Advertisement

Hydrobiologia

, Volume 810, Issue 1, pp 119–131 | Cite as

Sonar technology and underwater imagery analysis can enhance invasive Dreissena distribution assessment in large rivers

  • K. MehlerEmail author
  • L. E. Burlakova
  • A. Y. Karatayev
  • Z. Biesinger
  • A. Valle-Levinson
  • C. Castiglione
  • D. Gorsky
FRESHWATER BIVALVES

Abstract

Dreissena spp. are aggressive invaders of many waterbodies worldwide. However, the accurate assessment of their spatial distribution in large rivers is difficult using traditional sampling techniques such as Ponar grabs or SCUBA diving. The aim of this study was to use sonar technologies and underwater imagery (videos, still images) in tandem with traditional Ponar sampling to predict Dreissena presence, and produce a habitat suitability map to enhance our understanding of its spatial distribution in the lower Niagara River, New York, USA. Geo-referenced maps of environmental variables were generated using three sonar technologies: side scan sonar, multibeam sonar, and an Acoustic Doppler Current Profiler. Dreissena presence/absence was determined at 102 sites along a 10 km stretch using Ponar grabs supplemented by an underwater imagery. Substrate and near-bottom flow were the most important variables affecting Dreissena distribution. Habitats with coarse substrate and near-bottom flow of 0.6–0.80 m/s were predicted to be most often occupied. The habitat suitability model indicates that almost 90% of the stream bed in the river can be considered highly- or moderately suitable habitat. Our results demonstrate that supplementing traditional sampling with sonar technologies and underwater imagery can greatly improve Dreissena distribution assessment at the ecosystem scale.

Keywords

Dreissena spp. Large rivers Sonar technologies Underwater imagery Habitat suitability modelling 

Introduction

The exotic Dreissena polymorpha (Pallas 1771), the zebra mussel, and D. rostriformis bugensis (Andrusov 1897), the quagga mussel, are considered the most aggressive freshwater invaders in the Northern hemisphere (Karatayev et al., 2002, 2007, 2015a). Laurentian Great Lakes and particular Lake Erie were among the first waterbodies in North America colonized by zebra and quagga mussels (Mills et al., 1993; Carlton, 2008). Their proliferation has dramatic ecological and socioeconomic impacts, including a decrease in native benthic species (bivalves, amphipods; Howell et al., 1996; Vanderploeg et al., 2002; Karatayev et al., 2011, 2015a), shifts in benthic food webs (Higgins & Vander Zanden, 2010), and a dramatic increase in expenditures for drinking water facilities and power plants to control or remove Dreissena (Conelly et al., 2007). Large rivers are important vectors of spread and habitat for Dreissena (Orlova et al., 2004; Leuwen et al., 2009; Sanz-Ronda et al. 2014). Once Dreissena has established a stable population in a river, the ecological consequences are widespread, including the loss of native unionids (Ricciardi et al., 1996; Lucy et al., 2014), shifts in benthic communities (Ward & Ricciardi, 2007), and shifts in higher trophic levels (Strayer et al., 2004). On the other hand, it has been shown that native fish species could benefit by the increase of prey items usually associated with Dreissena colonies (Ricciardi et al., 1997; Rodriguez, 2006). The impacts of Dreissena in a given waterbody depend on their total population density, population dynamics, and the distribution within a waterbody (Karatayev et al., 2015a). Therefore, the accurate assessment of Dreissena distribution in rivers based on biotic and abiotic factors is essential. The most common way to study Dreissena is using a grab sampler or SCUBA diving. Traditional sampling methods such as Ponar grab or Ekman dredge sampler have been widely used in the assessment of Dreissena distribution in large rivers due to its simple handling and the possibility to obtain density and biomass information (Strayer et al., 2006; Strayer & Malcom, 2007; Strayer et al., 2011). However, Ponar sampling is restricted to soft and small-grain substrate (Van Rein et al., 2009) and low-energy habitats (Bingham et al., 1982). Ponar sampling can provide imprecise information of Dreissena abundance due to their relatively small surface areas (0.025–0.05 m2) and Dreissena’s patchy distribution makes is difficult to accurately assess their true abundance. While SCUBA diving is the most accurate sampling method for density and biomass estimation, its application is expensive and labor intense, restricted to warmer seasons, and may be limited by higher flow velocities (Mellina & Rasmussen, 1994). Ponar sampling and SCUBA diving are most efficient at the small scale; however, different techniques are required for Dreissena distribution assessment over larger scales (Van Rein et al., 2009). Sonar technology and underwater imagery have become powerful tools in aquatic ecology to enhance spatial distribution assessment and benthic mapping on the ecosystem scale (Cochrane & Lafferty, 2002; Kenney et al., 2003; Ninio et al., 2003; Yeung & McConnaughey, 2008; Lietz et al., 2015). The application of sonar technology is less limited by environmental factors, and environmental variables (e.g., substrate and depth) can be efficiently obtained over large areas.

