Invasion of the Allegheny River in Pennsylvania by the spiny water flea ( Bythotrephes longimanus Leydig , 1860 )

The spiny water flea (Bythotrephes longimanus), has successfully established populations in North American reservoirs and lakes, but little is known of its movement between lake and river systems. Here we document the first longitudinal movement of this aquatic invader in the Allegheny River, Pennsylvania. During spring 2013 we collected samples from the tailrace of Kinzua Dam to the confluence of Oil Creek, Oil City Pennsylvania, a distance of 100 km. The spiny water flea was present along 24 km of the sampled reach, comprising from 10 to 30% of the total zooplankton assemblage. In addition we collected two species of Daphnia, the dominant taxon, and Leptodora kindtii. All three developmental stages of the spiny water flea were present in samples collected (Instars I, II, and III). This study presents one of the first documented expansions of this species from an impoundment to downstream riverine reaches. Its presence warrants further monitoring and appropriate strategies to curb its spread and to investigate its potential impact on the lotic food web.


Introduction
There is increasing concern among biologists regarding the ecological impacts of aquatic invasive species.Among the more difficult to detect are those not readily seen by the naked eye from field collections -that is, until large aggregations form.One such invader, the spiny water flea (Bythotrephes longimanus Leydig, 1860), has successfully established populations in North American reservoirs and lakes (Branstrator et al. 2006;Cullis and Johnson 1988;Kelly et al.2013;Lange and Cap 1986;USGS 2013a).First detected in Lake Huron (Bur et al. 1986), this species quickly spread to all of the Great Lakes (Bur et al. 1986;Lange and Cap 1986;Cullis and Johnson 1988;Evans 1988), a likely result of Euro-Asian ship ballast water exchanges (Sprules et al. 1990;Berg et al. 2002;Colautti et al. 2005).Since its initial discovery, the spiny water flea has moved inland to a number of reservoirs and lakes (MacIsaac et al. 2000(MacIsaac et al. , 2004;;Branstrator et al. 2006;Kelly et al. 2013), a probable consequence of recreational boaters (MacIsaac et al. 2004;Johnson et al. 2008).
Reaching 12 mm in length when sexually mature (Branstrator 2005), B. longimanus has been implicated as causing major shifts in resident zooplankton, phytoplankton, and macroinvertebrate community composition (Lehman and Cáceres 1993;Foster and Sprules 2009;Strecker et al. 2011;Yan et al. 2002Yan et al. , 2011)).By consuming native cladocerans, copepods, and rotifers, the spiny water flea directly competes with larval fish (Berg and Garton 1988;Coulas et al. 1988;Evans 1988;Vanderploeg et al. 1993).However, due to its abundance in some systems, it has also become a significant food source for a variety of fish species during various portions of their life cycles (Nilsson 1979;Bur et al. 1986;Makarewicz and Jones 1990;Branstrator and Lehman 1996;Amtstaetter 2000;McRae 2001) while deterring consumption by others with its long spine  (Garton et al. 1993;Barnhisel and Harvey 1995;Compton and Kerfoot 2004).As a result, evidence suggests that the spiny water flea can exert both top-down and bottom-up control of predator and prey populations (MacIsaac et al. 2004;Brown et al. 2012;Kelly et al. 2013).
Because of its ability to disrupt aquatic ecosystems, the spread of this species has been of concern to state and federal regulatory agencies in the United States.While the movement of B. longimanus among lakes has been well documented (Colautti et al. 2005), few studies have examined the fate of this species in lotic systems.Previous studies indicate that impoundments can provide suitable sites for colonization (Johnson et al. 2008;Brown et al. 2012), as recreational boaters move among systems -often unknowingly from infested to pristine waters.
The discharge of water from a reservoir colonized by B. longimanus could allow for downstream movement of the species.While downstream dispersal has been documented for some invasive zooplankters (e.g.Daphnia lumholtzi; Shurin and Havel 2002), we were unable to find studies that detailed the potential extent of downstream movement in B. longimanus.For this study we had three main objectives: 1) to document the presence of B. longimanus in the Allegheny River; 2) to describe the longitudinal distribution of spiny water fleas from the tailrace of Kinzua Dam (KD), which forms the Allegheny Reservoir (AR), to its farthest extent downstream; and 3) to determine the relative abundance of B. longimanus in the zooplankton assemblage.

