Abstract
Spatial and temporal distribution data provide critical information for invasive species management. For example, distribution data can help managers with early detection and to guide other response actions. Environmental DNA (eDNA)-based sampling exists as one tool for monitoring invasive species. As part of bigheaded carp Hypophthalmichthys spp. monitoring efforts in the Illinois River, United States, we compared eDNA-based sampling results at multiple habitats across an invasion gradient in 2015. Greater densities of carp occurred downriver in the Illinois River, and lower densities occurred upriver. We sampled from five locations along this gradient and from three habitat types (backwater, main channel, and shoreline) within each location. We sampled each location in April and June. A priori, we hypothesized that 1) more eDNA detections would occur downriver, where higher densities of carp occur, 2) more eDNA detections would occur in backwater habitats than in areas of the river with more fish movement, and 3) more eDNA detections would occur in April because bigheaded carps are thought to use our sampling areas more during the spring. We compared the proportion of samples positive across this gradient, the habitat type, and the two sampling time periods. The most downriver location had the highest proportion of samples with eDNA detections, the backwater habitats had the highest proportion of samples with eDNA detections, and April had more positive detections than June. Our results highlight the importance of sampling across multiple habitat types and across time to gain a clear understanding of distribution when using eDNA-based sampling. Thus, being cognizant of the interactions between seasonal habitat use and eDNA-based detections is important for managers who rely on eDNA-based monitoring.
Introduction
Invasive species often degrade habitats, disrupt ecosystems, decrease native species abundances (Pejchar and Mooney 2009), and adversely affect economic activities (Cuthbert et al. 2021). Understanding distributions, movement, and relative changes in populations for invasive species can assist resource managers by giving them knowledge to inform management activities (Lodge et al. 2006; Reaser et al. 2020). Environmental DNA (eDNA)-based sampling methods can be used to help managers monitor invasive species (Darling and Blum 2007; Beng and Corlett 2020). eDNA-based methods extract DNA that organisms deposit in the environment (for example, in water, air, or soil) to determine which species, or at least their DNA, occur at a site (Ficetola et al. 2008). Monitoring programs using eDNA-based methods can be more efficient for detecting species than traditional sampling approaches (Fediajevaite et al. 2021), although this is not always the case (for example, refer to Ulibarri et al. 2017 or Randall et al. 2023). The application of eDNA-based sampling as a monitoring tool depends on multiple factors, including the ecology, biology, and chemistry of eDNA in the field (Barnes and Turner 2015); methodology for collection, isolation, and amplification of DNA (Goldberg et al. 2016; Ladell et al. 2018; Lance and Guan 2020); and statistical study design for sampling and replication (Erickson et al. 2019; Mize et al. 2019).
The allocation and placement of replicate samples are important components when designing broad-scale eDNA-based monitoring programs (Erickson et al. 2019; Mize et al. 2019). Specifically, all monitoring programs and studies have a finite number of replicates, and trade-offs exist with how replicates are used across sampling levels. Within monitoring programs, trade-offs exist among the number of locations sampled, the number of water samples per location, and the number of subsamples per water sample (for example, compare sampling 100 sites with 1 water sample per site and 1 quantitative polymerase chain reaction [qPCR] sample per site with sampling 1 site with 100 water samples and 1 qPCR replicate per sample or 1 site with 1 water sample and 100 qPCR replicates per water sample). Monitoring programs using eDNA-based methods need to balance these trade-offs. At the lowest level of replication, the number of molecular replicates affects the probability of detection for eDNA in a water sample, for example, the allocation of qPCR replicates per water sample. These detection probabilities influence the likelihood of detecting eDNA from a low-density species, such as at the leading edge of an invasion. Likewise, the number of water samples and molecular replicates must be balanced across spatial and temporal replication that dictate where and when sample collection occurs.
Sample placement through space and time, a topic related to replicate allocation, also affects eDNA-based monitoring programs because most species are distributed heterogeneously across landscapes, and species distributions may change seasonally for some species (Krebs 1999; Erickson et al. 2019; Mize et al. 2019). A mismatch between sampling design and project goals may limit conclusions from the data collected or potentially miss important patterns and trends. Understanding the distribution of eDNA across a landscape informs these design decisions, and many different possible designs exist. Some examples include random sampling designs (Dickie et al. 2018), grid-based sampling design (Mize et al. 2019), transect-based sampling design (Erickson et al. 2016, 2017), stratification or blocking sampling designs (Krebs 1999), and targeted or direct sampling designs (Harper et al. 2019). Thus, sampling design requires careful attention based on study or monitoring objectives (for example, early detection, population monitoring, or response to management actions).
