Movements and Habitat Use of Silver Carp in the Arkansas and White Rivers Free

In the decades since they first established in the lower Mississippi River in the early 1980s, many have felt the impact of Silver Carp Hypophthalmichthys molitrix throughout the Mississippi River basin (Schrank and Guy 2002; Hintz et al. 2017). Biologists, anglers, and conservationists are concerned over reduced condition and declining numbers of Gizzard Shad Dorosoma cepedianum, Bigmouth Buffalo Ictiobus cyprinellus, Paddlefish Polyodon spathula, and many juvenile sportfish following Silver Carp establishment (Williamson and Garvey 2005; Irons et al. 2007; Sampson et al. 2009). Silver Carp grow to 20 kg or more and possess gill rakers well-adapted for their planktivorous diet, which helps them outcompete native planktivores (Hayer et al. 2014; Ghosal et al. 2016; Lu et al. 2020). The U.S. Fish and Wildlife Service classified Silver Carp are as an “injurious species” (USFWS 2007) due to potential competition with native species as well as risk to humans (Kolar et al. 2007). The jumping behavior of Silver Carp can cause physical injury to people and expensive damage to boats and equipment. As Silver Carp population densities are high in their invaded range, human injury risk can be quite high (Williamson and Garvey 2005; Sass et al. 2010; Stuck et al. 2015). Local, state, and federal governments spend millions of dollars annually to control and eliminate Silver Carp populations (Tsehaye et al. 2013). People may avoid recreation in established Silver Carp waters. Therefore, economic losses when Silver Carp establish in important fisheries or recreation hotspots is of concern (Wittmann et al. 2015). If Silver Carp individuals continue to migrate and establish new populations, these impacts and concerns may become even more prevalent.

Measures to limit the establishment of new Silver Carp populations often involve harvest and migration prevention. In the Illinois River, Illinois, USA, where the population exceeds 2500 Silver Carp/river km (Sass et al. 2010), the state implemented a commercial harvest program to reduce populations in an effort to mitigate damage to ecosystems (Sass et al. 2010; Garvey et al. 2012; Stuck et al. 2015). Love et al. (2018) reported that Gizzard Shad populations improved in relative weight (W_r) and adult catch per unit effort following the commercial harvest program in the Illinois River. Other states have also implemented similar harvest programs. To limit migration of Silver Carp, Illinois installed electrical barriers on the Chicago Sanitary and Ship Canal to protect Lake Michigan from invasion (Moy et al. 2011). These barriers have been relatively effective with up to 99% deterrence (Noatch and Suski 2012). However, Silver Carp environmental DNA has been collected in the canal upstream of the electrical barriers (Jerde et al. 2011). These findings demonstrate that while current control measures have impacted the fight against Silver Carp, they are not perfect and should continue to be improved upon.

Increased knowledge of Silver Carp habitat preferences and movement activity patterns in many river systems will benefit the effort to improve control measures for Silver Carp. Silver Carp exhibited preference for habitats with low current velocity and warmer temperatures found in backwaters of the Upper Mississippi River and Ohio River (DeGrandchamp et al. 2008; Calkins et al. 2012; Prechtel et al. 2018). Seasonal trends in movement vary among studies, as some observed greater movement rates in the spring or fall, while others reported the longest movements in the summer (DeGrandchamp et al. 2008; Coulter and Goforth 2011; Coulter et al. 2016). Umland (2022) found that movement rates of Silver Carp did not differ between day and night in Kentucky Reservoir and Barkley Reservoir, Kentucky, USA. By understanding habitat preference and seasonal and diel movement patterns within an invaded river, fisheries managers may be able to more effectively harvest Silver Carp. For example, passive gear, such as gill nets, would be effective and practical when Silver Carp occupy deep water habitat or are actively moving, while active gear, like boat electrofishing, would benefit harvest in shallow water and dense cover during low movement periods (Mehdi et al. 2021). On larger scales, studies of annual movements conducted in some of the same river systems have found that on average Silver Carp roam <20 km/year at a rate of <1 km/d but are capable of moving distances exceeding 400 km in studies at rates of 60–90 km/d (DeGrandchamp et al. 2008; Coulter et al. 2016; Prechtel et al. 2018). To combat rare extreme movements, managers may seek to deploy some barriers far upstream of current invasion fronts. However, if extreme movements are nonexistent, operation of far-away barriers would be inefficient and costly. This is why managers will benefit from understanding Silver Carp behaviors in their systems when employing control strategies.

While a large collection of literature exists on habitat and movement behaviors of Silver Carp, a majority of studies took place at more northern latitudes, such as the Upper Mississippi River, Missouri River, Ohio River, and tributaries. Silver Carp behaviors in the Lower Mississippi River and its tributaries are far less understood. The goal of this study was to quantify Silver Carp movement activity and habitat use in the Arkansas River and White River, two large tributaries of the Lower Mississippi River. The first objective was to quantify annual movements and residency of Silver Carp and compare between the lock-and-dam fragmented Arkansas River and free-flowing lower White River. We hypothesized annual movements would be different between the two systems due to the influence of the dams on the Arkansas River’s hydrology. We also hypothesized residency periods of Silver Carp would differ between the two rivers and between seasons due to dam influences and seasonal climate differences. To test the hypotheses, we implanted Silver Carp from both rivers with acoustic tags then monitored their movement for a one-year period using a stationary acoustic receiver array. The second objective was to compare diel movements and habitat use between the two rivers during each of the four seasons. We hypothesized movements would be greater in one of the rivers, during some seasons, as well as during either day or night. We also hypothesized habitat use by Silver Carp would be in proportion to the availability of different habitat types and would differ between river, season, and diel period. To test these hypotheses, we tagged Silver Carp in both rivers during each season with radio transmitters and manually tracked them over multiple 24-h periods to collect data for calculating diel (day/night) movement rates and habitat usage. We also wanted to look at relationships between Silver Carp size and movement. We hypothesized larger Silver Carp would have greater movements than smaller individuals. The increased knowledge of Silver Carp movement and habitat use behaviors in the Arkansas River and White River will add to the literature on Silver Carp behaviors at more southern latitudes as well as arm fisheries managers of these systems with better knowledge to control these populations.