Side scan sonar in combination with underwater imagery analysis and Geographic Information Systems (GIS) has been successfully applied to assess Dreissena distributions in lakes (Coakley et al., 1997; Berkman et al., 1998; Haltuch & Berkman, 2000) and in large rivers (Strayer et al., 2006). Lietz et al. (2015) showed that the estimation of the overall nearshore area of Great Lakes infested with Dreissena increased by 15% when Ponar sampling was supplemented by underwater imagery compared to Ponar grabs alone. Haltuch & Berkman (2000) modeled Dreissena coverage in Lake Erie based on differences in the backscatter signal in side scan sonar images. While this method works well in soft substrate where Dreissena beds are easily visible in the sonar image due to distinctive differences in the backscattered signal (Berkman et al., 1998), it would be impossible to discriminate Dreissena beds from coarse substrate in sonar images due to similar backscatter signals from mussels and large substrate particles (e.g., gravel, cobble, or bedrock). The Niagara River is an important corridor for wildlife and spawning habitat for several fish species including the native lake sturgeon (Acipenser fulvescens). Dreissena was first detected in the Niagara River corridor near the outflow of Lake Erie in October 1989 and in summer of 1990 near the inflow of Lake Ontario (Howell et al., 1996). Recent studies showed that Dreissena can have negative impacts on lake sturgeon, such as reduced foraging and habitat use (McCabe et al., 2006), and it can modify the community structure and spatial distribution of benthic invertebrates (Ricciardi et al., 1997; Gonzales & Burkart, 2004; Burlakova et al., 2012; Kobak et al., 2014). Nevertheless, there is no current information on Dreissena spatial distribution in Niagara River. The size of the river, the morphology of the stream bed, and high-flow currents (Murdoch & Williams, 1989) however would hamper an accurate distribution assessment based simply on Ponar grab data.

The objective of this study was complementing sonar techniques with underwater imagery to improve Dreissena distribution assessment and habitat suitability modeling in the lower Niagara River. While side scan, multibeam, and acoustic Doppler provide full-coverage maps of substrate composition, bathymetry, and near-bottom flow over a large area, Ponar grab samples assess Dreissena presence/absence and density in point locations of soft substrate. On the other hand, video imagery provides information about the presence/absence of Dreissena in areas where the utilization of the Ponar is limited by coarse substrate or high flow. Coupling this information with habitat suitability models can enhance the assessment of Dreissena distribution on the ecosystems scale.

Methods

Study area

The study was carried out within a 10.5 km reach of the Niagara River below Niagara Falls (referred to as the lower Niagara River) between the towns of Lewiston and Youngstown, New York State (Fig. 1). The mean channel width was 700 m, with the mean and maximal water depth of 10 and 25 m, respectively. The substrate in the river is highly variable ranging from silty sand to large boulders and bedrock. The calcium concentration ranges from 26.6 to 41.9 mg/l, and the NO3-N and PO4-P range from 0.14–0.26 to 0.05–0.07 mg/l, respectively (New York Power Authority, 2005). The mean near-bottom flow is ~0.5 m/s with rapids reaching 1.7 m/s. The flow regime is disturbed by a hydroelectric power plant with strong diurnal water level fluctuations varying between 0.3 and 0.6 m (NYPA, 2005).
Fig. 1

Location of the study area in the lower Niagara River, New York, USA

Data collection

Substrate data were collected in 2011 by the U.S. Fish & Wildlife Service using a 400/900 kHZ Edgetech 4125-p side scan sonar. Raw sonar data were processed using SonarWiz (version 5, Chesapeake Technology Software, Mountain View, California), imported into ArcMap 10.1 (Esri, Redlands, California), and manually digitized based on backscatter intensity and surface roughness. The manual classification method was used according to Chapman (2015), and the substrate was classified by visual inspection of the side scan sonar image. Ninety-five geo-referenced underwater videos and 55 sediment samples were used to verify the accuracy of the substrate map (Table 1).
Table 1

Areal coverage of major substrate types in the lower Niagara River based on side scan sonar survey done in 2011 by the U.S. Fish & Wildlife Service

Substrate

Classification accuracy (%)

Coverage (km2)

Coverage (%)

Bedrock and boulder

90

4.5

30

Bedrock

60

0.6

4

Gravel–Cobble mixture

45

5.4

36

Predominantly gravel

90

1.3

9

Predominantly sand

50

0.8

5

Silty sand

86

0.3

2

Macrophytes

75

0.9

6

Shoreline

0.9

6

No data

0.3

2

Sum

 

15

100

The classification accuracy is calculated by comparing the classified substrate with the actual substrate based on underwater imagery and sediment samples

A substrate layer was generated in ArcMap with seven major classes: (i) bedrock, (ii) bedrock and boulder, (iii) gravel–cobble mixture, (iv) predominantly gravel, (v) predominantly sand, (vi) silty sand, and (vii) silty sand with submerged aquatic vegetation (referred to as macrophytes) (Fig. 2). The division of substrate into categories was based on still images of underwater imagery (for bedrock and boulder, bedrock, gravel–cobble mixture, and macrophytes), on sieving analyses of unconsolidated material from Ponar samples (gravel, sand, and silty sand). We calculated the area of each substrate type (Table 1) and then randomly chose sampling sites in each class with the number of sites proportional to class area, maintaining the minimum number of sampling sites for each substrate at 10. In total, 102 sites were surveyed in July and August of 2014. Water temperatures among sites ranged from 19°C to 23°C. Dreissena presence/absence was determined using a downward-looking GoPro Camera (Hero-3) secured to the line above a petite Ponar (0.0225 m2). A total of 95 underwater videos were collected, representing 93% of all sites sampled in the lower Niagara River. At 7 sites, the camera could not be used due to battery or light malfunction. The distance between the camera and the top of the Ponar was 0.5 m, and recording duration was set to at least 1 min. Ponar grab samples were taken from areas with unconsolidated material (e.g., sand and gravel), and Dreissena density was determined by extrapolating the number of individuals in the Ponar to an area of 1 m2. Primary bathymetry data were collected in 2014 and 2015 by the U.S. Fish & Wildlife Service using a Teledyne Odom Hydrographic ES-3 240 kHz multibeam echosounder (Seafloor Systems, Inc., Cameron Park, California). Raw sonar data were collected and processed using Triton Perspective (Triton Imaging, Inc., Capitola, California). Multibeam bathymetry data were augmented with depth measurements from the Acoustic Doppler Current Profiler. Both data sources were used to develop a complete depth map in ArcMap (Fig. 2). Near-bottom flow data were collected in July 2015 by the U.S. Fish & Wildlife Service using a Teledyne RDI 600 kHz Workhorse Monitor Acoustic Doppler Current Profiler (Teledyne RD Instruments, Poway, California). The profiler was floated downward-looking and towed at the surface making many transects covering the entire study area. Data from the deepest 2 m were averaged to give a single flow velocity at each location. Although biological and near-bottom flow data were taken in different years, records indicate that there was only slight variation in discharge (6200 m3/s in 2014 vs. 6700 m3/s in 2015, diff: 7%, International Niagara Board of Control 2014, 2015). Biological and environmental data were collected at the same time of the day (from 10 AM to noon) to avoid sampling bias due to the strong diurnal variation in discharge caused by the power plants. A complete flow map was interpolated using Delaunay triangulation (Fang & Piegl, 1993) from raw velocity data and imported into ArcMap (Fig. 2).
Fig. 2

