1 INTRODUCTION

Globally, hypersaline lakes serve as important habitats for migratory water birds and host unique microbial populations (Saccò et al., 2021a). However, increasing climate and atmospheric variations have led to environmental pressures faced by hypersaline lakes to strength over time (Wurtsbaugh et al., 2017). In some cases, this has initiated longer periods of dryness causing local biota to withstand extended desiccation periods (Schmidt et al., 2021). This, paired with the invasion of opportunistic species, can lead to long-term ecological damage and lowered biodiversity (Varó et al., 2015; Sánchez et al., 2016). Moreover, the true ecological importance regarding the diversity of hypersaline lakes is yet to be concisely and extensively researched (Saccò et al., 2021a). This lack of knowledge has contributed to uncontrolled human-induced exploitation of thesewetlands, some lakes being irreversibly damaged, with socio-economic benefits of lake use preferred over its ecological importance (Lerman, 2009; Zadereev et al., 2020).

The ecological knowledge of species allows the interpretation of biodiversity patterns to assess environmental pressures and impacts on ecosystems (Brantschen et al., 2021). Routine biomonitoring often classifies ecosystem states through biotic indices that succinctly summarise the information of species assemblages, allowing for comparison to reference states or systems (Diaz et al., 2004). The classification of ecological integrity based on biotic indices is generally focused on certain taxonomic groups (Pawlowski et al., 2018). In hypersaline ecosystems, macroinvertebrates—e.g., Diptera, Coleoptera, and Artemia species (Velasco et al., 2006; Mohebbi, 2010)—are among the most used bioindicators. For these groups, monitoring generally involves net capture, preservation, and morphological identification of specimens. This type of monitoring can be costly in terms of time and money, requiring expert taxonomic skills and often missing rare, small, or elusive species. In the last decade, DNA metabarcoding approaches have proven effective for the assessment of community assemblages (Brantschen et al., 2021). Thereby, DNA extracted from an environmental sample (environmental DNA, eDNA) provides information about the possible occurrence and distribution of species (Thomsen and Willerslev, 2015; Takahashi et al., 2023). In several instances, eDNA sampling was shown to complement traditional approaches for the assessment of biological indicators and the ecological state of ecosystems (Kelly et al., 2017; Brantschen et al., 2021). However, monitoring based on eDNA sampling has yet to be explored in hypersaline lakes for macroinvertebrates. As a proof of concept, this study investigates the outcomes of morphology-based taxonomic analyses and eDNA techniques by targeting macroinvertebrates communities in five hypersaline lakes from Rottnest Island (Wadjemup) in Western Australia.

2 MATERIAL AND METHOD 2.1 Study area and field work

Rottnest Island (Wadjemup) is a small island located 18 km off the coast of Perth, Western Australia (Fig. 1). It is a well-known tourist destination, with an average of 500 000 visitors per year, which creates a significant demand for water resources on the island (Jackson, 2008). Rottnest Island is a natural reserve providing habitat for the marsupials Setonix brachyurus (Lesson, 1842)—aka quokkas (Nicholls, 1971)—and reptiles (How and Dell, 1994), migratory wading birds (Saunders and De Rebeira, 2009), and subterranean fauna (Perina et al., 2022), among other animals. The island is protected under the highest level for public land in Western Australia (A-class reserve) under the Land Administration Act 1997, and hosts six permanent hypersaline lakes covering approximately 10% of the island's surface area with salinity ranges from 118–187 mg/L (Saccò et al., 2021b).

Water and net samples were collected from five hypersaline lakes (Garden Lake, Herschel Lake, Lake Baghdad, Lake Vincent, and Serpentine Lake) in November 2020, applying a sampling effort proportional to the size of the lakes and with a maximum of 500 m between sampling sites (Fig. 1). Macroinvertebrates from the water column were collected (and preserved in 70% ethanol) in each lake using a long-handled circular frame net with a mesh size of 250 µm and an opening of 30 cm in diameter, following a one-meter kick sample and sample storage method (utilizing 1-L plastic bottles) adapted from Chessman et al. (2002). Samples were collected across multiple sampling points (SP) within the five lakes (Garden: 1 SP; Herschel: 2 SP; Baghdad: 3 SP; Vincent: 1 SP; Serpentine: 3 SP). In addition, five 1-L samples (n=50) were collected by hand, and scooping through the water column, at each sampling point for eDNA metabarcoding. All samples were frozen until further processing in the Trace and Environmental DNA laboratory (TrEnD) at Curtin University in Perth, Western Australia. Invertebrates were counted and identified under a stereomicroscope to the lowest taxonomic level possible with the help of specific taxonomic keys (Davis and Christidis, 1997). For further information about the sampling site and physical-chemical parameters see Saccò et al. (2021b).

