Rapid geographic expansion of spiny water flea ( Bythotrephes longimanus ) in Manitoba , Canada , 2009 – 2015

The spiny water flea (Bythotrephes longimanus), an aquatic invasive zooplankton species native to Eurasia, was first recorded from Manitoba waters at the Pointe du Bois Generating Station on the Winnipeg River using a larval fish drift net with 950 μm mesh, on 18 July, 2009. Bythotrephes drift density upstream and downstream of the station was highly variable with maximum densities of 23.3 individuals/100 m (9 samples) in June 2010 and 9.4 individuals/100 m in June 2012 (60 samples). In August and October of 2011, Bythotrephes were identified from the stomachs of eight cisco (Coregonus artedi) collected from the South basin of Lake Winnipeg near the mouth of the Winnipeg River, indicating that the invader had become part of the local food web. Subsequent targeted sampling for Bythotrephes with ship-based vertical plankton net (76 μm mesh) tows at 65 offshore/pelagic sites in Lake Winnipeg during summer and fall of 2012 indicated that the invader had colonized the South basin by early August and had expanded its distribution into most of the North basin by early October. No individuals were captured at a site just past the lake outlet in the Nelson River. Densities of Bythotrephes in Lake Winnipeg ranged from 0 to 93.1 and 0 to 39.2 individuals/m among sites for the summer and fall sampling, respectively. The highest mean density (9.2 individuals/m) was observed for the South basin in the summer of 2012. Non-targeted kick-net sampling at Playgreen Lake, located on the Nelson River approximately 27 km north of the Lake Winnipeg outlet in August of 2012, confirmed the rapid northward expansion of Bythotrephes through Lake Winnipeg. This represents the first evidence for further downstream dispersal. However, as of August 2015, multiple kicknet and Ekman/Ponar grab samples from Cross and Sipiwesk lakes further downstream on the Nelson River have not captured any Bythotrephes.


Introduction
The spiny water flea Bythotrephes longimanus (Leydig, 1860), hereafter referred to as Bythotrephes, is native to large areas of Eurasia (Grigorovich et al. 1998;Strecker 2011), where it continues to expand its range (e.g., Ketelaars and Gille 1994), including into waterbodies not considered typical, high quality Bythotrephes habitat (MacIsaac et al. 2000).North-American lakes that support Bythotrephes populations were significantly larger, and had higher mean and maximum depths than those that did not; though Secchi disk transparency, which was one of the main variables separating lakes with (mean of 4.7 m) and without (1.7 m) Bythotrephes in Eurasia, did not significantly differ between the respective groups of North American lakes (MacIsaac et al. 2000).The presence of a low-light refuge from fish predation may also be important for the successful establishment of Bythotrephes in a lake (Branstrator et al. 2006;Young et al. 2009).
The dispersal mechanism(s) of Bythotrephes are not fully understood, but transport of resting eggs within sediments and ballast water of large ships are likely important pathways.Ballast water transport has been postulated as the invasion vector for Bythotrephes into North America (Sprules et al. 1990), where it was first recorded in Lake Ontario in 1982 (Johannsson et al. 1991).Over the next five years, Bythotrephes spread throughout the remaining Laurentian Great Lakes (Bur et al. 1986;Evans 1988;Cullis and Johnson 1988).By 1989By -1991, Bythotrephes had invaded inland lakes in southern Ontario, Canada (Yan et al. 1992) and in northern Minnesota, USA (Schuldt and Merrick, pers. comm. in Yan et al. 1992).As of 2015, Bythotrephes had been identified from 179 inland lakes in Ontario (Azan et al. 2015).
Invasions of Bythotrephes have caused changes in lake zooplankton communities, such as reduced diversity, abundance, and biomass (Yan et al. 2002;Boudreau and Yan 2003; Barbiero and Tuchman 2004;Kerfoot et al. 2016), and the displacement of native taxa (Weisz and Yan 2011).Bythotrephes also has been deemed responsible for ecosystem effects such as lower epilimnetic zooplankton productivity (Strecker and Arnott 2008) and reduced fish growth (Feiner et al. 2015).Some changes in ecosystem function associated with high Bythotrephes abundance have been linked to reduced ecosystem services, and economic losses amounting to approximately 100 million US dollars for a single lake (Walsh et al. 2016).