The Arkansas River and White River are both large tributaries of the lower Mississippi River. These rivers empty into the Mississippi River at km 1501 and 1530, respectively, in southeast Arkansas. The Arkansas Post Canal marks the start of the McClellan-Kerr Arkansas River Navigation System and connects the lower Arkansas River to the lower White River. The navigation system separates the Arkansas River across the state of Arkansas into a series of pools through numerous lock-and-dams. All lock-and-dams contain numerous gates that approximate free-flowing river conditions when fully opened and remain partially opened during a large portion of the year. While multiple dams create Beaver, Table Rock, and Bull Shoals reservoirs on the upper White River, a low head dam at Batesville, Arkansas United States (35.7593°N, 91.6548°W) marks the beginning of a ∼ 480 km free-flowing river reach. In pools of the Arkansas River, water level differs between pools but usually only ranges 3–4 m annually between maximum and minimum within pools. By contrast, the free-flowing White River ranges 6 m between maximum and minimum levels annually. We monitored the free-flowing river reach of the lower White River as well as the ∼450 km of the Arkansas River contained within the state of Arkansas with acoustic receivers (Figure 1). We conducted radio-tagging and tracking within a single pool or reach in both rivers. In the Arkansas River, Pool 4 spans ∼32 km from dam to dam and contains an abundance of different habitat types (Figure 2). In the White River, tagging and tracking efforts occurred within a ∼32 km reach containing many of the same habitat types as Arkansas River Pool 4 from DeValls Bluff, Arkansas United States (34.7944°N, 91.4450°W) to near the Cache River confluence (34.7026°N, 91.3253°W) (Figure 3).

We conducted range testing of our acoustic receivers (Innovasea Systems Inc., Halifax, Nova Scotia, Models VR2tx, VR2W) on Pool 4 of the Arkansas River on December 8 and 9, 2020. The Pool 4 on those dates was approximately 9.6 m and the discharge averaged 919.1 m³/s (Lock and Dam 5, Matthew Moix, U.S. Army Corps of Engineers – Little Rock Office, personal communication). We deployed receivers using an anchor, rope, and buoy at 100 m increments out to 500 m from an Innovasea V16 range test tag. An Innovasea VUE software-equipped computer enabled us to observe detections of the tag. We performed a second trial at 200 m increments out to 1000 m. We estimated the maximum detection range of the receivers to be ∼700 m. We chose receiver deployment locations where river width was <450 m to increase detection probability of passing Silver Carp.

In total, we selected 14 sites for receiver deployment, with seven sites in each river (Figure 1). These sites covered the entire span of the lower White River and the Arkansas River within the state of Arkansas. There was no overlap in coverage between receivers, so multiple receivers could not detect a single Silver Carp simultaneously. The battery life of each receiver was ∼15 months. We placed one receiver above the Batesville low-head dam on the White River to monitor potential Silver Carp movement upstream past the dam. We also placed receivers around the Arkansas Post Canal to detect movement between the two rivers. We secured receivers with an eye bolt in a 7.62 cm diameter PVC pipe affixed to a 23 kg concrete block with U-bolts set in the concrete. We secured the receiver and block to a large tree onshore using 30–40 m of 6.35 mm diameter stainless-steel cable and stainless-steel wire clamps. We lowered receivers and blocks to the riverbed from a boat at water depths <10 m with a relatively flat bottom contour.

We generally followed the habitat classification system of DeGrandchamp et al. (2008) to delineate habitat types within each tributary with modifications. We also identified new habitats in the two rivers. The main channel habitat consisted of deeper portions of the mainstem contained between channel ledges. Channel border consisted of shallower, flat areas of the mainstem located between main channel ledges and river banks. We defined side channels as lotic separations from the mainstem by islands. Backwaters were lentic pools and tributaries adjacent to the mainstem. Dike field habitat was only present in the Arkansas River and consisted of lentic, mainstem areas downstream of wing dikes. Floodplain habitat was inundated land outside of the normal riverbanks above bankfull when rivers flooded. This classification of habitat accounts for a majority of available water in the Arkansas River and White River, as well as a variety of lentic and lotic, main-stem or off-channel habitats throughout the year. In Pool 4 of the Arkansas River, main channel, channel border, side channel, dike field, and backwater habitat are available year-round. Lake Langhofer is a large, connected oxbow lake that provides a majority of backwater habitat in Pool 4. A small amount of floodplain habitat (∼3 km²) is available during spring floods. The White River reach near DeValls Bluff contains abundant main channel and channel border habitat, as well as side channels, backwaters, and an expansive spring floodplain.