An example of a large-scale monitoring effort in the United States is ongoing work by the U.S. Fish and Wildlife Service (USFWS) that focuses on tracking the distribution of invasive carp across the Mississippi River and Great Lakes Basins (Mize et al. 2019). This study focuses only on silver carp Hypophthalmichthys molitrix, bighead carp H. nobilis, and their hybrid species Hypophthalmichthys spp. (bigheaded carp). Within this paper, we refer to both species and their hybrids as “bigheaded carp” unless describing a species-specific method or result. These species, as well as grass carp Ctenopharyngodon idella, black carp Mylopharyngodon piceus, and sometimes common carp Cyprinus carpio, were recently known as “Asian carp,” but this term has fallen out of favor due to being biologically uninformative, linguistically inaccurate, and culturally insensitive (Kočovský et al. 2018). However, the generic bigheaded carp markers use the “AC” as part of the markers’ names, and this history provides context for the markers’ names. Both bigheaded species hybridize in North America, unlike their native range (Lamer et al. 2010), and are typically managed as one invasive species because both species occupy similar habitats, and control actions typically affect both species (Cupp et al. 2021). Kolar et al. (2005) provide an overview of the invasion history of these species. Briefly, bigheaded carp are currently established in at least 20 countries, including the United States. The species, native to central and eastern Asia, were imported to the United States in the early 1970s for use in aquaculture and wastewater treatment ponds to control algae. Due to flooding, the two species escaped and are now widespread in the Mississippi River Basin where concern exists that both species could become established in the Laurentian Great Lakes through the Illinois River/Chicago Area Waterway System (Ivan et al. 2020; Chapman et al. 2021).
Concern of bigheaded carp infesting the Great Lakes has led to the use of eDNA-based sampling as a monitoring tool for these species for over a decade in the Illinois River (Jerde et al. 2011, 2013). The adult population in Dresden Island Pool serves as the invasion front for bigheaded carp in the Illinois River, with more individuals occurring downriver in the Illinois River system that is closer to the source population in the Mississippi River, creating an invasion gradient (Invasive Carp Regional Coordinating Committee 2022; Figure 1). Currently, the USFWS, in coordination with partner state and federal agencies, leads eDNA-based monitoring for bigheaded carp in the upper Mississippi River and its subbasins. In addition to the upper Mississippi River, the USFWS samples in subbasins of the Great Lakes for bigheaded carp as part of early detection work (Asian Carp Regional Coordinating Committee 2015a). The intense monitoring effort for bigheaded carp in the Illinois River highlights the importance of assessing the factors that can improve detection and optimize study design. Optimizing sampling efforts is important for the Illinois River and similar systems with bigheaded carp monitoring programs because of the immense efforts put forth in these systems. Hence, efficiently collecting detection data can help manage bigheaded carp in the Illinois River and may also provide insight for other bigheaded carp monitoring programs and a model for eDNA-based monitoring programs for other species.
Map denoting the extent of silver carp Hypophthalmichthys molitrix and bighead carp H. nobilis environmental DNA sampling by the U.S. Fish and Wildlife Service within the Illinois Waterway in 2015. The U.S. Fish and Wildlife Service collected water samples from the lower Kankakee River and the Illinois River from the Marseilles Pool to the Lockport Pool on two separate sampling events; the first occurred from 21 to 23 April 2015, and the second occurred from 2 to 4 June 2015.
Map denoting the extent of silver carp Hypophthalmichthys molitrix and bighead carp H. nobilis environmental DNA sampling by the U.S. Fish and Wildlife Service within the Illinois Waterway in 2015. The U.S. Fish and Wildlife Service collected water samples from the lower Kankakee River and the Illinois River from the Marseilles Pool to the Lockport Pool on two separate sampling events; the first occurred from 21 to 23 April 2015, and the second occurred from 2 to 4 June 2015.
