Disordered Translocation is Hastening Local Extinction of the Chinese Giant Salamander

Guocheng SHU ,Ping LIU ,Tian ZHAO ,Cheng LI ,Yinmeng HOU,4 ,Chunlin ZHAO,4 ,Jie WANG,Xiaoxiao SHU,4,Jiang CHANG,Jianping JIANG＊ and Feng XIE＊

1CAS Key Laboratory of Mountain Ecological Restoration and Bioresource Utilization &Ecological Restoration Biodiversity Conservation Key Laboratory of Sichuan Province,Chengdu Institute of Biology,Chinese Academy of Sciences,Chengdu 610041,Sichuan,China

2Yibin University,Yibin 644000,Sichuan,China

3University of Chinese Academy of Sciences,Beijing 100049,China

4Sichuan University,Chengdu 610065,Sichuan,China

5State Key Laboratory of Environmental Criteria and Risk Assessment,Chinese Research Academy of Environmental Sciences,Beijing 100012,China

Abstract Biodiversity is declining globally by an unprecedented extinction rate.This is especially true for amphibians,accounting for 24.3% of all threatened vertebrates.As the largest extant amphibian species in the world,wild populations of the Chinese giant salamander (Genus Andrias) (CGS) have decreased dramatically because of overexploitation and habitat degradation.Translocation has become an important strategy for restoring threatened wild populations worldwide.However,disordered translocation usually has negative effects on the native populations.We provide an overview of CGS translocation and show that disordered translocation can increase local population extinction.Nearly four times the estimated number of wild individuals have been released across China.There are three types of translocation used for CGS,namely,reinforcement,reintroduction and ecological replacement,the last of which accounts for over one-third of translocations.Our genetic screening revealed that most released individuals were not from local populations,with one to four lineages detected in every release site (n =6).This disordered translocation can potentially reduce the genetic integrity of original populations.Hence,we suggest suspending current CGS translocation activities immediately,until more robust measures can be developed and implemented to improve the current translocation program,especially with respect to lineage identification and the identification of appropriate release sites.

Keywords Chinese giant salamander,genetic test,translocation,conservation,wild population

1.Introduction

Global biodiversity is in decline and we are currently experiencing unprecedented rates of extinction,for example,current rates of extinction are about 1000 times the likely background rate of extinction and future rates are likely to be 10 000 times higher (Pimm

et al.

,2014).Amphibians are disproportionately the most threatened vertebrate group,with 41% of species considered threatened (Hoffmann

et al

.,2010).In the face of rapid biodiversity loss,effective strategies for the conservation of threatened species are urgently required(Bainbridge

et al

.,2014;Taylor

et al

.,2017).Translocation has become an important tool to manage wildlife over the past few decades,which involves the humanmediated,intentional movements of living organisms from one area to another (Armstrong and Seddon,2007;IUCN/SSC,2013;Batson

et al

.,2015;Swan

et al

.,2019).If the primary objective is a conservation benefit,the translocation is called a conservation translocation.A conservation translocation is usually undertaken with the goal of improving the conservation status of the target species locally or globally,and/or restoring natural ecosystem functions or processes (IUCN/SSC,2013).Whilst translocations are being used with increased frequency,there is an obvious bias towards mammalian and avian taxa(Seddon

et al

.,2005;Bajomi

et al

.,2010).The Chinese giant salamander (CGS) have attracted widespread attention because they represent an evolutionary ancient and distinct amphibian lineage and have few close relatives.Also,they are the largest extant amphibian species in the world,with a maximum length and weight of 2 m and 50 kg,respectively (Zhao,1998).Historically,this species was widely distributed in the middle and lower tributaries of the Yangtze River,Yellow River,and Pearl River,including 17 provinces(or municipalities) in central,southern and southwestern China (Wang

et al

.,2004).However,the CGS populations have declined dramatically since the 1950s due to habitat loss and overharvesting for food (Wang

et al

.,2004).The CGS is listed as a National Special Protected Animal (category II) and as Critically Endangered on the IUCN Red List of Threatened Species (Liang

et al

.,2004),Biodiversity Red List of China and in included in CITES Appendix I (Jiang

et al

.,2016;CITES,2020).Therefore,the Chinese government has heavily invested to protect and conserve the CGS.So far,there are more than fifty nature reserves associated with protection of CGS in China(Liang

et al.

