Introduction
The Yangtze giant softshell turtle Rafetus swinhoei (Trionychidae; Gray, Reference Gray1873) is one of the largest freshwater turtles, with some males weighing > 160 kg. The species occurs exclusively in the Red River basin in China and Viet Nam, and the lower Yangtze River floodplain in China (Pritchard, Reference Pritchard2012; Wang et al., Reference Wang, Shi, Wen and Han2013) and is of particular cultural significance in Vietnamese mythology and folklore (Bettelheim, Reference Bettelheim2012). Until recently, R. swinhoei has been known primarily for cultural reasons (Pritchard, Reference Pritchard2012), with few studies describing the species’ ecology and natural history (Gray, Reference Gray1873; Heude, Reference Heude1880; Meylan & Webb, Reference Meylan and Webb1988; Pritchard, Reference Pritchard2001, Reference Pritchard2005). Large dams have been constructed within the species’ range, and potential negative effects, including impacts on the nesting habitats of the species and effects caused by isolation and overfishing, were not considered (Le Duc et al., Reference Le Duc, Pham, Zuklin, Bordes, Leprince and Ducotterd2020b). Rafetus swinhoei is categorized as Critically Endangered on the IUCN Red List (Fong et al., Reference Fong, Hoang, Kuchling, Li, Mccormack, Rao, Timmins and Wang2021) and is one of the most threatened turtle species globally (Stanford et al., Reference Stanford, Iverson, Rhodin, van Dijk, Mittermeier and Kuchling2020). It began to decline in the 1980s, and became rare in the 1990s as a result of habitat alteration and intensive capture for local consumption (Le Duc et al., Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a). Currently only one male individual remains in captivity, in China, and there is one known individual in the wild in Viet Nam, which was recently identified as a female (Le Duc et al., Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a; Stanford et al., Reference Stanford, Iverson, Rhodin, van Dijk, Mittermeier and Kuchling2020, WCS, 2020). Intensive efforts to locate more individuals in both Viet Nam and China have failed (Wang et al., Reference Wang, Shi, Wen and Han2013; Kuchling et al., Reference Kuchling, Lu and Rao2014; Pham et al., Reference Pham, Le Duc, Leprince, Bordes, Luu and Luiselli2020). Most data available on the species is in unpublished reports and documents, or has been published in unverified reports in social media. In China, R. swinhoei has reportedly never been observed in the wild (Stanford et al., Reference Stanford, Rhodin, van Dijk, Horne, Blanck and Goode2018). However, the species was historically reported to occur in the Red River, c. 300 km from Yuanjiang County (Wang et al., Reference Wang, Shi, Wen and Han2013). The species is also believed to have been present along the Yangtze River until a few decades ago (Pritchard, Reference Pritchard2012), but there is no evidence confirming its current presence in this river (Wang et al., Reference Wang, Shi, Wen and Han2013). Trapping and observation surveys have been conducted in China (Wang et al., Reference Wang, Shi, Wen and Han2013; Kuchling et al., Reference Kuchling, Lu and Rao2014), but only limited information from the surveys has been published.
Interview surveys have been carried out in both Viet Nam (Asian Turtle Program, Reference .2008) and China (Kuchling et al., Reference Kuchling, Lu and Rao2014). The results of these interviews (> 3,000 according to Timmins, 2016, unpubl. data) have not been published, the raw data are unavailable and the interview questions are unknown (Anadón et al., Reference Anadón, Giménez, Ballestar and Pérez2009; Turvey et al., Reference Turvey, Trung, Quyet, Nhu, Thoai and Tuan2015). However, information about the ecology and population biology of R. swinhoei has been obtained using local ecological knowledge, gathered in standardized interviews of former hunters of R. swinhoei in Viet Nam, raising the possibility that there may be more individuals in the wild (Le Duc et al., Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a,Reference Le Duc, Pham, Zuklin, Bordes, Leprince and Ducotterdb; Pham et al., Reference Pham, Le Duc, Leprince, Bordes, Luu and Luiselli2020). Several priority sites have been identified where trapping/observational surveys are urgently needed to confirm the presence of the species (Le Duc et al., Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a,Reference Le Duc, Pham, Zuklin, Bordes, Leprince and Ducotterdb; Pham et al., Reference Pham, Le Duc, Leprince, Bordes, Luu and Luiselli2020).