Bythotrephes was first documented in the Lake Winnipeg watershed at Saganaga Lake, Ontario in September of 2003 (USGS 2016a), and thereafter in several other northern Minnesota and northwestern Ontario lakes (Branstrator et al. 2006;Kerfoot et al. 2016).In August 2007 the species was recorded approximately 350 km to the west of Saganaga Lake in Lake of the Woods (USGS 2016b), the source of the Winnipeg River (Figure 1).No barriers exist for the natural dispersal of Bythotrephes from Lake of the Woods down the Winnipeg River into Lake Winnipeg and further into the Nelson River.Thus, downstream expansion of the invader from Lake of the Woods was to be expected.
We report here on the first records of Bythotrephes in Manitoba, its rapid expansion throughout Lake Winnipeg, the 10 th largest lake in the world by surface area, and further downstream into the upper reaches of the Nelson River.

Material and methods
Collections of Bythotrephes came from targeted and non-targeted sampling for the species.All captures of Bythotrephes from locations outside of Lake Winnipeg originated from non-targeted sampling that was primarily designed to capture fish early life stages with large drift traps, and/or benthic invertebrates with Ponar or Ekman samplers (henceforth referred to as grab samples) and by kick-netting.Non-targeted sampling for Bythotrephes within Lake Winnipeg included open water sampling of live individuals with kick-nets and grab samples.In addition, Bythotrephes was documented from the diet analysis of two zooplanktivorous fish species in Lake Winnipeg captured by trawling at several locations in the South basin of the lake (Olynyk 2013).Once Bythotrephes was documented from Lake Winnipeg in 2011, systematic and targeted sampling of the species in the lake was conducted with plankton nets starting in the summer of 2012.
Drift traps were deployed yearly between 2007 and 2012 with the exception of 2011.Traps consisted of a stainless steel frame with a 43 × 85 cm (0.3655 m 2 ) opening attached to a 3 m long, 0.95 mm mesh bag and a 9 cm diameter cod end with 0.95 mm mesh screen.Traps were deployed as pairs of surface (surface to 0.4 m depth) and bottom (4-13 m depth, depending on the depth of the water column) traps at several locations in the Winnipeg River upstream (maximum distance 2.8 km) and downstream (0.7 km) of the Pointe du Bois Hydroelectric Generating Station (GS).Traps were sampled after 24-hours.In situ water velocity was measured at set and at pull of each trap and was averaged for the calculation of drift density (see below).Velocity at the mouth of surface traps was measured using a Swoffer Model 2100 current velocity meter, and an AA Price-Type Model 6200 current meter was used for bottom traps.After the presence of Bythotrephes in the Winnipeg River was confirmed from samples collected in 2009, quantitative counts of Bythotrephes in drift trap catches were obtained in 2010 and 2012.Because early instars of Bythotrephes may evade a net with 0.95 mm mesh size, the counts (and resulting drift densities) are not fully quantitative, and represent a conservative estimate.In 2010, drift was collected at three sites downstream of the Pointe du Bois GS on 1 June, and at two sites upstream of the station on 9 June.In 2012, drift was collected at the same locations as in 2010, with an increase in the number of traps and consecutive sets to three 24-hour periods from 3-8 June.
Grab samplers consisted of petite Ponar and Ekman samplers, both measuring 0.15 × 0.15 m (0.0225m 2 ).2).The first record within the Lake Winnipeg watershed was in 2003 in Saganaga Lake, Ontario (350 km east of Lake of the Woods).
