Introduction

Benthic macroinvertebrates are a key biological quality element for the assessment of the ecological quality of rivers worldwide, because of their ability to indicate pollution and ecosystem degradation (Hering et al. 2006; Metcalfe-Smith 2009; Kalogianni et al. 2017). The methodologies followed, so far, for sampling, sorting and identification of benthic macroinvertebrates are time consuming and involve intense labor and high levels of expertise. Moreover, the morphological identification of river benthic macroinvertebrates to genus or species level is a challenging task because benthic macroinvertebrates are a very diverse group of organisms and many of the specimens collected are in larval stages and sometimes, they are not in pristine condition for efficient morphological identification. Ntislidou et al. (2020) highlighted the impact of human error during the identification process on water quality assessment and Haase et al. (2010) showed that 29% of the specimens are being overlooked during the sorting process, more than 30% of taxa differ between the analysts at the identification process and 16% of the samples differ at the final ecological assessment. These methodological caveats could have serious implications on river management and restoration plans.

Over the last fifteen years, the development of environmental genomics combined with high-throughput sequencing (HTS) technologies has led to an advance in surveying biodiversity and monitoring natural ecosystems at a broad spatiotemporal scale (Carraro et al. 2020; Brantschen et al. 2021; Seymour et al. 2021). Specifically, environmental DNA (eDNA) which is defined by Taberlet et al. (2012) as the “DNA that can be extracted from environmental samples (such as soil, water or air), without first isolating any target organisms” has become a valuable tool in a wide range of ecological studies (Taberlet et al. 2018). eDNA is released into the environment through cells, tissues, gametes, mucus, etc. or through the decomposition of dead organisms. It consists of a complex mixture of intracellular and extracellular DNA from many different organisms, and it is possibly degraded (Taberlet et al. 2012).

Recent studies have shown that eDNA metabarcoding can contribute towards new eco-genetic tools that will allow a more efficient, less subjective, cost-effective, and non-invasive alternative to traditional water quality assessments (Leese et al. 2016; Elbrecht et al. 2017; Hering et al. 2018; Pawlowski et al. 2018; Ruppert et al. 2019). It has been repeatedly shown that more species of benthic macroinvertebrates are being detected through eDNA metabarcoding compared to the conventional kick-net sampling followed by morphological identification, but community composition differs among the two methods (Wang et al. 2021; Brantschen et al. 2022; Ji et al. 2022). The dissimilarities in community composition could derive from the fact that eDNA can be transported from the upstream part of the river (Deiner and Altermatt 2014), DNA from different taxa can have different persistence times in the water (Goldberg et al. 2013) and biases of PCR primers during the amplification process can also play a significant role. However, pollution gradient and river ecological quality can be effectively assessed using eDNA metabarcoding data of benthic macroinvertebrates (Fernández et al. 2019; Ji et al. 2022). Brantschen et al. (2021) who compared eDNA metabarcoding versus traditional sampling and taxonomy methods of benthic invertebrates found that ecological classification was the same for 72% of the sampling sites.

Nevertheless, bulk sample DNA metabarcoding through tissue-homogenization has been proposed as a better alternative to eDNA metabarcoding because it can reveal higher diversity than traditional methods and represent local stream macroinvertebrate communities more accurately (Gleason et al. 2021; Pereira-da-Conceicoa et al. 2021). Still, this method, as well as eDNA metabarcoding from preservative ethanol of bulk samples, rely on the conventional kick-net sampling which is highly invasive as it alters the river substrate and extracts the specimens from the ecosystem. It also requires great sampling effort and has some practical limitations regarding access problems to deep river sites. Collecting and filtering water samples simplifies field workflows and more importantly it is a less invasive procedure towards the ecosystem.