We conducted coarse estimates of proportion of each habitat type in the Arkansas and White rivers on a seasonal basis. We acquired the river stage readings of each river from USGS river gauges at Pine Bluff (Arkansas River) and DeValls Bluff (White River) during each sampling period. We identified Copernicus Sentinel (Scorpius Labs d.o.o., Cvetkova ulica 29, SI-1000 Ljubljana, Slovenia) satellite images of the river reaches taken while the rivers were at stages corresponding to river stages observed during our sampling periods. Because the Arkansas River is a run-of-the-river type system managed primarily for commercial navigation, stages do not vary substantially annually, with summer, fall, and winter stages being relatively similar (Figure S1). Thus, we chose to use the same satellite imagery to assess aquatic habitat types for all three seasons. We used this same approach during summer and fall in the White River for the same reason (i.e., minimal stage variation, Figure S1). We cross-referenced all satellite images with topographic maps (Garmin Navionics, Garmin Ltd. Olathe, KS USA) of the rivers to best estimate proportions of contour-dependent habitats, such as main channels and channel borders. In small portions of the White River, the river channel deviated from the topographic map, so down imaging sonar (Garmin, Garmin Ltd. Olathe, KS USA) supplemented habitat estimation at those sites. Down imaging sonar provided depth and bottom contour information needed to characterize habitat types when the rivers deviated from the maps. We overlayed 200 m × 200 m grids over satellite images of the two river reaches in ArcGIS Pro (ESRI, Redlands, CA, USA). We estimated proportional availability of each habitat type within each river during each season as the count of grid squares containing respective habitat types divided by the total count of grid squares for each river reach. It was difficult to quantify floodplain habitat in the White River during spring due to canopy cover in the surrounding watershed. Thus, we estimated floodplain habitat in the White River by counting grid spaces immediately adjacent to the White River main channel. Hence, the White River floodplain proportion during spring was a conservative proportional estimate of that habitat type.

We differentiated Silver Carp from Bighead Carp Hypophthalmichthys nobilis by their spongy gill rakers and ventral keel (Kolar et al. 2007). We captured all Silver Carp using a Midwest Lakes Electrofishing System (30–60 Hz, 40–60% duty cycle, 7–15 A, 250–375 V; Midwest Lake Management Inc., Polo, MO, USA). For annual monitoring with acoustic telemetry, we captured 52 Silver Carp in the Arkansas River and 53 in the White River. We captured most Silver Carp between June and September 2021, though we tagged additional fish in March 2022. We confined capture and surgery during the summer months to the mornings to avoid the hottest periods of the day. We captured and released all Silver Carp within the acoustic receiver array and no more than five individuals from any one location at one time to limit effects of schooling behavior. We captured and released Silver Carp for radio tracking within the study reaches during each of the four seasons. We used the dates of the solstices and equinoxes to define the four seasons. We captured and tagged five Silver Carp in each river reach during October –November 2021 (fall), February 2022 (winter), May 2022 (spring), and September 2022 (summer). We set sample size for each river-season combination (ex. Arkansas-Summer) at five to ensure even sample sizes between groups, as capture rates were considerably low during certain river-season pairs.

When a Silver Carp was caught, we placed it in a 100 L holding tank filled with river water. We monitored water in the tank throughout the day for temperature and dissolved oxygen (DO) concentration to minimize stress on the fish prior to surgery. We maintained appropriate DO in the tanks with an air pump and air stones. We controlled temperature by replacing water in the tank when temperature differed by >1°C from the river. We collected length and weight data for each Silver Carp prior to surgery to ensure transmitters would not exceed 1.75% of the fish’s body weight (Summerfelt and Mosier 1984).

We sedated Silver Carp using a Smith-Root Portable Electroanesthesia System (Smith-Root Inc., Vancouver, WA, USA) at 60 Hz, 25% duty cycle, and 150 V for 5 s. This setting successfully sedated individuals for the 6–8 min required for surgery. We reduced voltage to 125 V in highly conductive water. Once sedated, we moved Silver Carp to a V-shaped trough, where a small, modified bilge pump pumped water over the gills. We removed a small patch of scales from the ventral left side posterior to the pelvic fin (DeGrandchamp et al. 2008). We disinfected this area with Ovadine solution (Syndel Laboratories Ltd., Nanaimo, Canada) diluted to 100 ppm with distilled water. We made a 2.5 cm incision at the site into the coelomic cavity with a sterilized scalpel. For acoustic telemetry, we inserted an Innovasea V16–6H (95 mm, 34 g) acoustic transmitter through the incision. For radio surgeries, we inserted a hollow-point needle posterior to the incision site to enable the antenna of an F1835B radio transmitter (48 mm, 16 g, Advanced Telemetry Systems Inc., Isanti, MN, USA) to pass through the body wall away from the incision. We closed the incision using 3–4 simple interrupted sutures (Ethicon Inc., Somerville, NJ, USA, CP-1 absorbable). We treated the surgery site with a povidone iodine ointment antibiotic before moving the fish back to the holding tank to recover. Individual surgeries took a mean (±SD) 421 ± 60 s to complete. We marked recovery when Silver Carp were able to maintain an upright posture and resumed active swimming in the tank. We kept all Silver Carp in the tank for a minimum of 10 min to account for potential false recoveries. Mean recovery of Silver Carp lasted 644 ± 421 (SD) s. Once Silver Carp recovered, we released them back into the river and marked the GPS coordinates.

We began manual tracking of radio-tagged Silver Carp two days after tagging. We delayed tracking for one day based on the assumption that Silver Carp may behave abnormally in the first day post surgeries. Tracking efforts began at 0630 hours and continued for 24 h. During that period, we attempted to locate all five Silver Carp every 4 h. We repeated this process every second day until three full tracking days were completed. We began tracking at the release site of the five Silver Carp and expanded upstream and downstream until all five were located or the first 4 h period had expired. We concentrated subsequent tracking periods around the river sections of last known locations. We used a 4500 series receiver and omni-directional dipole antenna (Advanced Telemetry Systems Inc., Isanti, MN, USA) to first detect tagged Silver Carp. When we detected a transmitter, we used a 3-element Yagi directional antenna to refine the position of tagged Silver Carp. Tagged Silver Carp often swam away from the approaching boat noise, so we determined refined position when a signal strength ∼150 was attained with the Yagi antenna pointed straight down. Transmitters’ signal strengths ranged 150–160 directly adjacent to the antenna during open-air testing. We marked the location with a Garmin ECHOMAP Plus 75sv (Garmin, Ltd., Olathe, KS, USA) and recorded the time and habitat.

