Abstract
Northern Snakehead Channa argus an invasive freshwater piscivore discovered in the Potomac River in 2004, has spread throughout the Chesapeake Bay (Maryland, USA). The first incidental reports of Northern Snakehead in Blackwater National Wildlife Refuge (Chesapeake Bay) occurred during 2012. Since that time, Northern Snakehead have become established and has formed the basis for a popular harvest fishery in the Blackwater River drainage. This increase in abundance has caused concern about competition and predation on other species. To determine whether the fish community has changed in composition or relative abundance in Blackwater River drainage (Blackwater River and Little Blackwater River) since the introduction of Northern Snakehead, we replicated fish community surveys from 2006 and 2007 (pre-snakehead) and compared those fish community composition with data collected from 2018 to 2019 (post-snakehead). Subsequent seasonal surveys of fishes between 2021 and 2023 helped to substantiate our findings. Including pre-snakehead and post-snakehead survey periods, we caught 35 species (32 fish species and three invertebrate species) totaling 51,781 individuals. With few exceptions, species richness was similar between periods, with 27 fishes caught pre-snakehead and 26 species caught post-snakehead. However, of 22 species that we captured during both pre-snakehead and post-snakehead survey periods, 19 declined in relative abundance, which was supported during subsequent seasonal surveys. These changes led to four of six sites having significantly different fish communities between survey periods. Fish communities experienced declines in relative abundances of popular prey species for Northern Snakehead: White Perch Morone americana, sunfish Lepomis, and killifishes Fundulus. Our study is among the first to document long-term fish community changes following establishment of Northern Snakehead in its introduced range.
Introduction
Throughout the world, introduced fishes have been a significant source of concern where they become established (Cucherousset and Olden 2011). The focus of research for introduced species often includes investigation into control and management methods, determining current negative impacts to invaded ecosystems, and modeling potential future community and species changes because of the invader (Simberloff et al. 2013). Not all fish introductions lead to negative consequences (Cucherousset and Olden 2011; Martin and Valentine 2019), but some negative impacts have included: reductions in abundance of native fauna, loss of biodiversity (Gurevitch and Padilla 2004) and alteration of trophic dynamics (Britton et al. 2010, Gallardo et al. 2016, Kramer et al. 2019). As invasive fishes become abundant in an area, they could hinder resource use by native fishes, which can reduce biomass and growth of native fish (Carey and Wahl 2010; Hughes and Herlihy 2012; Kramer et al. 2019). Experimentally, Carey and Wahl (2010) showed that increasing density of invasive Common Carp Cyprinus carpio reduced native fish growth. In addition to declines in biomass owing to reduced growth, declines in biomass can occur because of predation. Invasive lionfishes Pterois volitans and P. miles in the Western Atlantic have been shown to cause declines in prey fish biomass and their competitors for those prey (Green et al. 2012; Albins 2013). Such changes in biodiversity and biomass contribute to community homogenization, which could reduce resiliency of the ecosystem to natural or human disturbances and significantly alter the evolutionary trajectory of native species within a community (Olden et al. 2004; Petsch 2016).
Northern Snakehead Channa argus is a freshwater fish native to parts of Asia (Courtenay and Williams 2004) but illegally introduced by the public multiple times to waterways of the Chesapeake Bay (United States). It was first reported from the Potomac River in 2004 (Odenkirk and Owens 2005) and has since spread throughout tidal waters of the Chesapeake Bay (Odenkirk and Owens 2007; Love and Newhard 2018; Fuller et al. 2019). Its introduction and subsequent spread have caused concern among state and federal agencies about its impact to ecosystems. To date, most studies have focused on competition with sportfish in the Potomac River (Saylor et al. 2012; Love et al. 2015) and modeling potential consequences (Love and Newhard 2012; Love and Newhard 2021). Love and Newhard (2021) used published consumption rates (Liu et al. 1998) and diet habits (Saylor et al. 2012) to calculate consumption for a single population of Northern Snakehead to nearly 2,000 kg/year of its principal prey, including primarily euryhaline species: sunfish Lepomis, Yellow Perch Perca flavescens, White Perch Morone americana, and Banded Killifish Fundulus diaphanus. Whether such levels of hypothetical predation change the composition of aquatic communities, however, remains unknown. We investigated differences in fish communities before and after the establishment of Northern Snakehead.
Northern Snakehead was first reported within the Blackwater River national Wildlife Refuge (eastern Chesapeake Bay watershed) in 2012 (Fuller et al. 2019; Figure 1). Reports of Northern Snakehead in the area coincide with their expansion in the eastern Chesapeake Bay watershed from illegal introductions (King and Johnson 2011; Wegleitner et al. 2016) in Delaware, where the species was first observed in 2010 (personal communication, C. Martin, Delaware Department of Natural Resources and Environmental Control). The abundance of Northern Snakehead in Blackwater River drainage, which includes Blackwater River and Little Blackwater River has increased since becoming established. Fishery dependent surveys using creel cards in 2020 and 2023 in Blackwater River showed kayak and boat anglers caught approximately one Northern Snakehead every two or three hours (Data S1, Supplemental Material) and support widespread recognition that Blackwater River drainage has become popular for harvesting Northern Snakehead (Ciekot 2019). Additionally, commercial harvesters, although limited in effort throughout the drainage by the Refuge’s status, began harvesting Northern Snakehead in 2015 and annual harvest has increased an order of magnitude since then (Data S2, Supplemental Material).
Map of sites for fish community surveys completed in 2006, 2007, 2018–2019, as well as subsequent seasonal surveys (2021–2023) in the Blackwater River (sites BC1, BW1, BW2) and Little Blackwater River (LB1, LB2, LB3; Chesapeake Bay watershed, Maryland). Solid black outline depicts approximate boundary for the Blackwater National Wildlife Refuge. Dashed outline and inset depicts approximate location of the water control weir built within Parsons Creek in upper Blackwater River to prevent saltwater influx from the Chesapeake Bay into upper Blackwater River.
