Towards the conservation of Borneo’s freshwater mussels: rediscovery of the endemic Ctenodesma borneensis and first record of the non-native Sinanodonta lauta

The study area covered about 13,000 km², spanning the Kingdom of Brunei and Limbang Division, Sarawak (Fig. 1). The area is characterised by a tropical rainforest climate and is dominated by shale, siltstone and sandstone deposits, with some limestone in the upper reaches of the Limbang River (Mathew et al. 2016). It is located within the freshwater ecoregion of Northwestern Borneo, which features a diverse and highly endemic fish fauna (Abell et al. 2008; Rahim 2012; Sulaiman et al. 2018). The area includes several protected areas, most notably the National Parks of Tasek Merimbun and Ulu Temburong in Brunei, and Gunung Mulu and Gunung Buda in Limbang Division (Fig. 1).

Field surveys were conducted in the dry season, when water levels are lowest and access to mussels is therefore optimal, in March, April and July 2018. A total of 81 sites were surveyed (inset in Fig. 1), situated in the basins of the Belait, Tutong, Limbang, Temburong, Trusan, Mulam and Merapok rivers (from west to east) (Fig. 1, Online Resource 1). Sites included a range of freshwater habitats, from small, shallow streams (< 1 m width), to large lowland rivers (> 30 m width), as well as standing water bodies such as man-made ponds. Surveys were completed in two stages, i.e. (1) interviews and (2) physical surveys. Access to the site was not deemed safe due to the presence of Estuarine Crocodiles (Crocodylus porosus) at 38 sites, which are therefore limited to data from interviews (empty circles in inset of Fig. 1); both interviews and physical surveys were conducted at 43 sites (full circles in inset of Fig. 1).

At each site, covering about 100 m river length, we initially approached local people living or operating in the vicinity of the river (e.g. fishermen) to ask about safe access to the river and the presence of freshwater mussels (see Online Resource 2 for a list of questions put to interviewees). Whenever possible, at least three individuals from diverse demographic groups (age, sex and ethnicity) were interviewed informally at a locality of their choice (usually outside their house or on the river bank) to minimise disruption to them and thus maximise willingness to share their knowledge. Interviewees were thereby shown pictures of the freshwater mussel species previously recorded from northern Borneo. Where access was deemed safe (i.e. absence of crocodiles and river accessible with minimal risk of snake bites or falls), we surveyed for mussels by hand, net and rake following the same protocol as described in Zieritz et al. (2018c). Where mussels were found, voucher specimens and tissue snips were collected and preserved in absolute ethanol and deposited at the Research Collection of the Universiti Malaysia Sarawak (UNIMAS). Except for vouchers, all other specimens were returned to their habitat.

Several water quality parameters known to affect mussel distribution (see Zieritz et al. 2016; Zieritz et al. 2018c and references therein) were collected at each site: (1) pH, (2) dissolved oxygen (DO), (3) temperature and (4) conductivity were measured in situ using EUTECH pH 5 + (Eutech Instruments Europe BV, Nijkerk, Netherlands), and EXTECH instruments (Extech Instruments Inc., Nahua, NH, USA) SDL150, SDL100 and EC210, respectively. In addition, 300 ml-water samples were collected, of which at least 200 ml were filtered through a Whatman GF/C filter on the evening of the day of collection and 25 ml were immediately processed for analysis of concentrations of (5) total phosphorus (TP) by adding sulfuric acid. GF/C filters were folded, individually wrapped in labelled tin foil and stored together with water samples in the dark on ice whenever possible. Concentrations of (6) soluble reactive phosphorus (SRP) and (7) total ammoniacal nitrogen (TAN) were determined in the laboratory using standard spectrophotometric/colorimetric methods (Lorenzen 1967; Mackereth et al. 1989). (8) Organic matter (OM) concentrations of suspended solids were determined as (W_bi − W_ai)/V, where W_bi is weight [mg] of GF/C filters and residue after drying at 105 °C for 24 h, W_ai is weight [mg] of GF/C filters and residue after 4 h at 550 °C, and V is the volume of water filtered [l]. Concentrations of (9) calcium (Ca⁺) and (10) potassium (K⁺) ions were determined through ion chromatography (IC) analysis using a Metrohm Basic 792 ion chromatography system (Metrosep A Supp 4–250 column, with 1 mmol sodium bicarbonate and 3.2 mmol of sodium carbonate eluent at 1.0 ml min⁻¹).