We tested three hypotheses about eDNA-based sampling using data from a 2015 collection project by the USFWS. Our first hypothesis was that the proportion of eDNA samples containing eDNA would change along an invasion gradient, with fewer detections occurring farther upriver where bigheaded carp densities are lower based on work by Coulter et al. (2018). Second, we tested the hypothesis that a higher proportion of samples would contain bigheaded carp eDNA from sampling along the edges of the river channel and within low-flow habitats including backwaters and bays based on previous observations by Coulter et al. (2017), MacNamara et al. (2018), and Glubzinski et al. (2021). Third, we tested the hypothesis that more detection of bigheaded carp eDNA would occur during the spring (April) sampling event than in the summer (June) sampling event because we expected the species to use the sampled areas more during these seasons based on their seasonal movements and spawning behaviors.
Materials and Methods
Sampling locations
As part of the USFWS bigheaded carp monitoring program in the Illinois River system, the USFWS collected eDNA samples along transects (where habitats were not assigned until after the sample geographic locations were selected) in different locations across the bigheaded carp invasion gradient in the Illinois River Basin in 2015. The USFWS collected these samples to determine if eDNA occurrence changed across an invasion gradient and if detections were related to seasonal differences. USFWS personnel collected water samples from the lower Kankakee River and in the Illinois River from the Marseilles Pool to the lower Lockport Pool on two separate sampling events, from 21 to 23 April 2015 and from 2 to 4 June 2015 (Table 1; Figure 1). The USFWS Fish and Wildlife Conservation Offices collected the samples at evenly spaced transects between lock and dams relative to total river length (Asian Carp Regional Coordinating Committee 2015a), sampled transects perpendicular to the flow, and collected three samples per transect, one near the left descending bank, one in the midchannel, and one near the right descending bank. The USFWS Fish and Wildlife Conservation Offices collected samples either upstream from the boat or off the bow of the boat following methods described in U.S. Fish and Wildlife Service (2015). They designated these sample locations as one of three habitat types: bay, main channel, or shoreline. USFWS personnel collected the main channel samples from the thalweg of the main channel or side channel, the shoreline samples from the nearshore of the main channel or side channel, and the bay samples from off-channel, nonflowing areas, such as marinas, backwaters, barge slips, and similar habitats. Each sample consisted of 50-mL centrifuge tubes of surface water collected at each unique sample location, unlike the standard 250 mL collected for monitoring in the Chicago Area Waterway System (U.S. Fish and Wildlife Service 2015) because bigheaded carp were known to occur in the Illinois River Basin, and this allowed the USFWS to spread sampling efforts along a great length of the river (Asian Carp Regional Coordinating Committee 2015b). USFWS personnel centrifuged and preserved the water samples with 95% ethanol and shipped them to the USFWS Whitney Genetics Laboratory for analysis (Asian Carp Regional Coordinating Committee 2015a, 2015c; U.S. Fish and Wildlife Service 2015). At the field sites, they randomly selected 2 of 23 sampling sites for cooler blanks filled with deionized water (specifically, 3 2-L bottles and 15 50-mL centrifuge tubes per site). These negative controls did not replace the collection of water samples at the site but served as negative controls (U.S. Fish and Wildlife Service 2015).
Summary of positive detections of bigheaded carp environmental DNA from Hypophthalmichthys spp. (AC), bighead carp H. nobilis (BH), and silver carp H. molitrix (SC) from the U.S. Fish and Wildlife Service Monitoring program in the Illinois River Basin in 2015 grouped by pool, month, and quantitative polymerase chain reaction markers (refer to the Laboratory methods section for an explanation of the markers). We designated a sample as positive if fluorescence was detected in one or more of eight quantitative polymerase chain reaction replicates. n indicates the total number of samples per site.

