Methods |
Corresponding author: Kathleen E. Kyle ( katkyle914@gmail.com ) Academic editor: Sven Bacher
© 2024 Kathleen E. Kyle, Michael C. Allen, Nathan W. Siegert, Jason Grabosky, Julie L. Lockwood.
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:
Kyle KE, Allen MC, Siegert NW, Grabosky J, Lockwood JL (2024) Design of an eDNA sampling method for detection of an endophagous forest pest. NeoBiota 95: 149-164. https://doi.org/10.3897/neobiota.95.118267
|
Invasive wood-boring insects are a major economic and ecological concern worldwide as they impact native woody plant populations. These pest species are increasing in prevalence, with devastating impact, as global trade leads to higher rates of introduction and establishment. The emerald ash borer (Agrilus planipennis; EAB) is one such species, which has caused widespread damage across much of the United States and is now spreading across Europe. Non-indigenous woodborers such as EAB are difficult to detect at early stages of invasion, which is when management and eradication efforts are most effective and cost efficient. Environmental DNA (eDNA) surveys have demonstrated power in detecting invasive species when rare in the landscape due to their ability to detect trace amounts of DNA and identify to species. Here, we trialled a novel eDNA method for collecting environmental samples within host trees where invasive pest larvae are feeding, using EAB as a case study. We extracted tree cores approximately 1 cm in length using an increment hammer to assess detectability of eDNA from larvae feeding under the bark. In trees visibly infested with EAB, we observed a seasonal peak in EAB DNA detection probability (~ 64%; towards the end of the growing season), indicating a potential impact of ash tree phenology or EAB phenology on detection. When we trialled the method in a site with ash trees of low or uncertain EAB abundance, we did not record positive EAB eDNA detections. This outcome may have resulted from differing EAB phenology at the northern latitude of this survey site or because larval galleries were less numerous causing EAB DNA to be scarcer within the tree. Our results, however, provide preliminary evidence that increment hammer tree cores can be used to detect eDNA of EAB and, perhaps, other wood-boring pests. Further work is needed to clarify false negative survey detections at ash trees showing little to no signs or symptoms of infestation, as well as investigating the deposition, transport and persistence dynamics of EAB eDNA within trees.
Agrilus planipennis, forest management, early detection, emerald ash borer, environmental DNA, wood-boring insects
Phloem- and wood-boring insects are the most economically costly amongst invasive forest pests globally, resulting in over $2 billion in damage annually in the US alone (
EAB is the most economically costly invasive insect in the United States (
We posit that DNA from EAB larvae feeding within ash trees may be accessed via the collection and processing of 1 cm tree cores (Fig.
Schematic of our expectation that larval emerald ash borers (Agrilus planipennis; EAB) deposit sufficient DNA as they develop under the bark to be detectable in tree cores from within or nearby feeding galleries. EAB DNA deposited in feeding galleries may also become incorporated into the tree’s transport tissues surrounding the galleries and, if so, cores collected at some presently unknown distance from a gallery may correspondingly contain detectable EAB DNA. No matter the source, we expect the availability of EAB DNA from tree cores to be highly seasonal and peak either when larvae are most actively feeding or when the host tree moves large quantities of nutrients and water to the leaves or roots – shown here by downward arrows – in the transport tissues (beginning and end of the growing season, respectively) or a combination of the two. In the present study, core samples were extracted from sample trees at breast height (1.37 m) above ground level.
To test our hypotheses, we first had to design a qPCR assay that could detect trace amounts of DNA from an environmental source and accurately assign species identity to EAB and no other co-occurring species. This process involves, first, identifying a candidate species-specific amplicon and then measuring assay sensitivity and specificity (
We amplified part of the COI mitochondrial DNA region via polymerase chain reaction (PCR) using primers LepF1 and LepR2 (
All qPCR reactions consisted of 500 nM each primer, 250 nM probe, 1× TaqMan® Environmental Mastermix II with no UNG and 2 μl DNA. The optimised reaction protocol included an initial denaturing step of 96 °C for 10 min, followed by 45 cycles of denaturation at 96 °C for 15 s and annealing and extension at 60 °C for 1 min. All reactions were run on an Applied Biosystems StepOne Plus Real-Time PCR System (Applied Biosystems, Life Technologies, Carlsbad, CA). Each sample was tested for the presence of EAB eDNA in triplicate and considered positive if at least one of three technical replicates successfully amplified.