Study location
In 1965, the U.S. Army Corps of Engineers (USACE) completed construction of one of its largest flood control structures near the Pennsylvania/New York border, Kinzua Dam (KD) (Figure 1).Designed to regulate discharge on the Allegheny River, KD created the nearly 86 km 2 AR (Cowell and Stoudt 2007).From the reservoir, water discharged through KD flows another 300 km west and south to its Pittsburgh confluence with the Monongahela River to form the Ohio River.Approximately 195 km downstream of KD is the first of a series of eight navigational lock and dam structures and their attendant pools.The AR, in addition to flood control, provides hydroelectric power and a variety of recreational benefits to the region.Because of its proximity to the Great Lakes and availability to boat access (Leung et al. 2006), it is logical to think that the spiny water flea would eventually be documented here (R. Hoskin U.S. Army Corps of Engineers, personal communication).However, because sluice gates on KD remain open during periods of low and high flow (median flow 85m 3 /s), the potential also exists for this invader to move directly into the AR system in Pennsylvania.Our study reach extended from the tailrace of KD to an area 2 km downstream of the AR's confluence with Oil Creek, a distance of approximately 100 km (Figure 1, Table 1).Within this reach we established seven sampling stations.

Sampling protocol and analysis
As part of our sampling protocol to collect larval paddlefish (Polyodon spathula Walbaum, 1792), we deployed paired 1000 µm mesh ichthyoplankton drift nets between May and June 2013, at mid-water (approximately 1-2 m deep) and benthic locations (up to 4 m deep) (Figure 1), as described by Braaten et al. (2009).Samples were taken under varying flow and temperature regimes, from 42 to 255 m 3 /s and from 9 to 16 o C, respectively (USGS 2013b); however, we did not attempt to quantify zooplankton in relation to flow or temperature as neither were recorded coincident with each sampling event.Tailrace sampling locations below KD were passively fished for 10 minutes.Here, we anchored our boat at fixed points and deployed the nets in a stationary fashion.However, because stations farther downstream did not provide the needed current to fish these nets, we switched to an active sampling method by towing the paired-drift nets through the water column for 10 minutes.
Collected material was retrieved and stored in 10% formalin for post-processing.Due to the large mesh size of our nets, it is likely that many small zooplankters (e.g., Bosmina spp.and other Daphnids) were excluded from our samples.
In the laboratory, zooplankton and ichthyoplankton were identified to the lowest practicable taxonomic level and enumerated.Here, we report the zooplankton results.The relative abundance of zooplankton species at each station was determined from three replicate samples collected at each of the seven sampling stations.From each sample, we enumerated all individuals found in a series of subsamples until the coefficient of variation for each species fell below 30%.Subsamples were collected using a Hensen-Stempel pipette with a minimum of four 5-10 ml subsamples, from each sample.B. longimanus individuals were counted by eye in gridded petri dishes, while all other specimens were examined and enumerated under a dissecting microscope (~25-40 magnification).In addition, we took a subsample of 100 spiny water fleas from each station to determine life cycle stage (Instar I, II, or III).All collected organisms were identified according to Hebert (1995), Therriault et al. (2002), and Haney et al. (2013).We used Analysis of Variance (ANOVA) to determine if significant differences in proportional species and developmental stage composition existed among sampling locations.Prior to conducting the ANOVA, data were tested for normality using a Shapiro-Wilks test.In order to meet the assumption of normality, the proportional species data were arcsine square root transformed.If differences were detected in the ANOVAs, we employed a Tukey test (p-value<0.05), to ascertain statistical significance.

Results
The relative abundance of spiny water fleas was significantly higher in samples collected closer to the tailrace of KD (Figure 2; Tukey Test.Pvalue < 0.05).Our station at Warren (nearly 11 km downstream from KD) appeared to represent the initial point of a downstream decline in the relative abundance of B. longimanus, which comprised 18% of the sample here.The discharge from Conewango Creek (from 113 to 395 m 3 /s) may have produced some dilution of the zooplankton assemblage at this station.Densities of the spiny water flea downstream of Betts Park were not significantly different (Tukey Test, p-value>0.05)from stations at Starbrick and Buckaloons, with spiny water fleas comprising about 10% of the total community composition (Figure 2).Instars I, II, and III were present (Figure 3) with the majority occurring as Instar III (Tukey Test.p-value < 0.05).Gravid females carrying developing embryos were also present.
No zooplankton taxa were detected from stations at Tionesta or Oil City, 65 and 100 km downstream from KD, respectively.In addition to the spiny water flea, we collected three other zooplankton species: Daphnia pulicaria Forbes, 1893, Daphnia mendotae Birge, 1918, and Leptodora kindtii (Focke, 1844) (Figure 1).Daphnia spp.were the dominant taxa collected, comprising over 80% of the total zooplankton abundance among four of our five sampling locations (Figure 2).By contrast, L. kindtii was collected from only two locations, below KD and at Betts Park, comprising < 2% of the total abundance at these stations (Figure 2).Proportional differences were detected among taxa across sampling stations (ANOVA, p-value = 0.016).Most of the B. longimanus (~80-90%) appeared to be in excellent condition when examined under the microscope.The remaining 10-20% had broken tail spines and other damage typically found upon examination of samples collected by plankton nets.