Currently, there is no consensus regarding eDNA sampling, capture, preservation and extraction methods or the choice of PCR primers, sequencing platforms, reading depths and bioinformatic pipelines. This leads to sampling, laboratory and bioinformatic biases and contradicting results among different studies. The standardization and intercalibration of the new molecular methods for freshwater macroinvertebrate biomonitoring is of paramount importance for increased chances of future uptake of those methods into routine monitoring programs like the Water Framework Directive (2000/60/EU) (Blackman et al. 2019; Lefrançois et al. 2020). To this end, the development of standards under the European Committee for Standardization (CEN TC230/WG28) regarding water sampling for capture of macrobial environmental DNA in aquatic environments is currently in progress (oSIST prEN 17805:2022).

Since the use of river macrobial eDNA in ecology is a fast-evolving field, the main objective of this scoping review is to provide a synthesis of the existing methodologies of eDNA sampling, extraction, and amplification, focusing on river benthic macroinvertebrates. The two primary aims of this study are (a) to provide a comprehensible overview of methods that will help identify possible knowledge gaps, and techniques that need harmonization, and (b) to discuss the advantages and limitations of the methods currently used.

Methods

This review follows the reporting guidelines of PRISMA Extension for Scoping Reviews (PRISMA-ScR) (Tricco et al. 2018).

For this study, literature searches were performed in the Web of Science on 20 May 2022. The results of the search were refined by trial-and-error queries, that is, by screening the first 100 records and identifying the search terms that would minimize the number of irrelevant records. The sensitivity of the query was increased by adding more synonyms to each of the search terms and the specificity by excluding references that addressed other groups of organisms such as fish and diatoms.

This procedure resulted in the following search being the most effective for the objectives of the current review.

ALL=(“environmental DNA” OR eDNA OR bulk) AND ALL=(river* OR freshwater OR waterc* OR watersh* OR waterw* OR lotic OR stream OR tributar* OR channel* OR aquatic) AND ALL=(invertebrate* OR macroinvertebrate* OR benth* OR “aquatic insect*”) AND ALL=(metabarcoding OR “high throughput sequencing” OR HTS OR “next generation sequencing” OR NGS) NOT TI=(diatom*) NOT TI=(fish).

This query returned 121 results. Twelve of them were review articles and were excluded from further processing.

Additionally, the “Related records” tool in the Web of Science was used to search for articles that share the same references with the “parent” record i.e., the most relevant record to the search terms at the top of the list, in order to identify records that could be included in this review, but their titles, abstracts, or keywords did not meet the search terms. This search returned 49,718 records and, following a citations title screening, the first 146 records that shared at least 5 references with the “parent” record were retained, after excluding 1,318 review articles.

The same search query was performed in the Scopus citation database on 20 May 2022 and returned 89 documents, 7 of which were review papers and were excluded. The search query was:

(TITLE-ABS-KEY ({environmental DNA} OR edna OR bulk) AND TITLE-ABS-KEY (river* OR freshwater OR waterc* OR watersh* OR waterw* OR lotic OR stream OR tributar* OR channel* OR aquatic) AND TITLE-ABS-KEY (invertebrate? OR macroinvertebrate? OR benth* OR {aquatic insect?}) AND TITLE-ABS-KEY (metabarcoding OR {high throughput sequencing} OR hts OR {next generation sequencing} OR ngs) AND NOT TITLE (fish) AND NOT TITLE (diatom?))

Furthermore, an alert for our searches was created both in the Web of Science and the Scopus database from 20 to 2022 until September 2022, which generated 8 additional references.

The above lists of 109, 146, 82 and 8 records were imported in Mendeley Desktop Reference Manager and duplicates were erased resulting in the final reference list of 262 records. Subsequently, a screening process was performed to identify references eligible for inclusion in the review. For a study to meet the inclusion criteria for this review it should:

  1. i.

    address eDNA and river benthic macroinvertebrates, and

  2. ii.

    employ eDNA metabarcoding, and

  3. iii.

    include information on eDNA sampling, extraction, and amplification methods, and

  4. iv.

    be written in English.

After the title and abstract screening, 173 references were retained, based on the aforementioned criteria, followed by a full text screening that retained the 46 references included in this review (Online Resource 1; List of references).