Samplers were deployed to collect benthic sediment in water depths of 5-10 m; Ponar samplers over coarse substrates (e.g., gravel and cobble), and Ekman samplers over soft substrates (e.g., sand and clay).The kick-net had a 0.5 m long bag (400 µm mesh) with a cod end and was used to kick and sweep the substrate along perpendicular transects from the shoreline up to 1 m water depth.At each sampling location, 15 grabs and 15 kick-nets were executed, three of which were combined to create a total of five replicate grab or kick-net samples.Grab and kick-net sampling at Pointe du Bois was conducted in the forebay, approximately 11 and 8 km upstream of the station, respectively; grab sampling started in 2009, kick-net sampling in 2010.
Targeted sampling of Bythotrephes in Lake Winnipeg was conducted in 2012 during the summer (13 July-9 August) and fall (13 September-10 October) cruises of the research vessel "Namao" operated by the Lake Winnipeg Research Consortium.Sampling was conducted at 65 sites, 26 in the South basin, 12 in the Narrows, and 25 in the North basin, including one site (NWL) where the lake outflow creates the Nelson River (Figure 2).At each sampling site, two vertical zooplankton hauls were taken by pulling a conical zooplankton net with 25 cm diameter and 76 µm mesh size through the water column, starting approximately 2 m off the lake bottom, as determined by an International Offshore model Datamarine depth finder.The two vertical hauls at each site were combined into a composite sample.Density (individuals/m 3 ) of Bythotrephes at each station was determined from total volume of water filtered.Plankton net samples were preserved in 70% ethanol.
Grab, kick-net and drift trap samples were preserved in 4% formaldehyde solution.Ekman and Ponar samples were processed using sieves with 500 µm mesh.Drift samples were divided into subsamples using a Folsom Plankton Splitter.Each fraction was sorted until 300 organisms were counted.When the 300 organism count was achieved, the remaining organisms were counted so that a known fraction of the original sample was completely sorted.Total abundance of Bythotrephes for the whole sample was estimated by multiplying the split factor by the number of individuals counted in the split portion(s).Drift densities (individuals/100 m 3 ) were calculated for each drift trap set based on the sampling area of the trap, the duration of the set, and the average water velocity (m/s) following the formula: Drift density = n × 100/(t × a × v), where: n = number of individuals; t = set duration in seconds, a = mouth area of sampler (0.3655 m 2 ), and v = water velocity in meter per second.Mean drift densities did not differ statistically for locations upstream and downstream of the Pointe du Bois GS in 2010 (Kruskal-Wallis test, H = 0.54, df = 1, P = 0.56) and 2012 (Kruskal-Wallis test, H = 0.87, df = 1, P = 0.35) and were pooled for further analysis.As data failed to meet normality and/or homogeneity of variance assumptions, differences in densities between upstream and downstream locations, surface and bottom traps and between years were ascertained by Kruskal-Wallis one-way analysis on ranks.All analyses were conducted using the SigmaPlot 11.0 statistics package.

Occurrence and geographical expansion of Bythotrephes
Bythotrephes was recorded for the first time in the province of Manitoba on 18 July, 2009 when 18 individuals were recovered from a bottom drift trap set at the Pointe du Bois GS on the Winnipeg River (Table 1; Figure 1).This location is approximately 260 km downstream of the south end of Lake of the Woods, where Bythotrephes was reported in the summer of 2007 (Figure 1).Grab samples taken upstream of the station on 18 September 2009 did not yield any Bythotrephes (Table 1).In June of 2010, several thousand individuals of the invader were obtained from drift trap captures (Table 1).Again, no Bythotrephes were found in the kick-net and grab samples taken in the Pointe du Bois forebay on 15 September, 2010.Drift trap sampling was not conducted in the Winnipeg River in 2011.However, small numbers of Bythotrephes were recorded from the kick-net (5 individuals) and grab (4 individuals) samples taken in the Pointe du Bois reservoir on 15 September (Table 1).On 17 September 2011, a single Bythotrephes was found in a kick-net sample from the Winnipeg River at Lac du Bonnet, approximately 70 km downstream of Pointe du Bois GS (Figure 1), where the species had not been documented in kick-net or grab samples in the previous two years (Table 1).Also in 2011, several Bythotrephes were retrieved from the stomach of eight cisco (Coregonus artedi La Sueur, 1818) captured in the summer (1 August) and fall (27-28 September) from three locations (Stn 9, Stn W11, Stn 6), in Lake Winnipeg close to the mouth of the Winnipeg River, approximately 55-60 km downstream of Lac du Bonnet (Figures 1 and 2, Table S1).These findings represented the first confirmation of the presence of Bythotrephes in Lake Winnipeg and show that the invader had become part of the local food web.On September 25 and 27, 2011, four Bythotrephes were caught with plankton nets in the open water of Lake Winnipeg, very close to where some of the cisco were captured (Manitoba Sustainable Development, AIS Program, pers. comm., January 17, 2017).