,2013).Hunting wild CGS is prohibited and breeding the species in captivity requires permission.Recent field surveys have shown that the CGS has been critically depleted or extirpated in most of their original habitats (Xu

et al

.,2018;Turvey

et al

.,2018).With the development of CGS farming industry,the CGS farms can supply sufficient source populations for translocation programs of CGS,which provides hope and opportunity to save this threatened species.By the end of 2014,there were 2622 CGS farms in China and up to 1 249 000 individuals across the country (China Aquatic Wildlife Conservation,2015).As a result,government-promoted translocations for the protection of the CGS have been conducted widely in China.In 2014,the Ministry of Agriculture and Rural Affairs of the People’s Republic of China (MARAPRC) enacted the technical specification for the stock enhancement of Chinese giant salamander (SC/T 9414-2014).A lack of detailed knowledge on CGS population genetics meant that previous translocations did not consider CGS from different river basins represented as different lineages and the species was treated as a single population.Current studies revealed that the CGS is comprised of at least seven lineages (Yan

et al

.,2018;Liang

et al

.,2019).Turvey

et al

.(2019) argued that the wild populations from Pearl River/Nanling regions should be restored as a valid species

Andrias sligoi

(Boulenger,1924).Mix-source reintroductions can lead to outbreeding depression in hybrid offspring or reduce the genetic integrity of the original populations (Huff

et al

.,2011).Therefore,a detailed understanding of CGS translocations is important as it will guide future conservation management decisions.

In this study,we aimed to (a) understand the current CGS translocations including the number of individuals involved and the locations where CGS were released,and the translocation type,(b) to conduct genetic screening on released individuals and assigned them to CGS lineages or species in order to assess the effectiveness and risk of current translocations,and (c) to make recommendations for future CGS translocations.

2.Materials and Methods

2.1.Release data collection

The CGS often grabs the attention of the general public when released because it is regarded as a flagship species in China.Therefore,all CGS translocation news are generally released to the public.Data on CGS translocation were obtained from the internet (mainly media outlets) using the search terms Chinese giant salamander,giant salamander,and baby fish,along with the keywords reintroduction,translocation,release,and stock enhancement.The time frame of the search was from the last century to the end of October 2019.The language of the search was Chinese (Mandarin).We also obtained supplementary data from two fishery bureaus,including the Fishery Bureau of Tongjiang County,Sichuan and the Fishery Bureau of Cili County,Hunan.Following the definitions of IUCN translocations,we classified CGS translocations into 3 different types.Reinforcement,as the intentional movement and release of CGS into these existing populations of conspecifics;reintroduction,as the intentional movement and release of CGS inside their indigenous ranges from which they have disappeared;ecological replacement,as the intentional movement and release of CGS outside their indigenous ranges to perform a specific ecological function.

2.2.Survey area and methodology

We conducted sampling surveys at six release sites:Kaihua (4532 released individuals),Zhejiang;Yuexi (1890 released individuals),Anhui;Liannan(400 released individuals),Guangdong;Heyuan (1480 released individuals),Guangdong;Tongjiang (4600 released individuals),Sichuan,and Wanyuan (110 released individuals),Sichuan.These sites represent three distinct distribution regions of Chinese giant salamander in Turvey

et al.

(2019):the southwestern(Pearl/Nanling),northern (Yangtze/Sichuan) and southeastern(Huangshan and Zhejiang).Released individuals were collected in natural habitat through visual encounter methods and wire-mesh baited traps in September 2018 and 2019.The visual encounter methods consisted of wading,nocturnal spotlighting,and netting.Wire-mesh baited traps are reasonably effective for surveying

Andrias japonicus

and

Cryptobranchus alleganiensis

alleganiensis

(Foster

et al

.,2008;Brigger

et al

.,2013;Turvey

et al

.,2018).We selected survey areas according to habitat conditions during the day and searched for CGSsin the evening (from 18:30 to 23:00).The traps consisted of 110 cm × 60 cm × 30 cm rectangular boxes made of 1.5 cm plastic-coated hardware cloth and a funnel trap.Fresh fish and chicken were used as bait for trapping (Browne

et al

.,2011).The traps were set in the upper,middle,and lower reaches of the surveyed river and checked them the next day (between 5:00 and 7:00 AM).