Geo-referenced raster layers (substrate, depth, and near-bottom flow) used to generate the Dreissena habitat suitability map in ArcMap 10.1

Data analyses

Underwater imagery analysis followed the approach of Lietz et al. (2015). Video quality was classified as excellent (Dreissena presence/absence could be clearly assessed), marginal (Dreissena presence/absence could not be confidentially assessed and should be used with caution), or poor (Dreissena presence/absence could not be assessed). In areas where Dreissena were identified using the video, we did not distinguish between D. polymorpha and D. bugensis. To determine whether Dreissena detectability (1 = yes, 0 = no) was dependent on Dreissena density (tested for sand and gravel substrate only), a logistic regression was used according to Lietz et al. (2015). Dreissena detectability in video was regressed against Dreissena density from sand and gravel substrate using JMP SAS (JMP®, Version 12, SAS Institute Inc., Cary, North Carolina). In the logistic regression model, only videos of excellent and marginal quality were used. Sites where Dreissena were detected in the video but were not found in the Ponar samples were excluded from the model (2 sites). A multiple logistic regression model was used to predict Dreissena presence based on a set of environmental variables. In the model, depth, substrate, and near-bottom flow were cast as model effects and were used separately as well interactively. The response variable was Dreissena presence/absence. Then, a classification tree was created using recursive partitioning to determine the probability of Dreissena presence based on the range of each environmental variable. The classification tree is split based on the predictor variable that explains the most in the response variable. The data will be partitioned until further splits will not increase R 2.

Dreissena habitat preference indices were calculated using the ratio between habitat use (i.e., frequency of Dreissena occurrence) and habitat availability. This was done for substrate, depth, and near-bottom flow. For habitat preference analysis, continuous depth measurements were categorized into twelve depth zones: 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, and 24 m. Also, continuous near-bottom flow measurements were divided into seven velocity categories: 0–0.2, 0.21–0.4, 0.41–0.6, 0.61–0.8, 0.81–1.00, 1.01–1.20, and 1.21–1.28 m/s. Preference indices were normalized by dividing each preference value by the maximum value. A value of 0 represents least preferred habitat, while a value of 1 represents the most preferred habitat.

In ArcMap a Dreissena habitat suitability map was generated using the weighted overlay analysis within the model builder function. Habitat preference indices were used to reclassify the environmental variables. Results from statistical analyses were used to determine the weight of importance of each of the three reclassified input raster based on the importance of each environmental variable. The reclassify function has the advantage to reassign values in each input raster into a common evaluation scale of suitability or preference. Substrate, near-bottom flow, and depth values were reclassified (Table 2) ranging from 1 (least favorable) to 5 (most favorable). The reclassified environmental variables were then used in the weighted overlay analysis, and the weight of importance for each variable was assumed (e.g., influence of each environmental variable in %). Then, the final output raster (Suitability Index map) was generated with a suitability score for each pixel. Weights of importance were assumed: near-bottom flow: 40%, substrate: 50%, and depth: 10%. To evaluate the accuracy of the map, the number of sites with Dreissena presence/absence for each suitability index was divided by the total number of sites.
Table 2

Reclassified values for substrate, near-bottom flow, and depth used in the habitat suitability model. Reclassified values are dimensionless and ranges from 1 (least favorable) to 5 (most favorable)

Environmental variable

Reclassified as

Substrate

 Bedrock and boulder

5

 Bedrock

3

 Gravel–Cobble mixture

5

 Predominantly gravel

4

 Predominantly sand

3

 Silty sand

1

 Macrophytes

1

Near-bottom flow (m/s)

 0–0.2

3

 0.21–0.4

3

 0.41–0.6

4

 0.61–0.8

5

 0.81–1.00

3

 1.01–1.20

2

 0 1.21–1.28

1

Depth (m)

 2

1

 4

3

 6

5

 8

4

 10

4

 12

5

 14

4

 16

4

 18

3

 20

4

 22

5

 24

5

Results

Underwater video performance

Of all videos, 58% had an excellent quality rating, 21% a marginal rating, and 21% a poor rating; 53% of the marginal and poor videos had controllable reason for their rating, i.e., the camera light was not adjusted properly or the camera was not in focus, and 47% had uncontrollable reasons, including (1) mussels could not be properly distinguished from the substrate, (2) macrophytes obscured the view of the mussels, and (3) the water was too turbid. Dreissena were present at 68 out 102 sampling sites (67%). At 15 sites, Dreissena was visible in the video but was either not present in the Ponar (2 sites) or the Ponar did not work due to rocky substrate (13 sites). Conversely, at 13 sites, Dreissena was detected in Ponar samples but was not visible in the video. However, 50% of those sites had only 1–4 mussels in Ponar samples (44–178 mussels/m2). The detectability of Dreissena in videos is significantly related to its abundance for both substrates (sand: Χ 2 = 13, P < 0.01; gravel: Χ 2 = 6.75, P < 0.05, Fig. 3).
Fig. 3