Collected specimens were identified to species level through an integrative morphological-molecular approach. DNA was extracted from three specimens of each population using Macherey–Nagel Nucleo-Spin Tissue Kit following the manufacturer’s instructions. Cytochrome c oxidase subunit I (COI) mitochondrial DNA was amplified and sequenced in both directions from all extracts using primers and protocol detailed in Zieritz et al. (2016). For native species, we additionally sequenced 28S ribosomal DNA using primers and protocol detailed in Lopes-Lima et al. (2017). Sequences were cleaned up in program MEGA X (Kumar et al. 2018) and deposited on Genbank under Accession Numbers MN900788-900794 and MN902292-902301.

Initial BLAST results of the novel Ctenodesma sequences suggested phylogenetic affinities to representatives of the subfamily Gonideinae (sensu Pfeiffer et al. (2019)), especially members of the tribes Rectidentini and Contradentini. We designed our taxon sampling to recover the phylogenetic position of Ctenodesma by sampling all representative species of the tribes Rectidentini and Contradentini available on Genbank, as well as representatives of the other tribes of the Gonideinae (Table 2). Molecular matrices were aligned using ClustalW v. 2.1 (Larkin et al. 2007). PartitionFinder2 v 2.1.1 (Lanfear et al. 2017) was used to find the best partition scheme and Generalised time-reversible (GTR) model of nucleotide evolution using the AICc and greedy search algorithm (Lanfear et al. 2012). Maximum-Likelihood (ML) searches were then conducted in IQ-TREE v 1.6.10 (Nguyen et al. 2015) with an initial tree search followed by 10 independent runs and 10,000 ultrafast bootstrap replicates using the best codon partitioning scheme and model of nucleotide evolution. Bayesian inference (BI) analysis was performed using MrBayes v 3.2 (Huelsenbeck and Ronquist 2001; Ronquist et al. 2012) using the best codon partitioning scheme and model of nucleotide evolution with 20 × 10⁶ generations sampling every 1000 generations for a total of 20,000 trees with a burnin of 5000. Convergence of the two runs was monitored by the average standard deviation of split frequencies, the potential scale reduction factor (PSRF) and the effective sample size (ESS) of the estimated parameters.

Out of the 43 sites where both interviews and physical surveys were conducted, only one population of native mussels was found in a tributary of the Limbang River near the village of Kuala Medalam (alternative spelling Mendalam; Fig. 2). The species was identified as Ctenodesma borneensis by morphological characters. The species is small and elongated elliptical (Haas 1969), with dimensions of collected specimens being 4.3, 4.8 and 5.0 cm in length, and 2.0, 2.4 and 2.5 cm in height (Fig. 2a). Shells are coloured yellow-greenish to brown with rays and exhibit fine, irregular zig-zag sculpture. Umbos were heavily eroded in all specimens. Inhalant siphons exhibit one row of papillae (Fig. 2b). The one gravid specimen we collected exhibited glochidia in all four demibranchs (tetragenous).

COI and 28S sequences generated for the three specimens were identical (Genbank Accession numbers in Table 2). They were highly divergent from available sequences on Genbank, with the closest match being Sinoyriopsis cumingii with 86.3% similarity for COI, and Hyriopsis sp., Ensidens sp. and R. sumatrensis with 97.9–98.1% for 28S.

The population was found based on the knowledge and guidance of a local Lun Bawang (an indigenous tribe of this region) and is located in a relatively pristine patch of secondary rainforest only accessible by hiking. The stream was small (< 2 m width) and shallow (< 1 m depth), with a muddy substrate and abundant dead wood and leaf litter material (Fig. 2c), which was also recorded as high TP and OM concentrations in the water (Fig. 3). Other water chemistry parameters indicated good water quality with almost neutral pH, and low conductivity, nutrients and K⁺ concentrations (Fig. 3). Nevertheless, C. borneensis population density was low, as only three specimens were found over a 4-person hour survey time.

