Low genetic diversity and predation threaten a rediscovered marine sponge

Discovered in 1819 in the tropical waters off Singapore, the magnificent Neptune’s cup sponge Cliona patera (Hardwicke, 1820) was harvested for museums and collectors until it was presumed extinct worldwide for over a century since 1907. Recently in 2011, seven living individuals were rediscovered in Singapore with six relocated to a marine protected area in an effort to better monitor and protect the population, as well as to enhance external fertilisation success. To determine genetic diversity within the population, we sequenced the complete mitochondrial genomes and nuclear ribosomal DNA of these six individuals and found extremely limited variability in their genes. The low genetic diversity of this rediscovered population is confirmed by comparisons with close relatives of C. patera and could compromise the population’s ability to recover from environmental and anthropogenic pressures associated with the highly urbanised coastlines of Singapore. This lack of resilience is compounded by severe predation which has been shrinking sponge sizes by up to 5.6% every month. Recovery of this highly endangered population may require ex situ approaches and crossbreeding with other populations, which are also rare.

Since the 1990s, specimens of C. patera from Australia and Southeast Asia have emerged from trawling and biodiversity surveys, suggesting that at least some remnant populations persisted 26,[34][35][36][37] . Indeed, C. patera was rediscovered in 2011 from its type locality, Singapore, with just seven living individuals that are currently known to exist. Coastal populations of marine fauna are more adversely impacted by human activities 31,32 , which could reduce the rate of population recovery of C. patera in Singapore. In a bid to conserve this species, six of the seven sponges have been relocated to the Sisters' Islands Marine Park (SIMP), a marine protected area managed by Singapore's National Parks Board. Within that area, strict protection laws are enforced, and the site facilitates regular monitoring efforts. There is an added benefit of the aggregation which enhances the probability of external fertilisation in benthic organisms [38][39][40][41] . However, aggregating the sponges in a protected area might result in increased predation, particularly from large predators such as turtles, thus adversely hampering sponge recovery. Furthermore, small populations generally suffer from low genetic diversity 42 , limiting the long-term viability of this sponge population. The efficacy of this measure in recovering the local population of this critically endangered species has hitherto not been assessed 43 .
This study aims to determine the genetic diversity of the local C. patera population across multiple loci, thereby estimating their resilience. We further characterised its ecology at the relocated site to assess risks and threats to the threatened population. With some individuals facing severe predatory stress on relocation, exacerbated by negligible population genetic variability, we propose urgent recovery and conservation measures for this enigmatic species to prevent it from becoming extinct, again.

Methods
Sampling and DNA sequencing. Tissue samples of 2 cm 3 were taken from six relocated C. patera individuals and fixed in 100% molecular-grade ethanol. Genomic DNA and library preparation were conducted following Quek et al. 44 . Briefly, genomic DNA (gDNA) was extracted using EZNA Mollusc DNA Kit with a modified elution protocol to ensure high quality gDNA was obtained. Following which, purification of gDNA was conducted using Zymo DNA Clean-up and Concentrator Kit. Bioruptor Pico (Diagenode) and KAPA dualindexed adapters were ligated to sheared fragments using KAPA HyperPrep kit. Final libraries were size-selected following manufacturers' protocol using Agencourt AMPure XP beads (Beckman Coulter). The samples were then pooled in equimolar concentrations with other libraries not associated with this study and sequenced on a single Illumina HiSeq 4000 (150 × 150 bp) lane to recover the complete mitochondrial genome and nuclear rRNA sequences.
Separately, amplification of the internal transcribed spacer (ITS) for sponges was conducted using ITSRA2 (5′-GTC CCT GCC CTT TGT ACA CA-3′) and ITS2.2 (5′-CCT GGT TAG TTT CTT TTC CTC CGC-3′) 45 . Amplified products were purified using AMPure XP (Beckman Coulter) and cycle sequenced in both directions separately using the BigDye™ terminator method and sequenced using the ABI 3130 XL Genetic Analyzer (Applied Biosystems). Chromatograms produced were then edited on Geneious Prime (Biomatters) and assembled de novo.
Genome skimming and loci identification. Raw reads were demultiplexed and both low quality bases and adapters were trimmed and assembled using fastp v0.20.1 46 and SPAdes v3.12.0 47 respectively, under default settings. Mitochondrial genomes were identified by BLASTn (e-value 10 −6 ), searching against other clionaid cytochrome c oxidase subunit I (COI) sequences downloaded from GenBank (Fig. 2), with the longest contig selected as the mitochondrial genome, and circularisation of the genome was executed by circules.py 48 . Annotation of mitochondrial genomes was conducted using MITOS2 49 .