The targeted sampling in the summer and fall of 2012 confirmed the incidental records of 2011 and clearly demonstrated the rapid expansion of Bythotrephes in Lake Winnipeg.In July and August, the species was found most abundantly near the bay at the mouth of the Winnipeg River, but also occurred in most areas of the South basin (Figure 2; Table S2).Bythotrephes was also captured at one site in the south end of the Narrows during summer, but not at any sampling site further north.The fall sampling approximately two months later, showed that Bythotrephes had dispersed throughout the Narrows and had colonized most of the North basin; no individuals were captured at the single sites in Mossy Bay near the north shore and the lake outflow forming the Nelson River in 2012 (Figure 2).However, in the same year, Bythotrephes was found further downstream in the Nelson River at Playgreen Lake approximately 27 km north of the outlet of Lake Winnipeg (Figure 1), in each of five kick-net samples collected on 16 August 2012.Per sample abundance was variable but generally high, ranging from 6 to 112 individuals.Previous kick-net and grab sampling at Playgreen Lake in August 2009 had not resulted in any captures of Bythotrephes (Table 1).
Non-targeted sampling also confirmed expansion of Bythotrephes into the north end of the North basin of Lake Winnipeg.On 26 August, 2014 relatively large numbers of Bythotrephes (n = 133-286 individuals) were obtained in the five kick-net samples taken from Mossy Bay (Figure 1; Table 1).In 2015, small numbers (1 to 4 individuals) of Bythotrephes were again collected in grab and kick-net samples at sites located in Lake Winnipeg (Mossy Bay) and Playgreen Lake (Table 1).No Bythotrephes were found in kick-net and grab samples conducted further downstream on the Nelson River at Cross and Sipiwesk lakes between 2012 and 2015 (Table 1; Figure 1).

Densities of Bythotrephes at Pointe du Bois GS and in Lake Winnipeg
The drift traps set in the Winnipeg River at Pointe du Bois in 2010 and 2012 that were counted quantitatively for Bythotrephes provided some indication of the number of individuals present in the surface and bottom layers of the water column.Drift densities varied between 0 and 22.3 individuals/100 m 3 for the 69 samples taken from five sites located upstream and downstream of the Pointe du Bois GS (Table 2), representing up to 10,800 Bythotrephes in a 24 hour net set.The mean density of surface traps was significantly higher than that of bottom traps in both years (Kruskal-Wallis test, H = 14.6, df = 1, P < 0.001), and the mean surface density in 2010 was significantly higher than in 2012 (Kruskal-Wallis test, H = 5.01, df = 1, P = 0.025; Table 2).The relationship between drift density and water velocity was explored for surface traps.In 2012, when velocities averaged 0.37 m/s (range: 0.06-0.75m/s), Bythotrephes was not found in traps set in areas where velocities exceeded 0.3 m/s (Figure 3).However, in 2010, when the only four available trap sets were associated with higher velocities (mean: 0.95 m/s, range: 0.27-1.32m/s) compared to 2012, drift densities of 5.4 and 23.3 individuals/100 m 3 were recorded at velocities of 1.0 and 1.3 m/s, respectively.