To investigate whether larval EAB DNA was present in tissues of ash trees, we took core samples (containing cambium, phloem and xylem) from green ash (Fraxinus pennsylvanica) throughout the 2021 growing season at two sites in New Brunswick, New Jersey, USA. We sampled 7–13 trees per day on 13 dates between 21 May and 15 October (Table
New Jersey core collection dates along with ordinal date and accumulated growing degree days and sample size for each sampling date.
Calendar Date (2021) | Ordinal Date | GDD Accumulation Base 50 °F (10 °C) | Sample Size (n) |
---|---|---|---|
21 May | 141 | 366 | 20 |
30 June | 181 | 1185 | 20 |
6 Aug | 218 | 2105 | 14 |
13 Aug | 225 | 2307 | 24 |
20 Aug | 232 | 2504 | 22 |
27 Aug | 239 | 2702 | 22 |
3 Sept | 246 | 2853 | 16 |
10 Sept | 253 | 2990 | 26 |
17 Sept | 260 | 3148 | 26 |
24 Sept | 267 | 3284 | 26 |
1 Oct | 274 | 3367 | 26 |
8 Oct | 281 | 3474 | 20 |
15 Oct | 288 | 3584 | 20 |
Core samples were taken at breast height (~ 1.37 m above the ground) using a Haglöf 2.5 cm (1-inch) increment hammer at ash trees ranging from ~ 15 to 90 cm diameter at breast height (dbh). All trees showed visible signs of EAB colonisation (e.g., canopy dieback, epicormic sprouting); however, no bark splits or adult exit holes were visibly evident within 1 m of locations where we extracted tree cores. We were not able to strip bark from our sample trees after core extractions as they were on private property and we also wished to resample the same trees. Thus, we cannot be sure how close core samples were to larval feeding galleries or pupation chambers. Before taking each sample, we flame-sterilised the cutting tube of the increment hammer using 100% EtOH to remove any surface DNA between sample collections. At each tree, we took two 1 cm core samples (one each from north and south aspects). We placed the core samples into tissue disruption tubes for later DNA extraction. In total, we took 282 cores from 21 ash trees. On each sampling date, we also extracted 1-cm cores (n = 2 or 4, depending on the number of sample sites) from nearby oak trees (Quercus spp.) expecting that these could serve as ‘field negative control’ cores where we would not expect EAB DNA to be present.
In the laboratory, we extracted DNA from core samples using the DNeasy Plant Pro Kit (Qiagen) following manufacturer’s protocols and tested all samples via qPCR with our EAB COI assay as described above. Each DNA extraction and qPCR run included negative controls to ensure no in-lab contamination occurred.
On 22–23 September 2021 (ordinal dates 265–266), we collected tree cores from white ash trees (Fraxinus americana) in Loudon, New Hampshire, USA to explore the ability of the method to detect EAB when present at lower densities. This EAB infestation was much less advanced than at our New Jersey sampling sites. Host trees at our New Hampshire sites exhibited a wide array of decline, characterised on a spectrum from no visible signs of infestation (‘no damage’) to minimal signs of infestation (‘light damage’) to a small degree of dieback and epicormic sprouting (‘moderate damage’). Samples were taken using the same tree core methods as described above. Accumulated growing degree days at this site during the two-day sampling effort were 2224 and 2245. We collected samples from a total of 30 trees, 10 each from three tree damage categories: no damage, light damage and moderate damage. We extracted four tree cores, one from each cardinal direction, from each tree to increase probability of EAB detection. In total, 120 cores were taken from ash trees. Negative control cores were also taken from nearby birch trees (Betula spp.) to confirm complete decontamination of the increment hammer between samples.
We fitted a Bayesian generalised additive model (GAM) with a Bernoulli error distribution and a logit link function to describe the phenology of EAB eDNA detection over the course of the 2021 sampling season in New Jersey, as well as to assess the effects of covariates on detection rates (GDD accumulation and side of tree). Additionally, due to potential contamination (i.e. amplification of small quantities of EAB DNA) found in some negative control samples (see Results), we also conducted a parallel and more conservative set of analyses that were identical to those described above, but that only treated samples with at least two of the three qPCR technical replicates amplifying to be true detections (see Appendix
We used the modelled detection probabilities from the GAM (i.e. the posterior distributions) to conduct an additional analysis to estimate how many core samples would be required to detect EAB in an infested tree with 95% certainty. To do this, we used the formula p* = 1-(1-p)n, where p* is the probability of obtaining at least one positive core sample, p is the modelled per-sample detection probability and n is the number of core samples taken. We set p* equal to 0.95 and solved for n to estimate the number of cores required for each sampling event to have a 95% probability of detecting EAB presence.