Logistic regression to predict Dreissena detection probability in excellent and marginal video images for sand (a Χ 2 = 13.0, df = 1, P < 0.01) and gravel (b Χ 2 = 6.75, df = 1, P < 0.05)

Effects of environmental variables on Dreissena presence/absence

Of the environmental variables ‘substrate,’ ‘near-bottom flow,’ and ‘substrate * near-bottom flow’ were best predictors of Dreissena presence, while depth had no significant effects on Dreissena presence. The full model was highly significant (χ 2 = 18.7, P = 0.0009), and the area under the ROC curve was 0.7527 indicating good accuracy of the model. The decision tree (Fig. 4) showed that the highest probabilities of Dreissena presence (>95%) were predicted for three different combinations of the environmental variables: (i) all types of substrate—except silty sand—and near-bottom flow between ≥0.57 and ≤0.8 m s−1; (ii) near-bottom flow <0.26 m s−1 and larger substrate (bedrock and boulder and gravel–cobble mixture); and (iii) near-bottom flow ≥0.26 m s−1, larger substrate (bedrock and boulder and gravel–cobble mixture), and depths > 9.1 m. The probability of Dreissena presence approached zero in areas deeper than 6 m with substrate composed of silty sand. Results from the decision tree were similar to those from the habitat preference indices (Fig. 5). Dreissena in the lower Niagara River were found on all types of substrate, preferably bedrock and gravel–cobble mixture, throughout all depths, except very shallow sites (<1.5 m) and near-bottom flows between 0.2 and 1.0 m/s. No Dreissena were observed in areas where the flow velocity was above 1.10 m/s.
Fig. 4

Classification tree to predict Dreissena presence/absence based on environmental variables substrate, near-bottom flow, and depth. For each terminal split, the probability for Dreissena presence is given in brackets. There was no significant increase in R 2 after eleven splits

Fig. 5

Dreissena habitat preference indices for substrate (a), depth (b), and near-bottom flow (c). A value of 0 represents least preferred habitat, while a value of 1 represents the most preferred habitat

Habitat suitability map

The habitat suitability map (Fig. 6) indicated that high-suitability habitats (18%) were along the entire length of the river and corresponded very well with areas where the near-bottom flow ranged between 0.41 m/s and 0.80 m/s. The dominant substrate in these high-suitability habitats was mostly gravel–cobble mixture and bedrock and boulder. About 70% of the habitat fell in the medium-suitability habitat range and was typically located in the center of the stream channel and the nearshore areas. Although medium-suitability habitats in the center of the stream channel often provided the most favorable substrate such as bedrock and gravel–cobble mixture, near- bottom flow was often exceeding the optimal range for Dreissena. Contrarily, at nearshore areas, even though the near-bottom flow was within the optimal range for Dreissena, these areas were classified as medium-suitability habitats. Only 12% of the study area was classified as low-suitability habitat. Areas within this category were located near the outflow of the lower Niagara River into Lake Ontario and along the shallow nearshore areas of the river. The dominant substrate type in these habitats was silty sand and silty sand covered with dense stands of macrophytes or bare bedrock areas with near-bottom flow above 0.80 m/s. Accuracy evaluation of the map revealed that 92% of the sites that had Dreissena were located in areas with a high- or medium-suitability index, while only 8% of sites that had Dreissena were located in areas with a low-suitability index. On the other hand, 45% of the sites that had no Dreissena were located in areas with a low-suitability index, while 48% of sites that had no Dreissena were located in areas with high- or medium-suitability index. However, these results should be used with caution as the same data that were used to develop the habitat suitability map were also used to assess the accuracy. Therefore, the error may be underestimated.
Fig. 6

Map of habitat suitability index for Dreissena in the lower Niagara River

Discussion

Understanding factors controlling species abundance is crucial to improve the management of both natural populations and introduced species. Identifying limiting factors using the approach shown here is not only useful for Dreissena distribution but can also be used to assess the spatial distribution of native bivalves, mollusks, and other sessile benthic organism. Our study found that the most important limiting factors for Dreissena distribution in a large river were substrate and near-bottom flow.

Depth as an environmental variable did not seem to play a major role in Dreissena distribution except for shallow areas. The absence of Dreissena in these nearshore areas might be caused by physical factors, such as ice formation during winter or temperatures above their optimum during hot summer months. The lack of Dreissena in nearshore areas may also be explained by fish predation, as a number of fish, including lake sturgeon and round goby, are known to consume Dreissena (Molloy et al., 1997; Bartsch et al., 2005).

We do not know to what extent the factor ‘depth’ affects Dreissena biomass or density as these parameters were not determined at all sites. Jones & Ricciardi (2005) found a positive relationship between depth and Dreissena bugensis biomass in Saint Lawrence River. They assumed that this species is more susceptible to wave action and ice scour, and thus avoids shallow areas. Our analysis showed that the one of the limiting factors for Dreissena distribution in the lower Niagara River was near-bottom flow. In rivers, strong unidirectional flow can pose several challenges for a successful Dreissena settlement and population (Karatayev et al., 1998). Dreissena was absent in areas where the flow velocity exceeded 1.10 m/s. This is similar with a study by Sanz-Ronda et al. (2014) who showed for the Ebro River, Spain that D. polymorpha was absent in areas where the flow velocities exceeded 1.20 m/s. Dreissena has an obligate planktonic life stage, and higher flow velocities can significantly reduce successful settlement of larvae in rivers (Stoeckl et al., 1997; Karatayev et al., 1998; Horvath & Crane, 2010). Several studies showed that the abundance of Dreissena in a river depends on larval recruitment from an upstream lake population (Thorp et al., 2002). The upstream source of the Niagara River is Lake Erie with D. bugensis being the dominant species (Patterson et al., 2005; Karatayev et al., 2015b). Although we believe that both Lake Erie and the upper Niagara River are the dominant sources of larval supply for the lower Niagara River, the 35 m drop at Niagara Falls might represent a significant source of larval mortality. Comparing larval abundance upstream and downstream of Niagara Falls could quantify the falls’ effect on larval survival.

