Introduction

Scientists unanimously recognize the widespread presence of multiple stressors on lake systems (Ormerord et al. 2010). An active area of aquatic science is quantification of the relative importance of the unique and synergistic impact that these stressors have on lake functioning and biodiversity. Long-term data are essential to address this issue (Smol 2010) and paleolimnology has a well-established reputation for providing valuable insights into the drivers of biological assemblages over long time scales (Gregory-Eaves and Beisner 2011). There is, however, still room for considerable expansion of the field of paleolimnology, as many taxa that do not leave diagnostic morphological features in sediments could leave a DNA signature (Fig. 1). In addition, DNA-based methods have the potential to drive advances in the study of intra-specific diversity and ecosystem functioning. Here, we provide an assessment of the state of DNA-based methods in paleolimnology and their applications to explore past lacustrine biodiversity. We also identify future research needs related to scientific challenges with the integration of DNA-based methods into paleolimnology. We intentionally limited the scope of this paper to lacustrine organisms and therefore do not present a full overview of all ancient DNA work.

Fig. 1
figure 1

Schematic of the lacustrine pelagic food web, highlighting how DNA analysis can expand the range of organisms that can be targeted in paleolimnology. The biological groups that could be targeted with DNA, but typically do not leave a morphological subfossil, and consequently are not used as proxies in paleolimnology, are highlighted with an orange frame. The type of DNA is specified as being intracellular DNA in cysts, propagules, resting stages, or extracellular DNA (extraC. DNA). (Color figure online)

Full size image

Potential of DNA tools in paleolimnology

DNA-based methods complement classical paleolimnology proxies and expand the range of organisms detectable in lake sediments

An increasing number of publications show that aquatic sediment archives act as DNA repositories, preserving genetic information about organisms that live (or once lived) in the sediments or the water column above (Boere et al. 2011; Coolen et al. 2013; Lejzerowicz et al. 2013). In some cases, the DNA is preserved within intact cells (often a resting stage) and in other cases the DNA is extracellular, but bound to organic or mineral matrices. The diversity of biological groups that are archived in the sediments over centuries to millennia and that can be detected from their DNA is considerable. For example, Capo et al. (2015) demonstrated that 71% of the micro-eukaryotic taxonomic units present (and detectable) in the water column over a ~2-year period were also identified by extracting and sequencing DNA from lake sediment cores that extended back in time ~100 years. Numerous studies have also shown that aquatic organisms can be identified from DNA extracted from lake sediments, both from recent deposits, i.e. deposits from the last centuries (Epp et al. 2010, 2015; Stoof-Leichsenring et al. 2012, 2014, 2015; Domaizon et al. 2013) and more ancient deposits, i.e. deposits older than 1000 years (Coolen et al. 2004, 2013; D’Andrea et al. 2006) and up to 270,000 years old (Randlett et al. 2014). The key advantages of DNA-based paleoecological approaches include: (1) an increase in the number of species that can potentially be studied at high taxonomic resolution, including those without diagnostic features preserved in the sediment record, such as picocyanobacteria (Domaizon et al. 2013; Hou et al. 2014), (2) the wide range of explorations of lacustrine biodiversity trajectories possible with this approach, which vary from the study of population and species succession over time (Epp et al. 2010), to host–parasite dynamics (Kyle et al. 2015a) and the study of breakpoints in lacustrine biodiversity (Capo et al. 2016), (3) the possibility to target specific genes that are indicative of key ecosystem functions like the DNA of methane-oxidizing bacteria as a proxy for methanotrophy (Belle et al. 2014, 2015). With careful consideration of the methodologies needed to work with ancient DNA (aka aDNA; Cooper and Poinar 2000; Hofreiter et al. 2001; Rizzi et al. 2012), there are many exciting new insights that can be gained from the study of sedimentary DNA.

DNA-based methods are uniquely positioned to answer critical questions in paleolimnology, some of which are not possible with classic paleoecological indicators

In an attempt to define a future vision for the field of paleoecology, Seddon et al. (2014) identified 50 priority research questions, which are generally applicable to paleolimnology. Many of these questions would benefit from the integration of DNA-based methods. For example, DNA approaches would be very useful for addressing the question “How can the paleoecological record be applied to understand the interactions between native and invasive species?” (Seddon et al. 2014), because of their potential to track taxonomic groups at a finer resolution than can be achieved using morphological fossils. To date, few paired comparisons of an organismal group, identified using morphological remains and genetic markers, have been conducted (for diatoms: Stoof-Leichsenring et al. 2012, 2015) and, in some cases, considerably more taxonomic units were identified within particular genera based on DNA approaches, but not necessarily overall (Stoof-Leichsenring et al. 2012).

On a more general level, the expansion of paleolimnology to include DNA-based approaches on a more routine basis has the potential to challenge existing limits within the field, as well as to further promote the integration of paleolimnology with other domains. For example, Seddon et al. (2014) asked “What common questions can be addressed by ecologists and paleoecologists to bridge the contrasting spatial and temporal scales between the two disciplines effectively?” This is an important question as there is still sometimes a gap between the fields of paleolimnology and other neoecological fields (e.g. community ecology, population ecology, etc.) (Rull 2014). This is not to say that the fields do not interact significantly, but rather that paleolimnological data are not always used to their full potential in other fields of aquatic ecology. The integration of DNA-based approaches into paleolimnology will continue to expand the field and bridge communication with molecular ecologists, enriching the questions addressed and approaches utilized by paleolimnologists. The opposite is also true—the paleolimnological record offers ecologists a unique temporal perspective to address both fundamental and applied questions.