Map of sites for fish community surveys completed in 2006, 2007, 2018–2019, as well as subsequent seasonal surveys (2021–2023) in the Blackwater River (sites BC1, BW1, BW2) and Little Blackwater River (LB1, LB2, LB3; Chesapeake Bay watershed, Maryland). Solid black outline depicts approximate boundary for the Blackwater National Wildlife Refuge. Dashed outline and inset depicts approximate location of the water control weir built within Parsons Creek in upper Blackwater River to prevent saltwater influx from the Chesapeake Bay into upper Blackwater River.
Fish communities of Blackwater River drainage had been inventoried prior to establishment of Northern Snakehead (Love et al. 2008; Newhard et al. 2012). This inventory occurred to provide a baseline for evaluating the restoration of the upper Blackwater River to freshwater with the goal of supporting a greater biomass of anadromous fishes. In the upper Blackwater River, a water control structure or weir had been completed in 2007 for the purpose of preventing saltwater of Little Choptank River from entering upper Blackwater River, a condition caused by digging a canal in the 1800s to connect both ecosystems. Love et al. (2008) documented upper Blackwater River as an oligohaline or brackish water ecosystem and Little Blackwater River as one of the last remnants of freshwater in the drainage. Though relatively few fish species inhabited the drainage, Love et al. (2008) documented use of Little Blackwater River as an important nursery for abundant freshwater-dependent species.
Long-lasting differences in composition of fish species in Blackwater River drainage (both presence or absence and relative abundance) could arise from progressive or intense environmental disturbances (O’Connell et al. 2004; Matthews et al. 2013; Sobocinski et al. 2013), such as the establishment of Northern Snakehead in the drainage and freshening of the upper Blackwater River owed to a weir. Predation by Northern Snakehead could contribute to declines in abundance of freshwater and euryhaline fishes (Choi and Kim 2021; Rohrback et al. 2023). Based upon risk assessments (Courtenay and Williams 2004) and ability of Northern Snakehead to lower relative abundances of fishes (Choi and Kim 2021; Rohrback et al. 2023), we hypothesized that: 1) drainage-wide declines in relative abundance occurred for fishes that are prey for Northern Snakehead, supporting theoretical expectations (Love and Newhard 2021); and 2) such declines would result in significant differences in fish community composition and biodiversity in the drainage before and after Northern Snakehead establishment. We also hypothesized that the weir structure might have helped lower salinity for upper Blackwater River, resulting in a greater abundance of anadromous or freshwater-dependent fishes in the upper Blackwater River.
Methods
Fish collection
We sampled six sites using consistent methodology for each site, three in Blackwater River and three in Little Blackwater River (eastern Chesapeake Bay, Maryland; Figure 1), areas that were inventoried before Northern Snakehead was introduced into the Blackwater River drainage (hereafter, pre-snakehead survey period; Love et al. 2008, Newhard et al. 2012; Data S3, Supplemental Material). Pre-snakehead survey period data for most sites were available monthly for two full years in 2006 and 2007, except for the third site in Blackwater River, Buttons Creek (BC), where data were only collected in 2007. Due to ancillary objectives to document anadromous fishes, we sampled more frequently (weekly) using the same sampling methods in April and May for 2006 and 2007. For analyses (see below), we randomly selected a single week in each month to represent a monthly collection for 2006 and 2007. Analyzing monthly data for these two years afforded the opportunity to examine monthly variability in community structure across two years during the pre-snakehead survey period. We also sampled sites monthly for one year from June 2018 to May 2019 (hereafter, post-snakehead survey period). Because the post-snakehead survey period included just one year and might not be considered representative of the post-snakehead survey period, we conducted additional surveys using the same methods at each site in spring (April or May), summer (July or August) and fall (October or November) between 2021 and 2023 (i.e., the subsequent seasonal survey period). We undertook this additional effort to determine if surveys conducted between 2018 and 2019 were representative of the fish community. Only seasonal sampling occurred between 2021 and 2023 because resources were not available for monthly sampling after 2019. Given the decline in sampling effort during subsequent seasonal survey period, we used these data to solely determine if species occurrence and relative abundances differed between the post-snakehead survey period and subsequent seasonal survey period.
At each site, we set fyke nets (1.25 cm mesh, 15.2 m lead line) for approximately 24 hours. Net set and pull times were recorded to standardize catch by hours fished (i.e., catch per unit effort or CPUE). We calculated CPUE per month for each species by dividing monthly abundances by the number of fyke net hours fished. Upon retrieving the nets, all fish were identified to species and individually counted. We measured Northern Snakehead for total length and placed specimens on ice for later dissection for ancillary prey inspection. To determine if water quality conditions changed between time periods, water quality was measured at each sampling event using a YSI Pro 2030. Water quality variables included: temperature (°C), dissolved oxygen (DO; mg/L), and salinity (psu; Data S4, Supplemental Material).
Analysis
We analyzed monthly CPUE as a measure for relative abundances for analyses. We computed an average CPUE for each taxon during the pre-snakehead survey period. This point estimate was compared with a sampling distribution of bootstrap averages determined for the post-snakehead survey period for each taxon (Haukoos and Lewis 2005). To create bootstraps, we drew randomly (with replacement) from CPUEs observed for each taxon during the post-snakehead survey period and generated a distribution of 1000 possible CPUE values (i.e., bootstrap sample). We replicated each bootstrap 500 times and determined the average CPUE for each of these replications. From this distribution of 500 bootstrapped statistics, we computed 97.5th and 2.5th percentiles. Because the survey conducted during the post-snakehead survey period was like that as during the pre-snakehead survey, we assumed that the data provided a reasonable distribution of bootstrapped statistics. We used this approach with a two-tailed test because relative abundance may have increased or decreased between survey periods. The observed average CPUE from the pre-snakehead survey period was compared with the percentiles and when it was lower than the 2.5th percentile, or greater than the 97.5th percentile, then the difference between the observed average and bootstrapped average (or effect) was deemed significant.