Discussion
Our results suggest that B. longimanus is present in the AR, at least 24 km downstream from KD.However, samples collected farther from KD contained fewer B. longimanus in relation to the native Daphnia species.There are at least two possible explanations for this result.First, B. longimanus may not have established in the Allegheny River, but may be present due to downstream movement from the AR.Mortality and dilution along the course of the river may then account for their decrease in relation to Daphnia.Alternatively, B. longimanus may have established in the Allegheny River, but changing environmental conditions along the course of the river may have favored Daphnia.Unfortunately, the condition of B. longimanus individuals in our samples did not allow us to differentiate between downstream movement and successful establishment.Most individuals examined under the microscope appeared to be in excellent condition, many carried brood pouches with developing embryos, and all three instars of B. longimanus were represented in our samples (Figure 3).More study will be needed to determine if spiny water fleas have actually established a selfsustaining population in the river.
As Johnson et al. (2008) suggest, reservoir construction and the conversion of lotic to lentic waters may ultimately facilitate the spread of invasive species across the landscape.The presence of B. longimanus in areas downstream of KD appears to support Johnson et al. (2008).However, the impact that this species has on the local fauna remains unknown.As other authors have reported, this invader can cause major shifts in zooplankton, phytoplankton, and macroinvertebrate community composition in introduced waters (Yan et al. 2002;Foster and Sprules 2009;Strecker et al. 2011).
Of immediate concern is the potential for this invader to cause ecological shifts at higher trophic levels.Gill net surveys performed concurrently with zooplankton surveys yielded 22 fish species (Argent and Kimmel 2013).Among these were paddlefish, a "Species of Special Concern" in Pennsylvania, which is an obligate planktivore during portions of its life cycle (Jennings and Zigler 2009;Phelps et al. 2009).As Pegg et al. (2009) report, the spiny water flea is established in waters known to contain paddlefish and may pose a threat to their preferred forage.
A review of the literature reveals a paucity of information regarding the exploitation of the spiny water flea as a forage base by native Allegheny River fishes (see Argent and Kimmel (2013) for a list of native fishes).It is likely that larval resident fishes, if not gape limited, would utilize all zooplankton species with varying degrees of preference.However, as previous studies have demonstrated, B. longimanus can reduce the richness, diversity, and abundance of zooplankton in an invaded system (Yan et al. 2001(Yan et al. , 2002;;Strecker et al. 2006).If zooplankton assemblages respond similarly in the Allegheny River, resident fish populations could be negatively impacted.
We believe this study documents for the first time, the presence of the spiny water flea in the Allegheny River, an inland river system.The potential for this aquatic invader to facilitate an ecological shift exists in the areas below KD.Future studies should attempt to quantify this species' seasonal responses to varying flow regimes and its impacts on the riverine food web.Toward that end, more sampling effort should target the reach between Buckaloons and Tionesta, as well as downstream of Tionesta Reservoir, as this impoundment may serve as another inland source of expansion for the spiny water flea in the AR.Our farthest downstream locations (Tionesta and Oil City; Figure 1) did not yield any zooplankton taxa, which may be an artifact of the high flow conditions experienced during sampling events.
With the implementation of an appropriate sampling design and gear, more precise estimates of zooplankton density and species composition can be derived.Perhaps most important is the need to determine not only the linear extent of the spiny water flea, but how its presence alters the resident zooplankton assemblage of the AR.Over the lifespan of the spiny water flea (several days to a few weeks) and the flows experienced during this collection period, it is reasonable to conclude that these organisms may have traversed the entire 24km reach below the reservoir; however, it is also possible that reproduction occurred over downstream passage (as evidenced by the presence of all life cycle stages) and descendants arrived at the farthest documented downstream point.If the spiny water flea can maintain selfsustaining populations over long distances in a free-flowing river as this study suggests, it may eventually utilize the downstream series of Lock and Dam navigational pools on the AR to extend its range into the Monongahela and Ohio rivers.

Figure 1 .
Figure 1.Map of study reach and sampling stations along the Allegheny River.Star on inset map of Pennsylvania denotes the approximate location of Kinzua Dam.

Figure 2 .
Figure 2. Mean percent community composition of zooplankton taxa among the five sampling stations.Error bars denote standard deviation about the mean.

Table 1 .
Allegheny River sampling stations.Geo-referenced locations are given in decimal degrees.Corresponding station location map is presented in Figure1.