Finally, an additional question highlighted by Seddon et al. (2014) was “How do paleoecologists encourage hypothesis testing rather than data-dredging approaches when exploring relationships between proxies and records?” The introduction of molecular data into paleolimnology dramatically increases the amount of data available. This likely means an initial increase in exploratory studies in paleolimnology because of the need to develop the tools and machinery to handle this amount of data in the paleolimnology realm. The potential for hypothesis-driven research may, however, grow as new and different data are available for generating hypotheses over long temporal scales. We envision that DNA paleolimnological approaches will not replace the more classical physical, geochemical and morphological indicators that are widely used (Smol 2008), but rather the molecular methods will enable investigators to identify mechanisms for the observed changes more fully (e.g. methane-oxidizing bacteria supporting chironomid diets; Belle et al. 2014).

Common questions regarding the analysis of sedimentary DNA

What type of DNA is archived in sediments?

DNA of lacustrine taxa may be archived in sediments in different forms, depending on its biological origin (Fig. 2). This variation affects the protection and preservation of DNA, and consequently the analytical strategy needed to analyze the DNA, e.g. extraction methods, length of fragments targeted, DNA markers targeted, etc. DNA archived in sediments can exist as either extracellular or intracellular DNA, i.e. DNA outside or inside structurally intact cells. Many microscopic aquatic organisms produce some form of resting stage, which affords them physical protection and allows them to withstand stressful conditions. Resting stages vary across organismal groups from the akinetes produced by some cyanobacteria, to the resting propagules formed by some protists (Härnström et al. 2011) and the ephippia produced by cladocerans (Morton et al. 2015; Piscia et al. 2014; Rogalski 2015). DNA may be extracted directly from these resting stages, or organisms can be resurrected and the DNA can then be extracted. In contrast, other aquatic organisms may have DNA that is less protected (e.g. found within easily lysed cell membranes) or may release their DNA directly into the environment, through the shedding of tissues, release of metabolic wastes, or via decomposition after death. Given these contrasting origins, one can target primarily the intracellular or extracellular DNA fraction. Aquatic sediments are generally characterized by relatively high DNA concentrations in comparison to concentrations in the water column (Eichmiller et al. 2014; Turner et al. 2015), and most of this is thought to be extracellular DNA as demonstrated in marine sediments (Corinaldesi et al. 2005). Extracellular DNA can be either free from matrix or adsorbed onto mineral and organic particles, which limits its degradation by nucleases (Dell’Anno et al. 2002; Corinaldesi et al. 2005, 2008). Dell’Anno et al. (2002) reported that, in marine sediments, the free extracellular DNA fraction represented <5% of the total extracellular DNA pool, and that the dominant extracellular DNA fraction has a long residence time in sediments.

Fig. 2
figure 2

Possible DNA analyses according to the goal of the study and the way DNA is archived. SNP single nucleotide polymorphism, DGGE denaturing gradient gel electrophoresis, T-RFLP terminal restriction fragment length polymorphism, Barcoding identification of species using a specific DNA region, Metabarcoding identification of multiple taxa from a complex environmental sample using a specific DNA region

Full size image

Despite the general dominance of extracellular DNA, it is generally hypothesized that taxa known to produce rigid cell walls, such as diatoms, dinophyceae, some chlorophyceae like Desmids, should be well represented in the sediment record. In contrast, groups like Cryptophyta, whose DNA is protected solely by a simple membrane, were identified as being underrepresented in recent sediments of a deep alpine lake, compared to its water column (Capo et al. 2015). There is, however, considerable variability in cell wall types, even within taxa that might be considered to have “protected” DNA (e.g. organic-walled dinoflagellates tend to be sensitive to chemical conditions; Zonneveld et al. 1997), which in turn likely affects DNA preservation in the sediments. As such, we caution that categorizing organismal groups as having protected or unprotected DNA, i.e. cyst versus noncyst-forming species, is rather simplistic. For example, Boere et al. (2011) confirmed previous observations from Coolen et al. (2006) and showed that DNA preservation for green sulfur bacteria, an organismal group without a protective resting stage, appeared to be more robust over time compared to DNA preservation of dinoflagellates (which produce cysts), based on the analysis of a sediment core from an Antarctic fjord.

The ability to separate soluble extracellular DNA pools from intracellular and particle-complexed DNA pools would help clarify (1) which type of genetic information is held within each of these DNA pools (Corinaldesi et al. 2005; Lever et al. 2015; Torti et al. 2015), and (2) if differential preservation of DNA exists between the primarily intracellular and extracellular fractions, which could in turn obscure the full complement of source organisms to these pools (Boere et al. 2011; Lejzerowicz et al. 2013; Torti et al. 2015).

Is DNA modified when aging in sediments?

