Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Hydrogen Isotopes as a Sentinel of Biological Invasion by the Japanese Beetle, Popillia japonica (Newman)

  • Bruce A. Hungate ,

    Contributed equally to this work with: Bruce A. Hungate, Diana N. Kearns

    Affiliation Center for Ecosystem Science and Society, Colorado Plateau Stable Isotope Laboratory, and Department of Biological Sciences, Northern Arizona University, Flagstaff, AZ, 86011, United States of America

  • Diana N. Kearns ,

    Contributed equally to this work with: Bruce A. Hungate, Diana N. Kearns

    Affiliation Oregon Department of Agriculture, Plant Division, Salem, OR, 97301, United States of America

  • Kiona Ogle,

    Affiliations Center for Ecosystem Science and Society, Colorado Plateau Stable Isotope Laboratory, and Department of Biological Sciences, Northern Arizona University, Flagstaff, AZ, 86011, United States of America, Informatics and Computing Program, Northern Arizona University, Flagstaff, AZ, 86011, United States of America

  • Melanie Caron,

    Affiliation Center for Ecosystem Science and Society, Colorado Plateau Stable Isotope Laboratory, and Department of Biological Sciences, Northern Arizona University, Flagstaff, AZ, 86011, United States of America

  • Jane C. Marks,

    Affiliation Center for Ecosystem Science and Society, Colorado Plateau Stable Isotope Laboratory, and Department of Biological Sciences, Northern Arizona University, Flagstaff, AZ, 86011, United States of America

  • Helmuth W. Rogg

    Affiliation Oregon Department of Agriculture, Plant Division, Salem, OR, 97301, United States of America


Invasive species alter ecosystems, threaten native and endangered species, and have negative economic impacts. Knowing where invading individuals are from and when they arrive to a new site can guide management. Here, we evaluated how well the stable hydrogen isotope composition (δ2H) records the recent origin and time since arrival of specimens of the invasive Japanese beetle (Popillia japonica Newman) captured near the Portland International Airport (Oregon, U.S.A.). The δ2H of Japanese beetle specimens collected from sites across the contiguous U.S.A. reflected the δ2H of local precipitation, a relationship similar to that documented for other organisms, and one confirming the utility of δ2H as a geographic fingerprint. Within weeks after experimental relocation to a new isotopic environment, the δ2H of beetles changed linearly with time, demonstrating the potential for δ2H to also mark the timing of arrival to a new location. We used a hierarchical Bayesian model to estimate the recent geographical origin and timing of arrival of each specimen based on its δ2H value. The geographic resolution was broad, with values consistent with multiple regions of origin in the eastern U.S.A., slightly favoring the southeastern U.S.A. as the more likely source. Beetles trapped from 2007–2010 had arrived 30 or more days prior to trapping, whereas the median time since arrival declined to 3–7 days for beetles trapped from 2012–2014. This reduction in the time between arrival and trapping at the Portland International Airport supports the efficacy of trapping and spraying to prevent establishment. More generally, our analysis shows how stable isotopes can serve as sentinels of biological invasions, verifying the efficacy of control measures, or, alternatively, indicating when those measures show signs of failure.


In regions where a particular invasive species is not known to be established, capturing an individual of that species raises a key question: is it from an established population, or is it a new arrival [1,2]? The answer is important, because management actions can focus on documenting the extent of the invasion and eradication if the capture is known to come from an established population [1]. On the other hand, if an individual is a recent arrival from a source population elsewhere, monitoring for new arrivals and improved containment at the source populations may be more effective in reducing the probability that the invasion establishes. In other words, managing invasions improves when the status of the invasion is well understood [3,4]. Yet, for species known to threaten invasion, current techniques often do not make it possible to distinguish between new arrivals and members of establishing populations. The goal of this work was to test whether stable isotopes of hydrogen can serve as sentinels of biological invasion by the Japanese beetle (Popillia japonica Newman), distinguishing new arrivals from established residents.

The Japanese beetle was first found in the continental U.S.A. in southern New Jersey in 1916 [5], and it is now established throughout the eastern U.S.A. (Fig 1), threatening western states. This voracious beetle is damaging because it feeds on plant roots, foliage, and flowers, and is a generalist herbivore with over 300 plant species, including grasses used for turf [6], various crops [7], and horticultural plants [8]. Grubs (the immature form) overwinter until spring and then feed on roots until they pupate in the soil for two to three weeks. When they emerge as adults, they mate and continue to feed on plant foliage and flowers for up to 60 days, after which they lay eggs in the soil that hatch that same year; grubs continue to feed in the soil until fall when they overwinter [5]. Damage caused by Japanese beetles is likely to increase with rising CO2 and global warming because of increased foliage consumption rate [9], and they also exacerbate damage caused by other herbivores [10].

Fig 1. Map of continental US showing the status of the Japanese beetle invasion, by county: well established (red), establishing (orange), occurrences reported (yellow), and no occurrences known (white).

Black circles indicate sites of beetle collections used to develop the relationship between beetle and precipitation δ2H values, and the star indicates the study site at the Portland International Airport. Data compiled from refs 1519, as described in the Methods (Potential Sources of Origin).

Preventing expansion of this pest into the western U.S.A. is a high priority. Extensive measures are used in airports in the eastern U.S.A. to prevent individual beetles from entering cargo planes, including inspections and exclusion devices around doors, hatches, and other openings prior to departure, as well as pesticide applications at sites of capture [11]. Despite these measures, Japanese beetles have been caught at the Portland International Airport in Oregon since 1992, and every year since 2000. Each year, ground pesticide treatments that target the immature soil-dwelling beetles are applied in May. Foliar sprays targeting adult beetles were conducted in July and August 2002–2014. Other western states (Washington, California, Idaho, Utah, and Colorado) have also intercepted this quarantine pest at airports, in imported nursery stock, and in urban areas in recent years [1117]. The distinction between new arrivals or members of establishing populations becomes more difficult in areas around airports that receive many flights harboring potential invaders. Multiple individual captures could indicate strong propagule pressure, or alternatively, the growth of a nascent population, both of concern for invasive species, but with different management implications [18].

Variation in the isotope composition of animals reflects their natural history–what they eat, what they drink, and where they travel–providing a forensic tool for animal ecology [1921]. Hydrogen in animal tissue comes from ingested hydrogen, both as water and organic matter. Plant tissue hydrogen isotope composition (δ2H) reflects that of local precipitation, offset by isotope effects associated with photosynthesis and evaporation [22]. Thus, the hydrogen isotope signature in animal tissue should also reflect the isotopic composition of local precipitation [23,24], which displays broad geographic variation [25,26]. Across the U.S.A., regional variation in δ2H in precipitation is strong, with high values in the southeastern U.S.A., and decreasing systematically to the northwest.