The only other population of freshwater mussels found in our surveys was that of Sinanodonta lauta in a pond in Lawas District, Limbang Division, about 2.5 km east of the border to Brunei (Fig. 1). Being morphologically very similar to the widespread S. cf. woodiana, this population was identified through DNA barcoding. COI sequences were generated for three specimens (Genbank Accession Numbers MN900791-93), all of which were identical and matched with Haplotype 54 of S. lauta (Lopes-Lima et al. 2020). The specimens had been intentionally introduced to the pond by the owner a few years before, who had bought them from a market in Lawas town. Density was very high and several dozen specimens were sampled within 15 min. Water chemical parameters indicated slightly elevated TAN and K⁺ concentrations, and a water temperature of > 30 °C (Fig. 3).

Interviews at sites that could not be surveyed physically did not reveal any additional potential occurrences of mussels except for one anecdotal record of dense mussel populations in the upper Baram River basin in Gunung Mulu National Park (Fig. 1). Despite interviewing > 100 people at 20 sites across Brunei, there was no indication of the presence of freshwater mussels in Brunei in the past or present.

The most likely partitioning and modeling scheme was to treat all subsets (28S, COI_pos1, COI_pos2, COI_pos3) as separate partitions using the GTR + G model of nucleotide evolution. Convergence of the BI analysis was supported by the average standard deviation of split frequencies (0.001876), average PRSF values (1.000), and high ESS values (> 3000). The BI and ML reconstructions recovered identical supraspecific relationships except for a minor difference in the sister relationships with Hyriopsis (Fig. 4). Ctenodesma was recovered as sister to sampled representatives of the tribe Rectidentini (Hyriopsis, Rectidens and Ensidens), however, this clade had limited support (82 PP/72 UFBS).

Our work has resulted in the first confirmation of a population of C. borneensis for 57 years. The species was first described in 1874 from Lake (Danau) Seriang (alternative spelling Siriang), Kapuas basin, West Kalimantan, Indonesia, and has a presumed native distribution covering West Kalimantan, Sarawak, and Sabah (Fig. 1; Table 1). To our knowledge, two museum lots exist from “Sarawak”, dated from 1937 and 1945, respectively (Graf and Cummings 2019). The latest record is that by N.S. Haile from 1962 from Gomantong, Sandakan Residency, Sabah, Malaysia.

Although our find is certainly encouraging for freshwater mussel conservation, C. borneensis has likely declined in its area of occupancy and is now extremely rare, consisting of only isolated populations in the north and potentially other parts of Borneo. The population presented in this paper remains the only one we found across an area of 150,000 km² and 115 sites that have been surveyed for freshwater mussels from 2016 to 2018 (Zieritz et al. 2018c) (Fig. 1). Furthermore, although several other sites in the immediate proximity of the site where C. borneensis was found were surveyed, no further populations could be found (inset of Fig. 1). Finally, the population density of the identified population was extremely low.

The reasons for the apparent decline of this species may be linked to the large-scale deforestation across Borneo. The fact that the only two populations of C. borneensis recorded by scientists in the past 60 years inhabit streams within dense rainforests (the second site being Gomantong, which has been a designated Protection Forest Reserve since 1984; UNEP-WCMC and IUCN 2014–2020) suggests that the species may be particularly sensitive to human pressures. The Limbang River is known for its geological complexity, and susceptibility to erosion and flash floods (Krishnan et al. 2017). Ongoing and expanding deforestation activities in this area (World Resources Institute 2014) may therefore pose a particular threat to the integrity of the potential habitats for C. borneensis and other freshwater taxa.