At intervals of 1–3 months, we retrieved acoustic receivers and offloaded data. Data included detection date, time, transmitter ID, receiver ID, latitude, and longitude. For detections to be considered valid, we required multiple continuous detections of a Silver Carp at a receiver. We used detection data to calculate four dependent movement variables: cumulative movement distance, average movement rate, linear home range, and residency period. We measured distances between Silver Carp release sites and receiver locations using a combination of ArcGIS (ESRI, Redlands, CA, USA), Google Maps (Google LLC, Mountain View, CA, USA), and river kilometer data provided by the U.S. Army Corps of Engineers. Cumulative movement distance was the total observed distance a Silver Carp traveled between its release and last detection. We calculated average movement rate by dividing the cumulative distance by the total days elapsed between release and last detection. Linear home range was the distance along the river from the most upstream to most downstream detection for each individual Silver Carp. If a Silver Carp occupied two rivers, the linear home range for that individual was the sum of the linear home ranges in the two rivers. This included the length of the Arkansas Post Canal if movement occurred between the Arkansas River and White River. Cumulative movement distance, average movement rate, and linear home range were colinear, so we included only average movement rate in further analysis. Residency was the period a Silver Carp was in continuous proximity of a single receiver. We assigned a 12 h gap in detections as the limit to separate residency periods, and most detection gaps exceeded 12 h. Average residency was the mean of all the residency observations for each individual Silver Carp.

We analyzed data from acoustic telemetry using multiple approaches. We compared Silver Carp length and weight between rivers using two-sample t-tests. Average movement rate was tested for normality. We used a Wilcoxon rank-sum tests for non-normal data (Harris and Hardin 2013) to test for differences between the rivers where fish were tagged on average movement rate. We calculated Pearson correlation coefficients to examine relationships between Silver Carp length or weight and average movement rate after log_e-transforming the data to better fit a normal distribution. We conducted these correlation tests separately for each river and collectively for all Silver Carp. We used a mixed-model ANOVA (Rogers and White 2007) to test for effects of river, season, and river × season interaction on average residency period for individuals located in the spring, summer, and fall seasons. We set a random effect of individual/season in the ANOVA test. We excluded winter residency data from the model as no detections were made in the Arkansas River during winter months.

We summarized and analyzed characteristics and movement rates of radio-tagged Silver Carp with several approaches. We calculated overall mean lengths and weights of Silver Carp by river and compared them between rivers using two-sample t-tests. We used times and GPS locations to calculate hourly movement rates by dividing the distance between successive locations of a Silver Carp by the time elapsed. We determined distance between successive locations through ArcGIS (ESRI, Redlands, CA, USA) as the shortest river distance connecting the two locations. We then calculated the average hourly movement rate per individual. We experienced technical issues during winter tracking in the White River that precluded us from sequentially detecting Silver Carp enough times to calculate reliable hourly movement rates for that season. We classified hourly movement rates calculated entirely between sunrise and sunset as day hourly movement rates. Likewise, we classified hourly movement rates calculated entirely between sunset and sunrise as night hourly movement rates. We were therefore able to calculate average day and average night hourly movement rates for most fish. We used two mixed model ANOVAs (Rogers and White 2007) to test for effects of season and diel period with a random effect of individual/diel period. We also tested for an interaction effect between season and diel period. Pearson correlations examined relationships between fish lengths and weights, and day or night average hourly movement rates. We did not include in analyses individuals that were not detected frequently enough to calculate a day and night average hourly movement rate for the individual.

We categorized each observation of habitat use as either a day or a night observation. For each river, season, and diel period, we determined the proportion of observations in each habitat. To compare Silver Carp habitat use to habitat proportional availability in both rivers, we consolidated habitat data for each habitat type in each river and averaged the proportional availability of each habitat type per river over the course of a year. We then conducted two chi-square goodness-of-fit tests (Proc Freq, SAS Institute Inc. 2013) to compare habitat use proportions to average proportion of availability for each habitat type in the Arkansas River and White River. We used a multinomial logistic regression (Proc Logistic, SAS Institute Inc. 2013, Douma and Weedon 2019) to test for effects of river, season, and diel period on the probability of Silver Carp occupying a habitat type. We did not compare dike field habitat locations between rivers as dike fields were not available in the sampled White River reach. We performed statistical tests in program R (R Core Team, Vienna, Austria) and SAS software (SAS Institute Inc., Cary, NC USA) using an α = 0.05 significance level.

We located 20 of the 52 acoustic-tagged Arkansas River Silver Carp and 28 of 53 White River Silver Carp at least one time during the study. Acoustic telemetry arrays used in other systems observed eight Silver Carp tagged in this study outside of our studied river regions (Table 1). We did not observe any movement upstream past the Batesville low-head dam or between rivers through the Arkansas Post Canal. We detected Silver Carp as far upstream in the White River as Oil Trough, AR USA and up to the David D. Terry Lock and Dam #6 on the Arkansas River (Figure 1). Two Silver Carp moved upstream past a lock-and-dam in the Arkansas River and four Silver Carp moved downstream past a lock-and-dam. The upstream dam passages by Silver Carp occurred in spring 2022. During the time periods of passage, the Arkansas river rose ∼0.6 m, and a majority of 18-m dam gates were open 1–3 m or open entirely (Matthew Moix, personal communication). We were relatively successful at relocating radio-tagged Silver Carp, with notable exceptions during the fall season and the White River during winter, when technical issues were had with the radio receiver (Table 2).

