• Open Access

Rapid Invasive Species Detection by Combining Environmental DNA with Light Transmission Spectroscopy


  • Scott P. Egan,

    Corresponding author
    1. Department of Biological Sciences, University of Notre Dame, Notre Dame, IN, USA
    2. Advanced Diagnostics and Therapeutics Initiative, University of Notre Dame, Notre Dame, IN, USA
    3. Environmental Change Initiative, University of Notre Dame, Notre Dame, IN, USA
    • Correspondence

      Scott Egan, Department of Biological Sciences, University of Notre Dame, Notre Dame, IN 46556, USA. Tel: 615-618-6601; Fax: 574-631-7413. E-mail: scott.p.egan@nd.edu

    Search for more papers by this author
  • Matthew A. Barnes,

    1. Department of Biological Sciences, University of Notre Dame, Notre Dame, IN, USA
    Search for more papers by this author
  • Ching-Ting Hwang,

    1. Department of Biological Sciences, University of Notre Dame, Notre Dame, IN, USA
    2. Department of Physics, University of Notre Dame, Notre Dame, IN, USA
    Search for more papers by this author
  • Andrew R. Mahon,

    1. Department of Biological Sciences, University of Notre Dame, Notre Dame, IN, USA
    2. Department of Biology and Institute for Great Lakes Research, Central Michigan University, Mount Pleasant, MI, USA
    Search for more papers by this author
  • Jeffery L. Feder,

    1. Department of Biological Sciences, University of Notre Dame, Notre Dame, IN, USA
    2. Advanced Diagnostics and Therapeutics Initiative, University of Notre Dame, Notre Dame, IN, USA
    3. Environmental Change Initiative, University of Notre Dame, Notre Dame, IN, USA
    Search for more papers by this author
  • Steven T. Ruggiero,

    1. Environmental Change Initiative, University of Notre Dame, Notre Dame, IN, USA
    2. Department of Physics, University of Notre Dame, Notre Dame, IN, USA
    Search for more papers by this author
  • Carol E. Tanner,

    1. Environmental Change Initiative, University of Notre Dame, Notre Dame, IN, USA
    2. Department of Physics, University of Notre Dame, Notre Dame, IN, USA
    Search for more papers by this author
  • David M. Lodge

    1. Department of Biological Sciences, University of Notre Dame, Notre Dame, IN, USA
    2. Advanced Diagnostics and Therapeutics Initiative, University of Notre Dame, Notre Dame, IN, USA
    3. Environmental Change Initiative, University of Notre Dame, Notre Dame, IN, USA
    Search for more papers by this author

  • Editor Dr. Julie Lockwood


Invasive aquatic species introductions cause tremendous environmental and economic damage. Conservation and management efforts will benefit from rapid, inexpensive, and accurate on-site methods to detect harmful aquatic species to prevent their introduction and spread. Here, two technologies, environmental DNA (eDNA) sampling and Light Transmission Spectroscopy (LTS), were combined to address this need. Specifically, eDNA filtering and extraction methods were used to isolate DNA from: (1) lake water samples that were seeded with a microscopic fragment of five high-risk invasive species and (2) untreated samples from lakes infested with the invasive zebra mussel, Dreissena polymorpha, followed by polymerase chain reaction (PCR) amplification. LTS was then used to detect size shifts resulting from hybridization of PCR products with nanobeads covered with species-specific oligonucleotide probes. The results demonstrate that coupling eDNA sampling with LTS species detection can provide a sensitive and real-time solution for screening real-world water samples for invasive species.


Invasive species threaten biodiversity and the functioning of ecosystems and economies worldwide (Pejchar & Mooney 2009). In the U.S. alone, invasive species cause damage in excess of $120 billion annually (Pimentel et al. 2005). Perhaps nowhere have the harmful ecological and economic effects of aquatic invasive species been better documented than in the Great Lakes (Keller et al. 2009). Over 180 nonindigenous species have become established in the Great Lakes since European settlement, with about 70% arriving through the ballast water of transoceanic ships (Drake & Lodge 2007; Bailey et al. 2011). A subset of those species cause $138 million to $800 million in annual damages in the U.S. (Rothlisberger et al. 2012). With ever increasing global commerce, and in the absence of effective surveillance and management, the danger posed by invasive species introduced by a variety of vectors to aquatic ecosystems worldwide will only increase (Lodge et al. 2006; Keller et al. 2010).