Our phylogenetic analyses for the first time include an endemic Bornean freshwater mussel genus, i.e. Ctenodesma, which revealed that this genus represents (part of) a highly distinct and previously unsampled molecular phylogenetic lineage (Fig. 4), thereby identifying it as a particularly valuable conservation target (Faith 1996). Specifically, Ctenodesma was recovered as the sister-group to the Rectidentini and exhibited morphological characteristics typical of this tribe, including tetragenous marsupia (Lopes-Lima et al. 2017). Of the three Rectidentini genera sequenced to date, Ensidens and Hyriopsis are distributed across SE-Asia, whilst Rectidens is restricted to Sundaland (i.e. Java, Sumatra, Peninsular Malaysia and Borneo) (Graf and Cummings 2019). The non-monophyly of the Borneo samples included in our analysis (Rectidens and Ctenodesma) suggest multiple lineages have colonised Borneo, and appear to have done so by different mode and tempo. Rectidens appears to have more recently and more completely dispersed across Sundaland, whereas Ctenodesma appears to represent a relatively older colonisation event. Several other Sundaland endemics have been hypothesised to belong to various clades of the subfamily Gonideinae, including Caudiculatus (Chamberlainini), Elongaria and Prohyriopsis (Rectidentini), and Pressidens (Contradentini) (Graf and Cummings 2019; Pfeiffer et al. 2019); however, these taxa have yet to be included in a molecular phylogenetic reconstruction. The inclusion of these phylogenetically unsampled lineages will be useful in more completely estimating the evolutionary and biogeographic history of the freshwater mussels of Sundaland. Improved sampling of the Bornean and greater Sundaland fauna will also elucidate whether the recovered long branch of Ctenodesma represents a relict and depauperate lineage or a more diverse, but currently undersampled, radiation.

Whilst native mussels are likely declining in northern Borneo, our study revealed the introduction of a second non-native freshwater mussel species (in addition to S. cf. woodiana) to this island, i.e. S. lauta. The population from a pond in Limbang Division represents the first record of this species outside its native distribution in Japan, Korea and Southeast Russia (Lopes-Lima et al. 2020). This species is distinct but closely related to S. cf. woodiana, native to the Yangtze basin, which is actively spread, repeatedly introduced and very common in Sabah, and is also present in the Suai basin, Sarawak (Zieritz et al. 2018c; Lopes-Lima et al. 2020). The main vectors for ongoing introductions and spread of Sinanodonta in Borneo is indirect through fish that are infected with mussel larvae, which live as ectoparasites on fish gills and fins (Wächtler et al. 2001), as well as direct and intentional spread by humans for ornamental purposes and as a food source (Zieritz et al. 2018c). Introductions appear to be usually confined to ponds or other man-made water bodies, where native species are unable to survive and reproduce. As such, the current threat posed by non-native freshwater mussels to the native mussel fauna of Borneo appears to be small when compared to the scale of destruction posed by ongoing logging activities across the island (Miettinen et al. 2011).

Based on the combined dataset from our previous (Zieritz et al. 2018c) and present work, the current situation of northern Borneo’s freshwater mussel fauna is very concerning. As discussed in detail by Zieritz et al. (2018c), historical records indicate the native distribution of five species within the boundaries of the area spanning from the Rajang basin in the west to the upper Sembakung, Kinabatangan and Bengkoka basins to the east, i.e. C. borneensis, P. exanthematicus, Pseudodon walpolei, R. sumatrensis and Schepmania niewenhuisi. Of these, only R. sumatrensis appears to occur in healthy population sizes across several river basins of central Sarawak. Whilst C. borneensis has now been confirmed from a single site in the Limbang basin, and S. niewenhuisi was relatively recently (i.e. in 2000) collected from Sabah (albeit as dead shells only), P. exanthematicus and P. walpolei have not been found for at least 50 years. The fact that historical records of these two species are largely confined to the area already surveyed by us indicates that they are very rare or potentially already extinct. Future surveys across Borneo will be needed to revalidate this statement and identify any remaining populations of these endemic Bornean mussel species. Currently available data suggest that endemic Bornean mussels can only sustain populations in relatively pristine habitats. Future surveys targeted towards these species should thus focus on protected areas of northern Borneo, including Labi Hills (Sungai Ingei Conservation) in Brunei Darussalam, Gunung Mulu, Gunung Buda, Pulong Tau, Usun Apau and Lanjak Entimau in Sarawak. Remaining populations of Bornean endemic mussel species in unprotected areas would benefit from retaining or re-establishing riparian buffer zones, which have been shown to be effective in protecting freshwater ecosystems and biodiversity (Luke et al. 2017a, b). Although legislation guidelines in Sarawak state a buffer zone of 5–50 m along all rivers depending on river width, current implementation rate is poor, as non-compliance with these guidelines has limited to no legal consequences (Ligtermoet et al. 2009; Luke et al. 2019).