Variation in relative abundance of all species over time per site was also analyzed using nonmetric multidimensional scaling (NMDS). Each monthly sample of species’ relative abundances for each site was assigned a site score using separate NMDS analyses. Site scores were calculated using Sӧrenson (Bray-Curtis) distances that measured dissimilarity in ranked, relative abundances of species among monthly surveys. Prior to analysis, we relativized the data to maximum value per species to reduce the influence of very rare species. To relativize the data, monthly samples for two sites were removed because no fishes were collected (LB3, November 2007; LB1, May 2019). Additionally, during one sample (May 2006) at one site (BW1), just one Inland Silverside was collected, and this sample was considered an outlier and removed from the NMDS analysis. The NMDS analysis is a multidimensional ordination analysis that ranks and places site scores on k-axes to reduce stress of the final k-dimensional configuration (McCune and Grace 2002). The final number of dimensions was determined using a Scree Plot and we limited the number of axes for interpretation to the first two because they explained the greatest proportion of variance in the analysis (r2). In practice, final stress values ranging between 10-20 are common with community data and are generally considered acceptable (McCune and Grace 2002). We used the measure of final stress and a Monte Carlo significance test to evaluate the ordination. We used 250 runs of our data and 250 Monte Carlo simulations to assess the significance of obtaining an equal or lesser stress value than our observed final stress value. The final k-dimensional configuration was used to determine visually whether site scores produced for the pre-snakehead period differed from those produced for the post-snakehead period.
We learned which species were most correlated with scores from each NMDS axis by computing Spearman’s rank correlation coefficients (ρ) for monthly axis site scores and monthly CPUEs for the most abundant taxa: 1) catfishes (Amerius spp. and Ictalurus spp.); 2) Common Carp; 3) Gizzard Shad Dorosoma cepedianum; 4) killifish (Fundulus spp.); 5) sunfishes, crappies, and bass (Pomoxis spp., Lepomis spp., Micropterus spp.); and 6) White Perch. This analysis does not assign cause in the change of fish community structure but instead, helps identify the principal fishes associated with the change. A significance test of the ρ was performed to determine if it differed significantly from zero, which would indicate that the taxons’ relative abundance significantly correlated with the NMDS axis. Because water quality differences between periods may also influence community structure, monthly environmental data (water temperature, dissolved oxygen, and salinity) were correlated with site scores for all fish using ρ. Owed to relatively weak ( ρ <0.50) and insignificant correlations of the variables and site scores within a year, environmental variables reported here were not considered important predictors of changes in community structure. Spearman’s Rank correlations were conducted using SYSTAT for Windows (Version 13.0, SYSTAT Software, Inc.).
The NMDS analysis explored changes in fish community structure per site, and we specifically tested the hypothesis that aquatic communities differed significantly between pre-snakehead and post-snakehead periods using a multi-response permutation procedure (MRPP). The MRPP is a nonparametric analysis method designed to test the null hypothesis of no differences between communities of species (McCune and Grace 2002). A distance matrix between periodic fish communities based on ranked relative abundances of species was constructed using Sӧrenson distances. The within group distances were weighted by the relative size of the group. The magnitude of difference between groups (or effect size; A) was measured by the chance-corrected within-group agreement, which describes within group homogeneity compared to a random expectation. The within group distances were compared to that expected by random assignment of an observed relative abundance of a species to either a pre-snakehead or post-snakehead group. The randomized dataset was permuted 1000 times using a Monte Carlo test to determine the average within-group distance in scores for each group. The analysis provided a test statistic (T), a measure of effect size (A), and a P-value. The T is the difference between average within-group distances for observed and randomized datasets, divided by the square-root in variance of within-group distances. An A of 0 would indicate no distinct grouping effect and significant values of A < 0.1 commonly occur in community ecology (McCune and Grace 2002); given our modest number of monthly observations per site (N = 20 to 34), we regarded A > 0.01 as ecologically meaningful when statistically significant. All NMDS and MRPP analyses were conducted using PC-ORD (Version 7.07, McCune and Mefford 2018).
We also analyzed community structure by using rarefaction to compare two measures of biodiversity, species richness (i.e., the number of different species) and dominance (i.e., the fraction represented by the most common species) between pre-snakehead and post-snakehead periods for Blackwater River and Little Blackwater River. Because measurements of species diversity can increase with increasing abundance in the sample, we used rarefaction to standardize species diversity and compare it between survey periods at the lowest, common abundance (Sanders 1968; Hurlbert 1971). Given the limited biodiversity of estuaries such as Chesapeake Bay watershed, and thousands of individuals collected for each sampling time frame, it was assumed that collections were sufficient to represent available fish species. Rarefaction standardizes biodiversity metrics for differences in abundance. To do this, individuals were drawn at random from each community of organisms until reaching a user-specified number of individuals (i.e., lowest, common abundance). Biodiversity metrics were then computed for that level of abundance. This process was iterated 1,000 times to generate a mean and variance for each biodiversity metric. When iterations were finished, we compared metrics using 95% confidences. We determined if seasonal species richness or dominance differed between the post-snakehead period and subsequent seasonal survey by comparing rarefied estimates of diversity for the latter to that expected at the observed abundance on the rarefaction curves. Rarefaction and rarefaction curves (x-axis = interpolated abundance; y-axis = interpolated diversity index) were done using EcoSim (Version 7.0; Gotelli and Entsminger 2001).