Laboratory methods
The Whitney Genetics Laboratory extracted and analyzed all samples according to protocols described in the 2015 Quality Assurance Project Plan for the eDNA-based monitoring of bigheaded carp (U.S. Fish and Wildlife Service 2015). Upon receipt, laboratory personnel centrifuged the samples for an additional 5 min to reconcentrate suspended debris and placed the samples in laminar flow hoods to allow complete evaporation of alcohol before DNA extraction. They then used a cotton swab to gather the dried pellet from each sample centrifuge tube and extracted the DNA from each concentrated sample using IBI Scientific gMAX kits (IBI Scientific, Dubuque, IA). Each batch of tubes included a hood negative control to test for contamination during the drying process. The investigators included positive and negative extraction controls. Laboratory personnel initially analyzed the samples in eight replicate duplex qPCR reactions using a pair of markers labelled AC1 (FAM-labelled probe) and AC3 (HEX-labelled probe), both designed to detect loci within the cytochrome oxidase 1 gene of silver and bighead carp. If they detected either AC1 or AC3 fluorescence in at least one of the eight sample replicates, they considered the sample a positive detection and reanalyzed it using the following four additional species-specific markers: SC4, SC5, BH1, and BH2. The AC1 and AC3 markers are located on either end of the cytochrome oxidase 1 gene, the BH1 and BH2 markers detect loci on the ND2 and ND4 mitochondria regions, and the SC4 and SC5 markers detect loci on the ND6 and ND2 mitochondria regions. Both SC4 and SC5 markers amplify mitochondrial genome regions specific to silver carp, whereas BH1 and BH2 markers amplify two loci specific to bighead carp. For additional details, refer to Farrington et al. (2015) who provide a detailed description of marker development and validation. The original sampling protocol (U.S. Fish and Wildlife Service 2015) required the confirmation of species eDNA by the detection of at least one silver- or bighead-specific locus as well as a species-specific probe positive in one of eight replicate reactions. More recent sampling protocols dropped this two-step process and use all six assays on all samples (U.S. Fish and Wildlife Service 2022). Laboratory personnel also spiked each sample well with known concentrations of an internal positive control to test for inhibition. The internal positive control template addition included 1.0 µL of the novel mouse gene HemT diluted to 100 copies and subsequently quantified using HemT-specific standard curves. They assumed inhibition if they measured fewer copies of the spike in a sample, and they cleaned the extracts using a PCR clean-up kit (OneStep Inhibitor Removal kit by Zymo Research, Irvine, CA) and reamplified them (Asian Carp Regional Coordinating Committee 2015b; U.S. Fish and Wildlife Service 2015).
Statistical methods
Initially, we planned a single-occupancy modeling framework, such as the model presented by Dorazio and Erickson (2017), or a multispecies framework, such as the methods presented by Rota et al. (2016) or Devarajan et al. (2020), for this project. However, we had numerical convergence issues and difficulty interpreting model results. Instead, we used a generalized linear model (glm) to analyze the proportion of samples that contained at least one positive qPCR detection for eDNA (Bolker 2008). We specifically modeled the proportion of samples that were positive weighted by the number of samples using a quasibinomial model. We used a quasibinomial model to account for overdispersion in the data. We estimated coefficients for each waterbody, contrast coefficients for the bay habitat and shoreline habitat compared with the main channel habitat, a contrast coefficient for June compared with April, and a contrast coefficient for the second marker compared with the first marker for each species. During model construction, we explored interactions among coefficients, and none differed from 0. Thus, we did not include interaction terms in the final models. We constructed three glms, one for the bigheaded carp markers, one for the silver carp markers, and one for the bighead carp markers. We excluded Brandon Road Pool from all three models because there were no positive detections in this pool, thereby preventing glm estimates for this location. Likewise, we excluded Lockport Pool from the bighead carp marker and silver carp marker models because these markers had no detections in this location. The bighead carp and silver carp models exist within the context that we only ran the qPCR assays used to generate data for the models if the bigheaded carp assay detected eDNA. Statistically and synonymously, detection of bighead or silver carp eDNA also could be described as being conditional on a positive bigheaded carp marker detection. We used R version 4.4.0, including the core R glm() function, for our models (R Core Team 2024). We used ggplot2 version 3.5.1 for plotting our results (Wickham 2016). The data used in this article were published as a U.S. Geological Survey (USGS) data release (Peterman et al. 2024). The R code used in this manuscript has been released as a USGS software release (Peterman and Erickson 2024).