2.2 Genetic investigation

Five 1-L water sample replicates were taken from each sampling location from the five lakes (n=50) and filtered using a Sentino peristaltic microbiology pump (Pall Life 126 Sciences; New York, NY, USA) through 0.45-µm sterile membrane filters (Pall Life Sciences; New York, NY, USA). Membrane filters were incubated (56 ℃) overnight in a solution of ATL buffer (540 µL) and Proteinase K (60 µL). Digests were extracted with a QIAcube (Qiagen; Venlo, The Netherlands) with the final DNA eluted off the silica column in 100-µL AE buffer. Macroinvertebrate DNA was targeted with the fwh COI assay (Vamos et al., 2017). The quality and quantity of the DNA extracts (neat extract, as well as 1/10 and 1/100 dilutions) was assessed via qPCR (Applied Biosystems [ABI]; Foster City, CA, USA). The cycling conditions were as follows: initial denaturation at 95 ℃ for 5 min, followed by 40 cycles of 95 ℃ for 30 s, 52 ℃ for 30 s, 72 ℃ for 45 s and a final extension at 72 ℃ for 10 min. Additionally, an extraction control and negative PCR control were processed along with samples. Details regarding the PCR master mix recipes and library preparation are described in Saccò et al. (2021b). The final library concentration was determined using a QuBitTM 4 Fluorometer (Thermo Fischer; Melbourne, Australia) and sequenced using a 300-cycle V2 kit on an Illumina MiSeq platform (Illumina; San Diego, CA, USA).

2.3 Bioinformatics and data analysis

Sequence data (1 835 985 reads) was processed using the automated workflow ' eDNAFlow' (Mousavi-Derazmahalleh et al., 2021) which comprises USEARCH (Edgar, 2016) and BLASTN (Altschul et al., 1990). The fastx-uniques, unoise3 (with minimum abundance of 4) and otutab commands of USEARCH were applied to generate unique sequences, ZOTUs (zero-radius OTUs) and abundance table, respectively. The ZOTUs were compared against the BOLD and NCBI databases using the following parameters in BLASTN: perc_identity94, evalue 1×10^-3, best_hit_score edge 0.05, best_hit_overhang 0.25, qcov_hsp_perc 100, max_target_seqs=5. An inhouse Python script was used to assign the ZOTUs to their lowest common ancestor (LCA). A custom python script (Mousavi-Derazmahalleh et al., 2021) was then used to taxonomically assign ZOTUs by the lowest common ancestor (LCA) approach, with taxonomic assignment collapsed to the LCA if the percent identity of two hits with 97% query cover and 96% identity, differed by < 1%. Low abundant ZOTUs without at least one sample above 0.003% sequences assigned, were considered unreliable and excluded from the dataset (Vamos et al., 2017). An average of 27 774 reads were achieved per sample. Any ZOTUs from extraction and negative controls were removed from the dataset. Non-metric multidimensional scaling (nMDS) was performed on the community matrix involving OTUs and taxa levels, as well as on microbial community data described in Saccò et al. (2021b). Calculations were performed on both OTUs and taxa because more than 1 OTU can be assigned to the same taxa. Patterns inferred from these two taxonomic levels can thus differ. The correlation between nMDS ordinations was calculated with a correlation-like statistic derived from the symmetric Procrustes sum of squares and tested with a protest (Jackson, 1995) with the vegan package (Oksanen et al., 2020) of the R Statistical software (R Core Team, 2021).

3 RESULT

The combination of eDNA metabarcoding with morphology-based taxonomic analysis described 16 taxa belonging to the orders Anostraca, Diptera, Isopoda, and Coleoptera (Table 1). Metabarcoding accounted for 11 of the described taxa, whereas the morphology-based taxonomic analysis accounted for 5. Metabarcoding detected not only higher richness but also increased taxonomic resolution compared to morphological identification (Table 2). More in detail, metabarcoding analysis revealed 30 OTUs and 13 taxa of which 9 at species level, 3 at genus level, and 1 at family level. The 5 taxa found through net sampling referred to 1 at species level, 1 at genus level and 3 at family level (Table 1). Taxa richness in each lake was low, ranging from 2.9 (Hershel) to 4.2 (Serpentine) with metabarcoding and from 2 (Baghdad, Hershel, Serpentine, Vincent) to 4 (Garden) with morphological identification (Table 2). For metabarcoding data, richness was calculated as the average richness of samples collected within each lake. For eDNA metabarcoding, OTU richness was higher than taxa richness and there was no apparent relationship between the two estimates (e.g., the lowest OTU richness resulted in the second higher taxa richness).