Densities of Bythotrephes based on vertical plankton tows from Lake Winnipeg in 2012 were also highly variable, ranging from 0 to 93.1 and 0 to 39.2 individuals/m 3 among sites for the summer and fall sampling, respectively.Considering the three different regions of the lake, the highest mean density

Discussion
While drift net and benthic grab sampling at Pointe du Bois in the Winnipeg River started in 2007, Bythotrephes was not detected until 18 July, 2009.These results suggest that Bythotrephes had expanded its distribution from Lake of the Woods (where first detected in August of 2007) into the Winnipeg River to Pointe du Bois as early as the spring of 2009.An earlier year for this expansion is unlikely based on the absence of Bythotrephes in the drift samples from the previous two summers.Although Bythotrephes are difficult to detect at low densities (Mills et al. 1992), it seems unlikely that they would have been absent in drift traps filtering hundreds of thousands of cubic meters of water and set at locations suitable for capturing the species in large numbers in 2010 and 2012.From Pointe du Bois, the invader seems to have rapidly expanded its distribution downstream via the Winnipeg River into Lake Winnipeg.Considering an average current speed of the Winnipeg River of 0.3 m/s, individuals that stay entrained in the main flow would reach Lake Winnipeg from Pointe du Bois within 5 days.Thus, the presence of Bythotrephes in the diet of cisco captured in Lake Winnipeg near the mouth of the Winnipeg River starting in early August of 2011, i.e. more than two years after first detection approximately 140 km upstream in the Winnipeg River, is not surprising.It is likely that the earliest records of Bythotrephes from Lake Winnipeg postdate the expansion of the invader into the lake by some time.This conclusion is indirectly supported by observations that selection of Bythotrephes by cisco increases once the species becomes abundant (Young et al. 2009).Furthermore, results from ship-based zooplankton net tows at several sites indicate that Bythotrephes occurred in the South basin of Lake Winnipeg at low abundance (2 individuals in each of two samples) and very close to the mouth of the Winnipeg River by the end of September 2011 (Manitoba Sustainable Development, AIS Program, pers. comm., January 17, 2017).This post-dates our first record for the South basin from cisco stomach content by approximately two months.In the summer (late July) of 2012, we found Bythotrephes in plankton tows at the southernmost of 12 locations in the Narrows between the north and the South basins of Lake Winnipeg, and by mid-September of 2012, the invader was first detected in the North basin of the lake.
Bythotrephes are reliably captured in standard plankton tows except when their abundance is very low (Boudreau and Yan 2004).However, plankton tows sample a much lower volume than drift traps and, therefore, false negatives are more likely to occur with plankton tows.Hence, the invader may have reached the lake almost two years prior to first detection in 2011, but abundance seems to have remained low with a distribution largely restricted to the area close to the Winnipeg River mouth.Population densities would have increased as a consequence of continuous seeding from the river and reproduction within the lake during the summer and fall of 2010 and 2011 in a similar manner as described by Sprules et al. (1990) for the Laurentian Great Lakes.
The available evidence suggest that Bythotrephes colonized Lake Winnipeg, which has a surface area of 24,514 km 2 and distance north to south of 431 km, probably within two years once population density near the seeding location had increased over the previous two years.This invader expanded its distribution throughout the Laurentian Great Lakes (approximately 244,106 km 2 ) within five years of its first detection in Lake Ontario, likely aided by secondary intercontinental invasions (Sprules et al. 1990) and dispersal by boats, bait buckets, and other anthropogenic factors (Weisz and Yan 2010).These vectors may have also contributed to the rapid colonization by Bythotrephes of the Winnipeg River and parts of Lake Winnipeg.The river and the South basin of the lake have a relatively high shoreline coverage with cottages (compared with the North basin), shown to be the strongest predictor of Bythotrephes presence in a model that included physical, chemical, and human use variables (Weisz and Yan 2010).