Results
A spatial pattern emerged for detections of bigheaded carp (Figure 2; Table 1). The farthest downriver location, Marseilles Pool, had the highest proportion of positive samples, with 0.19 (95 of 513) samples positive for the bigheaded carp marker, and of those positives, 0.66 were positive for the bighead carp marker (63 of 95), and 0.52 (49 of 95) were positive for the silver carp marker. The next location upriver, the Dresden Island Pool, had a proportion of 0.06 samples positive (41 of 642) for the bigheaded carp marker, 0.39 (16 of 41) positive for the bighead carp marker, and 0.51 (21 of 41) positive for the silver carp marker. Kankakee River had a proportion of 0.03 (9 of 309) samples positive for the bigheaded carp markers, 0.56 (5 of 9) positive for bighead carp, and 0.22 (2 of 9) positive for silver carp. Lockport Pool had 0.01 (5 of 336) positive for the bigheaded carp assay and none positive for the bighead carp and silver carp markers (both 0 of 5). Brandon Road Pool had no samples positive (0 of 339) for the bigheaded carp marker. The glm coefficient estimates for Marseilles Pool were also higher than those from the other pools across all three markers (Figure 3; Supplemental Reference 1).
Plot of the proportion of samples with environmental DNA detections (y axis) across habitat types (x axis) for two sampling months in 2015 (facet columns) for different locations in the Illinois River Basin (facet rows) from the U.S. Fish and Wildlife Service bigheaded carp Hypophthalmichthys spp. monitoring program. We used multiple environmental DNA markers (plot color), including generic bigheaded carp Hypophthalmichthys spp. markers (AC; then referred to as “Asian carp”), and confirmatory assays for bighead carp Hypophthalmichthys nobilis (BH) and silver carp H. molitrix sampling (SC). The bighead and silver carp detections in the plot are conditional on detection of generic bigheaded carp in a sample.
Plot of the proportion of samples with environmental DNA detections (y axis) across habitat types (x axis) for two sampling months in 2015 (facet columns) for different locations in the Illinois River Basin (facet rows) from the U.S. Fish and Wildlife Service bigheaded carp Hypophthalmichthys spp. monitoring program. We used multiple environmental DNA markers (plot color), including generic bigheaded carp Hypophthalmichthys spp. markers (AC; then referred to as “Asian carp”), and confirmatory assays for bighead carp Hypophthalmichthys nobilis (BH) and silver carp H. molitrix sampling (SC). The bighead and silver carp detections in the plot are conditional on detection of generic bigheaded carp in a sample.
Forest plot of coefficient estimates from generalized linear models (one per taxa for a total of three models; colors in the plot) with quasibinomial error family for generic bigheaded carp Hypophthalmichthys spp. markers (AC), bighead carp H. nobilis markers (BH), and silver carp H. molitrix markers (SC) from the U.S. Fish and Wildlife Service Monitoring program in the Illinois River Basin in 2015. We only tested samples that tested positive for a generic bigheaded carp marker for other species. The models predicted if a sample would have at least one positive marker detection per sample. We estimated an intercept for each waterbody. The habitats are contrasted observations from the main channel. Observations in June are compared with April. The SC5.HEX marker is compared with the SC4.FAM marker, the BH2.HEX marker is compared with the BH1.FAM marker, and the AC3.HEX marker is compared with the AC1.FAM marker.
Forest plot of coefficient estimates from generalized linear models (one per taxa for a total of three models; colors in the plot) with quasibinomial error family for generic bigheaded carp Hypophthalmichthys spp. markers (AC), bighead carp H. nobilis markers (BH), and silver carp H. molitrix markers (SC) from the U.S. Fish and Wildlife Service Monitoring program in the Illinois River Basin in 2015. We only tested samples that tested positive for a generic bigheaded carp marker for other species. The models predicted if a sample would have at least one positive marker detection per sample. We estimated an intercept for each waterbody. The habitats are contrasted observations from the main channel. Observations in June are compared with April. The SC5.HEX marker is compared with the SC4.FAM marker, the BH2.HEX marker is compared with the BH1.FAM marker, and the AC3.HEX marker is compared with the AC1.FAM marker.