As noted above, data from the post-snakehead survey period (2018 to 2019) were limited to a year. To determine whether this post-snakehead survey period was anomalous, we compared species occurrence and CPUEs from the post-snakehead survey period to the subsequent seasonal survey period (2021 to 2023). Because subsequent seasonal surveys were conducted seasonally, we averaged post-snakehead CPUE data between months in spring (April to May), summer (July to August), and fall (October to November) for each species and site to make appropriate comparisons. As above, we compared observed CPUEs from the post-snakehead survey period to the subsequent seasonal survey period using bootstrapped averages and percentiles for the latter. We also used MRPP (as above) to test the hypothesis that community structure differed between the post-snakehead survey period and the subsequent seasonal survey period.
Results
Across pre-snakehead and post-snakehead survey periods, we captured 35 species (32 fish species and 3 invertebrate species) totaling 51,781 individuals in Blackwater River drainage (Table 1). Water quality conditions were generally similar between both periods, though salinity across sites was lower between 2018 and 2019 than 2006 or 2007 (Figure 2). Annual averages of salinity across all sites were 5.2 psu (SE = 0.5) in 2006 and 2.8 psu (SE = 0.3) in 2007, freshening to 1.1 psu (SE = 0.2) between 2018 and 2019. While freshening was not isolated to the upper Blackwater River (or BW1; Figure 2), salinity measured in 2006 (7.2 psu; SE = 0.3) and 2007 (4.8 psu; SE = 0.7) at BW1 remarkably decreased following weir construction (2018 to 2019: 1.3 psu; SE = 0.2; also, 2021 to 2023: 3.0 psu; SE = 0.6).
Range of water quality data collected at all sites sampled within the Blackwater River (sites BC, BW1, BW2) and Little Blackwater River (LB1, LB2, LB3; Chesapeake Bay watershed, Maryland) from 2006, 2007, and 2018 to 2019. Water quality data collected included: salinity (psu; inset A), dissolved oxygen (DO, mg/L; inset B), and water temperature (°C; inset C).
Range of water quality data collected at all sites sampled within the Blackwater River (sites BC, BW1, BW2) and Little Blackwater River (LB1, LB2, LB3; Chesapeake Bay watershed, Maryland) from 2006, 2007, and 2018 to 2019. Water quality data collected included: salinity (psu; inset A), dissolved oxygen (DO, mg/L; inset B), and water temperature (°C; inset C).
List of all species captured at all sites and surveyed locations of Blackwater River drainage (Blackwater River and Little Blackwater River in Chesapeake Bay watershed, Maryland) in 2006, 2007, 2018, and 2019, ordered from most abundant to least abundant in 2006.

Prior to the introduction of the Northern Snakehead, the most abundant species (in rank order) were: White Perch, Brown Bullhead Amerius nebulosus, and Black Crappie Pomoxis nigromaculatus. In fact, the species that were 90% of annual catch were the same between 2006 and 2007 (White Perch, Brown Bullhead, Black Crappie, Pumpkinseed Lepomis gibbosus, and Bluegill Lepomis macrochirus). After the establishment of the Northern Snakehead, the three most abundant species (in rank order) were Common Carp, Gizzard Shad, and White Perch. Also, during the post-snakehead survey period, we captured 125 Northern Snakehead in fyke nets (Table 1). They were encountered at the same frequency (45% of months sampled; or 5 of the 12 months) in both Blackwater River and Little Blackwater River, though we collected an order of magnitude more fish in Little Blackwater River (N = 110) than Blackwater River (N = 15). Total length range of all Northern Snakehead captured ranged between 121 mm and 680 mm (average = 351 mm; SE = 7.0).
Most fish species had lower relative abundances in the post-snakehead survey period than in the pre-snakehead survey period (Table 2). Of the 22 species that were captured in both pre-snakehead and post-snakehead survey periods, 19 species exhibited declines in average relative abundance. The greatest declines were observed for Black Crappie, White Perch, Bluegill, Brown Bullhead, and Pumpkinseed. Killifishes were less abundant or not captured during the post-snakehead survey period. During the pre-snakehead survey period, three species of killifish had been captured: Banded Killifish, Mummichog F. heteroclitus, and Striped Killifish F. majalis. Neither Banded Killifish nor Striped Killifish were collected during the post-snakehead survey period and relative abundance of Mummichog significantly declined. Other species not encountered during the post-snakehead survey period included: Atlantic Silverside Menidia menidia, Largemouth Bass Micropterus salmoides, Longear Sunfish Lepomis megalotis, Redfin Pickerel Esox americanus, and Spot Leiostomus xanthurus. With exception to Northern Snakehead, only two species significantly increased in relative abundance between survey periods, Gizzard Shad and Common Carp (Table 2).
Monthly catches of fish species of Blackwater River drainage (Blackwater River and Little Blackwater River in Chesapeake Bay watershed, Maryland) analyzed between 2006 and 2007 (pre-snakehead period) and 2018 and 2019 (post-snakehead period). Average catch for each species of fish for each sampling period is provided with standard error (SE). The magnitude of difference (i.e., effect) in average catch between the pre-snakehead period and the post-snakehead period was tested for significance (noted * when P < 0.05) using confidence intervals developed from bootstrapped means.

Differences in species’ relative abundances between periods resulted in differences in community structure and biodiversity. Fish communities significantly differed between pre-snakehead and post-snakehead survey periods (N = 177; T = −11.83; A = 0.04; P < 0.01; Table 3). Significant differences in community structure were observed at one site in Blackwater River (BW1) and all sites in Little Blackwater River (Table 3). Differences in fish communities were largely correlated with differences in overall abundance and changes in abundance of dominant species, particularly White Perch and sunfishes (Table 4). Sites sampled during the post-snakehead period had species with lower overall abundances compared to the pre-snakehead period (see Tables 1 and 2). However, sites differed in the types of changes. Score separation along NMDS axes largely reflected differences between pre-snakehead and post-snakehead survey periods in White Perch and sunfish relative abundances (Figure 3). For BC, LB2, and LB3, similar scores among surveys during the post-snakehead survey period reflected similar monthly assemblages. Most sites had trends associated with lower relative abundances of White Perch and sunfish during the post-snakehead survey period, with exception of BW1.