Chitin and keratin, key substances in insect structure (in the formation of the exoskeleton and wings) are biosynthesized with the incorporation of hydrogen isotopes from diet and water in the environment [21]. Once chitin and keratin are synthesized, the majority of hydrogen atoms are non-exchangeable with ambient water vapor [27]. For this reason, hydrogen isotope ratios in chitinous and keratinous tissues have proven valuable for determining the natal origins of organisms [19,21,24,2832], including of invasive species [33].

Other insect tissues are metabolically active and there is capacity for turnover of isotope signatures in new environments [34]. Isotope signatures in these tissues could determine whether an individual is a recent arrival, reflecting the isotopic signature of its region of origin, or whether it represents an established population, isotopically equilibrated with the new environment [28,35]. This could also provide useful information about invasive species.

Here, we analyzed δ2H signatures of Japanese beetles of known origin across the U.S.A. as a foundation for examining the continental-scale variation in δ2H of Japanese beetles. We tested experimentally how beetle δ2H shifts over a 5-week period when exposed to a new isotopic environment, and incorporated the findings into a hierarchical Bayesian model that describes the timing and origins of Japanese beetles caught at the Portland International Airport. This work assesses the utility of δ2H signatures of invasive organisms as sentinels of the invasion status, and as a potential foundation for making decisions about control and eradication efforts.

Materials and Methods

Hydrogen Isotope Composition of Field-Collected Japanese Beetles

Voucher specimens of Japanese beetles collected from field traps were sent from Illinois, Iowa, Kentucky, and Ohio (n = 10 for each location), and from multiple sites in Colorado (n = 15), Utah (n = 5), and Idaho (n = 11), a total of 71 samples from known populations. Specimens from Iowa, Kentucky, Ohio, and Illinois were provided in fall, 2010 by MTA Horticultural Consultants in Middletown, KY. The Utah samples were sent in November 2012 by the Utah Department of Agriculture, and the Colorado samples in 2012 by the Colorado Department of Agriculture. The samples from Idaho were collected between July and August 2012, and were provided by the Idaho Department of Agriculture. Specimens were kept frozen at -10°C in glass vials to reduce isotopic exchange with ambient water vapor [36] and were shipped overnight to the Colorado Plateau Stable Isotope Laboratory (CPSIL,, where they were dried before analysis. No permits or approvals were required for the field sampling, because the Japanese beetle is an invasive organism in all of the sites sampled, not requiring permission for removal. To compare δ2H values of beetles with δ2H of precipitation, we used the on-line isotopes in precipitation calculator (OPIC, [26,37]) to estimate the annual mean and standard error of δ2H in precipitation for each site where beetle specimens were collected. For the hierarchical Bayesian modeling, monthly values for each site were also estimated from OPIC and error terms estimated as described below.

Japanese beetles were collected from the Portland International Airport (PDX) and in several sites within 10 miles of the airport from 2006–2014. The beetles were collected in Japanese beetle traps placed in the field, checked on average every 3–4 weeks in earlier years (2006–2011) and every 3 days by camera in later years (2012–2015). Beetles were removed as soon as they were detected in the traps. Traps consist of a metal funnel top that has crossed panels that lead down to a collection can on the bottom. The upper portion of the trap is baited with a scented lure, a Trece Pherocon dual lure for Japanese beetle, which has two components, a floral attractant and a pheromone attractant. Based on observations of traps equipped with cameras installed since 2012, beetles typically die in the traps within 24 hours, minimizing isotope exchange with the surrounding environment due to metabolism. Furthermore, there was no relationship between the number of days between setting the trap and finding a dead beetle inside (r = 0.16), indicating that time in the trap had little influence on the isotope value of the beetle. Specimens recovered from traps were placed in glass vials and stored at -100 C.

Isotope Turnover Experiment

Approximately 200 beetles were shipped overnight on 16 July 2013 from the Virginia Department of Agriculture (Madison, Virginia, U.S.A.) for an experiment designed to determine whether beetle δ2H changes after relocation to a region with distinctly different δ2H in precipitation and plant material, and, if a change occurs, at what rate. These beetles were housed in a quarantine facility starting 17 July 2013. The facility was established in Salem, Oregon, U.S.A. in Spring 2013; the facility met all requirements for containment by the U.S. Department of Agriculture Animal and Plant Health Inspection Service Plant Protection and Quarantine agency (USDA APHIS PPQ). This agency granted a permit to move live plant pests. Within the facility, beetles were placed in a terracosm with a screen top in the quarantine area on July 17, 2013, and were provided with local rose plants (δ2H = -121‰ for leaf tissue) and water from a local well (δ2H = -66‰) during a 5-week experiment. The plant δ2H value is consistent with that expected for local vegetation, though the water δ2H was lower than that of local precipitation (δ2H = -88‰, CI 6‰, from OPIC [37]). Thus, to the extent ingestion of water influences beetle δ2H, our experiment likely underestimated isotope turnover.

Ten beetles were immediately frozen to determine initial δ2H. Each week over a 5-week period, 10 beetles were harvested from the terracosm. Specimens were frozen immediately and then dried. Individual beetles collected in the first and last week’s samples were dissected to separate the elytra, membranous wings, and bodies, and each of these body parts was analyzed separately for δ2H. This tested the idea that elytra and wings, high in chitin and keratin concentration, are less prone to changes in δ2H in response to a change in the δ2H of water and food sources, because non-exchangeable hydrogen is thought to be stable in chitin and keratin once they are synthesized [27].

Isotope Analyses

All samples were analyzed for δ2H at CPSIL at Northern Arizona University. All field specimens were analyzed using the entire beetle; some samples from the isotope turnover experiment were divided into membranous wings, elytra, and bodies (as described above), fractions that were analyzed separately for δ2H. All samples were ground to a fine powder, and 350 μg of each was weighed into a 3.5 mm × 5 mm silver capsule and then analyzed by coupled pyrolysis gas isotope ratio mass spectrometry. Samples and standards were pyrolyzed at 1400°C in an alumina:glassy carbon reactor (Thermo-Finnigan TC/EA) producing H2 gas which was separated chromatographically in a 0.6-m molecular sieve gas chromatography (GC) column and analyzed for stable isotope composition using an isotope-ratio mass spectrometer (Thermo-Finnigan TC/EA and DeltaPLUS-XL ™, Bremen, Germany). Non-exchangeable stable hydrogen isotopic ratios (2H/1H) of the Japanese beetles were reported in delta (δ) notation in parts per thousand (per mil, ‰) deviation from the VSMOW on the VSMOW-SLAP standard scale (Vienna Standard Mean Ocean Water-Standard Light Antarctic Precipitation) where δ2H = [(2H/1Hsample) / (2H/1Hstandard) − 1]. In order to determine the δ2H of non-exchangeable H, we used the comparative equilibration method of [38]. Three keratin laboratory reference materials from Environment Canada CBS (Caribou hoof, -197.0 ‰), KHS (Kudu horn, -54.1 ‰), and Keratin SJ (-121.6 ‰), were used to normalize the results. CBS and KHS are described in [30]. Keratin SJ is Spectrum Keratin Powder Lot # SJ1400 from Spectrum Chemical Manufacturing Corporation, #K3030, Lot SJ1400 (Len Wassenaar, personal communication). The standard deviation for repeated analyses of an internal standard was consistently < 0.2 ‰.