The other important limiting environmental factor for Dreissena distribution indicated by our analysis was substrate type. It is known that the availability of suitable substrate can significantly affect Dreissena presence in lakes and rivers (Karatayev et al., 1998). Dreissena in the lower Niagara River was observed more frequently in areas with hard substrate such as fractured bedrock compared to soft substrate such as sand and silty sand. In the Saint Lawrence River, Jones & Ricciardi (2005) observed a positive relationship between Dreissena biomass and substrate size. In our study, Dreissena also showed a high preference for habitats with gravel–cobble mixture as expected for lotic ecosystems with unidirectional flow. Shear stress and movement of unconsolidated substrate can be important factors limiting the settlement of Dreissena larvae in lotic environments even in the presence of appropriate substrate (Horvath & Lamberti, 1999; Quinn & Ackerman, 2014). Indeed, Sanz-Ronda et al. (2014) found D. polymorpha beds in the Ebro River, Spain were chiefly located in areas with stable substrate. The lower presence of Dreissena in the lower Niagara River at sites with gravelly substrate compared to those with bedrock and boulder might therefore be attributed to gravel movement. Also, areas with bare bedrock were less preferred; the lack of flow refuges in combination with high-flow velocities may prevent Dreissena larvae from settling successfully in those areas.

The effect of the large diurnal variation in discharge on Dreissena distribution at nearshore areas of the lower Niagara River caused by the New York Power Authority power projects is unknown. Dreissena in these areas were virtually absent even in the presence of hard substrate, such as riprap or gravel–cobble mixture. Baumgaertner et al. (2008) showed that the hydraulic stress caused by water level fluctuations had tremendous effects on the benthic communities at the nearshore areas of Lake Constance. Aerial exposures during low water levels pose another stress factor for Dreissena, although water level fluctuations occur usually on a diurnal basis and Dreissena can survive aerial exposure for at least 42 h (Collas et al., 2014). We therefore believe that the physical disturbance caused by water level fluctuations may be one of the driving forces affecting the distribution of Dreissena in the nearshore areas. Jones & Ricciardi (2005) argued that especially D. bugensis is less able to adhere tighter to hard substrate due to their round shells and might therefore be more susceptible to wave action.

Although our habitat suitability model predicts that 95% of the lower Niagara River is considered suitable Dreissena habitat, 16% of our sampling sites classified as highly suitable had no Dreissena, suggesting Dreissena are either not habitat limited as shown in other studies (e.g., Hastie et al., 2000) or that other factors are important. Calcium concentration has been identified as a potentially limiting factor in rivers and lakes (Mellina & Rasmussen, 1994; Karatayev et al., 2015b). However, the calcium concentration in the Niagara River (35.5–37.5 mg/l, EPA 1976; 26.6-41.9 mg/l, New York Power Authority, 2005) is much higher than the limits reported in the literature, 8–23.1 mg/l (D. polymorpha) and 12 mg/l (Dreissena bugensis), respectively, in the St. Lawrence River (Jones & Ricciardi, 2005; Karatayev et al., 2015b). Although low chronic oxygen levels can affect the survivorship of Dreissena (Johnson & McMahon, 1998), frequent measurements of dissolved oxygen concentration in the lower Niagara River suggest hypoxic conditions do not occur and would therefore not limit Dreissena. We also cannot exclude the presence of uninhabitable areas within high-suitability habitats unfavorable for Dreissena due to small-scale variability of the environmental factors not captured in our model.

Overall, this study demonstrated the advantage of augmenting traditional sampling methods combined with underwater video and sonar technologies to better understand the distribution of Dreissena at the ecosystem scale. In this study, 70% of the streambed in the lower Niagara River was composed of hard substrate—the preferred habitat for Dreissena. These areas could not be sampled by Ponar grabs, and Dreissena would have gone undetected. However, coupling underwater imagery with sonar technologies can overcome those limitations. While underwater imagery enhanced Dreissena detection at the small scale, the application of sonar technology and remote sensing allowed extrapolating physical habitat information and the presence/absence to the ecosystem scale by generating habitat suitability maps.

Ponar sampling cannot be completely replaced by sonar technologies nor underwater imagery as it allows for density and size-frequency estimation. Underwater imagery can also be used as a supplement to Ponar sampling by providing important habitat information beyond the sampling point. Underwater imagery is an affordable and effective way to provide a broader insight in habitat heterogeneity, including information about substrate composition and bed forms, the presence of submerged aquatic vegetation, and the presence of potential predators.

Large-scale habitat preference information for Dreissena is essential to predict its potential spread and can be used to support appropriate management actions such as early detection monitoring, flow control (in regulated rivers), or the application of biocides. Our model provides the management framework by incorporating the most important environmental factors and their interactions to describe Dreissena distribution in the lower Niagara River. The model can be used as a tool for river managers to assess future colonization and potential ecological consequences for fish spawning habitats or feeding grounds. It can be applied as a basis for large-scale habitat management to link environmental factors and the spatial distribution of benthic species in a cost-effective and time-efficient way. Another advantage of our approach used is that it is not restricted to Dreissena but can be applied to assess the distribution of other benthic macroinvertebrates or benthivorous fish in river, estuaries, nearshore coastal areas, or lakes.