Nonmetric multidimensional scaling (NMDS) analysis for fish community surveys conducted at sites in the Blackwater River (BC1, BW1, BW2) and Little Blackwater River (LB1, LB2, LB3; Chesapeake Bay watershed, Maryland) in 2006 and 2007, prior to introduction of Northern Snakehead Channa argus (PRE), and 2018 to 2019 (POST), after the introduction of Northern Snakehead. Multivariate data were structured along NMDS axes with stress, P-values, and r2 reflecting the ability of reduced dimensionality to represent multivariate space. Axes were correlated significantly with relative abundances of certain species: White Perch Morone americana, Gizzard Shad Dorosoma cepedianum, catfishes Ictalurus and Ameiurus, and sunfish Lepomis. Directionality of correlation is given as either increasing (inc.) or decreasing (dec.) along the given axis.
Nonmetric multidimensional scaling (NMDS) analysis for fish community surveys conducted at sites in the Blackwater River (BC1, BW1, BW2) and Little Blackwater River (LB1, LB2, LB3; Chesapeake Bay watershed, Maryland) in 2006 and 2007, prior to introduction of Northern Snakehead Channa argus (PRE), and 2018 to 2019 (POST), after the introduction of Northern Snakehead. Multivariate data were structured along NMDS axes with stress, P-values, and r2 reflecting the ability of reduced dimensionality to represent multivariate space. Axes were correlated significantly with relative abundances of certain species: White Perch Morone americana, Gizzard Shad Dorosoma cepedianum, catfishes Ictalurus and Ameiurus, and sunfish Lepomis. Directionality of correlation is given as either increasing (inc.) or decreasing (dec.) along the given axis.
Statistics from a multi-response permutation procedure (MRPP) that compared fish communities before the introduction of Northern Snakehead (Channa argus; 2006 and 2007) and afterwards (2018 and 2019) at each of the six sites in Blackwater River (BW or BC) and Little Blackwater River (LB; see Figure 1 for site location) in the Blackwater River drainage (Chesapeake Bay watershed, Maryland). The MRPP analysis a total number of observations (N), the test statistic (T), the effect of grouping by time periods (A) relative to no grouping, and the level of the effect for significance (P-value).

Scores from a nonmetric multidimensional scaling (NMDS) analysis correlated with fish species abundances (2006, 2007, and 2018 to 2019) from Blackwater River drainage (Chesapeake Bay watershed, Maryland). For each of six sites in Blackwater River (BC or BW) and Little Blackwater River (LB) within the drainage, NMDS scores correlated with monthly abundances for catfishes Ictalurus spp. and Amerius spp. (CA), Common Carp Cyprinus carpio (CC), Gizzard Shad Dorosoma cepedianum (GS), killifishes Fundulus spp. (KL), sunfishes Lepomis spp. (SN), and White Perch Morone americana (WP). Asterisks (*) denote coefficients that significantly demonstrate correlation between variables (P < 0.05).

Subsequent seasonal survey data supported the observed reductions in species relative abundance during the post-snakehead survey period (Table 5). Relative abundances for most species did not differ between post-snakehead and subsequent seasonal survey periods. However, thirteen species had significant changes in relative abundance since the post-snakehead survey period and communities differed significantly (T = −2.28; A = 0.02; P = 0.03). Of these species, ten experienced decreases in relative abundance: American Eel Anguilla rostrata, Black Crappie, Bluegill, Brown Bullhead, Common Carp, Gizzard Shad, Northern Snakehead, Pumpkinseed, Redbreast Sunfish Lepomis auratus, and White Perch. Both Banded Killifish and Mummichog were rare, being collected just one day and site each. Therefore, except for decreases in relative abundance of Common Carp, Gizzard Shad and Northern Snakehead, relative abundances from the post-snakehead survey period were deemed to be similar to and in some cases, further reflect continuing long-term declines of relative abundances in Blackwater River drainage.
Fishes of Blackwater River drainage (Chesapeake Bay watershed, Maryland) collected during seasonal surveys between 2018 and 2019 (post-snakehead period) and 2021 to 2023. Average catch for each species of fish for each sampling period is provided with standard error (SE). The magnitude of difference (i.e., effect) in average catch during the pre-snakehead period from that observed during the post-snakehead period was tested for significance (noted * when P < 0.05) using confidence intervals developed from bootstrapped means.

Species richness (S) for Blackwater River but not Little Blackwater River, was significantly greater before the introduction of Northern Snakehead than during the post-snakehead period (Figure 4). Species richness during the pre-snakehead survey period in Blackwater River averaged 26.7 (95% CI: 25–27) and decreased during the post-snakehead surveys of 2018 and 2019 to 21.0 (lowest common abundance = 4,602). In contrast, species richness in Little Blackwater River did not differ before or after Northern Snakehead establishment (Figure 4) and the number of species during the post-snakehead survey period in Little Blackwater River (S = 23.0; lowest common abundance = 3,313) was not different than the number of species for the pre-snakehead survey period (S = 22.0; 95% CI: 19-25).
Comparison of diversity measures (species richness (upper panel) and species dominance (lower panel)) for Blackwater River (right) and Little Blackwater River (left; Chesapeake Bay watershed, Maryland) before and after establishment of Northern Snakehead Channa argus. Rarefaction curves interpolated diversity at different levels of total abundance (x-axis) from that estimated at observed total abundance (i.e., the greatest abundance on the x-axis), which varied between Blackwater River and Little Blackwater River. Included in figure are the number of sampling events (n) analyzed for diversity measures and the error bars representing 95% confidence intervals.