Potential Sources of Origin

For the beetles captured at PDX, we assessed potential source regions using information about the status of the Japanese beetle invasion, as reported in the U.S. Japanese Beetle Harmonization Plan [15], updated June 2013, and supplemented with information from state and university extension agencies about specific reports of Japanese beetle captures [1214,16,17]. We focused on domestic sources only, because cargo air traffic to PDX is nearly exclusively domestic, as are the closest well-established populations of the Japanese beetle. Potential source regions included counties assigned to category 3 (Partially or Generally Infested) or to category 2 (Uninfested or Partially Infested) of the JB Harmonization Plan. Counties assigned to either category 1 (uninfested, under quarantine for import) or category 4 (not known to be infested) were not considered to be potential source regions. In Fig 1, areas assigned to categories 1 or 4 but with more recent reports of occasional captures of Japanese beetles are identified as sites experiencing potential invasion pressure (yellow regions), but these were not considered potential sources in the model. Major airports within each source region were identified and air traffic volume was recorded from publicly available data from the Federal Aviation Administration [39]. For all counties considered possible sources and for beetles whose isotope signatures identified the county as a possible source, the probability of a particular source county was weighted by air traffic from major airports. We also relaxed this prior, testing the model with equal probabilities of origin from all airports regardless of air traffic, an approach that assessed how well the stable isotope data could alone predict origin.

Monitoring and Control Measures

The Oregon Department of Agriculture (ODA) implements a state-wide detection program for the Japanese beetle each year to prevent its establishment in the state. Japanese beetle traps are placed at high-risk areas throughout the state, including major transportation hubs. The greatest density of traps is placed at PDX. Traps are deployed regularly at air cargo sites and other sites around PDX and the surrounding area, and sites are sprayed each year with pesticides where adult beetles have been trapped. Records from these trapping efforts and pesticide applications were provided by the ODA for the work presented here. Aircraft inspections began in 1991. On average, 45 aircraft per year are inspected for Japanese beetles by ODA employees; the inspections report the number of beetles found and whether the beetles found were alive or dead. All data are included in the Supporting Information files: the isotope turnover experiment (S1 and S2 Datasets), the field-collected beetle specimens from voucher and PDX trapping collections (S3 Dataset), and data on trapping and spraying efforts (S4 Dataset).

Bayesian Modeling

We constructed a hierarchical Bayesian model to estimate the time elapsed between arrival and capture and potential sources of origin for each beetle trapped at PDX. The model includes 1) an isotope mixing model for beetles collected at or near PDX to estimate potential sources of origin compiled by county (Fig 1), selected based on the status of the Japanese beetle invasion within the conterminous U.S.A., and weighted by air traffic through major airports; 2) latent variable models for each beetle’s unknown source of origin and time-since-arrival; 3) a beetle end-member model that incorporates the observed relationship between δ2H of precipitation and δ2H of the voucher beetles (see Fig 2); and 4) an isotope mixing model that describes the change in beetle δ2H after experimental relocation to a novel isotope environment. The model explicitly accounts for multiple sources of variation and uncertainty associated with observations measured with error and predicted quantities produced with error.

Fig 2.

A) Relationship between δ2H in Japanese beetles and the δ2H of mean annual precipitation from the region where beetles were collected. Small open symbols represent individual beetle collections. Grey circles are means of beetle samples from sites of similar δ2H in precipitation, with the size of the grey circle scaled to the number of samples from the site. Vertical and horizontal bars indicate 95% confidence intervals. B) Relationship between δ2H in Japanese beetles and the posterior estimate of monthly δ2H value of precipitation, with the most important month of year shown in color codes. Horizontal bars are 95% credible intervals. For both A and B, whole beetles were used for the δ2H analysis.

Mixing model for PDX beetles.

For each beetle i (i = 1, 2, …, 101) collected at or near PDX, we assume that the observed δ2H value of the beetle is normally distributed: (1.1) where μδi is the expected δ2H value and σi2 is the residual error variance for beetle i. The expected value is based on a linear mixing model that assumes that the δ2H value of each beetle is a mixture of (1) a beetle’s δ2H value from its potential source region (origin), denoted by δ2HSrc, and (2) the δ2H value of a beetle equilibrated with the PDX environment, denoted by δ2HPDX. Thus, μδi is given by: (1.2) where p is the relative contribution (0 ≤ p ≤ 1) of the PDX “end-member,” d(i) denotes the date that beetle i was sampled, and Srci denotes beetle i’s (unknown) source region. The date-specific end-members (δ2HPDX,d(i) and ) were computed as a weighted average of the monthly values predicted for the past 90 days (a period selected to encompass the life-span of an adult Japanese beetle [5]): (1.3) where wj,i is the weight associated with month index j relative to m(i), the month of collection, m, for beetle i (j = 0 = month of collection, j = 1 = previous month, etc.); the wj,i’s are computed as the proportional number of days (out of 90) that fall within each month period j. A similar equation is applied for .

We also employed a mixing model for σi2 to account for the possibility that the residual variance differs among sources: (1.4) where is the variance in δ2H among beetles at or near PDX, is variance in δ2H among beetles originating from one of the 57 possible source regions, and is variance in δ2H among beetles originating from a composite region whose δ2H values are not represented in the observed sources (see below). The indicator variable, IX, returns 0 if the logical function X is false, and 1 if X is true.

Based on the isotope turnover experiment, we modeled the relative contribution of the PDX end-member based on how long the beetle has been at or near PDX (i.e., time-since-arrive, or Time). In particular, (1.5) where a and b are parameters describing the rate of change in a beetle’s δ2H value following introduction to a new environment. T0 = -a/b + 14 is the number of days before the δ2H of the beetle starts to shift in response to a new environment (i.e., days for which p = 0), and T1 = (1-a)/b + 14 is the number of days it takes for the beetle to equilibrate with its new environment (i.e., when p = 1).

Latent beetle quantities.

The above, beetle-specific isotope mixing model requires Timei (time-since-arrival to PDX) and Srci (potential source region) for each beetle, but these quantities are latent (i.e., unknown). We implemented a latent variable model for each. Based on information about Japanese beetle invasion and air traffic, we assumed that there were 57 potential source areas. The δ2H of the precipitation from these source areas could not account for the high δ2H values that were measured for ∼20% of the beetles collected at or near PDX. Thus, we introduced a 58th composite region to account for an unknown source. For each beetle, we assume the simplest path from one potential source region to PDX. We specified an informative prior for Srci such that for source region j (j = 1, 2, …, 58): (1.6) where πj is proportional to the annual number of flights arriving at PDX from region j, and π58 is equal to 0.20 for the composite region. Eq 1.6 serves as an informative prior for the unknown Srci’s; the posterior solution distribution for each beetle’s Src may deviate from this prior once all data components and model assumptions are considered.

