Methods |
Corresponding author: Alexander M. Weigand ( alexander.weigand@mnhn.lu ) Academic editor: Owen S. Wangensteen
© 2018 Alexander M. Weigand, Jan-Niklas Macher.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Weigand AM, Macher J-N (2018) A DNA metabarcoding protocol for hyporheic freshwater meiofauna: Evaluating highly degenerate COI primers and replication strategy. Metabarcoding and Metagenomics 2: e26869. https://doi.org/10.3897/mbmg.2.26869
|
The hyporheic zone, i.e. the ecotone between surface water and the groundwater, is a rarely studied freshwater ecosystem. Hyporheic taxa are often meiofaunal (<1 mm) in size and difficult to identify based on morphology. Metabarcoding approaches are promising for the study of these environments and taxa, but it is yet unclear if commonly applied metabarcoding primers and replication strategies can be used. In this study, we took sediment cores from two near natural upstream (NNU) and two ecologically improved downstream (EID) sites in the Boye catchment (Emscher River, Germany), metabarcoding their meiofaunal communities. We evaluated the usability of a commonly used, highly degenerate COI primer pair (BF2/BR2) and tested how sequencing three PCR replicates per sample and removing MOTUs present in only one out of three replicates impacts the inferred community composition. A total of 22,514 MOTUs were detected, of which only 263 were identified as Metazoa. Our results highlight the gaps in reference databases for meiofaunal taxa and the potential problems of using highly degenerate primers for studying samples containing a high number of non-metazoan taxa. Alpha diversity was higher in EID sites and showed higher community similarity when compared to NNU sites. Beta diversity analyses showed that removing MOTUs detected in only one out of three replicates per site greatly increased community similarity in samples. Sequencing three sample replicates and removing rare MOTUs is seen as a good compromise between retaining too many false-positives and introducing too many false-negatives. We conclude that metabarcoding hyporheic communities using highly degenerate COI primers can provide valuable first insights into the diversity of these ecosystems and highlight some potential application scenarios.
Hyporheos DNA Metabarcoding, Streambed Health Assessment, Biological Quality Element, Meiofaunal Enrichment
DNA barcoding has become a standard technique in ecological and biodiversity studies (
To date, the assessment of stream biodiversity and quality often relies on the study of presence, absence and abundance of diatoms, macrophytes, fish and macrozoobenthic invertebrates (MZB), i.e. invertebrates living on or in the riverbed (
So far, the hyporheic freshwater fauna has been scarcely studied mainly due to difficulties in sampling (mostly pumping and digging;
We here present and evaluate a DNA metabarcoding protocol for the assessment of hyporheic freshwater communities, targeting the meiofauna (i.e. organisms <1 mm but >42 µm) of streambed sediments (i.e. the hyporheic zone). Further, we test the performance of a highly degenerate primer set, amplifying a fragment of the mitochondrial cytochrome c oxidase subunit I (COI) gene for the assessment of hyporheic communities and investigate the effect of PCR replication strategy on the inferred biodiversity parameters and community composition.
Testing primer performance and replication strategy is of high importance, as even though it is widely accepted that the mitochondrial COI marker is the gene of choice for metabarcoding of freshwater invertebrates, no consensus has yet been reached regarding primer sets and the number of PCR replicates needed for reliable detection of taxa. Further, to the best of our knowledge, no studies have tested the performance of highly degenerate COI primers for metabarcoding hyporheic freshwater communities and replication strategy for studying these habitats has also not been tested.
Different primers have been used and evaluated for metabarcoding of freshwater invertebrates (
Different replication strategies for metabarcoding studies have been tested (
We here present and evaluate a DNA metabarcoding protocol for the assessment of meiofaunal hyporheic freshwater communities. Further, we test the performance of a highly degenerate primer set amplifying a fragment of the mitochondrial cytochrome c oxidase subunit I (COI) gene for the assessment of hyporheic communities and investigate the effect of PCR replication strategy on the inferred community composition. The here presented metabarcoding protocol is further discussed in the context of potential application scenarios.