Full text and supplementary data of each of the 46 references were thoroughly studied and the following information was extracted: authors, title, year of publication, type of environmental samples (i.e., water, sediment etc.), sample volume (ml), number of sample replicates, filter type and pore size (µm), sample preservation and storage conditions, negative and positive controls, number of PCR replicates, methods of DNA extraction, primers used for PCR amplification and sequencing platforms. Publications that used more than one type of environmental samples or more than one method of extraction and/or amplification resulted in multiple entries in the database.

Results

Systematic literature search

The systematic literature search resulted in 46 references that were eligible to be included in this review (Online Resource 1; List of references). The flow diagram in Fig. 1 provides an overview of the sources screened, the number of the references assessed on each step of the systematic literature search and the reasons for exclusion. The results showed that over the last decade the number of published papers that use eDNA metabarcoding to study river benthic macroinvertebrate communities have been rapidly increasing from 1 paper/year in 2012, 2015 and 2016 to 18 papers on 2021 (Fig. 2).

Fig. 1
figure 1

PRISMA diagram with the key steps of the systematic literature search

Fig. 2
figure 2

Number of publications per year regarding eDNA metabarcoding applications on river benthic macroinvertebrates

eDNA sampling

There are many aspects to be considered as far as eDNA sampling is concerned, such as the type and volume of the environmental sample (i.e., water, sediment, etc.), the use of sample replicates and field blanks, the type and pore size of the filter (when applicable) and the storage conditions of the samples. The results of this review show that lotic freshwater research is focused primarily on the use of eDNA from water samples, followed by eDNA from preservative ethanol of bulk macroinvertebrate samples, while the use of sediment eDNA for monitoring freshwater ecosystems has been relatively limited (Fig. 3). Furthermore, innovative studies conduct research on new types of macroinvertebrate eDNA samples such as biofilm (Fig. 3) proposed by Rivera et al. (2021).

Fig. 3
figure 3

Sankey diagram depicting the types of environmental samples, and the type and pore size of the filters used during the filtration process (when applicable). Relative size of the final nodes indicates the proportion of publications that employ each type of environmental sample and the number of times each type of filter and each pore size were used in the reviewed literature. Diagram made using SankeyMATIC online (https://sankeymatic.com/build/)

The vast majority of the publications (32 out of 46) list water filtration as a method for eDNA isolation, and the most frequently used filters are various types of filters of 0.2–0.22 μm pore size, followed by 0.7 μm GF/F filters and 0.45 μm Cellulose Nitrate filters (Fig. 3, Online Resource 1; Table S1). 63% of the river water eDNA studies use sample replicates and the mean volume of water filtered is 858ml. On the other hand, only 33% of sediment and preservative ethanol eDNA studies use sample replicates (Table 1). The preservation of the samples is achieved mostly by freezing at -20oC or below (51% of publications), or by ethanol of at least 95% concentration (16% of the publications) or by a combination of both (18% of publications) (Online Resource 1; Table S2).

Table 1 Proportion of publications that use sample replicates and mean sample volume for each of the four types of environmental samples in the reviewed literature

eDNA extraction, amplification and sequencing

Commercial extraction kits such as DNeasy Blood and Tissue Kit (Qiagen) and PowerWater DNA Isolation Kit (Mobio laboratories) are commonly used in studies that target river benthic macroinvertebrate communities for the extraction of DNA from environmental samples. However, phase separation and precipitation methods such as Phenol-Chloroform Isoamyl DNA extraction, followed by ethanol precipitation and salt-precipitation protocol (Weiss and Leese 2016) are also frequently used (Fig. 4, Online Resource 1; Table S3).

Fig. 4
figure 4

DNA extraction methods from environmental samples regarding eDNA metabarcoding on river benthic invertebrates. Bar chart shows the number of times each type of extraction method was used, in the reviewed literature

As shown in Fig. 5 (Online Resource 1; Table S4), a variety of primer pairs are used for the amplification of benthic macroinvertebrate eDNA ranging from more universal primer pairs such as mlCOIlintF/jgHCO2198 and BF2/BR2 to primer pairs that target specific groups of benthic macroinvertebrates such as Crust16S_F(short)/Crust16S_R(short) and MOL16S_F/MOL16S_R. Most of the primers used target fragments of the mitochondrial cytochrome c oxidase subunit I (COI) gene, which is the standard DNA barcode marker for animals (Fig. 5, Online Resource 1; Table S4).