Early detection is critical for conservation and management efforts aimed at mitigating the threat posed by aquatic invasive species. Rapid and sensitive on-site detection can: (1) prevent invasions through the interception of harmful taxa at points of entry before they become established and, once present, (2) help curtail their further spread through the identification and monitoring of affected areas. Here, we focus on the latter need by combining advances in environmental DNA (eDNA) sampling (Jerde et al. 2011) with a new Light Transmission Spectroscopy (LTS) DNA detection technique (Li et al. 2010, 2011; Mahon et al. 2013) to screen natural lake water samples for target invasive species (Figure S1; we hereafter use the abbreviation LTS to refer to both Laser and Light Transmission Spectroscopy, with the latter based on the same physical principles as the former but implemented as a transportable instrument with a modified optical package). LTS works by measuring shifts in the size of carboxylated nanobeads of a known size (typically 200 nm in diameter) that are functionalized with 18–26 bp long oligonucleotide tags encoding a species specific diagnostic sequence. Beads in solution increase in size when DNA from the target organism anneals to the oligonucleotide tags, which can be detected by LTS.

Traditional surveillance methods (e.g., nets, for aquatic taxa) usually have low capture probabilities, making them reliable only for species that are moderately to highly abundant (Jerde et al. 2011). Thus, traditional methods produce many false negative inferences (Gu & Swihart 2004). Genetic monitoring continues to grow in importance for a variety of conservation and management applications (Schwartz et al. 2006), and advancements in noninvasive genetic sampling (Beja-Pereira et al. 2009) represent particularly important strides in the management of rare species. For example, screening water samples for rare invasive species can be a needle in a haystack problem, but the eDNA approach reduces the severity of that challenge (Lodge et al. 2012).

Species surveillance using eDNA exploits an advantage of aquatic sampling that can aid in detection: the aqueous environment often contains microscopic bits of tissue in suspension (e.g., sloughed tissues or cells, larvae or adults that are microscopic, milt, eggs, extracellular DNA from degraded tissues, scales, and invertebrate exoskeletons). As a result, it is possible to sample DNA from even rare taxa that are present, but not detectable by traditional means. The eDNA method has been successfully developed and applied to Asian carp invasions (Jerde et al. 2011) and for other species surveillance questions (Amos et al. 1992; Ficetola et al. 2008; Dejean et al. 2011; Goldberg et al. 2011; Thomsen et al. 2012). Here, we use eDNA filtering and extraction methods to isolate DNA for PCR and LTS analysis to attempt to detect five invasive target organisms in two different types of water samples. The first sample type consisted of lake water collected from the field and seeded with microscopic tissue samples from five high-risk invasive species. The second type was water collected from two different lakes where the zebra mussel, D. polymorpha, was known to be present and absent, respectively, to test the effectiveness of LTS on unseeded samples. Our work expands upon previous work (Mahon et al. 2011; Mahon et al. 2013) by demonstrating that combining eDNA with LTS can provide a sensitive and real-time solution for screening real-world water samples for multiple high-risk invasive species. This approach will be useful for future research as well as conservation and management efforts involving detection of rare species such as incipient invasions or remnant populations of imperiled species. Our approach will also benefit efforts by the private sector or regulatory agencies to detect and minimize contamination of ballast water or live organism shipments with harmful species.


Target species, water sources, and eDNA sampling

For the experiment in which we seeded lake water samples with DNA from target organisms, water was collected from St. Joseph Lake, Notre Dame, IN (41° 42’ 21.55” N, 86° 14’ 20.65” W) in February 2012. The water samples were seeded with tissue from one of five different species that have either invaded or pose a high-risk of invading the Great Lakes and/or estuaries in North America via ships’ ballast waters: (1) Dreissena polymorpha (zebra mussel) and (2) D. bugensis (quagga mussel), which have viable populations in the Great Lakes basin, (3) Eriocheir sinensis (Chinese mitten crab), which has been reported in the Great Lakes and is likely able to establish (Herborg et al. 2007), but is not yet reproducing (Tepolt et al. 2007), (4) Limnoperna fortunei (golden mussel), a native to central and southeastern Asia, which has invaded multiple river drainages in South America (Pie et al. 2006), and is likely to establish and outcompete zebra and quagga mussels in the Great Lakes if introduced (Boltovskoy et al. 2009), and (5) Carcinus maenas (green crab), a marine crab that has often been economically and environmentally damaging where it has been introduced, but is not reported in the Great Lakes (Deagle et al. 2003; Patil et al. 2005). None of these five species have been detected in St. Joseph Lake.