Sediment was collected at five lotic freshwater sites in the Boye tributary of the Emscher River catchment (Ruhr Metropolitan Region, North Rhine-Westphalia, Germany), which is one of Europe’s largest river management and restoration projects (
To extract hyporheic meiofaunal organisms from the sediment samples, the sediment was sieved through 1 mm mesh size in the laboratory, using high-pressure water to flush sediment through sieves. Similar sieving methods have been tested before (
The following protocol for genomic DNA extraction from lotic freshwater sediments is available as a step-by-step laboratory handout (Suppl. material
We also tried to extract DNA according to the standard PowerMax protocol, i.e. without the additional steps described above. However, this resulted in minimal DNA yield and we thus developed the protocol described here.
The resuspended pellets (10 mL volume) were transferred into the provided BeadTubes and extraction was carried out as described in the kit manual, with a final elution volume of 5 mL. The eluted DNA was finally concentrated by another salt precipitation as described above and the DNA pellet was resuspended in 50 µL PCR grade water.
Extracted DNA was quantified on a Fragment Analyzer with the Standard Sensitivity Genomic Kit (Advanced Analytical, Oak Tree, USA) and a volume containing 15 ng of DNA was used for PCR. A two-step PCR protocol was followed: For the first PCR, untailed primers BF2 and BR2 (
All raw reads were processed with R scripts as in
MEGAN (v.6.8.18,
Metazoan species richness was determined using the R package vegan (
No DNA was observed in the negative controls processed together with samples and thus, no contamination was suspected. A total of 14,128,336 read pairs were obtained after sequencing, retaining 2,313,243 high quality reads after bioinformatic processing and quality filtering (Suppl. material
A total of 22,514 MOTUs were recovered after bioinformatic processing. Of these, 3,935 (17.5%) could be assigned to a taxonomic name on Kingdom level with the majority being identified as bacteria (2,806; 71.3%). After applying the taxonomic assignment procedure described above, a total of 263 metazoan MOTUs were retained across all five sampling sites (Suppl. material
Taxonomic composition of hyporheic community. Taxonomic composition of the hyporheic community revealed for the filtered dataset with the 2-out-of-3 replicate strategy applied, excluding the water-depleted site NNU_61.
Taxonomic resolution of hyporheic community. Taxonomic resolution of the 2-out-of-3 replicate dataset, excluding the water-depleted site NNU_61. Numbers on the x-axis indicate the number of MOTUs per taxonomic group.
Overview of taxonomic resolution for metazoan MOTUs. Number of Metazoa, Diptera and OligochaetaMOTUs identified per taxonomic level when only MOTUs present in at least two out of three replicates per sampling site were considered.
MOTUs | Kingdom level | Order level | Family level | Genus level | Species level |
Metazoa | 180 | 154 | 145 | 124 | 109 |
Oligochaeta | 57 | 53 | 47 | 43 | |
Diptera | 38 | 37 | 33 | 28 |
Alpha diversity of all metazoan taxa was highest in the ecologically improved downstream (EID) sites, with 150 MOTUs found in EID_57 and 134 MOTUs in EID_27 when analysing the full dataset, i.e. all MOTUs found at a sampling site, respectively, 99 metazoan MOTUs in EID_57 and 68 MOTUs in EID_27, when only MOTUsfound in two out of three replicates were analysed. Both near natural upstream (NNU) sites demonstrated a considerably lower number of MOTUs than the EID sites (full dataset: 82 for NNU_101, respectively, 65 for NNU_58; 2-out-of-3 dataset: both 44 MOTUs) (Figure
Overlap of MOTUs between PCR replicates and between sites for all metazoan taxa. Venn diagrams of metazoan MOTUs found per PCR replicate, number of MOTUs shared between replicates and number of MOTUs shared between sampling sites when only MOTUs present in at least two of three replicates per sampling site were considered.
Overlap of MOTUs between PCR replicates and between sites for Diptera and Oligochaeta. Venn diagrams of Diptera and OligochaetaMOTUs found per replicate, number of MOTUs shared between PCR replicates and number of MOTUs shared between sampling sites when only MOTUs present in at least two out of three replicates per sampling site were considered.
Overview of alpha diversity estimates. Alpha diversity of Metazoa, Oligochaeta and DipteraMOTUs for the complete dataset including all MOTUs per site, respectively, the dataset excluding MOTUs found in only one out of three PCR replicates.
