Assessing biodiversity of a freshwater benthic macroinvertebrate community through non-destructive environmental barcoding of DNA from preservative ethanol

Next-generation sequencing (NGS) technologies are rapidly increasing the applicability of genomics approaches in biological sciences. Aside from the sheer volume of sequence information obtained from NGS devices, the massively parallelized capacity of these machines is a significant advantage for the analysis of mixtures of DNA fragments. In PCR-based applications of NGS technologies amplicons are generated from target genes, often from multiple template genomes, and are sequenced in parallel without the need for separating target organisms or their DNA through sample sorting or cloning[1]. For example, microbial ecologists have taken advantage of NGS technologies in amplicon-based (marker genes) metagenomic studies such as the analysis of 16S rDNA for prokaryotic biodiversity assessments in various ecological settings[2]. Similar approaches have been developed for many situations where multi-template (environmental) samples are targets of investigations[3].

Environmental barcoding[4] seeks to gain sequence information for standardized species-specific DNA markers—DNA barcodes—directly from environmental samples such as soil, water, air, benthos or gut contents of various organisms[4]. This approach has the potential to dramatically increase the applicability of DNA barcoding in situations where rapid assessment of biodiversity at high resolution (e.g. species-level) is required at a wide spatiotemporal scale or in places where access to individual specimens is impossible or impractical. For example, environmental assessment through biomonitoring relies on biodiversity patterns of bioindicator (sentinel) species such as benthic macroinvertebrates, often at larval stage. However, due to difficulties in robust species-level identification of target groups, biomonitoring programs are faced with an identification bottleneck that can then lead to difficulties in implementing these programs[5]. We have recently demonstrated the potential of using NGS-based environmental barcoding in identifying species of fresh water benthos[4]. This approach has triggered a wholly new biomonitoring paradigm--Biomonitoring 2.0--for environmental assessment[6].

An important concern in the analysis of environmental samples is obtaining DNA templates from all target organisms in the mixed sample. Methods have been developed to extract and purify DNA from environmental samples such as soil or water. However, for bulk material such as benthos--obtained using kick nets--orpassively sampled arthropods collected in a Malaise trap, DNA extraction often requires homogenizing the biomass from all organisms and then performing a standard DNA extraction protocol on this homogenized slurry. Although this approach has been effective in gaining DNA from organisms in the mixture, it results in loss of all individual specimens, thereby rendering any subsequent analysis on these individuals impossible. Recently, we demonstrated that ethanol, commonly used as a preservative medium for storing specimens, contains DNA of stored organism and that this “free DNA” can be directly used for downstream amplification and sequencing without the need for conventional DNA extraction approaches[7]. Although we have shown the utility of this approach in individual specimens stored in ethanol, it is not known whether ethanol-based free DNA of many different taxa, with various biomasses, in a mixed environmental sample such as benthos, could be sequenced in an NGS workflow.

Another important concern in the use of NGS for analysis of environmental samples is the issue of bias in multi-template PCR amplification[8]. Current workflows often require PCR amplification of target templates from mixed samples, which can result in differential amplification of sequences from some species, leading to qualitative and quantitative biases in sequence representation from target organisms in the mixture. In other words, because of PCR bias, some species may not be amplified and sequenced while others may be amplified and sequenced in excess. This can obscure identity and abundance measures from bulk environmental samples. Although modified amplification regimes have been developed for offsetting the effect of PCR bias[8] and methods based on direct sequencing of DNA (without the need for PCR amplification) are on the horizon[9], PCR amplification bias remains an important issue. This problem is especially important when environmental samples are used for surveillance and monitoring applications where comparative analysis of biodiversity should be performed objectively and reproducibly[6].

Here we introduce an enhanced approach for environmental barcoding, which will aid sampling, DNA extraction and PCR steps in biodiversity analysis of benthic macroinvertebrate taxa commonly used for biomonitoring applications. We firstly incorporate a non-destructive sample preparation approach by using preservative media (ethanol) as the source of target DNA (hereafter referred to as ethanol-based DNA) and compare it with a conventional tissue-based DNA extraction (hereafter referred to as tissue-DNA). Secondly, to increase the recovery of species’ DNA barcode sequences in bulk environmental samples and to offset specific primer-binding biases, we introduce a multiplex PCR approach targeting multiple amplicons within the standard cytochrome c oxidase 1 (COI) DNA barcode region. We develop and test three wide-range primer sets for NGS analysis. To show the utility of this approach, we test it in parallel with Sanger sequencing individual specimens from a typical biomonitoring benthic sample containing several groups of macroinvertebrates.

Although NGS approaches have dramatically increased the capacity of genomics applications in biodiversity science, these applications have so far focused on discovering biota rather than monitoring their changes. As such, issues related to efficiency, repeatability and robustness have not been fully explored. A biomonitoring application based on NGS analysis of communities[6] requires robustness and reproducibility. This work was an attempt in demonstrating the utility of ethanol as a source of DNA in a multiplex PCR approach for NGS-based environmental barcoding.

Our previous work has demonstrated the utility of leaked or ethanol-based DNA from preservative ethanol for direct PCR amplification and subsequent Sanger sequencing of single specimens[7]. Here we extended this approach for a mixed community of benthic taxa collected using standard aquatic biomonitoring approaches and analyzed using an NGS workflow. Although this approach is somewhat similar to using environmental DNA (e-DNA), typical e-DNA does not involve collecting and storing organisms even as a mixture that we used here[10]. In fact, we take advantage of ethanol as a widely used preservative medium to access DNA from organisms nondestructively. Our approach will leave physical samples (i.e. individual larval samples) intact and accessible for subsequent molecular or morphological examinations, if needed. Additionally, this ethanol-based DNA does not require any DNA extraction procedure and can be directly amplified and sequenced, reducing the cost and time required for analysis.