The mean densities of 2.2-9.2individuals/m 3 for plankton net tows from different areas of Lake Winnipeg are similar, but at the lower end of ranges of densities reported from plankton net tows at other lakes in North America.Walsh et al. (2016) reported fall densities of > 150 individuals/m 3 from Lake Mendota in Wisconsin (USA) using a combination of standard 30-cm diameter and larger 50-cm diameter nets.Barbiero and Tuchman (2004) recorded an average summer density for years 1986-1999 of 41 individuals/m 3 and 32 individuals/m 3 in the central and eastern basins of Lake Erie, and 13 and 10 individuals/m 3 in Lakes Huron and Michigan respectively.These estimates are based on 50-cm diameter and 64 μm mesh nets (1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996) and since 1997 using 153 μm mesh nets.Lower densities of 1-13 individuals/m 3 sampled using a 75-cm diameter net with 285 μm mesh were reported for 17 Canadian Shield lakes approximately 10 years after their invasion by Bythotrephes (Boudreau and Yan 2003).The relatively low densities of Bythotrephes in Lake Winnipeg may be indicative of its recent invasion and densities are expected to increase in the future, at least in some parts of the lake.However, we acknowledge that the low densities may, in part, be a consequence of using the standard zooplankton net instead of a modified Bythotrephes-specific net and of sampling only during the day (Yan and Pawson 1997;Armenio et al. 2017).
Although the number of Bythotrephes collected from kick-net sampling at Mossy Bay varied substantially between 2014 and 2015, the more than 1100 individuals obtained from the five samples in 2014 suggest a high abundance of the invader in the North basin of Lake Winnipeg.This conclusion is supported by recent presence-absence data from zooplankton net tows from identical lake locations as reported here for 2012; Bythotrephes is more regularly sampled from the North basin than the South basin (Manitoba Sustainable Development, AIS Program, pers. comm., 6 April, 2016).Differences in abundance of the invader between the two basins may be expected based on some of their limnological characteristics.The South basin is shallower (mean depth of 9.0 m) and has a lower Secchi disk depth (mean of 0.6 m) than the North basin (depth=13.3m, Secchi=1.4m; EC and MWS 2011), conditions that are perhaps not conducive to high abundances of Bythotrephes.Similar differences in the abundance of Bythotrephes between lake basins have previously been observed for Lake Erie, where the invader failed to establish in the more turbid western basin but thrived in the central and the eastern basins (Barbiero and Tuchman 2004).Lake Winnipeg is currently undergoing cultural eutrophication (EC and MWS 2011), particularly in the South basin, which may further reduce water clarity.Bythotrephes declined in Lago Maggiore (Italy) as the lake experienced eutrophication and increased in abundance subsequent to lake re-oligotrophication (Manca and Ruggiu 1998).Thus, unless nutrient loads (particularly phosphorus) into the lake (particularly the South basin) are reduced in the future, higher abundances of Bythotrephes may be limited to the North basin of Lake Winnipeg.
Bythotrephes has invaded the upper reaches of the Nelson River, and its further downstream dispersal along the approximately 410 km of river to the estuary into Hudson Bay seems inevitable.Over this course the Nelson River widens into several riverine lakes and is regulated by six hydroelectric generating stations with associated reservoirs.However, kicknet and grab samples taken at two of the lakes (Cross and Sipiwesk, see Figure 1), approximately 31 and 95 km further downstream on the Nelson River from Playgreen Lake, respectively, in 2014 and 2015 did not result in any captures of Bythotrephes.This does not necessarily indicate that Bythotrephes has not yet colonized these waterbodies; instead it may still be below detection limits.The Nelson River and most of its lake expansions and reservoirs may be a habitat less suited for abundant populations of Bythotrephes; the water is generally turbid with mean Secchi depths of 0.3-0.8m, only occasionally reaching Secchi depths of 1.9-2.4m in some waterbodies, as measured during summer and fall of 2008-2010(CAMP 2014)).Bythotrephes mainly occurs in large, deep, and clear lakes (MacIsaac et al. 2000;Branstrator et al. 2006), likely reflecting its strong reliance on visual feeding (Pangle and Peacor 2009;Jokela et al. 2013).