Taxa | EID_57 | EID_27 | NNU_58 | NNU_101 |
---|---|---|---|---|
All metazoan MOTUs | 150 | 134 | 65 | 82 |
Metazoan MOTUs in 2 out of 3 replicates | 99 | 68 | 44 | 44 |
All OligochaetaMOTUs | 52 | 53 | 11 | 30 |
OligochaetaMOTUs in 2 out of 3 replicates | 34 | 35 | 7 | 14 |
All DipteraMOTUs | 25 | 24 | 19 | 16 |
DipteraMOTUs in 2 out of 3 replicates | 17 | 13 | 15 | 9 |
When all MOTUs found at a sampling site were analysed, Sørensen dissimilarity within sampling sites ranged from 0.15 (Diptera, site NNU_58) to 0.31 (Metazoa and Diptera, site 27). In contrast, when only MOTUs found in at least two of three replicates per site were included in the analyses, communities in replicates were much more similar, with Sørensen dissimilarity ranging from 0.02 (Oligochaeta, sites 27 and 57) to 0.11 (Oligochaeta, site 58 and Diptera, site 27) (all values: Table
Beta diversity (Sørensen dissimilarity) across all sampling sites was 0.60 (Metazoa), 0.52 (Oligochaeta) and 0.64 (Diptera) when MOTUs from all three replicates per site were analysed and increased when only MOTUs present in at least two replicates per site were analysed: 0.75 (Metazoa), 0.71 (Oligochaeta) and 0.77 (Diptera) (Table
Metazoan communities were found to be more similar between the two ecologically improved downstream (EID) sites (Sørensen dissimilarity: 0.60) than between the two near natural upstream (NNU) sites (Sørensen dissimilarity: 0.84). The NNU sites shared only seven MOTUs, all of which also occurred at EID sites: Limnodrilus hoffmeisteri lineage VII (MOTU_4), Micropsectra notescens (MOTU_10), two MOTUs from the Hydra vulgaris complex (MOTU_160, MOTU_2580), Homo sapiens (MOTU_186), a Chaetonotidae (MOTU_592) and an unidentified metazoan (MOTU_1959). In contrast, the EID sites shared 33 MOTUs primarily comprising Oligochaeta (18/33; 55%) and Diptera (5/33; 15%), with 17 of these exclusively occurring at EID sites.
Overview of beta diversity estimates. Sørensen dissimilarity between replicates within sampling sites for all Metazoa, Oligochaeta and DipteraMOTUs and Sørensen dissimilarity calculated across sampling sites.
All metazoan MOTUs | All 3 replicates | 2-out-of-3 replicates |
---|---|---|
Site | Sørensen dissimilarity | |
EID_27 | 0.31 | 0.09 |
EID_57 | 0.22 | 0.08 |
NNU_58 | 0.23 | 0.10 |
NNU_101 | 0.29 | 0.07 |
Across sampling sites | 0.60 | 0.75 |
Oligochaeta MOTUs | ||
EID_27 | 0.18 | 0.02 |
EID_57 | 0.17 | 0.02 |
NNU_58 | 0.26 | 0.11 |
NNU_101 | 0.29 | 0.05 |
Across sampling sites | 0.52 | 0.71 |
Diptera MOTUs | ||
EID_27 | 0.31 | 0.11 |
EID_57 | 0.21 | 0.09 |
NNU_58 | 0.15 | 0.07 |
NNU_101 | 0.28 | 0.08 |
Across sampling sites | 0.64 | 0.77 |
The applied metabarcoding approach detected taxa characteristic for the hyporheic community (Pacioglu 2010), with water mites, platyhelminthes and nematodes comprising prominent exceptions. However, these taxa were observed in preliminary morphological studies performed in the course of two B.Sc. theses at the same sites and using the same sampling technique, seven weeks prior to DNA metabarcoding sampling (
Several possible explanations exist for their non-detection/absence: first, the lack of detection could be attributed to the failure of the BF2/BR2 primer pair to effectively amplify some taxonomic groups, as has been found in previous studies (
In addition to the detection of meiofaunal organisms, many macrozoobenthic organisms and several taxa not present in the preliminary morphological dataset were detected in the metabarcoding dataset. The detection of some taxa can likely be attributed to free DNA, cells or small body fragments of these taxa being present in the sampled sediments, e.g. Homo sapiens, Gasterosteus aculeatus, Agelastica alni, Elodes minuta, Pediobius sauliusand Psocoptera. This corresponds to findings in previous studies showing that cells or DNA of metazoans can be extracted and amplified from soil samples (
The most apparent advantage of the developed DNA metabarcoding approach for hyporheic communities is the detection of a large number of species that might not be found when applying standard kick-net sampling and omitting size sorting of samples to prevent over-representation of taxa with a high biomass per specimen (
Further, the use of degenerate primers can help to detect and identify metazoan and non-metazoan MOTUs potentially suitable for biomonitoring and ecological studies. However, COI might not be the ideal marker gene for studying some metazoan (