Sequences of nuclear ribosomal gene 28S were identified in a similar fashion to that of COI, by searching against poriferan 28S sequences downloaded from GenBank using BLASTn (e-value 10 −6 ) (Fig. 2). Only contigs with ≥ 200 bp overlap of ≥ 90% similarity to the references and minimum kmer coverage of 100 (determined by SPAdes v3.12.0) were extracted and separately assembled in Geneious Prime (Biomatters). Finally, to ensure    50 . Alignments were manually inspected to check for overlap between individuals of each nominal taxon. The best model of DNA evolution was identified using ModelTest-NG 51 and specified in ML phylogeny reconstruction using RAxML-NG v0.8.1 with 10 random and 10 parsimonious trees, and 200 bootstrap pseudoreplicates 52 . Each gene tree was inspected, and we only kept sequences from named taxa that were recovered as a monophyly with a minimum bootstrap support value of 50 for downstream pairwise comparisons. Retained sequences were realigned by MAFFT as above and the alignments were imported into MEGA X 53 to compute intraspecific pairwise distances.
Ecological data. Growth. Monthly photographs of relocated sponges were tracked each month from June 2020 to December 2020 to estimate growth rates via SCUBA dive surveys. Photos were taken with a scale from a fixed distance and angle using an Olympus TG-5. The total height of each sponge was measured using ImageJ 54 . For months where visibility in the photo was low, the data were not used. Growth rate (cm/month) was computed based on the height data obtained monthly.
Regeneration capability. To determine the regeneration ability of C. patera, we cored the cup section of all six relocated individuals using a 2 cm diameter stainless steel apple corer and estimated the rate of tissue recovery in relocated C. patera individuals. Weekly surveys were performed until the hole had fully recovered. During each survey, the diameter of the cored hole was measured, and any other visual observations noted. The surface area of the remaining hole was calculated based on the measured diameter. Recovery rate was measured by taking the difference in surface area every week, normalised by the number of days between two surveys. Average recovery rate of each sponge was calculated until the holes had completely sealed during the survey.

Results
Mitochondrial genome assembly and genetic diversity. The mitochondrial genomes of all six samples sequenced were identical, with a length of 19,133 bp. Comparison of gene order of ribosomal and proteincoding genes (PCGs) between C. patera and congener C. varians found slight differences, with the block of genes comprising nad6 and nad3 adjacent to the 12S rRNA gene in C. varians, whereas it is between nad4 and nad4L in C. patera (Fig. 3).
The phylogeny reconstructed showed a number of inconsistencies in taxon identification based on GenBank data (Fig. 3), corroborating studies such as those identifying cryptic species complexes in C. celata 55,56 . After extracting and analysing sequences only from taxa that form monophyletic groups, we found higher intraspecific pairwise differences in other clionaids, compared to genetic distances from nuclear and mitochondrial loci of C. patera that were nearly zero (Fig. 4, Tables S1-S3). Specifically, mean intraspecific pairwise distances for 28S rRNA locus in C. patera stood at 0.0423% (± SD 0.0324%), and no variability was detected for ITS and COI sequences.   www.nature.com/scientificreports/ the remaining two individuals (E and F), we recorded a loss of between 12 and 14 cm (24-33%), likely due to severe predation. Their cups were almost entirely consumed and visible bite marks were recorded on their stems (Fig. 5A,B). Nevertheless, C. patera generally exhibited remarkable rates of recovery after boring, sealing the 20 mm diameter core hole within three weeks (Fig. 5C,D). However, one of the individuals (F) did not show any sign of recovery after two weeks, and by the third week of observation, the tissue at its core had been consumed by predators.