To isolate eDNA from samples, 500 ml of lake water was filtered through a 20 micron polycarbonate membrane filter (GE Water & Process Technologies, Trevose, PA, USA). For the seeded samples, a microscopic fragment of tissue was removed from an adult specimen of a target species (previously preserved in 95% EtOH and frozen at –20°C) using fine-tipped dissecting forceps and added to the water sample prior to filtering (see Table 1 for tissue masses). Adding tissue prior to filtering simulated the normal sampling process of the eDNA approach. We generated four replicate seeded eDNA samples for each of the five target species from four different individuals. In addition, one control eDNA sample with no target tissue was analyzed in parallel with each species, resulting in a total of 25 eDNA samples for the seed experiment.

Table 1. Five target species analyzed in the study (four specimens of each taxon). Also given are the mass of tissue from the specimen used to seed a water sample, the concentration of eDNA extracted from the sample, and the concentration of the PCR amplification product generated using universal mtDNA invertebrate primers
Target high-risk speciesInd. IDTissue mass (μg)eDNA [C] (ng/μl)PCR [C] (ng/μl)
Limnoperna fortuneiL1128101.7118.6
(golden mussel)L214884.0117.9
Dreissena polymorphaZ166323.8124.0
(zebra mussel)Z2530117.5114.7
Dreissena bugensisQ132050.8118.2
(quagga mussel)Q242056.5115.6
Eriocheir sinensisM152896.1122.9
(Chinese mittenM21003.8122.2
Carcinus maenusG115675.3120.4
(green crab)G222285.2117.4

For the unseeded experiment, water was sampled in June 2012 from Eagle Lake, MI (41° 48’ 22.0314” N, 86° 1’ 31.2666” W), where D. polymorpha is known to occur (Perry et al. 1997) and in December 2011 from an unnamed lake at the USGS research station in Columbia, MO (38° 54’ 49.4742” N, 92° 16’ 22.3746” W), which has no zebra mussels. A total of six different 500 ml water samples were analyzed for each of the two lakes.

All water samples were put on ice immediately after collection and filtered with in 24–48 hours. All filters were frozen at –20°C and extracted within 2 weeks. We have found that these time windows provide high quality samples for screening (Jerde et al. 2011; Mahon et al. 2013).

eDNA extraction and PCR

To isolate and purify eDNA, the filter paper for each water sample was combined with 700 μl of CTAB Buffer and 20 μl of Proteinase K, vortexed for 15 seconds, and incubated at 63°C for 1 hour. After incubation, 700 μl of a 24:1 chloroform : isoamylalcohol solution was added and vortexed for 5 seconds, centrifuged at 14,000 RPM for 10 minutes, and the supernatant transferred to a new tube. Next, 1 ml of isopropanol was added, the sample inverted gently to mix, and incubated at –20°C for ≥4 hours. After this, the samples were centrifuged at 14,000 rpm for 20 minutes to form pellets that were washed twice with 500 μl 70% ethanol. Pellets were dried for 10 minutes at 45°C in a vacuum centrifuge and then resuspended in 50 μl of TE buffer and stored at 4°C. The DNA concentration of each extraction was measured on a Nanodrop 2,000 per the manufacturer's instructions (Thermo Scientific, Waltham, MA, USA) (Table 1).

Each eDNA sample was first screened using a species-specific set of primers to confirm the presence of the target species in each of the four seeded samples for the five high risk species. PCR was then performed on eDNA samples using universal invertebrate mtDNA primers that amplify a ∼600 bp fragment of the COI gene (Table 2). The total PCR reaction volume was 25 μl and included: 1 μl DNA template, 2.5 μl 10X buffer, 2.5 μl MgCl2, 0.5 μl universal invertebrate forward primer (LCO-1490), 0.5 μl universal invertebrate reverse primer (HCO-2198), 0.5 μl dNTPs, 0.15 μl Taq polymerase, and 17.35 μl ddH2O. PCR reaction conditions were: (1) 94°C for 1 minute, (2) 94°C for 30 seconds, (3) 41°C for 45 seconds, (4) 72°C for 1 minutes, (5) repeat steps 2–4 for 30 cycles, and (6) 72°C for 8 minutes (Folmer et al. 1994). To alleviate possible stochastic effects in PCR, we ran eight replicate 25 μl reactions per eDNA sample, pooling the amplification products in a total volume of 200 μl. A 10 μl subsample of the pooled product was electrophoretically run on a 1% agarose gel to confirm amplification.