Sampling locations within the river also had different proportions of positive detections (Figure 2; Table 1). Bay samples had the highest proportion of samples with detections of the bigheaded carp marker (0.15; 29 of 189), bighead carp marker (0.66; 19 of 29), and silver carp marker (0.66; 19 of 29). Main channel and shoreline samples had the same proportion of samples with a positive detection (0.06; 39 of 639 for the main channel samples and 82 of 1,311 for the shoreline samples) of the bigheaded carp marker. The main channel bighead carp marker samples were positive in 22 of 38 (0.56) samples, and the main channel silver carp marker samples were positive in 17 of 39 (0.44) samples. The shoreline samples for the bighead carp marker were positive in 43 of 82 (0.52) samples and for the silver carp marker in 36 of 82 (0.44) samples. The glm coefficient estimates for shoreline compared with the main channel were near 0, but the bay compared with the main channel coefficient was greater than 0 (Figure 3; Supplemental Reference 1).
A temporal trend also emerged. Samples from April had the highest proportion of positive samples (0.11 [116 of 1,065] for the bigheaded carp marker, 0.68 [79 of 116] for bighead carp marker samples, and 0.52 [60 of 116] for the silver carp marker samples). A low proportion (0.03; 34 of 1,074) of samples were positive for the bigheaded carp marker in June. A proportion of 0.15 (5 of 34) were positive for bighead carp and 0.35 (12 of 34) were positive for silver carp. The glm coefficient estimates for June compared with April were less than 0, and the 95% confidence interval did not contain 0; thus June samples were less likely than April samples to have positive detections (Figure 3; Supplemental Reference 1).
Discussion
We observed support for our first hypothesis for eDNA detections varying along a gradient. Specifically, a higher proportion of samples from the downriver location (with established, reproducing populations and also near the source population) had positive detections than observed in upriver locations (near the invasion front). This provides support for eDNA-based sampling methods to at least make relative comparisons in species density for this system with these species. This finding agrees with other eDNA-based samplings for the system (Coulter et al. 2018) and the broader eDNA-based sampling literature (Lacoursière-Roussel et al. 2016; Yates et al. 2019; Spear et al. 2020). For management or monitoring, this finding supports the use of eDNA-based sampling methods to observe spatial, and presumably temporal, trends for species management, especially at low densities. The information about sampling at low densities would be applicable for the Illinois River as well as other systems using the same or similar eDNA-based sampling and monitoring protocols for bigheaded carp, such as other USFWS samplings in the Mississippi River Basin and Great Lakes.
We observed support for our second hypothesis that bigheaded carp eDNA samples would have a higher proportion of positive samples in low-flow habitats, such as the bays of the Illinois River. Our findings demonstrate the importance of sampling different habitats within sites. Our observation that bay habitats had a higher proportion of samples with positive eDNA detections agrees with Glubzinski et al. (2021), who also observed that bays are the most preferred habitat for silver carp in the upstream sites of the Illinois River. Our within-site habitat comparisons supplement the work by Mize et al. (2019), who observed differences in habitat use across sites in the upper Mississippi River. Erickson et al. (2017) did not find differences in eDNA detections when sampling along transects in the Wabash River. However, the Wabash River differs greatly from the Illinois River. First, the Wabash River, especially the sampling sites visited by Erickson et al. (2017), is a smaller river without the diverse habitats of the Illinois River, such as backwaters, navigation channels, and tributaries. Second, the Wabash River is channelized, and the water in the Wabash River would be homogeneous compared with waters in the more heterogeneous Illinois River.
We observed support for our third hypothesis that the timing of sampling for eDNA affects detection results and inferences about species presence. These findings also agree with those from previous studies that found seasonal changes of eDNA detection for bigheaded carp in the Wabash River (Erickson et al. 2017) and the Mississippi River (Mize et al. 2019). Specifically, Erickson et al. (2017) found that bigheaded carp eDNA detection probability and site use varied across seasons, particularly in the most upstream site. Likewise, Mize et al. (2019) found that bigheaded carp change site usage through time, and detection probabilities within sites change across sampling time. This seasonal change in bigheaded carp habitat use agrees with research by Glubzinski et al. (2021) who observed silver carp changing habitat usage in the Illinois River. Coulter et al. (2019) also conducted eDNA sampling in the Illinois River and had overlap with our study sites; however, Coulter et al. (2019) focused on more downriver sites with higher bigheaded carp densities, and our study focused on upriver sites with lower bigheaded carp densities.