Fig. 5
figure 5

Primer pairs used for eDNA amplification in river benthic macroinvertebrate studies. Bar chart shows the number of times each primer pair was used in the reviewed literature

Negative controls are usually included in eDNA analysis workflows to check for sample contamination. In the reviewed literature, most of the studies employ negative controls for water filtration (when applicable), DNA extraction and/or PCR amplification (Fig. 6, Online Resource 1; Table S5). On the other hand, only 26.5% of the studies use amplification positive controls (Fig. 6, Online Resource 1; Table S5). Moreover, 65% of the studies use 3 PCR replicates on average and the remaining 35% of the studies does not utilize any replicates during the amplification process (Online Resource 1; Table S5).

Fig. 6
figure 6

Proportion of publications that use controls at each process step

Regarding the choice of sequencing platforms in the metabarcoding analysis, Illumina MiSeq is the most used sequencing platform in the reviewed literature (76% of the publications) followed by Illumina HiSeq (10%), Ion Torrent Proton (6%), Ion Torrent PGM (4%), Illumina NovaSeq (2%) and Roche 454 (2%) (Online Resource 1; Table S6).

Discussion

eDNA sampling

The main purpose of this review was to summarize the existing knowledge of the methods used in eDNA metabarcoding river benthic macroinvertebrate studies. The results showed that efforts in this field are primarily focused on eDNA captured from river water rather than river sediment or bulk sample preservative ethanol. However, many studies have found that eDNA metabarcoding of bulk sample preservative ethanol is a reliable surrogate for traditional kick-net samples in terms of being able to detect spatial stream communities’ variation (Erdozain et al. 2019; Martins et al. 2019, 2021). Moreover, Wang et al. (2021) found that preservative ethanol eDNA outperforms eDNA from water samples on detecting benthic invertebrate local biodiversity, whereas eDNA from water samples performed better considering broad scale biodiversity patterns. On the other hand, eDNA metabarcoding of preservative ethanol tends to underestimate or miss taxa that are rare, small-bodied or have pronounced exoskeleton when compared to conventional morphotaxonomy or tissue-homogenisation metabarcoding of bulk samples (Zizka et al. 2019b; Martins et al. 2021).

A few studies found that water eDNA does not adequately represent local aquatic macroinvertebrate communities and suggest that capturing eDNA from sediment samples should be further explored as it might be more representative of local benthic invertebrate communities (Gleason et al. 2021; Wang et al. 2021). Most benthic invertebrates inhabit the river’s substrate and macroinvertebrate eDNA from the water column could also accumulate in the substrate. To this end, Ji et al. (2022) captured eDNA from sediment samples and found that the number of macroinvertebrate species identified via eDNA metabarcoding at each site was higher than the number of species identified by the traditional morphology method, and Arthropoda, Annelida, and Mollusca, in particular, had statistically significant higher species richness. This could be explained by the fact that eDNA can be accumulated in the sediment and be preserved there even if the source of DNA is no longer present (Turner et al. 2015). Contrary, when studying fish river biodiversity, Sakata et al. (2021) showed that eDNA from sediment samples exhibited high spatial heterogeneity in species composition within a survey site, whereas water eDNA was found to be more homogeneously distributed. Consequently, sediment samples should have many replicates within a site to reflect the local biotic communities. It should also be noted that a newly proposed method that captures macroinvertebrate eDNA in rivers from biofilms shows the capacity of biofilms to retain macroinvertebrate eDNA and should be further explored (Rivera et al. 2021).