Table 2. Universal mtDNA PCR primers and species-specific tags used in this study
Universal PCR primers
Universal invertebrate forward (LCO-1490)5′-GGTCAACAAATCATAAAGATATTG-3′Folmer et al. 1994
Universal invertebrate reverse (HCO-2198)5′-TAAACTTCAGGGTGACCAAAAAATCA-3′Folmer et al. 1994
Species-specific tag for LTS
Limnoperna fortunei (golden mussel)5′-TCCAACCAGTCCCTACTCCACCCTCT-3′Pie et al. 2006
Dreissena polymorpha (zebra mussel)5′-GAATCTGGTCACACCAATAGATGTGC-3′Mahon et al. 2011
Dreissena bugensis (quagga mussel)5′-TGTTCAACCCCCACCAAATCCGCCCT-3′Mahon et al. 2011
Eriocheir sinensis (Chinese mitten crab)5′-AGGTGGGTAGACAGTCCACCCAGTA-3′Mahon et al. 2011
Carcinus maenus (green crab)5′-GGCAAYGATAATAATAAAAGGAT-3′Senapati et al. 2009

Nanoparticle preparation and hybridization to PCR products

Polystyrene nanoparticle beads were functionalized with species-specific oligonucleotide tags as described in Li et al. (2011) and Mahon et al. (2013). Table 2 lists the species specific tags used in the study. To detect target species, double-stranded PCR amplified DNA was denatured by heating to 95°C for 2 minutes, then immediately chilled on ice for 2 minutes. After chilling, 10 μl of denatured PCR product was combined with 20 μl of functionalized bead solution and incubated at 48°C for 1 minute prior to LTS measurement.

LTS and nanoparticle measurement

LTS measures the size increase of functionalized nanoparticles in suspension when the oligonucleotide tags on their surface hybridize with the complementary strand of a target species in the PCR amplification product. LTS is based on measuring the wavelength-dependent light transmittance through a sample cell containing nanoparticles plus suspension fluid compared with that of a similar cell containing only the suspension fluid (see Li et al. 2010; Li et al. 2011 for details). For each sample, the diameter of the hybridized nanobeads was measured four times, twice at each of two different concentrations. The first concentration was ∼50 ng DNA/μl (range: 48–53 ng/μl) and was achieved by mixing 150 μl of pooled PCR product with 700 μl of ddH2O. Previous studies have shown this to be an effective concentration for LTS measurement (Mahon et al. 2013). The second concentration of ∼25 ng/μl was half of the first (75 μl of pooled PCR product with 775 μl of ddH2O). LTS measurements were performed for functionalized nanoparticles both prior to and following hybridization to PCR products.


Seeded water samples

The concentration of DNA extractions and PCR amplifications did not differ between seeded versus control samples from St. Joseph Lake (mean ± SE presented throughout; DNA extractions: seeded = 103 ± 23 ng/μl, n = 20; control = 116 ± 2.6 ng/μl, n = 5; tdf = 23 = 0.26, P = 0.80; PCR product: seeded = 118 ± 0.7 ng/μl, n = 20; control = 118 ± 1.0 ng/μl, n = 5; tdf = 23 = 0.07, P = 0.95; Table 1). These results imply that the samples from St. Joseph Lake contained eDNA from other organisms that was PCR amplified by the universal mtDNA primers along with the added DNA from the target species to similar total concentrations.