For both extracellular DNA and intracellular DNA, aging in sediment means increased damage to DNA molecules. As a consequence, analyses become more difficult because of DNA strand breakage, abasic sites, miscoding lesions and DNA crosslinks (Lindahl 1993; Hofreiter et al. 2001). This results in fewer amplifiable templates, sequencing artefacts and preferential amplification of undamaged DNA, with a potential increase in false results or misidentifications if proper sequence authentication measures are not followed (Cooper and Poinar 2000). The degree of DNA fragmentation can vary, and in some cases rather long DNA fragments (>300 bp) were amplified from ancient deposits. Coolen et al. (2013) found ~500-bp-long fragments preserved in ~10-ka-old sediments. Loss of such long DNA fragments in lacustrine sediment, however, was reported even for sediments <1 ka old, whereas shorter DNA sequences were successfully amplified (Capo et al. 2016; Kyle et al. 2015b). For instance, in Norwegian lakes, Kyle et al. (2015b) showed that DNA fragments >350 bp could only be amplified if they came from the younger sediment layers, deposited recently (<20 years for cyanobacterial DNA), whereas short fragments (50 bp) were amplified in cores spanning 300 years.

How do lake characteristics affect DNA preservation?

There are now numerous studies in which DNA has been extracted successfully from lake sediments and inference of past biological dynamics has been generated from these natural archives (Table 1). Our list of study site examples varies from shallow polar lakes to deep temperate water bodies. Furthermore, our paper illustrates the variety of biological groups targeted and the diversity of issues addressed. From this pool of literature, we now know something about the extent to which DNA is preserved and the conditions that control the preservation of DNA in lacustrine sediments. Stable stratification, anoxic conditions and low and stable temperature appear to be among the most crucial factors that promote DNA preservation (Coolen et al. 2006; Corinaldesi et al. 2008, 2011), although other abiotic factors (e.g. mineralogical composition, organic matter load) and biotic factors (e.g. degradation via DNase activity) represent important constraints that influence the preservation of the DNA in sediments (Coolen and Overmann 1998; Coolen et al. 2004; Dell’Anno et al. 2005). Nonetheless, sedimentary ancient DNA is not restricted to anoxic aquatic systems, as successful results have been obtained from sediments deposited under fully oxic conditions, e.g. in marine systems (Lejzerowicz et al. 2013). Previous research revealed that adsorption of extracellular DNA on minerals is tight and stable once established (Dell’Anno et al. 2002), but several additional factors have proven to be influential, including the type of mineral, the proportion of clay, silt, and sand, porewater pH, as well as the valence and concentration of the bridging cations (Levy-Booth et al. 2007; Torti et al. 2015). Environmental measures such as pH and temperature are indeed known to influence the rate of depurination, i.e. hydrolysis responsible for the loss of the purine base (Lindahl and Nyberg 1972).

Table 1 Our review of the literature, focusing on the study of sedimentary DNA analyses in freshwater systems
Full size table

Deep hardwater lakes were found to have very well preserved cyanobacteria DNA, with long sequences archived and strong congruence observed between paleolimnological and historical records (Domaizon et al. 2013; Monchamp et al. 2016). In these hardwater lakes, calcite precipitation that occurs each year during spring and summer allows for the rapid sedimentation and burial of organic matter, including planktonic taxa from the water column (Dittrich et al. 2004), which in turn, are processes that likely promote DNA preservation. Likewise, organic matter preservation is likely to be favored in lakes with a concave morphology, as these conditions tend to result in direct particle sedimentation and limit sediment remobilisation and resuspension. Moreover, colder, deeper habitats and a lack of bioturbating organisms likely promote preservation of DNA that is buried in the sediments.

Despite these advances in understanding, no systematic field or laboratory study has quantified the degree to which DNA preservation is optimized in sediments, and thus this area is ripe for further investigation. A recent study of DNA from water provides interesting insights, showing evidence of slower decay rates at lower temperature and in waters with higher concentrations of dissolved organic carbon (Eichmiller et al. 2016). Additional insights could be gained from tapping into earlier studies that examined how different environmental conditions affected the preservation of classically studied morphological remains. We know, for example, that dinoflagellate cysts generally preserve less well under oxygen-rich conditions (Zonneveld et al. 1997), and by extension one could infer that DNA from dinoflagellates would also be poorly preserved under such conditions. Likewise, the silica frustules of diatoms are often degraded at high pH and it might be surmised that diatom DNA would not preserve well under these conditions.

We have learned a great deal over the past few decades with respect to taphonomy and the sediment record (Behrensmeyer et al. 2000), and new advances hold promise for exploring the influence of environmental factors that regulate the composition of the total sedimentary DNA assemblage (Vuillemin et al. 2015). We recommend comparative studies of DNA in the water column, sediment traps and sediment cores, to assess taphonomic biases. Study of sediment cores from the same site, but sampled in different years (Galman et al. 2008) will enable investigators to measure how the DNA signal is modified during burial.

Is there post-depositional transport of DNA between sediment strata?