Discussion
This study has provided clear evidence for two main threats against the long-term viability of the C. patera population in Singapore: low genetic variability and heavy predation pressure. Predation was so severe in some individuals that the entire cup was consumed, and would likely take a long time to recover, if at all possible. Indeed, sponge E is no longer alive (died sometime in February 2022) and only a small area of living tissue remains for sponge F. The known population in Singapore now stands at just six individuals. The mitochondrial genome of C. patera presented a novel gene rearrangement when compared to confamilials C. varians and Spheciospongia vesparium, with the block nad6-nad3 being translocated to between nad4 and nad4l in the former, and being between nad5 and 12S rRNA in C. varians. Examination of mitochondrial gene arrangements in Demospongiae has revealed multiple rearrangements, and is particularly rife in Heteroscleromorpha that contains Clionaida 57 . Rapid and unique mitochondrial evolution has also been observed in other sponges, such as in Calcaronea with linear chromosomes 58,59 , and in Hexactinellida with a frameshifting translation strategy to cope with their deep-water environments 60 . Furthermore, phylogeny reconstructions of Clionaidae have found C. varians to be more closely related to Spheciospongia than C. patera (Fig. 2), though all are nested within Clionaidae. These patterns support the identical gene arrangement between C. varians and S. vesparium, but also suggest that the rearrangement occurred recently. Nevertheless, this is only the second complete mitochondrial genome sequenced for Cliona and third for Clionaidae to date. More representative data for Clionaida are needed to determine if there is indeed rapid mitochondrial gene evolution within the order (Fig. 3).
The lack of variability for the COI gene among sponge species concurs with numerous past studies, as sponges have relatively slow rates of mitochondrial evolution (Fig. 4)   www.nature.com/scientificreports/ long generation times and low metabolic rates 64 . Nevertheless, in contrast to other members of Clionaidae, the C. patera population in Singapore had virtually no variation for the three gene loci analysed (Fig. 4, Tables S1-S3). For example, León-Pech et al. 65 found a minimum of 0.2% divergence within C. vermifera, and we detected at least some sequence variability in all species compared except C. yorkin. Alternative markers commonly used in phylogenetic and taxonomic studies of sponges include nuclear markers such as ITS and 28S loci of the ribosomal RNA (rRNA) region 66 , with the former being common in phylogeographic studies due to their higher degree of intraspecific variability 65,67 (but see Wörheide et al. 45 ). The lack of variability across the whole mitochondrial genome and ribosomal genes examined here could be caused by common maternal inheritance within C. patera. Considering that this species was once thought to be extinct, the population bottleneck in the early twentieth century and the resultant small remnant population that arose from a few ancestral individuals would have diminished the gene pool markedly. Low genetic diversity increases the risk of extinction by reducing the capacity of populations to adapt to environmental changes due to the lack of variation. Furthermore, high levels of urbanisation (e.g., land reclamation) and associated anthropogenic pressures (e.g., pollution, high sedimentation rates, ship groundings) along the coastlines of the city state exert further duress on the C. patera population in Singapore. These threats are known to be detrimental for marine invertebrates 68,69 , even leading to the local extirpation of several species 8,70 . The relocation of the sponges in Singapore to a marine park would partially ameliorate some of the threats faced by the C. patera population, but this measure (i.e., the no-fishing zone) may introduce other stressors such as increased abundance of large spongivores.
With advanced sequencing technologies, it is now possible to trace population genetic patterns from museum samples collected centuries ago, based on remnant ancient DNA (aDNA) using highly rigorous techniques 71 . For sponges, mini-barcodes have been developed for sponge identification of museum specimens 72,73 that can be extended in future studies to multi-marker assays for reconstructing population histories. Considering the large collections of C. patera available in museums around the world due to the wanton harvesting in the nineteenth century, population genetic studies of C. patera specimens around the world can be performed to detect bottlenecks in the past 74 . This would shed more light on the detrimental effects that anthropogenic activities have on sponge populations 75,76 and potentially drive conservation efforts to limit overharvesting globally.
Predation by spongivores, especially larger taxa, are able to shape sponge communities [77][78][79] . A recent review showed that fishes are known to influence sponge distributions in the Atlantic, with some 50 sponge genera serving as a prey 80 . Furthermore, large predators such as the hawksbill turtle appear to have a preference for certain sponge prey, such as those with a lower spicule content 81 . More specifically, predators of clionaids include both vertebrates (e.g., parrotfish, pufferfish, damselfish and turtles) and invertebrates (e.g., gastropods, decapods, isopods and echinoids) 82 www.nature.com/scientificreports/ fishes in Wakatobi Marine National Park, Indonesia. Clionaids appear to be palatable to spongivores as numerous bite marks on C. patera have been recorded (Table S4). Rapid regeneration and anti-predatory mechanisms are critical for the survival of sponges, particularly in large reef sponges such as Neofibularia nolitangere, Ircinia strobilina and Agelas clathrodes in the Caribbean 88 . In C. celata, rapid regeneration enabled papillae consumed to regenerate as quickly as within 12 days 89 . Similarly, we found that C. patera generally had remarkably high rates of recovery, taking only about three weeks to seal a 20 mm diameter core hole (Fig. 5C,D). In C. celata, regeneration rates were generally correlated with high current flow (present in SIMP), possibly aiding regeneration by reducing the amount of energy expended for feeding or waste removal 89 . Cliona patera could employ a similar strategy to their Caribbean counterparts, coupling mechanical defences (spicules) with rapid regeneration to maximise their survival chances under predatory stress.