2.3.Habitat suitability model

To infer the potential distribution range of the CGS under current climatic conditions,we produced a habitat suitability model according to CGS locality records and environmental data (Chen

et al

.,2018).We combined GPS information on the current distribution of the species and environmental variable modeling to project its historical distribution.For distribution information,we obtained sampling records of wild individuals from the literatures (Murphy

et al

.,2000;Fei

et al

.,2006;Zheng,2006;Li

et al

.,2009;Luo

et al

.,2009;Guo

et al

.,2011;Yang

et al

.,2011;Chen

et al

.,2017;Wu

et al

.,2017;Wang

et al

.,2018;Liang

et al

.,2019) and the Global Biodiversity Information Facility (GBIF;https://www.gbif.org) and excluded duplicate GPS sites within each pixel (30 arc-seconds,1 km ×1 km).Climatic predictor variables were downloaded from WorldClim (www.worldclim.org/current).We downloaded maps of Chinese administrative areas,elevation,and land use from DIVA-GIS (available at http://www.diva-gis.org/Data).Vegetation cover was extracted from the land use data.Human footprint data were downloaded from http://sedac.ciesin.columbia.edu/wildareas/.To ensure consistency in climate layer resolution,we increased the resolution of last glacial maximum(LGM) layers from 2.5 min to 30 arc-seconds by the ‘resample’method in ArcGIS v.10.3.According to the sampling records,all of China was selected as the range of interest.

Before constructing the habitat suitability model,bioclimatic raster layers were clipped into the study area and converted to ASC II format for use in MaxEnt.To prevent overfitting,we used the band collection statistic of the spatial analysis tool to calculate the correlation coefficients between bioclimatic layers in ArcGIS v.10.3 and removed the variables with strong correlations (correlation coefficient ≥ 0.8) and low contributions to the habitat suitability model with 19 bioclimatic factors used for preliminary analysis.Twelve bioclimatic predictor variables were ultimately included in our habitat suitability model:elevation,vegetation cover,human footprint data,annual mean temperature,mean diurnal range,isothermality,temperature seasonality,maximum temperature of the warmest month,minimum temperature of the coldest month,annual precipitation,precipitation seasonality,and precipitation of the coldest quarter.

The habitat suitability model was constructed by MaxEnt v.3.4.1 (Phillips

et al

.,2018) with 8 uncorrelated bioclimatic data layers using the maximum entropy method based on the optimal model,with 25% of occurrence records as the test set and 75% as the training set.Other conditions included 10 000 background points,a logistic output format and 10 crossvalidation replicates for each run,with the average of those replicate models representing the final prediction results.The area under the receiver operating characteristic curve (AUC)was used to evaluate the overall fit of the model,and AUC values ＞ 0.9 indicated that the model fit well (Araujo and Guisan,2006);the higher the AUC score is,the stronger the predictive power of the model.Jackknife analysis was employed to identify which bioclimatic factors were most important and contributed the most to the model.

2.4.Molecular sampling and sequencing

We swabbed oral mucosa cells or collected exfoliated skin cuticles from caught and trapped individuals.All animals were released after tissue collection at the original site of capture.The collected tissues were immediately stored in 95% ethanol at -20°C for DNA extraction.DNA was extracted using a TIANamp Genomic DNA Kit (Qiagen).Three mitochondrial DNA (mtDNA) fragments from the complete mitochondrial cytochrome b (

Cytb

) gene,the partial control region (D-loop),and the partial cytochrome C oxidase 1 (

COI

) gene were amplified by polymerase chain reaction (PCR).The primers and cycling parameters were as described by literatures (Tao

et al

.,2005,2006;Che

et al

.,2012).PCR products were purified using a Qiagen PCR purification kit and sequenced on an ABI 3100 automated sequencer.We also downloaded nucleotide sequences available in recent studies (Yan

et al

.,2018;Liang

et al

.,2019) and other references from GenBank (Table S1) and aligned them with ClustalW built into BioEdit version 7.1.9 (Thompson

et al

.,1997) using the default parameters.Alignments were also visually checked for ambiguous alignment in MEGA version 7.0.26 (Tamura

et al

.,2013).All new sequences were deposited in GenBank with the accession numbers listed in Table S1.