Here, we present the results of interview surveys with fishers on two river systems in Viet Nam where interviews have not previously been undertaken, categorize sites according to the likelihood that R. swinhoei is present, and recommend further research actions.
Study area
We conducted our survey in areas along the Da River (Black River), from Son La dam across the provinces of Son La, Lai Chau and Dien Bien, covering the entire 500 km of the Da River in Viet Nam (for interview data from areas along the Da River outside Viet Nam, see Le Duc et al., Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a). The Da River (total length 910 km) originates in Yunnan province in China and flows through Viet Nam before joining the Red River in Hanoi (Dao, Reference Dao2010; Vinh et al., Reference Vinh, Ouillon, Thanh and Chu2014; Le Duc et al., Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a).
Methods
We conducted 80 interviews during 13–24 November 2019, using the same questionnaire (Supplementary Material 1) and methodology as Le Duc et al. (Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a). Based on interviews with former hunters, fishers and local people during two periods in four areas (this study and Le Duc et al., Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a), we assigned a presence score to each site: score = 1: confirmed presence of R. swinhoei, based on shells, skulls or photographic evidence; score = 2: high probability of presence, where > 3 independent interviewees accurately described the species' morphological characteristics, there had been multiple sightings within the previous 10 years, the precise location of these sightings was known and the sighting was confirmed by multiple household members; score = 3: medium probability of presence, where only one interviewee described morphological characteristics accurately, and any sightings had occurred > 10 years ago; score = 4: low or no probability of presence, where interviewees did not know the species, or their comments were contradictory and/or unreliable.
We categorized the sites where R. swinhoei may be present according to the level of habitat modification that had occurred during the past 30 years: score = 1: no or little habitat alteration, where habitat has been modified by human activities, but still retains large patches of natural vegetation cover in the waterbody and on banks, and the size and shape of the site have not changed significantly; score = 2: moderate habitat alteration, where habitat has been modified by dams and/or land conversion, water level, area and/or shape have been changed, vegetation cover has declined, but patches of natural vegetation remain; score = 3: severe habitat alteration, where habitat has been completely modified by dams and/or land conversion, water level, shape, and/or size of the site have changed significantly, and natural vegetation has largely disappeared.
We assigned a score to estimate fishing intensity at each site: score = 1: no fishing or low fishing intensity, where < 20 fishers were active at the site; score = 2: medium fishing intensity, where 21–40 fishers were active; score = 3: high fishing intensity, with > 40 fishers. The scores for habitat alteration and fishing intensity were assigned by three authors (OLD, TVP, LL), who together had > 45 years of experience in field surveys.
Our analyses included four sites newly surveyed in this study, in addition to the sites previously described by Le Duc et al. (Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a) and Pham et al. (Reference Pham, Le Duc, Leprince, Bordes, Luu and Luiselli2020). For each site, we calculated an overall conservation priority score that combined the presence, habitat alteration and fishing intensity scores, whereby a low overall score indicates a site of high conservation priority. The conservation priority score was calculated with the formula: presence score + habitat modification score + fishing intensity score. The highest possible priority would thus be a score of 3 (1 + 1 + 1), and the lowest possible priority would be a score of 10 (4 + 3 + 3). We analysed correlations between presence, habitat alteration, fishing intensity and overall priority scores using a Spearman's rank correlation coefficient (r s). We used PAST 3.0 (Hammer et al., Reference Hammer, Harper and Ryan2001) for all analyses, with alpha set at 5%.
Results
Interviewees in Son La and Lai Chau provinces reported 13 reliable sightings from 2010–2019, seven of which were in 2018–2019 (Table 1; see Supplementary Material 1 for a full summary of interview responses). Overall, we identified 13 sites where R. swinhoei may be present in the Da, Ma, Chu, and Red River systems, at a mean altitude of 98.6 ± SD 80.4 m (Table 2). One site had a presence score of 1, with one individual confirmed and possibly two others present, three sites a score of 2 (high probability of turtle presence), and nine sites a score of 3 (Fig. 1). If we combine the findings from our surveys with those from areas along the Da River and in Thanh Hoa province (Le Duc et al., Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a), there is a total of 13 sites where R. swinhoei could potentially be present (Fig. 1), with sites 1 and 2 having the highest probability of presence based on the degree of habitat alteration and fishing intensity (Table 2), and sites 10, 11, 12 and 13 with multiple sightings reported over the last decade. Several of these reported sightings were in 2018 and 2019, predominantly at sites with more severe habitat alteration and higher fishing intensity (Table 1). In addition, these surveys combined provide evidence of 25 sites where R. swinhoei probably occurred until a few decades ago (Fig. 1). The assigned conservation priority scores for the different localities, with all variables, are shown in Table 2.