Lengths (N = 48, t = 5.99, df = 36.25, P < 0.001) and weights (N = 48, t = 6.18, df = 28.31, P < 0.001) of relocated, acoustic-tagged Silver Carp were significantly greater in the Arkansas River than in the White River. In the Arkansas River, tagged Silver Carp averaged 930 ± 50 (SD) mm and 10.7 ± 2.3 (SD) kg, while White River Silver Carp averaged 847 ± 41 (SD) mm and 7.1 ± 1.3 (SD) kg. Lengths (N = 32, t = 5.65, df = 24.5, P < 0.001) and weights (N = 32, t = 3.03, df = 27.1, P = 0.0053) of radio-tagged Silver Carp were greater in the Arkansas River as well. Mean (±SD) lengths of Silver Carp in the Arkansas River and White River were 959 ± 51 mm and 830 ± 67 mm, respectively. Mean (±SD) weights of Silver Carp in the Arkansas River and White River were 12.0 ± 2.3 kg and 9.0 ± 2.5 kg, respectively. While radio-tagged Silver Carp in the Arkansas River were generally larger, radio-tagged Silver Carp during certain seasons in the White River were as large or larger than some Arkansas River groups.

For all Silver Carp from both rivers, the mean of average movement rates was 1.78 ± 2.83 (SD) km/d. In total, Silver Carp averaged 233.4 ± 101.0 (SD) days between release and last detection. White River Silver Carp averaged 217.7 ± 101.1 (SD) days between release and last detection, while Arkansas River Silver Carp averaged 254.7 ± 99.4 (SD) days. A White River Silver Carp had the maximum observed average movement rate of 10.8 km/d. Average movement rates were variable and a majority were <2 km/d (Figure 4). Additionally, average movement rate data failed the normality test. Average movement rates were higher for Silver Carp tagged in the White River than in the Arkansas River (N = 48, W = 133, P = 0.0033). Length (N = 48, t = −2.04, df = 37, P = 0.049) was negatively correlated with log_e-transformed average movement rate for all Silver Carp but we did not observe correlations for either river.

The overall mean residency for the study was 18.9 ± 35.9 (SD) h. There was no effect of river (N = 8, Table 3) or season (N = 8, Table 3), and no interaction between main effects (N = 8, Table 3). Average residency periods of Silver Carp were similar between rivers and across seasons (Figure 5). A majority of Silver Carp had average residencies <10 h, with many Silver Carp residing near receivers for <1 h. The maximum average residencies observed ranged between 1–2 days across seasons (Figure 5).

Average hourly movement rates of individual Silver Carp ranged from 0.014 km/h to 2.564 km/h. The mean (±SD) of average hourly movement rates for all individuals was 0.232 ± 0.447 km/h, indicating most individuals had lower average hourly movement rates. The mean (±SD) of average hourly movement rates during the day was 0.151 ± 0.224 km/h, while the mean (±SD) of average hourly movement rates at night was 0.307 ± 0.574 km/h. The mixed-model ANOVA indicated there was a significant seasonal effect on average hourly movement rate in the Arkansas River (N = 18, Table 4). There was a significant interaction effect between season and diel period (N = 14, Table 5) on average hourly movement rate in the White River, as well as a significant seasonal effect (N = 14, Table 5). In both rivers, average hourly movement rates were noticeably higher in fall compared to the other seasons (Figure 6). Higher rates in fall would explain why location proportions were lower in the fall (Table 2), as Silver Carp moved around more compared to other seasons. In the White River, night average hourly movement rates were higher than day rates in the fall, but night and day rates were similar in spring and summer (Figure 6). Overall, average hourly movement rates were relatively low, with many fish moving <0.5 km/h. Length of Silver Carp was not correlated with day average hourly movement rate (N = 32, t = −2.01, df = 30, P = 0.054), but was negatively correlated with night average hourly movement rate (N = 32, t = −2.42, df = 30, P = 0.022). Silver Carp weight was negatively correlated with both day (N = 32, t = −3.30, df = 29, P = 0.003) and night (N = 32, t = −3.86, df = 29, P < 0.001) average hourly movement rates.

In the Arkansas River, Silver Carp did not use each habitat type in proportion to the average availability of the habitat type over the course of a year (χ² = 169.8, df = 5, P < 0.001). Overall, we located Silver Carp in backwater and channel border habitat at a higher proportion than those habitats were available, while Silver Carp underutilized floodplain, dike field, main channel, and side channel habitats (Figure 7). We also did not locate White River Silver Carp in each habitat type in proportion to each habitat’s availability (χ² = 213.3, df = 4, P < 0.001). In the White River, Silver Carp occupied backwater and side channel habitats to a greater degree than available, while Silver Carp occupied floodplain, main channel, and channel border habitats at proportions less than they were available (Figure 7).

The multinomial logistic regression determined that river (χ² = 86.2, df = 5, P < 0.001) and season (χ² = 145.6, df = 15, P < 0.001) significantly affected the probability of Silver Carp occupying a specific habitat type. Diel period had no effect (χ² = 3.89, df = 5, P = 0.57). We located Silver Carp in the Arkansas River more often in backwater, channel border, and dike field (not present in White River) habitats, while White River Silver Carp used more floodplain, main channel, and side channel habitats (Figure 8). Silver Carp used backwater habitat heavily in winter and spring seasons across both rivers compared to fall and summer and occupied main channel habitat more in the fall and summer (Figure 9). Silver Carp only occupied floodplain habitat in spring, as floodplain was only available during the spring season. Silver Carp occupied channel border and side channel habitat most often in the summer, while occupying dike field habitat in the Arkansas River rarely in fall, spring, and summer.