Notes

Acknowledgements

This study was funded by Ecological Greenway Fund entitled ‘Investigating Lake Sturgeon habitat use, feeding ecology and benthic resource availability in the lower Niagara River’, (Project 1113459, Award 66141). The authors thank Eric Bruestle, Joshua Fisher, and Mark Clapsadl from the Great Lakes Center at SUNY Buffalo State for their assistance in the field. The authors are grateful to Master student Anthony Cevaer from the Great Lakes Center at SUNY Buffalo State who helped processing sampled in the laboratory. Finally, the authors sincerely thank the anonymous reviewers for their constructive comments on this manuscript.

References

  1. Bartsch, M. R., L. A. Bartsch & S. Gutreuter, 2005. Strong effects of predation by fishes on an invasive macroinvertebrate in a large floodplain river. Journal of North American Benthological Society 24: 168–177.CrossRefGoogle Scholar
  2. Baumgaertner, D., M. Mörtl & K.-O. Rothhaupt, 2008. Effects of water-depth and water-level fluctuations on the macroinvertebrate community structure in the littoral zone of Lake Constance. Hydrobiologia 613: 97–107.CrossRefGoogle Scholar
  3. Berkman, P. A., M. A. Haltuch, E. Tichich, D. W. Garton, G. W. Kennedy, J. E. Gannon, S. D. Mackey, J. A. Fuller & D. L. Liebenthal, 1998. Zebra mussels invade Lake Erie muds. Nature 393: 27–28.CrossRefGoogle Scholar
  4. Bingham, C. R., Mathis, D.B., L. G. Sanders & E. McLemore, 1982. Grab samplers for benthic macroinvertebrates in the lower Mississippi River Environmental and Water Quality Operational Studies. Final Report. Miscellaneous Paper E-82-3.Google Scholar
  5. Burlakova, L. E., A. Y. Karatayev & V. A. Karatayev, 2012. Invasive mussels induce community changes by increasing habitat complexity. Hydrobiologia 685: 121–134.CrossRefGoogle Scholar
  6. Carlton, J. T., 2008. The zebra mussel Dreissena polymorpha found in North America in 1986 and 1987. Journal of Great Lakes Research 34: 770–773.CrossRefGoogle Scholar
  7. Chapman, S., 2015. Comparison of side scan sonar substrate classification methods to assess accuracy. Master thesis. State University of New York. University of Buffalo.Google Scholar
  8. Coakley, T. P., G. L. Brown, S. E. Ioannou & M. N. Charlton, 1997. Colonization patterns and densities of Zebra Mussel Dreissena in muddy offshore sediments of western Lake Erie, Canada. Water Air Soil Pollution 99: 623–632.Google Scholar
  9. Cochrane, G. R. & K. D. Lafferty, 2002. Use of acoustic classification of sidescan sonar data for mapping benthic habitats in the Northern Channel Islands, California. Continetal Shelf Research 22: 683–690.CrossRefGoogle Scholar
  10. Collas, F. P. L., K. R. Koopman, A. J. Hendriks, G. van der Velde, L. N. H. Verbrugge & R. S. E. W. Leuven, 2014. Effects of desiccation on native and non-native molluscs in rivers. Freshwater Biology 59: 41–55.CrossRefGoogle Scholar
  11. Connelly, N. A., C. R. O’Neill Jr., B. A. Knuth & T. L. Brown, 2007. Economic impacts of zebra mussels on drinking water treatment and electric power generation facilities. Environmental Management 40: 105–112.CrossRefPubMedGoogle Scholar
  12. Fang, T. & L. Piegl, 1993. Delaunay triangulation using a uniform grid. IEEE Computer Graphics and Applications 13: 36–47.CrossRefGoogle Scholar
  13. Gonzalez, M. & G. Burkart, 2004. Effects of food type, habitat and fish predation on the relative abundance of two amphipod species, Gammarus fasciatus and Echinogammarus ischnus. Journal of Great Lakes Research 30: 100–113.CrossRefGoogle Scholar
  14. Haltuch, M. A. & P. A. Berkman, 2000. Geographic information system (GIS) analysis of ecosystems invasion: Exotic mussels in Lake Erie. Limnology and Oceanography 45: 1778–1787.CrossRefGoogle Scholar
  15. Hastie, L. C., P. J. Boon & M. R. Young, 2000. Physical microhabitat requirements of freshwater pearl mussels, Margaritifera margaritifera (L.). Hydrobiologia 429: 59–71.CrossRefGoogle Scholar
  16. Higgins, S. N. & M. J. Vander Zanden, 2010. What a difference a species makes: a meta–analysis of dreissenid mussel impacts on freshwater ecosystems. Ecological Monographs 80: 179–196.CrossRefGoogle Scholar
  17. Horvath, T. G. & L. Crane, 2010. Hydrodynamic forces affect larval zebra mussel (Dreissena polymorpha) mortality in a laboratory setting. Aquatic Invasions 5: 379–385.CrossRefGoogle Scholar
  18. Horvath, T. G. & G. A. Lamberti, 1999. Mortality of zebra mussel, Dreissena polymorpha, veligers during downstream transport. Freshwater Biology 42: 69–76.CrossRefGoogle Scholar
  19. Howell, E. T., C. H. Marvin, R. W. Bilyea, P. B. Kauss & K. Somers, 1996. Changes in environmental conditions during Dreissena colonialization of a monitoring station in Eastern Lake Erie. Journal of Great Lakes Research 22: 744–756.CrossRefGoogle Scholar
  20. Johnson, P. D. & R. F. McMahon, 1998. Effects of temperature and chronic hypoxia on survivorship of the zebra mussel (Dreissena polymorpha) and Asian clam (Corbicula fluminea). Canadian Journal of Fisheries and Aquatic Sciences 55: 1564–1572.CrossRefGoogle Scholar