This review found that, for the majority of the publications, the capture of river macroinvertebrates eDNA, is done through water filtration (Fig. 3), which has been found to have higher detection rate for macroinvertebrate species compared to capture by precipitation (Deiner et al. 2015). Since eDNA is mostly captured from river water samples, another aspect to be considered during sampling is the volume of the water that needs to be filtered to efficiently represent the local macroinvertebrate communities. One liter is a commonly used volume of water filtered for the assessment of river macroinvertebrates through eDNA metabarcoding (Fernández et al. 2018, 2019; Macher et al. 2018; Bagley et al. 2019; Hajibabaei et al. 2019; Uchida et al. 2020; Clusa et al. 2021; Hupalo et al. 2022; Reinholdt et al. 2021; Seymour et al. 2021). Machler et al. (2016) who sampled different volumes of river water ranging from 0.25 to 2 L and studied the effect of sampled water volume on the detection of three macroinvertebrate species, suggest a minimum water volume of 1 L to reduce false negative detections.

The pore size of the filter used to capture eDNA is also an important factor that could affect the results of eDNA metabarcoding. The findings of this review showed that in river macroinvertebrate eDNA metabarcoding studies, filters of 0.2–0.22 μm, 0.45 and 0.7 μm pore sizes are mostly used (Fig. 3). In general, the amount of eDNA yielded increases with the decrease of filter pore size (Renshaw et al. 2015) but small pore size filters can easily clog in turbid environments (Deiner et al. 2016; Jeunen et al. 2022; Keck et al. 2022). Pre-filtration of water can be applied to overcome this issue. However, Majaneva et al. (2018) showed that although pre-filtration gives higher diversity index values, it lowers the amount of DNA yielded. Another way to overcome this issue is by using multiple filters until the desired volume of water filtered is achieved, which can increase the price per sample drastically.

Regarding the different filter membrane materials, Majaneva et al. (2018) found that mixed cellulose ester membrane filters (CN, 0.45 μm) yielded more DNA compared to polyethersulfone (PES, 0.2 μm) filters, while Spens et al. (2017) showed that mixed cellulose ester membrane filters (CN, 0.45 μm) and sterivex-GP capsule filters (polyethersulfone, 0.2 μm) yielded higher amount of eDNA compared with glass fibre (GF, 0.6 μm) and polycarbonate track etched filters (PCTE, 0.2 μm), in freshwater samples. Djurhuus et al. (2017) found nitrocellulose (NC, 0.2 μm) and polyvinylidene difluoride (PVDF, 0.2 μm) filters yielded more DNA than polycarbonate track etched (PCTE, 0.2 μm) and glass microfiber (GFF, 0.7 μm) filters in marine environments without the material of any of the 0.2 μm filters affecting the estimates of species richness and community composition.

Collecting field replicate samples, in each sampling site, is a way to increase the volume of water sampled and thus the detection probabilities of taxa. The number of field replicates to be collected depends on whether the system is lotic or lentic, the study objective i.e., the level of confidence needed to the detection of taxa, the budget, and the target species (Hobbs et al. 2017). In any case, when spatial heterogeneity of eDNA is expected (i.e., different flow velocity microhabitats in lotic systems), it is important to collect eDNA samples from various spots in a sampling site, either pooled or not, to better represent the biodiversity of the aquatic communities.

Another finding of this review is that the preservation of eDNA samples, before extraction, is usually done by freezing (-20oC or -80oC) (Carraro et al. 2020; Uchida et al. 2020; Persaud et al. 2021; Reinholdt Jensen et al. 2021; Keck et al. 2022) or by the addition of a preservative (ethanol or lysis buffer) to the filters followed by freezing (Bagley et al. 2019; Brantschen et al. 2021; Gleason et al. 2021; Hupało et al. 2022). To better prevent eDNA degradation, immediate addition of a preservative (molecular grade ethanol or Longmire’s buffer) is recommended by Spens et al. (2017) compared to freezing or extraction of eDNA within 5 h after filtration. Moreover, closed type filters i.e., Sterivex capsules, were found to have higher total eDNA yields and retain eDNA integrity at room temperature for at least 2 weeks (Spens et al. 2017). These attributes along with the reduced risk of contamination, make filter capsules preferable for long sampling campaigns in remote environments.