Artemia parthenogenetica (Bowen, 1978) and Ephydridae were the only taxa identified by both techniques. Within the order Anostraca, metabarcoding analysis assigned OTUs to Artemiidae (family level), Artemia parthenogenetica and Artemia tibetiana (Abatzopoulos, 1998) in all the lakes. Contrarily, the net-based survey enabled identification of Artemia parthenogenetica solely. Morphology-based data indicated that Garden Lake hosted dipterans families of Chironomidae, Ceratopogonidae, and Ephydridae. The latter taxon was also present in all the other lakes except for Serpentine, where family Stratiomyidae was uniquely detected. Coleoptera (Ochthebius queenslandicus (Hansen, 1998)) was also solely found at Serpentine Lake, but only via metabarcoding.

Taxonomic assessment of Diptera through eDNA metabarcoding allowed identification at genus/ species level in all the lakes, except for Lake Vincent where no dipterans were identified. These taxa included Aedes atropalpus (Coquillett, 1902) (Garden), Tanytarsus barbitarsis (Freeman, 1933) (Garden and Serpentine), Bryophaenocladius (Serpentine), Culicoides waringi (Lee and Reye, 1955) (Herschel, Baghdad and Serpentine), Philygria nigrescens (Cresson, 1930) (Baghdad and Serpentine), and Eristalinus aeneus (Scopoli, 1763) (Serpentine). We found an inherent variability in the occurrence probability (calculated as the number of samples in which a taxon was detected over the total number of samples) of taxa within each sampling point when using metabarcoding (Fig. 2). The highest occurrence probability, averaged across all sampling points, was found for Artemia parthenogenetica (100%), followed by Artemia tibetana (96%), Artemia (88%) and Tanytarsus barbitaris (60%). The other taxa were detected in less than 50% of the samples within each sampling point.

nMDS performed on OTUs and taxa had a stress of 0.09 and 0.06. The correlation between ordinations at OTU and taxa level was 0.50 (P < 0.001). Samples did not cluster according to lakes for both taxonomic levels (Fig. 3a & c), contrarily to microbial community as highlighted by the Procrustes analysis (Fig. 3b & d). Ordinations at OTU R=0.38, P < 0.001) and taxa (R=0.24, P > 0.1) levels for arthropods showed weak correlations with the ordination based on the microbial community.

4 DISCUSSION AND CONCLUSION

Aquatic invertebrates provide key food supplies for wading birds in vegetated and un-vegetated wetlands (Saccò et al., 2020 and references therein), and due the multiple ecosystem services they deliver (e.g., control of microbial and algal proliferation, maintenance of the energy flows), they are frequently used as ecological indicators in freshwater ecosystem studies (e.g., Wallace and Webster, 1996). However, despite the importance of saline and hypersaline wetlands for the conservation of biodiversity and the fight against climate change, significant less effort in designing efficient macroinvertebrate-based assessment protocols has been implemented. Our preliminary study is, to our best knowledge, the first exploring hypersaline macroinvertebrate communities through a combination of conventional net-based surveys and eDNA metabarcoding data.

The aquatic invertebrate communities at Rottnest Island (Wadjemup) lakes were like other coastal saline water fauna in Western Australia (Halse, 1981) and South Australia (De Deckker and Geddes, 1980), being characterised by a dominance of brine shrimps Artemia. When compared, our morphological results were in line with metabarcoding records at family level. However, molecular data also indicated the presence of two different species of brine shrimps (A. parthenogenetica and A. tibetiana) across all the lakes. Populations of A. parthenogenetica have been recorded across Western Australia including Rottnest Island (Wadjemup) via morphologically assessments (McMaster et al., 2007), however taxonomists regard this species as being used incorrectly referring to parthenogenetic populations of Artemia that do not form a true species (Asem et al., 2010, 2020). Similarly, A. tibetiana is considered as a A. urmiana based on the occurrence of nuclear gene flow between the type locality of A. tibetiana and populations of A. urmiana (Sainz-Escudero et al., 2021). Therefore, more in depth molecular studies (e.g., mtDNA) will be necessary to confirm the exact species of Artemia occurring on Rottnest Island (Wadjemup) and to further collaborate the hypothesis that Artemia were transported to island via the East Asian-Australasian Flyway (McMaster et al., 2007).

Our eDNA-based diversity assessment revealed the presence of the endangered oniscid isopod Haloniscus searlei (Chilton, 1920) at Garden Lake—the least saline system out of the five lakes. Endemic to southern Australia but widespread to Western Australia as well, this halobiont with terrestrial origins has considerable osmoregulatory capabilities that allow him to survive extreme and instable environments such as hypersaline lakes (Williams, 1983). When compared with a previous report by Edward (1983), who found this species in a vast range of permanent saline waters at Rottnest Island (Wadjemup), our results suggest a decline in distribution demanding for urgent further assessment.