To constrain the estimates of Timei, we employed a hierarchical prior that beetles collected during the same year are likely to have similar days-since-arrival: (1.7) where yr(i) denotes sampling year associated with beetle i, and I(0.5, 60) indicates that the lognormal distribution is truncated at 0.5 and 60 days to avoid unrealistically extreme values for Timei (the “isotope turnover experiment” indicated that beetles equilibrate with their new environments well before 60 days). The arguments of the lognormal distribution (μTime and ) take on values on the log-scale and are assumed to vary by year.

δ2H end-members.

The end-members represent the δ2H value of beetles equilibrated with the PDX environment (δ2HPDX) and with the source environment (δ2HSrc). These are not known directly, but are informed by the δ2H values of precipitation in each location, and the relationship between beetle δ2H and precipitation δ2H based on the voucher specimens (see Fig 1). Following Eq 1.1, for each beetle voucher specimen v (v = 1, 2, …, 100), we assume: (1.8) where is the is the expected δ2H value for beetle v, and describes the residual error variance.

We modeled beetle δ2H based on the δ2H of the precipitation of each known collection location, such that the mean beetle δ2H is given by: (1.9) where Δ is the offset between beetle and precipitation δ2H, and δ2HPpt is the “true” δ2H value of precipitation, indexed by l(v), the location (l) associated with specimen v, and m(v), the month (m) during which specimen v was collected. l(v) is known, m(v) is not, so m(v) is treated as a latent variable, with a prior such that each of the 12 possible months (Jan, Feb, …, Dec) is equally probable.

The δ2H values describing the PDX and potential 58 source end-members (δ2HPDX and δ2HSrc, Eq 1.2) are computed based on the δ2H of precipitation in each area and the estimated offset derived from the voucher specimen model (Eq 1.9) such that the monthly end-member values are: (1.10) The voucher data and precipitation data inform the offset (Δ); uncertainty in Δ is propagated to Eq 1.10 by modularizing the voucher-precipitation model [4042].

We employed a Berkson-type model [43,44] to quantify uncertainty in the true δ2H of precipitation, expected to be normally distributed around location-specific estimates: (1.11) where are the predicted δ2H values, and are the associated variances, both obtained from an on-line isotopes-in-precipitation calculator [26,37], which provides month-specific estimates () and annual standard deviations for each location. We computed the variance describing variation in monthly δ2H () based on the formula for the variance of a mean () of N random variables [45]: (1.12) Eq 1.12 assumes that Var(xi) = σ2 (i.e., a common variance) for all i = 1, 2, …, N. Cov(xi,xj) is the covariance between xi and xj, and ρi,j is the correlation coefficient. Here, xi is the δ2H of precipitation in month i (for a given location), for N = 12 months, and σ2 is the variance in the monthly δ2H values (i.e., in Eq 1.11). The variance associated with the annual values, , is provided by the on-line isotopes in precipitation calculator (OPIC, [26,37]). The correlation, ρi,j, was estimated by computing Pearson correlations between the monthly δ2H of precipitation for the 72 locations (57 source locations plus 15 voucher locations), which varied from ρ = 0.819 (Jan vs. Aug) to ρ = 0.999 (Apr vs. Jun), with an average ρ of 0.961 (based on 66 pairwise correlations). Plugging in values for N and each ρi,j into Eq 1.12, the monthly variance () can be estimated as a function of the location-specific (l) annual variance: (1.13) Given the high ρ values, the monthly variances are slightly higher than the annual variances.

Mixing model for isotope turnover experiment.

The isotope turnover experiment informs the parameters a and b in Eq 1.5 For beetle j (j = 1, 2, …, 62) in this experiment, we assume a likelihood similar to Eqs 1.1 and 1.11 such that: (1.14) where is the predicted δ2H value for beetle j, and is the residual variance for “group” g associated with beetle j (groups are defined below). Based on the experimental data, the predicted δ2H of beetle j () is expected to reflect the source environment for 14 days following relocation, with complete equilibration with the new environment after 35 days, and intermediate values between 14 and 35 days. Thus, the model for is: (1.15) where δ2HInit is the beetle δ2H value in the source environment, δ2HFinal is the value equilibrated with PDX, and qj is the proportional contribution of the new isotope environment (δ2HFinal) to beetle j’s δ2H value. The residual variance varies among the three categories of TimeExp in Eq 1.15 (≤ 14 days, > 14 and < 35 days, or ≥ 35 days).

The proportional contribution of the new environment is expected to vary linearly with TimeExp, following Eq 1.5: (1.16) with a and b from Eq 1.5 (informed by the isotope turnover data), and the model described by Eqs 1.14 and 1.15 is modularized so that other datasets (such as the PDX beetle data) do not influence the a and b estimates. Uncertainty in these parameters (as quantified by their posterior distribution) is propagated to the model for pi in Eq 1.5.


We implemented the model in a Bayesian framework, specifying priors for all remaining unknown parameters (i.e., those not specified in Eqs 1.11.15, above). Standard deviation terms (σ) were assigned wide, uniform priors, Uniform(0,100), including the deviations for σPDX, σSource, and σComposite (Eq 1.4), σTime (Eq 1.7), σVS (Eq 1.8), and σExp (Eq 1.14). We assigned a diffuse normal prior, Normal(0, 1×105), to the offset between beetle δ2H and precipitation δ2H, Δ, in Eq 1.9. We assigned Normal(0, 1×107) priors to the initial and final beetle δ2H values, δ2HInit and δ2HFinal, and to the a and b parameters in the isotope turnover experiment model in Eq 1.15; b was restricted to positive values. Finally, we assigned a Normal(0, 1×106) prior to the year-specific mean time-since-arrival, μTime in Eq 1.7. Given the large variances of these normal priors, they are considered sufficiently vague.

We programmed the model in the Bayesian software package OpenBUGS [40]. Modularization of the voucher versus precipitation δ2H model (Eqs 1.8 and 1.9, the Berkson model for precipitation δ2H (Eqs 1.111.13), and the isotope turnover model (Eqs 1.141.16) was achieved by using the built-in cut function in OpenBUGS [40,42]. OpenBUGS employs Markov chain Monte Carlo (MCMC) to sample from the joint posterior of all unknown quantities. We ran three parallel MCMC chains with relatively dispersed starting values and at least 20,000 iterations. Convergence was evaluated with the built-in Brooks-Rubin-Gelman convergence diagnostic tool (chains converged by 5,000 iterations). The marginal posterior distribution for each quantity was summarized by its posterior mean and central 95% credible interval (CI), defined by the 2.5th and 97.5th percentiles. Annotated OpenBUGS code is provided as Supporting Information (S1 Text), as well as the data input file (S5 Dataset), and model results (S6 Dataset).