In conclusion, most invasive species will not have the same level of intense monitoring and control as bigheaded carp in the Illinois River. For example, ongoing efforts that either monitor or capture bigheaded carp in the river include eDNA-based surveillance, hydroacoustic-based surveys, netting, electrofishing, telemetry, and commercial harvest (Asian Carp Regional Coordinating Committee 2015d). Thus, our study provides opportunities and insight into invasive species monitoring programs that may not be present for other systems and their monitoring programs. Specifically, three insights emerged: 1) eDNA-based sampling could detect an invasion gradient in the Illinois River, 2) location and habitat within a site affects eDNA-based sampling, and 3) time of year matters for eDNA-based sampling. Therefore, species detected by eDNA-based monitoring programs may miss different species if they only target or select a few habitat types and lack temporal replication. This is especially true at locations with low species densities (for example, locations near an invasion front). Thus, less intense sampling at more habitat sites would be more likely to detect a species than intense sampling at fewer sites for species that vary their habitat usage.
Supplemental Materials
Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.
Reference S1. Peterman LL, Erickson RA. 2024. Analysis of eDNA gradient in Illinois River: U.S. Geological Survey software release. Reston, VA: U.S. Geological Survey.
Available: https://doi.org/10.5066/P9WUBWZM and https://doi.org/10.3996/JFWM-23-038.S1 (80 KB PDF)
Reference S2.Asian Carp Regional Coordinating Committee. 2015b. 2015 Asian carp monitoring and response plan interim summary report. Asian Carp Regional Coordinating Committee.
Available: https://doi.org/10.3996/JFWM-23-038.S3 (7.38 MB PDF)
Reference S3.Asian Carp Regional Coordinating Committee. 2015a. 2015 Monitoring and response plan for Asian carp in the upper Illinois River and Chicago Area Waterway System. Asian Carp Regional Coordinating Committee.
Available: https://doi.org/10.3996/JFWM-23-038.S4 (4.51 MB PDF)
Reference S4.Asian Carp Regional Coordinating Committee. 2015c. 2015 Asian carp control strategy framework. Asian Carp Regional Coordinating Committee.
Available: https://doi.org/10.3996/JFWM-23-038.S5 (4.37 MB PDF)
Reference S5.Asian Carp Regional Coordinating Committee. 2015d. 2021 Asian carp action plan. Asian Carp Regional Coordinating Committee.
Available: https://doi.org/10.3996/JFWM-23-038.S6 (6.07 MB PDF)
Reference S6.Invasive Carp Regional Coordinating Committee. 2022. 2022 Invasive carp action plan. Invasive Carp Regional Coordinating Committee.
Available: https://doi.org/10.3996/JFWM-23-038.S7 (4.92 MB PDF)
Reference S7. Kolar CS, Chapman DC, Courtenay WR Jr, Housel CM, William JD, Jennings DP. 2005. Asian carps of the genus Hypophthalmichthys (Pisces, Cyprinidae)—a biological synopsis and environmental risk assessment. Washington, D.C.: U.S. Fish and Wildlife Service.
Available: https://doi.org/10.3996/JFWM-23-038.S8 (4.15 MB PDF)
Reference S8.[USFWS] U.S. Fish and Wildlife Service. 2015. Quality assurance project plan: eDNA monitoring of bighead and silver carps. Washington, D.C.: U.S. Fish and Wildlife Service.
Available: https://doi.org/10.3996/JFWM-23-038.S9 (5.2 MB PDF)
Reference S9.[USFWS] U.S. Fish and Wildlife Service. 2022. Quality assurance project plan: eDNA monitoring of bighead and silver carps. Washington, D.C.: U.S. Fish and Wildlife Service.
Available: https://doi.org/10.3996/JFWM-23-038.S10 (4.62 MB PDF)
Acknowledgments
We thank all reviewers and editors for their feedback on drafts of this paper. We thank the USGS Biological Threats and Invasive Species Research Program for funding. We thank Austin Hannah for assistance with Figure 1 and early manuscript discussions. We also thank the USFWS Fish and Wildlife Conservation Offices, including the La Crosse, Wisconsin, and Carterville, Illinois, offices, and the Wilmington, Illinois, substation for collection of the data and the Illinois Department of Natural Resources for field support. The findings and opinions expressed in this manuscript are those of the authors and do not necessarily represent the views of the U.S. Fish and Wildlife Service.
Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.
References
Author notes
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.