Based on the sequences we generated and those publicly available on GenBank, we designed primers EAB_COI-F (TTCGAGCAGAATTAGGAAATCCA) and EAB_COI-R (AAGCATGAGCAGTAACAATAACATTATAGA) and probe EAB_COI-Probe (CATTAATTGGCAATGACC), which target a 78 bp fragment within the COI mtDNA region of EAB. Our specificity testing indicated that our assay was specific to our target species (see Appendix
Of the 282 tree cores collected over the 2021 growing season, 120 tested positive for EAB DNA. Comparison of the GAMs explaining EAB eDNA detection probability indicated that the model including only the random effect of tree and the fixed effect of GDD was most parsimonious, as indicated by LOOIC comparison of models (Table
Comparison of Bayesian generalised additive models (GAMs) describing the phenology of emerald ash borer tree core eDNA detection probability, based on leave-one-out information criterion (LOOIC).
Model | ΔELPDa | ΔELPD SE | ELPD | LOOIC |
---|---|---|---|---|
Tree + GDD | 0.0 | 0.0 | -176.4 | 352.8 |
Tree + GDD + direction | -0.8 | 0.3 | -177.2 | 354.4 |
Tree | -13.2 | 5.8 | -189.6 | 379.2 |
Tree + direction | -14.4 | 5.8 | -190.8 | 381.6 |
Top: probability of detecting eDNA from emerald ash borer (Agrilus planipennis; EAB) using the tree core sampling method at sites in New Jersey, USA over time. Bottom: the number of ash tree core samples needed per tree to ensure 95% confidence in detection of EAB larval presence within a tree. Time is represented as growing degree day (GDD) accumulation, base 50 °F (10 °C). Black lines represent the posterior medians from a Bayesian general additive model (GAM); grey shading indicates 80% and 95% credible intervals. Black circles show the proportion of samples with detections on each sampling date calculated from the raw data.
None of our laboratory negative controls revealed contamination. However, three of our 30 ‘negative control’ Quercus tree cores returned weak positive EAB detections. These samples had cycle threshold (Ct) values of 39–41 and all had only one of three technical replicates amplify. The parallel analysis of detection probability that used stricter guidelines for declaring samples as ‘positive’ for EAB DNA revealed very similar phenological patterns (Appendix
Of the 120 core samples we took from ash trees at the lower EAB density site in New Hampshire, we recorded no positive qPCR detections of EAB. All negative controls also returned negative qPCR results. This sampling effort took place at ~ 2200 accumulated GDD, potentially missing the point when peak detection was witnessed during our high-density sampling season that took place at a more southern latitude.
Our results indicate that DNA deposited by EAB larvae can be recovered from ash tree cores using a standard increment hammer and confirmed via species-specific qPCR detection. We also show that there may be strong seasonality to this method of EAB detection. This seasonal pattern follows our hypothesis that, as they feed, EAB larvae deposit their DNA within galleries or it becomes incorporated within ash tree tissues. Further, the peak in seasonal eDNA detection probability of larvae was about three months later than the peak trap catch dates in conventional survey methods for capturing adults (
Phenology is an important consideration in any population assessment method (e.g.
The exact causes for the seasonality we observed in tree core eDNA detection rates are unknown. If EAB eDNA is indeed transported throughout the tree along with water and nutrients, the role of tree physiology becomes a key consideration in survey design. The ecophysiology of sap flow in trees has a long research history (e.g.
Nutrient flow within ash trees is restricted when larval EAB densities are high because larval galleries serve to interrupt phloem movement (
The extent to which the phenology of tree core eDNA detection does not overlap with the phenology of detection from conventional methods suggests an opportunity to use this new tool, once fully vetted, to extend the window of EAB detection for delimiting surveys or landscape-scale surveillance. Additionally, this combination of eDNA and conventional trapping programmes may apply to invasive wood-boring insects more generally, as most traps effectively catch the ephemeral flying adult stage of a target species, but are ineffective at detecting larval stages, which our eDNA tree core method could be well-suited to do. The eDNA tree core method could also complement other forms of EAB surveys besides traps, for example, by concentrating tree core eDNA sampling activity on or near girdled ‘trap trees’, which represent the best-known attractant for the species (
Our study introduces a promising and novel survey tool for detecting invasive wood-boring insects, along with proof-of-concept testing and insights into its performance. The key to realising if and when this method provides benefits to detecting EAB or other invasive wood-boring forest insects is: (1) executing an explicit test of the sampling effort necessary to detect EAB across different larval abundances; (2) a controlled repeat of our methods across a more inclusive array of EAB densities and latitudes to identify seasonal peaks and the degree to which these peaks diminish with lower EAB infestation levels; (3) identification of how much higher or lower detection probability of EAB is as a core sample is taken at varying distances away from larval feeding galleries and 4) a side by side comparison of landscape-level detection probability with the tree core method versus traditional trapping and detection techniques. It will only be possible for this tool to become operational after investigating these questions to better provide information about the power of the method as well as the potential shortcomings. Given the impacts of invasive wood-boring insects and the increasing number of invasions that are occurring globally (
We thank Patrick Tobin for his advice and input in comparing our methods to traditional methods of EAB detection. We thank Claire Rutledge at the Connecticut Agricultural Experiment Station for identification and specimens of co-occurring species for assay specificity testing. We thank Anne Nielson and Katherine (Tabby) Fenn for their advice and support as this project developed.