Fig. 1 Study area in northern Viet Nam, showing (a) the 13 sites where Rafetus swinhoei may be present, and (b) the sites where it was present before the 2000s (based on interviews with reliable descriptions of the species). For numbers of the various sites, see Table 2. To protect the sites, the precise locations are not shown; the circles indicate only the approximate locations. Scores for probability of the species' presence: 1: confirmed presence of R. swinhoei; 2: high probability of presence; 3: medium probability of presence (see Methods).
Table 1 Recent sightings (2010–2019) of Rafetus swinhoei that were reported in November 2019 by fishers from Son La and Lai Chau provinces, Viet Nam, along the Da River (only reliable reports were included; see Methods for details).
1 For site numbers, see Table 2.
2 A dash between two months (e.g. July–Aug.) indicates that the interviewee was unable to remember exactly the month of their observation, whereas an ampersand between two months (e.g. Apr. & Nov.) indicates that the interviewee reported distinct observations in different months.
Table 2 Details of the 13 sites with potential presence of Rafetus swinhoei, including the attributed scores for species presence, habitat alteration and fishing intensity (see Methods).
The scores for habitat alteration and fishing intensity were not significantly correlated (r s = 0.472, n = 13, P = 0.103), and neither were those for fishing intensity and species presence (r s = 0.068, n = 13, P = 0.827), nor for habitat alteration and species presence (r s = −0.058, n = 13, P = 0.851). To protect the sites, their geographical coordinates are not reported here, but can be obtained from the authors on request. The general landscape of these sites is shown in Plate 1.
Plate 1 Landscape of some of the sites with potential presence of R. swinhoei in northern Viet Nam. (a) De lake, Thanh Hoa province; (b) Dap Nai dam, Buoi River, Hoa Binh province; (c) Dat hydro dam, Chu River, Thanh Hoa province; (d) Suoi Bung stream, Da River, Son La province; (e) Suoi Mu stream, near Hoa Binh dam, Da River, Hoa Binh province; (f) Dong Mo lake, Hanoi; (g) Suoi Hai lake, Hanoi; (h) Mu stream, Da River, Son La province; (i) Muong Sai section, Da River, Son La province.
The overall conservation priority score was statistically more affected by fishing intensity (r s = 0.682, n = 13, P < 0.01) than by habitat alteration (r s = 0.600, n = 13, P < 0.05) and species presence (r s = 0.608, n = 13, P < 0.05), but overall the three variables had a similar weight.
Discussion
Our surveys provide new information about the potential presence of R. swinhoei in addition to the single location where the species is currently known, supporting the evidence previously reported by Le Duc et al. (Reference Le Duc, Pham, Leprince, Bordes, Tuan and Benansio2020a) and Pham et al. (Reference Pham, Le Duc, Leprince, Bordes, Luu and Luiselli2020). At least two sites (sites 1 and 3; Table 2) appear to be potentially suitable, and there have been reliable reports of occasional, recent sightings of the species at several other sites. We therefore consider it likely that R. swinhoei persists in additional sites in northern Viet Nam.
Several recent sightings were at sites with a relatively high density of fishers, perhaps because the probability of sightings increases when there are more observers. Sightings reported from sites with highly altered habitat cannot, however, be explained by collinearity effects between habitat alteration and density of fishers, as these two variables were not significantly correlated. The most likely interpretation is that R. swinhoei is an elusive, shy species that is rarely detected in the wild. It is therefore possible that the population of R. swinhoei is larger than believed and that the species can persist in remote areas where habitat alteration is low and fishing intensity moderate.