Comparison of diversity measures (species richness (upper panel) and species dominance (lower panel)) for Blackwater River (right) and Little Blackwater River (left; Chesapeake Bay watershed, Maryland) before and after establishment of Northern Snakehead Channa argus. Rarefaction curves interpolated diversity at different levels of total abundance (x-axis) from that estimated at observed total abundance (i.e., the greatest abundance on the x-axis), which varied between Blackwater River and Little Blackwater River. Included in figure are the number of sampling events (n) analyzed for diversity measures and the error bars representing 95% confidence intervals.
Changes in the number of species between pre-snakehead and post-snakehead survey periods did not necessarily represent extirpations. During the subsequent seasonal survey period, we collected species that had been collected during the pre-snakehead survey period, but not between 2018 and 2019. Of those species, we collected: Banded Killifish (N = 1), Bluespotted Sunfish Enneacanthus gloriosus (N = 1), Channel Catfish Ictaurus punctatus (N = 1), Hogchoker Trinectes maculatus (N = 1), Largemouth Bass (N = 4), Mummichog (N = 1), and Silver Perch Bairdiella chrysoura (N = 1). However, subsequent seasonal surveys did not collect Chain Pickerel Esox niger, Redfin Pickerel Esox americanus, or Longear Sunfish Lepomis megalotis. In addition to these significant absences, a new invasive fish was collected, Blue Catfish Ictalurus furcatus (N = 1), the first report of this species in the watershed. Therefore, since the establishment of Northern Snakehead in the Blackwater River drainage, most fish species have persisted in the drainage, but some new species (Blue Catfish) and absent species (e.g., Longear Sunfish and pickerel) characterize species-specific differences.
Species dominance (D) was significantly greater for the post-snakehead survey period for both rivers (Figure 4). Blackwater River had the largest change in species dominance, which increased from 0.38 (95% CI: 0.38-0.39) during the pre-snakehead survey period to 0.60 (lowest common abundance = 4,602) during the post-snakehead survey period. Species dominance calculated from subsequent seasonal surveys (D = 0.31) was lower than comparable surveys between 2018 and 2019 (D = 0.42; 95% CI: 0.41–0.43), suggesting a short period of such high species dominance. Species dominance for the fish community in Little Blackwater River likewise increased from the pre-snakehead survey period (D = 0.28; 95% CI: 0.20–0.29) to the post-snakehead survey period (D = 0.32; lowest common abundance = 3,313). However, subsequent seasonal surveys show similarly high species dominance (D = 0.38, lowest common abundance = 1,869; post-snakehead survey period: D = 0.36; 95% CI: 0.35-0.37).
Discussion
Our observations of fish communities following introduction and establishment of Northern Snakehead in historically freshwater nursery habitats of Blackwater River drainage generally support concerns over introducing Northern Snakehead (Courtenay and Williams 2004) and findings from other studies (Choi and Kim 2021; Rohrback et al. 2023). Major reductions in relative abundance occurred for several notable freshwater-dependent fish species via predator-prey interactions. Prior to Northern Snakehead introduction, there were high abundances (>1,000 individuals) of several species captured within a year. In fact, surveys completed in 2006 and 2007 were dominated by White Perch, killifishes, and sunfishes (Centrarchidae). Banded Killifish had been frequently caught before Northern Snakehead were introduced but were not observed in 2018 and 2019, and rarely caught in subsequent seasonal surveys. Killifishes were the numerically dominant prey items in Northern Snakehead guts sampled from Little Blackwater River (unpublished data, J. Thompson, Maryland Department of Natural Resources) and other river systems (Saylor et al. 2012). After Northern Snakehead establishment, reductions in relative abundance also occurred for White Perch, Brown Bullhead, Black Crappie, Bluegill and other sunfishes. In their native range, Northern Snakehead reduced abundance of Bluegill via predation (Choi and Kim 2021). Prey species such as White Perch and Bluegill also decreased in areas of Potomac River, whereas Northern Snakehead and Blue Catfish increased in abundance (de Mutsert et al. 2017). The observed reductions in relative abundances of perch, sunfish and killifishes had been also predicted from consumption modeling of Northern Snakehead from Potomac River (Love and Newhard 2021). Impacts differ across ecosystems, though. Despite frequent consumption of Bluegill by Northern Snakehead in impoundments of Virginia, reductions in relative abundance of Bluegill were not observed (Isel and Odenkirk 2019). Usually, researchers highlight impacts by predation but competitive displacement of other top predators by Northern Snakehead could also occur. Northern Snakehead could compete with Largemouth Bass (Saylor et al. 2012) if resources become limiting. While Love and Newhard (2012) reported that impacts owed to competition with Northern Snakehead were minimal in Potomac River, such impacts could occur for other species in ecosystems with highly limited resources.