The authors have declared that no competing interests exist.
No ethical statement was reported.
This work was supported by the USDA National Institute of Food and Agriculture McIntire–Stennis Project Accession Number 1017685 through the New Jersey Agricultural Experiment Station, McIntire–Stennis Project NJ17335.
All authors have contributed equally.
Kathleen E. Kyle https://orcid.org/0000-0001-8830-4460
Michael C. Allen https://orcid.org/0000-0002-6632-4337
Jason Grabosky https://orcid.org/0000-0003-4226-2888
Julie L. Lockwood https://orcid.org/0000-0003-0177-449X
The protocol for this newly-designed eDNA sampling method has been published on our lab environmental DNA website (https://sites.rutgers.edu/edna/). All sequences we generated during this study will be publicly available on GenBank (Accession numbers PP373086–PP373114).
EAB COI qPCR assay design, specificity and sensitivity testing
Methods
For the design of our EAB COI assay, we generated 21 sequences from specimens collected in the species’ invasive range and combined with 35 sequences that were publicly available on GenBank. The GenBank accession numbers for those sequences are listed below:
AY756137, AY864194, DQ861319, DQ861320, GU013563, JF887747, KM845113, KT250461, KT250462, KT250467, KT250473, KT250476, KT250479, KT250480, KT250487, KT250490, KT250504, KT250508, MF286180, MG477998, MH159080, MH159105, MN548248–MN548260.
We conducted in silico specificity testing on 10 closely-related sympatric species to ensure none would cross-amplify with the assay we designed. We also performed a nucleotide blast in GenBank to ensure no published sequences had high percent identity with our target amplicon sequence. Finally, we conducted in vitro specificity testing on seven co-occurring Coleoptera with specimens trapped together at EAB invasion sites in Connecticut.
We evaluated our assay’s lower limits of detection (LOD) by creating an 8-level 10-fold dilution series using genomic DNA (gDNA) extractions from EAB specimen legs with attached muscle tissue. We carried out qPCR analysis in 20 μl reactions with 11 replicates of each concentration, which ranged from 1.6 ng to 1.6 fg as quantified using a Qubit Fluorometer (Invitrogen v. 2.0). We estimated the 95% limit of detection (LOD) of the assay by fitting a 3-parameter log-logistic dose–response curve to the resulting concentration and detection data following
Results
Our in silico specificity test evaluated sequences of 10 relatives co-occurring in the US northeast and showed evidence that there are sufficient polymorphisms in the primer and probe regions such that they would not cross-amplify with our designed assay (Appendix
Based on published genetic data in conjunction with species sequences we generated, we created a Geneious Prime sequence alignment to analyse how many base pair polymorphisms existed between each of 10 co-occurring Buprestid species in the primer and probe regions we designed for our EAB COI qPCR assay (highlighted in grey). The sequences listed are consensus sequences based on a combination of those we generated ourselves as well as those accessed on GenBank.
The modelled 95% LOD of our EAB qPCR assay (
We break down below the number and source of sequences used to generate these consensuses as follows:
Agrilus bilineatus: 1 sequence we generated + HQ582712, HQ582713, MF286166, MF805329, MH159018, MH159107
Agrilus arcuatus: MF286192, MF286193, MF805139, MF805156, MF805159, MF805179, MF805196, MF805245, MF805295
Chrysobothris femorata: 1 sequence we generated + JF888345, KR126263, KR481996
Agrilus quadriguttatus: 1 sequence we generated
Dicerca lurida: 2 sequences we generated + MG057907
Spectralia gracilipes: 1 sequence we generated + KM364375, KM847081
Agrilus subcinctus: 1 sequence we generated
Buprestis striata: KR482483
Brachys ovatus: HQ582477
Accounting for potential contamination
This plot shows the same model output as in Fig.