Hydrogen isotope signatures (δ2H) in Japanese beetle specimens collected from sites across the continental U.S.A. varied by region, and were correlated with δ2H of annual precipitation (Fig 2A, R2 = 0.81, S3 Dataset). The relationship improved when monthly data were used (R2 = 0.97), and the latent month was estimated, which indicated that the months spanning April to October were most likely to have contributed to beetle δ2H (Fig 2B, S6 Dataset). Based on this relationship, the Japanese beetles are expected to be 47.9‰ lower in δ2H compared to precipitation (posterior mean and 95% CI for Δ = -47.9 [-51.3, -44.3]).

The δ2H of beetles from Virginia experimentally reared in a terracosm with Oregon plants and water sources (isotope turnover experiment) began changing after about 14 days (T0 = 13.4 [7.5, 17.4], Fig 3, S1 Dataset). The δ2H values declined after that point at a rate of 4.7% per day (posterior estimate of slope, b, mean = 0.047, 95% CI = [0.031 to 0.063]). After about 35 days, beetles reached δ2H values expected for equilibrium with their new isotope environment (T1 = 35.6 [31.4, 42.2]). After five weeks, wings and elytra exhibited a smaller shift in δ2H compared to the rest of the body (Fig 3), likely indicating slower turnover of H through keratin and chitin. Still, the isotopic shift was apparent in these tissues, indicating that hydrogen incorporation into or synthesis of new keratin and chitin occurred during the 5-week experiment.

Fig 3. δ2H of Japanese beetles during a 5-week experiment in which beetles from Virginia were reared in terracosms with plant material and environmental water from Oregon.

Symbols show means and bars two standard errors. Whole-body tissue samples were analyzed each week (circles), whereas wings (squares) and elytra (triangles) were sampled in week 0 and week 5.

The δ2H signatures of Japanese beetles collected in the area around the Portland International Airport (PDX) increased over time from 2007 to 2014 (R2 = 0.63, slope = 9.2‰ per year), initially reflecting the δ2H of organisms equilibrated with Portland precipitation, and later matching the δ2H of organisms from the eastern U.S.A. (Fig 4, S3 Dataset). These changes in the δ2H value of beetles trapped at PDX indicated a reduction in the average time-since-arrival over the sampling interval (Fig 5, S6 Dataset). For example, the average time-since-arrival varied from ∼ 44 days in 2008 to ∼ 4 to 8 days in 2012–2014 (Fig 5). Given that the beetles retain their source origin isotope signature for ∼14 days after arriving at PDX, it is difficult to resolve finer scale changes in time-since-arrival. Nevertheless, beetles collected in 2012–2014 were recent arrivals, whereas beetles collected early in the study (2007–2010) had been present longer before trapping, long enough that they may have begun to establish resident populations (95% CI for mean time-since-arrive contains 35 days). 2011 marks a transition period where beetles collected at PDX had arrived ∼20 days before capture (95% CI for mean time-since-arrival = [16, 24] days) (Fig 5).

Fig 4. δ2H of Japanese beetles collected at the Portland International Airport (Oregon, USA) (circles) and in locations nearby (triangles) from 2006–2013.

Increased trap densities around the airport have increased the probability of catching beetles as they fly from planes.

Fig 5. Estimated days between arrival and capture of each individual beetle captured at the Portland International Airport (open circles) and the average for each year (blue squares, with bars showing 95% confidence intervals; note year with combined data).

The solid line shows the exponential fit for the temporal relationship.

Assigning likelihoods to specific source regions of newly arrived Japanese beetles at PDX was possible with our hierarchical Bayesian model. The model run with the “flat prior”, which assigned equal likelihoods to each potential source city, resulted in a strong pattern: more southerly locations had higher probabilities of being sources (Fig 6A, weighted regression, P<0.001, S6 Dataset), though the credible interval around the probabilities for each potential source were large. In the model run including the airline traffic data, the probability of specific locations was more strongly predicted by the prior (i.e., air traffic density), than by the isotope data (Fig 6B). Including the isotope data tended to increase the value of the posterior compared to the prior for more southerly locations, whereas the posterior tended to decline compared to the prior for more northerly latitudes (Fig 6B): the difference (posterior–prior) declined with latitude (slope = -0.00072 ± 0.00032, P = 0.031). Together, the δ2H data indicate that the southeastern U.S.A. is the more likely source of Japanese beetles arriving at PDX.

Fig 6.

A. Probability of specific locations as sources for Japanese beetles arriving at PDX as a function of latitude, using the “flat prior” model where all cities are assigned an equal prior likelihood. Filled circles show the mean posterior estimate from the hierarchical Bayesian model and vertical lines show 95% credible intervals. The linear fit is the weighted regression (P<0.001, slope = -0.00091 ± 0.00007), where each data point is weighted by the inverse of the credible interval (thus, smaller credible intervals have more weight). B. Same as A, but using the hierarchical Bayesian model in which the probability of each city is proportional to the cargo air traffic density from [39]. The prior assignments are shown in small white circles. The weighted regression is not significant (P = 0.486, slope = -0.00044 ± 0.00064).

The number of beetles per cargo aircraft at PDX declined exponentially from 1991 to 2014 at a rate of 15% per year (Fig 7A, S4 Dataset). From 2003 to 2014, air cargo traffic to PDX declined from approximately 90 to 50 cargo flight operations per day [46] (S4 Dataset). Efforts to control the Japanese beetle around PDX through pesticide application was higher during the sampling period (2006–2014) compared to the previous 15 years, and density of traps increased during this period as well, from 100 in the 1990s, to 1000 or more in the 2000s (S4 Dataset). The area sprayed each year did not exhibit any temporal trend from 2006 to 2014 (Fig 7B). Similarly, the numbers of beetles trapped per year did not exhibit strong temporal trends over the sampling period (2006–2014) but was higher during this period compared to the previous 15 years (1991–2005, Fig 7B), coincident with the increase in trap deployment.

Fig 7.

A. Number of Japanese beetles encountered per aircraft inspected over time. Inspections began in 1991 and an average of 45.1 aircraft were inspected per year. B. Number of Japanese beetles trapped per year, and number of acres sprayed per year with pesticides by the Oregon Department of Agriculture in the vicinity of the Portland International Airport (PDX).


Japanese beetles collected from established populations in the eastern U.S.A. were consistently higher in δ2H compared to individuals collected from the western U.S.A. (Fig 2), indicating fidelity between δ2H of body tissues and δ2H of precipitation at the site of origin for the Japanese beetle, consistent with numerous other studies that have documented such relationships between δ2H of precipitation and insect tissues [21,24,28]. The relationship we found for Japanese beetles has a similar coefficient of determination and slightly larger offset (-47.9‰, R2 = 0.81) between beetle δ2H and precipitation δ2H compared to past analyses for butterfly wings in Europe (-40.6‰, R2 = 0.87 [28]) and dragonfly wings in North America -42.5‰, R2 = 0.75 [31]), though overall the relationships are quite similar. Our results confirm that, in regions with sufficient geographic variation in the δ2H value of precipitation, stable isotopes are useful for tracing the natal origins of organisms, including butterflies [19], birds [32,47], bats [48], and invasive beetles (Fig 2). Our work also confirms the potential for stable isotopes to provide information about the more recent movements of organisms [28,49], which could prove valuable in tracking and managing invasive organisms [50].