The Blackwater River drainage was also affected by the construction of a water control structure or weir in 2007, which was intended to restore the upper Blackwater River to a freshwater habitat. Bessler and Whitbeck (2012) compared salinities prior to (2004) and post (2011) weir construction and showed a reduction of salinity of upper Blackwater River from 8 psu to 4 psu (May to December), but this did not increase freshwater plant recruitment. In support of our hypothesis, we also observed a similar reduction of salinity in upper Blackwater River; and although sunfish species increased in relative abundance at upper Blackwater River (BW1), we did not observe greater relative abundances of freshwater-dependent species, such as river herring (Alosa spp.). In addition to the weir, greater levels of precipitation likely reduced salinity for BW1 during the post-snakehead survey period, as they did throughout the drainage. Average monthly rainfall Blackwater National Wildlife Refuge in 2007 (64 mm/month; SE = 10.5) was less than during post-snakehead survey season (87 mm/month; SE = 13.3; data from Western Regional Climate Center, RAWS USA Climate Archive; accessed September 24, 2024). Despite the freshening of the drainage in 2018 and 2019, freshwater and euryhaline fish species generally showed declines in relative abundance throughout the drainage, a pattern observed during subsequent seasonal surveys; and two freshwater species, Redfin Pickerel and Longear Sunfish, were neither collected during the post-snakehead nor subsequent survey periods. The freshening of the upper Blackwater River during the post-snakehead survey period may be explained by the additive effects of the weir and precipitation. Between 2021 and 2023, the drainage varied between oligohaline and mesohaline conditions, with average salinity ranging between 2.4 psu (SE = 0.7) and 8.1 psu (SE = 1.0), and 3.0 psu (SE = 0.6) at BW1, specifically. Therefore, although lower salinity has persisted in the upper Blackwater River since weir installation, it has neither been restored to freshwater nor resulted in intended changes in freshwater plant or fish communities.
Additional introduced species present in Blackwater River drainage have also altered the environment. Common Carp, which are widespread in tidal freshwater habitats of Chesapeake Bay, are considered “ecosystem engineers” that can alter native habitats through consumption of aquatic vegetation, which can influence primary production, water clarity and nutrient cycling, among others (Carey and Wahl 2010, Cucherousset and Olden 2011). Because Common Carp was common during all survey periods, it is likely the ecosystem engineering effects owed to this species had long occurred prior to our work. Blue Catfish, another invasive fish predator in Chesapeake Bay, was collected for the first time in Blackwater River drainage in May 2023. Because Blue Catfish depredates macroinvertebrates and fishes (Schmitt et al. 2021), it could be an additional threat in the future for Blackwater River drainage. While Blue Catfish could not account for changes in community structure between pre-snakehead and post-snakehead survey periods, its novel occurrence in the drainage in 2023 will raise greater concern for fishes, particularly Gizzard Shad and anadromous fishes, and could confound any future studies of ecological impacts owed to Northern Snakehead.
Fish abundance for some species in Blackwater River drainage likely varied because of changes in commercial harvest. Commercial harvest of White Perch in Blackwater River drainage in 2006 was 237 kg, approximately 4-times less than in 2018 (837 kg; Data S2, Supplemental Material), possibly explaining some reduction in catch of White Perch during our post-snakehead survey. The harvest of White Perch ranked far lower than other commercial fishery targets during the post-snakehead survey. Between 2006 and 2018, commercial harvest increased from over 200 kg to over 24,000 kg primarily for: Gizzard Shad (20,806 kg); Channel Catfish (1,365 kg); Common Carp (1,219 kg); and Northern Snakehead (1,097 kg). Despite this higher level of harvest, our surveys reflected greater catches of Gizzard Shad and Common Carp in 2018. Therefore, we deemed the commercial fishery for White Perch negligible in explaining the observed declines in its relative abundance. Differences in abundance could also be explained by intense or progressive environmental disturbances, such as changes in watershed land use, water quality, and habitat availability. Land use changes in a watershed such as impervious surface development may reduce water quality for fishes (Uphoff et al. 2011). This disturbance is less likely a factor causing change in fish communities of Blackwater River drainage because land use has not greatly changed for the Blackwater River drainage (Dorchester County 2019). A significant portion of the Blackwater River drainage is currently owned and protected by the U.S. Government (M. Whitbeck, Blackwater National Wildlife Refuge; see also, Figure 1). We measured water quality during surveys and except for lower salinity that favored freshwater-dependent species in 2018 and 2019, we found no progressive changes in water quality across the study period. Other unmeasured habitat factors may have affected fish distribution, such as the availability of underwater structure. For a fish community in Potomac River, de Mutsert et al. (2017) noted increases in Banded Killifish abundance over 30 years because of increases in habitat quality and area of submerged aquatic vegetation (SAV). However, SAV does not occur at appreciable levels in Blackwater River and Little Blackwater River (personal observation, J. Newhard, U.S. Fish and Wildlife Service; Orth et al. 2018).
Most species that occurred during the pre-snakehead survey period also occurred after the establishment of Northern Snakehead. Though absent in collections during the post-snakehead survey period, brackish tolerant species (e.g., Spot; Atlantic Silverside) were later caught in subsequent seasonal surveys when salinity levels were more similar to levels observed in 2006 and 2007. Likewise, some freshwater species, Largemouth Bass and Channel Catfish, were not collected during the post-snakehead survey period, but were collected in subsequent seasonal surveys and reported by anglers in 2018 and 2019 (personal communication, J. Love, Maryland Department of Natural Resources). Widespread extirpations were not observed following introduction of Northern Snakehead, which is arguably the case with many invasive species (Gurevitch and Padilla 2004). Species of pickerel and Longear Sunfish, however, were not detected during post-snakehead or subsequent seasonal survey periods and it is unknown whether these species have become extirpated.