Few prior Silver Carp studies have specifically compared movement patterns between free-flowing rivers and fragmented rivers (e.g., Stuck et al. 2015), such as those with a lock-and-dam navigation system. Silver Carp in the free-flowing lower White River had significantly higher average movement rates than Arkansas River Silver Carp. This aligns with our alternative hypothesis that Silver Carp annual movement behaviors would differ between the two rivers. In contrast, reported movements of Silver Carp appear similar between the free-flowing Wabash River in Indiana and the lock-and-dam fragmented Illinois River in Illinois (DeGrandchamp et al. 2008; Coulter et al. 2016). The Wabash and Illinois rivers are located in a similar region and latitude, but the studies took place several years apart, so interannual variability in weather and river conditions could confound comparisons between the two studies. Lubejko et al. (2017) observed that Silver Carp moved past lock-and-dams most often through dam gates during open river conditions. We confirmed with the U.S. Army Corps of Engineers – Little Rock Office that open river conditions occurred during parts of the spring 2022 season. Gillespie et al. (2016) also concluded that upstream dam passage of Silver Carp was likely inhibited more than downstream passage. We observed a small fraction of Arkansas River Silver Carp moving between pools, with most dam passages directed downstream. Upstream passages occurred during periods when dam gates were opened substantially (Matthew Moix, personal communication). It is plausible that the lock-and-dams on the Arkansas River provide a potential barrier to upstream movement of Silver Carp that is not present in the White River, limiting movements and resulting in lower average movement rates in the Arkansas River. If dams in the Arkansas River inhibit Silver Carp movements, harvest may have a significant impact on populations if pools with large populations are extensively targeted.

The mixed-model ANOVA results found significant seasonal effects on hourly movement rates in both the Arkansas River and White River, as well as a season and diel period interaction effect in the White River. These findings support our hypothesis that hourly movement rates differ based on season, but not by diel period. Seasonal differences in movement by Silver Carp have been observed in other studies (DeGrandchamp et al. 2008; Coulter et al. 2016). Kolar et al. (2007) reported Silver Carp remained active throughout the year in the Missouri River, only decreasing activity at temperatures below 4°C. We never observed temperatures below 10°C in the White River or Arkansas River. As with this study, Umland (2022) did not observe a diel pattern in Silver Carp movements in Kentucky Reservoir or Barkley Reservoir. Overall, hourly movement rates were fairly low across all seasons and diel periods, so active gears like boat electrofishing may be effective across multiple seasons and throughout the day (Mehdi et al. 2021). Radio tracking occurred within a few days of surgical implantation of the radio transmitters. Low hourly movement rates could be associated with physiological changes or modified behaviors related to surgical stress, though these changes seem to return to normal within ∼24 h after surgery, depending upon species, anesthesia, and transmitter implantation method (Bridger and Booth 2003).

We reject our null hypothesis that Silver Carp habitat occupation would be proportional to the availability of habitat types. Silver Carp occupied backwater habitat in both rivers significantly more than backwater habitat was available and underutilized main channel habitat. Previous studies have also observed Silver Carp preference of backwater habitat and avoidance of main channel habitat (DeGrandchamp et al. 2008; Calkins et al. 2012; Prechtel et al. 2018). In contrast, our results support our hypothesis that habitat use would differ between rivers and seasons. Differences in the availability of dike field and floodplain habitats explain observed differences in the use of those habitats by Silver Carp. While Silver Carp did not use these habitats in proportion to their availability, Silver Carp used dike field and floodplain habitat more in the rivers where each habitat was more prevalent. Differences in backwater and side channel habitat between rivers is likely due to their prevalence near Silver Carp release locations, as Arkansas River Silver Carp were released near backwater habitat, and White River Silver Carp were often released near side channel habitat. Across seasons, Silver Carp used backwater habitat more in winter and spring, and main channel habitat in summer and fall. Degrandchamp et al. (2008) also observed a difference in habitat use between seasons in 2005, as Silver Carp avoided the main channel in spring, as well as backwater and main channel habitat in summer. Despite differences in use among seasons, backwater habitat use was high compared to other habitats throughout the year in the Arkansas River and White River. Attempted harvest of Silver Carp in the Arkansas River and White River should yield high catch per unit effort in backwater habitats occupied heavily by Silver Carp across all four seasons.

Contrary to previous work, we did not observe differences in seasonal residency of acoustic-tagged Silver Carp in either the White River or Arkansas River. We found no support for our alternative hypothesis that residency period would differ between the two rivers or among seasons. Some studies have observed higher occupancy and longer residency periods during the hot summer months (Coulter et al. 2017; Glubzinski et al. 2021). In these studies, residency was evaluated in lentic, backwater habitat. While the large detection range of the acoustic receivers could cover some nearby backwaters, we set receivers at pinch points in the lotic main stem. Research suggests that Silver Carp not only prefer backwater habitat, but they also tend to avoid main channel habitat (DeGrandchamp et al. 2008; Calkins et al. 2012; Prechtel et al. 2018). Radio-tagged Silver Carp showed a distinct preference for lentic habitats like backwaters, side channels, and dike fields, as relative frequency of use exceeded relative proportions of those habitats by 30–70% in some seasons. Many radio-tagged Silver Carp remained in those habitats across multiple days. Calculated seasonal residencies of acoustic-tagged Silver Carp in this study were mostly short in duration, which could indicate an aversion to main channel habitat if assessed residencies occurred in the main river channel. It is possible the Silver Carp in the Arkansas River and White River follow the “highway analogy” outlined in Junk et al. (1989), largely using the main river channel as a “highway” to travel between preferred backwater habitat patches. These results further highlight selection of lentic, off-channel habitats like backwaters as prime targets for Silver Carp harvest.