Post-depositional transport of DNA within the sediments would complicate the interpretation of results from extracellular DNA analysis. Such DNA leaching in lake sediments is thought to be unlikely (Hofreiter et al. 2001; Anderson-Carpenter et al. 2011; Sjögren et al. 2016), but is a concern in terrestrial sequences (Haile et al. 2007). The most compelling evidence against DNA leaching in sediments comes from the numerous records that show temporal congruence between paired DNA and morphological (subfossil) or geochemical records (Coolen et al. 2006; Stoof-Leichsenring et al. 2012; Savichtcheva et al. 2014; Epp et al. 2015; Monchamp et al. 2016). We are, however, unable to rule out the possibility of DNA mobility in sediments based on our current state of knowledge and suggest that this is a fruitful area for experimental testing.

What precautions are necessary to ensure the reliability of results?

Several issues should be addressed to ensure that findings based on sedimentary DNA are robust. Some protocols are well established, such as the need to: (1) apply strict clean-lab procedures (Fulton 2012); (2) prevent contamination by recent DNA and replicate analyses (Hofreiter et al. 2001; Pääbo et al. 2004); (3) optimize procedures for fragmented ancient DNA (Hofreiter et al. 2001; Rizzi et al. 2012). Precautions that should be considered at each stage of analysis in a DNA paleolimnological project to optimize the reliability of results are summarized in Table 2. A common risk at each analytical step is potential contamination by ‘modern’ DNA, i.e. DNA from outside the sediments. Thomsen and Willerslev (2015) recently summarized the main pitfalls regarding eDNA studies, which are linked to contamination, inhibition, erroneous DNA sequences and reliability and completeness of reference DNA databases; this manuscript, however, did not focus on paleolimnological studies and we therefore provide some additional details here.

Table 2 Main analytical steps, risks and possible solutions for implementation of DNA analyses in paleolimnological studies. (Color table online)
Full size table

Sampling and handling of samples

When collecting a core, a recognizable genetic tracer can be used to test for coring and sampling contamination (Willerslev et al. 2003; Rawlence et al. 2014; Kirkpatrick et al. 2016). Alternatively, the outside 1–2 cm of sediment, which is potentially exposed to exogenous contamination during coring, can be removed; sub-samples can then be taken from the interior of the core. Regardless of sampling technique, an ideal sampling strategy should include taking paired samples (duplicate cores) to address intra-site variation.

Isolation of DNA from the core should be carried out in a dedicated DNA lab, physically separated from other molecular biology laboratories, especially because of the potential for contamination by PCR products (Knapp et al. 2012). Rigorous controls to prevent contamination include (1) use of separate, i.e. physically isolated labs for sampling, DNA extractions, and PCR preparations, (2) use of sterile disposable labware, (3) introduction of negative controls at all analytical steps, including sampling, extraction and amplification, (4) replication (at least duplicates) for extractions and sequencing of each sediment sample, in particular if inferences drawn from the DNA are not corroborated by other, independent data. Recommendations for the analysis of ancient DNA (Cooper and Poinar 2000; Gilbert et al. 2005) are applicable to work on DNA archived in sediments.

DNA extraction and molecular analyses

The type of extraction, with or without a cell lysis step, and the DNA region used to track the identity of biological taxa, might vary according to the targeted biological groups and the age of the sediments analyzed (Table 2). When metabarcoding is applied to infer a list of taxa from sediment DNA (Taberlet et al. 2012), the choice of barcodes to be sequenced is a key issue. Microbiologists have long used metabarcoding approaches, but the Consortium for the Barcode of Life (CBOL http://www.barcodeoflife.org/) is standardizing this method by defining the barcode markers to be used for each taxonomic group. To achieve high resolution at the species level, barcodes longer than 500 bp are generally needed (Hebert et al. 2003); however very short barcodes (<150 bp) are usually better suited to examine degraded DNA (Valentini et al. 2009). It is not always possible to find such short barcodes that enable precise taxonomic assignment (i.e. identification to the species level), and consequently identification at a higher taxonomic level (e.g. genus or family) must be considered sufficient for some metabarcoding studies.

Phylogenetic analyses are generally not possible with very short barcodes and the choice of a barcode is therefore a compromise between length (short for ancient/fragmented DNA) and the need for high taxonomic resolution and phylogenetic analyses. Several bioinformatics tools have been designed for finding optimal barcodes and primers dedicated to metabarcoding, and for testing their characteristics in silico (Ficetola et al. 2010; Riaz et al. 2011; Elbrecht and Leese 2015).

In cases where only highly degraded DNA is present in low quantities, many PCR cycles are required and false positive results are thus more likely with DNA barcoding. To limit and detect false positives, various quality assurance practices can be applied, including use of negative controls, adoption of a DNA sequence only if it was confirmed by at least two independent PCRs, and adjustment of replication level, e.g. increasing the total number of PCR replicates for each sample (Ficetola et al. 2015).

The level of PCR inhibition caused by co-extracted molecules, e.g. humic acid, also needs to be estimated to limit false negative results. As described by Savichtcheva et al. (2011), the level of potential inhibition can be verified using qPCR assays, performed with a series of increasing amounts of genomic DNA from a biological target such as an isolated strain, in the presence of increasing amounts of sediment DNA, potentially containing inhibitors. Comparison of the efficiencies of qPCR performed with the resulting mixtures can be used to estimate the level of inhibition; in this case, the approach applied was based on the assumption that inhibitors are diluted out when a log-linear relationship is achieved between Cq, i.e. the cycle threshold value, which is the number of cycles needed to reach a specified threshold value in signal strength, and the dilution factor (Lloyd et al. 2010).