eDNA extraction, amplification and sequencing

Previous studies evaluating how different eDNA capture and extraction methods affect biodiversity results of aquatic communities found a strong effect of the choice of methods on biodiversity detection (Deiner at al. 2015; Spens et al. 2017). The current review found that for eDNA extraction, many studies have used the DNeasy Blood and Tissue and PowerWater DNA Isolation Kits (Fig. 4). In rivers, eDNA capture by filtration combined with phenol–chloroform–isoamyl (PCI) extraction followed by ethanol precipitation yielded the highest amount of eDNA compared to other protocol combinations that had been tested (Deiner et al. 2015). Moreover, the combination of filtration with DNeasy Blood and Tissue Kit or PCI extraction has been shown to detect higher eukaryotes diversity (Deiner et al. 2015, Ruan et al. 2022). Renshaw et al. (2015) showed that PCI extraction protocol potentially yields more eDNA than DNeasy Blood and Tissue Kit when they used 0.45 μm CN filters for the capture of eDNA. Additionally, protocols where the DNA extraction is done directly from filter cartridges have been developed (Miya et al. 2016; Spens et al. 2017) and optimized (Wong et al. 2020) to efficiently extract high DNA yields from environmental samples while reducing the risk of contamination.

For eDNA amplification, most of the river macroinvertebrate eDNA metabarcoding studies utilize the standard DNA barcode marker for animals, mitochondrial cytochrome c oxidase subunit I (COI), which is highly efficient in discriminating vertebrate and invertebrate species (Fig. 5). However, a few studies, comparing different markers that amplify regions in the 18S, 16S and COI genes, identified that the performance of each marker varied, and recommend the use of multiple markers in freshwater macroinvertebrate samples to detect a wider range of taxa (Meyer et al. 2021; Ficetola et al. 2021; Martins et al. 2021). However, the COI marker has a notable reference library and in Europe, 64.5% of the freshwater macroinvertebrates used in ecological quality assessment is represented in BOLD database with at least 1 COI barcode and 41.8% with at least 5 COI barcodes (Weigand et al. 2019). Still, there are many unassigned reads in metabarcoding studies, and more effort should be put towards filling the reference libraries gaps.

Various primer pairs, targeting COI, are used for the amplification of macroinvertebrate eDNA with mlCOIlintF/jgHCO2198 and BF2/BR2 being the most popular (Fig. 5). It has been reported many times that the BF2/BR2 primer pair mostly amplifies non-target eDNA in river macroinvertebrate studies (Persaud et al. 2012; Macher et al. 2018; Zizka et al. 2019b; Leese et al. 2021; Gleason et al. 2021; Pereira-da-Conceicoa et al. 2021; Rivera et al. 2021) probably because it was designed to target a long (421 bp) fragment of the COI gene and thus it is difficult to amplify degraded extracellular DNA. On the other hand, the fwhF2/EPTDr2n primer pair that was designed by Leese et al. (2021) to optimize the detection of freshwater invertebrates in eDNA samples, achieved to yield on average > 99% of the reads from benthic invertebrate taxa. Furthermore, Brantschen et al. (2022) showed that the fwhF2/EPTDr2n primer pair can detect more indicator species belonging to the Ephemeroptera, Plecoptera and Trichoptera families compared to the mICOIintF/jgHCO2198 primer pair, resulting in a sufficient characterization of macroinvertebrate communities.

Furthermore, in metabarcoding studies, biodiversity was found to vary between PCR replicates (Deiner et al. 2015; Alberdi et al. 2018) but currently there are no set criteria for the minimum number of technical replicates necessary to optimize biodiversity detection. This review showed that studies use 3 PCR replicates on average to counteract stochastic effects of PCR amplification. Many more parameters of PCR amplification, like the number of PCR cycles and PCR inhibition that can affect the results of eDNA studies are explored in the bibliography (Kelly et al. 2019; Zizka et al. 2019a; Doi et al. 2021), while other studies propose the replacement of PCR amplification with bait capture enrichment as a way to overcome primer bias (Dowle et al. 2016; Gauthier et al. 2020). The use of positive amplification controls that is currently not widely adopted (Fig. 6) is important to detect possible PCR inhibition and the use of negative controls in every step of the metabarcoding process is also important to check for sample contamination and ensure the reliability of the metabarcoding results.