The detection of Silver Carp tagged in this study in tributaries of the Upper Mississippi River and Ohio River attests to a Silver Carp metapopulation (Levins 1969) in the greater Mississippi River basin. Data sharing through an unpublished acoustic telemetry data sharing tool that included our study and other telemetry studies throughout the Mississippi River basin made this documentation possible (C.A. Aldridge and E.C. Boone, U.S. Fish and Wildlife Service, personal communication). Silver Carp were initially discovered in the Lower Mississippi River, though subpopulations now occur throughout much of the Upper Mississippi River (Freeze and Henderson 1982; Lohmeyer and Garvey 2009). Extreme movements of a few individuals often link subpopulations of a species (Bennett 1987; Cooke et al. 2005). Other studies have observed maximum individual movement distances and rates that exceeded their sample averages by 20–30 times (DeGrandchamp et al. 2008; Coulter et al. 2016; Gillespie et al. 2016). In this study, the maximum total movements were 12.5 times and 6.8 times the average total movements for the Arkansas River and White River, respectively. Although detection outside of stationary acoustic arrays may be possible with mobile tracking, some distances are simply too great to cover with mobile tracking. Coordinating with other telemetry researchers may therefore be the best method for accurate estimations of Silver Carp movement in subpopulations of the Mississippi River and its tributaries.

We detected 38% of acoustic-tagged Silver Carp in the Arkansas River and 53% of Silver Carp tagged in the White River. We also located a high proportion of radio-tagged Silver Carp during most seasons, with one confirmed mortality. These detection percentages fall within the range observed in other studies. DeGrandchamp et al. (2008) reported detecting 20% of individuals tagged in the Illinois River in Spring 2004 and 80% of individuals tagged in fall 2004 and spring 2005. Coulter et al. (2020) reported a range of 15–85% detection of tagged Silver Carp in various pools of the Illinois River from 2013 to 2015. There are many potential factors affecting detection percentage, including mortality, small linear home ranges, and receiver line of sight. Across acoustic telemetry studies of fish, average tagging mortality was ∼11% (Klinard and Matley 2020). We likely experienced high mortality in our study, as tagging occurred during warmer months when Silver Carp have been found to be more stressed (Liss et al. 2013; Jeffery et al. 2019). Additionally, we released many Silver Carp >10 km from a stationary receiver. Linear home ranges <10 km would have resulted in no detections of these individuals. We estimated some linear home ranges to be <10 km, so we likely did not detect some individuals with small linear home ranges. Acoustic receivers must also have a clear line of sight between the receiver and transmitter to receive a transmission. We had multiple instances where debris was deposited around a receiver by river currents that may have blocked detections until the debris was removed. Increased mortality, small annual movements, and loss of receivers’ line of sight certainly prevented us from obtaining more detections of tagged Silver Carp and more accurate results.

The primary purpose of this research was to observe the movement and habitat use of Silver Carp over the course of a year, providing managers with insights into their movement patterns in the Lower Mississippi River in both altered (Arkansas River) and relatively unaltered (White River) systems. These data may inform harvest or other population control actions. We did not collect water temperature or discharge data while actively tracking fish. Passive tracking using acoustic receivers did not readily allow linking hydrography data and explicit fish movements. Therefore, a detailed examination of environmental cues and movement are beyond the scope of this study. Spring high flows (Erickson et al. 2016) and warming spring temperatures (Abdusamadov 1986) are considered cues for upstream movements associated with spawning. However, Coulter et al. (2013) suggest that such cues for spawning movements outside the species’ native range can vary considerably. The gage height of the Arkansas River varies less than gage height in an unregulated system like the White River. The McClellan Kerr Arkansas River Navigation System is managed for transportation and flood control. Thirty-year average gage height in May and June are about two feet higher than average gage height in other months. During the period of our study, average gage height during April and May was similar to the long-term average during April and May. Even if a spring pulse cued movement, the series of dams in the navigation system would have prohibited large movements in association with spawning. A fifty-six-year average of monthly discharge in the White River clearly shows a spring pulse, with discharge during April and May about 1.7 times the average discharge during the rest of the year. During our study period, however, discharge during the April–May period was 2.2 times the average discharge during the rest of the year and 1.5 times the long-term average for the April–May period. Hence, our observations of movement in the White River, especially during the spring, might be biased upward by the strong cue of rising water during the spring in 2022.

Our observations of movement and habitat use can inform management decisions. For example, we observed that Silver Carp prefer backwater habitat in both river systems. Removal efforts (Tsehaye et al. 2013, Cupp et al. 2021) might be more effectively targeted at backwaters in these two systems. These efforts in backwaters might be enhanced by using olfactory attractants, such as food attractants (Claus and Sorensen 2017) or pheromones (Sorensen et al. 2019). Attractants might be used in backwaters in conjunction with highly efficient removal approaches such as the unified method outlined in Li and Xu (1995). Targeting backwaters in the Arkansas River for large-scale removal efforts would eliminate the concern of interference with navigation in the main channel of the McClellan Kerr Arkansas River Navigation System. We observed fish originally tagged in the White River moving upstream into a tributary of the White River, as well as downstream to the Lower Mississippi River and then upstream into other tributaries during April and May. These movements at this time suggest that population dynamics models should consider Silver Carp in the Lower Mississippi River as a metapopulation, similar to the approach used by Tsehaye et al. (2013) for Silver Carp in the Illinois River. We observed little movement among pools of the Arkansas River Navigation System and no movement between river systems. This information could inform population models linked to movement probabilities, like the Spatially Explicit Invasive Carp Population model for the Illinois River developed by Coulter et al. (2018). The movements we observed between the White River and the Lower Mississippi River during spring could be influenced by changes to the operation of the Montgomery Point Lock and Dam. This dam features doors that rest on the riverbed during high flows, like in spring, and are lifted from during low flow periods, requiring the use of the Montgomery Point lock for navigation. If the doors were raised during the spring, it might minimize Silver Carp movement into the White River, similar to how movement is minimized year-round in the rest of the McClellan Kerr Arkansas River Navigation System by a series of dams.