In certain situations, checking the physiological state, i.e. active versus inactive, of the biological group that is targeted may be important. For example, Capo et al. (2015) compared molecular inventories obtained from DNA and from RNA, that is transcripts of the gene, indicating if the cells are active, to differentiate the active and inactive biosphere. This step was done to verify that the presence of active micro-eukaryotes in the sediments did not induce bias in the inventory of past eukaryotic diversity because active taxa may be over-represented. Such a verification step is even more important when studying bacteria of the past. Clearly, sediments represent an environment in which many bacteria and archaea can be active, even in relatively ancient sediments (Haglund et al. 2003). Therefore, RNA must be targeted along with DNA to differentiate between past bacterial communities and those living in the sediments.

Sequencing technologies and bioinformatics process

Once the targets are established, a decision must be made regarding the sequencing platform and bioinformatics approaches. High throughput sequencing (HTS) technologies applied for metabarcoding are rapidly evolving, and 454 pyrosequencing has largely been replaced by Illumina and Ion Torrent techniques that generate more DNA sequences at lower costs. Recent studies compared the main HTS platforms and showed that they perform equally well for HTS metabarcoding (Frey et al. 2014). It seems, however, that the bioinformatics treatment, as well as the library preparation, have larger effects on the molecular diversity inventories generated (Esling et al. 2015; Pylro et al. 2014).

With metabarcoding, each sample is multiplexed in one run of the analysis (each sample is tagged by a small DNA word called an MID or tag). Guidelines are to be observed to produce sets of tags, and standard multiplexing sample kits proposed by the manufacturers or programs, e.g. the OBITools package (http://www.grenoble.prabi.fr/trac/OBITools), can be used to design tags. Esling et al. (2015) recently proposed that researchers follow Latin Square Designs for multiplexing samples to optimize the detection of mistags, because amplicon library preparation is subject to pervasive sample sequence cross-contaminations as a result of tag-switching events (i.e. mis-tagging). Finally, there are recommended protocols for treating the resulting data for multiplexed reads, as reviewed by Coissac et al. (2012).

Before assigning a taxonomic identification to the obtained DNA sequences, a major issue is eliminating and or at least minimizing sequence errors that can have various origins, but mostly occur during PCR or sequencing. Numerous precautions should be applied all along the bioinformatics processing to limit false positives, which include applying stringent filtering and cleaning procedures to raw data. PCR-generated errors include formation of chimeras (Acinas et al. 2005) that can be detected by specific applications (Uchime; Edgar et al. 2011). In addition, several programs have been developed to remove sequencing and PCR errors (Coissac et al. 2012).

The artefacts generally represent low-frequency DNA sequences that may be discarded from further analyses (Reeder and Knight 2010). In addition, one could run replicate analyses (from extractions to PCR), and then minimize the presence of falsely detected rare DNA sequences that are a consequence of PCR bias (Lange et al. 2015) by retaining only DNA sequences that are common to the replicates.

Throughout the bioinformatics process, there are trade-offs between error trimming and keeping as much information as possible. Decisions depend on a variety of factors, including the gene target, molecular tools used, level of taxonomic resolution sought, and interest in phylogenetic analyses.

For taxonomic assignment, the comparison of obtained DNA sequences with a reference DNA database from known species can be performed using different methods, including analyses based on similarity or phylogeny, as well as different classification or coalescent-based algorithms. Several software programs designed specifically for metabarcoding can be applied to assign taxonomic names to sequences or to cluster sequences in operational taxonomic units (OTU) (Coissac et al. 2012).

Finally, completeness of the reference DNA sequence databases is critical to provide a relevant taxonomic list. As has been demonstrated in numerous publications, there is still a lack of references for numerous planktonic taxa found in environmental molecular inventories, and a strong need to continue to develop barcode reference libraries, especially for microbes (del Campo et al. 2014). For larger organisms, research programs such as the International Barcode of Life (IBOL; http://ibol.org/) have facilitated the standardization and reusability of barcodes. Open-access and curated reference barcoding databases are now becoming available for microbial indicators, as is the case for diatoms [R-Syst::diatom, linking DNA-barcodes to their taxonomic identifications (Rimet et al. 2016)] and other microbial eukaryotes [PR2 Protist Ribosomal Reference database for micro-eukaryotes (Guillou et al. 2013); PhytoRef database of the plastidial 16S rRNA gene of photosynthetic eukaryotes (Decelle et al. 2015)], and they offer valuable references for taxonomic assignment that complement other online public databases (e.g. NCBI https://www.ncbi.nlm.nih.gov/, Silva https://www.arb-silva.de/).