2.5.Phylogenetic inference

Bayesian inference (BI) was used to reconstruct phylogenetic relationships.The optimal partitioning scheme and best-fitting nucleotide substitution model for each partition of the mtDNA genomes were estimated using PartitionFinder v.2.1.1 (Lanfear

et al

.,2016).Bayesian analyses were conducted in MrBayes v.3.2.2 (Ronquist

et al

.,2012) executed in the CIPRES Science Gateway (http://www.phylo.org/index.php).All searches were performed with two independent runs,each initiated with a random tree.Each run consisted of four Markov chain Monte Carlo (MCMC) chains (three heated chains and one cold chain) estimated for 5 million generations and sampled every 1000 generations.Stationarity was assessed in Tracer 1.6,and the first 25% of samples were discarded as burn-in (Rambaut

et al

.,2014).The remaining trees were used to create a 50% majority-rule consensus tree and estimate Bayesian posterior probabilities (BPPs).

3.Results

We collected a total of 268 records of CGS translocations from official reports (internet searches,233 records),and fishery stations (35 records),and found that the first releases were undertaken in the Zhangjiajie National Nature Reserve,in Hunan Province,in 2002.By the end of October 2019,a total of 287 840 farmed individuals were released into the wild across 98 counties and cities (Figure 1A,1C) in central and southern China.All records of CGS being released were distrbuted in 16 provinces or municipalities (Figure 1B).In fact,the numbers of released animals may likely be an underestimate as not all releases were reported.Even so,this number is almost four times the estimated wild population size of

Andrias

species in China (75 000) (Liang

et al

.,2013).The number of released animals increased gradually between 2008 and 2012.This was followed by a sharp increase in the number of released animals between 2013 and 2015.The number of released animals peaked in 2016 and has subsequently declined (Figure 1A).The greatest number of individuals were released in Shaanxi Province since 2002,reaching up to 94 464 in total;by contrast,the minimum number of released individuals occurred in Beijing,where 200 individuals were released (Figure 1B).

There are three types of CGS translocation according to definitions of IUCN:reinforcement (40.8%),reintroduction(24.5%) and ecological replacement (34.7%) (Figure 1C).This meant more than one-third of releases took place at sites outside the indigenous range of CGS.Our habitat suitability model showed that the suitable habitat of CGS was concentrated in central and southern China and was scattered in the east.However,more than 10% of the release sites were not inside predicted suitable habitat (Figure 1C).

Figure 1 Translocation overview of the Chinese giant salamander.(A) Trends in the translocations of Chinese giant salamanders from 2002-2019.(B)The number of translocated giant salamanders across different Provinces in China.(C) Translocation sites (triangles) and types of the Chinese giant salamanders.The green areas present the predicted suitable habitat for Chinese giant salamander.The pie chart shows the proportion of three translocations.

In total,48 CGS individuals were recaptured from five of the six sites we sampled for released animals.The details of sites and number of individuals captured at each site were as followed:Yuexi,Anhui (5 adults,2 larvae);Liannan,Guangdong (3 adults,2 larvae);Kaihua,Zhejiang (28 adults,5 larvae);Heyuan,Guangdong (1 adult,1 larva) and Wanyuan,Sichuan (1 adult).Our molecular data showed that all the animals we captured could be assigned to non-native lineages defined by recent studies (Yan

et al

.,2018;Liang

et al

.,2019).Released individuals from Kaihua belonged to four non-native clades,namely,B,C,D and U1,described in Yan

et al

.(2018) (Figure 2).A nonindigenous lineage from Shaanxi was released in Liannan and Heyuan (B clade;Figure 2).Also,Wanyuan’s animal clustered into a clade with northern Guangxi (U1 clade;Figure 2),and Yuexi’s individuals represent three different non-local clades(U1,U2 and B;Figure 2).

Figure 2 Bayesian tree based on mtDNA and six sampling release sites of Chinese giant salamander.(A) A simplified Bayesian inference tree based on concatenated mtDNA (Cytb,D-loop and COI).Numbers near branches indicate Bayesian posterior probabilities.The capital letters outside and inside the brackets show corresponding clades for Chinese giant salamander in Yanet al. (2018）and Liang et al.(2019),respectively.Clade D corresponds to A.sligoi in Turvey et al. (2019).The distribution areas for different clades are marked after subulate branches.(B) The sampling release sites of Chinese giant salamander.The numbers represent sampling sites,including Wanyuan,Tongjiang,Yuexi,Kaihua,Liannan,and Heyuan in numerical order.The colored circles correspond to the different clades in the left.