Upon relocation, three of the six sponges were able to either maintain constant height, and one even grew by 8% (Fig. 5A). Nevertheless, two individuals fared poorly, with their cups almost entirely consumed (e.g. Fig. 5B). A recent paper by Wulff 79 debunked the popular binary of sponges being either "palatable" or "deterrent" based on frequency of consumption of sponge pellets, proposing that sponge defenses are predator-and habitatspecific. By transplanting the sponges to SIMP, the sponges might have inadvertently been exposed to increased opportunistic predation, particularly by aggregating them together. In addition, wounding of the sponge during regeneration experiments could have altered their gene expression, requiring allocation of metabolic resources toward regeneration and possibly triggering the release of other spongivore-attracting metabolites 90,91 . Future conservation efforts on this sponge need to carefully consider the diversity of potential spongivores in the target site prior to any relocation of individuals. We note that only two out of the six individuals at SIMP were preyed on severely, so more data on the effects of sponge aggregation on predation are needed to further assess this strategy. Finally, considering the trade-off in resource allocation between growth and gametogenesis in clionaids 92 , it is highly likely that individuals E and F (Fig. 5A,B) would prioritise recovery and growth and not be sexually active for some time.
The reproduction biology of Cliona has been studied extensively for several species, including Cliothosa delitrix 93 , C. tenuis 92 , C. vermifera 94 , C. celata and C. viridis 95 . In these studies, C. vermifera and C. tenuis were determined to be gonochoric, C. celata and C. viridis hermaphoroditic, and Cliothosa delitrix mostly gonochoric with some hermaphrodites observed. With the exception of C. vermifera, all other clionaids were oviparous. In addition, gamete release generally occurred under warmer temperatures, with up to 90% of sexual reproduction occurring annually. Interestingly, despite the high percentage of Cliothosa delitrix individuals found to contain reproductive structures 93 , a population study across the Greater Caribbean found > 12% of the samples were clones (n = 495) 96 , contrary to an earlier study which found no clones (n = 47) 97 . Based on available literature, we propose that C. patera is unlikely to reproduce by fragmentation, similar to other stalked sponges due to its inability to regenerate the attachment stalk and poor attachment 98,99 . Therefore, it is critical for cross-fertilization to be the main driver for increasing not only the threatened species' genetic diversity but also population abundance to ensure its long-term viability.
The low abundance and genetic diversity of this sponge population are pressing concerns for the species. Despite stringent legislative protection from associated anthropogenic impacts (e.g., seabed dredging, anchoring, coastal reclamation and harvesting) on the transplanted individuals, inbreeding between the small pool of individuals could continue to erode their genetic fitness, resulting in increased risk of extinction. In the future, by enacting partnerships between countries harbouring C. patera populations, with in situ individuals currently found only in Cambodia, Singapore and Thailand, the global genetic diversity of this species can be estimated across populations. Using high-throughput sequencing techniques such as those applied in this study, the global population genomics of C. patera can be examined to guide conservation action plans and enhance the species' genetic diversity. Critically, its exceptional rarity throughout the region may require ex situ conservation efforts, including transplanting and propagating genetically distinct individuals in aquarium settings. Public aquaria harbour a rich diversity of marine organisms, often including threatened species, and many have been involved in conservation and reintroduction programmes 100 . However, while some clionaids have been kept in laboratory aquaria successfully for experiments, these are mostly performed for encrusting forms [101][102][103] . Clearly, the reproductive biology of C. patera and its viability in the aquarium must be carefully assessed to mitigate the risk of losing the entire population. If successful, individuals reared in captivity may be relocated back to the wild, within suitably identified sites after careful assessment as outlined here.