Applications of DNA-based analyses to advance lacustrine ecology

Intra-specific diversity and complex biological assemblages can be described from DNA preserved in sediments

To date, the sequencing of sedimentary DNA and quantitative PCR have been applied to a large spectrum of organisms, ranging from cyanobacteria to microbial eukaryotes, including phytoplankton taxa (Coolen et al. 2004; Stoof-Leichsenring et al. 2012; Domaizon et al. 2013; Savichtcheva et al. 2014; Pal et al. 2015; Monchamp et al. 2016), zooplankton groups [Copepods (Bisset et al. 2005); Rotifers (Epp et al. 2010); Cladocerans (Ishida et al. 2012)] and fish (Stager et al. 2015; Turner et al. 2015). The methodological approaches applied across these different studies also vary, as in some cases DNA is extracted from individual resting stages, e.g. studies of the DNA from cladoceran resting eggs (Mergeay et al. 2007), whereas in other cases DNA is extracted from bulk sediment samples to target a wide range of phylogenetic groups, e.g. studies of 100s of eukaryotic groups (Coolen et al. 2013). Likewise, studies vary in their focus, from interspecific to intraspecific diversity. There have been numerous intra-specific studies of cladocerans, as this field has been active for some time (Hairston et al. 1999; Pollard et al. 2003), but other organismal groups are also targeted now, such as diatoms (Stoof-Leichsenring et al. 2012), cyanobacteria (Domaizon et al. 2013; Monchamp et al. 2016) and complex assemblages composed by various heterotrophic and autotrophic taxa (Coolen et al. 2013; Capo et al. 2016). Overall, the knowledge gained from this work is diverse in scope, ranging from quantifying natural variability in population and community dynamics, to understanding how these variables respond to anthropogenic disturbances locally (at the watershed scale) or globally (e.g. climate change).

DNA from zooplankton resting stages

One of the most established applications of extracting DNA from sediments and sediment cores comes from the study of zooplankton diapause (resting) stages. Zooplankton resting stages come in many forms; bare diapause eggs in the case of copepods, diapause eggs encased in a protective ephippium in the case of cladocerans, and desiccated individuals or bare eggs in the case of rotifers (Hairston and Fox 2009). These stages are produced cyclically, as well as sporadically, in times of intense competition or stress (Cáceres 1998). Many lacustrine systems rely on the resting egg bank both as a source of new individuals each growing season, as well as a source of genetic diversity.

Kerfoot et al. (1999) and Hairston et al. (1999) were among the first studies to characterize cladoceran egg assemblages through resurrection ecology, whereby the organisms were revived from resting eggs and their DNA extracted to identify their population identity. More recently, DNA-based approaches for studying resting eggs in the paleoecological record have expanded such that the DNA is extracted directly from the resting eggs, thus avoiding the somewhat tricky work of resurrecting the organism. This work was pioneered by Limburg and Weider (2002). Mergeay et al. (2007) extracted DNA from Daphnia eggs to identify changes in population dynamics following a large water-level change in Lake Naivasha, Africa. In addition, genetic analyses can be useful to identify the species present within the broader cladoceran assemblage because there are only a few references on the taxonomy of cladoceran ephippia (Vandekerkhove et al. 2004). The combined use of subfossil morphological identification and sequenced 12S rRNA from Daphnia ephippia cases was shown to be quite powerful in a study of Daphnia dynamics in an alpine lake (Hamrová et al. 2010). In that case, the authors found that the presence of a long tail spine on ephippia is not a species-specific trait, and the genetic analysis provided greater taxonomic resolution. In particular, they found that the ephippia in question were from the species D. longispina, not D. lacustris as previously thought.

New methods are also expanding horizons in terms of the sample and organism type that can be targeted. For example, Ishida et al. (2012) described a method for extraction and amplification of DNA from empty cladoceran ephippia cases, in addition to those with viable diapause eggs from sediment cores. Success of these methods in subsequent applications has varied (Briski et al. 2011; Ishida, pers. commun.). If analytical methods can be improved, such that investigators can reliably extract and amplify DNA from just ephippia cases, then we have the potential to evaluate whether there are any genetic differences between individuals that hatched, and thus leave only a case in the sediment record, and those that did not hatch, and thus left an intact ephippium in the sediment record. Recent progress has also been made in sequencing copepod eggs from sediments (Xu et al. 2011a, b; Makino et al. 2013), which is particularly exciting as the carapaces of these organisms do not preserve well in lake sediments and the eggs are very difficult to identify morphologically.

Targeting specific functions in the ecosystem

Sedimentary DNA studies also offer the opportunity to target specific genes and genotypes, which are linked to specific functions in an ecosystem. For example, Belle et al. (2014, 2015) used ancient DNA from the bacterial methanotroph community to provide information about past methane oxidation in two freshwater lakes. By pairing the analysis of methane-oxidizing-bacteria DNA with stable carbon isotope analyses of both chironomid head capsules and bulk sediment organic matter from the same sediment intervals, Belle et al. (2014, 2015) showed convincingly that these bacteria were an important source of carbon for the chironomids and that the contribution of the methanogenesis pathway to benthic production varied considerably over time.