  21. Jones, L. A. & A. Ricciardi, 2005. Influence of physicochemical factors on the distribution and biomass of invasive mussels (Dreissena polymorpha and Dreissena bugensis) in the St. Lawrence River. Canadian Journal of Fisheries and Aquatic Sciences 62: 1953–1962.CrossRefGoogle Scholar
  22. Karatayev, A. Y., L. E. Burlakova & D. K. Padilla, 1998. Physical factors that limit the distribution and abundance of Dreissena polymorpha (pall.). Journal of Shellfish Research 17: 1219–1235.Google Scholar
  23. Karatayev, A. Y., L. E. Burlakova & D. K. Padilla, 2002. Impacts of zebra mussels on aquatic communities and their role as ecosystem engineers. In Leppakoski, E., S. Gollach & S. Olenin (eds), Invasive Aquatic Species of Europe: Distribution, Impacts and Management. Kluwer Academic Publishers, Dordreicht.Google Scholar
  24. Karatayev, A. Y., D. K. Padilla, D. Minchin, D. Boltovskoy & L. E. Burlakova, 2007. Changes in global economies and trade: the potential spread of exotic freshwater bivalves. Biological Invasions 9: 161–180.CrossRefGoogle Scholar
  25. Karatayev, A. Y., L. E. Burlakova, S. E. Mastitsky, D. K. Padilla & E. L. Mills, 2011. Contrasting rates of spread of two congeners, Dreissena polymorpha and Dreissena rostriformis bugensis, at different spatial scales. Journal of Shellfish Research 30: 923–931.CrossRefGoogle Scholar
  26. Karatayev, A. Y., L. E. Burlakova & D. K. Padilla, 2015a. Zebra versus quagga mussels: a review of their spread, population dynamics, and ecosystem impacts. Hydrobiologia. 746: 97–112.CrossRefGoogle Scholar
  27. Karatayev, A. Y., L. E. Burlakova, S. E. Mastitsky & D. K. Padilla, 2015b. Predicting the spread of aquatic invaders: insight from 200 years of invasion by zebra mussels. Ecological Applications 25: 430–440.CrossRefPubMedGoogle Scholar
  28. Kenney, A. J., I. Cato, M. Desperez, G. Fader, R. T. F. Schuettenhelm & J. Side, 2003. An overview of seabed-mapping technologies in the context of marine habitat classification. ICES Journal of Marine Science 60: 411–448.CrossRefGoogle Scholar
  29. Kobak, J., L. Jermacz & D. Plachocki, 2014. Effectiveness of zebra mussels to act as shelters from fish predators differs between native and invasive amphipod prey. Aquatic Ecology 48: 397–408.CrossRefGoogle Scholar
  30. Leuven, R. S. E. W., G. van der Velde, I. Baijens, J. Snijders, C. van der Zwart, H. J. R. Lenders & A. bij de Vaate, 2009. The river Rhine: a global highway for dispersal of aquatic invasive species. Biological Invasions 11: 1989–2008.CrossRefGoogle Scholar
  31. Lietz, J. E., J. R. Kelly, J. V. Scharold & P. M. Yurista, 2015. Can a rapid underwater video approach enhance the benthic assessment capability of the national coastal condition assessment in the Great Lakes? Environmental Management 55: 1446–1456.CrossRefPubMedGoogle Scholar
  32. Lucy, F. E., L. E. Burlakova, A. Y. Karatayev, S. E. Mastitsky & D. T. Zanatta, 2014. Zebra mussel impacts on unionids: A synthesis of trends in North America and Europe. In Nalepa, T. F. & D. W. Schloesser (eds.), Quagga and Zebra Mussels: Biology, Impacts, and Control, 2nd ed. CRC Press, Boca Raton: 623–646.Google Scholar
  33. McCabe, D. J., M. A. Beekey, A. Mazloff & J. E. Marsden, 2006. Negative effect of zebra mussels on foraging and habitat use by lake sturgeon (Acipenser fulvescens). Aquatic Conservation: Marine and Freshwater Ecosystems 16: 493–500.CrossRefGoogle Scholar
  34. Mellina, E. & J. B. Rasmussen, 1994. Patterns in the distribution and abundance of zebra mussel (Dreissena polymorpha) in rivers and lakes in relation to substrate and other physicochemical factors. Canadian Journal of Fisheries and Aquatic Science 51: 1024–1036.CrossRefGoogle Scholar
  35. Mills, E. L., R. M. Dermott, E. F. Roseman, D. Dustin, E. Mellina, D. B. Corm & A. P. Spidle, 1993. Colonization, ecology, and population structure of the “quagga” mussel (Bivalvia: Dreissenidae) in the lower Great Lakes. Canadian Journal of Fisheries and Aquatic Sciences 50: 2305–2314.CrossRefGoogle Scholar
  36. Molloy, D. P., A. Y. Karatayev, L. E. Burlakova, D. P. Kurandina & F. Laruelle, 1997. Natural enemies of zebra mussels: Predators, parasites, and ecological competitors. Reviews in Fisheries Science 5: 27–97.CrossRefGoogle Scholar
  37. Murdoch, A. & D. Williams, 1989. Suspended sediments and the distribution of bottom sediment in the Niagara River. Journal of Great Lakes Research 15: 427–436.CrossRefGoogle Scholar
  38. Ninio, R., S. Delean, K. Osborne & H. Sweatman, 2003. Estimating cover of benthic organisms from underwater video images: variability associated with multiple observers. Marine Ecology Progress Series 265: 107–116.CrossRefGoogle Scholar
  39. Niagara Board of Control. 2014. One Hundred Twenty Second Semi-Annual Progress Report to the International Joint Commission. Covering the Period September 20, 2013 through March 25, 2014.Google Scholar
