et al., 2018). Overall, however, very few people in these communities have ever actually used these animals for medicinal purposes, indicating that such usage was probably never a major threat to wild populations. Conversely, a substantial number of rural respondents reported having eaten giant salamanders, and this reported level might represent an underestimate due to the potential sensitivity of this activity. Our results indicate that consumption of giant salamanders in rural communities was more common in the past, with about a quarter of older respondents having formerly eaten them in contrast to much lower numbers of younger respondents, and with a marked drop-off in consumption since the 1980s.
It is difficult to reconstruct population declines retrospectively in the absence of contemporary longitudinal monitoring data, and this shift in giant salamander consumption might reflect cultural or economic changes in China during recent history. For example, lowincome rural communities may have been more dependent upon wild protein sources before implementation of free-market reforms that prompted nation-wide economic growth from the 1980s onwards.
Indeed, respondents from both Shaanxi and Sichuan reported that giant salamanders were low-class meat, with another respondent from Sichuan reporting that giant salamander meat cost 0.6 yuan per kg in the 1970s and was far cheaper than pork. However, the midlate 20th century is regarded as a period of largely uncontrolled wildlife exploitation in China, suggesting that giant salamanders are also likely to have been overexploited during this period (Corlett, 2007;Lau et al., 2010;Turvey, Crees, et al., 2017). Our results thus provide indirect evidence that giant salamander populations in areas of good-quality habitat were probably already declining from at least the 1980s, due to local consumption and/or exploitation for trade, and several decades before the development of the modern farming industry.
Our results confirm that giant salamanders were considered 'dirty' and associated with bad luck in rural communities across many parts of China, usually due to a perceived association with dead children, and with species-specific cultural taboos reported against eating them by several respondents. Interestingly, some other 'liminal' animals are also associated with dead children (e.g. nightjars in historical English tradition; Greenoak, 1997). Comparable taboos are widespread across other social-ecological systems, and can play important conservation roles by influencing societal preferences and supporting sustainable behaviours at local scales (Colding & Folke, 2001;Jones et al., 2008;Lingard et al., 2003). However, our results also demonstrate that reported taboos against eating giant salamanders have apparently not been effective during recent history; the negative cultural values associated with these animals have not prevented local people from eating them, and respondent knowledge of such taboos is actually associated with a higher likelihood of past consumption. Although patterns and drivers of taboo stability or erosion are complex, increasing non-adherence to existing wildlife consumption taboos has been associated with rapid social change in other systems, as has taken place in China throughout recent decades (Golden & Comaroff, 2015).  (Gaillard et al., 2017;Shi et al., 2007Shi et al., , 2008. Comparable patterns of illegal acquisition of wild-caught stock to supply commercial farms are also seen across other exploited wildlife species in eastern and southeast Asia, such as in farming of porcupines Hystrix brachyura in Vietnam (Brooks et al., 2010).
There were 2,622 company-model salamander farms, of which 2,080 were legally licensed, across China at the end of 2013 (Li, 2015). Using this figure to extrapolate from the results of our farm  . However, we consider that this level of extraction of wild individuals is probably unsustainable for a long-lived, slow-reproducing apex predator (cf. Cheng, 1998;Huang, 1982).
Finally, our urban interviews reveal that consumption of giant salamanders in Chinese cities is relatively widespread, and shows a positive correlation with respondent education level that is likely associated with the luxury status of this commodity. Respondents in Guiyang, the smallest and least developed of the Chinese urban centres we surveyed, were more likely than respondents in Kunming or Xi'an to know someone who had eaten giant salamander and to want to eat giant salamander because of perceived good flavour.