Once widely known for its being threatened by sea level rise (Rogers and McCarty 2000), Blackwater River drainage and specifically, the function of Little Blackwater River as a freshwater nursery (Love et al. 2008) has further diminished in part because of Northern Snakehead, and possibly will continue if Blue Catfish establishes a population. Invasive fishes cause widespread reductions in biomass once they become established (Albins 2013, Carey and Wahl 2010, Gallardo et al. 2016), principally through direct predation, but also competition or habitat alteration (Albins 2013, Pelicice et al. 2015, Kramer et al. 2019). Managing invasive species costs the United States over 100 billion dollars per year (Pimentel et al. 2005; Cuthbert et al. 2022). Cost-effective control options for invasive fishes in the Chesapeake Bay watershed have largely been limited to incentivized harvest (Love and Genovese 2019) for the purpose of reducing population size (Newhard et al. 2019; Hoff and Odenkirk 2019). Northern Snakehead supports limited commercial and popular recreational harvest fisheries (Love and Genovese 2019) and with sufficiently high removal rates of invasive fishes (Kramer et al. 2019), biomass of native fishes can increase (Abekura et al. 2004). These restorative actions to maintain biodiversity could help restore the original function of Blackwater River drainage as a nursery. But, just as the water control structure in upper Blackwater River has not restored its freshwater nursery function, it is possible that harvest of invasive fishes will likewise fail in its restorative value, leaving these intense environmental disturbances within the Chesapeake Bay to further homogenize and alter fish communities (Rahel 2007; Daga et al. 2015, Gallien and Carboni 2017; Bezerra et al. 2019).
Supplemental Material
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
Data S1. Data submitted during a creel card survey conducted in Blackwater River drainage in Maryland between May 2023 and November 2023 to determine catch and harvest rates of Northern Snakehead Channa argus. The first column is the timestamp or the date when the survey was submitted. The second column and third column tallied whether anglers fished the area and whether they caught a Northern Snakehead. The third column tallied how many hours were fished. The fourth and fifth columns quantified the number of Northern Snakehead caught and harvested, respectively. The final two columns reflected comments from anglers.
Available: https://doi.org/10.3996/JFWM-24-028.S1 (11.9 KB)
Data S2. Annual commercial harvest data reported to Maryland Department of Natural Resources from Blackwater River drainage in Maryland between 2006 and 2018. The first three columns provide the year, month, and day. The fourth column provides the gear type used to harvest the species. The fifth column is duration (in hours) of gear use; note that values of 0 represent reports with no duration reported. The sixth column displays the species harvested, and the last column provides the catch (in pounds).
Available: https://doi.org/10.3996/JFWM-24-028.S2 (71.2 KB)
Data S3. Fishes collected from Blackwater River and Little Blackwater River in Maryland between 2006 and 2007, 2017 and 2018, and 2020 through 2023. The first column represents the siteID, which is a numerical identifier for the row of information. The second column represents the station or site for Blackwater River (BW) or Little Blackwater River (LB) or Buttons Creek (BC). The third and fourth columns represent a numerical month, beginning with January as 1 and ending with December as 12, and the year, respectively. The fifth column represents effort as the hours a fyke net spent in the water at a site. The sixth and seventh columns represent the species caught (by common name) and the total catch of that species, respectively.
Available: https://doi.org/10.3996/JFWM-24-028.S3 (269 KB)
Data S4. Water quality measured in Blackwater River and Little Blackwater River in Maryland between 2006 and 2007, 2017 and 2018, and 2020 through 2023. The first column represents the siteID, which is a numerical identifier for the row of information. The second column represents the station or site for Blackwater River (BW) or Little Blackwater River (LB) or Buttons Creek (BC). The third through sixth columns represent water temperature (in degrees Celsius), dissolved oxygen (in milligrams per liter), ambient conductivity (in microSiemens), and salinity. The seventh and eighth columns note the date a fyke net was set at the site and the date that the net was pulled. The ninth column represents effort as the hours a fyke net spent in the water at a site. The final column highlights comments made by the biologists at the time.
Available: https://doi.org/10.3996/JFWM-24-028.S4 (66.5 KB)
Reference S1. Bessler AM, Whitbeck M. 2012. Stewart’s Canal Restoration Project Report Comparison of Pre- and Post-Construction Monitoring on Vegetation and Salinity. U.S. Fish and Wildlife Service, Blackwater National Wildlife Refuge, Cambridge, Maryland.
Available: (4.42 MB PDF)
Reference S2. Snakeheads go from pariahs to prized catch. Delmarva Daily Times (April 1).
Available: (101 KB PDF) and https://www.dorchestercountymd.com/zoning/comprehensive-plan/ (May 2024)
Reference S3. 2004. Snakeheads (Pisces, Channidae) – A biological synopsis and risk assessment. U.S. Geological Survey Circular 1251.
Available: https://doi.org/10.3133/cir1251 (7.6 MB PDF) and https://pubs.usgs.gov/publication/cir1251 (October 2024)
Reference S4. Dorchester County. 2019. 2019 draft Comprehensive Plan. Chapter 3-Land Use.
Available: (82.6 MB PDF) and https://dorchestercountymd.com/planning-zoning/comprehensive-plan/ (October 2024)
Reference S5. Fuller PL, Benson AJ, Nunez G, Fusaro A, Neilson M. 2019. Channa argus (Cantor, 1842): U.S. Geological Survey, Nonindigenous Aquatic Species Database, Gainesville, FL.
Available: (946 GB PDF) and https://nas.er.usgs.gov/queries/FactSheet.aspx?SpeciesID=2265 (October 2024)
Reference S6. Green SJ, Akins JL, Maljković A, Côté IM. 2012. Invasive lionfish drive Atlantic coral reef fish declines.
Available: (115 GB PDF) and https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0032596 (October 2024).
Acknowledgments
We thank the staff of Blackwater National Wildlife Refuge, especially Matt Whitbeck for coordination of boating and sampling activities. The authors were assisted in the field by many people, including but not limited to: Mike Mangold, Alexis Walls, David Savage, Tanner Stoker, Scott Catton, Andrew Furness, Shannon Amato, and Jason Hanlon. The authors extend sincere gratitude for the reviewers and the associate editor who aided in developing this manuscript and improving the quality of the work. This research was funded by the U.S. Fish and Wildlife Service, Maryland Department of Natural Resources, and the Aquatic Nuisance Species Task Force (Agreement Award F19AP00311).
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
This Online Early paper will appear in its final typeset version in a future issue of the Journal of Fish and Wildlife Management. 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.