LTS successfully distinguished seeded versus control samples for all five target species. Figure 1 illustrates one individual LTS measurement for (A) the control with no target DNA present and (B–F) one seeded sample per target species. Figure 2 shows the mean peak shift across all replicates per species. The baseline diameter of functionalized nanobeads prior to hybridization with PCR products was 230 ± 0.4 nm. There was no significant shift observed from this baseline in the control lake water samples lacking target species tissue (229 ± 0.9 nm; tdf = 38 = 0.95, P = 0.35, n = 5) (Figures 1A and 2). Thus, the presence of PCR product from nontarget species in the control samples did not affect nanoparticle size, presumably because the PCR product from these organisms did not anneal with the species-specific tags on the nanobeads. In contrast, significant shifts averaging 146 ± 1.8 nm were observed for all 80 of the measured seeded samples (Figures 1B–F and 2; Table S1). These shifts were evident as secondary peaks of nanobeads that were distinct from the control peaks (∼229 nm); secondary peaks resulted from annealing between amplified target species sequences and the species-specific oligonucleotide tags.

Figure 1.

DNA-based invasive species detection using LTS. Each panel represents the density distribution generated by LTS for species detection. (A) LTS particle size distribution for nanobeads functionalized with the zebra mussel tag and hybridized with an eDNA extraction without the zebra mussel tissue added (control). The other five panels (B–F) represent one of the 16 measurements made on a specimen of each target species (see Methods). In each trial, there is a peak at ∼230 nm representing nanobeads that did not bind to the target and a second peak (between 300 and 450 nm) representing functionalized beads bound to the target DNA fragment (positive detection).

Figure 2.

Average shift (± standard deviation) in diameter between baseline beads attached with species specific oligonucleotide tags (beads + tag) and the seeded (+) and control (–) samples for each of the five target high-risk invasive species. All means presented are least-square means of the four replicates, which remove the significant, but small effect of concentration. All seeded and control samples differed significantly (P < 0.001) for all species (see Table S1 for details).

The size shift of nanobeads was not affected by the amount of target tissue initially added to the seeded samples, but there was a subtle (∼9 nm) increase in the mean size shift detected for LTS conducted at the higher 50 ng/μl DNA concentration (151 ± 2.4 nm, n = 40) compared to the lower 25 ng/μl concentrations (142 ± 2.4 nm, n = 40; tdf = 78 = –2.57, P = 0.01) consistent with previous studies (Mahon et al. 2013). As concentration was not the focus of the present study, all means presented throughout are least-square means for all four replicates per sample adjusted for the small effect of concentration. The size shift was not different among the four replicate individuals of a given target species (ANOVA: all P-values >> 0.05). However, the size shift did vary significantly among target species, ranging from 132 (±1.3) nm for Carcinus maenus up to 168 (±1.2) nm for E. sinensis (ANOVA: F4,75 = 39.9, P < 0.0001; Figure 2; Table S1).

Untreated water samples

Zebra mussel was detected in all six water samples from Eagle Lake, MI, which is known to be infested with zebra mussel (Figure 3). In contrast, no sample from the USGS research station lake without zebra mussel tested positive (Figure 3). In addition, as reported earlier for the seeded water experiment, zebra mussel was not detected in any of the control samples from St. Joseph Lake, IN, which is also believed to be without zebra mussel (Figures 1a and 2). Across all replicates, the mean size shift was greater for infested (Eagle Lake, 129 ± 1.3 nm, n = 24) than uninfested (0.5 ± 2.7, n = 24) lake water (ANOVA: F1,47 = 1358.0, P < 0.0001; Figure 3).

Figure 3.

DNA-based zebra mussel detection using LTS from naturally infested or uninfested lake water. One of twenty-four individual LTS measurements from (A) the USGS research station in Columbia, MO, where no D. polymorpha have been reported and (B) Eagle Lake, MI, where D. polymorpha have been documented for over 20 years (Perry et al. 1997). In panels (A) and (B), there is a peak at ∼230 nm representing functionalized beads and, in the Eagle Lake panel (B), a second peak at ∼340 nm representing functionalized beads that have bound to the target DNA fragment and represent a positive detection. In panel (C) average shift (±standard deviation) in the diameter of nanoparticles relative to the baseline beads attached with species specific oligonucleotide tags (beads + tag) for the infested (+ = Eagle Lake, MI) and uninfested (– = Columbia, MO) lake water samples. Means presented are least-square means. Infested and uninfested samples differed significantly (P < 0.001) in panel (c) (see Results for details).