For invasive species, isotopes can provide information complementary to other techniques. Molecular signals can resolve geographic patterns of invasive species’ range expansions as invasions proceed, distinguishing invaders from different source populations [51], as long as the populations have been separate for enough generations to develop genetic differences. For more rapidly expanding invasive species, genetic differentiation among populations may be absent and therefore not useful in identifying proximate sources [52], and, molecular signals provide little information about the short-term temporal dynamics of migrating organisms relevant to the time-scale of biological invasions. In contrast, the stable isotope composition of organisms can change within the lifetime of an individual, depending on the timescale of element turnover through body tissues [53] or body water [54]. Compared to molecular genetic tools, stable isotopes can reveal information about the recent origin of invasive organisms on a much finer timescale, likely more suitable to the timescale of decisions about controlling populations of invasive organisms. For example, δ2H analyses showed that populations of the marmalade hoverfly (Episyrphus balteatus) occurring in agricultural fields in France comprise both locally established individuals and new arrivals from more southern regions [55]. This species of hoverfly eats aphids so is important in biological control of an agricultural pest. The finding of low immigration indicates that focusing management on overwintering populations may be the most effective means to ensure continued biocontrol from this species [55]. For pests, as in our study, using stable isotopes can also help identify the immediate origins of individual organisms, whether they are locally established or intercepted vagrants.

The hierarchical Bayesian model did not clearly distinguish specific source regions of Japanese beetles, likely because of the many disparate sources of Japanese beetles throughout the eastern U.S.A. The resolution of the model could be improved in several ways: (1) finer resolution between the development of the beetles and the tissue δ2H, possibly resulting in a better regression based on monthly precipitation; (2) a model that more fully accounts for the beetles’ ontogeny, tracking, for example, the higher likelihood that earlier arrivals at PDX come from more southern sites with warmer climates and thus earlier beetle emergence; (3) including beetle tissues of varying turnover time and using the differences among tissues in δ2H to provide even greater temporal resolution; and (4) including other tracers that exhibit more geographic variation among potential source regions (possibly genetic markers, or other elements and isotopes). Despite the weak spatial resolution, the estimates of time since arrival were robust (R2 = 0.92 and slope = 0.92, for a comparison of time since arrival predicted between model runs with and without the air traffic prior). There is less geographic variation in δ2H among potential source regions compared to the difference in δ2H between all eastern source populations and sites in the western U.S.A. vulnerable to invasion. For this reason, estimates of arrival time were more precise that were estimates of geographical origin.

The increasing δ2H signatures of Japanese beetles captured around PDX (Fig 4) indicate that individuals collected during the earlier years (2006–2009) of control efforts may have been part of an establishing population, partially or fully equilibrated with the new (Portland) isotopic environment. By contrast, organisms captured more recently (2010–2014) were newer arrivals (Fig 5), still reflecting the isotopic signature of the source populations from the likely origin sites in the eastern USA (Fig 2). The shift in δ2H signatures toward source populations coincident with multiple years of pesticide ground treatments suggests that these management strategies in the area surrounding PDX has helped prevent the establishment of a local viable population. The decline in the frequency of beetles found on inspected aircraft (Fig 7) is consistent with more effective control measures at the source populations [56], especially given the decline in cargo air traffic to PDX from 2003 to 2014 [46]. Together, these indicate a reduction in the invasion pressure from cargo aircraft to PDX over this time period. Despite better control at source populations, new individuals continue to arrive with distinctly eastern δ2H signatures. Increased trap deployment may have reduced the time between the beetles’ arrival and trapping. δ2H appears to have been an independent and sensitive indicator of the status of the invasion. While it is difficult to attribute the change in δ2H over time to specific causes, the decline is consistent with control measures that have reduced the time between the arrival and capture of potentially invading individuals.

The anomalously low δ2H signatures for some individual Japanese beetles captured in 2010 and 2013 were statistical outliers for the populations of beetle captures during those years, though they fell within the range of δ2H values expected for organisms equilibrated with the PDX environment. In the world of invasions, outliers may be the few individuals required to establish a viable population [57]: these low δ2H values indicate that these individuals were present in the region long enough to equilibrate with the new environment. Given the heterogeneous distribution of Japanese beetle populations across the landscape, a low number of trapped beetles could indicate the presence of a viable population [58].

Our finding that δ2H of beetle tissues shifted after experimentally controlled migration from Virginia to Oregon (Fig 3) demonstrates that bulk tissue hydrogen isotope analysis can resolve recent movements of this organism: the shift in δ2H began after two weeks in the novel environment, was statistically significant after three weeks, and changed over a period of five weeks before reaching the expected δ2H value of the new environment. Individual Japanese beetles that preserve a clear δ2H signature of a distant source population are likely to be new arrivals, where ‘new’ for this system means having arrived within the preceding two weeks. Compared to the shift in beetle body δ2H after five weeks, the smaller shift in δ2H in wings and elytra by week five is consistent with these tissues being composed of keratin and chitin, thought to be stable after synthesis; yet, the decline in δ2H suggests that hydrogen exchange or new tissue formation occurred in these organisms during the experiment (Fig 3B). Combining δ2H analysis of stable tissues to record natal origin and of tissues that turn over more rapidly to record the timing of relocation would likely provide more power in resolving both the origins and movements of invasive organisms.

Isotope monitoring can guide management strategies: any shift toward local precipitation would be evidence that colonists are beginning to establish a new population, and would justify more intensive local control measures. By contrast, persistent trapping of individuals with the δ2H value consistent with the source populations would support increased inspection and prevention of colonists’ being able to board the aircraft in the first place. Continued trapping and monitoring the isotopic signatures of these potentially invasive organisms can provide managers with critical information about the status of the invasion. In general, our results point to the utility of isotope sentinels for invasive species: in cases where the δ2H of precipitation differs between source populations and sites at risk for invasion, monitoring captured individuals for isotopic composition can distinguish between recent arrivals and establishing populations.

Supporting Information

S1 Dataset. Data from the isotope turnover experiment in which beetles were experimentally transplanted to Oregon from their native environment in Virginia and grown in a terrarium.

Beetle samples were collected weekly for δ2H determinations to determine the shift in δ2H over time. Excel workbook with 2 worksheets.


S2 Dataset. Mass of Japanese beetle wings, elytra, and remainder of the body.

These masses were used to reconstruct whole-beetle δ2H values in the isotope turnover for weeks 1 and 5, cases where each beetle part was analyzed separately for δ2H. The δ2H value of the whole beetle sample was then calculated by mass balance. Excel workbook with 2 worksheets.


S3 Dataset. Japanese beetle δ2H values from samples collected around the conterminous U.S.A. and from within the study region, the area around the Portland International Airport (PDX, Portland, OR, U.S.A.).