4.Discussion

4.1.Factors that may influence CGS translocation

The success of amphibian translocations are related to multiple factors,such as habitat condition,age,food limitation,disease,the number of animals released,etc.(Griffiths and Pavajeau,2008;Germano and Bishop,2009).Generally,translocations are most successful when more than 1000 individuals were released (Germano and Bishop,2009).On the contrary,most reports indicated that causes of amphibian translocation failure were homing and migration of introduced individuals out of release sites and poor habitat (Germano and Bishop,2009).Current studies show that some CGS translocations have not succeeded in restoring wild CGS populations (Luo

et al

.,2009;Lin

et al

.,2017;Lu

et al

.,2020).Turvey

et al.

(2018) reported that dead individuals of CGS were found at one translocation site.Researchers suggested that most translocations of CGS failed to restore the wild populations due to the limitation of factors such as water quality,food supply,human disturbance,release season,individual age,poaching,and subsequent management measures (Luo

et al

.,2009;Lin

et al

.,2017;Zhang

et al

.,2017;Turvey

et al

.,2018).Thus,a variety of factors influence the success rate of translocation of CGS.So far,the number of released individuals of CGS (287 840) has far exceeded the number of wild individuals (75 000) (Liang

et al

.,2013),but release sites are outside the indigenous range (ecological replacement) and outside the predicted suitable habitat which accounted for more than 33% and 10% of sites,respectively.In general,our habitat suitability model is similar to the previously published model (Chen

et al

.,2018),both models predicted that suitable CGS habitat was primarily located in central and southern China.However,there were some differences between our model and the Chen

et al

.(2018) model.Our suitable habitats were mainly distributed in Shaanxi,Guizhou,Hunan,Sichuan,Chongqing,Hubei,Henan,and Zhejiang,while the suitable habitats from Chen

et al

.(2018)were concentrated in Guizhou,Hunan,Zhejiang,Hubei,Anhui,and Sichuan.This difference could be the result of the additional predictive variables (i.e.human footprint) and the inclusion of more historical records in our model.In fact,the conservation biologists’ original intention was to restore wild CGS population by reinforcement or reintroduction and not to conduct ecological replacement.Currently,there are no good management measures (e.g.monitoring and habitat protection)after releasing CGS back into the wild.Therefore,whether these translocations will succeed or not is completely uncertain and requires mark-recapture studies.In recent years,the scale of CGS translocation has declined since its peak in 2016.This decline might be associated with fluctuations of CGS market value.From 2008 to 2010,the CGS price gradually increased and reached peak at the end of 2010,when the highest price was 3000 RMB/kg (Cheng,2015).During this period the CGS industry expanded and was able to provide a greater number of both larvae and adult animals for translocation.In 2011,driven by huge profits,many farmers started to raise CGS (Cheng,2015),and since 2012,the price of CGS drastically declined (Li

et al

.,2018).However,because the sexual maturity cycle of captive takes about 4 years (Luo

et al

.,2005).The supply of salamander exceeded the demand in 2016,which was when the CGS translocations reached the largest scale.Meanwhile,the price of CGS has almost bottomed out in 2017 (140 RMB/kg),therefore the CGS farming industry has gone into decline (Li

et al

.,2018).Subsequently,the number of CGS translocations also declined.