Previous research suggested that the detection sensitivity of LTS was orders of magnitude greater than traditional surveillance methods (Mahon et al. 2013). Specifically, LTS successfully detected PCR amplified DNA isolated directly from target species tissue under laboratory conditions for samples containing: (1) only the target species, even at low DNA concentrations (Li et al. 2011); and (2) target species along with DNA from closely related species (Li et al. 2011; Mahon et al. 2013). Here, we present results demonstrating that a combined eDNA isolation and LTS approach (Figure S1) will work directly for real-world applications on field collected water samples from lakes typically containing large numbers of background species of plants, animals, microbes, etc. LTS was 100% accurate in all cases for: (1) seeded versus control natural water samples containing or lacking five high-risk invasive species; and (2) for all untreated samples directly surveyed from the field for presence or absence of zebra mussels. Thus, the eDNA-LTS approach delivered zero false positive and zero false negative rates for species detection.

Our findings highlight two areas for additional research to further optimize the application of LTS to species detection. First, while LTS was effective in all tests where target species tissue was added, there was variation in the peak shift among taxa. This may have been due to using universal mtDNA primers that PCR amplified the five different species to different final DNA concentrations in the eDNA samples. It may therefore be possible to design species specific PCR primers to improve the specificity and quantitative aspects of target taxa detection. Second, there was variation in the peak shift associated with the concentration of DNA measured. Further analysis is therefore warranted to determine the minimal threshold of DNA needed to positively confirm the presence of a target taxa, potentially reducing the number of PCR cycles needed for LTS, and possibly eliminating PCR altogether in some applications.

Species detection, including screening pathways for invasive propagules and monitoring habitats for newly introduced species (Lodge et al. 2006) as well as monitoring the distribution and abundance of threatened and endangered populations (Joseph et al. 2006), represents a fundamental tool for effective conservation and management. Without accurate, up-to-date knowledge of where rare species are, cost effective management is impossible. Genetic tools such as eDNA sampling and LTS have become increasingly prevalent in conservation and management efforts (Schwartz et al. 2007), and such tools have recently experienced rapid development in their application to monitoring biodiversity (Lodge et al. 2012; Thomsen et al. 2012). In a management context, there are at least two major advantages of eDNA methods over traditional methods: (i) increased sensitivity, that is, increased probability of detecting a species if it is present; and (ii) potentially lower costs, especially as eDNA methods increase in portability and automation. The overarching goal of our work is to enable accurate, sensitive, timely, and cost effective detection of target species.

We now have demonstrated the efficacy of eDNA sampling and LTS measurement as a laboratory-based detection method. Our ultimate objective is to apply the technology for rapid point of sampling detection in nature. To this end, LTS technology is portable, fast, and relatively inexpensive; we have recently developed a beta-model LTS device that together with a field-based PCR unit fit in a “carry-on” sized suitcase, run for a day from a rechargeable 12-volt battery, and following amplification can measure a sample in less than 10 seconds. The next major phase of research is therefore to conduct LTS testing on site and eventually to transfer the technology to personnel on the front line of invasive species detection in the field. In this regard, one extremely valuable application of LTS would be to rapidly screen the ballast tanks of transoceanic ships for invasive species. Currently, the most common management strategy, ballast water exchange, is of unknown effectiveness (Costello et al. 2007; cf. Bailey et al. 2011). Moreover, ballast water treatment technologies are still in the research and development phase (Tsolaki & Diamadopoulos 2010), and are many years from implementation for most ships worldwide. An urgent need therefore exists in the meantime for rapid detection methods like LTS to identify harmful invasive species in ballast prior to port entry to provide adequate time for implementing appropriate management actions.

In conclusion, our work implies that eDNA sampling and LTS could enable rapid species detection in the field in the context of research, voluntary or regulatory surveillance, and management actions to lower the risk of the introduction or spread of harmful species. Ballast water monitoring is one of many potential applications for LTS with ramifications for environmental protection, public health, and economic health. Although we have focused on aquatic habitats, the technology could be used to test for any organism in any habitat. Future research will expand analyses to include threatened or endangered species, pest species of agricultural concern, infectious species of health concern (e.g., pathogens and parasites), and species that pose biosecurity risks.


We would like to thank the Great Lakes Protection Fund, the U.S. Environmental Protection Agency, and the University of Notre Dame's Environmental Change Initiative and its Advanced Diagnostics & Therapeutics Initiative for providing funding for this research.