On the other hand, Smith and Peay (2014) suggest that PCR replication is not as important as reading depth in detecting diversity and show that the number of PCR replicates, when pooled together, does not affect a- and b-diversity metrics, whilst pseudo-b-diversity between sequencing replicates decreases with an increasing number of reads per sample. Other studies have also shown that the observed diversity increased with increasing sequencing depth (Alberdi et al. 2018; Shirazi et al. 2021) but at a large number of reads it reaches a plateau (Altermatt et al. 2023). As the technology of HTS advances and the efficiency of sequencing platforms increases, greater biodiversity can be uncovered as shown by Singer et al. (2019) who exhibited that Illumina NovaSeq can detect more DNA sequence diversity than the Illumina MiSeq in a high diversity ocean environment. Since the cost of these platforms is high, a cost-efficient alternative is Oxford Nanopore MinION sequencer that has been used to successfully identify aquatic invertebrates of a mock community (Baloğlu et al. 2021). However, previously reported high raw read error rates hinder the MinION’s use in metabarcoding (Krehenwinkel et al. 2019). Currently, an updated enzyme combined with the new R10.4.1 nanopore provides ≥ 99% raw read accuracy which is yet to be tested in freshwater invertebrate eDNA studies.

The last and crucial step in the metabarcoding workflow is the bioinformatic analysis of the high throughput sequencing output. Many bioinformatic pipelines like QIIME (Caporaso et al. 2010), OBITools (Boyer et al. 2016), JAMP (https://github.com/VascoElbrecht/JAMP) and APSCALE (Buchner et al. 2022) are used for quality control of the reads, assembling paired reads, demultiplexing, quality filtering, denoising and taxonomic assignment. However, caution is needed in the parametrization of each step because as showcased by Majaneva et al. (2015) different bioinformatic strategies can lead to different results when interpreting diversity and taxonomic composition of eukaryotic communities.

Conclusions

By reviewing the relevant literature, a great effort has been made the last fifteen years towards the advance of molecular methodologies concerning river benthic macroinvertebrates. It has been shown by many studies that eDNA metabarcoding can complement conventional water quality assessment methods by providing increased taxonomic resolution and revealing a higher amount of river benthic macroinvertebrate diversity. Importantly, our review emphasizes the need for further research and continued efforts to develop robust and standardized methodologies depending on the objectives of each study (i.e., broad scale biodiversity, water quality biomonitoring, monitoring of invasive species etc.). First of all, experimental studies should be conducted to compare eDNA of benthic macroinvertebrates from river water to eDNA from sediment and biofilm and test the efficiency of each sampling strategy, since the predominant method of capturing eDNA of benthic macroinvertebrates is through water filtration and little is known about eDNA from other sources. Secondly, this review has identified 15 different primer pairs that were used more than once in benthic macroinvertebrate eDNA studies and underlines the need for further testing of different primer pairs that will help optimize the taxonomic resolution of benthic macroinvertebrates and increase the robustness of the method. Thirdly, the exponential increase of eDNA metabarcoding studies and consequently the vast number of datasets that will be produced call for some standardization in reporting. It is crucial at this early point to follow the FAIR (Findable, Accessible, Interoperable, Reusable) data principles (Wilkinson et al. 2016) in order to maximize the impact that eDNA data can have on protecting the natural environment (Berry et al. 2021). Finally, as it seems impossible and impractical to standardize all aspects concerning metabarcoding protocols, calibration experiments, should be conducted to identify the key steps and the factors that introduce biases and biological variation (Blackman et al. 2019; Zaiko et al. 2021). The maximum possible standardization and harmonization of sampling and laboratory framework for macroinvertebrate eDNA metabarcoding studies would constitute the cornerstone towards the implementation of molecular techniques in river monitoring programs.