4.2.The risk of current CGS translocation

All the populations of CGS were treated as a single species during the previous translocations,but recent studies revealed that there are at least seven distinct lineages (Yan

et al

.,2018;Liang

et al

.,2019).Furthermore,a new study revealed that the wild populations from Pearl River/Nanling regions were a valid species

Andrias sligoi

(Turvey

et al

.,2019).In this study,even though we only recaptured 48 individuals from six release sites(Kaihua,Yuexi,Liannan,Heyuan,Tongjiang,and Wanyuan),all individuals belonged to non-native genetic lineages.For example,the native wild populations from Zhejiang belong to clade E according to the recent results (Yan

et al

.,2018;Liang

et al

.,2019),however,all sampled released individuals were divided into four non-native genetic lineages (B,C,D and U1) in Kaihua,Zhejiang,where the clade D is the representative of

Andrias sligoi

.This is quite contrary the genetic purity of karst cave populations (Wang

et al

.,2017).We don’t know if our captures were all the original released individuals or their offspring that were born in the wild.But our results suggest that some released individuals have survived at least four years (e.g.952 individuals were released in Kaihua,Zhejiang in October 2015.)and could compromise the genetic integrity of the native population.Meanwhile,some surveys suggest that the released individuals have established new populations in the wild,such as in Guangdong,but no evidence suggested that interbreeding has occurred between the released and wild populations (Su,2018).However,interbreeding has already occurred in

Andrias japonicus

in Japan owing to the introduction of

Andrias davidianus

(Wang

et al

.,2015),which resulted in severe genetic contamination in the wild.With so many non-native genetic lineages identified at our sampling sites,there is a strong indication that the extensive commercial farming has led to the movement of animals between farms throughout the range of CGS in China (Yan

et al

.,2018),and their progeny are being released into the wild without genetic testing.Due to their overwhelming advantage of population size,there is a great possibility that non-native strains or species may eventually eradicate the unique evolutionary characteristics of original populations.Although some hybrids are more fit than their parents and the pursuit of hybrid vigor or “heterosis” has been important in agriculture and animal breeding (Edmands,1999),empirical work indicates that multi-generational outbreeding depression can be sufficiently severe in some cases (e.g.,Houde

et al

.,2011;Huff

et al

.,2011).In addition,mixing genetically,divergent sources are often proposed to attenuate genetic diversity in reintroduced populations that may result from small effective population sizes (Huff

et al

.,2011).However,outbreeding depression is a possible negative tradeoff in hybrid offspring for mixing sources (Huff

et al

.,2011).Outbreeding becomes unhelpful with regard to the restoration of the wild population of CGS due to adaptability decline in hybrid offspring.Therefore,the current mixed-source translocation could accelerate the extinction of wild local CGS.

Besides,the release of non-native lineages may lead to the risk of ecological invasion.In our investigation,more than onethird of release sites were not within the historical distribution of CGS according to the historical records.So,these released individuals should be treated as an alien invasive species at these sites.There may be competitive or predator-prey relationships between released individuals and local aquatic animals and this could destroy the local ecological balance.

4.3.Translocation suggestions

Based on our findings,we suggest that the administrative department suspend CGS translocation and collect genetic information for all released populations across the country in order to take immediate remedial measures.Meanwhile,for standardizing the procedures of CGS translocation,we recommend that the following points should be considered:

Establish dedicated conservation breeding facilities.

Although CGS farms can be used for translocation programs,commercial breeding activities have resulted in extensive movements of CGS between different farms (Cunningham

et al

.,2016).Some evidence suggests that the introgression has occurred in some sampled individuals from different localities(Yan

et al

.,2018).This study showed that different non-native lineages were present at the same translocation site,where released animals were all from CGS farms.Furthermore,the stocks held on the private farms may be unsuitable for translocation due to the potential risk of pathogens (e.g.

Andrias

davidianus ranavirus

,

Aeromonas hydrophila

) and the unknown provenance of the salamanders held on the farms.Therefore,establishing dedicated conservation breeding facilities for CGS of known provenance should be conducted urgently (Turvey

et al

.,2018,2019).Current release programs must improve to identify the origin of captive animals and prevent extramural introductions of non-native lineages or different species (Turvey

et al

.,2019).

Pre-preparation for releasing.