  40. Niagara Board of Control. 2015. One Hundred Twenty Second Semi-Annual Progress Report to the International Joint Commission. Covering the Period September 25, 2014 through March 5, 2015.Google Scholar
  41. NYPA, 2005. New York Power Authority. Surface water quality of the Niagara River and its U.S. tributaries. Niagara Power Project FERC No. 2216.Google Scholar
  42. Orlova, M. I., J. R. Muirhead, P. I. Antonov, G Kh Shcherbina, Y. I. Starobogatov, G. I. Biochino, T. W. Therriault & H. J. MacIsaac, 2004. Range expansion of quagga mussels Dreissena rostriformis bugensis in the Volga River and Caspian Sea basin. Aquatic Ecology 38: 561–573.CrossRefGoogle Scholar
  43. Patterson, M. W. R., J. J. H. Ciborowski & D. R. Barton, 2005. The distribution and abundance of Dreissena species (Dreissenidae) in Lake Erie. Journal of Great Lakes Research 31: 223–237.CrossRefGoogle Scholar
  44. Quinn, N. P. & J. D. Ackerman, 2014. Effects of near-bed turbulence on the suspension and settlement of freshwater dreissenid mussel larvae. Freshwater Biology 59: 614–629.CrossRefGoogle Scholar
  45. Ricciardi, A., F. G. Whoriskey & J. B. Rasmussen, 1996. Impact of the (Dreissena) invasion on native unionid bivalves in the upper St. Lawrence River. Canadian Journal of Fisheries and Aquatic Sciences 53: 1434–1444.CrossRefGoogle Scholar
  46. Ricciardi, A., F. G. Whoriskey & J. B. Rasmussen, 1997. The role of the zebra mussel (Dreissena polymorpha) in structuring macroinvertebrate communities on hard substrates. Canadian Journal of Fisheries and Aquatic Sciences 54: 2596–2608.CrossRefGoogle Scholar
  47. Rodriguez, L. F., 2006. Can invasive species facilitate native species? Evidence of how, when, and why these impacts occur. Biological Invasions 8: 927–939.CrossRefGoogle Scholar
  48. Sanz-Ronda, F. J., S. Lopez-Saenz, R. San-Martın & A. Palau-Ibars, 2014. Physical habitat of zebra mussel (Dreissena polymorpha) in the lower Ebro River (Northeastern Spain): influence of hydraulic parameters in their distribution. Hydrobiologia 735: 137–147.CrossRefGoogle Scholar
  49. Stoeckel, J. A., D. W. Schneider, L. A. Soeken, K. D. Blodgett & R. E. Sparks, 1997. Larval dynamics of a riverine metapopulation: implications for zebra mussel recruitment, dispersal and control in a large-river system. Journal of the North American Benthological Society 16: 586–601.CrossRefGoogle Scholar
  50. Strayer, D. L. & H. M. Malcom, 2007. Effects of zebra mussels (Dreissena polymorpha) on native bivalves: the beginning of the end or the end of the beginning? Journal of North American Benthological Society 26: 111–122.CrossRefGoogle Scholar
  51. Strayer, D. L., K. A. Hattala & A. W. Kahnle, 2004. Effects of an invasive bivalve (Dreissena polymorpha) on fish in the Hudson River estuary. Canadian Journal of Fisheries and Aquatic Sciences 61: 924–941.CrossRefGoogle Scholar
  52. Strayer, D. L., H. M. Malcom, R. E. Bell, S. M. Carbotte & F. O. Nitsche, 2006. Using geophysical information to define benthic habitats in a large river. Freshwater Biology 53: 25–38.CrossRefGoogle Scholar
  53. Strayer, D. L., N. Cid & H. M. Malcom, 2011. Long-term changes in a population of an invasive bivalve and its effects. Oecologia 165: 1063–1072.CrossRefPubMedGoogle Scholar
  54. Thorp, J. H., J. E. Alexander Jr. & G. A. Cobbs, 2002. Coping with warmer, large rivers: a field experiment on potential range expansion of northern quagga mussels (Dreissena bugensis). Freshwater Biology 47: 1779–1790.CrossRefGoogle Scholar
  55. Van Rein, H. B., C. J. Brown & R. Quinn, 2009. A review of sublittoral monitoring methods in temperate waters: a focus on scale. International Journal of the Society for Underwater Technology 28: 99–113.CrossRefGoogle Scholar
  56. Vanderploeg, H. A., T. F. Nalepa, D. J. Jude, E. L. Mills, K. T. Holeck, J. R. Liebig, I. A. Grigorovich & H. Ojaveer, 2002. Dispersal and emerging ecological impacts of Ponto-Caspian species in the Laurentian Great Lakes. Canadian Journal of Fisheries and Aquatic Sciences 59: 1209–1228.CrossRefGoogle Scholar
  57. Ward, J. M. & A. Ricciardi, 2007. Impacts of Dreissena invasions on benthic macroinvertebrate communities: a meta-analysis. Diversity and Distribution 13: 155–165.CrossRefGoogle Scholar
  58. Yeung, C. & R. A. McConnaughey, 2008. Using acoustic backscatter from a sidescan sonar to explain fish and invertebrate distributions: a case study in Bristol Bay, Alaska. ICES Journal of Marine Science 65: 242–254.CrossRefGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2016

Authors and Affiliations

  1. 1.Great Lakes CenterBuffalo State CollegeBuffaloUSA
  2. 2.The Research Foundation of The State University of New YorkBuffalo State College, Office of Sponsored ProgramsBuffaloUSA
  3. 3.US Fish and Wildlife Service, Lower Great Lakes Fish and Wildlife Conservation OfficeBasomUSA
  4. 4.Civil and Coastal Engineering DepartmentUniversity of FloridaGainesvilleUSA

Personalised recommendations