The fact that ethanol used for accessing DNA contains a mixture of taxa with different sizes and abundances, does not seem to affect recovery of their sequences from ethanol-based DNA, as compared to tissue-DNA. We were able to recover many organisms that were only represented in small quantities in the mixture through the analysis of their DNA from ethanol. Missing taxa in both DNA sources were mainly from species with low abundance in our benthic sample and taxa with more than 1% were seldom missed. This pattern is strongly reflected in high correlation between the number of reads from two sources of DNA using different primer sets and leads us to deduce that primers bind to each species in the mixture with the same efficiency between two different sources of DNA. In other words, ethanol-based DNA seems to be a good replacement for DNA extracted from tissue mixtures, opening the door to a significant advantage for nondestructive analysis of benthic samples preserved in ethanol and can potentially be used in similar situations where a group of organisms are stored in a preservative media (see below).

Our results from both tissue extracted DNA and ethanol-based DNA suggest the presence of taxa not originally sequenced in our Sanger-based DNA barcode analysis. While two individual larval samples did not produce Sanger DNA barcodes in our analysis, 14 benthic taxa were detected that were not present as individuals in our mixture. Since our analysis was based on a single sampling event using all precautions to avoid possible contamination we can potentially link the presence of residual DNA to carry over through the organisms present in the larval community we sampled. This is in line with earlier observations[4].

Our study only included live field-sorted larval mixtures (a common sampling approach used for aquatic biomonitoring) and needs to be verified in other types of mixed samples such as Malaise traps (used for terrestrial arthropods), pitfall traps (used for soil arthropods) and different types of light traps (used for insects). A somewhat comparable approach for accessing extracellular DNA has been advocated for soil analysis, although it uses a phosphate buffer followed by a DNA extraction protocol[11].

Oligonucleotide primers are an integral factor in PCR-based NGS analysis, yet there has not been much attention to empirical optimization and testing of primers for specific genes commonly used in NGS analysis. Conversely, genes have mainly been selected based on availability of conserved binding sites[12]. It is, however, well known that differential primer binding can lead to quantitative and qualitative bias in taxa recovered from a mixed environmental sample even when primer-binding sites are conserved in a wide range of taxa[8]. Consequently, to offset the effect of biases in primer binding, in this study we used three newly developed primer sets for targeting benthic macroinvertebrate taxa commonly used in biomonitoring. Our results further show the effect of primers in taxa recovery but our combined approach does seem to alleviate PCR primer binding bias as we were able to recover up to 90% of all taxa and 100% of taxa with higher than 1% abundance, even though we used a relatively small sequencing throughput for this analysis (e.g. 1/16 of a 454FLX Titanium run for each DNA template from mixture). An increase in sequencing depth in a PCR regime using multiple primers may reduce or eliminate false negatives, especially for species with a smaller number of individuals or lower biomass in a sample. Although our study did not aim at quantitative analysis of taxa, we see high recovery (more sequence reads) from multiple primers in species that are represented by more individuals in the mixture (Figures2 and3). A similar pattern was observed in our previous study using only a single shorter mini-barcode fragment[4]. These early observations may provide the basis for developing a multiplex PCR regime for quantitative analysis of abundance as measured through biomass and considering gene copy number and other factors in different taxa. Measuring species abundance has been a key metric in traditional biomonitoring and ecological investigations but a recent study provides evidence that presence or absence alone can be used as a data source for a benthic response index used for biomonitoring[13].

The difficulty in large-scale spatiotemporal analysis of biodiversity has resulted in bottlenecks in executing biomonitoring programs and several other types of ecological and environmental investigations[6]. Next generation sequencing technologies coupled with high-resolution marker genes such as species-specific DNA barcodes--environmental barcoding--can address data rarity and low taxonomic resolution often crippling biodiversity analysis. Our study, additionally, allows nondestructive environmental barcoding analysis of benthic samples preserved in ethanol. If our observations are confirmed in a wider variety of biological specimens that are regularly collected in “ethanol jars”, this approach can pave the way for different applications, which require scanning the contents of these specimens for target taxa such as pests, pathogens, vectors, and rare or endangered organisms. The nondestructive nature of the ethanol-based analysis allows for subsequent thorough investigations to potentially separate and study these targeted taxa using different molecular or morphological techniques.

Approaches such as environmental barcoding are relatively new and require additional scrutiny to make them suitable for large-scale studies and real-world applications. Here we show the utility of ethanol-based DNA in simplifying sample preparation and alleviating destructive analysis of ethanol-preserved organisms. This procedure also allows preservation of specimens for further scrutiny. We also introduced a new multiplex PCR approach to neutralize specific primer-binding biases, thereby allowing the recovery of more taxa. We recognize that multiple samplings are required for comprehensive analysis of biota in an ecosystem through NGS. The availability of NGS devices with higher throughput and lower costs per analysis will facilitate generating sample sizes required for sound statistical analysis of NGS results[9]. We note, however, that our entire NGS data sets presented here were assembled using only two lanes of a 16 lane Roche 454 FLX Titanium run. Coupled with robust bioinformatics approaches, enhancements in NGS workflow will generate a framework for executing large-scale and realistic pilot projects in biomonitoring and related applications and provide insights for analyzing various bulk environmental samples in a similar manner.