Excel workbook with 2 worksheets.


S4 Dataset. Control measures taken by the Oregon Department of Agriculture to limit the invasion of the Japanese beetle, focusing on the Portland International Airport (PDX, Portland, OR, U.S.A.) as a potential point of entry.

Data include the numbers of beetles found during aircraft inspections, the numbers of beetles trapped in the field in the PDX region, the ground area treated with pesticide applications, the number of traps deployed each year in the PDX region, and air cargo traffic data available from the Port of Portland website. Excel workbook with 3 worksheets.


S5 Dataset. Input data required to run the hierarchical Bayesian model.

See readme file for more information. Excel workbook with 7 worksheets.


S6 Dataset. Output data from the hierarchical Bayesian model.

See readme file for more information. Excel workbook with 9 worksheets.


S1 Text. The hierarchical Bayesian model code, written in the program OpenBUGS.

PDF file with 6 pages.



We thank the following individuals for helping with this study: G. Brown, USDA APHIS PPQ, approved the quarantine facility. T. Finn, VA Department of Agriculture, collected and shipped overnight live Japanese beetles to Oregon for the experiment. ODA staff, E. Coombs, T. Adams, M. Savelich, and W. Staben constructed the quarantine facility for the experiment. ODA staff (P. Mitchell, E. Corey) collected trapped specimens at PDX. This research was supported by the Oregon State Department of Agriculture and the Northern Arizona University Technology Research Initiative Fund.

Author Contributions

Conceived and designed the experiments: DNK HWR BAH. Performed the experiments: DNK MC BAH. Analyzed the data: KO BAH JCM MC. Contributed reagents/materials/analysis tools: KO BAH MC HWR DNK. Wrote the paper: BAH DNK KO MC JCM.