The absence of Bythotrephes in several grab and kick-net samples at locations in the Winnipeg River in years when drift trap sampling resulted in substantial catches of the invader indicated that the benthic samplers and kick-nets used in our study may have a reduced probability of capturing Bythotrephes, particularly at low abundances.The latter two capture methods are not used for targeted sampling of Bythotrephes and we found no other published studies with information on their capture success.However, the first record of Bythotrephes in Manitoba, its first occurrence in Lake Winnipeg, and the further northward dispersal of the invader was documented using methods that did not specifically target the species.While allowing only limited estimates on the abundance or density of Bythotrephes, these methods provided critical information on the geographical distribution of this ecologically and economically important invasive species covering a distance along the invasion route of more than 600 km.These findings emphasize the important contribution of non-targeted sampling methods for the early detection of aquatic invasive species, potentially allowing for a timely implementation of management strategies.
The invasion of Bythotrephes has been associated with alterations of the zooplankton community, such as displacement of native taxa (Weisz and Yan 2011) and lower diversity, abundance, and biomass (Yan et al. 2002;Boudreau and Yan 2003;Barbiero and Tuchman 2004).At the ecosystem level, Bythotrephes has been associated with reduced fish growth (Feiner et al. 2015) and lower epilimnetic zooplankton productivity (Strecker and Arnott 2008), resulting in a reduction in the availability of zooplankton biomass as prey for planktivorous fish (Foster and Sprules 2009;Walsh et al. 2016).Lake Winnipeg has also been recently (2013) invaded by the zebra mussel Dreissena polymorpha (Pallas, 1771).This mollusc is known to alter ecosystem function by, for example, increasing water clarity (Geisler et al. 2016) and increasing benthic production at the cost of pelagic production (Johannsson et al. 2000;Rennie and Evans 2012;Higgins et al. 2014).With both these keystone invaders expanding throughout the lake at the same time, the ecosystem of Lake Winnipeg (and downstream habitats) will likely undergo some rapid and profound changes over the next decade.Because of the complexity of potential physical and biological interactions, the direction and magnitude of these changes are difficult to predict.For example, Dreissena may actually increase water clarity in Lake Winnipeg, particularly in the South basin, despite opposing trends due to eutrophication, thus creating habitat conditions that would favor higher Bythotrephes abundance.Synergistic negative effects of Bythotrephes and Dreissena on zooplankton abundance and production (Azan et al. 2015;Walsh et al. 2016) may make it difficult to predict consequences on the planktivorous fish community, including young-ofthe year fishes (Vanderploeg et al. 2012;Azan et al. 2015), as well as other foodweb interactions.

Figure 1 .
Figure 1.Invasion history of the spiny water flea Bythotrephes longimanus in Manitoba waterbodies from non-targeted sampling.Years indicate the first record at a location.Note that targeted sampling in 2012 documented B. longimanus in Lake Winnipeg just south of Mossy Bay (see Figure2).The first record within the Lake Winnipeg watershed was in 2003 in Saganaga Lake, Ontario (350 km east of Lake of the Woods).

Figure 2 .
Figure 2. Distribution and density of Bythotrephes longimanus in Lake Winnipeg based on ship-based pelagic plankton tows at 65 sites in the summer (13 July-9 August) and fall (13 September-10 October) of 2012; three lake areas are separated by lines; n = number of individuals.

Figure 3 .
Figure 3. Relationship between drift density of Bythotrephes longimanus and water velocity for surface drift traps (see photo insert) deployed in the Winnipeg River at Point du Bois from 3-8 June, 2012; n = number of individuals.

Table 1 .
Location and information for chronological sample collections in three waterbodies with indication of presence of Bythotrephes longimanus.Samples were collected from 2007 to 2015, however, Bythotrephes was not detected until 2009.Numbers in parentheses indicate sample size for each gear type.

Table 2 .
Drift density (no. of individuals/100 m 3 ) of Bythotrephes longimanus in surface and bottom drift traps set in the Winnipeg River at Pointe du Bois in 2010 and 2012; SE = standard error of the mean, N = number of samples.