Best practices for CGS translocation needs to involve selection of appropriate release sites (Luo

et al

.,2009;Lin

et al

.,2017),assessment of habitat (Lin

et al

.,2017),selection of animals to be released (Lin

et al

.,2017),genetic screening of animals to be released (Lin

et al

.,2017),prerelease health screening (SC/T 9414-2014),determination of release scale and time (Luo

et al

.,2009),adaptability training for releasing individuals (Gao

et al

.,2017;Kenison &Williams,2018),measurement of body parameters,marking with passive integrated transponder (PIT) tags (Dodd,2009;Liu

et al

.,2020)and preparation of appropriate food (e.g.small fish,shrimp)(Luo

et al

.,2009).Since the 1950s,and especially in the 1980s or human activities have caused damage,loss and reduction of

Andrias davidianus

habitat (Zhang

et al

.,2002).Our results indicate that some historical sites where CGS were present are no longer suitable for CGS due to the expansion of the human footprint and pollution.Release sites should be chosen according to the current distribution ranges and habitat suitability model,rather than historical distribution records of CGS.After determining the appropriate translocation sites,habitat assessement,including water quality,food supply,human disturbance and predators are required,because these environmental factors can affect the survival,growth and development (Luo

et al

.,2009;Zhang

et al

.,2017).Released individuals need be to strictly selected,as older individuals have higher survival (Luo

et al

.,2009;Zhang

et al

.,2016).As mentioned above,all animals destined for release should be subject to genetic screening to avoid genetic contamination.To increase survival,we suggest that the suitable release time in autumn and spring,because the food is plentiful and the water temperature is suitable during this period (Luo

et al

.,2009;Lin

et al

.,2017).Sex ratios of the released individuals should be between 1:1 and 1:1.5,which is consistent with the sex ratio of natural reproduction of giant salamanders (Luo

et al

.,2009).The number of translocated individuals should be determined by the carrying capacity of release site for CGS,because the translocation may fail if the scale of release is too large.The healthy and experienced individuals will likely adapt to new environments quickly.Quarantine and adaptability training are also critical for increasing the survival (Gao

et al

.,2017).Adaptability training mainly includes acclimatization to foods commonly found in their proposed release sites,the acclimatization to an array of environmental parameters that individuals will be exposed to at release sites (e.g.running water,temperature and photoperiod,etc.Gao

et al

.,2017),and behavioral adaptation (Crane and Mathis,2011;Gao

et al

.,2017).Body condition should be assessed for translocated animals,therefore the development and validation of a body scoring system is an important area of future research.Lastly,released individuals could be traced if they are marked with passive integrated transponders (PIT).

Releasing and post release

.Releasing cohorts should comprise of individuals of the same size or age class.Continuously track and monitor the released animals.We suggest that the animals should be monitored twice a week after releasing,and the monitoring frequency can be reduced after a month(based on our preliminary translocation project of CGS).The follow-up monitoring should continue for 5 years,since CGSs take at least three or four years to mature (Luo

et al

.,2005).Meanwhile,protecting the release sites from habitat destruction and poaching is essential.Any CGS deaths at release sites should be investigated and if possible,mitigated.Finally,all release programs should be systematically evaluated in order to continuously improve the translocations of CGS.Currently there is little evidence that CGS translocation results in the establishment of viable populations of native CGS;we suggest that translocations can only be deemed successful once there is evidence of breeding for several generations and the establishment of self-sustaining populations that have originated from translocated stock.

Acknowledgements

We would like to thank an anonymous reviewer and Nicholas Wu for their useful comments and suggestions.We thank all of the field assistants for participating in surveys and sampling.We appreciate the help in collecting data from the Sichuan Provincial Department of Agriculture and Rural Affairs,Gutian Mountain National Nature Reserve,Liannan Chinese Giant Salamander Provincial Nature Reserve，Heyuan Chinese Giant Salamander Provincial Nature Reserve,Heyuan Guishan Chinese Giant Salamander Municipal Nature Reserve.This work was supported by the National Key Programme of Research and Development,Ministry of Science and Technology (2016YFC0503200);the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA23080101);the Biodiversity Survey and Assessment Project of the Ministry of Ecology and Environment,China (2019HJ2096001006);the Open Project of CAS Key Laboratory of Mountain Ecological Restoration and Bioresource Utilization &Ecological Restoration Biodiversity Conservation Key Laboratory of Sichuan Province (KXYSWS2006);the Construction of Basic Conditions Platform of Sichuan Science and Technology Department(2019JDPT0020);and China Biodiversity Observation Networks (Sino BON).

Appendix

Table S1 Detailed information of samples and referenced sequences for genetic analysis in this study.Letters in front of brackets are corresponding clades in Yan et al. (2018).The capital letters in brackets present corresponding clades in Liang et al.(2019).

(Continued Table S1)

(Continued Table S1)