  1. 1. Carey JR (1991) Establishment of the Mediterranean fruit fly in California. Science 253: 1369–1373. pmid:1896848
  2. 2. Liebhold AM, Work TT, McCullough DG, Cavey JF (2006) Airline Baggage as a Pathway for Alien Insect Species Invading the United States. American Entomologist 52: 48–54.
  3. 3. Byers JE, Reichard S, Randall JM, Seastedt TR, Williamson M, Chornesky E, et al. (2002) Directing Research to Reduce the Impacts of Nonindigenous Species. Conservation Biology 16: 630–640.
  4. 4. Bodey TW, Bearhop S, McDonald RA (2011) Invasions and stable isotope analysis − informing ecology and management. In: Veitch CR, Clout MN, Towns DR, editors. Island invasives: eradication and management Gland, Switzerland: IUCN. pp. 148–151.
  5. 5. Fleming WE (1972) Biology of the Japanese beetle. Washington, DC.: United States Department of Agriculture.
  6. 6. Potter DA, Braman SK (1991) Ecology and Management of Turfgrass Insects. Annual Review of Entomology 36: 383–406.
  7. 7. Potter DA, Held DW (2002) Biology and management of the Japanese beetle. Annual Review of Entomology 47: 175–205. pmid:11729073
  8. 8. Tiddens PT, Cloyd RA (2006) Susceptibility of three rose genotypes to Japanese beetle (Coleoptera: Scarabaeidae) adult feeding. Arboriculture & Urban Forestry 32: 108–113.
  9. 9. Niziolek OK, Berenbaum MR, DeLucia EH (2013) Impact of elevated CO2 and increased temperature on Japanese beetle herbivory. Insect Science 20: 513–523. pmid:23955947
  10. 10. Hammons DL, Kurtural SK, Newman MC, Potter DA (2009) Invasive Japanese beetles facilitate aggregation and injury by a native scarab pest of ripening fruits. Proceedings of the National Academy of Sciences of the United States of America 106: 3686–3691. pmid:19234133
  11. 11. USDA (2010) Japanese Beetle Program Manual For Airports. In: USDA, editor: USDA. pp. 124.
  12. 12. Bethke JA. Insect Hot Topics: Japanese beetles. Davis, CA: University of California. 2014. Available:
  13. 13. Bullock S, de Yong R. Quarrantine Name: Japanese beetle. 2013. Montana Department of Agriculture. Helena, Montana. Available:
  14. 14. Castrovillo P. Japanese Beetle in Idaho. 2014. Idaho State Department of Agriculture. 2 p. Available:
  15. 15. Cooper M, Bech R, Geary MV. U.S. Domestic National Japanese Beetle Harmonization Plan. 1998. National Plant Board. Revised June 2013. 43 p. Available:
  16. 16. Hodgson E, Alston D, Vorel CA (2010) Japanese Beetle (Popillia japonica). Utah Pests Fact Sheet. Ogden, Utah: Extension, Utah State University. Available:
  17. 17. Shanklin D, Loftin K, Hopkins J. Common Questions About Japanese Beetles in Arkansas. 2013. University of Arkansas, Division of Agriculture, Cooperative Extension Service. 4 p. Available:
  18. 18. Ruiz G, Carton J (2003) Invasion vectors: a conceptual framework for management. In: Ruiz G, Carlton J, editors. INvasive Species: Vectors and Management Strategies. Washington: Island Press. pp. 457–504.
  19. 19. Wassenaar LI, Hobson KA (1998) Natal origins of migratory monarch butterflies at wintering colonies in Mexico: New isotopic evidence. Proceedings of the National Academy of Sciences of the United States of America 95: 15436–15439. pmid:9860986
  20. 20. Vander Zanden MJ, Casselman JM, Rasmussen JB (1999) Stable isotope evidence for the food web consequences of species invasions in lakes. Nature 401: 464–467.
  21. 21. Hobson KA (1999) Tracing origins and migration of wildlife using stable isotopes: a review. Oecologia 120: 314–326.
  22. 22. Yakir D, DeNiro MJ (1990) Oxygen and hydrogen isotope fractionation during cellulos metabolism in Lemna gibba L. Plant Physiology 93: 325–332. pmid:16667454
  23. 23. Bowen GJ, Liu Z, Vander Zanden HB, Zhao L, Takahashi G (2014) Geographic assignment with stable isotopes in IsoMAP. Methods in Ecology and Evolution 5: 201–206.
  24. 24. Bowen GJ, Wassenaar LI, Hobson KA (2005) Global application of stable hydrogen and oxygen isotopes to wildlife forensics. Oecologia 143: 337–348. pmid:15726429
  25. 25. Craig H (1961) Isotopic Variations in Meteoric Waters. Science 133: 1702–1703. pmid:17814749
  26. 26. Bowen GJ, Revenaugh J (2003) Interpolating the isotopic composition of modern meteoric precipitation. Water Resources Research 39: 1299.
  27. 27. Gröcke DR, van Hardenbroek M, Sauer PE, Elias SA (2011) Hydrogen Isotopes in Beetle Chitin. In: Gupta NS, editor. Chitin: Formation Amd Diagenesis. Dordrecht: Springer. pp. 105–116.
  28. 28. Brattstrom O, Bensch S, Wassenaar LI, Hobson KA, Akesson S (2010) Understanding the migration ecology of European red admirals Vanessa atalanta using stable hydrogen isotopes. Ecography 33: 720–729.
  29. 29. Flockhart DTT, Wassenaar LI, Martin TG, Hobson KA, Wunder MB, Norris DR (2013) Tracking multi-generational colonization of the breeding grounds by monarch butterflies in eastern North America. Proceedings of the Royal Society B-Biological Sciences 280.
  30. 30. Hobson KA, Koehler G (2015) On the use of stable oxygen isotope (delta O-18) measurements for tracking avian movements in North America. Ecology and Evolution 5: 799–806. pmid:25691999
  31. 31. Hobson KA, Soto DX, Paulson DR, Wassenaar LI, Matthews JH (2012) A dragonfly (δ2H) isoscape for North America: a new tool for determining natal origins of migratory aquatic emergent insects. Methods in Ecology and Evolution 3: 766–772.
  32. 32. Hobson KA, Van Wilgenburg SL, Faaborg J, Toms JD, Rengifo C, Sosa AL (2014) Connecting breeding and wintering grounds of Neotropical migrant songbirds using stable hydrogen isotopes: a call for an isotopic atlas of migratory connectivity. Journal of Field Ornithology 85: 237–257.
  33. 33. Holder PW, Armstrong K, Hale RV, Millet M-A, Frew R, Clough TJ (2014) Isotopes and Trace Elements as Natal Origin Markers of Helicoverpa armigera–An Experimental Model for Biosecurity Pests. Plos One 9: e92384. pmid:24664236
  34. 34. Forbes KJ, Gratton C (2011) Stable isotopes reveal different patterns of inter-crop dispersal in two ladybeetle species. Ecological Entomology 36: 396–400.
  35. 35. Podlesak DW, Torregrossa A- M, Ehleringer JR, Dearing MD, Passey BH, Cerling TE (2008) Turnover of oxygen and hydrogen isotopes in the body water, CO2, hair, and enamel of a small mammal. Geochimica et Cosmochimica Acta 72: 19–35.
  36. 36. Barrow LM, Bjorndal KA, Reich KJ (2008) Effects of preservation method on stable carbon and nitrogen isotope values. Physiological and Biochemical Zoology 81: 688–693. pmid:18752417
  37. 37. Bowen GJ. The Online Isotopes in Precipitation Calculator, version 2.2. 2015. Available:
  38. 38. Wassenaar LI, Hobson KA (2003) Comparative equilibration and online technique for determination of non-exchangeable hydrogen of keratins for use in animal migration studies. Isotopes in Environmental and Health Studies 39: 211–217. pmid:14521282
  39. 39. Federal Aviation Administration. Passenger Boarding (Enplanement) and All-Cargo Data for U.S. Airports Airports. 2013. Washington DC: Federal Aviation Administration. Available:
  40. 40. Lunn D, Spiegelhalter D, Thomas A, Best N (2009) The BUGS project: Evolution, critique and future directions. Statistics in Medicine 28: 3049–3067. pmid:19630097
  41. 41. Liu F, Bayarri MJ, Berger JO (2009) Modularization in Bayesian analysis, with emphasis on analysis of computer models. Bayesian Analysis 4: 119–150.
  42. 42. Ogle K, Barber J, Sartor K (2013) Feedback and modularization in a Bayesian meta-analysis of tree traits affecting forest dynamics. Bayesian Analysis 8: 133–168.
  43. 43. Dellaportas P, Stephens DA (1995) Bayesian analysis of errors-in-variables regression models. Biometrics 51: 1085–1095.
  44. 44. Ogle K, P E. (2015) Isotope partitioning of soil respiration: Bayesian solution to accommodate multiple sources of variability. Journal of Geophysical Research–Biogeosciences 120: 221–236.
  45. 45. Casella G, Berger RL (2002) Statistical Inference. Pacific Grove, CA: Duxbury. 660 p.
  46. 46. Po Portland (2016) PDX Aviation Statistics. In: Co Portland, editor. Portland, OR: City of Portland.
  47. 47. Bridge ES, Kelly JF, Xiao XM, Takekawa JY, Hill NJ, Yamage M. (2014) Bird migration and avian influenza: A comparison of hydrogen stable isotopes and satellite tracking methods. Ecological Indicators 45: 266–273. pmid:25045322
  48. 48. Baerwald EF, Patterson WP, Barclay RMR (2014) Origins and migratory patterns of bats killed by wind turbines in southern Alberta: evidence from stable isotopes. Ecosphere 5.
  49. 49. Hobson KA, Wassenaar LI, Bayne EM (2004) Using isotopic variance to detect long-distance dispersal and philopatry in birds: An example with Ovenbirds and American Redstarts. Condor 106: 732–743.
  50. 50. Poland TM, McCullough DG (2006) Emerald ash borer: Invasion of the urban forest and the threat to North America's ash resource. Journal of Forestry 104: 118–124.
  51. 51. Virgilio M, Delatte H, Backeljau T, De Meyer M (2010) Macrogeographic population structuring in the cosmopolitan agricultural pest Bactrocera cucurbitae (Diptera: Tephritidae). Molecular Ecology 19: 2713–2724. pmid:20561200
  52. 52. Aketarawong N, Isasawin S, Thanaphum S (2014) Evidence of weak genetic structure and recent gene flow between Bactrocera dorsalis s.s. and B. papayae, across Southern Thailand and West Malaysia, supporting a single target pest for SIT applications. Bmc Genetics 15.
  53. 53. Tieszen LL, Boutton TW, Tesdahl KG, Slade NA (1983) Fractionation and turnover of stable carbon isotopes in animal-tissues—implications for delta-C-13 analysis of diet. Oecologia 57: 32–37.
  54. 54. Podlesak DW, Bowen GJ, O'Grady S, Cerling TE, Ehleringer JR (2012) delta H-2 and delta O-18 of human body water: a GIS model to distinguish residents from non-residents in the contiguous USA. Isotopes in Environmental and Health Studies 48: 259–279. pmid:22397457
  55. 55. Raymond L, Vialatte A, Plantegenest M (2014) Combination of morphometric and isotopic tools for studying spring migration dynamics in Episyrphus balteatus. Ecosphere 5.
  56. 56. Hamilton RM, Foster RE, Gibb TJ, Sadof CS, Holland JD, Engel BA (2007) Distribution and dynamics of Japanese beetles along the Indianapolis airport perimeter and the influence of land use on trap catch. Environmental Entomology 36: 287–296. pmid:17445363
  57. 57. Aranda C, Eritja R, Roiz D (2006) First record and establishment of the mosquito Aedes albopictus in Spain. Medical and Veterinary Entomology 20: 150–152. pmid:16608499
  58. 58. Dalthorp D, Nyrop J, Villani M (1999) Estimation of local mean population densities of Japanese beetle grubs (Scarabaeidae: Coleoptera). Environmental Entomology 28: 255–265.