Beta diversity and, as such, site heterogeneity were considerably high for Metazoa, but also when investigating Diptera and Oligochaeta only. In total, 70% of all identified metazoan taxa (126/180) were exclusively present at a single sampling site each, despite pooling nine sediment cores at each sampling site. Thereby, the finding that beta diversity estimates between sites increased when only MOTUs found in at least two out of three replicates were included in analyses could hint to a real biological phenomenon: taxa abundant at one sampling site might be rare at another site, therefore being removed by restrictive MOTU filtering. Another explanation would be the presence of taxa due to contamination or tag switching, which is a common phenomenon on Illumina sequencing platforms (
The findings that EID sites harboured more taxa than NNU sites and that communities in EID sites were more similar than communities in NNU sites could be explained by different phenomena. It might show that upstream sites are more isolated from each other, therefore harbouring more distinct metazoan communities. In addition, very small streams in upstream sites tend to dry out during times of low rainfall, therefore communities in these streams might be adapted to more extreme conditions and might more frequently face community fluctuations. In contrast, communities in downstream sites are expected to be more often connected, receive drifting organisms from different upstream regions and rarely become dry and hence, communities could be more constant and similar in EID sites. The finding that freshwater communities are more different in upstream and generally isolated sites has been reported before (
The results of our study highlight that replication strategy and removal of MOTUs present in only one PCR replicate can greatly impact the inferred diversity within and across sampling sites. Although the dataset analysed in this study is small and does not allow drawing definite conclusions, the results suggest that replication and removal of rare MOTUs found only in one out of three replicates could be beneficial for future studies and monitoring of hyporheic communities. As previously outlined in
Hyporheic communities of stream ecosystems are often highly diverse. At the same time, taxonomic keys for genus and/or species level identification are widely lacking (
Stream restoration management: The hyporheic zone is often rapidly colonised by meiofaunal organisms (
Streambed health assessment: Studies have shown that substrate characteristics are highly important for ecological processes in streams and freshwater communities are strongly influenced by substrate characteristics (
Improvement of existing indices: As hyporheic fauna composition largely depends on small-scale environmental factors such as streambed substratum, a high resolution for ecological health assessments of streams could be reached. Some existing indices use information on certain taxa to assess ecosystem quality, e.g. the Oligochaete Index of Sediment Bioindication (IOBS;
Assessment of intermittent stream and spring ecosystems: Standard assessment protocols for streams require the presence of water. However, intermittent rivers are common in many arid parts of the world and the hyporheic zone is an important refuge for aquatic taxa during droughts (
All those aspects are currently tackled in the two EU COST Actions ‘DNAqua-Net - Developing new genetic tools for bioassessment of aquatic ecosystems in Europe’ (
Highly degenerate COI primers are known to amplify a multitude of non-metazoan organisms, especially when applied to samples with a low ratio of metazoan to non-metazoan taxa (
Raw data is available from the Short Read Archive (SRA), accession number SRR7623732.
AMW and JNM designed the study, performed sampling, analysed the data and wrote the manuscript.
We thank the Flora-Immerschitt Foundation for financial support. Cristina Hartmann-Fatu is acknowledged for her help in the lab. This article is based upon work from the COST Action DNAqua-Net (CA15219), supported by the EU COST (European Cooperation in Science and Technology) program.
DNA extraction protocol
Figure S1: Overview of sediment characteristics
Figure S2: Taxonomic composition of hyporheic community with all metazoan MOTUs
Table S1: Sampling site coordinates
Table S2: Primer combinations
Table S3: Overview of raw reads and quality filtered reads
Table S4: MetazoaMOTU list with all replicates and including site 61
Table S5: MetazoaMOTU list with 2-out-of-3 replicate strategy and excluding site 61
Table S6: Taxa identified based on morphology