Similarly, the dynamics of toxic cyanobacteria (e.g. Planktothrix) have been examined by studying a portion of the gene that encodes for microcystins (McyA) (Savichtcheva et al. 2011, Monchamp et al. 2016). For example, Savichtcheva et al. (2011) used the McyA gene as well as a more general cyanobacterial gene to quantify total and microcystin-producing Planktothrix populations from sediment cores spanning the past 70 years in three lakes. Each of these study lakes has long-term water-column monitoring data (spanning > 20 years), and thus comparisons were drawn between the coherence of the sedimentary DNA and contemporaneous water-column data. Overall, Savichtcheva et al. (2011, 2014) demonstrated strong congruence among the time series from any single study lake, and the lake with the greatest amount of Planktothrix in the water column also had the greatest amount of McyA copies in the sediments. Using PCR detection of the McyA gene, Monchamp et al. (2016) recently confirmed the presence of potentially toxic cyanobacterial taxa throughout the last century in Lake Zurich (Switzerland). Further applications of this methodology to a wide range of sites will advance our understanding of the spatio-temporal dynamics of toxic and non-toxic cyanobacteria. When combined with other paleolimnological indicators, both genetic and classical tools, we can also gain greater insight into the impact of cyanobacteria on lake ecology.

Emerging topics and challenges in applying sedimentary DNA to the field of lacustrine ecology

DNA approaches to track the dynamics of invasive species

Given the intense interest in invasive aquatic species, this is an area that is ripe for paleolimnological study. To date, there are few examples of how classical paleolimnological methods have been applied to address questions related to invasive species, e.g. invasion by the cladoceran Bythotrephes of a Canadian lake (Hall and Yan 1997). Application of sedimentary DNA methods offers an opportunity to expand the list of invasive taxa that could be targeted, and the possibility to study the consequences of the invasion. Molecular techniques were employed successfully to detect other invasive organisms, including vertebrates, invertebrates, plants and pathogens (Jerde et al. 2011; Mahon et al. 2013; Goldberg et al. 2015), by analysis of water samples; consequently some public agencies are now implementing eDNA analyses as part of their routine surveillance of invasive species, e.g. asian carp in the USA and Canada (https://biosurvey.ku.edu/research/asian-carp-edna). Paleolimnological studies have the potential to further advance our understanding of invasion dynamics, as they can help answer the following questions: (1) when was the species introduced? and (2) which taxa were concurrent with this exotic species? Already, genetic analyses of sediment egg banks have proven to be a key source of information regarding the success of invasive Daphnia populations (Most et al. 2015). Ultimately, we think that performing genetic and classical paleolimnological studies across many sites in a landscape is a promising approach to investigate biogeographic patterns of invasive taxa through time.

DNA approaches to reveal the presence of fish populations or communities over time

Fish are often of great interest to society because they provide numerous ecosystem services and are key components of lacustrine food webs. As such, it would be helpful to be able to quantify past changes in fish populations or communities from analysis of lake sediments. This has proven to be challenging using classical methods alone. Morphological structures of fish such as scales and bones are rarely preserved, but see Davidson et al. (2003). The temporal dynamics of fish populations and their role as predators, however, have been studied indirectly by measuring the size of key prey taxa, e.g. Daphnia (Jeppesen et al. 2002) and by considering changes in zooplankton community structure (Palm et al. 2005; Palm and Svensson 2010). In some instances, migratory fishes may have distinct geochemical signatures and return in numbers sufficiently large that the fish themselves change a lake’s nutrient budget. The reconstruction of historic Sockeye Salmon population abundances from analyses of stable nitrogen isotopes in lake sediment was shown to track historical fish population records accurately, and this approach was complemented by other classical paleoindicators of lake trophic status, such as diatoms and cladocerans (Finney et al. 2000; Gregory-Eaves et al. 2009). The potential to expand this area of study was demonstrated recently by analysis of environmental DNA in both water (Takahara et al. 2012; Lacoursière-Roussel et al. 2016) and sediments (Eichmiller et al. 2014; Stager et al. 2015; Turner et al. 2015). Eichmiller et al. (2015) were able to track the presence and composition of fish assemblages from environmental samples. Furthermore, Eichmiller et al. (2014) reported that concentrations of fish eDNA in sediment samples were several orders of magnitude greater than in water on a per mass basis. Nevertheless, Eichmiller et al. (2014) reported low detection rates for fish. To date, only a few studies were able to detect fish DNA in old sediment records (Matisoo-Smith et al. 2008; Stager et al. 2015). To reconstruct past population dynamics of fish, qPCR must be applied to estimate the target species biomass and population dynamics. However, the relationship between DNA quantity in the aquatic environment and fish biomass can vary depending on the target species, the barcode used (Takahara et al. 2012; Lacoursière-Roussel et al. 2016) and the methods used for sample processing (Eichmiller et al. 2016). Therefore, the DNA quantity recovered from qPCR would need to be related to historical data or be calculated from experimental data to ensure the reliability of the results. Nonetheless, it is a promising and exciting area of study.