All sites with multiple recent sightings (10, 11, 12 and 13) were clustered along the northern course of the Da River system, far from other sites where the species could occur (Fig. 1). This suggests that R. swinhoei may persist throughout the area, but has gone undetected because of the remoteness of the areas and limited survey efforts. Further research needs to focus on the northern course of the Da River system.
Our evaluation of the conservation priority of sites where this rare and elusive freshwater turtle may occur was based on an assessment of the likelihood that the species was present, the degree of habitat alteration and fishing intensity. This method could also be applied elsewhere. It requires the acquisition of interview-based data and a general inspection of habitat features and densities of fishers, variables that are relatively easy to obtain. Our approach could be applied in other areas where field surveys of chelonians are logistically difficult or time-consuming. For example, our method could be used to study species such as Erymnochelys madagascariensis in Madagascar, Cyclanorbis elegans in continental Africa (Demaya et al., Reference Demaya, Benansio, Lado, Diagne, Dendi and Luiselli2019a,Reference Demaya, Benansio, Lado, Jubarah, Ladu and Luisellib), and Chitra chitra, Chitra vandjiki, Chitra indica, Nilssonia formosa, Nilssonia nigricans, Pelochelys cantorii, Batagur affinis, Batagur kachuga and Batagur borneoensis in Asia. The method could also be used to study other threatened freshwater animals for which habitat loss and overfishing/overhunting are drivers of population declines (e.g. freshwater dolphins and large freshwater fishes).
However, there are also potential limitations of our method. Firstly, it is based on a combination of local ecological knowledge (interviews) and expert-based assessment of scores for habitat alteration and fishing intensity. As expert-based scores are influenced by researchers’ field experience, we recommend that only experts with > 5 years’ experience conduct field surveys and that the cumulative number of years of field experience of the surveyors is specified when publishing results. Averaging the scores independently assigned by multiple experts (three in our case) could also increase reliability. Seasonality can have a marked effect on perceived habitat characteristics and the density of fishers, and thus for highly seasonal environments we recommend that surveys are made in both dry and wet seasons or, if it is logistically difficult to survey in both seasons, to survey only in the wet season.
Based on the current knowledge of the species' distribution and population size, R. swinhoei is on the brink of extinction, with just two known living individuals. Further research should focus on locating additional live individuals in the wild and developing a robust conservation strategy. With respect to the sites we surveyed, site 1 has dense aquatic vegetation, hampering observational surveys. The remaining 12 sites have clear water surfaces, but sites 6, 7, 8 and 9 present logistical difficulties for observational surveys because of their complex spatial structure, with a number of islands present within the lakes. We recommend using trapping surveys at sites 1, 6, 7, 8 and 9, and observational surveys at other sites. Drones could facilitate sightings (Oliveira-da-Costa et al., Reference Oliveira-da-Costa, Marmontel, da-Rosa, Coelho, Wich, Mosquera-Guerra and Trujillo2020) and complement riverbank transects and visual searches by boat. In addition, we recommend monitoring fishers' catches, as newborn and juvenile R. swinhoei that can be mistaken for Pelodiscus sinensis because of the morphological similarities between the two species. Environmental DNA could also be used to survey for R. swinhoei (Rees et al., Reference Rees, Maddison, Middleditch, Patmore and Gough2014; Davy et al., Reference Davy, Kidd and Wilson2015; Kundu et al., Reference Kundu, Kumar, Tyagi and Chandra2018), but the efficiency of this method to detect freshwater turtles has been questioned (Raemy & Ursenbacher, Reference Raemy and Ursenbacher2018), and its applicability requires further investigation.
Acknowledgements
This study was funded entirely by Turtle Sanctuary and Conservation Centre, Paris, France. We thank the Vietnam National University of Forestry for granting the necessary permit to conduct the interview survey, and two anonymous referees for their helpful comments.
Author contributions
Study design: LL, OLD, TVP; fieldwork: OLD, TVP; data analysis: CD, OLD, LL, TVP; writing: LL, OLD, TVP, TZ; administrative and research support: BL, CB, DHD, TZ, VLQ.
Conflicts of interest
None.
Ethical standards
This research was authorized by the Government of the Socialist Republic of Viet Nam (research permit code: 613/DHLN-HTQT and 2562/UBND-LC) and abided by the guidelines on ethical standards of Oryx and the British Sociological Association.
Open access