Both methods—eDNA and nets—revealed Diptera to be the most diverse order overall, being most abundant in Garden Lake (least saline) and Lake Serpentine (most saline). Within this Order, Tanytarsus barbutarsis, a taxon previously found to have the widest range of salinities ever recorded (Edward, 1983), was only detected via eDNA metabarcoding in these two lakes. Previous studies have indicated that aquatic Diptera, such as Ephydridae, occur in marginal habitats that limit other sets of life forms, as it provides an enemy-free space to specialize in (i.e., Adler and Courtney, 2019). Overall, this suggests that salinity may not be a direct driver for the prevalence of dipteran species in the Rottnest Island (Wadjemup) lakes, with further work needed to understand the ecological drivers shaping the invertebrate communities in these systems.

Interestingly, our eDNA metabarcoding analysis detected Aedes atropalpus in Garden Lake, a species that has been flagged as exotic in Australia, most frequently detected through inspection and surveillance of imported cargo and international conveyances (Schaffner et al., 2013). Additionally, Ochthebius queenslandicus, a coastal marsh water beetle found across Australia (Perkins, 2007), was also detected with metabarcoding approaches. The detection of these species validates the importance of eDNA surveying as a tool to monitor the potential exotic species entering the island, and to identify the use of habitat by species that are not strictly halophile.

The detection of among-lake differences involving arthropods was less effective than by using the microbial community, and this can be ascribable to a series of reasons. Differences in salinity as well as other hydro-chemical characteristics can readily affect the microbial community composition as demonstrated by Saccò et al. (2021b) in the same lakes studied in this research. On the contrary, arthropod richness is reduced by salinity (Timms, 2022) probably because high salt concentration affects the osmoregulatory physiology of freshwater organisms (Griffith, 2017). Moreover, cross-tolerance to other stressors is likely in halophile species thus hindering the effect of differences in other chemical characteristics of lakes (Pallarés et al., 2017). However, given the functional links between these two crucial components (Mieczan et al., 2018; Blanchette et al., 2020), combined investigations on both microbial and invertebrate communities are essential to improve the quality of the ecological assessment in hypersaline ecosystems.

Overall, saline and hypersaline waters provide challenging conditions for eDNA analysis due to the high salt concentrations (often coupled with elevated UV radiations) affecting the DNA preservation and the filtering of water (Barnes et al., 2014). However, recent advances in filtering (e.g., use of syringes, see Muha et al., 2019) and in field extracting techniques (e.g., Yates et al., 2019) are opening new exciting doors to the use of molecular tools often remote ecosystems such as saline lakes, as well as sinkholes (Elbourne et al., 2022), subterranean environments (Saccò et al., 2022), deep oceans (McClenaghan et al., 2020), and coral reefs (West et al., 2020). In fact, despite the above limitations, we consider the combination of conventional sampling approaches and novel molecular tests hosts great potential for the ecological assessment of saline and hypersaline lakes. For instance, as indicated in freshwater ecosystems, the identification at species level as well as the identification of OTUs and even haplotypes can enhance our capacity to infer community dynamics (Laini et al., 2020, 2023). Moreover, genetic characterization of organisms can contribute to identification of key taxa and species with complex diagnostic characters and facilitate the early detection of alien species in freshwater systems (Macher et al., 2016; Blackman et al., 2022). On the other side, DNA metabarcoding to date prevents the assessment of quantitative data (e.g., abundance, biomass) mainly because of species-specific primer efficiency (Elbrecht and Leese, 2015; Takahashi et al., 2023). However, the quantitative information provided by morphological characterization, once combined with the increased taxonomic resolution obtained with metabarcoding, can provide a timely, cost-effective, and insightful approach for a better ecological assessment of saline and hypersaline lakes.

5 DATA AVAILABILITY STATEMENT

Taxonomy data can be found at github.com/ MACampbell91/Nets-meet-eDNA-in-salt-lakes.

6 ACKNOWLEDGMENT

We wish to acknowledge the Traditional Custodians of this Island, the Whadjuk people of the Noongar Nation, their ancestors and their Elders past, present and emerging. We acknowledge and respect their continuing culture and the contribution they make to the life of the Perth and Rottnest Island (Wadjemup) regions. This work was supported by resources provided by the Pawsey Supercomputing Centre with funding from the Australian Government and the Government of Western Australia. M. S. and N. E. W acknowledge support from the BHP-Curtin alliance within the framework of the "eDNA for Global Environment Studies (eDGES)" program.

Open Access

This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.