Inferring ecological networks from molecular inventories from complex assemblages

High throughput sequencing (HTS) approaches are becoming popular tools to estimate plankton diversity in both marine (Massana et al. 2015; de Vargas et al. 2015) and lacustrine samples (Nolte et al. 2010; Debroas et al. 2015; Kammerlander et al. 2015; Filker et al. 2016). In studies focusing on plankton, numerous authors have reported previously undescribed diversity in both eukaryotic and prokaryotic communities (Eiler et al. 2012; Mangot et al. 2013; Oikonomou et al. 2015). To date, most HTS inventories from lakes have focused on water samples, with only rare analysis of sediments. This body of work has revealed a high amount of eukaryotic diversity, including poorly studied biological groups, such as Apicomplexa, Chytridiomycota, and Cryptomycota. The taxa detected in these biodiversity inventories can then be assigned to different functional traits, e.g. phototrophs, parasites, phagotrophs, etc. This trait-based approach facilitates the inference of ecological dynamics from molecular inventories of the thousands of taxa/OTU detected (de Vargas et al. 2015), and may be further explored through network analyses. Recent analyses of HTS data from water-column samples provided evidence of the kinds of insights that can be gained through association networks, which can be indicative of food web structure (Chaffron et al. 2010; Chow et al. 2014; Kammerlander et al. 2015; Lima-Mendez et al. 2015). In particular, network analyses reveal co-occurrence patterns between taxa and environmental factors, identify taxa that have a central role in the network (important connectors of the network), and enable definition of emergent properties of the network, such as interconnectedness and average path length (Kara et al. 2013). To our knowledge, these types of analyses have not yet been applied to sedimentary DNA records, but methods are available to deal with time series data, and were recently reviewed by Faust et al. (2015). Application of statistical analyses to molecular inventories generated from sedimentary DNA records offers the possibility of gaining insights into long-term community processes, through alpha diversity (number of species and variation in their abundances at any one site or time period), beta diversity (differences in species composition among sites or periods) and network analysis, which are crucial for understanding the impacts of multiple stressors during the Anthropocene.

Coupling spatial and temporal approaches: temporal biogeography

Paleobiogeography is another important area of research, and the contribution of phylogenetic studies to the field of paleobiogeography has been significant over the last two decades (Lieberman 2003; Neubauer et al. 2016). Temporal approaches in biogeographic analyses, however, have generally been restricted to macro-organisms. As such, the description of planktonic biodiversity patterns has generally been performed without a temporal framework. Emergence of DNA-based methods, to study global lacustrine diversity from dated sediments, could help achieve key advances in plankton biogeography, as was done by Stoof-Leichsenring et al. (2015) for diatom lineages.

Use of molecular data for paleolimnological transfer functions

Transfer functions are an important tool in paleolimnology (Smol 2008; Smol and Stoermer 2010). In a transfer function, contemporary relationships between organisms preserved in surface sediments and environmental data (‘calibration sets’) are developed and applied to sediment records to infer past environmental conditions (Smol 2008). In paleolimnology, the most common organisms used for these models are diatoms because they are ubiquitous, well preserved in lake sediments, and their ecologies are well known. Seddon et al. (2014) also asked the question “When using modern analogues, what measures can be taken to be sure that the training set is sufficient to reconstruct the full range of likely past conditions, and if not, what else should be used to supplement these methods?” We suggest that DNA-based approaches could bring new information to the study of traditional paleoecological indicators in calibration sets. For instance, the addition of molecular inventories to morphological count data could provide a more complete view of past diatom assemblages, as genetic analyses have revealed inter- and intra-specific variation that is partially hidden behind single morphotypes (Stoof-Leichsenring et al. 2012).

At a more exploratory level, DNA-based approaches could enable investigators to consider new indicators or even a broader lacustrine community than the traditional approach of developing calibration sets that focus on a single indicator at a time. For instance, ciliates would be an interesting focal group, as their community composition is known to reflect the trophic state of the host water (Foissner and Berger 1996) and their DNA seems to be well preserved (Boere et al. 2011; Capo et al. 2016). A substantial amount of work will be required to validate key questions regarding the quality of preservation across different sites and how confidently one could apply the “space for time substitution” assumption that is foundational in transfer functions. To our knowledge, this research direction is not being actively pursued, but might be considered in the near future.

Integration of sedimentary DNA in biological collections and long-term environmental observations: the sediment core repository as a tool for conserving biodiversity

Because sediment cores have long been recognized as fundamental sources of environmental information, there has been a movement towards storing sediment cores in repositories and making them available to the scientific community. These repositories are incredibly valuable because new questions often arise and new techniques become available long after the sediment cores were collected, and can only be addressed by re-sampling such archived cores. Many existing sediment core repositories, however, do not routinely store sediments under ideal conditions for DNA analysis. Cores are typically stored at 4 °C in a refrigerator. Sedimentary DNA analyses could benefit from past coring expeditions, such as those funded by the International Continental Scientific Drilling Program (ICDP), but would require that sediments be stored under conditions more appropriate for DNA work (−20 °C).

Conclusions

Although early sedimentary DNA studies were initiated in the late 1990s, this field is experiencing rapid growth as new tools become available. This growth is facilitating study of a broader range of taxa in a more efficient manner. As with any paleolimnological variable, careful studies need to be conducted to evaluate preservation biases and appropriate lab methods need to be followed. The handful of calibration studies that have been performed to date, for which morphological or geochemical data have been compared with results of sedimentary DNA analyses, are very encouraging (Boere et al. 2011; Stoof-Leichsenring et al. 2012; Pal et al. 2015). There are many topics that are ripe for application of sedimentary DNA methods, including questions about invasive species, fish population dynamics and trophic interactions within lacustrine communities.