Overall, this study demonstrates that subpopulations of Silver Carp in Arkansas exhibit large and small-scale movements that have many similarities with few minor differences from subpopulations in other regions of the greater Mississippi River basin. Furthermore, Silver Carp dispersal upstream in highly fragmented systems, such as the Arkansas River, has the potential to be considerably slower due to lock-and-dam barrier potential than in free-flowing systems, such as the lower White River. Movement rates on short time scales are fairly small but may vary depending on season. Additionally, lentic habitats appear to be largely preferred by Silver Carp in the Arkansas River and White River like in other populations. Managers interested in controlling Silver Carp in the Arkansas River and White River should look to harvest Silver Carp in more lentic habitats like backwaters or side channels with boat electrofishing, as it should yield high catch per unit effort as Silver Carp occupy these habitats year-round for long periods of time. Harvest efforts should be concentrated more in the White River, as Silver Carp in the Arkansas River were less mobile and appear to be less able to migrate throughout the river. Because Silver Carp in the White River moved around more, gill nets may also yield high catch per unit effort for mobile individuals moving throughout the system. We would recommend focusing capture and harvest efforts in the late summer–early winter, when water levels in both rivers are at their lowest, reducing the volume of water and concentrating Silver Carp populations in occupied habitats.

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. Raw data of acoustic-tagging and detection of invasive Silver Carp Hypophthalmichthys molitrix in the Arkansas River and White River during 2021–2022. We used this data to determine movement variables of average movement rate and average seasonal residency of Silver Carp. This data was collected as part of a study to quantify movements and habitat use of Silver Carp in the Arkansas River and White River using telemetry during 2021 and 2022.

Available: https://doi.org/10.3996/JFWM-23-066.S1 (42.4 KB XLSX)

Data S2. Raw data used to calculate diel movement rates and quantify habitat locations of Silver Carp Hypophthalmichthys molitrix tagged with radio tags in Pool 4 of the Arkansas River and DeValls Bluff reach of the White River in central Arkansas from fall 2021 to summer 2022. This data was collected as part of a study to quantify movements and habitat use of Silver Carp in the Arkansas River and White River using telemetry during 2021 and 2022.

Available: https://doi.org/10.3996/JFWM-23-066.S2 (44.4 KB XLSX)

Figure S1. Hydrographs of (A) Arkansas River Pool 4 and (B) White River at DeValls Bluff over the study period of diel movement of Silver Carp Hypophthalmichthys molitrix from October 2021 through September 2022. This was part of a broader study of Silver Carp movements and habitat use in the Arkansas River and White River in Arkansas, USA during 2021 and 2022. Dotted line represents action stage on the Arkansas River. Hydrographs were obtained from USGS gauges 07263650 (Arkansas – Pine Bluff) and 07077000 (White – DeValls Bluff) available at waterdata.usgs.gov (accessed 8/30/2023).

Available: https://doi.org/10.3996/JFWM-23-066.S3 (44.7 KB JPG)

Reference S1. Gillespie N, Glover D, Stewart J, et al. 2016. Telemetry of Asian carp in the Ohio River. US Fish and Wildlife Service 2016 Technical Report. Available: https://micrarivers.org/wp-content/uploads/2018/08/2016-Ohio-River-AC-Telemetry-Annual-Project-Report-Final.pdf (August 2023).

Available: https://doi.org/10.3996/JFWM-23-066.S4 (231 KB PDF)

Reference S2.USFWS. 2007. Summary of species currently listed as injurious under (18 U.S.C. 42) Lacey Act. US Fish & Wildlife Service. Available: https://www.fws.gov/node/266035 (August 2023).

Available: https://doi.org/10.3996/JFWM-23-066.S5 (1.01 MB PDF)

This project and preparation of the report was funded in part by the U.S. Fish and Wildlife Service’s Fish and Aquatic Conservation Program: Invasive Carp Control and Management (grant award number F20AP11494 (for Lower Mississippi River) and/or F20AP11538 (for Arkansas River and White River)), through a sub-grant agreement with Arkansas Game and Fish Commission. Methods described in the manuscript were approved by the University of Arkansas at Pine Bluff Institutional Animal Care and Use Committee under protocol UAPB2021-03. We thank the staff and graduate students at the University of Arkansas at Pine Bluff for their assistance in gathering field data. We also thank the Arkansas Game and Fish Commission for their assistance in retrieving receivers as well as providing materials for this project. Thank you to Matthew Moix and the U.S. Army Corps of Engineers – Little Rock Office for providing information on Arkansas River dam operations. Thank you to the people at Iowa State University, Missouri Department of Conservation, and Murray State University for sending us our data through the acoustic telemetry lookup tool provided by the U.S. Fish and Wildlife Service. We thank those who took time to review this manuscript. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. The Nebraska Cooperative Fish and Wildlife Research Unit is jointly supported by a cooperative agreement among U.S. Geological Survey, the Nebraska Game and Parks Commission, the University of Nebraska, the U.S. Fish and Wildlife Service, and the Wildlife Management Institute.

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.