This might reflect the high habitat suitability and likely former abundance of giant salamanders in Guizhou Province , and their documented local trade for food in Guiyang before the recent establishment of the salamander farming industry, suggesting a possible long local tradition of consumption (Sowerby, 1925a(Sowerby, , 1925b; however, over a quarter of respondents in Kunming or Xi'an also knew people who had eaten giant salamanders. Concerningly, the most commonly reported reason by urban respondents for wanting to eat a giant salamander was the animal's rarity or expensiveness. Other threatened species also constitute luxury foods in China, and rare meat types are often preferentially consumed as indicators of wealth and status in Asian cultures (Fabinyi, 2011;Fabinyi & Liu, 2014;Shairp et al., 2016); however, placing value on rarity can drive disproportionate exploitation of rare species, potentially leading to an extinction vortex (Courchamp et al., 2006). Whereas both demographic and geographic variables are associated with respondent knowledge of giant salamander consumption, the desire to eat salamanders due to either rarity or curiosity was not correlated with any of our variables. Our results thus suggest that this driver of demand for giant salamanders may be widespread across the potential urban consumer market in China, making it difficult to reduce such demand through educational targeting or other interventions aimed at specific user groups. Once again, these findings are similar to freshwater turtle trade and consumption dynamics in China, where wild turtle meat commands a higher price than farmed turtle meat (Gaillard et al., 2017;Shi et al., 2007).
Our comprehensive assessment of patterns, levels and drivers of interactions with giant salamanders by multiple user-groups across China establishes an important new baseline for assessing the impacts of past and present exploitation, and our findings raise worrying concerns for the future of wild populations of these remarkable animals. It is likely that giant salamander populations were already in decline several decades ago as a result of historical exploitation; and such exploitation has clearly intensified recently due to largely unregulated illegal hunting to stock and maintain large wild-caught salamander populations in farms at a country-wide scale to support demand by urban consumers for high-prestige rare meat. These findings highlight an important and concerning gap in the effectiveness of China's existing conservation protection for some of its highestpriority threatened species; they will be used to inform updated extinction risk assessments and national conservation action planning for Chinese giant salamander species, and to develop regional and national management interventions that involve both government and civil society. Tackling this problem will likely require multiple coordinated approaches, including further research into sustainable salamander offtake levels and how to counter consumer preference for rare species, strengthened enforcement of existing legislation to prevent ongoing poaching, increased penalties for removing giant salamanders from the wild, and the permanent identification of captive-bred giant salamanders. Existing wild-caught farm stocks could be used to establish separately managed species-specific conservation breeding programmes should wild founding stock be unavailable (Cunningham et al., 2016;Turvey et al., 2018). We also recommend consumer-focused interventions to reduce urban demand, for example by encouraging socially responsible consumption (Yan & She, 2011).
Our research was conducted after Chinese president Xi Jinping initiated an ongoing nation-wide anti-corruption campaign in 2012, which targeted consumption of high-prestige rare animal products at official banquets; the high level of exploitation of giant salamanders demonstrated in our study is thus from the period when demand for consumption at banquets had likely already declined. We note that the COVID-19 pandemic has led to further restrictions on trade in rare and luxury wild animal products in China. Although these restrictions do not apply specifically to giant salamanders, we hope that the new regulations will benefit giant salamander conservation by reducing wider exploitation, demand and perceptions of wild meat. However, additional targeted and proactive interventions are now imperative to reduce the ongoing pressure on wild populations of giant salamanders across China, in order to save these unique 'living fossils'.

ACK N OWLED G EM ENTS
Funding was provided by the Darwin Initiative (Project No. 19-003), the National Natural Science Foundation of China (31360144, 31860600), Ocean Park Conservation Foundation Hong Kong and ZSL's EDGE of Existence programme. We thank all field teams for participating in surveys.

CO N FLI C T O F I NTE R E S T
The authors report no conflict of interest.

AUTH O R S ' CO NTR I B UTI O N S
S.T.T., A.A.C., S.C. and B.T. designed research; S.C., A.A.C. and S.T.T. coordinated data collection, and all authors collected data; S.T.T. interpreted and analysed data, and wrote the paper with assistance from A.A.C. and B.T.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets supporting this paper (interview data for different respondent groups in China) are available in the Supporting